Proc. Nati. Acad. Sci. USAVol. 86, pp. 2844-2848, April 1989Medical Sciences

GLQ223: An inhibitor of human immunodeficiency virus replicationin acutely and chronically infected cells of lymphocyte andmononuclear phagocyte lineage

(AIDS/macrophages/T lymphocytes/antiviral compound)

MICHAEL S. MCGRATHaOb, Kou M. HWANGc"d, SUSAN E. CALDWELLc, ISABELLE GASTONb, KA-CHEUNG LUKc,PAUL WUc'd VALERIE L. NGb e, SUZANNE CROWEb f, JAN DANIELSc, JANE MARSHb, TRCY DEINHARTC,PATRICIA V. LEKASb, JOANN C. VENNARIc, HIN-WING YEUNGg, AND JEFFREY D. LIFSONc'haDivision of AIDS/Oncology, Department of Medicine, and 'Department of Laboratory Medicine, University of California-San Francisco, San Francisco, CA94110; bAIDS/Immunobiology Research Laboratory, and fDivision of Infectious Diseases, Department of Medicine, San Francisco General Hospital, SanFrancisco, CA 94110; Divisions of cCellular Immunology and dMedicinal Biochemistry, Genelabs Incorporated, Redwood City, CA 94063;and gDepartment of Biochemistry and Chinese Medicinal Material Research Center, Chinese University of Hong Kong, Shaitin, NT, Hong Kong

Communicated by Leonard A. Herzenberg, December 27, 1988

ABSTRACT GLQ223 is a highly purified, formulatedpreparation of trichosanthin, a 26-kDa plant-derived ribo-some-inactivating protein with potent inhibitory activityagainst human immunodeficiency virus (HIV) in vitro. Thecompound produced concentration-dependent inhibition ofHIV replication in acutely infected cultures of T-lymphoblas-toid cells (VB cell line). Treatment with GLQ223 selectivelyreduced levels of detectable viral proteins compared to totalcellular protein synthesis and produced a selective decrease inlevels'of viral RNA relative to total cellular RNA in acutelyinfected cells. Substantial inhibition of viral replication wasobserved at concentrations of GLQ223 that showed littleinhibition of parallel uninfected cultures. Selective anti-HIVactivity was also observed in cultures of primary monocyte/macrophages chronically infected with HIV in vitro. Whenfreshly drawn blood samples from HIV-infected patients weretreated with a single 3-hr exposure to GLQ223, HIV replicationwas blocked for at least 5 days in subsequently culturedmonocyte/macrophages, without further treatment. The anti-HIV activity of GLQ223 in both acutely and chronicallyinfected cells and its activity in cells of both lymphoid andmononuclear phagocytic lineage make it an interesting candi-date as a potential therapeutic agent in HIV infection andAIDS.

GLQ223 is a highly purified, formulated preparation oftrichosanthin, a 26-kDa basic protein isolated from roottubers of Trichosanthes kirilowii (1, 20). Based on structuraland functional properties, the protein belongs to the family ofsingle-chain ribosome-inactivating proteins, which inhibit invitro translation in cell-free systems (2). Trichosanthin hasbeen reported to show a56% similarity at the amino acid levelwith the A chain of ricin toxin when both identical andconservative residues are considered (2) and shows substan-tial amino acid similarity with several other single-chainribosome-inactivating proteins (Michael Piatak, personalcommunication). Partially purified preparations of trichosan-thin have been administered in China, in single doses from 5to 12.5 mg, as a midtrimester abortifacient (3) and, in multipledose regimens involving 5-12 mg per dose, for treatment oftrophoblastic tumors (4). In general, these doses have beenreported to be both effective for the intended clinical indi-cation and well tolerated (3, 4).

In this communication, we report that GLQ223 selectivelyinhibits replication of human immunodeficiency virus (HIV),

the etiologic agent for AIDS, in cells of both lymphoid andmononuclear phagocyte cell lineages in vitro.

MATERIALS AND METHODSSource and Purification of GLQ223. Root tubers of T.

kirilowii were obtained from southern China. GLQ223 waspurified from an aqueous extract of homogenized tubermaterial by using modifications of a previously describedprocedure (1). Purity ofthe material obtained was determinedto be >98% by laser densitometric scanning of Coomassieblue-stained gels (NaDodSO4/PAGE) and by size-exclusionHPLC analysis (data not shown). Authenticity ofthe materialobtained was confirmed by Western blot analysis with rabbitantiserum raised against a reference preparation and byN-terminal amino acid sequence analysis (data not shown).

Cell Lines and Preparation ofPrimary Cell Populations. TheVB cell line (5) was used for acute infectivity assays; cellswere maintained in RPMI-1640 supplemented with 10%(vol/vol) heat inactivated fetal calfserum. Primary peripheralblood-derived monocyte/macrophages were prepared andcultured from either HIV-seronegative healthy donors orfrom donors known to be infected with HIV by usingprocedures described in detail elsewhere (6). The researchprotocol was approved by the Human Subjects Committee,University of California, San Francisco.

Bioassays for Anti-HIV Activity. T-cell assays: Acutelyinfected cells. The VB cell line was used to assess the effectsof GLQ223 on primary HIV infection of T-lymphoblastoidcells. Briefly, cells were inoculated with a titered cryopre-served virus stock [isolate HIV-lDV (6)] at a multiplicity ofinfection of -0.005. Cells were incubated at a density of 1-5 x 107 cells per ml with the inoculum for 60 min at 370C topermit adsorption of viral particles and then washed toremove unbound virus. Cells were then resuspended at 1.0 x105 cells per ml in RPMI-1640 supplemented with 10% fetalcalf serum and cultured in 24-well culture plates (1 ml perwell) with or without GLQ223 added to the desired concen-tration for the duration of culture. Cells were cultured for 4days, when supernatants were harvested for quantitation ofHIV p24 antigen content by using a commercially availablecapture immunoassay (Coulter). In this assay system, at themultiplicity of infection used, virally mediated cytopathiceffects and HIV p24 levels peak at day 4; anti-HIV activitywas thus evaluated at the time of maximum virus production.

Abbreviations: HIV, human immunodeficiency virus; AZT, 3'-azido-3'-deoxythymidine.hTo whom reprint requests should be addressed at: Division ofCellular Immunology, Genelabs Incorporated, 505 PenobscotDrive, Redwood City, CA 94063.

2844

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Proc. Natl. Acad. Sci. USA 86 (1989) 2845

Treated and untreated, infected and uninfected cultures (i.e.,four different types of cultures) were also plated in 96-wellplates for determination of incorporation of [3H]leucine and[3H]thymidine into trichloroacetic acid-precipitable proteinand cellular DNA, respectively, as determined by scintilla-tion counting of harvested cells. As the extent of viralcytopathology by day 4 was too great to allow reliablequantitation of [3H]leucine and [3H]thymidine incorporationor meaningful comparison to treated cultures, cells forincorporation studies were harvested on day 3.For RNA studies, cells were inoculated at a multiplicity of

infection of 0.03 and then cultured in 75-cm2 culture flasks inthe continuous presence of the desired concentration ofGLQ223. Aliquots of these cultures were harvested at 48 hr,the cells were lysed, and RNA was extracted (7) for analysisof viral and cellular RNA by Northern blot (8). Cell-freeculture supernatants were harvested for determination ofHIV p24 content at the time RNA was prepared. [3H]Leucineand [3H]thymidine incorporation were determined for ali-quots of the cultures, plated as above in 96-well plates, pulsedwith [3H]leucine or [3H]thymidine, and harvested at 48 hr.Macrophage assays: Cells chronically infected in vitro

with exogenous virus. Monocyte/macrophages were isolatedfrom peripheral blood of HIV-seronegative healthy donors,infected in vitro with HIV (HIV-lDV isolate; multiplicity ofinfection of 1.0), and cultured as described (6). HIV antigenexpression was assessed by flow cytometric analysis. Cellswere stained with a p24-specific murine monoclonal antibodyafter using an isotype-matched monoclonal antibody ofirrelevant specificity to set the threshold for backgroundfluorescence. The percentage of cells with HIV-specific(above background-fluorescence channel 150) fluorescencewas then determined. Cultures were used for studies de-scribed below only if the percentage of HIV antigen-express-ing cells, as determined by flow cytometric analysis, was.30%. After establishing the extent of infection, cells weretreated with a 3-hr pulsed exposure to GLQ223. Alternatively,cells received continuous exposure to 3'-azido-3'-deoxythy-midine (AZT, 40 p.M; Burroughs Wellcome). Treated cellswere maintained in suspension culture in Teflon jars (PTFE;Savillex, Minnetonka, MN) as described. After 4 days ofculture, HIV antigen expression was assessed in GLQ223-treated cultures and buffer-treated control cultures by flowcytometric analysis (6).Macrophage assays: Cells infected in vivo. Experiments

were also conducted with monocyte/macrophages infectedin vivo with HIV. Freshly drawn EDTA anti-coagulatedwhole blood from HIV-seropositive patients was treated for3 hr at 37°C with 500 ng of GLQ223 per ml or sham-treatedwith buffer. Mononuclear cells were then isolated overFicoll/Hypaque, with extensive washing, and placed inculture in Teflon jars. Five days later, HIV antigen expres-sion was quantitated by flow cytometric analysis (6).

RESULTSGLQ223 was tested for inhibition of HIV replication in avariety of assays using both primary and transformed targetcells and cells representing both lymphoid and monocytoidlineages. The compound was tested in assays of both acuteand chronic HIV infection and in experiments using primarycells from several distinct in vivo-infected donors, eachpresumably infected with a different strain of HIV.

Cultures ofthe highly susceptible T-lymphoblastoid cell lineVB were inoculated with HIV and then treated with GLQ223.Infection and viral replication were monitored in the culturesby following the appearance of characteristic HIV-inducedcytopathic changes, by measuring HIV p24 in culture super-natants, and by assessing production of viral antigens byimmunofluorescence analysis. As shown in Table 1, in mul-tiple experiments, GLQ223 treatment reproducibly resulted in

concentration-dependent inhibition ofHIV replication as mea-sured in this acute primary infection assay system. Inhibitionof HIV replication was observed by measurement of super-natant p24 content (Table 1) and by decreased viral cytopathiceffects (data not shown). The majority ofHIV p24 productionwas abolished at relatively low concentrations ofGLQ223 (16-63 ng/ml), with essentially complete inhibition seen at higherconcentrations (1-2 ,ug/ml). Greater than 75% inhibition wasobserved at concentrations that produced little inhibition(<20%) of total cellular protein synthesis or cellular DNAsynthesis in infected cells or parallel cultures ofuninfected VBcells, but inhibition of these parameters was seen at higherconcentrations with 3 days ofcontinuous exposure to GLQ223(Tables 1 and 2).A similar acute infection experiment was used to assess

effects ofGLQ223 on HIV replication at the RNA level. Cellswere harvested for RNA extraction at day 2 of culture toprevent the extensive RNA degradation observed at latertime points in untreated cultures, presumably as a conse-quence of advanced viral cytopathic effects and cell lysis.Concentration-dependent inhibition of HIV replication wasagain observed, as measured by supernatant HIV p24 levels(Fig. 1) and decreased virally induced cytopathology (datanot shown). Essentially complete inhibition of HIV p24production was observed at concentrations of GLQ223 thatdid not measurably inhibit total cellular protein synthesis orcellularDNA synthesis in infected cells or parallel cultures ofuninfected VB cells in this 2-day assay (Fig. 1C). GLQ223treatment also resulted in a selective proportional decrease inthe amount of viral RNA present in treated cells relative tototal RNA extracted from the cells (Fig. 1 A and B). RNAfrom infected untreated cells probed with a 32P-labelednick-translated full-length HIV DNA probe derived frompHXB-2 (9) showed a strong signal with characteristicallysized hybridizing bands corresponding to full-length (=9.5kilobases) and spliced forms [4.3 and =2.0 kilobases (10)] ofHIV RNA (Fig. LA, lane 1), with the hybridizing species ofless than 2 kiobases probably representing degraded viralmessage. No hybridization of the HIV probe to RNA fromuninfected cells was seen, regardless of GLQ223 treatment(Fig. LA, lanes 3 and 4). GLQ223 treatment resulted in adecrease in the amount of viral RNA detectable by hybrid-ization (Fig. LA, compare lanes 1 and 2), despite equivalentamounts of total RNA visible in each lane in the ethidiumbromide-stained gel prior to transfer (data not shown). Incontrast, GLQ223 treatment did not affect levels of messagedetectable by hybridization with a full-length y-actin cDNAprobe from pHF A-1 (11) in either HIV-infected treated cells(Fig. 1B, lanes 1 and 2) or uninfected treated cells (Fig. 1B,lanes 3 and 4).

Table 1. Inhibition of viral replication by GLQ223 inT-lymphoblastoid cells: 96-hr continuous exposure

GLQ223 treat-ment, ,ug/ml p24, ng/ml x 10-2 p24, % control

0 18.5 ± 1.4 100 ± 80.016 4.9 ± 1.6 27 ± 90.031 3.7 ± 1.4 20 + 80.063 2.3 ± 1.2 13 ± 60.126 1.4 ± 0.7 6 ± 40.251 0.7 ± 1.4 4 ± 20.502 0.3 ± 1.2 2 _ 11.005 <0.3 02.010 <0.3 0

HIV p24 production is shown relative to untreated infected controlcultures as a function of GLQ223 concentration for acutely infectedcells of the T-cell line VB. Cells were continuously exposed toGLQ223 for 96 hr. p24 values are means ± SD for eight experimentsat each concentration.

Medical Sciences: McGrath et al.

2846 Medical Sciences: McGrath et al.

Table 2. Effect of GLQ223 on [3H]thymidine and [3H]leucineincorporation in uninfected and acutely HIV-infected VB cultures:72-hr continuous exposureGLQ223 treatment,

/Lg/ml

00.0120.0250.0490.0980.1%0.3930.7851.5703.140

00.0120.0250.0490.0980.1%0.3930.7851.5703.140

[3H]Thymidine,% control

Uninfected cells100 ± 487 ± 391 ± 386 ± 484 ± 384 ± 280 ± 573 ± 260 ± 046 ± 2

HIV-infected cells78 ± 590 ± 389 ± 391 ± 489 ± 288 ± 387 ± 375 ± 469 ± 354 ± 2

[3H]Leucine,% control

100 ± 985 ± 880± 883 ± 283 ± 490 ± 1371 ± 461 ± 548 ± 433 ± 3

89 ± 6104 ± 1098 ± 799± 598 ± 593 ± 995 ± 562 ± 450 ± 345 ± 4

Cells (2 x 104 per well) were seeded in 96-well plates and thenpulsed either with 1 gCi (1 Ci = 37 GBq) of (3H]leucine or[3H]thymidine 8 hr prior to harvesting cells at 72 hr of culture.Labeled precursor incorporated into trichloroacetic acid-pre-cipitable cell-associated macromolecules was determined by scintil-lation counting. Values shown are the means SD for six experi-ments. Results are normalized to counts obtained for untreateduninfected control cultures ([3H]thymidine control value = 136,500

5007; [3H]leucine control value = 7897 691).

GLQ223 treatment of monocyte/macrophages chronicallyinfected with HIV in vitro also inhibited HIV replication, asassessed by flow cytometric analysis 4 days after a shortexposure of infected cells to GLQ223. Cells were infectedwith HIV-lDv and then cultured for 20 days prior to either apulsed 3-hr exposure to GLQ223 or initiation of continuousexposure to AZT (40 AM). After a subsequent 4 days ofculture, flow cytometric analysis was performed. A repre-sentative experiment is shown in Fig. 2, where untreatedcultures (Fig. 2A) showed 34% of the cells displaying HIV-specific fluorescence (solid lines) above the backgroundfluorescence threshold set with irrelevant control antibody(dashed lines; background threshold set at fluorescencechannel 150). No HIV-specific fluorescence was detectablein GLQ223-treated cells (Fig. 2B). In contrast to the completeabrogation of HIV antigen expression seen in GLQ223-treated cultures, AZT-treated cultures (Fig. 2C) showed 36%of cells displaying HIV-specific fluorescence. No change inviability was seen in GLQ223-treated cultures at the time ofanalysis, 4 days after treatment. However, 2 weeks followingtreatment, an =40% loss of cellular viability was noted byflow cytometry in GLQ223-treated, in vitro-infected cultureswhen HIV antigen expression remained undetectable (datanot shown). Identically treated uninfected cultures studied inparallel showed no change in cellular viability.

Additional studies were conducted using monocyte/mac-rophages isolated from individuals infected with HIV in vivo.In an ex vivo experimental system designed to simulate asclosely as possible the potential in vivo administration ofGLQ223, freshly drawn EDTA anticoagulated whole bloodfrom HIV-infected patients was treated with GLQ223 (500ng/ml) or sham-treated with buffer for 3 hr at 370C. Afterisolation of mononuclear cells by density centrifugation with

Akb9.5 - 07.54.4 - to

1.4 -

0.24- 1

1 2 3 4

C

120

100E

0-

I

8060

40

20

Bkb9.5 -7.5 -4.4 -

2.4 -rn

1.4-

0.24-

1 2 3 4

Untreated: 106.2 ng/ml/

10v.01 .1 1

GLQ223 (ig/ml)

FIG. 1. GLQ223 treatment selectively decreases relative levelsof HIV RNA: 48-hr continuous exposure. (A) HIV probe. Nohybridization of the HIV probe to RNA from uninfected cells thatwere untreated (lane 3) or treated with GLQ223 at 3.14 Ag/ml (lane4) was seen, whereas a strong signal with characteristically sizedbands corresponding to full length and spliced forms of HIV RNAwas seen in untreated infected cells (lane 1). GLQ223 treatment at3.14 Ag/ml (lane 2) resulted in a decrease in the amount of viral RNAdetectable by hybridization. (B) y-Actin probe. GLQ223 treatmentdid not affect levels of y-actin message detectable by hybridizationin untreated infected cells (lane 1), infected cells treated withGLQ223 at 3.14 ,ug/ml (lane 2), untreated uninfected cells (lane 3),and uninfected cells treated with GLQ223 at 3.14 j.g/ml (lane 4). Thesimilarity in size of the hybridizing bands for y-actin and spliced HIVmessage is coincidental and does not represent cross-hybridization;despite hybridization of the y-actin probe to RNA from uninfectedcells (lanes 3 and 4), there is no hybridization ofthe HIV probe to thesame RNA preparations (lanes 3 and 4 in A). (C) Aliquots of thecultures used for RNA studies were also tested for supernatant p24content at 48 hr. p24 values are single determinations at eachconcentration; variability of the assay employed is <10% for repli-

cate samples. [3H]Leucine incorporation values were as follows:uninfected untreated cells, 19,673 ± 3137; uninfected cells (3.14,ug/ml GLQ223), 20,285 ± 2019; infected untreated cells, 15,3251116; infected cells (3.14 /hg/ml GLQ223), 15,220 ± 482.[3H]Thymidine incorporation values were as follows: uninfecteduntreated cells, 54,840 ± 5763; uninfected cells (3.14 ,ug/mlGLQ223), 60,600 ± 2589; infected untreated cells 32,889 ± 2995;infected cells (3.14 ,ug/ml GLQ223), 55,021 ± 2905. (Mean valuesSD in cpm; n = 4.)

extensive washing, cells were cultured with no furtherexposure to GLQ223. HIV antigen expression was quanti-tated on day 5 of culture by flow cytometric analysis. Asshown in Table 3, this single pulsed treatment with GLQ223completely abrogated HIV antigen expression by monocyte/inacrophages from five of eight evaluable patients (quantifi-able HIV-specific immunofluorescence detectable in paralleluntreated cultures) and substantially inhibited expression inthe other three cultures. The treatment did not affect cellularviability at day 5 (data not shown).

DISCUSSIONHIV is the etiologic agent for AIDS and a spectrum of relateddisorders. Although a variety of potentially vulnerable stepsin the viral life cycle could theoretically serve as targets for

I...

Proc. Natl. Acad. Sci. USA 86 (1989)

Proc. Natl. Acad. Sci. USA 86 (1989) 2847

200 rA

100

CIOT-x0

a).0

Ez

0

200

100

1200 400 600 800

a

1 200

200 C

100

400 600 800

1 200 400 600 800Fluorescence

FIG. 2. Selective inhibition of HIV replication in chronicallyinfected monocyte/macrophage cultures. Flow cytometric analysis(linear fluorescence scale) for HIV p24 expression in cytoplasm ofinfected cells is shown for GLQ223 and AZT-treated monocyte/macrophage cultures chronically infected, in vitro, with HIV. Cellswere either untreated (A), pulse treated with GLQ223 (500 ng/ml for3 hr followed by washing, without further GLQ223 exposure) (B), ortreated continuously with 40 AM AZT (C).

therapeutic intervention, the in vivo pathobiology of HIVinfection will probably determine the ultimate efficacy ofpotential therapies. CD4-expressing lymphoid cells are sus-ceptible to HIV infection, and depletion of CD4-bearing Tlymphocytes is both the hallmark of HIV-associated diseaseand the apparent cause of many of the clinical consequencesof HIV infection. However, cells of the mononuclear phago-Table 3. Effect of pulsed GLQ223 treatment of whole blood onHIV expression in subsequently cultured monocyte/macrophagesfrom infected donors

Integrated total% p24-expressing HIV

cells* fluorescencetPatient Control Treated Control Treated % inhibition

1 6.3 0 3502 0 1002 2.1 0.1 1158 71 953 5.7 0 3156 0 1004 3.1 0 1813 0 1005 3.0 0 1797 0 1006 5.3 0.9 3332 613 827 7.1 0 4371 0 1008 4.4 0.8 3158 584 83The results shown are for representative experiments involving

eight different donors. Under the culture conditions employed (noexposure of cells to mitogens or other activation stimuli), p24-expressing cells were observed only among cells in the macrophage,but not lymphocyte, gating region.*Percent p24-expressing cells indicates the percentage of cells incultures expressing p24-specific (above background-fluorescencechannel 150) fluorescence on day 5 of culture.

tIntegrated total HIV fluorescence refers to the integrated area ofthefluorescence histogram obtained with the p24-specific antibody,above fluorescence channel 150, and reflects both the percentage ofp24-expressing cells and the intensity offluorescence seen for thesecells.

cyte lineage actually appear to be a major in vivo reservoir forHIV (6, 12), as is the case for most other retrolentiviruses (13,14). Therefore, agents that are not active in mononuclearphagocytes are unlikely to exert significant clinical effects, nomatter how promising the activity observed in vitro inT-cell-based assay systems. Furthermore, monocyte/macro-phages appear to be relatively more resistant to the cyto-pathic effects of HIV infection than T cells, more readilyallowing establishment of chronic infection with potential forpersistence of an in vivo viral reservoir. The importance ofchronic infection of cells, especially cells of the mononuclearphagocytic lineage, in the in vivo pathobiology of HIVinfection may represent an intrinsic limitation to the ultimateclinical efficacy of nucleoside analogs such as AZT. As AZTappears to act by inhibiting reverse transcription, an earlyevent in the HIV life cycle, it would be expected to have littleor no effect on chronically infected cells presumably produc-ing viral proteins from integrated proviral DNA. In vitrostudies show this to be the case in both macrophages and Tcells (Fig. 2; M.S.M., unpublished observations; ref. 15).GLQ223 inhibits HIV replication in both T cells and

monocyte/macrophages and shows activity in in vitro assaysof both acute and chronic infection. Inhibition of HIV repli-cation is manifest as a selective inhibition of levels of viralantigen and viral RNA relative to host cell protein synthesisand total cellular RNA levels in treated cells. GLQ223 showsanti-HIV activity in in vitro assays against both cells infectedin vitro with the virus and against cells infected with HIV invivo. Activity of GLQ223 against in vivo-infected cells fromdifferent patients (Table 3) demonstrates the in vitro activity ofthe compound against a variety ofpresumptively distinct virusstrains. Typically, monocytes freshly isolated from HIV-infected patients do not express detectable levels of HIVantigens. Brief in vitro cultivation of these cells allows HIVexpression in a subpopulation of monocyte-derived cells fromsuch donors; expressed viral antigens are readily detectable byflow cytometric analysis (M.S.M., unpublished observations).A pulsed exposure of whole peripheral blood from HIV-infected patients to GLQ223 resulted in essentially completesuppression of viral antigen expression by subsequently cul-tured cells from a majority of donors. No viral antigens weredetectable 5 days after a single brief exposure to the com-pound. In preliminary studies, a single 3-hr pulsed exposure toGLQ223 inhibits expression of HIV antigens in subsequentlycultured treated infected cells for up to 4 weeks. As a decreasein cellular viability is noted 2-3 weeks after treatment, whenviral antigen expression remains undetectable, but not atearlier time points (4-5 days after treatment) or in identicallytreated uninfected cultures, GLQ223 appears to exert a selec-tive inhibitory activity that initially suppresses viral replicationand ultimately selectively kills HIV-infected monocyte/macrophages. Despite the fact that freshly isolated peripheralblood monocytes from HIV-infected subjects do not typicallyexpress viral antigens, treatment of freshly drawn whole bloodfrom such subjects with GLQ223 prevented expression ofHIVantigens by subsequently isolated and cultured cells, suggest-ing that the compound may be capable of acting on at leastsome infected cell populations that do not express detectableviral antigens. Based on its potent anti-HIV activity in both Tcells and monocyte/macrophages, as well as its activity inassays of both acute and chronic infection, GLQ223 is aninteresting agent for evaluation for anti-HIV activity in vivo.Animal testing will permit assessment of any toxicities asso-ciated with the compound, including any toxic effects relatedto immunogenicity. Although essentially complete inhibitionof viral antigen expression was observed, in vitro, at concen-trations ofGLQ223 that did not appear to inhibit incorporationof [3H]leucine or [3H]thymidine, higher concentrations of thecompound did inhibit [3H]leucine and [3H]thymidine incorpo-ration, especially when present continuously in culture for

Medical Sciences: McGrath et al.

:.1,'I--

2848 Medical Sciences: McGrath et al.

several days. However, relatively brief, pulsed exposure tothe compound appears to be sufficient for anti-HIV effects andseems to minimize any inhibitory effects on noninfected cells,in vitro, a finding of potential relevance to minimizing any

observed in vivo toxicities.The mechanism of action whereby GLQ223 inhibits HIV

replication is currently unknown. Although the compoundbelongs to the family of single-chain ribosome-inactivatingproteins, it has not been established whether the anti-HIVactivity is directly related to this property. Two generalclasses of mechanisms can be envisioned. In the first, theselective antiviral activity of the compound may be due toselective binding or uptake by virally infected cells (16, 17).Once inside the infected cells, the compound may exertnonspecific effects through catalytic inactivation of ribo-somes (18, 19). In the second class of mechanism, selectivebinding or uptake by infected cells may or may not occur, butthe anti-HIV activity of the compound may be attributed todifferential effects on viral as opposed to host cell nucleicacid or protein synthesis, processing, or stability. Treatmentof acutely infected T-lymphoblastoid cells with GLQ223resulted in decreased levels of viral proteins at concentra-tions ofGLQ223 that did not affect cellular protein synthesis(Fig. 1C), at a time when immunofluorescence analysisshowed -75% of untreated infected cells expressing viralantigens (48 hr after infection; immunofluorescence data notshown). As selective binding or uptake by infected cellsfollowed by nonspecific ribosomal inactivation might beexpected to result in a measurable inhibition of total cellularprotein synthesis under these conditions, this observationargues against a mechanism based on nonspecific inhibitionof protein synthesis through ribosomal inactivation. Treatedcells showed a selective proportional decrease in the amountof viral RNA relative to total cellular RNA at 48 hr of culture(when -75% of untreated cells were productively infected;data not shown), with no effect on levels of RNA encodingthe cellular gene for -actin, suggesting a possible selectiveeffect ofGLQ223 on viral nucleic acid synthesis, processing,or stability. Additional studies are needed to clarify themechanism that underlies the observed selective anti-HIVactivity of GLQ223 and other related compounds that showsimilar activity in vitro.

We thank Dr. Leonard A. Herzenberg (Stanford University) forthoughtful review of the manuscript, Drs. Gregory Reyes andMichael Piatak (Genelabs) for helpful discussions, Dr. Holly Abrams(Genelabs) for providing virus stocks used in some studies, and theGenelabs Process Development Group for providing purifiedGLQ223.

1. Yeung, H. W., Wong, D. M., Ng, T. B. & Li, W. W. (1985)Int. J. Pept. Protein Res. 27, 325-333.

2. Zhang, X. & Wang, J. (1986) Nature (London) 321, 477-478.3. Cheng, K. F. (1982) Obstet. Gynecol. 59, 494-498.4. Chan, Y., Tong, M. K. & Lau, M. J. (1982) Shanghai J. Chin.

Med. 3, 30-31 (in Chinese).5. Lifson, J. D., Reyes, G. R., McGrath, M. S., Stein, B. S. &

Engleman, E. G. (1986) Science 232, 1123-1127.6. Crowe, S., Mills, J. & McGrath, M. (1987) AIDS Res. Hum.

Retrov. 3, 135-145.7. Chomczynski, P. & Sacchi, N. (1987) Anal. Biochem. 162,156-

159.8. Maniatis, T., Fritsch, E. F. & Sambrook, J. (1982) Molecular

Cloning:A Laboratory Manual (Cold Spring Harbor Lab., ColdSpring Harbor, NY), pp. 202-203.

9. Fisher, A. G., Collati, E., Ratner, L., Gallo, R. C. & Wong-Staal, F. (1985) Nature (London) 316, 262-265.

10. Muesing, M. A., Smith, D. H., Cabridilla, C. D., Benton,C. V., Lasky, L. A. & Capon, D. J. (1985) Nature (London)313, 450-457.

11. Gunning, P., Ponte, P., Okayama, H., Engel, J., Blau, H. &Kedes, L. (1983) Mol. Cell. Biol. 3, 787-795.

12. Gartner, S., Markovits, P., Markovits, D. M., Kaplan, M. H.,Gallo, R. C. & Popovic, M. (1986) Science 233, 215-219.

13. Haase, A. T. (1986) Nature. 322, 130-136.14. Narayan, O., Kennedy-Stoskopf, S. & Zink, M. C. (1988) Ann.

Neurol. (Suppl.) 23, S95-S100.15. Perno, C.-F., Yarchoan, R., Cooney, D. A., Hartman, N. R.,

Gartner, S., Popovic, M., Hao, Z., Gerrard, T. L., Wilson,Y. A., Johns, D. G. & Broder, S. (1988) J. Exp. Med. 168,1111-1125.

16. Fernandez-Puentes, C. & Carrasco, L. (1980) Cell 20, 769-775.17. Foa-Tomasi, L., Campadelli-Fiume, G., Barbieri, L. & Stirpe,

F. (1982) Arch. Virol. 74, 322-332.18. Endo, Y. & Tsurugi, K. (1987) J. Biol. Chem. 262, 8128-8130.19. Stirpe, F., Bailey, S., Miller, S. P. & Bodley, J. W. (1988)

Nucleic Acids Res. 16, 1349-1357.20. Wang, Y., Qian, R. Q., Gu, Z. W., Jin, S. W., Zhang, L. Q.,

Xia, Z. X., Tian, G. Y. & Ni, C. Z. (1986) Pure Appl. Chem.58, 789-798.

Proc. Natl. Acad. Sci. USA 86 (1989)

Coregulation of NADPH Oxidase Activation and Phosphorylation of a 48-kDProtein(s) by a Cytosolic Factor Defective in Autosomal RecessiveChronic Granulomatous DiseaseSusan E. Caldwell, Charles E. McCall, Cynthia L. Hendricks, Peter A. Leone, David A. Bass, and Unda C. McPhailDepartments ofBiochemistry and Medicine, Bowman Gray School ofMedicine of Wake Forest University,Winston-Salem, North Carolina 27103

Abstract

The mechanisms regulating activation of the respiratory burstenzyme, NADPH oxidase, of human neutrophils (PMN) arenot yet understood, but protein phosphorylation may play arole. We have utilized a defect in a cytosolic factor required forNADPH oxidase activation observed in two patients with theautosomal recessive form of chronic granulomatous disease(CGD) to examine the role of protein phosphorylation in acti-vation of NADPH oxidase in a cell-free system.NADPH oxidase could be activated by SDS in reconstitu-

tion mixtures of cytosolic and membrane subcellular fractionsfrom normal PMN, and SDS also enhanced phosphorylationof at least 16 cytosolic and 14 membrane-associated proteins.However, subcellular fractions from CGD PMN plus SDS ex-pressed little NADPH oxidase activity, and phosphorylationof a 48-kD protein(s) was selectively defective. The membranefraction from CGD cells could be activated for NADPH oxi-dase when mixed with normal cytosol and phosphorylation ofthe 48-kD protein(s) was restored. In contrast, the membranefraction from normal cells expressed almost no NADPH oxi-dase activity when mixed with CGD cytosol, and phosphoryla-tion of the 48-kD protein(s) was again markedly decreased.Protein kinase C (PKC) activity in PMN from the two patientsappeared to be normal, suggesting that a deficiency of PKC isnot the cause of the defective 48-kD protein phosphorylationand that the cytosolic factor is not PKC.

These results demonstrate that the cytosolic factor re-quired for activation of NADPH oxidase also regulates phos-phorylation of a specific protein, or family of proteins, at 48kD. Although the nature of this protein(s) is still unknown, itmay be related to the functional and phosphorylation defectspresent in CGD PMN and to the activation of NADPH oxi-dase in the cell-free system.

Introduction

Human neutrophils (PMN) play a critical role in host defenseagainst microorganisms and in the inflammatory response (1).One ofthe primary mechanisms for effecting these reactions is

Address reprint requests to Dr. McPhail, Department ofBiochemistry,Bowman Gray School of Medicine of Wake Forest University, 300 S.Hawthorne Rd., Winston-Salem, NC 27103.

Published in abstract form in 1987. (Clin. Res. 35:655a.)Received for publication 9 January 1987 and in revised form 5

October 1987.

the generation of toxic oxygen radicals by the PMN. The en-zyme in PMN that triggers oxygen radical production is anactivatable NADPH oxidase, whose mechanism of activationis still unknown. The active enzyme appears to be a multicom-ponent complex, most likely consisting ofa flavoprotein and aunique low potential cytochrome, cytochrome b559 (2).

Until recently, activation of NADPH oxidase could onlybe achieved by stimulation of intact PMN with a wide varietyof agents, including opsonized particles, phorbol esters, che-motactic peptides, and certain detergents (2). One characteris-tic these stimuli share is the ability to trigger activation ofprotein kinases inside the cell resulting in the phosphorylationof endogenous proteins (3-8). Protein kinase C (PKC)' hasbeen ofparticular interest as an intermediate. Direct activatorsof PKC, such as phorbol esters and cell-permeable diacylglyc-erols (9), trigger oxidase activation in the intact cell by anapparent PKC-dependent mechanism (10-13). In addition,stimuli, such as concanavalin A and chemotactic peptides,induce the release of endogenous stores of diacylglycerol frommembrane phospholipids and thus could activate PKC (14).

Further evidence supporting a role for protein phosphory-lation in the activation ofNADPH oxidase is given by recentreports of a phosphorylation defect in the PMN from severalpatients with chronic granulomatous disease (CGD), observedby stimulation of the intact cell (15-17). CGD is an inheriteddisease characterized by increased susceptibility to bacterialinfection, in which the only known functional defect is theinability oftheir phagocytic cells to produce oxygen radicals inresponse to cell stimulation. Two modes of inheritance of thedisease are observed, X-linked recessive and autosomal reces-sive (18, 19). Recently, the gene defect in the X-linked formhas been described (20) and the gene product appears to be asubunit of the cytochrome b559 (21, 22). PMN from most pa-tients with this form of the disease lack the characteristic spec-trum ofthe cytochrome (23). The gene defect in the autosomalrecessive form has yet to be identified. However, the phosphor-ylation defect in CGD has been attributed solely to the auto-somal recessive form (15), although, this has been disputed(16, 24).

Recently, a cell-free system for activation ofNADPH oxi-dase has been developed by ourselves (25) and others (26-28).Activation requires the apparent interaction of cytosolic andmembrane cellular cofactors and can be achieved by the addi-tion of an amphiphile, either arachidonic acid or SDS (29), tothe isolated cell fractions. The identities of the cytosolic andmembrane cofactors are unknown, although at least one nec-

1. Abbreviations used in this paper: CB, cytochalasin B; CGD, chronicgranulomatous disease; DFP, diisopropylfluorophosphate; EP, extract-able particulate fraction; NEP, nonextractable particulate fraction;PKC, protein kinase C; PMA, phorbol myristate acetate.

Cell-free Protein Phosphorylation and Neutrophil Oxidase Activation 1485

J. Clin. Invest.X The American Society for Clinical Investigation, Inc.0021-9738/88/05/1485/12 $2.00Volume 8 1, May 1988, 1485-1496

essary membrane component is cytochrome b559 (30). In addi-tion, a recent abstract reports that the cytosolic factor activityis absent in a CGD patient with the autosomal recessiveform (31).

It is not yet clear if a protein kinase is involved in oxidaseactivation in this system. Arachidonate can activate PKC (32),but inhibition or depletion ofPKC does not block arachidon-ate-mediated activation of NADPH oxidase in the cell-freesystem (30, 33, 34). Contradictory reports as to the require-ment for ATP in the system have appeared (30, 35). Finally, ithas not yet been determined if arachidonate or SDS can in-duce phosphorylation of endogenous protein substrates inconjunction with activation ofNADPH oxidase in the cell-freesystem.

In this report, the role of protein phosphorylation in theactivation ofNADPH oxidase in the cell-free system is exam-ined. Using both normal PMN and PMN lacking cytosolicfactor activity from patients with the autosomal recessive formof CGD, we have compared the ability of SDS to induce oxi-dase activation and phosphorylation of endogenous proteinsin cell-free mixtures of cytosolic and membrane fractions. Re-sults suggest that the cytosolic factor required for NADPHoxidase activation in the cell-free system is also specificallyrequired for the SDS-dependent phosphorylation of a 48-kDprotein, or family of proteins.

Methods

Isolation of cells. PMN (> 95% purity) were purified from humanperipheral blood by dextran sedimentation, Ficoll-Hypaque centrifu-gation, and hypotonic lysis as described (36). Two patients with auto-somal recessive CGD were studied, one of which (S.S.) is a 27-yr-oldmale reported previously (37). The other (S.H.) is a 25-yr-old femalepatient of Dr. Michael Cohen (University ofNorth Carolina School ofMedicine, Chapel Hill, NC) and was diagnosed based on clinical his-tory and neutrophil oxidative functional studies (defective in vitrobactericidal activity and the absence of a luminol-dependent chemilu-minescence response to phorbol myristate acetate, PMA, or opsonizedzymosan).

Subcellularfractionation. Purified PMN were suspended at 5 X 107PMN/ml in Hanks' balanced salt solution containing 4.2 mM sodiumbicarbonate and 10 mM Hepes, pH 7.4 (HBSS). To inhibit proteolysis(38), cells were treated with 2 mM diisopropylfluorophosphate (DFP)for 5 min at 4VC, and then washed with a 15-fold volume of ice-coldHBSS. PMN were resuspended at 1 X 108/ml in a sonication buffercontaining 0.34M (11 %, wt/vol) sucrose, 10mM Hepes, 1 mM EGTA,1 mM NaN3, 130 mM NaCl, 100 mM NaF, and 10 mM sodiumpyrophosphate. In some experiments, in which NADPH oxidase activ-ity only was measured, the DFP treatment was omitted, 0.5 mM PMSFwas included, and NaF and pyrophosphate were omitted. These alter-ations had no effect on the final oxidase activity obtained. l-ml ali-quots of cells were sonicated in 5-s bursts in a melting ice bath with asetting of 1.5 using a sonicator (model W-220; Heat Systems-Ultra-sonics, Inc., Farmingdale, NY) equipped with a stepped microtip. Cellbreakage of - 90% was achieved as monitored by phase microscopy.

Sonicates were centrifuged at 800 g for 10 min at 40C to pelletunbroken cells and nuclei. Supernatants were fractionated by discon-tinuous sucrose density gradient centrifugation. In the first experi-ment, supernatants in the 11% sucrose buffer were layered over 40%(wt/vol) sucrose in a ratio of 2:1 (vol/vol), as described by Gabig et al.(39). In subsequent experiments, a 15% (wt/vol) sucrose interface layerwas included to more easily separate membrane and cytosolic fractions(40), the final ratio being 2: 1:1 (vol/vol) of 11, 15, and 40%, respec-tively. Gradients were centrifuged at 4°C (SW 60.1 rotor; BeckmanInstruments, Inc., Palo Alto, CA) at 150,000 g for 30 min (39). Aftercentrifugation, the top layer of each discontinuous gradient (crudecytosol) was collected and centrifuged again at 150,000 g in a BeckmanType 50 rotor for 60 min at 4°C to pellet any contaminating particu-late material (cleared cytosol). The white band at the 1 1%/40% or the15%/40% interface from each discontinuous gradient and most of the40% sucrose layer was collected (membrane fraction). The discontin-uous gradients also yielded granule fractions as pellets, which wereresuspended in sonication buffer to the original volume layered ontothe gradient. All fractions were either used immediately or storedat -700C.

Gradient fractions were analyzed for purity by measuring the dis-tribution of the following marker enzymes: lactic dehydrogenase(LDH; 41) for cytosol, alkaline phosphatase (42) for plasma mem-brane, vitamin B12-binding protein (43) for specific granules, lysozyme(44) for both specific and azurophil granules, and myeloperoxidase(MPO; 45) for azurophil granules. Table I summarizes the distribution

Table L Subcellular Marker Enzyme Distribution in Fractions Isolated by Discontinuous Sucrose Gradient Fractionation

% Distribution of enzyme*

Alkaline B12-bindingFraction assayedt LDH phosphatase protein Lysozyme MPO

11%/40% gradientCytosolic 95.9 0.9 ND 0 7.1Membrane 2.5 93.8 ND 8.7 3.7Granule 2.0 5.3 ND 91.3 89.2

1 1%/ 15%/40% gradientCytosolic 86.7 1.1 2.4 5.3 2.915% layer 8.3 7.3 0.4 0.1 0.4Membrane 1.4 83.2 12.1 8.2 3.7Granule 0.6 5.6 85.1 86.3 92.7

* Marker enzyme levels were determined in subcellular fractions as described in Methods. Numbers given are the percentage of total activity re-covered in each fraction for one experiment with the 11%/40% gradient and the mean of two experiments with the 11 %/ 1 5%/40% gradient. Inthe 1I%/15%/40% gradient, total recovery for each marker (mean oftwo experiments) was: LDH, 85.7%; alkaline phosphatase, 97.4%; vitaminB12-binding protein, 79.9%; lysozyme, 102.9%; MPO, 68.2%; and protein (46), 87.3%. Distribution of protein in the 11%/40% gradient was: cy-tosolic, 50.4%; membrane, 7.5%; and granule, 42.2%. Protein distribution in the 1I%/l15%/40% gradient was: cytosolic, 43.8%; 15% layer, 5.3%;membrane, 7.8%; and granule, 41.4%. * Normal PMN were sonicated and fractionated by either an 1l%/40% or an 1l%/15%/40% discontin-uous sucrose gradient as described in Methods. Fractions were assayed immediately for LDH and stored at -70'C before the remainder of themarker enzyme assays.

1486 Caldwell, McCall, Hendricks, Leone, Bass, and McPhail

of these markers. In one experiment with the 1 1%/40% gradient, thecleared cytosol fraction contained 95.9% of the recovered LDH, 0.9%of alkaline phosphatase, and a mean of 3.5% of lysozyme and MPO.The membrane fraction contained 93.8% of the recovered alkalinephosphatase, 2.5% of LDH, 8.7% of lysozyme, and 3.7% of MPO. Inthe 1 1%/15%/40% gradient procedure, the cleared cytosol contained86.7% of the recovered LDH, 1.1% of alkaline phosphatase, 2.4% ofB12-binding protein, 5.3% of lysozyme, and 2.0% of MPO. The mem-brane fraction contained 83.2% of the recovered alkaline phosphatase,1.4% of LDH, 12.1% of B,2-binding protein, 8.2% of lysozyme, and3.7% of MPO. The granule fractions from both gradient procedurescontained > 85% of the recovered granule enzyme markers. The dis-continuous sucrose gradient fractionation procedure, therefore, wasconsidered to yield cytosolic and membrane fractions of acceptablepurity for use in the cell-free oxidase and phosphorylation systems.

NADPH oxidase assay in the cell-free system. Assays measured theSOD-inhibitable reduction of cytochrome c, and were performed aspreviously described (25, 36) with slight modifications. The assaymixture consisted of 0.048 mM potassium phosphate (pH 7.0), 0.076mM cytochrome c (type III), 1.0 mM EGTA, 10.0 MM flavin adeninenucleotide (all from Sigma Chemical Co., St. Louis, MO), 7.5 mMMgCl2, and either a 2:1 or 4:1 (vol/vol) ratio of cytosol to membranefraction. Cell equivalence in the assay was 1.5 X 107/ml of cytosol andeither 0.75 or 1.5 X 10'/ml of membrane fraction. The final assayvolume was 0.105 ml. Activation ofNADPH oxidase was achieved byaddition of 0 (H20 control) to 175 MM SDS (Bio-Rad Laboratories,Richmond, CA) and incubation at room temperature for 4 min. Theconcentration of SDS yielding optimal activation ofNADPH oxidasein reconstitution mixtures of control fractions varied from 120 to 175MM and was determined in each experiment. Equal aliquots of assaymixture were placed into two cuvettes: the reference cuvette, contain-ing 48 Mg/ml SOD (Diagnostic Data, Inc., Mountain View, CA), andthe sample cuvette, containing H20. NADPH oxidase activity wasmeasured by addition of 0.19 mM NADPH to both cuvettes, and thechange in absorbance at 550 nm recorded by a dual beam spectropho-tometer (Cary 2390; Varian Instruments, Sunnyvale, CA). Initial rateswere used for calculations and activity was expressed as nmol/min/mgmembrane protein (46) using an extinction coefficient of 21 mM-'cm-' for cytochrome c (47). Assays were performed on both fresh andfrozen cell fractions with no differences in results obtained.

Assayfor cytochrome b559. The method used was a modification ofpublished procedures (48, 49). PMN from normal donors and theautosomal recessive CGD patients were sonicated and fractionated bydiscontinuous sucrose density gradient centrifugation with sodium py-rophosphate and NaF omitted from, and PMSF included in, the soni-cation buffer. The membrane and granule fractions were analyzed forcytochrome b559 content by difference spectroscopy after dithionitetitration of an aerobic sample. Fractions were diluted 1:1 with 2Xbuffer, yielding final concentrations of 0.25 M potassium phosphate(pH 7.5), 0.12 mM sodium deoxycholate, 5.0 mM NaN3, and 0.5 mMH202. The reduced minus oxidized spectrum of each diluted fractionwas determined on a Cary 2390 spectrophotometer after completereduction of the sample cuvette by 6-10 mM anaerobically prepareddithionite. Amounts of cytochrome b559 were expressed as pmol/108cell equivalents using an extinction coefficient of 106 mM-' cm-' atthe Soret peak (49).

Cell-free protein phosphorylation. Cell-free mixtures for phosphor-ylation studies consisted of 1.4 mM MgCl2, 6.0 X 10' cell equivalents/ml each of cytosolic fraction and membrane fraction, and 51 MCi/ml[.y-32P]ATP (12.88 Ci/mmol; New England Nuclear, Boston, MA).SDS at a final concentration of460 MM or an equal volume ofdistilledwater was added and the samples were incubated 5 min at room tem-perature. The higher concentration of SDS was used to keep the ratioof SDS to cellular material similar to that used in the procedure foractivation of NADPH oxidase. Reactions were stopped by adding atwo-fold volume of ice-cold sonication buffer, which also reduced theviscosity of the mixtures for subsequent centrifugation. Samples werethen centrifuged at 150,000 g for 12 h at 40C in a Beckman Type 50

rotor. Supernatants (cytosolic fractions) were reserved and pellets(membrane fractions) were resuspended by sonication in a small vol-ume (< 1 ml) of sonication buffer. Part of each sample (200-300 Ml)was solubilized in electrophoresis sample buffer (50), boiled for 2 minin a water bath, and stored for no more than 3 d at -70'C beforeanalysis by SDS-polyacrylamide gel electrophoresis, as describedbelow. The remainder of each sample was frozen at -70'C withoutsolubilization; these nondenatured fractions were assayed for proteinby Peterson's modification (51) of the method of Lowry et al. (46).

Endogenous protein phosphorylation in intact PMN. Cells at 1X 108/ml from a normal donor and a CGD patient (S.S.) were sus-pended in buffer (4) containing 6 mM Hepes/Tris (pH 7.4), 0.15 MNaCl, 10mM glucose, 5 mM KCI, 1 mM MgCI2, 0.25 mM CaCI2, and32p; (1 mCi/1-2 X 0I cells; New England Nuclear). PMN were incu-bated at 370C for 30 min, washed, and resuspended at 5 X 107/ml inthe Hepes/Tris buffer without 32p;. Cells were prewarmed for 5 min at370C in the presence of 10-1 M cytochalasin B (CB), then treated witheither dimethylsulfoxide (DMSO) or 100 ng/ml PMA for 30 s. Reac-tions were terminated by addition of a 15-fold excess volume of ice-cold incubation buffer containing 1.0M NaFand 10mM EDTA. Cellswere washed and then treated with DFP and sonicated as describedabove. Sonicates were centrifuged at 500 g for 10 min at 4°C, andpostnuclear supernatants were centrifuged at 100,000 g for 1 h at 4°Cin a fixed angle rotor. Supernatants (cytosolic fractions) were removedand stored at -70°C: one aliquot was solubilized in electrophoresissample buffer as described above and the remainder analyzed for pro-tein content (51).

SDS-polyacrylamide gel electrophoresis and autoradiography. Re-duced, denatured samples (30-100 Mg protein/lane) were electropho-resed on 8-15% gradient polyacrylamide slab gels (1.5 X 140 X 320mm) using the discontinuous buffer system of Laemmli (50). Exceptwhere noted, within each gel, equal amounts of protein were loadedonto gel lanes. After electrophoresis, gels were silver stained by themethod ofWray et al. (52), photographed, and then dried between twosheets of cellophane using a slab gel dryer (model 443; Bio-Rad). Eachdried gel was exposed, typically for 1-3 d, to preflashed Kodak X-Omat x-ray film with a DuPont Cronex Lightning Plus intensifyingscreen at -70°C (53) to generate autoradiograms of the radioactivephosphoproteins. Autoradiograms were scanned with a laser densi-tometer (UltroScan XL) and scans were analyzed by the softwarepackage (GelScan XL; LKB-Produkter AB, Bromma, Sweden).

PKC assay. PKC activity and distribution in normal PMN andPMN from the patients with CGD were determined in crude subcellu-lar fractions as previously described (10). Briefly, PMN at 5 X 107/mlin HBSS were exposed, in the presence of I0-'M CB at 370C, to eithersolvent alone (DMSO), 10-6 M fMLP, or 100 ng/ml PMA for 30 s andreactions were terminated by addition of a 10-fold excess of ice-coldHBSS. After centrifugation, PMN were resuspended in an extractionbuffer containing 50 mM Tris-HCI (pH 7.5), 50 mM 2-mercaptoeth-anol, 1 mM PMSF, and 2 mM EGTA. PMN were sonicated as de-scribed above, centrifuged at 500 g, and postnuclear supernatants werecentrifuged at 150,000 g for 90 min. Supernatants (cytosolic fractions)were collected; pellets were resuspended by sonication in extractionbuffer containing 1.0% Triton X-100 (Sigma Chemical Co.) and cen-

trifuged at 150,000 g for 90 min. Supernatants, the extractable particu-late fractions (EPs), were saved and pellets, the non-extractable partic-ulate fractions (NEPs), were resuspended in extraction buffer contain-ing 0.1% Triton X-100 and saved. Fractions were stored at 0°C andassayed within 2 d of preparation.

Protein kinase assay mixtures (10) contained, in a final volume of0.25 ml: 35 mM Tris-HCl (pH 7.5), 10 mM MgCl2, 160 Mg/ml histonetype HI (type III, Sigma), 50 MM ATP with 2 X 106 cpm [y-32P]ATP,0.4 mM EGTA, 0.2 mM PMSF, 10 mM 2-mercaptoethanol, 0.01%Triton X-100, and 50 Ml diluted subcellular fraction. Cytosolic frac-tions were diluted 10-fold, EP fractions 20-fold, and NEP fractions2-fold. Assays were performed in the presence or absence ofactivators(0.6 mM added CaC12, 20 Mg/ml phosphatidylserine, and 2 ,ug/mldiolein) at 30°C for either 5 min (cytosolic fractions), 60 min (EP

Cell-free Protein Phosphorylation and Neutrophil Oxidase Activation 1487

fractions), or 30 min (NEP fractions). Reactions were stopped with20% (wt/vol) TCA, precipitates collected and counted, and activity wasexpressed as pmol 32p incorporated/min per 107 cell equivalents. Pre-vious studies in normal PMN (10, 54) had demonstrated that proteinkinase activity in cytosolic and EP fractions was completely dependenton the presence of phosphatidylserine, with no activity above back-ground in the presence of calcium alone. Activity in NEP fractionsshowed no dependence on added Ca2l and lipids (10).

Results

Activation ofNADPH oxidase in a cell-free system from nor-mal PMN and PMNfrom patients with autosomal recessiveCGD. Activation ofNADPH oxidase in a cell-free system wasmeasured using cell fractions from normal PMN and PMNfrom two patients with the autosomal recessive form ofCGD.As shown in Table II, substantial activation ofNADPH oxidaseby SDS occurred using reconstitution mixtures of normalmembrane and cytosolic fractions. In contrast, mixtures ofCGD membrane and cytosolic fractions had very low levels ofNADPH oxidase activity in the presence of SDS. If correctedfor activity observed in the absence of SDS in reconstitutionmixtures of normal cell fractions (2.5±0.7 nmol O/min permg, mean±SEM, n = 6), oxidase activity induced in CGDreconstituted fractions was only 2.1% of the normal control.This defect in the ability to activate NADPH oxidase appearedto reside in the cytosolic fractions from the patients with CGD,since (a) reconstitution mixtures ofCGD cytosol with normalmembrane supported very little activation of the oxidase bySDS, and (b) the CGD membrane fraction in combinationwith normal cytosol did allow normal levels of NADPH oxi-dase activation to occur. The membrane fractions of the twopatients showed equal ability to support NADPH oxidase ac-tivation in the presence of normal cytosol (S.H., 84% of activ-ity seen with paired normal membrane, n = 2; S.S., 106% of

Table II. Deficient Cytosolic Factorfor Activation ofNADPHOxidase in Two Patients with Autosomal Recessive CGD

Reconstitution mixture* NADPH oxidase activity$

nmol 0j/min/mgmembrane protein

Normal cytosol + normal membrane 159.0±52.7 (6)CGD cytosol + CGD membrane 5.8±0.4 (3)Normal cytosol + CGD membrane 153.4±62.4 (6)CGD cytosol + normal membrane 8.2±0.8 (4)

* Cytosolic and membrane fractions were isolated from two patients(S.S. and S.H.) with the autosomal recessive form of chronic granulo-matous disease and from six normal donors as described in Methods.One fractionation with CGD (S.S) and one with a normal donor uti-lized the 1 1%/40% (wt/vol) discontinuous sucrose gradient. The re-maining fractionations (one with S.S., one with S.H., and five nor-mal donors) utilized the 1I%/15%/40% (wt/vol) gradient separation.* NADPH oxidase activation was achieved by an optimal concentra-tion of SDS (120-175 MM) and activity was measured as described inMethods. Data are given as the mean±SEM with the number of ex-periments performed, each done in triplicate on a separate day, inparentheses. At n = 3, two experiments were with S.S. fractions andone was with S.H. fractions. In all other instances involving CGDfractions, two experiments were performed with S.H. fractions andthe remainder were with S.S. fractions.

normal, n = 4). In one experiment (not shown), an excess ofCGD cytosol (5.6 X 107 cell equivalents/ml versus the usual1.5 X 107/ml) was mixed with normal membrane fraction totest the possibility of a quantitative defect. The markedly de-creased level of activation of NADPH oxidase was not en-hanced, suggesting that the cytosolic factor is not present indecreased amounts, at least to the level of sensitivity of ourdetection system.

To further exclude an oxidase defect in the CGD mem-brane, membrane and granule fractions from both patientswere examined for their content of cytochrome b559. No cy-tochrome b was detected in cytosolic fractions from normalPMN. The level ofcytochrome b in normal PMN was 495±80pmol/ 108 cells (mean±SD, n = 7) and the cytochrome wasdistributed such that 32% was present in the membrane frac-tion and 68% in the granule fraction. Both patients had normal(S.H.) or slightly elevated (S.S.) levels of cytochrome b (S.H.,492 pmol/108 cells; S.S., 757±3 pmol/108 cells, mean±SD,n = 2) and a similar distribution as a paired normal control.Taken together, these results indicate that the PMN ofthe twopatients with autosomal recessive CGD do not have a defect incytochrome b559 and, instead, are defective in the cytosolicfactor required for the activation of NADPH oxidase in thecell-free system.

Protein phosphorylation in the cell-free system forNADPHoxidase activation. The cell-free system for activation ofNADPH oxidase was examined for protein phosphorylationchanges that might correlate with oxidase activation. Recon-stitution mixtures of membrane and cytosolic fractions iso-lated from normal PMN and the PMN ofthe two patients withautosomal recessive CGD were incubated with or without SDSin the presence of [y-32P]ATP and then subjected to ultracen-trifugation to allow separate analysis ofphosphoproteins in thetwo fractions. Representative results obtained with reseparatedcytosolic fractions are shown in Fig. 1 and with reseparatedmembrane fractions in Fig. 2. The left panel of each figureshows the silver-stained protein pattern ofthe gel and the rightpanel shows the autoradiogram obtained from the same gel.No apparent differences in the pattern of silver-stained pro-teins was observed, comparing fractions obtained from thenormal donor with those from the CGD patient.

As many as 24 phosphoproteins were detected in cytosolicfractions following reconstitution of normal fractions in thepresence of SDS (Fig. 1 B, lane 1), of which 16 were phos-phorylated more intensely and 3 were dephosphorylated bythe SDS treatment (compare to lane 2). An increase in phos-phorylation of a 48,000-D protein(s) (arrow) was observed incytosolic fractions from the normal reconstitution mixture bySDS treatment (lanes I and 2), but not in cytosolic fractionsfrom the CGD reconstitution mixture (lanes 3 and 4). Thiswas the only obvious phosphorylation defect observed in thecytosolic fraction from the CGD cell-free reconstitution.Crossover mixing of normal cytosol with CGD membranefraction (lanes 5 and 6) restored the SDS-dependent phosphor-ylation of the 48-kD protein(s), suggesting that any CGDmembrane components participating in the phosphorylationreaction are functional. In contrast, when the CGD cytosolicfraction was mixed with normal membrane fraction (lanes 7and 8), SDS induced no increase in background phosphoryla-tion in the 48-50-kD region of the gel. These results indicatethat a cytosolic factor necessary for phosphorylation of a48-kD phosphoprotein(s) is either defective or absent in the

1488 Caldwell, McCall, Hendricks, Leone, Bass, and McPhail

A. Cytosolic Fractions-Silver Stain B. Cytosolic Fractions-Autoradiogram

-200

4

I U

9Al

*i:

I

_.

- ~~~~~~~~~~~~~~~~~~~~~~~~.....i. I

I1U .I U

- 200-116

-92.5

-66

-45

Ii i ~~~~~~~~j~31* in WYb-21,

1*tj **511

__ _ _ ~~~~~-1

mm":t*

.5

\lii \±n \t J \Irn.W,CtrI.-C CGD---C 0t - CGD-CCtrI.-M CGD M CGD M (trl.-N1

Figure 1. Phosphorylation of normal and CGD PMN cytosolic pro-teins in the cell-free system. PMN from a normal donor and a CGDpatient (S.S.) were fractionated; cytosolic and membrane fractionswere mixed with 50 MCi [y-32P]ATP in the presence of 460 gM SDS(+) or of H20 as solvent control (-); and fractions were reseparatedas described in Methods. Cytosolic fractions (31 ug protein) were an-alyzed by SDS-polyacrylamide gel electrophoresis, silver staining,

CGD cytosolic fraction, or that the 48-kD phosphoprotein(s) isnot present in the cytosblic fraction from the CGD PMN.

Analysis of membrane fractions, reisolated after reconsti-tution ofnormal membrane and cytosolic fractions in the pres-ence of SDS, by SDS-polyacrylamide gel electrophoresis andautoradiography showed that - 32 phosphoproteins werepresent (Fig. 2 B, lane 1). Of these phosphoproteins, 14 had

94aIa!

S4

- 116

- 92.5

4.t i i -66

sii i^L ^ ^ -~45

-'Wi -~~~~~~~21.5

* , <IP -4

v\ C ,Qcz2\+/ \±z1.Ctrl.-C CGD-C Ctrl.-C CGD-CCtrl.-M CGD-M CGD-M Ctrl.-M

and indirect autoradiography. Ctrl.-C/Ctrl.-M, control cytosol + con-trol membrane; CGD-C/CGD-M, CGD cytosol + CGD membrane;Ctrl.-C/CGD-M, control cytosol + CGD membrane; COD-C/Ctrl.-M, CGD cytosol + control membrane. Arrow marks the position ofthe 48-kd phosphoprotein(s). Molecular weight markers are in kilo-daltons. (A) Silver-stained polyacrylamide gel. (B) Corresponding au-toradiogram.

increased phosphorylation and three had decreased phosphor-ylation compared with the same fraction isolated from a re-constitution mixture incubated in the absence of SDS (lane Ivs lane 2). Analogous to the results with the cytosolic fractions,phosphorylation of a 48-kD protein(s) (arrow) was markedlyenhanced by SDS treatment in the mixture ofnormal cytosolicand membrane fractions (lanes 1 and 2), but not during re-

Cell-free Protein Phosphorylation and Neutrophil Oxidase Activation 1489

I

1.

14

%,

a-qw I

A. Membrane Fractions-Silver Stain

so 4s

_. ilA.E4 i.. W

f.i

Os6-iF:n

1

Ii

II

-200

-116

,-92.5

am

i -66

-45

-31

, w ~~~~~21.5

z t s14

_IIX\ + -/ \+ -I

Ctrl.- C CGD-CCtrl.-M CGD-M

Ctrl.-C CGD- CCGD-M Ctrl.-M

\+±/=I \_

Ctrl.-C CGD-CCtrl.-M CGD-M

\i+ / \+

Ctrl.-C CGD-CCGD-M Ctrl.-M

Figure 2. Phosphorylation of normal and CGD PMN membraneproteins in the cell-free system. Membrane fractions (70.3 ,Ag pro-tein) reseparated from cytosolic fractions in the reconstitution experi-ment described in Fig. 1 were analyzed by SDS-polyacrylamide gelelectrophoresis, silver staining, and indirect autoradiography. Abbre-

constitution ofthe CGD cytosol and membrane fraction (lanes3 and 4). Reconstitution of CGD membrane fractions withnormal cytosol fully restored SDS-dependent phosphorylationof the 48-kD phosphoprotein(s) (lanes 5 and 6). In contrast,minimal phosphorylation of the 48-kD phosphoprotein(s) bySDS occurred when the CGD cytosolic fraction was reconsti-tuted with normal membrane fraction (lanes 7 and 8), furtherindicating the requirement for a cytosolic component that isnot functional in autosomal recessive CGD PMN.

Autoradiograms of membrane fractions from two addi-tional cell-free reconstitution experiments, one from each pa-

viations and methods are as described in the legend to Fig. 1. Thearrow marks the position of the 48-kd phosphoprotein(s). Molecularweight markers are in kilodaltons. (A) Silver-stained polyacrylamidegel. (B) Corresponding autoradiogram.

tient having autosomal recessive CGD, were subjected to den-sitometric analysis. Phosphoproteins in the region of 48 kD(arrows) are shown in Fig. 3. In the presence ofSDS, phosphor-ylation of a 48-kD protein(s) as a result of reconstitution ofnormal cytosol with normal membrane fractions was ob-served. No phosphorylation of that protein(s) was observedduring reconstitution of CGD cytosol with CGD membranefractions in the presence of SDS. Reconstitution of normalcytosol with CGD membrane fractions and SDS restoredphosphorylation ofthe 48-kD protein(s). In contrast, reconsti-tution of CGD cytosol with normal membrane fractions and

1490 Caldwell, McCall, Hendricks, Leone, Bass, and McPhail

- 200

-116

-92.5

-66

- 45

-31

I -21.5

-14

w.lA.

wA.A.

-_;

L

--1!iFA::::ts:

I.

B. Membrane Fractions - Autoradiogram

00C

0.0

Rf

Figure 3. Densitometry scans of autoradiograms ofmembrane fractions from normal and CGD PMNfollowing reconstitution with SDS in the cell-free sys-tem. Unstimulated PMN from two normal donorsand two patients with autosomal recessive CGD (S.S.and S.H.) were fractionated; cytosolic and membranefractions were reconstituted in the presence of [y-32P]-ATP and 460 MM SDS and reseparated as describedin Methods. The experiment with S.S. is separatefrom the one described in Fig. 1. Membrane fractions(S.S., 100 ,ug protein; S.H., 41 gg protein) were sub-jected to SDS-polyacrylamide gel electrophoresis,silver-staining, and autoradiography. Autoradiogramswere scanned by a soft laser densitometer in theregion of the 48 kD phosphoprotein(s) (arrows).In all scans the bottom of the gel is to the left. (A)CGD 1 (S.S.). (B) CGD 2 (S.H.). (Top panels) Nor-mal cytosol + normal membrane; (second panels)CGD cytosol + CGD membrane; (third panels) nor-mal cytosol + CGD membrane; (bottom panels)CGD cytosol + normal membrane.

SDS resulted in minimal, or no, phosphorylation in that re-gion of the gel.

The area under the curve at 48 kD on the scans shown inFig. 3 and on scans of the lanes containing fractions isolatedfrom SDS-containing mixtures in the autoradiogram shown inFig. 2 was determined. If the area for the reconstitution mix-ture of normal cytosol and CGD membrane is set at 100%, thesummarized data were as follows: normal cytosol + normalmembrane = 75±22; CGD cytosol + CGD membrane= 1.5±1.5; CGD cytosol + normal membrane = 14±8 (per-cent, mean±SEM, n = 3). The inability of the CGD cytosoland membrane to phosphorylate the membrane-associated48-kd protein(s), and the markedly decreased ability of theCGD cytosol to support phosphorylation when mixed withnormal membrane thus was observed in three separate experi-ments. Taken together, these results indicate that a cytosolicfactor activity necessary for SDS-dependent activation ofNADPH oxidase and defective in two patients with autosomalrecessive CGD, is also necessary to support SDS-dependentphosphorylation of a 48-kD phosphoprotein(s).

Since phosphorylation of a 48-kD protein(s) by SDS in

reconstitution mixtures ofnormal cell fractions and the lack ofphosphorylation in mixtures of CGD cell fractions was ob-served in both cytosolic and membrane fractions, it is possiblethat the same protein(s) is distributed between both fractions.Cytosolic and membrane fractions reisolated from the normaland CGD reconstitution mixtures described in Figs. 1 and 2were analyzed on the same gel to test this possibility. Theresults are shown in Fig. 4. The position of the 48-kD phos-phoprotein(s) can be identified by the increase in phosphoryla-tion induced by SDS in the normal cell fractions and the ab-sence of that increase in the fractions from the CGD PMN.The 48-kD phosphoprotein(s) (arrow) in both cytosolic andmembrane fractions comigrated (lanes I and 2 from left), sup-porting the possibility that they are the same protein or familyof proteins.

It has been reported previously that neutrophils from pa-tients with the autosomal recessive form ofCGD demonstratea defect in phosphorylation of a protein(s) in the 47-48-kDregion when intact cells are exposed to a stimulus (15, 17). It isthus possible that the 48-kD protein(s) observed in our cell-freesystem is the same as that in the intact cell system. We tested

Cell-free Protein Phosphorylation and Neutrophil Oxidase Activation 1491

Normal CGD Normal CGD/ ~\ I

W,:1R

a.

-d *

-SDS

C NI+ SDS

eea:4w

.:

1CM,/~

-SD.-.r

0.

40.';Aw'

::c

- SD:

-66

-45

-31

-21.5

-14

Recon.Figure 4. Comigration in SDS-polyacrylamide gel of the 48 kd phos-phoprotein(s) in both cytosolic and membrane fractions following re-

constitution with SDS in the cell-free system. Cytosolic (C, 31 ygprotein) and membrane (M, 70 Mig protein) fractions from the experi-ment described in Fig. 1 were electrophoresed on the same polyacryl-amide gel and autoradiography was performed. +SDS, fractions re-

constituted in the presence of460 MM SDS; -SDS, fractions recon-

stituted in the presence of H20 (solvent control); Normal, PMNfractions from a normal donor; CGD, PMN fractions from a patientwith autosomal recessive CGD (S.S.). Molecular weight markers are

in kilodaltons. The arrow indicates the 48-kD position.

that possibility by analyzing, in the same gel, phosphoproteinpatterns of cytosolic fractions prepared from cell-free reconsti-tution mixtures incubated with [y-32P]ATP and cytosolic frac-tions prepared from resting or PMA-stimulated 32Pi-labeledintact cells (Fig. 5). Interestingly, the phosphoprotein patternsof the cell-free versus the intact cell cytosolic fractions were

quite dissimilar. However, a defect in both SDS-dependentphosphorylation of a 48-kD protein(s) (arrow) in reconstitu-tion mixtures and in PMA-stimulated phosphorylation of a

Figure 5. Comparative phosphorylation patterns in cytosolic frac-tions from cell-free reconstitution mixtures and from intact normaland CGD PMN. Cytosolic fractions from the reconstitution experi-ment described in Fig. 1 (31 Mg protein; Recon.) and from intact nor-

mal or CGD PMN that had been loaded with 32P; and exposed to ei-ther DMSO or 100 ng/ml PMA for 30 s (64 Mg protein; Intact) wereanalyzed in the same SDS-polyacrylamide gel and subjected to auto-radiography. SDS, +460 MAM SDS; H20, solvent control; Normal,fractions isolated from normal PMN; CGD, fractions isolated from a

patient (S.S.) with autosomal recessive CGD. Molecular weightmarkers are in kilodaltons. Arrow marks the position of the 48-kDphosphoprotein(s).

48-kD protein(s) was observed in cytosolic fractions fromCGD cells. In addition, the 48-kD phosphoprotein(s) in cyto-solic fractions from normal reconstitution mixtures and fromnormal intact cells appeared to comigrate, supporting the pos-

sibility that they are the same protein(s).PKC activity in autosomal recessive CGD PMN. It was

possible that the lack of phosphorylation of the 48-kD phos-phoprotein(s) and the absence of NADPH oxidase activitywere due to deficient levels of PKC. The activity and distribu-

1492 Caldwell, McCall, Hendricks, Leone, Bass, and McPhail

-200

-116

-92.5

g A.

V.-

'.f'

i I..!A.

11.'1rI

- 200

-116

-92.5

-66

-45

-31

-21.5

-14

Ha.t s1 F .i j. j F *g g

t IF a as s XA* * :**:

^ us. .F.

'I * *:SiiaA:..S

_

.w. A; _

(0 <

Intact

.i

Co °cI) I'

Recon.+SDS

&i

W

0<

Intact

.i',;

.t

I.-le.'

I

d

I.1

Um

i?

1? ~

..

tion of PKC were determined in crude subcellular fractionsfrom unstimulated and stimulated PMN of normal controlsand the two patients with autosomal recessive CGD. As shownin Fig. 6, the level and distribution ofPKC activity in cytosolicand particulate EP and NEP fractions was similar in normaland CGD PMN, under all conditions examined. PMA stimu-lation of PMN from both patients with CGD induced theredistribution and loss ofPKC activity similar to that observedin each paired normal control (Fig. 6, B). The effect of fMLPstimulation was examined in one patient with CGD and thenormal increase in particulate-associated (EP and NEP) kinaseactivity was observed. These studies suggest that PKC is notdefective in the two patients with CGD.

Discussion

The mechanisms that regulate NADPH oxidase, the respira-tory burst enzyme in human PMN, are not yet well under-stood; although, progress in elucidating the nature of the en-

800 A. DMSO Normal 2Normal 1 CGD 2

] Cytood

2W ~~~NEP

B. PMA

fat Normala IOr D Normal2 CGD 2

1i4002-

.-G 200

E C.fMLPNormal 1 CGD 1

600

400

200

0Fraction Assayed

Figure 6. PKC activity and distribution in subcellular fr-actions, fromnormal and CGD PMN. PKC activity was measured in subcellularfr-actions isolated from normal or CGD PMN that had been treatedwith either DMSO (A), 100 ng/ml PMA (B) or IO' M fMLP (C) for30 s at 370C as described in Methods. Fractions assayed were the cy-tosolic (o, Cytosol), the detergent extract of the particulate fraction(Es, EP), and the resuspended residual particulate pellet from the de-tergent extract (-, NEP). Activity shown is the mean of triplicate de-terminations on each fraction and is given as net activity (the activityin the presence of calcium and lipids minus the activity in the ab-sence of calcium and lipids) for the cytosolic and EP fractions and asactivity in the absence of calcium and lipids for the NEP fractions(see Methods). Activity in the absence of calcium and lipids in theEP and cytosolic fractions was not different in the fractions fromnormal and CGD PMN and was < 91 pmol/min per I07 PMN in cy-tosolic and 17 pmol/min per 107 PMN in EP fractions. Two pairedexperiments, each with PMN from a different normal and a differentpatient with autosomal recessive CGD, were performed: CGD 1,S.S.; CGD 2, S.H.

zyme complex at the molecular level is being made (20-22).Recently, a cell-free system for activation ofNADPH oxidasewas developed by us (25) and others (26-28) and can now beutilized to more directly study the mechanisms involved. Ofparticular interest is the role of a newly discovered cytosoliccellular cofactor required for activation of the oxidase in thissystem (25-28). Protein phosphorylation has been implicatedas a mechanism ofoxidase activation (3-17) and phosphoryla-tion has not been directly studied in the cell-free system.Therefore, we have examined whether or not phosphorylationofendogenous proteins correlated with the cell-free activationofNADPH oxidase and have investigated the role of the cyto-solic factor in the regulation of the observed phosphorylationevents.

Activation ofNADPH oxidase in the cell-free system cor-related specifically with phosphorylation ofa 48-kD protein(s).NADPH oxidase activation was achieved by the addition ofSDS to a reconstituted mixture of cytosolic and membranefractions from unstimulated normal PMN, in agreement withprevious reports in guinea pig macrophages (29) and humanPMN (33), and similar to results obtained with arachidonicacid instead ofSDS as the activator (25-28). SDS also inducedthe phosphorylation or dephosphorylation of at least 19 cyto-solic and 17 membrane-associated proteins, including a 48-kDprotein(s) seen in both fractions, out of 56 phosphoproteinsobserved in the reconstituted cell-free system. Thus, SDS ap-pears capable of activating a protein kinase(s), as well as eitheractivating a phosphatase(s) or inhibiting a protein kinase(s).Although the identity of the kinase or kinases affected by SDSis not yet known, we have found that SDS can activate PKC incytosolic fractions from human PMN as measured by exoge-nous histone phosphorylation (data not shown). Therefore, atleast some of the phosphorylation reactions observed may becatalyzed by PKC.

Phosphorylation ofthe 48-kD protein(s) by SDS was mark-edly decreased in cell-free reconstitution mixtures of cytosolicand membrane fractions from PMN of two patients with au-tosomal recessive CGD. This was the only phosphorylationdefect noted and it correlated with the dramatic decrease in theability of SDS to activate NADPH oxidase in the same frac-tions. The defects in activation of NADPH oxidase and inphosphorylation of the 48-kD protein(s) could both be attrib-uted to a defect in the cytosolic fraction from the CGD PMN.The membrane fraction from CGD PMN appeared to be nor-mal, in that it supported normal levels of both phosphoryla-tion of the membrane-associated 48-kD protein(s) and activa-tion of NADPH oxidase when reconstituted with a normalcytosolic fraction. In addition, reconstitution of CGD cyto-solic fraction with normal membrane fraction mimicked thedefects in phosphorylation ofthe 48-kD protein and in activa-tion of NADPH oxidase seen with the reconstituted CGDfractions. Thus, the cytosolic factor required for activation ofNADPH oxidase in the cell-free system is defective in PMN oftwo patients with the autosomal recessive form of CGD, asoriginally suggested by Curnutte et al. (31), and this defectcorrelates with defective phosphorylation of a 48-kD pro-tein(s).

These results clearly indicate that the cytosolic factor re-quired for activation of NADPH oxidase also is required forphosphorylation of the 48-kD protein(s). Our results do notdifferentiate between two possible explanations for the rela-tionship between the cytosolic factor and the 48-kD protein(s).

Cell-free Protein Phosphorylation and Neutrophil OxidaseActivation 1493

The cytosolic factor could be the 48-kD phosphoprotein(s) andits presence in the membrane due to partial translocation in-duced by SDS. If so, then the presence of the 48-kD phospho-protein in the CGD membrane fraction could be caused bytranslocation from the normal cytosolic fraction during re-constitution. In fact, the 48-kD protein(s) in normal mem-brane fractions comigrated with the 48-kD protein(s) in nor-mal cytosolic fractions, supporting the possibility that they arethe same protein(s) and could either have dual localization orbe translocated by SDS treatment. Observations consistentwith the possibility that translocation of the cytosolic factoroccurs are that, under conditions that prime intact normalPMN or macrophages, NADPH oxidase in membrane frac-tions can be activated in the cell-free system without the addi-tion of cytosolic fractions (25, 29).

It is also possible that the cytosolic factor regulates phos-phorylation ofthe 48-kD protein(s) because the factor is itselfakinase or it activates a kinase. Our results suggest that thecytosolic factor is not PKC, since PKC activity appeared to bepresent at normal levels and to respond normally to stimula-tion by either PMA or fMLP in CGD PMN. However, wecannot exclude the possibility of a more subtle defect in PKCactivity or regulation not detected by our assay procedure.That the cytosolic factor is not PKC is also supported by othersusing inhibitors (33, 34, 55) or column fractionation (30) todissociate PKC activity from the oxidase activating factor innormal PMN cytosol. It is interesting that 46-48-kD proteinsin guinea pig (56) and human (57) PMN have been reported tobe substrates for PKC or its catalytic subunit, although it is notyet known if these and the protein(s) observed here are thesame. Further understanding of the relationship between the48-kD protein, the cytosolic factor required for NADPH oxi-dase activation, and PKC or other kinases will necessitate puri-fication ofthe components and more rigorous physical charac-terization of the 48-kD protein or proteins.

A defect in phosphorylation of a 47-48-kD protein, orfamily of proteins, in PMN from patients with CGD has beenobserved previously by stimulation of intact cells (15-17).Segal et al. (15), using one-dimensional polyacrylamide gelelectrophoretic separation, reported that the defect was re-stricted to patients with the autosomal recessive form of thedisease. However, Hayakawa et al. (16) and Okamura et al.(17), using two-dimensional gel electrophoresis, found thatboth forms ofthe disease expressed defects in phosphorylationof a 48-kD family of proteins and suggested that a more pro-nounced defect is present in the autosomal recessive form. Athird group (24), using two-dimensional gel electrophoresis,reported no defect in protein phosphorylation in six patientswith CGD (presumably representing both forms of inheri-tance). Our results, using one-dimensional gel electrophoresis,indicate, for the first time, that a defect in phosphorylation ofa48-kD protein(s) can be observed in a cell-free system fromPMN of patients with the autosomal recessive form of CGDand relate the defect to a deficient cytosolic factor required foractivation ofNADPH oxidase. The 48-kD protein(s) observedin our cell-free system by SDS treatment comigrated with a48-kD protein observed by stimulating intactPMN with PMA,consistent with the possibility that the phosphorylation defectin intact CGD PMN is the same as in the cell-free system.However, further analysis by two-dimensional gel electropho-resis, peptide mapping, and phosphoamino acid determina-tion will be necessary to fully identify and characterize the

48-kd protein(s) observed by the various groups and to deter-mine if all groups are studying the same family of proteins.

The relationship of the phosphorylation of the 48-kD pro-tein(s) to the activation ofNADPH oxidase in the intact cell orby SDS in the cell-free system is still not clear. Phosphoryla-tion of a protein or proteins at 46-48 kD has been correlatedwith activation ofthe respiratory burst in intact cells by severallaboratories (8, 56-58). However, other data dissociates phos-phorylation of a 48-kD protein(s) from stimulation of oxida-tive metabolism (59-61). With the exception of Ohtsuka et al.(56), these studies have utilized one-dimensional gel separa-tion systems and, if the 48-kD region contains a family ofproteins (16, 17, 24), such discrepancies might be expected.

A requirement for a protein kinase activity in the SDS-me-diated activation of NADPH oxidase in the cell-free systemhas not yet been demonstrated nor conclusively eliminated.Oxidase activation by SDS or arachidonate does not requireexogenously added ATP (30, 35) and depletion ofendogenousATP has been reported by two groups to inhibit (35, 55) andby another to have no effect (30) on oxidase activation. PKChas been shown to activate NADPH oxidase in a cell-free sys-tem (62); however, the mechanism utilized by PKC appears tobe different than that utilized by SDS and arachidonate (33)and PKC does not seem to be required for the latter com-pounds to exert their activating effect (30, 33, 34). These re-sults and our own suggest that if a kinase is involved in theSDS-activated cell-free system, it is not PKC.

Based on the present state of knowledge and the work re-ported here, at least three possibilities for the role ofthe 48-kDprotein(s) in the activation of NADPH oxidase are apparent.The 48-kD protein(s) may be unrelated to oxidase activity inPMN, but its phosphorylation may depend upon a normalcytosolic factor activity. Alternatively, the 48-kD protein(s)may be a component(s) ofNADPH oxidase; a protein of thatmolecular weight appears, albeit variably, in preparations ofthe purified oxidase flavoprotein (63). Finally, the 48-kD pro-tein(s) could be a regulatory molecule(s), influencing one ormore of several pathways (2, 11, 18, 19, 33, 36) ofactivation ofoxidative metabolism. Further efforts will be made to distin-guish between these possibilities and to resolve the currentdiscrepancies in the literature by fully characterizing the48-kD protein(s).

Acknowledgments

We thank Sue Cousart, William Crone, and Gary Nabors for theirexcellent technical assistance and Susan Britt for valuable secretarialhelp. We especially thank Dr. Mike Cohen for providing access to hispatient with autosomal recessive CGD.

This work was supported in part by National Institutes of Healthgrants AI-22564, AI-10732, AI-09169, AI-14929, and HL-29293.

References

1. Babior, B. M. 1978. Oxygen-dependent microbial killing byphagocytes. N. Engl. J. Med. 298:659-668.

2. McPhail, L. C., and R. Snyderman. 1984. Mechanisms of regu-lating the respiratory burst in leukocytes. In Regulation of LeukocyteFunction. R. Snyderman, editor. Plenum Publishing Corp., NewYork. 247-281.

3. Schneider, C., M. Zanetti, and D. Romeo. 1981. Surface-reactivestimuli selectively increase protein phosphorylation in human neutro-phils. FEBS (Fed. Eur. BioL. Soc.) Lett. 127:4-8.

1494 Caldwell, McCall, Hendricks, Leone, Bass, and McPhail

4. Andrews, P. C., and B. M. Babior. 1983. Endogenous proteinphosphorylation by resting and activated human neutrophils. Blood.61:333-340.

5. Andrews, P. C., and B. M. Babior. 1984. Phosphorylation ofcytosolic proteins by resting and activated human neutrophils. Blood.64:883-890.

6. White, J. R., C.-K. Huang, J. M. Hill, Jr., P. H. Naccache, E. L.Becker, and R. I. Sha'afi. 1984. Effect of phorbol 12-myristate 13-ace-tate and its analogue 4-a-phorbol 12,13-didecanoate on protein phos-phorylation and lysosomal enzyme release in rabbit neutrophils. J.Biol. Chem. 259:8605-8611.

7. Okamura, N., S. Ohashi, N. Nagahisa, and S. Ishibashi. 1984.Changes in protein phosphorylation in guinea pig polymorphonuclearleukocytes by treatment with membrane-perturbing agents whichstimulate superoxide anion production. Arch. Biochem. Biophys.228:270-277.

8. Pontremoli, S., E. Melloni, F. Salamino, B. Sparatore, M. Mi-chetti, 0. Sacco, and B. L. Horecker. 1986. Phosphorylation of pro-teins in human neutrophils activated with phorbol myristate acetate orwith chemotactic factor. Arch. Biochem. Biophys. 250:23-29.

9. Nishizuka, Y. 1984. The role of protein kinase C in cell surfacesignal transduction and tumour promotion. Nature (Lond.). 308:693-698.

10. Wolfson, M., L. C. McPhail, V. N. Nasrallah, and R. Snyder-man. 1985. Phorbol myristate acetate mediates redistribution of pro-tein kinase C in human neutrophils: potential role in the activation ofthe respiratory burst enzyme. J. Immunol. 135:2057-2062.

11. Gerard, C., L. C. McPhail, A. Marfat, N. P. Stimler-Gerard,D. A. Bass, and C. E. McCall. 1986. Role ofprotein kinases in stimula-tion ofhuman polymorphonuclear leukocyte oxidative metabolism byvarious agonists. Differential effects ofa novel protein kinase inhibitor.J. Clin. Invest. 77:61-65.

12. Gennaro, R., C. Florio, and D. Romeo. 1986. Co-activation ofprotein kinase C and NADPH oxidase in the plasma membrane ofneutrophil cytoplasts. Biochem. Biophys. Res. Commun. 134:305-312.

13. Cox, C. C., R. W. Dougherty, B. R. Ganong, R. M. Bell, J. E.Niedel, and R. Snyderman. 1986. Differential stimulation ofthe respi-ratory burst and lysosomal enzyme secretion in human polymorpho-nuclear leukocytes by synthetic diacylglycerols. J. Immunol.136:4611-4616.

14. Rider, L. G., and J. E. Niedel. 1987. Diacylglycerol accumula-tion and superoxide anion production in stimulated human neutro-phils. J. Biol. Chem. 262:5603-5608.

15. Segal, A. W., P. G. Heyworth, S. Cockcroft, and M. M. Bar-rowman. 1985. Stimulated neutrophils from patients with autosomalrecessive chronic granulomatous disease fail to phosphorylate aM,-44,000 protein. Nature (Lond.). 316:547-549.

16. Hayakawa, T., K. Suzuki, S. Suzuki, P. C. Andrews, and B. M.Babior. 1986. A possible role for protein phosphorylation in the acti-vation of the respiratory burst in human neutrophils. Evidence fromstudies with cells from patients with chronic granulomatous disease. J.Biol. Chem. 261:9109-9115.

17. Okamura, N., J. T. Curnutte, and B. M. Babior. 1987. Phos-phoproteins and the activation ofthe neutrophil respiratory burst oxi-dase. Fed. Proc. 46:1952. (Abstr.)

18. Tauber, A. I., N. Borregaard, E. Simons, and J. Wright. 1983.Chronic granulomatous disease: a syndrome of phagocyte oxidase de-ficiencies. Medicine (Baltimore). 62:286-309.

19. Curnutte, J. T., and B. M. Babior. 1987. Chronic granuloma-tous disease. In Advances in Human Genetics. Vol. 16. H. Harris andK. Hirschhorn, editors. Plenum Publishing Corp., New York.229-297.

20. Royer-Pokora, B., L. M. Kunkel, A. P. Monoaco, S. C. Goff,P. E. Newburger, R. L. Baehner, F. S. Cole, J. T. Curnutte, and S. H.Orkin. 1986. Cloning the gene for an inherited human disorder-chronic granulomatous disease-on the basis of its chromosomal lo-cation. Nature (Lon4.). 322:32-38.

21. Dinauer, M. C., S. H. Orkin, R. Brown, A. J. Jesaitis, and C. A.Parkos. 1987. The glycoprotein encoded by the X-linked chronicgranulomatous disease locus is a component of the neutrophil cy-tochrome b complex. Nature (Lond.). 327:717-720.

22. Teahan, C., P. Rowe, P. Parker, N. Totty, and A. W. Segal.1987. The X-linked chronic granulomatous disease gene codes for theB-chain of cytochrome b-245. Nature (Lond.). 327:720-72 1.

23. Segal, A. W., A. R. Cross, R. C. Garcia, N. Borregaard, N. H.Valerius, J. F. Soothill, and 0. T. G. Jones. 1983. Absence of cy-tochrome b-245 in chronic granulomatous disease. A multicenter Euro-pean evaluation of its incidence and relevance. N. Engl. J. Med.308:245-251.

24. Ishii, E., K. Irita, I. Fujita, K. Takeshige, M. Kobayashi, T.Usui, and K. Ueda. 1986. Protein phosphorylation ofneutrophils fromnormal children and patients with chronic granulomatous disease.Eur. J. Pediatr. 145:22-26.

25. McPhail, L. C., P. S. Shirley, C. C. Clayton, and R. Snyderman.1985. Activation of the respiratory burst enzyme from human neutro-phils in a cell-free system. Evidence for a soluble cofactor. J. Clin.Invest. 75:1735-1739.

26. Bromberg, Y., and E. Pick. 1984. Unsaturated fatty acids stim-ulate NADPH-dependent superoxide production by cell-free systemderived from macrophages. Cell. Immunol. 88:213-221.

27. Heyneman, R. A., and R. E. Vercauteren. 1984. Activation ofaNADPH oxidase from horse polymorphonuclear leucocytes in a cell-free system. J. Leukocyte Biol. 36:751-759.

28. Curnutte, J. T. 1985. Activation ofhuman neutrophil nicotin-amide adenine dinucleotide phosphate, reduced (triphosphopyridinenucleotide, reduced) oxidase by arachidonic acid in a cell-free system.J. Clin. Invest. 75:1740-1743.

29. Bromberg, Y., and E. Pick. 1985. Activation of NADPH-de-pendent superoxide production in a cell-free system by sodium dode-cyl sulfate. J. Biol. Chem. 260:13539-13545.

30. Curnutte, J. T., R. Kuver, and P. J. Scott. 1987. Activation ofneutrophil NADPH oxidase in a cell-free system. Partial purificationof components and characterization of the activation process. J. Biol.Chem. 262:5563-5569.

31. Cumutte, J. T., P. J. Scott, R. Kuver, and R. Berkow. 1986.NADPH oxidase activation cofactor (ACF): partial purification andabsent activity in a patient with chronic granulomatous disease (CGD).Clin. Res. 34:455a. (Abstr.)

32. McPhail, L. C., C. C. Clayton, and R. Snyderman. 1984. Apotential second messenger role for unsaturated fatty acids: activationof Ca2'-dependent protein kinase. Science (Wash. DC). 224:622-625.

33. Cox, J. A., A. Y. Jeng, P. M. Blumberg, and A. I. Tauber. 1987.Comparison of subcellular activation of the human neutrophilNADPH-oxidase by arachidonic acid, sodium dodecyl sulfate (SDS),and phorbol myristate acetate (PMA). J. Immunol. 138:1884-1888.

34. McPhail, L. C., C. L. Hendricks, M. C. Seeds, and D. A. Bass.1986. A possible role for calmodulin in activation of the respiratoryburst enzyme ofhuman neutrophils. Clin. Res. 34:465a. (Abstr.)

35. Clark, R. A., D. G. Leidal, D. W. Pearson, and W. M. Nauseef.1987. NADPH oxidase ofhuman neutrophils. Subcellular localizationand characterization of an arachidonate-activatable superoxide-gener-ating system. J. Biol. Chem. 262:4065-4074.

36. McPhail, L. C., and R. Snyderman. 1983. Activation of therespiratory burst enzyme in human polymorphonuclear leukocytes bychemoattractants and other soluble stimuli. Evidence that the sameoxidase is activated by different transductional mechanisms. J. Clin.Invest. 72:192-200.

37. Sanders, D. Y., M. R. Cooper, C. E. McCall, and L. R. DeCha-telet. 1972. Chronic granulomatous disease of childhood with onset ofsymptoms at age 11 years. J. Pediatr. 80:104-106.

38. Amrein, P. C., and T. P. Stossel. 1980. Prevention of degrada-tion ofhuman polymorphonuclear leucocyte proteins by diisopropyl-fluorophosphate. Blood. 56:442-447.

39. Gabig, T. G. D. English, L. P. Akard, and M. J. Schell. 1987.Regulation of neutrophil NADPH oxidase activation in a cell-free

Cell-free Protein Phosphorylation and Neutrophil Oxidase Activation 1495

system by guanine nucleotides and fluoride. Evidence for participationof a pertussis and cholera toxin-insensitive G protein. J. Biol. Chem.262:1685-1690.

40. Huey, R., and T. E. Hugh. 1985. Characterization of a C5areceptor on human polymorphonuclear leukocytes (PMN). J. Im-munol. 135:2063-2068.

41. Wroblewski, F., and J. S. LaDue. 1955. Lactic dehydrogenaseactivity in blood. Proc. Soc. Exp. Biol. Med. 90:210-213.

42. DeChatelet, L. R., J. V. Volk, C. E. McCall, and M. R. Cooper.1971. Studies on leukocyte phosphatases. II. Inhibition of leukocytealkaline phosphatase by amino acids and its reversal by zinc. Clin.Chem. 17:210-213.

43. Gottlieb, C., K. Lau, L. R. Wasserman, and V. Herbert. 1965.Rapid charcoal assay for intrinsic factor (IF), gastric juice unsaturatedB12 binding capacity, antibody to IF, and serum unsaturated B12 bind-ing capacity. Blood. 25:875-883.

44. Zeya, H. I., and J. K. Spitznagel. 1971. Characterization ofcationic protein-bearing granules of polymorphonuclear leukocytes.Lab. Invest. 24:229-236.

45. Migler, R., and L. R. DeChatelet. 1978. Human eosinophilicperoxidase: Biochemical characterization. Biochem. Med. 19:16-26.

46. Lowry, 0. H., N. J. Rosebrough, A. L. Farr, and R. J. Randall.1951. Protein measurement with the Folin phenol reagent. J. Biol.Chem. 193:265-275.

47. Massey, V. 1959. The microestimation of succinate and theextinction coefficient of cytochrome c. Biochim. Biophys. Acta.34:255-256.

48. Gabig, T. G., E. W. Schervish, and J. T. Santinga. 1982. Func-tional relationship of the cytochrome b to the superoxide-generatingoxidase of human neutrophils. J. Biol. Chem. 257:4114-4119.

49. Ohno, Y., E. S. Buescher, R. Roberts, J. A. Metcalf, and J. I.Gallin. 1986. Reevaluation of cytochrome b and flavin adenine dinu-cleotide in neutrophils from patients with chronic granulomatous dis-ease and description of a family with probable autosomal recessiveinheritance of cytochrome b deficiency. Blood. 67:1132-1138.

50. Laemmli, U. K. 1970. Cleavage of structural proteins duringassembly of the head of bacteriophage T4. Nature (Lond.). 227:680-685.

51. Peterson, G. L. 1977. A simplification of the protein assaymethod of Lowry et al. which is more generally applicable. Anal. Bio-chem. 83:346-356.

52. Wray, W., T. Boulikas, V. P. Wray, and R. Hancock. 1981.Silver staining of proteins in polyacrylamide gels. Anal. Biochem.118:197-203.

53. Laskey, R., and A. D. Mills. 1977. Enhanced autoradiographicdetection of 32P and 1251 using intensifying screens and hypersensitizedfilm. FEBS (Fed. Eur. Biol. Soc.) Lett. 82:314-316.

54. Pike, M. C., L. Jakoi, L. C. McPhail, and R. Snyderman. 1986.Chemoattractant-mediated stimulation of the respiratory burst inhuman polymorphonuclear leukocytes may require appearance ofprotein kinase activity in the cells' particulate fraction. Blood. 67:909-913.

55. Seifert, R., and G. Schultz. 1987. Fatty-acid-induced activationof NADPH oxidase in plasma membranes of human neutrophils de-pends on neutrophil cytosol and is potentiated by stable guanine nu-cleotides. Eur. J. Biochem. 162:563-569.

56. Ohtsuka, T., N. Okamura, and S. Ishibashi. 1986. Involvementof protein kinase C in the phosphorylation of 46 kDa proteins whichare phosphorylated in parallel with activation of NADPH oxidase inintact guinea-pig polymorphonuclear leukocytes. Biochim. Biophys.Acta. 888:332-337.

57. Pontremoli, S., E. Melloni, M. Michetti, B. Sparatore, F. Sala-mino, 0. Sacco, and B. L. Horecker. 1987. Phosphorylation and pro-teolytic modification ofspecific cytoskeletal proteins in human neutro-phils stimulated by phorbol 12-myristate 13-acetate. Proc. Nati. Acad.Sci. USA. 84:3604-3608.

58. Heyworth, P. G., and A. W. Segal. 1986. Further evidence forthe involvement of a phosphoprotein in the respiratory burst oxidaseofhuman neutrophils. Biochem. J. 239:723-73 1.

59. Caldwell, S. E., D. A. Bass, L. C. McPhail, and C. E. McCall.1987. Phosphorylation of a 48 kD protein, defective in chronic granu-lomatous disease (CGD) neutrophils, is not required for activation ofthe respiratory burst. Fed. Proc. 46:1033. (Abstr.)

60. Sha'afi, R. I., T. F. P. Molski, and C.-K. Huang. 1987. Dissocia-tion of the 47 kDa protein phosphorylation from degranulation andsuperoxide production in neutrophils. Fed. Proc. 46:1037. (Abstr.)

61. Reibman, J., H. M. Korchak, K. A. Haines, L. B. Vosshall, andG. Weissmann. 1987. Phosphorylation of the 47 kDa target protein isnot sufficient for superoxide anion production in neutrophils. Clin.Res. 35:537a. (Abstr.)

62. Cox, J. A., A. Y. Jeng, N. A. Sharkey, P. M. Blumberg, and A. I.Tauber. 1985. Activation of the human neutrophil nicotinamide ade-nine dinucleotide phosphate (NADPH)-oxidase by protein kinase C. J.Clin. Invest. 76:1932-1938.

63. Glass, G. A., D. M. DeLisle, P. DeTogni, T. G. Gabig, B. H.Magee, M. Markert, and B. M. Babior. 1986. The respiratory burstoxidase ofhuman neutrophils. Further studies ofthe purified enzyme.J. Biol. Chem. 261:13247-13251.

1496 Caldwell, McCall, Hendricks, Leone, Bass, and McPhail

0022-1767/88/14010-3560$02.00/0 THE JOURNAL OF IMMUNOLOGY Copyright 0 1988 by The American Assodation of Immunologists

Vol. 140.3560-3567. No. 10. May 15.1988 Printed (n U . S . A .

ALTERATIONS IN CELL PROTEIN PHOSPHORYLATION IN HUMAN NEUTROPHILS EXPOSED TO INFLUENZA A VIRUS

A Possible Mechanism for Depressed Cellular End-Stage Functions’

SUSAN E. CALDWELL,2’ LISA F. CASSIDY,+ AND JON S. ABRAMSON3* From the Departments of ‘Medicine, ‘Microbiology and Immunology, and ‘Pediatrics, Wake Forest University Medical Center,

Winston-Salem. NC 27103

When polymorphonuclear leukocytes (PMN) are exposed to most harvests of influenza A virus (de- pressing virus, DV) for 20 min, chemotactic, secre- tory, and oxidative functions are depressed upon subsequent exposure to soluble or particulate stim- uli. Other harvests of influenza A virus (non-DV) do not alter these activities. The DV-induced changes in multiple functions suggest the virus may inter- fere with steps involved in PMN activation. Because some of these steps may be regulated by protein phosphorylation, we examined the effect of non-DV and DV on cellular protein phosphorylation.

PMN loaded with 32P-labeled inorganic orthophos- phate were exposed to non-DV, DV, or buffer for 30 min: cells were then treated with buffer, FMLP ( M), or PMA (100 ng/ml) for 30 s. Samples were soni- cated and centrifuged: cytosolic and particulate fractions were analyzed by SDS-PAGE and autora- diography. Exposure of PMN to either non-DV or DV caused phosphorylation of several cell proteins. However, when DV-treated PMN were then stimu- lated with FMLP or PMA, further phosphorylation was inhibited compared to non-DV- or buffer-treated cells, This suggests that DV-induced depression of PMN end-stage functions may be due to changes in cell protein phosphorylation.

DV could interfere with phosphorylation of PMN proteins by altering protein kinase activity. We therefore examined the influence of non-DV and DV on some parameters that could affect kinase func- tion. PMN intracellular [Ca”] was monitored by us- ing the fluorescent Ca2+ indicator, Indo 1, and cAMP levels were measured by RIA. PMN treated with DV alone or DV plus FMLP had higher intracellular [CA2+) than PMN similarly treated with non-DV or buffer. Exposure of PMN to non-DV, DV. or buffer

Received for publication December 1, 1987. Accepted for publicatton February 16. 1988. The costs of publication of this article were defrayed in part by the

aduertfsernent in accordance with 18 U.S.C. Section 1734 solely to indi- payment of page charges. This article must therefore be hereby marked

cate this fact.

HL-29293. J. S . A. is the recipient of Research Career Development Award ‘ This study was supported in part by Grants AI-20506, AI-09169, and

AI-00670 from the National lnstitutes of Health. This work has been published in abstract form, entitled “Alterations in Protein Phosphoryla- tion During Influenza Virus Infection of Neutrophils May Induce De- pressed Cellular End-Stage Functions” [Clin. Res. 35:470A. 1987).

94063. Present address: Genelabs. lnc., 505 Penobsot Dr.. Redwood City, CA

Author to whom reprint requests should be addressed: Jon S . Abram-

Center, 300 S . Hawthorne Road. Winston-Salem, NC 27103. son. M.D.. Department of Pediatrics. Wake Forest University Medical

caused minimal changes in cAMP levels, and similar increases occurred in cAMP levels upon FMLP stim- ulation,

To determine whether DV interferes with trans- membrane signaling, the effect of influenza virus on PMN transmembrane potential was studied by using a fluorescent cyanine dye. Transmembrane potential changes were greater in PMN exposed to DV than to non-DV or buffer: however, subsequent stimulation with FMLP caused equivalent changes in transmembrane potential.

Our data show that protein phosphorylation in PMN is induced by DV and non-DV infection: upon subsequent stimulation with FMLP or PMA, there is inhibited cellular phosphorylation only in PMN pre- viously exposed to DV. Studies on cAMP levels and transmembrane potential changes suggest these pa- rameters are not involved in DV-induced depression of phosphorylation. The role of intracellular [Ca”] in inhibiting DV-induced phosphorylation remains to be clarified.

Exposure of PMN4 to most harvests of influenza A virus causes inhibition of in vitro end-stage PMN functions upon stimulation with secondary soluble and particulate stimuli (1 -7). Those depressed functions include respi- ratory burst activity, secretion, and chemotaxis. In addi- tion, the harvests of virus which inhibit these functions also cause disruption of lysosome-phagosome fusion (4); these viruses are designated DV. Maximal depression of PMN function occurs within 30 min of exposure of PMN to the virus. The inhibition of PMN function caused by DV is therefore occurring while virus particles are undergoing endocytosis into the PMN (4.6) but before the production of viral proteins (8).

Occasional harvests of influenza virus do not interfere with lysosome-phagosome function or PMN end-stage functions: these harvests are designated non-DV (4). Studies have shown that at least as many non-DV parti- cles are taken up into PMN when compared with DV, and that DV, but not non-DV, causes inhibition of fluid- and receptor-mediated endocytosis of PMN (7). Thus, the dif- fering effects that DV and non-DV have on PMN function is not simply due to increased numbers of DV particles taken up by the cell. Indeed, the cumulative data suggest

CB, cytochalasin 9; DV. depressing virus; CGD. chronic granulomatous Abbreviations used in this paper: PMN. polymorphonuclear leukocyte:

disease; non-DV, non-depressing virus: Di0C5(3). 3.3-dipentyloxacarbo- cyanine: N,. guanine nucleotide protein: NI. non-infected.

3560

VIRUS-INDUCED CHANGES IN NEUTROPHIL PROTEIN PHOSPHORYLATION 356 1

that influenza virus is interfering with one or more steps in the activation pathway of PMN.

Protein phosphorylation may have a major regulator role in activation of PMN responses to stimulating agents such as the chemotactic factor, FMLP, or the tumor pro- moter, PMA. Indeed, phosphoproteins in uninfected PMN undergo numerous changes in phosphorylation upon stimulation of the cells with FMLP, PMA, and other ag- onists (9-18). changes which precede such responses as activation of the respiratory burst, chemotaxis, and de- granulation.

Recently, phosphorylation of a 48-kDa protein in hu- man PMN has been correlated with activation of the respiratory burst enzyme, NADPH oxidase, in studies done with PMN from patients with CGD (15.17). Such cells cannot mount a respiratory burst upon stimulation: therefore, a phosphorylation defect in PMN from CGD patients could indicate a specific phosphoprotein in- volved in regulating the respiratory burst. A phosphopro- tein with a m.w. of 44,000 (15) or 48,000 (17), phos- phorylated in response to various stimuli in control PMN, is not phosphorylated in PMN from patients with CGD. We have shown that, in a cell-free, reconstituted system derived from PMN from a patient with CGD, a 48-kDa protein is not phosphorylated upon activation of the sys- tem with SDS.4 Additionally, phosphorylation of a 48- kDa protein present in purified membranes from CGD PMN was restored in the presence of normal cytosol under conditions in which a respiratory burst was also generated. These results further correlate phosphoryla- tion of this protein with respiratory burst activation, although a causal relationship has not yet been demon- strated. In the studies described here, depression of phos- phorylation of the 48-kDa cytosolic protein in influenza virus-treated PMN occurs in parallel with depression of PMN oxidative metabolism.

Inasmuch as exposure of PMN to influenza virus is known to depress multiple end-stage PMN responses [ 1 - 7). the effect of DV and non-DV on PMN protein phos- phorylation was studied. DV, but not non-DV, was found to cause marked inhibition in protein phosphorylation when cells were stimulated with FMLP or PMA. Further studies were then done to examine the effect of influenza virus on other parameters which could affect protein phosphorylation: intracellular [Ca"'], cAMP levels, and PMN transmembrane potential. Interference with the first two parameters could directly affect protein phos- phorylation in virus-infected PMN by influencing activity of the Ca2+- or CAMP-dependent protein kinases. Changes in transmembrane signaling could indirectly influence protein phosphorylation by affecting steps in cell acti- vation before phosphorylation. The data suggest DV does not alter protein phosphorylation by affecting cAMP lev- els or transmembrane potential. However, the role of intracellular [Ca2+] in DV-induced changes in protein phosphorylation requires further study.

MATERIALS AND METHODS

Isolation of PMN. Purified human PMN (2 97%) were obtained from heparinized whole blood by dextran sedimentation, separation

4Caldwell. S. E., C. E. McCall, C. L. Hendricks. P. A. Leone, D. A. Bass, and L. C. McPhail. Reconstitution of defect in phosphorylation of a 48 kd protein and NADPH oxidase activity in neutrophils from a patient with chronic granulomatous disease. J. Clin. Inuest. Submittedfor pub- Iication.

of the buffy coat component cells on Lymphocyte Separation Medium

described (3). Cells were resuspended in HBSS (pH 7.4) with 1.3 mM (Bionetics, Inc., Bethesda, MD). and hypotonic lysis of E as previously

ea2+. 0.9 mM MgZ+, and 0.1 % gelatin unless noted otherwise.

from allantoic fluid of embryonated chicken eggs (19). purified by Preparation of influenza virus. Influenza A virus was harvested

centrifugation on 15 to 60% (w/w) linear sucrose gradients, and dialyzed against PBS (10 mM sodium phosphate in 0.9% NaCl, pH 7.4) overnight. A recombinant virus (X-47) containing the internal proteins of influenza A/Puerto Rico/8/34 (H,NI) virus and the sur- face glycoproteins of influenza A/Victoria/3/75 (&&?) virus was used in all experiments. Most harvests of the X-47 virus inhibited oxidative, secretory, and chemotactic activities in PMN (DV). Several harvests of this virus did not depress PMN functions (non-DV). Determination of the 50% egg infectivity dose. hemagglutinin and neuraminidase activities, total protein content, and SDS-PAGE pat- terns have not established the difference between DV and non-DV harvests (our unpublished data). The amount of DV and non-DV used in each experiment was adjusted to achieve equal virus protein concentration (100 pg viral protein/l x 10' PMN unless otherwise stated) as determined by the method of Lowry et al. (20). Each virus harvest was aliquoted and stored at -70°C until use.

Protein phosphorylation in influenza virus-infected PMN. Pro- tein phosphorylation experiments were modifications of those de- scribed by s. E. Caldwell et al. (see footnote 4) for uninfected, intact PMN. Purified PMN (5 X lo7 PMN/sample) were suspended at a concentration of 1 X 10' PMN/ml in phosphate-free buffer contain- ing 6 mM HEPES-Tris (pH 7.4), 0.15 mM NaCI, 1 0 mM glucose, 5 mM KCI, 1 mM MgCl,, 0.25 mM CaCI2 (1 0). and 32P, ( 1 75 &i/5 X 1 O7 PMN: New England Nuclear. Boston, MA) and incubated for 30 min at 37°C with constant gentle agitation. Non-DV or DV in PBS (40 pg viral protein/5 x 1 O7 PMN) or PES alone was added to PMN so that cells were in a total volume of 0.6 to 0.7 ml, depending upon virus concentration. The incubation was continued with gentle agitation for 30 min at 37°C: PMN were then further diluted with warmed phosphate-free buffer to a volume of 1.0 ml (5 X lo7 cells/ml) and stimulated for 30 sec at 37°C with FMLP M) without CB. or with PMA ( 100 ng/ml). Reactions were stopped with a 15-fold volume of ice-cold phosphate-free buffer also containing 10 mM EDTA and 100 mM NaF as previously described (12). PMN were washed and incubated on ice for 5 min at a concentration of 5 X IO7 cells/ml in 2 mM diisopropylfluorophosphate to inhibit proteases (21) and washed once more the same cold buffer.

PMN were resuspended in 0.7 ml sonication buffer containing 0.34 M sucrose, 1 mM EDTA, 1 mM EGTA, 10 mM sodium pyro- phosphate, and 100 mM NaF. Cells were sonicated in a melting ice bath, with a model W-220 sonicator (Heat Systems-Ultrasonics, Inc.. Farmingdale, NY) fitted with a stepped microtip. Sonication was done in 5- or 10-s bursts for a total of 25 s . and breakage was monitored by phase microscopy. Centrifugation at 500 x g for 10 min at 4°C removed intact cells and nuclei; the supernatant was centrifuged at 100.000 X g for 1 h at 4°C in a Beckman type 50 rotor. Resulting cytosolic fractions (supernatants] were removed and the pelleted particulate fractions resuspended in 0.4 ml of the same sucrose buffer by sonication. Cytosols and particulate fractions were divided into roughly equal aliquots: one aliquot of each was frozen as native protein a t -70°C and later assayed for protein concentra- tion by Peterson's modification (22) of the method of Lowry et al. (20). These aliquots were also used for liquid scintillation counting to monitor 32P incorporation. The remainder of each sample was solubilized immediately by adding Laemmli electrophoresis sample buffer (23) and boiling in a water bath for 2 min. Solubilized fractions were stored at -70°C for no more than 3 days, and then analyzed by SDS-PAGE and indirect autoradiography, described below.

SDS-PAGE and indirect autoradiography. Reduced, denatured samples (30 to 70 pg protein/lane) were electrophoresed by using the discontinuous buffer system of Laemmii (23) on linear 8 to 15% gradient slab gels as described by S . E. Caldwell et al. (see footnote 4). Electrophoresis was carried out for 12 h at 10°C. by using equal amounts of protein in each gel lane within an experiment. Gels were silver-stained (24) to detect total protein, and then dried between two sheets of cellophane by using a model 443 slab gel dryer (Bio- Rad: Rockville Centre, NY). Dried gels were exposed to pre-flashed

ning Plus intensifying screen at -70°C for 1 to 3 days to generate Kodak X-Omat x-ray film in the presence of a Dupont Cronex Light-

autoradiograms (25). Autoradiograms were scanned with an Ultro- scan XL laser densitometer (LKB. Bromma, Sweden).

P M N intracellular [Ca"]. Intracellular [Ca2'] was measured in PMN washed in HBSS without ea2+ or Mg2' and then resuspended in HBSS with 1.6 mM CaC12. The acetoxymethyl ester of the fluores- cent compound, Indo 1 (Molecular Probes, Junction City, OR; kept as a 1 mM stock solution in DMSO at -70°C). was added to PMN (1 X 10' cells/ml) at a final concentration of 2 X M. and the cells

3562 VIRUS-INDUCED CHANGES IN NEUTROPHIL PROTEIN PHOSPHORYLATION

were incubated at 25OC for 20 min. PMN were washed twice in ice- cold buffer and resuspended a t a concentration of 1 X 10' PMN/ml in HBSS with 1.6 mM CaCI, (in certain experiments 2 mM EGTA was added to the buffer to remove extracellular Ca2+). Immediately after washing PMN were exposed to PBS. non-DV. or DV for 20 min a t 25°C. with or without subsequent stimulation with FMLP (5 X 10-'M) without CB for 10 min. Fluorescence intensity was monitored continuously on a Spex Fluorolog spectrofluorometer (Metuchen. N J ) with a n excitation wavelength of 355 nm and an emission wave- length of either 410 or 490 nm. Intracellular [Ca2+] was estimated using the following formula as described by Grynkiewicz et al. (26): ICs"] = Kd (R-R,,./R,-R)(Sfz/Sba): where R = the ratio of fluores- cence of 410/490. Kd = 250 nM. R,,, = 0.12. R,, = 2.6 and Sf2/Sb2 = 2.01.

CAMP levels. cAMP levels were measured in PMN (2 X lo7 cells/ ml) exposed to non-DV, DV. or buffer for 0 to 20 min a t 25'C with or without subsequent stimulation with FMLP (1 X 1 O-' M) without CB for 0 to 1 min. At various times. aliquots of cells were removed and cAMP levels determined using a RIA kit (New England Nuclear) as described by Smolen et al. (27). This method allows reliable measurements of cAMP with time intervals as short as 5 s.

Transmembrane potential. Membrane potential changes were monitored as described by Seligmann and Gallin (28) with a spectro- fluorometric technique utilizing the cyanine fluorescent dye. Di0C5(3) (Molecular Probes). PMN (5 X 10' cells/mi) were incubated with Di0C5(3) (5 x M) at 25°C for 5 min with constant stirring: this time period allowed the fluorescence intensity to reach a steady state. Cells were then exposed to PBS. non-DV. or DV for 20 min a t 25°C. with or without subsequent stimulation with FMLP (5 X lo-' M) without CB for 10 min. Fluorescence emission was monitored on a continuous recorder attached to a Spex Fluorolog spectroflu- orometer with a n excitation wavelength of 470 nm and emission wavelength of 495 nm.

Statfstfcal eualuatlons. Statistical evaluations were done by us- ing Student's t-test.

RESULTS

Protein phosphorylation in PMN exposed to influenza virus alone. To determine what effect exposure of PMN to virus alone has on protein phosphorylation in PMN, cells were loaded with 32P, and incubated with buffer, non-DV, or DV. PMN were also stimulated with FMLP or PMA within each experiment as a positive control. Cells were then fractionated. and cytosolic and particulate fractions analyzed by SDS-PAGE. autoradiography, and scanning densitometry. Figure 1 is an autoradiogram from a representative experiment, showing the effect of influenza virus alone on cytosolic phosphoproteins. Ex- posure of cells to both non-DV (third lane) and DV (fijth lane)-stimulated phosphorylation of some proteins when compared with cytosols from buffer-treated cells (first lane).

Superimposed densitometry scans of thefirst and third lanes of Figure 1. comparing buffer and non-DV treated cytosolic fractions (Fig. 2A). show that non-DV induced phosphorylation of several mid to high m.w. proteins (primarily those peaks on the right half of the scans). Figure 2B shows scans of cytosols from buffer- and DV- treated PMN in Figure 1 (first and fifth lanes). The scans clearly show DV also causes phosphorylation of several proteins. The differences in phosphorylation induced by non-DV and DV are shown in Figure 2C, which compares scans of the representative autoradiogram lanes (Fig. 1. third andfijth lanes). Although the scans are nearly identical in the low m.w. region of the autoradiogram (left end of the scans), some mid to high m.w. proteins are phosphorylated more intensely in PMN exposed to DV compared with non-DV (Figs. 1 and 2). Similar results were obtained with particulate fractions: Le.. both DV and non-DV caused phosphorylation of some proteins in particulate fractions (data not shown).

4 c

-1 16.2

-92

-66.2

-45

-3 1

-21.5 c

- I

+ L - + , , - + , NI Non-DV DV

to influenza virus without or with subsequent stimulation with FMLP. Flgure 1. Phosphorylation of PMN cytosolic proteins in cells exposed

PMN were prelabeled with "P, treated with buffer, non-DV. or DV for 30

ated as described in Materlals and Methods. An autoradiogram from a min and then wlth IO-' M FMLP or solvent (DMSO) for 30 s. and fraction-

representative ( n = 3) SDS-PAGE of PMN cytosollc proteins is shown. An aliquot of 50 rg of protein was analyzed in each gel lane. NI. cytosols derived from noninfected [buffer-treated] PMN: Non-DV. cytosols derived from non-DV-treated PMN; DV. cytosols derived from DV-treated cells: -. cytosols from solvent [DMSO)-treated control PMN: +. cytosols from cells stimulated with IO-'M FMLP. M, markers, to the right. are in kilodaltons. Arrowhead shows the position of the 48 kDa phosphoproteln.

Protein phosphorylation of influenza virus-treated PMN induced by secondary stimulation with FMLP. The autoradiogram in Figure 1 also illustrates the effect on protein phosphorylation of PMN exposed to influenza virus followed by stimulation with M FMLP. Com- parison of each set of cells before and after FMLP stim- ulation shows that the agonist induces similar, though not identical, patterns of phosphorylation in buffer and non-DV-treated PMN cytosols (second andfourth lanes). This effect is also shown in densitometry scans (of Fig. 1) in Figure 3A. (cytosols from buffer-treated cells with- out and with FMLP stimulation). PMN pretreated with non-DV. with or without subsequent FMLP stimulation, showed equivalent or increased protein phosphorylation as compared to buffer-treated cells (Fig. 3B). In contrast, phosphorylation of cytosolic proteins in FMLP-stimulated cells was partially inhibited by pretreatment of cells with DV (Fig. 3C). One of the proteins with inhibited phos- phorylation in DV-treated PMN cytosols (arrows, Fig. 3) was the 48-kDa phosphoprotein. Similar overall effects

VIRUS-INDUCED CHANGES IN NEUTROPHIL PROTEIN PHOSPHORYLATION 3563

A A

s e C (d

0 (I) D d

L

I C

Ffgure 2. Effect of influenza virus alone on phosphorylation of cyto- sollc proteins in unstimulated PMN. Densitometry scans of the first, third, and fifth lanes of Figure 1 are superimposed to show the effect of virus

compared with those from PMN exposed to non-DV (- - - - -): B, cytosols on phosphorylation. A , cytosols from PMN exposed to buffer (-) are

from PMN exposed to buffer (-) are compared with those from PMN treated with DV (- - - - -); C. cytosols from non-DV-treated PMN (-) are compared with those from DV-treated cells (- - - - -1. Top. top of autoradi- ogram. Arrows show the position of the 48 kDa protein.

of FMLP stimulation were observed in particulate frac- tions (data not shown).

Protein phosphorylation of influenza virus-treated PMN induced by secondary Stimulation with PMA. The inhibition of protein phosphorylation in DV-treated PMN upon FMLP stimulation was a striking observation, and therefore other studies were done to determine whether a different secondary stimulus (Le., PMA) caused a simi- lar effect. A representative autoradiogram of cytosolic phosphoproteins from buffer, non-DV, and DV-treated PMN without and with 100 ng/ml PMA stimulation is shown in Figure 4. As with FMLP treatment, the patterns and degree of PMA-induced phosphorylation were similar for cytosolic proteins from buffer and non-DV-treated

p i

C

proteins in PMN exposed to influenza virus. Densltometry scans of all Figure 3. Effect of FMLP stimulation on phosphorylation of cytosolic

lanes in Figure 1 are superimposed to illustrate the effect of FMLP

buffer-treated PMN without (-) and with [- - - - -) FMLP stimulation: B. stimulation on phosphorylation of cytosolic proteins. A , Cytosols from

cytosols from non-DV treated PMN without (-) and with (- - - - - ) FMLP treatment; C . cytosols from DV-treated PMN without (-)and with

position of the 48 kDa phosphoprotein. (- - - - -1 FMLP treatment. Top. top of autoradtogram. Arrows show the

PMN (second and fourth lanes). In contrast, and in agree- ment with the results described above with FMLP stim- ulation, phosphorylation was depressed after PMA treat- ment only in cytosolic proteins from DV-treated PMN. Again, one of the phosphoproteins with inhibited phos- phorylation was the 48-kDa protein (arrowhead). These effects are illustrated clearly in the superimposed densi- tometry scans shown in Figure 5 (scans derived from autoradiogram in Fig. 4). The degree of inhibition of phosphorylation in DV-treated PMN cytosols in this ex- periment (Fig. 5 C ) was even more dramatic than that observed in the experiment described in Figure 1, in which FMLP was the stimulus. Indeed, with PMA stimu- lation, very few proteins were phosphorylated above rest-

-200

-1 16.2

-92

-66.2

15

31

-21.5

-1 4.4

- + - + - + "U

NI Non-DV DV cells to influenza vlrus and subsequent stimulation with PMA. PMN were

FLgure 4. Phosphorylatlon of PMN cytosolic proteins after exposure of

prelabeled. stlmulated wlth 100 ng/ml PMA for 30 s. and analyzed as described in the legend to Figure 1. An autoradiogram from a represent- ative experiment In = 3) is shown. An allquot of 46 fig of protein was analyzed in each gel lane. NI. Cytosols from noninfected (buffer-treated] PMN: Non-DV. cytosols from non-DV-treated PMN: DV. cytosols from DV- treated cells: -. cytosols from control PMN treated wlth solvent (DMSO): +. cytosols from PMA-stlmulated cells. M. markers. to the rlght. are in kllodaltons. Arrow shows the position of the 48 kDa phosphoproteln.

ing levels in DV-treated PMN. Particulate fractions exhib- ited similar changes in phosphorylation with PMA treat- ment (data not shown).

Studies were done to examine whether the differences noted in the amount of phosphorylation of proteins in DV- vs non-DV-treated PMN. with or without secondary stimulation, was due to different amounts of radioactivity in the various cell preparations. These studies made use of our observation that in all the gel experiments (>20) that we have done using the method described in this paper, there.had been several proteins that appeared to have constant amounts of radioactivity regardless of the stimuli the cells had been exposed to. Scans done on the autoradiograms of these particular proteins revealed that the area under the curve was similar in a given gel for each of the conditions (data not shown). These data sug- gest that differences in 32P, uptake into the cell and loading onto the various lanes of the gel does not account for the differences in protein phosphorylation in DV- treated versus non-DV-treated PMN.

A A

n I

n B

sollc proteins following exposure to influenza virus. Densitometry scans Figure 5. Effect of PMA stlmulatlon on phosphorylation of PMN cyto-

of the autoradiogram In Flgure 4 are superimposed for ease of comparison. A. Cytosols from PMN exposed to buffer wlthout (-1 and with (- - - - -) PMA stlmulatlon: E. cytosols from non-DV-treated PMN without (-) and with (- - - - -1 PMA treatment: C. cytosols from DV-treated PMN with- out (-) and with (- - - - -) PMA treatment. Top, top of autoradiogram. Arrows show the positlon of the 48 kDa phosphoprotein.

Intracellular [Ca2']. The effect of influenza virus, with or without subsequent stimulation with FMLP. on intra- cellular [Ca"'] in PMN was measured using the fluores- cent probe, Indo 1. The peak increase in intracellular [Ca"] occurred by 10 s in PMN exposed to DV or non-DV: Ca2+ levels returned to base line by 20 min. When PMN were exposed to virus for 20 min and subsequently stim- ulated with FMLP for 10 min. the maximal increase in intracellular [Ca"] occurred by 10 sec. and was still greater than control levels at 15 min.

There was a significant (p < 0.01) increase in intracel-

VIRUS-INDUCED CHANGES IN NEUTROPHIL PROTEIN PHOSPHORYLATION 3565

lular [Ca'+] in PMN treated with DV or non-DV over that observed in buffer-treated cells (Table I). When PMN were preincubated with DV for 20 min and then stimulated with FMLP, intracellular [Ca"] increased significantly ( p < 0.01) over levels seen in non-DV or buffer-treated PMN. The estimated peak intracellular [Ca"'], calculated by using the formula of Grynkiewicz et al. (261 is shown for PMN with or without virus and FMLP treatment [Table I). When EGTA was preincubated with PMN before treat- ment with virus or FMLP, the immediate [( 10 s) rise in intracellular [Ca"] upon subsequent exposure to virus alone or after subsequent stimulation with FMLP was decreased 15 to 25%.

cAMP levels. Both DV and non-DV caused very little increase in cAMP levels in PMN, as measured with RIA. cAMP concentrations in response to FMLP stimulation of PMN preincubated with virus or buffer for 20 min was measured at 0 , 5 . 10, 15.20.30 and 60 s. The maximum cAMP level occurred within 15 s, and there was no sig- nificant difference between control and virus-treated cells at any of the time points (data not shown).

Transmembrane potential. The effect of influenza vi- rus or buffer on transmembrane potential of PMN was measured by monitoring fluorescence intensity in PMN by using DiOC5(3) (28). The peak changes in fluorescence intensity induced by non-DV or DV occurred within 10 s and returned to base line by 1 min (Fig. 6). The peak change in fluorescence intensity was greater in PMN exposed to DV than to non-DV or buffer. However, no differences in the fluorescence intensity changes were seen when PMN were incubated with non-DV, DV, or buffer for 20 min and then stimulated with FMLP for 10 min.

DISCUSSION

To our knowledge, this report constitutes the first ob- servation of early (i.e-, 30-min postexposure) alterations in cellular protein phosphorylation induced in immune cells by viruses. In experiments described here, several proteins underwent phosphorylation (few, if any, were dephosphorylated] when PMN were exposed to non-DV or DV alone for 30 min. Subsequent stimulation of these cells with FMLP or PMA resulted in similar patterns of phosphorylation in buffer and non-DV-treated PMN. In contrast, DV-treated PMN showed inhibited levels of pro-

TABLE I Peak Lntracellular [Ca2+j in resting and stlmulated PMN

Stimulant Fluorescence Ratio at 10 Peak [Caz+lb

410/490 nm 1 S t 2nd S a (nM)

Buffer None 0.8 f 0.10 190 DV None 1.3 ? 0.06' 456 Non-DV None 1.2 f 0.06' 388

Buffer DV

FMLP FMLP

1 . 3 f 0.05 388 1.5 f 0.05d 630

Non-DV FMLP 1.3 f 0.06 388 "Peak rises in intracellular [Ca2'1 occurred at the 10-s time point.

Incubation conditions and methods for measuring intracellular [Ca"+-] were as described in Materials and Methods. Data are the means f SEM for four experiments.

kfewicz et al. (26). Peak [Ca2'1 was estimated by using the formula described by Gryn-

buffer. ' p < 0.01 when cells treated with DV are compared with non-DV or

non-DV or buffer plus FMLP. p < 0.01 when cells treated with DV plus FMLP are compared with

J ~ m e ( m i d

Figure 6. Transmembrane potential changes in buffer and influenza virus-treated PMN before and after FMLP stimulation. The kinetics of the

non-DV (-1. or DV (a"0) for 20 min followed by FMLP treatment transmembrane potential changes in PMN exposed to buffer [A-A),

for 10 mln are shown. DiOC,(3) was used to measure transmembrane potential changes as described in the Materlals and Methods. Data are the mean f SEM (bars) from four experiments. At 0 .1 min. the fluores- cence intensity was greater for DV than for buffer or non-DV ( p < 0.05

time point. and 0.07. respectively). No significant differences were noted at any other

tein phosphorylation upon stimulation with either ago- nist. To date, the biophysical difference between DV and non-DV has not been elucidated, however, the use of non-DV in these experiments provides a useful control for determining if the effect of influenza virus on a particular PMN activation step might be responsible for the depression of end-stage functions. For example, if the changes in protein phosphorylation that occurred upon secondary stimulation of DV and non-DV pretreated PMN were identical, then a cause-effect relationship be- tween altered protein phosphorylation and depressed end-stage PMN functions would be very unlikely. Al- though DV-induced inhibition of phosphorylation corre- lates nicely with depression of PMN end-stage functions, a direct cause-effect relationship remains to be proven.

It is possible that DV alters phosphorylation of certain regulatory proteins, which could then suppress cellular responses to subsequent stimulation. For example, virus- induced changes in phosphorylation of N, proteins could cause inhibition of end-stage responses similar to that seen in DV-treated cells. Specifically, studies have shown that pertussis toxin, which inhibits the N, protein in PMN, causes depression of FMLP-induced chemotactic, secre- tory, and oxidative functions in PMN, while not inhibitlng phagocytosis (29-34). DV also inhibits chemotactic, se- cretory, and oxidative functions in vitro, but does not alter the phagocytic response (3-7). Inhibited phos- phorylation of N,-type proteins is unlikely to be the com- plete explanation for depression of end-stage functions by DV, because the depressed phosphorylation occurs both with FMLP and PMA treatment, whereas FMLP (reviewed in Ref. 35). but not PMA (29, 33, 36). has been associated with N, protein-mediated responses. Addition- ally, the role of phosphorylation in controlling N, protein- mediated steps remains speculative.

The 48-kDa phosphoprotein is another possible regu- latory protein through which DV could interfere with PMN end-stage functions, particularly actlvation of the respiratory burst. (We can identify this protein in our

3566 VIRUS-INDUCED CHANGES IN NEUTROPHIL PROTEIN PHOSPHORYLATION

autoradiograms because of experiments recently reported in which PMN from a CGD patients failed to phosphoryl- ate this protein, both in a cell-free system and in intact PMN (see footnote 4).) The finding that phosphorylation of the 48-kDa protein in DV-treated cells, in response to FMLP or PMA treatment, is inhibited is particularly in- teresting, since this protein is defectively phosphorylated in PMN from CGD patients. Inasmuch as the only func- tional defect described in PMN from CGD patients is the inability to mount a respiratory burst, this phosphopro- tein has recently been regarded as potentially important in regulation of oxidative metabolism in normal PMN. However, a definitive role for the 48-kDa phosphoprotein, or any other phosphoprotein, in the regulation of a spe- cific PMN end-stage response remains to be proven.

Our data are consistent with the idea that this protein may participate in regulation of the respiratory burst. The evidence for this is that I ) exposure of PMN to either virus stimulated some phosphorylation of this protein, an event accompanied by a respiratory burst (4, 5, 19); 2) non-DV-treated PMN, when stimulated by FMLP or PMA. underwent normal phosphorylation of this protein under conditions which also elicit normal respiratory burst activation (5, 37); and 3) in contrast, DV-treated PMN, upon stimulation, exhibit depressed respiratory burst activity (3-5, 37) and also show decreased phos- phorylation of this protein. The inhibited phosphoryla- tion of multiple cellular proteins by DV suggests that: 1) DV may depress PMN end-stage functions through alter- ations in phosphorylation of several regulatory proteins; and 2) the actual step(s) at which the virus interferes with PMN functions occurs prior to protein phosphoryl- ation.

The effect of DV on intracellular [Ca"], cAMP levels, and transmembrane signaling was also examined, be- cause various studies suggest that these steps are impor- tant in regulating cellular protein phosphorylation (re- viewed in Refs. 38 and 39). PMN have both Ca2+- and CAMP-dependent protein kinases which mediate phos- phorylation of multiple proteins upon cell stimulation. For example, recent studies suggest that protein kinase C is involved in the selective phosphorylation of PMN proteins stimulated with FMLP or PMA (40). and that Ca"' participates in the regulation of this kinase by in- ducing its translocation from the cytosol to the mem- brane (41, 42). The activation of protein kinase C and other kinases appears to be dependent upon transmem- brane signalling that occurs when a ligand binds to its receptor (38, 39).

Studies were done to determine if DV interferes with the rise in intracellular [Ca"'], which occurs when PMN are stimulated with FMLP (discussed in Refs. 43 and 44). The finding that virus alone caused a rise in intracellular [ca"'] was not surprising, since influenza virus can ini- tiate an oxidative burst in PMN (4, 5, 19). However, the data showing increased intracellular [Ca"'] in FMLP- stimulated cells preincubated with DV, as compared with non-DV or buffer, were unexpected. Although it is pos- sible that a marked increase in intracellular [Ca"+] could inhibit cell activation, we are unaware of evidence in the literature which supports this hypothesis. Additionally, the estimated threefold increase of intracellular [Ca"] above base line in cells treated with DV followed by FMLP is within the range reported by other investigators for

cells stimulated with FMLP (44, 45). The enhanced rise in intracellular [Ca"+] in cells preincubated with DV was not due to a higher base line [Ca"'], because cells stimu- lated with either virus have equivalent intracellular [Ca"'] by 20 min (Table I). The initial rise in intracellular [Ca"] (i.e., the increase which occurs within the first 10 s) caused by virus alone or by virus plus FMLP came mainly from internal sources, inasmuch as addition of EGTA to the cell suspension caused only a small decrease in the rise of intracellular [Ca2+] upon cell stimulation. Similar findings have been reported for PMN stimulated with FMLP (45). Sawyer et al. (46) have recently shown that, upon stimulation of PMN with opsonized zymosan, there are regional differences in the subcellular localization of Ca2+. We plan to do similar studies to determine what role, if any, the augmented rise in intracellular [Ca"'] has in DV-induced disruption of PMN functions.

An increase in cAMP levels in response to stimulation of PMN with FMLP does not appear to be a fundamental requirement for activating the cell (47), but may have a role in modulating the cell's response to FMLP (47.48) and in activating CAMP-dependent protein kinases. Ex- posure of PMN to virus alone had very little effect on cAMP levels. Upon subsequent stimulation of these cells with FMLP, both the kinetics and degree of increase in cAMP levels were similar in buffer and virus-treated PMN, suggesting that virus was not causing PMN dys- function by altering CAMP levels.

PMN transmembrane potential changes occur before end-stage functions, and are thought to be due to ionic fluxes across cell membranes (28, 49). The initial (0.1 to 0.2 min) change in transmembrane potential was greater in PMN exposed to DV than in cells treated with non-DV or buffer. However, when cells preincubated with virus or buffer were subsequently stimulated with FMLP, the responses did not differ significantly. One explanation for the finding that DV caused a greater initial change in transmembrane potential could be that the virus induces permeability changes in PMN membranes. A precedent for this is that Sendai virus, which is structurally similar to influenza virus, is known to cause leakage of intracel- lular metabolites from cells (50). Although influenza vi- rus was not found to cause such membrane permeability changes (50), those studies were not done using PMN; thus, depending on cell type, permeability changes might still be possible. Our own data using PMN preincubated with carboxyfluorescein diacetate or carboxy-4' ,5'-di- methylfluorescein diacetate, followed by exposure to in- fluenza virus did not, however, show leakage of these dyes out of the cell (our unpublished data). Regardless of the mechanism by which DV induces the change in trans- membrane potential, the data from experiments by using FMLP as a secondary stimulus suggest DV is not depress- ing PMN end-stage functions by interfering with trans- membrane potential changes.

The data obtained in this study show that both DV and non-DV induce protein phosphorylation in PMN, but that only DV suppresses subsequent phosphorylation upon treatment with a secondary stimulus. Although cAMP levels and transmembrane potential changes do not ap- pear to be causing this effect, the role of intracellular [Ca"'] is still unclear. Further work is also needed to determine 1) which, if any, of the proteins which undergo altered phosphorylation are important in causing abnor-

VIRUS-INDUCED CHANGES IN NEUTROPHIL PROTEIN PHOSPHORYLATION 3567 mal PMN end-stage responses, and 2) the mechanism by tection of 32P and lZ51 using intensifying screens and hypersensitized

film. FEES Lett. 82:314.

of Ca2+ indicators with greatly improved fluorescence properties. J .

Acknow'edgments* We thank Drs' David Bass and 27. Smolen. J. E., H. M. Korchak, and G . Weissman. 1980. Increased Biol. Chem. 260:3440.

levels of cyclic adenosine-3',5'-monophosphate in human polymor- phonuclear leukocytes after surface stimulation. J . Clin. Invest.

28. Seligmann. B. E., and J. I. Gallin. 1980. Use of lipophilic probes of 65: 1077.

which DV inhibits cellular protein phosphorylation. 26. Grynkiewicz. G., M. Poenie, and R. Y. Tsien. 1985. A new generation

Douglas Lyles for helpful discussions, Mrs. Jannell Rowe and Mrs. Lisa Winkler for technical assistance. and MrsSandy Alexander for secretarial assistance.

1

2.

3.

4.

5.

6.

7.

8.

9.

10.

11.

12.

13.

14.

15.

16

17

18.

19.

20.

21.

22.

23.

24.

25.

REFERENCES

Larson, H. E., and R. Blades. 1976. Impairment of human polymor-

Ruutu. P., A. Vaheri, and T. U. Kosunen. 1977. Depression of phonuclear leucocyte function by influenza virus. Lancet 1:283.

human neutrophil motility by influenza virus in vitro. Scand. J . "

Immunol. 6~897. Abramson, J. S. . E. L. Mills, G . S . Giebink, and P. G. Quie. 1982 Depression of monocyte and polymorphonuclear leukocyte oxidative metabolism and bactericidal capacity by influenza A virus. Infect. Immun. 35:350. Abramson, J. S., J. C. Lewis. D. S . Lyles. K. A. Heller, E. L. Mills. and D. A. Bass. 1982. Inhibition of neutrophil lysosome-phagosome fusion associated with influenza virus infection in vitro: role in

Abramson, J. S.. D. S. Lyles, K. A. Heller, and D. A. Bass.1982. depressed bactericidal activity. J . Clin. Invest. 69:1393.

Influenza A virus induced polymorphonuclear leukocyte dysfunction. Infect. Immun. 37:794. Abramson. J. S., J. W. Parce, J. C. Lewis. D. S. Lyles, E. L. Mills. R. D. Nelson, and D. A. Bass. 1984. Characterization of the effect of

sponses. Blood 64: 131. influenza virus on polymorphonuclear leukocyte membrane re-

Abramson. J. s., J. G. Wheeler, J. W. Parce. M. J. Rowe, D. s. Lyles, M. Seeds, and D. A. Bass. 1986. Suppression of endocytosis

Abramson. J. S., L. F. Cassidy, and D. S. Lyles. 1987. Synthesis of in neutrophils by influenza A virus in vitro. J. Infect. Dis. 154:456.

influenza A virus and endogenous proteins in polymorphonuclear leukocytes. Pediatr. Res. 21:308A [Abstr.). Schneider, C., M. Zanetti, and D. Romeo. 1981. Surface-reactive stimuli selectively increase protein phosphorylation in human neu-

Andrews, P. C., and B. M. Babior. 1983. Endogenous protein phos- trophils. FEES Lett. 127:4.

phorylation by resting and activated human neutrophils. Blood 61:333. Huang, C. K., J. M. Hill, Jr.. B.-J. Bormann. W. M. Mackin, and E. L. Becker. 1983. Endogenous substrates for cyclic AMP-dependent and calcium-dependent protein phosphorylation in rabbit peritoneal neutrophils. Biochim. Biophys. Acta 760: 126. Andrews, P. C., and B. M. Babior. 1984. Phosphorylation of cytosolic proteins by resting and activated human neutrophils. Blood 64:883. White, J. R.. C.-K. Huang, J. M. Hill, Jr.. P. H. Naccache. E. L.

acetate and its analogue 4-a-phorbol 12.13-didecanoate on protein Becker, and R. I. Sha'afi. 1984. Effect of phorbol 12-myristate 13-

phosphorylation and lysosomal enzyme release in rabbit neutrophils. J. BioL Chem. 259:8605. Okamura. N.. S. Ohashi, N. Nagahisa. and S . Ishibashi. 1983. Changes in protein phosphorylation in guinea pig polymorphonu- clear leukocytes by treatment with membrane-perturbing agents which stimulate superoxide anion production. Arch. Biochem. Bio-

Segal, A. W.. P. G . Heyworth, s. Cockcroft, and M. M. Barrowman. p h y s . 228:270.

1985. Stimulated neutrophils from patients with autosomal recessive chronic granulomatous disease fail to phosphorylate a M,-44,000 protein. Nature 316:547. Ishii, E., K. Irita. I. Fujita, K. Takeshige, M. Kobayashi, T. Usui. and K. Ueda. 1986. Protein phosphorylation of neutrophils from normal children and patients with chronic granulomatous disease.

Hayakawa, T., K. Suzuki. S. Suzuki, P. C. Andrews, and B. M. Eur. J. Pediatr. 145~22.

activation of the respiratory burst in human neutrophils. Evidence Babior. 1986. A possible role for protein phosphorylation in the

from studies with cells from patients with chronic granulomatous disease. J . Biol. Chem. 261 :9109. Heyworth, P. G.. and A. W. Segal. 1986. Further evidence for the involvement of a phosphoprotein in the respiratory burst oxidase of human neutrophils. Bfochem. J. 239:723. Mills, E. L.. Y. Debets-Ossenkopp, H. A. Verbrugh, and J. Verhoef. 1981. Initiation of the respiratory burst of human neutrophils by influenza vlrus. Infect. Immun. 32:1200. Lowry. D. H., N. J. Rosebrough, A. L. Farr, and R. J. Randall. 195 1, Protein measurements with the Folin phenol reagent. J. Biol. Chem. 193:265. Amrein. P. C.. and T. P. Stossel. 1980. Prevention of degradation of human polymorphonuclear leukocytes proteins by diisopropylfluo-

Peterson, G . L. 1977. A simplification of the protein assay method rophosphate. Blood 46:442.

of Lowry et al . which is more generally applicable. Anal. Biochem. 83:346. Laemmli. U. K. 1970. Cleavage of structural proteins during assem- bly of the head of bacteriophage T4. Nature 227:680. Wray. W.. T. Boulikas, V. P. Wray, and R. Hancock. 1981. Silver staining of proteins in polyacrylamide gels. Anal. Biochem. 1 18: 197. Laskey, R., and A. D. Mills. 1977. Enhanced autoradiographic de-

membrane potential to assess human neutrophil activation. Abnor- mality in chronic granulomatous disease. J. CIin. Invest. 66:493.

29. Becker, E. L.. J. C. Kermode, P. H. Naccache, R. Yassin. M. L. Marsh, J. J. Munoz, and R. I. Sha'afi. 1985. The inhibition of neutrophil granule enzyme secretion and chemotaxis by pertussis toxin. J . Cell Biol. 100: 1641.

30. Lad. P. M.. C. V. Olson, I. S . Grewal, and S . J. Scott. 1985. A pertussis toxin-sensitive GTP-binding protein in the human neutro- phil regulates multiple receptors, calcium mobilization, and lectin- induced capping. Proc. Natl. Acad. Sci. USA 82:8643.

3 1. Krause, K.-H., W. Schlegel, C. B. Wollheim. T. Anderson. F. A. Waldvogel. and P. D. Lew. 1985. Chemotactic peptide activation of human neutrophils and HL-60 cells. Pertussis toxin reveals correla- tion between inositol trisphosphate generation, calcium ion tran-

32. Verghese, M. W.. K. Fox, L. C. McPhail. and R. Snyderman. 1985. sients. and cellular activation. J . Clin. Invest. 76: 1348.

Chemoattractant-elicited alterations of CAMP levels in human poly- morphonuclear leukocytes require a Ca2+-dependent mechanism which is independent of transmembrane activation of adenylate cyclase. J. Biol. Chem. 260:6769.

33. Barrowman. M. M., S . Cockcroft. and B. D. Gomperts. 1986. Two roles for guanine nucleotides in the stimulus-secretion sequence of

34. Lad. P. M., C. V. Olson. and I. S . Grewal. 1986. A step sensitive to neutrophils. Nature 319:504.

pertussis toxin and phobol ester in human neutrophils regulates

35. Snyderman, R.. C. D. Smith, and M. W. Verghese. 1986. Model for chemotaxis and capping but not phagocytosis. FEBS Lett. 200:91.

leukocyte regulation by chemoattractant receptors. roles of a guanine nucleotide regulatory protein and polyphosphoinositide metabolism. J . Leukocyte-Biol. 401785.

. " .

36. Verghese, M. W.. C. D. Smith, and R. Snyderman. 1985. Potential

receptor mediated polyphosphoinositide metabolism, Ca'+ mobiliza- role for a guanine nucleotide regulatory piotein in chemoattractant

Commun. 127:450. tion and cellular responses by leukocytes. Biochem. Biophys. Res.

37. Abramson, J. S., G. L. Wiegand, and D. S . Lyles. 1985. Neuramin-

dysfunction. J . Clin. Mlcrobtol. 22: 129. idase activity is not the cause of influenza virus-induced neutrophil

38. Nishizuka, Y. 1984. The role of protein kinase C in cell surface signal

39. Ashendel, C. L. 1985. The phorbol ester receptor: a phospholipid- transduction and tumour promotion. Nature 308:693.

40. Pontremoli. S., E. Melloni, F. Salamino, B. Sparatore. M. Michetti, regulated protein kinase. Biochim. Biophys. Acta 822:219.

0 Sacco, and B. L. Horecker. 1986. Phosphorylation of proteins in human neutrophils activated with phorbol myristate acetate or with

41. May, W. S.. Jr.. N. Sahyoune. M. Wolf, and P. Cuatrecasas. 1985. chemotactic factor. Arch. Biochem. Biophys. 250:23.

Role of intracellular calcium mobilization in the regulation of mem- brane processes. Nature 31 7:549.

42. Wolf, M.. H. Levine 111. W. S . May, Jr.. P. Cuatrecasas, and N. Sahyoune. 1985. A model for intracellular translocation of proteln kinase C involving synergism between CaZ' and phorbol esters. Nature 31 7:546.

43. Pozzan, T.. D. P. Lew. C. B. Wollheim, and R. Y. Tsien. 1983. Is

221:1413. cytosolic ionized calcium regulatlng neutrophil activation? Science

44. White. J. R.. P. H. Naccache, T. F. P. Molski. P. Borgeat, and R. I. Sha'afi. 1983. Direct demonstration of increased intracellular con-

stimulation by chemotactic factor. Biochem. Biophys. Res. Com- centration of free calcium in rabbit and human neutrophils following

mun. 113:44. 45. Sklar. L. A., and 2. G . Oades. 1985. Signal transduction and Hgand-

receptor dynamics in the neutrophil. Ca2+ modulation and restora- tion. J . Biol. Chem. 260: 1 1468.

46. Sawyer, D. W.. J. A. Sullivan, and G. L. Mandell. 1985. Intracellular free calcium localization in neutrophils during phagocytosis. Science 2303363.

47. Simchowitz, L.. I. Spilberg. and J. P. Atkinson. 1983. Evidence that the functional responses of human neutrophils occur independently

Protein Phosphor. Res. 9:35. of transient elevations in cyclic AMP levels. J . Cyclic Nucleotide

48. Grady, P. G.. and L. L. Thomas. 1986. Characterization of cyclic- nucleotide phosphodiesterase activities in resting and N-formyl-

Biochfm. Biophys. Acta 885:282. methionyl-leucylphenylalanine-stimulated human neutrophils.

49. Kitagawa. S., and R. B. Johnston, Jr . 1985. Relationship between membrane potential changes and superoxide-releasing capacity in resident and activated mouse peritoneal macrophages. J. Immunol. 135:3417.

50. Foster, K. A.. K. Gill, K. J. Micklem, and C. A. Pasternak. 1980. Survey of virally mediated permeability changes. Biochem. J . 190:639.

THE JOURNAL OF BIOLOGICAL CHEMISTRYVol. 256, No. 10, Issue of May 25, pp. 4838-4842, 1981Printed in U. S.A.

Interaction of Sendai Virus Proteins with the Cytoplasmic Surface ofErythrocyte Membranes following Viral Envelope Fusion*

(Received for publication, October 30, 1980)

Susan E. Caldwell and Douglas S. LylesFrom the Department ofMicrobiology and Immunology, The Bowman Gray School ofMedicine of Wake Forest University,Winston-Salem, North Carolina 27103

Radiolabeled Sendai virus was allowed to fuse withhuman erythrocytes and inside out vesicles were pre-pared from the erythrocyte membranes. Viral proteinsincorporated into inside out vesicles were releasedfrom the membrane following treatment with variousagents which perturb protein structure. Disruptionproducts were analyzed by sucrose gradient centrifu-gation and sodium dodecyl sulfate-polyacrylamide gelelectrophoresis to identify the viral proteins. The re-sults indicated that 6 M guanidine, 40 mm lithium diio-dosalicylate (pH 11-13), and 4 M potassium thiocyanateremoved viral nucleocapsid and M proteins from thevesicles. Urea (6.0 M) and 5 mi p-chloromercuriben-zenesulfonate removed only viral nucleocapsid pro-teins. KCI (1.0 or 0.1 M), 5 or 0.5 mm sodium phosphate,10 mm EDTA (pH 10), 10 mm lithium diiodosalicylate, 1M urea, 0.5 mw ATP, or freeze-thaw cycles did notrelease any viral proteins. By contrast, treatment ofinside out vesicles with Triton X-100 solubilizes thelipid bilayer, releasing viral integral membrane glyco-proteins and leaving viral nucleocapsids intact. Nucleo-capsids isolated from virions do not adsorb nonspecif-ically to control inside out vesicles. These data implythat Sendai viral nueleocapsid and M proteins interactextensively with the cytoplasmic surface of infectedcell membranes, in the manner ofperipheral membraneproteins, whereas the glycoproteins, HN and F1, behaveas integral membrane proteins.

Sendai virus, a parainfluenza virus, penetrates into the hostcell by a mechanism which probably involves fusion of theviral envelope with the membrane of the host cell (discussedin Scheid and Choppin, 1976). Sendai virus has two glycopro-teins on the surface of the envelope. One, designated HN,possesses both hemagglutinating and neuraminidase activitiesand allows the virus to attach to receptors on the host cellmembrane (Tozawa et al., 1973;, Scheid and Choppin, 1974).The other glycoprotein, F, enables the attached virus to fusewith the host cell (Homma and Ohuchi, 1973; Scheid andChoppin, 1974). Virions grown in most tissue culture cellscontain F in the form of an inactive precursor, F0, which canbe activated in vitro by trypsinization. Sendai virus alsocontains a nonglycosylated envelope protein, M, the mem-

* This research was supported by a grant from the North CarolinaUnited Way and Forsyth Cancer Service and by Grant AI 15892 fromthe National Institutes of Health. Preliminary results of this workwere presented at the 1980 meeting of the American Society forMicrobiology (Caldwell, S., and Lyles, D. S. (1980) Abstr. Annu. Meet.Am. Soc. Microbiol. 265). The costs of publication of this article weredefrayed in part by the payment of page charges. This article musttherefore be hereby marked "advertisement" in accordance with 18U.S.C. Section 1734 solely to indicate this fact.

brane or matrix protein, which coats the inner surface of theviral envelope. The viral nucleocapsid proteins include NP,the major nucleocapsid protein, and P and L, probably sub-units of the virion RNA polymerase (nomenclature discussedin Scheid and Choppin, 1974, 1977).

Sendai virus fuses readily with erythrocytes, thus incorpo-rating viral proteins into the erythrocyte membrane (Bachi etal., 1973; Buechi and Bachi, 1979). Virus envelope fusion witherythrocyte membranes provides a model for virus penetra-tion. The advantage in using erythrocyte membranes is thatthey can be made into inside out vesicles containing viralproteins. These IO1 vesicles have all of the internal viralproteins on their external (cytoplasmic) surface (Lyles, 1979),where the mechanisms of membrane attachment of theseproteins can be investigated. The associations characterizedin the erythrocyte model should also be relevant to themechanism of virus maturation by budding from the hostplasma membrane. In the process of budding, the viral gly-coproteins are inserted into the host membrane biosyntheti-cally (rather than by fusion); the nucleocapsid then associateswith areas of membrane containing viral envelope componentsprior to evagination of the membrane to form the maturevirion (Choppin and Compans, 1975).The forces binding proteins to cell membranes can be

characterized in terms of the difficulty of removal of suchproteins from the membranes. Membranes may be treatedwith various agents which perturb membrane structure butleave the lipid bilayer relatively intact. The proteins removedfrom the membranes by these agents are functionally definedas peripheral membrane proteins. This technique has beendeveloped by Steck and others in studies of erythrocyte mem-brane proteins (reviewed by Steck, 1974a). Some peripheralproteins are easily eluted, some are more difficult to elute, andthe integral membrane proteins do not elute from the mem-brane without disrupting the lipid bilayer. In the presentwork, it is shown that the Sendai virus nucleocapsid and Mproteins behave as peripheral membrane proteins in the eryth-rocyte membrane, while HN and F behave as integral mem-brane proteins.

MATERIALS AND METHODS

Virus and Cells-The Z strain of Sendai virus was grown in MDBKcells with Dulbecco's minimal essential medium supplemented with10% newborn calf serum (Flow Laboratories) in the presence of[35S]methionine (10OCi/ml, New England Nuclear) (1 Ci = 3.7 x 1010Bq) and purified as described (Lamb and Choppin, 1977). The F.glycoprotein was activated with tosylphenylalanyl chloromethyl ke-tone-treated trypsin (5 ,ug/ml; trypsin-TPCK, Worthington) for 10min at 37 'C as described (Scheid and Choppin, 1974).

Erythrocyte Membranes Containing Viral Proteins-The fusionof radiolabeled Sendai virus with human erythrocytes and the pro-

'The abbreviations used are: IO, inside out; SDS, sodium dodecylsulfate.

4838

by on May 29, 2009

ww

w.jbc.org

Dow

nloaded from

Sendai Virus Protein Interaction with Inside Out Vesicles

duction of IO vesicles from these membranes has been described indetail (Lyles, 1979) and is briefly discussed under "Results."

Elution Procedure-IO vesicles containing viral proteins werepelleted for 1 h at 15,000 rpm in a Sorvall SS 34 rotor. Each resultingpellet was resuspended in 0.5 ml of a given eluting agent in 5 mmsodium phosphate, pH 8.0 (58P buffer), unless otherwise specified,and incubated on ice for 30 min. The preparation was then furtherdiluted with 0.5 ml of buffer to reduce the density of concentratedsolutions and was layered over a linear gradient of 15-45% (w/w)sucrose over a 60% sucrose cushion, all in 10 mM phosphate buffer,pH 7.4. The gradients were centrifuged at 40,000 rpm for 2 h in anSW 50L rotor. Each gradient was then fractionated in 20-drop frac-tions; 25-p1l aliquots of each fraction were counted by liquid scintilla-tion counting and peak fractions were pooled. These pooled fractionswere precipitated with 10% (w/w) trichloroacetic acid (Fisher) after100 pl of bovine serum albumin (1 mg/ml) was added to each as acarrier. Samples were spun at 2,000 rpm for 30 min and then washedwith 10% (w/w) trichloroacetic acid and finally with butanol (Fisher).Samples were evaporated to dryness under a thin stream of nitrogen.

Nucleocapsid Purification and Association with IO Vesicles-Radiolabeled virus was added to 2% Triton X-100 in 10 mM sodiumphosphate (pH 7.4) and potassium chloride (0.0, 0.1, or 1.0 M) andincubated on ice for 30 min. The preparations were then spun throughdiscontinuous gradients consisting of 2 ml each of 15 and 45% sucroseand 0.3 ml of 60% sucrose (all w/w) in an SW 50L rotor at 40,000 rpmfor 2 h. The bottom 1.0 ml of each gradient was collected and dialyzedovernight in 10 mm phosphate buffer, pH 7.4.

Nucleocapsid preparations (0.75-3.0 ml) were then added to apellet of IO vesicles which did not contain viral protein. The mixturewas incubated on ice 5 min and then spun through a linear 15-45%(w/w) sucrose gradient with a 60% sucrose cushion for 2.5 h at 40,000rpm in an SW 50L rotor. Gradients were fractionated in 20-dropfractions; 25-p1 aliquots were counted by liquid scintiUation counting,and peak fractions were pooled and precipitated with trichloroaceticacid as described for the elution procedure. SDS-polyacrylamide gelelectrophoresis was performed with the samples as described below.

SDS-Polyacrylamide Gel Electrophoresis-The pooled, trichlo-roacetic acid-precipitated samples were electrophoresed on 7.5% poly-acrylamide slab gels containing SDS as described (Laemrnli, 1970).The gels were processed for fluorography as described (Bonner andLaskey, 1974) and exposed to Kodak SB-5 x-ray film.

RESULTS

Radioactive Sendai virus was allowed to fuse with humanerythrocytes, thus incorporating viral proteins into the cellmembranes. The erythrocytes were lysed in 5 mi sodiumphosphate, pH 8.0. IO vesicles were prepared by incubationin low ionic strength buffer, which causes the membrane toinvaginate, followed by gentle homogenization to generate IOvesicles from the invaginations. The IO vesicles were purifiedfrom unsealed membrane fragments by centrifugation on dex-tran gradients, which separates membranes based on theirosmotic properties (Steck, 1974a, 1974b). These vesicles havebeen shown to have the internal viral proteins on their exter-nal (cytoplasmic) surface (Lyles, 1979). The extent of associ-ation of these viral proteins with the vesicle membrane can beexamined by treating vesicles with a variety of perturbantswhich disrupt membrane protein structure, leaving the lipidbilayer intact. The IO vesicles were treated with variouseluting agents, and the disruption products were analyzed bysedimentation in sucrose gradients (Fig. 1).Some treatments which elute erythrocyte membrane pro-

teins (Steck and Yu, 1973) did not elute viral proteins (Fig. 1,A and B), such as 1 M KCI or 3 freeze-thaw cycles in 5 mMsodium phosphate, pH 8.0. JO vesicles treated with theseagents sedimented to density equilibrium in the middle of thegradient similar to the control, untreated vesicles. Other re-agents, such as 4 M potassium thiocyanate, 40 mm lithiumdiiodosalicylate, 6 M urea, and 5 mM p-chloromercuriben-zenesulfonate eluted viral proteins from the membrane (Fig.1, C-F). Slowly sedimenting proteins released from the mem-brane remained at the top of the gradient, while other proteinsremained associated with the vesicle fraction. Of all the re-

25.

20-

5.

6

*0 0 0 0/t/-1Q... 0 0.. o.....

0

68 6

26 6 0

4 ~~~~~~~~4

3 5 7 9 11 13 3 5 7 9 I, 13FRACTION NO. top FRACTION NO. top

FIG. 1. Sedimentation analysis of disruption products frominside out vesicles containing Sendai viral proteins. Inside outvesicles containing Sendai viral proteins labeled with [nS]methioninewere incubated in the indicated solutions and centrifuged on a sucrosegradient as described under "Materials and Methods." The gradientswere collected from the bottom. The radioactivity of an aliquot ofeach fraction was determined by liquid scintillation counting and wasplotted on the ordinate. O-- -0, control vesicles incubated in 5 mMsodium phosphate, pH 8.0; * -*, vesicles incubated in: A, 1.0 MKCI; B, 5 mm sodium phosphate, then frozen and thawed 3 times; C,4 M KSCN; D, 40 mM lithium diiodosalicylate; E, 5 mM p-chloromer-curibenzenesulfonate; F, 6 M urea.

agents tested, none released rapidly sedimenting material suchas, for example, intact viral nucleocapsids.Peak fractions from Fig. 1 were pooled and analyzed by

SDS-polyacrylamide gel electrophoresis with fluorography todetect the labeled viral proteins, as shown in Figs. 2 and 3.Note that, as shown previously, each viral structural proteinis present in control 10 vesicles. In addition, a proteolyticfragment of NP, designated NP', is occasionally observed bothin virions and in IO vesicles, presumably due to low levels ofendogenous proteases in the preparation (Lyles, 1979). Asshown in Fig. 2, vesicles treated with 40 mm lithium diiodo-salicylate, 6 M guanidine, 4 M KSCN and high pH (pH 12-13)had little nucleocapsid (NP) or M proteins, which were foundin the top fractions of the gradient. Viral glycoproteins (HNand F1) remained associated with the vesicle fraction aftertreatment, as is typical of integral membrane proteins. Treat-ment of vesicle preparations with high pH could conceivablyhydrolyze phospholipids in the membrane; however, Steckand Yu (1973) have shown that this does not occur with brieftreatments at low temperature.

Electrophoretic analysis of IO vesicles containing viral pro-teins treated with 5 mmp-chloromercuribenzenesulfonate and6 M urea is shown in Fig. 3. These agents removed only viralnucleocapsid proteins, leaving the M protein associated withthe vesicle fraction. Lower concentrations of reagents de-scribed in Figs. 2 and 3, such as 10 mM lithium diiodosalicylate,1 M urea, or pH 10, did not elute any viral proteins under thesame experimental conditions (data not shown). Electropho-

4839

3-

0,A

2-9

.cl

A

by on May 29, 2009

ww

w.jbc.org

Dow

nloaded from

Sendai Virus Protein Interaction uith Inside Out Vesicles

4.-44 ,--

C V T V T

GUAN KSCN

NP- __Fi how** IS

no

S

*:~

V T

LISV T

pH 12

FIG. 2 (left). Reagents that remove both nucleocapsid and mproteins from inside out vesicles containing Sendai viral pro-teins. Inside out xesicles containing Sendai viral proteins labeledwith [PSjmethionine were incubated with the indicated reagents, andsubjected to sucrose gradient centrifugation as in Fig. 1. Peak fractionswere pooled, acid-precipitated, and subjected to SDS-polxacr\llam,,ldegel electrophoresis. Fluorographs of the dried gels are shown. C,control vesicles incubated in 5 mM sodium phosphate, pH 8.0. V,pooled fractions containing treated vesicles; T, pooled fractions nearthe top of the gradient; GUAN, 6 M guanidine; KSCN, 4 M KSCN;LIS, 40 mM lithium diiodosalicylate; pH 12, 50 mM glycine, pH 12;pH

retic analvsis of vesicles treated with agents that do not releasevliral proteins confirmed that the content of viral proteins inthe vlesicles was unchanged compared to untreated vesicles.Quantitation of the amount of NP or M remaining associatedwith the vesicle fraction following elution with the agentsdescribed in Figs. 2 and 3 by densitometry revealed that 70-90%2c of NP and M could be removed from the membrane. Theremaining NP and M may be due to unfused virions trappedwithin the 10 vesicles and carried through the preparation(Lyles, 1979).The eluting agents shown in Figs. 2 and 3 removed periph-

eral viral proteins from the vesicle membrane, leaving theintegral membrane glvcoproteins associated with the vesiclelipid bilayer. By contrast, complementarv results are observed

C V T

pH 13

S^

V TPCMBS

V TUREA

13, 0.1 N NaOH.FIG. 3 (right). Reagents which selectively remove the nu-

cleocapsid from inside out vesicles containing Sendai viralproteins. Inside out vesicles containing Sendai viral proteins labeledwith [ 8]methionine were incubated with the indicated reagents andsuhjeeted to sucrose gradient centrifugation. P-eak fractions werepooled, precipitated, and subjected to SDS-polvacrvlamide gel elec-trophoresis as in Fig. 2. Fluorographs of the dried gels are shown. V,pooled fractions containing treated vesicles; T, pooled fractions nearthe top of the gradient; PCM1BS, 5 mM p-chloroniercuribenzenesul-fonate: lREA, 6 M urea.

when similar 10 vesicle preparations are treated with TritonX-100, a nonionic detergent which solubilizes the lipid bilayerand leaves protein structure largely intact. Fig. 4A shows theanalysis of Triton-treated IO vesicles by sucrose gradientcentrifugation. The fluorogram in Fig. 4B shows that theintegral membrane proteins HN and F, remain at the top ofthe gradient, nucleocapsid proteins sediment to the bottom inthe form of intact nucleocapsids, and, depending upon the saltconcentration, M protein is distributed between the top andbottom fractions. This disruption pattern (Fig. 4) resemblesthat obtained when intact virions are treated with Triton, asdescribed by Scheid et al. (1972).

Control experiments tested whether viral nucleocapsids as-sociate nonspecificallv with IO erythrocyte membranes. Nu-

F0- *#4

NP-- (_I;,_F-

NP

am_,

S_W m6^ - 40

40 09o *-tXbe'.P

95

4840

by on May 29, 2009

ww

w.jbc.org

Dow

nloaded from

Sendai Virus Protein Interaction with In side Out Vesicles

FEs2-

x

-@\ o/.~ -m.5 ~ ~~~~~~10

FRACTION10 15

fop

FIG. 4. Analysis of disruption products frinside out vesicles containing Sendai viral proX-100. Inside out vesicles containing Sendai virnwith [ 'S]methionine were incubated in the presen100 in 0.1 M KCl on ice for 5 min and centrifuged oi

as described under "Materials and Methods." A, an

products bv sucrose gradient centrifugation. Thelected from the bottom. The ordinate is a plot of tan aliquot of each fraction as determined bxcounting.O --, control vesicles incubated in 5

phate, pH 8.0; v-S,xesicles incubated in Tritrograph of dried gel is shown. SDS-polvacrxlamidewas performed on peak fractions which were pooLEitated: C, control vesicles incubated in 5 mM sodi8.0; T, pooled fractions near the top of the grfractions of treated vesicles; B, pooled fractions fithe gradient.

cleocapsids were purified from intact viionsto JO vesicles which did not contain viramixture was then analyzed by sedimentatioidients to detect association of labeled nucleovesicle fraction. Nucleocapsids were preparecof 1 M KCl, 0.1 M KCl, or 10 mm phosphatEThe salt concentration influences the exterassociation with nucleocapsids; 1 M KCI re,

being present, and low salt concentration resu

sids containing greater amounts of M protei1972). Nucleocapsid preparations were alsoIO vesicles in different salt concentrations (IKCl in 10 mM sodium phosphate) since thethe nucleocapsid has been shown to be depeconcentration (Heggeness et al., 1980). Sucroysis of nucleocapsids prepared in 1M KCI or 1buffer which were added to 0 vesicles notproteins demonstrated that less than 10%8cprotein associated nonspecifically with IO v

shown).

DISCUSSION

A model for the plasma membrane of Sendcells is generated by allowing the envelopes c

grown in the presence of [ 'S]methionine to

rocyte membranes. 10 vesicles, prepared fibranes, have viral nucleocapsid and M pi

external (cytoplasmic) surface (Lyles, 1979)association of these proteins with the vesidlEby treatment with agents which dissociate pro

branes. The disruption products were analygation on sucrose gradients and identified by

m ! rP ws a amide gel electrophoresis. Agents eluting peripheral proteins1 from erythrocvte membranes and leaving the lipid bilayerintact (reviewed by Steck, 1974a) selectively remove the nu-cleocapsid and M proteins; the viral glvcoproteins remainassociated with the lipid bilayer, as would be expected forintegral membrane proteins (Figs. 2 and 3). Several treatmentsthat remove peripheral proteins from erythrocyte membranes

presulted in no elution of viral proteins, such as extremes of

HN- l * ionic strength (0.5 mm sodium phosphate, 1.0 M KCl), 10 mMEDTA, and pH 10. Unexpectedly, I M KCl did not elute viral

V w proteins despite its ability to dissociate M protein from nu-NP- cleocapsids following detergent extraction of virions (ScheidNP'I * w w et al., 1972). No elution of viral proteins was observed follow-

ing freezing and thawing of 10 vesicles, despite changes in-~- 0 * * * duced in the association of nucleocapsids with the viral en-

velope by freezing and thawing (Homma et al., 1976; Kim etal. 1979).

C T \1The ability to remove the nucleocapsid and M proteins

from 10 vesicles (Figs. 2 and 3) implies that their membraneteins eaithmTnitof attachment involves the maintenance of protein structure. Inal proteins labeled contrast, disruption of interactions occurring in the aqueousce of 1V. Triton X- phase does not dissociate the interactions in the lipid phasen sucrose gradients that bind HN and F, to the membrane bilayer. This is alsoalysis of disruption true of integral proteins of the erythrocvte membrane (Steck,gradients were cl 1974a). t is not known with what membrane componentsthe radloactixvitv inliquid scintillation xiral proteins are associated on the cytoplasmic surface of theimM soedium pho)s- 10 vesicles. It appears that there is little or no nonspecifictoIi X-100. B, fluo- association of nucleocapsids with the cytoplasmic surface ofgel electrophoresis the membrane since nucleocapsids isolated from virions undered and acid precip- a variety of conditions do not bind to J0 vesicles lacking viralurn phosphate, pH-adient, V pooled proteins. An association between NP and M of Sendai virus'rom the bottom of has been identified by treatment of intact virions with chem-

ical crosslinking reagents (Markwell and Fox, 1980). Such anassociation could account in part for the membrane interac-tions observed here. One or both of these components may

and were added also interact with other proteins in the membrane, such asil proteins. The the cytoplasmically exposed portion ofHN or F, (Lyles, 1979),n in sucrose gra- as hypothesized for this and other virus types (Blough and)capsids with the Tiffanv, 1975; Simons and Garoff, 1980). In addition, thesein the presence components could interact directil with the membrane lipidsbuffer, pH 7.4. in a conformation-dependent manner. We have not found

nt of M protein conditions which allow removal of intact nucleocapsids fromsults in little M the membranes without disruption of the lipid bilayer, as withilts in nucleocap- Triton X-100. This suggests that the forces binding nucleo-in (Scheid et al., capsids to membranes are similar in magnitude to thoseadded to similar responsible for nucleocapsid integrity.0.0, 0.1, or 1.0 M Some intriguing possibilities are suggested by these results.conformation of Since nucleocapsids remain associated with the membranetndent upon salt after fusion of the viral envelope with the cell membrane, these gradient anal- initial transcription events in virus replication may occur onL0 mM phosphate membrane-bound nucleocapsids or, alternatively, an addi-containing viral tional step after viral envelope fusion may be necessary forof nucleocapsid releasing nucleocapsids into the cytoplasm. Our preliminaryesicles (data not data favor the former possibility, suggesting that nucleocap-

sids remain membrane-bound following penetration of Sendaivirus into cell types other than erythrocytes. 10 vesiclescontaining viral proteins also provide a model for the struc-

lai virus-infected tural organization of the viral envelope. It has been suggested:fmature virions recently (Li and Fox, 1980) that M is a peripheral membranefuse with eryth- protein in virus envelopes, based upon disruption of virionsrom such mem- with lithium diiodosalicylate. Likewise, in electron micro-roteins on their graphs, nucleocapsids are frequently observed to be periph-The extent of erallv associated with the viral envelope (Kim et al., 1979)

es was examined and with cell membranes following viral envelope fusion (Bue-teins from mem- chi and Bachi, 1979). Mechanisms of membrane attachment,zed by centrifu- similar to those studied here may also be involved in virusSDS-polvacryl- maturation by budding from host cell membranes, in which

4841

5

by on May 29, 2009

ww

w.jbc.org

Dow

nloaded from

Sendai Virus Protein Interaction with Inside Out Vesicles

nucleocapsids and M protein associate with the cytoplasmicsurface of the host plasma membrane.

REFERENCESBachi, T., Aguet, M., and Howe, C. (1973) J. Virol. 11, 1004-1012Blough, H. A., and Tiffany, J. M. (1975) Curr. Top. Microbiol.Immunol. 70, 1-30

Bonner, W. M., and Laskey, R. A. (1974) Eur. J. Biochem. 46, 83-88Buechi, M., and Bachi, T. (1979) J. Cell Biol. 83, 338-347Choppin, P. W., and Compans, R. W. (1975) in Comprehensive

Virology (Fraenkel-Conrat, H., and Wagner, R. R., eds) Vol. 4, pp.

95-178, Plenum Press, New YorkHeggeness, M. H., Scheid, A., and Choppin, P. W. (1980) Proc. Natl.Acad. Sci. U. S. A. 77, 2631-2635

Homma, M., and Ohuchi, M. (1973) J. Virol. 12, 1457-1465Homma, M., Shimizu, K., Shimizu, Y. K., and Ishida, N. (1976)

Virology 71, 41-47

Kim, J., Hama, K., Miyake, Y., and Okada, Y. (1979) Virology 95,523-535

Laemmli, U. K. (1970) Nature (Lond.) 227, 680-685Lamb, R. A., and Choppin, P. W. (1977) Virology 81, 382-397Li, J. K.-K., Miyakawa, T., and Fox, C. F. (1980) J. Virol. 34, 268-271Lyles, D. S. (1979) Proc. Natl. Acad. Sci. U. S. A. 76, 5621-5625Markwell, M. A. K., and Fox, C. F. (1980) J. Virol. 33, 152-166Scheid, A., Caliguiri, L. A., Compans, R. W., and Choppin, P. W.

(1972) Virology 50, 640-652Scheid, A., and Choppin, P. W. (1974) Virology 57, 475-490Scheid, A., and Choppin, P. W. (1976) Virology 69, 265-277Scheid, A., and Choppin, P. W. (1977) Virology 80, 54-66Simons, K., and Garoff, H. (1980) J. Gen. Virol. 50, 1-21Steck, T. L. (1974a) J. Cell Biol. 62, 1-19Steck, T. L. (1974b) Methods Membr. Biol. 2, 245-281Steck, T. L., and Yu, J. (1973) J. Supramol. Struct. 1, 220-232Tozawa, H., Watanabe, M., and Ishida, N. (1973) Virology 55, 242-

253

4842

by on May 29, 2009

ww

w.jbc.org

Dow

nloaded from