the journal of vol. 267, no. 34, of 5, pp. 24544-24553 ... · the journal of biological chemistry q...

10
THE JOURNAL OF BIOLOGICAL CHEMISTRY Q 1992 by The American Society for Biochemistry and Molecular Biology, Inc. Vol. 267, No. 34, Issue of December 5, pp. 24544-24553,1992 Printed in U.S.A. Preferential Interaction of Albumin-binding Proteins, gp30 and gp18, with Conformationally Modified Albumins PRESENCEIN MANY CELLS AND TISSUESWITH A POSSIBLE ROLE IN CATABOLISM* (Received for publication, February 28, 1992, and in revised form, August 4, 1992) Jan E. Schnitzer$OllII, Arthur Sung$, Reinhard Howat$, and Julisa Bravo$ From the Departments of $Medicine and $Pathology, Division of Cellular and Molecular Medicine, IlZnstitute for Biomedical Engineering, University of California, San Diego, School of Medicine, La Jolla, California 92093-0651 Albumin binding to the endothelial surface appar- ently initiates its transcytosis via plasmalemmal vesi- cles andalso increases capillary permselectivity. Sev- eral albumin-binding proteins, which we call gp60, gp30, and gpl8, have been identified; however, their functional relationship to each other is unclear. In this study, we show that gp30 and gp18 are both variably expressed by cultured rat fibroblasts, smooth muscle cells, and endothelial cells and are present in all tissues examined (heart, lung, skeletal muscle, diaphragm, duodenum, kidney, fat, brain, adrenal, pancreas, and liver). The binding of albumin-gold complexes (A-Au) to gp30 and gp18 was compared with that of native and modified albumins. Monomeric native bovine serum albumin (BSA) interacted much less avidly than A-Au and BSA that was chemically modified byform- aldehyde (Fm-BSA) or maleic anhydride (Mal-BSA). Mal-BSA and A-Au have similar affinity constants for gp30 and gp18 (KO - 3-7 pg/ml(60-100 nM)), which is 1000-fold greaterthan BSA. These interactions were Ca’+-independent but sensitive to pH (~6.0) and high salt concentrations (21.0 M). Comparative bio- chemical characterization provided evidence of con- formational changes for Mal-BSA, Fm-BSA, and A- Au. Anti-native BSA serum recognizes BSA much more avidly thanmodified BSA. Mal-BSA, Fm-BSA, and A- Au are much more sensitive to trypsin digestion than BSA. Cellular processing was also examined. A-Au and Mal-BSA bound at the endothelial cell surface were degraded, whereas BSA was not. Our results indicate that: (i) gp30 and gpl8, unlike gp60, are expressed in all tissues tested regardless of the type of endothelia lining the microvasculature and the local mechanism of transendothelial albumin transport; (ii) BSA confor- mationally modified by either surface adsorption or chemical means not only interacts more avidly with gp30 and gpl8 than nativealbumin but also is prefer- entially degraded by the cells; (iii) A-Au and native albumin are not equivalent probes for detecting albu- min interaction sites; and (iv) gp30 and gpl8 exhibit * This work was supported in part by National Institutes of Health Grant HL43278 (to J. E. S.), a Max Kade Fellowship (to R. H.), a Howard Hughes undergraduate research fellowship (to J. B.), and a grant-in-aid from the American Heart Association, National and California affiliates (to J. E. S.). This work was presented in part at the 1991 and 1992 FASEB meetings (44,45). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 11 To whom correspondence should be addressed University of California, San Diego, School of Medicine, Cellular & Molecular Medicine, 0651,9500 Gilman Dr., La Jolla, CA 92093-0651. Tel.: 619- 534-0918. binding behavior resembling scavenger receptors. The possible roles of gp30 and gp18 in albumin binding, transcytosis, endocytosis, and even protein catabolism are discussed. Albumin binding to continuous microvascular endothelium appears to mediate albumin transcytosis across the endothe- lium via plasmalemmal vesicles (1,2) and appears to maintain normal capillary permeability (3-8, 46). Specific albumin binding to cultured endothelium derived from microvessels with a continuous endothelium has been quantitatedand partially characterized (9). Albumin binding to continuous capillary endothelia of many tissues has also been demon- strated in situ (1,2,7). Studies using albumin-gold complexes (1) and antibody recognition of monomeric albumin (2, 7, 9) have demonstrated notonly albumin binding (10,ll) but also its transport across continuous microvascular endothelium via plasmalemmal vesicles. Albumin acts as a carrier or trans- port protein within the circulation for many different ligands (12), and the binding of certain ligands such as fatty acids appears to increase albumin binding and transcytosis across continuous endothelium (13). Furthermore, it appears that microvascular beds lined with continuous endothelium require the binding of albumin to the endothelial glycocalyx in order to establish normal capillary permeability (3-8,46). Its bind- ing within transendothelial transport pathways (1, 2, 10) apparently forms an additional permselective barrier that reduces capillary permeability (14). Several albumin-binding proteins that we call gp60, gp30, and gp18 have been identified on the endothelial surface (11, 15-17). The functional and structural relationship of these albumin-binding proteins to each other is unclear. Recently, one of these proteins, gp60, was shown to be expressed selec- tively in tissues with microvascular beds lined with a contin- uous endothelium that are known to bind albumin not only for transcytosis via plasmalemmal vesicles but also to main- tain normal capillary permeability (11). Rat gp60 is an endo- thelial sialoglycoprotein, apparently with 0-linked oligosac- charides (18), present on the cell surface both in situ (18) and inculture (11, 15, 18, 19). Microvascular endothelial cells isolated from the continuous endothelium of rat heart, lung, and epididymal fat pad both express gp60 and bind albumin, whereas nonendothelial cells and brain-derived endothelial cells do not (11). The transendothelial transport pathway for various mole- cules including albumin varies considerably between organs and appears to depend greatly on the type of endothelium lining the microvasculature of the organ (20, 21). Based on this information, it appears logical that albumin-binding pro- 24544

Upload: others

Post on 15-Apr-2020

1 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: THE JOURNAL OF Vol. 267, No. 34, of 5, pp. 24544-24553 ... · THE JOURNAL OF BIOLOGICAL CHEMISTRY Q 1992 by The American Society for Biochemistry and Molecular Biology, Inc. Vol

THE JOURNAL OF BIOLOGICAL CHEMISTRY Q 1992 by The American Society for Biochemistry and Molecular Biology, Inc. Vol. 267, No. 34, Issue of December 5, pp. 24544-24553,1992

Printed in U.S.A.

Preferential Interaction of Albumin-binding Proteins, gp30 and gp18, with Conformationally Modified Albumins PRESENCE IN MANY CELLS AND TISSUES WITH A POSSIBLE ROLE IN CATABOLISM*

(Received for publication, February 28, 1992, and in revised form, August 4, 1992)

Jan E. Schnitzer$OllII, Arthur Sung$, Reinhard Howat$, and Julisa Bravo$ From the Departments of $Medicine and $Pathology, Division of Cellular and Molecular Medicine, IlZnstitute for Biomedical Engineering, University of California, San Diego, School of Medicine, La Jolla, California 92093-0651

Albumin binding to the endothelial surface appar- ently initiates its transcytosis via plasmalemmal vesi- cles and also increases capillary permselectivity. Sev- eral albumin-binding proteins, which we call gp60, gp30, and gpl8, have been identified; however, their functional relationship to each other is unclear. In this study, we show that gp30 and gp18 are both variably expressed by cultured rat fibroblasts, smooth muscle cells, and endothelial cells and are present in all tissues examined (heart, lung, skeletal muscle, diaphragm, duodenum, kidney, fat, brain, adrenal, pancreas, and liver). The binding of albumin-gold complexes (A-Au) to gp30 and gp18 was compared with that of native and modified albumins. Monomeric native bovine serum albumin (BSA) interacted much less avidly than A-Au and BSA that was chemically modified by form- aldehyde (Fm-BSA) or maleic anhydride (Mal-BSA). Mal-BSA and A-Au have similar affinity constants for gp30 and gp18 (KO - 3-7 pg/ml(60-100 nM)), which is 1000-fold greater than BSA. These interactions were Ca’+-independent but sensitive to pH ( ~ 6 . 0 ) and high salt concentrations (21.0 M). Comparative bio- chemical characterization provided evidence of con- formational changes for Mal-BSA, Fm-BSA, and A- Au. Anti-native BSA serum recognizes BSA much more avidly than modified BSA. Mal-BSA, Fm-BSA, and A- Au are much more sensitive to trypsin digestion than BSA. Cellular processing was also examined. A-Au and Mal-BSA bound at the endothelial cell surface were degraded, whereas BSA was not. Our results indicate that: (i) gp30 and gpl8, unlike gp60, are expressed in all tissues tested regardless of the type of endothelia lining the microvasculature and the local mechanism of transendothelial albumin transport; (ii) BSA confor- mationally modified by either surface adsorption or chemical means not only interacts more avidly with gp30 and gpl8 than native albumin but also is prefer- entially degraded by the cells; (iii) A-Au and native albumin are not equivalent probes for detecting albu- min interaction sites; and (iv) gp30 and gpl8 exhibit

* This work was supported in part by National Institutes of Health Grant HL43278 (to J. E. S.), a Max Kade Fellowship (to R. H.), a Howard Hughes undergraduate research fellowship (to J. B.), and a grant-in-aid from the American Heart Association, National and California affiliates (to J. E. S.). This work was presented in part a t the 1991 and 1992 FASEB meetings (44,45). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

11 To whom correspondence should be addressed University of California, San Diego, School of Medicine, Cellular & Molecular Medicine, 0651,9500 Gilman Dr., La Jolla, CA 92093-0651. Tel.: 619- 534-0918.

binding behavior resembling scavenger receptors. The possible roles of gp30 and gp18 in albumin binding, transcytosis, endocytosis, and even protein catabolism are discussed.

Albumin binding to continuous microvascular endothelium appears to mediate albumin transcytosis across the endothe- lium via plasmalemmal vesicles (1,2) and appears to maintain normal capillary permeability (3-8, 46). Specific albumin binding to cultured endothelium derived from microvessels with a continuous endothelium has been quantitated and partially characterized (9). Albumin binding to continuous capillary endothelia of many tissues has also been demon- strated in situ (1,2,7). Studies using albumin-gold complexes (1) and antibody recognition of monomeric albumin (2, 7, 9) have demonstrated not only albumin binding (10, l l ) but also its transport across continuous microvascular endothelium via plasmalemmal vesicles. Albumin acts as a carrier or trans- port protein within the circulation for many different ligands (12), and the binding of certain ligands such as fatty acids appears to increase albumin binding and transcytosis across continuous endothelium (13). Furthermore, it appears that microvascular beds lined with continuous endothelium require the binding of albumin to the endothelial glycocalyx in order to establish normal capillary permeability (3-8,46). Its bind- ing within transendothelial transport pathways (1, 2, 10) apparently forms an additional permselective barrier that reduces capillary permeability (14).

Several albumin-binding proteins that we call gp60, gp30, and gp18 have been identified on the endothelial surface (11, 15-17). The functional and structural relationship of these albumin-binding proteins to each other is unclear. Recently, one of these proteins, gp60, was shown to be expressed selec- tively in tissues with microvascular beds lined with a contin- uous endothelium that are known to bind albumin not only for transcytosis via plasmalemmal vesicles but also to main- tain normal capillary permeability (11). Rat gp60 is an endo- thelial sialoglycoprotein, apparently with 0-linked oligosac- charides (18), present on the cell surface both in situ (18) and in culture (11, 15, 18, 19). Microvascular endothelial cells isolated from the continuous endothelium of rat heart, lung, and epididymal fat pad both express gp60 and bind albumin, whereas nonendothelial cells and brain-derived endothelial cells do not (11).

The transendothelial transport pathway for various mole- cules including albumin varies considerably between organs and appears to depend greatly on the type of endothelium lining the microvasculature of the organ (20, 21). Based on this information, it appears logical that albumin-binding pro-

24544

Page 2: THE JOURNAL OF Vol. 267, No. 34, of 5, pp. 24544-24553 ... · THE JOURNAL OF BIOLOGICAL CHEMISTRY Q 1992 by The American Society for Biochemistry and Molecular Biology, Inc. Vol

Avid Binding of Modified Albumins to gp30 and gp18 24545

teins should be distributed selectively in microvascular beds that require albumin either to initiate transcytosis or to reduce transcapillary exchange. Therefore, in this study, we exam- ined the cell and tissue expression of gp3O and gp18 by blotting various rat cell and tissue extracts with albumin conjugated to colloidal gold particles (A-Au).' In addition, because A-Au may be different from native albumin, we compared their binding to gp30 and gp18 and their processing by cultured endothelial cells. The physical nature of these interactions was also partially characterized. Conformational changes in- duced by either chemical means or surface adsorption to gold particles caused not only much more avid binding to gp30 and gp18 than observed with native albumin but also preferential degradation.

EXPERIMENTAL PROCEDURES

Materials-Reagents and other supplies were obtained from the following sources: fetal calf serum, and phosphate-buffered saline (PBS) from GIBCO; gelatin from Difco (Detroit, MI); crystallized bovine serum albumin (BSA) from ICN Biochemicals (Cleveland, OH); EDTA, rabbit and bovine immunoglobulins, bovine orosomu- coid, ovalbumin, bovine transferrin, and polyclonal rabbit antiserum raised against BSA from Sigma; Dulbecco's modified Eagle's medium (DMEM) and trypsin/EDTA solution from Irvine Scientific (Irvine, CA); IODO-GEN (1,4,5,6-tetrachloro-3a,6a-diphenylglycouril), Tri- ton X-100, bicinchoninic acid protein assay from Pierce Chemical Co.; NalZ5I from Amersham Corp.; TPCK-treated trypsin from Wor- thington Biochemicals (Malvern, Pa); all tissue culture plasticware from Costar (Cambridge, MA) or Corning (Wilmington, DE); rat aortic smooth muscle cells (A-10) and rat kidney fibroblasts (NRK- F) from American Type Culture Collection (Rockville, MD); Immo- bilon filters from Millipore (Bedford, MA); nitrocellulose filters from Schleicher & Schuell (Keene, NH); maleylated BSA (Mal-BSA) was a kind gift from Dr. T. Carew (University of California, San Diego).

Cell Culture-Normal rat kidney fibroblasts (NRK-F) and rat aortic smooth muscle cells (A-10) were grown as per the instructions from ATCC. Microvascular endothelial cells from the rat epididymal fat pad (RFC) were isolated and grown in culture as in Refs. 10 and 19. The RFC cells were tested periodically for various standard endothelial markers as in Ref. 10.

Cell Lysates-All cell types were plated onto 35-mm plastic dishes using 100,OOO cells/dish as in Refs. 10 and 19. After 3-4 days, the cells were confluent and the RFC, NRK-F, and A-10 cells averaged 5.4 (k .8) X lo6, 1.3 (k.3) X lo6, and 7.1 (k 1.0) X lo5 cells per dish, respectively. Some cells were lysed and solubilized directly with 500 p1 of cold solubilization buffer (SB) containing 0.17 M Tris-HC1 (pH 6.8), 3% (w/v) SDS, 1.2% (v/v) 8-mercaptoethanol, 2 M urea, and 3 mM EDTA in double distilled water as in Refs. 17 and 18. Other cells were processed as in Ref. 16. Briefly, they were first homogenized in 1% SDS with 1 mM CaC12 in the presence of protease inhibitors (5 mM benzamidine and 1 mM phenylmethylsulfonyl fluoride) and the homogenate was sonicated and then centrifuged. The supernatant WBS fractionated by trichloroacetic acid precipitation, and the pellet was solubilized in 1% SDS. Both protocols gave similar blots.

Tissue Lysates-The vasculature of anesthetized male albino rats (Sprague-Dawley, 200-250 g) was flushed free of blood with DMEM as in Ref. 11. The following tissues were excised, cut into small pieces, and frozen in liquid nitrogen: brain, heart, lung, liver, kidney, adrenal, duodenum, pancreas, diaphragm, skeletal muscle, and epididymal fat pad. Weighed tissue samples were processed as described above for the cell lysates and as in Refs. 11 and 16.

Albumin Conjugation to Gold-Gold particles of 5-nm diameter were complexed with BSA as in Ref. 1. The concentration of the A-

thelia1 Fm-BSA, formaldehyde-BSA; Mal-BSA, maleic anhydride- ' The abbreviations used are: A-Au; albumin-gold complex; endo-

BSA; TPCK, L-1-tosylamido-2-phenylethyl chloromethyl ketone; PBS, phosphate-buffered saline; ELISA, enzyme-linked immunosor- bent assay; BSA, bovine serum albumin; DMEM, Dulbecco's modified Eagle's medium; PAGE, polyacrylamide gel electrophoresis; RCA, R. communis agglutinin; LFA, L. flavus agglutinin, WGA, wheat germ agglutinin; SBA, G. mar agglutinin; ConA, concanavalin A; RFC, rat epididymal fat pad endothelial cells; aBSA, anti-BSA antiserum; EWB, ELISA wash buffer; A-10, rat aortic smooth muscle cells; NRK-F, rat kidney fibroblasts.

Au suspension was measured by its absorbance at 515 nm. The BSA concentration present in the A-Au suspensions at a given absorbance was determined by performing several A-Au conjugations in the presence of trace IZ5I-BSA (1:lOOO ratio) and calculated from the sample radioactivity and the specific activity of the lZ5I-BSA. The amount of BSA adsorption to the gold particles was quite reproducible from batch to batch. An A-Au solution with an absorbance of 1.0 had a BSA concentration consistently near 20 pg/ml. Final calculations showed that an average of 4 albumin molecules were absorbed to each gold particle. Monodispersity of the probe was checked by electron microscopy in suspension negatively stained with 8% M?+-uranyl acetate in double distilled water.

Chemically Modified Albumim-BSA was reacted with formalde- hyde as in Ref. 22, and maleic anhydride-treated BSA (Mal-BSA) was generously provided by Dr. T. Carew, who utilizes this probe for investigating scavenger receptors and made it as described in Ref. 23. The extent of lysine modification was measured as in Ref. 24 and was consistent between different batches.

Radioiodination of Albumim-One-half milliliter of PBS contain- ing 300 pg of BSA or Mal-BSA was added to a glass screw-cap tube, which had been coated with 0.4 mg of IODO-GEN by evaporating 0.5 ml of this reagent under nitrogen in chloroform. NaIZ6I stock (50 ng, 1.0 mCi) was added, and the mixture was incubated for 30 min at room temperature with mild circular agitation every 3 min. The mixture was then removed and filtered through a Bio-Rad G6PD column to eliminate free lZ5I. The final protein concentration was determined using the bicinchoninic acid protein assay with calibra- tion performed using BSA. The specific activity of each batch was similar, ranging from 5 to 10 pCi/pg.

A-Au Blotting-Cell and tissue proteins were separated by SDS- PAGE (using equivalent protein loads for each lane (100 pg for cell lysates and 200 pg for tissue lysates)) and electrotransferred onto Immobilon or nitrocellulose filters as in Refs. 14 and 17. Filter strips were blocked overnight with 1-2 mg/ml rabbit or bovine IgG and then blotted with the A-Au complexes as in Ref. 16. For competition studies, filters with rat lung proteins were blotted with A-Au in the presence of different concentrations of potential competitors. Quan- titation of the band intensities for gp30 and gp18 was performed using a gel scanning video densitometer (Bio-Rad model 620). We tested a number of different blockers and found that A-Au blotting was sensitive to certain blockers and was inhibited by 1% gelatin and 5% Blotto but not IgG or ovalbumin. Therefore, we used IgG for blocking in all experiments described herein.

'ZSZ-Mal-BSA Blotting-The filter strips containing the separated cell or tissue proteins were blocked at 4 "C for about 4-8 h with either 1-2 mg/ml bovine IgG or 5% Blotto in PBS with 0.05% Nonidet P- 40 and then incubated at 4 "C overnight with 1 pg/ml lZ5I-Mal-BSA in the blocking solution in the presence or absence of potential competitors. The filters were washed three times for 10 min in PBS with 0.05% Nonidet P-40 and then air-dried and exposed for autora- diography at -70 "C using Kodak X-Omat AR film. Quantitation of the band intensities on the autoradiograms was performed as above using a video densitometer and/or counting of the radioactivity in each band using a Beckman Gamma 5500B counter. We tested the blockers mentioned above and found that none of them interfered with the binding of Iz5I-Mal-BSA to gp30 and gp18. Blotto appeared to give the cleanest results with a minimum of background staining. For all experiments comparing A-Au and Mal-BSA binding, only IgG was used as the blocker in order to be consistent.

Direct Binding Titration-Filters containing separated lung pro- teins were blocked with IgG and then blotted as described above with Iz5I-Mal-BSA, '"1-BSA, and Iz51-A-Au (or A-Au) at various concen- trations ranging from 0.1 to 40 pg/ml. The regions of the filter containing gp30 and gp18 were cut out along with an equivalent nonspecific region (near 140 kDa), which was used to estimate back- ground binding. The radioactivity in each band was counted, and the background radioactivity was subtracted from the radioactivity meas- ured in the gp30 and gp18 filters. For A-Au, the signal was determined densitometrically with subtraction of background.

Competitive ELZSA and Radioassay-Wells of 96-well plates were coated with 10 pg of BSA in PBS and allowed to air-dry. After washing and blocking with ELISA wash buffer (EWB) (1.0% oval- bumin in PBS) (3 X 1 min; 1 X 1 h), 50 p1 of albumin probe at various concentrations in EWB was added to each well and followed imme- diately with 50 p1 of polyclonal rabbit antiserum raised against BSA (aBSA) at a 1:lOOO dilution. After 1 h, all wells were washed three times with EWB, and then 100 p1 of goat anti-rabbit antiserum either radioiodinated or conjugated to horseradish peroxidase (1:lOOO dilu-

Page 3: THE JOURNAL OF Vol. 267, No. 34, of 5, pp. 24544-24553 ... · THE JOURNAL OF BIOLOGICAL CHEMISTRY Q 1992 by The American Society for Biochemistry and Molecular Biology, Inc. Vol

24546 Avid Binding of Modified Albumins to gp30 and gp18

tion in EWB) was added and incubated for 1 h. All wells were washed three times with EWB. For wells exposed to the '2sII-antiserum, 1% SDS in PBS was added and then the contents were removed for counting of y radioactivity. For the other plates, the ELISA was completed by adding 1 mg/ml o-phenylenediamine HC1 in 25 mM citric acid and 0.05 Na2POI with 0.03% H202. The reaction was stopped with 4 M sulfuric acid, and the absorbance at 405 nm was measured in each well using a Cambridge Technologies model 750 microplate reader. It was necessary to perform both ELISA and radioassays because one of the probes tested (A-Au) interfered with the absorbance in the ELISA.

Agarose Gel Electrophoresis-A 1% agarose gel was poured in running buffer (0.2 M glycine and 25 mM Tris base at pH 7.8), and 1 and 10 pg of BSA, Fm-BSA, and Mal-BSA were added to separate wells and run at constant voltage (50 V) for about 1 h. The migration distance of the proteins was measured from the bottom of the well to the leading edge of the Coomassie-stained band. The relative mobility was calculated by dividing the observed migration distance for each protein by that for BSA.

Enzymatic Digestion of Albumin Probes-The albumin probes a t 1 mg/ml were digested with 0.01 mg/ml of TPCK-treated trypsin with only minor modification of the procedure given in Ref. 25. Briefly, after 1, 30, and 60 min of digestion, equal aliquots were removed, added immediately to SB (3 X usual concentration to make 1 X final concentration) and heated in boiling water for 10 min. 25 pg of each sample was loaded per gel lane for SDS-PAGE on a 12% gel.

Warm-up Internalization and Degradation Assay-Confluent RFC monolayers, which were extensively washed with DMEM a t 37 "C (2 times, 1 min; 1 time, 10 min) and then a t 4 "C (3 times, 1 min), were incubated at 4 "C for 20 min (time necessary to achieve steady state (Ref. 10, data not shown)) with 1 pg/ml of lZ5I-ligand in DMEM. After washing with DMEM (3 X 1 min) a t 4 "C, the cells were then warmed to 37 "C by the addition of warm DMEM and placement in a 37 "C warm room. A t various time points, the DMEM in each well was collected and saved. The cells were immediately lysed with PBS containing 5% Triton X-100 and 1% SDS. The lysates were collected and counted using a Beckman Gamma 5500B counter. After the addition of BSA to a final concentration of 10 mg/ml, the saved media were subjected to 10% trichloroacetic acid precipitation to determine the extent of degradation (trichloroacetic acid-soluble counts) uersus the undegraded fraction (trichloroacetic acid-insoluble counts (pellet)). Background levels were subtracted from the samples.

RESULTS

In this study, we began to investigate the interrelationships of the albumin-binding proteins by characterizing gp30 and gp18 in terms of: (i) their cell and tissue distribution; (ii) their comparative ability to interact with monomeric native albu- min, chemically modified albumins, and albumin conjugated to colloidal gold particles (A-Au); and (iii) the physical char- acteristics of their interaction with A-Au by testing the effects of pH, calcium, and ionic strength on binding. The albumin probes were biochemically characterized in part to assess for possible changes in protein conformation. In addition, we report experiments that begin to examine potential differ- ences in the cellular processing of native and modified albu- mins.

Tissue Distribution of gp30 and gpl8"The pathways and mechanisms for albumin transport vary considerably from organ to organ depending on the type of endothelium lining the local microvascular bed (1,2,11,20,21,26-28). Therefore, determination of the tissue distribution of the albumin-bind- ing proteins may be an important first step in differentiating the functional roles of these proteins. Proteins from several rat organs were separated by SDS-PAGE and electrotrans- ferred to filters. The method originally used to identify gp30 and gp18 as albumin-binding proteins was applied (16), and these filters were blotted with A-Au. Fig. 1 shows the results for a representative panel of the tissues tested. gp30 and gp18 were detected in the heart, lung, brain, liver, and kidney. Depending on the protein loaded onto the gel, we found that the diffuse bands seen for gp30 and gp18 could actually be

180 kDa .

E g $ e $ $5~~~ FIG. 1. Detection of gp30 and gp18 in extracts from various

rat tissues. Proteins from the indicated tissues were extracted, solubilized, separated by SDS-PAGE, electrotransferred onto filters, and then blotted with A-Au (see "Experimental Procedures"). This figure shows a representative panel of the tissues tested. gp30 and gp18 were detected in all of the tissues tested.

FIG. 2. Detection of gp3O and gp18 in lysates from cultured cells associated with the vascular wall. Proteins from rat fibro- blasts (NRK-F), endothelial cells (RFC) , and smooth muscle cells (A-20) were blotted with A-Au (see "Experimental Procedures").

resolved into doublets. Fig. 1 shows the doublets most clearly in the strips from kidney. We also tested a variety of other organs by A-Au blotting. gp30 and gp18 were detected simi- larly in all tissues tested including diaphragm, skeletal muscle, small intestinal muscularis and mucosa, mesentery, fat, ad- renal, and pancreas. It appears that gp30 and gp18 have an ubiquitous rather than selective tissue distribution. Their tissue expression does not appear to depend on the type of endothelium lining the microvasculature since gp30 and gp18 were detected in tissues with continuous (i.e. heart, lung, skeletal muscle, and brain), fenestrated (i.e. duodenal mucosa, adrenal, and pancreas), and/or sinusoidal endothelium (i.e. liver).

Cell Distribution of gp30 and gp18-gp30 and gp18 were first identified as endothelial albumin-binding proteins (16); however, it is unclear whether they are specific for endothelial cells or are present also in other cells associated with the vascular wall. Therefore, we have tested rat aortic smooth muscle cells (A-10) and kidney fibroblasts (NRK-F) both grown in culture. The proteins of these cells were solubilized, separated by SDS-PAGE, and electrotransferred to filters followed by blotting with A-Au. Fig. 2 shows that protein bands with apparent molecular mass values of 30 and 18 kDa are detected in these lysates. Endothelial cells derived from the rat epididymal fat pad (RFC) also express gp30 and gp18 and were used as a positive control in these experiments. Because each filter represents an equivalent amount of cell protein loaded onto the gel, it would appear that RFC cells express more gp30 and gp18 than the other cell types. We

Page 4: THE JOURNAL OF Vol. 267, No. 34, of 5, pp. 24544-24553 ... · THE JOURNAL OF BIOLOGICAL CHEMISTRY Q 1992 by The American Society for Biochemistry and Molecular Biology, Inc. Vol

Avid Binding of Modified Albumins to gp30 and gpI8 24547

quantitated densitometrically the detected signal for A-Au blotting of gp30 and gp18 (with subtraction of background) by integration of the signal intensity of each band, normalized it to the RFC signal, and found that the NRK-F appear to express similar levels of gp30 and gp18 as the RFC cells with a 20% decrease for gp30 and a 25% increase for gp18. Con- versely, the A-10 cells expressed much less gp30 and gp18 with a 70 and 50% decrease, respectively. These results indi- cate that gp30 and gp18 are not expressed specifically by endothelial cells and are variably expressed by other cells associated with the vascular wall.

Characterization of Interaction of A-Au with gp30 and gpI8"Blotting with protein conjugated to gold is a simple technique that provides ample signal amplification. However, it must be clearly established that the protein-gold complex, which constitutes a polymeric probe with protein absorbed to its surface, interacts in a manner analogous to the monomeric native protein. Therefore, we attempted to use native mono- meric BSA to compete with the binding of A-Au to gp30 and gp18 on filters. As shown in Fig. 3, BSA a t 50-fold molar excess did not compete effectively with the observed binding of A-Au to gp30 and gp18. Because the same BSA was used t o make A-Au and to compete with A-Au binding, this result indicates that the adsorption of BSA to gold particles effec- tively increases its affinity for gp30 and gp18 and that the A- Au probe is therefore different from native BSA.

Protein adsorption to a variety of surfaces frequently causes considerable changes in the conformation of proteins such as albumin (29, 30). It is possible that adsorption of albumin to the surface of gold particles creates not only a polymeric albumin probe but also a conformationally modified albumin probe that results in a significantly higher affinity for gp30 and gp18 than that of native monomeric albumin. Therefore, we decided to test this hypothesis by checking two confor- mationally modified forms of albumin for their ability to compete with the interaction of A-Au to gp30 and gp18. For these albumins, the conformational modification was chemi- cally induced and not caused by surface adsorption. From the results shown in Fig. 3, it is clear that the modified albumins were able to compete much more effectively than native albumin for the binding of A-Au to gp30 and gp18. In these experiments, formaldehyde-treated albumin (Fm-BSA) and maleic anhydride-treated albumin (Mal-BSA) at 50-fold mo- lar excess to the A-Au almost totally ablated the observed signal for A-Au binding to these proteins, whereas BSA at the same concentration had little, if any, effect. Various proteins including orosomucoid, transferrin, and ovalbumin did not compete with A-Au binding to gp30 and gp18 (data not shown). These results suggest that conformational changes in albumin induced by either chemical modification or surface adsorption to gold are somehow equivalent a t least in terms of their ability to create a ligand with a much greater affinity

gp30 - I I I

I

9P18 -I I "

a a a py; ; m FIG. 3. Competition of A-Au binding to gp30 and gp18 with

modified and native albumins. The filter strips containing sepa- rated rat lung proteins were blotted with A-Au mixed with PBS alone (Control) or with PBS containing the indicated protein at 50-fold molar excess over the BSA attached to the colloidal gold particles.

for gp30 and gp18 than native BSA. In order to quantitate more precisely the relative affinities

of these competitors for gp30 and gp18, we examined a wide range of competitor concentrations for their ability to inter- fere with A-Au blotting of gp30 and gp18. The competition curves are shown in Fig. 4 ( A and B ) for gp30 and gp18, respectively. In both cases, much greater concentrations of native albumin were required to achieve the same degree of competition exhibited by the modified albumins, Fm-BSA and Mal-BSA. Transferrin, used as a negative control, affects A-Au binding insignificantly. These competition curves allow estimation of the concentration at which 50% competition is achieved (derived simply from the intersection of the compe- tition curve with the dotted line signifying 50% competition). Under most experimental conditions (31), this value provides a useful approximation of the equilibrium binding affinity as defined by the apparent dissociation constant ( K D ) of the competitor for the receptor. BSA had the lowest affinity with KD values for gp30 and gp18 of about 4 and 5 mg/ml, respec- tively. Mal-BSA had the highest affinity for gp30 and gp18 with KD values of about 4 and 5 pglml, respectively. The K D of Fm-BSA was intermediate a t 75 pg/ml for gp30 and 90 pg/ ml for gp18. These results give a quantitative estimate of the relative affinities of these albumins and show that: (i) chem- ically modified albumins react avidly with gp30 and gp18 and (ii) Mal-BSA has an affinity for gp30 and gp18 that is about 1000-fold greater than BSA.

Characterization of '2sII"al-BSA Interaction with gp30 and gpl8"Since the competition studies show that Mal-BSA interacts avidly with gp30 and gp18, we decided to perform

1201 A Transferrin

0 0.0 1 10 100 1000 loo00 1 m

Competitor - pg/ml

COMPETITOR - pg/ml

FIG. 4. Competition curves for A-Au binding to gp30 and gp18. Lung filters were blotted with A-Au (at a final concentration of 0.5 OD at 515 nm, which equaled 10 pg/ml BSA on the gold particles) mixed with the indicated protein to achieve the concentra- tion shown on the x-axis. The detected signal for gp30 and gp18 was quantitated densitometrically and was expressed as a percentage of the signal detected for the control without competitor. The results for gp30 ( A ) and gp18 ( B ) are presented as means for at least three observations (n 2 3).

Page 5: THE JOURNAL OF Vol. 267, No. 34, of 5, pp. 24544-24553 ... · THE JOURNAL OF BIOLOGICAL CHEMISTRY Q 1992 by The American Society for Biochemistry and Molecular Biology, Inc. Vol

24548 Avid Binding of Modified Albumins to gp30 a n d gp18

ligand blots directly with radioiodinated Mal-BSA in the presence of various potential competitors. Binding competi- tions were performed over a wide range of concentrations. Fig. 5 shows typical autoradiograms of filters, which were blotted with 12'I-Mal-BSA alone or in the presence of A-Au, Fm-BSA, Mal-BSA, orosomucoid, or BSA, while Fig. 6 shows the detailed competition curves. '2'II"al-BSA blotted gp30 and gp18 very specifically and with high affinity. Mal-BSA

200 kD;

0

a

FIG. 5. Specific interaction of 1261-Mal-BSA with gp30 and gpl8. Lung filter strips were blotted with 12sII"al-BSA (1.0 pg/ml) in the absence or presence of the indicated proteins at the following final concentrations (mg/ml): BSA (lo), Fm-BSA (lo), Mal-BSA (0.11, A-Au (.04), and orosomucoid (10).

lz0 l A Orosomucoid

0.0 1 10 l o o loo0 loo00 1 m

Competitor - pg/ml

and A-Au were the best competitors for the binding of "'I- Mal-BSA to gp30 and gp18 and gave very similar competition curves. Conversely, BSA did not compete a t all even a t high concentrations (>100,000 molar excess). Fm-BSA gave an intermediate profile and was able to ablate the signal a t higher concentrations. Orosomucoid was used as a negative control, showed little ability to compete, and produced a competition- concentration profile equivalent to BSA. From these compe- tition curves for the 12sII"al-BSA, the approximate KD for each competitor was estimated to be as follows: Mal-BSA (3 pg/ml for gp30 and gp18), A-Au (7 for gp30 and 4 pg/ml for gp18), Fm-BSA (3 mg/ml for both), and BSA and orosomu- coid (infinite because of a lack of competition). I t appears that Mal-BSA and A-Au have very similar affinities for gp30 and gp18. Furthermore, because A-Au can compete completely with the binding of '2sI-Mal-BSA to both gp30 and gp18 and because the reverse is also true (Mal-BSA can compete totally with A-Au binding to gp30 and gp18) (31), both modified albumins appear not only to bind to the same protein but may also interact with the same binding domain within the recep- tor molecule.

Additional Analysis of Ligand Binding-We performed di- rect titrations using each probe on its own a t concentrations ranging from 0.1 to 40 pg/ml. Analysis of several binding experiments revealed that: (i) A-Au, I2'I-A-Au, and '2'I-Mal- BSA exhibited very similar binding behavior with nearly equivalent binding kinetics; and (ii) both A-Au and12'I-Mal- BSA binding to both gp30 and gp18 saturated a t 10 pg/ml with an apparent KD of 5 pg/ml. For l*'I-BSA, only a small signal over background was detected and this binding did not reach saturation. These results confirm the results of the competition studies.

Effect of Electrolyte Conditions-The physical nature of the interaction of gp30 and gp18 with A-Au was investigated by determining the effects of calcium, pH, and ionic strength on binding. When A-Au blots were performed in the presence of 2 mM calcium in PBS or 2 mM EDTA in PBS, the binding was the same. This result indicates that the presence of calcium is not an essential requirement for binding. The effect of pH on A-Au binding to gp30 and gp18 is shown in Fig. 7 and was similar for both gp30 and gp18. At pH values ranging from 6.0 to 10.0, very little effect on the binding of A-Au to gp30 and gp18 was observed. However, at pH values less than 6.0, there was a rapid decrease in the observed binding of A- Au to both gp30 and gp18. The binding was ablated a t a pH

Orosomucoid

a

0.0 1 10 loo loo0 loo00 1 m

Competitor - pg/ml

FIG. 6. Competition curves for l2'1-Mal-BSA binding to gp30 and gp18. Lung filters were blotted with 12sII"al-BSA (1.0 pg/ml) mixed with the indicated protein at the concentration shown on the x-axis. The signal for gp30 and gp18 was quantitated and was expressed as a percentage of the signal detected for the control without competitor. The results for gp30 ( A ) and gp18 ( B ) are presented as means (n 2 3).

I / r l 0 ' I - I . v ' n . - . (

4 5 6 7 8 9 1 0

PH FIG. 7. The pH sensitivity of A-Au binding to gp30 and

gp18. Lung filter strips were blotted at the indicated pH with A-Au. The results were quantitated by densitometry and normalized to the A-Au signal detected at pH 7.0. Since the curves for both gp30 and gp18 were almost identical, the results were combined and expressed as the mean & S.D. (n = 3).

Page 6: THE JOURNAL OF Vol. 267, No. 34, of 5, pp. 24544-24553 ... · THE JOURNAL OF BIOLOGICAL CHEMISTRY Q 1992 by The American Society for Biochemistry and Molecular Biology, Inc. Vol

Avid Binding of Modified Albumins to gp30 and gp18 24549

equal to 4.0. Binding of A-Au to gp30 and gp18 was also sensitive to ionic strength conditions. Fig. 8 shows that low to normal physiological NaCl concentrations (10.15 M) did not affect the binding but higher salt concentrations (21.0 M) reduced it significantly. Binding to gp30 was more sensitive to the NaCl concentration than binding to gp18. This sensi- tivity to electrolyte concentration suggests that some aspects of A-Au binding to gp30 and gp18 may be electrostatically mediated.

Lectin Competition-Similar to our past work (15), we attempted to use several lectins (including Ricinus communis (RCA), Limax flaws (LFA), Triticum vulgaris (WGA), con- canavalin A (ConA), and Glycine max (SBA) agglutinins) to compete with albumin interactions, in this case with A-Au blotting of gp30 and gp18. Concentrations as high as 1 mg/ ml were used and only ConA competed significantly with A- Au binding to gp30 and gp18. Quantitation of the blots by densitometry revealed a 93 f 7 % (n = 4) decrease in signal as compared with controls done without lectins; this value

.a 8 .i? *I ............................................................. @\\8 ....... ~ ..... ............................

\ ! 4 1 \ \

*O 1 0

.01 .1 1 10

[NaCIl - M FIG. 8. The ionic strength sensitivity of A-Au binding to

gp30 and gpl8. Lung filter strips were blotted with A-Au at the indicated NaCl concentrations (pH = 7.4). The results were quanti- tated by densitometry, normalized to the A-Au signal detected a t 0.15 M, and expressed as the mean of duplicate filter strips.

0 - 1 10 loo loo0 loo00

Protein Concentration (pg/ml) FIG. 9. Comparative analysis of the recognition of modified

BSA by anti-native BSA serum. The indicated albumin probes were used as competitors in solution for the binding of aBSA to immobilized BSA (see “Experimental Procedures”). The maximum concentration of the BSA of the A-Au probe was limited to 1 mg/ml because of the nature of the colloidal gold solution. Also note that 100 pl volumes were used for each well, so that the total amount of protein is 1/10 of the concentration given. Each point represents the mean value for 4 c n 5 12. Standard deviations ranged from 3 to 17% with a median of about 7%.

represents the combined results for gp30 and gp18 because they showed similar levels of competition. All other lectins did not compete. We also lectin-blotted these filters as in Ref. 18. All of these lectins except for LFA interacted with bands corresponding to gp30 and gp18 (data not shown). Therefore, these albumin-binding proteins appear to be glycoproteins interacting with a variety of lectins; however, only ConA was able to compete with A-Au binding to them.

Biochemical Characterization of Albumin Probes-Although it is clear from the literature that both surface absorption and chemical modifications such as the ones used in this study can extensively alter protein structure (25, 29, 30), it is necessary to establish whether such changes have occurred for our albumin probes. We examined the albumin probes by assessing: (i) the extent of modification of their lysine resi- dues, (ii) their relative electrophoretic mobilities, (iii) their comparative ability to be recognized in solution by polyclonal antiserum raised to native BSA, and (iv) their sensitivity to trypsin digestion.

Extent of Lysine Modification-The chemical modification of lysine residues in Mal-BSA and Fm-BSA was measured by a decrease in the number of amino groups detected colorimet- rically with 2,4,6-trinitrobenzenesulfonic acid’ (24). Compar- ative analysis with BSA demonstrated that about 98 and 43% of the lysines detected by the assay were modified in Mal- BSA and Fm-BSA, respectively. These values are consistent with previous studies wherein the extent of lysine modifica- tion was 46% for Fm-BSA (32) and 99% for Mal-BSA (33).

Differential Antibody Recognition-Polyclonal antiserum against BSA (aBSA) was used to assess if the BSA probes had been modified sufficiently to reduce antibody interaction. The aBSA was incubated with the BSA probes in solution in the presence of a fixed amount of immobilized BSA (see “Experimental Procedures”). Fig. 9 shows the competition curves. Native BSA was clearly the best competitor. Fm-BSA and A-Au competed similarly and were much less effective than BSA but significantly more than Mal-BSA. Transferrin did not compete. The amount of albumin probe necessary to achieve a 50% decrease in observed aBSA binding was ex- amined. For BSA, it was about 7 pg (a concentration of 70 pg/ml with 100 pllwell). For both Fm-BSA and A-Au, it was 100 pg. Mal-BSA did not achieve 50% competition. Therefore, BSA in solution was recognized by aBSA about 15 times more avidly than the other albumins. This assay provides evidence that aBSA interacts more avidly in solution with native BSA than modified BSA, which suggests that conformational changes may have occurred.

Differential Resistance to Tryptic Digestion-The albumin probes were subjected to trypsin digestion over a 1-h period and then analyzed by SDS-PAGE (see “Experimental Proce- dures”). Bovine trypsin hydrolyzes only bonds involving the

Native BSA Fm-BSA A-AU Mal-BSA

Trypsin Digestion (Min)

FIG. 10. SDS-PAGE analysis of tryptic digestion of BSA probes. The indicated albumin probes were digested with trypsin for the indicated times and then analyzed by SDS-PAGE followed by Coomassie staining (see “Experimental Procedures”). The samples in the lunes marked 0 were never exposed to trypsin.

Page 7: THE JOURNAL OF Vol. 267, No. 34, of 5, pp. 24544-24553 ... · THE JOURNAL OF BIOLOGICAL CHEMISTRY Q 1992 by The American Society for Biochemistry and Molecular Biology, Inc. Vol

24550 Avid Binding of Modified Albumins to gp30 and gp18

carboxyl groups of arginine and lysine that are accessible to the protease. Fig. 10 shows that little, if any, of the native BSA appeared to be digested, whereas both chemically modi- fied albumins, in spite of having a significant number of lysine groups blocked (see above), which renders those sites resistant to cleavage by trypsin, were clearly more sensitive to trypsin digestion than native BSA. After 30 min, the band represent- ing intact Mal-BSA nearly disappeared and new bands rep- resenting smaller peptide fragments appeared. Mal-BSA was almost completely proteolyzed. Even the BSA absorbed to colloidal gold particles became more sensitive to trypsin diges- tion, despite the expected protective effect that the gold surface should provide to the absorbed protein because of steric hindrances. In essence, the trypsin molecule must have limited access to the albumin on the gold because it cannot approach the albumin from the side of the molecule that is adjacent to the gold. Unlike Mal-BSA, neither A-Au nor Fm- BSA was digested completely after 1 h; however, there was a significant decrease in the intensity of the band representing the whole molecule with a concomitant appearance of novel bands representing smaller peptides not initially present. Both chemical modification and surface absorption of BSA caused an increased sensitivity to trypsin digestion. Normally, BSA appears to have few sites accessible to trypsin; however, after either chemical modification or surface absorption, tryp- sin appeared to gain access to arginine residues and/or any unblocked lysines and a significant increase in trypsin cleav- age is observed. Conformational changes in BSA must have occurred to provide access to internal digestion sites that are not available for native BSA.

Differential Electrophoretic Mobilities-Chemical modifi- cation of proteins can alter their charge. The relative electro- phoretic mobility of Mal-BSA and Fm-BSA through a 1% agarose gel at pH of 7.8 was compared with that of native BSA based on the observed migration distance. The relative mobility for BSA, Mal-BSA, and Fm-BSA was 1.00, 1.44, and 1.17, respectively. These results suggest that both modified albumins are more polyanionic than native BSA and agree with the known effects of these chemical modification proce- dures. For Mal-BSA, a significant increase in electronegativ- ity is expected because of the addition of new negative charges with the formation of maleylysine residues (34). For Fm-BSA, the effect is less dramatic. Covalent linkage of formaldehyde to amine groups causes an apparent lowering in their pK of 2-3 units, which at neutral to basic pH decreases the number of positive charges and increases the overall electronegativity of the protein (34). Furthermore, our results agree fairly well with past work showing that BSA maleylated by a similar procedure had a relative electrophoretic mobility of 1.7 at a pH of 8.4 (33).

Different Cellular Processing of Modified and Natiue Albu- min-There is ample evidence in the literature that modified albumins are processed differently than native albumin and are not only internalized by endothelium but also accumulate in lysosomes in situ (see “Discussion”). However, it seems necessary to clearly establish whether modified albumins are indeed preferentially degraded. In order to begin to investigate potential differences in the cellular processing of the albumin probes, we incubated RFC cell monolayers with either Y - A - Au, ’261-Mal-BSA, or T -BSA a t 4 “C before allowing the cells to internalize and potentially degrade the ligand specifically bound at the cell surface by warming them to 37 “C (see “Experimental Procedures”). After incubating the cells at 37 “C for different times ranging from 0 to 60 min, a mass balance was performed by following the radioactivity associ- ated with the cellular fraction (cell-associated counts) and

released by the cells into the media either in the degraded fraction (trichloroacetic acid-soluble counts) or in the unde- graded fraction (trichloroacetic acid-insoluble counts; pellet). This type of “warm-up” experiment allows a focused assess- ment of the cellular processing of the ligand specifically bound at the RFC cell surface, examines directly the relationship between cell surface binding and subsequent internalization and possible degradation, and minimizes the background com- ponent derived from nonspecific fluid-phase endocytosis.

Fig. 11 shows a typical composite analysis of the time course for the cellular processing of each surface-bound ligand. For A-Au, after just 1 h at 37 “C, more than 50% of the total surface-bound A-Au was degraded by the cells, whereas less than 5% appeared to be released back into the media unde- graded. The A-Au associated with the cell decreased rapidly in direct parallel with the increase in degradation products found in the medium. This result suggests that once A-Au was bound to the cell surface, it was efficiently endocytosed for degradative purposes and was not internalized and/or released back into the medium in undegraded form. Surface- bound A-Au appeared to be rapidly and predominately proc- essed for degradation. Mal-BSA was processed similarly to A- Au. At least initially, Mal-BSA appeared to be internalized and degraded more quickly than A-Au with a shorter lag period. However, the percentage of the surface-bound Mal- BSA that was degraded after the full l-h period was quantia- tively very similar to that of A-Au. About 40% of the Mal- BSA originally bound at the cell surface was degraded with a concomitant proportional decrease in Mal-BSA associated with the cells. Unlike A-Au and Mal-BSA, very little surface- bound BSA was degraded by the RFC cells with less than 2% of the total BSA originally bound at the cell surface degraded after 1 h. About 20% of the surface-bound BSA is released back to the medium in an undegraded form after 60 min, which is much greater than that observed for A-Au. This undegraded albumin in the medium may result from reversi- bility of surface-bound albumin and/or from internalization and subsequent exocytosis without degradation (for example, via transcytosis). By comparing these curves, it is evident that: (i) about 25 times less of the BSA specifically bound at the cell surface was degraded when warmed up to 37 “C than the surface-bound A-Au or Mal-BSA; and (ii) the absorption of BSA onto the surface of colloidal gold particles and chem- ical modification of BSA by maleylation clearly altered the cellular processing of the BSA probes resulting in avid uptake and preferential degradation.’

DISCUSSION

Implications for Transendothelial Transport-Vascular en- dothelium is heterogeneous with considerable structural var- iation among the vascular beds of different organs and can be

These results cannot be explained simply by the polymeric nature of A-Au. First, monomeric Mal-BSA was processed similarly to the A-Au. Second, this assay was specifically designed to eliminate this “multivalency” factor as a variable by examining internalization and degradation of the surface-bound probes in terms of a percentage of the total ligand detected at the cell surface. The auxiliary albumin molecules, which are not bound directly to the receptor but are attached to the gold particles, potentially get “a free ride” that could increase the amount of albumin degraded by the cells in a dispropor- tionate manner relative to the monomeric probes. By assessing the data in terms of the original amount bound at the cell surface, these “accessory” albumins are factored out in the final percentage and do not directly skew the comparison with the monomeric probes. Al- though all of these albumin probes bound well to the RFC cell surface (our unpublished observations and Ref. 121, modified BSA were indeed processed differently and were degraded, whereas native BSA was not.

Page 8: THE JOURNAL OF Vol. 267, No. 34, of 5, pp. 24544-24553 ... · THE JOURNAL OF BIOLOGICAL CHEMISTRY Q 1992 by The American Society for Biochemistry and Molecular Biology, Inc. Vol

Avid Binding of Modified Albumins to gp30 and gp18 24551

Total

Cell-Dissociated """""-.

I I "

0 10 20 30 40 50 60 Incubation Time - min

'. %. -.- -. -. -. *. -. ".-. Cell Associated -. -. ---._. D e g r a ~ ~ " . - . '

_""" -"w" """"""".

,*=" - - - - - -). - "-"""-* Cell-Dissociated

jO'

v 0.t"' -

# I I i

0 10 20 30 40 50 60

Incubation Time - min

k

an Total

-. -.- Cell-Associated -. -" '~.-.-.""""""""-.-.-~ .. - H 401

m Cell-Dissociated 20-

"

8 """""*

/"""""" Degraded ~~ o- """""""- *""- """"""""_

I I 0- ~

v 0 10 20 30 40 50 60 Incubation Time - min

FIG. 11. RFC processing of modified and native albumin

bated at 4 "C with "'1-A-Au (A) , "'1-Mal-BSA ( B ) , or "'1-BSA (C) bound initially at the cell surface. RFC monolayers were incu-

and then incubated with DMEM at 37 "C for the time indicated. A mass balance was performed for the different fractions (cell-associ- ated uersus degraded uersus cell-dissociated), which were summated (Total). The results were normalized to the I2'II-ligand bound initially at the cell surface at 4 "C (incubation time = 0). For each time point, t.he mean of duplicates is given.

grouped into three main types: continuous, sinusoidal, and fenestrated (21). Albumin crosses each of these endothelial barriers via different transport pathways (for a more complete discussion, see Ref. 11). Briefly, many vascular beds lined with continuous endothelium transport albumin via plasma- lemma1 vesicles, possibly as a result of receptor-mediated transcytosis (1, 2). The only exception so far noted in the literature for continuous endothelium has been that of the brain. This highly restrictive endothelium neither binds nor transports albumin (35, 36), which may be due in part to the lack of plasmalemmal vesicles. Albumin crosses sinusoidal endothelium freely via the large gaps between the cells. Si- nusoidal endothelia of the liver and bone marrow endocytose A-Au complexes via coated vesicles but do not transcytose albumin via plasmalemmal vesicles (26, 27). Some fenestrated endothelia internalize albumin via coated vesicles; however, A-Au complexes cross the endothelium via diaphragmed fe- nestrations and not plasmalemmal vesicles (28).

The mechanism and pathway for albumin transendothelial transport depends on the type of endothelium lining the microvasculature of the organ. Therefore, it appears logical to expect that albumin-binding proteins involved in albumin transcytosis would be selectively distributed in a manner consistent with their function in each organ. If albumin- binding proteins are involved in specific transcytosis of al- bumin, then they should be present in tissues with microvas- cular beds with continuous endothelia that bind and transport albumin via plasmalemmal vesicles and absent in those tissues with other endothelia that do not. In addition, as discussed in more detail previously ( l l ) , capillaries with continuous endothelium also require albumin binding to the glycocalyx to maintain normal permeability (3,4,7,8,14), whereas other types of endothelia may not. Our results show that gp30 and gp18 are distributed ubiquitously in all of the tissues tested regardless of both the type of endothelium lining the micro- vasculature and the local mechanism of transendothelial transport of albumin. Furthermore, they are not specifically expressed by vascular endothelium but are also in lysates of fibroblasts and smooth muscle cells. The other albumin- binding protein gp60, unlike gp30 and gp18, appears to be expressed only in certain tissues with microvascular beds lined with continuous vascular endothelium that are known to bind and transcytose albumin (11). gp60 is not detected in cortical brain or in tissues solely with microvascular beds lined with sinusoidal or fenestrated endothelium (11). Of the cells tested so far, gp60 is expressed only by endothelial cells that bind native albumin (11). The tissue and cell distribution of gp60 correlates well with its apparent function mediating albumin binding and possibly transcytosis.

Avid Preferential Binding of Modified Albumins to gp30 and gpl8"Since most of the past work identifying gp30 and gp18 as albumin-binding proteins was performed using A-Au probes (16), it appeared to be prudent to examine carefully for potential differences in the interaction of gp30 and gp18 with A-Au and native monomeric albumin (BSA). Our com- petition studies show that BSA is not a very effective com- petitor for A-Au binding to gp30 and gp18 (see Figs. 3 and 4), especially when compared to chemically modified albumins. BSA binds at least 1000-fold less avidly to gp30 and gp18 than either Mal-BSA or A-Au. Therefore, neither A-Au nor Mal-BSA can be considered equivalent or appropriate probes for defining the interaction of native albumin with the endo- thelial cell surface and its binding proteins. Conversely, A-Au and Mal-BSA appear in many respects to be nearly equivalent probes. They not only bind to gp30 and gp18 with very similar high affinities but also appear to interact with the same

Page 9: THE JOURNAL OF Vol. 267, No. 34, of 5, pp. 24544-24553 ... · THE JOURNAL OF BIOLOGICAL CHEMISTRY Q 1992 by The American Society for Biochemistry and Molecular Biology, Inc. Vol

24552 Avid Binding of Modified Albumins to gp30 and gp18

binding domain (as indicated by their ability to compete fully with each other's binding).

Physicochemical Basis of Avid Binding of Modified Albu- mins-The difference in binding to gp30 and gp18 of A-Au and BSA could be related to the polymeric nature of the A- Au. Multiple ligands on each gold particle could react with several albumin binding sites and thereby have a higher effective binding af f in i t~ .~ However, even when BSA was present at 50-100-fold molar excess relative to the total BSA attached to the gold particles (A-Au), little competition was observed (see Figs. 2 and 3). Furthermore, although Mal-BSA is also a monomer, it not only competes excellently but interacts with similar affinity.* A-Au can abolish the inter- action of '251-Mal-BSA with gp30 and gp18, whereas BSA does not.

These results suggest an alternate, more plausible expla- nation for the difference in binding behavior among the various albumins. Surface adsorption of albumin to gold may cause the protein to unfold and alter its native conformation, which ultimately unmasks a binding domain(s) that is recog- nized avidly by gp30 and gp18. The conformational change induced by surface adsorption may resemble the chemically induced conformer sufficiently to be recognized in common by gp30 and gp18. It is well established that both chemical modification and surface adsorption of proteins such as al- bumin frequently result in changes in protein conformation (25, 29, 30). Although Mal-BSA and Fm-BSA have been extensively used over the past decade to study scavenger receptors (22, 32, 33, 37), it appeared necessary to test the albumin probes for possible changes in conformation. Both chemical modification and surface absorption of albumin caused sufficient changes in the molecule to decrease its recognition by polyclonal anti-BSA serum (Fig. 9) and in- crease its sensitivity to trypsin digestion (Fig. 10). Surface- induced unfolding of albumin is well documented (29) and is quite consistent with the results of our trypsin digestion assay. Although we cannot rule out that increased electronegativity plays a role in the interaction with gp30 and gp18 since Fm- BSA and especially Mal-BSA were more negatively charged than BSA, it seems more likely that conformation is the primary factor because: (i) the effects of formaldehyde treat- ment did not cause a very large change in apparent charge; and (ii) surface-induced unfolding is expected to change the apparent charge of the protein to a limited extent (29), espe- cially when compared to Mal-BSA. The avid recognition of these three different albumin probes may best be explained by an induced conformational change in albumin that be- comes a prerequisite for the unmasking of a recognition domain(s) not normally accessible in the native state to gp30 and gp18. Because changes in protein charge and conforma- tion can be very integrally interrelated (34), future work will be necessary to resolve this issue definitively and will require identifying the putative unmasked region of primary structure

Multiple ligands on the gold particle may increase the apparent effective affinity for the binding sites by decreasing its dissociation rate. The dissociation probability for the probe with multiple bound ligands is dependent on the dissociation rate of each ligand and requires that all ligands on the gold become detached at the same time. Furthermore, the presence of the other ligands may increase the probability of re-attachment. Therefore, although the dissociation rate for each albumin molecule on the gold particle could be the same, the overall effective affinity could increase.

Mal-BSA was analyzed by gel chromatography for possible polym- erization as a result of maleylation. Mal-BSA and BSA were greater than 99% monomer and gave almost identical chromatograms. The apparent small difference in affinity between Mal-BSA and A-Au may be caused by steric hindrances because of obvious differences in size.

of albumin that is the recognition domain responsible for its binding to gp30 and gp18. It is interesting to note that at least for Mal-BSA, protein conformational changes may play a more important role in macrophage scavenger recognition than increased electronegativity (33).

Both chemical modification and surface absorption of al- bumin can alter its interaction with cell surfaces (25, 30). In fact, a surface-induced conformer of albumin has recently been shown to avidly bind to hepatocytes, whereas the un- modified form does not (30). Since hepatocytes do express gp30 and gp18; they may mediate this interaction. Further- more, it is possible that most of the observed binding of albumin to gp30 and gp18 may actually stem from a small subpopulation of albumin which is modified (i.e. old albumin or albumin absorbed to plastic). This small percentage may react preferentially with gp30 and gp18, whereas the predom- inant nonmodified native form does not. We are currently beginning to investigate these possibilities.

Possible Relation to Scavenger Receptors-The phenomena of avid recognition of modified albumins by gp30 and gp18 suggests binding behavior akin to various known scavenger receptors (22,32,33,37). Native proteins interact poorly with scavenger receptors but when specific proteins are modified (i.e. acetylation or oxidation of LDL, maleylation of albumin), specific scavenger receptors bind them avidly for internaliza- tion by receptor-mediated endocytosis and ultimately for deg- radation (22, 32, 33, 37). Our current work shows that many ligands known to interact with scavenger receptors inhibit both A-Au binding to gp30 and gp18 and its internalization and degradation by RFC cells (38).'j

Different Internalization Pathways for Native Monomeric and Modified Albumins-Recently, a number of investigators (1 ,2, 26-28,39,40) have found that modified albumins such as A-Au can be internalized by vascular endothelium. In some microvascular beds lined with discontinuous endothelium such as the adrenal, liver, and bone marrow, A-Au apparently binds to coated pits and vesicles for internalization and ac- cumulation in lysosomal compartments (26-28). In other mi- crovascular beds lined with continuous endothelium, inter- nalization of both glycated A-Au and A-Au appears to occur via uncoated plasmalemmal vesicles and to a much lesser extent via coated vesicles (1, 2, 39, 40). These probes ulti- mately either accumulate within endosomes and lysosomes or are transported directly across the cell by plasmalemmal vesicles. Conversely, native albumin is only transcytosed via plasmalemmal vesicles without any detectable accumulation in endosomal compartments (2). This evidence suggests that plasmalemmal vesicles may be involved in both cellular en- docytosis and transcytosis and that there may be two popu- lations of functionally distinct plasmalemmal vesicles which deliver their contents to different parts of the cell.

Role of Albumin-binding Proteins in Endocytosis and Tran- scytosk-The binding of albumin probes to the endothelial cell surface appears to mediate their transcytosis or endocy- tosis. Native albumin binding to the surface of cultured mi- crovascular endothelium has been quantitated (10, ll) and can be inhibited by the interaction of lectins, such as RCA, LFA, and WGA, but not ConA and SBA, with gp60 (14). Furthermore, RCA interaction with gp60 appears to inhibit specific albumin transport across bovine pulmonary artery endothelial cell monolayers (41). In this study, we found that ConA but not RCA, LFA, WGA, or SBA inhibited A-AU binding to gp30 and gp18. Although all three albumin-binding proteins are glycoproteins, these apparent differences in lectin

J. E. Schnitzer and A. Sung, unpublished observations. J. E. Schnitzer and J. Bravo, manuscript in preparation.

Page 10: THE JOURNAL OF Vol. 267, No. 34, of 5, pp. 24544-24553 ... · THE JOURNAL OF BIOLOGICAL CHEMISTRY Q 1992 by The American Society for Biochemistry and Molecular Biology, Inc. Vol

Avid Binding of Modified Albumins to gp30 and gp18 24553

inhibition profiles suggest that gp60, gp30, and gp18 play distinctly different roles in native and modified albumin bind- ing and transport.

The differences in tissue and cell distribution of the albu- min-binding proteins also suggest different functions for these proteins. The morphological studies discussed above show that: (i) continuous, fenestrated, and sinusoidal endothelium can endocytose and apparently degrade A-Au; and (ii) a variety of organs have endothelia which bind A-Au (1, 2, 26- 28,39,40). These observations are much more consistent with the tissue distribution of gp30 and gp18; which we found in many different organs, and not gp60, which appears to be expressed selectively in certain tissues with continuous en- dothelium (11). Since albumin degradation occurs in many tissues and the vascular wall (the endothelium and possibly other associated cells) could be a major site for albumin endocytosis and catabolism (26,27,40,42,43), gp30 and gp18, which clearly bind conformationally altered albumins, may be involved primarily in this process. In this study, we show that RFC cells process A-Au and Mal-BSA in a manner different from that of native BSA. Both Mal-BSA and A-Au bound at the cell surface were internalized and degraded, whereas sur- face-bound BSA was not degraded. These results are consist- ent with both the work in situ demonstrating internalization and lysosomal targeting of A-Au bound at the cell surface (2, 26-28, 39, 40) and a possible role in albumin catabolism. Fibroblasts (NRK-F) also degrade lZ5I-A-Au and its degrada- tion is inhibited by Mal-BSA but not BSA (44).6 These in vitro studies indicate that conformationally altered albumin is preferentially degraded and suggest that examining the cellular processing of these albumin probes may be useful in differentiating the role of albumin-binding proteins and plas- malemmal vesicles in endocytosis and transcytosis.

Our observations that gp30 and gp18 are present in many cells and tissues along with their preferential interaction with conformationally modified albumins seem to decrease the likelihood that gp30 and gp18 have a primary role in the specific binding and transcytosis of native albumin which appears to occur only in tissues with continuous endothelium. However, this evidence does not preclude some involvement of gp30 and gp18 in albumin binding and even possibly transcytosis, especially in light of the high albumin concen- tration in plasma. With the data available at this time, it would appear that gp30 and gp18 may function preferentially in endocytosis and degradation of various modified albumins in a variety of cells associated with the vascular wall, whereas the role of gp60 in the binding and transcytosis of native albumin may be limited to certain continuous endothelia.

Although gp30 and gp18 are expressed by cultured RFC cells, it is not established at this time that all endothelial cell types express gp30 and gp18. The apparent ubiquitous tissue distribution of gp30 and gp18 may result from the expression by fibroblasts and/or smooth muscle cells and not in each case by endothelium.

REFERENCES 1. Ghitescu. L.. Fixman. A.. Simionescu. M.. and Simionescu. N. (1986) J.

Cell Bid. 102,1304-1311

Cell Biol. 106,2603-2612

68,87-97

, , , . , 2. Milici, A. J., Watrous, N. E., Stukenbrok, H., and Palade, G. E. (1987) J.

3. Levick, J. R., and Michel, C. C. (1973) Q. J. Exp. Physiol. Cogn. Med. Sci.

4. Michel, C. C. (1980) J. Physiol. 3 0 9 , 341-355 5. Mann, G. E. (1981) J. Physiol. 319,311-323 6. Gamble, J. (1983) J. Physiol. 339,505-518 7. Schneeberger, E. E., and Hamelin, M. (1984) Am. J. Physiol. 247 , H206-

u917 8. 9.

10.

11. 12. 13.

Hiire;', V. H., and Curry, F. E. (1985) Am. J. Physiol. 2 4 8 , H264-H273 Yokota, S. (1983) Biomed. Res. 4 , 577-586 Schnitzer, J. E., Carley, W. W., and Palade, G. E. (1988) Am. J. Physiol.

Schnitzer, J. E. (1992) Am. J. Physiol. 262 , H246-H254 Peters, T., Jr. (1985) Adu. Protein Chem. 37,161-245 Galis, Z., Ghitescu, L., and Simionescu, M. (1988) Eur. J. Cell Biol. 4 7 ,

2 6 4 , H425-H437

358-365

Hiire;', V. H., and Curry, F. E. (1985) Am. J. Physiol. 2 4 8 , H264-H273 Yokota, S. (1983) Biomed. Res. 4 , 577-586 Schnitzer, J. E., Carley, W. W., and Palade, G. E. (1988) Am. J. Physiol.

Schnitzer, J. E. (1992) Am. J. Physiol. 262 , H246-H254 Peters. T.. Jr. (1985) Adu. Protein Chem. 37.161-245

2 6 4 , H425-H437

Galis, Z., Ghit;?scu, 'L., and Simionescu, M.'(1988) Eur. J. Cell Biol. 4 7 , 358-365

14. Cur&, F . E . (1985) Fed. Proc. 44,2610-2613 15. Schnitzer, J. E., Carley, W. W., and Palade, G. E. (1988) Proc. Natl. Acad.

16. Ghinea, N., Fixman, A,, Alexandru, D., Popov, D., Hasu, M., Ghitescu, L., Sci. U. S. A. 86,6773-6777

Eskenasv. M.. Simionescu. M.. and Simionescu. N. (1988) J. Cell Biol.

17.

18.

19.

20.

107,23i1239 Ghinea, N., Eskenasy, M., Simionescu, M., and Simionescu, N. (1989) J.

Schnitzer, J. E., Shen, C.-P. J., and Palade, G. E. (1990) Eur. J . Cell Biol. Biol. Chem. 264,4755-4158

Schnitzer, J. E., Ulmer, J. B., and Palade, G. E. (1990) Proc. Natl. Acad. 62,241-251

Taylor, A. E., and D. N. Granger. (1984) in Handbook of Physiology (Renkin, Sci. U. S. A. 87,6843-6847

E. M.. and Michel. C. C.. eds) DD. 467-520. American Phvsioloeical

. , , . .

I . .~ Society, Bethesda, MD

21. Simionescu, M., and Simionescu, N. (1984) in Handbook of Physiology (Renkin, E. M., and Michel, C. C., eds) pp. 41-101, American Physiolog- ical Society, Bethesda, MD

22. Horiuchi, S., Takata, K., and Morino, Y. (1985) J. Biol. Chem. 260,475-

23. Butler, P. J. G., Harris, J. I., Hartley, B. S., and Leberman, R. (1969) 481

24. Habeeh, A. F. S. A. (1966) A d . Biochem. 14,328-336 Biochem. J. 112,679-689

25. Schneeberger, E. E., Lynch, R. D., and Neary, B. A. (1990) Am. J. Physiol.

"

26. Geoffrey, J. S., and Becker, R. P. (1984) J. Ultrastruct. Res. 89, 223-239 27. De Bruvn. P. P. H.. Michelson. S.. and Bankston. P. W. (1985) Cell Tissue

2 6 8 , L89-L98

28.

29.

30. 31.

32.

33.

34.

35.

36.

37.

38. 39.

40.

41.

42.

43.

44. 45. 46.

Bumbasirevic, V., Pappas, G. D., and Becker, R. P. (1990) J. Submicrosc. Res. 2 4 0 , l - 7

AnJade, J. D.'(1985) Surface and Interfacial As ects of Biomedical Poly- C tol Patbl 22,135-145

Reed, R. G., and Burrm on, C. M. (1989) J. Biol. Chem., 264,9867-9872 mers: Protein Adsorption, Plenum Press, New eork

Boeynaems, J. M., and gumont, J. E. (1980) Outline of Receptor Theory, pp. 1-120,165-200, Elsevier North-Holland Biomedical Press, New York

Horiuchi, S., Murakami, M., Takata, K., and Morino, Y. (1986) J. Biol. Chem. 261,4962-4966

Haberland, M. E., and Fogelman, A. M. (1985) Proc. Natl. Acad. Sci. U. S. A. 82,2693-2697

Means, G. E., and Feeney, R. E. (1971) Chemical Modification of Proteins, p. 125, Holden-Day Publishers San Francisco

Pardri e, W M , Elsenberg, J., 'and Cefalu, W. T. (1985) Am. J. Physiol. 249S2641E267

Simionescu, M., Ghinea, N., Fixman, A., Lasser M., Kukes, L., Simionescu, N., and Palade, G. E. (1988) J. Submierosc. Cytol. Pathol. 20,243-261

Sparrow, C. P., Parthasarathy, S., and Steinberg, D. (1989) J. Biol. Chem. 264,2599-2604

Bravo, J., Suni? A;, and Schnitzer, J. (1992) FASEB J. 6, A1322 Predescu, D., lmlonescu, M., Simionescu, N., and Palade, G. E. (1988) J .

Villaschi, S., Johns, L., Cirigliano, M., and Pietra, G. G. (1986) Microuasc. Cell Biol. 107 , 1729-1738

Res. 32. 190-199 Sifl-<n.er-Birnboim A,, Schnitzer J. E Lum H., Blumenstock F. A. Shen

C:%. J., Del VecAhio, P. J., and Maiik, A. B. (1991) J. Cell. hys io i . 149: 575-584

Waldmann, T. A. (1977) in Albumin Structure, Function and Uses (Rosen- oer, V. M., Oratz, M., and Rothschild, M. A., eds) pp. 255-273, Pergamon

Yedgar, S., Carew, T. E. Pittman R. C Beltz, W. F., and Steinberg, D. Press, New York

Bravo, J., Sun A , and Schnitzer, J. (1992) FASEB J. 6, A1322 (1983) Am. J. Physiol. 2 4 4 , Eldl-ElO?

Schnitzer, J. E!: and Horvat, R. (1991) FASEB J. 6 , A529 Curry, F. E., Michel, C. C., and Phillips, M. E. (1987) J. Physiol. 387 , 69-

". - - - - - -

82