Vol. 59, No. 2INFECTION AND IMMUNITY, Feb. 1991, p. 671-6780019-9567/91/020671-08$02.00/0Copyright © 1991, American Society for Microbiology

Intracellular Localization of Borrelia burgdorferiwithin Human Endothelial CellsYING MA,' ANNE STURROCK,2 AND JANIS J. WEIS'*

Division of Cell Biology and Immunology, Department of Pathology,' and Division ofPulmonary Medicine,Department of Medicine,2 University of Utah School of Medicine, Salt Lake City, Utah 84132

Received 26 June 1990/Accepted 14 November 1990

The later stages of infection by the Lyme disease pathogen, Borrelia burgdorferi, are characterized by thepersistence of the organism in individuals possessing a strong anti-Borrelia immune response. This suggests thatthe organism is sequestered in a tissue protected from the immune system of the host or there is a reservoir ofthe organism residing within the cells of the host. In this report, the ability of B. burgdorfieri to gain entranceinto human umbilical vein endothelial cells was explored as a model for invasion. Incubation of B. burgdorferiwith human umbilical vein endothelial cells at ratios ranging from 200:1 to 5,000:1 resulted in the intracellularlocalization of 10 to 25% of B. burgdorfieri in 24 h. The intracellular location of the spirochetes was

demonstrated by the incorporation of radiolabeled B. burgdorferi into a trypsin-resistant compartment and was

confirmed by double-immunofluorescence staining which differentiated intracellular from extracellular organ-isms. Actin-containing microfflaments were required for the intracellular localization, indicating that the hostcell participates in the internalization process. Activation of endothelial cells by agents known to increase theexpression of several adhesion molecules had no effect on the interaction of B. burgdorfieri with the endothelialmonolayer. This indicates that the endothelial receptor for B. burgdorfieri is constitutively expressed and thatinternalization is not dependent upon adhesion molecules whose expression is induced by inflammatorymediators. The demonstration of B. burgdorfieri within endothelial cells suggests that intracellular localizationmay be a potential mechanism by which the organism escapes from the immune response of the host and maycontribute to persistence of the organism during the later stages of Lyme disease.

Lyme disease is caused by the tick-borne spirocheteBorrelia burgdorferi (8, 19). Infection by this organismcauses multiple and varied symptoms which have beencategorized into three stages (17, 27, 29). The first stageinvolves the initial symptoms of the disease includingerythema chronicum migrans at the site of the tick bite andmore generalized flulike symptoms. The second and thirdstages of disease occur weeks to months after the first stageand can involve several different organs including cardiactissue, the central nervous system, and joints. In general,infected humans and animals experiencing symptoms ofstage II and III disease have an active anti-Borrelia immuneresponse including both humoral and cellular immunity.Antibiotic therapy resolves stage II and III symptoms inmost individuals, indicating that these later stages of diseaseare due to persistence of the organism. The ability of thespirochete to maintain infection in the presence of a specificimmune response suggests that the organism may hide fromthe defenses of the host in the later stages of infection. Theneurological involvement seen in some individuals and theimproved success in treatment of patients with stage II andIII disease with ceftriaxone, an antibiotic which crosses theblood-brain barrier, implies that the central nervous systemmay serve as a protective niche (11). Other investigatorshave shown that B. burgdorferi can transcytose an endothe-lial monolayer, providing a mechanism by which thesespirochetes could invade a variety of tissues (10, 30). It hasalso been speculated that B. burgdorferi resides within cellsof certain tissues during or between stages of disease, whichcould provide a second mechanism contributing to persis-tence by the organism (5, 26, 27).

* Corresponding author.

In this study we have addressed the possibility of B.burgdorferi localization within mammalian cells by deter-mining if the spirochete could be demonstrated within hu-man endothelial cells. Endothelial cells were chosen becausespirochetes can penetrate endothelial monolayers and ad-here to endothelial cells in vitro (10, 30, 31) and becausedamage to the endothelium has been observed in animals andpatients infected with B. burgdorferi (12, 28). Spirocheteinternalization by endothelial cells was demonstrated by theuptake of radiolabeled bacteria into a trypsin-resistant endo-thelial cell compartment. This was verified by visualizationof the bacterium within the endothelial cell.

MATERIALS AND METHODS

Bacteria. Low-passage frozen stocks of the JD-1 strain ofB. burgdorferi, provided by Sam Telford and Jose Ribeiro ofthe Harvard School of Public Health, were maintained at-70°C (23). Frozen stocks were seeded in 16 ml of BSK-II(1) containing 40 ,uCi of [3H]adenine and cultured for 3 to 4days at 32°C until reaching a density of approximately 107spirochetes per ml. This gave optimal labeling of spiro-chetes, with approximately 5,000 cpm incorporated per 107spirochetes. A derivative of B. burgdorferi SH2-82 that hasbeen passaged 300 times in the laboratory and is no longerinfective for mice was obtained from Tom Schwan of theRocky Mountain Laboratories (26).

Endothelial cells. Human umbilical vein endothelial cells(HUVECs) were isolated from human umbilical cords by thestandard procedure that uses collagenase digestion (18) andwere cultured in endothelial growth medium (Clonetics) ontissue culture flasks coated with 2% gelatin. HUVECs up topassage six were used in internalization experiments. Virtu-ally 100% of HUVECs were able to internalize fluorescein-

671

on January 20, 2021 by guesthttp://iai.asm

.org/D

ownloaded from

672 MA ET AL.

ated, acetylated low-density lipoprotein (LDL), a ligandbound specifically by the scavenger receptor of endothelialcells and macrophages (33), thus ensuring that the HUVECcultures were free from smooth-muscle or other contaminat-ing cell types.

Internalization assay. Internalization of B. burgdorferi byendothelial cells was measured in an assay that allowsdiscrimination between spirochetes that are tightly adherentto the extracellular surface of the HUVECs from thoseinside the HUVECs. The low-passage JD-1 strain of B.burgdorferi was labeled by growth for 3 to 5 days in BSK-IIcontaining 2.5 ,uCi of [3H]adenine per ml and washed twicewith RPMI 1640 containing 20% heat-inactivated fetal calfserum (RPMI-FCS). Under optimal conditions, the spiro-chetes were labeled to a specific activity of 2 x 103 bacteriaper cpm. The spirochetes were counted, and the indicatednumber was added to the cultured HUVECs. HUVECs wereseeded at 2 x 105 cells per 2-cm2 well in a 24-well cluster dishand allowed to form a confluent monolayer overnight. Theindicated number of spirochetes was added to each well, andsamples were incubated at 37°C for the indicated times. Fourwells were prepared for each sample. Supernatant and cellsfrom two wells for each sample were pooled to have ade-quate cells for a pellet. This allowed duplicate data points tobe collected for each sample. Samples were harvested byremoving the supernatants, rinsing the surface of cells withRPMI 1640 lacking fetal calf serum, and pooling the super-natant and washes for determination of counts per minute ofthe supernatant. Trypsin was added to the HUVEC mono-layers to loosen the cells from the dish and to releaseadherent extracellular bacteria from the HUVECs. Thewells were washed with RPMI-FCS, and the washes werepooled with the trypsin solution to inactivate the trypsin.The HUVECs were then washed twice with RPMI-FCS bycentrifugation at 200 x g, conditions which pellet HUVECsbut not spirochetes. The supernatants from these centrifu-gations were pooled and counted for the determination oftrypsin-sensitive counts representing extracellular spiro-chetes. The HUVEC cell pellets were also counted fordetermination of trypsin-resistant counts representing intra-cellular spirochetes. The sum of trypsin-sensitive andtrypsin-resistant counts per minute represent the totalcounts of cell-associated spirochetes.

Antisera. Immune rabbit sera was raised by twp immuni-zations, separated by 7 days, with 2 x 108 B. burgdorferi thathad been killed by heating at 56°C for 30 min. After 10 days,the rabbit was bled at weekly intervals for 4 weeks and thetiter for immunofluorescence staining was determined to beat least 1:250. Fluorescein isothiocyanate (FITC),conjugatedgoat anti-rabbit immunoglobulin G(IgG) and rhodamine-conjugated goat anti-rabbit IgG were purchased from TAGO(Burlingame, Calif.). The monoclonal antibodies H5332 (rec-ognizing OspA [5]), H5TS (recognizing OspB [4]), andH9724 (recognizing flagellin [3]) were obtained from TomSchwan.

Microscopic demonstration of intracellular organisms. Dou-ble-staining experiments were designed to discriminate in-tracellular from extracellular bacteria as described previ-ously for invasion of mammalian cells by Listeriamonocytogenes (25). HUVECs were cultured on gelatin-coated glass coverslips placed in six-well cluster dishes, andB. burgdorferi was added at a concentration of 107/ml andallowed to bind and infect during a 24-h incubation period at37°C. Coverslips were washed with phosphate-buffered sa-line (PBS) to remove loosely adherent bacteria and fixedwith 4% Formalin for 20 min at room temperature. Cover-

>

0.

0 2 3 4

Time (days)FIG. 1. Time course for binding and internalization of B. burg-

dorferi by HUVECs. A total of 2 x 108 labeled spirochetes wasincubated with approximately 2 x 105 adherent HUVECs per well ofa 24-well cluster dish. Cell-associated (U) and trypsin-resistant (0)counts per minute were determined as described in Materials andMethods. Duplicate samples were included for each time point. Theactual numbers for the 24-h time point were 46,763 + 272 cpm in thesupernatant, 1,290 + 59 cpm in the trypsin-sensitive fraction, and6,084 ± 140 cpm in the trypsin-resistant fraction.

slips were incubated in a 1:250 dilution of rabbit anti-B.burgdorferi antiserum for 30 min. All antibody incubationswere for 30 min at room temperature, and all antibodies werediluted 1:250 with 20% normal human serum in PBS. Cov-erslips were washed in PBS and incubated in FITC-conju-gated goat anti-rabbit IgG for 30 min. Coverslips werewashed in PBS, and plasma membranes of the endothelialcells were permeabilized by incubation with methanol for 2min. After an additional PBS rinse, coverslips were stainedwith the same rabbit anti-B. burgdorferi antiserum used inthe first step, washed, and stained with rhodamine-conju-gated goat anti-rabbit IgG. In this protocol, any spirochetesadherent to the extracellular surface of the HUVECs willstain with both antibodies, whereas those bacteria within thecell are not stained prior to permeabilization and only stainwith rhodamine. Samples were visualized with a Zeissfluorescent microscope equipped with Epi fluorescence.

RESULTS

Demonstration of intracellular localization within HUVECsby B. burgdorferi. Several experiments have been performedwhich demonstrate that B. burgdorferi adhere and invadehuman endothelial cells. The results shown in Fig. 1 are froma typical experiment and demonstrate that both cell associ-ation and internalization continue to increase until 48 h,when both are maximal. In the experiment shown in Fig. 1,14% of the added spirochetes were adherent at 48 h and 11%were intracellular by this time. After a 24-h incubationperiod, 50 to 75% of spirochetes associated with HUVECshad reached a trypsin-resistant, intracellular location.An experiment was performed to establish that spiro-

chetes incubated in tissue culture medium were not clumpingtogether and forming large complexes that would pelletunder the conditions used to separate spirochetes fromHUVECs. In this experiment, spirochetes were incubatedfor 24 h in wells containing endothelial growth medium but

INFECT. IMMUN.

on January 20, 2021 by guesthttp://iai.asm

.org/D

ownloaded from

INTRACELLULAR LOCALIZATION OF B. BURGDORFERI 673

no HUVECs, centrifuged, and washed alone or mixed withHUVECs prior to centrifugation and washes. Of the spiro-chete counts, 3% was found in the pellet fraction when spunby themselves and 4% was found when the spirochetes werespun with HUVECs. In the experimental samples (Fig. 1 to5), the nonassociated spirochetes were removed first andcounted as the supernatant fraction. Of the remaining cell-associated spirochetes, only 4% could be placed in thetrypsin-resistant fraction because of trapping of spirochetesby the larger HUVEC cells or pelleting of spirochetes duringthe low-speed centrifugation. This is between 0.5 and 2% ofthe total spirochete counts added to the wells in the samplesin Fig. 1 to 5.

Effect of agents that alter endothelial cell function oninternalization of B. burgdorferi. Further experiments wereperformed to characterize the invasion of HUVECs byspirochetes. Tumor necrosis factor (TNF) is known toenhance the binding of leukocytes to endothelial cells by itsability to rapidly increase endothelial expression of intercel-lular adhesion molecule 1 (24) and to induce synthesis andexpression of endothelial cellular adhesion molecule 1 (6)and vascular cell adhesion molecule 1 (21) with a timeoptimum of 2 to 4 h. Since leukocyte binding to intercellularadhesion molecule 1 and endothelial adhesion molecule 1 onendothelial cells is followed by penetration of the endothe-lium, it is possible that spirochetes might follow a similarmechanism involving the same endothelial cell receptors.HUVECs were incubated with 100 U ofTNF (Genentec) perml or with media alone for 3 h at 37°C. Radiolabeled B.burgdorferi were then added to the HUVEC monolayers,and the monolayers were incubated for 30 min, 4 h, and 24 hto determine if increased expression of either of thesereceptors could enhance the endothelial cell binding and/orinternalization of B. burgdorferi. In a parallel experimentperformed with the same batch of HUVECs and TNF, thepercentage of neutrophils bound by endothelial cells in-creased from 14 to 79% with TNF treatment, indicating thatreceptors for neutrophil adhesion had been induced. At earlytime points, TNF had no effect on the binding or internal-ization by HUVECs of B. burgdorferi, while at 24 h theinternalization was reduced by about one third (Fig. 2). Theeffect at 24 h is probably not the result of reduced expressionof receptors but rather the result of destabilizing effects ofTNF on cytoskeletal structures, as has been reported forbovine aortic endothelial cells (7). This is consistent withfindings presented in Fig. 3 (discussed below) in whichdisruption of the cytoskeleton prevented internalization ofthe spirochetes. The experiment presented in Fig. 2 providesstrong evidence that none of the TNF-inducible adhesionmolecules on endothelial cells serve as receptors for B.burgdorferi binding or internalization.

Stimulation of endothelial cells with histamine or withphorbol esters induces the rapid expression of the neutrophiladhesion molecule GMP-140 from preformed intracellularpools (16). Incubation of endothelial cells with 60 ng ofphorbol myristate acetate (Sigma) per ml for 30 min in Hanksbalanced salt solution containing 0.5% albumin, conditionswhich stimulate endothelial cell binding of neutrophils, hadno effect on the binding or internalization of B. burgdorferi(data not shown), suggesting that GMP-140 is not a receptorfor this organism.We next determined if inhibition of actin polymerization

with cytochalasin D could inhibit HUVEC internalization ofB. burgdorferi. A 12-h pretreatment of HUVECs with 1 agof cytochalasin D (Sigma) per ml completely blocked inter-nalization of spirochetes (Fig. 3). Cytochalasin D also

Z 20

0

0 0 20 30

Time (hrs)

FIG. 2. Effect of TNF on the binding and internalization of B.burgdorferi by HUVECs. Adherent HUVECs were treated with 100U of TNF per ml for 2 h (0, cell associated; O, trypsin resistant) orwith medium alone (0, cell associated; U, trypsin resistant) prior tothe addition of 2 x 108 spirochetes per well. Duplicate samples wereincluded for each time point, and values were determined asdescribed in Materials and Methods.

greatly reduced the binding of spirochetes to HUVECs,suggesting either that normal binding and internalizationinvolves receptor recycling which cannot occur in treatedcells or that clustering of receptors on the plasma membraneis required for optimal spirochete binding and that this isinhibited by cytochalasin D. The complete inhibition oflocalization of bacteria to a trypsin-resistant compartmentby cytochalasin D provides further evidence that thesebacteria were indeed intracellular rather than attached to theplasma membrane via a trypsin-insensitive structure. These

40

30

20

0

CA

-0;

v D's

*

co1; *

L.

50

Time (hrs)FIG. 3. Effect of cytochalasin D treatment of HUVECs on the

binding and internalization of B. burgdorferi. Endothelial cell mono-layers were treated with 1 ,ug of cytochalasin D per ml (0, cellassociated; 0, trypsin resistant) or with medium alone (U, cellassociated; 0, trypsin resistant) for 12 h prior to the addition of 4 x

107 labeled spirochetes. Duplicate samples were included for eachtime point, and values were determined as described in Materialsand Methods.

VOL. 59, 1991

on January 20, 2021 by guesthttp://iai.asm

.org/D

ownloaded from

674 MA ET AL.

30 -

20

10

o

T

Cell Assoc Trp Res Cell Assoc Trp Res

4 hours L 24 hours

FIG. 4. Effect of rabbit antisera on the binding and uptake of B.burgdorferi. Binding of 3.5 x 107 B. burgdorferi to HUVECscultured in 24-well cluster dishes was measured in the presence of a1:100 dilution of normal rabbit serum (solid bars) or of rabbit anti-B.burgdorferi (hatched bars). Samples were harvested at 4 and 24 h,and the percentage of cell-associated or trypsin-resistant counts perminute was determined as described in Materials and Methods.Values represent the average of duplicates, and the variation be-tween the samples is indicated by the error bars.

findings also indicate that the host is an active participant inthe internalization process and that the spirochetes maymake use of the actin polymers of the host in their uptake.Many invasive bacteria do require intact microfilaments forinvasion, and, thus, this finding is consistent with invasion ofmammalian cells by other pathogenic organisms (14, 32).The ability of trypsin to completely remove cell-associated

bacteria from the cytochalasin D-treated endothelial cells(Fig. 3, bottom line) demonstrates that this protocol success-fully removes extracellular bacteria from endothelial cellsand validates its usefulness in quantification of intracellularorganisms. These findings are further supported by thefluorescent micrographs in Fig. 6E and F and in Table 1, inwhich all spirochetes were extracellular in slides preparedfrom cytochalasin D-treated endothelial cells. CytochalasinD appears to have no effect on the viability of the spiro-chetes, as the addition of 1 ,ug of cytochalasin D per ml tobacterial cultures for 24 h had no effect on the percentage ofspirochetes that were motile (>95%).

Inhibition of internalization by rabbit anti-B. burgdorferiantiserum. The specificity of the interaction between B.burgdorferi and HUVECs was analyzed by testing for theability of rabbit polyclonal antibody to B. burgdorferi toblock binding and internalization. The addition of rabbitanti-B. burgdorferi antiserum reduced binding to 67% of thecontrol, incubated with normal rabbit serum, and reducedinternalization to 34% of the control at the 4-h time point(Fig. 4). At the 24-h time point, no difference was detectedbetween antibody-treated samples and controls in the exper-iment shown in Fig. 4, while in some experiments slightenhancement in the antibody-treated samples was observedat this time. This could be the result of aggregation of thebacteria by the antibody, leading to enhanced uptake andmasking of the inhibitory effect seen at the earlier time

points. Monoclonal antibodies directed against OspA (5),OspB (4), and flagellin (3) had no effect on binding orinternalization (data not shown); however, others haveshown that two different monoclonal antibodies to OspB canpartially block adherence to HUVECs (31).

Cell specificity of internalization. The specificity of uptakewas examined by analysis of different host cells. Twodose-response experiments are shown in Fig. 5 in which thebinding and internalization by HUVECs and HeLa cells ofincreasing numbers of B. burgdorferi were measured.Greater that 60% of those spirochetes associated with HU-VECs were intracellular for each dose of spirochetes in Fig.5A. This is in contrast to the binding to HeLa cells (Fig. SB),where B. burgdorferi bound very well, but only 10 to 15% ofspirochetes that bound to HeLa cells were able to achieve anintracellular location. This represented 2 to 5% of the totalnumber of spirochetes added to the cultures. This suggeststhat B. burgdorferi may bind to distinct receptors on dif-ferent cell types, only some of which are involved ininternalization, or that HeLa cells do not have the cytoskel-etal structures required for the spirochete to invade. Otherinvasive bacteria can invade HeLa cells (13, 34), indicatingthat this cell type can serve as a host for intracellularlocalization by some bacteria.

Strain specificity of bacterial uptake. A derivative of B.burgdorferi Sh-2-82, which has been passaged 300 times inthe laboratory and shown to have lost the ability to infectwhite-footed mice, was tested for the ability to achieve anintracellular localization (26). After a 24-h incubation periodwith HUVECs, 23% ± 0.2% of labeled spirochetes wereassociated with endothelial cells and 15% ± 0.1% of theorganisms were intracellular. Therefore, the ability toachieve an intracellular localization is a property shared byboth virulent and avirulent strains of B. burgdorferi. This isin further agreement with the endocytic nature of the inter-nalization process.

Fluorescent-antibody staining of intracellular spirochetes.Double-staining techniques were used in which spirochete-infected endothelial cells were stained before and afterpermeabilization with methanol (25). Apparently intact spi-rochetes staining with both FITC- and rhodamine-conju-gated goat anti-rabbit IgG were observed (Fig. 6A to D), aswere spirochetes staining only with rhodamine-conjugatedsecond antibody (Fig. 6B and D, arrows). Brightly stainedintracellular debris was occasionally observed and was pre-sumed to represent fragments of intracellular spirochetes.Neither intact spirochetes nor brightly staining speckleswere identified in HUVECs not exposed to spirochetes (datanot shown), confirming that the intracellular speckles weredue to specific staining of spirochetal antigens rather thantrapping of conjugated antibody by the HUVECs. Manyspirochetes were found adhering to cytochalasin D-treatedHUVECs; however, they always stained with both antibod-ies (Fig. 6E and F). This is consistent with the finding that allradiolabeled spirochetes adherent to cytochalasin D-treatedendothelial cells could be released by trypsin. The data froma representative experiment are presented in tabular form inTable 1. The fluorescent antibody studies permitted a muchlower ratio of spirochetes to endothelial cells (approximately30:1) than did the experiments assessing uptake of radiola-beled bacteria. This difference and differences in the integ-rity of the monlayer obtained with endothelial cells plated onglass coverslips rather on plastic are probably responsiblefor the lower ratio of spirochetes to endothelial cells de-tected in the stained samples relative to the results withuptake of radiolabeled spirochetes. The results of the fluo-

u

-la

*)

(Um

4)

u4

INFECT. IMMUN.

on January 20, 2021 by guesthttp://iai.asm

.org/D

ownloaded from

INTRACELLULAR LOCALIZATION OF B. BURGDORFERI 675

A.

40 a

.3

'Is

1-uW

a.

u,

u .

30

20

10,

0 4-10

B.

0.5 10 0.1

Sprochetes x 108 Spirochetes x 10

FIG. 5. Effect of spirochete number on the binding and internalization of B. burgdorferi by HUVECs and HeLa cells. Binding of theindicated numbers of labeled spirochetes to HUVECs (A) and HeLa cells (B). Cell-associated (solid bars) and trypsin-resistant (hatched bars)counts per minute were determined as described in Materials and Methods. Values represent the average of duplicates, and the variationbetween the samples is indicated by the error bars.

rescent antibody staining (Fig. 6 and Table 1) provideconfirmation of the intracellular localization of B. burgdor-feri within normal endothelial cells and the requirement forhost microfilaments for this localization.

DISCUSSION

The achievement of an intracellular location within mam-malian cells is a potent virulence factor for many bacterialspecies, including many which grow well in the extracellularenvironment. These include Yersinia spp., Salmonella spp.,Shigella spp., invasive Escherichia coli, and L. monocyto-genes (15). Invasion by many of these organisms is associ-ated with intracellular multiplication and cell-to-cell spread-ing. This may also result in invasion of other tissues(Salmonella spp. and L. monocytogenes) (14, 32) or mayenhance survival in the localized site of infection (Shigellaspp.) (9). Other organisms appear to invade without repli-cating (Yersinia and Bordetella spp.) within the invaded celland may use invasion as a mechanism of passing from onetissue to another (13-15). The ability to invade cells providesthe bacterium with a mechanism by which to escape destruc-tion by the innate and specific aspects of the immuneresponse of the host. An intracellular location may alsoprovide protection from antibiotics because the concentra-tion of many antibiotics within cells will be less than that inthe extracellular milieu (15, 25).Because of the virulence potential of host cell invasion, we

wanted to know if B. burgdorferi could reside within amammalian cell. We wished to demonstrate this as a prop-erty distinct from the ability to transcytose an endothelialcell monolayer. Our results demonstrate that B. burgdorferican achieve an intracellular localization in vitro and suggeststhat intracellular localization may occur in vivo. The abilityto localize within tissue cells may allow the spirochete,which has a very slow division time (10 to 12 h under optimal

conditions), a real opportunity to avoid killing by antibiotics,antibody-plus-complement-mediated lysis, or antibody-me-diated phagocytosis, and allow survival of low levels oforganisms for prolonged periods of time. One characteristicof the internalization process is a requirement for intact hostcell microfilaments. This is consistent with the requirementsfor invasion by several other pathogenic bacteria and sug-gests that the host cell is actively participating in theinvasion process (15). Others have referred to this type ofevent as "parasite directed endocytosis," reflecting theparticipation of both the microbe and the host cell in theinternalization process (20). The tissue normally serving as areservoir in vivo is not known, but the contrast in efficiencybetween the HeLa cell line and HUVECs suggests that somecell specificity will exist.

In the experiments presented in Fig. 1 to 5, the lowspecific activity of the metabolically labeled bacteria re-quired a relatively high ratio of spirochetes to endothelialcells to measure internalization. The validity of these find-ings was confirmed by immunofluorescent analysis (Fig. 6and Table 1) in which the binding and internalization of farfewer spirochetes could be measured. In both types ofanalysis, 50 to 75% of spirochetes associated with endothe-lial cells were intracellular and treatment with cytochalasinD completely prevented the intracellular localization ofbacteria. A further indication of the importance of the host inmediating the uptake of this organism comes from the findingthat noninfective organisms are internalized and that heat-killed spirochetes which form large clumps are also readilyinternalized.The existence of an intracellular reservoir of viable B.

burgdorferi could explain the clinical observations of Lymedisease, in which the bacteria can be demonstrated in lownumbers in infected tissues and synovial fluids but aredifficult for the host to eliminate (17, 27, 29). The slowgrowth of the organism further suggests that a small number

VOL. 59, 1991

on January 20, 2021 by guesthttp://iai.asm

.org/D

ownloaded from

676 MA ET AL.

FIG. 6. Intracellular localization of B. burgdorferi within cultured endothelial cells. Endothelial cell monolayers were cultured on gelatin-coated glass coverslips in six-well cluster dishes and incubated with 107 B. burgdorferi for 24 h at 37°C in the absence (A to D) or presence(E and F) of cytochalasin D. Coverslips were fixed with Formalin and stained with rabbit anti-B. burgdorferi antiserum and FITC-conjugatedgoat anti-rabbit IgG. Endothelial cells were then permeabilized with methanol and stained with rabbit anti-B. burgdorferi antiserum andrhodamine-conjugated goat anti-rabbit IgG. Photomicrographs of spirochetes stained with the FITC-conjugated antibody (A, C, and E) or withrhodamine-conjugated antibody (B, D, and F) are shown. Panels A and B are of the same region of the slide, panels C and D are of the sameregion of a different slide, and panels E and F are the same region of the cytochalasin D control slide. Arrows in panels B and D indicateintracellular spirochetes staining only with the rhodamine-conjugated antibody. All spirochetes detected in panel E were also detected in panelF.

of bacteria might survive for some time within a cell beforeaccumulating sufficient density of spirochetal peptideswithin class I or class II antigen, presenting structures tocause the host cell to be recognized as foreign. Thus, thedemonstration of B. burgdorferi within human cells would

provide a model for studying the role of intracellular local-ization on the virulence of this organism. A further indica-tion that the organism might be invading and survivingwithin cells of infected tissue comes from the recent dem-onstration of viable B. burgdorferi in cardiac tissue from a

INFECT. IMMUN.

on January 20, 2021 by guesthttp://iai.asm

.org/D

ownloaded from

INTRACELLULAR LOCALIZATION OF B. BURGDORFERI 677

TABLE 1. Quantification of intracellular B. burgdorferi by indirect immunofluorescent microscopy

Total no. of spirochetesSample ~~~~~~~~~~~~~~~~~~~~~~~~%SpirochetesSample Endothelial Spirochetes stained with FITC- Spirochetes stained with rhodamine- intracellulard

cellsa conjugated anti-rabbit IgGb conjugated anti-rabbit IgGc

Spirochetes + HUVECs 586 125 300 58Spirochetes + cytochalasin D-treated HUVECs 466 183 184 0.5

a Total number of endothelial cells in 20 distinct microscopic fields. Spirochete counts were made from the same 20 fields.b Total number of spirochetes that stained prior to endothelial cell permeabilization with rabbit anti-B. burgdorferi antiserum and FITC-conjugated goat

anti-rabbit IgG (detects extracellular bacteria).' Total number of spirochetes that stained after endothelial cell permeabilization with rabbit anti-B. burgdorferi antiserum and rhodamine-conjugated goat

anti-rabbit IgG (detects both intracellular and extracellular bacteria).d {[(Spirochetes stained with rhodamine-conjugated anti-rabbit IgG) - (Spirochetes stained with FITC-conjugated anti-rabbit IgG)]/[Spirochetes stained with

rhodamine-conjugated anti-rabbit IgGJ} x 100.

patient with dilated cardiomyopathy and the apparent local-ization of at least one spirochete within a myocardial cell (12,28).There are two reports documenting another important

interaction between B. burgdorferi and human endothelialcells, i.e., the ability to transcytose monolayers of endothe-lial cells (10, 30). These findings are important as theyprovide a mechanism by which the spirochete can escapefrom the blood and move into a variety of tissues in theinfected animal. In one report, it was suggested that B.burgdorferi interacts with fibronectin in the subendothelialmatrix (30); however, the bacteria used in that study werelabeled with [3H]thymidine, which others have shown not tobe incorporated into B. burgdorferi (22). In that study it ispossible that much of the [3H]thymidine was associated withthe surface of the bacterial cell instead of metabolicallyincorporated into bacterial DNA; therefore, it is difficult tointerpret those antibody inhibition studies (30).We were not able to identify the cellular receptor for B.

burgdorferi involved in endothelial cell uptake of this organ-ism. Our results suggest that the expression of the endothe-lial cell receptor was not induced by incubation with thespirochetes because the slow kinetics did not reveal a lagtime for binding or internalization (Fig. 1 to 3). Additionally,two agents which induce inflammatory signals, TNF andphorbol myristate acetate, had no effect on binding orinternalization, indicating that inflammatory stimulus wasnot required for expression of receptor or endocytic activityby the endothelial cells. In our studies, binding and internal-ization were seen with one noninfectious strain of B. burg-dorferi, suggesting this is not the only determinant of infec-tivity. It will be interesting to compare the intracellularsurvival of infectious and noninfectious strains of B. burg-dorferi.The experiments described in this paper provide a model

by which to study the role of intracellular survival inpersistence of the spirochete during infection of the host. Bystudying this interaction in vitro, it should be possible todetermine how long spirochetes can persist within endothe-lial cells and whether viable spirochetes can be recoveredfrom endothelial cells. This model should allow characteri-zation of the spirochete and endothelial cell moleculesinvolved in infection. Further analysis of this interactionmay provide information on the type of intracellular com-partment in which the spirochetes reside and whether intactspirochetes and degraded spirochetes are associated withdifferent types of cellular vacuoles. Information from thesestudies may also provide insight into the involvement ofendothelial cell localization in the pathological aspects ofLyme disease.

ACKNOWLEDGMENTS

We thank David Low and Tom McIntyre for helpful discussions.This work was supported by Public Health Service grant GM-

40085 from the National Institutes of Health (J.J.W.) and a BRSGfrom the University of Utah (J.J.W). J.J.W is a Lucille P. MarkeyScholar, and this work was supported in part by a grant from theLucille P. Markey Charitable Trust.

REFERENCES1. Barbour, A. G. 1984. Immunochemical analysis for Lyme

disease spirochetes. Yale J. Biol. Med. 57:581-586.2. Barbour, A. G., and S. F. Hayes. 1986. Biology of Borrelia

species. Microbiol. Rev. 50:381-400.3. Barbour, A. G., S. F. Hayes, R. A. Heiland, M. E. Schrumpf,

and S. L. Tessier. 1986. A Borrelia-specific monoclonal antibodybinds to a flagellar epitope. Infect. Immun. 52:549-554.

4. Barbour, A. G., S. L. Tessier, and S. F. Hayes. 1984. Variationin a major surface protein of Lyme Disease spirochetes. Infect.Immun. 45:94-100.

5. Barbour, A. G., S. L. Tessier, and W. J. Todd. 1983. Lymedisease spirochetes and Ixodid tick spirochetes share a commonsurface antigenic determinant defined by a monoclonal anti-body. Infect. Immun. 41:795-804.

6. Bevilacqua, M P., J. S. Pober, D. L. Mendrick, R. S. Cotran, andM. A. Gimbrone. 1987. Identification of an inducible endotheli-al-leukocyte adhesion molecule. Proc. Natl. Acad. Sci. USA84:9238-9242.

7. Brett, J., H. Gerlach, P. Nawroth, S. Steinberg, G. Godman, andD. Stern. 1989. Tumor necrosis factor/cachectin increases per-meability of endothelial cell monolayers by a mechanism involv-ing regulatory G proteins. J. Exp. Med. 169:1977-1991.

8. Burgdorferi, W., A. G. Barbour, S. F. Hayes, E. Grunwaldt, andJ. P. Davis. 1982. Lyme disease-a tick-borne spirochetosis?Science 216:1317-1319.

9. Clerc, P., B. Baudry, and P. J. Sansonetti. 1988. Molecularmechanisms of entry, intracellular multiplication and killing ofhost cells by Shigellae. Curr. Top. Microbiol. Immunol. 138:3-13.

10. Comstock, L. E., and D. D. Thomas. 1989. Penetration ofendothelial cell monolayers by Borrelia burgdorferi. Infect.Immun. 57:1626-1628.

11. Dattwyler, R. J., J. J. Halperin, D. J. Volkman, and B. J. Luft.1988. Treatment of late Lyme borreliosis-randomised compar-ison of ceftriaxone and penicillin. Lancet May 28, i:1191-1194.

12. deKoning, J., J. A. A. Hoogkamp-Korstanje, M. R. van derLinde, and H. J. G. M. Crijns. 1989. Demonstration of spiro-chetes in cardiac biopsies of patients with Lyme disease. J.Infect. Dis. 160:150-153.

13. Ewanowich, C. A., A. R. Melton, A. A. Weiss, R. K. Sherburne,and M. S. Peppler. 1989. Invasion of HeLa 229 cells by virulentBordetella pertussis. Infect. Immun. 57:2698-2704.

14. Finlay, B. B., and S. Falkow. 1988. Comparison of the invasionstrategies used by Salmonella cholerae-suis, Shigella flexneri,and Yersinia enterocolitica to enter cultured animal cells: endo-some acidification is not required for bacterial invasion or

VOL. 59, 1991

on January 20, 2021 by guesthttp://iai.asm

.org/D

ownloaded from

678 MA ET AL.

intracellular replication. Biochimie 70:1089-1099.15. Finlay, B. B., and S. Falkow. 1989. Common themes in microbial

pathogenicity. Microbiol. Rev. 53:210-230.16. Geng, J.-G., M. P. Bevilacqua, K. L. Moore, T. M. McIntyre,

S. M. Prescott, J. M. Kim, G. A. Bliss, G. A. Zimmerman, andR. P. McEver. Rapid neutrophil adhesion to activated endothe-lium mediated by GMP-140. Nature (London) 343:757-760.

17. Goldings, E. A., and J. Jericho. 1986. Lyme disease. Clin.Rheum. Dis. 12:343-367.

18. Jaffe, E. A., R. L. Nachman, C. G. Becker, and C. R. Minick.1973. Culture of human endothelial cells derived from umbilicalveins. J. Clin. Invest. 52:2745-2756.

19. Johnson, R. C., G. P. Schmid, F. W. Hyde, A. G. Steigerwalt,and D. J. Brenner. 1984. Borrelia burgdorferi sp. nov.: etiologicagent of Lyme disease. Int. J. System. Bacteriol. 34:496-497.

20. McGee, Z. A., G. L. Gorby, P. B. Wyrick, R. Hodinka, andL. H. Hoffman. 1988. Parasite-directed endocytosis. Rev. In-fect. Dis. 10:S311-S316.

21. Osborn, L., C. Hession, R. Tizard, C. Vassallo, S. Luhowskyj, G.Chi-Rosso, and R. Lobb. 1989. Direct expression cloning ofvascular cell adhesion molecule 1, a cytokine-induced endothe-lial protein that binds to lymphocytes. Cell 59:1203-1211.

22. Pavia, C. S., and S. Bittker. 1988. Assessment of in vitro growthof Borrelia burgdorferi by tritiated adenine incorporation. Ann.N.Y. Acad. Sci. 539:410-413.

23. Piesman, J., T. N. Mather, R. J. Sinsky, and A. Spielman. 1987.Duration of tick attachment and Borrelia burgdorferi transmis-sion. J. Clin. Microbiol. 25:557-558.

24. Pober, J. S., M. A. Gimbone, L. A. Lapierre, D. L. Mendrick,W. Fiers, R. Rothlein, and T. A. Springer. 1986. Overlappingpatterns of activation of human endothelial cells by interleukin1, tumor necrosis factor, and immune interferon. J. Immunol.

137:1893-1896.25. Portnoy, D. A., P. S. Jacks, and D. J. Hinrichs. 1988. Role of

hemolysin for the intracellular growth of Listeria monocyto-genes. J. Exp. Med. 167:1459-1471.

26. Schwan, T. G., W. Burgdorfer, M. E. Schrumpf, and R. H.Karstens. 1988. The urinary bladder, a consistent source ofBorrelia burgdorferi in experimentally infected white-footedmice (Peromyscus leucopus). J. Clin. Microbiol. 26:893-895.

27. Sigal, L. H. 1989. Lyme disease, 1988: immunologic manifesta-tions and possible immunopathogenetic mechanisms. Semin.Arthritis Rheum. 18:151-167.

28. Stanek, G., J. Klein, R. Bittner, and D. Glogar. 1990. Isolationof Borrelia burgdorferi from the myocardium of a patient withlongstanding cardiomyopathy. N. Engl. J. Med. 322:249-252.

29. Steere, A. C. 1989. Lyme disease. N. Engl. J. Med. 321:586-596.30. Szczepanski, A., M. B. Furie, J. L. Benach, B. P. Lane, and

H. B. Fleit. 1990. Interaction between Borrelia burgdorferi andendothelium in vitro. J. Clin. Invest. 85:1637-1647.

31. Thomas, D. D., and L. E. Comstock. 1989. Interaction of Lymedisease spirochetes with cultured eucaryotic cells. Infect. Im-mun. 57:1324-1326.

32. Tilney, L. G., and D. A. Portnoy. 1989. Actin filaments and thegrowth, movement, and spread of the intracellular bacterialparasite, Listeria monocytogenes. J. Cell Biol. 109:1597-1608.

33. Voyta, J. C., D. P. Via, C. E. Butterfield, and B. R. Zetter. 1984.Identification and isolation of endothelial cells based on theirincreased uptake of acetylated-low density lipoprotein. J. CellBiol. 99:2034-2040.

34. Wyrick, P. R., J. Choong, C. H. Davis, S. T. Knight, M. 0.Royal, A. S. Maslow, and C. R. Bagnell. 1989. Entry of genitalChlamydia trachomatis into polarized human epithelial cells.Infect. Immun. 57:2378-2389.

INFECT. IMMUN.

on January 20, 2021 by guesthttp://iai.asm

.org/D

ownloaded from