Internalin B is essential for adhesion and mediates the invasion of Listeria monocytogenes into human endothelial cells

Internalin B is essential for adhesion and mediates theinvasion of Listeria monocytogenes into humanendothelial cells

Shreemanta K. Parida, 1* Eugen Domann, 1 ManfredRohde, 2 Simone Mu ller, 3 Ayub Darji, 3 Torsten Hain, 1

Ju rgen Wehland 3 and Trinad Chakraborty 1

1Institute for Medical Microbiology, Justus-LiebigUniversity, Frankfurterstrasse 107, D-35392 Giessen,Germany.Departments of 2Microbiology and 3Cell Biology andImmunology, National Centre for Biotechnology, D-38124Braunschweig, Germany.

Summary

Listeria monocytogenes causes rhombencephalitis inhumans and animals and also affects the fetus inutero , causing disseminated sepsis. In both instances,the infection occurs by the crossing of endothelial cellslining a physiological barrier, the blood–brain barrieror the transplacental barrier. In this study, the abilityof L. monocytogenes wild-type EGD to invade humanumbilical vein endothelial cells (HUVECs) was evalu-ated using wild-type bacteria and isogenic Listeriamutants. Here, we show that invasion of HUVECs byL. monocytogenes is dependent on the expression ofthe internalin B gene product. This was demonstratedin several ways. First, L. monocytogenes strains lack-ing the inl B gene did not invade HUVECs. Secondly,avid invasion was obtained when a strain deleted forinl AB was complemented with a plasmid harbouringinl B only, whereas strains expressing inl A did notenter HUVECs. Thirdly, entry of wild-type EGD couldbe blocked effectively with antibodies to InlB. Fourthly,cell binding assays and flow cytometry with HUVECsshowed binding of purified InlB, but not InlA, suggest-ing a tropism of InlB for this cell type. Finally, physicalassociation of purified native InlB with the surface ofnon-invasive mutants dramatically increased theirability to invade HUVECs. In laser-scanning confocalmicroscopy, binding of InlB was observed as focaland localized patches on the cell surface of HUVECs.Qualitative examination of the entry process by scan-ning electron microscopy revealed that both wild-type

EGD and a recombinant strain overexpressing onlyInlB enter HUVECs in a similar fashion. The entry pro-cess was polarized, involved single bacteria andoccurred over the entire surface of endothelial cells.

Introduction

Listeria monocytogenes is one of the major pathogenscausing bacterial meningitis in infants and the elderly, inimmunocompromised individuals, such as transplantationpatients, cancer patients on therapy, patients with connec-tive tissue disorders and subjects with chronic alcoholismor taking high-dose steroids. A population-based surveil-lance study of HIV-infected persons has indicated anapproximately 100-fold greater risk of listeriosis (meningi-tis and bacteraemia) compared with the general population(Jurado et al., 1993). A growing population of patientsimmunocompromised owing to advanced HIV infection,leukaemia and renal transplantation is at 500–1000 timesgreater risk of invasive listeriosis than the general population(Jensen et al., 1994). Most cases of Listeria infection are,however, sporadic, and outbreaks have been traced toingestion of contaminated food, especially unpasteurizeddairy products (Schuchat et al., 1991).

L. monocytogenes is a motile, Gram-positive bacillusand a facultative intracellular organism capable of invadinga wide variety of non-professional phagocytic cells, includ-ing enterocytes, fibroblasts, dendritic cells and hepato-cytes (reviewed in Dramsi et al., 1996; Chakraborty andWehland, 1997). The normal route of infection in listeriosisis through the epithelial barrier of the intestine – in particu-lar, by translocation through the M cells of Peyer’s patchesfollowed by basolateral invasion of enterocytes (Marco etal., 1992). After entry into the gut, L. monocytogenes dis-seminate via lymphatics and blood to the other tissues.Listeria has been shown to invade host parenchymalcells, such as fibroblasts and hepatocytes (Dramsi et al.,1995). Endothelial cells are early and active participantsin the non-specific inflammatory response during the hae-matogenous spread and are likely to be primary targetsduring Listeria infection and spread (Drevets et al., 1995).

L. monocytogenes is one of the few bacteria that cancross the transplacental barrier to infect the fetus, thusresulting in stillbirths, preterm labour or an infant bornalive with systemic L. monocytogenes infection (Gray and

Molecular Microbiology (1998) 28(1), 81–93

Q 1998 Blackwell Science Ltd

Received 23 September, 1997; revised 4 January, 1998; accepted 9January, 1998. *For correspondence. E-mail [email protected]; Tel. (641) 99 41265; Fax (641) 9941259.

m

Killinger, 1966). Neonatal infection acquired in utero is char-acterized by disseminated abscesses and granulomas andis termed ‘granulomatosis infantisepticum’. Delayed man-ifestation can occur in infants with bacteraemia with orwithout sepsis syndrome and febrile illness without a recog-nizable source. Meningitis caused by L. monocytogenesaffects infants under 2 months of age, who acquired theinfection during delivery. The clinical spectrum of centralnervous system (CNS) infection by L. monocytogenesranges from meningitis to diffuse encephalitis or intracel-lular abscesses, predominantly located in the rhombence-phalum during natural infections in both human beings andanimals (Uldry et al., 1993; Berche, 1995), suggestingcrossing of the blood–brain barrier. In humans and all ver-tebrates, both the transplacental barrier and the blood–brain barrier exist at the level of the endothelial cells, liningthese barriers.

A distinguishing feature of pathogenic Listeria is theirability to invade non-phagocytic cells. Recently, a chromo-somal genetic locus, the internalin operon, has been iden-tified, whose gene products are involved in entry intovarious cell types, i.e. internalization (Dramsi et al., 1995;Gaillard et al., 1991; Lingnau et al., 1995). The gene pro-ducts of this operon, InlA and InlB, are cell wall-associatedproteins of 88 kDa and 65 kDa respectively. InlA associ-ates with the bacterial cell wall via a conventional cellwall anchoring sequence (Lebrun et al., 1996), while thecell wall association of InlB requires a larger segment,comprising the last 232 amino acids of the C-terminus(Braun et al., 1997). A striking structural feature of bothproteins is the presence of consecutive leucine-rich repeats(LRRs), present 15 times in InlA and seven times in InlB(Lingnau et al., 1995; Domann et al., 1997). Homologysearches have revealed that InlA and InlB are membersof a superfamily of LRR-containing proteins. These LRRsare found only in pathogenic bacteria, such as the ipaHgene family in Shigella flexneri, the YopM protein fromYersinia pestis, the filamentous haemagglutinin proteinof Bordetella pertussis and the TpLRR protein of Trepo-nema pallidum (Kobe and Deisenhofer, 1994; Makhov etal., 1994; Shevchenko et al., 1997). It has been shownthat pathogenic Listeria also harbours an internalin-likeprotein gene (irpA), encoding a secreted protein of 30 kDa(Engelbrecht et al., 1996; Lingnau et al., 1996; Domannet al., 1997). Recently, four more internalin-like proteinshave been found in L. monocytogenes, although the roleof these proteins in infection is at present unclear (Dramsiet al., 1997).

It has been shown previously that InlA and InlB are dif-ferentially required for entry into various tissue and culturecell lines (reviewed in Dramsi et al., 1996; Chakrabortyand Wehland, 1997). In this study, we have examinedthe interaction of Listeria with human umbilical vein endo-thelial cells (HUVECs) to elucidate mechanisms by which

pathogenic Listeria breaches the endothelial lining of theplacenta and blood–brain barrier. More precisely, we haveexamined the genetic basis of internalization of L. mono-cytogenes into HUVECs using isogenic deletion mutantsin known internalin genes and complemented mutants.These data have been corroborated in biochemical assaysusing purified InlA and InlB. Here, we report the unequivo-cal evidence of InlB-mediated entry into endothelial cells.

Results

Infection assays with deletion mutants reveal a crucialrole of InlB

We examined the ability of wild-type L. monocytogenesEGD and its isogenic mutants, L. monocytogenes DinlA2,DinlB2 and DinlAB2, to invade and grow in endothelialcells. HUVECs were infected with a multiplicity of infection(MOI) of 100:1, and growth was monitored at 2, 4, 6 and8 h by incubation with 50 mg ml¹1 gentamicin in the med-ium after 2 h of infection (co-incubation) with the bacteria.The results of this experiment are shown in Fig. 1. Therecovery of intracellular L. monocytogenes demonstratedactive infection of HUVECs by Listeria (Fig. 1A). Therewas a steady increase in bacterial cfu within the cells upto 6 h after infection, followed by a decline in the total num-bers of bacteria at 8 h. The invasion level of L. monocyto-genes EGD for HUVECs was quite low – only 0.31% of theinitial inoculum entered the host cells. Mutants deletedeither for internalin B (DinlB2) or for both internalins(DinlAB2) did not infect HUVECs, whereas the mutantDinlA2 strain did infect the cells, albeit at a lower levelthan in the wild-type strain. For comparison, the relativepercentage of invasion for each strain was calculatedwith wild-type EGD infection as 100% for each time point(Fig. 1B). At 2 h, invasion of DinlA2 mutant reached 76%of the wild-type strain, whereas the DinlB2 mutant wasabout 13-fold less at 8%. The DinlAB2 mutant was about90-fold less invasive than the parental strain and aboutsevenfold less invasive than the DinlB2 mutant at 1.1%.This trend was reflected at the other time points usedwithin the experiment and suggests that InlB is the primaryinvasin for this cell type.

Complementation of DinlAB2 strain with inlB, but notinlA, promotes invasion

To assess the individual abilities of InlA and InlB in promot-ing entry into HUVECs, we complemented the DinlAB2mutant with plasmids harbouring either the inlA gene,the inlB gene or the entire inlAB operon. In our initialstudies, we used two types of InlB-expressing plasmids.In the pERL3::PBinlB construct, the expression of InlB isdependent on its low-level prfA-independent promoter,

Q 1998 Blackwell Science Ltd, Molecular Microbiology, 28, 81–93

82 S. K. Parida et al.

while in the pERL3::PABinlB construct, the production ofInlB is the result of transcription originating from theprfA-regulated promoters P1, P2, P3 and P4 originally pre-sent upstream of the inlAB operon (Lingnau et al., 1995).Figure 2 describes the result after 2 h of infection. Entryof the DinlAB2 mutant was about 1% of that observed forthe wild-type EGD strain. Complementation of this mutantwith the inlA gene did not increase entry significantly(1.6%). Complementation with PBinlB alone also did notpromote invasion (< 1%). However, the expression of InlBfrom pERL3::PABinlB increased invasion to 600% comparedwith the wild-type strain. In a similar manner, complemen-tation with the pERL3::inlAB operon enhanced invasion to340%. Similar results were also obtained at other timepoints within the experiments (data not shown). Theseresults further confirmed that InlB is the primary invasinfor HUVECs.

Next, we used immunoblotting with specific monoclonalantibody (mAb) raised against either InlA or InlB to verifythe level of expression of both internalins in the strainsused (Fig. 3). From this analysis, we found that strainsharbouring additional copies of the respective internalingene indeed showed increased production of these proteins.In particular, the amount of InlB produced by both recombi-nant Listeria harbouring either the plasmid pERL3::PABinlBor the plasmid pERL3::inlAB was significantly higher thanthat produced by a strain harbouring pERL3::PBinlB andthe wild-type strain.

These results were also assessed qualitatively usingimmunofluorescence studies. Coverslips fixed during theinfection assay were stained for both F-actin and bacteria.Bacteria were seen in infected HUVECs after 2 h of infec-tion, and some of the bacteria had actin tails. Five- to 10-fold more bacteria were observed in the case of infection

Q 1998 Blackwell Science Ltd, Molecular Microbiology, 28, 81–93

Fig. 1. A. Intracellular growth of Listeria monocytogenes EGD serotype 1/2a and isogenic mutants in endothelial cells. HUVECs were infectedwith L. monocytogenes wild-type EGD, EGDDinlA2, EGDDinlB2, and EGDDinlAB2 strains for 2 h, and growth was monitored at various timepoints after incubation with 50 mg ml¹1 gentamicin in the culture medium. At the indicated times, cfu of Listeriae were determined after lysis ofthe infected cells and plating on BHI-containing agar plates. Values represent mean cfu per well from four different wells, and bars representsSEM. This figure represents one of the three independent experiments with similar results.B. Relative percentage of invasion of L. monocytogenes wild-type EGD, EGDDinlA2, EGDDinlB2 and EGDDinlAB2 strains. Values along thevertical axis are given relative to the invasion of wild-type strain EGD, which is taken as 100% for each time point after calculating the rate ofentry on the basis of cfu of inoculum. Results are relative percentages of the same experiment shown in A.

Fig. 2. Invasion of endothelial cells by L.monocytogenes requires InlB. Relative ratesof entry of L. monocytogenes wild-type EGD,EGDDinlAB2 complemented with pERL-3plasmid harbouring either inlA, inlB orPABinlB and EGDDinlAB2 strains respectively.HUVECs were infected for 2 h, followed byincubation for 2 h with gentamicin. Results at2 h after incubation with gentamicin areshown. Relative percentages were calculatedfrom the mean cfu per well from four differentwells and the inoculum. The y-axis is depictedin logarithmic scale for better presentation.This figure represents one of the threeindependent experiments with similar results.

L. monocytogenes invasion is dependent on internalin B 83

with overexpressed InlB strain (the recombinant mutantpERL3::PABinlB) than with the wild-type EGD (data notshown).

Antibodies against InlB inhibit infection of HUVECs byListeria

To corroborate our findings above, we examined the abilityof affinity-purified polyclonal antibodies, specific for eitherInlA or InlB, to inhibit the uptake of Listeria into HUVECs(Table 1). Polyclonal antiserum raised against InlA inhi-bited Listeria uptake by about 68%, whereas the use ofthe anti-InlB-specific antibodies inhibited infection by 93%.Combined use of both the antibodies reduced invasionby 95% of that observed with the wild-type EGD. Flowcytometric analysis performed with the InlA antibodiesused in this study showed that they were cross-reactivewith InlB and would explain their inhibitory action onInlB-mediated entry (data not shown). Control antibodiesdid not inhibit infection (data not shown). These findingsprovided further evidence for a dominant role of InlB ininfection of HUVECs by L. monocytogenes.

Purified InlB, but not InlA, binds to HUVECs

The ability of the purified internalins to bind HUVECs wasanalysed by coating increasing concentrations of nativeInlB protein to a 96-well microtitre plate followed by incuba-tion with HUVECs. The presence of bound cells was deter-mined using a colorimetric assay for the quantification ofcell viability based on the hexosaminidase colorimetrytechnique. The results presented in Fig. 4A show clearlythat InlB protein binds to HUVECs in a concentration-dependent manner. Binding was inhibited in the presenceof EDTA, and cells did not bind to uncoated wells. HUVECsdid not bind to the control wells coated with BSA alone.

This binding was subsequently validated by the more

Q 1998 Blackwell Science Ltd, Molecular Microbiology, 28, 81–93

Fig. 3. Expression level of InlA and InlB in the strains used.Immunoblot of polypeptides in the whole-cell extracts of L.monocytogenes wild-type EGD (lane 1), DinlA2 (lane 2), DinlB2(lane 3), DinlAB2 (lane 4), DinlAB2 strains harbouring eitherplasmid pERL3::inlA (lane 5), pERL3::PBinlB (lane 6), pERL3::inlAB(lane 7) and pERL3::PABinlB (lane 8). InlA at the top and InlB atthe bottom.

Table 1. Inhibition of invasion by affinity-purified anti-internalinantibodies.

L. monocytogenes EGDPercentageinhibition

þPBS ¹

þ a-InlA 68þ a-InlB 93þ a-InlA a-InlB 95

L. monocytogenes wild-type EGD were incubated with affinity-puri-fied anti-InlA, anti-InlB or with both antibodies for 15 min beforebeing added to HUVEC monolayers. After incubation of HUVECsand bacteria for 2 h in presence of the antibodies, gentamicin wasadded for a further 2 h. The table shows the levels of relative percent-age of inhibition of L. monocytogenes EGD (infection arbitrarily fixedat 100%) in the presence of anti-InlA, anti-InlB or a combination ofboth anti-InlA and anti-InlB affinity-purified polyclonal antibodies ata dilution of 1:100 (final concentration at 5 mg ml¹1).

Fig. 4. A. Adhesion of HUVECs to surface coated with InlB.Microtitre plates (96-well) were coated with purified InlB, and asuspension of HUVECs was added either in the absence or in thepresence of 10 mM EDTA (indicated). After incubation for 1 h, cellswere washed, and the number of bound cells was determinedusing the lysosomal hexosaminidase assay. Values representmean 6 SEM of quadruplicate measurements of optical density at450 nm of a single representative experiment out of three similarindependent experiments.B. Binding of purified InlB to endothelial cell membrane analysedby flow cytometry. Different concentrations of purified InlB (A–C) orpurified InlA (D–F); 0.1, 1.0 and 2.5 mg per well were incubatedwith HUVECs. Specific antibodies and corresponding fluoresceinisothiocyanate-labelled secondary antibodies were used to detectthe bound internalin. Thin lines represent the backgroundfluorescence in the absence of internalin protein. Thick linesrepresent the binding in the presence of the internalin protein.

84 S. K. Parida et al.

sensitive technique of flow cytometry. The cells were firstincubated with purified InlB (0.1 mg, 1.0 mg and 2.5 mg),followed by anti-InlB-specific polyclonal antibodies andsubsequently stained with fluorescein-labelled secondaryantibodies. The intensity of the positive staining wasobserved only with cells incubated with InlB, whereas nostaining was observed when cells were either incubatedwith ovalbumin in the same manner or incubated withInlB without the first antibodies (data not shown). Binding,detected as a shift in the fluorescence intensity, wasobserved only when InlB was present and was concen-tration dependent (Fig. 4B, A–C). The same resultswere obtained when we used monoclonal antibodies toInlB instead of affinity-purified polyclonal antibodies. In asubsequent experiment, the binding capacities of InlAand InlB for HUVECs were compared. Unlike InlB, InlAdid not bind to HUVECs at concentrations of 0.1, 1.0 and2.5 mg (Fig. 4B, D–F). InlA was detected using mouseanti-InlA monoclonal antibodies, and b-galactosidase wasused as a negative control. To assay the co-operativeeffects between InlA and InlB, cells were incubated with

1 mg of each internalin and stained with either anti-InlBmonoclonal antibodies or anti-InlA monoclonal antibodies.No difference in staining was observed compared with theincubation with InlB or InlA alone. When HUVECs weretreated with 1 mg of each internalin and later stained withmonoclonal antibodies to InlA followed by affinity-purifiedpolyclonal antibodies to InlB, we did not observe anychange in the staining pattern compared with the signalobserved with InlB alone in parallel. This indicates thatthe presence of InlA does not influence InlB binding toHUVECs (data not shown).

Binding of purified internalin B to HUVECs in the flowcytometry experiments was also verified by indirectimmunofluorescence. The fluorescent images represent-ing optical sections scanned at 2 mm were collected simul-taneously using a laser scanning confocal microscope andare shown in Fig. 5A–E. Binding of InlB to HUVECs wasclearly evident. However, it appeared to be unequally dis-tributed and was observed as focal and localized patcheson the cell surface. No binding was detectable with puri-fied InlA in these studies (data not shown).

Q 1998 Blackwell Science Ltd, Molecular Microbiology, 28, 81–93

Fig. 5. Laser-scanning confocal micrographs of HUVECs in suspension incubated with InlB and labelled with anti-InlB antibodies.A–E. A z-axis section series from top to bottom of a representative cell; section thickness is 2 mm. Note the focal patchy fluorescencedistribution only on the cell surface.F. Single view projection of 12 different sections. Bar represents 10 mm.

L. monocytogenes invasion is dependent on internalin B 85

Addition of native InlB enhances invasion of wild-typeas well as non-invasive strains

Surface association of purified recombinant InlB hasbeen shown previously to induce invasion of HeLa cellsby normally non-invasive strains (Braun et al., 1997). Weexamined the ability of purified native InlB to induceuptake of the wild type as well as the non-invasivemutants, EGDDinlB2 and EGDDinlAB2, in HUVECs. Incu-bation of native protein InlB (5 mg ml¹1) with wild-type EGDenhances internalization by about 24-fold, i.e. 10 6 0.5%versus 0.4 6 0.04% for EGD alone. In the non-invasivemutant strains, DinlB2 and DinlAB2, the addition of extra-cellular InlB had a dramatic effect on internalization. Thus,the addition of InlB to DinlB2 increased its invasive abilityby 28-fold (11.59 6 0.39%) over the wild-type EGD aloneand about 300-fold over the mutant strain alone (0.04 6

0.01%). The increase was 23- and 300-fold, respectively,for DinlAB2 with added InlB (9.40 6 0.49% and 0.03 6

0.005 respectively). Addition of purified InlA did not affectthe relative abilities of the wild-type and mutant strains ininvasion (data not shown).

Electron microscopic studies

Scanning electron microscopic observations of early eventsof infection allowed us to visualize the different steps in theinvasion pathway. Entry of the strain overexpressing InlBwas very similar to that observed with wild-type bacteriadespite its increased ability to invade HUVECs (Fig. 6A,C and I versus Fig. 6B and D–H). The binding and inva-sion of Listeria was not localized and could take place atvarious sites on the surface of HUVECs (Fig. 6A). Oncounting 400 bacteria that attached to HUVECs, 25%were present at the outermost cell edge (Fig. 6A, 1), 40%were close to the cell periphery (Fig. 6A, 2), and about35% were present on the other areas on the surface ofthe cells (Fig. 6A, 3). Bacteria always attached to the sur-face of the cell with singular pole and were subsequentlyengulfed within tightly apposed membranous structures.As shown in Fig. 6B, there was induction of localizedmembrane processes, demonstrated as the formation ofa ‘flap’ in the vicinity of the bacteria.

Discussion

The major conclusion derived from the studies presentedherein is that the entry of L. monocytogenes into HUVECsis dependent on InlB. However, current data regarding theability of L. monocytogenes to invade HUVECs is contro-versial. In a previous study on the entry of L. monocyto-genes into HUVECs, it was suggested that InlA is requiredfor the invasion of HUVECs (Drevets et al., 1995). In thatstudy, a transposon-induced mutant, in which the trans-poson had inserted into the promoter region of the inlAB

operon, was used. As a result of the insertion, the expres-sion of both InlA and InlB was greatly reduced. A recentreport has suggested that the invasion of HUVECs by L.monocytogenes is not dependent on either InlA or InlB(Greiffenberg et al., 1997). A higher concentration (20%)of pooled human serum used in that study may have com-promised the results obtained. This concentration had beenshown previously to increase the binding of a transposon-induced internalin mutant to HUVECs by 40-fold (Drevetset al., 1995). Also, it had not been established whetherantibodies to L. monocytogenes were present in the pooledhuman serum used, which could have led to internalin-independent association of bacteria with HUVECs. Thus,for this study, it was imperative to use independentapproaches to verify that InlB is indeed required for adhe-sion and entry into HUVECs.

Invasion studies using Listeria mutants lacking eitherthe inlA or inlB gene demonstrated that InlB, but not InlA,is essential for entry into the endothelial cells. A strainlacking both the inlA and inlB genes was unable to enterHUVECs. Using this strain, the role of InlB in invasioncould be clearly discerned. Hence, the DinlAB2 strain com-plemented with a plasmid harbouring inlB entered HUVECsmanifold over the wild-type strain, while no entry wasobserved with the same strain complemented with aninlA-carrying plasmid. As entry of Listeria into HUVECsis the sum of both bacterial adherence and subsequentinternalization, we used purified InlA and InlB to assesstheir functions as primary adhesins in this process. Thecell binding assay on the microtitre plate coated with puri-fied InlB and flow cytometric analysis demonstrated unam-biguously that InlB is an adhesin, mediating interaction ofthe bacteria with the HUVEC surface. Finally, physicalassociation of purified native InlB with the surface of twonon-invasive strains, DinlB2 and DinlAB2, increased theirability to invade HUVECs by about 28- and 23-fold, res-pectively, more than the wild-type strain. Although thedistribution of L. monocytogenes is centripetal in Caco-2cells, with invasion taking place from the cell periphery(Mengaud et al., 1996a), in the case of HUVECs, it wasnot localized only to the cell periphery but also occurredthroughout the surface of the infected cell.

A hallmark of L. monocytogenes infection is the ability ofthe bacterium to cross different physiological barriers inthe body, such as the epithelial barrier in the intestine, thetransplacental barrier and the blood–brain barrier. Thismodus operandi of pathogenesis is a Listeria-inducedactive phenomenon with exploitation of the host cell forits successful intracellular lifestyle. Only pathogenic Lis-teria possess the ability to induce their uptake by eukaryo-tic cells, engaging a host cell receptor with concomitantstimulation of host cell signal transduction mechanismsand rearrangements of the cortical actin cytoskeleton (Ire-ton et al., 1996; Mengaud et al., 1996a; Sibelius et al., 1996).

Q 1998 Blackwell Science Ltd, Molecular Microbiology, 28, 81–93

86 S. K. Parida et al.

Following uptake, L. monocytogenes escape from thehost vacuole to replicate intracytoplasmically and subvertthe host cell’s microfilament system to facilitate intracellu-lar movement and cell-to-cell spreading (Tilney and Port-noy, 1989).

In this study, we have used isogenic in-frame deletionmutants to assess the ability of L. monocytogenes toenter HUVECs. As has been observed previously, L. mono-cytogenes invaded HUVECs and was capable of dividingintracellularly (Ebel et al., 1994; Drevets et al., 1995). Inva-sion levels were low (0.31%). Nevertheless, bacteria, once

internalized, were capable of escaping from the phago-lysosome and intracellular motility as evidenced by thepresence of F-actin tails induced by internalized bacteria.The low level of infection observed could be derived fromthe fact that HUVECs are primary cells. In a previousreport, it has been demonstrated that cell immortalizationenhances L. monocytogenes invasion, probably as a resultof increased expression of a cell receptor depending on thestate of proliferation or differentiation of the cells (Velge etal., 1994). The intracellular growth of Listeria was observedfor up to 6 h after infection and was followed by a decline in

Q 1998 Blackwell Science Ltd, Molecular Microbiology, 28, 81–93

Fig. 6. Scanning electron micrographs of the entry process of L. monocytogenes in HUVECs.A. An overview of the attachment sites on HUVECs with wild-type EGD. (1) At the outermost cell edge; (2) at the cell periphery; (3) on theinner cell surface. Bar represents 2 mm.B. Induction of membranous flap (arrow pointed) in HUVEC membrane 5 min after attachment with the strain overexpressing InlB. Barrepresents 1 mm.C. The initial event of engulfment of the wild-type bacteria by HUVECs; arrow shows the ‘flap’. Bar represents 1 mm.D–G. Serial events of entry of Listeria overexpressing InlB.H. Completely intracellular overexpressing InlB Listeria in HUVECs. Bar represents 1 mm.I. Intracellular wild-type EGD in HUVEC. Bar represents 1 mm.

L. monocytogenes invasion is dependent on internalin B 87

the total numbers of bacteria. This decrease in bacterialnumbers may be explained by the cytotoxic effect of theinfection on the HUVEC monolayer, as bacteria releasedinto the medium do not grow and are killed by the bacter-icidal action of gentamicin.

During the course of the experiments, we observed thatcentrifuging the bacteria onto the HUVEC monolayer atthe beginning of co-incubation increased invasion bythree- to fivefold (data not shown). Cells that were centri-fuged and infected after centrifugation did not show anyaltered level of infection, demonstrating that the centrifu-gation does not damage or influence any other interactionwith bacteria except for enhancing contact.

All strains overexpressing InlB were proficient for entryinto HUVECs. The level of entry of L. monocytogeneswas dependent on the amount of InlB expressed. Thus,complementation of the DinlAB2 mutant with PBinlB didnot increase its invasive ability. Complementation withthe PABinlB plasmid increased invasion dramatically, sur-passing even that observed with the wild-type strain. Thiswas confirmed by immunoblotting whole-cell lysates ofthese strains with InlA- and InlB- specific monoclonal anti-bodies (Fig. 3). The increased levels of invasion of com-plemented mutant strains harbouring either the PABinlB(lane 8) or the PinlAB (lane 7) with both internalin genesover the wild-type strain is likely to be the consequenceof the high copy number of these plasmids in the respec-tive complemented strains.

Affinity-purified polyclonal antibodies to InlB reducedinfection of the wild-type strain by 93%, emphasizing therole of InlB in the infection of HUVECs. In our studies, poly-clonal antibodies against InlA reduced the entry of thebacteria by 68%, and a combination of both antibodiesreduced invasion by 95%. As the LRR region of both mol-ecules shares significant homology and has recently beenimplicated in mediating specific interaction between InlAand Caco-2 cells (Mengaud et al., 1996b), it is conceivablethat cross-reactivity of the antibodies in the LRR regioncould also affect invasion by either molecule. Flow cyto-metric analysis performed with the InlA antibodies usedin this study showed that they were cross-reactive withInlB, and this would explain their inhibitory action on InlB-mediated entry (data not shown).

To infect endothelial cells directly, bacteria must firstadhere to the surface of these cells before invading themthrough a bacteria-induced phagocytic process. To dis-sociate the process of adhesion from that of invasion, wehave used purified InlA and InlB to examine their bindingto the surface of the endothelial cell. Thus, purified InlBwas shown to bind to HUVECs by both a cell bindingassay and the more sensitive flow cytometric assay. Bind-ing of InlB to HUVECs was inhibited by the use of EDTA,suggesting a possible role of divalent ions in this inter-action. This inhibitory effect was also observed previously

in the binding of InlA to Caco-2 cells (Mengaud et al.,1996a). We have observed this inhibitory effect of EDTAqualitatively by immunofluorescence study in the adhesionof the wild-type EGD to HUVECs. The binding of InlB toHUVECs was specific and was not observed with InlA orunrelated proteins, such as ovalbumin and b-galactosi-dase. Thus, InlB is the primary adhesin for the attachmentof L. monocytogenes to HUVECs. This could be corrobo-rated by indirect immunofluorescence using laser-scan-ning confocal microscopy. Binding was already observedat 48C and occurred in the presence of sodium azide, indi-cating that this binding is not energy dependent. Patchesof InlB were found on the surface of HUVECs, suggestingthat either clustering occurred prior to binding or that thereceptor engaged has a localized distribution.

The results presented in this study lend further supportto the notion that the internalins of L. monocytogenes areused differentially for entry into cell types. It is therefore ofinterest to note that, while InlA is required for entry into theepithelial-like cell line Caco-2, entry into primary endothel-ial cells, such as HUVECs, requires the product of the inl Bgene. Entry into the hepatocyte-like cell line HepG2 hasalso been shown to be mediated by InlB. However, therole of internalins in the in vivo invasion of hepatocytesremains controversial (Gaillard et al., 1996; Gregory etal., 1996).

Scanning electron microscopic studies revealed detailsof the entry process of L. monocytogenes into HUVECs.We found that the DinlAB2 mutant complemented withthe PABinlB plasmid entered HUVECs in a similar fashionto that observed with the wild-type strain. As expected, therecombinant strain showed a higher level of invasion.Unlike previous observations with InlA-mediated infectionof Caco-2 cells, in which infection progressed centripetallyfrom the bacteria entering via the cell periphery (TemmGrove et al., 1994; Mengaud et al., 1996a), the invasionof HUVECs was more generalized and occurred over theentire surface of the cell. Nevertheless, the mode of entryof bacteria into both cell types appeared to be structurallysimilar. Thus, L. monocytogenes attached individually oras dividing bacteria with one pole engaging the cell sur-face. Bacteria appeared to induce the formation of amini-ruffle or flap, which resulted finally in the engulfmentof the bacteria within tightly apposed cellular membranes.For Haemophilus influenzae type b, endothelial invasion isalso not preceded by extensive plasma membrane activityin the host, and uptake involves processes arising in theclose vicinity of bacteria similar to that seen in Fig. 6(C–F) (Virji, 1996).

As has been observed recently for Vero cells (Ireton etal., 1996), we have found that in the invasion of both thewild-type strain and its InlB-overproducing counterpart,entry can be inhibited significantly by the phosphoinosi-tide-3-kinase inhibitor LY 294002 or Wortmannin (S. K.

Q 1998 Blackwell Science Ltd, Molecular Microbiology, 28, 81–93

88 S. K. Parida et al.

Parida, unpublished data). Thus, cellular processes regu-lated by this kinase appear to be induced upon interactionof these bacteria with a cellular receptor(s). These resultsare in accordance with previous data postulating a zippertype of entry of L. monocytogenes (Swanson and Baer,1995; Mengaud et al., 1996a; Finlay and Cossart, 1997).

The data presented here address one particular aspectof infection, whose end result is the release of pathogenicListeria into normal sterile sites, such as fetal tissue or thebrain. Although we have shown direct invasion of endo-thelial cells by L. monocytogenes, it is presently not clearif bacteria are actually transcytosed or whether lateralinfection occurs cell to cell, creating a depot for the subse-quent propagation of infection. We observed previouslythat two principal virulence factors of L. monocytogenes,i.e. listeriolysin and the inositol-specific phospholipase C,act in concert potently to induce mediators, such as plate-let-activating factor (PAF) and prostaglandin (PGI2), inHUVECs (Sibelius et al., 1996). More recently, we havefound that the adhesion molecules, P-selectin, E-selectin,ICAM-1 and VCAM-1, are specifically upregulated inHUVECs after incubation with the wild-type EGD strain(Krull et al., 1997). The induction of P-selectin is depen-dent on listeriolysin production, whereas the induction pro-cess for the other molecules is not yet known. Thus, thetropism of L. monocytogenes for endothelial cells andtheir ability to induce signalling events leading to the gene-ration of inflammatory and vasoactive mediators, such asthose encountered in sepsis, emphasizes the relevance ofendothelial cells in the pathophysiology of listerial infections.

Experimental procedures

HUVEC isolation

HUVECs were harvested from fresh human newborn umbili-cal cords with 0.025% collagenase (Worthington) as describedby Jaffe et al. (1973). Cells were maintained at 378C with 5%CO2 in MCDB131 medium (Gibco BRL, Life Technologies)supplemented with 10% fetal calf serum (FCS), epidermalgrowth factor (EGF) (5 ng ml¹1), 2 mM glutamine (Gibco) andantibiotics – penicillin and streptomycin (Gibco) on gelatinematrix in tissue culture flasks (Nunc). Flasks or tissue cultureplates were coated with 0.5% gelatine for 1 h at room tem-perature (RT) before plating the cells. All the experimentswere performed with HUVECs after the first passage. Flowcytometry experiments were carried out with first and secondpassages with similar results.

Characterization of HUVECs

Cells isolated and grown to confluency were characterized byimmunostaining with the antibodies to Von Willebrand factor(VWF) – (factor VIII), CD31, CD34, CD44 (all three antibodiesfrom Dianova), CD62P (Immunotech) and vimentin (Dianova).Cells were also analysed by flow cytometry for endothelial cellcharacterization. Using flow cytometry, strongly positive

staining was found in the case of VWF, vimentin and CD31.CD34 and CD44 were positive, but the staining was not aspronounced as that observed with VWF, vimentin and CD31.The cell population was found to be pure, as more than 99%of cells were found to be positive for VWF. These stainingcharacteristics did not change during the first five passages.

Bacterial strains, media and reagents

Listeria monocytogenes serotype 1/2a strain EGD (Kaufmann,1984) and the isogenic mutants (Table 2) were grown rou-tinely in brain–heart infusion (BHI) medium (Difco Labora-tories) overnight at 378C with continuous shaking at 200r.p.m. and, in the case of complemented strains harbouringa pERL-3 derivative, supplemented with 5 mg ml¹1 erythro-mycin (Gibco). Then, the exponential culture was grown in a1:50 dilution of the overnight culture suspension for 3 h.After 3 h of exponential culture, the optical density measuredat 600 nm was 0.8–1.0. An optical density of 1.0 correspondsto 1 ×109 bacteria ml¹1. Bacteria were recovered by centrifu-gation (5500 × g for 2 min), washed once with PBS, then oncewith MCDB 131 medium without any additive and, finally, sus-pended in the same medium. Each time, 50 ml of the bacterialsuspension was taken and plated after appropriate dilution of10¹5 in PBS in duplicates on BHI agar plates using the Spiralplatter (Autoplate model 3000; Spiral Biotech) to enumeratethe exact number of viable bacteria added to the cells. All che-mical and immunological reagents were of the highest purityavailable and were purchased from Sigma unless indicatedotherwise.

Construction of chromosomal deletion mutants

The isogenic mutants were created with a chromosomalin-frame deletion in either inlA or inlB (Lingnau et al., 1995).In order to generate the isogenic mutant strain DinlAB2, theflanking regions of the inlAB operon were amplified separatelyby polymerase chain reaction (PCR) using chromosomal DNAfrom L. monocytogenes as a template. The upstream anddownstream regions of the inlAB operon were amplified withthe oligonucleotide pairs 4616–4617 and 5046–4865, result-ing in products of 1.25 kb and 0.73 kb respectively. Both PCRproducts were digested with the restriction endonucleaseBamHI, ligated and amplified again by PCR with the flanking

Q 1998 Blackwell Science Ltd, Molecular Microbiology, 28, 81–93

Table 2. Listeria strains and plasmids used in this study.

Reference

StrainsListeria monocytogenes EGD Kaufmann (1984)Listeria monocytogenes DinlA2 Lingnau et al. (1995)Listeria monocytogenes DinlB2 Lingnau et al. (1995)Listeria monocytogenes DinlAB2 This study

PlasmidpERL3 Leimeister-Wachter et al.

(1990)pERL3::inlA This studypERL3::PBinlB This studypERL3::PABinlB This studypERL3::PinlAB This study

L. monocytogenes invasion is dependent on internalin B 89

oligonucleotides 4616 and 4865 to obtain a DNA fragment of1.95 kb. This PCR product was deficient for both the inlA andthe inlB genes including the intergenic region between inlAand inlB. Finally, InlA lacked the amino acids 73–800 andInlB lacked the amino acids 1–629. The 1.95 kb large PCRproduct was digested with the restriction endonucleases KpnIand XbaI and was cloned directly into the KpnI and XbaI restric-tion sites of the multiple cloning site of the suicide vectorpAUL-A (Chakraborty et al., 1992). After transformation ofthe wild-type strain EGD using a protoplast transformationprocedure (Wuenscher et al., 1991), the inlAB2 deletion mutantwas generated as described previously (Lingnau et al., 1995).To confirm the deletion of the entire internalin operon, themutant strain DinlAB2 was examined by immunoblotting withthe monoclonal antibodies against InlA and InlB and, addition-ally, by sequencing the deletion (Lingnau et al., 1995).

Cloning of the inlA, inlB, PAB inlB and inlAB genes intothe expression plasmid pERL3

To express the inlA and inlB genes and the entire inlABoperon in several Listeria strains, the corresponding DNA frag-ments were cloned separately into the Gram-positive expres-sion plasmid pERL3 (Leimeister-Wachter et al., 1992). TheinlA gene was amplified with the oligonucleotides 1317–4141, the inlB gene with the oligonucleotides 1320–4142 andthe entire inlAB operon with the oligonucleotides 1317–4142by PCR using chromosomal DNA from L. monocytogenesEGD as a template, resulting in PCR products with sizes of3.42 kb, 2.77 kb and 5.89 kb respectively. The PCR productswere cloned directly into the cloning plasmid pCRII (Invitro-gen). The recombinant pCRII plasmids were digested withXhoI and BamHI, and the resulting restriction fragments har-bouring the different internalin genes were cloned into theSal I and Bgl II sites of pERL3 respectively. All the oligonu-cleotides are described in Table 3.

As the inlB gene was cloned into the pERL3 vector withouta prfA-dependent promoter (Lingnau et al., 1995), we decidedto generate a new recombinant expressing the inlB gene withthe prfA-dependent promoter located upstream of the inter-nalin operon to establish a prfA-dependent expression. Forthis purpose, a PCR product of 3.87 bp was amplified usingthe oligonucleotides 1317–4142 by PCR with chromosomalDNA of the DinlA2 mutant strain (Lingnau et al., 1995). Thecorresponding PCR product was cloned directly into thepCRII plasmid as described above. After digestion with XhoIand Nsi I, the PABinlB-specific DNA fragment was cloned intothe compatible Sal I and Pst I restriction sites of pERL3. Therecombinant pERL3 plasmids harbouring inlA, inlB, PABinlB

and the inlAB operon were transformed using protoplasts asdescribed previously (Wuenscher et al., 1991).

Protein preparation, SDS–PAGE and immunoblotting

Cell wall extracts were prepared as described previously(Domann et al., 1997). Protein samples (10 ml) were boiledfor 3 min, loaded onto a 10% polyacrylamide gel and electro-phoresed in the presence of SDS. For immunoblot reactions,proteins were transferred to a nitrocellulose filter and reactedwith either InlA- or InlB-specific antibodies (1:1000). Immuno-blots were developed as described previously (Lingnau et al.,1995).

Infection assay

Cells in the first passage were seeded at a concentration of1 ×105 in 1 ml of MCDB 131 medium per well in 24-well tissueculture plates coated with 0.5% gelatine 1 h before seeding.Cells were allowed to grow for about 48 h. The medium waschanged, and fresh medium with all supplements except anti-biotics was added. The cells were grown to confluency over-night in the antibiotic-free medium. Infection was performedwith the Listeria wild-type EGD, isogenic mutants or comple-mented mutants at a cell–bacteria ratio of 1:100. After 2 h ofco-incubation, the medium was aspirated, and the cells werewashed three times with HEPES–Hanks’ buffer (with Ca2þ

and Mg2þ), replaced with MCDB 131 medium supplementedwith 50 mg ml¹1 gentamicin to kill the remaining extracellularListeria and incubated at 378C for different lengths of time(2, 4, 6 and 8 h). Thereafter, the supernatant fluids were dis-carded, and the cells were washed three times with HEPES–Hanks’ buffer and lysed at different time points with 0.2% TritonX-100 in distilled water. Cells were incubated with 0.2% TritonX-100 in distilled water for 20 min at RT, followed by thoroughmixing to lyse the cells completely. The lysate was diluted 10times in PBS and plated on BHI agar plates using Spiral plat-ter for the enumeration of intracellular Listeria. After 24 h ofincubation, the colonies were counted and cfu ml¹1 determined.

From the plates of the bacterial inoculum, the concentrationof the bacteria in the inoculum was determined. The meanpercentage of invasion for each bacterial strain for each timepoint was determined from the intracellular bacteria recoveredin cfu ml¹1 after lysis of the HUVECs and the numbers of bac-teria in the inoculum. Relative percentage of invasion wasdetermined for each mutant, taking the invasion of the wild-type EGD as 100%.

Before lysing the cells, the coverslips used in the infection

Q 1998 Blackwell Science Ltd, Molecular Microbiology, 28, 81–93

Table 3. Oligonucleotides used in theconstruction of the inlAB2 mutant and therecombinant plasmids in this study.

No. 58–38 sequence

1317 GGAATGACGAGCTCATACAAACAATATGG1320 GGCGATAGCGATAATGAGCTCTACC4141 CACGGTGATAGTCTCCGCTTGTA4142 TATAGGCGATTCCATACA4616 ACTTCATCTGCTGCAGGCTTAAAAGCA4617 AACTTGGTCTGGATCCGTTTGCGAGAC4865 GTCATTAAATCTAGACGATTCCATACA5046 GCAGCTAATTTAAGGGATCCGAAATAACTGAAAAAGACCT

90 S. K. Parida et al.

assay were fixed with 3% paraformaldehyde for 5 min. Fluor-escence staining of actin and bacteria was performed asdescribed previously (Niebuhr et al., 1993).

Infection assay with bacteria associated with native InlB

Listeria monocyotogenes wild-type EGD, DinlB2 and DinlAB2mutant strains from exponentially growing cultures werewashed in PBS twice, suspended in plain MCDB131 mediumwith 5 mg ml¹1 native InlB and incubated at 378C for 30 min.Dose-dependent invasion was detected, which peaked at5 mg ml¹1 purified native InlB in a pilot experiment using differ-ent concentrations of InlB (data not shown). This concentra-tion was used in the assays described here. Infection wascarried out with InlB-associated bacteria after incubatingwith purified InlB with MOI of 100:1. The plate was centrifugedat 1000 ×g for 1 min to enhance the contact of bacteria withHUVECs. The invasion assay was performed as describedabove. Infection was also carried out with the strains alonein parallel for comparison.

Pretreatment of bacteria and HUVECs with blockingantibodies

HUVECs were grown to confluency in the 24-well plates, andthe cells were grown in antibiotic-free medium overnight. Beforeinfection, the cells were washed with fresh MCDB 131 medium.L. monocytogenes EGD were cultured overnight and culti-vated in exponential culture for 3 h in BHI medium. Bacteriawere preincubated for 15 min with antibodies before infectionto ensure neutralization of the respective internalin protein.The antibodies were also present in the medium during theinteraction of bacteria with the endothelial cells. Four tubeswith 100 ml of bacterial suspension each from log culture weretaken and centrifuged at 5500 ×g for 2 min followed by twowashes in PBS and two washes in fresh MCDB 131 medium.The bacterial pellets were suspended in 100 ml of PBS, 100 mlof PBS with antibodies to InlA in 1:100 dilution, 100 ml of PBSwith antibodies to InlB in 1:100 dilution and 100 ml of PBS witheach antibody in a final dilution of 1:100. The final concentrationof the antibodies used was 5 mg ml¹1. The tubes were incubatedat 378C for 15 min. The medium was aspirated, and 500 ml offresh MCDB 131 medium was added with antibodies in a finaldilution of 1:100 (PBS, a-InlA, a-InlB and a-InlA þ a-InlB).Bacterial suspension (10 ml) was added to cells in the wells.The assay was performed in quadruplicate. The infectionwas carried out for 2 h, followed by washing in HEPES–Hanks’ buffer and incubation with 50 mg ml¹1 gentamicin inMCDB 131 medium for 2 h. The cells were washed with buffer,the coverslips were fixed and the other wells were treated with0.2% Triton-X-100 for cell lysis. The lysis fluid was thendiluted 10 times in PBS and plated for cfu determination.

Purification of InlA and InlB

Internalin A protein was purified from supernatant cultures ofL. monocytogenes EGD DactA2 by affinity chromatographyusing specific monoclonal antibodies to InlA. Internalin B pro-tein was purified from L. monocytogenes strain actA2 þ pERL3

50-1 (Leimeister-Wachter et al., 1990) by affinity chromato-graphy purification using specific monoclonal antibodies toInlB (S. Muller et al., manuscript in preparation).

Antisera to InlA and InlB were raised by immunizing rabbitswith 25 mg of purified InlA or InlB emulsified with CFA. At 3and 6 weeks after injection, animals were boosted with furtherdoses (25 mg) of antigen in IFA. Sera were collected at vari-ous times after immunization and tested for their reactivitiesto purified protein by immunoblot analysis or enzyme-linkedimmunosorbent assay (ELISA). Affinity purification was per-formed as described previously (Lingnau et al., 1996).

Cell binding assay

Binding of internalin to HUVECs was determined by two dif-ferent methods – first, by binding to the purified internalin pro-tein coated onto 96-well Maxisorp microtitre plates (Nunc)and, subsequently, by flow cytometry analysis. A Maxisorpmicrotitre plate was coated with 100 ml of internalin B solutionin HEPES–Hanks’ buffer at concentrations ranging from0.05 mg to 1.6 mg for 1 h at 378C, followed by blocking with0.1% BSA in HEPES–Hanks’ buffer for 2 h at 378C. All experi-ments were performed in quadruplicate. The wells were thenwashed three times with buffer. HUVECs, at a concentrationof 106 ml¹1 in MCDB 131 medium with 0.4% BSA (a volumeof 50 ml; 50 000 cells), were added per well and incubated at378C for 1 h to assay attachment to internalin. In another setin parallel, HUVECs were put in the medium containing10 mM EDTA. After incubation, cells were washed with plainmedium or with medium containing 10 mM EDTA. Medium(100 ml) containing 10 ml of WST-1 (Boehringer Mannheim)was added per well, and the plate was incubated for 5 h at378C with CO2 before the optical density was measured in amicroplate autoreader (model EL310; Bio-Tek Instruments)at 450 nm.

Flow cytometric analysis

Confluent HUVECs were treated with 5 mM EDTA/PBS for20 min at RT, and cells were flushed from the flask. Cellswere washed with PBS and later with FACS buffer (PBS con-taining 2% FCS and 0.02% sodium azide). The cell pellet wassuspended in FACS buffer, distributed in the V-bottomed-wellmicrotitre plates and treated with purified InlA, InlB and oval-bumin/b-galactosidase proteins in different concentrationsfor 30 min on ice. After centrifugation, cells were washedwith FACS buffer before adding 50 ml of the first antibodies[mouse monoclonal anti-InlA (Lingnau et al., 1995)/rabbitpolyclonal anti-InlB/anti-b-galactosidase] per well in FACSbuffer (final antibody concentration was 0.5 mg) for 20 minon ice. The cells were washed and treated with appropriatefluorescein-conjugated secondary antibodies (goat anti-mousewhole IgG/goat anti-rabbit IgG) at a dilution of 1:100 for20 min on ice. Then, the cells were washed and suspendedin 1 mM propidium iodide solution before being transferred toand analysed by FACSort (Becton Dickinson flow cytometer).

After the analysis, the remaining cells were spun andmounted on slides with mowiol–glycerol for immunofluores-cence microscopy. Confocal microscopy was performedusing the MRC1024 imaging system (Bio-Rad) with the UBHSand E2 filter set using a krypton–argon laser.

Q 1998 Blackwell Science Ltd, Molecular Microbiology, 28, 81–93

L. monocytogenes invasion is dependent on internalin B 91

Electron microscopy

HUVECs were grown on the gelatine-coated Thermanox cover-slips (Nunc) until confluency. Thermanox coverslips were dippedin 70% alcohol for 2 h, followed by washing in sterile distilledwater, before the gelatine coating was made. Then, the cellswere infected with L. monocytogenes EGD and other mutantsin the cell–bacteria ratio of 1:100. The plates were centrifugedfor 4 min at 200 ×g at RT to enhance contact between the bac-teria and the cells. Then, the infection was continued for differ-ent time periods of 0, 5, 10, 15 and 30 min, followed by rinsingof the coverslips in HEPES–Hanks’ buffer. The coverslips werefixed with 3% glutaraldehyde in 0.1 mM cacodylate buffer for30 min at 48C, followed by three washings in cacodylate buf-fer. The coverslips were kept at 48C in cacodylate bufferuntil the preparation for electron microscopy was done.

The fixed samples were dehydrated using a graded ace-tone series (10–100%) on ice, allowed to warm up to roomtemperature and critical point dried using liquid CO2. Subse-quently, the samples were sputter coated with an approxi-mately 10-nm-thick gold film. Samples were examined in aZeiss DSM982 Gemini field emission electron microscopeusing the in lens detector and at an acceleration voltage of5 kV or 10 kV at calibrated magnifications.

Acknowledgements

We acknowledge the support of Professor Norbert Suttorp(Department of Internal Medicine) in providing us with umbili-cal cords for HUVEC isolation, Ms Heike Geisel for co-ordinat-ing the cord collection, Ms S. zur Lage and Ms Sabine Klein fortheir expert technical assistance in flow cytometry experi-ments. We thank Dr F. Ebel for his critical comments on themanuscript. This work was supported by grants from theDeutsche Forschungsgemeinschaft through SFB 249 (TPA13)and funding from the Commission of the European Union(contract BMH-CT96-0659) to T.C.

References

Berche, P. (1995) Bacteremia is required for invasion of themurine central nervous system by Listeria monocytogenes.Microb Pathog 18: 323–336.

Braun, L., Dramsi, S., Dehoux, P., Bierne, H., Lindahl, G.,and Cossart, P. (1997) InlB: an invasion protein of Listeriamonocytogenes with a novel type of surface association.Mol Microbiol 25: 285–294.

Chakraborty, T., and Wehland, J. (1997) The host cellinfected with Listeria monocytogenes. In Host Responseto Intracellular Pathogens, Kaufmann, S.H.E. (ed.). Austin,TX: R.G. Landes Company, pp. 271–290.

Chakraborty, T., Leimeister-Wachter, M., Domann, E., Hartl,M., Goebel, W., Nichterlein, T., et al. (1992) Coordinateregulation of virulence genes in Listeria monocytogenesrequires the product of the prfA gene. J Bacteriol 174:568–574.

Domann, E., Zechel, S., Lingnau, A., Hain, T., Darji, A., Nich-terlein, T., et al. (1997) Identification and characterizationof a novel prfA-regulated gene in Listeria monocytogeneswhose product, IrpA, is highly homologous to internalin

proteins, which contain leucine-rich repeats. Infect Immun65: 101–109.

Dramsi, S., Biswas, I., Maguin, E., Braun, L., Mastroeni, P.,and Cossart, P. (1995) Entry of Listeria monocytogenesinto hepatocytes requires expression of inIB, a surface pro-tein of the internalin multigene family. Mol Microbiol 16:251–261.

Dramsi, S., Lebrun, M., and Cossart, P. (1996) Molecular andGenetic determinants interfere in invasion of mammaliancells by Listeria monocytogenes. Curr Topics MicrobiolImmunol 209: 61–77.

Dramsi, S., Dehoux, P., Lebrun, M., Goossens, P.L., andCossart, P. (1997) Identification of four new members ofthe Internalin multigene family of Listeria monocytogenesEGD. Infect Immun 65: 1615–1625.

Drevets, D.A., Sawyer, R.T., Potter, T.A., and Campbell,P.A. (1995) Listeria monocytogenes infects human endo-thelial cells by two distinct mechanisms. Infect Immun63: 4268–4276.

Ebel, F., Sibelius, U., Grimminger, F., Wehland, J., andChakraborty, T. (1994) Human umbilical cord cells, a novelinfection model for Listeria monocytogenes (abstract). Bio-engineering 3: 69.

Engelbrecht, F., Chun, S.K., Ochs, C., Hess, J., Lottspeich,F., Goebel, W., et al. (1996) A new prfA-regulated geneof Listeria monocytogenes encoding a small, secreted pro-tein which belongs to the family of internalins. Mol Microbiol21: 823–837.

Finlay, B.B., and Cossart, P. (1997) Exploitation of mammal-ian host cell functions by bacterial pathogens. Science276: 718–725.

Gaillard, J.L., Berche, P., Frehel, C., Gouin, E., and Cossart,P. (1991) Entry of L. monocytogenes into cells is mediatedby internalin, a repeat protein reminiscent of surface anti-gens from Gram-positive cocci. Cell 65: 1127–1141.

Gaillard, J.L., Jaubert, F., and Berche, P. (1996) The inlABlocus mediates the entry of Listeria monocytogenes intohepatocytes in vivo. J Exp Med 183: 359–369.

Gray, M.L., and Killinger, A.H. (1966) Listeria monocyto-genes and listeric infections. Bacteriol Rev 30: 309–382.

Gregory, S.H., Sagnimeni, A.J., and Wing, E.J. (1996)Expression on the inlAB operon by Listeria monocyto-genes is not required for entry into hepatic cells in vivo.Infect Immun 64: 3983–3986.

Greiffenberg, G., Sokolovic, Z., Schnittler, H.-J., Spory, A.,Boeckmann, R., Goebel, W., et al. (1997) Listeria mono-cytogenes-infected human umbilical vein endothelial cells:internalin-independent invasion, intracellular growth, move-ment, and host cell responses. FEMS Microbiol Lett 157:163–170.

Ireton, K., Payrastre, B., Chap, H., Ogawa, W., Sakaue, H.,Kasuga, M., et al. (1996) A role for phosphoinositide 3kinase in bacterial invasion. Science 274: 780–782.

Jaffe, E.A., Nachman, R.L., Becker, C.G., and Minick, C.R.(1973) Culture of human endothelial cells derived fromumbilical veins. Identification by morphologic and immuno-logic criteria. J Clin Invest 52: 2745–2756.

Jensen, A., Frederiksen, W., and Gerner Smidt, P. (1994)Risk factors for listeriosis in Denmark, 1989–90. Scand JInfect Dis 26: 171–178.

Jurado, R.L., Farley, M.M., Pereira, E., Harvey, R.C.,

Q 1998 Blackwell Science Ltd, Molecular Microbiology, 28, 81–93

92 S. K. Parida et al.

Schuchat, A., Wenger, J.D., et al. (1993) Increased risk ofmeningitis and bacteremia due to Listeria monocytogenesin patients with human immunodeficiency virus infection.Clin Infect Dis 17: 224–227.

Kaufmann, S.H.E. (1984) Acquired resistance to facultativeintracellular bacteria: relationship between persistence,cross-reactivity at the T-cell level, and capacity to stimulatecellular immunity of different Listeria strains. Infect Immun45: 234–241.

Kobe, B., and Deisenhofer, J. (1994) The leucine-rich repeat:a versatile binding motif. Trends Biochem Sci 19: 415–421.

Krull, M., Nost, R., Hippenstiel, S., Domann, E., Chakraborty,T., and Suttorp, N. (1997) Listeria monocytogenes potentlyinduces upregulation of endothelial adhesion moleculesand neutrophil adhesion to cultured human endothelialcells. J Immunol 159: 1970–76.

Lebrun, M., Mengaud, J., Ohayon, H., Nato, F., and Cossart,P. (1996) Internalin must be on the bacterial surface tomediate entry of Listeria monocytogenes into epithelialcells. Mol Microbiol 21: 579–592.

Leimeister-Wachter, M., Haffner, C., Domann, E., Goebel,W., and Chakraborty, T. (1990) Identification of a genethat positively regulates expression of listeriolysin, themajor virulence factor of Listeria monocytogenes. ProcNatl Acad Sci USA 87: 8336–8340.

Leimeister-Wachter, M., Domann, E., and Chakraborty, T.(1992) The expression of virulence genes in Listeria mono-cytogenes is thermoregulated. J Bacteriol 174: 947–952.

Lingnau, A., Domann, E., Hudel, M., Bock, M., Nichterlein, T.,Wehland, J., et al. (1995) Expression of the Listeria mono-cytogenes EGD inlA and inlB genes, whose productsmediate bacterial entry into tissue culture cell lines, byprfA-dependent and -independent mechanisms. InfectImmun 63: 3896–3903.

Lingnau, A., Chakraborty, T., Niebuhr, K., Domann, E., andWehland, J. (1996) Identification and purification of novelinternalin related proteins in Listeria monocytogenes andListeria ivanovii. Infect Immun 64: 1002–1006.

Makhov, A.M., Hannah, J.H., Brennan, M.J., Trus, B.L., Koc-sis, E., Conway, J.F., et al. (1994) Filamentous hemagglu-tinin of Bordetella pertussis. A bacterial adhesin formed asa 50-nm monomeric rigid rod based on a 19-residue repeatmotif rich in beta strands and turns. J Mol Biol 241: 110–124.

Marco, A.J., Prats, N., Ramos, J.A., Briones, V., Blanco, M.,Dominguez, L., et al. (1992) A microbiological, histopatho-logical and immunohistological study of the intragastricinoculation of Listeria monocytogenes in mice. J CompPathol 107: 1–9.

Mengaud, J., Ohayon, H., Gounon, P., Mege, R.M., and

Cossart, P. (1996a) E cadherin is the receptor for inter-nalin, a surface protein required for entry of L. monocyto-genes into epithelial cells. Cell 84: 923–932.

Mengaud, J., Lecuit, M., Lebrun, M., Nato, F., Mazie, J.C.,and Cossart, P. (1996b) Antibodies to the leucine richrepeat region of internalin block entry of Listeria monocyto-genes into cells expressing E cadherin. Infect Immun 64:5430–5433.

Niebuhr, K., Chakraborty, T., Rohde, M., Gazlig, T., Jansen,B., Kollner, P., et al. (1993) Localization of the ActA poly-peptide of Listeria monocytogenes in infected tissue cul-ture cell lines: ActA is not associated with actin ‘comets’.Infect Immun 61: 2793–2802.

Schuchat, A., Swaminathan, B., and Broome, C.V. (1991)Epidemiology of human listeriosis. Clin Microbiol Rev 4:169–183.

Shevchenko, D.V., Akins, D.R., Robinson, E., Li, M., Popova,T.G., Cox, D.L., et al. (1997) Molecular characterizationand cellular localization of TpLRR, a processed leucine-rich repeat protein of Treponema pallidum, the syphilisspirochete. J Bacteriol 179: 3188–3195.

Sibelius, U., Rose, F., Chakraborty, T., Darji, A., Wehland, J.,Weiss, S., et al. (1996) Listeriolysin is a potent inducer ofthe phosphatidylinositol response and lipid mediator gene-ration in human endotheIial cells. Infect Immun 64: 674–676.

Swanson, J.A., and Baer, S.C. (1995) Phagocytosis by zip-pers and triggers. Trends Cell Biol 5: 89–93.

Temm Grove, C.J., Jockusch, B.M., Rohde, M., Niebuhr, K.,Chakraborty, T., and Wehland, J. (1994) Exploitation ofmicrofilament proteins by Listeria monocytogenes: micro-villus-like composition of the comet tails and vectorialspreading in polarized epithelial sheets. J Cell Sci 107:2951–2960.

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

Uldry, P.-A., Kuntzer, T., Bogousslavsky, J., Regli, F., Mik-lossy, J., Bille, J., et al. (1993) Early symptoms and out-come of Listeria monocytogenes rhombencephalitis: 14adult cases. J Neurol 240: 235–242.

Velge, P., Bottreau, E., Kaeffer, B., and Pardon, P. (1994)Cell immortalization enhances Listeria monocytogenesinvasion. Med Microbiol Immunol 183: 145–158.

Virji, M. (1996) Microbial utilization of human signalling mol-ecules. Microbiology 142: 3319–3336.

Wuenscher, M.D., Kohler, S., Goebel, W., and Chakraborty,T. (1991) Gene disruption by plasmid integration in Listeriamonocytogenes: insertional inactivation of the listeriolysindeterminant lisA. Mol Gen Genet 228: 177–182.

Q 1998 Blackwell Science Ltd, Molecular Microbiology, 28, 81–93

L. monocytogenes invasion is dependent on internalin B 93