The Staphylococcus aureus Epidermal Cell Differentiation InhibitorToxin Promotes Formation of Infection Foci in a
Mouse Model of Bacteremia�
Patrick Munro,1* Maxime Benchetrit,5 Marie-Anne Nahori,2,3,4 Caroline Stefani,1 Rene Clement,1Jean-Francois Michiels,5 Luce Landraud,1,6 Olivier Dussurget,2,3,4 and Emmanuel Lemichez1,6
INSERM, U895, C3M, Toxines Microbiennes dans la Relation Hote Pathogenes, Universite de Nice-Sophia-Antipolis, UFR Medecine,IFR50, Nice 06204, France1; Institut Pasteur, Unite des Interactions Bacteries-Cellules, Paris F-75015, France2; INSERM, U604,
Paris 75015, France3; INRA, USC2020, Paris 75015, France4; Service d’Anatomie Pathologique, Hopital Pasteur, CHU deNice, 30 Avenue de la Voie-Romaine, BP 69, Nice 06002, France5; and Laboratoire de Bacteriologie,
CHU de Nice, Hopital l’Archet, Nice 06204, France6
Received 30 March 2010/Returned for modification 3 May 2010/Accepted 7 May 2010
Inactivation of the host GTPase RhoA by staphylococcal epidermal cell differentiation inhibitor (EDIN)exotoxins triggers the formation of large transcellular tunnels, named macroapertures, in endothelial cells. Weused bioluminescent strains of Staphylococcus aureus to monitor the formation of infection foci during the first24 h of hematogenous bacterial dissemination. Clinically derived EDIN-expressing S. aureus strains S25 andXen36 produced many disseminated foci. EDIN had no detectable impact on infection foci in terms ofhistopathology or the intensity of emitted light. Moreover, EDIN did not modify the course of bacterialclearance from the bloodstream. In contrast, we show that EDIN expression promotes a 5-fold increase in thenumber of infection foci produced by Xen36. This virulence activity of EDIN requires RhoA ADP-ribosyltran-ferase activity. These results suggest that EDIN is a risk factor for S. aureus dissemination through thevasculature by virtue of its ability to promote the formation of infection foci in deep-seated tissues.
Defining how pathogenic bacteria interact with the endothe-lium is of major relevance to evaluating the risk of severe formsof infections (21). Recent findings show that the epidermal celldifferentiation inhibitor (EDIN) exotoxins trigger the forma-tion of large transcellular tunnels, termed macroapertures, inendothelial cells of various vascular beds (5). EDINs belong toa large group of virulence factors produced by human patho-genic bacteria that target host Rho proteins (2, 4). As RhoGTPases are master regulators of the host cell actin cytoskel-eton, they play a central role in controlling cell adhesion,migration, and phagocytosis (18). Moreover, Rho proteins con-trol the formation and integrity of both intercellular adherenceand tight junctions (11) and are thus major regulators of theendothelium barrier function (27, 36). Rho proteins also playessential roles in the transcellular or intercellular modes ofdiapedesis of leukocytes through the endothelium (6).
Staphylococcal colonization of the skin and mucosa is a riskfactor for bacterial translocation to underlying tissues and thebloodstream. Failure to contain the initial infection can lead tosepsis or invasion of deeper tissues, leading to endocarditis,septic arthritis, and osteomyelitis (34). Staphylococcal infec-tions involve the combined actions of a large panel of virulencefactors, promoting bacterial colonization, destruction of tis-sues, and immune evasion (9, 12). These bacterial virulencefactors comprise secreted exotoxins and cell surface-associatedfactors (22). Pathogenic Staphylococcus aureus can produce
enzymes such as proteases, nucleases, hyaluronidases, lipases,and cytolytic toxins. These factors participate in the destruc-tion of host tissues and may favor bacterial dissemination inthe host tissues.
Several pathogenic strains of S. aureus produce EDIN typeA (EDIN-A), EDIN-B, or EDIN-C (3, 15, 33, 37–39). Whereasthe gene encoding EDIN-B is located on the chromosomewithin a pathogenicity island (15, 26, 38), genes encodingEDIN-A and EDIN-C are plasmid borne (37, 38). Strainscarrying the exfoliative toxin type D gene (etd) also harbor theedin-B gene (3, 15, 38). These strains correspond to the se-quence type 80 clone of Panton-Valentine leukocidin-positivecommunity-acquired (CA) methicillin-resistant S. aureus(MRSA), which continues to spread through Europe (39) andTunisia (3). Although genes encoding EDIN have a higherprevalence in pathogenic isolates of S. aureus (10), the contri-bution of EDINs to bacterial virulence remains to be defined.
EDIN exotoxins translocate into the host cell cytosol fromacidic compartments following macropinocytosis or after inter-nalization into phagosomes (24). Upon reaching the cytosol,EDINs preferentially mono-ADP-ribosylate RhoA, with thismodification occurring at asparagine-41 (8, 31, 35). Posttrans-lational modification of RhoA by ADP-ribosylation induces itstight association with RhoGDI, producing its release frommembranes, where RhoA transduces signals (16, 17). Inacti-vation of RhoA blocks a major pathway responsible for bothactin filament elongation and assembly into contractile acto-myosin cables (8, 18, 28). In endothelial cells, activation ofRhoA controls the formation of intercellular gaps by inducingcontractile actomyosin fibers, which pull on intercellular bor-ders (23). Thus, inactivation of RhoA by ADP-ribosylationreinforces the cohesion of adherence junctions in the endothe-
* Corresponding author. Mailing address: Batiment UniversitaireArchimed, INSERM U895, Equipe 6, C3M, 151 Route de Saint An-toine de Ginestiere, BP 2 3194, Nice F-06204 Cedex 3, France. Phone:33 4 89 06 42 61. Fax: 33 4 89 06 42 21. E-mail: [email protected].
lium (36). Nevertheless, recent advances show that inhibitionof RhoA impairs endothelium barrier function by producingtranscellular tunnels, referred to as macroapertures (5). In-deed, endothelial cell intoxication either by any of the threeisoforms of EDIN or by infection by EDIN-producing S. aureustriggers the formation of these transcellular tunnels. Macroap-ertures are also formed in the endothelium lining rat arteriesinfected ex vivo by EDIN-producing S. aureus (5). Macroaper-tures unmask the fibril matrix underneath the endotheliumwhere bacteria adhere.
The capacity of EDIN to open macroapertures in vitro andex vivo raised the possibility that this family of exotoxins mightfavor S. aureus hematogenous spread to deeper tissues (21).We investigated this by using bioluminescent strains of S. au-reus in a mouse model of infection. Here we show that EDINexpression confers on S. aureus a greater capacity to formdisseminated infection foci during bacteremia.
MATERIALS AND METHODS
Bacterial strains and growth conditions. The staphylococcal S25 strain waspreviously isolated from a bacteremic patient suffering from spondylodiscitis (5).This strain was previously cured of its edin-bearing plasmid, using the plasmid-associated cadmium resistance, and transformed with either pMK4-pPROTexpressing the EDIN wild type (S25-EDIN-Wt), pMK4-pPROT expressing cat-alytically inactive EDINR185E (EDIN-RE) (S25-EDIN-RE), or the pMK4-pPROT empty vector (S25-EV) (5). Bioluminescent S25 was generated by trans-formation with pXen5 (pAUL-A Tn4001 luxABCDE Kmr) (14). Stablebioluminescent transformants were selected as described previously (19).
The Xen36 bioluminescent strain (Caliper Life Sciences Inc.) was derivedfrom S. aureus ATCC 49525 (Wright), a clinical isolate from a bacteremicpatient. We established by PCR that this strain does not express the edin-A, -B,or -C gene. The Xen36 strain was transformed with pMK4-pPROT expressingthe EDIN wild type (Xen36 EDIN-Wt), pMK4-pPROT expressing catalyticallyinactive EDINR185E (Xen36 EDIN-RE), or the pMK4-pPROT empty vector(Xen36-EV), as described previously (5). The strains were grown at 37°C in brainheart infusion (BHI) medium (Xen36) or BHI medium supplemented with 20�g/ml chloramphenicol (Xen36 EDIN-Wt, Xen36 EDIN-RE, Xen36-EV). AllXen36-derived strains had similar growth rates in BHI broth and luminescenceintensities.
Animals and experimental model of infection. Six- to 8-week-old femaleBALB/c mice were purchased from Charles River (L’Arbresle, France). Themice were maintained and handled according to the regulations of the EuropeanUnion and the French Department of Health. For infections, overnight culturesof each strain of S. aureus were centrifuged and washed three times with phos-phate-buffered saline (PBS). The pellets were diluted in PBS to obtain thedesired bacterial concentrations. The lateral tail vein of each mouse was injectedwith 100 �l of bacterial suspension. For determination of bacteremia, blood wascollected in a heparin-containing syringe from the tail vein, serially diluted inPBS, and plated on BHI medium plates, which were incubated for 24 h at 37°C.
Infection monitoring by bioluminescence imaging. The lateral tail vein of eachmouse was injected with 100 �l of a suspension of bioluminescent S. aureusstrains. Imaging was performed at different time points after inoculation, usingan IVIS100 imaging system, according to the manufacturer’s instructions (Cali-per Life Sciences Inc.). Briefly, groups of five mice were anesthetized with 2.5%(vol/vol) isoflurane and placed in the IVIS100. Photons were counted over 5 min,and data analysis was performed using the Living Image (version 3.0) program(Caliper Life Sciences Inc.).
Histopathology. After identification of infection foci by bioluminescence im-aging, the mice were euthanized and organs corresponding to the infection fociwere collected, fixed in formalin, and embedded in paraffin. Consecutive 3-�mparaffin sections were stained with hematoxylin-eosin-saffron (HES). In parallel,Gram staining was performed to visualize the presence of bacteria.
Statistics. Analysis of variance with the Bonferroni post hoc test was used forstatistical analyses, unless indicated otherwise in the figure legends. Analyseswere performed with Prism (version 5.0b) software. Values were consideredsignificant if P was �0.05.
RESULTS
EDIN-positive S. aureus strains produce disseminated bio-luminescent foci during bacteremia. Induction of macroaper-tures in endothelial cells by EDIN exotoxins represents a po-tential virulence mechanism that may favor the disseminationof staphylococci in tissues via a hematogenous route. We in-vestigated this by monitoring infection in living mice using apreviously described bioluminescence technology (13). Isola-tion of clinical strain S. aureus S25 from the blood of a patientsuffering from a spondylodiscitis was described previously (5).We transformed this strain with a plasmid containing a Pho-torhabdus luminescens lux operon (luxABCDE) and selectedclones on the basis of their capacity to emit light. This allowedus to generate bioluminescent strain S25lux. We verified thatS25lux triggered actin cable disruption and induction of mac-roapertures in endothelial cells, as previously observed for S25(5) (Fig. 1A). Mice were infected by intravenous inoculation of5 � 107 CFU of S25lux. The light emitted was recorded 10 minafter infection and then 3 h and 24 h after infection. These dataindicate that bioluminescent foci formed 24 h after infection(Fig. 1B). Figure 1C shows two examples of histology analysisof bioluminescent foci that confirm the inflammation asso-ciated with the presence of bacteria. This revealed the ca-pacity of S25lux staphylococci to produce disseminated bio-luminescent infection foci as early as 24 h after induction ofbacteremia.
Clearance of S. aureus from the bloodstream. We suspectedthat EDIN might favor bacterial dissemination in tissues bydelaying the clearance of S. aureus from the bloodstream. Weused previously described recombinant strains engineeredfrom S25 to test this hypothesis (5). Briefly, S25 was cured ofthe native edin-bearing plasmid and transformed with theempty vector (S25) or plasmids encoding either wild-typeEDIN (S25-EDIN-Wt) or a catalytically inactive mutant ofEDIN (S25-EDIN-RE). The recombinant strains had similarkinetics of growth and the stable retention of plasmids 24 hafter the onset of bacteremia (data not shown). Mice wereinfected by intravenous inoculation with 106 CFU, deliveredthrough the tail vein. Survival of the bacteria in the blood-stream was assessed 3 h and 24 h after infection. The staphy-lococcal load was enumerated by plating blood samples onagar plates and measuring colony formation. For all strains, wemeasured at 3 h a dramatic decrease (at least 4 log units) in thenumber of viable bacteria in the blood (Fig. 2). No significantdifferences in the recovery of CFU were measured betweenthese strains 3 h and 24 h after infection (Fig. 2). Similar resultswere obtained with mice intravenously infected with 2 � 107
CFU (data not shown). Collectively, this shows an absence ofa detectable effect of EDIN on the kinetics of clearance ofstaphylococci from the bloodstream.
Formation of infection foci. Our findings prompted us toinvestigate the role of EDIN in the formation of disseminatedbioluminescent foci during staphylococcal bacteremia. Owingto technical limitations associated with introducing the Luxoperon in the genomic DNA of S25 and S25 cured of theedin-bearing plasmid, we made use of staphylococcal isolateXen36, which was initially recovered from a patient with bac-teremia and previously engineered to express luxABCDE (Cal-iper Life Sciences Inc.). We determined by PCR that Xen36 is
negative for edin-A, -B, and -C (data not shown). We thusaddressed the question of whether acquisition of an edin-bear-ing plasmid confers to the bacterium a higher capacity to formdisseminated infection foci during bacteremia. Mice were first
infected intravenously with 108 CFU of Xen36. The light emit-ted was recorded prior to infection (0 h) and then 3 h and 24 hafter infection. Bioluminescent foci appeared 24 h after infec-tion (Fig. 3A). Bioluminescent tissues were next analyzed after
FIG. 1. Induction of disseminated bioluminescent foci by EDIN-producing S25. (A) S. aureus bioluminescent S25 strain (S25lux) triggers thedisruption of actin cables and the formation of macroapertures (inset). Human umbilical vein endothelial cells were treated overnight with the S25luxsupernatant (dilution, 1/50). The supernatant was prepared from bacteria grown to an optical density at 600 nm of 1. The actin cytoskeleton was visualizedusing tetramethyl rhodamine isocyanate-conjugated phalloidin. Bar, 20 �m. (B) Bioluminescence imaging of infection. Five BALB/c mice were injectedintravenously with 5 � 107 S25lux bacteria. Dorsal and ventral imaging of the mice was performed at 10 min, 3 h, and 24 h postinfection. The color scaleindicates the signal intensity in numbers of photons/second/cm2. (C) Histopathology of infection foci. Pictures show hematoxylin-eosin-saffron stainingand the corresponding Gram staining of selected foci (arrows) in the heart at a high magnification (�400). BALB/c mice were injected intravenously with5 � 107 CFU/animal of S25lux. The color scale indicates the signal intensity in numbers of photons/second/cm2.
hematoxylin-eosin-saffron staining and Gram staining of thinsections. Figure 3B shows two examples of histology analysis ofinfection foci that confirm the inflammation associated withthe presence of clusters of bacteria. Taken together, these datashow the capacity of Xen36 to produce bioluminescent infec-tion foci.
Expression of EDIN favors formation of disseminated in-fection foci. We next examined whether acquisition of theedin gene might favor the formation of infection foci. Xen36was transformed with the empty pMK4-pPROT vector(Xen36-EV) or vectors encoding wild-type EDIN (Xen36EDIN-Wt) or catalytically inactive EDIN (Xen36 EDIN-RE). All recombinant strains of S. aureus showed similargrowth kinetics and levels of light emission (data not
FIG. 2. Bacterial clearance from the bloodstream by enumerationof bacteria in the blood of mice infected by S. aureus S25 derivatives.Three groups of 8 to 10 BALB/c mice were infected with 106 CFU/animal of S25, S25-EDIN-Wt, or S25-EDIN-RE. Enumeration of thebacteria was performed by measuring the numbers of CFU on LB agarplates in the absence of antibiotics. Values for each mouse are shown.Statistical analysis shows no significant differences (ns; P � 0.05).
FIG. 3. Monitoring of infection in a mouse model of Xen36 bacteremia. (A) Five BALB/c mice were injected intravenously with 108 CFU/animal ofXen36-EV. Dorsal and ventral imaging was performed on noninfected (0 h) mice and at 3 h and 24 h postinfection. The color scale indicates the signalintensity in numbers of photons/second/cm2. (B) Histopathology of infection foci. Pictures show hematoxylin-eosin-saffron staining and the correspondingGram staining of selected foci (arrows) in the kidney and muscle at high magnification (�400). BALB/c mice were injected intravenously with 108
CFU/animal of Xen36(pMK4-pPROT). The color scale indicates the signal intensity in numbers of photons/second/cm2.
shown). We also verified that expression of wild-type EDINspecifically triggered disruption of the actin cables and for-mation of macroapertures in endothelial cells (Fig. 4A, ar-rows). We next investigated the effect of EDIN in a mousemodel of bacteremia. Mice were intravenously infected with5 � 107 CFU of Xen36 EDIN-Wt or Xen36 EDIN-RE.Control experiments confirmed the stability of the edin-expressing plasmids over the 24 h following infection (datanot shown). Under these conditions, we visualized the for-mation of disseminated bioluminescent foci 24 h after in-fection (Fig. 4B). Foci were defined for light intensities
2-fold higher than the background (�104 photons/s). Mostfoci of the different bioluminescent strains were detected atthe level of the sacral part of the spinal column and the legjoints. Importantly, we measured differences in the numberof foci between Xen36 EDIN-Wt- and Xen36 EDIN-RE-infected animals (Fig. 4C). Xen36 EDIN-Wt produced up tosix infection foci per mouse. The mean value was �2.5foci/mouse for Xen36 EDIN-Wt, whereas the mean valuewas �0.5 focus/mouse for Xen36 EDIN-RE (Fig. 4C). Thus,expression of catalytically active EDIN in Xen36 signifi-cantly increases the rate of formation of disseminated bi-
FIG. 4. Induction of disseminated bioluminescent foci by EDIN-producing Xen36. (A) Supernatant of Xen36-expressing EDIN-Wt specificallytriggers actin depolymerization and formation of macroapertures. Human umbilical vein endothelial cells were treated overnight with Xen36-EV,Xen36 EDIN-Wt, or Xen36 EDIN-RE supernatants (dilution, 1/50) prepared from bacteria grown to an optical density at 600 nm of 1. The actincytoskeleton was visualized using tetramethyl rhodamine isocyanate-conjugated phalloidin. Bar, 20 �m. (B) Example of BALB/c mice injectedintravenously with 5 � 107 CFU/animal of Xen36 EDIN-Wt. Dorsal and ventral imaging of mice was performed 24 h after infection. The colorscale indicates the signal intensity in numbers of photons/second/cm2. (C and D) Groups of BALB/c mice were injected intravenously with 5 �107 CFU/animal of Xen36-EV, Xen36 EDIN-Wt, or Xen36 EDIN-RE. (C) Number of infection (nb. inf.) foci for each animal. The number ofinfection foci was estimated for 10 mice (EDIN-EV) and 15 mice (EDIN-Wt and EDIN-RE) in two independent experiments. Infection foci werecounted for signals of a relative intensity of 2-fold over the background. (D) Mean values of bioluminescent signals at infection foci measured forone constant selected tissue surface � standard error of the mean. Infection foci were counted for signals of a relative intensity of 2-fold over thebackground (�104 photons/second). Statistical analysis shows significant (P � 0.01) and nonsignificant (ns; P � 0.05) differences.
oluminescent foci during bacteremia (P � 0.01). We nextrecorded and quantified the light signals emitted at the foci.These measurements revealed that the intensities of lightemitted at the foci were similar under all conditions (Fig.4D). We verified that the light emitted by infection foci atthe level of the sacral part of the spinal column or leg jointsbetween the EDIN-Wt and RE conditions were also of sim-ilar intensities (data not shown). This suggested that EDINhad no detectable effect on the growth and survival of thebacteria in tissues under these conditions. Thus, acquisitionof an edin-expressing plasmid confers to Xen36 a greatercapacity to form disseminated infection foci.
We then investigated whether the expression of EDINunder these conditions might affect the course of bacteremiaby examining the possible impact of EDIN on bacterialclearance from the bloodstream. We observed that the lev-els of viable bacteria dropped dramatically as early as 24 hafter infection (Fig. 5). Importantly, we measured no signif-icant differences in bacterial clearance in the bloodstream ofmice infected with Xen36 producing either catalytically ac-tive or inactive EDIN.
Finally, we conducted histopathological analysis of the bio-luminescent foci. This analysis was conducted on mice infectedby Xen36 EDIN-Wt (six samples), Xen36 EDIN-RE (foursamples), and Xen36 containing the empty vector (four sam-ples). Representative examples of the infection foci analyzedare shown in Fig. 6. Microscopic examination of samples re-vealed acute inflammatory processes characterized by a local-ized infiltrate of polymorphonuclear leukocytes with scatteredmacrophages. Infection foci localized at different sites, i.e.,renal interstitium and tubules (pyelonephritis), as well as sacralfibrocartilage (spondylodiscitis) and femoral bone marrow (os-teomyelitis). No microscopic differences between infection fociproduced by the various strains derived from Xen36 were ob-served. Moreover, we noticed a good correlation between theintensity of the fluorescent signal and the extent of the inflam-matory reaction observed by histopathological analysis.
Together these observations show that expression of EDINconfers to S. aureus a greater capacity to form disseminatedinfection foci during bacteremia.
DISCUSSION
Determining the mechanisms of dissemination of S. aureusin host tissues is of major interest, given the high incidence andrate of mortality from staphylococcal infections. In order toaddress this challenge, we used a bioluminescence technologyto investigate the role of EDIN in S. aureus-induced bactere-mia. We show that expression of catalytic active EDIN inclinical strain Xen36 specifically increases the capacity of bac-teria to induce disseminated infection focus formation. Thebioluminescent foci produced by Xen36 EDIN-Wt and Xen36EDIN-RE were of similar light intensities. Moreover, his-topathological analysis revealed no differences between theinfection foci produced by either strain. Finally, we show thatEDIN plays no role in bacterial survival in the bloodstream.Collectively, these data show that EDIN favors the formationof disseminated infection foci during S. aureus-induced bacte-remia. We propose that EDIN is a risk factor for deep-seatedstaphylococcal tissue infections following bacteremia.
Several potent virulence factors found in human pathogenicbacteria target Rho proteins or actin itself (2, 4). Rho proteinsare implicated in a large number of key cellular processes, byvirtue of their ability to control the dynamics of the actincytoskeleton. The targeting of Rho proteins by bacterial viru-lence factors thus likely confers multiple advantages to patho-genic bacteria for host colonization and invasion. Several cellbiology studies have clearly established that EDIN exotoxinsand their close homologues, the C3 exoenzymes of Clostridiumbotulinum, can inhibit immune cell migration and complementreceptor phagocytosis (2, 7). Rather than triggering direct in-vasion of tissues by bacteria, it seems possible that EDIN mightfavor the formation of infection foci by blocking immune cellchemotaxis and function. This hypothesis is unlikely, consider-ing that both EDIN-Wt- and EDIN-RE-expressing S. aureusstrains have similar kinetics of clearing from the bloodstream.Moreover, the infection foci triggered by EDIN-Wt andEDIN-RE had similar light intensities and histopathologiesand were recorded early after infection. Together our datasupport the hypothesis that EDIN might favor the exit ofbacteria from the bloodstream rather than bacterial survival.
S. aureus can colonize the skin and mucosa of healthy indi-viduals (22). This gives to S. aureus a high propensity to reachthe bloodstream when these physical barriers are breached. S.aureus is frequently isolated from patients with bacteremia.These infections can evolve to cause endocarditis or infectionsof the bones, joints, kidneys, and lungs. Various clinical strainsof S. aureus produce EDIN exotoxins (3, 10, 26, 39). Epidemi-ological studies related to EDIN have not yet shed light on itsvirulent function in infection (15). Contrasting with this, cellbiology studies of EDIN and the highly homologous C3 exoen-zyme of Clostridium botulinum indicate that these factors canplay roles that range from hijacking immune cell responses tocompromising the integrity of the epithelium and endotheliumbarriers (1, 25). Indeed, we recently reported that EDIN,through inhibition of RhoA, specifically triggers the formationof large transcellular tunnels in endothelial cells (5). We haveshown that infection of endothelial cells with S. aureus produc-ing catalytically active EDIN also triggers the formation ofmacroapertures. Our analysis of the effect of EDIN in a modelof rat arteries infected ex vivo with S. aureus producing EDIN
FIG. 5. Bacterial clearance from the bloodstream by enumera-tion of bacteria in the blood of mice infected by derivatives of theS. aureus Xen36 strain. Groups of 7 to 10 BALB/c mice wereinfected with either 107 or 5 � 107 CFU/animal of S. aureus Xen36EDIN-Wt or Xen36 EDIN-RE. Enumeration of the bacteria wasperformed by measuring the numbers of CFU on LB agar plateswithout antibiotics. Values are shown for each mouse. Statisticalanalysis was performed with unpaired two-tailed t tests (ns, nonsig-nificant; P � 0.5).
revealed the formation of large macroapertures (5). This un-masks the subendothelial matrix where S. aureus binds (5).Here we provide the first demonstration that expression ofEDIN favors invasion of tissues by S. aureus and that thisvirulence activity of EDIN requires RhoA ADP-ribosyltrans-ferase activity.
An animal study showed that following bacteremia, S. aureuscells preferably adhere to capillaries and postcapillary venules(20). Postcapillary venules form a zone with a high level ofexchange between the blood and tissues. They are particularlyenriched in fenestrae and large discontinuities or gaps (20). Byexposing matrix proteins, these gaps may offer to S. aureus cellsboth a surface to which they can bind and a way to disseminateinto the surrounding tissue. Other ways of endothelium cross-
ing, such as transcytosis, might also participate in bacterialdissemination. S. aureus can bind to endothelial cell �5�1-integrin via fibronectin and the bacterial fibronectin-bindingproteins (32). Nevertheless, especially in light of the fact thatbacteria entering endothelial cells can be killed on a massivescale by the bactericidal activity of lysosomes, the efficiency ofS. aureus transcytosis remains unclear. Rather than providing amechanism of cell invasion, it has been proposed that thisphenomenon dramatically increases the mobility of bacteria atthe endothelial cell surface and postpones their endocytosis(30). It was suggested that this might give bacteria enough timeto produce cell-damaging toxins in order to survive and/ordisseminate (29). It is possible that EDIN plays a key role atthis step, allowing bacteria to dig transcellular tunnels and
FIG. 6. Histopathology of bioluminescent foci. Selected renal (A and B), osseous (C and D), and sacral (E and F) infection foci are shown.BALB/c mice were injected intravenously with 5 � 107 CFU/animal of Xen36-EV (control; A and C), Xen36 EDIN-Wt (B, D, and F), or Xen36EDIN-RE (E). Pictures of mice show the localization of infection foci analyzed (arrows), as well as their relative intensities. Microscopicexamination of selected foci shows the acute inflammatory process. Magnifications: �200 (A and B), �40 (C and D), and �100 (E and F).
allowing them to access the endothelium basement membranedirectly. In addition, EDIN might enlarge preexisting porespresent in postcapillary venules. Consistent with this, it hasbeen shown that the inhibition of RhoA in fenestrated hepaticsinusoidal endothelial cells leads to a dilatation and fusion offenestrae into large gaps (40).
Together, our data ascribe to EDIN an invasive virulenceproperty during staphylococcal bacteremia.
ACKNOWLEDGMENTS
Our laboratory, Toxines Microbiennes dan la Relation HôtePathogènes, is supported by institutional funding from theINSERM, a grant from the Agence Nationale de la Recherche (grantANR RPV07055ASA), and a grant from the Association pour la Re-cherche sur le Cancer (grant ARC 4906).
REFERENCES
1. Aktories, K., C. Wilde, and M. Vogelsgesang. 2004. Rho-modifying C3-likeADP-ribosyltransferases. Rev. Physiol. Biochem. Pharmacol. 152:1–22.
2. Barbieri, J. T., M. J. Riese, and K. Aktories. 2002. Bacterial toxins thatmodify the actin cytoskeleton. Annu. Rev. Cell Dev. Biol. 18:315–344.
3. Ben Nejma, M., M. Mastouri, B. Bel Hadj Jrad, and M. Nour. 2 April 2008.Characterization of ST80 Panton-Valentine leukocidin-positive community-acquired methicillin-resistant Staphylococcus aureus clone in Tunisia. Diagn.Microbiol. Infect. Dis. [Epub ahead of print.]
4. Boquet, P., and E. Lemichez. 2003. Bacterial virulence factors targeting RhoGTPases: parasitism or symbiosis? Trends Cell Biol. 13:238–246.
5. Boyer, L., A. Doye, M. Rolando, G. Flatau, P. Munro, P. Gounon, R. Clem-ent, C. Pulcini, M. R. Popoff, A. Mettouchi, L. Landraud, O. Dussurget, andE. Lemichez. 2006. Induction of transient macroapertures in endothelial cellsthrough RhoA inhibition by Staphylococcus aureus factors. J. Cell Biol.173:809–819.
6. Carman, C. V., P. T. Sage, T. E. Sciuto, M. A. de la Fuente, R. S. Geha, H. D.Ochs, H. F. Dvorak, A. M. Dvorak, and T. A. Springer. 2007. Transcellulardiapedesis is initiated by invasive podosomes. Immunity 26:784–797.
7. Caron, E., and A. Hall. 1998. Identification of two distinct mechanisms ofphagocytosis controlled by different Rho GTPases. Science 282:1717–1721.
8. Chardin, P., P. Boquet, P. Madaule, M. R. Popoff, E. J. Rubin, and D. M.Gill. 1989. The mammalian G protein RhoC is ADP-ribosylated by Clostrid-ium botulinum exoenzyme C3 and affects actin microfilaments in Vero cells.EMBO J. 8:1087–1092.
9. Clarke, S. R., and S. J. Foster. 2006. Surface adhesins of Staphylococcusaureus. Adv. Microb. Physiol. 51:187–224.
10. Czech, A., T. Yamaguchi, L. Bader, S. Linder, K. Kaminski, M. Sugai, andM. Aepfelbacher. 2001. Prevalence of Rho-inactivating epidermal cell dif-ferentiation inhibitor toxins in clinical Staphylococcus aureus isolates. J. In-fect. Dis. 184:785–788.
12. Foster, T. J. 2005. Immune evasion by staphylococci. Nat. Rev. Microbiol.3:948–958.
13. Francis, K. P., D. Joh, C. Bellinger-Kawahara, M. J. Hawkinson, T. F.Purchio, and P. R. Contag. 2000. Monitoring bioluminescent Staphylococcusaureus infections in living mice using a novel luxABCDE construct. Infect.Immun. 68:3594–3600.
14. Francis, K. P., J. Yu, C. Bellinger-Kawahara, D. Joh, M. J. Hawkinson, G.Xiao, T. F. Purchio, M. G. Caparon, M. Lipsitch, and P. R. Contag. 2001.Visualizing pneumococcal infections in the lungs of live mice using biolumi-nescent Streptococcus pneumoniae transformed with a novel gram-positivelux transposon. Infect. Immun. 69:3350–3358.
15. Franke, G. C., A. Bockenholt, M. Sugai, H. Rohde, and M. Aepfelbacher.2010. Epidemiology, variable genetic organization and regulation of theEDIN-B toxin in Staphylococcus aureus from bacteraemic patients. Micro-biology 156(Pt 3):860–872.
16. Fujihara, H., L. A. Walker, M. C. Gong, E. Lemichez, P. Boquet, A. V.Somlyo, and A. P. Somlyo. 1997. Inhibition of RhoA translocation andcalcium sensitization by in vivo ADP-ribosylation with the chimeric toxinDC3B. Mol. Biol. Cell 8:2437–2447.
17. Genth, H., M. Schmidt, R. Gerhard, K. Aktories, and I. Just. 2003. Activa-tion of phospholipase D1 by ADP-ribosylated RhoA. Biochem. Biophys.Res. Commun. 302:127–132.
18. Jaffe, A. B., and A. Hall. 2005. Rho GTPases: biochemistry and biology.Annu. Rev. Cell Dev. Biol. 21:247–269.
19. Kadurugamuwa, J. L., L. Sin, E. Albert, J. Yu, K. Francis, M. DeBoer, M.Rubin, C. Bellinger-Kawahara, T. R. J. Parr, Jr., and P. R. Contag. 2003.Direct continuous method for monitoring biofilm infection in a mousemodel. Infect. Immun. 71:882–890.
20. Laschke, M. W., S. Kerdudou, M. Herrmann, and M. D. Menger. 2005.Intravital fluorescence microscopy: a novel tool for the study of the interac-tion of Staphylococcus aureus with the microvascular endothelium in vivo.J. Infect. Dis. 191:435–443.
21. Lemichez, E., M. Lecuit, X. Nassif, and S. Bourdoulous. 2010. Breaking thewall: targeting of the endothelium by pathogenic bacteria. Nat. Rev. Micro-biol. 8:93–104.
22. Lowy, F. D. 1998. Staphylococcus aureus infections. N. Engl. J. Med. 339:520–532.
23. Millan, J., R. J. Cain, N. Reglero-Real, C. Bigarella, B. Marcos-Ramiro, L.Fernandez-Martin, I. Correas, and A. J. Ridley. 2010. Adherens junctionsconnect stress fibers between adjacent endothelial cells. BMC Biol. 8:11.
24. Molinari, G., M. Rohde, C. Wilde, I. Just, K. Aktories, and G. S. Chhatwal.2006. Localization of the C3-Like ADP-ribosyltransferase from Staphylococ-cus aureus during bacterial invasion of mammalian cells. Infect. Immun.74:3673–3677.
25. Nusrat, A., M. Giry, J. R. Turner, S. P. Colgan, C. A. Parkos, D. Carnes, E.Lemichez, P. Boquet, and J. L. Madara. 1995. Rho protein regulates tightjunctions and perijunctional actin organization in polarized epithelia. Proc.Natl. Acad. Sci. U. S. A. 92:10629–10633.
26. O’Neill, A. J., A. R. Larsen, R. Skov, A. S. Henriksen, and I. Chopra. 2007.Characterization of the epidemic European fusidic acid-resistant impetigoclone of Staphylococcus aureus. J. Clin. Microbiol. 45:1505–1510.
27. Pacaud, P., V. Sauzeau, and G. Loirand. 2005. Rho proteins and vasculardiseases. Arch. Mal. Coeur Vaiss. 98:249–254.
28. Ridley, A. J., and A. Hall. 1992. The small GTP-binding protein rho regulatesthe assembly of focal adhesions and actin stress fibers in response to growthfactors. Cell 70:389–399.
29. Schroder, A., B. Schroder, B. Roppenser, S. Linder, B. Sinha, R. Fassler, andM. Aepfelbacher. 2006. Staphylococcus aureus fibronectin binding protein-Ainduces motile attachment sites and complex actin remodeling in livingendothelial cells. Mol. Biol. Cell 17:5198–5210.
30. Schroder, A., R. Kland, A. Peschel, C. von Eiff, and M. Aepfelbacher. 2006.Live cell imaging of phagosome maturation in Staphylococcus aureus infectedhuman endothelial cells: small colony variants are able to survive in lyso-somes. Med. Microbiol. Immunol. 195:185–194.
31. Sekine, A., M. Fujiwara, and S. Narumiya. 1989. Asparagine residue in theRho gene product is the modification site for botulinum ADP-ribosyltrans-ferase. J. Biol. Chem. 264:8602–8605.
32. Sinha, B., P. P. Francois, O. Nusse, M. Foti, O. M. Hartford, P. Vaudaux,T. J. Foster, D. P. Lew, M. Herrmann, and K. H. Krause. 1999. Fibronectin-binding protein acts as Staphylococcus aureus invasin via fibronectin bridgingto integrin alpha5beta1. Cell. Microbiol. 1:101–117.
33. Sugai, M., T. Enomoto, K. Hashimoto, K. Matsumoto, Y. Matsuo, H. Ohgai,Y. M. Hong, S. Inoue, K. Yoshikawa, and H. Suginaka. 1990. A novelepidermal cell differentiation inhibitor (EDIN): purification and character-ization from Staphylococcus aureus. Biochem. Biophys. Res. Commun. 173:92–98.
34. Wertheim, H. F., D. C. Melles, M. C. Vos, W. van Leeuwen, A. van Belkum,H. A. Verbrugh, and J. L. Nouwen. 2005. The role of nasal carriage inStaphylococcus aureus infections. Lancet Infect. Dis. 5:751–762.
35. Wilde, C., M. Vogelsgesang, and K. Aktories. 2003. Rho-specific Bacilluscereus ADP-ribosyltransferase C3cer cloning and characterization. Biochem-istry 42:9694–9702.
36. Wojciak-Stothard, B., and A. J. Ridley. 2003. Rho GTPases and the regula-tion of endothelial permeability. Vascul. Pharmacol. 39:187–199.
37. Yamaguchi, T., T. Hayashi, H. Takami, M. Ohnishi, T. Murata, K. Na-kayama, K. Asakawa, M. Ohara, H. Komatsuzawa, and M. Sugai. 2001.Complete nucleotide sequence of a Staphylococcus aureus exfoliative toxin Bplasmid and identification of a novel ADP-ribosyltransferase, EDIN-C. In-fect. Immun. 69:7760–7771.
38. Yamaguchi, T., K. Nishifuji, M. Sasaki, Y. Fudaba, M. Aepfelbacher, T.Takata, M. Ohara, H. Komatsuzawa, M. Amagai, and M. Sugai. 2002.Identification of the Staphylococcus aureus etd pathogenicity island whichencodes a novel exfoliative toxin, ETD, and EDIN-B. Infect. Immun. 70:5835–5845.
39. Yamasaki, O., A. Tristan, T. Yamaguchi, M. Sugai, G. Lina, M. Bes, F.Vandenesch, and J. Etienne. 2006. Distribution of the exfoliative toxin Dgene in clinical Staphylococcus aureus isolates in France. Clin. Microbiol.Infect. 12:585–588.
40. Yokomori, H., K. Yoshimura, S. Funakoshi, T. Nagai, K. Fujimaki, M.Nomura, H. Ishii, and M. Oda. 2004. Rho modulates hepatic sinusoidalendothelial fenestrae via regulation of the actin cytoskeleton in rat endothe-lial cells. Lab. Invest. 84:857–864.
