Activation of EGFR on monocytes is requiredfor human cytomegalovirus entry and mediatescellular motilityGary Chana, Maciej T. Nogalskia, and Andrew D. Yurochkoa,b,1

aDepartment of Microbiology and Immunology, Center for Molecular and Tumor Virology, and bFeist–Weiller Cancer Center, Louisiana State UniversityHealth Sciences Center, Shreveport, LA 71130-3932

Edited by Thomas E. Shenk, Princeton University, Princeton, NJ, and approved October 16, 2009 (received for review August 10, 2009)

Human cytomegalovirus (HCMV) rapidly induces a mobile and func-tionally unique proinflammatory monocyte following infection that isproposed to mediate viral spread. The cellular pathways used byHCMV to initiate these biological changes remain unknown. Here, wedocument the expression of the epidermal growth factor receptor(EGFR) on the surface of human peripheral blood monocytes but noton other blood leukocyte populations. Inhibition of EGFR signalingabrogated viral entry into monocytes, indicating that EGFR can serveas a cellular tropism receptor. Moreover, HCMV-activated EGFR wasrequired for the induction of monocyte motility and transendothelialmigration, two biological events required for monocyte extravasa-tion into peripheral tissue, and thus viral spread. Transcriptomeanalysis revealed that HCMV-mediated EGFR signaling up-regulatedneural Wiskott–Aldrich syndrome protein (N-WASP), an actin nucle-ator whose expression and function are normally limited in leuko-cytes. Knockdown of N-WASP expression blocked HCMV-induced butnot phorbol 12-myristate 13-acetate (PMA)-induced monocyte motil-ity, suggesting that a switch to and/or the distinct use of a new actinnucleator controlling motility occurs during HCMV infection of mono-cytes. Together, these data provide evidence that EGFR plays anessential role in the immunopathobiology of HCMV by mediatingviral entry into monocytes and stimulating the aberrant biologicalactivity that promotes hematogenous dissemination.

The wide range of pathological complications associated withhuman cytomegalovirus (HCMV) infection is a direct conse-

quence of viral spread to peripheral organ sites and the broadcellular tropism of the virus (1). Monocytes are primary in vivotargets for HCMV and are believed to be responsible for hema-togenous dissemination of HCMV to multiple organ systems (2).We previously showed that HCMV infection of monocytes polar-ized the infected cell toward a distinct proinflammatory phenotypethat possessed the distinct biological changes necessary to promoteviral spread (3–5). Specifically, HCMV infection induced polariza-tion of infected monocytes toward an M1 proinflammatory cell typethat simultaneously exhibited characteristics associated with an M2antiinflammatory macrophage (6). The infected monocytes alsodisplayed a high level of chemokinesis when compared with mono-cytes activated by LPS or phorbol 12-myristate 13-acetate (PMA)(4, 5). Moreover, an accelerated rate of differentiation fromshort-lived monocytes (nonpermissive for HCMV replication) intolong-lived macrophages (permissive for HCMV replication) wasobserved following infection with HCMV (4). Based on our studies,we suggest that during primary infection, newly infected peripheralmonocytes acquire a motile phenotype that promotes exit of theinfected cell from the circulating blood into multiple organ tissuedespite the absence of a chemotactic gradient. Once in the sur-rounding tissue, differentiation into HCMV replication-competentmacrophages occurs, resulting in viral spread and lifelong persis-tence in the organs that serve as portals of viral exit.

The biological changes induced in HCMV-infected monocytesoccurred within a time frame in which no new viral gene expressionwas observed (4), suggesting that receptor–ligand interactionstriggered the rapid activation of infected monocytes. Although viral

glycoproteins have been documented to stimulate proinflammatorycellular signaling pathways in infected monocytes (4, 7), the host cellreceptors used by HCMV to modulate the cellular signaling net-work, and thus the unique functional changes in monocytes, remainunknown. The major HCMV glycoprotein B has been shown tobind to and directly activate epidermal growth factor receptor(EGFR), leading to the induction of PI(3)K activity in model celltypes (8). In support, rapid phosphorylation of EGFR was dem-onstrated in endothelial cells (ECs) (9) and trophoblasts (10)following infection with HCMV. However, unlike the transientactivation observed in fibroblasts (8) and trophoblasts (10), HCMVinfection stimulated chronic activation of EGFR in ECs (9),suggesting that distinct signaling profiles originating from the samecellular receptor can occur in different cell types. EGFR expressionwas required for viral entry/infection of breast cancer cells andtrophoblasts (8, 11). However, the role that EGFR plays duringviral infection remains controversial, because it has been reportedthat EGFR does not mediate entry into fibroblasts (12) and othershave shown that PDGF receptor (PDGFR)-� rather than EGFRwas required for efficient viral entry (13). Deciphering whichcellular receptor is used during viral entry is critical to unravelinghow HCMV infection modulates monocyte biology and conse-quently dictates viral spread and disease.

The presence of EGFR on the surface of human peripheral bloodmonocytes is unclear. Examination of mixed populations of periph-eral blood mononuclear cells (PBMCs) did not detect EGFRexpression (14). However, functional EGFR was detected on somemonocytic leukemic cell lines and macrophage subpopulations (15).Here, we report the expression and functionality of EGFR onhuman peripheral blood monocytes. Rapid activation of the EGFRsignaling pathway following HCMV infection was necessary forviral entry into monocytes and viral modulation of monocytemotility. Mechanistically, viral activation of EGFR on monocytesstimulates cell movement via the specific up-regulation of a highlyactive actin nucleator, neural Wiskott-Aldrich syndrome protein(N-WASP), that is normally expressed at near-undetectable levelsin leukocytes (16). Because N-WASP possesses a significantlygreater actin nucleating potential than WASP (17), the actinnucleator normally thought to control actin growth in leukocytes(18), our data suggest that HCMV has evolved a strategy involvinga switch in key actin nucleators following infection to generate

Author contributions: G.C., M.T.N., and A.D.Y. designed research; G.C. and M.T.N. per-formed research; G.C., M.T.N., and A.D.Y. analyzed data; and G.C. and A.D.Y. wrote thepaper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Data deposition: The data reported in this paper have been deposited in the GeneExpression Omnibus (GEO) database, www.ncbi.nlm.nih.gov/geo (accession no. GSE17948).

1To whom correspondence should be addressed at: Department of Microbiology andImmunology, Louisiana State University Health Sciences Center, 1501 Kings Highway,Shreveport, LA, 71130-3932. E-mail: ayuroc@lsuhsc.edu.

This article contains supporting information online at www.pnas.org/cgi/content/full/0908787106/DCSupplemental.

www.pnas.org�cgi�doi�10.1073�pnas.0908787106 PNAS � December 29, 2009 � vol. 106 � no. 52 � 22369–22374

IMM

UN

OLO

GY

monocytes exhibiting high levels of chemokinesis. Overall, thesefindings indicate that EGFR plays a dual role in the immunopatho-biology of HCMV by mediating viral entry into monocytes and theinduction of a unique proinflammatory motile phenotype, thuspromoting viral spread into distinct organ sites.

ResultsMonocytes Express a Bona Fide Functional EGFR That Is Rapidly Phos-phorylated upon HCMV Infection. The acquisition of a motile mono-cyte phenotype within 1 h postinfection (hpi) suggests that thedirect activation of a cellular receptor during viral binding/entrymediates the PI(3)K-dependent induction of infected monocytemotility (3). EGFR and PDGFR-� have both been shown tostimulate the PI(3)K signaling pathway and to be engaged duringHCMV entry into some cell types (8, 13). PDGFR-� is notexpressed, however, in cells of the myeloid lineage (19); we con-firmed this lack of expression on human primary monocytes (Fig.S1). Similarly, it has been suggested that EGFR is not a majorHCMV entry receptor, because monocytes are reported to lack thisreceptor (12, 20), although some myeloid leukemic cells expressEGFR (15). To date, EGFR expression on primary monocytesremains unclear, because previous studies were performed onPBMCs (14), in which the monocyte subpopulation constitutes only5% of the total PBMC population (21). Consequently, we nextfocused on determining whether EGFR could be responsible forHCMV-induced PI(3)K activity in infected monocytes. We testedEGFR expression on different blood leukocyte populations by flowcytometry. We failed to detect EGFR expression on T cells (CD3�)and B cells (CD20�); however, EGFR was expressed on monocytes(CD14�), albeit at lower levels than in the breast cancer cell lineMDA-MB-468 (Fig. 1A).

Next, we examined whether EGFR engagement by EGF, HCMVTowne/E, or HCMV TB40/E resulted in receptor activation. Con-sistent with receptor–ligand kinetics, EGFR was rapidly activatedwithin 10 min of treatment with EGF or infection with eitherHCMV strain (Fig. 1B). We found concomitant phosphorylation ofthe downstream PI(3)K target, Akt, within 10 min of treatment withEGF or infection with HCMV. Treatment with a potent EGFR-specific kinase inhibitor, AG1478, or a neutralizing anti-EGFRantibody before culture with EGF or infection with HCMV dimin-ished the rapid phosphorylation of EGFR and downstream targetsPI(3)K and Akt (Fig. 1C). This slight induction of PI(3)K and Aktin the presence of either EGFR inhibitor following infectionsuggests that other HCMV entry receptors, such as integrins (22,23), may also be involved in the activation of the PI(3)K pathway(24). Nonetheless, our results indicate that peripheral blood mono-cytes express bona fide functional cell surface EGFR.

HCMV-Induced EGFR Engagement and Activation Are Required for Effi-cient Viral Entry into Monocytes. With the controversy surroundingthe role that EGFR plays in viral entry (8, 12) and the lack of anidentifiable cell-specific receptor on blood leukocytes, we nextassessed whether the expression of EGFR on monocytes was adeterminant factor required for HCMV binding/entry to mono-cytes. TB40/E, a recombinant HCMV strain with GFP fused to theC-terminal end of the capsid-bound tegument protein pUL32 (25,26), was used to infect PBMCs, and flow cytometric analyses wereperformed to detect the presence of bound GFP-labeled viralparticles associated with CD3� (T cells), CD20� (B cells), andCD14� (monocytes) leukocyte subpopulations (Fig. 2 A–C). Wefound high levels of GFP-labeled virus associated with CD14� cells(Fig. 2B) but not with CD3� or CD20� cells (Fig. 2 A and C).

Is expression of EGFR responsible for the selective binding ofvirus to different populations of blood leukocytes, or does it play adifferent role in viral tropism, such as in the regulation of viralfusion? The presence of soluble heparin, a known inhibitor ofHCMV attachment (20), significantly decreased the presence ofbound GFP-tagged TB40/E with monocytes, as determined by flow

cytometric analysis, indicating that viral binding was blocked (Fig.2D). However, the presence of a blocking EGFR antibody did notsignificantly affect the levels of GFP-tagged virus associated withmonocytes, indicating that EGFR expression alone does not dictateviral attachment to monocytes.

Because flow cytometric analysis cannot differentiate betweensurface-bound and internalized virus, we next used immunofluo-rescent microscopy to examine whether GFP-labeled HCMV par-ticles bound to the monocyte cell surface required EGFR activityto be internalized into the host cell cytoplasm. Infection of mono-cytes with TB40/E carried out at 4 °C showed viral particleslocalized at the cell surface (Fig. 2E). Following a temperature shiftto 37 °C to initiate internalization, we found TB40E in the cyto-

Fig. 1. HCMV and EGF activate EGFR found on primary human peripheralblood monocytes. (A) PBMCs were stained with anti-CD3-PE-Cy5 (T cells),anti-CD14-APC-Cy7 (monocytes), and anti-CD20-PE (B cells) antibodies andwere then costained with either an anti-EGFR-GFP antibody (green) or a GFPisotype-matched control antibody (blue). The red line indicates unlabeledcells. MDA-MB-468 breast cancer cells were used as a positive control cell line.Examination of EGFR expression was performed by flow cytometry. (B) Iso-lated monocytes were treated with EGF or infected with HCMV (Towne/E orTB40/E) for 0, 10, 30, 45, and 60 min. Phosphorylated EGFR (pEGFR), total EGFR,phosphorylated Akt (pAkt), and total Akt were detected by immunoblotting.(C) Monocytes were treated with an anti-EGFR antibody or AG1478 beforetreatment with EGF or infection with Towne/E or TB40E. Total lysate washarvested, and pEGFR, total EGFR, phosphorylated PI(3)K [pPI(3)K], totalPI(3)K, pAkt, and total Akt were determined by immunoblotting. Membraneswere reprobed with antibody against �-actin (B and C).

22370 � www.pnas.org�cgi�doi�10.1073�pnas.0908787106 Chan et al.

plasm of the infected cell. Pretreatment of monocytes with AG1478or a neutralizing EGFR antibody resulted in the inhibition ofinternalization of the viral capsid and the clustering of viral particlesat the cell surface, which was also observed in EGFR-negativecytotrophoblasts (11). To assess the contribution of EGFR in theprocess of viral entry further, cells were examined for the presenceof internalized HCMV genomic DNA. We detected low levels ofviral DNA in samples infected at 4 °C following removal of surface-bound viral particles by proteinase K (PK) treatment, whereas highlevels of viral DNA were detected in samples with the temperatureshifted to 37 °C before the addition of PK, demonstrating that viralentry had occurred (Fig. 2F). Consistent with the immunofluores-cent analysis, PK treatment of HCMV-infected monocytes dem-onstrated a significant reduction in HCMV entry in the absence ofEGFR activation, although the lack of EGFR signaling did notcompletely eliminate viral entry. LY294002, an inhibitor of PI(3)Kactivity, did not block HCMV entry, suggesting that althoughPI(3)K is directly downstream of EGFR and is rapidly activatedfollowing infection, a divergent signaling pathway initiated from theEGFR kinase must mediate viral entry. All samples from infectedmonocytes not treated with PK exhibited similar amounts ofdetectable HCMV genomic DNA, indicating that viral binding wasnot affected by the different pretreatments (Fig. S2). qRT-PCRanalysis confirmed EGFR-mediated viral entry (Fig. 2G) andrevealed a 41% and 59% decrease in viral entry in the presence ofneutralizing-EGFR antibody and AG1478, respectively, which isconsistent with the 63% reduction in infectivity observed in tro-phoblasts treated with an anti-EGFR antibody before HCMVinfection (11). Other growth factor-like receptor tyrosine kinases,including PDGFR-�, VEGF receptor 1 (VEGFR1), and VEGFR2,are not involved in mediating viral entry into monocytes (Fig. S3).Taken together, these results suggest that EGFR-mediated signal-ing is initiated following viral binding to the receptor and that thissignaling dictates monocyte tropism by directing efficient internal-ization of the viral particle.

HCMV Promotes Monocyte Motility and Transendothelial Migration viaEGFR. Because our early data showed that viral binding triggersfunctional changes in monocytes (4, 7), we hypothesized thatHCMV engagement of EGFR during viral binding/entry wasresponsible for the HCMV-induced polarization of infected mono-cytes. Infection of monocytes by HCMV induced lamellipodiumand tail formation, which are hallmarks of cell motility (Fig. 3A).However, monocytes pretreated with EGFR inhibitors beforeinfection exhibited morphological characteristics similar to those ofquiescent mock-infected cells. Quantitative evaluation of motilityby phagokinetic motility track assays revealed that treatment ofmonocytes with AG1478 or an anti-EGFR antibody before infec-tion abrogated virus-induced chemokinesis (Fig. 3B). Pretreatmentwith the DMSO solvent control (3, 5) or the IgG isotype-matchedcontrol did not affect HCMV-induced motility. Similar results wereobserved when EGF was used as the activating ligand, furthersupporting the role that EGFR plays in directly modulating mono-cyte motility (Fig. 3B). In agreement with the viral entry assays, thepresence of neutralizing PDGFR-�, VEGFR1, or VEGFR2 anti-bodies during infection was not able to block HCMV-inducedmonocyte chemokinesis (Fig. S4). Overall, our data suggest thatEGFR plays two essential roles during HCMV infection of mono-cytes. First, EGFR activity is required for efficient viral entry,although other receptors may partially compensate for the lack ofEGFR signaling, as indicated by only a 50% inhibition of viral entryin the presence of EGFR inhibitors. Second, EGFR activationduring viral entry mediates HCMV-induced motility, becausemonocyte chemokinesis was inhibited nearly 100% when EGFRsignaling was abrogated.

Increased cell motility stimulates diapedesis (transendothelialmigration) by promoting monocytes to push between tight junctionsof adjacent ECs (27). At 24 hpi of EGF treatment, we found �50%

Fig. 2. EGFR acts as a tropism receptor directing HCMV internalization. PBMCswere mock-infected (red) or infected with a GFP tegument-tagged TB40/E virus(blue) and then labeled with anti-CD3-PE-Cy5 (T cells) (A), anti-CD14-APC-Cy7(monocytes) (B), or anti-CD20-PE (B cells) (C) antibody and examined by flowcytometry to detect GFP-positive viral particles. (D) Purified CD14� monocyteswere mock infected (red), TB40/E infected (blue), or TB40/E infected in thepresence of anti-EGFR antibody (orange) or soluble heparin (green) and thenanalyzed by flow cytometry. (E) Monocytes were pretreated with an anti-EGFRantibody or AG1478, cooled to 4 °C, and infected with TB40/E. Cells were thenimmediately fixed or temperature-shifted to 37 °C for 1 h and then fixed. Redfluorescence represents the labeledEGFRstaining,greenfluorescence representsTB40/E-GFP staining, blue fluorescence represents DAPI-stained nuclei, and thewhite dashed line represents the outer monocyte cell membrane. Images arerepresentative of three independent experiments from different donors. (F andG) Monocytes were pretreated with an anti-EGFR antibody, an anti-IgG controlantibody, AG1478, or LY294002; were cooled to 4 °C; and were infected withTowne/E. Cells were then treated with PK or incubated at 37 °C for 1 h before PKtreatment. The presence of genomic HCMV DNA was determined by RT-PCR (F)and qRT-PCR (G) analysis using HCMV genomic immediate early-specific primers.

Chan et al. PNAS � December 29, 2009 � vol. 106 � no. 52 � 22371

IMM

UN

OLO

GY

of monocytes undergoing diapedesis vs. only 20% of mock-infectedcells (Fig. 3C). The 2.5-fold induction in transendothelial migrationby HCMV or EGF was abrogated by inhibition of EGFR signaling.Pretreatment with the DMSO solvent control (3, 5) or the IgGisotype-matched control antibody did not affect HCMV-inducedmonocyte transendothelial migration. Together, these data showthat EGFR possesses biological functions in human peripheralblood monocytes and are consistent with our hypothesis that EGFRengagement by the virus triggers changes in monocytes to promotehematogenous dissemination.

HCMV Up-Regulates N-WASP Expression in an EGFR-Dependent Mannerto Promote Monocyte Motility. To identify the EGFR-dependentmechanism by which HCMV stimulates monocyte chemokinesis,

we examined the HCMV-infected EGFR-dependent monocytetranscriptome at 24 hpi. Because HCMV-induced monocyte mo-tility was abrogated by pretreatment with either the anti-EGFRantibody (inhibition of the extracellular activation domain ofEGFR) or AG1478 (inhibition of the intracellular kinase activity),we focused on HCMV-induced genes that were dependent ondirect binding to and signaling through EGFR. Ontology analysesof cellular genes known to be involved in cell motility identifiedN-WASP as a potential mediator of the changes observed inHCMV-infected monocytes (Table S1). N-WASP is an actin nu-cleator that mediates endocytosis, lamellipodia formation, andmotility in EGF-stimulated carcinoma cells (28–30); however,because N-WASP is generally expressed at near-undetectable levelsin monocytes (16), we wanted to confirm the induction of N-WASPprotein expression in monocytes following HCMV infection (Fig.4A). We also found that inhibition of the EGFR downstream targetPI(3)K blocked HCMV-induced expression of N-WASP.

To dissect the biological consequence(s) of N-WASP up-regulation during HCMV infection, siRNA knockdown was used.We observed a 90% reduction in N-WASP protein levels followingdensitometric analyses in N-WASP siRNA-transfected monocytesbut not in control siRNA-transfected monocytes when comparedwith mock-transfected cells by 24 h posttransfection (Fig. 4B). Thedown-regulation continued for up to 72 h and had no effect onmonocyte viability. Because N-WASP deficiency impairs EGFR-mediated endocytosis (28), we first tested whether HCMV entryinto monocytes was reduced in N-WASP-deficient monocytes.Similar levels of internalized viral genomes in monocytes expressinglow and high levels of N-WASP were observed (Fig. 4C), suggestingthat N-WASP was not required for efficient viral entry.

Because EGF stimulation can target N-WASP to the leadingmembrane edge of a motile cell to mediate movement (31), weexamined whether N-WASP plays a functional role in the ‘‘hyper’’-motility observed in HCMV-infected monocytes. Examination ofcellular morphology revealed that, unlike HCMV-infected WTmonocytes, N-WASP-deficient monocytes lacked distinct tail andlamellipodia formation (Fig. 4D). Quantitative phagokinetic mo-tility track assays demonstrated a 6-fold increase in monocytechemokinesis following HCMV infection, whereas only a 4-fold and3-fold induction in motility was observed following PMA and EGFtreatment, respectively (Fig. 4E). An 80% reduction in motility wasseen in N-WASP-deficient HCMV-infected monocytes. Similarly,N-WASP-deficient monocytes derived from a different N-WASP-specific siRNA exhibited analogous decreases in HCMV-inducedchemokinesis (Fig. S5), demonstrating that the decrease in the viralinduction of monocyte motility was not likely attributable tooff-target effects. Because 100% inhibition of EGF-induced mono-cyte motility was observed in the absence of N-WASP activity, ourdata suggest that low-level EGFR-independent cell motility mayalso be generated in HCMV-infected monocytes. The utilization ofN-WASP to induce hypermotility in cells appeared to be specific toHCMV infection, because the less proficient PMA induction ofmonocyte motility was not affected by the decrease in N-WASPexpression. Overall, these data suggest that HCMV specificallystimulates the rapid and aberrant up-regulation of N-WASP activityin monocytes (dependent on the EGFR/PI(3)K signaling cascade)to induce a distinct multidirectional motility.

DiscussionHematogenous dissemination of HCMV to multiple host organsites is critical to viral persistence within the infected host followingprimary infection. Viral spread occurs in a cell-dependent manner,with monocytes serving as direct mediators of viral spread toperipheral tissue (2, 32, 33). In our investigation of the mechanismfor how infected monocytes promote the simultaneous spread ofHCMV to biologically distinct tissue, we reported that monocytesacquire a unique chemokinetic phenotype via receptor–ligandinteractions during viral binding/entry (4, 7). In this study, we

Fig. 3. HCMV stimulates monocyte motility and transendothelial migrationin an EGFR-dependent manner. Monocytes were pretreated with an anti-EGFRantibody, an IgG isotype-matched control antibody, or AG1478. (A) Mono-cytes were infected with Towne/E, and photomicrographs were captured at 6hpi. Images are representative of three independent experiments from dif-ferent donors. (B) Cells were plated on colloidal gold-coated coverslips andinfected with Towne/E or treated with EGF for 24 h. The average area ofcolloidal gold cleared per monocyte was determined. (C) Monocytes weremock infected, Towne/E infected, or EGF treated for 6 h nonadherently, andthey were subsequently labeled with CellTracker Green. Labeled monocyteswere added to cell-culture inserts containing confluent monolayers of ECs andincubated for 24 h. Following incubation, the percentage of monocytes thatunderwent diapedesis was determined by fluorescence microscopy. (B and C)Results are from three independent experiments from different donors. Sta-tistical significance between experimental means (P value) was determinedusing the Student’s t test. *, P � 0.05.

22372 � www.pnas.org�cgi�doi�10.1073�pnas.0908787106 Chan et al.

demonstrate the expression of EGFR on the surface of primaryperipheral blood monocytes and that HCMV engagement ofEGFR results in rapid autophosphorylation and subsequent phos-phorylation of downstream targets. The activation of the EGFRsignaling pathway was required for efficient viral entry and mono-cyte chemokinesis, thus providing a link between viral entry and thepathogenic motility required for viral spread during primary infec-tion despite the absence of a chemotactic gradient.

The relation between EGFR expression and HCMV infection ofthe monocyte subpopulation in PBMCs indicated that EGFR couldbe an integral player determining monocyte tropism for HCMV.Because the presence of neutralizing anti-EGFR antibodies signif-

icantly reduced virus internalization into monocytes but did notaffect attachment of HCMV, we suggest that EGFR is a viraltropism receptor that functions at the stage of internalization.Integrins and PDGFR-� have also been reported to be required forefficient viral entry (13, 22, 23), but because of ubiquitous integrinexpression on PBMCs and the lack of PDGFR-� expression onmonocytes and macrophages (in this study and ref. 19), it is unlikelythat these receptors direct HCMV infection toward the monocytesubpopulation. EGFR has been shown to act as a viral tropismreceptor for targeting HCMV entry into biologically distinct pop-ulations of trophoblast cells (11). Our data now provide evidencethat EGFR expression and the ensuing downstream signalingfollowing viral engagement on monocytes provide cell type speci-ficity for viral internalization.

Several studies have reported the rapid activation of EGFR tooccur in various cell lines following HCMV infection (8–10) andthat the activation of EGFR signaling was required for efficientinfection (8, 11). We similarly found that EGFR activity wasnecessary for HCMV entry into primary monocytes, although incontrast to model cell types (23), activation of the downstreamtarget PI(3)K was not required for HCMV entry. This differencehighlights how unique pathways originating from the same cellularreceptor in biologically distinct cell types are used by HCMV toensure self-survival in multiple cell types. To add to the complexityof HCMV entry, other studies report that EGFR expression is notrequired for viral entry (12, 34). The reasons for these conflictingresults remain unresolved. In contrast to endotheliotropic/clinicalHCMV strains, such as Towne/E and TB40/E, laboratory/fibroblast-adapted strains, such as AD169, Towne, and TB40/F,have reduced infectivity of endothelial/epithelial cells and mono-cytes because of losses or mutations in the ULB� region of the viralgenome (35–37). Expression of distinct viral glycoproteins from theULB� region allows viral infection of endothelial/epithelial cells butnot of fibroblasts, indicating the existence of cell type-specificreceptors for different strains of HCMV (38). Although more workis needed to address whether strain variation can account for therequirement of EGFR signaling during entry, our data demonstratethat endotheliotropic strains Towne/E and TB40/E signal and enterthrough an EGFR-dependent mechanism on monocytes.

Analysis of the EGFR-dependent HCMV-infected monocytetranscriptome revealed a complex regulation of monocyte geneexpression that originates from engagement of and signalingthrough EGFR. More genes were regulated by the intracellularsignaling of EGFR when compared with genes regulated by theextracellular signaling of EGFR (Fig. S6; R � 0.78 vs. 0.93), whichis likely attributable to the cross-talk phosphorylation of the cyto-plasmic domain of EGFR by integrins activated during HCMVentry (22, 23, 39). We suggest that HCMV initiates a multireceptoractivation process during viral entry that depends on the expressionof cell-specific HCMV receptors on different cell types.

Our global transcriptional and functional analysis identified theactin nucleator N-WASP as a potential mediator of EGFR-dependent HCMV-induced monocyte chemokinesis. Although N-WASP is expressed in many different cell types, homeostaticexpression of N-WASP is low in hematopoietic cells (16), in whichthe related WASP is the dominant actin nucleator (18); thus, therole that N-WASP plays in HCMV-infected monocytes is unclear.Perhaps specific targeting of N-WASP by HCMV is based on the20 times higher actin-polymerizing potential of N-WASP whencompared with WASP activity (17). Indeed, it has been observedthat actin-based spreading of Shigella to adjacent cells is dependenton the exclusive recruitment of N-WASP (16). Because of lowN-WASP expression in hematopoietic cells, spread of Shigella wasnot observed from infected macrophages and was only restored byectopic expression of N-WASP (16). Our data indicate that HCMVhas devised a specific strategy to up-regulate the expression ofN-WASP rapidly, thus circumventing the normally low levels of thisprotein in monocytes. This requirement of N-WASP to mediate

Fig. 4. HCMV up-regulation of N-WASP occurs in an EGFR-dependent man-ner and is required for HCMV-induced monocyte motility. (A) Monocytes weretreated with an anti-EGFR antibody, AG1478, or LY294002 before infectionwith HCMV. At 24 hpi, total N-WASP expression was detected by immuno-blotting. (B) Cells were transfected with N-WASP siRNA or control siRNA, andN-WASP protein levels were analyzed by immunoblotting. (C) Monocyteswere mock-transfected (lanes 1–3), N-WASP siRNA-transfected (lane 4), orcontrol siRNA-transfected (lane 5) and incubated for 24 h. Cells were thencooled to 4 °C for 30 min and mock-infected (lane 1) or Towne/E-infected(lanes 2–5) for 90 min at 4 °C. Cells were either immediately treated with PK(lane 3) or incubated for 1 h at 37 °C (lanes 1, 2, 4, and 5) before treatment withPK. The presence of genomic HCMV DNA was determined by RT-PCR usingHCMV genomic immediate early-specific primers. RT-PCR analysis of GAPDHwas done as a loading control. (D) N-WASP siRNA, control siRNA, or untrans-fected monocytes were infected with Towne/E, and photomicrographs cap-tured at 6 hpi. Images are representative of three independent experimentsfrom different donors. (E) N-WASP siRNA, control siRNA, or untransfectedmonocytes were plated on colloidal gold-coated coverslips and infected withTowne/E (multiplicity of infection of 5), treated with PMA (10 ng/mL), ortreated with EGF (100 ng/mL). The average area of colloidal gold cleared permonocyte was determined. Results are from three independent experimentsfrom different donors. Statistical significance between experimental means (Pvalue) was determined using the Student’s t test. *, P � 0.05.

Chan et al. PNAS � December 29, 2009 � vol. 106 � no. 52 � 22373

IMM

UN

OLO

GY

HCMV-induced chemokinetic monocyte motility is in contrast tochemokine-induced chemotactic motility, which requires WASPactivity (40). Our data also indicate that PMA induction of mono-cyte motility was not affected by N-WASP deficiency in monocytes,thus providing further evidence for the select utilization of N-WASP during HCMV infection to induce hypermotility in cells.These observations provide a potential mechanism by which theN-WASP-dependent chemokinetic movement of monocytes favorsthe hematogenous dissemination of HCMV to multiple organ sitesduring primary infection.

Although the heightened motility induced following HCMVinfection is dependent on EGFR activity, our data indicate thatdivergent signaling pathways are generated from EGF- andHCMV-stimulated EGFR. The inhibition of EGFR by the pres-ence an anti-EGFR antibody or AG1478 completely abrogatedHCMV- and EGF-induced monocyte motility. However, unlikeEGF treatment, HCMV infection was able to induce motilitypartially in N-WASP-deficient monocytes, indicating that HCMVinduction of EGFR leads to cellular signals different from those ofEGF-induced EGFR activation. Indeed, other studies have alsoshown that HCMV infection stimulates a phosphorylation patternon the cytoplasmic tail of EGFR different from the phosphoryla-tion signature induced by EGF treatment (10), indicating thatHCMV acts as a unique ligand able to initiate an aberrant EGFRcellular signaling profile not observed with EGF treatment.

In summary, we have documented the expression of biologicallyfunctional EGFR on primary human peripheral blood monocytesand suggest that HCMV has evolved to use this cellular receptor to

initiate the multiple early events required for viral dissemination.EGFR acts as a tropism receptor by mediating efficient HCMVentry in a PI(3)K-independent manner to the monocyte subpopu-lation of PBMCs. In addition, the activation of EGFR induces thepathogenic chemokinetic movement of monocytes followingHCMV infection, which is dependent on PI(3)K activity and isessential for the migration of infected cells into peripheral tissue inwhich differentiation into replication-permissive macrophages canoccur. Our identification of EGFR on primary human monocytesprovides a possible mechanistic link between HCMV infection andthe development of cardiovascular diseases, such as atherosclerosis,in which pathogenesis is coupled to the migration of proinflam-matory monocytes into atherosclerotic plaques (41).

Materials and MethodsHuman Peripheral Blood Monocyte Isolation. Blood was drawn by venipunctureand centrifuged through a Ficoll Histopaque 1077 gradient (Sigma). Mono-cytes were then isolated by centrifugation through a Percoll (AmershamPharmacia) gradient and suspended in RPMI 1640 (Cellgro; Mediatech) sup-plemented with 1% human serum (Sigma).

Virus Preparation. HCMV strains Towne/E (passages 35–45) (7) and GFP-labeledTB40/E-UL32 (TB40/E) (25, 26) were cultured in human embryonic lung fibro-blasts. Virus was purified on a 0.5-M sucrose cushion, resuspended in RPMI1640 media (Cellgro), and used to infect monocytes at a multiplicity ofinfection of 5 for each experiment unless otherwise stated.

For complete details of all protocols, see SI Materials and Methods.

ACKNOWLEDGMENTS. This work was supported by a Malcolm Feist Cardio-vascular Research Fellowship and the National Institutes of Health (GrantsAI56077, HD051998, and 1-P20-RR018724).

1. Toorkey CB, Carrigan DR (1989) Immunohistochemical detection of an immediate earlyantigen of human cytomegalovirus in normal tissues. J Infect Dis 160:741–751.

2. Sinzger C, Jahn G (1996) Human cytomegalovirus cell tropism and pathogenesis.Intervirology 39:302–319.

3. Smith MS, Bentz GL, Smith PM, Bivins ER, Yurochko AD (2004) HCMV activates PI(3)Kin monocytes and promotes monocyte motility and transendothelial migration in aPI(3)K-dependent manner. J Leukocyte Biol 76:65–76.

4. Smith MS, Bentz GL, Alexander JS, Yurochko AD (2004) Human cytomegalovirusinduces monocyte differentiation and migration as a strategy for dissemination andpersistence. J Virol 78:4444–4453.

5. Smith MS, et al. (2007) Roles of phosphatidylinositol 3-kinase and NF-�B in humancytomegalovirus-mediated monocyte diapedesis and adhesion: Strategy for viral per-sistence. J Virol 81:7683–7694.

6. Chan G, Bivins-Smith ER, Smith MS, Smith PM, Yurochko AD (2008) Transcriptomeanalysis reveals human cytomegalovirus reprograms monocyte differentiation towardan M1 macrophage. J Immunol 181:698–711.

7. Yurochko AD, Huang ES (1999) Human cytomegalovirus binding to human monocytesinduces immunoregulatory gene expression. J Immunol 162:4806–4816.

8. Wang X, Huong SM, Chiu ML, Raab-Traub N, Huang ES (2003) Epidermal growth factorreceptor is a cellular receptor for human cytomegalovirus. Nature 424:456–461.

9. Bentz GL, Yurochko AD (2008) Human CMV infection of endothelial cells induces anangiogenic response through viral binding to EGF receptor and beta1 and beta3integrins. Proc Natl Acad Sci USA 105:5531–5536.

10. LaMarca HL, Nelson AB, Scandurro AB, Whitley GS, Morris CA (2006) Human cytomeg-alovirus-induced inhibition of cytotrophoblast invasion in a first trimester extravillouscytotrophoblast cell line. Placenta 27:137–147.

11. Maidji E, Genbacev O, Chang HT, Pereira L (2007) Developmental regulation of humancytomegalovirus receptors in cytotrophoblasts correlates with distinct replication sitesin the placenta. J Virol 81:4701–4712.

12. Isaacson MK, Feire AL, Compton T (2007) Epidermal growth factor receptor is notrequired for human cytomegalovirus entry or signaling. J Virol 81:6241–6247.

13. Soroceanu L, Akhavan A, Cobbs CS (2008) Platelet-derived growth factor-alpha recep-tor activation is required for human cytomegalovirus infection. Nature 455:391–395.

14. Real FX, et al. (1986) Expression of epidermal growth factor receptor in human cultured cellsandtissues:Relationshiptocell lineageandstageofdifferentiation.CancerRes46:4726–4731.

15. DreuxAC,LambDJ,ModjtahediH,FernsGA(2006)Theepidermalgrowthfactorreceptorsandtheir family of ligands: Their putative role in atherogenesis. Atherosclerosis 186:38–53.

16. Suzuki T, et al. (2002) Neural Wiskott-Aldrich syndrome protein (N-WASP) is the specificligand for Shigella VirG among the WASP family and determines the host cell typeallowing actin-based spreading. Cell Microbiol 4:223–233.

17. Yamaguchi H, et al. (2000) Two tandem verprolin homology domains are necessary fora strong activation of Arp2/3 complex-induced actin polymerization and induction ofmicrospike formation by N-WASP. Proc Natl Acad Sci USA 97:12631–12636.

18. Badolato R, et al. (1998) Monocytes from Wiskott-Aldrich patients display reducedchemotaxis and lack of cell polarization in response to monocyte chemoattractantprotein-1 and formyl-methionyl-leucyl-phenylalanine. J Immunol 161:1026–1033.

19. Inaba T, et al. (1993) Expression of platelet-derived growth factor beta receptor onhuman monocyte-derived macrophages and effects of platelet-derived growth factorBB dimer on the cellular function. J Biol Chem 268:24353–24360.

20. Compton T (2004) Receptors and immune sensors: The complex entry path of humancytomegalovirus. Trends Cell Biol 14:5–8.

21. Sipka S, et al. (2001) Decreased arachidonic acid release in peripheral blood monocytesof patients with systemic lupus erythematosus. J Rheumatol 28:2012–2017.

22. Feire AL, Koss H, Compton T (2004) Cellular integrins function as entry receptors forhuman cytomegalovirus via a highly conserved disintegrin-like domain. Proc Natl AcadSci USA 101:15470–15475.

23. Wang X, Huang DY, Huong SM, Huang ES (2005) Integrin ��3 is a coreceptor for humancytomegalovirus. Nat Med 11:515–521.

24. Velling T, Nilsson S, Stefansson A, Johansson S (2004) beta1-Integrins induce phos-phorylation of Akt on serine 473 independently of focal adhesion kinase and Src familykinases. EMBO Rep 5:901–905.

25. Sinzger C, et al. (1999) Modification of human cytomegalovirus tropism through propa-gation in vitro is associated with changes in the viral genome. J Gen Virol 80:2867–2877.

26. Bentz GL, et al. (2006) Human cytomegalovirus (HCMV) infection of endothelial cellspromotes naive monocyte extravasation and transfer of productive virus to enhancehematogenous dissemination of HCMV. J Virol 80:11539–11555.

27. Muller WA (2001) Migration of leukocytes across endothelial junctions: Some conceptsand controversies. Microcirculation 8:181–193.

28. Benesch S, et al. (2005) N-WASP deficiency impairs EGF internalization and actinassembly at clathrin-coated pits. J Cell Sci 118:3103–3115.

29. El-Sibai M, et al. (2007) Cdc42 is required for EGF-stimulated protrusion and motility inMTLn3 carcinoma cells. J Cell Sci 120:3465–3474.

30. Kurokawa K, Itoh RE, Yoshizaki H, Nakamura YO, Matsuda M (2004) Coactivation ofRac1 and Cdc42 at lamellipodia and membrane ruffles induced by epidermal growthfactor. Mol Biol Cell 15:1003–1010.

31. Svitkina T (2007) N-WASP generates a buzz at membranes on the move. Cell 128:828–830.32. Manez R, et al. (1996) Time to detection of cytomegalovirus (CMV) DNA in blood

leukocytes is a predictor for the development of CMV disease in CMV-seronegativerecipients of allografts from CMV-seropositive donors following liver transplantation.J Infect Dis 173:1072–1076.

33. Sinclair J, Sissons P (1996) Latent and persistent infections of monocytes and macro-phages. Intervirology 39:293–301.

34. Cobbs CS, et al. (2007) Human cytomegalovirus induces cellular tyrosine kinase signal-ing and promotes glioma cell invasiveness. J Neurooncol 85:271–280.

35. Dolan A, et al. (2004) Genetic content of wild-type human cytomegalovirus. J Gen Virol85:1301–1312.

36. Cha TA, et al. (1996) Human cytomegalovirus clinical isolates carry at least 19 genes notfound in laboratory strains. J Virol 70:78–83.

37. Murphy E, et al. (2003) Coding potential of laboratory and clinical strains of humancytomegalovirus. Proc Natl Acad Sci USA 100:14976–14981.

38. Ryckman BJ, Chase MC, Johnson DC (2008) HCMV gH/gL/UL128–131 interferes withvirus entry into epithelial cells: Evidence for cell type-specific receptors. Proc Natl AcadSci USA 105:14118–14123.

39. Edick MJ, Tesfay L, Lamb LE, Knudsen BS, Miranti CK (2007) Inhibition of integrin-mediated crosstalk with epidermal growth factor receptor/Erk or Src signaling path-ways in autophagic prostate epithelial cells induces caspase-independent death. MolBiol Cell 18:2481–2490.

40. Jones GE (2000) Cellular signaling in macrophage migration and chemotaxis. J Leuko-cyte Biol 68:593–602.

41. Ross R (1999) Atherosclerosis—An inflammatory disease. N Engl J Med 340:115–126.

22374 � www.pnas.org�cgi�doi�10.1073�pnas.0908787106 Chan et al.