Transcriptome Analysis Reveals Human Cytomegalovirus Reprograms Monocyte Differentiation toward an M1 Macrophage

Infection by human CMV (HCMV),³ a member of the Herpesviridae family, leads to morbidity and mortality in immunocompromised individuals, including AIDS and organ transplant patients, congenitally infected neonates, and cancer patients undergoing chemotherapy (1, 2, 3). HCMV infection causes a wide range of overt organ diseases including retinitis, gastrointestinal disease, hepatitis, and interstitial pneumonia due to the broad cellular tropism of the virus in vivo (4). Following initial primary infection of host epithelial cells by contact with HCMV-containing bodily fluids, HCMV replicates and spreads to the peripheral blood where viral dissemination to multiple organ systems occurs (4).

Peripheral blood leukocytes, specifically monocytes, are thought to be responsible for hematogenous spread of HCMV (4, 5). Monocytes are primary in vivo targets for HCMV (6), are a site of viral latency and persistence (4, 7), are the primary infiltrating cell type found in HCMV-infected organs (8, 9), and their aberrant function following HCMV infection is implicated in atherosclerosis, an inflammatory disease whose development and severity is associated with HCMV infection (10, 11). Furthermore, animal studies indicate that monocyte-associated viremia is a prerequisite for viral pathogenesis (12, 13). In accord with these in vivo observations, we previously provided in vitro evidence suggesting HCMV uses monocytes as a vehicle for spreading, infiltrating into, and persisting in host tissue (14, 15, 16, 17). Our studies have demonstrated that primary infection of peripheral blood monocytes by HCMV induces a proinflammatory state resulting in increased cell motility, firm adhesion to endothelial cells, and transendothelial migration (14, 15, 16). These functional attributes are acquired concomitant with the critical process of monocyte-to-macrophage differentiation, where short-lived monocytes (nonpermissive for viral replication) differentiate into long-lived macrophages (permissive for viral replication) (15). Although we have shown increased expression of multiple macrophage markers, including HLA-DR and CD68 (15), heterogeneity and plasticity are hallmarks of cells belonging to the mononuclear phagocyte system, and overall little is known about the phenotypic characteristics of the HCMV-activated monocyte/macrophage.

Monocyte and macrophage plasticity is apparent by the distinct morphological and functional responses to particular tissues and to the immunological microenvironment (18, 19, 20, 21). Macrophages can be functionally polarized into classically activated M1 macrophages by treatment with IFN-γ alone or in concert with LPS (22) or alternatively activated M2 macrophages by treatment with IL-4 or IL-10 (18, 23). Classically activated macrophages are characterized by an IL-12^high, IL-23^high, IL-10^low phenotype (24) and the production of toxic intermediates (reactive oxygen and nitrogen intermediates) (22) and proinflammatory cytokines (IL-1β and TNF-α) (25). M1 macrophages are potent effector cells efficient at eliminating pathogens and tumor cells. In contrast, alternatively activated macrophages exhibit an IL-12^low, IL-23^low phenotype (24), produce antiinflammatory molecules (IL-10 and IL-1 receptor antagonist (IL-1Ra)) (21, 26), and express high levels of scavenger, mannose, and galactose-type receptors (21, 23). M2 macrophages counteract inflammatory response and promote angiogenesis and tissue remodeling (21, 26, 27). M1 and M2 macrophages, however, are likely representatives of two extremes along a continuum of possible macrophage biological phenotypes.

“Classically” activated monocytes exhibit enhanced antimicrobial activities in a stimulus-dependent (particularly in response to IFN-γ) (28) but Ag-nonspecific manner through the increased expression of cell surface adhesion receptors and secretion of cytokines and chemokines (29, 30). Elevated levels of cell adhesion molecules promote adhesion to the endothelium, while the increased release of cytokines such as TNF-α and IL-1 can activate the endothelium to further promote transendothelial migration (31, 32). Similar to other pathogenic agents such as Gram-negative bacteria (via LPS), we previously showed that HCMV-activated monocytes exhibit increased expression of cytokines and chemokines, adhesion to endothelial cells, and transendothelial migration (14, 15, 33), suggesting a polarization toward an M1 macrophage phenotype. However, unlike other pathogens, HCMV gains a selective replication advantage from a strong host inflammatory response (15, 34, 35, 36, 37). Moreover, LPS- and HCMV-activated monocytes are morphologically distinct (15); thus, while it appears that monocytes activated by a different directing stimuli may share some functional traits, HCMV must stimulate a distinct M1/M2 macrophage reprogramming to meet its own specific replicative needs. Indeed, analysis of monocyte/macrophage activation by other stimuli such as CCL5 (38), LPS (38), Escherichia coli (39), and Streptococcus pyogenes (40) revealed individually unique expression profiles of macrophage markers characteristic of both the M1 and M2 phenotypes.

To obtain an understanding of the unique changes in monocytes following HCMV infection, particularly in the M1/M2 macrophage reprogramming, we examined the global dysregulation of the infected monocyte transcriptome. Although other studies have examined M1/M2 monocyte/macrophage differentiation induced by various cytokines and bacteria, we for the first time used a microarray gene profiling approach to examine the M1/M2 differentiation reprogramming in virally infected monocytes. A cDNA microarray containing 12,626 unique probe sets was used to assess HCMV modulation of genes in peripheral blood monocytes at 4 h postinfection (hpi), which our data has identified as a key temporal point when viral immediate-early proteins are not expressed (15) but when a number of critical cellular transcripts associated with early cellular responses are significantly induced (33). HCMV significantly altered the levels of 10.7% (1204 genes) cellular mRNAs in which 5.2% (583 genes) mRNAs were up-regulated and 5.5% (621 genes) mRNAs were down-regulated. Transcriptional profile comparison analysis revealed a majority of genes strictly associated with the M1 phenotype were induced by HCMV, while most genes associated with the M2 phenotype exhibited no change or a down-regulation. The up-regulation of monocytic genes implicated in M1 macrophage polarization suggest that HCMV modulates a rapid transition toward an M1 differentiation lineage, supporting our model for hematogenous dissemination of HCMV; that is, HCMV infection of peripheral blood monocytes forces cells to acquire an M1 proinflammatory phenotype to promote infected monocyte infiltration into peripheral tissue, differentiation into long-lived macrophages, and the subsequent establishment of a life-long viral persistence.

HCMV (Towne/E strain; passages 35–45) was cultured as previously described in human embryonic lung fibroblasts (33). Virus was purified on a 0.5 M sucrose cushion, resuspended in RPMI 1640 medium (Cellgro/Mediatech), and used to infect monocytes at a multiplicity of infection (MOI) of 15 for each experiment (14, 15, 33). Monocytes were diluted in RPMI 1640 medium to prevent homotypic aggregation of monocytes and rocked for 4 h at 37°C during infection; thus, a high MOI was used to ensure that all monocytes would be infected during this short incubation time. We have shown similar HCMV-induced phenotypic changes in monocytes using MOIs of 0.1–20 in previous publications (15).

Blood was drawn by venipuncture and centrifuged through a Ficoll-Histopaque 1077 gradient (Sigma-Aldrich). Mononuclear cells were collected and washed with saline (14, 15, 33). Monocytes were then isolated by centrifugation through a Percoll (Pharmacia) gradient (41). More than 95% of isolated PBMC were monocytes as determined by CD14-positive staining (data not shown). Cells were washed and suspended in RPMI 1640 medium (Cellgro/Mediatech) supplemented with 10% human serum (Sigma-Aldrich). University Institutional Review Board and Health Insurance Portability and Accountability Act guidelines were followed for all experimental protocols.

Isolated monocytes were HCMV infected or mock infected and incubated nonadherently at 37°C and total RNA was harvested 4 hpi with the RNA STAT-60 isolation kit (Tel-Test) according to the manufacturer’s protocol. Affymetrix human genome U95Av2 arrays, which contain 12,626 characterized sequences, were used to examine the cellular changes in primary human monocytes from six different human donors. Total RNA from mock-infected and HCMV-infected monocytes was harvested as described above. A minimum of 5 μg of RNA was used for each array. RNA integrity was assessed by electrophoresis on the Agilent 2100 bioanalyzer (Agilent Technologies). RNA was reverse-transcribed into cDNA and then transcribed in vitro to biotin-labeled cRNA. cRNA from each sample was fragmented and hybridized to GeneChip expression arrays following Affymetrix-recommended protocols. After washing and staining, the microarrays were scanned and the data quantified. Affymetrix Microarray Suite version 5.0 was used to determine changes in gene expression. Data Mining Tool version 3.0 was used to compile data from each of the replicates, and one-way ANOVA tests were performed to calculate p values. Spotfire DecisionSite 8.1.1 was used to group genes by ontology, generate scatter plots, and calculate correlation coefficients. The GEO accession number for these data is GSE11408.

Isolated monocytes were HCMV infected or mock infected and incubated nonadherently at 37°C on an orbital shaker. At 4 hpi, cells were pelleted and total RNA was isolated using the RNA STAT-60 isolation kit. RNA (0.5 μg) from each sample was reverse-transcribed by mixing RNA with random hexamers (0.1 μg/μl; Invitrogen), 1 mM deoxynucleotide triphosphates (Amersham Biosciences), and double-distilled H₂O and incubating at 65°C for 15 min. The RNA solution was then chilled for 1 min before 1× RT-buffer (Invitrogen), DTT (Invitrogen), 80 U of RNasin (Promega), 400 U of Moloney murine leukemia virus reverse transcriptase (Invitrogen), and double-distilled H₂O were added. After 1 h of incubation at 37°C, 2 U of RNaseH (Stratagene) was added and the samples were incubated for an additional 30 min. PCR was performed using an iCycler (Bio-Rad model 1708720). Template DNA (2 μg) was mixed with 1× iTAQ buffer (Bio-Rad), 15 mM MgCl₂ (Invitrogen), 50 μM deoxynucleoside triphosphates (Amersham Biosciences), 1.25 U of TAQ polymerase (Bio-Rad), double-distilled H₂O, and 20 μM forward and reverse primers (Integrated DNA Technologies). ICAM-1, integrin β₈, integrin α₁, IL-1β, TNF-α, CCL2, and GAPDH samples were incubated for 5 min at 95°C and cycled 35 times for 1 min at 95°C, 58°C, and 72°C. Samples were then incubated for 7 min at 72°C before storage at 4°C. Controls including samples lacking reverse transcriptase in the reverse transcription reactions and template DNA in the PCR reactions were also performed (data not shown). The following forward and reverse primers were designed with the support of Integrated DNA Technologies: ICAM-1, 5′-AAGCCAAGAGGAAGGAGCAAGACT-3′ (forward) and 5′-TGAACCATGATTGCACCACTGCAC-3′ (reverse); integrin β₈, 5′-GCTGATTGATGCGCCACAGACTTT-3′ (forward) and 5′-CAGGCAGACAAATGCAGCGGTAAA-3′ (reverse); integrin α₁, 5′-ACGCTCAGTGGAGAACAGATTGGT-3′ (forward) and 5′-AATTGTGCTGCCGAGATGACCAGC-3′ (reverse); IL-1β, 5′-AACAGGCTGCTCTGGGATTCTCTT-3′ (forward) and 5′-TGAAGGGAAAGAAGGTGCTCAGGT-3′ (reverse); TNF-α, 5′-ACCCTCAACCTCTTCTGGCTCAAA-3′ (forward) and 5′-AGGCCTAAGGTCCACTTGTGTCAA-3′ (reverse); CCL2, 5′-TCGCTCAGCCAGATGCAATCAATG-3′ (forward) and 5′-AGTTTGGGTTTGCTTGTCCAGGTG-3′ (reverse). To confirm equal cDNA loading, GAPDH RNA was amplified with 5′-GAAGGTGAAGGTCGGAGTC-3′ (forward) and 5′-GAAGATGGTGATGGGATTTC-3′ (reverse) primers. DNA bands were resolved on an agarose gel and images captured on a GelDoc system using Quantity One software (Bio-Rad).

Monocytes were isolated as described above and HCMV infected or mock infected nonadherently at 37°C. At 4 hpi, cells were centrifuged and total RNA was isolated with the RNA STAT-60 isolation kit. RNA (2 μg) from each sample was hybridized with a [α³²P]UTP-labeled human cytokine multiprobe (hCK-2b) template set (BD Biosciences) for 12 h. RNase protection assays were performed with the rabiolabeled RNA using the Multiprobe RNase Protection Assay System (BD Biosciences, 7th edition) according to the manufacturer’s protocol. Following hybridization, the samples were digested with RNaseA (Promega) and resolved on a denaturing polyacrylamide gel. The gel was then dried, and the images were captured with a PhosphorImager (Bio-Rad).

Monocytes were isolated as described above and HCMV infected or mock infected nonadherently for 6 h at 37°C on an orbital shaker. Following incubation, cells were removed by centrifugation and the supernatant was collected and stored at −80°C until use in the protein microarray assay. Cell-free culture supernatants were assayed for 37 different chemokines by using a RayBio human chemokine Ab array (RayBiotech) according to the manufacturer’s protocol. Data were analyzed by densitometry using Quantity One image analysis software (Bio-Rad).

Colloidal gold-coated coverslips were prepared as previously described (15). Briefly, glass coverslips were immersed in a 300 Bloom gelatin solution (0.5 g in 300 ml; Sigma-Aldrich), heated at 90°C for 10 min, and dried at 70°C for 45 min. A colloidal gold suspension was prepared by adding 11 ml of tissue culture water (Sigma-Aldrich) and 6 ml Na₂CO₃ (36.5 mM) to 1.8 ml AuHCl₄ (14.5 mM; Fisher Scientific), bringing the solution to a boil, and rapidly adding 1.8 ml of 0.1% formaldehyde (Fisher Scientific). While hot, 2 ml of the colloidal gold suspension was added to each coverslip and incubated at 37°C for 1 h. The coverslips were washed and transferred to 24-well plates.

Monocytes were mock infected or HCMV infected and incubated for 45 min at 37°C. Cells were then washed extensively with PBS to remove unbound virus and further incubated nonadherently for 6 h. Supernatants were collected following centrifugation to remove HCMV-infected monocytes. One milliliter of supernatants from mock-infected and HCMV-infected monocytes was added to colloidal gold-covered coverslips in 24-well plates. Next, 500 naive monocytes from the same donor were added to each well and incubated at 37°C. After 6 h of incubation, the cells were fixed in 1.5% paraformaldehyde for 15 min and mounted onto glass slides with glycerol. Track images of cells were captured using an inverted microscope at ×200 original magnification. Average area cleared per cell out of 20 cells per sample was determined by Scion imaging, and random motility was plotted as mean in arbitrary units (pixels cleared) ± SEM. The results are representative of three independent experiments from different human donors.

For compilation of the data from all experiments, the following criteria were used. Genes that had an absent call in more than two of the six HCMV-infected samples were removed from the pool of genes. The detection algorithm uses probe pair intensities to generate a detection p value, and a predefined Affymetrix threshold was used to assign a present, marginal, or absent call. Next, ANOVA tests were performed on the HCMV-infected vs mock-infected samples and p values were calculated for genes up-regulated or down-regulated 1.5-fold. A p value of ≤0.05 was used to generate a pool of genes that was statistically significant. Moreover, a fold change of 1.5 up or down in at least four of six HCMV-infected vs mock-infected samples was considered to be regulated by infection; thus, genes that may otherwise have been eliminated due to anomalous expression by a single donor were accepted. Analysis of six independent donors minimized the number of false-positives, thus allowing for a lower fold change cutoff threshold than for other microarray studies (42, 43).

We then tested the validity and reproducibility of hybridization signals from independent preparations of isolated monocyte RNAs from six different donors. Scatter plot analysis on the basis of signal was performed on genes from six mock-infected and six HCMV-infected samples that had a present call in at least four of the six HCMV-infected samples (Fig. 1). Because of the use of primary cells from different donors, these selection criteria were used to account for potential differences in gene expression between donors. Genes from three representative pairs of mock-infected vs mock-infected, HCMV-infected vs HCMV-infected, and mock-infected vs HCMV-infected samples were plotted on the basis of signal and the coefficients of determination ascertained (all possible pairwise analyses were done with similar results obtained (data not shown)). We found coefficients of determination to be ≥0.88 between multiple sample comparisons. Conversely, pairwise comparisons of mock-infected and HCMV-infected samples resulted in a lower coefficient of determination (∼0.74), verifying that the observed changes in mRNA levels by the HCMV-infected monocyte transcriptome were highly reproducible between donors.

At 4 hpi, HCMV significantly altered the levels of 1204 (10.7% of the total genes examined) cellular mRNAs, where 583 (5.2%) genes were up-regulated and 621 (5.5%) genes were down-regulated ≥1.5-fold in at least four of the six replicates (see supplemental data S1 for the complete list of altered transcripts).⁴ RT-PCR confirmed that HCMV stimulated ICAM-1, integrin β₈, integrin α₁, IL-1β, TNF-α, and CCL2 mRNA expression by 4 hpi (Fig. 2). Of the 1204 genes modulated in the monocyte transcriptome following infection, 48.4% were up-regulated and 51.6% were down-regulated (Fig. 3,A). To identify the trends in total cellular gene expression, those genes that were up-regulated and down-regulated ≥1.5-fold following HCMV infection were grouped by their behavioral patterns. Analysis of specific ontology groups in HCMV-infected monocytes showed altered levels of transcripts involved in the antiviral response (45% of the total antiviral response genes examined), apoptosis (18%), cell cycle (10%), cellular metabolism (12%), inflammation (31%), signal transduction (11%), and transcription factor activity (10%) (Fig. 3,A). Relative to the total number of genes regulated in the monocyte transcriptome during HCMV binding and entry (10.7%), the ontology analysis indicated genes involved in the antiviral (45%) and inflammatory responses (31%) to be more substantially modulated. Moreover, within these two ontology clusters, genes were disproportionately up-regulated (95% and 77% of genes involved in the antiviral and inflammatory responses, respectively). Because we have previously shown that HCMV induces a proinflammatory monocyte to promote the required functional changes in the infected cells necessary for hematogenous dissemination into tissue (14, 15, 16), we examined in more detail the gene ontologies known to be involved in inflammation (Fig. 3,B, Table I). The results presented in Table I are discussed in more detail below, where we specifically focus on genes implicated in monocyte extravasation into peripheral tissue.

HCMV encodes a number of immediate-early gene products with antiapoptotic function (44, 45, 46); however, a 3–4-wk delay in viral gene expression in HCMV-infected monocytes (15) suggests that HCMV regulates monocyte apoptosis via cellular antiapoptotic factors. Indeed, detailed examination of transcripts encoding for antiapoptotic factors revealed that 16% were up-regulated and only 4% were down-regulated at 4 hpi. These data are consistent with previous findings showing that a number of antiapoptotic-related genes such as BCL2A1 (4.9-fold increase) and TNFRSF6 (3.1-fold increase) are specifically up-regulated in M1-polarized macrophages (47). Moreover, NF-κB, a transcription factor responsible for the expression of a number of antiapoptotic and inflammatory genes (48, 49), is up-regulated 3.3-fold. These finding suggests that, even in the absence of de novo viral gene expression, HCMV can create a cellular environment skewed toward the inhibition of apoptosis that likely favors monocyte-to-macrophage differentiation and the long-term survival of the infected cells.

Adhesion molecules are required for monocyte transendothelial migration during normal and inflammatory processes (50). The tethering, rolling, and firm adhesion of monocytes to the apical surface of vascular endothelial cells are dependent on adhesion molecule expression. Monocytes must first adhere to endothelial cells and then move along the surface of the endothelial cell in search of the endothelial cell junctions (51). Selectins, ICAMs, and integrins are critical mediators of monocyte adhesion to endothelial cells before extravasation (52). M1-activated monocytes express significantly higher levels of these receptors; therefore, monocytes with a proinflammatory phenotype have a higher propensity to adhere and transmigrate into peripheral tissue (25, 29, 53). We found ICAM-1, which is necessary for the firm adhesion of monocytes to endothelial cells (54), to be up-regulated 4-fold at the transcriptional level following HCMV infection (Table I and supplemental data S1). This observation is in agreement with our previous data showing increased ICAM-1 expression on the surface of monocytes following infection, as well as with a number of other studies that have demonstrated strong up-regulation of surface ICAM-1 expression on endothelial and epithelial cells following infection (55, 56). Additionally, the microarray data indicate that the mRNA of a number of additional cell adhesion molecules such as ninjurin 1, laminin β₃, and integrins α₁, α₆, and β₈ were elevated. Conversely, α₅, α_M, and β₂ integrins were down-regulated following infection, indicating that HCMV can selectively alter integrin message expression. The specific up-regulation and down-regulation of integrins by HCMV have been previously observed in HUVECs and fibroblasts; however, it remains unclear of the consequences or reasons for such regulation (55, 57, 58). Nonetheless, our previous functional studies showing that HCMV-infected monocytes exhibited increased firm adhesion to endothelial cells 6 hpi and increased transendothelial migration 24 hpi demonstrate that the net result of the HCMV-mediated modulation of cell adhesion molecules is to increase monocyte adhesion to endothelial cells (14, 15).

An important step in monocyte migration to the tissue, following extravasation, is the degradation of the basal lamina. Matrix metalloproteinases (MMPs) are involved in the breakdown of extracellular matrix components. HCMV up-regulated the expression of both MMP1 and MMP10 ∼12-fold. MMP1 dissolves collagen types I, II, and III, (59), and MMP10 breaks down proteoglycans and fibronectin (60). HCMV may up-regulate the expression of these proteinases in monocytes to promote migration through the extracellular matrix en route to the tissue.

A classical feature of an M1 proinflammatory phenotype is the secretion of a milieu of cytokines and chemokines that promote monocyte activation along with cellular recruitment to sites of inflammation (53). Indeed, we find that HCMV infection significantly alters the levels of 25% of cytokine and chemokine mRNAs in monocytes. Consistent with our previous studies, as well as with others, proinflammatory cytokines IL-1β, IL-6, IL-12p40, IL-15, inhibin β_A, and TNF-α, which are associated with the classical activation M1 phenotype, were up-regulated following infection (14, 21, 25, 33, 61). HCMV potently induced IL-6, TNF-α, and inhibin β_A mRNA, reaching a 280-, 14.6- and 9.9-fold increase. HCMV infection also led to a down-regulation of certain cytokine genes such as vascular endothelial growth factor, which can act as a chemoattractant for T cells, and TNF ligand superfamily member 8 (CD30), which can stimulate T cell proliferation. Interestingly, M2 polarization-associated anti-inflammatory cytokines IL-1Ra and IL-10 were both up-regulated following HCMV infection (21, 25).

To confirm this unusual M1/M2 cytokine profile, RNase protection assays examining M1- and M2-associated cytokines were performed on RNA harvested from mock-infected and HCMV-infected monocytes 4 hpi isolated from multiple donors. Consistent with our gene profiling analysis, RNase protection assays show the up-regulation of cytokine genes implicated in the M1 phenotype (IL-12B, IL-1β, and IL-6) and the M2 phenotype (IL-10, IL-1Ra, and IL-18) (Fig. 4). These data highlight a unique reprogramming induced in monocytes following infection with HCMV. We next examine in depth the polarization gene profile of the HCMV-infected monocyte.

The rapid production of numerous cytokines is characteristic of a classical M1 monocyte/macrophage activation/differentiation lineage, and our transcriptome analysis indicates that HCMV-infected monocytes acquire a proinflammatory M1 phenotype. However, the up-regulation of transcripts (IL-1Ra and IL-10) implicated in the alternative M2 differentiation lineage suggests a complex and atypical M1/M2 reprogramming of monocytes following infection with HCMV. To further investigate HCMV-induced monocyte-to-macrophage differentiation, we compared the HCMV-infected monocyte transcriptome to the M1/M2 immunophenotypic profile induced by IFN-γ (M1 phenotype) and IL-4 (M2 phenotype) as described by Martinez et al. (47). Tables II and III list genes that were found to be strictly associated with M1 or M2 monocyte polarization, respectively. Transcriptional analysis of monocytes following infection with HCMV revealed that 30 (65%) M1-associated genes were up-regulated, 14 (30%) genes showed no change, and 2 (5%) genes were down-regulated by 4 hpi. Moreover, examination of functional categories indicate that the first M1-associated genes to be modulated during HCMV-induced monocyte polarization are membrane receptors (100%), apoptosis-related genes (100%), and cytokines and chemokines (66%). In contrast, of the genes associated with the antiinflammatory M2 phenotype, only 2 (4%) were up-regulated, 37 (84%) displayed no change, and 6 (14%) were down-regulated. Taken together, these data show a clear transcriptional bias toward an M1 macrophage activation phenotype following infection of monocytes with HCMV, although several transcripts implicated in antiinflammatory responses typical of alternatively activated macrophages (such as IL-1Ra, IL-10, CCL18, and CCL23) were also up-regulated.

Distinct patterns of chemokines are produced depending on whether a macrophage undergoes an M1 or M2 polarization (47). We found that 16 (44%) chemokine genes were up-regulated in monocytes within 4 hpi, while none was down-regulated (Table I). Differential analysis showed that 44% and 33% of M1- and M2-associated chemokines to be up-regulated, respectively. To confirm that the up-regulation of chemokine genes at the level of transcription conferred increased protein production, secretion of chemokines into the supernatant by monocytes 6 hpi was assayed using human chemokine Ab arrays. Examination of 37 chemokines showed that those identified by gene profiling to exhibit elevated gene expression levels also demonstrated the greatest fold increases in protein secretion with the exception of CXCL5 and IL-8, indicating that chemokine protein production strongly correlated with increased gene expression in infected cells (Fig. 5). Conversely, a few chemokines such as CCL19 and CCL9 not identified by the HCMV-infected monocyte transcriptome analysis displayed increased protein secretion, highlighting a complex biological process that likely involves multiple levels of regulation, including transcription, translation, and a regulation of the intercellular trafficking. We next examined whether the secreted chemokines from the infected monocyte samples were functional and could induce motility in naive monocytes. Treatment of uninfected monocytes with supernatant collected from HCMV-infected monocytes significantly stimulated cell motility (Fig. 6). Overall, although HCMV induces an atypical monocyte chemokine signature that does not strictly adhere to characteristics of the M1 or M2 phenotype, we suggest the overwhelming up-regulation of chemoattactants promotes viral spread by aiding infected monocyte migration to peripheral tissue and/or naive monocyte migration to sites of infection to enhance the infected cell pool.

Monocyte extravasation into tissue is a sequential process involving tethering, rolling, adhesion, and diapedesis (52). Quiescent monocytes bind to and transmigrate across endothelial cells at very low rates, but the rate of adhesion and transmigration can be significantly increased following classical activation with different pathogenic stimuli including LPS and HCMV (15, 62). However, LPS- and HCMV-activated monocytes are morphologically distinct (15); thus, while it appears that monocytes activated with different stimuli display certain similar phenotypical traits, HCMV must stimulate a distinct M1/M2 monocyte/macrophage polarization, which we advocate serves to promote viral survival and persistence. To examine this possibility, we initiated a study to examine the monocyte transcriptome shortly after infection to obtain a global understanding of the rapid M1/M2 reprogramming of HCMV-infected monocytes.

Global DNA microarray analyses demonstrated that HCMV dramatically altered the transcriptional profile of the monocyte transcriptome following infection. The regulation of gene expression was a highly specific process in which 583 mRNAs were significantly up-regulated and 621 mRNAs were significantly down-regulated. Ontology analysis revealed that most of the induced genes are associated with classically activated monocytes/macrophages. Generally, inflammation is used by the infected host to help clear infection via the release of antiviral regulatory factors such as IL-1β and TNF-α (63, 64). However, HCMV targets monocytes in vivo (4, 5, 65) and once infected appear to benefit from the induction of a strong host inflammatory response (15, 34, 35, 36, 37). Prostaglandin E₂, IL-1β, and TNF-α, key mediators in the inflammatory response, promote viral replication via stimulation of the HCMV IE genes (34, 36, 37). Moreover, the induction of M1-associated cytokines and chemokines such as IL-1β, IL-6, and TNF-α can also contribute to the shaping of the classically activated or proinflammatory macrophage phenotype, which is important for viral replication (15, 66, 67) and, as our present evidence identifies, for viral dissemination (14, 15).

The absence of viral IE gene expression (15) together with the previous observations that UV-inactivated HCMV and the major viral glycoprotein gB exhibited similar proinflammatory modulatory effects on monocyte function (33) suggest the involvement of receptor-ligand interactions. Engagement and activation of cellular surface receptors such as TLR2, epidermal growth factor, and integrins by HCMV (68) followed by the subsequent stimulation of cellular NF-κB and PI3K activity (14, 16, 33, 69) are likely the major mechanisms for the rapid monocyte gene regulation following infection, although we cannot preclude the possible involvement of some tegument proteins, which exhibit transactivator activity following viral entry (68). Overall, these studies identify that HCMV binding and/or entry is sufficient to induce a proinflammatory monocyte state exhibiting increased cytokine/chemokine secretion, motility, endothelial adhesion, transendothelial migration, and monocyte-to-macrophage differentiation.

Although transcriptional profile comparisons of genes differentially expressed in the M1 or M2 phenotype reveal a distinct bias toward the classical activation, a number of genes distinctly associated with alternatively activated monocytes were also up-regulated. The antiinflammatory cytokines IL-1Ra and IL-10 were both up-regulated following HCMV infection. Mice lacking endogenous IL-1Ra were less susceptible to infection by intracellular pathogens, supporting the importance of IL-1 in resistance to infection with intracellular organisms (70). HCMV and EBV encode viral homologs to IL-10 (71, 72), which can activate signaling pathways that dampen IFN-γ-triggered microbicidal pathways, inhibit Ag processing and presentation by APCs, and inhibit T cell cytokine production and cytotoxic activity (73). Because the HCMV IL-10 homolog is not synthesized in this early timeframe in monocytes, we propose that HCMV uses host cytokines instead to inhibit T cell function. Overall, our data hint that HCMV rapidly and selectively modulates M1- and M2-associated factors, resulting in an activated monocyte lying between the ends of the M1/M2 polarization spectrum to balance viral spread with immune evasion.

In agreement with this idea, examination of the HCMV-infected monocyte chemokine signature showed that the up-regulation of chemokines implicated both the classically and alternatively activated phenotype, although again a bias toward the M1 phenotype was observed. Thus, our data indicate that HCMV induces multiple chemoattractants regardless of M1 or M2 association, which we speculate promotes viral spread by driving infected monocyte migration into peripheral tissue and/or the recruitment of naive monocytes to sites of infection to enhance the infected cell pool. Although beneficial to the virus, chronic secretion of these chemokines could have pathological consequences to the host, such as the development of atherosclerosis where HCMV infection is strongly linked to disease development (10, 11). For example, chemokine (C-C motif) ligand 2 (4.1-fold increase) is thought to be one of the most powerful inducers of monocyte migration into atherosclerotic lesions (74).

Overall, we show that HCMV infection rapidly induces a proinflammatory monocyte. Is this activated phenotype a direct consequence of specific M1/M2 reprogramming by the virus or a general response to viruses? HIV studies including microarray analyses indicate that proinflammatory gene expression resulting in chronically activated HIV-infected monocytes/macrophages occurs in vivo (75, 76, 77). Similar to HCMV-infected monocytes, release of chemotatic cytokines such as MCP-1 from HIV-infected cells enhances dissemination by stimulating recruitment of target CD4 T cells and monocytes/macrophages to sites of infection (78). Furthermore, HIV appears to induce antiapoptotic and cell cycle-associated genes, leading to greater cell survival and viral replication, respectively (79, 80, 81). Nevertheless, although both HCMV- and HIV-infected monocytes display an activated invasive monocyte phenotype, a comparative analysis of gene expression in monocytes infected with HIV to our results revealed fundamental differences between the two viruses. In HCMV-infected monocytes, expression of inflammatory cytokine genes such as IL-1β, IL-6, and TNF-α were significantly up-regulated (presented herein), whereas in HIV-infected monocytes these genes remained unchanged (75). IL-1β and TNF-α have strong antiviral activity, and thus down-regulation by HIV is likely a strategy for immune evasion; however, as mentioned above, HCMV has devised a strategy using the NF-κB-activating properties of these cytokines to stimulate its own viral replication (36, 37). The chemokine receptor CCR5 is the major entry coreceptor for HIV and is up-regulated in HIV (82) but not in HCMV-infected monocytes. These differences are but a few examples used to highlight the transcriptional differences that occur following infection of monocytes with HCMV and HIV. Furthermore, HIV does not appear to force monocyte differentiation into macrophages, as we observed for HCMV (15), but rather as monocytes differentiate they become more susceptible to HIV infection (82). Thus, HCMV- and HIV-infected monocytes are not activated identically in the classical sense and lie at different points along the M1/M2 macrophage polarization continuum. By using microarray technology, we can dissect the unique strategies that these different viruses have evolved to promote dissemination and persistence.

The data presented in this global transcriptional profile study are consistent with our previously proposed model of HCMV dissemination, where HCMV infection classically activates monocytes to promote migration into host organ tissue and differentiation into replication permissive macrophages (14, 15, 16). Our data also confirm morphological studies suggesting that HCMV infection does not induce all of the same characteristics associated with the classically activated M1 phenotype induced by LPS. The HCMV-infected monocyte transcriptome revealed an atypical M1/M2 polarization, with a defined phenotype biased toward the M1 polarization that also incorporated selected attributes of the alternatively activated M2 phenotype. Presumably the up-regulation of specific genes associated with the antiinflammatory M2 phenotype, and in particular chemoattractants, would be advantageous for the virus. Overall, our study provides insight into how HCMV reprograms monocytes as a mechanism to promote viral dissemination and demonstrates the complexity of HCMV-induced signaling events, even in the absence of viral gene expression.

The authors have no financial conflicts of interest.