At present, only one region of the HCMV genome has been shown to code for homologs of cellular cytokines. Initial work by two groups identified a transcript expressed from HCMV UL111A during productive infection as coding for a homolog of human IL-10 (hIL-10), which was termed cmvIL-10 [5,6]. Analysis of HCMV latent infection of myeloid progenitor cells subsequently identified a second hIL-10-like transcript expressed from UL111A, termed latency associated cmvIL-10 (LAcmvIL-10) [7]. In addition to cmvIL-10 and LAcmvIL-10 transcripts, five other transcripts from the UL111A region have been reported to be expressed during the productive phase of infection [8], but as these transcripts have not yet been fully characterized and their biological function not reported, this section on cytokine homologs encoded by HCMV will focus on cmvIL-10 and LAcmvIL-10.

The cmvIL-10 transcript is comprised of three exons and two introns that encodes a protein of 175 amino acids which shares 27% amino acid identity with hIL-10, but retains the capacity to bind and signal through the hIL-10 receptor [5,6,9]. As a result, the repertoire of cmvIL-10 mediated immunomodulatory functions appears to mimic those of hIL-10 (Figure 2). For example, cmvIL-10 inhibits production of pro-inflammatory cytokines TNF-α, IL-1β and IL-6 in lipopolysaccharide (LPS) stimulated peripheral blood mononuclear cells (PBMCs) and monocytes, and decreases major histocompatibility complex (MHC) class I and MHC class II expression by monocytes [10,11,12]. Aside from inhibition of MHC molecules, cmvIL-10 also inhibits monocyte derived dendritic cell (DC) maturation and inhibits the upregulation of a number of DC costimulatory molecules thus disrupting DC function [13,14,15]. Suppression of pro-inflammatory mediators by cmvIL-10 was also demonstrated in other immune cells, including repressed synthesis of type I interferons by plasmacytoid DCs [16] and suppression of pro-inflammatory CXC chemokine ligand 10 (CXCL10) by microglial macrophages, resulting in limited chemotaxis of activated T lymphocytes towards these cells [17].

Apart from these immunosuppressive properties manifested on cells of myeloid origin, cmvIL-10 has been shown to stimulate proliferation and differentiation of B lymphocytes [18] and monocyte differentiation towards a phagocytic phenotype [19]. In addition, a role for cmvIL-10 has also been suggested in enhancing congenital HCMV disease by impairing cytotrophoblast remodeling of the uterine vasculature to limit fetal growth, as it has been demonstrated that cmvIL-10 decreased matrix metalloproteinase activity in endothelial cells and cytotrophoblasts and could impair endothelial cell migration and cytotrophoblast invasiveness in vitro [20]. This combination of cmvIL-10 controlled suppression of pro-inflammatory mediators and other important cellular functions, accompanied by stimulation of B cells may work together to tilt the host immune response towards a phenotype that is less effective at controlling virus replication. This may enhance the capacity of HCMV to not only more effectively establish primary productive infection, but may also contribute to chronic productive infection at sites of persistent viral shedding.

The LAcmvIL-10 transcript was initially identified during HCMV latency, and is a variant of the cmvIL-10 transcript [7]. Subsequent analysis has revealed that the LAcmvIL-10 transcript is also expressed during productive infection of human fibroblasts [21]. Both the site of initiation of translation and the transcriptional termination site are predicted to be the same as that of cmvIL-10, however differential splicing of the LAcmvIL-10 transcript results in the second intron of the cmvIL-10 mRNA being incorporated into the LAcmvIL-10 mRNA, resulting in an in-frame stop codon. As a consequence, the LAcmvIL-10 protein is predicted to be identical with the cmvIL-10 protein for the first 127 amino acids but then diverges for its final 12 amino acids at the C terminus, resulting in a truncated protein of 139 amino acids [7]. Nonetheless, like cmvIL-10, LAcmvIL-10 shares 27% identity with hIL-10.

Whilst cmvIL-10 exerts a broad range of immunomodulatory properties, the function of LAcmvIL-10 appears to be much more limited (Figure 2). Both cmvIL-10 and LAcmvIL-10 suppress MHC class II by primary human myeloid progenitor cells and monocytes [12,22]. However, unlike cmvIL-10 which impairs DC maturation [13,14,22], LAcmvIL-10 does not possess this function, as indicted by a lack of capacity to inhibit expression of proinflammatory cytokines TNF-α, IL-1α, IL-1β and IL-6 and costimulatory molecules CD40, CD80, and CD86 by DC stimulated to mature with LPS [22]. Furthermore, whilst cmvIL-10 stimulates B cells and phagocytic macrophages, LAcmvIL-10 does not [18,19]. The basis of the differential functional effects of cmvIL-10 and LAcmvIL-10 has not been fully elucidated, but the structural differences between these two viral cytokines provide some insights. In particular, it has been shown that whilst cmvIL-10 binds and signals through the hIL-10 receptor in a manner comparable to that of hIL-10, LAcmvIL-10 appears to exert its function differently. Studies using antibody to neutralize the hIL-10 receptor and analysis of cellular signaling events triggered by engagement of the hIL-10 receptor have demonstrated that LAcmvIL-10 either does not bind and signal through the hIL-10 receptor, or it does so in a manner different to that of cmvIL-10 and hIL-10 [22]. It therefore remains to be determined whether LAcmvIL-10 modulates MHC class II via engagement of a different cellular receptor, or whether it binds to the hIL-10 receptor in a unique manner which results in a restricted repertoire of functions.

Studies of the functions of cmvIL-10 and LAcmvIL-10 have mostly relied upon the use of recombinant proteins synthesized either from bacterial or mammalian expression systems. However, several studies have examined biological function in the context of productive or latent infections utilizing UL111A deletion viruses unable to express both cmvIL-10 and LAcmvIL-10, and collectively referred to as viral IL-10 (vIL-10) deletion viruses. Deletion of the vIL-10 gene from HCMV does not affect virus replication during in vitro HCMV productive infection of human fibroblasts [23], nor does it impinge upon the establishment and maintenance of HCMV latency in vitro [24]. However, vIL-10 acts to actively restrain the host’s immune response against HCMV during experimental latent infection, as shown by suppression of surface MHC class II on latently infected cells with accompanied suppression of CD4⁺ T cell recognition of latently infected cells [24]. Furthermore, in comparison to parental virus (with intact UL111A), vIL-10 deletion virus upregulates cytokines associated with DC formation during experimental latency and increases the proportion of myeloid DCs, some of which are Langerhans cells [25]. Thus, expression of vIL-10 during latency modifies differentiation of latently infected myeloid progenitors to inhibit formation of DCs, and this has been suggested to represent an additional means by which HCMV may restrict immune clearance of latently infected cells. It is not yet clear whether the inhibition of differentiation to a DC is due to vIL-10 mediated blockade of myeloid cell differentiation, as opposed to a skewing of differentiation to a different, less immunogenic myeloid cell type. Our hypothesis is that vIL-10 expressed during HCMV infection promotes an anti-inflammatory macrophage phenotype to aid in virus pathogenesis and this is a current focus of broader analyses of vIL-10 function in the context of HCMV infection.

Due to the ability of HCMV to cause severe or fatal disease in high risk groups such as congenitally infected fetuses and immunocompromised individuals, and the corresponding high health care costs associated with their treatment and care, development of an HCMV vaccine has been made a top priority by the Institute of Medicine in the US [26]. However, the quest for a licensed HCMV vaccine remains unfulfilled, despite a number of attempts over the last half a century (for comprehensive reviews of the HCMV vaccine field, see [27,28]).

It could be argued that the immunomodulatory properties of HCMV encoded vIL-10 proteins may represent a potential problem for the development of any live attenuated HCMV vaccine candidate as the host’s immune response towards such a vaccine may be suboptimal due to the expression of vIL-10 (present in all laboratory and wild type strains sequenced so far), resulting in decreased production of host’s pro-inflammatory cytokines and chemokines that drive the recruitment of immune cells towards the site of vaccination and/or suppression of DC maturation and limited antigen presentation due to decreased levels of MHC. Thus, the functions of vIL-10 should be considered in the development of such a vaccine to prevent HCMV disease.

In this respect, recent attempts have been made to engineer a live vaccine virus based upon deletion of a CMV encoded IL-10 homolog. Due to its strict species specificity, these investigations have, to date, relied upon the use of an animal model of HCMV infection, using rhesus CMV (RhCMV) [29,30]. Unlike murine CMV (MCMV), RhCMV encodes its own functional IL-10 homolog (RhcmvIL-10) [6]. RhCMV infection of rhesus macaques in many aspects mirrors natural HCMV infection [31] and this model enables vaccination-challenge experiments to be undertaken. The significance of CMV encoded IL-10 function in vivo is emphasized by a recent study that compared the effects of wild type RhCMV to the RhcmvIL-10 deletion mutant. In comparison to infection with wild-type RhCMV, rhesus macaques infected with a RhcmvIL-10 deletion virus displayed increased innate and adaptive immune responses, characterized by a higher inflammatory response, decreased macrophage infiltration but a significant increase in DC numbers and stronger CD4⁺ T cell proliferation, as well as higher IgG titers [30].

In addition to live vaccine design based upon deletion of RhcmvIL-10, a DNA vaccine encoding for RhcmvIL-10 protein has also been explored. Whilst this initial approach resulted in relatively weak immunogenicity [32], during natural RhCMV infection, RhcmvIL-10 was found to be highly immunogenic, resulting in strong induction of anti-RhcmvIL-10 antibodies in plasma that were capable of neutralizing RhcmvIL-10 function without cross reactivity with rhesus cellular IL-10, with anti-RhcmvIL-10 specific T cells identified in 22% of infected animals [33]. In the context of HCMV infection, antibody responses have also been detected against cmvIL-10 in 28% of infected individuals [34]. Anti-cmvIL-10 antibodies also demonstrated potent neutralizing function exclusively against cmvIL-10, with no cross reactivity with either hIL-10 or Epstein-Barr virus encoded IL-10 (ebvIL-10) [34].

To further investigate RhcmvIL-10 as a surrogate vaccine candidate for vIL-10 encoded by HCMV, mutated versions of RhcmvIL-10 unable to bind to the IL-10R but still preserving tridimensional structure of wild type RhcmvIL-10 were generated [29]. Although these mutated RhcmvIL-10 proteins lacked immunomodulatory functions of wild type RhcmvIL-10, they generated a strong anti-RhcmvIL-10 antibody response in infected animals that received DNA vaccines encoding these mutated forms of RhcmvIL-10. Importantly, these antibodies were capable of blocking the biological activity of wild type RhcmvIL-10 and did not cross react with cellular rhesus IL-10 [29]. In addition, immunization of seronegative animals with mutated forms of RhcmvIL-10 prior to infection with RhCMV resulted in prominent decrease in the number of cells infected with RhCMV as well as the frequency and amount of detectable RhCMV in these animals compared to mock immunized animals [35].

Whilst this review focuses on HCMV encoded cytokines/chemokines and their receptors, these studies of RhcmvIL-10 highlight the biological relevance of CMV encoded IL-10 in vivo and also provide an excellent example of how functional analyses of HCMV encoded IL-10 in vitro have helped to inform the development of novel potential vaccine strategies.

HCMV encodes two genes, UL146 and UL147, whose protein products exhibit limited identity with CXC-chemokines [44]. This homology includes putative signal sequences, cysteine spacing and for pUL146 an ELR-CXC sequence (glutamic acid, leucine, arginine [ELR] upstream of the CXC motif) that is common in a number of CXC-chemokines, including IL-8 and Gro-α. HCMV strain Toledo UL146 and UL147 exhibit 24% and 16% amino acid identity, respectively, to IL-8. ELR-CXC chemokine family members are known to play important roles in neutrophil metabolism [45]. Although the virus encodes two potential homologs of CXC chemokines, to date only the functional effects of pUL146 (vCXCL-1) have been reported in any detail.

Disruption/deletion of the UL146 gene from the HCMV strain Toledo genome limited the ability of HCMV infected fibroblasts to promote neutrophil chemotaxis. In addition, bacterially expressed pUL146 could also induce chemotaxis of human neutrophils to similar levels as IL-8 and Gro-α [44]. It was suggested that pUL146 mediated its effects through binding of CXCR2, rather than CXCR1, based on stable cell lines expressing each of the chemokine receptors, and could compete with IL-8 for binding to CXCR2 [44]. A more recent study examined the ability of pUL146 to activate signaling from a panel of 18 defined human chemokine receptors and demonstrated via both chemotaxis and calcium mobilization that pUL146 could induce signaling through both CXCR1 and CXCR2 [46], although the binding of pUL146 to CXCR1 was of lower affinity and potency. It would seem counterintuitive that HCMV should encode a function to recruit immune effectors such as neutrophils, however it has been postulated that neutrophils could act as potential passive carriers of HCMV to facilitate virus dissemination [44,46], providing at least a hypothetical role for pUL146 in enhancing virus spread.

The UL128 ORF encodes a spliced transcript whose protein product forms part of a pentameric complex with pUL130, pUL131A, gH (pUL115) and gL (pUL75) that plays a vital role in mediating entry of HCMV into both epithelial and endothelial cells [47]. Genes in the UL128 locus (UL128, UL130 and UL131A) are rapidly mutated following viral propagation in human fibroblasts but are maintained following passage in endothelial/epithelial cells [48]. In addition to its characterized role in HCMV entry, the UL128 protein (pUL128) contains four conserved cysteine amino acids at its N-terminus that are characteristic of the motif found in CC-chemokines [49]. On the basis of this motif it has been proposed that pUL128 may act as a chemokine [49]. Most recently, Zheng et al. expressed a soluble version of pUL128 and showed that it was capable of promoting migration of human PBMC to a similar level as another CC-chemokine MIP-1α/CCL3. Treatment of human PBMC with the soluble UL128 protein (pUL128) also induced increased expression of the pro-inflammatory cytokines TNF-α and IL-6 [50] and it is known that CC-chemokines can induce such cytokines [51]. How pUL128 promotes such activity, if it is related to its CC chemokine-like structure and if it is capable of signaling via CC-chemokine receptors has not yet been reported.

A number of publications have reviewed US28 functions [69,70,71] and a complete discussion of the pleiotropic roles of US28 is beyond the scope of this review. Here, we will focus on more recent reports detailing US28 function as a potential oncogene as well its potential immunomodulatory role during viral infection.

US28 is an early (E) gene with mRNA detected as early as 2 hours post infection in human fibroblasts [72]. US28 exhibits 30% homology with the β-chemokine receptor CCR1 and also exhibits homology in its ligand binding domain for other chemokine receptors including CCR2 and CCR5 [73]. pUS28 is the only viral family member that has been shown to bind chemokines, and can act as a receptor for a wide range of chemokines including the CC chemokines RANTES/CCL5, MCP-1/CCL2, MCP-3/CCL7, MIP-1α/CCL3 and MIP-1β/CCL4 as well as the CX3C chemokine, fractalkine/CX3CL1 [73,74,75]. Chemokine binding to pUS28 can induce an array of cellular responses including MAP kinase activation, calcium flux, cell migration and activation of transcription factors such as cyclic AMP responsive element binding protein (CREB) and NF-κB [73,76,77,78,79]. pUS28 also displays agonist independent, constitutive signaling activity inducing phospholipase C activation leading to production of the second messengers, inositol triphosphate and diacyl glycerol, and as well as promoting NF-κB activation [80,81,82,83,84]. Through such signaling pathways pUS28 can induce a wide variety of effects on host cell metabolism including promoting smooth muscle cell migration [76], which has been implicated in the association between HCMV infection and atherosclerosis [85]. A recent study also demonstrated that US28 from HCMV can be functional following viral expression in an in vivo setting [86]. Deletion of the UL33 homolog (M33), encoded by MCMV, from the viral genome limits salivary gland replication and is associated with reduced reactivation from spleen and lung explants [86,87]. Replacing M33 with US28 from HCMV at least partially complements the activity of M33 [86] and provides a system to study US28 activity in an animal model.

Essentially all known signaling functions of the protein have been first demonstrated using pUS28 expressed in isolation with much more limited numbers of reports demonstrating/confirming such activities in the context of a productive HCMV infection. It is also possible to detect US28 transcription following HCMV infection in a number of different models of latency [88,89,90] although the potential role of US28 expression in this setting has not yet been elucidated.

Compared to US28 relatively little is known about the function of the three other virally encoded members of the 7 transmembrane family, and to date these are orphan receptors with no known chemokine ligands. However, recent publications may provide insights into how US27, UL33 and UL78 may function as they appear to heteromerize with both viral and cellular G-protein coupled receptors modulating their function.

pUL21.5 (also known as pUL22A) is a small secreted glycoprotein encoded by HCMV that is dispensable for growth of both laboratory strains and clinical isolates in human fibroblasts [109]. Screening of a small panel of chemokines identified that pUL21.5 is capable of binding to the CC chemokine CCL5 (RANTES) but was unable to bind two other CC chemokines; CCL3 and CCL2 or to the CXC chemokine IL-8 [109]. Thus pUL21.5 may act as a sink to sequester free RANTES and is capable of competing for CCL5 binding to cell surface receptors on THP-1 cells [109]. Interestingly the mRNA encoding pUL21.5 is one of a limited subset of viral encoded RNAs that are incorporated into the mature virion [110] and thus should be available for translation immediately following infection without the requirement for de novo gene transcription.

CCL5 is a potent chemoattractant of immune effector cells such as monocytes and T-cells that bear the CCR5 receptor and can also signal through both CCR3 and CCR1 to recruit other cells including both macrophages and dendritic cells [111]. HCMV encodes at least five independent functions to modulate CCL5 expression and function, strongly suggesting an important role for this chemokine in the anti-HCMV immune response. In addition to pUL21.5, pUS28 has also been shown bind CCL5 potentially further sequestering this chemokine [73,92]. CCL5 mRNA transcription is induced by infection with UV irradiated HCMV that can be inhibited by expression of the IE86 immediate early protein following productive infection [112]. HCMV also encodes a miRNA, mir-UL148D, not expressed by laboratory strains of the virus, that can target CCL5 mRNA for degradation at late times post infection [113]. In addition, both pUL33 and pUL78 can modulate the response to CCL5 by modulating signaling through CCR5 (see section 6.2.2).