Functional characterization of chimpanzee cytomegalovirus chemokine, vCXCL-1_CCMV

Abstract

Human cytomegaloviruses (HCMVs) are important pathogens in immunocompromised patients and newborns. The viral chemokine, vCXCL-1, of the Toledo (Tol) strain of HCMV has been implicated in HCMV virulence. Chimpanzee CMV (CCMV) has several genes with similarity to the vCXCL-1_Tol gene, UL146. In order to test whether the CCMV viral chemokine, vCXCL-1_CCMV, is similar to vCXCL-1_Tol, we characterized its function in vitro. Receptor binding, activation, chemotaxis, signaling, and apoptosis in neutrophils were compared between vCXCL-1_Tol and vCXCL-1_CCMV and host chemokines. Although the homologues had similar activation potentials, chemotactic properties, and signaling, vCXCL-1_CCMV had a ~70-fold lower affinity for CXCR2 and displayed differences in integrin upregulation and neutrophil apoptosis. These data demonstrate that in spite of extensive amino acid variability in vCXCL-1, CCMV may provide a model for assessing the role of vCXCL-1 in CMV pathogenesis in vivo.

Introduction

Human cytomegalovirus (HCMV) is a ubiquitous β-herpesvirus, infecting between 50% and 90% of the population (Pass, 1985). Although asymptomatic in most cases, HCMV can cause serious complications in immunocompromised patients and newborns. Reactivation of latent HCMV or primary HCMV infection increases graft versus host disease (McCarthy et al., 1992), organ rejection in transplant patients (Ljungman, 1996), and inflammatory diseases such as gastroenteritis and retinitis in AIDS patients (Cheung and Teich, 1999). Congenital HCMV infection is the leading cause of infectious hearing loss in infants (Pass, 2001) and can lead to long-term sequelae such as learning disabilities and mental retardation. Due to the life-long effects associated with CMV congenital infections, the Institute of Medicine lists the development of a CMV vaccine as one of its top five priorities (Arvin et al., 2004). In an effort to develop a safe and effective vaccine, a more thorough understanding of HCMV pathogenesis is needed.

HCMV has a number of genes that encode proteins capable of suppressing or modulating the immune system that have been implicated in pathogenesis (Mocarski, 2002; Wiertz et al., 1997). Every aspect of the immune response can be altered including class I and class II antigen processing and presentation as well as inflammatory responses (Mocarski, 2002). Proteins that alter immune responses likely contribute to HCMV’s pathogenesis. Therefore understanding HCMV immune modulators may aid in designing safe and effective vaccines or potential antiviral therapeutics.

The Toledo strain of HCMV, a strain that is known to cause disease in humans (Quinnan et al., 1984), was shown to produce a functional viral chemokine, vCXCL-1 (Penfold et al., 1999). Although chemokines are divided into four families, the majority of chemokines fall into the two families, CC and CXC chemokines, based on the spacing of their amino terminal cysteines. The presence of a glutamate, leucine, arginine (ELR) motif amino terminal to the CXC further subclassifies the CXC chemokines into ELR-CXC or non-ELR families. The ELR-CXC family of chemokines was initially thought to function exclusively on neutrophils (Hebert et al., 1991) but also induces angiogenesis via endothelial cells (Addison et al., 2000; Coughlan et al., 2000; Strieter et al., 1995). The vCXCL-1 protein is a member of the ELR-CXC chemokine family (Sparer et al., 2004). Indeed, vCXCL-1_Tol (the subscript indicates the strain from which the protein is derived, e.g. Toledo) was shown to bind exclusively to CXCR2 and to induce calcium flux and chemotaxis of neutrophils (Penfold et al., 1999). Penfold et al. speculated that the induction of neutrophils could be used as a means for viral dissemination, thus increasing virulence. In fact HCMV can be isolated from neutrophils in immunocompromised patients (Martin et al., 1984; Saltzman et al., 1988; van der Bij et al., 1988) and neutrophilic infiltrates are found in AIDS patients who have CMV related retinitis (Holland et al., 1983; Pepose et al., 1985). The UL146 gene, which produces the vCXCL-1 protein, is absent in an attenuated lab strain, AD169, and based on these findings, the vCXCL-1 protein has been hypothesized to be a potential virulence factor (Cha et al., 1996). Testing this hypothesis in vivo is limited by the species specificity of cytomegaloviruses. In this study we wished to further evaluate the effects of vCXCL-1_Tol on neutrophil function in vitro and to compare those findings with host chemokines and a related chemokine from chimpanzee CMV (CCMV).

vCXCL-1_CCMV was chosen for comparison to vCXCL-1_Tol to evaluate whether the conservative substitution in vCXCL-1_CCMV’s ELR motif and variability in the N-loop region, both of which are considered important to receptor binding and triggering for host chemokines (Clark-Lewis et al., 1993, 1994; Lowman et al., 1996), affect protein function. In our work presented here, we compare differences in receptor binding affinity, cellular activation, signal transduction, chemotaxis, upregulation of adhesion molecules, and anti-apoptotic functions, between vCXCL-1_Tol, vCXCL-1_CCMV, and host chemokines.

Results

Comparison of amino acid sequences

Sequence analysis indicates vCXCL-1_CCMV has limited homology to vCXCL-1_Tol, CXCL1, and CXCL8 (Fig. 1). Although the amino acid sequences show relatively low homology, all share important characteristics of CXC chemokines, including an N terminal signal peptide, spacing of the four conserved cysteines, and ELR like motif. The ELR-CXC motif is present in all CXC chemokines that bind CXCR1 and CXCR2. While the vCXCL-1 from clinical isolates sequenced previously display conservation of this ELR-CXC motif (Arav-Boger et al., 2005; Lurain et al., 2006; Prichard et al., 2001), it is modified in vCXCL-1_CCMV with an L→M conservative substitution. Even though this is a conservative substitution, this replacement could affect neutrophil function based on published results showing that host chemokines with a modified ELR motif are subsequently impaired for receptor binding and calcium mobilization (Clark-Lewis et al., 1993; Hebert et al., 1991). Overall, the vCXCL-1_Tol and vCXCL-1_CCMV proteins are 52% similar with only 22% identity. This observation mirrors the low identity previously observed in the sampling of human clinical isolates, suggesting that only core, conserved regions of the protein are required for functionality. The N-loop region is another domain that is important for receptor engagement and function. This region lies between the second cysteine and the 3₁₀ helix usually from amino acids 10–20 (Fernandez and Lolis, 2002). Only a conserved proline residue within the N loop is conserved between the two proteins (Fig. 1). These differences in primary amino acid sequence may have pleiotropic effects. Alterations in these domains may lead to differences in affinity that subsequently alter binding and/or function. We set out to determine whether the variability within vCXCL-1_CCMV results in differences in function.

Fig. 1.

Amino acid alignment of the mature forms of CXCL1, CXCL8, vCXCL-1_CCMV, and vCXCL-1_Tol. Residues that are 100% conserved are indicated above the alignment. The N-loop region is residues 10–20. vCXCL-1_CCMV is 22% identical and 52% similar to vCXCL-1_Tol, 17% identical and 33% similar to CXCL1, and 18% identical and 40% similar to CXCL8.

vCXCL-1 proteins can be produced and purified using the baculovirus expression system

In order to address functional differences between the viral chemokines, we generated both proteins using recombinant baculovirus. We chose this system because baculovirus expression systems utilize the mammalian signal peptidase cleavage site and generate eukaryotic glycosylation patterns and folding. Following infection of HiFive cells with recombinant baculovirus expressing the His-tagged viral chemokines, the secreted protein was purified using Ni-NTA agarose beads. A silver stain of the isolated proteins indicates the high purity achieved in the isolation procedure (Fig. 2A). The apparent size of both vCXCL-1_CCMV and vCXCL-1_Tol at approximately 14 kDa is larger than the predicted 10 kDa. The differences in molecular weight were due to glycosylations (both have two predicted N-linked glycosylation sites) as subsequent digestion with the glycosidase, PNGase F, generated proteins of the predicted size (10 kDa) (data not shown). Mass mapping of trypsin-digested vCXCL-1_CCMV and vCXCL-1_Tol proteins confirmed the identities of the isolated proteins (Supplemental data 1). N-terminal sequencing confirmed proper cleavage of the signal peptide (Fig. 2B). Proper cleavage of the signal sequence is critical for chemokine function (Clark-Lewis et al., 1991). In fact, alteration by the addition or deletion of a single amino acid has been shown to abrogate the function of some chemokines (Proudfoot et al., 1996,1999). N-terminal sequencing indicated the signal peptide of vCXCL-1_CCMV was cleaved at the threonine residue directly preceding the EMR-CXC motif as predicted by the SignalP program (Fig. 2B). Of the two possible cleavage sites predicted for vCXCL-1_Tol, cleavage occurred at the threonine residue (T23) corresponding to that of vCXCL-1_CCMV. Our data show that the T23 site is the preferential cleavage site. A synthetic vCXCL-1_Tol based on the other signal sequence cleavage site (Y18) was previously shown to be 100-fold less active than that cleaved at the threonine residue corresponding to the N terminus in our preparations (Penfold et al., 1999). This is the first confirmation that this cleavage site is used for naturally processed vCXCL-1_Tol indicating that the active form of the protein is preferentially synthesized.

Fig. 2.

vCXCL-1 protein production. (A) vCXCL-1_CCMV and vCXCL-1_Tol production in baculovirus. 1 μg of 6His-tagged protein eluted from nickel beads was run on a 15% SDS PAGE gel and silver-stained. (B) N-terminal sequencing confirmed the predicted cleavage of the signal peptide of vCXCL-1_CCMV and vCXCL-1_Tol.

vCXCL-1_CCMV induces intracellular calcium release in PBN

With the differences noted in primary amino acid sequence, we assessed whether these differences correlate with differential functions. The release of intracellular calcium is often used as an indicator of chemokine induction of cellular activation (Murphy, 2003). We examined the ability of vCXCL-1_CCMV to induce intracellular calcium mobilization in freshly isolated PBN. The dose response analyses of vCXCL-1_CCMV and vCXCL-1_Tol showed similar activation potential (Fig. 3A), demonstrating that the amino acid differences did not alter triggering of intracellular calcium release.

Fig. 3.

vCXCL-1_CCMV has a similar activation threshold as vCXCL-1_Tol and can be desensitized by CXCL1. (A) Changes in fluorescence were measured over time after exposure of human PBN to different concentrations of vCXCL-1_CCMV or vCXCL-1_Tol. Arrows indicate the time of addition of the chemokine. (B) Desensitization profile of vCXCL-1_Tol/CXCL1 and (C) vCXCL-1_CCMV/CXCL1. 100 nM of CXCL1, vCXCL-1_CCMV, or CCL5 were added at the times indicated by the arrow. A second addition of 100 nM of the indicated chemokine was added at 90 s. Changes in fluorescence were measured over time. Data is representative of five experiments.

Previous work demonstrated the ability of vCXCL-1_Tol to desensitize PBN activation following exposure to a number of human chemokines, including CXCL1 (Penfold et al., 1999). Receptor desensitization is determined by the receptor affinity for the chemokine, phosphorylation of the carboxyl terminus, and subsequent β arrestin binding. We confirmed the ability of vCXCL-1_Tol to desensitize PBNs to CXCL1 activation. However, CXCL1 is not able to completely desensitize to vCXCL-1_Tol (Fig. 3B).

The desensitization profile of vCXCL-1_CCMV (Fig. 3C) was compared to that of vCXCL-1_Tol. Unlike vCXCL-1_Tol, vCXCL-1_CCMV is unable to completely desensitize PBN to activation via CXCL1. In addition CXCL1 is able to completely desensitize to vCXCL-1_CCMV added to PBN following CXCL1. Because CXCL1 binds only to CXCR2, its ability to inhibit vCXCL-1_CCMV calcium mobilization suggests that vCXCL-1_CCMV, like vCXCL-1_Tol, binds via CXCR2. Treatment of cells with the CXCR2 inhibitor, SB 225002, prior to exposure to vCXCL-1_CCMV resulted in minimal calcium mobilization further indicating that vCXCL-1_CCMV functions via CXCR2 (data not shown). The incomplete desensitization to CXCL1 is likely due to a lower affinity of human CXCR2 (hCXCR2) for vCXCL-1_CCMV. Alternatively, a reduced ability of vCXCL-1_CCMV to initiate phosphorylation of CXCR2 to the same degree as vCXCL-1_Tol could result in lower levels of receptor desensitization (Richardson et al., 2003). These results indicate that vCXCL-1_Tol engages hCXCR2 differently than vCXCL-1_CCMV such that different desensitization patterns are triggered.

vCXCL-1_CCMV has reduced affinity for CXCR2 compared to vCXCL-1_Tol or CXCL1

The ability of CXCL1 to desensitize to vCXCL-1_CCMV in the calcium mobilization assays suggests this viral chemokine binds to CXCR2. However, the lack of complete desensitization when vCXCL-1_CCMV was used to desensitize against CXCL1, could reflect differences in receptor binding affinity for the viral chemokine. To test this possibility, competition-binding assays were used to evaluate the ability of vCXCL-1_CCMV to compete for binding to CXCR2. vCXCL-1_CCMV was able to compete with ¹²⁵I CXCL1 for binding to hCXCR2, albeit at a higher 50% inhibitory concentration (IC₅₀) than that required for CXCL1 or vCXCL-1_Tol (Fig. 4). We conclude that vCXCL-1_CCMV binds to hCXCR2 but with a lower affinity than vCXCL-1_Tol (vCXCL-1_Tol IC₅₀ =0.9 nM vs. vCXCL-1_CCMV IC₅₀ =65 nM) and this may explain the differences in desensitization patterns seen in Fig. 3C. (i.e. CXCL1 can prevent subsequent calcium flux via vCXCL-1_CCMV while vCXCL-1_CCMV cannot desensitize to exposure to CXCL1). Affinity may not be the sole explanation for the differences between vCXCL-1_Tol desensitization patterns with CXCL1 (Fig. 3B), yet no of difference in the calcium flux activation potentials (Fig. 3A). The viral chemokines could be engaging the receptors such that β arrestins are not activated as efficiently. This would explain the differences in desensitization patterns while not affecting the activation potentials.

Fig. 4.

vCXCL-1_CCMV binds with less affinity than vCXCL-1_Tol to CXCR2. 293 cells stably expressing only CXCR2 were incubated with a constant amount of ¹²⁵I-CXCL1 and increasing amounts of unlabeled chemokines. Calculated IC₅₀ values are indicated in molar concentrations in parentheses. Data is representative of two experiments.

vCXCL-1_CCMV induces migration of human neutrophils

An important function of chemokines is their ability to induce chemotaxis of a variety of cells. ELR-CXC chemokines and vCXCL-1Tol trigger migration of neutrophils (Penfold et al., 1999). We tested the ability of vCXCL-1_CCMV to induce migration of human PBN using a modified Boyden chamber. vCXCL-1_CCMV induces migration of neutrophils with equal potency as the host chemokine CXCL1 (Fig. 5) and vCXCL-1Tol (Penfold et al., 1999). Once again, in spite of differences in affinity for hCXCR2 and primary amino acid sequences, vCXCL-1_CCMV chemoattracts PBNs efficiently.

Fig. 5.

vCXCL-1_CCMV induces migration of neutrophils. Migration assays were performed by loading human PBN onto the upper chamber of a 96-well plate fitted with a 5 μm filter with increasing amounts of chemokine loaded in the lower chamber. Results are reported as a chemotactic index of the number of cells migrated divided by the random migration±S.D. Data is representative from three experiments.

vCXCL-1_CCMV activates ERK and Akt signaling pathways

The differences in affinity between the chemokines may affect functional activation (Clark-Lewis et al., 1991, 1994; Geiser et al., 1993; Wu et al., 1996) as we observed with desensitization patterns (Fig. 3). Upon CXCR2 binding, G protein signal transduction cascades are activated (De Lean et al., 1980). Two of the central signal transduction cascades activated upon CXCL1 binding to CXCR2 are the extracellular signal-related protein kinase (ERK), a member of the mitogen-activated protein (MAP) kinase family (Shyamala and Khoja, 1998) and protein kinase B/Akt pathway, which is activated via the phophoinositide-3 kinase signaling cascade. The activation of the ERK and Akt signaling pathways are triggered following exposure to chemokines and β-arrestin activation (Lefkowitz and Shenoy, 2005). In order to assess whether the differences in binding to CXCR2 induce different signaling cascades, we analyzed the activation profile of differentiated HL-60 cells stably expressing CXCR2 following exposure to vCXCL-1_CCMV. HL60s were used due to the difficulty of isolating fresh PBNs without inadvertent signal transduction activation and host-to-host variation. Exposure of differentiated HL-60 cells to vCXCL-1_CCMV and vCXCL-1_Tol resulted in phosphorylation of both ERK and Akt to similar levels as the CXCL8 and CXCL1 controls (Fig. 6). At least for these two important signaling pathways, the affinity for hCXCR2 does not markedly impact activation.

Fig. 6.

Both viral and host chemokines activate the ERK and Akt pathways. Differentiated HL-60 cells stably expressing CXCR2 were stimulated with 100 nM chemokine for 1 min at room temperature and lysed instantly. Lysates were immunoblotted using antibodies against the phosphorylated (p) and unphosphorylated ERK and Akt proteins as indicated. This is representative of two experiments.

vCXCL-1_CCMV upregulates CD11b and CD11c on the surface of PBN

β₂ integrins, including CD11a, CD11b, and CD11c, are membrane bound receptors present on leukocytes that are necessary for PBN adhesion to and migration across the endothelium (Mayadas and Cullere, 2005). Because of their potential importance in recruitment of PBN to the site of CMV infection, we tested the ability of vCXCL-1_CCMV and vCXCL-1_Tol to alter surface expression of these receptors on PBN. Exposure of human PBN to vCXCL-1_CCMV and vCXCL-1_Tol resulted in no change in the levels of CD11a expressed on the cell surface (Fig. 7). However, CD11b and CD11c levels increased upon exposure to vCXCL-1_CCMV (30% and 12% above control respectively) and to similar levels as CXCL1 (51% and 11%) (Fig. 7). Interestingly, vCXCL-1_Tol also increased expression of these proteins on PBN but to somewhat higher levels (106% and 36%) compared to CXCL1 or vCXCL-1_CCMV. PBN exposed to CXCL8 showed similar levels of CD11b/c expression (106% and 46%) as those exposed to vCXCL-1_Tol. These results imply that the two viral chemokines differentially induce adhesion molecule upregulation: vCXCL-1_Tol is similar to CXCL8 stimulation and vCXCL-1_CCMV is similar to CXCL1. These data show that amino acid differences between these chemokines can lead to different activation profiles on PBN that could affect their functions in vivo.

Fig. 7.

vCXCL-1_Tol and vCXCL-1_CCMV differentially increase CD11b and CD11c expression on the surface of neutrophils. Human PBN were incubated in the presence of 100 nM of different chemokines (vCXCL-1_Tol, vCXCL-1_CCMV, CXCL1, or CXCL8) and labeled with fluorescently-conjugated antibodies against CD11a, CD11b, and CD11c. The table summarizes the flow cytometry data and is expressed as percent change in the mean fluorescent intensity compared to non-stimulated PBN. This is a representative experiment of three experiments.

vCXCL-1_CCMV decreases apoptosis of human PBN

PBN normally undergo apoptosis in 24–48 h (Haslett, 1992). Because PBN play a role in the dissemination of HCMV (Gerna et al., 1992, 2000), HCMV may have evolved viral chemokines that alter PBN viability. We assessed the capacity of vCXCL-1_CCMV and vCXCL-1_Tol to alter apoptosis of PBN. We found that vCXCL-1_CCMV and vCXCL-1_Tol decreased apoptosis (17% vs. 29% respectively) (Fig. 8). This decrease was similar to CXCL1 (21%). However, it was not as great as the reduction following exposure to CXCL8 (44%). Any differences in the longevity of neutrophils could affect viral survival or dissemination in vivo, which eventually affects HCMV pathogenesis.

Fig. 8.

vCXCL-1_CCMV and vCXCL-1_Tol reduce apoptosis in neutrophils. Human PBN were incubated for 24 h in the presence of either viral or human chemokines (100 nM). After incubation, cells were labeled with annexin-FITC and propidium iodide and then analyzed by flow cytometry. The percent reduction in apoptosis relative to the non-stimulated cells is summarized in the table. This is a representative experiment of five experiments.

Discussion

The significant findings of this paper are that two distantly related viral chemokines (vCXCL-1_Tol and vCXCL-1_CCMV) have similar functions in spite of distinct differences in affinity for CXCR2 and amino acid composition in the N loop and ELR motifs. These regions of host chemokines have been shown to play a role in receptor binding and activation (Clark-Lewis et al., 1993; Fernandez and Lolis, 2002; Hebert et al., 1991). Although vCXCL-1_CCMV was able to bind CXCR2, it was not able to bind with as high of an affinity as vCXCL-1_Tol or CXCL1 in competition binding studies (Fig. 4). This difference could reflect the difference in affinity of vCXCL-1_CCMV for human CXCR2 versus chimpanzee CXCR2 in these assays. However, the CXCR2 receptor in these two species is highly conserved. Chimpanzee CXCR2 is 99% identical to human CXCR2 based on amino acid comparisons. Only 2 amino acids out of 355 are different. These two amino acids are located in transmembrane regions 5 and 7 (Alvarez et al., 1996). Importantly, the transmembrane regions of CXCR2 that are known to be important for receptor structure are conserved between the human and chimpanzee receptors (Baggiolini et al., 1997). It is possible that binding motifs and structural differences not yet identified could result in differential binding of vCXCL-1_CCMV to hCXCR2, although this seems unlikely.

It is perhaps more likely that differences in the viral chemokines themselves are responsible for the differences in binding affinity. This may be due in part to the mutated ELR motif (L→M) present in vCXCL-1_CCMV. The ELR motif is known to be important in receptor binding and signaling (Clark-Lewis et al., 1991; Schraufstatter et al., 1993). Mutagenesis of each of the ELR residues of CXCL8 reduced the ability of CXCL8 to competitively bind to neutrophils and CXCR2 expressing cell lines (Hebert et al., 1991). However, these mutations did not completely eliminate receptor binding and other regions, such as the N-loop region, are also important in receptor binding (Fernandez and Lolis, 2002). vCXCL-1_CCMV only shares the single amino acid, proline, with CXCL1 in this region and the sequence diversity in this region may contribute to the differences seen in binding affinity. Additionally, the presence of a glycine at residue 31 has been shown to be important for receptor binding (Clark-Lewis et al., 1994, 1995; Li et al., 2002). vCXCL-1_C956 from a clinical isolate of HCMV has a substitution at the G31 residue but shows equivalent binding to CXCR2 and an equal capacity for inducing calcium mobilization to levels comparable to CXCL1 (unpublished data). This data demonstrates that the results seen with vCXCL-1_CCMV, which also has a substitution at this residue, is not solely responsible for its lower binding affinity to CXCR2. Finally, the carboxyl terminus plays some role in the function of chemokines. Although not thought to be important in all cases, the carboxyl terminal region of CXCL1 has been implicated in receptor binding (Roby and Page, 1995). The C-terminal region of the viral chemokines is considerably longer than that of CXCL1 and CXCL8 and may contribute to differential binding or signaling of the receptor (Fig. 1 and Penfold et al., 1999).

Evolution has directed the variation of HCMV genes such that viral genes have similar functions with limited homology to host genes. As discussed above, vCXCL-1 shows sequence variation from its cellular counterpart, CXCL1. Additionally, sequencing of human CMV clinical isolates has revealed extreme sequence variability between strains, with only limited regions of conservation and identities below 30% (Arav-Boger et al., 2006b; Prichard et al., 2001). Similarly UL144, a viral tumor necrosis factor (TNF)-alpha-like receptor gene, displays high variance in its sequence between clinical strains and correlates with severity of congenital disease (Arav-Boger et al., 2006a; Arav-Boger and Pass, 2002). Both human and rhesus CMVs encode interleukin 10 homologues, which retain functionality despite a sequence identity of less than 30% to their cellular counterparts (Kotenko et al., 2000; Spencer et al., 2002). Such variability among secreted and surface-displayed proteins may be driven by selective immune pressures, although why certain genes appear more susceptible to these pressures than others is unknown.

An important question is whether the reduced binding affinity by vCXCL-1_CCMV results in differences in the functional capacity of this protein. Relative differences in activity between CXCL8 and vCXCL-1_CCMV can be difficult to assess based on the ability of CXCL8 to bind and signal via both CXCR1 and CXCR2, whereas vCXCL-1_CCMV, based on our data, only activates CXCR2. Therefore, we also compared the biological activity of vCXCL-1_CCMV to the CXCR2-specific chemokine, CXCL1. As we have shown, vCXCL-1_CCMV induces activity in most biological assays comparable to that of CXCL1. Given the lower binding affinity of vCXCL-1_CCMV, we also tested the activation profile of vCXCL-1_CCMV at 1 μM, which is 15 fold above the IC₅₀, to ensure complete saturation of all expressed receptors. However, the higher concentration of vCXCL-1_CCMV did not result in a greater calcium flux than at 100 nM, therefore it is reasonable to assume that the receptors were saturated at the 100 nM concentration used in all experiments. This leads us to conclude that lower receptor binding affinity does not correlate with a reduction in biological activity when compared with CXCL1. One exception where reduced binding affinity may be important is the ability of vCXCL-1_CCMV to completely desensitize neutrophils upon subsequent exposure to CXCL1. In this case, the ability of CXCL1 to compete more effectively (i.e. higher binding affinity) for binding to CXCR2 allows it to displace vCXCL-1_CCMV and induce a second wave of calcium mobilization. Interestingly, vCXCL-1_CCMV was unable to desensitize PBN to CXCL1 even when used at a 10-fold higher concentration compared to CXCL1 (data not shown) suggesting the lack of complete desensitization may be due in part to differential activation and phosphorylation of CXCR2. Phosphorylation of G protein-coupled receptors prevents reassociation of the G protein and the receptor, resulting in a receptor that is desensitized to further stimulation by ligand (Neel et al., 2005).

vCXCL-1_Tol was previously found to functions via CXCR2, inducing calcium mobilization and chemotaxis of PBN with a potency comparable to CXCL8 (Penfold et al., 1999). We have verified similar findings for the CCMV viral chemokine, vCXCL-1_CCMV. vCXCL-1_CCMV showed a similar profile of calcium mobilization as vCXCL-1_Tol at the concentrations tested. Additionally, vCXCL-1_CCMV was able to induce migration of neutrophils as effectively as CXCL1. The ability of vCXCL-1_CCMV to induce PBN chemotaxis could enable the virus to modulate the immune response to benefit viral survival or dissemination. An example of a viral chemokine that functions in viral dissemination is the MCK2 chemokine from murine cytomegalovirus (MCMV) (Fleming et al., 1999; Saederup et al., 1999, 2001). Infection of mice with MCMV lacking the MCK chemokine gene resulted in decreased inflammation at the site of infection and dissemination of the virus. This evidence illustrates the ability of viral chemokines to alter the immune response in vivo (Saederup et al., 2001). Although these chemokines are from two different chemokine families (CXC for vCXCL-1_CCMV and CC for MCK2), they could be functional homologues. This is a common theme for CMVs. Other immune modulatory proteins such as the class I down regulating proteins from the M152 gene in MCMV (Ziegler et al., 1997) and the US2, US3, and US11 genes in HCMV (Gewurz et al., 2001; Landolfo et al., 2003; Rehm et al., 2002), lack sequence homology yet both viruses achieve the same end result of down-regulation of antigen presentation.

Chemotaxis of circulating neutrophils is dependent on adhesion to and extravasation across the endothelium. This process can be mediated by the interaction of β₂ integrins present on the surface of PBN with their ligands on endothelial cells. The β₂ integrin family includes CD11a, CD11b, CD11c, and CD11d. Previous studies demonstrated the ability of the chemokines, CXCL8 and CXCL1, to upregulate CD11b expression (Detmers et al., 1990). Here we demonstrate that both vCXCL-1_CCMV and vCXCL-1_Tol increase CD11b and CD11c expression on the cell surface. Furthermore, CXCL8 upregulation of CD11b on PBN involves the ERK signaling pathway (Takami et al., 2002), which was shown to be upregulated by both viral chemokines (Fig. 6). Therefore, these viral chemokines could enhance the ability of PBN to adhere to and cross the endothelium allowing migration of neutrophils through the tissue to the site of infection.

Regulation of apoptosis of neutrophils could also provide an advantage to HCMV by allowing PBN to survive and disseminate HCMV. In fact HCMV produces several proteins that alter apoptosis in infected cells (McCormick et al., 2005; Smith and Mocarski, 2005). Our results show that vCXCL-1_CCMV, vCXCL-1_Tol, and CXCL1 reduce apoptosis to similar levels. CXCL8 shows the highest reduction in apoptosis. However this chemokine has also been shown to alter apoptosis primarily via CXCR1 (Glynn et al., 2002) through activation of the ERK and Akt pathways in neutrophils (Akgul et al., 2001; Khreiss et al., 2004). The similarity in vCXCL-1_CCMV and vCXCL-1_Tol activation of the ERK and Akt pathways suggests that other pathways must also be involved. Prolonging PBN survival would be advantageous for HCMV. The longer the neutrophil is viable, the greater the chance it will be used for dissemination (Grundy et al., 1998).

In conclusion, we have shown that the CCMV viral chemokine has similar biological activities to vCXCL-1_Tol and CXCL1 except for integrin upregulation and inhibition of apoptosis. This is in spite of differences in binding affinities and amino acid variability in the N-loop region and ELR motif. A similar example is found in human herpesvirus 8 (HHV8) viral chemokines vMIP I and vMIP II (Boshoff et al., 1997; Kledal et al., 1997). This chemokine has limited homology (38% and 41% respectively) to host chemokines CCL2 (MIP1α) and CCL5 (RANTES) (Fernandez et al., 2000). These chemokines have similar functions to host chemokines yet can act as antagonists and bind to additional host chemokine receptors. Although we have not found antagonistic functions for vCXCL-1s, vMIP II is yet another example of a herpesvirus evolving a chemokine with unique functions.

We can infer from our data that very simple structural motifs guide receptor binding and activation rather than specific amino acid-to-amino acid interactions. The eotaxin family of host chemokines provides a similar example. These chemokines also have considerable variability yet slight differences in binding and activation (Forssmann et al., 1997; Kitaura et al., 1999; Shinkai et al., 1999; White et al., 1997). All eotaxins bind and/or activate the same eotaxin receptor, CCR3, in spite of very little amino acid identity (Duchesnes et al., 2006). There is some variability in the responses to the different eotaxins leading to speculation that these proteins may exist to “fine-tune” their interaction with their receptor to produce different biological outcomes. We propose that the variability in the viral chemokines helps to fine tune the neutrophil response, which allows for more efficient HCMV dissemination. This explains the high degree of variability that occurs naturally in HCMV isolates (He et al., 2006) and between their homologous host chemokines. The vCXCL-1s from different strains of HCMVs could allow for increased dissemination via fine-tuned alterations of apoptosis and integrin expression that provides a survival advantage in the specific hosts.

Materials and methods

Cells and viral DNA

SF9 insect cells were cultured in SF900 II media (Gibco) supplemented with 5% FBS, 2 mM L-glutamine, and 1% gentamicin (Biowhittaker). HiFive insect cells (Cellgro) were cultured in Insect Express media (Biowhittaker). Chimpanzee cytomegalovirus DNA was generously provided by Dr. Gary Hayward (Johns Hopkins School of Medicine).

Sequence analysis

Amino acid sequence alignments of vCXCL-1_Tol and vCXCL-1_CCMV were performed using the clustalW algorithm of MegAlign sequence analysis software (DNASTAR). SignalP 3.0 (http://www.cbs.dtu.dk/services/SignalP/) was used to predict the signal sequence cleavage sites of vCXCL-1_CCMV and vCXCL-1_Tol.

Protein purification and analysis

The vCXCL-1 gene, UL146, was PCR amplified from either HCMV or CCMV viral DNA and cloned into the baculovirus transfer plasmid 1392 (Invitrogen), which contains homologous regions for recombination into the baculovirus genome. PCR primers were designed to include the coding sequence for six histidines on the carboxyl terminus of the proteins for purification and Western blotting. SF9 cells were transfected with the plasmid construct and Sapphire linearized baculovirus DNA (Orbigen). Recombinant baculovirus containing the UL146 gene from either HCMV or CCMV was titered and used to infect HiFive cells for optimum protein expression. Cells were harvested after 48 h of incubation and protein was isolated from the supernatants using Ni-NTA agarose beads (Qiagen) and resuspended in PBS. Protein concentration was quantified using the Bradford assay (Biorad) according to the manufacturer’s instructions.

N-terminal sequencing was used to verify cleavage of the signal peptide. 100 pmol of vCXCL-1_CCMV and vCXCL-1_Tol was run on a 15% SDS-PAGE gel and then blotted onto PVDF membrane. The protein bands were excised from the membrane and analyzed by the Stanford PAN facility using de novo N-terminal sequencing.

Receptor binding analysis

The ability of vCXCL-1_CCMV and vCXCL-1_Tol to compete for binding to CXCR2 was evaluated as previously described (Penfold et al., 1999). Briefly, HEK293 cell transfectants over-expressing CXCR2 were incubated with ¹²⁵I-labeled CXCL1 and increasing concentrations of unlabeled chemokines. Cells were collected on glass filters, washed, and bound radioactivity was measured by liquid scintillation counting.

Activation of signal transduction pathways

Differentiated HL-60 cells stably expressing CXCR2 were stimulated with different chemokines at 100 nM for 1 min at room temperature and lysed in 2× RIPA buffer containing protease and phosphatase inhibitors (Walker et al., 2005). 20 ug of total protein was then run on a 10% SDS PAGE and immunoblotted with antibodies against phosphorylated Akt (Thr 308) (Cell Signal Technology), Akt, phosphorylated ERK, and ERK (Santa Cruz Biotechnology).

Neutrophil isolation

Peripheral blood neutrophils (PBN) were isolated from EDTA-treated blood from healthy human volunteers using dextran sedimentation and density gradient centrifugation as previously described (Markert et al., 1984). Erythrocytes were removed by hypotonic lysis. Neutrophils were resuspended in buffer or media as stated for the individual assays. Viable neutrophils were quantified by trypan blue exclusion. The use of human subjects has been approved by the University of Tennessee Institutional Review Board (IRB# 6476B).

Intracellular calcium mobilization assays

Release of calcium from intracellular stores was determined in freshly isolated PBN resuspended in PBS. Cells were loaded with 3 μg/ml Indo-1-AM (Molecular Probes) for 60 min at 37 °C. Cells were washed one time with PBS and diluted to 3–5×10⁶ cells/ml in Hanks’ balanced salt solution (HBSS) containing Ca²⁺ and Mg²⁺ and 1% FBS. Chemokines were added at varying concentrations to 2 ml of cells. All host chemokines were purchased from Pep-rotech (Rocky Hill, NJ) and endotoxin levels were less than 0.1 ng per μg (1 EU/μg). Calcium release was measured using a Photon Technology International Spectrophotometer (New Jersey) at an excitation of 350 nm. Relative intracellular calcium levels were expressed as the ratio of emissions at 490 nm to the emissions at 400 nm (Sparer et al., 2004).

PBN migration assays

Migration assays were performed on human PBN resuspended in HBSS with 0.1% BSA and 10 mM HEPES. Assays were performed in triplicate in 96-well chemotaxis plates. 20 μl of chemokines were loaded at varying concentrations into the lower well of the modified Boyden chamber (Neuroprobe) and fitted with a 5 μM filter. 1×10⁶ PBN in 30 μl were added to the upper well. The cells were incubated for 3 h at 37 °C. Migration of PBN was measured by direct count of cells that had migrated through the filter into the lower chamber. The chemotactic index was calculated as the number of cells that migrated in the presence of chemokine divided by the number of cells that migrated due to buffer only.

Apoptosis assays

PBN were resuspended in RPMI-1640 (Biowhittaker) with 1% FBS and added to a 96-well plate at 1×10⁶ cells/well. Chemokines were added to the wells at a final concentration of 100 nM. Plates were incubated at 37 °C, 5% CO₂ for 24 h. Cells were labeled with AnnexinV and propidium iodide (PI) using the Annexin V Kit (Caltag) according to the manufacturer’s instructions. Cells were analyzed by flow cytometry using FACS Calibur (Beckman-Coulter) within 1 h.

CD11 staining

PBN were resuspended in RPMI-1640 (Biowhittaker) with 1% FBS and added to a 96-well plate at 1×10⁶ cells/well. PBN were exposed to 100 nM of chemokines for 2 h at 37 °C and 5% CO₂. Cells were washed with PBS and blocked with 1% goat serum. PBN were incubated with fluorescently conjugated CD11a, CD11b, and CD11c antibodies (Caltag) on ice and fixed with 4% paraformaldehyde. Cells were analyzed by flow cytometry.