Report

A Broad-Spectrum Chemokine Antagonist Encoded by Kaposi's Sarcoma-Associated Herpesvirus

See allHide authors and affiliations

Science  12 Sep 1997:
Vol. 277, Issue 5332, pp. 1656-1659
DOI: 10.1126/science.277.5332.1656

Abstract

Kaposi's sarcoma–associated herpesvirus encodes a chemokine called vMIP-II. This protein displayed a broader spectrum of receptor activities than any mammalian chemokine as it bound with high affinity to a number of both CC and CXC chemokine receptors. Binding of vMIP-II, however, was not associated with the normal, rapid mobilization of calcium from intracellular stores; instead, it blocked calcium mobilization induced by endogenous chemokines. In freshly isolated human monocytes the virally encoded vMIP-II acted as a potent and efficient antagonist of chemotaxis induced by chemokines. Because vMIP-II could inhibit cell entry of human immunodeficiency virus (HIV) mediated through CCR3 and CCR5 as well as CXCR4, this protein may serve as a lead for development of broad-spectrum anti-HIV agents.

Accumulating evidence indicates that human herpesvirus 8 (HHV-8) is the infectious agent responsible for Kaposi's sarcoma in patients with and without HIV infection (1). Several viral proteins with homologous human counterparts involved in cellular signaling are encoded by HHV-8 (2). Three open reading frames of the HHV-8 genome code for chemokine-like proteins with 28 to 56% amino acid identity to each other and about 40% identity to human CC chemokines (3) (Fig. 1A). We cloned the vMIP-II chemokine by polymerase chain reaction (PCR) amplification from a biopsy of a Kaposi's sarcoma lesion in an HIV patient and found the nucleotide sequence was identical to the one in GenBank (4). The vMIP-II gene was expressed in COS-7 cells, and we tested the secreted, recombinant protein in binding assays with various human chemokine receptors and with the human cytomegalovirus-encoded US28 receptor (5). Conditioned medium containing recombinant vMIP-II displaced 125I-labeled MIP-1α (125I–MIP-1α) from the human CC chemokine receptors CCR1 and CCR5 and from the US28 receptor as well as radiolabeled MCP-1 from the human CCR2 receptor (Fig. 2A). Surprisingly, vMIP-II also bound to CXCR4 in competition with the labeled CXC chemokine SDF-1. However, vMIP-II showed low activity on the CXCR2 [interleukin-8B (IL-8B) receptor]. It did not interfere with binding of labeled IL-8 to CXCR1 (IL-8A receptor) or binding of the unrelated peptide substance P to its 7TM receptor NK-1. Conditioned medium from COS-7 cells transfected with the empty expression vector did not displace any of the radioactively labeled chemokines from the indicated receptors (Fig. 2A).

Figure 1

(A) Alignment of vMIP-II with various human CC chemokines (MIP-1α, MIP-1β, RANTES, and MCP-1) and human CXC chemokines (IL-8 and SDF-1) (32). The sequence of vMIP-II is shown on gray background and residues in the other chemokines that are identical to those of vMIP-II are indicated in white on black. Asterisks indicate conserved Cys residues. Only the sequences of the presumed mature, secreted proteins are shown. For vMIP-II NH2-terminal sequence analysis and mass spectroscopy of this study showed that a 23-residue signal peptide and a COOH-terminal Arg residue are removed from the virally encoded protein during synthesis and secretion in mammalian cells. (B) HPLC purification of recombinant vMIP-II expressed in COS-7 cells (solid line) monitored by testing aliquots for interference with 125I–MIP-1α binding to CCR5 expressed stably in HEK-293 cells (dashed line). A 214, absorbance at 214 nm; AU, absorbance units.

Figure 2

Competition binding experiments with recombinant vMIP-II in various chemokine receptors. (A) Whole-cell binding experiments were performed with recombinant vMIP-II in conditioned medium from transiently transfected COS-7 cells diluted 1:10 (closed bars) and 1:50 (hatched bars) in transiently transfected COS-7 cells expressing CCR1, CCR2, US28, or CXCR4; stably transfected HEK-293 cells expressing CCR5; or CHO cells stably expressing either CXCR1, CXCR2, or the 7TM receptor for the nonchemokine neuropeptide substance P (NK-1). Conditioned medium from mock-transfected COS-7 cells was used as a control (open bars). Below the name of each of the receptors the chemokine peptide used as radioactively labeled ligand is indicated. Inhibition of 100% was defined by homologous displacement. (B) Competition binding experiments with HPLC-purified vMIP-II on CCR5 and CXCR4 expressed stably in HEK-293 cells and on US28 expressed transiently in COS-7 cells. Values on ordinate represent percent of maximum bound.

The recombinant vMIP-II protein was purified by Bio-Gel P-30 gel filtration followed by reversed-phase high-pressure liquid chromatography (HPLC), and the chromatography was monitored by receptor analysis with human CCR5 stably transfected into HEK-293 cells (Fig.1B) (6). NH2-terminal sequence analysis of the first 10 residues of the recombinant vMIP-II demonstrated that a 23–amino acid signal peptide is removed from the precursor protein. Mass spectroscopy of the HPLC-purified vMIP-II gave a molecular mass of 7964, which indicates that a COOH-terminal Arg residue encoded by the open reading frame is also removed in the secretory pathway (theoretical molecular mass of the des-Arg protein is 7968). Thus, the vMIP-II sequence in Fig. 1A is likely to be the naturally occurring, secretory form of the viral protein (7).

In competition binding assays, HPLC-purified vMIP-II bound with high affinity to human CCR1 [median inhibitory concentration (IC50) = 8.0 ± 3.6 nM], CCR5 (IC50 = 5.2 ± 1.9 nM), and CXCR4 (IC50 = 5.8 ± 2.8 nM) receptors and with very high affinity to human CCR2 (IC50 = 0.82 ± 0.16 nM) and the cytomegalovirus-encoded US28 receptor (IC50 = 0.63 ± 0.20 nM) (binding curves for CCR5, US28, and CXCR4 are shown in Fig. 2B). Chemically synthesized vMIP-II (8) gave very similar results (data not shown).

Receptor binding of endogenous chemokines is normally associated with a rapid, pertussis toxin–sensitive calcium response. Instead, HPLC-purified recombinant vMIP-II induced a slow, prolonged calcium response in CHO (Fig. 3A, upper left) but not HEK-293 cells stably transfected with human chemokine receptors (data not shown) (9). In contrast to the endogenous chemokines, vMIP-II was unable to cause rapid mobilization of calcium from intracellular stores—that is, increase intracellular calcium in the presence of EDTA (Fig. 3A). Both the HPLC-purified recombinant peptide and vMIP-II prepared by chemical synthesis in a dose-dependent manner blocked the calcium mobilization elicited by the relevant human chemokines through CCR1, CCR2, CCR3, CCR5, and CXCR4 (Fig. 3A). In contrast, vMIP-II had no effect on the calcium response induced by IL-8 through CXCR2 (data not shown). Thus, in some cellular contexts vMIP-II appears as an agonist—that is, it elicits slow, prolonged leakage of extracellular calcium into the cell (10). But with respect to mobilization of intracellular calcium, vMIP-II is a pure antagonist against the action of endogenous chemokines. Activation of two distinct calcium signaling pathways through the same receptor has recently been shown for the mammalian chemokines IL-8 and GRO-α acting through CXCR2 (11). It has been demonstrated that the primary biological effect of chemokine chemotaxis is associated with the rapid, pertussis toxin–sensitive calcium response, in contrast to the slow, prolonged calcium response (12).

Figure 3

Effect of vMIP-II on [Ca2+]i in transfected cells and on chemotactic activity of human monocytes. (A) Traces of fluctuations in [Ca2+]i measured by Fura-2 fluorescence in CHO cells stably transfected with various chemokine receptors. All experiments except the one shown in the upper left were done with 10 mM EDTA in the extracellular medium to study calcium mobilization only from intracellular stores. (Upper left) Upper trace shows effect of 10−9 M vMIP-II followed by 10−7 M RANTES in cells expressing CCR5; lower trace shows an experiment performed under the same conditions but with 10 mM EDTA added to the medium. All other panels show the effect of pretreatment with vMIP-II (either a single dose of 10−7 M or a full dose-response series on the calcium mobilization induced by a submaximal dose of an appropriate endogenous human chemokine in cells expressing various chemokine receptors: CCR1, 10−10 M RANTES; CCR2, 10−7 M MCP-1; CCR3, 10−9 M MCP-3; CCR5, 10−9 M RANTES; CXCR4, 10−8.5 M SDF-1α). HPLC-purified recombinant vMIP-II and synthetic vMIP-II gave similar results; the experiments with CCR5 were done with recombinant material, and the rest were done with synthetic peptide. (B) Chemotaxis of freshly isolated human monocytes. (Upper) Dose-response curve for vMIP-II (n = 8) given alone. A single dose-response curve for RANTES is shown for comparison. (Lower) Dose-response curves for inhibition by vMIP-II of chemotaxis induced by 10−8 M RANTES (IC50 for vMIP-II = 3.1 nM;n = 5), MIP-1α (IC50 = 2.7 nM; n= 2), or MIP-1β (IC50 = 6.2 nM; n = 2).

The chemotactic property of vMIP-II has been probed in freshly prepared human monocytes (13). As expected from its lack of ability to cause rapid mobilization of intracellular calcium and in contrast to the endogenous chemokine RANTES, vMIP-II when given alone did not stimulate chemotaxis in the monocytes (Fig. 3B, upper). However, vMIP-II was an efficient and potent inhibitor of the chemotactic response to RANTES, MIP-1α, or MIP-1β in these cells (Fig. 3B, lower). Thus, it can be envisioned that vMIP-II is used by HHV-8 to block recruitment of leukocytes as part of the viral defense mechanism against the immune system of its host. Recently, it was suggested but not demonstrated that two other putative chemokines discovered as open reading frames in the genome of the human molluscum contagiosum poxvirus and a murine cytomegalovirus may act as anti-inflammatory agents by blocking chemokine receptors (14).

Although chemokines are often able to bind with high affinity to more than one receptor subtype, no known human chemokine has as broad a spectrum of activities as vMIP-II (15). RANTES and MCP-3, which bind to multiple CC chemokine receptors, are probably the most promiscuous of the human chemokines. But these CC chemokines do not bind with high affinity to any CXC chemokine receptor. Conversely, CXC chemokines such as IL-8 have negligible affinity for CC chemokine receptors. Nevertheless, despite relatively low levels of primary sequence identity, the monomeric forms of CC and CXC chemokines have very similar three-dimensional structures (16). Mutation and chemical modification of a single amino acid residue in IL-8 can introduce high affinity for this CXC chemokine on the CCR1 receptor where the affinity on its natural CXCR1 and -2 receptors is decreased two orders of magnitude (17). In the case of vMIP-II, this peptide appears to have been optimized by the virus to bind to and block multiple human chemokine receptor subtypes. HHV-8 also encodes a chemokine receptor called ORF74 (18). Although ORF74 binds IL-8 with high affinity, this virally encoded receptor is in fact also surprisingly promiscuous, as it binds both CXC and CC chemokines (19)—just as vMIP-II acts as a ligand for both CC and CXC chemokine receptors.

Although CCR5 and CXCR4 are the main co-receptors for HIV cell entry (20), CCR3 also appears to be important for HIV infection of microglia (21). Because vMIP-II binds to all these receptors (22), its ability to block HIV cell entry was tested (23). The 89.6 strain of HIV-1 was used as a convenient probe as it is able to exploit multiple chemokine receptors for cell entry (24). In U87/CD4 cells, vMIP-II could block infection of strain 89.6 mediated through CCR3, CCR5, or CXCR4 (Fig.4). On CCR5 and CXCR4, vMIP-II was less potent but equally efficient (at higher doses) compared with RANTES and SDF-1, respectively, whereas on CCR3, vMIP-II was highly potent and even more efficient than RANTES in blocking infection with strain 89.6 (Fig. 4). Against the SL-2 strain of HIV-1, which selectively uses CCR5 as co-receptor (25), vMIP-II and especially RANTES appeared less potent, whereas the semisynthetic analog AOP-RANTES as previously reported blocked infection very potently (26) (Fig. 4). Similar results were obtained with SF-162, a second NSI strain that uses CCR5 exclusively as co-receptor (data not shown). In peripheral blood mononuclear cells stimulated with phytohemagglutinin (PHA) and IL-2, vMIP-II at 200 nM reduced infectivity of SL-2 and SF-162 by a maximum of 50 to 75%, whereas RANTES blocked infectivity >98%. In this system, no significant reduction in infectivity with strain 89.6 was observed with either vMIP-II or RANTES (data not shown). Recently it was shown by cotransfection with CCR5 in CD4+ cells that one of the other HHV-8–encoded chemokines, vMIP-I, can also inhibit transmission of CCR5-dependent HIV-1 strains (2). The binding profile of at least vMIP-II, including several co-receptors for HIV cell entry, could make this protein a possible lead for development of broad-spectrum therapeutic agents against HIV infection.

Figure 4

Inhibition of HIV-1 infection of US87/CD4 cells stably expressing CCR5, CCR3, and CXCR4 with vMIP-II (•), RANTES (□), SDF-1α (○), and AOP-RANTES (▵). The 89.6 strain of HIV-1 can use all three chemokine receptors as co-receptors for cell entry (A to C), whereas the SL-2 strain selectively uses CCR5 (D). Nanomolar concentrations of chemokines are indicated for direct comparison with receptor binding and signal transduction studies (100 nM ∼ 800 ng/ml).

Conceivably, the original mammalian counterpart of the viral MIP was obtained by HHV-8 through an act of molecular piracy, as has been described for other herpesviruses (27). Genetic combinatorial chemistry combined with phenotypic selection has optimized the mammalian chemokines to benefit the virus—in analogy with methods used in pharmaceutical drug development. In the case of vMIP-II, it will be interesting to determine the structural basis for the promiscuous receptor recognition as well as the chemical modifications the virus has exploited to transform the protein into an antagonist.

  • * These authors contributed equally to this investigation.

  • To whom correspondence should be addressed. E-mail: schwartz{at}molpharm.dk or tncw5312{at}ggr.co.uk

REFERENCES AND NOTES

View Abstract

Navigate This Article