Technical Comments

β-Chemokine MDC and HIV-1 Infection

Science  24 Jul 1998:
Vol. 281, Issue 5376, pp. 487
DOI: 10.1126/science.281.5376.487a

The human immunodeficiency virus type–1 (HIV-1) uses the chemokine receptors CCR5 or CXCR4 (in conjunction with CD4) to enter healthy cells, and the chemokine ligands to these receptors generally block virus infection (1–3). R. Pal et al. studied a truncated form of the β-chemokine macrophage-derived chemokine (MDC), which lacked the first two NH2-terminal amino acids [(−2)MDC] that they purified from an immortalized CD8+ T-cell clone. They found that (−2)MDC inhibited both R5 and X4 HIV-1 strains in vitro (4). The effects of full-length MDC, a CCR4 ligand (5), on HIV-1 infection have not been described.

To investigate the mechanism by which a chemokine to CCR4 [which is not an HIV-1 coreceptor (6, 7)] could inhibit infection by diverse HIV-1 isolates, we prepared MDC with the use of a bacterial expression system that we have used successfully to produce other anti-viral chemokines. MDC produced in this way was a potent CCR4-ligand, which induced a calcium flux in cells that transiently express CCR4 (Fig. 1A), naturally CCR4-positive cells such as Jurkat (Fig. 1B) and UT-7 (8), and activated peripheral blood mononuclear cells (PBMCs) (Fig. 1, C and D). MDC did not stimulate a calcium flux when added to cells that transiently express other chemokine or orphan receptors, including CCR1, -2, -3, -5, or -8, CXCR2, CXCR4, GPR1, GPR15, and STRL33 (8). Further, addition of MDC to PBMCs desensitized their subsequent response to the CCR4 ligand TARC (Fig. 1D). MDC did not desensitize subsequent CCR5 or CXCR4 responses, which suggests that it does not down-regulate these receptors (Fig. 1, B and C). MDC was also a potent chemoattractant, as previously described (9). By contrast, recombinant (−2)MDC did not induce a Ca flux in multiple cell types that express CCR4 or other receptors, including PBMCs (Fig. 1, A to C), and it was 1000-fold less effective than full-length MDC in chemotaxis assays. We also found that synthetic MDC, while biologically active, lacked anti-viral properties. We conclude that recombinant (−2)MDC, produced by the same method used to generate biologically active MDC (and also other chemokines such as RANTES and SDF-1), is neither a CCR4 ligand or antagonist, nor an agonist for the relatively large number of chemokine and orphan receptors tested or for receptors present on PBMCs.

Figure 1

Cellular calcium response to MDC and (−2)MDC. Approximately 1 to 3 × 106cells were loaded with 2.5 μM Fura-2/AM (Molecular Probes) in the dark at 37°C for 1 to 2 hours. Cells were washed and resuspended in 1.5 ml colorless RPMI supplemented with 2% FBS. Ca2+mobilization was measured with the use of an Aminco-Bowman luminescence spectrometer in a constantly stirring cuvette. Excitation of loaded cells was monitored at 340 nm and 380 nm, and intracellular calcium concentration was calculated as previously described (11). MDC, but not (−2)MDC, mobilized calcium in (A) 293T cells transiently transfected with CCR4 under the control of the CMV promoter, (B) Jurkat cells, and (C) PBMCs that were stimulated with OKT3 and maintained in IL-2 (100 U/ml). In addition, treatment with MDC desensitized PBMCs with respect to further MDC stimulation as well as stimulation with TARC, another CCR4 ligand (D), but did not desensitize CCR5 or CXCR4. All chemokines were used at 500 ng/ml except for (−2)MDC in (B), which was used at 5 μg/ml. Chemokines were expressed as fusion proteins with a 30-kDa leader sequence in E. coli. (The recombinant fusion proteins are predominantly found as insoluble proteins in inclusion bodies). Subsequent to solubilization and refolding, chemokines were cleaved from the fusion protein by factor Xa treatment. Final purification of the chemokines was achieved by sequential column chromatography using ion exchange and reverse-phase HPLC columns.

Even high concentrations of MDC did not inhibit productive infection of PBMCs (Fig. 2A), or macrophages (8) by X4, R5, or R5X4 HIV-1 strains. MDC also did not inhibit virus entry into PBMCs as determined by a PCR entry assay; also, it did not inhibit infection of multiple cell lines by pseudotyped reporter viruses bearing X4, R5, or R5X4 Env proteins, including cell lines that express CCR4 in addition to coreceptors such as CXCR4 and CCR5 (8). MDC did not inhibit fusion of cells expressing either X4 or R5 Env proteins with PHA- or PHA+IL-2–activated PBMCs (8), PHA-activated primary CD4+ T-cells (Fig. 2B), macrophages (Fig. 2B), PM-1, or UT7 cells (8) with the use of a well-characterized gene reporter fusion assay (10). By contrast, SDF-1 inhibited fusion by X4 Env proteins and RANTES inhibited fusion by R5 Env proteins. Recombinant (−2)MDC also did not inhibit infection of PBMCs by X4, R5, or R5X4 virus strains, whereas AOP-RANTES inhibited CCR5–dependent virus infection in the same cells (Fig. 2A).

Figure 2

Virus infection and cell-cell fusion assays. (A) PBMC were isolated by Ficoll hypaque and stimulated with 5μg/ml PHA for 72 hours, and subsequently expanded in the presence of 50 U/ml IL-2 for 48 hours before infection by HIV-1 IIIB [multiplicity of infection (m.o.i.) 0.2], HIV-1 Ba-L (m.o.i. 0.2), or HIV-1 89.6 (m.o.i. 0.2) in the presence of 50 U/ml IL-2 alone (that is, “None”) or in combination with 500 ng/ml MDC (donor shown on the left), 500 ng/ml (−2)MDC (donor shown on the right) or 100 ng/ml AOP-RANTES (both donors). Cultures were sampled every 3 days, at which time medium with cytokines, or chemokines, or both was replenished. Reverse transcriptase (RT) activity within supernatants was determined against a standard curve of recombinant HIV-1 BH10RT (NIH AIDS Reference Reagent Program). Data is shown for day 10 after infection as picograms of RT activity per 0.01 mls. Similar results were obtained with either MDC or (−2)MDC with the use of the same infection protocol with cells from different donors (B). To measure cell-cell fusion, HeLa cells were infected with recombinant vaccinia viruses expressing the desired HIV-1 Env protein, as well as with vCB21R, a recombinant virus containing the E. coli lacZ gene under the control of the T7 promoter. HeLa cells were mixed with either macrophages or CD4+ T cells. Both cell types were infected with a recombinant vaccinia virus expressing T7 RNA polymerase. Macrophages were prepared from elutriated monocytes by 14 days of differentiation. Purified CD4+ T cells were obtained from the NIH Blood Bank and cultured for 72 hours in the presence of PHA-L (Boehringer, 5 μg/ml). Indicated chemokines at a final concentration of 200 nM were added to the target cells at room temperature 20 min before the addition of Env-expressing HeLa cells. Fusion was determined 2 hours after cell-cell mixing by measuring β-galactosidase activity. Results for the macrophage fusion experiment were divided by two so as to keep the scale similar to that obtained with the CD4+ T cells.

MDC has not been shown to interact with any known HIV coreceptor, which suggests that any anti-viral activity it may have is likely to be indirect, perhaps as a result of intracellular signaling. The recombinant MDC we produced, however, did not inhibit any virus strain tested, despite being fully capable of signaling through CCR4, nor did it affect CXCR4 or CCR5. The recombinant (−2)MDC was also inactive against HIV-1 and did not induce signals in PBMCs or by any receptor we tested it against in cell lines. The reasons underlying the discrepancies between our results and those of Pal et al. (4) are not clear. In our study, we used recombinant chemokines, whereas Pal et al. purified (−2)MDC from CD8+ cell-conditioned medium. Each of these approaches has potential weaknesses. In principle, our recombinant (−2)MDC might fold or be processed differently from (−2)MDC made in CD8+cells, preventing it from having activity against HIV-1. Recombinant RANTES and SDF-1 made by the same method, however, retain antiviral activity, and our recombinant MDC was a highly effective CCR4 agonist. Conversely, (−2)MDC isolated from conditioned medium may not have the purity of a recombinant protein, raising the possibility that the antiviral activity noted by Pal et al. in their (−2)MDC preparation might be attributable to an undetected contaminant. Further studies might resolve these issues.


Table 21

Isoforms of native MDC.

View this table:

β-Chemokine MDC and HIV-1 Infection

Supernatants of activated CD8+ T lymphocytes suppress HIV without cell lysis. The CC chemokines RANTES, MIP-1α, and MIP-1β, are known to prevent the entry into healthy cells and consequent replication of CCR5-dependent (R5) HIV strains. (1). Other factors, however, are thought to be involved, because residual suppressive activity remains after blockade of these chemokines with selective antibodies. Moreover, the presence of these chemokines does not explain the capacity of the CD8+ T cell supernatants, which do not contain SDF-1 (2), to inhibit entry and replication of CXCR4-dependent (X4) HIV strains.

R. Pal et al. report (3) a CD8+T lymphocyte–derived factor that appeared to suppress X4 or R5 HIV isolates; they found that this factor corresponded to a truncated form of MDC, MDC(3-69), which lacked two NH2-terminal residues (3). MDC, a CC chemokine with the NH2-terminal sequence gpyganmedsvcc… was recently isolated from human macrophages (4). We have synthesized wild-type MDC and MDC(3-69) and have tested their activities in comparison with that of TARC, which, like MDC, is selective of the CC chemokine receptor 4, CCR4 (5). MDC- and TARC-induced chemotaxis and Ca2+ mobilization in cells bearing CCR4 and showed cross-desensitization (Fig. 1). In contrast, MDC(3-69) was inactive and did not desensitize or block CCR4. Furthermore, MDC did not cause down-regulation of CCR5 or CXCR4, in contrast to their respective ligands, MIP-1β and SDF-1. HIV suppression in activated, CD8+ T lymphocyte–depleted, PBMCs was also tested. MDC and MDC(3-69), as well as TARC, had no activity, whereas marked suppression of either X4 or R5 strains was obtained with SDF-1 and MIP-1β, respectively. As with single tropic X4 or R5 viruses, MDC(3-69) also did not prevent replication of dual tropic R5X4 HIV isolates. Our results show that MDC, MDC(3-69), and TARC lack suppressive activity against CCR5 or CXCR4-dependent HIV strains, or both.

Figure 1

Migration and calcium mobilization in response to MDC, MDC(3-69), and TARC, and lack of HIV-suppressive activity. (A) Chemotaxis of murine 300-19 pre-B cells stably transfected with CCR4 induced by MDC (•), MDC(3-69) (○), and TARC (▴). Numbers of migrated cells ± SD per five high-power fields in triplicate assays are shown. Parental cells did not respond. (B) Receptor cross-desensitization. Fura-2 loaded 300-19 cells that express CCR4 were sequentially stimulated with 100 nM MDC or MDC(3-69), followed by 100 nM TARC or MDC (▴), and [Ca2+]i-dependent fluorescence changes were recorded. The same results were obtained when MDC(3-69) was used at 1,000 nM. (C) Chemotaxis of CEM cells induced by MDC (•), MDC(3-69) (○), and TARC (▴). For methods used in A to C, see (7). (D) Downregulation of CXCR4 and CCR5 in phytohemagglutinin (PHA)-and IL2-induced blood CD4+ T cells. Surface expression (relative units) of the receptors is shown on induction with SDF-1, MIP-1β, MDC, or MDC(3-69), as previously described (8). CXCR4 and CCR5 were detected with the use of the monoclonal antibodies 12G5 and 2D5, respectively (NIH AIDS Repository Reagent Program). (E) Effect of 200 nM SDF-1, MIP-1β, MDC, MDC(3-69), or TARC on HIV replication. 2×106 PHA- and IL2-activated PBMCs depleted of CD8+ T lymphocytes (antibody to CD8 attached to the surface of magnetic beads) were infected with 100 to 200 TCID50 of X4 (NL4-3), R5 (BaL and SF162), or R5X4 (SF2) HIV isolates. Chemokines were added before infection and maintained throughout the culture period. Culture supernatants were collected every 3 days, and medium changes incorporated 200 nM of chemokines. Viral replication was assessed by measuring soluble HIV p24gag antigen. Results corresponding to day 10 after infection are shown. All the isolates were tested at least twice with the use of different healthy blood donors. Representative results are shown.

We find no evidence for an interaction of MDC or MDC(3-69) with either CXCR4 or CCR5, and for the HIV isolates tested we confirm previous observations showing that CCR4, the selective receptor for MDC and TARC, does not function as an HIV coreceptor (6). These results do not support the notion that MDC or MDC(3-69) is responsible for HIV inhibitory activity of CD8 T lymphocyte supernatants.


β-Chemokine MDC and HIV-1 Infection

Response: In our report (1), we identified an HIV-suppressive activity, produced by an HTLV-transformed cell line, that blocked a variety of HIV-1 isolates (some of these isolates induce syncytia, some do not). The HIV-suppressive activity was associated with a protein that was determined by sequence analyses to be the β chemokine MDC (1). In more recent experiments, we tested recombinant or synthetic full-length MDC (GPYGANM…) as well as forms missing two NH2-terminal residues (−2; YGANM…) and found that all lacked antiviral activity (2). The truncated forms also did not stimulate cytoplasmic calcium mobilization in PBMCs (2). Thus, with respect to recombinant and synthetic MDC, our findings are in agreement with those of the other groups. Our studies indicate, however, that one or more of the native truncated forms are associated with both types of activity.

In our report, NH2-terminal amino acid sequencing of a native MDC (1) preparation showed a mixture of polypeptides that were missing one (−1; PYGANM…) or two NH2- terminal residues relative to the sequence originally described (3). Additional sequencing of the preparation indicated the presence of another form (ANM…) missing four (−4) residues. These findings were confirmed and extended by mass spectrometry performed on a subsequent preparation with HIV suppressive activity (4). The only molecules we detected (Table 1) had molecular masses matching that predicted for the −2 and −4 forms as well as a minor form consistent with MDC missing the first eight NH2-terminal residues (−8; DVSCC…). We did not detect the −8 form by NH2-terminal sequence analyses of the earlier preparation, indicating that recovery of this polypeptide may be variable. The same may be the case for the −1 species, which was not detected by mass spectrometry in the later preparation. In no case have we detected full-length MDC. In view of recent studies, published after our report (5–9), it now seems more likely that the various polypeptides are MDC isoforms generated by natural mechanisms of posttranslational processing. Thus, the maturation of MDC may be analogous to native RANTES (6), MCP-1, and MCP-2 (8), which are modified by proteolysis of their NH2-termini. In the case of RANTES, such processing greatly reduces its activity with CCR1 and CCR3, but not with CCR5 (5,6) and enhances its antiviral activity (7). Thus, NH2-terminal modification can induce shifts in receptor binding and activation profiles. Therefore, it is important to understand how such truncations might affect the various activities of MDC.

As we showed earlier, the collection of MDC isoforms consistently stimulates cytoplasmic calcium mobilization in activated PBMCs after long- or short-term culture (1). The activity is equivalent to that which is seen with the same concentration of synthetic full-length MDC (Fig. 1A), which strongly indicates that the native MDC isoforms, or one of them, is functional. The alternative explanation, which seems less likely, is the existence of a contaminant, undetected by sequencing or mass spectrometry, that triggers G-protein–coupled receptors to the same level as synthetic MDC. However, a monoclonal antibody against MDC that blocks signaling inhibited the induction of calcium mobilization by native MDC (Fig. 1B), although it did not directly block the antiviral effect (2). These results indicate that the signaling activity in our preparations is not a result of a contaminant that induces calcium mobilization, but is inconclusive with respect to antiviral activity. One explanation for the differential inhibition is that the monoclonal antibody does not recognize the antiviral domain or isoform. Further experiments with additional monoclonal and polyclonal antibodies might resolve this issue.

Figure 1

Intracellular Ca2+ mobilization induced by native and synthetic MDC. (A) Primary PBMC were activated for 72 hours in 5 μg/ml phytohemagglutinin (PHA) and 20 U/ml recombinant human IL-2 (rIL-2; Boehringer-Mannheim, Indianapolis, Indiana) and then cultured in rIL-2. Cells were treated with either synthetic or native MDC (33 nM each) and assayed as described previously (1). Data were acquired by a FACSCalibur (Becton-Dickinson, San Jose, California) flow cytometer, gating cells by forward and side scatter properties. Ca2+ mobilization was determined by analysis in a two-parameter density plot of linear emission at 530 nm collected in the FL-1 window over time. Time units are equal to 0.1 seconds. (B) Assay was repeated with the use of synthetic or native MDC (3 nM each) that was untreated or treated with a five-fold molar excess of monoclonal antibody against MDC for 1 hour at 4°C. Time units shown are equal to 0.05 seconds.

The major question now centers on which molecule in the native MDC preparations is responsible for the antiviral effect. On the basis of available evidence, it seems reasonable to eliminate full-length MDC from consideration as the major antiviral species because it is functional with respect to receptor activation. It is unclear, however, whether the −2 isoform can be characterized as inactive. Given our results with native material, at this time we cannot conclude that the −2 isoform is naturally inert. It seems equally possible that the −2 MDC isoform is at least partially functional, but prone to misfold and lose activity under certain conditions. Thus, it may not be unexpected that such an inert form of −2 MDC lacks antiviral activity. We also cannot conclude that the −2 isoform does not bind known HIV coreceptors because one commercial preparation (Peprotech, Rocky Hill, New Jersey) competes with MIP-1α for binding to CCR5, albeit at high concentrations (IC50 for MDC 202 nM). These results (11) are not necessarily at odds with those of Arenzana-Seisdados et al. and Lee et al., because it is possible that the isoform exhibits promiscuous but nonfunctional receptor interactions that may not have been detectable in their assay systems. Such interactions were demonstrated for the human herpesvirus 8 chemokine analog, vMIP-II (10, 12). Promiscuous receptor binding is consistent with the broad antiviral effect we observed; however, material we obtained from this source before commercialization was not antiviral in our assays (2). We are currently evaluating more recent preparations of this material in order to address this issue.

It is also possible that the activities we observe with the native MDC are a result of other isoforms we have not yet analyzed separately. Finally, although there is no conclusive evidence, it is possible that MDC may be closely associated with a cryptic factor that is the determinant for antiviral activity.

In summary, we agree that no recombinant form of MDC has yet shown antiviral activity, while the native truncated forms show both antiviral activity and the capacity to stimulate cytoplasmic calcium mobilization. Thus, the −4 and −8 forms, and even the −2 form, cannot be definitively excluded as the active molecule or molecules. Further experimentation is required to resolve these issues.


Navigate This Article