LD78β, A Non-allelic Variant of Human MIP-1α (LD78α), Has Enhanced Receptor Interactions and Potent HIV Suppressive Activity*

Chemokines play diverse roles in inflammatory and non-inflammatory situations via activation of heptahelical G-protein-coupled receptors. Also, many chemokine receptors can act as cofactors for cellular entry of human immunodeficiency virus (HIV)in vitro. CCR5, a receptor for chemokines MIP-1α (LD78α), MIP-1β, RANTES, and MCP2, is of particular importancein vivo as polymorphisms in this gene affect HIV infection and rate of progression to AIDS. Moreover, the CCR5 ligands can prevent HIV entry through this receptor and likely contribute to the control of HIV infection. Here we show that a non-allelic isoform of human MIP-1α (LD78α), termed LD78β or MIP-1αP, has enhanced receptor binding affinities to CCR5 (∼6-fold) and the promiscuous β-chemokine receptor, D6 (∼15–20-fold). We demonstrate that a proline residue at position 2 of MIP-1αP is responsible for this enhanced activity. Moreover, MIP-1αP is by far the most potent natural CCR5 agonist described to date, and importantly, displays markedly higher HIV1 suppressive activity than all other human MIP-1α isoforms examined. In addition, while RANTES has been described as the most potent inhibitor of CCR5-mediated HIV entry, MIP-1αP was as potent as, if not more potent than, RANTES in HIV-1 suppressive assays. This property suggests that MIP-1αP may be of importance in controlling viral spread in HIV-infected individuals.

Chemokines are a family of structurally and functionally related proteins that play a central role in the regulation of hemopoietic cell migration during development, immune surveillance, and the establishment of inflammatory and immune responses (1,2). These biological effects are mediated through a family of cell surface G-protein-coupled heptahelical receptors (3). Recently, these receptors have been implicated in HIV 1 pathogenesis with the discovery that a large number of the members of this family are able to act as cofactors for the entry of HIV into cells in vitro (for reviews, see Refs. 4 and 5). The ligands for these receptors are able to compete for binding with the HIV envelope protein to abrogate viral entry (6 -8). While an in vivo role for many of these receptors has yet to be demonstrated in the context of HIV infection, CCR5, a receptor for RANTES (regulated on activation, normal T cell expressed and secreted), MCP2 (monocyte chemotactic protein-2), and MIP-1␣ and -␤ (macrophage inflammatory protein-1␣ and -1␤) appears to be critical in HIV pathogenesis as individuals who are homozygous or heterozygous for a CCR5 null mutation are relatively resistant to HIV infection and progression to AIDS, respectively (9 -11). Moreover, the CCR5 ligands are produced by activated HIV-specific cytotoxic and helper T cells, whereby they are likely to play an important role in controlling viral spread (12)(13)(14)(15)(16). Current data show RANTES to be the most effective antagonist of CCR5-mediated HIV entry, with MCP2 and MIP-1␣ and -1␤ being less effective (17)(18)(19).
We have been interested in characterizing receptors mediating the biological effects of MIP-1␣, a member of the ␤ subfamily of chemokines able to induce chemotaxis of many mature leukocyte types (1, 2) and a potent inhibitor of hemopoietic stem cell proliferation (20,21). Mice lacking this gene display dramatic alterations in responses to several infectious agents (22) likely due to leukocyte chemotaxis abrogation. We have recently cloned several receptors for this ␤ chemokine from mouse and human sources (23,24) and found a number of discrepancies with respect to their interaction with murine MIP-1␣ and its presumed human homologue, LD78␣. Thus, while the murine and human forms of the promiscuous ␤-chemokine receptor D6 (23,24) bind murine MIP-1␣ with high affinity, LD78␣ interacts poorly with these receptors. Similarly, murine CCR5 binds murine MIP-1␣ with high affinity but does not recognize the putative human homologue.
The basis for this marked selectivity has not previously been addressed, however, it is notable that, while the murine genome contains only one copy of MIP-1␣ and -␤, in humans these two genes have been duplicated and mutated to produce two different non-allelic isoforms which still retain Ͼ90% homology (25,26). 2 For MIP-1␣, these have been called LD78␣ and -␤, with LD78␣ being the predominant experimentally used isoform and the one shown to interact poorly with the cloned chemokine receptors. Both forms are transcribed (26) and can be secreted from mammalian cells (6,21,27,28). Interestingly, consideration of differences in the putative signal sequences of the two LD78 isoforms by predictive algorithms, 2 suggests that there is variation in the site of signal peptidase cleavage between the two isoforms during secretion, and that LD78␣ is in fact produced without the four anticipated amino-terminal amino acids, ASLA (Fig. 1). Indeed, human MIP-1␣ has been shown to be produced naturally by CD8ϩ T cells as a Ϫ4 variant with the amino-terminal sequence ADTPT (6,27). We have worked extensively with the LD78␤ isoform (21,28), and numerous amino-terminal sequencing exercises have consistently revealed a "full-length" amino terminus APLAADTPT.
Using full-length and truncated variants of LD78␣ and -␤, we show here that the murine/human MIP-1␣ binding discrepancies are resolved by studying the properties of LD78␤ which consistently behaves more like murine MIP-1␣ than does LD78␣. It appears therefore that LD78␤ more accurately represents the functional human homologue of murine MIP-1␣ and that LD78␣ should be considered to be a related but functional distinct chemokine. Importantly, we also show that LD78␤ is the most potent natural CCR5 agonist described to date. Furthermore, LD78␤ exhibits a much greater ability to antagonize HIV entry through CCR5 than other forms of human MIP-1␣. In fact, LD78␤ is consistently better at HIV1 antagonism than RANTES, previously described as the most potent CCR5-dependant HIV1 entry inhibitor. We demonstrate that the enhanced activity of LD78␤ is solely due to the presence of a proline residue at position 2 of the mature protein. We propose renaming LD78␣ and -␤ to MIP-1␣S and MIP-1␣P, respectively, to reflect the importance of this residue in the functional differences between these two proteins.

EXPERIMENTAL PROCEDURES
Chemokines-MIP-1␣S is "human MIP-1␣" purchased from Peprotech, London, United Kingdom. It is derived from the LD78␣ cDNA and has the amino-terminal sequence of ASLAADTPT. MIP-1␣S-4 is human MIP-1␣ purchased from R&D Systems, Abingdon, UK. It is derived from the LD78␣ cDNA and starts as ADTPT. MIP-1␣P is derived from the LD78␤ cDNA and was prepared as described previously (28). Sequencing revealed APLAADTPT at the amino-terminal. MIP-1␣P-4 was produced from a modified LD78␤ cDNA in bacteria 3 with the amino terminus of ADTPT.
Receptor Binding Studies-CHO cells expressing chemokine receptors were prepared, and binding assays performed, as outlined previously (23). In short, binding assays were performed using 125 I-labeled murine MIP-1␣ at a constant concentration of 600 pM (for D6), 3 nM (for CCR1), or 9 nM (for CCR5), while varying the concentration of unlabeled human MIP-1␣ competitor protein. Remaining radioactivity bound after 90 min, and three ice-cold phosphate-buffered saline washes, was determined. Each point was done in triplicate, the average taken, and converted into a percentage of radioactivity bound in the absence of any unlabeled competitor chemokine. Data were analyzed using the LI-GAND software (29).
Ca 2ϩ Flux Assays-HEK 293 cells stably expressing human receptors CCR1 and CCR5 were derived by subcloning the cDNAs into pcDNA3 and transfection using Transfectam (Promega, Southampton, UK) according to the manufacturer's methods. Stably transfected cells were selected in 0.8 mg/ml G418. To detect ligand-induced calcium ion fluxes, cells were loaded with Fura-2-AM, then, ϳ6 ϫ 10 6 cells were incubated at 37°C in a continuously stirred cuvette in a Perkin-Elmer LS50 Spectrometer (340 nm ( ex ); 500 nm ( em )) and fluorescence emission recorded every 100 ms. After 2 min, ligand was added to a defined concentration and fluorescence recorded every 100 ms for a further 2 min. To control for day-to-day experimental variation, a full doseresponse curve for LD78␤ was performed each time a different ligand was tested.
HIV Entry Assays-5 ϫ 10 4 CEMx174-CCR5 cells, pretreated with chemokine for 30 min, were incubated for 4 h with luciferase virus (5 ng of p24), pseudotyped by either JR.FL or ADA according to a previously described protocol (17). Medium was then changed and the luciferase activity measured 3 to 5 days post-infection. The extent of inhibition of HIV entry was determined by comparing luciferase activity of chemokine-treated cells, with untreated cells. For studies with the replicationcompetent SF162 virus, protocols were similar to those described elsewhere (30). Briefly, CEMx174-CCR5 cells, or PHA/IL-2-activated PBMC from HIV1 seronegative donors, were pretreated with chemokine then infected overnight at 37°C with SF162 virus (m.o.i. ϳ0.04). The medium was then changed, and the p24 concentration subsequently determined after 5 days at 37°C using enzyme-linked immunosorbent assay. Percentage inhibition of HIV entry was calculated relative to chemokine-untreated controls.
Statistical Methods-Analysis of the dose-response relationship between the MIP-1␣ isoforms and other chemokines in binding, calcium flux, and HIV suppressive assays were tested by "log likelihood" methodology essentially as described previously (31).

RESULTS
In our recent studies of MIP-1␣ receptors (23,24) that revealed discrepancies between the binding of murine and human MIP-1␣, we used the LD78␣ isoform of human MIP-1␣ with the amino acids ASLAADTPT at the amino terminus ( Fig.  1). The close sequence similarity between LD78␣ and murine MIP-1␣ led us to examine the possible amino acid residues that may be responsible for the discriminatory binding of these ligands. The only consistent feature present in the mCCR5 and hD6 ligands but absent in human LD78␣ is a proline residue at position 2 of the mature protein (Fig. 1). The serine residue at position 2 of the LD78␣ we have used may therefore be preventing optimal receptor interaction. The alternative human MIP-1␣ isoform, LD78␤, however, does have a proline residue in position 2, with two reciprocal serine/glycine swaps in the region between cysteines 3 and 4 being the only other differences between the LD78␣ and -␤ proteins ( Fig. 1) (25,26). To simplify the nomenclature, and emphasize the serine/proline residue difference at position 2 in the two isoforms, we henceforth refer to LD78␣ and -␤ as MIP-1␣S and MIP-1␣P, respectively.
MIP-1␣P Interacts with High Affinity with Both Murine and Human CCR5 and D6 -To test the importance of these isoform differences, we have now compared the ability of MIP-1␣P and MIP-1␣S to interact with known murine MIP-1␣ receptors stably expressed on CHO cells (Table I). In addition, we have tested the Ϫ4 presumed naturally secreted form of MIP-1␣S FIG. 1. Alignment of the predicted protein sequences of mature ␤-chemokines. Human and murine sequences are prefixed with h and m, respectively. The second residue of each protein is enlarged, enboldened, and underlined. Ability to bind to human D6 and murine CCR5 is indicated with a "؉" to the right of the alignment. Non-binding chemokines are labeled with a "Ϫ." Binding information is from Refs. 23  . Murine MIP-1␣, which carries a proline residue at position 2 of the mature protein, binds to murine CCR5 in a manner similar to that seen with MIP-1␣P, suggesting that this human variant behaves more like the murine MIP-1␣ on CCR5 than does the more common MIP-1␣S isoform.
Similarly, the presence of the proline residue at position 2 of MIP-1␣P allowed it to bind with high affinity to murine D6 (K d 5.5 nM) in contrast to the other three isoforms tested (K d Ͼ 200 nM). Again the comparison with murine MIP-1␣ indicates that the MIP-1␣P isoform binds to D6 with an affinity that more resembles murine MIP-1␣ than does any of the other human variants tested.
In contrast to the data obtained with murine CCR5 and D6, studies with murine CCR1 expressing cells indicate that the proline at position 2 is not essential for high affinity binding of the chemokines to this receptor, indeed, the K d observed for the interaction of full-length proteins with murine CCR1 was markedly higher than that seen with the Ϫ4 truncated isoforms. Additionally, MIP-1␣P demonstrates a substantially higher binding affinity to that seen with murine MIP-1␣ indicating that the apparent equivalence of MIP-1␣P and muMIP-1␣ on murine CCR5 and D6 does not extend to muCCR1.
CHO cells expressing human receptors were also tested and showed similar results ( Fig. 2 and Table II). Thus, with human D6, while the MIP-1␣P isoform binds with high affinity (K d 5.5 nM), the Ϫ4 variants of the two isoforms bind only weakly (K d 77 nM for MIP-1␣S-4 and K d 124 nM for MIP-1␣P-4) and MIP-1␣S does not exhibit an enhanced binding affinity over its Ϫ4 variant ( Fig. 2A). Again, as seen with the murine receptors, the K d of MIP-1␣P for D6 is more similar to that of murine MIP-1␣ than any of the other isoforms tested (Table II). The differences between MIP-1␣S and MIP-1␣P are, however, somewhat less stark than those seen with the murine receptor. With human CCR5, there is an approximately 4 -6-fold higher binding affinity of MIP-1␣P (K d 6.2 nM) than is seen for the Ϫ4 MIP-1␣ proteins (K d 25 and 36 nM), and the addition of the ASLA amino acids to MIP-1␣S-4 results in a further reduction in binding affinity (Fig. 2B). Curiously, muMIP-1␣ binds in a manner more similar to that seen with either of the Ϫ4 variants than that seen with MIP-1␣P. The clear requirement for the proline residue in position 2 for enhanced binding of human MIP-1␣ to human CCR5 suggests that the simple presence of the analogous residue in murine MIP-1␣ is insufficient in the context of the other evolutionary changes in this protein, to mediate high affinity binding to human CCR5. Again, in a manner similar to that observed for murine CCR1, the proline at position 2 of MIP-1␣P appears to be unnecessary for the interaction of this ligand with human CCR1, with the full-length versions of MIP-1␣S and MIP-1␣P, showing a consistently lower affinity interaction with human CCR1 than either of the Ϫ4 variants (Fig.  2C). Again, in common with the data obtained using murine CCR1, murine MIP-1␣ binds more like the human Ϫ4 variants than either of the two full-length human MIP-1␣ isoforms (Table II).
It is important to note that while all the above binding data was obtained from displacement studies utilizing radiolabeled murine MIP-1␣ and the relevant cold competitor, very similar dissociation constants have been obtained from direct binding studies using radiolabeled human MIP-1␣ isoforms.
MIP-1␣ Isoform Binding Variations Are Reflected in Signaling Potency-To test whether the observed isoform binding variations are also reflected in the dose response for signaling through human CCR5 and CCR1, ligand-induced mobilization of Ca 2ϩ was studied. The transfected CHO cell lines used in the above binding studies are inefficient at fluxing Ca 2ϩ and for this reason we have used stably transfected HEK cells in the   Ca 2ϩ fluxing studies reported in this paper. Binding analyses using these cell lines indicates that the tested ligands for CCR1 and CCR5 bind with equivalent affinities to the receptors expressed in either CHO cells or HEK cells suggesting the likely equivalence of the receptors expressed on these different heterologous cells types. With full-length proteins on CCR5, halfmaximal signaling potency is seen at approximately 10 nM for MIP-1␣S and at approximately 500 pM for MIP-1␣P (Fig. 3A). The dose-response curves obtained for these two full-length chemokines are highly significantly different (p Ͻ 0.0002) indicating, as was seen with the binding studies, that MIP-1␣P is a markedly better ligand for CCR5 than MIP-1␣S. The MIP-1␣S-4 variant shows a slight, but significant (p Ͻ 0.001), increase in signaling potency through CCR5 compared with MIP-1␣S, while removal of the terminal 4 amino acids from the MIP-1␣P isoform significantly reduces its activity approximately 10-fold (p Ͻ 0.0002). Dose-response curves obtained using the two Ϫ4 variants were not significantly different and half-maximal activity was observed for both at concentrations of approximately 5 nM. The potency of signaling with MIP-1␣P is significantly higher (approximately 10-fold; p Ͻ 0.001) than that seen with RANTES (half-maximal activity seen at 2.5 nM) and ϳ100-fold higher than that seen with MIP-1␤ (p Ͻ 0.0002). MIP-1␣P is therefore the most active natural CCR5 agonist described to date. We have detected Ca 2ϩ fluxes with MCP2, which has recently been identified as a CCR5 ligand (32), but this ligand is not as potent as MIP-1␤ in this assay (data not shown). Murine MIP-1␣ signaling potency through human CCR5 accurately reflected the binding data with a half-maximal activity being observed at 5 nM, a concentration that was indistinguishable from that seen with either of the Ϫ4 variants (data not shown).
Signaling through human CCR1 again reflects the results from the binding studies and indicates no significant differences in signaling potency between the full-length isoforms of human MIP-1␣ which both display half-maximal fluxing capacity at approximately 2.5 nM (Fig. 3B). The Ϫ4 variants, however, while they are not different from each other in calcium fluxing ability, have a significantly higher signaling potency than the full-length peptides (p Ͻ 0.02) with half-maximal activity being observed at approximately 750 pM. In our hands, RANTES demonstrates a half-maximal fluxing potency at approximately 15 nM on human CCR1 and is thus significantly less potent than any of the MIP-1␣ isoforms tested (p Ͻ 0.001). Murine MIP-1␣ mediated a half-maximal flux at approximately 600 pM, again reflecting the similar binding potency of this ligand to the two Ϫ4 variants on human CCR1.
We have so far been unable to demonstrate a signaling role for human D6 (24) and, despite initial encouraging data (23), have been unable to reproducibly confirm a signaling role for murine D6. Thus the functional consequences of the high affinity binding of MIP-1␣P to either the murine or human orthologues of this receptor remain to be determined.
In summary, these results show that the proline residue at position 2 of MIP-1␣P is responsible for the enhanced binding of this isoform to murine and human D6 and CCR5, and for its strong activation of human CCR5. Conversely, interaction with CCR1 is lessened in the presence of APLA or ASLA at the NH 2 terminus of human MIP-1␣. These conclusions have been supported using Ϫ1, Ϫ2, and Ϫ3 forms of MIP-1␣S with NH 2 termini of SLAADTPT, LAADTPT, and AADTPT, respectively. In short, on human CCR1 and -5, and D6, the Ϫ2 and Ϫ3 forms behave like the Ϫ4 variants, while -1MIP-1␣S acts like fulllength MIP-1␣S protein (data not shown). Truncation beyond the Ϫ4 position reduces CCR1 and -5 activation properties further (not shown).
MIP-1␣P Is a Potent Natural Antagonist of HIV Interactions with CCR5-Given the strong affinity of MIP-1␣P for CCR5, we investigated the potency of this chemokine as an HIV suppressive agent. Studies were carried out to investigate the potency of full-length and Ϫ4 variants of both isoforms of human MIP-1␣ in suppressing entry of JRFL envelope pseudotyped virus into CEMx174-CCR5 cells. These studies demonstrated MIP-1␣P to be a significantly (p Ͻ 0.002) more potent suppressor of HIV entry than either MIP-1␣S or the Ϫ4 variants (Fig. 4A). Greater than half-maximal inhibition of JRFL pseudotyped viral entry was achieved with concentrations of MIP-1␣P as low as 5 ng/ml while 10-20-fold higher concentrations of the other human MIP-1␣ variants were required before 50% inhibition of viral entry was achieved. Similar results were seen with virus pseudotyped with the envelope of the ADA HIV1 strain, although all the variants were less able to inhibit entry (Fig. 4B). RANTES has been reported to be a much more potent antagonist of HIV entry through CCR5 than MIP-1␣S, MIP-1␤, and MCP2 (17)(18)(19). Therefore, further experiments were performed using replication-competent virus entry into CEMx174-CCR5 cells, and peripheral blood mononuclear cells (shown in Fig. 4, C and D, respectively), to compare MIP-1␣P to RANTES. These studies suggest that in addition to being by far the most effective human MIP-1␣ variant in suppressing HIV entry into target cells, MIP-1␣P is also at least as potent as RANTES. In fact in repeated experiments, MIP-1␣P was consistently better than RANTES as an HIV suppressive chemokine, however, these differences did not reach levels of statistical significance. Again, while MIP-1␣P was consistently less active as an HIV suppressive chemokine than the NH 2 -terminal modified form of RANTES, AOP-RANTES (33), this difference did not reach levels of statistical significance (Fig. 4, C and D, and see "Discussion"). Note that RANTES and MIP-1␣S-4 can enhance HIV entry into peripheral blood mononuclear cells at the lower concentrations tested, an effect not seen with MIP-1␣P or AOP-RANTES (Fig. 4D). In contrast to the CCR5-dependent viruses, no inhibition of CXCR4-dependent viral entry was observed using any of the MIP-1␣ isoforms tested thus confirming the CCR5 dependent selectivity of the MIP-1␣P inhibition (data not shown). These results present MIP-1␣P as one of the most potent naturally occurring inhibitors of HIV1 entry through CCR5.

DISCUSSION
Studies from our own and other laboratories have demonstrated the curious inability of human MIP-1␣ to bind with high affinity to CCR5 and D6 to which the closely related (74% identity at the amino acid level) murine MIP-1␣ protein binds strongly (23, 24, 34 -36). We now reveal that the basis for this is the absence of a proline residue at position 2 of the most commonly used isoform of human MIP-1␣ (MIP-1␣S/LD78␣). Our studies on the naturally occurring non-allelic variant of human MIP-1␣ (MIP-1␣P/LD78␤) which has a proline at position 2, reveal this to be active as a potent CCR5 and D6 ligand. It therefore more closely resembles murine MIP-1␣ than do any of the other human MIP-1␣ isoforms, suggesting that MIP-1␣P may be a closer functional human homologue of murine MIP-1␣ than MIP-1␣S. It appears, therefore, that despite the similar levels of identity between MIP-1␣S and P and murine MIP-1␣, MIP-1␣S is a functionally evolving variant of MIP-1␣. Our results suggest that MIP-1␣S should be regarded as a structurally related but functionally distinct chemokine which has substantially lost its ability to interact with the CCR5 and D6 receptors but which has apparently enhanced binding to CCR1 due to the proposed truncation of MIP-1␣S by 4 amino acids during release from the cell. Neither murine MIP-1␣ nor MIP-1␣P are released from mammalian cells as Ϫ4 variants (20,28,37) 4 and thus this putative alternative signal peptidase cleavage may be regarded as a component of the functional evolution of this human chemokine. The importance of the amino terminus to chemokine function is well documented (32, 38 -41) and it is therefore an ideal position for evolutionary modulation of function.
A major consequence of the enhanced interactions between MIP-1␣P and CCR5 is that this chemokine variant is now identified as being as potent as RANTES and AOP-RANTES in antagonizing CCR5-mediated cellular entry by HIV1. It is important, however, to point out that MIP-1␣P while not reaching levels of significance, was a consistently more effective HIV suppressive agent than RANTES. This coupled with the markedly and significantly lower ability of RANTES compared with MIP-1␣P to flux Ca 2ϩ following CCR5 binding, may suggest that the inability to demonstrate statistical significance in HIV suppression is a limitation of the assay system used and this is currently being investigated in more detail in our laboratories. Intriguingly, and as shown in Fig. 4, A and B, while it is consistently difficult to achieve higher than 50% inhibition of entry by JRFL or ADA pseudotyped viruses using other MIP-1␣ variants, MIP-1␣P can mediate near 100% inhibition with reasonable ease. Similar effects have been demonstrated for the synthetic NH 2 -terminal variant of RANTES, AOP-RAN-TES, with recent data suggesting this to be a consequence of alternative subcellular deposition of internalized receptors, and associated impairment of recycling of receptors (42). This may suggest that MIP-1␣P, in contrast to MIP-1␣S, alters receptor trafficking post-ligand binding and this is currently being investigated in our laboratory.
These results attesting to the importance of a proline residue at position 2 of MIP-1␣ are likely to have implications for our understanding of the interactions between other chemokines and their receptors. For example, it is likely that proline 2 in other ␤-chemokines is necessary for high affinity binding to D6. This, however, is not sufficient for D6 interaction, as SDF1 also contains a proline residue at position 2 yet shows no potential to bind to D6 (data not shown). Thus, proline 2 must be presented in the context of a ␤-chemokine to permit high affinity D6 interaction, while with CCR5 other domains are likely important in restricting the ligands for this receptor to MIP-1␣, -1␤, RANTES, and MCP2. Interestingly, the NH 2 terminus Xaa-Pro in chemokines has been demonstrated to be a target of dipeptidyl peptidase IV (CD26) (41,43,44). We would predict that MIP-1␣P, but not full-length nor Ϫ4 forms of MIP-1␣S, would be cleavable by this protease with dramatic changes in its properties, specifically a near total loss of interaction with D6, a reduced signaling capacity (and inhibition of HIV entry) through CCR5, and enhanced CCR1 activity. In fact, studies we have performed on a Ϫ2 form of MIP-1␣S (not shown) suggest that these properties will be seen with CD26-cleaved MIP-1␣P.
The role of the proline residue in CCR5 activation is of interest in light of recent data indicating the crucial role of proline 2 in SDF1 activation of its receptor CXCR4, which acts as an entry cofactor for T-tropic HIV1 strains. Indeed, mutation of this residue to glycine generates a high affinity CXCR4 antagonist (45). It is intriguing that both major HIV entry co-receptors have a strong preference for a proline residue at position 2 during receptor activation. Moreover, as proposed above for MIP-1␣P, SDF1 has been demonstrated to lose its anti-HIV and chemotactic activities upon removal of the first two amino acids by CD26 cleavage (43). Targeted reduction of CD26 activity in vivo may therefore enhance the activity of these HIV suppressive chemokines.
In addition to being of general interest to workers in the chemokine and HIV fields, our data also open up a number of potential therapeutic avenues of research. For example, earlier studies have demonstrated that the MIP-1␣P copy number varies between individuals, and this gene can in fact be absent from some individuals (25,26). ␤-Chemokine production has been reported to be associated with a number of inflammatory and autoimmune diseases (1,2), and deletion of the MIP-1␣ gene in mice dramatically alters responses to several infectious agents (22). In addition, ␤-chemokine production is clearly of importance in regulating the pathogenesis of AIDS (6,(12)(13)(14)(15)(16). Thus, given the ease with which MIP-1␣P is transcribed and translated (21,26,28) 5  ticular, in light of the potent HIV entry inhibition by MIP-1␣P described here, it would be of interest to test whether gene copy number affects the rate of progression to AIDS in HIV-infected individuals.
The demonstration of the potent HIV suppressive action of MIP-1␣P also indicates that it is possible that, as has been demonstrated with the chemokine RANTES (33,39), aminoterminal variants of this protein, such as AOP-linked MIP-1␣P, may exhibit enhanced HIV1 entry inhibition and/or receptor antagonism and have potential as HIV1 therapeutics.