A3₁₀ helical turn is essential for the proliferation-inhibiting properties of macrophage inflammatory protein-1 alpha (CCL3)

The hematopoietic system is characterized by a high turnover of mature blood cells from as many as 8 different lineages, which are constantly replenished throughout the entire life of the organism from a pool of multipotential stem cells resident in the bone marrow.¹  These hematopoietic stem cells (HSCs) therefore occupy a pivotal position within the hematopoietic hierarchy, and it is at the level of these cells that all hematopoietic function is ultimately regulated. In addition, HSCs and some less-primitive hematopoietic progenitor cells (HPCs) are responsible for the successful short- and long-term engraftment of peripheral blood or bone marrow transplant recipients, and are also attractive targets for hematopoietic-directed gene therapy. Clearly, therefore, it is important to be able to experimentally and therapeutically manipulate these cells, and for these reasons much effort has gone into understanding the regulation of HSC and HPC function.

We, and others, have been interested in the regulation of proliferation of primitive hematopoietic cells and evidence suggests that this regulation is controlled by positive as well as negative regulators. Prominent among the negative regulators of primitive hematopoietic-cell proliferation are members of the chemokine family of proinflammatory mediators.^2,3  Chemokines belong to a large family of proteins and are characterized by an ability to induce the directional migration of leukocytes^4-6  and other cell types.^7,8  Chemokines are biochemically characterized by the presence of a conserved cysteine motif in the mature proteins. The family is divided into 4 subfamilies on the basis of the specific nature of this motif. The 2 largest subfamilies are the CC and the CXC chemokine families. The other 2 families are the XC and the CXXXC families, which are represented by only single members. The numerous functions of chemokines are mediated through members of the 7-transmembrane-spanning G-protein-coupled receptor family (GPCRs), approximately 21 of which are currently known to bind chemokines.⁹  Receptors are referred to according to the subfamily with which they interact. Thus, CCRs interact with the CC chemokines, and CXCRs interact with the CXC chemokines. While the majority of chemokine receptors are G protein coupled and signal following ligand binding, there exists a small population of atypical and apparently signaling-incompetent receptors that may act as chemokine decoy receptors or alternatively may play roles in chemokine presentation.^10,11 

While a number of CC and CXC chemokines have been shown to inhibit the proliferation of primitive hematopoietic cells,^3,12,13  we have been particularly interested in the CC chemokine macrophage inflammatory protein-1α (MIP-1α/CCL3) in this context.^14-16  This protein functions in vitro^16,17  and in vivo^18-20  to inhibit proliferation of transiently engrafting hematopoietic stem/progenitor cells (HSPCs). Despite extensive biologic analyses, little information has been reported regarding the structural basis for this potentially important chemokine function. To date, structure/function analyses of chemokines have identified key structural requirements for varied chemokine functions. For example, many chemokines bind to proteoglycans. However, while there are indications that this interaction may be functionally important in some contexts,^21-23  it appears to be irrelevant for inhibition by MIP-1α.^15,24  In addition, there is clearly an important role for the N-terminus of chemokines in defining receptor binding ability and even subtle manipulations of this region, through mutagenesis²⁵  or enzymatic cleavage,^26-28  can radically alter chemokine function. Again, while this rule holds true for many chemokines, including MIP-1α in certain contexts,^29,30  the N-terminus appears to be less important for the proliferation inhibitory properties of this chemokine.³¹  Thus, a number of studies indicate that residues critical for HSPC inhibition by MIP-1α differ from those required for other chemokine functions. Indeed these structure/function studies, along with studies using receptor blockers and receptor-null mice, have led us to conclude that MIP-1α mediates HSPC inhibition through a novel receptor.³¹ 

The increasing availability of 3-dimensional structural data on members of the chemokine family has allowed more intuitive approaches to be taken to structure/function analyses. The existing structural data attest to the striking similarity of the monomeric chemokine units.^32-37  This common “chemokine fold” consists of an extended N-terminal region (which consists of an extended loop leading into a 3₁₀ helical turn), a triple-stranded, antiparallel β-sheet, and a C-terminal α-helix that lies on top of the β-sheet. Despite the similarity in chemokine monomeric structures, the dimeric structures vary considerably between the CC and CXC chemokine families. Whether this dimeric arrangement is functionally important has not yet been determined. Certainly most data would appear to indicate that chemokines typically interact with their surface receptors as monomers,^38,39  and the physiologic relevance of higher-order aggregates remains to be demonstrated.

Given the ability of MIP-1α to inhibit the proliferation of HSPCs and the potential clinical usefulness of this function, it was decided to attempt to identify the domain(s) within murine MIP-1α that are associated with this function of the chemokine. Identification of such domains, particularly if they are distinct from those involved in regulation of leukocyte migration, could potentially lead to the generation of MIP-1α variants specifically designed to retain HSPC inhibitory properties in the absence of leukocyte chemotactic properties. Such MIP-1α variants may have improved usefulness for the therapeutic inhibition of HSPCs.^2,12,40  Here we report the results of studies aimed at identifying the peptide regions that are most important for the inhibitory properties of MIP-1α. Specifically, we report that the helical turn directly preceding the first β-strand is a key peptide motif for HSPC inhibition by MIP-1α.

Recombinant human (h) RANTES/CCL5 was purchased from R&D Systems (Minneapolis, MN), and recombinant murine (mu) MIP-1α was generated as described previously.³⁸  All antibodies were purchased from R&D Systems. The peptide spanning the 3₁₀ helical turn was synthesised by CSS-Albachem (Edinburgh, United Kingdom).

The chimeric muMip1a and hRANTES cDNA molecules were generated by overlap PCR, using primers to the 5′ and 3′ ends of muMip1a and hRANTES and a number of chimeric primers spanning the borders of domain exchanges. Table 1 gives a complete list of the primers and Table 2 shows the order of primer use in chimera generation. Polymerase chain reactions (PCRs) were carried out in a final volume of 100 μL, containing 10 μL10 × Pfu buffer, 10 μL DMSO, 10 μL 50% glycerol, 4 μL dNTPs at 10 mM, 3 μL of each primer at 330 ng/mL (Promega, Southampton, United Kingdom), 1 μL of the template at 110 ng/mL, and 1 μL Pfu polymerase (Stratagene, Cambridge, United Kingdom), and made up to the final volume with water. In the second round of PCR where the double-stranded product of the first round was included as an additional primer, the amount of this double-stranded primer and of the template added to the reaction was reduced to 10 ng, and the concentrations of the single-stranded extreme 5′ and 3′ end primers was 0.1 μM. Products were obtained after 25 cycles of a 1-minute denaturing step (94°C), a 1-minute annealing step (55°C), and a 1-minute extension step (72°C), followed by a 10-minute incubation at 72°C. The final products were cloned into pCR-Script (Stratagene), sequenced, and subcloned into the NotI/BamHI sites of the baculovirus transfer vectors pVL1392/1393 (Becton Dickinson, Plymouth, United Kingdom).

Baculovirus transfer vectors containing the mutant cDNAs were cotransfected with linearized DNA derived from the defective AcNPV baculovirus (Becton Dickinson) into Sf9 insect cells grown in TNM-FH (BioWhittaker, Wokingham, United Kingdom), following the manufacturer's instructions. Viral clones were isolated by plaque assay, amplified, and used to infect Sf9 cells at a multiplicity of infection of 5. Recombinant protein-containing cell supernatants were collected after 6 days and viral particles removed by centrifugation at 18 000 rpm for 30 minutes.

The chimeric proteins were purified using an fast performance liquid chromatography (FPLC) system (AmershamPharmacia Biotech, St Albans, United Kingdom). The supernatant was first applied to a heparin-affinity column (AmershamPharmacia Biotech), the proteins eluted by developing a gradient between 0.1 M and 1.2 M NaCl and positive fractions identified by Western blotting. Positive fractions were further purified by reversed-phase chromatography (AmershamPharmacia Biotech), the proteins eluted by developing a gradient between 0% and 60% Acetonitrile/0.1% TFA and positive fractions again identified by Western blotting. The protein in these fractions was lyophilized, resuspended in phosphate-buffered saline (PBS), and the concentration estimated by analyzing aliquots in the presence of protein standards on a 17.5% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). The protein bands were visualised by silver staining or by Western blotting.

The swap of the alanine-histidine (AH) and glutamine-phenylalanine (QF) dipeptide motifs between hRANTES and muMIP-1α was achieved using wild-type hRANTES and muMip1a cDNAs. Mutagenesis was carried out using a QuickChange site-directed mutagenesis kit (Stratagene) according to the manufacturer's instructions. Chemokine mutants were partially purified by heparin-affinity chromatography, as described, and desalted into PBS prior to assaying in the colony-forming unitagar assay (CFU-A) as described in “CFU-A assay.”

HEK293 cells expressing human CCR5²⁹  were loaded with Fura-2-am for 1 hour at 37°C in SR buffer (136 mM NaCl, 4.8 mM KCl, 5 mM glucose, 20 mM HEPES, 1mM CaCl₂, and 0.05% bovine serum albumin [BSA; pH 7.2]). The cells were then washed and warmed at 37°C for 2 minutes before starting fluorescence measurements in an LS50 spectrophotometer (PerkinElmer Life Sciences, Norwalk, CT) at an excitation wavelength of 340 nm and an emission wavelength of 500 nm. Recombinant proteins were added after 50 to 60 seconds, the fluorescent emission recorded for a further 1 to 2 minutes and, in the case of desensitization measurements, additional chemokine added where appropriate.

CFU-A assays were carried out as described previously.⁴¹  Briefly, murine bone marrow was obtained by flushing the femoral bones of B6D2F1 mice (Harlan UK Ltd, Oxon, United Kingdom) with PBS. Assays were carried out in 3-cm plates with a bottom feeder-layer (equal volumes of 2 × medium [21 mL αMEM, Gibco Life Technologies, Paisley, United Kingdom; 25 mL donor horse serum, Sigma Chemical, Poole, United Kingdom, 1 mL L-glutamine, 200 mM and 3 mL sodium bicarbonate, 7.5%, per 50 mL] and 1.2% agar noble; DIFCO Laboratories, Detroit, MI) containing 12 ng/mL murine stem-cell factor (muSCF), 6 ng/mL human macrophage colony-stimulating factor (hM-CSF), and 0.2 ng/mL murine granulocyte-macrophage CSF (muGM-CSF). Murine bone marrow cells were added to the top layer (equal volumes of 2 × medium and 0.6% agar noble) at a concentration of 5 to 10 × 10³cells/mL. The plates were incubated at 37°C in a humid atmosphere of 5% O₂/10% CO₂ for 10 to 11 days, and colonies stained with 2-(p-iodophenyl)-3-(p-nitrophenyl)-5-phenyl tetrazolium chloride (INT) (Merck, Poole, United Kingdom).⁴¹  To assay for inhibitory activity, the chemokines and chimeras were added directly to the bottom-layer agar at a final concentration of up to 100 to 500ng/mL.

Previous detailed analyses^42-44  have indicated that cells giving rise to colonies with a diameter in excess of 2 mm in the CFU-A are phenotypically similar to spleen colony-forming unit (CFU-S) day-12 cells and thus represent transiently engrafting stem cells. The smaller colonies represent more committed progenitors, which are refractory to inhibition by MIP-1α. The extent of inhibition is therefore routinely measured by selectively counting colonies with a diameter of 2 mm or greater in each plate and comparable to the number of similar colonies in control plates.⁴¹ 

All data were analyzed using Student t test and all statistical analyses were performed using Graph Pad Prism software (GraphPad Software, San Diego, CA).

While MIP-1α is active as an inhibitor of HSPC proliferation, the related chemokine RANTES is inactive in this context. To determine the structural domains responsible for this activity in MIP-1α we have capitalized on the availability of the 3-dimensional structure of both of these molecules (Skelton et al⁴⁵ ; also J.M., G.J.G., N.W.I.; The structure of aggregated variations of the chemokine Mip1α, manuscript in preparation and PhD thesis, Glasgow University, 1997). The similarity of the primary and tertiary structures of these 2 chemokines has allowed us to design “domain-swap” mutants in which individual hRANTES domains are replaced with corresponding muMIP-1α domains, and vice versa.

In the first instance, and as shown in Figure 1A, muMIP-1α and hRANTES were divided into 3 major regions: (1) the N-terminal region up to the CC-motif; (2) the main body of the chemokine between the second and the fourth conserved cysteines, including the loop following the CC motif and the triple-stranded β-sheet; and (3) the C-terminal α-helix (following the last conserved cysteine). These were exchanged in all possible combinations to produce the first generation of mutants (Figure 1B). The chimeric molecules were generated by overlap PCR and expressed in the baculovirus system as described. The expression levels of the individual recombinant chimeric proteins (as measured by Western blotting) proved to be highly variable (Table 3). While there was no clear and consistent structural explanation for these marked differences, the combination of a hRANTES C-terminus and a muMIP-1α main body was clearly associated with poor expression. Wild-type hRANTES was consistently poorly expressed and thus, commercially available hRANTES has been used throughout the subsequent studies as the wild-type hRANTES control. All other proteins were subjected to a 2-step purification involving heparin-affinity and reversed-phase chromatography. The purity and approximate relative concentrations of the proteins were assessed by analysis of SDS-PAGE gels. The identity of the individual proteins was confirmed by N-terminal sequencing (data not shown).

To confirm the activity of the purified chimeras, they were tested for their ability to induce a calcium flux in HEK293 cells expressing human CCR5. We have previously shown³¹  this receptor not to be involved in HSPC inhibition by MIP-1α. However, CCR5 was chosen for initial assay of the chimeras because both muMIP-1α and hRANTES bind to it with high affinity and induce calcium fluxes with similar dose responses.^29,47  Successfully folded chimeras should therefore also interact functionally with this receptor. These calcium flux assays should therefore be regarded as indicative of productive folding of the chimeras and not as predictive of HSPC inhibitory function. Functional interaction with CCR5 was further confirmed by testing the ability of signaling chimeras to desensitize responses to subsequent stimulation by wild-type muMIP-1α. As summarized in Table 3, both muMIP-1α (effective concentration for 50% inhibition [EC₅₀] of 5.8 × 10^-8 M) and hRANTES (EC₅₀ of 2.8 × 10^-7 M) induced a calcium flux through CCR5 although MIP-1α was seen to be more potent. These potency differences have meant that EC₅₀ values for calcium flux measured for the chimeras would be difficult to interpret depending on the relative contribution of the muMIP-1α and hRANTES domains. For this reason we have presented the calcium flux data as a percentage of response induced by 10 μg/mL chimera compared with 10 μg/mL muMIP-1α. These figures should therefore be regarded as approximate indicators of chimera function. When fluxes in response to 10 μg/mL each of the chimeras were measured, MRR, MMR, RRM, and MRM (all as defined in Figure 1) all gave good responses, indicating that these chimeras interacted effectively with CCR5 and can thus be assumed to be properly folded. MRR, MMR, and MRM gave good desensitization to subsequent muMIP-1α responses, although RRM displayed only partial ability in this regard. With RMR, the induced calcium flux following CCR5 binding was weaker and, while there was desensitization to subsequent challenges with muMIP-1α, this was incomplete. A calcium flux was also observed following RMM binding to CCR5 but again was weaker, and this chimera had to be added at a 10-fold higher concentration than the other proteins to give a flux. Again, the RMM-mediated desensitization to subsequent muMIP-1α challenges was incomplete. The partial desensitization mediated by RRM, RMR, and RMM suggested that these chimeras were partially compromised in their ability to interact with CCR5. We are unable to say from the current data whether this relates to a potency difference or to combinations of specific muMIP-1α and hRANTES domains being less efficacious than the respective wild-type proteins. It is curious to note that the less active of the chimeras all share a hRANTES N-terminus, suggesting that this region of hRANTES may have less flexibility than the equivalent region of muMIP-1α to properly fold in the presence of heterologous peptide motifs.

Chemokines and their chimeras were tested for hematopoietic stem/progenitor-cell inhibitory activity in the in vitro CFU-A. The results from these assays (Figure 2) confirm the ability of muMIP-1α to inhibit CFU-A cell proliferation (inhibitory concentration of 50% [IC₅₀] of 9.2 × 10^-10 M) and further attest to the inactivity of hRANTES in this assay. The inhibitory activity of muMIP-1α in the CFU-A also correlates with a dramatic reduction in the number of cells per colony compared with colonies from control plates (data not shown). This confirms that the inhibition seen in the CFU-As does not simply reflect a change in colony morphology but is indeed an inhibition of proliferation. It is important to note that in the present study, in contrast to other studies,¹⁷  100% inhibition of colony formation is routinely achieved at the highest concentrations of chemokine. This relates to the fact that, in each assay plate, we specifically scored colonies derived from CFU-A cells, the equivalent of CFU-S day-12 cells,^42-44  and not other progenitor-cell-derived colonies. These “CFU-A” cells, in contrast to progenitor cells, are fully inhibitable by MIP-1α. The scoring methodology is detailed in “Materials and methods.”

Of the chimeras tested, RRM (IC₅₀ of 6.7 × 10^-9 M) and RMM (IC₅₀ of 1.02 × 10^-8 M) were seen to be weakly inhibitory, whereas RMR showed consistent marked inhibition (IC₅₀ of 1.2 × 10^-9 M) of CFU-A colony formation (Figure 2) and displayed a dose-response relationship that was similar to that seen with wild-type muMIP-1α (data not shown). Interestingly, while the inhibited colonies in the CFU-As treated with RMR and RMM had reduced numbers of cells compared with control colonies, the colonies in the RRM-treated plates did not display any reduction in component cell number (data not shown). Analysis of MMR, MRR, and MRM, which had all produced a calcium flux in CCR5-transfected cells, showed them to be inactive as HSPC inhibitors.

The observation that RMR and RMM were the only chimeric proteins that displayed inhibitory activity as measured by both total colony counting and evaluation of colony cell number suggests that the main determinants for CFU-A cell inhibition are located within the main body of muMIP-1α between the CC motif and the C-terminal α-helix. This conclusion is further supported by the fact that MRM, which lacks the muMIP-1α main body but retains the rest of the muMIP-1α framework, was inactive in the CFU-A despite being a potent mobilizer of calcium following CCR5 binding. It is important to note that, while the inactivity of MMR in the CFU-A appears to lessen the case for the main body of muMIP-1α being a key determinant for inhibitory activity, our subsequent second-generation chimera studies (see “Design of second-generation chimeras”) further confirm this to be the major domain for inhibitory function and suggest that MMR is compromised in some structural or functional way.

Having identified the main body of muMIP-1α (between the second and the fourth conserved cysteine residues) as containing important structural determinants for HSPC inhibition, further mutagenesis studies were initiated with the aim of more accurately defining this inhibitory domain. Alignment of the sequences of human and murine MIP-1α and RANTES indicate that the majority of nonconservative sequence differences are seen in the region between the second and the third cysteine residues (Figure 1). This region in MIP-1α and RANTES is composed of the following structural features: a loop (residues 12-19), a helical turn (residues 20-24), and the first β-strand (residues 25-31). These muMIP-1α structural domains were individually swapped with equivalent regions in hRANTES using essentially the same methodology described previously. These second-generation mutants are referred to henceforth as RLR, RTR, and RBR, where L is the loop, T the helical turn, and B the first β-strand (Figure 3A). These chimeras were expressed and purified as before.

All 3 second-generation chimeras expressed well in the baculovirus system. In calcium flux functional assays, RLR, RTR, and RBR all induced a flux through CCR5, although RBR produced only a partial desensitization of the receptor (Table 3).

The 3 second-generation chimeras were tested for proliferation inhibitory activity in the CFU-A. As the intention of these experiments was to identify key domains important for HSPC inhibitory activity, the second-generation mutants were only tested at a single high dose in the CFU-A and IC₅₀ values, which may be difficult to interpret, are not quoted. As before, MIP-1α and RMR produced significant inhibition, whereas RANTES was inactive in these inhibitory assays. As shown in Figure 3B, neither RLR nor RBR inhibited CFU-A colony formation at 100 ng/mL, and both were also inactive at concentrations up to 1 μg/mL (data not shown). In contrast, RTR was clearly active in the inhibitory assays, and addition of this chimera to the assay plates resulted in the characteristic dense and small colonies seen following muMIP-1α addition to the CFU-A (data not shown). Colony morphology in plates with added RLR or RBR was indistinguishable from that seen in control plates. Furthermore, as with muMIP-1α and RMR, the number of cells in the RTR-inhibited colonies is reduced compared with control colonies without any detectable alterations in cell composition of the colonies (data not shown). In contrast to RTR, neither RLR nor RBR had any effects on the number or types of cells within the CFU-A colonies. Thus these results clearly demonstrate that a major determinant of the inhibitory activity of muMIP-1α is located in the helical turn preceding the first β-strand.

Having identified the helical turn preceding the first β-strand as being key to the inhibitory activity of muMIP-1α, we have aligned the helical turn regions from inhibitory and noninhibitory CC chemokines for comparison. Examination of the turn regions in these chemokines highlights a dipeptide motif comprising an asparagine (in hMIP-1α and hMIP-1β) or a glutamine (in muMIP-1α) followed by a phenylalanine (N/QF) in the inhibitory chemokines (mu and hMIP-1α and hMIP-1β)⁴⁸  that is not present in the noninhibitory chemokines (muMIP-1β, mu and hRANTES, and hMCP1; Table 4). To examine the importance of these 2 amino acids for inhibitory activity we swapped the QF motif in muMIP-1α with the AH motif in hRANTES to generate MIPAH and RANTESQF. These were tested for inhibitory activity as before (Figure 4). The results indicate that introduction of the QF motif into the helical turn in hRANTES is sufficient to convert it to an inhibitor of HSPC proliferation. Curiously, introduction of the AH dipeptide into muMIP-1α did not block its inhibitory activity as expected. While we have no clear explanation for this we assume that it indicates that the turn region of muMIP-1α has greater tolerance to structural alteration than the equivalent region of hRANTES. This may also suggest that the overall structure of the turn region, rather than the specific amino acids, is the key feature important for the inhibitory properties of the chemokines. This conclusion is further supported by data obtained using a linear peptide (IPRQFI) spanning the turn region in muMIP1α. When tested either alone or in combination with muMIP-1α in the CFU-A, this peptide displayed neither agonistic nor antagonistic activity even when used at a 2000-fold molar excess over muMIP-1α (Figure 5). This peptide also does not display either agonistic or antagonistic activity on CCR5 as measured in Calcium flux assays (data not shown).

Together, these data suggest that while 2 amino acids are sufficient to introduce inhibitory activity to hRANTES, a significant secondary or tertiary structural element is also important.

The HSPC inhibitory activities of members of the chemokine family have been known for many years.¹²  We have been particularly interested in the ability of MIP-1α to inhibit the proliferation of transiently engrafting stem/progenitor cells. This property is not shared with the related chemokine RANTES, and more recent studies indicate that this effect relies on the interaction of MIP-1α with an as-yet uncharacterized receptor.³¹  These observations therefore set the inhibitory activities of MIP-1α apart from its proinflammatory roles. Here we have used detailed knowledge of the 3-dimensional structures of muMIP-1α and hRANTES to investigate the molecular basis for inhibition by MIP-1α. By exchanging domains between muMIP-1α and hRANTES, we have been able to demonstrate that the residues responsible for the inhibitory activity lie between the second and fourth cysteines in muMIP-1α. Further domain swapping has allowed us to more accurately define the inhibitory motif as being within the helical turn that precedes the first β-strand in muMIP-1α (residues 20-24). The importance of this motif in the context of proliferation inhibition contrasts with much evidence indicating the general importance of the N-terminus of chemokines in a range of other functions.^25,29,30,49  Indeed, our previous studies have demonstrated a lack of effect of N-terminal truncations on the inhibitory function of huMIP-1α.³¹  This relative unimportance of the N-terminus for inhibitory activity also correlates with the lack of involvement of the known muMIP-1α receptors in HSPC inhibition.

Intriguingly, not all chimeras bearing the region between the second and fourth cysteines were active in the inhibitory assays. Indeed, MMR, while capable of inducing a calcium flux following CCR5 binding, appeared inactive in the CFU-A inhibitory assay. The fact that RMR and RMM were active suggests that the pairing of the muMIP-1α N-terminus with the hRANTES C-terminus is unfavorable in terms of allowing folding appropriate for the inhibitory activity of these chimeric proteins. This, along with the inactivity of the peptide spanning the turn region, indicates that there are likely to be exacting tertiary structural requirements for MIP-1α to function as an HSPC inhibitor that are different from those required for its proinflammatory functions.

Further, more detailed experiments aimed at identifying individual amino acids in the 3₁₀ helical turn that are involved in the inhibitory functions of muMIP-1α have yielded confusing results. The main difference between muMIP-1α and hRANTES in this region is a 2-amino acid motif that is QF in muMIP-1α and AH in hRANTES. Swapping the AH in hRANTES for QF allows hRANTES to function as an HSPC inhibitor, which strongly argues for the importance of these residues in contributing to this functional property in muMIP-1α. However, the reciprocal swap of AH for QF in muMIP-1α does not, as might have been expected, remove the inhibitory activity of muMIP-1α. We interpret this as suggesting that, in the context of the muMIP-1α protein, there is sufficient structural tolerance in the 3₁₀ helical turn to allow this change to be made without blocking the inhibitory activity. Attempts at structural modeling of these swaps based on the known structures of MIP-1α and RANTES have failed to provide insights into the tolerance of muMIP-1α to the AH/QF swap.

Together these data suggest that the Q/NF motif on its own is not an absolute prerequisite for the inhibitory activity and that other primary or higher-order structural determinants of the helical turn are also important. Clearly, however, this dipeptide motif discriminates between the inhibitory and noninhibitory CC chemokines. Curiously, a positionally conserved glutamine residue is present in many other inflammatory CC chemokines but is typically followed by an arginine residue (see MCP1/CCL2 in Figure 5). The reason for this relatively strong evolutionary conservation of the QR motif is unclear at present.

To our knowledge, this is the first time that the helical turn preceding the first β-strand has been demonstrated to harbor determinants for a specific biologic activity of a chemokine. Amino acids in the vicinity of this region have been implicated in binding of eotaxin/CCL11 to its receptor, CCR3,^50,51  although the relevance of these amino acids to the biologic function of eotaxin remains to be demonstrated. In addition, studies by Bondue et al⁵¹  demonstrate that residues within this region contribute to the binding of MIP-1β/CCL4 to CCR5, further highlighting the importance of this helical turn for chemokine/receptor interactions. While, as mentioned in “Introduction” and earlier in “Discussion,” most studies indicate that receptor-activating residues in CC chemokines are found in the flexible N-terminal region preceding the CC motif, the relative conservation of the QR sequence in the 3₁₀ helical turn of inflammatory CC chemokines suggests that this region is likely to be important in other chemokines for direct interaction with receptors or for proper structural folding. Studies on the structural basis for the hematopoietic inhibitory properties of platelet factor 4(PF4)/CXCL4 have identified an alternative region within this chemokine that appears to be important for inhibitory activity. This region spans residues 34 to 58 of PF4, and further studies have identified a DLQ motif at positions 54 to 56 of human PF4 as well as a helical motif incorporating residues 38 to 46 as being essential for inhibitory activity.⁵²  The relationship of these regions to PF4 receptor binding⁵³  remains to be determined.

In conclusion, we report here the identification of the 3₁₀ helical turn preceding the first β-strand as being a specific motif within muMIP-1α that appears to be important for inhibitory activity. When swapped for equivalent regions in hRANTES, this small region can confer inhibitory activity on hRANTES. These studies have therefore highlighted the turn preceding the first β-sheet of muMIP-1α as a prime determinant of biologic function and open up the possibility of generating MIP-1α variants that are preferentially active as inhibitors of HSPC proliferation. For clinical use, these studies need to be repeated using human MIP-1α; such studies are under way in our laboratory. While issues of possible antigenicity of chimeric MIP-1α/RANTES proteins remain to be addressed, the present data highlight the potential value in further studying such molecules in the clinical context.