[Frontiers in Bioscience 3, d44-58, January 1, 1998]

	[Frontiers in Bioscience 3, d44-58, January 1, 1998] Reprints PubMed CAVEAT LECTOR
	CHEMOKINE RECEPTORS AND HUMAN IMMUNODEFICIENCY VIRUS INFECTION Paul D. Bieniasz and Bryan R. Cullen Department of Genetics and Howard Hughes Medical Institute, Duke University Medical Center, Durham, NC 27710 Received 12/15/97 Accepted 12/19/97 5. MECHANISM OF ACTION OF CORECEPTORS Accumulating evidence indicates that the SU envelope glycoproteins of primate lentiviruses directly interact with their cognate coreceptors (52-56). Soluble gp120 from T-tropic HIV-1 has been reported to form a precipitable complex with CXCR-4 (52) and specifically bind to cells expressing CXCR-4 with an affinity in the nanomolar range (53). In addition, a few groups have demonstrated that HIV-1 and SIV envelope glycoproteins are able to compete with beta chemokines for CCR-5 binding (54-56). Significantly, this competition is much more efficient in the presence of CD4, either coexpressed with CCR-5 on the cell surface, or in soluble form as a pre-formed complex with gp120. The inference from these data is that the interaction of HIV or SIV with a target cell is initially with CD4, thereby inducing a conformational change in gp120 which facilitates subsequent binding of envelope to the coreceptor. This scenario is supported by the previous observation that a number of HIV-2 strains can be induced to infect CD4 negative cells by pre-treatment with sCD4 (57). In this case a virion/sCD4 interaction presumably induces a stable envelope conformation that enables interaction with a coreceptor in the absence of cell surface CD4. Why this phenomenon should be restricted to HIV-2 remains obscure, but perhaps sCD4 complexes with the gp120 of other primate lentiviruses adopt a coreceptor-binding conformation that is less stable than that obtained with HIV-2. The ability of HIV-2 to be adapted to efficiently infect CD4 negative cells, using either CXCR-4, CCR-3 or the orphan receptor V28 (57-59), probably represents an extension of this phenomenon. In this case, the viral envelope presumably adopts a stable CXCR-4 binding conformation in which the coreceptor binding site is revealed without the need for prior 'activation' by CD4. 5.1 Envelope binding sites on virus coreceptors All of the identified HIV/SIV coreceptors are members of the seven transmembrane spanning G-protein coupled receptor family. As such, they contain an N-terminal extracellular domain and three extracellular loops (ECL-1-3). A number of studies have attempted to determine which of these extracellular domains are determinants of coreceptor function, and by implication, what are the amino acid residues that interact with the virus envelope (60-67). In most cases, the functional coreceptor phenotype cannot be mapped to a single linear sequence. It is therefore likely that the virus binding site or sites on the coreceptor involve multiple extracellular domains, and that within each of these domains multiple amino acids contribute to coreceptor function. While this complexity has largely frustrated attempts to define precise virus binding sites, studies of chimeric and mutant coreceptors have, to some extent, illuminated our understanding of how HIV/SIV interact with coreceptors and suggested that several, functionally redundant, interactions between coreceptor and virus envelope are likely to occur. As we will discuss, this has implications both for the feasibility of using coreceptor targeted inhibitors of virus entry as therapies and for understanding the evolution of coreceptor usage in vivo. 5.2 Determinants of CCR-5 coreceptor function Historically, the generation of chimeric molecules consisting of a functional retrovirus receptor and a closely related, but non-functional homologue has proven to be a successful strategy for rapidly mapping the determinants of virus receptor function. Several studies have attempted to map the extracellular determinants of CCR-5 coreceptor function using such an approach (60-66). Fortuitously, a murine homologue of CCR-5 had been previously identified that is 82% identical to the human receptor but non-functional as a coreceptor for a all HIV-1 strains tested so far. At least four groups have used human/murine CCR-5 chimeras to map functional domains (60-63). There is broad agreement in the general conclusions among these three studies, some discrepancies are probably the result of methodological difference, details in the derivation of the chimeric coreceptors and polymorphisms in murine CCR-5 alleles. Of the four extracellular CCR-5 domains, human and murine proteins differ in sequence in three. Remarkably, each of these extracellular human CCR-5 domains is independently capable of conferring some degree of coreceptor function on an otherwise inactive murine receptor (60-62). Conversely there is no single extracellular domain of human CCR-5 that cannot be functionally substituted by its murine counterpart. (at least in the case of most murine CCR-5 alleles) Importantly, these chimeric coreceptors are often functional for a significantly more restricted range of HIV-1 strains; the differential abilities of virus strains to utilize various chimeric receptors strongly implies that different virus strains interact with CCR-5 in different ways (61, 62). Examples of the strain dependent differences in the ability of HIV-1 isolates to utilize human/mouse chimeric CCR-5 receptors are presented in figure 2. Figure 2. Examples of differential utilisation of human/murine CCR-5 chimeras by M-tropic and dual-tropic HIV-1 strains. The relative ability of human/murine CCR-5 chimeras to support fusion induced by the envelopes of two M-tropic (ADA and Ba-L) and one dual-tropic (89.6) HIV-1 strain is indicated. While ADA is efficiently able to utilize chimeras containing a single human extracellular domain, BaL requires the presence of two, and 89.6 requires an intact human CCR-5 molecule. Some strains, in particular those which are dual- tropic (i.e. are also capable of using CXCR-4 as a coreceptor) appear to be particularly sensitive to perturbations in the human CCR-5 extracellular domains. In fact, none of the three human CCR-5 extracellular domains that differ in sequence to the murine homologue can be replaced by murine sequences without (at least partially) compromising dual-tropic coreceptor function (61, 62). This is a suprising observation, since human and mouse CCR-5 receptors are much more similar in sequence than are CCR-5 and CXCR-4, a fully functional coreceptor for dual tropic HIV-1 strains. In contrast the M-tropic strains ADA and SF162 and are able to quite efficiently utilize chimeric coreceptors where any one of the 3 divergent extracellular domains is of human origin (61, 62, figure 2). Other M-tropic strains such as Ba-L, M23 and E80 have intermediate properties, in that a larger number of chimeras are functional than for dua-tropic strains, but some chimeras which are functional ADA or SF162 coreceptors do not support fusion mediated by these additional envelopes. For example, while no single human extracellular domain is sufficient to confer high level Ba-L coreceptor function on murine CCR-5, (figure 2) inter-domain synergy can be observed; chimeras containing two extracellular domains of human CCR-5 origin exhibit near wild type Ba-L coreceptor function, while each individually has little or no activity in the context of an otherwise murine receptor. Interestingly the results obtained using a CCR-5 allele from an NIH Swiss mouse for chimera construction differ significantly from those obtained using other murine CCR-5 alleles (63). It is apparent that murine CCR-5 polymorphisms are largely responsible for this discrepancy, and implicate specific amino acids in the envelope/CCR-5 interaction (see below). It is noteworthy, therefore that given the complexity of envelope coreceptor interactions, in particular the participation of multiple coreceptor domains, non-functional CCR-5 homologues are unlikely to provide a neutral background for analysis using chimeras or site directed mutant receptors. Thus, which extracellular domains in CCR-5 that are 'flagged' as being important for coreceptor function is dependent not only on the virus strain, but also on the non-functional partner selected for the generation of chimeras. This premise is largely born out when the above studies are viewed in conjunction with others which have employed CCR-5/CCR-1, CCR-5/CCR-2b, CCR-5/CXCR-2 and CCR-5/CXCR-4 chimeras to map determinants of coreceptor function for both HIV-1 and SIV strains (64-67). However, it is clear that the N-terminal extracellular domain of CCR-5 is capable of conferring coreceptor function for some HIV-1 M-tropic and dual-tropic strains on multiple, otherwise inactive, 7TM receptors (65). Loss of function as a result of partial truncations in the CCR-5 N-terminus, in the context of a CCR-5/CCR-2b chimera, support the role of this domain in coreceptor function, although the particular amino acids involved are clearly strain dependent (64). As with human/murine CCR-5 chimeras however, the N-terminus is dispensable in an otherwise intact human CCR-5 molecule, since it can be truncated or replaced with those of CCR-2b, CCR-1, CXCR-2 and CXCR-4 without drastically compromising M-tropic coreceptor function (64, 65). Dual-tropic strains are again more fastidious than M-tropic strains in their ability to interact with N-terminally substituted CCR-5 chimeras; of the aforementioned 7TM receptors, only the CCR-1 N-terminus can functionally replace that of CCR-5, and then only for one of the two dua-tropic strains analysed (65). These data have been interpreted as suggesting that dual tropism evolves from M-tropic viruses that retain the ability to interact specifically with the N-terminus of CCR-5 while acquiring the ability to interact with the ECLs of CXCR-4 at the expense of interaction with CCR-5 ECLs. We would argue that the interaction of dua-tropic envelopes with CCR-5 also remains critically dependent on CCR-5 ECLs, since replacement of these with corresponding murine CCR-5 sequences results in a dramatic loss of coreceptor function (61, 62, figure 2). Mutational analyses of CCR-5 extracellular domains in the context of an otherwise intact molecule has not proven to be particularly informative for M-tropic HIV-1, since many point mutants retain substantial coreceptor function (66). Nevertheless, partial or total loss of dua-tropic coreceptor function (dependent on the particular strain) as a result of an aspartate to alanine mutation at position 11 in the N-terminus has been documented (66). The partial effect of this mutation for the 89.6 strain is markedly enhanced when combined with alanine substitution of lysine 197 (ECL2) and/or aspartate 276 (ECL-3). Again this, indicates that the CCR-5 N-terminus is a critical, but not the sole, determinant of dua-tropic coreceptor function. In the case of M-tropic HIV-1 strains, studies in our laboratory have exploited human/murine CCR-5 chimeric coreceptors whose activity is dependent on the presence of a single human CCR-5 extracellular domain (68). Residues of human origin were substituted, either individually or in combination, with corresponding murine sequences. These experiments indicate an important role for residues serine 7, asparagine 13, and tyrosine 15 in the N-terminus and serine 180 in ECL-2. In addition, individual amino acids were identified (for example tyrosine 184 and serine 185) that had little or no effect on coreceptor function when mutated individually, but dramatic losses in coreceptor result when both are substituted. Thus, it is evident that not only do multiple CCR-5 domains contribute to coreceptor function, but within these domains multiple amino acids play an important role. In addition, the identification of non functional CCR-5 alleles from African green monkey and NIH Swiss mouse has indicated an especially critical role for tyrosine 14 in the N-terminus, glutamine 93 in ECL-1 and proline 183 in ECL-2 (63). Indeed, mutation of tyrosine 14 or proline 183 in an otherwise intact CCR-5 molecule renders it non-functional as a coreceptor for several M-tropic HIV strains. A summary of the mutational analyses performed to date is depicted in figure 3. Although residues which are important for function are somewhat scattered throughout the extracellular domains, several important residues are closely clustered within the N-terminus and in ECL-2. As yet, we cannot determine whether these sequence motifs constitute spacially distinct envelope recognition domains, or form an single envelope binding site in the context of a folded protein. Alternatively some of these amino acids could be responsible for influlencing the overall structure of the CCR-5 extracellular domains and modulate coreceptor function without directly contacting the viral envelope. Figure 3. Summary of mutational analyses of CCR-5. Amino acid residues that differ between human and murine (BALB/c) CCR-5 are indicated by shading. Arrows indicate ammino acids that when mutated either individually or in combination compromise coreceptor function for M-tropic (M) or dua-tropic (D) HIV strains. Note that some mutations are performed in the context of human/mouse CCR-5 chimeras, where coreceptor activity is dependent on the presence of a single human extraceellular domain. (data from refs 63, 66, 67)). The CCR-5 sequences involved in recognition by other primate lentiviruses are less well characterized. However, in the context of CCR-5/CCR-2b chimeras, ECL-2 is required for recognition by several SIV strains (67). Like HIV-1 strains, different SIV strains use CCR-5 in different ways: For other SIV strains, ECL-2 is dispensable provided that an intact CCR-5 N-terminus is present. 5.3 Determinants of CXCR-4 coreceptor function With the aim of adopting similar approaches to analyse structure function relationships in CXCR-4, we and others have cloned the murine homolog of this receptor (69-71). Unfortunately, murine CXCR-4 was found to constitute a functional coreceptor for at least some T-tropic and dua-tropic HIV-1 strains (71,72), precluding the construction of informative chimeras. In addition, both rat and feline CXCR-4 receptors were also found to support infection by at least some T- tropic HIV-1 strains (73, 74). These observations were suprising given the large body of literature indicating that almost all non-human cells remain refractory to HIV-1 infection when engineered to express human CD4, whereas most human cell lines are highly susceptible to T-tropic HIV-1 infection when CD4 is expressed (1-4). This apparent anomaly can only partially be explained by species specific differences in expression patterns. While we obtained no evidence for differential expression of CXCR-4 mRNA in vivo (in mouse versus human tissues) it does appear that CXCR-4 expression is somewhat more restricted in immortalized murine cell lines as compared to human counterparts (71). However, there are examples of both murine and feline cells that contain abundant CXCR-4 mRNA but after CD4 expression, remain resistant to T-tropic HIV-1(74, 75). Species specific post-translational processes are also unable to account for this anomaly, since transfection of CXCR-4 negative fibroblasts with CD4 plus either human or murine CXCR-4 renders them permissive for HIV-1 fusion or infection (71, 72). It remains unclear why some cells remain resistant to infection when they apparently express a full complement of receptors that are capable of supporting fusion/infection in the context of a different cell line of the same species. Indeed, human macrophages express CXCR-4 but are not susceptible to T-tropic HIV-1 infection. It may be that the apparent need for relatively high levels of CD4 expression by CXCR-4 utilizing HIV-1 strains is a limiting factor for infection in some cases (76). Alternatively, it is possible that there are cell type specific, post-fusion blocks to virus infection (77, 78). Although a rat CXCR-4 homolog was found to be a functional coreceptor for the T-tropic isolate LAI, it is non-permissive for the T-tropic HIV-1 isolate NDK and the HIV-2 isolate ROD. Thus, human/rat CXCR-4 chimeras have been used to map functional determinants for these strains (79). Suprisingly, the sequence requirements for ROD and NDK are similar, in that both require the presence of a human CXCR-4 derived ECL-2, These two strains and LAI appear rather insensitive to deletions in the N-terminus, almost total removal of this region is required before complete loss of function is observed, although partial effects on NDK and ROD are evident with smaller deletions. A variant of ROD (ROD/B) which is capable of infecting cells expressing CXCR-4 in a CD4 independent manner shows a similar partial dependence on an intact N-terminal CXCR-4 domain (80). CXCR-4 chimeras with more distantly related 7TM receptors such as CXCR-2 and CCR-5 indicate an absolute requirement for ECL-2 in this context (65). However, chimeras that, in addition, contain the CXCR-4 N-terminus and/or ECL-1 are more active, and are recognised by a wider range of strains. Thus, it appears that envelope-CXCR-4 interactions are as complex and strain dependent as is the case with CCR-5. 5.4 Viral determinants of coreceptor recognition In many cases, transplantation of the V3 loop of HIV-1 can, in a absolute manner, change the tropism of a T-tropic to that of an M-tropic virus and vice versa (13-16). In fact, this phenotypic switch can be accomplished by minor changes in the V3 sequence, and correlates closely with a switch between CXCR-4 and CCR-5 coreceptor utilization (25, 61, 81). Furthermore, dua-tropic viruses that use both CXCR-4 and CCR-5 can be derived simply by manipulation of V3 loop sequences (81). Thus, the V3 loop is an excellent candidate for being a component of a coreceptor binding site. The additional observation that minor changes in V3 loop sequences influence the ability of envelopes to interact with chimeric coreceptors supports this hypothesis (61). For instance, the V3 loop sequences of ADA and Ba-L differ by a single amino acid, and while both are able to confer CCR-5 utilization on the T-tropic IIIB envelope, only the ADA V3 loop permits utililisation of an human/murine CCR-5 chimera containing the human CCR-5 N-terminus in an otherwise murine background. Similarly, two amino acid changes in the V3 loop of SIV mac236, influence the ability of this strain to recognise a panel of CCR5/CCR-2b chimeras. An example of the role of V3 sequences in determining coreceptor selection and influencing specificity for chimeric coreceptors is given in figure 4. Figure 4. The role of the V3 loop in determining coreceptor selection and in modulating envelope CCR-5 interactions. The ability of the T-tropic strain IIIB to use CXCR-4 but not CCR-5 is reversed by the substitution of V3 sequences with those of M-tropic strains ADA or BaL. IIIB strains containing V3 loops of either Ba-L or ADA are differentially able to utilise human/murine CCR-5 chimeras. A single amino acid change in the V3 loop influences the viruses ability to use a chimera containing a human N-terminal domain, but has little effect on utilization of a chimera containing human ECL-2 in an otherwise murine CCR-5 context. Nevertheless, the V3 loop is not the sole determinant of coreceptor recognition. In some cases, alternative tropisms can be mapped to non-V3 sequences (17-19), and while the V3 loop sequences of ADA and JRFL are identical, only the former envelope protein is capable of efficiently utilising BOB/GPR15 (33). In addition, the M-tropic Ba-L envelope does not efficiently utilise a human/murine CCR-5 chimera where only ECL-2 is of human origin. However, a chimeric virus containing a Ba-L V3 sequence in the context of the T-tropic IIIB envelope is able to use this chimera more efficiently (figure 4). This suggests, somewhat paradoxically, that the T-tropic IIIB envelope contains non-V3 sequences that facilitate interaction with the CCR-5 ECL-2. While these observations provide genetic evidence for the interaction of multiple envelope domains with multiple coreceptor domains, it remains true that binding sites on coreceptors and envelopes cannot yet be unambiguously identified. In the context of proteins that are capable of induced conformational change and for which little or no structural information is available, the effect of introducing sequence changes on the conformation of distal regions of envelope and coreceptor is not easily predicted. It is also important to recognize that there is considerable scope for the introduction of laboratory artifacts in such experiments: The very act of culturing virus in the laboratory imposes different selective pressures on viral populations that are likely to impact on the nature of their interaction with receptors. In addition there are examples of very minor envelope sequence changes drastically influencing coreceptor recognition (60, 67, 81), thus different results are likely to be obtained with cloned versus uncloned forms of the same virus isolate. Furthermore, the relative expression levels of coreceptors on transfected cells, as compared to the natural targets of virus infection, remain largely unexplored and clearly affect the efficiency of infection. Indeed, over-expression of some chemokine receptors render cells permissive for viral envelope induced cell fusion, while cells expressing lower levels of the same coreceptor remain resistant to infection by cell-free virus (82). Therefore, a degree of caution is warranted in interpreting the results of the aforementioned studies. Despite these caveats, it is abundantly clear that there is considerable plasticity in coreceptor usage, both in terms of the viruses ability to use one or more of a number of 'optional' coreceptors and the ability of strains to use a given coreceptor in different ways.

[Frontiers in Bioscience 3, d44-58, January 1, 1998]
Reprints
PubMed
CAVEAT LECTOR

CHEMOKINE RECEPTORS AND HUMAN IMMUNODEFICIENCY VIRUS INFECTION

Paul D. Bieniasz and Bryan R. Cullen

Department of Genetics and Howard Hughes Medical Institute, Duke University Medical Center, Durham, NC 27710

Received 12/15/97 Accepted 12/19/97

5. MECHANISM OF ACTION OF CORECEPTORS

Accumulating evidence indicates that the SU envelope glycoproteins of primate lentiviruses directly interact with their cognate coreceptors (52-56). Soluble gp120 from T-tropic HIV-1 has been reported to form a precipitable complex with CXCR-4 (52) and specifically bind to cells expressing CXCR-4 with an affinity in the nanomolar range (53). In addition, a few groups have demonstrated that HIV-1 and SIV envelope glycoproteins are able to compete with beta chemokines for CCR-5 binding (54-56). Significantly, this competition is much more efficient in the presence of CD4, either coexpressed with CCR-5 on the cell surface, or in soluble form as a pre-formed complex with gp120. The inference from these data is that the interaction of HIV or SIV with a target cell is initially with CD4, thereby inducing a conformational change in gp120 which facilitates subsequent binding of envelope to the coreceptor. This scenario is supported by the previous observation that a number of HIV-2 strains can be induced to infect CD4 negative cells by pre-treatment with sCD4 (57). In this case a virion/sCD4 interaction presumably induces a stable envelope conformation that enables interaction with a coreceptor in the absence of cell surface CD4. Why this phenomenon should be restricted to HIV-2 remains obscure, but perhaps sCD4 complexes with the gp120 of other primate lentiviruses adopt a coreceptor-binding conformation that is less stable than that obtained with HIV-2. The ability of HIV-2 to be adapted to efficiently infect CD4 negative cells, using either CXCR-4, CCR-3 or the orphan receptor V28 (57-59), probably represents an extension of this phenomenon. In this case, the viral envelope presumably adopts a stable CXCR-4 binding conformation in which the coreceptor binding site is revealed without the need for prior 'activation' by CD4.

5.1 Envelope binding sites on virus coreceptors

All of the identified HIV/SIV coreceptors are members of the seven transmembrane spanning G-protein coupled receptor family. As such, they contain an N-terminal extracellular domain and three extracellular loops (ECL-1-3). A number of studies have attempted to determine which of these extracellular domains are determinants of coreceptor function, and by implication, what are the amino acid residues that interact with the virus envelope (60-67). In most cases, the functional coreceptor phenotype cannot be mapped to a single linear sequence. It is therefore likely that the virus binding site or sites on the coreceptor involve multiple extracellular domains, and that within each of these domains multiple amino acids contribute to coreceptor function. While this complexity has largely frustrated attempts to define precise virus binding sites, studies of chimeric and mutant coreceptors have, to some extent, illuminated our understanding of how HIV/SIV interact with coreceptors and suggested that several, functionally redundant, interactions between coreceptor and virus envelope are likely to occur. As we will discuss, this has implications both for the feasibility of using coreceptor targeted inhibitors of virus entry as therapies and for understanding the evolution of coreceptor usage in vivo.

5.2 Determinants of CCR-5 coreceptor function

Historically, the generation of chimeric molecules consisting of a functional retrovirus receptor and a closely related, but non-functional homologue has proven to be a successful strategy for rapidly mapping the determinants of virus receptor function. Several studies have attempted to map the extracellular determinants of CCR-5 coreceptor function using such an approach (60-66). Fortuitously, a murine homologue of CCR-5 had been previously identified that is 82% identical to the human receptor but non-functional as a coreceptor for a all HIV-1 strains tested so far. At least four groups have used human/murine CCR-5 chimeras to map functional domains (60-63). There is broad agreement in the general conclusions among these three studies, some discrepancies are probably the result of methodological difference, details in the derivation of the chimeric coreceptors and polymorphisms in murine CCR-5 alleles. Of the four extracellular CCR-5 domains, human and murine proteins differ in sequence in three. Remarkably, each of these extracellular human CCR-5 domains is independently capable of conferring some degree of coreceptor function on an otherwise inactive murine receptor (60-62). Conversely there is no single extracellular domain of human CCR-5 that cannot be functionally substituted by its murine counterpart. (at least in the case of most murine CCR-5 alleles) Importantly, these chimeric coreceptors are often functional for a significantly more restricted range of HIV-1 strains; the differential abilities of virus strains to utilize various chimeric receptors strongly implies that different virus strains interact with CCR-5 in different ways (61, 62). Examples of the strain dependent differences in the ability of HIV-1 isolates to utilize human/mouse chimeric CCR-5 receptors are presented in figure 2.

Figure 2. Examples of differential utilisation of human/murine CCR-5 chimeras by M-tropic and dual-tropic HIV-1 strains. The relative ability of human/murine CCR-5 chimeras to support fusion induced by the envelopes of two M-tropic (ADA and Ba-L) and one dual-tropic (89.6) HIV-1 strain is indicated. While ADA is efficiently able to utilize chimeras containing a single human extracellular domain, BaL requires the presence of two, and 89.6 requires an intact human CCR-5 molecule.

Some strains, in particular those which are dual- tropic (i.e. are also capable of using CXCR-4 as a coreceptor) appear to be particularly sensitive to perturbations in the human CCR-5 extracellular domains. In fact, none of the three human CCR-5 extracellular domains that differ in sequence to the murine homologue can be replaced by murine sequences without (at least partially) compromising dual-tropic coreceptor function (61, 62). This is a suprising observation, since human and mouse CCR-5 receptors are much more similar in sequence than are CCR-5 and CXCR-4, a fully functional coreceptor for dual tropic HIV-1 strains. In contrast the M-tropic strains ADA and SF162 and are able to quite efficiently utilize chimeric coreceptors where any one of the 3 divergent extracellular domains is of human origin (61, 62, figure 2). Other M-tropic strains such as Ba-L, M23 and E80 have intermediate properties, in that a larger number of chimeras are functional than for dua-tropic strains, but some chimeras which are functional ADA or SF162 coreceptors do not support fusion mediated by these additional envelopes. For example, while no single human extracellular domain is sufficient to confer high level Ba-L coreceptor function on murine CCR-5, (figure 2) inter-domain synergy can be observed; chimeras containing two extracellular domains of human CCR-5 origin exhibit near wild type Ba-L coreceptor function, while each individually has little or no activity in the context of an otherwise murine receptor.

Interestingly the results obtained using a CCR-5 allele from an NIH Swiss mouse for chimera construction differ significantly from those obtained using other murine CCR-5 alleles (63). It is apparent that murine CCR-5 polymorphisms are largely responsible for this discrepancy, and implicate specific amino acids in the envelope/CCR-5 interaction (see below). It is noteworthy, therefore that given the complexity of envelope coreceptor interactions, in particular the participation of multiple coreceptor domains, non-functional CCR-5 homologues are unlikely to provide a neutral background for analysis using chimeras or site directed mutant receptors. Thus, which extracellular domains in CCR-5 that are 'flagged' as being important for coreceptor function is dependent not only on the virus strain, but also on the non-functional partner selected for the generation of chimeras.

This premise is largely born out when the above studies are viewed in conjunction with others which have employed CCR-5/CCR-1, CCR-5/CCR-2b, CCR-5/CXCR-2 and CCR-5/CXCR-4 chimeras to map determinants of coreceptor function for both HIV-1 and SIV strains (64-67). However, it is clear that the N-terminal extracellular domain of CCR-5 is capable of conferring coreceptor function for some HIV-1 M-tropic and dual-tropic strains on multiple, otherwise inactive, 7TM receptors (65). Loss of function as a result of partial truncations in the CCR-5 N-terminus, in the context of a CCR-5/CCR-2b chimera, support the role of this domain in coreceptor function, although the particular amino acids involved are clearly strain dependent (64). As with human/murine CCR-5 chimeras however, the N-terminus is dispensable in an otherwise intact human CCR-5 molecule, since it can be truncated or replaced with those of CCR-2b, CCR-1, CXCR-2 and CXCR-4 without drastically compromising M-tropic coreceptor function (64, 65). Dual-tropic strains are again more fastidious than M-tropic strains in their ability to interact with N-terminally substituted CCR-5 chimeras; of the aforementioned 7TM receptors, only the CCR-1 N-terminus can functionally replace that of CCR-5, and then only for one of the two dua-tropic strains analysed (65). These data have been interpreted as suggesting that dual tropism evolves from M-tropic viruses that retain the ability to interact specifically with the N-terminus of CCR-5 while acquiring the ability to interact with the ECLs of CXCR-4 at the expense of interaction with CCR-5 ECLs. We would argue that the interaction of dua-tropic envelopes with CCR-5 also remains critically dependent on CCR-5 ECLs, since replacement of these with corresponding murine CCR-5 sequences results in a dramatic loss of coreceptor function (61, 62, figure 2).

Mutational analyses of CCR-5 extracellular domains in the context of an otherwise intact molecule has not proven to be particularly informative for M-tropic HIV-1, since many point mutants retain substantial coreceptor function (66). Nevertheless, partial or total loss of dua-tropic coreceptor function (dependent on the particular strain) as a result of an aspartate to alanine mutation at position 11 in the N-terminus has been documented (66). The partial effect of this mutation for the 89.6 strain is markedly enhanced when combined with alanine substitution of lysine 197 (ECL2) and/or aspartate 276 (ECL-3). Again this, indicates that the CCR-5 N-terminus is a critical, but not the sole, determinant of dua-tropic coreceptor function.

In the case of M-tropic HIV-1 strains, studies in our laboratory have exploited human/murine CCR-5 chimeric coreceptors whose activity is dependent on the presence of a single human CCR-5 extracellular domain (68). Residues of human origin were substituted, either individually or in combination, with corresponding murine sequences. These experiments indicate an important role for residues serine 7, asparagine 13, and tyrosine 15 in the N-terminus and serine 180 in ECL-2. In addition, individual amino acids were identified (for example tyrosine 184 and serine 185) that had little or no effect on coreceptor function when mutated individually, but dramatic losses in coreceptor result when both are substituted. Thus, it is evident that not only do multiple CCR-5 domains contribute to coreceptor function, but within these domains multiple amino acids play an important role.

In addition, the identification of non functional CCR-5 alleles from African green monkey and NIH Swiss mouse has indicated an especially critical role for tyrosine 14 in the N-terminus, glutamine 93 in ECL-1 and proline 183 in ECL-2 (63). Indeed, mutation of tyrosine 14 or proline 183 in an otherwise intact CCR-5 molecule renders it non-functional as a coreceptor for several M-tropic HIV strains.

A summary of the mutational analyses performed to date is depicted in figure 3. Although residues which are important for function are somewhat scattered throughout the extracellular domains, several important residues are closely clustered within the N-terminus and in ECL-2. As yet, we cannot determine whether these sequence motifs constitute spacially distinct envelope recognition domains, or form an single envelope binding site in the context of a folded protein. Alternatively some of these amino acids could be responsible for influlencing the overall structure of the CCR-5 extracellular domains and modulate coreceptor function without directly contacting the viral envelope.

Figure 3. Summary of mutational analyses of CCR-5. Amino acid residues that differ between human and murine (BALB/c) CCR-5 are indicated by shading. Arrows indicate ammino acids that when mutated either individually or in combination compromise coreceptor function for M-tropic (M) or dua-tropic (D) HIV strains. Note that some mutations are performed in the context of human/mouse CCR-5 chimeras, where coreceptor activity is dependent on the presence of a single human extraceellular domain. (data from refs 63, 66, 67)).

The CCR-5 sequences involved in recognition by other primate lentiviruses are less well characterized. However, in the context of CCR-5/CCR-2b chimeras, ECL-2 is required for recognition by several SIV strains (67). Like HIV-1 strains, different SIV strains use CCR-5 in different ways: For other SIV strains, ECL-2 is dispensable provided that an intact CCR-5 N-terminus is present.

5.3 Determinants of CXCR-4 coreceptor function

With the aim of adopting similar approaches to analyse structure function relationships in CXCR-4, we and others have cloned the murine homolog of this receptor (69-71). Unfortunately, murine CXCR-4 was found to constitute a functional coreceptor for at least some T-tropic and dua-tropic HIV-1 strains (71,72), precluding the construction of informative chimeras. In addition, both rat and feline CXCR-4 receptors were also found to support infection by at least some T- tropic HIV-1 strains (73, 74). These observations were suprising given the large body of literature indicating that almost all non-human cells remain refractory to HIV-1 infection when engineered to express human CD4, whereas most human cell lines are highly susceptible to T-tropic HIV-1 infection when CD4 is expressed (1-4). This apparent anomaly can only partially be explained by species specific differences in expression patterns. While we obtained no evidence for differential expression of CXCR-4 mRNA in vivo (in mouse versus human tissues) it does appear that CXCR-4 expression is somewhat more restricted in immortalized murine cell lines as compared to human counterparts (71). However, there are examples of both murine and feline cells that contain abundant CXCR-4 mRNA but after CD4 expression, remain resistant to T-tropic HIV-1(74, 75). Species specific post-translational processes are also unable to account for this anomaly, since transfection of CXCR-4 negative fibroblasts with CD4 plus either human or murine CXCR-4 renders them permissive for HIV-1 fusion or infection (71, 72). It remains unclear why some cells remain resistant to infection when they apparently express a full complement of receptors that are capable of supporting fusion/infection in the context of a different cell line of the same species. Indeed, human macrophages express CXCR-4 but are not susceptible to T-tropic HIV-1 infection. It may be that the apparent need for relatively high levels of CD4 expression by CXCR-4 utilizing HIV-1 strains is a limiting factor for infection in some cases (76). Alternatively, it is possible that there are cell type specific, post-fusion blocks to virus infection (77, 78).

Although a rat CXCR-4 homolog was found to be a functional coreceptor for the T-tropic isolate LAI, it is non-permissive for the T-tropic HIV-1 isolate NDK and the HIV-2 isolate ROD. Thus, human/rat CXCR-4 chimeras have been used to map functional determinants for these strains (79). Suprisingly, the sequence requirements for ROD and NDK are similar, in that both require the presence of a human CXCR-4 derived ECL-2, These two strains and LAI appear rather insensitive to deletions in the N-terminus, almost total removal of this region is required before complete loss of function is observed, although partial effects on NDK and ROD are evident with smaller deletions. A variant of ROD (ROD/B) which is capable of infecting cells expressing CXCR-4 in a CD4 independent manner shows a similar partial dependence on an intact N-terminal CXCR-4 domain (80).

CXCR-4 chimeras with more distantly related 7TM receptors such as CXCR-2 and CCR-5 indicate an absolute requirement for ECL-2 in this context (65). However, chimeras that, in addition, contain the CXCR-4 N-terminus and/or ECL-1 are more active, and are recognised by a wider range of strains. Thus, it appears that envelope-CXCR-4 interactions are as complex and strain dependent as is the case with CCR-5.

5.4 Viral determinants of coreceptor recognition

In many cases, transplantation of the V3 loop of HIV-1 can, in a absolute manner, change the tropism of a T-tropic to that of an M-tropic virus and vice versa (13-16). In fact, this phenotypic switch can be accomplished by minor changes in the V3 sequence, and correlates closely with a switch between CXCR-4 and CCR-5 coreceptor utilization (25, 61, 81). Furthermore, dua-tropic viruses that use both CXCR-4 and CCR-5 can be derived simply by manipulation of V3 loop sequences (81). Thus, the V3 loop is an excellent candidate for being a component of a coreceptor binding site. The additional observation that minor changes in V3 loop sequences influence the ability of envelopes to interact with chimeric coreceptors supports this hypothesis (61). For instance, the V3 loop sequences of ADA and Ba-L differ by a single amino acid, and while both are able to confer CCR-5 utilization on the T-tropic IIIB envelope, only the ADA V3 loop permits utililisation of an human/murine CCR-5 chimera containing the human CCR-5 N-terminus in an otherwise murine background. Similarly, two amino acid changes in the V3 loop of SIV mac236, influence the ability of this strain to recognise a panel of CCR5/CCR-2b chimeras. An example of the role of V3 sequences in determining coreceptor selection and influencing specificity for chimeric coreceptors is given in figure 4.

Figure 4. The role of the V3 loop in determining coreceptor selection and in modulating envelope CCR-5 interactions. The ability of the T-tropic strain IIIB to use CXCR-4 but not CCR-5 is reversed by the substitution of V3 sequences with those of M-tropic strains ADA or BaL. IIIB strains containing V3 loops of either Ba-L or ADA are differentially able to utilise human/murine CCR-5 chimeras. A single amino acid change in the V3 loop influences the viruses ability to use a chimera containing a human N-terminal domain, but has little effect on utilization of a chimera containing human ECL-2 in an otherwise murine CCR-5 context.

Nevertheless, the V3 loop is not the sole determinant of coreceptor recognition. In some cases, alternative tropisms can be mapped to non-V3 sequences (17-19), and while the V3 loop sequences of ADA and JRFL are identical, only the former envelope protein is capable of efficiently utilising BOB/GPR15 (33). In addition, the M-tropic Ba-L envelope does not efficiently utilise a human/murine CCR-5 chimera where only ECL-2 is of human origin. However, a chimeric virus containing a Ba-L V3 sequence in the context of the T-tropic IIIB envelope is able to use this chimera more efficiently (figure 4). This suggests, somewhat paradoxically, that the T-tropic IIIB envelope contains non-V3 sequences that facilitate interaction with the CCR-5 ECL-2.

While these observations provide genetic evidence for the interaction of multiple envelope domains with multiple coreceptor domains, it remains true that binding sites on coreceptors and envelopes cannot yet be unambiguously identified. In the context of proteins that are capable of induced conformational change and for which little or no structural information is available, the effect of introducing sequence changes on the conformation of distal regions of envelope and coreceptor is not easily predicted. It is also important to recognize that there is considerable scope for the introduction of laboratory artifacts in such experiments: The very act of culturing virus in the laboratory imposes different selective pressures on viral populations that are likely to impact on the nature of their interaction with receptors. In addition there are examples of very minor envelope sequence changes drastically influencing coreceptor recognition (60, 67, 81), thus different results are likely to be obtained with cloned versus uncloned forms of the same virus isolate. Furthermore, the relative expression levels of coreceptors on transfected cells, as compared to the natural targets of virus infection, remain largely unexplored and clearly affect the efficiency of infection. Indeed, over-expression of some chemokine receptors render cells permissive for viral envelope induced cell fusion, while cells expressing lower levels of the same coreceptor remain resistant to infection by cell-free virus (82). Therefore, a degree of caution is warranted in interpreting the results of the aforementioned studies. Despite these caveats, it is abundantly clear that there is considerable plasticity in coreceptor usage, both in terms of the viruses ability to use one or more of a number of 'optional' coreceptors and the ability of strains to use a given coreceptor in different ways.