Several approaches have been used to study HIV pathogenesis in vivo. The availability of non-human primate models has largely advanced our understanding of the field. Studies with animal models have highlighted the importance of the so-called viral "accessory genes" in HIV disease progression. These genes were initially deemed non-essential in in vitro studies because the virus would be able to replicate despite their removal from the viral genome 9. Despite the usefulness of animal models to define viral determinants of pathogenesis, the genetic differences between human and non-human primates, have made the latter less amenable for the study the role of host factors.

Long-term nonprogressors (LTNP) have provided a unique opportunity to study the mechanisms of HIV disease. LTNPs are HIV-infected individuals who have lived free of symptoms for extended periods of time, in the absence of antiretroviral treatment. A standard criterion for LTNP status is to have had a documented infection for ten years or more, stable CD4-positive T cell counts above 500 cells/ml, and plasma viral load below 10,000 RNA copies/ml. Depending on the definition of "nonprogression" used, this population has been estimated to represent 2–4% of all infected patients 10. The recruitment of LTNP cohorts is a formidable task, because until recently, most patients with well documented clinical histories had been treated before the onset of symptoms.

An additional approach to examine disease progression is to investigate highly exposed uninfected (EU) individuals. EUs are subjects who resist HIV infection and seroconversion, despite being at high-risk for transmission. EU cohorts have been gathered from groups of intravenous drug users (IDU), sex workers, children born to seropositive mothers, individuals performing unprotected sex with multiple partners, and health care workers undergoing accidental exposure to the virus 11.

Important insight into HIV pathogenesis can also be gained by studying the natural course of infection in seropositive patients. Clinical variables (decline in CD4 counts, increase in viral load) have been used to monitor the rate of progression to disease in untreated patients, or to establish prognosis in terms of virologic and immunologic success in patients following antiretroviral regimes. These variables can be statistically associated with host genotypic variants or specific phenotypic traits.

Finally, the study of healthy HIV-seronegative patients who may bear genetic markers of interest, can also shed light into the mechanisms of HIV pathogenesis. The role of cellular factors influencing HIV replication and immunity can be addressed by exposing primary cells from healthy seronegative individuals to virus in vitro. Likewise, statistical associations between haplotypes or single-nucleotide polymorphisms (SNP) can be drawn by monitoring the extent of viral replication in vitro. When available, genetic associations with the rate of replication in these ex-vivo models can also be validated with in vivo data monitoring disease progression 12.

Factors influencing susceptibility to infection and the course of disease can be grouped into three categories: 1) viral factors determining the replicative properties of the virus or its ability to escape immune responses; 2) cellular factors modulating the innate or acquired immune responses to infection; and 3) cellular factors co-operating with viral products that govern the ability of the virus to replicate in human cells. Thus, the rate at which an untreated HIV-infected patient progresses to AIDS may be explained by a combination of these factors, which ultimately dictate how fast HIV replicates and/or how efficiently it overcomes the immune defenses posed by the host. Below, we will discuss cellular factors influencing various steps of the HIV life cycle, and those modulating the innate immune response. The role of the HLA system in AIDS progression is beyond the scope of this review, and has been amply discussed in other reviews 131415.

Entry into target cells occurs by a multi-step process that culminates with the fusion of viral and cellular membranes. HIV-1 utilizes CD4 as its primary receptor. Binding to CD4 is followed by conformational changes in the viral envelope that lead to the engagement of one of the viral co-receptors (CCR5 or CXCR4) 16. Based on their functionality in vitro, other chemokine receptors may also work as HIV-1 co-receptors. Among them are CCR2, CCR3, CCR8, CCR9, CXCR6, CX3CR1, ChemR23, APJ, and RDC1. However, CCR5 and CXCR4 constitute the major co-receptors in vivo (for a recent review see 17).

CXCR4 was the first co-receptor identified 18. Afterwards, the role of CCR5 in HIV-1 infection was soon elucidated 1920212223. Soon after these findings, researchers sought genetic polymorphisms that could confer protection to HIV-1 infection 1920212324. These studies characterized the CCR5Δ32 allele, which has been unequivocally associated with protection to HIV-1 infection in homozygotic individuals 242526. This discovery provided the first conclusive evidence for the existence of genetic resistance to HIV-1 infection. CCR5Δ32 expresses a truncated co-receptor that is not transported to the cell surface and thus is incompetent for viral entry 27. Individuals homozygous for the Δ32 allele seem to have a normal life expectancy, although immunological differences have been reported, which may influence the outcome of infections with other pathogens, such as West Nile and hepatitis C virus (reviewed in 1728). The apparent lack of immunological dysfunction in individuals with the homozygousΔ32 genotype may be explained by the functional redundancy in the chemokine receptors and their ligands. The CCR5Δ32 allele occurs at a frequency of 4–15% in the Caucasian population, with higher frequencies in Northern European populations. However, the CCR5Δ32 allele is rarely found in Asians and Africans. Approximately 1% of Caucasians carry two copies of the Δ32 allele 29. These individuals are overrepresented in cohorts of high-risk HIV-seronegative individuals (EUs) 253031. Protection against HIV-1 infection in homozygous CCR5Δ32 individuals, however, is not complete. Although rare, infections of homozygous CCR5Δ32 have been reported, but always in patients infected with virus strains utilizing the CXCR4 co-receptor 323334353637. Other studies have reported increased frequencies of CCR5Δ32 heterozygotes among LTNP, and in patients progressing to disease at slower than normal rates 38394041. However, these associations have not been observed in all cohorts, suggesting that the CCR5Δ32 allele alone may not universally slow progression to AIDS 42434445. Genetic differences among the ethnicities evaluated, or different transmission routes in the studied cohorts may explain these discrepancies.

The observation that high level of CCR5 expression on CD4-positive primary T cells is associated with high viral loads and accelerated disease progression further highlights the contribution of CCR5 to disease progression 4647. Generally, lymphocytes from Δ32 heterozygous individuals express lower surface levels of CCR5, as compared to those observed on cells from individuals homozygous for the wild type gene 48. In the previous CCR5 expression in heterozygous individuals was lower than the expected 50%, relative to wild-type homozygous. This observation led the authors to hypothesize that the truncated co-receptor may dimerize with the full-length protein and interfere with its transport.

Other mutations in the CCR5 coding region have been described. Some of them introduce frame-shifts that result in truncated proteins that, similarly to the Δ32 variant, fail to be transported to the cell surface (e.g., FS299). Other mutations (e.g. C20S and C269F) affect the formation of disulfide bridges, altering surface expression of the receptor and the ability to bind ligands 49. Some mutations result in undetectable or very low levels of surface receptor (C269F, G106R, C101X). Unlike Δ32, most of these CCR5 variants are relatively rare (frequencies below 1–2%) and only present in specific populations. Thus, their role in HIV-1 disease progression has not been properly established 174950. Interestingly, the rare C20S, C101X (also called m303), and T303A alleles are over-represented in EU individuals also carrying the more common CCR5Δ32 allele 5152. These findings suggest that alleles that may, in the context of a wild type allele, have a weak effect, may exert a more profound protection in combination with CCR5Δ32.

Genetic variants have also been described in the CCR5 promoter. Most changes are single base substitutions that could potentially alter the level of expression. CCR5P1, one of the multi-site haplotypes identified in the CCR5 promoter, is composed of 13 distinct SNPs. This haplotype has been associated with faster progression to AIDS in individuals carrying the wild type coding region for CCR5 and CCR2, and homozygous for CCR5P1 53. A similar association with rapid progression has been reported for an A/G polymorphism located in the first CCR5 intron. Individuals containing A in both copies (59029-A/A) progress more rapidly to AIDS 54. Interestingly, the Δ32 phenotype seems to be largely influenced by the presence of mutations in the CCR5 promoter region, with some combinations resulting in poor co-receptor expression and protection against HIV-1 transmission 55.

Antibodies against CCR5 may provide another mechanism of interference with CCR5 function in vivo. Anti-CCR5 antibodies have been reported in a subset of LTNP (24%) but not in other populations studied. These antibodies induce internalization of the co-receptor in vivo and block HIV entry of R5 strains in vitro 56. Anti-CCR5 antibodies have also been detected in the milk of 66% and 83% of HIV-seronegative and seropositive women, respectively. Anti-CCR5 antibodies purified from these women also protect against infection of R5 strains in vitro 57. These findings suggest that some degree of protection against vertical transmission of HIV-1 may be mediated by anti-CCR5 antibodies present in the milk. The mechanism governing induction of anti-CCR5 antibodies in humans is unknown, but these observations underscore the interesting prospect of preventing HIV transmission with vaccines targeting the CCR5 co-receptor, an approach that is being examined in murine models 58.

The compelling evidence supporting the role of CCR5Δ32 in protection from disease in homozygous individuals, the apparently healthy characteristics of these individuals, and the ubiquitous presence of CCR5-tropic HIV-1 strains throughout most of the disease, have prompted efforts to target CCR5 with novel antiretroviral therapies. To date, at least nine small-molecule inhibitors and monoclonal antibodies are under development, being tested in clinical trials, or awaiting imminent FDA review 5960. Aplaviroc, maraviroc, and vivriviroc are noncompetitive, allosteric CCR5 antagonists that have been tested in large clinical trials. Unlike CCR5 agonists (e.g. PSC-RANTES), none of these compounds induces signaling through CCR5 or receptor internalization. There are potential problems associated with the treatment with CCR5 antagonists. For example, use of this therapy may lead to the emergence of CXCR4-tropic strains, which could accelerate disease progression 61. In addition, CCR5 inhibition could interfere with the normal immune and inflammatory responses. Despite the apparent normal phenotype of CCR5Δ32 homozygotic individuals, it is not clear how interference with CCR5 will affect the already impaired immune systems in HIV-infected patients. Thus, many questions need to be answered before CCR5 inhibitors are safely used in humans.

Despite the pivotal role of CCR5 as the major co-receptor for HIV-1, polymorphisms in other chemokine receptors appear to also exert a certain degree of protection against HIV-1 infection and/or disease progression. The most compelling evidence comes from the CCR2-64I variant, an allelic variant in which isoleucine 64 is replaced by valine 62. Heterozygous individuals for CCR2-64I progress slower to AIDS, although no clear effect in protecting against HIV-1 infection has been documented. Not all studies have confirmed this association 4363, and the effect of the CCR2-64I remains controversial. Intriguingly, the CCR2 receptor is used only by a few strains in vivo, thus the mechanism of action of the CCR2-64I variant is unknown. CCR2 lies 17.5 kb upstream of the CCR5 promoter, and it has been suggested that the 64I variant may be in linkage disequilibrium with genetic variations in the CCR5 region 64. To date, most reports have observed no changes in the levels of CCR2 or CCR5 surface expression in CCR2-64I individuals 6465. One study, however, reported lower levels of surface CCR5 and suggested, though it did not prove, that CCR2-64I may bind with increased affinity to CCR5 intracellularly and thus interfere with the expression of CCR5 at the cell surface 66. Another report suggested interference with CXCR4 as an alternative explanation, demonstrating that the 64I gene product dimerizes with CXCR4 more efficiently than the wild-type CCR2 67.

CX3CR1, the receptor for the chemokine fractalkine 68, has also been associated with HIV-1 disease progression. Its role as a co-receptor for HIV-1 in vivo is not clear, but it has been suggested that CX3CR1 could affect HIV-1 replication by influencing the recruitment of immunomodulatory cells. Two SNPs have been identified that form part of the haplotype I249-M280. Originally, this haplotype was found at higher frequency in a cohort of Caucasian HIV-infected patients progressing to AIDS faster than normal 69. These results were later confirmed with a cohort of HIV-infected children 70. In another study evaluating individuals entering HAART treatment during one year, the I249 polymorphism was found at a higher frequency among those displaying early immunological failure, estimated as a decline in CD4 counts 71. A study with a Spanish cohort found the haplotype composed of I249 and T280 was overrepresented among LTNPs, as compared to normal progressors, but the same study reported no significant effect on the distribution of the M280 SNP 72. Adding controversy to the role of CX3CR1, the results with the M280 SNP have not been confirmed in other studies 737475. A possible explanation has been proposed to explain these discrepancies, as follows. Due to the deleterious effect of the M280 allele, this SNP may have disappeared from some cohorts due to premature death of patients before recruitment. The low frequency of the allele observed in a French cohort supports this explanation 76. Additional studies will be needed to address the importance of CX3CR1 in HIV-1 pathogenesis and to understand the biological basis behind the observed phenotypes.

The HIV-1 strains that are most often responsible for transmission utilize CCR5. These so-called M-tropic (R5) viruses predominate during the asymptomatic stage and infect CD4-positive lymphocytes and macrophages. In approximately half of the patients with advanced disease, the viral quasiespecies are dominated by viruses that utilize CXCR4 77. These viruses are called T-tropic (or X4) and infect macrophages inefficiently 7879. The emergence of X4 viruses ("phenotypic switch") is associated with accelerated decline in CD4-positive lymphocyte counts and faster progression to disease. Thus, the phenotypic switch has been thought of as a causal factor leading to accelerated disease. In support of a role for X4 viruses in disease progression, studies with macaques infected with SIV carrying CXCR4-tropic HIV-1 envelope (SHIV chimera) display rapid loss of CD4 counts and develop AIDS faster than the R5 counterpart 80.

It is possible that the emergence of X4 viruses may be the consequence, rather than the cause of immune deterioration and disease progression 77. Supporting this idea, not all patients who develop full-blown AIDS experience the switch to X4 viruses, and yet the R5 strains present in these individuals late during disease are more pathogenic than early viruses 81. The mechanisms governing HIV co-receptor switch are poorly understood. Furthermore, it is not clear whether the so called "switch", emergence of X4 tropic viruses, occurs by acquisition of mutations in R5 envelopes, or rather X4 viruses are transmitted during infection but replicates poorly during the asymptomatic stages. The emergence of viruses with dual tropism (R5-X4) suggest the accumulation of gradual changes. More sensitive phenotypic assays able to detect very small fractions of X4 and dual-tropic viruses will allow us to understand how viral tropism changes throughout infection. Several selective forces have been suggested to explain why the emergence of X4 strains is restricted during initial phases of infection (reviewed in 82). High levels of CCR5-positive activated and memory cells present in gut-associated lymphoid tissue, an important site during acute infection, may provide fertile ground for the proliferation of R5 tropic strains. In addition, the constitutive levels of expression of SDF-1 in mucosal tissues could act to restrict transmission of X4 viruses 83. This, however, contradicts the observation that parenteral transmission of HIV-1 also results in the early predominance of R5 viruses. Clearly, some selective forces must keep X4 viruses under control at early stages of infection. Later in infection, the selective pressure achieved by increased levels of β-chemokines (active against R5 strains), and the reduced levels of neutralizing antibodies to which X4 tropic viruses are more sensitive 84 may trigger the phenotypic switch. The observation that many pathogens are potent inducers of HIV-suppressive β-chemokines suggests that opportunistic infections could contribute to the appearance of X4 strains during the symptomatic stage 85.

The beta-chemokines MIP-1α(CCL3), MIP-1β(CCL4), and RANTES (CCL5) are the natural ligands of CCR5. Two additional variants named CCL3L1 and CCL4L1, encoded by genes arising from the duplication of CCL3 and CCL4, respectively, have also been described 86. The role of β-chemokines in HIV infection was first proposed in a seminal article in which the anti-HIV-1 effect of these molecules was reported just a few months before the discovery of CCR5 as co-receptor for HIV-1 87. Soon thereafter, several reports found inverse correlations between the levels of β-chemokines in plasma and the rate of disease progression 8889. Elevated levels of RANTES have also been associated with protection against HIV transmission in some EU cohorts 90. Interestingly, no differences in the plasma levels of beta-chemokines have been observed in some EU cohorts. These individuals display normal levels of CCR5 on their CD4-positive T cells. However, CD4-positive T cells from these EUs appear more sensitive to the HIV-1 inhibitory effect of β-chemokines, suggesting the existence of yet unknown mechanisms influencing the role of CCR5 in infection 91.

When bound to CCR5, β-chemokines induce internalization of the receptor, abrogating its ability to promote HIV-1 infection 92. The mechanisms governing inter-individual variations in β-chemokine expression are not completely understood. In support of a role for β-chemokines, SNPs in the MIP-1α gene are found at elevated frequencies among EU individuals 93. SNPs in the promoter regions of the RANTES gene have also been described and they could affect expression of RANTES. However, their role in disease progression remains controversial 94.

CCL3L1, also known as MIP-1αP, is the more potent CCR5 agonist and the strongest inhibitor of infection by R5 HIV-1 strains 95. Interestingly, the levels of CCL3L1 are determined, in part, by the number of tandem copies of the CCL3L1 gene, which varies from 2–10 among individuals, with highest copy numbers found in African populations 96. When analyzed by itself, the number of CCL3L1 copies is not significantly associated with HIV susceptibility. However, significant trends are found when the number of copies is analyzed in the context of a specific population. Thus, people with higher number of copies than their ethnic background average are less susceptible to HIV-1 infection and progress slower to AIDS 96. These findings underscore the important role of CCL3L1 in disease progression.

Another study has revealed a role for MIP-1β in HIV-1 pathogenesis 97. This chemokine is encoded by two highly related genes (CCL4 and CCL4L1). Two polymorphisms of the CCL4L1 gene have been described (L1 and L2). Individuals homozygous for L2 display reduced levels of CCL4 transcripts when compared to homozygous for L1. A higher frequency of the L2 allele has been observed in a Spanish cohort of HIV-infected patients, as compared to healthy controls, suggesting that the levels of MIP-1β may influence HIV-1 susceptibility 97. Polymorphisms in the RANTES gene have also been reported. Elevated levels of circulating RANTES are common in EU individuals and in HIV-seropositive patients displaying slow progression (reviewed in 9899). Two changes in the promoter (-403G/A and -28C/G) appear to modulate in vitro transcription of RANTES. In vivo, the presence of the -403A and -28G haplotype has been associated with slower disease progression in Japanese and Thai cohorts 100101, and lower susceptibility to infection in a Chinese cohort 102. A second study has analyzed only the -403A allele, confirming its protective role in HIV progression, but also describing it as a risk factor for HIV transmission 103. Thus, the role of polymorphisms in the RANTES promoter remains controversial, with at least two other reports failing to confirm these associations in Spanish cohorts 94104. The reason for these discrepancies may be due to the quite different distribution frequencies of RANTES alleles across ethnic groups. Further complicating the study of RANTES polymorphisms, some alleles appear to mitigate the effect of others. Thus, the -28G allele, common in Asians, rare in Euopean-Americans (EA) and absent in Africans, mitigates the disease-accelerating effects of another RANTES variant, In1.1.C, which have been described in EAs 105. This latter allelic variant by itself potently down-regulates RANTES transcription 106.

SDF-1 (also known as CXCL12) is the only known ligand of CXCR4 107108. As with β-chemokines and CCR5, occupation of CXCR4 by SDF-1 induces internalization of the receptor 109. Both CXCR4 and SDF-1 are essential during development, and knock out of either of these genes leads to lethal phenotypes in mice 110111. Not surprisingly, alleles leading to lack of expression of SDF-1 or CXCR4 have not been identified. Nevertheless, the role of of SDF-1 and CXCR4 in the adult life, recirculating leukocytes and hematopoietic precursors, may be less vital. A polymorphism in the noncoding region of SDF-1 has been reported (SDF1-3'A). In the homozygous form, the presence of an A at position 801 has been associated with slower progression to AIDS, as compared to heterozygous or wild-type homozygous. The biological basis for this association is not clear, since no differences in SDF-1 levels have been observed 112113. A number of reports have failed to confirm this association in other cohorts 114115116117. Thus, it is not clear whether the SDF1-3'A variant play a role disease progression.

Other chemokines such as MCP1 (CCL2), MCP3 (CCL7), and eotaxin (CCL11) bind CCR2 and CCR3, but not CCR5. Each of these chemokines have been associated with HIV pathogenesis. It has been suggested that they control migration of immune cells to sites of HIV-1 infection, thus contributing to virus propagation in vivo 118. Table 1 summarizes the role of known variants of chemokine receptors and their ligands in HIV pathogenesis.

DC-SIGN is a mannose-binding, calcium-dependent lectin that has been involved in transmission of HIV-1 from dendritic cells (DC) to T lymphocytes, a phenomenom named "trans-enhancement" (reviewed in 119120). DC-SIGN is expressed on immature DCs and activated B lymphocytes. Trans-enhancement requires binding of HIV-1 particles to DC-SIGN via the high mannose glycans present in gp120. The mechanism of transfer of HIV-1 to T cells remains controversial. First, it was proposed that transfer requires internalization and transient storage of HIV-1 particles in subcellular compartments 121. Recent evidence suggests that infection of DC cells is required for efficient transfer of HIV-1 to other cells 122.

DCs are thought to be among the first cells infected by HIV on the genital mucosa. Infected DCs migrate to lymph nodes where they transfer viruses to T cells. By infecting the very same cells implicated in protection against infection in mucosal tissue, HIV-1 utilizes DCs as Trojan horses that spread the virus to the lymph nodes. The role played by DCs in facilitating infection suggests a possible role for DC-SIGN variants and other C-type lectins in HIV-1 disease progression and transmission. A polymorphism in the DC-SIGN promoter at positions -336 has been identified. Individuals at risk of HIV carrying the -336C allele are more susceptible to infection than persons with the -336T variant 123. This association, however, has been observed for parenteral transmission of HIV-1, but not for mucosally acquired infection. Variants in the coding region of DC-SIGN and DC-SIGNR have also been identified. However, the importance of these alleles in protecting from HIV-1 infection has yet not been fully elucidated 124125126.

Langerin, also called CD207, is selectively expressed in Langerhans cells, which are spread over the mucosa through which HIV transmission occurs. Under some experimental conditions, Langerhans cells are infected with HIV-1 and transmit virions to T cells 127. However, recent evidence suggests that Langerin, in contrast to DC-SIGN, prevents HIV-1 transmission. HIV-1 particles captured by Langerin are internalized and degraded into Birbeck granules 128. Thus, Langerhans cells appear to present a first barrier against infection. This study does not exclude the possibility that these cells transmit HIV-1 at high viral inocula 129. The role of Langerin variants in HIV-1 transmission has not been studied, although a mutation in the langerin gene in a person deficient in Birbeck granules has been described 130.

In addition to DC SIGN, other C-type lectin receptors may also act as receptors for HIV-1: DC-SIGN-related (DC-SIGNR), the mannose receptor (MR), and Langerin can also bind gp120 120.

Defensins are small antimicrobial and antiviral cysteine-rich cationic peptides produced by leukocytes and epithelial cells. The role of defensins in innate immunity to fight bacterial, viral and fungal infections has long been known 131132(reviewed in 133134135). The first report on the anti-HIV-1 activity of defensins dates back to 1993 136.

Mammalian defensins are classified into alpha-, beta-, and theta defensins, and differ in their size and distribution of disulfide bridges 135. Alpha and β-defensins are peptides typically composed of 30–45 residues, and both display anti HIV-1 activity 135. Theta-defensins are cyclic peptides composed of two alpha-like precursor peptides. Active theta-defensin products are found only in some non-human primates, and are typically composed of 16–18 residues 137. Given their smaller size theta-defensins have been included in the family of minidefensins, which include molecules also found in other species such as horseshoe crabs and spiders 138. In humans and chimpanzees theta-defensins are found only as inactive pseudogenes. These pseudogenes are transcribed into mRNAs, but they harbor premature stop codons that preclude expression of functional products. Interestingly, when the putative human ancestral gene for a human theta-defensin was reconstituted, it was found to display in vitro potent anti-HIV-1 activity against R5 and X4 strains 139. The artificially reconstituted product was named human retrocyclin-1. This molecule displays lectin-like properties and binds to CD4 and gp120, thus preventing entry of HIV-1 into target cells 140.

So far six human α-defensins (also known as human neutrophil peptides or HNP) have been identified. α-defensins also bind CD4 and the viral envelope glycoprotein. Treatment of permissive cells with α-defensins induces down-modulation of CD4 141. Additionally, in the absence of serum (e.g. at mucosal surfaces), α-defensins may inactivate virion particles by a mechanism that includes membrane disruption 142. These findings indicate that α-defensins block HIV-1 entry at several steps, by directly inactivating virions and by blocking or eliminating the viral receptor from the cell surface. The mechanism of action of α-defensins and their inhibition profile led researchers to suggest that these molecules could constitute the long-sought CD8-positive cell anti-HIV factor (CAF) 143. The existence of CAF was first suggested in 1986, as a soluble factor derived from CD8-positive cells in LTNP individuals with the ability to achieve durable immune responses controlling HIV-1 infection 144. A report in 2002 suggested that based on specific antibody recognition, α-defensins 1, 2, and 3 were responsible for the antiviral activity of the eluded CAF factor. Experiments demonstrated that CAF activity could be eliminated with anti-α-defensins antibodies, and α-defensins could be detected inside CD8-positive cells 143. Later reports failed to confirm these results and found no evidence for the production of α-defensins in CD8-positive cells, although CD8-positive cells may internalize α-defensins. However, it is not clear whether the uptake of these molecules is needed for their function 145146. Beta-chemokines produced by CD8-positive cells may also contribute to the activity of CAF, which may no longer be ascribed to a single host factor. It is likely that CD8-positive cells express an unknown array of novel HIV-suppressive factors. Thus, the search to elucidate the CAF factor activities is still ongoing.

Six human β-defensins have been identified in epithelial cells, although genomic searches indicate that up to 28 different genes may be present in humans 147. The mechanism of action of β-defensins shares some similarities with that of α-defensins. They block viral entry of both X4- and R5-tropic HIV-1 strains, although their effect is more potent with T-tropic isolates. Beta-defensins 2 and 3 achieve their inhibitory effect in part by direct inactivation of viral particles, and also by down-modulation of CXCR4, but not CCR5, on T cells 148. This latter mechanism, however, has not been confirmed by others 149. Interestingly, HIV-1 induces expression of human β-defensins 2 and 3, but not 1, the latter of which displays poor antiviral activity. Induction of these defensins occurs by a mechanism that does not require viral replication 148150151. Recent findings suggest that β-defensin 2 may mediate its effect via CCR6. Beta-defensin 2 binds CCR6, and its inhibitory effect is abrogated after treatment of cells with anti-CCR6 antibodies. In agreement with this, MIP-3-α(also known as CCL20), the cognate ligand for CCR6, blocks HIV-1 infection in a manner similar to β-defensin 2 152. CCR6 is not a co-receptor for HIV-1, but is expressed in CD4+CD45RO+CCR5+ lymphocytes and dendritic cells, where it may play an important role governing the movement of immune cells to mucosal surfaces, the first tissues encountered by HIV during sexual transmission. Thus, β-defensins may also control HIV-1 replication by modulating the immune system. The mechanisms of defense against HIV infection mediated by α-, β-, and θ-defensins are summarized in Table 2.

The demonstrated antiviral activity of defensins have encouraged the search for polymorphisms influencing HIV-1 infection and disease progression. Both, α-, and β-defensins have been found in human breast milk, suggesting that they could play a role in protecting infants from infection 153154. One study has associated the amount of α-defensins in breast milk with protection against intrapartum and postnatal transmission of HIV-1 155. SNPs in the 5'-untranslated region of the human β-defensin 1 gene (DEFB1) have been associated with HIV-1 infection in an Italian pediatric cohort. The presence of a C/C allele at position -44 in HIV-1 infected mothers and their children is associated with higher risk of maternal-fetal transmission 156157. These results have been confirmed by the same group in another cohort of Brazilian children 158. However, studies with HIV-1 infected adults are still missing. Further highlighting the role of human defensins in HIV pathogenesis, elevated levels of α-defensins have been reported in EU individuals and HIV-infected women, as compared to healthy controls 159. Interestingly, human genes for both α-defensins (DEFA1 and DEFA3), and β-defensins (DEFB4, DEFB103, and DEFB104) are known to be polymorphic in copy number 160161162. Transcription of these genes correlates with copy number, and it is tempting to speculate that, as demonstrated for the CCL3L1 and CCL4L1 chemokines, polymorphisms in the copy number of defensin genes may modulate HIV-1 susceptibility and disease progression.

Studies in the early '80s identified cyclophilin A (CypA) as a cytosolic protein binding to the immunosuppressant molecule cyclosporin A 163. Subsequent studies identified an anti-HIV activity of cyclosporin A 164165, although its mechanism of action was not elucidated at that time. The role of CypA in HIV-1 replication was investigated in detail after discovering its HIV-1 capsid (CA) binding properties in a yeast two-hybrid screen 166. As a result of this interaction CypA is incorporated into virion particles 167168169. After years of studies the mechanism of action of CypA has just begun to be unraveled. HIV-1 has a limited host range that appears to be explained in part by the existence of saturable inhibitory factors that block virus replication at early steps, before reverse transcription occurs 170171172. CypA promotes HIV-1 infectivity in target cells by a mechanism that does not require CypA incorporation into virions 173174. The exact mechanism of action remains an enigma. CypA appears to modulate the action of other restriction factor(s), perhaps by altering CA conformation in a manner that makes it less sensitive to their inhibitory effect (reviewed in 175). The nature of the proposed conformational change may relate to the cis-trans isomerization of peptidyl-prolyl bonds mediated by CypA. However, the significance of this activity has not been conclusively demonstrated 176. Interestingly, cyclophilins may constitute modulators of the innate immune response in other eukaryotic systems. Recently, an innate mechanism of defense against Pseudomonas syringae infection has been described in Arabidopsis. Upon infection, a plant cyclophilin activates a bacterial protease, which then triggers a cascade of events that activates a plant's defense response 177. Thus, cyclophilins may represent evolutionarily conserved mechanisms of innate defense.

Several SNPs have been identified in the human CypA gene (PP1A). Their associations with HIV-1 infection and disease progression in vivo have not been studied to date. However, one group has used an ex vivo approach to evaluate polymorphisms in PP1A. In this strategy, CD4 T cells were purified from healthy donors and challenged in vitro with HIV-1. The extent of viral replication was correlated with specific alleles, and when in vivo data is available, correlations with disease progression in a cohort of HIV-1 infected individuals were sought. Following this approach a polymorphism in the PP1A promoter (1650A/G) was associated with lower ex vivo virus replication in cells derived from PP1A homozygous individuals (AA), and slower disease progression in vivo 12. The ex vivo replication profile in cells carrying alleles previously associated with slow progression (CCR5Δ32 and CCR264I) or rapid disease progression (RANTES In1.1C) followed the expected trend, confirming the validity of this ex vivo approach to identify relevant alleles. Nevertheless, validation studies with other cohorts will be needed to define the role of CypA variants in HIV-1 pathogenesis.

Another cellular factor implicated in early steps of HIV-1 replication is TRIM5α. This protein was identified from a rhesus macaque library screened for simian factors restricting HIV-1 replication upon transfer into otherwise permissive human cells 178. Early studies suggested that the action of TRIM5α and CypA were somehow related, and as mentioned above, it was suggested that CypA protects CA from the deleterious effects of human TRIM5α 175. TRIM5α also binds HIV-1 CA, though apparently in a CypA-independent manner 179180181. An alternative model proposes that common restriction factors may be shared in the cascade of events that leads to inhibition of HIV-1 replication by TRIM5α, and promotion of infectivity by CypA 175.

Unlike simian forms, the human ortholog of TRIM5α appears to only modestly inhibit HIV-1 replication. This has led to the hypothesis that polymorphisms in the human TRIM5α gene might result in increased restriction and modulate HIV-1 infection. One recent report has identified a TRIM5α haplotype (hap 9) containing a nonsynonymous SNP at position 136 (R136Q) that occurs with higher frequency among HIV-infected subjects than in exposed seronegative individuals 182. None of the individual SNPs found in this haplotype influenced HIV transmission, suggesting that the genetic sequence responsible for the observed phenotype is in linkage disequilibrium with this haplotype, lying in TRIM5 or one of the TRIM genes adjacent to it. Another study evaluated the 136Q allele and found the frequency of this allele elevated in uninfected African-Americans, as compared to HIV-1 infected subjects. However, this correlation was not observed in individuals of European descent. In agreement with a role for TRIM5α, the 136Q variant exhibited slightly stronger inhibitory activity than the 136R counterpart 183. The same study reported the identification of two additional non-coding SNPs present in one TRIM5α haplotype that is found in HIV-1 infected patients at higher frequencies than in uninfected individuals. The above studies suggest that polymorphisms in TRIM5α may influence susceptibility to HIV-1 infection. Only one study has evaluated the role of common TRIM5α variants on disease progression. The authors analyzed a cohort of 979 HIV-infected patients and found only modest associations with the rate of CD4-positive T cell decline before the onset of treatment 184. The viral determinants for sensitivity to inhibition by TRIM5α lie in the viral CA region. In vitro, mutations in HIV-1 CA overcome the restriction of HIV-1 to replicate in simian cells. A study by the Schuitemaker group has evaluated the presence of TRIM5α escape mutants in HIV-1 infected individuals, as an indicator of TRIM5α-mediated inhibition. Interestingly, CA escape mutants emerging late in the infection process could be found in 14% of the infected patients (N.A. Kootstra and H. Shuitemaker, personal communication). These findings suggest the existence of selective pressure on the virus to avoid TRIM5α-mediated restriction.

Host cells are endowed with another mechanism to halt HIV-1 infection before integration occurs. The human apolipoprotein B mRNA-editing enzyme-catalytic polypeptide-like-3G (APOBEC3G), formerly known as CEM15, is an endogenous inhibitor of HIV-1 replication 185186187. In the absence of the viral protein Vif, APOBEC3G is incorporated into HIV-1 particles in the producer cell, and during reverse transcription deaminates cytosine bases to uracil in the negative-sense single-stranded DNA, resulting in G to A hypermutations in the complementary, positive sense DNA strand. This hypermutation leaves the viral cDNA vulnerable to degradation by nucleases. Those cDNAs that manage to integrate into the host chromosomes carry multiple mutations that likely result in aberrant viral products 188189190. Recent reports appear to suggest that APOBEC3G also exerts antiviral activities through mechanisms independent of its cytidine deaminase activity 191192. APOBEC3G can be found in cells in two different forms of low and high molecular-mass (LMM and HMM, respectively). APOBEC3G in LMM form is enzymatically active and restricts HIV-1 infection. The HMM form is a catalytically inactive ribonucleoprotein complex that appears to protect against mobilization of endogenous cellular retroelements such as Alu and hY 193194195. Endogenous retroelements can influence transcription of adjacent genes and induce genetic disease in humans by insertional inactivation 196197198. This APOBEC3G-induced mechanism of restriction of Alu and hY mobilization is mediated by sequestration of the RNA retroelements into HMM complexes, which are kept away from the L1-dependent retrotransposition machinery 193. APOBEC3G and related deaminases may also protect against other retroviruses and hepatitis B virus, which also relies on a reverse transcription step to complete its life cycle 186191.

The Vif protein binds APOBEC3G in virus producer cells and targets it for degradation in the proteasome, thus preventing APOBEC3G incorporation into virions 199200201. To induce APOBEC3G degradation Vif binds the cellular proteins Cul5, elonginB, elonginC and Rbx1, to form a cullin5-based E3 ubiquitin ligase complex that leads to polyubiquitination and ultimately to proteasomal degradation of APOBEC3G and APOBEC3F, which also displays anti-HIV activity 200202.

The importance of the APOBEC3G-mediated innate response against HIV-1 is underscored by the existence of both APOBEC3G and Vif genetic variants influencing disease progression. Studies with LTNP have revealed an association between low viral load and a serine residue at position 132 of Vif. When tested in vitro, HIV-1 carrying Ser132 displayed a five-fold decrease in replication in PBMC, as compared to virus containing the wild type allele (Arg) at the same position 203. Another study revealed the presence of a two-amino-acid insertion in the Vif protein of a mother's virus and her child's, both LTNP. This amino acid insertion results in an aberrant Vif protein that severely impairs replication when expressed in a recombinant HIV-1 204. Furthermore, Vif-defective proteins have been isolated from strains from other LTNP individuals 205. The reason why some of these LTNP Vif variants do not revert to encode a functional product is not understood. It is possible that Vif plays other roles in viral replication that constrain its ability to mutate. Alternatively, the genetic background of these LTNPs may have weakened the virus' ability to replicate and mutate in these subjects.

With regards to APOBEC3G, it is plausible that genetic variants with altered inhibitory activity or partially resistant to Vif might alter the course of HIV-1 disease. A variant of APOBEC3G common in African-Americans has been identified, which contains a non-synonymous substitution. This allele carries Arg instead of His at position 186 (186R) and is strongly associated with faster decline of CD4 T cells and accelerated progression to AIDS 206. This variant, however, appears to have the same in vitro inhibitory activity than the common 186H allele. An extensive analysis of a French cohort failed to identify significant associations between 29 APOBEC3G polymorphisms and disease progression, although discrepancies with previous studies may be explained by the different variables used to analyze disease progression 207. Another study identified an APOBEC3G variant (C40693T) in a cohort of EU Caucasians subjects. This allele was found to be strongly associated with increased risk of infection 208. A recent report revealed an inverse correlation between APOBEC3G mRNA levels and viral load in infected individuals, with the highest APOBEC3G levels observed in LTNP patients 209. However, no associations with specific variants in the APOBEC3G gene have been identified in these individuals. The controversy over APOBEC3G variants, and the lack of a potential mechanism to explain their phenotypes warrants the need for additional studies.

A recent study has evaluated polymorphisms in the CUL5 gene encoding the Cullin 5 protein. Haplotypes encompassing 12 SNPs were clustered into two different groups that segregated with different profiles of CD4 T cell decline rates and viral load 210. One individual SNP (SNP6 A/G) appears responsible for the rapid CD4 T cell depletion and progression to AIDS. This correlation was observed in African-Americans, but not found in European-American and Asian populations. Interestingly, a significant additive effect was observed between some of the SNPs found in CUL5 and the APOBEC3G-186H allele. This study highlights the role of Cullin 5 in HIV-1 pathogenesis. Additional studies will be needed to understand the roles of CUL5 variants. Table 3 summarizes the effect on HIV infection and progression to AIDS of known genetic variants of human genes participating in post-entry steps of the virus life cycle.

Many cellular proteins known to play important roles in HIV-1 pathogenesis are incorporated into virions. For instance, CypA, APOBEC3G, TSG101, and VPS28, which are required at different steps of the viral cycle, are encapsidated into virions 167189217238. Most cellular proteins are incorporated into viral membranes in a passive manner if they associate with areas of the cell membrane through which the virus buds (reviewed in 239). However, some proteins are specifically incorporated into viral membranes through interaction with viral products. Specific incorporation of cellular proteins inside virions or into viral membranes has generally been seen as suggestive of a functional role in the HIV-1 life cycle. Thus, it is expected that many other proteins known to associate with virions play a role in HIV-1 transmission and disease progression. Among these potential candidates are the integral membrane-spanning intracellular adhesion molecule 1 (ICAM-1). ICAM-1 expression drastically enhances HIV-1 infectivity, and makes virions more resistant to neutralization with gp120-specific antibodies or sera from HIV-infected patients 240241242243244245. Enhancement of infectivity can be blocked with anti-ICAM-1 mAbs, and is augmented in cells expressing higher levels of LFA-1, the ICAM-1 receptor. These findings suggest that HIV-1 may use an ICAM-1/LFA-1 dependent route to promote cellular entry and escape the humoral immune responses targeting the viral envelope protein. The interaction between ICAM-1 and LFA-1 may also play an important role in infection of CD4-positive memory T cells, which are considered a latent reservoir of the virus 246. Polymorphisms in the ICAM-1 gene have been associated with susceptibility to a range of degenerative and inflammatory diseases. However their role in HIV pathogenesis has not been evaluated 247.

Specific incorporation into HIV-1 virions of host-derived HLA-DR, β2-microglobulin and class I MHC molecules has long been known 248249. However, the physiological relevance of these associations is poorly understood. Several explanations have been suggested. Interestingly, immunization with HLA-DR and class I MHC proteins confer protection against SIV infection, probably by generating class-I and class-II-specific antibodies 250251. Additionally, incorporation of HLA-DR, one of the most abundant cellular proteins, into viral membranes, renders virions capable of stimulating purified T lymphocytes in the presence of the staphylococcal enterotoxin A superantigen 252. Evidence that these mechanisms may influence HIV-1 pathogenesis is still missing, and these studies should be interpreted with caution. It has also been suggested that virion incorporation of MHC molecules could enhance virus infectivity 239.

Other proteins specifically incorporated into lentivirus capsids are hsp70, Ini1/SNF5, and the lysyl-tRNA synthetase. Different roles in the viral life cycle have been ascribed to these proteins. Hsp70 enhances the infectivity of HIV-1 253, whereas Ini1/SNF5 appears to modulate virus yield and infectivity 254. Lysyl-tRNA synthetase may participate in the recruitment of tRNALys to prime viral DNA synthesis 255256257. The role of these proteins in HIV-1 pathogenesis in vivo has not been addressed.

There is mounting evidence that the levels of envelope in HIV-1 particles are finely modulated and could play an important role in viral pathogenesis. Recent studies have changed the common view of the HIV-1 virion. Initially, it was believed that, like in other enveloped viruses, HIV-1 viral membranes were abundant in envelope proteins. It is now known that the relative density of envelope spikes in HIV-1 virions is surprisingly low. Compared to influenza, which contains 200–300 spikes per particle, it is estimated that HIV-1 virions incorporate only 4 to 7 spikes per particle 258. The reason for maintaining envelope levels to a minimum is unknown, though it may be related to the need to evade the antibody neutralizing response directed against the viral envelope.

In HIV-1, specific signal sequences in the cytoplasmic and membrane domain of gp41 modulate expression of envelope at the cell surface. Interactions between the cytoplasmic domain of gp41 and the viral matrix protein modulates envelope incorporation by a mechanism that requires the cellular protein TIP47 259260261262263264. The CD4 receptor, which plays an essential role allowing viral entry, inhibits envelope incorporation late in the viral cycle (reviewed in 265). Excessive amounts of CD4 in virus-producer cells interfere with envelope transport to the cell surface, preventing its incorporation into viral membranes 266267. Additionally, CD4, which is normally excluded from viral membranes, is then incorporated into HIV-1, interfering with the ability of envelope to bind CD4 in target cells. To overcome the inhibitory effects of CD4, HIV-1 utilizes three viral products, Nef, Vpu, and Envelope, to ensure CD4 is removed after its function in viral entry has been performed 265268269. The importance of CD4 down-modulation in HIV-1 replication has been highlighted in recent studies. Nef-mediated CD4 down-regulation correlates with the ability of HIV-1 to replicate in blood-derived primary lymphocytes and human lymphoid tissue 270271272. Other studies have evaluated the Nef-dependent CD4 down-modulation activity through different stages of disease. Nef alleles isolated from late symptomatic stages are more potent CD4 down-modulators than alleles isolated from the chronic asymptomatic stage 273274. This enhanced activity translates in faster HIV-1 replicative properties in vitro, and presumably in enhanced pathogenicity in vivo. Many other studies have revealed the existence of mutations or deletions in nef genes isolated from LTNP patients. In some instances, these changes have been associated with diminished CD4 down-modulation activity 275276277278.

No enzymatic activities have been associated to HIV-1 Nef, and it is widely accepted that this viral product exerts its CD4 down-modulation activity by binding to host proteins. Nef is known to interact with more than 25 cellular proteins. The physiological relevance of these interactions, however, has not always been evaluated 279. Despite the potential role of cellular partners involved in Nef activities, to date no studies evaluating their influence in HIV-1 pathogenesis have been performed.

Another key family of proteins participating in the innate immune response against pathogens is the Toll-like-receptors (TLR) family. During HIV infection, TLRs may play an important role in modulating the extent of viral replication (reviewed in 302303). There is mounting evidence that in addition to direct activation of TLRs by HIV, indirect activation mediated by opportunistic pathogens may influence the outcome of disease 304305. Signals mediated by pathogens can activate HIV transcription in trans, via the production of proinflammatory cytokines such as TNF-α, IL-1, or in cis, directly inducing expression of cellular factors activating the viral LTR (NF-κB and AP-1) 302.

Eleven TLR members have been identified so far in humans 306. TLRs recognize specific motifs expressed by pathogens such as lipid moieties, protein motifs, and nucleotide sequences. The first piece of evidence suggesting a role for TLRs in HIV-1 pathogenesis dates back to 1990, when it was shown that bacterial LPS activates the viral LTR, a process later found to be mediated by TLR4 307. To date, at least five TLR members (TLR2, TLR4, TLR7, TLR8, and TLR9) have been implicated in induction of HIV-1 expression during opportunistic co-infections 308309310311312313. Several studies support a role for co-infecting pathogens in HIV transmission and disease. TNF-α levels are elevated in HIV-infected patients co-infected with Mycobacterium tuberculosis 314. In another report, a clinical trial evaluating antisense molecules targeting HIV-1 Gag produced unexpected results, and revealed elevated viral loads in treated patients. Several clues led the authors to conclude that the unwanted effect had been the result of TLR9 activation by CpG motifs present in the antisense molecules 309315316. Polymorphisms in the TLR9 gene have been associated with HIV-1 disease progression. Two SNPs (1635A/G and +1174G/A) are found at higher frequency in HAART-naïve rapid progressors (estimated as CD4 count decline) than in normal progressors 317.

Further highlighting the influence of co-infecting pathogens, malaria episodes result in increased viral loads in HIV-infected individuals, a factor that may have influenced the spread of HIV in sub-Saharan Africa 318319320. Likewise, a study of Kenyan pregnant women has shown that susceptibility to malaria is enhanced up to 2-fold in HIV-infected patients 321. Interestingly, although malaria did not significantly affect the frequency of mother-to-child transmission according to this report, co-infection with helminth parasites was associated with increased risk of transmission by a mechanism likely involving activation of intrauterine lymphocytes 321. Importantly, not all co-infecting pathogens increase HIV replication. Infections with GB virus, measles virus, Orientia tsutsugamushi and HTLV type 1 and 2 viruses are known to attenuate HIV infection 322. Human herpesvirus 6 (HHV-6) and 7 (HHV-7) also suppress HIV replication. Interestingly, these viruses inhibit infection with R5 tropic strains, but minimally affect replication of X4 virus. HHV-6 interferes with HIV-1 by upregulating the CCR5 ligand RANTES 323. Unlike HHV-6, HHV-7 exerts its effect by downregulating CD4, the cellular receptor shared by HHV-7 and HIV-1 324. The R5-specific effect of the inhibition of HIV-1 replication by HHV-7 may be explained by lower affinities for CD4 in R5 envelopes, as compared to T-tropic envelopes.

The FC gamma receptor IIa (CD32) has also been shown to influence HIV pathogenesis. This receptor is expressed in B cells and monocytes, and plays an important role in the removal of antibody-coated infectious agents 325. One group has evaluated the effect of polymorphisms of the FC gamma receptor IIa gene in HIV-1 perinatal transmission in Kenyan women. The frequency of a His/His131 SNP is higher in HIV-positive than in HIV-negative children, whereas distribution of an Arg/Arg131 allele is reversed 326. Another study with a Kenyan cohort found similar conclusions for the same alleles in transmission of placental malaria 327. Unlike the His131 form, the Arg131 variant does not bind IgG2, however, further experimental evidence is needed to evaluate whether this property of the FC gamma receptor is responsible for the observed phenotype.

Cytotoxic T lymphocytes (CTL) play an essential role in the response to HIV infection. It is well accepted that the efficacy of the CTL responses vary among infected individuals and modulate disease progression in vivo 1314. In addition to HLA variants, other molecules modulate the efficacy of the CTL response. CTL-dependent killing of target cells is mediated by the content of storage granules released near the junction of the CTL and target cell. These granules contain the pore-forming protein perforin, and several serine proteases called granzymes 328. HIV-specific CTLs from LTNP achieving immunological control proliferate extensively in response to autologous HIV-infected CD4-positive T cells. These LTNP CTLs, unlike those in normal progressors, express higher levels of perforin upon stimulation and proliferation with autologous cells 329. This suggests that the quality of the CTL response plays an essential role in the response against HIV infection. Supporting this, polymorphisms in the perforin gene have been associated with slower disease progression in a French cohort. The presence of the SNP 1012T in the proximal perforin promoter correlates with fewer perforin-positive CTL cells, whereas homozygous individuals (1012T/T) display a trend towards slower disease progression 330.