HIV-1 isolates were initially classified on the basis of their replicative, cytopathic, and tropic properties. A majority of primary virus isolates, which replicated slowly and generated small amounts of progeny particles, were labeled “slow/low” to distinguish them from “rapid/high” HIV-1 strains, which exhibit faster infection kinetics and release high titers of virions. Some HIV-1 primary isolates, particularly those recovered from symptomatic individuals, induced syncytium formation (see Fig. 49.2B). These isolates were designated syncytium-inducing (SI) to distinguish them from non–syncytium-inducing (NSI) strains. Some SI isolates are able to infect CD4+ T-cell lines but not monocyte-derived macrophages (MDMs) and have been designated T-cell-line or TCL-tropic viruses. Cell tropism was a third property used to classify HIV-1 primary isolates. All HIV-1 strains are able to productively infect activated PBMCs;
some can also replicate in cultures of MDMs and have been classified as macrophage- or M-tropic. Other primary HIV-1 isolates are also able to infect both MDMs and T-cell lines; such strains have been classified as dual-tropic. Although exceptions exist, M-tropic isolates are usually NSI and “slow/low,” whereas TCL-tropic strains are frequently SI and exhibit the “rapid/high” replication phenotype. Forced passage of primary isolates in continuous CD4+ human T-cell lines has given rise to T-cell-line–adapted (TCLA) HIV-1 strains, which, in contrast to their progenitors, possess increased sensitivity to both neutralizing antibodies and soluble CD4.⁷³⁰

The discovery that α and β chemokine receptors function as co-receptors (in addition to CD4) for virus entry¹²⁷,¹⁸¹,²¹⁶ led to a revised classification system for HIV-1 strains (Table 49.1). Isolates are now categorized on the basis of chemokine receptor usage in fusion assays, in which CD4 and an individual co-receptor protein are expressed in reporter cells, or in entry assays, in which small-molecule co-receptor–targeted inhibitors, specific for CCR5 or CXCR4, block single-cycle or spreading viral infections in T-cell lines or human PBMCs. As indicated in Table 49.1, HIV-1 isolates using the CXCR4 receptor (now designated X4 viruses⁴⁷) are frequently TCL-tropic, SI strains that exhibit enhanced cytopathicity and more vigorous replicative properties. In contrast, HIV-1 strains exclusively using the CCR5 receptor (R5 viruses) may be M-tropic and NSI. Dual-tropic R5/X4 strains exhibiting a continuum of tropic phenotypes are frequently SI. Interestingly, most T-cell leukemia cell lines express high levels of CXCR4 but are CCR5 negative, explaining in part why primary HIV-1 isolates fail to replicate in these cells.

CCR5-using viruses are the predominant HIV-1 strains detected in recently infected individuals and remain the dominant virus throughout the asymptomatic phase of HIV-1 infection. In vivo, CCR5 is exclusively expressed on memory CD4+ T cells, a subset present at high levels in nonlymphoid tissues (e.g., in the gastrointestinal tract and lung).¹⁹¹ Many of these tissue-infiltrating memory CD4+ T cells are also activated. In contrast, CXCR4 is expressed on both naïve and memory CD4+ T-cell populations. In peripheral blood, where a majority of the circulating lymphocytes are naïve, 80% to 95% of the CD4+ T cells may express CXCR4, whereas only 5% to 10% produce detectable CCR5.²⁸⁰,⁵⁶⁰ Because R5 HIV-1 strains predominate in recently infected individuals, marked depletion of CCR5-expressing memory CD4+ T lymphocytes in the gastrointestinal tract has been reported, whereas more modest reductions of total CD4+ T-cell numbers in the blood occur during the acute infection.⁵¹⁶ The emergence of X4 variants during the symptomatic phase of infection, which occurs in approximately 50% of individuals infected with clade B HIV-1, is referred to as co-receptor switching. It is associated with accelerated CD4+ T-cell depletion and more rapid progression to AIDS. Such infected individuals have 30- to 70-fold higher frequencies of virus-infected naïve cells than persons who harbor only R5 virus, consistent with the co-receptor usage of each HIV-1 strain.⁶⁴ The accelerated and specific decline of the naïve CD4+ T cells after the emergence of X4 strains represents an additional insult to the immune system that severely impedes CD4+ T-lymphocyte renewal and heralds the beginning of a downhill clinical course.

Studies of acute HIV and SIV/SHIV infections have shown that nonactivated CD4+ T lymphocytes in the mucosal lamina propria and draining lymph nodes are the principal virus targets during this phase of the infection.⁴⁴⁷,⁵⁶²,⁸⁵¹ Paradoxically, the innate immune responses directed against HIV following virus acquisition induce a cascade of inflammatory cytokines that greatly increases the numbers and activation status of CD4+ T cells. In contrast to these findings, multiple studies have shown that resting PBMCs in culture are highly refractory to HIV-1 infection; no viral proteins or progeny virions are produced.⁷¹⁸,⁸⁴³ PBMCs, activated with mitogens or following incubation with anti-CD3 and anti-CD28 monoclonal antibodies (mAbs) and propagated in the presence of interleukin-2 (IL-2), are quite susceptible to HIV-1 and produce high titers of infectious virus. The block to HIV-1 replication in noncycling cultured PBMCs is not presently understood. Indeed, cell cycle studies have shown that progression into G₁b is required to achieve productive HIV-1 infections.⁴⁰⁵ In vivo, activated CD4+ T lymphocytes are the main source of progeny virions. Nonetheless, an inducible reservoir of resting “latently infected” memory CD4+ T cells, containing integrated HIV-1 DNA and expressing no viral proteins, is established in exposed individuals within days of transmission and is thought to be the driving factor preventing the eradication of this viral infection.¹³⁵,¹³⁷

During infections of human T-cell lines or CD4+ T lymphocytes, HIV-1 can spread through the cultures by two different mechanisms: (a) fluid phase diffusion of cell-free particles or (b) virological synapse (VS)-mediated cell-to-cell contact. Early studies reported that cell-to-cell transmission of HIV-1 was 10²– to 10³-fold more efficient than that of cell-free virus.¹⁸⁴ The extremely low particle-to-infectivity ratio (10⁻³ to 10⁻⁵) for HIV-1 undoubtedly contributed to the inefficient spread of cell-free virions. The formation of a VS with T cells, first reported for HTLV-1,³⁴⁴ utilizes some of the components of the immunologic synapse machinery involved in the interaction between lymphocytes and antigen-presenting cells.⁷⁰⁹ The formation of the HIV-1 VS is initiated by the binding of
gp120 in the infected cell to CD4 molecules on the surface of the target cell.³⁶⁵ This conjugate is further stabilized by interactions between leukocyte function–associated antigen (LFA) and intercellular adhesion molecules (ICAMs), which promote additional recruitment of actin, viral proteins (Gag and Env), CD4, and integrins to cell–cell contact regions. Cytoskeletal remodeling drives virus assembly in the infected cell to the plasma membrane and toward the target cell. Progeny particles released into the synaptic space enter the recipient cell by fusion with the plasma membrane or following passage through endosomal compartments.⁶⁶⁰ The co-polarization of clustered viral proteins in the donor cell with receptors in multiple adjacent recipients can also lead to the formation of polysynaptic structures.³⁶⁶ Filopodial bridges and other actin-containing tubular extensions of the plasma membrane have been reported for delivering virions to distant target cells.⁶⁹³ Inhibitors of HIV-1, including neutralizing antibodies, are able to block cell-free and cell-to-cell–mediated virus infectivity, although transfer across the VS may offer some protection from these agents.¹²¹

Although HIV-1 causes the loss of CD4+ cells in culture and depletion of this T-lymphocyte subset in virus-infected individuals, the mechanisms responsible for this cytopathic effect are still incompletely understood. HIV-1–induced killing of cultured CD4+ T cells (primary T cells and T-cell lines) occurs by necrosis or cytolysis rather than by classical apoptosis, which involves the activation of cysteinyl aspartate-requiring proteases (known as caspases).⁷² Some viral-encoded proteins including Vpr and Env are cytotoxic when expressed alone or in the context of a productive HIV infection. In vivo, the massive depletion of CD4+ T lymphocytes in blood and tissues during acute HIV and SIV infections is now recognized to be the result of direct virus-induced cell killing.⁵⁰⁶,⁵¹⁶ Nonetheless, apoptotic CD4+ T lymphocytes are present in the peripheral blood and lymph nodes of HIV-1 seropositive individuals during the chronic phase of infection. These cells are uninfected and are thought to represent the effects of bystander cell damage due to the unrelenting and generalized immune activation, a major factor contributing to HIV-1 pathogenesis.²⁸²

All known lentiviruses have retained the capacity to replicate in nonproliferating, terminally differentiated tissue macrophage. Unlike the oncoretroviruses, which require mitosis and dissolution of the nuclear membrane to complete a cycle of replication, HIV-1 is able to productively infect G₁ (but not G₀)-arrested macrophages. This property requires that the HIV-1 preintegration complex (PIC) be translocated through intact nuclear pores, a process in which the capsid (CA) protein modulates interactions with the nuclear transport machinery in nondividing cells⁴³² (see section on Nuclear Import). In HIV-1–positive individuals, infected macrophages contribute to viral persistence and dissemination to the central nervous system (CNS). HIV-1 is thought to enter the CNS very soon following transmission as virus-free particles or by the trafficking of infected lymphocytes (and possibly infected monocytes). HIV-1 does not productively infect neurons but instead targets perivascular macrophages in the choroid plexus and parenchymal microglial cells. Multinucleated giant cells, suggestive of virus-induced fusion of microglia and/or macrophages, are a histopathologic characteristic of HIV-1–associated neuropathy. Neurologic dysfunction affecting the CNS and the peripheral nervous system is observed in 40% to 70% of virus-positive individuals. HIV-1 replication in monocyte/macrophage lineage microglia within the CNS may persist in patients receiving HAART because of inefficient transfer of drugs across the blood–brain barrier. This viral reservoir, as well as HIV-1 present in latently infected memory CD4+ T cells, could contribute to the rebound of plasma viremia invariably observed after cessation of antiretroviral drug administration.

Virtually all information pertaining to HIV-1 replication in macrophages derives from studies of an in vitro surrogate: monocyte-derived macrophage (MDM). The vigorous replication by M-tropic viral strains in MDM requires that blood-derived monocytes undergo differentiation in vitro.⁶⁸² MDMs and tissue macrophages express lower surface levels of CD4, CCR5, and CXCR4 than CD4+ T lymphocytes.⁴⁴¹,⁷⁸⁸ Nonetheless, only M-tropic R5 HIV-1 strains are able to initiate infections of MDM and only a subset of M-tropic viruses are also neurotropic.³²⁶ Most primary HIV-1 isolates from blood are weakly M-tropic; strong macrophage tropism is mainly seen with variants from tissues such as the CNS, where macrophage dependence is paramount.⁵⁹⁸ Like productive infections of T lymphocytes, HIV-1 replication in MDMs requires integration of the reverse transcript.²⁰⁶ However, in contrast to numerous reports demonstrating that assembly and release of progeny virions occur at the plasma membrane of productively infected T cells, electron microscopy of HIV-1–infected MDM revealed accumulations of particles within intracellular vacuoles of endosomal origin.⁵⁷⁶,⁵⁸⁹ This interpretation has been questioned by reports showing that particles within putative internal vesicles were in a compartment, distinct from multivesicular bodies, which communicates with the exterior through tubular channels continuous with the highly invaginated macrophage plasma membrane.⁴⁵,¹⁸⁰,⁷⁹⁸ In fact, a recent study employing electron tomography and electron microscopy with stereology has reported that MDMs release progeny virions into a pre-existing membranous compartment, contiguous with the plasma membrane, which expands in size during HIV-1 infection.⁷⁹⁷ Virus particles that assemble in this internal compartment in MDMs are rapidly translocated to the VS following contact with susceptible T cells.²⁷⁷

DCs are antigen-presenting cells that capture, transport, and present antigens to CD4+ and CD8+ T lymphocytes. Two main subtypes of DCs have been identified: myeloid DCs (MDCs) and plasmacytoid DCs (PDCs). MDCs reside in multiple tissues including the skin (where they are called Langerhans cells) and the intestinal and genital tract mucosa. The less abundant PDCs are found in the blood, T-cell zones of lymph nodes, and thymus and can be recruited to sites of inflammation. After activation, MDCs secrete IL-12, whereas PDCs produce type I interferons and inflammatory cytokines. MDCs present near mucosal surfaces may represent the first line of defense against sexually transmitted HIV-1, transporting virus particles from this portal of entry to draining lymph nodes where they are degraded into antigenic peptides for presentation to CD4+ T lymphocytes. Paradoxically, this critical initial response exacerbates the HIV-1 acute infection by delivering virus to the very T-cell subset that is susceptible to and can massively amplify progeny virus production. Because of the low abundance of DCs in tissues and the small number of these cells recoverable from clinical specimens, most studies evaluating the interaction
of DCs with HIV-1 have used monocytes differentiated in the presence of granulocyte-macrophage colony-stimulating factor (GM-CSF) and IL-4. These derivatives are known as monocyte-derived DCs (MDDCs).

Productive HIV-1 infections of MDDCs are very inefficient compared to infections of activated CD4+ T cells, although vigorous virus replication is observed when DCs are first pulsed with virus and then co-cultivated with CD4+ T lymphocytes, a process called trans-infection.²⁵⁹ Immature DCs (IDCs) express relatively low levels of CD4, CCR5, and CXCR4 and larger amounts of C-type lectin receptors (CLRs) such as dendritic Cell-Specific Intercellular adhesion molecule-3-Grabbing Non-integrin (DC-SIGN) and dendritic cell immunoreceptor (DCIR), which capture HIV by binding to gp120-associated high-mannose glycans. Functionally, maturation of DCs leads to (a) increased capacity to degrade and present antigen to T cells by forming immunologic synapses, (b) augmented secretion of T-cell co-stimulatory molecules, and (c) potent suppression of endolysosomal degradation. Cell surface CLR expression, including that of DC-SIGN, is markedly reduced in mature DCs; capture of HIV-1 is now considered to be CLR, gp120, and CD4 independent in mature DCs.

Early studies reported that HIV-1 was internalized in trypsin-resistant, low-pH compartments for several days in MDDCs or cells expressing DC-SIGN.⁴¹⁵ Co-cultivation with activated CD4+ T cells resulted in rapid trans-infection.²⁵⁹ EM analyses of infected MDDCs indicated that intact HIV particles were present within cytoplasmic vesicles. Subsequent imaging studies, using green fluorescent protein (GFP)-labeled HIV, revealed that DCs transferred virus to T cells via a VS, utilizing adhesion molecules similar to those used in the formation of the immunologic synapse.⁴¹⁵ Virions in the infected DC were co-polarized with CD4 and chemokine co-receptors in the T lymphocyte at the cell interfaces. More recent experiments have shown that in DCs, HIV-1 resides in a unique surface-accessible intracellular compartment, distinct from endocytic vesicles, and can rapidly traffic to the VS following contact with T cells.⁸³⁷ This unusual DC compartment communicates with the extracellular environment and, similar to structures present in HIV-1–infected macrophages, is composed of highly folded invaginations of the plasma membrane.⁴⁵,⁷⁹⁸ DC-SIGN/gp120-mediated HIV-1 entry into IDCs by endocytosis very likely results in extensive degradation of viral proteins by an active endocytic system primed for antigen presentation. Consequently, IDCs would only transmit residual nondegraded HIV remaining on their surfaces to T lymphocytes. In contrast, mature DCs, unable to endocytose HIV in a normal fashion, sequester virions in specialized intracellular compartments and, during antigen presentation, rapidly transfer infectious particles to CD4+ T cells. Recent studies indicate that HIV-1 infection of DCs, and to a lesser extent mature DCs, is restricted by the interferon-induced host factor SAM domain and HD domain–containing protein 1 (SAMHD1).⁴⁶,³²⁸,⁴¹⁸ The ability of SAMHD1 to interfere with lentiviral infection is counteracted by Vpx (see section on Vpx).

In the search for an HIV-1 animal model during the early phase of the AIDS epidemic, cell suspensions from virus-infected individuals were inoculated into a variety of mammalian species including nonhuman primates, but only chimpanzees consistently became infected. It was subsequently shown that asymptomatic HIV-1 infections could readily be established in chimpanzees, but viremia was not maintained and no long-standing impairment to the immune system occurred.²⁴ This could reflect genetic changes, which occurred in the SIV_cpz progenitor of HIV-1 following its transmission from chimpanzees and adaptation to humans.⁶⁶³ The failure of chimpanzees to develop disease in a timely fashion following inoculation with tissue culture–adapted or patient-derived HIV-1 isolates, coupled with their endangered species status, has curtailed their use as an animal model of HIV-1–induced immunodeficiency.

When it became apparent that HIV-1 infections could only be established in chimpanzees and humans, potentially more tractable and novel rodent systems were developed to model specific steps of in vivo HIV-1 infections. In one approach, transgenic mice containing the entire HIV-1 genome or individual viral genes under the control of the HIV-1 long terminal repeat (LTR) were constructed. Because mice carrying such transgenes contain HIV-1 DNA in every cell, they model the postintegration phase of the virus life cycle and provide a means for assessing possible deleterious effects of viral proteins in vivo. In general, these first-generation transgenic animals expressed very low levels of HIV-1 proteins and occasionally developed disease.

To overcome the intrinsically low levels of HIV-1 LTR-directed expression in rodents, transgenic mice were subsequently generated carrying viral genes under the control of cellular/tissue-specific promoters. In particular, mice carrying a complete copy of the HIV-1 genome or just the nef gene, directed by a chimeric (human-mouse) CD4 promoter/enhancer element, spontaneously developed an AIDS-like disease characterized by marked depletion of CD4+ T lymphocytes, thymic atrophy, T-cell activation, and immunodeficiency.²⁹²,⁶¹⁸ Thus, targeting HIV-1 Nef synthesis to mouse CD4+ cells, in the absence of a spreading viral infection or the expression of other HIV-1 gene products, was sufficient to elicit a disease with some of the features of human AIDS.

In addition to their use for investigations of human hematopoiesis, innate and adaptive immunity, cancer biology, and regenerative medicine, mice engrafted with human cells have been utilized as a small-animal model for studies of HIV-1 pathogenesis and vaccine development. In very early studies, severe combined immunodeficiency (SCID) mice, engrafted with human PBMC or fragments of human fetal thymus and fetal liver engrafted under the renal capsule, were shown to support HIV-1 replication and moderate to profound depletions of CD4+ T lymphocytes but generated no immune responses to the virus.⁵⁴³,⁵⁵⁴ Relatively low levels of human cell engraftment were achieved in these first-generation humanized mice.

Although the subsequent development of nonobese diabetic (NOD) SCID animals resulted in improved human PBMC engraftment, the relatively short life span of the mice and residual murine natural killer (NK) cell activity limited their usefulness. Targeted mutations of the mouse interleukin-2 receptor-γ chain (γ_c^−/−) and reconstitutions with PBMC, hematopoietic stem cells (HSCs), or human cord blood greatly increased engraftment of human tissue.³⁴⁸ Inoculation of these mice with HIV-1 by parenteral and mucosal routes
resulted in sustained disseminated infections.²⁷,⁴⁹ The recently described surgical implantation of fetal thymic and liver tissues in humanized mice previously engrafted with HSCs resulted in animals able to support high levels of HIV viremia, experience CD4+ T-cell depletion, sustain generalized immune activation, and generate virus-specific antibody and CD4+/CD8+ T-cell responses.⁸¹ Although great progress has been made in the development of xenotransplant mouse models over the past 20 years, they are technically demanding to make and not currently amenable for widespread use.

When it became apparent that humans and chimpanzees were the only mammalian species that could be infected by HIV-1, attention turned to the SIV/Asian macaque model with the idea of constructing a virus containing both SIV and HIV gene segments. The earliest versions of these SIV/HIV chimeric viruses (SHIVs) consisted of the genetic backbone of SIV_mac239, into which the HIV-1 tat, rev, vpu, and env and, in some instances, portions of vpr and nef genes were inserted (e-Fig. 49.1). SHIVs were constructed because, unlike SIVs, they could be used as challenge viruses in vaccine experiments to monitor the effectiveness of immunogens eliciting antiviral neutralizing antibodies directed against the HIV-1 Env glycoprotein. The first pathogenic SHIVs developed all carried X4-tropic env genes and caused a rapid, systemic, and complete loss of CD4+ T lymphocytes within a few weeks of inoculation, high and sustained levels of viremia (>10⁷ RNA copies/mL of plasma), and death from immunodeficiency within 3 to 6 months of inoculation.³⁴⁵,³⁶¹,⁶²² This aggressive disease phenotype appeared to be ideally suited for vaccine studies and, for a few years (2000 to 2004), SHIVs rather than SIVs became the challenge virus of choice for such experiments. However, concern was raised when the results from many studies showed that although most vaccine regimens suppressed SHIV replication and disease induction, they failed to protect monkeys from acquisition of either SHIV or SIV infection. In addition, differences were demonstrated between the chemokine receptor utilization of SIV (CCR5) and most transmitted HIV-1 (CCR5) versus the early SHIVs (CXCR4) and the CD4+ T-cell subsets targeted by these viruses in vivo (memory [SIV and HIV-1] versus naïve [SHIV]).⁵⁶³ The development and use of R5-tropic SHIVs as challenge viruses in macaques were shown to be better surrogates for HIV-1 infections of humans.²⁹⁸

Derivatized HIVs (also known as minimal SHIVs), containing more than 90% of the HIV-1 genome and the SIV vif gene (to counteract macaque APOBEC3 proteins) plus a seven–amino acid segment of the SIV Gag analog of the HIV-1 cyclophilin A (CypA)-binding loop (to escape macaque TRIM5α [for tripartite motif-containing α] restriction), have recently been reported to establish transient infections in pig-tailed macaques.³⁴⁶ When it was subsequently appreciated that pig-tailed macaques lacked TRIM5α restriction activity against HIV-1, derivatized HIVs, containing only the SIV vif gene, were made and shown to transiently infect this monkey species.³⁰³

The HIV-1 replication cycle (Fig. 49.8) begins with the binding of virus particles to CD4 molecules on the surface of susceptible cells. In the context of tissue culture systems, susceptible cells include activated CD4+ T lymphocytes present in PBMC collected from uninfected individuals, T-cell leukemia cell lines, and MDMs. In vivo, activation of CD4+ T cells is not required for HIV-1 replication. Although binding of virions to CD4 is essential for HIV infectivity, their subsequent interaction with a co-receptor, now known to be members of the seven-membrane-spanning CC or CXC families of chemokine receptors, is required for membrane fusion and entry (reviewed in¹³). Multiple chemokine receptors exhibit activity in fusion/entry assays with HIV-1 gp120 (see later section on Env); however, the two most physiologically important co-receptors are CXCR4 and CCR5. Unlike some other enveloped viruses that enter cells by receptor-mediated endocytosis, HIV-1 and many other retroviruses fuse directly with the plasma membrane. Once inside the cell, HIV-1 encounters a cortical actin barrier, which can be remodeled by the virally encoded Nef protein. Partially uncoated virions migrate from cortical actin to the microtubule network where reverse transcription is initiated and movement to the nucleus begins (Fig. 49.8, steps 2 to 4) (reviewed in¹⁸). As with other retroelements, the partially double-stranded DNA reverse transcription product is transported through the cytoplasm and to the nucleus as a component of a PIC, which contains a subset of the Gag and Pol proteins. As noted earlier, the lentiviruses are unique among retroviruses in generating PICs that are actively transported by the nuclear import machinery into the interphase nucleus of nondividing cells arrested in the G₁ phase of the cell cycle (Fig. 49.8, step 5). After the import of the PIC into the nucleus, full-length linear copies of the reverse transcript are integrated into the chromosomal DNA of the infected cell (Fig. 49.8, step 6), a step required for efficient viral RNA synthesis and infectious particle production.

In activated T lymphocytes, integrated copies of HIV DNA serve as templates for RNA polymerase II (Pol II)-directed viral RNA synthesis (Fig. 49.8, step 7). The coordinated interaction of the HIV-encoded Tat protein and the cellular NF-κB and Sp1 transcriptional transactivating proteins with the RNA Pol II transcriptional apparatus ensures the production of high levels of viral RNA. Unspliced or partially spliced HIV transcripts are exported from the nucleus to the cytoplasm by a unique transport mechanism mediated by the virus-encoded Rev protein, which shuttles between these compartments (Fig. 49.8, step 8). The subsequent translation of the gp160 Env precursor occurs in the endoplasmic reticulum, whereas the Gag and Gag-Pol polyproteins are synthesized on free cytoplasmic ribosomes, and each is independently transported to the plasma membrane (Fig. 49.8, steps 9 and 10). The Gag and Gag-Pol polyproteins, in association with dimers of genomic RNA, condense at the plasma membrane to form an electron-dense “bud” (see Fig. 49.2) that gives rise to a spherical immature particle containing the mature TM and SU Env glycoproteins. Proteolytic processing of the Gag and Pol proteins by HIV PR during or immediately after particle release generates the cone-shaped core characteristic of mature HIV virions (Fig. 49.8, steps 11 and 12).

Thus, HIV-1 uses the same replicative strategy as the so-called simple retroviruses, which encode only the Gag, Pol, and Env proteins. However, during its evolution, HIV-1 has acquired six additional genes to carry out functions that are either (a) performed by cellular proteins already present in the cells infected by the simple retroviruses or (b) uniquely required for virus replication, transmission, and survival in hematopoietic
cells targeted by the primate lentiviruses. Some of the HIV-1 accessory proteins (Vif, Vpr, Vpu, and Nef) are not required for replication in certain human T-cell lines, although virus infectivity may be affected up to several thousandfold, depending on the accessory gene and the type of infected cell. In view of their conservation, however, these additional genes must be required for successful HIV-1 infectivity in vivo.

HIV-1 encodes only 15 proteins and like all other viruses must utilize a large number of cellular proteins for successful replication. Over the years, several techniques, such as immunoprecipitation/mass spectrometry and yeast two-hybrid screening, have been used to identify such host cell factors. The recent development of genome-wide small-interfering RNA (siRNA) or short hairpin (shRNA) screening has provided genetic approaches to identify potential human proteins required for HIV-1 replication, based on loss of function readouts.⁸²,⁴⁰⁴,⁸³⁶,⁸⁵⁷ In these experiments, siRNAs are transfected into target cells, whereas shRNAs are delivered using lentiviral vectors and subsequently processed into siRNAs intracellularly.

Of the 842 genes identified as being required for HIV-1 replication in three siRNA screens, 34 were also reported in at least two screens and three were detected in all three screens. The overlap between the three siRNAs and the single shRNA screen ranged from zero to three genes in pairwise comparisons. Eleven of the 34 genes found in two siRNA screens had also been previously reported to encode HIV-1–interacting proteins and 29 of the 34 genes are expressed in cells susceptible to HIV-1 (CD4 and chemokine coreceptor positive). The overlap between individual screens is unexpectedly small yet is statistically significant.⁹² This result is not surprising given that different challenge viruses, cell lines, siRNA libraries, periods of gene silencing prior to infection, and readouts of infection were used in the four studies. Nonetheless, all three siRNA screens identified factors associated with transport through the nuclear pores, trafficking between cellular compartments, DNA repair, RNA binding, and interaction with ubiquitin/proteasomal degradation machinery.⁹² Two of the siRNA screens identified previously reported cellular proteins known to be critical for HIV-1 infectivity including CD4, CXCR4, RelA, elongin B, and cyclin T1. Cyclophilin A, Sp1, and lens epithelium-derived growth factor (LEDGF) are among the known HIV-1 co-factors that were not indentified in any of the screens.

By definition, retroviruses possess the ability to convert their single-stranded RNA genomes into double-stranded DNA during the early stages of the infection process.³¹,⁷⁴⁷ The enzyme that catalyzes this reaction is RT, in conjunction with its associated ribonuclease H (RNaseH) activity. Retroviral RTs have two enzymatic centers: (a) a DNA polymerase that can copy either RNA templates (for “minus-strand” DNA synthesis) or DNA templates (for second- or “plus-strand” DNA synthesis) and (b) RNaseH (for the degradation of the tRNA primer and genomic RNA present in DNA-RNA hybrid intermediates). RT was the first viral protein targeted by antiretroviral therapy, and RT inhibitors remain central to the treatment of HIV-infected patients (for reviews see³³⁷,⁷⁰³).

Uncoating occurs after membrane fusion and entry of the viral nucleoprotein complex into the cytosol, giving rise to the reverse transcription complex (RTC). This multiprotein complex is composed not only of RNA and RT but also CA, NC, IN, and Vpr. While it was initially proposed that most if not all of the CA protein was lost from the RTC during uncoating, a variety of observations, including the fact that CA is the target for host restriction factors that disrupt reverse transcription (see later), support the hypothesis that at least some CA protein is retained in this complex. Recent findings suggest that uncoating and reverse transcription are linked; in the presence of an inhibitor of reverse transcription, uncoating is significantly delayed.³³⁸

The retroviral genome is packaged into the virion as two copies of single-stranded RNA. As discussed earlier, the two RNAs are held together in part by the DIS near their 5′ ends (see Fig. 49.7). Although each retroviral particle carries two strands of RNA, it appears that only one provirus is formed per virion.³³⁰ Retroviruses are therefore referred to as pseudodiploid. As with most other DNA polymerases, RT is dependent on the 3′-OH group of an RNA or DNA primer to initiate polymerization. Retroviruses use specific tRNAs (tRNA^Lys3 in the case of HIV-1) to initiate DNA synthesis. The mechanism of tRNA selection and placement on the template is complex, involving interactions with RT and NC as well as with the 18-nt pbs near the 5′ end of the viral genome (see Fig. 49.7).

Reverse transcription of the retroviral RNA genome to a double-stranded DNA copy takes place after viral entry into the target cell and proceeds via a series of steps that are outlined briefly here (see e-Fig. 49.2A). This is also described in Figure 47.7 of Chapter 47.

In addition to providing an essential function in the virus life cycle, the enzymatic activity of RT is routinely used in the laboratory to quantitatively monitor levels of progeny virions present in the supernatant of infected cultures and to elucidate the molecular details of virus replication. Also, the detection of viral DNA postinfection by real-time polymerase chain reaction (PCR) provides one of the most reliable methods to date for detecting entry and postentry steps in the virus replication cycle. Once inside the nucleus, some double-stranded viral DNA is ligated to generate circular forms; although these circular DNA products do not integrate, their formation is often used to monitor nuclear import (see later).

The mature HIV-1 RT holoenzyme is a heterodimer of 66- and 51-kD subunits. The 51-kD subunit (p51) is derived from the 66-kD (p66) subunit or some larger precursor by proteolytic removal of the C-terminal 15-kD fragment of p66 by PR (Fig. 49.6). The structure of HIV-1 RT has been determined by x-ray crystallography in a number of studies. The enzyme has been crystallized in several contexts: (a) unliganded,⁶³³ (b) bound to
a short DNA duplex and the Fab portion of an anti-RT antibody,³⁵⁴ (c) covalently linked to a complex of primer/template and deoxyribonucleotide triphosphate (dNTP),³³² and (d) bound to an RNA-DNA duplex.⁶⁵⁵ A number of structures are also available for RT bound to nonnucleoside RT inhibitors (NNRTIs). The crystal structure of HIV-1 RT reveals that the p66 and p51 subunits are folded into similar subdomains, but these subdomains are arranged quite differently in p66 and p51. The p66 subunit can be visualized as a right hand, with the polymerase active site within the palm, and a deep template-binding cleft formed by the palm, fingers, and thumb subdomains³⁵⁵ (Fig. 49.13 and 49.14). The polymerase domain is linked to RNaseH by the connection subdomain. The active site, located in the palm, contains three critical Asp residues (Asp-110, Asp-185, and Asp-186) in close proximity, and two coordinated Mg²⁺ ions. Mutation of these Asp residues abolishes the polymerase activity of RT. The orientation of the three active-site Asp residues is similar to that observed in other DNA polymerases (e.g., the Klenow fragment of Escherichia coli DNA Pol I).²² The p51 subunit appears to play a structural role and does not form a polymerizing cleft; Asp-110, Asp-185, and Asp-186 of p51 are buried within the subunit. Approximately 18 bp of the primer/template duplex lie in the nucleic acid–binding cleft, stretching from the polymerase to the RNaseH active sites. In the RT–primer/template–dNTP structure,³³² the presence of a dideoxynucleotide (ddNTP) at the 3′ end of the primer allows visualization of the catalytic complex blocked just prior to attack on the incoming dNTP. Comparison with previously obtained structures suggests a model whereby the fingers close in to trap the template and dNTP prior to nucleophilic attack of the 3′-OH of the primer on the incoming dNTP. After the addition of the incoming dNTP to the growing chain, the fingers adopt a more open configuration, thereby releasing the pyrophosphate and enabling RT to bind the next dNTP. HIV-1 RNaseH, whose structure has been determined by x-ray crystallography, displays a global folding similar to that of E. coli RNaseH.¹⁶⁹ The structure of human RNaseH1 bound to an RNA-DNA hybrid has allowed the modeling of HIV-1 RT RNaseH complexed with its substrate.⁵⁶⁶

The high rate of variation among HIV-1 populations poses one of the fundamental challenges to effectively controlling this pathogen. Highly variable virus populations are largely a consequence of rapid virus replication rates, coupled with the error-prone nature of RT (which lacks a proofreading function) and frequent template switching during reverse transcription¹⁴¹ (for review see⁷³³). Cell-free error rates for purified RTs have been determined,³⁸,⁶¹⁶,⁶³¹ as have in vivo retrovirus mutation rates.²,¹⁹²,⁴⁹¹,⁵⁸⁵,⁵⁸⁶ Errors during retrovirus replication are also likely to be introduced by the cellular DNA-dependent RNA Pol II during transcription of the viral RNA. The cell-free error
rate measured for HIV-1 RT is more than 10-fold higher than the mutation rate, suggesting that the in vitro reaction lacks factor(s) that increase fidelity.⁴⁹¹ The total HIV-1 in vivo mutation rate (a composite of substitutions, frameshifts, simple deletions, and deletions with insertions) was measured at ∼1 to 3 × 10⁻⁵ per cycle of replication.²,⁴⁹¹

As discussed earlier, retroviral particles contain two copies of single-stranded RNA. During reverse transcription, RT frequently switches from one template to the other; this occurs on the order of ∼30 times per infection event.³⁴ RT also undergoes intramolecular jumps.⁵⁸⁴ Intramolecular jumps lead to mutations (e.g., deletions, insertions, and duplications), whereas intermolecular jumps generate recombinants if the two packaged RNAs are not genetically identical.³²⁹ The intermolecular template switches that lead to recombination can occur when RT encounters a break in the RNA during polymerization—the so-called forced copy choice model.¹⁴² Jumps during minus-strand synthesis can occur in the absence of strand breaks,³⁷¹ and the stability of RT association with template is influenced by the balance between the enzyme’s RNaseH activity and polymerase activity—the “dynamic copy-choice” model.³⁴² The impact of high rates of recombination and high replication rates on HIV-1 populations is substantial. As noted earlier, a significant percentage (∼20%³¹²) of sequenced virus isolates are intersubtype recombinants, and recombinants constitute the predominant strains currently circulating in certain parts of the world. Recombination also provides a mechanism for the rapid generation of multiple-drug-resistant HIV-1 variants.

RT has long been a target in the search for antiviral compounds. RT inhibitors are now routinely administered to AIDS patients, usually in combination with PR and, more recently, IN inhibitors. There are two major classes of drugs that block reverse transcription: nucleoside RT inhibitors (NRTIs) and NNRTIs. Examples of NRTIs include AZT (zidovudine); 2′-deoxy-3′-thiacytidine (3TC or lamivudine); 2′,3′ dideoxyinosine (ddI or didanosine); and 2′,3′ dideoxycytidine (ddC or zalcitabine). NNRTIs include nevirapine, efavirenz, and delavirdine. NRTIs are nucleotide mimics that lack the 3′-OH and thus act as chain terminators upon incorporation into DNA, whereas the NNRTIs inhibit DNA polymerization by binding a small hydrophobic pocket near the RT active site and inducing a change in the structure of RT that blocks DNA synthesis.⁴⁰¹ Resistance to RT inhibitors often develops in patients as viral variants arise encoding mutant forms of RT that are no longer blocked by the inhibitors (for review see⁶⁵⁶). Resistance to NRTIs generally takes place by one of two mechanisms: (a) resistant RTs acquire the ability to selectively incorporate natural dNTPs but exclude the NRTI, or (b) mutant RTs incorporate the NRTI but subsequently excise it from the terminated primer. Resistance to NNRTIs occurs when mutations in RT interfere with the binding of the drug to the enzyme by disrupting key drug–enzyme interactions, changing the shape of the NNRTI binding pocket, or preventing entry of the drug into the binding pocket.⁶⁵⁷ Efforts continue to develop new NRTIs that are more difficult for drug-resistant RTs to exclude, and it may ultimately be possible to develop compounds that interfere with NRTI excision. NNRTIs with increased conformational flexibility (“wiggling and jiggling”) may also be more effective and more difficult for HIV-1 to evade if the flexible inhibitor can still bind to the altered pocket in the resistant RTs.¹⁶²,¹⁶³ Finally, a significant effort has been directed toward developing inhibitors that target the RNaseH activity of the enzyme.

As mentioned previously, recent studies have demonstrated that the host protein SAMHD1 restricts an early step in lentiviral infection in myeloid cells and is counteracted by the accessory protein Vpx. This topic will be discussed in more detail later in the section on Vpx.

Most retroviral genomes gain access to host cell chromosomal DNA during progression of the target cell through the cell cycle.⁴⁴⁴,⁶³⁴ These retroviruses are therefore unable to efficiently infect noncycling cells. In contrast, lentiviruses, including HIV-1, are able to productively infect nondividing cells (e.g., macrophages). During the events leading to its integration in the nucleus, the newly synthesized viral DNA remains associated with the PIC. This complex, which is composed of both viral and cellular proteins,⁷⁹,⁸⁷ must presumably possess the ability to cross the nuclear membrane during interphase to access the host cell genome. Because the HIV-1 PIC—estimated to be ∼56 nm in diameter⁵²³—is much larger than the passive diffusion limit for nuclear pores (∼9 nm),⁵⁰⁵ it has long been assumed that the PIC must harbor determinants that promote active nuclear import. Initially, it was proposed that MA contains the determinants of PIC nuclear import⁹⁰,⁷⁷⁷; however, this model was not validated.²²⁶,²³⁰,²³² Subsequent findings implicated Vpr in nuclear import (see section on Vpr). Numerous studies have observed that IN localizes to the nucleus, giving rise to the hypothesis that IN regulates PIC transfer across the nuclear membrane (e.g.,⁷⁸). This model has also been controversial, because mutations that disrupt the nuclear import of IN can cause defects in multiple aspects of IN function.³⁴⁷,⁴⁶⁷ It has also been proposed that a cis-acting, centrally located, triple-stranded DNA structure described earlier and known as the central DNA flap, which is a product of lentiviral reverse transcription (see section on Reverse Transcription), promotes nuclear import of the PIC.⁸⁴⁶ Although inclusion of the central flap improves the infectivity of lentiviral vectors, there is some disagreement as to whether this sequence is required in the context of replication-competent, full-length HIV-1 molecular clones.¹⁹,¹⁹⁷,⁴⁵⁴ An HIV-1 chimera-containing MLV IN and lacking all other proposed determinants of nuclear import (e.g., MA, Vpr, and the central DNA flap) retains the ability to infect nondividing cells,⁸²⁵ indicating that none of these elements is critical for HIV-1 nuclear import.

Although the viral determinants of nuclear import, and the mechanism by which they function, remain to be clarified, there is compelling evidence for a connection between CA and nuclear import. It was initially believed that HIV-1 CA is lost from the reverse transcription complex during uncoating; it now appears, however, that a significant amount of CA remains associated with this complex.²¹,¹⁸⁷,⁵¹² Analysis of chimeras between MLV (which is unable to infect nondividing cells) and HIV-1 (which can infect nondividing cells) revealed that the ability to infect nondividing cells maps to CA.⁸²³ Furthermore, mutations in CA can impose on HIV-1 infection a requirement for cell cycle progression⁸²⁶ and can inhibit nuclear import.¹⁸⁷

As mentioned earlier, several genome-wide RNA interference (RNAi) screens have identified a large number of host factors that are putatively required for HIV-1 infection.⁸²,⁴⁰⁴,⁸⁵⁷ Among these are nuclear pore proteins and karyopherins—including transportin 3 (TNPO3; also known as TRN-SR2), NUP153, and NUP358/RanBP2—that could play a role in PIC transfer across the nuclear pore. TNPO3 was shown to bind HIV-1 IN and promote PIC nuclear import.¹³¹ Subsequent work⁴¹⁰ confirmed a role for TNPO3 in HIV-1 infection and demonstrated that most lentiviruses showed a requirement for TNPO3 (feline immunodeficiency virus [FIV] being the exception) but that other retroviruses (e.g., RSV and MLV) did not. This study, however, failed to observe a role for IN–TNPO3 binding in the ability of TNPO3 to promote infection. Rather, the use of the HIV-1/MLV chimeras mentioned earlier
pointed to a role for CA in determining TNPO3 dependency.⁴¹⁰ NUP358/RanBP2 bears a cyclophilin domain that has been reported to bind directly to CA; this interaction promotes PIC nuclear import and also influences integration site preference.⁶⁷¹

In a screen designed to identify dominant factors that interfere with HIV-1 infection, a C-terminally truncated form of cleavage and polyadenylation factor 6 (CPSF6) was shown to block an early step in the HIV-1 infection process—specifically nuclear import—but had no effect on MLV infectivity.⁴³² Selection experiments gave rise to a resistant mutant that, intriguingly, contained a single amino acid change in CA (N74D). This N74D CA mutant was not only resistant to inhibition by the truncated CPSF6 but also displayed an altered requirement for karyopherins and nuclear pore factors. For example, wild-type (WT) HIV-1 infection was inhibited by depletion of TNPO3, RanBP2, and NUP153, but infectivity of the N74D mutant was relatively unaffected.⁴³²,⁵⁰⁴ Conversely, N74D infection was more sensitive than WT to depletion of nuclear pore proteins NUP85 and NUP155. These data again support a role for CA in PIC nuclear import and demonstrate that CA regulates HIV’s utilization of nuclear transport and pore components. CPSF6 is typically localized to the nucleus and overexpression of the full-length, nuclear form has little effect on HIV-1 infection.⁴³² The truncated form that inhibits HIV-1 infection lacks the CPSF6 nuclear localization signal (NLS) and is thus localized to the cytosol. Thus, the shift in CPSF6 localization from nucleus to cytosol may play a key role in its inhibitory activity. Consistent with this model, CPSF6 becomes cytosolic in TNPO3-depleted cells.⁴³³ Kewalramani⁴³³ showed that in TNPO3-depleted cells CPSF6 becomes cytosolic. Furthermore, CPSF6 depletion reverses the inhibitory effect of TNPO3 depletion. These data suggest that the effect of TNPO3 on HIV-1 is mediated through CPSF6. Because CPSF6 interacts with HIV-1 CA, it is conceivable that this protein plays a regulatory role in HIV-1 infection, potentially as a PIC uncoating or transport factor.

A distinguishing feature of retrovirus replication is the insertion of a DNA copy of the viral genome into the host cell chromosome after reverse transcription. The integrated viral DNA (the provirus) serves as the template for the synthesis of viral RNAs and is maintained as part of the host cell genome for the lifetime of the infected cell. Retroviral mutants deficient in the ability to integrate generally fail to establish a productive infection.

The integration of viral DNA is catalyzed by IN, a 32-kD protein generated by PR-mediated cleavage of the C-terminal portion of the HIV Gag-Pol polyprotein (see Fig. 49.6). The steps in the integration process were originally elucidated in studies using MLV,⁸⁷,²⁴² but these findings apply to HIV as well. In all retroviral systems, integration proceeds in the same series of steps (Fig. 49.17):

Our understanding of the chemistry of the retroviral integration reaction was greatly assisted by the development of in vitro integration assays. Purified IN, expressed in E. coli, can carry out 3′ processing and strand transfer reactions when combined with short synthetic oligonucleotides that mimic the viral DNA ends and a divalent metal ion (Mg²⁺ or Mn²⁺).¹⁵¹,³⁷⁹ Using such systems, IN also catalyzes a reaction known as disintegration, which in essence is the strand transfer reaction in reverse.¹²⁸ While disintegration may not reflect a physiologically relevant activity of IN, it nevertheless has been useful in understanding the biochemistry of integration. Initial studies observed that the predominant product in in vitro assays using purified HIV-1 IN was a single end joined to one strand of the target, rather than the more physiologically relevant product in which both ends are integrated into the target (concerted or full-site integration).

Retroviral IN proteins are composed of three structurally and functionally distinct domains: an N-terminal zinc-binding domain (NTD), a catalytic core domain (CCD), and a C-terminal domain (CTD) (Fig. 49.18). Problems with low solubility and propensity to aggregate hindered progress in defining the structure of the IN holoenzyme in its active (tetrameric) state. However, the structures of all three domains independently and in various combinations were solved by x-ray crystallography or NMR methods. The NMR structure of the NTD⁹⁶ revealed three helical bundles with a zinc coordinated by invariant amino acids His-12, His-16, Cys-40, and Cys-43. The structure of the NTD is reminiscent of helical DNA-binding proteins that contain a helix-turn-helix motif. Initially, poor solubility hampered efforts to solve the structure of the CCD. However, attempts at crystallography were successful when a mutation was identified that greatly increased solubility without disrupting in vitro catalytic activity.¹⁹⁸ Each monomer of the core domain is composed of a five-stranded β-sheet flanked by helices; this structure bears striking resemblance to that of other polynucleotidyl transferases including RNaseH and the bacteriophage MuA transposase.¹⁹⁸,⁶²⁶ Three highly conserved residues are found in analogous positions at the catalytic center of these enzymes; in HIV-1 IN, these residues are Asp-64, Asp-116, and Glu-152—often referred to as the D,D-35-E motif (Fig. 49.18). Mutations in this catalytic triad block HIV-1 IN function both in vivo and in vitro. The IN CTD, whose structure was initially solved by NMR,²⁰⁰,⁴⁶¹ adopts a five-stranded β-barrel folding topology reminiscent of Src homology 3 (SH3) domains. The crystal structure of an HIV-1 NTD-CCD IN fragment⁷⁸⁹ revealed a dimer of dimers, with two inner monomers and two outer monomers, with an extensive intermolecular NTD-CTD interface that provided a preview of the interface evident in the structure of the full retroviral tetramer.

A breakthrough in IN structural biology took place in 2010 with the realization that the IN of prototype foamy virus (PFV) is highly soluble and capable of concerted integration in vitro with short DNA substrates.⁷⁶³ Although divergent at the overall amino acid sequence level, HIV-1 and PFV IN are structurally conserved. The IN tetramer bound to DNA oligonucleotides that mimic the viral DNA ends—which together form the active nucleoprotein complex referred to as the “intasome”—was crystallized and the structure solved.²⁹⁴ This structure revealed that two monomers in the tetramer interact at a dimer interface and are responsible for all contacts with the DNA substrate; the other two monomers are on the outside of the complex and do not interact with each other or with DNA (Fig. 49.19). The residues of the catalytic triad (D,D-35-E) of the inner monomers are oriented close to the 3′-OH of the viral DNA ends. Two metal ions are coordinated at the active site by the residues of the catalytic triad. The addition of model target DNA to the intasome allowed for crystallization of a complex containing both viral and target DNAs.⁴⁷⁷ In the presence of Mg²⁺ the strand transfer reaction took place during crystallization, whereas in the absence of divalent metal ions strand transfer was blocked. This allowed capture of postcatalytic strand transfer and prestrand transfer complexes. These structures revealed that the target DNA, which binds in a cleft between the two halves of the intasome complex, adopts a highly bent conformation, allowing the well-separated intasome active sites to access the scissile phosphodiester bonds in the target DNA (Fig. 49.19). The viral (donor) DNA duplex is unpaired at the active site. Conformational changes that occur during catalysis are predicted to render the strand transfer reaction irreversible.⁴⁷⁷ Following the original publication of the structures described earlier,²⁹⁴,⁴⁷⁷ a number of additional PFV structures that contained potent inhibitors were solved (discussed later; for review see⁴⁴⁹); together these reports allowed the development of a structural model for HIV-1 integration that was validated by extensive mutational analysis.⁴⁰⁹

Although the basic mechanism by which retroviruses integrate their DNAs into the host cell genome is highly conserved, different retroviruses display clear differences in the selection of a cellular target site.⁵²⁵ For example, HIV-1 prefers to
integrate in actively transcribed genes. MLV favors transcription start sites but only weakly selects active genes. ASLV displays a strong preference for neither active genes nor transcription start sites; indeed, for ASLV, high-level transcription appears to be inhibitory for integration.⁵⁰⁷ Target site selectivity, which impacts upon the development of retroviruses as vehicles for gene therapy, is likely influenced by interactions between IN or other components of the PIC and chromatin-associated host factors. Several cellular proteins have been studied extensively for their contribution to integration and target site selection. The most important among these is a 75-kD, predominantly nuclear, chromatin-associated protein known as LEDGF/p75. A direct interaction between HIV-1 IN and LEDGF/p75 was reported,¹²⁴,²⁰⁴ and this protein was found to stimulate the strand transfer step of the integration reaction in vitro¹²⁴ and increase binding of IN to DNA.⁹³ However, early studies were discordant in terms of the effect of LEDGF/p75 depletion on HIV-1 replication.⁴⁵⁹,⁷⁶⁹ It soon became apparent that low levels of chromatin-bound LEDGF/p75 were sufficient to promote integration in cells. Near-complete knock-down of LEDGF/p75 levels⁴⁵⁸ or use of a murine LEDGF/p75 knockout cell line⁶⁹⁷ revealed markedly reduced (but not fully abolished) HIV-1 infectivity in the absence of detectable LEDGF/p75 expression. Infectivity was also severely compromised by overexpression of the IN-binding domain (IBD) of LEDGF/p75.¹⁷⁴,⁴⁵⁸ In the case of both LEDGF/p75 depletion and LEDGF/p75-IBD overexpression, infectivity was disrupted at the integration step, and the effect was specific for lentiviruses, establishing LEDGF/p75 as a cellular co-factor for lentiviral integration. Studies in which siRNA-resistant LEDGF/p75 mutants were added back to LEDGF/p75-depleted cells identified domains of LEDGF/p75 required for its activity as an integration co-factor.⁴⁵⁸,⁶⁹⁷,⁷⁷⁰ The observation that the LEDGF/p75-IBD and the chromatin-binding regions (the PWWP domain and AT-hook) (Fig. 49.20) were required for function supported the hypothesis that LEDGF/p75 functions to tether the PIC to target chromosomal DNA. The crystal structure of the IN-CCD and LEDGF/p75-IBD complex provided further details of the IN–LEDGF/p75 interaction.¹²³ In addition to a clear role for LEDGF/p75 in directing HIV-1 integration to transcription units, TNPO3 and the nuclear pore protein RanBP2/Nup358 have also been implicated in the targeting of HIV-1 integration. Ocwieja and colleagues⁵⁶⁷ reported that depleting these genes shifted HIV-1 integration out of gene-dense regions. In contrast, TNPO3 depletion had no effect on the targeting of MLV integration. Use of the aforementioned HIV/MLV chimeras once again mapped the requirement for TNPO3 to Gag.⁵⁶⁷

One of the notable findings to emerge from the LEDGF/p75 depletion studies is that in the absence of LEDGF/p75 expression, HIV’s preference for integrating into transcriptional units is lost.¹³⁸,⁶⁹⁷ These results are consistent with the hypothesis that the ability of LEDGF/p75 to tether lentiviral PICs to chromatin is a major determinant for target-site selection. This hypothesis in turn suggests that the targeting of lentiviral integration could be shifted out of transcription units by substituting the chromatin-binding sequences of LEDGF/p75 with chromatin-binding domains from other proteins. Indeed, the infectivity defect conferred by LEDGF/p75 depletion can be rescued with a chimeric protein bearing the LEDGF/p75 IBD
and foreign chromatin-binding domains,⁵¹⁵ and such chimeric PIC-tethering factors are able to shift HIV-1 integration target site specificity away from actively transcribed genes.²¹⁷,²⁶¹,⁷⁰⁰ The ability to engineer such custom-designed PIC–chromatin tethering factors and control integration site specificity may have applications in lentiviral-based gene therapy. The function of LEDGF/p75 in promoting HIV-1 integration raises the possibility that the IN–LEDGF/p75 interaction could serve as a target for developing novel antiviral therapeutics. By using the IN-CCD–LEDGF/p75-IBD co-crystal structure in an in silico screen, Debyser and colleagues¹³² identified small-molecule “LEDGINs” (LEDGF/p75-IN inhibitors) that block the protein interaction in vitro and HIV-1 infection at the integration step.

Early studies showed that several classes of agents inhibit IN function in vitro; these included polyanions, nucleotide analogs, and DNA-binding compounds.²⁰⁹ Most of these agents are fairly nonspecific in their mode of action, and many compounds that block activity of purified IN are not inhibitory in assays using purified PICs.²¹⁰ However, di-keto acid inhibitors were found to be specific and potent inhibitors of the IN-mediated strand transfer reaction both in vitro and in cell-based assays.³⁰⁷ These IN strand transfer inhibitors (INSTIs) bind IN only in the presence of divalent metal and viral DNA and prevent the complex from associating with cellular (target) DNA. Although structurally diverse, INSTIs appear to have common features, including (a) a hydrophobic moiety composed of halobenzyl groups and (b) co-planar oxygen atoms presumed to chelate the metal ions in the active site.⁵⁰⁹ Work in the early 2000s showed that naphthyridine carboxamides inhibit the replication of an SHIV in rhesus macaques, setting the stage for in vivo use of INSTIs.³⁰⁸ Further structure-activity relationship studies led to the development of raltegravir, also known as MK-0518 or Isentress (Merck Research Laboratories),⁷³¹ and elvitegravir (Gilead Sciences and Japan Tobacco).⁶⁵⁹ Raltegravir was approved for clinical use in HIV-infected patients by the U.S. Food and Drug Administration (FDA) in 2007 and elvitegravir was approved in 2012 (reviewed in⁵⁰⁹).