Bringing the global HIV epidemic under control will require more effective approaches to prevent the spread of the retrovirus, as well as broader use of existing and future antiretroviral drugs. These interventions must be applicable in the developing world, where HIV has the most severe impact. Understanding the dynamic interplay of HIV with its cellular host provides the biological basis for controlling the epidemic. This chapter reviews current understanding of the HIV life cycle, with particular attention to the interactions between viral proteins and cellular machinery, and highlights promising future points of attack.
|Binding and Entry|
The genetic material of HIV, an RNA molecule 9 kilobases in length, contains 9 different genes encoding 15 proteins. Considerable insights have been gained into the function of these different gene products.(Figure 1) To productively infect a target cell, HIV must introduce its genetic material into the cytoplasm of this cell. The process of viral entry involves fusion of the viral envelope with the host cell membrane and requires the specific interaction of the envelope with specific cell surface receptors. The two viral envelope proteins, gp120 and gp41, are conformationally associated to form a trimeric functional unit consisting of three molecules of gp120 exposed on the virion surface and associated with three molecules of gp41 inserted into the viral lipid membrane. Trimeric gp120 on the surface of the virion binds CD4 on the surface of the target cell, inducing a conformational change in the envelope proteins that in turn allows binding of the virion to a specific subset of chemokine receptors on the cell surface.(1)(Figure 2) These receptors normally play a role in chemoattraction, in which hematopoietic cells move along chemokine gradients to specific sites. Although these receptors, which contain seven membrane-spanning domains, normally transduce signals through G proteins,(2) signaling is not required for HIV infection.
Twelve chemokine receptors can function as HIV coreceptors in cultured cells, but only two are known to play a role in vivo.(2) One of these, CCR5, binds macrophage-tropic, non-syncytium-inducing (R5) viruses, which are associated with mucosal and intravenous transmission of HIV infection. The other, CXCR4, binds T-cell-tropic, syncytium-inducing (X4) viruses, which are frequently found during the later stages of disease.(3) In up to 13% of individuals of northern European descent, a naturally occurring deletion of 32 base pairs in the CCR5 gene results in a mutant CCR5 receptor that never reaches the cell surface.(4,5) Individuals homozygous for this mutation (1-2% of the Caucasian population) are almost completely resistant to HIV infection.(4,5) These observations emphasize the pivotal role of CCR5 in the spread of HIV and suggest that small molecules that prevent HIV interaction with CCR5 might form a promising new class of antiretroviral drugs.
Both CD4 and chemokine coreceptors for HIV are found disproportionately in lipid rafts in the cell membrane.(6) These cholesterol- and sphingolipid-enriched microdomains likely provide a better environment for membrane fusion, perhaps by mirroring the optimal lipid bilayer of the virus.(7) Removing cholesterol from virions, producer cells, or target cells greatly decreases the infectivity of HIV.(8) Studies currently under way are exploring whether cholesterol-depleting compounds might be efficacious as topically applied microbicides to inhibit HIV transmission at mucosal surfaces. The development of effective microbicides represents an important component of future HIV prevention strategies.
The binding of surface gp120, CD4, and the chemokine coreceptors produces an additional radical conformational change in gp41.(9) Assembled as a trimer on the virion membrane, this coiled-coil protein springs open, projecting three peptide fusion domains that "harpoon" the lipid bilayer of the target cell. The fusion domains then form hairpin-like structures that draw the virion and cell membranes together to promote fusion, leading to the release of the viral core into the cell interior.(9) The fusion inhibitors T-20 and T-1249 act to prevent fusion by blocking the formation of these hairpin structures.
HIV virions can also enter cells by endocytosis. Usually, productive infection does not result, presumably reflecting inactivation of these virions within endosomes. However, a special form of endocytosis has been demonstrated in submucosal dendritic cells. These cells, which normally process and present antigens to immune cells, express a specialized attachment structure termed DC-SIGN.(10) This C-type lectin binds HIV gp120 with high affinity but does not trigger the conformational changes required for fusion. Instead, virions bound to DC-SIGN are internalized into an acidic compartment and subsequently displayed on the cell surface after the dendritic cell has matured and migrated to regional lymph nodes, where it engages T cells.(11) Thus, dendritic cells expressing DC-SIGN appear to act as "Trojan horses" facilitating the spread of HIV from mucosal surfaces to T cells in lymphatic organs.
Once inside the cell, the virion undergoes uncoating, likely while still associated with the plasma membrane.(Figure 2) This poorly understood process may involve phosphorylation of viral matrix proteins by a mitogen-activated protein (MAP) kinase(12) and additional actions of cyclophilin A(13) and the viral proteins Nef(14) and Vif.(15) Nef associates with a universal proton pump, V-ATPase,(16) which could promote uncoating by inducing local changes in pH in a manner similar to that of the M2 protein of influenza.(17) After the virion is uncoated, the viral reverse transcription complex is released from the plasma membrane.(18) This complex includes the diploid viral RNA genome, lysine transfer RNA (tRNALys) which acts as a primer for reverse transcription, viral reverse transcriptase, integrase, matrix and nucleocapsid proteins, viral protein R (Vpr), and various host proteins. The reverse transcription complex docks with actin microfilaments.(19) This interaction, mediated by the phosphorylated matrix, is required for efficient viral DNA synthesis. By overcoming destabilizing effects of a recently identified protein termed CEM15/APOBEC3G, Vif stabilizes the reverse transcription complex in most human cells.(15-20)
Reverse transcription yields the HIV preintegration complex (PIC), composed of double-stranded viral cDNA, integrase, matrix, Vpr, reverse transcriptase, and the high mobility group DNA-binding cellular protein HMGI(Y).(21) The PIC may move toward the nucleus by using microtubules as a conduit.(22) Adenovirus and herpes simplex virus 1 also dock with microtubules and use the microtubule-associated dynein molecular motor for cytoplasmic transport. This finding suggests that many viruses use these cytoskeletal structures for directional movement. How the switch from actin microfilaments to microtubules is orchestrated remains unknown.
Recent studies have revealed a mechanism by which the target cell defends against the HIV intruder.(23,24) Within 30 minutes of infection, select host proteins including the integrase interactor 1 (also known as INI-1, SNF5, or BAF47), a component of the SWI/SNF chromatin remodeling complex, and PML, a protein present in promyelocytic oncogenic domains, translocate from the nucleus into the cytoplasm.(24)(Figure 2) Addition of arsenic trioxide sharply blocks PML movement and enhances the susceptibility of cells to HIV infection raising the possibility that the normal function of PML is to oppose viral infection.(24) The binding of integrase to integrase interactor 1 may be a viral adaptation that recruits additional chromatin remodeling factors. Whether these complexes influence the site of viral integration or improve subsequent proviral gene expression is not known.
|Crossing the Nuclear Pore|
Unlike most animal retroviruses, HIV can infect nondividing cells, such as terminally differentiated macrophages.(25) This requires an ability to cross the intact nuclear membrane. With a Stokes radius of approximately 28 nm or roughly the size of a ribosome, the PIC is roughly twice as large as the maximal diameter of the central aqueous channel in the nuclear pore.(26) The 3 µm contour length of viral DNA must undergo significant compaction, and the import process must involve considerable molecular gymnastics.
One of the most contentious areas of HIV research involves the identification of key viral proteins that mediate the nuclear import of the PIC. Integrase,(27) matrix,(28) and Vpr(29) have been implicated.(Figure 2) Because plus-strand synthesis is discontinuous in reverse transcription, a triple helical DNA domain or "DNA flap" results that may bind a host protein containing a nuclear targeting signal.(30) Matrix contains a canonical nuclear localization signal that is recognized by the importins alpha and beta, which are components of the classical nuclear import pathway. However, a recent publication calls into question the contributions both of the nuclear import signal in integrase and of the DNA flap to the nuclear uptake of the PIC.(31) The HIV Vpr gene product contains at least three noncanonical nuclear targeting signals.(32) Vpr may bypass the importin system altogether, perhaps mediating the direct docking of the PIC with one or more components of the nuclear pore complex. The multiple nuclear targeting signals within the PIC may function in a cooperative manner or play larger roles individually in different target cells. For example, while Vpr is not needed for infection of nondividing, resting T cells,(33) it enhances viral infection in nondividing macrophages.(34) The finding that both matrix(35) and Vpr(32) shuttle between the nucleus and cytoplasm explains their availability for incorporation into new virions.
Once inside the nucleus, the viral PIC can establish a functional provirus.(Figure 2) Integration of double-stranded viral DNA into the host chromosome is mediated by integrase, which binds the ends of the viral DNA.(21) The host proteins HMGI(Y) and barrier to autointegration (BAF) are required for efficient integration, although their precise functions remain unknown.(36) Integrase removes terminal nucleotides from the viral DNA, producing a two-base recess and thereby correcting the ragged ends generated by the terminal transferase activity of reverse transcriptase.(21) Integrase also catalyzes the subsequent joining reaction that establishes the HIV provirus within the chromosome.
Not all PICs that enter the nucleus result in a functional provirus. The ends of the viral DNA may be joined to form a 2-LTR circle containing long terminal repeat sequences from both ends of the viral genome, or the viral genome may undergo homologous recombination yielding a single-LTR circle. Finally, the viral DNA may auto-integrate into itself, producing a rearranged circular structure. Although some circular forms may direct the synthesis of the transcriptional transactivator Tat or the accessory protein Nef, none produces infectious virus.(37) In a normal cellular response to DNA fragments, the nonhomologous end-joining (NHEJ) system may form 2-LTR circles to protect the cell.(38) This system is responsible for rapid repair of double-strand breaks, thereby preventing an apoptotic response. A single double-strand break within the cell can induce G1 cell-cycle arrest. The ability of the free ends of the viral DNA to mimic such double-strand chromosomal breaks may contribute to the direct cytopathic effects observed with HIV.
Integration can lead to latent or transcriptionally active forms of infection.(39) HIV's transcriptional latency explains the inability of potent antiviral therapies to eradicate the virus from the body. Moreover, despite a vigorous immune response early in infection, these silent proviruses are a reservoir that allows reemergence of HIV when the body's defenses grow weaker. Understanding latency and developing approaches to target latent virus are essential goals if eradication of HIV infection is ever to be achieved.
The chromosomal environment likely shapes the transcriptional activity of the provirus.(40) For example, proviral integration into repressed heterochromatin might result in latency.(Figure 3) Other causes of latency may include cell type differences in the availability of activators that bind to the transcriptional enhancer in the HIV LTR or the lack of Tat. However, of the multiple copies of provirus that are usually integrated in a given infected cell, at least one is likely to be transcriptionally active. This fact may explain why the number of latently infected cells (105-106) in infected patients is small.
In the host genome, the 5´ LTR functions like other eukaryotic transcriptional units. It contains downstream and upstream promoter elements, which include the initiator (Inr), TATA-box (T), and three Sp1 sites.(41) These regions help position the RNA polymerase II (RNAPII) at the site of initiation of transcription and to assemble the preinitiation complex. Slightly upstream of the promoter is the transcriptional enhancer, which in HIV-1 binds nuclear factor [kappa]B (NF-[kappa]B), nuclear factor of activated T cells (NFAT), and Ets family members.(42) NF-[kappa]B and NFAT relocalize to the nucleus after cellular activation. NF-[kappa]B is liberated from its cytoplasmic inhibitor, I[kappa]B, by stimulus-coupled phosphorylation, ubiquitination, and proteosomal degradation of the inhibitor.(43) NFAT is dephosphorylated by calcineurin (a reaction inhibited by cyclosporin A) and, after its nuclear import, assembles with AP1 to form the fully active transcriptional complex.(44) NF-[kappa]B, which is composed of p50 and p65 (RelA) subunits, increases the rates of initiation and elongation of viral transcription.(45) Since NF-[kappa]B is activated after several antigen-specific and cytokine-mediated events, it may play a key role in rousing transcriptionally silent proviruses
When these factors engage the LTR, transcription begins, but in the absence of Tat described below the polymerase fails to elongate efficiently along the viral genome.(Figure 3) In the process, short nonpolyadenylated transcripts are synthesized, which are stable and persist in cells due to the formation of an RNA stem loop called the transactivation response (TAR) element.(46)
Tat significantly increases the rate of viral gene expression. With cyclin T1 (CycT1), Tat binds to the TAR RNA stem-loop structure and recruits the cellular cyclin-dependent kinase 9 (Cdk9) to the HIV LTR.(47)(Figure 3) Within the positive transcription elongation factor b (P-TEFb) complex, Cdk9 phosphorylates the C-terminal domain of RNAPII, marking the transition from initiation to elongation of eukaryotic transcription.(48) Other targets of P-TEFb include negative transcription elongation factors (N-TEF), such as the DRB-sensitivity inducing (DSIF) and negative elongation (NELF) factors.(48) The high efficiency with which the HIV LTR attracts these negative transcription factors in vivo may explain why the LTR is a poor promoter in the absence of Tat. The arginine-rich motif (ARM) within Tat binds the 5´ bulge region in TAR. A shorter ARM in cyclin T1, which is also called the Tat-TAR recognition motif (TRM), binds the central loop of TAR.(47)
Binding of the Tat cyclin T1 complex to both the bulge and loop regions of TAR strengthens the affinity of this interaction. All of these components are required for Tat transactivation. In the presence of the complex between Tat and P-TEFb, the RNAPII elongates efficiently. Because murine CycT1 contains a cysteine at position 261, the complex between Tat and murine P-TEFb binds TAR weakly.(49) Thus, Tat transactivation is severely compromised in murine cells. Cdk9 also must undergo autophosphorylation of several serine and threonine residues near its C-terminus to allow productive interactions between Tat, P-TEFb, and TAR.(50) Additionally, basal levels of P-TEFb may be low in resting cells or only weakly active due to the interaction between P-TEFb and 7SK RNA.(51) All of these events may contribute to postintegration latency.
Transcription of the viral genome results in more than a dozen different HIV-specific transcripts.(52) Some are processed cotranscriptionally and, in the absence of inhibitory RNA sequences (IRS), transported rapidly into the cytoplasm.(53) These multiply spliced transcripts encode Nef, Tat, and Rev. Other singly spliced or unspliced viral transcripts remain in the nucleus and are relatively stable. These viral transcripts encode the structural, enzymatic, and accessory proteins and represent viral genomic RNAs that are needed for the assembly of fully infectious virions.
Incomplete splicing likely results from suboptimal splice donor and acceptor sites in viral transcripts. In addition, the regulator of virion gene expression, Rev, may inhibit splicing by its interaction with alternate splicing factor/splicing factor 2 (ASF/SF2)(54) and its associated p32 protein.(55)
Transport of the incompletely spliced viral transcripts to the cytoplasm depends on an adequate supply of Rev.(53) Rev is a small shuttling protein that binds a complex RNA stem-loop termed the Rev response element (RRE), which is located in the env gene. Rev binds first with high affinity to a small region of the RRE termed the stem-loop IIB.(56)(Figure 4) This binding leads to the multimerization of Rev on the remainder of the RRE. In addition to a nuclear localization signal, Rev contains a leucine-rich nuclear export sequence (NES).(53) Of note, the study of Rev was the catalyst for the discovery of such NES in many cellular proteins and led to identification of the complex formed between CRM1/exportin-1 and this sequence.(53)
The nuclear export of this assembly (viral RNA transcript, Rev, and CRM1/exportin 1) depends critically on yet another host factor, RanGTP. Ran is a small guanine nucleotide-binding protein that switches between GTP- and GDP-bound states. RanGDP is found predominantly in the cytoplasm because the GTPase activating protein specific for Ran (RanGAP) is expressed in this cellular compartment. Conversely, the Ran nucleotide exchange factor, RCC1, which charges Ran with GTP, is expressed predominantly in the nucleus. The inverse nucleocytoplasmic gradients of RanGTP and RanGDP produced by the subcellular localization of these enzymes likely plays a major role in determining the directional transport of proteins into and out of the nucleus. Outbound cargo is only effectively loaded onto CRM1/exportin-1 in the presence of RanGTP. However, when the complex reaches the cytoplasm, GTP is hydrolyzed to GDP, resulting in release of the bound cargo. The opposite relationship regulates the nuclear import by importins alpha and beta, where nuclear RanGTP stimulates cargo release.(53)
For HIV infection to spread, a balance between splicing and transport of viral mRNA species must be achieved. If splicing is too efficient, then only the multiply spliced transcripts appear in the cytoplasm. Although required, the regulatory proteins encoded by multiply spliced transcripts are insufficient to support full viral replication. However, if splicing is impaired, adequate synthesis of Tat, Rev, and Nef will not occur. In many non-primate cells, HIV transcripts may be overly spliced, effectively preventing viral replication in these hosts.(57)
In contrast to Tat and Rev, which act directly on viral RNA structures, Nef modifies the environment of the infected cell to optimize viral replication.(2)(Figure 4) The absence of Nef in infected monkeys and humans is associated with much slower clinical progression to AIDS.(58,59) This virulence caused by Nef appears to be associated with its ability to affect signaling cascades, including the activation of T-cell antigen receptor,(60) and to decrease the expression of CD4 on the cell surface.(61,62) Nef also promotes the production and release of virions that are more infectious.(63,64) Effects of Nef on the PI3-K signaling cascade--which involves the guanine nucleotide exchange factor Vav, the small GTPases Cdc42 and Rac1, and p21-activated kinase PAK--cause marked changes in the intracellular actin network, promoting lipid raft movement and the formation of larger raft structures that have been implicated in T-cell receptor signaling.(65) Indeed, Nef and viral structural proteins colocalize in lipid rafts.(64,66) Two other HIV proteins assist Nef in downregulating expression of CD4.(67) The envelope protein gp120 binds CD4 in the endoplasmic reticulum, slowing its export to the plasma membrane,(68) and Vpu binds the cytoplasmic tail of CD4, promoting recruitment of TrCP and Skp1p.(Figure 5) These events target CD4 for ubiquitination and proteasomal degradation before it reaches the cell surface.(69)
Nef acts by several mechanisms to impair immunological responses to HIV. In T cells, Nef activates the expression of FasL, which induces apoptosis in bystander cells that express Fas,(70) thereby killing cytotoxic T cells that might otherwise eliminate HIV-1 infected cells. Nef also reduces the expression of MHC I determinants on the surface of the infected cell(71)(Figure 4) and so decreases the recognition and killing of infected cells by CD8 cytotoxic T cells. However, Nef does not decrease the expression of HLA-C,(72) which prevents recognition and killing of these infected cells by natural killer cells.
Nef also inhibits apoptosis. It binds and inhibits the intermediate apoptosis signal regulating kinase-1 (ASK-1)(73) that functions in the Fas and TNFR death signaling pathways and stimulates the phosphorylation of Bad leading to its sequestration by 14-3-3 proteins.(74)(Figure 4) Nef also binds the tumor suppressor protein p53, inhibiting another potiential initator of apoptosis.(75) Via these different mechanisms, Nef prolongs the life of the infected host cell, thereby optimizing viral replication.
Other viral proteins also participate in the modification of the environment in infected cells. Rev-dependent expression of Vpr induces the arrest of proliferating infected cells at the G2/M phase of the cell cycle.(76) Since the viral LTR is more active during G2, this arrest likely enhances viral gene expression.(77) These cell-cycle arresting properties involve localized defects in the structure of the nuclear lamina that lead to dynamic, DNA-filled herniations that project from the nuclear envelope into the cytoplasm.(78)(Figure 4) Intermittently, these herniations rupture, causing the mixing of soluble nuclear and cytoplasmic proteins. Either alterations in the lamina structure or the inappropriate mixing of cell cycle regulators that are normally sequestered in specific cellular compartments could explain the G2 arresting properties of Vpr.
New viral particles are assembled at the plasma membrane.(Figure 5) Each virion consists of roughly 1500 molecules of Gag and 100 Gag-Pol polyproteins,(79) two copies of the viral RNA genome, and Vpr.(80) Several proteins participate in the assembly process, including Gag polyproteins and Gag-Pol, as well as Nef and Env. A human ATP-binding protein, HP68 (previously identified as an RNase L inhibitor), likely acts as a molecular chaperone, facilitating conformational changes in Gag needed for the assembly of viral capsids.(81) In primary CD4 T lymphocytes, Vif plays a key but poorly understood role in the assembly of infectious virions. In the absence of Vif, normal levels of virus are produced, but these virions are noninfectious, displaying arrest at the level of reverse transcription in the subsequent target cell. Heterokaryon analyses of cells formed by the fusion of nonpermissive (requiring Vif for viral growth) and permissive (supporting growth of Vif-deficient viruses) cells have revealed that Vif overcomes the effects of a natural inhibitor of HIV replication.(20,82) Recently this factor, initially termed CEM15/APOBEC3G, was identified(83) and shown to share homology with APOBEC1, an enzyme involved in RNA editing. Whether the intrinsic antiviral activity of CEM15 involves such an RNA editing function remains unknown. CEM15 is expressed in non-permissive but not in permissive cells and when introduced alone is sufficient to render permissive cells nonpermissive.
The Gag polyproteins are subject to myristylation,(84) and thus associate preferentially with cholesterol- and glycolipid-enriched membrane microdomains.(85) Virion budding occurs through these specialized regions in the lipid bilayer, yielding virions with cholesterol-rich membranes. This lipid composition likely favors release, stability, and fusion of virions with the subsequent target cell.(7)
The budding reaction involves the action of several proteins, including the "late domain"(86) sequence (PTAP) present in the p6 portion of Gag.(87)(Figure 5) The p6 protein also appears to be modified by ubiquitination. The product of the tumor suppressor gene 101 (TSG101) binds the PTAP motif of p6 Gag and also recognizes ubiquitin through its ubiquitin enzyme 2 (UEV) domain.(88,89) The TSG101 protein normally associates with other cellular proteins in the vacuolar protein sorting pathway to form the ESCRT-1 complex that selects cargo for incorporation into the multivesicular body (MVB).(90) The MVB is produced when surface patches on late endosomes bud away from the cytoplasm and fuse with lysosomes, releasing their contents for degradation within this organelle. In the case of HIV, TSG101 appears to be "hijacked" to participate in the budding of virions into the extracellular space away from the cytoplasm.
|Summary and Conclusions|
As the AIDS pandemic continues, advances in antiretroviral therapies have slowed its advance in the industrialized world, but have had little effect in developing countries. Because of its high rate of mutation, HIV is able to refine and optimize its interactions with various host proteins and pathways, thereby promoting its growth and spread. The virus ensures that the host cell survives until the viral replicative cycle is completed. Possibly even more damaging, HIV establishes stable latent forms that support the chronic nature of infection. Eradication of the virus appears unlikely until effective methods are developed to purge these latent viral reservoirs.
Basic science will clearly play a leading role in future attempts to solve the mysteries of viral latency and replication. A small-animal model that recapitulates the pathogenic mechanisms of HIV is sorely needed to study the mechanisms underlying viral cytopathogenesis. Virally induced cell death is not limited to infected targets but also involves uninfected bystander cells.(91) Murine cells support neither efficient virion assembly nor release of virions from the cell surface.(92) Currently, this defect represents a major impediment to the successful development of a rodent model of AIDS.
Proposed mechanisms for HIV killing of T cells include the formation of giant cell syncytia through the interactions of gp120 with CD4 and chemokine receptors,(93) the accumulation of unintegrated linear forms of viral DNA, the proapoptotic effects of the Tat,(94) Nef,(95) and Vpr(96) proteins, and the adverse effects conferred by the metabolic burden that HIV replication places on the infected cell.(97) Of note, expression of Nef alone as a transgene in mice recapitulates many of the clinical features of AIDS, including immunodeficiency and loss of CD4-positive cells.(98) All of these mechanisms suggest potential points of therapeutic intervention. Finally, future therapies will likely target viral proteins other than the reverse transcriptase, protease, and integrase enzymes. Clinical trials are already underway to study small molecules or short peptides that block the binding of HIV to cell-surface chemokine receptors or interfere with the machinery of viral-host cell fusion. Although not as advanced in development, small molecules have been found that block Tat transactivation(99) and Rev-dependent export of viral transcripts from the nucleus to the cytoplasm.(100) As a proof of principle, dominant-negative mutants of Tat, Rev, and Gag proteins have been shown to block viral replication. By increasing the number of antiviral compounds available to target different steps in the viral replicative cycle, in particular drugs that can be deployed in developing countries, research at the cellular level can serve to extend survival and to improve the quality of life for infected individuals, and to inhibit the spread of AIDS.
Warner C. Greene thanks Gary Howard and Stephen Ordway for editorial support, Robin Givens for administrative support, and the National Institutes of Health (R01 AI45234-02, R01 CA86814-02, P01 HD40543), the UCSF California AIDS Research Center (C99-SF-002), the James B. Pendleton Charitable Trust, and the J. David Gladstone Institutes for funding support.
B. Matija Peterlin thanks the National Institutes of Health (R01-AI38532, R01-AI46967, RO1-AI49104, and R01-AI51165-01) and the Universitywide AIDS Research Program (R00-SF-006) for funding support.