Although invasive cytomegalovirus (CMV) disease is uncommon in the era of antiretroviral therapy (ART), asymptomatic CMV coinfection is nearly ubiquitous in HIV infected individuals. While microbial translocation and gut epithelial barrier dysfunction may promote persistent immune activation in treated HIV infection, potentially contributing to morbidity and mortality, it has been unclear whether CMV replication in individuals with no symptoms of CMV disease might play a role in this process. We hypothesized that persistent CMV replication in the intestinal epithelium of HIV/CMV-coinfected individuals impairs gut epithelial barrier function. Using a combination of state-of-the-art in situ hybridization technology (RNAscope) and immunohistochemistry, we detected CMV DNA and proteins and evidence of intestinal damage in rectosigmoid samples from CMV-positive individuals with both untreated and ART-suppressed HIV infection. Two different model systems, primary human intestinal cells differentiated in vitro to form polarized monolayers and a humanized mouse model of human gut, together demonstrated that intestinal epithelial cells are fully permissive to CMV replication. Independent of HIV, CMV disrupted tight junctions of polarized intestinal cells, significantly reducing transepithelial electrical resistance, a measure of monolayer integrity, and enhancing transepithelial permeability. The effect of CMV infection on the intestinal epithelium is mediated, at least in part, by the CMV-induced proinflammatory cytokine IL-6. Furthermore, letermovir, a novel anti-CMV drug, dampened the effects of CMV on the epithelium. Together, our data strongly suggest that CMV can disrupt epithelial junctions, leading to bacterial translocation and chronic inflammation in the gut and that CMV could serve as a target for therapeutic intervention to prevent or treat gut epithelial barrier dysfunction during HIV infection. Intestinal epithelial barrier dysfunction is a well-known consequence of HIV infection that persists in spite of ART. The underlying mechanisms by which HIV perturbs intestinal epithelial junctions remain unclear, and the impact of opportunistic viral pathogens in the gut has not been fully appreciated. HIV-infected individuals are almost universally coinfected with CMV. While ART has resulted in a dramatic decline in the occurrence of end-organ CMV diseases, CMV remains an independent contributor to systemic inflammation in HIV-infected people. In our analysis of rectosigmoid biopsies from CMV/HIV-coinfected individuals, we found active CMV replication associated with intestinal damage in the gut of ART-suppressed HIV-infected individuals with no symptoms of CMV disease. We demonstrated that CMV productively infects intestinal epithelial cells and, independent of HIV, disrupts their tight junctions and compromises epithelial barrier function. Furthermore, the CMV-induced proinflammatory cytokine IL-6 is a key factor in this process, and attenuation of CMV replication by letermovir, a new anti-CMV agent currently in clinical development, was sufficient to prevent CMV-induced loss of epithelial integrity. Our data highlight the role of CMV as a cofactor in intestinal epithelial barrier dysfunction in asymptomatic HIV infection and suggest a novel treatment strategy to prevent intestinal epithelial barrier dysfunction and inflammation in HIV infection.

Like normal cellular nucleosides, the nucleoside reverse transcriptase (RT) inhibitor (NRTI) 4′-ethynyl-2-fluoro-2′-deoxyadenosine (EFdA) has a 3′-hydroxyl moiety, and yet EFdA is a highly potent inhibitor of human immunodeficiency virus (HIV) and simian immunodeficiency virus (SIV) replication with activity against a broad range of clinically important drug-resistant HIV isolates. We evaluated the anti-HIV activity of EFdA in primary human cells and in HIV-infected humanized mice. EFdA exhibited excellent potency against HIV _JR-CSF in phytohemagglutinin-stimulated peripheral blood mononuclear cells (PBMCs), with a 50% inhibitory concentration of 0.25 nM and a selectivity index of 184,000; similar antiviral potency was found against 12 different HIV clinical isolates from multiple clades (A, B, C, D, and CRF01_AE). EFdA was readily absorbed after oral dosing (5 mg/kg of body weight) in both mice and the rhesus macaque, with micromolar levels of the maximum concentration of drug in serum ( C _max ) attained at 30 min and 90 min, respectively. Trough levels were at or above 90% inhibitory concentration (IC ₉₀ ) levels in the macaque at 24 h, suggesting once-daily dosing. EFdA showed reasonable penetration of the blood-brain barrier in the rhesus macaque, with cerebrospinal fluid levels at approximately 25% of plasma levels 8 h after single oral dosing. Rhesus PBMCs isolated 24 h following a single oral dose of 5 mg/kg EFdA were refractory to SIV infection due to sufficiently high intracellular EFdA-triphosphate levels. The intracellular half-life of EFdA-triphosphate in PBMCs was determined to be >72 h following a single exposure to EFdA. Daily oral administration of EFdA at low dosage levels (1 to 10 mg/kg/day) was highly effective in protecting humanized mice from HIV infection, and 10 mg/kg/day oral EFdA completely suppressed HIV RNA to undetectable levels within 2 weeks of treatment.

Entry inhibitors of human immunodeficiency virus, type 1 (HIV-1) have been the focus of much recent research. C34, a potent fusion inhibitor derived from the HR2 region of gp41, was engineered into a 1:1 human serum albumin conjugate through stable covalent attachment of a maleimido-C34 analog onto cysteine 34 of albumin. This bioconjugate, PC-1505, was designed to require less frequent dosing and less peptide than T-20 and was assessed for its antifusogenic activity both in vitro and in vivo in the SCID-hu Thy/Liv mouse model. PC-1505 was essentially equipotent to the original C34 peptide and to T-20 in vitro. In HIV-1-infected SCID-hu Thy/Liv mice, T-20 lost activity with infrequent dosing, whereas the antiviral potency of PC-1505 was sustained, and PC-1505 was active against T-20-resistant (“DIV”) virus with a G36D substitution in gp41. The in vivo results are the direct result of a significantly improved pharmacokinetic profile for the C34 peptide following albumin conjugation. Contrary to previous reports that the gp41 NHR trimer is poorly accessible to C34 fused to protein cargoes of increasing size (Hamburger, A. E., Kim, S., Welch, B. D., and Kay, M. S. (2005) J. Biol. Chem. 280, 12567–12572), these results are the first demonstration of the capacity for a large, endogenous serum protein to gain unobstructed access to the transient gp41 intermediates that exist during the HIV fusion process, and it supports further development of albumin conjugation as a promising approach to inhibit HIV-1 entry.

HIV evades eradication because transcriptionally dormant proviral genomes persist in long-lived reservoirs of resting CD4⁺ T cells and myeloid cells, which are the source of viral rebound after cessation of antiretroviral therapy. Dormant HIV genomes readily produce infectious virus upon cellular activation because host transcription factors activated specifically by cell stress and heat shock mediate full-length HIV transcription. The molecular chaperone heat shock protein 90 (Hsp90) is overexpressed during heat shock and activates inducible cellular transcription factors. Here we show that heat shock accelerates HIV transcription through induction of Hsp90 activity, which activates essential HIV-specific cellular transcription factors (NF-κB, NFAT, and STAT5), and that inhibition of Hsp90 greatly reduces gene expression mediated by these factors. More importantly, we show that Hsp90 controls virus transcription in vivo by specific Hsp90 inhibitors in clinical development, tanespimycin (17-(allylamino)-17-demethoxygeldanamycin) and AUY922, which durably prevented viral rebound in HIV-infected humanized NOD scid IL-2Rγ^−/− bone marrow-liver-thymus mice up to 11 weeks after treatment cessation. Despite the absence of rebound viremia, we were able to recover infectious HIV from PBMC with heat shock. Replication-competent virus was detected in spleen cells from these nonviremic Hsp90 inhibitor-treated mice, indicating the presence of a tissue reservoir of persistent infection. Our novel findings provide in vivo evidence that inhibition of Hsp90 activity prevents HIV gene expression in replication-competent cellular reservoirs that would typically cause rebound in plasma viremia after antiretroviral therapy cessation. Alternating or supplementing Hsp90 inhibitors with current antiretroviral therapy regimens could conceivably suppress rebound viremia from persistent HIV reservoirs.

The SCID-hu Thy/Liv mouse model of HIV-1 infection is a useful platform for the preclinical evaluation of antiviral efficacy in vivo. We performed this study to validate the model with representatives of all four classes of licensed antiretrovirals. Endpoint analyses for quantification of Thy/Liv implant viral load included ELISA for cell-associated p24, branched DNA assay for HIV-1 RNA, and detection of infected thymocytes by intracellular staining for Gag-p24. Antiviral protection from HIV-1-mediated thymocyte depletion was assessed by multicolor flow cytometric analysis of thymocyte subpopulations based on surface expression of CD3, CD4, and CD8. These mice can be productively infected with molecular clones of HIV-1 (e.g., the X4 clone NL4-3) as well as with primary R5 and R5X4 isolates. To determine whether results in this model are concordant with those found in humans, we performed direct comparisons of two drugs in the same class, each of which has known potency and dosing levels in humans. Here we show that second-generation antiretrovirals were, as expected, more potent than their first-generation predecessors: emtricitabine was more potent than lamivudine, efavirenz was more potent than nevirapine, and atazanavir was more potent than indinavir. After interspecies pharmacodynamic scaling, the dose ranges found to inhibit viral replication in the SCID-hu Thy/Liv mouse were similar to those used in humans. Moreover, HIV-1 replication in these mice was genetically stable; treatment of the mice with lamivudine did not result in the M184V substitution in reverse transcriptase, and the multidrug-resistant NY index case HIV-1 retained its drug-resistance substitutions. Given the fidelity of such comparisons, we conclude that this highly reproducible mouse model is likely to predict clinical antiviral efficacy in humans.