Mapping the HIV Integration Landscape: Cancer Genes and the Persistence of the Latent Reservoir

April 24, 2015

Several presentations at last year's CROI offered evidence that the proliferation and persistence of latently infected CD4 T cells is associated with HIV integration into certain genes that have also been implicated in the development of cancers. The results prompted the investigators to hypothesize that HIV integration into cancer-associated genes plays a causative role in sustaining the reservoir of latently infected CD4 T cells by promoting their survival. Some of the scientists involved in the research went so far as to suggest that the findings might somehow contribute to explaining the elevated risk of cancer in HIV-positive people (see the CROI 2014 summary on the blog). The studies were eventually published in Science last July, along with a commentary by David Margolis and Frederic Bushman pointing out that "most HIV-related malignancies are not T cell cancers, and even most HIV-related lymphomas are of B cell origin" and gently noting: "alternative interpretations of the data are possible."

Recently, in the journal Cell, a comprehensive analysis of HIV integration sites was published and the data argues against the idea that HIV targets cancer-associated genes as a means to persist; rather, integration into these genes is shown to be consistent with HIV's well-documented preference for integrating into genes that are active in the cell it has infected. A separate but complementary paper in Nature provides a topological explanation for HIV's integration preferences.

To start with the study published in Cell (led by Lillian B. Cohn from the laboratory of Michel Nussenzweig at Rockefeller University), it identified a staggering 6,719 unique HIV integration sites in CD4 T cells from 13 individuals. Three participants provided samples before and after starting ART, two were already on ART, four were untreated and the remaining four were viremic controllers (viral loads less than 2000 copies in the absence of ART). In all cases, there was evidence of proliferation of CD4 T cells containing integrated HIV DNA. Termed "clonal expansions," these cells are identifiable because the HIV DNA they contain is integrated into the same exact location in the genome.

Consistent with previous studies, clonal expansions of CD4 T cells containing integrated HIV DNA were more common in participants in whom HIV replication was suppressed by ART compared to those with ongoing virus replication (perhaps because homeostatic proliferation of CD4 T cells contributes to immune reconstitution after ART initiation). There was also a bias for clonal expansions to involve HIV integrated into genes with lower expression levels compared to single, non-expanded HIV integrations, which more often occurred in highly expressed genes.

Based on this finding, the authors suggest that the latent HIV in clonally expanded CD4 T cells is less likely to be capable of expression and may not contribute substantially to the replication-competent HIV reservoir. In additional support for this argument, they note that all 75 of the clonally expanded HIV genomes they tested were defective, consistent with two previously published reports demonstrating homeostatic proliferation of CD4 T cells containing integrated HIV DNA; in both of those studies the viruses were also grossly defective (see Imamichi et al., 2014 and Josefsson et al., 2013). However they acknowledge: "we cannot rule out the possibility that a rare clone of cells contains an active virus," and this proved a wise caveat because at CROI 2015 several presentations described examples of clonally expanded HIV DNA that appeared replication-competent (e.g. Sarah Palmer, who showed that in one participant in a clinical trial of panobinostat, a clonally expanded HIV contributed to viral load rebound after an ART interruption).

The Cell paper addresses the issue of HIV integration into cancer-associated genes in several ways. An analysis of all HIV integration sites confirms that these genes are overrepresented, but the authors explain: "this preference does not seem to be significant because it is similar to the overall preference for integration into highly expressed genes." Importantly, there were no associations between HIV integrations into cancer-associated genes and persistence or clonal expansion, and a longitudinal study involving three participants who initiated ART demonstrated a significant decrease in integrations into these genes over time.

A logical interpretation is that HIV's preference for integrating into active genes causes the virus to frequently end up located in genes that are being expressed in the CD4 T cells it is infecting; these genes inevitably include those involved in the cell cycle and proliferation because HIV primarily replicates in activated CD4 T cells. And it is cell cycle and proliferation genes that commonly go awry in cancer, hence the association between HIV integration sites and cancer-associated genes.

The paper in Nature, by Bruna Marini and colleagues, offers a relatively prosaic explanation for HIV's integration site preferences, albeit one that required complex experiments to reveal. Essentially, it appears that the three-dimensional composition of a cell's nucleus dictates the gene regions that are most rapidly reachable by HIV and, upon entry into the nucleus, the virus plumps for the nearest options. These areas are on the periphery of the nucleus, and there is a strong bias against integration into genes located more centrally. In addition to the location, the genes also need to be accessible for integration, and because genes that are not being expressed are cloaked in chromatin, this means the genes have to be active to some extent. The authors' suggestion is: "the virus simply integrates into the first open chromatin regions it meets along its route into the nucleus."

While these recent papers offer some evidence contrary to the notion that HIV integrates into cancer-associated genes as a strategy to facilitate its persistence, it appears likely that the topic will continue to be the subject of debate, at least until more data become available. A senior author of one of the Science papers, Stephen Hughes from the National Cancer Institute (NCI), continued to argue the original hypothesis in a plenary presentation at CROI 2015. Hughes showed a slide with the following bullet point: "Data from several labs show that there are genes in which HIV integration can cause clonal expansion (MKL2 and BACH2)." But a causative relationship between HIV integration into these genes and clonal expansion of CD4 T cells is not established; the reported data only supports an association. Rather than reflecting cause, the association could be an effect of these genes being active in CD4 T cells differentiating into a long-lived memory phenotype (BACH2 is known to have a role in CD4 T cell differentiation).

Additionally, the last bullet point of Hughes's presentation states: "There are reports of HIV integrations into human cancers, including a report of a tumor with an integration in BACH2 (Mack et al.)." Hughes's Science paper includes a similar statement: "there are reports of a small number of lymphomas with HIV proviruses integrated at defined sites; one lymphoma had a provirus integrated in BACH2 (15, 29, 30)." The three numbered cites are all from the same laboratory (of Michael McGrath), and are dated 2003, 1994 and 1992 respectively. To the best of my knowledge, no other similar findings have been reported. The Mack et al citation is #15 (full text PDF is available free), and the problem is that it does not show that "one lymphoma had a provirus integrated in BACH2." The paper clearly states: "The integration assay used had a relatively low sensitivity, indicating that identified integration sites were from multiple clonal cell populations. Some of these HIV-infected clonal cell populations may be associated with induction of pathogenesis in these tissues." Nowhere does it confirm that the HIV integration was in the lymphoma.

Although it's understandable that the low sensitivity of the assay used in the study might lead someone to assume that the integration must have been in the cancer cells (because those cells would be present at high frequency), that is not the only potential explanation. An alternative possibility is that the HIV integration was in the genome of a clonally expanded CD4 T cell population infiltrating the cancerous tissue. Coincidentally, a study reporting exactly that phenomenon was presented at CROI 2015 by Francesco Simonetti from NCI. Simonetti demonstrated the clonal expansion of a latently infected CD4 T cell in tumor tissue -- perhaps due to the CD4 T cell specifically responding to a cancer antigen -- in an HIV-positive individual with squamous cell carcinoma.

At this point, it appears that the specter of an additional cause of cancer in HIV-positive people has been raised with no compelling supporting evidence. And the most recent of the three papers that have been cited as offering inferential evidence was miscited. When you consider how many people are living with HIV and the amount of HIV integration events occurring in those individuals, contrasted with the relative rarity of T cell cancers, it never seemed a very plausible idea to begin with -- as David Margolis and Frederic Bushman imply in their Science commentary. This issue is not only relevant to HIV-positive people, but also for gene therapy researchers using HIV-based vectors to modify T cell receptors as a means to target cancers for immune-mediated destruction (an approach that has had considerable success in clinical trials).

For cure research, a central issue is whether latently infected memory CD4 T cells behave any differently from other memory CD4 T cells. Certainly, long-term persistence and homeostatic proliferation are completely normal facets of memory CD4 T cell biology; this is why, for example, smallpox-specific memory CD4 T cells remain detectable for decades after vaccination despite the absence of any exposure to smallpox antigens. Also, if HIV integration into specific genes exerts some kind of control over memory CD4 T cell behavior, it does not lead to notable growth of the latent HIV reservoir -- longitudinal studies show that the reservoir decays slowly with a half-life of around 43-44 months. The latest longitudinal study was just published in the Journal of Infectious Diseases by David Margolis's research group and, as is noted in an accompanying commentary by Bob and Janet Siliciano, the estimated half-life of 43 months matches previous data almost exactly (the prior estimate was 44 months).

Because the techniques that allow comprehensive analyses of HIV integration sites are new, it's perhaps inevitable that there is some wrangling over how to interpret the data. The confusion is likely to resolve as more results become available, and there is clearly room for larger longitudinal studies to track clonally expanded latently infected CD4 T cells and their HIV integration sites over time, and better define what proportion of these expanded virus genomes are replication-competent.

Richard Jefferys is the coordinator of the Michael Palm HIV Basic Science, Vaccines & Prevention Project Weblog at the Treatment Action Group (TAG). The original blog post may be viewed here.

This article was provided by Treatment Action Group. It is a part of the publication Michael Palm HIV Basic Science, Vaccines & Cure Project.

No comments have been made.

Add Your Comment:
(Please note: Your name and comment will be public, and may even show up in
Internet search results. Be careful when providing personal information! Before
adding your comment, please read's Comment Policy.)

Your Name:

Your Location:

(ex: San Francisco, CA)

Your Comment:

Characters remaining:


The content on this page is free of advertiser influence and was produced by our editorial team. See our content and advertising policies.