Beast in the Belly
A new focus on early HIV infection in the gut and other mucosal tissues may generate novel strategies to study, treat, and prevent infection
By Philip Cohen, PhD
An untreated HIV infection is often pictured as a long, bloody battle. The virus infects a new host, the immune system strikes back and the fight enters into a prolonged struggle. Then, after many years, the immune system finally collapses and the virus is declared the winner.
At least that was the story told as judged from the perspective of peripheral blood samples, which is how researchers and clinicians have usually followed the infection. But evidence is accumulating that immunity's struggle with HIV may actually be fought—and lost—within the first few weeks of acute infection on battlegrounds in the gut and mucosal tissues.
"Early on, the immune system's boat is sunk and it is left in the middle of the ocean swimming for its life," is how Louis Picker of the Oregon Health and Science University describes the new model of HIV infection. "The only question after that is: how long will it last?"
If early HIV infection does deliver a mortal belly wound to the immune system, it suggests that tipping the balance in that first battle will be important for understanding AIDS, directing future research into the disease, as well as holding implications for drug treatment and vaccine design.
Researchers have long known that the primary targets of HIV (and its nonhuman primate-infecting relative SIV) are CD4+ T cells. During the chronic phase of infection, peripheral blood levels of these cells can be fairly constant, and it is only when their numbers drop that the infection develops into the symptoms and severe consequences of AIDS. Clinicians and researchers generally followed the battle in peripheral blood because it is a tissue that is easy to collect. But it hasn't been clear that the picture there was representative of the infection or the status of the immune system, given that the majority of CD4+ T cells reside in the gut and other mucosal tissues.
An early hint that peripheral blood wasn't telling the whole story came in a study from Ronald Veazey of Tulane National Primate Research Center and his colleagues. They were studying the effects of SIV infection in rhesus macaques in the gut-associated lymphoid tissue (GALT), which is host to different developmental stages of CD4+ T cells. In inductive sites such as intestinal Peyer's patches, these cells are naïve and resting. But on exposure to antigen, they traffic through the blood and home to effector tissues, the largest of these being the lamina propria (LP) in the intestinal villi.
The researchers infected animals with SIVmac239, a pathogenic strain, and then at one week intervals drew blood and performed biopsies on the animals from peripheral lymph nodes, spleen and the LP. After only one week into the infection, CD4+ T cells in the LP dropped to about half the level seen in uninfected controls and continued to slide down after that, nearly disappearing in some animals. However, at the same early time points in the same animals, CD4+ T cells in the blood and other tissues remained relatively constant (Science 280, 427, 1998). Later work by Veazey and collaborators showed the killing was focused on memory CD4+ T cells that also expressed SIV's (and HIV's) coreceptor CCR5 on their surface. Since these cells account for a high proportion of the cells in the LP, but a smaller percentage of CD4+ T cells in the periphery, this helped explain why killing may have been focused in this tissue (J. Virol. 74, 11001, 2000).
While the results suggested a similar rapid, early depletion of CD4+ T cells in the gut could play a role in HIV infection, some experts were skeptical. "What I heard from other investigators was: SIV in monkeys wasn't going to be like HIV in humans," says Veazey. "The monkey disease had a faster course, so it could involve a distinct mechanism of pathology."
Eventually though, reports from human clinical studies supported the idea that the modus operandi of HIV was very similar to SIV. Satya Dandekar at the University of California, Davis and her colleagues found that levels of CD4+ T cells all but disappeared from the gut following HIV infection, even in two patients who were infected only four and six weeks before the biopsies were taken (J. Virol. 77, 11708, 2003). Martin Markowitz and his team at the Aaron Diamond Research Center found the same remarkable depletion of the gut CD4+ CCR5+ effector memory T cells (J. Exp. Med. 200, 761, 2004). Ashley Haase of the University of Minnesota Medical School and Daniel Douek of the Vaccine Research Center (VRC) of the National Institute of Allergy and Infectious Diseases and their colleagues looked at the levels of CD4+ CCR5+ T cells in the gastrointestinal tract, lymph nodes and peripheral blood in patients who had been infected from less than one month to greater than five years. At all points of infection there was massive CD4+ CCR5+ T-cell depletion from the GALT. In contrast, gut CD8+ CCR5+ T cells were not depleted and CD4+ CCR5+ T cells in the blood and lymph nodes were nearly the same as in uninfected people (J. Exp. Med. 200, 749, 2004).
Mechanism of destruction
Then, last year, two high profile papers tackled the question of mechanism by examining in fine detail the early interplay between virus and CD4+ T cells in the SIV/rhesus macaque model. The two papers used different techniques and stressed different conclusions. Douek, Mario Roederer (also of the VRC) and their colleagues used PCR capable of detecting single molecules of HIV DNA to examine the extent of viral colonization in the first few weeks of an infection. They concluded that 30-60% of CD4+ memory T cells contained SIV DNA at the peak of infection and most of these cells could be detected during a 4-day period. And by using flow cytometry to narrowly focus on CD4+ memory T cells (CD45RA- or CD45RA+ CD95+) throughout the body, they found that the depletion of these cells was apparent at early stages even in peripheral blood, although the largest area of the depletion was the gut mucosa. They concluded that direct infection accounted for CD4+ memory T-cell killing, either due to virally-induced cell lysis or cytotoxic T-lymphocyte killing of infected cells (Nature 434, 1093, 2005).
In the second report, Haase's team looked for SIV RNA in GALT sections. They also found that most of the viral gene expression was restricted to CD4+ memory T cells. But by their calculations, at the peak of infection only 7% of CD4+ T cells had detectable viral RNA, and infected cells could account for only 20% of the measured depletion. Haase's team favors a model where direct exposure to viral particles, but not infection, drives cells to undergo apoptosis (programmed cell death). In support of this model the researchers found cellular markers of apoptosis in LP but not in inductive sites (Nature 434, 1148, 2005).
Both groups also discovered a surprising population of cells involved in the infection. Douek and Roederer found that more cells were infected with the virus than express FACS-detectable levels of CCR5, suggesting that levels of CCR5 expression below detectable levels can still facilitate viral entry. Haase found that more than 90% of productively infected cells were not activated or proliferating, cells which are normally assumed to be the preferred replication ground for SIV and HIV. Even though these "resting" cells produce 5-fold less virus per cell they are about 10-fold more abundant in the LP, creating a large substrate for transmitted virus to target. All these results suggest that the gut may contain a close packing of cells that are not traditionally thought of as supportive of HIV replication. "What I think is most impressive about the GALT is that it is chock full of viral targets," says Douek. "The virus gets in there and it's like a swarm of locusts going through a cornfield."
While in agreement on the extent of GALT CD4+ T-cell depletion, the two different conclusions on mechanism are driven by the limits of each group's technology. PCR can detect a single HIV DNA genome, but can't determine whether that genome is involved in an active infection or an aborted one. HIV RNA production is a direct measure of viral gene activity, but not as sensitive as PCR and therefore might miss low levels of HIV gene expression that could still prove lethal to cells. "It's an interesting and potentially important question whether many cells are graveyards for HIV/SIV genomes or whether many cells are covertly infected," says Haase. He says that he has set up a consortium including the VRC group to combine techniques to follow the fate of the virus in the GALT even more closely. Defining the exact mechanism of killing may have implications for early therapy to treat infections.
Too little, too late
As well as being hard hit in the first wave of viral infection, GALT cells may be ill prepared to defend themselves. In separate work, Haase's team has used the SIV/rhesus macaque model of intravaginal transmission to compare CD8+ T-cell responses in GALT, vaginal and other tissues. They documented strong immune responses in vaginal tissues: the frequency of CD8+ T-cell responses against immunodominant SIV Gag or Tat epitopes was above 5% at 21 days post-infection. But this response occurred after the peak of viral infection, doing little to contain the infection until after a persistent infection was well established. In contrast, the CD8+ T-cell response in the GALT was consistently the lowest response of any lymphatic tissue and never reached 5%, an immune response Haase and his colleagues characterize as "too late and too little" (J. Virol. 79, 9228, 2005).
Haase has since shown that even in animals infected rectally—a mode of delivery that normally favors the induction of intestinal immunity—the CD8+ T-cell response to SIV is weak and still lags behind that in the vaginal tract. "It's stunning how really crappy the GALT immune response is," says Haase. The results raise hope for protection of the vaginal mucosa if a vaccine could hasten the development of the immune response there—good news since it is a major entry point for the virus. But the work also suggests new strategies may be necessary to adequately protect the gut mucosa. One possible reason why the GALT may mount a poor defense is suggested by another recent paper from Haase, Picker, and Jeffrey Lifson of SAIC Frederick, Inc. at the National Cancer Institute and other researchers which shows that SIV infection appears to induce a premature regulatory T cell (Treg) response. Normally, a Treg response develops to limit immune responses to prevent damage due to excessive immune activity. But in the case of SIV-infected GALT, Tregs might arise too soon and blunt an immune response before an adequate one even develops (J. Infect. Dis. 193, 703, 2006).
There is also evidence that once damaged the GALT may be slow to recover. In their published work Dandekar and Markowitz both studied HIV-infected individuals on HAART. They found that while years of drug therapy was effective at restoring CD4+ T-cell levels in blood and lymph nodes, reconstitution of the CD4+ T-cell population in the GALT lagged far behind for patients who had entered the chronic phase of infection. Interestingly, for patients treated within the first few weeks of infection, GALT CD4+ T-cell restoration was far better but still incomplete. Similar results were recently reported at the 13th Conference on Retroviruses and Opportunistic Infections (CROI) based on longitudinal studies of patients. Dandekar reported that the suppression of viral replication and inflammation in gut tissue during therapy best correlated with the degree of mucosal CD4+ T-cell restoration during therapy. She has also recently reported that long term non-progressors (LTNPs) actually have slightly higher levels of GALT CD4+ T cells than healthy controls, suggesting this may contribute to the lack of disease progression in these individuals (Proc. Natl Acad. Sci. USA 102, 9860, 2005).
Such results have thrown new focus on the issue of when to commence drug treatment for HIV infection. The benefits of early treatment versus delaying exposure to drugs and their possible side effects has long been discussed in the context of chronically infected patients. Now the question is whether initiating treatment in the first few weeks of infection would confer the added benefit of preserving GALT CD4+ T cells. Experts agree, though, that it is unrealistic to expect to identify most HIV-infected individuals at such an early stage. One exception is in the case of perinatal HIV infection. A recent study of 205 children found that very early treatment—by age 2 months, as opposed to 3 to 4 months—was associated with delayed and decreased progression of disease. The study did not, however, directly assess the effect on the GALT immunodepletion (JAMA 293, 2221, 2005).
A situation where this goal of exceptionally early treatment might be achieved is within studies designed to identify acute infections. This includes pre-exposure prophylaxis (PrEP) clinical trials, which are testing the ability of tenofovir therapy to prevent HIV infection in volunteers at high risk. Trial participants with breakthrough infections would thus already be on drug therapy before their infection is detected. In an SIV-macaque model, Lifson, Veazey and their colleagues have shown that even early post-infection tenofivir treatment can result in preservation of GALT CD4+ T cells and low to undetectable blood levels of virus after 30 days of post-infection therapy (J. Med. Primatol. 32, 201, 2003). It isn't yet clear if there would be any benefit of trying to restore GALT at later phases of disease or even how to achieve restoration. "The gut work tells us that what happens early is actually quite striking," says Michael Lederman of Case Western Reserve University. "The question now is if the deterioration of immunity and AIDS that we see later on is related to this early depletion and how." It will be important, for example, to see if the degree of GALT CD4+ T-cell depletion is predictive of disease progression.
While researchers are trying to make that link by conducting prospective studies of HIV infection, some general theories are emerging of how devastation of CD4+ T cells in the GALT may determine the course of disease. As Veazey and his Tulane colleague Andrew Lackner write in a recent commentary (Nat. Med. 11, 469, 2005) the revelation that HIV infection eliminates much of the memory CD4+ T-cell population in the first days of infection suggests the subsequent decline of the immune system is the result of a "valiant but futile effort to replace these cells." Indeed Picker, Lifson, and their colleagues have reported that the success of this effort may be a good indicator of disease progression. They assessed the aftermath of CD4+ T-cell depletion in SIVmac239 infected macaques using bronchoalveolar lymphocytes obtained by lung lavages as a window into the mucosal immune system. In some animals the depletion was followed by a significant boost in the production of short-lived CD4+ memory T cells which migrated to the lungs. In these animals the infection was stable and did not rapidly progress to disease. But in monkeys where this capacity to partially repopulate the mucosal CD4+ T-cell compartment quickly failed, disease progression was rapid (J. Exp. Med. 200, 1299, 2004).
This work suggests that the very immune activation that holds off immediate collapse of the immune system in the aftermath of GALT immunodepletion may also have two negative effects: creating activated cells to feed the next round of viral replication and setting the rate at which crucial immune cells are used up and at which disease progresses. "Once the intial depletion occurs, the regenerative capacity of the immune system is likely engaged in a war of attrition it cannot win in the long term," says Lifson. "While it may take a year or two in a macaque and a decade or more in a person to progress to AIDS, much of the eventual outcome is probably determined in the earliest stages of the infection." In another recent perspective article, Douek and other VRC team members also focus on subsequent immune activation as a possible link to disease progression and suggest how mucosal depletion itself may drive that activation (Nat. Immunol. 7, 235, 2006). These authors cite some of their own unpublished findings that immune activation persists even in HIV-infected people whose viral replication is highly suppressed by drug therapy. They propose that the massive depletion of CD4+ T cells during acute infection and possibly other changes in the GALT destroy the integrity of the normally tight immunological barrier in the intestine. As a result the systemic immune system becomes chronically subject to antigens from intestinal flora leading to increased immune system activation and other pathological changes such as fibrosis of lymph nodes.
Clearly, connecting the dots between early events and pathological changes that result in AIDS won't be easy. Many of the crucial events may occur in the GALT and other mucosal tissues that can only be accurately followed with invasive tissue biopsies. "You can't biopsy a thousand patients in a vaccine trial," says Veazey. One alternative strategy now being explored is to catch the cells en route to these destinations by using flow cytometry of T cells in the blood to examine cell surface proteins that determine where effector memory T cells migrate, so-called "homing markers." A number of promising markers are already being explored, among them CCR9, CCR10, and CD103, but a great deal of careful work is still needed to see how expression of these proteins relates to the immunological state of the GALT, especially in the context of HIV infection.
For vaccine designers, an important goal will be vaccines that induce secretion of antibodies and elicitation of cytotoxic T lymphocytes in this tissue. "I think the handwriting is on the wall," says Markowitz. "If a vaccine is going to be effective, it better be effective at mucosal sites." Roederer says that if appropriate markers could be found to monitor the immunological state of the GALT using samples of peripheral blood, it would be a valuable indicator of vaccine effectiveness. "Of course, the ultimate goal is for a completely protective vaccine," says Roederer. "But if people who are vaccinated become infected it's important to know if they're protected in any meaningful way. If we can prove a good correlation between the degree of early GALT depletion and later disease progression, for instance, then we'd have a rapid marker we can use in vaccine trials and know the benefit in weeks, rather than years."