HIV’s siege of the mucosa establishes the systemic crisis of immune-hyperactivation and may have semi-permanent effects

Gareth Hardy, Case Western Reserve University, Cleveland, Ohio

One of the major undiscovered anomalies of HIV disease that has defied explanation until now is the cause of immune hyper-activation. The initial perspective of HIV as an immunodeficiency disease was confounded by observations of polyclonal B-cell activation, hyper-gammaglobulinaemia, abundant plasma proinflammatory cytokines and the clinical resemblence of cachexia. Subsequently the immune activation story unfolded with reports describing heightened expression and frequency of cellular activation markers, increased activation induced cell death, spontaneous T cell apoptosis, and in more recent years increased T cell turn over and production, shortened telomere lengths, perturbations in the TCR repertoire, and clonal exhaustion and senescence.

Alongside the abundance of pro-inflammatory cytokines detected in the blood of HIV-positive patients, such as interferon-alpha, interferon-gamma and tumor necrosis factor-alpha, which have never been adequately explained, the high level of CD38 expression, and parallel HLA-DR positivity, on CD8 T cells is associated with a highly pathogenic state of immune activation and subsequent exhaustion of the circulating T cell pool. It is probable that all these manifestations of immune activation do not share the same pathological mechanism. For example, clonal exhaustion and perturbations in the TCR repertoire may result directly from the cellular immune response to an ever changing and overwhelming HIV antigen burden. Whereas the disease-associated heightened expression of activation markers by T cells is polyclonal and not HIV-specific.

Understanding the causes which underlie immune activation are important because there is a strong logical connection between hyper-activation of the immune system, rampant destructive viral replication, giving rise to the 109 viral particles produced every day in the untreated individual, and destruction of the CD4 T cell population which is ultimately fatal. Regardless of whether precipitous CD4 T cell loss is the result of the preferential infection and replication of HIV, and SIV, in activated CD4 T cells, or the result of such an increase in the proportions of T cells destined to die from becoming activated that homeostatic replacement becomes unsustainable, the key player in both mechanistic models is excessive immune activation, which is a better indicator disease progression than viral load. [1]

Last December, Jason M. Brenchley and colleagues at the Vaccine Research Centre (VRC), National Institutes of Health (NIH), Bethesda, USA, published the latest results of their continued investigations into the implications of the ferocious war HIV wages at the gastrointestinal mucosa, revealing what appears to be a very convincing explanation for the immune activation which plays a pivotal role in the pathogenesis of HIV infection. [2]

The group at the VRC, headed by Daniel C. Douek, have been conducting a series of intensive investigations into events in the gut mucosa from the first stages of infection. Initially, Douek’s group overturned the previously established view that CD4 cell loss in HIV-infection is a gradual process of attrition which takes years. This understanding of the rate of CD4 cell destruction was largely the result of measurements of CD4 cell numbers in the peripheral blood, where absolute CD4 cell numbers fall relatively slowly, at approximately 60-100 cells/mL blood per year. Since HIV predominantly infects memory CD4 T cells, and the majority of memory CD4 T cells reside in lymphoid tissues in the gut, the VRC group focused their attention on gut mucosal associated lymphoid tissue (MALT) finding that around 80% of memory CD4 T cells are depleted from this area in the first few days of infection. [3]

This represents a massive loss of the overall numbers of memory CD4 T cells in the body, in a very short period of time. Earlier similar reports describe huge losses in CD4 T cell numbers in the gastrointestinal lamina propria of rhesus macaques in SIV infection. [4]

In an editorial on the Brenchley paper in December edition of Nature Medicine [5], Barton Haynes, of Duke University, Durham, North Carolina, USA, outlines the role of MALT and the implications of Brenchley’s most recent results, connecting these events with immune activation. MALT consist of follicle areas in the mucosa located in the intestines, tonsils, appendix and peribronchial areas. These areas are exposed to large quantities of foreign antigens and must distinguish between antigens of innoquous origin and those of pathogenic microbial origin to which T and B cell responses must be instigated. On the mucosal surface follicle-associated epithelial cells (FAE) pick up antigens and pass them to dendritic cells which await in the underlying lamina propria (LP), accompanied by effector T cells surveying for microbial infiltration. Alternate recognition of different types of antigen by various of the thirteen described mammalian toll-like receptors (TLRs) on dendritic cells is probably responsible for the differentiation between pathogen associated and innoquous antigen sources. Thus the interaction between dendritic cells and lymphocytes in mucosal follicles, together with the physical barrier of the epithelium and coating mucous, prevent the large abundance of foreign antigens present at these sites from making contact with the systemic immune system, as this would otherwise result in massive inflammatory responses.

Following observations that the lamina propria is a major site for viral replication and CD4 T cell destruction in SIV infection, Qingsheng Li in Ashley T. Haase’s group at the University of Minnesota in Minneapolis, USA, reported that SIV largely replicated in the abundant pool of resting memory CD4 T cells at this site, rather than the considerably smaller pool of activated memory CD4 T cells. [6]

Furthermore, the inflammatory consequences of viral replication in the gastrointestinal mucosa gave rise to high levels of T cell and epithelial cell apoptosis. This process of epithelial apoptosis was related to enteropathic changes, which are associated with increased permeability of the gut mucosal epithelium in HIV disease. Haase’s immunohistochemistry of biopsied colonic MALT sections shows red chromogen-stained apoptotis-associated caspase-3 expression on CD4 T cells and apoptosis-associated cytokeratin-18 expression on epithelial cells. Like a burning necklace around the mucosa, Haase refers to this as HIV’s “red ribbon of apoptotic death”. In the March 2006 edition of Nature Immunology, Jason M. Brenchley, David A. Price and Daniel C. Douek explained that the consequence of heavy losses in mucosal CD4 T cell numbers as a direct result of HIV-mediated destruction of these cells in the first days of infection could result in significant damage to the integrity of the mucosal barrier. [7]

The implication of this damage to the microenvironment of the intestinal mucosa, the gastrointestinal lymphoid tissue, and loss of CD4 memory T cell numbers, is that ensuing breaches in the integrity of the gut wall may allow translocation of bacterial products into the systemic circulation. The gastrointestinal tract is by far the largest, and more or less the only large reservoir in the body for gram-negative bacteria, which synthesise lipopolysaccharide (LPS), a potent immune activator, and a component of the gram-negative bacterial cell wall. Translocation of bacteria and bacterial products, such as LPS, across the gut wall is a known cause of immune activation in other disease settings such as bone marrow transplantation associated graft versus host disease (GVHD) and also in inflammatory bowel disease (IBD), even without any sign of overt bacteraemia.

The first reported finding in this latest publication from the Douek group at the VRC is the ground-breaking observation that circulating LPS is significantly increased in both HIV and SIV infection. However LPS is only used an indicator of microbial translocation in this study and in fact a number of other microbial products, such as double-stranded and single-stranded RNA and DNA, peptidoglycan and flagellin may translocate across the gut wall inducing effects additional to those observed that result from LPS translocation. HIV-infected patients with chronic progressive disease, including those reaching the CD4 definition for AIDS, had significantly higher plasma levels of LPS than uninfected healthy controls (p = <0.0001). None of the HIV-infected patients in this study had opportunistic infections or overt signs of bacteriaemia. Patients with both chronic HIV infection and with AIDS had significantly higher plasma LPS levels than patients with acute or early HIV infection (p = <0.0001). There was no significant difference in plasma LPS levels between patients with chronic HIV infection and those reaching AIDS definition, nor was there a difference between patients with acute early HIV-infection and uninfected healthy controls. The lack of any difference between acutely infected patients and uninfected controls demonstrates that plasma LPS levels do not become elevated until the chronic phase of infection. Assuming a direct relationship between HIV-mediated destruction of memory CD4 T cells in the gastrointestinal musosa, the resulting inflammatory damage to the mucosal epithelium, and translocation of LPS to the peripheral blood, it is apparent that manifestation of this latter effect is considerably delayed, possibly due to immunological counter actions that neutralise LPS.

In addition to the human data, 11 rhesus macaques were infected with pathogenic SIVmac. 100 days later plasma LPS levels had risen in all but one animal compared to baseline for infection (p = 0.002). In order to confirm the gastrointestinal source of LPS, two of these animals were treated with a cocktail of “bowel sterlising” antibiotics – neomycin (10mg per kg body weight), metronidazole (40mg/kg) and cefotaxime (150 mg/kg). Fecal bacterial loads and plasma LPS levels were monitored over two weeks. One week into treatment plasma LPS levels fell and 11 of 12 observed Gram-negative bacterial species disappeared from fecal matter along with one of four Gram-positive species. By two weeks plasma LPS levels began to rise again presumably because of the outgrowth of other bacterial species.

Innate and adaptive immune responses to such increased levels of LPS include a number of soluble factors. CD14 is a receptor for LPS expressed by monocytes and macrophages, and can be secreted in soluble form to bind LPS. In addition monocytes express the pro-inflammatory cytokines interleukin-1 beta (IL-1b) and tumor necrosis factor (TNF) as a response to the presence of LPS. Systemically LPS binding protein (LPB) is secreted by the liver and gastrointestinal epithelial cells in response to LPS. Antibodies to the LPS core oligosaccharide, endotoxin core antibodies (EndoCAb), increase in conditions of acute microbial translocation to neutralise LPS. Brenchley and colleagues investigated the levels of these soluble factors in the plasma of HIV-infected patients and uninfected controls by ELISA, and the expression of TNF and IL-1 in monocytes by intracytoplasmic flow cytometry.

Plasma soluble CD14 (sCD14) levels were significantly elevated in both acutely/early infected and progressor patients (pooled chronically infected and AIDS definition patients) compared to controls (both p = <0.0001), suggesting chronic stimulation by LPS in vivo. In progressors there was a statistically significant relationship between plasma sCD14 levels and plasma LPS levels (r = 0.3, p = 0.001) which the authors suggest implies that LPS directly stimulates sCD14 secretion in vivo. Although previous reports have suggested that HIV gp120 could stimulate monocytes to secrete sCD14 Brenchley and colleagues could not find a relationship between viral load and sCD14 in these patients (p = 0.5). Plasma levels of LPB were also significantly elevated in progressors compared to controls (p = 0.0099), but not in acute/early infected patients (p = 0.57). Similarly there was a positive correlation between LPB and sCD14 levels in all patients. Following in vivo stimulation with LPS, healthy monocytes subsequently become unresponsive to LPS, diminishing the in vitro response in stimulation assays. In concordance with this the percentage of monocytes that responded to LPS stimulation in vitro by expressing IL-1b and TNF was reduced in individuals with higher plasma LPS levels and inversely correlated with plasma LPS levels for all patients (r = -0.5263, p = 0.0171). In contrast EndoCAb titres were significantly reduced in both progressors and early/acutely infected patients compared to uninfected controls (both p = <0.0001). There was also a significant reduction in EndoCAb titres in progressor patients compared to acutely/early infected patients (p = 0.0002). EndoCAb titres significantly inversely correlated with plasma LPS levels in all patients (r = -3190, p = 0.0005). The authors suggest this implies that microbial translocation across the gut wall occurs in early HIV-infection (as evidenced by increased sCD14 in the early/acute cohort, with apparently normal LPS levels), and LPS is neutralised by EndoCAbs. In other conditions of chronic microbial translocation, such as inflammatory bowel disease (IBD), EndoCAb titres are elevated. However the decrease in EndoCAb titres in chronic progressive HIV infection, coupled with increasing LPS levels, suggests, according to the authors, that EndoCAbs titres are insufficient to neutralise circulating LPS and prevent immune activation, which may be the result of abnormalities in B-cell responses in HIV-infection.

Since LPS is an indicator of microbial translocation which may involve multiple different additional products derived from bacteria, fungi and viruses, Brenchley and colleagues investigated whether elevated LPS levels were associated with other forms of immunological activation not directly associated with LPS. Levels of interferon alpha (IFN-alpha) have been associated with disease progression and immune activation. The principle cell type responsible for producing IFN-alpha is the plasmacytoid dendritic cell (pDC). However pDCs do not express known receptors for LPS, such as Toll-like receptor-4 (TLR-4). Brenchley and colleagues found a significant relationship between plasma LPS levels and plasma IFN-alpha levels in patients with progressive disease (r = 0.624, p = <0.0001). This suggests either that pDCs are being stimulated by other microbial products, known as pathogen (or microbial) associated molecular patterns (PAMPs), or that IFN-alpha is being produced in excess by other cells either in response to LPS or other microbial products. The former is the more likely explanation since pDCs characteristically produce orders of magnitude more IFN-alpha than other cell types. Furthermore Brenchleyand colleagues also found a significant relationship between plasma LPS levels and a principle marker of immune activation in HIV-infection, the co-expression of CD38 and HLA-DR by CD8 T cells, in a cohort of chronically infected and AIDS defined patients (r = 0.6970, p = 0.0306) and a separate cohort of chronically infected and uninfected individuals (r = 0.5553 p = 0.0059). This indicates that translocated microbial products may induce the T cell hyper-activation that is a hallmark of pathogenic immunodeficiency virus infection, directly, or indirectly via an overload of inflammatory cytokines.

HAART partially reduced plasma levels of LPS in 28 chronically infected patients after 48 weeks treatment. The difference in plasma LPS between week 48 and baseline for HAART was significant (p = 0.0107), although the difference between plasma LPS levels after 48 weeks of HAART and normal controls was also significant (p = 0.0026) suggesting that repair of mucosal immunity and the gastrointestinal epithelium is not sufficient to prevent microbial translocation following arrest of viral replication. Plasma sCD14 levels also did not change with HAART further suggesting that microbial translocation persists despite HAART. Interestingly the change in CD4 count after 48 weeks HAART was significantly inversely associated with plasma LPS levels at that time (r = -0.4628, p = 0.0151). Since no relationship was found between plasma LPS levels and either viral load or CD4 count the authors argue that a relationship between microbial translocation and CD4 T cell reconstitution is likely to be complex.

The cohort of long-term non-progressors (LTNPs) displayed elevated levels of plasma LPS in comparison to uninfected controls (p = <0.0001), but slightly lower levels than progressors which was not significant. Of note was the observation that in contrast to progressors, LTNPs maintained higher levels of EndoCAbs (p = 0.06), even with substantially raised plasma LPS levels. LTNPs also maintained lower levels of sCD14 than progressors (p = 0.0011) and LBP than progressors (p = 0.0025), suggesting that the immunostimulatory activity of LPS is attenuated in LTNPs compared to progressors, probably due to effective inactivation of LPS by higher levels of EndoCAb titres.

An alternative model of attenuated pathogenesis of immunodeficiency virus infection is the sooty mangabey model. These primates exhibit the same high viral loads as in pathogenic immunodeficiency virus infection, but do not experience disease progression, immunosuppression, CD4 T cell depletion, or immune activation. Unlike HIV infection in humans and pathogenic SIV infection in rhesus macaques, non-pathogenic SIV infection in sooty mangabeys gave rise to no increase in either plasma LPS, sCD14 or EndoCAb levels compared to uninfected animals. Thus it appears that a defining difference between pathogenic and non-pathogenic immunodeficiency virus infection is the integrity of the gastrointestinal mucosal barrier, which in turn determines protection or susceptibility to consequent immune activation and disease progression.

Aside from LTNPs, the one group of HIV-infected patients most likely to avoid this damage to the gastrointestinal mucosa and its consequent pathological effects are patients who initiate HAART early during acute infection and remain on treatment without interruption. Saurabh Mehandru and colleagues in Martin Markowitz’s group at the Aaron Diamond AIDS Research Centre, Rockefeller University, New York, addressed this issue in a separate paper published in December. [8]

Mehandru studied 40 patients who had initiated HAART during acute or early HIV infection, finding that 70% of this cohort demonstrated continued loss of CD4 T cells in the lamina propria of the gastro-intestinal tract after 5-7 years of uninterrupted therapy with continuous viral suppression. A total of 54 patients with acute or early HIV-infection were studied of whom 40 were assessed for changes in the gastro-intestinal mucosa after initiation of HAART. Of these 40, 18 were followed longitudinally from HAART initiation with sequential colonic biopsies over a three year period. The remaining 22 were examined cross sectionally at between 1 and 7 years of HAART, with no prior assessment at baseline for HAART. Eighteen uninfected controls were also studied. The size of the CD4 T cell population was assessed in biopsies from both gut lymphatic tissue (organised lymphatic tissue – OLT) and the lamina propria (LP), where effector memory T cells recirculate for surveillance of microbial infiltration, thus referred to in the paper as the “effector site”. On the one hand percentages of cells were assessed by flow cytometry, following enzymatic removal of mucosal mononuclear cells (MMC) and physical disruption of the biopsy and on the other hand by immunohistochemistry of paraffin-embedded biopsied tissue sections. The percentages of CD4 T cells found in the mucosa and blood are compared between the 32 patients biopsied before HAART and the cross sectional patients biopsied between 1 and 7 years after HAART initiation. The mean peripheral blood CD4 percentage in uninfected controls was 59.6, and the mean CD4 percentage in mucosal mononuclear cells was 56.3. Thus in uninfected individuals the relative CD4 percentages in blood and the gut mucosa are equivocal. In untreated acute/early infected (AEI) patients this relationship was very different, the mean blood CD4 percentage was 41.5, where as in the mucosal mononuclear cells it was 19.3. By 3 – 7 years HAART in the cross sectionally studied patients, these numbers were 58.1% in the peripheral blood and 42.3% in mucosal mononuclear cells.

Of the 32 patients who were assessed prior to initiation of HAART, 18 were subsequently followed longitudinally. However only a small number of these patients opted for sequential colonic mucosal biopsies, an invasive and uncomfortable procedure. By one year of HAART (n = 5) the difference between mean blood CD4 percentage, 57.9, and mucosal mononuclear cell CD4 percentage, 34.6, was significant (p = <0.001). By two years (n = 9) this difference retained the same significance and by three years HAART (n = 4) the mean blood CD4 percentage was 67.3 and the mean mucosal mononuclear cell CD4 percentage was 41.3 (p = 0.02).

Using biopsies from cross sectionally and longitudinally studied patients (n = 30), as well as controls, these results were confirmed by immunohistochemistry of the lamina propria and of organised lymphatic tissue. Multiple standard set areas were assessed for each tissue section. After one, two and three years of HAART the numbers of CD4 T cells in the mucosal lymphoid tissue were not different to those for uninfected controls, however the numbers of CD4 T cells in the lamina propria were significantly depleted even out to three years, though with moderate improvement by this time.

Nine of these patients demonstrated normalisation of CD4 numbers in the lamina propria, these were termed “reconstitutors”. In the remaining 21 patients (70%) persistent depletion of CD4 T cells was observed (mean CD4 count 5.5 cells, p = <0.001), these patients were termed “non-reconstitutors”. The authors found that the only factor associated with non-reconstitution in the lamina propria was the percentage of memory (CD45RO) CD8 T cells expressing the activation marker HLA-DR at baseline for HAART and post-treatment. Unfortunately the authors did not examine the percentages of memory CD8 T cells expressing the activation marker CD38, which may have been a more pertinent marker of pathological activation. Interestingly the numbers of T cells expressing Ki-67, a marker of cell cycle progression indicating cellular activation and proliferation, was increased in the gastro-intestinal mucosa. In uninfected controls immunohistochemistry for Ki-67 revealed expression levels of 3.6% in CD4 T cells and 1.5% in CD8 T cells, whereas 12.4% of CD4 T cells expressed Ki-67 in the same site in untreated acute/early infected patients and 14.2% of CD8 T cells expressed Ki-67. After HAART these levels persisted, with gradual reductions over time, which approached normal levels by 3-7 years treatment – 4.4% of CD4 T cells expressed Ki-67 and 7.25% of CD8 T cells expressed Ki-67.

The sustained depletion of CD4 T cells in the gut mucosa even after as long as 7 years of HAART in patients who initiated therapy about as early as reasonably achievable, and the associated persistent increase in activation markers associated with disease pathology, is a considerable concern for the long-term well being of people living with HIV on virologically successful antiretroviral therapy. Mehandru et al explain that while patients receiving HAART may have up to 50% reductions in CD4 T cells in the lamina propria, there do not appear to be short-term adverse consequences of this, which may be a result of immunological redundancy. However the long-term clinical consequences of persistent depletion of mucosal immunity may manifest as patients on HAART grow older, particularly given the possibility that prolonged mucosal CD4 lymphopenia may accelerate ageing of the immune system.

It is now evident the events in the gut mucosa are central and pivotal to the pathogenesis of HIV and other immunodeficiency virus infections. In his editorial on the Brenchley paper, Barton Haynes points out that while the association between HIV-infection and microbial translocation is compelling and robust, Brenchley et al have not yet directly shown that microbial translocation causes immune activation in HIV-infection. [5]

Two important inconsistencies remain to be explained. Firstly, in non-pathogenically SIV-infected sooty mangabeys gut CD4 T cells still become depleted, yet plasma LPS levels do not become elevated. Secondly, immune activation is apparent in acute HIV-infection yet plasma LPS levels are not elevated. Brenchley et al suggest that immune activation in early HIV-infection may be a result of other stimuli, most likely the massive early burden of HIV itself. Haynes goes on to explain that it is now important to establish if bacterial translocation is a result of loss of immune cells in the gut, epithelial cells or both. Brenchley et al conclude that their findings open up new areas for therapeutic intervention, namely to “prevent or reduce the propogation of HIV at mucosal surfaces, to restore the immunological and epithelial integrity of the mucosal barrier and to block the cellular and molecular pathways through which microbial products cause systemic immune activation”.

Characterisation of the mechanistic link between HIV-mediated destruction of gut mucosal epithelial CD4 T cells and translocation of microbial products, and why this link is abrogated in sooty mangabeys, may possibly reveal a pivotal switch between pathogenic and non-pathogenic immunodeficiency virus infection. Furthermore, the link between circulating microbial products and immune activation is now set to become a key area for continued research. Therapeutic approaches aimed at intervening in this pathological chain of events should be anticipated.


  1. Janis V. Giorgi, Lance E. Hultin, Jane A. McKeating, Timothy D. Johnson, Bronwyn Owens,1 Lisa P. Jacobson, Roger Shih, Julie Lewis, Dorothy J. Wiley, John P. Phair, Steven M. Wolinsky, and Roger Detels. Shorter Survival in Advanced Human Immunodeficiency Virus Type 1 Infection Is More Closely Associated with T Lymphocyte Activation than with Plasma Virus Burden or Virus Chemokine Coreceptor Usage. The Journal of Infectious Diseases.1999;179, 859-870
  2. Jason M Brenchley, David A Price, Timothy W Schacker, Tedi E Asher, Guido Silvestri, Srinivas Rao, Zachary Kazzaz, Ethan Bornstein, Olivier Lambotte, Daniel Altmann, Bruce R Blazar, Benigno Rodriguez, Leia Teixeira-Johnson, Alan Landay, Jeffrey N Martin, Frederick M Hecht, Louis J Picker, Michael M Lederman, Steven G Deeks & Daniel C Douek. Microbial translocation is a cause of systemic immune activation in chronic HIV infection. Nature Medicine. 2006:12, 1365 – 1371.
  3. Jason M. Brenchley, Timothy W. Schacker, Laura E. Ruff, David A. Price, Jodie H. Taylor, Gregory J. Beilman, Phuong L. Nguyen, Alexander Khoruts, Matthew Larson, Ashley T. Haase, and Daniel C. Douek. CD4+ T Cell Depletion during all Stages of HIV Disease Occurs Predominantly in the Gastrointestinal Tract. J Exp Med. 2004:200, 749-59.
  4. Ronald S. Veazey, MaryAnn DeMaria, Laura V. Chalifoux, Daniel E. Shvetz, Douglas R. Pauley, Heather L. Knight, Michael Rosenzweig, R. Paul Johnson, Ronald C. Desrosiers, Andrew A. Lackner. Gastrointestinal Tract as a Major Site of CD4+ T Cell Depletion and Viral Replication in SIV Infection. Science. 1998:280, 427 – 431.
  5. Barton F Haynes. Gut microbes out of control in HIV infection. Nature Medicine. 2006:12, 1351 – 1352.
  6. Qingsheng Li, Lijie Duan, Jacob D. Estes, Zhong-Min Ma, Tracy Rourke, Yichuan Wang, Cavan Reilly, John Carlis, Christopher J. Miller and Ashley T. Haase. Peak SIV replication in resting memory CD4+ T cells depletes gut lamina propria CD4+ T cells. Nature. 2005:434, 1148-1152.
  7. Jason M Brenchley, David A Price & Daniel C Douek. HIV disease: fallout from a mucosal catastrophe? Nature Immunology. 2006:7, 235 – 239.
  8. Saurabh Mehandru, Michael A. Poles, Klara Tenner-Racz, Patrick Jean-Pierre, Victoria Manuelli, Peter Lopez, Anita Shet, Andrea Low, Hiroshi Mohri, Daniel Boden, Paul Racz, Martin Markowitz. Lack of Mucosal Immune Reconstitution during Prolonged Treatment of Acute and Early HIV-1 Infection. PLoS. 2006:3, e484

Links to other websites are current at date of posting but not maintained.