|Home | About | Journals | Submit | Contact Us | Français|
Despite suppression of human immunodeficiency virus (HIV) replication by antiretroviral therapy, reconstitution of CD4+ cells is variable and incomplete, particularly in gut-associated lymphatic tissues (GALT). We have previously shown that immune activation and inflammation in HIV-infected and simian immunodeficiency virus–infected lymph nodes results in collagen deposition and disruption of the lymphatic tissue architecture, and this damage contributes to CD4+ cell depletion before treatment and affects the extent of immune reconstitution after treatment. In the present study, we compared collagen deposition and the extent of depletion and reconstitution of total CD4+ cells and subsets in peripheral blood, lymph nodes, and inductive and effector sites in GALT. We show that CD4+ cell depletion in GALT correlates with the rapidity and greater magnitude of collagen deposition in this compartment, compared with that in peripheral lymph nodes, and that although treatment does not restore CD4+ cells to effector sites, treatment in the early stages of infection can increase CD4+ central memory cells in Peyer patches.
Significant increases in peripheral blood CD4+ T cell counts with antiretroviral therapy (ART) have been extensively documented [1–3]. However, the dynamics and extent of CD4+ cell depletion and reconstitution during treatment may be substantially different in secondary lymphatic tissues and gut-associated lymphatic tissue (GALT), which collectively contain most CD4+ cells (98%). GALT suffers greater losses of CD4+ cells, compared with peripheral blood, in both simian immunodeficiency virus (SIV) infections and HIV infections; restoration in GALT, compared with that in the blood, is slow and incomplete when treatment is initiated in the chronic stage of infection [4 – 8]. Even if treatment is initiated in the early stages of infection, it is not clear whether it will substantially increase restoration of CD4+ cells in the gut. There have been reports of increases of gut CD4+ cells [9, 10], albeit delayed compared with the increases in peripheral blood, but even after 1–7 years of ART most patients continue to have substantial depletion (50%– 60%) of gut lamina propria lymphocytes .
In this article, we report studies of collagen deposition as one mechanism that contributes to the depletion of GALT CD4+ cells and limits reconstitution of these cells during treatment. We have previously shown that collagen deposition in lymph nodes of HIV-1–infected patients (hereafter, “HIV+ patients”) is correlated with CD4+ T cell depletion and with the extent of reconstitution observed during treatment, as well as that collagen deposition in the early stages of SIV infection is also correlated with the extent of CD4+ cell depletion [12–15]. In this study, we first compared differences in CD4+ cell counts in HIV− and HIV+ patients and changes in cell counts in CD4+ cell subsets before and after therapy in peripheral blood, lymph nodes, and gut lymphoid tissue. With these studies, we establish on a compartment and subset-specific basis the preferential depletion and relatively limited reconstitution of CD4+ cells in gut and that the extent of collagen deposition in the gut is correlated with this preferential depletion and impaired reconstitution.
HIV− individuals and ART-naive HIV+ persons were recruited into a study approved by the University of Minnesota institutional review board . Patients at all CD4+ cell count levels were enrolled, and patients were placed in 1 of the following 4 groups: HIV−, acute-early (individuals who were either HIV-antibody− and plasma HIV RNA+, or within 6 months after a documented seroconversion), presymptomatic (individuals with CD4+T cell counts >200 cells/mm3 and a positive HIV antibody test >6 months ago), or AIDS (individuals with CD4+ T cell counts <200 cells/mm3).
All study procedures were carried out at the University of Minnesota Medical Center (UMMC) and the National Institutes of Health–funded General Clinical Research Center and have been previously reviewed . Subjects were asked to undergo an inguinal lymph node biopsy to excise a lymph node (LN) and a colonoscopy to the terminal ileum with biopsy. Immediately after baseline samples of peripheral blood, LN tissue, and GALT were collected, HIV+ persons were started on ART. After 6 months, 19 HIV+ subjects again underwent all study procedures (15 were receiving ART and 4 were not).
Venous blood was used to measure CD4+ cell count by flow cytometry in the UMMC clinical laboratory (certified by the AIDS Clinical Trials Group for such procedures). Tissue biopsies (LN and ileal GALT samples) were divided; one portion was processed by immunohistochemical staining for quantitative image analysis to determine the absolute size of the total CD4+ cell population, and the remaining portion was processed by flow cytometry to proportionately quantify CD4+ cells (total, naive, central memory [CM], and effector memory [EM] T cells). These methods have been published elsewhere [13–16]. For flow cytometry, a portion of LN tissue and GALT was placed on ice and CD4+ cells were isolated within 24 h by gently separating the cells from surrounding tissue by using a mesh screen. One million cells were washed once in FACS wash (PBS supplemented with 0.1% sodium azide and 2% bovine serum albumin; Sigma). After aspiration of the supernatant, cells were stained with peridin clorophylla protein– conjugated CD4, allophycocyanin conjugated CD8, phycoerythrin conjugated CD27, and fluorescein isothiocyanate– conjugated CD45RO (all BD Pharmingen), and incubated for 30 min at 4°C, followed by another wash. Cells were fixed with 1% paraformaldahyde (Electron Microscopy Sciences) and analyzed on a FACS Calibur flow cytometer (BD Pharmingen). Lymphocytes were gated on the basis of characteristic forward and side scatter properties, followed by separation into CD4+ T cells and CD8+ T cells on the basis of expression of CD4 and CD8. Naive T cells were classified by expression of CD27 without expression of CD45RO, as described elsewhere . CM T cells were classified by coexpression of CD27 and CD45RO, and EM T cells were classified by lack of CD27 expression.
To quantify the total CD4+T cell population in each compartment, 4-μm sections were prepared from the fixed tissues and stained with antibody for CD4 by using either clone 1F6 (Ventana Medical Systems) or clone 4B12 (Neomarkers; Lab Vision). Images were captured to quantify the percentage of tissue area occupied by CD4 by using Photoshop (CS2, version 9.0; Adobe Systems) with plug-ins from Reindeer Graphics.
Because only 7 of the 15 patients who received ART for 6 months had sufficient numbers of GALT cells both at baseline and after 6 months of ART to allow quantification of CD4+ cell subsets by flow analysis, we developed a second method to quantify central memory CD4+ cells in Peyer patches of GALT. We obtained immunofluorescent images of sections triple labeled with antibodies against CD4, CD27, and CD45R0 and combined them in Photoshop (Adobe Systems) to unambiguously label CM cells in Peyer patches by using fixed tissue available for 8 of the patients (figure 1). Cells that were CD4+, CD27+, and CD45R0+ were manually counted.
To test for differences in cell count between HIV+ and HIV− subjects, the 2-sample t test with equal variance was used. Hotteling’s T2 test was used to test for changes in all 4 compartments investigated simultaneously, and confidence regions based on this test indicated changes in only 1 compartment. Regression and correlation analysis were used to examine the relationship between peripheral blood and other compartments and to test for associations between baseline levels in blood and changes in the other compartments. Because these methods rely on normality assumptions, standard diagnostics were employed to assess the normality assumption.
To determine the size of the central memory cell population in GALT, 2 assays were available (flow analysis combined with quantitative image analysis and direct counting). Because both assays yielded missing data for some patients at certain time points, the expectation-maximization (EM) algorithm was used to find the maximum likelihood estimates of the model parameters averaging over the missing data . For each patient, there were a maximum of 4 measurements (i.e., the day 0 and month 6 measurements for both assays), so the linear model presupposed that the 4 vector of measurements for each patient had a stage-specific mean (there were 3 stages: acute-early, pre-symptomatic, and AIDS) and a common covariance matrix. The EM algorithm converged to 4 decimal places within 50 iterations. The parametric bootstrap (assuming the errors in the linear model were normally distributed) was used to compute the P value for the test of no difference over time for the measurements obtained from the direct counting method. The change observed using the flow cytometry method was also statistically significant. The role of the direct counting method was to improve inference for the second method, because the 2 assays are highly correlated (r = 0.71), and different samples have different patterns of missing data for the 2 assays.
We recruited 46 individuals, 35 of whom were HIV+and 11 of whom were HIV−. All were sampled at baseline, and 19 of the HIV+ persons were sampled again after 6 months (of those, 15 received ART and 4 did not). The 4 who did not receive ART were not included in the analysis of CD4+ cell populations at baseline, but were reserved to act as a control group for those who had had tissue biopsies performed at baseline and again after 6 months of ART (described below). Among the 31 HIV+ individuals who contributed to the analysis of CD4+ cell populations before ART, the majority were in the chronic stage of disease (n = 18); 8 were in the acute-early stage and 5 had AIDS. Of those sampled in the acute-early stage, 3 underwent tissue biopsy during the seroconversion period (at which time they were HIV antibody−, HIV RNA+, and symptomatic), and 8 underwent biopsy during the periseroconversion period (within 4 – 6 months after a serologically proven seroconversion). The median age for the acute-early group was 39 years (range, 26 –59 years). The median CD4+ cell count in peripheral blood for the HIV+ individuals was 368 cells/mm3 (range, 42–939 cells/mm3), and the median count for the HIV− individuals was 820 cells/mm3 (range, 290 –1351 cells/mm3). The median plasma HIV RNA level was 32,047 copies/mL (range, 129 –500,000 copies/mL). Thirty HIV+ and 10 HIV− individuals were male; 23 (72%) of the HIV+and 11 (100%) of the HIV+ individuals were white and the remaining individuals were African American. The mean ages for HIV+ and HIV− individuals was 39 years and 37 years, respectively.
We compared the extent to which the total and subset populations of CD4+ cells were affected by HIV infection in peripheral blood mononuclear cells, inguinal LNs, and gut tissues, and we measured immune reconstitution (i.e., increase in CD4+ cell count) in the population that received ART. We documented significant depletion of CD4+ cells in each compartment in HIV+ individuals, compared with HIV− individuals (figure 2); specifically, we observed a 0.50-fold reduction in CD4+ cells in the peripheral blood (P < .001), a 0.59-fold reduction in the LN tissue (P = .001), a 0.72-fold reduction in the Peyer patches (P = .02), and a 0.66-fold reduction in the lamina propria (P = .008).
Next, we determined the extent of reconstitution of CD4+ cells in each compartment in individuals who completed 6 months of ART. The median increase in CD4+ cells in peripheral blood was 97 cells/mm3 (interquartile range [IQR], 45–196 CD4+ cells/mm3; P = .001); 18 patients (72%) achieved an increase of ≥50 CD4+ cells/mm3, and 12 (48%) achieved an increase of ≥100 CD4+ cells/mm3. In LN tissue, the median increase in area occupied by CD4+ cells was 5.56% (IQR, 2.75%–6.65%; P = .094). There was no significant increase in Peyer patches (median increase, −3.57% [IQR, −6.45% to 9.27%]; P = .68) or in the lamina propria (median increase, −0.54% [IQR, −1.81% to −0.065]; P = .26). Thus, in the 15 patients who were treated for 6 months with ART, we documented no significant increase in the total CD4+ population in lymphatic tissue.
We were interested to know whether significant reconstitution of the CD4+ cell population in lymphoid compartments was dependent on the timing of ART initiation and compared compartmental changes between patients grouped by stage of disease at baseline. If ART was initiated in the acute-early stage of disease, the mean increase in peripheral blood CD4+ cell count was 388 cells/mm3; if ART was initiated in the chronic stage, the mean increase was 176 cells/mm3 (P > .05). In LN tissue, the corresponding increases in area were 12.50% for those whose ART was initiated in the acute-early stage and 13.65% for those who had ART initiated in the chronic stage (P > .05). There was no significant increase in CD4+ cells in either GALT compartment.
A limitation of this analysis was the relatively small number of patients in each of the disease stage groups. We therefore reasoned that baseline (nadir) peripheral blood CD4+ cell count could be used as a surrogate for infection duration [19, 20] and modeled the effect of baseline peripheral blood CD4+ cell count on changes due to treatment. These models were fit to data that included the 4 HIV+ patients who did not receive ART and used samples obtained at baseline and 6 months to avoid regression to the mean. In the linear model that regressed changes in LN tissue and baseline peripheral blood CD4+ cell count for the group that received ART, we found that the higher the peripheral blood CD4+ cell count at the time ART is started, the greater the recovery of CD4+ cells in LN tissue (P = .020) We had insufficient tissue from the 4 HIV+ control patients to assess changes in GALT by this method.
In figure 3 we compare the size of the naive, CM, and EM CD4+ cell population in each compartment for HIV+ and HIV− individuals, as well as the extent to which each cell population recovers during ART for the HIV+ individuals. There were increases in the naive CD4+ cell population in peripheral blood (P = .006) and in LN tissue (P = .015), but in both compartments naive CD4+ cell populations were smaller than those in HIV− individuals before treatment (P = .001 for both compartments) and were still significantly smaller after 6 months of ART (P = .001 for both compartments).
The population of EM CD4+ cells in the lamina propria (the primary population of CD4+ cells in this compartment) was significantly greater in HIV− persons (P = .008), and in HIV+ persons there was no recovery of this population during treatment (P = .24). Moreover, in a subset of 4 HIV+ individuals who had an additional biopsy 36 months after starting ART, the mean percentage of the lamina propria that was occupied by CD4+ cells was 4.8%, similar to the area measured at baseline and still <50% of that observed in healthy control subjects. Thus, even with persistent suppression of viral replication and an increase in CD4+ cells in the peripheral blood, we found no evidence for reconstitution of this population in the lamina propria.
The size of the naive CD4+ cell population in Peyer patches was similar for HIV− and HIV+ individuals, and for HIV+ individuals both before and after ART (P = .45 and P = .14, respectively). However, when HIV− and HIV+ persons were compared, there were significant differences at baseline in the CM CD4+ cell population in peripheral blood, LN tissue, and Peyer patches (P < .001, P = .02, and P = .001, respectively), and the only significant increase after ART occurred in the peripheral blood (P = .013). Importantly, we did find a significant increase in the CM CD4+ cell population in Peyer patches when ART was initiated in the acute-early stage of disease (P < .01) (figure 4).
A central feature of HIV pathogenesis is the profound loss of CD4+ cells in GALT without evidence for immune reconstitution during receipt of ART, for reasons that have yet to be determined. We have previously described a process of pathologic fibrosis in secondary LNs, the magnitude of which is inversely correlated with the size of the naive CD4+ cell population in the paracortical T cell zone [13, 15]. Because fibrosis is correlated with CD4+T cell depletion and the magnitude of reconstitution in peripheral blood, we hypothesized that fibrosis occurred more rapidly and extensively in GALT, thereby accounting for the greater depletion and more limited reconstitution of CD4+ cells in GALT, compared with LN tissue. We stained GALT samples from HIV+ persons with trichrome, and the magnitude of fibrosis in Peyer patches and the lamina propria was determined by using quantitative image analysis. The mean percentage of the GALT area that stained positive for collagen in samples from HIV+ and HIV− individuals was 15.5% and 4.4%, respectively (P = .002) (figure 5). There was no significant difference in GALT specimens when the baseline sample and the sample obtained after 6 months of ART were compared. To determine whether fibrosis occurred more rapidly in GALT than in LN tissue, we compared levels of collagen in the T cell zone of LNs and Peyer patches for patients sampled during the acute-early stage of infection. We found a significantly greater level of collagen in Peyer patches, compared with LN tissue (figure 6) (P = .03); the extent of collagen deposition in the T cell zone of Peyer patches was negatively correlated with the size of the total CD4+ cell population (r = −0.60; P = .004) and naive CD4+ cell population (r = −0.75; P = .052) in that compartment (figure 7).
In this prospective, longitudinal analysis of the impact of HIV infection and ART therapy on CD4+ T cell populations in both the inductive and effector sites of GALT, peripheral blood, and LNs, we provide quantitative measures of the extent of depletion and reconstitution in each compartment for the same individual. Importantly, we show that in the earliest stages of infection, CD4+ cell populations have been depleted in LN tissue, Peyer patches, and the lamina propria to an extent not evident in peripheral blood, such that when the patient presents with symptoms of acute HIV sero-conversion, the population of CD4+ cells in GALT and secondary LN tissue is already reduced by approximately 50%, compared with that in Peyer patches, LN tissue, or peripheral blood compartments. This result is in agreement with those of other studies that examined the lamina propria in contrast with the Peyer patches, LN, or peripheral blood compartments [6, 21, 22].
We also quantified the differential impact of ART on CD4+ cell reconstitution in peripheral blood, compared with the other compartments. After 6 months of ART, the population of peripheral blood CD4+ cells had increased 41%, whereas the CD4+ cells in LN tissue increased only 15%, those in Peyer patches decreased by 22%, and those in the lamina propria decreased by 10%. We show, by combining flow cytometric analysis of cell suspensions with immunohistochemistry and quantitative image analysis, that this decrease occurs in EM cells, the major CD4+ cell subset population in the lamina propria. The reasons why the EM population in the lamina propria decreased in size are likely complex, but ongoing responses to microbes in the gut , frequent reactivation of a latent mucosal infections such as HSV2 [24–26], and/or a possible response to continuing low level of HIV replication despite ART could all contribute to the continuing drain on this population relative to naive and CM cells, which are also compromised, as discussed below.
The major new finding in this study is the early and extensive collagen deposition in GALT, to a greater extent than that which occurs in secondary lymphatic tissues, which we show to be correlated with greater depletion and limited reconstitution of CD4+ cells in the gut. We have previously shown in peripheral lymphatic tissues that fibrosis in the T cell zone is associated with reduced numbers of CD4+ cells and the potential for reconstituting naive CD4+ cells in particular, by a mechanism we refer to as the damaged niche, in which fibrosis disrupts the lymphatic tissue architecture and compromises the ability of lymphatic tissues to support T cell survival and proliferation [12–15, 27]. Of interest, levels of collagen in GALT do not predict immune reconstitution in peripheral blood. We think that the rapid and more extensive collagen deposition in the gut may be an important mechanism that contributes to the disproportionate levels of early and sustained depletion of CD4+ cells in GALT.
Under this model, early fibrotic damage to the Peyer patches contributes to the depletion and limits the reconstitution of the naive and CM CD4+T cell populations; it also contributes to the sustained depletion of EM cells in the lamina propria by compromising the source of these cells in the face of continuing drains on the population, as described above. Prior to initiation of ART, the relationship between fibrotic damage to the inductive source and EM CD4+ T cell populations in the lamina propria may be important in determining rates of progression to disease, based on studies of SIV infection in the nonhuman primate model [28 –30]; this relationship may also be important in explaining clinical benefit despite apparent small increases in reconstitution. In SIV infection of rhesus macaques, there is massive depletion of memory CD4+ cells in early infection, mainly in the lamina propria, and there is little recovery with treatment; the greatest reconstitution occurs if therapy is initiated in the early stage of infection. Picker et al.  have shown that the best predictor of disease progression in the SIV–rhesus macaque model is preservation of CM CD4+ cells, which presumably can continue to supply sufficient numbers of cells at effector sites to defend the host against pathogens in the gut and elsewhere. Similarly, less fibrotic damage and relatively greater preservation of an inductive source— here shown to include naive CD4+ cells—may better balance continuing drains on EM cells in the gut and produce clinical benefit despite apparently small increases in CD4+ cells.
We further confirm that ART initiated at later stages of HIV-1 infection does not result in gut reconstitution, but we provide new evidence in support of the conclusion that reconstitution of subpopulations of gut CD4+ cells is possible with early treatment, in agreement with some— but not most—recent studies. What we show here is that reconstitution is inductive site specific and subset specific, which may be the explanation for the discrepancy between our results and those obtained in these previous studies. Moreover, we show that early initiation of ART supports the greatest reconstitution of this population, as well as that of the peripheral blood and peripheral lymphatic tissues.
The benefits of early treatment for immune reconstitution—especially in the gut, the largest lymphoid organ of the immune system–suggests to us that current recommendations to wait for a CD4+ T cell count of 350 cells/mm3 before initiation of ART may not be optimal for the restoration of immunity to the extent that may prove to be necessary for immunosurveillance against tumors and pathogens over a relatively normal life span. Although current guidelines support the use of ART if the patient is diagnosed in the acute stage of disease, they are silent with respect to duration of therapy; some advocate stopping therapy in the chronic stage of disease. The strategy of interrupted therapy was recently tested in a large, international trial and was discontinued early as it was associated with more rapid progression of HIV infection and worse clinical outcome . Our observations suggest that earlier therapy might better preserve and restore the critical CM CD4+ cell population in GALT. We also speculate that antifibrotic drugs might have a role as adjunctive therapy in HIV-1 infection, both in limiting depletion and improving reconstitution during ART.
We acknowledge and thank the patients who participated in this study; Dr. Frank Rhame, Dr. Leslie Baken, Dr. Ronald Schut, and Dr. Peter Bornstein for referring patients to these studies; Marc Jenkins, PhD, for his advice and assistance; and Tim Leonard for assistance with manuscript preparation
Financial support: National Institutes of Health (P130-CA79458 – 01 to A.T.H., 1RO1DE12934 – 01 to T.W.S., MO1 RR00400 to A.T.H., 2UO1 AI041535 to A.T.H., RO1 AI54232– 01A2 to T.W.S., and R37 AI 28246 to T.W.S.).
Potential conflicts of interest: none reported.