Sexual Transmission and Propagation of SIV and HIV in Resting and Activated CD4+ T Cells

See allHide authors and affiliations

Science  12 Nov 1999:
Vol. 286, Issue 5443, pp. 1353-1357
DOI: 10.1126/science.286.5443.1353

This article has a correction. Please see:


In sexual transmission of simian immunodeficiency virus, and early and later stages of human immunodeficiency virus–type 1 (HIV-1) infection, both viruses were found to replicate predominantly in CD4+ T cells at the portal of entry and in lymphoid tissues. Infection was propagated not only in activated and proliferating T cells but also, surprisingly, in resting T cells. The infected proliferating cells correspond to the short-lived population that produces the bulk of HIV-1. Most of the HIV-1–infected resting T cells persisted after antiretroviral therapy. Latently and chronically infected cells that may be derived from this population pose challenges to eradicating infection and developing an effective vaccine.

HIV-1 is usually transmitted by heterosexual contact (1) and, in prevailing hypothetical reconstructions of the initial events in infection, is believed to first infect dendritic cells (DCs) or macrophages (Mφs) (2–11). At a later, undefined time these cell types pass infection to CD4+ T cells in postulated cell-cell interactions that activate the T cells so that they become susceptible to and can support virus replication (3, 11). We experimentally examined this reconstruction of heterosexual transmission in an animal model. We inoculated 14 rhesus monkeys (Macaca mulatta) intravaginally with uncloned simian immunodeficiency virus (SIV) monoclonal antibody mac251, a dual tropic strain that replicates in cultured Mφs or T cell lines (12,13). To identify the types of cells in which SIV first replicates we euthanized animals 1, 3, 7, and 12 days after inoculation and looked in tissue sections by in situ hybridization (ISH) for cells with detectable SIV RNA (14, 15). In screening ≥25 sections of tissues from the site of inoculation and distal sites in the lymphatic and other organ systems, we first detected SIV RNA in a small number of cells 3 days after inoculation only in the endocervical region. Between days 3 and 7 there was a large increase in viral (v) RNA+cells in the endocervical region (16) followed by an additional expansion in the infected population at that site by day 12. At this time vRNA+ cells were also found throughout the lymphatic tissues (LTs) and many other organs to which infection had been disseminated, with the exception of the central nervous system (Table 1).

Table 1

Tissues with viral RNA+. Cells after intravaginal inoculation of SIV. NT, not tested; −, not detected; +, ≥6 × 10−3 vRNA+ cells per square millimeter of tissue (>2 copies per cell); ++, ≥4.5 × 10−1 vRNA+ cells per square millimeter of tissue; ++++, ≥3.5 vRNA+ cells per square millimeter of tissue.

View this table:

To unequivocally identify the types of cells in which SIV was replicating, we combined ISH with immunohistochemical (IHC) staining of Mφs, DCs, and T cells (17, 18). Nearly all the vRNA+ cells 3 days after inoculation were intraepithelial lymphocytes or T cells in the lamina propria of the endocervix (Fig. 1, panel CD3, and Table 2) where they may have been exposed to virus through breaks in the single layer of epithelium lining the cervix. We characterized the host cell range for productive infection more extensively at days 7 and 12 (when there were sufficient numbers of vRNA+ cells to accurately determine the frequency of the types of vRNA+ cells) and documented infection in CD4+ T cells (Fig. 1, panel CD4) as well as Mφs and DCs. However, close to 90% (Table 2) of the vRNA+ cells at the portal of entry, LTs, and other sites to which infection had been disseminated were CD4+ T cells (Table 2). Because the apparent preponderance of vRNA+cells in the CD4+ T cell population might be the result of underestimating vRNA+ Mφs or DCs with only a few copies of vRNA, we lengthened the autoradiographic exposure to 7 days to be able to detect cells with two copies of vRNA per cell. There was no increase in the frequency of vRNA+ cells and thus no evidence to support models of transmission (8–11) in which virus replicates at low levels in these other cell types, which then transmit infection to CD4+ T cells. The preponderance of productive infection in CD4+ T cells shortly after intravaginal transmission and similar recent observations for the intrarectal, oral, and intravenous routes collectively point to the importance of CD4+ T cells in the propagation of infection after mucosal or parenteral exposure (11, 19).

Figure 1

SIV RNA and p27 in CD3+and CD4+ T cells at the portal of entry and in T cells in LNs stained or not stained by antibodies to markers of cellular activation and proliferation. Cells with SIV RNA detectable by ISH were typed IHC with antibodies that stain CD3+ and CD4+ T cells. Arrows in panel CD3 point to a brown-stained intraepithelial lymphocyte at the base of epithelium lining the lumen (L) of an endocervical gland and to another lymphocyte in the lamina propria with overlying collections of black silver grains indicating vRNA. In panel CD4 arrows point to CD4+ vRNA+cells in lymphocytes adjacent to the lumen of a gland. In panels HLA-DR and Ki67, ISH to detect vRNA was combined with antibodies that stain activated or proliferating T cells. Longer arrows point to double-positive cells that are vRNA+ and HLA-DR+ or Ki67+. Short arrows point to HLD-DR or Ki67 vRNA+ cells. In panels Cyclin D3, cdk6, and CD25, longer arrows point to vRNA+ cells that are also stained by antibodies to these markers of proliferation and activation. Short arrows point to vRNA+ cells that lack these markers. In panel SIVp27, the major GAG core protein of SIV, there are double-positive viral p27+ vRNA+ cells with lower (fewer silver grains) and higher (larger numbers of silver grains) levels of vRNA.

Table 2

Percent SIV RNA+ cell types within endocervix (CVX) and axillary LNs. SIV RNA+ cell types were determined by combined IHC–ISH assays of subjacent sections of endocervix in which vRNA+ cells were first identified at day 3. Percentages were determined at days 7 and 12 after intravaginal inoculation of SIV by scoring ≥100 T cells stained by antibodies to CD3 and CD4+ T cells stained by antibodies to CD4, Mφs stained by anti-CD68, and DCs stained by anti-S100. Because the percentages were determined independently in subjacent sections, the total amount of cell RNA+ cells could be >100%. At day 3 there were sufficient numbers of vRNA+ cells to accurately determine only a percentage for IHC staining with anti-CD3. The criteria for scoring vRNA+ cells and the limits of sensitivity are described in (16). The longer autoradiographic exposures did not reveal larger numbers of CD68+ or S100+vRNA+ cells.

View this table:

We expected the vRNA+ T cells to be activated and in cycle because in vitro T cell activation and lentiviral replication are generally thought to be inseparable (20). We set out to confirm that virus replication in vivo is also confined to activated and proliferating T cells by combining ISH to detect SIV RNA with IHC staining with antibodies to markers of T cell activation (HLA-DR) and proliferation (Ki67). As expected, we detected SIV RNA in cells that were activated and proliferating but, surprisingly, found also that many SIV RNA+ cells were HLA-DR and Ki67. Although both populations of SIV RNA+HLA-DR+ and Ki67+ and HLA-DR and Ki67 T cells expanded exponentially at the portal of entry and distal sites in the LTs, the predominant expansion was in the HLA DR and Ki67 T cell population (Fig. 1, panels HLA-DR and Ki67, and Tables 3 and4). The major difference between the populations that were HLA-DR+ and Ki67+ or HLA-DR and Ki67 was in the levels of viral gene expression. The mean number of silver grains determined by quantitative image analysis (QIA) (15, 21), which is proportional to the intracellular concentration of SIV RNA, was about four to seven times higher in the activated and proliferating HLA-DR+ and Ki67+ population (Tables 3 and 4). Despite the different levels of vRNA the SIV major core protein (Fig. 1, panel SIVp27) was detectable in both populations as further evidence of productive infection, even at low levels of vRNA.

Table 3

SIV RNA levels in HLA-DR+ or HLA-DR and Ki67+ or Ki67 T cells. Infected cells that were stained or not stained immunohistochemically with antibodies to the indicated markers were easily identified visually in sections of the indicated tissues 12 days after intravaginal inoculation by overlying clusters of ≥10 silver grains over background, satisfying the probability criterion of >0.995 that the cell was infected. Percentages of vRNA+cells by this criterion that were not stained by antibodies to the indicated markers were determined from 100 cells. Grain counts per cell were determined for 100 stained or unstained cells.

View this table:
Table 4

Expansion of two infected populations of T cells after sexual mucosal transmission of SIV. Cells that met criteria for infection defined in Table 1 that were stained or not stained immunohistochemically by antibodies for the indicated markers were first detectable 3 days after intravaginal inoculation after extensive search of 20 sections of endocervix with an area of 1660 mm2. For comparison, infected cell counts at later times were normalized to this area. Semilogarithmic plots of the frequency of vRNA+ cells that are HLA-DR+ or HLA-DR and Ki67+ or Ki67versus days of infection are linear; thus, both populations are expanding exponentially.

View this table:

The unexpected observation that SIV could replicate, albeit at low levels, in HLA-DR and Ki67T cells led us to seek evidence of a similar population in HIV-1 infection. We subsequently found and present evidence that, in HIV-1 infection from the earliest stages forward, most of the vRNA+ cells are T cells; HIV-1 replicates in these cells whether or not they are demonstrably activated and proliferating; and the infected cells that are not proliferating have lower levels of viral gene expression and are longer lived.

We analyzed lymph node (LN) biopsies obtained from four individuals 2 to 15 days (mean 8 days) from the onset of symptoms of the acute retroviral syndrome, which occurs after an average incubation period of 12 days (22); five LN biopsies from individuals ≥63 days to 242 days postseroconversion; and 10 tonsillar biopsies from a previously described cohort (23) of individuals in the late stages of HIV-1 infection (<200 CD4+ T cells per cubic millimeter of peripheral blood) (Table 5). In the acute and early stages of infection about 90% of the vRNA+ cells are CD3+T cells. This was also the case in the late stages of HIV-1 infection and 2 days after HAART. In the acute and early stages of HIV-1 infection, like in SIV infection, about half the vRNA+ cells were HLA-DR and Ki67, with a mean concentration of intracellular vRNA about two to five times lower than the activated population (Table 5). By the late stages of infection >75% of the vRNA+ cells were HLA-DR+ and Ki67+. The high percentage of infection of activated T cells in the late stages is consistent with the increased activation state characteristic of late infection (24). The increase in the percentage of T cells scored as activated vRNA+ T cells over the course of infection also indicates that the combined IHC staining technique is sufficiently sensitive to monitor changes in the activation status of infected cells.

Table 5

HIV-1 RNA in HLA-DR+ or HLA-DR and Ki67+ or Ki67 T cells at early and late stages of infection and post-HAART LNs and tonsil biopsies were obtained from HIV-1–infected individuals in the acute, early and late stages of infection and post-HAART (22, 23,25). With the specific activity and complexity of the HIV probes for autoradiographic exposures of 24 hours, infected cells with ≥10 copies of HIV-1 RNA were detected. The frequency of T cells with HIV-1 RNA was determined by counting vRNA+cells in sections whose weight was calculated from the area, thickness, and density (16). This frequency × percent CD3+ cells gives the T cell frequency. Reduction after HAART was calculated from changes from baseline Ki67+ or Ki67 population size in late-stage patients 2 days after initiating HAART. The percentage of CD3+ T cells that were HLA-DR± or Ki67± and grain counts in these populations were determined as described (16).

View this table:

To assess the dynamics of infection in Ki67+ or Ki67 populations, we determined changes in the relative population sizes post-HAART. Before treatment most of the vRNA+ cells are Ki67+ but after treatment most of the vRNA+ T cells are Ki67 (Fig. 2 and Table 5) because of the disproportionate life spans of infected T cells in the two populations. Two days after HAART, the Ki67+ vRNA+population shrinks by a factor of 9, whereas there is only a 1.5-fold change in the Ki67 population. We had previously shown that the decay of vRNA+ cells in tonsillar tissues in the cohort designated late-stage infection and day 2 post-HAART in Table 5was multiphasic (23). The initial decline occurred in a population of cells with high concentrations of vRNA that produce the bulk of HIV-1 and turn over with a t 1/2 of about 1 day. There was a slower decline in cells with smaller amounts of vRNA, and this chronically infected population persisted for 6 to 30 months after initiation of HAART (23, 25). The dramatic decrease in Ki67+ vRNA+ T cells immediately after HAART is consistent with the conclusion that the short-lived population of productively infected cells is composed of infected Ki67+ T cells. We had found (23) that the longer lived population was not IHC stained by antibodies that identify Mφs and now conclude, based on the relatively stable population size after HAART, that this population consists mainly of Ki67-infected T cells.

Figure 2

Population size changes post-HAART in Ki67+ or Ki67 vRNA+ cells. Data from Table 5 are plotted at baseline and 2 days after initiation of HAART.

To further narrow down the position in the cell cycle of the infected T cells in the Ki67 population, we used antibodies to earlier markers of stages of the cycle (Table 6). Quiescent cells in G0that reenter the cell cycle resynthesize D cyclins and cyclin-dependent kinases (cdks) in early stages of the cycle (26). We found in the populations of vRNA+ cells that there were also cyclin D3and cdk6 cells as well as double-labeled vRNA+ cyclin D3+ or cdk6+ cells (Fig. 1, panels Cyclin D3 and cdk6, and Table 6). Moreover, there were vRNA+ PCNA cells (Table 6) under conditions in which cells in any stage of the cycle and even some cells in G0 that recently exited the cycle will be labeled (27). We also characterized the activation state more fully and showed that there were vRNA+ cells that either had or lacked early and other activation markers such as CD25, CD71, CD30, CD38, and CD134 (Table 6). Panel CD25 in Fig. 1 has examples of vRNA+ cells that were CD25+ or CD25.

Table 6

Early markers of T cell activation and proliferation in SIV or HIV RNA+ cells. Percentage of viral RNA+ cells that were not stained by the indicated antibodies was determined as described in the legend to Table 2A. NT, not tested.

View this table:

By all these criteria, the infected T cells would be categorized as resting T cells. However, in vivo, the infected cells must be in a different state than cultured T cells where, under most (20) conditions, resting T cells cannot be productively infected. One possible difference in vivo is that cytokines provide a semipermissive milieu in resting T cells for viral replication, just as G0cells can be transduced by HIV-1–based vectors when cultivated with cytokines (28). Another possibility is that the state of the infected T cells in vivo differs because survival of resting T cells in vivo depends on contact with major histocompatibility complex molecules (29). This low level of stimulation does not commit the cell to full activation and progression through the cell cycle but alters the state of the cell sufficiently to support viral replication at low levels. This altered state in vivo may be one explanation of why naı̈ve CD4+ T cells in a constitutive resting state are infectable (30).

Infection of activated and resting populations of CD4+ T cells could play critical and different roles in infection. The ability to infect and replicate in resting or perhaps minimally activated T cells could be particularly advantageous at transmission and in the early stages of infection. At these times, local propagation of infection must be sustained in an environment where it is less likely that the next host cell in the chain of transmission will be an activated T cell in the vicinity of the productively infected cell. The low level of viral gene expression could also provide a mechanism common to other lentivirus infections (31) for cells to elude host defenses and in this way favor establishment of a persistent infection. Persistent infection in the early stages of infection stands as an obstacle to prevention of infection by a vaccine. Infection of an activated T cell, on the other hand, is important because higher levels of virus production disseminate infection systemically and provide access to large numbers of target cells for continued replication. This creates conditions in LT in which activated cells are continuously generated to maintain the high levels of replication, sustain infection, and increase the chances of transmitting infection to another individual. The high levels of virus production and storage in LT cells also directly or indirectly compromise control of infection by the immune system. HAART can disrupt this vicious cycle, with documented benefits to the immune system (32), but leaves long-lived populations of latently (33) and chronically infected cells (23, 25) as obstacles to the long-term goal of eradication of virus and vaccine development.


View Abstract

Navigate This Article