Macrophages as a Source of HIV During Opportunistic Infections

See allHide authors and affiliations

Science  20 Jun 1997:
Vol. 276, Issue 5320, pp. 1857-1861
DOI: 10.1126/science.276.5320.1857


The source of increasing viremia that characterizes the latter stages of human immunodeficiency virus (HIV) disease has remained a paradox because it occurs at a time when lymphoid tissue is quantitatively and qualitatively impaired, and the patients' CD4 T lymphocytes are steadily declining. Here, macrophages, both infected and uninfected with common opportunistic pathogens of HIV disease such as Mycobacterium avium complex and Pneumocystis carinii, were identified as highly productive sources of HIV in coinfected lymph nodes. These observations indicate that tissue macrophages are not only infected with HIV, but that common pathogens of HIV disease can dramatically increase their production of virus. Thus, prevention or successful treatment of opportunistic coinfections, or both, potentially benefits the patient twofold by limiting the pathology caused by opportunistic infection and by controlling induction of HIV replication.

Opportunistic infections (OIs) and rising levels of circulating human immunodeficiency virus–type 1 (HIV-1) are hallmarks of progressing HIV-1 disease (1). After infection with HIV-1, a variably paced loss of CD4 T lymphocytes leads to immunodeficiency and increasing susceptibility to a broad range of opportunistic viral, bacterial, fungal, and protozoal pathogens. A direct correlation between OIs and HIV-1 expression in vitro is paralleled by emerging in vivo evidence (2). Infections with common opportunistic pathogens, such as herpesvirus type-1,Mycobacterium avium complex, and M. tuberculosisare accompanied by mostly reversible increases in HIV-1 viremia (2-4). Bacterial pneumonia, an increasing cause of morbidity and mortality in HIV-1 infection, is also associated with increased circulating HIV-1 (5).

The source of the increased circulating HIV-1 has remained a puzzle because progression of HIV-1 disease is also characterized by involution of lymphoid tissue. Lymph nodes characteristically display progressive loss of germinal centers, leaving little if any remaining follicular dendritic cell–associated virus and only scattered productively infected CD4+ lymphocytes. The role of the long-lived, mobile phagocyte as a source of virus during HIV-1 disease has been considered minimal, as few if any productively infected cells expressing monocyte/macrophage markers have been documented either in the circulation or in lymphoid tissue (6-8). However, CD4+ macrophages are productively infected in the central nervous system of patients with acquired immunodeficiency syndrome (AIDS) dementia complex and vacuolar myelopathy (8). Furthermore, their numbers are maintained in HIV-1 disease and, both in vitro and in vivo, monocyte/macrophages have been shown, unlike lymphocytes, to be capable of producing large amounts of both intracellular and extracellular HIV-1 without necessarily succumbing to the lethal effects of productive viral infection. To resolve the inconsistency of enhanced HIV-1 replication despite decreasing CD4+ T cells, we focused on the potential role of the macrophage.

Probing of two Pneumocystis carinii (PC)–infected lymph nodes, excised from a patient with disseminated disease (9), by HIV-1 RNA in situ hybridization (ISH) for HIV-1–specific RNA revealed marked amounts of virus. The HIV-1–positive cells in these specimens far exceeded the number of productively infected lymphocytes characteristic of typical OI-free, hyperplastic lymph nodes from HIV-1–infected patients (Fig. 1, A and B). The PC lymph node specimens were almost totally replaced by necrotizing, hemorrhagic PC infection with only small amounts of residual lymphoid tissue (Fig.1B). Whereas HIV-1 RNA expression was especially associated with collections of inflammatory cells within the OI, it was also present in apparently cell-free areas (Fig. 1B). In non-OI infected lymph nodes, the bulk of the HIV-1 signal was characteristically associated with follicular dendritic cell (FDC) processes as virus:complement–antibody complexes and was by far best observed when ISH was performed after predigestion of the tissue sections with protease, which enhances penetration of the probe (Fig. 1A) (7). After protease and ISH, FDC-entrapped HIV was seen as a linear display of grains along myriad FDC processes, and productively infected cells were distinguished by the superimposition of the signal over the cell in a “starburst”-like image. In contrast, protease pretreatment neither substantially increased the number of HIV-1 RNA–positive cells nor the amount of signal in the PC co-infected lymph nodes, and did not reveal areas of residual FDC-associated HIV-1 (Fig. 1B). Rather, HIV-1 signal was demonstrated to be very intense both with or without protease digestion and to be associated with mononuclear and multinucleated cells.

Figure 1

HIV RNA in situ hybridization on OI-infected and noninfected lymph nodes. Lymphoid tissues from HIV-1–infected individuals, obtained surgically at the George Washington University Medical Center, Washington, D.C., were fixed in 10% neutral buffered formalin, embedded in paraffin, sectioned, and processed for in situ hybridization. For identification of HIV-1, protease-treated and untreated tissue sections (6 μm) (7) were hybridized to35S- or 33P-labeled antisense RNA representing 9 kb of the HIV-1 genome (IIIB) to detect the presence of HIV-1 viral RNA. Both sense (control) and antisense probes were used. After hybridization, the slides were dipped in NTB 2 Kodak emulsion, exposed over dessicant for 4 days at 4°C, and developed in D-19 at 15°C for 4 min. (A) In an HIV-positive, OI-negative, hyperplastic lymph node, protease pretreatment was necessary to reveal the bulk of the HIV-1 RNA (black grains) which was associated with the cytoplasmic processes of FDC in germinal centers (arrowheads). Scattered HIV-positive lymphocytes were seen around and within germinal centers (arrows). The linear FDC-associated signal is difficult to discern with brightfield optics below this magnification (original magnification, ×400). (B and C) In marked contrast, the intensity of the HIV-1 signal was so pronounced in the PC- and MAC-infected node specimens that it was readily visible at original magnifications of ×60 (B) and ×40 (C), respectively. The nodal architecture was disrupted, lymphocytes were depleted, and dense and diffuse collections of HIV-1 signal were prominent (black grains). The ISH grains were especially concentrated over cells but were also seen in adjacent acellular areas. (D) In the MAC-infected lymph node, the HIV signal coincided with mononuclear and especially, multinucleated phagocytes rich in internalized acid-fast (red) MAC bacilli. The tissues were first hybridized and developed, and acid-fast Kinyoun's staining was performed on the emulsion-coated slides (original magnification, ×160).

Mycobacterium avium complex (MAC)–infected lymph nodes excised from a patient with widely disseminated disease (10) and analyzed by identical ISH methods demonstrated that the marked HIV-1 expression was not unique to PC infection and in fact, the HIV-1 burden was even more pronounced in the MAC-infected specimens (Fig. 1, C and D). The HIV-1 signal varied considerably among cells, from severalfold above the background level to an amount of grains that obscured the cells, and which was too numerous to resolve for the majority of cells (Fig. 2, A and B).

Figure 2

Macrophages as the source of the HIV-1 RNA. (A) Collection of HIV-1–positive macrophages (cells obscured by grains; arrows indicate two of these) surrounded by erythrocytes in a PC-infected node. Original magnifications for (A) through (D) are ×640. (B) Typical starburst images of HIV-1–expressing lymphocytes (arrows) in the PC-infected node. (C) HIV p24–positive (red stain) MAC-infected multinucleated Langhans'-type giant cells. The clear needle-like intracytoplasmic structures (arrows) represent negative-stained images of bacilli. For immunohistochemistry, deparaffinized sections were washed in tris-buffered saline (pH 7.4) and preincubated for 10 min with blocking buffer (0.5% albumin w/v in tris-buffered saline or nonimmune serum) to inhibit nonspecific protein binding, followed by incubation with the primary antibody. The secondary antibodies were counterstained with New Fuchsin. (D) HAM 56 (a macrophage-associated specific antibody) costained (pink) the ISH-positive (black grains) Langhans'-type multinucleated macrophages (arrow). (E) Higher magnification of Fig. 1D showing colocalization of HIV-1 signal in mononuclear (arrows) and multinucleated (arrowhead) macrophages containing acid-fast (red) bacilli (Kinyoun's stain) (original magnification, ×400).

Immunohistochemical staining with HIV-1 p24-specific antibody paralleled the ISH results, revealing p24-positive mononuclear and multinucleated cells in the MAC- (Fig. 2C) and PC-infected specimens. In both cases, the cells producing the bulk of HIV-1 RNA were identified as mononuclear and Langhans'-type multinucleated macrophages by morphology and ISH combined with immunohistochemical staining with macrophage-associated antibodies (HAM56) (Fig. 2D) and CD68 (11). By sequentially performing HIV-1 ISH and staining sections with Kinyoun's stain (acid- fast), which specifically stains rod-shaped mycobacteria red, we documented that HIV-1 production occurs in both mononuclear and multinucleated giant cells that have internalized and are coinfected by MAC (Figs. 1D and 2E).

Transmission electron microscopic (TEM) analysis of PC-infected lymphoid tissue revealed cysts interspersed with inflammatory cells and remnants of PC organisms within phagocytic cells. Numerous developing, and especially, complete mature virions were readily identified by TEM (Fig. 3). In contrast, earlier studies (8), limited to OI-negative nodes, revealed relatively few mature virions, and those were invariably associated with FDC processes. In these coinfected specimens, budding virions arose from the complicated plasma membranes of mononuclear and multinucleated macrophages, whereas free mature virions were concentrated around these cells, within cytoplasmic vacuoles, and dispersed in acellular areas. Although virions were found in the vicinity of lymphocytes, none were observed budding from their surfaces. Productive infection with HIV-1 did not appear to be associated with death of macrophages.

Figure 3

HIV-1 identification in PC-infected lymph node by TEM. Glutaraldehyde-fixed surgical tissue was postfixed in OsO4, dehydrated through graded ethanol and propylene oxide, embedded in Spurr's epoxy, and thick- and thin-sectioned. Thin sections were placed on copper grids, stained with uranyl acetate and lead citrate, and viewed in a Zeiss EM10 microscope (LEO Electron Microscopy, Thornwood, New York). (A) In this field, at least 20 mature HIV particles (several indicated by arrows) are associated with only a small area of a macrophage plasma membrane. The cell abuts a PC cyst (P) (original magnification, ×25,000). (B) Two spike-covered virions bud from a single macrophage process (arrows) (original magnification, ×130,000). (C) Five mature virions (arrows) are intermixed with membranous vesicles shed from PC organisms (original magnification, ×50,000). (D) Several mature virions are within each of two Golgi vacuoles (arrows) in the cytoplasm of a macrophage (original magnification, ×66,000).

To ascertain whether this phenomenon is a general characteristic of lymphoid tissues infected with opportunistic pathogens, we compared HIV-1 production in lymph nodes with OIs from 11 HIV-1–positive subjects to a large number of HIV-1–positive, but OI-free lymph nodes. All coinfected biopsy specimens, with a single exception, had substantially greater numbers of HIV-1–expressing cells per unit area relevant to parallel analyses of non-OI lymph nodes (Table1). Compared to non-OI lymph nodes, which possess minimal numbers of HIV-1–infected cells of monocyte/macrophage lineage, these cells become a major, but not exclusive, source of HIV-1 in the OI-infected tissues. Levels ranging from a fivefold to more than a hundredfold increase in the number of productively infected cells were consistent with OI-induced HIV-1 expression. The differences in actual HIV-1 RNA copies per cell present in the OI- and non-OI–infected tissues are actually even more notable because the density of signal in the OI-infected specimens typically precludes resolution of individual grains.

Table 1

HIV-1–expressing cells in lymph nodes with and without opportunistic infections. Excisional biopsies were obtained with consent from 10 males, aged 28 to 50, and one 33-year-old female with AIDS-defining opportunistic infections (OIs) and from 67 HIV-1 seropositive subjects without evidence of OIs. Tissue sections were processed without and with (shown) protease and in situ hybridization using HIV-1 sense and antisense probes performed as described (7). After in situ hybridization, the entire tissue sections were imaged with a CCD camera and 60-mm macro lens. Using NIH Image and a Macintosh 8100/80AV computer, the areas of the sections were determined by planimetry based on a 10-mm bar standard. Subsequently, HIV-1 expressing mononuclear cells (not FDC associated HIV) in the entire tissue sections were counted in a darkfield microscope at a numerical aperture of 0.32. Positive cells were considered to have more than 20 silver grains per 200 μm2 and were expressed as the number of productively infected cells per 10 mm2 of lymph node tissue. Most of the cells in the OI-infected tissues had far more than this number of grains directly over the cell, making it impossible to quantitate their numbers. Cell numbers in highly infected tissues represent underestimates due to giant cells, the presence of multiple cell aggregation, and intense positivity of signal. MAC, Mycobacterium avium complex; MK, Mycobacterium kansasii; PC,Pneumocystis carinii; MTB, Mycobacterium tuberculosis; Histo, Histoplasma capsulatum.

View this table:

As judged by multiple parameters, the macrophage, whether infected (for example, MAC, Mycobacterium kansasii, Histoplasma capsulatum, M. tuberculosis) or uninfected (for example, PC) by opportunistic pathogens, can be responsible for production of large amounts of HIV-1 at the very time CD4 lymphocytes are minimal. Furthermore, the extensive viral replication by tissue macrophages provides compelling evidence that OIs can markedly change HIV-1 expression by this population of target cells. OIs apparently have the capacity to establish foci of HIV-1 production by recruitment of HIV-1–infected and infectible macrophages and stimulating their viral expression. The mechanism of up-regulation of HIV-1 RNA expression by OIs is still to be resolved (12, 13).

At all stages of HIV-1 disease (from hyperplasia through involution and lymphocyte depletion) lymphoid tissues free of OIs typically display few if any HIV-1–expressing macrophages, as detected by dual ISH and immunohistochemistry (Table 1) (7, 13). Nonetheless, our observations suggest that a large reservoir of such macrophages exists in the host, and that under conditions characteristic of progressive HIV-1 disease, these cells may become a major source of virus. Although the incidence and severity of OIs accelerate rapidly as the peripheral CD4 lymphocyte count falls below 200 cells per cubic millimeter (14), significant infections—such as M. tuberculosis, candida, herpes simplex virus, and herpes zoster—can occur at relatively high CD4 lymphocyte counts (1) and may influence HIV-1 replication in both lymphocyte and macrophage populations. Furthermore, the effect of nonopportunistic infections, such as sexually transmitted diseases, on HIV-1 viremia must also be taken into account (15). The influence of the OI-macrophage phenomenon on antiviral induction and maintenance therapy thus becomes a key issue.

A major impact on morbidity and mortality in HIV-1 disease has come from successful prevention and treatment of a broad range of OIs, most notably Pneumocystis carinii pneumonia, MAC, and cytomegalovirus. Based on our observations, controlling HIV-1 production by this means may be as significant as elimination of the direct pathogenicity of the OIs (13, 16). Our findings on viral production by macrophages, together with the newly identified role of the macrophage CCR5 coreceptor on initial HIV-1 selection and infection (17), suggest the need for a reassessment of the overall contributions of the macrophage in HIV-1 disease and as a therapeutic target.

  • * To whom correspondence should be addressed. E-mail: jmo{at}


View Abstract

Navigate This Article