It has been thought that HIV particles assemble and bud at the plasma membrane (PM) in T lymphocytes and HeLa cells, but at the endosomes in macrophages, suggesting that such endosomal targeting may be essential for HIV budding in macrophages 123456. However, recent studies using the inhibitors of the endocytic pathway and membrane-impermeant dyes have revealed that the PM is the primary site for HIV assembly and particle budding even in macrophages and that particles accumulate at the endosomes through endocytosis 789. Nevertheless, these studies are based on observations in acutely infected cells and little is known about HIV budding concomitant with reactivation from latency. Latently infected resting T cells are known to serve as a stable reservoir for HIV during anti-retroviral therapy and to produce infectious particles upon cell reactivation. Studies on HIV production from latently infected cells upon reactivation are necessary for a better understanding of HIV pathogenesis, although some studies have indicated intracellular accumulation of particles in chronically or latently infected cells 1011. Here, we employed J1.1 cells that were Jurkat T lymphocytic cells latently infected with HIV-1, and U1 cells that were U937 monocytic cells latently infected with HIV-1, and observed HIV particle budding following reactivation.

We initially tested the dose of TNF-α, and temporally monitored cell growth and HIV particle production after stimulation (Fig. 1A). J1.1 cells proliferated equally regardless of the dose of TNF-α, and the particle production levels increased to 50 ng/ml TNF-α. In contrast, proliferation of U1 cells was inhibited in a dose-dependent manner, and the highest level of particle production was observed at 50 ng/ml. We thus used 50 ng/ml TNF-α for further experiments. To avoid nonspecific stimulation by changing the medium, we added TNF-α directly to the culture medium, and this led to the higher dose of TNF-α required in our study than in other reports 1213.

Electron microscopy was carried out to examine where particle budding occurred in J1.1 and U1 cells upon reactivation (Fig. 1B). Little or no particles were produced in either cell line before TNF-α stimulation (Fig. 1B, most left panels), consistent with previous reports 11121314. Upon stimulation, nascent budding particles were visible on the surface of nearly all J1.1 cells, similar to the case with U1 cells (Fig. 1B, arrowheads). Unexpectedly, particles in intracellular vesicles were rarely seen in both J1.1 and U1 cells (Fig. 1B, arrow). The findings were confirmed by immunoelectric microscopy using anti-HIV-1 p17MA antibody (Fig. 1B, most right panels). Next, their parental cell lines, Jurkat and U937 cells, were infected with HIV-1, and particle production in acute infection were examined by electron microscopy. Particle budding was observed predominantly at the PM of both Jurkat and U937 cells (Fig. 1C, arrowheads) but some U937 cells displayed budding into intracellular compartments (Fig. 1C, arrows). Immunoelectric microscopy indicated similar results (Fig. 1C, most right panels). For quantification, we counted the number of cells containing particles at the PM alone or that of cells containing particles at both intracellular vesicles and the PM (Fig. 1D). Budding at the PM was prominent, regardless of whether cells were acutely or latently infected, or T lymphocytic or monocytic, suggesting that unlike chronically infected cells 10, HIV particles are most likely budded from the PM in latently infected cells, although it cannot be ruled out in this experiment that the particles observed in extracellular spaces might be released by exocytosis.

Gene expression analysis based on cDNA microarrays has extensively been employed and has provided evidence for the modulation of host cellular gene expression upon HIV infection (replication and latency) 151617181920. Although numerous host genes are modulated upon HIV infection, it is conceivable that expression levels of host membrane components may change by feedback regulation upon HIV reactivation, as HIV requires host cell membrane for particle budding. A membrane contains a number of microdomains, enriched in cholesterol (i.e., rafts) and in tetraspanins (e.g., CD63 and CD81), which accumulate at sites of HIV budding 7212223242526. It has been shown that TSG101, a component of endosomal sorting complex required for transport (ESCRT) is recruited to the sites of particle assembly and is responsible for HIV particle budding 2728. Thus we chose endosomal (EEA1, CD63, HRS, TSG101, and Syntaxin12) and PM (CD44 and SNAP23) markers and quantified their mRNA levels by qRT-PCR (Fig. 2A and 2B) using the primer sets shown in Additional File 1. Their properties and functions are as follows: EEA1 is a marker molecule for early endosome; HRS is an initial molecule for the ESCRT pathway; Syntaxin12 is a SNARE molecule for endosomal membrane fusion; CD44 is an adhesion molecule implicated in cell migration; SNAP23 is a SNARE molecule for PM fusion in the exocytic pathway. When the mRNA levels in J1.1 cells stimulated with TNF-α were compared with those in unstimulated J1.1 cells, CD44 gene expression was increased, but the other genes tested were largely unaltered. No significant upregulation of CD44 was observed when cells of its uninfected parental line, Jurkat, were similarly stimulated with TNF-α, indicating that the CD44 upregulation was not simply due to cell stimulation (Fig. 2A, upper). CD44 has been reported hardly expressed even at mRNA level in unstimulated Jurkat cells 29. A similar analysis was carried out for U1 cells. Downregulation of CD44 has been reported for chronically infected monocytic cells 30. Upon reactivation, CD44 upregulation was apparent but the endocytic molecules (CD63 and HRS) and SNAP23 were also upregulated in U1 cells. The modulation of others such as TSG101 was not statistically significant. These upregulations were not observed when uninfected U937 cells were stimulated (Fig. 2A, lower). The gene expression profiles upon reactivation were consistent with protein expression levels of the molecules when analyzed by Western blotting (Fig. 2C). We cannot simply compare the data of J1.1 and U1 cells, since expression levels of individual genes differ between the cell lines, but these significant upregulations were not observed in their parental but uninfected cells, suggesting that the upregulations might be linked with HIV expression. To test this possibility, we quantified expression levels of the same genes in acutely infected Jurkat and U937 cells and compared them with the levels in uninfected Jurkat and U937 cells. Upregulation of CD44 was observed in acutely infected Jurkat cells (Fig. 2B, upper), and this magnitude fold of upregulation was likely due to a very low level of CD44 expression in uninfected Jurkat cells 29. In acutely infected U937 cells, besides CD44 upregulation, upregulation of other genes (CD63, HRS, and SNAP23) was observed (Fig. 2B, lower). Together, the results indicate that the upregulation of host molecules observed here was likely to be linked with HIV production. Higher levels of gag mRNA than tat mRNA observed in this study were possibly because we analyzed at a late stage of HIV replication. Western blotting confirmed HIV antigens, p55Gag precursor and its processing products, p24CA and p17MA, appeared upon reactivation or infection and showed that unlike anti-p24CA antibody, anti-p17MA antibody used in this study (against the C-terminal region of p17MA) recognized the mature p17MA domain but not the unprocessed p55Gag (Fig. 2C).

Confocal microscopy revealed that the anti-p17MA antibody specifically detected mature p17MA produced upon HIV protease-mediated Gag processing (Fig. 3A). Since Gag processing occurs concomitant with particle budding, the p17MA signal obtained with the antibody most likely represents the sites of particle budding 431. No p17MA signal was seen on the day after infection, indicating that it was not derived from residual HIV (Fig. 3A). When J1.1 and U1 cells were reactivated, the p17MA antigens were observed at the cell periphery, likely at the PM, but no signals were seen in unstimulated cells (Fig. 3B). Similarly, the p17MA antigens were observed at the PM in acutely infected Jurkat and U937 cells (Fig. 3C). For quantification, we counted the numbers of cells based on p17MA distribution patterns (PM, intracellular+PM, or negative) and confirmed that HIV particles were preferentially formed at the PM (Fig. 3D), consistent with the data obtained by electron microscopy (Fig. 1D).

To understand the significance of the upregulation of host molecules observed here, we examined intracellular localization of the molecules by immunostaining. No CD44 staining was found in unstimulated J1.1 and U1 cells, consistent with previous reports indicating CD44 downregulation during latency 3032. Following reactivation, CD44 was visible and colocalized with the p17MA antigens at the PM. Similarly, CD63 and HRS stainings were rarely seen in unstimulated cells but became visible and colocalized with the 17MA signals, especially in U1 cells (Fig. 3B). These findings were very apparent in acutely infected cells (Fig. 3C): CD63 recruitment, as reported previously for acutely infected Jurkat T cells and macrophages 7253334, and HRS and CD44 accumulations to the sites where mature particles were formed. Together, our data suggest that HIV expression may lead to the upregulation of certain host molecules that are recruited to the sites of particle assembly, possibly to coordinate particle production. Because CD44 is a cell adhesion molecule that mediates lymphocyte aggregation and homing 3536, it is conceivable that the CD44 recruitment to HIV assembly sites may lead to an efficient cell-to-cell transmission of HIV and infected cell migration to lymph nodes.

In conclusion, despite numerous literature on HIV budding to intracellular compartments especially in macrophages, our data indicate that upon reactivation from latent infection, HIV predominantly buds at the PM, a morphology that is similar to particle budding in acute infection, suggesting that HIV latency have a potential for robust production of HIV observed for acute infection.

HIV: human immunodeficiency virus; TNF-α: tumor necrosis factor-α; CA: capsid; MA: matrix; qRT-PCR: quantitative RT-PCR; TSG101: tumor susceptibility gene-101; HRS: hepatocyte growth factor regulated tyrosine kinase substrate; EEA1: early endosomal antigen 1; SNAP23: synaptosome associated 23 kDa protein.

The authors declare that they have no competing interests.

MS performed the qRT-PCR analysis and confocal study. ED, TG, and KS carried out the electron microscopy analysis. YM designed the experiment and wrote the manuscript.