INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Upon infection, the RNA genome of human immunodeficiency virus type 1 (HIV-1) is reverse transcribed into a linear, double-stranded DNA molecule which is destined to integrate into the cellular genome to produce new viral RNA. The proteins involved in reverse transcription are assembled in the cytoplasm together with the viral genome into the preintegration complex, which is actively transported into the nucleus through the nuclear pores, a process unique to lentiviruses which allows them to access the host cell DNA independent of the cell cycle (10). Nuclear localization signals within the matrix, Vpr, and integrase proteins have been described (for a review, see references 9 and 25), and recently a central DNA flap of the HIV-1 genome was reported to be essential for nuclear uptake (62). Once in the nucleus, the proviral DNA has to integrate into the host cell genome for viral replication. The degree to which integration is a prerequisite for viral transcription at all has been widely discussed, but it appears that productive infection, at least, requires integrated viral genomes (12, 13, 21, 22, 54, 56, 59).

However, not all of the reverse-transcribed HIV-1 DNA will actually become integrated into the chromosomal DNA. A large proportion of the viral DNA molecules that are synthesized after infection remain unintegrated, a phenomenon that seems to be common to all cytopathic animal retroviruses (for a review, see references 8 and 14). Unintegrated HIV-1 DNA has gained clinical importance because its accumulation within infected cells might be associated with cytopathic effects or might be used as a marker for disease progression (28, 46, 47, 49, 50, 61). Whereas the cellular distribution of HIV-1 RNA has been investigated in several studies by in situ hybridization (6, 7, 23, 24, 41, 63), to our knowledge no data are currently available about the intranuclear localization of HIV-1 DNA, integrated as well as unintegrated, in relation to nuclear domains. Two types of domains within the cell nucleus are of particular interest when studying the distribution of viral genomes and transcripts, i.e., nuclear domain 10 (ND10) (also termed the promyelocytic leukemia protein [PML] oncogenic domain or PML body) and SC35 domains (also known as interchromatinic granule clusters).

The ND10 contain several interferon-upregulated proteins such as PML and Sp100 and have gained increasing interest as the nuclear site where DNA viruses start transcription and replication. Although these viruses apparently enter the nucleus randomly, they transcribe and replicate only in immediate association with ND10 (for a review, see reference 37). Moreover, several DNA viruses have the ability to modify or disperse ND10 after starting their transcription at this site (30, 31, 38). RNA viruses also appear to interact with ND10. Hepatitis delta virus, a satellite virus of human hepatitis B virus, aggregates its antigenomic RNA preferentially at ND10 and thereby causes a segregation of ND10 proteins (3). Overexpression of PML, an essential constituent of ND10, reduces replication of vesicular stomatitis virus and influenza A virus (15), suggesting that ND10 are involved in an antiviral defense mechanism.

SC35 domains represent the second nuclear component which is often used as a landmark in intranuclear localization studies. These domains are defined by aggregations of splicing factor SC35 (26) but also contain other components of the splicing machinery (for a review, see reference 43). SC35 domains are located in the interchromosomal space; transcripts of many genes, both cellular and viral, show a specific type of association with this domain (20, 31, 40, 55). The relationship between HIV-1 transcripts and SC35 domains has been studied mainly by transfecting cells with plasmids containing the proviral genome. Unspliced HIV-1 RNA and SC35 domains appear to be more or less randomly distributed within the nucleus without being specifically associated with each other (6, 7, 63), although a certain degree of specific association of both components has been described (23). In contrast, spliced HIV-1 RNAs may show a different localization pattern by accumulating within the splicing factor domains as has been shown for the multiply spliced transcripts encoding the tat gene (6).

In this report, we studied the fate of unintegrated HIV-1 DNA within the nuclei of infected cells by in situ hybridization and compared its nuclear distribution with the localization pattern of integrated DNA and viral transcripts as well as the distribution of ND10 and SC35 domains. Our studies were performed at a single-cell level allowing the direct visualization of viral genomes instead of studying whole cell populations. Also, we freshly infected cells with infectious virus particles rather than using transfections, as in earlier studies addressing the intranuclear distribution of HIV-1 RNA. To distinguish in situ unintegrated from integrated HIV-1 DNA, we employed stress-induced chromosome condensation (SICC) (51), which is a combination of heat shock and osmotic stress that leads to the contraction of interphase chromosomes within 15 to 20 min and thus reveals the interchromosomal space which is otherwise not visible during interphase. We show that unintegrated DNA accumulates between the chromosomes, a distribution pattern that is in sharp contrast to the specific chromosomal localization of integrated DNA. We demonstrate that unintegrated DNA aggregates into clusters and accumulates in the nucleus before viral transcripts are detectable. In contrast to several other viruses, neither HIV-1 RNA nor DNA affiliates with ND10. However, unintegrated HIV-1 DNA genomes, but not integrated genomes or transcription sites, specifically associate with SC35 domains. The implications of these localization patterns with regard to the possible cytopathic effect of unintegrated HIV-1 DNA are discussed.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Cell culture and infection. The following cell lines were obtained from the National Institutes of Health (NIH) AIDS Research and Reference Program: HeLa/LAV, a permanently HIV-1-infected HeLa cell line that carries integrated viral genomes (5); HOS-CD4.CCR5, a cell line that expresses CD4 and CCR5 (19); and U373-MAGI-CXCR4_CEM, which expresses CD4 and CXCR4 (58). Cells were grown in Dulbecco's modified Eagle medium supplemented with 10% fetal bovine serum and 1% penicillin-streptomycin at 37°C in a humidified atmosphere containing 5% CO₂. For the maintenance of HOS-CD4.CCR5 and U373-MAGI-CXCR4_CEM cells, puromycin (Sigma) was added to the medium at a concentration of 1 µg/ml.

Immunofluorescence and microscopy. ND10 were visualized with rabbit sera against Sp100 or PML (34), and SC35 domains were detected using a monoclonal antibody against splicing factor SC35 (Sigma-Aldrich) (26). Monoclonal antibody HIV.OT 34A specific for p24 was obtained from ICN.

FISH. Cells were fixed in 4% PFA or, when fluorescent in situ hybridization (FISH) was combined with immunocytochemistry, were first immunostained as described above and then refixed in 4% PFA to cross-link bound antibodies (39). Cells were then permeabilized in 0.2% Triton X-100 and, for detection of DNA, treated with RNase (Boehringer) at a concentration of 100 µg/ml in PBS for 30 min at 37°C. After equilibration in 2× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate), cells were dehydrated in an ethanol series (70, 80, and 100% ethanol for 3 min each at 20°C), air dried, and incubated overnight at 37°C with the hybridization mixture. Plasmid p89.6 FS (kindly provided by R. Collman) (16), which contains the HIV-1 89.6 genome, was labeled with biotin-16-dUTP by nick translation and used as a probe. The DNase concentration was adjusted to yield probe DNA with a fragment length of 200 to 500 bp. Probe DNA was dissolved at a concentration of 10 ng/µl in 50% formamide in 2× SSC containing 10% dextran sulfate, 100 ng of salmon sperm DNA (Gibco BRL)/µl, 1 µg of yeast tRNA (Sigma)/µl, and 0.5 mg of cot1 DNA (Gibco BRL)/µl. For DNA detection, probe and cells were simultaneously heated at 94°C for 4 min to denature DNA. To detect RNA, only the probe DNA was denatured at 90°C for 5 min. After hybridization, specimens were washed at 37°C with 55% formamide in 2× SSC (twice for 15 min each time), 2× SSC (10 min), and 0.25× SSC (twice for 5 min each time). Hybridized probes were labeled with FITC-avidin (1:500 in 4× SSC plus 0.5% bovine serum albumin; Vector Laboratories), and signals were amplified with biotinylated anti-avidin (1:250; Vector Laboratories), followed by another round of FITC-avidin staining. Finally, cells were equilibrated in PBS, stained for DNA with Hoechst 33258 (0.5 µg/ml) or propidium iodide (1 µg/ml), and mounted with Fluoromount G (Fisher Scientific).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Unintegrated HIV-1 DNA accumulates within the interchromosomal space. Flat, adherent cell lines are far more amenable to the in situ hybridization procedure than round suspension cells such as lymphocytes. To visualize HIV-1 genomes by FISH, we therefore used cell lines derived from human glioblastoma cells (U373-MAGI-CXCR4_CEM) and human osteosarcoma cells (HOS-CD4.CCR5) which express CD4 and the coreceptors CXCR4 and CCR5, respectively. Staining with antibodies against p24 2 to 3 days after a 15-h infection with HIV-1 strain 89.6 demonstrated high infection rates in both cell lines with up to 95% p24-positive cells (data not shown). Generally, U373-MAGI-CXCR4_CEM cells yielded slightly higher infection rates than HOS-CD4.CCR5 cells. When hybridized with an HIV-1-specific probe to detect reverse-transcribed DNA, cells incubated for 15 h in HIV-1-containing culture supernatant usually displayed 10 to 50 nuclear signals of variable size (Fig. 1A). As controls, FISH was performed with uninfected cells, or infected cells were hybridized with a probe not specific for HIV (biotinylated bacteriophage  DNA). In either case, no nuclear or cytoplasmic signals could be detected. DNase treatment prior to hybridization also eliminated any detectable signals, demonstrating that all visible spots actually represented HIV-1 DNA and not viral RNA. We further infected cells in the presence of AZT-5'-triphosphate at concentrations of 0.01, 0.1, and 1 µM to see whether the detected viral DNA molecules were synthesized de novo by reverse transcription. As expected, AZT-treated cells showed a dose-dependent reduction of detectable DNA signals. Most nuclei of cells treated with 1 µM AZT lacked viral DNA. In most of the untreated cells, viral DNA could be found only within the nucleus, whereas cells which usually displayed 3 to 10 cytoplasmic signals were observed only occasionally (data not shown).

View larger version (48K): [in this window] [in a new window]   FIG. 1.   FISH and immunostaining of HIV-1-infected cells. Labeled components are denoted in their respective color in each panel. In some images (A, B, F, K, and L), Hoechst 33258 staining of cellular DNA is shown in blue to visualize nuclei. Images C through E and G through J were acquired by confocal microscopy, whereas images A, B, F, K, and L represent nonconfocal micrographs. (A) HOS-CD4.CCR5 cell after 15 h of infection with HIV-1 and hybridization with an HIV-1-specific probe to detect viral DNA. The nucleus displays numerous HIV-1 DNA signals of various sizes. (B) In HeLa/LAV, a permanently HIV-1-infected cell line that carries integrated viral genomes, individual HIV-1 DNA signals of uniform size are detectable by FISH (arrows). (C) Infected HOS-CD4.CCR5 cell that underwent the SICC procedure to condense interphase chromosomes and was probed for HIV-1 DNA. Viral DNA can be found almost exclusively between the contracted chromosomes stained by propidium iodide. (D) HeLa/LAV cell processed as described in the legend for panel C. In contrast to DNA in infected HOS-CD4.CCR5 cells, most viral DNA in HeLa/LAV cells localizes on the condensed chromosomes (arrow). Note that the HIV-1 signal is present as a double dot, indicating the presence of an integrated viral genome in both sister chromatids after S phase. (E) Visualization of HIV-1 transcripts in an infected HOS-CD4.CCR5 cell in which SICC was induced. Viral RNA foci denote the sites where HIV-1 transcripts are released from the integrated proviral DNA into the interchromosomal space. The insert shows at higher magnification that these RNA foci actually emerge from the chromosomal DNA. Note that for detection of transcripts, cells are not treated with RNase, resulting in additional staining of RNA by propidium iodide as seen in part of the cytoplasm. (F) Accumulation of HIV-1 DNA in the nucleus of a HOS-CD4.CCR5 cell upon multiple infection. Unintegrated HIV-1 DNA aggregates into clusters as revealed by signals of different sizes. (G) Infected HOS-CD4.CCR5 cell probed for HIV-1 RNA and immunostained with antibodies against the ND10 protein Sp100. HIV-1 transcripts are concentrated into foci (arrows) which are not associated with ND10. (H) Infected HOS-CD4.CCR5 cell probed for HIV-1 DNA and stained with antibodies against Sp100 to visualize ND10. DNA signals are not associated with ND10. (I) Visualization of HIV-1 RNA by FISH and splicing factor SC35 by immunostaining of an infected HOS-CD4.CCR5 cell. RNA foci and SC35 domains are not specifically associated with each other but show a random nuclear distribution. The right cell is uninfected. (J) Infected HOS-CD4.CCR5 cell stained for HIV-1 DNA by FISH and SC35 by immunofluorescence. HIV-1 DNA associates with SC35 domains. (K) U373-MAGI-CXCR4CEM cells infected for 15 h and probed for HIV-1 DNA. Virtually all cells contain HIV-1 DNA. (L) The same experiment described in the legend to panel K was performed, but cells were probed for HIV-1 RNA. Only 13% of the cells display signals specific for HIV-1 RNA (arrow).

Nuclear unintegrated HIV-1 DNA aggregates into clusters. A striking feature of nuclear HIV-1 DNA in HOS-CD4.CCR5 and U373-MAGI-CXCR4_CEM cells was the variable size of signals obtained by FISH, since the size of in situ signals strictly correlates with size and/or number of the target molecules. Unintegrated DNA in both cell lines could be visualized as dots with a wide range of different sizes, often in the form of irregularly shaped aggregates up to several microns in diameter (Fig. 1F). In contrast, integrated viral DNA in HeLa/LAV cells was visible as more or less uniform signals of smaller size (Fig. 1B). Quantification by counting the number of pixels for each signal confirmed that integrated DNA was represented only by small-sized signals (mostly 2 to 4 pixels with an average signal size of 2.7 pixels), whereas unintegrated DNA yielded both small and large dots without any obvious size limit (average signal size, 7.8 pixels; the largest signal observed comprised 155 pixels) (Fig. 2). Since integrated DNA corresponds to single proviral genomes, we assume that the small signals observed in HeLa/LAV cells (two to four pixels) represent the size of one individual HIV-1 DNA molecule. Therefore, about half of the signals from unintegrated viral DNA (53%) appear to represent single HIV-1 genomes, whereas 47% of the DNA dots seem to be composed of multiple, closely associated copies of unintegrated DNA. The great majority of the total amount of unintegrated DNA molecules is therefore aggregated as clusters in the nucleus. The mean size of those DNA clusters correlated to the length of the infection time, with an average of 3.0 pixels per signal after 6 h of infection and 7.8 pixels per signal after 15 h of infection in HOS-CD4.CCR5 cells, indicating that viral DNA entering the nucleus is continuously forming aggregates.

View larger version (13K): [in this window] [in a new window]   FIG. 2.   Sizes of hybridization signals representing integrated DNA (in HeLa/LAV cells) and unintegrated DNA (in HOS-CD4.CCR5 cells infected for 15 h). The sizes of 150 randomly chosen signals in each cell line were quantified by counting the number of pixels of each signal (see Materials and Methods). Integrated DNA is detected as more or less uniform signals of small sizes (two to seven pixels), whereas unintegrated DNA comprises a wide variety of sizes, indicating its tendency to aggregate into clusters.

HIV-1 DNA and transcripts are not specifically associated with ND10. To study the localization of HIV-1 genomes in relation to ND10, we performed immunohistochemistry combined with FISH with infected HOS-CD4.CCR5 and HeLa/LAV cells. HOS-CD4.CCR5 cells possess considerably larger ND10 than U373-MAGI-CXCR4_CEM cells and thus allowed an easier identification of the ND10 domain by immunostaining. When we identified ND10 with antibodies against PML or Sp100 and simultaneously stained unintegrated HIV-1 DNA by FISH, cells which contained HIV-1 DNA still possessed intact ND10, whose number and appearance were not different from those in uninfected cells. ND10 remained unaffected both at early times after infection (30 min to 6 h) and after infection periods of 15 to 18 h. Unintegrated DNA was not specifically associated with ND10, and only occasionally were ND10 and viral DNA found in close association, obviously reflecting a random distribution of both components (Fig. 1H). Thus, in contrast to DNA viruses, HIV-1 DNA neither destroys nor associates with ND10. Since most of the unintegrated HIV-1 DNA probably is transcriptionally inactive or supports only low-level transcription, we wondered whether transcribing (i.e., integrated) viral genomes behaved differently and interacted with ND10. Visualization of total HIV-1 RNA by FISH of infected HOS-CD4.CCR5 cells usually revealed one to four RNA foci, most likely denoting sites of integrated HIV-1 genomes, and lower concentrations of transcripts more or less evenly distributed within the nucleus except for the nucleoli which lacked any visible signals. As in the case of HIV-1 DNA, cells positive for HIV-1 transcripts possessed intact ND10, as shown by costaining for PML or Sp100. Also, the transcription foci could not be found to be specifically associated with ND10 (Fig. 1G). We furthermore tested viral DNA in HeLa/LAV cells, i.e., integrated DNA, for its association with ND10. As expected from the absence of transcription sites from ND10 in HOS-CD4.CCR5 cells, integrated proviral DNA also did not show any specific localization pattern in relation to ND10 (data not shown). We conclude that HIV-1, in contrast to many other viruses, does not interact with ND10 irrespective of its status as DNA or RNA genome.

Unintegrated HIV-1 DNA, but not RNA, associates with SC35 domains. The interaction between unspliced HIV-1 transcripts and SC35 domains has been previously investigated, mostly by transfection of cells with plasmids containing the HIV-1 genome. Both colocalization (23) and a random distribution of both components (6, 7, 63) have been described. We reexamined the behavior of HIV-1 RNA in relation to SC35 domains in infected instead of transfected cells using a probe that comprises the whole HIV-1 genome and also tested whether HIV-1 DNA interacted with this domain. Concentrations of viral RNA denoting integrated genomes after infection of HOS-CD4.CCR5 or U373-MAGI-CXCR4_CEM cells did not specifically associate with SC35 domains. Although about 50% of the RNA foci were observed in direct contact with the splicing factor domains, this finding obviously reflects a random distribution of both components, given the relatively large nuclear volume occupied by these domains (Fig. 1I). Nucleoli lacked both SC35 domains and detectable HIV-1 transcripts and therefore appeared as empty holes within the nucleus (Fig. 1I). Likewise, hybridization signals in HeLa/LAV cells representing integrated HIV genomes were also associated only in a random fashion with SC35 domains (data not shown).

Accumulation of HIV-1 DNA precedes viral transcription. Although unintegrated HIV-1 DNA is probably not able to support productive viral replication, several studies suggest that it is able to transcribe (for a review, see reference 14). Since we observed high levels of unintegrated DNA in infected cells, we wondered how much the accumulation of unintegrated DNA correlated with the production of HIV-1 RNA. We incubated U373-MAGI-CXCR4_CEM cells in HIV-1-containing culture supernatant for 15 h and then further in virus-free medium. Viral DNA could be detected in 15% of the cells 6 h after the infection was started; however, no HIV-1 transcripts were visible at this time point. After a 15-h incubation with HIV-1, virtually all nuclei contained viral DNA, whereas viral RNA was observed in only 13% of the cells (Fig. 1K and L). Forty hours after infection, the number of HIV-1 RNA-positive cells finally reached levels similar to those of cells displaying viral DNA with 95% RNA-positive and 100% DNA-positive cells (Fig. 3). These findings suggest that the production of HIV-1 transcripts is a relatively slow process compared to the accumulation of viral DNA. Even after 40 h, although 100% of the cells contained high numbers of viral DNA for at least 25 h, 5% of the cells still lacked detectable transcripts.

View larger version (16K): [in this window] [in a new window]   FIG. 3.   U373-MAGI-CXCR4CEM cells were infected with HIV-1 for 15 h and then further incubated in virus-free medium. Cells were fixed 6, 15, and 40 h after starting the infection and probed for HIV-1 DNA or RNA. Six hours after infection, viral DNA, but no RNA, was visible in 15% of the cells. After 15 h, virtually all cells contained viral DNA whereas RNA was detectable in only 13% of the cells. Forty hours after infection, the number of cells containing HIV-1 transcripts increased to 95%. Viral DNA remained detectable in all cells after 40 h even though the HIV-1-containing supernatant had been removed earlier at 15 h postinfection.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Since efforts are currently being made to develop integrase inhibitors for AIDS therapy (18, 48, 52), a model system that allows the visualization of unintegrated HIV-1 DNA at the single-cell level might represent a useful tool to study the dynamics of the integration process. Here, we employed a novel method to distinguish in situ integrated from unintegrated HIV-1 DNA. Cells treated to perform SICC (51) contract their chromatin strands and thus clearly show the two major nuclear compartments, i.e., chromosomes and the interchromosomal space. In contrast to mitotic cells, SICC-treated cells are still in interphase and thus retain their nuclear envelope, allowing a distinction between nucleus and cytoplasm. FISH revealed the great majority of the nuclear proviral genomes either in the interchromosomal space in infected HOS-CD4.CCR5 and U373-MAGI-CXCR4_CEM cells or colocalizing with the cellular DNA in HeLa/LAV cells. Therefore, using this approach, we showed that nearly all HIV-1 DNA visible in freshly infected cells represents the unintegrated form. Here, we demonstrated specific localization features of unintegrated HIV-1 DNA, i.e., it aggregated into clusters within the nucleus and localized within or in close association with SC35 domains but showed no affiliation with ND10.

Transcription and replication of many DNA viruses is visible only at ND10. This localization pattern is believed to represent a cellular segregation mechanism to confine or restrict these viral activities within the nucleus. This view is supported by the fact that ND10 contain several interferon-upregulated proteins and that herpesviruses as well as adenovirus type 5 destroy or modify ND10 after having started their replication at these sites. Alternatively, ND10 might represent nuclear sites required by viruses to replicate (for a review, see reference 37). These nuclear domains have therefore gained interest as a possible target to interfere with or manipulate viral replication. However, with HIV-1 we never observed unintegrated or integrated DNA or viral transcription foci affiliated with ND10, except for what appeared to be random associations. This domain, therefore, appears not to be directly involved in the replicative cycle of HIV.

In contrast, unintegrated HIV-1 DNA showed a clear association with splicing factor domains containing the marker protein SC35. Over 90% of the unintegrated DNA molecules were observed to colocalize with SC35 domains, whereas only about 50% of viral transcription foci (or integrated viral DNA) were found to associate with this domain. The association between HIV-1 transcription sites and SC35 domains appears to be the result of a random distribution of both structures, due to the relatively large nuclear volume occupied by the splicing factor domains. In contrast, the almost exclusive localization of unintegrated DNA in or at these domains obviously reflects a nonrandom distribution pattern. A similar behavior has been described for microinjected simian virus 40 DNA, which also accumulated within the SC35 domains (17). However, viral DNA is not generally targeted into SC35 domains, although viral transcripts may show a clear association with this structure (31). The specificity of the colocalization of HIV-1 DNA with SC35 domains becomes even more evident when compared with the nuclear distribution of Epstein-Barr virus episomes. As we have shown previously, the circular, unintegrated genomes of Epstein-Barr virus localize almost exclusively outside of SC35 domains in latently infected cells (4). Moreover, they do not form aggregates but remain as individual molecules in the host cell nucleus, in contrast to unintegrated HIV-1 DNA, which is prone to form clusters containing many copies of the viral genomes.

It is noteworthy in this context that morphological changes in the three-dimensional structure of SC35 domains have been described for HIV-1-infected cells (36). SC35 domains contain high concentrations of splicing factors and small nuclear RNAs (snRNAs), as well as transcription factors, although splicing itself appears not to occur within these domains. SC35 domains are therefore believed to be necessary for the regulation of splicing factor ratios and concentrations, for recycling of splicing components, and for the assembly of the transcription and processing machinery (42-44). The presence of large numbers of unintegrated HIV-1 molecules within this domain could therefore interfere with the regulation of splicing factor concentrations in the nucleus.

The large amount of unintegrated HIV-1 DNA in the nuclei of infected cells is generally believed to be the result of multiple infections (50, 53). When we infected cells in the presence of the DNA replication inhibitor aphidicolin, replication-arrested and control cells displayed equal amounts of unintegrated DNA, also indicating that infection but not viral DNA replication is the only source for HIV-1 DNA. The first appearance of reverse-transcribed DNA has been reported at 4 h (2, 32) or even at 1 to 2 h (27) after infection. Together with our results showing that after 6 h viral DNA is already present in 15% of U373-MAGI-CXCR4_CEM cells, these data indicate that the generation of HIV-1 DNA represents a rapid process probably starting immediately after infection. In a lymphocyte cell line, the amount of viral DNA was found to gradually increase under single-cycle growth conditions for HIV-1 until a peak was reached at 8 to 12 h postinfection (32). Interestingly, the amount of HIV-1 DNA decreased after having reached the peak (32), whereas in our system the viral DNA appeared to be more or less stable at least until 40 h after infection. The number of unintegrated DNA molecules in infected lymphocytes has been estimated to be between 20 and more than 400 copies per cell (32, 45, 53), similar to our observations of infected HOS-CD4.CCR5 and U373-MAGI-CXCR4_CEM cells. The advantage of in situ hybridization combined with SICC is the direct visualization of unintegrated HIV-1 DNA molecules, although due to the aggregation of the viral DNA, the exact number of molecules contained in clusters can only be roughly estimated.

After a 15-h infection, virtually all U373-MAGI-CXCR4_CEM cells contained unintegrated DNA, but only a relatively small fraction of these cells displayed viral transcripts. Our FISH procedure allows the visualization of individual full-length HIV-1 transcripts; however, smaller RNA species may escape detection at the single-molecule level. We therefore conclude that unintegrated DNA molecules are unable to produce full-length transcripts or, alternatively, high concentrations of smaller transcripts which would be equally detectable. Since the first appearance of viral RNA from certain genes (tat, nef) could already be observed by PCR 2 to 3 h after infection of H9 cells (27), the generation of low levels of small RNA species of subgenomic size may have remained undetectable in our study. The number of transcription foci (which most likely denote the site of the integrated provirus) was considerably lower than the number of viral DNA molecules; integration, as a prerequisite for detectable transcription, therefore appeared to be a relatively rare event. Also, the generation of detectable transcripts clearly lagged behind the first appearance of unintegrated DNA, suggesting that integration or at least production of full-length viral RNA represents a relatively slow process. Using a coculture system for infection of MT4 cells, it has been estimated that at least one third of all input reverse-transcribed DNA will become integrated (2), whereas our data suggest a much lower integration rate. We further observed a considerable increase (from 15 to 95%) of cells containing viral transcripts in the absence of HIV-1-containing supernatants after an initial infection, indicating either that previously formed unintegrated DNA molecules (present in 100% of the U373-MAGI-CXCR4_CEM cells) may serve as a reservoir for later integration events or that integrated DNA may not immediately start transcription, at least at high levels. Earlier reports support both possibilities. In quiescent T cells, unintegrated DNA was found to retain the ability to integrate later upon T-cell activation in vitro (11). In a lymphocyte cell line with single-cycle growth conditions for HIV-1, the level of unspliced transcripts remained low until after 24 h of infection; whereas subgenomic RNAs, which might be less easier to detect by FISH, accumulated earlier (32). Although there seems to exist a considerable amount of integration-competent DNA, it has also been shown that newly synthesized viral DNA may be highly labile at least before being imported into the nucleus (33, 60). This might account for our observation that HIV-1 DNA was only occasionally detectable within the cytoplasm despite being present in the nucleus in large amounts. It remains to be determined whether future therapeutic interventions which block the HIV-1 integrase will be affected by a stable pool of integration-competent DNA.