We first sought to establish a quantitative relationship between the ability of HIV-1 virions to enter T-cells of mouse and human origin and, subsequently, to express a reporter gene in these target cells. To ensure comparable conditions in the cross-species comparisons, we employed an HIV-1 based lentiviral vector encoding for GFP driven by a cytomegalovirus immediate early promoter (HIV-CMV-GFP), which was pseudotyped with the vesicular stomatitis virus glyco-protein (VSV-G). Notably, the expression of GFP from this vector is not influenced by HIV-1 Tat/cyclin T1-dependent, potentially species-specific differences in LTR transactivation 3. Through incorporation of enzymatically active β-lactamase-Vpr fusion proteins (BlaM-Vpr) during virus production the efficiency of HIV-1 entry into target cells was specifically measured by CCF2 substrate cleavage in a flow cytometry-based virion-fusion assay 2223.

Following a single challenge with this dual HIV-1 reporter virus, T-cell lines of human and mouse origin were analyzed for virion fusion and early gene expression, 6 h p.i. and on day 3 p.i., respectively. Fig. 1 depicts representative flow cytometric data of both of these analyses for MT-4 (human) and S1A.TB (mouse) T-cells, in which gate R2 defines the cleaved CCF2 (blue fluorescence emission)-positive subpopulation (Fig. 1A, B; upper panels) or the GFP-positive subpopulation (Fig. 1A, B; lower panels) of all viable cells (gate R1), respectively. The specificity of virion entry and viral gene expression was confirmed by neutralization with an anti-VSV-G monoclonal antibody 24 or by pretreatment with the reverse transcriptase inhibitor efavirenz, respectively (Fig. 1; right panels).

T-cell lines from both species allowed quite similar levels of entry of the BlaM-Vpr-loaded HIV-CMV-GFP virus (Figs. 1A, B; upper panels), ranging on average from 10 to 31% (Fig. 2A). In stark contrast, analysis of GFP reporter expression showed a 67- to 290-fold reduction in the percentage of infected mouse T-cell lines (TIMI.4; R1.1, S1A.TB) expressing the reporter transgene compared to human MT-4 T-cells (Figs. 1A, B; lower panels; Fig. 2B). This degree of impairment in gene expression was also seen when equal titres of VSV-G HIV-CMV-GFP, that did not carry BlaM-Vpr, were used, or when gene expression was assessed on day 7 p.i. (data not shown). In a more refined analysis, the ratio of the percentages of cells that scored positive for gene expression (Fig. 2B) relative to virion entry (Fig. 2A) was defined for each T-cell line as a cumulative Relative Post-Entry Efficiency revealing a mouse-human species differerence of 55- to 235-fold (Fig. 2C). In summary, these consecutive analysis of virion entry and CMV-driven reporter gene expression from a single infection corroborate the observation of a severe post-entry block for HIV-1 in mouse T-cell lines 2021.

To characterize at which step of the replication cycle following entry HIV-1 encounters a block in murine T-cells, levels of late HIV-1 cDNAs and episomal 2-LTR circles were analyzed as markers for reverse transcription and nuclear import of the pre-integration complex, respectively. DNA was extracted from infected mouse and human T-cell lines (from the experiment shown in Figs. 1, 2), aliquots of which were harvested 24 h p.i. and analyzed by real-time PCR. The HIV-1 cDNA species were quantified using established protocols, specificity controls, and quantitative standards for either HIV-1 cDNA species and normalized to cellular DNA levels, which were determined in a parallel reaction by amplification of a cellular gene 625.

Following comparable levels of virion entry (Fig. 2A), levels of total HIV-1 cDNA were found to be quite similar in the cross-species comparison (Fig. 3A), suggesting an efficient reverse transcription process in these rodent cells. Similarly, levels of episomal 2-LTR circles were in the same range or slightly elevated in mouse T-cell lines relative to the human counterpart (Fig. 3B). Following normalization with levels of de novo synthesized HIV-1 cDNA (Fig. 3A), 2-LTR circle levels (Fig. 3B) turned out to be statistically indistinguishable (data not shown). Importantly, 2-LTR circles were also detected in HIV-1-infected primary mouse T-cells derived from splenocyte pools of 3 BALB/c mice. Levels of 2-LTR circles were comparable (Fig. 3D; mouse donor pool #1) or ~16-fold higher (Fig. 3D; mouse donor pool #2) than in primary human T-cell cultures, indicating that following virion entry (Fig. 3C) the processes of reverse transcription and nuclear import are intact in these primary mouse targets. As a specificity control, no 2-LTR circles could be detected in efavirenz-treated cultures, demonstrating that the amplified episomal HIV-1 cDNAs had been generated from de novo synthesized viral DNA and were not present in the inoculum (data not shown). Due to the generally low infection level and residual DNase-resistant, production-related plasmid contaminations in virus stocks, levels of de novo synthesized viral DNA could not be quantified separately in primary T-cells (data not shown). In summary, these results suggest that following entry of virions, reverse transcription occurs efficiently in mouse T-cells. Furthermore, abundant levels of 2-LTR circles suggest robust import of the pre-integration complex into the nucleus. This tentatively maps the replication barrier in mouse T-cells to a step after nuclear entry.

Next, we quantified provirus formation in infected mouse and human T-cells. In principle, a defect at the level of integration can drastically diminish or completely abrogate viral gene expression 2627. Similar to reported nested PCR strategies to amplify HIV-1 integrated in proximity to highly abundant genomic repeat elements in human cells (Alu elements) 28, or in rat cells (BC elements) 6, we designed a nested real-time PCR to specifically quantify integrated HIV-1 provirus in mouse cells using the most abundant consensus sequence B1 within mouse SINE elements 29, as the repeat target for the cellular anchor primer pair in the genome of this species (Fig. 4A).

To establish a standard for quantitative analyses of integration into the mouse genome, a stable polyclonal population of NIH3T3 fibroblasts containing integrated HIV-1 provirus was generated by infection with VSV-G HIV-1_NL4-3 GFP at a low MOI, cell passage for 7 weeks to allow complete loss of all unintegrated HIV-1 cDNA species, and subsequent enrichment of GFP-positive cells by flow cytometric sorting (thereafter referred to as NIH3T3P^int cells), in principle as reported previously for the rat species 6. Since these NIH3T3P^int cells no longer contain unintegrated HIV-1 cDNA species, the absolute number of HIV-1 integrants was accurately quantified by the number of total HIV-1 cDNA copies (54 HIV-1 cDNA copies per ng DNA), thus providing a faithful reference for the integration PCR standard in the mouse genome. Fig. 4B depicts a typical mouse HIV-1 integration standard plotted as a function of the natural logarithm of the concentration of HIV-1 versus the PCR cycle threshold. This standard has a dynamic range of over 3 logs with a highest copy number of 36.741, and both the slope and R² value were considered as quality controls in individual experiments.

The nested PCR strategy for quantification of HIV-1 integrants in mouse cells is depicted in Fig. 4A, and described in the figure legend and under Methods. This mouse integration PCR and a human integration PCR, the latter essentially following a published protocol 28, were validated side-by-side using genomic DNA from NIH3T3^int or HeLa^int cells 6, respectively (Fig. 5A). Here, the number of HIV-1 integrants per ng DNA for the complete PCR reaction was set to 100% for each species. First, omission of LTR primer #1521 from the first-round PCR reaction resulted in a loss of the amplification signal in both species. Second, a reaction mix without the cellular anchor primer pair (Fig. 5A; #2194 and #2231 (mouse); or #1519 and #1520 (human), 6) yielded low residual signals, most likely due to the partial formation of single-stranded DNA from LTR-containing HIV-1 cDNA by the first-round LTR primer, as reported 628. Finally, omission of the first-round PCR reaction did not give a signal above background (Fig. 5A), indicating that second-round amplification of non-preamplified DNA is not of significant concern.

Next, the mouse integration PCR was further validated in a dynamic infection context. Levels of total HIV-1 cDNA and integrants were quantified in parental NIH3T3 fibroblasts infected with either the integration-competent (IN wt) HIV-CMV-GFP or an integration-defective isogenic HIV-1 vector (IN(D64V)), the latter carrying a catalytic core mutation in integrase 30. One day after infection, high levels of total HIV-1 cDNA were found in NIH3T3 cells challenged with either lentiviral vector, while only background levels could be amplified from efavirenz-treated controls (Fig. 5B). In DNA extracted on day 7 p.i., integrants were readily amplified by the newly developed real-time PCR protocol in NIH3T3 cells infected with the IN wt vector, while the level of provirus formation was severely reduced with the IN(D64V) mutant (Fig. 5C). Collectively, a real-time PCR for the specific detection and absolute quantification of HIV-1 integrants in the genome of infected mouse cells was established and validated.

Next, we applied the mouse integration PCR to a dynamic VSV-G HIV-CMV-GFP infection in mouse T-cell lines. While virion entry (Fig. 2A), reverse transcription (Fig. 3A), and nuclear import (Fig. 3B) appeared to be intact in infected mouse T-cells, HIV-1 integrants from the identical experiment were undetectable in the genome of TIMI.4 cells and reduced 17- to > 29-fold in R1.1 and S1A.TB cells relative to human MT-4 T-cells (Fig. 6A). To establish a signature characteristic for individual T-cell lines in respect to their ability to allow HIV-1 integration, we calculated the Relative-Integration-Frequency defined as the number of integrants on day 7 p.i. (Fig. 6A) relative to the total amount of HIV-1 cDNA quantified on day 1 p.i. (Fig. 3A). Based on results from 2–5 independent experiments, infected TIMI.4 cells were grossly impaired in their Relative-Integration-Frequency value, > 305-fold compared to infected MT-4 cells (Fig. 6B). The two other mouse T-cell lines, R1.1 and S1A.TB, displayed an intermediate phenotype with Relative-Integration-Frequency values 18- to 54-fold lower compared to the human reference. Thus, integration into the host cell genome appears to be the paramount barrier imposed by mouse T-cells in the early phase of HIV-1 replication.

To gain insight into the quantitative relationship between vector integrants and transgene expression, the ratio of the percentage of T-cells expressing GFP from the CMV IE promoter and levels of integrants per cell was calculated, for convenience termed Transgene-Expression-per-Integrant. For the residual integrants in mouse R1.1 T-cells (Fig. 6A, B), reporter gene expression was efficient, resulting in a Transgene-Expression-per-Integrant value in a range similar to human MT-4 cells (Fig. 6C). In contrast, the Transgene-Expression-per-Integrant value in infected S1A.TB cells was 34-fold lower than R1.1 (Fig. 6C), indicating that integrants in this mouse T-cell line frequently do not result in a detectable gene expression. This suggests that at least in some T-cell lines a defect in transgene expression from residual integrants may impose an additional peri-integrational limitation in the mouse species.

Finally, we sought to confirm the key findings also in the context of an infection with a near-full length HIV-1_NL4-3. This replication-deficient, envelope-deleted molecular proviral clone carries an egfp reporter gene within the nef locus driven by the 5'-LTR, and virions were also VSV-G pseudotyped and loaded with BlaM-Vpr during production. Following infection with this virus, the Relative Post-Entry Efficiency in TIMI.4 and R1.1 mouse T-cells was 62- to 65-fold lower compared to human MT-4 (Fig. 7C) or Jurkat T-cells (data not shown), and thus in a range similar to that seen for the lentiviral vector (55- to 235-fold; Fig. 2C), despite the cyclin T1-dependence of gene expression. Of note, the post-entry restriction was not overcome at an MOI > 5, determined by saturating virion fusion conditions in the respective mouse T-cell lines (data not shown). Importantly, relative to T-cells from a human donor, a pool of primary T-cells from BALB/c mice also displayed a severely reduced Relative-Post-Entry-Efficiency following infection with VSV-G HIV-1_NL4-3 GFP (55-fold; Fig. 7F).

Furthermore, in cell aliquots removed on day 1 and 2 p.i. from the experiment shown in Figs. 7A–C, levels of total HIV-1 cDNA were quite similar in the cross-species comparison with infected mouse T-cell lines harbouring 1.4- to 4-fold higher levels than human MT-4 T-cells (Fig. 8A, upper panel). Most notably, levels of episomal 2-LTR circles were found to be drastically elevated in TIMI.4 T-cells (51- to 245-fold) and modestly enhanced in R1.1 T-cells (5- to 7-fold) relative to the human counterpart (Fig. 8A, lower panel). Following normalization with levels of de novo synthesized HIV-1 cDNA, 2-LTR circle levels were statistically indistinguishable for R1.1 and MT-4 T-cells. For reasons unknown, the DNase-resistant, production-related plasmid contamination in stocks of VSV-G HIV-1_NL4-3 GFP was typically 2–3 orders of magnitude higher compared to VSV-G HIV-CMV-GFP, and, as a consequence, integration could not be reliably quantified for the former virus in T-cells from either species (data not shown).

In an attempt to circumvent this technical limitation, we reasoned that the reduced ability of mouse T-cells to form proviruses should be reflected in a loss of cell-associated HIV-1 cDNA during prolonged passage. Consequently, cells were challenged with the replication-defective VSV-G HIV-1_NL4-3GFP carrying BlaM-Vpr and continuously cultivated for four weeks. First, levels of virus entry (6 h p.i.; data not shown), de novo synthesized HIV-1 cDNA (Fig. 8B), and 2-LTR circles (data not shown) were fairly comparable on 1 day p.i., confirming the intact early phase of the replication cycle and successful nuclear import of the preintegration complex. In contrast, a massive loss of cell-associated HIV-1 cDNA was observed in all three mouse T-cell lines at day 14 p.i. (Fig. 8B; 102- to > 2317-fold reduction) with levels in passaged TIMI.4 cells at background. On day 28 p.i., levels of amplified HIV-1 cDNA were further reduced in R1.1 and S1A.TB cells. Collectively, these results provide additional evidence that HIV-1 cDNA, despite nuclear localization, cannot efficiently integrate into the genome of mouse T-cell lines, resulting in a largely abortive infection.

Female BALB/c mice were obtained from Charles River Laboratories (Sulzfeld, Germany). Animals were kept in the central animal facility of the University of Heidelberg.

The mouse T lymphoma cell lines S1A.TB.4.8.2 (S1A.TB) (TIB-27; BALB/c strain), TIMI.4 (TIB-37; C57BL/6 strain) and R1.1 (TIB-42; C58.J strain) were obtained from the American Type Culture Collection and cultivated in Dulbecco's modified Eagle medium supplemented with 10% fetal calf serum or horse serum (R1.1), 1% penicillin-streptomycin, and 1% L-glutamine (all from GIBCO). Human T-cell lines MT-4 and Jurkat, the adherent cell lines TZM-bl, 293T, HeLa^int, and mouse fibroblast cell line NIH3T3 cells (CRL-1658; NIH/Swiss strain) were cultivated as reported 67535455. Primary T-cells from Ficoll-gradient purified peripheral blood mononuclear cells from healthy human donors were activated by phytohaemagglutinin-P (1 μg/ml) and human recombinant interleukin-2 (IL-2) (20 nM, BioMol, Germany) as reported 25355356. Primary T-cells from BALB/c mice were prepared from spleens, which had been removed aseptically from sacrificed animals. Single-cell suspensions were prepared by pushing tissue pieces through a nylon mesh screen (70-μm-pore-size nylon, Becton Dickinson). The cell suspension was washed once with PBS and erythrocytes were lysed for 3 min with ACK lysis buffer, followed by another PBS wash. Single-cell splenocyte suspensions were cultivated in supplemented medium and activated with Concanavalin A (1 μg/ml) and IL-2 as reported 56, yielding proliferating cultures containing ~90% CD3-positive T-cells.

Cellular and viral DNAs were extracted using Qiagen DNeasy Tissue Kits^®. Samples for quantitative HIV-1 integration PCR were extracted according to the method of Hirt 5758. Briefly, 5 × 10⁵ cells were pelleted and carefully resuspended in 160 μl HIRT Solution I. Next, 10 μl Proteinase K (20 mg/ml) and 200 μl HIRT Solution II were added and the cells were incubated for 30 min at 56°C. Then, 100 μl of 5 M sodium chloride were added to the reaction prior to overnight incubation at 4°C. Afterwards, chromosomal DNA was pelleted and the supernatant removed. The HIRT pellets were then extracted with phenol-chloroform-isoamyl alcohol (25:24:1), ethanol precipitated in the presence of glycogen, and washed in 70% ethanol prior to resuspending the DNA pellet in elution buffer (Qiagen).

VSV-G pseudotyped stocks of a three-plasmid HIV-1-based GFP vector 59 (HIV-CMV-GFP) or of HIV-1_NL4-3 E^- GFP 60, carrying BlaM-Vpr, were generated in 293 T cells by calcium phosphate DNA precipitation in principle as described 67. Virus stocks were characterized for p24 CA concentration by antigen enzyme-linked immunosorbent assay and for infectious titer on TZM-bl cells, respectively 53. Virus stocks were treated with DNase (Turbo DNase, Ambion, Dresden, Germany) for 1 h at 37°C (5 IU per 10 μl of concentrated virus stock) to reduce the degree of plasmid contamination prior to challenge of target cells.

Target T-cells (5 × 10⁶) were left untreated or pretreated with efavirenz (100 μM, Bristol-Myers Squibb, Uxbridge, UK) for 1 h, or a hybridoma supernatant containing anti-VSV-G monoclonal antibody I1 24 for 15 min at 37°C. Subsequently, cells were infected with BlaM-Vpr-loaded VSV-G HIV-1 for 6 h, washed and stained overnight with CCF2/AM dye. Fusion was monitored with a three-laser BD FACSAria Cell Sorting System (Becton Dickinson, San Jose, CA) as reported 6222325. On day 3 p.i., the percentage of GFP-positive cells was measured in the identical cultures on a FACSCalibur using BD CellQuest Pro 4.0.2 Software (BD Pharmingen).

Levels of total HIV-1 cDNA and 2-LTR circles in infected cell lines were measured by quantitative PCR using the ABI 7500 sequence detection system (Applied Biosystems, Foster City, CA) as reported in detail 25. For PCR standard curves, dilutions of pHIV-1_NL4-3 and pU3U5, respectively, covering 5 logs were used, supplemented with DNA extracted from uninfected cells. Results for HIV-1 cDNA species were normalized to the amount of cellular DNA, which was quantified in a parallel amplification of the mouse GAPDH gene or human RNaseP gene (Applied Biosystems), respectively. Genomic standards were derived from dilutions of genomic DNA extracted from uninfected cells. All samples were run in duplicate. Data analysis was performed with the 7500 System Software (Applied Biosystems). Of note, for all HIV-1 cDNA analyses values obtained for parallel infections of efavirenz-treated cultures were subtracted from values of untreated cultures to account for residual proviral plasmid contamination in DNAse-treated virus inocula. For total HIV-1 cDNA analyses shown in this manuscript for VSV-G HIV-CMV-GFP (Figs. 1, 2, 3, 4, 5, 6) or VSV-G HIV-1_NL4-3 GFP (Figs. 7, 8), the subtracted signals ranged between 0.07 – 0.65% or 0.3 – 4.6% of the untreated infection, respectively. However, in several experiments with other VSV-G HIV-1_NL4-3 GFP stocks this background signal reached levels above 60% (despite DNAse treatment), not allowing a reliable quantification of total HIV-1 cDNA or integrants.

The strategy to quantify HIV-1 integrants in the mouse genome by real-time PCR and the generation of a mouse integration standard cell line, NIH3T3^int cells, was adapted from the protocol established for the rat species 6. Briefly, HIV-1 integrants were amplified in the first reaction by one primer annealing to the LTR (primer #1521 28), which contains a lambda-phage heel sequence at the 5'-end, and by two outward-facing primers that target the highly redundant consensus sequence within the mouse B1 gene 40 (primers #2194, 5'-ACAGCCAGGGCTACACAGAG-3' and #2231, 5'-CCTCCCAAGTGCTGGGATTAAAG-3'). The first-round amplicon was then amplified in a second reaction with a lambda-specific primer (primer #1522 28) and an LTR primer (primer #1523, 6) and an LTR-specific probe (probe #1524, 6). The second-round cycling conditions were identical to those used to determine the total amounts of HIV-1 cDNA and 2-LTR circles 25. For the detection of HIV-1 integrants in infected human cells, the conditions and reagents were identical to those reported 6. For each integration PCR, a specificity control reaction was run in parallel in which the cellular primer pair was omitted during the first round reaction and this value was subtracted from the total signal (not more than 20% of the absolute signal). For integration standard curves, dilutions of genomic DNA derived from NIH3T3^int and HeLa^int cells covering 3.1 logs were used (see also Fig. 4B). Criteria for standard validity included the slope, which indicates the efficiency of the PCR reaction within a defined copy range using diluted cellular DNA carrying defined amounts of integrated provirus (ideally 3.32, corresponding to a 100% efficient reaction). In addition, only integration PCR assays in which the standard reached a R² coefficient > 0.9 were analyzed. The lowest detection standard of the PCR was 0.04 and 0.12 HIV-1 integrants per ng DNA for integrated provirus in mouse and human genomic DNA, respectively.