Introduction

New and re-emerging infectious diseases represent a serious medical problem that demands the development of general approaches for the rapid isolation of efficient pathogen inhibitors. We have previously described a technology for isolating efficient gene suppressors by functional expression selection of genetic suppressor elements (GSEs) from random fragment libraries (RFL) prepared from a gene or genome of interest.1,2 This technology is based on the assumption that a gene contains segments that can inhibit its own function when the isolated segment is expressed in a cell. GSEs can exert their effect as antisense RNAs, structural RNAs, or peptides acting as dominant-negative mutants. In general, the GSE approach involves (1) generation of a representative RFL from a gene or genome of interest in an expression vector; (2) transfer of the library into target cells; (3) selection for a desired phenotype; (4) identification and characterization of the GSEs responsible for the phenotype. The major advantage of this technology is that it does not require any previous knowledge of which gene(s) or protein(s) will represent the best targets for the development of efficient genetic inhibitors or which type of inhibitors (antisense RNAs, RNA decoys or transdominant mutants) will be most potent in suppression of a specific gene. To date, the GSE technology has been successfully used to isolate efficient inhibitors in bacterial and mammalian cells.1,2,3,4,5,6,7,8

The pathogen-derived resistance principle or ‘intracellular immunization’, based on the hypothesis that cells can be made resistant to intracellular pathogens by expression of mutant proteins, decoy RNAs or anti-sense RNAs, has been suggested as applicable to the treatment of HIV infection.9,10 A number of genetic inhibitors against HIV-1 have been designed rationally on the basis of the current knowledge of the viral life cycle, structure of the viral genome and functional organization of the viral protein domains. These include decoy RNA of TAR or RRE, ribozymes, antisense RNA of gag, tat, rev, env and transdominant mutants including Rev, Tat, Gag, Env and protease.11,12,13,14 These inhibitors have been suggested for use in the gene therapy of AIDS. The existence of these many inhibitors indicates that a variety of genetic inhibitors against HIV-1 can be isolated. Therefore, HIV-1 represented an attractive mammalian viral system for application of the GSE technology for the isolation of new genetic inhibitors or confirmation of existing inhibitors.

The goal of this study was to develop functional selection approaches for isolating genetic inhibitors of HIV-1. Two flow cytometry-based selection procedures were developed for identifying HIV-1 suppressors that target productive and latent stages of the viral life cycle. Multiple GSEs capable of interfering with both stages were isolated. These elements are potential candidates for development in the gene therapy of AIDS and also indicate functionally important regions of the viral genes that could be targeted by small molecule drugs.

Results

Selection approaches for HIV GSEs

The HIV-1 life cycle consists of two distinct stages, productive or acute infection and latency. An ideal anti-HIV inhibitor should be effective against both stages. Accordingly, two selection approaches were developed. The first approach, designed to inhibit virus induction in latently infected cells, was based on the unique properties of OM10.1 cells.15 The OM10.1 cell line is a chronically infected promyelocytic clone of HL-60 which contains a single copy of the HIV-1BRU isolate. Unlike other chronically infected tumor cell lines, OM10.1 cells remains approximately 99% CD4-positive until induction of the latent HIV. The addition of an inducer (eg TNF-α) causes the loss of CD4 in approximately 90% of the cells due to intracellular complexing with gp160/120 and viral replication.9 We hypothesized that expression of an anti-HIV GSE capable of interfering with induction would result in retention of surface CD4. Such cells could be separated from the CD4-negative population by fluorescent activated cell sorting (FACS), allowing enrichment of the GSEs responsible for the inhibition of viral induction (Figure 1a and b).

Figure 1
figure 1

Approaches for the selection of GSEs against HIV. (a) Scheme for OM10.1 selection of GSEs against HIV-1. (b) Representative flow cytometric profile for sorting of TNF-α induced OM10.1 cells containing vector or random fragment library (RFL). PE-conjugated isotype control monoclonal antibody (IgG1) was used as a negative control for gating. Uninduced RFL containing cells stained for surface CD4 are shown as a positive control. Surface CD4 expression of the induced vector population represents the selection phenotype background. Bar in library profile indicates the population of CD4-positive cells that were sorted. In the profiles shown, vector and library are 10% and 16% CD4-positive, respectively. The uninduced library population is 99% CD4-positive. (c) Scheme for productive infection selection of GSEs against HIV-1. (d) Representative flow cytometric profile of surface CD4 and intracellular p24 expression of CEM-ss cells containing vector or RFL 9 days after infection with HIV-1IIIB. Mock-infected profile shows an uninfected library population. Box indicates the CD4-positive, p24-negative cells that were sorted in the library population. Box is also shown in the mock-infected and vector profile for comparative purposes.

Full size image

The second selection scheme was designed to select GSEs capable of inhibiting productive HIV infection. Replication of HIV in susceptible cells is associated with accumulation of intracellular p24, concomitant with down-modulation of surface CD4. We assumed that expression of GSEs capable of interfering with productive infection should result in enrichment of protected cells displaying the CD4+, p24− phenotype. Such cells can be separated by FACS from the infected (p24+, CD4−) population (Figure 1c and d).

RFL libraries

It is well documented that HIV-1 has a very high mutation rate and substantial sequence diversity.16,17 Thus, regions that are conserved among different isolates may represent the most desirable targets for inhibition and provided the rationale for using various HIV-1 isolates in constructing the RFLs, in selections and in challenge of CEM-ss cells (Table 1). RFLs from various HIV-1 isolates were cloned into the retroviral vectors LXSN or LXSNgfr. A RFL from isolates HIV-1BRU and HIV-1SF2 in LXSN was used in the first OM10.1 selection (X1, approximately 100000 recombinant clones). A RFL from HIV-1HXB2 in LXSN was used in the productive infection selection (PI, approximately 100000 recombinant clones). Another RFL from HIV-1HXB2 (approximately 80000 recombinant clones) in LXSNgfr was used in two independent experiments in OM10.1 cells (X2 and X3).

GSE clusters

Each RFL was transferred into the target cells (OM10.1 or CEM-ss) and two rounds of selection were performed (Figure 1). The reproducibility of the system was demonstrated by independent transfers and selections of the same RFL (X2 and X3). After the second round of selection in OM10.1 cells, 40–50% of the elements were from a few short regions of the HIV-1 genome, indicating selection for these sequences. Sequence comparison of individual elements enriched in all selections revealed seven clusters of overlapping sequences, five in the sense orientation and two in the antisense orientation (Table 2, Figure 2). Two clusters are from areas of the HIV-1 genome where two or more viral genes overlap (vpr/tat, rev/tat). Some elements were nearly identical to those isolated in the productive infection selection. Thus, three clusters (the nef sense cluster the rev/tat sense cluster and the vpr/tat antisense cluster) were identified in all libraries and selections. We also identified a cluster of sense-oriented elements from the RT gene, which was found only in the productive infection selection.

Figure 2
figure 2

Positions of the isolated GSEs on the HIV-1BRU genome. Scale of the HIV-1 genome is in kilobases. Locations of the GSEs are shown below the scale. Shaded boxes indicate GSEs in the sense orientation and arrows indicate GSEs in the antisense orientation.

Full size image

Anti-HIV activity of individual GSEs

Putative GSEs from the first OM10.1 selection (X1) were transferred into OM10.1 cells and bulk populations were analyzed for their ability to prevent HIV-1 induction by TNF-α. In all cases, the growth rates and viability of the cells transduced with putative GSEs were undistinguishable from that of the controls (data not shown). Similarly, no differences were detected in CD4 levels before induction as measured by flow cytometry (data not shown). After induction, negative controls (OM10.1 cells and OM10.1 cells with the LXSN vector) revealed background levels of 10% CD4-positive cells (Figure 3). Two additional negative controls containing anonymous DNA inserts from plasmid vectors (34 and 220) were 8–11% CD4 positive. However, individual GSEs showed a constant and reproducible inhibitory effect by allowing 18–26% of the transduced cells to retain CD4 upon induction with TNF-α (Figure 3). In contrast, OM10.1 cells containing the transdominant mutant, RevM10,18,19,20 were not effective at inhibiting induction of virus from latency.

Figure 3
figure 3

CD4 retention levels of OM10.1 cells containing GSEs after TNF-α induction. OM10.1 cells containing GSEs in LXSN from the X1 selection were induced and analyzed as described in Materials and methods. Results are presented as the percentage of cells retaining surface CD4 expression 24 h after induction with TNF-α. Negative controls include OM10.1 cells, LXSN, 34 and 220. RevM10 is a Rev transdominant mutant. Results are of a representative experiment.

Full size image

To determine whether GSEs isolated in the OM10.1 selections (X1, X2 and X3) would protect naive cells from HIV-1 infection, individual elements were transferred into the human T cell line, CEM-ss. Once again, no differences in growth rates or viability were seen in populations containing controls and GSEs (data not shown). Also, before infection, CD4 levels of cells containing the GSEs and controls were analyzed by flow cytometry and found to be 99% positive and of equal density (data not shown). Upon infection of bulk populations with HIV-1SF2, a significant delay was observed in the development of the productive infection as determined by intracellular p24 staining (Figure 4a and b). GSEs isolated in productive infection selection had sequences overlapping with those from the OM10.1 selection and had similar effects upon viral challenge (data not shown). The reverse transcriptase (RT) cluster of elements isolated in productive infection selection showed inhibitory effects when challenged with HIV-1IIIB (Figure 4c). In all infections, the GSEs demonstrated similar or better effects than the transdominant mutant, RevM10, a known inhibitor of HIV-1 replication in T cell lines and primary cells.18,19,20 Thus, GSEs representing all clusters were found to be functionally active.

Figure 4
figure 4

Time-course of viral infection of CEM-ss cells containing GSEs. Cells were infected with HIV-1 and analyzed for intracellular p24 as described in the Materials and methods. Results are presented as the percentage of p24-positive cells at specified days after infection. Negative controls are 34 or LXSNgfr. RevM10 is the positive control. All results are of a representative experiment. (a) Infection of GSEs from the X1 selection. CEM-ss cells containing IGX-009 (nef sense), IGX-230 (rev/tat sense) and controls were infected with a TCID50 of 1000 of HIV-1SF2. (b) Infection of GSEs from the X2 and X3 selection. CEM-ss cells containing IGX-117 (vpr/tat sense), IGX-201 (RRE antisense) and controls were infected with a TCID50 of 500 of HIV-1SF2. (c) Infection of GSE from the productive infection (PI) selection. CEM-ss cells containing IGX-104 (RT sense) and controls were infected with a TCID50 of 3000 of HIV-1IIIB.

Full size image

Analysis of rev/tat GSE constructs

One region of similarity between the GSEs and previously reported genetic inhibitors is in the rev and tat regions. A cluster of sense-oriented GSEs were isolated from the second exon of the rev and tat genes, a sequence contained in both Tat and Rev transdominants.21,22 This sequence also overlaps with the sequence of the envelope. A GSE from the rev/tat sense cluster (IGX-230) showed a comparable profile in productive infection as the transdominant, RevM10 (Figure 4a). Conversely, induction from latency of OM10.1 cells showed that IGX-230 inhibited induction, while RevM10 had no effect (Figure 3). This suggested that the rev/tat GSE had a different mechanism of action than RevM10. IGX-230 was constructed using an adaptor with three start codons representing all three open reading frames, therefore three constructs were made using a single start codon for each of the three reading frames. The three IGX-230 sense- oriented constructs (ORFs for Rev, Tat and Env) showed similar effects in their ability to protect against HIV challenge (Figure 5) and similar to that previously seen for IGX-230. An antisense construct showed no effect (data not shown). In addition, IGX-103 from this cluster has also been shown to inhibit infection and lacks a 5′ adaptor and consequently a translation start codon (data not shown). This data would suggest that the GSEs from this cluster probably exert their effect as structural RNAs, with a different mechanism of inhibition than the transdominants.

Figure 5
figure 5

Time course of viral infection of CEM-ss cells containing rev/tat constructs. The 5′ primers were constructed to have only one start codon and represent the three possible reading frames, Tat (A), Env (B), Rev (C). The 5′ primers were the following: 5′-GGAATTCAAGCTTG CCGCCACCATGGGCCCGACGGAATCGAA-3′; 5′-GGAATTCAAG CTTGCCGCCACCATGGACGGGCCCGACGGAATCGAA-3′; 5′-GG AATTCAAGCTTGCCGCCACCATGGACGGCTGGGCCCGACGGAA TCGA-3′ for A, B and C, respectively. All constructs used the same 3′ primer (5′-GGATCCATCGATTCACTCACTCA-3′). Cells were infected with a TCID50 of 1000 of HIV-1SF2 and analyzed for intracellular p24 as described in Materials and methods. Results are presented as the percentage of p24-positive cells at specified days after infection. Negative control is 34. All results are of a representative experiment.

Full size image

Discussion

We report here the development of a general approach to identify efficient genetic inhibitors of a viral pathogen (HIV-1). This represents the first mammalian virus system in which the GSE technology was applied, in addition to the first use of flow cytometry for the selection of GSEs. Two selection procedures that target the productive and latent stages of the viral life cycle were used to isolate GSEs to HIV-1. Using independent selections with RFLs from various HIV-1 isolates, nearly identical elements from several highly conserved regions of the viral genome were identified. For example, the rev/tat sense GSE was isolated from the HIV-1BRU/HIV-1SF2 and the HIV-1HXB2 RFLs in selections with OM10.1 cells (which contain HIV-1BRU) and CEM-ss cells productively infected with HIV-1IIIB. The fact that elements with overlapping sequences from narrowly defined regions of the HIV-1 genome were isolated in different selection systems using different HIV-1 isolates strongly implies that these elements are from functionally important and conserved regions of the viral genome.

Given the limited number of viral genes encoded by HIV-1, it was not unexpected that we identified GSEs from the same regions targeted by others (rev, RRE, tat, gag), in addition to clusters of GSEs from previously untargeted regions (RT and nef). Direct comparisons revealed that the GSEs worked as well or better than a rationally designed transdominant mutant, RevM10, in inhibiting productive infection. However, the GSEs were able to inhibit induction from latency in OM10.1 cells, while RevM10 was ineffective in this regard. The rev/tat cluster has sequences that are encompassed by RevM10 and displayed similar effects in productive infection, however, the lack of effectiveness of RevM10 in the OM10.1 assay suggested that the rev/tat GSE has a different mechanism of action to RevM10. This points to the fact that even though the GSEs may overlap pre- viously reported inhibitors, the mechanism may be quite different.

With the exception of RevM10, it is difficult to compare the GSEs isolated here to those genetic inhibitors reported by others. Numerous variables can drastically effect the extent of the inhibition and necessitate a direct comparison of genetic inhibitors before deciding whether one is more effective than another. Expression levels of genetic inhibitors have been shown to effect HIV suppression.23,24,25,26,27,28 Clonal populations offer the greatest resistance to HIV-1, however, this inhibition can vary widely from clone to clone.18,19,29,30 Cells in bulk populations are unable to inhibit infection completely, but give a more representative picture of what is likely to occur in the clinical application.29 The amount of virus added has also been reported to effect the outcome.28 Different assay systems also can make comparisons difficult. The intracellular p24 assay has been reported to be more informative than the p24 ELISA due to its ability to distinguish between infected and uninfected cells as opposed to measurement of average p24 secretion.29

In the case of HIV-1, one of the best understood human viruses, a number of highly specific and effective antiviral strategies has been developed. These inhibitors are rationally designed with a potential mechanism of action in mind. This assumes that the mechanism is well understood and that the inhibitor will work in the manner envisioned. In the case of the GSE technology, efficacy of an inhibitor is known, but the mode of action is not initially known and may require extensive experimentation. The advantage of GSEs is that selection does not require any information on mechanism. This is also a disadvantage when a functional GSE is found. The determination of the mechanisms of action of GSEs may reveal novel functions in the viral life cycle, especially for genes whose functions are not well understood. The rev/tat GSE is suggested to act as a structural RNA. Other GSEs reported here are undergoing studies to determine potential mechanisms.

The outcome of GSE selections depends on the following parameters: (1) representation of gene fragments in the RFL; (2) efficiency of transfer into the target cells; and (3) levels of GSE expression. Based on our experience, a 10-kb viral genome can be sufficiently represented by an RFL of approximately 100000 recombinant clones with DNA fragments averaging 100 to 300 bases.1,2 To monitor the efficiency of transfer of RFLs into target cells, we designed retroviral vectors expressing nerve growth factor receptor (NGFR) and used flow cytometry to quantify each transfer. It is expected that the outcome of selection will depend in part on the efficiency of expression of a GSE. We monitored GSE expression by replacing the GSE with a reporter gene, green fluorescent protein (gfp). It was found that expression of gfp in NGFR-positive cells varied from 20 to 40% in OM10.1 cells and 50–60% in CEM-ss cells, suggesting that the modest inhibitory effects of the GSEs were due to low expression (data not shown). Other investigators have also reported that suboptimal expression levels of genetic inhibitors limit the effectiveness of HIV suppression.23,24,25,26,27,28 Vectors allowing increased expression may result in better inhibitory effects.

Several characteristic features of the GSE technology make it applicable to a broad spectrum of phenotypes. First, any phenotype associated with expression of a surface or intracellular protein can be used for functional selection of genetic inhibitors by FACS. Second, GSEs can be isolated despite high backgrounds and low expression levels of the GSE by iterative selection. Thus, the GSE technology represents a powerful tool of functional genomics that allows the identification of functionally important regions of known or unknown genes from the genome. Similar approaches are applicable to pathogens with unknown or uncharacterized genomes as long as a selection system can be devised. We are currently extending the use of the developed approaches to identify human cellular genes supporting the HIV life cycle.

Materials and methods

Plasmids

The retroviral vector, LXSN31 was obtained from Dr Dusty Miller (Fred Hutchinson Cancer Center, Seattle, WA, USA). LXSNgfr was constructed by replacing the neomycin resistance gene of LXSN with a truncated low affinity nerve growth factor (NGFR) gene. RevM1023 from Dr Donald Kohn (University of Southern California School of Medicine, Los Angeles, CA, USA) was cloned into the LXSNgfr vector. The following plasmids were the source of the HIV-1 genomic DNA: pBENN732 derived from 5′ half of HIV-1BRU (AIDS Research and Reference Reagent Program (ARRRP), Rockville, MD, USA), 9B/6R derived from the 3′ half of HIV-1SF2 (a gift from Dr C Cheng-Mayer), HIV-1HXB2 genomic DNA.33

Generation of random fragment library (RFL)

Construction of the RFLs in LXSN or LXSNgfr was as described.34 Two HIV-derived RFLs were produced. The first used HIV-1BRU- and HIV-1SF2-derived genomes as the DNA source, while the second was made from HIV-1HXB2 genomic DNA. The adaptor sequences for the first library contained three ATGs (5′-GAATTCAAGCTT ATGGATGGATG-3′). For the second library, only one ATG was used (5′-GATTCAGCTTGCCGCACC ATG GCT-3′). Both libraries used the same 3′ primer, containing three stop codons (5′-GGATCCATCGATTCACTCA CTCA-3′).

Cell lines and HIV-1 strains

Cell lines OM10.1 (ATCC, Rockville, MD, USA) and CEM-ss (ARRRP) were maintained in RPMI 1640 medium supplemented with 10% fetal bovine serum (FBS) at 37°C and 5% CO2. The amphotropic packaging cell line BING, obtained from Dr W Pear (Rockefeller University, New York), were maintained in DMEM medium supplemented with 10% FBS at 37°C and 5% CO2. Persistently HIV-infected cells, HUT78/HIV-1SF2 and H9/HIV-1IIIB (ARRRP), were used to prepare HIV-1SF2 and HIV-1IIIB viral stocks. 50% tissue culture infectious doses (TCID50) of these culture supernatants were determined on HUT 78 cells.35

Transfection and transduction of OM10.1 and CEM cells

BING packaging cells were transfected with plasmid DNA using a standard calcium phosphate method.36 Two methods were used for transduction. For LXSN-based libraries and GSEs, the packaging cells were cocultivated with the target cells (OM10.1 or CEM-ss) for 2–3 h. Three cocultivations were used at 24, 48 and 72 h after transfection. Cells were then grown under neomycin selection for 2 weeks. The surviving cells were purified using Ficoll and grown in neomycin-free media before any further manipulations. A different method was used for LXSNgfr-based libraries and GSEs. Filtered retroviral supernatants from BING cells (24- and 48-h virus) were used to infect the target cells by centrifugation at 1200 g for 90 min. One week later, cells were stained with a NGFR monoclonal antibody (mAb) (20.4, ATCC) and the transduced cells represented by the NGFR-positive population were sorted using a FACS Vantage (Becton Dickinson) equipped with a 488 nm argon laser. Cells were recultured in media and checked for NGFR before use.

Cell staining

For CD4 or NGFR staining, cells were washed twice (5% FBS, 1.5% BSA, 0.0055% EDTA), blocked with 5% normal mouse serum, followed by the addition of phycoerythrin (PE)-conjugated CD4 mAbs Q4120 (Sigma, St Louis, MO, USA) or L120 (a gift from Becton Dickinson Immunocytometry Systems, San Jose, CA, USA), or PE-conjugated NGFR mAb 20.4. Following a 30-min incubation, the cells were washed twice and analyzed or sorted by flow cytometry.

Intracellular p24 analysis was based on previously described methods.29,37 Cells were stained for CD4 (L120) as above, resuspended in 100 μl PBS, and PermeaFix (Ortho Diagnostics, Raritan, NJ, USA) added (1 ml). After a 40-min incubation at room temperature, the cells were pelleted, blocked and incubated for 30 min with FITC-conjugated anti-p24 (KC-57; Coulter, Hialeah, FL, USA). Cells were then washed twice and analyzed or sorted by flow cytometry.

Selection of GSEs in OM10.1 cells

The transduced populations of the OM10.1 cells (GSE library or insert-free vector) were washed once with PBS, then induced with 0.1 ng/ml of TNF-α (Boehringer Mannheim, Indianapolis, IN, USA) in RPMI supplemented with 10% FBS at a density 5 × 105 cells/ml. After 24 h, cells were stained with the PE-conjugated CD4 monoclonal antibody (Q4120). Propidium iodide was added to a final concentration of 10 μg/ml immediately before sorting. The propidium iodide negative, CD4-positive population was sorted by flow cytometry. The cells were lysed and the genomic DNA was purified.36 Inserts were amplified by PCR using vector-derived primers, the mixture was digested with BamHI and EcoRI, column purified, ligated to BamHI/EcoRI digested vector and transformed into E. coli. Purified DNA from transformants were either used as a pool for subsequent rounds of selection and/or individually isolated and sequenced using the ALF DNA Sequencer (Pharmacia LKB, Piscataway, NJ, USA).

Productive infection selection of GSEs

Transduced CEM-ss cells (GSE library or insert-free vector) were infected with HIV-1IIIB at a TCID50 of 3000 per 106 cells. At 9 days after infection, cells were stained for CD4 and p24 and the p24-negative, CD4-positive population was sorted. Genomic DNA purification, insert amplification, subcloning and sequencing were done as described above. Two rounds of selection were performed.

Analysis of GSEs in OM10.1 cells

Bulk populations of OM10.1 cells containing GSEs were induced with TNF-α as above. After 24 h at 37°C, the cells were stained with the PE-conjugated CD4 monoclonal antibody (Q4120) and CD4 retention levels were analyzed by flow cytometry.

Analysis of GSEs in CEM-ss cells

Bulk populations of CEM cells (1 × 106) containing GSEs were infected in a 1 ml volume for 2 h with a TCID50 of 500 or 1000 of HIV-1SF2 or 3000 of HIV-1IIIB in the presence of 4 μg/ml of polybrene. Cells were washed once with PBS and resuspended in 5 ml (2 × 105 cells/ml) of media. Samples were withdrawn every 3 to 4 days for intracellular p24 analysis and cells were passaged to 2 × 105 cells/ml in fresh medium.

Acknowledgements

We are indebted to Drs Clayton Smith and Eli Gilboa for their suggestions, advice and continued interest in this work. We thank Drs Mark Furth and Jay Levy for their support and Drs Eugene Mechetner and Bey-Dih Chang for their contributions at the early stages of this work. We also thank Dr Donald Kohn for his kind gift of the RevM10 transdominant mutant. The following reagents were obtained through the AIDS Research and Reference Reagent program, Division of AIDS, NIAID, NIH: H9/HTLV-IIIB NIH 1983 from Dr Robert Gallo, HUT78/HIV-1SF2 from Dr Jay Levy, CEM-ss from Dr Peter L Nara and pBENN7 from Dr Malcolm Martin. This work was supported in part by NIH SBIR grant R44 AI37381–03, National Cancer Institute grant RO1 CA56736 and a grant from Ingenex Inc.