Development of a nonintegrating Rev-dependent lentiviral vector carrying diphtheria toxin A chain and human TRAF6 to target HIV reservoirs

Abstract

Persistence of human immunodeficiency virus (HIV) despite highly active antiretroviral therapy (HAART) is a lasting challenge to virus eradication. To develop a strategy complementary to HAART, we constructed a series of Rev-dependent lentiviral vectors carrying diphtheria toxin A chain (DT-A) and its attenuated mutants, as well as human tumor necrosis factor receptor-associated factor 6 (TRAF6). Expression of these suicide genes following delivery through viral particles is dependent on Rev, which exists only in infected cells. Among these toxins, DT-A has been known to trigger cell death with as little as a single molecule, whereas two of the attenuated mutants in this study, DT-A(176) and DT-A(Delta N), were well tolerated by cells at low levels. TRAF6 induced apoptosis only with persistent overexpression. Thus, these suicide genes, which induce cell death at different expression levels, offer a balance between efficacy and safety. To minimize possible mutagenesis introduced by retroviral integration in nontarget cells, we further developed a nonintegrating Rev-dependent (NIRD) lentiviral vector to deliver these genes. In addition, we constructed a DT-A-resistant human cell line by introducing a human elongation factor 2 mutant into HEK293T cells. This allowed us to manufacture the first high-titer NIRD lentiviral particles carrying DT-A to target HIV-positive cells.

Figure 1. Specificity of the Rev-dependent lentiviral vectors in mediating HIV-dependent gene expression

(A) Schematic representation of the HIV-1 genome, a Rev-dependent lentiviral construct (pNL-GFP-RRE-SA), and the HIV-1 helper construct, pCMVΔR8.2, in which both the viral package signal (Δψ) and the envelope gene (Env) were deleted. Shown are the HIV-1 5' LTR, packaging signal (ψ), splice donors (D1, D4) and acceptors (A5, A7), IRES, RRE, and 3' LTR. (B) Specificity of the Rev-dependent lentiviral vector in HIV-1-positive T cells. CEM-SS cells were not infected (Cell) or infected with NL4-3.HSA.R+E-(VSV-G) (NL4-3.HSA, 1 µg p24 per million cells), a VSV-G pseudotyped HIV-1 strain with the murine heat-stable antigen CD24 (HSA) gene inserted into the nef region that allows HIV-1-positive cells to be monitored by surface staining of HSA. At 24 hours, cells were superinfected with lentivirus vNL-GFP-RRE-SA (1x, m.o.i. 0.2). For comparison, cells were also singly infected with either vNL-GFP-RRE-SA (No HIV infection) or NL4-3.HSA.R+E-(VSV-G) (NL4-3.HSA). At 72 hours, cells were harvested, stained with a PE-labeled rat monoclonal antibody against mouse CD24 (HSA), and then analyzed on a flow cytometer for both HSA and GFP expression. Isotype staining is not shown.

Figure 2. Rev-dependent killing of HIV-positive cells by DT-A, TRAF6, and AnlO

(A) Schematic representation of the Rev-dependent vectors carrying DT-A, TRAF6, and AnlO, and the helper construct, pCMVΔR8.2, that were used to cotransfect HEK293T or HeLa cells. (B) HeLa or HEK293T cells (1 million) were cotransfected with pCMVΔR8.2 (1 µg) plus pNL-DT-GFP-RRE-SA, pNL-TRAF6-GFP-RRE-SA, pNL-AlnO-GFP-RRE-SA, or pNL-GFP-RRE-SA (3 µg). As controls, these Rev-dependent vectors were identically cotransfected with an empty vector, pMSCVneo (1 µg). Cells were also cotransfected with pCMVΔR8.2 without the Rev-dependent vectors (1 µg pCMVΔR8.2 plus 3 µg pMSCVneo). GFP expression was measured at 48 hours post cotransfection by flow cytometry. Propidium Iodide (P.I.) was added to identify viable GFP-expressing cells.

Figure 3. Rev-dependent killing of HIV-positive cells by DT-A mutants

(A) DT-A mutagenesis and the construction of the Rev-dependent vectors carrying these mutants were described in Materials and Methods. HEK293T cells (1 million) were cotransfected with pCMVΔR8.2 (1 µg) and one of the Rev-dependent constructs carrying the DT-A mutants, pNL-DT(E148S)-GFP-RRE-SA, pNL-DT(E148D)-GFP-RRE-(SA), pNL-DT(176)-GFP-RRE-SA, or pNL-DTΔN-GFP-RRE-(SA) (3 µg). Cell killing was monitored by GFP expression at 48 hours post infection using flow cytometry. (B) The same cotransfection experiments were repeated in a DT-A resistant cell line, 5H7.

Figure 4. Testing of the DT-A-resistant HEK293T cells

(A) The human EF-2 mutant (G717R) was introduced into HEK293T cells by retroviral vector transduction. Cells were screened for the EF-2 mutation. Originally, 100 clones were selected, and 5 of them turned GFP-positive when cotransfected with pCMVΔR8.2 (1 µg for 1 million cells) plus the DT-A containing Rev-dependent vector, pNL-DT-GFP-RRE-SA (3 µg for 1 million cells). While the parental HEK293T cells generate 0% GFP-positive cells after the cotransfection, the DT-A resistant clones generate GFP-positive cells at different percentages: 46% in 5H7, 28% in CB2, 24% in AB1, 14% in 4H10, and 9% in 5E12, respectively. (B) To further measure the degree of DT-A resistance, one of the clones, 5H7, was cotransfected with pCMVΔR8.2 plus pNL-DT-GFP-RRE-SA. As a control, the cells were also identically cotransfected with pCMVΔR8.2 plus pNL-DT(R)-GFP-RRE-SA in which the DT-A gene was placed in a reverse orientation to prevent protein expression. The parental HEK293T cells were also identically cotransfected with these constructs. (C) Western blot analysis of both 5H7 and HEK393T cells cotransfected with either pCMVΔR8.2 plus pNL-DT-GFP-RRE-SA (lanes 2 and 4, DT) or pCMVΔR8.2 plus pNL-DT(R)-GFP-RRE-SA [lanes 1 and 3, DT(R)]. Untansfected cells (lanes 5 to 7) and a purified, recombinant DT-A protein (CRM9) (lane 8) were used as controls. A monoclonal antibody against DT was used for Western blot, and this antibody was also reactivated with a cellular protein (10–15KD) that was used as the loading control.

Figure 5. Development of the NIRD vector carrying DT-A and TRAF6

(A) Schematic representation of the Rev-dependent vector carrying luciferase, the non-integrating helper construct, pCMVΔR8.2(D116N), and pHCMV-G expressing VSV-G. (B) To demonstrate HIV-dependent expression of reporter genes from the NIRD vector, viral particles vNL-Luc-RRE-SA(D116N) and vNL-Luc-RRE-SA were generated by cotransfection of HEK293T cells with pCMVΔR8.2(D116N) or pCMVΔR8.2 plus pNL-Luc-RRE-SA plus pHCMV-G, and then used to superinfect an HIV-1-positive T cell line, J1.1 or the uninfected, parental Jurkat T cells (0.2 million cells). Luciferase was measured at 48 hours in J1.1 and Jurkat cells following infection. Both J1.1 and Jurkat were stimulated with 50 ng/ml PMA before infection. (C) To determine the background luciferase present in the HIV-negative Jurkat cells during infection with vNL-Luc-RRE-SA(D116N), cells (0.5 million cells) were pre-treated with azidothymidine (AZT) (50 µM) overnight, and then uninfected (lane 1) or infected with vNL-Luc-RRE-SA(D116N) for 2 hours (lane 2). Cells were washed and immediately lysed for Western blot analysis using a goat polyclonal anti-luciferase antibody. The blot was also probed with a goat polyclonal antibody to GAPDH for loading controls. (D) The background luciferase activity present in the vNL-Luc-RRE-SA(D116N) viral preparation was reduced by purifying the virion through anion exchange (Sartobind Q75) and size-exclusion (Vivaspin 20 and 500) columns. The relative luciferase activities (RLU) present in virion before and after purification were measured (normalized by virion p24). (E) Specific targeting of HIV-1-infected lymphocytes by TRAF6 NIRD vector. Human PBMC (1 million cells) were infected with a replication-competent virus NL4-3.HSA.R+ (10⁴ TCID_50/Rev-CEM). Aliquots of the infected cells were then superinfected at days 1, 4 and 7 post HIV infection with vNL-TRAF6-GFP-RRE-SA(D116N) (5 µg p24). HIV-1-positive cells were measured by surface staining of mouse HSA followed by flow cytometry at day 9 post HIV-1 infection. (F) To measure non-specific killing of cells by TRAF6 NIRD vector, HIV-uninfected cells were infected with only vNL-TRAF6-GFP-RRE-SA(D116N) as described in (E) (panel d, f, h). Following infection at day 0, 3 and 6, cells were analyzed one day later by propidium iodide (P.I.) staining and flow cytometry (panel d, f, h, respectively). As controls, cells were also mock infected with medium (panel a, c, e, and g), or treated with puromycin to induce non-specific killing (panel b).