(A) Primary human CD4+ T cells were stimulated and transduced with an Ad5/F35 vector expressing either the X4-ZFNs (top) or GFP (bottom) at MOIs from 100–1000. Total live cells were counted at different times after stimulation, and compared to an untransduced control. Data is from one of two independent experiments. (B) Cxcr4 disruption was determined four and eight days post treatment with the X4-ZFNs by the surveyor nuclease assay (cel1).

(A) Human CD4+ T cells were treated with the X4-ZFNs or R5-ZFNs expressed by Ad5/F35 vectors or were non-transduced (NTD). Four days later cells were infected with a primary X4 HIV-1 (Bk132), lab-adapted X4 HIV-1 (HxB2), primary R5X4 HIV-1 (R3A) or mock infected. The number of viable cells were measured at various times after stimulation. Cells were re-stimulated on day 13 (arrows). (B) The proportion of disrupted cxcr4 alleles was determined at the indicated times post-stimulation by 454 deep sequencing. The frequency of cxcr4 disruption was relatively constant in the mock-treated cells, but increased dramatically in the presence of HIV-1. (C) FACS analysis using a CXCR4-specific monoclonal antibody was performed at 19 days post infection (24 days post-stimulation). Mock HIV-infected cultures are shown on the left and HIV infected cultures on the right. Data shown is one of three independent experiments.

NSG mice were injected with human CD4+ T cells treated with X4-ZFNs or R5-ZFNs. 28 days post injection, mice were infected with primary X4 HIV-1 (Bk132) or were mock-infected. (A) CD4+ T cell counts were measured every 7–10 days post infection. In the presence of Bk132, treatment with X4-ZFNs conferred protection at 14 d.p.i (p = .05); however, this protection wanes by 34 d.p.i. (p = .88) (B) Cxcr4 disruption frequency was assessed by the surveyor nuclease assay in both peripheral blood (p<.001) and spleen (p<.001). At day 34 post infection, human CD4+ T cells were purified by positive selection prior to analysis to reduce any bias from low frequency contaminating human cells. Only samples with a detectable PCR signal are shown. Disruption frequency did not deviate significantly from the cell innoculum in either the blood or spleen. Data in (A) and (B) were analyzed by a general estimating equation (GEE). (C) HIV-1 Env from X4-ZFN mouse plasma was sequenced revealing a consensus Y302N mutation. To evaluate coreceptor tropism, a representative Env from the X4-ZFN mice and the viral innoculum were pseudotyped and used to infect NP2 cell lines expressing CD4 and either CCR5 or CXCR4. R5 HIV-1 (JRFL), R5X4 HIV-1 (R3A), and X4 HIV-1 (TYBE) controls are shown. Infectivity on NP2/CD4/CXCR4 cells was divided by that on NP2/CD4/CCR5 cells to determine relative coreceptor use. Data is an average of three independent experiments each done in triplicate. Error bars represent standard error.

(A) A CXCR4-specific ZFN pair was generated, comprised of two DNA-binding zinc finger proteins (ZFPs) each fused with a FokI endonuclease monomer. Each ZFP was designed to target 12 bp of cxcr4 sequence (in red), separated by 6 bp (in blue), conferring 24 bp of total specificity. Upon binding of both ZFPs, the FokI domains can dimerize and cleave the double stranded DNA. The subsequent double strand break is then repaired by error prone non-homologous end-joining resulting in various targeted mutations and a non-functional protein product. (B) The most common mutations induced by the X4-ZFNs, as detected by 454 deep sequencing, are indicated with their frequencies among all ZFN-induced lesions. In-frame deletions were preferentially generated with the most common being an 18 bp deletion, referred to as CXCR4Δ18. Frequences were averaged across five independent experiments in the absence of HIV infection.

The most common lesion induced by the X4-ZFNs was an 18 bp deletion, cxcr4Δ18, that results in deletion of the amino acid sequence RFYPND from the second extracellular loop of CXCR4 (see Figure 1B). To determine if CXCR4Δ18 was expressed on the cell surface, a mock, wild type cxcr4, or cxcr4Δ18 plasmid was transiently transfected into 293T cells that have low endogenous CXCR4 expression. Cells were then analyzed by flow cytometry after being stained simultaneously with anti-CXCR4 clone 4G10, which recognizes the N-terminus, and clone 12G5 whose epitope includes the second extracellular loop that is disrupted by the X4-ZFNs. WtCXCR4 was detected equally by both antibodies on the cell surface (middle panel, top row) and intracellularly (middle panel, lower row). However, CXCR4Δ18 was not detected by the N-terminal antibody on the cell surface (right panel, top row), but was detected when cells were permeabilized (right panel, bottom row) suggesting the 18 bp deletion prevents its expression on the cell surface.

(A) Ccr5Δ32 CD4+ T cells were stimulated on day 0 and transduced on day 1 with an Ad5/F35 vector expressing the X4-ZFNs, R5-ZFNs, or an untransduced control. On day 5, cells were HIV-infected with a mock, primary X4 HIV-1 (Bk132), lab-adapted X4 HIV-1 (HxB2), or a primary R5X4 HIV-1 (R3A). Live cells were counted approximately every two days. Cells were restimulated on day 13 (arrows). (B) Cxcr4 disruption frequency was assessed at various times by 454 deep sequencing. Disruption remained stable in the absence of HIV-1 infection, but profoundly increased in the presence of the three HIV-1 strains examined. Data shown is from one of two representative experiments.

The X4-ZFN pair's 24 bp binding site is conserved between humans and rhesus macaques. However, the human and rhesus R5-ZFNs have different binding sites; thus, a novel CCR5-ZFN pair was generated targeting rhesus ccr5. The rhesus R5-ZFNs and X4-ZFNs were delivered by Ad5/F35 vector at MOIs from 600–2000 into rhesus CD4+ T cells. Disruption frequency was measured by the surveyor nuclease assay. Data shown is an average of three independent experiments in cells from two different animals. Error bars represent standard error.