Efficient modification of CCR5 in primary human hematopoietic cells using a megaTAL nuclease and AAV donor template

Abstract

Genetic mutations or engineered nucleases that disrupt the HIV co-receptor CCR5 block HIV infection of CD4⁺ T cells. These findings have motivated the engineering of CCR5-specific nucleases for application as HIV therapies. The efficacy of this approach relies on efficient biallelic disruption of CCR5, and the ability to efficiently target sequences that confer HIV resistance to the CCR5 locus has the potential to further improve clinical outcomes. We used RNA-based nuclease expression paired with adeno-associated virus (AAV) – mediated delivery of a CCR5-targeting donor template to achieve highly efficient targeted recombination in primary human T cells. This method consistently achieved 8 to 60% rates of homology-directed recombination into the CCR5 locus in T cells, with over 80% of cells modified with an MND-GFP expression cassette exhibiting biallelic modification. MND-GFP – modified T cells maintained a diverse repertoire and engrafted in immune-deficient mice as efficiently as unmodified cells. Using this method, we integrated sequences coding chimeric antigen receptors (CARs) into the CCR5 locus, and the resulting targeted CAR T cells exhibited antitumor or anti-HIV activity. Alternatively, we introduced the C46 HIV fusion inhibitor, generating T cell populations with high rates of biallelic CCR5 disruption paired with potential protection from HIV with CXCR4 co-receptor tropism. Finally, this protocol was applied to adult human mobilized CD34⁺ cells, resulting in 15 to 20% homologous gene targeting. Our results demonstrate that high-efficiency targeted integration is feasible in primary human hematopoietic cells and highlight the potential of gene editing to engineer T cell products with myriad functional properties.

INTRODUCTION

HIV entry into human T cells requires binding to both CD4 and one of several G protein (heterotrimeric guanine nucleotide–binding protein)–coupled chemokine receptors that act as co-receptors for HIV infection. CCR5 is the major co-receptor used by transmitted HIV-1 viruses (1). Highlighting the importance of CCR5 in HIV infection, a naturally occurring human CCR5 allele conferring HIV resistance creates a protein variant (CCR5 Δ32) that is nonfunctional (2–4). One strategy for treating HIV-infected patients is the use of engineered nucleases to disrupt CCR5 expression in patient T cells. Patients rein-fused with autologous T cells after CCR5 disruption with zinc-finger nucleases (ZFNs) showed improved CD4 T cell survival during HIV viremia induced by temporary cessation of antiretroviral drugs (5). The key to the method’s success is that CCR5 expression appears to be dispensable for normal immune responses, as evidenced in individuals who are homozygous for the CCR5 Δ32 allele. Thus, biallelic disruption of CCR5, desirable for effective HIV resistance, should not adversely affect T cell function. This feature also makes the CCR5 locus a potentially advantageous site to target for other genetic T cell therapies because this site does not affect cell survival or growth and is within open, transcriptionally active chromatin. Coding sequences that might be usefully targeted to this locus would include, but not be limited to, agents previously shown to help control or eradicate HIV (6).

Gene editing relies on the use of engineered nucleases to induce double-strand breaks (DSBs) in specific target genes. DSBs are repaired by endogenous cellular enzymes through one of two pathways: nonhomologous end joining (NHEJ), an error-prone pathway that results in a high frequency of nucleotide insertions or deletions (indels), or homology-directed repair (HDR), which seamlessly repairs DSBs by using homologous DNA as a template. HDR can be subverted to insert nonhomologous DNA into specific DSB sites by using an exogenous donor template, with the desired nonhomologous sequence flanked with homologous ones. Although for some applications, the goal of gene editing is to disrupt gene function by creating indel mutations, in other cases, HDR is required to insert a novel coding sequence or to repair a gene mutation.

Therapeutic application of HDR requires both an engineered, site-specific nuclease and an efficient method for transient delivery of this nuclease and a relevant DNA donor template into primary cells. We have described a hybrid nuclease platform that combines a transcription activator–like effector (TALE) DNA binding domain with an engineered, sequence-specific homing endonuclease (HE), referred to as a megaTAL (7). These nucleases promote efficient cleavage of genomic DNA (gDNA) with high sequence specificity, and the single megaTAL coding sequence can be efficiently delivered by mRNA transfection, allowing high-level transient expression. When HDR is the desired outcome, a suitable donor DNA template that satisfies key criteria must also be optimized. These criteria are that it must be easy to deliver and nontoxic to primary cells; it should be efficiently recognized as a candidate repair template by the HDR machinery; and it should not integrate randomly into the host chromatin. Here, we describe the optimization of gene editing at the CCR5 locus of primary human T cells, using a CCR5-specific megaTAL and recombinant adeno-associated virus (rAAV) delivered donor template. The resulting protocol allowed us to efficiently target a range of expression constructs to the CCR5 locus in primary T cells and adult mobilized CD34⁺ peripheral blood stem cells (PBSCs).

RESULTS

Activity of alternative nucleases editing the human CCR5 locus

Site-specific insertion of therapeutic coding sequences in primary cells via HDR requires efficient delivery of both a high-efficiency designer nuclease and a DNA donor template. In addition, we hypothesized that overhang architecture at the nuclease cleavage site might differentially bias DSB repair toward either HDR or NHEJ. We tested this idea for two alternative nuclease platforms that efficiently target the CCR5 locus: a megaTAL nuclease (CCR5 megaTAL; fig. S1) that generates 3’ DNA overhangs or a CCR5–TALEN (TALE nuclease) pair where a dimerized Fok1 nuclease creates 5’ overhangs. We used reagents that targeted the same region of CCR5 (Fig. 1A) to optimally compare rates of HDR.

Fig. 1. CCR5-TALEN and CCR5-megaTAL activity in T cells and comparison of NHEJ and HDR events in a TLR reporter line.

(A) Location of the CCR5-TALEN (red line) and CCR5-megaTAL nuclease (blue highlight) binding sites within the human CCR5 gene. Nuclease architecture at the target site is schematically diagrammed beneath. CCR5 sequence used as homology arms in donor templates in subsequent figures is indicated at top. (B) Percentage of NHEJ events in primary human T cells detected using the T7 and TIDE sequencing 18 to 21 days after transfection with megaTAL, TALEN, or BFP control mRNA (1 μg of each mRNA and 1 μg of RNA per TALEN monomer) (n = 5). (C) Schematic diagram of traffic light reporter (TLR) cassette and representative data showing fluorescent marker expression based on repair pathway. Translational reading frame is indicated in parenthesis for each coding sequence. Fluorescence-activated cell sorting (FACS) plots beneath show green fluorescent protein (GFP) and mCherry expression in representative CCR5 TLR assay. GibberishFP, open reading frame (ORF) encoded by out-of-frame translation of GFP; eGFP, enhanced GFP. (D) HDR versus NHEJ events at CCR5 TLR locus in human embryonic kidney (HEK) 293T cells after treatment with CCR5-megaTAL or CCR5-TALEN mRNA and increasing MOI of AAV.GFP repair template. Bars show the means ± SEM; significance was calculated using the unpaired two-tailed t test; P values <0.05 are in red. CDS, coding sequences; FOK1, Flavobacterium okeanokoites nuclease.

To verify that both nucleases efficiently targeted the CCR5 locus, we used polyadenylate-tailed mRNA to deliver nuclease or a blue fluorescent protein (BFP) mRNA (as transfection control) into prestimulated purified primary human CD4⁺ T cells. Samples receiving BFP control RNA were 98% transfected as measured by FACS 48 hours after trans-fection. In the absence of a repair template, DNA breaks are resolved by the error-prone NHEJ pathway, resulting in either seamless religation or indels. We first estimated indel frequency at the CCR5 locus using the T7 endonuclease assay [Fig. 1B; (8)] and detected about two times greater activity with CCR5-TALEN as with megaTAL (82% versus 44%, respectively). Because the T7 assay may underestimate NHEJ events in the presence of a uniform spectrum of indels, we also assessed NHEJ rates using tracking of indels by decomposition (TIDE) sequencing [Fig. 1B and fig. S2; (9)]. TIDE sequencing allows the quantitation and characterization of mutations within a capillary sequence analysis relative to a reference sequence analysis by using a decomposition algorithm. Using this assay, we observed a rate of 42 and 32% NHEJ (TALEN versus megaTAL, respectively). Thus, NHEJ rates in primary T cells were equivalent or slightly higher with the CCR5-TALEN versus megaTAL.

To first test whether these nucleases facilitate HDR repair, we used the TLR assay (10). In this assay, the CCR5-megaTAL and CCR5-TALEN binding and cleavage sites were inserted within the GFP coding sequence of the GFP-T2A-mCherry expression cassette, and lentiviral delivery was used to stably introduce this TLR cassette into the chromatin of HEK 293T cells (HEK-CCR5-TLR). Reporter lines were treated with gene-editing reagents (including various doses of an AAV.GFP repair template) followed by flow cytometry at 72 hours to simultaneously detect NHEJ (read as indels causing frame shift mutations that allow mCherry expression) and HDR (read as repaired GFP expression; downstream mCherry is out of frame; Fig. 1C). Transfection of HEK-CCR5-TLR cells with BFP mRNA alone, or with the AAV.GFP donor alone, did not affect GFP or mCherry expression (Fig. 1D). HEK 293T cells transfected with TALEN mRNA alone exhibited 3.9 times higher NHEJ rates than did cells treated with megaTAL alone. Although NHEJ rates dropped ~50% for both enzymes when AAV.GFP donor template was present, the NHEJ rates of the TALEN remained about threefold higher than those of the megaTAL at all AAV doses. Notably, although the CCR5-TALENs exhibited higher NHEJ rates, HDR events were similar at all AAV dosages in megaTAL- and TALEN-treated cells [or slightly higher for the megaTAL at 50,000 multiplicity of infection (MOI)]. On the basis of these data, we calculated a more than threefold higher HDR/NHEJ ratio (at all AAV MOIs, with the exception of the highest) when cutting was performed by the megaTAL than by the TALEN enzyme.

High-efficiency HDR at the CCR5 locus in primary T cells

Having established the efficiency of our nucleases, we next tested methods for transient delivery of a DNA donor template into primary human T cells using recombinant AAV. To identify an optimal AAV sero-type, we transduced enriched primary human CD4 T cells using self-complementary (sc) AAV.MND.GFP [scAAV expressing GFP downstream of the modified retroviral promoter/enhancer (MND)] packaged using seven different AAV serotypes: 1, 2, 2.5, 5, 6, 8, and DJ. AAV transduction was read out as transient GFP expression at 48 hours (fig. S3), and maximal expression was achieved with virus packaged with serotype 6 at all AAV doses.

To assess HDR at the CCR5 locus in primary human T cells, we generated AAV6-serotyped vectors containing an MND-GFP expression cassette inserted between 1.3 kb (kilobase) CCR5 homology arms (adjacent to the region containing both the megaTAL and TALEN cleavage sites; AAV.CCR5.GFP; Fig. 2A). A control AAV containing an MND-BFP expression cassette without homology arms (AAV.BFP) was also generated to track rates of non–homology-driven AAV insertion at gDNA breaks. We tested multiple parameters for combining CCR5-megaTAL mRNA transfection with AAV transduction to optimize both cell viability and HDR (stable GFP expression over time) including the following: timing and sequence of cell stimulation, transfection and transduction, and incubation temperature after transfection. We found robust HDR using the sequence of procedures shown in Fig. 2A (see also Materials and Methods); we refer to the nuclease mRNA transfection (preceding the AAV donor template transduction) as time 0 of gene editing. Using optimized conditions, the percentage of cells transfected with BFP mRNA was 98% at day 2 and decreased to <1% by day 16. Expression of the fluorochrome, in AAV.BFP- or AAV.CCR5.GFP-only–transduced cells, averaged 80 and 64%, respectively, after 2 days (Fig. 2, B and C). This expression was unstable over time (presumably because of the presence of episomal AAV), decreasing to <1 and 2% for AAV.BFP or AAV.CCR5, respectively, by day 16. In contrast, in cells treated with both megaTAL mRNA and AAV.CCR5.GFP donor, high-level GFP expression was present at early time points (81% at day 2) and decreased somewhat over time. Consistent with stable integration of the donor template into host chromatin, most cells maintained robust GFP expression (63% at day 16). To determine whether stable GFP expression resulted from HDR or from direct insertion of the AAV DNA into the CCR5 cleavage site, we tested whether CCR5 homology arms were required. Cells were transduced with an AAV.BFP construct lacking homology arms. Although a large fraction of cells were positive for BFP at day 2 (86%), this proportion decreased to ~7% by day 16. Although more BFP⁺ cells at day 16 were observed in the presence of the megaTAL (7% versus <1%), indicative of on-target insertion events, together these data indicated that most of the cells with sustained GFP expression after megaTAL and AAV.CCR5.GFP co-delivery likely underwent HDR (Fig. 2C).

Fig. 2. Efficient HDR in primary T cells leading to introduction of a GFP expression cassette within the CCR5 locus.

(A) (Top) Diagram of AAV constructs used as donor templates. (Bottom) Timeline of gene-editing procedure beginning with bead stimulation of CD4⁺ T cells. pA, SV40 poly-adenylation signal; HA, homology arm. (B) Representative flow cytometry plots showing BFP versus GFP expression at days 2 and 16 after gene editing. The percentage of live single lymphocytes within the relevant quadrant is indicated. (C) Cell viability and BFP and GFP expression over time after gene editing (n = 5). Significance was calculated using the unpaired t test with the Holm-Sidak correction for multiple comparisons; significant P values (<0.05) are in red. (D) Diagram of human CCR5 locus before and after HDR with the AAV.CCR5.GFP donor template. Primers binding outside homology arms were used to amplify gDNA from treated cells visualized on agarose gel; expected sizes of bands for unedited and edited alleles are indicated; right chart shows indel % (determined by TIDE sequencing) within the lower–molecular weight (MW) bands (n = 3). P value was calculated using the unpaired two-tailed t test. (E) Nested polymerase chain reaction (PCR) of single cells (31 to 100 cells per condition) edited using megaTAL and AAV.CCR5.GFP donor at indicted MOIs. Graph shows percentages of amplicons with a single higher–molecular weight band (biallelic HDR) and a single lower–molecular weight band (no HDR detected) or both bands (monoallelic HDR).

To confirm the occurrence of HDR, we amplified gDNA from untreated cells or from cells treated with megaTAL mRNA (with or without the AAV.CCR5.GFP donor template) using oligonucleotides targeted to sequence outside of the CCR5 homology arms (Fig. 2D). Whereas gDNA from all cells produced bands that matched the size of the wild-type allele (or alleles with small indels), we also detected a band that matched the size predicted for HDR-modified loci in cells treated with both the donor template and CCR5-megaTAL. The products of PCRs were cloned into plasmids and colonies sequenced to confirm the presence of HDR-modified alleles (fig. S4). Colony sequencing of PCR products also confirmed the presence of viral insertions within the CCR5 locus when AAV.BFP was used in place of AAV.CCR5.GFP as a donor template after megaTAL treatment (fig. S4). This amplification strategy was also used in single-cell PCR studies to estimate the frequency of biallelic and monoallelic HDR achieved with our gene-editing protocol. We identified amplicons consistent with biallelic HDR events in 42% of GFP⁺-engineered T cells at the lowest MOI of the AAV donor, and the frequency of biallelic events increased to 82% at our maximum MOI (Fig. 2E). Although CD4 T cells are the primary targets for therapeutic disruption of CCR5, the ability to target genes to the CCR5 locus has applications in both CD4 and CD8 T cells. For example, CD4-specific applications include introduction of gene products that limit HIV entry, processing, or integration; applications in CD8 cells include targeted introduction of chimeric antigen receptors (CARs). We thus tested the efficacy of this CCR5 gene–editing protocol in total CD3- and in CD8-enriched human peripheral blood cells. We observed similar high rates of HDR and similar viability in CD3-, CD4-, or CD8-enriched T cells that underwent our CCR5-editing protocol (fig. S5).

Integration-deficient lentivirus (IDLV) delivery has previously been used to facilitate HDR in hematopoietic cells (11, 12). We therefore also tested HDR rates using IDLV designed to deliver a GFP donor template. Initial studies in the HEK-CCR5-TLR reporter cell line showed that IDLV delivery of the GPF repair template resulted in lower HDR rates in comparison with AAV delivery (fig. S6A). Consistent with this finding, IDLV containing the CCR5.GFP reporter cassette also led to only a low level (0.5%) of stable GFP expression in primary T cells (fig. S6, B and C).

The coding sequence of CCR5 shares high sequence identity (74%) to that of CCR2, particularly within the region encoding the seven trans-membrane domains (84%), where the megaTAL and TALEN target sites are located. Sequences within CCR2 and KIAA1257 (an uncharacterized gene) were identified as potential megaTAL off-targets on the basis of homology to the CCR5 target site combined with the CCR5-megaTAL one-off specificity profile (13). We thus tested whether CCR5 gene editing resulted in off-target cleavage or HDR at the CCR2 locus. Quantification of indels at the potential target site in CCR2 by TIDE sequencing showed that NHEJ rates for the megaTAL and TALEN here were 21.8 and 1.8%, respectively, compared with 42% at the CCR5 locus for both (fig. S7A). Indels were not detected at KIAA1257 by TIDE analysis (fig. S7A). Similar to these findings, an early iteration of the ZFN now in clinical trials for CCR5 disruption also exhibited off-target activity within CCR2 (14). Despite off-target activity at CCR2, neither HDR nor capture of the AAV.CCR5.GFP donor events was detected at the CCR2 locus by PCR analysis (fig. S7B).

Comparison of HDR rates using the CCR5-megaTAL or CCR5-TALENs

Using our optimized procedures, we compared the HDR efficacy with CCR5-megaTAL to that with TALENs in primary T cells. We trans-fected primary CD4 T cells with mRNA to express either the CCR5-megaTAL (1 μg of RNA) or both TALEN half-sites (1 μg of RNA for each half-site), followed by AAV.CCR5.GFP at a range of dosages or by an AAV.BFP control (Fig. 3, A and B). Cell viability throughout the experiment was not significantly different between cells treated with the megaTAL or TALEN (Fig. 3B). Expression of the AAV.BFP control was also equivalent in megaTAL- and TALEN-treated cells, indicating that AAV transduction, expression, and insertion rates were independent of the nuclease used (Fig. 3B). We found higher GFP expression at all dosages of AAV and at each time point analyzed with the CCR5-megaTAL. Direct comparison of the proportion of cells with sustained GFP expression (day 16) revealed a 1.3- to 4.5-fold increase (across all AAV doses) using megaTAL versus the TALEN mRNA (Fig. 3, C and D). Because the CCR5-megaTAL nuclease uses a single coding sequence (a potentially important cost consideration in clinical application) and generates higher rates of HDR with less donor template in these studies, we pursued further characterization of gene targeting to the CCR5 locus using this platform.

Fig. 3. Comparison of HDR rates in human primary T cells with CCR5-megaTAL versus those with CCR5-TALENs.

(A) Representative FACS plots showing HDR events observed 16 days after treatment with CCR5-megaTAL versus those with CCR5-TALEN at varying MOI of the AAV.CCR5. GFP donor. The percentage of live single lymphocytes within a relevant quadrant is indicated. (B) Time course of cell viability (top), % BFP⁺ cells (middle), and % GFP⁺ cells (bottom) after treatment with control BFP (1 μg), CCR5-megaTAL (MT; 1 μg), or CCR5-TALEN (TAL; 1 μg of each TALEN half-site) mRNA with and without various MOI of the AAV.CCR5.GFP donor (or negative control AAV.BFP lacking CCR5 homology arms). (C) Comparison of % GFP 16 days after gene editing with CCR5-megaTAL versus that with TALEN at various MOI of AAV.CCR5.GFP donor template. (D) Ratio of % GFP⁺ cells after megaTAL- versus that of TALEN-mediated HDR over time and at various MOI of the AAV donor. All bars show means ± SEM of n = 5; significance was calculated using unpaired t test with the Holm-Sidak correction for multiple comparisons; P values (<0.05) are in red.

Maintenance of TCR diversity and sustained engraftment of gene-edited T cells

To determine whether our gene-editing protocol targets T cells representing a broad T cell receptor (TCR) repertoire, we analyzed the TCR spectratype of cells that had undergone gene editing using the CCR5-megaTAL and AAV.CCR5.GFP repair template. Edited cells were collected 8 to 16 days after gene editing for this analysis. We found that treatment with either CCR5-megaTAL alone or CCR5-megaTAL with AAV.CCR5.GFP had no substantial impact on the total TCR complexity of the sample. The complexity scores for each Vβ subfamily were also unchanged (Fig. 4A and fig. S8).

Fig. 4. Maintenance of TCR diversity and sustained engraftment of gene-edited T cells.

(A) Representative TCR spectratypes of Vβ subfamilies 1 to 25 in mock-treated T cells (bottom panel) or T cells 8 to 19 days after treatment with CCR5-megaTAL mRNA with and without AAV.CCR5. GFP donor. (B) Total number and relative percentage of CD4 lymphocytes (upper panels), GFP⁺ CD4 lymphocytes (lower left panel), and CD4⁺ CCR5⁺ lymphocytes (lower right panel) obtained from spleens of NSG mice 4 weeks after transfer of gene-edited T cells; numbers of mice for each condition is shown below x axis of GFP⁺ panel and is the same for all panels. Bars show the means ± SEM; P values were calculated using the unpaired two-tailed t test; P values <0.05 are in red. n = number of mice. (C) % GFP variation versus splenocyte number for each mouse receiving CCR5-megaTAL and AAV.CCR5.GFP donor–edited T cells. (D) gDNA from total splenocytes obtained from two recipient mice for each condition at 4 weeks after T cell transfer was PCR-amplified using primers outside of the donor template homology arms and run on an agarose gel. The predicted size of HDR-modified (4.4 kb) and NHEJ/unmodified alleles (3.2 kb) are indicated. Splenocytes from mice receiving T cells edited with megaTAL and AAV.CCR5.GFP donor co-delivery were flow-sorted before obtaining gDNA to detect HDR and NHEJ events in GFP⁺ and GFP⁻ cells. WT, wild-type.

We next tested the ability of gene-edited T cells to engraft immune-deficient, nonobese diabetic–severe combined immunodeficient–interleukin-2 receptor (IL-2R)γ^null (NSG) mice. T cells treated with the CCR5-megaTAL alone, or in combination with AAV.CCR5 GFP, were injected into NSG mice 8 days after gene editing. One month after injection, lymphocytes were collected from the spleen and bone marrow for analysis. Mice remained healthy throughout this period with no overt signs of graft-versus-host disease by coat appearance or weight loss. An average of 18% of splenocytes recovered from all treatment groups were CD4⁺ CD45⁺ human lymphocytes, with no statistically significant differences in percentage or absolute number between recipients of gene-edited versus those of mock-treated cells. The MOI of the AAV.CCR5.GFP donor used in these experiments was 59,000; therefore, the input cells transplanted were about 30% GFP⁺. This proportion of GFP⁺ cells was only slightly reduced after 1 month in vivo, indicating that edited cells maintained long-term viability (Fig. 4B). There was no correlation between % GFP-modified cells in vivo and the number of human CD45⁺ splenocytes (Fig. 4C). HDR was confirmed with PCR and colony sequencing (Fig. 4D). There was a 39% reduction in CCR5⁺ CD4⁺ splenic lymphocytes in recipients of CCR5-megaTAL only when compared to mock-treated cells, suggesting that altered CCR5 expression did not affect in vivo viability. An even greater reduction in CCR5 expressing CD4⁺ cells was observed in mice receiving cells treated with both megaTAL and AAV donor, with 52 and 70% reduction in CCR5 surface expression in GFP⁻ and GFP⁺ cells, respectively.

Ex vivo CCR5 gene editing in adult mobilized CD34⁺ cells

The ability to target genes to the CCR5 locus in CD34⁺ cells is an important goal for therapies aimed at hematopoietic diseases including HIV. We optimized the use of our CCR5 gene–editing reagents for mobilized CD34⁺ cells apheresed from adult human volunteers. As seen in T cells, AAV serotype 6 enabled the highest delivery of scAAV.GFP expression (fig. S3). Optimal HDR and cell viability were observed when the timing of mRNA transfection and AAV transduction was reversed compared to our T cell protocol, a result of the slower cell cycling of the CD34⁺ cell population. Additionally, although it increased our transfection efficiency, incubating the cells at 30°C greatly reduced CD34⁺ cell viability. Using conditions optimized for the CD34⁺ cells (Fig. 5A), 89% of cells expressed the BFP mRNA control 1 day after transfection (Fig. 5B). However, cells receiving the CCR5-megaTAL mRNA had NHEJ rates of only ~16% by TIDE sequencing, suggesting that CCR5-megaTAL activity is reduced in CD34⁺ cells in comparison to primary T cells and/or that this population promoted more efficient DSB seamless repair. Transduction with AAV.CCR5.GFP alone resulted in 16% of cells expressing GFP at day 4, but GFP was undetectable by day 10. Co-delivery of megaTAL mRNA and CCR5 AAV donor resulted in 35% GFP expression at day 4 and 14% sustained GFP expression at day 10. Higher AAV.CCR5.GFP expression with nuclease co-delivery at day 4 likely results from disruption of cell cycling due to electroporation and/or nuclease-mediated DNA breaks, thus delaying dilution of episomal AAV. Notably, the viability of CD34⁺ cells was more affected by the combination of AAV transduction and megaTAL transfection than that of T cells (Fig. 5, C and D). The presence of HDR events was verified by colony sequencing of PCR-amplified gDNA from cells receiving the gene-editing reagents (Fig. 5E).

Fig. 5. Gene editing of mobilized adult CD34⁺ cells using co-delivery of megaTAL and AAV.CCR5.GFP donor.

(A) Timeline of CD34⁺ cell gene-editing procedures and analyses. (B) (Left) % NHEJ events detected (by TIDE sequencing) 10 days after treatment with megaTAL mRNA and % of cells separately transfected with BFP mRNA 24 hours after transfection. (Right) representative flow plots (gated on live singlets) showing BFP expression 24 hours after transfection. (C) Representative flow analysis (gated on live singlets) of BFP and GFP expression in CD34⁺ cells transduced with the AAV.CCR5.GFP template with and without CCR5-megaTAL mRNA transfection at 4 and 10 days post-transduction. (D) Viability and % BFP⁺ and GFP⁺ CD34⁺ cells after treatment with the specified gene-editing reagents. (E) DNA gel of PCR-amplified gDNA obtained from gene-edited CD34⁺ cells. Predicted sizes of HDR-modified (4.4 kb) and NHEJ/unmodified alleles (3.2 kb) are indicated. Chart shows the indel % (by TIDE sequencing) within the lower–molecular weight band (n = 3). All P values were calculated using the unpaired two-tailed t test; n = number of independent experiments, performed using cells from two donors.

Delivery of genetic therapies to the CCR5 locus in primary human T cells

Clinical trials disrupting CCR5 expression in HIV patient autologous T cells are currently underway. We tested the capacity to simultaneously disrupt the CCR5 locus and insert an expression cassette for candidate HIV therapeutics into the CCR5 locus, an approach anticipated to enhance the clinical use of CCR5 disruption. Previous studies have shown that expression of the HIV restriction protein C46 protects cells from infection by HIV (15, 16) and may offer protection against HIV strains with CXCR4 tropism. We designed a donor template to test whether we could engineer C46 expression into targeted CD3 T cells (Fig. 6A). Notably, our preliminary analysis introducing fluorescent reporters suggested that decreasing homology arm lengths to 0.6 kb decreased HDR rates by ~30% (fig. S9). The homology arms used in this vector were 0.8 kb, the maximum length we could include while remaining within the AAV packaging size limit. Sixteen days after gene editing, 28% of T cells expressed GFP linked by the T2A cleavage peptide; C46 expression at the cell surface was confirmed with the monoclonal antibody (mAb) 2F5. We next designed a second-generation anti–HIV-CAR gene repair construct on the basis of a high-affinity broadly neutralizing HIV antibody [PGT145; (17)]. The HIV-CAR included the single-chain variable fragment of PGT145 upstream of both CD3ζ peptide and CD137 (4-1-BB) costimulatory domains (Fig. 6B). Downstream of the HIV-CAR protein, we placed a T2A peptide linked to an RQR8, a peptide consisting of epitopes derived from both CD34 and CD20 (18). RQR8 allows for the enrichment of cells expressing the HIV-CAR with commercially available clinical-grade anti-CD34 affinity beads (whereas the CD20 antigen can be used as a suicide gene in the presence of rituximab). As a candidate therapeutic CAR in the clinical setting of B cell malignancies in subjects with HIV, as well as positive control for target cell killing assays, we also created a donor template to express an anti–CD19-CAR and T2A-BFP cassette (Fig. 6C). Targeted insertions of anti–HIV-CARs and anti–CD19-CARs at day 16 were 14 and 9% in CD3 T cells as assessed by anti-CD34 and BFP expression, respectively. Anti–HIV-CAR cells were enriched by sorting on anti-CD34 beads, leading to 40% stable RQR8 expression, and similarly, anti–CD19-CAR cells were sorted for BFP expression, achieving a stable 51% expressing population. We verified the functional ability of our engineered T cells to become activated and to lyse target cells downstream of CAR antigen binding. First, we incubated edited or mock control CD3 T cells with cell lines stably expressing either HIV envelope or CD19 (Fig. 6D), observing that engineered T cells up-regulated the activation marker CD137 only when incubated with cells expressing their target antigen. We then tested the ability of CD19-CAR⁺ T cells to lyse K562 cells expressing either CD19 and cis-linked GFP or a nonspecific target [B cell maturation antigen (BCMA)] and near-infrared fluorescent protein (iRFP) (a 1:1 ratio of CD19 target cell to irrelevant target expressing cell; Fig. 6E). Although the ratio of GFP⁺/iRFP⁺ target cells was not significantly altered in the presence of anti–CD19-CAR⁻ T cells, addition of anti–CD19-CAR⁺ T cells resulted in complete loss of GFP⁺ target cells at effector/target ratios of 3:1 or greater, confirming that CCR5 gene–edited T cells maintain the ability to respond to CAR engagement and direct cell killing in vitro. PCR amplification of the CCR5 allele verified the presence of bands consistent with HDR using the C46 and CAR donor templates (Fig. 6F).

Fig. 6. Efficient expression of genetic therapies targeted to the CCR5 locus in primary CD3 T cells.

(A to C) AAV donor templates used to target C46 (A), HIV-CAR (B), and CD19-CAR (C) to the CCR5 locus, including viral packaging size and MOI. FACS plots show expression 16 days after gene editing. For each, we show negative controls that were mock-transfected or transduced with only the AAV donor and the % of cells in relevant quadrant for transgene expression. HIV-CAR and CD19-CAR populations were column- or sort-purified before analysis at day 16. (D) CD137 expression in engineered T cells (x axis) cultured for 24 hours in the presence (blue or pink bars) or absence (black bars) of target cells expressing their indicated cognate antigen; P values were calculated using the unpaired two-tailed t test, and significant P values (<0.05) are in red; n = 3 independent experiments using T cells from different subjects. (E) Selective loss of GFP⁺ CD19–expressing K562 target cells (T) relative to CD19⁻ iRFP⁺ K562 cells (y axis) 48 hours after treatment with CD19-CAR⁺ T cell effectors (E) (ratio of E/T indicated on the x axis); n = 3. (F) DNA gels showing PCR amplicons of gDNA using primers binding outside of the CCR5 homology arms. gDNA was obtained from T cells edited with the AAV donors diagrammed in (A), after purification by flow cytometry. Predicted sizes of HDR-modified alleles (CD19-CAR. T2A.BFP, 6.4 kb; HIV-CAR.T2A.RQR8, 6.3 kb; C46.T2A.GFP, 6.0 kb) are indicated by arrowheads at the right; open arrow indicates size of unmodified allele (3.2 kb). WPRE, woodchuck hepatitis virus posttranscriptional regulatory element; PE, phycoerythrin.

DISCUSSION

Effective clinical application of gene editing requires efficient and safe methods for modification of primary human cell populations including hematopoietic cell lineages. The capacity to achieve HDR in primary cell populations, however, has lagged behind the recent rapid advancements in nuclease engineering. Here, we describe a cell engineering platform capable of driving high levels of gene editing in primary human T cells and CD34⁺ hematopoietic stem cell (HSC) through the use of transient co-delivery of nuclease mRNA and a single-stranded rAAV donor template. Identification of an optimal DNA donor platform for use in primary hematopoietic cells, in parallel with careful timing and optimization of mRNA delivery, allowed us to successfully apply editing technologies. We show that rAAV serotype 6 comprises an excellent DNA donor system that exhibits relatively limited toxicity in T cells, as measured by cell viability, even in the setting of very high MOI. Using this co-delivery method, we found both high cell viability and high levels of HDR in primary T cells with both megaTAL and TALENs. Most important, our methods easily introduced clinically relevant gene products into the CCR5 locus in T cells, pointing to a potentially safe clinically translatable gene-editing platform.

The HDR rates described in this study in primary T cells are very high. To date, using 14 independent donors in 25 separate experiments, we observed an average knock-in rate of the MND-GFP cassette of 60% (range, 52 to 71%; SD, 4.9%). Single-cell molecular analysis of megaTAL-modified cells demonstrated high levels of biallelic HDR in cells expressing the GFP cassette, with rates greater than 80% at our highest level of AAV delivery. Similar HDR rates were observed in bulk CD3 as well as in enriched CD4 or CD8 T cell populations. Further, there was little loss of cell viability during co-delivery or subsequent culture manipulation. The engineered cells retained a broad TCR repertoire, consistent with modification of multiple independent cells, and exhibited sustained genomic modification without loss or altered competitive advantage in vivo in the NSG model. In comparison to our findings, a previous study assessed HDR in primary T cells using DNA transfection of ZFNs and donor, finding ~5% modification of IL2R [reported as 20% on the basis of a correction for transfection efficiency; (19)]. Other work in human T cells has focused primarily on gene disruption. Plasmid delivery using Cas9 and guide RNAs led to NHEJ rates of <5% using a single guide and up to ~35% with multiple-guide delivery (20). Notably, the authors reported that primary cells, in contrast to transformed lines, were highly sensitive to plasmid-based DNA delivery, consistent with activation of innate DNA sensing pathways. mRNA-based nuclease delivery, as described here, also avoids the potential for insertion of a nuclease coding cassette into the cell genome, a scenario that could lead to genomic instability through accumulation of off-target cleavage events. mRNA has previously been used to deliver TALENs for TCR disruption, followed by lentiviral delivery of a CAR (21). TCR knockout frequency (~50%), however, was lower than the CCR5 disruption rates achieved here with TALEN and megaTAL reagents, and as discussed below, our co-delivery method allowed both the desired CCR5 knockdown and targeted insertion of a CAR expression cassette at the same site.

The use of rAAV in this study was based on previous seminal work demonstrating its inherent capacity to facilitate HDR [reviewed in (22, 23)]. Although the precise mechanisms that promote this process remain to be defined, rAAV gene targeting requires homologous recombination pathway proteins including RAD51/RAD54 (24) and properties specific to the structure or cellular processing of the single-stranded AAV genome (25). Additionally, as donor DNA is rate-limiting in driving HDR, the capacity to generate very high titer vector stocks combined with the limited toxicity of rAAV in most cell types provides the opportunity to drive mass action donor effects via delivery of very large numbers of viral genomes. Recently, several groups have leveraged these initial observations showing that after a nuclease-mediated DSB, rAAV-mediated HDR is markedly enhanced (>100-fold) in transformed cell lines, primary fibroblasts, or in vivo after liver-directed delivery (26–32). Although editing efficiencies of 1 to 10% were reported in most studies, levels as high as 65% were observed in the U2OS osteo-sarcoma line (27) after co-delivery of nuclease and donor in a single rAAV at an MOI similar to that used in our study. Application of these promising approaches in primary cells has been limited, in part, by identification of serotypes that efficiently transduce hematopoietic cell populations. We screened a panel of scAAV vectors containing a robust internal promoter driving GFP reporter gene expression and identified serotype 6 as optimal for targeting of both human primary T cells and CD34⁺ progenitors. Although our serotype findings in CD34⁺ cells differ from earlier studies (33, 34), they are consistent with work from another group (35, 36). On the basis of the serotype screen, we used AAV6 to successfully apply rAAV donor delivery in primary cells. In addition to templating targeted gene addition, rAAV also exhibits nonspecific integration at DNA break sites (37). Although this feature is prominent in transformed cell lines, reaching 20 to 30%, for example, in the osteosarcoma line described above (27), we found no evidence, on the basis of flow cytometry, for integration of AAV lacking homology arms in T cells in the absence of nuclease delivery. Low BFP reporter expression (~10-fold below HDR rates) was stably maintained, however, using control AAV with nuclease co-delivery, consistent with on-target AAV insertion. Notably, on the basis of genomic PCR analysis, targeted insertion was not observed after co-delivery of megaTAL and AAV donors with relevant homology arms, suggesting that HDR likely outcompetes random capture events. Our data cannot, however, exclude the possibility that non–HDR-directed integration of AAV occurs, either on-target at CCR5 or off-target at the highly similar CCR2 locus or other sites. Such integrations are unlikely to be any less safe than delivering CAR therapies by randomly integrating retroviruses, a method used in clinical trials without evidence of insertional mutagenesis [reviewed in (38)].

A key feature unique to the megaTAL platform is generation of DSB with 3’ DNA overhangs. We speculate that increased HDR rates at the CCR5 target site using the CCR5-megaTAL reflect a platform-specific bias in repair pathway choice due to differences in DNA processing required to promote HDR and/or involvement of single-strand annealing pathways. However, to fully address this idea will require comparison of HDR efficiency at additional target sites using alternative TALENs, as well as other FOK1-based nucleases (ZFNs) and blunt-end cutters including Cas9/CRISPR reagents. Despite our limited mechanistic understanding, the efficiency of HDR mediated by the CCR5-megaTAL at this target site and the ability to deliver this nuclease from a single mRNA therapeutic validate consideration of this nuclease for clinical translation.

As an initial test for clinically relevant HDR gene-editing applications in HIV, we used several alternative gene expression cassettes including the C46 peptide and disease-relevant CAR constructs. We paired these knock-in cassettes with CCR5 gene disruption. The CCR5 disruption rates obtained with megaTAL or TALEN mRNA delivery were equivalent or superior to previous in vitro studies and as well as a ZFN-based clinical trial (5). We succeeded in pairing high-efficiency disruption with targeted integration of therapeutic gene cassettes, with unselected HDR efficiencies of ~8 to 30%. Variability in HDR efficiencies between viral constructs was partially explained by homology arm length (fig. S9) and was also affected by viral titer and size of nonhomologous insert; an important goal for clinical vectors will be to optimize these parameters of the donor template, as well as features of the therapeutic cassette that affect mRNA transport and stability. In our candidate anti–HIV-CAR, we included a cis-linked, clinically relevant surface peptide (18). This allowed us to enrich gene-modified T cells by using an anti-CD34 antibody and provides the capacity for subsequent specific in vivo depletion of modified T cells with an anti–CD20 mAb—an approach readily applicable to other therapeutic cassettes. We demonstrated in vitro functional activity using both CAR knock-in constructs. The MND, a retroviral promoter engineered to resist transcriptional silencing (39, 40), was used to achieve robust and stable transgene expression in lymphocytes. Although this promoter has been shown to be safe and effective in a clinical gene therapy trial (41), future therapeutics may use natural mammalian promoters or include insulators to reduce the potential for insertional mutagenesis or transcriptional silencing (42). It will also be important to determine whether CCR5 site-specific integration manifests unanticipated safety concerns beyond CCR5 disruption alone. Although further work is required to assess sustained in vivo activity of these edited T cell populations, our capacity to achieve one-step, high-level modification and CCR5 disruption is likely to be rapidly translatable into clinical application. For example, on the basis of the now well-documented activity of anti–CD19-CAR T cells [reviewed in (43)], the capacity to endow patient T cells with both resistance to HIV and the capacity to target B cell tumors represent a potential therapeutic for malignant complications encountered in HIV subjects.

We also applied our co-delivery methods to promote HDR in mobilized human CD34⁺ PBSCs, a cell population relevant to clinical application across a wide range of hematopoietic disorders. We observed HDR rates of ~14% across multiple independent donors using co-delivery of megaTAL mRNA and rAAV. Although significantly less robust than in primary T cells, this rate is similar to previous work with ZFN mRNA and IDLV donor co-delivery to edit human cord blood–or bone marrow–derived CD34⁺ cells (11, 12). However, the rate at which we are successfully editing and maintaining the pluripotency of true HSCs within this population remains to be determined. Notably, our co-delivery approach in CD34⁺ cells promoted substantially greater cell toxicity than in primary T cells, similar to results reported with ZFN and TALEN editing. Optimization of HDR in CD34⁺ cells using this and other approaches [including Cas9 with dual-guide RNAs for efficient gene disruption (20)] will likely require a combinatorial manipulation of cell culture parameters to influence cell cycle and DNA repair pathway choice. Additionally, application of AAV serotypes with improved HSC transduction (34) and/or additional screening of recombinant serotypes may also facilitate editing outcomes.

In summary, we have established highly efficient methods to achieve targeted recombination within primary human hematopoietic cells by using co-delivery of megaTAL or TALEN-encoding mRNA and rAAV donor template. Although the packaging capacity of rAAV may limit some applications of this method, our findings highlight its potential for use in gene editing to engineer T cell products with a range of clinically relevant features and suggest that this platform may be translatable for use with alternative nucleases and cell types.

MATERIALS AND METHODS

Experimental design (objectives, design, and prespecified components)

The objective of this study was to develop a protocol for directing the integration of exogenous coding sequences into the CCR5 locus of primary human T cells. Additional aims included assessing the impact of this gene editing on the homeostasis and cytotoxic T cell killing function in gene-edited cells and testing gene-editing reagents in adult human mobilized CD34⁺ peripheral blood mononuclear cells (PBMCs). We first performed pilot experiments to optimize our gene-editing protocols in primary T cells and determined sample sizes needed to achieve an 80% statistical power at a P value of 0.05 to detect differences in HDR rates by flow cytometry detection of stable GFP expression. For NSG experiments, we calculated the sample size needed to detect a 1 SD change in the number of CD4⁺ cells obtained at the end of the experiment with an 80% power and a P value of 0.05.

Nuclease design

The CCR5-targeting HE was identified and engineered on the basis of previously described methodologies (13, 44). After yeast-based selection, the HE was assembled with a 10.5–repeat variable diresidue (RVD) TAL array as previously described (7) to generate the megaTAL enzyme (fig. S1). The resulting megaTAL enzyme recognized a 38-bp stretch in the sixth transmembrane domain of the CCR5 gene (Fig. 1). The CCR5-specific TALEN pair was assembled with a Golden Gate cloning strategy (45) into a pthXO1 scaffold truncated at positions N154 and C63 as described (46). The Xanthomonas pthXo1 Golden Gate destination ORF and an RVD plasmid library were a gift from D. Voytas. The final TALEN pair and the megaTAL nucleases were cloned into a basic mRNA expression vector.

Production of recombinant AAV and lentiviral vectors

See Supplementary Materials for details of AAV and lentiviral constructs. AAV stocks were produced by triple transfection of AAV vector, serotype helper, and adenoviral helper [HGT1-adeno] plasmids in HEK 293T cells. Transfected cells were collected 48 hours later, lysed by freeze-thaw, benzonase-treated, and purified over iodixanol density gradient as previously described (47). Titers of the viral stocks were determined by qPCR of AAV genomes (48) and ranged from 1 × 10¹¹ to 1 × 10¹² per microliter. Lentivirus (LV) was prepared as described (49). Titers for IDLVs were determined with Lenti-X p24 Rapid Titer ELISA Kit (Clontech) according to the manufacturer’s instructions.

Primary human T cell gene editing

T cells were thawed using drop-wise addition of cold deoxyribonuclease (DNAse) I buffer [phosphate-buffered saline (PBS), 5 mM MgCl₂, DNase I (20 Kunitz U/ml) (EMD-Millipore)], followed by centrifugation. Cells were resuspended at 1 × 10⁶ live cells/ml in T cell growth medium (basic culture medium supplemented with IL-2, IL-7, and IL-15 at 50, 5, and 5 ng/ml, respectively) and stimulated by using CD3/CD28 beads (Dynabeads, Life Technologies) for 48 hours at a 1:1 cell-bead ratio. Beads were then removed, and cells were allowed to rest in T cell growth medium for 16 hours at 5 × 10⁵ cells/ml. Next, cells were electroporated with mRNA using the Neon Transfection System and 10-μl tip as follows. Cells were washed twice with PBS, resuspended in Neon Buffer T at a density of 2.4 × 10⁷ cells/ml, then 1 μg of mRNA (CCR5-megaTAL and BFP) or 2 μg of mRNA (1 μg of each CCR5-TALEN half-site) was added for every 2.5 × 10⁵ to 3 × 10⁵ cells. After mixing, cells were electroporated (1400 V, 10 ms, and three pulses) and immediately dispensed into 200 μl of prewarmed T cell growth medium in a 96-well plate. For samples transduced with AAV, AAV was added to the culture 2 to 4 hours after electroporation, followed by continued 30°C incubation for 20 additional hours. AAV donor was added as 20% of the final culture volume regardless of titer (~1 × 10⁵ MOI) to optimize HDR events, unless specified otherwise. Samples not transduced with AAV were cultured continuously at 30°C for 24 hours. Subsequently, edited cells were cultured using standard conditions (37°C and expanded in T cell growth medium, replenished as needed to maintain a density of ~1 × 10⁶ cells/ml every 2 to 3 days).

To scale up editing of primary human T cells for transplant into NSG mice, we used the above editing procedure with the following modifications: 10 μg of CCR5-megaTAL mRNA was used to transfect 3 × 10⁶ primary human CD4⁺ T cells using the Neon Transfection Kit 100-μl tips (using the same electrical parameters as above). Cells were transduced 2 to 4 hours after electroporation by addition of 20% culture volume of AAV (6 × 10⁵ MOI), returned to 30°C for 22 hours, then expanded in a G-Rex10 flask (Wilson Wolf Manufacturing). During expansion, 75% of the medium volume was exchanged every 3 to 4 days with fresh T cell growth medium. Eight days after electroporation, cells were collected, washed, and used for transplantation.

TCR spectratyping

Spectratypes of TCR Vβ CDR3 subfamilies present in gene-edited CD4⁺ T cell samples were determined by the Immune Monitoring Core at the Fred Hutchinson Cancer Research Center using multiplex reverse transcription PCR as described (50). Complexity scores of each subfamily were determined as described (51).

Murine transplantation and analyses

NSG mice (Jackson Laboratory) were maintained in a specific pathogen–free Association for Assessment and Accreditation of Laboratory Animal Care–accredited facility in accordance with National Institutes of Health (NIH) Guide for the Care and Use of Laboratory Animals and the Seattle Children’s Research Institute’s Institutional Animal Care and Use Committee. Animals (8 to 15 weeks old) were injected intraperitoneally with 2 × 10⁷ engineered T cells in PBS, monitored daily and weighed twice-weekly, and euthanized at 4 weeks after transplant for analysis of engrafted cells.

Adult human mobilized CD34⁺ cell culture and transfection

Cryopreserved CD34⁺ cells enriched from PBSC mobilized adult donors were obtained from the Core Center for Excellence in Hematology at the Fred Hutchinson Cancer Research Center. Cells were thawed as described above. After centrifugation, CD34⁺ cells were resuspended to a concentration of 1 × 10⁶ cells/ml in serum-free stem cell medium [StemSpan Medium (STEMCELL Technologies) with thrombopoietin, stem cell factor, IL-6, and FLT3 ligand (PeproTech), all at 100 ng/ml]. CD34⁺ cells were prestimulated in this cytokine mix for 24 hour at 37°C, followed by the addition of the AAV (10% of culture volume) for 24 hour, washed with PBS, resuspended in Neon Buffer T at a density of 3 × 10⁵ cells/ml, and transfected using 1 μg of mRNA for every 3 × 10⁵ cells using the following settings: 1900 V, 20 ms, and 1 pulse. Cells were then dispensed into 3 ml of prewarmed stem cell medium in a six-well plate, and then incubated at 37°C. StemRegenin (1 ng/ml ) (Cellagen Technology) (52) was added 24 hours after gene editing and on alternate days in culture. Analysis of GFP and BFP expression was performed 4 and 10 days after gene editing. gDNA was extracted from cells at each condition on day 10 for molecular analyses of HDR.

T cell killing assays

An expression cassette for CD19-CAR.2A.BFP was targeted to the CCR5 locus of CD3⁺ T cells using the above gene-editing protocol, grown in T cell growth medium for 9 days, and FACS-sorted into BFP-positive and BFP-negative fractions. Sorted cells were expanded for 1 week in T cell growth medium in a T25 flask with OKT3 antibody (30 ng/ml), 25 million irradiated human PBMCs, and 5 million irradiated transformed lym-phoblastoid cell lines (53) as described (54). Target and control (CD19 or BCMA expressing, respectively) cells were generated by LV transduction of K562 cells using either pRRL.MND.CD19.2A.GFP or pRRL.MND. mBCMA.2A.iRFP, respectively. For killing assays, BFP-positive or BFP-negative T cells (effector) were plated in T cell growth medium in a 96-well plate with a 1:1 mix of CD19/GFP⁺K562 (target) and 5 × 10⁴ mBCMA/iRFP⁺ K562 cells at effector to target ratios of 1:1, 3:1, 5:1, and 10:1. After 48 hours, cells were analyzed by FACS. CAR-mediated killing was reported as a decrease in the percentage of GFP⁺ target cells relative to iRFP⁺ control cells.

CD3⁺ T cells that underwent CCR5 gene editing using the AAV.CCR5.HIV-CAR.2A.RQR8 donor template were enriched by flow sorting RQR8⁺ cells (labeled with anti-human CD34 followed by goat anti-mouse IgG1-PE secondary; table S1) and grown as described above. CD3⁺ T cells with the CD19-CAR.2A.BFP construct were generated in parallel and enriched by sorting on BFP. To assess activation of HIV-CAR–expressing cells, edited cells were cocultured at a 1:1 ratio with either a chronically HIV-infected T cell line (ACH-2) or the parental control, A3.01 T cell line (final concentration, 1 × 10⁶ cells/ml). Before use in this assay, ACH-2 and A3.01 cells were grown separately for 24 hours in T cell culture medium with 1 μM phorbol 12-myristate 13-acetate and antiretroviral therapy (tenofovir, azidothymidine, and nevirapine, 20 μM each; NIH AIDS Reagent Program). After 24 hours in coculture-edited T cells were stained using anti-CD137, followed by LIVE/DEAD Fixable Near-IR Viability Stain, incubated for 15 min in 4% paraformaldehyde and analyzed by flow cytometry.

Statistical analysis

Statistical analyses were performed with Prism 6 (GraphPad Software). Data are shown as means ± SEM unless otherwise noted. Tests of statistical significance were performed using the unpaired two-tailed Student’s t test; where noted, we corrected for multiple comparisons using the Holm-Sidak method with α = 5.0%. For multiple comparisons of normally distributed data, statistical analysis was determined with a two-way analysis of variance using Tukey’s multiple comparison test, α = 5.0%.

Supplementary Material

S

Fig. S1. CCR5-megaTAL amino acid sequence.

Fig. S2. Spectrum of indels at CCR5 target site in human T cells after megaTAL or TALEN treatment.

Fig. S3. Robust transduction of human primary hematopoietic cells using AAV6.

Fig. S4. Sequence analyses verifying seamless HDR using AAV.CCR5.GFP donor template.

Fig. S5. Comparison of transfection, transduction, and HDR events using CCR5 gene–editing reagents in CD3, CD4, and CD8 T cells.

Fig. S6. Efficacy of IDLV delivered donor template for HDR gene editing.

Fig. S7. Evaluation of off-target editing by CCR5 nucleases in primary human T cells.

Fig. S8. TCR spectratyping demonstrates maintenance of TCR complexity in gene-edited T cells.

Fig. S9. Variation of HDR rates with CCR5 homology arm size.

Table S1. Antibody sources.

References (55–62)