CRISPR/Cas9-mediated insertion of HIV LTR within BACH2 promotes expansion of T regulatory-like cells

Cloning of AAV Repair Templates.

Gibson Assembly or T4 DNA Ligase (New England Biolabs) were used to clone gene synthesized or PCR amplified fragments (IDT; Integrated DNA Technologies or Twist Bioscience). For each genomic target, plasmids were created to have 600bp homology arms homologous to the respective cut sites. All inserted nucleotide sequences and ligation products were confirmed by Sanger sequencing. Stellar competent cells (TaKaRa) were transformed with the finalized plasmids to generate stocks for AAV production.

AAV Production.

All donor templates designed for editing experiments were cloned into AAV plasmid backbones as previously described^12,13. Recombinant AAV6 stocks were produced in HEK293T cells as previously described¹³. Viral titers were determined by qPCR using AAV-specific primers and probe as described previously¹².

Human Peripheral blood samples.

Cryopreserved PBMC isolated from volunteers were purchased from the Fred Hutchinson Cancer Research Center. Upon thawing, CD4+ cells were isolated by negative selection (EasySep Human CD4+ Enrichment Kit; STEMCELL Technologies) and cultured in RPMI 1640 media with 20% FBS, 1x GlutaMAX (Life Technologies), 10nM HEPES (Life Technologies), and 1x beta mercaptoethanol (Gibco). Cells were cultured in 50ng/ml human IL-2 (Peprotech) between 0.5–1 × 10e6/ml in flat-bottom culture plates at 37°C with 5% CO₂.

CRISPR/Cas9 target design and reagents.

Single guide RNAs (sgRNAs) were designed to target genomic sites at the BACH2, STAT5B, and AAVS1 loci using the CCTop online tool¹⁴, and synthesized by Synthego. Guides were reconstituted in buffer TE (Synthego) with a final concentration of 50 pmol/ul, aliquoted and stored at −80C. Alt-R HiFi Cas9 was purchased from IDT (Newark, NJ).

Editing of Primary Human T Cells.

Following negative CD4+ selection, cells were activated for 3 days with Dynabeads Human T-Activator CD3/CD28 for T Cell Expansion and Activation (Gibco) according to the manufacturer’s instructions. After activation, beads were magnetically removed and cells were cultured overnight. Cells were then washed twice in PBS and resuspended in LONZA buffer P3 prior to nucleofection. Cas9 nuclease and sgRNAs were complexed at a 1:2.5 ratio and delivered to cells by LONZA 4-D Nucleofection System according to the manufacturer’s instructions, at a concentration of 2uM per 1.0 × 10⁶ cells. Cells were then transferred into prewarmed media formulated as described above, supplemented with 2.5% FBS. AAV6 vectors carrying the corresponding repair templates were used to transduce edited cells at 30% culture volume. Twenty-four hours later, standard media supplemented with 20% FBS was added to all culture wells and cells. Cells were maintained at 0.5–1.0 × 10⁶ cells/ml for the duration of the experiments. Media including fresh IL-2 was replaced every 2–3 days. The following sgRNA sequences were used in editing experiments: BACH2 G1 AACTGCTTGAGCCCAAAAGG, BACH2 G3 CCAGCAGTAAGTCTGTTGTA, STAT5B G3 GAGGCTACCACCTCACCTAG, and AAVS1 P1 ATTCCCAGGGCCGGTTAATG.

Genomic and Transcriptomic Analysis.

Genomic DNA was extracted from all experimental samples (DNeasy Blood and Tissue, Qiagen) and used in ddPCR to assess HDR or in conventional PCR to quantify NHEJ). To assess NHEJ, gDNA was amplified by PCR using PrimeSTAR GXL Polymerase (Clonetech) and custom primers sitting 200–300bp flanking the predicted cut site. Amplicons were then Sanger sequenced (Genewiz) and the resulting .ab1 sequence files served as input for ICE Analysis (Inference of CRISPR Edits) developed by Synthego. Primer sequences available upon request.

Each sgRNA generated in this study was submitted to CCTop (https://crispr.cos.uni-heidelberg.de/) to examine putative off-target CRISPR/Cas9 localization to the human genome (hg38) with the additional parameters: 20bp target site length, 12bp core length, and 4 total potential mismatches (maximum of 2 core mismatches). The top 8 predicted cut sites were investigated by sequencing as described above for evidence of off target cutting (Fig. S1), and the primer sequences are available upon request.

RNA was extracted from 3 to 5 × 10⁵ cells from all experimental samples using the RNeasy Micro Kit (Qiagen). 100 – 200 ng of RNA was used to synthesize cDNA using the iScript Advanced cDNA Synthesis Kit for RT-qPCR (Bio-Rad Laboratories) according to the manufacturer’s instructions. Synthesized cDNA was used as the template for hybrid/endogenous transcript ddPCR, as well as sequencing reactions.

Flow Cytometry.

Flow cytometry was performed on an LSR II (BD Biosciences) flow cytometer, and data analyzed using FlowJo software (TreeStar). Cells were labeled with fluorescent antibodies and the viability stain Alexa Fluor 350 carboxyl acid NHS ester (Invitrogen). Surface proteins were stained by incubation with Abs for 30mins at 4°C. In cases where intracellular protein expression was assessed, cells were fixed and permeabilized following surface stain. For FOXP3, CTLA-4, and HELIOS, fixation/permeabilization was carried out with the True-Nuclear Transcription Factor Buffer Set (BioLegend) according to the manufacturer’s instructions. For intracellular cytokine production, cells were cultured in media containing 50ng/ml PMA and 1ug/ml Ionomycin (Millipore Sigma), and 1ug/ml Golgistop (BD Bioscience) for 5h at 37°C. This was then followed by fixation/permeabilization with Cytofix/Cytoperm (BD Bioscience) and then incubated with cytokine-specific antibodies. To assess responsiveness to IL-2, stimulation and staining for pStat5 was performed as described¹⁵. 5ng/ml IL-2 (PeproTech) was used to stimulate the cells during the pStat5 stain. To assess cellular proliferation by flow cytometry, the CellTrace Far Red Cell Proliferation Kit (Invitrogen) was used according to the manufacturer’s instructions. Briefly, cells were incubated in PBS with the appropriate concentration of CellTrace stain for 20 min at 37°C. 5 times the staining volume of media was then added to all stained wells to remove free dye for 5 min, and then cells were pelleted by centrifugation and resuspended in fresh media. Cells were cultured for 96h before proliferation was assessed by flow cytometry.

Antibodies used in this study include, from BioLegend, IFNg (cat. 502516), TNFa (cat. 502944), CTLA-4 (cat. 349908), FOXP3 (cat. 320208), CD4 (cat. 300530), LAG3 (cat. 369314), ICOS (cat. 313528), PD-1 (cat. 329922 and 329920), CD69 (cat. 310910), CD44 (cat. 103028); from BD Biosciences, pSTAT5 (cat. 612599), CD3 (cat. 349201), CD25 (cat. 335789), CD127 (cat. 563086), CD62L (cat. 562720), CD8 (cat. 560662); eBioscience, HELIOS (cat. 501124891/61988342), TIGIT (cat. 5011268 / 12950042); and from Life Technologies, IL-2 (cat. 25–7029-42). All antibodies were used at a 1:100 dilution except for pSTAT5 (1:10) and TIGIT (1:50).

Digital droplet (dd)PCR HDR Assays.

Droplet Digital PCR was used to assess targeting efficiency in edited samples (Bio-Rad Laboratories). All primers and probes were designed in Primer-Blast (https://www.ncbi.nlm.nih.gov/tools/primer-blast/) using the Homo sapiens genome as the reference database, and were purchased from IDT. Briefly, a forward primer and HEX probe were placed on the LTR region of the inserted cassette, and a reverse primer was located downstream of the 3’ homology arm on the endogenous locus. The same forward primer and probe were used to assess editing at all genomic loci, while the reverse primer was specific to the gene targeted. The same primer and probe sets were used to assess editing events targeted by both the Solo-LTR and LTR-MSD constructs. A single control set of primers and a probe were designed to amplify a region of the STAT5B gene to assess the total number of alleles interrogated in each ddPCR reaction. All ddPCR reactions were duplexed with the insert set of primers and probes as well as the control set of primers and probes to assess the allelic frequency of HDR (homology-directed repair). Targeting efficiency was expressed as the ratio of the concentration of targeted alleles (positive for LTR insert) to that of the total alleles (positive for STAT5B control) present. Primer and probe sequences are available upon request. ddPCR was performed according to the manufacturers’ instructions.

Gene expression analysis.

The expression of hybrid HIV/BACH2 and HIV/STAT5B transcripts from edited samples were quantified using custom ddPCR primers and probes designed by Cesana et al.¹⁰, and the identities were confirmed by Sanger sequencing. Expression of BACH2, STAT5B, IL10, TGFB1 and HPRT was quantified using TaqMan Gene Expression Assays (Applied Biosystems): BACH2 (Hs00222364_m1, exon boundary 8–9), STAT5B (Hs00273500_m1, exon boundary 16–17), IL10 (Hs00961622_m1, exon boundary 4–5), TGFB1 (Hs00998133_m1, exon boundary 6–7), and HPRT (Hs99999909_m1, exon boundary 6–7). The concentration of hybrid or endogenous transcripts were normalized to concentration of HPRT transcripts duplexed in the same reaction.

Western blot analysis.

Chemiluminescent Western blot images and Ponceau S stain images were captured on the ChemiDoc MP imaging system. Quantification was performed using the ImageJ program. The Ponceau S stain band was used for quantification and normalization. BACH2 MW: 130 kDa. ACTB (b-actin) MW: 42 kDa.

Bulk RNA-sequencing and differential gene expression analysis.

Approximately 1 × 10e6 CD4+ T cells from a late culture timepoint (day 48 post-editing) were collected for RNA extraction (Qiagen) and poly-A selected mRNA library preparation (TruSeq RNA Library Prep Kit v2; Illumina). Cells were collected from 2 editing experiments. 33–41 million raw paired-end reads per sample were generated. Reads were mapped to a custom hg38 human reference genome which divided the BACH2 locus into 2 regions: BACH2_5p reads mapping upstream of the target site (hg38 gene position: 115,649) and BACH2_3p reads mapping downstream of the target site. Additionally, the LTR-MSD sequence was appended to hg38. STAR 2.7 was used to filter and map reads¹⁶. HTSeq was used to count the overlap of reads with genes¹⁷. Count normalization and differential gene expression analysis was performed using DESeq2¹⁸. The data produced by this experiment was deposited in the NCBI BioSample database for public access with the accession number PRJNA733175 (https://www.ncbi.nlm.nih.gov/bioproject/733175).

PANTHER Gene-ontology biological processes overrepresentation test.

Gene lists determined by differential expression were passed through the gene ontology (GO)^19,20 biological process complete list and analyzed by PANTHER overrepresentation test (Released 20200407)²¹ using the Fisher’s exact test with FDR correction. An FDR P<0.05 was used as the cutoff. The following GO annotation version and release date was: GO Ontology database DOI: 10.5281/zenodo.3727280 Released 2020–03-23.

Statistical analysis.

Statistical analyses were performed in GraphPad Prism 7 (GraphPad) and R Software package. The analyses performed are indicated in figure legends. Combined experimental values are shown as mean ± SD. (p values: ns = ≥ 0.05, * = 0.01 to 0.05, ** = 0.001 to 0.01, *** = 0.0001 to 0.001, **** = ≤ 0.00001).

Targeted insertion of HIV proviral elements into integration hotspots in primary human T cells.

To directly assess the impact of HIV integration into the BACH2 locus, we used HDR-based gene editing to insert different HIV component constructs into primary human CD4 T cells. Designer nucleases and rAAV6 donor templates were co-delivered using a highly efficient HDR-editing platform previously developed in our laboratory^{12,13,22–24}. The CRISPR/Cas9 system was used to achieve efficient nuclease-based DNA targeting using synthetic single guide RNAs (sgRNAs) targeting a region within intron 5 of BACH2 (BACH2 G1 and G3) and within intron 1 of STAT5B (STAT5B G1; Fig. 1a and Fig. S1), respectively, previously found to have multiple distinct HIV integrations in patient cohorts⁴.

Minor limitations in sgRNA design, due to oligo specificity and the 5’-NGG-3’ Cas9 PAM site requirement, restricted targeting to 2-bp upstream (hg38 BACH2 gene position: 115,649) of an in vivo recovered BACH2 IS⁶ (Fig. 1a). Similarly, targeting was directed to intron 1, the intron preceding the start of the coding sequence of STAT5B (hg38 STAT5B gene position: 29,125) 1-bp downstream of an in vivo recovered IS⁶.

As shown in Fig. 1b, CD4+ T cells were activated, electroporated for delivery of CAS9 protein and gRNA as ribonucleoproteins (RNP) complexes and maintained in culture for up to 7 weeks for analyses. Initial Cas9 cutting efficiencies at the BACH2 and STAT5B loci were assessed by non-homologous end-joining (NHEJ) repair readout, a measure of the NHEJ-mediated insertions and deletions (indels) that manifest after repair of double-strand DNA breaks. An average of 97% (+/− 2.4 SD) of indels were found at the intended BACH2 target site and 96% (+/− 1.2) of indels at the STAT5B target site, indicating highly efficient on-target cutting at both loci (Fig. 2c). Off-target cleavage was also measured for BACH2 G3 and STAT5B G1 based on in silico predictions and found to be minimal (Fig. S1). Additional BACH2 sgRNAs were assessed for both on-target and off-target NHEJ-repair based on in-silico predicted guide design; however, appreciable off-target cleavage was observed for BACH2 G1 and this guide was therefore excluded from further testing (Fig. S1).

Next, to recapitulate HIV-IS that faithfully mirrored the IS identified in relevant patient populations, four alternative AAV donor cassettes were generated. These donors contained either a single HIV-LTR (Solo-LTR) or a 5’ HIV-LTR continuing a downstream major splice-donor (MSD) sequence (LTR-MSD), and were designed to target either BACH2 or STAT5B. As shown in Fig. 1b, CD4+ T cells were initially activated and cultured for four days prior to electroporation for ribonucleoprotein (RNP) delivery and subsequently infected with rAAV6 donor viruses. HDR-mediated targeting of the AAV cassettes was measured by ddPCR, spanning the 600bp donor template homology arms. HDR rates at the BACH2 locus were 33.5% (+/− 1.2%) for Solo-LTR and 23.4% (+/− 2.1%) for LTR-MSD. Slightly higher targeting rates of 49.5% (+/− 6.8%) and 37.7% (+/− 7.5%), respectively, were obtained within the STAT5B (Fig. 1d). Regardless of initial editing efficiency or targeted loci, cellular viability was minimally impacted by RNP and rAAV co-delivery or the resulting genomic manipulations (Fig. S1–S2).

Insertion of HIV LTR-MSD within BACH2 uniquely promotes outgrowth of HDR-edited T cells.

To determine the effects of HIV LTR insertions at specific cellular loci, HDR-edited and control primary CD4+ T cell populations were cultured for up to 7 weeks post-editing. Strikingly, the proportion of BACH2 LTR-MSD-edited cells progressively increased over time, comprising ~70% of the culture at 7 weeks (Fig. 2a). When normalized to initial targeting frequency, BACH2 LTR-MSD-edited cells expanded by nearly 3-fold (Fig. 2b). These growth characteristics were dependent on the presence of the HIV MSD, as Solo-LTR-edited cells at the BACH2 locus did not expand (Fig. 2a–b). Absolute cell counts showed BACH2 LTR-MSD-edited cells continued to expand over 40 days of culture, while Solo-LTR-edited cell numbers, similar to controls, plateaued after ~14 days (Fig. 2c). In contrast, STAT5B-edited cells did not exhibit a proliferative phenotype with either construct (Fig. 2d–e). Consistent with ddPCR-based HDR tracking, STAT5B edited-cells, targeted with either the LTR-MSD or the Solo-LTR AAV donor, did not differ in cell counts when compared to controls (Fig. 2f). As an additional control, we utilized HDR-based editing of AAVS1, a ‘safe-harbor’ chromosomal location not associated with HIV IS or hybrid transcripts²⁵. Similar to control T cells and STAT5B edited-cells, AAVS1 LTR-edited cells did not exhibit differences in T cell expansion during 4 weeks of culture (Fig. S2). To control for an idiosyncratic impact of the specific CRISPR/Cas9 target site within intron 5 of BACH2, an alternative CRISPR guide (BACH2 G1) and AAV targeting cassette was used to introduce to the LTR-MSD cassette into an alternative location in intron 5. This led to a similar selective expansion of BACH2 LTR-MSD-edited T cells further validating these findings (Fig. S1). Together, these results demonstrate that targeting of an HIV LTR-MSD within intron 1 of BACH2 uniquely promotes the competitive outgrowth of HDR targeted primary CD4+ T cells.

Characterization of BACH2 and STAT5B targeted hybrid mRNAs.

Next, we investigated whether LTR hybrid transcripts were produced by the targeted cell populations. Primers were positioned in the LTR trans-cassette and in either exon 6 of BACH2 (Fig. 3a) or exon 2 of STAT5B (Fig. 3b). As expected, only the BACH2 and STAT5B-edited cells targeted with the LTR-MSD cassette exhibited qualitative levels of LTR hybrid transcripts. The identities of these hybrid transcripts were confirmed by DNA sequencing (Data not shown).

To determine whether the selective outgrowth of BACH2 LTR-MSD edited T cells correlated with production of hybrid transcripts, the relative amount of hybrid vs. endogenous transcript in BACH2 or STAT5B HDR-edited T cell populations was quantified using RT-ddPCR (Fig. 3a–b) at early (Day 7) and late (Day 35) cell culture timepoints. In all assays, transcript levels were standardized to those of a housekeeping gene, hypoxanthine guanine phosphoribosyl transferase (HPRT). In BACH2 LTR-MSD edited cells the level of hybrid transcripts, as a fraction of total BACH2 transcripts, increased 2.2-fold from early to late culture timepoints (p < 0.01) (Fig. 3c). In contrast, no change in hybrid transcript was observed in the STAT5B-edited cells (Fig. 3c). The levels of endogenous BACH2 and STAT5B transcripts were also evaluated over time, and interestingly, BACH2 transcript levels increased modestly in BACH2 LTR-MSD edited cells (1.3-fold +/− 0.2; p < 0.05) (Fig. 3d). This increase was not observed in BACH2 Solo-LTR-edited samples, suggesting that generation of hybrid BACH2 transcripts may facilitate this change. In contrast, there was no significant change in endogenous STAT5B transcripts in LTR-MSD edited cells (1.2-fold +/− 0.2; p > 0.05) compared to control cells (Fig. 3e). Consistent with these findings, BACH2 protein levels increased by up to 2.5-fold by day 21 in LTR-MSD edited-cells (p < 0.01), while there was no significant change in STAT5B LTR-MSD edited-cells (Fig. 3f–g). Together, these data demonstrate that expression of BACH2 can be driven by transcription initiated at the HDR integrated LTR and is likely responsible, at least in part, for the increased proliferation and/or selective survival of HDR-edited CD4+ T cells.

HIV LTR-driven BACH2 expression leads to a Treg-like population with increased proliferative capacity.

We hypothesized that the alterations in BACH2 gene expression in edited cells would alter the phenotype of these populations. We performed RNA gene expression analysis in HDR edited populations at ~7 weeks post editing. Insertion of the LTR-MSD resulted in differential expression of 452 genes relative to cells edited with Solo-LTR (p-adj. < 0.05): 309 upregulated and 142 downregulated in LTR-MSD, respectively. Established gene sets were characterized including: i) Treg cell markers, ii) T-cell exhaustion cell markers, and iii) cell cycle associated genes (Fig 4a and data not shown). IL7R (CD127) was 7.5-fold lower (adj. p = 1.1E-07) and IL2RB (CD25) was 5.9-fold higher (adj. p = 7.6E-04) consistent with a Treg phenotype²⁶. Markers for T cell exhaustion, LAG-3 and TIGIT, were also more abundantly expressed in the LTR-MSD population (11.3-fold; adj. p-value = 3.1E-04 and 3.7-fold; adj. p-value = 0.0051, respectively). Finally, at least 72 genes associated with cell cycle processes (GO:0022402) were differentially expressed in LTR-MSD-edited cells, including: MKI67 (3.7-fold; adj. p-value = 1.8E-05), TOP2A (2.9-fold; adj. p-value = 0.012), FOS (35.6-fold; adj. p-value = 3.5E-04), ASPM (3.1-fold; adj. p-value = 4.5E-04) and H2AFX (2.4-fold; adj. p-value = 6.1E-04) (Fig. 4a and data not shown). An unbiased gene-ontology biological processes overrepresentation test for the 452 differentially expressed genes in LTR-MSD edited cells revealed a list of 246 gene-sets with a FDR < 0.02 (Data available upon request). The top 30 overrepresented biological processes gene-sets with fold-enrichment are shown, the majority of these indicating genes enriched in cell cycle processes (Fig. 4b). An additional comparison between LTR-MSD and unedited cells revealed 154 differentially expressed genes (p-adj. < 0.05): 82 upregulated and 72 downregulated in LTR-MSD, respectively (Figure 4b).

Next, differential expression at the BACH2 locus (Fig 4c) was interrogated. Reads mapping to the inserted LTR cassette region (designated gene name: LTR) were 16.3-fold higher in LTR-MSD-edited cells (adj. p-value = 6.9E-04). This was associated with a 1.3-fold increase in transcript levels downstream of the insertion site (designated gene name: BACH2_3p; adj. p-value = 0.73), mirroring the RT-ddPCR results (Fig 4c). Interestingly, reads mapped to the region upstream of the BACH2 insertion site (designated gene name: BACH2_5p) were 4.1-fold lower (adj. p-value = 3.7E-03) in LTR-MSD-edited cells, implying that LTR promoter-initiated transcription dominates over the host BACH2 promoter (Fig. 4c).

To further investigate this altered cell phenotype, multicolor flow cytometry was used to evaluate the expression of proteins associated with T-cell activation, exhaustion, and differentiation. BACH2 LTR-MSD edited cells showed increased expression of multiple surface markers, including PD-1, CD69, LAG-3, TIGIT, and CD25 (Fig 5a and b). BACH2 LTR-MSD edited-cells also exhibited a shift from a predominantly naïve (Tn: CD44-/CD62L+) or memory (Tmem: CD44+/CD62L-) phenotype to a predominant T effector (Teff: CD44-/CD62L-) phenotype (Fig 5a).

Notably, expression of proteins associated with a Treg phenotype were also impacted by LTR-driven BACH2 expression: LTR-MSD BACH2-edited cells showed an increase in the proportion of CD25^high/CD127^low cells compared to controls (1.4-fold +/− 0.21; p < 0.001) (Fig 5a and b). Strikingly, expression of intracellular proteins associated with Treg programming including the key transcription factors, FOXP3 and HELIOS, as well as the cytoplasmic pool of the inhibitory receptor, CTLA-4, were increased in LTR-MSD edited-cells (Fig 5a). FOXP3 and HELIOS were increased by 1.5-fold (+/− 0.3, p < 0.05) and 1.6-fold (+/− 0.28, p < 0.01) respectively, while CTLA-4 expression was increased by 2.1-fold (+/− 0.32, p < 0.001) compared to controls (Fig. 5b). To further evaluate this Treg-like phenotype, edited cells were evaluated for suppressive cytokine production, including IL-10 and TGFβ commonly expressed by multiple Treg subsets²⁷. IL-10 transcript levels were significantly increased in the LTR-MSD BACH2-edited cells (1.8-fold +/− 0.29, p < 0.001), while TGFβ1 transcripts were decreased (1.3-fold lower +/− 0.16; p < 0.001) compared to controls (Fig. 5c).

The multiple surface and transcriptional phenotypic changes observed in LTR-MSD BACH2-edited cells prompted assessment of the capacity to express other key cytokines. Conventional Treg cells consistently exhibit a reduction in IL-2 expression and variable expression of inflammatory cytokines. IL-2 production in BACH2-edited cells was measured and found to be significantly reduced (1.2-fold lower +/− 0.11; p < 0.05) compared to controls (Fig. 5d). There were no discernable differences in the production of two other key inflammatory cytokines: INFγ or TNFα (Fig. 5d).

Interestingly, flow cytometry revealed a reduction in CD4 expression following HDR-editing, a finding specific to BACH2 LTR-MSD edited cells. Over 3 weeks in culture, CD4+ cells decreased from 98% to 73% in the LTR-MSD edited-cells (Fig. 5e; average of 4 donors, p < 0.01). While CD4 loss varied amongst PBMC donors, a significant and reproducible loss of CD4 expression was observed in all but one donor, with the greatest reduction reaching 50% CD4⁻ cells (Data not shown). To investigate whether this process reflected downregulation of CD4 production vs. surface internalization, a competitive flow cytometry based CD4 mAb assay was used. Intracellular CD4 expression was not detected in the CD4 surface negative population, suggesting that CD4 loss reflected reduced transcription and/or altered post-transcriptional events (Data not shown).

To begin to investigate the mechanism by which LTR-MSD BACH2-edited cells persisted and expanded in culture (Fig. 2c), we measured cellular proliferation using a flow cytometry-based proliferation dye dilution assay. At 5 weeks post-editing and growth in T cell expansion conditions, we observed that LTR-MSD cells proliferated 2.6-fold (p < 0.01) more than control cells (Fig. 5f); findings that correlate with enhanced growth rates of BACH2 LTR-MSD-targeted cells in culture.

Next, similar to CD25^hi conventional Tregs²⁸, it was posited that increased CD25 expression in HDR edited cells would facilitate an increase in signaling in response to IL-2. To initially test this idea, we performed flow analysis of phosphorylated STAT5 (pSTAT5) to measure IL-2 signaling. Cell populations were cultured in the absence of exogenous IL-2 for 24h, stimulated with IL-2 and assessed by flow cytometry²⁹. No significant differences in pSTAT5 expression was observed in cells with integrated LTR-MSD over controls (Fig. S3a). However, we reasoned that IL-2 starvation might differentially impact BACH2 LTR-MSD-targeted vs control cell populations. To this end, cells were exposed to different IL-2 doses (5ng/ml and 50ng/ml IL-2) post-editing and proliferation assessed by flow cytometry. Notably, in limiting IL-2 conditions, LTR-MSD edited cells proliferated 1.9-fold more (p<0.001) than the control cells (Fig. S3b–d). Additionally, as shown in Fig. S4, we performed LTR-MSD editing of BACH2 using input CD4 T cells that were depleted of natural (n)Treg. In this setting, HDR-editing again led to the generation and expansion of a Treg-like population. Thus, while LTR-MSD edited Treg-like cells are not derived from nTreg, they exhibit phenotypic similarities to nTreg including increased competitive fitness in an IL-2 limited environment.

In summary, our combined observations support the conclusion that BACH2 LTR-MSD editing results in increased BACH2 expression, thereby leading to generation of a unique T cell population exhibiting multiple Treg-like features: high levels of transcription factors including FOXP3 and Helios; increased expression of inhibitory receptors including CTLA-4 and LAG-3; upregulation of the high affinity IL-2 receptor (CD25^high/CD127^low); reduced expression of IL-2; and increased expression of IL-10. In parallel, this population exhibits evidence for increased cell activation (upregulation of CD69, PD-1, and acquisition of Teff phenotype), that correlate with enhanced proliferation and selective expansion in vitro.