HIV-1 promoter is gradually silenced when integrated into BACH2

The persistence of the latent HIV-1 reservoir is a major obstacle to cure HIV-1 infection. HIV-1 integrates into the cellular genome and some targeted genomic loci are frequently detected in clonally expanded latently HIV-1 infected cells, for instance, the gene BTB domain and CNC homology 2 (BACH2). We investigated HIV-1 promoter activity after integration into specific sites in BACH2. The HIV-1-based vector LTatCL[M] contains two fluorophores: 1.) Cerulean, which reports the activity of the HIV-1 promoter, and 2.) mCherry driven by a constitutive promotor and flanked by genetic insulators. This vector was inserted into introns 2 and 5 of BACH2 of Jurkat T-cells via CRISPR/Cas9 technology in the same and convergent transcriptional orientation of BACH2, and into the genomic safe harbour AAVS1. Single cell clones representing active (Cerulean+/mCherry+) and inactive (Cerulean−/mCherry+) HIV-1 promoters were characterized. Upon targeted integration of the 5.3 kb vector LTatCL[M] into BACH2, active HIV-1 promoters were gradually silenced as reflected by decrease in Cerulean expression over a period of 162 days in culture. Silenced HIV-1 promoters could be reactivated by TNF-α and Romidepsin. This observation was independent of the targeted intron and the transcriptional orientation. BACH2 mRNA and protein expression was not impaired by mono-allelic integration of LTatCL[M]. Our results show that the HIV-1 promoter is silenced when integrated into BACH2 without impairing BACH2 mRNA and protein expression. This might contribute to HIV-1 persistence, enabling infected T-cells to complete differentiation into a memory phenotype, persist, and clonally expand over time.


Introduction 51
Antiretroviral therapy (ART) blocks efficiently virus replication, however, does not cure 52 HIV-1 infection due to the presence of replication-competent but silenced proviruses 53 preferentially integrated in long-lived resting CD4 + T-cells (Finzi et al. 1997, Wong et al. 54 1997. Various factors and molecular mechanisms that result in HIV-1 latency have been 55 proposed (Ruelas and Greene 2013). One such factor might be the integration site of the 56 provirus, which has been suggested to not only be responsible for silencing the provirus, but 57 also supporting cell expansion, thus maintaining the size of the HIV-1 latent reservoir 58 (Maldarelli et  infected individuals who have been on ART for several years, it is conceivable that these 76 proviruses are inactive, although it remains unknown whether this presumed inactivity is a 77 result of integration site-dependent silencing of replication-competent proviruses or due to 78 defective proviruses. 79 To address the question of whether HIV-1 would be silenced upon integration into intron 5 of 80 BACH2 in the same transcription orientation, we employed a modified version of our dual-81 fluorophore HIV-1-based vector, LTatC [M], which reproduces features of active and latent 82 HIV-1 infections (Kok et al. 2018). This vector comprises two fluorescent reporter genes: 1.) 83 Cerulean, which reports the activity of the HIV-1 promoter, and 2.) mCherry, the expression 84 of which is driven by a constitutive promoter and further protected from position-effect 85 variegation by a pair of flanking genetic insulators to identify cells harbouring an integrated 86 vector (Uchida et  transcriptional orientations (s and c) ( Figure 1A). 122 Plasmids encoding Cas9 and guide RNAs were based on pX458 (Addgene plasmid 48138) 123 (Supplementary Table 1). Annealing of 100 µM 5ʹ′ phosphorylated primers for the guide  124  RNAs was performed using the following conditions: 80°C for 5 min, 65°C for 7 min, 60°C  125 for 7 min, 55°C for 7 min, 50°C for 7 min, 45°C for 7 min, 40°C for 7 min, 35°C for 7 min, 126 30°C for 7 min, 25°C for 7 min, and 4°C hold. The annealed guide RNA primers were 127 separately cloned into pX458. 128 All plasmid sequences were confirmed by next-generation sequencing. Briefly, plasmids were 129 gel purified, diluted to 0.2 ng/µl, and 1 ng of the plasmids were processed using the Illumina 130 Nextera XT DNA library Prep kit. The DNA libraries were subsequently paired-end or single-131 end sequenced with an Illumina MiSeq using the MiSeq Reagent Kit v3 (150-cycle). Plasmid 132 maps are shown in Figure 1A. in RPMI-1640 medium media supplemented with 10% fetal bovine serum (FBS) and 1% 144 Penicillin Streptomycin (10'000 units/ml Penicillin, 10 mg/ml Streptomycin). 145 At day 9 post transfection, cells were analysed using fluorescence-activated cell sorting 146 (FACS) and sorted at 20 cells per well in 96-well plates using a BD FACSAria™ III (BD 147 Biosciences). Two cell phenotypes were sorted: 1. Cerulean + /mCherry + and 2. Cerulean -148 /mCherry + . This was done for each nucleofected cell population: BACH2_i5s, BACH2_i5c, 149 BACH2_i2s, BACH2_i2c, AAVS1_s, and AAVS1_c. After 50 days, expanded cell cultures 150 were analysed by flow cytometry and subsequently single cell sorted. Longitudinal flow 151 cytometric analysis of the cells were done with the LSRFortessa II (BD Biosciences) and data 152 were analysed using the FlowJo Software v.10.0.8. (FLOWJO, LLC) ( Figure 1C). 153

Amplifying the junctions of LTatCL[M] integration into BACH2 and AAVS1 158
Genomic DNA was extracted from 5 million transfected Jurkat T-cells using the DNeasy 159 Blood and Tissue kit (Qiagen) and quantified by Quant-iT™ PicoGreen™ dsDNA Assay kit 160 (ThermoFisher Scientific). To verify the targeted integration of the dual-fluorophore vector 161 LTatCL (1/400 diluted rabbit anti-BACH2 antibody (PA5-23642, ThermoFisher Scientific) and 218 1/5'000 diluted rabbit-anti-GAPDH antibody (ab9485, Abcam)) overnight at 4°C. After 219 washing the membrane four times with 1x PBS-T, the secondary antibody, 1/10'000 diluted 220 goat anti-rabbit IgG H&L (HRP) (ab97051, Abcam) was added and the membrane was 221 incubated at RT for 1 h. The membrane was then washed four times with 1x PB-T. To 222 visualize the protein, Immobilon Crescendo Western HRP substrate (Merck) was added to the 223 membrane. Chemiluminescence was captured with the Stella system (model 3200, Matlab 224 Group Companies). GSH and we expect that the HIV-1 promoter will not be silenced when integrated in this 246 GSH. These six vector constructs and subsequently generated Jurkat T-cell clones were 247 named BACH2_i5s, BACH2_i5c, BACH2_i2s, BACH2_i2c, AAVS1_c, and AAVS1_s 248 ( Figure 1A and B). 249 Jurkat T-cells were nucleofected with linearized plasmids of the six LTatCL[M] constructs. 250 Nine days post transfection, the frequencies of stably transfected Cerulean + /mCherry + and 251 single mCherry + cells were between 0.1 and 1.8% ( Figure 1C). Cells were sorted to enrich 252 Cerulean + /mCherry + (i.e., active HIV-1 promoter) and single mCherry + (i.e., inactive HIV-1 253 promoter) cells ( Figure 1D). 254 To obtain monoclonal cell lines, a sequential flow cytometric sorting strategy was employed. 255 For all targeted integration variants, Cerulean + /mCherry + and single mCherry + cells were first  homologeous arms ( Figure 1A) and short PCR extension times. All monoclonal cell lines 274 contained a not targeted allele of the respective targeted integration site (Table 1). 275 Third, BACH2 mRNA expression was quantified in 11 Cerulean + /mCherry + monolonal cell 276 lines, representing each targeted integration variant. A decrease of BACH2 mRNA expression 277 could not be observed in any Cerulean + /mCherry + monolonal cell line (Figure 2A, Table 1). 278 Forth, the potential impact on BACH2 protein expression after targeted integration of 279 LTatCL[M] into BACH2 was investigated by Western blot analyses. In line with unaltered 280 BACH2 mRNA expression, BACH2 protein expression was not impaired compared to non-281 transfected Jurkat T-cells in all 35 monoclonal cell lines tested ( Figure 2B, Table 1).

282
In summary, we successfully performed targeted integration of the 5.3 kb HIV-1-based, dual-283 fluorophore vector LTatCL[M] into various loci of the human genome via CRISPR/Cas9-284 mediated homology-directed repair. We subsequently generated monoclonal cell lines 285 modelling

actively and latently HIV-1-infected cells with integrated LTatCL[M] in either 286
transcriptional orientation in the targeted genomic loci. 287

Silencing of the HIV-1 promoter in cells with integrated LTatCL[M] in BACH2 288
To study the HIV-1 promoter activity over time in Cerulean + /mCherry + monoclonal cell lines, 289 i.e., active HIV-1 promoter, the fluorescence profile was frequently monitored for 162 days. relative to BACH2. In contrast, in all 5 AAVS1_s and AAVS1_c Cerulean + /mCherry + 297 monoclonal cell lines the frequency of Cerulean + /mCherry + cells remained relatively stable 298 for 162 days (>80% in 4/5 cell clones) as observed for one BACH2_i5s clone ( Figure 3A). 299 Overall, the HIV-1 promoter when integrated into the BACH2 gene is silenced over time in 300 the majority of Cerulean + /mCherry + monoclonal cell lines whereas it remains active when 301 integrated into AAVS1. 302

Monoclonal single mCherry + cell lines harbour large internal deletions in LTatCL[M] 313
A total of 17 single mCherry + monoclonal cell lines, presumably representing latently HIV-1- This raises the question, whether BACH2 is an exceptional HIV-1 integration site promoting 347 HIV-1 latency or, vice versa, AASV1 is an exceptional HIV-1 integration site preventing 348 HIV-1 latency. In a future study, we will be expanding the repertoire of investigated HIV-1 349 integration sites. 350 In virally suppressed HIV-1-infected individuals, the provirus integrated into BACH2 has 351 been found predominantly in intron 5 in the same orientation as the gene and has been linked 352 with clonal expansion. However, in vitro in HIV-1 infected cell lines, integration has been 353 found to occur randomly in the BACH2 gene (Maldarelli, et al. 2014, Wagner, et al. 2014).

354
The preference of HIV-1 integration into BACH2 intron 5 in the same orientation in vivo Monoclonal cell lines expressing initially only mCherry, i.e., modelling latently HIV-1 393 infected cells, were found to contain large deletions in the LTR-tat-Cerulean cassette. Similar 394 deletions were already observed by us when we used a variant of our vector to generate 395 retroviruses for infection of cells (Kok, et al. 2018). There, we speculated that the deletions 396 were caused by error-prone HIV-1 reverse transcription (Bebenek et al. 1989, Patel and 397 Preston 1994) or copy-choice recombination (Sanchez et al. 1997, Simon-Loriere and Holmes 398 2011), which is also observed in cells from HIV-1-infected individuals (Ho et al. 2013). 399 However, our current system does not require reverse transcriptase prior to integration since 400 we are using CRISPR/Cas9-mediated targeted integration. Therefore, the deletions might be 401 due to recombination events triggered by the LTRs flanking the Cerulean cassette (Mager and 402 Goodchild 1989). Together with our previous observation (Kok, et al. 2018  points of amplicons were due to using different primer pairs for amplifications. 513