Targeted Chromatinization and Repression of HIV-1 Provirus Transcription with Repurposed CRISPR/Cas9

The major barrier to HIV-1 cure is the persistence of latent provirus, which is not eradicated by antiretroviral therapy. The “shock and kill” approach entails stimulating viral production with latency-reversing agents followed by the killing of cells actively producing the virus by immune clearance. However, this approach does not induce all intact proviruses, leaving a residual reservoir. CRISPR/Cas9 has been utilized to excise integrated Human Immunodeficiency Virus (HIV) DNA from infected cells in an RNA-guided, sequence-specific manner. Here, we seek to epigenetically silence the proviral DNA by introducing nuclease-deficient disabled Cas9 (dCas9) coupled with a transcriptional repressor domain derived from Kruppel-associated box (KRAB). We show that specific guide RNAs (gRNAs) and dCas9-KRAB repress HIV-1 transcription and reactivation of latent HIV-1 provirus. This repression is correlated with chromatin changes, including decreased H3 histone acetylation and increased histone H3 lysine 9 trimethylation, histone marks that are associated with transcriptional repression. dCas9-KRAB-mediated inhibition of HIV-1 transcription suggests that CRISPR can be engineered as a tool for block-and-lock strategies.


Introduction
Human immunodeficiency virus type 1 (HIV-1) is the causative agent of acquired immunodeficiency syndrome (AIDS). The primary targets of HIV-1 are CD4+ T cells, which are diminished by HIV-1 infection, directly contributing to immunodeficiency. Current antiretroviral therapy (ART) halts viral replication and disease progression. However, discontinuation of ART leads to rapid rebound of HIV-1 replication, revealing that ART does not target persistent latent HIV-1 infection. This latent reservoir, with an estimated half-life of 44 months [1,2], has been shown to be established early at the time of infection and/or at the time of ART initiation [3]. Resting CD4+ T cells and other long-lived memory CD4+ T cell populations are posited to harbor the bulk of latent HIV infection [1,4,5]. This reservoir is established as a result of direct infection of quiescent naive or memory CD4+ T cells [6][7][8]; infection of CD4+ T cells differentially activated [9][10][11]; and the transition of infected, activated CD4+ T cells to a resting memory phenotype [12][13][14][15]. Multiple mechanisms, including lack of transcriptional activators, abundance of transcriptional repressors, stochastic fluctuation of Tat,

dCas9 Constructs
gRNA expression constructs were designed for HIV NL4-3 genome sequence using the online molecular biology software Benchling Software, 2018 (www.benchling.com). HIV-1 specific sequences were 21 bp in length, preceded by a protospacer adjacent motif (PAM) sequence "NGRRT" on the complementary strand, and followed by the Cas9 scaffolding region. Single gRNA sequences spanning the HIV genome were assigned specificity and activity scores on a scale of 01-00 using Benchling software. Specificity scores describe the off-target potential for gRNA binding compared to DNA libraries of interest [44][45][46]. The activity score predicts endonuclease activity of Cas9 at the target site, which is influenced by size and secondary structure of gRNA [44][45][46]. Guide sequences met a threshold specificity score greater than 50 (range 67.29-6.4), whereas there was a range of activity scores (range 3.16-6.4) ( Table 1), suggesting that all guides are specific for the HIV genome but may not be optimally active. Sequences were either cloned into expression vectors as described below or obtained as Ultra RNA Oligos (Integrated DNA Technologies, Coralville, IA, USA).  [44,46]. b Activity scores (01-00) are predicted endonuclease activity at the target site based on gRNA secondary structures. Higher scores indicate greater activity [44,45].

Lentivirus Packaging
Lentiviruses were generated in HEK293T cells by Polyethylenimine (PEI; Sigma Aldrich 408727, Natick, MA, USA) transfection. 24 h prior to transfection, 5 × 10 6 cells were plated in a T75 tissue culture flask. For transfection, 15 mg of total plasmid DNA was combined with 45 mg of PEI in 1.2 mL serum-free DMEM solution and incubated for 30 min prior to addition to cells. Single round HIV-1 was generated by cotransfection of pNL43-∆env-luciferase. After 48 h, viral supernatants were collected and syringe-filtered using 0.45 µm filter (Fisher SLHV033RS, Waltham, MA, USA) prior to storage at −80 • C. Viruses were titrated on CEM cells containing a HIV-1 Tat driven GFP reporter and measured by flow cytometry. The multiplicity of infection (MOI) was estimated by multiplying percent of GFP-positive cells by the total number of cells infected, and divided by volume of inoculum.

Transductions and Infections
Three million cells were transduced or infected with lentiviruses (MOI 0.1) by spinoculation (60 min, 1200× g) in a six-well tissue culture plate with 1 mL of media. Cells were washed with PBS and resuspended in the media. Jurkat T cell and J-Lat 6.3 cell lines stably expressing dCas9-KRAB were generated by transducing cells with a dCas9-KRAB lentivirus containing blasticidin resistance and selected in RPMI complete media containing 10 µg/mL blasticidin (InvivoGen ant-bl-1, San Diego, CA, USA).

HEK293T Transfections
HEK293T cells were plated 24 h prior to being transfected using a 4:1 molecular weight ratio of PEI to plasmid DNA. For 96-well assay, 10,000 cells were plated and transfected with 125 ng of total DNA. For gRNA screens, equal parts of dCas9-KRAB, gRNA, and HIV NL43-Denv-Luciferase plasmids were used. pCDN3.1 vector was a control for DNA input. After 48 h, transfected cells were lysed for luciferase assay using Bright-Glo (Promega E2610, Madison, WI, USA). Cell viability was monitored with CellTiter Glo (Promega C7570, Madison, WI, USA). Since Bright-Glo and CellTiter-Glo are directly proportional, interassay variability was adjusted by normalizing to cell input.

JLat Reactivation
JLat 6.3 stably expressing dCas9-KRAB was removed from media containing blasticidin and split at 200,000 cells/mL 2 days prior to nucleofection. Cells were electroporated per manufacturer protocol (Lonza #VACA-1003, Portsmouth, NH, USA) with specifications and minor modifications as described below. Three hundred nM of Ultramer RNA Oligos gRNA constructs were used per condition using program X-005 on Nucleofector 2b Device (Lonza, Portsmouth, NH, USA). Following electroporation, 500 mL of warmed RPMI complete with 20% FBS was added to the cuvette and allowed to rest for 10 min prior to being transferred to a 12-well tissue culture plate in total volume of 1.5 mL overnight. Cells were washed with PBS and resuspended in RPMI complete with 10% FBS for 4 days. On day 5, 100,000-200,000 cells were stimulated in duplicate with PMA (10 mg/mL) and ionomycin (2.5 nM) for 18 h in 96 well-plate. Conditions were pooled and harvested for flow cytometry or qPCR. Percentage of reactivation was determined by GFP expression compared to a scramble control. RNA expression was determined using qPCR targeting HIV-GFP reporter transcription to avoid any lentiviral interference. HIV-1 RNA expression was normalized to GAPDH expression.

Statistical Analysis
Statistical analyses were performed using GraphPad Prism (GraphPad Prism v 8.4.3 for windows, GraphPad Software, San Diego, California USA, www.graphpad.com). Student t-tests were performed. Statistical significance was considered to be p ≤ 0.05.

Repressing HIV Transcription with dCas9-KRAB
We wanted to explore mechanisms by which dCas9-KRAB inhibits HIV-1 transcription. We cotransfected HEK293T cells with NL43-∆env-Luciferase and dCas9-KRAB chimeric expression constructs plus candidate gRNAs to identify repressive targets. Guides 397 and 518, which flank the transcriptional start site (TSS) in the LTR, repressed HIV-luciferase expression by greater than 70%, compared to scrambled control and gRNAs targeting regions outside of LTR (Figure 1). Transfection of gRNAs alone did not significantly influence HIV-luciferase expression.
To determine whether dCas9-KRAB and gRNA repressed HIV expression in infected cells, HEK 293T cells were infected with VSVg pseudotyped NL4-3-∆env-Luciferase and transfected with dCas9-KRAB and respective gRNAs. Compared to the control gRNA, LTR targeting gRNA resulted in an approximately 50% decrease in luciferase expression (Figure 2A). Combining gRNAs did not enhance repression. Diminished HIV-1 proviral transcription was confirmed by RT-qPCR. HIV RNA transcription per HIV DNA template was decreased by 50% to 90% with gRNA targeting the LTR, compared to the control gRNA ( Figure 2B). Proviral HIV-1 transcription was monitored over time by assessing luciferase and RNA expression. Proviral transcription was repressed in infected HEK293T cells for at least 5 days and up to 14 days, suggesting that dCas9-KRAB and guide RNAs facilitated a persistent decrease in HIV transcription ( Figure S1). These results demonstrate the ability to actively repress HIV-1 proviral transcription with dCas9-KRAB.

dCas9-KRAB Inhibits Reactivation of Latent HIV
We examined if dCas9-KRAB inhibited reactivation of transcriptionally repressed latent provirus. For this experiment, we electroporated J-Lat 6.3 cells that stably express dCas9-KRAB with gRNAs. J-Lat 6.3 are a clonally selected cell line that contains a single copy of HIV-1 with GFP inserted in nef and has low basal HIV expression, which is induced in response to T cell activation and latency reversing agents [43]. J-Lat-dCas9-KRAB cells electroporated with specific gRNAs were stimulated with PMA plus ionomycin, and HIV-1 transcription was assessed by RT-qPCR to measure HIV-1 mRNA. RNA expression in which the LTR specific gRNA prevents induction of transcription compared to control gRNA was reduced by 60% (Figure 3). Blocking of reactivation of HIV transcription indicates that dCas9-KRAB prevents reactivation and maintains repression of latent HIV-1 provirus. Figure 3. dCas9-KRAB inhibits reactivation of latent proviral HIV. J-Lat 6.3 cells that stably express dsaCas9-KRAB were electroporated with short RNA gRNA targeting LTR (397) or scrambled control. Cells were activated with PMA + Ionomycin, and HIV expression was monitored by qRT-PCR. These data include results from five independent experiments. Error bars represent standard error, and Student t-test was performed to determine significance. ** indicates p ≤ 0.01.

dCas9-KRAB Repression of HIV-1 Transcription Correlates with Epigenetic Modifications
The KRAB domain has been shown to function as a transcriptional repressor through its ability to interact with co-repressors, such as KAP-1, which in turn recruits chromatin modifiers like histone deacetylases and histone methyltransferases [49][50][51][52]. We hypothesized that dCas9-KRAB specifically represses transcription by changing the chromatin landscape at the HIV-1 LTR.
We performed ChIP-qPCR on chromatin prepared from HEK293T cells infected with HIV-1 and transfected with dCas9-KRAB and gRNAs. Protein-DNA complexes from infected cells were precipitated with antibodies against acetylated histone3, a histone mark that correlates with active transcription, and H3K9me3, a histone mark that correlates with transcriptional repression. ChIPs show that cells harboring dCas9-KRAB and specific gRNAs targeting the 5 LTR of HIV-1 had approximately 50% decrease in acetylated histone-3 at +30/+134 base pairs where the repressive nuc-1 is located [53], compared to control gRNAs (Figure 4 and Figure S2). Consistent with dCas9-KRAB facilitating repressive epigenetic changes at the 5 LTR, we observed three-to five-fold higher levels of H3K9me3 at nuc-1 when transfected with LTR gRNA compared to control gRNA (Figure 4 and Figure S2). As acetylated Histone 3 is a mark of open chromatin and H3K9me3 is a mark of repressed or closed chromatin, these data indicate that dCas9-KRAB represses transcription by recruiting epigenetic modifiers such as histone deacetylases and histone methyltransferases.  Figure S2). Error bars represent standard deviation, and a student t-test was performed for statistical significance. **** indicates p < 0.001.

Discussion and Conclusions
The presence of latent but replication-competent HIV-1 provirus in long-lived resting and memory CD4+ T cell populations is a major barrier to curing HIV/AIDS. Strategies to purge or shock and kill latent provirus include inhibitors of histone acetyltransferases, methyltransferase, Bromodomain (BRD) factors, and T-cell-signaling agonist, as well as engineering transcriptional activators, including using chimeric CRISPR/Cas9 or ZnF transcriptional activators [36,37,54]. One drawback to these approaches is that only a fraction of the HIV-1 provirus is transcriptionally induced or is accessible to DNA-modifying agents. CRISPR/Cas9 and targeted endonucleases have been successfully used to excise HIV-1 provirus in vitro [32][33][34] and in animal models [31,35]. Efforts to repress or maintain a transcriptionally silent HIV-1 provirus have also been explored. Block and lock strategies include using molecules that sequester the HIV transcriptional activator Tat, including inhibitor RNAs [55,56] and didehydro-cortistatin A (dCA) [57]. We demonstrate that a chimeric protein that includes dCas9 fused to the KRAB repression domain inhibits HIV-1 transcription and prevents the induction of latent provirus through epigenetic-mediated mechanisms that target the LTR.
We show that targeting dCas9-KRAB to the 5 LTR of HIV-1 results in robust repression of HIV transcription, compared to targeting other regions of HIV-1 genome. These results are consistent with recent findings showing that repurposed CRISPR/Cas9 fused to transcriptional activators targeted to the 5 LTR reverse HIV latency in vitro [36,37,54]. Furthermore, our results demonstrate that dCas9-KRAB are facilitating post-translational histone modifications at the LTR, including increased levels of H3K9me3 and decreased levels of acetylated H3, compared to control cells consistent with epigenetic changes associated with HIV-1 latency. KRAB recruits the cofactor KAP1 (also known as TRIM 28 and TIF1b) to promote durable heritable epigenetic chromatin modifications, and has been proposed to maintain heterochromatin and repress transcription of endogenous retroelements [38][39][40][41][42]58,59]. It should be noted that KAP1 activity may be context-dependent, and has been suggested to be an activator as well as a repressor for HIV-1 transcription [60].
In summary, we demonstrate a proof of concept that CRISPR technologies can be repurposed to directly target HIV-1 provirus to repress proviral transcription by promoting repressive epigenetic modifications. These studies provide an additional tool to extinguish HIV-1 proviral expression and suggest the potential of using engineered targeting factors as a strategy for long-term repression and permanent inactivation of the HIV-1 provirus.
Author Contributions: A.O. and B.B. contributed equally to this work, including designing and performing experiments, evaluating data and assembling and writing the manuscript. S.L. generated dCas9 reagents and helped write sections of the manuscript. M.G. did initial exploratory experiments that provided the foundation for the project. W.W.W. supervised the creation of critical reagents and was involved in conceiving the project, designing experiments and editing of the manuscript. A.J.H. was the principal investigator and involved in all aspects of the project. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by grants to A.H. and W.W. from amfAR and NIH (NIH R01 AI138960). In addition, services were provided by the Providence/Boston CFAR Basic Science Core (P30AI042853).

Conflicts of Interest:
The authors declare no conflict of interests.