Repression of HIV-1 Reactivation Mediated By KRAB Fused CRISPR/dCas9 Proteins In Lymphoid And Myeloid Cell Models

Background: Despite antiretroviral treatment ecacy, it does not lead to the complete eradication of HIV infection. In addition, HIV-1 latency reactivation is a major challenge towards cure efforts. Two strategies to cure HIV-1 infection, the “shock and kill” is based on the eradication of the HIV-1 from the patient, and the “block and lock". The “Block and Lock” methodology aims to control HIV-1 latency reactivation, promoting a functional cure. The KRAB fused CRISPR/dCas9 (pdCas9KRAB) system was previously produced to control cell transcription. Based on this construct we developed a CRISPR RNAs (sgRNAs), to guide the pdCas9KRAB up to ve different sites in HIV-1 provirus sites to block HIV-1 latency reactivation. This process was mediated by phorbol esters and HDAC inhibitors. Results: We found ve sites in the HIV-1 provirus genome (LTR1-LTR5) that minimize CRISPR off-targets and transduced them in the lymphoid and myeloid HIV-1 latency models. One of the ve sgRNAs (LTR5) which binds specically in the HIV-1 LTR NFκB binding site was able to promote a robust repression of reactivation pattern in a HIV-1 latency lymphoid model stimulated with Phorbol 12-Myristate 13-Acetate (PMA) and Ingenol B (IngB), both potent protein kinase C (PKC) stimulators. Reactivation with HDAC inhibitors, such as SAHA and Panobinostat, showed the same strong inhibition of reactivation. Additionally, we observed a reduction of 100 times in HIV-1 RNA molecules, when reactivated IngB in myeloid HIV-1 latently infected U1 cells. Conclusion: Taken together, our results show that the KRAB fused CRISPR/dCas9 system can robustly prevent the HIV-1 latency reactivation process, mediated by PMA or IngB and SAHA or Panobinostat, both in myeloid and lymphoid HIV-1 latency. In addition, we demonstrated that KRAB repressor protein is crucial to

functioning as a transcription repression factor. KRAB plays a crucial role inducing heterochromatin state of the proximal DNA molecule by recruiting KRAB associated protein 1 (KAP1). This protein interacts with other repressive complexes such as heterochromatin protein 1 (HP1) and histone methyltransferases (HMT), which promotes the ampli cation of trimethylated H3K9 marker and chromatin remodeling. Altogether, these tools based on DNA speci c recognition and repressive epigenetic markers induction could act as a deep latency strategy [26,27]. Hence, the present study aims to provide a proof of concept of a possible use of CRISPR/dCas9 DNA recognition system fused with KRAB domain to maintain a repressive state in a latent-infected lymphoid and myeloid cell, independently of the reactivation process.

Materials and reagents
The cell line HEK293T used in the present study was obtained from the cell bank ATCC, from the English American type of culture collection, and maintained in DMEM medium (Dulbelcco's Modi ed Eagle Medium -11995073 -ThermoFisher -USA) supplemented with 10% of fetal bovine serum (A31608, ThermoFisher, USA).
The lymphocyte lineage cells used in this study was J.Lat 10.6 obtained from the Aids reagent program of the National Health Institute of the United States of America (NIH, USA). The myeloid strain was the U1 cell, which comes from U937 myeloid cells chronically infected with an HIV-1 clone. The lymphocyte and myeloid lines described above were kindly provided by Dr. Lucio Gama (Johns Hopking Medical School, MD, US). They were maintained in RPMI 1640 medium supplemented with 10% fetal bovine serum. All cell lines were cultured without antibiotics and kept in a conventional cell cultural incubator at 37 °C and 5% CO 2 . The cultured cells were tested for mycoplasma contamination using MycoAlert® Mycoplasma detection kit (LT07-318, Lonza, USA).
Selection of the best sgRNAs for the HIV-1 proviral genome and for AAVS1 control gene To identify and design sgRNAs that would be complementary to the HIV-1 genome, the HIV-1 subtype B sequence, clone HXB2, accession number AF033819 was previously acquired from the Los Alamos database was submitted in the CRISPR Pick tool which is made available in the Broad Institute genetic disorder platform (GPP). This platform designs the possible sgRNAs, align against the human genome sequence, GRCh38 (NCBI RefSeq v.10920210514), for minimize the off target and rank them. We also performed the same process described above to design a control sgRNA for AAVS1 DNA sequence.

Assembly of sgRNAs in the pdCas9Krab expression vector
The pdCas9Krab vector expresses the fused dCas9 and Krab proteins, both being regulated by the UbC promoter. The sgRNA is regulated by the U6 promoter that occurs in the same vector. To insert the nucleotide sequence for the sgRNA of choice, a digestion of 1.5 μg of the vector pdCas9KRAB with 10 UI/ μL of BsmBI (R0739S, New England Biolabs, USA) was performed following the manufacture's procedure. The sgRNA strings was synthetized as independent oligos by (Integrated DNA technologies, USA). After the digestion, lyophilized sgRNAs oligos (additional le 1) were resuspended for 100 μM in H 2 O DEPC.
The phosphorylation step was proceeded with the enzyme T4 PNK (M0201S, New England Biolabs, USA), accordingly manufacturer's protocols. Then, the reaction was submitted to 95 ºC for 5 min and 5 ºC for 1 min to join the two DNA strands.
The ligation of the sgRNAs sequences in the digested expression vector was performed T4 DNA Ligase enzyme (M0202S, New England Biolabs, USA), following manufacturer's procedure.
The ligation reaction was transformed in chemically competent bacteria JM109, previously prepared using Mix & Go Competent Cells -JM109 strain kit (T3003, Zymo Research, USA). Bacterial transformations were performed according to the manufacturer's recommendations. These were plated in LB medium supplemented with 50 μg/ml ampicillin and maintained at 37ºC for 24 h.
Con rmation of positive colonies was performed by colony PCR using the U6 forward primer and the one of sgRNA strand as a reverse primer (LTR (1-5) Reverse). The PCR cycling consisted of 10 min at 95 ºC, 35 cycles of 95 ºC for 30 s, 55 ºC for 30 s and 72 ºC for 30 s, ending with 10 min at 72 ºC. The PCR product was evaluated by electrophoresis on a 2% agarose gel and fragments of approximately 280 bp were observed (data not shown).

KRAB sequence excision
The sequence of the repressor protein KRAB was removed using 10 UI of NheI (Promega, USA) and after the resulting plasmid was relinked with 1 UI of T4 ligase (M1794, PROMEGA, USA), generating the vector pdCas9koKRAB. The excision was con rmed by sequencing the plasmid using the BigDye Terminator protocol (ThermoScienti c, USA), according to the manufacturer's recommendation (data not shown).

Lentiviral vectors production
HEK293T cells were seeded at 8.0x10 5 cells/mL per well in 6-well plates, 24 h before transfection procedures. The LTR1-5sgRNA/dCas9KRAB expression vectors were incubated with a mix solution containing 9 μL of 2.5 M CaCl2, 8 μg of total plasmid (3.53 μg of each expression vectors, 1.76 μg of VSV-G and 2.7 μg of psPAX) and H 2 O DEPC for a nal volume of 90 μL, nally 90 μL of 2X Hepes Buffer Saline (HBS) was added for DNA precipitation. The mixture was incubated at room temperature for 15 min and dripped into the well, which had the culture medium previously replaced by a 1 mL of Opti-MEM (31985070, ThermoFisher, USA). The transfection solution was kept in contact with cells for 6 h and after this time it was replaced by 2 ml of complete medium. The transfected cells were kept for 48 h in an incubator at 37 ºC, in an atmosphere with 5% CO 2 . After two days, the supernatant from each well was collected and ltered in a 0.22 μm. The solution with the lentiviruses was frozen at -80 ºC until the moment of transduction.

Transduction of lentiviral vectors in lymphoid and myeloid cells
The cells used for virus transduction containing sgRNA/dCas9KRAB construct were J.Lat 10.6 and U1.
For this purpose, 2.5x10 5 cells were plated in a 12-well plate in a nal volume of 1 mL. Then, 8 μg/mL polybrene (Merck -H9868) and 500 μL of the lentivirus solution was added to the culture medium The virus and cell solution were then submitted to centrifugation at 1000 g for 2 h at room temperature. The plates were then kept for 24 h in an oven at 37 ºC, in an atmosphere with 5% CO 2 , and 0.5 μg / mL of puromycin (Merck -USA -P8833) was added to select only the cells in which the construct was present. The optimum puromycin concentration was obtained by testing the drug sensitivity curve (data not shown). The transduced culture was kept in check until all sensitive cells had been removed and satisfactory cell growth was observed. This entire selection process took up to two weeks.
In addition to puromycin selection, transduction was con rmed by PCR. The PCR product was evaluated by electrophoresis on a 2% agarose gel and fragments of approximately 100 bp were observed (supplemental gure 1).

Measurement of green uorescent protein (GFP) in J-Lat 10.6 cell line model
Wild-type (WT) and transduced (LTRs) J.Lat 10.6 cells were plated at a concentration of 10 5 cells/mL, in 24-well plates and separated into three experimental groups: the untreated group, which did not receive the reactivating drug (negative control), the PMA group which received 1 μM of PMA and the IngB group which received 1 μM of ingenol B. The plates were kept for 24h at 37ºC, in an atmosphere with 5% CO 2 .
After incubation, cells were centrifuged at 300 G for 4 min and resuspended in 100 μL of PBS. The analysis was performed using ow cytometry through BD Accuri C6 device (BD Biosciences, USA).

HIV-1 RNA detection in U1 cell line model
The U1 cell activation process was conducted as previously described by Abreu et al [29]. U1 WT cells and U1 LTRs were plated, in duplicate, at a concentration of 2.0x10 5 per well. Two conditions per cell type were adopted: without treatment and with 1 µM IngB. Each group was kept for 48 h and then frozen at -80ºC. HIV-1 viral load was estimated by HIV-1 RealTime Ampli cation Kit through m2000 RealTime System (Abbott, IL, USA).

Statistical analysis
Statistical calculations were performed with one-way or two-tailored unpaired Student T test using Graph Pad Prism version 9.0.0. P values ≤ 0.05 were considered statistically signi cant. Experiments were performed in three independents replicates with two replicates per samples. Statistical data analysis of untreated J.Lat 10.6 ko/KRAB cells experiments were performed in two independent replicates with two replicates per samples.

CRISPR sgRNA binds speci c to HIV-1 LTR in silico
In order to nd the best sgRNAs that bind to HIV-1 provirus genome with maximum on target and minimum human genome off target ligation, we performed an in silico screening using the GPP (Genetic Perturbation Platform) sgRNA designer from Broad Institute. Through in silico analysis of the les generated by the platform, 98 possible sgRNAs were obtained (additional le 2). Firstly, we performed a screen to analyse only the sgRNA which bind in HIV-1 LTR or near locations. Among these, the ve best ranked sgRNAs were chosen based on the major parameter established by the database: sgRNA's ability to do not bind within the human genome. More off-targets mean a lower score assigned by the program and a lower placement of this sgRNA. Therefore, the sgRNAs chosen for the subsequent assays were LTR1 to LTR5 (table 1).
LTR1 and LTR2 sgRNAs bind to the U3 region of the 3' LTR. LTR5 sgRNA binds to NF-kB transcription factor binding site in the 5'LTR U3R region. In turn, LTR4 binds to the U3 region of the 5'LTR, and LTRs 3 bind to the region encoding NEF viral protein (Fig. 1A).

HIV-1 latency can be maintained by speci c CRISPR action in a lymphoid model
To understand whether the sgRNA inserted in the repression vector could limit HIV-1 replication, J.Lat 10.6 cells were transduced with the repressor constructs LTRs and subjected to PMA and IngB reactivation. A statistically signi cant reduction in GFP expression was observed in clones LTR1, LTR2, LTR4 and LTR5, in both reactivation treatments compared to treated WT and AAVS1 control. There is a signi cant reduction (p <0.0001) in the percentage of positive GFP LTR5 cells with a decrease of 72.31% and 78.66% in the PMA and IngB treatments, respectively, when compared with the WT treated with reactivators ( Fig. 1). LTR1, LTR2, LTR4 clones treated with LRA drugs also had a statistically signi cant reduction in GFP expression (p <0.05) compared to the treated WT, of 28.70%, 16.87% and 38.27%, respectively, despite being less effective than LTR5 in inducing a repressive state. LTR3 clone showed no statistically signi cant difference in reducing GFP expression when compared to treated WT and AAVS1 control (Fig. 1). The cells which were transduced with the empty control vector and with the AAVS1 control had no repressive effect in the GFP positive cells amount when compared with the WT clone.
KRAB protein impact in the maintenance of HIV-1 latency To understand whether the suppression of GFP expression in J.Lat 10.6 cells would be only by the presence of the CRISPR/KRAB construct or it would be possible that only the sgRNA and dCas9 expression could impact the HIV reactivation. Thus, we removed KRAB repressor sequence of the LTR2, LTR3, LTR4, LTR5 vectors, generating sgRNA knockout KRAB (LTR2-LTR5koKRAB) and treated them with PMA or IngB. For LTR2koKRAB, LTR3koKRAB we did not observe a signi cant reduction of GFP positive cells when comparing with WT and sgRNA negative control. In turn, both LTR4koKRAB and LTR5koKRAB have shown signi cant GFP reduction, 44.04% (PMA) / 53.6% (IngB) and 33.43% (PMA) / 31.95% (IngB), when treated with PMA or IngB, respectively (Fig. 2).
CRISPR sgRNA LTR5 repressor prevents HIV-1 reactivation by NFκB independent manner After con rmation of which was the best clone cell for repressing HIV-1 reactivation, we investigated whether this repression was obtained only because we administered PKC stimulation drugs. To eliminate the bias created due to use of phorbol ester drugs, LTR5 J.Lat 10.6 cells were treated with SAHA and Panobinostat (Pano), both potent HDAC inhibitors. We found that the LTR5 J.Lat 10.6 cells when treated with SAHA and Pano express an amount of GFP positive cells 35 and 24 times less, respectively when compared with WT cells (Fig. 3). These results show that the CRISPRdCas9-KRAB system can repress the latency reactivation through a NFκB independent mechanism.

HIV-1 latency can be maintained by speci c CRISPR action in a myeloid cell model
To examine whether the reactivation repression visualized in the J.Lat cell model could be replicated in a myeloid lineage, we transduced U1 cells with the LTR repressors that worked best in J.Lat10.6 (LTR1, LTR4, LTR5), stimulated or not IngB and measured HIV-1 RNA reactivation by qRT PCR (Fig. 4). In a nonreactivated state, U1 WT cells express 400000 RNA copies/mL, while LTR1 and LTR4 repressor cells express respectively 159446 and 190005 RNA copies/mL, almost two times less HIV-1 RNA molecules when compared to U1 WT cells. Additionally, when we observed the LTR5 repressor cells, in nonreactivated state, the expression of HIV-1 RNA molecules was a mean of 157036 RNA copies/mL, more than 2 times less expression compared with U1 WT cells in all three replicates (p<0.09). When we reactivated the U1 cells with IngB, we observed an expression reduction about 100 times in HIV-1 RNA molecules only for the U1 LTR5 cells, 1.5x10 6 RNA molecules/mL for IngB treatments compared with U1 Wt cells, 1,5x10 8 molecules/mL IngB treated. There was no signi cant alteration in the RNA molecules production of U1 LTR1 and LTR4 cells in comparison of U1 Wt cells in terms of HIV-1 RNA expression.

Discussion
Here, we describe a potential use of the repressive KRAB protein in association of CRISPR/dCas9 DNA recognition system. This system is addressed to HIV-1 proviral genome, almost all in the LTR sequence, in order to maintain a latent state despite LRA induction as a strategy of deep latency induction in lymphocytes and monocytes lineages. The HIV-1 latency reactivation of reservoirs is the major issue against a cure for HIV-1 infection. HIV-1 latency maintenance resembles endogenous retrovirus (HERVs) controlled by cells through repressive epigenetics mechanisms, such as HDCAs, DNMTs and HMTs, which recruit chromatin remodelling systems inducing heterochromatin formation [30]. After such alterations, the chromatin conformation imposes a structural impermeant to the access of transcription factors in HERVs and HIV-1 promoter, which implicates in transcription disruption. The molecular interplay between HERVs and HIV-1 has been demonstrated by the reactivation of some types of HERVs by HIV-1 Tat protein [31]. In contrast, somatic cells can e ciently control viral reactivation, unlike HIV-1, which under certain T cell stimulation overcomes these repressive epigenetic mechanisms and can be successfully replicated [32,33]. J.Lat is the most used lymphocyte lineage to test LRA response since it carries a latent integrated HIV-1 with a GFP reporter instead of the Env gene. Different types of J.Lat cells show different reactivation patterns. J.Lat 10.6 is one of the most responsive clone, with 80-90% GFP+ cells after phorbol ester (PMA and IngB) stimulation on average [28,29,34] [25][26][27]. These LRAs act through the PKC pathway, stimulating nuclear internalization of the NFκB transcription factor and simultaneously HIV-1 LTR activation. Saayman et al. have shown that CRISPR-sgRNA dCas9, when associated with VP64 or with a synergistic activation mediator (SAM) is capable of stimulating HIV-1 latency reactivation in J.Lat clones cells. J.Lat 10.6 showed 40-fold activation when compared with the control sgRNA cell [35]. Nevertheless, rather the reactivation induction, the present study shows a distinct approach, to prevent HIV-1 RNA reactivation even in the presence of an optimal concentration of LRAs. One of our six designed CRISPR constructs for the HIV-1 subtype B genome could inhibit latency reactivation process 160 times. The best sgRNA interacts exactly in the LTR enhancer region, in the NFκB ligation point thus it could explain why the repression rate was so robust when J.Lat 10.6 LTR5 cells were treated with PMA and IngB, both NFκB stimulation drugs. This nding is consistent with the results obtained by Saayman et al., which showed that the most responsive sgRNA was designed to the NFκB ligation LTR's sequence [35].
The bias produced for using a NFκB agonist to analyse the reactivation capability of latent-integrated HIV-1 in cells with the LTR5 construct was mitigated by using HDACi, which promotes the HIV-1 reactivation by a NFκB independent mechanism [36,37]. When we used SAHA and Panobinostat, both HDACi, in LTR5 JLat cells, we observed respectively 1% and 3% of GFP+ cells, showing a robust repression.
Despite being controversial whether myeloid cells act as latency reservoirs, some research showed that myeloid cells could promote viral latency in Gut Associated Lymphoid Tissue (GALT), lung, adipose tissue, and central nervous system (CNS) Myeloid cells as a latency sanctuary were also demonstrated in SIVmac infections in vivo [38]. Given the importance of myeloid cells in HIV-1 latency, our work shows that LTR5 sgRNA/dCas9 KRAB can attenuate phorbol-ester HIV-1 reactivation. Although we did not see the same range of repression as in J.Lat 10.6 cells, we realize that the same LTR5 sgRNA can function similarly in both lineages. The difference between the repressive state in J.Lat and in U1 cells could be explained by the distinct molecular mechanism of latency establishment [39,40]. In terms of U1 untreated cells we were unable to observe a statistically signi cant reduction (p <0.09) of HIV-1 RNA molecules in U1 LTR5 clones due to one of three replicates showed an outlier result. Probably, if we have performed more replicates, we may have observed a decrease of HIV-1 RNA molecules in U1 LTR5 clone supernatant.
The impact of KRAB protein in the HIV-1 latency establishment was demonstrated by several studies. Genome-intact provirus reservoirs of 64 HIV-1 elite controllers were often integrated in centromeric satellite DNA and in genes encoding for KRAB-ZNF proteins. Jiang and colleagues observed HIV-1 provirus from elite controllers being harboured in a repressive location and showing the robust repressive histone markers H3K9me3 and H3K4me1, which induce heterochromatin formation and impact provirus reactivation [41]. Interestingly, both repressive markers found in elite controllers are also induced by KRAB proteins that are encoded in our repressive vector. Moreover, KRAB-containing zinc nger protein ZNF304 was found as a naturally robust HIV-1 latency inductor in a genome-wide CRISPR knockout screening of HIV-1 infected Jurkat cell line [42]. Another KRAB-containing zinc nger, ZNF10, was demonstrated to repress LTR activity through interaction with NF-kB and SP1 binding sequence [43]. According to our hypothesis, KAP1, a protein recruited by KRAB, was reported to be responsible by the impairment of HIV-1 gene transcription, especially by inhibiting Tat's P-TEFb induction in both myeloid and lymphoid cells [44,45]. In myeloid cells, KAP1, in association with CTIP2, was observed to stimulated Tat degradation by SUMOylating throughout the proteasome pathway [44,46]. Recent ndings, in addition of corroborate our results of CRISPR/dCas9 fused KRAB protein, it improved the knowledge of the impact of KRAB protein in retrovirus transcription such as prototype foamy virus (PFV) showing that this protein negatively regulates PFV transactivator protein (Tas) and leading its to degradation through ubiquitination [47]. Our KRAB fused CRISPR constructs, in agreement with previous studies, show a robust HIV-1 transcription impairment, especially when targeted to the NFkB region. Furthermore, when KRAB protein was removed from the construct, the impact of HIV-1 reactivation was more prominent for all the LTR recognition CRISPR clones, although in J.Lat 10.6 LTR5 ko/KRAB clone we observed a signi cant decrease in HIV-1 reactivation compared with Wt clone. This effect is in according with previous works that demonstrated the allosteric inhibition of transcription promoted by dCas9 [48,49].
Our results are also consistent with those observed by Olson et al. which suggest a relevant inhibition of KRAB-fused CRISPR/dCas9 system in J.Lat 6.3 model, when reactivated with PMA plus ionomycin. Moreover, this study showed the promotion of H3K9me3 modi cation mediated by their constructs in HEK293T cells [50]. The present study does not access all epigenetics patterns stimulated by the ligation of KRAB fused CRISPR/dCas9 systems on HIV-1 LTR. However, we hypothesize that KRAB-KAP1 interaction plays a critical role in the reactivation repression since removing the KRAB protein from the system does not show the same repressive state upon LRA stimulus.
Further experiments such as a chromatin immunoprecipitation assay associated with a next generation sequencing could be done to access the LTR5 sgRNA ligation speci city and off-target possibility and whether this repressor could impact any cellular gene expression. An ex vivo approach using CD4 + T cells and macrophage cells from HIV-1 infected patients could be done to test with the repression we observed in cell culture could be replicated in vivo.

Conclusion
In summary, we demonstrate the robust impact of our KRAB-fused CRISPRdCas9 system in HIV-1 latency maintenance in both lymphoid and myeloid HIV-1 latency models. Additionally, we demonstrate that this system can be resistant to both types of HIV-1 LRA such as PKC agonists and HDAC inhibitors. Altogether, our results contribute to further studies investigating the development of a useful block and lock strategy and to make a step forward on the direction of a cure for HIV-1 infection.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.