key: cord-0024167-4zdm3e2v
authors: Herskovitz, Jonathan; Hasan, Mahmudul; Patel, Milankumar; Blomberg, Wilson R.; Cohen, Jacob D.; Machhi, Jatin; Shahjin, Farah; Mosley, R. Lee; McMillan, JoEllyn; Kevadiya, Bhavesh D.; Gendelman, Howard E.
title: CRISPR-Cas9 Mediated Exonic Disruption for HIV-1 Elimination
date: 2021-11-10
journal: EBioMedicine
DOI: 10.1016/j.ebiom.2021.103678
sha: dbaf683bb884209c0e18d65c0070bd341417eb24
doc_id: 24167
cord_uid: 4zdm3e2v

BACKGROUND: A barrier to HIV-1 cure rests in the persistence of proviral DNA in infected CD4+ leukocytes. The high HIV-1 mutation rate leads to viral diversity, immune evasion, and consequent antiretroviral drug resistance. While CRISPR-spCas9 can eliminate latent proviral DNA, its efficacy is limited by HIV strain diversity and precision target cell delivery. METHODS: A library of guide RNAs (gRNAs) designed to disrupt five HIV-1 exons (tat(1-2)/rev(1-2)/gp41) was constructed. The gRNAs were derived from a conseensus sequence of the transcriptional regulator tat from 4004 HIV-1 strains. Efficacy was affirmed by gRNA cell entry through transfection, electroporation, or by lentivirus or lipid nanoparticle (LNP) delivery. Treated cells were evaluated for viral excision by monitoring HIV-1 DNA, RNA, protein, and progeny virus levels. FINDINGS: Virus was reduced in all transmitted founder strains by 82 and 94% after CRISPR TatDE transfection or lentivirus treatments, respectively. No recorded off-target cleavages were detected. Electroporation of TatDE ribonucleoprotein and delivery of LNP TatDE gRNA and spCas9 mRNA to latently infected cells resulted in up to 100% viral excision. Protection against HIV-1-challenge or induction of virus during latent infection, in primary or transformed CD4+ T cells or monocytes was achieved. We propose that multi-exon gRNA TatDE disruption delivered by LNPs enables translation for animal and human testing. INTERPRETATION: These results provide “proof of concept’ for CRISPR gRNA treatments for HIV-1 elimination. The absence of full-length viral DNA by LNP delivery paired with undetectable off-target affirms the importance of payload delivery for effective viral gene editing. FUNDING: The work was supported by the University of Nebraska Foundation, including donations from the Carol Swarts, M.D. Emerging Neuroscience Research Laboratory, the Margaret R. Larson Professorship, and individual donor support from the Frances and Louie Blumkin Foundation and from Harriet Singer. The research received support from National Institutes of Health grants T32 NS105594, 5R01MH121402, 1R01Al158160, R01 DA054535, PO1 DA028555, R01 NS126089, R01 NS36126, PO1 MH64570, P30 MH062261, and 2R01 NS034239.

Defining the scientific pathways required to eliminate the human immunodeficiency virus type one (HIV-1) from its infected human host remains a global health concern for 38 million infected people. To avoid HIV-1 reactivation from latently integrated proviral DNA in CD4+ T cells and mononuclear phagocytes, lifelong antiretroviral therapy (ART) is required. ART protects against virus-induced reductions in numbers and function of CD4+ T lymphocytes and progression to the acquired immunodeficiency syndrome (AIDS) [1] . Thus, a functional cure for HIV-1 is needed to eliminate dependence on and stigma of lifelong ART.

While curative measures for HIV-1 are under development, gene editing stands alone in having the ability to remove the virus independently from host antiviral adaptive immunity. Although concurrent viral elimination strategies that include broadly neutralizing antibodies (bnAbs) [2] , immune modulation [3] , and chimeric antigen receptor (CAR)-T cells [4, 5] show promise in mice, selection for bnAb resistant strains [6] , viral rebound [3, 7] , and adverse side effects [8] have been reported in monkeys and humans. Treatments notably depend on host FcR-mediated cellular killing or antigen presentation in MHC-I and, as such, rely on host immunity for viral clearance. Thus, gene therapy uniquely offers the ability to block the propagation of HIV-1 infection by directly targeting either host cell receptors or HIV-1 proviral DNA itself. The viral co-receptor CCR5 can be ablated ex vivo [9, 10] while latent viral DNA can be excised in live animals [11] . Based on these results, gene therapy for HIV-1 elimination is under active clinical investigation (NCT02140944, NCT03164135, NCT03666871).

Nonetheless, applying gene therapy towards HIV-1 cure strategies has met several obstacles. One of the most significant rests in viral DNA sequence diversity. With a mutation rate of 1 in 1700 nucleotides [12] , HIV-1 quasispecies easily accumulate during the course of viral infection [13] . Divergent strains with increased replicative fitness evade host immunity in individuals and circulate through populations [13] as evidenced by the emergence of 11 global HIV-1 subtypes (also referred to as clades; A-L) [14] . HIV-1 subtypes B and C are the most prevalent clades in Europe and the Americas (10.2%) [15] and worldwide (46%) [16] , respectively. Thus, gene therapies designed to broadly target then eliminate HIV-1 must overcome strain diversity.

We posit that clustered regularly interspaced short palindromic repeat (CRISPR) approaches can meet the challenge of viral diversity. Potential broad spectrum proviral DNA targets of the ten kilobase pair (kb) HIV-1 genome include the viral promoter long terminal repeat (LTR) [17] [18] [19] [20] and viral transcription factor tat [21] [22] [23] . CRISPR gene editing can cleave two alternatively spliced HIV-1 tat exons. Thus, CRISPR can simultaneously cleave essential rev and env genes that are transcribed from the same portion of the genome but in alternate open reading frames [21] .

Defining the optimal gRNAs to eliminate HIV-1 infection and implementing the approach towards human trials is a challenge. First, computational assessments suggest that proviral targeting choices require affirmation in patient-derived HIV-1 isolates. Indeed, previously published LTR-directed gRNAs likely possess limited cleavage efficiencies during natural infection [24] . Second, gRNAs must inactivate HIV-1 transcription precluding the assembly of functional virions en trans from intact viral operons. Third, optimal combination gRNAs must minimize any emergence of CRISPR-resistant escape mutants [20, 23] . While multiplexed clustered regularly interspaced short palindromic repeats (CRISPR)-associated protein 9 (Cas9) treatments were designed to account for HIV-1 strain variation [25] and minimize viral escape [21] , testing against transmitted founder viral strains was not completed. Fourth, a CRISPR-Cas9 delivery system seeking to find and eliminate the virus from its natural target cells must be employed.

Accepting each of these needs, we developed then characterized a CRISPR-Cas9 system capable of eliminating diverse proviral HIV-1 strains. The approach taken mirrors the "mosaic" HIV-1 vaccine design currently in late-phase clinical trials (NCT03964415, NCT03060629) serving to immunize those persons with a high risk of acquiring infection. The vaccine employs HIV-1 gag/pol/env antigens encoded from a global composite of viral strains [26] . The data offered support the concept that CRISPR-Cas9 gRNAs designed against a multiclade HIV-1 consensus sequence, hereafter termed "mosaic gRNAs", can prove effective against latent viral infection. The gRNAs targeting genetically conserved and overlapping tat/rev exons maximized CRISPR-based attenuation of HIV-1 infection. Both lentiviral and lipid nanoparticle (LNP) delivery of CRISPR-Cas9 HIV-1 excision therapies proved successful in blocking HIV-1 reactivation from latently infected CD4+ T and monocytes and primary human macrophages. Delivery of the potent gRNAs is a critical barrier to translation and hence we verified and affirmed four different delivery modalities for validation of the findings. The data, taken together, bring the delivery of mosaic gRNAs to excise HIV-1 proviral DNA towards a workable HIV-1 elimination strategy that can be deployed in an infected human host.

The HIV-1 genome was analysed, and DNA multiple sequence alignments were made for each LTR or exon. This was completed according to the known HXB2 reference strain (https://www.hiv.lanl. gov/content/sequence/HIV/MAP/landmark.html) from the Los Alamos National Library (LANL) HIV sequence database (https://www. hiv.lanl.gov/) using all 4004 complete sequences available through 2018. The exported FASTA files were input to WebLogo 3 (http://

Evidence before this study Clustered regularly interspaced short palindromic repeat (CRISPR)-associated protein 9 (Cas9) approaches were used to eliminate latent integrated proviral DNA from human cells in infected humanized mice. These prior works promoted proof of concept data sets that the human immunodeficiency virus type one could be removed from cells and tissues. However, the perils of CRISPR-Cas9 gene-editing strategies, including potential off target toxicities, recombination, viral excision sensitivity and specificity, and precision delivery to viral target cells were notable approach limitations. The detection and consequent excision of diverse sequences from a broad range of viral strains, the avoidance of secondary adverse events, and the generation of optimal therapeutic cargo delivery to groups of latently infected cells were known limitations. All would enable HIV-1 DNA elimination from infected animals and ultimately for infected humans.

CRISPR-Cas9 gRNAs designed against multiclade HIV-1 consensus sequences were proven effective against a broad range of viral strains. The gRNAs which targeted genetically conserved and overlapping tat/rev exons facilitated CRISPR-based activities facilitating mosaic viral gene disruptions. Lentiviral and lipid nanoparticle (LNP) delivery of the CRISPR-Cas9 HIV-1 gene editing systems successfully block HIV-1 reactivation from prior latently infected CD4+ T cell and monocyte-macrophage cell lines. Cell-directed delivery of CRISPR-Cas9 is an essential element for HIV-1 elimination.

The HIV-1 global epidemic continues unabated even with the widespread use of antiretroviral therapy. While those infected live longer and maintain a quality-of-life, viral growth is sustained as antiretroviral drugs are virostatic and viral mutations continue in the absence of sterilizing therapy. These data demonstrate that it is possible to eliminate HIV-1 infection through gene editing approaches. A sterilizing HIV-1 cure is possible.

weblogo.threeplusone.com/create.cgi) and the resulting plain text data table of positional entropy was graphed in heat-map form. Consensus sequences lacking gaps for the entire HIV-1 genome and tat using the LANL multiple sequence alignments were generated using SnapGene Ò software (GSL Biotech). CRISPR gRNAs for HIV-1 tat were designed using CHOPCHOP v3 [27] (https://chopchop.cbu.uib.no) and Broad Institute GPP sgRNA Designer (https://portals.broadinstitute. org/gpp/public/analysis-tools/sgrna-design) using tat consensus sequence as the specified target and human GRCh38 as the host organism. The top 35 gRNA candidates were inspected in WebLogo to identify the 8 gRNAs with the highest degrees of conservation among HIV-1 strains. Guide RNA conservation percentages for the full 20 base pair (bp) target were also determined using SnapGene Ò . The specificities of tat-and control-targeting gRNAs were predicted using Broad Institute GPP sgRNA Designer and CRISPR-OFF webserver v1.1 (https://rth.dk/resources/crispr/crisproff/).

Putative off-target loci for top gRNA candidates were selected from top-ranked candidates in CRISPR-OFF and Cas-OFFinder (http:// www.rgenome.net/cas-offinder/) tools. These sites were selected based on highest predicted off target toxicity. Testing in these sites were selected for primary screening.

Escherichia Coli (E.Coli, STBL3, Stellar) was used for transformation and streak plated on selective antibiotic-containing (ampicillin, 100mg/mL, or kanamycin, 25mg/mL) on Luria Bertani (LB)-agar plates then incubated for 16 hours at 37 o C. Single clones were then inoculated in LB broth with noted antibiotic and cultured (16 hours, 32 o C, 185 rpm orbital shaking). Clones were stored in 1:1 glycerol/water at -80 o C. CRISPR plasmid constructs containing mosaic gRNAs were cloned by T4 ligase oligonucleotide insertion (Table S1) in digested px333 (Addgene #64073) or pLentiCRISPR-RFP657 (Addgene #75162) and transformed into STBL3 E. coli. The pX333 used was a gift from Andrea Ventura (Addgene plasmid #64073) http://n2t.net/ addgene:64073 ; (Addgene #64073) [28] . In-Fusion HD Cloning (Takara #638920) was utilized to synthesize tat mutants on pHIV-1 NL4-3-Dnef-eGFP background. Maxiprep purified plasmids were transfected into HEK293FT cells using PEI or into ACH2 / U1 leukocytes via TransIT-Jurkat reagent (MirusBio #2120) or TransIT-X2 reagent (Mir-usBio #6000). Supernatants were collected 72 hours after media replacement (containing PEI) or plasmid transfection (TransIT-Jurkat or TransIT-X2).

Human embryonic kidney cells (HEK293FT) cells were grown in Dulbecco's Modified Eagle's Media (DMEM) containing 10% v/v fetal bovine serum (FBS), 100 units/mL penicillin and streptomycin (Pen-Strep), 0.5 mg/mL Geneticin Ò , 1% v/v MEM non-essential amino acids (MEM NEAA), and 1 mM sodium pyruvate. Adherent HEK293FT were detached from culture plates using 3 mL Trypsin-EDTA (0.25%) and passaged at 3 x 10 6 cells in 15mL growth medium every 3-4 days. CEM-ss CD4+ T cells (NIH ARP #776), HIV-1 latently infected ACH2 T cells (NIH ARP #349), full length HIV-1 latently infected lymphocytic Jurkat JLat 8.4 cells (NIH ARP #9847) and HIV-1 chronically infected U1 pro-monocytes (NIH ARP #165) were cultured in RPMI containing 10% v/v FBS, 100 units/mL PenStrep, and 100 units/mL L-glutamine. Suspension CEMss, ACH2, and U1 cells were maintained at 0.2-1 x 10 6 cells/mL by passaging in 15 mL growth medium every 3-4 days. All cells listed were propagated in vented cap 75 cm 2 U-shaped flasks at 37°C and 5% CO 2 .

Primary human monocytes were obtained by leukapheresis from HIV-1,2 and hepatitis B seronegative donors and purified by countercurrent centrifugal elutriation. Monocytes were cultured in Dulbecco's Modified Eagle Medium (DMEM) containing 4.5 g/L glucose, 200 mM L-glutamine, and 1 mM sodium pyruvate supplemented with 10% heat-inactivated human serum, 50 mg/mL gentamicin, and 10 mg/mL ciprofloxacin. Human macrophage colony-stimulating factor (MCSF, 1000 U/mL) was included in the culture media for seven days to facilitate monocyte to macrophage differentiation. Half media exchange was performed at alternate days. After differentiation was achieved, cells were switched to the same media sans MCSF. MDMs were then used for in vitro assays to study TatDE LNP induced protection against viral infection. All cells were grown in plate format at 37°C in a 5% CO 2 incubator.

HEK293FT cells were seeded 24 hours prior to transfection in DMEM growth medium lacking geneticin at 5 x 10 5 cells / well in 12well culture plates (px333-pHIV-1 co-transfection screen) or 6 x 10 6 cells / 10-cm culture dish (stock HIV-1 generation). Four hours prior to transfection, medium was changed to OptiMEM containing 250 mM chloroquine diphosphate (1 mL per well or 8 mL per 10-cm dish). Polyethyleneimine (PEI; linear MW 25,000 Da) stock (1 mg/ mL) was diluted in OptiMEM to account for a 3:1 w/w PEI:DNA ratio.

For gRNA co-transfection screening, 1.5 mg px333 plus 1.5 mg pHIV-1 were diluted in 150 mL OptiMEM, which was then further diluted with 150 mL OptiMEM / PEI. Viral stock transfections were similarly prepared using 10 mg pHIV-1 diluted in 1.5 mL OptiMEM, which was then further diluted with 1.5 mL OptiMEM / PEI. DNA / PEI mixtures were incubated for 15 minutes at 25°C prior to addition to cell cultures (100 mL / well in biological triplicates = 1 mg DNA / well; 3 mL / 10 cm dish = 10 mg DNA / 10 cm dish). Sixteen hours post-transfection, DMEM growth medium lacking geneticin was replaced and cells were cultured for 72 hours until experimental or lentivirus harvesting. ACH2 and U1 cells were seeded in 90 mL at 5 x 10 4 cells / well in round-bottom 96-well culture plates 24 hours prior to transfection with pLentiCRISPR-RFP657. pLenti-CRISPR-RFP657 plasmid (1.08 mg) was diluted in 81 mL OptiMEM, then supplemented with 3.24 mL TransIT Ò -Jurkat transfection reagent. After 20 minutes incubation at 25 o C, 10 mL of the transfection mixtures were added per well in biological replicates (n = 8) and cultured for 72 hours followed by chemical stimulation to assay viral outgrowth.

Electroporation was performed in 4D-Nucleofector TM X Unit (Lonza # AAF-1002X) device using Lonza Nucleocuvette TM and Ingenio Ò Electroporation solution. The optimal program was used after finalizing assay conditions and analysing the resultant data sets by flow cytometry ( Figure S1 ). Optimization was based around using the Cell line Optimization 4D-Nucleofector TM X Kit (Lonza # V4XC-9064) according to manufacturer protocol to find the optimal solution + pulse condition. Mirus Ingenio electroporation solution with different pulse code was also tested and used. For optimization, cells were spun down (150xg, 10 minutes) and supernatants removed. Cells were resuspended in either Cell Line solution from Lonza with supplements or Ingenio electroporation solution. pmaxGFP from Lonza was added to the cells at an equivalent ratio and cells were electroporated using different pulse codes. Cells were then incubated at RT for 10 minutes, then supplemented with preincubated RPMI media and maintained at 37°C in 5% CO 2 incubator. 48 hours post electroporation, cells were analysed by flow cytometry and the results were calculated as % GFP-positive of gated singlet cells.

HIV-1 NL4-3-Dnef-eGFP and HIV-1 NL4-3-Dnef-eGFP-Dtat were synthesized by reverse-genetics from supernatants of transfected HEK293FT cells.

Supernatants were passed through 0.45 mm filters, then ultracentrifuged to purify and concentrate the virus. HIV-1 viral stocks were titered by reverse transcriptase (RT)-qPCR (Takara #631235) to determine viral RNA copies / mL. Multiplicities of infection (MOI) for HIV-1 infection in CEMss T cells (NIH ARP #776) were calculated as viral RNA copies/cell. CRISPR-transducing lentivirus was commercially prepared at the University of Iowa Viral Vector Core Facility and titered using digital droplet PCR. ACH2 T cells (NIH ARP #349) were transduced with CRISPR-encoding lentivirus at designated MOIs. Primary human monocyte derived macrophages were infected using HIV-1 ADA at an MOI of 0.01.

ACH2, Jlat and U1 cells were stimulated three days following transfection or transduction or LNP treatment with recombinant tumour necrosis factor-alpha (TNFa; 15 or 20 ng / mL) or by phorbol 12-myristate 13-acetate (PMA; 20 ng/mL or 50 nM) in both treated and untreated group. Subsequently, RT assay from cell supernatant or ddPCR from extracted RNA was performed to measure stimulation index.

Nucleic Acid Isolation: Biological replicates were pooled and fractioned for DNA and RNA extraction. DNA was harvested from experimental pooled biological triplicates using NucleoSpin Tissue XS Micro kit (Machery-Nagel #740901). RNA was harvested from pooled biological triplicates (cellular pellets) using TRIzol TM solution (Invitrogen #15596026). In short, cells were lysed using TRIzol followed by a DNase I treatment (Zymo Research #E1010), and mechanical lysis (30 Hz, 10 minutes) using the TissueLyzer II instrument (Qiagen). The subsequent RNA isolation was performed according to the TRIzol User Guide provided by Thermofisher and scaled proportionally based on the initial TRIzol volume used. Isolated RNA was then converted to cDNA (Thermo Scientific #AB1453B).

PCR: CRISPR-mediated excision of proviral DNA was determined by PCR using 150 ng template DNA, 12.5 mL PrimeTime Gene Expression Master Mix (IDT #1055772), 1 mL (10 mM) each of forward and reverse primers, and water bringing the total volume to 25 mL. Table S2 . PCR final contents were run in 1.5% (w/v) agarose/TAE gels by electrophoresis at 75V for 35 minutes. Gel extracted PCR amplicons, and total PCR contents were sequenced and subjected to Synthego Inference of CRISPR Edits (ICE) Analysis v2.0 (https://ice.synthego.com/#/). RT-q-PCR: For quantitative reverse transcriptase (RT) PCR (RT-qPCR), 10 ng RNA was treated with DNase I (Zymo Research #E1010). Verso cDNA synthesis kits (Thermo Scientific # AB1453A) were used combining 1mL of DNase I treated RNA and random hexamers up to a volume of 10mL. Reverse transcriptase reactions were then carried out (42 o C for 30 minutes, 95 o C for 2 minutes) with the Verso kit. RT-qPCR was performed in technical triplicates using standard curve absolute quantitation for gapdh (1 diploid for genome = 6.6 picograms DNA extracted from ACH2 cells) and spCas9 (pLenti-CRISPR-RFP657; 10 9 copies equal 12.01 ng according to MW 7.22 x 10 6 Da). Reactions were performed in 10 mL volumes containing 1 mL cDNA template, 5 mL PrimeTime Gene Expression Master Mix (IDT), 0.5 mL of forward and reverse primer (10mM) (5'-CCCAAGAG-GAACAGCGATAAG-3'; 5'-CCACCACCAGCACAGAATAG-3'), and 0.5 mL 20x GAPDH endogenous control probe (Applied Biosystems #4333764T) or 200 nM spCas9 probe (56-FAM/ATCGCCAGA/ZEN-/ AAGAAGGACTGGGAC/3IABkFQ). All RT-qPCR reactions were performed as technical triplicates using the following thermal cycler conditions: 3 minutes 95 o C; 60 cycles of 95 o C for 15 seconds, 60 o C for 60 seconds. The particles were analysed by transmission electron microscope (TEM) (FEI Tecnai G2 Spirit) after Nanovan Ò (Nanoprobes #2011-5ML) stain. Diluted LNP samples were deposited and incubated on mica for two minutes, subsequently washed and dried before performing atomic force microscopy. LNPs were also analysed by dynamic light scattering using Malvern Zetasizer Nano (Malvern Panalytical) for size and polydispersity index after preparation using a previously described method [29] .

Tracking of the LNPs was achieved by including 80 mg of rhodamine DHPE (Invitrogen TM Lissamine TM Rhodamine B 1,2-Dihexadecanoyl-sn-Glycero-3 Phospho-ethanolamine, Triethylammonium Salt; DHPE) in the lipid film making process, followed by a LNP fluorescent tag. Next, px333DE DNA was added to the labelled LNPs, then purified as described for unlabelled particles (section 2.8). Subsequently, px333DE plasmid was chemically modified using click chemistry for covalent labelling using Label IT Ò Nucleic Acid Labeling Kit -Rhodamine-TM (Mirus #MIR4125). The labelled DNA was checked for fluorescence using a fluorescence spectrophotometer (Excitation/ Emission 560/580 nm). Finally, covalently labelled DNA was used to prepare LNPs as described above.

Culture supernatants were assayed for RT activity by surveying for was added to each well and then read by scintillation counting using TopCount Scintillation counter (Perkin Elmer). Radioactive counts per minute (CPM), proportionate to the amount of virus contained in sample culture supernatant, were normalized to trypan blue-based cell counts taken at time of harvest.

Primary human monocytes recovered after elutriation were plated on coverslips in 12 well plates. After 7 days of differentiation in MCSF treated media A, cells were maintained in media supplemented with human serum. Rhodamine DHPE tagged LNPs were used for nanoparticle cell trafficking, whereas px333-rhodamine-TM loaded LNPs were used to track the localization of CRISPR cargo inside the nucleus. MDMs were inoculated for 12 hr with Rhodamine TM labelled px333 containing CRISPR LNPs at 2 mg/10 6 cells dose and post inoculation were washed in PBS buffer and then fixed in 4% PFA. For immunostaining, cells were then permeabilized with 0.25% Tween 20 in 1x PBS for 15 minutes on rocker, followed by blocking (blocking solution: 5% Normal Goat Serum (NGS) (abcam #ab7481), 1% Bovine Serum Albumin (BSA) (Fisher Scientific #BP1600-100), 0.25% Tween 20 (Fisher Scientific #BP337-500) in PBS for 1 hour. Primary antibodies were diluted in antibody dilution buffer (10% NGS, 1% BSA, 0.25% Tween 20 in 1x PBS) at manufacturer recommended dilutions and added to respective wells. The following primary antibodies were used to label different endosomal compartments: Rabbit anti-human Rab5, (Abcam #ab18211), early endosomal marker, rabbit anti-human Rab7 (Santacruz biotechnology #sc10767), a late endosome marker, and rabbit anti-human Lamp-1 (lysosomal-associated membrane protein-1) (Abcam #ab24170) for lysosomes. Cells were incubated overnight at 4 o C in a rocker. Cells were washed for at least 3 times with 1x PBS for 10 minutes each in the rocker. Next, secondary antibody (Goat anti-rabbit Alexa Fluor ࣨ 488 (Invitrogen #A11008)) was prepared in antibody dilution buffer (1:500), with DAPI (Thermofisher, #D1306) as nuclear stain and Phalloidin (Abcam #ab176760) as cell boundary stain. Cells were incubated for 1.5 h in secondary antibody preparations for Rab 5, Rab7, and Lamp-1. Cells were adequately washed with 1x PBS. Coverslips were taken out and mounted on microscope slides using prolong gold antifade reagent. Compartmental colocalization of LNPs was assessed using a 63£ oil objective on an LSM 710 confocal microscope [30] .

mRNA loaded LNPs were formulated using non-turbulent microfluidic mixing using NanoAssmblr Ignite (Precision Nanosystem #NIN0001) using NxGen Cartridge (Precision Nanosystem #NIN0061) following manufacturer provided protocol. In short, 25mM lipid mix containing ionizable lipid (50 mol%), DSPC (10 mol%), Cholesterol (37.5 mol%), DMG PEG 2.5K (2.5 mol%) dissolved in ethanol was mixed with mRNA dissolved in a 100 mM acidic formulation buffer (pH 4). The lipid mix is available as Genvoy ILM (Precision Nanosystem #NWW0041) and the formulation buffer is also available commercially(#NWW0043). The cationic lipid amine to mRNA phosphate ratio (N/P) ratio was close to 5:1 (5:1:7:1) for all the formulations. FLuc LNP was formulated with 300 mg of CleanCap Ò FLuc mRNA (5moU) -(TriLink Biotechnologies #L-7202-100) (#L-7206-100) and GFP LNP was formulated with 200 mg of DasherGFP Ò mRNA (Aldevron #3870-0200). The RNAs were dissolved in aqueous formulation buffer and subsequently formulated using microfluidic mixing. For the CRISPR mRNA loaded LNP, termed as TatDE rLNP, we mixed 100 mg of CleanCap Ò Cas9 mRNA (5moU) -(TriLink Biotechnologies #L-7206) with 75mg each of TatD and TatE sgRNA. The sgRNAs were custom ordered to produce end modified chemically synthesized stable RNA. Then we dissolved them in 1500 mL of formulation buffer to produce the aqueous phase.

The aqueous to lipid phase flow ratio was selected at 3:1 with a flow rate of 12mL/min. After formulation, rLNPs were collected and immediately diluted in 40x volume of PBS and subsequently concentrated using an Amicon Ò Ultra-15 (Millipore Ò # C7715, MWCO 10kDa) centrifugation filter to remove any residual ethanol from the formulation. The concentrated purified formulation was approximately 2 mL for all the formulations. Next, 10 mL of the LNP was diluted in 990 mL of PBS and particle properties (size, polydispersity etc) were measured using Malvern Zetasizer. Finally, the purified LNPs were quantified using the Quant-IT TM RiboGreen TM RNA Assay Kit (Thermofisher # R11490) following manufacturer recommendation. For accurate quantitation, respective mRNA (Fluc or GFP) dissolved in formulation buffer was used to prepare standards. LNPs were quantitated neat as well as after Triton-X degradation. Encapsulation efficiency was calculated from the ratio of encapsulated RNA to total RNA.

Jurkat JLat 8.4 Jurkat T cell line and U1 cell lines were plated at a density of 5x10 5 cells/well in 12 well plate in RPMI media supplemented with 10% FBS and 1% Pen/Strep for 18 hours before mRNA LNP treatment and maintained at 37°C in a 5% CO2 incubator. They were then treated at 1 mg equivalent LNP dose of FLuc LNP or GFP LNP per well. All the RNA loaded LNP treatments were done in the presence of Apolipoprotein E4 to simulate the presence of such protein in blood that aids in nanoparticle uptake. Forty-eight hours post treatment, the Fluc LNP treated cells were analysed using Luciferase Assay System (Promega # E1500). In short, cells were spun down at 650xg, washed in PBS and lysed for release of intracellular protein. D-Luciferin substrate was added to the lysates and luminescence was processed and analysed according to manufacturer recommendation using a SpectraMax M3 Plate reader (Molecular Devices).

GFP LNP treated cells were spun down, washed with PBS and stained with Live/Dead fixable far read stain (Invitrogen # L10120) and fixed with 2% PFA in FACS buffer. Subsequently flow cytometry was performed on them and the data is reported as % positive live gated singlet cells. The detailed gating strategies are listed in the supplementary materials ( Figure S2-S3 ). Beads were used as a compensation control and compensation matrices for each cell line was created and applied to data analysis using FlowJo v10.8.

TatDE rLNPs were treated at 1 mg Cas9 mRNA equivalent per 5x10 5 cells/well similar to the GFP and TatDE LNPs. All the conditions in each cell line were ran in biological triplicates. Seventy-two hours post treatment, 2.5x10 5 of the cells were collected and the rest of the cells were stimulated with TNF-a or PMA for viral outgrowth assay. Half of the collected cells were spun down and seeded in new plate and maintained in the incubator. Subsequently they were subjected to MTT viability assay to assess the TatDE rLNP mediated toxicity. Untreated cells served as a control for the MTT assay. Furthermore, the remaining cells were harvested, and genomic DNA was extracted using NucleoSpin Tissue XS kit as described above. To account for the low abundance of HIV genome in the genomic DNA content, nested PCR was performed with HIV specific primers as previously described in both treated and untreated group. Agarose gel electrophoresis was performed to show the PCR product. To further confirm the findings, U1 and JLat cells were harvested 72 hours post stimulation, and RNA was extracted using TRIzol method. Digital droplet PCR (ddPCR) for HIV transcript was performed on the extracted DNA with seminested primers as reported [31] .

The experiments performed in this study did not involve any human or animal subjects. Hence, it was exempt from the requirement of UNMC IRB (institutional review board) and UNMC IACUC (institutional animal care and use committee) approvals.

The funding authorities of this study did not participate in any part of the experimental design, execution, or data analysis. The study was not changed, affected, or influenced in any scale by the funding bodies.

Cell Lines were obtained from NIH HIV reagent program or commercial vendors. Each cell line obtained was verified against the lot specific certificate of analysis to ensure that they were free from mycoplasma, bacteria and fungi. Routine mycoplasma testing is performed every six months in our laboratories.

All other reagents used in this study are listed with their vendors and catalogue numbers. We checked to make sure that these numbers are also RRID for the said reagents. To keep uniformity, we used the same format throughout the document to list the company name and catalogue number. We also verified that these strings identify a single reagent.

We only used commercial primary and secondary antibodies from reputed vendors and the RRID is listed with the specific antibodies.

Experiments were run in biological replicates (n = 3 or 4) in at least two separate trials, as indicated in figure legends. RT activity assays and RT-qPCR were further performed with technical triplicates. Data are presented as mean § standard error of the mean (SEM) or standard deviation (SD). Pearson correlation between mean RT activity reductions and percent gRNA conservation was performed using two-tailed p-value determination. Pearson correlation was chosen to measure whether the two variables (conservation and efficacy) were positively correlated. Statistical differences between unstimulated compared to stimulated RT activity values and stimulated control compared to stimulated treatment RT activity values were calculated with two-way ANOVA followed by Sidak's post-hoc test for multiple comparisons. Two-way ANOVA measured whether stimulation and/or the treatment has effect on the RT activity. Sidak's post hoc test was performed to analyse the variance between groups and adjust P value and confidence interval. As experimental and control groups were independent of each other and comparisons were pre-selected for stimulated or unstimulated, and amongst stimulated groups alone, Sidak's test was chosen over the Bonferroni method as control of entire family-wise error with the latter method was unnecessary. Normal distribution of data was determined by Shapiro-Wilk or Kolmogorov-Smirnov tests. Homoscedasticity of data was determined by the Cochran C, Hartley, and Bartlett test. One-way ANOVA with Dunnett's post-hoc test for multiple comparisons was employed to assess differences between the stimulation indices of control-and tat-targeted treatments. This approach was used to evaluate RT activity and cell vitality in infected control (untreated) and treated groups. Where ANOVA was used to test significance for the multivariate experiments and whether at least two groups are different and control of total family-wise error was unnecessary, P values were adjusted by Dunnett's correction for comparing groups against a single control. For these tests, each group was assumed to have equal variance as determined by the Cochran C, Hartley, and Bartlett or Levene tests with values normally distributed as confirmed by Q-Q plots and Shapiro-Wilk or Kolmogorov-Smirnov tests. The experiments were performed so that each of the recorded responses were independent of one another. Additionally, we also performed nonparametric Mann Whitney U test for pairwise comparison by adjusting P value to number of comparisons and significance was still shown. Cohen's d or Effect size eta squared is reported where applicable. Progeny virion production as measured by RT activity in the electroporation experiments were compared using unpaired t-tests. All the control and treatment groups were assessed independently of each other. Statistical calculations set type I error cut-off a = 0.05 and were performed using GraphPad Prism v7.0a or 9 for Mac OS X (GraphPad Software, San Diego, California USA, www.graphpad.com).

Targeted HIV-1 gene therapy requires specific recognition of diverse proviral sequences. HIV-1 sequence diversity occurs both at the population level and during disease progression, thus lessening the effectiveness of host immune responses and promoting the emergence of antiretroviral drug resistance [32] . While the 5' LTR is a promoter for viral transcription and is susceptible to CRISPR-based HIV-1 inactivation, we reasoned that targeting the driver of LTR-directed full-length transcription, HIV-1 tat, would improve CRISPR therapeutic efficacy. To test this notion, a multiple sequence alignment heat map, constructed from 4004 HIV-1 strains, was utilized to observe the degree of nucleotide heterogeneity across the viral genome ( Figure 1 ). Conserved regions of the HIV-1 genome include portions of flanking LTRs, gp120-encoding segments of env, and multiple exon overlaps (Figure 1a) . Low entropy in HIV-1 genes encoding p6/protease, tat/rev, vpu/gp120, and nef/3' LTR highlight sequence preservation in loci that govern multiple phases of the viral lifecycle.

As comparators, two CRISPR-Cas9 reference systems were adopted. First, LTR-1 and GagD are gRNAs that target the U3 portion of the LTR and MA region of gag, respectively (Figures 1b, S4) . LTR-1/ GagD was selected as a reference control because of its utility in sterilizing one-third of HIV-1 infected humanized mice, thereby establishing proof-of-concept for a CRISPR-based HIV cure in an animal model [11] . Second, DLTR-1 and DLTR-2 are top-rated gRNAs from a quadruplex panel designed to cleave most known HIV-1 strains with minimal off-target human genome recognition [25] . Delivery of DLTR-1 and DLTR-2 gRNAs to cells by lentivirus remains a challenge, however, because their targeting of HIV-1 LTR ( Figure S5 ) simultaneously also thwarts the synthesis of all CRISPR-transducing lentiviruses. Control gRNAs display complete 20 base pair (bp) recognition of 11-30% of curated HIV-1 strains (Table S3) . A minimum off-target threshold of 2.266 was set for tat gRNA design in accordance with a CRISPR-specificity algorithm score for the DLTR-2 control.

A library of eight HIV-1 tat-targeting gRNAs designed for deployment with Streptococcus pyogenes Cas9 (spCas9) endonuclease was created. These "mosaic" gRNAs were constructed against the tat consensus sequence synthesized from 4004 HIV-1 strains ( Figure S4) . Five of the eight produced gRNAs target the sense strand (Figure 1c) . Because portions of tat overlap with rev and the gp41 portion of env, up to three exons could be simultaneously disrupted with these gRNAs ( Figure S5) . Notably, tat-directed gRNAs retain full 20 bp complementarity for 6-67% of all known HIV-1 strains (Table S3) . Duplexed mosaic gRNA treatments were found to target at least 56-62% of strains, assuming no nucleotide mismatch tolerance (Table S4) . Taken together, these data demonstrate that gRNAs can be produced with broad HIV-1 excision potential while, at the same time, limiting off-target effects.

To test the hypothesis that conserved gRNAs would most effectively limit viral replication, a plasmid co-transfection screen was implemented. Human embryonic kidney (HEK293FT) cells were transfected with plasmids encoding HIV-1 in the absence (untreated control) or presence of spCas9-gRNA constructs. Seven HIV-1 molecular clones, one laboratory (NL4-3) plus six clade B founder strains, were included to reflect predominant North American and European viral subtype sequence heterogeneity. Four independent co-transfection screens were performed to ensure experimental accuracy. Supernatants were measured after 72 hours for HIV reverse transcriptase (RT) activity. The single mosaic gRNA CRISPR TatE and TatD constructs demonstrated the most significant suppression of HIV-1 replication (Figure 2a) . TatE gRNA reduced RT activity by an average of 76% ( § 6.5%, standard error of the mean (SEM)), with more than 75% reduction in 6 of 7 tested HIV-1 strains. Notably, TatE outperformed all single gRNA controls, including DLTR-2. Pearson correlation analysis evaluated whether CRISPR gRNA activity was associated with target sequence conservation (Figure 2b) . While positive trends were observed, significance was achieved when TatE was excluded from the analysis. The data support the idea that although gRNA targeting conserved genomic regions of HIV-1 proviral DNA are generally more efficacious, the TatE locus is critical to viral replication. Although, TatE2 and TatF target the same region as TatE, they showed comparatively lower efficacy. This can be attributable to mismatches between target DNA and gRNA protospacer adjacent motif (PAM, seen with TatF) or in seed sequence proximal to PAM (seen with TatE2) ( Figure S6) . Mismatches within or close to PAM have been previously characterized as detrimental to Cas9 endonuclease activity [33] .

We next sought to determine whether proviral DNA excision induced by two CRISPR-Cas9 gRNAs would further suppress HIV-1 replication. The noted efficacy of TatE, which targets a region of tat that overlaps with both rev and env, prompted us to posit that simultaneous disruption of multiple viral exons would most drastically blunt HIV-1's lifecycle. Plasmids were subcloned to express various combinations of the top three performing tat-directed gRNAs. Table S4 summarizes the percent conservation, excision fragment length, and the number of deactivated exons elicited by each pairing.

Increased antiretroviral activity was observed when TatD and TatE gRNAs were co-expressed (TatDE). Viral replication in TatDE-treated cells was lowered on average by 82% ( § 4.6% SEM), with a range of 65-98% across all tested viral strains (Figure 2a) . TatDE demonstrated greater levels of viral suppression compared to LTR-1/GagD ("LG") and DLTR-1/DLTR-2 ("DLTR1-2") duplexed reference CRISPR controls by measured RT activity. The »2.5 kilobase pair (kb) dropout of intervening proviral DNA was observed against all seven HIV-1 molecular clones (Figures 2c, S7) . To confirm the accuracy of TatDE CRISPR, excision bands were Sanger sequenced and analysed by the Inference of CRISPR Edits (ICE) algorithm ( Figure S8) .

Sequencing data qualitatively depict nucleotide degeneracy in both target loci, with insertion/deletion (indel) mutations at a rate of 25-51% in excision bands. Pairing our two most conserved efficacious gRNAs, TatDH, and of our largest excising duplex, TatEH, proved inferior to 5-exon inactivating TatDE in suppressing RT activity. Taken together, these data support the hypothesis that disrupting a maximal number of viral exons by CRISPR-Cas9 leads to the greatest suppression of HIV-1 replication.

We next evaluated the specificity of our tat-directed CRISPR-Cas9 by comparing the degree of proviral genome editing to that observed in off-target loci. ICE analysis performed on Sanger sequences from these reactions revealed that dual gRNA systems improve indel mutation rates to >40% as compared to <10% seen with single gRNA controls (Figure 2d) . The high efficiency may be attributed to DNA excision, which overrides the potential for host DNA repair. TatDE therapy resulted in an average of 45% ( § 3.0% SEM) gene modification rate with 40-50% editing in 6 of 7 tested strains, supporting TatDE's ability to deactivate a broad variety of HIV-1 proviral species. Next generation sequencing will be employed in future studies to validate the accuracy of these findings.

To ensure TatDE cleavage is restricted to HIV-1 proviral DNA, we investigated the potential for aberrant indels at off-target loci in the human genome. These covered five possible loci recognized by each TatD, TatE, LTR-1, and GagD predicted in sillico during gRNA design. No editing was observed at top putative off-target positions for TatD and TatE (Table S5) . This was sustained regardless of whether cells were treated with single or dual gRNAs. Similarly, off-target analysis by ICE for LTR-1 and GagD failed to demonstrate indel mutations in host genes ( Table S6) . Seven of 10 possible off-target regions for TatDE fell in intronic regions, whereas only 2 of 10 for LTR-1/GagD were noncoding. These results suggest that TatDE CRISPR is not more likely to affect host gene expression than the LTR-1/GagD system. Further sequence analysis of the edited genomes will elucidate whether mutations were inadvertently introduced at off-target sites.

Having demonstrated the ability of TatDE CRISPR to suppress plasmid-encoded viral transcription, we next investigated whether latent HIV-1 could be excised in leukocytes. For these experiments, TatD, TatE, or non-targeting control gRNAs were cloned onto a lentiviral-CRISPR expression plasmid. To first assess CRISPR-Cas9 function, ACH2 T cells and U1 promonocytes bearing 1-2 copies of latent HIV-1 proviral DNA copies per cell were transfected with cloned lentiviral-CRISPR constructs. Subsequent stimulation of ACH2 T cells with TNFa induced viral rebound in control-and LTR-1/GagD-treated cultures but not in those receiving single-or dual TatD plus TatE treatments ( Figure S9 ). U1 promonocytes were equally responsive to LTR-1/ GagD and TatD/TatE treatments ( Figure S10) . HIV-1 proviral DNA excision was present in both cell types and was confirmed by sequencing. These data, in aggregate, validate the TatDE CRISPR-Cas9 system for lentiviral delivery.

We next ascertained whether lentiviral transduction of TatDE CRISPR-Cas9 could halt HIV-1 induction from latently infected ACH2 T cells. CRISPR Transgene expression was measured by reverse-transcriptase quantitative PCR (RT-qPCR) and determined to be above detection limits (Figure 3a) . Dual TatD/TatE lentiviral treatments blocked TNFa-induced stimulation. Co-transduction of TatD/TatE at MOIs of 10 and 1 significantly reduced the release of HIV-1 into culture supernatants by 80% and 94%, respectively (Figures 3b-c) . We also evaluated genomic DNA by nested PCR for TatDE HIV-1 excision (Figures 3e, S11) . A predictable »525 bp excision band was present for cells co-transduced at an MOI of 1, supporting the reduction in RT activity at this MOI. Shorter amplicons were observed with MOI 0.1 treatment; however, the shortened amplicons disappeared in stimulated conditions. Paired with the RT activity results (Figure 3d) , it is likely that transduction at a low MOI was subtherapeutic, as few cells bear the CRISPR-induced indel. Conversely, we cannot exclude the possibility that the efficacy observed at MOI 10 in the absence of excision bands results from cytopathicity encountered with high levels of lentiviral treatment. These results cumulatively support the utility of TatDE CRISPR-Cas9 in eliminating latent proviral HIV-1.

As a potential mechanism to explain TatE's high antiretroviral activity despite low sequence conservation, we investigated how disruption of different numbers of viral exons affects HIV-1 replicative fitness. Three HIV-1 NL4-3-Dnef-eGFP non-frameshift point mutants were created to parallel CRISPR mutation profiles (Figure 4a) , in which two-(DTatD), three-(DTatE), or five (DTatDE) exons were altered (Figure 4b ). Virions were produced by transfecting mutated NL4-3 plasmids in HEK293FT cells. The resulting virions were imaged by transmission electron microscopy (TEM) (Figure 4c) and titered by RT-qPCR. The size of the HIV-1 tat mutants ranged from 110.6-150.5 nm in diameter, closely approximating that of wild-type HIV-1 measured at 125.4 nm. Likewise, the HIV-1 tat mutants' nearly spherical appearance and inner conical capsid matched the morphology of the wild-type control virus. Next, CD4+ T lymphocytes were infected at an MOI of 0.1. At ten days following HIV-1 infection, significant differences in RT activity were detected in culture supernatants (Figure 4d) . HIV-1 NL4-3-DtatD displayed nearly equivalent levels of viral replication as wild-type virus, while mutants bearing 3 disrupted exons were significantly lower in RT activity compared to unmutated control. As CD4+ T cell proliferation rates differed after ten days due to HIV-1 induced cytopathicity, the percent of cultured HIV-1 infected cells was measured by flow cytometry during a 28day time course (Figure 4e) . Whereas HIV-1 NL4-3-DtatD proliferation approximated that of control HIV-1, HIV-1 NL4-3-DtatE and HIV-1 NL4-3-DtatDE remained at or around baseline for four weeks. These data suggest that the locus targeted by TatE gRNA is critical for maximal CRISPR activity. In aggregate, we conclude that the high efficacy of TatDE CRISPR-Cas9 therapy against numerous HIV-1 strains results from disrupting five viral exons simultaneously.

Next, we designed and synthesized TatDE LNPs capable of loading the TatDE gRNA plasmid using thin-film hydration methods (Figure 5a) . The TatDE LNPs were seen morphologically as roundshaped particles by TEM and AFM (Figure 5b-d) . They had an average size of 180 nm and polydispersity index (PDI) of 0.2 using dynamic hoc test) (g) PCR was performed in extracted DNA from cell lysates using HIV-1 specific primers, followed by agarose gel electrophoresis. The excised band at 500 bp is highlighted in yellow confirms "putative" protection. (h) The sized band was confirmed as the appropriate CRISPR-Cas9 excised subgenomic viral DNA by Sanger sequencing and the multiple alignment program for nucleotide sequences. Data points, in e, depict mean § SEM from triplicate biological replicates. light scattering (DLS) and were stable during storage (Figure 5e and S12). CRISPR LNP treatment of primary human MDM protected against HIV-1 ADA infection (MOI 0.02) in a dose-dependent manner for seven days. This was affirmed by monitoring culture cell supernatants for progeny virion production by RT activity (Figure 5f ) and by polymerase chain reaction (PCR) of cell lysates with HIV-1 specific primers spanning the subgenomic viral DNA fragments of the gRNAs. Agarose gel electrophoresis demonstrated excision of a »2.4 kb fragment of the viral genome (Figure 5g) . Extraction of this excised amplicon followed by Sanger sequencing was deployed to validate the CRISPR editing. The recovered Sanger traces were subjected to sequence alignment to confirm the correct viral excision fragment (Figure 5h ).

We next sought out the mechanism of TatDE LNPs trafficking into the cells using confocal laser scanning microscopy. For this, we used a dual approach to label nanoparticles. In the first set of experiments, we labelled the TatDE LNPs (Figure 6a-c) with rhodamine DHPE phospholipid, and in another set, we labelled the nucleic acid cargo (Figure 6d) . For the first set of experiments, the goal was to assess the route of LNP migration towards the nucleus. Cells were treated with respective LNPs in membrane-permeabilizing solution, fixed and stained 12h post-treatment. TatDE LNPs colocalized with Rab5 ( Figure 6a ) and Rab7 (Figure 6b ) but not Lamp1 (Figure 6c ). Rab 5 and Rab7 are early and late endosomal markers, respectively. LNP colocalization with the endosomal compartments suggest that the TatDE LNPs travel through the endosomal compartments. Also, there was minimal colocalization with the Lamp1 lysosomal compartments, confirming that the CRISPR LNPs are not degraded via lysosomes. To further survey payload delivery to the nucleus, parallel experiments were performed with fluorescently tagged nucleic acid. We observed that rhodamine-tagged nucleic acid reached the DAPItagged nucleus inside macrophages. Z-stack analysis was performed to show that the TatDE LNPs were indeed inside the nucleus, not on top in a different plane. Pink colocalization between the blue nucleus and red nucleic acid in a single plane confirmed that the CRISPR payload was delivered to the nucleus by the LNPs.

Combination TatD and TatE gRNA-cas9 ribonucleoproteins (TatDE RNP) were next assessed for antiretroviral efficacy. TatDE RNP and two HIV molecular clones (NL4-3 & pCH040.c/2625) were co-transfected HEK293FT cells using TransIT-X2 transfection reagent (Figure 7a ). 48 hours post-transfection supernatants were tested for RT activity as an indicator of progeny virion production and extracted cellular DNA was assessed for proviral DNA excision. TatDE RNP coelectroporation decreased the virion production in the supernatant by 96% ( §1% SD) compared to positive control (Figure 7b and S10). Nested PCR showed full-length amplicon in untreated cells, whereas no full-length amplicon was seen in the RNP-treated cells (Figure 7c) . Electroporation did not adversely impact cell vitality ( Figure 7d and S11). Next, we used electroporation to nucleofect the TatD/E RNPs into ACH2 cells (Figure 7e) . Treated cells were resistant to viral outgrowth after stimulation with TNFa ( Figure 7f) . PCR showed excision of integrated HIV-1 DNA present in the infected T cell line (Figure 7g ).

Lipid nanoparticles were optimized for mRNA delivery. Green Fluorescent Protein (GFP) and firefly Luciferase (Fluc) mRNA LNP was formulated using microfluidic mixing (Figure 8a) . Subsequently, they were purified and concentrated. The diameter of luciferase LNP was 73.08 nm and GFP LNP was 76.62nm measured by dynamic light scattering with PDI values of 0.083 and 0.098 respectively (Figure 8b and c). The nanoparticles were then lysed using Triton-X and RNA quantity was measured in the solution using Quant-IT RiboGreen RNA assay kit. Neat nanoparticles served for quantitation of encapsulation efficiency. FLuc and GFP LNPs had an encapsulation efficiency of 94.4% and 83.8% respectively (Figure 8d ). Then we treated U1 and JLat cells with luciferase and GFP LNP at a dose of 2 ug per million cells. These cell lines are latently infected cell models where each cell has at least one copy of HIV genome. We saw robust expression of luciferase in the cells (Figure 8e ). In addition, more than 85% of the gated singlet cells were found to be GFP-positive (Figure 8f-g) . 

Over 79,000 unique human-or simian immunodeficiency virus (HIV/SIV) sequences spanning eleven global subtypes have been identified [14] . The documented 6% mutation rate even among elite controllers highlights the presence of a multitude of intrapatient HIV-1 quasispecies [34] . These considerations led us to design mosaic CRISPR-Cas9 gRNAs against conserved and multi-exon regions of a consensus HIV-1 sequence to facilitate viral elimination. Herein, we introduce exonic disruption as a guiding principle for CRISPR therapeutics as a required link to control residual viral infection. Our data establish an association between tat/rev target DNA conservation and CRISPR-Cas9 efficacy across multiple HIV-1 transmitted founder strains (Figure 2b) . The impact of this finding is twofold. First, it validates our mosaic gRNA approach as a tenable means of using CRISPR to inactivate genetically heterogeneous viral targets. Second, the correlation reveals that conserved tat/rev-directed gRNAs outperform LTR-based reference controls. Given that these were selected for their potential to recognize divergent HIV-1 quasispecies in patients [25] and facilitate elimination from infected hosts [11] , the comparative superiority of TatDE represents an advance in CRISPR therapy as a strategy for HIV-1 elimination.

A second advantage of the TatDE mosaic gRNA approach is that it identifies exonic disruption as a potential mechanism for antiretroviral efficacy. Assessed variables include CRISPR on-target cleavage efficiency, target DNA conservation, excision length, and numbers of disrupted exons during CRISPR-Cas9 treatment. Non-frameshift sitedirected mutagenesis of 3 or more exons at the TatE locus, but not the two exons at TatD, significantly suppressed viral replication. In our gRNA co-transfection screen, the most broadly active CRISPR treatments included tat/rev/gp41-directed TatE. One implication that follows is that usage of TatE-directed CRISPR may help overcome low potency [35] and viral escape [36, 37] otherwise associated with single gRNA treatment of HIV-1. These results also suggest that cleavage at other multiexon locations along the HIV-1 genome, such as gag/pol and nef/LTR, or tat/rev/gp41 by other Cas nucleases [20, 38] may afford broader antiviral coverage. Thus, exonic disruption paired with target DNA conservation is integral in generating effective HIV-1 CRISPR therapeutics.

Delivery of HIV-1 specific CRISPR payloads remains one final parameter yet to be optimized prior to human administration. Current ex vivo approaches ablate the HIV-1 co-receptor CCR5 with CRISPR-transducing lentiviruses [39, 40] . However, the 5% editing Figure 9 . mRNA LNP carriage (rLNP) of HIV-1 TatDE and Cas9 Attenuate Viral Replication. CleanCap Cas9 mRNA, TatDE gRNA was combined and formulated using optimized lipid mix aided by microfluidic mixing to formulate the TatDE rLNP. They were characterized and tested for antiviral efficacy. (a) Schematic representation of antiviral efficacy and toxicity testing. (b) TatDE rLNP showed a narrower size distribution with a PDI of 0.045 and an average diameter of 76.11 nm. (d) They also had a high encapsulation efficiency of 92.9%. TatDE rLNP was treated to cells at 2ug Cas9 mRNA equivalent per million of U1 or JLat cells. (c) Seventy-two hours post-treatment, cells were tested for rLNP mediated toxicity using MTT vitality assay. TatDE rLNP was non-toxic for the U1 cells (»100% vitality in the treated group) and JLat cells (85% vitality in the treated group) compared to untreated controls. U1 and JLat cells contain one or more integrated copies of the viral genome in each cell. Cell genomic DNA was isolated, and nested PCR was performed with HIV specific primers. Agarose gel electrophoresis of PCR product showed in both U1 cells (e) and JLat cells (f) full-length HIV was present in untreated cells but not treated cells. Treated cells rather had an expected 525 bp excised fragment. PCR was completed from three independent samples. Subsequently, treated cells were stimulated with 20 ng/mL TNFa (JLat 8.4 cells) or 50nM phorbol myristate acetate (PMA) (U1 cells). Seventy-two hours post-stimulation, cells were harvested, and RNA was extracted. (g-h) Highly sensitive digital droplet PCR showed induction of HIV-1 RNA production in untreated cells, whereas in the case of treated groups, even after stimulation, HIV-1 RNA production was near the baseline. (i) Further analysis shows that stimulation led to up to 100 (JLat) to 300 (U1) fold increases in RNA production. efficiency in transduced human hematopoietic stem cells failed to render viral suppression upon antiretroviral drug withdrawal [39] . Our strategy is to deploy TatDE CRISPR using a highly efficient vector to viral target cells. With lentiviral transduction at an MOI of 1, the TatDE CRISPR transgene was detected at an average of 33.2 spCas9 RNA copies per ACH2 T cell, with a 94% reduction in RT activity and clear viral excision. Using a single broad-spectrum gRNA targeting LTR transduced at this same level but allowed viral rebound in »20% latently infected JLat cells [41] . Other studies had achieved >90% reduction in HIV-1 reactivation and prevention of viral escape when CRISPR-Cas9 was lentivirally transduced in T cells at MOIs of 10-15 [21, 23, 42] . These differences highlight the likelihood that TatDE CRISPR-Cas9 maintains greater potency against HIV-1 compared to other CRISPR-Cas9 therapies. A possible critique of our delivery choice is that patients will likely already be receiving combination antiretroviral drugs that inhibit lentivirus' ability to transduce CRISPR. We employed LNP delivery as a strategy to achieve viral excision even in the face of concurrent antiretroviral therapy. To optimize the physicochemical characteristics and loading, specific molar lipid ratios were selected for CRISPR TatDE plasmid encased LNPs that efficiently enabled the cargo delivery to the nucleus. Lipids included DOTAP as a cationic lipid which complexed with the negatively charged DNA and easily incorporated with other lipids to form a stable colloidal suspension. Cholesterol provided stability to the LNPs. Macrophages are ubiquitous as they are found in all lymphoid and non-lymphoid tissues and contribute both to cell homeostasis while serving as reservoirs for replication competent HIV-1 [43] . Additionally, macrophages can readily spread HIV-1 to autologous CD4+ T cells [44] . Therefore, human MDMs were next chosen to evaluate the excision efficiency of TatDE LNPs to prevent viral infection. We showed that the level of expression of CRISPR LNPs was high and excision of integrated HIV-1 proviral DNA occurred in virus-challenged macrophages. Localization of CRISPR LNPs in the cell nucleus was confirmed both by viral excision and confocal microscopy studies. We purport that the LNPs employed to represent an excellent platform for both safety and efficacy. To affirm these results and to bring this technology to the next level for translation from cells to animals then to humans, we evaluated the abilities to reactive latent HIV-1 infection in U1 and JLat 8.4 cell lines, used to model latent viral infection after mRNA LNP treatment. The cells were PMA treated and when used in coordination with TatDE rLNP, it prevented HIV-1 RNA reactivation with up to complete viral elimination as tested by ddPCR.

For delivery of CRISPR therapeutics to the target cells, the form of payload and vector were both were tested. CRISPR payload can be DNA (TatDE LNP and transfection), mRNA+sgRNA (TatDE) or CRISPR RNP (electroporation). RNP and mRNA showed superior efficacy compared to DNA. mRNA loaded TatDE LNPs are similar to other mRNA loaded LNP therapeutics used in the industry including the Moderna and Pfizer SARS-CoV-2 vaccine. Furthermore, early-stage success of Intellia's Clinical Trial (NCT04601051) for in vivo CRISPR Cas9 editing for patients with hereditary transthyretin amyloidosis with polyneuropathy (ATTRv-PN) and shows the promise of using in vivo gene therapy [45] .

The selectivity of the CRISPR-Cas system for HIV-1 proviral DNA is a key factor in assessing our treatment's therapeutic window. No offtarget edits were observed with TatDE treatment in the ten loci surveyed by a Sanger sequencing-based algorithm. By contrast, ICE data demonstrate a mean 45% editing efficiency of proviral DNA template with TatDE CRISPR. This led to detectable excision of »2.5 kb of viral DNA and to the nearly complete abolition of viral release. These findings are aligned with reports that »10% and 39-63% editing efficiencies from multiplex endonuclease treatments resulted in more than 90% knockdown in herpes simplex virus and HIV-1, respectively [23, 46] . Dysregulation of positive feedback loops, as in the case of tat/ rev-directed CRISPR-Cas9, and the formation of dominant-negative proteins encoded after gene editing may explain the disproportionately high efficacy in each of these cases. The HIV-1 proviral genomes isolated from elite controllers display large deletions of env sequences flanked by tat/rev amongst other regulatory genes [47] . It is postulated that defective virus in elite controllers primes the host immune system to clear cells bearing replication-competent HIV-1 proviral DNA [47] . Therefore, we posit that TatDE CRISPR therapy may similarly compromise latent HIV-1 sequences, thereby inducing immune-mediated clearance in vivo.

Future investigations will facilitate the translation of these strategies for human use. TatDE's multistrain potential needs to be further validated against additional HIV-1 strains and clades for antiviral activity. The interrogation of the preliminary lack of off-target editing within the human genome is an inherent limitation. Future works must employ CIRCLE-seq [48, 49] , Guide-seq [41, 50] and wholegenome sequencing [11] to each of the TatDE CRISPR-treated samples. Further characterization of the TatDE's mechanism of action needs be determined in regard to its ability to downregulate fulllength viral transcripts due to tat, diminish rev-associated nuclear export, and impede gp41-mediated viral entry. These additional studies will clarify the value in expanding our conserved, multi-exon targeting CRISPR library to other Cas endonucleases. Finally, precision decorated LNPs need to be developed in order to speed entry and excision of all CD4+ T cells and macrophages cell reservoirs. Every infected cell must be excised in order to cure HIV. We will explore leukocyte targeted LNPs as well as modifying payload compositions to the lipid nanoparticles to further optimized TatDE CRISPR therapy.

A major challenge to gene therapy-based treatment of HIV lies in finding suitable therapeutic mode (DNA, mRNA, or RNP) and vehicle (such as LNPs) capable of inactivating divergent HIV strains at anatomic reservoir site. These studies provide proof-of-concept of our formulations' potential to fully excise the integrated HIV that might thereby lead to HIV sterilization. To conclude, the combination of "diverse" strain-inactivating TatD and "tri-exon"-directed TatE gRNAs prompted maximal suppression of HIV-1 replication by CRISPR-Cas9. Further refinement of these methodologies along with advance antibody-based targeting techniques [51] will advance the utility of gene therapy in eliminating latent viral infection from infected hosts across the globe. 19040PCT; Serial No. PCT/US2020/016126; International Publication No. WO 2020/160418 A1). H.E.G is a member of the scientific advisory board at Longevity Biotech and a co-founder of Exavir Therapeutics, Inc.

Virological failure, HIV-1 drug resistance, and early mortality in adults admitted to hospital in Malawi: an observational cohort study

Tandem bispecific neutralizing antibody eliminates HIV-1 infection in humanized mice

Antibody and TLR7 agonist delay viral rebound in SHIV-infected monkeys

Multispecific anti-HIV duoCAR-T cells display broad in vitro antiviral activity and potent in vivo elimination of HIV-infected cells in a humanized mouse model

Attacking Latent HIV with convertibleCAR-T Cells, a Highly Adaptable Killing Platform

Maternal Broadly Neutralizing Antibodies Can Select for Neutralization-Resistant, Infant-Transmitted/Founder HIV Variants

Rapid Elimination of Broadly Neutralizing Antibodies Correlates with Treatment Failure in the Acute Phase of Simian-Human Immunodeficiency Virus Infection

Advances in Developing CAR T-Cell Therapy for HIV Cure

Control of HIV despite the discontinuation of antiretroviral therapy

Evidence for HIV-1 cure after CCR5Delta32/Delta32 allogeneic haemopoietic stem-cell transplantation 30 months post analytical treatment interruption: a case report

Sequential LASER ART and CRISPR Treatments Eliminate HIV-1 in a Subset of Infected Humanized Mice

The accuracy of reverse transcriptase from HIV-1

Variation in HIV-1 set-point viral load: epidemiological analysis and an evolutionary hypothesis

HIV sequence databases

The challenge of HIV-1 subtype diversity

Understanding the mechanisms driving the spread of subtype C HIV-1

Use of the CRISPR/Cas9 system as an intracellular defense against HIV-1 infection in human cells

CRISPR-Cas9 Can Inhibit HIV-1 Replication but NHEJ Repair Facilitates Virus Escape

Genome scale screening identification of SaCas9/gRNAs for targeting HIV-1 provirus and suppression of HIV-1 infection

Extinction of all infectious HIV in cell culture by the CRISPR-Cas12a system with only a single crRNA

CRISPR/Cas9 system targeting regulatory genes of HIV-1 inhibits viral replication in infected T-cell cultures

The CRISPR/Cas9 system inactivates latent HIV-1 proviral DNA

Multiplexed tat-Targeting CRISPR-Cas9 Protects T Cells from Acute HIV-1 Infection with Inhibition of Viral Escape

Designing broad-spectrum anti-HIV-1 gRNAs to target patient-derived variants

Novel gRNA design pipeline to develop broad-spectrum CRISPR/Cas9 gRNAs for safe targeting of the HIV-1 quasispecies in patients

Comparison of shortened mosaic HIV-1 vaccine schedules: a randomised, doubleblind, placebo-controlled phase 1 trial (IPCAVD010/HPX1002) and a preclinical study in rhesus monkeys (NHP 17-22

CHOP-CHOP v3: expanding the CRISPR web toolbox beyond genome editing

In vivo engineering of oncogenic chromosomal rearrangements with the CRISPR/Cas9 system

Bioimaging predictors of rilpivirine biodistribution and antiretroviral activities

Rilpivirine-associated aggregation-induced emission enables cell-based nanoparticle tracking

HIV-1 cellular and tissue replication patterns in infected humanized mice

HIV-1 sequence variation: drift, shift, and attenuation

Novel gRNA design pipeline to develop broad-spectrum CRISPR/Cas9 gRNAs for safe targeting of the HIV-1 quasispecies in patients

Highly divergent patterns of genetic diversity and evolution in proviral quasispecies from HIV controllers

Anti-HIV-1 potency of the CRISPR/Cas9 system insufficient to fully inhibit viral replication

A combinational CRISPR/Cas9 gene-editing approach can halt HIV replication and prevent viral escape

The Impact of HIV-1 Genetic Diversity on CRISPR-Cas9 Antiviral Activity and Viral Escape

CRISPR-CasPhi from huge phages is a hypercompact genome editor

CRISPR-Edited Stem Cells in a Patient with HIV and Acute Lymphocytic Leukemia

CCR5 editing by Staphylococcus aureus Cas9 in human primary CD4(+) T cells and hematopoietic stem/progenitor cells promotes HIV-1 resistance and CD4(+) T cell enrichment in humanized mice

Safe CRISPR-Cas9 Inhibition of HIV-1 with High Specificity and Broad-Spectrum Activity by Targeting LTR NF-kappaB Binding Sites

Elimination of HIV-1 Genomes from Human T-lymphoid Cells by CRISPR/Cas9 Gene Editing

Macrophages and HIV-1: An Unhealthy Constellation

Efficient HIV-1 transmission from macrophages to T cells across transient virological synapses

CRISPR-Cas9 In Vivo Gene Editing for Transthyretin Amyloidosis

Gene editing and elimination of latent herpes simplex virus in vivo

Distinct viral reservoirs in individuals with spontaneous control of HIV-1

Analysis of CRISPR/Cas9 Guide RNA Efficiency and Specificity Against Genetically Diverse HIV-1 Isolates

CIRCLE-seq: a highly sensitive in vitro screen for genome-wide CRISPR-Cas9 nuclease off-targets

GUIDE-seq enables genome-wide profiling of off-target cleavage by CRISPR-Cas nucleases

Cell specific delivery of modified mRNA expressing therapeutic proteins to leukocytes

The work was supported by the University of Nebraska Foundation, which includes donations from the Carol Swarts, M.D. Emerging Neuroscience Research Laboratory, the Margaret R. Larson Professorship and individual donor support from the Frances and Louie Blumkin Foundation and from Harriet Singer. The research received support from National Institutes of Health grants T32 NS105594, 5R01MH121402, 1R01Al158160, R01 DA054535, PO1 DA028555, R01 NS126089, R01 NS36126, PO1 MH64570, P30 MH062261, and 2R01 NS034239. The authors thank Drs. Beat Bornhauser (U. of Zurich) and Won-Bin Young (U. of Pittsburgh) for kindly providing plasmids used in this project. We want to thank Edward Makarov and Dr. Santhi Gorantla for their help with the ddPCR. Some of the figure panels were made with BioRender.com. The following reagent was obtained through the NIH HIV Reagent Program, Division of AIDS, NIAID, NIH: NIH ARP 776, 349, 165, 11919. Gratitude is extended to Nicholas Conoan (UNMC Electron Microscopy Core) for his technical assistance. Sanger sequencing was performed with the help of UNMC Genomics Core through Ronald J. Redder. Genomics Core receives partial support from the National Institute for General Medical Science (NIGMS) INBRE-P20GM103427-19 grant as well as The Fred & Pamela Buffett Cancer Center Support Grant -P30 CA036727. This publication's contents are the sole responsibility of the authors and do not necessarily represent the official views of the NIH or NIGMS. We are also thankful to Evan A. Schroder for technical support. A final word of appreciation is directed to Daniel Chadash and the engineering team at Genome Compiler for making open access cloning software freely available online.

DNA Sanger sequence chromatograms and spreadsheets used for Inference of CRISPR Edits v2 (ICEv2), along with algorithm results and raw flow cytometry data are uploaded to Mendeley Data repository are publicly accessible at doi:10.17632/phyy89w9c2.1.

Supplementary material associated with this article can be found in the online version at doi:10.1016/j.ebiom.2021.103678.