Computational analysis of cas proteins unlocks new potential in HIV-1 targeted gene therapy

Abstract

Introduction: The human immunodeficiency virus type 1 (HIV-1) pandemic has been slowed with the advent of anti-retroviral therapy (ART). However, ART is not a cure and as such has pushed the disease into a chronic infection. One potential cure strategy that has shown promise is the Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR)/Cas gene editing system. It has recently been shown to successfully edit and/or excise the integrated provirus from infected cells and inhibit HIV-1 in vitro, ex vivo, and in vivo. These studies have primarily been conducted with SpCas9 or SaCas9. However, additional Cas proteins are discovered regularly and modifications to these known proteins are being engineered. The alternative Cas molecules have different requirements for protospacer adjacent motifs (PAMs) which impact the possible targetable regions of HIV-1. Other modifications to the Cas protein or gRNA handle impact the tolerance for mismatches between gRNA and the target. While reducing off-target risk, this impacts the ability to fully account for HIV-1 genetic variability. Methods: This manuscript strives to examine these parameter choices using a computational approach for surveying the suitability of a Cas editor for HIV-1 gene editing. The Nominate, Diversify, Narrow, Filter (NDNF) pipeline measures the safety, broadness, and effectiveness of a pool of potential gRNAs for any PAM. This technique was used to evaluate 46 different potential Cas editors for their HIV therapeutic potential. Results: Our examination revealed that broader PAMs that improve the targeting potential of editors like SaCas9 and LbCas12a have larger pools of useful gRNAs, while broader PAMs reduced the pool of useful SpCas9 gRNAs yet increased the breadth of targetable locations. Investigation of the mismatch tolerance of Cas editors indicates a 2-missmatch tolerance is an ideal balance between on-target sensitivity and off-target specificity. Of all of the Cas editors examined, SpCas-NG and SPRY-Cas9 had the highest number of overall safe, broad, and effective gRNAs against HIV. Discussion: Currently, larger proteins and wider PAMs lead to better targeting capacity. This implies that research should either be targeted towards delivering longer payloads or towards increasing the breadth of currently available small Cas editors. With the discovery and adoption of additional Cas editors, it is important for researchers in the HIV-1 gene editing field to explore the wider world of Cas editors.

Keywords: CRiSPR/Cas; HIV-1; cure strategy; gene therapy; guideRNA.

FIGURE 1

Cas9 editor influences the number of safe, broad, and effective (SBE) targets. Each column indicates the results of the NDNF pipeline for SpCas9 (left), SaCas9 (middle), and LbCas12a (right). The PAM sequence used is indicated in the graph labels. The top row indicates the broad gRNAs by plotting the on-target rate for each gRNA, those in green are above a 75% cutoff and considered broad. The second row indicates the safety of each gRNA by plotting the off-target count for each gRNA. Those with no hits are considered safe (above the blue line). The third row indicates the effectiveness of each gRNA by plotting the average RC index of the 40 bp window centered on the cut-site. The fourth row plots the counts of SBE gRNAs and their overlaps.

FIGURE 2

Higher promiscuity leads to more broad targets but fewer safe ones. The number of safe gRNAs was calculated and plotted for the wild-type PAMs indicated in the key for SpCas9 (blue), SaCas9 (orange), and LbCas12a (green) in solid lines using the left axis. The number of broad gRNAs was calculated and plotted for the wild-type PAMs for SpCas9 (blue), SaCas9 (orange), and LbCas12a (green) in dotted lines using the right axis.

FIGURE 3

The pattern of targetable regions is altered by PAM mutations. For each of the SpCas9, SaCas9, and LbCas12a variants identified, the pattern of broad (A), safe (B), and effective (C) gRNA is plotted against the target position in the HXB2 reference genome. (D) Indicates the overlapping sites of safe, broad, and effective gRNAs. The counts of each category are provided in Table 2. E) A bar graph showing the number of unique genomic positions targetable by SBE gRNAs for each PAM choice.

FIGURE 4

Increase PAM specificity leads to a decrease in SBE gRNAs and loss of targetable sites. The number of (A) SBE gRNAs and (B) targetable sites was plotted against the PAM specificity. The X-ticks are representative PAMs for each level of specificity. The blue line represents a y ∼ log(x) regression and the shadow indicates a 95% CI. A regression was performed for SBE ∼ log₁₀(PAM-promiscuity) and found no significant correlation. Alternatively, Unique-Sites ∼ log₁₀(PAM-promiscuity) found a significant correlation (p = 0.0039) with a modest R ² = 0.26.

FIGURE 5

Nomination and validation stages are unimpacted by random samplings. (A) The number of samplings each protospacer was nominated in during cross-validation. (B) A histogram of the standard deviation across each sampling for each protospacer. For (A) and (B) SpCas9 is in blue, SaCas9 is in orange, and Cpf1 is in green. (C) The mean validation hit rate was calculated across all samplings (x) and then was plotted against the hit rate obtained in each S(i) sampling (y). (D) A residual plot for the regression of Hit Rate S(i) ∼ Hit Rate Mean.

FIGURE 6

HIV-1 subtype distribution of full genomic sequences in the LANL dataset. (A) The entire 2021 LANL full genome dataset. (B) A subset of the LANL database randomly selected to be a globally representative validation dataset. (C) The remaining sequences used as a nomination set.

FIGURE 7

The effect of mutations on HIV-1 replication capacity. (A) The RC index of each induced mutation was plotted in blue. The black line represents a 40 bp rolling average centered on each point. The red line indicates the maximum RC index of a lethal mutation (0.1) while the green line indicates the minimum value of a tolerated mutation (0.2). Between the green and red lines indicates attenuated. (B) The number of lethal, attenuated, and tolerated mutations is plotted.