Application of CRISPR-Cas9 Editing for Virus Engineering and the Development of Recombinant Viral Vaccines

Loss of function

Several studies have focused on loss of function through CRISPR-Cas9-based gene knockout strategies. Different types of virus genomes, including live viruses, ³² viral genomic DNA, ³³ BAC DNA, ³⁴ and integrated viral genes in cell genomes such as human immunodeficiency virus (HIV), ³⁵ have been successfully modified via the CRISPR-Cas9 system. The knockout efficiency (number of plaques with deletion/total number of plaques) can reach 75% mediated by the NHEJ repair mechanism for Marek's disease virus (MDV) in chicken embryo fibroblast cells ³² or 85% by HDR pathway for vaccinia virus (VACV). ³⁶ Multiple gene knockouts can also be achieved by simultaneous editing using several gRNAs. ³⁷ Besides Cas9, other Cas endonucleases such as Cas12a and Cas13a also have the capability of inducing strong inhibition of virus in infected cells.^38,39 Base editing using a Cas9 nickase or dCas9–cytidine deaminase fusion protein has also been performed to induce premature stop codon mutagenesis in virus genomes with 100% efficiency, but the editing has been reported only in E. coli based on the BAC system at present. ⁴⁰ The majority of previous studies have obtained good (>50%) knockout efficiency, while a few reported low recombination rates and difficulty in purification of edited clones. ⁴¹ For some viral infections where multiple viral genomes need to be targeted within the same cell, such as Kaposi's sarcoma herpesvirus in endothelial cells^42,43 and MDV in lymphocytes, ⁴⁴ there are challenges in using CRISPR to target all copies. Therefore, the research results should be cautiously tested to confirm whether the residual wild-type viral gene products influence the phenotype of interest. Cell lines stably expressing Cas9 endonuclease and sgRNAs have been constructed for successful viral gene editing. However, whether the cell lines are suitable for long-time antiviral application should be carefully verified, since some reports have shown that the virus could escape from CRISPR-Cas9-mediated inhibition after passages. ⁴⁵

Gene insertion and replacement

The last few years have witnessed increased application of CRISPR-Cas9-based gene editing in virology for structural and functional studies of viral genes and to understand virus–host interactions. ⁴⁶ These include the insertion of foreign genes as well as the replacement of viral genes in multiple viruses (Table 1). Efficiency of inserting foreign genes via the HDR pathway with homologous arms^36,47,48 and with non-homologous strategies have also been explored to enhance the efficiency of insertion.^49,50 Markers and reporters have been widely used in CRISPR-Cas9 knock-in platforms to enable visualization and easy purification of viruses. For example, a stable recombinant pseudorabies virus (PRV) strain expressing firefly luciferase and enhanced green fluorescent protein (eGFP), developed through CRISPR-Cas9-based knock-in, was very useful for the screening of antiviral compounds and gRNAs. ⁵¹ Similarly, reciprocal swapping of gD and gI glycoprotein genes of two PRV strains to generate chimeric viruses was valuable for comparative study. ⁵² Assisted with single-cell fluorescence-activated cell sorting (FACS), pure recombinant virus clones can be obtained in a single round of screening. ⁴⁷ Gene replacement and interchange are usually done through HDR using single or dual gRNAs and donor-containing homologous arms. A number of studies have used reporter gene ³⁶ or intermediates ⁴⁷ for exchange. Compared to other traditional recombination methods, CRISPR-Cas9-based gene exchange is a much easier and rapid approach for the comparative study of viral gene function. ⁵³

Gene activation or deactivation

Infection by several viruses results in a chronic infection where the viral genome is held in a latent, epigenetically silenced state, with periodic reactivation. While latency is one of the hallmarks of herpesvirus infections, latent reservoirs of viruses, such as HIV type 1 (HIV-1), is also very significant, since latency is one of the main obstacles for elimination of HIV by antiretroviral therapy. An effective therapeutic strategy would be the reactivation of HIV-1 from its latent state, making patients susceptible to antiretroviral therapy. Modified Cas9 without the endonuclease activity fused with transcriptional activators and repressors has been used for targeted CRISPR activation (CRISPRa) and CRISPR interference (CRISPRi), respectively. Cas9 variants with mutations in their two endonuclease domains deactivate the nuclease activity of Cas9 (nuclease-deficient disabled Cas9 [dCas9]), enabling the development of programmable RNA-dependent DNA-binding proteins. ⁵⁴ Different protein or functional domains have been successfully fused to dCas9 to develop sequence-specific gene expression or cellular localization tools. ⁵⁵ For example, several repurposed CRISPR-Cas9 systems have been employed to activate latent HIV. The dCas9 fused with a tetrameric repeat of the herpes simplex virus (HSV) VP16 transactivation domain VP64⁵⁶ and/or a synergistic activation mediator (SAM) complex^56,57 has been introduced to target the long terminal repeat enhancer region of HIV-1, resulting in robust activation of latent HIV-1 infection in human T cells. Conversely, repurposed CRISPR-dCas9 fused to transcriptional repression domain derived from Kruppel-associated box (KRAB) has also been effective for the epigenetic silencing of the HIV proviral DNA. ⁵⁸

RNA targeting

CRISPR technology has also been recently developed for RNA targeting. Although most CRISPR-Cas9-based editing focuses on DNA viruses or their DNA replication intermediates,^59,60 there are also some RNA targeting Cas nuclease-mediated genome editing of ssRNA viruses^61–63 and even dsRNA viruses. ⁶⁴ The Cas9 endonuclease from Francisella novicida can target diverse viruses including both +ssRNA viruses and –ssRNA viruses via an engineered RNA-targeting guide RNA, ⁶¹ though low RNA-cleavage efficiency and possibility of inducing off-target effect on cellular DNA have been reported. ⁶⁵ The ability of class 2 type VI CRISPR effector Cas13 to target a wide range of ssRNA viruses has been demonstrated. Computational analysis of more than 350 mammalian ssRNA viral genomes has been explored for the broad utility of Cas13. ⁶² This study has demonstrated potent activity against three distinct ssRNA viruses: lymphocytic choriomeningitis virus, influenza A virus (IAV), and vesicular stomatitis virus. ⁶² Lately, antiviral strategy to combat SARS-CoV-2 and H1N1 IAV has been developed based on Cas13d and defined crRNAs to target conserved viral gene for RNA cleavage or reduce virus replication. ⁶³ A strategy of abrogation of porcine reproductive and respiratory syndrome virus infectivity by CRISPR-Cas13b-mediated viral RNA cleavage in MARC-145 cells has been developed to target two viral genes with an all-in-one system expressing Cas13b and duplexed crRNA cassettes. ⁶⁶

Modification of cell lines for vaccine production

With most of the new generation of vaccines produced in cell culture systems, increasing the viral titers or production of antigens is a major priority for the vaccine manufacturing industry. Recent advances in genome-editing techniques also open up the prospects of a relatively unexplored area of research for increasing vaccine yield through genetic engineering of host cell lines.^109–112 Viruses grown in such engineered cell lines can reach much higher titers (up to 5–7 logs higher in Vero cell line ¹¹² compared to wild-type cells). Furthermore, some of the engineered cell lines may have unexpected capability of enhancing replication of other viruses. ¹⁵ CRISPR-Cas-mediated metabolic engineering has also been extended to enhance the production of vaccine antigens.^113,114 For example, knocking out a glutamine synthetase encoding the glul gene from the Chinese hamster ovary (CHO) cell line has been done to improve the glutamine metabolism selection system, ¹¹⁵ which in turn can potentially increase the production of vaccine antigen in these cells. Knockout of DNA methyltransferase Dnmt3a in CHO cells provided an enhanced long-term stable cell line of transgene expression with low methylation rate of the global DNA, providing the potential for improving the production of recombinant protein, including vaccine antigens in CHO and other mammalian cells. ¹¹⁶ Using HDR, gene expression cassettes can also be integrated into specific genomic locus in CHO cells to generate stable cell lines with manageable and predictable features for large-scale production. ¹¹⁷

CRISPR-Cas9 libraries

High-throughput screening based on CRISPR-Cas9 libraries has become an attractive and powerful technique to screen target genes for functional studies. Genome-wide CRISPR-Cas9 screens are increasingly used for identifying receptors, antiviral proteins, and other important genes in virus–host interaction and virus restriction factors.^89,118–120 CRISPR-Cas9 sgRNA libraries can be designed using online tools and databases such as Cas-OFFinder (www.biootools.com), ¹²¹ Addgene (https://www.addgene.org/crispr/libraries/), Cas-Database, ¹²² CRISP-view, ¹²³ DepMap, ¹²⁴ BioGRID ORCS, ¹²⁵ and lately iCSDB database (https://www.kobic.re.kr/icsdb/), DepMap portal, and BioGRID ORCS. ¹²⁶

Herpesviruses

Herpesvirus-based vectors have been widely used in the development of a variety of vaccines and therapeutics due to their inherent advantages, including limited host range, large genome size capable of accommodating multiple inserts, stable attenuation, ability for inducing strong cell-mediated and humoral immune responses, stable and robust expression of foreign genes, as well as good safety records. For these reasons, a number of human and animal herpesviruses have been successfully developed as recombinant vectors.

HSV type 1 (HSV-1) harbors a 150 kb large genome, and half of its 77 genes are dispensable for virus replication, ¹²⁸ giving abundant vector capacity. Additionally, early studies have shown that the viruses lacking certain genes have replication restricted to cancer cells, demonstrating the potential for developing HSV-1 for oncolytic virus therapeutics. ¹²⁹ Large fragments have been efficiently knocked into sites of HSV-1¹²⁷ and HSV-2 genomes, ⁸¹ sometimes as multiple copies, ⁴⁸ suggesting the feasibility of targeted CRISPR-Cas9-mediated genome-editing strategies for developing HSV-1-based attenuated vaccines and oncolytic vectors. Lately, a non-homologous insertion (NHI) strategy has also been explored for rapid and high-efficiency knock-in of foreign genes. ⁵⁰ The NHI strategy also paved a way for accelerating recombinant virus selection based on its inhibitory effect on wild-type and reverse-integrated viral genomes.

In animal herpesviruses, CRISPR-Cas9 has dramatically enhanced the homologous recombination efficiencies between PRV genomes and a linearized BAC transfer vector, ¹³⁰ or PRV and a double gene substitution donor. ⁴⁷ We have recently developed a simple and rapid CRISPR-Cas9-based strategy for constructing a recombinant herpesvirus of turkey (HVT) vector delivering the VP2 protein of infectious bursal disease virus (IBDV) through a NHEJ-dependent repair pathway and a Cre-LoxP system to eliminate the reporter protein.^49,131 This strategy, for incorporation of genes encoding foreign antigens into the HVT genome, can lead to the rapid construction of recombinant vaccines for efficacy studies. Using this strategy, expression cassettes of infectious laryngotracheitis virus (ILTV) gD, gI, and the H9N2 AIV hemagglutinin have been inserted sequentially into the distinct locations of the recombinant HVT-IBDV-VP2 viral genome, generating a triple insert recombinant vaccine candidate. ¹³² This offers the prospect of rapidly developing multivalent HVT-vectored vaccines capable of inducing simultaneous protection against multiple avian diseases, particularly to overcome the interference between individual recombinant HVT vaccines expressing different antigens are administered simultaneously. Other herpesvirus vaccine vectors such as duck enteritis virus^133,134 and ILTV ¹³⁵ have also been developed via NHEJ-CRISPR-Cas9 and Cre-Lox systems. Recombinant MDV vector carrying all the CRISPR-Cas9 components using an attenuated MDV vaccine strain (814 strain) for in vivo editing of reticuloendotheliosis virus (REV) and avian leukosis virus subgroup J (ALV-J) infection have recently been reported,^136,137 providing efficient delivery of CRISPR components into chickens, inducing a drastic reduction of REV and ALV-J viral load, significantly diminishing clinical disease.

Poxviruses

After the successful eradication of smallpox, VACV has been modified as a vector for vaccine and immunotherapeutic candidates. The most widely used method for modification of VACV is based on homologous recombination. ¹³⁸ A highly efficient CRISPR-Cas9 method accompanied with HDR repair, using a modified Cas9 without the nuclear localization signal (NLS) to target the cytoplasmic location of VACV replication, has been reported to generate mutations in the N1L and A46R genes of VACV. ³⁶ Simultaneous and rapid homologous recombination of multiple target sites in both N1L and A46R genes have also been successful achieved. To obtain a marker-free VACV vector, Cre-LoxP and Flp-FRET systems have been introduced into the strategy above for marker gene excision. ⁶⁷ A marker-free TK and N1L double gene–deleted VACV vector was efficiently constructed by these marker-free systems, providing significant potential for developing new vaccines and vectors for clinical application. However, all studies using CRISPR-Cas9 on VACV have not given satisfactory results. For example, one recent study showed that the Cas9/gRNA complexes can cut VACV genomes efficiently, but generated very few (from <1% to no more than 10%) recombinant virus compared to the relatively high efficiency of gene editing in other viruses such as HSV (40–60%). ¹³⁹ This is thought to be due to the low efficiency of HDR and NHEJ repair pathways that leaves the cut viral genomes unable to be replicated. However, in a recent study, the inefficiency of DNA repair of the VACV genome after Cas9 cleavage was used as an advantage for efficient selection method for the generation of VACV recombinants or other genetic variants. ¹³⁹

Avipox vectors such as Canarypox ALVAC have been widely studied and used as an efficacious and safe viral vector for mammals as several canarypox-vectored recombinant vaccines have been commercialized for years. Based on a bivalent ALVAC-vectored recombinant vaccine carrying the F gene and H gene of canine distemper virus (CDV), a new ALVAC recombinant virus has been constructed with the insertion of the matrix (M) gene through HDR-mediated CRISPR-Cas9 gene editing. ¹⁴⁰ ALVAC recombinant with inserted CDV M gene has shown rapid and greater antibody responses compared to the existing vaccines, possibly because of the VLP structure assembled in the presence of the M protein.

Transgene delivery vectors

CRISPR-Cas9 complexes for targeted gene editing are currently used mainly in preclinical trials, with a few examples of therapeutics use in genetic diseases. However, CRISPR-Cas9 complexes delivered through different viral vectors are also valuable for antiviral therapies. Viral vectors used as transgene delivery vehicles of CRISPR-Cas9 complexes can be classified into two classes: integrating viruses such as murine leukemia virus, ¹⁴¹ foamy virus, ¹⁴² HIV, and lentivirus, and non-integrating viruses such as adeno-associated viruses (AAV), ¹⁴³ AdV, ¹⁴⁴ and Sendai virus. ¹⁴⁵ Up to now, lentivirus vectors are usually the first choice for ex vivo gene correction and screening, while AAVs are preferred option for in vivo gene transfer. ¹⁴⁶ Lentiviruses are highly effective for genome-wide CRISPR-Cas9 screening and delivering the CRISPR-Cas9 machinery to the more difficult-to-transfect cells. Despite their advantages, lentivirus vectors which carry an RNA genome can integrate into the host genome, inducing long-term constitutive CRISPR-Cas9 expression. ¹⁴⁷ However, lentivirus-like particles have been developed to achieve transient expression of genome-editing proteins,^148,149 introducing an alternative solution for lentiviruses. AAV can efficiently infect a broad range of tissue types and yield very high viral titers, making it ideal for the delivery of antiviral CRISPR-Cas9 to the virus-infected cells. ¹⁵⁰ However, AAV can integrate randomly into the genome at low rates, ¹⁵¹ and the risk of such integration into the Cas9-induced DSBs during editing should be considered. ¹⁵²

Efficiency

Several factors influencing the gene editing and transgene insertion efficiency have been identified. The cleavage and repair efficiency at the target site are dependent on the characteristics of gRNA and target DNA. To achieve the most efficient gRNA recognition and cleavage, redundant sgRNAs targeting different cleavage sites can be designed and compared via several simple methods such as T7 endonuclease 1 mismatch assay, ¹⁵⁷ surveyor mismatch cleavage assays, ¹⁵⁸ High-Resolution Melting assay, ¹⁵⁹ indel detection by amplicon analysis, ¹⁶⁰ universal donor as reporter, ¹⁶¹ or targeted next-generation sequencing. ¹⁶² Dual or multiple gRNAs strategies are also helpful in improving editing efficiency.^37,48 With the expanding repertoire of CRISPR-Cas endonucleases (SpCas9, SaCas9, Cas12a, CjCas9, etc.), the most efficacious system for target editing can be selected and optimized. ¹⁶³ Compared to the expression of Cas9 from plasmid-based system or from the Cas9 mRNA, direct delivery of the synthetic Cas9 ribonucleoprotein complex offers several advantages, including its transient activity and rapid degradation with reduced off-target effects, ¹⁶⁴ in addition to the high efficiency and rapid cleavage, without the need for promoter or codon optimization.^165,166

CRISPR-Cas9 editing involves a two-step process of cleavage of the DNA followed by the DNA repair. In the editing of viral genomes, while the cleavage efficiency of Cas9/gRNAs can be high (often reaching 90% as detected by polymerase chain reaction), the efficiency of repair of cleaved fragments is always much lower (1–7%). Although many of the multiple copies of the viral genomes are efficiently cleaved by the CRISPR-Cas9 complex, large proportions of the viral genomes cannot be efficiently repaired, resulting in inhibition of viral replication. While the exact reasons for the inefficient repair of the viral genomes are not known, access of the components of the repair machinery at the sites of viral replication could be a major reason. In most cases, the efficiency of NHEJ-based strategies was significantly higher than that of the HDR-based strategies,^50,167,168 especially when insertion of large foreign genes is required. ¹⁶⁹ However, there are also some unsuccessful cases of viral genome editing via NHEJ repair.^41,139 For example, NHEJ repair at low efficiency following CRISPR-Cas9 targeting of VACV has been reported recently. ¹³⁹ Cas9 cleavage has been shown to be efficiently achieved, but recombination has been inefficiently made by either NHEJ or HDR. As VACV is a nuclear-independent virus that has the ability to replicate in cytoplasm, even in enucleated cells, ¹⁷⁰ one presumption is that VACV genome might have little opportunity to obtain access to NHEJ repair mediators such as DNA ligase IV. ¹⁷¹ For viruses replicating in the cytoplasm, removing N- and C-terminus NLS from the Cas protein sequence can enhance the editing efficiency. ¹⁷² In addition, when NHEJ is efficiently induced during genome editing, it is likely that the factors restricting the frequency of inducing HDR pathway are affected. However, chemicals such as the NHEJ inhibitor SCR7 ¹⁷³ and NU7026 ¹⁷⁴ have been used to increase on-target deletions and recombination efficiency.

Compared to gene editing of the host genome, the generation of recombinant progeny virus is relatively complex, since the viral genome replicates very fast and constantly produce abundant target sites. Therefore, the proliferative characteristics of virus should be considered during the optimization of recombinant strategy. After transfection, Cas9 protein is detected at 4–8 h,^139,140 reaching the maximum expression at approximately 20 h. ¹⁴⁰ Cas9 cleavage of the target is potentially happening very early and fast, until the limit at which genome or donor DNA could no longer be cut is reached. The difference in the speed of CRISPR-Cas9 cleavage compared to the virus replication rates, as well as the relative inefficiency of repair pathways on the viral genome, are important factors affecting viral genome editing. Adjusting of the multiplicity of infection of viruses, the inoculation time, and the harvest time may also be helpful to enhance viral genome editing efficiency. ¹²⁷ Genetic variations in the viral genome may also affect the editing efficacy. Studies demonstrated that a 1 nt mismatch at the cleavage site or two mismatches anywhere in target sequence significantly reduced the inhibitory effect on HIV-1. ¹⁰¹ It is also important to note that Cas9/gRNA complexes in the cells might influence the proliferation of wild-type viruses. It has been reported in different recombinant virus studies that the CRISPR-Cas9 system in eukaryotic cells significantly inhibited wild-type virus replication, leading to a higher proportion of recombinant viruses among progeny viruses.^{127,139,175,176}

When knock-in of foreign genes is carried out, the efficiency of recombination is also related to the features of donor construct. The length of the foreign fragment has been shown to be inversely associated with insertion efficiency of both HDR ¹⁷⁷ and NHEJ strategies. ⁵⁰ The mass ratio of Cas9 plasmid to donor plasmid can be adjusted during co-transfection to maximize the recombination efficiency.

Several delivery systems for the Cas9/sgRNA complex have achieved remarkable editing efficiencies in “easy to transfect” cultured cells. However, the efficiency may drop sharply when switching to transfection of recalcitrant cell lines, exacerbated when long DNA templates were co-transfected with gRNA and Cas9 to achieve gene insertion or replacement, often resulting in no editing event, useless indels, or error insertions. To achieve better delivery and gene-editing efficacies, parameters such as cell confluency and culturing methods could be adjusted; varied delivery systems could be tested as well. Compared to non-viral delivery systems, viral delivery systems such as lentivirus, adenovirus, or AAVs have been characterized by high transduction efficiency both ex vivo and in vivo. However, the application of viral vectors in developing recombinant vaccines are constrained by their exogenous viral substance, limited cargo capacity, and risk of genome integration. Baculoviral delivery vectors have a good transduction efficiency. They also have distinct characteristics that they are both replication- and integration-deficient in mammalian cells, giving good prospects for the delivery of CRISPR tools. Baculoviral vectors have large heterologous DNA cargo holding capacity, rendering them highly suitable for accommodating all components of a CRISPR machinery in a single vector.^178,179

Target specificity and off-target effects

Potential off-target effects remain one of the major concerns with CRISPR technology. The RNA-guided nucleases used in CRISPR systems have been shown to bind to several mismatch sequences in the binding of sgRNA and target DNA. ¹⁸⁰ These off-target effects have the potential of generating unexpected gene mutations or chromosomal translocations. To tame off-target mutagenesis, a number of studies have identified many factors influencing off-target CRISPR-Cas9 editing. Sequence complementarity, PAM recognition specificity, target sequence homology, and Cas9 expression level are the main factors influencing off-target effects.^181,182 In addition, other factors such as gRNA binding stability and substrate availability can also influence the likelihood of off-target editing.^183,184 Therefore, a number of strategies have been adopted to fine-tune the components of CRISPR-Cas9 system to enhance target specificity and eliminate off-target effect.

However, when it comes to viral genome editing, fewer off-target effects can be expected. The genome of viruses (that ranges from 1 kb to 2.47 Mb ¹⁸⁵ ) is far smaller than it is for the host genome (3 × 10⁹ nt), and no potential off-target effects were found in many viral editing studies.^37,48,68,127 It is, however, worth noting that in viral genomes containing overlapping genes, multiple genes might be cleaved and repaired.

Fidelity of editing

CRISPR-Cas9 editing outcomes have been thought to be highly unpredictable due to its error-prone DSB repair systems such as canonical NHEJ, ¹⁸⁶ theta-mediated end joining, microhomology-mediated end joining, and even HDR repair. ¹⁶⁷ Because of this, the NHEJ-based CRISPR-Cas9 system could generate mutant viruses resistant to Cas9/sgRNA. ¹⁸⁷ The fidelity of NHEJ can be affected by the DNA sequences rather than the length of donor. ⁵⁰ Some indels, especially those in coding regions, may not reduce virus replication, as these may not be recognized by the same Cas9/sgRNA complex as a result of escaping Cas9 cleavage. ¹⁸⁸

To reduce the occurrence of undesired mutations, certain machine learning algorithms have been developed based on the non-random nature of Cas9-induced DSB repair and abundant data available in the public domain, including inDelphi (https://indelphi.giffordlab.mit.edu), ¹⁸⁹ FORECasT (https://partslab.sanger.ac.uk/FORECasT), ¹⁹⁰ and SPROUT (https://github.com/amirmohan/SPROUT). ¹⁹¹ These databases available online can effectively perform predictions of the activity of gRNAs and their repair outcomes with reasonable accuracy. ¹⁸⁶ Besides, escape variants can be prohibited using other strategies such as targeting multiple essential genes and using redundant sgRNAs,^37,172,192 modification of sgRNAs, and through development of new Cas enzymes such as Cas12a ¹⁹³ to cleave at site outside the target. To avoid the invalid iterative cut-and-repair process of imprecise NHEJ-mediated repair, the junction sequence formed at the CRISPR-Cas9 recognition site after cleavage and integration should be carefully designed to prevent re-cuttings.