Multiplexed Gene Editing and Protein Overexpression Using a Tobacco mosaic virus Viral Vector

A new plant virus-based method enables gene knockout screening and plant genetic engineering using transient expression methods.

Abstract

Development of CRISPR/Cas9 transient gene editing screening tools in plant biology has been hindered by difficulty of delivering high quantities of biologically active single guide RNAs (sgRNAs). Furthermore, it has been largely accepted that in vivo generated sgRNAs need to be devoid of extraneous nucleotides, which has limited sgRNA expression by delivery vectors. Here, we increased cellular concentrations of sgRNA by transiently delivering sgRNAs using a Tobacco mosaic virus-derived vector (TRBO) designed with 5′ and 3′ sgRNA proximal nucleotide-processing capabilities. To demonstrate proof-of-principle, we used the TRBO-sgRNA delivery platform to target GFP in Nicotiana benthamiana (16c) plants, and gene editing was accompanied by loss of GFP expression. Surprisingly, indel (insertions and deletions) percentages averaged nearly 70% within 7 d postinoculation using the TRBO-sgRNA constructs, which retained 5′ nucleotide overhangs. In contrast, and in accordance with current models, in vitro Cas9 cleavage assays only edited DNA when 5′ sgRNA nucleotide overhangs were removed, suggesting a novel processing mechanism is occurring in planta. Since the Cas9/TRBO-sgRNA platform demonstrated sgRNA flexibility, we targeted the N. benthamiana NbAGO1 paralogs with one sgRNA and also multiplexed two sgRNAs using a single TRBO construct, resulting in indels in three genes. TRBO-mediated expression of an RNA transcript consisting of an sgRNA adjoining a GFP protein coding region produced indels and viral-based GFP overexpression. In conclusion, multiplexed delivery of sgRNAs using the TRBO system offers flexibility for gene expression and editing and uncovered novel aspects of CRISPR/Cas9 biology.

Gene editing tools allow for the precise targeting of DNA for purposes of changing nucleic acid sequences. This is performed by a variety of systems relying on DNA binding proteins fused to nucleases (Porteus and Carroll, 2005; Cermak et al., 2011), ribonucleic acids (Dong et al., 2006), and protein-nucleic acid complexes (Cong et al., 2013; Jinek et al., 2012; Mali et al., 2013). Recently, the Streptococcus pyogenes CRISPR/Cas9 system has been adapted as a programmable DNA targeting RNA-guided nuclease complex for gene editing (Doudna and Charpentier, 2014). The S. pyogenes-based system has been widely adopted primarily due to the simplicity of its parts, consisting of a sequence-specific synthetic single guide RNA (sgRNA) and the sgRNA programmable Cas9 nuclease (Mali et al., 2013). CRISPR gene editing technology creates DNA double-stranded breaks (DSBs), which activate the native host DNA repair mechanisms of nonhomologous end joining (NHEJ) or homologous dependent repair. NHEJ repairs DSBs by inserting or deleting (indels) nucleic acids to restore the integrity of the host dsDNA. These indels cause localized DNA disruption and have been used for sequence-specific disruption of downstream gene products, such as proteins and long noncoding RNAs (Schiml et al., 2014; Ran et al., 2013).

Current CRISPR-based editing processes in plants have focused on delivery of the Cas9 nuclease and sgRNAs by transformation technologies or transient delivery to protoplasts (Cong et al., 2013; Nekrasov et al., 2013). Transient protoplast editing, while relatively efficient, requires the regeneration of edited protoplast cells to produce plant lines. Alternatively, crop plants carrying T-DNA with coding regions for Cas9 and sgRNAs are predisposed to regulation or require traditional breeding techniques to remove unwanted DNA inserts prior to commercialization (National Academies of Science, Engineering, and Medicine, 2016). For these reasons, an efficient transient method for generating stable mutations in tissue is desirable for the development of edited plant lines and as a rapid screening tool.

Viral vectors are used in biotechnology for their ability to replicate and produce recombinant products in a wide range of hosts (Scholthof et al., 1996, 2002). The use of viral vectors as delivery devices for gene editing tools offers the potential of producing a high incidence of edited cells without the incorporation of recombinant DNA (Scholthof et al., 1996). DNA (Baltes et al., 2014; Yin et al., 2015) and RNA (Ali et al., 2015) viral-based replicons have been tested as vectors for delivery of CRISPR components to create gene knockouts and gene insertions. To date, the efficiency of transient gene editing by viral vectors has been relatively low, and delivery methods demonstrating a knockout phenotype have relied on transgenic technology in some capacity (Yin et al., 2015; Ali et al., 2015). CRISPR-based transient screens could be used for functional genetic studies, much like viral-induced gene silencing (VIGS) screens (Liu et al., 2002). For example, a Nicotiana benthamiana GFP-expressing transgenic line (16c; Ruiz et al., 1998) has been used extensively as proof-of-concept for RNA silencing studies due to the phenotype associated with a knockdown of GFP expression (Voinnet et al., 2000; Anandalakshmi et al., 1998, 2000) and was selected in our studies for its potential to establish a proof-of-principle for using viral delivery of CRISPR components for editing efficiency tests.

TMV RNA-based overexpression (TRBO), is a coat protein (CP) deletion mutant of the Tobacco mosaic virus (TMV) U1 strain, which was initially developed as an agroinfiltration-based delivery tool for expressing recombinant protein coding sequences in host cells (Lindbo, 2007). The CP deletion prevents the virus from systemically moving throughout infected plants, while still allowing localized cell-to-cell movement through the movement protein. TRBO has been exploited in biotechnology for its ability to produce large amounts of a protein of interest in hosts such as N. benthamiana, while consolidating the infection to infiltrated leaves (Lindbo, 2007). Recombinant protein production within infiltrated leaves is due to high quantities of transcript produced through the TRBO CP subgenomic RNA promoter. The transient Agrobacterium tumefaciens-based delivery of CRISPR components, in current delivery systems that utilize constitutive promoters, is limited by the relatively low production of sgRNAs that lack 5′ and 3′ nucleotide overhangs (Cong et al., 2013; Nekrasov et al., 2013). We speculated that the potentially high output of sgRNAs by TRBO could provide a very effective editing platform to boost efficiency when used in combination with Cas9.

In this study, we show that TRBO is a suitable vector for transient delivery of high concentrations of sgRNAs in N. benthamiana for Cas9 programming through the quantification of gene-specific indel percentages. We hypothesized that through targeting the GFP coding region in transgenic N. benthamiana 16c plants that there would be a phenotypic response along with an increase in indel percentage. We designed a sgRNA delivery platform involving the incorporation of RNA catalytic ribozymes to be transcribed from TRBO (Gao and Zhao, 2014). After optimizing deployment of sgRNAs, which surprisingly did not necessitate the use of ribozymes, NHEJ-derived indel values of nearly 70% in the GFP coding region were achieved in Cas9 coinfiltrated tissue. These results, from in planta experiments, contradict what we found using in vitro Cas9 cleavage assays, in which Cas9-based DNA DSBs did not occur when predicted TRBO-encoded subgenomic sgRNAs did not have 5′ and 3′ sgRNA overhang removal capabilities, suggesting that an unknown processing event is occurring in plants. Further in planta experimentation showed the majority of mutations occurred during 2 to 3 d postinfiltration (dpi), and the occurrence of indels within mgfp5 was reliant on TRBO replication. Furthermore, these plants showed a reduction in GFP expression as well as a noticeable chlorophyll-associated red autofluorescent phenotype, confirming GFP protein expression reduction. We also successfully demonstrated the ability to target two paralogs with a single sgRNA, as well as the codelivery of two adjoining sgRNAs using a single TRBO construct. Additionally, we report the previously unexplored ability of TRBO to deliver a single subgenomic RNA encoding both a biologically functional protein along with a sgRNA for editing a host gene in planta.

RESULTS

Development of a Genomic Target and sgRNA Deployment Optimization

Constitutive GFP-expressing transgenic N. benthamiana 16c plants were selected for our proof-of-concept study for the TRBO-sgRNA delivery system due to the easily detectable red auto-fluorescent phenotype exhibited upon disruption of GFP production (Ruiz et al., 1998). In addition to the phenotype observed with GFP inactivation, GFP protein expression levels can be verified with assays using commercially available antibodies. The mgfp5 sequence was used to design a sgRNA around a BsgI restriction site for downstream detection of indels using a restriction enzyme resistance assay (Cong et al., 2013; Nekrasov et al., 2013). To evaluate whether RNA replicons can deliver high quantities of sgRNAs in planta, we used a TMV-based CP deletion mutant, TRBO, which replicates to high concentrations in infiltrated N. benthamiana (Lindbo, 2007; Fu et al., 2014). Due to the specificity of the sgRNA in CRISPR-based gene editing systems, we hypothesized that nucleotide overhangs 5′ proximal to the spacer sequence (Dahlman et al., 2015; Fu et al., 2014) or 3′ proximal to the scaffold sgRNA (Konermann et al., 2015; Zalatan et al., 2015) would lower the efficiency of the system and the subsequent occurrence of DSBs in planta. The 5′ and 3′ sgRNA overhangs were expected based on what is known about the TRBO CP subgenomic RNA, which contains both a 5′ 63-nucleotide sequence as well as the TMV 3′ UTR (∼200 nucleotides) downstream of the inserted sequence (Dawson et al., 1989; Buck, 1999). To circumvent possible compatibility issues of Cas9-sgRNA-DNA binding and subsequent DSBs due to 5′ and 3′ sgRNA nucleotide overhangs, the construction of three independent sgRNA delivery devices were designed using catalytic RNAs (ribozymes; Fig. 1A).

Figure 1.

TRBO-sgRNA delivery in planta. A, Schematic of TRBO-sgRNA constructs. The predicted subgenomic RNA transcriptional start site is indicated by “+1.” Three TRBO-sgRNA constructs were made to optimize sgRNA delivery: TRBO-gGFP consisting of the guide GFP (gGFP) only (red), TRBO-gHDV consisting of gGFP with a 3′ proximal hepatitis delta virus (HDV) ribozyme (blue), and TRBO-RGR consisting of gGFP with a 5′ hammerhead ribozyme (yellow) and a 3′ HDV ribozyme. B, The pHcoCas9 construct contains an N-terminal triple FLAG-tag (3×FLAG), NLS, and a human codon-optimized Cas9 nuclease followed by a C-terminal NLS. Transcription is initiated by a 35S promoter and terminated by a 35S terminator. The pHcoCas9 transcripts contain both a 5′ TEV UTR and 3′ TEV UTR for increased translation efficiency. C, Immunoblot assays from N. benthamiana leaf sampled at 3 dpi were taken from two replicates of two different Cas9 expression plasmids. Anti-Flag probes were used for Cas9 protein expression detection of a previously constructed vector (pFGC) and pHcoCas9. D, Three technical replicates of BsgI restriction enzyme resistance assay performed on mgfp5 amplicons from genomic DNA samples taken 12 dpi. Infiltration treatments are as follows: pHcoCas9 only (Control), pHcoCas9/TRBO-gGFP (TRBO-gGFP), pHcoCas9/TRBO-gHDV (TRBO-gHDV), and pHcoCas9/TRBO-RGR (TRBO-RGR) codelivery. Numbers under each lane indicate predicted indel percentages quantified using ImageJ. E, PCR and RT-PCR products, respectively amplified from TRBO plasmid for each sgRNA delivery construct (P) or from 14 dpi coinfiltrated pHcoCas9/TRBO-sgRNA 16c N. benthamiana plants cDNA (RT) for detection of construct expression and retention in planta. Each plasmid amplicon (P) serves as a construct specific molecular weight control for the coinciding RT-PCR products.

The ribozymes were designed and placed to yield sgRNAs that either lacked both 5′ and 3′ overhangs, only removed 3′ overhangs, or that resulted in sgRNAs without any flanking ribozymes and thus predicted to yield a substrate with substantial 5′ and 3′ overhangs (Fig. 1A). With each of these sgRNA delivery constructs, it could be readily tested if overhangs affect the generation of Cas9/sgRNA DSBs in planta. The first construct, aimed at 5′ and 3′ overhang removal, has the sgRNA sequence flanked by a 5′ hammerhead (HH) ribozyme and a 3′ hepatitis delta virus (HDV) ribozyme (RGR), as described previously (Gao and Zhao, 2014). The second construct contained only a 3′ HDV ribozyme located 3′ proximal to the sgRNA (gHDV) sequence. The third construct consisted of a sgRNA without the presence of catalytic HH and HDV units (gGFP) predicted to result in sgRNA that carries both 5′ and 3′ nucleotide overhangs. Each of the three independent sgRNA delivery combinations (RGR, gHDV, and gGFP) were subcloned downstream of the TRBO CP subgenomic promoter for expression (Fig. 1A). The activity of the ribozymes was verified in vitro (Supplemental Fig. S1).

TRBO-Mediated Delivery of Biologically Active sgRNAs

To test the biological activity of TRBO-delivered sgRNAs in planta, the Cas9 nuclease must be expressed along with the TRBO-sgRNA delivery constructs. The presence of indels can then be assayed and quantified to estimate the efficiency of each sgRNA delivery construct. For Cas9/sgRNA codelivery, expression of the Cas9 protein within N. benthamiana using a minimal binary plasmid was considered to be optimal for the highest possible transient expression. A Cas9 cassette was constructed using a previously synthesized human codon-optimized Cas9 consisting of an N terminus 3× FLAG tag and N and C terminus nuclear localization signal (NLS; Addgene plasmid 42230) (Cong et al., 2013). Optimal Cas9 expression in plants was achieved using a CaMV 35S double promoter and translation enhancers including the 5′ and 3′ UTR regions of Tobacco etch virus (TEV; Fig. 1B). Protein expression was confirmed by western blot assays, following agroinfiltration of N. benthamiana. Our newly constructed Cas9 delivery cassette (pHcoCas9) had much higher rates of protein expression than a previously developed plant Cas9 expression vector (pFGC-pcoCas9; Addgene plasmid 52256; Li et al., 2013), which did not produce detectable Cas9 protein under our experimental conditions (Fig. 1C).

N. benthamiana 16c plants were coinfiltrated with A. tumefaciens cultures harboring the pHcoCas9 expression vector and either gGFP, gHDV, or RGR containing TRBO constructs. Host genomic DNA was sampled at 12 dpi and analyzed for indels. (Nekrasov et al., 2013; Cong et al., 2013). PCR amplicons of mgfp5 from each TRBO-sgRNA delivery system were incubated with BsgI restriction enzyme. Restriction-enzyme-resistant PCR products were present for each of the TRBO-sgRNA delivery treatments, indicating indel accumulation in planta, while control pHcoCas9-only treatments were readily digested by BsgI (Fig. 1D). BsgI resistant DNA was cloned and sequenced to confirm the presence of indels at the predicted Cas9 DNA nucleation sites (Supplemental Fig. S2). Image analysis software (ImageJ) was used to quantify the BsgI resistant amplicons or the predicted percentage of mutated DNA compared to the digested “wild-type” mgfp5 DNA. Upon gel quantification analysis, both gGFP and gHDV showed greater than 60% mutated mgfp5 DNA and RGR had a mutation rate of 46% (Fig. 1D).

We anticipated that the lower indel efficiency of the RGR construct, as compared to gGFP and gHDV, was due to a decreased fitness level of the RNA replicon. To test the stability of the TRBO-sgRNA delivery constructs in planta, RNA was isolated from Cas9/TRBO-sgRNA coinfiltrated tissue at 14 dpi from plants previously confirmed to contain indels by BsgI restriction enzyme resistance assays (Fig. 1D). RNA extractions from each treatment were subject to TRBO-specific reverse transcription (RT) reactions using equal quantities of total RNA. RT-PCR analysis was used to confirm the integrity of the expected sgRNA constructs followed by cloning and sequencing of each amplicon (Fig. 1E). Both gGFP and gHDV delivery constructs were confirmed to be stable in planta through sequencing analysis, but the RGR construct contained a point mutation in the HH ribozyme (Supplemental Fig. S3). Additionally, when TRBO-RGR was delivered without Cas9, a chimera of RT-PCR products accumulated in N. benthamiana 16c plants (Supplemental Fig. S3). The sequencing of one of the most prevalent amplicons from the RGR RT-PCR products indicated a complete deletion of the sgRNA from the TRBO genome (Supplemental Fig. S3). Following these results, we concluded that ribozymes, specifically the 5′ HH ribozyme, had a negative effect on the replication of TRBO. Importantly, the TRBO-sgRNA delivery construct without ribozymes was both sufficient and efficient for delivery of biologically active sgRNAs (Fig. 1D). In summary, the results show a novel feature that in planta, Cas9 tolerates the subgenomic RNA-mediated delivery of GFP-sgRNA carrying both 5′ and 3′ overhangs

5′ and 3′ sgRNA Overhangs Negate Cas9 DNA Nuclease Activity In Vitro

Since the in planta results using TRBO-sgRNA delivery containing sgRNA nucleotide overhangs was unexpected, we anticipated that these results might have larger implications in Cas9/sgRNA biology in in vivo systems. While it has been largely assumed that sgRNA overhangs prevent Cas9 catalytic activity, the proper experiments have not been performed in vitro to confirm this assumption (Jinek et al., 2012). However, there has been considerable evidence that insinuates that sgRNA overhangs within the delivery construct do not impede DSBs in both bacteria (Karvelis et al., 2013) and human cell lines (Cong et al., 2013) and also has been hinted at in plants (Jia and Wang, 2014).

To understand if this phenomenon is fundamental to Cas9/sgRNA or if it is a unique interaction that only occurs in vivo, we aimed at using our predicted subgenomic RNAs through the TRBO-sgRNA delivery methods previously tested in planta (Fig. 1D) and provide them along with Cas9 in vitro to determine if they create catalytically active complexes outside of cells. Subgenomic transcripts were produced by designing a forward primer with a 5′ T7 promoter sequence at the subgenomic transcriptional start site and a reverse primer located in the 3′ UTR. PCR amplification of TRBO-gGFP, TRBO-gHDV, and TRBO-RGR constructs created the predicted subgenomic RNA templates sub-gGFP, sub-HDV, and sub-RGR, respectively. Each of the templates was used for T7 RNA synthesis. In addition to the subgenomic sgRNAs, a template encompassing only the 100-nucleotide gGFP region was created (gGFP) as a positive control (Fig. 2A).

Figure 2.

In vitro Cas9 cleavage assay using a mgfp5 DNA template. A, Depiction of sgRNAs synthesized for loading into Cas9 in vitro Cas9 cleavage assays. The gGFP RNA carries no Cas9 overhangs and serves as the positive control. Predicted subgenomic RNAs (sub) were constructed for each of the construct used for the in vivo assay. TRBO-gGFP (sub-gGFP), TRBO-gHDV (sub-HDV), and TRBO-RGR (sub-RGR) all contain the 63-nucleotide subgenomic promoter sequence and the downstream 3′ UTR from the TRBO vector. B, mgfp5 template was PCR amplified from 16c genomic DNA, cloned into a TA plasmid, and used as a template for Cas9/gGFP cleavage. The plasmid was linearized with ScaI restriction enzyme when coincubated with Cas9 and either gGFP, sub-gGFP, sub-HDV, or sub-RGR (RNAs depicted in A). Red arrows indicate expected DNA cleavage product sizes while the blue arrow represents uncut mgfp5 template.

A DNA template was created by amplifying the mgfp5 gene containing the sgRNA target from 16c genomic DNA and cloned into a TA vector. Purified Cas9 was incubated with each synthesized sgRNA at room temperature to create Cas9/sgRNA complexes. Purified plasmid was then coincubated with each of the Cas9/sgRNA treatments along with ScaI restriction enzyme, which was used to linearize the plasmid. A positive Cas9/sgRNA reaction can be visualized through gel electrophoresis and the presence of digested linearized plasmid, yielding three DNA bands (Fig. 2B). Following these assays, it was confirmed that the subgenomic sgRNAs retaining the 5′ overhangs (sub-gGFP and sub-HDV) did not yield DNA cleavage while sgRNAs from which 5′ overhangs were removed (gGFP and sub-RGR) did cleave the DNA target (Fig. 2B). These in vitro results, along with the above results in planta, demonstrate that some unique interaction must be occurring to allow for TRBO-gGFP and TRBO-gHDV to create DSBs in vivo and that a comparable interaction and concomitant activity is lacking during the in vitro experiments.

Targeted Introduction of Indels Reduces GFP Expression

Following confirmation of indel disruption for the TRBO-delivered gGFP, gHDV, and RGR constructs in planta (Fig. 1D), we further tested if GFP protein production would be significantly hindered by the development of indels. Half-leaf assays were conducted with half of the leaf being infiltrated with one of the three TRBO-sgRNA delivery methods alone and the other half being subjected to coinfiltrations of a TRBO-sgRNA treatment and pHcoCas9. Protein samples taken 14 dpi were analyzed using western blotting with a GFP-specific antibody. GFP expression showed large reductions in GFP protein expression in pHcoCas9 coinfiltrated tissue with both TRBO-gGFP and TRBO-gHDV (Fig. 3), confirming our observations for both TRBO-gGFP and TRBO-gHDV restriction enzyme resistance assay. Since TRBO-gGFP was performing optimally and because it allows for simplicity of design compared to constructs carrying ribozymes, we focused on this construct for delivery from here on.

Figure 3.

16c N. benthamiana GFP protein expression in leaf tissue treated with TRBO-sgRNA constructs with and without Cas9. Anti-GFP HRP western blot for TRBO-sgRNA delivery alone (−) as well as pHcoCas9/TRBO-sgRNA co-delivery (+). Top, Western blot used to detect GFP (white arrow). Bottom, Coomassie stain of Rubisco (black arrow) used as a protein loading control for each sample. Abbreviations are the same as in previous figures.

Rapid TRBO-gGFP Derived Indels and the Corresponding Loss-of-Function Phenotype

To determine if mutations within the mgfp5 coding region of 16c plants are a direct consequence of viral replication or basal levels of expression from the 35S promoter used to launch the initial TRBO transcripts, a TRBO-gGFP replicase mutant (RM) was constructed by deleting a 1419-bp region within the replicase coding sequence (Fig. 4A). It was previously reported that TRBO-based recombinant protein expression was detected after just 3 dpi and would continue for several days thereafter (Lindbo, 2007). Accordingly, 16c plants were co-infiltrated with pHcoCas9 and either TRBO-gGFP or RM-gGFP. Genomic DNA was isolated from these treatments daily over the course of 10 d to have a better understanding of quantity and timing of indel development. BsgI restriction enzyme resistance assays were used to both verify and quantify the percentage of indels in infiltrated tissue (Supplemental Fig. S4). As expected, the negative control pHcoCas9/RM-gGFP samples showed an indel detection of less than 3%, which is low enough to be accounted for as background from nonindel-containing DNA that was undigested or due to the 35S promoter driving the expression of the replication-negative construct. Cas9/TRBO-gGFP showed no measurable indel buildup until 3 dpi, where percentages increased from less than 2% to 48%, and then to ∼70% between 6 and 7 dpi (Fig. 4B). Thus, the time course analysis confirms that the pHcoCas9/TRBO-sgRNA transient knockout screening method is virus replication dependent, and TRBO subgenomic RNAs provide high levels of sgRNA expression for efficient and rapid gene editing.

Figure 4.

TRBO-gGFP delivery indel time course and analysis of GFP protein expression in N. benthamiana 16c plants. A, TRBO-based constructs used in time course and imaging study. TRBO-gGFP is replication competent and carries gGFP. RM-gGFP is a TRBO-gGFP mutant that has a 1,419-bp deletion in the replicase protein coding sequence that does not support vector replication or transcription of subgenomic RNA. B, Indel percentages over 10 d calculated using restriction resistance assay for TRBO-gGFP and a TRBO replicase mutant carrying gGFP. C and D are 16c N. benthamiana leaf images using a dissecting microscope with GFP filter. Ten days postinfiltration leaf samples were treated with under the following conditions: C, RM-gGFP coinfiltrated with pHcoCas9, and D, TRBO-gGFP coinfiltrated with pHcoCas9.

To visualize how high levels of indel percentages result in a loss-of-function phenotype, at 10 dpi leaves from both RM-gGFP and TRBO-gGFP infiltrations were viewed using a stereomicroscope with a GFP filter (Olympus SZX7). Although GFP expression was present, a clear knockdown of GFP fluorescence was observed in pHcoCas9/TRBO-gGFP coinfiltrated tissue compared to pHcoCas9/RM-gGFP plants (Fig. 4, C and D). It should be noted that in absence of Cas9 delivery, GFP expression was still present (Fig. 3), indicating the reliance for both a replication-competent TRBO along with codelivery of Cas9 protein, as expected. Therefore, the clear difference in phenotype based upon a strong constitutively expressed plant chromosomal localized gene demonstrates the potential value of Cas9/TRBO-sgRNA as a transient knockout screening tool.

TRBO-sgRNA Targeting of Native Genes

Following our success with TRBO-gGFP to induce high percentages of indels in the mgfp5 transgene, we next tested the ability to target endogenous genes within N. benthamiana. ARGONAUTE1 (NbAGO1) was selected for gene targeting because of its central role in endogenous RNA silencing in N. benthamiana, by presumably serving as the catalytic portion of the RNA-induced silencing complex. VIGS-based knockdown screens show severe systemic phenotypic responses when NbAGO1 paralogs are knocked down (Jones et al., 2006). The two NbAGO1 paralogs within N. benthamiana share 97.61% transcript identity and have been termed NbAGO1-H (KR942296) and NbAGO1-L (KR942297) (Jones et al., 2006; Gursinsky et al., 2015). While VIGS assays allow for systemic responses, here we aimed to understand if localized gene editing of a native gene could be observed through NbAGO1 targeting in mature leaf tissue. To test this with the Cas9/TRBO-sgRNA coinfiltration system, we designed an sgRNA spacer using a complementary region of both NbAGO1-H and NbAGO1-L within the protein coding region for the PAZ domain (Supplemental Fig. S5). The sgRNA target for NbAGO1 (gAGO1) was cloned into TRBO expression vector, using the same method as described for the TRBO-gGFP, to yield TRBO-gAGO1.

Using genomic DNA sequences, specific PCR amplification of NbAGO1-H and NbAGO1-L were designed by aligning primers within allele specific intron regions that vary significantly between the paralogs. Half-leaf assays consisting of pHcoCas9/TRBO-gAGO1 and TRBO-gAGO1 only were infiltrated into N. benthamiana 16c plants. NbAGO1-H and NbAGO1-L alleles were amplified from genomic DNA at 7 dpi, and amplicons were subject to the surveyor nuclease assay, where digested products contain indels. Both NbAGO1-H and NbAGO1-L showed significant indel percentages in both paralogs (Fig. 5B), albeit much lower percentages than observed when previously targeting mgfp5 (Fig. 1D). An observable phenotype was yet to be associated with the mutated NbAGO1 alleles at 7 dpi, suggesting that an NbAGO1 knockout phenotype might take longer to establish or might not develop in mature leaves. Nevertheless, our virus-mediated system successfully achieved editing of a native gene in mature leaves.

Figure 5.

Indel analysis for TRBO delivery of single and multiple sgRNAs. A, TRBO-gGFP-gAGO1 co-delivery construct used for simultaneous delivery of both gGFP and gAGO1. B and C, Treatments included pHcoCas9 coinfiltrated with one of the following: TRBO-gGFP (gGFP), TRBO-gGFP-gAGO1 (gGFP-gAGO1), and TRBO-gAGO1 (gAGO1). M indicates DNA marker. Blue arrows indicate the presence of non-indel-containing DNA, and red arrows are indel-containing DNA. B, NbAGO1-H and NbAGO1-L PCR amplicons treated with surveyor nuclease for indel detection. Marker (M) is same molecular weights as in Fig. 1D. C, The mgfp5 amplicons when treated with BsgI restriction enzyme to detect indels.

TRBO Can Efficiently Deliver Multiple sgRNAs on a Single Subgenomic RNA

After successfully editing endogenous genes and a transgene using the TRBO delivery system, we further explored the system’s utility through the delivery of multiple sgRNAs from one construct (multiplex). The ability to edit multiple genes at once, upon transient delivery, could greatly aid functional genetic studies and regenerating plants with multiple gene knockouts. Due to the ability of the Cas9/TRBO-sgRNA system in N. benthamiana 16c plants to tolerate sgRNAs with 5′ and 3′ nucleotide overhangs while still producing indels, we hypothesized that multiple sgRNAs could be co-delivered using a single TRBO construct. We predicted that both the gGFP and gAGO1 sgRNA constructs could be positioned “side-by-side” under control of the same CP subgenomic promoter, without a linker sequence separating the sgRNAs, and efficiently induce Cas9-mediated DSBs. Additionally, we also tested if the sgRNA orientation proximity to the 5′ or 3′ end of the subgenomic RNA effected the efficiency of indel production.

The TRBO-gGFP vector was used as a template for the addition of gAGO1 directly 3′ to the sgRNA scaffolding of gGFP, creating TRBO-gGFP-gAGO1 (Fig. 5A). Half-leaf assays were carried out with coinfiltrations of TRBO-gGFP-gAGO1 either with or without Cas9 in 16c N. benthamiana plants. Cas9/TRBO-gGFP and Cas9/TRBO-gAGO1 were used to compare delivery of a singular guide compared to the multiplexed gRNA delivery vector TRBO-gGFP-gAGO1. At 10 dpi, genomic DNA was sampled for indel analysis of mgfp5 and both NbAGO1 alleles. NbAGO1-H and NbAGO1-L amplicons were analyzed for indels using the surveyor nuclease assay (Fig. 5B), while mgfp5 was subjected to the BsgI restriction enzyme resistance assay (Fig. 5C). Remarkably, following indel quantification analysis there was not a decrease, but rather an increase in observed indel percentages for all three genes when sgRNAs are multiplexed within TRBO; from 15% to 27% for NbAGO1-L, from 8% to 11% for NbAGO1-H, and from 53% to 64% for mgfp5. These results indicate that at least two sgRNAs can be delivered using the TRBO-sgRNA delivery system with an increase in delivery efficiency. Due to these results, we concluded that sgRNA positioning does not affect indel percentages during sgRNA multiplexing, offering a substantial level of flexibility not previously known for sgRNA design.

TRBO-Based Protein Overexpression and sgRNA Delivery Using a Single TRBO Delivery Construct

The Cas9/TRBO-sgRNA transient knockout system appears to tolerate both 5′ and 3′ overhangs, which was exhibited initially by the efficiency of the TRBO-sgRNA delivery without ribozymes and then through the multiplex delivery of sgRNAs in planta through a singular TRBO construct. These data suggest the possibility to deliver sgRNAs along with larger RNA molecules, such as a protein-encoding region. Delivery of biologically active sgRNAs while retaining TRBO-based overexpression of proteins would be of great benefit for biotechnology purposes and as a fundamental tool for functional genomic studies. To examine if a protein open reading frame could be co-delivered with a sgRNA (ORF-sgRNA), gGFP was selected to be delivered along with the gfpc3 protein coding gene within the TRBO vector (TRBO-G; Lindbo, 2007). gGFP was designed to target only the 16c (host) chromosomal mgfp5 sequence and therefore should not interfere with TRBO gfpc3 protein expression due to self-targeting from Cas9/gGFP. Viral-based GFP overexpression, which is indicative of TRBO-G replication and subgenomic RNA transcript delivery, should induce several fold higher expression than 16c-based GFP expression, allowing for direct visualization of viral dependent GFP expression.

To test if sgRNA proximity to the protein-coding region would disrupt indel percentages, gGFP was introduced directly 5′ or 3′ proximal to the gfpc3 coding sequence, forming TRBO-G-5′gGFP and TRBO-G-3′gGFP, respectively (Fig. 6A). It is important to note that the gGFP spacer sequence carries a start codon (AUG) and a stop codon (UGA) in the gRNA scaffolding within the same reading frame. We predicted that TRBO-dependent GFP expression would not occur using the construct TRBO-G-5′gGFP due to the small ORF being upstream of the GFP start site. However, by delivering a gGFP 5′ proximal to gfpc3 sequence, we could determine if gGFP proximity to the protein coding sequence would affect indel rates in planta.

Figure 6.

Co-delivery of GFP protein and gGFP using TRBO in 16c N. benthamiana. A, Depiction of GFP protein and gGFP co-expression constructs expressed from the TMV CP subgenomic promoter. The 5′ and 3′ preference for gGFP delivery was analyzed by creating constructs that carried both a 5′ and 3′ GFP protein coding sequence proximal to gGFP. B, 16c N. benthamiana leaves coinfiltrated with pHcoCas9 and either TRBO-gGFP, TRBO-G-5′gGFP, or TRBO-G-3′gGFP followed by visualization of GFP protein under UV light at 3 dpi. TRBO-G-5′gGFP was not expected to show viral vector-based expression of GFP due to a start and stop codon within the gGFP sequence. C, BsgI restriction enzyme resistance assay at 6 dpi. Lane treatments are as follows: pHcoCas9 alone (C), pHcoCas9/TRBO-gGFP co-infiltration (gGFP), pHcoCas9/TRBO-G-5′gGFP co-infiltration (5′gGFP), and pHcoCas9/TRBO-G-3ʹgGFP co-infiltration (3′gGFP).

Agroinfiltrations of 16c plants with pHcoCas9 alone or codelivery of pHcoCas9 and TRBO-gGFP, TRBO-G-5′gGFP, or TRBO-G-3′gGFP were carried out. At 3 dpi, infiltrated leaves were viewed under UV light to visualize TRBO-dependent GFP expression (Fig. 6B). As we anticipated, neither pHcoCas9 alone nor the coinfiltrations of pHcoCas9 and TRBO-gGFP or TRBO-G-5′gGFP showed GFP expression at levels expected for viral delivery. However, TRBO-G-3′gGFP showed a dramatic increase of GFP expression at 3 dpi (Fig. 6B). Following protein expression visualization, genomic DNA was sampled at 6 dpi then subject to mgfp5 amplification followed by a BsgI restriction enzyme resistance assay for indel quantification analysis. Remarkably, coinfiltrations of pHcoCas9 and TRBO-G-5′gGFP as well as pHcoCas9 and TRBO-G-3′gGFP showed 5% and 11% higher levels of indel formation, respectively, in comparison with the pHcoCas9/TRBO-gGFP control (Fig. 6C). BsgI-resistant bands were subsequently cloned and sequenced to verify the presence of indels at the mgfp5 target locus. Our results suggest that the additional GFP-coding RNA does not hinder gGFP biological activity but might actually increase the occurrence of DSBs through an unknown mechanism. Notably, indel rates of mgfp5 were observed at an 11% increase for both the ORF-sgRNA, TRBO-G-3′gGFP (Fig. 6C), and multiplex delivery of sgRNAs construct, TRBO-gGFP-gAGO1 (Fig. 5C), compared to the TRBO-gGFP control. These results confirm those above that sgRNA positioning has limited consequence for its activity. Moreover, the developed virus-mediated system offers the novel ability to overexpress and inactivate genes simultaneously.

DISCUSSION

Several biological systems have benefited from the development of the CRISPR/Cas9 for targeted transient knockdown of gene expression levels, such as bacteria, yeast, and human cell lines, but a similar localized transient system has yet to be routinely implemented for plant systems (Larson et al., 2013; Qi et al., 2013; Gilbert et al., 2013). Previously developed viral gene editing delivery methods focused on the creation of gene knockout plant lines (Baltes et al., 2014; Ali et al., 2015). Here, we use CRISPR/Cas9 as a method to locally knock out genes in plants and demonstrate its potential use as a transient screening method for virus-plant interactions. First, we explored the potential of a TMV-based delivery vector, TRBO, for its ability to deliver biologically functional sgRNAs through the CP subgenomic promoter in planta. Through the design and testing of multiple sgRNA delivery systems in N. benthamiana, we were surprised to find that 5′ and 3′ sgRNA overhangs did not impede Cas9/sgRNA-mediated DNA cleavage, while sgRNA overhangs negated Cas9 nuclease activity in vitro. In fact, the HH ribozyme in the TRBO-RGR construct appears to reduce the fitness of replicons while increasing the accumulation of mutations within the HH sequence in planta.

The Cas9/TRBO-sgRNA system transiently delivered and stably sustained indel rates at 60% to 70% within 7 dpi, creating knockout levels potentially sufficient for transient functional genetic screens. Disruption of the coding regions in the gene paralogs NbAGO1-H and NbAGO1-L demonstrated the ability to transiently target complementary regions among genes. Remarkably, the delivery of multiple biologically active sgRNAs within the same TRBO construct resulted in an increase in indel induction. Due to the successful delivery of two fused sgRNAs in planta, we anticipated that a larger RNA transcript caring both an sgRNA (gGFP) and a protein coding sequence (gfpc3) could be delivered using TRBO-G-3′gGFP. Delivery of Cas9/TRBO-G-3′gGFP successfully expressed GFP protein and increased the percentage of indels compared to the TRBO-gGFP control. One potential explanation for the increase in indels in the ORF-sgRNA codelivery method might be due to the longer TRBO 3′ end of TRBO-G-3′gGFP that aided in TMV replicase fidelity and extension from the TRBO CP subgenomic promoter as compared to TRBO-gGFP. For instance, increased distance between the upstream RNA promoter and the 3′ tRNA-like structure has been shown to stimulate transcription (Culver et al., 1993).

As mentioned previously, the current Cas9/sgRNA-based expression systems are limited by low expression of sgRNAs in vivo (Li et al., 2013; Nekrasov et al., 2013). Using our Cas9/TRBO-sgRNA delivery approach, we aimed at saturating the system with sgRNAs to allow for more efficient editing. We believe that the current limitations of the Cas9/TRBO-sgRNA delivery system, explained here, may now be the availability of Cas9 protein. We suggest that increasing Cas9 expression, such as using transgenic Cas9 plants (Ali et al., 2015), could increase the efficiency of the system. However, while transgenic Cas9 expression might improve the accumulation of gene-specific indels, it removes the benefit of a transient delivery system. Even though the Cas9/TRBO-sgRNA platform is still reliant on T-DNA that could be integrated into the genome, this may represent a key first step toward separating gene-editing tools from transformation technology. Instead of a transgenic-based system, viral vectors could co-deliver Cas9 and sgRNA to host plants. While such a method might be appealing, there are a limited number of viral vectors that will tolerate an insertion the size of the S. pyogenes Cas9 ORF (∼4.4 kb). Additional complications can arise through the delivery of multiple viral vectors, such as spatial expression differences within tissues and the requirement for shared susceptible hosts among the vectors used.

The TRBO-sgRNA transient gene-editing platform varies from other viral-induced gene editing platforms such as geminivirus-based delivery systems (Baltes et al., 2014; Yin et al., 2015), and the Tobacco rattle virus (TRV) sgRNA delivery system (Ali et al., 2015). Compared to those initial reports, we aimed to extend the concept by primarily focusing on exploiting the high titer of TMV for the purpose of solving the relatively low yield dilemma of sgRNAs from current U6 promoter-based delivery systems. While we did not directly demonstrate that the Cas9/TRBO-sgRNA system increases indel percentages compared to constitutive expression of sgRNAs, the time course analysis of TRBO-sgRNA in comparison with the replication-negative constructs did address this issue in some capacity considering the vector was still driven by the constitutive 35S promoter. Additionally, we aimed to investigate the influence of 5′ and 3′ guide RNA overhangs through the utilization of ribozyme delivery systems (Li et al., 2013; Nekrasov et al., 2013; Xie and Yang, 2013). The TRBO vector was of particular interest due to high expression from the 3′ coterminal CP subgenomic RNA while also restricting sgRNA expression to only infiltrated leaves. This approach differs from the TRV and geminivirus sgRNA delivery tools, where the primary goal was to produce seeds with gene knockouts through virus systemic movement, not localized knockouts with the capacity to be used as a transient screening tool (Ali et al., 2015; Yin et al., 2015). Additionally, the geminivirus-based DNA replicon was used to deliver Cas9, sgRNAs, and a DNA donor template to create a desired DNA sequence modification (Baltes et al., 2014). The addition of such a large insertion may have possibly reduced the efficiency of genomic DSB events and subsequent genomic recombination events as well as localized movement of the replicon. While these prior pioneering methods created knockout and plant lines with sequence specific changes, our strategy aimed to provide localized transient screening while avoiding systemic movement of the viral vector. It is also worth noting that the systemic movement of the aforementioned viruses relies on CP expression, thus allowing the recombinant virus to be encapsidated and perhaps more easily disseminated (Ali et al., 2015; Yin et al., 2015). The potential unintended biological consequences of creating functional virions should be a consideration when using these technologies. The RNA-based TRBO-sgRNA delivery system may alleviate many of these concerns.

The delivery of multiple sgRNAs as well as the simultaneous delivery of a protein coding sequence and an sgRNA fusion transcript using a single TRBO vector was effective even with the predicted large sgRNA nucleotide overhangs from the subgenomic delivered sgRNA. While the specificity of the complementary region within the 5′ end of the spacer sequence has been studied thoroughly (Dahlman et al., 2015; Fu et al., 2014), it has not been documented, to our knowledge, that significant (>1 nucleotide) 5′ spacer proximal overhangs are tolerated without affecting the induction of DSBs in vivo. The effects of sgRNA 3′ overhangs have been passively examined previously with a mutant Cas9 nuclease with deactivated nuclease domains (dCas9; Konermann et al., 2015; Zalatan et al., 2015). These studies have led to some indication that binding of Cas9 might be hindered by the presence of 3′ overhangs due to the incorporation of RNA binding protein RNA motifs at the 3′ end of the sgRNA scaffold RNA (Zalatan et al., 2015), but this is far from being conclusive.

Complications resulting from sgRNAs harboring 5′ and 3′ overhangs has been an area of concern for researchers since the inception of CRISPR/Cas9 as a gene-editing technology (Cong et al., 2013; Jinek et al., 2012; Mali et al., 2013). This is an issue we expected but did not encounter when developing the TRBO-sgRNA delivery system in planta. One of three activities could explain the tolerance of 5′ and 3′ sgRNA nucleotide overhangs observed in the Cas9/TRBO-sgRNA delivery system, none of which have been previously explored in detail: (1) Cas9 tolerates large noncomplementary overhangs 5′ proximal to the spacer and 3′ to the scaffold RNA in planta, (2) Cas9 has the ability to cleave sgRNA overhangs in planta, or (3) endogenous proteins in N. benthamiana cleave 5′ and 3′ sgRNA nucleotide overhangs in planta. Extensive in vitro and in vivo studies have been conducted on the activity of Cas9 that have demonstrated the importance of the sgRNA spacer sequence on the nuclease activity of Cas9 (Sternberg et al., 2014; Dahlman et al., 2015; Mali et al., 2013; Fu et al., 2014; Jinek et al., 2012). Largely, these studies have assumed that 5′ overhangs reduce the incidence of DSBs created by Cas9, with the exception of the addition of guanine for increased T7 RNA polymerase transcription and RNA polymerase III promoter expression, which does not appear to affect Cas9 catalytic activity (Hsu et al., 2014; Nekrasov et al., 2013; Xie and Yang, 2013). Based on these assumptions, a reduction or nullification in Cas9 DNA cleavage associated with sgRNA 5′ overhangs was expected to occur in planta for the gGFP and gHDV constructs. However, here we demonstrated that sgRNA overhangs within the delivery construct do not seem to effect catalytic activity in planta, while as expected they do abolish Cas9/sgRNA-mediated cleavage in vitro. These observations taken together would indicate that the high indel percentages observed using the TRBO-gGFP and TRBO-gHDV delivery systems suggest that the removal of 5′ overhangs must be occurring in planta. Although sgRNA delivery systems have been adapted to remove RNA overhangs within the delivery system itself (Gao and Zhao, 2014) or to rely on host RNases for programmed nucleotide overhang cleavage (Xie et al., 2015), removal of sgRNA overhangs through existing in vivo processes has not been reported, although it has been suggested (Cong et al., 2013; Karvelis et al., 2013).

The most popular method currently used to assess genetic factors and their effect on viral infection has been the VIGS screening method, where the host is typically infected with TRV before being subjected to the pathogen of interest (Burch-Smith et al., 2004; Scholthof et al., 2011). Subjecting a plant to a bacterium and a virus to study the host effects of another virus elicits concern for the quality of the screen. Results from traditional VIGS screening are often probed for verification through transgenic knockdown techniques such as dsRNA hairpin technology (Odokonyero et al., 2015), T-DNA knockout insertion lines, or more recently CRISPR/Cas9 knockouts (Brooks et al., 2014). The adoption of our viral-delivered sgRNAs technique could shorten the time of screening, increase the reliability of the transient screen, and decrease the space and plants needed to conduct screening. Further genetic studies, which are not necessarily accessible when using VIGS screening, such as targeting promoters, enhancers, and insulator sequences, is now feasible for understanding functionality within the context of a native gene environment (Zhang et al., 2016; Flavahan et al., 2016; Basak and Nithin, 2015). Additionally, the adaptation of sgRNA delivery to include different viruses, viral strains, or genetic variants has the potential to help understand host symptom development and viral disease onset (Mandadi and Scholthof, 2013).

Although a TRBO-sgRNA screening method is promising, there are certainly some constraints when it comes to CRISPR-based gene targeting, particularly when considering large or polyploid plant genomes such as N. benthamiana. Additionally, transient CRISPR screens create a chimera of cell types including heterozygous, wide-type, and true knockout cells, which can greatly reduce the quality and effectiveness of the screening method (Morgens et al., 2016). These complications could even be a larger problem when targeting genes that are functionally redundant. In these instances, the use of both VIGS- and CRISPR-based screening methods in parallel could increase the reliability of experimental results and provide findings within one screen that is overlooked through the others (Morgens et al., 2016; Barrangou et al., 2015, Shalem et al., 2015).

In summary, this study provides new insight into the previously unexplored flexibility of delivering one or more sgRNAs launched from the TRBO vector without the necessity of designing processing elements. Furthermore, it is shown that the same vector can be used for the simultaneous delivery of an sgRNA for gene editing along with an ORF for overexpressing a protein of interest. These findings illustrate new properties associated with virus-mediated delivered CRISPR/Cas9 gene editing.

MATERIALS AND METHODS

Design of sgRNAs and Construct Development

The sgRNA targets were identified using the CRISPR design toolset (benchling.com). Gene sequences were used from previously reported and deposited sequences for mgfp5 (Haseloff et al., 1997), NbAGO-L, and NbAGO-H (Gursinsky et al., 2015). The NbAGO-L and NbAGO-H sgRNA (gAGO1) was designed through gene sequence alignment and identification of complementary areas for potential co-targeting (benchling.com). Sequences for each of the sgRNA sequences can be found in Supplemental Fig. S6.

RGR was designed as described previously (Gao and Zhao, 2014), using an mgfp5-targeting sgRNA as mentioned above. The RGR construct was synthesized using custom gene synthesis (GenScript) with the inclusion of a 5′ PacI and a 3′ NotI site for cloning into TRBO, as described by Lindbo (2007). Toward this, the RGR construct was used as a template to create both the HDV and gGFP construct using PCR fragment cloning. Forward primers were designed carrying an additional 5′ PacI site and complementary regions overlapping the mgfp5 spacer sequence of the sgRNA. Reverse primers were designed for both the gHDV and gGFP constructs using the 3′ NotI site with complementary regions to the 3′ HDV and scaffold portion of the sgRNA, respectively. The high-fidelity Q5 polymerase (New England Biolabs) was used to produce PCR amplicons, corresponding to each construct, and cloned into pGEM-T Easy (Promega). The cloned fragments and the TRBO vector were then subjected to a PacI and NotI double digest and a subsequent ligation step to produce TRBO-RGR, TRBO-gHDV, and TRBO-gGFP.

The gAGO1 target was constructed using TRBO-gGFP as a template for site-directed mutagenesis using the Q5 Site-Directed Mutagenesis Kit (New England Biolabs). Primers for site-directed mutagenesis were designed using the online NEBaseChanger (New England Biolabs) tool, and the online generated protocol was used to carry out the mutagenesis reaction. TRBO-gGFP-gAGO1 was constructed using DNA assembly, where TRBO-gGFP was used as the destination vector. Using TRBO-gGFP overlapping primers, gAGO1 was amplified from TRBO-gAGO1 and inserted 3′ to the gGFP scaffold of NotI linearized TRBO-gGFP using NEBuilder HiFi DNA Assembly master mix (New England Biolabs).

RM was constructed using the TRBO-gGFP plasmid. TRBO-gGFP was digested with SmaI and StuI restriction enzymes to excise 1419-bp region in the replicase coding region. This DNA fragment cleaned using the Zymoclean Gel DNA Recovery kit (Zymo Research) and used for a blunt-end ligation using T4 ligase (New England Biolabs) to create the final replication incompetent (RM) construct.

TRBO-G was used as destination cloning backbone for both the TRBO-G-5′gGFP and TRBO-G-3′gGFP constructs. TRBO-G-5′gGFP was constructed by linearization of TRBO-G with PacI. The gGFP construct was amplified from TRBO-gGFP using forward and reverse primers that overlapped with the PacI linearized TRBO-G plasmid. The forward primer consisted of a spacer sequence (TAA) between the TRBO-G overlapping sequence and gGFP overlapping sequence to maintain the PacI site within the final TRBO-G-5′gGFP construct. The TRBO-G digestion and gGFP amplification were visualized using gel electrophoresis followed by excision of bands at the expected M_r and cleaned using the Zymoclean Gel DNA Recovery kit (Zymo Research). The final TRBO-G-5′gGFP construct was made using HiFi DNA Assembly master mix (New England Biolabs) with linearized TRBO-G and gGFP fragments following the manufacture’s recommended protocol. TRBO-G-3′gGFP was constructed essentially as TRBO-G-5′gGFP, but TRBO-G was instead linearized using the NotI site 3′ to the GFP stop codon. Additionally, the reverse primer for gGFP amplification consisted of a spacer sequence (GGCCGC) between the TRBO-G overlapping region and the gGFP complementary region to repair the NotI site within the final TRBO-G-3′gGFP construct.

The Cas9 expression construct was assembled using the human codon-optimized Cas9 nuclease (HcoCas9; Addgene 42230; Cong et al., 2013). HcoCas9 was PCR amplified using a forward primer with a BamHI site extension and a reverse primer carrying an extension with an XhoI site. A modified pRTL22 (Restrepo et al., 1990) carrying the 3′ TEV sequence upstream of the CaMV 35S terminator sequence was used as the subcloning vector. The 35S-Cas9-term cassette was then transferred into the binary destination plasmid pBINPLUS-sel through the HindIII site.

Agroinfiltrations

Agrobacterium tumefaciens strain GV3101 (pMP90RK) was used for agroinfiltration of all binary vectors used, as described previously (Odokonyero et al., 2015). In brief, cultures were grown overnight (16–20 h) under constant shaking (250 rpm) at 28°C in LB media with 50 mg/L kanamycin. Cells were pelleted by centrifugation and resuspended in infiltration buffer (10 mm MgCl₂, 10 mm MES, pH 5.7, and 200 µm acetosyringone). TRBO-based cultures were resuspended to a final infiltration concentration of OD₆₀₀ 0.4 and pBINPLUS-sel Cas9 expression vector at OD₆₀₀ 0.5. Coinfiltrations were carried out by mixing equal volumes of resuspended TRBO and Cas9 Nicotiana benthamiana cultures to the above-mentioned final concentrations. Four-week-old 16c plants were infiltrated with A. tumefaciens suspensions on the abaxial side of the leaf and returned to normal growth conditions.

DNA and Indel Assays

Single-plant DNA samples for indel assays were carried using 50 mg of leaf tissue from three infiltrated leaves, totaling 150 mg of tissue, to avoid tissue-dependent effects as well as to create a pooled biological replicate. DNA extractions were then carried out using the ZR Plant/Seed DNA Miniprep kit (Zymo Research). For the BsgI restriction enzyme resistance assay and the Surveyor Nuclease (Integrated DNA Technologies), 100 ng of genomic DNA was used for PCR amplification of either mgfp5, NbAGO1-H, or NbAGO1-L. Amplicons were then cleaned using DNA Clean and Concentrator-5 (Zymo Research) kit and resuspended in DNase and RNase-free water. Amplicon concentrations were measured using a Nanodrop, and 200 to 400 ng DNA used for final Surveyor Nuclease and 400 to 500 ng DNA for BsgI digestions. Amplicons of genomic mgfp5 were subject to BsgI restriction enzyme and incubated at 37°C overnight. NbAGO1-H and NbAGO1-L amplicons were treated with Surveyor Nuclease, a DNA mismatch endonuclease. Surveyor Nuclease digestions were carried out using the manufacturer’s instructions (IDT). Both the BsgI restriction enzyme resistance assay and the Surveyor Nuclease reactions were visualized using 2% agarose gel electrophoresis stained with ethidium bromide. Image files (.tif) were uploaded in the image analysis software ImageJ (NIH). The background signal was subtracted from gel images, and band intensities were measured using standard gel peak analysis workflow.

RNA Extractions and RT-PCR

Plant RNA extractions were carried out using the Direct-zol RNA Miniprep kit (Zymo Research) following the manufacturer’s instructions. cDNA was synthesized with equal volumes of total RNA using the SuperScript III first-strand synthesis kit (Invitrogen) along with a TMV 3′-UTR-specific reverse primer, RT-PCR was carried out using the TMV 3′-UTR-specific reverse primer and a TMV-specific forward primer 5′ to the TRBO PacI restriction cite using Taq polymerase (New England Biolabs). RT-PCR bands were gel extracted using Zymoclean Gel DNA Recovery kit (Zymo Research) and cloned into the pGEM-T Easy vector system (Promega) for DNA sequencing.

GFP Imaging

N. benthamiana 16c plants or leaves of TRBO-G infiltrated plants were visualized under a handheld UV mercury lamp as previously described (Odokonyero et al., 2015, Everett et al., 2010). TRBO-gGFP- and RM-gGFP-treated leaf images were visualized using an Olympus SZX7 stereomicroscope with a GFP filter essentially as described previously (Gao et al., 2013).

Protein Extractions and Western Blotting

For Cas9 protein detection, extractions were carried out using 50 mg of 3 dpi 16c leaf tissue infiltrated with both pFGC-pcoCas9 (Addgene plasmid 52256) and pHcoCas9. Tissue was ground in liquid nitrogen followed by suspension in 250 µL of 1:1 1× TE buffer and 5× Laemmli SDS buffer. Samples were boiled for 5 min before centrifugation for 2 min at 10,000g at room temperature. The cleared lysate (supernatant) was kept for SDS-PAGE, where 20 µL of each treatment was electrophoresed through a 4% to 20% gradient ExpressPlus PAGE gel (GenScript) in 1× NuPAGE MOPS SDS running buffer (ThermoFisher Scientific) at 100 V until adequate protein ladder separation occurred (∼3 h). Proteins were then transferred onto a nitrocellulose membrane in Tris-Gly transfer buffer (25 mm Tris and 192 mm Gly) at 100 V for 60 min. The membrane was then blocked in 5% nonfat milk in TBS-Tween with gentle agitation for 30 min at room temperature followed by an overnight incubation in DYKDDDDK tag mouse mAb (GenScript) at a dilution of 1:2,000 at 4°C. Membranes were washed with TBS-Tween three times for 5 min each wash. Secondary goat anti-mouse IgG conjugated with horseradish peroxidase (Sigma-Aldrich) was incubated for 1 h at a dilution of 1:2,000. Membranes were then washed briefly with TBS-Tween before three final wash steps in TBS for 5 min each. Colorimetric detection of Cas9 protein was achieved by dissolving 30 mg of 4-chloro-1-naphthol in 10 mL of cold methanol and adding 30 µL of 30% hydrogen peroxide to 50 mL of 1× TBS. The two substrates were mixed and added to the nictrocellulose membrane. This was followed by a dark incubation period of at least 15 min with gentle agitation for signal development. Reactions were stopped by washing the membrane with distilled water, blotted dry, and then photographed.

GFP expression was measured essentially as stated above. However, protein samples after extraction were diluted 1:1 with equal quantities of 1× TE buffer and 5× Laemmli SDS buffer for increased downstream detection sensitivity. Following dilution, 20 µL of each sample was loaded and electrophoresed. GFP signal was detected as described previously except rabbit pAb to GFP (GenScript) at a dilution of 1:1,000 was used as the primary antibody followed by using an IgG goat anti-rabbit horseradish peroxidase conjugate (Sigma-Aldrich) at a dilution of 1:2,000. Protein polyacrylamide gel staining was carried out by adding 50 mL of QC Colloidal Coomassie stain (BioRadiation) following the manufacturer’s protocol.

Ribozyme Efficiency Assay

In vitro ribozyme efficiency activity assays for the RGR, gHDV, and gGFP constructs were carried out using the final TRBO destination cloning vector as the template for PCR extension encompassing the cloning sites. A forward primer including the T7 promoter sequence was designed in the TMV movement protein ORF followed by the TMV 3′ UTR reverse primer. In planta ribozyme activity for the corresponding TRBO-sgRNA constructs was assessed by cDNA synthesis followed by an RT-PCR step using the T7-TMV complementary forward primer and TMV 3′ UTR reverse primer. PCR amplicons were then used as a template for T7 in vitro transcription reactions (New England Biolabs). Following transcript synthesis, RNA was visualized using 2% agarose gel electrophoresis stained with ethidium bromide.

Cas9/sgRNA in Vitro Cleavage Assays

Subgenomic sgRNAs and gGFP RNAs were synthesized using T7 RNA synthesis methods as described above. In brief, a forward primer starting at the CP subgenomic promoter transcriptional start site of TRBO caring a 5′ T7 promoter was used to PCR amplify each of the TRBO-gGFP, TRBO-gHDV, and TRBO-RGR constructs along with a reverse primer within the 3′ UTR. One hundred fifty nanograms of each of the PCR amplicons (sub-gGFP, sub-HDV, and sub-RGR, respectively) were then used as a template for T7 in vitro transcription reactions (New England Biolabs). RNA synthesis reactions were verified using 1% agarose gel electrophoresis stained with ethidium bromide and quantified using a Nanodrop.

A PCR mgfp5 fragment was amplified from untreated 16c genomic DNA and TA cloned into pGEM-T Easy (Promega). TA-mgfp5 was used as the DNA template for each of the cleavage reactions. One hundred nanomolar of purified Cas9 Nuclease (New England Biolabs) was first incubated in Cas9 Nuclease reaction buffer (New England Biolabs) and 100 nm with either gGFP, subgGFP, subgHDV, or subRGR RNAs at room temperature for 5 min. One hundred nanograms of purified TA-mgfp5 plasmid and 1 µL of ScaI restriction enzyme (New England Biolabs) was then added to each reaction (20 µL total volume) and incubated for 30 min at 37°C. Reactions were visualized using 1% agarose gel electrophoresis stained with ethidium bromide.

Supplemental Data

The following supplemental materials are available.

• Supplemental Figure S1. Predicted subgenomic RNAs used for in vitro Cas9 nuclease activity assays.

• Supplemental Figure S2. Cas9/TRBO-gGFP indel conformation.

• Supplemental Figure S3. TRBO-sgRNA delivery in planta.

• Supplemental Figure S4. Indel percentage measured over 10 d.

• Supplemental Figure S5. NbAGO1 paralogs NbAGO1-H (KR942296) and NbAGO1-L (KR942297) sequence alignment.

• Supplemental Figure S6. Sequences used for each guide RNA delivery tool used in this study.

• Supplemental Table S1. Primers used in this study.

Glossary

VIGS

viral-induced gene silencing

DSB

double-stranded break

NLS

nuclear localization signal