Drosophila expression plasmids, transfections and RNA isolation

To generate Drosophila plasmids that co-express a Cas13 effector along with its associated guide RNA, site directed mutagenesis was first used to remove the existing BsmBI site from the previously described pUb 3xFLAG MCS plasmid backbone (37). A control guide RNA sequence containing BsmBI sites was then inserted downstream of the Drosophila snRNA:U6:96Ab promoter using the NdeI restriction site. HA-tagged versions of RxCas13d (8), PspCas13b (7), or LwaCas13a (6) (or their associated catalytic dead versions) were then cloned downstream of the constitutive Drosophila Ubi-p63e promoter. Lastly, each guide RNA of interest was cloned into the BsmBI sites. Full details of the cloning, including full plasmid sequences, are described in detail in the Supplementary Material. The sequences of all guide RNAs used in Drosophila are listed in Supplementary Table S1.

Drosophila reporter plasmids expressing eGFP (Hy_pMT eGFP SV40; Addgene #69911), Laccase2 Exons 1–3 (Hy_pMT Laccase2 Exons 1–3; Addgene #91799), and nLuc (Hy_pMtnA nLuc SV40; Addgene #132654) under the control of the inducible Metallothionein A promoter were described previously (38–40). Reporter plasmids expressing mCherry (Hy_pMtnA mCherry SV40; Addgene #176302) and FFLuc (Hy_pMtnA FFLuc SV40; Addgene #176299) under the control of the Metallothionein A promoter were generated by replacing the eGFP ORF of Hy_pMT eGFP SV40 with the indicated ORF. Full details of the cloning, including full plasmid sequences, are described in detail in the Supplementary Material. Drosophila reporter plasmids expressing eGFP (Hy_pUbi-p63e eGFP SV40; Addgene #132650) under the control of the Ubi-p63e promoter was described previously (40). The associated mCherry plasmid (Hy_pUbi-p63e mCherry SV40; Addgene #176300) was generated by replacing the eGFP ORF with mCherry.

DL1 cells were seeded in 12-well plates (5 × 10⁵ cells per well) in complete Schneider's Drosophila medium and cultured overnight. On the following day, 500 ng of plasmid DNA was transfected (using a master mix) into each well using Effectene (4 μl of enhancer and 5 μl of Effectene reagent; Qiagen 301427). Unless otherwise noted, 50 ng of Cas13 expression plasmid was transfected along with 225 ng of each of the on-/off-target expression plasmids. pUb-3xFLAG-MCS (No BsmBI) was used as the empty vector plasmid. Total RNA was isolated ~40 h later using TRIzol (Thermo Fisher Scientific 15596018) according to the manufacturer's instructions. When examining reporters driven by the Metallothionein A promoter, a final concentration of 500 μM copper sulfate (Fisher BioReagents BP346-500) was added to cells for the last 14 h prior to RNA isolation.

S2 cells were seeded in six-well plates (4 × 10⁶ cells per well) in complete Schneider's Drosophila medium and cultured overnight. On the following day, 2000 ng of plasmid DNA was transfected into each well using Effectene (16 μl of enhancer and 30 μl of Effectene reagent; Qiagen 301427). As above, total RNA was isolated ~40 h later using TRIzol (Thermo Fisher Scientific 15596018) according to the manufacturer's instructions.

Human expression plasmids, transfections and RNA isolation

The RxCas13d-2A-eGFP expression plasmid (pXR001; Addgene #109049) and the corresponding guide RNA expression vector (pXR003; Addgene #109053) were described previously (8). The PspCas13b-2A-eGFP expression vector was a gift from Eric Wang (University of Florida) and made by replacing the RxCas13d ORF in pXR001 with PspCas13b. The corresponding PspCas13b guide RNA expression vector was described previously (pC0043; Addgene #103854) (7). All guide RNAs were cloned into the BbsI sites and their sequences are listed in Supplementary Table S2.

pcDNA3.1(+)/AUG-nLuc-3XFLAG reporter (Addgene #127299) was previously described (41) and the FFLuc expression plasmid (pGL4.13) was obtained from Promega. The pcDNA3.1(+) mCherry plasmid (Addgene #176301) was generated by replacing the eGFP ORF in pcDNA3.1(+) eGFP (Addgene #109020) with mCherry. Humanized Renilla luciferase reporter (p CIneo-RL; Addgene #115366) was described previously (42).

HeLa cells (3 × 10⁵ cells per well) or HEK293T cells (5.5 × 10⁵ cells per well) were seeded in 6-well plates in complete DMEM medium and cultured overnight. On the following day, 1000 ng of plasmid DNA was transfected (using a master mix) into each well using Lipofectamine 2000 (Thermo Fisher Scientific 11668019) according to the manufacturer's instructions. Unless otherwise noted, 300 ng of Cas13 expression plasmid and 200 ng of guide RNA expression plasmid were transfected along with 250 ng of each of the on-/off-target expression plasmids. pBEVY-L was used as the empty vector plasmid (43). Total RNA was isolated 40–48 h later, as indicated, using TRIzol (Thermo Fisher Scientific 15596018) according to the manufacturer's instructions.

Northern blotting

Northern blots using 1.2% denaturing formaldehyde agarose gels, NorthernMax reagents (Thermo Fisher Scientific), and oligonucleotide probes were performed as previously described (44). Oligonucleotide probe sequences are provided in Supplementary Table S3. Blots were hybridized with individual oligonucleotide probes, washed, and viewed with the Typhoon 9500 scanner (GE Healthcare) followed by quantification using ImageQuant (GE Healthcare). They were then subsequently re-probed with additional oligonucleotide probes so that all blots shown in a figure panel are derived from the same experiment. To confirm expression of the guide RNAs, 8% polyacrylamide gels rather than denaturing formaldehyde agarose gels were used.

Western blotting

Cells were lysed 40 h after transfection using RIPA buffer (150 mM NaCl, 1% Triton X-100, 50 mM Tris pH 7.5, 0.1% SDS, 0.5% sodium deoxycholate and protease inhibitors [Roche 11836170001]) for 20 min on ice. Lysates were then spun at 12 000 × g for 15 min and the supernatants were moved to new tubes. Samples were resolved on NuPAGE 4–12% Bis–Tris gels (Thermo Fisher Scientific NP0323) and transferred to PVDF membranes (Bio-Rad 1620177). Membranes were blocked with 5% nonfat milk for 1 h before incubation in primary antibody (diluted in 1× TBST) overnight at 4°C. Membranes were then washed with 1× TBST (3 × 5 min) followed by incubation in secondary antibody at room temperature for 1 h. The following antibodies were used: rabbit anti-HA (1:1000, Abcam ab9110), mouse anti-α-tubulin (1:10 000, Sigma T6074), rabbit anti-GAPDH (1:10 000, Proteintech 10494-1-AP), sheep anti-mouse IgG/HRP (1:10 000, Cytiva NA931), and donkey anti-rabbit IgG/HRP (1:10 000, Cytiva NA934). Membranes were processed using SuperSignal West Pico PLUS Chemiluminescent Substrate (Thermo Fisher Scientific 34580) according to the manufacturer's protocol.

RT-qPCR

9 μg of total RNA (quantified by Nanodrop) was digested with TURBO DNase (Thermo Fisher Scientific AM2238) in a 50 μl reaction following the manufacturer's protocol. 1 μg of the digested RNA was then reverse transcribed to cDNA in a 20 μl reaction using SuperScript III (Thermo Fisher Scientific 18080051) with random hexamers following the manufacturer's protocol. cDNA was diluted 10-fold with DEPC-treated water and RT-qPCR was performed using Power SYBR Green PCR Master Mix (Thermo Fisher Scientific 4368708). RT-qPCR reactions were performed in 15 μl reactions that contained 1.5 μl of diluted cDNA, 7.5 μl 2× Power SYBR Green PCR Master Mix, and 6 μl 1.5 μM gene-specific primer pairs. Primer sequences were tested by serial dilution and melt curve analysis, and their sequences are provided in Supplementary Table S4.

Using the LightCycler 96 Real-Time PCR System (Roche) and white plates (VWR 10755–256), the following cycling conditions were used: 95°C for 10 min, 40 amplification cycles of 95°C for 15 s followed by 60°C for 1 min, and a final melting cycle of 95°C for 10 s, 65°C for 1 min, and 97°C for 1 s. Subsequently, a melt curve was performed to verify that amplified products were a single discrete species. Threshold cycle (CT) values were automatically calculated by the LightCycler system and relative transcript levels (compared to Drosophila Act42a or human GAPDH) were calculated using the 2^−ΔΔCT method. RT-qPCR reactions were performed using three independent biological replicates, with each replicate having two technical replicates.

Once activated, the nuclease activity of RxCas13d can be inherently non-specific in Drosophila cells

We aimed to explore the underlying cause of the RxCas13d off-target effects and determine if cellular conditions could be identified in which bystander RNA degradation (trans-cleavage) is minimized. First, the co-transfection assays were repeated using catalytic dead RxCas13d (denoted dRxCas13d) that harbors quadruple mutations (R239A/H244A/R858A/H863A) in the HEPN (Higher Eukaryotic and Prokaryotic Nucleotide-binding) nuclease domains (Figure ​(Figure1A).1A). These mutations abolish the nuclease activity of RxCas13d but do not affect its RNA binding activity (8,21). No significant changes in eGFP or mCherry mRNA levels were observed when dRxCas13d was employed (Figure ​(Figure1C).1C). This indicates that the HEPN nuclease domains are critical for both the on- and off-target (cis- and trans-cleavage) effects of RxCas13d in cells.

We next examined whether the off-target effects are dependent on the presence of the target RNA (Supplementary Figure S3). Similar to the assay diagrammed in Figure ​Figure1A,1A, DL1 cells were co-transfected with a plasmid expressing RxCas13d/guide RNA complementary to eGFP and a plasmid expressing eGFP (target RNA) driven by the inducible MtnA promoter. In this case, the co-transfected mCherry (off-target RNA) reporter was driven by the Ubi-p63e promoter, thereby allowing mCherry to be constitutively expressed. Off-target depletion (trans-cleavage) of mCherry by RxCas13d was only observed when expression of the eGFP target mRNA had been induced (Supplementary Figure S3). Collectively, these data confirm that the off-target effects of RxCas13d in cells are due to the RxCas13d/guide RNA complex recognizing its target mRNA, which activates the HEPN nuclease activity (Figure ​(Figure2A2A).

Open in a separate window

Figure 2.

Reducing the expression level of RxCas13d does not diminish the off-target effects. (A) Proposed model of Cas13 effector activation. Base pairing between the guide RNA and its target mRNA (green) leads to a conformational change in Cas13 (orange) that activates the endonuclease activity on the surface of the protein. This can result in cis-cleavage/degradation of the target RNA (left) but also trans-cleavage of bystander RNAs (red), resulting in off-target effects (right). This model predicts that increased levels of target mRNA should result in increased off-target effects due to increased numbers of active Cas13 protein in cells. (B) Drosophila DL1 cells were transfected with decreasing amounts of RxCas13d/guide RNA expression plasmid, while the amounts of the eGFP (225 ng) and mCherry (225 ng) plasmids transfected were kept constant. Empty vector (pUb-3xFLAG MCS (No BsmBI) plasmid) was added as needed so that 500 ng DNA was transfected in all samples. 24 h after transfection, CuSO₄ was added and total RNA was isolated after an additional 14 h. (C) Northern blots (20 μg of total RNA/lane) were used to quantify the relative expression levels of eGFP and mCherry mRNA. Data are shown as mean ± SD, N = 3. For statistical comparisons, data were compared to the random guide RNA samples. () P < 0.05. n.s., not significant.

To test whether RxCas13d generally has significant bystander (trans-cleavage) effects in Drosophila cells, we systematically altered each of the components of the co-transfection assay. Neither extending the RxCas13d guide RNA spacer length from 24 to 30 nucleotides (Supplementary Figure S4A) nor changing the promoter (Supplementary Figure S4B) or the open reading frame (Supplementary Figure S4C) of the on-/off-target reporter mRNAs diminished the off-target (trans-cleavage) effects observed with RxCas13d. In all cases, the extent of off-target effects remained as strong as the level of on-target knockdown observed (Supplementary Figure S4A-C). Similar results were observed in Drosophila S2 cells (Supplementary Figure S4D). It thus appears that the nuclease activity of activated (target RNA bound) RxCas13d may be inherently non-specific in Drosophila cells (Figure ​(Figure2A2A).

Off-target effects of RxCas13d are positively correlated with the expression level of the target mRNA

Although the nuclease activity of activated RxCas13d may lack strong specificity, we reasoned that limiting the amounts of RxCas13d/guide RNA and/or the activating target mRNA in cells may maximize on-target (cis-cleavage) effects and diminish off-target RNA degradation (trans-cleavage events). To first test the effect of reducing RxCas13d/guide RNA levels, the amount of RxCas13d/eGFP guide RNA plasmid transfected was reduced by 10-fold (to 5 ng from 50 ng, which was the amount transfected in Figure ​Figure1B),1B), while keeping the amounts of eGFP (225 ng) and mCherry (225 ng) expression plasmids transfected constant (Figure ​(Figure2B).2B). Less efficient on-target eGFP mRNA depletion was observed under these conditions (Figure ​(Figure2C),2C), confirming that RxCas13d/eGFP guide RNA were at limiting levels. However, the degree of off-target degradation of mCherry mRNA still mirrored the efficiency of on-target knockdown (Figure ​(Figure2C).2C). We thus conclude that reducing the level of RxCas13d/guide RNA was insufficient to enable specific on-target knockdown in the co-transfection assay. This is perhaps because the target RNA is expressed at levels high enough that the RxCas13d/guide RNA complexes present in cells are constantly bound by target RNA, thereby activating the non-specific RxCas13d nuclease (trans-cleavage) activity.

To directly test the importance of target RNA levels, we decreased the expression of the eGFP target mRNA and examined on-/off-target effects (Figure ​(Figure3).3). Previously, in Figures ​Figures1B1B and 2C, 225 ng of target mRNA expression plasmid was used, so the amount of eGFP plasmid used here was progressively decreased to as low as 2 ng while keeping the amounts of RxCas13d/guide RNA (50 ng) and mCherry (off-target) expression plasmids (225 ng) constant (Figure ​(Figure3A).3A). RT-qPCR confirmed that reducing the amount of eGFP plasmid transfected resulted in the expected gradient of eGFP mRNA levels (Supplementary Figure S5A) with no effect on mCherry mRNA levels (Supplementary Figure S5B). Compared to a control random guide RNA, guide RNAs complementary to eGFP enabled RxCas13d to deplete eGFP mRNA levels by >75% regardless of the amount of eGFP plasmid transfected (Figure ​(Figure3B,3B, Supplementary Figure S5C). At the highest eGFP expression levels (50 ng eGFP plasmid transfected), the mCherry and RxCas13d mRNAs were also depleted by RxCas13d by ~75%, thereby mirroring the degree of on-target knockdown (Figure ​(Figure3C).3C). This result suggests that RxCas13d/guide RNA complexes are constantly bound by target RNA under these transfection conditions, which activates the non-specific (trans-cleavage) RxCas13d nuclease activity. In contrast, as the eGFP mRNA levels were progressively decreased, the amounts of off-target effects also decreased and ultimately went away. For example, when 2 ng of eGFP plasmid was transfected, eGFP guide-2 resulted in >75% depletion of eGFP mRNA (Figure ​(Figure3B,3B, Supplementary Figure S5C) but no significant change in mCherry or RxCas13d mRNA levels (Figure ​(Figure3C).3C). This suggests that when RxCas13d is used to target lowly expressed RNAs, off-target (trans-cleavage) effects may be minimal as the RxCas13d/guide RNA complex is rarely in an activated state (bound to the target RNA). Furthermore, this observation regarding target RNA abundance likely helps explain in part why some groups have suggested RxCas13d has good specificity in cells (8,33–36), while others (including us earlier in this manuscript) have come to the opposite conclusion (29,31).

Open in a separate window

Figure 3.

A positive correlation was observed between target mRNA level and RxCas13d off-target effects. (A) Drosophila DL1 cells were co-transfected with a constant amount of RxCas13d/guide RNA (50 ng) and mCherry (225 ng) expression plasmids, but variable amounts of eGFP expression plasmid (2, 5, 10, 25 or 50 ng). Empty vector (pUb-3xFLAG MCS (No BsmBI) plasmid) was added as needed so that 500 ng DNA was transfected in all samples. 24 h after transfection, CuSO₄ was added and total RNA was isolated after an additional 14 h. RNA expression levels were then analyzed by RT-qPCR (B) or northern blotting (C). (B) RT-qPCR was used to quantify the expression of eGFP mRNA in cells transfected with the RxCas13d plasmid expressing a random guide RNA or a guide RNA complementary to eGFP. For each amount of eGFP plasmid transfected, the relative abundance of eGFP mRNA was normalized to the respective random guide RNA samples. Data are shown as mean ± SD, N = 3. () P < 0.05. (C) Northern blots (20 μg of total RNA/lane) were used to quantify the relative expression levels of mCherry and RxCas13d mRNAs. Data are shown as mean ± SD, N = 3. For statistical comparisons, data were compared to the random guide RNA samples. () P < 0.05. n.s., not significant. A complete table of P-values for all comparisons is provided in Supplementary Table S5.

PspCas13b has significantly improved specificity in Drosophila cells compared to RxCas13d

The potential for significant off-target effects with RxCas13d severely limits its ability to catalyze specific depletion of RNAs of interest in Drosophila cells. We thus next examined other commonly used Cas13 effectors, especially those in other Cas13 subtypes (Supplementary Figure S1A-B). Cas13 effectors all possess nucleic acid recognition and RNA cleavage activities, but they are divided into 6 subtypes that have low sequence similarity beyond the active sites for the two HEPN domains (10). In fact, Cas13 subtypes vary greatly in size and domain organization, including the locations of the HEPN domains. We thus reasoned that the on-/off-target (cis- and trans-cleavage) effects of different Cas13 effectors could be highly divergent in Drosophila cells.

LwaCas13a (isolated from Leptotrichia wadei) was originally shown in 2017 to be able to elicit knockdown of target RNAs in Escherichia coli and human HEK293FT cells, with an efficiency similar to that of short hairpin RNAs (shRNAs) in human cells (6). We found that the LwaCas13a effector protein and its associated guide RNA (30-nt spacer length) are able to be expressed in Drosophila DL1 cells (Supplementary Figure S1A and B). However, no depletion of target mRNAs could be observed when LwaCas13a was employed in the same co-transfection assay that had been used in Figure ​Figure11 to characterize RxCas13d (Supplementary Figure S6). This negative result is consistent with a recent report that also failed to observe target mRNA knockdown by LwaCas13a in human cells (34).

We instead obtained much more promising results when PspCas13b (from Prevotella sp. P5-125) (7) was used in the co-transfection assays (Figure ​(Figure4A).4A). Guide RNAs (30-nt spacer length) complementary to eGFP (Figure ​(Figure4B,4B, left) or mCherry (Figure ​(Figure4B,4B, right) resulted in 50–75% depletion of the target transcript, indicative of strong on-target knockdown (cis-cleavage) by PspCas13b. As expected, these on-target effects were lost when the guide RNAs alone (without the PspCas13b effector) were transfected (Supplementary Figure S7) or when catalytic dead PspCas13b (denoted dPspCas13b), which harbors mutations (H133A/H1058A) in the HEPN nuclease domains, was employed (7) (Figure ​(Figure4C).4C). In stark contrast to what we observed with RxCas13d (Figure ​(Figure1B),1B), all of the eGFP and mCherry guide RNAs tested with catalytically active PspCas13b resulted in no significant change in expression of the non-target reporter mRNA or the mRNA encoding the Cas13 effector itself (Figure ​(Figure4B).4B). This suggests that PspCas13b has better specificity than RxCas13d in Drosophila cells.

Open in a separate window

Figure 4.

Co-transfection assays revealed PspCas13b has better specificity in Drosophila cells. (A) Drosophila DL1 cells were co-transfected with (i) 50 ng of plasmid that constitutively expresses a guide RNA from the U6 promoter as well as HA-tagged catalytically active or dead (H133A and H1058A mutations) PspCas13b from the Ubi-p63e promoter, (ii) 225 ng of plasmid that expresses eGFP from the copper-inducible MtnA promoter, and (iii) 225 ng of plasmid that expresses mCherry from the MtnA promoter. 24 h after transfection, CuSO₄ was added and total RNA was isolated after an additional 14 h. Northern blots were then performed. (B) Plasmids expressing active PspCas13b and a guide RNA complementary to eGFP (left) or mCherry (right) were employed in the co-transfection assay. Representative Northern blots (20 μg of total RNA/lane) are shown. ImageQuant was used to quantify the relative expression levels of eGFP, mCherry and PspCas13b mRNAs from three independent experiments. eGFP and mCherry mRNA expression was normalized to the empty vector samples, while PspCas13b mRNA expression was normalized to the random guide RNA samples. RpL32 mRNA served as an endogenous loading control. Data are shown as mean ± SD. For statistical comparisons, data were compared to the random guide RNA samples. () P < 0.05. No significant changes in expression of the mRNAs encoding PspCas13b or the off-target fluorescent protein were found. (C) Same as (B) except that plasmids expressing catalytic dead PspCas13b (dPspCas13b) were used. n.s., not significant.

To test the robustness of PspCas13b, we systematically altered each of the components of the co-transfection assay in a manner analogous to how RxCas13d was characterized in Supplementary Figure S4. The PspCas13b guide RNA spacer length was decreased from 30 to 24 nucleotides (Supplementary Figure S8A), the promoter (Supplementary Figure S8B) or open reading frame (Supplementary Figure S8C) of the on-/off-target reporter mRNAs was changed, and the assays were performed in Drosophila S2 cells rather than DL1 cells (Supplementary Figure S8D). In all cases, we found that PspCas13b was able to catalyze on-target knockdown with no significant change in expression of the off-target reporter.

PspCas13b, but not RxCas13d, can be used to specifically deplete a circular RNA in Drosophila cells

Recent work has revealed that many eukaryotic protein-coding genes can be alternatively spliced to yield linear RNAs as well as covalently closed circular RNAs (49–52) (Figure ​(Figure5A,5A, left). Because the exonic sequences present in a mature circular RNA are almost always also present in its cognate linear mRNA, approaches for depleting circular RNAs have focused on the backsplicing junction (BSJ) as this is the only sequence unique to the circular RNA. Indeed, two recent reports have suggested that targeting RxCas13d to a circular RNA using a guide RNA complementary to its BSJ can be used to specifically deplete the transcript without affecting cognate linear mRNA levels (34,36). Given our data from the co-transfection assays in Drosophila cells, we wanted to revisit this approach and determine whether RxCas13d or PspCas13b could indeed be used to specifically deplete a circular RNA in Drosophila cells.

Open in a separate window

Figure 5.

PspCas13b, but not RxCas13d, can be used to specifically deplete a circular RNA in Drosophila cells. (A) A three-exon Laccase2 minigene driven by the copper-inducible MtnA promoter can be alternatively spliced to yield a linear mRNA or a circular RNA derived from exon 2. To test the ability of Cas13 effectors to catalyze isoform-specific depletion, guide RNAs were designed that should deplete only the linear RNA (Exon 3 guide), only the circular RNA (BSJ guide), or both the linear and circular RNAs (Exon 2 guide). Drosophila DL1 cells were co-transfected with (i) 50 ng of plasmid that constitutively expresses a guide RNA as well as RxCas13d or PspCas13b effector and (ii) 450 ng of the Laccase2 Exon 1–3 minigene expression plasmid. 24 h after transfection, CuSO₄ was added and total RNA was isolated after an additional 14 h. (B) Representative Northern blots (20 μg of total RNA/lane) using an oligonucleotide probe complementary to exon 2 of the Laccase2 minigene. (C) ImageQuant was used to quantify the expression levels of linear and circular RNA normalized to the empty vector samples. RpL32 mRNA served as an endogenous loading control. Data are shown as mean ± SD, N = 3. For statistical comparisons, data were compared to the random guide RNA samples. () P < 0.05.

We focused on the Drosophila laccase2 (straw) gene, which can be backspliced to generate a 490-nt circular RNA from exon 2 that accumulates to high levels in vivo (38,53). In prior work, we characterized the cis-acting sequences required for laccase2 backsplicing and generated a copper-inducible three-exon minigene plasmid (Hy_pMT Laccase2 Exons 1–3) that efficiently produces a three-exon linear RNA as well as a circular RNA from exon 2 (38,39) (Figure ​(Figure5A,5A, left). To test the ability of Cas13 effectors to modulate the output of this minigene, guide RNAs were designed that are complementary to (i) only the circular RNA (BSJ guide), (ii) only the linear RNA (Exon 3 guide) or (iii) both the linear and circular RNAs (Exon 2 guide). Co-transfection assays were then performed in Drosophila DL1 cells and Northern blots used to quantify the levels of linear and circular RNA derived from the minigene (Figure ​(Figure5A).5A). When RxCas13d was employed, the knockdown effects lacked specificity: the BSJ guide RNA (which should be specific for the circular RNA) also significantly depleted the linear mRNA and the Exon 3 guide RNA (which should be specific for the linear RNA) also strongly depleted the circular RNA (Figures ​(Figures5B,5B, ​,C,C, left). In contrast, PspCas13b was able to deplete only the transcript of interest (Figures ​(Figures5B, C,5B, C, right). The PspCas13b BSJ guide RNA depleted the target circular RNA by 60 ± 11% while having no significant effect on linear RNA levels. Similarly, the PspCas13b Exon 3 guide RNA depleted the target linear RNA by 60 ± 12% while having no significant effect on circular RNA levels. These observations suggest that PspCas13b, but not RxCas13d, can be used to specifically knock down individual spliced isoforms, including circular RNAs, in Drosophila cells.

RxCas13d can have strong off-target effects in human cells

Having defined the efficiency and specificity of several Cas13 effectors in Drosophila cells, we next aimed to perform similar co-transfection analyses in human cells, focusing first on RxCas13d. HeLa cells were transiently transfected with plasmids that constitutively express (i) HA-tagged RxCas13d (with nuclear localization sequences) followed by a 2A peptide and eGFP (8) (Supplementary Figure S9A), (ii) a guide RNA (Supplementary Figure S1C), (iii) 3xFLAG-tagged firefly luciferase (FFLuc) and (iv) 3xFLAG-tagged nanoLuciferase (nLuc) (Figure ​(Figure6A).6A). Guide RNAs complementary to FFLuc (Figure ​(Figure6B,6B, left) or nLuc (Supplementary Figure S9B) enabled RxCas13d to deplete the target transcripts by 50–75%. Nonetheless, similar to the results in Drosophila cells (Figure ​(Figure1B),1B), activation of RxCas13d in HeLa cells resulted in trans-cleavage events and significant degradation of the non-target reporter mRNA as well as RxCas13d-2A-eGFP transcripts (Figure ​(Figure6B,6B, Supplementary Figure S9B). Northern blots notably revealed no change in expression of endogenous GAPDH transcripts in these experiments. However, we noted that significantly less total RNA could be isolated from the RxCas13d co-transfection assays when the FFLuc (Figure ​(Figure6C)6C) or nLuc (Supplementary Figure S9C) guide RNAs were used. This suggests that the presence of active RxCas13d may cause cells to die or be subjected to a growth disadvantage, which is consistent with several recent reports that indicate RxCas13d can be toxic in Drosophila, human U87 glioblastoma cells, and mouse ES cells (29,31). We propose that this defect in transfected cells causes the signal from non-transfected cells to dominate the Northern blots, thereby making it difficult to see changes in expression of endogenous transcripts. Regardless, the non-target reporter and RxCas13d-2A-eGFP mRNAs make it clear that trans cleavage of bystander RNAs can be an issue for RxCas13d in human HeLa cells (Figure ​(Figure6B,6B, Supplementary Figure S9B). Consistent with this conclusion, we also observed significant off-target effects when examining the effect of RxCas13d on other combinations of reporter genes, including Renilla luciferase (RLuc)/mCherry (Supplementary Figure S10A–C) and nLuc/RLuc (Figure ​(Figure7A7A–C).

Open in a separate window

Figure 6.

Quantification of on- and off-target effects of RxCas13d and PspCas13b in human HeLa cells. (A) HeLa cells were co-transfected with (i) 300 ng of plasmid that constitutively expresses HA-tagged Cas13 protein followed by a 2A peptide and eGFP, (ii) 200 ng of plasmid that expresses a guide RNA, (iii) 250 ng of plasmid that expresses nanoLuciferase (nLuc) and (iv) 250 ng of plasmid that expresses firefly luciferase (FFLuc). 48 h after transfection, total RNA was isolated and Northern blots performed. (B, C) Guide RNAs complementary to FFLuc were employed in the co-transfection assay. (B) Representative Northern blots (20 μg of total RNA/lane) are shown. ImageQuant was used to quantify the relative expression levels of FFLuc, nLuc, and Cas13-2A-eGFP mRNAs. nLuc and FFLuc mRNA expression was normalized to the empty vector (pBEVY-L) samples, while Cas13-2A-eGFP mRNA expression was normalized to the random guide RNA samples. GAPDH mRNA served as an endogenous loading control. Data are shown as mean ± SD, N = 3. For statistical comparisons, data were compared to the random guide RNA samples. () P < 0.05. (C) Relative RNA concentrations obtained from the co-transfection assays when the RxCas13d or PspCas13b expression plasmids were used. Data are normalized to the empty vector samples and shown as mean ± SD, N = 5. () P < 0.05. (D, E) HeLa cells were co-transfected with constant amounts of RxCas13d, guide RNA, and nLuc expression plasmids, but variable amounts of FFLuc expression plasmid (20, 50 or 150 ng). (D) RT-qPCR was used to quantify depletion of FFLuc mRNA. For each amount of FFLuc plasmid transfected, the relative abundance of FFLuc mRNA was normalized to the respective random guide RNA samples. Data are shown as mean ± SD, N = 3. () P < 0.05. (E) Northern blots (20 μg of total RNA/lane) were used to quantify the relative expression levels of nLuc and RxCas13d-2A-eGFP mRNAs. Data are shown as mean ± SD, N = 3. For statistical comparisons, data were compared to the random guide RNA samples. () P < 0.05. A complete table of P-values for all comparisons in (D) and (E) is provided in Supplementary Table S5.

Open in a separate window

Figure 7.

The extent of RxCas13d off-target effects varies across human cell lines. (A) HeLa or HEK293T cells were co-transfected with (i) 300 ng of plasmid that constitutively expresses HA-tagged RxCas13d protein followed by a 2A peptide and eGFP, (ii) 200 ng of plasmid that expresses a guide RNA, (iii) 250 ng of plasmid that expresses Renilla luciferase (RLuc) and (iv) 250 ng of plasmid that expresses nanoLuciferase (nLuc). 48 h after transfection, total RNA was isolated and Northern blots performed. (B–E) Guide RNAs complementary to nLuc or RLuc were employed in the co-transfection assay in HeLa cells (B, C) or HEK293T cells (D, E). (B, D) Representative Northern blots (20 μg of total RNA/lane) are shown. ImageQuant was used to quantify the relative expression level of nLuc, RLuc, and Cas13-2A-eGFP mRNAs. nLuc and RLuc mRNA expression was normalized to the empty vector (pBEVY-L) samples, while Cas13-2A-eGFP mRNA expression was normalized to the random guide RNA samples. GAPDH mRNA served as an endogenous loading control. Data are shown as mean ± SD, N = 3. For statistical comparisons, data were compared to the random guide RNA samples. () P < 0.05. (C, E) Relative RNA concentrations obtained from the co-transfection assays. Data are normalized to the empty vector samples and shown as mean ± SD, N = 3. () P < 0.05. n.s., not significant.

To determine if the off-target (trans-cleavage) effects of RxCas13d in HeLa cells are diminished when the target mRNA level is reduced, we performed a titration assay analogous to that performed previously in Drosophila cells (Figure ​(Figure3).3). The amounts of the RxCas13d-2A-eGFP, guide RNA and nLuc (off-target mRNA) expression plasmids transfected were kept the same as in Figure ​Figure6B,6B, but the amount of the FFLuc (target mRNA) expression plasmid was reduced from 250 ng to as low as 20 ng (Figure ​(Figure6D6D and E). At the highest FFLuc expression level (150 ng FFLuc plasmid transfected), the nLuc and RxCas13d-2A-eGFP transcripts were depleted by 50–75% (Figure ​(Figure6E)6E) which mirrors the amount of on-target nLuc depletion (cis- cleavage) observed (Figure ​(Figure6D).6D). Meanwhile, when 20 ng FFLuc plasmid was transfected, RxCas13d was able to deplete FFLuc mRNA (Figure ​(Figure6D)6D) with no significant change in nLuc or RxCas13d-2A-eGFP transcript levels (Figure ​(Figure6E).6E). This suggests RxCas13d may be able to specifically deplete low abundance transcripts in HeLa cells, but these assays need to be interpreted cautiously and one must be aware of the potential for off-target (trans-cleavage) effects.

The extent of RxCas13d off-target effects differs between human cell lines

Given that several prior reports have indicated that RxCas13d lacks off-target effects in human cells (8,33–36), we next wanted to further explore the underlying reasons for the conflicting observations. A prominent difference between studies is often the cell type examined, so we wanted to perform the same co-transfection experiments (Figure ​(Figure7A)7A) side-by-side in HeLa (Figure ​(Figure7B, C)7B, C) and human embryonic kidney HEK293T cells (Figure ​(Figure7D,7D, ​,E).E). When guide RNAs complementary to nLuc or RLuc were employed in HeLa cells, significant depletion of the non-target reporter and RxCas13d-2A-eGFP mRNAs was observed (Figure ​(Figure7B)7B) and less total RNA could be isolated from the co-transfection assays (Figure ​(Figure7C).7C). In contrast, when these same experiments were performed in HEK293T cells, guide RNAs complementary to nLuc or RLuc did not decrease the expression of the non-target reporter (Figure ​(Figure7D)7D) or the total RNA yield (Figure ​(Figure7E),7E), but expression of the RxCas13d-2A-eGFP transcripts did notably decrease (Figure ​(Figure7D).7D). This indicates that the extent of off-target effects can differ depending on the cell type examined and that very different conclusions can be drawn depending on how one measures off-target (trans-cleavage) effects. Here, if one only focuses on the effects on RxCas13d-2A-eGFP mRNA levels, then it might be concluded that RxCas13d has the same degree of off-target effects between HeLa and HEK293T cells (Figure ​(Figure7B,7B, ​,D).D). On the other hand, if one only focuses on the non-target luminescent reporter, then it might be concluded that RxCas13d has no off-targets in HEK293T cells (Figure ​(Figure7D).7D). Both of these conclusions are incorrect, and our results instead reveal a more complicated picture regarding trans-cleavage events in human cells. It remains unclear why some transcripts like RxCas13d-2A-eGFP are more sensitive than others to collateral damage by active RxCas13d, but similar differences in the extent of selective RxCas13d off-target effects were observed when a different combination of reporter genes was employed (Supplementary Figure S10A-E). Therefore, although it has been assumed that trans-cleavage events catalyzed by RxCas13d should be sequence-independent and that all transcripts should decrease proportionally, this is not what we observed. Future work will hopefully reveal the underlying reasons for the transcript specific differences, which may be related to differences in subcellular localization or transcript accessibility.

We thank all the members of the Wilusz lab, Junwei Shi (University of Pennsylvania), as well as Chase Kelley and Eric T. Wang (University of Florida) for discussions and advice.

Author contributions: Y.A. and J.E.W. conceived and designed the project. Y.A. and D.L. performed experiments and analyzed data. Y.A. and J.E.W. wrote the manuscript with input from all authors.