CRISPR Genome Editing Systems in the Genus Clostridium: a Timely Advancement

The genus Clostridium is composed of bioproducers, which are important for the industrial production of chemicals, as well as pathogens, which are a significant burden to the patients and on the health care industry. Historically, even though these bacteria are well known and are commonly studied, the genetic technologies to advance our understanding of these microbes have lagged behind other systems. New tools would continue the advancement of our understanding of clostridial physiology.

ABSTRACT

The genus Clostridium is composed of bioproducers, which are important for the industrial production of chemicals, as well as pathogens, which are a significant burden to the patients and on the health care industry. Historically, even though these bacteria are well known and are commonly studied, the genetic technologies to advance our understanding of these microbes have lagged behind other systems. New tools would continue the advancement of our understanding of clostridial physiology. The genetic modification systems available in several clostridia are not as refined as in other organisms and each exhibit their own drawbacks. With the advent of the repurposing of the CRISPR-Cas systems for genetic modification, the tools available for clostridia have improved significantly over the past four years. Several CRISPR-Cas systems such as using wild-type Cas9, Cas9n, dCas9/CRISPR interference (CRISPRi) and a newly studied Cpf1/Cas12a, are reported. These have the potential to greatly advance the study of clostridial species leading to future therapies or the enhanced production of industrially relevant compounds. Here we discuss the details of the CRISPR-Cas systems as well as the advances and current issues in the developed clostridial systems.

INTRODUCTION

Clostridia are comprised of obligate anaerobic endospore-forming firmicutes, and the best studied act as either important producers of industrially relevant components (e.g., Clostridium acetobutylicum or C. cellulolyticum) or pathogens (e.g., C. botulinum or Clostridioides difficile) (1, 2). Genetic tool development in clostridia increased over the last few years and continues to expand (3). Counterselection tools for homologous recombination or allelic exchange are common across many bacterial organisms, including clostridia (4). Such counterselection markers include I-SceI (5), upp (6), mazF (7), codA (8), galK (9), pyrE (10), and pyrF (11). Though common, each tool has limitations, which have been discussed in length (12–15). More recently, single-stranded DNA annealing protein was used to engineer a recombineering system for C. acetobutylicum (16). This protein, RecT, is an ortholog from C. perfringens and is used by phage (i.e., Rac or lambda prophages) to help increase the efficiency of homologous recombination of the single-stranded DNA into bacterial genomes. Through the help of RecT, a short oligonucleotide is introduced into the target site (16, 17). This genetic tool shows the potential for exploitation of bacteriophage-derived single-stranded DNA binding proteins for homologous recombination (16).

In the mid-2000s, TargeTron technology (ClosTron) was repurposed for use in clostridia (18, 19). This development led to a surge in our understanding of the physiology and molecular processes that occur in clostridia, especially in C. difficile. The TargeTron system relies on the retargeting of mobilizable group II introns to create insertion mutations in the desired target site. The inclusion of retrotransposable activated markers (RAMs) permits the activation of an antibiotic resistance gene upon splicing from the intron RNA, thereby providing antibiotic resistance upon insertion into the genome (18, 19). Despite the popularity of this tool, it has major drawbacks compared to clean deletions of target genes. The TargeTron target sites can be difficult to identify in coding regions which are relatively small, <400 bp. Because of this, not every gene can be targeted for mutation (7, 12). Moreover, due to the nature of the mutation, the insertion results in polarity for downstream genes (12). Also, due to target sites being chosen by an algorithm, targeting at some sites results in aberrant insertions at other sites in the genome. Lastly, the antibiotic resistance marker has led to perceived hypervirulence of mutant strains in the hamster model of C. difficile infection (20, 21).

With the advent of the CRISPR-Cas systems for genome editing, there has been a surge in the application of this tool in many organisms. The ease of use and pliability of these systems make them an attractive target to be developed for use in clostridia. Here we discuss CRISPR-Cas9/Cpf1 genetic engineering and the CRISPR-cas tools developed so far in Clostridium species.

CRISPR

The clustered regularly interspaced short palindromic repeats (CRISPR), along with their CRISPR-associated (Cas) proteins, have been developed into one of the most promising and successful genome editing tools to date (22). Although the significance was unknown at the time, the CRISPR system was first discovered in 1987 (23). CRISPR loci have been found in ∼47% of bacteria and ∼87% of archaea and have been suggested to have spread through horizontal transfer among prokaryotes as a basis of adaptive immunity. CRISPR arrays within bacteria and archaea are divided into three major types and sixteen different subtypes based on the cas genes present within the organisms (24).

CRISPR systems function as defense mechanisms to evade attacks from phage and other mobile genetic elements by incorporating a small fragment of the invader’s DNA into the host’s genome. When an invading organism inserts its DNA into the bacterial cell, the native Cas proteins adapt to the infection by facilitating the incorporation of ∼30-bp fragments of this invading DNA (spacers) into the CRISPR locus on the host’s chromosome. Within the CRISPR locus is a series of spacers, which have identity to prior adaptation events, flanked by repeat sequences of ∼30 bp in length, forming the CRISPR array. The acquisition of spacers occurs in an ordered fashion, where each spacer is incorporated at the beginning of the array behind the leader repeat sequence, resulting in a timeline of the previous infections (25).

The CRISPR array is constitutively transcribed into pre-CRISPR RNAs (pre-crRNAs). These pre-crRNAs are processed and cleaved into individual crRNAs which are loaded onto a Cas endonuclease (26). In type II CRISPR systems, a trans-activating crRNA (tracrRNA) is upstream of the CRISPR locus and is essential for crRNA maturation by RNase III and Cas9 (27). The tracrRNA and crRNA form a duplex that is loaded onto the Cas9 endonuclease (26). The most well-known and characterized Cas endonuclease is Cas9 from Streptococcus pyogenes (22). Once the Cas proteins are loaded with duplexed tracrRNA:crRNAs, the Cas-RNA complex will find complementary sequences within invading nucleic acids (28). The ability of the Cas protein to discriminate this foreign DNA from the complementary sequence encoded by the CRISPR array of the host is through the use of a protospacer adjacent motif (PAM). The PAM sequence is two to five nucleotides in length and found at either the 5′ end or 3′ end of the protospacer sequence in the invader’s genome; the location and sequence of the PAM sequence recognized on the invading DNA is unique for each type of Cas endonuclease (29, 30). Upon recognition of the complementary sequence by the CRISPR RNA and the PAM by the Cas endonuclease, the Cas endonuclease introduces a double-stranded DNA break into the invading DNA or a single-stranded RNA break into the invading RNA (29–31).

Soon after the discovery of the function of the Cas9 endonuclease, the first CRISPR genome editing systems were developed in human cells, zebrafish embryos, and bacteria (Streptococcus pneumoniae and Escherichia coli) (32–39). In bacteria, a Cas endonuclease, a single guide RNA (sgRNA), and a donor region for homologous recombination are required for in vivo editing (Fig. 1A) (39). The Cas endonuclease cleaves double-stranded DNA. In order for the endonuclease to cleave at a specific location, a guide RNA is used to direct the endonuclease to the intended target site (40). To simplify the system, the gRNA has been engineered to be produced as a single RNA molecule by fusing the crRNA and the tracrRNA. The result is an sgRNA which loads onto the endonuclease and directs it to the target site by binding to the DNA (30). In silico identification of sgRNAs is chosen where a PAM site is directly up- or downstream from the target sequence (29). The sgRNA brings the Cas endonuclease to the intended target, and endonuclease recognizes the PAM sequence and cleaves the DNA (29–31). To date, the most commonly used Cas endonuclease for genome editing in bacteria is Cas9 from S. pyogenes; the Cpf1 endonuclease from Acidaminococcus sp. is being developed as an alternative to Cas9 (22, 41, 42). Because many bacteria either do not have or have inefficient nonhomologous end joining (NHEJ) repair systems, repair of double-stranded DNA breaks through homologous recombination is preferred (43). To facilitate this repair, a donor region that encodes the desired change in the genome is included in these systems (40). In prior work, Jiang et al. (39) demonstrated that homologous recombination post-Cas9 cleavage of the DNA resulted in only a small increase in editing. Their work suggests that Cas9 cleavage of the bacterial DNA counterselects the population of cells that has not recombined with the donor region; recombination of the donor region with the chromosome removes the sgRNA site from the chromosome. Here, we review the developed CRISPR systems for genetic modification in clostridia, which greatly expanded in 2015 and continue to expand today.

FIG 1.

Comparison of CRISPR system plasmids. Representative sample mutagenesis plasmids for CRISPR editing systems are shown. Each contains the three regions: a homology region (denoted by homology arms [HA]), an endonuclease (Cas9 or Cpf1), and a gRNA. Plasmids for editing using Cas9 or Cas9n (A), dCas9 with no necessary homology region (B), Cpf1/Cas12a with one target (C), and Cpf1/Cas12a with two targets (D) are shown. Examples of CRISPR-Cas mutagenesis plasmids are listed inside each plasmid.

WILD-TYPE Cas9

In 2012, the function of the type II Cas9 endonuclease was characterized and kicked off the CRISPR-Cas9 genome editing revolution (30, 32, 33). The Cas9 enzyme relies on the base pairing between the tracrRNA and the crRNA to cleave the cDNA adjacent to the PAM sequence (5′-NGG-3′ in S. pyogenes) located at the 3′ end of the target sequence (29–31). Cas9 has two nuclease domains (HNH and RuvC-like) that are essential for the nuclease to induce a double-stranded DNA break (Fig. 2A). The HNH domain, a ββα-metal fold which contains the active site, cleaves the DNA strand complementary to the targeting sequence located on the crRNA and the RuvC-like domain cleaves the opposite strand of DNA and shares an RNase H fold similar to the RuvC Holliday junction resolvase (30, 44, 45). Mutation of either domain results in a Cas9 nuclease that only cleaves a single strand of DNA (Fig. 2A). For example, introducing an aspartic acid to alanine mutation at the 10th amino acid (D10A) inactivates the RuvC-like domain (Cas9n) and results in an enzyme that only makes single-stranded DNA breaks (“nickase”) (30). In addition, mutation of both nuclease domains, D10A and H840A, results in a catalytically dead nuclease, also known as dCas9, which will be discussed later (34, 46).

FIG 2.

Review of CRISPR-Cas genetic modification systems. Shown are graphical representations of each CRISPR-Cas system discussed in this review. Each contains the different endonucleases: Cas9 (A), Cpf1/Cas12a (B), and dCas9 with an activator/repressor (C), as well as the sgRNA or crRNA, PAM site and sequence, and cleavage locations and their respective cleavage domains for each. A table is included in panel A to describe the different mutant alleles of Cas9. Panel C also shows the promoter region area and the start codon for reference.

In some organisms, including clostridia, the expression of the Cas9 endonuclease has a high cost on the bacterial cell in terms of toxicity (12, 13, 47–50). Despite this cost, CRISPR-Cas9 systems have been successfully applied in C. acetobutylicum (13, 14, 51), Clostridium autoethanogenum (52), Clostridium beijerinckii (12, 13, 50, 53), C. cellulolyticum (48, 54), C. difficile (15, 55), Clostridium ljungdahlii (56), Clostridium pasteurianum (49, 51), and Clostridium saccharoperbutylacetonicum (57) (Table 1) by tightly regulating the expression of the cas9 gene. Specifically, the implementation of inducible promoters (e.g., tetracycline-, xylose-, or lactose-inducible promoters) has helped to circumvent the issue of Cas9 toxicity (14, 15, 50, 55, 57). Except for C. cellulolyticum, Clostridium cellulovorans, and Clostridium tyrobutyricum, all of these clostridia use CRISPR-Cas9 systems that encode wild-type Cas9 and have mutation efficiencies exceeding ∼50% (Table 1).

TABLE 1.

CRISPR-Cas systems and their properties in clostridia

Species	Endonuclease used				Organism of endonuclease	No. of plasmids	Homology arm length (bp)	Endonuclease promoter	sgRNA promoter	Type(s) of mutationa	Editing efficiency or repression (%)	Reference
	Cas9	dCas9	Cpf1	Endogenous
C. acetobutylicum	X				S. pyogenes	1	500–1,000	thl	sRNA sCbei_5830	Replacements	100	51
	X				S. pyogenes (COb )	2	500–1,000	tet (inducible)	thl	NM, small deletions, small replacements	100	14
	Nc				S. pyogenes	1	500–1,000	ptb and thl	pJ23119	Small deletions	7 to 100	13
		X			S. pyogenes	1	—d	thl	sRNA sCbei_5830	Knockdown	20 to 90	51
		X			S. pyogenes	1	—	ptb	pJ23119	Knockdown	45	13
C. autoethanogenum	X				S. pyogenes (CO)	2	1,000	IPL12	Native Wood-Ljungdahl cluster	Deletions	50 to 83	52
C. beijerinckii	X				S. pyogenes	1	1,000	spoIIE	thl	Small deletions	47 to 100	12
	X				S. pyogenes	1	1,000	spoIIE and lac (inducible)	sRNA sCbei_5830	Small deletions, small insertions, NM	0 to >99	50
	N				S. pyogenes	1	500–1,000	thl	pJ23119	Small deletions	19 to 100	13
	N				S. pyogenes	1	—	thl	pJ23119	Random NM	20 to 94	53
		X			S. pyogenes	1	—	thl	sRNA sCbei_5830	Knockdown	65 to 95	59
		X			S. pyogenes	1	—	ptb	pJ23119	Knockdown	84	13
			X		Acidaminococcus sp.	1	500	lac (inducible)	sRNA sCbei_5830	Small deletions	100	61
C. cellulolyticum	N				S. pyogenes (CO)	1	100–1,000	fdx	P4 (synthetic)	Small deletions, small insertions	<95 to 100	48
	N				S. pyogenes (CO)	1	NRe	fdx	P4 (synthetic)	Small insertion	NR	54
C. cellulovorans		X			S. pyogenes	1	—	thl	pJ23119	Knockdown	77 to 95	68
C. difficile	X				S. pyogenes (CO)	1	1,000	tet (inducible)	gdh	Small deletions	20 to 50	15
	X				S. pyogenes	1	500–1,000	lac (inducible)	sRNA sCbei_5830	Small deletions and insertions	80 to 100	55
		X			S. pyogenes (CO)	1	—	xyl (inducible)	gdh	Knockdown	∼90	69
			X		Acidaminococcus sp.	1	500	lac (inducible)	sRNA sCbei_5830	Small and large deletions, double deletions	25 to 100	62
C. ljungdahlii	X				S. pyogenes	1	1,000	thl	araE	Small and large deletions	50 to 100	56
		X			S. pyogenes	1	—	lac (inducible) and tet (inducible, from C. acetobutylicum)	P4 (synthetic)	Knockdown	97 to 99	60
C. pasteurianum	X				S. pyogenes	1	1,000	thl	sRNA sCbei_5830	Small deletions	100	51
	X				S. pyogenes	1	1,000	thl	sRNA sCbei_5830	Small deletion	100	49
		X			S. pyogenes	1	—	thl	sRNA sCbei_5830	Knockdown	85	51
				X	C. pasteurianum	1	1,000	Native CRISPR leader	—	Small deletions	100	49
C. saccharoperbutylacetonicum	X				S. pyogenes	1	1,000	lac (inducible)	sRNA sCbei_5830 and pJ23119	Deletions	6 to 75	57
C. tyrobutyricum				X	C. tyrobutyricum	1	1,000	lac (inducible)	—	Small deletions	6.7 to 100	91

^aNM, nucleotide mutation; small deletion/insertion, smaller than 2 kb; large deletion/insertion, larger than 2 kb.

^bCO, codon optimized.

^cN, nickase.

^d—, not present.

^eNR, not reported.

A few clostridial CRISPR systems have used the Cas9n allele (13, 48, 53, 54). The major advantage of this allele of Cas9 is for use in bacteria which have poor double-stranded DNA break repair systems or in bacteria where expressing both wild-type Cas9 and the gRNA is lethal (48, 58). By using the Cas9n allele, the organism can overcome the toxic effects induced by wild-type Cas9. In C. cellulolyticum, Cas9n was required to merely obtain transformants of the mutagenesis plasmid (48). In that study, however, the endonuclease was under the control of a constitutive promoter (48). It is unclear if the use of an inducible system in C. cellulolyticum would allow for tighter regulation of the wild-type Cas9 endonuclease and overcome the limitation of using the wild-type cas9 allele. Interestingly, other clostridia that have developed Cas9n as a tool also have tools that use wild-type Cas9 (i.e., C. acetobutylicum [13, 14, 51] and C. beijerinckii [12, 13, 50, 53]). It is unclear if it is necessary to use the Cas9n allele in these systems; the editing efficiencies are high for both Cas9 alleles.

The most recent use of the Cas9n allele in C. beijerinckii is different than what is used in any other clostridia (53). Here, the nickase is fused to both a cytidine deaminase and a uracil DNA glycosylase inhibitor. In this unique system, the complex is guided to the target site by the sgRNA, a nick is made in the DNA, specific base pair substitutions (C/G to T/A) are introduced, and the site is repaired. These changes result in possible missense mutations or null mutations in the targeted gene/coding region. The efficiencies reported here for base editing are comparable to those seen for gene editing using Cas9. While this method avoids the use of a donor region on the plasmid (decreasing the size of the plasmid as well as obviating the need of the bacteria to repair the lesion), it does not introduce specific mutations within the chromosome (53).

One aspect of the CRISPR-Cas9 system is the composition of the target sequence and the design of the sgRNA that recognizes this sequence (e.g., G+C content and specific nucleotide positions within the intrinsic sequence). Oddly, none of the developed clostridial CRISPR-Cas9 systems have discussed this aspect of the system. The authors report differences in editing efficiencies when targeting different genes, but there are no direct discussions about the sgRNAs used or if this impacted the efficiencies of gene editing. In all of these systems, one sgRNA is used to make one mutation. Whether or not the authors tested other potential target sites within the mutated gene is unclear. The inclusion of multiple sgRNAs within the same plasmid to make multiple mutations at one time has not been accomplished in clostridia using the wild-type Cas9 allele. To accomplish this, each sgRNA would require its own promoter to drive its expression, or different spacers could be used if the tracrRNA is expressed separately. Unfortunately, this would result in an increase in the size of the mutagenesis plasmid, which may result in issues when working with such a large plasmid. Smaller promoter regions such as the C. beijerinckii sCbei_5830 small RNA promoter (49–51, 59) or the C. cellulolyticum P4 synthetic promoter (48, 54, 60) used in some clostridia CRISPR systems could be functional in other Clostridium species. Both promoters, the sCbei_5830 small RNA promoter and P4 promoter, are small (42 bp and 36 bp, respectively) and are functional in several clostridia (Table 1) (48–51, 54, 55, 57, 59–62). This is promising for those clostridia which do not have many genetic tools and for which strong promoters are unknown.

The use of a donor region for homologous recombination is often necessary in bacteria to achieve high editing efficiencies (39). A double-stranded chromosomal break is highly toxic to cells, and recombination of the homology region with the chromosome protects against Cas9 targeting by removing the targeting site (48, 63). Nearly 28% of bacteria have the NHEJ component Ku (64). NHEJ components either are not present in some bacteria or have low expression. This poses a problem for genome editing when trying to repair a Cas9-induced double-stranded DNA break. Without added expression of NHEJ components, such as on a multicopy plasmid, repairing the lesion using this method is not practical (65). Therefore, homologous recombination-based protection is much more efficient and sensible. The most commonly used homology arm length in clostridia CRISPR-Cas systems is 1 kb each, with 500 bp each being the second most common(12–15, 48–52, 55–57). The size of the homology arms may differ for each bacteria or CRISPR-Cas9 system, but the lowest reported homology region that achieves reasonable editing efficiency in clostridia is 100 bp (48).

dCas9/CRISPRi

CRISPR-Cas9-mediated genome editing creates a markerless edit within the target DNA (e.g., bacterial chromosome or plasmid) (22). Not all genes can be targeted, as some genes are essential for the survival of the bacteria. For these reasons, a genetic system which does not make chromosomal deletions and only regulates the expression of genes is necessary (66). CRISPR interference (CRISPRi) has been developed, where both catalytic sites of Cas9 have been mutated: D10A (located in the RuvC-like domain) and H840A (located in the HNH domain) (Fig. 2A) (30, 46). These mutations render Cas9 catalytically dead. Here, Cas9 and the sgRNA function to bind and block a region of DNA from transcriptional activity. First developed and implemented in E. coli (46), the authors suggest that the Cas9:sgRNA complex collides with the elongating RNA polymerase (RNAP). They found that targeting the template strand permits RNAP to read through and not come in contact with the Cas9:sgRNA complex. Thus, targeting the nontemplate strand of DNA has been reported to yield higher repression than targeting the template strand in bacteria (46). A common use of this tool is to target the promoter region of a gene or operon to downregulate expression upon induction of the CRISPRi system (46, 67). A positive aspect to this system is that the plasmid size for the CRISPRi system is smaller than that of the original genome editing CRISPR-Cas systems, in that no donor region for DNA repair is necessary (Fig. 1B). More importantly the functionality or characterization of essential genes can be explored using the CRISPRi system (66).

A few Clostridium species have developed CRISPRi systems: C. acetobutylicum (13, 51), C. beijerinckii (13, 59), C. cellulovorans (68), C. difficile (69), C. ljungdahlii (60), and C. pasteurianum (51) (Table 1). In all cases, the successful repression of genes in these organisms ranges from 20% (e.g., plasmid-carried afp in C. acetobutylicum [51]) to 99% (e.g., pta in C. ljungdahlii [60]). These values depended on the gene being targeted and, more likely, the target sequence used for the sgRNA. The target sequence chosen for the implementation of CRISPRi is more important here due to the function of repression; an inefficient target sequence may not hold dCas9 on the DNA as tightly or consistently as a more efficient target sequence. Li et al. targeted the spo0A gene in both C. acetobutylicum and C. beijerinckii, using the same system, and obtained different repression percentages, 45% and 84%, respectively (13). The two sgRNA sequences are only 35% identical. Because the two sequences are different, and in different organisms, direct comparisons between the two studies cannot be made. Similarly, the location of targeting has an effect on the efficiency of repression. Target sites that are farther away from the transcription start site have lower efficiencies than those closer to the transcription start site. Moreover, the use of multiple sgRNAs increases the efficiency of repression, as long as the target sequences do not overlap. Finally, truncated sgRNA target sequences, those with less than 12 bp, do not result in repression of the target gene; full-length sgRNAs, those of 20 bp in length, are preferred for efficient repression (46). The only obvious issue with the CRISPRi genetic tool is that of polarity for downstream genes. The effects of knocking down one gene in an operon will likely have an effect on the downstream genes.

Cpf1/Cas12a

Cpf1 (Cas12a) is a type V CRISPR system effector protein which has been studied in Francisella novicida, Acidaminococcus sp. (AsCpf1), Moraxella bovoculi, and Lachnospiraceae bacterium (47). The Cpf1 endonuclease specifically recognizes T-rich PAMs instead of G-rich PAMs, as in the case for Cas9 (45). In another divergence from how the type II CRISPR system works, Cpf1 itself is responsible for the maturation of pre-crRNA; no tracrRNA is needed. AsCpf1 is guided to its target by the mature crRNA to recognize the PAM sequence 5′-TTTN-N₂₃-3′ (i.e., 5′-TTTN-3′ followed by a 23-bp protospacer). AsCpf1 then cleaves the double-stranded DNA resulting in a staggered, 5-nucleotide (nt) 5′ overhang that is 18 to 23 bp downstream from the PAM site (42). Unlike Cas9, AsCpf1 only uses one domain (RuvC-like) to digest both DNA strands rather than one domain for each strand of DNA (Fig. 2B) (42, 70). Even though this endonuclease is not as well-studied/characterized as the Cas9 endonuclease, Cpf1 has been used for genome engineering in Escherichia coli, Yersinia pestis, Mycobacterium smegmatis, and Corynebacterium glutamicum (71, 72) as well as in C. beijerinckii and C. difficile (Table 1) (61, 62).

One advantage of using Cpf1 over Cas9 is that the T-rich PAM sequence used by Cpf1 could be more probable in AT-rich organisms, such as clostridia, as well as for promoter regions, which are commonly AT-rich (45). It has been suggested that Cpf1 is better suited for bacteria, such as C. difficile, which have low DNA conjugation efficiencies. This suggestion is based on the described lower toxicity of the Cpf1 endonuclease (62). These toxicity claims were based upon a study of Corynebacterium glutamicum where the authors could not obtain transconjugants when introducing their CRISPR-Cas9 plasmids. However, they were able to obtain several transformants from the same system using Cpf1 instead of Cas9 (72). A separate study found that Cas9 is toxic compared to Cpf1 in Synechococcus sp. strain 2973 (73). However, because this is not an exhaustive list, the lower toxicity of Cpf1 than of Cas9 may be organism dependent. Finally, Cpf1 was suggested to have lower off-target effects than Cas9 (62, 74). However, the study referenced, as well as other studies, found this to be true in human cells and was not tested in bacteria (74, 75). In C. difficile, Cas9 has no known or detectable off-target effects as of the date of this review (15).

Another proposed advantage of using Cpf1 is that since a tracrRNA is not needed, the cost for constructing and using plasmids would be less due to the shorter gRNA (Fig. 1C). For Cpf1, a gRNA consisting of only 42 nucleotides would need to be synthesized compared to the >100-nucleotide sgRNA (crRNA and tracrRNA) needed for other type II endonucleases, such as S. pyogenes Cas9 (45, 73). Moreover, the smaller size of the gRNA could lead to the use of multiple gRNAs on a single plasmid to simultaneously make multiple mutations in a single application of this system (Fig. 1D) (61, 62, 76).

Only two clostridia have developed CRISPR-Cpf1 systems, C. beijerinckii and C. difficile, and the efficiencies of mutagenesis of these systems range from 25% (i.e., C. difficile ermB1-ermB2) to 100% (i.e., C. beijerinckii spo0A, C. beijerinckii pta, and C. difficile fur) (61, 62). These values are similar to those obtained from using wild-type Cas9 (Table 1). More genes would have to be targeted to identify significant differences between using these two endonucleases in their respective CRISPR systems. Using C. difficile as an example, the promoters used by Wang et al. (55) (Cas9) and Hong et al. (62) (Cpf1) for the endonuclease and the sgRNA are the same. Even though the efficiencies for the CRISPR-Cas9 system are higher than that of Cpf1, few genes were targeted and the same genes were not targeted between the studies (Table 1) (55, 62). As for C. beijerinckii, the same mutation efficiencies were reported for two genes, spo0A and pta, using either Cas9 or Cpf1 as the endonuclease (61). In that study, the same gRNA was not used in each system (61). Based on these few studies, there is low evidence of either endonuclease, Cas9 or Cpf1, being superior to the other in clostridia.

ENDOGENOUS CRISPR SYSTEMS

To overcome the toxic effects of S. pyogenes Cas9 or Acidaminococcus sp. Cpf1 endonucleases, some genome editing systems were designed that rely on the endogenous CRISPR array for editing (77–79). Seventy-four percent of clostridial species harbor CRISPR-Cas loci, including many of the well-studied clostridia (49, 80–91). Of these, C. pasteurianum and C. tyrobutyricum have developed CRISPR-Cas genome editing systems based on their respective endogenous systems (Fig. 3, Table 1) (49, 91).

FIG 3.

Endogenous CRISPR genome editing in clostridia. Shown is a graphical representation of CRISPR-Cas genome editing in a clostridial vegetative cell. (1) The endogenous CRISPR region contains a cas operon which encodes a Cas endonuclease that is subsequently generated. (2) Meanwhile, a plasmid contains a synthesized CRISPR array under the control of an inducible promoter and a donor region for homologous recombination containing an upstream homology arm (HA) and a downstream HA. When induced, the plasmid transcribes the crRNAs to be used by the endogenous system to generate individual crRNAs. (3) The endogenous Cas endonuclease complexes with a crRNA to target and cleave the DNA. This is then repaired by the donor region located on the CRISPR plasmid.

As a general starting place to develop endogenous CRISPR systems as genetic tools, Pyne et al. (49) analyzed the Cas proteins within the CRISPR array in C. pasteurianum and determined the PAM sequences by analyzing spacer sequences within the CRISPR array to find common nucleotide sequences among them. The authors then developed a single-plasmid system which mimics the native CRISPR system by including a synthetic CRISPR array and crRNAs with spacers corresponding to the target region which was to be mutated. The authors also predicted PAM sequences for three other clostridia: C. autoethanogenum, C. tetani, and C. thermocellum (49). Zhang et al. also developed an endogenous CRISPR-Cas system but for C. tyrobutyricum (91). The authors used similar approaches described above to analyze the CRISPR arrays and determined the PAM sequence for C. tyrobutyricum. In addition, the authors multiplexed the gRNAs and made simultaneous genomic deletions using one CRISPR-Cas plasmid construct (91).

Both of the tools using endogenous CRISPR-Cas systems differ from those using Cas9 or Cpf1 in that the endogenous systems rely on the endonucleases encoded in the genome to form mature crRNAs as well as make the double-stranded DNA breaks within the genome; no endonucleases are encoded by the plasmid. Genome editing control lies within the design of the CRISPR array containing the pre-crRNAs as well as the homology arms to be used as donor regions for homologous recombination. The CRISPR array would contain multiple pre-crRNAs, all under the control of a single promoter. Once transcribed, the endogenous system processes the pre-crRNAs into mature crRNAs that are then loaded onto the endogenous Cas endonucleases. This unique system potentially allows for multiple targets to be engineered using a single plasmid system. This avoids the use of multiple different promoters for each gRNA (49, 78, 91).

The development of a CRISPR-Cas gene editing tool using an endogenous system requires the presence of a known CRISPR system within that organism. Due to a large number of bacteria harboring CRISPR-Cas systems, there is great potential for the development of this tool in nonmodel organisms (24). For those organisms which have previously identified or well-studied endogenous CRISPR systems, the development for genetic modification is streamlined. For those that do not have characterized systems, there is work to be done prior to developing the genetic tool. The type/class of CRISPR array needs to be determined, and this will help determine how the system will function. Another vital part of developing an endogenous CRISPR system is identifying the PAM recognition sequence for the encoded endonuclease. Once the CRISPR array and PAM are identified, the endogenous CRISPR system could be exploited for genome editing (49, 91). Unfortunately, different strains of an organism might harbor slightly different components to their CRISPR systems and, therefore, may have different requirements for editing (e.g., PAM recognition). This could be seen as a drawback for exploiting the endogenous CRISPR locus, but the benefits could outweigh the long development time to establish other genetic systems. Because the majority of clostridia have endogenous CRISPR systems, these could be exploited for use as genetic tools (49, 92). In particular, those bacteria which have few to no genetic tools available are good candidates for using this type of genetic engineering, since it relies on existing components of the genome, provided they can easily be transformed or have good conjugation systems.

FUTURE DIRECTIONS AND CONCLUSIONS

In the past few years, the CRISPR-Cas systems have become a fast-growing and beneficial contribution to the Clostridium field as genetic tools. These tools have the potential to be applied to all Clostridium species (as well as most other bacteria). While significant strides have been made toward developing CRISPR-Cas systems, there is a lot to be learned about the basic biology of CRISPR-Cas systems as well as the endogenous systems within clostridia in order to develop genetic tools.

A main consideration in clostridia is ensuring tight regulation of Cas9, or any other endonuclease, in order to overcome any potential toxic side effects. The use of a strong regulated promoter that drives the expression of each Cas9 and the sgRNA seems to be key to having a successful CRISPR-Cas genetic tool. In particular, the small RNA promoter from C. beijerinckii has been successfully used in four other clostridia to regulate expression of the sgRNA (49–51, 55, 57, 59, 61, 62). It is possible that this promoter could be efficient in other clostridia.

Application of the CRISPR-Cas9 system to other organisms may require some optimization. As observed in C. saccharoperbutylacetonicum, the C. beijerinckii CRISPR-Cas9 system, developed by Wang et al., was directly applied to this organism and was successful (50, 57). Unfortunately, there is probably not a “universal” CRISPR-Cas system for use in clostridia, because not all promoters or plasmid replicons will likely work in every organism. The future of using this system in other clostridia would be to first start with a preexisting system of a closely related Clostridium species and then modify this system as needed. Should this strategy not prove fruitful, the exploitation of the endogenous CRISPR-Cas system from the organism of interest is a viable option. With the successful application in C. pasteurianum and C. tyrobutyricum, these systems work well and with high efficiency (49, 91).

Another aspect of the system which will need to be worked on is understanding what makes a quality sgRNA, i.e., what makes one sgRNA result in a higher mutation efficiency than another sgRNA. The “rules” for designing sgRNAs developed for eukaryotes, or even other prokaryotes, may not be applicable to clostridia. Currently, many studies have designed multiple sgRNAs with different target sequences in order to see which will result in a mutant at a high efficiency. Elucidating a set of rules for selecting target sequences will aid in more rapid gene editing.

It was previously demonstrated that a system can be made where dCas9 can be coupled to an activator or repressor to regulate the transcription of a specific gene (Fig. 2C) (67, 93). While it has been demonstrated that CRISPRi works effectively in some clostridial species, it would be interesting to apply a system which uses a transcriptional activator coupled to dCas9 to enhance the transcription of specific genes, for example, for the fine-tuning of biofuel production and increasing the use of clostridia to generate valuable end products.

Overall, several labs have laid the groundwork for developing CRISPR-Cas systems as genetic tools (i.e., Cas9, Cas9n, dCas9, Cpf1/Cas12a, and endogenous systems) in clostridia. The wide range of mutations and regulatory control that can be made with these systems has already proved beneficial, and the future use of these technologies is promising.