ABSTRACT
CRISPR-Cas systems are prokaryotic adaptive immune systems that facilitate protection of bacteria and archaea against infection by external mobile genetic elements. The model pathogen Francisella novicida encodes a CRISPR-Cas12a (FnoCas12a) system and a CRISPR-Cas9 (FnoCas9) system, the latter of which has an additional and noncanonical function in bacterial virulence. Here, we investigated and compared the functional roles of the FnoCas12a and FnoCas9 systems in transformation inhibition and bacterial virulence. Unlike FnoCas9, FnoCas12a was not required for F. novicida virulence. However, both systems were highly effective at plasmid restriction and acted independently of each other. We further identified a critical protospacer-adjacent motif (PAM) necessary for transformation inhibition by FnoCas12a, demonstrating a greater flexibility for target identification by FnoCas12a than previously appreciated and a specificity that is distinct from that of FnoCas9. The effectors of the two systems exhibited different patterns of expression at the mRNA level, suggesting that they may confer distinct benefits to the bacterium in diverse environments. These data suggest that due to the differences between the two CRISPR-Cas systems, together they may provide F. novicida with a more comprehensive defense against foreign nucleic acids. Finally, we demonstrated that the FnoCas12a and FnoCas9 machineries could be simultaneously engineered to restrict the same nonnative target, thereby expanding the toolset for prokaryotic genome manipulation.
IMPORTANCE CRISPR-Cas9 and CRISPR-Cas12a systems have been widely commandeered for genome engineering. However, they originate in prokaryotes, where they function as adaptive immune systems. The details of this activity and relationship between these systems within native host organisms have been minimally explored. The human pathogen Francisella novicida contains both of these systems, with the Cas9 system also exhibiting a second activity, modulating virulence through transcriptional regulation. We compared and contrasted the ability of these two systems to control virulence and restrict DNA within their native host bacterium, highlighting differences and similarities in these two functions. Collectively, our results indicate that these two distinct and reprogrammable endogenous systems provide F. novicida with a more comprehensive defense against mobile genetic elements.
INTRODUCTION
CRISPR (clustered regularly interspaced short palindromic repeats)-Cas (CRISPR-associated) systems are prokaryotic adaptive immune systems that protect against mobile genetic elements like bacteriophages and plasmids (1, 2). These conserved genomic loci contain a CRISPR array and adjacent cas genes (3–5). The CRISPR array consists of alternating repetitive sequences (repeats), interspaced by unique spacer sequences that often correspond with sequences of foreign nucleic acid targets, and these spacer targets are referred to as protospacers (6–8).
Upon introduction of a foreign nucleic acid, a part of the invading sequence is selected, processed, and integrated into the CRISPR array with flanking repeat sequences (9). To protect against subsequent infection by a nucleic acid that shares the same sequence, the CRISPR array is transcribed and processed into individual CRISPR RNAs (crRNAs), each comprised of one spacer and part, if not all, of the repeat sequence. The crRNAs form complexes with Cas protein(s), and when the spacer sequence of the crRNA binds to the protospacer in the target, the associated Cas protein(s) interferes with the nucleic acid target (10–12).
CRISPR-Cas systems are mechanistically diverse and can be divided into multiple classes (5, 13). Class I systems require a multiprotein complex for target recognition and cleavage, while class II systems consist of a single effector protein for these processes (13). Both can be further subdivided into types based on phylogeny, gene clustering, and associated RNAs (13). The class II system that uses the Cas9 protein for DNA recognition and cleavage has been extensively characterized, and the relative simplicity of a single effector protein has enabled the development of revolutionary Cas9-derived tools for genetic engineering (10, 14, 15). Other systems with single effector proteins have been discovered that have critical differences compared to Cas9 (16–18). Of these systems, the CRISPR-Cas12a system (also known as Cpf1) has been the focus of notable research due to unique characteristics that benefit genome engineering applications. Distinctions of Cas12a include processing of its own crRNAs by the Cas12a protein, use of a single crRNA and no accessory RNAs to cleave DNA targets, and production of staggered double-strand breaks (19–21). Despite the extensive study of these Cas9 and Cas12a CRISPR systems in nonnative cell types and for engineering, few of these functions have been studied in their native bacterial contexts, which is important for understanding the natural functions and nuanced biology of these widespread prokaryotic defense systems.
Interestingly, the intracellular pathogen Francisella novicida encodes both a CRISPR-Cas9 (FnoCas9) and a CRISPR-Cas12a (FnoCas12a) system, facilitating the study of how these two systems operate concurrently (22). In addition, F. novicida is a model for native functions of Cas9 beyond DNA cleavage, such as its role in virulence (23, 24). Cas9 transcriptionally represses four immunostimulatory putative bacterial lipoproteins in F. novicida, promoting evasion of the host innate immune response and thus disease pathogenesis (24–26). Transcriptional repression by Cas9 occurs via scaRNA, a non-crRNA small RNA, which binds to the bacterial chromosome in multiple locations with partial complementarity to the DNA (24, 27). The incomplete complementarity between the scaRNA and the endogenous DNA targets fails to promote the cleavage activity of Cas9, which would otherwise induce a lethal break in the chromosome at the targeted sites (24). The RNA, rather than protein, directed nature of this second function suggests that if other RNAs of a similar composition are present elsewhere, this function may exist beyond this specific strain-system combination.
Although the role of FnoCas9 in virulence has been established, the role of FnoCas9 and FnoCas12a in DNA defense and the role of FnoCas12a in virulence have yet to be explored in a native bacterial host. Both systems in F. novicida encode CRISPR repeats targeting various sites throughout the same putative F. novicida prophage genome, suggesting that both are capable of DNA defense functions. The prophage has only been found in strains lacking complementary spacers, suggesting functional in vivo activity of these systems in bacteriophage defense (22). The cas12a-containing CRISPR-Cas loci are conserved in diverse bacterial species (e.g., Prevotella, Flavobacterium) and are found almost exclusively in mammalian host-associated (commensal and pathogenic) bacteria, similar to the ecological distribution of CRISPR-Cas9 systems (5, 16, 22, 27–30). Interestingly, unlike CRISPR-Cas9 systems that are unique to bacteria, Cas12a-associated systems are also found in archaea and have both DNase and RNase activity, allowing DNA targeting and crRNA processing by the Cas12a protein alone (5, 19).
Herein, we demonstrate that the CRISPR-Cas12a system defends against foreign DNA invasion within its native bacterial host. Since F. novicida contains both CRISPR-Cas9 and -Cas12a systems, it provides a unique opportunity to explore the endogenous functions of two class II systems in relation to each other. We compared their physiological activities, revealing that their functions in DNA defense were independent and that each effector protein exhibited a different pattern of expression, providing the first description of how two class II systems function relative to one another in the same bacterium. Conversely, we found that only FnoCas9 was important for F. novicida virulence, while a loss of FnoCas12a did not alter the pathogenesis of the bacterium in a mammalian host. To better elucidate the functional differences between these two systems in DNA cleavage, we studied the protospacer adjacent motif (PAM), which discriminates between self and nonself DNA (6, 31–33), finding that the FnoCas12a PAM has underappreciated flexibility. We further compared the baseline efficiency of the F. novicida Cas12a and Cas9 systems in DNA defense by engineering the CRISPR arrays for the same, nonnative target and found that they protected the bacterium with remarkably similar efficacy. This similarity suggests that functional differences between FnoCas12a and FnoCas9 are not in their baseline ability to restrict infection or the molecular differences in their DNA cleavage mechanism but rather in other aspects of their function, including their regulation and distinct PAM requirements. These findings indicate that multiple class II CRISPR-Cas systems can exist in a bacterium to provide both robust and comprehensive protection from mobile genetic elements and that they can be further utilized to modulate bacterial physiology through the programmed targeting of DNA for cleavage.
RESULTS
FnoCas9 contributes to F. novicida virulence independently of FnoCas12a.Given the function of FnoCas9 in F. novicida virulence, we first characterized the role of FnoCas9 and FnoCas12a during infection, together and independently. Mice were infected subcutaneously, and the bacterial burden in the skin and spleen was measured 48 h postinfection (Fig. 1A and B). Wild-type (WT) F. novicida established robust infection in both organs, while a Δcas9 mutant was highly attenuated. Previous work demonstrated that complementation of cas9 into a Δcas9 mutant restores virulence to WT levels in survival experiments (27). Further, since the virulence defect of the Δcas9 mutant was due to the increased expression of FnoCas9-regulated genes, virulence was also recovered when the Cas9-regulated genes were deleted from the Δcas9 mutant (24). Mutation of cas12a from WT F. novicida had no effect on the ability of the pathogen to establish infection nor was a Δcas9 Δcas12a double mutant further attenuated from the levels of the Δcas9 strain in vivo. These results are the first to indicate a divergence in the functional roles of these two systems in their native contexts, with FnoCas9 exhibiting an alternative function in virulence, while FnoCas12a did not alter bacterial pathogenesis in vivo.
Cas9 contributes to F. novicida virulence independently of Cas12a. CFU burden in mouse skin (A) and spleens (B) 48 h after infection with U112 (WT), Δcas12a, Δcas9, and Δcas9 Δcas12a (n = 5; bars represent geometric means; *, P ≤ 0.05; **, P ≤ 0.005). n.s., not significant.
Endogenous F. novicida CRISPR systems function independently in DNA defense.The first indication that FnoCas9 and FnoCas12a protect F. novicida against external DNA was in the spacer sequences in the CRISPR arrays. The F. novicida U112 cas12a-associated CRISPR array contains 9 spacers, two of which target a putative prophage that was identified in F. novicida strain 3523, also targeted by the F. novicida U112 CRISPR-Cas9 system (22). To determine if F. novicida U112 is resistant to infection by protospacer-containing sequences, and in the absence of viable bacteriophage to use in such experiments, we performed a plasmid inhibition transformation assay to mimic infection by a mobile genetic element. Wild-type F. novicida was transformed with an empty vector control plasmid or a plasmid containing a 50-bp sequence from the putative prophage encompassing the 30-bp protospacer and 10 bases on each side (referred to as the “Cas12a_target_5” plasmid, as it corresponds to the 5th spacer within the Cas12a-CRISPR array) inserted into a nontranscribed portion of the plasmid (Fig. 2A). To avoid potential complications of copy number, both the target plasmid and the control plasmid were made in the same broad-host-range vector (pBAV1K-T5-GFP; pBAV). Wild-type bacteria resisted transformation with the Cas12a_target_5 plasmid while remaining permissive to transformation by the control plasmid.
Cas12a and Cas9 have distinct targets that they inhibit with similar efficiencies. (A) Wild-type F. novicida, Δcas9, and Δcas12a were transformed with the control and the Cas12a_target_5 plasmid. (B) Wild-type F. novicida, Δcas9, and Δcas12a were transformed with the control and a Cas9_target_13 plasmid that contains a protospacer that is complementary to spacer 13 from the cas9-associated CRISPR array. Both Cas9 and Cas12a inhibit transformation by spacers in their respective loci with similar efficiencies, and inhibition occurs in the absence of the effector from the other locus (bars represent standard error [SE]; **, P ≤ 0.005; ***, P ≤ 0.0001).
We next tested whether there was interdependency between the endogenous F. novicida Cas9 and Cas12a systems in DNA cleavage by measuring the ability of Cas12a to inhibit transformation with a Cas12a_target_5 plasmid in the presence and absence of Cas9. F. novicida restricted transformation with the Cas12a_target_5 plasmids while remaining permissive to transformation with the control, while the Δcas12a strain was permissive to transformation with both plasmids. However, when a Δcas9 strain was transformed with the Cas12a_target_5 plasmid, it was able to restrict transformation with this plasmid, suggesting that Cas12a-dependent targeting occurs independently of Cas9 activity (Fig. 2A). We next tested the independence of the Cas9 system of F. novicida in DNA cleavage. Transformation inhibition assays were performed with a Cas9_target_13 plasmid. This spacer was used in previous studies as a control for the cleavage activity of Cas9 (24, 34). The crRNA containing this spacer is more abundant than crRNAs produced from the middle of the array based on RNA sequencing (24). The Cas9_target_13 plasmid contained the protospacer for spacer no. 13 of the CRISPR-Cas9 array flanked by a PAM sequence that had previously been determined for F. novicida Cas9 by in vitro and in vivo DNA targeting assays (30, 34, 35). We observed that F. novicida could inhibit transformation by the Cas9_target_13 plasmid and that this inhibition was dependent on an intact Cas9 (Fig. 2B). Interestingly, F. novicida inhibited transformation by both Cas12a- and Cas9-targeted plasmids by approximately 3.5 logs, suggesting that in these transformation conditions, both proteins exhibited similar abilities to restrict foreign DNA (Fig. 2). This is the first examination of plasmid restriction by two distinct class II CRISPR systems within the same native host, indicating that restriction occurs efficiently by both effector proteins and with neither protein interfering with or improving the ability of the other system to restrict DNA. Thus, the presence of these two systems provides F. novicida with dual defenses against invading nucleic acid threats.
Cas12a exhibits PAM promiscuity in native host.The ability of F. novicida to protect against transformation by DNA sequences targeted by Cas12a-associated spacers led us to investigate the flexibility of the PAM sequence in the CRISPR-Cas12a system in vivo (2, 36–38). In vitro structural and heterologous expression studies have identified that the FnoCas12a spacer sequence required for plasmid interference is 5′-TN-3′ (20). To identify the preferred PAM of FnoCas12a within Francisella, we aligned the 5′ and 3′ flanking nucleotides of putative prophage regions that are complementary to 9 unique spacers found in different strains of Francisella (22). This analysis predicted a 5′-TTTN-3′ PAM immediately upstream of the matching target sequence (Fig. 3A).
Cas12a requires a 5′ protospacer-adjacent motif (PAM). (A) Sequence logo plot for the 5′ flanking sequence of 9 protospacers complementary to 9 unique spacers from different strains of Francisella. Letter height represents the nucleotide frequency at each position relative to the protospacer for these samples. (B) 5′ sequence alignment for PAM mutations made in the Cas12a_Target_5 plasmid (a, n = 16; b, n = 12; c, n = 9; d, n = 5; e, n = 11; bar represent SE). Bases that have been mutated from the predicted PAM are underlined and bold. (C) Wild-type F. novicida U112 and a Δcas12a mutant were transformed with the control (n = 17) and Cas12a_Target_5 (n = 19) plasmids and the PAM derivatives of the Cas12a_Target_5 plasmid in panel B. Significant differences in the transformation efficiency between each PAM mutant plasmid and both the control plasmid (red) and the Cpf1_target_5 plasmid (blue) were determined using a Mann-Whitney test. *, P ≤ 0.05; **, P ≤ 0.005; ***, P ≤ 0.0001.
Given the high AT content of this PAM and the high AT content of Francisella spp., we investigated the stringency of FnoCas12a PAM requirements in vivo (39). We determined whether each of the three T residues within the putative PAM was required for target recognition by conducting a transformation assay with point mutants in the predicted PAM sequence of the Cas12a_target_5 plasmid. We compared the transformation efficiencies with each PAM variant using the ratio of CFU recovered from the Δcas12a mutant to that of the CFU recovered from the wild type. We mutated the −2 position T (5′ to the protospacer) to a C in the Cas12a_target_5 plasmid, resulting in a 5′-TTCN mutant PAM (Fig. 3B, construct a). Targeting of the 5′-TTCN plasmid was largely abrogated compared to the Cas12a_target_5 plasmid (5′-TTTN PAM), indicating that the −2 T is important for target recognition. However, this mutation did not fully recover the transformation of the plasmid to the level of the untargeted control plasmid, suggesting that a 5′-TN did not account for the full PAM sequence in vivo (Fig. 3C). There was also a 24-h delay in growth of the transformants with the 5′-TTCN mutant PAM, further suggesting either a partial targeting event or an increased level of mutations that enable escape from targeting compared to that observed with the other constructs. Conversely, mutation of the −3 position T to C (construct b) did not alter plasmid targeting efficiency compared to that of the Cas12a_target_5 plasmid, indicating that this residue is nonessential for targeting (Fig. 3B and C). We then tested a target plasmid harboring a −4 T to C mutation (construct c). Targeting of this plasmid was also similarly efficient compared to that of the Cas12a_target_5 plasmid encoding the 5′-TTTN PAM, suggesting that the −4 PAM residue does not play a large independent role in DNA recognition (Fig. 3B and C). The combined mutation of the −3 and −4 position TT residues to CC yielded a 5′-CCTN mutant PAM (construct d), which unlike the −4 T to C (construct c) or −3 T to C (construct b) mutations alone, significantly reduced the targeting of the plasmid compared to that of the Cas12a_target_5 plasmid (Fig. 3B and C). The 5′-CCTN (construct d) mutation was still inhibited in transformation efficiency compared to that of the control plasmid, consistent with the demonstrated role of the −2 position T demonstrated with construct a. This suggests that the −3 and −4 nucleotides together play a role in improving target recognition and do so in tandem with the −2 position base (Fig. 3B and C, constructs a to d). Because each of the −2 to −4 position bases in the predicted 5′-TTTN PAM had a partial contribution to plasmid targeting, we generated a protospacer construct with the 5′-TTTN PAM nucleotides mutated to 5′-CCCN (Fig. 3B, construct e). The 5′-CCCN plasmid was not targeted effectively, and transformation efficiency was recovered to almost that of the control plasmid (Fig. 3C). From these data, we conclude that F. novicida Cas12a requires a 5′-TTTN PAM and that the three T residues play an additive role in target DNA inhibition. This is distinct from the PAM required for the other class II CRISPR system in F. novicida, Cas9, which requires a 3′-NGG, highlighting the mechanistic differences between these systems that together provide more comprehensive protection from mobile genetic elements (30, 34, 35).
Cas12a and Cas9 exhibit different trends in relative mRNA level during transformation.Due to the functional redundancy in DNA cleavage and distinct differences in PAM requirements of the F. novicida Cas9 and Cas12a systems, we considered the possibility that the two CRISPR systems are specialized to protect F. novicida from mobile genetic elements that arise in different environmental niches. F. novicida likely utilizes adaptive immunity in the soil, brackish water, and mammalian hosts, among other environments. To test whether the two systems can exhibit different patterns of expression, we quantified the mRNA levels of cas9 and cas12a during the transformation conditions used in plasmid interference experiments relative to the starting culture of log-phase F. novicida U112 (optical density at 600 nm [OD600], 1.0) (competent cells prior to transformation, transformation, and during recovery from transformation in growth medium) (Fig. 4). Interestingly, we observed that both cas12a and cas9 mRNA increased similarly between the starting culture and the competent cell conditions. However, during transformation, cas9 mRNA levels decreased toward the level of the starting culture, while the cas12a mRNA levels remained at the competent cell level. At the recovery stage of transformation, the cas9 mRNA levels dropped further to below the levels of the starting culture, while the cas12a mRNA levels did not change. These results suggest that the F. novicida CRISPR-Cas systems may be regulated to provide protection in different environments and that the CRISPR-Cas systems’ mechanistic differences may specialize them for threats specific to those different conditions.
cas12a and cas9 exhibit different trends in relative mRNA level during transformation. Relative mRNA levels of both cas12a (blue) and cas9 (red) were examined by qRT-PCR (quantitative reverse transcription-PCR) during plasmid transformation conditions in the starting culture, competent cells, transformation, and after recovery. Each experiment was normalized to a housekeeping gene and the mRNA level in the starting culture (OD600, 1.0), resulting in the fold change in mRNA level between conditions. n = 11 to 15; error bars represent SE; *, P ≤ 0.05.
Cas9 and Cas12a have the same baseline ability to restrict foreign DNA in F. novicida.We next sought to directly compare the efficiency of plasmid restriction of the F. novicida CRISPR systems, without the variables of the competing spacers in the CRISPR arrays or the sequence of the spacer, by using both systems to target the same nonnative target sequence. The PAM sequences for Cas12a and Cas9 are located on opposing sides of the protospacers, allowing for the same sequence to be targeted by both systems. Each CRISPR array was replaced with a single nonnative spacer flanked on both sides by the complete crRNA repeat sequence of the respective system. The two CRISPR systems were otherwise unmodified, with the repeat-spacer-repeat sequences placed downstream of the native Cas9 or Cas12a CRISPR array promoters (Fig. 5A). A strain with the Cas12a CRISPR array engineered for the new target and an intact WT Cas9 system and a strain with Cas9 engineered for the new target but with an intact WT Cas12a system were transformed with a plasmid containing a protospacer target for the engineered CRISPR arrays (Fig. 5B). In this vector, called the “Cas12a&Cas9_target” plasmid, the protospacer was flanked by the PAM for both CRISPR systems to provide a universal target plasmid.
Cas9 and Cas12a exhibit similar endogenous DNA targeting efficiencies when engineered for the same artificial target. (A) Schematic of a CRISPR array and the engineered Cas9 and Cas12a CRISPR loci. In the engineered loci, the ΔcrRNA was complemented with a repeat-(nonnative spacer)-repeat, using repeats specific for each CRISPR system. The engineered Cas12a spacer is transcribed and processed into a mature crRNA that interacts with Cas12a independently of accessory RNAs. The engineered Cas9 spacer is transcribed and processed into a mature crRNA that interacts with the tracrRNA, and the resulting RNA duplex binds to Cas9 and guides the effector to its target. (B) Wild-type F. novicida U112, a strain with Cas12a engineered, a strain with Cas9 engineered, and a strain with both Cas12a and Cas9 loci engineered for the same nonnative spacer were transformed with control and Cas12a&Cas9_Target plasmids. The percentage of colonies escaping targeting was calculated relative to the untargeted control plasmid (n = 7 to 9; bars represent SE; ***, P ≤ 0.0001).
Reprogramming both the Cas12a and Cas9 systems for the new target successfully restricted transformation with the target plasmid to just above the limit of detection of the assay (Fig. 5B). We then engineered both CRISPR loci for the new target in the same strain and evaluated the ability to restrict transformation with the target plasmid, finding that this strain also restricted transformation with high efficiency (Fig. 5B). This first controlled comparison of two class II CRISPR system activities in their native bacterial host suggests that in spite of mechanistic differences, in the absence of competing crRNAs, the Cas9 and Cas12a CRISPR systems of F. novicida have remarkably similar abilities to restrict infection with a spacer-encoded target with an almost undetectable level of plasmid escape.
DISCUSSION
F. novicida harbors two distinct endogenous class II CRISPR systems, CRISPR-Cas12a and CRISPR-Cas9, providing a unique model for studying natural class II CRISPR system functions relative to one another (22). FnoCas9 has a described function distinct from DNA defense, repressing four endogenous genes to enable virulence and providing a model and mechanism for noncanonical CRISPR-Cas9 roles in bacterial physiology (24, 40). Not only does the FnoCas9 system enable virulence in a mammalian host but also the native presence of two CRISPR-Cas systems in this bacterium represents a two-pronged CRISPR defense; both systems contain spacers that target the same putative prophage, which suggests that there is a fitness benefit to retaining the two DNA targeting class II systems (22). We demonstrated that these two systems inhibit transformation with their respective targets with similar efficiencies and that they do so independently (Fig. 2). Conversely, FnoCas9, but not FnoCas12a, is involved in virulence in a murine model of infection, highlighting a clear delineation in their endogenous functions (Fig. 1).
Furthermore, we show that both Cas12a and Cas9 CRISPR systems can be engineered for a new target. Upon reprogramming the CRISPR array, in the absence of competing variables, both systems restrict transformation with a target plasmid at the same efficiency, with an extremely low frequency of plasmid escape (Fig. 5). When both systems are reprogrammed for the same target in a single strain, the efficiency of plasmid restriction is not increased above that of each individual system (Fig. 5). This suggests that there may not be significant importance to having the second system available to compensate for possible PAM mutations in the target that would prevent recognition by one of the systems in the conditions that we tested. Therefore, the repurposing of both systems for different targets within a bacterium could provide maximally efficient protection from two sequences. These results indicate that with simple genome manipulations, bacteria and archaea encoding CRISPR-Cas12a and/or Cas9 systems can be reprogrammed to target invading and emerging nucleic acid threats that are not already targeted by complementary sequences in the native spacer array.
In spite of their baseline similarities in DNA restriction upon reprogramming, we identify key differences between the two systems that allow them to provide F. novicida with more comprehensive protection from nucleic acid threats. We show that the nonoverlapping PAM requirements of Cas9 and Cas12a and the flexibility of the Cas12a PAM provide a level of target diversity that far exceeds that of either system individually (Fig. 3 and 5). Likewise, we show that the two systems follow distinct patterns of effector protein mRNA level over the course of transformation (Fig. 4). We hypothesize that these differences allow the bacterium to take maximal advantage of the mechanistic differences between the two systems for adaptive immunity without expending energy on producing large effector proteins unnecessarily. Additional investigation into the differential regulation of these two CRISPR-Cas systems in the presence of different types of nucleic acid predation may provide insight into the unique specializations of these two systems, connecting differences in expression to differences in function.
There are differences in the molecular DNA cleavage mechanisms of Cas9 and Cas12a, which include inducing different double-stranded breaks in their targets, crRNA processing, and presence of a tracrRNA (trans-activating crRNA) (which hybridizes into an RNA duplex with the crRNA and is necessary for binding to Cas9) (19, 20, 41). These distinctions likely result in functional differences in exogenous DNA protection. We confirmed and interrogated aspects of the FnoCas12a mechanism in the native bacterial host, a context they had not been examined in previously. However, due to the lack of a viable phage for Francisella, we are unable to test how these differences effect adaptive viral immunity in F. novicida. Nevertheless, we can make informed predictions based on the composition of the CRISPR arrays about the Cas9 and Cas12a system functions. Given the presence of spacers for the same putative prophage in both CRISPR arrays, both systems can likely protect against overlapping invading elements (22). This suggests that the conditions in which each system is expressed and differences in their target selection (PAM) and adaptation requirements likely play an important role in their differences in DNA defense. This therefore expands the important functional differences in the activities of these systems beyond their distinct molecular mechanisms of crRNA processing and DNA cleavage.
Interestingly, unlike Cas9, the FnoCas12a PAM is located immediately to the 5′ of the protospacer (20, 42). Because of this, the same highly virulent sequence could theoretically be targeted by both the Cas9 and Cas12a systems to increase protection of the bacterium. We demonstrated this by reprogramming both systems in the same strain for the same nonnative target and transforming with a plasmid containing the protospacer flanked by the PAMs for both systems (Fig. 5). Interestingly, this did not increase the efficiency of protection from the strains with individually reprogrammed loci, suggesting that the difference in PAM orientation is the product of the cleavage mechanism of each system, which evolved independently of the sustained presence of the other system.
However, one benefit of two similarly functional class II systems is the recognition diversity that is enabled by two Cas effectors with distinct PAMs. In vitro and heterologous expression FnoCas12a studies have suggested that 5′-TN is the necessary PAM sequence (19, 20). We predicted that the preferred PAM for FnoCas12a is 5′-TTTN PAM using bioinformatics analyses and further interrogated the PAM requirements of FnoCas12a by evaluating its ability to restrict plasmids with systematic mutation in the 5′-TTTN PAM sequence while in its natural bacterial host. We found that FnoCas12a actually exhibits a high level of flexibility in the PAM sequences that it recognizes. We observed that of the three Ts in the PAM sequence, the −2 position T is the most important. However, mutation of this base is not sufficient to completely prevent DNA targeting (Fig. 3C). Interestingly, the −4 T and −3 T were independently dispensable for target DNA recognition and targeting in vivo but together significantly improved DNA targeting (Fig. 3C). We observe that a 5′-TTTN is the preferred PAM sequence in vivo, with the 5′-CCCN mutant PAM having the largest effect on abrogating target plasmid inhibition (Fig. 3C). This differs from the in vitro cleavage data for FnoCas12a, which suggest a 5′-TN PAM, more closely resembling the 5′-TTTN PAM of AsCas12a and the 5′-TTTN PAM that we predicted based on Fig. 3A (43). These results highlight the importance of studying CRISPR-Cas systems within their native bacterial hosts in addition to through heterologous expression and in vitro models. Not only have we identified differences in the requirements of the Cas12a system for target recognition, but we have demonstrated the importance of studying their interactions and functional characteristics for understanding their relative contributions to the immune system of a bacterial pathogen. These differences and comparisons are critical for both understanding basic bacterial physiology and expanding the existing CRISPR toolset.
The engineering of native CRISPR-Cas systems in bacteria, primarily Cas9-dependent systems, is an evolving tool to modify the bacterial genome, combat antibiotic resistance and virulence, and manipulate the population balance in complex environments (36, 38, 44–51). The diverse applications of engineering bacterial systems represent one of the new frontiers of CRISPR-Cas technologies, and Cas12a may provide an ideal alternative for these tools. Studies have shown that the reprogramming of native and heterologously expressed CRISPR-Cas systems can be used to make targeted mutations in the bacterial chromosome, conduct genetic screens, or alter transcription of a gene (36, 44, 46, 49–56). Cas12a and Cas9 are found primarily in mammalian commensals and pathogens and provide tools for making highly relevant microbes tractable for reverse genetic studies. Likewise, in both native and nonnative bacterial hosts, Cas12a and Cas9 systems have additional applications in human health and agriculture as an antimicrobial and antivirulence tool (46, 57, 58). The ability to transfer functional CRISPR-Cas systems between diverse bacteria and the lethality of CRISPR-Cas targeting of the bacterial genome has led to the highly sensitive removal of specific bacteria from mixed cultures (46, 48). Similarly, CRISPR systems can be used to selectively alter the fitness of specific organisms, for example, through the conjugated targeting of pathogenic strains as well as emerging virulence factors and antibiotic resistance cassettes (37, 46). The reprogramming of multiple systems for new targets could be combined, native or supplemented into a prokaryote of interest, to effectively avoid resistance mutations or anti-CRISPR proteins that exist for one CRISPR-Cas system (59–61). Therefore, the use of multiple single-effector CRISPR systems within a host provides a more comprehensive toolset, much like how naturally in F. novicida, as demonstrated herein, the presence of both Cas12a and Cas9 provides a more comprehensive DNA defense program.
MATERIALS AND METHODS
Bacterial strains and growth conditions.All strains and plasmids can be found in Table S1 in the supplemental material. All strains of Francisella novicida U112 were grown at 37°C, shaking, in tryptic soy broth (TSB) (VWR International, Inc.) supplemented with 0.2% cysteine (BD Biosciences, Sparks, MD). All strains were plated on tryptic soy agar (TSA) (VWR International, Inc.) plates supplemented with 0.1% cysteine. Escherichia coli DH5α was grown in tryptic soy broth and plated on tryptic soy agar. Kanamycin sulfate (30 μg/ml; Fisher Scientific Company) was used for selection for both liquid cultures and solid agar.
Mouse infections.Specific-pathogen-free female C57BL/6J mice (Jackson) at 8 weeks old were infected subcutaneously with ∼2 × 105 CFU bacteria. Skin and spleens were harvested at 48 h postinfection, homogenized in phosphate-buffered saline (PBS), and the bacterial burden per organ was determined as described by Ratner et al. (24). Animals were cared for as described by Ratner et al., and the Emory University Institutional Animal Care and Use Committee (protocol number YER-2000573-061314BN) approved all procedures (24, 27).
PAM prediction.Francisella sp. CRISPR-Cas12a spacers that target a putative prophage in F. novicida U112 3523 were confirmed using BLASTn (https://blast.ncbi.nlm.nih.gov/Blast.cgi?PAGE_TYPE=BlastSearch). The protospacer was identified by the location of the hairpin secondary structure in the CRISPR repeat relative to the spacer for each strain, and the sequence flanking both sides of the protospacer was identified by BLASTn. The flanking sequences for all Francisella sp. 3523 protospacers were compiled, and the relative nucleotide contribution of each base at each position in these 5′ and 3′ sequences was determined using a sequence logo plot (https://weblogo.berkeley.edu/logo.cgi), indicated by relative letter height.
Francisella novicida strain construction.Null deletion and point mutants listed in Table S1 were constructed by allelic exchange using the primers in Table S2 in the supplemental material. Homologous sequences (500 to 1,000 bp) up and downstream of the region of interest were amplified by PCR from genomic DNA of F. novicida U112, using Phusion high-fidelity DNA polymerase (New England Biolabs). These fragments were used to construct the allelic exchange substrate, and a kanamycin selectable marker containing Flp recombinase target (FRT) sites was inserted between the flanking sequences using overlapping PCR, as previously described (62). Linear fragments were transformed into chemically competent F. novicida, and mutants were selected on kanamycin plates. DNA was isolated using QIAquick tissue extraction kit (Qiagen, Inc.), and mutants were confirmed by PCR amplification from outside of the recombined region followed by sequencing (Genewiz). Strains were unmarked using a temperature-sensitive suicide vector, pFFlp, carrying the Flp recombinase in trans, as described previously (63).
Plasmid manipulations.Plasmids were transformed and isolated from competent E. coli (NEB 5-alpha) using Zyppy miniprep, midiprep, and maxiprep kits (Zymo Research). The broad-host-range vector pBAV1K-T5-GFP (pBAV) was used as the control plasmid and backbone for the target plasmids in all assays (64). Overlapping PCR was used to construct Cas12a_target and Cas9_target plasmids and the panel of PAM mutants in the Cas12a_target_5 plasmid (see Tables S1 and S2). In the target plasmids, the prophage region and/or protospacer sequence was inserted immediately downstream of the EcoRI site of the pBAV backbone with the primers indicated in Table S1.
Plasmid inhibition transformation assays.Overnight starting cultures of F. novicida were used to make competent cells by diluting them 1:100 and growing them to an OD600 of 1.0 to 1.2. To make competent cells, cultures were concentrated 10-fold in 4°C chemical transformation buffer (CTB) and incubated at 4°C for 15 min (27). Competent cells (300 μl) were incubated with plasmid DNA (500 ng or 1,000 ng) for 25 min with shaking at 37°C before recovering in 1 ml of TSB plus 0.2% cysteine for 2.5 h (shaking at 37°C). Transformations were plated on TSA plates with kanamycin selection and incubated at 37°C overnight, and CFU were enumerated for each transformation. The data were normalized to the transformation efficiency of the positive control (transformation with the control plasmid or transformation into Δcas12a [Fig. 5B and Fig. 3C, respectively]) and presented as proportion or percent. The data in Fig. 2 are presented as transformants per 100 ng of plasmid. A no-DNA transformation was used as a negative control.
Quantitative real-time PCR.RNA was isolated over the course of the plasmid inhibition transformation assay (at an OD600 of 1.0, competent cells, transformation, recovery [45 min], recovery [120 min]) using TRI Reagent and a Direct-zol RNA miniprep kit (Zymo Research, Irvine, CA). Wild-type F. novicida was grown to late log phase (OD600 of 1.0). This starting culture was concentrated 10-fold in 4°C chemical transformation buffer (CTB) and stored at 4°C for 20 min (competent cells). For each transformation, 300 μl of competent cells was shaken at 37°C for 25 min (transformation). After 25 min, 1 ml of F. novicida medium (TSB plus 0.2% Cys) was added to each transformation and shaken at 37°C for 45 min (recovery) and 120 min (plating). Transformations were plated on selective media at 120 minutes. DNA was removed from the samples using Turbo DNase I (Ambion Biosciences). A Power Sybr green RNA-to-CT 1-step kit (Applied Biosystems) was used to generate cDNA and conduct the quantitative reverse transcription-PCR (qRT-PCR) using the primers indicated in Table S2. Relative transcript levels were calculated by normalizing threshold cycle (CT) values to the Francisella novicida housekeeping gene DNA helicase II (uvrD) to determine 2−ΔΔCT for each condition (27). Each experiment was normalized to the mRNA level in the starting culture (OD600, 1.0), resulting in the fold change in mRNA between conditions. Results were plotted as fold change in mRNA relative to the starting culture (culture with OD600 of 1.0).
Statistical analysis.For each figure, the Shapiro-Wilk test was used to evaluate the distribution of the data. Significance was determined using the Kruskal-Wallis test followed by a Dunn’s comparison between the groups for comparisons between multiple groups with nonnormal distribution (Fig. 1). For Fig. 2 and 5, the Mann-Whitney test was used to determined significance between pairs of groups with nonnormal distribution.
ACKNOWLEDGMENTS
We thank Siddharth Jaggavarapu and Graeme Conn for thoughtful comments on the manuscript and Chui-Yoke Chin for her insight into assays on this topic. We also thank Timothy R. Sampson for early feedback on this work.
CRISPR work in the lab is supported by National Institutes of Health (NIH) grant R01-AI110701 to David S. Weiss, who is also supported by a Burroughs Wellcome Fund Investigator in the Pathogenesis of Infectious Disease award.
FOOTNOTES
- Received 24 October 2019.
- Accepted 11 March 2020.
- Accepted manuscript posted online 13 April 2020.
Supplemental material is available online only.
- Copyright © 2020 American Society for Microbiology.
