Inducible CRISPRi-Based Operon Silencing and Selective in Trans Gene Complementation in Borrelia burgdorferi

ABSTRACT

To accelerate genetic studies on the Lyme disease pathogen Borrelia burgdorferi, we developed an enhanced CRISPR interference (CRISPRi) approach for isopropyl-β-d-thiogalactopyranoside (IPTG)-inducible repression of specific B. burgdorferi genes. The entire system is encoded on a compact 11-kb shuttle vector plasmid that allows for inducible expression of both the sgRNA module and a nontoxic codon-optimized dCas9 protein. We validated this CRISPRi system by targeting the genes encoding OspA and OspB, abundant surface lipoproteins coexpressed by a single operon, and FlaB, the major subunit forming the periplasmic flagella. As in other systems, single guide RNAs (sgRNAs) complementary to the nontemplate strand were consistently effective in gene repression, with 4- to 994-fold reductions in targeted transcript levels and concomitant reductions of protein levels. Furthermore, we showed that ospAB knockdowns could be selectively complemented in trans for OspA expression via the insertion of CRISPRi-resistant, synonymously or nonsynonymously mutated protospacer adjacent motif (PAM*) ospA alleles into a unique site within the CRISPRi plasmid. Together, this establishes CRISPRi PAM* as a robust new genetic tool to simplify the study of B. burgdorferi genes, bypassing the need for gene disruptions by allelic exchange and avoiding rare codon toxicity from the heterologous expression of dCas9.

IMPORTANCE Borrelia burgdorferi, the spirochetal bacterium causing Lyme disease, is a tick-borne pathogen of global importance. Here, we expand the genetic toolbox for studying B. burgdorferi physiology and pathogenesis by establishing a single plasmid-based, fully inducible, and nontoxic CRISPR interference (CRISPRi) system for transcriptional silencing of B. burgdorferi genes and operons. We also show that alleles of CRISPRi-targeted genes with mutated protospacer-adjacent motif (PAM*) sites are CRISPRi resistant and can be used for simultaneous in trans gene complementation. The CRISPRi PAM* system will streamline the study of essential Borrelia proteins and accelerate investigations into their structure-function relationships.

INTRODUCTION

CRISPR-Cas systems naturally function for prokaryotic adaptive immunity against viruses and plasmids, but their synthetic applications continue to revolutionize medicine and research (1). The chromosomal loci encoding these systems contain arrays of short direct repeats separated by spacer sequences of foreign origin. This array is transcribed and processed into CRISPR RNAs that associate with RNA-guided Cas nucleases to locate and cleave complementary nucleic acid sequences (i.e., protospacers) in the invading genome of a virus or plasmid. To prevent the spacer from recognizing and cleaving its complement in the chromosome array, Cas nucleases must also recognize a protospacer-adjacent motif (PAM) immediately adjacent to the protospacer sequence on the nontargeted DNA strand. Class 2 CRISPR-Cas systems are particularly unique for their single multidomain effectors (Cas9, Cas12, and Cas13), which are functionally analogous to entire complexes of Cas proteins in class 1 systems (2). In 2013, Qi and colleagues (3) coexpressed a catalytically dead version of Cas9 (dCas9) from Streptococcus pyogenes with customizable single guide RNAs (sgRNAs) that sterically block transcription elongation of the 20-nucleotide protospacer targets with short 5′-NGG-3′ PAM sites. Owing to the technique’s superiority for simple, titratable, target-specific, and reversible gene silencing, this CRISPR interference (CRISPRi) method has since become a widely used technique for gene function studies across all three domains of life (3–5), including diverse bacterial species (6–10).

Borrelia burgdorferi, the causative spirochete of Lyme disease (11, 12), is evolutionarily distant from other tractable bacterial models (13). It is among the most genetically complex bacteria known with a distinctly segmented genome consisting of a megabase chromosome and large set of circular and linear plasmids (14, 15). Nonetheless, a robust genetic toolbox has been developed to investigate B. burgdorferi physiology and pathogenesis (16, 17). This includes (i) a growing number of antibiotic resistance cassettes that can be used for gene disruption and complementation by homologous recombination (18–22), (ii) Himar1-based transposon mutagenesis (23, 24), (iii) E. coli/B. burgdorferi shuttle vector plasmids (18, 25, 26), and (iv) hybrid lac or tet promoters for conditional gene expression (27–29). With these tools, functional assessment of essential genes classically requires the generation of merodiploid lac- or tet-promoter-driven overexpression intermediates that are then further mutated to disrupt the endogenous gene copy. Here, we developed a fully inducible, nontoxic CRISPRi system with optional in trans gene complementation that augments the current CRISPRi platform for Borrelia burgdorferi (10). We validated this system by targeting a polycistronic operon encoding OspA and OspB, two abundant and well-studied borrelial outer surface lipoproteins that play important roles in the pathogen’s enzootic cycle (30–35).

RESULTS

Like most bacterial phyla, spirochaetes are predicted to contain native CRISPR-Cas systems based on extensive computational analyses (2). However, Cas homologs found in various Treponema and Leptospira species are not detectable in the Borrelia genus with domain-based NCBI BLAST searches (36). This suggested that synthetic CRISPR systems can be successfully implemented in B. burgdorferi and other related species without any prior genomic alterations. To construct a CRISPRi system for B. burgdorferi, we modified the Mobile-CRISPRi system described by Peters et al. (37) for diverse bacteria. We aimed (i) to simplify the experimental workflow of generating knockdown mutants by providing all necessary elements on a single customizable plasmid, (ii) to eliminate the potential for rare codon usage toxicity in a low GC content organism by codon-optimizing any heterologous proteins, and (iii) to increase stringency of the system by placing both dCas9 and sgRNA CRISPRi modules under inducible promoters. Briefly, in two stepwise NEBuilder HiFi assemblies, Streptococcus pyogenes-derived dCas9-myc (SpydCas9-myc) and trc-driven sgRNA modules from the Mobile-CRISPRi system were amplified by PCR and combined with a third module containing the backbone of pJSB142, a shuttle vector for isopropyl-β-d-thiogalactopyranoside (IPTG)-inducible expression of genes under a hybrid T5/lac promoter (P_pQE30) in B. burgdorferi (28). B. burgdorferi transformants carrying the resulting recombinant plasmid pJJW100 showed a severe growth defect under inducing conditions (not shown), which suggested rare codon toxicity (38) from SpydCas9. We therefore codon-optimized SpydCas9 using the OPTIMIZER web-based server (Supplementary File S1 in the supplemental material) (39). The B. burgdorferi codon-optimized dCas9 (BbdCas9) nucleotide sequence was synthesized by GenScript and used to replace the original gene in pJJW100 by restriction ligation, yielding pJJW101 (Fig. 1A). This optimization eliminated any observable toxicity in B. burgdorferi (Fig. 1C).

FIG 1.

Single-plasmid-based CRISPRi system for Borrelia burgdorferi. (A) Plasmid map showing the functional elements of pJJW101. pJJW101 contains a pUC (ColE1) high-copy origin of replication and inverted repeat (IR)-flanked B. burgdorferi paralogous family (PF) 57, 50, and 49 ORFs, which allow for the plasmid’s autonomous replication in E. coli and B. burgdorferi, respectively (see the text). Codon-optimized LacI (BbLacI), constitutively expressed from a B. burgdorferi flaB promoter, represses both codon-optimized BbdCas9-myc and sgRNA production under pQE30 (T5/lac hybrid) and trc (trp/lac hybrid) promoters, respectively. Three sequential BsaI sites (black-filled circles) are used for the directional insertion of unique gene-targeting spacers in a “scarless” Golden Gate Assembly-like manner that allows digestion and ligation steps to be carried out simultaneously, with the elimination of BsaI recognition sites in the ligated product. A single BglI site (open circle) downstream of the sgRNA module can be used for directional insertion of complementing gene copies or reporters. The plasmid confers resistance to kanamycin (Kan^R) with the aph[3′]-IIIa gene. (B) Schematic of the pJJW101 knockdown functional region as an empty vector control without gene-targeting sgRNA spacer insert. (C) Liquid culture growth curves of recombinant B. burgdorferi B31-e2 carrying pJJW101 (empty vector control). Cultures were inoculated at either 1 × 10⁵ cells/mL (teal lines) or 1 × 10⁶ cells/mL (orange lines) and grown in the presence of different IPTG inducer concentrations. The shown sets of growth curves are representative of growth curve sets from three biological replicates.

To test the system’s knockdown efficiency, we first designed 20-nucleotide spacers targeting either the nontemplate (NT) or template (T) strands of flaB or ospA (Table 1) using the open-source CRISPy-web interface on an artificially concatenated sequence of the Borrelia burgdorferi B31 genome (Supplementary File S2) (40). Pairs of complementary primers with 5′-overhangs matching the BsaI cut site sticky ends (Table 2) were synthesized for each chosen spacer sequence and hybridized. The resulting double-stranded oligonucleotides were directly ligated into BsaI-cut pJJW101, generating a series of flaB and ospA knockdown plasmids (Table 3; plasmid IDs 2 to 10) with simple knockdown functional regions (Fig. 2A). Borrelia burgdorferi B31-e2 clones carrying the “empty” pJJW101 vector (lacking any targeting spacer sequence) or one of the nine derived knockdown plasmids were cultured, and their whole-cell protein profiles were determined by SDS-PAGE and immunoblotting at 36 h postinduction with 0.5 mM IPTG. In comparison to the empty pJJW101 vector and uninduced controls, the knockdown plasmids had varied efficiencies at silencing target protein expression when the CRISPRi system was induced (Fig. 2C). Notably, the 162-kDa BbdCas9-myc was visible as a distinct band after Coomassie staining, allowing for the rapid confirmation of system inductions. We also readily observed robust knockdowns of the highly abundant OspA and OspB (31 and 34 kDa, respectively) and to a lesser extent FlaB (37 kDa). Among the nine evaluated sgRNAs, ospAT1 was the only spacer sequence that did not appreciably affect target protein levels. ospAT1’s ineffectiveness is likely multifactorial: first, as observed in other systems, there appears to be an efficiency bias for NT sgRNAs over T sgRNAs (3, 6), supported by flaBNT2 and ospANT2 sgRNAs leading to the strongest reductions in target protein expression. Second, the specificity of the 3′-core (PAM-proximal) region of sgRNA spacer sequences for the intended target has been shown to drive knockdown efficiency (41–44) as off-target binding sequesters dCas9-sgRNA complexes. For this reason, the CRISPy-web platform calculates the number of off-target genomic hits containing 0-, 1-, or 2-bp mismatches for the 13-bp 3′ core of each potential 20-bp spacer sequence in a selected open reading frame (ORF). Third, a GC core content ≥30% leads to more efficient sgRNA annealing to the target sequences (41–44). ospAT1 had the least favorable values of the tested sgRNAs for silencing ospA as it was located on the T strand, had 10 potential off-target binding sites with 1- or 2-bp core mismatches, and had a low core GC content of 23.1% (Table 1).

TABLE 1.

sgRNA spacer sequences used in this study

sgRNA	Starta	Enda	Strand	ORF	PAM	Sequence (5′ to 3′)	Core %GC	Core 0-bp mismatch	Core 1-bp mismatch	Core 2-bp mismatch
flaBT1	76	98	T	BB0147	TGG	ACTCAAGAAAAGCTTTCTAG	30.8	0	0	6
flaBT2	127	149	T	BB0147	TGG	GCTGCTGGCATGGGAGTTTC	53.8	0	0	0
flaBNT1	393	415	NT	BB0147	TGG	TGTTTGATAACATGTGCATT	30.8	0	0	1
flaBNT2	481	503	NT	BB0147	TGG	TGAGACCCTGAAAGTGATGC	46.2	0	0	4
ospAT1	4	26	T	BBA15	AGG	AAAAAATATTTATTGGGAAT	23.1	0	2	8
ospANT1	38	60	NT	BBA15	AGG	ATTTTGCTTACATGCTATTA	23.1	0	0	2
ospANT2	69	91	NT	BBA15	AGG	AAACGCTGTTTTTCTCGTCA	38.5	0	0	0
ospANT3	103	125	NT	BBA15	AGG	ACAAGAACTTTCATTTCACC	38.5	0	0	1
ospANT4	290	312	NT	BBA15	TGG	TTCTTTGAAAACTTCAAGTG	30.8	0	0	3

^aCoordinates of the first and last nucleotide of the spacer sequence relative to the 5′ end of the ORF. See text for explanation of other parameters.

TABLE 2.

Oligonucleotides used in this study

Purpose (study)/ID	Name	Sequence (5′ to 3′)
OspA RT-qPCR (72)
1	OspA-F	ATATTTATTGGGAATAGGTCTAATAT
2	OspA-R	CTTTGTTTTTTTCTTTGCTTACAAG
FlaB RT-qPCR (73)
3	BorFlaLeo-R-ok	GCTGGTGTGTTAATTTTTGCAG
4	FlaW-sense4	AGCAACTTACAGACGAAATTAATAG
16S rRNA RT-qPCR (74)
5	Bor-16S-Fv2	GGCGGCACACTTAACACGTTAG
6	Bor-16S-Rv2	GGTCAAGACTGACGCTGAGTCA
pJJW101 assembly (this study)
7	pJSB142-F1	GTGTTTTTTCGTTTTGGTCCTAAAAGCTTGAAGATAGAGAGAGAAAAGTGGAAGGT
8	pJSB142-R1	ATGCTGTACTTCTTGTCCATATGTAATTTCTCCTCTTTAATGAATTCTGTGTGAAATTG
9	dCas9-F	TTAAAGAGGAGAAATTACATATGGACAAGAAGTACAGCATCGGC
10	dCas9-R	TCTCTATCTTCAAGCTTTTAGGACCAAAACGAAAAAACACCCT
11	pJSB142-F2	CAGAATATTTGCCAGAACCGGGATCCTCTAGAAAATCATAAAAAATTTATTTGCTTTGT
12	pJSB142-R2	GGTGCTTTTTTTTTGAATTCCCCGGGTACCGGC
13	sgRNA-F	CTTAGTGGCCGGTACCCGGGGAATTCAAAAAAAAAGCACCGACTCG
14	sgRNA-R	TATGATTTTCTAGAGGATCCCGGTTCTGGCAAATATTCTGAAATG
sgRNAs (this study)
15	flaBT1-F	TAGTACTCAAGAAAAGCTTTCTAG
16	flaBT1-R	AAACCTAGAAAGCTTTTCTTGAGT
17	flaBT2-F	TAGTGCTGCTGGCATGGGAGTTTC
18	flaBT2-R	AAACGAAACTCCCATGCCAGCAGC
19	flaBNT1-F	TAGTTGTTTGATAACATGTGCATT
20	flaBNT1-R	AAACAATGCACATGTTATCAAACA
21	flaBNT2-F	TAGTTGAGACCCTGAAAGTGATGC
22	flaBNT2-R	AAACGCATCACTTTCAGGGTCTCA
23	ospAT1-F	TAGTAAAAAATATTTATTGGGAAT
24	ospAT1-R	AAACATTCCCAATAAATATTTTTT
25	ospANT1-F	TAGTATTTTGCTTACATGCTATTA
26	ospANT1-R	AAACTAATAGCATGTAAGCAAAAT
27	ospANT2-F	TAGTAAACGCTGTTTTTCTCGTCA
28	ospANT2-R	AAACTGACGAGAAAAACAGCGTTT
29	ospANT3-F	TAGTACAAGAACTTTCATTTCACC
30	ospANT3-R	AAACGGTGAAATGAAAGTTCTTGT
31	ospANT4-F	TAGTTTCTTTGAAAACTTCAAGTG
32	ospANT4-R	AAACCACTTGAAGTTTTCAAAGAA
Complementation (this study)
33	BglI-PospA-F	CCCGGGTACCGGCCACTAAGGCAAGTCCCAAAACTGGGACTTTTTTTAAATAAAAAAT
34	BglI-ospA-R	TTTTTTTATTCCTAGGCCTTAGTGGCTTATTTTAAAGCGTTTTTAATTTCATCAAG
35	PAM1*-F	GAATAGGTCTAATATTAGCATTAATAGCATGTAAGC
36	PAM1*-R	GCTTACATGCTATTAATGCTAATATTAGACCTATTC
37	PAM2*-F	GTAAGCAAAATGTTAGCAGTCTTGACGAGAAAAACAGCG
38	PAM2*-R	CGCTGTTTTTCTCGTCAAGACTGCTAACATTTTGCTTAC
39	PAM3*ngg-F	GTTTCAGTAGATTTGCCCGGTGAAATGAAAG
40	PAM3*ngg-R	CTTTCATTTCACCGGGCAAATCTACTGAAAC
41	PAM3*P35G-F	GTTTCAGTAGATTTGGGTGGTGAAATGAAAGTTC
42	PAM3*P35G-R	GAACTTTCATTTCACCACCCAAATCTACTGAAAC
43	PAM3*P35S-F	GTTTCAGTAGATTTGTCTGGTGAAATGAAAG
44	PAM3*P35S-R	CTTTCATTTCACCAGACAAATCTACTGAAAC
45	PAM4*-F	CGATCTAGGTCAAACAACACTTGAAGTTTTC
46	PAM4*-R	GAAAACTTCAAGTGTTGTTTGACCTAGATCG
Culture PCR (this study)
47	dCas9-F2	AGCCTTAGCGTCAACTCCACTTGC
48	dCas9-R2	AGGAGCTCTGCTGTTTGATAGCGG

TABLE 3.

Plasmids used in this study

ID	Plasmid	sgRNA	Complementing allele	Knockdown
1	pJJW101	None	None	None
2	pJJW101-flaBT1	flaBT1	None	FlaB
3	pJJW101-flaBT2	flaBT2	None	FlaB
4	pJJW101-flaBNT1	flaBNT1	None	FlaB
5	pJJW101-flaBNT2	flaBNT2	None	FlaB
6	pJJW101-ospAT1	ospAT1	None	OspAB
7	pJJW101-ospANT1	ospANT1	None	OspAB
8	pJJW101-ospANT2	ospANT2	None	OspAB
9	pJJW101-ospANT3	ospANT3	None	OspAB
10	pJJW101-ospANT4	ospANT4	None	OspAB
11	pASP151	ospANT1	WT ospA	OspAB
12	pASP152	ospANT2	WT ospA	OspAB
13	pASP153	None	PAM1* ospA	None
14	pASP154	None	PAM2* ospA	None
15	pASP155	None	WT ospA	None
16	pASP156	ospANT1	PAM1* ospA	OspB only
17	pASP157	ospANT2	PAM1* ospA	OspAB
18	pASP158	ospANT1	PAM2* ospA	OspAB
19	pASP159	ospANT2	PAM2* ospA	OspB only
20	pBTM151	ospANT3	WT ospA	OspAB
21	pBTM152	ospANT4	WT ospA	OspAB
22	pBTM153	ospANT3	PAM3a* ospA	OspAB
23	pBTM153e1	ospANT3	PAM3b* ospA	OspB only
24	pBTM153e2	ospANT3	PAM3c* ospA	OspB only
25	pBTM154	ospANT4	PAM4* ospA	OspB only
26	pBTM155	ospANT3	PAM2* ospA	OspAB
27	pBTM156	ospANT4	PAM2* ospA	OspAB

FIG 2.

CRISPRi knockdowns of flaB and ospA. (A) Schematic of the pJJW101 knockdown functional region as an empty vector control or with template (T) and nontemplate (NT) strand gene-targeting sgRNAs. The target sites for each sgRNA (red triangles) are shown proportional to their respective ORF. (B) Borrelia cells carrying CRISPRi system plasmids having either no gene-targeting sgRNA (empty) or one of eight unique gene-targeting sgRNAs (against flaB or ospA) were induced with 0.5 mM IPTG for 36-h in comparison to uninduced controls. Quantification of flaB, ospA, and 16S rRNA transcripts was done by RT-qPCR. Targeted flaB and ospA transcripts were normalized against the 16S rRNA endogenous control. Data are expressed as the average fold change of targeted genes by different sgRNAs relative to the amount of transcript present in uninduced empty samples (indicated by arrows). Error bars show the mean ± SEM for three independent experiments. (C) Coomassie-stained SDS-PAGE gels and Western immunoblots of whole-cell lysates prepared from pJJW101-transformed Borrelia cells incubated with (+) or without (−) 0.5 mM IPTG for 36 h. Data shown are representative of three biological replicates.

We next harvested total RNA from spirochetes with CRISPRi systems that resulted in strong protein silencing to more directly analyze their knockdown efficiencies at the transcript level by reverse transcription-quantitative PCR (RT-qPCR) (Fig. 2B). While generally in line with the immunoblot data (Fig. 2C), the analysis showed 4-fold, 7-fold, and 74-fold decreases in ospA transcript abundance for ospANT4, ospANT3, and ospANT1 inductions, respectively. This supports the preference of effective sgRNAs for promoter-proximal targets, but it also indicates that even relatively low transcriptional silencing can be sufficient to produce a phenotype at the protein level. Conversely, the visually greater protein knockdown efficiency of flaBT1 compared to flaBNT1 was measured as only a 25% greater transcript knockdown. These inconsistencies may originate from unique posttranscriptional and posttranslational regulation mechanisms. As shown for the ospAB operon, variations of ospAB mRNA levels among strains or even within cultures of the same strain can still have equivalent protein levels, likely as a result of the same regulatory mechanisms (45–47). Thus, overall CRISPRi sgRNA efficiency should be measured at the protein level and based on phenotypic changes whenever possible.

At this point, our CRISPRi platform allowed for the controlled repression of B. burgdorferi genes. Yet, as the polar effects of ospA-targeting sgRNAs on OspB expression showed, strict fulfillment of Koch’s molecular postulates in this system was limited to monocistronic genes or the last gene in an operon. Thus, we explored whether we could augment the platform to allow for specific in trans complementation. Using the ospAB operon for proof-of-principle experiments, we took advantage of the strict PAM requirement for sgRNA-mediated gene targeting of dCas9 (48). The pJJW101 empty vector and its four ospANT sgRNA derivatives were modified by directional insertion of different ospA variants under their own promoter (P_ospA-ospA) into a BglI site downstream of the sgRNA module (Fig. 3A). The P_ospA-ospA copy PAM sites for each of the four ospANT sgRNAs were mutated individually, leading to ospA PAM* alleles that were expected to be resistant to targeting by their cognate sgRNAs (Table 4). The resulting series of recombinant pJJW101 derivatives (Table 3, plasmid IDs 13 to 27) contained both matched (e.g., ospANT2 and ospA PAM2*) and mismatched (e.g., ospANT2 and ospA PAM1*) combinations of sgRNA and complementing ospA alleles to evaluate the specificity of the complementation system. B. burgdorferi clones harboring the plasmids were cultured for 48 h with or without CRISPRi system induction, and their protein profiles were assessed by SDS-PAGE and immunoblotting (Fig. 3B and C). As expected, sgRNA-lacking (empty) control constructs with either wild-type (WT), PAM1*, or PAM2* ospA variants failed to block OspA and OspB expression. The observed increase in OspB levels upon complementation with the PAM2* ospA variant remains enigmatic but could be explained by a previously described OspA-independent OspB expression mechanism (49). Next, we compared the ability of matched and mismatched sgRNA/PAM* ospA combinations to selectively restore OspA expression, with WT ospA allele constructs used as internal negative controls. As expected, OspA expression was only restored in matched constructs while WT ospA-complemented and -mismatched constructs retained a dual OspAB knockdown phenotype, notably despite the plasmid-derived increase in ospA copy number. While the PAM1*, PAM2*, and PAM4* ospA mutations were synonymous, the PAM3 site overlapped with a proline codon, eliminating the possibility for a synonymous change. We therefore generated a series of three mutants: PAM3a* represented a synonymous PAM site “wobble” (NGG to NGG) mutation, which failed to complement OspA expression in an ospANT3-mediated knockdown. The other two nonsynonymous PAM3b* (Pro35Gly) and PAM3c* (Pro35Ser) ospA alleles complemented the ospANT3 knockdowns with fully mutant OspA proteins upon system induction. Notably, PAM3c* ospA is identical to a naturally occurring OspA homolog from B. burgdorferi strain CA7 (GenBank AAA20957.1). That OspA variant is predicted by Alphafold (50) to fold almost identically to B. burgdorferi B31 OspA (Fig. 3D) with a slight angular offset that is exacerbated for in the PAM3b* (Pro35Gly) variant (Supplementary File S3). This set of experiments fully validated the B. burgdorferi CRISPRi PAM* system for gene silencing and selective gene complementation.

FIG 3.

CRISPRi knockdowns of the ospAB operon with PAM* ospA allele-mediated complementation. (A) Schematic of the complemented pJJW101 knockdown functional region as an empty vector control or with gene-targeting sgRNAs. (B) Borrelia cells carrying either wild-type (WT) or PAM mutant (PAM*) ospA-complemented CRISPRi system plasmids having either no gene-targeting sgRNA (empty) or one of three unique ospA-targeting sgRNAs (NT2, NT3, or NT4) were induced (+) with 0.5 mM IPTG for 48 h in comparison to uninduced (−) controls. Whole-cell lysates were analyzed by Coomassie-stained SDS-PAGE gels and anti-OspA and anti-OspB immunoblots. Data shown are representative of three biological replicates. (C) Immunoblots for the ospA-complemented CRISPRi system inductions (including data in panel B) were quantified with ImageJ to visualize knockdown proportions as the ration of induced/uninduced protein levels. Error bars show the mean ± SEM for three independent experiments. (D) AlphaFold prediction for mature P35S OspA mutant is superimposed onto the prediction for mature WT OspA (transparent) and colored by predicted local difference distance test (pLDDT) score for each residue. PyMOL calculated root mean square deviation (RMSD) for the superimposition is shown.

TABLE 4.

PAM site mutations

sgRNA	sgRNA-matching PAM site	PAM* name	PAM* sitea	Amino acid change
ospANT1	AGG	PAM1*	ATG	None
ospANT2	AGG	PAM2*	AGA	None
ospANT3	AGG	PAM3a*	GGG	None
		PAM3b*	ACC	Pro to Gly
		PAM3c*	AGA	Pro to Ser
ospANT4	TGG	PAM4*	TGT	None

^aMismatching nucleotides are underlined.

DISCUSSION

The CRISPRi PAM* system described here further augments the genetic toolbox for the spirochetal pathogen B. burgdorferi, providing a method for inducible knockdowns with optional in trans complementation. All necessary factors for knockdown and complementation are encoded on a single recombinant plasmid, making the system convenient for any B. burgdorferi strain background and, with some minor adaptations, transferable to genetically amenable relapsing fever Borrelia (16, 51–54).

Takacs et al. (10) also observed toxicity of the original S. pyogenes dCas9 gene in B. burgdorferi during the development of their CRISPRi platform. While that study reduced SpydCas9 expression to tolerated levels by lowering the inducer concentration, we were able to overcome heterologous protein toxicity by codon optimization of the GC-rich SpydCas9 gene (56% GC) to match the AT-rich B. burgdorferi genome (Supplementary File S1). This suggests that SpydCas9’s toxicity in B. burgdorferi is due to rare codon usage (38), likely exacerbated by the large size of the transcript/protein. Codon-optimized BbdCas9 allowed us to maintain the use of the original P_pQE30 (T5/lac) hybrid promoter for IPTG-driven BbdCas9 expression under standard induction conditions. As a result, the BbdCas9 protein was clearly detectable in Coomassie-stained SDS-PAGE gels of induced whole-cell protein samples but remained undetectable even by immunoblot (via a C-terminal C-myc epitope tag) under noninducing conditions (Fig. 2C), indicating tightly controlled expression. Our system also added an additional level of control by putting the sgRNA module under a trc (trp/lac) hybrid promoter, which ensures that both dCas9 and sgRNA CRISPRi system components remain repressed until being simultaneously induced by IPTG. This concerted control of both dCas9 and sgRNA expression might be key to eliminating phenotypic changes under noninducing conditions; such baseline phenotypes were observed by Takacs and colleagues (10) when targeting essential B. burgdorferi genes and had to be skillfully resolved by mutational weakening of the dCas9-driving P_pQE30 promoter sequences. Building on those collective findings, additional modifications of the B. burgdorferi CRISPRi platform that replace the currently lac-based promoters with the tetracycline-responsive ospA/tetO hybrid P_ost promoter (29) might render the system inherently titratable to match the CRISPRi systems for other bacteria (6, 37).

The CRISPy-web interface provided an accessible and relatively simple tool for identifying potential sgRNA spacers. When asked to query a genomic sequence like the artificial concatenated B. burgdorferi strain B31 contig used here, it produces a ranked list of spacers for each selected ORF, using the lowest number of genomic off-target core hits as the most important parameter for selection and ranking (40). Our set of experiments continues to support the general rules that more efficient spacers also (i) target the nontemplate strand, (ii) bind closer to the targeted gene’s promoter region, and (iii) have ≥30% GC content within their core region.

As demonstrated in this study, the B. burgdorferi CRISPRi PAM* approach can be used to restore the expression of a targeted gene. Mutation of the PAM sequence alone is sufficient to generate a “CRISPRi-resistant” complementing allele. However, it should be noted that NT or T sgRNA PAM sites overlapping Pro or Gly codons, respectively, will require nonsynonymous mutations to produce PAM* sites. In these cases, structural information and comparative genomics can be used to guide sgRNA selection and PAM* mutant design. For example, in our ospANT3 PAM* complementation series, the affected OspA Pro35 lies in a loop connecting two β-sheets (Fig. 3D and Supplementary File S3) where substitution mutations are less likely to disrupt the structure and destabilize the protein. We first used a BLAST search to identify a naturally occurring OspA Pro35Ser variant in Borrelia burgdorferi strain CA7 (55), which informed the design of the PAM3c* ospA allele (Table 4). The PAM3b* allele replaced Pro35 with the small and flexible Gly residue. As predicted, both complementing OspA mutants were readily detected by SDS-PAGE and Western immunoblotting (Fig. 3B). Alternatively, CRISPRi-resistant alleles can be generated by disrupting complementarity to the core spacer sequence. A machine learning approach for spacer sequence mismatching has been used to determine which substitutions will most likely reduce sgRNA activity in E. coli and Bacillus subtilis (56), and a combination of PAM and spacer mutations was recently used in a screen for Mycobacterium tuberculosis antimicrobial resistance mechanisms (57).

One obvious application of the B. burgdorferi CRISPRi PAM* system is to ensure that a CRISPRi-mediated phenotype is indeed due to silencing of the targeted gene alone. This will be particularly important when assessing potential polar effects in phenotypical studies of genes that, based on RNASeq data (58), are either not at the end of a polycistronic transcript or not independently transcribed. If complementation with the PAM* allele does not restore the WT phenotype, the same pJJW101-based CRISPRi system can be used to complement with the operon’s nontargeted genes. The currently largest pJJW101-related recombinant plasmid is about 14 kb in size (25), which means that pJJW101 should accommodate at least 3 kb of complementing sequence. Larger operons might require placement of the system’s BbdCas9 and BbLacI (or TetR) expression cassettes on stable B. burgdorferi replicons as in the study by Takacs et al. (10).

Most intriguingly, the B. burgdorferi CRISPRi PAM* system can be employed to simultaneously silence a gene and complement it with a mutant allele, similar to the above-described replacement of strain B31 OspA by a mutant variant from strain CA7. This system application will be particularly helpful in structure-function studies of essential B. burgdorferi proteins. Traditionally, construction of an essential gene knockdown strain first required adding an inducible target gene allele on a recombinant plasmid, resulting in a merodiploid intermediate. This intermediate then had to be mutagenized further to disrupt the endogenous target gene locus by allelic exchange with an antibiotic cassette, finally yielding the haploid conditional knockout. In structure-function studies, this two-step protocol had to be repeated for each tested mutant. CRISPRi PAM* reduces this protocol to a single transformation with a single plasmid. This increased efficiency in B. burgdorferi molecular genetics also opens the door to generating CRISPRi PAM* plasmid-based mutant libraries in a Mut-Seq approach (59) to probe essential residues within essential proteins. To that end, we are currently using CRISPRi PAM* for structure-function studies of essential protein secretion and envelope homeostasis pathway components in B. burgdorferi.

MATERIALS AND METHODS

Bacterial strains, growth conditions, and recombinant plasmid purification.

Chemically competent NEB5-alpha cells (NEB; C2987H) were transformed with recombinant plasmids per manufacturer instructions and grown at 37°C on selective Luria-Bertani (LB) agar plates (BD; 244520) and in LB broth (Fisher; BP1426) containing 30 μg/mL kanamycin (Sigma; K4000). Plasmid DNA was isolated from E. coli clones using a Miniprep kit (Macherey-Nagel; 740588) and verified by single-pass primer extension sequencing (ACGT, Inc.) with either generic or custom DNA oligonucleotide primers (Integrated DNA Technologies). Verified plasmids were purified using a Miniprep kit (Macherey-Nagel; 740412) and concentrated by standard sodium acetate/ethanol precipitation (60) to ≥500 ng/μL for transformation into the B. burgdorferi strain type strain clone B31-e2 (61). B. burgdorferi was grown in liquid sterile filtered Barbour-Stonner-Kelly-II (BSK-II) medium (62, 63) containing 9.7 g/L CMRL-1066 (US Biological; C5900-01), 5.0 g/L neopeptone (Gibco; 211681), 6.6 g/L HEPES sodium salt (Fisher; BP410), 0.7 g/L citric acid (Sigma; C-8532), 5.0 g/L dextrose anhydrous (Fisher; BP350), 2.0 g/L yeastolate (Gibco; 255772), 2.2 g/L sodium bicarbonate (Fisher; BP328), 0.8 g/L sodium pyruvate (Fisher; AC132155000), 0.4 g/L N-acetylglucosamine (Sigma; A3286), 25 mg/L phenol red (Sigma; P-3532), and 50.0 g/L bovine serum albumin (Gemini; 700-104P) at pH 7.6 to 7.7, with 60 mL/L heat-inactivated rabbit serum (Pel Freez; 31126) and 200 mL/L 7% gelatin (Gibco; 214340) added before use. The cultures were incubated at 34°C under 5% CO₂ in a humidified incubator. Recombinant B31-e2 clones were propagated in BSK-II with 200 μg/mL kanamycin for plasmid maintenance.

Transformation and clonal selection of B. burgdorferi.

Electrocompetent B31-e2 cells were transformed by electroporation (64) with 1 to 5 μg of plasmid in 2-mm gap cuvettes (Thermo; 5520) using a Bio-Rad MicroPulser on EC2 setting, consistently measuring 2.49-kV/cm field strength and 5.20- to 5.60-ms pulse times. Electroporated cells were immediately resuspended in 12 mL of prewarmed BSK-II and allowed to recover at 34°C for 18 to 20 h. Clonal selection of transformants was carried out by adding the 12 mL recovered culture to 35 mL BSK-II with a final concentration of 200 μg/mL kanamycin, followed by plating into 96-well microtiter plates and 8 to 16 days of incubation (65). Culture-positive wells were expanded into 6 mL of selective BSK-II and allowed to reach stationary phase for verification of plasmid acquisition by direct PCR of cultured B. burgdorferi cells (66) using QuickLoad 2× Taq Master Mix (NEB, M0271) and primers 49 and 50 (Table 2). Positive clones were flash frozen on dry ice in BSK-II containing 10% DMSO (Sigma; D2438) and stored at −80°C. Maintenance of the endogenous plasmids carried by clone B. burgdorferi B31-e2 (cp26, cp32-1, cp32-3, cp32-4, lp17, lp38, and lp54) (61) was assessed by multiplex PCR using plasmid-specific oligonucleotide primers (67). This indicated that our clone B31-e2 stock had lost lp38. However, all other clone B31-e2 plasmids were maintained in the pJJW101 or daughter plasmid transformants (data not shown).

CRISPRi system plasmid construction.

pJJW101 was constructed using two stepwise NEBuilder HiFi assemblies (NEB; E5520S) followed by restriction and ligation. All used oligonucleotide primers are referred to by their ID number listed in Table 2. For the first assembly, the vector backbone of pJSB142, a kanamycin resistance-conferring derivative of pJSB104 (28), was amplified using primers 7 and 8 and combined with an SpydCas9 ORF module amplified by primers 9 and 10 from pTn7C107 (37). For the second assembly, the resulting pJSB142-dCas9 shuttle vector plasmid was amplified in its entirety with primers 11 and 12 to combine with a P_trc-driven sgRNA module amplified by primers 13 and 14 from pTn7C107, resulting in pJJW100. Next, a standard format codon usage table for B. burgdorferi B31 was obtained from the codon usage database (68) for species-specific codon optimization of dCas9 using the random guided codons parameter on OPTIMIZER (Supplementary Material S1) (39). This codon-optimized version of dcas9 (Bbdcas9) was synthesized and cloned in pBluescript II KS (+) with flanking 5′-NdeI and 3′-NotI restriction endonuclease sites (GenScript), yielding pBS-Bbdcas9. Finally, the NdeI-NotI SpydCas9 fragment of pJJW100 was replaced by the NdeI-NotI Bbdcas9 fragment from pBS-Bbdcas9, yielding pJJW101.

sgRNA spacer sequence design and insertion into pJJW101.

The 20-nucleotide spacer sequences for our sgRNAs were designed using the CRISPy-web interface on an artificially concatenated sequence of the entire Borrelia burgdorferi B31 genome having individual genetic elements separated by “NNNNNN” (Supplementary File S2) (40). Four sgRNAs, two each complementary to the nontemplate (NT) and template (T) strands, were chosen for targeting flaB. One T and four NT sgRNAs were chosen for targeting ospA (Table 1). The 5′- to 3′-spacer sequences listed in the CRISPy-web output were used to design oligonucleotide pairs: Forward oligonucleotides added 5′-TAGT overhangs to the spacer sequence, while reverse oligonucleotides added 5′-AAAC overhangs to the reverse complement of the spacer sequence (Table 2, oligonucleotide IDs 15 to 32). Complementary pairs of single-stranded oligonucleotides were mixed at a final concentration of 2.4 μM, denatured at 95°C for 5 min, and annealed by gradual cooling to room temperature (https://portals.broadinstitute.org/gpp/public/resources/protocols). The resulting double-stranded oligonucleotides had 5′ extensions that were complementary to the nonpalindromic 3′ overhangs generated by BsaI-HFv2 (NEB, R3733) digestion of pJJW101, facilitating efficient directional ligation of hybridized sgRNA primers and cut pJJW101 vector.

WT and PAM* ospA allele generation.

The coding region for OspA with its native intergenic promoter was amplified by PCR with Q5 thermostable high-fidelity polymerase (NEB; M0491) from B31-e2 genomic DNA with primers 35 and 36 (Table 2) containing 5′-BglI overhangs (Table 2), and the resulting amplicon was cloned into pCR-Blunt II-TOPO (Invitrogen; 451245). This plasmid was then used as the template for Splicing by Overlap extension (SOE) PCR with Q5 polymerase, flanking primers 35 and 36, and pairs of mutagenic complementary PAM* primers (Table 2, primers 37 to 48) to generate six PAM* versions of the BglI-flanked P_ospA-ospA amplicon. Synonymous mutations were introduced into the PAM sites wherever possible, e.g., the PAM1* site 5′-AGG-3′ to 5′-ATG-3′ corresponded to a synonymous 5′-GCC-3′ to 5′-GCA-3′ Ala codon change on the complementary strand. PAM3* could not be made synonymously, an issue for nontemplate sgRNAs that occurs when the reverse complement 5′-NGG-3′ PAM site aligns with a 5′-CCN-3′ Pro codon. To overcome this, the ospA (ORF BBA15) nucleotide sequence was used in an NCBI blastx query with default search parameters and a filter for Borrelia sequences. Sifting through the alignments tab with alignment view set to “Query-anchored with dots for identities” allowed for quick identification of a naturally occurring P35S allelic variant in Borrelia burgdorferi strain CA7 that was used for the PAM3c* ospA allele.

Complementation of CRISPRi constructs with WT and mutant PAM* alleles.

pJJW101 derivatives containing one of the four NT ospA-targeting sgRNAs were cut with BglI (NEB; R0143), dephosphorylated with Quick-CIP (NEB; M0525), and ligated with T4 ligase (NEB; M0202) at a 1:2 ratio with either WT or PAM* BglI-cut PospA-ospA amplicons. Isolated plasmid DNA from E. coli transformants was checked for insert by digestion with BglI prior to confirmatory sequencing with primer 14 (Table 2).

CRISPRi system inductions and growth curves.

Recombinant B. burgdorferi strain stocks grown in selective BSK-II were diluted in selective medium with or without 0.5 mM IPTG and harvested after 36 or 48 h of incubation. Whole-cell lysates were checked for overall changes in transcript and protein levels by SDS-PAGE and Western immunoblotting as detailed below. Cell density was assessed by direct counting of bacterial cultures diluted 10- to 100-fold in PBS in a Petroff Hauser counting chamber under phase-contrast microscope (Nikon Eclipse E400).

RT-qPCR for transcript expression analysis.

Total RNA was extracted from B. burgdorferi cells with TRIzol Reagent (Invitrogen; 15596026) according to the manufacturer’s instructions after a 30-minute 3,000 × g swinging-bucket centrifugation at room temperature. Residual DNA contamination was removed by a 1-h DNase I treatment (Thermo Scientific; 18047019) followed by phenol-chloroform extraction (Ambion; AM9720) and standard ammonium acetate/ethanol precipitation (69). RNA samples were used for reverse transcription and quantification of ospA, flaB, and 16s rRNA transcripts by the Luna Universal one-step RT-qPCR kit (NEB, E3005), according to the manufacturer’s instructions on an Applied Biosystems 7500 Fast real-time PCR system. The primers to amplify flaB, ospA, and 16S RNA transcripts are shown and referenced in Table 2 (primers 1 to 6). Transcript levels were validated and normalized against 16S rRNA transcript with fold changes calculated using the comparative CT (2^−ΔΔCT) method for quantification.

SDS-PAGE and immunoblotting for protein expression analysis.

B. burgdorferi cells grown to late logarithmic or early stationary phase were harvested by centrifugation in a swinging bucket at 3,000 × g for 30 min at room temperature. Harvested cells were washed twice with dPBS + 5 mM MgSO₄ to remove culture medium BSA and resuspended in standard 1× SDS sample buffer (69). Whole-cell lysates were resolved by 12% SDS-PAGE and visualized by Coomassie staining (Fisher; BP3620-1). For immunoblotting, proteins were transferred electrophoretically to nitrocellulose blotting membrane (GE, 10600002) using a Bio-Rad Mini Trans-Blot apparatus at 100 V for 1 h with prechilled transfer buffer (25 mM Tris, 200 mM glycine, and 20% methanol). Membranes were blocked for 30 min in TBST buffer (25 mM Tris, 150 mM NaCl, 0.05% Tween 20, pH 7.2) with 5% dry milk before incubating on a rocker overnight at 4°C with rat anti-FlaB polyclonal antiserum (1:4,000 dilution; a gift from M. Caimano, University of Connecticut Health Center), mouse anti-c-Myc 9E10 (1:1,000 dilution; Fisher; MA1-980), mouse anti-OspA (1:10,000 dilution; H5332; reference 70), or mouse anti-OspB (1:5,000 dilution; H6831; reference 71) primary antibodies. After three 10-minute washes with TBST, the blots were incubated at room temperature on a rocker for 1 h with secondary anti-mouse IgG-HRP (Sigma; A4416) or anti-rat IgG-HRP (Thermo; 31470) antibody. Three additional 10-minute TBST washes, membranes were allowed to react with Super Signal West Femto (Thermo; 34096) substrate per manufacturer instructions and the signal was detected by exposure to autoradiography film (Midsci; XC59X).