CRISPR typing of Campylobacter jejuni reveals a link between CRISPR array preservation and Guillain-Barré syndrome

Guillain-Barré syndrome (GBS) is a post-infection sequela of Campylobacter jejuni-induced enteritis. Clustered regularly interspaced short palindromic repeats and associated genes (CRISPR-Cas) confers adaptive immunity and plays role in virulence in many bacteria. We investigated C. jejuni CRISPR type (CT) to explore association of CRISPR-Cas with risk of developing GBS. We analysed CRISPR-Cas in C. jejuni isolated from 30 patients with GBS, 60 patients with enteritis and 52 healthy controls from Bangladesh. CRISPR types were determined by PCR followed by CRISPR array sequencing. Statistical and genomic analyses were performed using SPSS and multiple web-based software respectively. We found preserved CRISPR array was signicantly more frequent in GBS-related strains than healthy control-related strains (P = 0.02, OR = 2.95). Increased CRISPR array length was signicantly associated with GBS compared to healthy control- (P = 0.003, AUC = 0.7) and enteritis-related strains (P = 0.02, AUC = 0.65). We reported 38 new CT found among 70 CRISPR-preserving strains. CT of GBS-related strains were unique from enteritis-and healthy control-related strains. Eighty spacers, including 20 novel spacers, were identied among the CRISPR-preserving strains. CRISPR typing had more discriminatory power than PCR-based subtyping in enteritis- and healthy control-related strains. Further genomic analyses are warranted to elucidate role of the C. jejuni CRISPR array in GBS pathogenesis.


Introduction
Campylobacter jejuni is a leading cause of bacterial gastroenteritis worldwide and the most frequent antecedent event associated with the post-infection sequela, Guillain-Barré syndrome (GBS) [1][2][3] . Molecular mimicry between bacterial lipopolysaccharide (LPS) and host nerve gangliosides is the apparent cause of pathogen-derived GBS in a susceptible host 4,5 . However, not all C. jejuni-infected individuals develop GBS after enteritis 6 . In addition to host susceptibility factors 7-10 , bacterial virulence factors also function as key regulatory elements that may trigger GBS [11][12][13] . Moreover, the cst-II sialyltransferase gene of C. jejuni, which encodes ganglioside-like structures expressed on the bacterial cell surface, has been linked to GBS 14 . In addition to disease pathogenesis, cst-II has been shown to confer e cient bacteriophage resistance in C. jejuni along with a functionally complex adaptive immune system, the type II Clusters Regularly Interspaced Short Palindromic Repeat and associated genes (CRISPR-Cas) 15 .
The CRISPR-Cas system is a repeat region of numerous archaeal and bacterial DNA sequences that provides adaptive immunity against invading phages and plasmids by targeting their nucleic acids in a sequence-speci c manner [16][17][18] . The latest CRISPR-Cas system classi cation includes two major classes, six types and 33 subtypes based on organizational variation in the Cas genes 19 . The C. jejuni genome harbours a class 2, type II-C CRISPR-Cas system comprised of a Cas9-Cas1-Cas2 operon followed by a trans-activating CRISPR-RNA (tracrRNA) region and CRISPR repeat-spacer array, which is transcribed in the opposite direction 20 . The CRISPR array in C. jejuni is composed of multiple 36-bp tandem consensus direct repeats (DRs) separated by 'spacer' sequences of similar length (30 bp) acquired from an invading foreign phage or plasmid 21,22 .
In addition to conferring adaptive immunity against invading foreign DNA, the CRISPR-Cas system also plays a role in regulation of virulence genes, competition between mobile genetic elements (MGEs), DNA repair, gene regulation of group behaviour, and the acquisition of antibiotic resistance genes and pathogenicity islands 23,24 . The type II-C CRISPR-Cas system exerts regulatory roles in pathogenicity, cell surface immunity, stress response bacterial physiology and anti-microbial resistance in many bacteria, including C. jejuni 25,26 .
Due to the presence of polymorphisms in the spacer alleles within the CRISPR array, CRISPR-based genotyping (CRISPR typing) of bacteria enables high-resolution genotyping of pathogenic bacteria, including C. jejuni isolates. CRISPR typing also provides an effective framework for epidemiologic source tracking, studies of genome microevolution and host-virus population dynamics 27,28 . A previous comparative genotyping study revealed diversity among genotypes and clonal complexes of various strains of C. jejuni isolated from chickens, individuals with enteritis and patients with GBS in Bangladesh 29 . Serotyping and comparative genomic analyses demonstrated certain serotypes and genotypes were overrepresented in C. jejuni isolates obtained from Bangladeshi patients with GBS 30,31 . Thus, this study aimed to determine the CRISPR types (CTs) in C. jejuni-related strains and explore the association between preservation of the CRISPR-Cas sequence and the risk of developing GBS in Bangladesh.

Results
Distribution of CRISPR-Cas components in C. jejuni CRISPR array was detected in 67% of the GBS-related, 48% of the enteritis-related and 40% of the healthy control-related C. jejuni strains (Table 1). There were no signi cant differences in the frequencies of the Cas genes (Cas1, Cas2 and Cas9) between the groups of C. jejuni strains related to GBS, enteritis and healthy controls. All GBS-associated strains and almost 90% of the enteritis and healthy control-related C. jejuni strains were positive for at least one Cas gene (Table 1). CRISPR types in C. jejuni PCR-based rapid screening revealed a total of seven CRISPR subtypes across all 142 C. jejuni isolates ( Table 2). Amplicon lengths of 0 bp (negative in the F2 primer PCR assay) and 190 bp were not considered as subtypes, as they either did not harbour any CRISPR array or did not contain any spacer in their CRISPR array. Two subtypes were not detected in the GBS-related strains (330 bp and 720 bp; Table 2) and two subtypes were not detected in the enteritis-related strains (530 bp and 590 bp; Table 2); the other three subtypes were ubiquitously detected in the GBS-, enteritis-and healthy control-related strains. Table 2 PCR-based detection of CRISPR subtypes in C. jejuni strains isolated from patients with GBS, enteritis and healthy controls. a Indicates PCR amplicon length in PCR assay with F2 and R primers (Fig. 4) (Table 3). Six CTs were found in both the enteritis-and healthy control-related strains. However, all seven CTs found in the GBS-related strains were absent in the enteritis-and healthy control-related strains (see Supplementary Table S1 online). Moreover, based on the presence of unique spacer arrangements that did not match with any C. jejuni CRISPR array sequence in the NCBI nucleotide database, 38 novel CTs were detected among the 42 CTs 32 . The most frequent CT among the GBS-related strains was CT 1. CT 9 was most frequent among enteritis-related strains, and CT 18 and CT 42 were most frequent among healthy control-related strains (see Supplementary Table S1 online). Spacer polymorphism-based CRISPR typing had a signi cantly higher discriminatory power than PCR-based typing in both the enteritis-related and healthy control-related C. jejuni strains (SID = 0.97 and 0.99, respectively; Table 3). However, due to the low (or lack of) variation in spacer arrangement within CRISPR arrays of the same length, PCR-based typing and sequence-based typing had similar discriminatory power among the GBS-related C. jejuni strains (SID = 0.79 and 0.83, respectively; Table 3).  Table S2 online) contained at least one nucleotide polymorphism site. The UPGMA dendrogram revealed that the enteritis-and healthy control-related strains were more closely related than the GBS-related strains, indicating a distinctive pattern of spacer acquisition and arrangement within the CRISPR arrays of the GBS-related strains (Fig. 2).

Features of target proto-spacers
We found target proto-spacers within 62 of the 80 spacer sequences (see Supplementary Table S2 online).
The target proto-spacers detected in 19 spacers from GBS-related strains, 31 spacers from enteritis-related strains and 28 spacers from healthy control-related strains were found in Campylobacter phages or plasmids. The most abundant proto-spacers were found in Campylobacter phage DA10 (NCBI accession no. MN530981), which was targeted by 14 spacers from GBS-related strains, 27 spacers from enteritis-related strains and 26 spacers from healthy control-related strains (see Supplementary Table S1 and   Supplementary Table S2 online). Nine spacers from GBS-related strains, 10 spacers from enteritis-related strains and eight spacers from healthy control-related strains shared at least 80% homology with target proto-spacers found in non-Campylobacter phage or plasmids (see Supplementary Table S2 online). The sequence logo (Fig. 3) created by aligning the 3` downstream proto-spacer sequences 33 identi ed 5 -ℕℕAYAC -3 as a putative proto-spacer adjacent motif (PAM) for the C. jejuni Cas9 endonuclease.

Discussion
This study indicates that preservation of the CRISPR array in C. jejuni is strongly associated with the risk of developing GBS. In addition, the probability of developing GBS following C. jejuni infection increases signi cantly with the CRISPR array length. Moreover, we found the Campylobacter bacteriophage DA10 is targeted by the CRISPR-Cas system of the majority of C. jejuni strains with a preserved CRISPR array. Moreover, this study identi ed 38 novel CTs and 20 novel spacer sequences in the CRISPR array, and demonstrates that GBS-related C. jejuni strains contain unique CT patterns compared to enteritis-and healthy control-related strains.
The type II CRISPR-Cas system has previously been reported to be involved in endogenous gene expression and gene regulation, respond to envelope stress and regulate other physiological processes in bacteria 25,35 . The ability of the Cas9 endonuclease to regulate the virulence genes of bacteria containing the type II CRISPR-Cas system were described in experimental studies of C. jejuni 15 39 . However, the relationship between CRISPR array preservation and bacterial virulence or pathogenesis had not previously been assessed. This study provides the rst evidence that preservation of the CRISPR array in C. jejuni is associated with the development of GBS, which suggests a preserved CRISPR array may contribute to virulence or immunogenicity by controlling the expression of bacterial genes that increase the risk of developing GBS.
The CRISPR array length depends on the number of spacers acquired by the bacteria (Fig. 4). In this study, we observed longer C. jejuni CRISPR arrays were associated with a higher probability of developing GBS. This observation can potentially be explained by the ndings of Martynov et al., who reported that bacteria maintain the optimum numbers of spacers to gain an evolutionary bene t 40 . Additionally, Levin's mathematical model predicted that a preserved CRISPR array provides a competitive advantage to bacteria by facilitating long-term co-evolutionary arms races between phage and bacteria 41 . This study supports the mathematical prediction that bacteria prefer to maintain an optimum CRISPR array length.
We also con rmed that CRISPR array-based typing had more discriminatory power and sensitivity than PCR-based subtyping due to the highly polymorphic CRISPR arrays of enteritis-and healthy control-related C. jejuni strains; thus, CRISPR array-based typing may represent a superior genotyping method 42  This study has several limitations. Even though we used PCR-based CRISPR typing, the discriminatory power of CRISPR typing was not compared with traditional genotyping or serotyping methods such as pulsed-eld gel electrophoresis (PFGE), multilocus sequence typing (MLST) or aA serotyping. Additionally, the sample size was relatively small and the signi cance of the association between C. jejuni CRISPR array preservation and the development of GBS needs to be veri ed by an experimental study.
Overall, this study reveals that preservation of the CRISPR array in C. jejuni may provide an additional evolutionary bene t during the immunopathogenesis of GBS, which establishes a link between CRISPR array preservation and the risk of disease development. Further studies of larger, multicentric cohorts and genomic analysis of the entire C. jejuni CRISPR-Cas system using an experimental study design are required to con rm the role of the C. jejuni CRISPR array in bacterial virulence and the pathogenesis of GBS.

Study population
We isolated 30 C. jejuni strains from the stools specimen of patients with GBS and 52 C. jejuni strains from ethnically matched healthy controls as part of a prospective case-control study at the icddr,b. In addition, we included 60 historical C. jejuni strains from patients with enteritis in this study. All patients with GBS ful lled the National Institute of Neurological Disorders and Stroke (NINDS) diagnostic criteria for GBS 50 . This study was reviewed and approved by the Ethics Committee of the icddr,b, Dhaka, Bangladesh, and all participants provided written informed consent prior sample collection. All methods were carried out in accordance with relevant guidelines and regulations.
All con rmed strains were stored at -80°C in brain heart infusion broth containing 15% glycerol until analysis 30 .

Extraction of bacterial genomic DNA
Bacterial DNA was extracted from colonies collected from the blood agar plates using Qiagen Genomic DNA puri cation kits, according to the manufacturer's instructions (Qiagen, Venlo, the Netherlands). DNA samples were stored at -20°C prior CRISPR typing.

Determination of C. jejuni CRISPR-Cas components
The components of the C. jejuni CRISPR-Cas system were detected by PCR. The rst forward primer for the CRISPR array, F1 (see Supplementary Table S3 online), was designed to target the left anking region (leader sequence) of the CRISPR array to detect con rmed or true CRISPR array loci 28 . The second forward primer for the CRISPR array, F2 (see Supplementary Table S3 Supplementary Table S3 online and 40 sec extension at 72°C, and a nal extension step at 72°C for 10 min. PCR products were visualized by electrophoresis on 1.5% agarose gels using a Molecular Imager® Gel Doc™ XR + system (Bio-Rad Laboratories Inc., Hercules, CA, USA).

CRISPR-based genotyping
Initial subtyping of the C. jejuni isolates was based on the CRISPR array length determined by PCR screening 52,53 . The CRISPR array length was calculated by subtracting 155 from the size of the PCR product ampli ed using the F2 primer (Fig. 4). The number of spacers and DRs were determined from the size of the PCR products. The repeat number is always one greater than the spacer number, as each spacer is anked by two DRs 54 . C. jejuni strains containing at least one spacer sequence in their CRISPR array were classi ed as having a preserved CRISPR array; otherwise, the strain was termed degenerate 41 .
The PCR products ampli ed from 20 GBS-, 29 enteritis-and 21 healthy control-related CRISPR array preserving C. jejuni isolates were puri ed using the ExoSAP-IT® PCR Product Cleanup Reagent (Affymetrix, Cleaveland, OH, USA) at 37°C for 15 min, followed by 15 min inactivation at 80°C. Cycle sequencing was carried out using the BigDye Terminator Kit (Perkin-Elmer Applied Biosystems, Foster City, CA, USA) and the nal puri ed products were loaded onto an ABI 3500xL Automated Genetic Analyzer (Perkin-Elmer Applied Biosystems) to determine the spacer polymorphism-based CRISPR type (CT). Each unique spacer sequence was given an allele number based on its chronology of discovery (see Supplementary Table S2 online). Each unique spacer arrangement was annotated as a unique CRISPR type (CT). The sequences of the C. jejuni CRISPR arrays were deposited in GenBank under accession numbers MW561358 to MW561427.
CRISPR array sequence analysis CRISPR array sequences were detected using the web-based software CRISPRFinder (https://crispr.i2bc.paris-saclay.fr/Server/) 55 and queried using the web tool CRISPRTarget (http://crispr.otago.ac.nz/CRISPRTarget/crispr_analysis.html) to detect target proto-spacers in the spacer sequences using the GenBank-Phage, Refseq-Plasmid and Refseq-Viral databases 56 . The CRISPR array sequences were also screened using the Nucleotide BLAST web tool (https://blast.ncbi.nlm.nih.gov/Blast.cgi) to determine novel CTs 57 . Phylogenetic analysis was performed using the UPGMA (unweighted pair group method with arithmetic mean) function of MEGA X software 58 .

Statistical analysis
The discriminatory power of the various methods of CRISPR typing were compared by calculating   UPGMA dendrogram (prepared using MEGA X software) of C. jejuni CRISPR arrays from 20 strains isolated from patients with GBS, 29 strains isolated from patients with enteritis and 21 strains isolated from healthy controls. Clusters of GBS-related strains are indicated by red branches; clusters of enteritis and healthy control-related strains are indicated by green branches Figure 3 Sequence logo of the 3downstreampro → -spacersequences, ∈ dicat ∈ g3-NNNNAYAC-5` is the putative proto-spacer adjacent motif (PAM) of the Cas9 endonuclease in all 70 CRISPR array-containing C. jejuni strains isolated from patients with GBS, enteritis and healthy control Figure 4 Schematic illustration of the CRISPR array structure and the targeting primers. R indicates the consensus direct repeat (DR) of the C. jejuni CRISPR array (36 bp); S indicates a spacer sequence (30 bp), which is always anked by two DRs. The C. jejuni DR-speci c forward primer (F2) is 25 bases long and overlaps the rst 13 bases of the DR

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.