Main

Prokaryotes use clustered regularly interspaced short palindromic repeats and CRISPR-associated protein (CRISPR–Cas) systems to recognize and cleave foreign nucleic acid sequences to protect against phages and other mobile genetic elements1,2,3. However, with a few exceptions4,5, little is known about the regulation of these systems under physiological conditions. The limited understanding of CRISPR–Cas regulation is partly attributable to the absence of native CRISPR–Cas activity in cultured Enterobacteriaceae, a commonly studied family of bacteria, necessitating investigation using artificial overexpression systems6,7,8.

Citrobacter rodentium is a Gram-negative bacterial pathogen that naturally infects and causes colitis in mice9. Like most Enterobacteriaceae, C. rodentium is a facultative anaerobe, capable of growth in the presence or absence of oxygen (oxic or anoxic conditions, respectively). We performed RNA sequencing of the pathogen’s transcriptional response to anoxia and observed more transcripts from the type I-E cas locus (schematized in Fig. 1a) in anoxic versus oxic culture conditions (Fig. 1b, Extended Data Fig. 1 and Supplementary Table 1). To test whether increased cas expression correlates with CRISPR–Cas activity, we designed a functional assay10 that monitors the retention frequency of either a plasmid containing a sequence that is (target) or is not (control) recognized by the native C. rodentium CRISPR locus (plasmid retention assay; schematized in Extended Data Fig. 2).

Fig. 1: Anoxia causes Fnr-dependent activation of CRISPR–Cas immunity in C. rodentium.
figure 1

a, A schematic of the C. rodentium cas locus. b, RNA sequencing (RNA-seq) results from the cas locus of C. rodentium cultured under oxic or anoxic conditions for 3.5 h on solid LB agar. Data represent three biological replicates. FDR, false discovery rate. Additional analysis in Extended Data Fig. 1 and Supplementary Table 1. c,d, Fraction of cells from a single colony cultured for 24 h on solid LB agar that retained the CRISPR–Cas target or control plasmid. Assay design in Extended Data Fig. 2. Geometric mean of biological replicates. For c: 3 colonies each. For d: wild-type (WT), 5 colonies; Δcas3-2, Δfnr, and Fnr-site mut. (with mutation displayed at the bottom of f), 6 colonies. e, InducTn-seq volcano plot comparing fold change in the gene insertion frequency between cells that retained the CRISPR target or control plasmid. Points represent individual genes. P value from two-sided Mann–Whitney U test. Additional data in Supplementary Table 1. f, Fnr-binding sites. Top: consensus defined in E. coli15. Bottom: putative C. rodentium WT and mutated Fnr-binding sites relative to the cas3 start codon. g, qPCR of cas3 following 3.5 h of culture on solid LB agar. ΔΔCT (threshold cycle) analysis compared with rpoA and oxic culture conditions. Data represent three biological replicates (plates) with three technical replicates per sample. h,i, Consensus motif28 (h) and location relative to cas3 start codon (i) of putative Fnr-binding sites upstream of Enterobacteriaceae cas3 orthologues.

Source data

Full size image

In oxic culture conditions, 92% of cells retained the target plasmid, indicating that CRISPR–Cas immunity is inactive (Fig. 1c). By contrast, only 1% of cells retained the target plasmid during anoxic culture (Fig. 1c), demonstrating anoxic-specific immunity. To verify that this assay requires CRISPR–Cas activity, we deleted the entire C. rodentium cas locus (∆cas3-2) and repeated the assay. Deletion of cas3-2 eliminated CRISPR–Cas immunity during anoxic culture (Fig. 1d). We conclude that anoxia is required for both expression and activity of C. rodentium CRISPR–Cas immunity.

To discover anoxic regulators of the C. rodentium cas locus, we leveraged the plasmid retention assay for a transposon-insertion loss-of-function screen (InducTn-seq11). We anoxically cultured a transposon mutant population of C. rodentium carrying the target plasmid and then sequenced the mutants that retained the plasmid following antibiotic selection, comparing with a population containing a control plasmid (Fig. 1e and Supplementary Table 2). As expected, transposon insertions in the endonuclease cas3 were specifically enriched in the population retaining the target plasmid, indicating that disruption of cas3 prevents CRISPR–Cas immunity. By contrast, cas1 and cas2, which are not involved in CRISPR–Cas interference, were not enriched, demonstrating the specificity of the assay. One of the two most enriched non-cas-related genes was fnr, which encodes an oxygen-responsive transcriptional regulator widely conserved among facultative anaerobic bacteria12 (Fig. 1e). The other most enriched gene was a chaperone involved in the maturation of iron–sulfur cluster-containing proteins (hscA), previously demonstrated to be needed for full Fnr activity13. Based on these data, we hypothesized that Fnr regulates CRISPR–Cas expression.

In support of this hypothesis, CRISPR–Cas immunity was eliminated in the absence of fnr (Δfnr; Fig. 1d). Mutation of a putative Fnr-binding motif14,15 centred 69.5 nucleotides upstream of cas3 (Fig. 1f) also eliminated CRISPR–Cas immunity (Fig. 1d). Furthermore, quantitative PCR (qPCR) demonstrated that mutation of the Fnr-binding site upstream of cas3 eliminated the transcriptional response of cas3 to anoxia (Fig. 1g). These results indicate that Fnr directly activates CRISPR immunity during anoxia.

To determine the conservation of Fnr-mediated regulation of CRISPR–Cas immunity, we used OrthoDB16 to select 501 non-redundant Gammaproteobacteria genomes containing cas3 orthologues and interrogated the 300 nucleotides upstream of cas3 with motif enrichment analysis14. In total, 141 of 501 genomes contained at least one predicted Fnr-binding site upstream of cas3 (Supplementary Tables 3 and 4). These genomes were distributed among most orders of Gammaproteobacteria (Extended Data Fig. 3). Notably, the Fnr-binding sites in Enterobacteriaceae closely matched the Fnr-consensus sequence previously defined in Escherichia coli strain MG1655 (ref. 15) and were primarily centred at the same position as in C. rodentium (Fig. 1h,i and Extended Data Fig. 4). The positional conservation of an Fnr-binding motif in 13 out of 32 Enterobacteriaceae suggests conserved Fnr-dependent cas3 activation within a subset of this family.

Many of the Enterobacteriaceae with an Fnr-binding motif upstream of cas3 have been isolated from the mammalian intestine (for example, Escherichia, Citrobacter and Klebsiella), which is frequently an anoxic environment17. We hypothesized that activation of CRISPR–Cas immunity by anoxia may protect Enterobacteriaceae from threats encountered within the microbially rich intestine. Consistent with this hypothesis, RNA sequencing revealed that faecal-associated C. rodentium isolated from infected C57BL/6J mice had more transcripts from the cas locus than in oxic culture (Fig. 2a, Extended Data Fig. 5 and Supplementary Table 1).

Fig. 2: C. rodentium CRISPR–Cas immunity is activated by Fnr within the murine intestine.
figure 2

a, Cas transcripts from RNA sequencing (RNA-seq) of C. rodentium recovered from the faeces of infected, female, C57BL/6J mice 7 days after inoculation compared with bacteria from oxic culture. Data represent three biological replicates from mice and three from oxic culture. Additional analysis in Extended Data Fig. 5 and Supplementary Table 1. b, Retention of a CRISPR target plasmid by the indicated strains of C. rodentium measured by serial dilution and plating of faeces from infected mice (N = 16 wild type (WT), 8 ∆fnr, 8 ∆cas3-2 infected mice; equal mix of sexes). Lines represent the geometric mean, and shading represents the geometric standard deviation. Significance compared with wild-type by two-way analysis of variance with Dunnett’s multiple-comparison test on log10-transformed data (**P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001).

Source data

Full size image

To determine if increased cas expression results in CRISPR–Cas immunity within the intestine, we infected mice with C. rodentium strains carrying the CRISPR–Cas target plasmid and monitored plasmid retention in faecal bacteria. Over the first 24 h, most shed bacteria retained the plasmid (Fig. 2b). Subsequently, wild-type C. rodentium progressively lost the plasmid, with only 1% of shed bacteria retaining the plasmid 13 days after inoculation. By contrast, 85% of ∆cas3-2 cells and 23% of ∆fnr cells retained the plasmid over the same 13-day period (Fig. 2b). These results suggest that anoxia within the intestine results in Fnr-dependent activation of CRISPR–Cas immunity in C. rodentium. Furthermore, the discrepancy between the ∆cas3-2 and ∆fnr mutants suggests that intestinal signals beyond anoxia may regulate CRISPR activity—potentially by relieving H-NS-mediated transcriptional repression, as previously observed in other bacterial species7,8,18.

The intestine contains low concentrations of oxygen and dense communities of microbes and phage17,19. Density-dependent regulation is consistent with previous reports of quorum-sensing-regulated CRISPR–Cas activity in Serratia spp., Pseudomonas aeruginosa and Aliivibrio wodanis19,20,21. We therefore propose that, in dense microbial communities such as the host intestine, anoxic regulation of CRISPR–Cas immunity in C. rodentium and other Enterobacteriaceae represents an adaptation that protects these bacteria against predation.

Methods

Regulatory statement

All bacterial work was performed in biosafety level 2 facilities at the Brigham and Women’s Hospital according to protocols reviewed and approved by the Brigham and Women’s Hospital Institutional Biosafety Committee under protocol 2011B000082. All personnel working with bacteria were trained in relevant safety and protocol-specific procedures. Animal studies were conducted at Brigham and Women’s Hospital in compliance with the ‘Guide for the Care and Use of Laboratory Animals’ and according to protocols reviewed and approved by the Brigham and Women’s Hospital’s Institutional Animal Care and Use Committee under protocol 2016N000416.

Bacterial strains

The strains used in this study are listed in Supplementary Table 5. C. rodentium is a spontaneous streptomycin-resistant isolate of strain ICC168 (ref. 22), previously known as Citrobacter freundii biotype 4280 (ATCC 51459). Escherichia coli strain MFDpir was used for cloning23.

Oxic and anoxic culture

Bacteria were cultured at 37 °C in either liquid lysogeny broth (LB) shaking at 200 rotations per minute or on solid LB containing 1.5% agar. Oxic culture was performed in atmospheric conditions. Anoxic culture was performed in a Baker Concept 400M anaerobic workstation set to 0% oxygen, with media acclimated to the anoxic environment for at least one night.

Plasmid assembly

Plasmid fragments were amplified from plasmid or genomic DNA using single-stranded DNA primers (Integrated DNA Technologies). Fragments were assembled with NEBuilder HiFi DNA Assembly Master Mix (New England Biolabs). Assembled plasmids were transformed into E. coli strain MFDpir with electroporation and transferred to the recipient strains by conjugation.

Constructing mutant C. rodentium strains

The allelic exchange protocol from ref. 24 was used to create in-frame deletions, and the Fnr-site mutation in C. rodentium. pTOX5 (Genbank MK972845) was linearized with the restriction enzyme SwaI and assembled with ~1-kb homology arms flanking the desired mutation. Primer sequences are included in Supplementary Table 6. For deletions, two to three codons were left intact at both ends of the deletion. This plasmid was electroporated into MFDpir, checked by PCR and conjugated into C. rodentium. Transconjugants were purified by plating consecutively three times. Individual colonies were cultured for 1 h without antibiotic selection and then counterselected to isolate double crossovers lacking the plasmid backbone. Single colonies were selected, and PCR and whole-genome sequencing confirmed the identity and fidelity of the mutant strains.

Animal experiments

Adult (9–12 weeks) C57BL/6J mice were purchased from Jackson Laboratory (strain #000664) and acclimated for at least 72 h before experimentation. During infection, mice were housed under specific pathogen-free conditions at 68–75 °F, with 30–50% humidity and a 12-h light/dark cycle in a biosafety level 2 facility.

For C. rodentium infections, mice were deprived of food for 3–5 h before inoculation. Animals were then mildly sedated with isoflurane, and 100 µl of the indicated strain suspended in phosphate-buffered saline (PBS) was inoculated into the stomach with a sterile feeding needle (Cadence Science). Dose (streptomycin-resistant colony-forming units, CFU) and CRISPR plasmid retention (gentamicin-resistant CFU) were determined retrospectively by serial dilution and plating.

Animal health was monitored during infection by measuring weight, body condition and faecal appearance. C. rodentium colonization and plasmid retention were monitored by sampling faeces from infected animals. Fresh faecal pellets were suspended in sterile PBS and homogenized in a bead beater (BioSpec Products) with 3.2-mm stainless-steel beads. C. rodentium concentration (CFU g−1) and plasmid retention were determined by serial dilution and plating for CFUs.

RNA sequencing

For cultured samples, C. rodentium was grown overnight in liquid LB in oxic conditions. A total of 4 × 107 CFU from the stationary phase culture were seeded onto LB agar plates and cultured for 3.5 h in the presence or absence of oxygen. Cells were immediately diluted in two parts Qiagen RNAprotect Bacteria Reagent and stored at −80 °C until processing.

For faecal samples, female mice were infected with 5 × 109 CFU of C. rodentium, and colonization was monitored in the faeces to ensure engraftment. Seven days after inoculation, fresh faeces from infected mice were submerged in RNAprotect Bacteria reagent and manually disrupted with a sterile rod to release bacteria. To remove large debris and eukaryotic cells, samples were passed through a 5-µm filter and bacteria were frozen at −80 °C until processing.

RNA was released from bacteria using lysozyme and proteinase K digestion. RNA was extracted with a Qiagen RNeasy kit and purified with RNA Clean & Concentrator-5 (Zymo Research). SeqCenter performed library preparation and sequencing using the following method, provided by SeqCenter: “Samples were DNAse treated with Invitrogen DNAse (RNAse free). Library preparation was performed using Illumina’s Stranded Total RNA Prep Ligation with Ribo-Zero Plus kit and 10 bp unique dual indices (UDI). Sequencing was done on a NovaSeq X Plus, producing paired end 150 bp reads. Demultiplexing, quality control, and adapter trimming was performed with bcl convert.”

RNA sequencing data was processed using CLC Genomics Workbench (Qiagen). RNA-Seq Analysis 2.8 parameters: reference – ICC168 reference genome; mismatch cost = 2; insertion cost = 3; deletion cost = 3; length fraction = 0.8; similarity fraction = 0.8; global alignment; strand specific = reverse; max hits per read = 10; count paired reads as two = no; ignore broken pairs. Differential gene expression was determined with Differential Expression for RNA-Seq 2.8. Results are included in Supplementary Table 1.

Plasmid retention assay

The assay is schematized in Extended Data Fig. 2. The CRISPR-target plasmid contains the gentamicin resistance gene aaC1 and a protospacer adjacent motif (PAM; AAG) followed by a protospacer sequence matching the native CRISPR array (ATCTGTTTATAGCTGGCTATAAAATTTATAAA). The control plasmid is identical except that the protospacer sequence is replaced by a protospacer recognized by the E. coli MG1655 CRISPR–Cas system (GCAACGACGGTGAGATTTCACGCCTGACGCTG), and not the C. rodentium native CRISPR–Cas system.

For plasmid retention assays in culture, strains carrying the target or control plasmids were outgrown overnight in an oxic environment on solid LB plates with gentamicin. The next day, strains were restreaked onto LB plates without antibiotics and cultured in the presence or absence of oxygen for 24 h. Single colonies were resuspended in sterile PBS, and serial dilution was used to determine the fraction of the population that retained the plasmid (gentamicin-resistant divided by streptomycin-resistant CFU).

For plasmid retention assays during infection, an equal mix of male and female mice were inoculated with ~5 × 109 CFU of the indicated strain. Plasmid retention was measured in the inoculum and faeces throughout the infection.

InducTn-seq

Control or CRISPR-target plasmids were conjugated into a C. rodentium InducTn-seq mutant library11. A total of 3 × 108 transconjugants were expanded in oxic conditions on LB containing gentamicin, to select for the plasmid, and arabinose, to induce further miniTn5 transposition. The mutant libraries were stored at −80 °C in PBS with 20% glycerol. Subsequently, 5 × 107 CFU of the mutant libraries were seeded onto LB agar plates and cultured under anoxic conditions for 24 h. The population was then expanded in oxic conditions on LB plates containing gentamicin to select for mutant cells that retained the plasmid during anoxic culture. Cells were frozen at −80 °C until processing.

Sequencing libraries were prepared with the protocol from ref. 11. Genomic DNA was extracted using a DNeasy Blood and Tissue Kit (Qiagen) and sheared to approximately 400 bp using a M220 ultrasonicator (Covaris). The fragmented DNA was then end-repaired using the Quick Blunting Kit (NEB), polyadenylated with Taq polymerase and dATP, and Illumina P7 adapters were ligated using T4 DNA ligase (NEB). The end of the miniTn5 transposon within the integrated InducTn-seq vector was removed by double restriction enzyme digestion followed by SPRIselect size-selection. Transposon-adjacent sequences were amplified from 800 ng of DNA by PCR using an Illumina i7 index sequence on the reverse primer. Primer dimers were removed by size selection, and samples were sequenced on a NextSeq 1000 (Illumina).

InducTn-seq data were analysed with the protocol from ref. 11 using Python. MiniTn5 transposon-insertion frequency was compared between populations that retained the control or target plasmids. Significance was measured with the non-parametric Mann–Whitney U statistical test with Benjamini–Hochberg multiple testing correction. Results are included in Supplementary Table 2.

qPCR

Bacteria were cultured in liquid LB in oxic conditions. Approximately 107 CFU from the culture were seeded onto LB agar plates and cultured for 3.5 h in the presence or absence of oxygen. After culture, cells were diluted immediately in two parts Qiagen RNAprotect Bacteria reagent and frozen at −80 °C until processing.

RNA was released from bacteria using lysozyme and proteinase K digestion, and extracted using an RNeasy kit (Qiagen). qPCR was performed using the Luna Universal One-Step RT-qPCR kit on a StepOnePlus Real-Time PCR System (Applied Biosystems). At least three biological and three technical replicates were included per sample, with primers targeting both rpoA and cas3 transcripts. Primer sequences are included in Supplementary Table 6.

qPCR data were analysed by comparative critical threshold (CT) analysis. The average cas3 CT of three technical replicates was first normalized to the CT of rpoA from the same sample. Then, the CT was normalized using the cas3 CT from oxic culture, producing ΔΔCT.

Phylogenetic analysis of cas3 orthologues

In total, 578 cas3 orthologues from 500 Gammaproteobacteria were selected for analysis by OrthoDB (version 12.0)16. We added E. coli strain EDL933 to this list. Strains are listed in Supplementary Table 3. To construct a phylogeny, we retrieved the nucleotide sequence of dnaA from the National Center for Biotechnology Information (NCBI) for 482 of the Gammaproteobacteria and used MAFFT (strategy: FFT-NS-2; model: DNA200; v7.526)25 to align the sequences, FastTree (model: Jukes-Cantor with CAT rate heterogeneity; v2.1.11)26 to construct the phylogeny and iTOL (Interactive Tree of Life; version 7.1)27 to create a visualization.

For motif enrichment analysis, we retrieved 300 bp upstream of cas3 from NCBI for 553 of the 579 cas3 orthologues. These sequences were compared with the prokaryotic transcription factor motif database PRODORIC (release 2021.9) with Simple Enrichment Analysis (SEA; version 5.5.7)14 with the following parameters: differential enrichment analysis; both strands; Fisher exact test; control sequences from shuffled sequences, preserving 3-mer frequencies; hold-out 10% of sequences. This analysis determined that the Fnr motif defined in E. coli strain MG1655 (ID MX000004) was significantly enriched within the dataset (P = 6.87 × 10−5). The location relative to cas3 and the scores of putative Fnr-binding sites are included in Supplementary Table 4. The consensus logo of putative Enterobacteriaceae Fnr-binding sites was created with WebLogo version 2.8.2 (ref. 28).

Software and statistics

Data analysis was performed using CLC genomics workbench (version 24.0.1), GraphPad Prism (version 10.4.1), Python (version 3.12) and Microsoft Excel. The number of samples and statistical tests are described in the figure legends. Graphics were prepared with GraphPad Prism and Microsoft PowerPoint. Reads were mapped to the C. rodentium ICC168 genome FN543502.1.

Licence information

This Brief Communication is subject to the Howard Hughes Medical Institute (HHMI)’s Open Access to Publications policy. HHMI laboratory heads have previously granted a non-exclusive CC BY 4.0 licence to the public and a sublicensable licence to HHMI in their research articles. Pursuant to those licences, the author-accepted manuscript of this Brief Communication can be made freely available under a CC BY 4.0 licence immediately upon publication. This Brief Communication is the result of funding in whole or in part by the National Institutes of Health (NIH). It is subject to the NIH Public Access Policy. Through acceptance of this federal funding, the NIH has been given the right to make this Brief Communication publicly available in PubMed Central upon the Official Date of Publication, as defined by the NIH.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.