Creating a programmable gene-silencing system using Type I-C CRISPR for Neisseria gonorrhoeae

We designed the original Type I-C CRISPRi delivery vector to introduce components of the N. lactamica Type I-C Cascade system into the Gc genome using DNA transformation (Fig. 1A). This vector encodes the Cascade complex (cas5, cas8c, cas7, internal cas11 embedded within cas8, and a CRISPR array) adjacent to the lac regulatory sequences (lacI and the lac promoter/operator) (38), allowing for IPTG regulation of Cascade expression. The Nla I-C CRISPR array contains a 32 base pair (bp) repeat, 35 bp gene-specific spacer, and the leader sequence. The Type I-C CRISPRi delivery vector lacks the cas3 gene, rendering this an interference system. The delivery vector encodes an erythromycin resistance cassette to select transformants that integrated the Type I-C CRISPRi system into the Gc chromosome using the flanking homology to the lctP and aspC genes in the chromosomal complementation plasmid pGCC4 (39) (Fig. 1B). However, the transformation efficiency of the Type I-C CRISPRi construct into Gc was low (~3 × 10⁻⁸ transformants/CFU).

To allow for ease of cloning, we generated a minimal single-repeat, single-spacer (R-S) as opposed to the repeat-spacer-repeat (R-S-R) array. We found that the R-S and R-S-R spacer arrays displayed similar Nla Type I-C-mediated knockdown magnitude when targeting the lacI gene in an Escherichia coli platform (Fig. S1). This result indicates that the minimal spacer array is sufficient for knockdown with this system.

To allow the efficient reprogramming of the gene-specific spacer/guide sequence, we designed a new vector called the Spacer Switch Plasmid containing ~950 bp of the lctP gene and ~100 bp of the cas7 gene to allow replacing of the existing CRISPR array with a new one (Fig. 1B). Flanked by these homology arm regions are a chloramphenicol resistance cassette and the CRISPR array with new spacer(s). Selection for transformants on erythromycin and chloramphenicol replaces the resident spacer with the new spacer(s) (Fig. 1C). Transformation of this spacer switch plasmid occurs at an efficiency of 2 × 10⁻³ chloramphenicol-resistant transformants/CFU. We can also transfer the gDNA from a strain carrying the CRISPRi locus to any other Gc strain via DNA transformation, since using chromosomal DNA results in an efficiency of 2 × 10⁻² erythromycin-resistant transformants/CFU, which allows for general use of the CRISPRi system in Gc.

To determine whether our CRISPRi platform can be used to knockdown a highly expressed gene, we programmed NlaCascade to target the opaD gene in an OpaD constitutively expressing, non-variable strain (OpaD_nv) (40). The opaD gene is 1 of 11 opa loci that, in total, encode 7–9 unique Opa proteins. The opa genes are non-essential and all independently phase variable due to pentameric 5′CTCTT′3 repeats within the signal sequence of the open reading frame (ORF), which promote slipped-strand mispairing during DNA replication (41). These frameshifts result in the independent "on" or "off" expression of each Opa variant. The OpaD_nv strain was generated in a genetic background with in-frame, markerless deletions for all 11 opa gene loci (40). In this background, a full-length opaD gene with point mutations in the pentameric repeat region that prevents phase variation was introduced, resulting in the constitutively "on" expression of OpaD. OpaD expression imparts an opaque, dark-hued colony morphology (39).

We identified an opaD-specific protospacer sequence 3′ to a 5′TTC′3 PAM site and is 40 bp upstream of the opaD start codon (Fig. 2A). This spacer sequence overlaps the −10 element of the opaD promoter (42) and is designed to bind to the coding strand relative to the opaD mRNA transcript. To quantify opaD gene repression, we performed quantitative reverse transcription-PCR (qRT-PCR) and found a corresponding change in opaD transcripts with an IPTG dose response (Fig. 2B). Relative to the parental strain, the CRISPRi-opaD strain displayed a significant (~10-fold) decrease in opaD transcript with growth on 0.01 mM IPTG, with repression magnitude leveling out at higher IPTG concentrations. IPTG had no effect on opaD expression in the parental strain. Furthermore, OpaD_nv strains carrying the CRISPRi machinery expressing a spacer targeting the replication origin region (ori), a spacer that does not target the Gc chromosome (non-targeting), or no spacer (null) did not display any reductions in opaD gene expression (Fig. S2). These results show that the downregulation in opaD expression is specific for the targeting spacer sequence.

We used Western blot to analyze OpaD protein expression (Fig. 2C). As previously reported, the Opaless strain, deleted for all 11 opa loci, showed no OpaD expression. The CRISPRi-opaD strain displayed similar levels of OpaD expression as the parental strain following growth without IPTG. Addition of IPTG into the growth medium slightly decreased OpaD expression to intermediate levels at 0.005 mM IPTG but significantly decreased protein levels at 0.01 mM IPTG. No OpaD expression was detected at 0.05 mM, 0.1 mM, and 1.0 mM IPTG. These data show that this Type I-C CRISPRi system can be used to effectively knockdown expression of a highly expressed protein.

Gc interactions with PMNs are central to disease pathogenesis, with patient-derived purulent exudate samples commonly containing mixtures of bacteria associated with PMNs (43, 44). These interactions are mediated through CEACAM-Opa contact, where OpaD mediates Gc binding to the PMN surface via CEACAM1 or CEACAM3 interaction to elicit an oxidative burst (45, 46). We quantified binding of OpaD expressing Gc to CEACAM1 using imaging flow cytometry in which 4′,6-diamidino-2-phenylindole (DAPI)-labeled Gc was incubated with the GST-tagged recombinant N-terminus of CEACAM1 (47). As previously shown, the OpaD_nv strain displayed higher binding than the Opaless strain (76% vs 2.7%) (Fig. 3A). At 0 mM IPTG, 68% of the CRISPRi-opaD cells were AF488 positive. The percentage of AF488-positive cells dropped slightly to 61% following growth on 0.005 mM IPTG but significantly to 44% on 0.01 mM IPTG. Following growth on 0.05, 0.1, and 1.0 mM IPTG, the binding decreased to approximately 3%. These results show that CRISPRi-mediated repression of opaD results in a tunable regulation of Gc-CEACAM1 binding.

We tested whether reducing OpaD protein expression in the CRISPRi-opaD strain would affect PMN reactive oxygen species (ROS) generation in an IPTG-dependent manner using a luminol-dependent chemiluminescent assay (48). Consistent with previous reports, OpaD_nv incubation with PMNs resulted in a strong ROS response, and the Opaless strain elicited no detectable ROS (Fig. 3B). PMN ROS generation in response to the CRISPRi-opaD strain was slightly diminished compared to the OpaD_nv strain in the absence of IPTG, suggesting that there may be a small repression even without IPTG. This small repression effect in the absence of IPTG was also apparent in Fig. 3A. Increasing the IPTG concentration to 0.005 mM, 0.01 mM, and 0.05 mM IPTG corresponded to a further reduction in ROS release. No chemiluminescence was detected when Gc was grown with more than 0.05 mM IPTG. These experiments provide further evidence that the Type I-C CRISPRi system can repress gene expression in a concentration-dependent manner and regulate PMN responses.

We next tested whether the Type I-C CRISPRi system can be used to simultaneously repress multiple genes from a single CRISPR array. We synthesized a spacer switch vector with five unique spacers predicted to together target all 11 opa genes (Fig. 4A), and each spacer targeted at least two opa genes. We introduced this array into strain N-1-60, as this strain encodes all 11 opa loci. This array was designated CRISPRi-opaMS (Opa multi-spacer).

Every opa gene was repressed in the CRISPRi-opaMS strain following growth on 1.0 mM IPTG (Fig. 4B). The range in gene repression varied from 140-fold for opaF/H to 2.5-fold for opaE, and the difference in repression was significantly greater in opaF/H compared to opaB/G, opaI, opaA, opaK and opaD, and opaE.

While all 11 opa genes are transcribed (41), there are differences in transcript abundance (49). We postulated that the range in CRISPRi-mediated opa gene repression could be explained by differences in basal opa gene expression. To calculate basal expression, we compared the cycle threshold (Ct) value of each opa gene to the housekeeping gene rmp at 0 mM IPTG. There was a partial correlation between the basal level of opa gene expression and knockdown magnitude (Fig. S3A). The opaE gene was both the lowest expressed and minimally repressed gene. The opaF/H genes showed the second highest level of expression and the highest level of repression. Similarly, the opaA, opaB/H, and opaD genes all showed intermediate levels of repression and intermediate levels of basal expression relative to the other opa genes. However, these correlations between basal gene expression and CRISPRi repression did not hold true for all the opa genes. The opaI gene showed the highest level of transcripts but was only the fourth most repressed gene, and the opaK gene had the third highest gene expression but showed the sixth lowest level of repression. Thus, basal gene expression is one of the parameters explaining the differences in gene repression by the CRISPRi-opaMS strain, but there are likely other factors as well.

We also hypothesized that the number of spacers targeting each opa gene in the CRISPRi-opaMS array could explain the range in opa gene repression magnitudes. We generated strains encoding the single targeting spacers derived from the CRISPRi-opaMS array (Fig. S3B) and compared the opa gene repression in these single-spacer constructs to the CRISPRi-opaMS five-spacer array (Fig. 4A). We first tested the three opaF/H spacers (Spacer 2, Spacer 3, and Spacer 5) individually and found the CRISPRi-opaMS strain displayed significantly greater repression magnitude than the single-spacer targeting constructs (Fig. S3C). While the Spacer 2 and Spacer 5 single targeting constructs resulted in 25- and 85-fold gene repression, respectively, the Spacer 3 construct only repressed opaF/H twofold. In contrast, repression by the full CRISPRi-opaMS array resulted in a 181-fold opaF/H repression. The other opa genes displayed similar repression patterns by the single-spacer constructs compared to the full CRISPRi-opaMS array. These results indicate that genes are repressed at greater magnitudes by multiple spacers than by a single targeting spacer and that some predicted targeting spacers are ineffective at repression. Overall, the difference in opa gene repression by the CRISPRi-opaMS strain is likely due to differences in basal gene expression and the number of targeting spacers per gene.

We next quantified the duration of CRISPRi repression following IPTG removal. We targeted the pilE gene, which is not essential in vitro and encodes the major subunit of the type four pilus (Tfp). Expression of the Tfp is necessary for transformation competence, so we expected CRISPRi-mediated pilE repression to diminish DNA transformation (50–53). To control for Tfp antigenic variation, we performed these experiments in N-1-60, an FA1090 strain derivative defective for pilE antigenic variation and pilC phase variation (53). We used pilus-dependent colony morphology (54) to determine whether the CRISPRi-pilE strain yields a nonpiliated colony morphology when induced with IPTG. While uninduced CRISPRi-pilE colonies had a morphology consistent with piliated Gc (55), CRISPRi-pilE Gc plated on 0.1 mM IPTG showed colony morphology consistent with nonpiliated cells (Fig. 5A).

We used DNA transformation competence as a proxy for pilE expression status and measured the restoration of DNA competence over time following the removal of IPTG from the growth medium. To do this, we grew the CRISPRi-pilE strain overnight on gononoccal medium (GCB) plates containing 0 or 0.1 mM IPTG and transferred the bacteria from both conditions into IPTG-free liquid media. At 30-minute time points, we added pSY6, a plasmid containing a gyrB gene with a point mutation that confers nalidixic acid resistance once recombined with the chromosomal gyrB gene (56). Bacteria grown overnight without IPTG showed transformation of the mutant gyrB throughout the 4-hour time course (Fig. 5B). Gc grown overnight on 0.1 mM IPTG did not yield detectable numbers of Nal-resistant progeny at 0 and 30 minutes post IPTG removal. After 60 minutes of IPTG depletion, the Gc grown overnight on 0.1 mM IPTG began to display transformation competence, and the efficiency of transformation continued to increase at the 1.5-, 2.0-, and 2.5-hour time points. By 3 hours, the Gc achieved the same DNA transformation efficiency as the strain grown without IPTG. These results show that the Cascade stably bound to pilE promoter begins to turnover and lose repression ability of this target by 60 minutes.

We determined the effect of CRISPRi expression on Gc fitness by measuring and comparing the CFU/colony for the OpaD_nv and the OpaD_nv/CRISPRi-opaD, OpaD_nv/CRISPRi-ori, and OpaD_nv/CRISPRi-null strains on media with or without 1.0 mM IPTG. While there was no difference in growth among the strains grown with 0 mM IPTG (Fig. 6A), the addition of 1.0 mM IPTG to the growth medium slightly but significantly affected the growth of the CRISPRi-expressing strains (Fig. 6B). This effect occurred regardless of whether there was a spacer that directed the CRISPRi machinery to the genome. From these data, we conclude that the expression of the cas genes of Nla Cascade components causes a small fitness defect that should be accounted for when using Gc-CRISPRi with high IPTG concentrate.

One important use for the Type I-C CRISPRi system would be to regulate the expression of essential genes. To test this possibility, we generated N-1-60 strains where the CRISPRi system targeted dnaE (CRISPRi-dnaE) and dnaQ (CRISPRi-dnaQ), essential components of the DNA polymerase III holoenzyme (57). When the CRISPRi-dnaE and CRISPRi-dnaQ strains were grown without IPTG, both strains displayed normal growth (Fig. 7). Growth with IPTG in the media resulted in small reduction in colony-forming units per milliliter (CFU/mL) for both strains within 1 hour of CRISPRi induction. The viability of both strains decreased significantly over a 5-hour time course, reflecting the turnover of the DNA polymerase III complex. These results indicate that the Type I-C CRISPRi system can be used to repress essential gene expression in Gc.

Strains and their details are listed in Table S1. All strains are FA1090 derivatives, either in the OpaD_nv or N-1-60 backgrounds (40, 53). The OpaD_nv is a phase-locked OpaD constitutively expressing strain in an opaA-K deleted (Opaless) background (40). The N-1-60 strain contains multiple point mutations in the pilE G4 and is pilC2 phase-locked "on" with the 1-81-S2 pilin variant (53). Strains were grown on Gonococcal (Gc) Medium Base (BD Difco) (36.25 g/L), agar (1.25 g/L), Kellogg Supplement I (22.2 mM glucose, 0.68 mM glutamine, 0.45 mM cocarboxylase), and Kellogg Supplement II [1.23 µM Fe(NO₃)₃] (54). Gc was grown at 37°C at 5% CO₂. For liquid growth, Gc was grown in GCBL medium consisting of 15 g/L of Bacto protease peptone 3, 23 mM potassium phosphate dibasic, 7.35 mM potassium phosphate monobasic, and 17.11 mM sodium chloride, supplemented with Kellogg Supplement I and II and 5 mM sodium bicarbonate.

The cas5, cas8c, and cas7 genes were amplified from N. lactamica strain ATCC23970 gDNA using primers containing PacI and PmeI restriction cut sites (Table S1) (38). The amplified fragment was inserted into the PacI/PmeI-digested pGCC4 plasmid (designated pYZ1342) to allow insertion of the CRISPRi locus into an intragenic site of the Gc chromosome (34). To generate the opaD CRISPR array, two oligo pairs oYZ2266 + oYZ2267 and oYZ2268 + oYZ2269 were annealed separately. These two annealed oligonucleotides were ligated to the BsaI-cut-CRISPRi delivery vector (pYZ1389) by T4 ligase (NEB). Transformants were screened by PCR and Sanger sequencing. Transformation efficiency of Gc was obtained by dividing the calculated CFU grown on selective media by the total number of bacteria.

This assay was conducted as described earlier (92), except that the strains used do not contain the third pBAD-derived plasmid encoding anti-CRISPR genes. To construct the R-S-R SCANR spacer array, oligos oYZ3160 and oYZ3161 were ligated together in BsaI-cut pYZ1386, underwent PCR with primers oYZ22110 and oYZ2469, and then Gibson assembled into the Nla Cascade-containing pYZ1432 plasmid. The R-S SCANR array was constructed by Gibson assembling the oligo oYZ5355 into pYZ1432. To carry out the E. coli CRISPRi repression assay, an overnight culture of each final strain (3.0 mL Luria broth [LB] with 20 µg/mL chloramphenicol and 50 µg/mL kanamycin) was pelleted and resuspended in 1× PBS at OD₆₀₀ of 1.0 and imaged using a ChemiDoc MP imager (Bio-Rad) on the Alexa488 channel for 0.036 s. These samples were then quantified for OD₆₀₀-normalized GFP fluorescence using a microplate reader and plotted as described earlier (92).

SDS-PAGE and immunoblotting for Opa expression were conducted as described previously (93). Lysates were harvested from overnight lawns on GCB with various IPTG concentrations. Samples were normalized in PBS + 5 mM MgSO₄ to equivalent optical densities and lysed in sample buffer containing 12 mM Tris pH 6.8, 5% glycerol, 0.4% SDS, 1% β-mercaptoethanol, and 0.02% bromophenol blue. Lysates were boiled for 5 minutes and sheared with a syringe and 25-gauge needle (BD 305122). Bands were resolved using pre-cast 4–20% Criterion TGX (Bio-Rad 5671094) protein gels and transferred in CAPS methanol buffer to polyvinylidene difluoride membranes. Blots were blocked in PBS + 0.5% Tween-20 then incubated with mouse 4B12 pan-Opa antibody (94) and detected using a secondary antibody, goat anti-mouse H + L conjugated to DyLight 800 (Thermo Fisher Scientific PISA510176) (94). The loading control was detection of the uniformly expressed cytoplasmic protein Zwf by rabbit antiserum (95) and secondary antibody goat anti-rabbit H + L conjugated to Alexa Fluor 680 (Invitrogen A21109). Fluorescent blots were imaged with a LI-COR Odyssey CL-X, and relative band intensities were quantified with Image Studio software.

GST-tagged recombinant N-terminal domains of human CEACAM1 were purified and utilized as previously described (47). Gc was grown overnight with various IPTG concentrations, and 1 × 10⁸ bacteria were incubated with GST-tagged N-CEACAM1 for 30 minutes at 37°C with end-over-end rotation. Gc was washed and stained to detect the presence of CEACAM with a mouse anti-GST antibody p1A12 (BioLegend 640802) followed by an Alexa Fluor 488 conjugated goat anti-mouse antibody (Jackson ImmunoResearch 115-545-164). Gc was then resuspended in 2% paraformaldehyde containing DAPI DNA stain. To process and analyze N-CEACAM1 binding, ImageStream X Mk II imaging flow cytometer with INSPIRE and IDEAS v. 6.2 software was used. In each replicate, single-color Gc was used to create the respective data set's compensation matrix. Bacteria were gated by DAPI positivity, singlet size, focused images, and Alexa Fluor 488 positivity.

All human subjects research was conducted in accordance with a protocol approved by the University of Virginia Institutional Review Board for Health Sciences Research (#13909). Neutrophils were isolated from venous blood collected from healthy human subjects via dextran sedimentation followed by a Ficoll gradient as described previously (96). PMNs were resuspended in Dulbecco's phosphate-buffered saline (without calcium and magnesium; Thermo Fisher) with 0.1% dextrose and used within 2 hours after isolation.

Neutrophils were resuspended in Morse's defined medium containing 20 µM luminol (97). Gc was added at an MOI of 100, and the generation of reactive oxygen species was measured every 3 minutes for 1 hour via chemiluminescence using a Victor3 Wallac luminometer (98). For an additional negative control, uninfected neutrophils were used as before (40).

Bacteria were grown overnight on GCB supplemented with or without IPTG (Sigma-Aldrich) for 18–20 hours and swabbed into Bacteria RNA Protect (Qiagen), pelleted, and frozen at −80°C. RNA was isolated via the RNeasy Mini Kit (Qiagen) and treated with DNase I (Promega) to remove contaminating gDNA. RNA was incubated with Random Hexamers (Promega) and subsequently reverse transcribed with the SuperScript III Reverse Transcriptase (Thermo Fisher). A no-RT control was made for each sample to ensure that there was minimal to no gDNA contamination. The qPCR primers used for this study were designed with the IDT Primer Design tool and are listed in Table S1. qPCR was performed using iTaq Universal SYBR Green Supermix (Bio-Rad) in a Bio-Rad iQ5 Real-Time Detection system. The Gc gene, rmp, was used as an internal control (99). Gene expression change was calculated using the 2^−ΔΔCT cycle threshold method (100). qRT-PCR experiments were conducted following the Minimal Information for Publications on Quantitative Real-Time PCR Experiments guidelines (101).

For liquid medium growth, Gc was first grown overnight on solid media at 37°C at 5% CO₂ for 19 hours, then swabbed, resuspended into 10-mL GCBL with Kellogg supplements and sodium bicarbonate, and grown for 2 hours. Cells were then diluted to an OD₅₅₀ of 0.1 and grown for 2 hours to allow for consistent exponential growth. Following these two growth periods, the cells were diluted again to an OD₅₅₀ of 0.1 and treated with or without IPTG. At 1-hour time points, samples were taken and diluted for CFU/mL calculation. For solid growth, bacteria were plated at equal colony densities on GCB supplemented with 0 or 1.0 mM IPTG for 19 hours. At 2-hour increments, three colonies per plate were separately isolated via Whatman disks (Sigma-Aldrich) and suspended into 0.5-mL GCBL. Suspensions were vortexed and then diluted for CFU/colony calculations.

The DNA transformation efficiency was measured as published previously (56). Bacteria were swabbed from overnight growth, resuspended into GCBL supplemented with 5 mM sodium bicarbonate, and diluted to an OD of 0.15. Each sample was supplemented with various IPTG concentrations and grown for 4 hours to allow for pilE repression. At 30-minute time intervals, 30 µL was sampled and incubated at 37°C for 20 minutes with 100 ng of the pSY6 plasmid, which is a DNA containing a gyrB mutation rendering nalidixic acid resistance upon transformation (56). After 20 minutes, the samples were treated with 1 U DNase I for 10 minutes to degrade non-transformed DNA. The bacteria were then grown in GCBL for 2 hours at 37°C before being serially diluted and plated for CFU/mL on GCB in the presence or absence of 2 µg/mL nalidixic acid (Dot Scientific).

In all experiments, excluding the oxidative burst assay, the reported results are depicted as the means ± standard errors for three independent experiments or more. Statistics were calculated using GraphPad Prism analysis software; a P value of >0.05 was considered significant. The Student's t test and analysis of variance (one-way and two-way) were used as appropriate to determine significance, as specified in figure legends.