CRISPR interference in a Streptococcus agalactiae Multi-locus Sequence Type 17 Strain

34 Group B Streptococcus (GBS), a common colonizer of the human genital and gastrointestinal tracts, is a leading 35 cause of neonatal bacterial meningitis, which can lead to severe neurological complications. The hypervirulent 36 serotype III, sequence type 17 (ST-17) strain COH1 is strongly associated with late-onset disease due to its unique 37 set of virulence factors. However, genetic manipulation of ST-17 strains is notoriously challenging, limiting the ability 38 to study key pathogenic genes. In this study, we developed a CRISPR interference (CRISPRi) system utilizing an 39 endogenous catalytically inactivated Cas9 (dCas9) in the COH1 strain, enabling targeted and tunable gene 40 expression knockdown. We confirmed the efficacy of this system through hemolysis assays, qPCR transcriptional 41 analysis, and in vitro infection models using human brain endothelial cells. The CRISPRi system successfully 42 produced phenotypic knockdowns of essential virulence genes, including pilA, srr2, and iagA, reducing adhesion, 43 invasion, and inflammatory responses at the blood-brain barrier. This platform enables rapid gene knockdowns for 44 functional genomics in ST-17 GBS, enabling high-throughput screening and pathogenesis research. 45 Importance 46 Group B Streptococcus (GBS) remains the world’s leading cause of neonatal meningitis. GBS-host interactions at the 47 blood-brain barrier (BBB) are dependent on bacterial factors, including surface factors and two-component systems. 48 Multi-locus sequence type 17 (ST-17) GBS strains are highly associated with neonatal meningitis, and these strains 49 harbor many virulence factors for infection at the BBB. Historically, these factors have been studied using traditional 50 knockout mutagenesis, which has proven challenging in the most common ST-17 lab strain, COH1. This study 51 utilizes CRISPR interference (CRISPRi) to generate rapid expression knockdown. This study validates a CRISPRi-52 enabled COH1 dCas9 strain as a versatile tool for probing GBS pathogenesis at the BBB. 53

54 Group B Streptococcus (GBS), also known as Streptococcus agalactiae, is a Gram-positive bacterium that 55 asymptomatically colonizes the genital and gastrointestinal tracts of 20-30% of healthy individuals (1, 2). However, 56 during or shortly after birth, GBS can opportunistically infect neonates and infants, leading to conditions such as 57 sepsis, pneumonia, or meningitis (3-6). Worldwide, GBS is the leading cause of neonatal bacterial meningitis, which 58 is uniformly fatal without medical intervention (1, 2). While modern medical interventions have transformed GBS 59 meningitis from a uniformly fatal illness to an often curable one, mortality remains at 5% to 10%, with survivors facing 60 long-term neurological sequelae such as blindness, deafness, seizure, and stroke (1, 7-9). Additionally, GBS can 61 asymptomatically spread to other body sites and cause disease in adults, including soft tissue infections, bacteremia, 62 urinary tract infections, endocarditis, and meningitis (3-5). Neonatal invasive GBS disease is classified into two types: 63 early-onset disease (EOD), which occurs within the first seven days of life, and late-onset disease (LOD), which 64 develops between 7 and 90 days of life. 97% of neonatal invasive GBS diseases are caused by serotypes I–V, and 65 serotype III accounts for 43% of early-onset disease and 73% of late-onset disease (1, 5, 8). LOD presentation is 66 most commonly meningitis, but can also present as urinary tract, joint, bone, and soft tissue infection, as well as 67 pneumonia and bacteremia (8, 10, 11). 68 Some GBS strains are more strongly associated with neonatal disease than others. Many strains that fall within 69 serotype III sequence type 17 (ST-17), considered a hypervirulent sequence type, are significantly associated with 70 neonatal meningitis and particularly LOD (5, 8). A significant contributing factor to ST-17’s hypervirulence is the 71 number of adhesins and virulence factors shared across this clade. These include the serine-rich repeat proteins (Srr-72 1/2), GBS pilus tip adhesin (PilA), hypervirulent GBS adhesin (HvgA), laminin binding protein (Lmb), streptococcal 73 fibrinogen binding protein A (SfbA), and Group B streptococcal surface protein C (BspC), as well as the two-74 component systems CovR/S and CiaR/S, which have all been described previously through traditional allelic-75 exchange mutagenesis or transposon mutagenesis (2, 12-24). However, mutagenesis can prove laborious and time-76 consuming, particularly in ST-17 strains such as COH1, which are notoriously difficult to manipulate. 77 GBS utilizes an endogenous Type II-A CRISPR-Cas9 system, much like S. pyogenes, and this system has recently 78 been used to screen candidate genes using CRISPR interference (CRISPRi) (25). This process involves site-directed 79 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2025. ; https://doi.org/10.1101/2025.08.27.672580doi: bioRxiv preprint mutagenesis of key catalytic residues RuvC-like and HNH domains (D10 & H845, respectively), which yields the 80 ability to produce targeted, tunable expression knockdown in any given candidate gene containing the appropriate 81 NGG protospacer adjacent motif (PAM) sequence (25, 26). Until now, this has only been done in serotype Ia and 82 serotype V strains, which, while valid for their backgrounds, are not as translatable to study at the blood-brain barrier 83 (BBB). We hypothesize that such a system in COH1 will provide a faster path to in vitro loss-of-function phenotypes, 84 facilitating the speed of research on genes of interest. We report here the use of our hCMEC/D3 brain endothelial cell 85 model to study infection, specifically quantification of bacterial adhesion, invasion, and chemokine expression after 86 knocking down various virulence genes. Additionally, in the process of generating the COH1 dCas9 mutant for this 87 paper, we report the first use of a Cas12a-based system in the generation of a markerless point mutation in GBS. 88 These findings will be of use in facilitating GBS research. 89

90 Generation of Streptococcus agalactiae str. COH1 dCas9 91 To utilize CRISPRi in GBS, we opted to make use of the endogenous GBS cas9 gene, point-mutating it to generate a 92 COH1 strain with catalytically deadened Cas9 (dCas), into which we transform a single plasmid expressing the single 93 guide RNA (sgRNA) used to target dCas9 to the site of interest in the genome (Figure 2A) (25, 27). The mutant was 94 generated using two approaches, with the first point mutation inactivating the HNH-like domain (H845A) generated 95 using classical allelic exchange methods, and the second point mutation inactivating the RuvC-like domain (D10A) 96 generated with a Cas12a-based gene deletion method (25, 26, 28, 29). The Cas12a mutagenesis process required 97 targeting of the genome for cutting, and presentation of a repair cassette on pGBSedit that carried our target (D10A) 98 mutation as well as silent mutations altering the sgRNA target site and PAM to protect any successfully edited 99 mutants (Figure 1A). For these silent mutations, we selected codons frequently used by GBS to avoid introducing 100 rare codons that might alter gene expression (30). This is the first instance of this system being used to generate a 101 markerless point mutation in GBS (Figure 1A). Mutagenesis was confirmed via Oxford Nanopore whole-genome 102 sequencing (Plasmidsaurus) to verify the correct genotype (Figure 1B) (31). No growth kinetics changes were 103 observed with this mutant (Supp. Figure 1). 104 105 Verification of Knockdown Phenotypes 106 To confirm that the mutant strain would show an altered phenotype when targeted with an sgRNA, we performed a 107 hemolysis assay utilizing sgRNAs targeted to the cyl operon and the known virulence repressor covR, which 108 regulates cyl expression (2). To verify reduced hemolytic activity, a 1% red blood cell (RBC) solution was aliquoted 109 into a 96-well plate, infected, and then combined with a PBS suspension of a GBS + sgRNA transformant. Two 110 guides were used for each target gene to evaluate the potential for modulation of the knockdown phenotype. Control 111 groups were instead treated with an equal volume of PBS or with .1% TritonX-100, as well as an experimental control 112 carrying a “scramble” sgRNA sequence that lacks a genomic target in COH1 and should therefore produce no 113 change in phenotype. Triton-X is a detergent that will lyse the RBCs nearly completely, providing a maximum lysis 114 level to normalize data. After a one-hour incubation, cell debris was pelleted, and supernatant was transferred to a 115 new 96-well plate, and absorbance was read at 415 nm to read hemoglobin release as a proxy for the degree of RBC 116 lysis. Results were normalized to the reads of Triton-X-treated samples. Almost all the GBS-infected wells showed 117 some degree of lysis compared to PBS mock wells. The strains harboring cyl-targeted guides showed reduced 118 hemolysis relative to the scramble control, and the covR-targeted strains showed increased lysis, demonstrating that 119 the guides and CRISPRi system produce expected phenotypes (Figure 2B). 120 Next, to verify the possibility of generating knockdowns of essential genes, a noted strength of CRISPRi, an sgRNA 121 was produced to target the essential gene ccpA. After the transformation of this sgRNA into GBS, a growth curve was 122 performed to verify a change in growth kinetics, and the difference was further quantified using an area under the 123 curve (AUC) analysis. Again, we achieved the expected phenotype, as the bacterial lag phase was extended by 124 multiple hours (Figure 2C), and area under the curve analysis showed a reduction in AUC (Figure 2D). This further 125 validates the CRISPRi system for reducing essential gene function without being lethal to the bacteria (25, 26). 126 Taken together, the CRISPRi system produced targeted loss-of-function phenotypes as intended. 127 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2025. ; https://doi.org/10.1101/2025.08.27.672580doi: bioRxiv preprint qPCR Verification of Transcriptional Alteration 128 As CRISPRi reduces expression at the transcript level, qPCR was used to verify that the observed phenotypes were 129 attributed to a reduction in targeted mRNA abundance. RNA was isolated via bead beating and standard lysis 130 protocols, and cDNA was synthesized to be used in a qPCR reaction. Knockdown verification was performed using 131 the previously mentioned genes covR, cylE, and ccpA, as well as in the established virulence factors srr2, iagA, and 132 pilA. In each case, sgRNA targeted genes showed reduced expression levels (Figure 3), though some variability was 133 apparent, with distance from transcriptional start not always trending with expression reduction. The primers used are 134 reported in Table S1. 135 CRISPRi for in vitro Infection Studies 136 As GBS COH1 is associated with neonatal meningitis, we sought to ensure that knockdowns of known factors in the 137 CRISPRi strain would produce reproducible infection phenotypes in established in vitro blood-brain barrier modeling 138 systems. For adhesion/invasion assays and cell immune response verification, the hCMEC/D3 cell line was used. For 139 investigating the efficacy of the CRISPRi model in infection, sgRNAs were designed to target the known GBS 140 virulence factors pilA, srr2, and iagA. These genes all have verified roles in infection at the BBB in vitro and with in 141 vivo mouse models, as well as further investigation of the molecular mechanisms of these roles in infection (2, 14, 23, 142 24, 32). Adhesion (reported as cell-association) and invasion results were quantified by infecting hCMEC/D3 cells 143 seeded on a 12-well plate for 30 minutes and 4 hours, respectively. All knockdowns exhibited a reduction in both 144 adhesion (Figure 4A) and invasion (Figure 4B) relative to scramble. To investigate the known contribution of PilA to 145 immune activation, we performed an experiment using mock-infected cells, infected with scramble, or infected with a 146 PilA-targeted knockdown and collected RNA to quantify changes in chemokine and cytokine expression. We found 147 that the scramble control significantly increased chemokine and cytokine expression when compared to mock, and 148 that knocking down PilA reduced the overall induction of these proinflammatory genes (Figure 4C-E). Taken 149 together, these results demonstrate that use of the CRISPRi system mimics phenotypes previously described using 150 classical mutagenesis, thereby providing an opportunity to use CRISPRi to identify potential novel targets in the 151 future. 152 In silico GBS CRISPRi sgRNA Library Generation 153 Following the generation of the CRISPRi strain and verification of its utility, we compiled an in silico high coverage 154 sgRNA library for use in GBS infection research using CRISPRi. We have designed a double-coverage library of 155 potential sgRNAs using the software tool CHOPCHOP, covering 1944/2073 GBS genes (omitting rRNA and tRNA 156 genes), totaling 3595 sgRNA sequences (Supp. Table 2) to be used as a resource for GBS research. Each of these 157 guides has been generated to minimize off-target effects and consistency of knockdown effects, and evaluated for 158 homology against several other common GBS research strains, including A909, CNCTC 10/84, NEM316, BM110, 159 and CJB111 (Supp. Table 2). This in silico library can facilitate future GBS phenotypic research by speeding design 160 of synthetic protospacer oligonucleotides and could potentially inform development of a full-genome coverage 161 knockdown library in GBS. 162

163 CRISPRi is an increasingly popular technology for precise, tunable gene regulation, and it has been utilized across a 164 wide range of studies in both eukaryotic and prokaryotic systems. By using a catalytically inactive Cas9 (dCas9) 165 targeted to specific genomic sequences via single-guide RNAs, CRISPRi enables researchers to inhibit transcription 166 without permanently altering the genome, making it a powerful tool for investigating gene function (25, 26, 29, 33). 167 CRISPRi applications in GBS are advancing, with new tools for GBS rapidly being developed (25, 29). These 168 innovations create opportunities to advance the study of clinically relevant strains such as COH1 as well as the study 169 of complex host-pathogen interactions. 170 171 Our main priority in designing this mutant-based endogenous CRISPRi system was ease of use, leveraging tools 172 established in other GBS strains while balancing the knockdown efficiency. To that point, every guide tested in this 173 study produced a viable knockdown phenotype. Notably, these knockdowns were typically achieved, depending on 174 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2025. ; https://doi.org/10.1101/2025.08.27.672580doi: bioRxiv preprint transformation success, within 1-2 weeks from design to performing assays. There was, however, some variability in 175 some of the knockdown phenotypes that is worth noting, particularly in the case of the covR knockdowns. 176 177 Generally, the results were as expected, with cyl operon knockdowns showing reduced or nearly eliminated lytic 178 activity and covR knockdowns showing increased lysis relative to the “scramble” control sgRNA group (25, 34-36). 179 Additionally, variation in efficacy from guides was seen. This is expected and desired for regulation purposes via 180 varying the target site distance from the transcriptional start site (TSS), but other variations were observed. For 181 instance, the cyl guide targeted 283 bp from TSS showed a greater degree of knockdown than the guide targeted 384 182 bp from TSS (Figure 2A), in agreement with an established correlation (26). The covR guide pair seems to disagree 183 with this; however, more factors may be in play as the less effective covR 227 guide targets an AGG PAM sequence, 184 which has been reported elsewhere as being a less efficient PAM sequence in GBS (25). 185 186 Also of great importance is the viability of the dCas9 mutant as a tool for the study of COH1 infection modeling in 187 vitro. We and others utilize several models to this end, with the iBEC and hCMEC/D3 models increasing the 188 robustness of the findings (37-39). Three well-established virulence genes were selected for knockdown: pilA, iagA, 189 and srr2, as there is a great deal of literature to which we can compare our knockdown data for verification purposes. 190 The data presented here compares favorably to knockout phenotypes, though less stark in some cases (Figure 4A, 191 B, F) (2, 14, 23, 32). Additionally, we performed qPCR and found that CMEC/D3 cells had a reduced inflammatory 192 response to knockdown strains of virulence factors, with neutrophil recruitment markers (IL-8, CXCL1, and CXCL2) 193 expressed at lower levels after infection with the pilA knockdown strain compared to the control (Figure 4C–E). 194 195 While the system shows strong potential utility, some limitations are worth noting, particularly given the use of a 196 mutated endogenous Cas9 gene rather than purely plasmid-based systems of CRISPRi. The foremost of these is that 197 the system is not placed under an inducible promoter, leaving any knockdown ubiquitously expressed. This limits 198 utility and versatility, as inducible systems would permit a more nuanced experimental design. Additionally, antibiotic 199 selection is necessary to retain the plasmid, which is a limitation for in vivo applications. It is also possible that there 200 are unexpected non-transcriptional effects from mutating endogenous Cas9 rather than a plasmid system. With that 201 said, this is unlikely as previous literature has shown minimal transcriptional change in GBS dCas9 strains relative to 202 wild-type, as PAM scanning is unaffected by this mutation (25). The generation of the in silico sgRNA library may be 203 of use to other researchers studying GBS gene function, facilitating faster screening of genes and providing key 204 validation before investing the time and resources to generate mutants. Future innovation on these systems may 205 permit more complex studies, such as CRISPRi-seq, permitting the discovery and interrogation of new genes of 206 interest in GBS COH1. 207 208

209 Maintenance and differentiation of cell lines 210 hCMEC/D3 cells were cultured as described previously (40). Cells were maintained on tissue culture flasks coated 211 with 1% rat tail collagen in EndoGRO MV medium (Millipore-Sigma). Cells were grown until 85% confluent, then split 212 for infection experiments. Cells were seeded onto 12-well tissue culture plastic dishes coated with 1% rat tail at 1 x 213 105 cells/cm2 for experiments and allowed to grow to confluency (4 days) at 37°C + 5% CO2 before infection 214 experiments were conducted. 215 Bacterial strains and growth conditions 216 Group B Streptococcus (GBS; Streptococcus agalactiae) strain COH1 (serotype III, multi-locus sequence type 17 217 (MLST-17)) was used for this study and cultured in Todd-Hewitt broth (THB) at 37°C in static culture (41, 42). E. coli 218 strain DH5α competent cells were used as a reservoir for plasmids, and were grown in Luria-Bertani (LB) broth (LB 219 Miller formulation) at 37°C. 220 sgRNA design and protospacer cloning 221 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2025. ; https://doi.org/10.1101/2025.08.27.672580doi: bioRxiv preprint Guides are designed as described previously to maximize efficiency (26, 33). Great care was taken during design to 222 avoid off-target activity, avoiding seed-sequence homology wherever possible. The software tool CHOPCHOP was 223 used for sgRNA design with that in mind, and NCBI BLAST was used extensively to verify sgRNA efficacy. Guides for 224 cyl and covR were received from Dr. Thomas Hooven’s lab at the University of Pittsburgh Medical Center, as was the 225 sgRNA shuttle vector p3015b (25). sgRNA sequences are ordered from Eurofins as individual oligos with desired 226 overhangs matching to vector p3015b insertion site, then end phosphorylated (New England Biolabs T4 227 polynucleotide kinase). Phosphorylated oligos are annealed by heating to 95 °C for 5 minutes, followed by gradual 228 cooling to 4°C to produce the desired spacer. p3015b is miniprepped from DH5α E. coli before digestion with BsaI 229 (Eco31I) (Thermofisher). The digested plasmid sample is cleaned (Promega Wizard SV Gel and PCR Cleanup Kit), 230 then the annealed spacer is ligated into the digested shuttle vector (New England Biolabs T4 Ligase) for heat-shock 231 transformation into E. coli. Following E. coli transformation, the plasmid is miniprepped (ThermoFisher PureLink™ 232 Quick Plasmid Miniprep Kit) for transformation into GBS. 233 In silico GBS sgRNA Library Design 234 In general, guides were designed as described above. For library generation, wherever possible, two sgRNAs were 235 generated for every gene (omitting rRNAs and tRNAs) in the COH1 genome. For each gene, one sgRNA was 236 designed near the transcription start site (TSS) (-100 bp to 300 bp from TSS) and another further downstream 237 (>500bp downstream of TSS). Wherever possible AGG PAM sequences were avoided due to the noted reduced 238 efficiency of this PAM in GBS (25). The silico tool CHOPCHOP was used for sgRNA generation, as it is designed to 239 help minimize off-target effects. Following sgRNA sequence generation, the best sequences were then checked for 240 homology to other common GBS lab strains (A909, CNCTC 10/84, NEM316, BM110, CJB111), and sgRNAs that 241 would be useful in other strains were chosen wherever possible. This library is provided in Supplemental Table 2. 242 GBS competent cell preparation and transformation 243 Electrocompetent GBS were prepared as described previously, with minor adaptations (43-45). COH1 were grown 244 overnight in a 15 mL THB + 0.6% glycine broth at 37°C. The following day, this culture was diluted in an additional 35 245 mL of THB + 0.6% Glycine and grown to OD600= .6. This culture was pelleted via 4°C centrifugation at 3200 x g and 246 washed twice on ice with a cold 0.6% glycine solution. The remaining pellet was then resuspended in a 400 µL 247 solution of cold 25% PEG 6000 + 10% glycerol and used immediately. Electroporation was also performed as 248 described previously, with only minor alterations (43-45). When electroporating the bacteria, three 3 kV pulses were 249 performed with 5 second intervals between them. Bacteria were then permitted to recover for two hours in THB + 250 25% PEG 6000 in static culture before plating on THB + 5 µg/mL erythromycin agar plates. 251 Generation of GBS dCas9 mutations 252 Point mutations needed for catalytic deadening of Cas9 were performed in two steps due to the size of the gene and 253 the genomic distance between the two catalytic sites. The H845A missense mutation was generated as previously 254 described using allelic exchange mutagenesis (45). The sucrose-sensitive suicide vector pMBSacB, containing the 255 necessary homology repair cassette with the point mutation, was generously provided by Dr. Thomas Hooven’s lab at 256 the University of Pittsburgh Medical Center. This plasmid was transformed into GBS as described above, and after 257 screening via colony PCR, the transformants were grown overnight in THB + 5ug/mL erythromycin broths at 28 °C. 258 The following day, these broths were passed to fresh tubes of the same broth at 37 °C to begin temperature selection 259 of the first allelic crossover. Following PCR verification of the first allelic crossover event, broths of this strain were 260 passed into broths of THB lacking erythromycin to eliminate selection and permit curing of the plasmid. These broths 261 were repeatedly passed at 28°C and plated onto THB until colony PCR, using primers 750 base pairs up-and-262 downstream of the homology arm sites, could confirm that a second allelic crossover event occurred. The product of 263 this colony PCR was then inserted into a TOPO TA Vector (TOPO TA Cloning Kit for Sequencing, Thermofisher) and, 264 after being passed in DH5α E. coli, submitted to Plasmidsaurus for Oxford Nanopore Sequencing to confirm 265 generation of the first point mutation. 266 The second point mutation was generated using the Cas12a expressing vector pGBSedit, which was received from 267 Dr. Thomas Hooven’s lab at the University of Pittsburgh Medical Center. A 23 bp protospacer targeted to the Cas9 268 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2025. ; https://doi.org/10.1101/2025.08.27.672580doi: bioRxiv preprint gene was designed using the appropriate TTTV PAM. This was assembled into pGBSedit (NEB HiFi Assembly 2x 269 Master Mix), and then the homology repair cassette harboring the D10A point mutation was designed and also 270 assembled into the vector. In order to prevent self-cutting of either the plasmid or the final, mutated genome by 271 Cas12, the repair cassette also carries a series of silent mutations in the TTTV PAM used for targeting as well as the 272 targeted sequence itself. The remainder of the process was performed as previously described by Hillebrand et al 273 (29). The plasmid is electroporated into GBS, and, after screening and re-streaking, a colony is inoculated into THB + 274 5 µg/mL erythromycin broth and grown overnight. This broth is then sub-cultured, and eventually 250ng/mL 275 anhydrotetracycline is added, followed by a one-hour incubation period. The culture is then plated onto THB agar + 276 5ug/mL erythromycin + 250ng/mL anhydrotetracycline agar plates and allowed to grow overnight. Mutant PCR 277 screening for point mutation is performed by designing a primer to match the silent mutated sgRNA targeting 278 sequence of the mutant gene as the forward primer and using a primer 750 base pairs outside the homology region 279 as the reverse. PCR was performed using the Monserate Midas Quik-Load PCR Mix 2X (2004-G). To ensure 280 specificity for the point mutation, it is necessary to avoid using any polymerase mixes that contain isostabilizing 281 compounds. Raising the annealing temp 3-5°C above the norm for Taq PCR may also be necessary. The PCR-282 confirmed mutant was then re-streaked, cured of the plasmid by passes in media without erythromycin, and then the 283 genome was sent to Plasmidsaurus for whole-genome sequencing to confirm the desired genotype. 284 GBS infection assays 285 hCMEC/D3 or iBEC cells were seeded onto 12-well plates. Before infection assays, GBS dcas9 knockdowns were 286 grown overnight at 37°C in Todd Hewitt Broth (THB), supplemented with 5µg/mL erythromycin to aid in selection for 287 the p3015b vector containing the protospacer. From the overnight culture, bacteria were subcultured into fresh THB + 288 5µg/mL erythromycin and grown to OD600= 0.4-0.6. Bacteria were spun down and resuspended in PBS to OD600= 0.4. 289 Bacteria were then diluted 1:10 in EC medium to a multiplicity of infection (MOI) of 10. Adherence and invasion 290 assays were conducted following previously described protocols (46, 47). 291 For the bacterial adherence assays, bacteria and cells were incubated for 30 minutes at 37oC + 5% CO2. The cells 292 were then washed 5x with PBS to remove non-adherent bacteria, lysed with 0.025% Triton X-100, diluted in PBS, 293 and plated onto THB + 5µg/mL erythromycin plates. Plates were incubated overnight at 37°C, with colonies being 294 quantified the following day. For the invasion assays, bacteria and cells were incubated for 2 more hours before 295 adding 100 µg/mL gentamicin. After 2 more hours of incubation at 37°C + 5% CO2, cells were washed 3x with PBS to 296 remove antibiotics and plated onto THB plates + 5µg/mL erythromycin plates. Plates were incubated overnight as 297 above for quantification. Cell-association and invasion were quantified with the formula 298 (#𝐶𝐹𝑈∗𝑑𝑖𝑙𝑢𝑡𝑖𝑜𝑛 𝑐𝑜𝑟𝑟𝑒𝑐𝑡𝑖𝑜𝑛∗𝑣𝑜𝑙𝑢𝑚𝑒 𝑐𝑜𝑟𝑟𝑒𝑐𝑡𝑖𝑜𝑛) (𝐼𝑛𝑝𝑢𝑡 𝐶𝐹𝑈/𝑤𝑒𝑙𝑙) and normalized to MOI association/invasion rates. The true MOI was 299 determined for each experiment. 300 Hemolysis assays 301 Sheep red blood cells (RBCs) were prepared by washing 5 mL of defibrinated blood (Hemostat Labs) with an equal 302 volume of Hank’s Balanced Salt Solution (HBSS) (VWR)at 4°C. The RBCs were pelleted by centrifugation at 500 × g 303 for 15 min at 4°C and resuspended in 5 mL of HBSS. This washing step was repeated three times. A 0.1% Triton X-304 100 in PBS solution was prepared if not previously made to serve as a positive control. The washed RBCs were then 305 pelleted by centrifugation at 500 × g for 15 min at 4°C. The RBC pellet (50 μL) was diluted into 5 mL HBSS to 306 generate a 1% red blood cell suspension (RBCS). 307 A 96-well plate was used to set up the assay. Each well contained 100 μL of the 1% RBCS combined with 100 μL of 308 the bacterial PBS suspensions or control solutions. The plate was incubated at 37°C with 5% CO₂ for 1 hour. 309 Following incubation, the plate was centrifuged at 2000 rpm for 5 min at 4°C. Supernatants (100 μL) were transferred 310 to new wells on the same 96-well plate, and hemoglobin release was measured by recording absorbance at 415 nm 311 using a spectrophotometer. Percent hemolysis was calculated relative to a positive control consisting of RBCs treated 312 with 0.1% Triton X-100 (100 μL), which represented 100% hemolysis. 313 RNA isolation and qPCR 314 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2025. ; https://doi.org/10.1101/2025.08.27.672580doi: bioRxiv preprint When evaluating human cell inflammatory responses to knockdowns, hCMEC/D3 or iBEC cells were seeded onto 12-315 well plates and infected with GBS at an MOI of 10. Immediately after 5 hours of infection, the total RNA was collected 316 using the Macherey–Nagel NucleoSpin RNA kit (Macherey–Nagel). cDNA was synthesized with the qScript cDNA 317 Synthesis kit (Quantabio), followed by SYBR Green (PowerUp SYBR Green Master Mix, Thermo Fisher) qPCR for 318 each of the targets: IL-8, CXCL1, CXCL2, primers all listed in Table S1. 18S was used as the housekeeping gene for 319 human cell lines. 320 For bacterial knockdown qPCR verification, bacteria were grown as described for infection experiments to OD600= .4 321 to ensure mid-log growth phase. Bacteria were resuspended in RA3 lysis buffer, then bead-beaten to lyse before 322 being processed as described above. qPCR data were collected on the QuantStudio3 system (Applied Biosystems), 323 and data are presented as fold change using the delta–delta–CT calculation (48). 324 325 Acknowledgments 326 B.J.K. is supported by the National Institute of Neurological Disorders and Stroke grant R15NS131921, and the 327 National Institute of Allergy and Infectious Diseases grant R03AI185593 and received salary support from both 328 grants. B.J.K. is also supported by startup funding at the University of Texas at Dallas. T.A.H is supported by National 329 Institutes of Allergy and Infectious Disease (NIAID) R21AI178067, R01AI182835, and R01AI177991. G.H.H. is 330 supported by a UPMC Children’s Hospital of Pittsburgh Research Advisory Council graduate student grant. 331 T.A.H and G.H.H. contributed the p3015b plasmid and some of the constructs used in this study. W.D.C wrote the 332 manuscript with assistance from J.F.W. and performed all experiments alongside A.W.F, B.G., A.S. and J.F.W. 333 W.D.C., A.W.F., B.G., B.E., V.S., B.K., K.S., K.S., T.V., A.M., A.G., and K.H. all contributed to sgRNA library 334 generation. 335 336 Conflict of Interest Statement 337 The authors declare that the research was conducted in the absence of any commercial or financial relationships that 338 could be construed as a potential conflict of interest. 339 340 Data availability 341 The sgRNA library has been made available as supplementary table 2. 342 343 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2025. ; https://doi.org/10.1101/2025.08.27.672580doi: bioRxiv preprint

344 1. Armistead B, Oler E, Adams Waldorf K, Rajagopal L. The Double Life of Group B 345 Streptococcus: Asymptomatic Colonizer and Potent Pathogen. J Mol Biol. 2019;431(16):2914–346 31. 347 2. Doran KS, Fulde M, Gratz N, Kim BJ, Nau R, Prasadarao N, et al. Host-pathogen 348 interactions in bacterial meningitis. Acta Neuropathol. 2016;131(2):185–209. 349 3. Puopolo KM, Lynfield R, Cummings JJ. Management of Infants at Risk for Group B 350 Streptococcal Disease. Pediatrics. 2019;144(2). 351 4. Puopolo KM, Madoff LC, Eichenwald EC. Early-onset group B streptococcal disease in 352 the era of maternal screening. Pediatrics. 2005;115(5):1240–6. 353 5. Seale AC, Bianchi-Jassir F, Russell NJ, Kohli-Lynch M, Tann CJ, Hall J, et al. Estimates 354 of the Burden of Group B Streptococcal Disease Worldwide for Pregnant Women, Stillbirths, 355 and Children. Clin Infect Dis. 2017;65(suppl_2):S200–s19. 356 6. Jones DE, Kanarek KS, Lim DV. Group B streptococcal colonization patterns in mothers 357 and their infants. J Clin Microbiol. 1984;20(3):438–40. 358 7. Berardi A, Trevisani V, Di Caprio A, Bua J, China M, Perrone B, et al. Understanding 359 Factors in Group B Streptococcus Late-Onset Disease. Infect Drug Resist. 2021;14:3207–18. 360 8. Creti R, Imperi M, Berardi A, Lindh E, Alfarone G, Pataracchia M, et al. Invasive Group B 361 Streptococcal Disease in Neonates and Infants, Italy, Years 2015-2019. Microorganisms. 362 2021;9(12). 363 9. Mynarek M, Vik T, Andersen GL, Brigtsen AK, Hollung SJ, Larose TL, et al. Mortality and 364 neurodevelopmental outcome after invasive group B streptococcal infection in infants. Dev Med 365 Child Neurol. 2024;66(1):125–33. 366 10. (CDC) CfDCaP. Prevention of perinatal group B streptococcal disease: a public health 367 perspective. Centers for Disease Control and Prevention. Preprint at (1996). 368 11. Berardi A, Rossi C, Guidotti I, Vellani G, Lugli L, Bacchi Reggiani ML, et al. Factors 369 associated with intrapartum transmission of group B Streptococcus. Pediatr Infect Dis J. 370 2014;33(12):1211–5. 371 12. Dammann AN, Chamby AB, Catomeris AJ, Davidson KM, Tettelin H, van Pijkeren J-P, 372 et al. Genome-Wide fitness analysis of group B Streptococcus in human amniotic fluid reveals a 373 transcription factor that controls multiple virulence traits. PLoS Pathog. 2021;17(3):e1009116. 374 13. Deng L, Spencer BL, Holmes JA, Mu R, Rego S, Weston TA, et al. The Group B 375 Streptococcal surface antigen I/II protein, BspC, interacts with host vimentin to promote 376 adherence to brain endothelium and inflammation during the pathogenesis of meningitis. PLoS 377 Pathog [Internet]. 2019 2019/06//; 15(6):[e1007848 p.]. 378 14. Doran KS, Engelson EJ, Khosravi A, Maisey HC, Fedtke I, Equils O, et al. Blood-brain 379 barrier invasion by group B Streptococcus depends upon proper cell-surface anchoring of 380 lipoteichoic acid. The Journal of Clinical Investigation. 2005;115(9):2499–507. 381 15. Hensler M, Liu G, Sobczak S, Benirschke K, Nizet V, Heldt G. Virulence Role of Group 382 BStreptococcusβ‐Hemolysin/Cytolysin in a Neonatal Rabbit Model of Early‐Onset Pulmonary 383 Infection. The Journal of Infectious Diseases. 2005;191(8):1287–91. 384 16. Lembo A, Gurney MA, Burnside K, Banerjee A, De Los Reyes M, Connelly JE, et al. 385 Regulation of CovR expression in Group B Streptococcus impacts blood–brain barrier 386 penetration. Molecular Microbiology. 2010;77(2):431–43. 387 17. Martins ER, Andreu A, Melo-Cristino J, Ramirez M. Distribution of pilus islands in 388 Streptococcus agalactiae that cause human infections: insights into evolution and implication for 389 vaccine development. Clin Vaccine Immunol. 2013;20(2):313–6. 390 18. Mazzuoli MV, Daunesse M, Varet H, Rosinski-Chupin I, Legendre R, Sismeiro O, et al. 391 The CovR regulatory network drives the evolution of Group B Streptococcus virulence. PLoS 392 Genet. 2021;17(9):e1009761. 393 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2025. ; https://doi.org/10.1101/2025.08.27.672580doi: bioRxiv preprint 19. Randis TM, Gelber SE, Hooven TA, Abellar RG, Akabas LH, Lewis EL, et al. Group B 394 Streptococcus -hemolysin/Cytolysin Breaches Maternal-Fetal Barriers to Cause Preterm Birth 395 and Intrauterine Fetal Demise in Vivo. Journal of Infectious Diseases. 2014;210(2):265–73. 396 20. Schubert A, Zakikhany K, Pietrocola G, Meinke A, Speziale P, Eikmanns BJ, et al. The 397 fibrinogen receptor FbsA promotes adherence of Streptococcus agalactiae to human epithelial 398 cells. Infect Immun. 2004;72(11):6197–205. 399 21. Spencer BL, Deng L, Patras KA, Burcham ZM, Sanches GF, Nagao PE, et al. Cas9 400 Contributes to Group B Streptococcal Colonization and Disease. Frontiers in Microbiology. 401 2019;10. 402 22. Tenenbaum T, Bloier C, Adam R, Reinscheid DJ, Schroten H. Adherence to and 403 invasion of human brain microvascular endothelial cells are promoted by fibrinogen-binding 404 protein FbsA of Streptococcus agalactiae. Infect Immun. 2005;73(7):4404–9. 405 23. Banerjee A, Kim BJ, Carmona EM, Cutting AS, Gurney MA, Carlos C, et al. Bacterial Pili 406 exploit integrin machinery to promote immune activation and efficient blood-brain barrier 407 penetration. Nat Commun. 2011;2:462. 408 24. Maisey HC, Hensler M, Nizet V, Doran KS. Group B streptococcal pilus proteins 409 contribute to adherence to and invasion of brain microvascular endothelial cells. J Bacteriol. 410 2007;189(4):1464–7. 411 25. Gopalakrishna KP, Hillebrand GH, Bhavana VH, Elder JL, D’Mello A, Tettelin H, et al. 412 Group B Streptococcus Cas9 variants provide insight into programmable gene repression and 413 CRISPR-Cas transcriptional effects. Communications Biology. 2023;6(1):620. 414 26. Qi LS, Larson MH, Gilbert LA, Doudna JA, Weissman JS, Arkin AP, et al. Repurposing 415 CRISPR as an RNA-guided platform for sequence-specific control of gene expression. Cell. 416 2013;152(5):1173–83. 417 27. Jinek M, Chylinski K, Fonfara I, Hauer M, Doudna JA, Charpentier E. A programmable 418 dual-RNA-guided DNA endonuclease in adaptive bacterial immunity. Science. 419 2012;337(6096):816–21. 420 28. Chua MJ, Collins J. Rapid, Efficient, and Cost-Effective Gene Editing of Enterococcus 421 faecium with CRISPR-Cas12a. Microbiol Spectr. 2022;10(1):e0242721. 422 29. Hillebrand G, Carlin S, Giacobe E, Stephenson H, Collins J, Hooven T. A Cas12a 423 Toolbox for Rapid and Flexible Group B Streptococcus Genomic Editing and 424 CRISPRi. bioRxiv. 2025:2025.06.20.660720. 425 30. Ma YP, Ke H, Liang ZL, Liu ZX, Hao L, Ma JY, et al. Multiple Evolutionary Selections 426 Involved in Synonymous Codon Usages in the Streptococcus agalactiae Genome. Int J Mol Sci. 427 2016;17(3):277. 428 31. Wang Y, Zhao Y, Bollas A, Wang Y, Au KF. Nanopore sequencing technology, 429 bioinformatics and applications. Nature Biotechnology. 2021;39(11):1348–65. 430 32. Six A, Bellais S, Bouaboud A, Fouet A, Gabriel C, Tazi A, et al. Srr2, a multifaceted 431 adhesin expressed by ST-17 hypervirulent Group B Streptococcus involved in binding to both 432 fibrinogen and plasminogen. Mol Microbiol. 2015;97(6):1209–22. 433 33. Qu J, Prasad NK, Yu MA, Chen S, Lyden A, Herrera N, et al. Modulating Pathogenesis 434 with Mobile-CRISPRi. J Bacteriol. 2019;201(22). 435 34. Lamy M-C, Zouine M, Fert J, Vergassola M, Couve E, Pellegrini E, et al. CovS/CovR of 436 group B streptococcus: a two-component global regulatory system involved in virulence. 437 Molecular Microbiology. 2004;54(5):1250–68. 438 35. Jiang S-M, Cieslewicz Michael J, Kasper Dennis L, Wessels Michael R. Regulation of 439 Virulence by a Two-Component System in Group B Streptococcus. Journal of Bacteriology. 440 2005;187(3):1105–13. 441 36. Whidbey C, Burnside K, Martinez RM, Gendrin C, Vornhagen J, Frando A, et al. A 442 Hyperhemolytic/Hyperpigmented Group B Streptococcus Strain with a CovR Mutation Isolated 443 from an Adolescent Patient with Sore Throat. Clin Res Infect Dis. 2015;2(2). 444 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2025. ; https://doi.org/10.1101/2025.08.27.672580doi: bioRxiv preprint 37. Kim Brandon J, Bee Olivia B, McDonagh Maura A, Stebbins Matthew J, Palecek Sean 445 P, Doran Kelly S, et al. Modeling Group B Streptococcus and Blood-Brain Barrier Interaction by 446 Using Induced Pluripotent Stem Cell-Derived Brain Endothelial Cells. mSphere. 447 2017;2(6):10.1128/msphere.00398–17. 448 38. Kim BJ, McDonagh MA, Deng L, Gastfriend BD, Schubert-Unkmeir A, Doran KS, et al. 449 Streptococcus agalactiae disrupts P-glycoprotein function in brain endothelial cells. Fluids and 450 Barriers of the CNS. 2019;16(1):26. 451 39. Weksler B, Romero IA, Couraud PO. The hCMEC/D3 cell line as a model of the human 452 blood brain barrier. Fluids Barriers CNS. 2013;10(1):16. 453 40. Eigenmann DE, Xue G, Kim KS, Moses AV, Hamburger M, Oufir M. Comparative study 454 of four immortalized human brain capillary endothelial cell lines, hCMEC/D3, hBMEC, TY10, 455 and BB19, and optimization of culture conditions, for an in vitro blood–brain barrier model for 456 drug permeability studies. Fluids and Barriers of the CNS. 2013;10(1):33. 457 41. Tettelin H, Masignani V, Cieslewicz MJ, Donati C, Medini D, Ward NL, et al. Genome 458 analysis of multiple pathogenic isolates of Streptococcus agalactiae: implications for the 459 microbial "pan-genome". Proc Natl Acad Sci U S A. 2005;102(39):13950–5. 460 42. Wilson CB, Weaver WM. Comparative susceptibility of group B streptococci and 461 Staphylococcus aureus to killing by oxygen metabolites. J Infect Dis. 1985;152(2):323–9. 462 43. Framson PE, Nittayajarn A, Merry J, Youngman P, Rubens CE. New genetic techniques 463 for group B streptococci: high-efficiency transformation, maintenance of temperature-sensitive 464 pWV01 plasmids, and mutagenesis with Tn917. Applied and Environmental Microbiology. 465 1997;63(9):3539–47. 466 44. Holo H, Nes Ingolf F. High-Frequency Transformation, by Electroporation, of 467 Lactococcus lactis subsp. cremoris Grown with Glycine in Osmotically Stabilized Media. Applied 468 and Environmental Microbiology. 1989;55(12):3119–23. 469 45. Hooven Thomas A, Bonakdar M, Chamby Anna B, Ratner Adam J. A Counterselectable 470 Sucrose Sensitivity Marker Permits Efficient and Flexible Mutagenesis in Streptococcus 471 agalactiae. Applied and Environmental Microbiology. 2019;85(7):e03009–18. 472 46. Espinal ER, Matthews T, Holder BM, Bee OB, Humber GM, Brook CE, et al. Group B 473 Streptococcus-Induced Macropinocytosis Contributes to Bacterial Invasion of Brain Endothelial 474 Cells. Pathogens. 2022;11(4). 475 47. Vollmuth N, Bridgers BE, Armstrong ML, Wood JF, Gildea AR, Espinal ER, et al. Group 476 B Streptococcus transcriptome when interacting with brain endothelial cells. J Bacteriol. 477 2024;206(6):e0008724. 478 48. Livak KJ, Schmittgen TD. Analysis of Relative Gene Expression Data Using Real-Time 479 Quantitative PCR and the 2−ΔΔCT Method. Methods. 2001;25(4):402–8. 480 481 482 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2025. ; https://doi.org/10.1101/2025.08.27.672580doi: bioRxiv preprint Figure Legends 483 484 485 Figure 1, Cas12a-based Point Mutation for CRISPRi. (A) Graphical representation of GBS point mutation process, 486 showing anhydrotetracycline induction of Cas12a and sgRNA expression on the pGBSedit (pGBSedit) plasmid. Also 487 pictured is the inclusion of a repair cassette, containing the homology needed for repair and carrying the desired 488 mutation. (B) Depiction of the genotypic changes needed to generate a point mutation, showing both a hypothetical 489 point mutation (D360A) as well as the silent mutations in the sgRNA target site necessary to protect the plasmid and 490 mutant strain from Cas12a cutting. These silent mutations must be designed on the plasmid repair cassette for the 491 mutation process to be successful. 492 493 Figure 2, Verification of Phenotype Changes. (A) Graphical depiction of the knockdown generation process. 494 sgRNAs are cloned into the shuttle vector p3015b, then electroporated into Streptococcus agalactiae str. COH1 495 dCas9. These transformants should now have a knockdown phenotype and can be used for subsequent in vitro 496 experimentation. (B) Hemolysis assay data, with Mock being PBS-treated sheep’s blood, Scramble being a control 497 infection lacking a targeted sgRNA, as well as two variants of knockdowns of covR and the cyl operon. Beneath each 498 bar graph is a visual representation of the hemolytic phenotype of each transformant using bacterial growth on 499 sheep’s blood agar, with the covR knockdowns showing more hemolysis relative to the control and the cyl 500 knockdowns showing less hemolysis. One-way ANOVA was performed, **P < 0.01 and ***P < 0.001. All experiments 501 were performed in biological and technical triplicate, (n=9). 502 503 Figure 3, Transcriptional Changes. (A-F) qPCR analysis of target genes following sgRNA transformation into 504 Streptococcus agalactiae str. COH1 dCas9 for conformation of CRISPR interference expression modulation. sgRNAs 505 shown include the established virulence genes pilA (A), srr2 (B), iagA (C), covR (D), and cyl (E), as well as the 506 essential gene ccpA (F). (A-F). One-way ANOVA was performed, *P < 0.05, **P < 0.01, and ***P < 0.001. All 507 experiments were performed in biological and technical triplicate, (n=9). 508 509 Figure 4, In vitro Infection Application. GBS adhesion (A) and invasion (B) rates to hCMEC/D3 monolayers when 510 various well-established virulence genes are knocked down. (C-E) hCMEC/D3 inflammatory marker expression levels 511 when facing challenge with either mock infection, Scramble control, or pilA knockdown as a representative virulence 512 gene. Left to right, the qPCR targets are (C) CXCL8 (IL-8), (D) CXCL1, and (E) CXCL2. (A-E) One-way ANOVA was 513 performed, *P < 0.05, **P < 0.01, and ***P < 0.001. All experiments were performed in biological and technical 514 triplicate, (n=9). 515 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2025. ; https://doi.org/10.1101/2025.08.27.672580doi: bioRxiv preprint Figure 1 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2025. ; https://doi.org/10.1101/2025.08.27.672580doi: bioRxiv preprint A B B Mock Scrmbl covR 227 k/d covR 313 k/d cyl 283 k/d cyl 384 k/d D Scrmbl ccpA k/d C Scrmbl ccpA k/d Figure 2 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2025. ; https://doi.org/10.1101/2025.08.27.672580doi: bioRxiv preprint FF Scrmbl ccpA k/d FDD Scrmbl covR 227 k/d covR 313 k/d D E Scrmbl cyl 283 k/d cyl 384 k/d F E A Scrmbl pilA k/d A B Scrmbl srr2 k/d B B C Scrmbl iagA k/d C Figure 3 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2025. ; https://doi.org/10.1101/2025.08.27.672580doi: bioRxiv preprint C C Mock Scrmbl pilA k/d D D D Mock Scrmbl pilA k/d E E E Mock Scrmbl pilA k/d A A Scrmbl pilA k/d srr2 k/d iagA k/d B Scrmbl pilA k/d srr2 k/d iagA k/d Figure 4 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted August 27, 2025. ; https://doi.org/10.1101/2025.08.27.672580doi: bioRxiv preprint