Exploring and engineering PAM-diverse Streptococci Cas9 for PAM-directed bifunctional and titratable gene control in bacteria

Abstract

The RNA-guided Cas9s serve as powerful tools for programmable gene editing and regulation; their targeting scopes and efficacies, however, are always constrained by the PAM sequence stringency. Most Streptococci Cas9s, including the prototype SpCas9 from S. pyogenes, specifically recognize a canonical NGG PAM via a conserved RxR PAM-binding motif within the PAM-interaction (PI) domain. Here, SpCas9-based mining unveils three distinct and rarely presented PAM-binding motifs (QxxxR, QxQ and RxQ) among Streptococci Cas9 orthologs. With the catalytically-dead QxxxR-containing SedCas9 from S. equinus, we dissect its NAG PAM specificity and elucidate its underlying recognition mechanism via computational prediction and mutagenesis analysis. Replacing the SedCas9 PI domain with alternate PAM-binding motifs rewires its PAM specificity to NGG or NAA. Moreover, a semi-rational design with minimal mutation creates a SedCas9-NQ variant showing robust activity towards expanded NNG and NAA PAMs, based upon which we engineered a compact ω-SedCas9-NQ transcriptional regulator for PAM-directed bifunctional and titratable gene control. The ω-SedCas9-NQ mediated metabolic reprogramming of endogenous genes in Escherichia coli affords a 2.6-fold increase of 4-hydroxycoumarin production. This work reveals new Cas9 scaffolds with distinct PAM-binding motifs for PAM relaxation and creates a new PAM-diverse Cas9 variant for versatile gene control in bacteria.

Keywords: 4-hydroxycoumarin; CRISPR interference and activation; Cas9 engineering; Gene control; Metabolic engineering.

Fig. 1.. Characterization of the QxxxR PAM-binding motif containing SeCas9.

a. Sequence alignment of different PAM-binding motifs among representative SpCas9-like orthologs. Previously identified Cas9s are underlined. b. The cas locus and protein domain structure of SpCas9 from S. pyogenes M1 GAS and SeCas9 from S. equinus ATCC 9812. c. Pair-wise sequence comparison of crRNA and tracrRNA between Sp-sgRNA and Se-sgRNA. d. The dual-plasmid eGFP repression assay system: pZE12 with randomized NNN PAM library and sgRNA, and pCS27 containing catalytically inactive SedCas9. The NNN was located after the ATG start codon of eGFP. e. PAM profile determination of SedCas9 with eGFP repression assay. f. The influence of the 4th nucleotide of the PAM on SedCas9 activity. g. SeCas9-mediated in vivo genome cleavage at NNG PAM sites of aslB locus from E. coli BW25113 (F′) and eGFP locus from E. coli BW25113 (F′) with chromosomally integrated eGFP (E. coli::eGFP). Data indicated the mean ± standard deviation (n = 3 independent biological replicates).

Fig. 2.. Dissecting the PAM recognition mechanism via computational modeling and mutagenesis analysis.

a. Structural superimposition of SpCas9 (PDB ID: 4UN3, grey) and SeCas9 (pink) predicted by AlphaFold 2.0. Colored components include the targeted dsDNA (blue), sgRNA (purple), PAM sequence (yellow), and the QSNLR loop from SeCas9 (green). b. Molecular dynamics simulations of interactions between SeCas9 PI domain with TAG and TGG PAMs. Within the zoomed-in view of the left SeCas9-TAG and SeCas9-TGG panels, potential hydrogen bonds were denoted in dashed lines. The two right panels showed the simulations results of predicted hydrogen-bonding distances between the Q1340/R1344 and 2nd/3rd bases of TAG or TGG PAMs. c. The eGFP repression assay of SedCas9 PI domain variants targeting TAG and TGG PAMs. The dashed boxes indicate the mutation of S1341/N1342/L1343 to leucine (L) or lysine (K). d. Replacement of the PI domain of SedCas9 (grey) with that of SpCas9 (SpPI, orange), putative SmCas9 (VEF19407.1) from S. mutans NCTC10832 (SmPI, green), and putative SeCas9-HC5 (KEY47635.1) from S. equinus HC5 (HC5PI, blue). The chimeric mutants were created by fusing the N-terminal (1–1108 aa) of SedCas9 with the SpPI (1100–1368 aa), SmPI (1090–1350 aa), and HC5PI (1109–1375 aa), respectively. The eGFP repression assays were conducted with SedCas9 chimeric variants towards NNG and NAA PAMs. Data indicated the mean ± standard deviation (n = 3 independent biological replicates).

Fig. 3.. Semi-rational generation of SedCas9 mutants with expanded PAM scope.

a. PAM interacting (Q1340 and R1344) or proximal (L1120, D1147, S1148, T1230, E1231 and K1346) residues in SeCas9 identified via AlphaFold modeling and SpCas9-based alignment. b. Comparison of SedCas9 variants towards NNG PAMs via the eGFP repression assay. c. The schematic of the eGFP repression assay of SpdCas9 variants and SedCas9-NQ towards single-R PAMs. d. Comparison of eGFP repression activity between previously engineered PAM-flexible SpdCas9 variants (dxCas9-3.7 and SpdRY) towards NGN PAMs and SedCas9-NQ towards NNG PAMs. e. Comparison of eGFP repression activity between SpdRY towards NAN PAMs and SedCas9-NQ towards NNA PAMs. The data towards all PAMs tested were generated from three biological replicates (n = 3). In boxplots of d and e, the dash and ‘+’ within each box respectively represent medium and mean values. * P≤ 0.05, ** P≤ 0.01, ***P≤ 0.001, ****P≤ 0.0001 (two-tailed t-test; n = 12 independent biological replicates)

Fig. 4.. The fusion transcription regulator ω-SedCas9-NQ mediated bifunctional and titratable gene control.

a. Schematic of the ω-SedCas9-NQ mediated promoter walking for CRISPR activation (CRISPRa) and targeting at coding sequence for CRISPR interference (CRISPRi). For CRISPRa, ω-SedCas9-NQ binds to the upstream of transcription start site (TSS, +1 bp) every 8–12 bp on both non-template (N) and template (T) strands and recruits RNA polymerase (RNAP subunits β′βα₂). The sgRNAs are designed to target available NNG or NAA PAMs on Plpp0.03 and PchbB promoters. For CRISPRi, ω-SedCas9-NQ binds to representative NNG or NAA PAMs and blocks RNAP for transcriptional elongation. b. CRISPRa of eGFP under control of an engineered low-strength promoter Plpp0.03. NC, negative control. c. CRISPRa of eGFP under control of the PchbB promoter from the cryptic cellulobiose utilization locus chb. NC, negative control. d. ω-SedCas9-NQ mediated titratable CRISPRi of eGFP at NNG and NAA PAMs. * P≤ 0.05, ** P≤ 0.01, ***P≤ 0.001, ****P≤ 0.0001 (two-tailed t-test; n = 3 independent biological replicates). Data indicated the mean ± standard deviation (n = 3 independent biological replicates).

Fig. 5.. The ω-SedCas9-NQ mediated metabolic control for 4-hydroxycoumarin production enhancement in E. coli.

a. Metabolic pathway of 4-hydroxycoumarin (4HC) production from glycerol in E. coli. The 4HC pathway were assembled on two plasmids, pZE12-EP-APTA and pCS27-PS. sgRNAs were constructed on the pCS27-PS plasmid. Genes to be repressed are shown in red, including ptsI (phosphoenolpyruvate-protein phosphotransferase), eno (enolase), ppc (phosphoenolpyruvate carboxylase), pykA (pyruvate kinase II), pykF (pyruvate kinase I), gltA (citrate synthase), fabD (malonyl-CoA-acyl carrier protein transacylase), fabF (β-ketoacyl-(acyl carrier protein) synthases II), and csrA (carbon storage regulator). Genes to be activated are shown in green, including accB (biotin carboxyl carrier protein) and fadR (fatty acid degradation regulator). Genes to be over-expressed on plasmids are shown in blue, including ppsA (phosphoenolpyruvate synthetase), tktA (transketolase I), aroG^fbr (feedback-inhibition-resistant 3-deoxy-7-phosphoheptulonate synthase), aroL (shikimate kinase II), entC (isochorismate synthase), pchB (isochorismate pyruvate lyase), sdgA (salicoyl-CoA ligase) and pqsD (β-ketoacyl-ACP synthase III (FabH)-type quinolone synthase). Metabolite abbreviations: DHA, dihydroxyacetone; DHAP, dihydroxyacetone phosphate; G3P, glycerol 3-phosphate; F6P, fructose-6-phosphate; E4P, erythrose-4-phosphate; PEP, phosphoenolpyruvate; PYR, pyruvate; AcCoA, acetyl-CoA; α-KG, α-ketoglutarate; Mal, malate; OAA, oxalacetate; DAHP, 3-deoxy-arabino-heptulosonate 7-phosphate; Cho, chorismate; ICho, isochorismate. b. Schematic of the ω-SedCas9-NQ mediated CRISPRi and CRISPRa of genes of interest (GOI). sgRNAs for repression targets were targeting five neighboring NNG and NAA PAMs. sgRNAs for activation targets were targeting NNG and NAA PAMs every 5–10 bp upstream of transcription start site (TSS, +1 bp). c. The increase in folds of 4HC titers with first-round screening of repression targets at different PAMs in test tube cultures with 3 ml M9Y medium containing 10 g/l glycerol. d. The increase in folds of 4HC titers with first-round screening of activation targets at different positions on the promoters in test tube cultures with 3 ml M9Y medium containing 10 g/l glycerol. e. The second-round shake flask experiments for 4HC production from first-round CRISPRi and CRISPRa targets in shake flask cultures with 20 ml M9Y medium containing 10 g/l glycerol. The bars denoted 4HC titers and the circle dots denoted OD600. All production tests were performed with the engineered strain E. coli::ω-SedCas9-NQ. Negative control (NC), E. coli::ω-SedCas9-NQ with the 4HC pathway plasmids pZE-EP-APTA and pCS27-PS. * P≤ 0.05, ** P≤ 0.01, ***P≤ 0.001, ****P≤ 0.0001 (two-tailed t-test; n = 3 independent biological replicates). Data indicated the mean ± standard deviation (n = 3 independent biological replicates).