Glycine Cleavage System and cAMP Receptor Protein Co-Regulate CRISPR/cas3 Expression to Resist Bacteriophage

The CRISPR/Cas system protects bacteria against bacteriophage and plasmids through a sophisticated mechanism where cas operon plays a crucial role consisting of cse1 and cas3. However, comprehensive studies on the regulation of cas3 operon of the Type I-E CRISPR/Cas system are scarce. Herein, we investigated the regulation of cas3 in Escherichia coli. The mutation in gcvP or crp reduced the CRISPR/Cas system interference ability and increased bacterial susceptibility to phage, when the casA operon of the CRISPR/Cas system was activated. The silence of the glycine cleavage system (GCS) encoded by gcvTHP operon reduced cas3 expression. Adding N5, N10-methylene tetrahydrofolate (N5, N10-mTHF), which is the product of GCS-catalyzed glycine, was able to activate cas3 expression. In addition, a cAMP receptor protein (CRP) encoded by crp activated cas3 expression via binding to the cas3 promoter in response to cAMP concentration. Since N5, N10-mTHF provides one-carbon unit for purine, we assumed GCS regulates cas3 through associating with CRP. It was evident that the mutation of gcvP failed to further reduce the cas3 expression with the crp deletion. These results illustrated a novel regulatory pathway which GCS and CRP co-regulate cas3 of the CRISPR/Cas system and contribute to the defence against invasive genetic elements, where CRP is indispensable for GCS regulation of cas3 expression.


Introduction
Prokaryotic viruses occur ubiquitously and pose a serious threat to bacteria and archaea [1]. To resist these extremely large numbers of predators, bacteria have developed multiple resistant mechanisms [2][3][4]. As one of the important defence mechanisms, arrays of clustered regularly interspaced short palindromic repeats (CRISPR) and CRISPR-associated (Cas) proteins constitute the CRISPR/Cas system which has attracted much attention in terms of its function and application in recent years [5,6]. CRISPR arrays are composed of the spacer sequences acquired from foreign DNA between 26 and 72 bp and highly conserved repeat sequences flanked by the spacers [7,8]. A group of cas genes encoding Cas proteins is generally located near the CRISPR loci, and it displays immune function associated with CRISPR arrays to battle with foreign DNA invasion [9]. The entire immunization Figure 1. Schematic of the Esherichia coli (E. coli) Type I-E CRISPR/Cas system. The system consists of cas3 and cse1 operon. The cse1 operon contains 7 genes (blue) consisting of cse1, cse2, ces4, cas5e, cse3, cas1, and cas2, with transcription being initiated from the cse1 promoter. The H-NS represses and LeuO activates cse1 operon, respectively. The glycine cleavage system (GCS), related to many characters of bacteria, catalyzes the glycine to obtain N 5 , N 10 -methylene tetrahydrofolate (N 5 , N 10 -mTHF), which is a one carbon donor for the production of serine, thymidine, and purines. GCS consists of three enzymes and a carrier: gcvP (glycine decarboxylase), gcvH (lipoic acid-containing carrier), gcvT (tetrahydrofolate dependent aminomethyltransferase), and gcvL (dihydrolipoamide dehydrogenase) [18]. The CRP is a global regulator that has multiple regulatory effects on bacteria. It performs regulatory functions by forming a CRP-cAMP complex with cAMP and binding to the promoter region of the gene [19]. Our previous work has verified that overexpression or deletion of cas3 significantly affects bacterial susceptibility to phage infection. Therefore, we used the transposon mutation and DNA pull-down technology to screen the proteins that regulate cas3 in E. coli, and elucidated the mechanisms by which CRISPR/cas3 is regulated. Our study suggested that GCS affected the bacterial susceptibility to phage by altering cas3 expression, and CRP was dispensable for the GCS to regulate cas3 expression.

Strains, Plasmids, and Growth Conditions
The strains, plasmids, and oligonucleotides used in this study are shown in Tables S1-S3. The E. coli K-12 strain MG1655 and its derivatives were cultured at 37 • C in Luria-Bertani (LB) or minimal media containing 48 mM Na 2 HPO 4 , 22 mM KH 2 PO 4 , 9 mM NaCl, 19 mM NH 4 Cl, 2 mM MgSO 4 , 100 µM CaCl 2 , and 0.5% (w/v) glucose in a shaker at 200 rpm. When necessary, different amounts of nalidixic acid, chloramphenicol (Cm), ampicillin (Amp), glycine, serine, N 5 , N 10 -mTHF, glucose, or cAMP were added into the media. Bacterial concentration was measured in a SHIMADZU UV-1800 Spectrophotometer at OD 600 . At least three replicates were performed for all experiments.

β-Galactosidase Assays
The reporter plasmids containing cas3, gcv, cse1, or crp promoter were transferred into MG1655 and its mutants. The method for determining β-galactosidase (β-gal) activity is described previously [20]. The cultures were taken when the OD 600 was approximately 1.0. A modified procedure of β-gal assay was used in a transposon mutagenesis experiment to determine the β-gal activity of the reporter strain and its mutants. Briefly, 20 µL of each cultured bacterium was pipetted into 96-well plates and mixed with 80 µL of permeabilization solution (100 mM Na 2 HPO 4 , 20 mM KCl, 2 mM MgSO 4 , 0.8 mg/mL hexadecyltrimethylammonium bromide, 0.4 mg/mL sodium deoxycholate, 5.4 µL/mL beta-mercaptoethanol). These samples were incubated at 30 • C for 30 min. Subsequently, 140 µL of substrate solution (60 mM Na 2 HPO 4 , 40 mM NaH 2 PO 4 , 1 mg/mL o-nitrophenyl-β-D-galactoside, 2.7 µL/mL β-mercaptoethanol) was added into each well. After sufficient color had developed, 160 µL of stop solution (1 M Na 2 CO 3 ) was added, and duration of reaction time was noted. The OD 420 of each sample was recorded using a Biotek ELx800 Microplate Reader. The β-gal activity was calculated by the method described by Miller [20].

Construction and Identification of Transposon Mutants
S17-1 λ pir (pUTmini-Tn5) and reporter strain (E. coli MG1655∆lacZ∆cas3::lacZ) were cultured to mid-log phase as the donor and recipient, respectively (see Figure S1). The mutants were selected, and the promoter activity of cas3 (β-gal activity) was measured by the modified procedure of β-gal assay as described previously. The transposon insertion sites in each mutant were identified by genome walking (Genome Walking Kit, Takara, Kusatsu, Japan).

DNA Pull-Down Assays
The biotin labelled PCR primers for amplifying promoter regions of cas3 gene were commercially synthesized (RuiMian, Shanghai, China) (see Figure S1 in Supplemental Material). DNA pull-down assays were performed as described previously [21]. Briefly, the positive biotin labelled sequence (−172 to 0 of cas3) was amplified from MG1655 genomic DNA. The DNA sequence was affixed to beads and then incubated with a supernatant of bacterial lysate. Beads were subsequently washed with buffer containing non-specific DNA, and 50 mM NaCl was used to remove non-adhering and low-specificity DNA-binding proteins. Then 100, 300, and 1000 mM NaCl were used to elute specific DNA-binding proteins. The eluted proteins were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), followed by Brilliant Coomassie staining, and were then analyzed by mass spectrometry.

Western Blotting
Bacteria were cultured to mid-log phase, and 1 mL of bacteria was collected. Polyclonal anti-Cas3 and monoclonal anti-GroEL (Abcam, Cambridge, US) were used for subsequent immunodetection as primary antibodies (see Text. S1 in Supplemental Material). The samples were detected in Chemidoc equipment (BioRad, Hercules, US), and the relative greyscales of Cas3 in different strains were analyzed by software ImageJ (1.4.3.67).

Electrophoretic Mobility Shift Assays
To determine the binding specificity of CRP, the positive probe of the cas3 gene from MG1655 genomic DNA and the negative probe were cloned into pMD-18T (Takara, Kusatsu, Japan), respectively. The negative probe was from the mutated DNA fragment which the predicted CRP-binding site TATGAGCAGCATCGAA was altered to CTGTGGCAGCAGACTA. CRP was purified according to standard protein purification procedure with slight modification as previously reported [22]. The FAM labelled PCR primers for amplifying promoter regions of the cas3 gene were commercially synthesized (RuiMian, Shanghai, China). The reaction mixture containing binding buffer (TOLO Biotech, Hefei, China), 10 nM probe, 3 mM cAMP, and different amounts of CRP (0, 100, 200, and 400 nM) was incubated at 26 • C for 30 min. The samples were separated by 2.0% TBE agarose gel. The probes were detected by the ImageQuant LAS 4000 mini (GE Healthcare, Marlborough, US).

DNase I Footprinting Assay
DNase I footprinting assays were performed to identify the CRP-binding sequence described by Wang et al. [23]. Briefly, each probe was prepared with the same method as described in EMSA. Then, the probe (300 ng) was incubated with different amounts of recombinant CRP protein in a total volume of 40 µL. Furthermore, digestion, electrophoresis, and data analysis were carried out with the same procedure described by Wang et al. [23].

Lytic Infection Efficiency Assays
The susceptibility of MG1655 and its mutants to phage was confirmed by PFU assay with certain modifications [24,25]. In brief, strains either containing anti-phage plasmids pGEX3 (pGEX with phage vB_EcoS_SH2 spacer) or not were cultured in LB to mid-log phase. Bacteria were centrifuged at 6000× g for 5 min and then resuspended in 10 mM MgSO 4 . For lytic infection, each group of bacteria of approximately 2 × 10 8 CFU was mixed with 2 × 10 7 PFU phage, and incubated at 37 • C for 2 h. The mixture was pelleted at 12,000× g for 2 min. The supernatants were filtered followed by 10-time serially dilution to evaluate the phage titres on indicator bacterium MC1061.

Plasmid Transformation Assays
The competent state of MG1655 and its mutants at mid-log phase was prepared for chemical transformation. Competent cells were mixed with 500 ng plasmids pGEX, pGEX1 (pGEX with CRISPR1 loci of MG1655 and PAM), or pGEX2 (pGEX with CRISPR2 loci of MG1655 and PAM), respectively, cooled on ice for 30 min, and then heat-shocked followed by recovery in 1 mL LB for 3 h. After serial dilution, the mixture was plated on LBA (100 µg mL −1 Amp) and incubated at 37 • C for 16 h. The colonies were counted to determine the plasmid's transformation efficiency. The stability of the plasmid was determined by the ratio of the number of the pGEX1 or pGEX2 transformed colonies to the pGEX transformed colonies.

Mutation of gcvP or gcvT Decreases cas3 Expression to Affect Phage Infection
To screen the regulators of the Type I-E CRISPR/Cas system cas3, an E. coli reporter strain of which the cas3 open reading frame was replaced by lacZ was used to construct a transposon mutant library (see Figure S2 in Supplemental Material). The cas3 promoter activities of approximately 3000 mutants were measured by modified β-gal assay. The mutants with relatively high or low changes in galactosidase activity were selected as candidates, and some of which were sequenced including kdpA, waaR, atpD, gcvP, and so on. A gcvP gene, which encodes a pyridoxal phosphate-containing glycine decarboxylase belonging to the GCS, was identified by genome walking since insertion of the transposon led to decrease in cas3 promoter activity (Figure 2A). The GCS was encoded by the gcvTHP and gcvL operon [26]. To confirm the impact of gcvP and verify whether the reduced cas3 activity was due to the disruption of the GCS, gcvP, or gcvT deletion, mutants and their complemented strains were generated. To confirm the cas3 promotor activity, a reporter plasmid (pRCL1)-containing cas3 promoter was then transformed into each mutant. The results showed that the cas3 promoter activity in both ∆gcvP (EC20) and ∆gcvT (EC40) mutants reduced to half as high as in wild type (WT) (EC1001) and restored in complemented strains C∆gcvP (EC30) and C∆gcvT (EC50) ( Figure 2B). The expression level of cas3 was further determined by Western blotting, which is similar to the reporter plasmid results ( Figure 2C). These data indicated that cas3 expression was affected by the GCS. The cas3 promoter activity significantly reduced in ∆gcvP and ∆gcvT. (C) The expression level of cas3 by Western blotting. The value was calculated from the relative grayscale value of Cas3 to GroEL, and the value of WT was normalized as one. The cas3 expression significantly reduced in ∆gcvP and ∆gcvT. (D) In the control group, the plasmids without the anti-vB_EcoS_SH2 spacer (pGEX) were transferred into WT (EC15), ∆gcvP (EC205), ∆cas3 (EC135), ∆hns (EC95), ∆hns∆gcvP (EC105), and ∆hns∆cas3 (EC125). In the experimental group, the plasmids containing the anti-vB_EcoS_SH2 spacer (pGEX3) were transferred into WT (EC16), ∆gcvP (EC206), ∆cas3 (EC136), ∆hns (EC96), ∆hns∆gcvP (EC106), and ∆hns∆cas3 (EC126). The phage titer of ∆hns∆gcvP was approximately twice as high as that of ∆hns. (E) The LeuO overexpression plasmids were transferred into WT and ∆gcvP. In the control group, the plasmids without the anti-vB_EcoS_SH2 spacer (pGEX) were transferred into WT (EC141) and ∆gcvP (EC142). In the experimental group, the plasmids containing the anti-vB_EcoS_SH2 spacer (pGEX3) were transferred into WT (EC144) and ∆gcvP (EC145). The phage titre of pleuO∆gcvP was approximately twice as high as that of WT (pleuO). Phage titres were measured by lytic infection efficiency assay, as previously described. All the data were mean ± SEM of at least three replicates, and the p value (* p < 0.05) was analyzed by the t-test.
The integrated or engineered CRISPR spacers matching invasive DNA is indispensable for the CRISPR/Cas system to exert its immune function [12,27], and the entire cse1-cse2-cse4-cas5e-cse3 operon is repressed when E.coli is under normal biological conditions that led to the CRISPR/Cas remaining static. To activate the CRISPR/Cas, a ∆hns mutant was constructed. To determine the effect of GCS on bacterial susceptibility to phage, a plasmid containing the spacer was constructed to match the DNA of phage vB_EcoS_SH2, and was transformed into WT and its ∆gcvP, ∆cas3, ∆hns, ∆hns∆gcvP, ∆hns∆cas3 mutants followed by phage infection (see Text S1 and Figure S3 in Supplemental Material). The control group was transformed with an empty vector.
The results showed that the mutation of cas3 resulted in a similar susceptibility to phage with the control group. Compared with WT, the phage titer of ∆hns was significantly decreased. In contrast, the phage titer of ∆hns∆gcvP was approximately twice as high as that of ∆hns ( Figure 2D). In the control groups, there was no difference among the mutants, which suggested the mutation of these genes had no effect on the other phage-defending mechanisms. These results demonstrated that GCS affected susceptibility to phage by altering cas3 expression, and Cas3 was crucial for the CRISPR/Cas system to resist phage infection.
To further confirm the impact of gcvP on cas3 when cse1 was activated, the activator of cse1 encoded by leuO was overexpressed by transforming the pleuO to WT, ∆gcvP, and ∆crp. The results showed that the phage titre of WT (pleuO) was significantly decreased compared with WT. In contrast, the phage titre of pleuO∆gcvP was approximately twice as high as that of WT (pleuO) ( Figure 2E).

The cas3 and gcvTHP Promoter Activities are Promoted by Glycine
Since GCS functions via catalyzing glycine, cas3 expression of WT and mutant ∆gcvP cultured in LB media supplemented with glycine was determined to further investigate whether the down-regulation of cas3 expression was due to the defect of GCS function. In LB media, the addition of 20, 40, 80, and 100 mM glycine induced cas3 promoter expression of WT, however, the added glycine had no effect on the ∆gcvP groups ( Figure 3A). To minimize the effect of components of media, the cells were cultured in minimal media. The results showed that the cas3 expression was activated by the minimal of 1 mM glycine in WT but it failed to be activated by glycine in ∆gcvP ( Figure 3B), which indicated the knock-out of gcvP led to the silence of GCS that the CCS was unable to regulate cas3 expression through glycine usage.
It was reported previously that the expression of gcvTHP operon can be induced by glycine in E. coli as well [28]. Therefore, to further investigate the correlation between gcvTHP expression and cas3 expression, the gcvTHP promoter was cloned into a reporter plasmid (pRCL2) and transformed into WT. The supplemented glycine in LB media had no effect on the expression of the gcvTHP promoter, but the activity of the gcvTHP promoter exhibited a higher level in LB than in minimal media without any added glycine ( Figure 3C,D). In the minimal media, gcvTHP promoter activity started to increase with 1 mM glycine ( Figure 3D). These data demonstrated the supplemented glycine was able to promote cas3 and gcvTHP promoter activity in minimal media, and the amount of added glycine to enhance cas3 promoter activity was consistent with the amount needed to promote gcvTHP promoter activity in minimal media ( Figure 3B,D).

N 5 , N 10 -Methylene Tetrahydrofolate Promotes cas3 Promoter Activity
In one-carbon unit metabolic pathways, both serine and glycine can be catalyzed into N 5 , N 10 -methylene tetrahydrofolate (N 5 , N 10 -mTHF), the same intermediate metabolite derived from serine hydroxymethyltransferase-catalyzed serine or GCS-catalyzed glycine [29,30]. To clarify whether GCS affected cas3 expression by N 5 , N 10 -mTHF, cas3 promoter activity of ∆gcvP cultured in LB media supplemented with serine or minimal media with N 5 , N 10 -mTHF (Toronto Research Chemicals) was detected using pRCL1, since glycine activated the cas3 expression. For mutant ∆gcvP, the cas3 promoter can be activated by 40 mM serine in LB media ( Figure 4A). As expected, 0.1 mM N 5 , N 10 -mTHF promoted cas3 promoter activity in both WT and ∆gcvP ( Figure 4B). Our results demonstrated that GCS indirectly regulated cas3 promoter activity, and the N 5 , N 10 -mTHF derived from GCS-catalyzed glycine was able to promote cas3 promoter activity. The β-gal activity of the cas3 promoter was measured by reporter plasmids in WT or ∆gcvP cultured in minimal media supplemented with 0, 0.5, 1, 5, 10, or 20 mM glycine. The cas3 promoter was activated by the minimal of 1 mM glycine in WT. (C) The β-gal activity of the gcvTHP promoter was measured by reporter plasmids in WT (EC12) cultured in LB media supplemented with 0, 20, 40, 80, or 160 mM glycine. The supplemented glycine in LB media had no effect on the expression of the gcvTHP promoter. (D) The β-gal activity of the gcvTHP promoter was measured by reporter plasmids in WT cultured in minimal media supplemented with 0, 0.5, 1, 5, 10, or 20 mM glycine. The gcvTHP promoter activity started to increase with 1 mM glycine. All the data were mean ± SEM of at least three replicates, and the p value (* p < 0.05) was analyzed by the t-test. The β-gal activity of the cas3 promoter was measured by reporter plasmids in WT or ∆gcvP cultured in minimal media supplemented with 0, 0.01, or 0.1 mM N 5 , N 10 -mTHF. 0.1 mM N 5 , N 10 -mTHF promoted cas3 promoter activity in both WT and ∆gcvP. All the data were the mean ± SEM of at least three replicates, and the p value (* p < 0.05) was analyzed by the t-test.

CRP Positively Regulates the Transcription of cas3 Gene to Affect Phage Infection
To further explore the regulators which directly bind to the cas3 promoter, DNA pull-down assays were employed. The protein bands presented only in experimental group but not in control group were analyzed by mass spectrometry (MS) ( Figure 5A). In the upper protein band, eight potential regulators were identified, and in the lower protein band, 63 potential regulators were identified. Among these proteins, the regulatory effect of TnaA, kdgR, AtpD, and CRP on cas3 were verified. CRP encoded by crp was selected among the proteins identified by MS, because cAMP is the essential molecule for CRP function, and it may be associated with N 5 , N 10 -mTHF. To investigate whether CRP affects cas3 expression, a ∆crp mutant was constructed and pRCL1 was transformed into the WT and ∆crp (EC60). Then, cas3 promoter activity in WT and ∆crp cultured in LB supplemented with glucose was analyzed, since it was reported that glucose can reduce the concentration of intracellular cAMP, which is indispensable for CRP-cAMP complex binding to DNA fragments [31]. The cas3 promoter expression level in WT was almost 4 times higher than that in ∆crp. In WT, the presence of glucose led to an approximately 4 times decrease in cas3 promoter activity. As expected, cas3 promoter activity was not affected by glucose in ∆crp, which indicated cAMP is required by CRP to regulate cas3 promoter activity ( Figure 5B). In WT, the presence of glucose led to an approximately 4 times decrease in cas3 promoter activity. (C) In the control group, the plasmids without the anti-vB_EcoS_SH2 spacer (pGEX) were transferred into WT (EC15), ∆crp (EC605), ∆hns (EC95), and ∆hns∆crp (EC115). In the experimental group, the plasmids containing the anti-vB_EcoS_SH2 spacer (pGEX3) were transferred into WT (EC16), ∆crp (EC606), ∆hns (EC96), and ∆hns∆crp (EC116). The phage titre of ∆hns∆crp was approximately 3 times of ∆hns. (D) The plasmids overexpressing LeuO were transferred into ∆crp. In the control group, the plasmids without the anti-vB_EcoS_SH2 spacer (pGEX) were transferred into ∆crp (EC143). In the experimental group, the plasmids containing the anti-vB_EcoS_SH2 spacer (pGEX3) were transferred into ∆crp (EC146). The phage titre of pleuO∆crp was almost 3 times of WT (pleuO). Phage titres were measured by lytic infection efficiency assay, as described in Materials and Methods. All the data were mean ± SEM of at least three replicates, and the p value (* p < 0.05) was analyzed by the t-test.
To determine the effects of CRP on bacterial susceptibility to phage, the lytic infection efficiency assay was performed. The phage titre of ∆hns∆crp was approximately 3 times of ∆hns, and the phage titre of pleuO∆crp was almost 3 times of WT (pleuO). In addition, there was no difference between WT and ∆crp in the control group, which suggested the mutation of crp had no effect on the other phage-defending mechanisms ( Figure 5C,D). These results demonstrated that CRP affected bacterial susceptibility to phage by altering cas3 promoter activity.

CRP Activates cas3 Expression by Binding to Transcriptional Initiation Area
To elucidate the regulatory mechanism of CRP protein on cas3 expression, electrophoretic mobility shift assays (EMSA) were performed. The sequence of the cas3 (E. coli MG1655 strain) promoter was analyzed. Compared with the previously reported core motif of the CRP-binding site (TGTGAN 6 TCACA) of E. coli, the transcriptional initiation area of cas3 in MG1655 harbors a predicted core binding site (TATGAN 6 TCGAA) [32]. Between those, three varied nucleotides were observed. The putative binding site is located at −51 of the cas3 open reading frame close to the predicted cas3 promoter ( Figure 6A). According to the sequence of putative binding site, a probe amplified from the promoter of cas3 and its mutant was applied in EMSA. In the original probe group, at least 400 nM CRP protein was required for the entire shift of the 10 nM probe ( Figure 6B). The predicted core binding site mutant probe lost its ability to bind to the CRP-cAMP complex, such that even the 400 nM CRP protein was incapable of shifting the mutant probe ( Figure 6C).
To further verify this result and clarify the binding motif of CRP, the DNase I footprinting assay was performed. The binding site of CRP to the cas3 promoter was identified, which is located from −38 to −51 and from −58 to −66, overlapping the predicted core motif of the CRP-binding site ( Figure 6D).
To further investigate the regulation of CRP to cas3 protomer activity, the predicted core binding site mutant promoter was cloned into reporter plasmid (pRCL5). The activity of the mutant cas3 promoter was almost 3.5 times lower than the original cas3 promoter ( Figure 6E), which indicated that the CRP-cAMP activates cas3 expression by binding to the transcriptional initiation area (ATGAGCAGN 7 AAATAGCCCGCTG).

The Effect of GCS on cas3 Expression is Mediated by CRP
GCS catalyzes glycine to CO 2 and NH 3 and transfers the methylene to H 4 folate to form N 5 , N 10 -mTHF [26], which might provide one-carbon unit for synthesizing cAMP required for CRP function. Therefore, GCS might regulate cas3 expression through CRP protein. To investigate whether the absence of CRP would influence GCS regulation of cas3 expression, the mutant ∆crp∆gcvP and the complemented strain were constructed. The expression of cas3 was evaluated by reporter plasmid and Western blotting. Notably, compared to WT, cas3 expression of ∆crp∆gcvP and ∆crpC∆gcvP were significantly reduced. Nevertheless, they were not any lower or higher than the ∆crp (Figure 7). In addition, cas3 expression restored in the complemented strain C∆crp. These results demonstrated that the regulation of GCS on cas3 expression is mediated by CRP. The EMSA for evaluating the binding ability of CRP-cAMP to the putative binding site. In the original probe group, at least 400 nM CRP protein was required for the entire shift of the 10 nM probe. (C) The EMSA for evaluating the binding ability of CRP-cAMP to the probe, which was amplified from the mutant sequence of the putative binding site. 400 nM CRP protein was incapable to shift the mutant probe. (D) The DNase I footprinting assay for sequencing the exact binding motif of CRP-cAMP. The nucleotides in red were consistent with the predicted binding site. (E) The β-gal activity of the original and mutant cas3 promoter was measured by reporter plasmids in WT (EC19) or ∆crp (EC69). The activity of mutant cas3 promoter was almost 3.5 times lower than the original cas3 promoter. All the data were mean ± SEM of at least three replicates, and the p value (* p < 0.05) was analyzed by the t-test. The expression level of cas3 by Western blotting. The value was calculated from the relative grayscale value of cas3 to GroEL, and the value of WT was normalized as one. The result was consistent with β-gal assat. All the data were mean ± SEM of at least three replicates, and the p value (* p < 0.05) was analyzed by the t-test.

The Sufficient Concentration of cAMP Restores cas3 Promoter Activity Reduced by the Mutation of GCS
To clarify whether cas3 expression regulated by GCS was due to the concentration of cAMP or the level of crp expression, cas3 and crp expression was evaluated with reporter plasmids. The cas3 promoter activity was detected in WT, ∆gcvP, and ∆crp cultured in LB with different amounts of cAMP addition. Crp expression in WT and ∆gcvP was also evaluated by reporter plasmids containing the crp promoter. Interestingly, the addition of excess cAMP resulted in a significant increase in cas3 promoter activity in WT and ∆gcvP, whereas cas3 promoter activity was irresponsive to cAMP in ∆crp. Notably, cas3 promoter activity in both WT and ∆gcvP reached the same level, which suggested that a sufficient concentration of cAMP could compensate for the decrease in cas3 promoter activity caused by gcvP deletion ( Figure 8A). Moreover, in both WT and ∆gcvP, gene crp expression remained at the same level ( Figure 8B). These results indicated that cAMP restored cas3 promoter activity reduced by the mutation of GCS, and GCS regulated cas3 promoter activity not by altering crp expression. In addition, we utilized HPLC to determine the cAMP, ATP, ADP, and AMP concentration of WT and ∆gcvP. Since the cAMP concentration is too low, it is difficult to distinguish the concentration of cAMP in WT and gcvP by the HPLC method. Interestingly, the concentration of ATP in WT and ∆gcvP is the same, but the concentration of ADP and AMP in ∆gcvP is significantly lower than that in WT ( Figure S4). This result further confirmed that N 5 , N 10 -mTHF has a certain correlation with cAMP.

The Activation of cse1 by H-NS Has No Effect on the Regulation of cas3 by GCS and CRP
Previous work has reported that the promoter of the cse1 operon is silenced by heat-stable nucleoid-structuring protein (H-NS) in E. coli [15]. To clarify the impact of GCS and CRP on the cse1 and cas3 operons in the absence of hns, the cse1 and cas3 promoter activity was firstly evaluated in WT and mutants. The absence of hns led to a 20 times increase in cse1 expression, which indicated hns deletion activated cse1 operon. With hns mutation, the deletion of gcvP had no effect on cse1 expression, and the deletion of crp only led to slight increase in cse1 expression ( Figure 9A). In the ∆hns (EC90), cas3 promoter activity was slightly increased compared to WT. However, the expression levels of cas3 in ∆hns∆gcvP (EC100) and ∆hns∆crp (EC110) were significantly lower than that in ∆hns ( Figure 9B). These results suggested that GCS and CRP regulated cas3 promoter activity rather than cse1 expression when the H-NS was absent. The β-gal activity of the cas3 promoter was measured by reporter plasmids in WT, ∆gcvP, or ∆crp cultured in LB media supplemented with 0, 5, 7, 9, 11, 13, or 15 mM cAMP. The addition of excess cAMP resulted in a significant increase in cas3 promoter activity in WT and ∆gcvP. The cas3 promoter activity in both WT and ∆gcvP reached the same level. (B) The β-gal activity of crp promoter was measured by reporter plasmids in WT (EC13) or ∆gcvP (EC23). In both WT and ∆gcvP, gene crp expression remained at the same level. All the data were mean ± SEM of at least three replicates, and the p value (p < 0.05) was analyzed by the t-test. The absence of hns led to a 20 times increase in cse1 expression, and with hns mutation, the deletion of gcvP had no effect on cse1 expression, and the deletion of crp led to slight increase in cse1 expression. (B) The β-gal activity of the cas3 promoter was measured by reporter plasmids in WT, ∆gcvP, ∆crp, ∆hns (EC91), ∆hns∆gcvP (EC101), and ∆hns∆crp (EC111) cultured in LB. In the ∆hns, cas3 promoter activity was slightly increased compared to WT. The expression levels of cas3 in ∆hns∆gcvP (EC100) and ∆hns∆crp (EC110) were significantly lower than that in ∆hns. All the data were mean ± SEM of at least three replicates, and the p value (* p < 0.05) was analyzed by the t-test.

Mutation of gcvP or crp Increases the Stability of Foreign Plasmids
To further investigate the impact of GCS and CRP on foreign DNA within bacteria, the stability of plasmids was evaluated by plasmid transformation assay. The transformation efficiencies of both pGEX1 and pGEX2 which contain the spacer cloned from MG1655, were remarkably increased in ∆hns∆gcvP, ∆hns∆crp, and ∆hns∆cas3 compared with ∆hns ( Figure 10). This elucidated that GCS and CRP regulated the interference of CRISPR/Cas to defend against invasive foreign DNA. Furthermore, GCS and CRP affect interference of CRISPR/Cas by regulating cas3 expression. Figure 10. Mutation of gcvP or crp increases the stability of plasmids. The plasmids containing the CRISPR1 (pGEX1) or CRISPR2 (pGEX2) loci of MG1655 and the empty plasmids (pGEX) were transferred into WT, ∆hns, ∆hns∆gcvP, ∆hns∆crp, and ∆hns∆cas3. The stability of the plasmid was determined by calculating the percentage of bacteria containing the pGEX1 or pGEX2 in the bacteria containing the pGEX plasmids. The transformation efficiencies of both pGEX1 and pGEX2, which contain the spacer cloned from MG1655, were remarkably increased in ∆hns∆gcvP, ∆hns∆crp, and ∆hns∆cas3 compared with ∆hns. All the data were mean ± SEM of at least three replicates, and the p value (* p < 0.05) was analyzed by the t-test.

Discussion
CRISPR/Cas is one of the important defending mechanisms against bacterial phage infection, however, most research has focused on the regulation of the cse1 operon in E. coli [15,24,33]. Herein, we studied the regulation factors of cas3, which were screened by transposon mutation and DNA pull-down technology. Previous reports have shown that GCS is related to many biological properties of bacteria [34,35], however, the further mechanism has not been elucidated. Meanwhile, another regulator CRP was identified of which the mechanism may be associated with GCS. Thus, the GCS and the CRP were selected to clarify their mechanism in regulating cas3 expression.
Glycine is the substrate catalyzed by gcvP, and was reported to activate gcvTHP operon [28,35]. Nevertheless, our study indicated that supplemented glycine promoted gcvTHP expression in minimal media rather than LB. We assumed that the component of LB is complicated, containing glycine, which leads to high expression level of gcvTHP regardless of whether there has been supplemented glycine in LB. Thus, the highly expressed GcvTHP is capable of cleaving sufficient exogenous glycine to promote cas3 promoter expression. However, there is no glycine in the minimal media which keeps the gcvTHP operon expression at lower level. Then, with the addition of glycine, the expression of gcvTHP increased followed by elevated cas3 promoter activity.
The present study showed that the predicted core CRP-binding site in MG1655 shared 7 nucleotides with E. coli consensus (TGTAAN 6 TCATG), while a previous report shared the other 7 nucleotides with E. coli consensus [24]. The mutated probe of the entire predicted core CRP-binding site was constructed which led to the loss of CRP-binding capacity. However, according to the results of the DNase I footprint of the sequenced cas3 promoter, there were 4 nucleotides overlapped with the predicted CRP core-binding site. As a result, the mutated 4 nucleotides brought the significant decrease of cas3 expression. Interestingly, although there are few reports on the systematic study of cas3 regulation, an earlier study identified cas3 as a part of the CRP regulon, showing high identity with our sequenced CRP-binding site [36].
Using transcriptome and bioinformatic techniques, CRP is found able to regulate the transcriptional initiation of more than 200 promoters within E. coli, and many of their regulated mechanisms have been illustrated [36][37][38][39]. In most cases, CRP positively controls gene expression [40], and in our results, the regulation of CRP on cas3 was consistent with this report. The positive regulation of CRP on the CRISPR/Cas system was also reported previously in Thermus thermophiles of which Type I-E (both cas3 and cse1 operons) and Type III-A systems are activated by CRP [40]. However, an opposite discovery showed that CRP negatively regulates Type I-E cse1 operon expression within E. coli. This is mainly due to the CRP repressing cse1 expression by competing for a binding site with LeuO, which is an activator antagonizing H-NS. Moreover, their results mentioned that CRP has no significant influence on cas3 expression [24]. This inconsistency may be due to the different growing phases of the bacteria collected. In Pseudomonas aeruginosa, the quorum-sensing system was observed as regulating Type I-E cas3 expression reported by Høyland-Kroghsbo et al. In their report, minimal cas3 expression is found at low cell density, but at a higher cell density, cas3 expression is activated. In addition, the mutation in both AI synthase lasI and rhlI leads to a significant decrease in cas3 expression [41]. Another study reported that when bacteria grows into the stationary phase, the concentration of cAMP is augmented [42]. Our results demonstrate that given sufficient cAMP, cas3 expression is elevated. Besides, the QS system in E. coli was reported to be repressed by CRP. Therefore, according to the correlation among the CRP, QS, and CRISPR/Cas system, we proposed a pathway of the correlation between cell density and the activation of the CRISPR/Cas system, where the increased cAMP concentration at high cell density would enhance the regulatory capability of CRP, leading to an increase in cas3 expression.
The intermediate metabolite N 5 , N 10 -mTHF produced by the cleavage of glycine by GCS can provide one-carbon unit for the synthesis of purine nucleotides, which is synthesized in the de novo biosynthetic pathway. In this pathway, the first generated nucleotide is inosine monophosphate, which contributes to the synthesis of various intermediates, including AMP and GMP [43]. Under the continuous catalysis of kinases, ATP and GTP are generated by AMP and GMP, respectively. Therefore, glycine plays a significant role in the biosynthesis of purines, ATP and GTP [44]. In addition, ATP can be catalyzed to remove a pyrophosphate to form cAMP [45]. Moreover, the regulatory function of CRP can be inhibited by glucose because the inactivation of adenylate cyclase by glucose leads to a lack of cAMP [46]. The inhibition of CRP by glucose was verified in our study which suggested cAMP was essential for CRP to regulate the CRISPR/Cas system. Here comes the question whether there is a functional correlation between GCS and CRP when they are regulating cas3 expression? To answer this question, the gcvP further mutated on the ∆crp mutant which had no further impact on cas3 expression, which suggested the indispensable role of CRP for the GCS regulation of cas3 expression. More importantly, we found that a sufficient concentration of cAMP could restore the cas3 promoter activity that was reduced by the mutation of gcvP. However, the crp expression was not affected by gcvP, and the ∆crp mutant is irresponsive to cAMP. A previous study reported that GCS was found responsible for ATP generation. In one-carbon metabolism cycle, per mole of glycine is used to generate one mole of ATP, with NAD + and NADP + acting as cofactors [47]. In addition, our present study confirmed that ADP and AMP levels were significantly reduced after gcvP gene mutation. In summary, we presumed that the interruption of GCS leads to a decrease in cAMP concentration, which affects the formation of CRP-cAMP, and consequently leads to a decreased cas3 expression ( Figure 11).
In fast growing E. coli, the CRISPR/Cas system has not shown to be active in WT, either for immunity or for acquisition of new spacers. A previous work showed that by using an engineered spacer matching phage λ, the bacteria gained weak protection against phage infection. However, depression of the CRISPR/Cas system by mutation of the hns gene results in a higher level of protection [48]. The activation of CRISPR/Cas in E. coli is closely related to a heat-stable nucleoid-structuring protein (H-NS), which can bind non-specifically to the AT-rich region of dsDNA. Therefore, H-NS can inhibit the function of CRISPR/Cas by binding to the cas gene promoter [15,33]. Another transcription activator, LeuO, can relieve the H-NS-mediated repression of CRISPR/Cas by competitively binding to the promoter of the cas gene, however, LeuO has to be overexpressed to relieve H-NS-mediated repression and the silence of leuO has no impact on cse1 expression [15]. The influence of cse1 was ruled out, as hns was knocked out, when bacterial susceptibility to phage and the stability of foreign plasmids were studied. Given the result that WT and the mutants without the anti-vB_EcoS_SH2 spacer maintained the same level of bacterial susceptibility to phage, it was confirmed that the changed interference ability induced by GCS and CRP was due to the altered cas3 expression rather than other mechanisms against phage. Figure 11. The proposed regulation model of the CRISPR/cas3 within E. coli. The N 5 , N 10 -mTHF is produced in both serine and glycine metabolism pathways. The disruption of GCS leads to a decrease in N 5 , N 10 -mTHF. Intracellular cAMP concentration correlates with N 5 , N 10 -mTHF production. Reduced cAMP weakens the binding ability of the CRP-cAMP complex, which is capable of activating cas3 expression by binding to the cas3 promoter. The altered cas3 expression affects interference with invasive DNA.

Conclusions
Bacteria have to meet their metabolic requirements to maintain their physiological state including replication, pathogenesis, and defense. Despite the critical nature of this nutritional interaction, the contribution of many metabolic pathways to defend bacteriophage have not yet been examined. To our knowledge, this study identifies the first contribution of the bacterial GCS for preventing them from phage infection. This contribution is closely linked to the CRISPR/Cas system through interaction with CRP. Our study describes the relationship between bacterial metabolism and phage defense mechanisms from the perspective of GCS and CRP. In prokaryotic and eukaryotic cells, GCS serves conservatively as one-carbon provider linked to the ATP generation [49]. During the serine, one-carbon cycle, glycine synthesis pathway, the glycine can be imported from the environment and utilized to generate ATP while maintaining one-carbon cycle. Under resource shortage situation, the bacteria have to keep a lower proliferation rate to save more energy. However, compared to the resource shortage situation bacteria experienced, the pressure from phage seems extremely dangerous. In conditions of resource abundance, the increased bacterial cell density will suffer the high risk of phage infection. Therefore, CRISPR/Cas system expression is dependent on the QS system in some kinds of reported bacteria. Our study implies that when bacteria encounter resource shortages, the CRISPR/Cas system might be shut down to free up more resources to get the bacteria through the tough time, and bacteria will highly express phage-counter measures during the high cell density phase. Taken together, cas3 expression was regulated by both GCS and CRP in E. coli. The GCS and CRP deletion decreased the expression of cas3. The CRP regulates cas3 expression by directly binding to the cas3 promoter in response to cAMP concentration and is indispensable for GCS regulation of cas3 expression. The down-regulation of cas3 expression consequently resulted in the decreased bacterial resistance to phage infection and interference with foreign DNA. Our results highlighted a clear cas3 expression regulation pathway, which might provide insight into interaction between bacteria and phage.