The bacterial strains and plasmids used in this work were listed in Table S2. Z. mobilis ZM4 was routinely grown at 30˚C in RM broth (20.0 g/L glucose, 10.0 g/L yeast extract, 1.0 g/L MgSO₄, 1.0 g/L (NH₄)₂SO₄, 2.0 g/L KH₂PO₄) or RM agar plate. If needed, kanamycin was added in the RM medium at 200 μg/mL. The cultures of Escherichia coli DH5α were grown in Luria-Bertani (LB) broth (1% peptone, 0.5% yeast, 0.5% NaCl) at 37˚C or with addition of antibiotics if needed.

The total RNAs were isolated from 6 mL culture of Z. mobilis ZM4 growing in RM for 12 h and 16 h under stationary condition by using the Bacterial RNA Kit (Omega Biotech Co, Atlanta, USA) according to the manufacturer’s instructions. Then PrimeScript™ RT reagent Kit with g DNA Eraser (Takara Biotech Inc., Dalian, China) was used to perform the reverse transcriptional reaction using 1.0 μg total RNA prior to removal of the genomic DNA and the specific primers of ZmC1S3.F, ZmC1S6.R, ZmC2S2.F, ZmC2S5.R (Table S3), respectively. The resulting cDNAs were used to detect the target transcripts of CRISPR loci P1 and P2 by PCR with the specific primer pairs (ZmC1S3.F/ZmC1S6.R, ZmC2S2.F/ZmC2S5). The PCR cycles were initiated at 94˚C for 4 min to denature the template DNA, followed by 25 or 32 amplification cycles. Each amplification cycle consisted of 94˚C for 30 sec, 55˚C for 30 sec and 72˚C for 30 sec. And then, the PCR products were analysis on 2% agarose gel. In addition, the same primer pairs were used to amplify the RNA samples to examine if the RNA samples was contaminated with the genomic DNA or not.

The invasion plasmid was constructed by the PCR using pBBR1MCS2 [23] as template and the overlapped primers (Table S3), which contained the spacer 1 and 3 of the CRISPR locus P1 of Z. mobilis ZM4. In addition, TGG or GG were introduced at 3’-end of the spacer sequence to serve as protospacer, respectively, since TGG or GG was demonstrated as an effective PAM site in the another type I-F of CRISPR-Cas systems [24] -[26] . PCR amplification with the KOD Plus DNA polymerase (Toyobo Co., Ozaka, Japan) following cycling steps: pre-denaturation 94˚C for 4 min; denaturation 94˚C for 30 sec, annealing at 52˚C - 54˚C (for various primer pair) for 30 sec, extension 68˚C for 6 min, 30 cycles; and final extension at 68˚C for 10 min. The PCR product, after digested by DpnI (Thermo scientific, Massachusetts, USA) was transformed into E. coli DH5α competent cells. The inserted spacer sequence and PAM site were confirmed by DNA sequencing.

Transformation of Z. mobilis ZM4 was following the previous description [27] with minor modification. In brief, the cells were cultured in RM broth to the middle log-growth phase at 30˚C in stationary condition. The cells were collected and washed with ice-cold dd H₂O containing 10% glycerol by centrifugation at 4000 rpm for 5 min by 4 - 5 times. And then, 1 μg plasmid DNAs were added into 100 μL competent cells in a 2 mm-diameter cuvette. After standing by on ice for 10 min, electroshock was performed at 2500 V/cm, 50 μF capacitance and 10 ohms resistance on the Gene PulserX-cell (BioRad Co. California, USA). Afterwards, the cells were rapidly added with 3 mL RM medium and incubated for 6 h at 30˚C. Finally, 0.2 mL of cells were dispensed on RM agar plates containing 200 μg/mL kanamycin. After incubation at 30˚C for three to four days, the numbers of transformants were first counted. In addition, the entrance of plasmid DNA into the colonies was confirmed with the colony PCR. All the transformation experiments with each plasmid were independently performed in triplicates or more.

First, the full length of the CRISPR locus P1 of Z. mobilis ZM4 was amplified using the primers (Cri01.F and Cri01.R) with the genomic DNAs as template. After digesting with Kpn I and Hind III, the PCR fragment was cloned into the plasmid pBBR1MCS2 digested with the same restriction endonucleases. The resulting plasmid assigned as pCri01, was confirmed by DNA sequencing. And then, three 32-bp sequences with a nearby PAM of TGG at 3’-end were selected as the protospacer at different genomic sites, two located in the gene ZMO0293 and one in ZMO0875. The selected sequence as an alternative spacer was inserted into the pCri01 in displacement of the original spacer 1 and spacer 3, or both via overlapping PCR. The PCR reactions with the Master Mix Kit (Vazyme, Nanjing, China) were performed using pCri01 as template and the overlapping primers (Table S3) as following steps: for the first 6 cycles, pre-denaturation 94˚C for 4 min, denaturation 94˚C for 30 sec, annealing at 54˚C for 30 sec, extension at 72˚C for 6.5 min; and for the subsequent 24 cycles, denaturation 94˚C for 30 sec, annealing at 62˚C - 68˚C (for various primer pairs) for 30 sec, extension at 72˚C for 6.5 min with an additional extension at 72˚C for 5 min. And then, the PCR products were digested by DpnI (Thermo scientific) and transformed into E. coli DH5α. The resulting genome targeting plasmids were sequenced to confirm the inserted DNA sequence. For targeting the genome, these plasmids (Table S2) were transformed into the ZM4 cells by electroporation as described in the above section. The number of transformants was accounted for each plasmid. For each plasmid, the transformation was independently performed in triplicates.

According to the databases CRISPRdb (crispr.u-psud.fr) and CRISPI (www.crispi.it/it/index.php), Z. mobilis ZM4 contains the CRISPR-Cas system with three CRISPR loci named as P1 to 3 and a cas gene cluster. The organization and structure of the CRISPR/Cas system in the genome was present in Figure 1. The CRISPR loci (P1-3) share the identical repeat sequence (GTTCACTGCCGCACAGGCAGCTTAGAAA); and contain the unique spacers interspaced by 8, 6, and 1, respectively. It is noticeable that the CRISPR P3 is not complete due to the last repeat sequence. The repeat sequence in these CRISPR loci of Z. mobilis ZM4 is completely identical with the other CRISPR arrays from Pseudomonas aeruginosa [15] , Pectobacterium atrosepticum [28] , Yersinia

pestis [29] and E. coli [15] , which should be classified into group 4 [30] . The majority of spacer sequence are 32 or 33 nt in length, none of which matches to the known viruses or plasmids by BLASTn against the Nucleotide Collection database (www.blast.ncbi.nlm.nih.gov/Blast.cgi).

Unlike most CRISPR-Cas systems in the other bacteria or archaea in which the cas gene cluster is located immediately near by the CRISPR locus [3] , the cas gene clusterin Z. mobilis ZM4 is far distant from the CRISPR loci in the genome and consisted of six cas genes: cas1, cas3, csy1, csy2, csy3, and csy4 (or cas6f) (Figure 1). The components and organization of the cas gene cluster in Z. mobilis ZM4 are similar with that of Y. pestis Z176003 [29] . Accordingly, the CRISPR-Cas system in Z. mobilis ZM4 belongs to Type I-F or Ypest [31] .

In order to confirm if the CRISPR loci in Z. mobilis ZM4 is or not transcribed, a strategy based on the reverse transcription (RT) and PCR reaction was adopted (Figure 2(a)). Two total RNA samples, isolated from the cells growing for 10 and 16 h in RM broth, were reversely transcribed with each specific primer, e.g. Zm4C1S3.F, Zm4C1S6.R, Zm4C2S2.F or Zm4C2S5.R, which is specific to complement with a specific spacer sequence on the double strands of the CRISPR locus P1 and P2, respectively. First, PCR reaction was performed using the RNA samples as template, indicating that no genomic DNA was contaminated in these RNA samples (Figure 2(b), line 1). And then, the same PCR amplification with the routine 32 cycles was performed with two pairs of primers (ZmC1S3.F/ZmC1S6.R and ZmC2S2.F/ZmC2S5.R for CRISPR loci P1 and P2, respectively). Unexpectedly, the PCR products were obviously observed by the cDNA templates by reversely transcribed either with the forward primer or the reverse primer (Zm4C1S3.F or Zm4C1S6.R for the CRISPR locus P1, and Zm4C2S2.F or Zm4C2S5.R for the CRISPR locus P2). Subsequently, PCR reaction with reduced cycling number (25) was also preformed again. A representative result was shown in Figure 2(b), indicating that both of the cDNA templates reversely transcribed with the forward or reverse primer could be amplified to produce the PCR products. However, higher yield was obtained with the cDNA template from the reverse primer than that from the forward primer under the same condition. These data indicated that the CRISPR arrays of Z. mobilis ZM4 were transcribed on both strands, but with different expression level. The CRISPR locus in Sulfolobus acidocaldarius was also reported to be transcribed on both strands [32] . Conclusively, these experimental evidences indicated that the CRISPR loci are indeed transcribed in Z. mobilis ZM4 under the normal growing condition. Further, we can infer the transcriptional direction and the leader region of the CRISPR loci in this bacterium. Accordingly, the characteristic elements such as TATA box and AT-rich region can be inferred in the upstream region of the CRISPR loci P1 and P2.

Further, the transcripts of six cas genes in Z. mobilis ZM4 growing at various conditions could be easily detected in the previous transcriptional analysis [33] [34] . For example, the relative transcriptional levels of the cas genes are presented in Table 1, which is retrieved from the transcriptome data in the previous studies [35] . These data indicate that the cas genes may be transcribed constitutionally.

Since the cas gene cluster and the CRISPR loci are transcribed actively in Z. mobilis ZM4, this system is expected

^*Z. mobilis ZM4 was grown in RM medium at 30˚C in stationary condition for various times. And 220 g/l glucose was added into the medium at 6 h. The data were obtained from the microarray assay described as before [35] .

to be active to defense against the invader DNA elements. However, no virus has been reported to associate with Z. mobilis, so the plasmid-based invader assay was applied to this purpose [13] [36] . First, a series of invader plasmids were constructed (Table S2), with in which the first spacer and the third spacer of CRISPR locus P1 with the deduced PAM site (TGG or GG) at the 3’-end on the target strand were inserted in the shuttering plasmid pBBR1MCS2by overlapped PCR to serve as an artificial protospacer. All the constructed invasion plasmids and the parent plasmid were transformed into the competent cells of ZM4 by electroporation under the same condition. The transformants were selected on the RM agarplates with addition of 200 μg/mL kanamycin. In addition, the transformants were randomly picked up and used to perform PCR, which confirmed entrance of the plasmid DNA into the bacterial cells (data not shown).

Expectedly, the numbers of transformants of the invader plasmids containing the artificial protospacer with PAM motifs (TGG and GG) were significantly less than that of the parent plasmid (without no spacer inserted) (Figure 3(a)). The data of three individual transformation experiments were given in Figure 3(b). Reduce of the relative transformation rate was contributed to the immune defense of the invader plasmid DNA by the native CRISPR-Cas system in the bacterial cells [13] . However, the immune activity is less effective with the PAM motif of TGG than GG because the transformation rate trends to about 1% - 4% with GG as PAM in comparison to >20% with TGG as PAM (Figure 3(b)). Based on the structural model of the Cascade complex revealed in the Type I CRISPR-Cas system [7] [8] , the Cascade complex with the crRNA biogenerated from the CRISPR transcripts in Z. mobilis ZM4 would be formed, and then guide to target the protospacer (Figure 3(c)).

In addition, the invader plasmid DNAs were still escaped from the immune interference by the CRISPR-Cas system (Figure 3), which is consistence with the previous studies with the archaeal CRISPR-Cas system [13] .

In order to confirm the immune function of the CRISPR-Cas system in Z. mobilis ZM4, a series of targeting plasmids were constructed (Table S2), in which the selected target sequences at different genomic sites were cloned into the CRISPR array P1 in placement of the original spacer 1 and/or 3. These plasmids are expected to drive the modified CRISPR array to express and to generate the special crRNA in the bacterial cells. After transformation into cells, the targeting plasmids led to almost no transformant to form on the kanamycin-cont- aining RM agar plates (Figure 4(a)). In contrast, number of the transformants was much more with the control plasmid pCri01 which hosting a native CRISPR locus P1 (Figure 4(b)). These data indicate that the crRNA derived from the modified CRISPR array with spacer sequence targeting to the specific genomic sites in the plasmids could guide to cut the genome at the target protospacer site. As a result, the double-stranded breakage in the genome made by the interference activity will lead to the cells to die [31] . Noticeable, the transformation rate is usually less than 4% with various spacers or location in the CRISPR array (Figure 4). In contrast, the transformation rate to target the invader plasmids is in the range of about 20% when T/AGG PAM motifs were used (Figure 3). Therefore, the immune activity to target the genome is stronger than to target the invader plasmids under this condition.