Using Chromosome and Plasmid CRISPRi Systems to Turn Off Genes in Lactococcus lactis

To monitor expression levels, a fluorescent variant of dCas9 was generated by fusing its coding sequence at the 3′ end to the superfolder GFP (sfGFP) gene, connected through the flexible linker AGSGGEAEA [20,21]. For inducible expression, the dcas9 or dcas9-sfgfp gene was placed under the control of the nisin-responsive promoter P_nisA in a pNZ8048-derived plasmid [7], while the sgRNA cassette was constitutively expressed from the P_usp45 promoter on a pTLR-based backbone (Figure 1A,B) [20,22]. sgRNA design followed previously established criteria to ensure efficient silencing: targeting of the non-template DNA strand, a minimum 12 bp base-pairing region, and complete sequence complementarity within the 7 nucleotides adjacent to the PAM site (5′-NGG-3′). To minimize off-target interactions, the 12 bp region immediately upstream of the PAM was selected [13]. The inducible host strain L. lactis NZ9000 was used [23]. To verify inducibility, fluorescence intensity was measured in NZ9000 harboring pNZ-P_nisA-dcas9-sfgfp. Increasing concentrations of nisin (0–20 ng/mL) led to a dose-dependent rise in fluorescence, confirming controlled expression of dCas9-sfGFP (Figure 1C). Next, a luciferase reporter strain (NZ9000 pseudo10::P_usp45-luc) was constructed to evaluate CRISPRi function. The chromosomal locus pseudo10 is a commonly used silent region that allows stable chromosomal integration without affecting L. lactis growth. Cells carrying pNZ-P_nisA-dcas9-sfgfp and pTLR-P_usp45-sgRNAluc were analyzed for luciferase activity. Upon induction with 5 ng/mL nisin, luciferase activity decreased markedly, demonstrating that the dCas9-sfGFP fusion retained silencing activity (Figure 1D). At the end of exponential growth, a 74% reduction in luciferase expression was observed without detectable effects on bacterial growth.

To further evaluate the functionality of the CRISPRi system, its effects on cell growth and morphology were examined. The gene acmA, which encodes the major autolysin responsible for cell separation and autolysis in L. lactis NZ9000 was selected as a target. Deletion of acmA results in the formation of extremely long chains of cells [24].

Adding nisin to strains expressing only dCas9 or dCas9-sfGFP had minimal impact on growth (Figure 2B,C, growth curves), whereas co-expression of the corresponding sgRNA targeting acmA caused a noticeable reduction in growth rate and final OD (Figure 2D). Microscopy analysis revealed that, as expected, long chains of cells were formed in the acmA knockdown strain (right micrographs in Figure 2A–D).

No detectable fluorescence from dCas9-sfGFP was observed in uninduced cells (left micrograph in Figure 2C). Nevertheless, long chains were present even without nisin induction (Figure 2D, left micrograph). These findings indicate that basal dCas9 expression resulting from P_nisA leakiness, while not affecting growth rate or final OD, is sufficient to block transcription and elicit the corresponding mutant phenotype.

To minimize background activity of uninduced dCas9-sfGFP, the copy number of the dcas9-sfgfp cassette was reduced by chromosomal integration. The P_nisA-dcas9-sfgfp construct was inserted into the transcriptionally silent pseudo29 locus of L. lactis NZ9000 [25], while the sgRNA continued to be expressed constitutively from P_usp45 on the plasmid pTLR. Expression levels of dCas9-sfGFP were compared between plasmid- and chromosome-based systems. Nisin (10 ng/mL) was added to cultures, and relative fluorescence was measured after 3 h of induction. As shown in Figure 3B, dCas9-sfGFP expression from the chromosome was approximately 20-fold lower than from the plasmid.

When targeting acmA with sgRNA in cells carrying the chromosomal dcas9-sfgfp cassette, strain NZ9000 pseudo29::P_nisA-dcas9-sfgfp (pTLR-P_usp45-sgRNA_acmA) displayed normal morphology in the absence of nisin. In contrast, the plasmid-based system showed observable phenotypic changes under the same conditions (compare Figure 3C with Figure 2D). These results demonstrate that chromosomal insertion of dCas9-sfGFP significantly reduces basal expression, allowing a tighter, more controllable chromosome-basedCRISPRi system, cbCRISPRi.

YbeY is a highly conserved protein found across nearly all sequenced bacterial species. It is typically involved in ribosome maturation and quality control as an endoribonuclease, and implicated in biofilm formation, stress responses, and virulence [26,27,28,29,30]. In Sinorhizobium meliloti, YbeY is required for sRNA-mediated silencing of amino acid ABC transporters [28], and in Pseudomonas aeruginosa, it regulates RpoS via the sRNA ReaL [27]. Its role in L. lactis remains unexplored, though it may function as a candidate sRNA chaperone in a species lacking Hfq, CsrA, or ProQ homologs [31,32,33,34,35].

Although ybeY is essential in several bacteria [36,37,38], it can be deleted in P. aeruginosa, S. meliloti, and E. coli, with pleiotropic effects on growth, toxin production, or cell aggregation [39,40]. Attempts to delete ybeY in L. lactis using the temperature-sensitive pGhost system or the pCS1966 oroP-based selection/counterselection system failed, even after multiple attempts (Figure S1) [9,41]. Standard methods relying on double-crossover homologous recombination failed to delete ybeY from the chromosome, despite multiple alternative attempts (Figure S1).

Four sgRNAs targeting ybeY were designed for cbCRISPRi, and qPCR confirmed that all four (strains CHML013–CHML016) efficiently silenced transcription (Figure 4A,B). Three sgRNAs decreased ybeY expression by roughly 200-fold, while sgRNA2 exhibited weaker silencing. This lower efficiency is likely due to its 12-base region adjacent to the PAM having a GC content of about 33%, compared to 42–58% for the other sgRNAs. Silencing ybeY also reduced expression of downstream genes dgkA-llmg1485 but did not affect upstream genes. Knockdown of the ybeY-dgkA-llmg1485 operon caused severe growth defects (Figure 4D,E), while overexpression of ybeY in NZ9000 had no growth impact (Figure S2). To dissect the roles of the three genes in the observed growth defects, we constructed plasmid-based complementation strains. The first expressed YbeYp with a synonymous mutation at the sgRNA target site to prevent silencing (Figure 4A). The second complemented dgkA and llmg_1485 under their native operon promoter, and the third co-expressed the mutated YbeYp together with dgkA and llmg_1485. As shown in Figure 4D,E, restoring only ybeY or dgkA–llmg_1485 partially alleviated the growth defect, while co-expression of all three genes fully rescued it. These findings indicate that loss of YbeY has the strongest impact, but reduced expression of dgkA and/or llmg_1485 also compromises growth.

Plasmid-based complementation showed partial recovery when either ybeY or the two downstream genes were complemented, and complete recovery when all three were expressed (Figure 4D,E). To prevent overexpression, dgkA-llmg1485 driven by their native promoter were integrated into the pseudo10 locus (strain CHML20). This restored the growth defects while ybeY remained silenced (Figure 5B). No differences in the recovery of growth defects were observed between CHML020 and the plasmid-based dgkA–llmg_1485 complementation strain CHML017. qPCR confirmed transcriptional restoration (Figure 5C). Light microscopy with live/dead staining indicated no viability difference in exponential phase, but stationary-phase cell death ranged from 15 to 54%, higher than 0.5–3% in the parent strain CHML012 (Figure 5D). These results confirm ybeY essentiality and the importance of downstream operon genes for normal growth of L. lactis.

Deletion of ybeY in several bacterial species results in the appearance of an additional band corresponding to 17S rRNA on agarose gels, representing the 16S rRNA precursor [29]. In L. lactis, however, knockdown of ybeY did not produce such a band (Figure S3). qPCR analysis also revealed no significant differences in the 5′ ends, middle regions, or 3′ ends of immature 16S rRNA and 23S rRNA (Figure S4). These findings suggest that L. lactis YbeY may not be primarily involved in 16S rRNA maturation, although a very low level of residual YbeY expression in the knockdown strain could suffice to complete this process.

To further investigate L. lactis YbeY function, RNA-seq was performed comparing the transcriptomes of the ybeY-silenced strain and the control strain lacking the sgRNA-expressing plasmid, L. lactis NZ9000-dCas9. All strains were grown at 30 °C in GM17 medium to early log phase (OD600 = 0.4), followed by induction with 10 ng/mL nisin for 1 h to produce sgRNA and/or dCas9. Total RNA was isolated, rRNA-depleted, sequenced, and analyzed. This induction ensured sufficient dCas9 for function without causing severe side effects. RNA-seq revealed that ybeY silencing broadly affected the transcriptome, with most highly differentially expressed genes (DEGs) being downregulated. Functional and gene set enrichment analyses using FUNAGE-Pro [42] annotated DEGs to Gene Ontology (GO) terms (Figure 6). Many downregulated genes encode proteins of the 50S and 30S ribosomal subunits. Although no accumulation of immature 16S and 23S rRNA ends was detected, as was the case in E. coli and P. aeruginosa [27,29], ribosomal protein genes were largely repressed, indicating a role of YbeY in ribosome processing in L. lactis. Additionally, genes encoding membrane proteins, ATP synthase subunits, sugar/amino acid permeases, other transporters, and components of the fab and acc operons were downregulated. Some changes may reflect secondary effects after the 1 h induction.

The cbCRISPRi system was also applied to the 66-nt noncoding sRNA ArgX. Deletion of ArgX elevates arc and arg operon expression [43], and qPCR confirmed effective knockdown by CRISPRi with expected transcriptomic effects (Figure S5). Due to PAM constraints, only 42 of 184 sRNAs in L. lactis are potentially targetable (Table S2). Since YbeY has been reported to degrade an sRNA involved in oxidative stress in P. aeruginosa [27], sRNA expression in the ybeY-silenced strain was examined. Ten of 186 trans-encoded sRNAs and 2 of 60 cis-encoded antisense RNAs identified in L. lactis MG1363 [44] were differentially expressed. Whether YbeY directly or indirectly affects these sRNAs requires further study.

The bacterial strains used in this study are listed in Table 1. L. lactis NZ9000 and its derivatives were grown at 30 °C in Difco M17 medium (BD, Franklin Lakes, NJ, USA) containing 0.5% (wt/vol) glucose (GM17). For fluorescence and luminescence analyses, L. lactis strains were grown in a chemically defined medium supplemented with 1% (wt/vol) glucose (CDM) [16]. When needed, erythromycin or chloramphenicol was added at a final concentration of 5 μg/mL. Chemically defined SA medium supplemented with 0.5% (wt/vol) glucose and 20 μg/mL 5-fluoroorotic acid (5-FOA; Sigma-Aldrich, St. Louis, MO, USA) as the sole pyrimidine source was used for the selection of chromosomal insertions [9]. E. coli DH5α was used as a cloning host; it was grown aerobically at 37 °C in LB media (Formedium, Norfolk, UK) supplemented with, when needed, erythromycin at a final concentration of 200 μg/mL. All chemicals were obtained from Sigma-Aldrich.

Standard molecular cloning was performed essentially as described previously [50]. Chromosomal DNA from L. lactis was isolated using the GenElute Genomic DNA Kit (Sigma-Aldrich). Plasmids and PCR products were isolated and purified using a High Pure plasmid isolation and PCR purification kit (Roche Applied Science, Mannheim, Germany) and a NucleoSpin gel and PCR cleanup kit (Macherey-Nagel, Düren, Germany) according to the manufacturers’ instructions. PCR was performed with Phusion or DreamTaq polymerase (both from Fermentas, St. Leon Roth, Germany) according to the manufacturer’s protocol. The obtained PCR products were mixed and treated with a mixture of Quick-Fusion enzymes (BIO-Connect Services BV, Huissen, The Netherlands), yielding 15-nucleotide overhangs that were annealed to complementary overhangs. No ligation was needed; Quick-Fusion-treated mixtures were directly used to transform E. coli. The oligonucleotides used in this study are listed in Supplementary Table S1 and were purchased from Biolegio BV (Nijmegen, The Netherlands). Competent E. coli cells were transformed using heat shock [51], while electrocompetent L. lactis cells were transformed using electroporation [52] with a Bio-Rad gene pulser (Bio-Rad Laboratories, Richmond, CA, USA). All nucleotide sequencing was performed at Macrogen Europe (Amsterdam, The Netherlands).

Pertinent regions of all the plasmids constructed in this study were sequenced to confirm their proper nucleotide sequences. Streptococcus pyogenes dcas9 (dcas9) was obtained from plasmid pJWV102 [46] by PCR using the primers 0032-USER_Pnis_dCas9_F and 0220-dCas9-USER_R, while linearized pNZ8048 vector was obtained using the primers 0217-pNZ8048_USER_F/0221-Pnis_pNZ8048_R. The fragments were fused employing the USER cloning kit (New England Biolabs, Ipswich, MA, USA) according to the manufacturer’s instructions, with the exception of using only one half of the recommended volume per reaction. The reaction mixture was directly used to transform competent L. lactis NZ9000. The resulting vector was labeled pNZ-P_nisA-dcas9. The plasmid pNZ-P_nisA-dcas9-sfgfp, which expresses a Cas9 variant that couples C-terminally with sfGFP via the flexible linker sequence AGSGGEAEA, was generated as follows: dcas9 was amplified from pJWV102 using the primers 0032-USER_Pnis_dCas9_F/0218-dCas9_USER_R. The sfgfp gene with the polylinker-encoding sequence at its 5′ end was amplified from the plasmid pLG-MG1 [21] using the primers 0215-sfGFP_USER_F/0219-linker_sfGFP_USER_R. The fragments were fused through USER cloning, and the mixture was subsequently used to transform competent L. lactis NZ9000 to obtain the proper construct.

The DNA fragment encoding the single guide RNA targeting the luciferase gene (sgRNAluc) was amplified from the plasmid pPEPX-P3-sgRNAluc [46] by the primers 0149-sgRNA_F/0150-sgRNA_R. The constitutive L. lactis promoter P_usp45 was amplified from the plasmid pSEUDO::Pusp45-sfgfp(Bs) [20] with the primers 0147-Pusp45_F/0148-Pusp45_R, and the linearized pGHost vector was obtained after amplification using the primers 0145-pGHost_F/0146-pGHost_R. The three fragments were then fused with the USER cloning kit, and the reaction mixture was subsequently used to transform competent E. coli DH5α, resulting in the plasmid pGHost-P_usp45-sgRNAluc. The P_usp45-sgRNAluc cassette, obtained by double digestion of the latter plasmid with the restriction enzymes NcoI and XhoI, was ligated into pTLR [22] treated with the same enzymes. The ligation mixture was used to transform competent E. coli DH5α, yielding the plasmid pTLR-P_usp45-sgRNAluc. The sgRNAluc sequence is transcribed directly after +1 of the P_usp45 promoter and contains 20 nucleotides (nt) as the base-pairing region, which targets the luciferase gene and is followed by an optimized sgRNA [53]. The plasmid pTLR-P_usp45-sgRNAluc was used as the template for the generation of other sgRNA expression plasmids by quick-fusion cloning (see the next section).

The 20-nt guide sequences of sgRNAs targeting different genes were selected with the CRISPR Primer Designer [54]. Briefly, a search was performed within the coding sequence of each gene for a 14-nt specificity region consisting of the 12-nt “seed” region of the sgRNA and GG of the 3-nt PAM (GGN). sgRNAs with more than one binding site within the L. lactis genome, as determined by BLAST (v 2.12.0) searches, were discarded. The guide sequence was chosen to be as close as possible to the 5′ end of the coding sequence of the targeted gene [13]. Cloning of sgRNA sequences was performed by quick-fusion cloning (Figure S6). The primer pair 0153-sgRNA_backbone_FW/0154-sgRNA_backbone_RV was designed for linear amplification of the plasmid pTLR-P_usp45-sgRNAluc. The primers bind directly upstream and downstream of the 20 bp guide sequence to enable easy sgRNA sequence swapping. To fuse the new 20-nt guide sequence into the linearized vector, two 50-nt complementary primers were designed for each target gene. Each primer contained 15 nt at one end, overlapping with the sequence on the 5′ end of the linearized vector, followed by the 20-nt gene-specific guide sequence and then 15 nt overlapping with the sequence on the 3′ end of the linearized vector. The two 50-nt complementary primers were annealed in TEN buffer (10 mM Tris, 1 mM EDTA, 100 mM NaCl, pH 8) by heating at 95 °C for 5 min and slowly cooling to room temperature. The annealed product was fused with the linearized vector using the Quick-Fusion Cloning Kit (BIO-Connect Services BV) according to the manufacturer’s instructions, except for using only one half of the recommended volume per reaction. The reaction mixture was directly used to transform competent E. coli DH5α, and the designated plasmid was subsequently introduced into the strains CHML009 (NZ9000 pseudo29::P_nisA-dcas9) or CHML010 (NZ9000 pseudo29::P_nisA-dcas9-sfgfp).

The luc gene was amplified from the plasmid pPEP23 [55] using the primers 0189_luc_F/0190-luc_R, while the vector pSEUDO::Pusp45-sfgfp(Bs) [20] was linearized by amplification using the primers 0187-pseudo10_R/0188-pseudo10_F. The two fragments were fused by employing the Quick-Fusion Cloning Kit protocol. The reaction mixture was directly used to transform competent E. coli DH5α to pick up plasmid pSEUDO10-P_usp45-luc. This plasmid, which cannot replicate in L. lactis, was introduced into L. lactis NZ9000 via electroporation. Cells in which the plasmid had integrated into the genomic locus pseudo10 were obtained by selection on erythromycin. The proper integrant strain, NZ9000::pseudo10::P_usp45-luc, was subsequently selected by growing on selective SA media plates containing 20 μg/mL 5-FOA [9].

The flanking regions of pseudo29 were amplified using 0247-pseudo29_UP_F/0248-pseudo29_UP_R and 0249-pseudo29_DOWN_F/0250-pseudo29_DOWN_R, while the linearized vector pCS1966 was obtained by PCR amplification using the primer pair pCS1966_1FW/pCS1966_1RV. The fragments were fused by Quick-Fusion Cloning as described above, after which the reaction mixture was used to transform competent E. coli DH5α. The resulting vector was designated pSEUDO29. The P_nisA-dcas9-sfgfp cassette was amplified using the primers 03-dcas9-pseudo29-F/04-dcas9-pseudo29-R, while linearized pSEUDO29 was obtained by amplification employing the primers 01-pseudo29_F/02-pseudo29-R. The fragments were fused with a Quick-Fusion Cloning Kit. A transformant carrying pSEUDO29-P_nisA-dcas9-sfgfp was selected in E. coli DH5α. This plasmid, which does not replicate in L. lactis, was introduced into L. lactis NZ9000 via electroporation; cells in which a two-step homologous recombination event occurred were selected on selective SA medium plates supplemented with 20 μg/mL 5-FOA [9]. The obtained strain was labeled NZ9000::pseudo29::P_nisA-dcas9-sfgfp.

L. lactis was grown overnight in CDM at 30 °C and then diluted to an optical density (OD) at 600 nm (OD₆₀₀) of 0.1 in fresh CDM. The mixture was incubated until the mid-log phase (OD₆₀₀ = 0.5) was reached, after which the cells were diluted again to a starting OD₆₀₀ of 0.05 in fresh CDM supplemented with different concentrations of nisin. Then, each culture was dispensed to 200 μL, in triplicates, in wells of a 96-well microtiter plate (Polystyrol, transparent, flat, and clear bottom; Corning, New York, NY, USA). The OD₆₀₀ and fluorescence signal (GFP: excitation 485 nm/emission 535 nm) were measured every 10 min at 30 °C in an Infinite 200 Pro plate reader with I-control 1.10.4.0 software (Tecan Group Ltd., Männedorf, Switzerland). For luminescence measurements, the cells were grown as above, except that when the cells were diluted to OD₆₀₀ = 0.05, D-luciferin sodium salt (SYNCHEM OHG, Altenburg, Germany) was added to a final concentration of 2.5 mg/mL. The cultures were subsequently divided in triplicate (200 μL each sample) in 96-well polystyrol microtiter plates (white, flat, and clear bottom; Corning). The OD₆₀₀ and luminescence signals were measured every 10 min at 30 °C in an Infinite 200 Pro plate reader with I-control 1.10.4.0 software.

All micrographs were obtained with a DeltaVision Elite inverted epifluorescence microscope (Applied Precision, GE Healthcare, Issaquah, WA, USA) equipped with a stage holder, a climate chamber, a seven-color combined set InsightSSI solid-state illumination module, and a scientific complementary metal oxide semiconductor (sCMOS) camera (PCO AG, Kelheim, Germany). A 100× phase-contrast objective (numerical aperture 1.4, oil immersion, DV) was used for image capture in combination with SoftWorX 3.6.0 software (Applied Precision, GE Healthcare) to control the microscope setup. A fluorescence filter set with excitation at 475/28 nm and emission at 525/48 nm was used to visualize GFP. Fluorescence was imaged with 0.5 to 1 sec exposure times, while the maximum transmission of the light source was maintained for all exposure settings. A standard microscope slide was prepared with a layer of solidified agarose (1.5%, wt/vol in phosphate-buffered saline), and 1 μL of a bacterial cell culture was spotted onto the agar. The sample was covered with a standard microscope coverslip for microscopic observation.

Cells were cultured overnight at 30 °C in Difco GM17 medium, transferred to fresh medium to an OD₆₀₀ of 0.05 and grown to the early log phase (OD₆₀₀ = 0.4). dCas9 protein expression was induced for 3 h with 10 ng/mL nisin, after which the cells were 10-fold serially diluted. Four microliters of solution were dropped on GM17 agar supplemented with the proper antibiotic, and the mixture was incubated at 30 °C for 24 h for observation.

All procedures were executed at 4 °C, and all solutions were treated with DEPC and subsequently sterilized. The cells were grown to the required OD₆₀₀ via induction, if necessary, and pelleted by centrifugation. The cells were rapidly frozen in liquid nitrogen. For RNA isolation, the samples were resuspended in 400 μL of TE buffer (10 mM Tris-HCL, 1 mM EDTA, pH 7.4). SDS (to 1%), 0.5 g of glass beads (75–150 µm; Thermo Fischer Scientific, Rockford, IL, USA) and 500 μL of phenol/chloroform (1:1 v/v) were added, after which the cells were disrupted by shaking 2 times for 60 s, with a 1 min interval on ice, in a Biospec Mini-BeadBeater (Biospec Products, Bartlesville, OK, USA). The cell suspension was centrifuged at 14,000 rpm for 15 min, and the upper liquid containing nucleic acids was removed and treated with 250 μL of chloroform. The nucleic acids in the water phase were obtained after centrifugation at 400 μL and were precipitated by adding sodium acetate to 0.3 M and 2.5 volumes of 100% ethanol and incubating overnight at −80 °C. The nucleic acid pellet was resuspended in 100 μL DNase digestion buffer with 85 μL DEPC-Milli-Q water, 10 μL 10× DNase I buffer, and 5 μL DNase I (Roche Diagnostics GmbH, Mannheim, Germany) and incubated for 1 h at 37 °C. RNA was then purified using standard phenol/chloroform extraction and sodium acetate/ethanol precipitation. The RNA pellets were resuspended in 100 μL of DEPC-treated MiliQ water. The RNA concentration was measured with a Nanodrop ND-1000 (Thermo Fisher Scientific, Waltham, MA, USA). The integrity of the 16S/23S rRNA and the presence of any DNA contamination were assessed by both gel electrophoresis and analysis on an Agilent 2100 Bioanalyzer (Agilent Technologies, Waldbronn, Germany). The RNA samples were stored at −80 °C.

cDNAs were synthesized using the SuperScript™ III One-Step RT-PCR System and random hexamer primers (Thermo Fisher Scientific) in 20 μL reaction volumes. RT-PCR was performed using SYBR II Green Supermix and an iCycler Thermal Cycler with an iQ5 Multicolor Real-Time PCR Detection System (all three from Bio-Rad Laboratories). All the samples were analyzed in triplicate. Since ybeY might be involved in 16S maturation in L. lactis, the gyrA gene was used as an internal control instead of the 16S gene [56]. All the experiments were performed twice with 3 replicates.

All RNA preparations for sequencing were made from 2 biological replicates. rRNA depletion and cDNA library preparation were performed using RiboCop rRNA Depletion and total RNA-Seq Library Prep Kits (Lexogen GmbH, Wien, Austria). cDNA libraries were sequenced using Illumina Next-Generation Sequencing technology. The raw sequencing reads were mapped to the L. lactis MG1363 reference genome (accession: NC_009004.1↗). Read alignment was performed on Bowtie [57]. The reads per kilobase of transcript and per million mapped reads (RPKM) values were used as inputs for the T-REx2 analysis pipeline [58] and the FUNctional Analysis and Gene Set Enrichment for Prokaryotes online tool (FUNAGE-Pro) [42] for global statistical analyses. As a large number of genes were widely affected, DEGs were visualized according to the following criterion: >3-fold change in expression (|Log2FC| ≥ 1.5) was considered to indicate statistical significance. The top hit genes were weighted-selected from gene set enrichment analysis. sRNAs with a 1.5-fold change in expression and statistical significance were selected (p-value < 0.01).