Prophage protein RacR activates lysozyme LysN, causing the growth defect of E. coli JM83

Prophage enriched the prokaryotic genome, and their transcriptional factors improved the protein expression network of the host. In this study, we uncovered a new prophage-prophage interaction in E. coli JM83. The Rac prophage protein RacR (GenBank accession no. AVI55875.1) directly activated the transcription of φ80dlacZΔM15 prophage lysozyme encoding gene 19 (GenBank accession no. ACB02445.1, renamed it lysN, lysozyme nineteen), resulting in the growth defect of JM83. This phenomenon also occurred in DH5α, but not in BL21(DE3) and MG1655 due to the genotype differences. However, deletion of lysN could not completely rescued JM83 from the growth arrest, indicating that RacR may regulate other related targets. In addition, passivation of RacR regulation was found in the late period of growth of JM83, and it was transmissible to daughter cells. Altogether, our study revealed part of RacR regulatory network, which suggested some advanced genetic strategies in bacteria.

annotated as a lysozyme gene, which is highly probable to participate in the process of cells lysis. Subsequently, LysN was overexpressed via pBAD-lysN (plysN) in JM83. The growth curve showed that the turbidity of culture decreased significantly, while the number of living cells only decreased slightly (Fig. 3a,b). In addition, LysN can also lead to lysis of the other four strains (Fig. 3a), suggesting that LysN may be indeed the direct functional protein associated with lysis. As expected, the fractal pattern due to LysN overexpression observed in JM83, DH10B, MG1655 cells was quite similar to pracR/JM83 overexpression case (Figs 1c and 3c). Combined together, we hypothesize that LysN is one of the direct effectors in the RacR regulatory loop to damage the cells.
inactivation of lysN rescues JM83 from lysis. To further confirm whether lysN is under control of RacR and accordingly lead to cell lysis, the lysN mutant strain was tested, of which the region from +91 to +201 was replaced by a linker plus 3 × Flag-tag (111 bp) and ended with the stop codon TGA (Fig. 4a). Compared to JM83, cell density of lysN mutant strain (ΔlysN) stayed at a medium level (Fig. 4b), and SEM observation showed that cell debris almost disappeared (Fig. 4c). Taken together, these results showed that inactivation of lysN rescues JM83 from lysis. Meanwhile, when introducing the lysN expression cassette into pracR, as expected, cells would lyse, which indicated that RacR indeed regulates lysN in vivo ( Fig. 4b and Supplementary Fig. S1).
We also noticed the cell elongation phenomenon by RacR overexpression in wild type, ΔlysN strains via SEM, no matter with or without lysN expression (Figs 1c and 4c). This suggests that there may also exist targets other than lysN regulated by RacR, which is responsible for cell elongation instead of cell lysis. While although the elongated cells overexpressing RacR didn't show significant culture turbidity decrease (purple line in Fig. 4b) or cell debris (Fig. 4c) in the first 2-5 hours, the living cell number was decreased significantly (purple bars in Fig. 4d). The decrease of living cells overexpressing RacR was even more serious than lysN ovexpressing cells in the first 3 hours but ultimately became close after 5 hours. This may imply that the other regulatory pathway other than lysN by RacR is also quite important for cell survival.
RacR triggers transcription of the lysN directly. We then measured the transcription level of lysN in pracR/JM83. The accumulation of RacR led to a more than 4000-fold increase in lysN mRNA level, which is almost silent without RacR induced expression (Fig. 5a). It suggests that RacR has the ability to trigger the transcription of lysN. Then, electrophoretic mobility shift assay (EMSA) was performed with purified His 6 -RacR and the DNA probe (designated lysN * : from −235 to +7 relative to start codon of lysN). The top of Fig. 4a showed the potential binding motifs in lysN * that derived from half of the palindromic sequence 5′-GCCTAA-3′ and 5′-TTAGGC-3′ 27 . As shown in Fig. 5b, His 6 -RacR was observed to bind to lysN * probe in a concentration-dependent manner. This interaction was nearly completely blocked by addition of 150-fold unlabeled lysN * , while not by addition of the 150-fold unlabeled unspecific DNA.    Furthermore, the upstream region and 90 bp sequence at the 5′ end of lysN was fused to the reporter gene lacZ, and the mutations were introduced into the potential binding motifs, as shown in Supplementary Fig. S2. The result showed the activity of controls were extremely low, while overexpression of RacR led to a 140-fold increase (Fig. 5c), which was highly consistent with the transcriptional levels of lysN in the genome (Fig. 5a). In addition, base substitution mutation significantly decreased the activity of LacZ, although not completely inhibited ( Supplementary Fig. S2). These results demonstrated that RacR activates transcription of lysN by directly binding to the promoter region. JM83 restored after 18 h of induction due to passivation of regulation. However, we were soon puzzled by the subsequent behavior of pracR/JM83, since the cell density of culture would begin to recover after 8 h of induction (Fig. 6a). What's more, replenishment of strong inducer at 9 h point would not interfere with the original growth trend (Fig. 6a), indicated that the recovery of growth was not due to insufficiently induced expression of RacR. We then checked the morphology of pracR/JM83, plysN/JM83, and pracR/ΔlysN after 18 h of induction, as we expected, Fig. 6b confirmed the recovery of all these strains.
To answer this question, we performed qRT-PCR. We found that the mRNA level of lysN decreased significantly at 18 h after induction (Fig. 7a). Hence, we inferred that there is a strong connection between lysN mRNA level and cell status. We assumed the possibility that the decline of RacR or the passivation of its regulation causes a lower transcriptional level of lysN, and detected cellular RacR protein level of each time points. As shown in Fig. 7a, the His 6 -RacR protein level did not change by culture, which excluded the possibility of cascaded decrease of RacR and LysN. Even so, we decided to introduce another plasmid in those recovered cells to produce RacR, in case there are any undetectable changes in pracR. We selected monoclonal recovered pracR/JM83 and renamed it JM83-Anti (resist the overexpression of RacR, Anti for short). After checked the availability in JM83  www.nature.com/scientificreports www.nature.com/scientificreports/ (Fig. 7b), pCA24N-racR was selected to produce pCA24N-racR/Anti (contain pracR and pCA24N-racR). We did not observe a obvious difference in the expression levels of RacR between JM83 and Anti strain, however, the latter cell growth was no longer inhibited (Fig. 7b). This result strengthened the hypothesis of passivation of RacR regulatory, and suggested this negative effect is permanent. Indeed, continuous cells culture imply the heritable peculiarity of passivation, since all daughter cells grew normally in the presence of L-Arabinose (Fig. 7c, G2 to G5). To sum up, although this particular third factor was not identified, we have uncovered the mechanism of cell growth recovery. In the following research, we will explore more deeply about the passivation of RacR regulatory.

Discussion
RacR was previously predicted as a transcriptional repressor that belongs to MerR superfamily, whose inhibitory effect on Rac prophage toxin YdaS has been demonstrated recently 27 . In our study, RacR overexpression triggered ϕ80 lysozyme LysN and resulted in cell lysis. This kind of relationship is highly similar to the effect of RerR on the toxin genes in Clostridium difficile 28 . The phage ΦCD119 regulator RerR has been shown inhibit to the distant toxin genes tcdA and tcdR by directly binding of their promoter, and the PaLoc (pathogenicity locus) of these toxin genes is commonly thought to belong to mobile genetic elements 29 .
The activation and inhibition to LysN and YdaS mean that the host must provide a tight mechanism to adjust the concentration of RacR. In fact, the irregular palindrome in the racR-ydaS intergenic region raised the possibility that RacR might negatively regulate its own transcription ( Supplementary Fig. S2). This assumption is consistent with the characteristic of regulators of MerR superfamily, which bind to palindrome sequences and were reported to regulate their own expression 30 . Another characteristic of these proteins is the N-terminal located HTH domain and irregular C-terminal domain, the latter is commonly used to bind metal ions, such as Hg 2+ for Tn501 protein MerR 31 . The lysN transcription decreased significantly at 18 h, even reached the same level as the wild-type JM83, while the RacR was constant (Figs 5a and 7a). We propose a hypothesis that the N-terminal of RacR binding the promotor of lysN at the early stage, with the assistance of an unknown metal ion. While at the late stage, the concentration of these ions decreased and the free RacR no longer activates the expression of lysN. However, only three imperfect conserved motifs were found in the upstream region of lysN, and their base composition and spatial arrangement neither can be compared to the regulatory sequence of ydaS. More than that, obvious enzyme activity was detected when lacZ under the control of a mutant lysN promoter ( Supplementary  Fig. S2), which suggested that these "CCTA" containing motifs were not the key sequences for RacR regulation, and the promoter region of lysN contain some more important motifs.
Although we have uncovered the causes of RacR overexpression leading to JM83 lysis, the growth of ΔlysN strain cannot recover completely (Fig. 4b), indicated that RacR may also influence other genes. In the early stages of our research, the famous phage regulators CI, CII, and Cro were also considered as potential targets for RacR. We detected their mRNA in ΔlysN after 2 h and 18 h of RacR overexpression, as shown in Supplementary  Fig. S3. It seems that RacR tried to break the balance of these three regulators in E. coli to establish a phage lytic state 32 , since the concentration of primary repressor CI was reduced and its negative regulator Cro was increased. However, we did not find active phages in our solid or liquid medium, and the cultures containing broken JM83 cells were not infectious. We speculate that the defects of ϕ80dlacZΔM15 prophage prevent the assembly and packaging of ϕ80, but whether CII or Cro plays a role in lysN expression and growth defect of ΔlysN is unknown. On the other hand, the elongation of cells is probably related to division genes, as the proteins that constitute the divisome 33 , the ZapA-ZapB complex 34 , and the Tol-Pal system 35 . It has been reported that the interference with cell division leads to an elongation phenotype in E. coli 36,37 , which is extremely similar to the cell morphology of RacR overexpressing JM83 (Figs 1c and 4c).
In summary, the regulation of RacR to lysN is special, since they belong to two different prophages. While in the typical phage lytic cycle, the S holin and R transglycosylase are under the strictly controlled of their own major phage regulators, which activated in the late stage to release the phage 38 . Although the physiological significances for lysozyme activated by foreign regulator still unclear, the model of the cell response to LysN and eventual recovery (Fig. 6a) reveals a diversity of bacterial genetic strategies. We attempted to find clues of the temporal expression of LysN, by introducing a Flag-tag in JM83 genome (Fig. 4a). However, the target protein (LysN_30N-Flag,   Figure 7. The regulation of RacR is invalid in all daughter cells. (a,b) qRT-PCR analysis of pracR/JM83 and growth curves of JM83 and Anti after induction. Western blotting indicates the His 6 -RacR level in each strain, and the full-length blot is presented in Supplementary Fig. S4. The Anti strain was depicted in this study. (c) The continuous cell culture of pracR/JM83. The whole process contains five generations (G1, G2, G3, G4, and G5). Data represent means ± standard deviations of results from three independent experiments. (2019) 9:12537 | https://doi.org/10.1038/s41598-019-48690-4 www.nature.com/scientificreports www.nature.com/scientificreports/ 8.2 kDa) was not detected successfully, and we supposed that the short half-life period of a small artificial protein affects its detection, since the mRNA of LysN_30N-Flag was transcribed (Supplementary Fig. S5).
Growth curves and spotting assay. Growth was monitored by measuring the optical density (OD) at 600 nm (OD 600 ). A single colony of each strain was inoculated in LB and grown at 37 °C overnight. Then, the strains were transferred to 500 mL flasks containing 100 mL of LB medium and were cultured at 37 °C in a shaking incubator (190 rpm). 0.2% L-Arabinose and/or 1 mM IPTG was added when OD 600 reached about 0.6. We recorded the optical density of these strains at an hour intervals for 8 hours at 28 °C. Meanwhile, 100 μL bacterial suspensions after 1 hour, 3 hours and 5 hours of induction were harvested, 10-fold gradient diluted in fresh LB medium and spread on LB agar plates. The plates were incubated at 28 °C for 24 hours followed by calculating the average colony-forming units (CFU) per milliliter according to the formula [(viable count from each concentration × dilution fold × 10)/n]. Above assays were repeated in triplicate.
Scanning electron microscope (SEM). Equivalent cell densities of different E. coli strains were collected through centrifugation (2300 × g for 5 min at 4 °C) and washed three times with phosphate buffer (PBS, 0.1 M, pH 7.5). Then, the cell pellets were fixed with 2.5% glutaraldehyde at 4 °C for 5 hours. After washing three times at 4 °C, these samples were dehydrated for 10 min each in increasing concentrations of ethanol (30%, 50%, 70%, 80%, and 90% (V/V)). Subsequently, the samples were frozen at −80 °C for about 24 hours, dried with a vacuum freeze dryer, and then observed with S-3400N scanning electron microscopy.
construction of lysN in-frame deletion mutant ∆lysN. All primers used in mutant construction are listed in Supplementary Table S2. PCR amplifications were performed to generate the upstream fragment of lysN with primer pair 19SY-1/19SY-2 and the downstream part with primer pair 19XY-1/19XY-2. Otherwise, we introduced a linker plus 3 × Flag-tag sequences which replaced the in-frame deletion region from +91 to +201 in lysN (Fig. 4a). The PCR product containing a site-directed deletion of lysN was obtained via overlap PCR with primer pair 19SY-1/19XY-2 and ligated into the NheI/XbaI site of the suicide vector pDMKE (the insB deleted derivative of pDMK 39 ). The resulting plasmid, pDMKE-lysN, were duplicated in E. coli DH5α(λpir) and then electrotransformed into E. coli JM83. Single colonies selected on LB plate with kanamycin and chloramphenicol suggest that the plasmid was integrated into the chromosome by homologous recombination. The double-crossover recombination was selected on LB plate with 10% sucrose. The lysN in-frame deletion mutant was designated as ∆lysN and confirmed via PCR and sequencing.
Quantitative real-time PCR (qRT-PCR). RNA from E. coli JM83 or ∆lysN strains frozen at −80 °C was extracted using Pure RNA Isolation Kit according to the manufacturer's protocols. For removal of the remaining DNA, total RNA was incubated with RNase-free DNase I at 28 °C for 1 hour. 1 μg total RNA was used to generate cDNA using Reverse Transcription M-MLV (RNase H-) kit. Subsequently, quantitative real-time PCR was performed according to SYBR Green PCR Master Mix and each sample was made in triplicate. rpoD acts as the internal reference gene. To normalize data, transcription levels of the rpoD gene in all samples were set to 1.0. Relative mRNA levels were analyzed using the 2 −ΔCt (ΔCt = Ct tested genes − Ct rpoD ) method. The primers for qRT-PCR are listed in Supplementary Table S2.
Overexpression and purification of RacR protein. The racR gene was PCR-amplified from E. coli JM83 and cloned into the NdeI/EcoRI site of the pET-28a (+) vector to yield pET28a-racR with an N-terminal His 6 -tag. The RacR expression plasmid was transformed into E. coli BL21(DE3). The E. coli strain was induced with 1 mM IPTG until OD 600 reached about 0.6 and grown at 16 °C for 16 hours. Then, the strain was harvested by centrifugation (5900 × g for 5 min at 4 °C) and washed three times with phosphate buffer. Pellets were resuspended to a final concentration of 10 OD/mL, sonicated on ice, and centrifuged at 5900 × g for 5 min at 4 °C. The protein was then purified via nickeliminodiacetic acid-agarose chromatography and desalinated into 1 × binding buffer (10 mM Tris-HCl [pH 8.0], 1 mM EDTA, 0.1 M NaCl, 0.1 mM dithiothreitol, 5% glycerol, and 10 μg/mL bovine serum albumin 27 ). Purified protein was analyzed by 12% SDS-PAGE, and the protein concentration was determined by the Bradford assay.
Electrophoretic mobility shift assay (EMSA). EMSAs were carried out using the purified His 6 -RacR and PCR-amplified DNA probes. The biotin-labeled probes were obtained by PCR with primer head-biotin in Supplementary Table S2, then purified and quantified. Increasing amounts of RacR were added to the 1 × binding buffer that containing target lysN * probes (5 ng) and 50 μg/mL poly(dI·C), and incubated at 28 °C for 40 min. Samples were run on a 6% polyacrylamide gel in 0.5 × TBE buffer at 130 V for 1 hour, then transferred to a nylon membrane at 380 mA for 55 min, subsequently analyzed using chemiluminescent EMSA kits.