Heat shock transcription factor σ 32 defective in membrane transport can be suppressed by transposon insertion into genes encoding a restriction enzyme subunit or a putative autotransporter in Escherichia coli

Edited by Hironori Niki * Corresponding author. E-mail: yura.takashi.62X@st.kyoto-u.ac.jp † Corresponding author. E-mail: yakiyama@infront.kyoto-u.ac.jp ‡ These authors contributed equally to this work. DOI: http://doi.org/10.1266/ggs.18-00040 Heat shock transcription factor σ defective in membrane transport can be suppressed by transposon insertion into genes encoding a restriction enzyme subunit or a putative autotransporter in Escherichia coli


INTRODUCTION
The heat-shock response (HSR) is a major cellular response to protein unfolding, aggregation and damage, and involves rapid and transient induction of heat shock proteins (HSPs) in all organisms (Morimoto, 2011).Although much attention has been paid to understanding the role of well conserved HSPs, such as the DnaKJ (HSP70/40) and GroELS (HSP60/10) chaperones that remove misfolded or abnormal proteins in Escherichia coli and other organisms, the question of how this all takes place in an organized and dynamic fashion remains largely unsolved (Guisbert et al., 2004;Yura et al., 2007).
Regulation of the HSR is complex even in E. coli, and is accomplished by controlling the synthesis, activity and stability of the key transcription factor σ 32 , which is quite unstable and normally found in very small amounts (Straus et al., 1987;Yura et al., 2000).When cells are suddenly exposed to high temperature, the abundance of σ 32 is rapidly and transiently enhanced, both by increased translation following disruption of the secondary structure of rpoH mRNA (Morita et al., 1999a(Morita et al., , 1999b) ) and by transient stabilization of σ 32 (Gamer et al., 1992(Gamer et al., , 1996;;Liberek et al., 1992;Guisbert et al., 2004).Degradation of σ 32 depends primarily on the membrane-localized protease FtsH (Herman et al., 1995;Tomoyasu et al., 1995;Ito and Akiyama, 2005).Mutations in the conserved N-terminal control region of σ 32 affect both the activity and stability of σ 32 (Horikoshi et al., 2004;Obrist and Narberhaus, 2005), suggesting its involvement in the FtsH-mediated degradation.Extensive work on these deregulated mutants showed that they are defective in chaperone-mediated feedback control in vivo, but the defects could not be recapitulated in vitro (Yura et al., 2007), which eventually led to the finding that σ 32 must be localized to the inner membrane by a non-canonical SRP pathway to be properly regulated (Lim et al., 2013;Miyazaki et al., 2016).Although σ 32 does not contain a canonical membraneanchor sequence found in membrane proteins, the N-terminal control region of σ 32 binds directly to Ffh (the protein moiety of the SRP), which, together with the SecYEG translocon, brings σ 32 to the membrane (Lim et al., 2013;Miyazaki et al., 2016).Thus, we envisage that the membrane targeting of σ 32 is an important step for regulating the HSR, in which some additional factors could contribute to protein homeostasis in the membrane and/or in the cytoplasm.
We have now carried out random transposon mutagenesis of the E. coli chromosome to look for Tn5 insertions that could suppress the strongly deregulated phenotype of the I54N-σ 32 mutant, characterized previously (Yura et al., 2007;Lim et al., 2013;Miyazaki et al., 2016).Here, we characterized transposon insertion mutations in mcrC, encoding the DNA cleavage catalytic subunit of a type IV restriction enzyme, and in ydbA2, encoding part of a putative autotransporter.
Growth conditions and determination of σ 32 activity Cultures were grown in LB broth (Difco) with appropriate antibiotics and other supplements at 30 °C with or without shaking.σ 32 activity was determined by measuring β-galactosidase activity driven by the heat-shock promoter (PhtpG-lacZ), as described previously (Yura et al., 2007).
Construction of a plasmid carrying mcrC::Tn5 Multicopy plasmid pUC18 carrying mcrC::Tn5 under the control of the Plac promoter and lacI q repressor was constructed by Gibson assembly (Gibson, 2011) of several PCR products that include the mcrC::Tn5 region of the suppressor mutant DNA, the lac promoter and the lacI q repressor gene.The plasmid thus constructed was verified by digestion with restriction enzymes and sequence determination.This plasmid should allow regulated expression of the N-terminal segment of the mcrC gene interrupted with Tn5, which was identical to the chromosomal mcrC::Tn5 obtained in this study.

RESULTS
Transposon insertion into chromosomal genes can suppress the hyperactive mutant I54N-σ 32 and reduce its activity Escherichia coli cells producing hyperactive and stable I54N-σ 32 mutant protein defective in SRP-dependent membrane transport (Lim et al., 2013;Miyazaki et al., 2016) were electroporated with the Tn5-DNA-transposome complex, and plated on LB-lactosetetrazolium agar medium containing kanamycin to look for transposon insertions that reduced the σ 32 activity at 30 °C.The lacZ reporter (PhtpG-lacZ) and the indicator dye provided a sensitive means to detect these suppressor mutations (see Materials and Methods).Among the kanamycin-resistant transformants thus obtained, we picked 10 putative suppressor mutants (red colonies), purified them by single-colony isolation, and extracted the chromosomal DNA for PCR amplification and sequencing of the transposon-adjacent region.
Two sites of Tn5 insertion were identified (Fig. 1).One contained Tn5 within the mcrC gene encoding the DNA cleavage catalytic subunit of a type IV restriction enzyme (Pieper and Pingoud, 2002), and was expected to produce a truncated form of the McrC protein (McrC::Tn5) of 85 amino acids followed by 24 extra amino acid residues derived from the transposon sequence.The other site contained Tn5 inserted within the ydbA2 gene, which encodes part of a putative autotransporter, resulting in the production of a truncated form of the YdbA2 protein (YdbA2::Tn5) of 192 amino acids followed by 18 extra amino acid residues.Although ydbA2 has not been characterized, published ribosome profiling data indicate that it is indeed translated (Li et al., 2012;Michel et al., 2014).
To confirm that these insertions are directly responsible for the observed reduction in σ 32 activity, each of the Tn5 insertions was transduced into the parental strain by phage P1, selecting for kanamycin resistance at 30 °C.All the transductants obtained showed reduced σ 32 activity characteristic of the donor suppressor mutant used, indicating that the reduced σ 32 activity directly resulted from the Tn5 insertion rather than some secondary mutation.The results are summarized in Table 1 (upper panel).When these insertions were transduced into rpoH + cells producing wild-type σ 32 , the transductants obtained showed slightly lower σ 32 activities than the parent, but the extents of decrease were much lower than those with I54N-σ 32 mutant cells (Table 1, lower panel).If inactivation of the target gene by the transposon insertion was responsible for the suppression of the deregulated mutant σ 32 , deletion of the gene should also suppress the I54N-σ 32 phenotype.To test this point, we constructed rpoH(I54N) ΔmcrC and rpoH(I54N) ΔydbA2 double mutant strains.However, the σ 32 activity did not change appreciably from the parental rpoH(I54N) single mutant (Table 2), indicating that the suppressor phenotype is not due to loss-of-function of the McrC or YdbA2 protein.Moreover, the suppression of the deregulated I54N-σ 32 mcrC::Tn5 or ydbA2::Tn5 could not be complemented by plasmids expressing the wild-type mcrC + or ydbA2 + gene, respectively (data not shown).These results indicate that the suppression by the Tn5 insertion mutations cannot be ascribed to inactivation of the target mcrC or ydbA2 gene.
To determine whether the extent of chaperone-   ) to log phase at 30 °C, and σ 32 activity was determined by measuring β-galactosidase activity for each of the suppressor mutants.The enzymatic activities were compared with the wild-type control, which was set as 1.0, with means ± standard deviation calculated from more than three independent experiments.wt, wild-type.ND, not done.
mediated feedback inhibition was altered by the suppressor mutations, we examined the effects of overproduction of DnaK-DnaJ chaperones on σ 32 activity by adding a high concentration (1 mM) of IPTG to overproduce DnaK-DnaJ chaperones.No further reduction in the σ 32 activity was observed with any of the suppressor mutants tested.The deregulated σ 32 activity of the parental I54N-σ 32 strain was also unaffected by the chaperone overproduction.Thus, the suppressors do not appear to affect the response of I54N-σ 32 to the chaperone-mediated negative control of the HSR (Table 1, upper panel).The possibility that the suppressors altered the response of I54N-σ 32 to chaperones other than DnaKJ could not be excluded, however.
Tn5 insertion into the mcrC or ydbA2 gene suppresses I54N-σ 32 specifically To examine whether the mcrC::Tn5 or ydbA2::Tn5 insertion can affect σ 32 mutants other than I54N-σ 32 , each of the above Tn5 insertions was transduced into a set of isogenic strains producing various σ 32 mutant proteins (I54N, A50D or K51E), and the resulting transductants were compared to determine possible differential effects of the suppressors among different σ 32 mutants.Whereas the A50D mutant exhibits strong resistance to chaperone-mediated inhibition, comparable to I54N, the K51E mutant is only moderately resistant (Yura et al., 2007).All these mutants are significantly altered in their binding to Ffh, the protein moiety of the SRP that is required for their transport to the membrane (Miyazaki et al., 2016).As expected, marked reduction of σ 32 activity was observed with I54N-σ 32 by mcrC::Tn5 or ydbA2::Tn5 insertion, whereas little if any effect was found with the other σ 32 mutants (Table 3).Thus, only cells producing I54N-σ 32 were strongly affected by mcrC::Tn5 or ydbA2::Tn5 insertion, clearly indicating the I54N-allele specificity of the suppression.These results exclude a trivial explanation of the apparent suppression, such as inhibition of β-galactosidase activity by the Tn5truncated product.
Suppressor insertions (mcrC::Tn5 and ydbA2::Tn5) do not affect the cellular abundance of I54N-σ 32 To see whether the mcrC::Tn5 or ydbA2::Tn5 suppressor insertion affects the amount or stability of I54N-σ 32 , we compared the cellular amounts of I54N-σ 32 between strains with or without Tn5 insertions by SDS-PAGE analysis of crude cell extracts, followed by immunoblotting using antiserum against σ 32 .Despite the reduced σ 32 activity observed with the suppressor mutants, neither suppressor mutation affected the amount of σ 32 to any significant extent (Fig. 2A and 2C).Thus, the suppressor mutations lead to the reduced activity of σ 32 without affecting its accumulation levels.It is conceivable that the McrC or YdbA2 polypeptide fragments produced by the Tn5 insertion mutations somehow interfere with the activity of I54N-σ 32 , possibly by binding to it.It is noteworthy that the polypeptides do not significantly destabilize the hyperactive and stable I54N-σ 32 (Fig. 2A  and 2B).
A plasmid carrying the mcrC::Tn5 segment strongly reduces the σ 32 activity of cells producing I54N-σ 32 To further substantiate the effects of the truncated form of McrC protein on I54N-σ 32 , a pUC18 derivative carrying mcrC::Tn5 was constructed by Gibson assembly of PCR products that included the sequences for the Plac promoter, the mcrC::Tn5 suppressor version of mcrC and lacI q (see Materials and Methods).The resulting plasmid, which can express the truncated form of McrC protein from the lac promoter (inducible by IPTG), ΔydbA2::kan 0.9 ± 0.1 *The mcrC deletion or the ydbA2 deletion (Keio collection) was transduced into the parental strain producing I54N-σ 32 by selecting for kanamycin resistance.Three colonies were picked for each transduction, purified by single-colony isolation, and grown to late log phase to determine σ 32 activity as in Table 1.
*A set of strains carrying each of the feedback-resistant rpoH mutations on a multicopy plasmid and the wild-type rpoH + gene on the chromosome were used as the host to transduce each of the Tn5 insertions by phage P1 by selecting for kanamycin resistance.After purifying the transductants obtained, cells were grown and tested for σ 32 activity as in Table 1.σ 32 activity of the I54N mutant host (without suppressor mutation) was set as 1.0 for comparison.
was introduced into I54N-σ 32 mutant cells.The plasmid carrying mcrC::Tn5 inhibited the I54N-σ 32 activity very strongly in the presence of excess inducer (1 mM IPTG), and less strongly in its absence (10 μM IPTG) (Table 4).The control plasmid that lacks the mcrC gene showed no detectable effect on the I54N-σ 32 activity.Thus, the N-terminal segment of the McrC protein fused with a short Tn5-encoded amino acid sequence expressed from the plasmid reduces the activity of I54N-σ 32 very strongly.Such plasmids should provide effective means to further scrutinize the nature of the McrC-Tn5-mediated negative control of σ 32 activity.

DISCUSSION
We have isolated two distinct suppressor mutations that can specifically reduce the σ 32 activity of the deregulated/hyperactive I54N-σ 32 mutant protein, without affecting protein stability or level.The suppressor mutations have been identified as Tn5 insertions into the genes for the restriction enzyme subunit McrC and a putative autotransporter, YdbA2, of unknown function (Table 1).Although the mechanism of suppression remains unknown, loss of gene function is not responsible for the suppression (Table 2).The allele specificity for rpoH(I54N) (Table 3), the lack of complementation Accumulation and stability of wild-type and I54N-σ 32 in the presence or absence of the suppressor mutations.Cells were grown to log phase at 30 °C in LB medium with 5 μM IPTG.Spectinomycin (500 μg/ml) was added at time 0, and samples were taken at the times indicated.After treatment with 5% trichloroacetic acid, precipitates were collected and washed with acetone, and total proteins were solubilized in Laemmli sample buffer and analyzed by SDS-polyacrylamide gel electrophoresis (10% gel) essentially as described previously (Miyazaki et al., 2016).σ 32 levels were quantitated by immunoblotting with antiserum against σ 32 by the standard procedure, and typical gel patterns are presented in (A), which also shows some non-specific bands (*).(B) Stability of σ 32 , based on two experiments, is plotted against chase time, setting the time 0 values for each strain as 1.0, with standard deviations.(C) The levels of σ 32 at time 0 were calculated and compared among different strains, setting the value for the parental rpoH (I54N) strain as 1.0.4) raise the interesting possibility that the short N-terminal polypeptide fragments produced by the suppressors (Tn5 insertions) interact directly with I54N-σ 32 and reduce the latter's activity.We note that the suppressor mutations are expected to produce two polypeptides (McrC::Tn5 and YdbA2::Tn5) with a region of weak mutual similarity, encompassing 13 amino acids, based on Clustal W multiple alignment (Thompson et al., 1994).This region resides between Val-45 and Gly-57 of the McrC protein and between Val-143 and Gly-155 of YdbA2 (Fig. 1A and 1B).Whether these regions of similarity are actually involved in their interaction with I54N-σ 32 remains to be clarified.By contrast, the C-terminal sequences of the McrC::Tn5 and YdbA2::Tn5 polypeptides encoded by the transposon region are quite distinct from each other (Fig. 1A).We therefore assume that the transposon-encoded region is unlikely to contribute to the suppression.
Because the parental strain we used contained I54N-σ 32 , which had virtually lost the capacity to be transported to the membrane, the I54N-σ 32 should stay mostly in the cytoplasm and bind to RNA polymerase core enzyme to transcribe heat shock genes very strongly under growing conditions.The present results suggest that the short N-terminal fragments (109 amino acids for McrC::Tn5 and 210 amino acids for YdbA2::Tn5) of suppressor proteins specifically inactivate the I54N form of σ 32 by directly binding to it.One possibility would be that binding of the suppressor polypeptide inhibits the ability of I54N-σ 32 to associate with the core RNA polymerase.Alternatively, the suppressor polypeptides might inhibit transcription initiation by RNA polymerase containing I54N-σ 32 .In either case, the precise molecular mechanisms are left for future studies.In view of the remarkable property of σ 32 that it can be recognized non-canonically by the SRP and localized to the cytoplasmic membrane of the cell, another novel mechanism of I54N-σ 32 inactivation is conceivable.The truncated suppressor polypeptides could enhance SRP-dependent membrane localization and sequestration of I54N-σ 32 , without being presented to the FtsH protease.
The McrC and YdbA2 proteins may actually be involved in σ 32 activity regulation during the HSR and/ or under growing conditions in E. coli.However, the suppression is not due to loss of these proteins.Rather, both mcrC::Tn5 and ydbA2::Tn5 insertions may create σ 32 -interacting polypeptides that artificially interfere with the σ 32 activity or inactivate it by sequestration to the membrane.Nevertheless, it is possible that the N-terminal regions of these proteins have a regulatory function toward σ 32 and that such a function is expressed transiently during the synthesis of these proteins.Further work related to the function of the McrC::Tn5 or YdbA2::Tn5 polypeptide may lead to a unique opportunity to look into the mechanism of activity regulation of σ 32 and, hence, the control of the HSR in E. coli.In particular, the segments of sequence similarity (Fig. 1) merit further molecular analysis.Although the specific target, the I54N mutant form of σ 32 , used in this study inevitably limits the general significance of our finding, our analysis nevertheless raises the possibility that certain polypeptide fragments inhibit the wild-type σ 32 transcription factor.We could explore such polypeptides, in particular those expressed under stressful environments, as a unique antibiotic acting against bacteria (Blum et al., 2000).

Table 1 .
Effects of Tn5 insertion into the mcrC or ydbA2 gene on σ 32 activity of the hyperactive feedback-resistant I54N-σ 32 protein defective in membrane transport

Table 2 .
Effect of chromosomal mcrC or ydbA2 deletion on σ 32 activity of the strain producing I54N-σ 32

Table 3 .
Effects of Tn5 insertion into the mcrC or ydbA2 gene on the activity of various σ 32 mutant proteins

Table 4
the suppressor phenotype by plasmids carrying the mcrC + or ydbA2 + gene (data not shown), and suppression of rpoH(I54N) by the truncated form of McrC protein expressed from a plasmid in the presence of the chromosomal mcrC + gene (Table of