Control of expression of the ICE R391 encoded UV-inducible cell-sensitising function

orfs90/91 stimulate orf43transcription after exposure to UV irradiation

We previously demonstrated that over-expression of orf43 when cloned into the arabinose inducible pBAD33-orf43 construct was responsible for the UV-inducible sensitisation observed in ICE R391 and other ICEs of the SXT/R391 family [8]. Mutagenesis data also suggested that the putative transcriptional controller encoded by orfs90/91 was also involved, although not directly. To investigate the relationship between orfs90/91 and orf43, we utilised both qualitative and quantitative RT-PCR targeting these genes in different mutant backgrounds and with and without UV irradiation.

qRT-PCR analysis of expression levels pre and post exposure to 40 J.m^-2 UV-irradiation indicated that orfs90/91, orf43 and the previously documented UV-inducible orf4 (jef, Figure 1) [14] were up-regulated after exposure to UV irradiation. Analysis indicated that orf4 (jef) specific mRNA levels were up-regulated 0.78 fold, orf43, 0.513 fold and orfs9091, 0.339 fold. In contrast other ICE R391 genes not involved in cell sensitisation [8] were not up-regulated post exposure: aph (encoding Kanamycin resistance) was down-regulated 0.23 fold post-exposure while orf31 (encoding a putative Lon protease) was also down-regulated 0.19 fold post-exposure. Analysis of the up-regulated genes in mutant backgrounds indicated that in a Δorfs90/91 (∆26) background, orf43 up-regulation was abolished (Figure 2) while analysis of orfs90/91 transcription in a Δorf43 (∆14) background did not prevent orfs90/91 specific mRNA up-regulation following UV irradiation (orfs90/91 up-regulated 0.61 fold in AB1157 R391 ∆14). This indicated a dependency on orfs90/91 for orf43 up-regulation but not vice versa. Further analysis of orf43 transcription in a Δorfs40/41 mutant (Δ11) [8] demonstrated that deletion of these genes, upstream of orf43, did not prevent the UV-induced up-regulation of orf43 mRNA, suggesting that inducible orf43 transcription was stimulated through a region directly in front of the orf43 gene (Figure 2) and that this region should be investigated further. This observation is supported by previous deletion analysis where orfs40/41 (Δ11) and ∆orf42 (∆13) were deleted but retained the UV-inducible sensitising phenotype [8]. Analysis of the up-regulated orfs90/91 and orf43 mRNA decay rate post-exposure (Figure 3) revealed that orfs90/91 mRNA levels were maximally up-regulated directly after exposure and decayed rapidly with a return to basal levels within 5 minutes post-exposure. However orf43 mRNA levels were maximally up-regulated 7 minutes post-exposure and up-regulated levels were sustained for a longer period of time, minimally over 30 minutes (Figure 3). The observation of the rapid increase and decay of orfs90/91 specific mRNA levels followed by a slower and longer sustained increase in orf43 specific mRNA levels supports the hypothesis that UV irradiation acts as an inducing agent for orfs90/91, which subsequently up-regulates the transcription of orf43 possibly from a site preceding the gene.

Cytotoxic orf43 transcription is regulated through a region directly upstream of orf43

Based on previous observations with the Δ11 and ∆13 ICE R391 deletions, which deleted orfs40 to most of orf42 inclusive [8], the most likely location for an orf43 control site would be the last 36 bp specific to orf42 directly in front of orf43. Comparative bioinformatic analysis of this region and the previously documented orfs90/91 regulated orf4 (jef) [14] uncovered a short 7 bp homologous DNA sequence (5’-AGAAGAT-3’) present in front of both genes. This conserved sequence was located 77 bp upstream of orf4 (jef) but directly in front of orf43 where the last 2 base pairs of the sequence overlapped the first two base pairs of the predicted start codon of orf43. As no other recognisable promoter or operator region was predicted upstream of orf43, this 7 bp sequence may possibly represent a binding motif for the putative transcriptional enhancer (orfs90/91). However it is well known that transcriptional enhancer control sites can be difficult to predict [18] as they tend to be short DNA sequences lacking high sequence conservation even between enhancer types. To examine if the last 36 bp specific to orf42 and preceding orf43 did in fact contain a control site for orf43 transcription, orf43 specific mRNA expression was analysed in a number of specific deletion backgrounds spanning this putative control region [Table 1, Figure 4C]. Three directed ICE R391 deletion mutants were generated [Figure 4C] in an E. coli (AB1157 R391) background; the KOA deletion removed the genes orf32 to orf42 and placed the inserted ampicillin cassette on the reverse complement to ensure removal of all possible promoters of orf43 transcription except for the 36 bp directly in front of orf43, the KOB deletion removed the genes orf32 to orf42 similar to KOA but additionally removed the 36 bp directly in front of orf43 while the KOC deletion was identical to the KOA mutation, preserving the putative 36 bp control site but also contained an additional secondary zeocin resistant deletion which removed orfs90/91. These three deletion mutations were screened by both qualitative and quantitative UV survival assays to determine their effect on the cell-sensitising function [Figure 4A] and additionally were examined by RT-PCR to determine if orf43 specific mRNA transcription still occurred [Figure 4B]. The KOA mutant retained the UV-inducible sensitising function [Figure 4A] and orf43 mRNA transcription [Figure 4B], while the KOB and KOC mutations abolished the sensitising function as well as orf43 mRNA transcription. As the KOB and KOC mutations deleted either the proposed orfs90/91 control site or orfs90/91 genes respectively and both of these deletions prevented cytotoxic orf43 transcription, this analysis supported the hypothesis that orfs90/91 encode a transcriptional enhancer complex that activates cytotoxic orf43 transcription through the 36 bp sequence directly in front of orf43. In addition, cloning of orf43 with the predicted control site in front of the gene showed that the cytotoxic function could be repressed only in cells not containing orfs90/91 (data not shown), again supporting the hypothesis.

Strain	Genotype	Source
AB1157	F-, thr-1, araC14, leuB6, ∆(gpt-proA)62, lacY1, tsx-33, qsr’-0, glnV44, galK2, λ-, Rac-0, hisG4, rfbC1, mgl-51, rpoS396, rpsL31 (StrR), kdgK51, xylA5, mtl-1, argE3, thi-1	E. coli genetic stock centre (CGSC), Yale University, New Haven, Connecticut, USA
TOP10	F-, mcrA0, ∆(mrr-hsdRMS-mcrBC), φ80dlacZ58(M15), ∆lacX74, recA1, araD139, ∆(araA-leu)7697, galU-, galK0, rpsL - (StrR), endA1, nupG -	Bio-Sciences, Dun Laoghaire, Dublin, Ireland
P125109	S. Enteritidis PT4 wild type (NCTC 13349), NalR	National Collection of Type Cultures (NCTC), Salisbury, UK
Plasmid	Genotype	Source
pBAD33-orf43	CmR, p15A ori, PBAD L-arabinose inducible, orf43	Armshaw and Pembroke, 2013 [8]
pBAD33-orf43[SM12]	CmR, p15A ori, PBAD L-arabinose inducible, orf43 containing mutation converting two leucines to prolines at a.a. position 47 and 48.	This study
pBAD33-orf43[SM56]	CmR, p15A ori, PBAD L-arabinose inducible, orf43 containing mutation converting glutamine at position 115 to asparagine.	This study
pKOBEG	Ts, PBAD-gam-bet-exo cat (CmR)	Dr. P. Latour-Lambert, Institut Pasteur, 25 rue du Dr Roux, Paris, France
pUC18	AmR template for deletion mutant construction	Sigma-Aldrich, Arklow, Wicklow, Ireland
pcDNA3.1(+)	ZeR template for deletion mutant construction	Invitrogen, Bio-Sciences, Dun Laoghaire, Dublin, Ireland
ICE	Genotype	Source
R391	KmR, HgR	Dr R.W. Hedges, Royal Postgraduate Medical School, London, UK
R391 Mutant	Genotype	Source
AB1157 R391 ∆14 (∆orf43)	ICE R391 orf43 deletion strain, AmR, UV-, tra-	Armshaw and Pembroke, 2013 [8]
AB1157 R391 ∆26 (∆orfs90/91)	ICE R391 orfs90/91 deletion strain, AmR, UV-, tra-	Armshaw and Pembroke, 2013 [8]
AB1157 R391 ∆11 (∆orfs40/41)	ICE R391 orfs40/41 deletion strain, AmR, tra-	Armshaw and Pembroke, 2013 [8]
AB1157 R391 ∆25AmR∆14ZeR	ICE R391 orf90 – orf94 and orf43 deletion strain, AmR, ZeR, UV-, tra-	This study
AB1157 R391 KOA	ICE R391 orf32 - orf42 (29575 bp – 41491 bp) deletion strain, AmR, tra-	This study
AB1157 R391 KOB	ICE R391 orf32 - orf42 (29575 bp – 41527 bp) deletion strain, AmR, UV-, tra-	This study
AB1157 R391 KOC	ICE R391 orf32 - orf42 (29575 bp – 41491 bp) and orfs90/91 deletion strain, AmR, ZeR, UV-, tra-	This study

Str^R is streptomycin resistant; Cm^R is chloramphenicol resistant; Km^R is kanamycin resistant; Hg^R is mercury resistant; Ze^R is zeocin resistant; T_s is temperature sensitive; Nal^R is nalidixic acid resistant and Am^R is ampicillin resistant. UV^- is UV-inducible cell-sensitising function abolished; tra^- is conjugative transfer deficient mutant.

Site-directed mutagenesis of Orf43

Bioinformatic analysis of orf43 indicated that it belongs to a highly conserved TraV-like family of transfer proteins involved in type IV secretion systems required for conjugation [8]. Site-directed mutagenesis of pBAD33-orf43 was carried out to convert two leucines at a.a. positions 47 and 48 to prolines in the predicted Orf43 protein (GenBank: AAM08037). Insertion of two prolines was expected to disrupt the α-helical transmembrane spanning region of Orf43 by creating a 30° bend [19]. This mutation was found to cause loss of the cytotoxic function of pBAD33-orf43[8] as there was no observable decline in host cell growth rates after induction of the mutant clone compared to the wild type clone [Figure 5A,B]. Since introduction of membrane disruptive mutations abolish the effect, this is suggestive that membrane association is required in addition to over-expression of the Orf43 protein for sensitisation and cytotoxicity associated with this ICE product. Other directed mutations such as conversion of glutamine at position 115 (predicted in the periplasmic region of Orf43) to asparagine (Q115 to N115) was shown to have no effect on UV sensitisation or orf43 transcription [Figure 5C,D].

Hierarchical control of the ICE R391 UV-inducible sensitising effect

Many SXT/R391-like ICEs reduce post UV survival rates of E. coli host cells through the action of a recA-dependent process [6, 20]. Mutational analysis of the ICE R391 determined that the core genes orfs90/91 and orf43 were required for expression of the cell-sensitising function [8] while bioinformatic analysis indicated that orf96 likely encodes a λ cI-like repressor similar to RecA substrates in other phage systems that are cleaved following SOS induction [9]. Initial attempts to delete orf96 proved fruitless and no deletion could be isolated. However a Δorf96 (Δ28) deletion [8] could be isolated in an ∆orfs90/91 mutant background suggesting that orf96 may control expression of orfs90/91 which we have shown here directly control expression of orf43, the ultimate instigator of the cytotoxicity associated with ICE R391. The data presented here and in Armshaw and Pembroke (2013) [8] have led to the development of a model to explain the control of UV-inducible sensitisation (Figure 1). We hypothesise that UV irradiation of E. coli induces the host RecA protein which results in cleavage of the ICE R391 encoded product of orf96, the phage λ⁴³⁴ cI-like ICE repressor. We propose that cleavage of Orf96 in turn leads to expression of orfs90/91 which in turn leads to up-regulation of orf43 and other ICE R391 genes such as orf4 (jef) [14]. We have previously demonstrated that up-regulation of orf4 (jef) leads to increased ICE R391 transfer [14]. In the related ICE SXT, Beaber et al., (2004) [17] demonstrated that SetR, the SXT homolog of Orf96, acted as a repressor of ICE SXT transfer and that it is bound to ICE operators that controlled setC/D, SXT homologs of orfs90/91, in a similar way to our proposal for ICE R391. They also proposed that repression was lifted by induced RecA protein cleaving the SetR repressor in a similar manner to our proposal for orfs90/91. The recA dependence for the ICE R391 UV-sensitising effect [6], the similarity to the SXT system [17], the deletion data and qRT-PCR data presented here support the model presented.

It would thus appear that UV irradiation is the instigator of the control loop leading to over expression of orf43 which leads to cytotoxicity. Since normal levels of Orf43 are required for ICE transfer and play a key part in formation of the conjugative transfer system of ICE R391, it appears that the associated cytotoxicity is related to the induced overexpression. Evidence in support of this comes from data showing that overexpression of orf43 from the arabinose inducible clone pBAD33-orf43 leads directly to cytotoxicity [8]. The UV-inducible sensitising effect is conserved amongst many SXT/R391 ICE family members [6, 20]. A sophisticated control system is in place to control this effect yet the exact nature and reason for conservation of such an unusual apparently ‘evolutionary negative’ effect remains to be elucidated. We are currently examining the nature of the cytotoxicity and developing theories for its function and retention.

Bacterial strains, elements and media

The bacterial strains, plasmids and ICE R391 deletion mutants utilised as part of this study are listed in Table 1. Strains were stored at −80°C in either Luria-Bertani (LB) broth or M9 minimal media containing 50% (v/v) glycerol. Media was supplemented with appropriate antimicrobial agents: nalidixic acid, 30 μg ml^-1; ampicillin, 100 μg ml^-1; chloramphenicol, 25 μg ml^-1, kanamycin, 30 μg ml^-1, streptomycin, 100 μg ml^-1; mercuric chloride, 20 μg ml^-1; zeocin, 25 μg ml^-1 as required. For growth and analysis of strains containing pBAD33-orf43, M9 minimal media containing 0.4% (v/v) glycerol was used with either 0.4% (w/v) glucose or 0.02%-0.2% (w/v) L-arabinose to repress or induce gene expression respectively as previously described [8].

Directed deletions of ICE R391 and subsequent deletion mutant screening

ICE R391 specific deletions were generated as previously described [8]. Screening of resulting ICE R391 deletion mutants for loss of cell-sensitising function by qualitative and quantitative UV survival assays were carried out as described [8]. Screening of ICE R391 deletion mutants’ conjugative transfer ability to recipient Salmonella enterica serotype Enteritidis strain P125109 was performed as described [21].

Qualitative reverse transcriptase PCR

Cells were collected by centrifugation, washed twice with diethyl pyrocarbonate-treated distilled water and resuspended in 10 mM Tris, [pH8.0]. Total RNA was isolated using the Absolutely RNA Miniprep kit (Agilent Technologies) according to the manufacturer’s protocol. Absence of contaminating DNA was verified by PCR. Qualitative reverse transcriptase PCR was performed using the AccuScript High Fidelity 1^st Strand cDNA Synthesis Kit (Agilent Technologies) according to the manufacturer’s protocol. Resulting cDNA was analysed immediately by PCR using gene-specific primers or stored at −20°C.

Quantitative reverse transcriptase PCR (qRT-PCR)

Quantitative UV assays were carried out as described [8]. Unirradiated and irradiated cells were collected by centrifugation and total RNA isolated as described. Absence of contaminating DNA was verified by PCR. qRT-PCR was performed using the Brilliant III Ultra-Fast SYBR Green qRT-PCR Master Mix (Agilent Technologies) according to the manufacturer’s protocol using the Stratagene Mx3000P Real Time PCR System and appropriate gene-specific PCR primers with the following temperature profile: 1 cycle at 42°C for 30 minutes to convert RNA to cDNA, 1 denaturation cycle at 95°C for 3 minutes followed by 40 cycles at 95°C for 30 seconds, 54°C for 60 seconds, 72°C for 30 seconds followed by melting curve analysis from 65°C to 95°C to determine specificity of the PCR reaction. Specificity of the PCR reaction was verified by SYBR safe staining on a 2% (w/v) agarose gel. The internal standard curve using the unirradiated RNA sample to estimate the change in target RNA quantity consisted of: undiluted RNA, a 1 in 2 dilution, a 1 in 4 dilution and a 1 in 10 dilution of unirradiated RNA. A no template negative control was also included. In addition, qRT-PCR was also carried out on the known endogenous housekeeping gene proC as an internal control to quantify the relative change in transcription of the gene of interest [22].

Site-directed mutagenesis of pBAD33-orf43

Site-directed mutagenesis of pBAD33-orf43[8] was performed using specifically designed complementary mutagenic primers to linearly amplify pBAD33-orf43 to generate a mutated nicked DNA product. Non-mutated methylated template DNA was eliminated by incubation with the DpnI restriction enzyme. Mutated DNA products were then transformed into TOP10 and plated on appropriate media containing chloramphenicol, 25 μg ml^-1. Resulting TOP10 colonies were cultured, had plasmid content extracted using the QIAprep Spin Miniprep Plasmid extraction kit from QIAGEN (West Sussex, RH10, 9NQ, UK) according to the manufacturer’s protocol and screened for the presence of pBAD33-orf43 by restriction enzyme digestion. Mutated pBAD33-orf43 was verified by DNA sequencing to contain the desired mutation without additional mutations. Mutated pBAD33-orf43 was confirmed to still transcribe orf43 specific mRNA by RT-PCR as described. Determination of the effect of induction of mutated pBAD33-orf43 on host cell growth rate was carried out as described [8].