Transcriptional Organization of the Salmonella Typhimurium Phage P22 pid ORFan Locus

Many phage genes lack sequence similarity to any other open reading frame (ORF) in current databases. These enigmatic ORFan genes can have a tremendous impact on phage propagation and host interactions but often remain experimentally unexplored. We previously revealed a novel interaction between phage P22 and its Salmonella Typhimurium host, instigated by the ORFan gene pid (for phage P22 encoded instigator of dgo expression) and resulting in derepression of the host dgoRKAT operon. The pid gene is highly expressed in phage carrier cells that harbor a polarly located P22 episome that segregates asymmetrically among daughter cells. Here, we discovered that the pid locus is fitted with a weak promoter, has an exceptionally long 5′ untranslated region that is instructive for a secondary pid mRNA species, and has a 3′ Rho-independent termination loop that is responsible for stability of the pid transcript.


Introduction
With an estimated presence of around 10 31 virions, viruses of bacteria (i.e., bacteriophages or phages) are the most ubiquitous entities in our biosphere and have an unrivalled impact on the ecology and evolution of their hosts [1,2]. Furthermore, the many different phage species that exist constitute a tremendous sequence space that we have only just begun to explore [3]. Indeed, while sequencing efforts are massively sampling this space, it is clear that many phage genes lack homologs in the currently existing databases and have not been investigated experimentally [4][5][6]. However, these unknown phage genes (also referred to as phage ORFans) can profoundly impact phage propagation [7,8] and host interactions [9,10].
As such, the existence of many phage ORFan genes underscores that the full complexity of phage-host associations and interactions has yet to be established, even in well-studied model systems such as the temperate lambdoid phage P22 and its Salmonella Typhimurium host. Indeed, we have previously revealed the existence of a cryptic carrier state association of P22 that deviates from its strictly lytic or lysogenic proliferation [11]. We found that during the path towards lysogenisation of its host, P22 is able to form a polarly-located episome that persistently becomes segregated asymmetrically between daughter cells. In carrier cells inheriting this P22 episome, a P22 ORFan gene termed pid (for phage P22 encoded instigator of dgo expression [12]) is highly expressed. The corresponding Pid protein derepresses the host dgoRKAT operon, an operon that drives D-galactonate metabolism [12], although the impact of this Pid/dgo interaction and the regulation of the pid locus remain elusive.
The presence of the pid locus remained unnoticed for decades, until its identification as a functional gene, located between the orf25 and orf80 of P22 [12]. Although found in the late region of the P22 genome, the orientation of pid opposes the main late promoter. In this study, we report the findings of a genetic screen designed to reveal key regulatory functions of the P22 pid locus.

Genetic Screen for Functional Mutations in the P22 pid Locus
To functionally characterize the P22 pid locus, we set up a genetic screen in which a DES-mutagenized P22 population was plaqued on LB X-Gal on the LT2K7 indicator strain (i.e., LT2 dgoT::MudK; [12]) that reports on derepression of the dgo operon by producing the LacZ reporter protein. We chose plaques with an altered degree of blue color resulting from LacZ activity on the X-Gal and sequenced the pid locus of the corresponding phage. We excluded clear plaques from this screen since previous research revealed that obligate lytic variants of P22 do not enter the carrier state and do not express Pid, and therefore do not support the Pid/dgo interaction [12]. Our screen revealed (i) two clones (P22-pid C-183A and P22-pid C-182A ) that had more intense blue coloration than wild-type P22 (i.e., P22wt) and that carry mutations far upstream of the pid open reading frame (ORF), (ii) one blueattenuated clone (P22-pid ORF_C251T ) carrying a mutation within the pid ORF, and (iii) one blue-attenuated clone (P22-pid term_G292A ) carrying a mutation downstream of the pid ORF ( Figure 1).
We found that during the path towards lysogenisation of its host, P22 is able to f polarly-located episome that persistently becomes segregated asymmetrically be daughter cells. In carrier cells inheriting this P22 episome, a P22 ORFan gene term (for phage P22 encoded instigator of dgo expression [12]) is highly expressed. The sponding Pid protein derepresses the host dgoRKAT operon, an operon that driv galactonate metabolism [12], although the impact of this Pid/dgo interaction and th ulation of the pid locus remain elusive.
The presence of the pid locus remained unnoticed for decades, until its identifi as a functional gene, located between the orf25 and orf80 of P22 [12]. Although fou the late region of the P22 genome, the orientation of pid opposes the main late prom In this study, we report the findings of a genetic screen designed to reveal key regu functions of the P22 pid locus.

Genetic Screen for Functional Mutations in the P22 pid Locus
To functionally characterize the P22 pid locus, we set up a genetic screen in w DES-mutagenized P22 population was plaqued on LB X-Gal on the LT2K7 indicator (i.e., LT2 dgoT::MudK; [12]) that reports on derepression of the dgo operon by prod the LacZ reporter protein. We chose plaques with an altered degree of blue color res from LacZ activity on the X-Gal and sequenced the pid locus of the corresponding p We excluded clear plaques from this screen since previous research revealed that ob lytic variants of P22 do not enter the carrier state and do not express Pid, and therefo not support the Pid/dgo interaction [12]. Our screen revealed (i) two clones (P22-pi and P22-pid C-182A ) that had more intense blue coloration than wild-type P22 (i.e., P and that carry mutations far upstream of the pid open reading frame (ORF), (i blue-attenuated clone (P22-pid ORF_C251T ) carrying a mutation within the pid ORF, an one blue-attenuated clone (P22-pid term_G292A ) carrying a mutation downstream of t ORF ( Figure 1). Figure 1. DES-mutagenesis screen revealed functional mutations in the P22 pid locus. Selected mutagenized P22 clones plaqued on a lawn of the LT2K7 indicator (LT2 dgoT::MudK) on LB in which blue coloration reports on LacZ activity (panel (a)) and the corresponding mutation pid locus (panel (b)). Numbering of the mutations (indicated as *) is relative to the pid start co Figure 1. DES-mutagenesis screen revealed functional mutations in the P22 pid locus. Selected DES-mutagenized P22 clones plaqued on a lawn of the LT2K7 indicator (LT2 dgoT::MudK) on LB X-Gal in which blue coloration reports on LacZ activity (panel (a)) and the corresponding mutation in the pid locus (panel (b)). Numbering of the mutations (indicated as *) is relative to the pid start codon.
Of the two mutants attenuated in the Pid/dgo interaction, one contained a C-to-T mutation located 251 bp downstream of the pid start codon, leading to a T84M amino acid substitution in the C-terminal part of the Pid protein. However, since the structure-function relationship of Pid currently remains elusive, it was unclear as to how this amino acid change would compromise the Pid protein or the Pid/dgo interaction. The other mutant harbored a G-to-A mutation in the stem-loop structure downstream of the pid ORF (292 bp downstream of the pid start codon). This stem-loop (hairpin) structure is characterized by a G+C-rich stem followed by a stretch of poly(T) on the sense strand, likely corresponding to a Rho-independent termination site [13]. Such sites play an important role in the termination mechanism by regulating the release of the transcript and RNA polymerase from the DNA template strand. The structure and dimension of such factor-independent termination sites can vary, and can influence transcript termination [14]. The G-to-A mutation in the G+Crich stem structure of pid might disrupt stem-loop formation and impact upon termination functionality and mRNA stability.

Characterization of the P22 pid Promoter
Since the above-mentioned mutations within and downstream of the pid ORF are likely loss-of-function mutations that simply compromise the Pid/dgo interaction, subsequent focus was placed on the upstream mutations that appeared to boost the interaction, i.e., P22-pid C-183A (with a C-to-A mutation, located 183 bp upstream of the pid start codon) and P22-pid C-182A (with a C-to-A mutation, located 182 bp upstream of the pid start codon). Since both mutations were next to each other but far upstream of the pid start codon, the surrounding genetic context was scrutinized and predicted to be a promoter (97.1% probability score via the PhagePromoter prediction tool integrated in the Galaxy framework, [15]). Moreover, a possibly eroded Pribnow promoter box (i.e., 5 -AATCCT-3 vs. the canonical 5 -TATAAT-3 ) was identified 181-186 bp upstream of the start codon where both the Cto-A mutations were situated (p-value of 0.0014 via the SAPPHIRE 1 Promoter prediction tool, [16]).
To experimentally investigate the functionality of this putative promoter region, and of the C-182A and C-183A mutations, the fragment from 533 to 12 bp upstream of the pid start codon from P22wt, P22-pid C-182A , and P22-pid C-183A was cloned upstream of the gfp gene in the pFPV25 promoter-probe vector [17] and assayed for GFP fluorescence production ( Figure 2a). It was clear that the upstream region contained a basal promoter activity, which was boosted four-or sixfold by the C-182A or C-183A mutation, respectively ( Figure 2b). These mutations change the region into 5 -AATCAT-3 or 5 -AATACT-3 , respectively, and likely improve the functionality of the Pribnow promoter box by more closely resembling the optimal prokaryotic consensus sequence (i.e., 5 -TATAAT-3 ). Comparable studies on other genes have shown that mutations that make the Pribnow box more similar to the consensus sequence cause increased transcription of the downstream gene [17][18][19]. When the wild-type fragment cloned upstream of the gfp reporter gene was shortened from its 5 end (cf. P pid 338 and P pid 232 ; Figure 2a), promoter activity was lost when the 533 to 245 bp fragment upstream of the pid start codon was removed (i.e., P pid 232 ; Figure 2c), suggesting no other promoters were present within the 244 bp fragment immediately upstream of the pid ORF.
Finally, the plasmid data in Figure 2 also suggest that the wild-type pid promoter (P pid wt ) is active (or at least shows constitutive leaky expression), even in the absence of P22borne regulatory factors. Nevertheless, the leaky expression of pid from a P22wt lysogen (in the LT2K7 reporter) was insufficient to mount the Pid/dgo interaction ( Figure 3). In contrast, a P22-pid C-183A lysogen with upregulated P pid activity constitutively derepressed the dgo operon ( Figure 3).  P22-borne regulatory factors. Nevertheless, the leaky exp ogen (in the LT2K7 reporter) was insufficient to mount th In contrast, a P22-pid C-183A lysogen with upregulated P pressed the dgo operon ( Figure 3).

Characterization of the P22 pid Transcript
Because the Ppid promoter appeared to be far upstream script was predicted to have an extensive 5′ untranslated cent RNA-seq data from the Salmonella Typhimurium D2 naturally lysogenized with the BTP1 prophage that carrie P22, revealed that the transcriptional start site (TSS) of Ppid stream of the pid start codon. This TSS location further co tified in the DES-mutagenesis screen were located in the region. Specifically, the C-182A and C-183A mutations m of the Ppid promoter ( Figure 4). These data suggest that the nucleotide 40,880 on the genome of the P22 phag NC_002371.2).

Characterization of the P22 pid Transcript
Because the P pid promoter appeared to be far upstream of the start codon, the pid transcript was predicted to have an extensive 5 untranslated region (5 UTR). Analysis of recent RNA-seq data from the Salmonella Typhimurium D23580 strain [6,[20][21][22], which is naturally lysogenized with the BTP1 prophage that carries a pid locus identical to that of P22, revealed that the transcriptional start site (TSS) of P pid was indeed located 174 bp upstream of the pid start codon. This TSS location further confirms that the mutations identified in the DES-mutagenesis screen were located in the Pribnow box of the promoter region. Specifically, the C-182A and C-183A mutations modulated the -8 and -9 regions of the P pid promoter ( Figure 4). These data suggest that the TSS of the pid gene is located at nucleotide 40,880 on the genome of the P22 phage (NCBI Reference Sequence NC_002371.2).
The length of the pid transcript of P22 was examined in more detail via Northern blotting. To ensure pid expression, we exposed LT2 populations to a high MOI infection with wild-type or mutant P22 in order to favor lysogenisation and phage carrier state dynamics. After RNA extraction of these infected populations, Northern blotting involved either a ssDNA probe targeting the 5 UTR region of the pid mRNA or a ssDNA probe targeting the ORF region ( Figure 5a). Following P22wt infection, the 5 UTR probe revealed a transcript corresponding to the expected full-length of 476 bp spanning from the TSS to the Rho-independent termination site (referred to as F-band) (Figure 5b, lane 1). Following infection with P22-pid term-G292A (carrying the G-to-A terminator mutation 292 bp downstream of the pid start codon), the 5 UTR probe only revealed a vague F-band (Figure 5b, lane 7), supporting our earlier hypothesis ( Figure 1) that Rho-independent termination was important for pid mRNA stability. As expected, no pid transcript was observed following a negative control infection with P22-∆pid, in which the pid ORFan had been removed (from the promoter region, 240 bp upstream of the start codon, to the Rho-independent termination site) (Figure 5b, lane 6).
Interestingly, following P22wt infection, the ORF probe not only revealed the expected F-band, but also a slightly smaller fragment or transcript of between 350 and 400 bp (referred to as smaller or S-band) (Figure 5b, lane 1). To further investigate whether the 5 UTR was the cause of this S-band, we made truncations within the 5 UTR of the pid locus in P22, from the 3 end, while leaving the RBS of pid intact. The resulting mutant phages then contained the pid promoter region extending into either 109 bp (P22-pid 423 ), 57 bp (P22-pid 371 ), or 17 bp (P22-pid 331 ) of the 5 UTR region, followed by the pid RBS and ORF (Figure 5a). Northern blotting of the RNA extracted from high MOI infections with these truncated mutants yielded correspondingly shortened F-bands with both the 5 UTR or ORF probe (note that the annealing position of the 5 UTR probe has been removed in P22-pid 331 ) (Figure 5b, lanes 3-5). However, no parallel shortened S-bands could be detected with the ORF probe in these samples, suggesting that the 5 UTR region 110-162 bp downstream of the transcriptional start site is responsible for this S-band.
script was predicted to have an extensive 5′ untranslated region (5′UTR). Analysis o cent RNA-seq data from the Salmonella Typhimurium D23580 strain [6,[20][21][22], whic naturally lysogenized with the BTP1 prophage that carries a pid locus identical to tha P22, revealed that the transcriptional start site (TSS) of Ppid was indeed located 174 bp stream of the pid start codon. This TSS location further confirms that the mutations id tified in the DES-mutagenesis screen were located in the Pribnow box of the prom region. Specifically, the C-182A and C-183A mutations modulated the -8 and -9 reg of the Ppid promoter ( Figure 4). These data suggest that the TSS of the pid gene is locate nucleotide 40,880 on the genome of the P22 phage (NCBI Reference Seque NC_002371.2). This might in turn indicate that the 5 UTR region contains either secondary promoter elements or a processing site, leading to a smaller (S-band) transcript or fragment next to the full length (F-band) transcript. However, our promoter-probe studies detailed above ( Figure 2) did not indicate additional promoters to be present closer to the pid ORF. Moreover, boosted pid expression via infection with P22-pid C-183A (carrying the upregulating C-to-A mutation at -9 from the TSS) not only revealed the F-and S-bands with the ORF probe, but also a number of shorter fragments that might hint towards processing of the F-band (Figure 5b, lane 2). Interestingly, following P22wt infection, the ORF probe not only revealed the pected F-band, but also a slightly smaller fragment or transcript of between 350 and 4

Discussion
Our investigation of the transcriptional organization of the P22 pid locus showed that the promoter was slightly leaky, and that expression of the pid transcript was mainly attenuated because of an eroded Pribnow box. Mutations rendering the Pribnow promoter box more similar to the prokaryotic consensus sequence significantly increased expression, even within lysogens (cf. constitutively derepressed Pid/dgo interaction), and even in the absence of any other P22-related factors (cf. reporter plasmid studies). This in turn suggests that the Pid overexpression observed during the phage carrier state of P22wt [11] could stem from P22 copy-number (i.e., pid dosage) effects during episome formation, rather than from more complex regulatory effects.
The pid promoter, however, was found to be located far upstream of the start codon, and TSS analysis confirmed the presence of an extensive 174 bp 5 UTR. While long 5 UTR regions have been linked to mRNA stability [23][24][25] and expression regulation [26] in bacteria and archaea, the prevalence of long 5 UTRs in phages remains relatively uncommon and unexplored. The filamentous phage SW1, isolated from the deep-sea bacterium Shewanella piezotolerans WP3, contains two operons that have exceptionally long 5 UTRs (with lengths of 314 and 601 bp) that appear to thermoregulate the RNA stability of the transcripts [27]. Our results suggest that the 5 UTR of pid is responsible for the appearance of a second (and smaller) pid mRNA species, which could originate from an RNA processing site or the presence of a yet undetected secondary promoter. Although it currently remains to be resolved whether this smaller species serves a regulatory function, previous research has shown that mRNA processing in the 5 UTR region can affect transcript stability [28][29][30].
The 3 UTR of pid contains a functional Rho-independent termination site, as suggested by the mutation attenuating the Pid/dgo interaction and pid transcript instability. Previous studies suggest that disruption of the stem-loop structure of Rho-independent termination sites abolishes terminator activity [13,31,32]. As a consequence, the mRNA is degraded and expression of the ORF is drastically decreased. This is exemplified by the tI terminator of phage λ, in which disruption of the stem structure via the introduction of point mutations or deletions significantly reduces termination efficiency and decreases RNA stability [31,32]. Similarly, in E. coli, the disruption of a G+C-rich stem structure caused almost complete abolition of terminator activity and reduced expression of the crp gene due to increased mRNA degradation [13].
In summary, the expression regulation mechanism of the P22 pid ORFan gene was scrutinized, revealing a weak promoter region with an extensive 5 UTR region. The latter may harbor a transcript processing or secondary promoter site, although its biological significance still remains elusive. Termination of the pid transcript is controlled by a Rho-independent terminator that is important for proper pid expression. As such, this study provides functional characterization of one of the few remaining ORFans in the P22 model phage.

Strains, Phages, and Growth Conditions
Bacterial strains, phages, and plasmids used in this study are listed in Table 1. Lysogeny broth (LB) medium [33] was used to culture bacteria, either as a broth or as agar plates after the addition of 1.5% (for spreading plates) or 0.7% (for soft agar plates) agar. Stationary phase cultures were obtained by growing cultures in LB for 15-20 h at 37 • C under aerated conditions (agitation of 200 rpm on rotary shaker). Exponential phase cultures were obtained by 1:100 dilution of stationary phase cultures in pre-warmed broth and further grown at 37 • C to an OD 600 between 0.2 and 0.3. When required, the following chemicals (AppliChem GmbH, Darmstadt, Germany) were added to the growth medium at the indicated final concentrations: kanamycin (50 µg/mL), ampicillin (100 µg/mL), tetracycline (20 µg/mL), and 5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside (40 µg/mL; X-Gal).

Construction of Lysogens
Lysogens were constructed by propagation of a phage on indicator strain LT2K7 via the double agar overlay method [34]. Cells within the turbid zone of a plaque were isolated and purified on green indicator plates [39] and plates were incubated for 15-20 h at 37 • C. Pale green colonies were selected and cross streaked across the respective phage on green indicator plates to confirm lysogenisation.

Construction of Phage Mutants
Amplicons used in the construction of phage mutants were obtained via PCR with Phusion DNA polymerase (Thermo Fisher Scientific, Waltham, MA, USA). Primers were synthesized by IDT (Leuven, Belgium) and are listed in Table 2. P22-∆pid was constructed on the basis of the λ-red recombineering system, described by Datsenko and Wanner in the year 2000 [37]. The amplicon was generated via PCR amplification with pKD13 as template and primer pair P22_∆pid_Fw and P22_∆pid_Rev. The introduced kanamycin cassette was subsequently flipped out with pCP20, which harbors Flp to recombine two frt-sites flanking the kanamycin gene [38].
Phage mutants P22-pid 423 , P22-pid 371 , and P22-pid 331 were constructed on the basis of the λ-red recombineering system [37] and a two-step selection and counterselection process using the tetA-sacB cassette, described by Li et al. in the year 2013 [36]. The tetA-sacB cassette was amplified from E. coli XTL298 in a colony PCR with primers P22_5UTRpid_tetAsacB_Fw and P22_5UTRpid_tetAsacB_Rev. The obtained cassette was introduced in the 5 UTR region of pid via homologous recombination, replacing 13-157 bp upstream of the pid ORF. Correct clones were selected on tetracycline LB agar plates. The tetA-sacB cassette was subsequently substituted via homologous recombination with the truncated 5 UTR amplicons obtained via either primer pairs P22_5UTR423pid_Fw and P225UTRpid_Rev, P22_5UTR371pid_Fw and P225UTRpid_Rev, or P22_5UTR331pid_Fw and P225UTRpid_Rev. Correct clones were selected on counterselection agar plates and validated via Sanger sequencing (Macrogen Europe B.V., Amsterdam, the Netherlands).

Construction of Plasmids
The inserts that were ligated in the vectors were obtained via PCR with Phusion DNA polymerase (Thermo Fisher Scientific, Waltham, MA, USA). Primers were synthesized by IDT (Leuven, Belgium) and can be found in Table 2. Constructs were validated via Sanger sequencing (Macrogen Europe B.V., Amsterdam, the Netherlands).
Plasmids pFPV-P22-Ppid 338 -gfp and pFPV-P22-Ppid 232 -gfp were constructed by digesting pFPV25 with XbaI and BamHI (Thermo Fisher Scientific, Waltham, MA, USA). The digested vector was subsequently ligated with a truncated 5 regulatory region of pid of either 338 bp or 232 bp. The latter amplicons were obtained, respectively, using primer pairs P22_pid338_Fw and P22_pid_Rev or P22_pid232_Fw and P22_pid_Rev, and digested with XbaI and BamHI prior to ligation.

DES-Mutagenesis Screen
To screen for mutants in P22, we added 50 µL of the mutagen diethylsulfate (DES) to 50 µL of LT2 [P22] stationary phase culture and the mixture was incubated for 50 min at 37 • C without shaking. Next, 50 µL of the mixture was added to 2 mL of LB broth and grown for 6 h at 37 • C under shaking conditions. Phages were extracted and propagated on LT2 as plaques in LB soft agar or as lysates in LB. This procedure was repeated eight times to obtain eight pools of mutagenized phage lysate. The mutagenized phage lysate was propagated on stationary phase culture of LT2K7 on green indicator plates [39]. After overnight incubation, each indicator plate was replica plated on a new indicator plate and on a LB agar plate containing X-Gal. In total, 39,950 plaques were screened on LB agar plates supplemented with 40 µg/mL X-Gal and 26,740 plaques on LB agar plates supplemented with 20 µg/mL X-Gal, allowing the identification of more blue phenotypes compared to wild-type P22.

Fluorescence Population Level Measurement
Population-level fluorescence measurements were obtained in a Fluoroskan Ascent FL (Thermo Fisher Scientific, Waltham, MA, USA), equipped with a filter pair with an excitation wavelength of 485 nm and an emission wavelength of 520 nm. A multi-well microtiter plate was filled with 200 µL of stationary phase cultures grown in AB-glucose medium (2 g/L (NH 4 ) 2 SO 4 , 6 g/L Na 2 HPO 4 , 3 g/L KH 2 PO 4 , 3 g/L NaCl, 0.1 mM CaCl 2 , 1.0 mM MgCl 2 , and 0.003 mM FeCl 3 [40] and supplemented with 0.2% D-glucose, 1 µg/mL uracil, and 1 µg/mL thiamine and 0.2% casamino acids). The obtained fluorescence values were subsequently normalized over the optical density at OD 600 of the sample (obtained via the Multiskan RC, Thermo Labsystems OY, Helsinki, Finland).

Statistical Analysis
All fluorescence measurements were performed in triplicate. Statistical analyses, histograms, and Tukey HSD tests (p = 0.05) were carried out using the JMP software (version Pro 15.1.0; SAS, Cary, NC, USA), and differences were considered significant when p ≤ 0.05.

Northern Blotting
Exponential phase cultures of S. Typhimurium LT2 were infected with P22wt, P22pid 423 , P22-pid 371 , P22-pid 331 , P22-pid C-183A , P22-pid term_G292A , or P22-∆pid (MOI 10) for 90 min. Infection was stopped by adding 20% volume of stop mixture (95% EtOH, 5% phenol; pH 5.5), followed by snap-freezing in liquid nitrogen. Following this, total RNA was extracted according to the hot phenol method, and the remaining DNA was digested via DNase I treatment (Thermo Fisher Scientific, Waltham, MA, USA). Subsequently, 7 µg of total RNA per sample was resolved on a 6% polyacrylamide gel containing 7 M urea and transferred to an Amersham Hybond-XL (GE Healthcare, Chicago, IL, USA) membrane. After UV-crosslinking, transcripts of interest were detected by radiolabeled oligonucleotides Sal_5S_rRNA (5S rRNA probe for loading control), Pid_5UTR, and/or Pid_ORF ( Table 2). The labelling reaction was carried out as follows: 10 pmol of the oligonucleotides were 5 -labeled with 10 µCi 32 P-γ-ATP (Perkin Elmer, Waltham, MA, USA) by T4 polynucleotide kinase (PNK, Thermo Fisher Scientific, Waltham, MA, USA) for 1 h at 37 • C. After removing unincorporated 32 P-γ-ATP with a Microspin G-25 column (GE Healthcare, Chicago, IL, USA), 7 µL of the labelled oligonucleotide was added to 17 mL EKONO hybridization buffer (G-Biosciences, Saint Louis, MI, USA) and hybridized to the membrane, overnight at 42 • C in a hybridization oven. Radioactive signals were imaged with a Typhoon 9400 (Variable Mode Imager, Amersham Biosciences, Amersham, United Kingdom), and the membrane was stripped for re-probing by immersing it in 500 mL hot H 2 O/5 mL 10% SDS for 7 min.