Construction of an rsmX co-variance model and identification of five rsmX non-coding RNAs in Pseudomonas syringae pv. tomato DC3000

Abstract

Non-coding RNAs (ncRNAs) are important components of many regulatory pathways and have key roles in regulating diverse functions. In the Pseudomonads, the two-component system, GacA/S, directly regulates at least two well-characterized ncRNAs, RsmZ and RsmY, which act by sequestration of translation repressor proteins to control expression of various exoproducts. Pseudomonas fluorescens CHA0 possesses a third ncRNA, RsmX, which also participates in this regulatory pathway. In this study we confirmed expression of five rsmX ncRNAs in Pseudomonas syringae pv. tomato DC3000, and determined the distribution of the members of the rsmX ncRNA family by screening available genomic sequences of the Pseudomonads. Variable numbers of the rsmX family exist in Pseudomonas genomes, with up to five paralogs in Pseudomonas syringae strains. In Pseudomonas syringae pv. tomato DC3000, the rsmX genes are 112 to 120 nucleotides in size and are predicted by structural analysis to contain multiple exposed GGA motifs, which is consistent with structural features of the Rsm ncRNAs. We also found that these rsmX ncRNA genes share a conserved upstream region suggesting that their expression is dependent upon the global response regulator, GacA.

Introduction

Extensive sequencing and computational analyses have revealed that bacteria contain a wide variety of small, non-coding, RNA (ncRNA) molecules. Although the function of the majority of these ncRNAs remains to be determined, it has become increasingly clear that they play a critical role in regulatory networks, facilitating adaptation to diverse environmental stresses.¹ For a large number of ncRNAs, regulation occurs through an antisense mechanism. In this process, complementary base-pairing with a target mRNA either activates or represses translation of the transcripttargets the mRNA for degradation.² However, some ncRNAs are capable of interacting with proteins, influencing expression of all genes that the target protein activates or represses.²

Members of the Rsm/Csr family of ncRNAs are well-characterized examples of protein-binding ncRNAs. The first member of this family, CsrB, was discovered in Escherichia coli when it was copurified with its protein-binding partner, CsrA, a global regulator involved in the transition from exponential to stationary growth phase.^3,4 Genetic and overexpression analyses revealed that CsrB antagonizes CsrA, affecting glycogen biosynthesis, glycolysis, biofilm formation and motility.^4–6 Functional analogs of the CsrA/CsrB system have since been discovered in Erwinia,⁷ Salmonella,⁸ Vibrio⁹ and Pseudomonas.¹⁰

In many of these systems multiple, redundant, ncRNA antagonists, are involved. For example, in Pseudomonas fluorescens CHA0 (CHA0), three functional analogs of the E. coli csrB are present. These three ncRNAs are the 127 nucleotide rsmZ,¹⁰ the 118 nucleotide rsmY¹¹ and the 119 nucleotide rsmX.¹² Expression of rsmX and Y are cell density dependent, with expression increasing throughout the growth phase.^11,12 However, expression of rsmZ is delayed, with optimal expression occurring at the end of exponential phase.^10,12 Expression of all three ncRNAs is dependent upon a conserved upstream activating sequence (UAS), which is bound by the response regulator, GacA.¹³ Additional transcription factors are also thought to be involved in expression of the ncRNAs, including integration host factor (IHF) in the case of rsmZ.¹³ Production of the three ncRNAs leads to the sequestering of two CsrA homologs, RsmA and RsmE. Binding is dependent upon the presence of repeated, unpaired, GGA motifs, which are present in exposed stem loops of the ncRNAs.¹⁴ In the absence of the ncRNAs, RsmA and RsmE act as repressors, preventing translation of target mRNAs by blocking ribosome binding sites.^15,16 Thus, sequestering of these regulators leads to release and translation of a wide variety of downstream genes involved in secondary metabolism.^17,18

Homologs of rsmZ and rsmY have also been discovered in other Pseudomonas strains using laboratory and/or computational approaches. These include P. aeruginosa PA01,¹⁹ Pseudomonas fluorescens Pf-5, Pseudomonas putida KT2440 and Pseudomonas syringae pv. tomato DC3000 (DC3000).²⁰ Additionally, curated multiple sequence alignments and co-varaiance models for both rsmY and rsmZ are available in Rfam²¹ (http://rfam.janelia.org/) and these ncRNAs have been predicted in all fully sequenced Pseudomonas species strains, including P. aeruginosa, P. fluorescens, P. entomophila, P. stutzeri, P. mendocina and P. syringae. In contrast, rsmX has only been experimentally verified in CHA0,¹² and predicted in P. fluorescens Pf-5, P. fluorescens Pf0-1, P. putida KT2440 and DC3000.^12,20 Additionally, the current version of Rfam (version 9.1) does not contain a model for rsmX.

As the distribution of RsmX ncRNAs among the Pseudomonads has not yet been fully evaluated, the aim of this study was to determine if rsmX genes are conserved throughout all sequenced Pseudomonas strains, as for rsmY and rsmZ. In this study we report the identification of additional members of the Rsm family of ncRNAs and we show that the RsmX family is larger and more complex than previously thought. Furthermore, we provide a covariance model for the RsmX family of ncRNAs.

Results

Identification of rsmX genes in pseudomonas spp. using BLAST.

To determine if rsmX is conserved throughout the genus Pseudomonas, we employed a simple BLASTN analysis across all sequenced Pseudomonas strains using the experimentally verified sequence of rsmX from CHA0.¹² BLAST analysis revealed that variable numbers of rsmX homologs were present in some fully sequenced Pseudomonas species strains (Data not shown). In particular, the three fully sequenced P. syringae strains (DC3000, P. syringae pv. syringae B728a [B728a] and P. syringae pv. phaseolicola 1448a [1448a]), all appear to contain five rsmX paralogs (Table 2). All but one of these copies is located within an intergenic region. In the one exceptional case an rsmX candidate in 1448a is predicted to be located within the annotated open reading frame PSPHH_2778. Since this ORF encodes for a hypothetical protein, is relatively small and is absent from syntenic positions in the other strains (Fig. S1), it is highly probable that PSPHH_2778 is mis-annotated as a protein coding region. Furthermore, all copies, except DC3000 rsmX-1, displayed synteny across the three P. syringae genomes (Fig. S1). Interestingly, two of the rsmX ncRNAs (DC3000 rsmX-3 and rsmX-4) are located in tandem in all three syringae strains (Fig. S1).

Table 2.

rsmX candidates identified in fully sequenced Pseudomonas genomes using CMsearch

	Genomic coordinates
Strain	5′ end	3′ end
P. syringae DC3000	3244374c	3244256c
	4170369c	4170252c
	6144830	6144943
	6145122	6145235
	6198149c	6198038c
P. syringae B728a	3053360	3053480
	2024633	2024750
	5867169	5867282
	5867461	5867572
	5929638c	5929527c
P. syringae 1448a	3215850	3215970
	2002611	2002728
	160740c	160627c
	160448c	160336c
	5761100c	5760989c
P. fluorescens Pf-5	4774103c	4773991c
P. fluorescens Pf0-1	4402504c	4402389c
P. mendocina ymp	6376*	6478*
	3581176c	3581069c
P. stutzeri A1501	356904	357013
	357144	357250

^*This rsmX candidate was not identified with BLAST.

Expression of rsmX homologs in P. syringae pv. tomato DC3000.

5′ and 3′ RACE was used to determine whether the five rsmX paralogs found through BLAST analysis were expressed in DC3000 and to establish the size and genomic boundaries for these transcripts. All five rsmX transcripts in DC3000 were expressed under the condition tested and the co-ordinates are shown in Table 1. The sizes of the transcripts varied from 112 nt to 120 nt. The rsmY and rsmZ genes predicted by Rfam were also expressed under the same condition. These transcripts were slightly larger than the rsmX transcripts, with lengths of 126 nt and 132 nt, respectively. The genomic locations of rsmY and rsmZ established with 5′ and 3′ RACE are in overall agreement with Rfam predictions.

Table 1.

Pseudomonas syringae pv. tomato DC3000 rsm coordinates

ncRNA	5′ end	3′ end	Length
rsmX-1	3244374c	3244255c	120 nt
rsmX-2	4170369c	4170253c	117 nt
rsmX-3	6144830	6144943	114 nt
rsmX-4	6145122	6145235	114 nt
rsmX-5	6198149c	6198038c	112 nt
rsmY	555344	555469	126 nt
rsmZ	1728435	1728566	132 nt

^cExpression occurs on the negative, complementary, strand.

Sequence similarity and secondary structure of rsmX candidates.

To determine the degree of similarity among the five rsmX transcripts in DC3000 and the one confirmed copy of rsmX in CHA0,¹² we aligned the sequences with MUSCLE (Multiple Sequence Comparison by Log Expectation). The results of the MUSCLE alignment are shown in Figure 1A and reveal a large degree of sequence similarity, with approximately 54 to 58% of the nucleotides being conserved across the six transcripts. Furthermore, LocARNA predicts a consensus secondary structure for these six transcripts (Fig. 1), containing four stem loops with exposed GGA motifs. Additionally, an apparent Rho-independent terminator is conserved across the five transcripts (Fig. 1). These features are consistent with what was previously predicted for the CHA0 rsmX gene alone¹² and are indicative of rsm ncRNAs in general. Further, the presence of exposed GGA motifs suggest that these RsmX ncRNAs function like other Csr/Rsm ncRNAs by pairing with RNA binding proteins.¹⁴

Figure 1.

(A) MUSCLE alignment of the five rsmX genes in P. syringae pv. tomato DC3000 and the single rsmX gene from P. fluorescens CHA0. (B) LocARNA alignment of the sequences in (A). Matching nested parentheses, highlighted with boxes, indicate base pairing in the consensus secondary structure. Unpaired “GGA” motifs are highlighted in pink. The sequences highlighted in green are identical between all six rsmX genes and are predicted to form a stem-loop, which functions as a rho-independent terminator. A plain text version of the alignment is provided in Figure S2. (C) Consensus secondary structure of the sequences in (A), predicted by RNAalifold. The 5′; and 3′; ends of the ncRNA are labeled. The coloring of base pairs represents: Red—one type of base pair occurs in all sequences used to generate the consensus; Yellow—two types of base pairing occur; Green— three types of base pairing occur. The shading of base pairs represents: Saturated, no inconsistent sequences; Pale, one inconsistent sequence; Very pale, two inconsistent sequences.

Construction of an rsmX co-variance model.

Using the five rsmX sequences from DC3000 and the one from CHA0, a co-variance model (CM) was constructed using the Infernal toolset. This model was used to search all fully sequenced Pseudomonas genomes to locate additional rsmX candidates that may have been missed using BLASTN alone. The results of the CMsearch are shown in Table 2. The highest number of rsmX candidates were found in the Pseudomonas syringae pv. strains (DC3000, B728a and 1448a), with five in each genome. These five candidates are the same as those found using BLASTN. Construction of the CM allowed for identification of an additional rsmX candidate in P. mendocina ymp that was not identified previously by BLAST (Table 2). As with BLAST, we were unable to detect any copies of rsmX in P. fluorescens SBW25, P. entomophila or any of the four sequenced P. aeruginosa or P. putida strains.

Clustering of rsmX, Y and Z.

The 21 rsmX candidates identified with CMsearch and the P. fluorescens CHA0 rsmX sequence were clustered using RNAclust, along with 14 rsmY and 14 rsmZ sequences present in Rfam as either seed or predicted sequences (see Methods), to ascertain the relationship between the three classes. The resulting dendogram is shown in Figure 2. Overall, the sequences cluster in a manner consistent with their naming (i.e., all rsmX sequences cluster together, as do rsmY and rsmZ sequences).

Figure 2.

An RNAClust-generated dendrogram showing the similarity based upon secondary structure between the rsmX, rsmY and rsmZ genes from all fully sequenced Pseudomonads in Rfam version 9.1. Coordinates of the ncRNA genes are shown with the names and Genbank accession number of the host strain. Genes suffixed by “[RsmY]” and “[RsmZ]” were found using the Rfam models of the same names. A plus (“+”) before the family name indicates that the sequence is a seed sequence used to construct the Rfam model. Genes suffixed by “[=RsmX]” were found using the RsmX model discussed in this paper. Coordinates for the genes from CHA0 are not given since that organism has not been fully sequenced. The rsmX-1 through rsmX-5 genes of DC3000 are denoted in red font. The gene highlighted in blue is the predicted rsmX gene in P. putida KT2440,²⁰ not detected in our analysis (see discussion).

Identification of putative GacA binding sites upstream of DC3000 rsmX candidates.

The Rsm family of ncRNAs have been shown to be regulated by the two component response regulator GacA in several Pseudomonads.^10,13,22 In order to determine if the rsmX candidates from DC3000 share this conserved regulatory site, the upstream regions of these ncRNAs were aligned and a sequence logo was generated (Fig. 3). These analyses revealed the existence of three highly conserved motifs. As shown in Figure 3, one of these motifs begins approximately 75 nt upstream of the transcriptional start site and is highly similar to the GacA UAS previously described for the rsmX ncRNA of P. fluorescens CHA0 (TAT AGC GAA TTT CCT ACA).¹² Alignment of the sequences also revealed a highly conserved −10 box and a putative −35 box directly upstream of the rsmX ncRNAs (Fig. 3). This putative −35 box closely resembles the consensus −35 sequence (TTG ACA), however, we do acknowledge that it displays atypical spacing relative to the −10 box. Atypical spacing for a −35 box has been reported in Pseudomonas previously.²³

Figure 3.

(A) MUSCLE alignment of the promoter regions starting roughly 80bps upstream of each of the five rsmX genes in DC3000. The upstream activating sequence (UAS) and putative −10 and −35 boxes are outlined. The +1 arrow indicates the transcription start site. Asterisks denote conserved nucleotides. (B) Sequence logo showing the conservation of sequence found in the promoter regions starting roughly 80bps upstream of each of the five rsmX genes in DC3000. The x-axis denotes position relative to the approximate transcription start site.

Discussion

In this study we demonstrated that the rsmX ncRNA, first identified in P. fluorescens CHA0,¹² is common throughout the genus, Pseudomonas. Using BLASTN analysis with the experimentally confirmed rsmX gene from CHA0 as the query, we were able to predict that five paralogs of an rsmX ncRNA were present in P. syringae pv. tomato DC3000. RACE analysis confirmed the expression and genomic boundaries of these transcripts and structural analysis revealed the presence of stem loops displaying exposed GGA motifs (Fig. 1). These unpaired GGA motifs are thought to play a role in the sequestration of binding proteins.¹⁴ We observed four of these exposed motifs, consistent with what has been observed for the P. fluorescens CHA0 rsmX ncRNA.¹²

Using the five confirmed rsmX sequences in DC3000 and the one previously confirmed in CHA0,¹² we were able to develop a co-variance model for rsmX. This model was successfully applied to sequenced Pseudomonas genomes, confirming BLAST predictions for the presence of rsmX candidates in P. syringae, P. stutzeri, P. fluorescens and P. mendocina and the lack of rsmX candidates in all sequenced strains of P. aeruginosa, P. putida and the one sequenced strain of P. entomophila (Table 2). Furthermore, the model was able to successfully locate a second copy of rsmX in P. mendocina, which was not found with BLASTN alone. Interestingly, neither BLASTN nor the co-variance model were able to detect a previously predicted rsmX-like gene in P. putida KT2440.²⁰ When this predicted sequence was clustered with all rsmX, Y and Z sequences, it grouped separately, suggesting that this predicted gene in KT2440 may encode a representative of another, distinct, class of ncRNAs (Fig. 2). This may explain why we were unable to locate it with our rsmX-specific search.

Previous studies have shown that Rsm family ncRNAs, including RsmX in P. fluorescens CHA0, are regulated by GacA.^12,15,17,19 Further, it is thought that GacA-mediated regulation acts exclusively through the Rsm family ncRNAs, as has been documented for P. aeruginosa.²² In these cases, an upstream activating sequence (UAS) for GacA activation is present in the promoter region of these ncRNAs.^11,13,17 An alignment of the regions upstream of the rsmX genes in P. syringae pv. tomato DC3000 revealed that all five sequences contain a motif that resembles the GacA UAS of previously described Rsm gene families in other strains (Fig. 3). This strongly suggests that the newly identified rsmX ncRNAs are also regulated by GacA. This UAS was highly conserved among the five candidates, as were the putative −35 and −10 boxes (Fig. 3).

In P. fluorescens CHA0, rsmX, Y and Z have been demonstrated to possess similar affinities for their binding partners, RsmA and E.^12,15 However complete redundancy of the three rsm ncRNAs is unlikely, with in-depth analyses suggesting that they are regulated separately. For example, it has been demonstrated that rsmZ is produced in late growth phase, whereas rsmX and Y are expressed in parallel throughout the growth phase.¹² The potential for differential regulation was confirmed at the molecular level, with the rsmZ promoter region displaying significant disparity to that of rsmX and Y.¹³ For instance, the UAS of rsmZ is located further upstream of the transcription start site than for rsmX and Y and the promoter region contains additional regulatory elements in the form of a PsrA box and integration host factor recognition (IHF) sequences.¹³ Although no obvious differences exist in the temporal expression of rsmX and Y, differential regulation may be subtle. Analysis of the promoter regions of the two rsm ncRNAs show distinct differences in the UAS, as well as the linker region between the UAS and the −10 box.¹³ Further, our clustering analyses reveal that the rsmX and Y sub-families are distinct (Fig. 2), implying inherent differences between the two ncRNAs.

Although we may be able to assume that differential regulation of rsmX, Y and Z occurs in P. syringae pathovars as for P. fluorescens CHA0, the presence of five copies of an rsmX ncRNA raises the question of what benefits these bacteria obtain from harboring seemingly redundant ncRNAs. As demonstrated in this study, these rsmX ncRNAs share significant similarity at both the nucleotide and structural level (Fig. 1), and the high degree of similarity was further confirmed by the clustering analyses (Fig. 2). The lack of obvious differences provides little clue as to differing functions among the five copies.

Multiple ncRNAs controlling a single regulatory process is not without precedent and have been described as being both redundant and additive in function.¹⁸ In Vibrio spp., multiple qrr (quorum regulatory RNA) ncRNAs are produced at low cell density. Through an antisense binding mechanism, these ncRNAs destablize luxR mRNA, affecting transcription of all downstream genes whose expression is dependent upon the LuxR transcriptional activator.²⁴ In V. cholerae, four qrr ncRNAs are present and all four need to be deleted in order to eliminate quorum sensing control of downstream genes.²⁴ In such a way, redundant ncRNAs may be advantageous in that they promote a gene dosage effect, resulting in a rapid strong output of the ncRNA under certain conditions.¹⁸

In contrast, in V. harveyi, five such qrr ncRNAs are present.²⁵ Like V. cholerae, they appear highly similar at the nucleotide and structural level, however, these five ncRNAs have been demonstrated to not act redundantly, but additively, and it has been proposed that through differential regulation they may finely-tune quorum sensing regulation.²⁵ We can only speculate whether a similar scenario exists in DC3000 based on minor distinctions between the five RsmX ncRNAs, resulting in small differences in either expression or function. In such a way, rsmY and Z may provide the coarse tuning for GacA regulation and multiple copies of the highly similar rsmX are responsible for fine-tuning. Overall, such delicate variations could result in different binding affinities to the Rsm proteins. Interestingly, whereas only one or two copies of Rsm/Csr binding proteins have been identified in E. coli, Salmonella enterica, Erwinia spp., P. aeruginosa and P. fluorescens, P. syringae pathovars have an elevated number of binding protein homologs. Three homologs are present in B728a, four in DC3000 and five in 1448a.¹⁵ Therefore, subtle variations in the regulation or sequence of the five rsmX ncRNAs may target them to different binding proteins, which may themselves, be differentially expressed.

Clearly the complexity of Rsm/GacA regulation in the P. syringae pathovars, extends beyond rsmX alone and further analyses are required to determine the roles of individual members. We are currently assessing differential expression of the seven Rsm ncRNAs in DC3000. Further, future studies will be directed at more closely examining the four Rsm binding protein candidates in DC3000, including an analysis of their binding affinities to each Rsm ncRNA and their patterns of expression.

Materials and Methods

Bacterial growth conditions.

DC3000 was incubated at 28°C in 4 ml King's broth (KB)²⁶ medium for 16 h. Overnight cultures were used to inoculate 50 ml of fresh KB broth in a 250 ml baffled flask to a final OD₆₀₀ = 0.02. Flasks were incubated at 28°C with shaking at 200 rpm for 8 h.

RNA isolation.

Small RNA fractions (<200 nt) were isolated from 1 ml of cultured cells using the PureLink miRNA Isolation Kit (Invitrogen) according to the manufacturer's protocol. Isolated RNA was treated twice with 2 units of TURBO DNase (Ambion) to eliminate DNA contamination. RNA was purified from the DNase mixture by bringing the total volume of the sample up to 90 µl with RNase-free H₂O (Ambion) and reapplying the sample to a PureLink miRNA Isolation column.

5′ and 3′ RACE.

5′ RACE assays were performed using the 5′ RACE System for the Rapid Amplification of cDNA Ends, V2.0 kit (Invitrogen). Five hundred nanograms to 1 µg of isolated RNA were used in each reaction and the procedure was carried out following the manufacturer's protocol. Following amplification with a gene specific primer (GSP2), PCR products were separated on a 2% agarose gel and bands of interest were excised, geleluted (Zymoclean Gel DNA Recovery kit, Zymo), and cloned into pCR 2.1 TOPO vector (Invitrogen). Plasmids from three separate clones were purified (QIAprep spin, QIAGEN) and sequenced, using both M13 forward and reverse primers. Primers used in 5′ RACE analyses are shown in Table 2.

3′ RACE was performed using a protocol adapted from Argaman et al.²⁷ Briefly, 1 µg of RNA was mixed with 100 pmol of RNA adapter (5′-phosphate-UUC ACU GUU CUU AGC GGC CGC AUG CUC-idT-3′), heat-denatured at 95°C for 5 min, then quick-chilled on ice. The adapter was ligated at 17°C for 12 h in the presence of 40 units of T4 RNA ligase (New England Biolabs) and 40 units of RNase OUT (Ambion) in a buffer containing 50 mM Tris-HCl (pH 7.9), 10 mM MgCl₂, 4 mM DTT, 150 µM ATP and 10% DMSO. The ligated RNA product was purified from the reaction using the RNA Clean and Concentrator −5 kit (Zymo) and reverse-transcribed using 20 pmol of a single primer complementary to the RNA adapter (A1). Reverse transcription was performed using the Thermoscript RT system (Invitrogen) according to the manufacturer's protocol. The products of reverse transcription were amplified using a 2 µl aliquot of the RT reaction, 20 pmol of each gene-specific and adapter-specific primer (A2) and 1X Ex-Taq Polymerase mix (TaKaRa Bio). Cycling conditions were as follows: 95°C/2 min; 35 cycles of 94°C/30 sec, 57°C/45 sec, 72°C/30 sec; 72°C/10 min. PCR separation, gel extraction, cloning and sequencing were performed as described for 5′ RACE. Primers used in 3′ RACE analyses are shown in Table 2.

Computational analyses.

Version 2.2.22+ of BLASTN (http://blast.ncbi.nlm.nih.gov/Blast.cgi) was used to generate alignments of CHA0 rsmX¹² to fully sequenced Pseudomonas spp. genomes. A list of all genomes used in this study and their Genbank and Refseq accession numbers is provided in Table S1. The non-default parameters of word size = 7 and e-value <1 × 10^-5, were used to compute the alignments.

Multiple sequence alignments of candidate rsmX nucleotide sequences from DC3000 and the rsmX sequence from P. fluorescens CHA0,¹² were performed using MUSCLE v4.0.²⁸ Structural alignments were performed with LocARNA v1.5a,²⁹ using the default settings. A consensus secondary structure prediction of the same RsmX ncRNAs was generated using RNAalifold³⁰ from the ViennaRNA package (http://rna.tbi.univie.ac.at/cgi-bin/RNAalifold.cgi).

Co-variance models (CM) for the family of DC3000 rsmX ncRNAs and the CHA0 rsmX ncRNA were constructed using CMbuild, CMcalibrate and CMsearch from the Infernal software package v1.0.2,³¹ and calibrated separately for each of the fully sequenced Pseudomonas genomes in three steps. (1) A zeroth-order null model of each genome was constructed. (2) An uncalibrated CM was constructed using CMbuild from the aligned training set and null model computed in step 1. (3) The CM was calibrated using CMcalibrate and a fixed seed of 1 was used to ensure reproducible results. Following construction of each model, the corresponding genome was searched for rsmX ncRNAs using CMsearch. Results were filtered to retain matches with e-values less than 1 × 10^-7.

RNAclust v1.1 (www.bioinf.unileipzig.de/∼kristin/Software/RNAclust/) was used to cluster Pseudomonas rsmX, rsmY and rsmZ sequences. The set of input sequences consisted of the Pseudomonas seed sequences used to construct the rsmY and rsmZ co-variance models in the Rfam version 9.1 database, the Rfam version 9.1 database predictions for rsmY and rsmZ in all Pseudomonads, the rsmX, rsmY and rsmZ genes from CHA0,¹² the rsmX prediction from P. putida KT2440,²⁰ and the 21 high-quality matches to our rsmX model found by CMsearch in the Pseudomonads. Unlike our rsmX model and the rsmY model in Rfam, the Rfam rsmZ model contains part of the upstream, un-transcribed promoter region. Therefore, rsmZ sequences were trimmed to include only the actual transcript. After clustering was complete, NJplot v2.3,³² was used to render the results as a dendogram.

Nucleotide sequence alignments of the promoter regions 80 bps upstream of the transcriptional start sites for each of the five rsmX ncRNAs in DC3000 was performed using MUSCLE v4.0. Sequence logos were generated using WebLogo 3.0.³³

Table 3.

Oligonucleotides used in this study

Oligonucleotide name	Sequence (5′-3′)
3′ RACE
RNA Adapter	P-UUC ACU GUU CUU AGC GGC CGC AUG CUC-idT*^
A1	GAG CAT GCG GCC GCT AAG AAC AGT G*
A2	CAT GCG GCC GCT AAG AAC AGT*
RsmY	CGT AGC GCA GGA AGC GCA AC
RsmZ	CGA AGC TGT GCC AAC GGA CAG
RsmX-1	GCA GGA AGC GTT GCA GAG G
RsmX-2	CAA TTC AAC AGC TGC AGC GAA GG
RsmX-3	ACA AGG AGT TCA CCA GGA TCA GG
RsmX-4	ACA GGG AGT TCA CCA GGA TCA AG
RsmX-5	CAC CAG GGT CAA GGA TGA CCG
5′ RACE
RsmY GSP1	CTG TTT CCC TGA TTT CCC TTT CAC
RsmY GSP2	TTT CAC CCC GCC GTC CTG G
RsmZ GSP1	CTT GTA TTC CCT TGT CAT CGT CC
RsmZ GSP2	CTT GTC ATC GTC CTG ATG AAT CGC
RsmX-1 GSP1	TAC CAT CCC GAC GTC CTG TC
RsmX-1 GSP2	CGA CGT CCT GTC AGT AGC CA
RsmX-2 GSP1	CAA CGT CCT GTC AGT AGC CTC C
RsmX-2 GSP2	GTC AGT AGC CTC CTG GCA ATG G
RsmX-3 GSP1	ATT CCA ACT CCC TGT CGG CTG
RsmX-3 GSP2	TGT CGG CTG CCT TCC CGG
RsmX-4 GSP1	CTG TCG GCC GCC TCC GAG
RsmX-4 GSP2	GCC GCC TCC GAG GCG ATG
RsmX-5 GSP1	CCA TCC CGA CTC CCT GTC G
RsmX-5 GSP2	TGT CGG CTG CCT CCA AGG C
Sequencing
M13 Forward	GTA AAA CGA CGG CCA G
M13 Reverse	CAG GAA ACA GCT ATG AC

^*These oligonucleotides were used in the production and amplification of cDNA for 3′ RACE as described in the methods.

^{^}p, 5′ monophosphate. idT, inverted deoxythymidine.