Cloning of approximately 100 bp upstream sequence encompassing each of the three upstream motifs 1–3 in front of chimeric ncRNA::GFP reporter genes suggested that all three motifs have independent transcriptional activity. To test whether the PSE/UM1 is sufficient for ncRNA expression, we made a reporter construct consisting of a fragment of the CeN7 locus including 90 bp upstream and 67 bp transcribed (i.e. -90 to +67 bp; Additional file 1) sequence. The CeN7 locus encodes an SL2 RNA with a TATA-less PSE/UM1 upstream sequence. The fragment was inserted into plasmid pPD95_77, which contains a GFP ORF, thereby creating a CeN7::GFP chimeric reporter gene (henceforth CeN7_100). The recombined gene was co-injected with the rol-6 marker gene into young adult hermaphrodite gonads, and transgenic strains were selected based on the presence of the roller phenotype. Reporter gene expression was examined by reverse transcription polymerase chain reaction (RT-PCR) with one primer located in the CeN7 fragment and the other primer in the plasmid sequence, using RNA extracted from transgenic C. elegans strains (Figure 2). To confirm that the observed transcriptional activation was not a spurious result, we further selected an additional PSE/UM1 SL2 RNA locus (CeN16-1). The construct CeN16-1_100 was made as above from a fragment containing 119 bp upstream sequence and 60 bp transcribed sequence, and expression of the fusion reporter (CeN16-1::GFP) was confirmed with RT-PCR (Figure 2). Transgenic strains containing the empty plasmid pPD95_77 were also obtained the same way as described above, but no reporter gene expression was observed by RT-PCR.

The upstream motif 2 is composed of two sub-motifs with sequence and spacing similar to that of the A and B boxes of the tRNA promoter. To address the transcriptional activity of UM2, two constructs were made. One (CeN37_100) was made from a fragment containing 130 bp upstream sequence and 56 bp transcribed sequence from the snoRNA locus CeN37, the other (CeN55_100) from a fragment containing 134 bp upstream sequence and 89 bp transcribed sequence from the snoRNA locus CeN55. RT-PCR with one primer located in the CeN37 and CeN55 fragment and the other primer in the plasmid sequence demonstrated the expression of transgenes CeN37::GFP and CeN55::GFP, respectively (Figure 2).

Upstream motif 3 is exclusively found at stem-bulge RNA (sbRNA) loci. To determine the transcriptional activity of UM3, 140 bp and 120 bp of the upstream sequence of the sbRNA loci CeN74-2 and CeN72, respectively, was cloned in front of the GFP ORF in pPD95_77 to yield constructs CeN74-2_100 and CeN72_100 (the constructs also included 32 and 36 bp of the respective transcribed sbRNA sequences, fused to the GFP reporter gene). RT-PCR verified the expression of both reporter fusions (Figure 2).

To assay whether the sequence upstream of each motif might contain additional regulatory elements, we also constructed plasmids containing approximately 300 to 1000 bp fragments of the 5'-flanking sequence from the CeN7, CeN37 and CeN74-2 loci, respectively. Transgenes expression levels were assayed by quantitative RT-PCR (qRT-PCR), and for each construct, 2 to 5 transgenic lines were tested.

In the case of UM1, qRT-PCR of RNA from transgenic strains showed that the expression driven by CeN7_300 (containing a 264 bp 5' flanking fragment) was 4–5 fold higher than expression driven by CeN7_100 (Figure 3A), strongly suggesting the existence of an enhancer located within the 90~264 bp upstream region. However, a search for 5' end features in this region of PSE/UM1 ncRNAs in C. elegans failed to yield any common motifs (see Methods for details). The expression driven by CeN7_1k showed an almost identical level to that of CeN7_300, indicating that no additional regulatory elements exist within the 264~1000 bp upstream of the ncRNA locus. In the case of UM3, no significant increase in expression was found by increasing the length of the upstream fragment to 245 bp (CeN74-2_300; Figure 3B). Similarly, qRT-PCR indicated no increase in expression of the UM2 locus when the upstream region was extended to 294 bp (CeN37_300) and 989 bp (CeN37_1k; Figure 3C). These observations suggest that for the two assumed RNA polymerase III-driving promoters (i.e. UM2 and UM3) most or all the promoter activity resides within 100 bp of the 5'end of the respective annotated loci.

Each of the three upstream motifs contains two sub-motifs whose base pairs are particularly invariant among loci. To investigate the contribution of the PSEA and PSEB sub-motifs to the overall PSE/UM1 transcriptional activity, we mutated each of the two sub-motifs mainly by converting each purine and pyrimidine residue to its opposite purine and pyrimidine, respectively (i.e. A to G, and vice versa), in the context of CeN7_1k. The corresponding constructs were labeled as CeN7_A_mut, CeN7_B_mut and CeN7_(A+B)_mut. Mutation of either of the two sub-motifs (PSEA and PSEB) reduced expression to 3~7 % of that of CeN7_1k, suggesting that both PSEA and PSEB are required for the transcription of PSE/UM1 loci (Figure 3A).

The sub-motifs in UM2 promoter strongly resemble the A and B box motifs of the tRNA internal promoter. Mutation of either of these two motifs in the context of CeN37_1k caused a strongl reduction in the expression levels of the mutant constructs (18 % for CeN37_A_mut and 26 % for CeN37_B_mut compared with CeN37_1k; Figure 3B). Concomitant mutation of both sub-motifs (CeN37_(A+B)_mut) reduced the expression level to 7 % compared with CeN37_1k.

Of the two sub-motifs in UM3, one strongly resembles the PSEB of the PSE/UM1, whereas the second sub-motif has the consensus sequence GTATA. We mutated both motifs in the same fashion as described above, and observed the effect on expression in the context of CeN74-2_300. Abolishing the PSEB-like sub-motif (CeN74-2_A_mut) reduced the expression level to 49 % compared with that of non-mutated CeN74-2_300. Mutation of the GTATA sub-motif produced an even more modest reduction (79 % compared to that of non-mutated CeN74-2_300) than PSEB. Mutating both sub-motifs simultaneously, however, resulted in near abolishment of expression of the reporter (CeN74-2_(A+B)_mut; Figure 3C). The results suggest a certain redundancy in regulatory activity between the two sub-motifs of UM3, but also that at least one of the sub-motifs must be present for expression of the downstream locus to occur at an appreciable rate.

5' RACE performed on non-mutated constructs found that the TSS of CeN7::GFP and CeN74-2::GFP were identical to 5' ends of the endogenous RNAs. For CeN37::GFP, however, the TSS was apparently located 70 bp upstream of the 5' end of mature endogenous RNA and 9 bp upstream of UM2 sequence itself (Figure 4A).

As shown above, 5' RACE of the reporter transcript of a UM2 construct (CeN37_1k) found that transcription was initiated 70 bp upstream of the 5'end of the mature endogenous RNA, and even 9 bp upstream of the UM2 sequence. In plants, dicistronic tRNA-snoRNA transcripts, which are subsequently cleaved to yield mature tRNA and snoRNA have been described 32. To investigate whether the UM2 is also internally located with regards to the primary transcript of the endogenous RNAs, we visually inspected the tiling microarray data 1 for possible evidence of transcription of the UM2 site itself. Although we found no indication of expression of the UM2 sequence at the CeN37 locus, we found other 21 instances with indications of some level of expression, 16 at known snoRNA loci and five preceding unannotated TUFs. In addition to CeN37, we therefore selected five candidate loci (CeN50-2, CeN39, CeN53, CeN55 and CeN119) for analysis by 3' RACE as indicated in Figure 5. If the dicistronic tRNA-snoRNA model also applies to the UM2-snoRNA transcripts, two RACE bands should be expected, one corresponding to the UM2-snoRNA primary transcript, the other to the UM2 fragment remaining after cleavage. Four (CeN39, CeN53, CeN55 and CeN119) of the 6 loci yielded RACE fragments of length expected if transcription initiation occurred at the start of, or upstream of the UM2 sequence, and sequencing confirmed a joint UM2-snoRNA transcript in all four cases. A band corresponding to a smaller fragment was observed in three cases, but none of the sequences included the UM2 fragment. Thus, the part of the primary transcript that contains the UM2 sequence is either rapidly degraded after cleavage, or the UM2-containing fraction of the primary transcript is removed by 5'exonuclease digestion during maturation of the snoRNA. 5' RACE analysis showed that the transcription of the UM2 loci starts 9~13 bp upstream of the first base in the box A-like sub-motif.

Transcripts derived from UM2 and UM3 loci are usually terminated at a run of several consecutive thymidine (T) residues, which is a property of loci transcribed by RNA polymerase III. The minimal number of consecutive T residues sufficient for termination of RNA polymerase III transcription varies among organisms, but analysis of known UM2 and UM3 loci suggests 4 consecutive Ts are sufficient for termination in C. elegans (Figure 4B). This is similar to what has been found in human and mouse, but is less than the 5–7 normally needed for termination in the yeast species (Figure 4B) 14. In the plasmid pPD95_77 there are runs of four (or more) Ts located at variable distances from the plasmid multicloning site (MCS). To determine the actual termination sites of the UM2 and UM3 constructs, we performed 3'RACE on the resulting reporter transcripts. For CeN37_1k (UM2) the reporter transcript terminated at the first, second and third "TTTT" tracts (located 122, 177 and 236 nt downstream of TSS; Figure 4C). For the CeN74-2_300 (UM3) construct, reporter transcript termination was identical to what has been observed in CeN37_1k (Figure 4C). qRT-PCR of fragments corresponding to the different termination sites suggested that most of the UM2 and UM3 reporter transcripts terminated at the first "TTTT" tract (Figure 4D).

3'RACE of the UM1 reporter transcript gave no specific result, but RT-PCR indicated that at least a fraction of the GFP ORF was included in the transcript. We further investigated the expression of GFP in worms from 2–3 independent transgenic strains from each of the different constructs. No GFP expression was observed after genetic transformation with any of the UM2 (CeN37) and UM3 (CeN74-2) constructs. However, marked GFP expression was observed under fluorescent microscope for the CeN7_1k and CeN7_300 transgenic worms, and even the CeN7_100 transgenic worms showed weak GFP expression. To eliminate the possibility that the observed GFP expression was an effect of this particular PSE/UM1 locus, we tested another PSE/UM1 ncRNA locus, CeN16-1, whose upstream sequence was also able to drive GFP expression. 5'-end RACE performed on RNA extracted from CeN7_1k, CeN7_300 and CeN16-1_1k transgene worms showed that the transcription start site of the reporters were identical to that of the wild type ncRNA loci, suggesting that the transcriptional activity resulting in the observed GFP expression was the same as that driving the transcription of the endogenous ncRNAs.

Since PSE/UM1 could drive expression of GFP, we used UM1::GFP fusions to analyse the temporal-spatial expression pattern of related ncRNA loci having this upstream motif. One group of such loci are the C. elegans SL2 RNAs, which are a nematode-specific group of ncRNAs that function in trans-splicing of operonic mRNAs. There are around 20 SL2 RNA loci in C. elegans with slightly variable sequence characteristics. As far as is known, all participate in the same function, i.e. joining of an additional "exon" to the 5'end of internal (or non-first) mRNAs in operonic loci, but nothing appears to be known about the background for the numerous and variable SL2 RNA loci in the C. elegans genome. Previous experiments have demonstrated that different SL2 RNA genes show different temporal expression patterns 6343537, but little is known about their spatial expression patterns. We therefore examined the temporal-spatial expression of five SL2 RNA genes using promoter::GFP fusion constructs. The tested ncRNAs showed considerable variation in expression both in time and space. CeN7 is principally expressed in hypodermal cells, which is in agreement with previous in situ hybridization result 35, but also showed expression in skin muscles, excretory cells, head and ventral neuron cells (Figure 6A &6B; Additional file 2). The CeN16-1 locus was also active in excretory cells, and showed strong expression in the pharynx (Figure 6C; Additional file 2). Two other loci, CeN6 and CeN11, showed marked expression in intestinal muscles near the tail (Figure 6D &6F; Additional file 2), however, their temporal expression pattern differed, CeN11 showing visible expression from larval stage 3, while CeN6 showed no expression until the mature adult stage. The fifth SL2 RNA locus, CeN19, is expressed in intestinal muscles near the vulva (Figure 6E; Additional file 2). The temporal expression pattern is generally identical with previously northern and microarray analysis 6343537. CeN7, CeN16-1 and CeN19 showed strong expression from an early stage of embryo development (Additional file 2) through the mature adult stage, whereas CeN6 and CeN11 both showed stage specific expression.

An interesting question is whether the transcriptional activity of the various SL2 RNA loci correlates with their target loci (i.e., the mRNAs to which the SL2 exons are spliced). To this end we downloaded spliced leader 38 and expressional data 1939. However, mRNA spliced to a specific SL2 RNA do not show any well correlated spatial expression patterns (Additional file 3), and even though it is possible to find SL2 RNA loci whose expression resembles one or a few of SL2 RNA 34 and mRNA temporal expression data 204041 did not show any significant correlation between specific SL2 RNAs and their target mRNAs (Additional file 3).

We also observed that mutations in the PSEA and PSEB sub-motifs greatly changed the expression patterns of the CeN7 locus constructs (i.e. CeN7_A_mut, CeN7_B_mut and CeN7_(A+B)_mut) compared with that of non-mutated constructs (i.e. CeN7_1k and CeN7_300). The CeN7_1k and CeN7_300 constructs were clearly expressed in hypodermal cells, skin muscles, excretory cells, head and ventral neurons (Figure 6A &6B; Additional file 2), however, the expression driven by CeN7 mutations was restricted to amphid and tail neuron cells (Figure 6G &6H). This suggests that the sequence characteristics of PSEA and PSEB are not only important for the general expression level of a locus, but may also influence where and when a locus is expressed.

Standard nematode cultivating conditions were used as described in ref 58. The strains used were N2 and UNC-76.

Individual PCR reactions were performed in 50 or 100 ul reaction volume using wild type C. elegans (N2 strain) genomic DNA as template. PCR products corresponding to the desired fragments were digested with enzyme HindIII and BamHI, and then cloned into the promoter-less vector pPD95_77 (kindly provided by Andrew Fire) upstream of the GFP reporter gene. The constructs were verified by sequencing. Mutations were performed using standard PCR procedures. The primers used in this work are listed in Additional file 4.

Reporter constructs were injected at 50 ng/ul together with 50 ng/ul transformation marker pRF4 [rol-6(su1006)] or unc-76. Stable lines of transgenic worms were established as described previously 59. For each construct, 2~5 transgenic lines were analysed.

RNA was extracted from transgenic worms according to Trizol (Invitrogen) protocol. RNA digested with DNase I (Fermentas) was used as template for RT-PCR (Qiagen one step RT-PCR kit). For qRT-PCR, reverse transcription was performed using primers 95_77_2_R and U2_R, and the cDNA was used as template for the qPCR according to qPCR mix protocol (Qiagen QuantiTect SYBR Green PCR Kit). When the expression levels of transcripts of different length were to be compared, a nested approach using internal primer yielding identical length PCR products were used for the qPCR step (following reverse transcription). The reactions were carried out on an MJ Research Opticon TM 2.

3'-RACE was performed as described 1 with minor modifications. Briefly, DNase I digested total RNA was ligated to the 3' end adaptor 3AD, and the ligated RNA was reverse transcribed into complementary DNA (cDNA) with a primer (3RT) complementary to the 3' adaptor 3AD. First round PCRs were performed by using primers 3RT and pPD95_77_1_F. Nested PCR was then performed, using 3RT and pPD95_77_2_F as primers, and the first round PCR products as template. The PCR products were analysed on a PAGE gel, and candidate bands were recovered and sequenced.

DNase I digested total RNA was reverse transcribed into cDNA with primers pPD95_77_2_R, pPD95_77_3_R and pPD95_77_4_R, respectively. The cDNA was treated with terminal DNA transferase (Fermentas) to add a 3'end poly(A) tail. First round PCR was performed with 3'CDS primer and the corresponding primer used for reverse transcription. Second round PCR was then carried out using the diluted first round PCR products as template, and 3'CDS and pPD95_77_1_R as primers. The second round PCR products were analysed, cloned and sequenced as above.

C. elegans genome annotation and sequence data were downloaded from Wormbase (version WS140) 39. The MEME motif discovery tool (version 3.0.13) 25 was used to search for conserved motifs within and upstream of the ncRNA loci. The search for common motifs acting as enhancer on 54 UM1/PSE ncRNAs was carried out on 100~300 bp upstream sequences of these loci.