Experimental infections were performed at the Moredun Research Institute, UK as described previously (Laing et al., 2013). All experimental procedures were examined and approved by the Moredun Research Institute Experiments and Ethics Committee (MRI E46 11) and were conducted under approved UK Home Office license (PPL 60/03899) in accordance with the 1986 Animals (Scientific Procedures) Act.

The MHco3(ISE) strain is susceptible to all broad-spectrum anthelmintics (Roos et al., 2004) and was inbred to produce the material for the H. contortus genome sequencing project at the Wellcome Trust Sanger Institute (Laing et al., 2013). MHco4(WRS) originates from South Africa and is resistant to IVM and benzimidazole anthelmintics (van Wyk and Malan, 1988) and is also known as the White River Strain, while MHco10(CAVR) is an Australian strain otherwise known as the Chiswick AVermectin Resistant strain (Le Jambre et al., 1995) which is highly resistant to IVM but susceptible to benzimidazoles and levamisole. All strains were maintained at the Moredun Research Institute. Introgression of IVM resistance-conferring loci from both resistant strains into the genetic background of the susceptible strain MHco3(ISE) was carried out as described previously by Redman et al. (2012). In short, progeny from the crosses of the resistant strains with MHco3(ISE) underwent four rounds of back-crossing to MHco3(ISE), with IVM selection, to produce a fourth backcross generation: MHco3/4.BC₄ [MHco3(ISE) crossed with MHco4(WRS)] and MHco3/10.BC₄ [MHco3(ISE) crossed with MHco10(CAVR)]. In this manuscript the backcrossed strains are referred to as MHco3/4.BC₄ and MHco3/10.BC₄. For T. circumcincta, two strains were compared: a susceptible strain (MTci2) originally from Weybridge, that was isolated prior to the use of broad spectrum anthelmintics and a multi-drug resistant strain, MTci5, that shows decreased susceptibility to fenbendazole, levamisole, and IVM (Dicker et al., 2011). Female worms of these strains were available as frozen pellets from MRI.

For this study, MHco3/4.BC₄ and MHco3/10.BC₄ were used after an additional three rounds of selection with IVM (Oramec™) at 0.2 mg/Kg, with one exception. To re-establish MHco3/4.BC₄ and MHco3/10.BC₄, sheep were infected with 5,000 L3 of the appropriate line and treated with IVM at either 0.1 or 0.2 mg/Kg. At the first passage, no L3 were obtained from fecal cultures from the sheep infected with MHco3/10 BC₄ treated at 0.2 mg/Kg IVM. Consequently, for the next round of selection, the input L3 came from the animal treated with IVM at 0.1 mg/Kg. The origin and maintenance of MHco3/4.BC₄ and MHco3/10.BC₄ is described in data shown in Table S1.

Three sheep were infected per os with 5,000 L3 of each of the strains, MHco3(ISE), MHco4(WRS), MHco10(CAVR), MHco3/4.BC₄, and MHco3/10.BC₄ and were not drug treated. An additional three sheep per strain were infected with only the IVM-resistant strains MHco4(WRS), MHco10(CAVR), MHco3/4.BC₄, and MHco3/10.BC₄ and treated with IVM at 0.2 mg/Kg 7 days prior to euthanasia. Infected sheep were euthanized at 28 days post-infection. Surviving worms were isolated from the abomasum, washed extensively in PBS and picked into groups of 20 males or 20 female worms and immediately snap frozen in liquid nitrogen. RNA was prepared from each sample individually by grinding the pellet using a liquid nitrogen-cooled pestle and mortar and standard Trizol® protocol (Invitrogen). RNA quality and quantity was assessed using a spectrophotometer and Bioanalyzer (Agilent 2100 Bioanalyzer). 7.5 μg of male RNA and 7.5 μg of female worm RNA from paired samples (i.e., from the same sheep) was pooled and sent on dry ice to LC Sciences (Houston, Texas) for microarray. The arrays contained 609 predicted miRNA sequences from H. contortus (Winter et al., 2012), 369 sequences from C. elegans miRNAs (miRBase 16) and 50 control sequences. The array was probed with triplicate biological samples of pooled male and female RNA labeled with Cy5. miRNA expression was analyzed in different worm groups using two methods, ANOVA (Analysis of Variance) and t-test, and the degree of significance evaluated by comparing between-group variation with within-group variation. For an experiment involving more than two sample groups, ANOVA was first applied to produce a miRNA expression profile overview across all samples, then t-tests were performed to identify significantly differentiated miRNAs among all combinations of two groups.

To ascertain the expression profile of hco-miR-9551 throughout the H. contortus life cycle, data from a different microarray experiment were analyzed. This array contained exactly the same sequences as described above but the samples used to probe the array were all isolated from MHco3(ISE) worms, as follows: adult male, adult female, L4 isolated at day 7 post-infection, L3, in vitro activated L3 (exsheathed and cultured for 24 h in RPMI-1640 at 37°C and 5% CO₂ in air) and gut-enriched tissue (dissected from adult female worms). Biological triplicates of RNA were prepared for each sample as described previously, but in this experiment RNA was labeled with Cy3.

The microarray dataset supporting the expression of miRNAs in resistant vs. susceptible H. contortus can be found on NCBI GEO (https://www.ncbi.nlm.nih.gov/geo/): Series GSE79542. The microarray dataset showing the expression of miRNAs across the H. contorus life cycle can be found at (https://www.ncbi.nlm.nih.gov/geo/): Series GSE101501.

For qRT-PCR analyses, 1 μg of DNase I®-treated (Ambion) total RNA from female, male worms or L3 was used in polyadenylation and reverse transcription reactions (plus and minus enzyme) using the miRNA 1stStrand cDNA Synthesis® protocol (Agilent Technologies). Single RNA samples were also processed for qRT-PCR from female MHco3(ISE) worm gut tissue, from in vitro released eggs/L1 of MHco3 (ISE) worms or from parasite excretory-secretory (ES) products from MHco3/4.BC₄ and MHco3/10.BC₄ worms. qPCR was carried out following the miRNA qPCR Master Mix® protocol (Agilent Technologies) and results analyzed using MxPro qPCR Software, Version 4.10. Results were normalized to hco-miR-50, which was found by microarray to be present at similar levels throughout all the experimental conditions examined. The sequences of the HPLC-purified oligonucleotide primers (Eurofins) are as follows: hco-miR-50–5′-TGATATGTCTGGTATTCTTGG-3′; hco-miR-9551–5′-CACAGCATTTTACTGAGCC-3′. Agilent Universal Reverse miRNA primer® was used in all reactions.

Unless stated otherwise, qRT-PCR was performed on three biological samples and each PCR was carried out in triplicate. Lin-Reg® software (Ramakers et al., 2003; Ruijter et al., 2009) was used to calculate PCR efficiency using linear regression. These data were then processed by REST2009® software (qiagen.com) (Pfaffl, 2001; Pfaffl et al., 2002), which uses a mathematic model to account for any differences in PCR efficiencies between normaliser genes and genes of interest (http://rest.gene-quantification.info). This allows compensation for variations in starting material thus providing more accurate quantification of gene expression.

For this analysis two in silico 3′ UTR libraries were combined to generate a single searchable database. The first library contained 3′ UTR sequences predicted from the H. contortus transcriptomic data (Laing et al., 2013). This library contained 14,507 predicted 3′ UTRs from 11,264 genes and thus represented only ~52% of the genome (n = 21,799). The predicted 3′ UTRs ranged in length from 1–15,355 nt, although the majority were between 100 and 1,100 nt. 3′ UTRs of 50–1,000 nt were selected (11,328 sequences representing 8,992 genes) and combined with data from a second library that contained 1 kb of downstream sequence from each predicted gene in the genome, or sequence up to the end of a scaffold. As it was based on transcriptomic data, the first 3′ UTR database was considered the more reliable and sequences from the downstream database were used to “fill in” the missing 3′ UTRs.

The combined library, representing 3′ UTRs of 93% of all predicted H. contortus genes, was screened using the following online miRNA target prediction algorithms: PITA (Kertesz et al., 2007) (with default settings: seed sequence of 8 bases, ddG<−10³), miRANDA (Enright et al., 2003) (filtered for scores >145 and energies <−10), and RNAhybrid (Rehmsmeier et al., 2004) (p < 0.1 and energies <−22). Only 3′ UTRs that scored positive with all three programmes were retained. These were further analyzed to ensure that the binding sites predicted by each programme overlapped, resulting in a set of “high confidence” target mRNAs. These transcripts were then assessed using previously created RNAseq datasets generated from triplicate biological replicates of 20 pooled females worms of the MHco3(ISE), MHco4(WRS), and MHco10(CAVR) strains. Reads were mapped to the most current version (version 3) of the MHco3 H. contortus genome assembly (Laing et al., 2013) using Tophat2 (Kim et al., 2013). The following command line parameters were used: -I 40000 -r 200 -p 6 -a 6 -g 1 -N 5 –read-gap-length 3 –read-edit-dist 8 –no-discordant –no-mixed –min-intron 10 –microexon-search –mate-std-dev 40 –library-type fr-unstranded. A higher allowance for polymorphism between mapped RNAseq reads and the MHco3 genome assembly was allowed to ensure reads of other strains were not discarded due to the numbers of polymorphic sites. Mapped reads were counted using the htseq-count tool from HTSeq (Anders and Huber, 2010). DESeq2 was then used to call differential expression between strains of interest using default settings (Love et al., 2014). Transcripts that were down-regulated in both or one resistant strain compared to the susceptible MHco3(ISE) parental strain were identified. Transcriptomic data can be accessed via https://data.mendeley.com/datasets/mwvsjgt9jd/draft?a=2e2cfcc5-589f-4f93-ab6c-9330f92d13e8 (Rezansoff and Gilleard, unpublished).

Additionally, the top 400 predicted targets of hco-miR-9551 derived from PITA, miRANDA, and RNAhybrid were used in pathway analysis using DAVID bioinformatics resources (Huang da et al., 2009). C. elegans orthologs of the predicted H. contortus genes were identified and used to assemble a gene list which was entered into the online DAVID tool (https://david.ncifcrf.gov/) where this list was converted into associated biological pathways based on gene-annotation enrichment analysis. p values were adjusted for multiple testing to control the false discovery rate, using the Benjamini-Hochberg procedure (Anders and Huber, 2010) and pathways with a score of p < 0.05 were considered significant.

3′ UTRs from predicted hco-miR-9551 target genes HCOI00821400, HCOI00084600, and HCOI01910900 were amplified from H. contortus genomic DNA, isolated as described previously (Winter et al., 2013), using PfuUltra II Fusion HS DNA Polymerase (Agilent) and the following primer sequences (restriction sites underlined);

HCOI00821400 F Not I 5′GCGGCCGCCGTTGTTCCCATGCCACTTG-3′

HCOI00821400 R Not I 5′-GCGGCCGCGGATTCAAATCAGAAGC-3′

HCOI00084600 F Not I 5′-GCGGCCGCGCCCTACAATGTCTTCGCC-3′

HCOI00084600 R Not I 5′-GCGGCCGCGTGAGCAGGAAACAGC-3′

HCOI01910900 F Not I 5′-GCGGCCGCAGATTCAACATCCTCT-3′

HCOI01910900 R Not I 5′-GCGGCCGCTGATACTGGCTTTCCTCCA-3′

Products were cloned into pCR2.1-TOPO (Invitrogen) and then subcloned into the Not I site downstream of firefly luciferase in pMir-Target (Origene). Similarly, a 298 bp section of the hco-miR-9551 locus was amplified from H. contortous genomic DNA using the following primer sequences, mir-9551 F Kpn I 5′-GGTACCGTTACTTGCCGAT-3′ and mir-9551 R Kpn I 5′-GGTACCTGTCTGTCTCATCA-3′. This product was cloned into pCR2.1-TOPO and subcloned into vector pEGFP-N1 (Clontech) using Kpn I to generate plasmids in the forward and reverse orientation.

HEK293 cells were maintained as described previously (Winter et al., 2015). For transfections, 1 × 10⁴ cells/ml were seeded into the wells of 96-well plates in a volume of 100 μl using Dulbecco's Modified Eagle's Medium (containing 4500 mg/L glucose and sodium bicarbonate, Sigma D5671). Cells were transfected after 24 h when they were ~50% confluent using Lipofectamine LTX (Invitrogen) with 50 ng of a mir-9551-containing plasmid in either the forward or reverse direction, 6.25 ng of the relevant pMir-Target-derived plasmid, and 0.5 ng of phRG-TK (Renilla luciferase, Promega). Transfections were performed following the manufacturer's protocol for HEK293 cells. DNA was diluted to a volume of 20 μL using Opti-MEM I Reduced Serum Medium (Invitrogen) and 0.35 μL Lipofectamine LTX Reagent added. After incubation at room temperature for 30 min 20 μL of the DNA-Lipofectamine complex was then added directly to the cells in each well. Cells were grown for 48 h then analyzed using a Dual Luciferase Assay kit (Promega) following the manufacturer's protocol with six replicates used per test condition.

A microarray consisting of all identified H. contortus miRNAs (Winter et al., 2012) and all C. elegans miRNAs (miRBase 15) was constructed (LC Sciences). The array was probed with a 1:1 mix of RNA from male and female H. contortus worms recovered from three individual sheep per strain, yielding triplicate biological replicates of IVM sensitive strain MHco3(ISE) and IVM resistant strains MHco4(WRS) and MHco10(CAVR) and the relevant backcrosses, MHco3/4.BC₄ and MHco3/10.BC₄ (see section Materials and Methods for details and nomenclature). The backcrossed lines have been described previously (Redman et al., 2012) and had undergone three additional rounds of selection with IVM when used in the current study. For the four IVM-resistant strains, worms were also collected from three sheep per strain that had been treated with 0.2 mg/kg of IVM 7 days prior to worm collection.

Analysis of the microarray data demonstrated similar expression levels of most miRNAs between IVM resistant and sensitive strains. However, a single miRNA, hco-miR-9551, was found at ~6-fold greater abundance in the IVM-resistant parental strains MHco4(WRS) and MHco10(CAVR) compared to the susceptible parent. In the backcrossed lines, MHco3/4.BC₄ and MHco3/10.BC₄, hco-miR-9551 was also up-regulated 4- to 4.5-fold compared to IVM susceptible MHco3(ISE) worms (p < 0.05 for all samples; see Table 1, Figure 1). No other miRNA showed the same consistent up-regulation in every IVM-resistant strain and no miRNA was significantly decreased in all resistant strains (Figure 2). To investigate whether the expression of hco-miR-9551 was further induced by exposure to IVM, the same microarray was probed with RNA isolated from IVM-resistant worms recovered from sheep that had been treated with IVM 7 days previously. No significant difference in hco-miR-9551 expression levels was observed in any of the resistant strains compared to those recovered from untreated sheep (see Table 2 for details). Thus, hco-miR-9551 does not appear to be further induced in response to IVM.

As the probes for the array consisted of a mixture of male and female adult worms, we investigated whether the expression of hco-miR-9551 varied between sexes and life cycle stages. Additional RNA samples were prepared in triplicate from batches of male, female and L3 worms of all five strains and hco-miR-9551 levels measured by qRT-PCR relative to hco-miR-50, a constitutively expressed miRNA (see microarray dataset). The results for female worms broadly confirmed those from the original microarray; MHco4(WRS) and MHco10(CAVR) worms showed a 5- to 6-fold increase over MHco3(ISE) worms, while the backcross lines MHco3/4.BC₄ and MHco3/10.BC₄ showed ~a 2- to 3-fold increase in expression of hco-miR-9551 (Figure 3). In male worms and L3 larvae, levels of hco-miR-9551 were not significantly different between susceptible and resistant strains examined by qRT-PCR (data not shown). The results for these life cycle stages were much less consistent, reflecting the low levels of hco-miR-9551 in these stages (see Figure 4A). Thus, hco-miR-9551 appears to be a miRNA that is significantly up-regulated in female worms of all IVM-resistant strains tested in this study.

As qRT-PCR analysis suggested that hco-miR-9551 was highly expressed in female worms only, we investigated its expression in greater detail. Data were first analyzed from an identical microarray that had been probed with samples of RNA isolated from multiple life cycle stages of MHco3(ISE) worms. These included L3, in vitro activated L3 (exsheathed and cultured for 24 h, see section Materials and Methods for detail), L4 recovered at 7 days post-infection, male or female worms, or a preparation of gut tissue dissected from female worms (Marks et al. unpublished, but see microarray data set for details). This analysis confirmed that hco-miR-9551 was expressed in female worms of the susceptible MHco3(ISE) strain; male worms, L3, in vitro activated L3 or gut tissue did not express significant levels of hco-miR-9551 (see Figure 4A). The first signs of expression were evident in L4 stages recovered at 7 days post-infection, suggesting that this miRNA is developmentally regulated following infection of the mammalian host.

Expression levels of hco-miR-9551 were confirmed by qRT-PCR in triplicate biological replicates of male, female and L3 MHco3(ISE) worms. The data in Figure 4B show hco-miR-9551 levels relative to hco-miR-50 and confirm that hco-miR-9551 is essentially specific to female worms in the MHco3(ISE) strain with very low levels of expression in male worms or in L3. To determine if hco-miR-9551 is expressed in eggs developing within adult female worms, further qRT-PCR studies were carried out. No detectable signal was observed for hco-miR-9551 in a pool of eggs/L1 larvae collected from in vitro culture, nor was it detected in gut tissue isolated from single adult females, while a control miRNA could be amplified from all samples (data not shown). However, hco-miR-9551 was identified in a small RNA library prepared from excretory-secretory (ES) products of adult MHco3(ISE) worms and could be amplified by PCR from ES products of cultured adult worms of both backcrosses (Gu et al. unpublished data), suggesting it may be released from the secretory system and/or female gonad.

The hco-miR-9551 locus is located on scaffold_496 (Laing et al., 2013) and see WormBase ParaSite (http://parasite.wormbase.org). This scaffold is just over 140 Kb in length and encodes eight protein-coding gene models. Five of these gene models have predicted orthologs in C. elegans, all of which reside on the X chromosome, suggesting scaffold_496 is derived from the H. contortus X chromosome. All eight gene models are encoded on the reverse strand, while the hco-miR-9551 precursor sequence is encoded on the forward strand, lying 3,866 bp downstream (3′ end) of gene model HCOI01030600.t1 and 10,152 bp upstream (5′ end) of gene model HCOI01030500.t1 (See scaffold in Figure S1). Evidence in the literature has shown that that some miRNAs can affect the expression of genes in close proximity, providing a local negative feedback mechanism (van Rooij et al., 2009; Dill et al., 2012; Truscott et al., 2014; Yuva-Aydemir et al., 2015) and interestingly, both these genes are expressed at a lower level in MHco4(WRS) and MHco10(CAVR) than in MHco3(ISE) female worms, although the differences do not reach significance (data not shown). Searching for hco-miR-9551 binding sites in both genes demonstrated that HCOI01030600.t1 contained four potential sites in the 3′ UTR and one within the open reading frame. The predicted products of these genes are of unknown function and they have no conserved domains or orthologs in C. elegans. They do, however, have orthologs in other clade V parasitic nematodes including Oesophagostomum dentatum and Ancylostoma ceylanicum (for HCOI01030600.t1) and A. ceylanicum and Dictyocaulus viviparus (for HCOI01030500.t1).

There are no orthologs of hco-miR-9551 documented in miRBase, but scrutiny of the genomes of other clade V parasitic species identified an orthologous sequence in several species. In each case, RNA-fold analysis of the parasite mir-9551 using Mfold (http://unafold.rna.albany.edu/?q$=$mfold/RNA-Folding-Form) (Zuker, 2003), supports a hairpin structure with folding energies consistent with miRNA structure (data shown in Table S2). No such sequence was found in free-living nematodes, such as Caenorhabditis species and examination of parasitic nematode sequences from species belonging to clade I, III or IV, failed to identify any orthologs of hco-miR-9551. Thus, hco-miR-9551 appears to be specific to clade V parasitic nematodes belonging to the sub-order Strongylida. The only miRNA present in miRBase showing any degree of homology with hco-miR-9551 is miR-308-3p from various species of Drosophila, which shares the same seed sequence (nucleotides 2-7) with hco-miR-9551.

To investigate whether upregulation of miR-9551 occurred in other parasitic nematodes showing resistance to IVM, we focused on the related strongylid T. circumcincta. Strain MTci2 is susceptible to anthelmintic, while strain MTci5 shows resistance to IVM, fenbendazole and levamisole (Dicker et al., 2011). Expression levels of the T. circumcincta hco-miR-9551 ortholog, tci-miR-9551, were compared in the drug sensitive strain (MTci2) compared to the multi-drug resistant strain (MTci5). As can be seen from Figure 5, the expression of tci-miR-9551 was 16-fold higher in drug resistant T. circumcinta (p = 0.037).

Determining the function of a specific miRNA is difficult, even in organisms such as C. elegans, where genetic manipulation is feasible. In an attempt to define the biological roles of hco-miR-9551, we identified target mRNAs using computational programmes that search for miRNA binding sites. We screened an in silico 3′ UTR library using three separate programmes, miRANDA, PITA, and RNAhybrid to increase the likelihood of identifying true mRNA targets (see section Materials and Methods for details of settings). Only those 3′ UTRs scoring positive in all three programmes were considered further. The number of 3′ UTRs containing predicted hco-miR-9551 binding sites varied markedly from 2,255 using MIRANDA to 653 and 435 with RNAhybrid and PITA, respectively. In all, 73 UTRs were predicted by all three programmes to have hco-miR-9551 binding sites, of which 68 had overlapping binding sites predicted by all three programmes and therefore were considered “high confidence” targets (data shown in Table S3).

As miRNA expression frequently shows an inverse correlation with target mRNAs, we aimed to narrow down potential mRNA targets for further analysis, by scrutinizing a transcriptomic dataset prepared from adult female worms of the MHco3(ISE), MHco4(WRS) and MHco10(CAVR) strains (Rezsanoff, Laing and Gilleard, unpublished). Of the 26,064 genes in this transcriptomic dataset, a total of 694 were down-regulated in both MHc04(WRS) and MHc010(CAVR) female worms compared to MHc03(ISE). We focused on the expression profile of the 68 genes with predicted hco-miR-9551 binding sites in both resistant strains, MHc04(WRS) and MHc010(CAVR), compared to the susceptible strain, MHco3(ISE). Of these 68 genes, three showed significantly reduced expression in both resistant strains (Benjamini-Hochberg adjusted score p < 0.05) as listed in Table 3A. The expression of several other mRNAs was reduced in only one of the resistant strains compared to the IVM-susceptible female worms (see Table 3B). Of the three genes down-regulated in both resistant strains, HCOI00084600 is a predicted RNA-binding protein in both H. contortus and C. elegans, HCOI00821400 contains a ChaC domain with predicted gamma-glutamyl cyclotransferase (GGCT) function, while HCOI01910900 encodes a hypothetical protein. Two genes were significantly down-regulated only in MHco4(WRS) worms; these encoded a predicted protein with a 7-transmembrane G-protein-coupled receptor domain (HCOI00499800), the C. elegans homolog of which (ser-4) shows similarity to mammalian serotonin receptors. The second gene is a probable vacuolar ATPase (HCOI01954200). In MHco10(CAVR) worms, a collagen gene (HCOI02065400), a probable histone acetyltransferase gene (HCOI00706800) and a protein with a predicted glycolipid-transfer domain (HCOI0277200) were all significantly down-regulated. The putative C. elegans homologs of these genes and their RNAi phenotypes are shown in Tables 4A and B.

As the analysis described above relied upon the relatively stringent requirement for each mRNA target to be recognized by all three programmes, genes of interest could potentially be excluded. We addressed this issue by adopting a broader, overall view of the predicted targets to identify pathways enriched in the target gene data. The top 400 hits were selected from each of the three target prediction programmes, their orthologs identified in C. elegans and analyzed using the DAVID bioinformatic database. Allowing for genes that appeared in two or more lists and those which did not have orthologs in C. elegans, 485 WormBase identifiers were used to create a gene list, which was entered into the DAVID online tool. Functional annotation clusters were generated and significant pathways identified (Benjamini-Hochberg adjusted score p < 0.05, see Figure 6). Individual genes in each pathway are outlined in Table S4. Importantly, these included membrane/transmembrane associated genes, lipoproteins, metal-binding genes and transport-associated genes. Although P-glycoproteins are the classical drug transporters in nematodes and have been associated with drug resistance (Kerboeuf and Guegnard, 2011; Choi et al., 2017), these were not identified in the DAVID pathway analysis, suggesting involvement of a novel mechanism of resistance.

Three predicted hco-miR-9551/mRNA interactions from the transcriptomic analysis were further investigated by transfection of HEK293 cells in a dual luciferase assay. 3′ UTRs of HCOI00821400, HCOI00084600, and HCOI01910900 from H. contortus were cloned downstream of firefly luciferase and transiently transfected into HEK293 cells along with a plasmid containing hco-miR-9551 in the forward or reverse direction. The CHAC-domain containing protein HCOI00821400 contained two putative hco-miR-9551 binding sites in the 3′ UTR, while HCOI00084600 and HCOI01910900 contain one. This was determined using the PITA algorithm (Kertesz et al., 2007) using a minimum seed size of 8, allowing for a single G:U wobble and a single mismatch and results were refined by considering only those with PITA ΔΔG energies of < −7 (see Table 5).

Levels of luciferase in the presence of the forward or reverse hco-miR-9551 were compared. The 3′ UTR of HCOI00821400 showed a small (10.5%) but highly significant reduction (p = 1.9E-5) in expression in the presence of the forward hco-miR-9551, while there was no difference in expression for the 3′ UTRs of HCOI00084600 and HCOI01910900 when cells were transfected with the forward or reverse hco-miR-9551.The mean of three experiments comparing expression in the presence of the forward hco-miR-9551 relative to the reverse is shown in Figure 7.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.