Non-Coding RNAs are Differentially Expressed by Nocardia brasiliensis in Vitro and in Experimental Actinomycetoma

Bacterial Strain and Culture Conditions

The strain N. brasiliensis HUJEG-1 (ATCC700358) was used. Bacteria were grown on brain heart infusion (BHI) agar (BD-bioxon, cat. no. 210700) for eight days at 28°C. A suspension equivalent to 3X10⁹ colony-forming units (C.F.U.)/mL was prepared as an inoculum. Two strains of Escherichia coli (XL1-BLUE and Stellar^TM) were used as competent cells for cloning experiments.

Induction of Murine Actinomycetoma

Ten male CD-1 mice (25 g and eight weeks of age) were used in experiments. Animals were managed according to the Research and Ethics Committee Code of the Faculty of Medicine, UNAM (Project 011-2012). We followed the 8^th Edition of Guide for the Care and Use of Laboratory Animals when caring for and using the mice in our study [13]. To induce N. brasiliensis infection, 50 µl of the previously prepared suspension were inoculated in the murine plantar pad. After bacterial inoculation, mice were maintained for two months at room temperature (between 24 and 28°C) and were provided food and water ad libitum. After this period, the murine actinomycetoma was clinically evident (by swelling and distortion of the foot, sinus tracts and drainage of purulent exudate), and mice were sacrificed by cervical dislocation [13]. The plantar pads were washed, and under sterile conditions, purulent exudate was completely extracted and deposited into sterile 1.5 mL tubes on ice. This material was washed three times with sterile saline solution and recuperated by centrifugation.

RNA Next-Generation Sequencing

For total RNA extraction, one hundred milligrams of purulent exudate (formed mainly by bacterial micro-colonies, i.e., grains) were macerated in the presence of liquid nitrogen until obtaining a fine, whitish dust. To purify RNA that is highly enriched for small RNAs species (mainly 200 nt or less), the mirVana^TM PARIS^TM Kit (Life Technologies Corporation, cat. no. AM1556) was used following the manufacturer´s instructions. The quality of this RNA preparation was verified in a denaturing acrylamide gel (8 M urea (Sigma, cat. no. U1250) and 12% acrylamide (Bio-Rad, cat. no. 161-0146)). Five hundred nanograms of purified RNA were used to synthesize cDNA and subjected to next-generation sequencing in duplicate (LCSciences, Texas, USA) using an Illumina HiSeq 2500 platform and samples were run in the 50 cycle SE read configuration for 36 bp single-end reads. In this system the adapters (5´ and 3´) are ligated to the RNA and after an RT reaction is used to create single stranded cDNA which is then PCR amplified using the TruSeq Small RNA Sample Prep kits. Sequences were mapped against reference databases, such as miRBase, to separate sequences corresponding to mouse transcripts. Non-mapped sequences in miRBase were grouped as no-hits for mouse and were mapped against the N. brasiliensis HUJEG-1 genome. As a control, an N. brasiliensis culture grown on BHI agar and incubated at 28°C for eight days was used. This culture was also processed to obtain and sequence the small RNAs using the same conditions as mentioned above including 500 ng of sRNA as starting material.

Rapid Amplification of cDNA Ends (RACE)

RACE is a PCR-based method used to obtain the full-length sequence of the 5´ and 3´ ends of a transcript and therefore its size [14]. For this procedure, cDNA was obtained from purified ncRNAs using random primers following the manufacturer’s instructions (Clontech, cat. no. 634859). The transcripts of the 5’ and 3’ ends were amplified using the corresponding specific oligonucleotides (Table S2). The amplicons were cloned using the Thermo Scientific CloneJET PCR Cloning kit (Fermentas, cat. no. K1231). The plasmids were recovered using the PureLink Quick Plasmid Miniprep kit (Life Technologies Corporation, cat. no. K210010) and sequenced.

Quantitative Reverse Transcription-Polymerase Chain Reaction (qRT-PCR)

To confirm the presence of the transcripts found by RNA-Seq, qRT-PCR was performed using the NCode™ miRNA First-Strand cDNA Synthesis and qRT-PCR kit (Life Technologies Corporation, cat. no. MIRQ-100) according to the manufacturer´s instructions. As starting material we used 1 µg RNA obtained from infection or in vitro growth conditions. Briefly, a poly A tail was added to RNA of sizes <200 nt using the poly A polymerase and incubated at 37°C for 15 min. The polyadenylated RNA was used for cDNA synthesis using the SuperScriptIII Reverse Transcriptase, and the reaction was incubated for 50 min at 50°C. Finally, the synthesized cDNA was used for qRT-PCR reactions using the Platinum SYBR Green qPCR SuperMix-UDG with the corresponding oligonucleotides (Table S2). The cycling programme consisted of 50°C for 2 min; 95°C for 2 min; 40 cycles of 95°C for 15 seconds and 65°C for 60 seconds using the Step One Plus^TM System (Life Technologies Corporation, cat. no. 4376600). For relative quantification we used the method ΔΔ Ct. The C_T values from all qRT-PCR reactions were obtained and analysed in triplicate to confirm the existence of ncRNAs. 5S rRNA was used as endogenous standard to normalize the expression levels of the tested sRNA candidates.

In silico Analysis

The genome of N. brasiliensis was analysed to predict regulator ncRNAs using the non-coding RNA characterization (nocoRNAc) software, which incorporates several procedures to detect transcript characteristics. For example, for the detection of the transcript´s 3’-end, nocoRNAc uses the TransTermHP program, which predicts Rho independent end signals. In addition, transcript´s 5’-end are predicted by the detection of destabilized regions in the genomic DNA [15].

Induction of Murine Actinomycetoma

To induce infection, 10 male CD-1 mice (eight weeks of age, 25 g) were inoculated in the plantar pad with a N. brasiliensis suspension as described in Materials and Methods. Two months after bacterial inoculation, mice exhibited typical signs of actinomycetoma Fig. (1A) and were sacrificed. Under sterile conditions and by a skin-deep incision with a scalpel, the purulent material containing numerous grains (consisting of bacterial cumulus) was recovered. The presence of grains was verified by light microscopy Fig. (1B). sRNAs were run in a denaturing 12% poly-acrylamide gel to verify its quality (Fig. S1). The absorbance ratios are showed in (Table 1).

Global Non-coding RNA Profiles in N. brasiliensis

To detect the expression of ncRNAs in N. brasiliensis the transcriptome of < 200 nt isolated from actynomycetoma and from N. brasiliensis grown on BHI agar for eight days, was subjected to next-generation sequencing on an Illumina platform for 36 bp single-end reads. An outline of the procedures used in this study is presented in Fig. S1. The sequences showing no hits in the mouse genome (32,062,154) were subjected to a BLAST (basic local alignment search tool) search against the N. brasiliensis genome resulting in 11,024,619 reads. A Student T-test between reads obtained from actinomycetoma and N. brasiliensis grown in vitro was done for the in-depth analysis (p < 0.05 values) to determine non-coding RNA candidates. Under this statistical criterion, 70 sequences present in both conditions were identified as potential ncRNAs because they were localized in IGRs or corresponded to antisense of N. brasiliensis genes. Table (2) presents the localization of 18 of these sequences that were mapped in the N. brasiliensis genome and were chosen due to their differential expression in both conditions. Twelve out of these 18 transcripts are encoded in IGRs, three transcripts corresponded to tRNA-derived fragments, and one was overlapped between an IGR and the 5´end 23S rRNA. Finally, two transcripts mapped in the antisense direction of two genes, one of which corresponded to a putative non-ribosomal peptide synthetase and the other to an OmpR family two-component response regulator.

A searching of ORFs by the ORF finder program (NCBI) was performed in order to confirm these small RNAs are not encoding small proteins. After this analysis we found only Nbnc-8 encoding a possible small peptide having 11 amino acids, but it did not show any known protein similarity and its nucleotide sequence is only present in N. brasiliensis and N. seriolae.

The Expression of Most ncRNAs was Increased in vivo Compared with in Vitro

Table (3) presents the differential expression values under in vivo and in vitro conditions of 10 transcripts selected from Table (2). These transcripts were selected based on the highest fold changes between two conditions. The differential expression change of these RNAs was confirmed by qRT-PCR Fig. (2). The size of these transcripts was determined and RACE, as shown in Table (3). The transcripts Nb-tRFthr, Nb-tRFmet, Nbnc-3, Nbnc-5, Nb-tRFcys, Nbnc-10 Nb-rRF_23S exhibited increased expression in vivo compared with in vitro. The remaining ncRNAs (Nbnc-4, Nbnc-7 and Nbnc-8) exhibited an increased expression in vitro compared with in vivo conditions. Nbnc-3 and Nbnc-5 were upregulated in vivo by RNA-Seq; however results were inconsistent by qPCR.

tRNA-Derived Fragments were Highly Expressed in Vivo

In contrast to tRNA derived fragments in other organisms including archeae and humans that are derived from the 5´end, the Nocardia tRNA fragments belonging to threonine, methionine and cysteine tRNAs (Nb-tRFmet, Nb-tRFcys, and Nb-tRFthr) identified here are derived from the 3´ end of the tRNAs Fig. (3) [17]. After the analysis of diverse Nb-tRFthr clones we found that this transcript sequence corresponds to a fragment overlapping the N. brasiliensis tRNA^Thr and it ends 10 nt downstream (coordinates 8,659,485-8,659,534 in N. brasiliensis genome).

A Fragment Derived from 23S rRNA (Nb-rRF_23S) is Distributed in Different Actinomycetes

Among the transcripts that were mainly expressed under in vivo conditions, we found Nb-rRF_23S. Surprisingly, this fragment was located within three of the nine copies of the N. brasiliensis rRNA operon overlapping a part of an IGR and the 5´ end of the 23S rRNA gene Fig. (4). In contrast to most of ncRNAs detected that range in size from 50 to 300 nt, the transcript size was 466 nt, as determined by RACE experiments.

To investigate the phylogenetic distribution of this transcript in different actinomycetes, the BLAST algorithm was used. We identified the presence of a corresponding sequence in pathogenic and non-pathogenic bacteria belonging to the Actinomycetales order, including the Nocardia, Mycobacterium, Rhodococcus and Tsukamurella genera. The most conserved region obtained in the alignment was used to construct a phylogenetic tree, which is shown in Fig. (5). With the exception of Rhodococcus equi 103S, a tendency to form independent clades corresponding to each genus was observed. In several Nocardia and other actinomycetes species, the presence of a corresponding sequence has been observed in more than one copy of the rRNA operon.

Prediction of ncRNAs Using Bioinformatic Tools

The secondary structure of N. brasiliensis ncRNAs was predicted using the mfold programme [16]. Results of this application are showed in Figs. S3-S11.

As an alternative approach to searching for ncRNAs in N. brasiliensis, an in silico analysis was performed using the nocoRNAc programme [15]. According to this software, N. brasiliensis contains 146 ncRNAs genes (Table S1), 106 of which are encoded in IGRs and 40 are antisense-localized RNAs of genes with putative function, including three lipases and three oxidoreductases likely involved in virulence.

Interestingly, by comparing the ncRNAs predicted by the nocoRNAc programme and those found by RNA-Seq analysis, four sequences (4 of the 70 RNAs differentially expressed) were found in common Fig. (6), and these are localized to IGRs. The expression of three of these ncRNAs was increased in the in vitro conditions (including Nbnc-4), whereas one ncRNA was expressed at similar levels in both conditions. The low number of ncRNAs shared by both methods could be due to potential false-positive predictions of the nocoRNAc programme. Moreover, the expression of many other ncRNAs is probably dependent on other experimental conditions, which were not the purpose of this work.

Ten potential N. brasiliensis ncRNAs (showed in Table 3) of known sizes were subjected to the IntaRNA programme in order to predict the corresponding mRNAs target [18]. This programme enables the prediction of interactions in three regions of the mRNA target: start codon, coding sequence, and stop codon. The data from this analysis are shown in Tables S3-S6. The ncRNA Nbnc-3 (Tables S3A-C) was predicted to interact with the start codon of a Mce-family protein, with the start codon and coding sequence of a mycolyltransferase involved in Rhodococcus equi‘s virulence [19], and with the stop codon of a carboxylesterase, a known M. tuberculosis virulence factor [20]. Moreover, Nbnc-10 (Tables S4A-B) was predicted to interact with the start codon of a lipase and an epoxide hydrolase, both P. aeruginosa virulence factors [21]. Nbnc-10 also interacted with the stop codon of a glycosyl transferase that is essential in Streptococcus pneumoniae lung infections [22].

Furthermore, Nb-tRFcys (Tables S5A-C) was predicted to interact with the start codon of several mRNAs related to virulence, including dipeptidyl aminopeptidase [23], sodium hydrogen exchanger [24], TetR family of transcriptional repressors [25], and aldehyde dehydrogenase [26]. In addition, Nb-tRFcys was predicted to interact with the coding sequence of a dipeptidyl aminopeptidase, and with the stop codon of the protocatechuate 34-dioxygenase subunit alpha [27], a lipase [28], and an amidase [29]. Finally, Nb-rRF_23S was predicted to interact with the stop codon of a chloramphenicol resistance protein.