All animal experiments were performed in accordance with national guidelines (articles R214–87 to R214–90 from the French “rural code”) and European guidelines (directive 2010/63/EU of the European Parliament and of the council of September 22, 2010 on the protection of animals used for scientific purposes). All protocols were approved by the ethics committee of the Pays de la Loire region, under protocol nos. CEEA 2009.14 and CEEA 2012.145, and performed at the required biosafety level. Animals were maintained under specific pathogen-free conditions in the animal house facility of Angers University Hospital, France (agreement A 49 007 002).

Mycobacterium ulcerans strain 01G897 was originally isolated from patients from French Guiana (De Gentile et al., 1992). Bacterial suspensions were prepared as previously described (Marsollier et al., 2007b; Marion et al., 2016b), with adjustment to 2 × 10⁵ acid-fast bacilli/ml for inoculation (50 μl) into the tail of 6-week-old females of the inbred FVB/N mouse strain (Charles River Laboratories, Saint-Germain-Nuelles, France).

RNA was extracted from infected tail skins 30 days post-infection, by the Trizol/chloroform method (Method 1), which co-extracts host and bacterial RNA, or by the differential lysis method (Method 2), optimized and adapted from that described by (Rustad et al., 2009), for the isolation of bacterial RNA alone.

(i) Sample preparation: Tail skin was excised from infected mice and immediately placed in a Petri dish containing a mixture of 1 ml Trizol (Ambion) and 1 ml RLT buffer (Qiagen) supplemented with 1% β-mercaptoethanol. Skin tissues were cut into smaller pieces, transferred to round-bottomed tubes and broken up with a TissueRuptor (Qiagen).

(ii) RNA extraction and purification: Samples were transferred to two bead beating tubes (0.1 mm glass beads, MoBio) and were shaken with TissueLyser (Qiagen) at 4°C for 5 min at 30 Hz. The samples were immediately placed on ice and centrifuged at 10,000 × g for 5 min at 4°C to remove cell debris. The supernatant was transferred to a 15 ml tube containing 1 ml Trizol. 200 μl chloroform/isoamyl alcohol (24:1) was added to the tube, which was then repeatedly inverted to mix and centrifuged at 10,000 × g for 5 min. The aqueous phase was transferred to a clean tube containing 400 μl chloroform/isoamyl alcohol and centrifuged again, as in the previous step. The aqueous phase was transferred to a clean tube containing 1 volume of 70% ethanol, and the tube was repeatedly gently inverted to mix. Total RNA was purified with the RNeasy Midi kit (Qiagen), with DNase treatment, according to the manufacturer’s protocol, and eluted in 100 μl RNase- and DNase-free water.

(iii) DNase treatment: Contaminating DNA was removed by retreating the RNA with DNase, for 45 min at 37°C, with the TURBO DNA-free kit (Ambion), according to the manufacturer’s protocol.

(iv) Eukaryotic RNA removal: To reduce the levels of contaminating host RNA from the samples, MICROBEnrich kit (Ambion) was used according to the manufacturer’s protocol.

(i) Sample preparation: Tail skin was excised from infected mice and immediately placed in a Petri dish containing 2 ml Tris-EDTA (TE) buffer (10 mM Tris-HCl, 1 mM EDTA). Skin tissues were cut into smaller pieces, transferred to round-bottomed tubes and broken up with a TissueRuptor (Qiagen). 4 ml TE buffer was added and the tissue homogenates were digested with 1 ml of a 20 mg/ml proteinase K solution (Qiagen) for 10 min at 55°C (without shaking). Another 6 ml of TE buffer was added and the samples were centrifuged at 3,200 × g for 15 min at 4°C. The pellet, which contained the bacterial cells, was resuspended in 300 μl Tri-reagent (Zymo Research) and 300 μl RLT buffer (Qiagen) supplemented with 1% β-mercaptoethanol.

(ii) RNA extraction and purification: The samples were transferred into a bead beating tube (0.1 mm glass beads, MoBio) and shaken with TissueLyser (Qiagen) at 4°C for 5 min at 30 Hz. The samples were immediately placed on ice and centrifuged at 10,000 × g for 5 min at 4°C to remove cell debris. The supernatant was transferred into a clean tube containing 1 volume of 100% ethanol, with which it was mixed by repeated gentle inversion. The RNA was purified and treated with DNase with the Direct-zol RNA MiniPrep kit (Zymo Research), according to the manufacturer’s protocol, and eluted in 50 μl of RNase- and DNase-free water.

(iii) DNase treatment: Contaminating DNA was removed by retreating the RNA with DNase for 45 min at 37°C with the TURBO DNA-free kit (Ambion), according to the manufacturer’s protocol.

(i) Quantification and purity analysis: Total RNA concentration and purity (A_260nm/A_280nm) were assessed with 1 μl of the RNA preparation, on a NanoDrop 1000^TM (Thermo Scientific) spectrophotometer. A ratio greater than 1.8 is usually considered to indicate satisfactory RNA purity (Imbeaud et al., 2005).

(ii) Quality and integrity analysis: RNA quality and integrity were assessed with the Experion automated electrophoresis system (Bio-Rad). The total RNA sample (Method 1) was diluted 10-fold and the Experion RNA StdSens analysis chip was used (quantification of 5–500 ng/μl RNA). With enriched bacterial RNA preparations (Method 2), the RNA was diluted twofold and the Experion RNA HighSens analysis chip was used (quantification of 100–5,000 pg/μl RNA). The RNA quality indicator (RQI) method returns a number between 1 (highly degraded RNA) and 10 (intact RNA) for each RNA sample (Imbeaud et al., 2005; Schroeder et al., 2006).

RT-qPCR targeting the M. ulcerans ppk gene was performed to detect mycobacterial RNA transcripts in RNA samples. RT-qPCR targeting the mouse gapdh gene was performed to evaluate the contamination of RNA samples with host RNA. The ppk and gapdh genes were selected as housekeeping genes for the RT-qPCR analyses for M. ulcerans and mouse, respectively. Amplification efficiency (Table 1) was determined from the slope of a standard curve of cDNA serial dilutions.

(i) Reverse transcription: The first-strand cDNA was synthesized in a reaction volume of 20 μl containing 2 μl of total RNA, 500 ng of random primers (Invitrogen) and the M-MLV reverse transcriptase (Invitrogen). Contaminating DNA in RNA sample was checked by performing a negative control with no reverse transcriptase (RT-) for each sample.

(ii) Quantitative real-time PCR: qPCR was performed in a reaction volume of 10 μl containing Absolute Blue qPCR mix (Thermo Scientific), 300 nM primers, 100 nM Taqman probe (Table 1) and 2.5 μl of diluted twofold dilution of cDNA/RT-. The sequences of the primers and probes used are provided in Table 1. Reactions were run on a AriaMx Thermocycler (Agilent), with the following program: 10 min at 95°C and 40 cycles of 10 s at 95°C and 1 min at 60°C. Each sample was analyzed in duplicate.

(i) Ribosomal RNA (rRNA) removal: rRNA was depleted from the mycobacterial total RNA preparations with the RiboZero Epidemiology Illumina kit, which removes eukaryote and prokaryote rRNA in a single step.

(ii) Preparation of RNA-seq libraries: The RNA-seq libraries were prepared with the TruSeq Stranded Total RNA LT Sample Prep kit (Illumina). The quality of all libraries was checked with the DNA-1000 kit (Agilent) on a 2100 Bioanalyzer and quantification was performed with Quant-It assays on a Qubit 1.0 fluorometer (Invitrogen).

(iii) RNA-seq: Clusters were generated for the resulting libraries, with Illumina HiSeq SR Cluster Kit v4 reagents. Sequencing was performed with the Illumina HiSeq 2500 system and HiSeq SBS kit v4 reagents. Runs were carried out over 65 cycles, including seven indexing cycles, to obtain 65-bp single-end reads. Sequencing data were processed with Illumina Pipeline software (Casava version 1.9). All 65-bp reads were aligned against the complete genome and plasmid sequences of M. ulcerans (Agy99 strain) obtained from the Burulist database with Bowtie software, and against the mouse genome (GRCm38) obtained from the Ensembl database with STAR software. Data were normalized and analyzed in R, with the Bioconductor packages.

For studies of the interactions between the host and the bacteria during the infection of mice with M. ulcerans, we initially tried to extract bacterial and host RNA together, from the tail skin of mice infected with M. ulcerans. The infected tissues were broken up with a TissueRuptor and subjected to chemical and mechanical lysis, to release the bacterial and eukaryotic RNA. The RNA was then subjected to Trizol/chloroform extraction and purified on a Qiagen midi column which allows the purification of up to 1 mg RNA (Figure 1).

With this approach, we obtained 200 to 300 μg of total RNA per sample (Figure 2C). Electrophoretic analysis revealed strong bands corresponding to the 18S and 28S rRNA of eukaryotic cells, with no bacterial RNA bands corresponding to 16S and 23S rRNA in samples (Figure 2A and Supplementary Figure S1). The electropherogram trace confirmed this result, as it contained peaks only for mouse rRNA (Figure 2B and Supplementary Figure S1). The mycobacterial RNA transcripts in samples were assessed by a reverse transcription-polymerase chain reaction (RT-qPCR) targeting the M. ulcerans housekeeping gene, ppk. The samples contained only small amounts of mycobacterial RNA, with Ct values of 32–33 (Table 2). The A_260nm/A_280nm ratio exceeded 1.8, confirming the purity of the RNA in each sample (Figure 2C), and electrophoresis showed a complete absence of RNA degradation, with an RQI value greater than 9 (Figure 2C). However, we were unable to assess the integrity of bacterial RNA, because no bacterial RNA was visualized.

To eliminate the host RNA from the samples, MICROBEnrich kit (Ambion) was used which selectively removes 18S, 28S rRNA, and polyadenylated mRNA. Following MICROBEnrich treatment, electrophoresis was performed and showed this treatment to have been poorly effective in our hands, because strong bands corresponding to the 18S and 28S rRNAs were still detected (Figure 2D). These data confirmed the predominance of eukaryotic ribosomal RNA in samples, and suggested that mouse mRNA would probably also predominate over bacterial transcripts. However, based on our qPCR demonstrating the presence of mycobacterial RNA in samples, we decided to attempt a deep sequencing approach (RNA-seq) on a sample before and after bacterial RNA enrichment with the MICROBEnrich kit. The obtained RNA-seq data showed an absence of reads aligned with the M. ulcerans genome, even though mycobacterial RNA transcripts were detected by RT-qPCR, a more sensitive technique. In this case, sequencing depth was insufficient to obtain reads mapping to the M. ulcerans genome, confirming that the removal of mouse RNA was not effective enough to achieve an enrichment of the sample in mycobacterial RNA. In conclusion, these data demonstrate that studies of the M. ulcerans transcriptome in vivo will require a method capable of significantly decreasing eukaryotic RNA levels, to facilitate the detection of mycobacterial transcripts by RNA-seq.

We developed an alternative technique for the purification of bacterial RNA from infected host tissues by differential lysis, to achieve our goal. The principal challenge in this method was the choice of a lysis system resulting in host-cell lysis without damage to the bacterial cells. We recently demonstrated that the mechanical disruption of tissues had no effect on bacterial viability (Marion et al., 2016b). We therefore disrupted the host cells by mechanical lysis and treated them with proteinase K, to digest and degrade the mouse tissue whilst leaving the bacterial cells intact. The intact bacterial cells were then separated from the lysate containing host RNA by centrifugation, and the pellet containing the bacteria was resuspended in lysis buffer and subjected to bead beating. The bacterial RNA was purified on a Zymo column, making it possible to purify up to 100 μg of RNA directly from samples in Tri-reagent, a much less time-consuming approach (Figure 1).

We obtained 150 to 500 ng of total RNA per sample (Figure 3C), about one-thousandth the amount obtained with Method 1 (Figure 2C). However, gel electrophoresis on the total RNA obtained with this method revealed the presence of bands corresponding to the 16S and 23S bacterial rRNA whereas the mouse 28S and 18S rRNA were not detected (Figure 3A and Supplementary Figure S1). This finding was confirmed by the electropherogram trace (Figure 3B and Supplementary Figure S1).

The enrichment of samples in mycobacterial RNA by the differential lysis method (Method 2) was evaluated by RT-qPCR. The efficacy of eukaryotic RNA removal during sample preparation with Method 2 was evaluated by comparing equivalent fractions of enriched (Method 2) and non-enriched (Method 1) RNA samples. We quantified the expression of both bacterial and mouse genes in the samples obtained with the two methods.

No PCR product was obtained if the amplification was performed without prior reverse transcription (RT-). There was therefore no detectable DNA contamination. A comparison of ppk transcription levels showed that the transcript of this gene was present in both cases, with similar Ct values (30–33). An analysis of murine gapdh gene transcription revealed a difference in Ct of 15 between Method 1 and Method 2 (Table 2), corresponding to the presence of five orders of magnitude less host RNA with Method 2 than with Method 1. These results clearly demonstrate the efficacy of Method 2 for removing eukaryotic RNA. This method was, thus, considered suitable for the enrichment of samples in M. ulcerans RNA. The purity and integrity of bacterial RNA were assessed by determining the A_260nm/A_280nm ratio and the RQI value, respectively. The A_260nm/A_280nm ratio was greater than 1.8 (Figure 3C) and RQI values ranged from 7.3 to 9 (Figure 3C), indicating that the RNA was of sufficiently high quality for use in subsequent experiments, including transcriptional analyses.

Given the promising nature of these findings, we performed deep sequencing (RNA-seq) on the enriched samples. The removal of rRNA is a crucial step in RNA-seq, because rRNA signals can prevent adequate coverage of the bacterial transcriptome. We therefore used the Ribo-Zero gold rRNA Removal Kit (Epidemiology, Epicentre) to remove both mouse and bacterial rRNA before producing the cDNA library. The RNA-seq data confirmed that less than 5% of the reads mapped to rRNA, demonstrating the efficacy of the rRNA removal kit. The datasets for samples M2-1, M2-2 and M2-3 obtained by the optimized method (Method 2) contained 9.5, 7.6 and 7.9 million reads aligning with the M. ulcerans genome, respectively, whereas no reads aligned with this genome sequence were detected with the samples obtained by Method 1 (Figure 4). This enrichment procedure therefore provides sufficient coverage of sequences mapping to the M. ulcerans, because a minimum of 2 to 5 million reads from a ribosomal RNA-depleted library is required to provide adequate coverage of the gene expression profiles of bacteria in RNA-seq experiments (Rienksma et al., 2015).

Finally, two biological replicates, M2-1 and M2-2, were used to evaluate the reproducibility of the gene expression profiles obtained with RNA-seq technology. Spearman’s coefficient of correlation between samples (r = 0.9813) indicated a similar overall pattern of relative gene expression in the biological replicates, indicating that Method 2 was reproducible (Figure 5). This optimized method can therefore be used to study the mycobacterial transcriptome in a mouse model.