Mycoplasma hyopneumoniae transcription unit organization: genome survey and prediction.

Mycoplasma hyopneumoniae is associated with swine respiratory diseases. Although gene organization and regulation are well known in many prokaryotic organisms, knowledge on mycoplasma is limited. This study performed a comparative analysis of three strains of M. hyopneumoniae (7448, J and 232), with a focus on genome organization and gene comparison for open read frame (ORF) cluster (OC) identification. An in silico analysis of gene organization demonstrated 117 OCs and 34 single ORFs in M. hyopneumoniae 7448 and J, while 116 OCs and 36 single ORFs were identified in M. hyopneumoniae 232. Genomic comparison revealed high synteny and conservation of gene order between the OCs defined for 7448 and J strains as well as for 7448 and 232 strains. Twenty-one OCs were chosen and experimentally confirmed by reverse transcription-PCR from M. hyopneumoniae 7448 genome, validating our prediction. A subset of the ORFs within an OC could be independently transcribed due to the presence of internal promoters. Our results suggest that transcription occurs in 'run-on' from an upstream promoter in M. hyopneumoniae, thus forming large ORF clusters (from 2 to 29 ORFs in the same orientation) and indicating a complex transcriptional organization.

Introduction

Mycoplasma hyopneumoniae is the key aetiological agent of porcine enzootic pneumonia (EP) and is a major contributor to porcine respiratory disease complex.[1] EP is a chronic respiratory disease that mainly affects finishing pigs. Although major efforts to control M. hyopneumoniae infection and its detrimental effects have been made, swine production has suffered from significant economic loss.

Like other mycoplasmas, M. hyopneumoniae has a small genome with limited biosynthetic potential.[2-4] Up until now, four M. hyopneumoniae genome strains have been sequenced.[2-4] Despite the significant amount of data produced by genome sequencing, information on its open read frame (ORF) cluster (OC) organization, transcriptional unit (TU) formation and transcriptional regulation is very limited. Genes organized in an ORF cluster (OC) or TU are arranged in tandem in the genomic sequence, being delimited by the location of the upstream promoter and the downstream transcriptional terminator. TU identification is commonly based on bioinformatics approaches combined with experimental data. Mycoplasma studies have demonstrated the occurrence of polycistronic mRNA and TUs in some genomic regions.[5-12] However, information on characterized TUs is available for exceptional cases, making it difficult to judge the accuracy of current in silico TU or OC prediction methods in mycoplasma genome sequences.

Prediction and recognition of mycoplasma promoter elements are poorly developed. The unusual A + T content of intergenic regions (IRs) in mycoplasma[13] and the weak −35 consensus[14] prevent efficient prediction of promoters by current bioinformatics approaches. Moreover, in spite of the development of artificial transformation and other genetic tools for some species of Mycoplasma,[14-17] these methodologies are yet to be developed for M. hyopneumoniae. Therefore, strategies for the mapping of genome organization and gene transcription in M. hyopneumoniae can help understand regulatory mechanisms and form the basis for future work on mapping promoters and terminators.

Information regarding transcription promoter or terminator sequences have yet to be defined in M. hyopneumoniae. Therefore, any analyses at the level of transcription regulation are difficult. Recently, Gardner and Minion[18] demonstrated that transcription occurs in IRs of M. hyopneumoniae, indicating that M. hyopneumoniae does not strictly control transcription termination. An analysis of the average RNA folding energy near stop codons suggested that no stem-loops are formed in Mycoplasma genitalium and Mycoplasma pneumoniae TUs, indicating the existence of qualitatively different and uncharacterized mechanisms for transcription termination.[19] Defining a transcription termination region in mycoplasmas has revealed some contradictory results. In some operons of M. genitalium,[8] the presence of stem-loop structures have been found, and results from Hoon et al.[20] suggest that the Rho-independent terminator represents the main mode of transcriptional termination in mycoplasmas. However, it has been suggested that genes are arranged in long clusters in mycoplasmas, indicating the possibility that the starting point of transcription occurs in the first gene and stops at low frequencies due to the absence of specific regions for transcription termination.[21]

Although gene co-regulation occurs in mycoplasmas, to date only little information is available on ORF organization in the M. hyopneumoniae genome, as well as its relationship with pathogenicity. In order to establish differences between the gene organization of pathogenic and non-pathogenic M. hyopneumoniae and also to relate this gene organization to transcription regulatory mechanisms, we systematically analysed the synteny and conservation of ORF order among three M. hyopneumoniae genomes (two pathogenic strains: 7448 and 232; one non-pathogenic strain: J). We then performed the experimental characterization of some OCs predicted in the M. hyopneumoniae 7448 strain. This is the first report where gene organization analysis was systematically performed and compared in three M. hyopneumoniae genomes, two of which are pathogenic to swines and a significant number of predicted OCs are experimentally demonstrated and named TUs. These results could contribute to promote the understanding of M. hyopneumoniae transcription regulation.

Materials and methods

In silico analysis of TUs

OC prediction was performed by Artemis Release 10.5.2 software.[22] The criteria for the manual examination of possible OCs in M. hyopneumoniae genomes (MHP7448_GenBank: NC_007332; MHP232_GenBank: NC_006360 and MHPJ_GenBank: NC_007295) were established on the occurrence of clusters with two or more genes in tandem in the same DNA strand. This was performed by systematic annotation comparison of the protein sequences encoded in all ORFs from the analysed genomes. According to the complexity of adjacent ORF rearrangements, two groups were created: OCs characterized by the presence of two or more ORFs in the same DNA strand until the occurrence of ORFs in the opposite strand, and monocistronic (mC) group representing single ORFs. Differences in the annotation among the genomes were evaluated by comparing protein sequences using the NCBI/BLASTP program.

The comparative analysis of OC organization found among the three M. hyopneumoniae genomes (7448, 232 and J) was accomplished based on the M. hyopneumoniae 7448 genome results by Artemis Comparison tools Release 9.0.[23]

Primer design

The primers were designed in Vector NTI Advance 10 (Invitrogen) based on the M. hyopneumoniae 7448 genome sequence (GenBank: NC_007332). Primer pairs were chosen to enable synthesis by reverse transcriptase–PCR (RT–PCR), whose products span each IR, demonstrating the occurrence of polycistronic mRNA and confirming the prediction previously made. The OC7448111 (Fig. 1) was used as an example where three primer pairs were designed for utilization in the RT–PCR analysis. As shown in Fig. 1, primer pairs were distributed in the predicted OC and used in the following combination in the RT–PCRs: 1°F/1°R; 2°F/2°R; 3°F/3°R; 1°F/2°R and 2°F/3°R. To generate experimental data, a total of 71 primer pairs were projected and employed with this methodology (Supplementary Table S1).

Figure 1.

OCs in M. hyopneumoniae. A series of RT–PCRs using ORF-specific primers are designed to produce the products shown at the centre of the figure (black lines with the product size in bp). The black arrows 1°F/1°R, 2°F/2°R and 3°F/3°R correspond to primers 111 1°F, 111 1°R, 111 2°F, 111 2°R, 111 3°F and 111 3°R, respectively, and represent a schematic localization of primers pairs in OC7448111 designed to span the IRs between ORFs. The ORFs and tRNA30 nomenclature is in agreement with the GenBank NC_007332 and are represent by grey arrows.

DNA and RNA manipulations

Mycoplasma hyopneumoniae strain 7448 was grown in 5 and 25 ml of Friis broth[24] at 37°C for 48 h for DNA extraction and RNA isolation, respectively. Genomic DNA of M. hyopneumoniae was isolated using standard protocols.[25] The quality of the isolated DNA was verified by gel electrophoresis and displayed pure high-molecular-weight DNA.

Total RNA of M. hyopneumoniae extraction was performed with Trizol® (Invitrogen) following the manufacturer's guidelines, including 1 U/μg DNase/RNAse free (Promega) treatment, and then the extracted RNA was quantified in Qubit™ system (Invitrogen) and also analysed by gel electrophoresis. The RNA extraction was stored at −70°C.

Reverse transcriptase–PCR

A first-strand cDNA synthesis reaction was conducted by adding 1 µg of total RNA to 132.5 ng of pd(N)6 random hexamer (GE Healthcare) and 10 mM deoxynucleotide triphosphates. The mixture was heated to 65°C for 5 min and then incubated on ice for 1 min. First-strand buffer (250 mM Tris–HCl, pH 8.3, 375 mM KCl, 15 mM MgCl2), 0.1 M dithiothreitol, 40 U RNase inhibitor (Invitrogen) and 200 U M-MLV RT (Moloney Murine Leukemia Virus Reverse Transcriptase—Invitrogen) was then added to a total volume of 20 µl. The reaction mixture was incubated at 25°C for 10 min and at 37°C for 50 min followed by 15 min at 70°C. Negative control was prepared in parallel, differing only by the absence of the RT enzyme.

PCRs included 1 U Taq DNA polymerase (Cenbiot Enzimas), 10× buffer (100 mM Tris–HCl, pH 8.5, 500 mM KCl), 1 mM magnesium chloride, 100 mM deoxynucleotide triphosphates, 10 pmol of each primer (Supplementary Table S1) and 1 µl of the first-strand cDNA reaction in a final volume of 25 µl. Negative control of RT–PCR was prepared in parallel, which differed only by the absence of cDNA, and no genomic DNA was added to the reaction mixture for the PCR control. PCR positive control was prepared using the genomic DNA of M. hyopneumoniae as a template. The PCR conditions were as follows: 1 cycle at 94°C for 5 min was followed by 30 cycles of 94°C for 30 s; denaturation and extension temperature and time varied according to each primer pair (Supplementary Table S1). The final extension step was at 72°C for 10 min. Reaction products were analysed in 1.2% agarose gels and fragments of the expected size were precipitated with tRNA (Invitrogen) as follows. To each RT–PCR (22 µl), 1 µg of tRNA and 2.5 V of absolute ethanol (4°C) were added and then incubated for 16 h at −20°C. After 30 min of centrifugation and wash with 70% ethanol, the pellets were resuspended in 15 µl of ultrapure water and sequenced with the DYEnamic ET Dye Terminator Cycle Sequencing Kit (Amersham Biosciences) in the MegaBACE 1000, as recommended by the manufacturer.

Results and discussion

Genes in M. hyopneumoniae are preferably organized in OCs

This study represents a global assessment of OC organization with in silico and in vitro analysis of the M. hyopneumoniae genome. Our detailed analysis indicated that genes are organized preferably in OCs. The gene-by-gene genome organization of the three strains of M. hyopneumoniae was analysed, gene localization was compared to detect ORFs with order conservation and two groups were created as described in Materials and methods: the OC group (named OC74481–OC7448117 in the 7448 strain; OCJ1–OCJ117 in the J strain and OC2321–OC232116 in the 232 strain) and the mC group (named mC74481–mC744834 in the 7448 strain; mCJ1–mCJ34 in the J strain and mC2321–mC23236 in the 232 strain). In summary, 117 OCs and 34 mCs (strains 7448 and J) and 116 OCs and 36 mCs (strain 232) were detected (Table 1). The mC ORFs of the three M. hyopneumoniae strains are listed in Supplementary Table S2. Surprisingly, just 5% of the ORFs in the three genomes were mC. Therefore, the majority (95%) of M. hyopneumoniae ORFs are transcribed in long polycistronic mRNAs, thus forming extensive OCs. Moreover, the ORF numbers that constitute each OC was very variable (Table 2). As shown in Table 2, the majority of the OCs were composed of two ORFs (22%), with increasing numbers of up to 29 ORFs. Interestingly, in the three genomes analysed, approximately 85% of the OCs were composed of two to eight ORFs, demonstrating the presence of OCs with fewer numbers of ORFs.

Table 1.

General feature of the OCs’ organization in M.hyopneumoniae genomes

Feature	MHP7448	MHP J	MHP232
Total length (base pairs)	920 079	897 405	892 758
Total no. protein coding ORFs	657	657	691
Total no. of predicted OCs	117	117	116
Total no. of mC (% compared with total ORFs)	34 (5.1%)	34 (5.1%)	36 (5.2%)
Valid ORFs	17	12	17
Conserved hypothetical ORFs	8	7	10
Hypothetical ORFs	9	15	9
OCs organization similarity (%)a	100%	84%	71.00%
OCs organization similarity (%) without RIsb	100%	85.0%	82.0%
Location of region with inverted sequence (bp)	101 560–354 533	Not present	104 691–344 079
Transposases + ICEH-IIc	9 + 1	7 + 0	5 + 1
Validated predicted OCs (% compared with total predicted OCs)	21 (18%)	0	0

aSimilarity evaluated in relationship with MHP7448. The percentage represents the total OCs conservation (equality ORFs and RIs).

bSimilarity evaluated in relationship with MHP7448. The percentage represents the total OCs conservation leave out RIs differences.

cProbably relationeted with sites of recombination in comparative analysis of OC organization between 7448 strain and 232 or J strains.

Table 2.

ORF composition in OCs of M. hyopneumoniae genomes

Feature	MHP7448	MHP J	MHP232
Total no. of OCs	117	117	116
OCs composed of two ORFs	26	26	25
OCs composed of three ORFs	19	18	16
OCs composed of four ORFs	16	14	16
OCs composed of five ORFs	15	16	13
OCs composed of six ORFs	9	12	9
OCs composed of seven ORFs	11	10	10
OCs composed of eight ORFs	5	5	5
OCs composed of nine ORFs	2	1	4
OCs composed of 10 ORFs	3	4	6
OCs composed of 11 ORFs	3	4	2
OCs composed of 12 ORFs	2	2	2
OCs composed of 13 ORFs	1	1	3
OCs composed of 15 ORFs	1	1	2
OCs composed of 20 ORFs	1	0	0
OCs composed of 21 ORFs	0	0	1
OCs composed of 22 ORFs	1	0	0
OCs composed of 24 ORFs	0	1	0
OCs composed of 26 ORFs	1	1	0
OCs composed of 27 ORFs	0	0	1
OCs composed of 29 ORFs	1	1	1

Genome minimization can be explained by reducing the number of genes or by packing them into strings of clustered genes. In the genus Mycoplasma, during genome reduction both processes seem to have occurred. In M. hyopneumoniae, the genome has been reduced by a loss of genes as those related with the ability to formylate Met-tRNAi[3] among others. Moreover, as shown in Table 2, the density of genes in OC74487 and OC7448102 could be explained by the genome compaction. Our results in Table 2 are in agreement with those reported by Rogozin et al.,[26] suggesting that, in diverse organisms, the majority of gene clusters are formed by a strings of two to four genes. Therefore, this arrangement found in the three M. hyopneumoniae strains could be the result of the maintenance of the size of the gene clusters during genome reduction.

High level of synteny of the OCs is found among M. hyopneumoniae strains

Detailed analysis of each OC revealed a high level of synteny between the OCs defined for the 7448 and J strains as well as for the 7448 and 232 strains. Gene order was conserved with 85% similarity when comparing the 7448 and J strains, and 82% when comparing the 7448 and 232 strains (Table 1, Supplementary Tables S3–S5). Previous comparative analysis of these three genomes was performed at a genomic scale and strain-specific differences were demonstrated, but high rates of conservation were detected in regions of rearrangements, which were probably involved in pathogenesis.[3] In agreement with Vasconcelos et al.,[3] our work also demonstrated that the prevalence of ORFs transcribed in the DNA strand with conserved OCs reflected the high level of synteny among the strains' genomes. This occurs not only due to the same arrangement of ORFs but, more interestingly, also due to the presence of IRs of the same size. Although most of the strains display high synteny, low-degree conservation regions also occur. The analysis of genome regions with lower synteny evidences differences in ORF organization and size of intergenic distance. Some cases with common differences are shown in Fig. 2 and are related to: (Fig. 2A) ORFs that are separated into different OCs (for example OC23246, OC744810 and OC744847, Supplementary Tables S3 and S5); (Fig. 2B) addition or removal of OCs (see OC23202 and mC23201, mC744801 and mC744802, Supplementary Tables S2, S3 and S5); (Fig. 2C) OCs with addition or removal of ORFs, generally encoding hypothetic proteins (for example OC744806 and OC23207 or OC7448100 and OCJ100, Supplementary Tables S3–S5).

Figure 2.

Comparison of the OCs’ organization in three M. hyopneumoniae strains. (A) Homologous ORFs that are separated into different OCs in MHP7448 strain. (B) Addition or removal of OCs: the homologous ORF is mC in 7448 strain and is polycistronic in 232 strain. (C) OCs with addition or removal of ORFs, generally encoding hypothetic proteins. The bars connecting the arrows represent close matches. White arrowheads represent the ORFs and indicate the direction of transcription.

In M. hyopneumoniae, the overall distribution of ORFs within the OCs is highly variable in the number of ORFs and size of the intergenic distances, as well as in the functional categories of the encoded products. Approximately 92% of the OCs display some variations related to these parameters (see Supplementary Tables S3–S5). However, some of the OCs are composed of ORFs whose products are related functionally, some examples are: OC744822, OC744843, OC744886, OC7448105 and OC7448115, all OCs with orthologs in the 232 and J strains (Supplementary Tables S3–S5). Taken together, these results suggest that the classical definition of operon cannot be applied to M. hyopneumoniae genomes.

Despite the high level of synteny between the genomes of M. hyopneumoniae 7448, J and 232, strain-specific differences, such as inverted regions and rearrangements, were demonstrated, being potentially related to pathogenesis.[3] The inverted segment present in the 232 strain was further analysed and compared with the cognate 7448 region, taking into consideration the ORF string organization and OC distribution (Fig. 3). In M. hyopneumoniae 232, this region corresponds to genome positions from nucleotides 104 691–353 186 bp (OC23211–OC23246), with a total length of 248 495 bp. In M. hyopneumoniae 7448, the equivalent location is from nucleotides 101 560–354 533 bp (OC744810–OC744846), with a size of 252 973 bp. In this location, the overall conservation of gene order and OC distribution were preserved (Fig. 3). Typical examples of this organization are OC744828 and OC23228, each containing 29 ORFs. Other examples are OC23215 and OC744841 with 12 conserved ORFs each (see Supplementary Tables S3 and S5). A closer inspection of the regions bordering these inverted genome segments showed that, in each case, OC conservation was also preserved (see OC23210 and OC74489). Therefore, the genome was inverted but the ORF organization was maintained. Moreover, at the right border side, the distribution of the ORF order was maintained, even when OC23246 split from OC744810 and OC744847. It should be noted that the inversion event did not change the relative ORF order in the OCs. Taken together, these results demonstrate the similarity in ORF strings among the three strains, further supporting the hypotheses of Chen et al.[27] Price et al.[28] and Roback et al.[29] that the conservation in OC or TU organization in two or more phylogenetically related species is an important criterion for OC and after TU determination. Therefore, our results suggest that comparative genomics is a plausible approach for predicting OCs in mycoplasma genomes.

Figure 3.

A schematic diagram of the inverted region and rearrangement from M. hyopneumoniae 232 and 7448 strain. The bars connecting the maps represent close matches. The genomic limits of these regions are shown by arrows (OC23211, OC23246, OC744810 and OC744846). The other detached OCs are discussed in the text. Visualization by the software Artemis Compare.

Some of the differences in ORF strings found among the three M. hyopneumoniae strains could be related to the presence of transposases or to the integrative conjugative element.[3] Although the number of transposases is almost similar in all the three genomes (Table 1), their localization in some cases, probably, could be involved in the generation of instable regions where most changes occur, leaving the rest of the genome stable. The integrative conjugative element (ICEH), related to genome rearrangement and probably pathogenesis, is present in the genomes of M. hyopneumoniae 7448 and 232 strains but have a different organization.[3] The ICEH is located within OC744870, together with ORFs usually associated with bacterial conjugative elements and ORFs encoding hypothetical products, and within OC23286 with a similar organization.

Experimental analysis of OCs cotranscription

Aiming to validate some of the OCs predicted by our in silico analysis, 21 OCs encompassing 115 ORFs from M. hyopneumoniae 7448 were further characterized, corresponding to 17% of the total ORFs in the genome and 18% of the predicted OCs (Table 1 and Supplementary Table S6). The selection of these OCs was initially based on functional category variation among the ORFs and OCs with ORFs with long IRs. Usually, experimental validation of TUs is often established by measuring the length of mRNAs through Northern analysis. However, this approach failed to yield reliable transcript sizes in our experiments (data not shown). This experimental difficulty was also reported for other mycoplasmas and is probably due to the large size and instability of the transcripts in this genus.[7,8] Therefore, we performed gene-specific RT–PCR strategy to amplify the IRs between the ORFs of the predicted OC7448 for experimental validation. Briefly, as shown in Fig. 1, we designed primer pairs that would only amplify a product if adjacent genes could be found on a single mRNA molecule. In the example shown in Fig. 1, OC7448111 encompassed four ORFs and one tRNA gene and amplicons were generated from: MHP7448_0653 and ksgA (amplicon 263 bp), ksgA and rpsL (amplicon 706 bp), rpsL and tRNA30 (amplicon 357 bp), MHP7448_0653 and rpsL (amplicon 1555 bp), ksgA and tRNA30 (amplicon 1157 bp), indicating that a polycistronic mRNA spanning from MHP7448_0653 to tRNA30 is produced.

The experimental data from RT–PCR analysis of M. hyopneumoniae 7448, using primer pairs to each gene spanning the junctions between genes, demonstrated that the ORFs localized within each OC were linked. Therefore, 18% of the OCs characterized in silico in the 7448 strain were validated, corresponding to 21 of 117 OCs (Supplementary Table S6). To better demonstrate these results, three segments of the M. hyopneumoniae 7448 genome encompassing 63.2 kb (Fig. 4A), 27.5 kb (Fig. 4B) and 21.4 kb (Fig. 4C) are shown. The first genome region analysed contained nine OCs (OC7448109–OC7448117) with a total of 44 ORFs (Fig. 4A), representing the typical ORF string distribution as previously described: (i) ORF number: two ORFs in OC7448110, OC7448113, OC7448114, OC7448115 up until 11 ORFs in OC7448109; (ii) ORF product: OC7448114 products from the same functional category and OC7448112 products not functionally related and (iii) ORF string conservation: OC7448112 with conserved gene order in OCJ112 and OC232111; OC7448113 with same gene order as OCJ113 and OC232112 (Supplementary Tables S4–S6).

Figure 4.

Representation of experimentally analyzed OCs in M. hyopneumoniae genome. Some regions experimentally tested (A–C) are detached from the overall picture of the genome (labelled from 1 to 920 079 bp). The position on the chromosomal sequence is indicated in bp below both termini of the bars, with white arrowheads indicating the direction of transcription. (B and C) The size of some intergenic regions are marked (IR: followed by the number). (A) Genome coverage of 63.2 kb showing nine OCs. (B) Genome coverage of 27.5 kb showing three OCs. Two intergenic regions (IR: 711 bp and IR: 2757 bp) are indicated. (C) Genome coverage of 21.2 kb showing three OCs and two mC. The intergenic region (IR: 299 bp) in OC744890 represent the sequence between tuf and lon genes. Visualization by the software Artemis.

Some reports have suggested that genes co-transcribed as TUs or operons are likely to be compactly arranged on the genome with short IRs, although there are cases where regulatory elements can cause variations in the length of TUs.[30,31] Mycoplasma hyopneumoniae has a reduced genome with limited biosynthetic potential, and the presence of short intergenic sequences would help us to allow efficient transcription. Interestingly, the size of the IRs was extremely variable within different OCs (Supplementary Tables S3–S5). RT–PCRs with specific primers were applied to amplify all the IRs present in the 21 OCs and cotranscription was observed, even with large IRs (Fig. 4B, Supplementary Table S6). The OC744833 contained six ORFs (Supplementary Table S6) with a significant variation in the IR: (i) IR length of 2757 bp between glyA and nrdF genes; (ii) IR length of 2 bp between nrdF and nrdI genes; (iii) IR length of 76 bp between nrdI and nrdE genes; (iv) IR length of 89 bp between nrdE gene and an ORF encoding a hypothetical protein (MHP7448_0220) and (v) IR length of 711 bp between two ORFs encoding hypothetical proteins (MHP7448_0220 and MHP7448_0219).

Cotranscription of ORF strings in the same DNA strand

The genome region detailed in Fig. 4C illustrates the analysis to validate the criteria used in silico for OC definition, based on the presence of ORF strings in the same DNA strand until the occurrence of ORFs in the opposite strand. The ORF strings in OC744890 and OC744891 were interrupted by the presence of mC744825 in the opposite strand. For the RT–PCRs, two primer pairs were designed to evaluate the presence of a polycistronic transcript corresponding to the ORFs strings in OC744890 plus OC744891. The first primer pair (gyrF/gyrR, Supplementary Table S1) amplified a 941 bp DNA fragment of the MHP7448_0528_gyrA (mC744825). The other primer pair was localized within the coding sequence of first ORF (deo) of OC744890 and in the IR between OC744890 and OC744891 (gyr/deoC, Supplementary Table S1), potentially generating an amplification of 1338 bp. Although the presence of mRNA for the gyrA ORF was demonstrated, no transcript (single polycistronic mRNA) was observed, including both OC (OC744890 plus OC744891) ORF strings. Therefore, these results support the criteria used in silico for defining the OCs characterized by the presence of two or more ORFs in the same DNA strand until the occurrence of ORFs in the opposite strand. Taken together, these findings demonstrated that the typical gene organization present in other bacteria, as Escherichia coli or Bacillus subtilis, cannot be applied to M. hyopneumoniae genome.

Although in M. hyopneumoniae genes are arranged in long clusters, suggesting that the starting point of transcription occurs in the first gene, the transcription termination regions have yet to be defined. Aiming to contribute to the definition of the terminator sequences, one IR between OC7448112 and OC7448113 (see Fig. 1A) was analysed by RT–PCR. Primers pairs were designed to amplify the region between the ends of each OC (Supplementary Fig. S1). Our results demonstrate the presence of transcripts in this region, suggesting the absence of specific regions for transcription termination. These findings are supported by the observation of Gardner and Minion[18] that transcription occurs in IRs of M. hyopneumoniae, and that transcription run off the end of the last ORF.

The gene encoding the elongation factor Tu (tuf) is the last of five ORFs to constitute an OC (OC744890), expressed from a promoter upstream of the first ORF (deoC) (Supplementary Table S6, Fig. 4C). However, in the 299 bp IR between tuf and lon, a promoter-like sequence was found by primer extension analysis (data not shown). The OC744890 represents a typical OC in M. hyopneumoniae, with ORFs encoding products not related functionally as deoC (deoxyribose-phosphate aldolase), upp (uracil phosphoribosyltransferase), MHP7448_0525 (hypothetical protein), lon (heat shock ATP-dependent protease) and tuf (elongation factor Tu, EF-Tu).

EF-Tu is necessary for protein synthesis in metabolically active cells, being possibly involved in the synthesis of alternative transcripts in the M. hyopneumoniae 7448 strain. It is already known that, in TUs, discoordinate expression occurs when the product of an individual gene needs to be expressed apart from the others to mediate a different role.[31,32] The OC744890 has upstream and internal promoters resulting in two different sizes of clusters, indicating that the entire OC could be expressed from a common promoter or a subset of the OC (tuf) is separately transcribed in response to a specific signal. Previous reports have suggested a relationship between the expanded IRs and presence of internal regulatory elements in highly expressed TUs.[31,33] Güell et al.[12] in an M. pneumoniae transcriptome analysis demonstrated many alternative transcripts. The experimental evidence of the presence of regulatory elements such as internal promoters in the expanded IRs provided by the analysis of the tuf gene (IR of 299 bp between adjacent genes) and the variable size of the M. hyopneumoniae IRs analysed by RT–PCR suggest the presence of internal promoters in the IRs of M. hyopneumoniae OCs and a complexity in transcription regulation.

Conclusions

Different approaches have been used to analyse gene organization in bacterial genomes through the development of new algorithm[34] or prediction programs,[35,36] which are based on characteristics such as intergenic distance, closely related gene function, conservation of gene order in phylogenetically close genomes, or even the prediction of promoter region[37] and transcriptional terminators.[19,20] Although these algorithms have shown a great promise in terms of being able to predict operon structure with a high degree of specificity and sensitivity, the data that they rely on are available only for a small number of bacterial species[36] and their application to all bacterial species could present some difficulties. Furthermore, combinations of these prediction analyses with experimental data are scarce. Not surprisingly, analysis of OC prediction followed by experimental confirmation in mycoplasma genomes was not conducted prior to this study. Moreover, experimental data on TUs are also scarce.

Similar to M. hyopneumoniae genes, some M. pneumoniae genes are arranged in long OCs but only with short or almost no intergenic sequences.[12] Analysis of P97 and P102 paralog families have also suggested that a distance of up to 54 bp is an upper limit for defining the cotranscript structure in M. hyopneumoniae.[9] However, we have systematically mapped in silico all the OCs in three different strains of M. hyopneumoniae and experimentally validated 115 ORFs in 21 OCs with variably sized IRs in the 7448 strain. It has been proposed that intergenic distances between ORFs in OCs are similar in prokaryotes and can be used to predict OCs and to estimate the total number of OCs in genomes.[26,28,29,38,39] However, according to our analysis, the distance models do not seem to be applicable for OC prediction in M. hyopneumoniae genomes. Furthermore, the variability in size of intergenic distances within an OC can be related to the complexity of mechanisms of gene regulation. Therefore, the unusual distribution of intergenic distances between ORFs within OCs in M. hyopneumoniae may reflect a biological difference in the genome structure as previous suggested for Synechocystis.[28] Furthermore, transcription analysis across the IRs of the M. hyopneumoniae 232 genome demonstrated the presence of transcripts in the IRs and that transcription run off the end of the last ORF and RNA polymerase is gradually released from template.[18]

Taken together, these findings suggest that OCs are continuously transcribed until reaching the next OC or ORF present in the opposite strand. However, a subset of the OCs could be independently transcribed due to the presence of internal promoters. These new data suggest complex transcriptional organization in M. hyopneumoniae genomes and the probability of the occurrence of other not-yet-known mechanisms that are involved in transcription. Furthermore, the high synteny of gene organization between pathogenic and non-pathogenic strains indicates that pathogenicity is not related to gene organization. The accelerating pace of molecular research into this important pathogen is certain to provide additional data to refine the results described in this work, forming a solid empirical foundation for our future understanding of the relationships between gene organization and transcription regulation in M. hyopneumoniae.

Supplementary data: Supplementary data are available at www.dnaresearch.oxfordjournals.org.

Funding

This work was supported by grants from the Brazilian National Research Council (CNPq).