- Top of page
- Material and Methods
- Literature Cited
To date, few organisms have been shown to possess variable ribosomal RNA, otherwise considered a classic example of uniformity by concerted evolution. The polymorphism for the 16S rRNA in Nosema ceranae analysed here is striking as Microsporidia are intracellular parasites which have suffered a strong reduction in their genomes and cellular organization. Moreover, N. ceranae infects the honeybee Apis mellifera, and has been associated with the colony-loss phenomenon during the last decade. The variants of 16S rRNA include single nucleotide substitutions, one base insertion-deletion, plus a tetranucleotide indel. We show that different gene variants are expressed. The polymorphic sites tend to be located in particular regions of the rRNA molecule, and the comparison to the Escherichia coli 16S rRNA secondary structure indicates that most variations probably do not preclude ribosomal activity. The fact that the polymorphisms in such a minimal organism as N. ceranae are maintained in samples collected worldwide suggest that the existence of differently expressed rRNA may play an adaptive role in the microsporidian.
MICROSPORIDIA have been controversial for a long time, with both puzzling eukaryotic and prokaryotic features. Although they lack mitochondria, it is now clear that they have a mitochondriate ancestry but have lost those organelles as part of a number of adaptations to the intracellular parasitic way of life. A strong genome reduction has occurred in most microsporidian lineages, up to extreme cases such as the human and vertebrate species parasites Encephalitozoon cuniculi and E. intestinalis with 2.9 and 2.3 Mbp, respectively (about half the amount of DNA of Escherichia coli), organized in 11 eukaryotic chromosomes (Keeling and Corradi 2011). Their relationship with Fungi has received several lines of evidence (Adl et al. 2012; Lee et al. 2008) and their basal position in the Fungi phylum has recently been demonstrated (Capella-Gutiérrez et al. 2012). Apparently, most Microsporidia reproduce only in a clonal way but a sexual cycle, including meiosis, exists in various lineages (Ironside 2007).
The microsporidian parasite Nosema ceranae has become a severe pathogen in the last decade: it infects the cells of the gut epithelium of bees and is associated with colony-loss phenomenon in Apis mellifera, causing major economic damage in the honey industry (Bromenshenk et al. 2010; Eischen et al. 2012; Higes et al. 2008, 2009). It appears that the original host of N. ceranae is the Asian honeybee, Apis cerana, and it is this microsporidium which has infected the European honeybee A. mellifera in the last decade (Botías et al. 2012; Fries 2010; Higes et al. 2006; Huang et al. 2006; Klee et al. 2007; Martín-Hernández et al. 2007). Moreover, N. ceranae has proved to be not only more virulent but also more resistant to environmental changes than Nosema apis (Higes et al. 2010), though further studies are needed to establish whether or not this is the cause of N. ceranae spreading colony loss in warmer geographical areas.
According to Cornman et al. (2009) the genome size of N. ceranae may have around 8 Mbp. In an overview of the genome, these authors noticed that the ribosomal loci were “recalcitrant” to the genome assembly parameters because they seemed to be very polymorphic. Intraspecific variability of DNA sequences has been described in a number of Microsporidia. Part of such diversity may be related with the biological properties of the different strains (Haro et al. 2006) and adaptation to specific hosts (Xiao et al. 2001). In a previous work, Sagastume et al. (2011) showed that the rDNA of N. ceranae is highly polymorphic, especially in the intergenic spacer (IGS), and the haplotypes found suggested that recombination is occurring in this species. The diversity also extended to the small subunit (SSU), 16S ribosomal RNA gene (the first 600 bp of this gene were then analysed) and included single nucleotide polymorphisms (SNP) and small insertions-deletions (indels).
Ribosomal RNA genes are usually assumed to have an almost perfect sequence identity among the different repeats existing in the genomes (Ueno et al. 2007). Indeed, sequence conservation is considered to be necessary for the finest translational efficiency (Van Spaendonk et al. 2001). A number of mechanisms, such as unequal meiotic recombination, sister chromatid exchanges, and gene conversion, lead to concerted evolution (Eickbush and Eickbush 2007). The birth-and-death model can also lead to a similarity of duplicated genes: new copies are formed by gene duplications, some may persist for long time periods, while others are lost by deletion or simply become pseudogenes; the identity is, most of all, a consequence of a strong purifying selection. In actual fact, the data obtained from a number of rRNA genes fit better to the predictions of the birth-and-death than those of the concerted evolution model (Nishimoto et al. 2008; Rooney 2004; Rooney and Ward 2005).
Nevertheless there are some cases where different functional rDNAs are maintained. In Streptomyces strains the heterogeneity of rDNA sequences can be divided in two groups evolving by different mutation-selection mechanisms (Ueda et al. 1999). Mylvaganam and Dennis (1992) proposed the two adjacent different rRNA functional operons in Haloarcula marismorthi could be the result of either a lost ability to homogenize the repeated sequences, or to a hypothetical chimerical origin of this archaeobacterium. Carranza et al. (1999) reported that two different copies of rDNA exist in all the species analysed from a family of flatworms (but not in other related families). The two types evolve at different rates, and most probably both of them are functional. The case of the microalga Prototheca wickerhamii is especially interesting: Ueno et al. (2007) observed not only multiple sequences of 18S rRNA but also that recombination occurs to produce new types. The most studied example of different ribosomes in the same species is that of Plasmodium berghei, as three types of transcribed rRNA genes are expressed in different stages of the parasite cycle (Gunderson et al. 1987). However, Van Spaendonk et al. (2001) demonstrated that those ribosomes are functionally equivalent, suggesting that the maintenance of additional rRNA units simply represent a gene dosage phenomenon, although some functional significance could not be excluded. The intragenomic polymorphism for the ribosomal SSU of the foraminifer Elphidium macellum has recently been suggested to be a consequence of hybridization between differentiated populations (Pillet et al. 2012).
The discovery of different SSU genes co-existing in the same strains of N. apis (Gatehouse and Malone 1998), Nosema bombi (O'Mahony et al. 2007), N. ceranae (Sagastume et al. 2011) and Nosema bombycis (Liu et al. 2013) is even more unexpected given the extreme reduction of Microsporidia genome stated above. Under strong selective pressure to keep only a minimum of functions, the maintenance of different ribosomes suggests some biological significance.
Keeping those facts in mind, our first aim was to determine if the polymorphism was occurring all along the SSU gene of N. ceranae. Secondly, it was essential to ascertain that the different SSU gene variants were actually expressed. In E. cuniculi rDNA clusters are repeated and located in the telomeres of the chromosomes (Katinka et al. 2001), although it is not known whether all of them are expressed or not (intrastrain variation has not been described). This opens the possibility that some rDNA repeats could actually be pseudogenes with no more than some structural roles in the microsporidian genome. Thus, the existence of nucleotide sequence variation has also to be analysed in the 16S RNA. Finally, if different 16S RNAs exist in the same N. ceranae strain, the positions of the polymorphic sites in the secondary structure may provide important clues about their hypothetical biological significance. Since the complete studies of the E. coli ribosomal RNA secondary and tertiary structures specific roles have been ascribed to different helices (Schmeing and Ramakrishnan 2009). Mutations in specific points can produce different consequences in ribosomal functions such as defects in the start codon selection (the most deleterious mutations directly related with the initiation factors bind), defects in aminoacyl-tRNA selection (missense and nonsense suppressor mutations) or increase +1 frameshift (for review see McClory et al. 2011).
- Top of page
- Material and Methods
- Literature Cited
To ensure the in vitro fidelity of PCR and the lack of recombination, which is known to occur under some experimental conditions (Judo et al. 1998; Lahr and Katz 2009; Meyerhans et al. 1990), we performed a preliminary experiment. Two clones of known sequence differing in 6 bp spread along a 1279 bp fragment were used as a template for the control amplification. One single de novo mutation was found among the total 37,770 nucleotides (30 clones × 1259 bp). Except for that single mutation, all the fragments fitted completely to one of the two original sequences, that is, under the conditions tested, no in vitro recombination was detected.
Five different sequences of the complete 16S rRNA gene were obtained from N. ceranae infected honeybees: SSU-2, SSU-3, SSU-7, SSU-8 and SSU-13 (GenBank accessions JX205149 to JX205152, and JX205129, respectively); the differences include nine point substitutions, plus four indels of a single nucleotide and one indel of the tetranucleotide GATT (Fig. 2). Four SNPs and the latter indel had also been found in a previous work with N. ceranae isolates from very different sources (Sagastume et al. 2011). It is important to note that, given the preliminary validation, the results cannot be attributed to PCR errors or in vitro recombination between slightly different sequences.
Figure 2. Haplotypes of the different 16S RNA sequences found. The positions of the alternative nucleotides or indels are given relative to the complete sequence including all possible insertions (hence it corresponds to a total of 1266 bp instead of 1259 bp in the reference sequence U26533. The first five haplotypes were obtained from the entire 16S RNA gene sequence, while the remainder proceeds from a fragment of 750 bp of the ribosomal RNA (nucleotides 52–801). Note that, at least in their common 750 bp, the haplotypes 1 and 11 are identical.
Download figure to PowerPoint
After the isolation of total ribosomal RNA and retrotranscription to cDNA, 19 clones containing 750 bp of the 16S RNA (nucleotides 52–801) were analysed, revealing 13 different sequences: one of them was found in five clones (which was also identical to SSU-8 clone above), another was found three times and the remaining 11 were unique findings (GenBank accessions JX205130 to JX205148). The haplotypes are shown in Fig. 2.
In the final alignment of all sequences, 27 variable points were detected: 19 substitutions (17 transitions and 2 transversions), four single nucleotide insertion-deletions (indel) in poly-nucleotide segments, and the frequent indel of the GATT tetranucleotide. All of them were named following the alignment positions. Considering that the variations are not randomly located (see below) and that a number of them were also detected in different studies and in expressed 16S rRNA, it is clear that not only different sequences of the SSU gene exist in the genome of N. ceranae, but different ribosomal RNA sequences are also being expressed.
The polymorphism indexes (PI), calculated with the program DNAsp 5.10 for the two groups of sequences – the five complete SSU and the 750 bp fragment – were very similar: 0.0032 and 0.0029, respectively, indicating that, in terms of diversity, the smaller fragment seems to be representative of the entire gene. The average number of nucleotide differences (k) between two 16S rRNA genes is four. The polymorphic sites tend to be located at specific regions. Figure 3 shows the variation of the PI along the sequence (a sliding window of 25 nucleotides was considered): it is noteworthy that the variable zones coincide for the two groups of sequences.
Figure 3. Variation of the Polymorphism Index (PI, y-axis) along the sequence of the 16S rRNA gene, excluding gaps (x-axis). Full line: SSU sequences; discontinuous line: cDNAs. The positions of the polymorphic indels are indicated with arrows.
Download figure to PowerPoint
To locate the variable points in the SSU, the secondary structure was redrawn from the model available for N. ceranae (Chen and Huang 2010). Figure 4 shows the overview of the complete SSU secondary structure and details the helices where polymorphic sites were detected. The localization of the polymorphic sites in the 16S rRNA structure and the putative functions of the correspondent regions are summarized in Table 1.
Table 1. Summary of Nosema ceranae 16S rRNA mutations and insertion-deletions, with the correspondent location in single (ss) or double (ds) stranded conformation of the secondary structure, remarking the changes produced in double stranded pair bases, and main known characteristics of each area. The mutations are numbered relative to the positions in the alignment
|Variable sites||Structural location||Structural characteristics|
|G46A||ss||Mildly deleterious mutation in Escherichia colia|
|G74A||ds: changing U-A to U-G||–|
|GATT indel 117||ss||Elbow connecting h8 to the main structure|
|A288G||ds: changing G-C to A-C||–|
|C309T||ds: changing U-A to C-A||–|
|C328G||ds: changing G-C to C-C||–|
|A329G||ds: changing G-C to A-C||–|
|T indel 475||ss: polyT zone||–|
|G480A||ds: changing A-U to G-U||–|
|C622T||ds: changing A-U to A-C||–|
|T647C||ds: changing C-G to U-G||Highly variable G-U zone|
|G659A||ds: changing A-U to G-U||Highly variable G-U zone|
|C661G||ds: changing G-U to C-U||Highly variable G-U zone|
|G662A||ds: changing G-U to A-U||Highly variable G-U zone|
|A indel 666||ss: polyA zone||Different loop in Nosema ceranae and Nosema apis|
|A668G||ss||Different loop in N. ceranae and N. apis|
|G678A||ds: changing U-A to U-G||Highly variable G-U zone|
|T indel 846||ss- polyT zone||–|
|T indel 939||ss: polyT zone||Different loop in N. ceranae and N. apis|
|A970G||ss||Moderately deleterious mutation in E. colia|
Figure 4. Secondary structure of the 16S rRNA of Nosema ceranae. The dark circles denote the polymorphic positions detected. Enlarged areas (boxed) correspond to the highest variable regions: G-U zone and GAUU indel region. The variable points were named according with the global alignment position.
Download figure to PowerPoint
- Top of page
- Material and Methods
- Literature Cited
As in most organisms, ribosomal genes are repeated in the genomes of Microsporidia. In E. cuniculi, despite its reduced genome size, rDNA clusters are located near the telomeres of the chromosomes (Brugère et al. 2000). In N. apis, there are at least three repeated array copies of rDNA (Gatehouse and Malone 1998), but the number of rRNA genes in the genomes of both N. apis and N. ceranae is still to be precisely calculated. The heterogeneity of rDNA copies in the same genome has been known since the genome project of N. ceranae: Cornman et al. (2009) reported that, with the assembly algorithms currently used, it was impossible to achieve the complete sequence of the rDNA due to its variability. On the other hand, Sagastume et al. (2011) showed that the sequence diversity of rDNA in this species is very high, especially in the IGS but also in the first 600 bp analysed from the 16S RNA gene, where some SNPs and INDELs appear systematically in clones from isolates geographically very distant. The data also indicated that recombination was generating new variants.
After comparing five complete sequences of the 16S RNA gene obtained from the same isolate, we found polymorphic sites from nucleotide #46 to #970 from a total length of about 1259 bp. The approach to the variability in expressed 16S RNA was focused on the segment of 750 bp (ranging from nucleotide 52–801). This fragment seems to be representative of the whole molecule in terms of variability given their practically identical polymorphism indexes. In addition, the variable regions are the same in both cases and no polymorphisms were observed within extensive sequence domains (Fig. 3). The variations included point substitutions, the number of Thymidines or Adenines in a series, and a GATT tetranucleotide insertion (Fig. 3). Some differences were unique. Given the diversity previously described, this was expected after the analysis of only a small number of sequences; a more complete analysis would most probably reveal additional representatives of such alleles. For instance, the insertion A666 and the deletion T846 were unique here but were also found in clones from different isolates in the 2011 study, and A666 is also present in the GenBank reference sequence obtained by Fries et al. (1996). PCR errors and in vitro recombination can be discarded since in a parallel test carried out under the same experimental conditions, no recombination was observed, and a single point mutation occurred in more than 37,000 nucleotides. Moreover, as stated by Ueno et al. (2007), the changes are systematically located in specific regions that tend to be more variable and do not alter the overall secondary structure; in vitro errors were expected to be randomly distributed, irrespective of the in vivo relevance of a given region.
The ribosomal RNAs have been the subject of thorough analyses. Although 16S secondary structure of N. ceranae has already been described (Chen and Huang 2010), no studies have been carried out to analyse the consequences of mutations in the rRNA sequence, nor the specific role of each area of the secondary structure. Two- and three-dimensional models of E. coli rRNA are currently being used for comparison and reference to a wide range of organisms. Indeed, a number of constant features are maintained among very different organisms (Woese et al. 1983). Microsporidia, such as Nosema, have reduced genomes and, correspondingly, the SSU RNA is also much smaller than in “classic” eukaryotes. Relative to E. coli, some helices are missing but the general structure is equivalent (http://www.rna.ccbb.utexas.edu).
From the 19 substitutions found, 12 were located in double stranded conformation areas (ds) of the secondary structure, and 7 involved in single stranded loops (ss) (Table 1). All but one “ds-mutations” produced a weaker joint with their complementary bases. The exception is G662A, which changes a G-U pair to A-U (which appeared in 29% of the sequences of this study) and is located in a region of high variability (Fig. 4). It corresponds to helix 26 in E.coli (h26) which, according to Woese et al. (1983), has an unusually high frequency of contiguous G-U pairs. While mutations G659A, G662A and G678A in N. ceranae seem to maintain those contiguous noncanonical base pairs, C661G weakens the helix, changing G-U pair to C-U (Fig. 4). The terminal bulge next to h26 is one of the only two structural differences between N. ceranae and N. apis (Chen and Huang 2010), and house the mutation A668G. This change was detected in 46% of the sequences analysed, appearing systematically throughout the different trials, in different years, carried out in our laboratory. No mutations in h26 have been described in E. coli and, at present, we cannot shed light on the meaning of this polymorphism, but its recurrence strongly suggests a possible functional relevance.
The different indels spread along the secondary structure were all in single stranded conformation areas. Four are single nucleotide indels that vary the number of Thymidine or Adenosine in a series. The GATT tetranucleotide indel is especially interesting. The insertion was present in five of the haplotypes described here, and was also found in around 30% of the isolates of N. ceranae from very different origins as far apart as Spain, Central Europe, Kyrgyzstan in Asia (Sagastume et al. 2011) and even Australia (unpublished results), suggesting a hypothetical adaptive significance for the GATT polymorphism. In the secondary structure of the 16S RNA (Fig. 4), it is located in the single stranded elbow between helices h7 and h8 (E. coli nomenclature). Helix 8 plays important roles in E. coli and mutations have been described as missense and nonsense suppressors (McClory et al. 2010). E. coli h8 interacts with h14 near the EF-Tu binding site (McClory et al. 2010) which possibly has a direct role during the decoding process (Villa et al. 2009). Both helices h8 and h14 also interact with 50S subunit to form bridge B8 (McClory et al. 2011). In N. ceranae, no variability was found in the equivalent to h8, but the adjacent GATT indel modifies the length of the elbow which hypothetically could influence the capability of movement of h8. The fact that this polymorphism has been maintained worldwide, suggests some functional role.
Only two of the 19 mutations found have been demonstrated to have functional consequences: G46A and A970G. In 2005, Yassin et al. (2005) catalogued the equivalent positions in E. coli (positions 51 and 1181, respectively). The first is a “mildly deleterious mutation clustered in a region of less obvious functional significance”, and the second a “moderately deleterious mutation located in an area substantially far from the known functional centers that occupy the interface side”. In N. ceranae, the two mutations appeared in just one clone from the genomic DNA study and was not detected in the cDNA clones nor in previous studies. At present there is no information on the rest of the polymorphic points described here.
Intraspecies genetic variability has been described in a number of Microsporidia, such as the Encephalitozoon species, where different genotypes are associated to preferences for different hosts (Ghosh and Weiss 2009). In E. hellem, different isolates of human origin showed variable rates of in vitro proliferation (Haro et al. 2006). In all these cases, a given strain proved to be genetically uniform. By contrast, intrastrain diversity exists in N. bombi (O'Mahony et al. 2007), N. apis (Gatehouse and Malone 1998), N. bombycis (Liu et al. 2013) and N. ceranae (Sagastume et al. 2011). In this latter case, as in the algae P. wickerhamii (Ueno et al. 2007), it has been shown that new variants are originated by recombination of different sequences. To our experience, no matter how selected, purified and diluted an isolate of N. ceranae is, uniform sequences cannot be obtained, unless cloned. Therefore, the heterogeneity of certain domains is a characteristic feature of this microsporidian.
Different SSU genes in the same genome have been reported in different organisms including eubacteria (Ueda et al. 1999), archaeobacteria (Mylvaganam and Dennis 1992), Cryptosporidium parvum (Rooney 2004), P. berghei (Gunderson et al. 1987; Nishimoto et al. 2008), the microalga P. wickerhamii (Ueno et al. 2007) and even flatworms (Carranza et al. 1999). In all of them, the different variants of SSU have been demonstrated to be expressed, although whether or not they actually have different functional roles is still an open question.
Van Spaendonk et al. (2001) showed that, under experimental conditions, the different ribosomes of P. berghei were equally functional. In the cases above, the polymorphic sites are located in positions that do not affect the general structure of the SSU. Although some segments have more constraints than others, ribosomal genes are still highly conserved. Different rates of evolution have been observed for the duplicates of ribosomal genes (Carranza et al. 1999; Mylvaganam and Dennis 1992; Ueda et al. 1999) indicating positive selection, so it could also be suggested that each sequence is adapted to a somehow different function or to different environmental conditions. It is important to emphasize that if two ribosomal variants behave identically under a given set of experimental conditions, it does not imply the same result would be obtained in any environment.
N. ceranae possesses a considerable range of different rRNA sequences, which can be obtained from any isolate, from any geographical origin, and also from a single infected honeybee. This is remarkable considering that the genomes of Microsporidia are typically compacted and have lost large amounts of genes in adaptation to their strict parasitic life. Further studies need to examine why concerted evolution of rDNA is not occurring here, and whether the ribosomal diversity is related with the shockingly fast worldwide invasion of honeybee colonies by N. ceranae.