Analysis of adaptive ribosomal gene diversity in wild plant populations from contrasting climatic environments

Abstract

Plant populations may contain variation that reflects adaptation to local environmental conditions. Clues to adaptive evolution of plants may be found in the genomes of species growing in diverse environments or across steep environmental gradients, and under stress. We have examined populations of wild relatives of barley and rice across diverse environmental gradients. Greater diversity, in a nuclear biotic stress defense gene and in chloroplast genes, was found in the more stressed, hotter and dryer environments. This may reflect the greater heterogeneity of these environments. Adaptation of plants to different abiotic stresses (temperatures and levels of water availability) may also require significant adaptation to the different biotic (pest and disease) pressures in these environments.¹

 

Plants growing across environmental gradients revealed greater diversity in a defense gene (Isa) in more stressed, hotter and dryer environments.² Chloroplast genome diversity also exhibited a similar variation with environment.³ We now report analysis of nuclear ribosomal genes from the same wild population. Two contrasting environments did not show significant differences in the level of diversity. However the pattern of SNP distribution within the rDNA did vary with greater SNP density in the RNA coding sequences compared with the internal transcribed spacers.

Diversity within the Nuclear Ribosomal Genes of a Microlaena stipoides Individual

Ribosomal genes exist in the nuclear genome as thousands of tandem arrays,⁴ which according to the tenet of concerted evolution become identical over time.⁵ However in the case where concerted evolution is slower than speciation, divergent haplotypes of the region can be observed within an individual.⁶ As a result intra-individual variation has been identified in the ribosomal genes of many polyploids.^7,8 Until recently measuring intra-individual diversity has been accomplished using PCR, cloning and Sanger sequencing, each step of which has intrinsic limitations which would not have captured the breadth of diversity given the high copy numbers within the genome. Whole genome massively parallel shotgun sequencing data from a single M. stipoides plant² were analyzed to establish the level of nuclear ribosomal diversity within a tetraploid individual.

A reference assembly aligned 205,588 reads to a putative M. stipoides gene sequence using the following parameters; Mismatch cost of 2, insertion and deletion costs of 3, length fraction of 0.8 and similarity of 0.8, minimum distance for paired end reads was set at 180 with a maximum distance of 340, criteria was set to ignore non-specific matches and vote for ambiguous calls.

This resulted in an average coverage of 1,129 X and a minimum coverage of 56X. SNP were called for any polymorphism which occurred at a frequency of 1% or greater. SNP detection parameters were assigned as follows; window length of 21, Maximum number of gaps or mismatches of 2, Minimum quality score of 30 for SNP site and surrounding bases, minimum coverage was initially set as 1X. The SNP density for this individual M. stipoides plant was determined to be 0.65 SNP/100bp for the nuclear ribosomal genes.

M. stipoides Whole Genome Shotgun Sequencing of Population Bulks

DNA samples from two independent populations growing in two environmentally differentiated sites, site A and site B, were selected for massively parallel sequencing. Site B was at high altitude with lower temperatures and greater rainfall compared with the dryer and hotter site A near the coast. Eleven individuals were sampled per site. DNA from the 11 specimens from each site were first quantified by nanodrop spectroscopy and gel visualization and then equimolar amounts placed in one of two DNA pools corresponding to each site. Each individual sample contributed 270 ng of DNA to each pool, giving a total of 2.7 μg of DNA per pool. Each DNA pool was sequenced on an Illumina GAII Platform (Illumina) and processed as described by Nock et al.⁵

Sequence data were trimmed on a CLC genomics workbench (www.clcbio.com, last accessed 2nd January 2012) where reads with a quality score of less than 0.01 were discarded and paired end reads were trimmed to a minimum of 30bp and single end reads trimmed to a minimum of 20bp in length. Reads from the two sites were then assembled to the putative nuclear ribosomal consensus sequence using the reference assembly and SNP discovery parameters described above.

Secondary analysis of the CLC output was conducted using Microsoft Excel 2007 where SNP’s at any reference position were eliminated from the analysis if the number of variants was only 1, that is the polymorphism was between the reference and the pooled sample rather than within the pool. The minimum coverage (Fig. 1) was limited to 88 X or greater at any given locus for the SNP to be reported. Minimum acceptable SNP frequency at any given locus was set at 1%. SNP were reported as total number SNP/gene and as unique (defined as the SNP only occurring in the pooled DNA from one site for any given reference position) or as a shared SNP (defined as the SNP being identified in data pools from both Site A and Site B)(Table 1). SNP frequency/100bp was also calculated as; Total number of SNP/Number of bases in the alignment consensus to the reference gene sequence *100.

Figure 1. Sequence coverage of ribosomal genes from wild populations.

Table 1. Distribution of SNP across reference positions between sites. SNP are reported as unique (defined as the SNP only occurring in the pooled DNA from one site for any given reference position) or as a shared SNP (defined as the SNP being identified in data pools from both Site A and Site B).

Putative annotation	Shared SNP	SNP Unique to Site A	SNP Unique to Site B
17s2	72	22	38
ITS1	4	3	2
5.8s	1	4	2
ITS2	2	2	1
25s	73	55	45
Total	152	86	88

Diversity in M. stipoides rRNA Genes in Populations from Different Environments

The number of SNP in nuclear ribosomal gene sequences was compared between pools of 11 individuals from each of two populations growing in contrasting environments³ (Table 2). M. stipoides is tetraploid with a base chromosome number of 12, and an estimated genome size of 880 Mbp 3.⁴ Assuming the concentration of DNA derived from each individual was equal in each pool, a single copy nuclear allele would be expected to have a minimum frequency of 1/44, or 2.3% in a pool of 11 potentially heterozygous individuals. A SNP frequency threshold of greater than or equal to 1% was used for SNP from the nuclear genome as this was likely to be well above the frequency of sequencing errors. The rRNA gene SNP frequency in DNA pools from site A and site B were 3.97 SNP/100 bp and 4.03 SNP/100bp respectively. Application of the Fisher’s exact two-tailed test between to two sites indicates no significant difference in diversity in the nuclear ribosomal genes between locations.

Table 2. SNP distribution in the Nuclear RDNA identified in an individual M. stipoides plant and via pooled (11 wild plants) next generation re-sequencing from two isolated populations, Site A - hot dry, coastal locality, and Site B - wet, cool semi-alpine locality.

Nuclear Ribosomal Gene			Site A (pool of 11 plants)		Site B (pool of 11 plants)		Individual plant
	Size (bp)	Reference Position	Number of SNP	Number of SNP /100bp	Number of SNP	Number of SNP /100bp	Number of SNP	Number of SNP /100bp
17s2	1944	1–1944	94	4.84	110	5.66	7	0.36
ITS1	237	1945–2181	7	2.95	6	2.53	5	2.11
5.8s	164	2182–2345	5	3.05	3	1.83	1	0.61
ITS2	222	2346–2567	4	1.80	3	1.35	3	1.35
25s	3423	2568–5990	128	3.74	118	3.45	23	0.67
Total	5990		238	3.97	240	4.01	39	0.65

The high level of conservation of the functional 17s, 5.8s and 25s has been used for decades as a means to amplify the more variable internally transcribed spacer (ITS) regions which have been shown to have sufficient species specific variability to be used for phylogenetics in the Poaceae.^3-5 While the SNP distribution for an individual plant follows the expected pattern of being higher within the ITS than the RNA coding region, the number of SNP is far higher than previously reported within an individual for this locus. This is likely the result of utilizing whole genome shotgun massively parallel sequencing which can identify a far greater amount of intra-individual ribosomal diversity as it is not restricted by the limitations of PCR and detects very low frequency alleles in a population of sequences far beyond that which could be captured by cloning.

Distribution of SNP across the nuclear rDNA for both pools is not as expected. The intra-individual SNP density for the both pooled samples is higher in the RNA coding regions than the ITS. The higher diversity in the coding regions suggests that this variation is functional. While an overall difference in diversity was not found between the two sites the elevated diversity in the coding regions may be adaptive. This supports the potential for adaptive variation in ribosomal genes in wild plant populations as reported from other studies.^11-13 While conventional analysis of ribosomal gene variation by PCR and electrophoresis has been sucessfullly applied, the use of next generation sequencing is proving to be a more sensitive methodolgy capable of identifying sequence variation that may have previously been overlooked due to the repetitive nature of the ribosomal loci.