DNA methylation alteration is a major consequence of genome doubling in autotetraploid Brassica rapa

Polyploids are typically classified as autopolyploids or allopolyploids based on the origin of their chromosome sets. Autopolyploidy is much more common than traditionally believed. Allopolyploidization, accompanied by genomic and transcriptomic changes, has been well investigated. In this study, genetic, DNA methylation and gene expression changes in autotetraploid Brassica rapa were investigated. No genetic alteration was detected using an amplified fragment length polymorphism (AFLP) approach. Using a cDNA-AFLP approach, approximately 0.58% of fragments showed changes in gene expression in autotetraploid B. rapa. The methylation-sensitive amplification polymorphism (MSAP) analysis showed that approximately 1.7% of the fragments underwent DNA methylation changes upon genome doubling, with hypermethylation and demethylation changes equally affected. Fragments displaying changes in gene expression and methylation status were isolated and then sequenced and characterized, respectively. This study showed that variation in cytosine methylation is a major consequence of genome doubling in autotetraploid Brassica rapa.


INTRODUCTION
Two major types of polyploidy are known: autopolyploidy and allopolyploidy.Autopolyploids are generally considered to be derived from multiplication of the genome within a single genome, whereas allopolyploids arise via interspecific hybridization and chromosome doubling [1].Autopolyploids have long been considered rare in natural populations and viewed as evolutionary dead-ends [1].A dramatic resurgence in the study of polyploidy has increased our knowledge of autopolypoid dynamics and has elevated estimates of the frequency of autopolyploids in natural populations [1,2].Thus, autopolyploidy may contribute more to evolution and species diversification than was traditionally thought [1,3].
Genomic shock, including chromosomal rearrangement, the gain and loss of chromosome segments, gene repression and activation, subfunctionalization and transposon activation, as well as epigenetic changes, has been investigated repeatedly in allopolyploids [4].In autopolyploids, changes to the nuclear environment are not as profound as in allopolyploids [5,6].The extent to which autopolyploidy induces genetic, epigenetic and gene expression changes in different autopolyploids remains debatable.
The genus Brassica has long served as a model system for studying the molecular and phenotypic changes associated with both recent and ancient polyploidization events.Previous studies have reported rapid genetic, epigenetic, and expression changes in Brassica allopolyploids and found that homoeologous recombination is the major mechanism underlying these variations [6,[28][29][30].Studies aimed at separating hybridization and genome doubling effects have confirmed that genome doubling can ameliorate genomic and transcriptomic alterations induced by hybridization and can instigate additional alterations [6,29].No significant proteomic changes were detected between diploid and autopolyploid Brassica oleracea [31].Therefore, independent studies are necessary to clarify the genetic, epigenetic and gene expression responses in Brassica autopolyploids.
The objective of this work was to uncover genetic, DNA methylation and gene expression changes in Brassica autopolyploids.AFLP, methylation-sensitive amplification polymorphism (MSAP) and cDNA-AFLP approaches were applied in autotetraploid Brassica rapa and its diploid progenitors.Sequences showing changes in DNA methylation and gene expression in the autotetraploids were cloned and characterized, respectively.

Plant materials
Autotetraploid Brassica rapa was generated with colchicine treatment of fully homozygous diploid Brassica rapa L. cv.Duanbaigeng.Thirty days after germination, the plantlets were flooded with a 0.1% colchicine solution for approximately 6 h.The plantlets were rinsed in distilled H 2 O and grown to maturity.From each colchicine-treated plant (C1 generation of autotetraploids), seeds (C2 generation of autotetraploids) were harvested and sown.The ploidy of the C2 generation of autotetraploids was assessed using chromosome number and morphological characteristics (the size of and the number of chloroplasts in the guard cells of stoma).
The C2 generation of autotetraploids and the diploid progenitor were grown in a greenhouse under the same conditions.Young leaves (5 cm in length) of mature plants and similarly sized flowers (~0.5 cm in length, before opening) were collected from different autotetraploid and diploid individuals at comparable developmental stages.

DNA and RNA extraction and cDNA synthesis
Total DNA was isolated using the CTAB procedure.Total RNA was extracted using TRIzol reagent (Invitrogen, cat.no.15596026).The reverse transcription, synthesis of second-strand cDNA and purification of cDNA were performed by strictly following previously described procedures [6].

AFLP analysis
AFLP was performed according to previous research [6].The adaptors and primers are listed in Supple-mentary Table S1.Selective PCR products were mixed with 25 µl of formamide dye (98% formamide, 10 mM EDTA, 0.05% w/v bromophenol blue and xylene cyanol), denatured at 95°C for 4 min and separated by electrophoresis on 6% denaturing polyacrylamide (20:1 acrylamide:bisacrylamide, 7.5 M urea, and 1× Tris-borate-EDTA buffer, pH 7.8).The gels were prerun at 100 W for approximately 30 min before 4.5 µl of the mix was loaded; the gels were then run at 65 W for approximately 2 h.The gels were silver-stained using a DNA silver staining system (Promega, catalog number: Q4132).Gel images were taken using a CCD camera.

cDNA-AFLP analysis
The cDNA-AFLP procedure was described in previous research [29] and modified from a published method [32].The adaptors and primers are listed in Supplementary Table S1.Electrophoresis and staining were performed as AFLP analysis.To ensure that there was no DNA contamination in our RNA samples, a negative control was prepared without reverse transcriptase (RNA samples treated with RNase-free DNase I, starting from the digestion-ligation step).A clear cDNA-AFLP gel with no bands was obtained.

MSAP analysis
MSAP analysis was performed following the procedure described in previous research [29], which was modified from the general procedure [33].The adaptors and primers are listed in Supplementary Table S1.The electrophoresis and staining were performed as for AFLP analysis above.

Scoring of AFLP, cDNA-AFLP and MSAP bands
For AFLP, cDNA-AFLP and MSAP analyses, one biological replicate and three technical replicates were performed.The technical replicates were run from the same cDNA and DNA sample but from a different digestion-ligation-amplification reaction.Only clear and reproducible bands between the three replicates were used for scoring.Moreover, with AFLP, cDNA-AFLP and MSAP gels, the upper and lower regions of the gels, where resolution was not satisfactory, were not used for band scoring.The scored AFLP, cDNA-AFLP and MSAP bands were transformed into a binary character matrix, using "1" or "0" to indicate the presence or absence of a band at a particular position, respectively.

Cloning and characterization of differentially methylated and expressed fragments
Fragments that showed evidence of alteration in MSAP gels and cDNA-AFLP gels in the autotetraploids were excised, eluted and amplified with the same primer combinations used in the selective amplification.The PCR products were cloned with the pCR2.1-TOPOTA cloning kit (Invitrogen) and sequenced.The sequences obtained were analyzed for similarity to the Brassica rapa genome sequence via Phytozome 12 (https://phytozome.jgi.doe.gov/pz/portal.html).

The absence of genetic changes in autotetraploids
To assess genome doubling-induced genetic changes in a quantitative manner, AFLP was performed on DNA from leaves to track genetic changes in the autotetraploids compared to the diploid plants.The band patterns in the autotetraploids were expected to be similar to the diploids, and all cases of deviation from such additivity were scored as genetic changes induced by genome doubling.A total of 2069 AFLP fragments were obtained using 48 pairs of selective primer combinations from leaves.No novel or missing bands were detected between diploid and autotetraploid plants.It should be noted that the AFLP approach would mask any absence of a single gene copy in autotetraploids.

Alterations in gene expression in autotetraploids
To investigate the consequences of gene expression on genome doubling, cDNA-AFLP display was performed on the autotetraploids and their diploid progenitors.Using 48 different selective primer pairs, 1966 and 2033 transcripts were detected in leaves and flowers, respectively (Fig. 1).Of the 1966 transcripts that were detected in leaves, 7 transcripts were silenced in autotetraploids and 3 transcripts were ac-tivated in autotetraploids.In all, 0.51% of the transcripts showed altered expression in leaves (Table 1).Of the 2033 transcripts that were detected in flowers, 9 transcripts were silenced in autotetraploids and 4 transcripts were activated in autotetraploids.In all, 0.64% of the transcripts showed altered expression in flowers (Table 1).

Alterations in rapid DNA methylation in autotetraploids
The MSAP technique was performed by incorporating a pair of isoschizomers, HpaII/MspI, which possess differential sensitivity to cytosine methylation at a CCGG site.The MSAP band patterns were compared between the diploid and the autotetraploid samples to detect cytosine methylation changes at CCGG sites throughout the genome.In the absence of methyla-tion changes, the autotetraploid lines were expected to have the same MSAP patterns as the diploid parent.Any deviation from this expected additivity was considered to be the result of an alteration in methylation pattern related to genome doubling, in which the autotetraploid plants significantly differed from the diploid parent plants.In the autotetraploid and diploid B. rapa, 1798 and 1636 clear and reproducible bands were amplified, respectively, from leaves of mature plants and from flowers using 48 pairs of EcoRI + HpaII/MspI selective primer combinations (Fig. 2).Five major groups were identified according to methylation status in the autotetraploids and the diploids, as shown in Table 2.The first group (group A) contains additive band patterns between the autotetraploids and the diploids, among which 1270 and 1160 nonmethylation bands (A1), 120 and 103 hemi-methylation bands (A2), and 383 and 338 full     methylation bands (A3) were detected in leaves and flowers, respectively.
Four additional groups of non-additive band patterns were observed.Ten and fourteen non-methylated bands from diploid leaves and flowers were found to be hypermethylated in the autotetraploids (group B).Three and four hemi-methylated bands from diploid leaves and flowers displayed different methylation status in the autotetraploids (group C).Five and eight fully methylated bands from diploid leaves and flowers showed different methylation status in the autotetraploids (group D).Seven and nine novel methylation bands were detected from the diploid parental "00" band pattern in leaves and flowers, respectively (group E).This type of methylation change could be caused by full methylation of both cytosine residues or full methylation of the external cytosine to another methylation status, rather than by sequence mutation at non-CCGG to CCGG sites.In all, 25 of 1798 (1.39%) CCGG sites displayed altered DNA methylation states in leaves, and 35 of 1636 (2.14%) CCGG sites displayed altered DNA methylation states in flowers.
Methylation changes in autotetraploids were caused by either hypermethylation or demethylation.A band appearing in the MspI lane or HpaII lane indicated possible demethylation alteration and vice versa; a band missing in the MspI lane or HpaII lane indicated possible hypermethylation alteration.It should be noted that if internal CCGG site(s) exist in the MSAP fragments, the methylation states cannot be calculated according to the above rationale [34].However, most studies have indicated that internal CCGG site(s) in MSAP fragments occur infrequently [29,35].In this study, the B, C1 and D1 groups indicated hypermethylation changes, and the C2, D2 and E groups indicated demethylation changes.Thus, 14 bands were hypermethylated and 11 bands were demethylated in leaves, and 22 bands were hypermethylated and 13 bands were demethylated in flowers.It is interesting that the diploid genome methylation CCGG sites (10 and 14 in leaves and flowers, respectively) and diploid genome non-methylation CCGG sites (8 and 12 in leaves and flowers, respectively) had equal ability to undergo methylation status shifts during the process of genome doubling.

Molecular characterization of fragments showing changes in DNA methylation and gene expression
Transcript-derived fragments (TDFs) and MSAPisolated fragments (MIFs) showing alterations in the flowers of the autotetraploids were cloned and sequenced.The homologies of the MIFs and TDFs to the Brassica rapa genome sequence were detected via Phytozome 12.Among the 13 TDFs, two transposaselike proteins were detected (TDF9 and TDF10) (Table 3).The molecular characterizations of the MIFs are listed in Table 4.Of the 35 sequenced MIFs, 14 sequences showed similarity to known annotated genes in Brassica rapa (Table 4).MIF10, MIF25 and MIF30 had similarity to known retroelement sequences.

DISCUSSION
This study provides a comprehensive portrayal of genome doubling effects in autotetraploid B. rapa.No genetic alteration was detected in autotetraploid B. rapa using an AFLP approach.Approximately 0.58% and 1.7% of fragments showed changes in expression and changes in DNA methylation, respectively.This may underestimate genetic alterations in autotetraploids; because genome doubling increases the level of genomic redundancy and AFLP is a dominant marker, some recessive alterations may be masked by the dominant alleles.AFLP-derived cDNA-AFLP and MSAP approaches may underestimate gene expression and DNA methylation alteration in autotetraploids.Only approximately 0.02% of sites in the genome, 0.06% of cytosine sites in the genome and 4.2% of the total transcript fragments were analyzed, compared to whole genome sequencing, and thus some variation underlying genome doubling is missed.However, MSAP and AFLP can still provide insight into genetic and epigenetic changes in plant biology [36,37].

Gene expression alterations in autopolyploids
About 0.58% of gene expression changes occurred in autotetraploid B. rapa.This result is consistent with a previous study in amphihaploid and amphidiploid B. napus [29], trigenomic allohexaploid Brassica carinata × Brassica rapa [6] and autopolyploid Pasalum notatum [22], which confirmed that genome doubling per se could induce small but distinct gene expression alterations during polyploid formation.

Cytosine methylation changes in autopolyploids
MSAP alterations, either from genetic variation or from methylation status, shift the marker locus.AFLPs are typically interpreted as genetic variation.
In this study, the MSAP deviation from Mendelian expectation is much higher than that of the AFLP markers.This indicated that most DNA methylation changes in the autotetraploids were caused by a methylation status shift in the pre-existing CCGG sites.Compared to genetic and gene expression changes detected in this study, DNA methylation changes are a major consequence in autotetraploid B. rapa.A similar phenomenon was also detected in amphihaploid and amphidiploid B. napus [29].Studies in trigenomic allohexaploid Brassica carinata × Brassica rapa indicated that genomes per se could induce half of the methylation changes in the conversion from triploid to hexaploid [6].The DNA methylation-sensitive enzymes used in the MSAP assay are biased towards DNA methylation at CGG and CCG sites, while DNA methylation changes at CG, CHG and CHH sites are not detected by the MSAP assay.
Studies in Cymbopogon [17] and Eragrostis curvula [15] autopolyploids reported that autopolyploidy caused increased cytosine methylation.Variation in DNA methylation during the tetraploid-diploid conversion reverted during the diploid-tetraploid conversion in Eragrostis curvula [15].A cross between dandelion ploidy levels triggered de novo methylation variation [14].A study in Phlox drummondii provided evidence for an increase in DNA methylation changes through subsequent generations [9].Stochastic changes among individuals were observed in Solanum [16].However, no methylation changes were detected in watermelon with differing ploidy [18].The above data supported the hypothesis that variation in cytosine methylation might be a major consequence during genome doubling in polyploids.
Epigenetic landscape changes can affect gene expression.The role of epigenetic changes during polyploid formation and diploidization is still not com-pletely clear.Furthermore, the DNA methylation status of newly formed species appears to change consistently through several early generations [9].All these studies focused on signal cytosine methylation polymorphisms (SMP), which may not be linked to genotype per se.Future work should analyze differentially methylated regions (DMR) during genome doubling [38].

Table 1 .
Summary of cDNA-AFLP analysis in autotetraploids and their diploid progenitor.

Table 2
; d: arrow indicates B1 MSAP band pattern listed in Table2; e: arrow indicates B2 MSAP band pattern listed in Table2; f: arrow indicates C1 MSAP band pattern listed in Table2; g: Arrow indicates D1 MSAP band pattern listed in Table2; h: Arrow indicates D2 MSAP band pattern listed in Table2; i: Arrow indicates E1 MSAP band pattern listed in Table2; j: Arrow indicates E2 MSAP band pattern listed in Table2; c: Arrow indicates E3 MSAP band pattern listed in Table2

Table 2 .
Frequencies of MSAP band patterns and cytosine methylation alterations in various autotetraploid and diploid organs.

Table 3 .
Molecular characterization of fragments showing different expression in autotetraploids according to cDNA-AFLP analysis * Band pattern in cDNA-AFLP gels as follows: diploid lane, autotetraploid lane; + band present, -band absent

Table 4 .
Molecular characterization of fragments showing methylation changes in autotetraploids relative to their diploid progenitors.

Table 4 .
[26]inued dopsis[19], almost no ploidy-dependent changes in gene regulation were detected.A study in an autopolyploid sunflower series[26]reported that gene expression varied among different autopolyploid individuals.