Chromosomal loci important for cotyledon opening under UV-B in Arabidopsis thaliana
- 3.5k Downloads
Understanding of the genetic architecture of plant UV-B responses allows extensive targeted testing of candidate genes or regions, along with combinations of those genes, for placement in metabolic or signal transduction pathways.
Composite interval mapping and single-marker analysis methods were used to identify significant loci for cotyledon opening under UV-B in four sets of recombinant inbred lines. In addition, loci important for canalization (stability) of cotyledon opening were detected in two mapping populations. One candidate locus contained the gene HY5. Mutant analysis demonstrated that HY5 was required for UV-B-specific cotyledon opening.
Structured mapping populations provide key information on the degree of complexity in the genetic control of UV-B-induced cotyledon opening in Arabidopsis. The loci identified using quantitative trait analysis methods are useful for follow-up testing of candidate genes.
KeywordsQuantitative Trait Locus Mapping Population Recombinant Inbred Line Quantitative Trait Locus Analysis Quantitative Trait Locus Mapping
Higher plants have complex sensory mechanisms to detect changes in the light environment [1, 2, 3, 4, 5]. There is evidence from physiological experiments that low levels of UV-B radiation (280-315 nm) induce photomorphogenic responses in Arabidopsis thaliana [6, 7] and other species [8, 9]. Results from studies of UV-B-induced photomorphogenesis  and gene expression [10, 11, 12] point to the existence of a complex web of interactions involving phytochromes and cryptochromes as a part of the underlying perception, regulation and/or signaling system . However, several responses induced by low-dose UV-B appear to be to some extent independent of the photoreceptors that mediate responses to visible radiation [6, 7, 11, 13]. Cotyledon-opening under defined UV-B conditions is a key phenotype, as this morphological response is consistent with signalling through a photoreceptor . Fluence-response curves and the independence of the cotyledon-opening response from DNA repair make this defined cotyledon-opening phenotype especially useful for genetic screens .
Screens for isolating mutants with altered sensitivity to UV-B in hypocotyls have been performed and led to the identification of uli3, a UV-B hyposensitive mutant , and uvr8, a mutant hypersensitive to UV-B [14, 15]. Both mutants appear to be altered in UV-B-specific signaling pathways. Photomorphogenic mutant analyses implicate the b-ZIP transcriptional factor HY5 as a component in the UV-B regulation of the expression of selected genes , and the COP1 gene as a regulator of HY5 and additional gene expression under UV-B . Extensive analysis of UVR8 indicates that nuclear localization of the protein is important in UV-B signal transduction  and that UVR8 and HY5 are part of the same signal transduction pathway for gene expression responses .
Genome scanning is another useful approach for the isolation of loci that encode components of UV-B perception and signalling mechanisms. Genetic variability in wild ecotypes of A. thaliana provides an opportunity to discover genetic regions controlling phenotypic differences [19, 20]. Genetically complex or polygenic inheritance, quantitative measurement and the availability of DNA markers allow access to quantitative trait loci (QTL) [21, 22]. QTL mapping studies delimit the chromosomal regions controlling quantitative traits, and in some cases allow the identification of new alleles or causal mutations [23, 24, 25]. In the Arabidopsis model system QTL analysis has been performed with potentially adaptive traits like salt tolerance , aluminium tolerance , and powdery mildew disease resistance , as well as with developmental or physiological traits such as seed size , leaf architecture , inflorescence development  and growth rate . It is possible to compare genetic architecture in different environments, including controlled treatment environments, although the power to detect loci will be less than the detection power for overall genetic effects .
Metabolic and signalling pathways can be identified from epistatic interactions [34, 35, 36, 37], and gene interactions are widespread and important in theory and in most data sets. Epistatic interactions can explain substantial amount of variance , although power to detect epistasis is low in most recombinant inbred (RI) experiments with 100-200 lines. Better understanding of the genetic architecture of UV-B responses will thus be achieved by incorporation of epistatic locus interactions into the analysis.
Differences in phenotype measurements between individuals of the same RI genotype are due to random fluctuation and alleles genetically controlling stability (canalization, for developmental traits) or variance (anti-canalization) [39, 40]. Arabidopsis loci for canalization of flowering and plant growth have been found previously, by QTL mapping of RI lines . Full understanding of phenotype and environmental constraints on phenotype expression require consideration of the stability of the phenotype as well as the extent of the effect of important alleles.
No QTL studies have been reported with UV-B-induced responses in Arabidopsis. In rice Sato et al.  have reported three QTLs controlling UV-B resistance, with a more precise mapping of the location of one of the reported QTLs . Recombinant inbred lines, or in some species clones of F2 families, are the most useful for exploration of genotype and environment interactions, as the same genotype can be exposed to multiple environments .
In the experiments described in this paper we used four sets of RI lines to perform a QTL analysis of UV-B-induced photomorphogenesis in Arabidopsis. We measured cotyledon opening in de-etiolating seedlings as a model photomorphogenic response; under certain experimental conditions this response is induced by UV-B (but not UV-A) and the induction mechanism for cotyledon opening does not appear to involve signals derived from UV-B-induced DNA damage . We used composite interval mapping to locate additive, UV-B-specific and epistatic QTL, by incorporating all the data including replicates and multiple 'environments', which in this case are UV-B treatment and control (-UV-B) growth conditions. We also analyzed these data with a straightforward single-marker method to select the most robust loci for candidate gene hypothesis testing. Loci for cotyledon-opening stability were identified; these include loci at the same position as the cotyledon opening QTL and new loci that regulate stability but not extent of cotyledon opening. We combined our QTL with information from the literature to produce a list of candidate genes for further testing. Mutations in our top candidate gene, HY5, were defective in UV-B-induced cotyledon opening.
Results and Discussion
Chromosomal regions affecting cotyledon opening
Comparison of epistatic interactions
Gene interactions (epistasis) are likely to be important in explaining genetic control of phenotypic traits; however, these interactions are difficult to detect . We scanned for interactions between two loci, whether or not those loci had an effect on cotyledon opening considered singly. Significant two-locus epistatic interactions were detected in the BayxSha (Fig. 2), ColxKas (Fig. 3) and LerxCvi (Fig. 5) populations. Larger mapping populations are needed for detection of multiple loci by interval mapping methods , and these are the larger mapping populations in our study. In BayxSha and LerxCvi the epistatic interactions had lower heritability than any individual locus, suggesting only small contributions to genetic control of cotyledon opening. In contrast, for the ColxKas population epistatic interactions are important, with heritabilities as high as those seen in individual main-effect loci (Additional File 2).
The BayxSha loci with significant epistasis also had significant main effects (Fig. 2), although the main effects were opposite in their allele contributions (Sha allele with the higher effect in BS2_25 and the Bay allele having the larger effect at BS5_53, Additional File 1). Thus, the epistatic interaction is not a simple additive effect. The estimated epistatic high allele is Bay, which suggests that the Bay allele at BS5_53 interacts equally well with either allele from the BS2_25 region, but the Sha allele at BS5_53 is more specific.
As in BayxSha, the LerxCvi epistatic loci also have single-locus main effects. In this case, both single-locus Ler alleles contribute to cotyledon opening while the epistatic alleles are from Cvi. This suggests that the Cvi alleles contribute to a pathway or complex, while the products of the Ler alleles work independently.
In the ColxKas population with the CK2_75/5_27 pair the allele effect was of the same direction as the individual locus alleles and of intermediate effect size. In this population there was also an epistatic interaction between two new loci on chromosomes 4 and 5; these new loci had no significant effect considered separately (Fig. 3). This is an example of epistasis with no main effect, which may indicate the epistatic alleles have a negative effect on their interaction partner.
Detection of interactions between loci may be useful for predicting which genes under the QTL regions are causal, as metabolic networks and gene families can cause epistasis [34, 35, 36, 37]. The four mapping populations vary in extent of epistasis effects, indicating that particular allele interactions underlay the statistical epistasis we detected.
Chromosome regions with a UV-B-specific effect
Four loci with significant effects in the +UV-B-exposed treatment environment (cellulose di-acetate filter) were identified, BS2_25 from the BayxSha population, CK3_28 from the ColxKas population, and LC2_43 plus LC5_8 from the LerxCol population (Figs. 2, 3, 4, 5). Significant loci detected explained only about one-fourth of the genotype by environment variance. The remaining variance could be explained by presence of a number of additional small-effect loci or by epistatic interactions among more than two markers in these populations.
Architecture of canalization loci
Genomic Loci for Canalization
QTL Position in cM
QTL Position Range in cM
Comparison to cotyledon opening QTL
Bay x Sha population1
4 - 17.8
P = 6.9 × 10-4
Same locus; higher variation allele has lower effect on opening.
Ler x Col population2
41 - 54
Locus is the same or nearby (intervals overlap), has significant UV-B variation effect but no additive main effect. Ler allele median variance is high in +UV-B round 1 (P = 0.0039), and Ler allele median variance is low in -UV-B round 2 (P = 0.016).
55 - 58
P = 3.5 × 10-5
New locus, no UV-B-specific effect.
Comparisons of chromosomal regions detected in the different populations
We do expect to have the same alleles in geographically distinct accessions if there is strong selection. There is some overlap in the accession represented in the mapping populations, in that Ler is in two populations and Col in two populations. The Col common parent does not condition any common loci but in mapping populations with a Ler parent, a region in the top of Chromosome 5 is present. The additional support for these loci derived from detection in multiple populations illustrates the value of including common parents in mapping populations or using a diallel design to derive RI lines [49, 50].
Comparison of UV-B signaling gene position to QTL
Highest priority candidate genes under UV-B-specific QTL
Large QTL region that includes CEN2
RED1 is downregulated by PHYB  and has auxin phenotypes.
At2g26710 light signaling
Involved in multiple light signalling pathways; UV-regulated expression 
We selected criteria for searching for candidate genes under QTL by considering 1) loci found in multiple populations with common alleles 2) loci with UV-B-specific effects 3) loci confirmed with both QTL Network and single-marker methods 4) size of region, and 5) the availability of additional information such as expression data on genes located near significant markers and/or annotation of genes with suggestive biochemical functions . Based on these criteria, a region on the top of Chromosome 5 was selected.
Candidate gene mutant hy5has a defect in UV-B-induced cotyledon opening
In the LerxCol and LerxCvi populations there is a region at the top of chromosome 5 identified by QTLNetwork and single-marker analysis as controlling cotyledon opening. Examination of this region of the AGI Arabidopsis map http://www.arabidopsis.org suggested one obvious candidate gene, HY5/At5g11260, in this interval. Global gene expression experiments have implicated the HY5 gene in signal transduction from a UV-B receptor  and additional expression measurements have placed the UVR8 and HY5 genes in that same signal transduction pathway .
We examined the cotyledon-opening angles by calculating the medians split on significant parent markers for Ler x Cvi and Ler x Col lines. For those Ler x Cvi lines that have the BH.144L Ler marker the UV-B opening angle is greater (median 101) than when Cvi allele is present (median 74); thus the Ler allele of this marker is responsive and Cvi allele is less responsive. For the Ler x Col population, marker C5_017 has a higher median for the Ler allele group than the Col allele subset, although the UV-B treatment difference is not significant in the QTL Network analysis. This suggests that the actual locus conditioned by the Ler responsive allele is between anchor markers nga249 and nga151 on the physical map; this region includes the HY5 gene. We examined the available Ler genomic sequence  in this chromosomal region. There were several polymorphisms in both the coding region and in the 5' UTR region of the HY5 gene.
Extensive characterization of morphological responses such as cotyledon-opening, when combined with genotyped structured mapping populations, allow identification of chromosomal loci and specific locus interactions important for control of the growth response. In addition to the region near HY5, we have identified several other loci associated with UV-B-induced cotyledon opening as well as loci specifying general light-induced opening.
Four recombinant (RI) line sets of A. thaliana were used for QTL mapping of cotyledon opening induced by UV-B. Lines used in this work were obtained from the Arabidopsis Biological Resource Center (ABRC) in Columbus, Ohio http://www.arabidopsis.org. The accession numbers include CS57921 for Bay x Sha, with 165 RILs developed by , CS84999 for Col x Kas, with 128 RILs developed by , CS1899 for Ler x Col, with 99 RILs (one is redundant), developed by , and CS22000 for Ler x Cvi, with 162 RILs developed by . The hy5-215 mutant seeds in the Columbia background were kindly provided by X.-W. Deng (Yale University).
QTL experimental design
We designed our QTL-mapping experiment following the guidelines described by Belknap  and Lynch and Walsh , with the additional priority of ensuring that populations with multiple parent ecotypes were included . As our priority was detection of loci important for UV-B-induced cotyledon opening, we chose to measure more replicates. This allows better estimation of effect sizes (amount of cotyledon opening conferred by the two alternative alleles at a locus), at the cost of less precise estimation of the position of the QTL on the map and less ability to detect 'small' QTL that have relatively little effect on the response [56, 58]. Measurements on each RI line under each treatment (+UV-B and -UV-B) were repeated, with the two sets of measurements named round1 and round2 (Additional File 5, Additional File 6, Additional File 7, Additional File 8, Additional File 9, Additional File 10, Additional File 11, Additional File 12). Eight seeds were planted for each treatment-RI line combination for each round. The two rounds were analyzed separately, in a way similar to traditional crop QTL analyses where planting year is analyzed as a separate factor.
Plant growth conditions
Eight seeds of each RI line or 16 seeds of hy5-215 and Col wildtype [59, 60] were sown in 1 cm height plastic boxes containing 0.8% agar (w/v). Boxes were covered with UV-B transparent film (Rolopac, Buenos Aires, 0.025 mm) and stored for three days in darkness at 6°C. To induce germination, seeds were exposed to an R-light pulse (30 minutes) and incubated in darkness at 25°C for 24 hours before being transferred to the UV-B irradiation chambers. R light (30 μmoles m-2 s-1) was provided by red fluorescent tubes (40/15, Philips).
UV-B treatment and measurement of cotyledon opening were carried out as previously described . Briefly, one-day-old etiolated seedlings were exposed for 3 days to a daily period of 2.5 hours of UV-B followed by a 5 minute R pulse. Seedlings were transferred to the dark for 24 hours and then the aperture of the cotyledons was measured on the fourth day. UV-B was provided by two UV-B 313 bulbs (Q-Panel 313, Cleveland), with a 0.1-mm-thick cellulose di-acetate film (La Casa del Celuloide, Buenos Aires) placed between the tubes and the seedlings to filter out the UV-C radiation emitted by the fluorescent tubes, as previously described [12, 15]. UV-B level was measured with a IL1700 double-monochromator spectroradiometer (International Light, Newburyport, MA), integrating the spectral irradiance between 290 and 315 nm. The radiometer was calibrated against a standard lamp (OL-40, Optronic, Orlando, FL) in the short-wavelength range and a model 1800 calibrator (LI-COR, Lincoln, NE) for wavelengths greater than 320 nm. In order to obtain a control treatment (-UV-B treatment), the UV-B portion of the spectrum emitted by the Q-Panel 313 bulbs was removed using a clear polyester film (Mylar-D, DuPont, Wilmington, DE; 0.1-mm thick). Neutral density filters were placed below the filters to reduce the intensity of UV to 2.25 μmoles m-2 sec -1. This photon flux was identified as a suitably low level from the action spectrum for UV-B-induced-cotyledon-opening . UV spectra under each film type are provided in Additional File 13.
Measurement of cotyledon opening
Cotyledon aperture was measured with a protractor and magnifying glass. Accuracy of measurement for each assayer was checked by measuring test seedlings 6-9 times; the coefficient of variation of the measurements within and between raters was less than 5% (data not shown). Measurements were expressed as angle between the cotyledons (0 to 180°). The raw measurement data (in degrees) on all individual plants is available in Additional File 5, Additional File 6, Additional File 7, Additional File 8, Additional File 9, Additional File 10, Additional File 11 and Additional File 12, arranged by population and date of measurement.
These RI lines have been characterized for molecular markers and that data was accessed from the Natural resource at http://arabidopsis.info/BrowsePage. We removed uninformative markers from our analysis by checking each adjacent marker for difference in the line distribution pattern of the markers. If the line distribution pattern in two successive markers on the chromosome was exactly the same, the second marker was removed. In order to determine if there was any artificial (non-syntenic) correlation in the particular RI line sets that we used, we performed Pearson correlations on each line set, as previously described , using the SAS procedure CORR. There was little artificial correlation evident in the line sets we used (data not shown).
To adjust for the large numbers of zeros in the data set, 0.5 was added to each phenotype measurement. Levene's median natural log statistic for each individual was then calculated as previously described [41, 62]. Using the median form is a conservative choice as compared to use of the mean version of the Levene's statistics, as the median form is less sensitive and less susceptible to artefacts with increasing variance in larger trait measurements. The full data set was used for QTLNetwork analysis.
QTL Network v. 2.0 analysis
This composite interval mapping program  was accessed from http://ibi.zju.edu.cn/software/qtlnetwork/. This particular CIM mapping program was chosen as it has provision for appropriate analysis of multiple replicates of each line and for multiple environments (+UV-B and -UV-B, in our case) . This program estimates effect sizes using a Bayesian fitting method , and chooses cofactors automatically. Program default settings (test window of 10 cM, walk speed of 1 cM) were used, except for deselection of the option for best genotype prediction. The experiment-wise P value threshold (across the whole experiment, two rounds of two treatment environments for eight replicate genotypes of each RIL in each of the four populations) was kept at P = 0.05, with 1000 permutations for F-value threshold selection.
A mixed model was constructed in SAS v9.1 (SAS Inc, Cary, NC) for individual marker state and UV-B treatment. Each round of the experiment was analyzed separately. A genome-wide P value threshold of 10-4 was chosen [33, 65], as the marker density in these RI line sets was not large enough to require adjustment for marker correlation .
Cotyledon opening angles in mutant and Columbia wild-type were compared in individual experiments by permutation test and then in all three experiments. The modeling of all hy5 and Col measurements was performed using SAS PROC GENMOD with the angle measurements divided by ten to generate truncated values between 0 and 18 and specifying a Poisson distribution.
We gratefully acknowledge support from the Women's International Science Collaboration (WISC--a program of the American Association for the Advancement of Science) to A. S. in 2003-2004, and from Agencia Nacional de Promoción Cientifica y Tecnológica and Universidad de Buenos Aires to C. B. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. We are grateful to James E. Blum, Department of Mathematics and Applied Statistics at UNCW, who wrote the SAS code for correlation analysis and html display. We also thank Carlos Mazza, IFEVA, for providing the UV bulb spectra.
- 11.Ulm R, Baumann A, Oravecz A, Mate Z, Adam E, Oakeley EJ, Schafer E, Nagy F: Genome-wide analysis of gene expression reveals function of the bZIP transcription factor HY5 in the UV-B response of Arabidopsis. Proc Natl Acad Sci USA. 2004, 101: 1397-1402. 10.1073/pnas.0308044100.PubMedPubMedCentralCrossRefGoogle Scholar
- 12.Wade HK, Bibikova TN, CValentine WJ, Jenkins GI: Interactions within a network of phytochrome, cryptochrome and UV-B phototransduction pathways regulate chalcone synthase gene expression in Arabidopsis leaf tissue. Plant J. 2001, 25: 675-685. 10.1046/j.1365-313x.2001.01001.x.PubMedCrossRefGoogle Scholar
- 33.Lynch M, Walsh B: Genetics and Analysis of Quantitative Traits. Sunderland, MA: Sinauer Associates, Inc. 1998.Google Scholar
- 46.Park SK, Jung YJ, Lee JR, Lee YM, Jang HH, Lee SS, Park JH, Kim SY, Moon JC, Lee SY, et al: Heat-shock and redox-dependent functional switching of an h-Type Arabidopsis thioredoxin from a disulfide reductase to a molecular chaperone. Plant Physiol. 2009, 150: 552-561. 10.1104/pp.109.135426.PubMedPubMedCentralCrossRefGoogle Scholar
- 50.Simon M, Loudet O, Durand S, Berard A, Brunel D, Sennesal FX, Durand-Tardif M, Pelletier G, Camilleri C: Quantitative trait loci mapping in five new large recombinant inbred line populations of Arabidopsis thaliana genotyped with consensus single-nucleotide polymorphism markers. Genetics. 2008, 178: 2253-2264. 10.1534/genetics.107.083899.PubMedPubMedCentralCrossRefGoogle Scholar
- 55.Alonso-Blanco C, Peeters AJ, Koornneef M, Lister C, Dean C, van den Bosch N, Pot J, Kuiper MT: Development of an AFLP based linkage map of Ler, Col and Cvi Arabidopsis thaliana ecotypes and construction of a Ler/Cvi recombinant inbred line population. Plant J. 1998, 14: 259-271. 10.1046/j.1365-313X.1998.00115.x.PubMedCrossRefGoogle Scholar
- 58.Vales M, Schön C, Capettini F, Chen X, Corey A, Mather D, Mundt C, Richardson K, Sandoval-Islas J, Utz H, et al: Effect of population size on the estimation of QTL: a test using resistance to barley stripe rust. Theor Appl Genet. 2005, 111: 1260-1270. 10.1007/s00122-005-0043-y.PubMedCrossRefGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.