Combining linkage and association mapping identifies RECEPTOR-LIKE PROTEIN KINASE1 as an essential Arabidopsis shoot regeneration gene

De novo shoot organogenesis (i.e., the regeneration of shoots on nonmeristematic tissue) is widely applied in plant biotechnology. However, the capacity to regenerate shoots varies highly among plant species and cultivars, and the factors underlying it are still poorly understood. Here, we evaluated the shoot regeneration capacity of 88 Arabidopsis thaliana accessions and found that the process is blocked at different stages in different accessions. We show that the variation in regeneration capacity between the Arabidopsis accessions Nok-3 and Ga-0 is determined by five quantitative trait loci (QTL): REG-1 to REG-5. Fine mapping by local association analysis identified RECEPTOR-LIKE PROTEIN KINASE1 (RPK1), an abscisic acid-related receptor, as the most likely gene underlying REG-1, which was confirmed by quantitative failure of an RPK1 mutation to complement the high and low REG-1 QTL alleles. The importance of RPK1 in regeneration was further corroborated by mutant and expression analysis. Altogether, our results show that association mapping combined with linkage mapping is a powerful method to discover important genes implicated in a biological process as complex as shoot regeneration.The capacity to regenerate in vitro adventitious shoots is of major importance for biotechnological breeding and commercial in vitro initiation and propagation of plants. Unfortunately, shoot regeneration is not always easy to achieve: among plant species, varieties, and cultivars, it is highly variable and currently unpredictable. The impact of shoot regeneration for horticulture and agriculture is illustrated by the numerous studies that assess the natural allelic variation and map quantitative trait loci (QTL) for the regeneration capacity in diverse crops, such as tomato (Solanum lycopersicum), wheat (Triticum aestivum), rice (Oryza sativa), barley (Hordeum vulgare), sunflower (Helianthus annuus), cabbage (Brassica oleracea), and potato (Solanum tuberosum) (1–12). However, it is difficult to draw general conclusions from these studies because of the low-resolution linkage maps and little detailed knowledge about gene functions in these crops.Therefore, the use of the model plant Arabidopsis thaliana is more appropriate. In a widely applied two-step regeneration procedure, root explants are first incubated on an auxin-rich callus-inducing medium (CIM) and subsequently transferred to a cytokinin-rich shoot-inducing medium (SIM) (13). Genome-wide analyses of the gene expression profiles accompanying the successive steps in the regeneration process revealed multiple key regulators and genes implicated in phytohormonal signaling during shoot regeneration (14–18). Reporter gene fusions with marker genes allowed visualization of their spatiotemporal expression patterns during regeneration, contributing to the elucidation of the function of important shoot-related genes, such as CUP SHAPED COTYLEDON1, CUP SHAPED COTYLEDON2, SHOOT MERISTEMLESS, WUSCHEL (WUS), and CLAVATA3 (14, 19–22). By means of classical forward and reverse genetics approaches, additional genes involved in shoot regeneration have been identified (23).Shoot regeneration in Arabidopsis has also been studied by QTL mapping with recombinant inbred lines (RILs) of Ler × Col (24, 25) or Ler × Cvi (26). These studies revealed multiple QTL, but thus far, no quantitative trait gene (QTG) or quantitative trait nucleotide (QTN) responsible for any of these QTL has been reported. Indeed, linkage mapping studies often fail to identify the causal gene because of their limiting mapping resolution (27).Recently, genome-wide association studies with an increased mapping resolution have received much attention for the identification of QTL in plants, particularly in Arabidopsis, as an alternative to or combined with linkage mapping approaches (28–31). Here, we aimed at identifying QTGs underlying the natural shoot regeneration variation in Arabidopsis by using linkage mapping complemented with association mapping. Furthermore, early parameters, such as callus and root formation, explant greenness, and shoot primordia development, were examined in a set of 88 Arabidopsis accessions. We calculated pairwise correlations between the different parameters and shoot formation to assess whether the early observations could predict regeneration. We phenotyped 86 RILs derived from a cross of Nok-3 and Ga-0, accessions with high- and low-regeneration abilities, respectively, and mapped five regeneration QTL. A local association mapping revealed that RECEPTOR-LIKE PROTEIN KINASE1 (RPK1) is the most likely gene underlying the major QTL REG-1, which was supported by mutant analysis and a quantitative complementation test.