ABSTRACT: Persistence of the invasive perennial weed leafy spurge is mainly attributed to seasonal production of underground adventitious buds (UABs), which undergo well-defined phases of dormancy (para-, endo- and eco-dormancy). These well-defined phases of dormancy also allow UABs of leafy spurge to survive extreme seasonal variations, including dehydration-stress. Consequently, objectives of this study include understanding the effects of dehydration-stress on vegetative reproduction, flowering competence, and transcript profiles at different phases of bud dormancy. The vegetative growth potential of UABs was monitored by removing the aerial portion of dehydration-stressed plants and re-watering the root system. Further, microarray analysis was used to follow transcriptome profiles to identify critical defense- and signaling-pathways at different phases of dormancy in UABs. Surprisingly, only 3 days of dehydration-stress is required to break the endodormant phase in UABs. In leafy spurge, vernalization of endodormant UABs has previously been shown to induce flower competence, while breaking of endodormancy via dehydration-stress did not induce floral induction. Thus, these two environmental treatments open a unique approach to independently dissect molecular mechanisms involved in endodormancy maintenance and floral competence. Bioinformatics analysis of transcriptome data helped to identify models and overlapping pathways. During endodormancy break, LEC1, PHOTOSYSTEM I RC, and brassinosteroids were identified as ooverlapping hubs of up-regulated genes, and DREB1A, CBF2, GPA1, MYC2, BHLH, BZIP, and flavonoids were identified as overlapping hubs of down-regulated genes. Additionally, key genes involved in metabolic activity, chromatin modification, and cross-talk between growth regulators were identified as playing a role during endodormancy maintenance. Plant material, controlled environmental treatments, and vegetative growth: Three-month-old paradormant leafy spurge plants grown in a greenhouse were acclimated for one week, under growth chamber conditions, prior to being subjected to a ramp-down in temperature (27 °C → 10 °C) and photoperiod (16 h → 8 h light) for 12 weeks to induce endodormancy, as previously established (Doğramacı et al. 2010; Foley et al. 2009). Endodormancy status was confirmed by comparing vegetative growth rates of plants with and without a ramp-down treatment. To study the effects of dehydration-stress on vegetative growth from UABs, water was withheld from endodormant plants up to 21 days. After dehydration for 0, 1-, 3-, 7-, 14-, and 21-days, the aerial portion of the plants were decapitated at the soil surface and vegetative growth and floral competence of UABs were monitored under growth-conducive conditions by re-watering the root system. Vegetative shoot growth from six plants was recorded weekly for four weeks, and results of four replications were analyzed using SAS 9.2 (SAS Institute, Cary, NC, 2008) software as described by Doğramacı et al. (2010). Microarray analysis: At the end of each treatment, crown buds were collected, and RNA extraction, cDNA synthesis, fluorescent labeling, microarray hybridization using ~23K element arrays, and spot intensity analyses were performed as previously described by Doğramacı et al. (2010). Based on unexpected vegetative growth results obtained from this study, only microarray data obtained from 0-, 1-, and 3-day dehydration-stressed endodormant UABs, which were compared with microarray data for endodormant, and flowering competent ecodormant UABs from our previous study (Doğramacı et al. 2010; GSE19217), were used for bioinformatics analyses. However, to merge the datasets from these two separate experiments, T-tests were first performed on microarray data obtained from endodormant UABs from each experiment, and differentially-expressed genes (p<0.005) were removed from the dataset (~5% of the spots). After removing the outliers, the two endodormant transcriptome data sets were grouped together and used as the baseline control for further ANOVA, T-tests and other analyses. Array normalization, statistical analyses and clustering of the dataset and Venn diagram generation were done using GeneMaths XT 5.1 software as previously described by Doğramacı et al. (2010).