Creating a biomedical knowledge base by addressing GPT inaccurate responses and benchmarking context.

We created GNQA, a generative pre-trained transformer (GPT) knowledge base driven by a performant retrieval augmented generation (RAG) with a focus on aging, dementia, Alzheimer's and diabetes. We uploaded a corpus of three thousand peer reviewed publications on these topics into the RAG. To address concerns about inaccurate responses and GPT 'hallucinations', we implemented a context provenance tracking mechanism that enables researchers to validate responses against the original material and to get references to the original papers. To assess the effectiveness of contextual information we collected evaluations and feedback from both domain expert users and 'citizen scientists' on the relevance of GPT responses. A key innovation of our study is automated evaluation by way of a RAG assessment system (RAGAS). RAGAS combines human expert assessment with AI-driven evaluation to measure the effectiveness of RAG systems. When evaluating the responses to their questions, human respondents give a "thumbs-up" 76% of the time. Meanwhile, RAGAS scores 90% on answer relevance on questions posed by experts. And when GPT-generates questions, RAGAS scores 74% on answer relevance. With RAGAS we created a benchmark that can be used to continuously assess the performance of our knowledge base. Full GNQA functionality is embedded in the free GeneNetwork.org web service, an open-source system containing over 25 years of experimental data on model organisms and human. The code developed for this study is published under a free and open-source software license at https://git.genenetwork.org/gn-ai/tree/README.md .

Cross-species transcriptomics reveals differential regulation of essential photosynthesis genes in Hirschfeldia incana.

Photosynthesis is the only yield-related trait not yet substantially improved by plant breeding. Previously, we have established H. incana as the model plant for high photosynthetic light-use efficiency (LUE). Now we aim to unravel the genetic basis of this trait in H. incana, potentially contributing to the improvement of photosynthetic LUE in other species. Here, we compare its transcriptomic response to high light with that of Arabidopsis thaliana, Brassica rapa, and Brassica nigra, 3 fellow Brassicaceae members with lower photosynthetic LUE. We built a high-light, high-uniformity growing environment, in which the plants developed normally without signs of stress. We compared gene expression in contrasting light conditions across species, utilizing a panproteome to identify orthologous proteins. In-depth analysis of 3 key photosynthetic pathways showed a general trend of lower gene expression under high-light conditions for all 4 species. However, several photosynthesis-related genes in H. incana break this trend. We observed cases of constitutive higher expression (like antenna protein LHCB8), treatment-dependent differential expression (as for PSBE), and cumulative higher expression through simultaneous expression of multiple gene copies (like LHCA6). Thus, H. incana shows differential regulation of essential photosynthesis genes, with the light-harvesting complex as the first point of deviation. The effect of these expression differences on protein abundance and turnover, and ultimately the high photosynthetic LUE phenotype is relevant for further investigation. Furthermore, this transcriptomic resource of plants fully grown under, rather than briefly exposed to, a very high irradiance, will support the development of highly efficient photosynthesis in crops.

The effect of developmental variation on expression QTLs in a multi parental Caenorhabditis elegans population.

Regulation of gene expression plays a crucial role in developmental processes and adaptation to changing environments. expression quantitative trait locus (eQTL) mapping is a technique used to study the genetic regulation of gene expression using the transcriptomes of recombinant inbred lines (RILs). Typically, the age of the inbred lines at the time of RNA sampling is carefully controlled. This is necessary because the developmental process causes changes in gene expression, complicating the interpretation of eQTL mapping experiments. However, due to genetics and variation in ambient micro-environments, organisms can differ in their "developmental age," even if they are of the same chronological age. As a result, eQTL patterns are affected by developmental variation in gene expression. The model organism Caenorhabditis elegans is particularly suited for studying the effect of developmental variation on eQTL mapping patterns. In a span of days, C. elegans transitions from embryo through 4 larval stages to adult while undergoing massive changes to its transcriptome. Here, we use C. elegans to investigate the effect of developmental age variation on eQTL patterns and present a normalization procedure. We used dynamical eQTL mapping, which includes the developmental age as a cofactor, to separate the variation in development from genotypic variation and explain variation in gene expression levels. We compare classical single marker eQTL mapping and dynamical eQTL mapping using RNA-seq data of ∼200 multi-parental RILs of C. elegans. The results show that (1) many eQTLs are caused by developmental variation, (2) most trans-bands are developmental QTLs, and (3) dynamical eQTL mapping detects additional eQTLs not found with classical eQTL mapping. We recommend that correction for variation in developmental age should be strongly considered in eQTL mapping studies given the large impact of processes like development on the transcriptome.

Plasticity of maternal environment-dependent expression-QTLs of tomato seeds.

Seeds are essential for plant reproduction, survival, and dispersal. Germination ability and successful establishment of young seedlings strongly depend on seed quality and on environmental factors such as nutrient availability. In tomato (Solanum lycopersicum) and many other species, seed quality and seedling establishment characteristics are determined by genetic variation, as well as the maternal environment in which the seeds develop and mature. The genetic contribution to variation in seed and seedling quality traits and environmental responsiveness can be estimated at transcriptome level in the dry seed by mapping genomic loci that affect gene expression (expression QTLs) in contrasting maternal environments. In this study, we applied RNA-sequencing to construct a linkage map and measure gene expression of seeds of a tomato recombinant inbred line (RIL) population derived from a cross between S. lycopersicum (cv. Moneymaker) and S. pimpinellifolium (G1.1554). The seeds matured on plants cultivated under different nutritional environments, i.e., on high phosphorus or low nitrogen. The obtained single-nucleotide polymorphisms (SNPs) were subsequently used to construct a genetic map. We show how the genetic landscape of plasticity in gene regulation in dry seeds is affected by the maternal nutrient environment. The combined information on natural genetic variation mediating (variation in) responsiveness to the environment may contribute to knowledge-based breeding programs aiming to develop crop cultivars that are resilient to stressful environments.

Genetic mechanisms underlying increased microalgal thermotolerance, maximal growth rate, and yield on light following adaptive laboratory evolution.

BACKGROUND: Adaptive laboratory evolution (ALE) is a powerful method for strain optimization towards abiotic stress factors and for identifying adaptation mechanisms. In this study, the green microalga Picochlorum sp. BPE23 was cultured under supra-optimal temperature to force genetic adaptation. The robustness and adaptive capacity of Picochlorum strains turned them into an emerging model for evolutionary studies on abiotic stressors such as temperature, salinity, and light. RESULTS: Mutant strains showed an expanded maximal growth temperature of 44.6 °C, whereas the maximal growth temperature of the wild-type strain was 42 °C. Moreover, at the optimal growth temperature of 38 °C, the biomass yield on light was 22.3% higher, and the maximal growth rate was 70.5% higher than the wild type. Genome sequencing and transcriptome analysis were performed to elucidate the mechanisms behind the improved phenotype. A de novo assembled phased reference genome allowed the identification of 21 genic mutations involved in various processes. Moreover, approximately half of the genome contigs were found to be duplicated or even triplicated in all mutants, suggesting a causal role in adaptation. CONCLUSIONS: The developed tools and mutant strains provide a strong framework from whereupon Picochlorum sp. BPE23 can be further developed. Moreover, the extensive strain characterization provides evidence of how microalgae evolve to supra-optimal temperature and to photobioreactor growth conditions. With this study, microalgal evolutionary mechanisms were identified by combining ALE with genome sequencing.

Prioritizing candidate eQTL causal genes in Arabidopsis using RANDOM FORESTS.

Expression quantitative trait locus mapping has been widely used to study the genetic regulation of gene expression in Arabidopsis thaliana. As a result, a large amount of expression quantitative trait locus data has been generated for this model plant; however, only a few causal expression quantitative trait locus genes have been identified, and experimental validation is costly and laborious. A prioritization method could help speed up the identification of causal expression quantitative trait locus genes. This study extends the machine-learning-based QTG-Finder2 method for prioritizing candidate causal genes in phenotype quantitative trait loci to be used for expression quantitative trait loci by adding gene structure, protein interaction, and gene expression. Independent validation shows that the new algorithm can prioritize 16 out of 25 potential expression quantitative trait locus causal genes within the top 20% rank. Several new features are important in prioritizing causal expression quantitative trait locus genes, including the number of protein-protein interactions, unique domains, and introns. Overall, this study provides a foundation for developing computational methods to prioritize candidate expression quantitative trait locus causal genes. The prediction of all genes is available in the AraQTL workbench (https://www.bioinformatics.nl/AraQTL/) to support the identification of gene expression regulators in Arabidopsis.

Short-term physiologic response of the green microalga Picochlorum sp. (BPE23) to supra-optimal temperature.

Photobioreactors heat up significantly during the day due to irradiation by sunlight. High temperatures affect cell physiology negatively, causing reduced growth and productivity. To elucidate the microalgal response to stressful supra-optimal temperature, we studied the physiology of Picochlorum sp. (BPE23) after increasing the growth temperature from 30 °C to 42 °C, whereas 38 °C is its optimal growth temperature. Cell growth, cell composition and mRNA expression patterns were regularly analyzed for 120 h after increasing the temperature. The supra-optimal temperature caused cell cycle arrest for 8 h, with concomitant changes in metabolic activity. Accumulation of fatty acids was observed during this period to store unspent energy which was otherwise used for growth. In addition, the microalgae changed their pigment and fatty acid composition. For example, palmitic acid (C16:0) content in the polar fatty acid fraction increased by 30%, hypothetically to reduce membrane fluidity to counteract the effect of increased temperature. After the relief of cell cycle arrest, the metabolic activity of Picochlorum sp. (BPE23) reduced significantly over time. A strong response in gene expression was observed directly after the increase in temperature, which was dampened in the remainder of the experiment. mRNA expression levels associated with pathways associated with genes acting in photosynthesis, carbon fixation, ribosome, citrate cycle, and biosynthesis of metabolites and amino acids were downregulated, whereas the proteasome, autophagy and endocytosis were upregulated.

Genomic prediction in plants: opportunities for ensemble machine learning based approaches.

Background: Many studies have demonstrated the utility of machine learning (ML) methods for genomic prediction (GP) of various plant traits, but a clear rationale for choosing ML over conventionally used, often simpler parametric methods, is still lacking. Predictive performance of GP models might depend on a plethora of factors including sample size, number of markers, population structure and genetic architecture. Methods: Here, we investigate which problem and dataset characteristics are related to good performance of ML methods for genomic prediction. We compare the predictive performance of two frequently used ensemble ML methods (Random Forest and Extreme Gradient Boosting) with parametric methods including genomic best linear unbiased prediction (GBLUP), reproducing kernel Hilbert space regression (RKHS), BayesA and BayesB. To explore problem characteristics, we use simulated and real plant traits under different genetic complexity levels determined by the number of Quantitative Trait Loci (QTLs), heritability ( h 2 and h 2 e ), population structure and linkage disequilibrium between causal nucleotides and other SNPs. Results: Decision tree based ensemble ML methods are a better choice for nonlinear phenotypes and are comparable to Bayesian methods for linear phenotypes in the case of large effect Quantitative Trait Nucleotides (QTNs). Furthermore, we find that ML methods are susceptible to confounding due to population structure but less sensitive to low linkage disequilibrium than linear parametric methods. Conclusions: Overall, this provides insights into the role of ML in GP as well as guidelines for practitioners.

Delayed Protein Changes During Seed Germination.

Over the past decade, ample transcriptome data have been generated at different stages during seed germination; however, far less is known about protein synthesis during this important physiological process. Generally, the correlation between transcript levels and protein abundance is low, which strongly limits the use of transcriptome data to accurately estimate protein expression. Polysomal profiling has emerged as a tool to identify mRNAs that are actively translated. The association of the mRNA to the polysome, also referred to as translatome, provides a proxy for mRNA translation. In this study, the correlation between the changes in total mRNA, polysome-associated mRNA, and protein levels across seed germination was investigated. The direct correlation between polysomal mRNA and protein abundance at a single time-point during seed germination is low. However, once the polysomal mRNA of a time-point is compared to the proteome of the next time-point, the correlation is much higher. 35% of the investigated proteome has delayed changes at the protein level. Genes have been classified based on their delayed protein changes, and specific motifs in these genes have been identified. Moreover, mRNA and protein stability and mRNA length have been found as important predictors for changes in protein abundance. In conclusion, polysome association and/or dissociation predicts future changes in protein abundance in germinating seeds.

The genetics of gene expression in a Caenorhabditis elegans multiparental recombinant inbred line population.

Studying genetic variation of gene expression provides a powerful way to unravel the molecular components underlying complex traits. Expression quantitative trait locus (eQTL) studies have been performed in several different model species, yet most of these linkage studies have been based on the genetic segregation of two parental alleles. Recently, we developed a multiparental segregating population of 200 recombinant inbred lines (mpRILs) derived from four wild isolates (JU1511, JU1926, JU1931, and JU1941) in the nematode Caenorhabditis elegans. We used RNA-seq to investigate how multiple alleles affect gene expression in these mpRILs. We found 1789 genes differentially expressed between the parental lines. Transgression, expression beyond any of the parental lines in the mpRILs, was found for 7896 genes. For expression QTL mapping almost 9000 SNPs were available. By combining these SNPs and the RNA-seq profiles of the mpRILs, we detected almost 6800 eQTLs. Most trans-eQTLs (63%) co-locate in six newly identified trans-bands. The trans-eQTLs found in previous two-parental allele eQTL experiments and this study showed some overlap (17.5-46.8%), highlighting on the one hand that a large group of genes is affected by polymorphic regulators across populations and conditions, on the other hand, it shows that the mpRIL population allows identification of novel gene expression regulatory loci. Taken together, the analysis of our mpRIL population provides a more refined insight into C. elegans complex trait genetics and eQTLs in general, as well as a starting point to further test and develop advanced statistical models for detection of multiallelic eQTLs and systems genetics studying the genotype-phenotype relationship.

Network Analysis Prioritizes DEWAX and ICE1 as the Candidate Genes for Major eQTL Hotspots in Seed Germination of Arabidopsis thaliana.

Seed germination is characterized by a constant change of gene expression across different time points. These changes are related to specific processes, which eventually determine the onset of seed germination. To get a better understanding on the regulation of gene expression during seed germination, we performed a quantitative trait locus mapping of gene expression (eQTL) at four important seed germination stages (primary dormant, after-ripened, six-hour after imbibition, and radicle protrusion stage) using Arabidopsis thaliana Bay x Sha recombinant inbred lines (RILs). The mapping displayed the distinctness of the eQTL landscape for each stage. We found several eQTL hotspots across stages associated with the regulation of expression of a large number of genes. Interestingly, an eQTL hotspot on chromosome five collocates with hotspots for phenotypic and metabolic QTL in the same population. Finally, we constructed a gene co-expression network to prioritize the regulatory genes for two major eQTL hotspots. The network analysis prioritizes transcription factors DEWAX and ICE1 as the most likely regulatory genes for the hotspot. Together, we have revealed that the genetic regulation of gene expression is dynamic along the course of seed germination.

Detection of QTLs for genotype × environment interactions in tomato seeds and seedlings.

Seed quality and seedling establishment are the most important factors affecting successful crop development. They depend on the genetic background and are acquired during seed maturation and therefor, affected by the maternal environment under which the seeds develop. There is little knowledge about the genetic and environmental factors that affect seed quality and seedling establishment. The aim of this study is to identify the loci and possible molecular mechanisms involved in acquisition of seed quality and how these are controlled by adverse maternal conditions. For this, we used a tomato recombinant inbred line (RIL) population consisting of 100 lines which were grown under two different nutritional environmental conditions, high phosphate and low nitrate. Most of the seed germination traits such as maximum germination percentage (Gmax ), germination rate (t50 ) and uniformity (U8416 ) showed ample variation between genotypes and under different germination conditions. This phenotypic variation leads to identification of quantitative trait loci (QTLs) which were dependent on genetic factors, but also on the interaction with the maternal environment (QTL × E). Further studies of these QTLs may ultimately help to predict the effect of different maternal environmental conditions on seed quality and seedling establishment which will be very useful to improve the production of high-performance seeds.

WormQTL2: an interactive platform for systems genetics in Caenorhabditis elegans.

Quantitative genetics provides the tools for linking polymorphic loci to trait variation. Linkage analysis of gene expression is an established and widely applied method, leading to the identification of expression quantitative trait loci (eQTLs). (e)QTL detection facilitates the identification and understanding of the underlying molecular components and pathways, yet (e)QTL data access and mining often is a bottleneck. Here, we present WormQTL2, a database and platform for comparative investigations and meta-analyses of published (e)QTL data sets in the model nematode worm C. elegans. WormQTL2 integrates six eQTL studies spanning 11 conditions as well as over 1000 traits from 32 studies and allows experimental results to be compared, reused and extended upon to guide further experiments and conduct systems-genetic analyses. For example, one can easily screen a locus for specific cis-eQTLs that could be linked to variation in other traits, detect gene-by-environment interactions by comparing eQTLs under different conditions, or find correlations between QTL profiles of classical traits and gene expression. WormQTL2 makes data on natural variation in C. elegans and the identified QTLs interactively accessible, allowing studies beyond the original publications. Database URL: www.bioinformatics.nl/WormQTL2/.

Extracting the GEMs: Genotype, Environment, and Microbiome Interactions Shaping Host Phenotypes.

One of the fundamental tenets of biology is that the phenotype of an organism (Y) is determined by its genotype (G), the environment (E), and their interaction (GE). Quantitative phenotypes can then be modeled as Y = G + E + GE + e, where e is the biological variance. This simple and tractable model has long served as the basis for studies investigating the heritability of traits and decomposing the variability in fitness. The importance and contribution of microbe interactions to a given host phenotype is largely unclear, nor how this relates to the traditional GE model. Here we address this fundamental question and propose an expansion of the original model, referred to as GEM, which explicitly incorporates the contribution of the microbiome (M) to the host phenotype, while maintaining the simplicity and tractability of the original GE model. We show that by keeping host, environment, and microbiome as separate but interacting variables, the GEM model can capture the nuanced ecological interactions between these variables. Finally, we demonstrate with an in vitro experiment how the GEM model can be used to statistically disentangle the relative contributions of each component on specific host phenotypes.

Prior Biological Knowledge Improves Genomic Prediction of Growth-Related Traits in Arabidopsis thaliana.

Prediction of growth-related complex traits is highly important for crop breeding. Photosynthesis efficiency and biomass are direct indicators of overall plant performance and therefore even minor improvements in these traits can result in significant breeding gains. Crop breeding for complex traits has been revolutionized by technological developments in genomics and phenomics. Capitalizing on the growing availability of genomics data, genome-wide marker-based prediction models allow for efficient selection of the best parents for the next generation without the need for phenotypic information. Until now such models mostly predict the phenotype directly from the genotype and fail to make use of relevant biological knowledge. It is an open question to what extent the use of such biological knowledge is beneficial for improving genomic prediction accuracy and reliability. In this study, we explored the use of publicly available biological information for genomic prediction of photosynthetic light use efficiency (Φ PSII ) and projected leaf area (PLA) in Arabidopsis thaliana. To explore the use of various types of knowledge, we mapped genomic polymorphisms to Gene Ontology (GO) terms and transcriptomics-based gene clusters, and applied these in a Genomic Feature Best Linear Unbiased Predictor (GFBLUP) model, which is an extension to the traditional Genomic BLUP (GBLUP) benchmark. Our results suggest that incorporation of prior biological knowledge can improve genomic prediction accuracy for both Φ PSII and PLA. The improvement achieved depends on the trait, type of knowledge and trait heritability. Moreover, transcriptomics offers complementary evidence to the Gene Ontology for improvement when used to define functional groups of genes. In conclusion, prior knowledge about trait-specific groups of genes can be directly translated into improved genomic prediction.

Dormancy cycling: translation-related transcripts are the main difference between dormant and non-dormant seeds in the field.

Primary seed dormancy is a mechanism that orchestrates the timing of seed germination in order to prevent out-of-season germination. Secondary dormancy can be induced in imbibed seeds when they encounter prolonged unfavourable conditions. Secondary dormancy is not induced during dry storage, and therefore the mechanisms underlying this process have remained largely unexplored. Here, a 2-year seed burial experiment in which dormancy cycling was studied at the physiological and transcriptional level is presented. For these analyses six different Arabidopsis thaliana genotypes were used: Landsberg erecta (Ler) and the dormancy associated DELAY OF GERMINATION (DOG) near-isogenic lines 1, 2, 3, 6 and 22 (NILDOG1, 2, 3, 6 and 22). The germination potential of seeds exhumed from the field showed that these seeds go through dormancy cycling and that the dynamics of this cycling is genotype dependent. RNA-seq analysis revealed large transcriptional changes during dormancy cycling, especially at the time points preceding shifts in dormancy status. Dormancy cycling is driven by soil temperature and the endosperm is important in the perception of the environment. Genes that are upregulated in the low- to non-dormant stages are enriched for genes involved in translation, indicating that the non-dormant seeds are prepared for rapid seed germination.

A genetical metabolomics approach for bioprospecting plant biosynthetic gene clusters.

OBJECTIVE: Plants produce a plethora of specialized metabolites to defend themselves against pathogens and insects, to attract pollinators and to communicate with other organisms. Many of these are also applied in the clinic and in agriculture. Genes encoding the enzymes that drive the biosynthesis of these metabolites are sometimes physically grouped on the chromosome, in regions called biosynthetic gene clusters (BGCs). Several algorithms have been developed to identify plant BGCs, but a large percentage of predicted gene clusters upon further inspection do not show coexpression or do not encode a single functional biosynthetic pathway. Hence, further prioritization is needed. RESULTS: Here, we introduce a strategy to systematically evaluate potential functions of predicted BGCs by superimposing their locations on metabolite quantitative trait loci (mQTLs). We show the feasibility of such an approach by integrating automated BGC prediction with mQTL datasets originating from a recombinant inbred line (RIL) population of Oryza sativa and a genome-wide association study (GWAS) of Arabidopsis thaliana. In these data, we identified several links for which the enzyme content of the BGCs matches well with the chemical features observed in the metabolite structure, suggesting that this method can effectively guide bioprospecting of plant BGCs.

A multi-parent recombinant inbred line population of C. elegans allows identification of novel QTLs for complex life history traits.

BACKGROUND: The nematode Caenorhabditis elegans has been extensively used to explore the relationships between complex traits, genotypes, and environments. Complex traits can vary across different genotypes of a species, and the genetic regulators of trait variation can be mapped on the genome using quantitative trait locus (QTL) analysis of recombinant inbred lines (RILs) derived from genetically and phenotypically divergent parents. Most RILs have been derived from crossing two parents from globally distant locations. However, the genetic diversity between local C. elegans populations can be as diverse as between global populations and could thus provide means of identifying genetic variation associated with complex traits relevant on a broader scale. RESULTS: To investigate the effect of local genetic variation on heritable traits, we developed a new RIL population derived from 4 parental wild isolates collected from 2 closely located sites in France: Orsay and Santeuil. We crossed these 4 genetically diverse parental isolates to generate a population of 200 multi-parental RILs and used RNA-seq to obtain sequence polymorphisms identifying almost 9000 SNPs variable between the 4 genotypes with an average spacing of 11 kb, doubling the mapping resolution relative to currently available RIL panels for many loci. The SNPs were used to construct a genetic map to facilitate QTL analysis. We measured life history traits such as lifespan, stress resistance, developmental speed, and population growth in different environments, and found substantial variation for most traits. We detected multiple QTLs for most traits, including novel QTLs not found in previous QTL analysis, including those for lifespan and pathogen responses. This shows that recombining genetic variation across C. elegans populations that are in geographical close proximity provides ample variation for QTL mapping. CONCLUSION: Taken together, we show that using more parents than the classical two parental genotypes to construct a RIL population facilitates the detection of QTLs and that the use of wild isolates facilitates the detection of QTLs. The use of multi-parent RIL populations can further enhance our understanding of local adaptation and life history trade-offs.

Induction of desiccation tolerance in desiccation sensitive Citrus limon seeds.

Many economically important perennial species bear recalcitrant seeds, including tea, coffee, cocoa, mango, citrus, rubber, oil palm and coconut. Orthodox seeds can be dried almost completely without losing viability, but so-called recalcitrant seeds have a very limited storage life and die upon drying below a higher critical moisture content than orthodox seeds. As a result, the development of long-term storage methods for recalcitrant seeds is compromised. Lowering this critical moisture content would be very valuable since dry seed storage is the safest, most convenient and cheapest method for conserving plant genetic resources. Therefore, we have attempted to induce desiccation tolerance (DT) in the desiccation sensitive seeds of Citrus limon. We show that DT can be induced by paclobutrazol (an inhibitor of gibberellin biosynthesis) and we studied its associated transcriptome to delineate the molecular mechanisms underlying this induction of DT. Paclobutrazol not only interfered with gibberellin related gene expression but also caused extensive changes in expression of genes involved in the biosynthesis and signaling of other hormones. Paclobutrazol induced a transcriptomic switch encompassing suppression of biotic- and induction of abiotic responses. We hypothesize that this is the main driver of the induction of DT by paclobutrazol in C. limon seeds.

Tulipa gesneriana and Lilium longiflorum PEBP Genes and Their Putative Roles in Flowering Time Control.

Floral induction in Tulipa gesneriana and Lilium longiflorum is triggered by contrasting temperature conditions, high and low temperature, respectively. In Arabidopsis, the floral integrator FLOWERING LOCUS T (FT), a member of the PEBP (phosphatidyl ethanolamine-binding protein) gene family, is a key player in flowering time control. In this study, one PEBP gene was identified and characterized in lily (LlFT) and three PEBP genes were isolated from tulip (TgFT1, TgFT2 and TgFT3). Overexpression of these genes in Arabidopsis thaliana resulted in an early flowering phenotype for LlFT and TgFT2, but a late flowering phenotype for TgFT1 and TgFT3. Overexpression of LlFT in L. longiflorum also resulted in an early flowering phenotype, confirming its proposed role as a flowering time-controlling gene. The tulip PEBP genes TgFT2 and TgFT3 have a similar expression pattern in tulip, but show opposite effects on the timing of flowering in Arabidopsis. Therefore, the difference between these two proteins was further investigated by interchanging amino acids thought to be important for the FT function. This resulted in the conversion of phenotypes in Arabidopsis upon overexpressing the substituted TgFT2 and TgFT3 genes, revealing the importance of these interchanged amino acid residues. Based on all obtained results, we hypothesize that LlFT is involved in creating meristem competence to flowering-related cues in lily, and TgFT2 is considered to act as a florigen involved in the floral induction in tulip. The function of TgFT3 remains unclear, but, based on our observations and phylogenetic analysis, we propose a bulb-specific function for this gene.