Abstract
The decision of whether to grow and proliferate or to restrict growth and develop resilience to stress is a key biological trade-off. In plants, constitutive growth results in increased sensitivity to environmental stress1,2. The underlying mechanisms controlling this decision are however not well understood. We used temperature as a cue to discover regulators of this process in plants, as it both enhances growth and development rates within a specific range and is also a stress at extremes. We found that the conserved chromatin-associated protein DEK plays a central role in balancing the response between growth and arrest in Arabidopsis, and it does this via H2A.Z-nucleosomes. DEK target genes show two distinct categories of chromatin architecture based on the distribution of H2A.Z in +1 nucleosome and gene body, and these predict induction or repression by DEK. We show that these chromatin signatures of DEK target genes are conserved in human cells, suggesting that DEK may act through an evolutionarily conserved mechanism to control the balance between growth and arrest in plants and animals.
Main
Plants are exposed to daily fluctuations in ambient temperature, which influence growth, development and fitness3. Being sessile organisms, plants frequently need to opt between growth and stress resilience, making them good systems to analyse how these trade-offs are made.
Recently, it has become clear that the chromatin landscape—i.e. the positioning of the nucleosome containing histone variants such as H2A.Z and H3.3 across the gene body—is correlated to the responsiveness of a gene to environmental variation3–7. For instance, hypersensitive environmental genes are enriched in H2A.Z-nucleosomes in their gene bodies, suggesting its involvement in the regulation of environmental response4,7. Warm temperature results in H2A.Z-nucleosome eviction and large-scale transcriptional activation in plants8,9. Furthermore, H2A.Z is removed from +1 nucleosomes of temperature-induced genes upon temperature increase allowing activation of expression by transcription factors9. The molecular mechanism underlying this response likely involves a re-organisation of the chromatin landscape by chromatin remodelling enzymes. In this paper, we demonstrate that DEK-DOMAIN CONTAINING PROTEIN3 (DEK3), the most abundant member of a family of four DEK-domain containing chromatin re-modellers in Arabidopsis10, is a link between the chromatin landscape and environmental responsiveness.
The mammalian orthologue DEK is an oncoprotein involved in the development of cancer, inflammation and stem cell biology11,12. Because of its key role in cancer, DEK has been intensely studied in animals, and it has been described to have roles as an H3.3 chaperone, co-transcriptional regulator and in splicing13–16. Despite these studies, the genome-wide targets directly regulated by DEK are not known17, and it is also unclear whether DEK serves as an activator or inhibitor of transcription. In Drosophila and humans DEK can serve as a H3.3 chaperone, and it remodels nucleosomes into more transcriptionally active chromatin through its histone chaperone activity15. By contrast, in mammalian cell lines DEK acts as a positive regulator of heterochromatin formation, maintaining the balance between heterochromatin and euchromatin in vivo14. DEK therefore appears to act as either an activator or a repressor depending on the context.
In Arabidopsis, DEK was identified as a component of the nucleolus by mass spectrometry16. DEK3 can change the topology of protein free DNA in vitro and cause transcriptional repression of specific loci by increasing nucleosome density10. Correct DEK3 expression is critical for the degree of response of the plants to some stress conditions, including high salinity and heat shock. Plants with constitutively elevated levels of DEK3 are more sensitive to high salinity, whereas plants deficient in DEK3 are more salt tolerant10.
To further understand the global effects of DEK3 on gene expression in plants, we used chromatin immunoprecipitation DNA-sequencing (ChIP-seq) and RNA-seq to investigate how DEK3 occupancy influences the responsiveness of gene expression to temperature. We show that DEK3 affects temperature-dependent biological trade-offs in Arabidopsis, through the up-regulation of stress response genes and by the suppression of thermo-responsive induction of growth and development genes. Furthermore, inherent chromatin landscape features are sufficient to predict whether a gene will be up- or down-regulated by overexpression of DEK3. Additionally, we show for the first time that DEK3 genetically and physically interacts with H2A.Z and modifies H2A.Z-nucleosome distribution within the gene bodies of DEK3 target genes. We suggest a model whereby feedback between the chromatin landscape and chromatin re-modellers can affect trade-off between growth and arrest, and therefore influence developmental plasticity.
Results
H2A.Z and DEK interact
Since increasing temperature does not cause H2A.Z-nucleosome loss in vitro9, we sought to determine if other factors may contribute to this process in plants. To find H2A.Z interacting proteins, we performed H2A.Z affinity purification (from HTA11::HTA11-FLAG expressing lines) coupled with mass spectrometry. This approach identified three homologous DEK-domain containing chromatin re-modellers and 6 previously reported H2A.Z interactors18 among 263 proteins found in complex with H2A.Z in vivo (Supp. Fig. 1a, Supp. Table.1). Two of them, DEK3 and DEK2, have been found among 7 proteins enriched significantly in samples collected at 27°C compared to 17°C (Fig. 1a, Supp. Table.1) suggesting a possible role in temperature pathway regulated through H2A.Z nucleosomes. There are four DEK-domain containing proteins in Arabidopsis thaliana, among them, DEK3 shows a strong and abundant expression in all plant organs10. Interestingly, DEK3 shows one of the highest fold change when comparing the binding affinities to H2A.Z complexes between 27°C and 17°C (Fig. 1a, Supp. Table 1). We confirmed the interaction of DEK3 with H2A.Z by co-immunoprecipitating H2A.Z-containing nucleosomes and H2A.Z protein (purified from plants), with purified DEK3 (Fig. 1b). This validation set up mimics in vitro binding experiments while preserving post-translational modifications of H2A.Z and DEK3. We find that H2A.Z purified from histone extract coimmunoprecipitated with DEK3 in the comparable levels to H2A.Z nucleosomes obtained from nuclear extract, despite reduced levels of other histones as determined by H3 levels (Fig. 1b). This suggests that other protein factors might not be necessary for the interaction between these two proteins.
Previously, immunopurification of DEK3 has identified its interaction with histones H3 and H4, but not with H2A and H2B10. The use of an antibody directed against all H2A variants in this previous study made the detection of H2A.Z very difficult as H2A.Z levels are only approximately 10 % of those of H2A19.
Perturbing DEK expression levels affects growth and stress response
Temperature has a strong effect on plant growth and development. Growth in high ambient temperatures (below the threshold of inducing widespread heat stress), results in faster growth and accelerated development (thermomorphogenesis), illustrated by increased hypocotyl length and early flowering20. In contrast, heat stress induces metabolic imbalance, accumulation of toxic by-products and, adversely influencing reproductive growth and yield quality21. Plant resilience to stress conditions such as high salinity and heat shock requires the correct level of DEK310. Additionally, a DEK3 null allele (dek3-2) shows an exaggerated response to warmer (non-stressful) temperature, while overexpression of DEK3 reduces the thermal responsiveness of seedlings, both as measured by elongation of hypocotyl and flowering time (Fig. 1c-d, Supp. Fig. 1b-c). The altered responsiveness of DEK3 mis-expressing plants to temperature changes appears to be a general property, since DEK3 overexpressors are also unable to acclimate to cold temperature and show increased sensitivity to cold stress, applied with and without acclimation (Fig. 1e). dek3-2 plants are less sensitive to freezing without acclimation (Fig. 1e, middle panel), suggesting a key role of DEK3 in cold resistance pathways.
We observe similarly altered hypocotyl growth in dek2 and dek4 mutants, suggesting a general role for this family of genes in controlling response to temperature (Supp. Fig. 1d).
To further understand why plants with abnormal DEK3 protein levels exhibit altered temperature responses, we investigated gene expression in dek3-2 and 35S::DEK3 plants at 17 °C and 27 °C. Since the ambient temperature transcriptome is dynamic22, we sampled the transcriptome over the 24 h day-night cycle (Supp. Fig. 1e-g). Consistent with plant phenotypes, the largest perturbation in gene expression occurs in 35S::DEK3 at 27 °C (Supp. Fig. 1e). The overexpression of DEK3 caused ∼4,860 genes to be miss-expressed in at least one time-point compared to WT plants when grown at warm temperature (Supp. table 2).
We used Principal Component Analysis (PCA) to further understand differences in gene expression. Principle Components (PC) 1 and 2 together could explain 54% of the gene expression variance and primarily separate the samples based on time of day into day and night time samples (Supp. Fig. 1f). Among the night samples, the expression of differentially expressed genes in 35S::DEK3 plants at 27 °C resembles the 17 °C transcriptomes of all genotypes rather than that of warm temperature transcriptomes (Supp. Fig. 1f). Indeed, this is consistent with the phenotype of 35S::DEK3 plants, which at 27 °C show perturbed hypocotyl elongation and flowering similar to the plants grown at 17 °C (Fig. 1c-d, Supp. Fig. 1b-c).
Hierarchical clustering of differentially expressed genes reveals three major categories: the first group includes warm temperature induced genes in Col-0, whose expression was induced during the day (first group: clusters 2 and 3) or during the night (Second group: clusters 6 and 7), and genes whose expression was specifically induced only by DEK3 overexpression at 27 °C (third group: clusters 4 and 5) (Supp. Fig. 1g).
The first group is highly enriched for different GO terms related to metabolism, translation and ribosome biogenesis, RNA methylation, nucleosome assembly and histone modifications (Supp. Table 2); the second group of genes is highly enriched for GO terms connected to shoot and meristem development, regulation of growth, metabolism, transcription (RNA elongation and gene silencing) and photosynthesis (Supp. Table 2). This cluster contains genes encoding heat shock proteins such as HSP70, as well as genes related to hypocotyl growth, like LHY. Genes in this group are more expressed in dek3-2 toward the end of the night; Genes in the third group respond only in 35S::DEK3, and are specifically enriched in GO terms related to response to biotic stimuli and immune response (Supp. Fig. 1g, Supp. Table 2).
DEK3 direct targets can be activated or repressed by elevated DEK3 expression
Since the effects of DEK3 on transcription may be indirect, we performed CHIP-seq of DEK3-CFP expressed under its native promoter in dek3-2 mutant plants grown at 17 °C and 27 °C at the end of the night and day. These time points are the most distinct according to a PCA analysis (Supp. Fig. 1f), allowing us to capture the maximal diversity in DEK3 behaviour. We observed DEK3 binding primarily in active chromatin states (Supp. Fig. 2a, the chromatin states 1, 3 and 7) characterized by open chromatin and highly correlated with mRNA-encoding genes23.
Since DEK3 primarily binds to DNA in shallow, broad peaks (not sharp peaks like a transcription factor), a peak calling approach was not appropriate for analyzing DEK3 binding profiles. Instead, we identified targeted genes by clustering the DEK3 profiles over the gene bodies. DEK3 can be detected in the majority of gene bodies with defined boundaries at the beginning and the end of the genes (Fig. 2a, Supp. Fig. 2b). For many genes, there is an enrichment for DEK3 in the 3’ end of the gene (Fig. 2a, clusters 5 - 7), which is consistent with its role as an H3.3 chaperone in Drosophila and humans14,15. Since overexpression of DEK3 protein caused such a dramatic effect on plants physiology and transcriptome at warm temperatures (Fig. 1c-e, Supp. Fig. 1b-c, 1e-g), we checked its chromatin binding profiles and compared them to the profiles of the native DEK3 (Supp. Fig. 2b). Since native and overexpressed DEK3 bind widely throughout the genome (Fig. 2a, Supp. Fig. 2b) it is not possible to make a quantitative comparison in their binding to chromatin, however, it is clear that they have different binding patterns. Overexpression of DEK3 results in extension of binding beyond the ends of the gene body (Supp. Fig. 2b) which might contribute to the altered gene expression seen in 35S::DEK3 (Fig. 1b-d, Supp. Fig.1b-c). We do not observe a significant difference in the binding of DEK3 to chromatin that can be associated with temperature or the time of day (Fig. 2a, Supp. Fig. 2b), suggesting that other factors may be involved in temperature dependent gene expression pattern of DEK3 targets during the day. Hence, we used all four DEK3 ChIP-seq datasets from DEK3::DEK3-CFP dek3-2 plants as biological replicates for the following analysis in order to determine DEK3 direct targets controlled by temperature.
We identified 2079 genes as potential direct DEK3 targets (Fig. 2b), as defined by having a high DEK3 occupancy in both temperatures (clusters 3-5 in Fig. 2a) and being differentially expressed at least at one time point at 27 °C when DEK3 is perturbed (Supp. Table 2).
Since we observe the highest influence of DEK3 levels on gene expression based on PCA and hierarchical cluster analysis at ZT0 (Supp. Fig. 1f-g), we further analysed this timepoint. More than half of the potential direct DEK3 targets (1048 out of the 2079 genes) were differentially expressed at this time point (Supp. Table 3). To analyse the mechanism of DEK3 transcription regulation at this time point, we focused on these 1048 genes for further analysis.
These DEK3 targets can be separated into three groups with distinct transcriptional profiles (Fig. 2c, Supp. Table 3): Genes in Cluster 1 show greatly enhanced expression at 27 °C in 35S::DEK3 compared to Col-0 and dek3-2 (Fig. 2c). Since these genes are already expressed at low levels in Col-0, it would be challenging to detect a significant reduction in expression in dek3-2 compared to WT plants (Fig. 2c).
This group is dominated by seven members of the AP2-type transcription factors, mainly of the ETHYLENE RESPONSE FACTOR (ERF) class whose expression is induced by DEK3 overexpression in combination with temperature (Fig. 2d). Overall, the genes in this cluster are highly enriched for stress genes (Supp. Fig. 2c). The miss-expression of the stress transcriptome likely contributes to the observed enhanced sensitivity of 35S::DEK3 to freezing (Fig. 1e).
Cluster 2 contains genes induced during the night at 27 °C in Col-0. These genes are up-regulated in dek3-2 earlier in the night compared to Col-0, while not induced at all when DEK3 is overexpressed. For example, the temperature responsive auxin biosynthesis gene YUCCA8 that is necessary for hypocotyl elongation at 27 °C is directly repressed by DEK3. Prominent transcription factors in cluster 2 include the AUXIN RESPONSE FACTORS (ARF) 1 and 19, and the GATA transcription factors 2, 5 and 9. Many genes also implicated in auxin signalling, for example NAKED PINS IN YUCCA MUTANT (NPY) 1, 3 and 5 are also repressed by DEK3. At the same time these genes are induced even more during the night in dek3-2 plants in temperature dependent manner (Fig. 2d). Overall there is strong enrichment for the GO-terms associated with growth, development and auxin (Supp. Fig. 2c), indicating that influence on growth caused by DEK3 perturbations at 27 °C is direct. The final cluster is heterogeneous, but it is the only cluster that contains genes that decrease their expression levels at elevated temperatures.
DEK3 affects growth and survival in salinity stress10, heat stress10, elevated ambient temperature and cold stress (Fig. 1c-e, Supp. Fig. 1b-c), indicating that DEK3 controls genes playing a role in promoting stress responses and inhibiting growth in a broad range of conditions. Consistent with this, using publicly available transcriptomic data from the AtGenExpress consortium, we found that the Cluster 1 genes were up-regulated and the Cluster 2 were down-regulated in a wide range of abiotic stress conditions including UV-B, salt, osmotic, cold, wounding, heat, drought, genotoxic and oxidative stresses24 (Supp. Fig. 4a).
The activation or repression of DEK3 targets is related to gene body H2A.Z
DEK3 mediates the correct expression of a set of genes that are widely responsive to abiotic stresses, while simultaneously repressing the induction of genes promoting growth (Fig. 2c, Supp. Fig. 2c). This raises the question of how a single factor can both activate and repress gene expression in a locus specific manner. This behaviour is similar to that reported in mammals, where DEK behaves as repressor in some cases and as an activator in others17,25.
Since chromatin state is correlated with the degree of gene responsiveness4,6,7, we investigated whether the binding pattern of H2A.Z, H3.3 and DNA accessibility were predictive of whether a direct DEK3 target would be activated (“stress” genes) or repressed (“growth” genes). To do this, we built a conditional decision tree model that would take the distribution of DEK3, H2A.Z and H3.3 (ChIP-seq) and the DNA accessibility (MNase-seq) in WT as an input, and trained a model to predict whether a DEK3 bound gene would be up- or down-regulated in the 35S::DEK3 plants at 27 °C at the end of the night (Fig. 3a, Supp. Fig. 3a-d), where the biggest difference in expression could be observed (Supp. Fig. 1f).
Strikingly, the most important feature that distinguished between repressed genes (referred to as the “growth” genes) and induced genes (referred to as the “stress” genes) was the gene-body distribution of H2A.Z (p<0.001) with most DEK3 up-regulated “stress” genes having H2A.Z in the gene body, whereas the repressed “growth” genes did not have H2A.Z in the gene body. These “growth” genes predominantly have +1 H2A.Z nucleosomes (Fig. 3a), some “stress” genes were found to have both gene body H2A.Z and strong +1 H2A.Z. H3.3 and chromatin occupancy (via MNase-seq) also contribute to differences in “growth”, and “stress” genes. Specifically, genes with lower H2A.Z levels, but open promoter regions are more likely to be “stress” genes (Fig. 3a, node 2 and genes with lower levels of H3.3 are more likely to be “growth” genes (Fig. 3a, node 6).
While the level of DEK3 binding does not determine whether a gene will have H2A.Z in the gene body (Fig. 3a-b, Supp. Fig. 3e), perturbing DEK3 expression consistently changes the distribution of H2A.Z incorporation around the +1 nucleosome and gene body (Fig. 3c-d, Supp. Fig. 3f). In dek3-2, there is a larger ratio of gene body H2A.Z in “growth” genes compared to WT plants (Fig. 3c-d (lower panel), Supp. Fig. 3f (lower panel)), which is consistent with these genes being more easily induced in dek3-2 compared to Col-0 at 27 °C (Supp. Fig. 4b (upper panel)). This might explain the phenotypic differences between these plants (Fig. 1c-d, Supp. Fig. 1b-c). Conversely, there is lower enrichment for H2A.Z in both the +1 position and gene bodies of “stress” related genes in plants over-expressing DEK3 compared to WT background (Fig. 3c-d (upper panel), Supp. Fig. 3f (upper panel)).
It has been shown that transcription of genes with H2A.Z in gene bodies is more easily induced by environmental cues and stress4,7—these are the genes that are induced by overexpression of DEK3 and temperature (Fig. 3a, node 1, Supp. Fig. 4b (lower panel)). In contrast, genes that have only +1 H2A.Z nucleosomes and H3.3 around TTS, are expected not to be easily induced by environment but correlated with growth4. These genes are suppressed by overexpression of DEK3 and induced in dek3 in a temperature specific way manner compared to Col-0 (Supp. Fig. 4b (upper panel)).
These observations are consistent with the interaction of DEK3 with H2A.Z (Fig.1a-b, Supp. Fig. 1a), and with decreased expression of the “growth” genes and increased expression of the “stress” related genes in the absence of H2A.Z (Fig. 4d-e) or when its incorporation into chromatin is reduced (Fig. 4f-g). Given that H2A.Z is known to accumulate in the gene body of environmentally responsive genes (such as in drought stress responsive genes)4,7,6, a relationship between DEK3 and H2A.Z may suggest a mechanism by which DEK3 perturbations could affect H2A.Z distribution in +1 nucleosomes and gene bodies, and in this way control inducibility of gene expression by environment. This pattern doesn’t hold among strongly DEK3-bound genes in other nodes (Fig. 3c, Supp. Fig. 3f), suggesting that the chromatin landscape is likely to be changing as a result of the DEK3-H2A.Z interaction, rather than as a consequence of changes in transcription.
DEK3 and H2A.Z interact genetically
Our results suggest that DEK3 plays a role in driving H2A.Z-nucleosome removal from the gene body of stress responsive genes. Supporting this model (Supp. Fig. 4i), when H2A.Z levels or its deposition is perturbed in mutants lacking H2A.Z or ACTIN RELATED PROTEIN6 (ARP6), the transcriptome shows similar mis-expression as in 35S::DEK3 (Fig. 4d-e), but opposite to that of dek3-2 (Fig. 4f-g). Furthermore, the arp6-1 mutation is able to partially suppress the effect of dek3-2 on the transcriptome (Supp. Fig. 4g). Since ARP6 might be present in several protein complexes including SWR1 and may have other roles26,27, we also investigated the effect of removing H2A.Z in plants on the DEK3 dependent transcriptome. In the triple mutant for hta8,9,11, we observe a consistent pattern over the time course, with up-regulation of 35S::DEK3 activated genes and down-regulation of 35S::DEK3 suppressed genes (Fig. 4c-e), similarly to what is observed in arp6-1 (Fig. 4f-g).
In general, “growth” related genes have less H2A.Z signal, and this is most likely to be found near the TSS (Fig. 3a). These temperature dependent genes are further induced by warm ambient temperature in dek3-2 possibly by increasing the levels of their H2A.Z gene body binding (Fig. 4d, Supp. Fig. 4b (upper panel)). Reducing H2A.Z-nucleosome occupancy in the h2a.z and arp6-1 backgrounds slightly inhibits the temperature induction of “growth” genes (Fig. 4d, 4f). However, when the levels of both proteins, H2A.Z and DEK3, on chromatin are reduced in the dek3-2 arp6-1 double mutant, the temperature dependent activation of these genes is been impaired (Supp. Fig. 4g (upper panel)). In contrast, the “stress” related genes are enriched in H2A.Z in their gene bodies (Fig. 3a) and up-regulated in the 35S::DEK3 plants grown at 27°C, (Fig. 4e, Supp. Fig. 4b (lower panel)). These genes are also up-regulated in h2a.z and arp6-1 mutants in a temperature-dependent manner (Fig. 4e, 4g), indicating that H2A.Z-nucleosome occupancy may regulate their transcription. This increase is abrogated in dek3-2 arp6-1 plants, demonstrating the antagonistic interaction between DEK3 and ARP6 (Supp. Fig. 4g (lower panel)).
In light of the physical interaction between DEK3 and H2A.Z (Fig. 1a-b, Sup. Fig. 1a) and in line with the transcriptional patterns of dek3-2 arp6-1 double mutants and arp6-1 (Supp. Fig. 4g), we sought to determine if they also interact genetically. The partial rescue of the arp6-1 phenotype by the dek3-2 mutation at both an elevated temperature and cold stress (Fig. 4a, Supp. Fig. 4c-f) indicates that ARP6 and DEK3 might have opposite functions within the same pathway. Over-expression of DEK3 in arp6-1 led to temperature-dependent lethality (Supp. Fig. 4h), suggesting that normal levels of DEK3 expression are necessary for plants to survive with reduced H2A.Z-nucleosomes when exposed to warm or cold temperature.
H2A.Z deposition patterns in human DEK targets
DEK proteins are conserved through evolution found in almost all higher eukaryotes29,30, and show similarity in domain structure between plants and animals12. Three of the four plant DEK proteins, DEK2, DEK3 and DEK4, have been found in complex with H2A.Z by mass spec analysis (Fig. 1a, Supp. Fig. 1a, Supp. Table 1) and showed similar growth phenotypes in response to warm temperature (Fig. 1d, Sup. Fig. 1d), suggesting functional redundancy between them. We therefore asked if our finding that the pattern of H2A.Z-nucleosome occupancy on the gene body may be predictive of how a gene is regulated by DEK3 in Arabidopsis also applies in humans. Previously, human DEK (hDEK) has been shown to both activate and repress expression of the genes involved in cancer and hematopoiesis30–38 (summarised in Supp. Fig. 4j). We observe that hDEK up-regulated genes possess mainly gene body H2A.Z, while hDEK down-regulated genes have more +1 nucleosome H2A.Z binding (Supp. Fig. 4j, 4k). This suggests that the interaction between DEK and H2A.Z may be functionally conserved, and that the role of DEK in controlling the balance between growth and stress resilience is conserved between plants and vertebrates. DEK in vertebrates plays key roles in carcinogenesis7,9,10,13,25,31,36–38,39, haematopoiesis25,30,33 and inflammation11,46, processes involving a fine-control of proliferation and stress resilience. It will be interesting to determine if these functions of DEK are also mediated via H2A.Z-nucleosomes.
Discussion
We find that altered levels of DEK3, an Arabidopsis ortholog of the onco-protein DEK, perturbs developmental programming in Arabidopsis thaliana grown in warm temperature by influencing expression of environmental (stress related) and growth related (developmental) genes (Sup. Fig. 2c, Sup. Fig. 4b). The effect of DEK3 on the transcriptome and development appears to be mediated at least in part by changes in the relative distribution of H2A.Z-nucleosomes on the gene bodies of target genes (Fig. 3c-d, Sup. Fig. 3f). These results may explain the dual effect of DEK3 on the transcription of its target genes, shedding light on the longstanding question regarding the influence of DEK on gene expression12,17,25. We also provide evidence for physical and genetic interaction between DEK3 and H2A.Z in Arabidopsis (Fig. 1a-b, Sup. Fig. 1a, Sup. Fig. 4c-g). Since DEK proteins are found in most multicellular eukaryotes, as is the H2A variant H2A.Z, these findings may be of broad relevance, and contribute towards understanding of hDEK as a therapeutic target.
The chromatin landscape regulates accessibility of DNA to the transcriptional machinery and transcription factors, controlling the transcriptome and how it responds to a changing environment47. Incorporation of the histone variants, H2A.Z and H3.3, into chromatin is frequently associated with responses to environmental perturbations4,7,8. While it was reported previously that DEK is a H3.3 chaperone and controls H3.3 deposition into chromatin14, the effect of altered levels of DEK3 on more than half of the warm temperature transcriptome can be explained by initial pattern of H2A.Z distribution (Fig. 3a, Sup. Fig. 3f). Previous work suggests that DEK may interact with H3.314 — which is not inconsistent with this study as it is possible for a protein to have multiple binding partners. It will be interesting to see if DEK3 is able to bind H2A.Z and H3.3 simultaneously.
Our results suggest that DEK3 influences the chromatin landscape by modulating the distribution of H2A.Z at specific genes thus controlling their responsiveness to environmental signals (Fig. 3c-d, Sup. Fig. 3f). Our data confirm and extend previous reports suggesting that the initial pattern of H2A.Z on gene bodies is important for the proper induction of transcription in response to environmental stress4,7. We show that the levels of H2A.Z on gene bodies is altered by DEK3, influencing gene expression responsiveness in response to environmental signals (Fig. 3). DEK3 overexpression for example depletes H2A.Z-nucleosome occupancy, leading to enhanced transcriptional responses to higher temperature (Fig. 3c-d, Supp. Fig. 3f, Supp. Fig. 4b).
DEK3 may influence H2A.Z distribution through its role in H3.3 incorporation into chromatin. In animal systems, lack of DEK causes enhanced H3.3 incorporation into the chromatin by HIRA and DAXX/ATRX chaperones14. H3.3 incorporation into chromatin might in turn lead to elevated DNA methylation and subsequent prevention of H2A.Z incorporation into gene bodies6. This model is not consistent with our finding that increased levels of H2A.Z occur in the gene bodies of dek3 plants, and conversely, overexpressing DEK3 results in reduced gene body H2A.Z, at least in the subset of the DEK3 target genes (Fig. 3a, 3c-d, Sup. Fig. 3f). These observations suggest another mechanism of regulation. Interestingly, H3.3 distribution was able to explain the effect of DEK3 expression on the warm temperature transcriptome for only a subset of the genes (Fig. 3a, node 6), while the H2A.Z pattern alone predicts the behaviour of more than half of the DEK3 target genes (Fig. 3a). This effect could partially be explained by possible physical interaction between DEK3 and H2A.Z (Fig. 1a-b). Even though we could not exclude the possibility of indirect physical interaction between these two proteins, our genetic studies confirm the biological relevance of this interaction (Sup. Fig. 4c-f, Sup. Fig. 4h). Many cancers are associated with impaired DEK levels12,13,40,41,43,45,46,48–50,14,25,32–35,38,39, and it will be interesting to determine if these show a similar relationship for transcriptional response and the distribution of hDEK and H2A.Z-nucleosomes.
Our work suggests a model for growth promoting and growth inhibiting responses to environmental changes (Sup. Fig. 4i), where genes with high levels of gene body H2A.Z are responsive to environmental induction, and this sensitivity is enhanced by DEK which can facilitate a decrease in H2A.Z occupancy and higher gene expression. Genes with low gene body H2A.Z occupancy are more likely to be resistant to activation by environmental signals in presence of normal DEK3 levels. For plants this is important for example in tuning the response of the transcriptome to ambient temperature. For plants it is of particular importance to keep a balance between developmental and metabolic transcriptomes under different conditions. Enhanced growth as a result of elevated temperature may cause a reduction in size of adult plant and seed yield51. Additionally, plants need to attenuate their development rate in order to provide appropriate response to biotic stress to obtain optimised fitness2. However, it may also play a role in other contexts, for example in the case of inflammation and cancer where changes in the relative levels of nutrients and oxygen occur frequently, and cells must respond appropriately.