Discussion
A high resolution time series is required to capture fast transcriptional dynamics in the early morning
While previous researchers have noted that there is a peculiar burst of gene expression in the early morning, they were unable to fully characterise the transcriptional dynamics in this time period because the time points were not sampled frequently enough. We were able to identify multiple coordinated waves of gene expression (Figure 5A). This included a cluster containing HSFA1A (cluster 7-8) that had elevated expression at 27oC and a second cluster that had a large number of light induced genes and that decreased its expression levels at the higher ambient temperature (cluster 5-6). A large proportion of transcription factors have perturbed expression in the early morning, and most of these genes are perturbed in similar ways in both the circadian and light sensing mutants (Figure 5B). Specifically, we observed (i) elevated levels of temperature-responsive genes and (ii) a time delay in the expression in most clusters. In general, it is a useful strategy to conduct a few high resolution time course experiments, and then use this data to select informative time point experiments for future experiments.
However, it is important to note that we have characterised the gene expression changes that result from the kind of sudden onset of light that occur in growth chambers, rather than those that are found in natural conditions, such as those investigated in \citet{Annunziata2018}. Nevertheless, there are a number of benefits of this approach. Firstly, this allows us to more directly compare our results to other studies of biotic and abiotic responses in the early morning that were also performed in artificial growth chambers with sudden onset of light \cite{Grundy_2015,Dickinson2018,Ingle2015,Ezer2017a,Cortijo_2018,Michael2008,Seaton2018}. Secondly, this kind of study allows us to directly address how the transcriptome responds to a light stimulus in the morning. Specifically, since we know the exact time that plants were first able to detect light, we can measure the exact time delay between this exposure and a change in gene expression. Finally, farming in artificial lights is becoming common, and it is important to understand how crops respond to lighting conditions that are reminiscent of growth chambers \cite{Ibaraki_2016,Dutta_Gupta_2017,Olvera_Gonzalez_2013}.
Phytochromes and cryptochromes coordinate light responses in the early morning via HY5 and BBX transcription factors
Our time course data allowed us to infer a high-confidence transcriptional network that controls the dawn burst in gene expression. HY5 and BBX family proteins appear at the core of this network downstream of phytochrome and cryptochrome photoreceptors and are likely to coordinate the early and late waves of gene expression in the early morning. Notably, the photoreceptors induce both positive (HY5, HYH) and negative (BBX Set A) regulators of light responses, while they simultaneously repress other photomorphogenesis-promoting factors (BBX Set B). HY5 and HYH are not only induced by light at the transcriptional level, but the respective proteins are also stabilised through light-induced inactivation of the CONSTITUTIVELY PHOTOMORPHOGENIC 1 (COP1)/SUPPRESSOR OF PHYA-101 (SPA) repressor complex \cite{Osterlund2000,Holm2002,Saijo2003}. Boosting HY5 and HYH protein levels via both mechanisms will trigger light responses in the early morning, but this process may require fine-tuning through other factors including BBX proteins. BBX24, BBX25 and BBX32 interfere with HY5 transcriptional activity through direct interaction \cite{Holtan2011,Gangappa2013,Job2018}, while BBX30 and BBX31 act downstream of HY5 to promote elongation growth \cite{Heng2019}. BBX20 and BBX21 on the other hand promote HY5 function by increasing HY5 transcript level and post-translationally enhancing HY5 activity \cite{Job2018,Xu2018,Wei2016}, but these effects seem to be largely suppressed at dawn. Another layer of complexity is added by the fact that the COP1/SPA complex also promotes degradation of several BBX proteins \cite{Indorf2007,Yan2011,Xu2016,Gangappa2013}.
HY5 and BBX proteins interdependently tune photomorphogenic responses such as the reduction in elongation growth after dawn, but may exert additional functions independently of each other. BBX proteins have not yet been implicated in other HY5-mediated responses such as temperature signaling or biosynthesis of secondary metabolites. On the other hand, BBX30, BBX31 and BBX32 act as negative regulators of the floral transition independently of HY5 through the repression of FLOWERING LOCUS T (FT) \cite{Graeff2016,Tripathi2017}. Strong induction of these genes in the early morning may be involved in gating FT’s responsiveness to inductive signals in the photoperiodic control of flowering.
In summary, our results reveal a large transcriptional regulatory network that controls the dawn burst of gene expression. Phytochrome and cryptochrome photoreceptors coordinate a substantial portion of this network to adjust photomorphogenesis, photoperiodism and clock entrainment in accordance with the plant’s light environment