Regulation of tomato HsfA1s

In general, the expression of Hsfs is not only restricted to stress response, but also to developmental programs. This was explained in several cases where Hsf mutants show phenotypic alterations related to specific developmental processes under stress and non-stress conditions (von Koskull-Döring et al., 2007; Kotak et al., 2007a; Scharf et al., 2012; Fragkostefanakis et al., 2016). In tomato, among four members of subclass A1, HsfA1a was proposed as the master regulator of HSR based on physiological and molecular analysis of HsfA1a co-suppression transgenic lines which are more sensitive to high temperatures than wild-type plants (Mishra et al., 2002). However, no phenotypic alterations were observed in transgenic lines under non-stress conditions. In the light of this, the regulation of the recently identified closely related HsfA1 factors is of great interest.

The basis of the master regulator assumes a constitutively expressed protein that is maintained inactive under non-stress conditions mainly by interaction with molecular chaperones like Hsp70 and Hsp90, and can be only released and activated under stress conditions to induce a set of genes encoding for molecular chaperons, Hsfs and other proteins required for protection against stress (Hahn et al., 2011). This general system of regulation has been previously shown for Hsf1 among four vertebrate Hsfs (Åkerfelt et al., 2010).

52 From transcriptomic and qRT-PCR analysis, all HsfA1 members are by large constitutively expressed, and only in few cases preferentially induced in specific developmental stages (Fig. 4 and 5). In general, HsfA1c and HsfA1e are expressed in at very low levels as shown by RNAseq data, while HsfA1a is expressed at higher levels (Fig.4). It has been previously shown that HsfA1a protein is detectable in leaves, stems, seedlings and fruits (Mishra et al., 2002), while the lack of antibodies for the other HsfA1s does not allow to judge whether a low but constitutive mRNA synthesis is accompanied by a significant protein accumulation. This would be particularly interesting for HsfA1b which shows the highest variation in transcript abundance in different tissues, and is the highest expressed HsfA1 in ripening fruits and developing seeds (Fig. 4).

The preferential induction of HsfA1b and HsfA1e genes in specific tissues or developmental stages (Fig.

5) might hint to a possible developmental function and in turn explain the lack of an obvious phenotypic alteration of A1CS plants and fruits compared to wild type when grown under non-stress conditions (Mishra et al., 2002). However, further detection of their transcript levels but in A1CS transgenic tissues, especially fruits and seeds, or the individual knock-down of such members is required to support this conclusion.

The developmental regulation of HsfA1s might also be related to the priming of cells in case of an upcoming stress. This has been exemplified for HsfA2 which is pre-synthesized in non-stressed male meiocytes to confer protection in case of a stress in more advanced stages (Fragkostefanakis et al., 2016). In this manner, HsfA1b and HsfA1e could act as priming factors for thermotolerance in fruits and seeds. In support of this, Arabidopsis HsfA1e is preferentially expressed in seeds and is therefore involved in seeds thermotolerance (Liu et al., 2011). In addition, the earlier stages of fruit ripening, and particularly at the mature green pre-climacteric stage, are more sensitive to severe heat stress conditions when compared to more ripening stages (Mishra et al. 2002). The higher tolerance of the ripe fruit might be due to the presence of pre-synthesized HsfA1b and HsfA1e (Fig. 4 and 5).

Among HsfA1 genes, HsfA1b is highly induced in response to heat stress, in a similar manner to the HS-marker gene HsfA2. Interestingly, HsfA1b is rapidly induced, having a peak in expression within 30 minutes of stress, while HsfA1e shows a more late response with a slight increase after 4 hours of stress (Fig. 6). Recently, HsfA1a was shown to acts a positive transcriptional regulator of HsfA1b while HsfB1 as a competitive repressor (Fragkostefanakis et al. 2018). In leaves, suppression of HsfB1 leads to the very strong accumulation of HsfA1b transcripts (Fragkostefanakis et al. 2018). Therefore the developmental or the stress-dependent regulation of HsfA1b might be controlled by the activities of specific Hsfs, like HsfA1a and HsfB1.

53 Apart from transcriptional regulation, an important control mechanism affecting the fate of Hsfs is protein turnover. This has been shown for tomato HsfA2 (Hu, PhD thesis 2017), HsfB1 (Röth et al., 2016a) and HsfA7 (Mesihovic, PhD thesis 2018), for which the nuclear retention of these Hsfs has been directly related to their degradation rate. The presence of the NES and NLS in the C-terminus of Hsfs facilitate the nucleocytoplasmic shuttling of Hsfs as shown for HsfA1a and confirmed here (Scharf et al., 1998). HsfA1e is the only HsfA1 with no reported NES, and in agreement with this HsfA1e was detected only in the nucleus (Fig. 11). Interestingly, HsfA1b which has both NES and NLS shows also a nuclear retention, while HsfA1a and HsfA1c show nucleocytoplasmic equilibrium (Fig. 11). Both HsfA1b and HsfA1e showed a high turnover rate with half life time of about 3 hours, while HsfA1a and HsfA1c remain stable for the 6 hours of the experiment (Fig. 12). Thereby the accumulation of transcripts of HsfA1b or HsfA1e might not be accompanied by protein accumulation; however it could serve as a priming factor in case of heat stress to allow a faster response in specific tissues and developmental stages. HsfA2 is already involved in such a priming process during pollen development in tomato while other stress-induced Hsfs including HsfA1b might be also involved in such a process (Fragkostefanakis et al., 2016).

The protein stabilization of HsfA1b and HsfA1e might require the presence of co-factors which under specific conditions and developmental stages might allow the protein accumulation of these Hsfs. For example, HsfA1a and HsfA2 interaction in stressed cells leads to the stabilization of both factors, thereby contributing to the stimulation of transcriptional activity of the complex (Scharf et al., 1998).

However, co-expression of HsfA1s with HsfA2, HsfA3, HsfA7 or HsfB1 did not lead to significant changes in the abundance of the former, suggesting that are other factors even non-Hsfs might contribute to this. As for example, binding of multiple ubiquitin molecules tags the protein for degradation by 26S proteasome (Muratani and Tansey, 2003). This has been shown for many Hsfs including tomato HsfA2 (Hu, PhD thesis 2017), HsfB1 (Hahn et al., 2011; Röth et al., 2016a) and HsfA7 (Mesihovic, PhD thesis 2018). Further investigations on the protein levels of HsfA1b and HsfA1e in the presence of proteasome inhibitor like MG132 can prove such negative regulatory mechanism. Although, all proteins are targeted by ubiquitin ligase (E3), one of three enzymes that mediate ubiquitin targeting, and consequential proteasome-mediated turnover, the intracellular localization of the Hsf is an important determinant factor for such regulatory mechanism as cytosolic Hsfs can escape it. Another determinant factor is the structure and the presence of a degradation signal, specific sequence in the target protein, that signals proteolysis by representing a site for ubiquitin ligase (E3) binding (Muratani and Tansey, 2003). Whether, HsfA1b and HsfA1e have such degradation signals still needs to be further investigated.

54 All in all, these different regulation mechanisms, developmental or stress dependent, positive or negative, among HsfA1-members might be in part responsible for such functional diversity in tomato subclass A1 Hsfs.