Cold priming results in decreased cold activation of specific ZAT genes
ZAT10 showed in the previous study strongest primability of the selected cold-responsive genes [17]. To identify similarly regulated genes in Arabidopsis thaliana, publicly available data resources on transcript abundance regulation were scanned with GENEMANIA for ZAT10-like regulated genes [35]. The 15 highlighted genes (Fig. 1A) included BAP1, which is, like ZAT10, cold-priming sensitive and less inducible by cold 5 days after 24 h cold priming at 4 °C, as shown before [17]. Additionally, GENEMANIA revealed the genes for the zinc-finger transcription factors ZAT6 (Zinc finger protein 6; At5g04340), ZAT11 (At2g37430), ZAT12 (At5g59820), ZAT5 (At2g28200), ZAT18 (At3g53600), the WRKY transcription factors WRKY33 (At2g38470) and WRKY40 (At1g80840), the AP2-type transcription factors ERF6 (Ethylene response factor 6; At4g17490), ERF13 (At2g44840) and ERF104 (At5g61600), the mitochondrial uncoupling protein PUMP4 (At4g24570) and the Ca2+-binding protein encoding gene At4g272800. A similar analysis on the STRING v.11 platform [36] named also ACS6 (1-aminocyclopropane-1-carboxylate synthase 6; At4g11280), that is involved in ethylene biosynthesis, as a ZAT10 co-expressed gene (Fig. 1). All these genes respond, like ZAT10, to a wide range of abiotic stress stimuli and to oxidative stress [20, 37-43].
STRING v.11 further indicates protein-protein interactions (Fig. 1; orange lines). Via feed-back effects, they could impact on transcript abundance regulation. The ZAT10 transcription factor interacts with the MAP kinases MPK3 (At3g45640) and MPK6 (At2g43790), which are elements of a core plant stress signal transduction pathway responding to biotic and abiotic signals [44, 45]. MPK6 and MPK3 also phosphorylate ZAT6 [46], ERF6, ERF104 [41, 43], WRKY33 [47], WRKY40 [41] and ACS6 [48]. Additionally, ZAT10 interacts with the transcriptional co-repressors TOPLESS (TPL, At1g15750) and TOPLESS-RELATED-4 (At3g15880) [44, 49-51]. TPL binds also ZAT6 [50]. To test ZAT10-like regulated genes for the cold-primability of their cold regulation, we selected genes with different affinity to MPK6 / MPK3 and / or TPL, namely ZAT10, ZAT6, ACS6 and WRKY40 for a qPCR (quantitative polymerase chain reaction)-based priming analysis. We further included the gene for the bi-functional enolase LOS2 (At2g36530), which is a negative upstream transcriptional regulator of ZAT10 [52]. The transcript levels of these genes were analysed by qPCR immediately after triggering in previously naive plants (T) and in plants that were cold-primed 5 days before cold triggering (PT). As controls, untreated plants (C) and plants (P) that perceived 5 days earlier the priming cold-treatment, but were not cold-triggered, were analysed.
Like ZAT10, the transcript levels of ZAT12 and ZAT6 were significantly decreased in PT-plants as compared to T-plants, demonstrating priming-sensitivity (Fig 1B). ACS6 and WRKY40 were not sensitive to the triggering stimulus, independent of whether the plants were cold-primed or not. Regulation of LOS2, which binds the ZAT10 promoter and controls ZAT10-mediated cold-induction of the cold and drought marker gene RD29 [52], was strongly cold-inducible (comparison of transcript levels in C- and T-plants), but not priming-regulated (comparison of transcript levels in T- and PT-plants). The analysis gave no indication that interaction with known ZAT10-interacting proteins controls priming, but demonstrated that cold-priming affects specific genes, even in a group of genes which are otherwise widely co-regulated with ZAT10 [17-19] (Fig. 1).
The effect of cold priming on the regulation of the ZAT genes upon high light triggering
For comparison of the cold-priming effect on cold and high light triggering, we established a heat filtered high light set-up (800 µmol quanta m-2 s-1) (Fig. 2A), which increases H2O2 levels and damages photosystem II (as indicated by the maximum quantum yield of photosystem II) to a similar extent as the 4 °C treatment used for cold priming and cold triggering does (Fig. 2B and C).
The set-up was further evaluated by qPCR for its impact on regulation of well characterized light and heat regulated genes. After 2 h in high light, the transcript levels of the light-inducible genes ELIP2 (early light induced protein 2, At4g14690; [53]), GPX7 (glutathione peroxidase 7, At4g31870; [54]) and PAL1 (phenylalanine ammonium lyase 1, At2g37040) were increased (Fig. 2D). The heat filter was sufficient to counteract significant activation of the heat sensitive genes HSFA7a (At3g51910) and HSFA7b (At3g63350) [55, 56] (Fig. 2E).
Besides induction of ZAT10, the light treatment increased the ZAT6 transcript levels almost as strong as the 24 hours cold treatment (T-plants in Fig. 1B and 2F). ZAT12 showed only a very weak (but also significant) response to the light treatment (Fig. 2F). In cold-primed plants, the mean transcript levels of ZAT6 were lower in PT-plants than in T-plants, indicating primability, although the effect was not significant due to strong variation of the gene induction level. On the contrary, the transcript levels of ZAT10 and ZAT12 were more similarly regulated by light triggering in primed and non-primed plants (Fig 2F). Consequently, cold priming did not have any or had only very little effect on the light triggering response of these genes.
Photosynthetic performance after triggering
The differences between the cold and the light triggering response of the ZAT genes in cold-primed plants (Fig. 1B and 2F), especially ZAT10, could result from effects of priming on the photosynthetic electron transport efficiency. To test this hypothesis, we compared the photosynthetic performance of photosystem-II in cold-primed plants after cold and light triggering by chlorophyll-a fluorescence analysis. Triggered (T) and primed + triggered (PT) plants were analysed side-by-side by 2-dimensional chlorophyll-a fluorescence imaging in middle-aged leaves, which show strongest priming sensitivity in 4-week-old plants [30] (Fig. 3).
After cold and light triggering, the maximal quantum yield of photosystem-II (FV/FM; 0 min in Fig. 3 top) was similar in dark-acclimated T- and PT-plants, demonstrating that the triggering responses were unaffected by cold priming. Upon illumination with a photosynthetic photon flux density (PPFD) of 185 µmol quanta m-2 s-1, the quantum yields of photosystem II (ΦPS-II) and photochemical and non-photochemical quenching (qP and NPQ) also did not differ between primed and non-primed plants, both, after cold- and after light-triggering (T- and PT). It demonstrated that the priming treatment did not reduce the response to the triggering stress, although cold and light by itself differently impacted on ΦPS-II and the quenching parameters (Fig. 3). The similarity of the responses between the respective T- and PT-plants did not support the hypothesis that the priming-dependent differences in gene expression regulation result from differences in stress-induced damage or regulation of photosystem-II activity as caused by priming.
Effect of cold priming on cold- and high light-regulated gene expression
For more insight into the effect of cold priming on the stress responses, we maximally widened the target gene spectrum and performed a genome wide RNA-sequencing (RNASeq analysis) experiment 2 h after cold (4 °C) and light triggering (800 µmol quanta m-2 s-1) of 5 days earlier cold-primed and non-primed plants. RNA sequencing resulted in 23.76 – 24.14 million reads per sample (Suppl. Tab. 1). At minimum, 98.49 % of the reads could be mapped to the TAIR10 genome (Suppl. Tab. 1). Sequences were recorded for 24085 different genes. The transcript levels of many well-known, highly cold and light-responsive transcription factors, e.g. CBF1 (At4g25490) and CBF3 (At4g25480) [57], ANAC078 (At5g04410) [58] and ZAT10 [17, 21] and ZAT6 [59], were 2 h after cold or light triggering already strongly decreased (Suppl. Tab. 2). At the same time, the transcript levels of secondarily cold regulated genes, such as the CBF3-regulated gene COR15A (At2g42540) and the ANAC078 target genes At1g56650, At3g01600 and At5g58610 [60] still were induced (Suppl. Tab. 2). Genes that are well characterized for their heat inducibility, such as HSFA2 (At2g26150), HSFA7a (At3g51910), and HSA32 (At4g21320), were only very weakly expressed in all samples (Suppl. Tab. 2). The transcript level of the senescence regulating NAC transcription factor ORE1 (ANAC092; At5g39610) [61] was not increased in any sample (Suppl. Tab. 2). The expression pattern confirmed high responsiveness of stressor-specific target genes and showed that the treatments did not induce heat signalling or activate senescence.
61.7 % of the genes that were at least 2-fold up-regulated and 32.8 % of the genes at least 2-fold down-regulated in response to light in unprimed plants, were also at least 2-fold regulated by the cold treatment. On the contrary, only 0.3 and 5.5 % of the at least 2-fold regulated genes were inversely regulated by cold and light. Thus, our cold and light treatments widely regulated genes in the same direction in unprimed plants, similar as shown before by others [31].
Volcano plots (depicting the intensity of priming-dependent regulation based on the false discovery rate (FDR)) (Fig. 4 top) and blotting of the gene expression levels of primed plants (y-axes) against the gene expression levels of the respective unprimed plants (x-axes) (Fig. 4 bottom), showed that cold priming affected cold and light regulation of only specific genes. Cold triggering resulted in much less gene expression variability than light triggering in cold-primed plants (Fig. 4 bottom). In general, most significant priming-dependent regulation was observed for medium strongly expressed genes (Fig. 4 bottom).
Principal component analysis (PCA) (Fig. 5A) and clustering (Fig. 5B) of the relative transcript level in T- and PT-plants indicated that the priming effects on non-triggered, and cold- or light-triggered plants differed in direction and intensity. Already this first comparison let assume that the priming effects observed after triggering did not result from prolonged gene dysregulation in response to the priming stimulus, but that priming affected the response to the triggering stimulus in a stressor-specific manner.
Long-term, not triggering-dependent gene expression effects of cold priming
For more stringent gene regulation analysis, the 13775 genes were selected that were detected in all samples and were recorded with FPKM (fragments per kilobase of exon per million reads mapped) values of 5 or higher in at least one data set. The effects of priming on the transcript levels were calculated by dividing the FPKM-values of primed and non-primed plants at the end of the lag-phase (P / C) and in cold-triggered (PT-C / T-C) and light-triggered plants (PT-L / T-L).
Transcriptome comparison between C and P plants at the end of the 5-day-long lag-phase demonstrated that the transcriptome had widely reverted prior to application of the triggering stimuli. Only for 12 genes more than 2-fold higher and only for 4 genes more than 2-fold lower transcript levels were recorded in primed plants as compared to control plants (Fig. 5C top, Suppl. Tab. 3). At1g53870 (encoding a LURP (Late/sustained Up-regulation in Response to Hyaloperonospora parasitica)-like protein, At1g73260 (putative trypsin inhibitor), At4g12490 and At4g12480 (two bifunctional inhibitor proteins, AZI3 and EARLI1), a cation exchanger (At3g51860) and a haloacid dehalogenase-like hydrolase (HAD) superfamily protein (At5g36790) were strongest up-regulated. These genes were only weakly expressed under control conditions. Consequently, the absolute regulation of the transcript levels was low. On the contrary, the transcript levels of a transmembrane protein (At4g12495), the senescence and stress inducible gene SAG13 (At2g29350, encoding a short-chain alcohol dehydrogenase) and extensin-4 (At1g76930) were recorded with FPKM values higher than 10. Their transcript levels were more than 2-fold increased 5 days after cold priming reflecting a strong absolute effect (Supp. Tab. 3).
The four genes which were down-regulated in P compared to C encode lipid-transfer protein-4 (At5g59310), a glycine-rich protein (At1g04800), another LURP1-like protein (At1g53890) and an embryo development controlling gene (At4g29660) (Suppl. Tab. 3).
Analysing the transcript abundance patterns at lower threshold (FPKM ≥ 5 in at least one of the treatments and log2 (primed / unprimed) ≥ I 0.5 I) (Fig. 5C bottom) showed only for two of the 365 potentially long-term regulated genes, namely a hypothetical gene (At5g23411) and At1g53870 (encoding a LURP1-related protein), co-upregulation in not triggered and cold- or light-triggered plants. Only one hypothetical gene (At1g13470) was co-downregulated in cold-primed plants in all three treatment groups (Suppl. Tab. 4). The very low number of co-regulated genes demonstrates that the priming memory affects gene regulation in a stressor-specific manner.
Common triggering-dependent effects of cold priming on cold and light triggering
Since cold and excess light regulate the majority of genes in the same direction [31], regulation of common signal transduction elements would result in high similarity between the effect of cold and light triggering on priming sensitive genes. Already the analysis of a small selection of ZAT10-related genes showed differences (Fig. 1B and 2E). On the transcriptome level, RNASeq analysis identified under the more stringent conditions used for analysis (FPKM ≥ 10 and log2(PT/T) ≥ I 1 I) only a gene for a not further characterized transmembrane protein (At4g22510) as potentially (at least 2-fold) priming co-regulated in cold- and light-triggered plants (Fig. 5C top).
Lowering the threshold to FPKM ≥ 5 and log2(PT/T) ≥ I 0.5 I showed 29 genes as being co-regulated in a priming-dependent manner after light and cold triggering (Suppl. Tab. 4). Eight of the 17 co-up-regulated transcripts map to the same chromosome region and several of the short genes overlap in sense and antisense orientation. Consequently, the FPKM values (as calculated for these genes) may overstate the actual transcript abundances and the regulation amplitudes of individual genes. The remaining co-up-regulated genes encode (besides hypothetical proteins and proteins of unknown function) with ERD6-like 1 (early response to dehydration-6 like-1; At1g08920), a CC-NBS-LRR class immune receptor (At1g59218), the extensin OLE1 (At2g16630), a kinase inhibitor-like protein (At2g28870), plastome-encoded photoreceptor protein M (Atcg00220) and the plastid ribosomal subunit L32 (Atcg01020) a diverse spectrum of proteins.
In the group of the 12 genes, which are less expressed after light and cold triggering in primed plants (Suppl. Tab. 4), three encode disease associated genes, namely two β-glucanases (PR2 (BGL2; At3g57260) and BLG3 (At3g57240)) and one chitinase (At2g43570).
Specific effects of cold priming on cold and light triggering
Most priming-responsive genes were regulated by either cold or by light triggering (Fig. 5C). Under highly selective conditions (FPKM ≥ 10 and log2(PT/T) ≥ I 1 I), the transcript levels of only two genes, expansin-A8 (At2g40610) and glycine-rich protein 9 (At2g05440), were lower after cold triggering due to cold priming. In parallel, 13 genes were more strongly expressed after cold triggering in cold-primed plants than in non-primed ones. Three of them, Kunitz trypsin inhibitor 1 (At1g73260), NIT2 (At3g44300) and SAG13 (At2g29350), were already induced prior to application of the triggering stimulus. Nine of the remaining 10 genes encode (hypothetical) lipid transfer proteins or are not characterized for their function (Suppl. Tab. 3). The remaining, trigger-specifically regulated gene was OLE1 (At2g16630) that encodes an extensin.
On the contrary, light triggering resulted in cold-primed plants in specific accumulation of the transcripts for 9 genes, of which three encode heat shock proteins. Various defence-related genes, such as PR2 (pathogen responsive gene 2, At3g57260), PR4 (At3g04720), a pathogen and circadian controlled gene PCC1 (At3g22231), a chitinase (At3g12500) and five defensins, were less strongly induced by high light in primed plants than in naïve ones (Suppl. Tab. 3). Two genes, namely, At2g73260 and At4g12495, encoding a trypsin inhibitor and a transmembrane protein, showed inverse regulation in primed plants before and after light triggering. Inversion of the priming effect by the triggering response demonstrates that priming actively affected gene regulation by the triggering light stress event.
The quantitative differences between the priming impact on cold and light triggering were confirmed when the genes were filtered based on weaker criteria (FPKM ≥ 5 and log2(PT/T) ≥ I 0.5 I) (Fig. 5C): 130 genes were specifically induced and 121 down-regulated in cold-primed plants after cold triggering. Light triggering of cold-primed plants resulted in stronger induction of 613 and down-regulation of 334 genes in comparison to light-triggered non-primed plants.
Analysis of regulation patterns by qPCR
Regulation observed in the RNASeq experiment with pooled plant material from ten plants per treatment was evaluated by qPCR in at least 3 independently cultivated and treated biological replicates for 5 genes showing priming effects at the end of the lag-phase, for 5 genes which were regulated in a priming-dependent manner after cold triggering, and for 5 priming sensitive genes regulated by light (Fig. 6A). The priority was given to genes with high FPKM values. In the qPCR analysis, the transcript levels were normalized to the expression intensity of the constitutively expressed gene YLS8 (At5g08290) [62]. In all three gene sets, three genes were selected which are up-regulated in primed plants as compared to non-primed plants and two which were down-regulated. 13, out of the selected 15 genes, showed in the qPCR analysis significant regulation (Student t-Test, p<0.05) consistent with the RNASeq data. The transcript levels of the other two genes, namely At5g59720 (encoding the heat-shock protein HSP18.2) and At1g73260 (encoding a Kunitz factor protein) were by average (although not significantly) more than 2-fold regulated in the same direction as in the RNASeq experiment.
Of the five genes tested by qPCR for higher transcript levels 5 days after cold priming (Fig. 6A top), RNASeq analysis indicated only for SAG13 also higher transcript levels after cold triggering. qPCR in independently cultivated and treated biological replicates confirmed this effect (Fig. 6B). Additionally, it also showed down-regulation in primed plants after light triggering consistent with the RNASeq analysis (Suppl. Tab. 3; Fig. 6B). qPCR further confirmed the regulation observed by RNASeq for extensin-4 (At1g76930) and PR2 (At3g57260) before and after triggering (Fig. 6B). The ratios calculated from the FPKM values of primed and the respective unprimed plants (P/C; PT-C/T-C and PT-L/T-L) were for all treatments in the range of the values obtained by qPCR for the various biological replicates (Fig. 6B).
Functional categorization of the cold priming effect on the triggering response
Functional categorization of the priming-regulated genes based on analysing the enrichment of gene ontologies (GO) [63, 64] was performed with the wider data set (log2 (PT/T) > l 0.5 l; FPKM ≥ 5) on the AgriGO v2 platform (http://systemsbiology.cau.edu.cn/agriGOv2/). Data processing was evaluated using the Fischer test (F-test) and the Yekutieli method for α-level adjustment at a p-level of 0.05 [65]. The minimum threshold for statistical testing and multi-test adjustment was set to 5 genes per GO-term [66]. From the primary data, the subset of the most specific GOs within the hierarchical GO structure were extracted for the figures (Fig. 7 and 8). The full lists including information on the p-value and FDR (False Discovery Rate) and graphical images depicting all GO-terms in hierarchical order are provided in the supplements (Suppl. Tab. 5).
In the group of transcripts that were up-regulated in cold-primed plants after cold triggering, stress regulated genes were significantly enriched in comparison to non-primed cold stressed plants (Fig. 7). Especially genes responding to wounding, immune and programmed cell death regulation and / or genes under control of jasmonic acid signalling were over-represented. Additionally, priming preferentially affected the cold triggering response of genes involved in the starvation regulation and in flavonoid and anthocyanin biosynthesis (At4g22880, At4g09820, At2g02990, At3g29590, At5g17220, At5g42800, At4g14090, At5g54060). All eight genes of the latter group were also induced by excess light but were less induced or even inversely regulated in primed plants after light triggering compared to primed plants after cold triggering (Suppl. Tab. 5). CHS and PAL1, which regulate early steps of phenylpropanoid metabolism and were previously shown to be more strongly activated in cold-primed plants upon cold triggering [17], were also more strongly induced in cold-primed plants in response to cold triggering in the new dataset, although they did not pass the threshold criteria used here for the bioinformatics analysis. In parallel, cold triggering resulted in cold-primed plants in weaker expression of genes involved in transport organization, growth and morphogenesis (Fig. 7). Various of the less expressed genes respond to auxin-activated signalling and response pathways.
After light triggering, genes involved in organelle organization, morphogenesis and nucleic acid metabolism were more strongly induced in cold-primed plants than in non-primed ones. Genes responding to biotic stimuli, acids and oxygen-containing organic compounds (At5g44420, At3g15356, At3g22231, At2g14560 At1g73260, At4g10500, At3g16530) and genes involved in metabolic regulatory processes are less represented in primed plants (Fig. 8). In general, GO analysis showed that cold priming results in an inverse support of growth and biotic stress response upon cold and light triggering (Fig. 7 and 8, orange and dark green bars).
Sub-analysis of the priming-responsive genes inversely regulated by cold and high light
In the group of 159 genes with higher transcript levels in PT-C plants than in T-C plants and the 379 genes down-regulated in PT-L plants as compared to T-L plants (FPKM values > 5 and log2 (PT/T) > l0.5l ) 17 genes were inversely regulated (Supp. Tab. 6). Additionally, 12 genes were inversely regulated between the group of 145 genes down-regulated PT-C plants (as compared to T-C) and 633 genes up-regulated in PT-L (as compared to T-L) (Supp. Tab. 6).
Six of these (in total) 29 inversely regulated genes were not annotated in TAIR10, which is the data background used for functional categorization with AgriGO v2. Only one biological function was significantly overrepresented in the remaining group of 23 genes (Suppl. Tab. 7). Seven of the 23 genes, namely At2g29350, At4g37990, At1g73260, At2g43510, At3g22231, At3g04720 and At3g12500, respond to biotic stimuli. They all showed higher transcript levels after cold triggering and lower ones after light triggering if the plants were cold-primed before (Suppl. Tab. 6). Taking even slight regulation prior to triggering into account, all these genes show specific responses to light triggering (Suppl. Tab. 6). Three of them (At3g22231, At3g04720 and At3g12500) showed also up-regulation of the transcript levels after cold triggering and down-regulation in response to light. These three two-directionally regulated genes encode the plasma membrane protein Pathogen and Circadian Controlled 1 (PCC1; At3g22231), Pathogenesis Related 4 (PR4; At3g04720) and a basic chitinase (CHI-B; At3g12500). All three genes are associated with pathogen defence. Also CHS (At5g13930), but not the other core response gene PAL1 (At2g37040), showed stronger expression in cold-primed plants upon cold triggering and lower transcript levels after excess light triggering, although with lower amplitudes than PCC1, PR4 and CHI-B (Suppl. Tab. 2).
Expression network analysis on the GENEMANIA platform indicated only very faint co-expression between PR4 and CHI-B and no co-regulation of the two genes with PCC1. The impression that these genes are hardly co-regulated in naïve plants was confirmed by comparison of transcript abundance regulation using the compare-mode of the eFP browser [67] on publicly available transcript abundance regulation data for developmental regulation in Arabidopsis thaliana and the response to biotic and abiotic stress. qPCR analysis confirmed the inverse regulation of pathogen related genes PCC1, PR4 and Kunitz 1 after cold and light triggering of cold-primed plants (Fig. 9). For CHI-B, the transcript levels were below the detection level of qPCR.
The other 16 genes, which responded inversely to cold and light triggering in a priming-dependent manner, have diverse functions. Five encode transmembrane proteins (At4g12495, At1g79170, At1g16916, At5g65580 and At1g53035), two protease inhibitors (At1g73260 and At2g43510) and two protein phosphatases 2C (At5g02760 and At3g16800) (Suppl. Tab. 6).