Leaf physiology
In conjunction with this study, we recently reported the impact of drought stress and heat stress in the same experiment on shoot growth, leaf physiological traits, abscisic acid (ABA) and proline levels, and expression of key genes involved in ABA and proline biosynthesis in leaves (Lehr et al., 2022). Water availability dominated the growth and leaf physiological responses, while heat stress only played a minor role, even with daily maximum leaf temperatures close to 40°C. Compared with the control, WS significantly reduced shoot elongation, leaf water potential (Ψl), stomatal conductance (gs), photosynthesis, and transpiration, while strongly increasing leaf ABA. By contrast, with the exception of Ψl, heat stress rarely altered any of these physiological traits and did not impact shoot growth, though it tended to exacerbate the effect of water stress on leaf physiology. Riesling leaves seemed to be somewhat more sensitive to heat stress, especially in combination with water stress, than were Cabernet Sauvignon leaves. For example, based on the decrease in gs, Riesling vines experienced severe water stress (gs < 0.05 mol H2O m− 2 s− 1) under both WS and HWS, but Cabernet Sauvignon only reached extreme stress under HWS (Lehr et al., 2022). In both varieties, however, midday Ψl decreased from an average of -0.8 MPa in the control to -1.2 MPa under HS to -1.3 MPa under both WS and HWS. While sufficiently low to induce stomatal closure, these values are still high enough for grapevines to avoid xylem cavitation and leaf wilting (Gambetta et al., 2020). All vines were exposed to the stress treatments for seven days; thus the mature leaves on which our measurements were conducted had developed under non-stress conditions. These results confirm recent data obtained using Malbec grapevines and show that leaves are relatively resilient with respect to heat stress, at least in the temperature range (≤ 40°C) tested here (Galat Giorgi et al., 2019). Earlier work had found the optimum temperature for photosynthesis of grape leaves to be in the range 25‒30°C (Greer and Weedon, 2012).
Berry metabolites
In sharp contrast to most leaf physiological traits and shoot growth (Lehr et al., 2022) the variation in berry metabolite composition was dominated by the effect of temperature, while water availability played, at most, a minor role. The different measures of fruit composition in the berries collected after 7 days of stress (Fig. 1) indicated that these berries were in the middle of the ripening period during which sugars accumulate rapidly and malic acid is being degraded (Hernández-Montes et al., 2021). Despite the up to 70% reduction in leaf photosynthesis under WS (Lehr et al., 2022), neither berry weight nor TSS (which is a robust proxy for sugar concentration in ripening grapes) were significantly impacted by any of the stress treatments (Fig. 1A and 1B). Similar results were found in a heat stress experiment with Cabernet Sauvignon grapes (Lecourieux et al., 2017) and Muscat Hamburg grapes (Carbonell-Bejerano et al., 2013), and in a water stress experiment with different grape varieties (Keller et al., 2015). In another study, however, in which Sémillon grapes were exposed to 40/25°C day/night temperatures for 4 days at the beginning of ripening, heat stress inhibited both berry growth and sugar accumulation (Greer and Weston, 2010). Therefore, the effect of heat stress might depend on the variety or the way treatments are applied. For example, while our study applied realistic, diurnal irradiance and temperature profiles, many other growth chamber studies use static day/night conditions. In both of our varieties, heat stress increased the berry pH (Fig. 1C). In Riesling, the WS berries had an intermediate pH (Fig. 1C). Heat stress (HS and HWS) decreased TA by 26–32%, malic acid (MalA) by 40–52%, and oxalic acid (OxA) by 20–42% in both varieties (Fig. 1D), while water stress by itself had no effect (Fig. 1E). Among organic acids, MalA is metabolized during grape ripening, while OxA continues to accumulate (Keller and Shrestha, 2014). Two other organic acids, tartaric acid (TartA) and citric acid (CitA) displayed variety-specific responses with regard to the impact of stress on their accumulation. TartA, like OxA a derivative of ascorbic acid (vitamin C) metabolism (Burbidge et al., 2021), was not significantly impacted in Cabernet Sauvignon under any stress condition but was elevated in Riesling under heat stress (Fig. 1G). CitA was not significantly impacted by any stress condition in either variety (Fig. 1H).
In practice, the impact of elevated temperature during the growing season often requires acid addition to adjust wine acidity and reduce the pH to appropriate levels, an endeavor that can incur significant expenses. Unlike MalA, which is easily metabolized under heat stress (Lakso and Kliewer, 1975; Rienth et al., 2016), TartA is much more stable at high temperatures (Sweetman et al., 2009; Hernández-Montes et al., 2021). In our study, heat stress during the early ripening phase slightly increased TartA in Riesling but not Cabernet Sauvignon berries. In contrast, Lecourieux et al. (2017, 2020) reported an increase in Cabernet Sauvignon berries exposed to elevated temperature (~ 34°C vs. 26°C at veraison). It is conceivable that there may be varietal differences in the activation of antioxidative processes in an effort to maintain homeostasis under heat stress. Despite the limited response of TartA to temperature, the lower acidity of grape berries that accompanied exposure of grapevines to HS and HWS, but not WS, indicates that temperature has a particularly pronounced effect on fruit acidity (especially MalA and OxA), which is not brought about by water deficit alone. This conclusion supports the idea that perceived effects of water stress on grape composition, or at least on organic acid metabolism, are often indirect and mediated by temperature (and light), arising from a decrease in shoot growth under water stress and the associated increase in fruit exposure to sunlight, causing solar heating of the berries (Keller et al. 2016). Though water stress decreased shoot growth in our study as well (Lehr et al., 2022), we excluded solar heating as a complicating variable by applying temperature treatments inside environmentally-controlled growth chambers. Unlike in many field experiments, moreover, leaf senescence (yellowing and abscission) was not observed with the treatment structure used in our study.
In summary, whereas water stress had a direct effect on leaf physiological processes (which may be mediated by hydraulic properties as well as hormones such as ABA), it did not affect berry organic acid metabolism in a similar way. The latter, instead, was much more strongly impacted by temperature. It is likely that the pronounced effect of temperature in grape berries compared with leaves is a consequence of the low transpiration rate and, thus, poor evaporative cooling capacity of grape berries (Zhang and Keller, 2015). During ripening, moreover, water is supplied to the berries via the phloem rather than the xylem, so the berries are relatively well buffered against soil water deficit (Keller et al., 2015; Zhang and Keller, 2017).
Transcriptome assembly and annotation
Functional annotation of the 87,867 contigs from the final grape assembly dataset was resulted in assignation of blast results to 63,032 (71.73%) of contigs and functional annotation to a total of 55,182 contigs (62.80%) (Supplementary File 1).
Differentially Expressed Genes and Enriched Gene Ontologies
A total of 5,144; 6,347; and 5,283 annotated genes were significantly differentially expressed over time in HS, WS, and HWS treatments, respectively. Of these, 2,001 HS, 2,126 WS, and 2,056 HWS genes were differentially expressed in Cabernet Sauvignon, and 3,808 HS, 4,952 WS, and 3,912 HWS genes were differentially expressed in Riesling (Fig. 2). In addition, a set of genes that displayed shared differential expression across all treatments and in both grape varieties was identified—these were classified as “core stress genes” (Fig. 2, Supplementary File 2).
In Cabernet Sauvignon, gene expression response was consistent with the berry metabolite results, with the highest number of DEGs identified in the HWS and HS treatments, while WS elicited fewer expression changes in comparison with HS and HWS. Riesling displayed more DE genes in all treatments compared to Cabernet Sauvignon, with the highest number of DE genes identified in the HWS treatment, followed closely by HS and then WS. While global gene expression changes were lower in WS than HS and HWS overall, the response of Cabernet Sauvignon berries to WS was markedly lower than that of Riesling berries. Furthermore, the greater number of DEGs identified in Riesling (Fig. 2), and the elevated number of genes impacted by WS in this variety, suggest a heightened sensitivity of Riesling berries to both temperature and water stress compared to Cabernet Sauvignon, consistent with the physiological observations in leaves.
Gene ontology enrichment analysis was conducted using 9 lists of DE genes: 1.) core HS genes; 2.) core WS genes; 3.) core HWS genes; 4.) Cabernet Sauvignon unique HS genes; 5.) Cabernet Sauvignon unique WS genes, 6.) Cabernet Sauvignon unique HWS genes; 7.) Riesling unique HS genes; 8.) Riesling unique WS genes; 9.) Riesling unique HWS genes.
Key stress ontologies corresponding to the core DE genes are as follows: “RNA binding” (HS); “catalytic activity” and “regulation of molecular function” (WS, HWS), and “response to abiotic stimulus”, “catabolic processs”, “cellular component organization”, “chloroplast”, “cytosol”, “mitochondrion”, “protein binding”, “response to chemical” (HS, WS, HWS) (Table 1). The activation of core genes and corresponding shared enriched ontologies suggests that a key group of stress responsive pathways orchestrates variety and stress-specific gene expression responses downstream.
Several additional enriched ontologies were identified in analyzing the unique DE genes that were variety or stress regime-specific. Cabernet Sauvignon displayed unique enrichment of GO terms “lipid binding” and “nuclear lumen” (HS); “nucleus” (WS, HWS); and “chromatin binding’, “membrane”, “translation”, and “translation factor activity, RNA binding”. In Riesling, unique enriched GOs included “abscission”, “circadian rhythm”, “DNA binding”, “response to biotic stimulus”, “response to external stimulus”, and “signal transduction” (WS); “cell differentiation”, “growth”, “intracellular non-membrane-bound organelle”, and “vacuole; and “cell communication”, “DNA-binding transcription factor activity”, “embryo development”, “generation of precursor metabolites and energy”, “Golgi apparatus”, “lipid metabolic process”, “peroxisome”, “plasma membrane”, and “protein modification process” (HS, WS, HWS) (Table 1).
The berries of the two cultivars seem to adopt different strategies when exposed to individual or concomitant stress. In Cabernet Sauvignon, mobilization of metabolites, suggesting an osmotic homeostatis strategy seems to be predominant. The enriched ontologies indicate activation of protein synthesis and transport along with membrane binding and transmembrane transport of the synthesized products and other substrates to mitigate the effects of stress. In Riesling, regulation of redox signaling and energy, and growth and development seem to be strategy for stress mitigation. The enriched ontologies show a global stress response involving enhanced cellular signalling, energy production, redox homeostasis, and cellular growth and differentiation.
Interestingly, a majority of the GOs identified from the lists of unique DE genes were enriched for all stress treatments in both varieties. Furthermore, many of these unique genes corresponded to the same ontologies that were identified when conducting enrichment analysis for the core stress genes. This finding suggests that, while HS, WS, and HWS differentially impact gene expression responses in Cabernet Sauvignon and Riesling fruit, perhaps the unique genes activated correspond to the same or similar pathways, or activate different aspects of stress responsive pathways.
Of particular interest to this study were genes associated with the core stress GO term “carbohydrate metabolic process”, “mitochondrion”, “generation of precursor metabolites and energy”, and a number of GOs pertaining to response to environmental stimuli and signal transduction as the genes corresponding to these ontologies are directly related to the carbohydrate and organic acid metabolites assayed in this study. Moreover, the signaling associated genes are expected to lend insight into variety and stress regime-specific initial response to heat and drought.
Organic acid metabolism
Organic acid content is a critical component of fruit and wine organoleptic quality. Most acids accumulate until grape berries undergo a metabolic shift at the onset of ripening. However, acid content, and therefore TA and pH, are highly influenced by genotype and environmental conditions (Sweetman et al., 2014). High-temperature-driven reduction of organic acids, particularly malate, has been studied primarily in red grapevine varieties, including Shiraz (Sweetman et al., 2014), Cabernet Sauvignon (Ruffner et al., 1976; Lecourieux et al., 2020), Muscat Hamburg (Carbonell-Bejerano et al., 2013), and DCRF mutants (microvine) (Rienth et al., 2016). Recently, effects of combined temperature and drought stress on two white varieties, Chardonnay and Xynisteri, were also studied (Tzortzakis et al., 2020). Results of these studies and the present work suggest that WS and HS may impact grape berries of different varieties in different ways, with the metabolic and physiological responses much more pronounced under heat stress.
Several of the assayed organic acids are directly (MalA, CitA) or indirectly (OxA) associated with the TCA cycle and the glyoxylate cycle. Both processes take place primarily in the mitochondria, a cellular component that was enriched in the GO analysis. The TCA cycle is the pathway responsible for the breakdown of pyruvate produced during glycolysis, generation of donor molecules for mitochondrial electron transport and substrates for other metabolites, and production of CO2, while the glyoxylate cycle is involved in the breakdown of fatty acids to produce substrates for gluconeogenesis. Significantly elevated expression of a DEG encoding the mitochondrial isoform of malate dehydrogenase (mMDH), which catalyzes conversion of MalA to oxaloacetic acid (OAA), was observed in HS and HWS grapes, corresponding to reduced levels of MalA in the berries (Fig. 3). Glyoxysomal malate dehydrogenase (gMDH) also displayed high transcript abundance during stress. In contrast to MDH isoform expression, most other TCA and glyoxylate cycle enzyme-encoding genes displayed reduced expression during both HS and WS, and especially under HWS (Fig. 3). The enzyme in the TCA cycle downstream of mMDH, citrate synthase (mCS), displayed reduced transcript abundance during WS and HWS in Cabernet Sauvignon, and during HS and HWS in Riesling, with significant changes in mCS expression only observed in Riesling. In the glyoxysome, expression of gCS displayed a similar pattern in Cabernet Sauvignon and Riesling during HWS; after stress, however, expression increased significantly in Riesling WS and HWS (Fig. 3). The expression changes in CS isoforms during stress corresponded to reduced CitA levels in Riesling, while no significant change in CitA accumulation was observed in Cabernet Sauvignon.
The third metabolite assayed, OxA, can be produced via a number of pathways, including decarboxylation of OAA and metabolism of ascorbate and/or glyoxylate (Fig. 3) (Cai et al., 2018). Assuming a hypothetical, unidirectional production of TCA and glyoxylate cycle metabolites, the increased MDH and decreased CS transcript abundance would be expected to indicate a greater flux of OAA into the production of OxA. However, this is not what we observed in this study, as OxA levels decreased significantly in both Cabernet Sauvignon and Riesling under heat stress (Fig. 3), suggesting an alternative fate of OAA, such as a gluconeogenic precursor or biomineralization to calcium-oxalate (DeBolt et al., 2004; Franceschi and Nakata, 2005).
Venios et al. (2020) described a comprehensive physiological response of grape berries to heat stress, characterized by reduced TA, reduced MalA, increased sugar to acid ratio, reduced flavanol and anthocyanin production, and increased sugar content. The described compositional changes are consistent with those measured in the present study, particularly with regards to TA and MalA content. However, while previous studies have sought to elucidate effects of heat on different stages of berry development and in different temperature regimes (Pillet et al., 2012; Rienth et al., 2014; Sweetman et al., 2014), the present study explored the different effects that the interaction of heat and water stress have in both red and white grape varieties during the early ripening phase.
Carbohydrate metabolic processes – glycolysis and gluconeogenesis
It has been suggested that transcriptome remodeling in response to high temperature disrupts synchrony of sugar and organic acid metabolism during grape berry development (Rienth et al., 2016). Consistent with this idea, our study found heat stress effects on organic acids but not sugar in ripening grape berries (Fig. 1). Thus, in conjunction with assessment of organic acid metabolism at the transcriptome level, it was also of interest to observe expression patterns of genes involved in gluconeogenesis as well as glycolysis. Gluconeogenesis has a temperature optimum near 20°C, decreasing to half the maximum rate at 30°C (Ruffner et al., 1975), thus the effects of elevated temperature are expected to elicit a particularly notable effect on gluconeogenic processes.
Hexokinase, the first committed enzyme in glycolysis, displayed significant reduction in transcript abundance in Riesling during HS and HWS in comparison with the control; however, in Cabernet Sauvignon no significant change in expression of this gene was observed under any of the stress conditions (Fig. 3). A similar trend was observed for the second step in the glycolytic pathway, phosphohexose isomerase, where significant decreases in gene expression were measured during WS, HS, and HWS in berries of both grape varieties, particularly at the peak of stress treatment.
Mitochondrial electron transport
In general, reduced expression of genes associated with mitochondrial ETC complexes I, II, III, and IV was observed during stress in both varieties. In most cases, HS and HWS had the greatest impact on reduction of transcriptional activity. Transcription associated with alternative respiration (AOX homolog, ubiquinol oxidase 2) was elevated in Riesling but repressed in Cabernet Sauvignon—the AOX gene is activated in response to stress and plays a role in reduction of ROS (Hewitt and Dhingra, 2019). The heightened expression in Riesling may be indicative of an elevated ROS scavenging response in this variety. As Riesling berries, unlike those of Cabernet Sauvignon, do not accumulate anthocyanins, they might utilize an alternative strategy by which to mitigate oxidative stress, although both varieties displayed elevated expression of ROS scavenging genes in response to HS and HWS (Fig. 3).
Heat and water stress signal transduction pathways
In the general model for heat stress response in grape berries (Venios et al., 2020), stress sensor proteins respond to elevated temperatures and transmit heat stress signals via ROS and secondary messengers that activate signal transducers, such as mitogen-activated protein kinases (MAPKs), that further relay the stress signal. This activates a transcriptional network comprised of stress-related proteins and chaperones (e.g., heat shock proteins [HSPs], heat stress transcription factors [HSFs]; ascorbate peroxidase [APX], and dehydroascorbate reductase [DHAR]) that ultimately confers tolerance to heat stress.
In both Cabernet Sauvignon and Riesling, several MAPK genes, numerous HSP genes, and several HSF, APX, and DHAR genes displayed significant responses to HS and HWS treatments (Fig. 4, top). Unique to Riesling, many HSPs were significantly upregulated under WS as well, indicating these transcription factors may also play a role in the WS response, in addition to the HS response, of this variety. Several significant MAPKs were identified in both varieties, with more significant DEGs observed in Riesling, as well as a greater balance of upregulated to downregulated genes. Though Cabernet Sauvignon displayed fewer significant MAPK genes, nearly all were upregulated. HSPs, not surprisingly, represented a substantial portion of significant DEGs pertaining to the general heat stress pathway (Fig. 4, top). Both Cabernet Sauvignon and Riesling had high numbers (i.e., 52 and 48) of upregulated DEGs in the HS and HWS treatment. As with MAPKs Riesling displayed a high number of upregulated HSPs under WS, suggesting that HSPs play a role in the WS response in this variety (Fig. 4, top). In addition to the MAPK and HSP primary signal transducers, differentially expressed HSFs were identified in both varieties, with stronger representation in Riesling. Multiple differentially expressed APX and DHAR genes were identified in both varieties, with a higher representation of upregulated DEGs in Cabernet Sauvignon HS and HWS treatments, and a greater balance of upregulated and downregulated DEGs observed in Riesling. The greater balance of upregulated to downregulated genes in Riesling is suggestive of a more fine-tuned mechanism of stress response regulation, which may be necessary in the absence of pigmented antioxidants, like anthocyanins, that accumulate in red varieties. This is in contrast to Cabernet Sauvignon, which displayed fewer differentially expressed genes overall, but for which a greater percentage of the DEGs were upregulated under HS and HWS treatments (Fig. 4, top).
In the general model for response to water deficit (Boudsocq and Laurière, 2005), drought response may be activated in both ABA-dependent and independent manners. In the case of the former, water deficit triggers accumulation of ABA and ABA receptor activity (e.g., increased expression of NCED, PYR, and PYL), which leads to activation of MYB/MYC transcription factors, which together coordinate the hormone-mediated drought response. In the case of the latter, drought conditions activate dehydration-responsive binding elements and cold-binding factors (DREBs and CBFs), which coordinate the water stress response by MYC, MYB, and other transcription factors in a stress hormone-independent manner.
This study revealed that Cabernet Sauvignon berries displayed significant temporal upregulation of ABA receptor-encoding NCED and PYL/PYR genes in HS and HWS, but not WS (Fig. 4, middle), which contrasts with the response observed in leaves (Lehr et al., 2022). No significantly upregulated DEGs encoding ABA receptors were observed for Riesling, and conversely, all of the differentially expressed genes identified for this class were downregulated (primarily in the WS treatment). MYB/MYC genes were represented among the DEGs for both varieties; however, no significantly upregulated MYB/MYC DEGs were observed in Cabernet Sauvignon, and a larger number of these transcription factors, as well as a greater balance of upregulated and downregulated genes was observed in Riesling (Fig. 4, middle). Significant DREB genes were not represented in Cabernet Sauvignon at all, while Riesling displayed differential expression of DREB genes for all stress treatments (Fig. 4, middle). Taken together, these findings suggest that, unlike their leaves, the berries of both grape varieties likely respond to water stress in an ABA-independent manner, although ABA-dependent mechanisms may be at play in the heat stress response of Cabernet Sauvignon. Moreover, the heightened number of significant DEGs for the WS treatment, as well as the greater balance of upregulated and downregulated genes, in Riesling indicates that the berries of this variety may be more sensitive to water stress than those of Cabernet Sauvignon, and therefore may require more fine-tuned regulation of its responses (Fig. 4, bottom).