Molecular and genetic bases of heat stress responses in crop plants and breeding for increased resilience and productivity

This review provides a wide-ranging assessment of approaches to reduce the impact of heat stress on food crops.

Abstract

To ensure the food security of future generations and to address the challenge of the ‘no hunger zone’ proposed by the FAO (Food and Agriculture Organization), crop production must be doubled by 2050, but environmental stresses are counteracting this goal. Heat stress in particular is affecting agricultural crops more frequently and more severely. Since the discovery of the physiological, molecular, and genetic bases of heat stress responses, cultivated plants have become the subject of intense research on how they may avoid or tolerate heat stress by either using natural genetic variation or creating new variation with DNA technologies, mutational breeding, or genome editing. This review reports current understanding of the genetic and molecular bases of heat stress in crops together with recent approaches to creating heat-tolerant varieties. Research is close to a breakthrough of global relevance, breeding plants fitter to face the biggest challenge of our time.

Introduction

Current analysis conducted by scientific communities including NASA’s Goddard Institute for Space Studies (GISS) indicates that the average global temperature on Earth has increased by ~0.8 °C since 1880. Two-thirds of the warming has occurred since 1975, at a rate of roughly 0.15–0.20 °C per decade (Lorenz et al., 2019). Moreover, the Intergovernmental Panel on Climate Change (IPCC) in its last report assesses that even limiting global warming to just 1.5 °C would require an unprecedented change in many aspects of society (https://www.ipcc.ch/sr15/).

For many years, scientists have studied the physiological mechanisms underlying heat stress response (HSR) and tolerance in plants (Fahad et al., 2017; Prasad et al., 2017; Govindaraj et al., 2018). However, after the discovery of the molecular responses and regulation of heat shock proteins (HSPs) and the activation of other genes, understanding of the HSR process became more mechanistic (Key et al., 1981; Altschuler and Mascarenhas, 1982). Major advances have been made in the molecular mechanisms of HSR (Ohama et al., 2017) and identification of genes or quantitative trait loci (QTLs) associated with heat tolerance in plants.

The production of plants with new alleles in HS-responsive genes and the possibility of testing these plants on innovative phenotyping platforms offer the opportunity fully to establish the role of these genes in the HSR and to develop crop plants with an increased tolerance to heat (Comastri et al., 2018).

Temperature stress, crop yield, and food security

Temperature is one of the major environmental factors which affect plant growth, development, and yield. Temperatures persistently above optimal for plant growth may induce HS and reduce yields. At some threshold, they will be lethal. Extreme heat events can be measured in different ways: the maximum temperatures reached (intensity), how often the events occur (frequency), and how long they last (duration). HS has negative effects on crop physiology (e.g. decreased photosynthesis, increased respiration) and plant growth and yield (Prasad et al., 2017). Another adverse effect of HS is the negative influence on the plant root system, which provides support, nutrient and water uptake, and transport to other plant organs (Valdés-López et al., 2016), resulting in disrupted pollination, flowering, root development, and root growth stages (Sehgal et al., 2017; Cho, 2018).

In many species, for both cool and warm seasons, yield reduction after HS is observed as a consequence of decreasing seed- or fruit-set percentage (Otero et al., 2011; Shah et al., 2011; Singh and Jwa, 2013; Jangid and Dwivedi, 2016; Fahad et al., 2017; Sehgal et al., 2017; Comastri et al., 2018; Devasirvatham and Tan, 2018; Govindaraj et al., 2018). In wheat, exposure to short episodes (2–5 d) of HS (>24 °C) at the reproductive stage (start of heading) resulted in substantial damage to floret fertility, while a mean daily temperature of 35 °C caused total failure. Increasing the duration of high temperature at this stage reduced the grain weight linearly (Maestri et al., 2002; Prasad and Djanaguiraman, 2014); similarly for pea (Bhattacharya, 2019), lentil (Barghi et al., 2012), and chickpea (Wang et al., 2006). An extensive review on the threshold temperatures for several crop species was reported by Kaushal et al., 2016. Table 1 shows threshold temperatures in vegetative and reproductive development for several crop species, together with literature indications about the losses (or gains) attributed to HS alone.

Table 1.

Yield losses due to heat stress in cool and warm season crops

Species	Threshold temperatures for the speciesa	World production in 2017b (kg ha–1)	Average yield reduction (%)	Reference
Cool-season crops
Barley (Hordeum vulgare L.)	Not reported	3136	15	Weichert et al. (2017)
Chick pea (Cicer arietinum L.)	15–30 °C for growth, 25 °C for reproductive growth	1015	19–50	Devasirvatham and Tan (2018)
Citrus (Citrus spp.)	35 °C for vegetative growth	9600	N/A	N/A
Lentils (Lens culinaris Medik.)	Not reported	1153	38–58	Sita et al. (2018)
Spinach (Spinacia oleracea L.)	Not reported	29 993	50	Yan et al. (2016)
Wheat (Triticum spp.)	20–30 °C for vegetative growth, 15 °C for reproductive growth	3531	6	Lobell et al. (2011); Zampieri et al. (2017); Comastri et al. (2018)
Warm-season crops
Grapes (Vitis vinifera L.)	Not reported	10 716	35–50	Greer and Weedon (2013)
Maize (Zea mays L.)	33–38 °C for photosynthesis and pollen viability	5755	7–40	Valdés-López et al. (2016); Zhao et al. (2017); Meseka et al. (2018); Prasad et al. (2018)
Peanut (Arachis hypogaea L.)	29–33 °C for vegetative growth, 39–40 °C for seed set and yield	1686	6	Prasad et al. (2001)
Potato (Solanum tuberosum L.)	Not reported	20 111	18–23	Hancock et al. (2014)
Rapeseed (Brassica napus L.)	About 30 °C for flowering	2195	Up to 85%	Koscielny et al. (2018); Sparks (2018)
Rice (Oryza sativa L.)	33 °C for biomass, 35 °C limiting for grain formation and yield	4602	3	Zhao et al. (2017)
Sorghum [Sorghum bicolor (L.) Moench]	26–34 °C for vegetative growth, 40 °C for reproductive growth and yield	1416	17–44	Tack et al. (2017)
Soybean [Glycine max (L.) Merr.]	26–36 °C for reproductive development, 39 °C lethal	2854	3–7	Valdés-López et al., 2016; Zhao et al. (2017)
Sunflower (Helianthus annuus L.)	Not reported	1804	10–70	Debaeke et al. (2017)
Tomato (Solanum lycopersicum L.)	37 °C for vegetative growth, 28–30 °C for reproductive development	37 600	28	Snider et al. (2012); Lamaoui et al. (2018)

^a Data from Luo (2011) and Kaushal et al. (2016).

b Data obtained by world production and world cultivated extension for each crop, from FAOSTAT 2017, http://www.fao.org/faostat/en/#data/QC/visualize.

HS can impair several physiological processes linked to seed size and quality. HS during grain filling markedly decreased starch accumulation in wheat (Hurkman et al., 2003), rice (Yamakawa and Hakata, 2010), and maize (Yang et al., 2018). The levels of sugars such as fructose, sugar nucleotides, and hexose phosphate also declined due to HS (Yang et al., 2018). The decrease in sugars may be related to assimilate utilization for purposes other than edible component production (Asthir et al., 2012). In maize, waxy grain starch content decreased, whereas protein content increased, resulting in a change of grain quality (Yang et al., 2018). Moreover, increasing temperature and CO₂ reduced protein and micronutrient content in grain (Chakraborty and Newton, 2011) and soybean (Li et al., 2018). In soybean under HS, the nutritional value of total free amino acids was reduced together with total protein concentration, while the oil concentration was significantly increased. As a general evaluation, reductions in total yield are mainly due to an alteration in balance of the source and sink activities that take place under HS.

The capacity of crop plants to overcome temperature stress has been interpreted in terms of avoidance, escape, or tolerance (Osmond et al., 1987). Avoidance is any mechanism that permits the plant to be or to become non-susceptible to the stressor effect, annulling in this way most of its deleterious effects (Hasanuzzaman et al., 2013). In escape, plants can alter their growth cycle before the stress hits hard, such as anticipating reproduction or shedding vegetative structures. From the agronomic point of view, tolerance is certainly more interesting; being based on the endurance of the plant in stress conditions, it does not involve substantial modification of the growth habit with potential deleterious effects on yield. A heat-tolerant plant is therefore inclined to continue its growth cycle quite independently of the stress (Barnabás et al., 2008).

The contributions of ‘omics’

Conventional plant breeding strategies based mainly on phenotype selection and qualitative genetics have led breeders, after the green revolution, to achieve continuous increases in seed yield and improved yield stability, independently of the environmental cost of achieving them (Setia and Setia, 2008). However, the growing worldwide demand for enhancing yields of major crops is placing pressure on breeding programs to provide elite cultivars more adaptable to the ongoing changes. The complexity of the information needed to meet this challenge has prompted advances in understanding the biochemical and molecular processes that underlie important metabolic, physiological, and developmental traits which affect the ability of plants to cope with the upcoming climate change. This has provided new insights into how plants function, and promoted the development of new scientific disciplines, mainly based on high-throughput approaches.

The progress of ‘omics’ technologies, in particular genomics, transcriptomics, proteomics, and metabolomics, has enabled direct and unbiased monitoring of the factors affecting crop growth and yield in response to environmental threats. Overall, omics constitute powerful tools to reveal the complex molecular mechanisms underlying plant growth and development, and their interactions with the environment, which ultimately determine yield, nutritional value (Setia and Setia, 2008; Soda and Wallace, 2015), and the required level of agricultural inputs. The following sections provide some examples showing the successful application of omics in crop improvement.

Genomics and transcriptomics

Molecular responses to HS have been investigated in many plant species to identify the complex processes and pathways regulated in acclimation and protection against temperature stress (Qu et al., 2013; Driedonks et al., 2016). Regulation of gene expression through transcriptional and post-transcriptional mechanisms is certainly linked to both stress recognition and stress response; this aspect has been investigated at different stress levels and comparing different genetic resources. The conventional transcriptomic approach has been applied to study gene expression changes under stress conditions in many crop species, such as rice (Sarkar et al., 2014; González-Schain et al., 2016; Mangrauthia et al., 2017), tomato (Frank et al., 2009; Bita et al., 2011), barley (Mangelsen et al., 2011), maize (Frey et al., 2015; Shi et al., 2017b), wheat (Qin et al., 2008), soybean (Chung et al., 2013; Gillman et al., 2019), brassica (Dong et al., 2015), and grape (Liu et al., 2012a). More recently, application of advanced genome-wide transcriptome analyses has elucidated the roles of relevant genes from HS perception to signal transduction and stress response (Table 2). In general, most of the HS-responsive genes are involved in primary and secondary metabolism, translation, transcription, regulation, and responses to processes such as calcium, phytohormone, sugar, and lipid signaling, or protein modifications including phosphorylation (Jha et al., 2014; Takahashi and Shinozaki, 2019). Up-regulation under HS conditions is confined mainly to transcription factors and HSPs.

Table 2.

Review of the transcriptomic, proteomic and metabolomic analyses performed to dissect the mechanisms involved in the heat stress response

Species	Primary stress	Secondary stresses	Tissue or development stage analyzed	Simulation conditions	Molecular techniques applied	References
Barley	Heat	N/A	Anthesis	42 °C	Transcriptomics (Affymetrix Barley1 GeneChips)	Mangelsen et al. (2011)
Barley	Heat	Drought	Plant leaf	26 °C	Metabolomics IC-MS/MS	Templer et al., (2017)
Barley	Heat	Mild drought and combination	Flag leaves	30 °C+50% FC	Transcriptomics (RNA-seq)	Cantalapiedra et al. (2017)
Chickpea	Heat	N/A	Whole plant development	Up to 42 °C	Proteomics (LC-MS)	Parankusam et al. (2017)
Chickpea	Heat	General oxidative stress	Plant leaf	37 °C	Targeted metabolomics	Salvi et al. (2018)
Citrus	Heat	Drought	Plant leaf	40 °C	Metabolomics GC-MS LC-MS	Zandalinas et al. (2017)
Grape	Heat	Various temperatures	Leaves	35, 40, and 45 °C	Proteomics iTRAQ LC-MS/MS	Jiang et al. (2017)
Grape	Heat	Followed by recovery	Plant leaves	43 °C	Proteomics (iTRAQ LC-MS/MS)	Liu et al. (2014)
Grapes	Heat	N/A	Leaves	45 °C	Transcriptomics (Affymetrix gene chip)	Liu et al. (2012a)
Lentils	Heat	N/A	Pods	Up to 33 °C	Enzyme activities	Sita et al. (2018)
Lentils	Heat	N/A	Leaves and pods	30–50 °C	Enzyme activities	Chakraborty and Pradhan (2011)
Lentils	Heat	Drought, also combined	Leaves and pods	30 °C	Enzyme activities	Sehgal et al. (2017)
Maize	Heat	Drought	Leaves	42 °C	Proteomics (iTRAQ LC-MS/MS)	Hu et al. (2015); Zhao et al. (2016)
Maize	Heat	Drought	Chloroplasts	42 °C	Proteomics (MALDI TOF/ TOF MS/MS)	Hu et al. (2015)
Maize	Heat	N/A	Seedlings	32–38 °C	Transcriptomics (RNA-seq)	Frey et al. (2015)
Maize	Heat	N/A	Seedlings	42 °C	Transcriptomics (RNA-seq)	Shi et al. (2017b)
Maize	Heat	High CO2	Plant leaf	32 °C	Metabolomics (GC-MS)	Qu et al. (2018)
Mung bean	Heat	Exogenous glutathione	Plant leaves	42°	Targeted lipidomics	Nahar et al., 2015
Peanut	Heat	N/A	Seedlings	40 °C and 30 °C	Physiological measurements, metabolomics	Singh et al. (2016)
Peanut	Heat	N/A	Leaves	45 °C	Transcriptomics (qRT–PCR), physiological measurements	Chakraborty et al. (2018)
Potato	Heat	N/A	Leaves, tubers	Up to 30 °C	Metabolomic (GC/MS), physiological measurements, transcriptomic (microarray)	Hancock et al. (2014)
Potato	Heat, drought		Leaves	35 °C	Transcriptomics (RNAseq)	Tang et al. (2016)
Potato	Heat	Potato virus Y (PVY)	Leaves	28 °C	Transcriptomics (qRT–PCR)	Makarova et al. (2018)
Potato	Heat		Tubers, skin and phelloderm	33–35 °C	Transcriptomics (qRT–PCR)	Ginzberg et al. (2009)
Rapeseed	Heat	N/A	Flowering	Up to 35 °C	GWAs	Rahaman et al. (2018)
Rapeseed	Heat	N/A	Developing seed	Up to 35 °C	Transcriptomics (95k EST microarray)	Yu et al. (2014)
Rapeseed	Heat, drought	Combination of the two stresses		29±0.5 °C from the 38th day after sowing, 30% field capacity	Metabolomics, physiological measurements	Elferjani and Soolanayakanahally (2018)
Rice	Heat	N/A	Anthers	Up to 37 °C in	Proteomics (iTRAQ LC-MS/MS)	Mu et al. (2017)
Rice	Heat	N/A	Grain and Leaves	30 °C	Transcriptomics (qRT–PCR)	Phan et al. (2013)
Rice	Heat	Open field	Leaves, early milky stage of rice grains, spikelets	42/32, 38, 28 °C	Proteomics (MALDI TOF/TOF MS/MS)	Liao et al. (2014); Das et al. (2015)
Rice	Heat	Cold	Suspension cell	44 °C	Proteomics (label-free in-gel digestion together with LC-MS/MS)	Gammulla et al. (2010)
Rice	Heat	Various temperatures	Seedlings, anthesis	35 °C, 40 °C and 45 °C, 38 °C	Proteomics (MALDI TOF MS/MS)	Han et al. (2009); Jagadish et al. (2010)
Rice	Heat	N/A	Anthers	38 °C	Proteomics (label-free in-gel digestion together with LC-MS/MS)	Kim et al. (2015)
Rice	Heat	N/A	Plants at flowering	38 °C for 1 d	Transcriptomics (RNAseq, large-scale qRT–PCR)	González-Schain et al. (2016)
Rice	Heat	N/A	Seeds	Germination at 30 °C (control) and 42 °C (stress)	Transcriptomics (RNA-seq, qRT–PCR)	Mangrauthia et al. (2016)
Rice	Heat and drought	Sugar starvation	Floral organs		Metabolomic and Transcriptomics	Li et al. (2015)
Rice	Heat	Oxidative stress	10-day-old seedlings	42±1 °C for 30 min with 10 mM H2O2	Transcriptomics (60mer rice 44k array, qRT–PCR)	Mittal et al. (2012)
Rice	Heat	N/A	Pollen	37 °C	Targeted metabolomics	Feng et al. (2018)
Rice	Heat	N/A	Leaves	45 °C	Transcriptomics (RNA-seq)	Fang et al. (2018)
Rice	Heat	N/A	Grains	38 °C	Transcriptomics (RNA-seq)	Liao et al. (2015)
Rice	Heat	Drought exogenous IAA	Plant leaf	40 °C	Target lipidomics	Sharma et al. (2018)
Rice	Heat	Exogenous brassinosteroids	Plant leaf	47 °C	Target lipidomics	Thussagunpanit et al. (2015)
Sorghum	Heat	Drought	Seedlings	28 °C and 50 °C, water withholding. Combination of the stress	Transcriptomics (microarray)	Johnson et al. (2014)
Soybean	Heat	High humidity	Pre-harvest seed	40 °C	Proteomics (MALDI TOF/TOF MS/MS)	Wang et al. (2012)
Soybean	Heat	N/A	3- to 4-week-old plants	45 °C for 3, 6, 24 h	Transcriptomics (qRT–PCR)	Chung et al. (2013)
Soybean	Heat	High CO2	Plant leaf	36 °C	Metabolomics (GC-MS)	Xu et al. (2016)
Soybean	Heat	N/A	Roots	40 °C	Proteomics (LC-MS/MS); Transcriptomics (RNA-seq)	Valdés-López et al. (2016)
Soybean	Heat	N/A	Dry, imbibed, or germinated seeds from heat-tolerant and heat-sensitive cultivars	Field conditions with heat stress	Transcriptomics (RNA-seq)	Gillman et al. (2019)
Spinach	Heat	N/A	Leaves	37 °C, 35 °C	Physiological, enzyme activity, proteomic (trypsin digestion and iTRAQ labeling), transcriptomics (NGS)	Yan et al. (2016); Zhao et al. (2018)
Sunflower	Heat	Light	Leaves and immature seed	37 °C	Transcriptomics	Hewezi et al. (2008)
Tomato	Heat	Melatonin		Up to 42 °C	Metabolomics	Xu et al. (2016)
Tomato	Heat	Oxidative stress	Flower buds	36 °C\25 °C for 3 months	Proteomics (SDS–PAGE followed by LC-MS/MS)	Mazzeo et al. (2018)
Tomato	Heat	Lipid antioxidant and galactolipid remodeling	5- to 6-week-old plants	38 °C	Metabolomics (MS)	Spicher et al. (2016)
Tomato	Heat	N/A	Different developmental stages of tomato pollen	38 °C	Proteomics (mass accuracy precursor alignment (MAPA) plus LC/MS)	Chaturvedi et al. (2015)
Tomato	Heat	Waterlogging in open field	Leaves	39 °C	Proteomics (MALDI TOF/TOF MS/MS)	Lin et al. (2016)
Tomato	Heat	N/A	Anthers	32 °C for 2, 6, 16 or 30 h	Transcriptomics (cDNA-AFLP; 90 K Custom TomatoArray 1.0)	Bita et al. (2011)
Tomato	Heat	N/A	Flower buds	43–45 °C for 2 h	Transcriptomics (Affymetrix GeneChip® Tomato Genome Array)	Frank et al. (2009)
Tomato	Heat	N/A	week‐old tomato plants	1 h at 39 °C	Transcriptomics (massive analysis of cDNA ends (MACE)	Fragkostefanakis et al. (2015)
Tomato	Heat	N/A	Pollen	38 °C	Metabolomics (LC-QTOF-MS plus LTQ Orbitrap XL, mass spectrometer)	Paupière et al. (2017)
Tomato	Heat	N/A	Pollen tubes	37 °C	Targeted metabolomics	Muhlemann et al. (2018)
Tomato	Heat	Cold	Plant leaves	38, 20, 10 °C	Untargeted lipidomics LC/MS	Spicher et al. (2016)
Wheat	Heat	N/A	Seeds in filling stage	37 °C for 4 h	Metabolomics, transcriptomics	Wang et al. (2015)
Wheat	Heat	Lipid alteration	Flowering	Up to 35 °C	Metabolomics	Narayanan et al. (2016)
Wheat	Heat	N/A	Seedlings	Up to 42 °C	Transcriptomics	Comastri et al. (2018)
Wheat	Heat	N/A	Seedlings	35 °C	Proteomics (MALDI TOF/TOF MS/MS)	Gupta et al. (2015)
Wheat	Heat	N/A	Leaves, stems, and spikes, flag leaf	38 °C, 37 °C	Proteomics (iTRAQ LC-MS/MS)	Lu et al. (2017); Kumar et al. (2019)
Wheat	Heat	Open field	Grain	40 °C	Proteomics (MALDI TOF/TOF MS/MS)	Wang et al. (2018)
Wheat	Heat	Low nitrogen	Plant leaf	32 °C	Proteomics (MALDI TOF/TOF MS/MS)	Yousuf et al. (2017)
Wheat	Heat	Drought	Post-anthesis	35 °C	Proteomics (MALDI TOF/TOF MS/MS)	Zhang et al. (2018)
Wheat	Heat	N/A	Flag leaf	35 °C	Proteomics	Wang et al. (2015)
Wheat	Heat	N/A	Leaf of heat-susceptible ‘Chinese Spring’ (CS) and heat-tolerant ‘TAM107’ (TAM)	40 °C for 1 h with and without heat acclimation (34 °C, 3 h)	Transcriptomics (wheat genome array)	Qin et al. (2008)
Wheat	Heat	N/A	Flag leaf, seedlings	37/17 °C for 3 days in anthesis stage	Proteomics	Lu et al. (2017)
Wheat	Heat	N/A	Spikelet post-anthesis	35 °C	Metabolomics (LC/HRMS)	Thomason et al. (2018)
Wheat	Heat	Drought	Seedlings	40 °C, 20 % (w/v) PEG-6000	Transcriptomics (RNA-seq)	Liu et al. (2015)
Wheat	Heat	N/A	Plant leaf	35 °C	Lipidomics ICP-MS	Narayanan et al. (2016)
Wheat	Heat	N/A	Pollen	35 °C	Lipidomics ICP-MS	Narayanan et al. (2018)
Wheat	Heat	N/A	Plant leaf	35 °C	Lipidomics ICP-MS/MS	Djanaguiraman et al. (2018)

Abbreviations: cDNA-AFLP, cDNA-amplified fragment length polymorphism; GWAS, genome-wide association study; IAA, indole-3-acetic acid; IC-MS/MS, ion chromatography tandem MS; ICP-MS, inductively coupled plasma MS; iTRAQ, isobaric tags for relative and absolute quantitation; LC/HRMS, liquid chromatography-high resolution MS; MACE, massive analysis of cDNA ends; MALDI TOF/TOF MS/MS, matrix assisted laser desorption/ionization-time of flight tandem MS; MAPA, mass accuracy precursor alignment; N/A, not applicable; NGS, next-generation sequencing; qRT–PCR, quantitative reverse transcription–PCR; QTOF, quadrupole time of flight; RNA-seq, RNA sequencing,

The activation and production of heat shock factors (HSFs) and HSPs and the increase in reactive oxygen species (ROS)-scavenging activity play a key role in the responses and acclimation of plants to HS (Maestri et al., 2002; Driedonks et al., 2016; Comastri et al., 2018). ROS-related genes are within the main up-regulated category under HS in different plant species (Chao et al., 2008; Chou et al., 2012; Mittal et al., 2012; Suzuki et al., 2014) and have been associated with basal heat tolerance (Almeselmani et al., 2006, 2009; Kang et al., 2009; Bhattacharjee, 2012). Moreover, ROS play an important role as signal molecules involved in the transduction of intracellular and intercellular signals controlling gene expression and activity of anti-stress systems (Petrov and Van Breusegem, 2012), suggesting a role for ROS in the activation of HSFs during HS (Driedonks et al., 2016).

Both HSFs and HSPs are central to the HSR and in the acquisition of thermotolerance in plants (Scharf et al., 2012; Ohama et al., 2017). As a result of advances in whole-genome sequencing, additional information has been gathered for the Hsp gene family (Chen et al., 2018). HS transcription factors (TFs) involved in heat sensing and signaling are highly conserved both structurally and functionally throughout the eukaryotic kingdom and are key players for the expression of Hsp genes (Scharf et al., 2012). The HSF gene family has been identified and characterized in several crop species such as grape (Liu et al., 2012b), soybean (Scharf et al., 2012; Chung et al., 2013), maize (Lin et al., 2011), wheat (Xue et al., 2014), rice (Mittal et al., 2009; Jin et al., 2013; Lavania et al., 2018), and tomato (Yang et al., 2016). Progress in genome annotation has led to the development of the specialized HEATSTER database (http://www.cibiv.at/services/hsf/info#anfang) which can identify and correctly annotate new HSF genes in plant species (Berz et al., 2019).

The involvement of HSPs in the defense response following HS has been extensively reported in several plant species (Park and Seo, 2015; Comastri et al., 2018). Gene expression analysis was performed in tomato leaves applying MACE (massive analysis of cDNA ends), which showed that 2203 genes (9.6% of the total) had enhanced transcript abundance in response to HS (Fragkostefanakis et al., 2015). Those encoding small HSP (sHSP) family members showed the strongest induction, with nine out of 100 Hsp40 genes up-regulated in response to heat. HSF genes showed basal and stable expression in non-stressed conditions and, for some, a strong induction in response to HS. These studies also reported the importance of TFs in the regulation of HSR, since 92 of the genes up-regulated upon HS are TF genes (bHLH, bZIP, ERF, MYB, and WRKY families). These are well-known major elements in HSR and thermotolerance, highlighting the complex regulatory networks activated beyond those directly controlled by HSFs. The tomato HsfA1a TF seems to have a unique function as a master regulator for acquired thermotolerance and cannot be replaced by any other HSF (Scharf et al., 2012).

In wheat, 6560 probe sets displayed changes in expression after heat treatment at 40 °C, with or without acclimation at 34 °C; representing Hsp genes, HSF genes, and genes encoding proteins involved in phytohormone biosynthesis/signaling, calcium and sugar signal pathways, RNA metabolism and ribosomes, as well as primary and secondary metabolism (Kotak et al., 2007; Qin et al., 2008). In barley, the effects of a short-term HS on central developmental functions of the caryopsis were studied at the transcriptional level (Mangelsen et al., 2011). Over 2000 differentially expressed genes were identified, equally divided between up- and down-regulation. The core HSR includes some conserved processes such as abscisic acid- (ABA) responsive gene activation, HSP-mediated protein folding, calcium signaling, ROS scavenging, and the biosynthesis of compatible solutes, which are rapidly induced. Within the down-regulated genes, it is noticeable that those involved in carbon and nitrogen metabolism and cellular growth, which determine long-term negative effects, differ significantly depending on the plant developmental stage at the time of the stress.

During grain development, the main negative effects of HS are related to the decreased accumulation of storage compounds, which can have a detrimental effect on both seed quality and final yield (Mangelsen et al., 2011; Hurkman et al., 2013). In wheat, HSPs can play a crucial role in gluten formation as a ‘glue’ between the reserve proteins (glutenins and gliadins) and starch (Maestri et al., 2002). During flowering, a significant reduction in the expression of non-essential photosynthetic genes may constitute an energy-saving strategy to facilitate other key stress responses with the aim of maintaining sugar metabolism and consequently protecting pollen viability (X. Li et al., 2015). Unraveling the complexity of high temperature tolerance may benefit from a systems biology approach (see Fig. 1).

Fig. 1.

Scatterplots obtained by multidimensional scaling of the matrix of biological process Gene Ontology (GO) terms across transcriptomic data obtained from heat-stressed tomato (A) and maize (B). The color of each spot indicates the level of enrichment from red (lowest p) to blue (highest p). The labels refer only to the most frequent GO terms. (A) Tomato transcripts identified by Fragkostefanakis et al. (2015) are analyzed through their Arabidopsis orthologs, 67 genes in total; (B) maize transcripts have been identified in Frey et al. (2015) as significantly induced, 454 genes in total. Enriched GO classes have been identified with AgriGO v.2 (http://systemsbiology.cau.edu.cn/agriGOv2/;Du et al., 2010) and the results visualized with ReviGO (http://revigo.irb.hr;Supek et al., 2011).

Proteomics

Protein profiling under HS conditions helps to identify stress-responsive proteins, and the detailed investigation of these proteins reveals their function in stress tolerance mechanisms (Priya et al., 2019). The synthesis and accumulation of HSPs is a prompt response after exposure to high temperature and it is considered one of the most important adaptive strategies to overcome the deleterious effects of HS (Wahid et al., 2007; Keller and Simm, 2018). The majority of HSPs are molecular chaperones involved in protein stabilization and signal transduction during HS (Sung et al., 2001; Arce et al., 2018). HSPs prevent accumulation of proteins with anomalous conformations and eliminate non-native aggregations formed during stress, with ubiquitin-mediated degradation of these proteins (Kotak et al., 2007).

Proteomics allows the study of the direct gene products, which often differ with the level of gene regulation. This technique has been applied to investigate heat responses in rice, wheat, tomato, maize, soybean, grape, and chickpea (for references, see Table 2). The types of technique utilized for these studies have been mostly 2D gel electrophoresis (SDS–PAGE) for protein separation, followed by MS:MALDI-TOF (matrix-assisted laser desorption/ionization-time of flight), MS, or TOF/TOF MS/MS for protein isolation, and iTRAQ (isobaric tags for relative and absolute quantitation) with LC-MS/MS for relative quantification (Zhang et al., 2006; Wiese et al., 2007). Other techniques have been deployed including label-free in-gel digestion together with LC-MS/MS or Orbitrap; SDS–PAGE followed by LC-MS/MS; and mass accuracy precursor alignment (MAPA) plus LC-MS (Chaturvedi et al., 2015). Many successful examples of proteomics applied to the study of HSPs and other HS-induced proteins have been reported (Malcevschi and Marmiroli, 2012; Table 2).

Different stages of plant development have been investigated using proteomics, from seedling, to flower (anthers and pollen especially in rice and tomato), to fruit, but also spikelet and grain maturation. In rice, where HS has been studied in isolation, all classes of HSPs (high and low molecular weight) have been found among the predominant protein categories, although other protein types and Gene Ontology (GO) classes varied extensively according to the part of the plant investigated. For instance, in pollen and anthers, late embryogenesis abundant (LEA) proteins were present in high numbers. Interestingly, proteins related to oxidative stress were identified in most of the studies on high HS, alone or in combination with other stresses. ROS are produced during HSR, which results in an oxidative imbalance within the cell. Furthermore, chloroplast function and photosynthesis are both strongly affected by ROS generated during HS, and it seems that HSPs can prevent ROS damage to the photosystems or the organelle structure. As an example, in maize, it was found that sHSP26 protects chloroplasts under HS (Hu et al., 2015). Other GO categories involved were sugar metabolism, the tricarboxylic acid (TCA) cycle, regulatory proteins, and proteins that function in energy metabolism.

More information on stress-associated active proteins (SAAPs) in wheat has recently been reported (Kumar et al., 2019). This study identified ~4272 SAAPs in wheat using absolute quantification methods. Some of the differentially abundant SAAPs identified were Rubisco, Rubisco activase (RCA), oxygen-evolving extrinsic protein (OEEP), HSP17, superoxide dismutase (SOD), catalase (CAT), and calcium-dependent protein kinase (CDPK). Pathway analysis showed the carbon assimilation pathway, followed by starch metabolism, to be most perturbed under HS in wheat. A positive correlation was established between the expression of SAAPs at transcript and protein levels in wheat under HS (Kumar et al., 2019).

Adaptive responses to HS also involve various post-translational modifications (PTMs) of proteins. There is a detailed literature on the effects of HS on PTMs such as protein phosphorylation in rice and wheat (Chen et al., 2011; Wu et al., 2014). Currently, >300 PTMs have been detected and partly characterized, and the numbers are increasing (Wu et al., 2016). HS has also been shown to enhance the small ubiquitin-like modification (SUMOylation) of particular proteins (Miller and Vierstra, 2011). SUMOylation of accessible lysines on target proteins is mediated in plants by a sequential three enzyme conjugation pathway: the E1 activating enzyme heterodimer (SAE1a or b combined with SAE2), a single E2 conjugating enzyme (SCE1), and at least two E3 ligases (SIZ1 and MMS21/HPY2). In Arabidopsis thaliana L. (Heynh), the activity of the HS TF HSFA2 was shown to be regulated by SUMOylation (Cohen-Peer et al., 2010). Moreover, in potato, rapid changes in protein SUMOylation and serine phosphorylation were observed in response to HS (Colignon et al., 2019). In addition, the overexpression of SlSIZ1 E3 ligase was shown to enhance heat tolerance in tomato (Zhang et al., 2018).

Metabolomics and lipidomics

Major environmental stresses cause metabolic reorganization towards homeostasis, maintaining essential metabolism and synthesizing metabolites with stress-protective and signaling characteristics. In plants such as tomato (Paupière et al., 2017), maize (Qu et al., 2018), barley (Templer et al., 2017), wheat (Thomason et al., 2018), soybean (Xu et al., 2016), and citrus (Zandalinas et al., 2017), metabolism reprogramming under HS has been studied by untargeted metabolomics. For example, in maize, under CO₂ stress and a sudden HS, malate, valine, isoleucine, glucose, starch, sucrose, proline, glycine, and serine were successfully found as key players in the combined stress response (Qu et al., 2018). Under both heat and drought stress, amounts of amino acids, and antioxidants such as glutathione and α-tocopherol, were highly affected in two cultivars of barley (Templer et al., 2017); in particular, this combination of stresses led to accumulation of free amino acids in the leaf. In post-anthesis wheat, the amounts of some amino acids increased, while others decreased [l-arginine, l-histidine, l-tryptophan. l-threonine, 4-aminobutanoate (GABA), l-aspartate, and l-phenylalanine]; sugars, sugar alcohols, and other organic compounds were among the major groups of metabolites whose concentration decreased in tissue due to HS.

Thomason et al. (2018) identified the metabolites that exhibited the greatest decreases during HS in wheat: drummondol, anthranilate, dimethylmaleate, galactoglycerol, guanine, and also glycerone. Soybean subjected to both HS and excess CO₂ had decreased starch, fructose, glucose, sucrose, and maltose concentrations, whereas pinitol increased with temperature (Xu et al., 2016). Zandalinas et al. (2017) found in citrus that water stress and HS, and their combination, altered metabolite levels related to glycolysis and the TCA cycle, the chloroplastic phase of the glyoxylate/dicarboxylate cycle, and the quantity of polar metabolites participating in the phenylpropanoid pathway arising from shikimic acid, depending on the level of stress. In tomato pollen, under HS, most of the putatively identified secondary metabolites belonged to three major groups: flavonoids, polyamines, and alkaloids (Paupière et al., 2017). Other examples come from untargeted metabolomics experiments where it was found that salicylic acid, ascorbic acid, sulfur-containing molecules (glutathione), phenolic secondary metabolites, and in general all antioxidant molecules can reduce HS symptoms in many plant species (Mobin et al., 2017; Feng et al., 2018; Muhlemann et al., 2018; Niu and Xiang, 2018; Salvi et al., 2018; Ihsan et al., 2019). Other major aspects of high temperature stress are cell signaling, molecular transport, and the changes in lipid metabolism as a basic mechanism to regulate membrane fluidity.

Lipids, which are major components of the membranes of cells and organelles, are among the first targets of ROS produced during HS and directly by high temperatures (Narayanan et al., 2016, 2018). Stress-induced lipid peroxidation of unsaturated fatty acids and other changes in membrane lipid profiles can lead to membrane damage, electrolyte leakage, and cell death (Liu and Huang, 2000). Under severe HS and long-term HS conditions, both chloroplastic and extra-plastidial glycerolipids are oxidized, yielding oxylipin-containing glycerolipids and other cytotoxic molecules such as acrolein and methyl vinyl ketone (MVK), derived from peroxidation of trienoic ω-3 fatty acids (Vu et al., 2012).

Accumulation of highly saturated fatty acids helps confer HS tolerance by reducing structural membrane fluidity, which is increased at high temperatures (Escandón et al., 2018). There is an interesting correlation between the type of metabolites involved and the need to protect specific cellular functions or cell compartments from the adverse effects of stress. In addition to acyl oxidation, plants regulate other aspects of membrane lipid composition in response to changes in temperature; in particular, the chloroplast membranes that host the photosynthetic apparatus are thought to be highly vulnerable to damage caused by HS (Welti et al., 2007; Zheng et al., 2011). Severe HS results in increasing frequency of membrane phase separation of non-bilayer-forming glycerolipids. Moderate HS (30–45 °C) results in thylakoid grana membrane destacking (Higashi and Saito, 2019). During HS, the endoplasmic reticulum (ER) is the major site for membrane phosphatidylcholine (PC) synthesis in plants (Jessen et al., 2015). Free fatty acids are exported from chloroplasts and re-esterified to CoA in the cytosol. PC synthesis in the ER utilizes a mixed pool of acyl-CoA substrates via the Kennedy pathway and the PC acyl editing pathway (Lands cycle) (Wang et al., 2012a). Fatty acid desaturase 2 (FAD2) converts PC-bound 18:1 to 18:2 in the ER and, during HS, FAD2 mediates polyunsaturation of ER glycerolipids and plays a role in the plant ER stress response to HS (Martinière et al., 2011). Chloroplastic galactolipid biosynthesis through the eukaryotic pathway needs a transfer of glycerolipids from the ER to chloroplasts by phospholipid flippase ALA10 which interacts with FAD2 affecting the degree of PC saturation. LACS4/9 proteins are also involved in this mechanism (Botella et al., 2016).

At the chloroplastic outer and inner membranes, five trigalactosyldiaglycerols (TGDs) and proteins TGD1, 2, 3, 4, and 5 form a transporter complex that mediates ER to plastid lipid trafficking (Xu et al., 2008). However, there is no concerted induction of these membrane proteins at the transcriptional level under HS in Arabidopsis (Higashi et al., 2015). In chloroplasts, long-term HS (>1 d) decreases the levels of glycerolipids containing 16:3 and/or 18:3, and reduces the activity of all FAD enzymes (Higashi and Saito, 2019). In general, during HS and other environmental stresses, conversion of monogalactosyldiacylglycerol (MGDG) into digalactosyldiacylglycerol (DGDG) increases through the action of DGDG synthases, encoded by DGD1 and DGD2. Terrestrial plants increase the ratio of DGDG to MGDG to improve thylakoid membrane stability under various abiotic stresses (Higashi et al., 2018). MGDGs are also decreased through the action of specific lipases (DAD1; PLIP1, 2, and 3; and HIL1). These recent studies suggest that lipases which are localized in chloroplasts have an important role in the lipid remodeling process under HS (Higashi and Saito, 2019).

Glycolipids and phospholipids are converted to triacylglycerol (TAG) in the form of oil bodies, as a transient storage depot for acyl groups prior to membrane lipid recycling/degradation: phospholipid:diacylglycerol acyltransferase 1 (PDAT1) transfers an acyl group from PC to diacylglycerol (DAG), and produces TAG. Mutant pdat1 seedlings are more sensitive to HS than the wild type; therefore, PDAT1 contributes to HS-induced TAG accumulation, and plays a role in HS response in Arabidopsis seedlings (Mueller et al., 2017). Other chloroplast-localized enzymes that maintain the structure of the thylakoid membranes during HS are FAX 1, 2, and 3, and VIPP1(Zhang et al., 2012; N. Li et al., 2015). HS also changes the lipid composition of the plasma membrane and of the ER; PC and PE (phosphatidylethanolamine) phospholipids are major components of extra-plastidial membranes. A decrease in the PE polyunsaturation level is likely to improve membrane stability under HS (Narayanan et al., 2016). Long-term HS increases phosphatidylserine (PS) content in the leaves of wheat (Narayanan et al., 2016).

Most of the lipidomics investigations of the response to HS have been performed on Arabidopsis, with only a few studies on crops such as wheat and rice. One example of lipidomics applied to HS in wheat found that the decrease in the photosynthetic rate under HS is due to lipid desaturation, oxidation, acylation, and consequent damage to organelles (Djanaguiraman et al., 2018). The membrane protein–lipid associations are of great importance for the maintenance of membrane fluidity; in this context, overexpression of the gene OsFBN1 (coding for a fibrillin) facilitates the import of lipids to the chloroplast during HS and the consequent grain filling (Djanaguiraman et al., 2018).

It has been suggested that diminishing the ROS pool directly has positive consequences on the state of the lipids. For example, modulating antioxidant defenses and the detoxification of methylglyoxal (MG) through the application of exogenous glutathione to mung bean [Vigna radiata (L.) R.Wilczek] increased the tolerance of the plants to HS in the short term (48 h) (Nahar et al., 2015). The increase in certain plant hormones can also be beneficial to the lipid status in rice during HS (Lim et al., 2017). For example, high concentrations of indole-3-acetic acid (IAA) or brassinosteroids can successfully initiate a mechanism for the decrease of primary ROS caused by HS that attack lipid membranes (MGDGs and sulfoquinovosyldiacylglycerols) and thylakoids (Thussagunpanit et al., 2015; Sharma et al., 2018). In a non-targeted lipidomics experiment in rice under both HS and water stress, 298 lipids responded to the changes in conditions; of these, 128 were identified. Interestingly, in the case of HS, a decrease in the unsaturation of lipids was predominant and can be linked to an increase in the cell membranes’ rigidity (Navarro-Reig et al., 2019). In a lipidome-wide study of tomato leaf under HS, changes in 791 molecules were detected between 20 °C and 38 °C. These results indicate that the most important changes at the lipidome-wide level occur in tocopherols, plastoquinone/plastoquinol (and their metabolites), and in the degree of fatty acid saturation of galactolipids (Spicher et al., 2016).

Epigenetic modifications

The epigenetic changes that occur at the DNA level through methylation of cytosine residues or at the level of chromatin by post-translational modifications of histones can result in altered gene transcription, and are an important mechanism in regulating gene expression during development and in response to environmental stimuli (Zhang and Hsieh, 2013). The persistence of a memory of temperature stress is dependent not only on the persistence of HSPs and their TFs but also on the expression of certain genes relevant for epigenetic signatures (Liu et al., 2015). A primary stress episode can sensitize plants for acclimation to subsequent stresses, resulting in faster or stronger changes in gene expression upon repeated exposure (Vriet et al., 2015), constituting a physiological priming mechanism. This primary event is fundamental in the acquisition of thermotolerance (Sanyal et al., 2018). A comprehensive RNA-Seq (RNA sequencing) analysis of gene expression and splicing events in HS-primed and non-primed plants has been carried out (Ling et al., 2018). This study showed alternative splicing as a novel and significant part of HS priming-induced memory and that this is critical for enhanced stress tolerance. However, most studies have been carried out in Arabidopsis, and in cultivated plants the understanding of the molecular mechanisms involved still represents a challenge (Friedrich et al., 2019).

Further epigenetic control is due to miRNAs, which are emerging as factors involved in transcriptional regulation and HS memory. miRNAs are involved directly in HS adaptation by acting as post-transcriptional regulators. Plant miRNAs bind their target transcripts and cleave and/or degrade the target mRNA molecule (Budak and Akpinar, 2015; Alptekin et al., 2017). The regulation of stress-responsive genes through activation of miRNAs is particularly important under abiotic stress conditions (Liu et al., 2017; Gahlaut et al., 2018).

Experimental validation of miRNA targets is still a challenge in polyploid crops including wheat. Recently, HS-responsive miRNAs have been reported from several susceptible and tolerant cultivars, with responses primarily being invoked within 24 h of treatment (Qin et al., 2008; Kumar et al., 2015). Furthermore, Ravichandran et al. (2019) performed a detailed miRNA and isomiR (miRNA isoforms) annotation and validated the HS-regulated miRNA target genes associated with thermotolerance. They confirmed a high degree of conservation between dicots and monocots for miR156 regulation of squamosa promoter-binding-like protein (Stief et al., 2014), miR159 regulation of MYB transcription factor (Wang et al., 2012b), miR166 regulation of homeobox leucine-zipper protein (Arikit et al., 2014), and miR398 regulation of SOD (Jagadeeswaran et al., 2009). New miRNA targets were also identified for organelle-specific transcripts such as pentatricopeptide repeat-containing proteins and mitochondrial transcription termination factor-like proteins.

Interestingly, several miRNAs showed altered expression patterns under heat and cold stresses in wheat, including miR164 (targeting HSP17) and miR319 (targeting a MYB transcription factor). These are up-regulated during a cold stress response but down-regulated in response to HS. On the contrary, miR160 and miR164 were down-regulated in response to heat, resulting in the induction of HSP expression, and up-regulated under cold stress. Regulating mechanisms triggered by high temperature can be reversed by cold stress, although the underlying mechanisms may be similar (Alptekin et al., 2017).

Breeding for heat tolerance

To ensure food security for future generations, agriculture must double the current crop production rate despite the predicted threats, which include climate change (Sedeek et al., 2019), decrease in arable land and desertification, salinization, and emerging diseases. Plant breeders need affordable and scalable solutions to achieve a more sustainable production. Approaches to reach the essential goal of producing more with less include (i) harnessing natural and artificial mutations; (ii) exploiting the available genetic resources to produce new genetic material more tolerant to HS and related secondary stresses; (iii) improving the ability to screen and identify the available sources of resilience; and (iv) developing new breeding techniques.

Conventional breeding approach

The effects of HS and prolonged heat waves on food production are heightened by a greater genetic uniformity in crop plants resulting from the narrowing of the varieties used in developed countries (Fu, 2015; Lopes et al., 2015). This has prompted increased efforts to identify new genetic resources and useful traits to mitigate or counteract the effects of climate change on crop productivity (Pignone et al., 2015; Janni et al., 2018) (Table 3).

Table 3.

Harnessing plant genetic resources for heat stress breeding

Species	Countries of origin	Number of accessions	Source	Type of study	Reference
Barley	Spain and Germany	N/A	Saatzucht Breun GmbH	Gene expression studies	Cantalapiedra et al. (2017)
Chickpea	Several countries	300	N/A	GWAS	Thudi et al. (2014)
Chickpe0a	India	280	ICRISAT	Identification of heat tolerance superior lines	Krishnamurthy et al. (2011)
Maize	N/A	100 inbred lines	N/A	Field trial	Naveed et al. (2016)
Rice	Several countries	455	N/A	QTL identification	Ye et al. (2015)
Rice	Several countries	511	IRRI	Identification of heat tolerance superior lines	Tenorio et al. (2013)
Tomato	Several countries	81	University of Naples (Italy)	GWAS	Ruggieri et al. (2019)
Tomato	N/A	44	TGRC, University of California, Davis	Identification of heat tolerance superior lines	Alsamir et al. (2017)
Wheat	Several countries	1711	CYMMIT	GWAS	Singh et al. (2018)
Wheat	Mexico	2255	CYMMIT	Identification of heat tolerance superior lines	Hede et al. (1999)

CYMMIT, International Maize and Wheat Improvement Center; GWAS, genome-wide association study; ICRISAT, International Crops Research Institute for the Semi-Arid Tropics; IRRI, International Rice Research Institute; N/A not applicable; QTL, quantitative trait locus; TGRC, Tomato Genetics Resource Center

It is well established that responses in crop species to abiotic stress, including HS, form a complex quantitative trait whose inheritance is controlled by a synergy between genes identified as QTLs, distributed throughout the genome. Traditionally QTL mapping has been used to identify new genetic variability and new sources of tolerance for introduction to breeding programs. However, QTL regions can be quite large and may contain many genes to be investigated as potential candidate genes, and many QTL studies had limited value for breeding because of low marker density. More recently, genotyping-by-sequencing (GBS) has allowed an increase in the number of markers, in particular SNPs (single nuclear polymorphisms), evenly distributed throughout the genome (Spindel and Iwata, 2018). In this way, it is now possible to obtain genetic maps with high resolution and precise mapping of QTLs, and in some cases identifying candidate genes controlling associated quantitative traits (Bhat et al., 2016). The application of genome-wide association studies (GWAS) permits narrowing down the candidate regions to explore specific haplotypes in natural populations and even wild species (George and Cavanagh, 2015; Verdeprado et al., 2018).

In wheat, several genomic regions associated with heat tolerance have been mapped using QTL analysis, often combined with GWAS and GBS. Major QTL clusters associated with drought and heat tolerance have been mapped on several chromosomes (Vijayalakshmi et al., 2010; Paliwal et al., 2012; Talukder et al., 2014; Acuña-Galindo et al., 2015; Shirdelmoghanloo et al., 2016; Sharma et al., 2017; Maulana et al., 2018). GWAS of sorghum seedlings under HS has identified associated key genomic regions, and specific alleles from these regions can be used to develop heat-tolerant sorghum cultivars (Chopra et al., 2017). Through inheritance studies, Marfo and Hall (1992) reported two dominant genes controlling most of the heritable tolerance to heat at pod set in cowpea. Four QTLs were identified in cowpea as associated with pod set number per peduncle under HS, and markers were utilized in breeding applications (Lucas et al., 2013; Pottorff et al., 2014). Syntenic analysis of the closely related soybean genome identified HSPs and HSFs in these QTL regions (Liu et al., 2019). In rice, heat tolerance at the flowering stage is controlled by several QTLs that have been used in breeding programs based on gene pyramiding (Ye et al., 2012, 2015; Kilasi et al., 2018).

A major QTL for thermotolerance was successfully identified and cloned in African rice (Oryza glaberrima Steud.); thermo-tolerance 1 (TT1) encodes an α2 subunit of the 26S proteasome involved in the degradation of ubiquitinated proteins. The allele OgTT1 of the thermotolerant accession permitted increased thermotolerance through a more efficient elimination of cytotoxic denatured proteins and effective maintenance of heat response processes than the non-tolerant cultivar carrying the OsTT1 allele. Overexpression of OgTT1 was associated with enhanced thermotolerance in rice, Arabidopsis, and Festuca elata Keng f. ex E.B.Alexeev (X.M. Li et al., 2015).

In tomato, several QTLs involved in heat tolerance have been identified (Xu et al., 2017), while two QTL hot spots for heat tolerance with respect to grain yield were found in maize (Jha et al., 2014; Frey et al., 2016).

Screening for genotypic differences, and effective selection techniques, are crucial for identifying heat-tolerant parental sources and for inheritance studies, and also for crop improvement by combining other traits which are influenced by heat tolerance (Marfo and Hall, 1992; Hall, 2004). Since photosynthesis and reproductive development processes are directly adversely affected by HS (Prasad et al., 2008), desired characteristics of a heat-tolerant variety will be higher photosynthetic rates (e.g. stay-green leaves), enhanced membrane thermostability, and stable pod set or grain production under high temperature conditions (Bita and Gerats, 2013). Methods for breeding for heat tolerance in cowpea involved selection for abundant flower production and greater pod set under higher night temperatures and long-day growing conditions (Marfo and Hall, 1992). These efforts resulted in release of the HStolerant cowpea variety California Blackeye 27 (CB27) with better yield (Ehlers et al., 2000). This variety was crossed with Pima (a Nigerian heat-tolerant germplasm) giving Apagbaala, a cowpea variety intended for Ghana (Padi et al., 2004). However, this was not heat tolerant in Ghana even though it performed well under Californian growing conditions.

The common bean (Phaseolus vulgaris L.) has also been investigated. Initial screening at CIAT (Centro Internacional de Agricultura Tropical) of a germplasm core set for HS tolerance identified 30 resistant lines. Breeding programs utilizing these genetic resources helped to develop heat-tolerant bush (CIAT, 2006) and climbing bean lines (Blair et al., 2006). Moreover, introgression of Tepary bean (Phaseolus acutifolious A.Gray) genes into P. vulgaris gave an interspecific line that was used as a parent for breeding heat-tolerant lines (Polanía et al., 2017). Recent screening of chickpea genotypes indicated the existence of extensive genotypic variation for reproductive stage heat tolerance; further studies led to the release of heat-tolerant breeding line ICCV 92944 for late-sown conditions (Gaur et al., 2019).

Targeted breeding efforts will help to build heat tolerance in crops (Reynolds et al., 2011). A conceptual model to improve heat tolerance in wheat was proposed involving genetically determined physiological traits such as light interception, radiation use efficiency, and partitioning of total assimilates. The physiological breeding approach combines all these traits towards generating a cumulative genetic effect on yield (Cossani and Reynolds, 2012). Data sets from three different spring bread wheat nurseries at CIMMYT (International Maize and Wheat Improvement Center) with different breeding goals were extensively analyzed, and the results showed that spring wheat breeding targeted against abiotic stress delivers better genetic gains in warmer environments (Gourdji et al., 2013). The performance of yield traits under heat and non-stressed environments has been used to identify heat-tolerant genotypes for breeding programs (Ni et al., 2018; Gaur et al., 2019). The identification of critical genes controlling heat tolerance in common wheat led to the release of new cultivars of bread wheat and durum wheat capable of withstanding severe heat (Tadesse et al., 2019), for example the cultivar Faraj, which is able to maintain yield under heat and drought conditions El Hassouni et al. (2019).

It has been shown that compared with direct selection for grain yield, indirect selection through secondary traits with high heritability and significant association with grain yield under stress is a more effective approach in stress tolerance breeding (Bänziger and Lafitte, 1997; Bänziger et al., 2000; Bheemanahalli et al., 2017). In maize, traits associated with reproductive success under heat stress (anthesis–silking interval, pollen viability, stigma receptivity, tassel blast, tassel sterility, and seed set percentage under open pollinated conditions) and other morpho-physiological traits (leaf firing, senescence, and chlorophyll content) were studied along with grain yield for the selection of HS-tolerant germplasms (Alam et al., 2017). As a result, two maize genotypes, VL05728 and VL05799, with better seed setting due to reproductive success under stress were identified as tolerant lines for heat stress (Alam et al., 2017). Several maize breeding programs have successfully increased yields in HS conditions (Cairns and Prasanna, 2018). The identification of suitable donors, highly tolerant to HS and to the combination of HS and drought stress, was successfully achieved by evaluating 300 (Cairns et al., 2013) or 29 (CIMMYT) maize inbred lines (Dinesh et al., 2018). These studies are the best examples to show the significant molecular diversity among maize inbred lines selected for heat tolerance

Reproductive stage HS in rice causes substantial yield loss. Delayed flowering, reduced pollen dispersal, low pollen production, spikelet sterility due to poor anther dehiscence, and impaired starch synthesis during grain development are the major problem areas (Jagadish et al., 2016; Arshad et al., 2017). Natural genetic variation for HS tolerance exists in rice, and significant genetic components controlling these variations have been identified. Several QTLs and genes associated with HS tolerance have been reported (Ye et al., 2015;Shanmugavadivel et al., 2017; Kilasi et al., 2018), and some have been characterized (Krishnan et al., 2011), but more translational research is needed in this area. Guodao 6 and Xieyou 46 are heat-tolerant hybrids developed with stable high rates of grain setting and spikelet fertility under HS (Tao et al., 2008). Other varieties including Fusaotome (tolerant), Hanahikari, Koshijiwase, and Tentakaku (moderately tolerant) were used in breeding for grain quality under heat stress (Ishizaki, 2006). In rice, HS-tolerant traits from line N22 (Jagadish et al., 2008) have been used for introgression into other varieties for developing climate change-ready rice (Ye et al., 2015). QTLs for HS tolerance during the reproductive stage were identified using a recombinant inbred population between N22 and IR64 (Ye et al., 2012).

In tomatoes, reproduction is particularly sensitive to continuous mild heat, which mainly affects pollen viability. The Asian Vegetable Research and Development Center (AVRDC) has identified 39 tolerant lines after several years of natural genetic variation studies. Some of these were used in their tomato breeding programs which produced HS-tolerant lines, such as Equinox (Scott et al., 1995) and Sun Leaper (Gardner, 2000). Another example is the development of ‘Amelia’, a heat-tolerant tomato cultivar for tropical conditions (Gil et al., 2004). Identification of more QTLs associated with heat tolerance in tomatoes (Wen et al., 2019) could enhance breeding efforts for development of stress-tolerant tomato varieties.

The development of molecular tools to accelerate breeding is crucial for the identification of useful traits and their application in breeding programs. In particular, marker-assisted selection (MAS) to improve breeding efficiency has become commonplace. Many MAS strategies have been developed, including marker-assisted backcrossing with foreground and background selection, enrichment of favorable alleles in early generations, and selection for quantitative traits using markers at multiple loci and across multiple cycles of selection (Bassi et al., 2016). MAS requires the use of markers flanking the target locus, and is considered one of the most efficient methods when working on complex traits having a quantitative hereditary characteristic, such as HS tolerance (Tayade et al., 2018). MAS can be used in forward breeding to enrich the allelic frequency for a few desired traits with strong additive QTLs in early selection cycles, to introgress favorable alleles into an elite background, and for integration of (native) traits into a breeding pipeline. This approach has been used in maize to develop improved lines for stress-prone environments (Beyene et al., 2016).

The exploitation of GWAS has led to the approach called genomic selection (GS) a promising tool to design novel breeding programs and to develop new marker-based models for genetic evaluation. The most important factor for its successful and effective implementation in crop species is the availability of genome-wide high-throughput, cost-effective, and flexible molecular markers, having low ascertainment bias, suitable for large population sizes, as well as for both model and non-model crop species with or without the reference genome sequence (Bhat et al., 2016). As a pre-breeding tool, it can serve to identify genetic materials with beneficial variation for complex traits (Wang et al., 2018), predicting the breeding value of an individual within a breeding population. It also provides new opportunities to increase genetic gain of complex traits efficiently. GS has been applied to active breeding programs in several crop species including wheat (Song et al., 2017), rice (Spindel and Iwata, 2018), soybean (Duhnen et al., 2017; Matei et al., 2018), sunflower (Dimitrijevic and Horn, 2017), maize (Cerrudo et al., 2018), chickpea (Roorkiwal et al., 2016), and grapes (Fodor et al., 2014; Viana et al., 2016).

GS and marker-assisted recurrent selection (MARS), widely used in the private sector, are proving efficient for the development of novel cultivars in many crops (Tayade et al., 2018). The major advantage of GS with respect to MAS/MARS is that alleles with minor effects can be captured and used in selection (Cairns and Prasanna, 2018). For both approaches, success depends on excellent phenotypic characterization during the discovery or training phase, respectively.

New breeding techniques to increase heat tolerance

Genetic modification through biotechnology and other new breeding techniques (NBTs) is a powerful strategy that offers novel opportunities to improve crop adaptability. However, general public concerns and complex legislation are limiting the application of NBTs. Encouraging data collected from genetics can be exploited to significantly increase tolerance to both biotic stresses and abiotic stresses such as salinity, drought, heat, and cold (Zou et al., 2011; Raza et al., 2019). Several studies have used TF genes and other genes associated with abiotic stress tolerance as targets for development of new varieties (Lamaoui et al., 2018). Crops including wheat, maize, tomato, and rice have been genetically modified to enhance thermotolerance, targeting mainly HSPs and HSFs (Fu et al., 2008; Qi et al., 2011; Xue et al., 2015; Casaretto et al., 2016; Wang et al., 2016; Trapero-Mozos et al., 2018). Transgenic cotton plants developed to overexpress an HSP, AtHSP101, in pollen had improved pollen germination and pollen tube growth under high temperature. This significantly enhanced the overall heat tolerance of reproductive tissues and reduced yield losses due to high temperature, supported by both greenhouse and field evaluation of transgenic plants (Burke and Chen, 2015). A survey of the transgenic lines identified to enhance heat tolerance in several species is reported in Table 4.

Table 4.

List of selected heat stress- (HS) tolerant transgenic plants

Crop	Target gene or protein/sources	Promoter	Stress or trait	Reference
Maize	OsMYB55/rice	Maize ubiquitin Ubi1 promoter/overexpression	Increased drought and HS tolerance	Casaretto et al. (2016)
Maize	ZmNF-YB2/maize	Rice actin 1 constitutive promoter/overexpression	Enhanced drought tolerance and photosynthetic capacity	Nelson et al. (2007)
Potato	HSc70 allelic variant/ potato	HSc70 native promoter	Greater tolerance to HS as determined by improved yield	Trapero-Mozos et al. (2018)
Rice	HSP70/rice	CaMV 35S	Overexpression manifested enhanced tolerance to HS	Qi et al. (2011)
Rice	Athsp101 /Arabidopsis	CaMV 35S promoter	Increased tolerance to high temperature	Katiyar-Agarwal et al. (2003)
Rice	OsRab7	CaMV 35S	Greater tolerance to HS as determined by improved yield	El-Esawi and Alayafi (2019)
Rice	TaMBF1c/wheat	Maize ubiquitin 1	Higher thermotolerance than control plants at both seedling and reproductive stages	Qin et al. (2015)
Soybean	P5CR/Arabidopsis		Enhanced HS tolerance	De Ronde et al. (2004)
Tobacco (Nicotiana tabacum L.)	HSP70-1/tobacco	CaMV 35S	Transgenics possessed enhanced tolerance to HT stress	Montero-Barrientos et al. (2008)
Tobacco	HSP70-1/brassica		Enhanced tolerance to HT stress	Wang et al. (2016)
Tomato	HSP21/tomato	CaMV 35S	Overexpression protected PSII from temperature-dependent oxidative stress; early accumulation of carotenoids noted	Neta-Sharir et al. (2005)
Wheat	Hsf6A /wheat	Barley HVA1s promoter/drought inducible, up-regulated	Improved thermotolerance	Xue et al. (2014)
Wheat	EF-Tu/maize	Maize ubiquitin 1 promoter/overexpression	Improved thermotolerance	Fu et al. (2008)
Wheat	Sucrose transporter gene HvSUT1/barley	Hordein B1 promoter	Increased enhanced in sucrose transport and shows a superior performance for many yield-related traits compared with control	Weichert et al. (2017)
Wheat	TaHsfA6f/wheat	HVA1s	Improved thermotolerance	Xue et al. (2015)
Wheat	TaFER-5B/wheat	Maize ubiquitin 1	Enhance thermotolerance	Zang et al. (2017)
Wheat	TaGASR1/wheat	NA	Improved tolerance to HS and oxidative stress	Zhang et al. (2017)
Wheat	TaHsfC2a/wheat	NA	Improved thermotolerance	Hu et al. (2018)

The availability of genomic sequences for several crops together with genome editing techniques has opened up new breeding possibilities for almost any given desirable trait (Jaganathan et al., 2018). The most common technique available for genome editing, clustered regularly interspaced palindromic repeats (CRISPR)/CRISPR-associated protein 9 (Cas9), modifies a genome in a targeted manner, and has been used in nearly 20 crop species so far including rice, tomato, potato, cotton, soybean, maize, sorghum, and wheat (Table 5) (Ricroch et al., 2017; Jaganathan et al., 2018; Zaidi et al., 2018). There are at least 13 different patents on new CRISPR/Cas9 approaches, which makes editing even more simple and secure, and new perspectives in the production of novel genetic variability are opened up by its application as an allelic-drive tool, engineering and repairing pathways, and introducing specific point mutations or insertions (Guichard et al., 2019).

Table 5.

Genome editing approaches used for heat stress breeding

Crop	Target gene	Stress or trait	Reference
Maize	ARGOS8	Improved yield under drought stress condition	Shi et al. (2017a)
Rice	OsPDS, OsMPK2, OsBADH2	Abiotic stress tolerance	Shan et al. (2013)
Rice	OsMPK2, OsDEP1	Yield under stress	Shan et al. (2014)
Rice	GS3, Gn1a	Grain size and number increase	Shen et al. (2017)
Rice	GW2, GW5, TGW6	Grain weight increase	Xu et al. (2016)
Rice	Gn1a, DEP1, GS3	Grain size and number Increase in dense, erect panicles	Li et al. (2016)
Wheat	TaDREB2, TaERF3	Abiotic stress tolerance	Kim et al. (2018)

Currently, attempts exploiting genome editing to increase heat tolerance target several genes mainly involved in the ethylene response and TFs, with the final aim to increase yield under abiotic stresses including HS (Li et al., 2016; Shi et al., 2017a; Kim et al., 2018). However, despite the abundance of novel genetic material generated through CRISPR, few field experiments have been conducted associated with breeding programs.

Mutational breeding

Mutational breeding strategies have been applied since the late 1990s to generate new variability in plants (Gilliham et al., 2017; Uauy, 2017). An important breakthrough came with the development of the TILLING (targeting induced local lesions in genome) approach, which provides a relatively simple strategy to identify mutations (lesions) in a target sequence independently of their phenotypic effect (Uauy, 2017). TILLING approaches require a population of, typically, ethyl methanesulfonate- (EMS) induced mutants based on the capacity of the chemical agent to generate point mutations distributed randomly in the genome, and a screening method to identify individuals with mutations in the target gene (Wang et al., 2010). TILLING and its updated form, established for polyploidy species through the use of exome capture and the development of the in silico TILLING database (Krasileva et al., 2017), have been used extensively to investigate genetic variability in several species mainly targeting quality traits.

Although in TILLING approaches mutagenesis is untargeted and does not provide the versatility of genome editing, crops improved using chemical or radiation mutagenesis via TILLING are not regulated as transgenic organisms in most jurisdictions, increasing their commercial competitiveness with the more precise genome editing approaches (Kumar et al., 2017).

Recently, Comastri et al. (2018) identified a collection of four new small Hsp26 (sHsp26) alleles suitable for enhancing heat tolerance in durum wheat using both in silico and in vivo TILLING approaches. Following application of TILLING, the ability of a mutated HSP to enhance heat tolerance in tomato has also been demonstrated (Marko et al., 2019). In rice, a TILLING population has been screened for mutations in the HSP genes, and a number of lines showing preliminary enhanced tolerance to HS may be useful for future breeding programs (Yona, 2015; Table 6).

Table 6.

Mutational breeding in crops

Species	Target trait /genes	Reference
Barley	Starch increases	Sparla et al. (2014)
Bread wheat (Triticum aestivum L.)	Starch branching enzyme	Botticella et al. (2011)
Chickpea	Salt tolerance	Kaashyap et al. (2017)
Durum wheat (Triticum durum Desf)	Heat stress	Comastri et al., 2018
Durum wheat (Triticum durum Desf)	Amylose content	Sestili et al. (2015)
Durum wheat(Triticum durum Desf), bread wheat (Triticum aestivum L.)	High amylose	Slade et al. (2012)
Oat (Avena sativa L.)	Improved β-glucan, antioxidants and omega-3 fatty acid	Chawade et al. (2010)
Peanut	Allergen reduction	Knoll et al. (2011)
Peanut	LOX gene	Guo et al. (2015)
Potato	Waxy mutant	Muth et al. (2008)
Rapeseed	Erucic acid synthesis	Wang et al. (2008)
Rapeseed	Sinapine biosynthesis	Harloff et al. (2012)
Sorghum	Reduction of cyanogenic glucosides	Blomstedt et al. (2012)
Sunflower	Accumulation of fatty acids	Kumar et al. (2013)
Tobacco	Leaf yield	Reddy et al. (2012)
Tomato	Fruit biology and ripening	Okabe et al. (2011, 2013)
Tomato	Virus resistance	Piron et al. (2010)
Tomato	Increased pigment and nutrient content	Jones et al. (2012)
Tomato	Early flowering, a solitary flower	MacAlister et al. (2012)
Tomato	Reduced ethylene sensitivity, delayed fruit ripening, prolonged fruit shelf life	Okabe et al. (2011)
Tomato	Decreased ascorbate	Baldet et al. (2013)
Tomato	Decreased carotenoid content	Gady et al. (2012)
Tomato	Increased lycopene content	Silletti et al. (2013)
Tomato	Reduced phenolics content	Di Matteo et al. (2013)

Conclusions

There are several examples of heat-tolerant varieties in major crops such as wheat, maize, rice, tomato, and legumes successfully developed through conventional breeding. These efforts can be fast-tracked using modern genomic and phenotyping tools. The first HS gene was identified and cloned in tobacco with the pioneering work of Barnett et al. (1979). This gene was shown to encode HSP70, which strongly accumulated in heat-stressed tissue.

Subsequent molecular studies revealed a more complex system with the discovery of HSFs (reviewed in Guo et al., 2016), organelle-specific HSPs (Waters and Vierling, 1999; Waters, 2013), and linkage between HS and the induction of an alternative splicing system (Ling et al., 2018) capable of directing de novo HSP synthesis in conditions where most proteins were either produced abnormally or were degraded (Vierling, 1991), remodeling the global protein machinery in cells. However, neither these discoveries nor the networking of all the known elements have produced a satisfying picture of the plant heat stress response. QTLs for heat tolerance offer a different, potentially complementary, interpretation with a stronger emphasis on physiology of reproduction in heat stress conditions.

Current data indicate that at least two different genetic systems can protect plants from the otherwise lethal effects of heat stress: (i) a series of Mendelian HS genes acting possibly in a dominant or semi-dominant way; and (ii) a small number of QTLs which allow plant growth and reproduction under different stress conditions. There has been little progress using transgenic approaches in studies of model plant to crop plant translational genetics. Extensive knowledge has become available on physiological, biochemical, and molecular regulation through advanced phenotyping and multi-omics tools developed over the last decade. All of this suggests that significant advances in crop improvement will be achieved, resulting in the development of new high-yielding varieties with greater heat tolerance and adaptation to climate change.

Author contributions

The authors contributed equally to the preparation of the review. MJ took on the responsibility of connecting all the different tasks distributed as follows: EM, gene ontology and system biology; MM, protein and metabolites; MG, gene expression and epigenetic modification; BV, gene expression and breeding; and MJ, induction of HS response; NM and HTN coordinated the planning and the writing of the review.