2.1 Plant material, growth conditions and tissue collection

Two Phaseolus vulgaris L. cultivars, Red Hawk and Sacramento, were used in this study. Our previous research on these cultivars found that Red Hawk is heat-sensitive, while Sacramento is heat-tolerant (Soltani et al., 2019) but are otherwise very similar. Both cultivars belong to the kidney bean market class from the Andean gene pool indicating relative genetic similarity. For a comparative analysis of reproductive function and metabolism between Red Hawk and Sacramento genotypes, seeds were planted in Sure-Mix potting soil (Michigan Grower Products, Inc., Galesburg, MI) in three-gallon plastic nursery pots (Nursery Suppliers, Inc., Chambersburg, PA) in a completely randomized design. The plants were grown in a greenhouse with average day/night temperature of 27/20°C and fertilized twice per week with ½-strength Hoagland's solution starting at V1 growth stage. Plants subjected to heat stress were transferred to a greenhouse with an average day/night temperature of 34/27°C about 25 days after sowing (DAS), upon appearance of the first flower buds. The control plants remained in the original greenhouse with a day/night temperature of 27/20°C. During the treatment, heat-treated and control plants were watered daily to eliminate the effect of water limitation on reproductive development. Five days after the plants were subjected to heat treatment, a source leaf from the fourth node from the base of the plant was collected from control and heat-treated plants and immediately frozen in liquid nitrogen. The leaf harvest was performed within a window of 5 hr during the middle of the light period. Flowers at 5 days before anthesis were collected daily from both control and heat-treated plants and flash frozen in liquid nitrogen. The number of biological replicates for each experiment is indicated in the figure legends.

2.2 Collection of leaf exudates

After 5 days of exposure to elevated temperature, source leaf exudates were collected from heat stressed and control plants according to Zhang, Garneau, Majumdar, Grant, and Tegeder (2015) with some modifications. Briefly, one trifoliate source leaf with at least 4 cm of petiole was cut from plants in water containing 20 mM Na-EDTA to chelate calcium and prevent occlusion of phloem sieve tubes (Knoblauch, Peters, Ehlers, & van Bel, 2001). The leaves were immediately covered with a wet paper towel and plastic wrap to prevent transpiration and the petiole was re-cut while submerged in 20 mM Na-EDTA. The leaves were then allowed to sit in the EDTA solution for 5 min before the petioles were inserted into a 1.5 ml microcentrifuge tube with 600 μl of 20 mM Na-EDTA. The leaves were allowed to bleed in a box with high humidity for 1 hr. After bleeding, the leaves were carefully removed and 50 μl of 1 N HCl was added to the exudates, which were then vortexed, and centrifuged at 4°C and 15,000 x g for 30 min. A 200 μl aliquot was transferred to a new tube and vacuum dried (Savant Instruments, Hyderabad, India). The residue was then re-dissolved in 20 μl milliQ water and used for sucrose analysis.

2.3 Transport studies with ¹⁴CO₂

To test if sucrose phloem loading and allocation to anthers are affected by elevated temperature and if differences exist between bean cultivars, a radioactive tracer experiment was performed. Gaseous ¹⁴CO₂ was prepared from a reaction between 1 M acetic acid and a sodium bicarbonate (NaH[¹⁴C]O₃) solution. The radioactive ¹⁴CO₂ was first pulled into a previously evacuated stainless-steel tank. The tank was then pressurized with 5% ¹²CO₂ in air. The final specific activity was ~1 Ci/mol. On the day of the experiment, a fully expanded source leaf was clamped into the fluorescence head attachment of an LI-COR 6800 portable photosynthesis system (LI-COR, Lincoln, NE) and fed with 40 Pa unlabelled CO₂ until stomatal conductance and net photosynthetic rate stabilized under the following conditions: 1000 μmol m⁻² s⁻¹ light, 500 μmol s⁻¹ air flow rate, 25°C T_leaf, 1 kPa leaf to ambient air water vapour pressure difference (VPD). Once stabilized, ¹⁴CO₂ was fed to leaves by simultaneously turning on the tank with ¹⁴CO₂ connected to the LI-COR 6800 and turning off the machine's ¹²CO₂ mixer. The leaf was then fed with radioactive CO₂ for 10 min for tracer analysis. Flowers on the same branch as the leaf that was fed with ¹⁴CO₂ were harvested after 24 hr. The anthers were subsequently collected from these harvested flowers. Thirty anthers per plant were treated with 10 ml of Safety-Solve Complete Counting Cocktail (Research Products International, Mount Prospect, IL) and then the radioactivity was measured by Perkin Elmer Tri-Carb scintillation counter (Waltham, MA). These data are presented as disintegrations per minute (DPM).

2.4 Anther respiration measurement

The LI-COR 6800 was fitted with an adapter (LI-COR part # 6800-19) with the insect respiration chamber attached (LI-COR part # 6800-89) prior to respiration measurements. Flowers that developed under control and heat stress conditions were collected from plants at 5 days before anthesis and the anthers were carefully excised and weighed. After weighing, the anthers were transferred to a strip of moist paper towel and then inserted into the respiration chamber. The respiration chamber was exposed to 1,000 μmol m⁻² s⁻¹ white LED light during the measurement. The flow rate, chamber temperature, VPD and relative humidity were maintained as described above (see Section 2.3).

2.5 Pollen count and viability

Flowers that developed at control or elevated temperature were collected and anthers were excised. Nine anthers, each from an individual flower, were placed on a microscope slide with a drop of Lugol's solution and squished with a cover slip. Once the pollen grains were released from the anthers, the glass slide was placed under a stereomicroscope (Zeiss SteREO Discovery.V8, Karl Zeiss Microscopy, GmBH, Jena, Germany) and pollen count was performed.

Pollen viability was assayed by collecting flowers and anthers as described above. However, instead of using Lugol's solution, the anthers and pollen grains were stained with 1% triphenyl tetrazolium and then incubated for 1 hr in 30°C (Prasad, Boote, & Thomas, 2002). Viable pollen grains appeared as reddish-purple colour due to the formation of insoluble red formazan.

2.6 Carbohydrate extraction and analysis

Sucrose, glucose and starch were extracted following the method of Lu, Steichen, Weise, and Sharkey (2006). Briefly, frozen flower and anther tissues were ground to a fine powder and lyophilized. This lyophilized tissue was then weighed and placed in a microcentrifuge tube. Soluble carbohydrates were then extracted by adding 3.5% perchloric acid at 2 μl per mg tissue. The solution was homogenized by vortexing and allowed to sit on ice for 5 min. After incubation, samples were centrifuged at 18,000× g for 10 min at 4°C. The supernatant was transferred to a new microcentrifuge tube, while the remaining pellet was further extracted for starch. Neutralizing buffer (2 M KOH, 150 mM Hepes, 10 mM KCl) was added at ¼ the volume of the supernatant to bring up the pH to approximately pH 7 and mixed by vortexing. The mixture was then frozen in liquid nitrogen and thawed twice to precipitate the K-perchlorate, centrifuged at 18,000× g for 10 min at 4°C, and then the supernatant was transferred to a fresh 1.5 ml tube for soluble sugar analysis. Carbohydrate concentrations were determined using NADP(H)-linked assays, as described by Lu and Sharkey (2004).

2.7 Hydrogen peroxide measurement

Anther tissues were collected, as described above and immediately frozen in liquid N₂. A pool of anthers from 25 flowers represents each biological replicate. The anthers showed no significant difference in weight (Figure S1). The frozen tissues were transferred to 2 ml microfuge tubes with stainless steel beads and then ground using a tissue homogenizer. Hydrogen peroxide was extracted with 500 μl of 5% trichloroacetic acid in 50 mM sodium phosphate buffer (pH 7.4) and then vortexed. For hydrogen peroxide extraction and measurement with Amplex-Red reagent, a buffer of pH 7.4 was used as recommended (Chakraborty et al., 2016). The samples were then centrifuged at 14,000 x g and the supernatant was transferred to a new 1.5 ml tube. Hydrogen peroxide was quantified in 40 μl of tissue extract using the Amplex Red Hydrogen Peroxide/Peroxidase Assay Kit (ThermoFisher, Waltham, MA). The Amplex Red reagent reacts with H₂O₂ in a 1:1 stoichiometry producing a red-fluorescent oxidation product called resorufin in the presence of peroxidase (Zhou, Diwu, Panchuk-Voloshina, & Haugland, 1997). This kit had been shown to be sensitive enough as it was able to measure as little as 10 picomoles of H₂O₂ in a 100 μl sample volume. Absorbance was measured at 560 nm and H₂O₂ concentration was calculated using the molar absorptivity of resorufin (58,000 M⁻¹ cm⁻¹).

2.8 RNA extraction and expression analyses of PvSUT1.1

RNA-seq on leaf tissue was previously performed by Soltani et al. (Soltani et al., 2019; Figure S1). To verify the RNA-seq result on PvSUT1.1 expression, quantitative real time PCR (qRT-PCR) was carried out using an Eppendorf Realplex² Mastercycler (Eppendorf North America, Hauppauge, NY). Total RNA was extracted from leaves that were collected and immediately frozen in liquid nitrogen and ground into fine powder. Approximately 200 μl equivalent of frozen powder was aliquoted in a nuclease-free 1.5 ml microcentrifuge tube. Total RNA was extracted using E.Z.N.A. Plant RNA Kit (Omega Bio-Tek, Norcross, GA) and cDNA was synthesized using oligo(dT) primers and Moloney murine leukemia virus reverse transcriptase (Invitrogen, Waltham, MA; Santiago & Tegeder, 2016). A subsequent polymerase chain reaction was performed for 24 cycles. At least three biological replicates, consisting of 1:10 diluted cDNA, were used per run. Each reaction mix consisted of 1X Platinum Taq DNA polymerase buffer (Invitrogen, Waltham, MA), 200 μM deoxynucleotide diphosphate (Fisher Bio-Reagents, Waltham, MA), 2.5 mM MgCl₂ (Invitrogen), 0.25X SYBR Green I (Thermo Fisher Scientific, Waltham, MA), 50 nM ROX reference dye (Thermo Fisher Scientific, Waltham, MA), 0.5 U of Platinum Taq DNA Polymerase (Invitrogen, Waltham, MA), and 0.25 μM primers. The Phaseolus vulgaris ACTIN11 (PvACT11; Pélissier & Tegeder, 2007) gene was used as a control to determine the relative expression of PvSUT1.1.

2.9 Enzyme extraction and activity assays

Crude enzyme extracts were obtained from isolated anthers derived from flowers that had developed in control or elevated temperature conditions. Up to five biological replicates were made by collecting pools of anthers from 10 flowers at 5 days before anthesis stage. At this stage, the flower bud petals were still unopened and light green in colour. The anthers were extracted on ice with 400 μl of phosphate buffer (50 mM sodium phosphate buffer [Na₂HPO₄ + NaH₂PO₄] at pH 7.0, 1% polyvinylpolypyrrolidone) and macerated with a micropestle. The homogenized mixture was then centrifuged at 14,000× g for 5 min at 4°C and the supernatant transferred to a fresh 1.5 ml tube. The anther extract was then used immediately for the enzyme assay. Catalase activity was measured colorimetrically using 50 μl of enzyme extract and a catalase assay kit (Megazyme, County Wicklow, Ireland). For this kit, catalase activity was determined by measuring the decrease in H₂O₂ concentration in a two-step reaction. The first reaction involves incubation of the sample with a known concentration of H₂O₂ for 5 min at 25°C. The reaction is then stopped by addition of 15 mM sodium azide. The samples are then assayed for the formation of quinoneimine at 520 nm in a second reaction involving an enzyme-linked decomposition of H₂O₂ with 3,5-dichloro-2-hydroxy-benzenesulfonic acid, 4-aminoantipyrine and peroxidase. Glutathione reductase (GR) activity was measured following Cakmak and Marshner (1992). Briefly, 50 μl of enzyme extract was mixed with 25 mM EPPS buffer (pH 7.8), 0.5 mM oxidized glutathione, and 0.12 mM NADPH (dissolved in 100 mM carbonate buffer, pH 10.6) in a total volume of 200 μl. NADPH oxidation was measured by the decrease in absorbance at 340 nm over a period of 20 min using a 96-well plate reader (Molecular Devices Filter Max 5, San Jose, CA). The activity of GR was normalized by the total protein content of the anther extract. G6PDH activity was measured according to Preiser, Fisher, Banerjee, and Sharkey (2019) with some modifications. The reaction mixture included 50 mM Tris HCl (pH 7.4), 5 mM MgCl₂, 0.5 mM glucose 6-phosphate, 0.5 mM NADP, 10 mM DTT and 50 μl anther enzyme extract to a total volume of 1 ml. For this assay, an increase in absorbance due to NADPH production was measured over time using a spectrophotometer (Thermo Scientific GENESYS™ 10S UV–Vis, Waltham, MA). G6PDH activity was normalized by the total protein content of the anther extract. For measurement of glutathione peroxidase (GPX), ascorbate peroxidase (APX) and dehydroascorbate reductase (DHAR) activities, spectrophotometric enzyme assays were performed according to Noctor, Mhamdi, and Foyer (2016).

2.10 Statistical analysis

Graphs in each figure represent results from one plant growth set. Each experiment was repeated using tissues from an independently grown set of plants to ensure repeatability of the results. Each set of plants had biological replicates per genotype in each temperature condition. Data are presented as mean ± SEM. To determine the differences between population means, two-way ANOVA and Tukey HSD post hoc tests were performed using SigmaPlot version 14.0 (Systat Software, Inc., San Jose, CA).

3.1 Red Hawk produced fewer pollen grains than Sacramento at elevated temperature

Under control conditions, heat-tolerant (Sacramento) and heat-sensitive (Red Hawk) common beans appear to have the same amount of pollen produced and the anthers were able to release the pollen grains (Figure 1a). When these plants were subjected to heat treatment, Sacramento anthers appeared to produce similar amounts of pollen as anthers in control conditions (top, right panel; Figure 1a), whereas Red Hawk produced very few pollen grains (bottom, right panel; Figure 1a). Quantification of pollen grains show both Sacramento and Red Hawk produced similar amounts of pollen grains under at control temperatures. When exposed to heat stress, both genotypes exhibited a reduction in pollen production (Figure 1b). However, under heat stress Sacramento produced over twice as many pollen grains as did Red Hawk. Viability of the pollen grains that were produced under heat stress conditions was reduced to a similar degree in both Sacramento and Red Hawk (Figure 1c).

3.2 H₂O₂ accumulates in heat-sensitive bean anthers but not in the tolerant variety

To test if differences in anther ROS exists between bean genotypes, hydrogen peroxide content of anthers collected from heat-tolerant Sacramento and heat-sensitive Red Hawk was measured. Under the control temperature treatment, Red Hawk had a lower H₂O₂ content than Sacramento (Figure 2a). However, when subjected to heat stress, H₂O₂ content increased in Red Hawk to the levels observed in Sacramento under control conditions. For Sacramento, no change in ROS content was observed within the genotype when exposed to elevated temperature. As H₂O₂ can be degraded enzymatically, we next investigated anther catalase activity. In the control treatment, Sacramento had higher catalase activity compared with Red Hawk, even though anther hydrogen peroxide levels were slightly elevated (Figure 2b). Catalase activity responded differentially across the treatments for each cultivar, with it increasing in Sacramento but decreasing in Red Hawk in response to elevated temperature. As a result, Sacramento had a much higher catalase activity than Red Hawk in the heat treatment. Together, these data support the idea that Sacramento anthers are more heat-tolerant than Red Hawk. Further, these data suggest that Sacramento anthers are more responsive to increases in temperature than Red Hawk.

3.3 Effects of heat on enzymes related to ascorbate regulation of H₂O₂

In addition to catalase, H₂O₂ can also be removed through the glutathione and ascorbate systems (Figure 2c). The glutathione system depends on GPX, GR and NADPH (from G6PDH activity) while the ascorbate system depends on APX and DHAR activities. To determine if APX was involved in H₂O₂ quenching at elevated temperature in bean anthers, enzyme activity assays were performed. At control conditions, APX activity was significantly higher in Red Hawk than in Sacramento (Figure 3a). At high temperature, APX activity was increased in Sacramento but there was no change in Red Hawk. Further analysis of dehydroascorbate reductase (DHAR) activity showed similar activities between genotypes at control conditions (Figure 3b). At elevated temperature, enzyme activity was increased in both bean cultivars compared with controls but no significant difference was observed between genotypes.

3.4 Effects of heat on enzymes related to glutathione regulation of H₂O₂

Antioxidant enzymes require reductants from the OPPP, which is limited by the enzyme G6PDH. The activity of anther G6PDH was measured at control and high temperature conditions. Under control conditions, between the genotypes, the heat-sensitive Red Hawk had higher G6PDH activity than Sacramento (Figure 4a). In the elevated temperature treatment, G6PDH activity increased in Sacramento (about 78%) while it decreased in Red Hawk by about 37%. This resulted in Sacramento having higher G6PDH activity than Red Hawk at elevated temperature.

The NADPH produced by G6PDH is used by the glutathione reductase (GR) to cycle glutathione from the oxidized to the reduced form. Patterns of GR activity mirrored those of G6PDH activity (Figure 4b). Red Hawk had higher GR activity than Sacramento in the control treatment. However, Sacramento GR activity greatly increased and Red Hawk GR activity greatly decreased in the heat treatment (Figure 4b). As a result, Sacramento had higher GR activity than Red Hawk at elevated temperature. As GPX is connected to the glutathione cycle, we next measured GPX activity (Figure 4c), which showed no significant differences within and between genotypes and temperature conditions.

3.5 Leaf carbohydrate export correlates with heat sensitivity

The expression of a sucrose transporter gene hypothesized to be involved in phloem loading of sucrose (Santiago, Ward, & Sharkey, 2020; Soltani et al., 2019) was measured by qRT-PCR in both heat-sensitive and heat-tolerant genotypes. The expression of PvSUT1.1 was significantly reduced at high temperature in both bean varieties (Figure 4d; Figure S1). Interestingly, the expression of other characterized (PvSUF1/PvSUT1.4 and PvSUT1/PvSUT1.5) and putative sucrose transporters were not changed in both heat-tolerant Sacramento and heat-sensitive Red Hawk (Table S1). Leaf-exudate sucrose level was significantly different between genotypes at control conditions where the heat-sensitive Red Hawk's leaf exudate had higher sucrose content than Sacramento (Figure 4e). At high temperature, Sacramento did not show significant change versus control but a significant reduction of sucrose was detected in Red Hawk. Leaf-to-anther transport of carbon was assessed by ¹⁴CO₂ feeding. The two bean genotypes showed no significant difference in radioactivity at control conditions (Figure 4f). When plants were exposed to elevated temperature, no change in radioactivity was observed in Sacramento while a significant reduction was seen in Red Hawk.

3.6 Sucrose accumulation is altered in flowers and anthers of heat-treated plants

To determine the effect of reduced carbon export on sink soluble sugar levels under heat stress, sucrose and glucose were measured in floral tissues of both bean genotypes (Figure 5). In whole flowers, the temperature treatment did not affect sucrose levels in Sacramento, but a significant reduction was observed in Red Hawk (Figure 5a). As pollen grains develop inside the anther tissue, the amount of sucrose was also measured in anthers. In control conditions, Red Hawk had a higher sucrose content (Figure 5b). After exposure to elevated temperatures, no significant change was observed in Sacramento, while a significant reduction in sucrose level was found in heat-treated Red Hawk plants. This resulted in Sacramento having higher anther sucrose levels than Red Hawk in the elevated temperature treatment. Since sucrose is cleaved into glucose and fructose in sink tissues, we also quantified anther glucose content. Similar to sucrose levels, Red Hawk had higher glucose content than Sacramento at control conditions. When heat-treated, anther glucose was increased in Sacramento, while in Red Hawk anther glucose declined (Figure 5c). Sacramento had higher glucose content than Red Hawk in the elevated temperature treatment.

3.7 Anther cellular respiration rate is reduced in both common bean genotypes when subjected to high temperature stress

Genotype- and temperature-specific changes in free glucose accumulation in anthers suggest differences in its use during heat stress. Two parallel pathways that glucose is channelled to are the OPPP and glycolysis (and subsequent cellular respiration). We hypothesize competing allocation of glucose between these pathways. To test for changes in cellular respiration in both bean varieties, anther respiration rate was measured (Figure 6). At control conditions, respiration rate was similar between bean genotypes. After exposure to high temperature, the respiration rates in both genotypes were reduced. However, there was no significant difference between genotypes within high temperature treatment.

4.1 Hydrogen peroxide is scavenged via multiple pathways in common bean anthers

When H₂O₂ accumulation exceeds a plant cell's capacity to scavenge the ROS, the cells experience oxidative damage and initiate programmed cell death (Gill & Tuteja, 2010; Vacca et al., 2004). This has been demonstrated in cultured tobacco BY2 cells, where high temperatures increased levels of H₂O₂ and superoxide anions, which in turn induced programmed cell death (Vacca et al., 2004). There are multiple mechanisms by which H₂O₂ is scavenged to prevent cellular damage and programmed cell death. One of these pathways is through the enzyme catalase, where its overexpression was found to confer heat tolerance in plants (Chiang et al., 2014). This finding suggests that the ability to scavenge H₂O₂ via catalase during high temperature stress is a marker of a plant's heat tolerance capacity. This idea is further supported by our results, which found increased anther catalase activity in heat-tolerant Sacramento and reduced activity in heat-sensitive Red Hawk (Figure 2b).

Another way to scavenge H₂O₂ is through ascorbate and glutathione. These two antioxidants were traditionally considered to function in an integrated fashion, but recent work on Arabidopsis revealed that they have independent and specific functions in plants (see Foyer & Noctor, 2011 and references therein). As part of the ascorbate pathway, APX and DHAR enzyme activities were usually measured in response to various abiotic stresses (Badawi et al., 2004; Kaushal et al., 2011; Sairam, Srivastava, & Saxena, 2000). For heat stress, leaf APX activity was found to be lower in a heat-sensitive wheat (Triticum aestivum) genotype than the tolerant genotype (Sairam et al., 2000). As APX is considered a primary enzyme involved in H₂O₂ removal (Noctor & Foyer, 1998; Smirnoff, 2000), lower APX activity suggests heat-sensitive plants have lower ROS scavenging capacity. This, however, is not supported by our anther APX activity data in heat-sensitive Red-Hawk, and it is unclear why this is the case in Red-Hawk anthers. Conversely, plants with high APX activity were regarded as having higher ROS scavenging capability and stress tolerance, which agrees with our data for the heat-tolerant Sacramento bean variety. Between Sacramento and Red Hawk, no significant difference in APX activity was observed indicating that APX activity is not an important determinant of thermotolerance in anthers via H₂O₂ scavenging, as it is in leaves. This is further supported by the total ascorbate measured in anthers, where no significant difference was observed between temperature treatments within each common bean genotype (Figure S2).

The reduced (GSH) and oxidized (GSSG) forms of glutathione are interconverted by the GR enzyme. Like ascorbate, glutathione is an important antioxidant when plants experience stress, and the activity of the GR enzyme is a signature of oxidative stress resilience (Aono, Kubo, Saji, Tanaka, & Kondo, 1993; Foyer et al., 1995; Shu, Wang, Duan, Deng, & Meng, 2011). Consistent with previous reports that higher or lower GR activity is related to oxidative stress tolerance, we found higher GR activity in the more heat tolerant Sacramento cultivar in the elevated temperature treatment (Figure 4b). Interestingly, heat-tolerant Sacramento had lower GR activity at control temperature (Figure 4b). Based on this, it can be assumed that the heat-sensitive Red Hawk may always metabolically be on guard to stress by allocating higher resources (Figures 4e and 5b,c) to anthers at unstressed conditions. This is further supported by the fact that GR is involved in redox balance in plant cells (Foyer et al., 1995; Rao & Reddy, 2008).

In plants, GSH, rather than GSSG, is what is responsible for ROS balance and other biological processes including protection from effects of various environmental stress, and regulation of stress defense gene expression (Hasanuzzaman, Nahar, Anee, & Fujita, 2017; Wingate, Lawton, & Lamb, 1988). Ultimately, a high GSH:GSSG ratio prevents the formation of mixed disulphides, and is essential for maintaining glutathione as an antioxidant (Foyer, Lopez-Delgado, Dat, & Scott, 1997; Szalai, Kellős, Galiba, & Kocsy, 2009). GR activity is important for maintaining the cellular GSH pool and is an important determinant of GSH pool size. Therefore, it is likely that heat-tolerant bean anthers with high GR activity, like those of the Sacramento cultivar, also have a high GSH content that helps protect the developing tissue during heat stress. In contrast, the reduced GR activity at high temperature in Red Hawk implies that its anthers do not have enough GSH to protect them from heat stress. Regardless of environmental condition, antioxidant capacity is linked with carbon metabolism because GR activity is dependent on NADPH produced by the OPPP via the G6PDH enzyme. This relationship is especially crucial in sink tissues such as heat-sensitive anthers that receive carbon from source tissues like leaves. This idea is supported by studies that have found that heat-tolerant plants are able to export sucrose out of leaves for delivery to sinks, while heat-sensitive plants have reduced sucrose export at high temperatures (Basu & Minhas, 1991; Dinar & Rudich, 1985). Our results are consistent with these patterns, as the thermotolerance status of bean genotypes was correlated with their sucrose export capacity (Figure 4e,f). Taken together, our results indicate that thermotolerance of common anthers can be traced back to catalase and glutathione-dependent scavenging of ROS, and that the ascorbate pathway seems to play a minor role. The importance of glutathione in anther ROS levels, and by extension thermotolerance, appears to be connected with the supply of sugars from leaves and sugar metabolism in anthers, specifically through the OPPP.

4.2 Anther G6PDH activity at high temperature contributes to heat stress tolerance in heat-tolerant bean genotype

To date, a few carbon metabolism enzymes have already been shown to be involved in heat stress tolerance including invertase (Jiang et al., 2020; Rezaul et al., 2019; Sato et al., 2006), fructokinase and hexokinase (Karni & Aloni, 2002), sucrose-phosphate synthase and sucrose synthase (Kaushal et al., 2013), and glyceraldehyde-3-phosphate dehydrogenase (Kim, Guo, & Wang, 2020). In addition to these enzymes, we found that G6PDH contributes to heat tolerance in bean anthers. G6PDH catalyses the rate-limiting reaction in the OPPP and was previously shown to be involved in abiotic stress tolerance attributed to maintenance of H₂O₂ homeostasis as a result of increased reductase activity and ascorbate content (Cardi et al., 2015; Gong et al., 2012; Landi et al., 2016; Liu, Wang, Hu, Hu, & Bi, 2013). However, previous studies on G6PDH and stress tolerance were focused on the activity of the enzyme in leaves, roots, or callus tissue under drought and salt stresses. To the best of our knowledge, the role of G6PDH in anther resilience at high temperature is yet to be examined.

In general, G6PDH activity is regulated by substrate (G6P) availability (Hauschild & von Schaewen, 2003) and redox (Preiser et al., 2019). In our study, we found genotypic differences in G6PDH activity, even in the control treatment (Figure 4a). We also found that this is not due to substrate limitation since G6PDH activity was assayed in vitro with the same substrate (G6P) concentration in the reaction mixture for all samples (Figure 4a; see Materials and Methods). For potential differences in redox state of G6PDH between genotypes at high temperature, DTT was added to the assay reaction to fully reduce all enzyme extracts (see Materials and Methods). These conditions indicate that the response was seen in this study (Figure 4) is not due to differences in the substrate availability or redox state of the enzyme. Together, this means that the differences in G6PDH activity are due to differences in enzyme abundance between bean genotypes at elevated temperature. However, further studies will be needed determine whether the difference between the genotypes is due to differences in K_m or V_max. Our conclusion that G6PDH enzyme abundance might differ between genotypes is supported by a recent pollen proteomics study that revealed that high temperature negatively affects the protein synthesis and degradation machinery in plants (Jegadeesan et al., 2018). Based on that proteomics study by Jegadeesan et al. (2018), we hypothesized that the difference in G6PDH enzyme abundance is due to inherent differences in anther protein turnover between Sacramento and Red Hawk.

In addition to G6PDH, studies on other carbohydrate metabolism enzymes were reported to play a role in heat tolerance in plants and in pollen fertility. For example, repression of anther-localized cell wall invertase (CWIN; Goetz, Godt, Guivarc'h, Chriqui, & Roitsch, 2001; Sato et al., 2006) and sucrose-phosphate synthase (Kaushal et al., 2013) can cause male-sterility. Further, higher expression levels of invertase genes were observed in the tolerant versus sensitive varieties of rice (Jiang et al., 2020). The heat-tolerant rice cultivar in that study had higher NAD(H) and ATP content in the spikelets than the sensitive genotype, suggesting the heat-tolerant rice cultivar was able to better maintain energy homeostasis at elevated temperature. These findings are consistent with our results, especially for the heat-tolerant Sacramento. While we did not measure NADPH levels in anthers, the increased G6PDH activity and higher sugar supply within Sacramento points to better redox homeostasis of this genotype in the high temperature treatment than Red Hawk. Elevated NADPH levels theoretically can promote GR activity, which uses NADPH to cycle glutathione, and in turn helps regulate ROS levels. The connection between G6PDH and GR activities, and ROS regulation through redox balance is supported by a study in rice, where G6PDH was shown to be involved in the maintenance of cellular redox in cells subjected to salt stress (Zhang, Liu, Wang, & Bi, 2013).

Our study, together with studies on other carbon metabolism enzymes, emphasizes the importance of maintaining carbon metabolism in male reproductive tissues during periods of high temperature stress to prevent infertility. While our current data suggest a positive role of G6PDH in heat stress tolerance, further physiological and biochemical studies using g6pdh mutants are needed to confirm the role of the enzyme in heat stress tolerance of anthers.

4.3 Is there a shift in carbon flux between respiration and OPPP?

After the arrival of sucrose in the anthers, it is cleaved into fructose and glucose. Under normal conditions, glucose is channelled to both cellular respiration through glycolysis and the OPPP, which specifically produces NADPH, that can be used for ROS quenching and synthesis of amino acids, fatty acids etc. Between bean cultivars used in this study, differences in leaf sucrose export (Figure 1a), anther sucrose and glucose accumulation (Figure 2b–c), ROS accumulation (Figure 4), G6PDH and GR activities (Figure 5b,c), and anther respiration rate (Figure 6) suggest a shift in carbon flux within the anthers, at least for the heat-tolerant Sacramento genotype. Since leaf sucrose export and anther carbohydrates were unchanged but anther respiration was reduced in Sacramento, these results are consistent with our hypothesis that the partitioning of carbon between glycolysis and the OPPP is shifted in favour of the OPPP at high temperature (Figure 7). This is further supported by the increased G6PDH activity in Sacramento at elevated temperature, presumably to maintain ROS balance in anthers and prevent oxidative damage. In contrast, in heat-sensitive Red Hawk the reduced sucrose arrival in anthers means that all downstream pathways will also be negatively affected. In our study, these negative downstream effects included the accumulation of anther ROS, reduced anther respiration rate and decreased activities of G6PDH and GR. Reduced protein synthesis in anthers also might have contributed to reduced G6PDH activity in Red Hawk.