Lincomycin treatment
Lincomycin treatments have generally involved infiltration of detached leaves with the lincomycin solution (e.g. Davis et al., 2016; Yang et al., 2007). For the HT treatment in cowpea, however, we found that detached leaves resulted in strong declines in PSII efficiency contrary to what we observed in attached leaves. Therefore, our treatments were either through watering of soil with lincomycin solution or syringe infiltration of attached leaves with lincomycin solution. Both approaches gave similar results, although watering the soil had a slower response (in the order of several hours to days) since uptake was generally slow. For the syringe infiltration, effects were seen within a few hours. Briefly, the abaxial side of the leaf was infiltrated with 0.45 mM lincomycin hydrochloride using a syringe by the solution. Control plants were mock infiltrated with deionized water. Following infiltration, plants were kept under low light (50 μmol m-2 s-1) for 30 min to keep the stomata open and then 1 h in the dark before the HL treatment was imposed (GT+HL). For the HT+HL treatment, the temperature was increased to 40 °C 1 h after initiation of HL and to 45 °C after another 1 h. The temperature was kept at 45 °C for the next 5 h.
Determination of photorespiration
The velocity of rubisco for oxygenation (v o — used as a proxy for photorespiration) and for carboxylation (v c) were calculated from rubisco kinetics using the following equations as described in (Busch, 2013):
\(v_{c}=\frac{A+R_{l}}{1-\frac{\Gamma^{*}}{C_{c}}}\) (1)
\(v_{o}=2\bullet\left(v_{c}-A-R_{l}\right)\) (2)
where A is net assimilation as measured from gas exchange,C c is CO2 concentration in the chloroplast, and R l is the day (light) respiration, which is also referred to by R d.
The temperature response function of Bernacchi and others (Bernacchi, Singsaas, Pimentel, Portis Jr, & Long, 2001; Bernacchi, Portis, Nakano, von Caemmerer, & Long, 2002) were used g m(needed C c), Γ* and R l at different temperatures. vc calculated using other parameters from the gas exchange measurements or estimated based on data for soybean (Walker et al., 2017). A sample excel sheet for the calculation is attached as supplementary material.
The electron transport rate from gas exchange (ETRGE) was calculated using the following equation:
\(\text{ETR}_{\text{GE}}=4\bullet(v_{c}+v_{o})\) (3)
To enable direct comparison with the electron transport rate (ETR) obtained by chlorophyll fluorescence, the ETR assessed with LI-COR 6800 simultaneously with gas exchange parameters were used.
Data analysis
Depending on the experiment, we used 2-way (repeated measures) or 3-way ANOVA to determine differences and interactions between treatments (light intensity, temperature and/ or leaf maturity). Means separation was by Tukey’s honestly significant difference (HSD) test. Two-sample t-test was also used on specific datasets. Data analysis was performed in Genstat 18th Edition (VSNi, UK), Origin Pro 2019b (OriginLab Corp, Northampton, MA, USA) or in R version 3.6.1 (R Core Team, 2019).
RESULTS AND DISCUSSION
Effects of short-term exposure to elevated temperature on the partitioning of energy to photoprotection and photochemistry 
To determine the effect of light intensity on the temperature response of cowpea, 4-or 11- d-old-seedlings of two genotypes, Yacine (HL-tolerant) and 58-77 (HL-sensitive), were exposed to 300, 1000 or 1500 µmol photons m-2 s-1 PAR while the temperature was increased by 5 °C every 2 h starting from GT (30 °C) to HT (45 °C). Following these treatments, the plants were returned to GT for another 1 h to assess the ability to recover from the treatments.
Figure 1 shows the responses of photosynthetic parameters in Yacine to the temperature regime. Similar results were obtained in the HL-sensitive line, 58-77 (Supplementary Figure 1A). At LL (300 µmol m-2 s-1), ɸIIin young leaves remained nearly constant as the temperature was increased, consistent with electron flow predominantly light-limited. Mature leaves showed ɸII values that decreased significantly (p<0.001) with increasing temperature, down by ~10% of initial values at 45 °C, similar to earlier reports in tomato (Meng et al., 2017), At 1000 and 1500 µmol m-2 s-1, mature leaves showed only small changes in ɸII with temperature, but young leaves showed ɸII that increased significantly (by about 30% p<0.001) as temperature was increased to 45 °C, likely indicating an increase in the capacity for photochemistry, possibly reflecting an increase in heat tolerance with illumination as reported earlier (Havaux et al., 1991; Weis, 1985).
The relatively small effect of temperature on ɸIIis somewhat surprising given the expected strong temperature dependence of photosynthetic reactions and suggests that photosynthesis is subject to compensatory regulation or control by multiple processes. All of these effects were reversed , suggesting that they were not caused by irreversible changes.
In all cases, short-term exposure to increasing temperature led to reversible decreases in light-induced thylakoid pmf , as estimated by the amplitude of the dark-interval relaxation kinetics (DIRK) of the electrochromic shift [ECSt, Figure 1B (Takizawa et al., 2007)]. The decreases in pmf can be attributed to increased thylakoid proton conductivity, g H+, estimated from ECS decay kinetics (Baker, Harbinson, & Kramer, 2007; Kanazawa & Kramer, 2002; Kramer, Avenson, & Edwards, 2004) (see Figure 1C), and thus likely reflects increased ATP synthase activity (Kanazawa & Kramer, 2002) as was seen previously with moderate temperature increases in other species (Zhang & Sharkey, 2009; Zhang et al., 2009), although other factors such as proton slippage or membrane leakage cannot be ruled out. The pmf , acting through its pH component, is suggested to be responsible for activating q E (Li et al., 2002). The exception was mature leaves at low light, for whichpmf decreased, but g H+remained relatively constant with temperature. In this case, the decreases in pmf at higher temperature could be attributed to decreased ɸII , which should translate into lower proton influx by LEF.
Figure 1D shows the responses of NPQ to temperature. Note that these values were estimated using the NPQ(T) method (Tietz et al., 2017), which allows for more rapid and less invasive measurements, but results in NPQ values that are typically higher than those obtained with the classical method. In the case of older leaves, NPQ(T) (the sum of q I(T) andq E(T)) increased with temperature, but the opposite behavior was observed in younger leaves (Figure 1D), though in all cases the changes were relatively modest. In general, changes in total NPQ(T) were inversely related to changes inɸII . NPQ(T) was reversed during the recovery except under HL (1000 and 1500 µmol m-2s-1) where it remained elevated or increased, after return to GT. The q L fluorescence parameter, which estimates the fraction of PSII centers with oxidized QA (Kramer, Johnson, Kiirats, & Edwards, 2004) increased by ~25% at the highest temperature in HL, regardless of leaf age (Figure 1E). This result implies that electrons did not accumulate on the intersystem electron transfer chain, suggesting that changes in the capacity for LEF were compensated for by complementary adjustments in NPQ(T) so that the quantum yield of unregulated energy dissipation (ɸ NO) remained relatively constant over the course of the experiment (Supplementary Figure 1B).
CEF has been reported to increase under HT conditions in some species (Essemine, Qu, Mi, & Zhu, 2016; Sun, Geng, Du, Yang, & Zhai, 2017), possibly to provide extra ATP needed to maintain metabolic processes, or to maintain pmf and initiate pmf -related regulation of light capture and electron flow. Figure 1F plots the relative flux of protons through the ATP synthase (v H+), estimated by the decay of the DIRK or the ECS signal, against LEF. Increases in the ratio (v H+/LEF) would indicate elevated CEF relative to LEF (Livingston, Kanazawa, Cruz, & Kramer, 2010; Strand et al., 2015; (Walker, Strand, Kramer, & Cousins, 2014)). The slope of v H+/LEF is also dependent on the content of chloroplasts and ECS-active pigments. Thus, without calibration, we cannot directly compare slopes between leaves of different developmental stages. However, we can assess changes over short time periods, and thus we normalize the ratios to those obtained at 30 °C and note that in the mature leaves, generally the ratio increased with temperature (Figure 1G) at HL (both 1000 and 1500 µmol m-2 s-1), but declined in LL. Since we know that ɸII and thus LEF increased slightly or remained nearly constant in mature leaves with increasing temperature, the only way that the ratio would increase in HL is for CEF to increase. Thus, this data suggests that CEF increased with temperature in mature leaves under HL. On the other hand, the ratio declined with increasing temperature in young leaves, regardless of light intensity, indicating that CEF decreased with temperature. The observation in mature leaves is consistent, whereas that in the young leaves is contrary to earlier reports (Sun et al., 2017), suggesting different responses in leaves of different ages.
Photoinhibition of PSII is controlled by altering the rates of damage and repair processes (Aro et al., 1993; Murata, Takahashi, Nishiyama, & Allakhverdiev, 2007; Nishiyama & Murata, 2014), and it is possible that HT impacts one or more of these processes. We thus compared the effects of lincomycin, an inhibitor of chloroplast protein translation that abolishes PSII repair (Tyystjärvi & Aro, 1996), on changes in maximal PSII quantum efficiency (FV /FM ), measured 20 min after exposure to light, to dissipate q E (Figure 2). Exposure to 1500 μmol m-2 s-1 at GT (GT+HL) over a period of six hours resulted in a loss ofFV /FM of about 50% in the presence of lincomycin, but only ~20% in the absence of the inhibitor, suggesting that the HL treatments induced substantial photodamage, which was partly reversed by repair. On the other hand, the loss of FV /FM in the presence of lincomycin was only ~25% when the temperature was ramped from 30 to 45 °C during exposure to HL (HT+HL), implying that HT decreased the rate of PSII photodamage. The fact thatFV /FM deceased only slightly under HT in the absence of lincomycin, suggests that the lower rate of PSII photodamage and repair were able to maintain nearly full PSII efficiency.
Overall, the results in Figure 2 imply that photoinhibition was lower at HT+HL, both because of decreased rates of PSII damage and relatively unhindered rates of PSII repair. This view is consistent with the increased q L and decreased pmf at HT (discussed above), which could ameliorate PSII photodamage caused by recombination reactions that would otherwise lead to1O2 production (Davis et al., 2016). An increase in PSII repair with temperature was proposed to contribute to the tolerance to HT observed in tobacco with enhanced synthesis of the osmoprotectant, glycine betaine (Yang et al., 2007).
Effects of high temperature on CO2 assimilation and photorespiration
Figure 3 shows temperature response curves for net CO2assimilation (A ) under conditions similar to those used in Figure 1. For simplicity, we present data on young leaves in the main text, but found similar results with mature leaves, as presented in Supplementary Figure 2. At LL (300 μmol m-2 s-1),A declined steadily as temperature was increased from 30 to 45 °C (Figure 3A), but recovered fully (even increasing slightly) after returning to growth temperature. Under 1000 or 1500 μmol m-2 s-1, A remained relatively constant from 30-40 °C, but decreased significantly (p<0.01) by ~20% from 40-45 °C, consistent with previous work on other species (Crafts-Brandner & Salvucci, 2002; Lu et al., 2017). Similar to what was found in Loblolly pine (Pinus taeda ) and eastern cottonwood (Populus deltoides ) (Urban, Ingwers, McGuire, & Teskey, 2017a, 2017b), temperature-dependent decreases in A were not related tog s, which increased with temperature (p<0.01) under all light intensities but recovered when returned to growth temperature (Figure 3B). Indeed,C i increased with temperature (Figure 3C and Supplementary Figure 2C), due to increased gs and decreased A . These results imply that the drop in A was likely related to decreased carboxylation or increased oxygenation. It is possible that rubisco was deactivated (Salvucci & Crafts-Brandner, 2004) and/or, as is well established, the specificity for CO2 over O2 declined with increasing temperature (Brooks & Farquhar, 1985; Galmés et al., 2016; Prins et al., 2016).
velocities of rubisco carboxylation (v c) and oxygenation (v o — used as a proxy for photorespiration) (Figures 4 A and B) from the gas exchange data shown in Figure 3. Results from experiments taken at 1500 μmol m-2 s-1 were omitted because they were very similar to those observed at 1000 μmol m-2s-1. At LL (300 μmol m-2s-1, Figure 4A), v c andv o remained relatively constant over temperature with v o /v c ratio of about 0.3. However, at HL (1000 μmol m-2s-1, Figure 4B), v o more temperature-dependent and increased by more than 36% at 45 °C compared to v c, which only increased by about 16%. These effects were independent of leaf age.
This increase in photorespiration over CO2 assimilation can account for some of the reduction in net CO2assimilation at HT. It is also possible that some LEF was diverted into “alternative” electron sinks (not related to assimilation or photorespiration) at HT. These sinks include the Mehler-peroxidase reaction (MPR, water-water cycle) (Asada, 2000; Miyake, 2010), the plastid terminal oxidase (PTOX) (Rumeau, Peltier, & Cournac, 2007), nitrogen (N) reduction, and lipid biosynthesis.
To assess this possibility, we compared the electron transport rate (ETR) measured by chlorophyll fluorescence with that calculated from gas exchange (ETRGE). The difference between the two parameters indicates flow of electrons that cannot be accounted for by carbon fixation and photorespiration. Figure 4C shows that under LL, in mature leaves, there was no significant (p>0.05) difference between ETR and ETRGE at 30 °C, but the difference increased significantly (p=0.03) at 45 °C, due to a decrease in ETRGE which likely stems from the decrease invc observed in Figure 4A. A decrease inv c can affect ETRGEGE throughout the entire temperature range, except during recovery when ETRGE in both leaves increased to the level of ETR. This suggests the persistence of alternative electron acceptors, whose activity likely reduced following recovery.
In HL however, upon increasing the temperature to 45 °C, there was no significant difference (p=0.9) between ETR and ETRGE in mature leaves, although in young leaves, differences were significant (Figure 4D). This strongly suggests that, at HT+HL and especially in mature leaves, the utilization of electrons by CO2 assimilation and photorespiration could account for most of the electrons coming from the light reactions. Nevertheless, in young leaves, there is evidence for the existence of a large pool of alternate electron acceptors which were active regardless of temperature. These results are generally consistent with thev H+/LEF data in Figure 1F–G, which shows elevated CEF in mature leaves but a decline in young leaves under HT+HL. Thus, the case in the mature leaves is generally consistent with an increase in processes with higher ATP demand, such as photorespiration. The low CEF in young leaves is consistent with more active alternative electron acceptors, although photorespiration was also increased as shown in Figure 4B.
The larger alternative electron flow in young leaves could be explained by more electron consuming processes. Such processes include nitrogen metabolism and lipid synthesis, electrons from the light reactions, although these were not directly assayed in this study. Differential nitrogen metabolism between young and mature leaves has been reported in other species and stresses, e.g. in rice (Wang, Wu, Han, Zheng, & Yang, 2012). There have also been reports that nitrate (NO3-) reduction can use up to 25% of electrons from LEF (Bloom, Caldwell, Finazzo, Warner, & Weissbart, 1989). Considering that the Hoagland’s nutrient solution used to fertilize the cowpea plants in our experiments provide nitrogen mostly as NO3-, this explanation may account for at least some of the alternative electron sinks.
, photons going to alternative electron acceptors also increased. These observations suggest that the combined activity of photorespiration and other alternate electron acceptors could account for a significant portion of electrons from LEF under HT, especially in HL, although, photorespiration was the dominant electron acceptor when considering the relative change in photon or electron utilization.
Figure 5 shows A /C i curves with concurrent chlorophyll fluorescence measurements for young leaves under both GT (30 °C) and HT (45 °C) at 300 and 1000 μmol photons m-2s-1 and under ambient (21%) and low (1%) O2 atmospheres. Similar results were obtained for mature leaves (Supplementary Figure 5). A summary of the fitted parameters based on Sharkey et al. (Sharkey et al., 2007; Sharkey, 2016) is shown in Supplementary Figure 4.
As described in detail in the Materials and methods, the measurements were performed sequentially, starting at ambient CO2(400 ppm) followed by lowering CO2 to 50 ppm through 200 ppm and followed by a gradual rise to high CO2 levels (1000 ppm). Under 300 μmol photons m-2s-1 and 30 °C (Figure 5A), A increased with CO2 over the entire range measured. In low oxygen,A appeared to become saturated with CO2. In Figure 5B, ETR/4 was mostly independent of CO2indicating the RuBP regeneration was the rate determining process. In 1 and 21% O2 ETR/4 was about the same except that in 1% O2 ETR/4 declined at low CO2, the behavior that indicates RuBP-saturated rubisco kinetics was rate determining. The initial measurement at 400 ppm (circled data) fell on the line observed during the CO2 response curve measurement.
Under 300 μmol photons m-2 s-1 and 45 °C, the data were similar (Figure 5 C and D) except that the decline in ETR/4 at low CO2 was more pronounced, indicating the RuBP-saturated rubisco kinetics were more limiting than at 30 °C. Initial measurements (circled) were the same as measurements made during the response curve analysis. Because the maximum ETR/4 was the same at 1 and 21% O2 at both temperatures there was no indication of triose phosphate use limitation, as would be expected at this moderate light intensity.
The responses were different at 1000 μmol photons m-2s-1. In 21% O2 and 30 °C, the initial response of A to Ci was much steeper and the CO2 dependence of ETR/4 indicated a significant control by rubisco up to ~200 ppm (compare Figure 5 B with 5 F). In Figure 5E, A at 1 and 21% O2 reach the same rate and then are independent of CO2. This has been labeled the triose phosphate use limitation (Sharkey 1985a), which can often be observed at saturating light and slightly higherC i than the leaves likely experienced during growth. Because of photorespiration, when A in 1 and 21% O2 is the same at the same Ci , ETR/4 must be higher at 21% O2 as was observed (Figure 5F). Because A was the same at 1 and 21% O2 and no reverse sensitivity was observed there is no evidence for photorespiration increasing A by providing amino acids to the plant as is sometimes observed (Busch et al. 2018). Initial measurements (circled) fell on the same line as determined during the measurements from low to high CO2 (Figure 5 E and F). The behaviors reported in Figure 5 A through F show the expected relationships amongA , ETR/4, light, and Ci (Sharkey 1985b)
In 1000 μmol photons m-2 s-1 and 45 °C, novel behavior was observed. Especially at highCi , A was approximately the same in 1 and 21% O2 (Figure 5G). This would require higher ETR/4 in 21% O2, which was observed (Figure 5H). However, the O2 insensitivity of A was not correlated with aCi insensitivity (Figure 5G). This phenomenon was seen in both young and mature leaves (supplemental Figure 6G) and in both 1000 and 1500 μmol photons m-2s-1 (supplemental Figure 7 C and G). This behavior is not predicted by current gas exchange theory (Sharkey 2016). Also observed only in saturating light and high temperature is a decline inA and ETR/4 between the first measurement (circled) and the sameCi measured during the response curve measurements. We suspect this was a change in metabolic state with time but could also have been induced by the low Ciexperienced during measurements made to construct the curve.
The strong response of both A and ETR/4 toCi in 1000 μmol photons m-2s-1 and 45 °C indicates that rubisco activity was determining the behaviors. Rubisco activation is known to decline at high temperature (Salvucci and Crafts-Bradner 2004). The unusual behavior could reflect a variable rubisco activation state at saturating light and high temperature. Study of this unusual behavior including determining how widespread it is, the activation states of rubisco, metabolomics, and more detailed measurements of electron and proton flows may provide insight into how cowpea tolerates high temperature.
Longer-term effects of elevated temperature and fluctuating light on photosynthesis
We have shown that short term HT combined with HL caused a shift toward photochemistry with the increased electron flux attributed to decreased PSII photodamage, enhanced PSII repair and increased photorespiration as well as activity of other alternative electron acceptors. However, it was not clear whether the effects on PSII function were only for a short period or whether it persisted over a relatively longer time scale and under dynamic conditions as occurs in the field. We, therefore, tested the hypothesis that prolonged exposure to HT+HL especially under dynamic conditions could lead to more severe loss in efficiency of PSII for photochemistry (ɸ II) under the combined stress relative to HL stress alone. Dynamic (fluctuating) light conditions have been shown to produce phenotypes not detected under constant light conditions normally used in growth chambers (Cruz et al., 2016; J. Li et al., 2019). Figure 6 shows results from a four-day experiment, with altered temperature and light intensity patterns, as indicated above Figure 6A, to expose the plants to a range of light and temperature conditions. On Day 1, the light intensity was fluctuated between 300 and 1500 µmol photons m-2 s-1 in two steps. On Days 2 and 3, the light was held constant at 1500 and 300 µmol photons m-2 s-1 respectively, and on the final day the fluctuating pattern used on Day 1 was repeated. Plants were also exposed to either a constant GT (30 °C), a constant HT (45 °C), or a daily fluctuation in temperature between GT in the morning, HT at midday and returning to GT by end of the day (HTR — HT with recovery). The night temperature was kept constant (20 °C) for all treatments, except the constant HT, where it was kept at 35 °C.
Both lines showed decreased ɸ II when exposed to fluctuating and HL under constant GT, but was particularly evident in the HL-sensitive line (58-77) on days 2 through 4 (Figure 6A). The decreased ɸ II was accompanied by increases in long-lived NPQ (q I(T)), suggesting that the regiment of light treatments induced substantial photoinhibition or photodamage. By contrast, both regimes with elevated temperatures resulted in smaller losses of PSII activity on Day 4 (Figure 6A), when compared to the initial values at the beginning of Day 1, suggesting that the HT prevented the buildup of photoinhibition. Supplementary Figure 7 shows the log2 fold change ofɸ II which indicates that both HT treatments resulted in relatively higher PSII activity compared GT, especially in HL.
When the temperature was first increased in Day 1, there was an abrupt spike in NPQ(T) (Figure 6B); the rapid recovery of this spike suggests that it is caused by q E(T) (Figure 6D). This effect was observed in both lines, but was larger for Yacine, especially under HL, and was regardless of whether the light was constant or fluctuating. However, with time, NPQ declined, likely due to the decreasing q E(T), and reached values similar to the GT treatment at the end of the day. By the 4th day, NPQ(T) and photoprotection (q E(T)) had declined in Yacine below values of the GT plants. Since ɸ II increased with the decline in NPQ as was seen in Figure 1, it suggests that PSII photodamage was not increased and/or that PSII repair was able to maintain high activity, in line with Figure 2.
By contrast, q I(T) was persistently low in the HT treatments compared to the GT plants under HL regardless of the light intensity (Figure 6C). Thus, it is possible that the increasedq I(T) in GT resulted in impaired PSII activity as indicated by the low ɸ II in HL. These results suggest that HT prevented loss of PSII activity under HL by reducing the slowly relaxing form of NPQ. It should be noted that compared to values at the beginning of Day 1, the HT treatment slightly increasedq I(T). However, when compared to GT at the same time points, q I(T) was lower in both HT treatments. This reduced q I(IT) is consistent with enhanced activity of alternative electron acceptors, and particularly photorespiration.
The data presented imply that, at least in these cowpea lines and in particular younger leaves, the light reactions of photosynthesis were relatively unimpaired under combined HT+ HL as has been reported in tomatoes (Gerganova et al., 2016). Further, these effects apparently persist over a relatively long term and under dynamic light and temperature conditions. Using two genotypes with differing sensitivity to HL revealed that the effects of combined HT and HL were common regardless. In fact, the HL-sensitive genotype, 58-77, appeared to show more tolerance to HL when combined with HT. These observations allude to the fact that cowpea plants are better equipped to deal with the combination of HL and HT stresses since these often occur together in the field. However, it is not known whether the exhibited tolerance of the light reactions would lead to increased productivity since carbon metabolism was inhibited under these conditions.