Photosynthetic thermal acclimation is indicated by the shift of net photosynthetic assimilation rate (A_n) temperature-response curve and the shift of temperature at which maximum A_n was achieved (Tmax). When tobacco wild type (WT) and the ndhB mutant (ΔB, deficient in NDH-B subunit) were grown at normal temperature (25 °C day/20 °C night) and measured with an increasing temperature gradient of 20 °C to 45 °C, the A_n of both WT and ΔB showed similar temperature-response curve peaked between 30 °C and 35 °C (T_max between 30 °C and 35 °C). After the tobacco plants were placed at 35 °C day/30 °C night for 1 day (including night period), the temperature-response curve of WT shifted toward a higher temperature range, with the T_max increased to above 35 °C, and the A_n decreasing phase became much slower ( Figures 1A, C ). In ΔB mutant, the photosynthetic performance was suppressed severely, and without a shift of T_max at 1 day after heat-treatment (DAH) ( Figures 1B, D ). The result indicated that compared with the photosynthetic inhibition in the ndh mutant at 1DAH, WT plants were rapidly acclimated during the first day in moderate hot environment, and the photosynthesis performance was kept high at above 35°C ( Figures 1A, C ). The shift of temperature-response curve and T_max occurred 2 days after heat treatment in ΔB mutant, while the shifted curve and T_max were retained in WT ( Figures 1G ), indicating both materials achieved heat acclimated status after 2 days treatment.

At excessive hot temperature of 40°C, the photosynthesis rate of WT was enhanced at 1DAH and 2DAH ( Figure 1E ). The relative net photosynthesis rates of WT (normalised to maximum A_n) measured at 40°C 1 day after heat-treatment and 2 DAH, were increased from 70-80% to above 90% of the maximum A_n ( Figures 1F ); that is, WT tobacco had enhanced its photosynthesis performance at the high-temperature regime within the 1st day of temperature shift. However, the A_n of ΔB mutant measured at 40°C was suppressed at 1DAH and did not recover until 2 DAH, which indicates that the acclimation process in ΔB mutant was delayed ( Figures 1E, F ). The experiment using another ndh mutant, ΔCJK, presented a similar trend as ΔB compared to WT ( Supplementary Figure 1 ). Considering the potential impacts of night period, photosynthesis temperature-response curves were also measured on WT and ΔB plants that have been heat-treated under 24 hours full-light. Unlike the rapid shift of WT under 35 °C day/30 °C night within 1 day, the photosynthetic acclimation of WT plants under full-light treatment was slow and indistinct, and was not significantly different from ΔB mutant ( Figure 2 ).

To track down the cause and effect of this delayed photosynthetic thermal acclimation, more photosynthetic parameters were gathered at the treatment temperature of 35 °C. Light-saturated A_n in WT tobacco was increased after moderate heat treatment; however, the light-saturated A_n of ΔB was suppressed at 1 DAH, and staged a recovery at 2 DAH ( Figure 3A ). Differently, a decrease of the quantum yield in PSII [Y(II)] was observed earlier at 2 hours after heat treatment (HAH) in ΔB compared to WT, although it was recovered and slightly increased in both WT and ΔB at 1 DAH and 2 DAH ( Figure 3B ). The different trends between the two photosynthetic parameters in WT and ΔB mutant suggested asynchronous responses in light reactions and CBB cycle at 1 DAH. Other chlorophyll fluorescence parameters such as the maximum quantum efficiency of PSII (F_v/F_m) and Non-photochemical quenching (NPQ) were all similarly suppressed at the initial period of heating (2 HAH), and recovered at 1 DAH and 2 DAH, in both WT and ΔB ( Figures 3C, D ).

Post-illumination rise (PIR) in Chl fluorescence is associated with the reduction of PQ, likely caused by the electrons accumulated in the stroma or cytosol during illumination transferred back (Asada et al., 1993; Mi et al., 1995; Shikanai et al., 1998), and therefore be used to evaluate the activity of cyclic electron flow (Deng et al., 2003). The PIR in WT became more apparent under the heat treatment, while the ΔB mutant did not show substantial PIR ( Figure 3E ) despite the heat treatment. The enhanced CEF in WT appeared synchronous to the thermal acclimation, suggesting the absence of the NDH-mediated CEF could be associated with the delay of acclimation to a higher temperature.

The maximum rate of Rubisco carboxylase activity (V_{c max}), maximum rate of light-saturated photosynthetic electron transport (J _max), and dark respiration rate were calculated from A-Ci curves (Sharkey et al., 2007) ( Table 1 ; Supplementary Figure 2 ). The dark respiration rate of WT increased after heat treatment, but that of ΔB was not changed. V_{c max} at 25°C (V_{c max25}) in WT slightly increased at 1 DAH, while V_{c max25} was not changed in ΔB after heating. Notably, the increased difference of V_{c max} at 35°C (V_{c max35}) between WT and ΔB further indicated an inhibited Rubisco carboxylase activity in ΔB relative to WT. On the other hand, J _max at 35°C (J _max35) were slightly higher in WT than in ΔB after heat treatment.

To assess the gene expression changes of WT and ΔB tobacco leaves in response to heat treatment, an RNAseq analysis was undertaken at different time points, in comparison with the expression levels before treatment. At the total transcriptome level, principal component analysis (PCA) showed clear gene expression variation following the time-points of heat treatment ( Supplementary Figure 3A ). Although the PCA indicates that the expression variation was more affected among timepoint rather than between WT and ΔB, the total expression heatmap showed that the expression patterns of ΔB mutant at each time-point were mostly distinct from those of WT ( Figure 4A ). At 2 HAH, there were 4644 differentially expressed genes (DEGs) in WT, including 2424 up-regulated and 2220 down-regulated genes; and there were 5409 DEGs in ΔB mutant, which included 2728 up-regulated and 2681 down-regulated genes. At 6 HAH and 12 HAH, there were more DEGs in both WT (12181 and 13031 DEGs) and ΔB (10821 DEGs and 11377 DEGs) ( Supplementary Figure 3B ). Considering the appearance of significant physiological inhibition at early stage of 2 HAH ( Figure 3C ), Gene ontology (GO) term enrichment analysis was undertaken, which revealed that up-regulated DEGs in WT at 2 HAH were enriched in multiple photosynthesis-related pathways, such as “light-harvesting”, “chloroplast organization”, “Photosystems” and “chloroplast thylakoid membrane” ( Supplementary Figure 4A ). The up-regulated DEGs at 2 HAH in ΔB were enriched in pathways that respond to stress, such as “response to heat”, “response to temperature stimulus”, “response to reactive oxygen species” and “response to abiotic stimulus”, instead of photosynthetic pathways ( Supplementary Figure 4C ). The hierarchal clustering of photosynthesis-related DEGs indicated that around two-thirds were up-regulated in WT at 2 HAH, 6 HAH and 24 HAH (1 DAH), whereas many of these genes were down-regulated in ΔB ( Figure 4B ). Further classification of photosynthetic genes ( Supplementary Figure 5 ) showed that DEGs encoding PSII, PSI, and their light-harvesting proteins (Lhcbs and Lhcas) were most distinct between WT and ΔB.

Around two-thirds of nuclear-encoded NDH subunit genes were up-regulated after heat treatment. Some of the up-regulations were temporary, that reached their peaks around 12 HAH, and subsequently returned toward previous levels ( Figure 4C , Supplementary Figure 5 ). The temporary up-regulation was less observed in ΔB mutant. Another group of DEGs distinct between WT and ΔB were Rubisco-related genes (such as Rubisco small subunit gene and gene encoding Rubisco large subunit-binding protein), but not for Rubisco activase ( Supplementary Figure 5 ). Apart from photosynthetic genes, RNA-seq also indicated the up-regulation of alternative oxidase (AOX) participating in respiration, plastoquinol terminal oxidase (PTOX), and some thermos-responsive transcription factors, such as class A1 heat shock factor (HsfA1) and dehydration-responsive element-binding protein 2A/2C (DREB2A/DREB2C). The up-regulation of these genes lasted longer in ΔB than in WT. The expression of genes encoding plastidial NAD-dependent malate dehydrogenase (plNAD-MDH) and mitochondrial malate dehydrogenase (mMDH) was also found increased in both ΔB and WT ( Figure 5 ).

At the protein level, western blots indicated that the changes of photosystem proteins were different between WT and ΔB during the first 2 days of thermal acclimation. PsbA (D1) content of PSII tended to be increased in WT, but was transiently declined at 2 HAH in ΔB. PsaD of PSI was temporarily suppressed in WT, while it was in a downward trend in ΔB. Lhcb and Lhca were enhanced over the first 6 or 12 hours in WT but suppressed after 12 hours in ΔB ( Supplementary Figure 6B ). The protein abundances of NDH subunits in WT tobacco leaves showed temporary increases that correspond to gene expression pattern; ndhT, ndhS and ndhH from two different sub-complexes showed increased contents at the period from 2 HAH to 1 DAH, but returned to normal level at 2 DAH. Similar trends were found from other proteins contributing to CEF (PGR5/PGRL1) ( Supplementary Figures 6A, D ). In addition, a Blue-Native PAGE was undertaken, showing that the content of NDH-PSI supercomplex was increased at 1 DAH and 2 DAH compared to 0 DAH in WT, while it was absent in ΔB ( Supplementary Figure 6C ).

In order to further dissect the mechanism behind the delayed photosynthetic thermal acclimation in ΔB mutant, we focused on the first day-night-day shift (including time points shortly before night and shortly before dawn) and performed a set of metabolism analysis. The two products of light reactions, NADPH and ATP, were used to drive CBB cycle. The NADPH amount of WT and ΔB increased at 12 HAH (before night) and reduced at 18 HAH (before dawn); its dehydrogenated product NADP⁺ had the opposite trend. However, ΔB accumulated a larger amount of NADPH than WT, and was not able to consume NADPH sufficiently during the night time ( Figure 6A ); that is, the mutant tended to accumulate more reducing power at the end of the diurnal cycle under increased temperature. ATP content decreased after heating, but recovered to higher level in WT at 1 DAH, while stayed at lower level in ΔB ( Figure 6A ). Contrary to NADPH, NADH was accumulated at 18 HAH, and its amount was similar in WT and ΔB within the first 24 hours under heating. To evaluate the effect of reducing power accumulation, the amount of hydrogen peroxide (H₂O₂) was analysed. As expected, H₂O₂ content increased at 12 HAH, and remained relatively higher level in ΔB than in WT ( Figure 6C ).

Apart from light reactions, majority of metabolites participating in the CBB cycle were markedly decreased upon temperature increase in both WT and ΔB. Ribulose bisphosphate (RuBP), the initial substrate of CBB cycle, decreased at 12 HAH and 18 HAH, and re-accumulated at 1 DAH but with significantly lower level in ΔB than in WT. The subsequent metabolites in CBB cycle, 3-phosphoglyceric acid (PGA), glyceraldehyde 3-phosphate/dihydroxyacetone phosphate (GAP/DHAP), and fructose-1,6 bisphosphatase (FBP), exhibited similar trends as RuBP; GAP, as the prime end-product of photosynthesis, was with lower amount in ΔB than in WT at 1 DAH ( Figure 6B ). At the end of the glycolysis process, phosphoenolpyruvate (PEP) exhibited similar decrease and increase pattern as CBB cycle metabolites, while the subsequent metabolites involved in respiration process, pyruvate (Pyr), citric acid (CIT), isocitric acid (ICIT), fumarate (Fum) and malate (Mal), did not show markable differences between WT and ΔB mutant. Unlike respiration, most of the metabolites participating in photorespiration were induced at 12 HAH and decreased overnight; and the amount of glyceric acid (Glyce) showed remarkably higher level in ΔB than in WT from 12HAH ( Figure 6B ).

Wild type tobacco (Nicotiana tabacum cv Xanthi), ΔB mutants in which the chloroplastic NDHB gene was inactivated (Shikanai et al., 1998) and ΔCJK mutant in which the chloroplastic NDHC, NDHJ and NDHK genes were inactivated (Takabayashi et al., 2002) were cultivated in a phytotron (about 500–600 µmol photons m^-2 s^-1, 16 hours light at 25 °C and 8 hours dark at 20 °C, 65% relative humidity). Six weeks old tobacco plants were transferred into growth cabinets with independent temperature control for heat treatment, in which temperature was set to 35 °C day and 30 °C night (with a control cabinet set to 25 °C day and 20 °C night). For every batch of tobacco plants used for experiment, at least 18 pots each for WT and ndh mutants were planted, which ensures that there are 3-4 replicates for measurement or sampling at each time point. For heat treatment under continuous light, temperature was set to 35 °C all day.

Gas exchange was measured on mature tobacco leaves (third leaf from the top) prior to and after heat treatment. LI-6800 instrument equipped with 2 cm² cuvettes and a 6800-01A multiphase flash fluorometer (a combination light source and chamber for gas exchange measurement) (Li-COR, Lincoln, NE, USA) was used to measure the net photosynthetic CO₂ assimilation rates (A_n). A_n was measured with the following settings: 600 (for temperature-response curve) or 1000 (for light-saturated An) µmol photons m^-2 s^-1, 400 μmol mol^-1 CO₂, flow rate of 500 μmol s^-1, and relative humidity of 65%. All photosynthesis measurements were taken at least 2 hours after the growth cabinet lights were turned on.

During the measurements of the temperature-response curves, the measuring chamber temperature was heated at 1 °C per minute from 20 °C to 45 °C with the LI-6800 auto-control program, and An was recorded at 30-s intervals. The x-axis of temperature-response curve was leaf temperature, and the y-axis value corresponding to the vertex were taken as the maximum A_n.

The A-Ci curve measurements were performed using the auto program settings of LI-6400 instrument with the following sequence of CO₂ concentrations: 400, 200, 50, 100, 150, 300, 400, 500, 600, 800, 1000, 1200 μmol mol^-1 CO₂ ( Supplementary Figure 3 ). CO₂ levels were changed at 1-2 min intervals. A-Ci analyses were performed and related photosynthetic parameters were calculated according to Sharkey et al. (2007).

Chlorophyll fluorescence parameters were measured using a portable modulated chlorophyll fluorometer, PAM-2000 (Walz, Effeltrich, Germany). Slow kinetics curve was measured on dark-adapted tobacco plants (pre-dark-treatment for at least 2 hours), started with a light-saturated pulse, application of actinic light (970 µmol m^-2 s^-1) after 40 second interval, and then light-saturated pulses every 20 seconds until fluorescence reached steady state. Chlorophyll fluorescence parameters were calculated as previous studies described (Bjorkman and Demmig, 1987; Krall and Edwards, 1992; Schultz, 1996; Kramer et al., 2004). The parameters included maximum photochemical efficiency (F_v/F_m), quantum yield of Photosystem II (φ_II), and non-photochemical quenching (NPQ).

Post-illumination rise (PIR) of chlorophyll fluorescence was measured as previous studies described (Asada et al., 1993; Mi et al., 1995; Shikanai et al., 1998). PIR is the transient increase of chlorophyll fluorescence after actinic light turned off indicating re-reduction of the quinone pool in the dark, which is associated with the activity of NDH and correlates with the changes in NDH mediated CEF. The measurement with PAM-2000 was performed with an actinic light of 600 µmol m^-2 s^-1, illuminated until fluorescence reached steady-state, and the fluorescence curve was recorded for further 2 minutes after shut down of the actinic light.

Mature leaves of tobacco (third leaf from the top) were harvested prior to and after the temperature treatment, immediately frozen in liquid N₂ and stored at -80 °C. RNA samples were extracted from the leaf tissue using TRIzol® Reagent (Invitrogen, Massachusetts, USA) and genomic DNA was removed using DNase I (TaKara, Shiga, Japan). The RNA quality was determined by 2100 Bioanalyser (Agilent Technologies, CA, USA) and quantified using ND-2000 (NanoDrop Technologies, Delaware, USA).

Transcriptome library for RNAseq was prepared from 1 μg of total RNA for each sample, using TruSeqTM RNA sample preparation Kit from Illumina® (San Diego, CA, USA). After sequencing, the raw paired-end reads were trimmed and quality controlled by SeqPrep (https://github.com/jstjohn/SeqPrep) and Sickle (https://github.com/najoshi/sickle) with default parameters. The clean reads were aligned to Nicotiana tabacum reference genome (GCF_000715135.1, https://www.ncbi.nlm.nih.gov/genome/425?genome_assembly_id=274804) using HISAT2 (http://ccb.jhu.edu/software/hisat2/index.shtml) software (Kim et al., 2015), and reads were mapped assembled by StringTie (https://ccb.jhu.edu/software/stringtie/index.shtml?t=example) in a reference-based approach (Pertea et al., 2015). DEGs were identified according to the expression level of each transcript based on transcripts per million reads (TPM). Differential expression analysis was performed by DESeq2 (Love et al., 2014)/DEGseq (Wang et al., 2010)/EdgeR (Robinson et al., 2010); and if |log2FC| > 1 and Q value ≤ 0.05 (DESeq2 or EdgeR) or Q value ≤ 0.001 (DEGseq), the gene was identified as a DEG. Furthermore, functional-enrichment analysis including GO and KEGG were undertaken to identify which biological and metabolic pathways that DEGs were significantly enriched in. The statistical tests used were Fisher’s test (significance level of P < 0.05) and Benjamini-Yekutieli procedure (FDR < 0.05).

Total leaf protein was extracted using a protein extraction buffer containing 50 mM Tris-HCl (pH 8.0), 0.25 M sucrose, 2 mM dithiothreitol (DTT), 2 mM EDTA, and 1 mM phenylmethylsulfonyl fluoride (PMSF). Protein content was evaluated by western blot, where the extracted proteins (>20 μg for each sample loading) were separated by SDS-PAGE (12.5% SDS-PAGE gels) and transferred onto a PVDF membrane. After blocking the membrane with 5% w/v skim-milk in TBST (150 mM NaCl, 20 mM Tris, 0.1% Tween, pH 7.6) for 1 hour, it was incubated in a primary antibody solution (1:5000 primary antibody, 5% w/v skim-milk powder in TBST) for 1 hour. The membrane was then washed with TBST for 1 hour and incubated with a secondary antibody (1% w/v skim-milk powder in TBST) for an additional 1 hour. The membrane was finally visualized using ECL Plus reagent (Epizyme, SQ201). The specific antibodies used for detecting photosynthetic proteins were obtained from Orizymes Biotechnologies Company (Shanghai), including NDH subunits, PGR5/PGRL1, PsaD, PsbA, Lhca1, Lhcb1 and Actin.

BN-PAGE gel was prepared according to Jarvi et al. (2011). Tobacco leaves were sampled and homogenized in STN medium at 4 °C (0.4 M sucrose, 50 mM Tris-HCl pH 7.6, 10 mM NaCl). After low-speed (200 g for 3 min) and high-speed (5000 g for 10 min) centrifugation at 4°C, the chloroplasts collected were ruptured by TN medium (50 mM Tris–HCl pH 7.6, 10 mM NaCl). The thylakoid membranes were collected by centrifugation at 8000 g for 5 min at 4°C, and suspended in solubilization buffer [25 mM BisTris-HCl, pH 7.0, 10 mM MgCl2, 20% (v/v) glycerol]. Thylakoid membranes containing 0.5 mg ml⁻¹ chlorophyll were solubilized with 2% (w/v) n-dodecyl-β-maltoside (DDM) by gentle agitation on ice for 1 hour. The samples were separated by Native-PAGE undertaken at 4 °C by increasing the voltage gradually from 50V to 200 V during the 5.5 h run. After running, the gel was stained with a coomassie blue solution (0.1% Coomassie Blue R250 in 10% aceticacid, 40% methanol and 50% H₂O) for 1 hour and de-stained with a de-staining solution (10% acetic acid, 40% methanol and 50% H2O) for 12 hr. Band identifications are taken based on previous study (Jarvi et al., 2011).

Tobacco leaf discs (1-cm-diameter) were collected prior to and after heat treatment. The treatment was started at 9 am, and subsequent sampling was performed at 9 pm (12 hours, 2 hours before dark), 3 am (18 hours, 4 hours after dark), and 9 am (1DAH). At each time point 3 tissue replicates from WT and ΔB were sampled, immediately frozen in liquid nitrogen, and ground to powder with steel balls at later stage. To extract metabolites, 4°C pre-cooled methanol: chloroform (7:3, v/v) was added to the samples. The samples were then incubated at -20°C for 3-4 hours, and 560 μL cold deionized water was added to each sample. After centrifugation at 2200 g (4°C), the supernatant was transferred to new tubes. A 50% (v/v) methanol solution was mixed with the precipitation, centrifuged again, and the supernatant was transferred to the previous tubes. Extracts were collected and stored at -80°C after being filtered by an organic phase filter. For sample loading, the metabolite extract was injected into a QTRAP 6500+ system (Sciex, Danaher Corporation, USA). Separation was conducted using a Luna® NH₂ LC column (3 µm, 100 × 2 mm, Phenomenex, California, USA), with gradients from 20% (v/v) acetonitrile solution to solvent A (20mM ammonium acetate in 5% acetonitrile solution, pH 9.5), and to solvent B (acetonitrile). The elution started with different solvent percentages for different durations and ended with 85% solvent B for 3 minutes (0~1 min, 15% A and 85% B; 1~8 min, 70% A and 30% B; 8~22 min, 95% A and 5% B; 22~25 min, 15% A and 85% B). The column eluent was connected to the mass spectrometer fitted with an electrospray. The detection mode was set to negative ion mode, and the scan mode was set to multi reaction monitoring (MRM). For each sample, molecular features were defined based on retention time and mass (m/z) with Analyst® software 1.6.3 (Sciex, Danaher Corporation, USA). The carbon-metabolism metabolites as well as cofactors NAD(P)H、NAD(P)⁺、ADP and ATP were determined according to the chromatographic peaks calculated by the software, and the relative amounts of metabolites among different time-points were calculated by the chromatographic peak areas.