pgr5-1 was obtained by ethyl methyl sulfone mutagenesis, which likely introduces other mutations into its genome (Munekage et al., 2002). Mutant PGR5 protein (PGR5_G130_S) accumulates in pgr5-1 pgrl2-1 plants (Ruhle et al., 2021). pgr5-2 mutant has a serine-to-phenylalanine alteration in the middle of mature proteins and accumulates higher levels of PGR5 and PGRL1 proteins than pgr5-1 mutant (Nakano et al., 2019). To completely knock out PGR5 (AT2G05620) gene, we constructed a pgr5 mutant (pgr5-3, Col-0 background) using the CRISPR–Cas9 system to insert an adenine in the 243rd base pair downstream of the initiation codon in Arabidopsis ( Figure S1A ). We also used the CRISPR–Cas9 system to knock out PGRL1A (AT4G22890) (Col-0 background) and PGRL1B (AT4G11960) genes (crr2-2 background and Col-0 background). PGRL1A and PGRL1B are two highly homologous genes of the Arabidopsis thaliana. pgrl1a-2 has an adenine insertion in the first exon. pgrl1a-3 has a thymine insertion in the first exon ( Figure S1B ). pgrl1b-2, pgrl1b-3, pgrl1b-4, and pgrl1b-5 have adenine, cytosine, thymine, and guanine insertions in the first exon, respectively ( Figure S1C ). Insertional mutations in these genes result in premature termination of translation. To confirm the mutation effect, we checked the protein levels of PGR5 and PGRL1 in the mutants. PGR5 protein was hardly detected, and PGRL1 levels significantly decreased to roughly half of wild type (WT) in pgr5-1 and pgr5-3 ( Figure S1D ). This indicates that PGR5 gene is completely knocked out in pgr5-3 mutant and confirms that the deletion of PGRL1 affects PGR5 protein accumulation (DalCorso et al., 2008). There was no difference in growth phenotype among the mutants and WT under optimal growth conditions ( Figure S1E ). Complementation experiments showed that introducing corresponding genes in the background of pgr5, pgrl1a, and pgrl1b could restore CET activity ( Figure S2 ).

To investigate the functional link of different CET pathways, we first generated mutants that are deficient in the two partially redundant pathways of CET. pgr5 crr2 double mutant was obtained by crossing pgr5-3 with the crr2-2 mutant, and pgrl1a1b crr2 triple mutant was obtained by crossing pgrl1a-3 with pgrl1b-5 crr2-2. crr2-2 mutant is impaired in the NDH pathway of CET due to defective in a pentatricopeptide repeat (PPR) protein necessary for the expression of the plastid ndhB gene (Hashimoto et al., 2003). We then compared growth phenotypes of mutants with key CET genes knocked out under different light conditions. There was no evident difference in the growth phenotype among single mutants crr2 and pgr5, double mutants pgrl1a1b and pgr5 crr2, triple mutant pgrl1a1b crr2, or WT on Murashige and Skoog (MS) medium supplemented with 2% sucrose and 0.8% agar in a growth chamber with a 16/8-h light/dark cycle and a photosynthetic photon flux density of 60 μmol m⁻² s⁻¹ at 22°C for 10 days ( Figure 1A ). However, after the seedlings were transferred to soil with higher light conditions (120 μmol m⁻² s⁻¹) for 2 weeks, pgr5 crr2 mutant displayed small and yellowish leaves ( Figure 1B ), while the growth phenotype of triple mutant pgrl1a1b crr2 behaved like WT, even though the protein level of PGR5 and NDH was similar in both pgr5 crr2 and pgrl1a1b crr2 ( Figure S3 ). PGR5 was not detected in pgr5, pgrl1a1b, pgrl1a1b crr2, and pgr5 crr2, and loss of PGRL1 also blocked PGR5 accumulation. However, NdhJ levels were dramatically lower in crr2, pgr5 crr2, and pgrl1a1b crr2, confirming the deletion of crr2 in these mutants. Given that PGR5 plays a crucial role in fluctuating light (Suorsa et al., 2012), we also analyzed the growth phenotype of the above mutants under milder fluctuating light conditions (cycles of 5 min low light at 60 μmol photons m⁻² s⁻¹ and 1 min high light at 600 μmol photons m⁻²s⁻¹) after 13 days of growth under constant light. There was no significant difference in the growth phenotype before fluctuating light treatment in all mutants and WT ( Figure 1C , top). However, all the mutants showed visible photodamaged phenotype 6 days after treatment with the fluctuating light: slightly in crr2, obviously in pgr5 and pgrl1a1b, severely in pgrl1a1b crr2, and most severely in pgr5 crr2 ( Figure 1C , middle). This indicates a higher ability to tolerate fluctuating light in pgrl1a1b crr2 than in pgr5 crr2. When the fluctuating light treatment reached 14 days, the photodamaged phenotype of pgr5 and pgrl1a1b was further exacerbated, while lethal phenotype appeared in both pgrl1a1b crr2 and pgr5 crr2 ( Figure 1C , bottom). The results indicate that in addition to the PGR5 pathway, the NDH pathway helps protect against fluctuating light stress. We also analyzed the chloroplast ultrastructure of pgr5 crr2 and pgrl1a1b crr2 mutants and WT in 4-week-old leaves grown under 120 μmol m⁻² s⁻¹ light conditions. As shown in Figure S4 , the WT and pgrl1a1b crr2 chloroplasts displayed a well-developed membrane system consisting of interconnected stroma lamella and grana lamella, while those of pgr5 crr2 showed abnormal chloroplast structures, containing few stacking thylakoids and no starch grain ( Figure S4 ). The abnormal chloroplasts indicate that PGR5 is essential for photoprotection during chloroplast development. The above findings indicate that pgrl1a1b crr2 mutant possesses a higher tolerant ability than pgr5 crr2 mutant under different light conditions.

To understand the phenotypic differences between pgrl1a1b crr2 and pgr5 crr2, we first compared the CET activity among the seedlings of CET mutants and WT grown under low light (60 μmol m⁻² s⁻¹). As shown in Figure 2A , both pgr5 crr2 and pgrl1a1b crr2 lost a transient increase in chlorophyll fluorescence after the termination of actinic light illumination as crr2, as did other crr background mutants, which is similar to the previous results (Munekaga et al., 2004). We further compared the initial dark reduction rate of P700⁺ under far-red light, which reflects CET activity (Mi et al., 1992). The initial re-reduction rate of P700⁺ decreased by approximately 34% in pgr5 crr2, but there was no significant difference in pgrl1a1b crr2, compared with WT ( Figure 2B ). These results indicate that total CET activity is higher in pgrl1a1b crr2 than in pgr5 crr2. We also checked the electron donation activity of electrons from NADPH to PQ via Fd reflected in CET activity, which increased chlorophyll fluorescence by adding NADPH and Fd (Endo et al., 1997). CET activity remarkably decreased in both pgrl1a1b crr2 and pgr5 crr2 ( Figure 2C ).

We next compared chlorophyll fluorescence parameters to further examine the photosynthetic performance of the seedlings of CET mutants and WT grown under low light conditions. The light response of quantum yield of the PSII photochemistry [Yield (II)] decreased as light intensity increased; it declined the same faster in both pgr5 crr2 and pgrl1a1b crr2, compared with slower declines observed in WT ( Figure 2D ). NPQ is a chlorophyll fluorescence parameter that reflects the level of thermal dissipation. Increases in the light response of NPQ were detected in WT but were significantly suppressed to a similar degree in pgr5 crr2 and pgrl1a1b crr2 ( Figure 2E ), which is consistent with previous observations (Munekage et al., 2002). These results indicate that the light reaction of photosynthesis was suppressed in both mutants under unstressed conditions. The redox state of the PQ pool was also compared among the CET mutants and WT by measuring the chlorophyll parameter of 1 − qL, which reflects the reduction state of Q_A in PQ pools (Kramer et al., 2004). The result shows that the light-dependent reduction of Q_A increased to a similar degree in both pgr5 crr2 and pgrl1a1b crr2 grown under low light conditions ( Figure 2E ), indicating that inactivation of CET causes over-reduction of inter-photosystem electron carriers, such as PQ pool and Cytb₆f. To investigate the role of the higher CET activity in pgrl1a1b crr2, we compared the photosynthetic performance among the mature plants of mutants (pgr5 crr2 and pgrl1a1b crr2) and WT grown under high light conditions. The results show that the decrease in Yield (II) was faster in pgr5 crr2 (to zero at 80 μmol m⁻² s⁻¹) than in pgrl1a1b crr2 (to zero at 500 μmol m⁻² s⁻¹) ( Figure 3A ), indicating that pgrl1a1b crr2 possesses a higher quantum yield of PSII than pgr5 crr2 under high light conditions. However, NPQ was suppressed to a similar level in response to low light in pgr5 crr2 and pgrl1a1b crr2 and slightly lower in pgr5 crr2 than in pgrl1a1b crr2 in response to high light ( Figure 3B ). The reduction state of Q_A was slightly higher in pgr5 crr2 than in pgrl1a1b crr2, but the difference was not significant. We also compared CET activities among the mutants and WT grown under high light conditions. Both the initial rate of re-reduction of P700 ( Figure 3D ) and the electron donation activity of electrons from NADPH to PQ via Fd ( Figure 3E ) were higher in pgrl1a1b crr2 than pgr5 crr2. These results indicated that the higher stress tolerance of pgrl1a1b crr2 is attributed to its higher CET activity, compared with pgr5 crr2.

The above results demonstrate that PGR5 mutation caused lower CET activity, resulting in a more severe growth phenotype, and low photosynthetic capacities under high light conditions than PGRL1 mutation in crr2. Given that previous study has shown that PGR5 plays an important role in photorespiration (Munekage et al., 2008), we compared the photosynthetic rate monitored with photosynthetic oxygen evolution in leaf in the presence of NaHCO₃ among the CET mutants (pgr5 crr2 and pgrl1a1b crr2) and WT under high oxygen (21% O₂) conditions favorable to photorespiration and low oxygen (3% O₂) condition that inhibits photorespiration. We first carried out the experiments using seedlings grown under low light conditions. The photosynthetic oxygen evolution rate decreased significantly in pgrl1a1b crr2 and much more severely in pgr5 crr2 when compared with WT at 21% O₂ ( Figure 4A ), indicating that CET plays an important role in photosynthetic efficiency. However, the rate of oxygen evolution increased approximately twofold in both pgrl1a1b crr2 and WT and fourfold in pgr5 crr2 at 3% O₂ ( Figure 4A ), indicating that pgr5 crr2 has higher photorespiratory activity than both pgrl1a1b crr2 and WT. To see whether the upregulation of photorespiratory activity is related to the change in protein level, we compared the amount of several key proteins involved in the photorespiratory pathway. Figure 4B shows that there is no evident difference among the CET mutants and WT, suggesting that the upregulation of photorespiratory activity in pgr5 crr2 is not caused by the increased abundance of key proteins involved in the photorespiration pathway.

To obtain more comprehensive information on these mutants at the protein level, we performed a proteomic analysis of the CET mutant lines and WT grown under high light conditions. The results show that approximately seven proteins were downregulated and approximately six proteins were upregulated in pgr5 when compared with pgrl1a1b ( Figure 5A ), indicating that pgr5 and pgrl1a1b have similar changes in chloroplast proteome, which strongly correlates with their photosynthesis performances. However, under crr2 background mutation, both the downregulation and upregulation proteins significantly increased to 400 by approximately 57-fold and 326 by approximately 54-fold, respectively ( Figure 5B ). These results indicate that under crr2 background mutation, PGR5 mutation greatly changes the proteome. Differentially expressed proteins between pgr5 crr2 and pgrl1a1b crr2 were analyzed with Kyoto Encyclopedia of Genes and Genomes (KEGG), indicating that they are involved in photosynthesis; photosynthetic carbon fixation; carbon metabolism; glycine, serine, and threonine metabolism; glyoxylate dicarboxylate metabolism; pyrimidine metabolism; alanine, aspartate, and glutamate metabolism; ribosome; SNARE interactions in vesicular transport; and biosynthesis of amino acids. The proteins involved in the glyoxylate dicarboxylate metabolism pathway and carbon metabolism pathways were significantly changed. Among the significantly changed proteins, we found that the key proteins involved in photorespiration processes, such as phosphoglycolate phosphatase 1, glycolate oxidase 1, glycolate oxidase 2, peroxisomal catalase 2, glutamate–glyoxylate aminotransferase, and alanine–glyoxylate aminotransferase were dramatically upregulated in pgr5 crr2 (ranging from 86% to 215%) compared to pgrl1a1b crr2 ( Figure 5C ). The photosynthetic oxygen evolution rate also decreased significantly in pgrl1a1b crr2 and much more severely in pgr5 crr2 when compared with WT at 21% O₂ ( Figure 5D ), the rate of oxygen evolution increased approximately threefold in pgr5 crr2, and there was no significant difference among the CET mutants and WT at 3% O₂ when the plants were grown under high light conditions ( Figure 5D ). These results indicate that the impairment of both PGR5 and NDH upregulates photorespiration.

By comparing different CET mutants, we found that they are both interdependent and complementary. Our work supports previous conclusions that the PGR5/PGRL1 pathway is the dominant CET in C3 plants ( Figure 2 ). In the case of NDH deficiency, the loss of PGR5 resulted in the lowest photosynthetic performance (Figures 3A, B , 4A , 5D ) and the poorest tolerance to high light conditions when compared with PGRL1 loss in the same mutation background ( Figure 3 ). These results indicate that PGR5 was more important than PGRL1 in the PGR5/PGRL1 pathway, providing additional evidence that PGR5 can independently operate in CET in pgrl1a1b pgrl2-1 plants.

It has been reported that the localization of PGR5 to the thylakoid membrane is dependent on PGRL1 (DalCorso et al., 2008). When PGRL1 is deleted, the accumulation of PGR5 on the thylakoid membrane is affected, which in turn causes PGR5 to degrade, resulting in undetectable PGR5 protein ( Figure S3 ). Although PGR5 protein was not detected in either pgr5 crr2 or pgrl1a1b crr2, pgrl1a1b crr2 has higher photosynthetic activity ( Figures 3 , 4A ) and better growth phenotype under different light conditions ( Figure 1 ), compared with pgr5 crr2. The difference is attributed to the higher CET activity in pgrl1a1b crr2 than in pgr5 crr2 ( Figures 2B , 3D, E ), which is essential for efficient photosynthesis (Munekaga et al., 2004). The results indicate that PGRL1 deletion is not equivalent to PGR5 deficiency. To confirm this hypothesis, we generated a quadruple mutant pgr5 pgrl1a1b crr2. As expected, the growth phenotype of pgr5 pgrl1a1b crr2 was similar to pgr5 crr2 under greenhouse conditions (120 μmol m⁻² s⁻¹) ( Figure S5 ). We speculate that pgrl1a1b crr2 may contain a trace amount of PGR5 protein, with pgrl1a1b crr2 having a similar phenotype with WT and higher CET activity ( Figure 2B ) under optimal conditions.

Previous work had found that PGR5 mutations increased photorespiratory activity (Munekage et al., 2008) and that mutations in the Ndh-N or Ndh-O of maize increased photorespiration and decreased the carbon assimilation rate under high light and saturated CO₂ conditions (Peterson et al., 2016). However, the functional link between CET and photorespiration remains to be investigated. In this work, we found that growth under either low light or high light conditions, completely blocking both the PGR5 and NDH pathways, caused upregulation of photorespiratory activity ( Figures 4A , 5D ), suggesting that the photorespiratory pathway compensates for the role of CET either in the regulation of ATP/NADPH ratio or in photoprotection (Shikanai, 2014). Phenotype analysis for the plants grown under high light conditions revealed that the deletion of PGR5 under crr2 background caused photo-oxidative phenotype ( Figure 1B ), resulting in dramatic changes in protein levels ( Figure 5B ), implying the importance of both PGR5 and NDH pathways for multiple functions in addition to photosynthesis. The upregulation of the expression of photorespiration-related proteins ( Figure 5C ) in pgr5 crr2 grown under high light conditions might have resulted from the oxidative stress ( Figure 1B ) caused by low NPQ ( Figure 3B ). It was found that CET with photorespiration cooperatively regulates the redox state of P700 to suppress the over-reduction in PSI under environmental stress conditions (Takagi et al., 2016). The lowest photosynthetic capacity (oxygen evolution activity) caused by the inactivation of both PGR5 and CRR2 genes could be due to the high photorespiratory activity ( Figures 4A , 5D ). The previous studies and our results indicate that the higher the CET activity, the lower the photorespiratory activity. Therefore, the upregulation of photorespiration in pgr5 crr2 could be the functional complement for CET-dependent photoprotection and regulation of reductive powers for the loss of CET function. The coordination of CET and photorespiration is essential for maintaining the efficient operation of photosynthesis, especially under stressed conditions.

We generated knockout mutants of three nuclear genes, PGR5, PGRL1A, and PGRL1B, via CRISPR/Cas9 methods as previously described (Mao et al., 2013). The PGR5-, PGRL1A-, and PGRL1B-specific guide RNA expression sequences were introduced into the CRISPR/Cas9 construct using the primers listed in Table S1 . Hygromycin was then used to screen the above gene-edited plants without T-DNA, while mutants pgr5, pgrl1a, and pgrl1b containing no T-DNA were used in subsequent experiments.

To complement pgr5, pgrl1a, and pgrl1b, cDNA was subcloned into the pGWB17 plasmid with Myc-tag under the control of the 35S promoter. The resulting construct was transformed into the Agrobacterium tumefaciens GV3101 strain and introduced into pgr5, pgrl1a, and pgrl1b plants as previously described (Weigel and Glazebrook, 2006). Individual transgenic plants were selected based on resistance to 30 mg/L of hygromycin in half-strength MS medium and 0.8% agar. Resistant plants were transferred to the soil and grown in the growth chamber to produce seeds.

WT (Col-0) and Arabidopsis mutants (pgr5, crr2, pgrl1a1b, pgr5 crr2, and pgrl1a1b crr2) were grown on an MS medium containing 2% sucrose and 0.8% agar in a growth chamber under a 16/8-h light/dark cycle with a photosynthetic photon flux density of 60 μmol m⁻² s⁻¹ at 22°C for 10 days. For the high light treatment, the 10-day-old seedlings were transferred to soil with a 16/8-h light/dark cycle and a photosynthetic photon flux density of 120 μmol m⁻² s⁻¹ at 22°C. The seeds were incubated in darkness for 3 days at 4°C before sowing to ensure synchronized germination. For fluctuating light treatment, after the 7-day-old plants grown on MS medium were transferred to soil for 5 days at constant light (80 μmol m⁻² s⁻¹), the light condition was changed to fluctuating light with high light (600 μmol m⁻² s⁻¹) for a 1-min cycle and low light (60 μmol m⁻² s⁻¹) for a 5-min cycle at 22°C.

cDNAs encoding the mature PGR5 (amino acids 1–134) and PGRL1A proteins (amino acids 60–200) were amplified by PCR and cloned into the pET51b expression vector (Novagen, Darmstadt, Germany). The resulting plasmids were transformed into the Escherichia coli strain BL21. The fusion protein was purified on a nickel-nitrilotriacetic acid agarose resin matrix. The collected protein fraction (6 mg) was sent to Shanghai Immune Biotech to produce the antibodies.

For immunoblot analysis, the total protein samples were prepared from seedling leaves. Samples were denatured with sodium dodecyl sulfate (SDS) sample buffer in a boiling water bath for 5 min and separated by SDS–polyacrylamide gel electrophoresis (SDS-PAGE) in a 12.5% or 15% polyacrylamide gel. The proteins were then electrophoretically transferred onto a polyvinylidene difluoride (PVDF) membrane (Millipore, Billerica, MA, USA) and incubated with antibodies against chloroplast proteins. Signals were identified by an ECL plus Western blot detection system (Tanon).

Leaves used for transmission electron microscopy (TEM) were from the plants first grown on MS medium containing 2% sucrose and 0.8% agar in a growth chamber under a 16/8-h light/dark cycle with a photosynthetic photon flux density of 60 μmol m⁻² s⁻¹ at 22°C for 10 days. The seedlings were then transferred to soil with a 16/8-h light/dark cycle and a photosynthetic photon flux density of 120 μmol m⁻² s⁻¹ at 22°C for 3 weeks. Leaf segments were fixed as previously described (Lan et al., 2021) and observed with a transmission electron microscope (Hitachi H-7650, Tokyo, Japan).

Intact chloroplasts were isolated from 4-week-old leaves and purified at 4°C as previously described (Wang et al., 2006; Xu et al., 2014). The intact chloroplasts were osmotically ruptured to measure Fd and NADPH-dependent PQ reduction as previously described (Munekage et al., 2002).

The kinetics of chlorophyll (Chl) fluorescence was measured with Dual-PAM-100 or PAM-100 (Heinz Walz, Effeltrich, Germany). A transient post-illumination increase in Chl fluorescence in Arabidopsis leaves was measured after the termination of actinic light (120 μmol photons m⁻² s⁻¹ for 2 min) using a PAM 100 as previously described (Kofer et al., 1998).

The Data Acquisition Software installed in a computer connected to Dual-PAM-100 automatically calculates the Chl fluorescence parameters, Yield (II), NPQ, and Yield (I).

Fd-dependent PQ reduction activity in ruptured chloroplasts was detected by increases in Chl fluorescence by adding NADPH (0.25 mM) and Fd (5 mM) under the illumination of weak measuring light (1.0 μmol m⁻² s⁻¹) using a PAM 100, as described by a previous study (Endo et al., 1998).

The redox state of P700 was measured with a PAM chlorophyll fluorometer (Walz, Effeltrich, Germany) equipped with an emitter-detector ED-P700DW-E unit. P700 absorbance changes were monitored by absorbance at 810–830 nm (Schreiber et al., 1986; Klughammer and Schreiber, 1998). An initial reduction rate (0–1 s) of P700⁺ was estimated as CET activity (Wu et al., 2011).

Col-0 and Arabidopsis mutants (pgr5, crr2, pgrl1a1b, pgr5crr2, and pgrl1a1bcrr2) were grown on an MS medium containing 2% sucrose and 0.8% agar in a growth chamber under a 16/8-h light/dark cycle with a photosynthetic photon flux density of 60 μmol m⁻² s⁻¹ at 22°C for 10 days. We then transferred the seedlings to soil with a 16/8-h light/dark cycle and a photosynthetic photon flux density of 120 μmol m⁻² s⁻¹ at 22°C for 4 weeks. The leaves of plants from three independent biological replicates were homogenized in Rensink extraction buffer (50 mM Tris/HCl pH 7.5, 100 mM NaCl, 0.5% (v/v) TritonX-100, 2 mM of DTT, and protease inhibitor cocktail (Sigma–Aldrich, Darmstadt, Germany). The samples were then sent to the Orizymes Biotechnologies Company (Shanghai, China) for quantitative proteome analysis.

Whole leaves were cut from seedlings grown under low light for 10 days. In contrast, functional leaves from the plants grown under high light for 20 days were cut into small fragments (approximately 1 mm wide). The fragments were then stirred into a 1.8-ml suspension containing 50 mM^–1 of NaHCO₃ and 50 mM^–1 of Tris-HCl (pH 7.5) in the thermostated glass vessel of a Clark-type oxygen electrode. The photosynthetic O₂ evolution was normally detected several minutes after illumination (800 µmol photons m⁻² s⁻¹), as previously described (Wu et al., 2012).