The Arabidopsis thaliana T‐DNA insertion mutants gsa2 (GABI_364C09), ntrc (SALK_012208), trxf1 (SALK_128365) and ntrc/trxf1 were grown under short-day conditions (SD; 10 h light/14 h dark) and standard light intensity (110 μmol photons m⁻² s⁻¹) at 22°C.

To construct the 35S::GSAAT Cys substitution mutants, the AtGSA1 gene was amplified using appropriate primers (Supplementary Table 1) from Arabidopsis Col-0 genomic DNA, cloned into the entry vector pJet1.2 (Thermo Scientific) and used as a template for site-directed mutagenesis in which single cysteine codons were replaced by gene-specific serine codons (Supplementary Table 1) as suggested by Liable and Boonrod, 2009. The resulting constructs, i.e. GSAAT(C138S), GSAAT(C168S), GSAAT(C190S), GSAAT(C396S), and the GSAAT(WT) control were then transformed into gsa2 mutants using the plant transformation vector pCAMBIA-Strep (driven by a 35S promotor) and the Agrobacterium tumefaciens strain GV2260. These Arabidopsis seedlings were grown on soil under SD conditions for further analyses.

The cDNA sequence for the mature AtGSA1 (AT5G63570, without the transit peptide sequence 1–38 for plastid targeting, according to https://www.uniprot.org/uniprotkb/P42799/entry), was cloned into the pET-28a(+) vector, which includes an N-terminal 6X histidine (His)-tag (Novagen, Merck Millipore, Burlington, MA, USA). These constructs were then transformed into E. coli Rosetta cells, and their expression was induced by adding 0.4 mM isopropyl-D-1-thiogalactopyranoside (IPTG) at 16°C for 14–16 h. His-tagged TRX f1, TRX m1, and NTRC were expressed similarly and induced with IPTG for 3 h at 37°C. All the proteins were purified using nickel-nitrilotriacetic acid (Ni-NTA) resin (Thermo Fisher Scientific, Waltham, MA, USA)) according to the manufacturer’s protocol. Finally, the purified proteins were dialyzed and concentrated using Amicon^® Ultra-4 Centrifugal Filter Units (10K/30K) (Merck-Millipore, Burlington, MA, USA).

Three-week-old leaf tissue (20–30 mg) was frozen in liquid nitrogen, and homogenized. The samples were dissolved in 200-300 μl of 2×SDS‐PAGE sample buffer (100 mM Tris/HCl pH 6.8, 4% SDS, 20% glycerol and 2 mM DTT), denatured at 95°C for 5 min, and centrifuged for 1 min at 16,000 g at room temperature.

In-vivo alkylation assays were performed as previously described (Naranjo et al, 2016). Aliquots (25–30 mg) of leaf tissues were frozen in liquid nitrogen and homogenized directly in 300 µl of 10% (v/v) trichloroacetic acid (TCA) to prevent any oxidation/reduction. The samples were incubated on ice for 30 min and then centrifuged for 10 min at 16,000 g at 4°C. The pellets were then washed twice with 500 µl of acetone for 10 min each, centrifuged at 16,000 g at 4°C, resuspended in alkylation buffer (2% SDS, 50 mM TRIS–HCl pH 7.8, 2.5% glycerol, and 4 M urea), and incubated with either 10 mM methyl-maleimide polyethylene glycol (MM-PEG₅₀), 60 mM IAA (iodoacetamide) or 1 mM AMS (4-acetoamido-4-maleimidylstilbene-2,2-disulfonic acid) for 30 min at room temperature to alkylate protein thiols.

For in-vitro alkylation assays, 200–300 ng of purified recombinant His-tagged GSAAT proteins were pre-incubated with diamide, CuCl₂ or DTT in PBS buffer (150 mM NaCl, 20 mM Na₂HPO₄, pH 7.4) for 15 min at room temperature. Subsequently, the proteins were precipitated using TCA as described previously, and incubated with 1-3 µM TRXf1/TRXm1 or 100 mM N-ethylmaleimide (NEM) as indicated (Wittmann et al., 2018). The proteins to be treated with NEM were completely reduced with 100 mM DTT prior to TCA precipitation. Finally, the precipitated proteins were incubated with MM-PEG₅₀ to label the modified cysteines. Aliquots (10‐15 μl) of each sample were then loaded onto 10% or 12% reducing or non-reducing SDS PA-gels for subsequent immunoblot analysis using specific antibodies.

Tert-Butyl 4-{[(tert-butoxy)carbonyl]amino}-5-oxopentanoat (100 mg, 0.348 mmol) was dissolved in dry dichloromethane (2.5 ml). To this solution 1.5 ml 4N HCl in dioxane were added at room temperature. The mixture was stirred for 8h at room temperature. Thereafter the reaction mixture was diluted with 10 ml of diethylether. After sonication for 5 min, n-pentane (10 ml) was added and sonication was repeated for 5 min. After standing for 10 min, the supernatant was removed. The solid residue was mixed again with 5 ml of diethylether and sonicated for 5 min. After standing for 10 min, the supernatant was removed and the remaining solid was dried in high vacuum to yield glutamate 1-semialdehyde (4-amino-5-oxopentanoic acid) as its hydrochloride salt. The yield was 56mg (96%). Purity and identity were confirmed by LC-MS (pos. ESI-MS): m/z calculated for C₅H₉NNO₃ [M+H]⁺ 132.13, found 130.0 (Supplementary Figure 4). Mass spectra were recorded with an Agilent 1260 infinity liquid chromatography coupled quadrupole mass spectrometer 6120 detector.

In-planta GSAAT assays were performed as described previously (Hoober et al., 1988) with some modifications. The crude extracts for the enzyme assay were prepared by homogenizing the tissue in 0.1 M MOPS buffer (Na 2-(N-morpholino) ethanesulfonate-0.1 M Na phosphate) pH 6.8. Aliquots of the extract (200-300 µg) were pre-incubated with DTT or diamide or left untreated (UT) for 15 min at RT, then combined with 10-30 µM GSA, 10 µM pyridoxal phosphate (PLP) and 10 mM levulinic acid in a total volume of 1 ml, and incubated at 28°C for 10 min. The reaction was terminated by the addition of ethyl acetoacetate and adjusted to pH 6.8, followed by heating for 10 min at 100°C. The tubes were then cooled to room temperature and 1 volume of modified Ehrlich’s reagent (12.6% perchloric acid, 74.6% acetic acid, 11.4% HgCl₂ and 0.4% 4-NN-dimethylamino)benzaldehyde) was added, and finally absorption was recorded at 553 and 526 nm as described by Mauzerall and Granick (1956).

In-vitro GSAAT assays were performed with 5 µg of His-GSAAT, pre-incubated with either DTT or diamide, or left untreated (UT), prior to the addition of 1-3 µM His-TRX f1 or His-TRX m1 for 10 min at RT. Further steps in the assay were carried out as described above.

Full-length cDNA copies of Arabidopsis GSA, TRXf1 and NTRC genes were cloned into the pJET2.1 vector (Thermo Scientific) using appropriate primers (Supplementary Table 1). They were then fused with either the N-terminal or the C-terminal half of the YFP protein-containing plasmids pVyNE and pVyCE (Gehl et al., 2009; Invitrogen, Carlsbad, CA, USA), respectively. These fused plasmids were then transiently co-expressed by infiltration into tobacco (Nicotiana benthamiana) leaves via Agrobacterium tumefaciens GV2260. The tobacco leaf discs with the expressed proteins were then analyzed for yellow fluorescent signals after 2 days of dark incubation using an LSM 800 confocal microscope (Zeiss; λex 514 nm, λem (YFP) 530–555 nm, λem (Chl) 600–700 nm).

Chloroplast extracts (100 µg of Chl) were solubilized with 1% (w/v) dodecyl maltoside (DM) for 10 min at 4°C, and incubated with 50 μg of either purified His-TRX-f1, His-TRX-m1 and His-NTRC as bait proteins in binding buffer (BF, 25 mM Tris-HCl [pH 7.8], 150 mM NaCl, 5 mM MgCl₂, 10% [v/v], glycerol, and cOmplete protease inhibitor [Roche]) overnight at 4°C and 45 rpm. Then 50 μl of Ni-NTA agarose (Thermo Fisher Scientific) was added to each extract containing His-GSAAT proteins, and incubated for 2 h, as before. Ni-NTA resin-bound proteins were washed six times by centrifugation at 3,000 rpm for 5 min each at 4°C using BF supplemented with 10 mM imidazole. Finally, the Ni-NTA resin-bound proteins were eluted with BF + 200 mM imidazole, fractionated on a reducing 12% SDS-PA gel and probed with TRX f1 and NTRC antibodies following immunoblotting.

cDNAs were synthesized from 2 μg RNA pretreated with DNase I (Thermo Scientific) as described in Sinha et al. (2022). qRT-PCR primers used in this study are listed in Supplementary Table 1.

A X-ray crystallographic structure of Arabidopsis GSAAT1 (GSAAT_At) has been reported at 1.25 A˚ resolution (Song et al., 2016). In agreement with a previous structure for Synechocystis GSAAT (GSAAT_Syn), GSAAT_At forms an asymmetric dimer, which reflects the differential binding of its substrates pyridoxal 5’-phosphate (PLP) and pyridoxamine 5’-phosphate (PMP) as cofactors to the two subunits, respectively (Hennig et al., 1997; Song et al., 2016). The transit peptide is 40 amino acid residues (aa) long, while the mature AtGSAAT1 protein is comprised of 474 aa. The large catalytic pocket is made up of between residues 104 (Tyr) and 368 (Gly), which are flanked by a 63-aa N-terminal domain and a 106-aa C-terminal segment. The lysine residue K274 of the mature enzyme (also designated K314 in the sequence of the GSAAT_At precursor) is located close to the bound cofactor PLP, with which it forms a Schiff-base linkage (Hennig et al., 1997; Stetefeld et al., 2006; Song et al., 2016).

GSAAT_At has four conserved Cys residues, Cys138, Cys168, Cys190 (all present in the catalytic domain) and Cys396 (at the C-terminus). BLAST searches were carried out on the NCBI website (http://blast.ncbi.nlm.nih.gov/Blast.cgi) and sequence alignment of GSAATs from different species was performed using MUSCLE (https://www.ebi.ac.uk/Tools/msa/muscle/) and visualized using ESPript software (https://espript.ibcp.fr/ESPript/ESPript/) (Supplementary Figure 1). All four cysteine residues are conserved in higher plants, the first three Cys residues (Cys138, Cys168, Cys190) are conserved in the single C. reinhardtii GSAAT, whereas only Cys190 is retained in the GSAAT of Chlorobium, a genus of green sulfur bacteria. In summary, as conserved cysteine residues are potential candidates for the redox control of proteins due to their ability to undergo reversible oxidation-reduction reactions, participate in enzyme catalysis, maintain structural integrity, and sense the cellular redox environment, Cys190 conservation across different species underscores its essential role in GSAAT enzyme catalysis or its regulatory function (Supplementary Figure 1).

Fractionation of purified recombinant His-GSAAT_At (300 µM) expressed in E. coli on a non-reducing/non-denaturing polyacrylamide (PA) gel revealed that approximately half of the protein migrated as a dimer and the other half as a monomer (Figure 2A). Separation of proteins on a non-reducing SDS-PA gel, after addition of increasing amounts of the oxidizing agent copper chloride (CuCl₂₎ to GSAAT resulted in the predominance of the dimeric form, while GSAAT progressively reverted to the monomeric form as the content of the reducing agent dithiothreitol (DTT) was increased. Concentrations exceeding 0.5 mM DTT completely converted GSAAT into the monomeric form. GSAAT has been identified as a potential target of the reductants TRX-f and TRX-m in spinach chloroplasts (Balmer et al., 2003). Indeed, the disulfide bonds of GSAAT_At were also reduced when TRXs were added to the purified protein, thus preventing the potential formation of intra- and intermolecular disulfide linkages. The addition of 3 µM purified His-tagged TRX-f1 to His-GSAAT_At after pre-incubation with either an oxidant (CuCl₂) or a reductant (DTT) promoted the formation of reduced monomeric GSAAT. The GSAAT dimer was completely converted into the monomeric form at a concentration of 0.1 mM added DTT upon recycling of the oxidized TRX isoform (Figure 2B). Incubation of GSAAT_At with 5 µM TRX-m1 (plus 0.1 mM DTT), following the separation on a non-reducing SDS PA gel, still resulted in a residual amount of GSAAT dimers (Supplementary Figure 2). This finding suggests that the in vitro reducing capacity of TRX-m1 on GSAAT is lower than that of TRX-f1.

It can be assumed that GSAAT_At is able to switch between various reduced thiol groups of Cys residues and oxidized disulfide bond(s) involving Cys residues, glutathione or sulfides. To assess the number of redox-sensitive Cys residues in GSAAT, the enzyme was treated with hydrogen peroxide or DTT (0.1–10 mM DTT) and subsequently with N-ethylmaleimide (NEM) to irreversibly block the free thiol groups of Cys residues in GSAAT. Then, after incubation with 100 mM DTT, GSAAT was treated with methoxypolyethylene glycol maleimide 5000 (mPEG maleimide5000). mPEG-MAL binds to every free Cys and alters the mobility of the protein on a non-reducing SDS-PA gel electrophoresis (PAGE) depending on how many thiol-mPEG conjugates are formed. Two additional immunoreactive GSAAT bands were observed. This can be interpreted that two Cys residues were oxidized, so that they could not react with NEM. But it is not excluded that also three (or four) Cys residues could be labeled with methoxypolyethylene glycol maleimide (mPEG-MAL)-5000, due to the broader difference in molecular mass of the immune-reacting GSAAT-band with zero (0) or one (1) cysteine bound to mPED-MAL. This assessment is even more relevant if it is considered that the oxidized state would lead to an intramolecular disulfide bond, which usually could not lead to a single reduced Cys residue under reducing conditions, but must allow two bonds with mPEG-MAL, unless the second Cys residue is structurally or spatially hidden. Indeed, when GSAAT is reduced with gradually increasing amounts of DTT prior to incubation with NEM, either one Cys (Band 1, with 0.1 mM DTT) or no Cys residue of GSAAT is accessible for mPEG binding (with DTT ≥1 mM) (Figure 2C). Consequently, with increasing amounts of added DTT, GSAAT becomes more accessible to NEM prior to treatment with mPEG-MAL. We assume that at least two Cys residues of GSSAT are oxidized, which may form an intracellular disulfide bond under oxidizing conditions. The reasons for the detection of bands 1 and 2 in the oxidized state will be further discussed in the Discussion section.

His-GSAAT_At should also undergo redox-dependent structural alterations that may affect its enzymatic activity. The GSAAT activity was therefore examined under oxidized (diamide, CuCl₂) and reduced conditions (DTT and TRX’s). Oxidized GSAAT has a lower enzyme activity than GSAAT in the absence of any pretreatment (UT = untreated). Compared to the UT proteins, the dependence of DTT on His-GSAAT was determined, where the GSAAT activity was not considerably stimulated in the presence of DTT. Moreover, the GSAAT activity was activated by 2-fold when pre-treated with His-TRXf1 compared to UT. Interestingly, the ability of TRX-m1 to promote GSAAT activity was lower than that of TRX-f1, as suggested above (Figure 2D). Like GSAATs from barley, Sulfolobus solfataricus and Synechococcus PC6803, the activity of GSAAT_At was also inhibited by the inhibitor gabaculine (Smith et al., 1991; Palmieri et al., 1996; Berry-Lowe et al., 1992). Lastly, incubation of GSAAT_At at 95°C for 10 min denatured and completely inactivated the protein (Figure 2D).

In addition to the in-vitro analysis of recombinant GSAAT_At, the redox state of GSAAT in Arabidopsis leaf extracts was examined. Untreated (UT) and 30 min H₂O₂-treated leaf extracts from light-exposed wild-type seedlings contain some dimeric GSAAT, but the monomer is the dominant form. As diamide is a mild oxidizing agent compared to H₂O_2, less dimers were observed in this case, as shown by electrophoresis on a non-denaturing PA gel (Figure 3A). Another experimental approach was undertaken to determine the relative proportions of the reduced and oxidized forms of GSAAT in Arabidopsis leaf extracts. The total protein extract was pretreated with oxidizing agents (H₂O₂, diamide) or with DTT for 30 min. Then, the free Cys residues of GSAAT_At molecules in the leaf extract were labeled with the sulfhydryl-binding reagent 1mM 4-acetoamido-4-maleimidylstilbene-2,2-disulfonic acid (AMS), and subsequently separated by non-reducing denaturing sodium dodecyl sulfate–PA gel electrophoresis (SDS-PAGE). AMS binds to reduced thiol groups in Cys residues and increases the molecular mass of the protein, so that the reduced and oxidized forms of GSAAT are readily distinguishable on SDS-PA gels, since the labeled protein (reduced form) migrates more slowly than the unlabeled oxidized form. In wild-type Arabidopsis leaf extracts grown under standard short day (SD) conditions, only one distinctive immunoreactive GSAAT band – either AMS-treated or UT – was observed. In samples treated with 10 mM H₂O₂ or 200 µM diamide, the GSAAT proteins were completely oxidized and therefore exhibited a mobility like that of the UT control. In contrast to the oxidized variants, the UT and DTT-treated samples contained reduced, slowly migrating GSAAT (Figure 3B).

We also evaluated the redox state of GSAAT in light- and dark-incubated wild-type and ntrc (SALK_012208; Serrato et al., 2004) seedlings during photoperiodic growth. These experiments revealed that after transfer from dark to either high light (HL; 500 μmol photons m−2 s−1) or normal light (NL, 120 μmol photons m−2 s−1) and vice versa, for either 10 min or 30 min, ntrc extracts contained two immunoreactive GSAAT bands – a partially oxidized species and a dominant reduced form – while wild-type leaves contained only the reduced from of GSAAT, regardless of whether dark-exposed leaf samples were analyzed after a D to NL (D-NL) or a D to HL (D-HL) transition, or light-exposed samples had undergone a HL-D transition (Figure 3C). These results conclusively demonstrate that, in the absence of NTRC, a portion of GSAAT_At remains in the oxidized form. In both gel blots shown in Figures 3B, C, RbcL was used as a loading control, and it too reacts with AMS, as indicated by a mobility shift relative to the UT protein samples.

We also addressed the redox sensitivity of GSSAT in planta using three-week-old seedlings of ntrc, trxf1 (SALK_128365; Thormählen et al., 2015), ntrc/trxf1 and wild type grown under SD conditions under normal lighting (NL, 120 µmol photons m^-2 sec^-1, Figure 4A) to assay for GSAAT accumulation and activity. Knockout of the NTRC gene in Arabidopsis resulted in a growth-retarded, pale green mutant phenotype with 50% less Chl than in the wild type (Figure 4B). With 14% less Chl, the trxf1 mutant is phenotypically indistinguishable from the wild type, while ntrc/trxf1 exhibited an additive effect with severe growth retardation and 69% less Chl compared to wild type (Thormählen et al., 2015; and Figures 4A, B). Despite the decrease in Chl levels, the Chl a/b ratio did not change in either mutant or WT seedlings (Figure 4C). The ALA-synthesizing capacities of ntrc, trxf1 and ntrc/trxf1 mutants were diminished by 36%, 15% and 80% compared to WT seedlings, respectively (Figure 4D), thus confirming that the absence of NTRC is responsible for impaired ALA synthesis and decreased Chl content in single and double mutants (Richter et al., 2013; 2018; Wittmann et al., 2018; 2024).

We then analyzed the stability and activity of GSAAT in order to verify the contribution of redox-dependent control to the regulation of ALA synthesis. Immunoblots of the leaf extracts confirmed a lower content of GSAAT and some other TBS proteins (such as PORB, GluTR and uroporphyrinogen decarboxylase (UROD)) in the ntrc/trxf-1 double mutant relative to wild-type seedlings (Figures 4E, F), and also revealed a slight decrease in accumulation of GSAAT, as previously reported (Wittmann et al., 2018). The relative GSAAT protein content fell by 22%, 12% and 35% in ntrc, trxf1 and ntrc/trxf1, respectively, compared to WT, as quantified in three different immunoblot experiments (Figure 4F). These decreases in GSAAT content cannot be explained by reduced transcriptional activity of the corresponding GSA1 and GSA2 genes (Figure 4G). This observation is compatible with the lack of correlation between constant transcript content and lower levels of GluTR, UROD and PORB (Figures 4E, G). The GSAAT activity in leaf extracts decreased by 30%, 10% and 40% in ntrc, trxf1 and ntrc/trxf1, respectively, indicating a correlation between the decreases in plastidal reductants, GSAAT level and GSAAT enzymatic activity (Figures 4E, F, H). Pre-incubation of the leaf extracts with 2 mM DTT led to moderate increases of about 23%, 20%, and 30% in GSAAT activity in ntrc, trxf1, and ntrc/trxf1, respectively, compared to the UT extracts (Figure 4H). We assume that most of the GSAAT in all plant variants is present in the reduced form. Hence, the significant decrease in the rate of ALA synthesis in ntrc and ntrc/trxf1 mutants can be explained by decreased amounts of GluTR and GSAAT as the result of a relative lack of reductants.

We used different methodological approaches to confirm the interactions of GSAAT with TRXs and NTRC. First, a bimolecular fluorescence complementation (BiFC) assay was performed. Gene constructs encoding fusion proteins consisting of either the C- or N-terminal half of the yellow fluorescent protein (YFP) and the proteins of interest were transiently expressed together in leaves of Nicotiana benthamiana after infiltration with Agrobacterium strains. After two days of incubation in the dark, the YFP signal was observed by confocal laser-scanning microscopy (Figure 5A). The results confirm the interaction of GSAAT with the plastidal reductants NTRC, TRX-f1, TRX-f2 and TRX-m1. Parallel expression of GSAAT variants bearing either the N- or the C-terminal half of YFP and fusion constructs of TRXf1 and protoporphyrinogen oxidase 1 (PPOX1) were used as positive and negative controls, respectively.

Secondly, an in vitro pull-down experiment was performed with recombinant His-tagged TRX-f1, TRX-m1 and NTRC as bait proteins, in order to trap potential target proteins in the chloroplast extracts. The ALAD protein, which is a known target for TRXs and NTRC, was used as a positive control (Wittmann et al., 2024). GSAAT was found in the elution buffer after release from the specifically bound reductants in both experiments (Figures 5B, C). As NTRC, TRX-f1 and TRX-m1 were tagged with 6X-His tag, the immune reactions with the anti-His antibody were observed in all three His-tagged proteins, which were bound to the Ni-NTA resin. The light-harvesting chlorophyll a/b binding protein 1 of photosystem II (LHCB1) and photosystem I LHCA1 did not interact with NTRC, TRX-f1 or TRX-m1, respectively, in these assays.

In conclusion, these results indicate potential interaction of GSAAT with the plastidal reductants NTRC, TRX-f1 and TRX-m1. As it is obvious that TRX-interaction with its target proteins occurs only transient, we do not speculate on a tight binding. However, this finding is consistent with the outcomes of thiol-dependent affinity chromatography of TRX-f1 and TRX-m1, in which GSAAT was pulled down from extracts of Spinacia oleracea (Balmer et al., 2003), and with analyses of the Chlamydomonas thioredoxome (Perez-Perez et al., 2017).

To examine the relevance of the conserved Cys residues of GSAAT_At for its structure and activity, the WT and the four single Cys (C) to Ser (S) substitution mutant proteins - designated as GSAAT(WT), GSAAT(C138S), GSAAT(C168S), GSAAT(C190S) and GSAAT(C396S), respectively - and the double mutant GSAAT(C168S/C190S), were heterologously expressed as His-tagged proteins and subsequently purified. Although the majority of all recombinant GSAAT_At mutant variants were insoluble after induction of their expression in E. coli, a significant fraction of each of the (approximately 50 kDa) heterologous proteins remained soluble and were analyzed by non-reducing SDS-PAGE for redox-dependent switching between monomeric and dimeric states with CuCl₂ as an oxidizing agent and DTT as the reducing agent.

We verified dimer formation and in-vitro activity of purified recombinant wild-type and mutant GSAAT_At. GSAAT dimerization was examined in UT protein samples, and after the addition of 50 µM CuCl₂ or 2 mM DTT. All of the GSAAT variants were separated under reducing conditions as the 50-kDa monomer. Under oxidizing conditions, two monomeric redox states (red, ox1) could be distinguished, as the monomer migrates as a double band in GSAAT(WT) proteins and points to an independent intra-molecular redox modification. GSAAT(C396S) accumulated only in a monomeric form, while the other mutant variants were found in almost similar amounts of monomers and very small amounts of dimers in the respective UT sample and under oxidized conditions (Figure 6A). GSAAT(C138S) mimicked the structural properties and the mobility of GSAAT(WT) (not shown), therefore, a participation of Cys138 in the formation of redox dependent intra- and inter-molecular disulfide bridges are excluded. In summary, with the exception of GSAAT(C396S), fractions of GSAAT(WT) and, to a lesser extent, the other GSAAT mutants accumulate as 100 kDa dimers under non-reducing conditions.

We then examined the catalytic activity of the recombinant GSAAT_At substitution mutants under different redox conditions. GSAAT(WT), the four single and the double C→S mutants were initially oxidized in 50 µM CuCl₂ and then directly reduced by the addition of either 2 mM DTT or 3 µM recombinant His-TRX-f1. The oxidized forms of the six variants displayed similarly low GSAAT activities. Maximal GSAAT activity (relative to DTT treatment) was observed upon addition of TRX-f1. GSAAT(WT) and GSAAT(C138S) showed the biggest increases in activity – 215% and 238% and 426% and 384% in the presence of DTT and TRX-f1, respectively. This analysis also indicates that the purified GSAAT was not completely reduced. The reduced GSAAT(C168S), GSAAT(C190S) and GSAAT(C396S) mutants, as well as the double mutant GSAAT(C168S/C190S), each exhibited only half as much activity as that measured in the presence of DTT or TRX-f1 (Figures 6B, C). In fact, whether incubated with DTT or TRX-f1, the residual activity of the oxidized proteins GSAAT(C190S) and GSAAT(C168S/C190S) could scarcely be enhanced at all, which points to an impaired redox sensitivity of these mutants. We therefore assume that these mutant variants neither respond to CuCl₂ inhibition nor to DTT/TRX-f1 stimulation, and instead behave as redox-unresponsive proteins. However, the GSAAT(C396S) mutant showed a moderate stimulation of its activity (by 48% and 262% relative to the oxidized protein), when treated with DTT and TRX-f1, respectively. In light of their redox insensitivity, we suggest that GSAAT(C168S) and GSAAT(C190S) are the most likely candidates for the formation of an intramolecular disulfide bridge, while the Cys residues 138 and 396 apparently do not to contribute to the redox-sensitive activation of GSAAT.

We then constructed GSAAT_At mutants in which the conserved Cys residues had been individually replaced by Ser, and assessed their stability, activity and redox states in a gsa2 background. Previous studies revealed GSAAT2 as the dominant isoform in Arabidopsis leaf extracts, and gsa2 showed a stronger pale green phenotype than gsa1 (Sinha et al., 2022). To this end, a homozygous gsa2 strain (GK_362C09) was transformed with p35S:GSA1 mutant gene constructs by Agrobacterium-mediated transformation. The gsa2 strain itself is known to show a stronger GSAAT deficiency phenotype than its gsa1 counterpart (Sinha et al., 2022). We then selected single transgenic lines that expressed comparable levels of each of the GSAAT1 variants in the homozygous gsa2 background (Supplementary Figure 3A). All selected transgenic lines were morphologically wild-type-like. Moreover, none of the homozygous gsa2 mutants expressing any of the different transgenic GSA variants showed any visible phenotypic anomaly (Supplementary Figure 3A). Chl content and ALA-synthesizing capacity consistently remained wild-type-like (Supplementary Figures 3B, C) – with the striking exception of GSAAT(C190S), which exhibited a significantly lower rate of ALA synthesis. Owing to use of the 35S promoter for the expression of the GSAAT_At variants, GSAAT1 was overproduced in all the transgenic lines. However, GSAAT(C190S) accumulated to a lesser extent than any of the other mutants, albeit still more than that expressed in the gsa2 mutant. All other TBS proteins analyzed were expressed at similar levels in all transgenic lines (Supplementary Figure 3D). Because of the overexpression of the transgene, the complementation efficiency of the C→S substitution variants of GSAAT1 could not be assessed. Comparisons between the transgenic GSAAT1 substitution mutants and wild-type GSAAT1 expressed in the gsa2 background revealed a slightly reduced ALA synthesis capacity of leaf discs and in planta GSAAT activity of the leaf extracts only for the GSAAT(C190S) line (Supplementary Figures 3C, E). The slight decrease in the accumulation of GSAAT(C190S) is probably due to impaired redox-dependent protein stability or enzyme activity.

We also assessed the in-planta effects of altered redox conditions on the enzyme activities of the wild-type GSAAT1 and its substitution mutants, and analyzed GSAAT1 activity of chloroplasts in the presence or absence of DTT and CuCl₂. Under oxidizing conditions, a drastic decrease in GSAAT activity in leaf extracts is observed for all transgenic lines, which is similar to that seen with GSAAT(WT). UT extracts of all transgenic lines exhibited at least two- to three-fold higher GSAAT activity, indicating that GSAAT was mainly present in reduced form. Interestingly, GSAAT(WT), GSAAT(C138S) and GSAAT(C396S) lines were still markedly redox responsive, and exhibited up to 20% higher enzyme activity when supplied with DTT relative to the UT control conditions, while the two gsa2 lines expressing GSAAT(C168S) and GSAAT(C190S) resulted only in 9% and 2% increased GSAAT activity upon addition of DTT to the extracts (Supplementary Figure 3E). Hence, mutation of either C168 or C190 results in the loss of the redox responsiveness of GSAAT1 in planta.

In addition, the electrophoretic mobility of GSAAT(WT) and the GSAAT(C→S) lines is altered by the thiol-reactive compound AMS. As proposed as result of the in planta enzyme activities, this finding confirms that most of the GSAAT protein is always predominantly reduced in planta under standard light conditions (Supplementary Figure 3F).

The X-ray structure of the Arabidopsis GSAAT1 dimer has been reported at 1.25 Å resolution (Song et al., 2016). The structural model of GSAAT_At (PDB ID: 5hdm) revealed that Cys138, Cys168 and Cys190 are more buried within the dimeric structure, while Cys396 is exposed on the surface of GSAAT, albeit mirror symmetrically on opposite sides of the dimer (Figure 7A). This conformation would exclude intermolecular disulfide bonding for dimerization, as suggested above. The two residues Cys168 and Cys190, which are proposed to be highly sensitive to redox changes, show on one hand the closest proximity to each other, although the distance between the sulfur atoms of both Cys residues is still around 10.3Å. On the other hand, they are certainly more accessible for TRX in the monomeric form than in the dimeric form. But we still propose that TRX may have access to the side groups of Cys190 and likely Cys168, (Figure 7B). We speculate that conformational movement causing a structural rearrangement of the flexible loop may bring these two Cys residues closer. Testing modeling positions of the amino acid residues in the flexible loop without resulting in any steric clash using Coot (Emsley et al., 2010), allowing Cys190, situated in the flexible loop, to approach Cys168 without hindrance. The resultant distance between the sulfur atoms of the two Cys residues could decrease to less than 3Å (Figure 7C). The distance could even be further reduced to ≈2.2 Å through a slight movement and rotation of Cys168, which would permit the formation of an intra-molecular disulfide bond. Such a re-arrangement event has also been suggested for the disulfide bonding of oxidized Mg protoporphyrin methyltransferase (CHLM), another redox-controlled enzyme of Chl biosynthesis (Richter et al., 2017).

To identify the redox-sensitive Cys residues in GSAAT, we examined how the redox-dependent control of GSAAT affects its stability and catalytic activity. Mutant transgenes bearing single Cys→Ser substitutions in GSAAT_At were generated and subcloned for overproduction of the recombinant proteins in E. coli and for expression of the GSAAT1 mutants in Arabidopsis gsa2. The gsa2 mutant was selected because its seedlings show fainter green leaves than gsa1 (Sinha et al., 2022), which allows the efficiency of the GSAAT1 substitution mutants to be clearly demonstrated in terms of their ability to complement the deficiency of the dominant GSAAT2 enzyme.

When the soluble recombinant GSAAT_At variants were electrophoretically separated on a non-reducing PA gel, GSAAT(WT) was found to contain almost equal amounts of the monomer and dimer forms (Figure 6A), while GSAAT(C168S) and GSAAT(C190S) expressed only small amounts of the dimer, and GSAAT(C396S) did not migrate as a dimer (Figure 6A). The monomerization effect after adding DTT or TRX is of course interesting, especially, as the monomeric interface does not show any disulfide bond (Figure 7) and the GSAAT(C396S) mutant in E. coli was found only as monomer. It remains not clear why this surface residue Cys396, located far away from the monomer interaction could lead to monomerization. On the other hand, the GSAAT(C396S) mutant shows increased activity in the presence of a reducing agent, further confirming the redox function mediated by other cysteines (Figure 6). Apart from the low propensity of the GSAAT(C168S) and GSAAT(C190S) mutants to dimerize, these single substitution mutants not only failed to promote redox-dependent dimerization and oxidation of monomeric GSAAT, but even prevented it. Hence, we propose that the oxidation state of the monomeric GSAAT depends on the redox status of several Cys residues. Moreover, it is hypothesized that Cys168 and Cys190 can form an intra-molecular disulfide bond in wild-type GSAAT.

Neither the dimer nor the additional monomeric form (Ox) were detectable for GSAAT(C396S). The complete absence of dimerization in C396S suggests that Cys396 may be involved in an inter-molecular disulfide bond that results in the formation of an oligomeric structure that combines GSAAT with other enzymes, factors or molecular chaperones involved in ALA formation, such as GluTR, GBP (Sinha et al., 2022) and tetrapyrrole biosynthesis-regulating tetratricopeptide-repeat protein1 (TTP1) (Herbst et al., 2023). It is likely that the formation of a putative inter-molecular disulfide bond depends on the interaction between a surface Cys on GSAAT and Cys residues in other proteins, facilitating the assembly of a bimolecular or oligomeric complex which may play a critical role in stabilizing the protein structure in a dimer form.

The purified UT recombinant GSAAT variants were not entirely reduced, and additional reducing power enhanced their enzyme activities. The GSAAT(C138S) mutant displayed a level of catalytic activity similar to that of the wild-type enzyme. The enzymatic activity of GSAAT(WT) increased by 3.2 and 5.3-fold upon addition of DTT and TRX-f1, respectively, relative to the UT and oxidized protein. Similarly, GSAAT(C138S) showed 3.3- and 4.9-fold elevated activities in the presence of reductants relative to the UT GSAAT mutant. Regarding the redox-dependent structural changes in the recombinant GSAAT proteins, the mobility of the GSAAT(C138S) mutant was found to be similar to the wild type under various conditions, suggesting that a role of Cys138 in the formation of redox-dependent intra- and intermolecular disulfide bridges can be excluded (data not shown). These observations suggest that GSAAT(C138S) mimics the catalytic and structural properties of GSAAT(WT) (Figure 6B).

Supplementation with TRX-f1 enhanced the enzymatic activity of GSAAT(C396S) by 3.6-fold. However, the activities of its UT and oxidized forms were 1.3-fold lower than that of the wild-type. In contrast, the other two substitution mutants, GSAAT(C168S) and GSAAT(C190S), and the corresponding double mutant, displayed hardly any increase in activity under reducing conditions. These observations from enzyme assays lead us to propose that the Cys residues at positions 168 and 190 are essential for the protein activity and are likely sensitive to redox changes.

In gsa2 seedlings, the 35S promoter-driven expression of the Cys→S substitution mutants of GSSAT resulted in wild-type-like green pigmentation of the leaves, which indicates that the GSAAT2-deficient phenotype can be rescued by all GSAAT1 variants (Supplementary Figure 3A). This points to the capacity of all overproduced mutant GSAAT1 variants to compensate for the loss of GSAAT2. Assays performed with leaf extracts of these lines indicated that GSAAT(C138S) and GSAAT(C396) are weakly stimulated by the addition of DTT (by up to 20% compared to their UT samples). However, the GSAAT(C168S) and GSAAT(C190S) variants are hardly stimulated at all by additional reducing power. Moreover, GSAAT(C190S) also appears to be less stable than the other C→S substitution mutants (Supplementary Figure 3D).

Structural insights into GSAAT that led us to speculate about the mechanism of action of the thiol switch prompts other questions, which we would like to discuss. 1. How can this rearrangement of the loop be facilitated? The answer must take into account not only the formation of the GSAAT dimer, but also the recently proposed oligomeric structure of an ALA-synthesizing complex (Sinha et al., 2022). More specifically, one must consider that several factors could play a role in the conformational rearrangement, including changes in pH and redox states in the immediate environment. Other reasons could also promote this conformation change that facilitates the two Cys residues to come into close spatial proximity and facilitate the formation of disulfide bonds. Based on the published structure (Song et al., 2016), we suggest a possible hydrogen bonding between His192 and Asp286 with PMP due to their local proximity of a distance of 2.7 Å (Figure 7D). By sharing a hydrogen of His192 and Asp286, this possible hydrogen bond interaction could be part of a network of interactions that influences the conformation and dynamics of the protein. Thus, we propose that these structural rearrangements are due to the positioning of the hydrogen bond, which could affect the flexibility of nearby loops, and possibly trigger the movement of the Cys190-containing loop (Figure 7D).

2. How can the reduced activity of oxidized GSAAT be explained? How can disulfide bonding lead to reduced activity? Once the disulfide bridge is formed as a result of loop rearrangement, the bound PMP could potentially be accessible to additional interaction partners (Figure 7E). We suggest that the hydrogen rearrangement could take place between His192 in the flexible loop and Asp286. Both residues are within an ideal hydrogen bond distance from the N-atom of the ϵ-amino group of PMP after the formation of the Schiff-base linkage between K274 (K314 of the precursor protein) and PLP (the N-atom of the bound PMP/PLP). Such additional interaction, facilitated by the disulfide bridge, may altered the enzyme conformation, effecting its catalytic efficiency.

3. What consequences might a disulfide bond formation between Cys168 and Cys190 have for the action of the cofactor PMP? The structural change (Figure 7E) could potentially hinder the transamination reaction, because the loop movement might bring the ϵ-amino group of PMP into closer proximity to other amino-acid side groups, such as His192, promoting the hydrogen bond formation that may influence the reactivity of PMP with GSA. Such changes could reduce the transamination activity, as suggested by an interference with the formation of diaminovalerate, a proposed metabolic intermediate in this catalytic reaction of GSAAT (Smith et al., 1991). We are aware that this is a hypothesis for which there is currently no direct biophysical confirmation.

In light of experimental evidence showing the different monomeric and dimeric states of the Cys substitution mutants compared to GSAAT(WT), we propose that Cys168 and Cys190 are cruical for the protein activity and likely function as redox-active residues. The single mutants GSAAT(C168S), GSAAT(C190S) and the double mutant GSAAT(C168S/C190S) all showed decreased redox-induced dimerization, suggesting a potential formation of an intramolcular disulfide bridge. C396S substitution disables GSAAT homodimerization. It remained to be determined whether intermolecular bonding to another protein, such as GBP, TTP1 or ALAD (Sinha et al., 2022, Herbst et al., 2023, Wittmann et al., 2024) would stabilize the dimeric GSAAT in the ALA synthesizing complex.