Protein targeting to starch 1, a functional protein of starch biosynthesis in wheat (Triticum aestivum L.)

TaPTST1, a wheat homolog of AtPTST1 containing CBM can interact with GBSSI and regulate starch metabolism in wheat endosperm. In cereal endosperm, native starch comprising amylose and amylopectin is synthesized by the coordinated activities of several pathway enzymes. Amylose in starch influences its physio-chemical properties resulting in several human health benefits. The Granule-Bound Starch Synthase I (GBSSI) is the most abundant starch-associated protein. GBSSI lacks dedicated Carbohydrate-binding module (CBM). Previously, Protein Targeting To Starch 1 (PTST1) was identified as a crucial protein for the localization of GBSSI to the starch granules in Arabidopsis. The function of its homologous protein in the wheat endosperm is not known. In this study, TaPTST1, an AtPTST1 homolog, containing a CBM and a coiled-coil domain was identified in wheat. Protein-coding nucleotide sequence of TaPTST1 from Indian wheat variety ‘C 306’ was cloned and characterized. Homology modelling and molecular docking suggested the potential interaction of TaPTST1 with glucans and GBSSI. The TaPTST1 expression was higher in wheat grain than the other tissues, suggesting a grain-specific function. In vitro binding assays demonstrated different binding affinities of TaPTST1 for native starch, amylose, and amylopectin. Furthermore, the immunoaffinity pull-down assay revealed that TaPTST1 directly interacts with GBSSI, and the interaction is mediated by a coiled-coil domain. The direct protein–protein interaction was further confirmed by bimolecular fluorescence complementation assay (BiFC) in planta. Based on our findings we postulate a functional role for TaPTST1 in starch metabolism by targeting GBSSI to starch granules in wheat endosperm.


Introduction
Starch is a plant-derived carbohydrate and a significant source of calories for humans and livestock. In wheat grain, starch accounts for nearly 70% of the dry weight and is the key determinant of its nutritional quality and yield (Shewry and Hey 2015). The plant diurnal dynamics enable plants to form two types of starch, namely transient and storage starch. The transient starch is synthesized in leaves during the daytime and gets metabolized as a carbon source for non-photosynthetic tissues during the night. The storage starch gets deposited in the storage tissue like endosperm in cereal seeds and tubers of different crops and serves as the carbon reserve for germination (Bahaji et al. 2014). Starch is semi-crystalline granules, composed of two types of glucose polymers amylose and amylopectin. Amylose is mainly a linear polymer of glucose linked by α-1,4-glycosidic linkage, whereas amylopectin is a highly branched molecule (Pérez and Bertoft 2010).
The starch biosynthesis in cereal endosperm is resultant of the coordinated activities of different enzymes including ADP-glucose pyrophosphorylase (AGPase), soluble starch synthase (SS), granule-bound starch synthase (GBSS), starch branching and debranching enzymes, and starch phosphorylase. The detailed starch biosynthesis pathway in cereals is reviewed elsewhere (Keeling and Myers 2010;Tetlow 2018). Briefly, the first committed step towards the starch biosynthesis is the AGPase reaction resulting in the ADP-glucose (ADPG), a substrate for starch synthases, that is transported to amyloplast by Bt1 transporter (Ghosh and Preiss 1966;Beckles et al. 2001). Two isoforms of GBSS are expressed in a tissue-specific manner. GBSSI in the amyloplast solely elongates the linear glucose chain by transferring glucose moiety using ADPG as the substrate. The other isoform GBSSII is responsible for amylose biosynthesis in leaves contributing to the transitory starch (Vrinten and Nakamura 2000). Soluble starch synthase (SSs) is involved in the biosynthesis of amylopectin. SSI is involved in the elongation of the short-chain of glucans of DP 6-10, while SSII is responsible for the synthesis of chain DP 12-20 (Delvallé et al. 2005;Zhang et al. 2008). An earlier study in Arabidopsis reported the involvement of SSIII in the synthesis of longer glucan chains in leaves (Zhang et al. 2005). SSIV is required for the initiation of granules rather than the elongation of glucan chains (Roldán et al. 2007).
The proportion of amylose and amylopectin in starch largely influences its physio-chemical properties and nutritional quality by virtue of structural differences between amylose and amylopectin. The health benefits associated with starch rich in amylose have been studied and discussed elsewhere (Birt et al. 2013;Li et al. 2019;Lim et al. 2020).
GBSS is the only starch synthase responsible for the synthesis of longer amylose chain. The knockout or loss of function mutants of GBSSI is reported to synthesize amylose-free (waxy) starch in different species, like hexaploid wheat (Nakamura et al. 1998), rice (Wang et al. 1995), potato (Hovenkamp-Hermelink et al. 1987) and maize (Shure et al. 1983).
In cereals, amylose biosynthesis lags behind the amylopectin biosynthesis during seed development (Tomlinson and Denyer 2003;Radchuk et al. 2009). The amylose biosynthesis genes are mainly upregulated in the mid and later stages of seed development, unlike the expression of branching enzymes and soluble starch synthases which are higher during early stages of grain filling. The temporal arrangement of the amylose and amylopectin biosynthesis is hypothesized to be crucial since amylose biosynthesis occurs within the amylopectin matrix (Tatge et al. 1999;Tomlinson and Denyer 2003).
However, the relationship between the overexpression of GBSSI and amylose content is not linear in maize and potato (Tsai 1974;Flipse et al. 1996). GBSSI overexpression in wheat did not affect amylose content substantially (Sestili et al. 2012), suggesting the involvement of other factors associated with amylose biosynthesis. GBSS is one of the most abundant starch-bound proteins closely associated with starch granules. The enzyme is highly unstable in the soluble fraction (Seung et al. 2015;Hebelstrup et al. 2017). The enzyme lacks any dedicated CBM for its localization to the starch granules, a requisite to the amylose biosynthesis. A non-catalytic carbohydrate-binding protein PTST1 was identified as critical to localize GBSS in Arabidopsis leaves. The C-terminal catalytic domain of GBSS interacts with a protein PTST via coiled-coil domain and the interaction is crucial for the localization of enzyme to the starch granules. PTST was identified as a non-catalytic protein pertaining to CBM48 and the coiled-coil domain in Arabidopsis (Seung et al. 2015). The mechanism of starch biosynthesis in leave and endosperm differs (Tetlow et al. 2018). In rice, a GBSS binding protein, namely GBP was identified to affect the quality and yield of rice (Wang et al. 2020). A PTST1 homolog was found to be involved in targeting HvGBSSI and normal starch biosynthesis in barley endosperm (Zhong et al. 2019). However, there is no report of GBSS binding proteins and its targeting to starch granules in wheat endosperm and leaves. Despite the availability of whole-genome sequences of Triticum aestivum, there are many genes with no assigned functions. In this study, we used molecular and biochemical approaches to identify and characterize the homolog of AtPTST1 in wheat endosperm. The direct protein-protein interaction between wheat GBSSI and PTST was identified in vitro and in planta. Our study suggests TaPTST1 as a GBSS binding protein with a possible role in starch metabolism by ushering GBSSI to starch granules in wheat endosperm.

Plant materials
The Indian bread wheat variety 'C 306' was planted during the Rabi season 2019-2020 at the research farm of the National Agri-Food Biotechnology Institute, Mohali, India. This study utilized low amylose mutants 'TAC 6, 'TAC 358' and 'TAC 846' from the previous study from our lab (Mishra et al. 2016). The primary spikes from the individual plant in three biological replicates were harvested at 7, 14, 21 and 28 DPA (days post anthesis) and stored at -80 °C, till further use. For the bimolecular fluorescence complementation (BiFC) assay, Nicotiana. benthamiana wild-type plants were grown at 25 °C with a photoperiod of 10 h Light/14 h dark in the plant growth chamber.

Phylogenetic and bioinformatic analysis
The amino acid sequence of AtPTST1 was used to search EnsemblPlants database (https:// plants. ensem bl. org/ Triti cum_ aesti vum/ Info/ Index). The amino acid sequences of orthologous genes were retrieved with the highest similarity with AtPTST1. The alignment of the sequence was created using MEGA X (Kumar et al. 2018). The neighbour-joining method was used to construct the phylogenetic tree with the following parameters: p-distance model, pairwise deletion, and bootstrap (1,000 replicates). The alignment used for phylogenetic tree construction is provided in Data S1. The coiled-coil domain was predicted by the COILS/PCOILS server (https:// embnet. vital-it. ch/). The domain analysis was performed using the SMART webserver (Letunic et al. 2021).

Cloning and sequencing of TaPTST1
The cDNA libraries were prepared from the Indian wheat variety 'C 306' from the pooled samples of total RNA isolated from wheat grains at 7, 14 and 21 DAP. Full-length TaPTST1 protein-coding nucleotide sequence of 879 bp was amplified from cDNA by using primers (Table S1) and cloned to pJET1.2 cloning vector. A total of 30 clones were sequenced by the Sanger sequencing method (ABI3730XL instrument). Sequences were aligned to 'Chinese spring' reference sequences retrieved from the ensemble plant using multialign (Corpet 1988).

Homology modelling and docking
Homology model of wheat PTST_AMPK domain was built using I-TASSER webserver, and TOME V3 Server (Pons and Labesse 2009;Yang and Zhang 2015). The model was validated using various validation tools like ERRAT plot, Ramachandran plot, ProSA, ProQ, and verify-3D. The COACH server was used to predict protein-ligand binding sites (Yang et al. 2013a, b). For homology modelling of CC structure of TaPTST1 Swiss model was used and molecular docking of TaPTST1 coiled-coil domain and GBSSI was performed by the HDOCK webserver (Yan et al. 2017(Yan et al. , 2020.

Quantitative real-time PCR
Total RNA was extracted from the grains collected at different seed developmental stages using TRIzol method (Simms et al. 1993) (Simms et al. 1993). RNA concentration was measured on the Nano Droplite (Thermo Fisher Scientific, USA) and the integrity of RNA was checked on the 1.5% agarose gel. Genome-specific primers of the genes were designed using Prime 3.0 software. cDNA was prepared using 'SuperScript™ III First-Strand Synthesis System' following the manufacture's protocol. The relative quantitative gene expression analysis was conducted in three technical replicates on the CFX96 thermocycler (BIO-RAD, USA) system. For normalization of gene expression, the wheat tubulin gene was used as an internal control. All the primers used in the study are listed in Table S1.

Cloning and recombinant protein expression in E. coli
The full-length CDS of TaPTST1 was amplified using forward and reverse primers (Table S1) and CDS of GBSSI was previously cloned (Mishra et al. 2021). For each of the genes, 5 clones were sequenced and aligned with sequence retrieved from the EnsemblPlants database.
A 1605 bp nucleotide fragment coding for mature GBSSI protein, lacking 70 amino acids predicted transit peptide was cloned in pET28a expression vector using EcoRI and XhoI restriction endonucleases. For the full-length PTST expression, the expression vector was constructed using BamHI and SacI restriction enzymes. The N-terminal fragment of wheat PTST was cloned into the pET28a expression vector using EcoRI and XhoI enzymes to create PTST-N1-209 with N-terminal His-tag. Recombinant proteins with N-terminal His-tag were expressed in E. coli Rosetta cells harbouring each of the IPTG-inducible expression construct i.e., pET28a-TaGBSSI, pET28a-TaPTST1, and pET28a-TaPTST-N1-209. Cells were induced using 0.2 mM IPTG and grown in LB medium at 16 °C with constant agitation for 16 h. Sorbitol (0.5 M) was used in the culture media to promote protein solubilization (Prasad et al. 2011). Cell lysis and protein extraction was performed using an earlier described method (Spriestersbach et al. 2015). Proteins were purified by Ni-NTA affinity chromatography using standard protocol followed by desalting. GBSSI and TaPTST1 were eluted at the concentration of 100 and 150 mM imidazole in 100 mM Tricine buffer respectively. Purified proteins were detected by immunoblotting with anti-His antibodies.

Starch binding assay
The starch binding assay was performed using the protocol described previously (Seung et al. 2015). Briefly, 1 µg of the recombinant TaPTST1 was incubated in a binding medium with purified wheat starch, amylose, and amylopectin from potato starch (Sigma Aldrich) for 30 min at 20 °C with endover-end shaking. Control reaction was carried out in same condition without any glucan. The supernatant was collected by centrifugation of the mixture for 2 min at 13,000 g and the pellet was washed three times with wash buffer. The pellet, final wash, and supernatant were heated at 95 °C into the SDS-sample loading buffer, followed by SDS-PAGE and immunoblotting with anti His-tag rabbit monoclonal antibodies (RM146 Thermofisher scientific) at 1:3000 dilutions. Goat Anti-Rabbit IgG (H + L)-HRP conjugated secondary antibodies were used at the same dilution for the detection of expressed proteins.

Immunoaffinity pull-down assay with recombinant protein
Purified recombinant GBSSI protein (10 μM) was incubated with protein-A sepharose beads immobilized with equimolar concentration of anti-wheat GBSSI antibodies in 100 μl of binding buffer containing 100 mM tricine (pH 7.5) and 0.1% BSA for 4 h. GBSSI antibodies were raised in rabbits against the peptide CRGKTKEKIYGPDAGTDYED (ABGENEX Pvt. Ltd.). The recombinant TaPTST1 was added to the reaction and incubated for 4 h. The resin was washed five times with wash buffer, all steps were carried out at 4 °C. The resin was heated in SDS-sample loading buffer and the proteins were visualized using SDS-PAGE followed by immunoblotting. Immunoblotting was performed with the 6x-His tag antibodies. The pull-down assay was performed for the fulllength PTST peptide as well as for the N-terminal truncated peptide.

Generation of BiFC vector constructs and transient expression in tobacco
The full length coding sequence of wheat PTST was amplified and cloned in plant compatible BiFC vector pSPYCE, using BamHI and XmaI restriction sites, enabling the construction of C-terminal fusion protein with eYFP C-terminus. GBSSI was cloned in pSPYNE173 using XhoI and XmaI restriction sites to synthesise fusion protein with eYFP N-terminus. Both fusion proteins expressed under 35S promoter were used to transform Agrobacterium tumefaciens (strain GV3101). BiFC experiments were performed according to the method described earlier (Thakur et al. 2021). Briefly, positive transformants were grown in LB medium with the selection markers rifampicin, gentamycin, and kanamycin at 28 °C for 24 h. Cells were pelleted by centrifugation and resuspended in a resuspension medium and kept for 6 h before the infiltrations. The cell suspensions of both the expression constructs and blank vector control were infiltrated into the intercellular spaces between abaxial epidermal cells of intact N. benthamiana leaves (4-5-weekold) using a 1 ml syringe. Plants were kept for 48 h under the same growth condition in growth chambers prior to microscopy.

Microscopy and imaging
For the microscopic analysis, tobacco leaf discs were cut after 48 h of infiltration. Using water as the imaging medium, lower epidermis cells were analysed at room temperature. The fluorescent YFP-signals were detected using a Carl Zeiss confocal microscope (LSM880) (Karl Zeiss, Jena, Germany). YFP signals were monitored using argon laser for excitation at 514 nm and wavelength window of 518 to 557 nm for emission signals. ImageJ software was used for the processing of the images.

Identification of AtPTST1 homologs in wheat
Homologs of AtPTST1 were identified using BLASTp and AtPTST1 amino acid sequence as a query at NCBI and Uni-Prot database. There was no sequence with significant similarity in the Triticum species. The BLASTp search at ensemble plant database (Ensembl Plants release 47) found three peptides of 292 amino acid each, TraesCS6D02G090900.1, TraesCS6A02G102300.1 and TraesCS6B02G130700.1 with the 54.5, 55.1, and 60.9% identity with AtPTST1 respectively. The identified protein was found to be encoded by the three homeologues gene located at chromosome 6A, 6B and 6D.
Sequencing of all the three homeologues in the 'C 306' Indian wheat variety identified no sequence variation compared to the 'Chinese spring' reference genome. The nucleotide and amino acid alignments of all three TaPTST1 homeologues of 'C 306' with Chinese spring are shown in supplementary data (Figs. S1 and S2). The sequence similarity among the nucleotide sequences of three homeologues encoded by A, B, and D genome was A to B (90%), B to D (90%), and A to D (93%). The peptide encoded by the three genomes of wheat showed a high sequence identity of A to B (94.52%), B to D (94.18%), and A to D (92.18%). The TaPTST1 was encoded by 8 exons with transcript lengths of 1346, 1321, and 1282 bp of A, B, and D genome homeologues respectively. All the transcripts resulted in 879 bp long coding sequences encoding for 292 amino acids long peptide.
The candidate protein was found to have 72.88% identity with rice PTST (XP_015627711.1). The molecular phylogenetic analysis revealed the conservation of the protein sequences among the different species (Fig. 1). The PTST1 from monocots and dicots are grouped into two different clades. All three homeologues of TaPTST1 showed closed sequence similarity and grouped into the same clade and showed the highest identity with barley PTST1 sequence among all the monocots.

Domain analysis of TaPTST1
The amino acid sequence of candidate TaPTST1 was analysed using the Pfam and CDD database. The protein was found to have a 28 amino acid long coiled-coil domain (108-135) and an AMPKbeta-like C-terminal domain (cd02859). The AMPKbeta-like C-terminal domain was found to have carbohydrate-binding residues, suggesting a probable site that may interact directly with starch granules in amyloplast. The C-terminal domain was found to have conserved residues among the different species including AtPTST (Fig. 2a). The coiled-coil prediction using a coils web server (https:// embnet. vital-it. ch/ softw are/ COILS_ form. html) identified the strong probability of the residues from 110 to 140 to form coiled-coil structures (Fig. 2b). The coiled-coil structures are the potential interaction site for GBSS binding in AtPTST (Seung et al. 2015). Similar coiled-coil structures were found in the wheat GBSSI (Fig. 2c). The presence of the potential binding sites among both the proteins is indicative of the probable interactions between the TaGBSSI and TaPTST1.

Homology modelling and structural analysis
A detailed amino acid analysis for secondary structure predictions using the PSIPRED tool showed the strong potential of residues at N-terminal to form coiled structures. PSIPRED analysis showed the three helical structures in N-terminal one from residues 51-86, second from 88 to 117, and the third helix of 18 residues from 135 to 152 (Fig. S3).
The unavailability of decoded structural information of any homolog of the PTST type of proteins is a limitation to access the structural aspects. To identify the residues of PTST which are directly involved in interaction with glucans, the TaPTST1 secondary and tertiary structural homology model was constructed. TaPTST1 Psi-blast in uniprot did not result in any significant sequence similarity to any PDB structure. Therefore, AMPK domain and N-terminal CC structures were modelled separately. The homology model of AMPK domain of TaPTST1 was generated by i-Tasser server. The built model had the C-score of -4.26 with the estimated TM-score of 0.26 ± 0.08. Additionally, homology model of TaPTST1 was also built using AAKB2_RAT (PDB: 4Y0G) as Fig. 1 The neighbour-joining phylogenetic tree of PTST protein from MAFFT alignment. Tree was constructed with the p-distance model, pairwise deletion, and bootstrap (1,000 replicates). Bootstrap values are shown over the branches 1 3 a template with the TOME V3 Server. The selected template showed highest identity percentage to the query sequence and the homology protein model was built with the Atome score of 90.74. Both the homology models showed similar structures. TaPTST1 was then analysed for potential ligand binding sites and identified W232, K263, W270, N286 consensus alpha-D glucopyranose binding residues in C-terminal CBM. The modelled PTST structure docked with GLC is depicted (Fig. 3a). The identified residues were surface exposed and may be potential glucans-interacting residues.
The homology model of TaGBSSI without predicted signal peptide was generated using decoded rice GBSSI crystal (PDB: 3UVE). The CC-structure model generated by Swiss model server (Fig. 3b) was docked with TaGBSSI at Hawkdock server predicted coiled-coil mediated interaction between the proteins. Further, analysis of docked-molecule identified a coiled-coil helical structure in the C-terminal domain of TaGBSSI (Fig. 3c), where, some key surfaceexposed residues K365, E368, K372, E375 were identified as the possible interacting residues (Fig. 3d).

Homeologues specific expression profile of GBSSI and PTST
We investigated the expression profile of TaPTST1 encoding transcripts in six different tissues i.e., root, stem, leaf, spike, seedling and grain using gene expression values (TPM -Transcripts Per Million) of hexaploid wheat cultivar 'Chinese Spring' from publicly available wheat expression database. The three transcripts of homeologues of TaPTST1 was found to be expressed substantially in all the tested tissues. Interestingly, the TaPTST1 showed highest expression in grain, while the expression of all homeologues was comparable in other tissues. Among the three homeologues TaPTST1_6D (TraesCS6D02G090900) showed highest expression in all tissues (Fig. 4a).
To further analyse the expression profile of TaPTST1 and GBSSI homeologues at different seed developmental stages, i.e., 7 DAA (Days after anthesis), 14 DAA, 21 DAA, and 28 DAA, q-RT PCR was performed. Among the three homeologues of GBSSI, GBSSI_4AL was found to be null for transcript expression, while the expression of GBSSI_7A was consistent throughout the seed development. The expression of GBSSI_7D was found to be higher after the 21 DAA. The 6D and 6A homeologues of wheat TaPTST1 showed higher expression at the later seed developmental days, while the expression of PTST_6B was lower than other homeologues (Fig. 4b). The results indicated expression of the transcripts GBSSI and TaPTST1 in wheat grain throughout the grain filling stages. The highest expression of TaPTST1 in grain may be indicative of its probable function in wheat grain. GBSSI is known to be the most abundant

Direct binding of TaPTST1 with starch granules
The TaPTST1 was predicted to have a CBM module that could bind to starch. We analysed the PTST binding affinity against wild-type wheat starch, purified amylose, and amylopectin. The amylose and amylopectin were assayed separately to identify the specific interaction of TaPTST1 with two different polymers. His 6 -TaPTST1 was incubated with any of the native starch, amylose, and amylopectin. We found that the TaPTST1 can interact with wheat starch well when assayed with wheat starch. Although, the protein interacts with less affinity to purified amylose fraction of starch and showed no binding with amylopectin (Fig. 5). The interaction of TaPTST1 with starch suggested the probable role of the protein in starch biosynthesis.

The C-terminal domain is essential for carbohydrate-binding
The C-terminal AMPKbeta-like domain was predicted to have carbohydrate-binding residues and serve as the carbohydrate-binding module. To understand the significance of the domain in TaPTST1, a truncated peptide TaPTST1-209 lacking the AMPKbeta-like domain was expressed and purified. Interestingly, the TaPTST1-209 fails to interact with the starch granules in vitro, suggesting the crucial role of the domain as the carbohydrate-binding module in TaPTST1 (Fig. 5). Additionally, no interaction between TaPTST1-209

Identification of direct interaction between TaPTST1 and GBSSI
Domain analysis and docking results suggested the possible interaction between TaPTST1 and TaGBSSI mediated by the coiled-coil domain. We investigated the direct interaction between the proteins by immunoaffinity pull-down assay using recombinant proteins expressed in E. coli. N-terminal His-tagged GBSSI was incubated with N-terminal Histagged TaPTST1. GBSSI was recovered using protein-A sepharose beads coated with anti-TaGBSSI antibodies. Additionally, an immunoaffinity pull-down assay was also performed with TaPTST1-209 lacking the AMPK domain. Both TaPTST1 and TaPTST1-209 were co-eluted with GBSSI in a substantial amount, indicating a direct interaction between two recombinant proteins. The TaPTST1-209 was able to interact with GBSSI, indicating that the AMPK domain has no role to play in the GBSSI binding of TaPTST1 (Fig. 6).

Direct interaction between TaPTST1 and GBSSI in planta
Further, to confirm the direct protein-protein interaction between TaPTST1 and GBSSI in plants, BiFC assays were performed. The pSPYCE::TaPTST1 and pSPYNE173::GBSSI constructs encoding TaPTST1_YFP C and GBSSI_YFP N respectively were transiently expressed in Nicotiana benthamiana leaf epidermal cells. YFP protein was expressed under the control of constitutive CaMV 35S promoter as YFP control for the experiment and YFP signals were detected as expected. The YFP fluorescence signal was detected in chloroplasts when both plasmids were co-expressed in N. benthamiana leaves. The leaves co-infiltrated with either pSPYCE::TaPTST1 and pSPYNE173 or pSPYNE173::GBSSI/ pSPYCE resulted in non-detectable YFP signals (Fig. 7). These results confirmed that both the Fig. 5 C-terminal AMPK is essential for the carbohydratebinding of TaPTST1. a purified recombinant TaPTST1 with N-terminal His 6 -tag b Binding of His 6 -TaPTST1 to wild-starch, amylose, and amylopectin. Unbound proteins are represented as a soluble fraction (S), W f is the final wash, and starch-bound proteins are shown as a pellet (P) Control reaction was carried out without any glucan. c The truncated peptide His 6 -TaPTST1-209 lacking C-terminal AMPK. d Binding of His 6 -TaPTST1 His 6 -TaPTST-209 with wildtype starch in vitro Fig. 6 In vitro immunoaffinity pull-down assay using GBSSI antibodies. a with recombinant His 6 -TaPTST1 b recombinant His 6 -TaPTST1-209. Inputs represent the proteins used in the assay and S and B denote soluble and bound proteins. Immunoblotting was done using His 6 antibodies 1 3 proteins are localized to chloroplast and TaPTST1 directly interacts with GBSSI in leaves.
Therefore, the interaction between the GBSSI and TaPTST1 in vitro and in planta and starch binding affinity of TaPTST1 suggested the probable role of TaPTST1 in starch biosynthesis by targeting GBSSI to starch granules.

Lower expression of TaPTST1 in low amylose wheat mutants
To understand the effect of TaPTST1 on amylose content in wheat grain, expression analysis of TaPTST1 and GBSSI was performed in the low amylose mutant lines at different seed developmental stages. The amylose content in mature grains of mutant lines 'TAC 6','TAC 358' and 'TAC 846' was ~ 6%, ~ 3%, and 7% respectively, while the amylose content in wild wheat variety 'C 306' was ~ 26% (Mishra et al., 2016). Low amylose or amylose free phenotype is largely believed to be associated with the key enzyme GBSSI. The two transcripts of GBSSI were expressed upto 10 Log 2 fold lower in lines 'TAC 6' and 'TAC 358' than 'C 306', while the most of the transcripts of TaPTST1 was also found to be down regulated. Interestingly, in 'TAC 846' the expression of GBSSI-7A was up to 3 Log 2 fold lower than 'C 306', while the expression of TaPTST1 was upto 10 Log 2 fold down-regulated (Fig. 8).

Discussion
Understanding the starch metabolism pathways in cereals, mostly rice and wheat is at the centre to improve grain's nutritional and processing quality by targeting one or more pathway enzymes. GBSSI is the most abundant starchassociated protein in cereal endosperm and the enzyme is dependent on unknown protein for interaction with glucans, as it lacks any dedicated Carbohydrate-binding domain (CBM). Previously, PTST1-mediated targeting of GBSSI was reported in Arabidopsis (Seung et al. 2015). A homolog of AtPTST1 was identified and found crucial for the development of starchy endosperm in barley (Zhong et al. 2019). In this study, we identified and characterized TaPTST1, a homolog of AtPTST1 in hexaploid wheat.

TaPTST1 is a functional protein and can interact with starch via its CBM domain
Expression analysis of starch metabolism genes and related transporters and transcription factors in wheat grain at different grain developmental stages showed different patterns (Singh et al. 2015). Based on the highest expression of TaPTST1 wheat grain, the potential role of the protein in grain may be postulated. The identified TaPTST1 showed significant sequence similarity with known PTST sequences from other plant species. However, monocots and dicots were grouped in two different clades in the phylogenetic tree, which may be due to the early divergence of monocots from their dicot relatives (Schuyler et al. 1983).
The AMPK domain-containing proteins have a glycogenbinding domain (GBD) in mammals (McBride et al. 2009).
In plants, the GBD is known as CBMs that directly interacts with carbohydrates like starch (Christiansen et al. 2009). Different CBM containing proteins have been identified in plants (Seung et al. 2015(Seung et al. , 2017(Seung et al. , 2018Zhong et al. 2019;Wang et al. 2020) and their role was established in starch metabolism. The CBM was found to be located on the N or C-terminal of these proteins. The TaPTST1 peptide contains a C-terminal AMPK domain serving as a carbohydratebinding module and showed close sequence similarity with barely PTST1 (Zhong et al. 2019). Some of the residues are conserved (marked by * in Fig. 3) across the analysed sequences, suggesting the conserved role of the residues. We demonstrated the direct binding of TaPTST1 with starch and amylose with different affinities that may be attributed to the structural differences between the molecules. TaPTST1 showed no affinity for amylopectin, unlike AtPTST1, which showed binding with both amylose-free and wild-type starch (Seung et al. 2015). The TaPTST1 lacking C-terminal AMPK failed to bind the starch in vitro, suggesting that the AMPK may serve as a CBM. These findings demonstrated that TaPTST1 is a novel CBM containing protein with no assigned function in wheat and can interact with starch by its C-terminal CBM.

Potential role of TaPTST1 in starch biosynthesis by targeting GBSSI to starch granules
In cereal endosperm, the site of amylose biosynthesis is pre-existing, semi-crystalline amylopectin matrix, permeable to small molecules (Tatge et al. 1999). GBSSI is one of the most abundant starch-bound proteins in wheat grain. Although it lacks any CBM for its localization to starch granule, that is essential for the functioning of the enzyme. However, the mechanism of GBSSs entry into the matrix of the starch granule remains elusive. The TaGBSSI has a C-terminal coiled-coil structure that was previously found as a potential structure for interaction with other coiled-coil containing proteins via electrostatic interactions mediated by surface-exposed lysine and glutamate residues. Charged amino acid-mediated electrostatic interactions are shown to play a major role in coiled-coil interactions (Mason and Arndt 2004;Thompson et al. 2012).
In our study, the direct interaction between GBSSI and TaPTST1 identified by immunoaffinity pull-down assay indicated the potential role of TaPTST1 in targeting GBSSI to the starch granules. Further, the BiFC assay identified the interaction between TaPTST1 and GBSSI in the chloroplast of tobacco leaves. Interestingly, the lower expression of TaPTST1 and GBSSI was found in low amylose mutant lines 'TAC 6' and 'TAC 358'. The expression of GBSSI in 'TAC 846' was not significantly affected, while the expression of TaPTST1 was down-regulated to 10 log 2 fold change. Seung et al., 2015 reported the amylose-free phenotype of GBSSI knockouts as well as PTST1 knockouts in Arabidopsis. The lower expression of TaPTST1 in low amylose lines is evident to its role in amylose biosynthesis. Taken together the findings, we postulate that TaPTST1 interacts with TaGBSSI protein via the coiled-coil domain and targets it to the starch granules in wheat endosperm. Understanding the mechanism and role mediated by TaPTST1 in starch biosynthesis is important to develop or select better crop varieties with enhanced nutrition properties.

Acknowledgements
We would like to thank the Executive Director of the National Agri-Food Biotechnology Institute (NABI), Mohali, an Autonomous Institute of the Department of Biotechnology, Government of India for funds and support. VS acknowledges the Council of Scientific and Industrial Research for JRF and SRF fellowships for PhD work and Indian Institute of Science Education and Research (IISER) Mohali for PhD program. We thank Dr. Prafull Salvi for providing vectors for BiFC experiments. We also acknowledge DeLCON (DBT-electronic library consortium), Gurgaon, India, for the online journal access.
Author contributions VS and JR conceived and designed the experiment, and manuscript writing. VS performed the experimental work and data analysis. VF, PK, AP, AM, MB helped in experimental work. AM and HS helped to manuscript writing. VR and SBS helped to draft the final manuscript. All authors read and approved the final manuscript.
Data availability Enquiries about data availability should be directed to the authors.

Declarations
Conflict of interest All authors declare that they have no conflict of interest.