Protein Involved in Initiation 1 Interaction With Starch Synthase 4 From Arabidopsis thaliana Induces Inhibition of Elongating Activity
Mélanie Bossu, Rayan Osman, Guillaume Brysbaert, Marc Ferdinand Lensink, David Dauvillée, Coralie Bompard
TL;DR
This study shows how a protein called PII1 interacts with SS4 to regulate starch synthesis in plants.
Contribution
The study reveals that PII1 specifically inhibits SS4's glucan elongation activity through a coiled-coil domain interaction.
Findings
PII1 interacts with SS4 via a coiled-coil domain.
This interaction inhibits SS4's glucan elongation activity.
The effect is specific to SS4 and not other synthases.
Abstract
Starch is the major energy storage compound in plants. It accumulates in the form of insoluble, partly crystalline granules whose number and shape are specific to each plant species. These characteristics are defined very early in starch biosynthesis, at the initiation stage. Starch biosynthesis initiation is a complex process that relies on the coordinated action of several proteins that interact together in the so‐called complex of initiation. Starch Synthase 4 (SS4) is the only initiation protein with enzymatic activity. It catalyzes the formation of glucan primers, which serve as substrates for the enzymatic machinery that synthesizes starch granules. Previous studies have highlighted the importance of interactions between SS4 and regulatory proteins in this process. Among them, Protein Involved in Initiation 1 (PII1) interacts with SS4 but its function is not yet established. In…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
FIGURE 1
FIGURE 2
FIGURE 3
FIGURE 4
FIGURE 5
FIGURE 6
FIGURE 7| Oligonucleotide Primer name | DNA Sequence 5′‐3′ | Amplicon sizes (bp) | Encoded sequence |
|---|---|---|---|
| EntPII1For | CACCCTGCCATACTCGAATCTTGGC | 1396 | 102‐564 |
| EntPII1Rev | TTAGCGGTTAGTGAGCTCTGCAATACG | PII1 H2 and H3 helices | |
| BamΔ349For | CCC | 2094 | 350‐1040 |
| XhoSS4Rev | CC | ||
| BamPII | GG | ||
| SalPII | GGG | 2250 | Full length PII1 devoid of TP |
| BamΔ466For | CCC | ||
| XhoSS4Rev | CC | 2058 | 467‐1040 |
- —Région Hauts‐de‐France10.13039/501100010095
- —Centre National de la Recherche Scientifique10.13039/501100004794
- —Université de Lille10.13039/100015872
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsFood composition and properties · Polysaccharides and Plant Cell Walls · Enzyme Production and Characterization
Introduction
1
In higher plants, starch is the most abundant and widely distributed non‐structural carbohydrate that accumulates as water‐insoluble, partly crystalline granules. In heterotrophic tissues, starch functions as a long‐term carbohydrate reserve, supporting germination or seasonal regrowth. This “storage starch” is a key component of our staple crops, accounting for half of human caloric intake and is also extensively used as an industrial commodity. In leaves, “transitory starch” accumulates in chloroplasts during the day and is used as a carbon and energy source during the night.
Starch is composed of two polymers of glucose residues, namely amylose and amylopectin, organized as linear α‐1,4‐glucans covalently linked to one another by α‐1,6‐bonds also called branching points (for review [1]). Amylose molecules are mainly linear (< 1% α‐1,6 bonds) whereas amylopectin, the main component contains 5%–6% of branching points [1, 2].
Amylose linear chains are synthesized by the action of the Granule Bound Starch Synthase (GBSS) that is embedded in starch granules and belongs to the starch synthase class of glucosyltransferases [1, 3, 4, 5]. A second non‐enzymatic protein, namely PROTEIN TARGETING TO STARCH 1 (PTST1), is also involved in the process by targeting GBSS to starch granules [6].
Amylopectin is synthesized by the concerted activities of soluble starch synthases (SSs), starch branching enzymes (SBEs), and starch debranching enzymes (SDBEs) acting by forming transient complexes during the different stages of biosynthesis [7].
While the mechanisms of the enzymes catalyzing the biosynthesis of starch polymers are well known [8], the initial processes at the origin of biosynthesis are still not well understood.
However, an initiation step is required to produce the oligosaccharide substrates of the biosynthetic enzymes. The nature and origin of the glucans that serve as substrates for the initiation mechanisms have not been clearly identified but several proteins involved in initiation have been identified. They all have one or more predicted coiled coil regions that indicate their propensity to interact and Arabidopsis plants that do not express these proteins show a phenotype in which the number of granules per chloroplast is reduced attesting to their implication in the initiation step [9].
The first of them is an enzyme from the SSs family, starch synthase 4 (SS4). Arabidopsis mutant plants lacking SS4 show a phenotype in which the number of starch granules formed in the chloroplasts is greatly reduced and the remaining granules are much larger and spherical, in contrast to the lenticular granules observed in the wild‐type plants [10]. This 1040 amino acids protein is composed of three distinct domains: an N‐terminal domain containing a transit peptide (1–42) and a predicted coiled‐coil (cc) region (43–465), a so‐called dimerization domain (471–515), and the catalytic domain at the C‐terminus (540–1040) holding the synthase activity. Like all SSs in plants (SS1 to SS4 and GBSS) SS4 display the conserved catalytic domain consisting of two Rossman‐like domains (Glycosyl Transferase domain or GT) connected by a hinge region. The N‐terminal and C‐terminal GT subdomains of SS4 belong to the GT5 and GT1 family (http://www.casy.org/), respectively. It has been proposed that the GT5 domain binds the acceptor substrate and the GT1 domain binds the donor substrate [11].
The N‐terminal domain, containing four predicted cc regions [12] was shown to be required for the granule shape, while the catalytic domain is sufficient to regulate granule number [9]. SS4 is the only protein in the initiation complex that has enzymatic activity. Among the other initiation proteins, there are two members of the PTST family: PTST2 and PTST3, two conserved coiled‐coil and CBM48‐containing proteins [13], the thylakoid‐associated MAR‐BINDING FILAMENT‐LIKE PROTEIN 1 (MFP1) [14], MYOSIN‐RESEMBLING CHLOROPLAST PROTEIN or PROTEIN INVOLVED IN INITIATION 1 (MRC/PII1) [15, 16] and a pseudo‐starch synthase, SS5 [17]. All initiation proteins (PII1/MRC, SS4, SS5, PTST2, PTST3, MFP1) contain one or more predicted variable‐length cc‐folded regions and disordered regions. These particular characteristics give them the potential to form protein–protein interactions and to be involved in conformational changes associated with a complex regulatory mechanism.
It has been proposed and shown that these proteins interact specifically, probably sequentially and in some cases transiently, as part of an initiation complex whose molecular mechanism is still poorly understood [15]. The function of some of these proteins has been investigated.
The first event in starch initiation is likely linked to the action of the MFP1 that indirectly interacts with SS4 via PTST2 and specifically determines the subchloroplastic location of the starch granule initiation machinery [14]. Besides SS4, PTST2 seems to play a central position in the initiation process in Arabidopsis (At) as it also interacts with PII1 and SS5 [15, 17]. Through its Carbohydrate Binding Module (CBM), it binds and delivers glucans, able to adopt a helical secondary structure, to SS4. The function of PTST3 is partially redundant with that of PTST2 [13]. PTST2 is also involved in the starch initiation process of several cereals' endosperms and potatoes by interacting with many other enzymes involved in starch biosynthesis initiation [13, 18, 19, 20, 21]. The non‐canonical starch synthase, SS5, which contains an incomplete catalytic domain containing only the GT5 subdomain has no enzymatic activity and interacts with PII1 [17]. It remains to be determined whether proteins with multiple partners are able to interact with them simultaneously via multiple recognition sites, or whether these interactions are in competition with each other as part of their functional regulation.
PII1 has a predicted coiled‐coil region that spans most of the protein without any predicted catalytic or glucan binding domains [15]. In Arabidopsis, knockout mutants have a phenotype analogous to ss4 mutant plants with one large lenticular (rather than spherical in ss4) granule per chloroplast instead of the normal 5–7 granules. This phenotype is similar to the ptst2 phenotype, but is less severe, as PII1 chloroplasts still contain one starch granule in most chloroplasts, whereas ptst2 and ss4 mutants have many empty chloroplasts. PII1 function remains unknown, particularly the role of its interaction with SS4 in the initiation process.
In this work we studied both the structure of PII1 and SS4 as well as their interaction. Through biochemical experiments using the recombinant truncated proteins, we were able to reveal an effect of PII1 on the elongation capacity of SS4. This unexpected result is discussed in terms of its potential implication on the regulation of starch granule initiation in vivo.
Results
2
Expression of PII1 and SS4
2.1
To investigate the role of PII1 in the initiation steps leading to new starch granules, we aimed to study its structure as well as its molecular interaction with SS4. For these studies, we considered a biochemical and structural approach. As these approaches required the expression and purification of both proteins, we undertook the expression of both proteins in their entirety, with the exception of their chloroplast transit peptides (cTPs). To do this, we carried out a series of expression assays on plasmids containing the genes encoding each protein, optimized for expression in E. coli (see Section 4). We succeeded in expressing PII1 under these conditions, but none of the tests allowed us to obtain enough soluble protein, while SS4 was little or not expressed. At this stage, and in order to continue the study, we considered using truncated but functional constructs of each protein (see further). To identify the regions that might affect expression or solubility, we analyzed the sequence of PII1 and SS4 and calculated molecular models using AlphaFold3.
We have primarily used this tool to gain a better understanding of the structural organization of initiation proteins, in particular SS4 and PII1, and to make progress towards understanding how they interact and, beyond that, towards studying their function.
Structural Analysis of PII1 Molecular Model
2.2
Previous sequence analysis of PII1 predicted the presence of coiled‐coil regions along almost the entire length of the protein [15, 16] suggesting that the protein is involved in interactions with one or more other proteins (Figure S1A). To visualize their arrangement, we then analyzed molecular models of the protein created by AlphaFold3 [22], see the Section 4 for molecular modeling.
For the full‐length PII1, the five models calculated by AlphaFold3 were very similar. For the sake of clarity, only the model with the best score is presented. The model obtained is composed of 6 helices predicted with a pLDDT > 90 confidence index and numbered H1 to H6 from the N‐terminus (Figure 1A). The 70 amino acids at the N‐terminus and the region between helices H3 and H4 are predicted to be disordered. Helices H2 and H3 are very long helices of 203 and 245 amino acids respectively, organized in antiparallel coils as suggested by the sequence analysis (Figure 1A). The analysis of the Predicted Aligned Error (PAE, Figure 1B) matrix shows that the relative position of all helices except H4, is predicted with a high confidence index.
Molecular model of PII1. The structures are represented as ribbon diagram. (A) AlphaFold3 model of PII1, regions with pLDDT > 90 are colored navy blue, regions with 90 > pLDDT > 70 are colored cyan, regions with pLDDT < 70 are colored yellow and orange. The six helices of the model are labeled H1 to H6 from the N‐terminal end of the protein. (B) PAE matrix of the model, the region corresponding to the six helices are labeled. (C) Surface potential of the coiled‐coiled region (H2‐h3) of PII1, the positive and negative charges at the surface are colored blue and red, respectively. The hydrophobic side chains of amino acids involved in the coiled‐coil stabilization are shown in black sticks.
The H2 and H3 helices interact with each other via hydrophobic interactions involving amino acids predicted to be involved in predicted repeated heptads for coiled‐coils (Figure 1C). On the other hand, numerous charged residues are directed towards the protein surface and may be involved in interactions with PII1 partners. The numerous hydrophilic residues present on the surface of the H2‐H3 domain of PII1, may create interactions with the solvent, increasing the chances of obtaining a soluble protein sample for its structural and functional analysis (Figure 1C).
The Coiled‐Coil Region of PII1 Is Conserved Among Orthologs
2.3
To verify the conservation of the predicted coiled‐coil (abbreviated cc in the rest of the manuscript) H2‐H3 region of PII1, we investigated whether it was conserved in its orthologs. Twenty‐eight orthologs were identified on the Arabidopsis information resource website (TAIR https://www.arabidopsis.org). All the orthologs are proteins close in size to AtPII1 (between 720 and 820 residues), with the exception of the Setaria italica protein, which has a much shorter sequence (492 residues). Their amino acid sequences were aligned using blastp [23] (Figure 2A, Figure S2).
Primary and tertiary structure alignment of AtPII1 with 28 orthologs. (A) Amino acid sequence alignment of PII1 with the 27 orthologs having the same length computed with Blastp. The sequence of Setaria italica has been manually added from the alignment of PII1 with all tested orthologs. Both alignments are detailed in Figure S2. Only alignment positions with no gaps are colored. Red indicates highly conserved positions and blue indicates lower conservation, gaps are represented by red lines and grey regions correspond to non‐conserved regions. The position of conserved helices is indicated by arrows and labeled below. (B) Structure alignment of H2‐H3 coiled‐coil region of PII1 with the 28 orthologs. Structures are represented as ribbon diagram, helices and loops are colored red and green respectively.
The result shows a highly conserved sequence for the 28 AtPII1 ortholog proteins analyzed, all along the sequence. Setaria italica has a much shorter sequence than the other orthologs, corresponding only to the N‐terminal region containing H1, H2 and H3 in AtPII1 (Figure 2A Figure S2).
As with AtPII1, we calculated a molecular model for each of the identified PII1 orthologs using AlphaFold3. With the exception of the Setaria ortholog, which contains only the H1 and the coiled‐coil domain, analysis of the results shows the presence of the 6 predicted helices similar to those predicted for the Arabidopsis protein (Figure 2A).
Alignment of the structures of H2 and H3 of PII1 and those of its 28 orthologs shows a high degree of structural conservation of the conserved cc region (Figure 2B), which is, in agreement with an important role for this region in the function of PII1.
Dimerization Domain Has a Crucial Role in SS4 Dimer Structure
2.4
The structure of the catalytic domain of AtSS4 has been solved by crystallography in the presence of ADP‐glucose and acarbose, a competitive inhibitor used in the treatment of type‐2 diabetes [24]. Analysis of the structure revealed a monomeric protein and has allowed the precise characterization of its catalytic domain.
Four cc regions (cc1 to cc4 numbered from the N‐terminus) have been identified in the N‐terminal part of SS4 followed by a so‐called dimerization domain upstream of the catalytic domain (Figure S1B). SS4 has been shown to be dimeric and that the dimeric form has higher activity than the monomeric form. It has been shown that the dimerization is linked to the presence of the dimerization domain, and that predicted cc regions are not involved in the dimerization [12]. The cc1 and cc2 regions are required for SS4 localization to the thylakoid. Interestingly, the authors produced a truncated construct of the cc1 and cc2 regions (comprising the first 349 amino acids of the protein), which proved to be soluble and active [12]. In order to obtain soluble (and active) protein samples, we decided to work with the same construct, which we will call SS4‐Δ349, for the remainder of this work.
The dimerization domain has been predicted to be located between amino acids 471 and 515 and its structure is not known [12]. In order to identify this domain, to verify the possibility of dimerization and to visualize the position of the different regions that make up SS4, we created dimer models of SS4 and SS4‐Δ349 using AlphaFold3.
The two models obtained show the regions containing the predicted catalytic and dimerization domains with a high level of confidence (pLDDT above 90 and between 70 and 90, respectively) and the analysis of the PAE matrix shows a good level of confidence in the relative positions of these domains (Figure S3A–D). The AlphaFold3 predicted catalytic domain of SS4 is almost identical to that of the X‐ray‐crystallographic structure [24] with a root mean square deviation (rmsd) of 0.3 Å for all Cα. On the AlphaFold3 model, SS4 has a predicted dimeric structure in which the dimerization domains of each monomer interact with the GT5 subunit of the catalytic domain of the second monomer sharing a large surface area (Figure 3A). The pattern of the dimerization domain, originally predicted to be between residues 471 and 515, is composed of residues 466 to 533 in the model. Its structure is predicted as a globular domain composed of a bundle of 4 helices (dH1 to dH4) parallel to each other and an N‐terminal region of about 7 amino acids forming a short helix (Figure 3B). The two dimerization domains are organized into swap domains, each of which interacts with the GT5 domain of the neighboring molecule (Figure 3A, Figure S3E,F). This interaction involves the helices dH1 and dH2 of the dimerization domain and the GT5 domain of the adjacent molecule described in Figure 3B,D. The interaction between the dimerization domain and the GT5 domain involves 21 residues per monomer and is described in Table S1 and covers a surface of 2200 Å^2^, suggesting a strong interaction. This interaction between the dimerization domain and GT5 is conserved in the full‐length or truncated structure of SS4 (Figure 3), which is a further vote of confidence in the already very convincing pLDDT and PAE matrix values (Figure S3C,D). In the SS4 full‐length model, interactions are predicted between the two dimerization domains by the short N‐terminal helix (Figure S3E). In the truncated model SS4‐Δ349, the interaction between the dimerization domains and the GT5 domains is identical to the full SS4 model (rmsd 0.144 Å for all Cα), with the same confidence indices and positioning errors of the same order (Figure S3D, Figure 3D). In contrast, the interaction between the dimerization domains observed in the SS4 is not conserved in the model of SS4‐Δ349 in which the two dimerization domains are not in the same position and are not directly interacting (Figure S3E,F). This difference in the predictions is unlikely due to the full or truncated version of the dimer to predict, but rather to the uncertainty in the positioning of the dimerization domains between each other, as shown in both PAE matrix plots. This could indicate that, in the context of conformational changes associated either with the binding/release of glucans or with interaction with another protein in the initiation complex, the interaction between the dimerization domain of one monomer and the GT5 domain of the adjacent monomer would remain stable, whereas the interaction between the two dimerization domains could change.
Molecular models of dimeric SS4 and SS4‐Δ349. The structures are represented as ribbon diagram. (A) Structural organization of SS4 full length as predicted by AlphaFold 3. Only one monomer is colored (catalytic domain in red, dimerization domain in green, cc3‐4 helix in violet and cc1‐2 helix in orange). The catalytic sites are indicated by a black star. The second monomer is colored gray. (B) Molecular model of the dimerization domain of SS4. The position of helices is labeled dH1 to dH4 from the N‐terminal. (C) Structural organization of SS4‐Δ349 with the same color code as SS4. (D) Interaction between the dimerization domain (in green) and GT5 (in red) domains of SS4. The side chains of amino‐acids residues involved in the interaction are represented in sticks and conserved residues are colored yellow.
N‐Terminal Domain of SS4 Organizes in Coiled‐Coil
2.5
The N‐terminal domain, although predicted as helices for the cc1 to cc4 regions, has different configurations depending on whether the whole protein or the truncated protein is considered (and on the position of the dimeric domains in relation to each other, Figure 3 and Figure S3).
In the five SS4 models generated by AlphaFold3 the structure of the catalytic and dimerization domains is conserved in contrast to the N‐terminal domain. The latter is always predicted to be organized as a coiled‐coil, but the distribution of helices, predicted with a lower confidence index than the C‐terminal domain, is quite different. In the best model, as expected, the 1–180 region is predicted disordered, while the cc1/cc2 and cc3/cc4 regions are involved in two distinct 200 Å long helices. The cc3/cc4 helices of each monomer are predicted with good confidence (pLDDT between 90 and 70) whereas the cc1/cc2 helices are predicted with low confidence (pLDDT < 70) (Figure S3A–D). Altogether, in the model they form a tetrameric coiled‐coil confirmed by the PAE. In this organization, the two cc3/cc4 regions form a central dimeric coiled‐coil around which the cc1 and cc2 regions are supercoiled (Figure S4A). It seems likely that AlphaFold positioned cc1 and cc2 on cc3/cc4 to stabilize the latter. cc1 and cc2 have been shown to be involved in (and essential for) the localization of SS4 to thylakoids. In this case, they must form a domain distinct from cc3/cc4 in order to interact with MFP1 or PTST2. It is therefore highly likely that they are not organized as in the model described.
The five SS4‐Δ349 AlphaFold models were also very similar. In these models the cc3 and cc4 regions form distinct helices forming an antiparallel cc that assembles with the cc3/cc4 region of the second molecule, whose folding is identically predicted, to form a confident tetrameric cc (90 > pLDDT > 70 and 70 > pLDDT > 50) folded with low expected position error (Figures S3C,D and S4B). This configuration is probably adopted to stabilize the hydrophobic regions of cc3/cc4 that interact with cc1/cc2 in the model of the whole protein. Like for the full‐length protein, this could indicate that individual cc regions are able to adapt their folding depending on the partner with which they interact. This would give SS4 the ability to adapt to multiple partners (individually or together) by adjusting the conformation of its cc regions.
SS5 Has a Similar Dimeric Organization as SS4
2.6
Since SS5 is a protein involved in the initiation mechanisms, and interacts with PII1, we studied its structure using AlphaFold3 [22]. SS5, like SS4, is predicted to be dimeric [17], so we started by modeling a dimer. The model obtained has a high confidence index, as shown by the pLDDTs per residue and the PAE matrix (Figure S5A,B).
The molecular model shows a dimeric organization of SS5, similar to that of SS4, involving two dimerization swapping domains (from residue 121 to 139, Figure S5C, Table S2). The dimerization domain of SS4 is not conserved in the SS5 sequence [17], however a dimerization domain is observed with a conserved structure when compared with SS4 (Figure S5D). The helix H1 which interacts with the GT5 domain of the other monomer is involved in the predicted long N‐terminal helices forming the coiled‐coil region. It has been shown that when the previously predicted cc region of SS5 (residues 68–120) is absent, the protein is neither able to form dimers nor to interact with PII1 [17]. This result is consistent with the molecular model of SS5, with the absence of the cc region likely affecting the structure of the dimerization domain.
The GT5 domain of SS5 is extended by a helix (439–460) positioned at the site occupied by the C‐terminal helix originating from the GT1 domain of SS4 (which is absent in SS5) and interacting with the GT5 domain of SS5 (Figure S5E).
The first 59 amino acids are predicted to be disordered and residues 60–139 of the two molecules form two 115 Å long helices forming a coiled‐coil as observed for SS4. These helices are longer than the previously predicted cc region in SS5 (residues 68 to 120) [17].
The model for SS5 is very similar, in terms of its N‐terminal, cc, dimerization and GT5 domains, with the addition of the terminal helix, to SS4 at the dimerization site. The GT5 domain, the dimerization domain and the Helix 439–460 of SS5 superpose to the equivalent structures in SS4 with an rmsd of 1 Å on 250 Cα (Figure S5E). Only these two SSs have this dimeric organization and both interact with PII1, so it is conceivable that the dimerization region could be involved in the interaction with PII1.
PII1 H2‐H3 Is Soluble and Organized in CC
2.7
To study PII1 and SS4, we made truncated constructs to obtain soluble samples compatible with biochemical and biophysical approaches. For PII1, we produced a protein construct containing only the H2‐H3 helices that make up the conserved cc region, hereafter referred to as PII1‐H2‐H3.
PII1‐H2‐H3 is expressed in large amounts, most of which is found in the soluble fraction (Figure S6). The soluble fraction was incubated with IMAC beads and eluted with 500 mM imidazole. At this stage, and despite incubation with the beads, only a few parts of the soluble sample were able to bind to the resin and eluted with imidazole. The fraction that did not bind was subjected to a second unsuccessful incubation with Ni‐beads, suggesting that in solution the protein must adopt multiple conformations or form soluble aggregates and that in this form the 6‐histidine tag is no longer accessible. The eluted fraction was then subjected to a steric exclusion chromatography (SEC) step. PII1‐H2‐H3 eluted in a very broad peak, probably due to its non‐globular structure, and the presence of the long cc. This type of chromatography is probably not suitable for non‐globular proteins and does not allow us to determine the monodispersity of the sample, but it did allow us to eliminate the contaminants present at the end of the affinity chromatography step. We therefore kept it in the protein purification protocol (Figure 4A).
Purification and SR‐CD analysis of PII1‐H2‐H3. (A) SDS‐PAGE of the protein sample of PII1‐H2‐H3 after purification, used for SR‐CD experiments. (B) SR‐CD spectrum of PII1‐H2‐H3 (solid black line) and PII1‐H2‐H3 with 50% of TFE (broken line).
To experimentally verify the coiled‐coil structure of PII1‐H2‐H3, we used a Synchrotron radiation circular dichroism (SR‐CD) study approach on the purified PII1‐H2‐H3 protein sample at 3 mg/mL. CD is mostly used to analyze the content of secondary structure elements in a protein and to distinguish between individual helices and helices involved in a coiled‐coil assembly, which is particularly relevant to our study. With a CD approach, a typical spectrum obtained from alpha helices shows a positive peak around 190 nm and two negative peaks at 208 and 222 nm, the latter mainly due to helices, the others also coming from other structural elements. For free helices, the magnitude of the 208 nm peak is lower than the one at 222 nm; the opposite is true for helices involved in a coiled‐coil [25, 26]. SR‐CD analysis of PII1‐H2‐H3 presents a typical helix‐folded protein profile (Figure 4B) with a positive peak at 190 nm and two negative peaks at 208 and 222 nm. The ratio of the ellipticity measured at 222 nm to that measured at 208 nm gives a value lower than 1, indicating and confirming the organization of the helices into cc.
To go further in the validation of the presence of cc regions, we measured SR‐CD spectra of PII1‐H2‐H3 in the presence of 50% of 2,2,2‐trifluoroethanol (TFE), which has the property of separating helices organized in cc and stabilizing them as individual helices. The results show that at 50% TFE concentration all helices are independent (Θ222 /Θ208 > 1) (Figure 4B) thus definitely confirming the structuration of helices in cc in the structure of PII1.
SS4‐Δ349 Is Soluble and Displays Enzymatic Activity
2.8
Since SS4 cannot be produced in its full form, we decided to express a truncated version of the protein for the biochemical assays. We based this on the work of [12], who expressed a soluble and active SS4 construct containing the cc3, cc4, dimerization and catalytic domains. The deleted part corresponds to the first 349 residues, so we named this construct SS4‐Δ349.
SS4‐Δ349 was expressed in small quantity but in soluble form (Figure S6). It was purified by an IMAC affinity chromatography step (Figure S6 and Figure 5A).
Purification and characterization of starch synthase elongation activity of SS4‐Δ349. (A) SDS‐PAGE of the protein sample of SS4‐Δ349 after purification. (B) Zymogram revealing the elongation function of SS4‐Δ349. One‐hundred fifty micrograms of total soluble protein extract of BL21 (DE3) and BL21 (DE3) expressing SS4‐Δ349 were loaded onto a 10% non‐denaturing acrylamide gel containing 0.3% oyster glycogen (Sigma). After migration, the gel was incubated overnight at room temperature in synthase incubation buffer containing 2.4 mM ADP‐glucose. Starch synthase activities were visualized as brown bands after iodine staining.
We then verified that the deletion of cc1/cc2 in SS4 did not affect the ability of the enzyme to elongate glucan chains in our sample. To do this, we performed zymograms consisting of the electrophoretic separation of soluble protein extracts of bacteria expressing SS4‐Δ349 (150 μg of total soluble proteins), under native conditions, in a 10% polyacrylamide gel containing 1% glycogen. After migration, we observed the capacity of the enzyme to elongate the outer chains of glycogen as revealed after iodine staining of the gels after incubation in the presence of ADP‐Glc. This activity is revealed by the appearance of a brown/black coloring band with iodine, while the unmodified glycogen chains stain orange/light brown. To verify that the measured activity was not due to a bacterial enzyme, we performed the same experiment with a soluble extract of the same bacterial strain that does not express SS4‐Δ349. (Figure 5B). Analysis of the results showed very low elongation activity for the BL21 strain likely corresponding to the E. coli glycogen synthase (GlgA) at the top of the gel. To simplify the process, we attempted to express SS4 in a BL21 strain that does not express GlgA, which is often used for this type of approach. However, the protein was not expressed in this strain. Nevertheless, we exploited the presence of GlgA to assess the specificity of PII1for SS4.
In contrast, a band of high elongation activity was observed in the strain expressing SS4‐Δ349, indicating the high activity of the enzyme in its truncated form. This activity is shared by most of the starch synthase isoforms targeted to the chloroplast. Studies using high‐performance anion exchange chromatography coupled to pulsed amperometric detection (HPAEC‐PAD) showed that all SSs are capable of extending glucan chains from maltose or glucans with a higher degree of polymerization in the presence of ADP‐glucose (DP) [27]. None of them, including SS4, could synthesize de novo glucans from ADP‐glucose and glucose [27]. We tried to perform the same experiment by replacing glycogen with a mix of small malto‐oligosaccharides (MOS), but we were unable to detect any iodine staining.
In order to verify that the dimerization domain of SS4 is involved in the interaction with PII1, we also made a truncated construct of SS4 consisting only of the catalytic domain and the dimerization domain. To do this, we deleted the first 466 residues and named the construct SS4‐Δ466. Like SS4‐Δ349, the expression level is very low, but we were able to identify it by its activity on the zymogram (see next paragraph and Figure 6).
Zymogram revealing PII1‐H2‐H3 effect on SS4‐Δ349 starch synthase activity. Total soluble protein extract (175 or 350 μg) of BL21 (DE3) or BL21 (DE3) expressing PII1‐H2‐H3 or SS4‐Δ349 or a mixture of both were loaded onto a 10% non‐denaturing acrylamide gel containing 1% oyster glycogen (Sigma). After migration, the gel on the left was incubated 24h at room temperature in synthase incubation buffer containing 2.4 mM ADP‐glucose and the gel on the right in synthase incubation buffer without ADP‐glucose. Starch synthase activities were visualized as brown bands after iodine staining.
PII1H2‐H3 Inhibits SS4‐Δ349 and SS4‐Δ466 Activity
2.9
In order to check if the interaction with PII1 had an effect on the glucan elongating activity of SS4, we mixed soluble extracts from bacteria in which PII1‐H2‐H3 and SS4‐Δ349 had been expressed. Since we could not measure the exact amount of each protein in the soluble extract, we first mixed an equivalent amount of total protein for each extract (175 and 350 μg, Figure 6). After incubation for 15 min, the mixtures were loaded onto a native polyacrylamide gel and the SS4 elongation activity was assessed as previously by iodine staining of the zymogram after incubation overnight with ADP‐Glc (Figure 6).
The ratio of each protein was estimated from the intensity of the PII1‐H2H3 and SS4‐Δ349 bands on SDS‐PAGE. This analysis showed that in the soluble extracts the amount of PII1 was about 60 times higher than that of SS4. This value should be treated with caution as the level of expression (especially for SS4) varies from one expression to another. For this experiment, we performed several controls in order to analyze the results unambiguously. First, we performed the same experiment without ADP‐glucose to check that the observed activity was not linked to another substrate (Figure 6). We then performed the experiment with soluble proteins from expression bacteria that were not transformed by the plasmid to check that the observed activity was due to truncated forms of SS4 and not to a bacterial protein (Figure 6). We did the same with bacteria expressing PII1‐H2‐H3 to verify that this protein doesn't have any enzymatic activity (Figure 6). Finally, to verify that the change in catalytic activity was specific to the presence of PII1 and not due to the presence of a higher protein concentration in the medium we checked that the activity of SS4‐Δ349 remains unaffected in the presence of an excess amount of Bovine Serum Albumin (BSA).
Analysis of the results shows that no elongating activity is observed in the absence of ADP‐glucose, confirming that this precursor is used as a substrate by truncated forms of SS4 and GlgA and that no effect of the presence of BSA could be observed on the activity of both SS4‐Δ349 and GlgA (Figure 6). PII1‐H2‐H3 does not display any elongation activity.
In the presence of PII1‐H2‐H3, a strong inhibition of SS4‐Δ349 activity is observed under the tested conditions (Figure 6). In the presence of PII1‐H2‐H3, truncated forms of SS4 lose their elongation activity whereas GlgA remains fully active regardless of the amount of PII1‐H2‐H3. This demonstrates the specificity of PII1‐H2‐H3's inhibitory effect on SS4.
This experiment shows that PII1‐H2‐H3 interacts specifically with SS4‐Δ349 and that this interaction inhibits SS4‐Δ349 glucan elongation activity.
The same experiment has been done with SS4‐Δ466 and showed that this activity is also inhibited in the presence of PII1‐H2‐H3, Figure S7.
The fact that the activity of SS4‐Δ466 is inhibited by PII1‐H2‐H3 shows indirectly that there is an interaction between the two proteins, suggesting that the N‐terminal cc domain of SS4 is not involved or at least not crucial for the interaction with PII1 and reinforcing our hypothesis that PII1 specifically targets the dimerization domain or dimeric organization of SSs.
Molecular Modeling of the Interaction Between SS4 Proteins Involved in the Initiation Complex
2.10
Numerous interactions between the proteins involved in initiation have been demonstrated and are summarized in Figure 7. PII1 can interact strongly with SS4, PTST2 and SS5 [15, 17]. PTST2 interacts strongly with PTST3 and MFP1 weaker with PII1 and transiently with the C‐terminal domain of SS4 through its CBM [13, 15]. PTST3 interacts only with PTST2. To visualize the interaction modes of these proteins, we attempted to model all possible combinations of interaction using AlphaFold3 [22]. We calculated 50 structure models with AlphaFold3 for each complex and analyzed the results very carefully. Unfortunately, we were unable to obtain any models that met our selection criteria (as described in Section 4). In most cases, we obtained models with reasonable pLDDT values, of the same order as the models obtained for the individual proteins, but the PAE matrices did not validate the positions of each molecule. Furthermore, of the 50 structure models calculated, most proposed different structures, sometimes even clashes between the different chains. The ipTM and pTM values never exceeded 0.35 and 0.4 respectively. This is probably due to several factors: most of the proteins involved in the initiation complex (all except the catalytic domain of SS4) have sections with very few homologous sequences, especially in the cc regions—regions with repeats that can be organized in a large number of structural variations—which makes the training and subsequent inference harder, especially when several partners are involved, knowing the variety of binding they can be involved in (see discussion). Furthermore, all the proteins have disordered regions, which reduce the prediction scores of the obtained models.
Summary of known interactions in the initiation complex of Arabidopsis thaliana. The confirmed direct interactions between proteins are indicated by a double black arrow.
Discussion
3
All the proteins involved in starch binding initiation have been proposed to interact with each other to form an initiation complex [15, 17]. This proposal was supported by interaction studies between the identified molecules and by the fact that all these proteins have more or less important regions of their structure predicted to be organized in coiled‐coils, known to be involved in protein–protein interactions [28].
As their physical properties, particularly their length and flexibility, have important structural and functional properties, the cc are widely found in biological processes and multimolecular complexes (for review see [28]). Long cc, act often as molecular spacers, either separating functional domains or scaffolding large macromolecular complexes. Some cc act simply as molecular spacers or facilitate oligomerization, while others have evolved the ability to communicate conformational changes along their length. Some of them are able to bind membranes, linking two compartments in the cell [29] conferring conformational plasticity. The coiled‐coil domain of myosin, which is similar to PII1, has been shown to act as a mechanosensor and to transmit structural changes along its chain [30].
At the present stage of our knowledge of initiation mechanisms, we know that all the proteins involved have one or more cc regions predicted on the basis of sequence, and we have information on interactions between proteins that have been shown experimentally. However, apart from the crystallographic structure of the catalytic domain of SS4 and the CBM48 domains of the PTSTs, there is no structural information and very little knowledge about the regions involved in the interactions. Nor do we know whether the initiation complex involves all the proteins at once, or whether the interactions are transient or sequential. It is also possible that certain partners in the initiation complex have not yet been identified by the approaches used to date.
In the present work we focused on the structural study of PII1 and SS4 combined with structural and biochemical approaches to progress in the understanding of the effect of the binding of PII1 on SS4's enzymatic activity.
PII1 Contains a Long Conserved Coiled‐Coil Domain
3.1
We studied the structure of PII1 using molecular modeling and showed that the protein is composed of 6 helices that are conserved in all identified orthologs found on the TAIR website. The conserved sequences also adopt a conserved 3D structure, in particular the long H2–H3 helices, which account for almost 60% of the amino acids in the protein. These two helices join together to form a long cc region, which is about 300–350 Å in length. The presence of cc was demonstrated experimentally by SR‐CD. We have shown that this domain is conserved in all tested orthologs and that it may play an important role in the function of PII1.
At this stage, it is difficult to envisage a function for PII1 solely on the basis of its structure. However, it has been shown that PII1 interacts with PTST2, SS5 and SS4, and it may be involved in regulating SS4 activity.
What Is the Advantage of Organizing SS4 and SS5 Into Dimers?
3.2
Both SS4 and SS5 form dimers and it has been shown that dimeric organization is needed for interaction with PII1 [12, 17]. We showed that a structurally conserved dimerization region located between the N‐terminal cc region and the C‐terminal catalytic domain is involved in the dimer formation. Our results show with high confidence that these dimerization regions form a small globular domain that interacts with the GT5 subunit of the SS4 catalytic domain or the SS5 GT5 subdomain.
This organization, in which each monomer exchanges its dimerization domain, interacts strongly with the GT5 domain of the second monomer, offers some interesting properties. Firstly, it allows the two GT5 domains of the catalytic domains to be kept close together, leaving the catalytic region (for SS4) or the glucan‐binding domain (for SS5) accessible. Furthermore, having two GT5 subdomains close together could also augment the glycan concentration locally. The dimerization domain, within a monomer, is linked via unstructured loops to both the GT5 subdomain of the catalytic domain and the cc region. Within the dimer, it interacts with the GT5 of the other monomer (Figure 3). This key position has the dual advantage of maintaining the dimer and conferring dynamic properties that can allow signals to be transmitted to the catalytic domain via conformational changes induced by the binding of other partners.
It is therefore conceivable that the binding of PTST2 and/or PII1 to SS4 could influence its catalytic activity in the initiation complex.
SS4 and SS5 Both Interact Specifically With PII1 Which Does Not Recognize Other Families of SSs
3.3
All SSs share a catalytic domain (for functional SSs) or at least the GT5 domain and non‐conserved N‐terminus domains. SS1 and SS2 have no predicted cc regions, in contrast to SS3, SS4 and SS5 [17]. In SS3 (Uniprot F4IAG2), there is no predicted dimerization domain; it is a carbohydrate‐binding module (CBM) that links the predicted coiled‐coil region to the catalytic domain of the enzyme, which is otherwise monomeric.
Only SS4 and SS5 interact with PII1. The specificity of their recognition by PII1 cannot be entirely related to the presence of cc regions, since SS3 is not recognized. Nor can it be due to the structure of GT5, which is present in all SSs. What distinguishes SS4 and SS5 from other SSs is the dimerization associated with the presence of a structurally conserved dimerization domain that interacts strongly with the GT5 domain and acts as a hinge with the cc region. Since this dimerization is necessary for recognition by PII1 [17], it is conceivable that the latter interacts with SS4 and SS5 specifically with this part of the molecule and (due to the structure of PII1) with the coiled‐coil regions of the SSs. Our results showing that SS4‐Δ466, which contains only the dimerization and the catalytic domains, interacts with PII1 and induces a big modification of its catalytic properties are consistent with this hypothesis. Dimerization could therefore be the prerogative of SSs involved in the initiation of starch biosynthesis. This also implies that PII1 interacts with either SS4 or SS5, but not both at the same time. In this case, SS5 could have a regulatory effect on the binding of PII1 to SS4, by mimicking the PII1 binding domain of SS4.
PII1 Binding Can Modify SS4 Enzymatic Activity in Solution
3.4
Apart from the fact that it interacts with SS4, PTST2, and SS5 and that its absence in the plant results in a phenotype in which the initiation of starch granule synthesis is affected, the role of PII1 is not known. In this work, we have shown that PII1 conserved cc interacts with truncated forms of SS4 and that this interaction inhibits the elongating activity of SS4. We have also shown that this inhibition is specific to SS4 and does not affect the activity of the E. coli glycogen synthase GlgA. When compared with SS4, GlgA contains only the catalytic domain (GT5 and GT1) that shares 31% sequence homology and whose 3D structure superposes the SS4 catalytic domain structure with an rmsd of 1.3 Å based on 344 residues (Cα). Since GlgA and SS4 have similar catalytic domains, this result indicates that PII1 probably does not interact directly with the catalytic domain of SS4, but rather via its dimerization region.
It is rather counterintuitive to observe the inhibition of SS4 by PII1, since the absence of PII1 in the plant leads to a reduction in initiation events. Nevertheless, it is possible that this inhibition of SS4 by PII1 is a transient process within a complex and highly dynamic mechanism involving several partners, which needs to be regulated in space and time. This process is essential for correct initiation, as shown by the phenotype of the PII1 mutant.
As shown by the structural analysis, the dimerization domain of SS4 is located at the interface between the cc region and the GT5 subdomain of the catalytic domain. It is likely that the binding of PII1 to SS4 induces a conformational change in the catalytic domain leading to its loss of activity maybe breaking the link between the two catalytic domains or separating both GT domains. However, the strength of this inhibition must be qualified in the absence of other initiation proteins, in particular PTST2, which binds to the catalytic domain of SS4 via its CBM48 domain [13].
This is the first time that modulation of SS4 enzymatic activity by an initiation protein has been observed. This result needs to be analyzed in the context of full‐length proteins, but the fact that two truncated SS4 constructs behave in the same way is a favorable argument. This result is difficult to interpret in the current context of knowledge, but has the advantage of opening doors to new avenues for analyzing the behavior of the initiation complex.
In conclusion, the fact that each of these molecules has one or more cc regions, giving them the ability to interact with each other, but also to transmit conformational change signals within the initiation complex itself, suggests a complex and highly regulated mechanism of which we currently know only the basics. Our current knowledge is based on analyses of the phenotypes of mutants in which the initiation proteins have been inactivated or overexpressed, and on the identification of partners, which shed light on the nature of the interactions but do not allow us to go any further. In this work, we have studied the structural and enzymatic aspects of the current state of knowledge, which have allowed us to make progress in our understanding of the mechanisms and also to put forward some new hypotheses. The difficulty of producing the key players in the initiation complex in solution, and the presence of long cc regions that tend to aggregate in the absence of partners, complicates conventional structural biology tools that require monodisperse solutions. Determining the stoichiometry of each complex formed in solution and using cryo‐electron microscopy to study them will undoubtedly be an asset in the study of these complex mechanisms. The development of AI tools for modeling molecular models of multi‐protein complexes, which is rapidly expanding, should help to predict the interactions between the different initiation proteins and greatly improve our understanding of these mechanisms.
Methods
4
Molecular Modeling
4.1
Protein structures of all PII1 and SS4 constructs were initially modeled using both AlphaFold2 combined through MassiveFold (ref here: https://www.researchsquare.com/article/rs‐4319486/v1) that allows going further by generating a large number of molecular models, all the neural network model versions available to date (v1, v2 and v3) [22]. Initially, three predictions per neural network model were generated for monomers, resulting in 15 predictions per monomeric protein, and five predictions per neural network model for protein complexes, resulting in 75 predictions for each. Massive sampling and increased recycling were also attempted for some complexes, pushing predictions until 600, but did not show substantially improved results in these cases, compared to a basic prediction run.
When this was available, we used AlphaFold3 for the calculation of new molecular models and compared them to the predictions already obtained by MassiveFold. The structure models obtained with both approaches being similar for our proteins, we only kept AlphaFold3 predictions for the sake of clarity of the manuscript.
AlphaFold 3 computes five predictions and sorts them by confidence score. Several criteria for the validity of the structure are available: the predicted Local Distance Difference Test (pLDDTs), the Predicted Aligned Error (PAE) matrix and, for the multimers, the TM coefficients. The pLDDTs indicate the confidence estimate per atom on a scale of 0–100, where a higher value indicates a higher confidence for each protein amino acid. The PAE matrix estimates the error in the relative position and orientation between two parts in the predicted structure. Higher values indicate higher predicted error and therefore lower confidence. Then, predicted Template Modeling scores (pTM) and the interface predicted template modeling score (ipTM) are both derived from a measurement called the template modeling (TM) score. pTM is an integrated measure of how well AlphaFold‐Multimer has predicted the overall structure of the complex. In contrast, ipTM measures the accuracy of the predicted relative positions of the subunits forming the protein–protein complex. Disordered regions and regions with low pLDDT scores may negatively impact the ipTM score even if the structure of the complex is predicted correctly. Both measure the accuracy of the entire structure [31, 32], for high confidence it should be > 0.5 and > 0.8, respectively. TM scores should be interpreted cautiously [22] as they can be influenced by several factors. For this study all TM scores were below the “trustable” values and this can be attributed to the presence of large cc regions as well as predicted unfolded regions in proteins. Indeed, in cc regions, local contacts can be inferred with a good score, which explains the pLDDT values > 0.80 in these regions. However, the contacts can be shifted between repeats in the prediction, which makes the process less straightforward, resulting in low pTM scores. Disordered regions are always hard to predict, with high flexibility, resulting in low pLDDT values in these regions and decreasing the pTM scores. Therefore, we remained cautious in interpreting the results and models in which the PAE matrix showed no credible interactions between proteins were considered invalid.
Structures and electronic surfaces were visualized using the PyMOL Molecular Graphic System, Schrödinger LLC.
Synchrotron Radiation Circular Dichroism
4.2
Synchrotron radiation circular dichroism (SR‐CD) spectra were measured at the DISCO beamline of the SOLEIL Synchrotron (Gif‐sur‐Yvette, France). The beam size of 4 × 4 mm and the photon flux per nm step of 2 × 10^10^ photons s^−1^ in the spectral band from 270 to 170 nm prevented radiation‐induced damage [33]. CD spectra were acquired using IGOR software (WaveMetrics). Before measurements, the molar elliptical extinction coefficient of Ammonium d‐10‐Camphorsulfonate Ammonium (CSA) was measured on the beamline and used as standard for calibration of all data measurements [34]. Protein and buffer spectra were collected consecutively and are the mean of 3 accumulations. The buffer baseline was then subtracted from the spectra and the data processing was conducted using CDToolX software [35].
PII1‐H2‐H3 protein solutions were deposited between 2 CaF_2_ coverslips with a path length of 10 (TFE) or 50 (protein alone) μm [36]. The influence of TFE on the structure of PII1 was studied by mixing the protein with TFE (at a final concentration of 50%) and then measuring the spectra under the same conditions as for the native protein after a 10‐min incubation. Spectra containing protein buffer and 50% TFE were subtracted from the protein/TFE spectra before CSA calibration.
Cloning, Expression, and Purification of Proteins
4.3
PII1‐H2‐H3, SS4‐Δ349, and SS4‐Δ466 were cloned and expressed in Escherichia coli as recombinant proteins lacking their N‐terminal transit peptides. The complete cDNAs encoding the full‐length PII1 and SS4 lacking their N‐terminal transit peptides, cloned into the pENT‐D‐Topo plasmid (Thermo Fisher, Rochester, USA) [16] were used as templates for the construction of the vectors expressing the different truncated forms of SS4 and PII1. A fragment of the PII1 cDNA corresponding to the H2 and H3 helices was amplified with primers EntPII1For and EntPII1Rev using the ultra‐high Kapa Hifi Hot Start polymerase (Roche Diagnostics, Austria) and the 1.4 kb PCR product was transferred into the pENT‐D‐Topo plasmid (Thermo Fisher, Rochester, USA). This entry vector was used to construct the pET300‐PII1 vector by using the Gateway technology into the Champion pET300 plasmid (Thermo Fisher) that allows the generation of recombinant proteins fused to an N‐terminal 6‐Histidine tag. The PCR products corresponding to the SS4‐Δ349 or SS4‐Δ466 truncated ss4 cDNA fragment were amplified using the Kapa Hifi Hot Start polymerase and digested by BamHI and XhoI restriction enzymes before their cloning into the pET‐DUET‐1 (Novagen) vector's corresponding restriction sites allowing the expression of N‐terminal His‐tagged proteins. For co‐expression experiments, the PII1 sequence was amplified using the BamPII and SalPII primers allowing the introduction of BamHI and SalI restriction sites, which were used to clone the digested PCR product into the pET‐DUET‐1.
Primers sequences are detailed in Table 1, the restriction enzyme sites used for the cloning are underlined in the primers' sequences. Each construct was fully sequenced before use to ascertain the integrity of the cloned sequences.
The recombinant proteins were expressed in E. coli BL21 DE3 in 2 × 500 mL of LB medium supplemented with 100 μg/mL antibiotic then induced with 0.5 mM IPTG overnight at 20°C.
Cells were pelleted at 6000 g for 30 min at 4°C and stored at −80°C for further use.
Cells pellets were resuspended in lysis buffer (20 mM Hepes/NaOH pH 7, 150 mM NaCl), supplemented with one tablet per liter of EDTA‐free protease inhibitors (EDTA‐free, Roche), disrupted by Emulsiflex or sonication and centrifuged at 10 000 g for 30 min at 4°C. The supernatants were then subjected to purification.
Purification was performed by a first step of Immobilized Metal Affinity Chromatography (IMAC) using a 5 mL IMAC HisTrap Excel column (Cytiva, Amersham, UK). Protein sample was loaded onto the column pre‐equilibrated with buffer A (20 mM Hepes/NaOH pH 7, 150 mM NaCl). The beads were washed with 10 column volume of buffer A supplemented with 10 mM Imidazole and eluted with buffer A supplemented with 500 mM Imidazole and 10% (w/v) glycerol.
This initial step was followed by a second purification step through size exclusion chromatography Superdex200 10/300 (Cytiva) equilibrated with buffer A. Five‐hundred microliters of the IMAC eluted sample were injected on the column. The protein sample purity was assessed by SDS‐PAGE 10%.
For structural study, protein samples were concentrated using Vivaspin centrifugal concentrator with a 10 kDa cut‐off (Sartorius). Protein concentrations were determined using a Nanodrop Spectrophotometer (ND1000) from Thermo Scientific.
Zymogram
4.4
To study the elongation capacity of SS4, soluble proteins were separated by electrophoresis onto a 10% non‐denaturing acrylamide gel containing 0.3% oyster glycogen (Sigma). The run was carried out for 2 h at 200 mA at 4°C using the Bio‐Rad Mini Protean III system. The gel was then incubated overnight at room temperature in synthase incubation mix (67 mM Glycyl‐glycine/NaOH pH 9; 133 mM ammonium sulfate; 80 mM MgCl_2_; 0.6 mg mL^−1^ BSA; 25 mM β‐mercaptoethanol; 2.4 mM ADP‐glucose). Control experiments were carried out following the same procedure but omitting the ADP‐glc in the incubation mix. To test the capacity of SS4 to synthesize de novo glucan primers, the same procedure was used but without providing any polysaccharide substrate in the gel matrix. The initiation capacity of the enzyme was tested in a control experiment in which a mix of small MOS was provided in the incubation mix.
Author Contributions
Mélanie Bossu: investigation, conceptualization, writing – review and editing. Rayan Osman: investigation. Guillaume Brysbaert: investigation, writing – review and editing. Marc Ferdinand Lensink: writing – review and editing, investigation. David Dauvillée: conceptualization, writing – review and editing, investigation. Coralie Bompard: conceptualization, investigation, writing – original draft, writing – review and editing, supervision.
Funding
This work was supported by Région Hauts‐de‐France, Centre National de la Recherche Scientifique, and Université de Lille.
Supporting information
Data S1: prot70089‐sup‐0001‐supinfo.pdf.
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1B. Pfister and S. C. Zeeman , “Formation of Starch in Plant Cells,” Cellular and Molecular Life Sciences 73, no. 14 (2016): 2781–2807.27166931 10.1007/s 00018-016-2250-x PMC 4919380 · doi ↗ · pubmed ↗
- 2S. Pérez and E. Bertoft , “The Molecular Structures of Starch Components and Their Contribution to the Architecture of Starch Granules: A Comprehensive Review,” Starch 62 (2010): 389–420.
- 3S. G. Ball and M. K. Morell , “From Bacterial Glycogen to Starch: Understanding the Biogenesis of the Plant Starch Granule,” Annual Review of Plant Biology 54 (2003): 207–233.10.1146/annurev.arplant.54.031902.13492714502990 · doi ↗ · pubmed ↗
- 4S. Schwarte , H. Brust , M. Steup , and R. Tiedemann , “Intraspecific Sequence Variation and Differential Expression in Starch Synthase Genes of Arabidopsis Thaliana ,” BMC Research Notes 6 (2013): 84.23497496 10.1186/1756-0500-6-84PMC 3608163 · doi ↗ · pubmed ↗
- 5D. Seung , “Amylose in Starch: Towards an Understanding of Biosynthesis, Structure and Function,” New Phytologist 228, no. 5 (2020): 1490–1504.32767769 10.1111/nph.16858 · doi ↗ · pubmed ↗
- 6D. Seung , S. Soyk , M. Coiro , B. A. Maier , S. Eicke , and S. C. Zeeman , “PROTEIN TARGETING TO STARCH Is Required for Localising GRANULE‐BOUND STARCH SYNTHASE to Starch Granules and for Normal Amylose Synthesis in Arabidopsis,” P Lo S Biology 13, no. 2 (2015): e 1002080.25710501 10.1371/journal.pbio.1002080 PMC 4339375 · doi ↗ · pubmed ↗
- 7N. Crofts , N. Abe , N. F. Oitome , et al., “Amylopectin Biosynthetic Enzymes From Developing Rice Seed Form Enzymatically Active Protein Complexes,” Journal of Experimental Botany 66, no. 15 (2015): 4469–4482.25979995 10.1093/jxb/erv 212PMC 4507757 · doi ↗ · pubmed ↗
- 8A. M. Smith and S. C. Zeeman , “Starch: A Flexible, Adaptable Carbon Store Coupled to Plant Growth,” Annual Review of Plant Biology 71 (2020): 217–245.10.1146/annurev-arplant-050718-10024132075407 · doi ↗ · pubmed ↗