Abstract
Amylopectin, the primary form of starch in plant leaves, seeds and tubers, features a tree-like architecture with branched glucose chains. Excess branches result in the formation of soluble phytoglycogen instead of starch granules. In higher plants and green algae, the debranching enzyme isoamylase ISA1 forms either homomultimer or hetero-multimer with ISA2 to facilitate branch trimming and starch granule formation, but the molecular basis remains largely unknown. In this study, we reconstitute the rice OsISA1-ISA2 complex in vitro and determine the cryo-EM structures of the OsISA1 homodimer, as well as the malto-oligosaccharide (MOS)-free and MOS-bound OsISA1-ISA2 heterocomplex. The OsISA1 dimer shows a tail-to-tail rod-like architecture, whereas the OsISA1-ISA2 complex mainly exhibits as a trimer, with OsISA2 flanking on the N-terminal segments of the dimeric OsISA1. Combined with comprehensive biochemical analyses, these structural data elucidate the organization of the ISA1-ISA2 heterocomplex in higher plants and demonstrate how ISA1 and ISA2 cooperate during amylopectin biosynthesis.
Introduction
Starch is the primary storage carbohydrate in plants, constituting a significant portion of the human diet and having numerous industrial applications1,2. Amylopectin, the predominant glucose polymer in starch, consists of α−1,4-linked glucan chains that are branched via α−1,6-glycosidic bonds, forming a racemose or tree-like architecture1,3. The branched structure enables adjacent chains to form double helices, which pack into crystalline lamellae interspersed with amorphous lamellae, resulting in a semi-crystalline matrix that confers starch its insoluble characteristics1,4,5. The analogous storage compound in animals, fungi and bacteria is glycogen. Like starch, glycogen also consists of α-1,4 and α-1,6-linked glucans. However, glycogen has a higher proportion of α-1,6 linkages and shorter α-1,4-linked chains. These structural differences prevent the formation of crystalline lamellae, rendering glycogen soluble3.
Amylopectin synthesis in plants involves several key enzymes that likely work simultaneously and synergistically to build its complex branched structure6,7. ADP-glucose pyrophosphorylase (AGPase) catalyzes the synthesis of ADP-glucose from glucose-1-phosphate and ATP, with ADP-glucose serving as the precursor for starch biosynthesis8,9. Granule-bound starch synthase (GBSS) and soluble starch synthase (SS) extend the α-1–4 linked glucan chains at the non-reducing end using ADP-glucose as the substrate10. Starch branching enzymes (SBE) introduce α-1,6-glycosidic linkages into the linear α-1,4-glucan chains to create branches11. Additionally, starch debranching enzymes (DBE) play a critical role by removing some of the excess branches from pre-amylopectin. Mutations that abolish or reduce DBE activity result in the replacement of starch granules with a soluble glycogen-like glucan called phytoglycogen12,13. Phytoglycogen is similar in composition to amylopectin but has a higher degree of branching, shorter chains, and closer branch points, which hinder crystallization and prevent the formation of starch granules3.
Plants have two types of DBEs: isoamylases (ISA) and pullulanase (PUL, also known as limit dextrinase), both of which hydrolyze α-1,6-glycosidic bonds at the branch points of starch. Higher plants possess three ISA isozymes (ISA1, ISA2, and ISA3) and a single PUL. Interestingly, the eukaryotes that make starch typically contain ISA proteins, whereas those that make glycogen do not, suggesting the evolutionarily conserved roles of ISA in starch metabolism. These DBEs play distinct yet complementary functions in starch metabolism, with ISA3 and PUL focusing on amylopectin degradation14,15,16, while ISA1 and ISA2 facilitate amylopectin synthesis17,18. Mutants of ISA1 lead to reduced insoluble granular starch in various species such as Arabidopsis16,18,19, potato17, rice20,21, maize22,23, barley24, and Chlamydomonas25,26,27, resulting in a notable accumulation of soluble phytoglycogen. In contrast, ISA2 mutants show more varied phenotypes. In Arabidopsis and potato, repression or mutation of the ISA2 gene cause the same phenotype as repression or mutation of ISA117,18,28,29. In maize and rice endosperm, the loss of ISA2 does not result in a comparable phenotype to the loss of ISA121,22. Maize isa2 null mutant exhibits comparable starch levels with no accumulation of phytoglycogen compared to the wild type, though more smaller granules are observed30. The rice isa2 suppressed line has the same plumped seeds as the wild type contrast to the isa1 suppressed lines that exhibit shriveled seeds21. Thus, in maize and rice endosperm, ISA1 is sufficient for normal starch synthesis.
The role of ISA in starch biosynthesis is to increase the frequency at which soluble α-polyglucan converts to a semi-crystalline form19,31. This is achieved by removing excess or improperly positioned branches that disrupt double helix formation and impede starch crystallization3,4,32,33,34. Both ISA1 and ISA2 belong to the glycoside hydrolase family 13 (GH13), which is known for its role in hydrolyzing glycosidic bonds in carbohydrates35. Despite their shared classification, ISA1 and ISA2 exhibit distinct biochemical properties due to differences in their enzymatic activities. ISA1 is enzymatically active and capable of hydrolyzing α-1,6-glycosidic bonds at branch points within amylopectin, whereas ISA2 is likely enzymatically inactive, possibly due to non-conservative amino acid substitutions in its active site22,29. ISA1 is thought to function by forming either homomultimers or hetero-multimers with ISA217,21,22,36. In Arabidopsis leaves and potato tubers, ISA1 and ISA2 work together in a heteromeric complex, with both proteins being essential for their activity and physiological function17,37. In contrast, rice and maize endosperms contain both ISA1 homomultimers and ISA1-ISA2 hetero-multimers, which exhibit three activity bands on native-PAGE containing amylopectin21,22,30,36,38. In these cases, ISA1 is a component of all three forms of ISA multimers. For ISA1 homomultimer, different oligomeric states have been reported across species. In rice and maize, ISA1 extracted from maturing endosperm has been suggested to form pentamers36 or mulitimers22 analyzed based on gel permeation chromatography. In contrast, recombinant maize ISA1 exists as a dimer when analyzed by AUC39. Similarly, the crystal structure of recombinant Chlamydomonas Reinhardtti ISA1 also resolves as a dimer40. For the hetero-multimer, rice endosperm ISA1-ISA2 is proposed to assemble into hexamers36. To date, the quaternary structure has only been determined for the ISA1 homodimer in green algae, while the architecture of the ISA1-ISA2 heteromultimer remains undetermined. In higher plants, the molecular mechanisms by which ISA1 form homomultimers or hetero-multimers with ISA2, and how these complexes trim excess branches to facilitate amylopectin synthesis, are still not fully understood.
In this study, we reconstituted the rice ISA1-ISA2 heterocomplex in vitro and determined the cryo-EM structures of three states: the ISA1 homodimer, the MOS-free ISA1-ISA2 heterotrimer and the MOS-bound heterotrimer. Combined with comprehensive biochemical analyses, these structures provide unprecedented insights into ISA1-ISA2 hetero-multimers assembly in higher plants, elucidating the molecular basis by which ISA2 interacts with ISA1 to promote branch trimming and facilitate amylopectin biosynthesis.
Results
The assembly and in vitro enzymatic activity of OsISA1-ISA2 heterocomplex
In higher plants, it is known that isoamylase ISA1 forms either homomultimers or hetero-multimers with ISA221,22,29,36,37. These complexes facilitate the conversion of soluble phytoglycogen into semi-crystalline starch granule by removing excess branches31. However, the molecular basis underlying the formation of these ISA oligomers remains poorly understood. To investigate this, we cloned rice OsISA1 (amino acids: 55–803) and OsISA2 (amino acids: 35–800), excluding their N-terminal transit peptides (predicted by TargetP), into modified pET15D vectors. OsISA1 was successfully expressed and purified to homogeneity by gel filtration (Fig. 1a), which behaved the same as native OsISA136. In contrast, the purification of His-tagged OsISA2 was challenging due to its instability and tendency to form inclusion bodies. To address this, we constructed OsISA2 with solubilization tags. Strep-Sumo-tagged OsISA2 showed slightly improved solubility, with a visible protein band on the SDS-PAGE gel. However, the yield of OsISA2 protein was significantly lower compared to that of OsISA1. To obtain sufficient protein, 15–20 litres of bacterial culture were amplified, and OsISA2 was purified alongside contaminating proteins after gel filtration (Supplementary Fig. 1a, b), consistent with observations from the investigation of ZmISA239. In this case, re-injection of the OsISA2 peak fraction reduced contamination (Fig. 1a). Co-incubation of His-OsISA1 and Strep-Sumo-OsISA2 resulted in the formation of an OsISA1-ISA2 complex, albeit with contaminating proteins (Supplementary Fig. 1). Meanwhile, we explored the interactions of OsISA1 and OsISA2 by co-expression. Co-expressing His-tagged OsISA1 and Sumo-tagged OsISA2 in 1 liter of BL21 (DE3) cells successfully yielded an OsISA1-ISA2 heterocomplex (Fig. 1a), as confirmed by co-elution and gel filtration analysis. Overall, both co-incubation and co-expression strategies resulted in the formation of the OsISA1-ISA2 complex with a similar stoichiometry of OsISA1:OsISA2. However, the co-expression approach proved significantly more efficient, yielding higher protein amounts and purity (Fig. 1a, Supplementary Fig. 1). These in vitro reconstituted OsISA1-ISA2 heterocomplex behaved the same as the native complex22,36.
a Gel filtration analysis of OsISA1, OsISA2 and OsISA1-ISA2 complex. Left, gel filtration chromatography. Right, SDS-PAGE corresponding to the chromatography. The star symbol indicates the contaminate Hsp70. b Analytical ultracentrifugation characterization of protein oligomerizations. The peak corresponded molecular masses are shown. c Enzymatic examination of OsISA1, OsISA2, and OsISA1-ISA2 complex using the BCA method. 4 nM OsISA1, 4 nM OsISA2, and 1 nM OsISA1-ISA2 was used in each reaction, respectively. The vertical ordinate shows the enzymatic activity in a unit of mg/s/μmol. mg, the amount of the reducing ends in the reactions; s, seconds of the reaction time; μmol, the enzymes in the reactions. At least three biological replicates were performed (n = 9; *p < 0.05; **p < 0.01; ***p < 0.001; unpaired t-test, error bars = mean ± SEM). d Native PAGE coupled enzymatic examination of OsISA1, OsISA2, and OsISA1-ISA2 complex. 50 nM protein was used for each lane. The native gel, containing 0.1% (w/v) maize amylopectin, is stained brown with I2-KI solution. The debranching activity of isoamylase is indicated by white bands. Blank and BSA lanes were set as negative controls.
Distinct elution volumes for the OsISA1-ISA2 heterocomplex, OsISA1, and OsISA2 indicate different oligomerization states (Fig. 1a) To validate this, we performed analytical ultracentrifugation (AUC) experiment, a powerful method for the quantitative analysis of molecular mass in solution based on sedimentation equilibrium. The results revealed major peaks for OsISA1, OsISA2, and the OsISA1-ISA2 complexes, with molecular masses of 170 kD, 87.8 kD, and 290 kD, respectively (Fig. 1b). Given that the theoretical monomeric form of OsISA1 and OsISA2 are 83 kD and 85 kD, respectively, the AUC results suggest that the major solution states of OsISA1 and OsISA2 are dimeric and monomeric, respectively, while the OsISA1-ISA2 complex may be heterotrimeric. The observation of dimeric OsISA1 is consistent with previous findings on maize ZmISA139. We also noticed that there were other minor peaks for the individual sample, suggesting that there may be other oligomeric states, but at low levels.
Nano-differential scanning fluorimetry (nanoDSF) revealed that the rice OsISA1-ISA2 complex exhibits greater thermal stability compared to OsISA1 and OsISA2 individually (Supplementary Fig. 2). This suggests that the coordination between OsISA1 and OsISA2 stabilizes the complex, potentially enhancing enzymatic activity or substrate specificity. To test this hypothesis, we assessed the enzymatic activities of OsISA1, OsISA2, and the OsISA1-ISA2 complex using amylopectin, amylose, and glycogen as substrates, employing the Bicinchoninic Acid (BCA) method29. The results demonstrated that OsISA1 was active against both amylopectin and glycogen but showed no activity on amylose, while OsISA2 exhibited no activity on any of the substrates (Supplementary Fig. 3). Notably, the OsISA1-ISA2 complex displayed a four-fold increase in enzymatic activity compared to OsISA1 alone when hydrolyzing amylopectin and glycogen (Fig. 1c, Supplementary Fig. 3). Furthermore, isoamylase activity was visualized on native gels containing amylopectin as the substrate. After electrophoresis, the gels were incubated with iodine solution, revealing clear bands corresponding to starch-degrading enzyme activity against a dark background. The results confirmed that OsISA1 and the OsISA1-ISA2 complex exhibited amylolytic activity, while OsISA2 was enzymatically inactive (Fig. 1d). The OsISA1-ISA2 heterocomplex produced longer, smeared bands compared to OsISA1 alone, indicating enhanced hydrolytic activity on amylopectin (Fig. 1d). These findings demonstrate that OsISA2 enhances the debranching activity of OsISA1 through their interaction, highlighting the functional significance of the heterocomplex formation.
Structure of OsISA1 homodimer
To elucidate the molecular architecture of OsISA1 and its role in debranching activity, we determined the atomic structure of dimeric OsISA1 using single particle cryo-electron microscopy (cryo-EM), achieving a reconstruction with a nominal resolution of 2.7 Å (Supplementary Fig. 4). Details of the structure data collection and refinement are provided in Supplementary Table S1. The OsISA1 structure revealed an extended dimeric architecture (Fig. 2a), measuring ~185 Å in width and 65 Å in height (Fig. 2b). The dimeric configuration of OsISA1 aligns with the results from AUC (Fig. 1b). The OsISA1 structure is organized into three domains: the N-terminal carbohydrate-binding module (CBM) ___domain (residues 86–222), the central α-amylase (AMY) ___domain (residues 223–690), and the C-terminal dimerization ___domain (residues 695–803) (Supplementary Fig. 5a). The OsISA1 dimer adopts a tail-to-tail configuration, stabilized by interactions between the C-terminal domains (Fig. 2b). The dimeric assembly of OsISA1 is reminiscent of the crystal structure of CrISA1 dimer from Chlamydomonas reinhardtii40. OsISA1 homodimer is structurally superimposable with the CrISA1 dimer, with an RMSD value of 0.879 Å for one protomer between 617 Cα atom pairs (Fig. 2c). A closer inspection of the C-terminal dimerization domains of OsISA1 and CrISA1 revealed good alignments of the β-sheets, with minor structural variation in the α-helix (Fig. 2d). Overall, the structural similarities between OsISA1 and CrISA1 highlight an evolutionarily conserved feature among ISA1 enzymes, spanning from unicellular green algae to multicellular higher plants.
a Cryo-EM density map of the dimeric OsISA1. The two protomers are colored in slate gray and slate blue, respectively. b Structural model of dimeric ISA1, shown in cartoon representation in the same orientation as in (a). N and C termini are indicated. c Structural superposition of OsISA1 with CrISA1. OsISA1 protomers are colored in slate gray and slate blue, and CrISA1 protomers are in wheat and cyan, respectively. d Structural superposition of the dimerization ___domain of OsISA1 and CrISA1. e Structural alignment of the C-terminal dimerization ___domain of OsISA1 protomers A and B. The secondary structural elements are labeled. f Interaction interface of OsISA1 protomers. Key residues are labeled and shown in stick representation. g Enzymatic analysis of OsISA1 mutants using native gel analysis. At least three biological replicates were performed.
Comparisons of OsISA1 protomer with debranching enzyme orthologs from Pseudomonas amyloderamosa IAM41, Sulfolobus solfataricus TreX42, Streptomyces venezuelae GlgX43, and E.coli GlgX44 revealed RMSD values of 0.99 Å, 1.06 Å, 1.09 Å, and 1.09 Å, respectively (Supplementary Fig. 5b–e). These results indicate overall structural similarities among isoamylases from diverse species. ISA1 and these orthologs belong to the α-amylase family, which contain thousands of members comprising of a similar C-terminal β-sandwich fold. However, to our knowledge, none of these orthologs have been reported to form an elongated tail-to-tail dimer. Thus, the C-termini-mediated dimerization observed in ISA1 represents a unique structural feature.
The C-terminal dimerization ___domain (DD) of OsISA1 is characterized by a five-stranded antiparallel β-sheet (β1β2β3β4β5) and a two-stranded β-sheet (β4β6), along with an additional α-helix (α1) situated between β5 and β6 (Fig. 2e). Structural alignment of the DD domains of OsISA1 protomers A and B reveals a highly similar structure, with an RMSD value of 0.208 Å (Fig. 2e). The α-helix and the proceeding long loops are critical for dimerization. The basic residue H773, located on the α-helix, forms a hydrogen bond with Y761 from the adjacent protomer. Additionally, a salt bridge is formed between H773 of protomer A and D770 of protomer B (Fig. 2f). Both Y761 and D770 are conserved in ISA1s among higher plants and are situated on the long loop. To assess the importance of these residues in dimerization, several mutants were generated. Single mutants (Y761A and H773A), and the double mutant (H773A/D770A) did not disrupt dimerization (Supplementary Fig. 6) and retained enzymatic activity comparable to the wild-type (Fig. 2g). We also created a triple mutant (ISA1H773A/D770A/Y761) and constructs with deletions of α1 (residues 770–785), the adjacent loop (residues 761–770), or the entire C-terminal dimerization ___domain (residues 680–803) to investigate their roles in OsISA1 stability and catalytic activity. However, only minor amounts of soluble protein were recovered for these constructs (Supplementary Fig. 6), and they exhibited a complete loss of amylopectin hydrolysis activity, underscoring the essential role of these segments in maintaining OsISA1 stability (Fig. 2g). In addition, β4 (residues: A737–G740) and β6 (residues: Y787–L790) are located on the opposite site of the dimer interface and may mediate interactions. The sequences of β4, β6, α1, and the preceding loops are relatively conserved (Supplementary Fig. 7), collectively contributing to the dimerization of ISA1.
The AMY ___domain features the characteristic (β/α)8-barrel motif, a common structural feature among α-amylase family members, consisting of eight parallel β-strands surrounded by eight parallel α-helices (Supplementary Fig. 5a). As a member of the glycoside hydrolase family 13, OsISA1 contains an essential catalytic triad – comprising of two aspartates and one glutamate – located at the bottom of the active-site cleft in the AMY ___domain. Specifically, the residues D432, E488, and D561 are positioned within this cleft (Supplementary Fig. 8a). Site-directed mutagenesis by substituting these residues with alanine, led to a complete loss of OsISA1’s catalytic activity (Supplementary Fig. 8b). Furthermore, native gel analysis using amylopectin as the substrate, followed by staining with I2-KI solution, confirmed that these mutations abolished amylolytic activity (Supplementary Fig. 8c). These findings highlight the critical roles of these acidic residues in enzyme catalysis.
Structure of OsISA1-ISA2 heterotrimer
To investigate the mechanistic basis for the enhanced debranching activity of OsISA1 mediated by catalytically inactive OsISA2, we determined the cryo-EM structure of the OsISA1-ISA2 complex at a nominal resolution of 2.4 Å (Supplementary Fig. 9). The complex exhibits a 2:1 stoichiometry (OsISA1:OsISA2), with a single OsISA2 protomer bound to an OsISA1 dimer (Fig. 3a, b). The overall structure displayed an approximate dimension of 230 Å in width and 90 Å in height (Fig. 3a, b). 2D classification predominantly identified a stable heterotrimeric configuration, while a minor population suggested a potential heterotetrameric state with two OsISA2 protomers flanking the OsISA1 dimer. However, insufficient particle numbers precluded high-resolution reconstruction of this putative tetramer (Supplementary Fig. 9), implying dynamic behavior of the complex. The quaternary structure of the OsISA1-ISA2 heterocomplex, as revealed in this study, has not been previously reported. AlphaFold 3 modeling corroborated our findings45, producing a trimeric model (2:1) with relatively high confidence scores - predicted template modeling (pTM) score of 0.74 and interface predicted template modeling (ipTM) score of 0.67, closely resembling our experimental structure (Supplementary Fig. 10a, b). A tetrameric model (2:2) was also generated (Supplementary Fig. 10c), which showed lower prediction reliability (pTM = 0.56; ipTM = 0.45) and exhibited significant conformational deviations when superimposed with the cryo-EM heterotrimer structure (Supplementary Fig. 10d).
a Cryo-EM density map of the ISA1-ISA2 heterocomplex. ISA1 protomers are colored in slate gray and slate blue, while ISA2 is shown in violet. b Structural model of OsISA1-ISA2 heterocomplex in cartoon representation. Two side orientations are shown. The N and C termini of OsISA2 are indicated. Domain organizations of OsISA2 are shown on top. TP target peptide, NTD N-terminal ___domain; CBM, Carbohydrate-binding module. AMY, amylase; CT, C-terminal region. c Interactions between ISA1 AMY and ISA2 CBMs. d Interactions between ISA1 AMY and ISA2 AMY. e Interactions between ISA1 CBM and ISA2 AMY. The involved residues are shown in sticks. f, g Key residues validated by co-expression and pull-down assays. The His-tagged OsISA1 and HA-tagged OsISA2 are validated by anti-His and anti-HA, respectively. Asterisks indicate the relative mutants of either OsISA1 or OsISA2. At least three biological replicates were performed.
OsISA2, also a member of glycoside hydrolase family 13, exhibits a conserved ___domain architecture with OsISA1, as evidenced by their structural alignment with an RMSD value of 0.996 Å (Supplementary Fig. 11a). However, OsISA2 possesses an additional N-terminal ___domain (NTD) located upstream of its CBM (Fig. 3b, Supplementary Fig. 11a). The NTD adopts a β-sandwich fold comprising eight β–strands, resembling a CBM but differs in its composition and architecture (Supplementary Fig. 11b). While the CBM domains of OsISA1 and OsISA2 (classified as CBM48)46 exhibit strong structural conservation (Supplementary Fig. 11c), the OsISA2-NTD shows significant divergence from the OsISA1-CBM (Supplementary Fig. 11d). This distinction is further supported by sequence analysis, which demonstrates low similarity between the OsISA2-NTD and typical CBMs of both OsISA1 and OsISA2 (Supplementary Fig. 11e). Further investigation is needed to elucidate the biological role of this unique ___domain in starch metabolism.
Overall, ISA1s and ISA2s share relative conserved sequences (Supplementary Fig. 7), but their AMY domains exhibit lower conservation. In OsISA2, the potential catalytic residues are I458, D495, and S566, which correspond to D432, E488, and D561 in OsISA1, respectively (Supplementary Fig. 11f). To determine whether these differences affect enzymatic activity, we generated an OsISA2 triple mutant (I458D/D495E/S566D) designed to mimic the catalytic triad of OsISA1. However, the OsISA2 mutant still showed no detectable activity in native gels containing amylopectin (Supplementary Fig. 11g), suggesting that additional structural or sequence features are essential for catalysis. These results support a nonenzymatic role for OsISA2, where it likely assists ISA1 in facilitating amylopectin synthesis.
The interaction surface between OsISA1 and OsISA2
In the OsISA1-ISA2 complex, OsISA2 forms extensive contacts with OsISA1, encompassing a buried surface area of 1943 Å2. These interactions primarily involve the N-terminal carbohydrate-binding module (CBM) and middle amylase (AMY) domains of both proteins, while their C-terminal domains remain unengaged (Fig. 3b). Three distinct interaction regions have been identified (Fig. 3c–e).
In interface 1, residues from a long loop connecting the CBM and AMY domains of OsISA1 interacts with the N-terminal CBM segments of OsISA2 (Fig. 3c). Specifically, the side chain of R239 in OsISA1 forms hydrogen bonds with the backbones of residues P200, Y201, and R204 in OsISA2’s long loop. Additionally, the backbones of residues G234, P237, L238, and K349 in OsISA1 interact with R41, Y201, H134, and R43 in OsISA2, respectively. Furthermore, Q233 in OsISA1 forms a hydrogen bond with R41 in OsISA2. To assess the significance of these residues, we performed truncation and mutation experiments on OsISA1 and OsISA2. Deletions of residues 237-239 or 233-239 in OsISA1 (designated as OsISA1Δ237-239 and OsISA1Δ233-239) significantly reduced its interaction with OsISA2, highlighting the importance of residues 233-239 in OsISA1’s loop. Similarly, loop deletion and mutation in OsISA2 (OsISA2Δ200-204 and OsISA2R41A/R43A) resulted in decreased interaction with OsISA1 (Fig. 3f, g). These findings underscore the critical role of this interface in mediating the interaction between OsISA1 and OsISA2.
In interface 2, R410 of OsISA1 forms a salt bridge with E479 of OsISA2 (Fig. 3d). Additionally, the backbone of residues C441, L443, and Y450 in OsISA1 form hydrogen bonds with R474, W512, and T552 in OsISA2, respectively. The side chain of Y450 in OsISA1 also forms a hydrogen bond with the backbone of P508 in OsISA2 (Fig. 3d). Residues C441 to Y450 of OsISA1 are situated in a long loop within the AMY ___domain. Deletion of this segment (Δ441-450) has minimal impact on the interaction between OsISA1 and OsISA2 (Fig. 3f). Similarly, the double mutant OsISA2R474A/E479A shows weakened interaction with OsISA1.
In interface 3, the backbone of L149 and N155 in OsISA1 interact with residues R278 and P485 in OsISA2, respectively (Fig. 3e). Residues L149 to N155 are part of a loop within the N-terminal CBM of OsISA1. Deletion of these residues (Δ149-155) affects the protein’s fold and stability, thereby impairing the interaction between OsISA1 and OsISA2 (Fig. 3f). In contrast, alanine substitution of R278 in OsISA2 has little effect on the interaction (Fig. 3g).
Structure of oligosaccharide-bound OsISA1-ISA2 heterocomplex
To elucidate substrate binding and catalytic mechanisms, we incubated the OsISA1-ISA2 heterocomplex with maltopentaose, maltoheptaose, or amylopectin at 4 °C overnight, followed by cryo-EM grid preparation and data collection. After intensive data processing, ligand densities were only observed in the amylopectin-incubated OsISA1-ISA2 heterocomplex but not in maltopentaose- and maltoheptaose-incubated samples. Given the known amylolytic activity of OsISA1-ISA2 in cleaving α-1,6-glycosidic bonds to generate amylose chains or MOS (Fig. 1c, d), we thus captured a product-bound state. We determined the cryo-EM structure of MOS-bound complex at 2.7 Å resolution (Fig. 4a; Supplementary Fig. 12). Cryo-EM analysis demonstrated that the complex predominantly exists as a trimer, with minor populations of dimeric OsISA1 and tetrameric OsISA1-ISA2 (Supplementary Fig. 12). This oligomeric heterogeneity suggests dynamic behavior of the OsISA1-ISA2 complex, potentially reflecting its dissociation.
a Cryo-EM density map and surface representation of MOS-bound OsISA1-ISA2 heterocomplex. OsISA1 protomers are colored in yellow green and royal blue, OsISA2 in violet, and MOS in orange. b Structural superposition of MOS-free and MOS-bound OsISA1-ISA2 heterocomplex. MOS-free OsISA1-ISA2 is colored in light gray. c Structural superposition of MOS-bound OsISA1-ISA2 heterocomplex with MOS-bound CrISA1 dimer. MOS-bound CrISA1 dimer is colored in cyan. d The interaction networks of MOS in the catalytic pocket of OsISA1. The maltoheptaose and the free glucose molecules are labeled. Surrounding residues within 4 Å are shown in sticks. The catalytic triad are shown in green sticks. e, f Maltotriose-binding cleft within OsISA1. g Four glucose molecules are shown in the C-terminal segment of OsISA2. Surrounding residues within 4 Å are shown in sticks.
The cryo-EM structure revealed three distinct MOS-binding subsites in each OsISA1 protomer, but only one in OsISA2 (Fig. 4a). Comparative structural analysis with the ligand-free complex showed moderate global flexibility (RMSD = 2.4 Å over 2064 residues), with conformational changes predominantly occurring in loop regions and terminal segments of both subunits (Fig. 4b). Notably, structural alignment with the previously reported maltoheptaose-bound CrISA1 (PDB: 4OKD)40 revealed an RMSD value of 0.738 Å for one ISA1 protomer over 636 residues (Fig. 4c), demonstrating striking evolutionary conservation. The oligosaccharides occupied equivalent positions across all three binding sites (Fig. 4c), indicating evolutionary conservation of both the substrate binding mode and catalytic mechanism among these enzymes.
Structural analysis revealed distinct MOS-binding patterns in OsISA1 and OsISA2. In OsISA1, MOS molecules were observed in the AMY ___domain and loop regions that connect AMY and C-terminal DD (Fig. 4a). The AMY ___domain accommodated a maltoheptaose (G1-G7) molecule within a narrow substrate-binding cleft, while three additional glucose units adjacent to the catalytic triad were observed (Fig. 4d). Two maltotriose are present between the AMY and C-terminal DD (Fig. 4a, e, f). These MOS molecules are surrounded by a network of polar amino acids (Fig. 4d–f). In contrast, OsISA2 showed a simpler binding pattern, with only four glucose molecules localized at its C-terminal segments (Fig. 4g). No additional MOS molecules were observed elsewhere in OsISA2.
Molecular basis by which OsISA2 increases the amylolytic activity of OsISA1
Given that OsISA2 enhances the enzymatic activity of OsISA1, we investigated whether OsISA2 induces a conformational change in OsISA1. Structural superposition of MOS-free and MOS-bound OsISA1-ISA2 complexes with the OsISA1 homodimer showed minimal conformational changes in OsISA1 (Supplementary Fig. 13a, b). Crucially, the catalytic triad geometry of ISA1 remained similar (Supplementary Fig. 13c, d), indicating that the increased amylolytic activity of OsISA1 upon OsISA2 binding is not due to conformational alterations in OsISA1.
We then assessed the amylopectin-binding activities of OsISA1 and OsISA2. Both OsISA1 and OsISA2 contain CBM, AMY, and the C-terminal ___domain (termed CT in OsISA2, or DD in OsISA1). To evaluate the amylopectin-binding contribution, each segment of the protein was constructed (Fig. 5a). We purified the OsISA1CBM to homogeneity, but failed to obtain the OsISA1AMY and OsISA1AMY-DD (Fig. 5a). OsISA1 displayed weak amylopectin-binding activity, and the catalytic-dead mutant OsISA1D561A exhibited comparable binding activity to the WT (Fig. 5b). OsISA1CBM possessed no amylopectin-binding activity, suggesting the minimal amylopectin-binding activity of OsISA1 largely contributed by its AMY-DD segments, as observed by the structure of oligosaccharide-bound OsISA1-ISA2 heterocomplex (Fig. 4a). Contrary to OsISA1, the individual domains of OsISA2 were almost obtainable except the AMY ___domain (Fig. 5a). OsISA2 exhibited strong amylopectin-binding activity (Fig. 5c). Without NTD and CBM segments, OsISA2ΔNTD and OsISA2AMY-CT still possessed comparable amylopectin-binding activity to the wild type OsISA2. Both NTD and CBM had amylopectin-binding activity, also did the CT ___domain (Fig. 5c). Thus, the individual NTD, CBM, AMY and CT domains all contributed to the amylopectin-binding activity of OsISA2 (Fig. 5c). The oligosaccharide-binding activity of OsISA2CT was consistent with the structural observation (Fig. 4a). Meanwhile, the above protein samples incubated with Glutathione Sepharose 4B beads revealed no detectable binding activity (Supplementary Fig. 14), confirming their binding specificity against amylopectin. Collectively, these findings imply that the homologous OsISA1 and OsISA2 have evolved with divergent biochemical characters.
a Schematic ___domain of OsISA1, OsISA2 and related protein constructs. OsISA1 and related truncations or mutations are fused with a N-terminal 6×His and a C-terminal myc-tag. OsISA2 and related truncations are constructed with a N-terminal Sumo and a C-terminal Strep-tag. Amylopectin-binding activity of OsISA1 (b) and OsISA2 (c) analyzed by SDS-PAGE and Western blot validation. Bands are categorized as input (In), supernatant (S), wash (W), and pellet (P). The amylopectin-bound ratio (P/In) for each protein is calculated and shown below the gel, with band intensities estimated using ImageJ. 30 nM protein and 60 mg amylopectin were used in each assay. Asterisks indicate the position of the target protein. d OsISA2 enhances OsISA1’s debranching activity as analyzed by native PAGE containing amylopectin. Increasing amounts of OsISA2 result in sharper retardation and more pronounced smearing bands by the formation of OsISA1-ISA2 hetero-multimers. Equal amounts of OsISA1 (50 nM) were used in lanes 2–5. At least three biological replicates were performed.
To explore their coordinated roles in starch metabolism, we investigated the amylolytic activity of OsISA1 in the presence of OsISA2. The results showed that OsISA1’s debranching activity increased progressively with higher amounts of OsISA2 (Fig. 5d). Moreover, the addition of increasing amount of OsISA2 led to the formation of two kinds of OsISA1-ISA2 heterocomplexes that exhibit amylolytic activity, similar to the in vivo investigations22,36,39. From abovementioned structural and biochemical investigations, we proposed that the presence of OsISA2 might modify the substrate-binding affinity and catalytic efficiency of OsISA1, collectively facilitating the phase transition of soluble phytoglycogen into semi-crystalline amylopectin biosynthesis.
Evolutionary conservation of the ISA1-ISA2 heterocomplex from maize, potato and Arabidopsis
The isoamylase ISA1 and ISA2 are evolutionarily conserved in higher plants, exhibiting high protein sequence similarities (Supplementary Fig. 7). The presence of ISA1-ISA2 hetero-multimers had been reported from several species including rice36, maize22,30,39, potato17,29, and Arabidopsis, revealing the functional roles of the isoamylase heterocomplex. To further corroborate the assembly of ISA1-ISA2 heterocomplex in higher plants, we tried to purify the individual ISA1 and ISA2 members from maize, potato, and Arabidopsis. The expression and purification of ZmISA2 and AtISA2 in E.coli was hindered due to the extremely low yield, whereas StISA2 exhibited improved soluble character with higher protein amount (Supplementary Fig. 15). Co-expression of ISA1 with ISA2 resulted in the formation of ZmISA1-ISA2, AtISA1-ISA2, and StISA1-ISA2 complex (Fig. 6a–c). Our in vitro investigation revealed that the presence of ISA1 could improve the expression or stability of ISA2 in rice, maize, and Arabidopsis, respectively. This is consistent to the in vivo investigation that the protein levels of ISA2 is strikingly lower in the isa1 mutant than the wildtype21,22,30,37.
Gel filtration analysis of ISA1, ISA2, and ISA1-ISA2 heterocomplex of maize (a), Arabidopsis (b) and potato (c). Specific tags and boundaries are shown on top. The signal peptide was removed for protein expression. d–f Protein mass weight examined by analytical ultracentrifugation characterization. The values of molecular weight are shown. g, h Protein oligomerization examined by native PAGE analysis. The left panel indicates samples visualized by Coomassie blue staining. The right panel indicates samples visualized by silver solution staining. A total of 2.5 μL 50 nM protein was used for each sample. i Enzymatic activity of ISA1, ISA2, and ISA1-ISA2 complex of maize, Arabidopsis and potato. 4 nM ISA1, 4 nM ISA2, and 1 nM ISA1-ISA2 complex were used. At least three biological replicates were performed. Data are presented as mean values +/− SEM (n = 9; *p < 0.05; **p < 0.01; ***p < 0.001; unpaired t-test, error bars = mean ± SEM).
AUC experiments were performed to examine the oligomeric states of ISA1 and ISA1-ISA2 complex. The results revealed molecular masses of 167 kDa, 161 kDa, and 155 kDa for ZmISA1, AtISA1, and StISA1 (Fig. 6d–f), respectively, suggesting the dimeric states of these ISA1s, which are consistent with the results of OsISA1 by cryo-EM and AUC in this study, ZmISA1 by AUC39 and CrISA1 by crystal structure determination40. For the heterocomplex, ZmISA1-ISA2, AtISA1-ISA2, and StISA1-ISA2 exhibited major peaks of molecular weights of 261 kDa, 292 kDa, and 305 kDa (Fig. 6d–f), respectively, indicating their trimeric states. Meanwhile a minor peak corresponding to 574 kDa, 531 kDa, and 524 kDa were also observed (Fig. 6d–f), indicating a higher oligomeric state but with lower abundance. These results are consistent to the investigation of OsISA1-ISA2 by AUC (Fig. 1b) and cryo-EM structure determination (Figs. 3 and 4).
The oligomeric states of ISA1 and ISA1-ISA2 heterocomplex were also examined by native PAGE analysis. ISA1 showed a single band from all four species, whereas the ISA1-ISA2 heterocomplex exhibited three different bands from rice, maize, and Arabidopsis (Fig. 6g, h), consistent with the 2D classification and ab initial model (Supplementary Fig. 12a, b), as well as the previous in vivo investigations from rice21,36, maize22,30,39, and Arabidopsis16,18,37. The lower weak band corresponds to the ISA1 homodimer, whereas the upper bands represent two types of hetero-multimers, suggesting the dynamic characters of ISA1-ISA2 heterocomplex that higher-order hetero-multimers could dissociate into lower-order hetero-multimer and ISA1 homodimer39. Contrast to rice, maize, and Arabidopsis ISA1-ISA2 heterocomplex that may both dissociate into homodimer and hetero-multimers on the native PAGE, potato StISA1-ISA2 heterocomplex exists mainly as hetero-multimers but no StISA1 homodimers (Fig. 6h), consistent to the in vivo investigations21,29. Moreover, it seems that the StISA1-ISA2 heterocomplex exhibits higher oligomeric states, thus, further structural determination of StISA1-ISA2 hetero-multimers is needed to figure out whether there are higher-order oligomeric structures other than trimer or tetramer.
Enzymatic assays revealed that ZmISA1 exhibit some debranching activity (Fig. 6i), consistent to the results of OsISA1 (Fig. 1c, d). In contrast, AtISA1 and StISA1 showed no detectable activity (Fig. 6i). Strikingly, the ISA1-ISA2 heterocomplex in all tested species displayed significantly stronger enzymatic activity compared to ISA1 alone (Fig. 6i), underscoring the coordinated role of ISA1 and ISA2 in amylopectin biosynthesis. In summary, our structural and biochemical analyses of the ISA1-ISA2 heterocomplex across rice, maize, potato, and Arabidopsis revealed evolutionary conservation of the heterocomplex that drives the phase transition of soluble phytoglycogen into semi-crystalline amylopectin, a critical step in starch biosynthesis.
Discussion
In this study, we reconstituted the OsISA1-ISA2 heterocomplex in vitro, and determined the cryo-EM structures of the OsISA1 dimer, as well as MOS-free and MOS-bound OsISA1-ISA2 heterotrimers (Figs. 2–4). These high-resolution three-dimensional structures reveal the molecular architecture of ISA homo- and hetero-multimers in higher plants. Structural prediction using AlphaFold 3 yielded a spatial architecture of the OsISA1-ISA2 heterotrimer that closely matched our experimentally determined cryo-EM structure (Supplementary Fig. 10a, b), supporting the robustness of AlphaFold 3’s predictive capabilities. Further analysis by 2D classification and ab initio reconstruction suggested the presence of a heterotetrameric form, though limited particle numbers prevented high-resolution reconstruction (Supplementary Figs. 9, 12). Nevertheless, the formation of ISA1-ISA2 heterotetramers in vivo appears physiologically reasonable. Consistent with prior in vivo studies16,18,21,22,30,36, our in vitro native PAGE analysis revealed two distinct ISA1-ISA2 heterocomplexes, likely a heterotrimer and heterotetramer, in approximately equal proportions in rice, maize, and Arabidopsis (Fig. 6). Interestingly, the PAGE matrix appeared to increase tetrameric assembly compared to solution conditions, possibly due to confinement effects. These observations highlight the dynamic nature of the ISA1-ISA2 complex, with the heterotrimer representing a major, structurally resolved form. AlphaFold 3 generated a putative heterotetrameric model in which each OsISA2 subunit flanks the N-terminal segment of an OsISA1 protomer (Supplementary Fig. 10c). However, the modest ipTM scores for both trimeric (ipTM = 0.67) and tetrameric (ipTM = 0.45) assemblies fell below the high-confidence threshold (ipTM > 0.8)45, emphasizing the need for experimental validation when interpreting such predictions. Collectively, our integrated approach, combining cryo-EM structural determination, AlphaFold 3 modeling, and biochemical validations, establishes a robust architectural framework for the ISA1-ISA2 heterocomplex in higher plants.
The structural findings provide mechanistic insights into its assembly and functional regulation of amylopectin biosynthesis. A previous study of OsISA1 and OsISA1-ISA2 from rice endosperm revealed a molecular mass of 420–480 kDa for OsISA1 homo-oligomer, and 510–550 kDa for OsISA1-ISA2 hetero-multimer by gel filtration chromatography, suggesting pentamers of OsISA1 and hexamers of OsISA1-ISA236. The different results obtained by our study and the previous investigation is largely due to the different methods used. Similarly, the molecular mass of ZmISA1 was also larger examined by gel filtration (~300 kDa)22 than by AUC analysis (~160 kDa)39. It was suggested that the gel filtration method separates proteins based on hydrodynamic volume, which is more accurate for the estimation of globular proteins rather than elongated or rod-shaped proteins like ISA1 or ISA1-ISA2 heterocomplex39.
In rice and maize, both the enzymatic-active ISA1 homodimer and ISA1-ISA2 hetero-multimers are present22,47. ISA1 and ISA2 exhibit different expression patterns. ISA1 is strikingly higher expressed than ISA2 in the endosperm tissues22, and thus the major functional form maybe ISA1 homodimer that is sufficient for the storage starch synthesis. In contrast, ISA2 exhibits higher expression levels than ISA1 in the maize leaf30, and only ISA1-ISA2 hetero-multimers are detected in rice leaf that are responsible for the transitory starch synthesis21. It is reasonable that the leaves accumulate and degrade starch rapidly in a diurnal rhythm, and the presence of ISA1-ISA2 hetero-multimers might efficiently convert the phase transition of soluble phytoglycogen into semi-crystalline granules. The spatio-temporal gene expression profile from RiceXPro database (https://ricexpro.dna.affrc.go.jp/category-select.php) also revealed high expression levels of OsISA1 in ovary and endosperms but low levels in other tissues, on the contrary, OsISA2 are widely expressed except the endosperm (Supplementary Fig. 16). These results imply the ISA1-ISA2 hetero-multimers may function at other plant developmental stages in addition to the endosperm maturation.
Amylolytic activity revealed different enzymatic characters of the homologous ISA1s. In rice and maize, both ISA1 homodimer alone and ISA1-ISA2 hetero-multimers have amylolytic activity (Figs. 1 and 6i), explaining why the functional loss of ISA2 has no phenotypes on rice and maize endosperm development21,22,38. In both potato and Arabidopsis, amylolytic activity is exclusively observed in the ISA1-ISA2 heterocomplex, while ISA1 alone exhibits no detectable activity (Fig. 6i) This finding explains why genetic disruption of either ISA1 or ISA2 produces identical phenotypic consequences, demonstrating that both subunits are essential for functional amylopectin biosynthesis in these species17,18,29. The distinct enzymatic properties of ISA1 orthologs explain why ZmISA1 and OsISA1, but not AtISA1 or StISA1, can complement the isa1isa2 double mutant phenotype in Arabidopsis39,48.
ISA2 is evolutionarily conserved across higher plants, sharing ___domain architecture (CBM, AMY, and C-terminal segment) with ISA1 but exhibiting distinct functional properties (Supplementary Fig. 7). ISA2 contains valid mutations in the catalytic center of AMY that render it enzymatically inactive, even when the critical triad are converted to the same as ISA1 (Supplementary Fig. 11g). Substrate-binding assays revealed that OsISA2 possessed strikingly higher amylopectin-binding activity than OsISA1 (Fig. 5b, c). Notably, the individual ___domain of OsISA2 (NTD, CBM, AMY and CT) retained substrate-binding activity (Fig. 5c). Though structurally comparable to OsISA2-CBM and classified as CBM48 (Supplementary Fig. 11), OsISA1-CBM lacked detectable amylopectin binding (Fig. 5b), consistent with the absence of MOS in the ISA1-CBM in our cryo-EM structure of MOS-bound OsISA1-ISA2 complex (Fig. 4a). The OsISA2-NTD segment comprising a β-sandwich fold resembling yet distinct from canonical CBMs exhibits substrate-binding activity. However, the NTD shows lower sequence conservation among plant ISA2 orthologs, thus the evolutionarily role for this extra ___domain needs further investigated.
Our cryo-EM structure of the MOS-bound OsISA1-ISA2 complex, obtained after amylopectin incubation, reveals MOS density surrounding the OsISA2 CT segment (Fig. 4a). This structure suggests a potential substrate channel spanning from OsISA2 to OsISA1, which may modulate substrate affinity, specificity, or catalytic efficiency. These findings demonstrate how OsISA2 has evolved distinct biochemical properties from OsISA1, trading catalytic activity for enhanced substrate binding. To further uncover the molecular basis by which ISA2 modulates the amylolytic activity of ISA1, more MOS- and branched MOS-bound ISA1-ISA2 hetero-multimer structures from diverse plant species will be crucial to clarify their coordinated roles in promoting phase transition of soluble phytoglycogen into semi-crystalline amylopectin biosynthesis in the future.
Methods
Molecular cloning, protein expression, and purification
The coding sequences of OsISA1 (LOC_Os08g40930.1) and OsISA2 (LOC_Os05g32710.1) from oryza sativa were amplified. OsISA1 (amino acids 55–803) was subcloned into the pET15D vector with a N-terminal 6×His-tag. For soluble expression and purification of OsISA2, OsISA2 (amino acids 35–800) was cloned into pBB75 vector containing a N-terminal SUMO-tag and a C-terminal 2×Strep-tag. OsISA2 (amino acids 35–800) was subcloned into the pBB75 vector with a N-terminal SUMO-tag for co-expression with His-tagged OsISA1. The point mutants were constructed by overlapping PCR. All constructs were verified by Sanger sequencing. For single protein expression, the plasmids of His-tagged OsISA1 and SUMO-Strep-tagged OsISA2 were individually transformed into E. coli strain BL21 (DE3). For protein co-expression, the plasmids of His-tagged OsISA1 and SUMO-tagged OsISA2 were co-transformed into BL21 (DE3). One liter of LB medium containing 100 mg mL–1 ampicillin or (and; for co-expression) 50 mg mL–1 kanamycin was inoculated with a transformed bacterial pre-culture and shaken at 37 °C until the cell density reached an OD600 of ~1.0–1.2. Protein production was induced with 0.2 mM isopropyl-β-D-thiogalactopyranoside at 16 °C for 14–16 h. The cells were collected by centrifugation, homogenized in buffer A (25 mM Tris-HCl, pH 8.0, 150 mM NaCl), and lysed using a high-pressure cell disrupter (JNBIO, China). Cell debris was removed by centrifugation at 20,000 × g for 1 h at 4 °C.
To purify OsISA1 or OsISA1-ISA2 proteins with 6×His-tag, the supernatant was loaded onto a column equipped with Ni2+ affinity resin (Ni-NTA, Qiagen), washed with buffer B (25 mM Tris-HCl, pH 8.0, 150 mM NaCl, 15 mM imidazole), and eluted with buffer C (25 mM Tris-HCl, pH 8.0, 250 mM imidazole). To purify OsISA2 proteins with 2 × Strep-tag, the supernatant was loaded onto Strep-Tactin affinity resin and eluted with buffer containing 100 mM Tris-HCl, pH 8.0, 150 mM NaCl and 5 mM desthiobiotin. The protein was then separated by cation exchange chromatography (Source 15Q, GE Healthcare) using a linear NaCl gradient in buffer A. The purified protein was concentrated and subjected to gel filtration chromatography (Superdex-200 Increase 10/300, GE Healthcare) in a buffer containing 25 mM Tris-HCl, pH 8.0 and 150 mM NaCl. Purity of the proteins were examined using SDS-PAGE and visualized by Coomassie brilliant blue staining through all purification steps. The peak fractions of His-OsISA1 and His-OsISA1-SUMO-OsISA2 were collected for Cryo-EM grid preparation or stored at –80 °C for further biochemical assays. The mutant proteins were purified similarly as the wild-type proteins. Homologous isoamylase ISA1s and ISA2s from maize, potato, and Arabidopsis were also constructed. Maize ZmISA1 (UniProt: EU970890.1; residues 45–789), ZmISA2 (UniProt: AY172633.3; residues 23–799), potato StISA1 (UniProt: AY132996.1; residues 48–793), and StISA2 (UniProt: AY132997.1; residues 39–878) were individually cloned into pET15D vector with a N-terminal 6×His-tag. ZmISA2 and StISA2 were also cloned into pBB75 vector with no-tag for co-expression with His-tagged ZmISA1 and StISA1, respectively. Arabidopsis AtISA1 (AT2G39930; residues 44–783) and AtISA2 (AT1G03310; residues 63–882) were cloned into pET21B with a C-terminal 6×His-tag. AtISA1 was also cloned into pBB75 vector with no-tag for co-expression with His-tagged AtISA2. Protein production and purification of these homologous His-tagged ISA1s or ISA2s, and the ISA1-ISA2 heterocomplexes were performed similarly as described above for rice isoamylases.
In vitro pull-down assays and western blot
For pull-down assays, mutants of OsISA2 were cloned into pBB75 vector containing a N-terminal SUMO-tag and a C-terminal HA-tag. To validate the interaction interface between OsISA1 and OsISA2, His-tagged OsISA1 was used to pull down HA-Sumo-tagged OsISA2. Both the His-tagged OsISA1 and HA-Sumo-tagged OsISA2 were co-expressed in E. coli BL21 (DE3). Mutants or deletions of OsISA2 were also co-expressed with wild-type OsISA1, and vice versa. Following bacterial cell disruption, the supernatant was subjected to purification using either Ni-NTA columns. The pellet, supernatant, wash, and eluent fractions were then analyzed by SDS-PAGE, followed by Coomassie Brilliant Blue staining or Western blot analysis.
Isoamylase enzymatic assays using the copper-bicinchoninic acid (BCA)
Reducing end formation released by the debranching enzyme on different glucan substrates was quantified using a method adapted from a previous report29. The assays were performed in a 60 μL reaction system, 4 nM OsISA1, 4 nM OsISA2, and 1 nM OsISA1-ISA2 complex were each mixed with 2.5 mg/mL glucan substrate (amylopectin, amylose, or glycogen), respectively. Samples were diluted with 50 mM MES (pH 6.5) to ensure that the absorption readings were within the linear range and then incubated at 30 °C. After 30 minutes reactions, 60 μL aliquots were heated at 95 °C for 15 min to stop the reaction. An equal volume of BCA reagent was then added to the samples. The BCA reagent was prepared by mixing 0.5 M Na2CO3, 0.288 M NaHCO3, and 5 mM sodium bicinchoninic acid with 12 mM L-Ser and 5 mM CuSO4·5H2O at a 1:1 ratio. The mixture was incubated at 80 °C for 30 min and then allowed to cool to room temperature. Finally, 100 μL aliquots from each assay were transferred to a microtiter plate, and the absorbance was recorded at OD562. The amount of reducing sugar was determined by plotting the absorption value from the reaction against the maltose standard curve. Assays were checked to ensure that they were linear with respect to enzyme concentration and time and that substrate concentration was saturating. The absorption values are within the standard curves. The enzyme activity is in a unit of mg/s/μmol (mg, the amount of the reducing ends in the reactions; s, seconds of the reaction time; μmol, the enzymes in the reactions). The experiment was performed in triplicate to ensure reproducibility.
Native PAGE coupled enzymatic examination
Aliquots of the purified proteins (50 nM) were mixed with 2×loading buffer (50% [v/v] glycerol, 0.05% [v/v] Bromophenol Blue) in a 1:1 ratio to a 5 μL system. The samples were then loaded onto a 6% native PAGE containing 0.1% (w/v) maize amylopectin (Aladdin, China) and electrophoresed at 250 V for 3 h at 4 °C. After electrophoresis, the gels were incubated in 50 mM MES, pH 6.5 and 1 mM DTT for 2 h at 37 °C. Amylolytic activity was visualized by staining the gels with an I2-KI solution containing 1% (w/v) I2 and 2% (w/v) KI.
Polyglucan-binding assay
The polyglucan-binding assay was conducted following previously described method with minor modifications49. Water-insoluble amylopectin isolated from maize was purchased commercially (Aladdin, China) and used as the substrate. OsISA1 (amino acids 55–803) was cloned into pET15D vector containing a N-terminal 6×His-tag and a C-terminal myc-tag, and OsISA2 (amino acids 35–800) was cloned into pBB75 vector containing a N-terminal SUMO-tag and a C-terminal 2×Strep-tag. Recombinant proteins, including myc-tagged OsISA1, strep-tagged OsISA2 and their respective boundaries or mutants (~100 nM), were incubated with an excess of amylopectin (~60 mg) in a binding buffer (50 mM HEPES-NaOH, pH 7.4, 2 mM MgCl2, 1 mM DTT, 0.1% [w/v] BSA, and 0.01% [v/v] Triton X-100) in a total reaction volume of 250 μL. The mixture was rotated end-over-end at 25 °C for 30 min. After incubation, the unbound fraction in the supernatant was collected by centrifugation at 13,500 × g for 30 s. The pellets were washed three times with the binding buffer, and then resuspended in 200 μL elution buffer (50 mM HEPES-NaOH, pH 7.4, 2 mM MgCl2, 1 mM DTT, and 4% [v/v] SDS) to obtain the bound fraction. Both the unbound and bound fractions were analyzed by SDS-PAGE, followed by immunoblotting using anti-myc or anti-strep antibodies. Each experiment was performed in at least three biological replicates to ensure reproducibility.
Western blot
Proteins were transferred to 0.45 μm PVDF membranes (Merck) using the Trans-Blot Turbo transfer system (Bio-Rad). Membranes were blocked in 5% BSA/TBST overnight at 4 °C, followed by incubation with primary antibodies diluted in 5% BSA/TBST for 2 h at room temperature. After three 5-min washes with TBST, membranes were incubated with HRP-conjugated secondary antibodies (1:4000 in 5% BSA/TBST) for 1 h at room temperature. After three additional TBST washes, membranes were imaged using a Tanon Imaging System. Uncropped blots are provided in the Source Data file. Antibodies used in this study are as follows. Primary antibodies (1:3000 in 5% BSA/TBST): Anti-His (TransGen Biotech, #HT501); Anti-Myc (TransGen Biotech, #HT101); Anti-Strep (Abbkine, Lot# ABT2230); Anti-HA (Proteintech, #51064-2-AP). Secondary antibodies (1:4000 in 5% BSA/TBST): HRP-conjugated goat anti-mouse IgG (H + L) (Proteintech, #SA00001-1); HRP-conjugated goat anti-rabbit IgG (H + L) (Proteintech, #SA00001-2).
Analytical ultracentrifugation (AUC) analysis
The molecular masses of ISA1, ISA2, and ISA1-ISA2 were investigated by AUC experiments, which were performed in a Beckman Coulter XL-I analytical ultracentrifuge using two-channel centrepieces. The protein sample was in solutions containing 25 mM Tris-HCl, pH 8.0, 150 mM NaCl. Data were collected by absorbance detection at 18 °C for proteins at a concentration of ~0.6 mg mL−1 at a rotor speed of 45,000 × g. The SV-AUC data were globally analyzed using the SEDFIT program and fitted to a continuous c(s) distribution model to determine the molecular mass.
Grid preparation and data acquisition
Aliquots (3.5 μL) of the His-tagged OsISA1 (0.75 mg mL–1) and His-OsISA1-SUMO-ISA2 complex (0.6 mg mL–1) were dropped onto glow discharged Quantifoil R1.2/1.3 300 mesh Cu grids (Quantifoil, MicroTools GmbH, Germany), blotted with a Vitrobot Mark IV (ThemoFisher Scientific) using 3 s blotting time with 100% humidity at 8 °C, and plunged into liquid ethane cooled by liquid nitrogen. For ligand-bound samples, OsISA1-ISA2 complex (0.93 mg mL−1) was incubated with maltopentaose, maltoheptaose, or maize amylopectin (Aladdin, China) at a 1:5 stoichiometry for 8 h at 4 °C, respectively, and then subjected to grid preparation. The sample was imaged on an FEI Titan Krios transmission electron microscope at 300 kV with a magnification of 81,000×. Images were recorded by a Gatan K3 Summit direct electron detector using the counting mode. Defocus values varied from 1.2 µm to 2.2 μm. Each image was dose-fractionated to 40 frames with a total electron dose of 50 e− Å−2 and a total exposure time of 3.0 s. EPU was used for fully automated data collection50.
Data processing, model building and refinement
All stacks were motion corrected using MotionCor2 with a binning factor of 1, resulting in a pixel size of 0.84 Å and dose weighting was performed concurrently51. The defocus values were estimated using Gctf52. For OsISA1, a total of 956 good micrographs were selected, from which 723,145 particles were auto-picked using RELION53. After reference-free 2D classification, 715,065 good particles were selected for 3D classification. Multi-reference 3D classification was performed in RELION. Then, a total of 161,630 particles were selected from good classes and transferred to the cryoSPARC software package for further processing54, followed by several rounds of ab-initio reconstruction, heterogeneous refinement. Particles belonging to the best class were selected followed by non-uniform refinement and local refinement, applying C2 symmetry, yielding a particle density with an estimated resolution of 2.7 Å based on FSC55. Similar data processing strategy was applied for the ligand-free and ligand-bound OsISA1-ISA2 complex from a total of 1004 and 1052 good micrographs, respectively, applying C1 symmetry, resulting in a particle density with an estimated resolution of 2.4 Å and 2.7 Å. The atomic model was built in COOT56 and refined with PHENIX57. The structure was validated through the examination of Molprobity scores58. Resolution was estimated with the gold standard Fourier shell correlation 0.143 criterion. High-resolution images were prepared using ChimeraX and PyMOL. Data collection and refinement statistics are given in Supplementary Table S1. A diagram of the procedures for data processing is presented in Supplementary Figs. 4, 9 and 12.
Statistics and reproducibility
Data were analyzed using Prism (GraphPad) and Origin 8. Specific statistical tests are described in the figure legends. All *t*-tests were two-sided. Mean ± SD values are provided in the Source Data file (Figs. 1c, 6i and Supplementary Figs. S3, S8b). Significance levels: *p < 0.05, **p < 0.01, ***p < 0.001 (Figs. 1c and 6i). For isoamylase enzymatic assays, at least three independent biological replicates (each with three technical replicates) were performed using the copper-bicinchoninic acid (BCA) method.
Accession numbers
1BF2; 2VR5; 7U3A; 2WSK; Spatio-temporal gene expression of various organs throughout entire growth in the field. OsISA1 (LOC_Os08g40930/Os08g0520900): https://ricexpro.dna.affrc.go.jp/GGEP/graph-view.php?featurenum=20548 OsISA2 (LOC_Os05g32710/Os05g0393700): https://ricexpro.dna.affrc.go.jp/GGEP/graph-view.php?featurenum=36045.
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
Data availability
The cryo-EM density maps of OsISA1, OsISA1-ISA2 heterocomplex, and MOS-bound OsISA1-ISA2 heterocomplex have been deposited in the EM Data Bank (EMD-61158 [https://www.ebi.ac.uk/emdb/EMD-61158]; EMD-61188; EMD-63051) and the Protein Data Bank (9J60 [https://www.rcsb.org/structure/9J60]; 9J6X; 9LFN), respectively. Protein sequences and related tags of ISA1 and ISA2 used in this study are provided in the Supplementary Data 1. Source data is available for Figs. 1, 2g, 3f, g, 5b, c, d and 6 and Supplementary Figs. 1b, 2, 3, 6a, b 8b, c, 11g, 14, and 15. Source data are provided with this paper.
References
Zeeman, S. C., Kossmann, J. & Smith, A. M. Starch: its metabolism, evolution, and biotechnological modification in plants. Annu. Rev. Plant Biol. 61, 209–234 (2010).
Santelia, D. & Zeeman, S. C. Progress in Arabidopsis starch research and potential biotechnological applications. Curr. Opin. Biotechnol. 22, 271–280 (2011).
Streb, S. & Zeeman, S. C. Starch metabolism in Arabidopsis. Arabidopsis Book 10, e0160 (2012).
Smith, A. M. & Zeeman, S. C. Starch: a flexible, adaptable carbon store coupled to plant growth. Annu. Rev. Plant Biol. 71, 217–245 (2020).
James, M. G., Denyer, K. & Myers, A. M. Starch synthesis in the cereal endosperm. Curr. Opin. Plant Biol. 6, 215–222 (2003).
Keeling, P. L. & Myers, A. M. Biochemistry and genetics of starch synthesis. Annu. Rev. Food Sci. Technol. 1, 271–303 (2010).
Crofts, N., Nakamura, Y. & Fujita, N. Critical and speculative review of the roles of multi-protein complexes in starch biosynthesis in cereals. Plant Sci. 262, 1–8 (2017).
Georgelis, N., Braun, E. L., Shaw, J. R. & Hannah, L. C. The two AGPase subunits evolve at different rates in angiosperms, yet they are equally sensitive to activity-altering amino acid changes when expressed in bacteria. Plant Cell 19, 1458–1472 (2007).
Figueroa, C. M., Asencion Diez, M. D., Ballicora, M. A. & Iglesias, A. A. Structure, function, and evolution of plant ADP-glucose pyrophosphorylase. Plant Mol. Biol. 108, 307–323 (2022).
Szydlowski, N. et al. Starch granule initiation in Arabidopsis requires the presence of either class IV or class III starch synthases. Plant Cell 21, 2443–2457 (2009).
Tetlow, I. J. & Emes, M. J. A review of starch-branching enzymes and their role in amylopectin biosynthesis. IUBMB Life 66, 546–558 (2014).
Nakamura, Y. Towards a better understanding of the metabolic system for amylopectin biosynthesis in plants: rice endosperm as a model tissue. Plant Cell Physiol. 43, 718–725 (2002).
Nakamura, Y. & Kainuma, K. On the cluster structure of amylopectin. Plant Mol. Biol. 108, 291–306 (2022).
Delatte, T. et al. Evidence for distinct mechanisms of starch granule breakdown in plants. J. Biol. Chem. 281, 12050–12059 (2006).
Dinges, J. R., Colleoni, C., James, M. G. & Myers, A. M. Mutational analysis of the pullulanase-type debranching enzyme of maize indicates multiple functions in starch metabolism. Plant Cell 15, 666–680 (2003).
Wattebled, F. et al. Mutants of Arabidopsis lacking a chloroplastic isoamylase accumulate phytoglycogen and an abnormal form of amylopectin. Plant Physiol. 138, 184–195 (2005).
Bustos, R. et al. Starch granule initiation is controlled by a heteromultimeric isoamylase in potato tubers. Proc. Natl Acad. Sci. USA 101, 2215–2220 (2004).
Delatte, T., Trevisan, M., Parker, M. L. & Zeeman, S. C. Arabidopsis mutants Atisa1 and Atisa2 have identical phenotypes and lack the same multimeric isoamylase, which influences the branch point distribution of amylopectin during starch synthesis. Plant J. 41, 815–830 (2005).
Zeeman, S. C. et al. A mutant of Arabidopsis lacking a chloroplastic isoamylase accumulates both starch and phytoglycogen. Plant Cell 10, 1699–1712 (1998).
Kubo, A. et al. The starch-debranching enzymes isoamylase and pullulanase are both involved in amylopectin biosynthesis in rice endosperm. Plant Physiol. 121, 399–410 (1999).
Utsumi, Y., Utsumi, C., Sawada, T., Fujita, N. & Nakamura, Y. Functional diversity of isoamylase oligomers: the ISA1 homo-oligomer is essential for amylopectin biosynthesis in rice endosperm. Plant Physiol. 156, 61–77 (2011).
Kubo, A. et al. Functions of heteromeric and homomeric isoamylase-type starch-debranching enzymes in developing maize endosperm. Plant Physiol. 153, 956–969 (2010).
James, M. G., Robertson, D. S. & Myers, A. M. Characterization of the maize gene sugary1, a determinant of starch composition in kernels. Plant Cell 7, 417–429 (1995).
Burton, R. A. et al. Starch granule initiation and growth are altered in barley mutants that lack isoamylase activity. Plant J. 31, 97–112 (2002).
Dauvillee, D. et al. Biochemical characterization of wild-type and mutant isoamylases of Chlamydomonas reinhardtii supports a function of the multimeric enzyme organization in amylopectin maturation. Plant Physiol. 125, 1723–1731 (2001).
Posewitz, M. C. et al. Hydrogen photoproduction is attenuated by disruption of an isoamylase gene in Chlamydomonas reinhardtii. Plant Cell 16, 2151–2163 (2004).
Mouille, G. et al. Preamylopectin processing: a mandatory step for starch biosynthesis in plants. Plant Cell 8, 1353–1366 (1996).
Liu, C. et al. Like early starvation 1 and early starvation 1 promote and stabilize amylopectin phase transition in starch biosynthesis. Sci. Adv. 9, eadg7448 (2023).
Hussain, H. et al. Three isoforms of isoamylase contribute different catalytic properties for the debranching of potato glucans. Plant Cell 15, 133–149 (2003).
Lin, Q. et al. Function of isoamylase-type starch debranching enzymes ISA1 and ISA2 in the Zea mays leaf. N. Phytol. 200, 1009–1021 (2013).
Boehlein, S. K. et al. Soluble and insoluble alpha-glucan synthesis in yeast by enzyme suites derived exclusively from maize endosperm. Plant Physiol. 193, 1456–1478 (2023).
Abt, M. R. & Zeeman, S. C. Evolutionary innovations in starch metabolism. Curr. Opin. Plant Biol. 55, 109–117 (2020).
Ball, S. et al. From glycogen to amylopectin: a model for the biogenesis of the plant starch granule. Cell 86, 349–352 (1996).
Myers, A. M., Morell, M. K., James, M. G. & Ball, S. G. Recent progress toward understanding biosynthesis of the amylopectin crystal. Plant Physiol. 122, 989–997 (2000).
Moller, M. S., Henriksen, A. & Svensson, B. Structure and function of alpha-glucan debranching enzymes. Cell Mol. Life Sci. 73, 2619–2641 (2016).
Utsumi, Y. & Nakamura, Y. Structural and enzymatic characterization of the isoamylase1 homo-oligomer and the isoamylase1-isoamylase2 hetero-oligomer from rice endosperm. Planta 225, 75–87 (2006).
Sundberg, M. et al. The heteromultimeric debranching enzyme involved in starch synthesis in Arabidopsis requires both isoamylase1 and isoamylase2 subunits for complex stability and activity. PLoS One 8, e75223 (2013).
Lin, Q. et al. Functional interactions between starch synthase III and isoamylase-type starch-debranching enzyme in maize endosperm. Plant Physiol. 158, 679–692 (2012).
Facon, M. et al. Distinct functional properties of isoamylase-type starch debranching enzymes in monocot and dicot leaves. Plant Physiol. 163, 1363–1375 (2013).
Sim, L. et al. Crystal structure of the Chlamydomonas starch debranching enzyme isoamylase ISA1 reveals insights into the mechanism of branch trimming and complex assembly. J. Biol. Chem. 289, 22991–23003 (2014).
Katsuya, Y., Mezaki, Y., Kubota, M. & Matsuura, Y. Three-dimensional structure of Pseudomonas isoamylase at 2.2 A resolution. J. Mol. Biol. 281, 885–897 (1998).
Woo, E. J. et al. Structural insight into the bifunctional mechanism of the glycogen-debranching enzyme TreX from the archaeon Sulfolobus solfataricus. J. Biol. Chem. 283, 28641–28648 (2008).
Schumacher, M. A. et al. Allosteric regulation of glycogen breakdown by the second messenger cyclic di-GMP. Nat. Commun. 13, 5834 (2022).
Song, H. N. et al. Structural rationale for the short branched substrate specificity of the glycogen debranching enzyme GlgX. Proteins 78, 1847–1855 (2010).
Abramson, J. et al. Accurate structure prediction of biomolecular interactions with AlphaFold 3. Nature 630, 493–500 (2024).
Drula, E. et al. The carbohydrate-active enzyme database: functions and literature. Nucleic Acids Res 50, D571–D577 (2022).
Utsumi, Y. & Nakamura, Y. Structural and enzymatic characterization of the isoamylase1 homo-oligomer and the isoamylase1–isoamylase2 hetero-oligomer from rice endosperm. Planta 225, 75–87 (2006).
Streb, S. & Zeeman, S. C. Replacement of the endogenous starch debranching enzymes ISA1 and ISA2 of Arabidopsis with the rice orthologs reveals a degree of functional conservation during starch synthesis. PLoS One 9, e92174 (2014).
Yan, H. et al. Rice LIKE EARLY STARVATION1 cooperates with FLOURY ENDOSPERM6 to modulate starch biosynthesis and endosperm development. Plant Cell 36, 1892–1912 (2024).
Mastronarde, D. N. Automated electron microscope tomography using robust prediction of specimen movements. J. Struct. Biol. 152, 36–51 (2005).
Zheng, S. Q. et al. MotionCor2: anisotropic correction of beam-induced motion for improved cryo-electron microscopy. Nat. Methods 14, 331–332 (2017).
Zhang, K. Gctf: Real-time CTF determination and correction. J. Struct. Biol. 193, 1–12 (2016).
Kimanius, D., Forsberg, B. O., Scheres, S. H. & Lindahl, E. Accelerated cryo-EM structure determination with parallelisation using GPUs in RELION-2. Elife 5 e18722 (2016).
Punjani, A., Rubinstein, J. L., Fleet, D. J. & Brubaker, M. A. cryoSPARC: algorithms for rapid unsupervised cryo-EM structure determination. Nat. Methods 14, 290–296 (2017).
Chen, S. et al. High-resolution noise substitution to measure overfitting and validate resolution in 3D structure determination by single particle electron cryomicroscopy. Ultramicroscopy 135, 24–35 (2013).
Emsley, P. & Cowtan, K. Coot: model-building tools for molecular graphics. Acta Crystallogr. D. Biol. Crystallogr. 60, 2126–2132 (2004).
Adams, P. D. et al. PHENIX: a comprehensive Python-based system for macromolecular structure solution. Acta Crystallogr. D. Biol. Crystallogr. 66, 213–221 (2010).
Davis, I. W. et al. MolProbity: all-atom contacts and structure validation for proteins and nucleic acids. Nucleic Acids Res. 35, W375–W383 (2007).
Acknowledgements
The authors thank the cryo-EM Facility at Wuhan University, Public Laboratory of Electron Microscopy and Center for Protein Research (CPR) of Huazhong Agricultural University, for providing technical support during sample preparation and EM image acquisition. This work was supported by the National Key R&D Program of China (2023YFF1001100 to P.Y.), the National Natural Science Foundation of China (32270255, 32130011 to J.Y.), the Foundation of Hubei Hongshan Laboratory (2021HSKF003 to J.Y.), Fundamental Research Funds for the Central Universities (2662025SKPY001 to J.Y., 2662025SKPY008 to P.Y.), Open Research Fund of State Key Laboratory of Hybrid Rice (KF202103, Wuhan University, to J.Y.), and the Knowledge Innovation Program of Wuhan-Shuguang Project (2023020201020353, to Y.L.).
Author information
Authors and Affiliations
Contributions
J.Y. designed and supervised the project. R.F., G.Z., X.Y. and P.C. purified proteins, carried out the biochemical assays. F.Z., M.W., X.W. and J.L. prepared the cryo-EM grids, and acquired EM data. Z.G. built and refined the atomic model. Y.L., D.Z. and P.Y. analyzed the data. J.Y. wrote the manuscript.
Corresponding author
Ethics declarations
Competing interests
The authors declare no competing interests.
Peer review
Peer review information
Nature Communications thanks Alan Myers, who co-reviewed with Tracie Hennen-Bierwagen and Alan Culbertson and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Source data
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
About this article
Cite this article
Fan, R., Guan, Z., Zhou, G. et al. Amylopectin branch trimming and biosynthesis elucidated by the rice isoamylase ISA1-ISA2 heterocomplex. Nat Commun 16, 5638 (2025). https://doi.org/10.1038/s41467-025-60944-6
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41467-025-60944-6
