A new method for artificially designing scaffold proteins that mediate RuBisCO aggregation in crops

Abstract

This invention provides a method for designing scaffold proteins that mediate RuBisCO aggregation in crop plants. Using a known scaffold protein as a template, this method generates sequences through the following steps: a. RFdiffusion scaffold protein backbone generation; b. MPNN sequence generation; c. AlphaFold structure prediction; and d. Optimizing the optimal sequence by evaluating the generated scaffold protein sequences using one or more scoring metrics. Unlike previous research focused on RuBisCO itself, this invention shifts the focus to the design and optimization of scaffold proteins. By optimizing the parameter evaluation method for protein design, novel scaffold proteins were designed, addressing the bottleneck of traditional methods in forming the RuBisCO concentration module and opening up a new research direction.

Description

Technical Field

[0001] The present invention belongs to the field of agricultural biotechnology, and specifically relates to protein design and optimization based on artificial intelligence, and more specifically to a new method for artificially designing scaffold proteins that mediate the aggregation of crop RuBisCO. Background Art

[0002] The growth of C3 crops such as rice is limited by the low carbon fixation efficiency of ribulose-1,5-bisphosphate-carboxylase / oxygenase (RuBisCO), resulting in actual yields far lower than theoretical values. C4 plants, as well as cyanobacteria and eukaryotic algae, have evolved a carbon dioxide concentrating mechanism (CCM) that concentrates inorganic carbon around RuBisCO through the action of carbon transporters and carbon fixation chambers, forming a RuBisCO concentrating module, thereby increasing the substrate CO2 concentration and reducing the occurrence of photorespiration, thereby improving photosynthesis efficiency. Typical carbon fixation chambers include the carboxysomes of cyanobacteria and the pyrenoids in Chlamydomonas. (1) . C3 crops do not have a CCM mechanism. Introducing CCMs from algae and other organisms into C3 crops is considered an efficient strategy to increase yields, which is expected to alleviate the pressure caused by the reduction of arable land and insufficient food supply. However, how to form a functional RuBisCO concentration module in crops remains a technical bottleneck for implementing this strategy. Therefore, designing a new scaffold protein that can adapt to RuBisCO in C3 crops and mediate the concentration and aggregation of RuBisCO in chloroplasts is a key issue to be solved.

[0003] Existing studies have mainly focused on introducing exogenous RuBisCO or screening RuBisCO mutants, but the results have not met expectations, and the growth of mutant plants has not been superior to that of wild-type plants. (2) The main existing methods include:

[0004] 1. Introduction of exogenous RuBisCO

[0005] By replacing the large subunit of RuBisCO in tobacco with cyanobacterial RuBisCO and its scaffold protein, the researchers successfully observed the formation of a functional RuBisCO concentration module in tobacco leaves and obtained related mutant plants. (3–7) However, these mutants were unable to grow normally under air conditions, exhibiting a phenotype requiring high carbon dioxide levels, and their growth rates were significantly slower than those of the wild type. Even after the introduction of other CCM modules, they still exhibited growth defects.

[0006] 2. Construction of Hybrid RuBisCO

[0007] The cyanobacterial scaffold protein only interacts with the large subunit of RuBisCO. A study introduced the large subunit of RuBisCO and the scaffold protein from cyanobacteria into tobacco, and the results showed that it could form a hybrid enzyme with the small subunit of tobacco and form a RuBisCO concentration module together with the scaffold protein. (8) However, these mutants still exhibited a high CO2 requirement phenotype and slow growth.

[0008] 3. Gene mutation

[0009] The scaffold protein in Chlamydomonas reinhardtii only interacts with the RuBisCO small subunit and has a simpler structure than that of cyanobacteria. By fusing the interacting α-helix in the Chlamydomonas reinhardtii RuBisCO small subunit to the Arabidopsis RuBisCO small subunit, the researchers successfully made the Arabidopsis RuBisCO bind to the Chlamydomonas scaffold protein to form a condensation module in vitro. (9) However, due to the instability of the C. reinhardtii scaffold protein in higher plants, a functional RuBisCO concentrating module was ultimately not observed in the mutant plants.

[0010] While the aforementioned studies (Table 1) have made some progress in RuBisCO modification, no substantial breakthrough has yet been achieved. A new research approach is urgently needed to construct RuBisCO concentration modules in crops to improve carbon fixation efficiency.

[0011] Table 1 Existing RuBisCO concentration module construction system

[0012]

[0013] References:

[0014] 1. JH Hennacy, MC Jonikas, Prospects for EngineeringBiophysical CO2 Concentrating Mechanisms into Land Plants to Enhance Yields. Annu Rev Plant Biol 71, 461–485 (2020).

[0015] 2. ND Nguyen, et al., A carboxysome-based CO2 concentrating mechanism for C3 crop chloroplasts: advances and the road ahead. Plant J 118,940–952 (2024).

[0016] 3. M. T. Lin, A. Occhialini, P. J. Andralojc, M. A. J. Parry, M. R.Hanson, A faster Rubisco with potential to increase photosynthesis in crops.Nature 513, 547–550 (2014).

[0017] 4. A. Occhialini, M. T. Lin, P. J. Andralojc, M. R. Hanson, M. A. J.Parry, Transgenic tobacco plants with improved cyanobacterial Rubiscoexpression but no extra assembly factors grow at near wild-type rates ifprovided with elevated CO2. Plant J 85, 148–160 (2016).

[0018] 5. B. M. Long, et al., Carboxysome encapsulation of the CO2-fixingenzyme Rubisco in tobacco chloroplasts. Nat Commun 9, 3570 (2018).

[0019] 6. T. Chen, et al., Incorporation of Functional Rubisco Activasesinto Engineered Carboxysomes to Enhance Carbon Fixation. ACS Synth Biol 11,154–161 (2022).

[0020] 7. T. Chen, et al., Engineering α-carboxysomes into plantchloroplasts to support autotrophic photosynthesis. Nat Commun 14, 2118(2023).

[0021] 8. DJ Orr, et al., Hybrid Cyanobacterial-Tobacco Rubisco SupportsAutotrophic Growth and Procarboxysomal Aggregation. Plant Physiol 182, 807–818 (2020).

[0022] 9. N. Atkinson, et al., The pyrenoidal linker protein EPYC1 phaseseparates with hybrid Arabidopsis-Chlamydomonas Rubisco through interactions with the algal Rubisco small subunit. J Exp Bot 70, 5271–5285 (2019). Summary of the Invention

[0023] This invention aims to design and optimize scaffold proteins, key components of the RuBisCO concentrating module (RCM). Advanced artificial intelligence (AI) technology is used for protein design, optimizing screening parameters and metrics to generate a library of artificially designed protein sequences. We then screen and validate these proteins through wet experiments, identifying novel scaffold proteins capable of binding to RuBisCO in crops. These proteins can then be used to construct and optimize RuBisCO concentrating modules in crops. This approach provides a new solution for the application of the CO2 concentrating mechanism (CCM) in crops, aiming to improve crop photosynthesis efficiency and yield.

[0024] Specifically, the present invention provides the following technical solutions:

[0025] In one aspect, the present invention provides a method for designing a scaffold protein that mediates RuBisCO aggregation in crops. The method uses a known scaffold protein as a template and generates a sequence by the following steps:

[0026] a. RFdiffusion scaffold protein backbone generation: Based on the known complex structure of the scaffold protein and the corresponding RuBisCO, the interaction interface between the two is determined. The residues at the scaffold protein interface are set as hotspot residues during diffusion. The length of the sequence generated by diffusion in the hotspot residue region is limited to the same length as the known scaffold protein, while the amino acid residues at other positions remain unchanged to generate the scaffold protein backbone.

[0027] b. MPNN sequence generation: The scaffold protein backbone generated in step a is filled with amino acid side chains to ensure structural conformation stability. The scaffold protein backbone generated in step a is matched to the known binding sites of the scaffold protein and RuBisCO, taking into account the interaction between the scaffold protein and RuBisCO and steric hindrance, thereby generating the scaffold protein sequence.

[0028] c. AlphaFold structure prediction: The generated scaffold protein sequence is used to predict the complex structure with the crop RuBisCO to obtain the interaction interface information between the scaffold protein and RuBisCO;

[0029] d. Evaluate the generated scaffold protein sequences using one or more scoring metrics to select the best sequence.

[0030] In some embodiments, the scoring metrics include Distance, ipTM, ipae, and RMSD.

[0031] In some embodiments, sequences are selected with a Distance of less than 10 Å.

[0032] In some embodiments, sequences with an ipTM score greater than 0.8 are selected.

[0033] In some embodiments, sequences with an IPAE of less than 10 or less than 20 are selected. In the present invention, the range of this metric should be adjusted based on the IPAE scores of different template scaffold proteins and RuBisCO, as well as the IPAE scores of all corresponding designed scaffold proteins and RuBisCO. For example, when screening scaffold proteins designed based on EPYC1, CsoS2, and PYCO1, sequences with an IPAE of less than 10 are selected, while when screening scaffold proteins designed based on CcmM, the IPAE range should be expanded to sequences with less than 20.

[0034] In some embodiments, sequences are selected that have an RMSD of less than 2 Å.

[0035] In some embodiments, the known scaffold protein is selected from CsoS2 protein from α-cyanobacteria, CcmM protein from β-cyanobacteria, EPYC1 protein from Chlamydomonas reinhardtii, and PYCO1 protein from diatoms, but is not limited thereto.

[0036] In some embodiments, the method further comprises, after step d, rescreening sequences whose binding sites are identical to known scaffold protein binding sites, and analyzing the interaction interface between the scaffold protein and RuBisCO using the PDBsum website to screen out scaffold protein sequences that mediate the aggregation of crop RuBisCO.

[0037] In another aspect, the present invention provides a scaffold protein sequence obtained by screening according to the above method.

[0038] In some embodiments, the sequence of the scaffold protein is shown in SEQ ID NO: 1.

[0039] In another aspect, the present invention provides use of the scaffold protein described above in increasing crop yield.

[0040] In the present invention, the term template scaffold protein is used interchangeably with known scaffold proteins.

[0041] definition

[0042] Wet labs: Wet labs are in contrast to dry labs, which rely on computers and data. Biological wet labs refer to laboratory experiments involving biochemical reagents and biological sample processing. These experiments are characterized by the need for physical manipulation and handling of biological materials, often using various chemical reagents and precision instruments to process and test biological samples.

[0043] Scaffold proteins: During phase separation, scaffold proteins drive and stabilize the formation of phase-separated droplets or aggregates through multivalent interactions. Scaffold proteins typically possess multiple binding sites that can interact with other proteins or nucleic acids, forming complex network structures. For example, in this application, scaffold proteins can form multivalent interactions with RuBisCO, mediating the formation of the RuBisCO concentrating module.

[0044] Hotspot residues: In protein design, hotspot residues refer to amino acid residues that contribute significantly to the binding free energy at the protein-protein interaction interface. By targeting these hotspot residues, we can provide a proposed binding region for the interacting proteins, thereby reducing the difficulty of designing interacting proteins.

[0045] RuBisCO concentrating module: The RuBisCO concentrating module refers to localized RuBisCO aggregates mediated by multivalent interactions between scaffold proteins and RuBisCO, driving and stabilizing the phase separation formation of droplets.

[0046] Phase separation: Phase separation is the process by which a homogeneous mixture separates into two or more distinct phases. In biology, phase separation can form droplets or aggregates, which perform specific functions within cells. RuBisCO phase separation occurs when RuBisCO forms membraneless organelles or aggregates through liquid-liquid phase separation. This phase separation helps increase the local concentration of RuBisCO, which not only improves the efficiency of photosynthesis but also regulates carbon assimilation within the cell.

[0047] There are four existing scaffold proteins: CsoS2 from α-cyanobacteria (Uniprot: O85041), CcmM from β-cyanobacteria (Uniprot: Q03513), EPYC1 from Chlamydomonas reinhardtii (Uniprot: Q94ET8), and PYCO1 from diatoms (Uniprot: B7GCF7). These four scaffold proteins are derived from non-plant organisms. Due to differences in the interaction interfaces resulting from species differences, these scaffold proteins can only mediate the aggregation of RuBisCO from their own species and cannot mediate the aggregation of RuBisCO from plants, such as crops.

[0048] Flexible motif: The scaffold protein connects multiple RuBisCO binding domains in series through a flexible sequence or motif. The flexible sequence or motif of the scaffold protein in the present invention is derived from the flexible sequences or motifs in the above four known scaffold proteins.

[0049] Beneficial effects

[0050] (1) Novelty of the target: Unlike previous studies that focused on RuBisCO itself, this study shifted the focus to the design and optimization of scaffold proteins for the first time. By optimizing the parameter evaluation method for protein design, a new scaffold protein was designed, which solved the bottleneck of the RuBisCO concentration module in traditional methods and opened up a new research direction.

[0051] (2) Introducing and improving protein AI design methods: The present invention innovatively combines artificial intelligence technology for protein design, breaking through the limitations of traditional methods based on structural mutation and screening. By combining AI design with structural biology, the efficiency of design and optimization is significantly improved, and the R&D cycle is shortened. In addition, the current design methods for interacting proteins are often designed for interactions between two proteins, while the design of multi-subunit interactions between CcmM and RuBisCO, two large subunits and one small subunit in existing scaffold proteins, cannot be designed using the conventional RFdiffusion-ProteinMPNN-Alphafold protein design method. Therefore, when designing scaffold proteins based on CcmM, the present invention does not use RFdiffusion, which can only be used for the design of interactions between two proteins, but directly generates ProteinMPNN for the hotspot residue region of the scaffold protein to obtain the sequence used for subsequent screening.

[0052] (3) Innovation in screening indicators and parameters: The interaction between the scaffold protein and RuBisCO is relatively weak, which is also to maintain the fluidity and variability of the phase-separated droplets. Therefore, when performing IPAE scoring screening, the current common range of screening high-affinity interacting proteins was not used as an indicator. Instead, the IPAE score of the binding of the existing scaffold protein to RuBisCO was first performed, and the IPAE scores of all generated scaffold proteins binding to RuBisCO were referred to. The range of IPAE scoring indicators used for the final screening was determined by comprehensive consideration. The relevant indicators commonly used in protein design currently only evaluate the accuracy of protein prediction results, and do not analyze protein interactions. The present invention innovatively introduces PDBsum software that performs detailed analysis of protein interaction interfaces, and conducts a specific analysis of the interaction interface residues of the predicted protein complexes. The analysis results are used to screen high-affinity interacting proteins and the interaction residues can be mutated and optimized based on these results.

[0053] (4) Simple operation and low cost: Unlike traditional methods that are costly and complex, most of the work in this invention is completed through computer algorithms, which reduces the need for extensive wet experiments and reduces experimental costs. In addition, existing protein design and structure prediction tools are mature, and relevant literature and tutorials are abundant, further reducing the cost of trial and error.

[0054] (5) High throughput and rapid screening: With the help of AI design, the present invention can generate thousands of protein sequences in a short period of time and quickly screen out the best sequences through high-throughput scoring indicators, significantly improving experimental efficiency and shortening the research cycle.

[0055] (6) The present invention creatively introduces the software PDBsum for protein interaction interface analysis. This software can help analyze the specific interactions between the designed scaffold protein and RuBisCO, and can be accurate to specific residues, which facilitates the subsequent analysis, classification and optimization of the interacting residues. Among the four scoring indicators, Distance, ipTM, ipae and RMSD, iPTM is used to evaluate the accuracy of Alphafold's prediction results, RMSD can be used to evaluate the stability of the designed scaffold protein before and after binding to RuBisCO, Distance roughly evaluates whether the scaffold protein is bound to the specified position of the interaction interface of RuBisCO, and ipae can roughly evaluate the interaction between the scaffold protein and RuBisCO. The above indicators do not conduct detailed analysis and screening of the interaction interface between the designed scaffold protein and RuBisCO. As a result, thousands of sequences still meet the requirements after passing the indicator screening. The large number of sequence libraries has increased the workload and difficulty of subsequent wet experiment verification. The introduction of PDBsum can perform detailed residue and interaction area analysis on the interaction interface between the designed scaffold protein and RuBisCO in the predicted results. Selecting scaffold proteins with a large number of interacting residues and a large interaction area for subsequent wet experiments can reduce the workload and difficulty to a feasible range. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 The schematic diagram shows the results of SDS-PAGE detection of the composition and purity of rice RuBisCO. The first lane is the protein marker.

[0057] Figure 2 The results of SDS-PAGE detection of the expression and stability of the designed scaffold proteins are shown. The first lane is the protein marker, and the remaining lanes are the concentrated samples of the purified different scaffold proteins.

[0058] Figure 3 The results of the turbidity experiment of the designed scaffold protein and rice RuBisCO are shown. The ability of the scaffold protein to form a concentration module with RuBisCO was reflected by measuring the absorbance of the mixture of different scaffold proteins and crop RuBisCO at 340nm.

[0059] Figure 4 Negatively stained electron micrographs of a complex of scaffold protein and rice RuBisCO are shown. Specifically, a negatively stained electron micrograph of a sample incubated with 0.25 μM scaffold protein and 0.125 μM RuBisCO is shown. Scale bar, 100 nm, is indicated below the image. DETAILED DESCRIPTION

[0060] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific embodiments and with reference to the accompanying drawings.

[0061] The technical solution of the present invention includes the following parts:

[0062] 1. AI-Based Scaffold Protein Design. Compared to conventional AI design methods, the scaffold protein design of this application incorporates the following modifications and innovations: 1. Because scaffold proteins are composed of multiple RuBisCO binding domains connected by flexible motifs, we performed AI design targeting only the RuBisCO binding domains (also known as scaffold protein modules) rather than directly designing the entire scaffold protein. 2. After obtaining high-affinity RuBisCO binding domains, the ability and strength of the scaffold protein to mediate RuBisCO concentration module can be adjusted by adjusting the number and order of RuBisCO binding domains connected by flexible motifs. After screening for scaffold proteins with better evaluation indicators, we constructed plasmids with 3 and 4 binding domains combined with the characteristics of existing scaffold proteins, and purified the corresponding proteins. The results of the turbidity experiment showed that the scaffold protein with 4 binding domains had a stronger ability to mediate the formation of RuBisCO concentration modules. However, since the protein with 4 binding domains corresponding to P7 was expressed by precipitation during purification, it was misfolded and unstable. Therefore, the version with 3 binding domains was used in subsequent experiments, while the rest were all versions with 4 binding domains. Figure 2 It can also be seen that P7 is smaller than other scaffold proteins because it lacks a binding domain. P8 provided by the present invention has four binding domains connected by three flexible motifs. The flexible motifs are underlined in the following sequence.

[0063] If the design is not based on the scaffold protein, then the only option is to design for RuBisCO, another component in the RuBisCO concentration module. However, due to the complexity of RuBisCO itself as a hexamer and the fact that the design cannot affect its activity, the optimization of the RuBisCO concentration module designed for RuBisCO is more difficult. In addition, the current design modifications for RuBisCO often lead to a decrease in the biomass of the experimental plants, and no positive results have been obtained.

[0064] AI technology was used to design and generate new scaffold protein sequences. Using the four existing scaffold proteins as templates, the sequences were generated through the following steps:

[0065] RFdiffusion: This method leverages the powerful capabilities of the RoseTTAfold protein structure prediction network to generate new protein structures from random noise through an iterative denoising process. The interaction interface between the two proteins is determined based on the existing complex structure of the scaffold protein and the corresponding RuBisCO. Residues at the scaffold protein interface are designated as hotspot residues during diffusion. The length of the sequence generated by diffusion in the hotspot residue region is limited to the known scaffold protein, while the amino acid residues at other positions remain unchanged. RFdiffusion generates approximately 1,000 scaffold protein backbones.

[0066] MPNN sequence generation: The generated protein skeleton is filled with amino acid side chains to ensure that the structural conformation remains stable, thereby generating approximately 5,000 scaffold protein sequences.

[0067] AlphaFold structure prediction: The generated scaffold protein sequence is used to predict the complex structure with the crop RuBisCO to obtain the interaction interface information between the scaffold protein and RuBisCO.

[0068] 2. Dataset Evaluation and Screening

[0069] The generated scaffold proteins were evaluated using a variety of scoring metrics (iPTM and ipae are given in the Alphafold prediction results, while Distance and RMSD require additional calculations) to screen out the best sequences:

[0070] Distance: Use the distance parameter to score and preliminarily determine the relative position of the scaffold protein to RuBisCO. A smaller distance indicates that the scaffold protein is closer to RuBisCO. Sequences with a distance less than 10 Å are retained.

[0071] ipTM: This parameter measures the accuracy of the predicted relative positions of subunits within the complex. Sequences with an ipTM score greater than 0.8 are retained. These results suggest that the predicted relative positions of subunits are very accurate and the prediction quality is high.

[0072] IPAE: This parameter is used to assess the accuracy of the relative positions of residues in protein structure prediction. Lower IPAE scores indicate smaller prediction errors between the relative positions of two residues and higher confidence in the positions of these residues. The screening range was adjusted based on the IPAE scores of different template scaffold proteins and RuBisCO, and then screened for subsequent analysis.

[0073] RMSD: This parameter primarily compares the conformational changes of the scaffold protein before and after binding to RuBisCO. A smaller RMSD indicates a more stable scaffold protein structure, less susceptible to major conformational changes. Sequences with an RMSD value less than 2 Å were retained.

[0074] PDBsum interface analysis (this is a website outside of Alphafold, the URL is https: / / www.ebi.ac.uk / thornton-srv / databases / cgi-bin / pdbsum / GetPage.pl?pdbcode=index.html. It can be installed as a local version): After using the above parameters for screening, there may still be some sequences that meet the above parameter requirements, but the binding site is obviously wrong. In this case, it is necessary to manually screen the sequences whose binding position is the same as the binding position of the scaffold protein template, and use the PDBsum website to analyze the interaction interface between the scaffold protein and RuBisCO.

[0075] Finally, 20-30 optimal protein sequences were screened out and entered the subsequent wet experiment verification stage.

[0076] 3. Wet test verification

[0077] 3.1 Purification of RuBisCO

[0078] Oryza sativa L. spp. japonica leaves were ground and RuBisCO was finely purified from the leaf extracts by ion exchange and size exclusion chromatography.

[0079] (1) Crude purification: Freeze-dried rice leaves were ground in a mortar until pulverized. 100 mL of low-salt buffer (50 mM Tris pH 8.0, 50 mM NaCl, 1 mM EDTA, 1 mM DTT, 10 mM MgCl2) was added to extract the soluble contents, which were then filtered with gauze. The crude extract of the crop leaves was subjected to refrigerated high-speed centrifugation (12,000 rpm, 30 min, 4°C) in a centrifuge. The supernatant was the crude extract of the leaves, which was then finely purified.

[0080] (2) Anion exchange chromatography: The Hitrap Q HP (5 mL) chromatography column was pre-equilibrated with purification buffer, and the crude leaf extract was then loaded onto the chromatography column using a peristaltic pump. The sample was gradient eluted using a high salt buffer (50 mM Tris pH 8.0, 1 M NaCl, 1 mM EDTA, 1 mM DTT, 10 mM MgCl2). 20% to 30% of the eluted fraction was the target product.

[0081] (3) Size exclusion chromatography: The eluted product of anion exchange chromatography was subjected to SDS-PAGE electrophoresis, and the eluted sample with the highest purity was selected for size exclusion chromatography. A size exclusion chromatography column (HiLoad 16 / 600 Superdex 200 prepgrade) was pre-equilibrated with purification buffer (20 mM Tris pH 8.0, 50 NaCl, 5 mM MgCl2, 5% glycerol) for at least 1 column volume. After loading the purified product from the previous step, elution was continued for 1 column volume. The sample was collected based on the UV absorption spectrum at 280 nm. Finally, the sample was concentrated using a 100 kDa concentrator tube by low-temperature high-speed centrifugation (2500 rpm, 10 min / cycle, 4°C), quickly frozen in liquid nitrogen, and stored at -80°C for subsequent experiments.

[0082] 3.2 Scaffold protein purification

[0083] The scaffold protein with affinity tag was recombinantly expressed in Escherichia coli system and purified by nickel column affinity chromatography and molecular exclusion chromatography.

[0084] (1) Protein expression: Use the 42°C heat shock method to transform the exogenous protein recombinant expression plasmid into the Escherichia coli expression system (such as BL21(DE3), pKY206). After adding IPTG, induce the expression of the target protein at 16°C for 20-24 h or at 37°C for 4-6 h (Xia LY, Jiang YL, Kong WW, Sun H, Li WF, Chen Y, Zhou CZ. Molecular basis for the assembly of RuBisCO assisted by the chaperone Raf1. Nat Plants. 2020 Jun;6(6):708-717. doi: 10.1038 / s41477-020-0665-8. Epub 2020May 25. PMID: 32451445). The cells were harvested by low-temperature high-speed centrifugation (8000 rpm, 4 min, 4°C), resuspended in purification buffer (20 mM Tris pH 8.0, 50 NaCl, 5 mM MgCl2, 5% glycerol), and frozen at -20°C.

[0085] (2) Cell disruption: Dilute the bacterial suspension to 40 mL with purification buffer and disrupt the cells using an ultrasonic disruptor (power 40%, disruption for 2 s / interval for 2 s, disruption for 30 min) in an ice-water bath. Subsequently, the cell disruption solution was subjected to low-temperature high-speed centrifugation (12,000 rpm, 30 min, 4°C), and the supernatant was collected for affinity chromatography.

[0086] (3) Nickel column affinity chromatography: The centrifugal supernatant was transferred to a nickel chelate affinity chromatography column, and gradient concentrations of imidazole (0 mM, 20 mM, 50 mM) were added to the purification buffer for washing. Finally, the target product was eluted with a purification buffer containing 500 mM imidazole.

[0087] (4) Size Exclusion Chromatography: The eluted product from the nickel column was subjected to refrigerated high-speed centrifugation (12,000 rpm, 30 min, 4°C) in a centrifuge. The supernatant was loaded onto a pre-equilibrated size exclusion chromatography column (HiLoad 16 / 600 Superdex 75prep grade) and eluted for one column volume. The eluted sample was collected and concentrated using a 10 kDa concentrator tube by low-temperature high-speed centrifugation (2,000 rpm, 10 min / cycle, 4°C). The sample was quickly frozen in liquid nitrogen and stored at -80°C for subsequent experiments.

[0088] 3.3 Turbidity experiment

[0089] The 340 nm UV absorption of the mixed solution of RuBisCO and scaffold protein was measured to evaluate the ability of the scaffold protein to promote the turbidity formation of RuBisCO.

[0090] (1) Sample preparation: Calculate the molar concentration based on the mass volume concentration of RuBisCO (final concentration 0.5 / 1 / 2 μM) and scaffold protein (final concentration 5 / 10 / 20 μM) samples, and dilute to the target concentration using buffer (20 mM Tris pH 8.0, 50 mM NaCl, 5 mM MgCl2).

[0091] (2) Turbidity detection: Turbidity detection was performed using a DU800 spectrophotometer. 200 μL of buffer (20 mM Tris pH 8.0, 50 mM NaCl, 5 mM MgCl2) was added to the four chambers of a quartz cuvette and zeroed under 340 nm ultraviolet light. Subsequently, the absorption of 340 nm ultraviolet light by the mixed systems of 0.5 μM, 1 μM, and 2 μM RuBisCO with 5 μM, 10 μM, and 20 μM scaffold protein and the unmixed system (negative control) was continuously measured to reflect the formation of RuBisCO turbidity mediated by the scaffold protein. Data were measured every 12 s for a total of 20 min.

[0092] 3.4 Negative staining electron microscopy

[0093] Negative staining electron microscopy was used to detect the state of the complex between the scaffold protein and RuBisCO, confirming that it formed a RuBisCO concentration module.

[0094] (1) Sample preparation:

[0095] a. Copper mesh hydrophilization: Take a copper mesh with a common carbon support film, place it in an ultrasonic cleaning machine (Plasma System, Gatan) and glow discharge it for 10 seconds, then remove it and set aside.

[0096] b. Place 3.5 μL of purified RubisCO / scaffold protein sample or 1.5 μL of turbidity assay sample (the turbidity formed after the mixture of RuBisCO and scaffold protein is mixed) on a copper mesh, wait for 1 minute, and remove excess liquid with filter paper.

[0097] c. Add 7.5 μL of uranyl acetate (1% w / v) onto the copper grid to wash away excess buffer from the sample and immediately remove excess liquid with filter paper.

[0098] d. Add 7.5 μL of uranyl acetate (1% w / v) solution to the copper grid for counterstaining. Wait 1 minute, remove excess uranyl acetate with filter paper, and let it stand for 5 minutes to dry.

[0099] (2) Electron microscopy observation: Negatively stained samples were observed using a 120 kV transmission electron microscope (Technai G2 120 kV).

[0100] 3.5 Detection of phase separation components

[0101] Samples from turbidity experiments were analyzed by SDS-PAGE and quantitative mass spectrometry to assess the formation of phase-separated components.

[0102] (1) SDS-PAGE: Turbidity detection samples were added to protein electrophoresis buffer (60 mM Tris-HCl pH 6.8, 2% SDS, 0.1% bromophenol blue, 25% glycerol, 14.4 mM β-mercaptoethanol) and then loaded onto a 15% SDS-PAGE for electrophoresis. Electrophoresis was performed in Tris-Glycine buffer at a constant voltage of 250 V for 35 min. The gel was stained with Coomassie Brilliant Blue and imaged using an imager.

[0103] (2) Grayscale integration: Use Image J to perform grayscale integration on the sample bands in each lane of the electrophoresis detection imaging result diagram to analyze the stoichiometric ratios of different components in each sample.

[0104] (3) Quantitative mass spectrometry: Liquid chromatography-mass spectrometry (LC-MS) system. The protein sample was reduced with 100 mM DTT for 1 h at 37°C, then placed in the dark at room temperature and UV-irradiated, alkylated with 55 mM iodoacetamide for 15 min, and then transferred to a Microcon YM-30 (centrifugal filter device). The lysis buffer was replaced with 200 μL UA (8 M urea (Urea), 100 mM Tris-HCl pH 8.5) twice. The UA buffer was then replaced with 0.1 M triethylammonium bicarbonate (TEAB, Sigma-Aldrich) and digested with sequencing-grade trypsin 1:50 (v / v) at 37°C overnight. The resulting peptides were desalted with a StageTip, completely dried with a SpeedVac concentrator, and stored at -20°C for further analysis.

[0105] 3.6 Enzyme activity determination

[0106] The RuBisCO activity was determined using the Solebro ribulose diphosphate carboxylase / oxygenase (RuBisCO) activity assay kit (BC0440). The effects of different scaffold proteins on RuBisCO activity were calculated and compared.

[0107] Here are the steps:

[0108] (1) Preheat the spectrophotometer / microplate reader for at least 30 minutes, adjust the wavelength to 340 nm, and zero with distilled water.

[0109] (2) Add samples according to the steps in the table below:

[0110] Table 2 Sample addition system

[0111]

[0112] (Note: Reagent 3, Reagent 4 and working solution are all components of the activity detection kit)

[0113] (3) Record the absorbance at 340 nm at 20 seconds (A1) and at 5 minutes (A2) for 20 seconds. Calculate ΔA_test = A1_test - A2_test, ΔA_blank = A1_blank - A2_blank, and ΔA_test = ΔA_test - ΔA_blank. Maintain the reaction temperature at 25°C. (Only use 1-2 blank tubes.)

[0114] (4) Calculate RuBisCO activity according to sample protein concentration: RuBisCO activity (U / mg prot) = [ΔA x V total ÷ (ε xd) x 10 9] ÷ (Cpr x Vsample) ÷ T = 344×ΔA ÷ Cpr. (One unit of enzyme activity is the oxidation of 1 nmol NADH per milligram of protein per minute at 25°C)

[0115] V total: total volume of the reaction system, 1.07×10 -3 L; ε: NADH molar extinction coefficient, 6.22×10 3 L / mol / cm; d: cuvette optical path, 1 cm; V: sample volume added, 0.1 mL; T: reaction time, 5 min; 10 9 :Unit conversion factor, 1 mol=10 9 nmol; Cpr: protein concentration, mg / mL

[0116] 4. Result Processing and Analysis

[0117] Based on the wet lab results, a scaffold protein was identified that could be stably expressed and purified, effectively bind to crop RuBisCO, stably form a concentrating module, and enhance RuBisCO enzymatic activity. This scaffold protein can be used to construct RuBisCO concentrating modules in crops, and further validated and applied in vivo.

[0118] Example 1: Design of a Rice RuBisCO Concentration Module Based on the β-cyanobacterial Scaffold Protein CcmM

[0119] (1) Construction of protein sequence dataset based on AI design and homology search

[0120] Based on the reported structure of the β-cyanobacterial scaffold protein CcmM in complex with RuBisCO, we identified CcmM hotspot residues during RF diffusion. We generated new sequences using MPNN and simultaneously searched for CcmM homologous proteins using BLAST, ultimately obtaining 3,030 sequences for subsequent screening.

[0121] (2) Evaluation of scoring indicators of datasets

[0122] Screening was performed using the following criteria: Distance < 10 Å, ipTM > 0.8, ipae < 10, and RMSD < 2 Å. After screening using these four criteria, 1280 sequences still met the requirements. Therefore, PDBsum was used to score the interfaces between these scaffold proteins and RuBisCO. Sequences with a total interaction interface area greater than 1000 Å were selected. 2 There are 45 sequences in total (because the total interface area of ​​the existing scaffold proteins binding to RuBisCO is about 1000 Å 2), and then 11 sequences with the largest number of salt bridges and hydrogen bonds were selected and named P1-P11 (PDBsum related results of the 11 sequences are shown in Table 3) for wet experiment verification.

[0123] Table 3 PDBsum analysis results of 11 sequences used for wet experiment verification, including the interaction interface between the scaffold protein sequence and RuBisCO and the polar interactions (hydrogen bonds and salt bridges) formed.

[0124]

[0125] (3) Wet test verification

[0126] a. Purification of Rice RuBisCO: 20g of fresh rice leaves were freeze-dried with liquid nitrogen and ground in a mortar until finely pulverized. After adding buffer, the crude extract was filtered through gauze. The crude extract was subjected to refrigerated high-speed centrifugation to remove insoluble components. The supernatant was then subjected to anion exchange chromatography and molecular sieve chromatography to collect samples corresponding to the RuBisCO protein peak in the crop. The samples were concentrated and used in subsequent experiments. Figure 1 shown. Figure 1 The results showed that the bands corresponding to the large subunit (~53 kD) and small subunit (~15 kD) of RuBisCO were both detected, preliminarily proving that the RuBisCO holoenzyme was successfully obtained.

[0127] b. Purification of scaffold proteins: The 11 scaffold protein genes were synthesized and E. coli expression plasmids were constructed. The 11 scaffold proteins were expressed in E. coli. The E. coli cells were harvested and disrupted by ultrasound. The disrupted liquid was centrifuged at high speed and the supernatant was extracted. The corresponding scaffold protein samples were obtained after nickel column affinity chromatography and molecular exclusion chromatography. The samples were concentrated and used in subsequent experiments. Figure 2 shown. Figure 2 The results showed that SDS-PAGE detected target bands (~47kD) corresponding to 11 scaffold proteins, indicating that we successfully obtained 11 scaffold protein samples.

[0128] c. Turbidity experiment: Turbidity experiment can detect protein aggregation or phase separation by measuring the light scattering of the solution, and determine the phase separation ability of the protein under different conditions by analyzing the absorbance change. Using the protein sample obtained by the above purification, the crop RuBisCO protein with a final concentration of 1 μM was mixed with 10 μM samples of different scaffold proteins, and the absorbance change at 340 nm was measured. The results showed that the designed P8 protein has the ability to form strong turbidity with rice RuBisCO. The results are as follows Figure 3 shown. Figure 3The results showed that the addition of different scaffold proteins caused RuBisCO to aggregate to varying degrees, leading to an increase in absorbance. The increases were more pronounced for the P8, P6, and P3 scaffold proteins compared to the other scaffold proteins, indicating that all three scaffold proteins can interact with RuBisCO to form a RuBisCO concentrating module, with P8 being the most potent.

[0129] d. Negative staining electron microscopy: The experimental group samples obtained from the turbidity experiment were diluted and prepared into negative staining electron microscopy samples. It was observed that the scaffold protein mediated multiple RuBisCO molecules to form chain structures and effectively form aggregates. Figure 4 shown.

[0130] (4) Result processing and analysis

[0131] Based on the above wet experiment verification results, we finally designed a new scaffold protein that can mediate the effective aggregation of rice RuBisCO and is an ideal rice RuBisCO concentration module.

[0132] sequence

[0133] SEQ ID NO: 1 Amino acid sequence of P8 protein (the flexible motif connecting adjacent scaffold protein modules is underlined)

[0134]

[0135] The specific embodiments described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above are only specific embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for designing a scaffold protein that mediates the aggregation of RuBisCO in crops, characterized in that: The method uses a known scaffold protein as a template and generates a sequence through the following steps: a. RFdiffusion scaffold protein backbone generation: Based on the known complex structure of the scaffold protein and the corresponding RuBisCO, the interaction interface between the two is determined. The residues at the scaffold protein interface are set as hotspot residues during diffusion. The length of the sequence generated by diffusion in the hotspot residue region is limited to the same length as the known scaffold protein, while the amino acid residues at other positions remain unchanged to generate the scaffold protein backbone. b. MPNN sequence generation: The scaffold protein backbone generated in step a is filled with amino acid side chains to ensure structural conformation stability. The scaffold protein backbone generated in step a is matched to the known binding sites of the scaffold protein and RuBisCO, taking into account the interaction between the scaffold protein and RuBisCO and steric hindrance, thereby generating the scaffold protein sequence. c. AlphaFold structure prediction: The generated scaffold protein sequence is used to predict the complex structure with the crop RuBisCO to obtain the interaction interface information between the scaffold protein and RuBisCO; d. Evaluate the generated scaffold protein sequences using one or more scoring metrics to select the best sequence.

2. The method according to claim 1, characterized in that The scoring indicators include Distance, ipTM, ipae and RMSD.

3. The method according to claim 2, characterized in that Sequences with a distance less than 10 Å were selected.

4. The method according to claim 2, characterized in that Sequences with an ipTM score greater than 0.8 were selected.

5. The method according to claim 2, characterized in that Select sequences with IPAE less than 10 or less than 20.

6. The method according to claim 2, characterized in that Sequences with an RMSD less than 2 Å were selected.

7. The method according to any one of claims 1 to 6, characterized in that The known scaffold protein is selected from the group consisting of CsoS2 protein from α-cyanobacteria, CcmM protein from β-cyanobacteria, EPYC1 protein from Chlamydomonas reinhardtii, and PYCO1 protein from diatoms.

8. The method according to any one of claims 1 to 6, characterized in that The method further includes, after step d, rescreening a sequence whose binding site is identical to a known scaffold protein binding site, and analyzing the interaction interface between the scaffold protein and RuBisCO using the PDBsum website to screen out a scaffold protein sequence that mediates the aggregation of crop RuBisCO.

9. A scaffold protein obtained by screening according to any one of claims 1 to 8.

10. The scaffold protein according to claim 9, characterized in that The sequence of the scaffold protein is shown in SEQ ID NO:

1.

11. Use of the scaffold protein according to claim 9 or 10 in increasing crop yield.

Images