Homodimeric Protein Research Articles

An engineered transforming growth factor β (TGF-β) monomer that functions as a dominant negative to block TGF-β signaling

The transforming growth factor β isoforms, TGF-β1, -β2, and -β3, are small secreted homodimeric signaling proteins with essential roles in regulating the adaptive immune system and maintaining the extracellular matrix. However, dysregulation of the TGF-β pathway is responsible for promoting the progression of several human diseases, including cancer and fibrosis. Despite the known importance of TGF-βs in promoting disease progression, no inhibitors have been approved for use in humans. Herein, we describe an engineered TGF-β monomer, lacking the heel helix, a structural motif essential for binding the TGF-β type I receptor (TβRI) but dispensable for binding the other receptor required for TGF-β signaling, the TGF-β type II receptor (TβRII), as an alternative therapeutic modality for blocking TGF-β signaling in humans. As shown through binding studies and crystallography, the engineered monomer retained the same overall structure of native TGF-β monomers and bound TβRII in an identical manner. Cell-based luciferase assays showed that the engineered monomer functioned as a dominant negative to inhibit TGF-β signaling with a Ki of 20-70 nm Investigation of the mechanism showed that the high affinity of the engineered monomer for TβRII, coupled with its reduced ability to non-covalently dimerize and its inability to bind and recruit TβRI, enabled it to bind endogenous TβRII but prevented it from binding and recruiting TβRI to form a signaling complex. Such engineered monomers provide a new avenue to probe and manipulate TGF-β signaling and may inform similar modifications of other TGF-β family members.

Asymmetric configurations in a reengineered homodimer reveal multiple subunit communication pathways in protein allostery

Many allosteric proteins form homo-oligomeric complexes to regulate a biological function. In homo-oligomers, subunits establish communication pathways that are modulated by external stimuli like ligand binding. A challenge for dissecting the communication mechanisms in homo-oligomers is identifying intermediate liganded states, which are typically transiently populated. However, their identities provide the most mechanistic information on how ligand-induced signals propagate from bound to empty subunits. Here, we dissected the directionality and magnitude of subunit communication in a reengineered single-chain version of the homodimeric transcription factor cAMP receptor protein. By combining wild-type and mutant subunits in various asymmetric configurations, we revealed a linear relationship between the magnitude of cooperative effects and the number of mutant subunits. We found that a single mutation is sufficient to change the global allosteric behavior of the dimer even when one subunit was wild type. Dimers harboring two mutations with opposite cooperative effects had different allosteric properties depending on the arrangement of the mutations. When the two mutations were placed in the same subunit, the resulting cooperativity was neutral. In contrast, when placed in different subunits, the observed cooperativity was dominated by the mutation with strongest effects over cAMP affinity relative to wild type. These results highlight the distinct roles of intrasubunit interactions and intersubunit communication in allostery. Finally, dimers bound to either one or two cAMP molecules had similar DNA affinities, indicating that both asymmetric and symmetric liganded states activate DNA interactions. These studies have revealed the multiple communication pathways that homo-oligomers employ to transduce signals.

Insights into the Catalytic Mechanism and Transition State Stabilization of 5′‐Methylthioadenosine Nucleosidase using Neutron Crystallography

5′‐Methylthioadenosine nucleosidase (MTAN) is an obligatory homodimeric protein that catalyzes the hydrolysis of the N‐ribosidic bond of a variety of adenosine‐containing metabolites. The proposed catalytic mechanism of MTAN is initiated by D198 protonating the N7 position of the adenyl moiety, which consequently causes elongation and ultimate breaking of the N‐ribosidic bond, resulting in the formation of an oxocarbenium ion intermediate. Of particular interest is the pKa elevation of D198 that was proposed to have a value of 8.2, which is significantly higher than the theoretical pKa for an aspartic acid side chain. In this study we present the first joint X‐ray/neutron crystal structures of Helicobacter pylori MTAN (HpMTAN) in complex with a substrate (S‐adenosylhomocysteine), early (Formycin A) and late (p‐ClPh‐Thio‐DADMe‐ImmA) dissociative transition state analogues, as well as with the catalytic products (adenine and S‐ribosylhomocysteine). The solved X‐ray/neutron crystal structures provide the positions of exchangeable hydrogen atoms and protonation states of ionizable groups that further our understanding of the catalytic mechanism as well as the stabilization of the transition state for HpMTAN. These results confirm that the catalytic reaction is initiated by D198 functioning as a general acid through the observation of a shared deuterium atom with unconventional hydrogen bond geometry between the carboxylate group of D198 and the N7 position of adenine. Additionally, the transition state analogues neutron structures illustrate the minor differences in the hydrogen bond interactions resulting in the DADMe‐ImmA analogues being extremely potent inhibitors with Kd values in the picomolar range, despite HpMTAN shown to contain an early dissociative transition state structure.Support or Funding InformationThis work was supported by the Center for the Advancement of Science in Space via a cooperative agreement with National Aeronautics and Space Administration Grant N‐123528‐01 (to D.R.R) and by National Institute of Allergy and Infectious Disease/NIH Grant AI105084 (to D.R.R.).

Cell-free reconstitution reveals centriole cartwheel assembly mechanisms

How cellular organelles assemble is a fundamental question in biology. The centriole organelle organizes around a nine-fold symmetrical cartwheel structure typically ∼100 nm high comprising a stack of rings that each accommodates nine homodimers of SAS-6 proteins. Whether nine-fold symmetrical ring-like assemblies of SAS-6 proteins harbour more peripheral cartwheel elements is unclear. Furthermore, the mechanisms governing ring stacking are not known. Here we develop a cell-free reconstitution system for core cartwheel assembly. Using cryo-electron tomography, we uncover that the Chlamydomonas reinhardtii proteins CrSAS-6 and Bld10p together drive assembly of the core cartwheel. Moreover, we discover that CrSAS-6 possesses autonomous properties that ensure self-organized ring stacking. Mathematical fitting of reconstituted cartwheel height distribution suggests a mechanism whereby preferential addition of pairs of SAS-6 rings governs cartwheel growth. In conclusion, we have developed a cell-free reconstitution system that reveals fundamental assembly principles at the root of centriole biogenesis.

Silencing of galectin-1 inhibits retinal neovascularization and ameliorates retinal hypoxia in a murine model of oxygen-induced ischemic retinopathy

Aberrant neovascularization is a consequence of inappropriate angiogenic signaling and contributes to several diseases. Although many regulators of pathogenic angiogenesis have been identified, the understanding of this process remains incomplete. Galectin-1 (Gal-1), as a homodimeric protein with a single carbohydrate-recognition domain, is implicated in several pathologic processes, including angiogenesis; however, its involvement in retinal neovascularization (RNV) remains unknown. Here, we investigated the anti-angiogenic effect of silencing Gal-1 through intravitreal injection in a mouse model of oxygen-induced retinopathy (OIR). Our results revealed that Gal-1 was overexpressed and closely related to retinal neo-vessels in OIR retinas. After silencing Gal-1 via intravitreal injection of adenoviral-Gal-1-RNA interference (Ad-Gal-1-RNAi), RNV and retinal hypoxia were significantly attenuated, indicating the anti-angiogenic effect of Gal-1 inhibition. Western blot analysis and real-time polymerase chain reaction indicated that the expression of both neuropilin-1 (Nrp-1) and B cell lymphoma-2 (Bcl-2) decreased after intravitreal injection of Ad-Gal-1-RNAi, implying the possible involvement of Nrp-1 and Bcl-2 in Gal-1-related angiogenic processes. Additionally, whole-mount fluorescence and hematoxylin and eosin staining showed that intravitreal injection of Ad-Gal-1-RNAi did not significantly disrupt the retinal vasculature and neuronal structure of room air mice. Moreover, Ad-Gal-1-RNAi transfer promoted retinal vascular sprouting and increased retinal vascular perfusion, likely through decreased phosphorylation of myosin phosphatase target protein-1. Collectively, our results demonstrated that Gal-1 functions as an important regulator in RNV and offers a promising strategy for the treatment of RNV diseases, such as proliferative diabetic retinopathy and retinopathy of prematurity.

Structural and biochemical studies on Vibrio cholerae Hsp31 reveals a novel dimeric form and Glutathione-independent Glyoxalase activity.

Vibrio cholerae experiences a highly hostile environment at human intestine which triggers the induction of various heat shock genes. The hchA gene product of V. cholerae O395, referred to a hypothetical intracellular protease/amidase VcHsp31, is one such stress-inducible homodimeric protein. Our current study demonstrates that VcHsp31 is endowed with molecular chaperone, amidopeptidase and robust methylglyoxalase activities. Through site directed mutagenesis coupled with biochemical assays on VcHsp31, we have confirmed the role of residues in the vicinity of the active site towards amidopeptidase and methylglyoxalase activities. VcHsp31 suppresses the aggregation of insulin in vitro in a dose dependent manner. Through crystal structures of VcHsp31 and its mutants, grown at various temperatures, we demonstrate that VcHsp31 acquires two (Type-I and Type-II) dimeric forms. Type-I dimer is similar to EcHsp31 where two VcHsp31 monomers associate in eclipsed manner through several intersubunit hydrogen bonds involving their P-domains. Type-II dimer is a novel dimeric organization, where some of the intersubunit hydrogen bonds are abrogated and each monomer swings out in the opposite directions centering at their P-domains, like twisting of wet cloth. Normal mode analysis (NMA) of Type-I dimer shows similar movement of the individual monomers. Upon swinging, a dimeric surface of ~400Å2, mostly hydrophobic in nature, is uncovered which might bind partially unfolded protein substrates. We propose that, in solution, VcHsp31 remains as an equilibrium mixture of both the dimers. With increase in temperature, transformation to Type-II form having more exposed hydrophobic surface, occurs progressively accounting for the temperature dependent increase of chaperone activity of VcHsp31.

Formation and carbon monoxide-dependent dissociation of Allochromatium vinosum cytochrome c' oligomers using domain-swapped dimers.

The number of artificial protein supramolecules has been increasing; however, control of protein oligomer formation remains challenging. Cytochrome c' from Allochromatium vinosum (AVCP) is a homodimeric protein in its native form, where its protomer exhibits a four-helix bundle structure containing a covalently bound five-coordinate heme as a gas binding site. AVCP exhibits a unique reversible dimer-monomer transition according to the absence and presence of CO. Herein, domain-swapped dimeric AVCP was constructed and utilized to form a tetramer and high-order oligomers. The X-ray crystal structure of oxidized tetrameric AVCP consisted of two monomer subunits and one domain-swapped dimer subunit, which exchanged the region containing helices αA and αB between protomers. The active site structures of the domain-swapped dimer subunit and monomer subunits in the tetramer were similar to those of the monomer subunits in the native dimer. The subunit-subunit interactions at the interfaces of the domain-swapped dimer and monomer subunits in the tetramer were also similar to the subunit-subunit interaction in the native dimer. Reduced tetrameric AVCP dissociated to a domain-swapped dimer and two monomers upon CO binding. Without monomers, the domain-swapped dimers formed tetramers, hexamers, and higher-order oligomers in the absence of CO, whereas the oligomers dissociated to domain-swapped dimers in the presence of CO, demonstrating that the domain-swapped dimer maintains the CO-induced subunit dissociation behavior of native ACVP. These results suggest that protein oligomer formation may be controlled by utilizing domain swapping for a dimer-monomer transition protein.

Bifunctional Inhibitors as a New Tool To Reduce Cancer Cell Invasion by Impairing MMP-9 Homodimerization.

Protein homodimers play important roles in physiological and pathological processes, including cancer invasion and metastasis. Recently, MMP-9 natural homodimerization via the PEX domain has been correlated with high migration rates of aggressive cancer cells. Here we propose that bifunctional MMP-9 inhibitors designed to impair natural MMP-9 homodimerization promoted by PEX-PEX interactions might be an effective tool to fight cancer cell invasion. Elaborating a previously described dimeric hydroxamate inhibitor 1, new ligands were synthesized with different linker lengths and branch points. Evaluation of the modified bifunctional ligands by X-ray crystallography and biological assays showed that 7 and 8 could reduce invasion in three glioma cell lines expressing MMP-9 at different levels. To rationalize these results, we present a theoretical model of full-length MMP-9 in complex with 7. This pioneering study suggests that a new approach using MMP-9 selective bifunctional inhibitors might lead to an effective therapy to reduce cancer cell invasion.

Identification of Critical Residues for Anion Conductance in the TMEM16A Channel

TMEM16A (also called DOG-1 and Anoctamin-1) is a transmembrane-localized, homodimeric protein that makes up the pore forming subunit of a calcium-activated chloride channel found in epithelial cells, tumor cells and a subset of DRG neurons. However, the structural and biophysical properties of this channel remain poorly understood. Here, we aimed to identify residues that determine the conductance properties of TMEM16A, by using site-directed mutagenesis and heterologous expression in HEK293 cells followed by patch clamp electrophysiology. Using a recently crystallized fungal orthologue of the TMEM16 family as a model, we identified several residues clustered around the putative pore region that are critical for normal channel conductance of anions. Mutations at these residues also influenced the binding properties of several known channel inhibitors. We conclude that the identified residues are likely to be directly involved in dictating the physical properties of the anion conduction pathway, and may also help to confirm the as yet unknown location of the TMEM16A channel's anion pore.

Preparation and Characterization of a Novel Monoclonal Antibody Against the Extracellular Domain of Human Transferrin Receptor.

Human transferrin receptor (hTfR1), a type II homodimeric transmembrane protein, is extensively expressed at low levels in most normal human tissues. However, the expression level of hTfR1 in some tumor tissues and the blood-brain barrier (BBB) is comparatively higher. Therefore, we developed an hTfR1 monoclonal antibody 2C9, which may be utilized to deliver drugs across the BBB through receptor-mediated endocytosis in the future. 2C9 was obtained by hybridoma cells fused by the SP2/0-Ag14 cell line and splenic B cells from hTfR1-immunized BALB/c mice. In this study, we used indirect ELISA, Western blot, and immunofluorescence to explore the characterization and application of MAb 2C9. The results showed that the affinity constant (Kaff) of the MAb 2C9 produced from ascites was 2.85 × 10-8 M and its isotype is IgG2a (Kappa chain). Above all, this report provides a more comprehensive protocol to produce monoclonal antibodies against the extracellular domain of hTfR1.

Methyl-Labeling Assisted NMR Structure Determination of a 66 KDA Growth Factor-Receptor Complex

TGF-β1, TGF- β2, and TGF-β3 are 26 kDa disulfide-linked homodimeric signaling proteins. They all signal through the TGF-β type I and type II receptors, yet TGF-β2, which is well known to bind TβRII several-hundred fold more weakly than TGF-β1 and TGF-β3, has an additional requirement for the TGF-β type III receptor (TβRIII), a membrane-anchored non-signaling receptor that potentiates the binding of TβRII. Though it is known that TβRIII has two component domains that bind TGF-β2 non-cooperatively at independent sites, the structure of these domains bound to TGF-β2 and residues responsible for specific binding are not yet known.

Ciliary entry of KIF17 is dependent on its binding to the IFT-B complex via IFT46-IFT56 as well as on its nuclear localization signal.

Cilia function as cellular antennae to sense and transduce extracellular signals. A number of proteins are specifically localized in cilia. Anterograde and retrograde ciliary protein trafficking are mediated by the IFT-B and IFT-A complexes in concert with kinesin-2 and dynein-2 motors, respectively. However, the role of KIF17, a homodimeric kinesin-2 protein, in protein trafficking has not been fully understood in vertebrate cilia. In this study, we demonstrated, by using the visible immunoprecipitation assay, that KIF17 interacts with the IFT46-IFT56 dimer in the IFT-B complex through its C-terminal sequence located immediately upstream of the nuclear localization signal (NLS). We then showed that KIF17 reaches the ciliary tip independently of its motor domain and requires IFT-B binding for its entry into cilia rather than for its intraciliary trafficking. We further showed that KIF17 ciliary entry depends not only on its binding to IFT-B but also on its NLS, to which importin α proteins bind. Taking the results together, we conclude that in mammalian cells, KIF17 is dispensable for ciliogenesis and IFT-B trafficking but requires IFT-B, as well as its NLS, for its ciliary entry across the permeability barrier located at the ciliary base.

In Vitro Assembly and Stabilization of Dengue and Zika Virus Envelope Protein Homo-Dimers

In Vitro Assembly and Stabilization of Dengue and Zika Virus Envelope Protein Homo-Dimers

Multivalency Effects on the Immobilization of Sucrose Phosphorylase in Flow Microchannels and Their Use in the Development of a High‐Performance Biocatalytic Microreactor

AbstractMicrostructured reactors are emerging engineering tools for the development of biocatalytic conversions in continuous flow. A promising layout involves flow microchannels that are wall‐coated with enzyme. As protein immobilization within closed microstructures is challenging, we suggested a confluent design of enzyme and microreactor: fusion to the silica‐binding module Zbasic2 is used to engineer enzymes for high‐affinity oriented attachment to the plain wall surface of glass microchannels. In this study of sucrose phosphorylase, we examined the effects of multiple Zbasic2 modules in a single enzyme molecule on the activity and adsorption stability of the phosphorylase immobilized in a glass microchannel reactor. Compared to the “monovalent” enzyme, two Zbasic2 modules, present in tandem repeat at the N‐terminus, separated at the N‐ and C‐terminus of an enzyme monomer, or arranged N‐terminally in a protein homodimer, boosted the effectiveness of the immobilized phosphorylase by up to twofold. They attenuated (up to 12‐fold) the elution of the wall‐coated enzyme during continuous reactor operation. The divalent phosphorylase was distributed uniformly on the microchannel surface and approximately 70 % activity could still be retained after 690 reactor cycles. Reaction–diffusion regime analysis in terms of the second Damköhler number (DaII≤0.02) revealed the absence of mass transport limitations on the conversion rate. The synthesis of α‐d‐glucose 1‐phosphate occurred with a productivity of ∼14 mm min−1 at 50 % substrate conversion (50 mm). The use of wall‐coated enzyme microreactors in high‐performance biocatalytic reaction engineering is supported strongly.

Mutations at the Qo Site of the Cytochrome bc1 Complex Strongly Affect Oxygen Binding.

The homodimeric bc1 protein complex is embedded in membranes of mitochondria and photosynthetic bacteria, where it transports protons across the membrane to maintain an electrostatic potential used to drive ATP synthesis as part of the respiratory or photosynthetic pathways. The reaction cycle of the bc1 complex is driven by series of redox processes involving substrate molecules from the membrane, but occasional side reactions between an intermediate semiquinone substrate and molecular oxygen are suspected to be a source of toxic superoxide, which is believed to be a factor in aging. The present investigation employs molecular dynamics simulations to study the effect of mutations in the Qo binding sites of the bc1 complex on the ability of oxygen molecules to migrate to and bind at various locations within the complex. It is found that the mutations strongly affect the ability of oxygen to bind at the Qo sites, and moreover, different behavior of the two monomers of the bc1 complex is observed. The conformational differences at the Qo sites of the two monomers are studied in detail and discussed. The anionic form of semiquinone was identified as leading to the greatest opportunity for side reactions with oxygen.

Regulatory Conformational Coupling between CLC Anion Channel Membrane and Cytoplasmic Domains

CLC anion channels are homodimeric proteins. Each subunit is comprised of 18 α-helices designated “A-R” and an intracellular carboxy-terminus containing two cystathionine-β-synthase (CBS1 and CBS2) domains. Conformational coupling between membrane and intracellular domains via poorly understood mechanisms is required for CLC regulation. The activity of the C. elegans CLC channel CLH-3b is reduced by phosphorylation of a carboxy-terminus “activation domain,” which disrupts its interaction with CBS domains. CBS2 interfaces with a short intracellular loop, the H-I loop, connecting membrane helices H and I. Alanine mutation of a conserved H-I loop tyrosine residue, Y232, prevents regulation demonstrating that the loop functions to couple phosphorylation-dependent CBS domain conformational changes to channel membrane domains. To gain further insight into the mechanisms of this coupling, we mutated conserved amino acid residues in membrane helices H and I. Only mutation of the H-helix valine residue V228 to leucine prevented phosphorylation-dependent channel regulation. Structural and functional studies of other CLC proteins suggest that V228 may interact with Y529, a conserved R-helix tyrosine residue that forms part of the CLC ion conduction pathway. Mutation of Y529 to alanine also prevented CLH-3b regulation. Intracellular application of the sulfhydryl reactive reagent MTSET using CLH-3b channels engineered with single-cysteine residues in CBS2 indicate that V228L, Y529A, and Y232A disrupt putative regulatory intracellular conformational changes. Extracellular Zn2+ inhibits CLH-3b and alters the effects of intracellular MTSET on channel activity. The effects of Zn2+ are disrupted by V228L, Y529A, and Y232A. Collectively, our findings indicate that there is conformational coupling between CBS domains and the H and R membrane helices mediated by the H-I loop. We propose a simple model by which conformational changes in H and R helices mediate CLH-3b regulation by activation domain phosphorylation.

Role of CBS and Bateman Domains in Phosphorylation-Dependent Regulation of a CLC Anion Channel

Eukaryotic CLC anion channels and transporters are homodimeric proteins composed of multiple α-helical membrane domains and large cytoplasmic C-termini containing two cystathionine-β-synthase domains (CBS1 and CBS2) that dimerize to form a Bateman domain. The Bateman domains of adjacent CLC subunits interact to form a Bateman domain dimer. The functions of CLC CBS and Bateman domains are poorly understood. We utilized the Caenorhabditis elegans CLC-1/2/Ka/Kb anion channel homolog CLH-3b to characterize the regulatory roles of CLC cytoplasmic domains. CLH-3b activity is reduced by phosphorylation or deletion of a 14-amino-acid activation domain (AD) located on the linker connecting CBS1 and CBS2. We demonstrate here that phosphorylation-dependent reductions in channel activity require an intact Bateman domain dimer and concomitant phosphorylation or deletion of both ADs. Regulation of a CLH-3b AD deletion mutant is reconstituted by intracellular perfusion with recombinant 14-amino-acid AD peptides. The sulfhydryl reactive reagent 2-(trimethylammonium)ethyl methanethiosulfonate bromide (MTSET) alters in a phosphorylation-dependent manner the activity of channels containing single cysteine residues that are engineered into the short intracellular loop connecting membrane α-helices H and I (H-I loop), the AD, CBS1, and CBS2. In contrast, MTSET has no effect on channels in which cysteine residues are engineered into intracellular regions that are dispensable for regulation. These studies together with our previous work suggest that binding and unbinding of the AD to the Bateman domain dimer induces conformational changes that are transduced to channel membrane domains via the H-I loop. Our findings provide new, to our knowledge, insights into the roles of CLC Bateman domains and the structure-function relationships that govern the regulation of CLC protein activity by diverse ligands and signaling pathways.

Modeling and minimizing CAPRI round 30 symmetrical protein complexes from CASP-11 structural models.

Many of the modeling targets in the blind CASP-11/CAPRI-30 experiment were protein homo-dimers and homo-tetramers. Here, we perform a retrospective docking-based analysis of the perfectly symmetrical CAPRI Round 30 targets whose crystal structures have been published. Starting from the CASP "stage-2" fold prediction models, we show that using our recently developed "SAM" polar Fourier symmetry docking algorithm combined with NAMD energy minimization often gives acceptable or better 3D models of the target complexes. We also use SAM to analyze the overall quality of all CASP structural models for the selected targets from a docking-based perspective. We demonstrate that docking only CASP "center" structures for the selected targets provides a fruitful and economical docking strategy. Furthermore, our results show that many of the CASP models are dockable in the sense that they can lead to acceptable or better models of symmetrical complexes. Even though SAM is very fast, using docking and NAMD energy minimization to pull out acceptable docking models from a large ensemble of docked CASP models is computationally expensive. Nonetheless, thanks to our SAM docking algorithm, we expect that applying our docking protocol on a modern computer cluster will give us the ability to routinely model 3D structures of symmetrical protein complexes from CASP-quality models. Proteins 2017; 85:463-469. © 2016 Wiley Periodicals, Inc.

Disrupting the Constitutive, Homodimeric Protein-Protein Interface in CK2β Using a Biophysical Fragment-Based Approach.

Identifying small molecules that induce the disruption of constitutive protein–protein interfaces is a challenging objective. Here, a targeted biophysical screening cascade was employed to specifically identify small molecules that could disrupt the constitutive, homodimeric protein–protein interface within CK2β. This approach could potentially be applied to achieve subunit disassembly of other homo-oligomeric proteins as a means of modulating protein function.

Deposition mode of transforming growth factor-β expressed in transgenic rice seed.

Mouse TGF-β highly accumulated by expressing as a secretory homodimeric protein in transgenic rice endosperm. It was tightly deposited in ER-derived PBs by interaction with cysteine-rich prolamins. TGF-β is one of the key players involved in the induction and maintenance of mucosal immune tolerance to dietary proteins through the induction of regulatory T cells. In order to utilize rice-based TGF-β as a tool to promote oral immune tolerance induction, high production of TGF-β is essentially required. When the codon-optimized mTGF-β was expressed as a secretory protein by ligating an N-terminal signal peptide and C-terminal KDEL ER retention signal under the control of the endosperm-specific rice storage protein glutelin GluB-1 promoter, accumulation level was low in stable transgenic rice seeds. Then, to increase the accumulation level of mTGF-β, it was expressed as fusion proteins by inserting into the C terminus of acidic subunit of glutelin GluA and the variable region of 26kDa globulin. When fused with the glutelin, it could accumulate well as visible bands by CBB staining gel, but not for the 26kDa globulin. Unexpectedly, expression of homodimeric mTGF-β linked by a 6×Gly1×Ser linker as secretory protein resulted in higher level of accumulation. This expression level was further enhanced by reduction of some endogenous prolamins by RNA interference. The monomeric and dimeric mTGF-βs were deposited in ER-derived PBs containing prolamins. When highly produced in rice seed, it is notable that most of ER-derived PBs were distorted and granulated. Step-wise extraction of storage proteins from rice seeds suggested that the mTGF-β strongly interacted with cysteine-rich prolamins via disulfide bonds. This result was also supported by the finding that reducing agent was absolutely required for mTGF-β extraction.