Isolation and sequence analyses of cDNA clones for the large subunits of two isozymes of rabbit calcium-dependent protease

Two sets of cDNA clones were isolated from cDNA libraries prepared from poly(A+) RNA of rabbit lung and spleen by screening with the cDNA probe for the large subunit (80-kDa subunit) of chicken calcium-dependent protease (Ca2+-protease; Ohno, S., Emori, Y., Imajoh, S., Kawasaki, H., Kisaragi, M., and Suzuki, K. (1984) Nature 312, 566-570). The two sets of clones were identified as cDNA clones for two Ca2+-protease isozymes with high (mu-type) and low (m-type) calcium sensitivities from a comparison of the primary structures deduced from the nucleotide sequences with partial amino acid sequences from the two isozymes. The cDNA clones for the 80-kDa subunits of the mu- and m-type Ca2+-proteases contained, in total, about 1.5- and 2.2-kilobase cDNA inserts, respectively, which correspond roughly to the C-terminal halves of the coding regions and the entire 3'-noncoding regions. The two isozymes are encoded by two distinct mRNA species present in all the tissues examined, although the amount of mRNA significantly differs among the various tissues. Four E-F hand structures, typical calcium-binding structures in various calcium-binding proteins such as calmodulin, were detected in the C-terminal regions of both isozymes, as in the case of chicken Ca2+-protease. Comparison of the amino acid sequences of the two rabbit isozymes and the corresponding region of the chicken enzyme revealed marked homology, which indicates that these three enzymes have the same evolutionary origin. Furthermore, we suggest that the mu-type rabbit Ca2+-protease, rather than the m-type, is similar to chicken Ca2+-protease, which is regarded as an m-type enzyme in the C-terminal region. The evolution and molecular basis of the differences in calcium sensitivities of the Ca2+-proteases are discussed.     

Cloning and Characterization of the cDNA Encoding a Novel Brain-Specific 14-kDa Protein

A new acidic protein with a molecular weight of 14,000 was purified from rat brain, in which it was specifically expressed, and partially sequenced by protein sequencing. On the basis of results obtained from the amino acid sequences, mixed oligonucleotides were synthesized and used as probes to clone a cDNA from a rat brain cDNA library. The cloned cDNA provided the full-length sequence of the 14-kDa protein. Northern blot hybridization using total RNA from several tissues of the rat provided evidence that the 14-kDa protein was expressed specifically in rat brain. Transfection of this cDNA into mammalian cells resulted in expression of the 14-kDa protein. The amino acid sequence predicted from the cDNA of the rat brain 14-kDa protein contained 137 amino acid residues. A hydropathy profile revealed a hydrophobic domain (amino acids 60-80) flanked by highly hydrophilic stretches on both sides. Whereas the N-terminal region of the 14-kDa protein contained four repeating motifs, EKTKEGV, the C-terminal domain was rich in glutamic acid and proline. A computer search of the amino acid sequence of the 14-kDa protein indicated no homology to any other protein reported so far.     

Crocalbin: a new calcium-binding protein that is also a binding protein for crotoxin, a neurotoxic phospholipase A2

Utilizing Marathon-ready cDNA library and a gene-specific primer corresponding to a partial amino acid sequence determined previously, the complete nucleotide sequence for the cDNA of crocalbin, which binds crotoxin (a phospholipase A2) and Ca2+, was obtained by polymerase chain reaction. The open reading frame of the cDNA encodes a novel polypeptide of 315 amino acid residues, including a signal sequence of 19 residues. This protein contains six potential Ca(2+)-binding domains, one N-glycosylation site, and a large amount of acidic amino acid residues. The ability to bind Ca2+ has been ascertained by calcium overlay experiment. Evidenced by sequence similarity in addition, it is concluded that crocalbin is a new member of the reticulocalbin family of calcium-binding proteins.     

65-kilodalton protein phosphorylated by interleukin 2 stimulation bears two putative actin-binding sites and two calcium-binding sites

We have previously characterized a 65-kilodalton protein (p65) as an interleukin 2 stimulated phosphoprotein in human T cells and showed that three endopeptide sequences of p65 are present in the sequence of l-plastin [Zu et al. (1990) Biochemistry 29, 1055-1062]. In this paper, we present the complete primary structure of p65 based on the cDNA isolated from a human T lymphocyte (KUT-2) cDNA library. Analysis of p65 sequences and the amino acid composition of cleaved p65 N-terminal peptide indicated that the deduced p65 amino acid sequence exactly coincides with that of l-plastin over the C-terminal 580 residues [Lin et al. (1988) Mol. Cell. Biol. 8, 4659-4668] and has a 57-residue extension at the N-terminus to l-plastin. Computer-assisted structural analysis revealed that p65 is a multidomain molecule involving at least three intriguing functional domains: two putative calcium-binding sites along the N-terminal 80 amino acid residues; a putative calmodulin-binding site following the calcium-binding region; and two tandem repeats of putative actin-binding domains in its middle and C-terminal parts, each containing approximately 240 amino acid residues. These results suggest that p65 belongs to actin-binding proteins.     

The deduced protein sequence of the human carboxypeptidase N high molecular weight subunit reveals the presence of leucine-rich tandem repeats

Human plasma carboxypeptidase N is a 280-kDa tetramer with two high molecular mass (83-kDa) glycosylated subunits which protect the two 50-kDa catalytic subunits and keep them in the circulation. An initial clone for the 83-kDa subunit was obtained by screening two lambda gt11 human liver cDNA expression libraries with antiserum specific for carboxypeptidase N or the 83-kDa subunit. The libraries were rescreened with the labeled cloned cDNA, and the largest clone obtained (2536-base pair insert) was completely sequenced. The deduced protein sequence matched the sequence of several tryptic peptides from the 83-kDa subunit but did not contain the NH2-terminal sequence. The remaining portion of the protein coding sequence was synthesized by the polymerase chain reaction, cloned, and sequenced. The composite cDNA sequence is 2870 base pairs long with an open reading frame of 1608 base pair coding for a protein of 536 amino acids (Mr = 58,762). The protein sequence contains seven potential N-linked glycosylation sites and a threonine/serine-rich region which is a potential site for attachment of O-linked carbohydrate. The most striking feature is a region (residues 68-355) containing 12 leucine-rich tandem repeats of 24 residues with the following consensus sequence: P-X-X-alpha-F-X-X-L-X-X-L-X-X-L-X-L-X-X-N-X-L-X-X-L (X = any amino acid and alpha = aliphatic amino acids, I, L, or V). This repeating pattern is found in the leucine-rich alpha 2-glycoprotein and in other proteins where it might mediate interactions with macromolecules. This region also contains five sequences with heptad repeating leucine residues comprising a leucine zipper motif. The leucine-rich domain likely constitutes an important structural or functional element in the interactions of the 83- and 50-kDa subunits to form the active tetramer of carboxypeptidase N.     

Isolation and characterization of a cDNA encoding a chick alpha-actinin

We have isolated and sequenced a 2.1-kilobase cDNA encoding 86% of the sequence of alpha-actinin. The cDNA clone was isolated from a chick embryo fibroblast cDNA library constructed in the expression vector lambda gt11. Identification of this sequence as alpha-actinin was confirmed by immunological methods and by comparing the deduced protein sequence with the sequence of several CNBr fragments obtained from adult chicken smooth muscle (gizzard) alpha-actinin. The deduced protein sequence shows two distinct domains, one of which consists of four repeats of approximately 120 amino acids. This region corresponds to a previously identified 50-kDa tryptic peptide involved in formation of the alpha-actinin dimer. The last 19 residues of C-terminal sequence display an homology with the so-called E-F hand of Ca2+-binding proteins. Hybridization analysis reveals only one size of mRNA (approximately 3.5 kilobases) in fibroblasts, but multiple bands in genomic cDNA.     

Characterization of a novel plant PP2C-like protein Ser/Thr phosphatase as a calmodulin-binding protein

Protein phosphatases regulated by calmodulin (CaM) mediate the action of intracellular Ca2+ and modulate functions of various target proteins by dephosphorylation. In plants, however, the role of Ca2+ in the regulation of protein dephosphorylation is not well understood due to a lack of information on characteristics of CaM-regulated protein phosphatases. Screening of a cDNA library of the moss Physcomitrella patens by using 35S-labeled calmodulin as a ligand resulted in identification of a gene, PCaMPP, that encodes a protein serine/threonine phosphatase with 373 amino acids. PCaMPP had a catalytic domain with sequence similarity to type 2C protein phosphatases (PP2Cs) with six conserved metal-associating amino acid residues and also had an extra C-terminal domain. Recombinant GST fusion proteins of PCaMPP exhibited Mn2+-dependent phosphatase activity, and the activity was inhibited by pyrophosphate and 1 mm Ca2+ but not by okadaic acid, orthovanadate, or beta-glycerophosphate. Furthermore, the PCaMPP activity was increased 1.7-fold by addition of CaM at nanomolar concentrations. CaM binding assays using deletion proteins and a synthetic peptide revealed that the CaM-binding region resides within the basic amphiphilic amino acid region 324-346 in the C-terminal domain. The CaM-binding region had sequence similarity to amino acids in one of three alpha-helices in the C-terminal domain of human PP2Calpha, suggesting a novel role of the C-terminal domains for the phosphatase activity. These results provide the first evidence showing possible regulation of PP2C-related phosphatases by Ca2+/CaM in plants. Genes similar to PCaMPP were found in genomes of various higher plant species, suggesting that PCaMPP-type protein phosphatases are conserved in land plants.     

Molecular cloning of hippocalcin, a novel calcium-binding protein of the recoverin family exclusively expressed in hippocampus.

We have isolated a cDNA clone encoding a novel calcium-binding protein of the recoverin family from rat brain cDNA library. This clone (PCB11) has 588 nucleotides in the open reading frame including the termination codon, 174 nucleotides of the 5' leader and 800 nucleotides of the 3' noncoding region. The complete amino acid sequence deduced from the cDNA is composed of 195 residues, has a calculated molecular mass of 22,574 Daltons, and contains three putative calcium-binding domains of the EF-hand structure. The deduced amino acid sequence has a striking sequence homology to those of the retinal recoverin family (recoverin, visinin, P26, 23kD protein, S-modulin) and the brain-derived recoverin family (P23k, 21-kDa CaBP and neurocalcin). Northern blot, in situ hybridization, immunoblot and immunohistochemical analyses revealed that the protein is exclusively expressed in pyramidal layer of the hippocampus. The protein was therefore designated hippocalcin.     

Molecular cloning of a novel mammalian calcium-dependent protease distinct from both m- and mu-types. Specific expression of the mRNA in skeletal muscle

Two types of calcium-dependent protease with distinct calcium requirements (termed muCANP and mCANP) are known in mammalian tissues. These two isozymes consist of different large (80-kDa) subunits (mu- or m-types) and identical small (30-kDa) subunits. By screening human and rat muscle cDNA libraries with a cDNA probe for the chicken CANP large subunit, which has a structure similar to both the mammalian mu- and m-types, a cDNA clone encoding a novel member of the CANP large subunit family was obtained. The encoded protein (designated "p94") consists of 821 amino acid residues (Mr 94,084) and shows significant sequence homology with both human mu-type (54%) and m-type (51%) large subunits. p94 can be divided into four domains (I-IV) as reported for the CANP large subunit family. Domains II and IV are potential cysteine protease and calcium-binding domains, respectively, and have sequences homologous to the corresponding domains of other CANP large subunits. However, domain I of p94 is significantly different from others. Moreover, p94 contains two unique sequences of 62 and 77 residues in domains II and III, respectively. In contrast to the ubiquitous expression of mu- and m-types, Northern blot analysis revealed that the mRNA for p94 exists only in skeletal muscle with none detected in other tissues including heart muscle and smooth muscles such as intestine.     

Molecular cloning of a histidine-rich Ca2+-binding protein of sarcoplasmic reticulum that contains highly conserved repeated elements

We reported previously the purification of a 165-kDa muscle-specific protein identified by virtue of its ability to bind 125I-labeled low density lipoprotein with high affinity after sodium dodecyl sulfate-polyacrylamide gel electrophoresis (Hoffmann, S. L., Brown, M. S., Lee, E., Pathak, R. K., Anderson, R. G. W., and Goldstein, J. J. (1989) J. Biol. Chem. 264, 8260-8270). The protein is located in the lumen of the sarcoplasmic reticulum, where it has no access to plasma lipoproteins. It binds to 45Ca2+ on nitrocellulose blots and stains metachromatically blue with Stains-all, a cationic dye that stains Ca2+-binding proteins. In the current paper, we have isolated a full-length rabbit cDNA clone for the 165-kDa protein. The deduced amino acid sequence reveals a 852-amino acid protein with the following structural features: 1) an NH2-terminal 27-residue putative signal sequence; 2) a highly repetitive region containing nine nearly identical tandem repeats of 29 residues, each consisting of a histidine-rich sequence HRHRGH, a stretch of 10-11 acidic amino acids, and a sequence containing 2 serines and a threonine in a negatively charged context; 3) a 13-residue stretch of polyglutamic acid; and 4) a COOH-terminal cluster of 14 closely spaced cysteine residues with the repeating pattern of Cys-X-X-Cys suggestive of a heavy metal binding domain. Histidine, aspartic acid, and glutamic acid accounted, respectively, for 13, 12, and 19% of the amino acids. The protein does not share any significant sequence homology with the cell surface low density lipoprotein receptor. Stretches of acidic amino acids are a feature of two other luminal sarcoplasmic reticulum proteins, suggesting that these may be a general feature of luminal sarcoplasmic reticulum proteins. We suggest that the histidine-rich Ca2+-binding protein described in the current study be designated HCP. The role of HCP in Ca2+ homeostasis in the sarcoplasmic reticulum of skeletal and cardiac muscle remains to be determined.     

Digestion of mu- and m-calpain by trypsin and chymotrypsin

Proteolytic digestion by trypsin and chymotrypsin was used to probe conformation and domain structure of the mu- and m-calpain molecules in the presence and the absence of Ca(2+). Both calpains have a compact structure in the absence of Ca(2+); incubation with either protease for 120 min results in only three or four major fragments. A 24-kDa fragment was produced by removal of the Gly-rich area in domain V of the 28-kDa subunit. The other fragments were from the 80-kDa subunit. Except for trypsin digestion of m-calpain, the region between amino acids 245 and 265 (human sequence) was very susceptible to cleavage by both proteases in the absence of Ca(2+); this region is in domain II (IIb of the crystallographic structure). Although no proteolytically active fragments could be isolated from either tryptic or chymotryptic digests, the calpain molecule can remain assembled in a proteolytically active complex even after the 80-kDa subunit has been completely degraded. The results suggest that interaction among different regions of the entire calpain molecule is required for its full proteolytic activity. In the presence of 1 mM Ca(2+), both calpains are degraded to fragments less than 40-kDa in less than 5 min. The C-terminal ends of both subunits, from amino acids 503 to 506 to the end of the 80-kDa subunit and from amino acids 85 to 88 to the end of the 28-kDa subunit, were resistant to degradation by either protease in the presence or in the absence of Ca(2+). Hence, this part of the calpain molecule is in a compact structure that does not change significantly in the presence of Ca(2+).     

cDNA-derived amino acid sequence of the 86-kDa subunit of the Ku antigen

The Ku antigen is a DNA-associated nuclear protein recognized by sera from patients with autoimmune diseases. It consists of two polypeptides of 86 and 70 kDa. cDNA clones encoding the 86-kDa subunit of the Ku antigen were isolated by probing lambda gt11 recombinant cDNA expression libraries with a monoclonal antibody specific for this protein. The amino acid sequence deduced from the cDNA comprises 732 amino acids and corresponds to a protein with molecular weight of 81.914. Nineteen residues at the NH2 terminus determined by protein sequencing corresponded to the sequence deduced from the cDNA. The predicted amino acid sequence contains a region with repeating leucine residues similar to the "leucine zipper" structure observed in the c-myc, v-myc, and c-fos oncogene products. The largest cDNA hybridized to 2.7- and 3.4-kilobase poly(A)+ mRNAs from HeLa cells. The cDNA clones expressed fusion proteins immunoreactive with the monoclonal antibody and sera from patients with autoimmune diseases.     

The cDNA sequence of the 69-kDa subunit of the carrot vacuolar H+-ATPase. Homology to the beta-chain of F0F1-ATPases

Vacuolar ATPases constitute a novel class of N-ethylmaleimide- and nitrate-sensitive proton pumps associated with the endomembrane system of eukaryotic cells. They resemble F0F1-ATPases in that they are large multimeric proteins, 400-500 kDa, composed of three to nine different subunits. Previous studies have indicated that the active site is located on the approximately 70-kDa subunit. Using antibodies to the approximately 70-kDa subunit of corn to screen a carrot root lambda gt11 cDNA library, we have isolated cDNA clones of the carrot 69-kDa subunit. The complete primary structure of the 69-kDa subunit was then determined from the nucleotide sequence of its cDNA. The 69-kDa subunit consists of 623 amino acids (Mr 68,835), with no obvious membrane-spanning regions. The carrot cDNA sequence was over 70% homologous with exons of a Neurospora 69-kDa genomic clone. The protein sequence of the carrot 69-kDa subunit also exhibited 34.3% identity to four representative F0F1-ATPase beta-chains over a 275-amino-acid core stretch of similar sequence. Alignment studies revealed several regions which were highly homologous to beta-chains, including sequences previously implicated in catalytic function. This provides definitive evidence that the vacuolar ATPase is closely related to the F0F1-type ATPases. A major functional difference between the 69-kDa and beta-subunits is the location of 3 critical cysteine residues: two in the putative catalytic region (Cys-248 and Cys-256) and one in the proposed Mg2+-binding site (Cys-279). These cysteines (and two others) probably account for the sensitivity of the vacuolar H+-ATPase to the sulfhydryl reagent, N-ethylmaleimide. It is proposed that the two ATPases may have arisen from a common ancestor by the insertion or deletion of a large stretch of nonhomologous sequence near the amino-terminal end of the subunit.     

Characterization of cDNA clones encoding a novel calcium-activated neutral proteinase from Schistosoma mansoni

To identify and characterize Schistosoma mansoni proteins that are recognized by infected hosts, we have used a pool of sera from infected humans to screen cDNA libraries constructed from poly(A)+ mRNA of adult S. mansoni. The deduced amino acid sequences of the three isolated clones showed a high degree of similarity to the large subunit of calcium-activated neutral proteinase (CANP) from humans and chicken. These overlapping clones, which include a nearly full-length clone with an open reading frame of 758 amino acid residues, together encode the entire large subunit of CANP. The deduced sequence of this S. mansoni protein can be divided into four domains (I-IV) that include the two domains characteristic of other large subunits of CANP: a thiol-protease domain (II) and a calcium-binding domain (IV) containing EF hand motifs. However, the schistosome protein is unique in having only three EF hand motifs in the calcium-binding domain and in having an additional EF hand motif that is shared between domains II and III. We have shown that these EF hand motifs are capable of binding 45Ca2+. Furthermore, the large subunit is S. mansoni contains an NH2-terminal sequence of 28 residues that is absent from the mammalian CANPs and has a high degree of similarity to the presumed receptor binding sequence of colicin Ia and Ib.     

Gene structure of calcium-dependent protease retains the ancestral organization of the calcium-binding protein gene

The gene structure of calcium-dependent protease (Ca2+-protease) was determined. It comprises at least 21 exons, and these were assigned to the 4 functional domains of the protease. The protease domain does not show clear correlation between exons and functional units, but the calmodulin-like calcium-binding domain shows strong correlation. Each of the 4 consecutive calcium-binding regions in the C-terminal part of Ca2+-protease is encoded by one exon. This gene structure supports the idea that the 4 calcium-binding regions of calcium-binding proteins such as calmodulin arose by 2 steps of gene duplication.     

Ca(2+)-binding properties of the platelet glycoprotein IIb ligand-interacting domain.

Glycoprotein (GP) IIb is the alpha subunit of platelet integrin GPIIb-IIIa. Analysis of the primary structure of this subunit has indicated the presence of four stretches of amino acid residues that are highly conserved among various integrin alpha subunits and that have been suggested to be putative calcium-binding sites. To verify the Ca(2+)-binding capacity of these conserved domains and their implication in integrin adhesive functions, a fragment corresponding to the amino acid sequence of GPIIb from positions 171 to 464 was expressed. The nucleotide sequence coding for this GPIIb domain was generated by polymerase chain reaction, cloned into the pTG1924 expression vector, and expressed in Escherichia coli strain TGE901. The recombinant protein was purified by gel exclusion chromatography and used in equilibrium dialysis experiments. The results demonstrate that the four binding sites can be occupied by Ca2+. Two classes of binding sites can be detected, including two sites with a Kd of 30 microns and two sites of lower affinity with a Kd of 120 microns. Interaction of Ca2+ with these two classes of sites was inhibited by a large excess of Mg2+ or Mn2+, suggesting that these cations are competitive for the same sites on GPIIb. Thus, the four Ca(2+)-binding sites of GPIIb are not similar and exhibit different affinities for divalent ions. To verify the functional implication of these Ca(2+)-binding sites, the effect of Ca2+ on the binding of fibrinogen to the recombinant protein was analyzed using a solid-phase assay. The results indicate that optimal fibrinogen binding occurs when the four calcium-binding sites are occupied and establish the functional importance of this Ca(2+)-binding domain in the ligand-binding activity of GPIIb.     

Elucidation of a protease-sensitive site involved in the binding of calcium to C-reactive protein

C-reactive protein (CRP) is a Ca2+-binding protein composed of five identical 23-kDa subunits arranged as a cyclic pentamer, present in greatly elevated concentration in the blood during the acute phase of processes involving tissue injury and necrosis. In the present work, it was found that treatment of human CRP with Pronase or Nagarse protease produces two major fragments which remain associated in physiological buffers but are separable under denaturing conditions. To localize the cleavage site(s), the fragments were characterized according to molecular mass, amino acid composition, partial amino acid sequence, and reactivity with monoclonal antibodies specific for the fragments and for defined CRP epitopes including residues 147-152 and 199-206. Nagarse protease cleaves the CRP subunit between residues 145 and 146, producing two fragments, 16 and 6.5 kDa (calculated molecular mass). Pronase cleaves the CRP subunit between residues 146 and 147, producing a 16-kDa fragment (A1) and a 6.5-kDa fragment (B); an additional fragment (A2) approximately 1 kDa smaller than fragment A1 is also apparently produced due to a secondary cleavage site in fragment A1. Cleavage appears to be completely inhibited in the presence of 1 mM CaCl2. Ca2+ does not protect cleaved CRP from heat-induced aggregation (i.e., precipitation) as it does the intact protein. Protease-cleaved CRP loses the ability to bind to the Ca2+-dependent ligand phosphorylcholine but remains the ability to bind to the Ca2+-independent ligand arginine-rich histone. Equilibrium dialysis indicates that intact CRP binds 2 mol of Ca2+/mol of subunit with a Kd of 6 X 10(-5) M.(ABSTRACT TRUNCATED AT 250 WORDS)     

cDNA sequence of the small subunit of the hamster ribonucleotide reductase.

Ribonucleotide reductase activity is markedly elevated in cell lines selected for resistance to hydroxyurea, a cytotoxic drug known specifically to inhibit ribonucleotide reductase. From a cDNA library constructed from a highly hydroxyurea-resistant hamster lung cell line, 600H in which the activity is elevated more than 80-fold, we have isolated a full length cDNA for the small subunit of the reductase. The cDNA is 3.48 kb long with an open reading frame of 1158 nucleotides and a long 3' flanking region of 2169 nucleotides from the termination codon. The derived polypeptide sequence is closely similar to the small subunit of the mouse, differing from it in 20 amino acid positions. Most of these replacements occur in the N-terminal segment of the protein. The hamster subunit does not contain 4 amino acid residues found in the mouse small subunit near the C-terminal end. RNA blots probed with the cDNA show two poly(A)+ RNA species which are elevated in hydroxyurea-resistant cells.     

A novel plant ferritin subunit from soybean that is related to a mechanism in iron release

Ferritin is a multimeric iron storage protein composed of 24 subunits. Ferritin purified from dried soybean seed resolves into two peptides of 26.5 and 28 kDa. To date, the 26.5-kDa subunit has been supposed to be generated from the 28-kDa subunit by cleavage of the N-terminal region. We performed amino acid sequence analysis of the 28-kDa subunit and found that it had a different sequence from the 26.5-kDa subunit, thus rendering it novel among known soybean ferritins. We cloned a cDNA encoding this novel subunit from 10-day-old seedlings, each of which contained developed bifoliates, an epicotyl and a terminal bud. The 26.5-kDa subunit was found to be identical to that identified previously lacking the C-terminal 16 residues that correspond to the E helix of mammalian ferritin. However, the corresponding region in the 28-kDa soybean ferritin subunit identified in this study was not susceptible to cleavage. We present evidence that the two different ferritin subunits in soybean dry seeds show differential sensitivity to protease digestions and that the novel, uncleaved 28-kDa ferritin subunit appears to stabilize the ferritin shell by co-existing with the cleaved 26.5-kDa subunit. These data demonstrate that soybean ferritin is composed of at least two different subunits, which have cooperative functional roles in soybean seeds.