Monoclonal antibodies to human fibroblast procollagenase. Inhibition of enzymatic activity, affinity purification of the enzyme, and evidence for clustering of epitopes in the NH2-terminal end of the activated enzyme

This study describes 11 monoclonal antibodies (Mabs) against human fibroblast collagenase that (i) inhibit the specific catalytic activity of the enzyme and/or (ii) react with one or more forms of the enzyme on Western blots. Each of the Mabs specifically immunoprecipitated the Mr 57,000/52,000 procollagenase from [35S]methionine-labeled culture medium. Five Mabs, designated VI-3, VI-4, 2C5, 4A2, and 7C2, inhibited the activity of fibroblast-type collagenase against soluble monomeric collagen and against reconstituted collagen fibrils but did not inhibit the genetically distinct human PMN leukocyte collagenase. The interstitial collagenase produced by human mucosal keratinocytes (SCC-25) was also inhibited, whereas the corresponding enzyme from rat was not. Assignment of epitopes to structural domains within the molecule based on immunoperoxidase staining of Western blots of collagenase and its autocatalytic fragments revealed that 9 of 11 epitopes, including those recognized by 4 inhibitory Mabs, were clustered in a 169-residue domain, which constitutes the NH2-terminal part of the Mr 46,000/42,000 active enzyme. One Mab (X-2a) specifically recognized the Mr 57,000/52,000 zymogen species and failed to react with the active Mr 46,000/42,000 form. The inhibitory Mab VI-3 was used for immunoaffinity purification of procollagenase from culture media with a recovery better than 80% and a yield of approximately 1.4 mg of enzyme/L of medium.     

Monoclonal antibodies to an artificial immunogen, specifically demonstrating phosphotyrosine-containing proteins

The importance of protein phosphorylation at tyrosyl hydroxy groups in the control of cell proliferation has recently been established. For identification of tyrosine-phosphorylated proteins, monoclonal antibodies (Mabs) against artificial immunogens containing O-phosphotyrosine (pTyr) or tripeptide pTyr-Gly-Gly as haptens were generated; the haptens were coupled to carrier proteins (bovine serum albumin, human immunoglobulin, keyhole limpet hemocyanin). After immunization of mice with pTyr coupled to keyhole limpet hemocyanin, Mabs were generated which were highly specific for pTyr and did not cross-react with O-phosphoserine, O-phosphothreonine, tyrosine or nucleoside-5'-monophosphates. The Mabs specifically react with tyrosinephosphorylated proteins in the Rous sarcoma virus-transformed rat XC-cell.     

Light form of Morris hepatoma gamma-glutamyltransferase

Using solubilization with bromelain, the light form of gamma-glutamyltransferase was purified from Morris hepatoma 5123D. Some properties of this enzyme were compared to those of the light form rat kidney GGT. Anthglutin and its isomer inhibit competitively the former enzyme but non-competitively the latter. On zymograms of rat control sera, five GGT fractions were noted, but in sera of rats with hepatoma 5123D also the light form and the increase of GGT activity ain region of fraction II were observed. Only these two enzyme fractions react with antibody anti heavy form of Morris hepatoma GGT.     

Purification,kinetic studies,and homology model of Escherichia coli fructose-1,6-bisphosphatase

Previous kinetic characterization of Escherichia coli fructose 1,6-bisphosphatase (FBPase) was performed on enzyme with an estimated purity of only 50%. Contradictory kinetic properties of the partially purified E. coli FBPase have been reported in regard to AMP cooperativity and inactivation by fructose-2,6-bisphosphate. In this investigation, a new purification for E. coli FBPase has been devised yielding enzyme with purity levels as high as 98%. This highly purified E. coli FBPase was characterized and the data compared to that for the pig kidney enzyme. Also, a homology model was created based upon the known three-dimensional structure of the pig kidney enzyme. The k_cat of the E. coli FBPase was 14.6 s⁻¹ as compared to 21 s⁻¹ for the pig kidney enzyme, while the K_m of the E. coli enzyme was approximately 10-fold higher than that of the pig kidney enzyme. The concentration of Mg²⁺ required to bring E. coli FBPase to half maximal activity was estimated to be 0.62 mM Mg²⁺, which is twice that required for the pig kidney enzyme. Unlike the pig kidney enzyme, the Mg²⁺ activation of the E. coli FBPase is not cooperative. AMP inhibition of mammalian FBPases is cooperative with a Hill coefficient of 2; however, the E. coli FBPase displays no cooperativity. Although cooperativity is not observed, the E. coli and pig kidney enzymes show similar AMP affinity. The quaternary structure of the E. coli enzyme is tetrameric, although higher molecular mass aggregates were also observed. The homology model of the E. coli enzyme indicated slight variations in the ligand-binding pockets compared to the pig kidney enzyme. The homology model of the E. coli enzyme also identified significant changes in the interfaces between the subunits, indicating possible changes in the path of communication of the allosteric signal.     

Molecular cloning of cDNA for rat liver general acyl CoA dehydrogenase and homology between the rat liver and pig kidney enzymes

cDNA clone for general acyl CoA dehydrogenase (GAD) was isolated from a rat liver cDNA expression library in lambda gt11 using anti-pig kidney GAD antibody. Size of the isolated cDNA was estimated to be 1.5-1.6 kb. By immunological analysis of fusion protein and epitope selection, the cDNA clone was identified as that containing the GAD gene. Partial amino acid sequence deduced from nucleotide sequence of the cDNA coincided with that of the pig kidney enzyme. The antibody cross-reacted with rat liver enzyme and molecular weights of these enzyme proteins were shown to be almost the same. All these results indicate that rat liver GAD shares a common structure with pig kidney enzyme.     

Light-optical and ultrastructural immunohistochemistry of antigen H-4, a common marker of epithelial cells]

Epitope localization reacting with mice monoclonal antibodies (Mabs) H 4 was investigated using the specimen of epithelium of skin, human uterine cervix as well as in the culture of epithelium cells of guinea-pig duct deferent. Mice monoclonal antibodies against antigen H 4 obtained by the hybridoma method after immunization of mice with rat colon epithelium cytoskeletal fractions were used. Mabs H 4 were shown to react with antigen of intermediate filaments of all studied normal epithelial, cancer cells as well as culture epithelial cells. Mabs H 4 are supposed to be used as a unique immunohistochemical marker of epithelium cells under normal and malignant growth conditions.     

A monoclonal antibody against human kidney gamma-glutamyl transpeptidase: preparation, immunochemical, and immunohistochemical characterization

To perform immunohistochemical study of the distribution of gamma-glutamyl transpeptidase in human organs, a highly specific antibody against the human enzyme is required. We prepared monoclonal antibody against gamma-glutamyl transpeptidase from human kidney, using the hybridoma technique. The antibody was of the IgG1 type and the light chain belonged to the kappa class. The antibody reacted specifically with the 63 KD heavy subunit of the enzyme. Examination of the specificity of the antibody performed by immunohistochemical staining of human kidney sections revealed that the antigen was localized on the brush border and along the basolateral membrane of the epithelial cells of both the convoluted and the straight portions of the proximal tubule. This antibody was also reactive in several human organs other than kidney, including epididymis, prostate, seminal vesicle, pancreas, and normal liver, and in human hepatoma. These findings indicate the existence of an antigenic determinant common to human kidney and other organs. The monoclonal antibody did not crossreact with mouse, rat, guinea pig, rabbit, or pig kidney.     

The effect of 3-mercaptopicolinic acid on phosphoenolpyruvate carboxykinase (GTP) in the rat and guinea pig.

In vitro, 3-mercaptopicolinic acid inhibited phosphoenolpyruvate carboxykinase activity in supernatant fractions of liver, kidney cortex, and adipose tissue obtained from fasted rats. 3-Mercaptopicolinic acid also inhibited enzymatic activity in the mitochondrial and supernatant fractions of liver obtained from fasted guinea pigs. In the fasted rat, the oral administration of 3-mercaptopicolinic acid increased liver carboxykinase activity even though the blood glucose concentrations decreased. Kidney cortex carboxykinase decreased while adipose tissue enzyme was unchanged. In the fasted guinea pig, the oral administration of 3-mercaptopicolinic acid lowered blood glucose concentrations but had no effect on liver mitochondrial or supernatant carboxykinase activity. The elevation in rat liver enzymatic activity appears to be due to protein synthesis, since the concurrent administration of cycloheximide prevents the increase in enzyme activity. 3-Mercaptopicolinic acid appears to be noncompetitive with respect to Mn²⁺.     

The effect of pyrazines on the metabolism of tryptophan and nicotinamide adenine dinucleotide in the rat Evidence of the formation of a potent inhibitor of aminocarboxy-muconate-semialdehyde decarboxylase from pyrazinamide

The intraperitoneal or oral administration of pyrazinamide and pyrazinoic acid (pyrazine 2-carboxylic acid) resulted in a marked increase of the NAD content in rat liver. The injections of pyrazine and pyrazine 2,3-dicarboxylic acid exhibited no significant effect on the hepatic NAD content. The boiled extract obtained from liver and kidney of rat injected with either pyrazinamide or pyrazinoic acid exhibited a potent inhibitory effect on the aminocarboxymuconate-semialdehyde decarboxylase (EC 4.1.1.45) activity in either liver or kidney, although pyrazinamide or pyrazinoic acid per se did not inhibit the enzyme activity. The unknown inhibitor of aminocarboxymuconate-semialdehyde decarboxylase was dialysable and heat-stable, and mostly excreted in urine by 6 and 12 h after injection of pyrazinoic acid and pyrazinamide, respectively. Pyrazine 2,3-dicarboxylic acid, pyrazine, nicotinamide, nicotinic acid, tryptophan, anthranilic acid, 5-hydroxyanthranilic acid and quinolinic acid exhibited no significant effect on the aminocarboxymuconate-semialdehyde decarboxylase activity in liver and kidney at the concentration of 1 mM in the reaction mixture. The expired ¹⁴CO₂ from l-[benzen ring-U-¹⁴C]tryptophan was markedly decreased by the pyrazinamide injection, while the urinary excretion of ¹⁴C-labeled metabolites from l-tryptophan, mainly quinolinic acid, was markedly increased. These results suggest that the glutarate pathway of l-tryptophan was strongly inhibited by the inhibitor produced after the administration of pyrazinoic acid and pyrazinamide. Pyrazinamide but not pyrazinoic acid also exhibited a significant inhibition of the nuclear enzyme poly(ADP-ribose) synthetase in rat liver.     

Catalytic and structural characteristics of carp hepatopancreas D-amino acid oxidase expressed in Escherichia coli

D-amino acid oxidase of carp (Cyprinus carpio) hepatopancreas was overexpressed in Escherichia coli cells and purified to homogeneity for the first time in animal tissues other than pig kidney. The purified preparation had a specific activity of 293 units mg(-1) protein toward D-alanine as a substrate. It showed the highest activity toward D-alanine with a low Km of 0.23 mM and a high kcat of 190 s(-1) compared to 10 s(-1) of the pig kidney enzyme. Nonpolar and polar uncharged D-amino acids were preferable substrates to negatively or positively charged amino acids. The enzyme exhibited better thermal and pH stabilities than several yeast counterparts or the pig kidney enzyme. Secondary structure topology consisted of 11 alpha-helices and 17 beta-strands that differed slightly from pig kidney and Rhodotorula gracilis enzymes. A three-dimensional model of the carp enzyme constructed from a deduced amino acid sequence resembled that of pig kidney D-amino acid oxidase but with a shorter active site loop and a longer C-terminal loop. Judging from these characteristics, carp D-amino acid oxidase is close to the pig kidney enzyme structurally, but analogous to the R. gracilis enzyme enzymatically in turnover rate and pH and temperature stabilities.     

Differential expression of cytochrome P-450 1 and related forms in rabbit liver and kidney

Monoclonal antibodies developed to cytochrome P-450 1, some of which react with proteins in addition to P-450 1, were used to investigate the differential expression of P-450 1 dependent 21-hydroxylase activity in renal tissue of rabbits exhibiting differences in hepatic 21-hydroxylase activity. Using immunohistochemical techniques, the monoclonal antibodies, 2F5 and 3C3, localized protein in the S2 and S3 segments of the proximal tubule in the renal cortex. These two monoclonal antibodies, 2F5 and 3C3, reacted with a kidney protein that migrated on sodium dodecyl sulfate-polyacrylamide gel electrophoresis with a relative electrophoretic mobility that did not correspond to known rabbit hepatic isozymes and was termed P-450 K. Antibodies specific for P-450 1 and 3b, 1F11 and 8-27, respectively, produced no staining in kidney. The protein recognized by the 2F5 and 3C3 antibodies is immunologically distinct from cytochrome P-450s 1, 2, and 3b. The rate of 21-hydroxylation of progesterone was shown to be approximately 100-fold less in kidney than liver microsomes where this pathway is largely catalyzed by P-450 1. The activity of the kidney microsomes was not inhibited by antibodies directed to P-450 1. In addition, the variation observed for the 21-hydroxylase activity in the hepatic microsomal fraction of outbred New Zealand white rabbits was not evident in kidney microsomes from these same animals. The 2F5 antibody was found, however, to be inhibitory (about 50%) of the 11-hydroxylation of lauric acid in kidney microsomes. This suggests that P-450 K participates in lauric acid 11-hydroxylase activity. The treatment of rabbits with phenobarbital, but not 2,3,7,8-tetrachlorodibenzo-p-dioxin, was found to induce the levels of P-450 K.     

Bile salt sulfotransferase in guinea pig liver

Bile salt sulfotransferase from guinea pig liver is purified by the procedures of ammonium sulfate fractionation, Sephadex G-100 column chromatography, agarose-hexane-adenosine 3′,5′-diphosphate affinity chromatography and polyacrylamide gel electrophoresis. The purified enzyme exhibits a pH optimum of 6.8, an isoelectric point of 5.6 and a molecular weight of 7600 estimated by gel filtration technique. The apparent K_m values of the enzyme are 7.7·10^?5 M for taurolithocholate and 1.4·10^?6 M for 3′-phosphoadenosine 5′-phosphosulfate. It requires Mg²⁺ and free sulfohydryl group(s) for activity. The enzyme reacts with hydroxy groups of bile salts at both 3α and 3β positions. No activity is found in the kidney of guinea pig. The purified enzyme does not react with estrone, estradiol, testosterone, dehydroepiandrosterone, cholesterol, phenol, tyramine, and serotonin. The results indicate that bile salt sulfotransferase is distinct from other hepatic sulfotransferases.     

Human gamma-glutamyl transpeptidase cDNA: comparison of hepatoma and kidney mRNA in the human and rat

gamma-Glutamyl transpeptidase (GGT) is a glutathione-metabolizing enzyme that has been extensively studied in relation to hepatocarcinogenesis. Using a cDNA for rat kidney GGT as a probe, we have isolated a full-length cDNA for human GGT from a hepatoma cell-line library. Nucleotide sequence analysis of the clone revealed a 2326-bp insert that includes a 5'-untranslated region of 487 nucleotides (nt), an open reading frame (ORF) of 1707 nt, and a 3'-untranslated region of 132 nt. The ORF encodes a protein with an amino acid sequence that is highly similar to that of the rat GGT precursor peptide, with an overall identity of 79%. The cDNA clone was used to probe Northern blots of hepatoma and kidney RNA from both human and rat. In both species, the GGT mRNA is longer in hepatoma than in kidney. In addition, the human mRNAs were longer than their counterparts in the rat. None of three human hepatocellular carcinomas examined showed a marked elevation in GGT mRNA levels relative to surrounding liver tissue.     

Monoclonal antibodies to Rous sarcoma virus pp60src react with enzymatically active cellular pp60src of avian and mammalian origin

The derivation and characterization of 22 hybridoma clones producing monoclonal antibodies (Mabs) specific for the transforming protein of Rous sarcoma virus, pp60src, are described. All Mabs reacted with pp60v-src encoded by Prague, Schmidt-Ruppin, and Bratislava 77 strains of Rous sarcoma virus. Of these Mabs, 10 efficiently immunoprecipitated pp60c-src from chicken embryo cells. Of these 10 Mabs, 2 (GD11 and EB8) readily detected pp60c-src from a variety of rodent and human cultured cells and from rat brain tissue in an in vitro immune complex kinase assay. Mapping experiments have tentatively localized the determinant(s) recognized by GD11 and EB8 to a region of the src protein bounded by amino acid residues 82 to 169, whereas the remaining Mabs appeared to recognize determinants residing within residues 1 to 82 or 169 to 173. Most of the Mabs complexed denatured pp60v-src in a Western immunoblot, and several were used to localize pp60v-src in Rous sarcoma virus-transformed chicken embryo cells by indirect immunofluorescence microscopy.     

Expression of multiple proteins structurally related to gamma-glutamyl transpeptidase in non-neoplastic adult rat hepatocytes in vivo and in culture.

Freshly isolated hepatocytes from normal adult rat liver do not express measurable gamma-glutamyl transpeptidase (GGT) mRNA in contrast to the significant GGT mRNA levels expressed by normal adult rat kidney and hyperplastic bile ductular tissue from bile duct-ligated rats. However, the induction of GGT activity in rat hepatocytes by two-thirds hepatectomy was accompanied by the appearance of a high level of GGT mRNA. We are now able to demonstrate that normal adult rat hepatocytes express 5 protein bands which cross-react with 2 different anti-rat kidney GGT antisera. The apparent molecular weights were 26.9, 58.0, 63.9, 73.5, and 83.4 kDa, respectively. Expression of the 26.9- and 58.0-kDa proteins strikingly parallels the pattern of induction of GGT enzymatic activity. This suggests that these 2 proteins correspond to the active dimeric enzyme previously described in kidney and neoplastic hepatocellular tissue. In normal hepatocytes, the 73.5-kDa protein represents 50% of the total GGT-immunoreactive protein, in contrast to kidney, where this band contains less than 4% of the GGT protein. The kinetics of expression of the 73.5-kDa protein upon induction of GGT activity in hepatocytes, as well as in culture turnover studies, suggests that this protein is a precursor form of the active enzyme, such as the described 78/79-kDa single-chain glycoprotein propeptide of GGT. It appears that in normal hepatocytes, this precursor is not processed to the same extent as in kidney or in hyperplastic bile ductular tissue.     

Isolation, nucleotide sequence, and expression of a cDNA encoding pig citrate synthase

Citrate synthase is a key enzyme of the Krebs tricarboxylic acid cycle and catalyzes the stereospecific synthesis of citrate from acetyl coenzyme A and oxalacetate. The amino acid sequence and three-dimensional structure of pig citrate synthase dimers are known, and regions of the enzyme involved in substrate binding and catalysis have been identified. A cloned complementary DNA sequence encoding pig citrate synthase has been isolated from a pig kidney lambda gt11 cDNA library after screening with a synthetic oligonucleotide probe. The complete nucleotide sequence of the 1.5-kilobase cDNA was determined. The coding region consists of 1395 base pairs and confirms the amino acid sequence of purified pig citrate synthase. The derived amino acid sequence of pig citrate synthase predicts the presence of a 27 amino acid N-terminal leader peptide whose sequence is consistent with the sequences of other mitochondrial signal peptides. A conserved amino acid sequence in the mitochondrial leader peptides of pig citrate synthase and yeast mitochondrial citrate synthase was identified. To express the pig citrate synthase cDNA in Escherichia coli, we employed the inducible T7 RNA polymerase/promoter double plasmid expression vectors pGP1-2 and pT7-7 [Tabor, S., & Richardson, C. C. (1985) Proc. Natl. Acad. Sci. U.S.A. 82, 1074-1078]. The pig citrate synthase cDNA was modified to delete the N-terminal leader sequence; then by use of a synthetic oligonucleotide linker, the modified cDNA was cloned into pT7-7 immediately following the initiator Met. A glutamate-requiring (citrate synthase deficient), recA- E. coli mutant, DEK15, was transformed with pGP1-2 and then pT7-7PCS. pT7-7PCS complemented the E. coli gltA mutation.(ABSTRACT TRUNCATED AT 250 WORDS)     

Purification, structure, and characterization of caltrin proteins from seminal vesicle of the rat and mouse.

Caltrins, small basic proteins that inhibit calcium uptake by epididymal spermatozoa, have been purified from seminal vesicle content of the mouse and rat. Mouse caltrin (M(r) 8,476) contains 75 amino acid residues, 14 basic, 5 acidic, and 7 cysteines while rat caltrin (M(r) 6,217) has 56 residues, 10 basic, 5 acidic, and 6 cysteines; their pI values are 10.2 and 9.3, respectively. The proteins did not react with Ellman's reagent unless the cystine residues were previously reduced. The primary structures were determined by sequencing fragments generated by trypsin, clostripain, and endoproteinase Lys-C digestion. The sequences were ordered to give the total structural formula. The two molecules have no sequence similarity and are different from those of the bull and guinea pig previously reported. Only rat caltrin has a sequence of 13 residues nearly identical to that in guinea pig caltrin I. Both rat and mouse caltrin react with antibodies against bovine and guinea pig caltrins. Reduction and alkylation of cysteine residues suppressed the immunologic response of mouse caltrin; however, modified rat caltrin retained partially its immunoreactivity with the antiserum against guinea pig caltrin I. The same treatment abolished the calcium transport inhibitory activity of mouse caltrin and greatly reduced that of rat caltrin. It is likely that rat and mouse caltrins have the same physiological function as proposed for bovine caltrin; namely, to regulate the development of the Ca(2+)-dependent processes that "capacitate" sperm for fertilization.     

Increased sensitivity of the enzymatic isotopic assay of histamine: measurement of histamine in plasma and serum.

The use of rat kidney instead of guinea pig brain as the source of histamine-N-methyltransferase for the enzymatic assay of histamine was found to improve the sensitivity of the assay. A partially purified preparation (ammonium sulfate fractionation) of the kidney enzyme was 20- to 50-fold more active than the guinea pig preparation, and sufficient enzyme for 14,000 assays could be prepared from six rats. The kidney enzyme, unlike the guinea pig brain enzyme, was free of interfering enzyme activities and gave low values for assay blanks. The two enzymes otherwise had similar properties. The low blank values permitted direct measurement of histamine in normal plasma without the need to isolate and concentrate histamine from the sample. Plasma histamine levels in normal individuals ranged from 0.2-1.4 (mean 0.6, n = 19) ng/ml.     

3 beta-hydroxysteroid isomerase dehydrogenase in guinea-pig kidney: possible involvement in 11-deoxycorticosterone formation in situ

3 beta-Hydroxysteroid isomerase dehydrogenase, capable of acting on C21- and C19-3 beta-hydroxy-5-ene-steroids has been found in guinea-pig kidney at equivalent levels to those in guinea pig testes. Of the 3 beta-hydroxy-5-ene-steroids present in guinea pig serum, 21-hydroxypregnenolone occurs in highest concentration (17 nM) followed by pregnenolone (10 nM), whereas 17 alpha-hydroxy-pregnenolone and dehydroepiandrosterone occur in very low concentrations (less than 0.5 nM). Furthermore, the concentration of 21-hydroxypregnenolone relative to 11-deoxycorticosterone (the mineralocorticoid of the guinea pig), is 10:1 (Nishikawa and Strott, Steroids 41 (1983) 105-120). The apparent Km value for 21-hydroxypregnenolone, for the reaction yielding 11-deoxycorticosterone as catalysed by guinea pig kidney microsomes, was 85 nM and the Vmax 33 pmol/min per mg protein. Pregnenolone was a competitive inhibitor (apparent Ki = 5 microM) in the above reaction. A sex difference in the level of the enzyme in the kidney was found (activity in the female was one-third of that in the male) which may indicate that the enzyme is under partial androgen control. 3 beta-Hydroxysteroid isomerase dehydrogenase activity was also detected in guinea pig liver and again it was lower in the female. Whilst the exact role of 3 beta-hydroxysteroid isomerase dehydrogenase in guinea-pig kidney remains uncertain, the data suggest that it may utilise blood-borne 21-hydroxypregnenolone, the later then playing the role of a prohormone.     

Identification and characterization of rat intestinal keratins. Molecular cloning of cDNAs encoding cytokeratins 8, 19, and a new 49-kDa type I cytokeratin (cytokeratin 21) expressed by differentiated intestinal epithelial cells.

In the previous paper (Quaroni, A., Calnek, D., Quaroni, E., and Chandler, J.S. (1991) J. Biol. Chem. 266, 11923-11931) we describe the use of a panel of "antikeratin" monoclonal antibodies to study cytokeratin distribution in rat intestinal epithelium. In the present paper we describe the use of three antikeratin monoclonal antibodies to identify and recovery cDNA clones expressing immunologically specific fusion proteins from a rat intestinal cDNA library. DNA sequence analysis identified each cDNA encoded epitope including the carboxyl-terminal portions of cytokeratins 8 and 19 (as cataloged by Moll, R., Franke, W.W., and Schiller, D.L. (1982) Cell 31, 11-24) recognized by antibodies RK4 and RK7, respectively. In addition, antibody RK5 was used to recover a cDNA clone (pRK5) encoding a portion of a 48-kDa keratin-related protein with unique tissue and cellular distribution, designated cytokeratin 21. Translation of cDNA-selected mRNAs yielded individual proteins which could be resolved and identified by their specific immunoreactivities. The pRK5 cDNA was used to recover a larger (approximately 1.3 kilobase pairs) cDNA clone (KB2) from an independent cDNA library for DNA sequence analysis and for the recovery of additional overlapping cDNA clones. The resulting cDNA sequence (1519 base pairs) contains the complete coding region of cytokeratin 21 (49,387 daltons). The predicted amino acid sequence of cytokeratin 21 confirms its identity as a novel type I cytokeratin expressed predominantly in the intestinal epithelium.