Affinity labeling of the active site and the reactive sulfhydryl associated with activation of rat liver phenylalanine hydroxylase

A pterin analogue, 5-[(3-azido-6-nitrobenzylidene)amino]-2,6-diamino-4-pyrimidinone (ANBADP), was synthesized as a probe of the pterin binding site of phenylalanine hydroxylase. The photoaffinity label has been found to be a competitive inhibitor of the enzyme with respect to 6,7-dimethyltetrahydropterin, having a Ki of 8.8 +/- 1.1 microM. The irreversible labeling of phenylalanine hydroxylase by the photoaffinity label upon irradiation is both concentration and time dependent. Phenylalanine hydroxylase is covalently labeled with a stoichiometry of 0.87 +/- 0.08 mol of label/enzyme subunit. 5-Deaza-6-methyltetrahydropterin protects against inactivation and both 5-deaza-6-methyltetrahydropterin and 6-methyltetrahydropterin protect against covalent labeling, indicating that labeling occurs at the pterin binding site. Three tryptic peptides were isolated from [3H]ANBADP-photolabeled enzyme and sequenced. All peptides indicated the sequence Thr-Leu-Lys-Ala-Leu-Tyr-Lys (residues 192-198). The residues labeled with [3H]ANBADP were Lys198 and Lys194, with the majority of the radioactivity being associated with Lys198. The reactive sulfhydryl of phenylalanine hydroxylase associated with activation of the enzyme was also identified by labeling with the chromophoric label 5-(iodoacetamido)fluorescein [Parniak, M. A., & Kaufman, S. (1981) J. Biol. Chem. 256, 6876]. Labeling of the enzyme resulted in 1 mol of fluorescein bound per phenylalanine hydroxylase subunit and a concomitant activation of phenylalanine hydroxylase to 82% of the activity found with phenylalanine-activated enzyme. Tryptic and chymotryptic peptides were isolated from fluorescein-labeled enzyme and sequenced. The modified residue was identified as Cys236.     

Amino acid sequence of NADH-cytochrome b5 reductase of human erythrocytes

The amino acid sequence of soluble NADH-cytochrome b5 reductase purified from normal human erythrocytes was determined as one approach to understand the hereditary disease of a deficiency of this enzyme. The protein is hydrophilic as a whole, but two regions, from Phe-36 to Ile-71 and from Met-231 to Phe-275, were found to be highly hydrophobic. The sequence of the latter region is particularly unique, and rich in proline (20%). The sequence of the amino-terminal region was very similar to the partial sequences of the corresponding regions of the enzymes from pig and steer liver microsomes.     

Interaction of fluorescently labeled myosin subfragment 1 with nucleotides and actin

Isolated myosin heads (subfragment 1) were modified by covalent attachment of 5-(iodoacetamido)fluorescein or 5-(iodoacetamido)salicylic acid to the essential sulfhydryl group SH1. The extrinsic fluorescence of the modified proteins was sensitive to binding of nucleotides and F-actin. With the fluorescein derivative [subfragment 1 (S1) modified with 5-(iodoacetamido)fluorescein (IAF) at SH1 (S1-AF)], association with MgADP decreased the probe fluorescence by 30%, whereas binding to actin increased the emission by a factor of 2. In the ternary complex acto-S1-AF X MgADP, the effect of nucleotide on the intensity of the attached fluorescein canceled the effect of actin. The fluorescence state of this ternary complex was similar to that of S1-AF X MgADP. The emission of S1-AF was resolved into two components with lifetimes of 4.3 and 0.6 ns and relative contributions of 33% and 67%, respectively. Interaction of S1-AF with nucleotides and actin did not alter the lifetimes but significantly shifted their fractional contributions. Quenching studies showed that the short lifetime likely arose from the fluorescein moiety statically quenched by internal groups. Binding of MgADP to the salicylate derivative [S1 modified with 5-(iodoacetamido)salicylic acid at SH1 (S1-SAL)] induced a 25% enhancement of the probe fluorescence, whereas formation of acto-S1-SAL decreased the emission by 10% regardless of whether MgADP was bound to the protein. Both labeled S1 species bound MgADP with a similar affinity, comparable to that of unmodified S1 previously reported by other investigators.(ABSTRACT TRUNCATED AT 250 WORDS)     

Complete amino acid sequence of NADH-cytochrome b5 reductase purified from human erythrocytes

The complete amino acid sequence of soluble NADH-cytochrome b5 reductase purified from human erythrocytes was determined. The enzyme, which contained 8 methionine residues, was cleaved by cyanogen bromide. The resulting nine peptides were separated by gel filtration and purified further by high-performance liquid chromatography. The purified peptides were sequenced by automated Edman degradation. Three large CNBr peptides, residues 1-101, 109-151, and 169-231, were further fragmented with trypsin, Staphylococcus aureus V8 protease or a lysyl endopeptidase of Achromobacter lyticus. The peptides obtained from the tryptic digest of citraconylated FAD-depleted apoprotein completed the alignments of the other peptides. The enzyme was composed of 275 amino acid residues. The 4 functionally important cysteine residues were located in the COOH-terminal portion. The molecular weight of the protein was calculated to be 31,260 without FAD. A prediction of the secondary structure was made by the method of Chou and Fasman. The protein was hydrophilic as a whole (43% polarity), but some regions were rich in hydrophobic residues. From the sequence homology of this enzyme with the pyridine nucleotide-binding sites of other flavoproteins, three candidates for the FAD and NADH-binding domains were suggested.     

Solid-phase synthesis and applications of N-(S-acetylmercaptoacetyl) peptides

The reagent pentafluorophenyl S-acetylmercaptoacetate was used to modify the N-terminus of resin-bound side-chain-protected peptides. The modification was carried out in an automated cycle in the final stage of fluorenylmethoxycarbonyl (Fmoc)/polyamide-mediated solid-phase synthesis. Side-chain deprotection and cleavage from the resin with aqueous trifluoroacetic acid gave the N-(S-acetylmercaptoacetyl) peptides. The S-acetylmercaptoacetyl peptides were transformed into reactive thiol-containing peptides by incubation with hydroxylamine at neutral pH. The S-deacetylation was performed in the presence of a sulfhydryl-reactive compound (or intramolecular group) to enable immediate capture of the sensitive thiol. Three applications were investigated. An S-acetylmercaptoacetyl peptide, containing a sequence of a meningococcal membrane protein, was incubated with hydroxylamine in the presence of 5-(iodoacetamido)fluorescein to give the corresponding fluorescein-labeled peptide in 62% yield. The same peptide was also S-deacetylated in the presence of bromoacetylated poly-L-lysine to afford a peptide/polylysine conjugate. Finally, a peptide corresponding to a sequence of herpes simplex virus glycoprotein D was prepared. This peptide, containing an N-terminal-S-acetylmercaptoacetyl group and an additional C-terminal S-(3-nitro-2-pyridinesulfenyl)cysteine residue, was converted into a cyclic disulfide peptide (20%).     

Isolation, characterization, and mapping of a human acid beta-galactosidase cDNA

A lambda gt11 human testicular cDNA library was screened with degenerate oligonucleotide probe mixtures based on amino acid sequence data generated from cyanogen bromide fragments and tryptic fragments of purified human beta-galactosidase. Six positive clones were identified after screening 2 x 10(6) plaques. The sequences of these six clones were determined and found to be derived from two different cDNAs. The sequence of the longest of these cDNAs is nearly identical to that recently determined by Oshima et al. (1988). It codes for a 76-kD protein and all 11 peptides that were generated from the purified enzyme. The second clone is shorter by 393 bp in the central portion of the coding region. Analysis by Northern blotting revealed the presence of a single mRNA species of 2.45 kb in lymphoblasts and testicular tissue. It is deduced from the amino acid sequence data that proteolytic processing of the precursor form of beta-galactosidase must occur by cleavage in the carboxy-terminal portion of the polypeptide perhaps around amino acid 530 at a uniquely hydrophilic sequence. Using a probe generated from the 3' region of the cDNA, we have mapped the locus coding for human beta-galactosidase to chromosome 3p21-3pter.     

Molecular cloning of human lysyl oxidase and assignment of the gene to chromosome 5q23.3-31.2.

Lysyl oxidase (EC 1.4.3.13) initiates the crosslinking of collagens and elastin by catalyzing oxidative deamination of the epsilon-amino group in certain lysine and hydroxylysine residues. We report here on the isolation and characterization of cDNA clones for the enzyme from human placenta and rat aorta lambda gt11 cDNA libraries. A cDNA clone for human lysyl oxidase covers all the coding sequences, 230 nucleotides of the 5' and 299 nucleotides, of the 3' untranslated sequences, including a poly(A) tail of 23 nucleotides. This cDNA encodes a polypeptide of 417 amino acid residues, including a signal peptide of 21 amino acids. Sequencing of two rat lysyl oxidase cDNA clones indicated six differences between the present and the previously published sequence for the rat enzyme [Trackman et al. (1990) Biochemistry 29: 4863-4870], resulting in frameshifts in the translated sequence. The human lysyl oxidase sequence was found to be 78% identical to the revised rat sequence at the nucleotide level and 84% identical at the amino acid level, with the degree of identity unevenly distributed between various regions of the coded polypeptide. Northern blot analysis of human skin fibroblasts RNA indicated that the human lysyl oxidase cDNA hybridizes to at least four mRNA species; their sizes are about 5.5, 4.3, 2.4, and 2.0 kb. Analysis of a panel of 25 human x hamster cell hybrids by Southern blotting mapped the human lysyl oxidase gene to chromosome 5, and in situ hybridization mapped it to 5q23.3-31.2.(ABSTRACT TRUNCATED AT 250 WORDS)     

Purification to homogeneity and partial amino acid sequence of a fragment which includes the methyl acceptor site of the human DNA repair protein for O6-methylguanine.

DNA repair by O6-methylguanine-DNA methyltransferase (O6-MT) is accomplished by removal by the enzyme of the methyl group from premutagenic O6-methylguanine-DNA, thereby restoring native guanine in DNA. The methyl group is transferred to an acceptor site cysteine thiol group in the enzyme, which causes the irreversible inactivation of O6-MT. We detected a variety of different forms of the methylated, inactivated enzyme in crude extracts of human spleen of molecular weights higher and lower than the usually observed 21-24kDa for the human O6-MT. Several apparent fragments of the methylated form of the protein were purified to homogeneity following reaction of partially-purified extract enzyme with O6-[3H-CH3]methylguanine-DNA substrate. One of these fragments yielded amino acid sequence information spanning fifteen residues, which was identified as probably belonging to human methyltransferase by virtue of both its significant sequence homology to three procaryote forms of O6-MT encoded by the ada, ogt (both from E. coli) and dat (B. subtilis) genes, and sequence position of the radiolabelled methyl group which matched the position of the conserved procaryote methyl acceptor site cysteine residue. Statistical prediction of secondary structure indicated good homologies between the human fragment and corresponding regions of the constitutive form of O6-MT in procaryotes (ogt and dat gene products), but not with the inducible ada protein, indicating the possibility that we had obtained partial amino acid sequence for a non-inducible form of the human enzyme. The identity of the fragment sequence as belonging to human methyltransferase was more recently confirmed by comparison with cDNA-derived amino acid sequence from the cloned human O6-MT gene from HeLa cells (1). The two sequences compared well, with only three out of fifteen amino acids being different (and two of them by only one nucleotide in each codon).     

Primary structure and gene localization of human prolidase

Complementary DNA clones of prolidase (imidodipeptidase, EC 3.4.13.9) were isolated from human liver and placental cDNA libraries. Two clones named lambda PL21 and lambda PP6 from the liver and placental cDNA libraries, respectively, were analyzed in detail. The first clone, lambda PL21, carried a cDNA insert of 1.7 kilobase pairs and covered all the coding region of human prolidase mRNA. The second clone, lambda PP6, contained a 1.8-kilobase insert with a full-length 3'-untranslated region. Comparison of the amino acid sequence predicted from the nucleotide sequence of the cDNA insert of the two clones with the partial amino acid sequence determined by Edman degradation of peptides derived from human erythrocyte prolidase established that both clones code for human prolidase. The amino terminus of the human mature enzyme is blocked and seems to begin with the sequence X-Ala-Ala-Ala. Presumably no processing occurs at the carboxyl terminus. The mature enzyme is composed of 492 residues, corresponding to Mr 54,305. The sequence of prolidase is unique and not similar to any known protein, except for a significant similarity to regions of F1-ATPase alpha and beta subunits from various sources. The gene has been mapped to the short arm of chromosome 19 (19p13.2). Elucidation of the complete amino acid sequence and the gene location of prolidase should provide the basis for understanding structure-function relationships and also inherited disorders caused by deficiency of this metabolically important enzyme.     

Sequence of the cDNA and gene for angiogenin, a human angiogenesis factor

Human cDNAs coding for angiogenin, a human tumor derived angiogenesis factor, were isolated from a cDNA library prepared from human liver poly(A) mRNA employing a synthetic oligonucleotide as a hybridization probe. The largest cDNA insert (697 base pairs) contained a short 5'-noncoding sequence followed by a sequence coding for a signal peptide of 24 (or 22) amino acids, 369 nucleotides coding for the mature protein of 123 amino acids, a stop codon, a 3'-noncoding sequence of 175 nucleotides, and a poly(A) tail. The gene coding for human angiogenin was then isolated from a genomic lambda Charon 4A bacteriophage library employing the cDNA as a probe. The nucleotide sequence of the gene and the adjacent 5'- and 3'-flanking regions (4688 base pairs) was then determined. The coding and 3'-noncoding regions of the gene for human angiogenin were found to be free of introns, and the DNA sequence for the gene agreed well with that of the cDNA. The gene contained a potential TATA box in the 5' end in addition to two Alu repetitive sequences immediately flanking the 5' and 3' ends of the gene. The third Alu sequence was also found about 500 nucleotides downstream from the Alu sequence at the 3' end of the gene. The amino acid sequence of human angiogenin as predicted from the gene sequence was in complete agreement with that determined by amino acid sequence analysis. It is about 35% homologous with human pancreatic ribonuclease, and the amino acid residues that are essential for the activity of ribonuclease are also conserved in angiogenin. This provocative finding is thought to have important physiological implications.     

Primary structure of human poly(ADP-ribose) synthetase as deduced from cDNA sequence

Human poly(ADP-ribose) synthetase consists of three proteolytically separable domains, the first for binding of DNA, the second for automodification, and the third for binding of the substrate, NAD (Ushiro, H., Yokoyama, Y., and Shizuta, Y. (1987) J. Biol. Chem. 262, 2352-2357). We have isolated and sequenced cDNA clones for the enzyme using synthesized oligodeoxyribonucleotide probes based on the partial amino acid sequence of the protein. The open reading frame determined encodes a protein of 1,013 amino acid residues with a molecular weight of 113,203. The deduced amino acid sequence is consistent with the partial amino acid sequences of tryptic or alpha-chymotryptic peptides and the total amino acid composition of the purified enzyme. The native enzyme is relatively hydrophilic as judged from the hydrophilicity profile of the total amino acid sequence. The net charge of the NAD binding domain is neutral but the DNA binding domain and the automodification domain are considerably rich in lysine residue and quite basic. The DNA binding domain involves a homologous repeat in the sequence and exhibits a sequence homology with localized regions of transforming proteins such as c-fos and v-fos. Furthermore, this domain contains a unique sequence element which resembles the essential peptide sequences for nuclear location of SV40 and polyoma virus large T antigens. These facts suggest the possibility that the physiological function of poly(ADP-ribose) synthetase lies in its ability to bind to DNA and to control transformation of living eukaryotic cells like the cases of those oncogene products.     

Identification of human porins. II. Characterization and primary structure of a 31-lDa porin from human B lymphocytes (Porin 31HL)

We characterize and describe for the first time the primary structure of a human porin with the molecular mass of 31 kDa derived from the plasmalemm of B-lymphocytes (Porin 31HL). Porin 31HL is shown to be a basic, channel forming membrane protein. The protein chain is composed of 282 amino acids with a relative molecular mass of 30641 Da without derivatisation. It is not a glycoprotein. The N-terminus is acetylated. Altogether the amino-acid sequence shows 56% hydrophilic or charged amino acids arranged in alternating regions of hydrophilic or hydrophobic character as it is typical for porins. In addition the 18 N-terminal amino acids of Porin 31HL can be arranged to an amphilic alpha-helix like in other porins. Porin 31HL shows approx. 29% or 24% identity to the primary structure of mitochondrial porins of Neurospora crassa and Saccharomyces cerevisiae. Partial data on mitochondrial porins from rat kidney and beef heart show sequence identity of about 90% to the human B cell porin elaborated here.     

Cloning and expression of human 75-kDa inositol polyphosphate-5-phosphatase.

Inositol polyphosphate-5-phosphatase (5-phosphatase) hydrolyzes inositol 1,4,5-trisphosphate and inositol 1,3,4,5-tetrakisphosphate and thereby functions as a signal terminating enzyme in cellular calcium ion mobilization. A cDNA encoding human platelet 5-phosphatase has been isolated by screening for beta-galactosidase fusion proteins that bind to inositol 1,3,4,5-tetrakisphosphate. The sensitivity of the screening procedure was enhanced 50- to 100-fold by amplification of "sublibraries" prior to carrying out binding assays. The sequences derived from the "expression clone" were used to screen human erythroleukemia cell line and human megakaryocytic cell line cDNA libraries. We obtained two additional clones which together consist of 2381 base pairs. The amino-terminal amino acid sequence from the 75-kDa 5-phosphatase purified from platelets is identical to amino acids 38-56 predicted from the cDNA. This suggests that the platelet 5-phosphatase is formed by proteolytic processing of a larger precursor. The cDNA predicts that the mature enzyme contains 635 amino acids (Mr 72, 891). Antibodies directed against recombinant TrpE fusion proteins of either an amino-terminal region or a carboxyl-terminal region immunoprecipitate the enzyme activity from a preparation of the 75-kDa form of platelet 5-phosphatase (Type II) but do not precipitate the distinct 47-kDa 5-phosphatase (Type I) also found in platelets. In addition, the recombinant protein expressed in Cos-7 cells has the same 5-phosphatase activity as the platelet 5-phosphatase.     

Protein-protein interactions in contact activation of blood coagulation. Characterization of fluorescein-labeled human high molecular weight kininogen-light chain as a probe

Limited proteolysis of high molecular weight kininogen by kallikrein resulted in the generation of an inactive heavy chain of Mr = 64,000 and active light chains of Mr = 64,000 and 51,000 when analyzed by sodium dodecyl sulfate (SDS)-gel electrophoresis under reducing conditions. Starting with kininogen from outdated plasma, a light chain with an apparent molecular weight of 51,000 on 7.5% SDS gels was purified and characterized. Molecular weights of 28,900 +/- 1,100 and 30,500 +/- 1,600 were obtained by gel filtration of the reduced and alkylated protein in 6 M guanidine HCl and equilibrium sedimentation under nondenaturing conditions in the air-driven ultracentrifuge, respectively. The light chain stained positively with periodic acid-Schiff reagent on SDS gels indicating that covalently attached carbohydrate may be responsible for the anomalously high molecular weight estimated by SDS-gel electrophoresis. A single light chain thiol group reacted with 5,5'-dithiobis-(2-nitrobenzoic acid) (DTNB) in the presence and absence of 6 M guanidine HCl. Specific fluorescent labeling of the thiol group with 5-(iodoacetamido)fluorescein (IAF) occurred without loss of clotting activity. Addition of purified human plasma prekallikrein to the IAF-light chain resulted in a maximum increase in fluorescence anisotropy of 0.041 +/- 0.001 and no change in the fluorescence intensity. Fluorescence anisotropy measurements of the equilibrium binding of prekallikrein to the IAF-light chain yielded an average Kd of 17.3 +/- 2.5 nM and stoichiometry of 1.07 +/- 0.07 mol of prekallikrein/mol of IAF-light chain. Measurements of the interaction of prekallikrein with iodoacetamide-alkylated light chain using the IAF-light chain as a probe gave an average Kd of 16 +/- 4 nM and stoichiometry of 1.0 +/- 0.2 indicating indistinguishable affinities for prekallikrein.     

A cDNA clone for the precursor of rat mitochondrial ornithine transcarbamylase: comparison of rat and human leader sequences and conservation of catalytic sites.

We have cloned a DNA complementary to the messenger RNA encoding the precursor of ornithine transcarbamylase from rat liver. This complementary DNA contains the entire protein coding region of 1062 nucleotides and 86 nucleotides of 5'- and 298 nucleotides of 3'-untranslated sequences. The predicted amino acid sequence has been confirmed by extensive protein sequence data. The mature rat enzyme contains the same number of amino acid residues (322) as the human enzyme and their amino acid sequences are 93% homologous. The rat and human amino-terminal leader sequences of 32 amino acids, on the other hand, are only 69% homologous. The rat leader contains no acidic and seven basic residues compared to four basic residues found in the human leader. There is complete sequence homology (residues 58-62) among the ornithine and aspartate transcarbamylases from E. coli and the rat and human ornithine transcarbamylases at the carbamyl phosphate binding site. Finally, a cysteine containing hexapeptide (residues 268-273), the putative ornithine binding site in Streptococcus faecalis, Streptococcus faecium, and bovine transcarbamylases, is completely conserved among the two E. coli and the two mammalian transcarbamylases.     

Cloning and expression of human salivary-gland kallikrein in Escherichia coli.

CDNA clones for human kallikrein have been identified in a cDNA library constructed from mRNA of human salivary gland. The entire coding sequence for preprokallikrein and for the 5'- and 3'-untranslated regions were isolated by using a mixture of oligonucleotides corresponding to amino acids 51-56 of human urinary kallikrein and one oligonucleotide corresponding to amino acids 233-238 of human pancreatic kallikrein. The DNA sequence proved that, with the exception of two amino acid exchanges, kallikrein of the human salivary gland is identical with pancreatic kallikrein. Salivary gland and renal kallikrein was expressed in Escherichia coli from plasmid pKK223-3 under the control of the tac promoter. The protein was identified by Western-blot analysis and by demonstration of its specific proteolytic activity.     

Selective radiolabeling and isolation of the hydrophobic membrane-binding domain of human erythrocyte acetylcholinesterase

The hydrophobic, membrane-binding domain of purified human erythrocyte acetylcholinesterase was labeled with the photoactivated reagent 3-(trifluoromethyl)-3-(m-[125I]iodophenyl)diazirine. The radiolabel was incorporated when the enzyme was prepared in detergent-free aggregates, in detergent micelles, or in phospholipid liposomes, but the highest percentage of labeling occurred in the detergent-free aggregates. Papain digestion of the enzyme released the hydrophobic domain, and polyacrylamide gel electrophoresis in sodium dodecyl sulfate or gel exclusion chromatography demonstrated that the label was localized exclusively in the cleaved hydrophobic domain fragment. This fragment was purified in a three-step procedure. Digestion was conducted with papain attached to Sepharose CL-4B, and the supernatant was adsorbed to acridinium affinity resin to remove the hydrophilic enzyme fragment. The nonretained fragment associated with Triton X-100 micelles was then chromatographed on Sepharose CL-6B, and finally detergent was removed by chromatography on Sephadex LH-60 in an ethanol-formic acid solvent. The fragment exhibited an apparent molecular weight of 3100 on the Sephadex LH-60 column when compared with peptide standards. However, amino acid analysis of the purified fragment revealed only 1 mol each of histidine and glycine per mole of fragment in contrast to the 25-30 mole of amino acids expected on the basis of the molecular weight estimate. This result suggests a novel non-amino acid structure for the hydrophobic domain of human erythrocyte acetylcholinesterase.     

Isolation of cDNA clones and complete amino acid sequence of human erythrocyte glycophorin C

Two cDNA clones for glycophorin C, a transmembrane glycoprotein of the human erythrocyte which carries the blood group Gerbich antigens, have been isolated from a human reticulocyte cDNA library. The clones were identified with a mixture of 32 oligonucleotide probes (14-mer) which have been synthetized according to the amino acid sequence Asp-Pro-Gly-Met-Ala present in the N-terminal tryptic peptide of the molecule. The primary structure of glycophorin C deduced from the nucleotide sequence of the 460 base-pair insert of the pGCW5 clone indicates that the complete protein is a single polypeptide chain of 128 amino acids clearly organized in three distinct domains. The N-terminal part (residues 1-57, approximately) which is N- and O-glycosylated is connected to a hydrophilic C-terminal domain (residues 82-128, approximately) containing 4 tyrosine residues by a hydrophobic stretch of nonpolar amino acids (residues 58-81, approximately) probably interacting with the membrane lipids and permitting the whole molecule to span the lipid bilayer. Northern blot analysis using a 265-base-pair restriction fragment obtained by DdeI digestion of the inserted DNA shows that the glycophorin C mRNA from human erythroblasts is approximately 1.4 kilobases long and is present in the human fetal liver and the human K562 and HEL cell lines which exhibit erythroid features. The glycophorin C mRNA, however, is absent from adult liver and lymphocytes, indicating that this protein represents a new erythrocyte-specific probe which might be useful to study erythroid differentiation.