Cotranslational folding of globin.

Globin synthesis in a wheat germ cell-free translation system was performed in the presence of [3H]hemin and [35S]methionine to determine the minimal length of the nascent ribosome-bound globin chain capable of heme binding. Nascent polypeptides of predetermined size were synthesized on ribosomes by translation of truncated mRNA molecules. Analysis with the use of sucrose gradient centrifugation and puromycin reaction revealed that the ribosome-bound N-terminal alpha-globin fragments of 140, 100, and 86 amino acid residues are capable of an efficient heme binding, whereas those of 75, 65, and 34 amino acid residues display a significantly weaker, or just nonspecific, affinity to heme. This indicates that the ribosome-bound nascent chain of 86 amino acid residues has already acquired a spatial structure that allows its interaction with the heme group or that heme attachment promotes the formation of the proper tertiary structure in the ribosome-bound nascent peptide. In any case the cotranslational folding of globin is suggested.

Globin synthesis in a wheat germ cell-free translation system was performed in the presence of [ 3 H]hemin and [ 35 S]methionine to determine the minimal length of the nascent ribosome-bound globin chain capable of heme binding. Nascent polypeptides of predetermined size were synthesized on ribosomes by translation of truncated mRNA molecules. Analysis with the use of sucrose gradient centrifugation and puromycin reaction revealed that the ribosome-bound N-terminal ␣-globin fragments of 140, 100, and 86 amino acid residues are capable of an efficient heme binding, whereas those of 75, 65, and 34 amino acid residues display a significantly weaker, or just nonspecific, affinity to heme. This indicates that the ribosome-bound nascent chain of 86 amino acid residues has already acquired a spatial structure that allows its interaction with the heme group or that heme attachment promotes the formation of the proper tertiary structure in the ribosome-bound nascent peptide. In any case the cotranslational folding of globin is suggested.
The mechanism by which the growing polypeptide chain acquires its native conformation has been discussed in many recent reviews (1)(2)(3)(4)(5). Historically, most protein-folding studies were based on the analysis of protein refolding in vitro. These studies have provided basic insights into the principles and mechanisms governing the folding of polypeptides into compact three-dimensional structures (6 -9). However, the in vivo folding is assumed to proceed cotranslationally (10 -15). Evidence suggesting the cotranslational protein folding has come from experiments that demonstrated (i) the enzymatic activity of the growing polypeptide on the ribosome (16 -21), (ii) the formation of correct epitopes able to bind corresponding conformational antibodies (22), and (iii) the formation of correct disulfide crossbridges in the growing nascent chains (23)(24)(25)(26)(27)(28).
Previously we reported data indicating that heme attachment to the globin chains may proceed cotranslationally (29). Analysis of globin synthesis in cell-free extracts of rabbit reticulocytes carried out in the presence of 3 H-labeled hemin revealed the presence of [ 3 H]hemin in the polyribosome fraction synthesizing globin chains. The addition of puromycin resulted in the release of both [ 3 H]hemin and 14 C-labeled leucine polypeptide from the polyribosomes. The data obtained indicated cotranslational heme binding to the nascent globin chains and thus to the cotranslational folding of the globin molecule since heme binding in the case of globin molecules is a function of the definite protein structure (30,31). However, the possibility of globin tetramer assembly on the ribosome in those experiments could not be excluded. Cotranslational trimerization of the retrovirus cell attachment protein, 1, has been demonstrated recently (32), and thus the assembly of the nascent globin chains could as well be the case. If this was the case, the previously obtained data could be alternatively explained by the presence of complete globin chains (with labeled hemin) associated with the nascent chain.
To rule out this possibility, we have used the wheat germ cell-free translation system, which does not contain endogenous globin molecules. We have also performed the translation experiments with ␣-globin synthesis, which, in contrast to ␤-globin, does not form the tetramer structure (31). We have found that ␣-globin is capable of heme binding during its synthesis on the ribosome. In addition, we have demonstrated that incomplete ␣-globin molecules of 140, 100, and 86 amino acid residues (lengths are given excluding the first initiator methionine) are capable of cotranslational heme binding with an approximately equal efficiency, whereas polypeptide chains of 75, 65, and 34 amino acid residues display a significantly weaker, or just nonspecific, affinity to heme. This shows that the nascent chain of 86 amino acid residues possesses a spatial structure that allows its interaction with the heme group or that the heme attachment promotes the formation of the proper tertiary structure of the growing polypeptide on the ribosome. Hence, the cotranslational folding of globin molecule is suggested.

EXPERIMENTAL PROCEDURES
In Vitro Transcription of ␣-Globin cDNA-The plasmid PHST101 (rabbit ␣-globin subclone PSP64) containing ␣-globin cDNA under control of the SP6 promotor was the gift of Professor J. Ilan, Case Western Reserve University. The transcription reaction was carried out according to Gurevich et al. (33) in 500 l (total volume) of 80 mM HEPES-KOH buffer, pH 7.5, containing 16 mM MgCl 2 , 2 mM spermidine, 20 mM dithiothreitol (DTT), 1  Cell-free Protein Synthesis-Cell-free translation of the ␣-globin mRNA was performed using wheat germ extract as described by Clemens (35). The reaction mixture contained 20 mM HEPES-KOH buffer, pH 7.5, 3 mM Mg(CH 3 COO) 2 (29). The final reaction volume was 100 l. The translation was carried out at 24°C. After 25 min of incubation, the * This work was supported by the Russian Academy of Sciences and by Grant 93-04-6983 from the Russian Foundation for Fundamental Research. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
¶ To whom correspondence should be addressed. Tel./Fax: 7-095-924-0493; E-mail: spirin@sun.ipr.serpukhov.su. reaction mixtures were subdivided into two equal portions. Puromycin was added to a final concentration of 1.5 mM to one of the portions, and the incubation of both portions was continued for an additional 10 min.
10 g of full-sized globin mRNA was incubated at 37°C with a 50-fold molar excess of a complementary 20-mer oligodeoxyribonucleotide and 100 units of RNase H from Escherichia coli. The reaction was carried out for 1 h in 40 mM Tris-HCl buffer, pH 7.6, containing 1 mM DTT, 1 mM MgCl 2 , and 30 mg/ml bovine serum albumin. Truncated globin mRNAs obtained after digestion were used in the translation experiments (Fig. 1). Completeness of RNA digestion was controlled by 5% polyacrylamide gel electrophoresis in the presence of 7 M urea. To check the specificity of RNase H action, 3Ј-ends of the truncated RNA molecules were determined, and the truncated ␣86 mRNA molecule was sequenced using the Pharmacia RNA sequencing enzyme kit according to the procedure described in Ref. 37. Production of globin peptides of the defined lengths after translation of the truncated mRNAs in the cell-free system was controlled by electrophoresis as described (38).
Polymerase Chain Reaction (PCR)-PCR fragments were generated using the oligonucleotides 5Ј-GATTTAGGTGACACTATAGAATACA-3Ј as the downstream primer and 5Ј-GTCGCTGAGAGTAGACAG-3Ј as the upstream primer. The plasmid pHST101 was linearized with BamHI and then amplified using Taq DNA polymerase by 30 cycles of PCR consisting of 1 min at 94°C, 1 min at 40°C, and 1 min at 72°C each.
Sucrose Gradient Centrifugation-Sucrose gradient centrifugation was used to analyze the incorporation of radiolabeled methionine as

Full-length ␣-Globin Is Capable of Cotranslational Heme
Binding-It was shown earlier that globin molecules are capable of cotranslational binding of heme in a homologous rabbit reticulocyte cell-free system (29). Here we used the same methodology but with the wheat germ cell-free system. The question was whether the individual globin molecule (␣or ␤-chain) is capable of cotranslational heme binding. For the in vitro translation in a wheat germ system we chose the ␣-globin that has a 10-fold stronger ability to bind the heme group as compared with the ␤-globin (30,39) and, in contrast to ␤-globin, does not form the tetramer structure (31). approximately 100 amino acid residues and longer were capable of binding the heme (29).
To answer this question, the method of translation arrest by antisense oligodeoxynucleotides was chosen. The translation of mRNAs, truncated by RNase H in the presence of complementary oligonucleotides, results in nascent polypeptides of predetermined lengths attached to the ribosome as was reported previously (46). It was also reported that the dominant RNase H cut occurs at the RNA 5Ј-end in the RNA oligodeoxyribonucleotide complex in conditions of complete hydrolysis (47). Thus, the length of nascent peptide is predictable (48,49). We started with the globin mRNA lacking the last 3Ј-terminal coding triplet. Sucrose gradient centrifugation analysis revealed that such an incomplete nascent globin peptide (␣-globin 140; the length here does not include the first initiator methionine) is capable of heme binding during translation in the wheat germ system (not shown). Using the same approach, we found that an incomplete globin peptide of 100 amino acid residues (␣-globin 100) is also capable of cotranslational heme binding (Fig. 3A). Surprisingly, the same has been demonstrated for a shorter globin peptide of 86 amino acid residues lacking the heme-binding histidine residue at position 87 (Fig.  3B). Determination of the 3Ј-end and sequencing of the ␣86 mRNA showed slight heterogeneity of the ␣86 mRNA 3Ј-end (not shown). To prove that an ␣-globin peptide of 86 amino acid residues is capable of heme binding, we performed the experiments with ␣86 mRNA obtained after SP6 transcription of a PCR-generated template containing SP6 promotor and the corresponding part of the ␣-globin coding sequence. Results obtained in these additional experiments proved the initial observation (Fig. 3C).
Shorter ␣-Globin Peptides of 75, 65, and 34 Amino Acid Residues Display a Weaker or Nonspecific Cotranslational Heme Binding-As shown previously in in vitro experiments, the proteolytic fragment of the ␤-globin molecule comprising residues 31-104 is capable of binding the heme group (50). The same phenomenon has been also demonstrated recently for the mini-myoglobin polypeptide fragment (residues 32-139) (51). To determine the length of a globin chain sufficient and necessary for cotranslational heme binding, we translated truncated ␣-globin mRNAs (Fig. 1) and produced shorter nascent peptides of 75, 65, and 34 amino acid residues. Incorporation of radiolabeled methionine as well as hemin into the nascent globin peptides was controlled after sucrose gradient centrifugation of the wheat germ translation system as described above. In contrast to the experiments with longer polypeptides (Fig. 3, A  and B), we found much lower incorporation of [ 3 H]hemin either into the ribosome-bound polypeptides of 34 (not shown), 65, or 75 amino acid residues (Fig. 3, D and E).
Since all polypeptides under investigation have an equal number of methionine residues (at positions 1 and 33), the [ 3 H]hemin/[ 35 S]methionine ratio can be used as a measure of the efficiency of heme binding to the nascent polypeptides of various lengths. Fig. 4 presents the summary of puromycin effects on the release of nascent 35 S-labeled methionine polypeptide and [ 3 H]hemin from ribosomes after the translation of full-length and truncated mRNAs. Some amounts of hemin found attached to the shorter peptides can probably be attributed to a nonspecific heme adsorption. Alternatively, one can speculate that heme begins to dock to the globin polypeptide chain very early during its synthesis when the first amino acid residues competent for heme binding appear from the peptidyltransferase center, but at these stages the heme binding is not strong enough to form a stable heme-globin complex. DISCUSSION A number of experimental findings suggest cotranslational protein folding (16 -28, 32, 52). Thus, the possibility of cotranslational ligand binding cannot be excluded. Studies of biosynthesis of protein D1 of the membrane-bound chloroplast reaction center directly indicated the cotranslational binding of chlorophyll to an incomplete D1 molecule (53,54). Cotranslational binding was also assumed to take place for the heme group in the case of globins (29,55). It is well known that the detachment or displacement of heme groups is accompanied by denaturation of hemoglobin, whereas the addition of the heme group to apohemoglobin or apomyoglobin promotes the formation of the native structure of the molecules (30, 56 -58). It was shown that mini-apo-␤-globin (residues 31-104) or mini-apomyoglobin (residues 32-139) reconstituted with natural heme preserved conformations similar to those in the whole molecules (50,51). Moreover, it was evident that the heme orientation in the pocket and the coordination state of the ferrous iron in the mini-globins are just the same as in the whole molecules (51). On the grounds of these data we suggest that heme binding to the nascent globin chains can be used as a test of cotranslational folding of globins. The aim of this work is to determine the length of the nascent globin chain on which heme attachment occurs during translation.
We have demonstrated an equally efficient cotranslational incorporation of [ 3 H]hemin into nascent globin chains of 140, 100, and 86 amino acid residues (as well as into the full-length molecule). The fact that puromycin (known to release ribosomebound peptides) causes the release of both 35 S-labeled methionine polypeptide and [ 3 H]hemin from ribosomes (Figs. 2 and 3) indicates that either the nascent peptide of 86 amino acid residues possesses a spatial structure allowing its proper interaction with the heme group or the heme attachment promotes the formation of the proper tertiary structure. Hence, the cotranslational formation of the spatial structure of globin at the early stages of its synthesis is likely. It was recently shown that chemically synthesized peptides with chymotrypsin inhibitor-2 growing from the N terminus acquire the three-dimensional structure in vitro while achieving the length of 62-63 amino acid residues (59). We believe that the same occurs in the case of nascent ␣86 (and longer) globin polypeptides growing on the ribosome.
To illustrate our results, we present wire frame models of the well known three-dimensional structures of the human deoxyhemoglobin ␣-chain ( Fig. 5A and Ref. 60) and the incomplete ␣86-globin chain (Fig. 5B). The ␣86-globin model was produced from the crystal structure of ␣-chain by skipping the C-terminal residues. The length of the incomplete ␣86 molecule in the model is the same as the lengths of nascent peptides in our experiments. There are a number of amino acid residues in the ␣-globin molecule known to be involved in the formation of contacts with the heme group. Among them are Met-32, Tyr-42, Phe-43, His-45, Phe-46, Lys-61, Val-62, Ala-65, Leu-83, Leu-86, Leu-91, Val-93, Asn-97, Phe-98, Leu-101, Leu-136, and two heme-binding His residues, His-58 and His-87 (60). Our results show that the nascent peptide of 86 amino acid residues lacking the heme-binding histidine residue at position 87 is capable of heme binding. This may indicate that the contacts provided by the remaining residues are sufficient for specific heme binding. Eleven residues involved in heme binding remain in the incomplete polypeptide of 86 amino acid residues (Fig. 5B). Since the incomplete nascent peptide of 75 amino acid residues does not bind the heme group so efficiently (Fig. 3D), it can be speculated that the leucine residues at positions 83 and 86 provide the necessary contacts, thus forming part of the heme pocket and allowing the incomplete nascent chain of 86 amino acid residues to bind the heme group quite effectively.
If this is the case, all the C-terminal sections of the polypeptide must appear from the ribosome and form E and F helices. It presumes that a nascent polypeptide chain can fold immediately at the peptidyltransferase center of the ribosome and no intraribosomal tunnel exists, as was indicated previously (61)(62)(63).
The alternative case could be that a ribosomal tunnel or channel hides the C-terminal section of a growing polypeptide (64,65). Then the nascent ribosome-bound globin peptide of 86 amino acid residues should be able to bind hemin either without F helix, if just 15 amino acid residues are hidden (41,42), or without both E and F helices, if 30 -40 amino acids are accommodated within the ribosome (39,42,44). The latter seems unlikely. Rather, the shielding of the C-terminal part of the growing nascent peptide can now be explained by nascent polypeptide-associated complex binding (63,66), known to protect about 30 C-terminal amino acid residues from the proteolysis. As nascent polypeptide-associated complex cycle of binding and release was proposed (63), it can be speculated that nascent polypeptide-associated complex is released from the nascent peptides when it acquires the three-dimensional structure. Taking into account all the facts and considerations mentioned above, we argue that the globin polypeptide chain begins to correctly fold with the participation of heme rather early during its elongation on the ribosome.