Structure of an Antifreeze Polypeptide Precursor from the Sea Raven,

The cystine-rich antifreeze polypeptides (AFP) from sea raven were fractionated by reverse-phase high performance liquid chromatography into several components, with SR2 (M, 17,000) as the major AFP. Sea raven AFP cDNA clones were isolated from a liver cDNA library using a synthetic oligonucleotide, and the identity of one of the clones, C2-1, was confirmed by hybridization selection and cell-free translation. C2-1 encodes a pre-AFP of 195 amino acids with no evidence of any profragments. Comparison of the de- duced amino acid sequence with partial peptide sequences from SR2 showed substitutions in at least four amino acid positions, suggesting that C2-1 cDNA codes for a minor component. Both the primary and the predicted secondary structures of sea raven AFP are completely different from those of other fish AFP. This further confirms that sea raven AFP belongs to a different class of antifreezes. The high frequency of re- verse turns and the presence of paired hydrophilic amino acids in these structures are striking features of the protein and may contribute to their antifreeze ac-tion.

The abbreviations used are: AFP, antifreeze polypeptides; HPLC, high ~r f o~a n c e liquid chromatography; SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis; BSA, bovine serum albumin; ND, not determined. and its sequence does not indicate'the presence of any repeating amino acid unit (8). Sea raven AFP represents yet a different type of AFP. It contains average amounts of alanine; and to date, it is the only fish AFP which contains a significant amount of half-cystine. The role of cystine appears to be important as the activity of sea raven AFP is sensitive to sulfhydryl reagents. Circular dichroism studies indicate the presence of @-structure and little a-helical content (9). Antibodies to sea raven AFP do not cross-react with either winter flounder or shorthorn sculpin AFP. This immunological specificity is consistent with sea raven AFP being a different type of antifreeze protein (9).
In order to further understand the structure-function relationship in antifreeze polypeptides, we have determined the structure of a precursor to sea raven AFP.

Microheterogeneity and Size of Sea Raven AFP--We have
demonstrated earlier that sea raven AFP can be fractionated into two major and three minor components by reverse-phase HPLC (12). However, slightly different profiles were obtained from AFP isolated from different pools of fish sera (Fig. 1). The variation was demonstrated in the relative amounts of each component. Nonetheless, SR2 was the major component.
The sea raven AFP components showed similar molecular weights, as estimated by SDS-PAGE (Fig. 2). Sephadex G-75-purified AFP and all five AFP components purified by reverse-phase HPLC were analyzed with or without reduction and alkylation. The estimated molecular weight of nonreduced AFP was about 14,000 and was 17,000 for the reduced and alkylated AFP. These results are similar to the range of 14,500-16,000 for the AFP, as previously estimated from gel filtration chromatography by Slaughter et al. (9).
The amino acid composition of SR2 is presented in Table  1. There is a significant level of half-cystine, an amino acid not found in any other fish AFP, A~though present in modest amounts, alanine is still the most abundant amino acid, and its level is similar to that found in ocean pout AFP (29). This is in contrast to the 60 mol % of alanine in AFP from winter flounder and shorthorn sculpin (4,7 ) . The amino acid compositions of the other sea raven AFP components are similar Portions of this paper (including "Experimental Procedures," Figs. 1-6 and 9, and Table 1) are presented in miniprint at the end of this paper. Miniprint is easily read with the aid of a standard magnifying glass. Full size photocopies are available from the Journal of Biological Chemistry, 9650 Rockville Pike, Bethesda, MD 20814. Request Document No. 86M-1875, cite the authors, and include a check or money order for $6.00 per set of pho~copies. Full size photocopies are also included in the microfilm edition of the Journal that is available from Waverly Press.

Stru&ture of Sea
Raven A n t i f r e~~e Polypeptide 15691 to that of SR2, as is the case with the composition of the components from other fish AFP (4, 7,291. The number of cysteine residues in SR2 was estimated. SR2 was reduced and alkylated with varying ratios of charged and neutral alkylating agents, followed by elect~phoretic separation based on charge differences (Fig. 3). The number ( n ) of cysteine residues was determined by counting the number of bands ( n + 1). A total of 12 bands corresponding to 11 cysteine residues is shown (lune 6).
Specificity of Antibodies to SRZ-Polyclonal antibodies were produced against the major component SR2 and were used to check the immunological relationship of SR2 with the other sea raven AFP and AFP from other fish. Results from the immunoblot (Fig. 4) show that antibodies against SR2 crossreact"with the other sea raven AFP, thus establishing their immunological similarity. These antibodies do not react with AFP from winter flounder, shorthorn sculpin, or ocean pout, indicating the immunological or structural differences between sea raven AFP and AFP from these other fish.
Z~e n t i f~c u~~n of the 3iosyn~het~c Precursor-RNA enriched for the 10-11 S mRNA species was recovered from a sucrose density gradient (Fig. 5, lane 4 ) and translated in a rabbit reticulocyte-lysate cell-free translation system. The translat.ion products were immunoprecipi~ted with antibodies to SR2. The reduced and alkylated immunoprecipitates and the reduced and ~3H]iodoacetate-alkylated SR2 were subjected to SDS-PAGE followed by fluorography. Immunoprecipitation of the translation products gave one predominant 20,000dalton and one minor 14,000-dalton product (Fig. 6A, lane 1 ) . Before the immunoprecipitation, the 14,000-dalton product was the major species (see Fig. 6B, lane 2). Immunoprecipitation resulted in an enrichment of the larger species. In the presence of nonlabeled SR2, a significant decrease in the recovery of the larger product provides evidence for its competition with SR2 for antibodies (Fig. 6A, lane 2) and thus establishes its identity as the biosynthetic precursor for sea raven AFP. This precursor is about 20,000 daltons, whereas the mature AFP (Fig. 6A, lane 3 ) is about 17,000 daltons. The difference of about 3,000 daltons could be accounted for by the presence of a signal peptide on the precursor. It does not appear to contain an additional prosequence. Zsolation and Sequence of a Sea Raven A F P cDNA-A mixed 17-mer synthetic oligonucleotide was derived from the amino acid sequence of a tryptic peptide of SR2. This probe was used to screen clones from a cDNA library enriched for the liver poly(A)+ 10-11 S RNA. In the initial screening, the cDNA probe used was made by primer-extension of the synthetic oligonucleotide. Positives from this first screening were rescreened by colony hybridization to 32P end-labeled synthetic oligonucleotide. Use of the primer-extended cDNA probe enabled high stringency washes, resulting in lower background, and reduced the number of clones to be reexamined by hybridization with the synthetic 17-mer. The second screen with the 17-mer was more specific, although the background was higher due to the necessary decrease in washing stringency. Seven positive clones were obtained from a screening of 20,000 clones. One of the stronger positives, C2-1, was chosen for further analysis. Plasmid DNA was isolated from C2-1 and immobilized on nitrocellulose. The mRNA hybridizing to (22-1 cDNA was isolated by hybridization select.ion. It was then translated in a cell-free translation system, immunoprecipitated with anti-SR2 antibodies, and examined by SDS-PAGE and fluorography (Fig. 63). The sea raven AFP precursor was indicated by the 20,000-dalton component (tune 2). The hybri~zation-selected mRNA gave A change in the direction of the chain indicates a reverse or @-turn. The arrow indicates the position of the predicted signal peptidase cleavage site. a translation product which migrated with the AFP precursor (lane 3). This translation product was immunoprecipitable by anti-SR2 antibodies (lane 5). These results provided additional confirmation that C2-1 was a sea raven AFP clone.
The insert in clone C2-1 (874 base pairs) was sequenced by the dideoxy chain terminating procedure ( Fig. 7, Appendix). The reading frame was established by matching with chymotryptic, tryptic, and thermo~ysin peptide sequences from SR2.3 However, there are at least four amino acid substitutions between the deduced sequence and the partial peptide sequences. The DNA sequence in (22-1 encodes a 195-residue preprotein with a molecular weight of 20,009, which is in complete agreement with the size of the biosynthetic precursor identified from the cell-free transiation studies. There are 11 base pairs in the 5'-untranslated region and 270 base pairs in the 3'-untranslated region. The highly conserved polyadenylation signal AATAAA is not found in this 3"untranslated region. The predicted secondary structure for the deduced amino acid sequence is, shown in Fig. 8. The presence of a limited amount of &structure and little a-helix is in agreement with the earlier circular dichroism studies of sea raven AFP (9). No homology was found to exist between the sea raven AFP sequence and sequences listed in the MicroGenie Nucleotide and Protein Sequence Data Bank.

DISCUSSION
We have confirmed our earlier observations (12) of the microheterogeneity in sea raven AFP both by reverse-phase HPLC and cDNA sequence analysis (e.g. the amino acid substitutions between (22-1 and SR2). As with the other AFP studied, the source of the microheterogeneity may include post-translational modification and the expression of multigene families (5,6,8,30). Furthermore, the variability of sea raven AFP may be related to a recent observation made by Fletcher et al. (31) that phenotypic variations of plasma AFP levels are present in the sea raven. In that study, they found that significant levels of AFP were present during the summer and that 4040% of the fish did not increase their AFP levels during the winter. Population variation as a cause was ruled out as sea ravens from other geographical locations also demonstrated the same result. The authors postulated that since sea ravens do not normally encounter ice-laden seawater, AFP may not be essential to their survival, and thus, considerable genetic polymorphism may have developed. Re-

Structure of Sea Raven
Antifreeze Polypeptide cent evidence from genomic Southern blots4 indicates the presence of more than 40 AFP genes in the sea raven genome. Thus, it is possible that one of the reasons for the observed individual variability may be the differential expression of such polymorphic genes.
The N terminus of the AFP precursor contains a stretch of residues typical of a signal peptide. Based on observations of signal peptidase cleavage sites in 39 proteins (32), cleavage may occur after serine at position 29 or after threonine at position 30. These positions correspond to the end of a hydrophobic stretch of residues as seen on a hydrophilicity plot of the precursor (Fig. 9). Since the N terminus of the mature AFP is blocked, further studies wi11 be needed to identify the cleavage site. However, proteolytic processing at the above positions would generate.a mature polypeptide of about 17,000 daltons, which is in complete agreement with the molecular mass for sea raven AFP as estimated by SDS-PAGE in this study. The sea raven AFP precursor is therefore similar to that of the ocean pout AFP. Both of these precursors contain only the signal sequence with no evidence of a profragment as demonstrated in flounder prepro-AFP. The presence of substitutions in at least four amino acid positions between the deduced sequence of the cDNA and the peptide sequences of the major component SR2 suggests that this cDNA encodes a minor component of sea raven AFP.
The amino acid composition of one of the minor components, SR3, resembles closely the composition from C2-1 (Table 1). Thus, C2-1 cDNA may code for SR3. In the winter flounder, sequencing of the two major components has revealed three amino acid substitutions in otherwise identical sequences (5, 33). Similarly, the sequences of two ocean pout AFP components have shown differences in at least two amino acid positions. The amino acid substitutions in these fish AFP do not appear to affect antifreeze activity (8,33).
Comparison of the amino acid composition of SR2 and of C2-1 shows that they are similar ( Table 1). The number of cysteine residues in SR2, as estimated by reduction and alkylation, agrees with the number predicted for (22-1. This is not surprising, as the number of cysteine residues is expected to be conserved despite the presence of some amino acid substitutions. Examination of the predicted secondary structure (Fig. 8) shows a high frequency of reverse turns, with the turns accounting for about 37% of the protein. Reverse turns occur in proteins with an average frequency of about 19% (34) and have been implicated with roles in receptor binding, antibody recognition, and post-translational modifications (i.e., phosphorylation, glycosy~ation) (35). Due to their intrinsically polar nature and general occurrence at the surface of proteins (35), the reverse turns in sea raven AFP may be involved in forming hydrogen bonds with water. In addition, the presence of paired hydrophilic amino acids (Asp-Asp at positions 51-52, 160-161, and 179-180) on some of these turns may be important to the AFP-ice interaction. One way of investigating the importance of these turns would be to perform fragment deletion or site-specific mutagenesis in these regions and analyze the effect on AFP activity. Such studies (including chemical modifications) to elucidate the structure-function relationship of sea raven AFP are currently underway in our laboratory.

FIG. 7.
Sequence of a sea raven AFP precursor deduced from its cDNA. Half-arrows indicate sequences indentical with SR2, whereas residues in parentheses below the sequence indicate amino acid substitutions. The arrow above the nucleotide sequence indicates the predicted signal peptidase cleavage site.  eel mntsiains 10 mM methyl mercury hydroxide (19). The tractions enriched for the 10-11 S mRNA species were pmled. ethanol-precipitated. and IVOPhitilrd.

840
Sea raven poly (A)* mRNA was translated using a mRNA-dependent rabbit reticulocyte lysate all-free translation kit. 0.5-1.5 pg of mRNA and 13SSI-L-methionine (,IO00 Cilmmol) were used with the recnmmended reaclion mnditions. in a volume of 30 pl. and incubated at 37 OC for I hr.
Translation products were immunoprecipitated with antibodies to SR2. Following translation. samples were diluted with an equal volume of water. and two volumes of phosphate buffered saline (PBS. 50 mM sodium L-methionine). 3-5 pg of SR2 was added IO some samples IO mmpete with phosphate pH 7.2. 1.81 NaU. 0.6s BSA. IS SD5. 2 1 Triton X-100. I mglml followed by an wernight incubation at 4 %. Goat anti-rabbit IgG antibody antibody. I p1 of anti-SR2 serum was added, and incubated at 37 "C fa I hr.
immunoprecipitates were pelleted and washed four times with ice-mld PBS.
(IO p11 was added, and the samples were incubated overnight at 4 %. The Prior to SDS-PAGE, they =re reduced and alkylated as described by Scheele resuspcnded in I O p1 of reducing buffer (5s SI&. 60 mM Tris pH 6.8, 10s el ai (20). with some modification. The immunoprecipitates were glycerol. 0.1 M dithiothreitol. 0 IS bromophenol blue). boiled for 3 min, and incubated at 37 OC for 30 min. 1 pl of I M iodoacetate (prepared prior lo use) was added. follwed by Incubation at room temperature for 30 min. I 111 of 0.5 N NaOH was added to readjust the buffer pH. and the samples were subieded to SDS-PAGE. The translation prod~cts were detected by fluorography using EN3HANCE.
A CDNA library was made aaxlrding to the procedure of Heidecker and Messing (2 I ). using the Plasmid vector pUC 9. The cDNA was synthesized from sea raven poly (A)' mRNA enriched for the 10-1 1 S fraction.
An amino acid sequence ol 16 residues was Obtained from a tryptic peptide of SR2. SR2-T9 (C. L. Hew el d . unpublished results). The sequence mmplemenlary sequence of a mixed synthetic 17-mer oligonucleotide.
Ala Ala Cys Cys Met Gln gave the least degeneracy and was used IO derive the Following custom synthesis. it was purdied by polyacrylamldelurea gel electrophoresis.
The cDNA clones were prepared for hybridization on nitrocellulose filters as dercribed by Maniatis el ai (22). and further treated to reduce background. a-dw lo Wmds (23). The first r c r w used a cDNA probe made by primerixtension. with sea raven poly (A)' mRNA as the template and the synthetic oligonucleotide as the primer. Synthesis of this probe was essentially as demibed by Houghton e/ d. (24). except that the probe was Positives from the first rcreening were re-screened by hybridization at 42 % annealed to the mRNA in the presence of 0.1 M KU. at 42 % for 1 hr.
with the synthetic prcbe. which was end-labelled using lr32Pl-ATFJ b3000 Cilmmol) and T4 polynuctwtide kinase as described by Woods (23). The filters were washed with excess 6X SSC (IX SSC is 0.15 M NaU. 0.015 M sodaurn citrate. pH 7) and 0 05s sodium pyrophosphate for IO min at rmm temperature. three times at 42 %for 20 min each time. and finally on-at 47 % for 10 min.
One of the positives. clone C2-I. was chaen f a further analysis by fixed onto a nitrocellulose filter (22). The mRNA axresponding to the C2-l hybridization selection. C2-l plasmid DNA was prepared. and 20 pg was cDNA was purified by hybridization selection amrding to Ricciardi PIai(2S).
with hybridization muring at 47 % for 4 hr. The mRNA isolated was translaled immediatedly in a rabbit reticulqte lysate cell-free translation system. The translation products were immunoprewpitated with anti-SR2 serum. reduced and alkylated. and subjected to SDS-PAGE and fluorography as described above.
DNA sequencing was perlamed by the didmnl-chain termination method cDNA were subcloned into the MI3 mp8 and mp9 vectors.
(26.27). Restriction fragments from both a Sau 3A and an Alu 1 disest of the v DNA and protein sequence analyses includw the hydrophilicity plot. were performed using the software program DNA Inspector I1 from Textm. West Lebanon. New Hampshire. The seomday structure was predicted l a the amino acid sequence deduced from the sea raven AFP cDNA. Tbs prediction was performed on an Apple II mmputer using the program made

6
Eiel; Determination of the number of cysteine residues in SR2. SR2 mtpinins 0 IO n acidiccarboxymethyl groups. where n is the number of cysteine residues. was subjecled to PAGE using a hiih pH dismntinuous system. SR2 was reduced. and alkylated in the presence of neutral 0.25 M iodoacetamide (lane I). acidic 0.25 M iodoacetate (lane 5). and I:l. 1.3. and 1:9 ratios of iodoacetamide to iodoacetate (lanes 2.3, and 4, respectively). Lane 6 wntains a mixture of the Samples applied in lanes 1-3. The bands are wunted on the right by the number of acidic groups.

rn op SS
'Id I SR I 9. lmmunodetection of AQP by Western blotting. Sephadex G-75purilied sea raven (SR) AFP. all live AFP from HPLC (lanet 1-51, and AFP from winter flounder (WF). m a n pout (OP). and shorthorn SDS-polyamylamide gel. and then electrophoretically transferred sculpin (ssl were electrophoresed on a 9-22.5s exponential gradient onto a nitrmllulose filter. The blotted proteins were reacted with anti-SR2 antibodies,which were then reacted wilh12SI-Protein A. The filter was autoradmgraphed overnight at -70 OC.