- Email: [email protected]
Analysis of antifreeze glycoproteins in fish serum
ANALYTICAL
BIOCHEMISTRY
139, 197-204 (1984)
Analysis of Antifreeze TIMOTHYS.BURCHAM,DAVID Department
of Food
Science
Glycoproteins
in Fish Serum’
T. OSUGA,HARUOCHINO,~ of California, Received November 3, 1983
and Technology,
University
ANDROBERTLFEENEY~ Davis,
California
95616
A procedure utilizing high-pressure &e-exclusion chromatography that permits rapid screening for both the types of components present in and the quantity of antifreeze glycoprotein in fish serum or solution is described. The applicability of the method is demonstrated by a comparative study of five different fish species, four of which contain the antifreeze glycoprotein and one which does not contain this protein. The antifreeze glycoprotein compositions of two fish of the same species, collected at different locations or under different environmental conditions, are also compared. A linear molecular-weight versus elution-volume function is established for both standard native proteins and the antifreeze glycoproteins, but these two lines do not coincide. The differences in tertiary structure between the antifreeze glycoproteins and normal proteins are presented as an explanation for the nonequivalence of calibration lines.
The antifreeze glycoproteins (AFGP)4 of polar fish are a heterogeneous group of blood proteins that allow the survival of marine fish at ocean temperatures as low as - 1.8”C. The antifreeze glycoproteins are polymers of the triglycopeptide unit Ala-Ala-(P-Gal( 1 - 3)a-GalNAc)Thr, the components differing by the number of triglycopeptide units they contain. AFGP l-5 are large-molecular-weight components ranging in molecular mass from 32 to 10 kD, respectively; AFGP 6-8 are smaller (7.8 to 2.7 kD, respectively) and contain some proline replacing alanine in the triglycopeptide unit. For a review of the properties of the antifreeze glycoproteins see Feeney et al. (1) and DeVries (2).
’ This work was supported in part by National Institutes of Health Grant GM238 17 and National Science Foundation Grants DPP78260 15 and DPPS 116963. * Present address: Biochemical Laboratory, Institute of Low Temperature Science, Hokkaido University, Sapporo 060, Japan. 3 To whom correspondence should be addressed at the Department of Food Science and Technology, 1480 Chemistry Annex, University of California, Davis, Calif. 95616. 4 Abbreviations used: AFGP, antifreeze glycoprotein; aufs, absorbance units full scale; HPSEC, high-pressure size-exclusion chromatography; TCA, trichloroacetic acid.
Despite the fact that much work has been done on the antifreeze glycoproteins, there is not at the present time a rapid and sensitive detection procedure for their analysis in serum or solution. The antifreeze glycoproteins present an unique problem in analysis because they do not contain any aromatic residues, which would allow absorbance to be read in the near uv, and they do not stain by any of the common electrophoretic stains, such as Coomassie brilliant blue or amido black, possibly because of their extreme polarity (containing -50% by weight carbohydrate). The first assay of purity was an electrophoretic procedure used by DeVries et al. (3) which combined the discontinuous buffer system of Poulik (4) and the cr-napthol/H2S04 staining procedure of Feeney et al. (5). Although this electrophoretic procedure is the standard by which all antifreeze glycoproteins are and should be compared, the method lacks sensitivity and requires the use of concentrated sulfuric acid, a somewhat hazardous reagent, in the staining procedure. We have used an updated version of the method which employs the use of 2-mm-thick polyacrylamide slab gels with IO-111wells. The newer version increases sensitivity at least lo-fold, and electrophoretic time is less than one-fourth of the 197
0003-2697184 $3.00 Copyright 0 1984 by Academic Press, Inc. All rights of reproduction in any form reserved.
198
BURCHAM
initial analysis procedure; still, the method suffers from lack of sensitivity when compared to analytical procedures for other proteins and the use of sulfuric acid in the staining procedure. Other electrophoretic staining procedures, most notably the silver-staining procedures developed for normal proteins (6,7) and those developed for glycoproteins (8), also do not stain the antifreeze glycoproteins (T. S. Burcham, D. T. Osuga, and R. E. Feeney, unpublished results), although this could be a result of AFGP not being fixed to the polyacrylamide gels in the procedures as published. Another approach is the preelectrophoretic derivatization of AFGP with fluorescent compounds. We have derivatized AFGP using dansyl chloride prior to electrophoresis, and it resulted in lower detection limits over the a-napthol/HzSO,, procedure. Fluorescamine was employed by O’Grady et al. (9); our laboratory has also used this procedure and found it superior to derivatization with dansyl chloride in the ease of sample labeling and the lack of fluorescent by-products produced, yielding a clearer separation. However, preelectrophoretic derivatization of antifreeze glycoproteins has a large effect on their mobility in an electric field (most antifreeze glycoproteins contain only the carboxyand amino-terminals’ contributions to net charge), and results obtained from derivatization must be interpreted with caution. Relative mobilities increased approximately twofold, as compared to the standard cY-napthol/ H2S04 procedure, when using the fluorescamine procedure of O’Grady et al. (9) (T. S. Burcham, D. T. Osuga, and R. E. Feeney, unpublished results). In some of the first work done in characterization of the antifreeze, Scholander and colleagues (10) observed that the antifreeze substance was present in serum samples that had been “deproteinized,” i.e., treated with TCA. More recently, Fletcher et al. (11) used TCA precipitation of serum as a step in the purification of frostfish (Microgadus tomcod) antifreeze glycoprotein. In this paper we describe a method that utilizes the TCA precip-
ET
AL.
itation as adapted from Fletcher et al. (11) coupled with high-pressure size-exclusion chromatography (HPSEC) as a general method for the analysis of native antifreeze glycoproteins in both serum and solutions of preparations. The facile method is rapid, reproducible, and sensitive to microgram quantities. The utility of the method is demonstrated by the hitherto unpublished analysis of antifreeze glycoproteins in the unfractionated serums of different polar fishes, yielding an accurate comparison of the presence and relative amounts of the various AFGP components in the fish examined. EXPERIMENTAL
PROCEDURES
Materials. The chemicals used were sodium sulfate (anhydrous powder, J. T. Baker, lot 32239), phosphoric acid (85.9%, Mallinkrodt AR, lot KLDN), sodium hydroxide (Mallinkrodt AR, lot KPBN), TCA (Mallinkrodt AR, lot KJDH), bovine serum albumin (Sigma, lot 23G8 150), lysozyme (Sigma, lot lOF-8097), and standard proteins for HPSEC molecularweight calibration (Bio-Rad). Chicken ovotransferrin and chicken ovomucoid were isolated in this laboratory. All reagents were made with water purified as follows: Housedeionized water was further treated with an IWT Research ion-exchanger Model 2 deionizer (Illinois Water Treatment Co.) and then put through a 20 X 1.6~cm column of AG50 l-X8 analytical grade mixed-bed resin (Bio-Rad, lot 12088) and finally through a Norganic cartridge and 0.45~pm filter system (Millipore). After the buffers and reagents were made with this water, they also were filtered through a 0.45~pm membrane filter and degassed for 30 min with stirring under vacuum. Serum samples were obtained from live fish. Blood was taken from an incision in the caudal artery, except for Dissostichus mawsoni. Boreogadus saida serum was collected in the Barents Sea in 1982, and Pagothenia borchgrevinki were collected through the ice at McMurdo Sound, Antarctica, in November 1982. D. mawsoni serum was from fish held in a
ANALYSIS
OF ANTIFREEZE
storage tank at McMurdo Sound. Eleginus grucilis serum was collected at three different times: One set was obtained off the coast of Kotzebue, Alaska, in March 1982; samples were also taken in early November 198 1 and in February 1982 from fish by Mombetsu, Hokkaido, Japan. Rainbow trout serum (Sulmo guirdnerii) was from fish in Putah Creek at the outlet of Monticello Dam in California. All samples represent pooled serum from many fish, although single-fish serum samples were collected and compared for B. saida and E. gracilis, and that of D. mawsoni which represented serum from a single large specimen. All serum samples were filtered with a centrifugal microfilter (5 pm; Schleicher and Schuell, Inc.), and were kept frozen at -20°C until thawed just prior to analysis. Methods. The isocratic HPSEC system consisted of an Altex Model 110A single-piston pump (Beckman Instruments) with a highsensitivity pressure filter unit attached. Samples were injected with an Altex Model 2 10 injection valve, and the sample loop used throughout the entire experiment was 50 ~1. A Helicoid pressure gauge (O-5000 psig; Rainin Instruments) was used to measure inline pressure. The entire system was plumbed with l/ 16-in.-o.d., I/ l OO-in.-i.d. stainless-steel 3 16 tubing. A Gilson Holochrome HM variable-wavelength flow-through spectrophotometer (5-~1 high-pressure quartz cell) was used for detection. Signal output from the detector was recorded on a Linear Model 285 dual-pen strip chart recorder; one pen connected to the lOO-mV variable range output and the other to the fixed range lo-mV 2.0 aufs output of the Holochrome. When quantitation was necessary, the lOO-mV variable range output was also connected to a SP4270 computing integrator (Spectra-Physics). For the analyses reported in this paper, two TSK3000SW-TSK2000SW columns (60 X 0.75 cm; Kratos) were connected in series and an SW-type TSK precolumn was in-line between the injection valve and the two column series. Columns were also tested alone and other columns, for example, a Bio-Sil
GLYCOPROTEINS
199
TSK-250 (30 X 0.75 cm; Bio-Rad), were also used. The mobile phase buffer was 0.1 M sodium sulfate/O.02 M sodium phosphate, pH 6.8. The mobile phase of Hefti (12), that is, 0.08 M sodium phosphate/O.32 M sodium chloride/20% (v/v) ethanol, pH 7.0, was also examined for its utility in this separation. Typical analysis procedures were as follows. To an aliquot of serum (50 ~1) in an Eppendorf microcentrifuge polypropylene test tube (400 ~1; 6 X 45 mm), 50 ~1 of 0.5 M TCA was added and mixed rapidly with a vortex mixer. The samples were incubated at 40°C for 5 min, and then centrifuged for 5 min using a Beckman Microfuge B. The clear supematant was injected directly onto the HPSEC system using a 250-~1 blunt-end syringe (Hamilton). Flow rates of the experiments ranged from 0.1 to 1.O ml/min (pressure range, 70 to 850 psig) and the time of the analysis varied from 10 to 1 h with the respective flow rates using the two columns connected in series, as described above. No temperature control of the column was maintained; all chromatography was done at ambient temperatures. Peaks were identified by HPSEC of individual components of AFGP, whose identity and purity were ascertained by electrophoresis by staining with (Ynapthol/H2S04 as described above. Loss of AFGP due to treatment with acid was quantitated by treating a stock solution of purified AFGP 4 from P. borchgrevinki with water (control) or with the acid solution and then subjecting it to the procedure detailed above, but using the SP4270 to record area of the AFGP peak both with and without treatment with acid. An attempt to quantitate the amount of the various antifreeze glycoprotein components in the blood of P. borchgrevinki was done as follows: using the SP4270 computing integrator, the percentage relative areas of the antifreeze glycoprotein components were determined. These percentage relative areas were multiplied by the average total yield (34.7 g/ liter serum) of all the antifreeze glycoprotein components obtained by the standard AFGP isolation procedure from P. borchgrevinki (3).
200
BLJRCHAM
ET AL.
The standard procedure described above was used only on P. borchgrevinki serum using a flow rate of 1.0 ml/min. The partial specific volume (ij) was determined for AFGP 4 in 0.1 M KC1 using a Mettler precision digital density meter, DMA 02C, with a Tronac, Inc. Model 40 precision temperature controller together with a Tamson, Inc. reservoir (Model TXV45) connected to a Lauda K-2/R. RESULTS AND DISCUSSION
:: C
Chromatograms of each species’ serum are given in Figs. l-3. The final large peak in each chromatogram corresponds to the TCA present in the sample. The loss of AFGP due to the TCA treatment was -3%; the area of the control was 97% that of the untreated sample. The HPSEC data for the standard proteins and that for the antifreeze glycoproteins were treated using the method of Himmel and Squire (13). The method was developed for an accurate determination of the molecular weights of native proteins on the TSK (type SW) HPSEC columns, the column type used
“.h”
“.“”
Fraction
of Run
Time
2. Conditions: flow rate, 0.5 ml/min; chart speed, 10 cm/h. Only the 0.2-aufs traces are shown. (A) The 0.2aufs trace of P. borchgrevinki from Fig. I is included for comparison of AFGP content between fish species. (B) HPSEC of D. mawsoni serum. (C) HPSEC of B. saida serum. (D) HPSEC of S. guirdnerii (rainbow trout) serum. FIG.
in this procedure. The method utilizes a plot of the “volume function,” F, , against the cube root of the molecular weight of the protein. The volume function is equal to F, =
Jr'/3
-
,y3
$3
_
VA13
where V, = the void volume, V, = the total volume accessible to solvent, and V, = the i 4 elution volume of the protein. Following the recommendation of Himmel and Squire (13), we used thyroglobulin (Mr 660,000) to deterFIG. I. HPSEC of P. borchgrevinki serum. Conditions: mine the void volume, VC,.The V0 of our sysflow rate, 0.1 ml/min; chart speed, 2 cm/h; detection at tem was 20.5 ml. Yl was the elution volume 220 nm. Upper trace is 0.2 aufs; lower trace is 2.0 aufs. AFGP components are numbered. of cyanocobalamin (Mr 1355; V, = 52.0 ml);
ANALYSIS
OF ANTIFREEZE
-
4
I
I
0.25 Fraction
I
0.50 of Run
0.75 Time
FIG. 3. Conditions are the same as in Fig. 2. (A) HPSEC of E. grucilis serum obtained from Alaskan fish. (B) HPSEC of E. gracilis serum obtained from Northern Japanese fish.
although Himmel and Squire (13) used sodium azide to determine V,, they observed that bradykinnin (Mr 104 1) eluted at the same volume as sodium azide. Therefore, the use of cyanocobalamin for the determination of V, should give an accurate calculation of V,. Given in Fig. 4 is the plot of F, versus Mf” for both the standard proteins and the antifreeze glycoproteins. The linear least-squares correlation coefficients were -0.94 and -0.97 for the standard proteins and the antifreeze glycoproteins, respectively. Figure 1 is the chromatogram for P. borchgrevinki resulting from a flow rate of 0.1 ml/ min and analysis time of 10 h. These conditions of analysis result in the best resolution of the antifreeze components, although the length of the run is a disadvantage. Also, this chromatogram has both the 0.2 (bold line) and the 2.0 aufs (thin line) in the figure, while the 2.0-aufs chromatogram is deleted from the
GLYCOPROTEINS
201
other fish serum runs. The eight components of the antifreeze glycoprotein are clearly distinguished and are indicated on the figure. The figure clearly shows the heterogeneity of AFGP with multiple minor peaks throughout the chromatogram. Also shown are the four components of AFGP 6; this observation separately confirms the work of O’Grady et al. (9), who found four distinct components for AFGP 6 using electrophoresis. The chromatogram in Fig. 1 shows that AFGP 4 and 5 also have multiple forms. On comparing the chromatograms from P. borchgrevinki, D. mawsoni, and B. saida (Fig. 2) it can be seen that B. saida (Fig. 2C) does not contain any of the components of AFGP l-2 and very little 3, but contains primarily AFGP 4-8. The observation that B. saida does not contain antifreeze components 1 and 2 was also observed by Osuga and Feeney ( 14) using electrophoresis. D. mawsoni (Fig. 2B) contains a much higher proportion of AFGP 3 as compared to P. borchgrevinki; AFGP 3 is the dominant high-molecular-weight polymer of D. mawsoni antifreeze. The chromatogram in Fig. 2D is that of the rainbow trout serum; this chromatogram shows that
FIG. 4. Plot of the volume function versus the cube root of the molecular weight. 0, Normal proteins; W,antifreeze glycoproteins. The large differences in the two calibration lines are taken as evidence of an expanded structure of the antifreeze glycoproteins. V. = void volume; V, = total volume accessible to solvent; V, = elution volume of the specific protein.
202
BURCHAM
there are no appreciable amounts of TCAsoluble compounds of large molecular weight present in trout serum that coincide with AFGP peaks. Therefore this method will prove useful for screening many different species of polar fish for the presence of antifreeze glycoproteins. Relative to the other polar fish, E. grads serum (Fig. 3) contains components similar in molecular weight to AFGP 2, 4, 5, 6, and 7, but these may not be directly comparable to the other fish, as this fish has an AFGP with a slightly different amino acid composition compared to the other fish species used in this study (T. S. Burcham, D. T. Osuga, H. Chino, Y. Yeh, and R. E. Feeney, submitted for publication). The chromatograms resulting from the E. gracilis obtained from the two locations, Alaska (Fig. 3A) and Japan (Fig. 3B), were essentially identical in the type and amount of antifreeze components except for a small amount of large-molecular-weight peaks in the Japanese E. gracilis. The E. gracilis serum collected in March, however, had a fivefold higher concentration of the AFGP than that collected in November (data not shown). The seasonal variation of AFGP concentration in E. gracilis has never been shown; the method described in this paper will allow rapid analysis of serum to detect and quantitate seasonal variations in AFGP concentration for many other fish species. The resolution of the various AFGP components improved as the flow rate decreased, a result similar to that of Hefti ( 12). The triplet group of AFGP 6, as seen in the chromatogram of P. borchgrevinki done at 0.1 ml/min (Fig. 2A), becomes fused into one broad peak at 0.5 ml/min, as seen in Fig. 2B in the chromatogram of D. mawsoni. Although the resolution is better at the lower flow rates, adequate resolution occurs at the higher flow rates for most applications, allowing shorter analysis time. The resolution also decreased when using the columns separately, or when using the Bio-Sil TSK-250 column; the latter column did not resolve any of the minor components of AFGP 4, 5, or 6.
ET
AL.
The amounts of each component of AFGP, as determined using the method described above, are given in Table 1. When multiple components existed for an AFGP species, these components were grouped and treated mathematically as a homogenous component, e.g., the areas of the AFGP 6 peaks were summed, and the published molecular weight as given in Feeney et al. ( 1) was applied to the total summed area of those peaks labeled as AFGP 6. AFGP 7 and 8 were also grouped together as one peak, as at the high flow rate used ( 1.O ml/min) the peaks were not resolved from one another. Therefore, the values are an estimate of the actual concentration of the antifreeze glycoproteins in the blood of P. borchgrevinki. When the elution volumes of the AFGP components are compared to those of the proteins contained in the HPSEC calibration test mix (thyroglobulin, IgG, ovalbumin, and myoglobin), bovine serum albumin, ovotransfenin, and lysozyme, it was found that the antifreeze components have a much lower elution volume than expected from their published molecular weights; that is, they have an apparent molecular weight that is much larger than their known molecular weights (see Fig. 4). For example, AFGP 1, with a M, of 32,000, has an equivalent elution volume similar to that of a protein with a molecular weight five TABLE CONCENTRATION
1
OF ANTIFREEZE
GLYCOPROTEIN
IN
Pugothenia borchgreviinki SERUM a AFGP component 1 2 3 4 5 6 7+8 Total “See text concentration.
Area (X lo-‘) 1.6 1.0 8.0 20.6 21.6 18.1 435.1 506.0 for methods
Relative area (“lo)
Approx amount in serum (g/l)
0.3 0.2 1.6 4.1 4.3 3.6 85.9 100.0 used to determine
0.1 0.1 0.6 1.4 1.5 1.2 29.8 34.7 the serum
ANALYSIS
OF
ANTIFREEZE
times its actual value. This general pattern exists for all the antifreeze glycoprotein components. Another glycoprotein containing -25% carbohydrate, chicken ovomucoid, exhibits a twofold increase in the apparent molecular weight using the HPSEC described. This observation confirms the results of Fletcher et al. (1 l), who, although using a different buffer system, observed that the AFGP from Antarctic fish and from the Arctic Microgadus tomcod eluted at a much lower volume than expected from their molecular weights. Even with the use of the buffer system of Hefti (12), who claimed that the buffer allows for accurate comparison of molecular weights using HPSEC and that the glycosyl portion does not alter the elution behavior of the proteins, the antifreeze glycoproteins still eluted at a much lower volume than expected. (It should be noted that Hefti’s observations were based on glycoproteins containing < 16% carbohydrate; AFGP, on the contrary, contains - 50% by weight carbohydrate.) The importance of using proteins with similar charge and hydrodynamic shape when determining molecular weights using any gel-exclusion technique is well established. Accordingly, there could be two reasons for the great difference in elution volumes of standard proteins listed above and that of the antifreeze glycoproteins: First, AFGP could be retained in the column due to electrostatic interactions with the column matrix. The second possibility is that the hydrated radius of AFGP is actually comparable to a globular nonglycoprotein of fivefold molecular weight. These two possibilities will be examined in turn below. Electrostatic interactions are not likely in the case of the antifreeze because of the high ionic strength of the elution buffer. The ionic strength of the 0.1 M sodium sulfate, 0.02 M sodium phosphate buffer (pH 6.8) is -0.33. Kopaciewiez and Regnier (15) and Regnier (16) have shown that electrostatic effects are significant only below ionic strengths of 0.1. Also, the pH of the mobile phase (6.8) is near the pZ of all the proteins with the exception
203
GLYCOPROTEINS
of some components of E. gracilis antifreeze; the only charged residues of the antifreeze are the N- and C-terminals of the proteins, and the calculated pZ of these proteins is near pH 6.02. On the contrary, this effect can be explained as the result of the hydrodynamic shape of the antifreeze in solution. Using the s20,wand the Ozo determined by Ahmed et al. (17), AFGP 4 was calculated to have a f/h value of 2.1. [Note, for this calculation, a value for V of 0.7089, as calculated from density measurements of this laboratory, rather than a value of 0.7463, as calculated from composition ( 17), was used.] Assuming the antifreeze exists as a prolate ellipsoid in solution, these f/f0 values correspond to an axial ratio, a/b, of -25 (18). While we do not consider the antifreeze to be a prolate ellipsoid in solution, the point remains that the antifreeze has a much larger hydrodynamic radius than what would be expected from its molecular weight alone. By more quantitative methods of conformational analysis, the antifreeze glycoproteins may indeed exist as an extended molecule. Both early viscometric (3) and more recent vacuum uv CD measurements (19) indicate that AFGP exists in an extended conformation, as an extended random coil (3), or as a threefold left-handed helix of the collagen type (19). CONCLUSION
The HPSEC method is a sensitive and rapid analytical procedure for the determination of antifreeze glycoproteins in the serum of fish. The method can also be used for the determination of antifreeze glycoproteins in other mixtures, such as in an enzymatic digest, or the distribution of AFGP components in a purification scheme. (When the amount and concentration in purified AFGP solutions are determined, the TCA precipitation step may, of course, be omitted.) The data given in this paper clearly illustrate the differences in amounts and component types of AFGP in the fish species examined, which should be of
204
BURCHAM
use for determining changes in the amounts of antifreeze glycoproteins with the environment and for establishing phylogenetic relationships between polar fish species. ACKNOWLEDGMENTS We thank the Norwegian Institute of Marine Research, Bergen, and the Department of Biochemistry, University of Bergen, Norway, and especially the captain, crew, and the scientists of the ship, the “G.O.Sars,” for their haspitality and help to T.S.B. during the 1982 Barents Sea capelin survey and the collection of B. saida serum; the Institute of Low Temperature Science., Hokkaido University, Sapporo, Japan, and especially Dr. K. Tanno, Dr. M. Hoshi, and Dr. T. Sato for their assistance to R.E.F. and H.C. in collecting serum of E. gracilis at Mombetsu, Hokkaido Island, Japan in 1981 and 1982; M. H. Penner and R. Steiner for collection of E. grucilis serum at Kotzebue, Alaska, in 1982; M. Knauf and Y. Yeh for collecting serum of P. borchgrevinki in Antarctica in 1982. Also, sincere thanks to Dr. A. L. DeVries for help in obtaining the D. mawsoni serum in Antarctica and to M. Knauf for his discussions and technical help. Appreciation is expressed to C. Howland for her expert editorial advice, and to Diana Melboum for typing the manuscript. Note grevinki tomus.
in prooj The species Pagothenia borch(Boulanger) has also been described as Trema-
added
REFERENCES 1. Feeney, R. E., Osuga, D. T., and Yeh, Y. (1979) in Proteins at Low Temperatures (Fennema, O., ed.), Advances in Chemistry Series, Vol. 180, pp. 83107, Amer. Chem. Sot., Washington, D. C.
ET AL. 2.
3. 4. 5. 6. 7.
DeVries, A. L. (1983) Annu. Rev. Physiol. 45, 245260. DcVries, A. L., Komatsu, S. K., and Feeney, R. E. (1970) J. Biol. Chem. 245, 2901-2908. Poulik, M. D. (1957) Nature (London), 180, 14771479. Feeney, R. E., Osuga, D. T., and Ma&a, H. (1967) Arch. Biochem. Biophys. 119, 124-132. Menil, C. R., Switzer, R. C., and VanKeuren, M. L. (1979) Proc. Natl. Acad. Sci. USA 76,4335-4339. Oakley, B. R., Kirsch, D. R., and Morris, N. R. (1980) Anal.
Biochem.
105, 361-363.
8. Dubray, G., and Bezard, G. (1982) Anal. Biochem. 119, 325-329.
9. G’Grady, S. M., Schrag, J. D., Raymond, J. A., and DeVries,A. L. (1982) J. Exp. Zool. 224, 177-185. 10. Gordon, M. S., Amdur, B. H., and Scholander, P. F. (1962) Biol. Bull. 122, 52-62. 1 I. Fletcher, G. L., Hew, C. L., and Joshi, S. B. (1982) Canad.
J. Zool.
60, 348-355.
12. Hefti, F. (1982) Anal. Biochem. 121,378-381. 13. Himmel, M. E., and Squire, P. G. (198 1) Znt. J. Peptide Protein
Res. 17, 365-373.
14. Osuga, D. T., and Feeney, R. E. ( 1978) J. Biol. Chem. 253, 5338-5343. 15. Kopaciewiez, B., and Regnier, F. E. ( 198 I) lnt. Symp. HPLC Proteins Peptides, Paper No. 127, Washington, D. C. 16. Regnier, F. E. (1983) in Methods in Enzymology (Hits, C. H. W., and Timasheff, S. N., eds.), Vol. 91, pp. 137-190, Academic Press, New York. 17. Ahmed, A. I., Yeh, Y., Osuga, D. T., and Feeney, R. E. (1976) J. Biol. Chem. 251, 3033-3036. 18. Cantor, C. R., and Schimmel, P. R. (1980) Biophysical Chemistry, Parts II and Ill, Freeman, San Francisco. 19. Bush, C. A., Feeney, R. E., Gsuga, D. T., Ralapati, S., and Yeh, Y. (1981) Int. J. Peptide Protein Res. 17, 125-129.
BIOCHEMISTRY
139, 197-204 (1984)
Analysis of Antifreeze TIMOTHYS.BURCHAM,DAVID Department
of Food
Science
Glycoproteins
in Fish Serum’
T. OSUGA,HARUOCHINO,~ of California, Received November 3, 1983
and Technology,
University
ANDROBERTLFEENEY~ Davis,
California
95616
A procedure utilizing high-pressure &e-exclusion chromatography that permits rapid screening for both the types of components present in and the quantity of antifreeze glycoprotein in fish serum or solution is described. The applicability of the method is demonstrated by a comparative study of five different fish species, four of which contain the antifreeze glycoprotein and one which does not contain this protein. The antifreeze glycoprotein compositions of two fish of the same species, collected at different locations or under different environmental conditions, are also compared. A linear molecular-weight versus elution-volume function is established for both standard native proteins and the antifreeze glycoproteins, but these two lines do not coincide. The differences in tertiary structure between the antifreeze glycoproteins and normal proteins are presented as an explanation for the nonequivalence of calibration lines.
The antifreeze glycoproteins (AFGP)4 of polar fish are a heterogeneous group of blood proteins that allow the survival of marine fish at ocean temperatures as low as - 1.8”C. The antifreeze glycoproteins are polymers of the triglycopeptide unit Ala-Ala-(P-Gal( 1 - 3)a-GalNAc)Thr, the components differing by the number of triglycopeptide units they contain. AFGP l-5 are large-molecular-weight components ranging in molecular mass from 32 to 10 kD, respectively; AFGP 6-8 are smaller (7.8 to 2.7 kD, respectively) and contain some proline replacing alanine in the triglycopeptide unit. For a review of the properties of the antifreeze glycoproteins see Feeney et al. (1) and DeVries (2).
’ This work was supported in part by National Institutes of Health Grant GM238 17 and National Science Foundation Grants DPP78260 15 and DPPS 116963. * Present address: Biochemical Laboratory, Institute of Low Temperature Science, Hokkaido University, Sapporo 060, Japan. 3 To whom correspondence should be addressed at the Department of Food Science and Technology, 1480 Chemistry Annex, University of California, Davis, Calif. 95616. 4 Abbreviations used: AFGP, antifreeze glycoprotein; aufs, absorbance units full scale; HPSEC, high-pressure size-exclusion chromatography; TCA, trichloroacetic acid.
Despite the fact that much work has been done on the antifreeze glycoproteins, there is not at the present time a rapid and sensitive detection procedure for their analysis in serum or solution. The antifreeze glycoproteins present an unique problem in analysis because they do not contain any aromatic residues, which would allow absorbance to be read in the near uv, and they do not stain by any of the common electrophoretic stains, such as Coomassie brilliant blue or amido black, possibly because of their extreme polarity (containing -50% by weight carbohydrate). The first assay of purity was an electrophoretic procedure used by DeVries et al. (3) which combined the discontinuous buffer system of Poulik (4) and the cr-napthol/H2S04 staining procedure of Feeney et al. (5). Although this electrophoretic procedure is the standard by which all antifreeze glycoproteins are and should be compared, the method lacks sensitivity and requires the use of concentrated sulfuric acid, a somewhat hazardous reagent, in the staining procedure. We have used an updated version of the method which employs the use of 2-mm-thick polyacrylamide slab gels with IO-111wells. The newer version increases sensitivity at least lo-fold, and electrophoretic time is less than one-fourth of the 197
0003-2697184 $3.00 Copyright 0 1984 by Academic Press, Inc. All rights of reproduction in any form reserved.
198
BURCHAM
initial analysis procedure; still, the method suffers from lack of sensitivity when compared to analytical procedures for other proteins and the use of sulfuric acid in the staining procedure. Other electrophoretic staining procedures, most notably the silver-staining procedures developed for normal proteins (6,7) and those developed for glycoproteins (8), also do not stain the antifreeze glycoproteins (T. S. Burcham, D. T. Osuga, and R. E. Feeney, unpublished results), although this could be a result of AFGP not being fixed to the polyacrylamide gels in the procedures as published. Another approach is the preelectrophoretic derivatization of AFGP with fluorescent compounds. We have derivatized AFGP using dansyl chloride prior to electrophoresis, and it resulted in lower detection limits over the a-napthol/HzSO,, procedure. Fluorescamine was employed by O’Grady et al. (9); our laboratory has also used this procedure and found it superior to derivatization with dansyl chloride in the ease of sample labeling and the lack of fluorescent by-products produced, yielding a clearer separation. However, preelectrophoretic derivatization of antifreeze glycoproteins has a large effect on their mobility in an electric field (most antifreeze glycoproteins contain only the carboxyand amino-terminals’ contributions to net charge), and results obtained from derivatization must be interpreted with caution. Relative mobilities increased approximately twofold, as compared to the standard cY-napthol/ H2S04 procedure, when using the fluorescamine procedure of O’Grady et al. (9) (T. S. Burcham, D. T. Osuga, and R. E. Feeney, unpublished results). In some of the first work done in characterization of the antifreeze, Scholander and colleagues (10) observed that the antifreeze substance was present in serum samples that had been “deproteinized,” i.e., treated with TCA. More recently, Fletcher et al. (11) used TCA precipitation of serum as a step in the purification of frostfish (Microgadus tomcod) antifreeze glycoprotein. In this paper we describe a method that utilizes the TCA precip-
ET
AL.
itation as adapted from Fletcher et al. (11) coupled with high-pressure size-exclusion chromatography (HPSEC) as a general method for the analysis of native antifreeze glycoproteins in both serum and solutions of preparations. The facile method is rapid, reproducible, and sensitive to microgram quantities. The utility of the method is demonstrated by the hitherto unpublished analysis of antifreeze glycoproteins in the unfractionated serums of different polar fishes, yielding an accurate comparison of the presence and relative amounts of the various AFGP components in the fish examined. EXPERIMENTAL
PROCEDURES
Materials. The chemicals used were sodium sulfate (anhydrous powder, J. T. Baker, lot 32239), phosphoric acid (85.9%, Mallinkrodt AR, lot KLDN), sodium hydroxide (Mallinkrodt AR, lot KPBN), TCA (Mallinkrodt AR, lot KJDH), bovine serum albumin (Sigma, lot 23G8 150), lysozyme (Sigma, lot lOF-8097), and standard proteins for HPSEC molecularweight calibration (Bio-Rad). Chicken ovotransferrin and chicken ovomucoid were isolated in this laboratory. All reagents were made with water purified as follows: Housedeionized water was further treated with an IWT Research ion-exchanger Model 2 deionizer (Illinois Water Treatment Co.) and then put through a 20 X 1.6~cm column of AG50 l-X8 analytical grade mixed-bed resin (Bio-Rad, lot 12088) and finally through a Norganic cartridge and 0.45~pm filter system (Millipore). After the buffers and reagents were made with this water, they also were filtered through a 0.45~pm membrane filter and degassed for 30 min with stirring under vacuum. Serum samples were obtained from live fish. Blood was taken from an incision in the caudal artery, except for Dissostichus mawsoni. Boreogadus saida serum was collected in the Barents Sea in 1982, and Pagothenia borchgrevinki were collected through the ice at McMurdo Sound, Antarctica, in November 1982. D. mawsoni serum was from fish held in a
ANALYSIS
OF ANTIFREEZE
storage tank at McMurdo Sound. Eleginus grucilis serum was collected at three different times: One set was obtained off the coast of Kotzebue, Alaska, in March 1982; samples were also taken in early November 198 1 and in February 1982 from fish by Mombetsu, Hokkaido, Japan. Rainbow trout serum (Sulmo guirdnerii) was from fish in Putah Creek at the outlet of Monticello Dam in California. All samples represent pooled serum from many fish, although single-fish serum samples were collected and compared for B. saida and E. gracilis, and that of D. mawsoni which represented serum from a single large specimen. All serum samples were filtered with a centrifugal microfilter (5 pm; Schleicher and Schuell, Inc.), and were kept frozen at -20°C until thawed just prior to analysis. Methods. The isocratic HPSEC system consisted of an Altex Model 110A single-piston pump (Beckman Instruments) with a highsensitivity pressure filter unit attached. Samples were injected with an Altex Model 2 10 injection valve, and the sample loop used throughout the entire experiment was 50 ~1. A Helicoid pressure gauge (O-5000 psig; Rainin Instruments) was used to measure inline pressure. The entire system was plumbed with l/ 16-in.-o.d., I/ l OO-in.-i.d. stainless-steel 3 16 tubing. A Gilson Holochrome HM variable-wavelength flow-through spectrophotometer (5-~1 high-pressure quartz cell) was used for detection. Signal output from the detector was recorded on a Linear Model 285 dual-pen strip chart recorder; one pen connected to the lOO-mV variable range output and the other to the fixed range lo-mV 2.0 aufs output of the Holochrome. When quantitation was necessary, the lOO-mV variable range output was also connected to a SP4270 computing integrator (Spectra-Physics). For the analyses reported in this paper, two TSK3000SW-TSK2000SW columns (60 X 0.75 cm; Kratos) were connected in series and an SW-type TSK precolumn was in-line between the injection valve and the two column series. Columns were also tested alone and other columns, for example, a Bio-Sil
GLYCOPROTEINS
199
TSK-250 (30 X 0.75 cm; Bio-Rad), were also used. The mobile phase buffer was 0.1 M sodium sulfate/O.02 M sodium phosphate, pH 6.8. The mobile phase of Hefti (12), that is, 0.08 M sodium phosphate/O.32 M sodium chloride/20% (v/v) ethanol, pH 7.0, was also examined for its utility in this separation. Typical analysis procedures were as follows. To an aliquot of serum (50 ~1) in an Eppendorf microcentrifuge polypropylene test tube (400 ~1; 6 X 45 mm), 50 ~1 of 0.5 M TCA was added and mixed rapidly with a vortex mixer. The samples were incubated at 40°C for 5 min, and then centrifuged for 5 min using a Beckman Microfuge B. The clear supematant was injected directly onto the HPSEC system using a 250-~1 blunt-end syringe (Hamilton). Flow rates of the experiments ranged from 0.1 to 1.O ml/min (pressure range, 70 to 850 psig) and the time of the analysis varied from 10 to 1 h with the respective flow rates using the two columns connected in series, as described above. No temperature control of the column was maintained; all chromatography was done at ambient temperatures. Peaks were identified by HPSEC of individual components of AFGP, whose identity and purity were ascertained by electrophoresis by staining with (Ynapthol/H2S04 as described above. Loss of AFGP due to treatment with acid was quantitated by treating a stock solution of purified AFGP 4 from P. borchgrevinki with water (control) or with the acid solution and then subjecting it to the procedure detailed above, but using the SP4270 to record area of the AFGP peak both with and without treatment with acid. An attempt to quantitate the amount of the various antifreeze glycoprotein components in the blood of P. borchgrevinki was done as follows: using the SP4270 computing integrator, the percentage relative areas of the antifreeze glycoprotein components were determined. These percentage relative areas were multiplied by the average total yield (34.7 g/ liter serum) of all the antifreeze glycoprotein components obtained by the standard AFGP isolation procedure from P. borchgrevinki (3).
200
BLJRCHAM
ET AL.
The standard procedure described above was used only on P. borchgrevinki serum using a flow rate of 1.0 ml/min. The partial specific volume (ij) was determined for AFGP 4 in 0.1 M KC1 using a Mettler precision digital density meter, DMA 02C, with a Tronac, Inc. Model 40 precision temperature controller together with a Tamson, Inc. reservoir (Model TXV45) connected to a Lauda K-2/R. RESULTS AND DISCUSSION
:: C
Chromatograms of each species’ serum are given in Figs. l-3. The final large peak in each chromatogram corresponds to the TCA present in the sample. The loss of AFGP due to the TCA treatment was -3%; the area of the control was 97% that of the untreated sample. The HPSEC data for the standard proteins and that for the antifreeze glycoproteins were treated using the method of Himmel and Squire (13). The method was developed for an accurate determination of the molecular weights of native proteins on the TSK (type SW) HPSEC columns, the column type used
“.h”
“.“”
Fraction
of Run
Time
2. Conditions: flow rate, 0.5 ml/min; chart speed, 10 cm/h. Only the 0.2-aufs traces are shown. (A) The 0.2aufs trace of P. borchgrevinki from Fig. I is included for comparison of AFGP content between fish species. (B) HPSEC of D. mawsoni serum. (C) HPSEC of B. saida serum. (D) HPSEC of S. guirdnerii (rainbow trout) serum. FIG.
in this procedure. The method utilizes a plot of the “volume function,” F, , against the cube root of the molecular weight of the protein. The volume function is equal to F, =
Jr'/3
-
,y3
$3
_
VA13
where V, = the void volume, V, = the total volume accessible to solvent, and V, = the i 4 elution volume of the protein. Following the recommendation of Himmel and Squire (13), we used thyroglobulin (Mr 660,000) to deterFIG. I. HPSEC of P. borchgrevinki serum. Conditions: mine the void volume, VC,.The V0 of our sysflow rate, 0.1 ml/min; chart speed, 2 cm/h; detection at tem was 20.5 ml. Yl was the elution volume 220 nm. Upper trace is 0.2 aufs; lower trace is 2.0 aufs. AFGP components are numbered. of cyanocobalamin (Mr 1355; V, = 52.0 ml);
ANALYSIS
OF ANTIFREEZE
-
4
I
I
0.25 Fraction
I
0.50 of Run
0.75 Time
FIG. 3. Conditions are the same as in Fig. 2. (A) HPSEC of E. grucilis serum obtained from Alaskan fish. (B) HPSEC of E. gracilis serum obtained from Northern Japanese fish.
although Himmel and Squire (13) used sodium azide to determine V,, they observed that bradykinnin (Mr 104 1) eluted at the same volume as sodium azide. Therefore, the use of cyanocobalamin for the determination of V, should give an accurate calculation of V,. Given in Fig. 4 is the plot of F, versus Mf” for both the standard proteins and the antifreeze glycoproteins. The linear least-squares correlation coefficients were -0.94 and -0.97 for the standard proteins and the antifreeze glycoproteins, respectively. Figure 1 is the chromatogram for P. borchgrevinki resulting from a flow rate of 0.1 ml/ min and analysis time of 10 h. These conditions of analysis result in the best resolution of the antifreeze components, although the length of the run is a disadvantage. Also, this chromatogram has both the 0.2 (bold line) and the 2.0 aufs (thin line) in the figure, while the 2.0-aufs chromatogram is deleted from the
GLYCOPROTEINS
201
other fish serum runs. The eight components of the antifreeze glycoprotein are clearly distinguished and are indicated on the figure. The figure clearly shows the heterogeneity of AFGP with multiple minor peaks throughout the chromatogram. Also shown are the four components of AFGP 6; this observation separately confirms the work of O’Grady et al. (9), who found four distinct components for AFGP 6 using electrophoresis. The chromatogram in Fig. 1 shows that AFGP 4 and 5 also have multiple forms. On comparing the chromatograms from P. borchgrevinki, D. mawsoni, and B. saida (Fig. 2) it can be seen that B. saida (Fig. 2C) does not contain any of the components of AFGP l-2 and very little 3, but contains primarily AFGP 4-8. The observation that B. saida does not contain antifreeze components 1 and 2 was also observed by Osuga and Feeney ( 14) using electrophoresis. D. mawsoni (Fig. 2B) contains a much higher proportion of AFGP 3 as compared to P. borchgrevinki; AFGP 3 is the dominant high-molecular-weight polymer of D. mawsoni antifreeze. The chromatogram in Fig. 2D is that of the rainbow trout serum; this chromatogram shows that
FIG. 4. Plot of the volume function versus the cube root of the molecular weight. 0, Normal proteins; W,antifreeze glycoproteins. The large differences in the two calibration lines are taken as evidence of an expanded structure of the antifreeze glycoproteins. V. = void volume; V, = total volume accessible to solvent; V, = elution volume of the specific protein.
202
BURCHAM
there are no appreciable amounts of TCAsoluble compounds of large molecular weight present in trout serum that coincide with AFGP peaks. Therefore this method will prove useful for screening many different species of polar fish for the presence of antifreeze glycoproteins. Relative to the other polar fish, E. grads serum (Fig. 3) contains components similar in molecular weight to AFGP 2, 4, 5, 6, and 7, but these may not be directly comparable to the other fish, as this fish has an AFGP with a slightly different amino acid composition compared to the other fish species used in this study (T. S. Burcham, D. T. Osuga, H. Chino, Y. Yeh, and R. E. Feeney, submitted for publication). The chromatograms resulting from the E. gracilis obtained from the two locations, Alaska (Fig. 3A) and Japan (Fig. 3B), were essentially identical in the type and amount of antifreeze components except for a small amount of large-molecular-weight peaks in the Japanese E. gracilis. The E. gracilis serum collected in March, however, had a fivefold higher concentration of the AFGP than that collected in November (data not shown). The seasonal variation of AFGP concentration in E. gracilis has never been shown; the method described in this paper will allow rapid analysis of serum to detect and quantitate seasonal variations in AFGP concentration for many other fish species. The resolution of the various AFGP components improved as the flow rate decreased, a result similar to that of Hefti ( 12). The triplet group of AFGP 6, as seen in the chromatogram of P. borchgrevinki done at 0.1 ml/min (Fig. 2A), becomes fused into one broad peak at 0.5 ml/min, as seen in Fig. 2B in the chromatogram of D. mawsoni. Although the resolution is better at the lower flow rates, adequate resolution occurs at the higher flow rates for most applications, allowing shorter analysis time. The resolution also decreased when using the columns separately, or when using the Bio-Sil TSK-250 column; the latter column did not resolve any of the minor components of AFGP 4, 5, or 6.
ET
AL.
The amounts of each component of AFGP, as determined using the method described above, are given in Table 1. When multiple components existed for an AFGP species, these components were grouped and treated mathematically as a homogenous component, e.g., the areas of the AFGP 6 peaks were summed, and the published molecular weight as given in Feeney et al. ( 1) was applied to the total summed area of those peaks labeled as AFGP 6. AFGP 7 and 8 were also grouped together as one peak, as at the high flow rate used ( 1.O ml/min) the peaks were not resolved from one another. Therefore, the values are an estimate of the actual concentration of the antifreeze glycoproteins in the blood of P. borchgrevinki. When the elution volumes of the AFGP components are compared to those of the proteins contained in the HPSEC calibration test mix (thyroglobulin, IgG, ovalbumin, and myoglobin), bovine serum albumin, ovotransfenin, and lysozyme, it was found that the antifreeze components have a much lower elution volume than expected from their published molecular weights; that is, they have an apparent molecular weight that is much larger than their known molecular weights (see Fig. 4). For example, AFGP 1, with a M, of 32,000, has an equivalent elution volume similar to that of a protein with a molecular weight five TABLE CONCENTRATION
1
OF ANTIFREEZE
GLYCOPROTEIN
IN
Pugothenia borchgreviinki SERUM a AFGP component 1 2 3 4 5 6 7+8 Total “See text concentration.
Area (X lo-‘) 1.6 1.0 8.0 20.6 21.6 18.1 435.1 506.0 for methods
Relative area (“lo)
Approx amount in serum (g/l)
0.3 0.2 1.6 4.1 4.3 3.6 85.9 100.0 used to determine
0.1 0.1 0.6 1.4 1.5 1.2 29.8 34.7 the serum
ANALYSIS
OF
ANTIFREEZE
times its actual value. This general pattern exists for all the antifreeze glycoprotein components. Another glycoprotein containing -25% carbohydrate, chicken ovomucoid, exhibits a twofold increase in the apparent molecular weight using the HPSEC described. This observation confirms the results of Fletcher et al. (1 l), who, although using a different buffer system, observed that the AFGP from Antarctic fish and from the Arctic Microgadus tomcod eluted at a much lower volume than expected from their molecular weights. Even with the use of the buffer system of Hefti (12), who claimed that the buffer allows for accurate comparison of molecular weights using HPSEC and that the glycosyl portion does not alter the elution behavior of the proteins, the antifreeze glycoproteins still eluted at a much lower volume than expected. (It should be noted that Hefti’s observations were based on glycoproteins containing < 16% carbohydrate; AFGP, on the contrary, contains - 50% by weight carbohydrate.) The importance of using proteins with similar charge and hydrodynamic shape when determining molecular weights using any gel-exclusion technique is well established. Accordingly, there could be two reasons for the great difference in elution volumes of standard proteins listed above and that of the antifreeze glycoproteins: First, AFGP could be retained in the column due to electrostatic interactions with the column matrix. The second possibility is that the hydrated radius of AFGP is actually comparable to a globular nonglycoprotein of fivefold molecular weight. These two possibilities will be examined in turn below. Electrostatic interactions are not likely in the case of the antifreeze because of the high ionic strength of the elution buffer. The ionic strength of the 0.1 M sodium sulfate, 0.02 M sodium phosphate buffer (pH 6.8) is -0.33. Kopaciewiez and Regnier (15) and Regnier (16) have shown that electrostatic effects are significant only below ionic strengths of 0.1. Also, the pH of the mobile phase (6.8) is near the pZ of all the proteins with the exception
203
GLYCOPROTEINS
of some components of E. gracilis antifreeze; the only charged residues of the antifreeze are the N- and C-terminals of the proteins, and the calculated pZ of these proteins is near pH 6.02. On the contrary, this effect can be explained as the result of the hydrodynamic shape of the antifreeze in solution. Using the s20,wand the Ozo determined by Ahmed et al. (17), AFGP 4 was calculated to have a f/h value of 2.1. [Note, for this calculation, a value for V of 0.7089, as calculated from density measurements of this laboratory, rather than a value of 0.7463, as calculated from composition ( 17), was used.] Assuming the antifreeze exists as a prolate ellipsoid in solution, these f/f0 values correspond to an axial ratio, a/b, of -25 (18). While we do not consider the antifreeze to be a prolate ellipsoid in solution, the point remains that the antifreeze has a much larger hydrodynamic radius than what would be expected from its molecular weight alone. By more quantitative methods of conformational analysis, the antifreeze glycoproteins may indeed exist as an extended molecule. Both early viscometric (3) and more recent vacuum uv CD measurements (19) indicate that AFGP exists in an extended conformation, as an extended random coil (3), or as a threefold left-handed helix of the collagen type (19). CONCLUSION
The HPSEC method is a sensitive and rapid analytical procedure for the determination of antifreeze glycoproteins in the serum of fish. The method can also be used for the determination of antifreeze glycoproteins in other mixtures, such as in an enzymatic digest, or the distribution of AFGP components in a purification scheme. (When the amount and concentration in purified AFGP solutions are determined, the TCA precipitation step may, of course, be omitted.) The data given in this paper clearly illustrate the differences in amounts and component types of AFGP in the fish species examined, which should be of
204
BURCHAM
use for determining changes in the amounts of antifreeze glycoproteins with the environment and for establishing phylogenetic relationships between polar fish species. ACKNOWLEDGMENTS We thank the Norwegian Institute of Marine Research, Bergen, and the Department of Biochemistry, University of Bergen, Norway, and especially the captain, crew, and the scientists of the ship, the “G.O.Sars,” for their haspitality and help to T.S.B. during the 1982 Barents Sea capelin survey and the collection of B. saida serum; the Institute of Low Temperature Science., Hokkaido University, Sapporo, Japan, and especially Dr. K. Tanno, Dr. M. Hoshi, and Dr. T. Sato for their assistance to R.E.F. and H.C. in collecting serum of E. gracilis at Mombetsu, Hokkaido Island, Japan in 1981 and 1982; M. H. Penner and R. Steiner for collection of E. grucilis serum at Kotzebue, Alaska, in 1982; M. Knauf and Y. Yeh for collecting serum of P. borchgrevinki in Antarctica in 1982. Also, sincere thanks to Dr. A. L. DeVries for help in obtaining the D. mawsoni serum in Antarctica and to M. Knauf for his discussions and technical help. Appreciation is expressed to C. Howland for her expert editorial advice, and to Diana Melboum for typing the manuscript. Note grevinki tomus.
in prooj The species Pagothenia borch(Boulanger) has also been described as Trema-
added
REFERENCES 1. Feeney, R. E., Osuga, D. T., and Yeh, Y. (1979) in Proteins at Low Temperatures (Fennema, O., ed.), Advances in Chemistry Series, Vol. 180, pp. 83107, Amer. Chem. Sot., Washington, D. C.
ET AL. 2.
3. 4. 5. 6. 7.
DeVries, A. L. (1983) Annu. Rev. Physiol. 45, 245260. DcVries, A. L., Komatsu, S. K., and Feeney, R. E. (1970) J. Biol. Chem. 245, 2901-2908. Poulik, M. D. (1957) Nature (London), 180, 14771479. Feeney, R. E., Osuga, D. T., and Ma&a, H. (1967) Arch. Biochem. Biophys. 119, 124-132. Menil, C. R., Switzer, R. C., and VanKeuren, M. L. (1979) Proc. Natl. Acad. Sci. USA 76,4335-4339. Oakley, B. R., Kirsch, D. R., and Morris, N. R. (1980) Anal.
Biochem.
105, 361-363.
8. Dubray, G., and Bezard, G. (1982) Anal. Biochem. 119, 325-329.
9. G’Grady, S. M., Schrag, J. D., Raymond, J. A., and DeVries,A. L. (1982) J. Exp. Zool. 224, 177-185. 10. Gordon, M. S., Amdur, B. H., and Scholander, P. F. (1962) Biol. Bull. 122, 52-62. 1 I. Fletcher, G. L., Hew, C. L., and Joshi, S. B. (1982) Canad.
J. Zool.
60, 348-355.
12. Hefti, F. (1982) Anal. Biochem. 121,378-381. 13. Himmel, M. E., and Squire, P. G. (198 1) Znt. J. Peptide Protein
Res. 17, 365-373.
14. Osuga, D. T., and Feeney, R. E. ( 1978) J. Biol. Chem. 253, 5338-5343. 15. Kopaciewiez, B., and Regnier, F. E. ( 198 I) lnt. Symp. HPLC Proteins Peptides, Paper No. 127, Washington, D. C. 16. Regnier, F. E. (1983) in Methods in Enzymology (Hits, C. H. W., and Timasheff, S. N., eds.), Vol. 91, pp. 137-190, Academic Press, New York. 17. Ahmed, A. I., Yeh, Y., Osuga, D. T., and Feeney, R. E. (1976) J. Biol. Chem. 251, 3033-3036. 18. Cantor, C. R., and Schimmel, P. R. (1980) Biophysical Chemistry, Parts II and Ill, Freeman, San Francisco. 19. Bush, C. A., Feeney, R. E., Gsuga, D. T., Ralapati, S., and Yeh, Y. (1981) Int. J. Peptide Protein Res. 17, 125-129.