Staffan ERIKSSON2, Hans JORNVALL3, Mats CARLQUIST' and Hans EKLUND ' ..... The skilful technical assistance of Margareta Karlsson and Ella. Cederlund ...
Eur. J. Biochem. 150,423-427 (1985)
0FEBS 1985
Protein B 1 of ribonucleotide reductase Direct analytical data and comparisons with data indirectly deduced from the nucleotide sequence of the Escherichia coli nrdA gene Britt-Marie SJOBERG
Staffan ERIKSSON2, Hans JORNVALL3, Mats CARLQUIST’ and Hans EKLUND
’ Department of Molecular Biology, Swedish University of Agricultural Sciences, Uppsala Biomedical Center
’
Departments of Biochemistry and Chemistry, Karolinska Institute, Stockholm (Received March 7/May 7, 1985) - EJB 850220
The total composition, the N-terminal amino acid sequence, and the amino acid sequences of four internal regions have been determined for the ribonucleotide reductase large subunit, protein B 1, prepared from a recombinant I-lysogenic Escherichiu coli K strain, which overproduces the enzyme 30 - 50-fold. The data have been compared with those previously reported for B 1 prepared from a thymin-starved E. coli B strain and with the indirectly derived primary structure of B1 recently reported from the nucleotide sequence of the E. coli K nrdA gene. Two major differences to these results were found. First, the B1 polypeptides started with initiator Met-1 (45%0),Asn-2 (30%) or Gln-3 (Is%), demonstrating a differrent type of N-terminal heterogeneity than that found earlier. Secondly, the total amino acid composition as derived from hydrolyzed protein B 1 differed substantially from the amino acid composition derived from the nucleotide data. This has the consequence that Cys, Arg, Thr and possibly Val and Ser appear more frequently whereas Asx, Glx, Tyr and possibly Gly appear less frequently in the nucleotide-derived data as compared to direct protein hydrolysates. We suggest usage of other reading frames in the approximate area of residues 630 - 700 of the primary structure of the nrdA gene to compensate for these discrepancies and for the relatively high incidence of uncommon codons in the reading frame proposed for this area of the gene. Such changes have implications on the previously assigned putative active-site region of protein B 1.
Ribonucleotide reductase is an essential enzyme in all living cells catalyzing the reduction of ribonucleotides to their corresponding deoxyribonucleotides [1]. The enzyme from Escherichiu coli is the most extensively studied ribonucleotide reductase and it is a prototype for all known eucaryotic ribonucleotide reductases [2]. The E. coli enzyme consists of two non-identical subunits, proteins B1 and B2, coded for by an operon [3] consisting of the nrdA and the nrdB gene, respectively. Protein B 1 contains binding sites for the ribonucleoside diphosphate substrates and the allosteric nucleoside triphosphate effector molecules [l]. It contributes redox-active disulphides to the active holoenzyme which consists of a one-toone complex of B 1 and B2 [4]. Protein B2 provides a unique tyrosine radical stabilized by an adjacent binuclear iron center [5, 61. Because of the low abundance of ribonucleotide reductase in wild-type cells, overproducing systems have been utilized for the isolation of both subunits. The enzyme was first prepared from thymin-starved E. coli B3 cells [7], a condition found to derepress the level of ribonucleotide reductase apCorrespondence to B.-M. Sjoberg, Institutionen for Molekylar Biologi, Sveriges Lantbrukuniversitet, Box 590, S-75124 Uppsala, Sweden Abbreviations. SDS, sodium dodecyl sulfate; kb, lo3 bases; EBV, Epstein-Barr virus, HSV, herpes simplex virus; DABITC, dimethylaminoazobenzene isothiocyanate. Enzyme. Ribonucleoside diphosphate reductase (EC 1.17.4.1).
proximately 10-12-fold [8]. Later the genes coding for ribonucleotide reductase in E. coli K was cloned into a heatinducible defective A vector [9]. After specific induction, cells lysogenic for this recombinant I bacteriophage produced 30 50 times more ribonucleotide reductase than the wild-type cells. A simplified purification scheme for ribonucleotide reductase was developed and resulted in large amounts of pure enzyme with characteristics identical to those of the earlier studied enzyme 191. The most recent achievement was the cloning of the E. coli K nrdA gene into an R 1 ‘run-away’ replicon [ 101. Heat inactivation of the copy-number control results in 150- 200 times elevated levels of protein B 1, which constitutes 30-40% of the soluble protein in a crude extract. Recently the DNA sequence of an 8.6-kb E. coli K fragment covering the nrd operon was reported [3]. Primary structures for the nrdA and nrdB genes have been deduced from this fragment. In addition, other nucleotide sequences coding for ribonucleotide reductase subunits have been published. In the eucaryotic virus herpes simplex type 2, a 140-kDa protein and a 38-kDa protein have been shown to code for B1 and B2 equivalents, based on immunological and enzymatic studies [11 - 151. In Epstein-Barr virus, a 93-kDa protein and a 34-kDa protein have been assigned as ribonucleotide reductase subunits due to their high degree of homology with the HSV2 140-kDa and 38-kDa sequences [16]. In the marine mollusc Spisulu solidissirnu one of the maternally most abundant RNA species, specifically translated after fertilization, has been shown to code for a B 2 equivalent (a 42-kDa protein)
424 with characteristic immunological and enzymatic activities [17]. Due to structural similarities at the amino acid sequence level, it has been possible to assign functional importance to a few amino acid residues of the B2 subunit [18]. In this report, we have correlated the nucleotide sequence of the nrdA gene of E. coli K [3], the structure of protein B 1 purified from the A-lysogenic over-producing E. coli K strain [9] and the data published earlier on B 1 isolated from thyminstarved E. coli B cells [7]. Such a study was necessary in order to clarify the relatively large differences between the predicted amino acid composition as judged from the nucleotide sequence and the corresponding experimentally determined values for pure protein B 1.
MATERIALS AND METHODS Protein B1 of ribonucleotide reductase was isolated in homogeneous f o r m f r o m heat-induced E. coli KK546 as described before [9]. It had a specific activity [9] of 700 U/mg protein and displayed a single band upon analytical SDS/ polyacrylamide gel electrophoresis [19]. Homogeneous protein B 1 was reduced with dithiothreitol and carboxymethylated with 14C-labelled or unlabelled iodoacetate [7] prior to acid hydrolysis or sequence degradation. Carboxymethylated B 1 polypeptides were either digested with trypsin (tosylphenylalanylchloromethane-treated;Worthington) for 16 h at 37°C in 0.1 M ammonium bicarbonate, pH 8.0, at protease/substrate ratios of 1 :50 (w/w), or treated with CNBr (60-fold molar excess over total Met content) in 70% formic acid for 24 h at room temperature. Peptides after CNBr cleavage were purified by Sephadex G-50 chromatography in 30% acetic acid followed by reversephase high-performance liquid chromatography on Waters Nova-Pak CIS Radial-Pak cartridge (8 x 100 mm). In order to keep peptides soluble, the separation was carried out in 50% formic acid with a gradient (30-50% in 40 min; 1 ml/ min) of acetonitrile. Purification of CNBr peptides was difficult due to non-stoichiometric cleavage. This result was in fact anticipated, because a Ser (or Thr) residue often follows Met in the data deduced from the nucleotide sequence [3] and therefore participates in a peptide bond less sensitive to CNBr cleavage [20] in analogy with other recent studies on large proteins [21]. Tryptic peptides were purified by DEAESephadex chromatography followed by reverse-phase highperformance liquid chromatography in 0.1% H3P04 and a gradient of acetonitrile [22]. Total compositions were determined with a Beckman 121 M amino acid analyzer after hydrolysis in evacuated tubes at 110°C for 24 h with 6 M HCl, 0.5% phenol. Amino acid sequences were determined by liquid-phase sequencer (Beckman 890 D) degradations using a 0.1 M Quadrol peptide program, and application into glycine-precycled polybrene [23]. Phenylthiohydantoin derivaties were identified by highperformance liquid chromatography [24] coupled, where applicable, with thin-layer chromatography and radioactivity measurements [23]. Manual degradations were carried out with the dimethylaminoazobenne isothiocyanate (DABITC) method [25], utilizing byproducts in the identifications [26]. Amino acid sequences were aligned using the programs GAP, COMPARE, DOT PLOT and BESTFIT of the University of Wisconsin program package [27] on a VAX 11/750 computer. The program FRAMES of this package was used to plot the frequency of rare codons [28, 291 for the different reading frames. Plotter programs were modified for an
Table 1. Amino acid composition of protein BI as derived f r o m two independent hydrolytic analyses and as deduced from the nucleotide sequence of the nrdA gene All values have been correlated to the length of the nrdA gene, 776 amino acid residues and refer to nearest integral number, n.d. = not determined Residue
Protein B 1 prepared from overthyminproducing starved E. coli K [9] E. coli B [7]
Asx Thr Ser Glx Pro GIY Ala Val Met Ile Leu TYr Phe LYS His '4% TrP CYS
90 40 48 85 33 51 65 38 16 48 75 43 31 40 18 37 n. d 10
87 37 48 83 34 50 65 40 1s 50 77 40 29 44 19 41 6 12
Deduced from DNA sequence of the nrdA gene of E. coli K [3]
Residues 634 - 702 of the nrdA gene of E. coli K [3]
71 53 57 67 32 38 63 50 17 48 12 33 27 43 16 56 9 22
2 11 9 1 3 1 4 5 2 3 2 0 1 3 1 10 3 8
HP 7220 plotter by M. Sundvall (Dept of Medical Immunology, Uppsala University, Sweden).
RESULTS
Total composition Two different sets of experimentally derived amino acid compositions of E. coli ribonucleotide reductase are available. The amino acid composition of protein B 1 prepared from the overproducing 1-lysogen KK546 is shown in Table 1 and compared with published data for protein B 1 prepared from thymin-starved E. coli B cells [7]. Also included in Table 1 is the amino acid composition of protein B1 as deduced from the nucleotide sequence of the cloned nrdA gene derived from a K12 strain [3]. For simplification of the comparisons, the hydrolysis data have been adjusted to the approximate length of the cloned gene (776 amino acid residues). It is obvious that there is a compelling overall agreement between the B and the K hydrolysis data. It is, however, also noteworthy that there are unexpectedly large discrepancies between the hydrolysis data and the composition derived from the nucleotide sequence as regards to Cys, Thr, Arg and possibly Val and Ser being more frequent and Asx, Glx, Tyr and possibly Gly being less frequent in the nucleotide sequence than in the hydrolysis data. Some of these discrepancies could be explained by contaminations, free amino acids, hydrolytic destructions, slow release or other factors disturbing the direct analysis of the protein. It is also possible that the differences could be related to discrepancies between the nucleotide and protein sequences, as shown below, and summarized in the discussion.
425 Table 2. Results of sequence analysis of the intact protein and of four peptides The intact protein was analyzed by liquid-phase sequencer degradations, and revealed an N-terminal heterogeneity as shown. Three major sequences were obtained (marked I, I1 and 111) and could be separately interpreted since they were phase-shifted by one and two residues, respectively, of the longest structure and were obtained in different yields (corresponding to about 45%, 30% and 15% of the total respectively). The four peptides were three CNBr fragments analyzed by liquid-phase sequencer degradations and one tryptic peptide analyzed by the DABITC method. The three CNBr fragments were those in the two pools (A and B) of radioactive material after Sephadex G-50 fractionation of the nucleotide-labelled protein [30]. Pool A was a mixture of two peptides, I and 11, but they could be separately interpreted as shown, because of different amounts of the two peptides, and because of the known DNA sequence [3]. The tryptic peptide was the major peptide in a pool also containing the nucleotide-labelled structure. Values show approximate yields in nanomoles of phenylthiohydantoin derivatives, D indicates identification by the DABITC method. Residues within parentheses were incompletely identified Cycle
1 2 3 4 5 6 7 8 9 10 11 12
13 14 15 16 17 18
Intact protein
CNBr peptides (pool A)
I
I1
Met 18 Asn 15 Gln 12 Asn 13 Leu 14 Leu 15 Val 10 Thr 8 Lys 9 Arg 5
Asn 12 Gln 9
Asp 6 Gly 5 (Ser) 2
Asn 12 Leu 10
Leu 9 Val 9 Thr 6 Lys 7 Arg 3 Gly 4 (Thr) 3
111
I
I1
-
Gly 4.1
(As4 Leu 6
(A%) (A%) Thr 2.9 Leu 3.4 Gly 2.8 Ile 2.9
Ser 1.8 Gly 1.9 Val 2.8
Leu 4 Val 3 -
Lys 5 Arg 2 Asp 4 -
(Ser) 1 -
Amino-terminal amino acid sequence
The N-terminal sequence of protein B 1 as derived from a sequencer degradation of the full-length carboxymethylated polypeptide is shown in Table 2. In the first cycle, both Met and Asn were obtained. Despite this inhomogeneity, the sequencer degradation could be followed for more than seven or eight steps. The data establish that the protein preparation is composed of three major types of protein chains in approximate yields of 45%, 30% and 15% (peptides I, I1 and I11 in Table 2), respectively. The three chains are related, differing only by a phase-shift of one and two residues, respectively, thus establishing that the preparation is essentially pure protein B 1. The longest structure (peptide I) corresponds exactly to the DNA-derived amino acid sequence starting at the initiator Met [3]. The other two chains (peptides I1 and 111) start at positions 2 and 3 of the DNA-derived sequence and suggest an N-terminal processing heterogeneity in the intact protein. There is thus an excellent correlation between this direct sequencer results and the primary structure deduced from the nucleotide seuquence [3]. Approximately half of the overproduced protein B1 from strain KK546 starts at the same point as the translational start point in the nucleotide sequence. Internal amino acid sequences
In an attempt to localize allosterically important areas in protein B 1, the photoaffinity labelled subunit [30] was carboxymethylated and digested with trypsin or CNBr. In order to keep the CNBr fragments in solution during separation, they were submitted to reverse-phase high-performance
Gly 2.8 Val 3.0 Ile 2.6 Asn 2.2 Phe 2.4 Ala 2.4 Tyr 1.9
(TYr)
Leu 1.9 Ala 1.7 Lys 1.5
(Arg) Thr 2.2 Pro 2.1 Thr 2.1 (Arg) Gln 0.5 Phe 1.5 Ser 0.3 Ser 0.5
Cys(Cm) 0.4 Val 1.1 (Leu) Ile 0.9 Glu 0.4
[Cys(Cm)l
CNBr peptide (PO01 B)
Tryptic peptide
(Thr) Phe 0.6 (Ser) Tyr 0.6 Ala 0.7 Ala 0.8
Tyr D Val D Ser D Gln D Arg D
Val 0.6 Lys 0.4
(GI4
Leu 0.4
(GW
Gly 0.3 Lys 0.3 Tyr 0.2 Leu 0.1 Val 0.2 (A4
liquid chromatography in strong acid (cf. [31]). Consequently, the gradient for elution utilized acetonitrile in 50% formic acid. After this separation labelled peptides were analyzed for their amino acid sequence. The sequence of one tryptic fragment and sequences of three CNBr fragments are presented in Table 2. These stretches of primary structure are readily localized within the deduced amino acid sequence of the nrdA gene and match exactly with residues 142 - 157, 212 - 229, 246 -250 and 507 - 524 [3]. Our confirmation of the amino acid sequence of the N-terminal and these four internal regions as deduced from the DNA-derived data have been indicated in Fig. 1.
DISCUSSION Despite the generally good correlation between sequencer results on B 1 protein and the primary structure deduced from the nucleotide sequence of the nrdA gene [3], there still exist remarkable differences in the total composition data (Table 1). These large discrepancies cannot be explained by strain differences, because the E. coli B and K hydrolytic data are consistant, whereas the K nucleotide data differ from K as well as the B [7] hydrolysis data. Since the differences have the appearance of some specific residues being overrepresented in the nucleotide data, segments of nucleotidederived amino acid sequence were analysed for the relative abundance of the three residues Cys, Arg and Thr (Fig.1). The sum of these three residues amounts to approximately 11 -12% of the hydrolysis data (Table 1). It is clear from Fig. 1 that these residues have a non-random occurrence in the deduced amino acid sequence corresponding to the re-
7 ,./'
I
.
I
,'
,
0
.
Y 0
I
,.
/
'/
/'
.
.
0
,200 m
100
200
300
400
500
600
'
',*
I
m
700
. I
.
Residue number
i L
r
m
C .'0 Q cp
a
a:
5'
b.
-
'., ._
-
. . . .... .......................
.........
300
-.
600
900
.....
_
-
4
'i' I
*
. . . . . . . .-
,
'8
w
I
,
,'
/
*
"
, I
.
.. ,
I
0
3'
......
1200 1500 1800 2100
Nucleotide number
Fig. 1. Relative abundance of Cys, Arg and Thr and uncommon codon usage in E. coli protein B l , as deduced from the DNA sequence 131. Areas confirmed by direct sequence analyses have been indicated by bars at the top of the upper graph. The lower plot displays the codon usage in all three forward reading frames (codons AGG, CGG, CGA for Arg, UUG, CUA for Leu and UGA, UAG, UAA for stop, all have a codon frequency of less than or equal to 0.1 [27 - 291 and are marked with a dot) and indicates all open reading frames (boxed) initiated with an AUG codon and terminated with one of the three known stop codons [27]
ported nrdA gene. Most prominent is the region at approximately residues 590 - 700, where the relative abundance of Cys, Arg and Thr amounts to more than 40%. This segment, to which a putative active-site region (residues 667 670) has been assigned [3], thus has a very rare amino acid composition and sequence. We therefore undertook a search of codon usage in the reported nrdA gene taking into account that ribonucleotide reductase belongs to the class of moderately to low abundant proteins, and these results are also included in Fig. 1. There is a relatively high incidence of uncommon codons in the approximate area of residues 630 700 in the open reading frame used for the nrdA gene. In the other two frames there are long uninterrupted coding stretches (data not fully shown in Fig. 1 since the program [27] indicates open reading frames as initiated by an AUG codon) devoid of uncommon codons approximately from residues 590 - 610 and 650-710. We would therefore like to suggest that parts of the other reading frames be used for the B 1 polypeptide in the approximate area of residues 634-702. Such an exchange would result in an amino acid composition of the nrdA gene which is more similar to the hydrolysis data for E. coli B and K strains than the presently available composition deduced from the nucleotide sequence (Table 1). This suggestion, which offers a plausible explanation to most deviations will probably frustrate the assignment of the putative active site of B 1 [3]. A definite clarification of these discrepancies requires a complete amino acid sequence determination of the 776-residue B 1 polypeptide, which is beyond the scope of this report, or a determination of the nucleotide sequence of both DNA
0
200
LOO
600
f. coli protein B1 [residue number)
Fig. 2. Dot plot (271 comparison of the E. coli protein B1, as deduced from the DNA sequence [3J and the 93-kDaprotein ofEBV f 16). The peptide scoring comparison table of Staden was used [33], the window was 30 residues long and the stringency was set to 32
strands in this area of the nrdA gene, which is already under way (Carlsson, J., Fuchs, J. A. & Messing, J., unpublished). The complete nucleotide sequence of a B 1 equivalent has been reported for a 93-kDa protein of Epstein-Barr virus (EBV) [16]. Similarly, a 140-kDa protein in herpes simplex virus [ l l ] has been inferred as a B1 equivalent. We have compared (Fig. 2 ) and aligned (data not shown) the amino acid sequence of the nrdA gene product [3]with the EBV data. It is obvious from Fig.2 that the EBV 93-kDa protein and E. coli protein B1 share regions of distant homology over their entire lengths of polypeptides. However, some areas, especially at the C-terminal, bear negligable relationship. For example, the segment with a high Cys, Arg and Thr content, residues 630-700, in protein B1 shows no relationship to any C-terminal EBV sequence (Fig.2). Only a few scattered similarities are found in the segment of residues 634- 702 upon alignment of the nrdA protein sequence with the 93-kDa EBV protein (data not shown). In contrast, it was earlier demonstrated that the EBV and HSV sequences have a pronounced relationship in this region and throughout [16]. The similarities between E. coli and EBV sequences are primarily manifested in short conserved areas like Gln-Arg-Ala-Gly, Glu-Phe-Phe-Glu-Arg-Xaa-Tyr and Ala-Leu-Met-Pro, corresponding to residues 249-252, 371-377 and 616-619 in protein B 1. Because the relationship is distant, conserved areas are most likely encountering residues of functional importance like allosteric and catalytic sites of ribonucleotide reductase. It is quite concievable that the allosteric sites of ribonucleotide reductase should share sequence similarities of the order of the ones found for other nucleotide binding domains like the NAD site of dehydrogenases [32]. An alignment of even more B 1 equivalents from different species will undoubtedly help to substantiate these findings and may constitute the basis for future probing of such areas with sitedirected mutagenesis. It was earlier found with protein B1 prepared from thymin-starved cells that the dimeric protein had two different N-terminal amino acid residues in roughly equal amounts [7].
427 Two different chymotryptic N-terminal peptides were found and it was concluded that the N-terminal difference reached further into the polypeptide chain [7]. Recent support came from the reported sequence data of the nrdA gene, where processing of 2 and 25 residues, respectively, was proposed to explain this N-terminal heterogeneity of B 1 [3]. In overproduced protein B 1 from the KK 546 strain, roughly the same percentage of polypeptides as in B1 prepared from thymin-starved cells have undergone partial proteolysis, but only causing removal of one or two N-terminal residues. Overproduced protein B 1 shows no apparent difference to protein B 1 prepared from thymin-starved cells with regards to retention on ion-exchange columns, molecular mass on SDS/ polyacrylamide gel electrophoresis, specific activity, or binding of allosteric effectors [9]. This demonstrates that extensive post-translational modification of B 1 polypeptides is neither an a priori demand for activity nor does it affect the enzyme activity as assayed in our in vitro systems. It remains to be tested whether protein B 1 from wild-type E. coli cells resembles the thymin-starved or the over-produced protein B1. In summary, we have found that the primary structure of the nrdA gene as deduced from the DNA sequence fits by and large with data derived from the isolated homogeneous B1 protein. Two major differences from earlier published results [3, 71 are obvious. (a) The N-terminal part of over-produced protein B1 is heterogeneous, but in a different manner than previously suggested, and starts with initiator Met-1, Asn-2 or Gln-3, respectively. (b) In the approximate region of nucleotide positions 5408 - 5612 we suggest the usage of other reading frames, because of the high incidence of uncommon codon usage and the large discrepancy from the total composition of the real protein. Such changes will undoubtedly affect the previously [3] assigned putative active-site region of B 1. The skilful technical assistance of Margareta Karlsson and Ella Cederlund is gratefully acknowledged. This work was supported by grants from the Swedish Medical and Natural Science Research Councils and the Magn. Bergvall Foundation. Note. During the preparation of this paper, the primary structure of the mouse ribonucleotide reductase subunit M 1 has been deduced from DNA sequence data [34]. An alignment of subunits M I and B 1 shows very few similarities in the C-terminal region, whereas an alignment of subunits M 1 and EVB 93-kDa protein shows substantial similarity in the same area.
REFERENCES 1. Reichard, P. & Ehrenberg, A. (1983) Science (Wasch. DC) 221, 514-519. 2. Sjoberg, B.-M. & Graslund, A. (1983) Adv. Znorg. Biochem. 5, 87- 110. 3. Carlson, J., Fuchs, J. A. & Messing, J. (1984) Proc. Natl Acad. Sei. USA 81,4294 - 4297.
4. Thelander, L. (1974) J . Biol. Chem. 249,4858-4862. 5. Petterson, L., Graslund, A,, Ehrenberg, A,, Sjoberg, B.-M. & Reichard, P. (1980) J . Biol. Chem. 255, 6706-6712. 6. Sjoberg, B.-M., Sanders-Loehr, J. & Loehr, T. (1982) Biochemistry 21, 96 - 102. 7. Thelander, L. (1973) J . Biol. Chem. 248,4591 -4601. 8. Biswas, C., Hardy, J. & Beck, W. S. (1965) J . Biol. Chem. 240, 3631 - 3640. 9. Erikson, S., Sjoberg, B.-M., Hahne, S. & Karlstrom, 0. (1977) J . Biol. Chem. 252, 6132-6138. 10. Larsson, A. (1985) Acta Chem. Scand. B38,905-907. 11. McLauchlan, J. & Clements, J. B. (1983) EMBO J . 2, 19531961. 12. Dutia, B. M. (1983)J. Gen. Virol. 64, 513-521. 13. Huszar, D. & Bacchetti, S. (1983) Nature (Lond.) 302, 76-79. 14. Bacchetti, S., Evelegh, M. J., Muirhead, B., Sartori, C. S. & Huszar, D. (1984) J . Virol. 49, 591 -593. 15. Preston, V. G., Palfreyman, J. W. & Dutia, B. M. (1984) J . Gen. Virol. 65, 1457 - 1466. 16. Gibson, T., Stockwell, P., Ginsburg, M. & Barrell, B. (1984) Nucleic. Acids Res. 12, 5087 - 5099. 17. Standart, N. M., Bray, S. J., George, E. L., Hunt, T. & Ruderman, J. V. (1985) J . Cell. Biol. 100, 1968-1976. 18. Sjoberg, B.-M., Eklund, H., Fuchs, J. A., Carlson, J., Standart, N. M., Ruderman, J. V., Bray, S. J. & Hunt, T. (1985) FEBS Lett. 183, 99 - 102. 19. Laemmli, U. K. (1970) Nature (Lond.) 227, 680-685. 20. Schroeder, W. A., Shelton, J. B. & Shelton, J. R. (1969) Arch. Biochem. Biophys. 130,551 -555. 21. Hempel, J., Kaiser, R. & Jornvall, H. (1985) Eur. J . Biochem., in the press. 22. von Bahr-Lindstrom, H., Carlquist, M., Mutt, V. & Jornvall, H. (1982) in Methods in protein sequence analysis (Elzinger, M., ed.) pp. 455-462, Humana Press, Clifton. 23. Jornvall, H. & Philipson, L. (1980) Eur. J. Biochem. 104, 237247. 24. Zimmerman, C. L., Appella, E. & Pisano, J. J. (1977) Anal. Biochem. 77, 569 - 573. 25. Chang, J. Y., Brauer, D. & Wittmann-Liebold, B. (1978) FEBS Lett. 93, 205-214. 26. von Bahr-Lindstrom, H., Hempel, J. & Jornvall, H. (1982) J . Protein Chem. 1, 257-262. 27. Devereux, J., Haeberli, P. & Smithies, 0. (1984) Nucleic. Acids Res. 12, 387-395. 28. Grantham, R., Gautier, C. & Gouy, M. (1980) Nucleic. Acids Res. 8, 1893-1912. 29. Grantham, R., Gautier, C., Gouy, M., Jacobzone, M. & Mercier, M. (1981) Nucleic Acids Res. 9, r43-r74. 30. Erikson, S. (1983) J . Biol. Chem. 258, 5674-5678. 31. Kratzim, H. D., Kruse, T., Maywald, F., Thinnes, F. P., Gotz, H., Egert, G., Pauly, E., Friedrich, J. & Yang, C.-Y. (1984) J . Chromatogr. 297, 1 - 11. 32. Rossmann, M. G., Liljas, A., Branden, C.4. & Banaszak, L. J. (1975) in The enzymes (Boyer, P. D., ed.) 3rd edn, vol. 13, pp. 61 - 102, Academic Press, New York. 33. Staden, R. (1982) Nucleic Acids Res. 10, 2951 -2961. 34. Caras. I. W., Levinson, B. B., Fabry, M., Williams, S. R. & Martin, D. W., Jr (1985) J . Biol Chem. 260, 7015-7022.
