Linda K. Gowan

On the primary and tertiary structure of relaxin from the sand tiger shark (Odontaspis taurus)

The structural variability between the two known relaxin sequences from the pig [l-3] and rat [4] far exceed that observed for any insulin pair, yet their primary sequences can readily be folded into an insulin conformation [ 561. This raises the question of how much more variability in the primary sequence may be tolerated without preventing an insulin-like folding to occur and how much of the surface may be varied without loss of receptor interaction. The evolutionary diversion of relaxins from each other and the possible existence of a common ancestral gene for relaxins and insulinsled us to examine the structure of shark relaxin. Based on the assumption that sharks have existed in their present form for at least 500 million years, i.e., have lived close to the point at which fishes and the ancestors of mammals are assumed to have branched from each other and where the insulin gene might have undergone duplication to give rise to the relaxin gene, we expected to observe a closer relation between the relaxin of this quasi-prehistoric species and insulins.

ON THE THREE‐DIMENSIONAL STRUCTURE OF RELAXIN

The observation that porcine relaxin comprises two polypeptide chains linked together with disulfide bridges which have the same disposition as those in led to the suggestion that relaxin and insulin may have similar three-dimensional structures. Although only five further residues of porcine relaxin are identical to those in equivalent positions of porcine insulin, model building studies 4 , showed that relaxin may have a tertiary structure closely resembling that of insulin. More recently the sequence of rat relaxin6 and the partial sequence of shark relaxin 7 have been determined. Although these sequences are rather different from each other and from that of porcine relaxin, model building using computer graphics techniques demonstrates that they may have similar tertiary structures. In this paper we review the use of model building techniques for constructing three-dimensional structures of such homologous globular hormones and we describe and compare the predicted structures for porcine and shark relaxins. We then discuss the relevance of these models to the biology and evolution of relaxin.

Isolation and partial amino acid sequence of three subunit species of porcine spleen ferritin: Evidence of multiple H subunits

A partial amino acid sequence for three different subunits of the iron storage protein, ferritin, has been determined. Ferritin (Mr approximately 480,000) was isolated from porcine spleen and dissociated into its component subunits (Mr approximately 20,000). The subunits, in turn, were separated into three fractions by reversed-phase HPLC. The fractions appeared to be of equal size by sedimentation velocity, sodium dodecyl sulfate-polyacrylamide gel electrophoresis, and size-exclusion chromatography in 6 M guanidinium chloride. All three fractions were shown to be monomeric and to have no covalently attached carbohydrate (J. F. Collawn et al. (1984) Arch. Biochem. Biophys. 233, 260-266). Determination of the amino acid sequence of the C-terminal 70-80 residues from each of the fractions demonstrated three different sequences. Comparison with human liver H and L subunit sequences indicates that two of the porcine ferritin subunits are H-type subunits and one is an L-type subunit. Application of the Chou-Fasman algorithm on the three partial sequences suggests that these respective regions from each of the three subunits would probably adopt the same conformation.

DEVELOPMENT OF A SPECIFIC RADIOIMMUNO ASSAY FOR E DOMAIN CONTAINING FORMS OF INSULIN-LIKE GROWTH FACTOR II

The analysis of cDNA clones for human (1,2) and rat (3) insulin-like growth factor-II (IGF-II)1 has led to the prediction that the processed forms of the growth factors, i.e. Mr = 7422 (67 amino acids) for human IGF-II are synthesized as precursors with an extension of 89 amino acids at the carboxyl terminus. This extension is termed the E domain. Moses et al (4) identified two precursor forms of rat IGF-II (originally designated multiplication-stimulating activity) in the conditioned medium of Buffalo rat liver, i.e. BRL-3A, cells with appMrs = 16,270 (MSA-1) and 8,700 (MSA-II). Human serum, spinal fluid and tissue extracts also contain high Mr forms of IGF-II (5-8). Zumstein et al., (5) have purified a Mr = 10,000 variant form of IGF-II from serum that contained Cys-Gly-Asp for Ser33 in the C domain and an E domain extension of 21- amino acids. This 10 kDa IGF-II may be similar or identical to the “big IGF-II” that was reported to be present in human serum and in spinal fluid (6). We have isolated a still larger form of IGF-II from normal human serum (7). N-terminal amino acid sequence analysis through the first 28 residues and RRAs using rat placental membranes established it as a form of IGF-II. As evidenced from its mobility during SDS-PAGE, it has an appMr = 15,000. Very recently, Hudgins et al., (8) established that normal human serum contains several forms of precursor IGF- II with acidic isoelectric points, i.e. pI’s. The mass and acidic nature of one of these molecules with an apparent Mr= 15,000 was contributed, in part, by polysaccharides and sialic acids, respectively (8). These results may explain the observations of several investigators that extracts from the tissues and serum of patients with malignant tumors contain a broad size range of IGF-II (2,9,10).

Evolution, relaxin and insulin: a new perspective.

It seems appropriate to introduce a session on evolution with a brief discussion of currently accepted concepts about the development of complexity in biological systems. My discussion will center around the structures of relaxin and insulin, two functionally unrelated hormones that originate from different tissues within one organism. From a bird’s eye perspective relaxin and insulin look identical, and only upon closer examination do differences in the primary structures become apparent. If gene duplication had not been discovered, certainly the relaxin-insulin pair would have induced the idea (FIGURE 1 ) . The idea of gene duplication, in fact, led to the experiment I wish to report here and to the suggestion that the gene duplication that permitted diversification might have been an early event preceding the formation of multicellular organisms. In TABLES 1 and 2 some previously known and some new sequences of the A and B chains of relaxin are compared with insulin sequences from hogs, hagfish, and rats. The first three lines show the sequence of the A chain of shark relaxin recently determined in our laboratory,l that of porcine relaxin,z-5 and the sequence of rat relaxin.c The cysteine residues are invariant as are the crosslink^,^ and no large blocks of identity are apparent as they are in the sequences of the A chains of insulins shown in the next three lines. While the relaxins are related to each other by about 50% of their primary sequence, the relationship of any relaxin to insulins extends to no more than 25% of the residues, including the cysteines. Notably preserved are the hydrophobic residues corresponding to the A? position, which are isoleucine in all insulins but show a leucine or methionine in the relaxin chains. Other such hydrophobic areas preserved are leucine in A,, of the insulin chains (isoleucine in the relaxin chains) and tyrosine in A,, of insulin (leucine in the corresponding position in the relaxin chains). The same general impression is gained when one examines the primary structure of the B chains of various relaxins and insulins (TABLE 2). Again the cysteines are preserved as well as the glycines following each cysteine. In addition, a few positions remain hydrophobic across all the chains compared in this figure. While the sequence homology between the relaxin and insulin is not very impressive, the cysteine residues are retained in their appropriate positions and the area forming the hydrophobic core of insulin seems preserved as well. The results are summarized in TABLE 3, which shows a numerical difference between the phylogenetically most distant pair of insulins as 29% and 41 %, respectively for the A and B chains, whereas relaxin and insulin from a single species differ by 75%-78% (hog, rat). In general, the differences between insulins amount to