TIBS 15-JANUARY1990
CONSIDERABLE ATTENTION is presently focused on determining the functions of the lipocortin/calpactin family of membrane-associated Ca2+-binding proteins (for reviews see Refs 1,2). The structural feature which both characterizes and defines this protein family, is a repetitive conserved segment of 64 amino acid residues, that occurs four times in lipocortin I, calpactin I, protein If, endonexin II, lipocortin III and synexin; and eight times in p68 (67 kDa calelectrin). This structural feature is generally believed to reflect the capacity of each of these proteins to bind Ca~+, and to associate Ca2+-dependently with phospholipids. The lipocortins/calpactins apparently differ from other CaZ+-binding proteins in that they lack sequences conforming to the 'EF hand 3 helix-loop-helix structure common to calmodulin and troponin C, and the structure(s) of their Ca2+-binding sites have not yet been defined. However, it should be noted that secondary structure predictions based on the primary amino acid sequence of lipocortin !, have suggested that Ca2+-binding may be achieved through an alternative helix-loop--helix, in which most of the coordinating ligands are provided by main chain carbonyls 4. Important clues relating both to the nature of the Ca2+-binding site and to the evolution of this protein family, may be contained within the sequence of a Ca2+-binding protein from Streptomyces erythraeus described recently by Swan et al.~ These authors drew attention to the marked similarity of sequence of four peptide segments of the bacterial protein, with the four EF hands of human calmodulin. They also reported that the second of the four repeats of the bacterial protein was unlike the other three repeats, in that it lacked certain of the amino acid requirements of the classic EF hand structure. Thus, the Ca2+-binding ligands provided at positions 1 and 3 of the canonical EF hand helix-loop-helix structure, are almost invariably the [~-carboxyl groups of aspartic acid. However, in the second repeat of the bacterial Ca2+-binding protein, these were replaced by glycine residues. Despite it's non-conformity with the typical EF hand structure, the authors suggested that the second repeat might still bind Ca2+. S. E. Moss and M. J. Crumpton are at the Imperial Cancer Research Fund Laboratories, PO Box 123, Lincoln's Inn Fields, London WC2A 3PX, UK.
It is of considerable interest to note that the second repeat of the bacterial protein, that may represent a 'mutant' EF hand, bears a close similarity to a highly conserved consensus sequence which provided the first evidence for a structural relationship between the members of the lipocortin/calpactin family6. Thus, alignment of this region with the corresponding region of repeat 2 of human caipactin I, revealed almost 50% sequence homology (Fig. lb). As the basis of phylogenetic comparisons is that conserved structure reflects conserved function, the above similarity argues that this domain within calpactin 1 (and by inference, the other
members of the family) may harbour the Ca2+-binding site. In support of the homology, it should be noted that the penultimate residue of the aligned sequences in Fig. lb is a conserved arginine. The arginine in this position occurs without exception in every repeat of all seven lipocortin/calpactin family members, in all species investigated. Taylor and Geisow4 suggested that this invariant arginine forms a stabilizing ionic interaction with two of the acidic residues which provide Ca2+-binding ligands. Although the second repeat of the bacterial protein has the acidic residues necessary to provide coordinating ligands conforming to the Taylor
(a)
S. erythraeus
1 3 5 7 9 ~ F ~ G ~ G ~: L ~ :R ~ :
12 Repeat 1
Calmodulin
K
G
G
Z
T
Repeat 2 BA~
G
F
Repeat 3
Repeat 4
(b) S. erythraeus
L F D Y L A K E A G'V G S D G - S L T E E'Q F I ~ ~I
Calpactin I (repeat 2)
A S E L K A S *4 K |G L G'T D E D S,L • E I I C S,,,N
I
Figu• 1 (a) Alignment of the four proposed 'EF' hands of Streptomyces erythraeus with the corresponding four EF hands of human calmodulin. Boxed residues are those which conform to the requirements of an £F hand calcium-binding site. (b) Alignment of the second S. erythraeus £F hand (plus some additional flanking sequence) with the second repeat of human calpactin I. The boxed region corresponds to the consensus sequence described by Geisow et al. 6
© 1990,ElsevierSciencePublishersLtd,(UK) 0376-5067/90/$02.00
11
TIBS 15 -JANUARY1990
and Geisow model 4, insertion of a space (Fig. 1b) is required to bring the invariant arginine into alignment. It should also be noted that no regions of significant homology could be identified elsewhere between the sequences of calpactin I and the bacterial protein. Furthermore, the lipocortin/calpactin family proteins share structural homology over a 64 amino acid repetitive domain, of which the 17 amino acid sequence described by Geisow et al. 6 forms only a small part. As a result, it is clearly not obligatory that their shared Ca2+-binding site is contained within the 17 amino acid EF hand-related sequence.
Representatives of the lipocortin/ caipactin family have yet to be unequivocally identified in either lower eukaryotes or prokaryotes, although on the basis of the above deductions, this would now appear to be a possibility. Also, the presence within a single bacterial Ca2+-binding protein of both a lipocortin/ calpactin-related sequence and calmodulin-like EF hands, raises the intriguing possibility that these two distinct families of proteins may share common ancestry. Whilst this is an attractive hypothesis, it is clearly a relatively complex matter to account for the separation and evolution
of the lipocortin/calpactin-like proteins from the second repeat of a four-repeat calmodulin-related polypeptide.
References 1 Crornpton,M. R., Moss, S. E. and Crumpton, M. J. (1988) Ce1155,1-3 2 Klee,C. B. (1988) Biochemistry27, 66456653 3 Tufty, R. M. and Kretsinger,R. H. (1975) Science 187,167-169 4 Taylor,R. M. and Geisow,M. J. (1987) Protein Eng. 1,183-187 5 Swan,D. G., Hale, R. S., Dhillon, N. and Leadlay,P. F. (1987) Nature 329, 84-85 6 Geisow,M. J., Fritsche, U., Hexham,J. M., Dash, B. and Johnson,T. (1986) Nature 320, 636638
LETTERS Anchorin CII sequence now corrected As Moss and Crumpton had suggested correctly in their recent letter to TIBS ~, the sequence peculiarity in the second tetrad of chick anchorin Cll (residues 1 2 5 - 1 7 0 ) , which had no homology to the otherwise homologous protein human endonexin il, was a result of DNAsequencing artefacts, and not due to alternative splicing of a common gene as had been suggested by Haigler et al. 2 The result of new sequencing work on three new clones obtained by Drs R. Gropp and C. Hofmann in our laboratory clearly confirms
Correct use of Scatchard plots The problem of the misuse of non-linear Scatchard plots presented by Zierler 1,is part of a wider misunderstanding of complex ligand-binding and enzyme mechanisms. Ligand-acceptor binding data, in common with initial velocity data on enzymesubstrate interactions, transport processes and pharmacokinetic processes, are generally collected under steady state conditions. Such processes are generally described by a rational polynomial function of the type:
v
_alS+a2S2+...anS
n
(1)
1 -k blS d- b2S2 -F . . . bm Sm
where v = the variable under study (e.g. the concentration of bound ligand in a ligandbinding study or v/[E0];initial velocity/ [enzyme concentration] for an enzyme12
that the stretch of 45 amino acids (125-170) published by Fernandez et aL 3 and Pf/iffleet aL 4 is incorrect; one base was missing, and seven bases were incorrectly placed into the sequence. Errata giving the corrected sequence will appear in EMBOJ. and J. Biol. Chem. On the basis of the corrected sequence, anchorin Cll can now clearly be considered as the homologue of human endonexin II.
References 1 Moss, S. E. and Crurnpton,M. J. (1989) Trends
3 Fernandez,M. P., Selrnin, 0., Martin, G. R., Yamada,Y., Pf~ffle, M., Deutzmann,R., Mollenhauer,J. and yon der Mark, K (1988) J, Biol. Chem. 263, 5921-5925 4 Pf~iffle,M., Roggiero,R., Hofmann,H., Fernandez,M, P., Selmin, 0., Yarnada,Y., Garrone,R. and yon der Mark, K. (/988) EMBOJ. 7, 2335-2342
K. VON..DER MARK, C. HOFMANN, M. PFAFFLEAND R. GROPP
Biochem, ScL14,325 2 Haigler, H. T., Fitch, J. M., Jones,J. M. and Schlaepfer, D. D. (1989) TrendsBiochem.Sci. 14, 48-50
Max-Planck-Gesellschaftzur FOrderungder Wissenschaften e.V., MPGGruppeRheurnatologie/ Bindegewebsforschung,Schwabachanlage10, D-8520 Erlangen,FRG.
catalysed reaction); S = the concentration of free ligand or varied substrate; al ... a, and bl ... b,n are constants, all other ligand and product concentrations are constant, and there is no receptor or enzyme polymerization. The values of n and m (i.e. the degree of the function) reflect the number of receptor or enzyme species reacting with the ligand. Normally n < m. If n = m = l, then the equation simplifies to Scatchard or MichaelisMenten 1:1 functions where a~ = n/IQ (or V~/Km) and bl = 1//Q(or 1/Km). Complex rational polynomial functions which are models of biological steady state processes have been analysed in detail 2-s. The practical aim of the biochemist should be to determine experimentally binding or catalytic data over as wide a concentration range as is feasible, and then fit that data to an appropriate experimental model. One pitfall in such a study is to use an over-simplistic model - as Einstein said, everything should be reduced to as simple a level as possiblebut not more simple! What is required is the
determination of the number of receptor or enzyme species reacting with ligand (i.e. the values of n and m in the above equation- the degree of the rational polynomial function) and determination of the values of the binding or catalytic constants in the mechanism (i.e. the values ofal ... a, and bl ... bn, in Eqn 1). These quantitative parameters may be obtained either by computational or by graphical methods. Computational methods usually involve iterative non-linear regression followed by some technique to establish 'goodness of fit'. Many computer programs have been described which carry out this procedure (see Ref. 6 for a review); even common statistical programs such as SASmay be used in this way.Analysis of residuals (i.e. the difference between observed data points and data points predicted by a particular model) has been used to differentiate between simple and complex ligand-binding models 7. Graphical procedures are simple and accessible to those without computers, but are more
CONSIDERABLE ATTENTION is presently focused on determining the functions of the lipocortin/calpactin family of membrane-associated Ca2+-binding proteins (for reviews see Refs 1,2). The structural feature which both characterizes and defines this protein family, is a repetitive conserved segment of 64 amino acid residues, that occurs four times in lipocortin I, calpactin I, protein If, endonexin II, lipocortin III and synexin; and eight times in p68 (67 kDa calelectrin). This structural feature is generally believed to reflect the capacity of each of these proteins to bind Ca~+, and to associate Ca2+-dependently with phospholipids. The lipocortins/calpactins apparently differ from other CaZ+-binding proteins in that they lack sequences conforming to the 'EF hand 3 helix-loop-helix structure common to calmodulin and troponin C, and the structure(s) of their Ca2+-binding sites have not yet been defined. However, it should be noted that secondary structure predictions based on the primary amino acid sequence of lipocortin !, have suggested that Ca2+-binding may be achieved through an alternative helix-loop--helix, in which most of the coordinating ligands are provided by main chain carbonyls 4. Important clues relating both to the nature of the Ca2+-binding site and to the evolution of this protein family, may be contained within the sequence of a Ca2+-binding protein from Streptomyces erythraeus described recently by Swan et al.~ These authors drew attention to the marked similarity of sequence of four peptide segments of the bacterial protein, with the four EF hands of human calmodulin. They also reported that the second of the four repeats of the bacterial protein was unlike the other three repeats, in that it lacked certain of the amino acid requirements of the classic EF hand structure. Thus, the Ca2+-binding ligands provided at positions 1 and 3 of the canonical EF hand helix-loop-helix structure, are almost invariably the [~-carboxyl groups of aspartic acid. However, in the second repeat of the bacterial Ca2+-binding protein, these were replaced by glycine residues. Despite it's non-conformity with the typical EF hand structure, the authors suggested that the second repeat might still bind Ca2+. S. E. Moss and M. J. Crumpton are at the Imperial Cancer Research Fund Laboratories, PO Box 123, Lincoln's Inn Fields, London WC2A 3PX, UK.
It is of considerable interest to note that the second repeat of the bacterial protein, that may represent a 'mutant' EF hand, bears a close similarity to a highly conserved consensus sequence which provided the first evidence for a structural relationship between the members of the lipocortin/calpactin family6. Thus, alignment of this region with the corresponding region of repeat 2 of human caipactin I, revealed almost 50% sequence homology (Fig. lb). As the basis of phylogenetic comparisons is that conserved structure reflects conserved function, the above similarity argues that this domain within calpactin 1 (and by inference, the other
members of the family) may harbour the Ca2+-binding site. In support of the homology, it should be noted that the penultimate residue of the aligned sequences in Fig. lb is a conserved arginine. The arginine in this position occurs without exception in every repeat of all seven lipocortin/calpactin family members, in all species investigated. Taylor and Geisow4 suggested that this invariant arginine forms a stabilizing ionic interaction with two of the acidic residues which provide Ca2+-binding ligands. Although the second repeat of the bacterial protein has the acidic residues necessary to provide coordinating ligands conforming to the Taylor
(a)
S. erythraeus
1 3 5 7 9 ~ F ~ G ~ G ~: L ~ :R ~ :
12 Repeat 1
Calmodulin
K
G
G
Z
T
Repeat 2 BA~
G
F
Repeat 3
Repeat 4
(b) S. erythraeus
L F D Y L A K E A G'V G S D G - S L T E E'Q F I ~ ~I
Calpactin I (repeat 2)
A S E L K A S *4 K |G L G'T D E D S,L • E I I C S,,,N
I
Figu• 1 (a) Alignment of the four proposed 'EF' hands of Streptomyces erythraeus with the corresponding four EF hands of human calmodulin. Boxed residues are those which conform to the requirements of an £F hand calcium-binding site. (b) Alignment of the second S. erythraeus £F hand (plus some additional flanking sequence) with the second repeat of human calpactin I. The boxed region corresponds to the consensus sequence described by Geisow et al. 6
© 1990,ElsevierSciencePublishersLtd,(UK) 0376-5067/90/$02.00
11
TIBS 15 -JANUARY1990
and Geisow model 4, insertion of a space (Fig. 1b) is required to bring the invariant arginine into alignment. It should also be noted that no regions of significant homology could be identified elsewhere between the sequences of calpactin I and the bacterial protein. Furthermore, the lipocortin/calpactin family proteins share structural homology over a 64 amino acid repetitive domain, of which the 17 amino acid sequence described by Geisow et al. 6 forms only a small part. As a result, it is clearly not obligatory that their shared Ca2+-binding site is contained within the 17 amino acid EF hand-related sequence.
Representatives of the lipocortin/ caipactin family have yet to be unequivocally identified in either lower eukaryotes or prokaryotes, although on the basis of the above deductions, this would now appear to be a possibility. Also, the presence within a single bacterial Ca2+-binding protein of both a lipocortin/ calpactin-related sequence and calmodulin-like EF hands, raises the intriguing possibility that these two distinct families of proteins may share common ancestry. Whilst this is an attractive hypothesis, it is clearly a relatively complex matter to account for the separation and evolution
of the lipocortin/calpactin-like proteins from the second repeat of a four-repeat calmodulin-related polypeptide.
References 1 Crornpton,M. R., Moss, S. E. and Crumpton, M. J. (1988) Ce1155,1-3 2 Klee,C. B. (1988) Biochemistry27, 66456653 3 Tufty, R. M. and Kretsinger,R. H. (1975) Science 187,167-169 4 Taylor,R. M. and Geisow,M. J. (1987) Protein Eng. 1,183-187 5 Swan,D. G., Hale, R. S., Dhillon, N. and Leadlay,P. F. (1987) Nature 329, 84-85 6 Geisow,M. J., Fritsche, U., Hexham,J. M., Dash, B. and Johnson,T. (1986) Nature 320, 636638
LETTERS Anchorin CII sequence now corrected As Moss and Crumpton had suggested correctly in their recent letter to TIBS ~, the sequence peculiarity in the second tetrad of chick anchorin Cll (residues 1 2 5 - 1 7 0 ) , which had no homology to the otherwise homologous protein human endonexin il, was a result of DNAsequencing artefacts, and not due to alternative splicing of a common gene as had been suggested by Haigler et al. 2 The result of new sequencing work on three new clones obtained by Drs R. Gropp and C. Hofmann in our laboratory clearly confirms
Correct use of Scatchard plots The problem of the misuse of non-linear Scatchard plots presented by Zierler 1,is part of a wider misunderstanding of complex ligand-binding and enzyme mechanisms. Ligand-acceptor binding data, in common with initial velocity data on enzymesubstrate interactions, transport processes and pharmacokinetic processes, are generally collected under steady state conditions. Such processes are generally described by a rational polynomial function of the type:
v
_alS+a2S2+...anS
n
(1)
1 -k blS d- b2S2 -F . . . bm Sm
where v = the variable under study (e.g. the concentration of bound ligand in a ligandbinding study or v/[E0];initial velocity/ [enzyme concentration] for an enzyme12
that the stretch of 45 amino acids (125-170) published by Fernandez et aL 3 and Pf/iffleet aL 4 is incorrect; one base was missing, and seven bases were incorrectly placed into the sequence. Errata giving the corrected sequence will appear in EMBOJ. and J. Biol. Chem. On the basis of the corrected sequence, anchorin Cll can now clearly be considered as the homologue of human endonexin II.
References 1 Moss, S. E. and Crurnpton,M. J. (1989) Trends
3 Fernandez,M. P., Selrnin, 0., Martin, G. R., Yamada,Y., Pf~ffle, M., Deutzmann,R., Mollenhauer,J. and yon der Mark, K (1988) J, Biol. Chem. 263, 5921-5925 4 Pf~iffle,M., Roggiero,R., Hofmann,H., Fernandez,M, P., Selmin, 0., Yarnada,Y., Garrone,R. and yon der Mark, K. (/988) EMBOJ. 7, 2335-2342
K. VON..DER MARK, C. HOFMANN, M. PFAFFLEAND R. GROPP
Biochem, ScL14,325 2 Haigler, H. T., Fitch, J. M., Jones,J. M. and Schlaepfer, D. D. (1989) TrendsBiochem.Sci. 14, 48-50
Max-Planck-Gesellschaftzur FOrderungder Wissenschaften e.V., MPGGruppeRheurnatologie/ Bindegewebsforschung,Schwabachanlage10, D-8520 Erlangen,FRG.
catalysed reaction); S = the concentration of free ligand or varied substrate; al ... a, and bl ... b,n are constants, all other ligand and product concentrations are constant, and there is no receptor or enzyme polymerization. The values of n and m (i.e. the degree of the function) reflect the number of receptor or enzyme species reacting with the ligand. Normally n < m. If n = m = l, then the equation simplifies to Scatchard or MichaelisMenten 1:1 functions where a~ = n/IQ (or V~/Km) and bl = 1//Q(or 1/Km). Complex rational polynomial functions which are models of biological steady state processes have been analysed in detail 2-s. The practical aim of the biochemist should be to determine experimentally binding or catalytic data over as wide a concentration range as is feasible, and then fit that data to an appropriate experimental model. One pitfall in such a study is to use an over-simplistic model - as Einstein said, everything should be reduced to as simple a level as possiblebut not more simple! What is required is the
determination of the number of receptor or enzyme species reacting with ligand (i.e. the values of n and m in the above equation- the degree of the rational polynomial function) and determination of the values of the binding or catalytic constants in the mechanism (i.e. the values ofal ... a, and bl ... bn, in Eqn 1). These quantitative parameters may be obtained either by computational or by graphical methods. Computational methods usually involve iterative non-linear regression followed by some technique to establish 'goodness of fit'. Many computer programs have been described which carry out this procedure (see Ref. 6 for a review); even common statistical programs such as SASmay be used in this way.Analysis of residuals (i.e. the difference between observed data points and data points predicted by a particular model) has been used to differentiate between simple and complex ligand-binding models 7. Graphical procedures are simple and accessible to those without computers, but are more
