Several calcium-binding motifs have been characterized for proteins; especially those with regu- latory roles. The structural chemistry of EF hands and the coordination chemistry of calcium and other spherical metal ions, and related models, have been reviewed by Falke.14
Frequently, Ca2þ binding to such motifs resulted in perturbation of bonding interactions between the two principal elements of secondary structure. Such perturbations underlie the cooperativity often observed in proteins carrying such motifs. The best known is the so-called EF hand motif, first identified in carp parvalbumin,15and named after the E and F alpha helices of the structure that adopt a structural form resembling the thumb and index finger of a hand.
The EF hand captures Ca2þ in a loop region between two helices, with the side chains of residues 1, 3, 5, and 9 within the loop, residue 12 in the following helix, and the carbonyl of residue 7 making contacts (Figure 1).16,17 Residue 9 is often in the outer sphere and stabilizes a Ca2þ-bound H2O. Based on the structural understanding of calcium ion binding to EF hands, the binding affinity and metal ion selectivity have been tuned through rational amino acid substitu- tions.18–21
A-hairpin motif shows residues 1–7 in well-conserved structural positions relative to the EF hand (Figure 1). Residue 9 typically moves and delivers two oxygens from a carboxylate to take up the site previously taken by residue 12. Divalent calcium can also be bound by a structurally similar loop bound by an helix andstrand. Overall, the residues 1, 3, 5, and 9 are conserved in all of these structures and appear sufficient to define the Ca2þbinding motif. Putative Ca2þbinding sequences in integrins resemble those found in EF hands16and have been termed ‘‘EF hand-like’’ (Figure 1). The helix/strand defined motif common to integrins also lacks the need for residue 12.
No distinct structural motif is used by calcium-dependent nucleases such as staphylococcal nuclease.22Since these enzymes are extracellular, where Ca2þconcentrations are high, the binding affinities are normally low and in the millimolar range (Table 1).
8.5.3.2 Magnesium
For alkali and alkaline earth ions the most common protein ligands are the carboxylate groups in glutamate and aspartate residues. Figure 2 summarizes possible binding modes.23,24 Larger cations favor the syn configuration with anti only identified for smaller ions with MO distances
<1.98 A˚, and for very large ions such as Csþ.25 Alkali and alkaline earths, unlike transition metals, also bind the carboxylate out of plane. Some metal ions with MO distances in the range 2.3–2.8 A˚ may bind in the bidentate mode, including Ca2þ.25
Computational methods have been brought to bear on the question of what controls coordin- ation mode and geometry. Results from computational studies of crystallographically defined coordination complexes are compared with experimental structural data and further compared with similar structural configurations in protein structures. The assumption that gas-phase calculations might mimic protein environments has been tested on model complexes and generally found to be valid.23,24,26
(a) (b) (c)
(d)
(e) C-terminal
helix
N-terminal helix
Figure 1 Structural binding motifs for Ca2þ. (a)–(c) show the-strand motif, EF-hand motif, and helix turn motif, respectively. (d) shows a stereodiagram of the calcium binding domains in (a) and (b). (e) illustrates the ligand sites noted in the text. Backbone atoms are shown in yellow for (a) and green for (b) (reproduced
with permission from Springeret al.16and Reidet al.17).
Binuclear metal centers are commonly found for transition metal-containing systems.27Several have been cited for magnesium7and even calcium ion.28The relevance of these crystallographic- ally defined ions to the functional state in solution has been debated;7,29,30 however, in this chapter we shall focus simply on the details of coordination chemistry (Figure 2). The close proximity of two divalent metal ions in a low dielectric medium, such as a protein, requires the close proximity of negatively charged groups to alleviate the electrostatic repulsion. The occur- rence of a bridging carboxylate or carboxylates(s) yields an inter-metal distance of5.2 A˚;
however, an additional bridging hydroxide or phosphate oxygen can reduce the inter-site distance to3.3–3.6 A˚.24Typically when a single carboxylate bridges magnesium the metal lies in an anti orientation relative to the carboxylate. When two carboxylates bridge, the geometry is syn.24
Certain proteins may use two metal ions in close proximity to maintain a preferred shape, or assist in enzyme catalysis.7,27As noted previously, metal–ligand interactions are structurally more precise than H bonds, and so their use better assures a specific structural orientation that may be required for protein function. Such an idea may be extended to all of the examples described in this chapter. There has been some effort to synthetically model such binuclear magnesium centers.
For example, three novel carboxylate-bridged dimagnesium(II) complexes have been reported that use a ligand design based onm-xylenediamine bis(Kemp’s triacid imide) as the binucleating ligand.31 The crystallographically determined structure of the carboxylate complex shows a MgMg distance of 4.783(2) A˚. Reaction with sodium diphenyl phosphate yielded two new compounds. One contained the binuclear core with a bridging bidentate phosphate ligand and a metal–metal distance of 4.240(5) A˚, while the other showed one bridging and one terminal phosphate ligand with a metal–metal distance of 4.108(3) A˚.31The ligand exchange rates for the binuclear magnesium complexes were found to be two orders of magnitude slower relative to exchange results for the equivalent dinuclear zinc complex, consistent with the relative exchange rates for the individual Mg2þand Zn2þions.
5.2Å 5.2Å
M
H
H H
H
M
M M
M M M
M
IIe91
W23
W24
W25 W127
W204 W269 W240
Glu45 Mn504
Mn501 Asp74
O
O
O O
O O
O O O
O O
O
O R
R
R R
C C
C
C
3.7Å
3.2Å –
–
–
– C C
C
C O O
O
H
(a)
(b)
H H
H H
H
H H
O
O O
O O
O O
syn anti btdentate
O
O
O Mn+
Mn+
Mg2+
Mg2+
Mn+
Figure 2 Magnesium binding motifs on proteins illustrated for one metal (a) and two metal (b) binding sites. The proposed binuclear site fromE. coliEcoRV is illustrated in (b) (reproduced with permission from
Katzet al.23and Gluskeret al.24).
The role of Mg2þ and Ca2þin biological catalysis commonly involves mediating hydrolysis of phosphate esters.4,7,27To this end the divalent ion often contains a bound H2O that is then located proximally to the bound ester function (the so-called template effect). A second proposed role for the divalent metal ion is to promote ionization of the bound H2O.23It is well recognized that in completely hydrated ions, charged metal ions lower the acidity constant (pKa) of bound H2O. For Mg(H2O)62þ
the energy to remove a proton is 40% of that required to deprotonate bulk H2O (the pKa is lowered).23Commonly, however, alkali and alkaline earths are bound to proteins by carboxylates.
It has been shown that replacement of a bound H2O by carboxylate increases the energy to depro- tonate an Mg2þ-bound H2O by80 kcal/mole1, making it intermediate between the bulk water and fully hydrated Mg2þ(aq).23This change appears to originate entirely from electrostatics rather than the direct perturbation of the electronic structure of the metal ion. Further substitution by additional carboxylates further increases the pKa with clear implications for enzymes that carry such motifs (Figure 3) where ionization of bound water to form hydroxide is often invoked.
Computational studies of magnesium binding to proteins have established that the dielectric properties of the solvent medium play a critical role in defining whether magnesium binds in an inner or outer sphere fashion.26When"4, carboxylate may replace water, although additional stabilizing interactions are required for the carboxylate if deprotonation of the acid is required before coordination. This also rationalizes the observed outer sphere coordination preferred by divalent magnesium to nucleic acids32,33 since the binding sites tend to be extensively solvent accessible with a higher dielectric constant. Coordination of charged carboxylate is strongly preferred relative to neutral acids, amides, or alcohol ligands. Stabilization of bound carboxylate stems from electrostatic factors and hydrogen bonding from the Mg-bound carboxylate to either water or H-bond acceptor groups bound to Mg2þ. Electrostatic demands dictate a maximum of four carboxylate ligands to magnesium. Additional carboxylate may bind only if there is an additional metal cofactor to bridge to, or if the ligand binds in the neutral protonated form. The latter is quite feasible since the grouping of several charged carboxylates tends to increase the effective pKaof one of them. Water is the favored co-ligand for carboxyl ligated magnesium as a result of its small size and hydrogen bonding capacity. It was concluded also that magnesium binding to proteins most likely occurs stepwise with an initially hydrated metal entering the pocket with subsequent stepwise replacement of inner sphere waters.
Proteins typically bind metal ions with binding affinities (KD) that match physiological avail- ability.3,7,33 In the case of intracellular free Ca2þ this is sub-micromolar, and so such Ca2þ
catechol O-methyltransferase binds two OH of catechol 2(3) available adjacent binding sites
ASV integrase
2(4) available adjacent binding sites (W)
bacterial luciferase
3(4) available adjacent binding sites (W)
TNF RNase from T4 bacteriophage binds two OH of ribose
4 available adjacent binding sites (W) CheY
1(3) available binding sites (W) mandelate racemase binds COO-and OH of mandelate 2(3) available adjacent binding sites Asp169
Asp64
Asp132 Asp346
Asp121
Asp54 Asp170
Asp141 Asp195
Mg2+
Mg2+
Mg2+ Mg2+
Mg2+
Mg2+
C C
C
C C
C C
C
C
C
Glu247 Glu221C W W
W
W
W W
W W
W W
W W
W W
W
W W
W
X X X
mNO2
subtrate X I
I I
I
I I
I C O––
O C–– subtrate
active site acti
ve site
phosphorylated oNO2
bkbn Asn59 acti
ve site
Glu237
Figure 3 Coordination modes for magnesium ion binding to a variety ofE. coliproteins (reproduced with permission from Katzet al.23).
binding proteins show KD1mM, while extracellular Ca2þ proteins show KD>1 mM, again reflecting the higher extracellular concentrations of this cation. The intracellular concentration of free Mg2þ is 0.5 mM, and so KD’s typically lie in this range. Despite this similarity inKD, many metal binding proteins, and magnesium-dependent enzymes in particular, show a rich variety of coordination motifs. These variations tailor theKDas described, but also accommodate the diverse functional roles of the metal cofactor. In this regard, the availability of high-resolution structural data on several metal-dependent nuclease enzymes allows a critical analysis of the coordination chemistry of the bound cofactor and implications for the functional mechanism.7 There is an apparent and interesting relationship between the requirements for metal-bound solvent interactions with the substrate and the coordination number with protein ligands, which ultimately controls the hydration state of the bound metal.34
Metal binding pockets of surface magnesium sites typically show poor selectivity and can accom- modate a large variation in metal size.35,36Such relationships have been investigated intensively by Falke and co-workers, who have shown that such sites show strong charge selectivity, but weak size selectivity.37A comparison of calcium binding EF hands versus magnesium-dependent enzymes that Mg2þ and Ca2þ sites exhibit similar charge selectivity, but the CaII site is highly size-selective, preferring divalent and trivalent ions with radii similar to that of CaII. A structural comparison of the Mg2þand Ca2þsites suggested that these distinct size specificities stem from fundamentally different coordination schemes. The Ca2þ site surrounds the bound ion with a pentagonal bipyramidal array of seven protein oxygens, thereby fixing the coordination number and control- ling the radius of the substrate cavity, while the flexible nature of the Mg2þsite was proposed to stem from its use of solvent oxygens to coordinate one hemisphere of the bound ion with variable coordination numbers and radius resulting to accommodate substrate ions of different size.
The metal binding pockets for the magnesium-dependent enzymes (several are summarized in Figure 3) differ markedly. The coordination environment and charge density in the metal binding pockets are distinct. The number of carboxylate ligands varies from none to three (additional carboxylates may derive from bound substrate, such as the case for mandelate racemase). Never- theless, the binding affinity for Mg2þis similar (0.5 mM) for each of these magnesium binding proteins. How does this arise, and why should it be so?
It appears that the solvation state of the metal cofactor is of critical importance for the proper functioning of these enzymes. The solvation state in turn is defined by the coordination state.
Note that in the case of RNase H, not only the number of bound H2O, but also the facial geometry, was found to be required for catalytic activity.38 Since the metal binding affinities of metal-dependent nucleases are tuned to the normal physiological concentration, Nature has evolved a structural mechanism that permits permutation of the ligand environment of the essential metal cofactor, while optimizing the binding affinity to physiological requirements. In particular, the metal binding pockets of metal-dependent nucleases have evolved to allow a large variation in the number of protein ligands (and thereby solvent water), while optimizing the binding affinity to physiological requirements.
For many metallonucleases the metal cofactor appears to activate substrate by outer-sphere stabilization of the transition state. This is consistent with crystallographic precedent33,39based on the ground-state binding geometry of Mg2þ–oligonucleotide complexes. Experimental data clearly show that transition state stabilization is effected by hydrogen bonding with essentially no contribution from electrostatic stabilization through the charge on the cofactor.38