Effect of Carboxylate-Binding Mode on Metal Binding/Selectivity and Function in Proteins

Abstract / We delineate the factors governing the carboxylate-binding mode (monodentate vs bidentate) in metalloproteins. We reveal how the carboxylate-binding mode affects the binding affinity and selectivity of a metal ion as well as the function of a metalloprotein using Ca2+-binding proteins
and enzymes (ribonuclease H1, phosphoserine phosphatase) as examples. The collected data indicate that a carboxylate monodentate/bidentate switch, in addition to other structural factors, could be used to fine-tune the metal-binding site affinity and/or selectivity, thus modifying the function/properties of the metalloprotein.

Todor Dudev and Carmay Lim*

Institute of Biomedical Sciences, Academia Sinica

Introduction
Almost one-half of all known proteins contain metal cofactor(s), which perform a variety of tasks ranging from protein structure stabilization to enzyme catalysis, activating many essential life processes such as respiration and photosynthesis. Among the metal ions, Na, K, Mg, Ca, Zn, Cu, Fe, Co, and Mn are most frequently found to bind to proteins under physiological conditions. Because of the critical roles of these metal cofactors in protein function, many studies have been carried out to understand the factors governing metal binding and selectivity in metalloproteins. We recently summarized the fundamental princi2+ 2+ 2+ ples governing Mg , Ca , and Zn binding and selectivity in metalloproteins, while others reviewed the role of Li+, Cu2+, and Fe2+ in binding, selectivity and metal-induced folding in proteins. This account differs from previous reviews in summarizing recent experimental and theoretical evidence on how the carboxylate-binding mode (monodentate vs bidentate) of aspartate (Asp-) or glutamate (Glu-) amino acid residues

plays an important role in the binding affinity and/or selectivity of a metal cofactor and thus function of a metalloprotein. Aspartate and glutamate side chains are unique among the 20 amino acids in possessing a carboxylate group that can bind the metal cation via one of the carboxylate oxygen atoms (Figure 1a) or both oxygen atoms, forming a four-membered ring (Figure 1b). The carboxylate side chain can also bind to two metal ions by bridging them via an oxygen to each metal ion, forming a 'chelate ring'. We first delineate the factors governing the carboxylate-binding mode in proteins and show how these principles can be used to rationalize the observed differences in the carboxylate-bind2+ 2+ ing mode in Mg2 - and Ca proteins. We then reveal how the carboxylate-binding mode affects not only the binding affinity but also the selectivity of a metal ion. Finally, we show how the carboxylate-binding mode affects the function of a 2+ metalloprotein using Ca -binding proteins and several enzymes as examples.

53

Significant Reserach Achievements of Academia Sinica

Life Sciences

Institute of Biomedical Sciences

Fig 1Monodentate vs bidentate carboxylate binding to a metal cation. (a) Monodentate binding of Asp13 and Asp57 to hexacoordinated Mg2+ in bacterial chemotaxis protein CheY (1CHN, 1.76 ), where water molecules in the first shell stabilize the metal-free carboxylate oxygen atoms of Asp13 and Asp57. (b) Bidentate binding of Glu31 to heptacoordinated Ca2+ in site I of calmodulin (1OSA, 1.68 ).

The metal cation is relatively large and can accommodate bulky protein main chain/side chain dipoles that do not stabilize the second carboxylate oxygen or the metal cation. (5) The first- or second-shell ligands lack hydrogen-bond donors or provide only poor hydrogen-bond donors such as the peptide backbone that do not stabilize the second carboxylate oxygen or the metal cation. In addition to the competition between the metal cation and the first/secondshell residues for the second carboxylate oxygen, the relative rigidity of the metal-binding site structure can also affect the monodentate bidentate equilibrium.

Bidentate Carboxylate Binding Favors a Trivalent Lanthanide Cation over a Divalent Cation
The principles outlined above could help not only to rationalize the preferred carboxylatebinding mode found in different metalloproteins but also in the design of metal-binding sites that are specific for certain metal cations. For example, they could be used to design binding sites that are specific for trivalent lanthanide cations, 3+ Ln , instead of the natural divalent metal cofac2+ 2+ tors such as Mg and Ca . This is useful because alkaline-earth metal-binding sites in proteins 2+ (Ca -binding sites in particular) have few chemical properties that can be used to explore their biochemistry in situ. In contrast, luminescent lanthanide ions can be used in bioanalytical assays to determine the interdomain distance of proteins. The results (data not shown) show that 2+ bidentate carboxylate binding facilitates the Ca 3+ La exchange in fully or partially buried sites. The metal exchange free energy becomes more favorable with an increasing number of carboxy3+ lates bound bidentately to La . This is because, as compared to monodentately bound carboxylates, increasing the number of bidentately bound 3+ carboxylates to La increases the number of water molecules that are freed from the metal's first coordination sphere, which, in turn, results in a gain in the gas-phase entropy and solvation free energy.

Factors Governing the Carboxylate-Binding Mode to a Given Metal Cofactor in Proteins

The theoretical results suggest that the carboxylate-binding mode is determined mainly by competition between the metal cation, on one hand, and nonacidic neighboring ligands from the metal's inner or outer coordination sphere, on the other, for the second oxygen of the COO- moiety. Bidentate carboxylate binding is preferred over the monodentate mode when the second carboxylate oxygen's interactions with the metal cation are more favorable than those with first or second-shell ligand(s). This takes place when the following conditions occur. (1) The metal cation has a high coordination number (high steric repulsion among the ligands) and is thus more tolerant to the less space-demanding bidentate motif rather than the bulkier monodentate motif. (2) The metal cation is a good Lewis acid that can accept a charge from the second carboxylate oxygen. (3) The metal's positive charge is not neutralized by charge transfer from negatively charged ligands in the metal complex and can thus attract the second carboxylate oxygen. (4)

Significant Reserach Achievements of Academia Sinica

54

Life Sciences

Carbox ylate-Binding Mode Affects the Maximum Number of Metal-Bound Carboxylates

The number of metal-bound Asp - /Glu - , which determines the net charge of a carboxylate-rich metal-binding site, has been found to play an indirect role in enhancing the affinity and/or selectivity of a protein cavity for a given metal cofactor. For example, the three Asp /Glu 2+ side chains lining engineered EF-hand Ca -binding sites contribute a net ligand charge of -3, 2+ which helps to selectively bind divalent Ca against a much higher background concentration + + of monovalent cations such as Na and K . However, the maximum number of carboxylates that could bind to a metal ion of charge q, denoted by Max COO (M q + ), was not known. Furthermore, if and how the carboxylate-binding mode affects this upper limit was also not known. To evaluate if and how the carboxylateCOO q+ binding mode affects the Max (M ) in proteins, we computed the free energies for the successive exchange of metal-bound water molecules for carboxylates bound either monodentately or bidentately using a combined DFT and CDM approach. From the computed free energies the maximum number of carboxylates that could 2+ 2+ bind to the natural metal cofactors, Mg , Ca , 2+ and Zn , as well as non-natural metal cofactors, lithium (Li+), lanthanum (La3+), and zirconium (Zr4+), were determined. The findings from the calculations were validated by comparison with the number of Asp/Glu coordinated to univalent + + + + + 2+ 2+ (Li , Na , K , Rb , Cs ), divalent (Mg , Ca , 2+ 2+ 3+ Zn , Cd ), and trivalent (lanthanides, Ln ) metal ions in <3.0 X-ray/NMR structures in the PDB. Complexes of tetravalent metals such 4+ 4+ 4+ as Zr , Hf , or Ce were not found in any PDB structures if the metal complex were not additionally stabilized by interactions with outer-shell ligand(s) (Figure 2). On the basis of the combined results from the DFT/CDM calculations and the PDB survey, guidelines were derived for estimating the COO q+ Max (M ) in metalloproteins. The results sug-

Fig 2 Calculated free energies ( G 4) for the successive exchange of a metalbound water molecule for monodentately bound acetate in solvent inaccessible [M(H2O)m(CH3COO) n]q-n complexes (M = Li + , Zn 2+ , La 3+ , and Zr4+) as a function of the number of carboxylates in the product metal complex.

gest that the carboxylate-binding mode could affect the MaxCOO(Mq+). If all the carboxylates are bound monodentately to the metal cofactor in a fully/ partially buried protein cavity, a metal ion of charge q can bind no more than q + 1 Asp /Glu . However, if one or more acidic residues bind bidentately to the metal cofactor in a relaCOO q+ tively buried protein cavity the Max (M ) may be raised to q + 2.

Significant Reserach Achievements of Academia Sinica

Carboxylate-Binding Mode Can Control Protein Function

As the carboxylate-binding mode can affect the affinity and/or selectivity of a protein cavity for a given metal cofactor, it can also affect the protein's function. For example, ghe binding mode of a highly conserved Glu at the last posi2+ tion of the EF-hand Ca -binding loop plays a crucial role in discriminating between Ca 2+ (Figure 1b) is correlated with large conformation-

55

Life Sciences

Institute of Biomedical Sciences

Fig 3Monodentate vs bidentate carboxylate binding in substrate-bound phosphoserine phosphatase. (a) Monodentate binding of Asp11 to Mg2+ in PSP from Methanococcus jannaschii (1L7P, 1.90 ). (b) Bidentate binding of Asp20 to Ca 2+ in human PSP (1NNL, 1.53 ). The substrate is modeled in the active site and a ligating water molecule and Asp13/22 were omitted for clarity only.

al changes in the protein, which subsequently trigger a cascade of events along the signal transduction pathway. In contrast, monodentate binding of this Glu to a hexacoordinated Mg2+ results in a physiologically silent protein. The carboxylate-binding mode can not only regulate signal transduction but also abolish catalytic activity in some metalloenzymes such as E. coli ribonuclease H1 (RNase H1) and human phosphoserine phosphatase (PSP). Human PSP, 2+ which utilizes Mg as a cofactor, catalyzes the hydrolysis of phosphoserine to yield L-serine and inorganic phosphate. However, the enzymatic activity is abolished upon binding to Ca 2+, a potential competitor of Mg2+ in living cells. A comparison between the active-site X-ray structures of Mg2+-bound and Ca2+-bound PSP suggests that a switch in the carboxylate-binding mode of an essential Asp residue (from mono- to bidentate) contributes to the loss of catalytic activity. In the Mg 2+ -bound PSP from Methanococcus jannaschii, Asp11 is monoden2+ tately bound to Mg and its metal-free carboxylate oxygen can thus attack the phosphorus atom of the phosphoserine substrate (Figure 3a). In the 2+ Ca -bound human enzyme, the corresponding 2+ Asp20 is bidentately bound to Ca , thus preventing it from attacking the substrate, resulting in an inactive enzyme (Figure 3b). In analogy, bidentate binding of Asp70, a hypothesized general

Fig 4 Different Mg 2+- and Ca 2+-binding sites in RNase H1. (a) Mg 2+ bound to Asp10 and Glu48 monodentately (1RDD, 2.8 ). (b) Ca 2+ bound to Asp10 and Glu48 monodentately and additionally Asp70 bidentately

base in RNase H1 catalysis, to Ca2+ in addition to the protein ligands of the smaller native cofactor, Mg2+, enables Ca2+ to bind the enzyme tighter than Mg 2+ in vitro but prevents Asp70 from deprotonating a water nucleophile for subsequent phosphate attack, thus abolishing enzymatic activity (Figure 4). The results herein reveal that the carboxylate-binding mode plays an important role in the binding affinity and selectivity of a metal cofactor and thus function of a metalloprotein. Both theoretical and experimental results summarized herein indicate that a carboxylate monodentate/ bidentate switch, in addition to other structural factors, could be used to modulate the metalbinding site affinity/selectivity for a given metal cofactor, thus altering the function of the metalloprotein. Thus, switching the carboxylate-binding mode is potentially another design tool that could be employed to engineer new metal-binding sites with preprogrammed properties.
The original paper was published in Accounts of

Significant Reserach Achievements of Academia Sinica

Chemical Research 40.1 (2007): 85-93.

56