Can the propensity of protein crystallization be increased by using systematic screening with metals?

Abstract

Protein crystallization is one of the major bottlenecks in protein structure elucidation with new strategies being constantly developed to improve the chances of crystallization. Generally, well‐ordered epitopes possessing complementary surface and capable of producing stable inter‐protein interactions generate a regular three‐dimensional arrangement of protein molecules which eventually results in a crystal lattice. Metals, when used for crystallization, with their various coordination numbers and geometries, can generate such epitopes mediating protein oligomerization and/or establish crystal contacts. Some examples of metal‐mediated oligomerization and crystallization together with our experience on metal‐mediated crystallization of a putative rRNA methyltransferase from Sinorhizobium meliloti are presented. Analysis of crystal structures from protein data bank (PDB) using a non‐redundant data set with a 90% identity cutoff, reveals that around 67% of proteins contain at least one metal ion, with ∼14% containing combination of metal ions. Interestingly, metal containing conditions in most commercially available and popular crystallization kits generally contain only a single metal ion, with combinations of metals only in a very few conditions. Based on the results presented in this review, it appears that the crystallization screens need expansion with systematic screening of metal ions that could be crucial for stabilizing the protein structure or for establishing crystal contact and thereby aiding protein crystallization.

Introduction

The structural understanding of a protein plays a critical role in disseminating its biological function and is currently hindered by crystallization of proteins. The crystal structure determination of macromolecules has seen several advancements over the last two decades, with advent of sophisticated detectors, powerful synchrotrons and the advancement of computational methods.1, 2 In light of these progresses, it was suggested more than a decade ago that—to quote the words of the author—“the process of macromolecular crystal structure elucidation may become fully automatic”.1 However, crystallization of a protein is still a bottleneck. Protein crystallization has been known since early 19^th century and many strategies have been developed to increase the crystallization propensity.3, 4, 5, 6, 7 In addition to those discussed in the reviews mentioned, several other strategies have been attempted, for example, surface entropy reduction by either mutating highly entropic surface residues like lysines, glutamates, glutamines to alanine, threonine or tyrosine8 or by methylation of highly flexible lysine residues on the surface9; covalently linking target protein to a well‐studied crystallization friendly protein10; producing racemic mixture of proteins11, 12; removing the glycosylation on protein surface13, 14; co‐crystallizing target protein with nano‐bodies or with a Fab (fragment of antigen‐binding) of an antibody15, 16, 17; use of non‐protein nucleating agents (for e.g.18, 19, 20, 21, 22). Although, these methods require additional steps of introducing mutations, covalent modification or preparation of partner protein or an antibody or introduction of external agents, they have been very successful in crystallizing proteins that are difficult to crystallize. Despite these several strategies that have evolved over the years for protein crystallization, many protein crystals are still obtained by trial‐and‐error6 and based on the data from structural genomics consortia, it has been observed that only about 18% of purified proteins yield diffraction‐quality crystals.23 A recent article on analysis of crystallization conditions of ∼700 proteins from Structural Genomics Consortium at Oxford, reported that using three different screens had only marginal improvement in the overall success, over a single screen with multiple drops and with three different protein to precipitant ratio (2:1, 1:1, and 1:2).24 This is not surprising considering the fact that most conditions repeat across different screens or are slight variants of the same. These observations suggest that the quest for novel strategies and chemical agents for protein crystallization is a never‐ending endeavour.

Based on the analysis of systematically generated data on crystallization attempts at Northeast Structural Genomics Consortium, the authors conclude that the crystallization propensity is controlled by occurrence of well‐ordered surface epitopes that mediate inter‐protein interactions and is not strongly influenced by thermodynamic stability of the protein.25 They also suggest that even a well ordered and a stable protein can be recalcitrant to crystallization, if such epitopes capable of generating ordered three‐dimensional arrangement are absent. Crystal contacts are generally mediated by weaker non‐covalent interactions, and hence require complementary interface spanning several residues. In the absence of the above‐mentioned epitopes capable of mediating crystal contacts, metal ions present in the crystallization condition could aid the formation of crystal contacts by coordinating with residues on the protein surface. Since the metal‐ligand bonds are stronger but thermodynamically labile and highly directional as compared to the non‐covalent interactions that constitute protein‐protein interfaces, they are capable of stabilizing the crystal contacts, with limited number of properly positioned ligands on the protein surface, unlike the requirement for a patch of many interacting residues at a protein‐protein interface. In the subsequent sections we present examples of metal‐mediated crystallization/assembly and argue the case for the need of systematic screening with metals to increase the propensity of protein crystallization.

Metal Mediated Crystallization of Proteins

In general, metals are known to play an important role in oligomerization and crystallization of proteins. The ability of divalent metals to mediate protein crystallization has been observed long back26, 27, 28, 29 and was known before the birth of macromolecular crystallography. For example, the addition of CdSO₄ to horse spleen juice resulted in immediate crystallization of ferritin.30 Protein data bank (PDB31) contains many examples, where metals play critical role in the crystallization process. Here we present some of the interesting examples, for which clear analysis of role of metal in crystallization is available.

Crystallization of ligand bound form of leucine/isoleucine/valine‐binding protein (LIVBP) and leucine‐specific binding protein (LBP) highlights the role of divalent cations in crystallization.28 It took 17 years to obtain crystals of ligand bound form of LIVBP, after crystals of the apo LIVBP were obtained. After several attempts, they were able to obtain the crystals in condition containing organic precipitants such as 2‐methyl‐2,4‐pentanediol, ethanol or isopropanol. However, these crystals were not suitable for crystallographic analysis. Attempts to improve the quality of crystals by varying type and concentration of precipitating agents, protein concentration, type (leucine, isoleucine, or valine), and concentration of ligands, temperature, pH, and methods of crystallization met with failure. It was only when divalent cations were used were they able to obtained good quality crystals of LIVBP and also those of LBP.28 They also report the effects of different divalent cations on the crystallization of LIVBP bound with leucine and found variation in crystal morphology, quality and/or diffraction power, with the cation used. There have been other reports describing the influence of divalent metal ions in crystal formation via the mediation of interactions between adjacent molecules packed in the crystal,32 with examples like the oligomerization of human S100A12 in the presence of Ca²⁺ and Zn²⁺ 33 and of α‐synuclein in the presence of Cu²⁺.34 There have also been recent reports of a protein crystallizing in three different crystal forms with metal ions on the surface mediating crystal contacts.35, 36 Use of trivalent metal ions such as yttrium for the crystallization of β‐lactoglobulin has been reported by Zhang et al.37 In crystals of β‐lactoglobulin yttrium ions mediate crystal contacts through surface‐exposed glutamate and aspartate side chains. Metal ions were also used to direct protein self‐assembly (for example38, 39, 40, 41). A recent theoretical study together with experimental evidences argues how, ion‐mediated attractive patches drive the association of proteins and hence help in protein crystallization.42

Our experience with crystallization of a putative rRNA methyltransferase from the soil bacterium Sinorhizobium meliloti (Sm_Mtase) is also presented here. When Sm_Mtase, was screened with commercially available crystallization screens, of the 196 conditions screened, crystals were observed in only one condition containing 5 mM Cobalt(II) chloride hexahydrate (CoCl₂), 5 mM Nickel(II) chloride hexahydrate (NiCl₂), 5 mM Cadmium chloride hydrate (CdCl₂), 5 mM Magnesium chloride hexahydrate (MgCl₂), 0.1 M HEPES pH 7.5, 12% wt/vol PEG 3,350 [Fig. 1(A)], from the Index Screen (Hampton Research). Interestingly, apart from the buffer and PEG 3350, this condition has four divalent metal ions. The condition with four metals together seemed to be harsh for the protein as evidenced by excessive precipitation around the small crystals [Fig. 1(A)], suggesting that using too many metals in the same condition may not be appropriate. To test whether all four metals were required for crystallization, and to improve the quality of the crystals, in addition to the original condition, crystallization trials varying the number, combination of metals along with concentration of PEG 3350 were attempted. Crystals larger than those from the condition with all four metals, were observed in the condition with NiCl_2, CdCl₂ and MgCl₂ with 9% wt/vol PEG 3350 as precipitant [Fig. 1(B)]. Crystals did not appear in the absence of Mg²⁺, hence crystallization attempts were repeated by first incubating the protein with 100 mM MgCl₂ prior to crystallization trials with conditions containing 0.1 M HEPES pH 7.5, 9% wt/vol PEG 3,350 and one of five divalent metal ions (Ni²⁺, ${\text{Cd}}_{,}^{2+}$ Zn²⁺, Cu²⁺ or Co²⁺). Crystals were observed in all these conditions, although the quality of the crystals varied with the transition metal used in crystallization. Conditions containing Ni²⁺ or Co²⁺ yielded the best crystals in terms of appearance and size [Fig. 1(C,D)]. These experiments suggest that at least two metals, Mg²⁺ in combination with a transition metal are required for crystallization.

Figure 1.

Crystals of the putative rRNA methyltransferase from Sinorhizobium meliloti, (A) in the condition 5 mM CoCl₂, 5 mM NiCl₂, 5 mM CdCl₂, 5 mM MgCl₂, 0.1 M HEPES pH 7.5, 12% wt/vol PEG 3,350 (B) in the condition 5 mM NiCl₂, 5 mM CdCl₂, 5 mM MgCl₂, 0.1 M HEPES pH 7.5, 9% wt/vol PEG 3,350 (C) in the condition 5 mM CoCl₂, 0.1 M HEPES pH 7.5, 9% wt/vol PEG 3,350 and (D) in the condition 5 mM NiCl₂, 0.1 M HEPES pH 7.5, 9% wt/vol PEG 3,350. In (C) and (D) the protein was incubated with MgCl₂.The best crystals so far seem to be in the condition with 5 mM CoCl₂, 0.1 M HEPES pH 7.5, 9% wt/vol PEG 3,350 with MgCl₂ incubated protein.

Sm_Mtase is a S‐adenosyl‐L‐methionine (SAM)‐dependent methyltransferase composed of two domains; a smaller N‐terminal L30‐like RNA binding domain and a C‐terminal catalytic SPOUT domain.43, 44 Examining the crystal structure of Sm_Mtase revealed the C‐terminal SPOUT domain is involved in both intermolecular crystal contacts mediated by direct contact of surface residues from the adjacent SPOUT domain, as well as metal mediated interaction with the smaller L30 domain (Figs. 2 and 3). However, the L30 domain, which is relatively disordered, is not involved in any intermolecular interactions, directly via the protein chain, as clearly evidenced from the observation of empty space around that domain in figure 3. The only interaction of L30 domain is with an adjacent SPOUT domain from a symmetry related molecule, mediated through a metal ion (Fig. 3). From the analysis of crystal‐packing, it was clearly evident that the crystal could not pack without L30 domain, highlighting the important role of the divalent metals in crystallization of Sm_Mtase.

Figure 2.

Crystal contact mediated by the metal ion in Sm_Mtase. Residues H40 and E43 from one molecule and H129* from a symmetry related molecule coordinate with the metal forming intermolecular crystal contacts. 2F_o – F_c map is represented as a blue mesh and anomalous map as a red mesh; a different colour (yellow) is used for the protein chain from symmetry related molecule. The figure was generated using PyMOL.45

Figure 3.

2F_o – F_c density for Sm_Mtase after initial refinement. The L30 and SPOUT domains (marked with ellipses and indicated by labelled arrows), including a symmetry related SPOUT are involved in metal‐mediated interaction. The presence of anomalous peak between the L30 domain and an adjacent SPOUT domain from a symmetry related molecule, represented in red, clearly indicates that the metal involved in crystal contact is not Mg²⁺. 2F_o – F_c map is drawn at 1.5σ level and the anomalous map is drawn at 3.0σ level. The figure was generated using PyMOL.

Anomalous map (generated using FFT(CCP4)46, 47, 48) clearly shows the presence of a metal ion coordinating crystal contact between His40, Glu43 from one molecule with His129 of a symmetry related molecule (Fig. 2). The above said residues, in both molecules of the asymmetric unit, are involved in similar crystal contacts—metal coordination with His40, Glu43 residues of chain A mediate crystal contacts along crystallographic a‐axis and coordination with His40, Glu 43 residues of chain B mediate crystal contacts along crystallographic c‐axis. The association of dimers of this protein, along crystallographic b‐axis together with metal mediated contact in other two independent directions appear to help in the generation of 3D‐assembly.

It should be noted that when the protein was incubated with 5 mM Co²⁺ and Mg²⁺ and screened with first 48 conditions of Crystal Screen HT ((Hampton Research, USA) crystals with new morphology were observed in one of the conditions. Interestingly, initial screening using all the 98 conditions of Crystal Screen HT (Hampton Research, USA) did not produce any crystals, suggesting that metals could improve the chances of crystallization. Most interesting part of these experiments is that this protein crystallizes in a condition with a combination of metal ions. This kind of combination of metal ions, in this case, transition metal with an alkaline earth metal, is rare in most popular crystallization screens. Generally, Mg²⁺ is coordinated by the side chain carboxylate functionality of Asp and Glu, or main chain carbonyl oxygen atoms; the other metals in the crystallization cocktail, Ni²⁺, ${\text{Cd}}_{,}^{2+}$ Zn²⁺, Cu²⁺, and Co²⁺, are generally coordinated by His, Asp, Glu, and Cys.49, 50, 51 Similarly, few of these metals, for example Zn²⁺or Cd²⁺ can have their coordination fully satisfied by only cysteine residues. Hence, there are quantitative differences between Mg²⁺ and these other metals.

Although Mg²⁺ was required for the crystallization of Sm_Mtase, we were not able locate Mg²⁺ in the Sm_Mtase structure, probably due to highly disordered L30 domain or perhaps Mg²⁺ only plays a role in the initial nucleation of crystals. Similarly, for the crystallization of a membrane fusion protein ZneB from Cupriavidus metallidurans two metals were essential—zinc and one of the four ions Cd²⁺, Co²⁺, Ni²⁺, or Cu²⁺—highlighting the importance of a combination of metals for crystallization. Zinc is a structural metal in ZneB and it is observed in the crystal structure, however the second ion is not observed. Mass spectrometric analysis revealed that Cd²⁺, Co²⁺, Ni²⁺ do not interact with the protein and Cu²⁺ showed nonspecific binding to the protein. Similar to the case of Sm_Mtase the non‐structural metals perhaps aided in the nucleation of the crystals. It has to be noted that the protein to zinc ratio was about 1:2 (∼300 μM zinc) and the concentrations of the other metal ion is not mentioned for the case of ZneB.52

Structural Role of Metals in Proteins

As mentioned in the introduction, proteins with complementary epitopes on their surface capable of generating crystal contact are one of the important requirements in protein crystallization. Conformation of the protein in its apo from may be very different from when it is bound to a ligand or a cofactor or its cognate partner. It is known that about one third of proteins require metals for their activity or stability.53 In some cases, protein may be completely disordered in the absence of a metal co‐factor. Also, metals can induce conformational changes and stabilize protein structure, which in turn may expose novel crystallization epitopes on their surface and hence aid in the crystallization of the protein. A classic example of metal‐mediated stabilization is that of zinc finger proteins, implicated in various important cellular processes like gene regulation, signal transduction and so forth. Zinc plays a structural role in zinc finger domains, which though structurally diverse, are all stabilized by coordination with one or more zinc ions, usually with a characteristic pattern of cysteine and histidine residues.54, 55 Binding of either zinc or copper to the amyloid precursor protein (APP) E2 domain, induces large conformational changes, each metal stabilizing distinct conformations, of an otherwise flexible domain, converting the protein to different functional states.56 In human immunodeficiency virus type 1 (HIV‐1) integrase, two separate reversible metal‐induced conformational changes activate the enzyme.57 A striking case of metal induced conformational change is observed in metal fusion protein ZneB from Cupriavidus metallidurans.52 Zn²⁺ is essential for obtaining crystals of this protein and of the two molecules observed in the asymmetric unit, zinc is bound to only one molecule and is not seen in the other. The molecule with no metal bound is missing a domain of ∼50 amino acid residues, suggesting a disordered conformation of this domain, whereas the same domain is well ordered in the metal bound molecule (Fig. 4). APP E2 domain and ZneB crystallized only in the presence of metals thus suggesting the metal‐induced conformational change and stability were crucial for crystallization. Similarly, binding of copper induces conformational change in human Wilson's disease protein, a copper transporter. The N‐terminal domain of this protein has six metal‐binding sites in independent subdomains and conformational changes are observed in the entire N‐terminal domain on binding to copper.58 Another example of metal‐induced conformation change is that of an intrinsically disordered protein human metallothionein‐A—distinct conformations with varying degrees of orderliness are observed based on the number of metal ions bound to the protein, with upto seven metal ions per molecule required to induce a compact folded conformation.59 All the above examples highlight the role of metal either in stabilizing the protein or inducing distinct and stable conformational states of the proteins and hence aid in crystallization as discussed above.

Figure 4.

Metal mediated conformational change observed in the crystal structure of a membrane fusion protein ZneB from Cupriavidus metallidurans. Zinc is represented as a cyan sphere. Of the two molecules in the asymmetric unit one of them has metal bound (Chain A) and the other one does not (Chain B). One disordered domain is missing in Chain B and on binding to metal the same domain is ordered as seen in Chain A, illustrating metal‐mediated conformational change and stabilization. The figure was generated using PyMOL.

Metal‐Mediated Symmetrization of Proteins and Crystallization

It has been observed that symmetric (homooligomeric) proteins tend to crystallize more readily than asymmetric, monomeric proteins. Oligomers tend to crystallize in space groups that matches the internal symmetry of the oligomer.60, 61 Metal coordination with amino acid residues on the protein surface can be used to direct protein assembly and symmetric oligomerization, thus improving the chances of crystallization. A cytochrome cb562 variant with a bis‐His motif introduced via mutations resulted in different oligomeric forms, monomeric, dimeric and tetrameric, based on the concentration of zinc used.62 The symmetry of oligomerization is also governed by the metal coordination geometry of the specific metal used. Cu²⁺ induces a dimeric and Ni²⁺ induces a trimeric assembly of the modified cytochrome cb562 as opposed to the tetrameric assembly with Zn²⁺,38 indicating the supramolecular arrangement of the protein can be controlled by metal coordination obviating the need for large number of interacting residues on the surface.40 This and a few other examples are comprehensively reviewed by Salgado et al.40 A recently described crystallization strategy, termed metal‐mediated synthetic symmterization, inspired by the work of Salgado et al and their own study on synthetic symmetrization,61 utilizes rationally designed surface mutations that support the formation of metal‐mediated crystal contacts and bring about synthetic symmetrization that could facilitate crystallization63; T4 lysozyme and maltose binding protein (MBP) modified with judicious placement of pair of histidine or cysteine residues, produced crystals with more than ten isoforms for both the proteins. The proteins were cocrystallized in the presence of either Cu²⁺, Ni²⁺, or Zn²⁺ and crystallization was performed using 3–4 standard 96‐well format commercial crystallization screens. The double histidine mutants of both proteins could oligomerize through metal coordination. It has to be noted that not all possible crystal forms were characterized and there was a possibility of more isoforms of these proteins being produced through metal mediation. They also conclude that either the metal mediated symmetrization of proteins imparts these symmetrized oligomers to crystallize in different crystal forms or the metal mediated interaction between the monomeric or metal symmetrized oligomeric proteins can result in the generation of 3‐dimensional crystals. In about half of the structures instead of the mutated histidine one of the natural surface residues, usually Glu or Asp or water molecules were involved in metal binding, which begs the question, whether the metal mediated symmetrization would have occurred with the wild type protein itself. Unfortunately, this work did not describe the screening of these metals with respective apo proteins. It is also interesting to note that protein crystals were obtained at concentration lower than that is generally used for these proteins, suggesting that this strategy may be useful for proteins that can be overexpressed with low yield. Examples of metal mediated symmetrization can also be observed in protein structures available in protein data bank. For example, in the crystal structure of mouse C1QL1 globular domain (PDB entry 4D7Y) a cluster of nine negatively charged residues (Asp), which are otherwise expected to repel each other, come together due to their interaction with four metal ions in the interface and help in trimerization of the protein [Fig. 5(A)]. In this case three metals are assigned, Cd, Ni, Mg and only Cd and Ni are involved in oligomerization with no apparent role of Mg in either symmetrization or formation of crystal contacts. Similarly, crystal structure of glutathione S‐transferase‐III from Zea mays var. mutin (PDB entry 1AW9) illustrates both metal‐mediated symmetrization and formation of crystal contact [Fig. 5(B)]. Here Cd²⁺ ions mediate the formation of hexamers and crystal contacts in all three directions.

Figure 5.

Representative examples of metal‐mediated oligomerization and/or crystal contact formation from protein crystal structures in PDB: (A) Metal mediated symmetrization in the crystal structure of mouse C1QL1 globular domain (PDB entry 4D7Y). Three different metal ions are assigned in this structure, represented as colored spheres, Cd (red), Ni (grey) and Mg (magenta). The Cd and Ni ions are clearly seen to mediate the symmetrization and hence play role in the formation of crystals. (B) observation of both metal mediated symmetrization and crystal contact formation in the crystal structure of glutathione S‐transferase‐III from Zea mays var. mutin (PDB entry 1AW9). In this structure, Cd ions mediate both oligomerization (represented as yellow spheres) and crystal contact in all three directions (represented as red spheres). The figure was generated using PyMOL.

Prevalence of Metals in Protein Structures

Our observation that combination of metals was important for crystallization of Sm_Mtase led us to analyse the distribution of Mg, Ca, Mn, Zn, Ni, Co, Cu, Cd, Fe Na, and K, the commonly observed metals in protein structures. In many of the structures with metals in them, the metals were expected to aid crystallization either by stabilizing domains, promoting oligomerization or by mediating crystal contacts. Advanced Search option at http://www.rcsb.org was used with the following search queries with the requirement to match all the following conditions: Experimental Method—X‐ray, Molecule Type—Protein, Chemical ID—metal ID (for example, Mg). To search for the presence of a combination of two metals both were searched together. To rule out multiple structures of similar proteins non‐redundant data set with 90% identity cutoff was used. Table 1 represents the distribution of each of the metal atoms and that of a combination of two each of these metals. As of 2.09.2015 there were 93263 protein structures in the PDB, solved using X‐ray crystallography and 28243 structures in the non‐redundant data set with 90% identity cutoff. Only those 28,243 structures are considered in this study. Of these, 18,815 structures contain at least one metal (Table 1, sum of the values on the diagonal), hence ∼67% of the protein structures contain at least one metal in them. Among these, 4015 structures contain a combination of at least two metals (Table 1, sum of the off‐diagonal elements), which is 14.2% of the total number of structures. In many of these structures the metal is acquired from the crystallization condition and could be playing a role in crystallization, for example, of the 439 structures with Cd²⁺ in them, in ∼80% the metal was found in the crystallization condition. Examples, from PDB, where metals play role in crystallization by either stabilizing multimeric domains (metal mediated symmetrization) and/or forming crystal contacts are provided in the previous section (Fig. 5).

Table 1.

Distribution of Commonly Observed Metals in Protein Structures

Structures with metals	Mg	Ca	Mn	Zn	Ni	Co	Cu	Cd	Fe	Na	K
Mg	5002	199	81	673	48	38	16	104	38	458	278
Ca		2957	67	233	39	14	36	22	33	287	49
Mn			1033	54	6	0	3	5	18	121	28
Zn				4126	37	8	47	19	44	245	116
Ni					514	2	3	11	7	42	7
Co						310	1	8	2	23	11
Cu							274	6	4	30	4
Cd								439	3	100	75
Fe									433	33	9
Na										2774	169
K											953

The diagonal elements (shown in bold) represent the number of occurrences of each individual metal and the off‐diagonal elements represent the number of times a combination of the metals in the corresponding row and column occur in protein structures.

Systematic Screening with Metals as a New Protein Crystallization Strategy

One major approach to protein crystallization is the use of premixed cocktails in the form of commercially available crystallization screens. A recent survey of the successful crystallization conditions as reported in the PDB analysed 256 commercial screens amounting to 15906 conditions and found that there is on average a twofold redundancy in the commercial screens with over 100 conditions found more than 10 times.64 Despite the clear evidence of the role of metals in crystallization the use of metals to introduce crystal contacts and/or for oligomerization is not considered a common strategy in protein crystallization.7 It is known that the overall propensity for the metal ligands in proteins, Asp, Glu, Cys and His, together, adds up to ∼15%,65 and most of them being hydrophilic residues, their occurrence on the surface is substantial. Moreover, the involvement of main‐chain carbonyl moieties and residues such as Asn and Gln in metal coordination is frequently observed in proteins.49, 51 Hence, there is a possibility that the interaction of appropriate metal with these surface exposed ligands, can generate the interaction epitope that can result in three‐dimensional order without the deliberate modifications of proteins described in the previous section. This is also supported by our experience with Sm_Mtase and the examples of metal mediated crystallization and assembly presented here. These observations suggest that more conditions with different combinations of metals in them should be incorporated into crystallization screens to increase the success rate of metal mediated protein crystallization. However, among the crystallization conditions provided in several popular commercially available crystallization screens there are only some conditions with transition metals in them and the metal concentrations used are usually too high (as high as 100 mM) which could lead to protein precipitation in most cases, with only a few conditions containing metals at lower concentration and only very few contain combination of several metals.

Since, the coordination geometries of each metal and their preference for ligands in proteins vary, we are of the opinion that a much more systematic screening with metals, solo or in different combinations, together with pH and other variables would lead to the exploration of different possibilities of oligomerization and/or crystal contact formation which could translate to a higher success rate with metal mediated protein crystallization. Initial crystals of Sm_Mtase were obtained from a cocktail of four metals (Cd, Ni, Co and Mg) which is unique in classic screens. While it is tempting to propose the use of a cocktail of several metals, hoping that the protein would pick up an appropriate metal, it could be a case of too many metals in the broth spoiling the experiment. Due to the presence of several metal binding side chains on the surface, simultaneous binding of multiple metal ions on the protein surface could lead to the formation of heterogeneous mixtures of metal‐cross‐linked protein chains, and thus precipitation of the protein. This is indeed observed for Sm_Mtase—in the presence of four metals there was excessive precipitation of Sm_Mtase with only small crystals in the drop [Fig. 1(A)]. Better crystals were obtained in a condition with one transition metal and one alkali metal [Fig. 1(C,D)] but the crystal and diffraction quality varied with the transition metal used. This has also been observed with crystallization of LIVBP—different divalent cations yielded crystals of varying quality.28 Thus, while cocktails of metals could be useful in finding an initial hit, a better strategy would be to develop screens with individual transition metals in each condition or a combination of two transition metals or a combination of a transition metal and alkali metal. However, it has to be noted that high concentration of metals, especially transition metals can induce the precipitation of proteins and it would be appropriate to use low concentration of metals in crystallization trials. As reported by Lagonowsky et al.63 they had success in crystallizing the proteins having bis‐His motifs using a protein to metal ratio of 1:1.25–1.5. In the case of LIVBP protein crystallization, critical upper limit to cadmium concentration for formation of good quality hexagonal crystals was around 1 mM. At 3 mM concentration of cadmium, rod/needle shaped crystals were obtained. Interestingly, no crystals were obtained at 20 mM Cd²⁺. However, in the case of LBP higher concentrations of Cd²⁺ were required.28 Considering that protein concentrations used for crystallization are usually in the range of 1 mM (used at typical concentrations of 10–20 mg/ml), a metal concentration of 3 mM in the screen solutions (amounting to about 1.5 mM concentration in the drop containing protein) should suffice to begin with, for crystallization trials. However, a systematic study is required to understand the parameter such as metal concentration, concentration of the precipitants and so forth.

Another important advantage of use of metals in crystallization is that metal‐ligand interaction being stronger and directional, results in the generation of ordered contacts which could lead to better diffraction quality. As reported in Trakhanov and Quiocho,28 crystals of LIVBP bound to leucine was initially obtained only in the presence of divalent cations but eventually they were able to obtain crystals of the ligand bound protein in the absence of divalent cations. However, the crystals obtained without the divalent cations diffracted to a lower resolution than those grown in the presence of the cations. Similarly, as reported in Quistgaard et al.,36 severe anisotropy in data was observed with better diffraction in the direction where metal‐mediated crystal contacts were present. Co‐crystallization of E2 domain of amyloid precursor protein with metal ions resulted in improved diffraction quality and lower overall B‐factor.56 Similarly, in the case of Sm_Mtase, we obtained best quality crystals in conditions containing either Co²⁺ or Ni²⁺ and Mg²⁺. Crystals obtained with Ni²⁺ in the condition look better morphologically [Fig. 1(C,D)], however crystals obtained with Co²⁺ in the condition diffracted better. Hence, in addition to facilitating crystallization use of systematic metal screening could also result in better quality of X‐ray diffraction data. In addition to improving the quality of X‐ray diffraction data, the use of metals in crystallization could aid in structure determination. Weak anomalous signals from calcium, in routine datasets, has been exploited in determining protein structures.66 Hence, structure determination utilizing the anomalous signals from these metals, can be attempted in cases where there is no suitable model available for molecular replacement.

Conclusions

Metals, have been observed to be crucial in oligomerization and/or formation of crystal contacts thus aiding in crystallization of proteins. Also, proteins utilizing metal cofactors may produce superior crystals in the presence of the cognate metal and in some cases crystallization may be fully dependent upon the presence of the metal, due to metal‐dependent stabilization of the protein. Furthermore, additional metal(s) together with metal cofactor may be required to facilitate crystallization, when involvement of metal coordinated crystal contact is also crucial for crystal growth via the introduction of new crystal contacts. While it is common to use known metal cofactors in crystallization trials, the use of combination of metals is not a general strategy for the crystallization of proteins. The Sm_Mtase described in this report appears to have no metal dependency for its activity and the opportunity to obtain crystals would have been missed if not for the rare combination of metals and other parameters such as pH and precipitant. Interestingly, combinations of metals are infrequently present in commercially available crystallization kits. Various examples from literature and the results presented here highlight the importance of considering crystallization strategies involving systematic screening with different metals and combinations of metals. We believe, methods such as ligand based screening, crystallization in space and many other methods explained in the introduction, that require modification of proteins or help of an antibody and so forth, while useful, metal centric screening appears to be an economic and accessible strategy with wider applicability.