Glutamyl-glutamate – a tailor-made chelating ligand for the [Be₄O]⁶⁺ core in basic beryllium complexes and implications on investigations on the origins of chronic beryllium disease

1 Introduction

This work is a result of our continued efforts to investigate the coordination chemistry of the Be²⁺ cation both experimentally and theoretically [1–6]. The so-called basic beryllium salts are among the best-known beryllium compounds, and their structural chemistry is part of textbook knowledge in inorganic chemistry. Owing to this, and probably to their inertness, these compounds nowadays are not the subject of any intense research activities. They are not soluble in aqueous solutions and do not react with Brønstedt acids or bases. As they are not formed directly from aqueous solutions without the presence of organic solvents, and the isolated O^2– is an exceedingly strong proton acceptor, it appears likely that these highly symmetric complexes are only meta-stable with respect to H₂O, H₃O⁺, and OH^–, due to a kinetic hindrance of their decomposition, similar to the well-known case of SF₆versus H₂O.

In this light, the accessibility of a recently found mono-hydroxo derivate, Be₄O(RCO₂)₅OH is quite remarkable [4]. In this compound, the OH^– group replaces one carboxylate group and coordinates in µ²-mode to two beryllium atoms. It is the first described heteroleptic derivate of this compound class that contains less than six carboxylate functions. As the central Be₄O moiety shows significant deviations from the ideal tetrahedral symmetry and other bonding parameters also show deviations from ideal values, higher degrees of carboxylate substitution by OH^– appear unlikely or at least strongly unfavored.

Owing to the inertness of the basic beryllium compounds, it seems a purely academic question how this stability can still be enhanced, but biological considerations discussed in this work highlight its practical relevance.

2 Results and discussion

The stability of coordination compounds depends both on the electronic interaction between their ligands and coordination centers and on entropic influences. The latter are the basis for the well-known chelate effect. Thus, the question arose if there are suitable chelating ligands containing two carboxylate functions for the [Be₄O]⁶⁺ coordination unit that would lead to an even enhanced stability of these compounds. Geometrical considerations guided by molecular dynamics calculations have shown that an alkyl chain with about seven to eight saturated carbon atoms, i.e. -heptanediyl- or octanediyl, is sufficient in achieving an unstrained bridging. Instead of focusing on 1,7-bis(carboxy)heptane or 1,8-bis(carboxy)octane, we were interested in structurally related potential ligands of biological relevance. In this way, the glutamyl-glutamate dipeptide caught our attention. More specifically, we investigated its biologically relevant diastereomer l-glutamyl-l-glutamate (EE, see Fig. 1) as a potential chelating ligand for a [Be₄O]⁶⁺ aggregate.

Fig. 1:

The l-glutamyl-l-glutamate dianion EE^2–.

We have optimized the structures of four Be₄O(OAc)_6–2n (EE)_n complexes (n = 0,1,2,3) at the density functional theory (DFT) level (using Becke’s three-parameter functional in combination with the Lee-Yang-Parr exchange-correlation functional [B3LYP]-D3/triple-ζ quality basis sets augmented with two polarization functions [def2-TZVPP]; see the Methods section for further information and Fig. 2 for a depiction of their molecular structure).

Fig. 2:

The molecular structures of four basic beryllium carboxylate complexes as calculated at the DFT level. Oxygen (red), beryllium (green), nitrogen (blue), carbon (black), and hydrogen (white) atoms are displayed in different colors. (a) [Be₄O(OAc)₆], (b) [Be₄O(OAc)₄EE], (c) [Be₄O(OAc)₂(EE)₂], and (d) [Be₄O(EE)₃].

From a structural point of view, all four complexes are similar, and the transition from the homoleptic hexaacetates via heteroleptic acetate/EE^2– complexes to the homoleptic tris-EE^2– complex is not connected with large changes in coordination geometries. However, a closer inspection of the structural parameters (see Table 1) reveals some systematic changes. With an increasing degree of EE^2– substitution, a slight but significant expansion of the central Be₄O core occurs. The effect is reflected in an increase in the average Be–O1 distance, a tendency also observed in the Be···Be distances. At the same time, remarkably, the deviation from an ideal tetrahedral symmetry of the Be₄ tetrahedron gradually decreases (compare the variation within Be–O1 and within Be···Be distances). In contrast to the slight expansion of the core, the distances from Be to ligand oxygen atoms slightly decrease. A comparison of the angles between carboxylate carbon atoms (C1 to C6) and the unbound central oxygen atom O1 shows the perhaps most notable change in structure with increasing number of EE^2– ligands. It is a distinct trend toward an ideal octahedral arrangement of the C–C bonds of the six carboxylate groups with respect to the Be₄O core. The implications of these structural changes with respect to the stability of the complexes are discussed below.

Using computational data, we have calculated the energy balance of the substitution reactions starting from basic beryllium acetate, yielding in three subsequent steps the basic beryllium tris-l-glutamyl-l-glutamate:

The energy balance of these three reactions is summarized in Table 2.

Hence, each of the substitution steps of the two acetate anions for one EE^2– dianion yields about 10 kcal mol^–1 in electronic energy and is favored entropically. A primary conclusion is that the coordinative interaction between glutamate and Be²⁺ is significantly stronger than the interaction between acetate and Be²⁺ ions. In this respect, the Be²⁺ ion closely resembles a proton – an analogy based on the similar hardness of both cations, which has been noted and discussed previously [7] and for which one is tempted to use the term Schrägbeziehung (German for “diagonal relation”) in the periodic table of elements – and HOAc is a stronger Brønsted acid than glutamic acid (implying weaker cation base interactions). As it appears that the core Be₄O unit in Be₄O(OAc)₆, as discussed in [4], is somewhat too small to fulfill the geometrical requirements for an ideal µ²-carboxylate coordination, deviations from T_d or T symmetry are observed in these complexes. The stronger glutamate-Be²⁺ interaction weakens the interactions with the central oxygen atom to some extent, leading to an expansion of the Be₄O core and to a symmetrization of the whole complex. Hence, the symmetrization of the molecule is a consequence of the enhanced interaction between the Be²⁺ and the glutamate ions as compared to acetate. As expected, the chelating effect also contributes to a lowering of the free energy (Table 2) by affecting the entropy balance of the reaction. The chelating effect apparently is large enough to compensate for the generation of an additional stoichiometric equivalent in the reactions.

In the following, we attempt a speculative discussion on the possible implications of these findings on the still unknown molecular basis of the chronic beryllium disease (CBD). Recently, Clayton et al. [8] confirmed via protein crystallography the hypothesis [9, 10] that Be²⁺ ions bind to a complex of the major histocompatibility complex MHCII with a protein at a site where five carboxylate containing residues (-Glu-, -Glu-Glu-, -Asp-Leu-Phe-Glu-) from different peptide chains are accessible in a very small spatial region. In our opinion, the suggested coordination center with one Be²⁺ ion, a close-by Na⁺ ion, and three “crystallographic” water molecules in the center region of the five glutamic acid residues [8] is not a convincing structure model. Analytic evidence for a sodium ion at this position is lacking and the model does not explain in how far this aggregate could be as specific a toxin as it apparently is, operative in CBD. In the light of our findings, we suggest to test the possibility of a Be₄O coordination center in the structure model. As indicated by preliminary DFT calculations on a small section of the protein-beryllium complex of about 550 atoms, there is enough structural flexibility and a large enough cavity to form a Be₄O complex with the four glutamic acid and the aspartic acid residues. The central residual electron density Clayton et al. found in this cavity and have fitted with a Na⁺ ion could well be the central oxygen atom of a Be₄O moiety. Interestingly, as our calculations show, the sixth ligand position is filled by the amido-carbonyle function of the mentioned -Leu- residue, coordinating in µ²-mode. As we have pointed out in the introduction, such a heteroleptic coordination motive is not unprecedented in the chemistry of basic beryllium compounds [4].

This molecular model for the key protein complex in CBD implies (a) an irreversibility of the complex formation under physiological conditions, (b) the distortive impact of the beryllium on the protein structure, as the Be₄O strongly directs the carboxylate functions into octahedral arrangement and also affords their spatial close approximation, and (c) the necessity to carry the MHCII allele HLA-DP2, which has a glutamic acid residue at the position 69 of the ² chain. Otherwise, there would be a lysine residue that forms a hydrogen bond with one of the four remaining carboxylate functions, which would leave no possibility for the formation of a stable complex with the [Be₄O]⁶⁺. We see these points not explained by the current structure model [8].

Open questions remain also in this model: why do exclusively Be²⁺ cations bind so specifically at this protein site and how can the Be₄O-protein complex be formed under physiological conditions? To tentatively answer the former question, especially the alternative Zn²⁺ and Co²⁺ cations, which are known to form complexes, M₄OX₆ isostructural with basic beryllium compounds have to be regarded. We note that the M₄O cores of these metal cations would require a significantly larger cavity than the Be₄O core in the protein. For example, a metal–oxygen distance to the central oxygen atom in basic zinc o-tolylcarboxylate is 1.937 Å, and in an isostructural cobalt complex, we find a metal–oxygen distance of 1.932 Å, as compared to 1.666 Å in the basic beryllium acetate (found in CSD [11] entries BOHYIH, REHQOK, and BEOACT). Another aspect making the basic beryllium compounds a unique compound class is that only Be can practically not extend its coordination number (CN) from four to higher coordination, unlike Zn and Co, which appears to be a necessary requirement for possible decomposition processes, where the metal cation needs to undergo nucleophilic attacks. Oxygen-centered carboxylate- or carbamate-metal complexes with similar bonding motifs are known in Pb, Fe, Cu, and Ni (e.g. CSD [11] entries DOCNOZ, POFYAK, VULCAH, WALHAS); however, in all of these cases, the metal cation already has a CN higher than four, so the possibility for a specific bonding of these metals at the MHCII site where Be binds can be excluded.

With regard to the question of how the Be protein complex can be formed under physiological conditions, a slightly acidic pH (about 5.0) for the lysosome, the cell compartment where the MHCII–protein–beryllium complex is formed, can be assumed. Under these conditions, Be²⁺ is soluble and present in the form of hydrated Be²⁺ ions or multinuclear species like the ring-shaped {[(H₂O)₂BeOH]₃}³⁺ cation [12, 13], which could potentially act as a physiologically mobile form able to transport at least three beryllium atoms at once. However, further research has to be carried out to obtain definite answers these questions.

3 Conclusions

Based on our theoretical considerations, we conclude that, owing to both enhanced coordinative interactions and entropy reasons, basic beryllium glutamyl-glutamate complexes are significantly more stable than basic beryllium complexes with stronger Brønsted acids like HOAc. The six carboxylate groups of the glutamyl-glutamate ligands are arranged more symmetrically around the [Be₄O]⁶⁺ core than in comparable complexes with simple carboxylates.

Further research has to show if the [Be₄O]⁶⁺ moiety is indeed the responsible key agent in CBD. NMR-spectroscopic investigations are strongly encouraged in addition to crystallographic measurements because, owing to the local symmetry at the beryllium atoms, basic beryllium compounds show very sharp characteristic signals uncommon for aqueous solutions of beryllium salts in ⁹Be NMR spectra.

4 Methods

All molecular structures were optimized at the DFT level [14, 15] using B3LYP as implemented in Turbomole version 6.6 [14–17]. The def2-TZVPP and the m4 grid were employed [18, 19].

The converged structures were verified as minima on the potential energy surface by subsequent frequency analyses. For the frequency analyses, the local density functional BP86 [20] was used in connection with double-ζ basis sets (def-SVP) [16]. The entropy was calculated using Turbomole’s module freeh, treating the molecular rotations classically using the ideal gas model and the harmonic approximation to vibrational modes.