Downloaded by KTH ROYAL INST OF TECHNOLOGY on November 14, 2015 | http://pubs.acs.org Publication Date: May 21, 2009 | doi: 10.1021/bk-2009-1012.ch012
Chapter 12
Understanding the Biological Chemistry of Mercury Using a de novo Protein Design Strategy 1
Vincent L. Pecoraro*, Anna F. A. Peacock, Olga Iranzo , and Marek Łuczkowski 2
Department of Chemistry, University of Michigan, Ann Arbor, MI 48109-1055 Current addresses: Instituto de Tecnologia Química e Biológica, Avda. Da República (ΕΑΝ), 2781-901, Oeiras, Portugal; and University of Wrocław, 14 Joliot-Curie, 50-383 Wrocław, Poland 1
2
Hg(II) is a well known toxin that has a high affinity for protein thiolate functional groups. While an area of significance, a dearth of literature exists on the chemistry of Hg(II) with thiolate containing proteins. In this chapter we demonstrate the design of proteins that complex Hg(II) in linear, trigonal planar, and tetrahedral environments. Physical techniques such as Hg NMR, Hg PAC and UV-vis spectroscopy to characterize Hg(II) sites in proteins are also described along with the application of our understanding of Hg(II) interactions with designed proteins to address the binding of Hg(II) in protein sites such as MerA (2-coordinate), MerR (3coordinate) and Hg substituted rubredoxin (4-coordinate). Finally, knowledge from this system is used to predict the chemistry of Hg(II) bound forms of Hah1 at high pH. 199
199m
Although Lewis Carroll's famed hatter is the most notorious literary character thought inflicted by mercury poisoning, the phrase "mad as a hatter" is known to predate Alice in Wonderland by over fifty years. Despite the fact that the toxic effects of mercurials have been known for at least two hundred years, this element still remains a concern in modern society. Chief among the culprits © 2009 American Chemical Society
In Bioinorganic Chemistry; Long, E., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2009.
183
Downloaded by KTH ROYAL INST OF TECHNOLOGY on November 14, 2015 | http://pubs.acs.org Publication Date: May 21, 2009 | doi: 10.1021/bk-2009-1012.ch012
184 for dangerous mercuric complexes are alkyl mercurials such as dimethyl mercury (7). Of particular interest has been the release of mercury through industrial stacks, die burning of fossil fuels, the presence of mercury in vaccines and silver amalgam fillings and the accumulation of mercury in top of the food chain fish such as tuna (2). Given the broad deleterious health effects of this element, it is somewhat remarkable that so little is truly known about the specific biochemical sites of mercury intervention. Clearly Hg(II) has a high affinity for sulfhydryl groups of proteins; however, there are numerous targets that have been inferred, but not necessarily proven as the origin of the toxic effects (3, 4). Certainly, methyl mercury compounds, being more lipid soluble, easily cross the blood brain barrier, but once across, it is unclear which specific biochemical targets are impacted (e.g., tubulin, acetylcholinesterase, etc.). The same can be said for immunologic suppression by mercury, the effects are well established but the molecular targets remain elusive. Detoxification in humans is associated with metallothioneins which sequester the metal (5), whereas bacteria reduce Hg(II) to the less toxic Hg(0) using a reductase (6). In order to understand the biochemistry of mercury more completely, it is important to describe the chemistry of this element with sulfhydryl donors in the most common coordination environments. Ideally, one would examine mercuric complexes within a construct that was basically invariant, but which allowed for preparation of mercury compounds in the most common structures. The desired coordination modes include linear (or slightly bent) for 2-coordinate complexes, trigonal planar (or slightly T-shaped) for 3-coordinate compounds and tetrahedral for 4-coordinate complexes, Figure 1. Unfortunately, there are no known native systems that allow for this diversity of metal coordination geometry. However, given advances in peptide synthesis and the prediction of
2RS
RS^
\Hg
\
SR
RS.
Hg SR
Linear
SR Trigonal
SR Hg
RS
SR
Tetrahedral
Figure 1. Common coordination modes for Hg thiolate complexes.
In Bioinorganic Chemistry; Long, E., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2009.
185
Downloaded by KTH ROYAL INST OF TECHNOLOGY on November 14, 2015 | http://pubs.acs.org Publication Date: May 21, 2009 | doi: 10.1021/bk-2009-1012.ch012
protein structure, one should be able to exploit the emerging field of de novo protein design in order to prepare well defined scaffolds that should sequester mercury into each of these desired structures. This article describes how this objective may be met.
Preparation of 2-, 3- and 4-Coordinate Hg(II) Thiolate Complexes We have shown that the de novo designed TRI peptide family associates into specific a-helical aggregates depending on the pH, sequence composition and peptide length. The parent peptide T R I , with the linear sequence AcG(LKALEEK) G-NH , exploits the concept of a heptad repeat to place hydrophobic residues (in this case, leucine) in the 1 and 4 positions of a seven amino acid repeating sequence. The consequence of this sequence is that hydrophobes will occupy one face of an a-helix while the remaining positions can contain residues that both solubilize the peptide and stabilize a specific aggregation state through specific salt bridges. In the case of TRI, one forms at low pH values (< 5.5) predominantly two-stranded coiled coils and at pH values > 5.5 parallel, three-stranded coiled coils (7, 8). Similar pH dependent aggregation state behavior is observed for the related peptides, B A B Y and G R A N D , which have the same repeated heptad sequence but which are either shorter (three heptads) or longer (five heptads), respectively, than TRI. The stability of the aggregate is enhanced by lengthening the peptide (9). Thus, at the same concentrations, B A B Y peptides may be unassociated and unfolded, TRI peptides may be partially associated and folded and G R A N D peptides fully associated and folded. In order to introduce metal binding sites into these peptides, one or more leucine residues are replaced by cysteine residues making, for example, TRIL16C (see Table I for this and related sequences). When associated as a two-stranded coiled coil, one can prepare a scaffold presenting two cysteines to a metal, whereas the three-stranded coiled coil architecture provides a trigonal plane of three cysteinyl sulfur atoms. Alternative constructs include di-substituted peptides in which adjacent layers of leucines are substituted by cysteine residues yielding peptides that provide four or six sulfur donor atoms depending on whether they form two- or three-stranded coiled coils. Our initial foray into mercury binding with these peptides utilized TRIL16C and TRIL12C. Based on sequence, these peptides differ only by the site of cysteine substitution with TRIL16C having a cysteine in the 1 or a position and TRIL12C placing cysteine in the 4 or d position of a heptad. As we will see in a 4
2
st
th
st
th
In Bioinorganic Chemistry; Long, E., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2009.
186 Table I. Derivatives of TRI Peptides
Downloaded by KTH ROYAL INST OF TECHNOLOGY on November 14, 2015 | http://pubs.acs.org Publication Date: May 21, 2009 | doi: 10.1021/bk-2009-1012.ch012
Peptide
Sequence
TRI
Ac-G LKALEEK LKALEEK LKALEEK LKALEEK G-NH
TRIL9C
Ac-G L K A L E E K CKALEEK LKALEEK L K A L E E K G-NH
2
TRIL12C
Ac-G LKALEEK LKACEEK LKALEEK LKALEEK G-NH
2
TRIL16C
Ac-G LKALEEK LKALEEK CKALEEK LKALEEK G-NH
2
TRIL19C
Ac-G LKALEEK LKALEEK LKACEEK LKALEEK G-NH
2
TRIL9CL12C
Ac-G LKALEEK CKACEEK LKALEEK LKALEEK G-NH
2
TRIL12CL16C
Ac-G LKALEEK LKACEEK CKALEEK LKALEEK G-NH
2
2
moment, this slight shift will have a profound impact on trigonal metal coordination; however, we will first explore the 2-coordinate species formed by both systems (7, 8). At pH values below 5.5, both peptides prefer to aggregate as two-stranded coiled coils. Thus, it was expected, and subsequently confirmed through spectroscopic studies, that 2-coordinate bis thiolato Hg(II) compounds would exist regardless of the peptide:metal ratio. More interesting was the behavior at higher pH values where the peptide has a three-stranded coiled coil aggregation state preference. Three distinct behaviors were observed. If the stoichiometry of peptide to mercury was 2:1, only a two-stranded coiled coil with 2-coordinate Hg(II) was observed for either peptide. This observation demonstrated that under these conditions, the stability of Hg(II) in a 2coordinate structure exceeded that of the bundle to retain its preferred threestranded coiled coil aggregation state. If the ratio of peptide to mercury was increased to 3:1 and the pH was maintained between 5.5 and 7, the predominant species for both peptides was a three-stranded coiled coil that contained a two coordinate Hg(II) species (10). This suggested that the third cysteine of the aggregate remained protonated and uncoordinated to the Hg(II). If the pH of these solutions was now raised (to 8.6 for TRIL16C and 9.5 for TRIL12C) one obtained three-stranded aggregates that contained fully 3-coordinate, trigonal planar Hg(II) (9). We were delighted with these observations as these mercurated peptides provided the first peptidic system to bind Hg(II) as a trigonal thiolato complex in aqueous solution. Further analysis demonstrated that an equilibrium existed between 2- and 3coordinate Hg(II) within the three-stranded coiled coil according to the equation: Hg(II)(pep) (Hpep) -» Hg(pep) - + H 2
3
In Bioinorganic Chemistry; Long, E., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2009.
187 Quantitative assessment of these equilibria for TRIL16C and TRIL12C yielded pK 's of 7.8 and 8.5, respectively (77). These data provided the first inkling that metal binding to cysteine residues was dependent on the a vs. d substitution pattern of these peptides. Subsequent studies with Cd(II) and Pb(II) derivatives have further demonstrated this principle (72, 75). It should be noted that the pK for the 2- to 3-coordinate conversion is not dependent on the stability of the peptide aggregate (e.g., BABYL9C, TWL9C and GRANDL9C have the same pK values of 7.6±0.2) (9, 14), but instead tracks with the a or d site substitution pattern. A summary of these various equilibria is given in Figure 2. In a subsequent study, we first demonstrated that the simple addition of small thiolates such as P-mercaptoethanol to existing Hg(pep)2 did not lead to trigonal thiolato mercury complexes (75). Thus, the capability of forming these desired structures is a direct consequence of peptide recognition. We next addressed the relationship between the peptide self association affinity and the ability to form trigonal Hg(II) species. Using eleven different peptides, we found a linear free energy correlation between the self association affinity to form a three-stranded coiled coil and the energy of the formation of Hg(pep) " (75). These studies conclusively demonstrated that our ability to complex Hg(II) as a trigonal complex was a direct consequence of the designed peptide recognition and that we could titrate metal peptide affinities by controlling the self association affinities of the apopeptides. We next challenged our design strategy by testing Hg(II) complexation in solutions containing mixtures of TRIL2WL9C and TRIL2WL23C. In theory, mixtures of these peptides could form anti-parallel three-stranded coiled coils that yielded trigonal Hg(II) (e.g., Hg(TRlL2WL9C) (TRlL2WL23C)"). In fact, no such heterotrimeric complexes were observed by circular dichroism or Hg NMR spectroscopies (16). These studies demonstrated conclusively that we could define the coordination environment of the mercury ion while retaining exquisite control of protein aggregation state and orientation. With 2- and 3-coordinate Hg(II) complexes in hand, our next objective was to prepare Hg(SR) " complexes. Because our protein design does not allow for the formation of four-stranded aggregates, we needed to shift our strategy to peptides containing dual cysteine substitution. The two peptides that we chose to examine were TRIL12CL16C and TRIL9CL12C. These constructs allowed us to compare a -Cys -X-X-X-Cys - binding motif found in TRIL12CL16C to the more common -Cys -X-X-Cys - sequence of native proteins found in TRIL9CL12C. Studies of these dual substituted peptide systems are complicated by perturbing two adjacent leucine layers which stabilize the peptide aggregates. Thus, rather than achieving pure two-stranded or three-stranded coiled coils at pH = 8.5, we observe a mixture of species (77). Nonetheless, addition of Hg(II) leads to well defined structures. In the case of TRIL12CL16C one obtains twostranded coiled coils that yield spectroscopic parameters that are the hallmark of a
Downloaded by KTH ROYAL INST OF TECHNOLOGY on November 14, 2015 | http://pubs.acs.org Publication Date: May 21, 2009 | doi: 10.1021/bk-2009-1012.ch012
a
3
2
199
2
4
d
a
t
d
In Bioinorganic Chemistry; Long, E., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2009.
a
Downloaded by KTH ROYAL INST OF TECHNOLOGY on November 14, 2015 | http://pubs.acs.org Publication Date: May 21, 2009 | doi: 10.1021/bk-2009-1012.ch012
188
9 .
Increase in pH
Q B
8
A
pH * 8.6
pH = 6.5 199,
HgNMR: -844 ppm 199m Hg PAC: v = 1,539(10)
199
Hg NMR: -844 ppm
1 9 9 m
H g PAC:
Q
rj- 0,13(3}
tf — 0.11(3)
1 9 9
199
H g NMR: -908 ppm
199m
Hg p
A
C
.
V
Q
U
i
5
5
8