A designed heme-[4Fe-4S] metalloenzyme catalyzes sulfite reduction like the native enzyme

See allHide authors and affiliations

Science  14 Sep 2018:
Vol. 361, Issue 6407, pp. 1098-1101
DOI: 10.1126/science.aat8474

Metals brought together do more

Enzymatic reduction of oxyanions such as sulfite (SO32−) requires the delivery of multiple electrons and protons, a feat accomplished by cofactors tailored for catalysis and electron transport. Replicating this strategy in protein scaffolds may expand the range of enzymes that can be designed de novo. Mirts et al. selected a scaffold protein containing a natural heme cofactor and then engineered a cavity suitable for binding a second cofactor—an iron-sulfur cluster (see the Perspective by Lancaster). The resulting designed enzyme was optimized through rational mutation into a catalyst with spectral characteristics and activity similar to that of natural sulfite reductases.

Science, this issue p. 1098; see also p. 1071


Multielectron redox reactions often require multicofactor metalloenzymes to facilitate coupled electron and proton movement, but it is challenging to design artificial enzymes to catalyze these important reactions, owing to their structural and functional complexity. We report a designed heteronuclear heme-[4Fe-4S] cofactor in cytochrome c peroxidase as a structural and functional model of the enzyme sulfite reductase. The initial model exhibits spectroscopic and ligand-binding properties of the native enzyme, and sulfite reduction activity was improved—through rational tuning of the secondary sphere interactions around the [4Fe-4S] and the substrate-binding sites—to be close to that of the native enzyme. By offering insight into the requirements for a demanding six-electron, seven-proton reaction that has so far eluded synthetic catalysts, this study provides strategies for designing highly functional multicofactor artificial enzymes.

The presence of sulfite in the environment inhibits bioremediation of pervasive, toxic oxyanions such as (per)chlorate, arsenate, and nitrate (1) that are chemically challenging to reduce and have low binding affinity to transition metals. Bioinspired artificial catalysts have been developed to reduce nitrate and perchlorate (2), but sulfite reduction remains inaccessible to synthetic catalysts. As a result, complete oxyanion remediation has so far been accomplished only via biofilms undergoing anaerobic respiration (3, 4)

SO32− + 6 e + 7 H+ → HS + 3 H2O(1)

Sulfite reduction to sulfide (Eq. 1) is accomplished to completion by assimilatory sulfite reductase (SiR), which contains a structurally complex cofactor composed of a heme macrocycle (siroheme) and a cubane [4Fe-4S] cluster (5) bridged by a cysteine residue that serves as both the proximal ligand to the siroheme and a ligand to one Fe atom of the [4Fe-4S] cluster (Fig. 1). The [4Fe-4S] cluster is proposed to act as a molecular battery that facilitates electron transfer to the heme, enabling sequential 2e reduction (6), even though the precise role it plays in catalysis and tuning siroheme reactivity is not understood. Two synthetic models of the heme-[4Fe-4S] cofactor have been reported but lacked activity (7, 8). These models did not include elements of the SiR substrate-binding pocket, which is rich in positively charged residues thought to facilitate binding and protonation of the substrate concurrent with electron transfer and proposed to be required for activity (9, 10). Designing artificial enzymes that include mononuclear or homonuclear metal-binding sites has been successful in generating catalysts (1117); however, designed enzymes have rarely recapitulated the complex heteronuclear metal centers responsible for multielectron, multiproton reactions. Here we report success in replicating both structural and functional elements of SiR by designing a [4Fe-4S] cluster proximal to the heme center in cytochrome c peroxidase (CcP). By rationally building secondary sphere interactions in the metal- and substrate-binding sites, we increased the sulfite reductase activity of this artificial metalloenzyme to approach the activity of a native SiR.

Fig. 1 Design of a [4Fe-4S] binding site in CcP to mimic the heme-[4Fe-4S] center in native SiR.

From left to right: A search structure generated from the binding cavity of the [4Fe-4S] in the siroheme-[4Fe-4S] cofactor from the hemoprotein subunit of native Escherichia coli SiR [Protein Data Bank (PDB) ID 2GEP] (9) was used to search the PDB for suitable hemoprotein scaffolds. Yeast CcP was identified as a suitable scaffold, and a binding site for a heme-[4Fe-4S] cofactor was designed by a combination of computational and rational design methods. The resulting computational model of the designed heme-[4Fe-4S] center in SiRCcP.1 is shown on the right. Three Cys mutations (T180C, W191C, L232C) coordinate Fe atoms, and the H175C mutation acts as a bridging Cys ligand between the heme and [4Fe-4S] cofactors.

We chose CcP, a native heme-binding protein, as a protein scaffold because of its small size and stability. Furthermore, we found that its active site contains a cavity on the heme proximal face large enough to host a [4Fe-4S] cluster (figs. S1 and S2). This heme proximal cavity was essential in discriminating among suitable hemoprotein scaffolds, as described in the supplementary materials (fig. S3). We then used the Rosetta matcher and enzyme design algorithms (11) to select residues to mutate to Cys to coordinate the [4Fe-4S], resulting in mutation of four residues (His, Thr, Tyr, and Leu) to Cys [His175→Cys (H175C), T180C, W191C, and L232C] (Fig. 1) and two additional mutations for stability: M230A (M, Met; A, Ala), made to relieve steric clash, and D235V (D, Asp; V, Val), made to remove the nearby negative charge previously reported to interfere with Cys-heme coordination in CcP (18) (fig. S4). The computational model of this sextuple mutant CcP (called SiRCcP.1) is a close structural match for the cofactor-binding site in native SiR (fig. S5) and was readily expressed in the cofactor-free apo form (figs. S6 and S7).

We incorporated an iron-sulfur cluster into apo-SiRCcP.1 by in vitro reconstitution, and we hereafter refer to this protein as FeS-SiRCcP.1. We characterized FeS-SiRCcP.1 by ultraviolet (UV)–visible absorption, electron paramagnetic resonance (EPR), and extended x-ray absorption fine structure (EXAFS) spectroscopies. These characterization methods of both the oxidized and reduced states indicate proper incorporation of a [4Fe-4S] cluster, as designed (19, 20) (Fig. 2, A and B, and figs. S8 to S11). Elemental analysis for Fe and S in FeS-SiRCcP.1 matched well with the Fe and S content expected at full [4Fe-4S] incorporation (4 Fe:14 S, accounting for inorganic S and all S atoms in Cys and Met residues) with a ratio of 4.00 Fe:14.01 S observed experimentally.

Fig. 2 Spectroscopic properties of SiRCcP.1 with [4Fe-4S], heme, and heme-[4Fe-4S] cofactors confirm binding of [4Fe-4S] and heme-[4Fe-4S] cofactors.

(A) X-band EPR spectrum of FeS-SiRCcP.1 reduced with an excess of sodium dithionite (black) and simulated spectrum (magenta) indicating an S = ½ species consistent with a [4Fe-4S]+: gx = 1.891, gy = 1.919, gz = 2.035; linewidths (G) Ax = 42, Ay = 27, Az = 25. The spectrum shown was measured at a frequency of 9.173 GHz and modulation amplitude of 10 Gauss; a microwave power of 10 mW, and a temperature of 15 K. No paramagnetic species at g = 2 are observed before reduction, suggesting a [4Fe-4S]2+ (S = 0) state as prepared (oxidized). A [3Fe-4S]+ species could be generated by reoxidation of the reduced species with an excess of potassium ferricyanide (fig. S10). (B) Magnitude of the phase-uncorrected k3-weighted Fourier transform (FT) and EXAFS (inset) of the Fe K-edge spectra for FeS2+-SiRCcP.1 (black) and the best fit (magenta). The EXAFS spectrum is consistent with a symmetrical cubane structure (three scattering Fe atoms per absorber Fe and four scattering S atoms in the first coordination shell). Fitting parameters are provided in table S1, and the EXAFS of FeS+-SiRCcP.1 are plotted in fig. S11. R, scattering shell distance. (C) UV-visible spectra of Fe-S–reconstituted SiRCcP.1 (FeS-SiRCcP.1) and heme-Fe-S–reconstituted SiRCcP.1 (heme-FeS-SiRCcP.1) in the presence and absence of potassium cyanide. As prepared, FeS-SiRCcP.1 has weak, broad absorption at 400 nm, which decreases upon reduction (fig. S9). Ferric heme-FeS-SiRCcP.1 has a blue-shifted Soret peak (378 nm) relative to that of native CcP (408 nm) and exhibits a Soret maximum and Q-bands (inset) consistent with penta-coordinate, thiolate-ligated high-spin ferriheme. The ferrous heme-FeS-SiRCcP.1 spectrum obtained by incubation with sodium dithionite (fig. S12) is distinct from the species obtained by photoreduction (fig. S13) and is identical to the spectrum of heme2+-FeS+-SiRCcP.1 incubated with sulfite under nonturnover conditions (fig. S14), indicating sulfite-bound hexacoordinate ferroheme. The photoreduced spectrum (fig. S13) represents the ferrous pentacoordinate species. (D) X-band EPR spectra of heme-FeS-SiRCcP.1 in the presence of cyanide. CN-heme3+-FeS2+-SiRCcP.1 is predominantly low-spin ferric heme (gx = 1.87, gy = 2.26, gz = 2.45) with no visible [4Fe-4S] features. In the all-ferrous state, heme features disappear (presumably S = 0 ferrous heme) while the S = ½ [4Fe-4S] feature reappears. Similar behavior has been reported for CN-bound SiR. EPR spectra were measured at 18 K with a microwave power of 10 mW at a frequency of 9.24 GHz.

In addition to the [4Fe-4S], a heme cofactor is required to complete the SiR catalyst, and we found that heme-b, the native heme cofactor of CcP, could be incorporated into FeS-SiRCcP.1 in vitro, similarly to native CcP. We call this double-cofactor form heme-FeS-SiRCcP.1 (Fig. 2C). Elemental analysis of heme-FeS-SiRCcP.1 yielded 5.54 Fe per heme, confirming that the [4Fe-4S] remains intact following heme reconstitution. Heme-FeS-SiRCcP.1 displayed key spectroscopic features and ligand-binding properties of thiol-ligated hemoproteins and even native SiR (Fig. 2, C and D, and figs. S12 to S14) (2123). Notably, heme3+-FeS2+-SiRCcP.1 has low affinity for the cyanide anion (CN), requiring ~2000 equivalents, whereas 1e reduced heme2+-FeS2+-SiRCcP rapidly binds roughly one equivalent of CN (Fig. 2C). This behavior has been reported for SiR (24) but not for most other hemoproteins, including CcP, which suggests that our designed SiRCcP.1 substantially alters the heme character of the native scaffold protein without altering the heme chemical structure. The presence of the positively charged [4Fe-4S] could screen the increased negative charge in ferrous heme that tends to weaken the CN-Fe2+ interaction in hemoproteins, but we also observed higher CN affinity in heme2+-SiRCcP.1 with no reconstituted [4Fe-4S]. Therefore, local electrostatic effects, perhaps related to removal of the proximal His-Asp-Trp peroxidase triad in SiRCcP.1 (fig. S2), or changes in heme conformation that affect σ acceptance and π back-donation must also play a role (2527).

We measured the sulfite reductase activity of heme-FeS-SiRCcP.1 by the rate of oxidation of the electron mediator methyl viologen (MV+) in the presence of sodium sulfite using a protocol previously reported for native SiR (28). Heme-FeS-SiRCcP.1 oxidizes MV+ at a rate of 0.348 (± 0.15) min−1, whereas wild-type CcP displayed no measurable activity above background. Similarly, the rate by heme-FeS-SiRCcP.1 was nominally zero in the absence of sulfite. The initial design that incorporated the heme-[4Fe-4S] cofactor (SiRCcP.1), therefore, showed notable sulfite reduction activity compared with its native scaffold protein but far less than a native SiR, indicating that a heme-[4Fe-4S] cofactor alone is insufficient to promote rapid sulfite reduction. To improve activity, we sought to systematically examine the key features of native SiR active sites and determine which had the greatest impact on catalysis. Lys and Arg residues are conserved in SiR and are thought to be important for substrate binding and catalysis (9, 10) (fig. S15A). Therefore, we introduced analogous W51K, H52R, and P145K (K, Lys; R, Arg; P, Pro) mutations to SiRCcP.1, which, when combined with the native CcP residue Arg48, form a positively charged cavity that resembles SiR (Fig. 3A and fig. S15B). The reductase activity of the resulting mutant, W51K/H52R/P145K-SiRCcP.1, is 1.26 ± 0.20 min−1, 5.3-fold greater than that of SiRCcP.1.

Fig. 3 Sulfite reduction activity of SiRCcP mutants is modulated by substrate binding and [4Fe-4S] secondary sphere mutations.

(A) Computational model of the SiRCcP substrate-binding site with mutations W51K, H52R, and P145K (cyan) made to mimic native SiR residues K217, R153, and K215, respectively (fig. S15). The native CcP residue R48 (orange) is positioned similarly to R83 in native SiR. (B) Computational model of the mutations D235V (SiRCcP.1, cyan), D235N (SiRCcP.2, green), and D235C (SiRCcP.3, magenta) in SiRCcP. (C) Sulfite reduction activity of four SiRCcP mutants in comparison to a native sulfite reductase from M. tuberculosis (Mtb NiRA). “K+R” denotes the presence of the three mutations W51K, H52R, and P145K. Reported activities represent the average of triplicate measurements with standard error (see table S2).

In proteins, [4Fe-4S] clusters are stabilized by backbone and amino acid side-chain interactions in the secondary coordination sphere (29, 30). The crystal structures of SiRs reveal several side-chain hydrogen bond donors to inorganic S atoms, such as the amino group of N481 (N, Asn) in E. coli SiR (fig. S16A). We sought to increase the activity of SiRCcP by adding similar stabilizing interactions to the [4Fe-4S] cluster secondary coordination sphere by mutating residue D235—whose functional group is oriented toward the [4Fe-4S] in SiRCcP in a position similar to N481 in SiR—to Asn and Cys based on important secondary sphere residues in various [4Fe-4S] centers (29). We found that the D235N mutation (called SiRCcP.2) (Fig. 3B and fig. S16B) increased activity to 5.91 ± 1.6 min−1, 4.3-fold greater than that of W51K/H52R/P145K-SiRCcP.1 (a 17-fold increase over SiRCcP.1). We achieved an even greater increase in activity with the D235C mutant (called SiRCcP.3) (Fig. 3B), which, when combined with the substrate-binding site mutations given above (W51K/H52R/P145K-SiRCcP.3), exhibited sulfite reduction activity of 21.8 (± 2.4) min−1, a 63-fold increase over that of SiRCcP.1 (fig. S17 and table S2). We compared these rates to a native dissimilatory SiR, NiRA from Mycobacterium tuberculosis (Mtb NiRA) (28), whose activity has been reported under comparable reaction conditions, and found that the activity of W51K/H52R/P145K-SiRCcP.3 is ~18% of a native SiR’s activity (Fig. 3C and table S2). Assimilatory SiRs have been observed to reduce sulfite to sulfide with minimal side products, but dissimilatory SiRs typically achieve far less than 50% efficiency in vitro. We quantified sulfide as well as the side products trithionate (S3O62−) and thiosulfate (S2O32−) produced with sulfite as the substrate. Sulfide was quantified by methylene blue formation, and we found that ~10% of the products formed in the reaction were hydrogen sulfide (the six-electron reduction product), with the remainder being primarily the two- and four-electron–reduced side products in a distribution similar to native dissimilatory SiR. Together, these products account for more than 90% of the electron donors consumed (figs. S18 and S19). The product conversion efficiency of W51K/H52R/P145K-SiRCcP.3 is therefore similar to the efficiency of a natural dissimilatory SiR (31). SiRCcP is thus both a close structural model and a substantially active functional model of SiR capable of complete six-electron and seven-proton sulfite reduction.

Our progressive optimization of SiRCcP revealed structural features beyond the metallocofactor required for efficient multielectron reduction of an oxyanion, including secondary sphere interactions such as hydrogen bonding to the [4Fe-4S] (through D235V/N/C mutations) and residues that may promote substrate binding and protonation (through W51K, H52R, and P145K mutations). Models that did not incorporate these features had diminished activity, and the necessity of these secondary interactions in our SiRCcP model has important implications for optimizing design strategies to achieve high activity in multielectron redox reactions with artificial enzymes, a process simplified and accelerated in this study by cavity-based scaffold selection. Furthermore, the SiRCcP enzyme is functional with the heme-b cofactor present in native CcP instead of the biologically distinct siroheme in SiR, suggesting that siroheme is not absolutely required for sulfite reduction. These insights, combined with the ability to recapitulate a complex multielectron, multiproton delivery system in a simple, stable scaffold, suggest a strategy to design highly active catalysts for other difficult oxyanion reduction and multielectron reactions that may be broadly applicable.

Supplementary Materials

Materials and Methods

Supplementary Text

Figs. S1 to S19

Tables S1 and S2

Listings S1 to S4

References (3258)

References and Notes

Acknowledgments: We thank R. Schnell and G. Schneider from Karolinska Institutet (Stockholm, Sweden) for providing plasmids used to express the native SiR from M. tuberculosis (Mtb NiRA), T. B. Rauchfuss and A. R. Fout for helpful discussions, and Y. Lee for assistance with collecting enzyme activity data. X-ray absorption studies were conducted at the Stanford Synchrotron Radiation Lightsource at the SLAC National Accelerator Laboratory, and we thank beamline scientists M. Latimer and E. Nelson for their assistance. All work reported here was conducted at the University of Illinois at Urbana-Champaign. Funding: The work described in this paper is supported by the U.S. National Institutes of Health (R01-GM062211 to Y.L. and predoctoral training grant 5T32-GM827625 to E.N.M.). This work is also partially funded by the U.S. Department of Energy's Center for Advanced Bioenergy and Bioproducts Innovation (Office of Science, Office of Biological and Environmental Research, under DE-SC0018420). Any opinions, findings, and conclusions or recommendations expressed in this publication are those of the author(s) and do not necessarily reflect the views of the U.S. Department of Energy. Use of the Stanford Synchrotron Radiation Lightsource, SLAC National Accelerator Laboratory, is supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences under contract DE-AC02-76SF00515. The SSRL Structural Molecular Biology Program is supported by the DOE Office of Biological and Environmental Research, and by the National Institutes of Health, National Institute of General Medical Sciences (including P41GM103393). The contents of this publication are solely the responsibility of the authors and do not necessarily represent the official views of NIGMS or NIH. Author contributions: E.N.M., P.H., and Y.L. conceptualized the project; E.N.M. and I.D.P. conducted computational investigations; E.N.M and P.H. conducted experimental investigations and data analyses; E.N.M. and M.J.N. conducted EPR analyses; E.N.M. created data visualizations; Y.L. was responsible for project supervision, direction, resource acquisition, and data analyses and interpretation; E.N.M. drafted the original manuscript; and E.N.M., I.D.P., P.H., M.J.N., and Y.L. edited and reviewed the manuscript. Competing interests: E.N.M., P.H., and Y.L. are inventors on patent application 62/702,940, submitted by University of Illinois at Urbana-Champaign, that covers artificial metalloproteins as biocatalysts for sulfite reduction. I.D.P. is currently affiliated with the Bioinformatics Department, Ambry Genetics; P.H. is currently affiliated with the Department of Biochemistry and Institute for Protein Design, University of Washington; and Y.L. is also affiliated with the Department of Bioengineering, Materials Science and Engineering, and Beckman Institute for Advanced Science and Technology, University of Illinois at Urbana-Champaign. Data and materials availability: All data are available in the main text or supplementary materials. All SiRCcP variants described in this Report (SiRCcP.1, SiRCcP.2, and SiRCcP.3 and their W51K/H52R/P145K variants, as described in the main text) are available from the University of Illinois at Urbana-Champaign under a material transfer agreement with the university.

Correction (20 September 2018): G. Schneider's name was mistakenly omitted from the first sentence of the Acknowledgments. This error has been corrected.

View Abstract

Navigate This Article