Rational Design of Dual Active Sites in a Single Protein Scaffold: A Case Study of Heme Protein in Myoglobin.

Rational protein design has been proven to be a powerful tool for creating functional artificial proteins. Although many artificial metalloproteins with a single active site have been successfully created, those with dual active sites in a single protein scaffold are still relatively rare. In this study, we rationally designed dual active sites in a single heme protein scaffold, myoglobin (Mb), by retaining the native heme site and creating a copper-binding site remotely through a single mutation of Arg118 to His or Met. Isothermal titration calorimetry (ITC) and electron paramagnetic resonance (EPR) studies confirmed that a copper-binding site of [3-His] or [2-His-1-Met] motif was successfully created in the single mutant of R118H Mb and R118M Mb, respectively. UV/Vis kinetic spectroscopy and EPR studies further revealed that both the heme site and the designed copper site exhibited nitrite reductase activity. This study presents a new example for rational protein design with multiple active sites in a single protein scaffold, which also lays the groundwork for further investigation of the structure and function relationship of heme/non-heme proteins.

Metalloproteins with multiple active sites usually perform biological functions effectively. For example, in cytochrome c oxidase (CcO), heme a delivers electrons to the heme a3:CuB site, where O2 is reduced to H2O,1 and in copper nitrite reductase (NIR), type‐1 copper delivers electrons to type‐2 copper, the site of nitrite binding and reduction.2 Moreover, a di‐heme enzyme, MauG, possesses two hemes c in distinct spin states with a long‐range charge‐resonance model.3 It is thus attractive to create functional artificial proteins with multiple active sites for more efficient biocatalysis. Rational protein design has been proven to be a powerful tool, based on the scaffold of either natural proteins or de novo proteins.4 Although many artificial metalloproteins with a single active site have been successfully created,4 those with dual active sites in the same protein are still relatively rare. Based on a protein dimer, Roelfes and co‐workers designed an artificial metalloenzyme with dual copper sites capable of catalyzing the Diels–Alder reaction.5 Using domain swapping for horse heart myoglobin (Mb), Hirota and co‐workers designed a heterodimeric protein with two different heme active sites, which exhibits distinct ligand binding properties.6 These artificial metalloproteins were designed using two proteins by dimerization. Alternatively, Dutton and co‐workers designed 4‐helix bundles that can accommodate two heme groups or one heme and one other cofactor such as zinc chlorin.7 Moreover, dual Fe–S clusters or metal‐binding sites for Zn and Hg can be created in 3‐helix bundles.8 Therefore, development of other methods to design functional proteins with dual active sites in a single protein scaffold will broaden the variety of artificial enzymes. Note that although simply fusing two enzymes together might be a straightforward method,9 chemical modifications as commonly used for fusion, as well as protein–protein interactions, should be considered to minimize the side effects on the active sites.

Due to the small size (153 amino acids) with a single heme group, the easy‐to‐produce Mb has been favored as a protein model for heme protein design.10 By redesigning the heme active site, the O2 carrier Mb has been converted successfully into various artificial enzymes.4a,4d–4h, 10, 11 Copper (CuB) and iron (FeB) binding sites can also be designed close to the heme iron, mimicking the heteronuclear metal center in CcO and nitric oxide reductase (NOR), respectively.12 With these advances, we are interested in design of dual active sites remote from each other using Mb as a model protein to create multiple metal centers in the same protein scaffold, which potentially produces artificial enzymes more efficient than those with a single active site.

To this end, we adopted a new approach of retaining the native heme and creating another metal‐binding site remotely from the heme site (Figure 1). The previous experience of designing a CuB site in the heme pocket of Mb suggests that it is essential to create a metal‐binding motif, such as a motif of three histidine,[3‐His], as found for the type‐2 copper site in NIR.2 By a close inspection of the crystal structure of sperm whale Mb (PDB code 1JP613), we found that a single mutation of Arg118 to His would create a [3‐His] copper‐binding motif, since the other two His residues (His24 and His119) are available in the native Mb with a proper conformation to form the motif (Figure 1). Similarly, mutation to Met would likely create a [2‐His‐1‐Met] copper‐binding site, as in the active site of peptidylglycine monooxygenase (PHM).14

Figure 1

Design of dual active sites in a single protein scaffold of Mb.

The choice of Arg118 as the target for mutation to build the metal‐binding motif was further supported by a crystallographic study of the F33Y CuBMb mutant in the presence of CuII.15 The structure revealed that, when the protein crystal was soaked with an excess of copper ions (10 equiv CuSO4), in addition to the copper in the designed CuB site, there were four other copper‐binding sites in the crystal structure (Figure S1 in the Supporting Information). Notably, a copper (Cu1, ∼20 Å from the heme iron) is coordinated by both His24 (2.03 Å) and His119 (1.67 Å), as well as a water molecule (3.69 Å). We hypothesize that this Cu1‐binding site containing two His residues is more stable than the other three copper‐binding sites on the protein surface. Moreover, the CD atom of Arg118 is close to Cu1 with a distance of 3.79 Å, suggesting that a His or Met introduced at this position may serve as an additional ligand for the copper. These observations encouraged us to construct both R118H Mb and R118M Mb single mutants.

R118H Mb and R118M Mb mutants were expressed and purified using the procedure reported previously.12a The protein was confirmed by matrix‐assisted laser desorption time‐of‐fight (MALDI‐TOF) mass spectrometry studies (Figure S2 in the Supporting Information). To investigate whether any copper binds to the designed metal‐binding site, we performed isothermal titration calorimetry (ITC) studies for the two single mutants, with WT Mb as a control. Titration of CuII into the solution of WT Mb caused only a slight change in binding enthalpy. In contrast, major changes were observed when the CuII was titrated into a solution containing R118M Mb or R118H Mb under the same condition (Figure 2). These results strongly suggest that introducing the R118H or R118m mutation increases the CuII‐binding significantly over that of WT Mb. To obtain quantitative information of the binding, we used a single‐site binding model to fit the data, yielding a dissociation constant (K d) value of ∼81 and ∼11 μm, and a stoichiometry of CuII to protein of 0.83±0.31 and 1.02±0.05, for R118M Mb and R118H Mb, respectively. Note that due to the limitation of ITC technique as well as data fitting, these results should be treated as approximate values. Moreover, slight changes in binding enthalpy were still observed for the two mutants with CuII/Mb ratio higher than 2 (Figure 2 B,C), which suggests the existence of a weak binding site or nonspecific binding to the protein surface. The data fitting for WT Mb showed a larger K d value of ∼752 μm, with a stoichiometry of 1.73±0.28, likely due to the binding of CuII to other sites such as the Cu2 site (Figure S1 in the Supporting Information). These results demonstrate that the [3‐His] and [2‐His‐1‐Met] copper‐binding sites were successfully constructed in Mb by a single mutation, with both sites exhibiting an affinity for CuII much higher (∼68‐ and ∼9‐fold, respectively) than that for WT Mb, although another weak binding site may also exist on the protein surface of the mutants.

Figure 2

ITC data for titration of WT Mb (A), R118M Mb (B), and R118H Mb (C) with CuII at 25 °C. Top, raw data. Bottom, plot of integrated heats versus CuII/Mb ratio.

To further confirm the CuII binding to the designed metal‐binding site, we collected electron paramagnetic resonance (EPR) spectra. As shown in Figure 3 A, the EPR spectra of R118H Mb and R118M Mb mutants in presence of 1 equiv CuII have features typical of the type‐2 copper,16 which are distinct from that for the free CuII in solution, indicating the binding of copper to protein. Although a similar spectrum was observed for CuII‐WT Mb under the same conditions, shoulder peaks appeared in the parallel region, as indicated by arrows (Figure 3 B). EPR simulation suggested that there are two copper‐binding sites with a ratio of ∼1:0.7 in WT Mb, whereas site‐2 was not obvious for either CuII‐R118M Mb or CuII‐R118H Mb (Figure S3 in the Supporting Information and Table 1). It should be noted that although ITC studies suggested there were free CuII in WT Mb with 1 equiv CuII due to the large K d value, the EPR spectrum collected at a much higher (∼10‐fold) concentration and lower temperature than the ITC studies showed copper binding to the protein. Meanwhile, a free CuII signal was indeed observed in EPR titration with excess CuII ions (Figure 4 A). EPR titration further showed that additional CuII would bind to site‐2 on the protein surface, and the signal intensity of site‐1 with respect to site‐2 is higher for the mutants than that for WT Mb (Figure 4 B ). These observations suggest that upon mutation of Arg118 in site‐1, His118 or Met118 acts as an additional ligand for the copper, resulting in a higher copper binding affinity.

Figure 3

A) EPR spectra of R118H Mb (i), R118M Mb (ii), and WT Mb (iii) (0.3 mm) in the presence of 1 equiv CuII and free CuII in buffer (iv), collected at 30 K, 2 mW power, and 9.43 GHz. B) Expanded view comparison of the parallel hyperfine region of the spectra in (A).

Table 1

Comparison of spin Hamiltonian parameters deduced from the simulation of the EPR spectra.

Copper‐binding site	g x	g y	g z
WT Mb (Site‐1)	2.047	2.050	2.263
WT Mb (Site‐2)	2.052	2.052	2.305
R118H Mb ([3‐His])	2.047	2.048	2.250
R118M Mb ([2‐His‐1‐Met])	2.047	2.050	2.263
PHM ([3‐His])	2.043	2.070	2.263
PHM ([2‐His‐1‐Met])	2.045	2.071	2.280
NIR (Type‐2 Cu site)	2.076	2.076	2.358

Mb: myoglobin, PHM: peptidylglycine monooxygenase, NIR: nitrite reductase

Figure 4

A) EPR spectra for CuII (1, 2, 3, 5, and 10 equiv) titration of WT Mb (0.18 mm) and 5 equiv free CuII in the same buffer solution (the arrow indicates the free CuII and CuII bound to site‐2, respectively). B) EPR spectra comparison of R118H Mb, R118M Mb, and WT Mb in the presence of 5 equiv CuII under the same conditions (CuII binding sites are indicated by arrows).

Note that CuII‐R118M Mb has identical g values to the site‐1 of Mb (Table 1), which suggests that coordination of Met118 causes no geometry alteration for the copper. This may be attributed to that fact that Met has the same side chain of CA–C atoms to that of Arg, and the sulfur atom of Met118 is likely in a similar position of the CD atom of Arg118, with a short distance favorable for direct coordination (Figure S1 in the Supporting Information). Interestingly, the CuII‐R118H Mb has a g value slightly lower than that of CuII‐R118M Mb, which is consistent with the prediction that metal‐binding sites with more N donors have lower g values based on Peisach–Blumberg correlations.17 These findings also agree with a recent study of CuII binding to engineered PHM mutants with a [3‐His] or [2‐His‐1‐Met] copper‐binding site (Table 1).14 Meanwhile, when compared with the type‐2 copper site in native NIR from Alcalogenes faecalis (Table 1),18 large differences in g values were observed for both CuII‐R118H Mb and CuII‐R118M Mb, indicating that the artificially constructed copper site in Mb does not identically reproduce the geometry of the type‐2 copper in native NIR. These structural differences might have profound effects on the protein reactivity of nitrite reduction.

After confirming that CuII indeed binds to the single mutants of Mb, we then tested whether the copper site confers NIR activity, as it resembles the active site of native copper NIR,2 albeit not identically. We performed an ascorbate–nitrite assay, which was demonstrated by Pecoraro and co‐workers for testing the NIR activity of type‐2 copper center designed in α‐helical coiled coils.19 In this assay, ascorbate acts as a sacrificial reductant for CuII, and the resultant CuI‐bound protein catalytically reduces nitrite to NO. The oxidation of ascorbate was monitored by the decrease of its absorption at 265 nm (Figure 5 A). Based on the changes of A265nm as a function of time (Figure 5 B), the initial rate of ascorbate oxidation (vASC,Ox) for the first 2000 sec was calculated to be (0.90±0.05)×10−5 mm s−1 for CuII‐R118H Mb. This value is 2.25‐ and 1.55‐fold of that for CuII‐WT Mb ((0.40±0.02)×10−5 mm s−1) and CuII‐R118M Mb ((0.58±0.02)×10−5 mm s−1), respectively. These results suggest that the copper site in Mb with a higher copper affinity exhibits a higher NIR activity, presumably due to a higher occupancy of the copper site according to the ITC studies. The charge of the active site changed upon mutation of Arg118 might also affect the binding and reduction of nitrite. Note that CuII alone did not show catalytic activity in this assay, as confirmed in previous study.19 In the absence of CuII ions, a low oxidation rate ((0.10±0.01)×10−5 mm s−1) of ascorbate was observed for WT Mb and its mutants, which may be due to an auto‐oxidation of ascorbate,20 since the heme iron remained in the ferric state in these conditions, and no reaction occurred at the heme site. Additionally, the NO product in this essay was confirmed by addition of the headspace gas in the reaction cuvette for CuII‐R118H Mb to deoxy WT Mb, which resulted in the formation of the NO‐bound form, as shown in the UV/Vis spectrum (Figure S4 in the Supporting Information).

Figure 5

A) UV/Vis spectra of a solution of sodium ascorbate (45 μm) and NaNO2 (5 mm) collected each minute, containing R118H Mb (5 μm) and CuII (5 μm). B) Decreases of [Asc−] versus time in samples containing R118H Mb, R118M Mb, or WT Mb in the absence or presence of CuII.

To verify that the heme site in the CuII‐R118H Mb can also exhibit NIR activity, we increased the concentration of ascorbate to quickly reduce not only the CuII, but also the heme FeIII. As shown in Figure 6 A, the Soret band changed over time from 409 to 421 nm. Two visible bands appeared at ∼548 and ∼580 nm, indicating a conversion of the met form to the NO‐bound ferrous form (NO−Fe2+).21 No observation of the ferrous heme in the reaction further suggested fast reduction of NO2 − and binding of NO. With 1 equiv CuII added, the observed rate constant (k obs) of formation of the NO−Fe2+ complex ((8.73±0.05)×10−3 s−1) is ∼1.7‐fold higher than the protein alone (5.19±0.03)×10−3 s−1). A ∼1.4‐ and ∼1.1‐fold k obs was also observed for R118M Mb and WT Mb, respectively, upon CuII binding (Figure S5 in the Supporting Information). These results indicate that the NO produced at the Cu site may compete with NO2 − and rapidly bind to the ferrous heme center. Since the NO produced at the heme site will also readily bind to the ferrous heme center in situ, we did not expect too much for the enhancement of k obs contributed by the Cu site. Moreover, the NO product was confirmed by EPR studies (Figure 6 B and inset of Figure S5 in the Supporting Information), which exhibited characteristic signals of a nitrosyl ferrous heme complex (g 1∼2.078, g 2∼2.027, g 3∼1.988), similar to previous observations (g 1∼2.08, g 2∼2.01, g 3∼1.98).22 The enhanced intensity of the signals further indicates that both the Cu and heme sites are reactive.

Figure 6

A) UV/Vis spectral changes for formation of NO–heme iron(II) complex upon nitrite (5 mm) reduction by CuII‐R118H Mb (5 μm) with ascorbate (12 mm). Kinetic traces at 421 nm are shown as an inset. B) EPR spectra of nitrite reduction by R118H Mb (0.18 mm) in the presence (i) and absence (ii) of 1 equiv CuII for 5 min.

In conclusion, we demonstrated that dual active sites can be designed in Mb by retaining the native heme site and creating a remote copper‐binding site through a single mutation. This study presents a new example for rational protein design with multiple active sites in a single protein scaffold. In addition, the NIR activity was demonstrated for both the heme and Cu sites, making it possible for this study to lay the groundwork for further investigation of the structure and function relationship of heme/non‐heme proteins using the single protein scaffold, by redesigning not only the heme site, but also the Cu site such as introducing other metal ions.

Experimental Section

Experimental details, X‐ray crystal structure of CuII‐loaded F33Y CuBMb, [15] MALDI‐TOF mass spectra, EPR simulation, and extra UV/Vis, EPR spectra are provided in the Supporting Information.

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supplementary

Click here for additional data file.