A new structural model for NiFe hydrogenases: an unsaturated analogue of a classic hydrogenase model leads to more enzyme-like Ni—Fe distance and interplanar fold

The new structural NiFe hydrogenase model [Ni(L′)FeCp*(CO)][PF6] [L′ = S—C(Me)=C(Me)—S—(CH2)3—S—C(Me)=C(Me)—S] is reported.


Chemical context
Since the discovery and structural elucidation of nickel-iron hydrogenases, synthetic chemists have worked towards closer ISSN 2056-9890 and closer structural models for the NiFe hydrogen-splitting active site (Lubitz et al., 2014). This active site contains two terminal sulfur donors and two bridging sulfur donors coordinated to nickel, as well as a pseudo-octahedal coordination sphere around iron, which is completed by cyano and carbonyl ligands ( Fig. 1, left). Several closely related models of the active site have been prepared by combining an Ni('S 4 ') fragment ('S 4 ' = dianionic tetradentate sulfur ligand) with an [FeCp R (CO)] + fragment (Cp R = Cp, C 5 H 5 or Cp*, C 5 Me 5 ), as illustrated in Fig. 1 (right) (Canaguier et al., 2010;Yang et al., 2015;Zhu et al., 2005). These complexes have an overall mono-cationic charge, consistent with formal Ni II and Fe II oxidations states. The first 'S 4 ' ligand used in this capacity featured a saturated two-three-two carbon linker, in L 2À = [S-CH 2 -CH 2 -S-(CH 2 ) 3 -S-CH 2 -CH 2 -S] 2À (Fig. 2, left) (Yang et al., 2015;Zhu et al., 2005).
In the following discussion, we compare the structural features obtained with the unsaturated ligand L 02À with those    of literature complexes using the saturated ligand L 2À . The structures of [Ni(L)FeCp(CO)] + , as the PF 6 À salt/ CH 2 Cl 2 solvate (Zhu et al., 2005), and [Ni(L)FeCp*(CO)] + , as the PF 6 À salt (Yang et al., 2015), are known. Both saturated analogues [Ni(L)FeCp(CO)] + and [Ni(L)FeCp*(CO)] + show Ni-Fe distances that are similar for the two, 3.1727 (6)/ 3.1529 (7) Å (two independent molecules in the unit cell) and 3.111 (5) Å , respectively, for the two complexes. The [Ni(L 0 )FeCp*(CO)] + complex, on the other hand, has a much shorter Ni-Fe distance [2.9195 (8), see above]. Also, [Ni(L)FeCp(CO)] + and [Ni(L)FeCp*(CO)] + show interplanar fold angles that are similar for the two, 39.56 (5)/41.99 (5) (two independent molecules in the unit cell) and 47.22 (9) , respectively, while [Ni(L 0 )FeCp*(CO)] + has a much larger fold angle of 64.85 (6) (see above). The large fold angle and short Ni-Fe distance observed in the complex with the unsaturated ligand L 0 match the structure of the enzymatic active site more closely than the angles/distances of the complexes containing the saturated ligand L. For eight structurally characterized enzymes, the dihedral angles range from 59 to 99 and the Ni-Fe distances range from 2.53 to 2.97 Å (one outlier being desulfovibrio fructosovorans with 46 and 3.23 Å ; Zhu et al., 2005). We have thus provided evidence that unsaturation in an 'S 4 '-ligand of the type (S-C 2 -S-C 3 -S-C 2 -S) 2À can increase structural resemblance to the enzyme in models of the type [Ni('S 4 ')FeCp R (CO)] + . Structural similarity to the enzyme in models was, in alternative approaches, also favoured when additional donor atoms were incorporated into the ligand chain (such as 'S 3 N 2 ') or where two bidentate chelate ligands were used instead of one large 'S 4 ' ligand. (Zhu et al., 2005) Within the context of linear 'S 4 ' ligands, an [Ni(L 00 )FeCp*(CO)] + model with four carbon atoms, instead of three, in the remote portion of the backbone (see L 002À in Fig. 2, right) provided an Ni-Fe distance and fold angle very similar to those of the L 0 analogue, of 2.9611 (8) Å and 62.48 (4) , respectively (Canaguier et al., 2010). In terms of activity, [Ni(L 00 )FeCp*(CO)] + was shown to be active as a hydrogen-production catalyst (Canaguier et al., 2010), which suggests that the [Ni(L 0 )Cp * (CO)] + complex, with the unsaturated 'S 4 ' ligand L 0 , might warrant deeper investigation. We conclude that the introduction of unsaturation in the 'S 4 ' ligand led to a better structural model relative to the unsaturated ligand, highlighting a new variant of the classic [Ni('S 4 ')FeCp R (CO)] + -type hydrogenase model.

Database survey
The Cambridge Crystallographic Database (version 5.39 including updates up to February 2018; Groom et al., 2016) was surveyed. A search was performed aimed at finding Ni 1 Fe 1 complexes that contain at least one (possibly substituted) cyclopentadienyl unit, at least one carbonyl (CO) coordinated to iron, and a nickel center bonded to at least four sulfurs. The substructure that was used for the search contained a cyclo-C 5 unit (any type of bond allowed), a nickel atom bonded to four sulfur atoms (any type of bond allowed), as well as an Fe-C-O unit (any type of bond for Fe-C and for C-O). Out of the six hits, RULQEV, RULQOF and RULQUL are trimetallic (instead of dimetallic) complexes (and also do not contain a cyclopentadienyl but rather a saturated five-membered ring within a polycyclic structure). Since they are not very close analogues of [Ni(L 0 )FeCp*(CO)] + , they are not discussed further. LAZVUE (Zhu et al., 2005) contains [Ni(L)FeCp(CO)] + (as the PF 6 À salt, CH 2 Cl 2 solvate), MUDXOA (Yang et al., 2015) contains [Ni(L)FeCp*(CO)] + (as the PF 6 À salt), and SUWWAJ (Canaguier et al., 2010) contains [Ni(L 00 )FeCp*(CO)] + (as the BF 4 À salt, CH 2 Cl 2 solvate). These three complex cations are discussed in detail above.

Synthesis and crystallization
The syntheses were performed in dried solvents under an inert atmosphere (nitrogen or argon; vacuum) using standard glovebox (MBraun) and Schlenk techniques. Deuterated NMR solvents were from Cambridge Isotopes. [Cp*Fe(CO) 2 ] 2 was acquired from Alfa Aesar. All other chemicals were obtained from Sigma-Aldrich. Photolysis was performed using a 160 W mercury vapour lamp (model: Westron Mega-Ray Self-Ballasted Zoologist).
Ni(S 2 C 2 Me 2 ) 2 : This precursor for the nickel part of the complex was prepared as described in the literature (Schrauzer & Mayweg, 1965).

Fractional atomic coordinates and isotropic or equivalent isotropic displacement parameters (Å 2 )
x y z U iso */U eq