Structure of dimerized assimilatory NADPH-dependent sulfite reductase reveals the minimal interface for diflavin reductase binding

doi:10.21203/rs.3.rs-4758050/v1

Download PDF

Article

Structure of dimerized assimilatory NADPH-dependent sulfite reductase reveals the minimal interface for diflavin reductase binding

https://doi.org/10.21203/rs.3.rs-4758050/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Escherichia coli NADPH-dependent assimilatory sulfite reductase (SiR) fixes sulfur for incorporation into sulfur-containing biomolecules. SiR is composed of two subunits: an NADPH, FMN, and FAD-binding diflavin reductase and an iron siroheme/Fe₄S₄ cluster-containing oxidase. How they interact has been unknown for over 50 years because SiR is highly flexible, thus has been intransigent for traditional X-ray or cryo-EM structural analysis. A combination of the chameleon plunging system with a fluorinated lipid overcame the challenge of preserving a dimer between the subunits for high-resolution (2.84 Å) cryo-EM analysis. Here, we report the first structure of the reductase/oxidase complex, revealing how they interact in a minimal interface. Further, we determined the structural elements that discriminate between pairing a siroheme-containing oxidase with a diflavin reductase or a ferredoxin partner to channel the six electrons that reduce sulfite to sulfide.

Biological sciences/Structural biology

Biological sciences/Structural biology/Electron microscopy/Cryoelectron microscopy

NADPH-dependent assimilatory sulfite reductase

hemoflavoprotein

oxidoreductase

cryo-EM

Assimilatory sulfite reduction by NADPH-dependent sulfite reductase (SiR) is essential to produce sulfide for incorporation into sulfur-containing biomolecules. In g-proteobacteria like Escherichia coli, SiR is a multimeric oxidoreductase composed of an octameric diflavin reductase (SiRFP) and four independently binding subunits of a siroheme-containing hemoprotein (SiRHP)^1-3.

The E. coli SiRFP subunit is homologous to cytochrome P450 (CYP) reductase (CPR)⁴, the reductase domain of the bacterial CYP/CPR fusion CYP102A1/CYPBM3⁵, the reductase domain of nitric oxide synthase (NOSr)^6,7, and methionine synthase reductase (MSR)⁸. One of the hallmarks of this diflavin reductase family is that they are exceptionally conformationally malleable, which makes structural analysis challenging. For example, to date there are no high-resolution structures of the full-length NOS homodimer, the complex between methionine synthase and MSR, or the SiR heterododecameric holoenzyme (for simplicity, here referred to as a dodecamer). The structure of the CYP/CPR heterodimer and CYPBM3 are known^9-11, but CYP/CPR form a 1:1 heterodimer, whereas the oxidase and reductase domains are fused in CYPBM3, so little can be inferred about other homologs that function as higher-order protein complexes.

Despite its homology to other well-studied diflavin reductases, SiRFP is unique because it assembles into an octamer through its N-terminal 52 amino acids¹². Removing those amino acids results in a 60 kDa monomer (SiRFP-60), which binds SiRHP as a 1:1 heterodimer with reduced activity³. Further, removing the complete N-terminal FMN-binding flavodoxin (Fld) domain results in a 43 kDa monomer that contains just the NADPH- and FAD-binding NADP⁺ ferredoxin reductase (FNR) domain (SiRFP-43), which also binds SiRHP as a 1:1 heterodimer but is inactive for electron transfer¹³ (Abbreviations and theoretical molecular weights are summarized in Table S1).

SiRHP has few known homologs because of its unique siroheme/Fe₄S₄ cluster assembly that form the sulfite-binding active site¹⁴. Assimilatory SiRs or siroheme-dependent nitrite reductases (NiRs) from other bacterial species or plants have a similar hemoprotein but use a transiently bound ferredoxin as their electron source^15-17. SiRs that are responsible for energy conversion, dissimilatory sulfite reductases (DSRs), share a common siroheme binding fold but are heterotetrameric and are often fused to auxiliary domains¹⁸. Their electron donors are poorly understood. Here, we show the first high-resolution cryogenic electron microscopy (cryo-EM) structure of the minimal SiRFP/SiRHP dimer, which elucidates their binding interface to understand how SiRHP tightly binds SiRFP’s FNR domain.

The SiRHP-SiRFP interaction is highly sensitive to cryogenic TEM (cryo-EM) preparation

We determined the structure of the SiRFP/SiRHP dimer from three modified, minimal dimers, each of which is named by the change to SiRFP and its resulting molecular weight (Table S1). First, we truncated SiRFP to remove the N-terminal Fld domain (SiRFP-43/SiRHP). This is an inactive dimer, as the Fld domain is required for electron transfer. Nevertheless, the two subunits bind tightly and is the most simplified complex between SiRFP and SiRHP¹⁹. Second, we truncated both SiRFP’s N-terminal octamerization domain as well as the linker between the Fld and FNR domains to create a monomeric SiRFP that can be locked in an open position (SiRFP-60D/SiRHP)²⁰. Third, we generated a variant of monomeric SiRFP-60 lacking reactive cysteines into which we engineered a disulfide bond between the Fld and FNR domains (SiRFP-60X/SiRHP)²¹.

Each variant is highly sensitive to traditional blotting plunge-freezing methods for cryo-EM preservation. To overcome this sensitivity, we combined the protection of a high critical micelle concentration, fluorinated lipid, fos-choline-8 (FF8, Creative Biolabs, Shirley, NY, USA )²², with the blot-free, rapid plunging afforded by the chameleon system (SPT Labtech, Melbourn, UK)²³. This cryo-EM sample preparation helped us to retain each intact complex within near ideal ice thickness and avoid denaturation at the air water interface (Fig. S1). The smallest complex (SiRFP-43/SiRHP) showed well-aligned 2D class averages, however the 3D structure revealed structural anisotropy, either due to its small size/asymmetric geometry or from a preferred orientation, that limited high-resolution analysis despite the absence of mobile elements (Figs. S1A and S2). SiRFP-60D/SiRHP showed moderate-resolution density (3.54 Å) for the SiRFP FNR domain and SiRHP, however the N-terminal Fld domain was not visible (Figs. S1B and S3A). The 2.84 Å-resolution structure of SiRFP-60X/SiRHP revealed the most detail for SiRFP’s Fld and FNR domains, despite a lack of density for the linker between SiRFP’s domains in the highest-resolution reconstruction (Figs. 1A, S1C, and S3B). High resolution features for each of the cofactors in both subunits supported this reported resolution (Fig. S4). Therefore, we analyzed the SiRFP-SiRHP interface for this construct in detail.

SiRFP-SiRHP binding

The SiR dodecameric holoenzyme is composed of oligomers of the dimers discussed here and is about 800 kDa in mass. Despite this large mass, the binding interface between the minimal SiRFP/SiRHP dimer is small, reaching the surface area of 1,138 Å² relative to the overall surface of 43,610 Å². SiRHP alone has a solvent-exposed surface of 25,680 Å². SiRFP-60X alone has a solvent-exposed surface of 21,930 Å². That is, for a large complex only about 2.6% of the solvent-exposed surface is buried upon subunit binding. This is consistent with hydrogen-deuterium exchange data on the complex that reveals single, short peptides from each subunit that become occluded upon binding³, the sequences of which predicted the interface would be dominated by hydrophobic interactions (Fig. 1B).

The interface is governed by the N-terminus of SiRHP. The structure of this region is previously uncharacterized as it is proteolytically removed in the X-ray crystal structure of E. coli SiRHP²⁴. These 80 amino acids follow the topology helix 1 - loop - helix 2 - turn - helix 3 - b-strand 1 - helix 4 - loop - b-strand 2 (Fig. 1C). Only amino acids from the turn, helix 4, and surrounding loops directly interact with SiRFP. The regions that are N-terminal to the interface interact with domain 1 or the N-terminal half of the parachute domain (i.e. the first sulfite or nitrite reductase repeat (S/NiRR)^24,25), breaking SiRHP’s pseudo two-fold symmetry (Figs. 1A and S5A). An extension to the parachute domain that is not ordered in the original crystal structure, from amino acids 184–209, helps to hold the N-terminus in place (Fig. S5A). Those that are N-terminal to the interface approach the distal active site, but do not contribute significantly to anion or siroheme binding as they are held back by interactions to SiRFP, discussed below. Moving N-terminally along the peptide, it then turns back to form the loop that binds a pocket in SiRFP before moving away from SiRFP. The N-terminal most amino acids reach all the way to the other side of SiRHP, interacting with the N-terminus of the a-helix (h11) that precedes the linker that joins the two S/NiRRs and mimics the siroheme binding site (Fig. S5B)²⁴.

The central interaction that pegs the subunits together is a p-cation interaction between h-Lys27 from SiRHP (for simplicity, amino acids from SiRHP will be designated with the prefix “h-“) and f-His258 from SiRFP (similarly, amino acids from SiRFP will be designated with the prefix “f-“) (Figs. 1D and E). This interaction is buttressed by f-Phe496 and f-Val500, which have previously been shown essential for SiRFP-SiRHP binding¹³. Further hydrophobic and p-stacking interactions dominate the interface. For example, h-Leu40 inserts into a pocket in SiRFP formed between f-b-sheet 17 and f-a-helix 18, which includes f-Phe496. h-Ile65 Cg2 sits 3.3 Å from the plane of f-Arg250’s guanidinium group (Figs. 1D and F), which is rotated 90^o from its position in free SiRFP²⁰. h-Gln72Cg also packs into a pocket formed by the backbone atoms of f-Ile247, f-Thr248, and f-Gly249, pinned in place by the h-Ile65/f-Arg250 and h-Lys73/f-His258 interactions. Farther from the interface, there is another stacking interaction between the guanidinium group from h-Arg63 and the h-Phe437 aromatic ring that stabilized the deformed helix that includes h-Ile65 (Figs. 1D and G). h-Phe437 is adjacent, through h-A443, to SiRHP’s iron-sulfur cluster. In this way, the binding interface between SiRFP and SiRHP reaches to the SiRHP active site through a network of hydrophobic interactions (Fig. S6).

Neither ionic interactions nor hydrogen bonds play a direct role in the interface, but rather stabilize the amino acids and structural elements that mediate the interface (Figs. 1C, D and S6). For example, a hydrogen bond network from f-Thr404Og through f-Tyr498OH and finally to f-His258Nd1 position its imidazole ring for the interaction with h-Lys73. An ionic interaction between h-Lys127Nz and h-Asp38Od2 reach across the loop, presenting h-Leu40 to project into SiRFP’s pocket. An additional ionic interaction between h-Asp61d1 and h-Arg66NHd also stabilize the deformed helix that contains h-Ile65 and turn it towards f-Arg250.

The SiRHP interaction with SiRFP differs from SIR interaction with ferredoxin

g-proteobacteria couple a diflavin reductase, SiRFP, to SiRHP. In contrast, other organisms like Zea mays and Mycobacteria tuberculosis use a ferredoxin (Fd) as their reductase partner^15,17. In Z. mays SIR, Fd bridges SIR’s C-terminal domain 2 to a loop between the first two b-strands (amino acids Asp110 to Asn118), positioning the Fd iron-sulfur cluster near the SIR metal sites²⁶. The interaction is bolstered by en face stacking between Fd Tyr37 and SIR Arg324 (Fig. 2A).

In SiRHP, the equivalent element between the structurally conserved b-strands 1 and 2 is considerably longer, stretching from h-Asp62 to h-Arg77 and containing a short helix from h-Arg63 to h-Glu71 (Fig. 2B). This loop contains the critical residues h-Gln72 and h-Lys73 that anchor the interaction with SiRFP’s FNR domain, which faces away from where Fd binds to the Z. mays homolog (Fig. 2C). Further, the arginine in SIR that stacks with Fd Tyr37 is not conserved in SiRHP – the equivalent position is h-Gly262, despite the structural conservation of the loop between b-strands 7 and 8 that contributes to siroheme binding (Fig. 2D).

The SiRHP active site loop is locked in its closed conformation

When bound to SiRFP, SiRHP’s anion binding loop (h-Asn149 to h-Arg153) is in its closed position, held in place by a long, through-space interaction between h-Arg53 and h-Asn149 (Fig. 3A). h-Arg53 is, in turn, held in place by stacking between its guanidinium group and h-Tyr58. The ring of h-Tyr58 subsequently sits over the methyl group on the siroheme pyrroline A ring. The only other new protein/siroheme interaction identified in this now complete structure of SiRHP is an ionic bond between h-Gln60 and the propionyl group from siroheme pyrroline ring B. The siroheme is saddle-shaped, as in free SiRHP and unlike in dissimilatory SIR^27–30. h-Arg153 is flipped away from the bound phosphate. The other three anion binding amino acids, h-Arg83, h-Lys215, and h-Lys217, remain largely unchanged from free SiRHP (Figs. 3A-C)²⁴.

This conformation differs from the various redox and anion-bound structures of free SiRHP^24,31–33. In the phosphate-bound, free SiRHP structure that lacks the N-terminal 80 amino acid extension²⁴, the loop is flipped open such that h-Ala146-h-Ala148 are disordered (Fig. 3B). Upon reduction and sulfite binding, h-Arg153 flips over to interact with the smaller anion and the loop becomes ordered³². h-Asn149 points away from the active site. In this way, SiRFP binding to SiRHP, with the ordering of SiRHP’s N-terminus, induces an intermediate structure with elements of both the oxidized, phosphate bound and the reduced, sulfite bound conformations (Fig. 3D).

In the original SiRHP X-ray crystal structure, the siroheme iron is significantly domed above the siroheme nitrogens, indicative of an oxidized Fe³⁺ (Fig. 3B)²⁴. Subsequent chemical reduction experiments show the doming flattens upon conversion to Fe²⁺, commensurate with release of the phosphate to allow substrate binding (Fig. 3B)³². Contemporary X-ray diffraction experiments show that this reduction is beam-induced, but within the constraints of the crystal the phosphate remains bound to the siroheme iron in the active site³³. In this cryo-EM structure, the siroheme iron appears to be in the plane of the siroheme nitrogens, suggesting that it has also been reduced by the electron beam. The central density for the phosphate is 3.5 Å from the siroheme iron and its pyramidal shape is rotated such that the oxygen-iron bond is broken (Figs. 3A and S4C).

SiRFP is highly mobile

Although the 2D class averages in all three datasets appeared to show little orientation preference with discernable features, further analysis revealed each to have unique properties related to the SiRFP variant used to generate the dimer (Table S1, Figs. S1-3).

SiRFP-43/SiRHP (107 kDa in mass): The 2D class averages for SiRFP-43/SiRHP appeared to show high-resolution features, however the initial 3D models were poorly aligned, likely due to a combination of small mass, preferred orientation, and limitations in grid preparation, so the refined volume did not achieve high resolution despite its simplified form. 3D variability analysis in CryoSPARC³⁴ did not reveal significant conformational mobility, however orientation analysis and calculation of the 3DFSC³⁵ showed the particles harbored a preferred orientation (Fig. S2). Nevertheless, the absence of SiRFP’s Fld domain did not alter the SiRFP-SiRHP interaction.

SiRFP-60D/SiRHP (123 kDa): As with SiRFP-43/SiRHP, this dimer showed 2D class averages with high-resolution features for the core of the dimer (Figs. S1B and S3A). Nevertheless, the initial models were of inconsistent structure, so we checked for heterogeneity using “3D variability” in CryoSPARC, which revealed a distinctive degree of movement in the Fld domain. To better understand the degree of flexibility for the Fld domain from the SiRFP-60D/SiRHP dimer, we performed both “3D flex” in CryoSPARC as well as “heterogenous refinement” in cryoDRGN³⁶ with the particles from refinement on the main heterodimer body. This analysis, anchored on the SiRFP-FNR/SiRHP dimer, identified a dramatic movement of SiRFP’s Fld domain relative to the FNR domain, swinging 20^o between the most compact and most open forms (Video S1 and Fig. 4A). The density for the linker joining the domains is not visible. In its most compact conformation, the Fld domain reaches the canonical “closed” conformation in which the Fld domain tucks into a cavity in the FNR domain (Fig. 4B). In the most open conformation, the Fld domain assumes a different position to that seen in the “open” conformation determined by X-ray crystallography of the same monomeric variant and the average solution envelope determined from SANS, intermediate between the fully opened and closed conformations (Fig. 4C)^19,20.

SiRFP-60X/SiRHP (124 kDa): The Fld and FNR domains are covalently linked in this dimer, thus restricting the swinging domain motion of the Fld domain seen in the SiRFP-60D/SiRHP dimer. No preferred orientation was identified either in 2D or 3D analysis (Fig. S3B). Nevertheless, there is no density for the linker between the Fld and FNR domains in the highest-resolution reconstruction, which suggests that the 36 amino acid linker is exceptionally dynamic but allows SiRFP’s two domains to come into close contact to catch them in a crosslink. Locking the Fld and FNR domains in the closed conformation resulted in the most stable variant for high-resolution analysis (2.84 Å), discussed above.

With further classification of the SiRFP-60X/SiRHP structure, a sub-class of ~ 20,000 particles (10% of the particles used for highest resolution reconstruction for this dataset) emerged that revealed the linker between the two domains. The linker runs from the C-terminal end of the Fld domain (amino acid f-Ser207) to the N-terminal end of the FNR domain (amino acid f-Ile226) (Fig. 4D). There is neither defined secondary structure nor contact with either of the SiRFP domains, thus the linker is not constrained in a single conformation across the entire ensemble of particles even when the domains are crosslinked. The highly mobile nature of these 36 amino acids explains how the Fld domain can reposition for contact with the oxidase partner, SiRHP, presumably to mediate electron transfer.

The interface between the SiRHP and SiRFP subunits of the SiR holoenzyme is surprisingly minimal, driven by p-cation, hydrophobic, and planar stacking between aromatic amino acids. The non-canonical planar stacking interactions with strategic arginine residues are not as uncommon as would be expected from the canonical role that arginine plays in electrostatic interactions and are often found at important structural elements of the protein or enzyme³⁷. In the case of the SiRFP/SiRHP dimer, these interactions either mediate the dimer interface itself or establish the structural platform for other amino acids to mediate the interface. Polar or electrostatic interactions play supporting roles to align the non-polar interactions because the arginine amino groups are free to make independent, more canonical, interactions with other side chains.

The curious combination of non-ionic interactions that holds the complex together explains why cryo-EM analysis has been elusive – the hydrophobic air-water interface in a thin film for cryogenic plunging is destructive for such interactions^38–41. We performed extensive trial-and-error probing grid type, grid substrate, hole size, method for hydrophilizing the grid, detergents, and plunging methods to identify what seems to be a singular condition that allowed cryogenic preservation for cryo-EM analysis: the combination of FF8 and the chameleon blot-free, rapid plunging. Neither FF8 alone with traditional plunging nor chameleon alone without FF8 was sufficient. Such low-throughput screening to find the one combination of variables that allows for high-quality sample preparation highlights the need for further technology development on the front-end of cryo-EM structure determination. High-throughput data collection with direct electron detectors is not sufficient to overcome cryo-EM sample preparation challenges.

Surprisingly minimal structural elements determine the evolutionary pressure for siroheme-dependent SiR oxidases to partner with a diflavin reductase like SiRFP or a simpler Fd reductase. For example, in comparing the only known structures of monomeric assimilatory SiR oxidases with their reductase partners – E. coli SiRHP (this work) and Z. mays SIR²⁶ – a single loop extension between a core two-strand b-sheet discriminates between Fd binding in the case of Z. mays SIR and the binding of SiRHP to the FNR domain of SiRFP in the case of E. coli. E. coli has only one fdx gene that does not fulfill the reductase function for SiRHP, given that strains of E. coli lacking the gene encoding SiRFP (cysJ) cannot grow in a low sulfur environment¹⁴.

The SiRFP/SiRHP dimeric structure reported here shows that when SiRHP binds SiRFP, the anion-binding active site loop is locked in its closed conformation, constrained from disorder. Nevertheless, h-Arg153 is in its phosphate-binding conformation. The position of the loop is, thus, impacted by the presence of SiRHP’s N-terminus that is absent in the X-ray crystal structure. The importance of this observation is understood in comparison to a series of X-ray crystal structures in various redox states and bound to various anions^31,32. This series of structures shows how the active site re-orients with the changing state, namely that the loop between h-Asn149 to h-Arg153 is disordered in the oxidized state. Upon reduction the loop becomes ordered and h-Arg153 flips to bind the substrate.

The question that remains is how SiRFP’s Fld domain approaches SiRHP’s metallic active site for productive electron transfer as it does not mediate the structural interaction between the subunits. Clearly there is dramatic flexibility within SiRFP, even when the Fld and FNR domains are covalently attached by an engineered crosslink – we cannot visualize the linker between the domains without extensive classification. Within this minimal dimer, the crosslink is intramolecular but in the SiR dodecameric holoenzyme, we have observed both intermolecular and intramolecular crosslinks²¹. Further, despite our extensive efforts at determining the structure of the full complex, the dodecamer has proven to be highly heterogeneous. Together, these observations support the hypothesis that within the holoenzyme, there is not a singular inter-subunit interaction between the Fld and FNR domains of any given SiRFP subunit within the octamer.

The basis for SiRFP’s extreme flexibility is that the linker between the Fld and FNR domains is 36 amino acids long, compared to the 12 amino acid long linker in CPR⁴². Heterogeneity analysis of SiRFP-D in complex with SiRHP identified a dominant vector of motion parallel to the major axis of the FNR domain. This is in contrast to the highly extended conformation seen in the X-ray crystal structure of free SiRFP-D²⁰. In this way, there is not a single axis along which the Fld domain moves relative to the FNR domain such that the Fld domain makes a productive interaction with SiRHP within a SiRHP/SiRFP dimer. Nevertheless, prior biochemical experimentation shows that cross-subunit interactions occur within the SiRFP octamer²¹. Thus, it is likely that a unique, single electron transfer-pathway does not exist within the full, dodecameric holoenzyme complex.

The combined technological developments in cryo-EM over the past decade all played a role in determining the structure of this elusive complex, from cryogenic sample preservation to data collection to image analysis. In doing so, we show that the subunits of NADPH-dependent assimilatory sulfite reductase bind through an interface governed by the N-terminus of SiRHP and the FNR domain of SiRFP. The interaction is surprisingly minimal, governed by a set of hydrophobic and stacking interactions. Structural pairing between a siroheme-dependent oxidase and diflavin reductase or ferredoxin appears to fall to a short loop between a conserved 2-stranded b sheet. The high mobility of SiRFP’s Fld domain relative to its FNR domain likely explains how a minimal dimer maintains redox-dependent functionality and provides a mechanism for redundant electron transfer pathways within the dodecamer that efficiently performs the six-electron reduction of sulfite to sulfide.

Protein Expression and Purification

Each SiRFP/SiRHP dimer variant was generated and purified as previously described^2,3,13,19,21. Briefly, untagged SiRHP was co-purified with the following SiRFP variants: 1) SiRFP-43, a 43 kDa monomeric SiRFP variant lacking the N-terminal Fld domain and linker¹⁹; 2) SiRFP-60D, a 60 kDa monomeric variant of SiRFP generated by truncating its first 52 amino acids with an additional internal truncation of six amino acid (D-AAPSQS) in the linker that joins the Fld and FNR domains²⁰; and 3) SiRFP-60X, the 60 kDa monomer with an engineered disulfide crosslink between the Fld and FNR domains, as has previously been analyzed in octameric SiRFP²¹. In SiRFP-60X, two background cysteines (C162T and C552S) were altered to avoid unwanted crosslinking and two cysteines were added (E121C and N556C) to form a disulfide bond between Fld and FNR domains of SiRFP, previously confirmed by mass spectroscopy analysis²¹. DNA sequencing confirmed the presence of all deletions and mutations in the variants.

E. coli LMG194 cells (Invitrogen, Carlsbad, CA, USA) were transformed with the pBAD plasmid containing the genes encoding SiRFP-60D, SiRFP-60X, SiRFP-43 or SiRHP. All variants were expressed independently. Dimers were formed as followed: for SiRFP-43/SiRHP and SiRFP-60D/SiRHP, cells were mixed before lysis and the dimers co-purified. The SiRFP-60X/SiRHP dimer was assembled by reconstituting SiRFP-60X with 2 Eq SiRHP for 1 h on ice, as before^2,19,21. A six-histidine tag was present in all SiRFP variants whereas SiRHP was untagged. Each recombinant E. coli strain was grown and induced at 25°C with 0.05% L-arabinose. SiRFP-43, SiRFP-60D and SiRHP cells were expressed for 4 hr whereas SiRFP-60X was expressed overnight. All variants were purified using a combination of Ni-NTA affinity chromatography (Cytiva, Marlborough, MA, USA), anion exchange HiTrap-Q HP chromatography (Cytiva, Marlborough, MA, USA) and Sephacryl S300-HR size exclusion chromatography (Cytiva, Marlborough, MA, USA) using previously optimized SPG buffers (17 mM succinic acid, 574 mM sodium dihydrogen phosphate, pH 6.8, 374 mM glycine, 200 mM NaCl)^2,13,19,21. All variants have been extensively characterized with UV-Vis spectroscopy, size exclusion chromatography, SDS PAGE analysis for correct stoichiometry and with SANS for solution monodispersity (Fig. S7)^2,13,19,21.

Cryo-EM sample preparation

SiRFP-43/SiRHP, SiRFP-60D/SiRHP and SiRFP-60X/SiRHP were all plunged using the chameleon blotless plunging system (SPT Labtech, Melbourn, UK) at 10 mg/mL protein. chameleon self-wicking grids²³ were backed with 18-gauge gold (Ted Pella, Redding CA, USA) on an Auto 306 vacuum coater (BOC Edwards, West Sussex, UK). The following glow discharge (GD)/wicking times (WTs) were used: SiRFP-43/SiRHP: 30 s GD, 130 ms WT; SiRFP-60D/SiRHP: 80 s GD, 195 ms WTs; SiRFP-60X/SiRHP 45 s GD, 175 ms WTs. All samples were premixed with FC-8 detergent²² at 2 mM final concentration. The plunged grids were clipped and then screened for high quality with a Titan Krios operating the Leginon software package⁴³.

Data collection and processing

SiRFP-43/SiRHP: 14,600 movies were collected at 300 KV on a Titan Krios using a GATAN K3 camera with 0.844 Å/pixel. After motion correction with Motioncor2⁴⁴ in the Relion GUI⁴⁵, CTF estimation was performed using CTFFIND4⁴⁶. Particles were picked by “blob picker” followed by “template picker” in CryoSPARC³⁴. Initial 2D classification, followed by multiple rounds of 2D class selection/classification, identified 1,500,000 particles that were used for initial map building and non-uniform refinement in CryoSPARC. This process achieved a reported 3.6 Å-resolution map of the minimal dimer. However, the map features did not reflect the reported resolution. “Orientation diagnostic” job from CryoSPARC was performed to confirm this interpretation. To further confirm and diminish the preferred orientation issue, “Rebalance 2D” was performed to put particles into 7 super classes and limit the total particles in each superclass down to total of 350,000 and 105,000 particles. Non-uniform refinement⁴⁷ followed by orientation diagnostics performed in CryoSPARC produced a more accurate resolution (4.31 Å, 4.74 Å respectively) and better quality map. deepEMhancer⁴⁸ was used to sharpen these maps (Fig. S2). Particle picking performed by the TOPAZ algorithm⁴⁹ gave the same result.

SiRFP-60D/SiRHP: 25,488 movies were collected at 300 KV on a Titan Krios using a GATAN K3 camera with 0.844 Å/pixel. Motion correction, CTF estimation, and particle picking were performed as for SiRFP-43/SiRHP. After multiple rounds of 2D classification, ~ 550,000 particles were used for initial map building and non-uniform refinement in CryoSPARC to achieve the final 3.49 Å-resolution structure of the dimer, masked to omit SiRFP’s Fld domain. Multiple refinements with different masking were performed. Masking to include the whole complex, including SiRFP’s Fld domain, resulted in a low resolution reconstruction for the Fld domain. Particles were then down sampled and imported to cryoDRGN to perform heterogenous refinement, giving a series of volumes showing the Fld domain movement. CryoSPARC 3D Flex⁵⁰ gave the same result as cryoDRGN.

SiRFP-60X/SiRHP: ~10,000 movies were collected at 300 KV on a Titan Krios using a DE Apollo camera with 0.765 Å/pixel. Motion correction, CTF estimation, and particle picking were performed as above. Multiple rounds of 2D classifications identified ~ 185,000 particles that were used for initial map building and non-uniform refinement in CryoSPARC to achieve a 2.84 Å-resolution structure of the entire dimer, including SiRFP’s Fld domain. Further 3D variability was performed to track the Fld linker. After multiple rounds of 3D classification 10% of the particles (~ 20,000) were used to perform a non-uniform refinement, resulted in a 3.61 Å overall resolution structure, sharpened by deepEMhancer including the linker.

Model building and refinement.

Model building was initiated using the “fit in map” algorithm in Chimera⁵¹ using the atomic model of SiRFP from PDB ID 6EFV²⁰, fitting each of the Fld or FNR domains independently or the atomic model of SiRHP generated in AlphaFold^52,53 to capture its N-terminal 80 amino acids. Iterative real-space refinement in PHENIX⁵⁴ with manual fitting in Coot⁵⁵ yielded a model with a correlation of 0.85 (Table S3). The model deposited in PDB as 9C91and the map deposited in EMDB as EMD-45359. A stack from the final 3D reconstruction particles was deposited in the EMPIAR database under the accession number EMPIAR-12180.

Acknowledgements and funding

We kindly thank Dr.s Scott Stagg and Christopher Stroupe for helpful discussions. Some of this work was performed at the Simons Electron Microscopy Center at the New York Structural Biology Center, with major support from the Simons Foundation (SF349247). Florida State University supports cryo-EM in the Biological Imaging Resource Center, which houses the following equipment used in this study: a Gatan Solaris Plasma Cleaner (NIH grant S10 RR024564), a Hitachi HT7800 (NSF grant 2017869), a ThermoFisher Vitrobot Mark IV (NIH grant S10 RR024564), an SPI chameleon plunging system (NIH grant R24 GM145964), a ThermoFisher Titan Krios (NIH grant S10 RR025080), and a DE Apollo direct electron detector (NIH grant R35 GM139616). This work was further supported by National Science Foundation grants MCB1856502 and CHE1904612 to M.E.S.

Author contributions

BGE participated in data collection and performed the data analysis. NW and BGE prepared the protein specimen. KN and MA prepared cryogenic specimen and participated in data collection. IA optimized protein specimen preparation. AW and JHM supervised cryogenic specimen preparation and data collection. MES conceived of the study and oversaw experimentation.

Siegel LM, Davis PS (1974) Reduced nicotinamide adenine dinucleotide phosphate-sulfite reductase of enterobacteria. IV. The Escherichia coli hemoflavoprotein: subunit structure and dissociation into hemoprotein and flavoprotein components. J Biol Chem 249:1587–1598
Murray DT et al (2022) Neutron scattering maps the higher-order assembly of NADPH-dependent assimilatory sulfite reductase. Biophys J 121:1799–1812. https://doi.org/10.1016/j.bpj.2022.04.021
Askenasy I et al (2015) The N-terminal Domain of Escherichia coli Assimilatory NADPH-Sulfite Reductase Hemoprotein Is an Oligomerization Domain That Mediates Holoenzyme Assembly. J Biol Chem 290:19319–19333. https://doi.org/10.1074/jbc.M115.662379
Wang M et al (1997) Three-dimensional structure of NADPH-cytochrome P450 reductase: prototype for FMN- and FAD-containing enzymes. Proc Natl Acad Sci U S A 94:8411–8416
Girvan HM et al (2006) Flavocytochrome P450 BM3 and the origin of CYP102 fusion species. Biochem Soc Trans 34:1173–1177. https://doi.org/10.1042/BST0341173
Garcin ED et al (2004) Structural basis for isozyme-specific regulation of electron transfer in nitric-oxide synthase. J Biol Chem 279:37918–37927. https://doi.org/10.1074/jbc.M406204200
Zhang J et al (2001) Crystal structure of the FAD/NADPH-binding domain of rat neuronal nitric-oxide synthase. Comparisons with NADPH-cytochrome P450 oxidoreductase. J Biol Chem 276:37506–37513. https://doi.org/10.1074/jbc.M105503200
Olteanu H, Banerjee R (2001) Human methionine synthase reductase, a soluble P-450 reductase-like dual flavoprotein, is sufficient for NADPH-dependent methionine synthase activation. J Biol Chem 276:35558–35563. https://doi.org/10.1074/jbc.M103707200
Freeman SL et al (2018) Solution structure of the cytochrome P450 reductase-cytochrome. J Biol Chem 293:5210–5219. https://doi.org/10.1074/jbc.RA118.001941
Zhang L et al (2020) Structural insight into the electron transfer pathway of a self-sufficient P450 monooxygenase. Nat Commun 11:2676. https://doi.org/10.1038/s41467-020-16500-5
Zhang H et al (2018) The full-length cytochrome P450 enzyme CYP102A1 dimerizes at its reductase domains and has flexible heme domains for efficient catalysis. J Biol Chem 293:7727–7736. https://doi.org/10.1074/jbc.RA117.000600
Zeghouf M, Fontecave M, Coves J (2000) A simplifed functional version of the Escherichia coli sulfite reductase. J Biol Chem 275:37651–37656. https://doi.org/10.1074/jbc.M005619200M005619200 [pii]
Askenasy I et al (2018) Structure-Function Relationships in the Oligomeric NADPH-Dependent Assimilatory Sulfite Reductase. Biochemistry 57:3764–3772. https://doi.org/10.1021/acs.biochem.8b00446
Siegel LM, Murphy MJ, Kamin H (1973) Reduced nicotinamide adenine dinucleotide phosphate-sulfite reductase of enterobacteria. I. The Escherichia coli hemoflavoprotein: molecular parameters and prosthetic groups. J Biol Chem 248:251–264
Nakayama M, Akashi T, Hase T (2000) Plant sulfite reductase: molecular structure, catalytic function and interaction with ferredoxin. J Inorg Biochem 82:27–32
Schnell R, Sandalova T, Hellman U, Lindqvist Y, Schneider G (2005) Siroheme- and [Fe4-S4]-dependent NirA from Mycobacterium tuberculosis is a sulfite reductase with a covalent Cys-Tyr bond in the active site. J Biol Chem 280:27319–27328. https://doi.org/M502560200
Pinto R, Harrison JS, Hsu T, Jacobs WR Jr., Leyh TS (2007) Sulfite reduction in mycobacteria. J Bacteriol 189:6714–6722. https://doi.org/10.1128/jb.00487-07
Jespersen M, Pierik AJ, Wagner T (2023) Structures of the sulfite detoxifying F 420-dependent enzyme from Methanococcales. Nat Chem Biol 19:695–702. https://doi.org/10.1038/s41589-022-01232-y
Murray DT, Weiss KL, Stanley CB, Nagy G, Stroupe ME (2021) Small-angle neutron scattering solution structures of NADPH-dependent sulfite reductase. J Struct Biol 213:107724. https://doi.org/10.1016/j.jsb.2021.107724
Tavolieri AM et al (2019) NADPH-dependent sulfite reductase flavoprotein adopts an extended conformation unique to this diflavin reductase. J Struct Biol. https://doi.org/10.1016/j.jsb.2019.01.001
Walia N et al (2023) Domain crossover in the reductase subunit of NADPH-dependent assimilatory sulfite reductase. J Struct Biol 215:108028. https://doi.org/10.1016/j.jsb.2023.108028
Kampjut D, Steiner J, Sazanov LA (2021) Cryo-EM grid optimization for membrane proteins. iScience 24:102139. https://doi.org/10.1016/j.isci.2021.102139
Levitz TS et al (2022) Approaches to Using the Chameleon: Robust, Automated, Fast-Plunge cryoEM Specimen Preparation. Front Mol Biosci 9:903148. https://doi.org/10.3389/fmolb.2022.903148
Crane BR, Siegel LM, Getzoff ED (1995) Sulfite reductase structure at 1.6 A: evolution and catalysis for reduction of inorganic anions. Science 270:59–67
Crane BR, Getzoff ED (1996) The relationship between structure and function for the sulfite reductases. Curr Opin Struct Biol 6:744–756. https://doi.org/S0959-440X (96)80003-0 [pii]
Kim JY, Nakayama M, Toyota H, Kurisu G, Hase T (2016) Structural and mutational studies of an electron transfer complex of maize sulfite reductase and ferredoxin. J Biochem. https://doi.org/10.1093/jb/mvw016
Oliveira TF et al (2008) The crystal structure of Desulfovibrio vulgaris dissimilatory sulfite reductase bound to DsrC provides novel insights into the mechanism of sulfate respiration. J Biol Chem 283:34141–34149. https://doi.org/M805643200
Oliveira TF et al (2011) Structural insights into dissimilatory sulfite reductases: structure of desulforubidin from desulfomicrobium norvegicum. Front Microbiol 2:71. https://doi.org/10.3389/fmicb.2011.00071
Parey K, Warkentin E, Kroneck PM, Ermler U (2010) Reaction cycle of the dissimilatory sulfite reductase from Archaeoglobus fulgidus. Biochemistry 49:8912–8921. https://doi.org/10.1021/bi100781f
Schiffer A et al (2008) Structure of the dissimilatory sulfite reductase from the hyperthermophilic archaeon Archaeoglobus fulgidus. J Mol Biol 379, 1063–1074 https://doi.org/S0022-2836(08)00456-7 [pii] 10.1016/j.jmb.2008.04.027
Crane BR, Siegel LM, Getzoff ED (1997) Probing the catalytic mechanism of sulfite reductase by X-ray crystallography: structures of the Escherichia coli hemoprotein in complex with substrates, inhibitors, intermediates, and products. Biochemistry 36:12120–12137. https://doi.org/10.1021/bi971066i bi971066i[pii]
Crane BR, Siegel LM, Getzoff ED (1997) Structures of the siroheme- and Fe4S4-containing active center of sulfite reductase in different states of oxidation: heme activation via reduction-gated exogenous ligand exchange. Biochemistry 36:12101–12119. https://doi.org/10.1021/bi971065q bi971065q[pii]
Smith KW, Stroupe ME (2012) Mutational analysis of sulfite reductase hemoprotein reveals the mechanism for coordinated electron and proton transfer. Biochemistry 51:9857–9868. https://doi.org/10.1021/bi300947a
Punjani A, Rubinstein JL, Fleet DJ, Brubaker MA (2017) cryoSPARC: algorithms for rapid unsupervised cryo-EM structure determination. Nat Methods 14:290–296. https://doi.org/10.1038/nmeth.4169
Aiyer S, Zhang C, Baldwin PR, Lyumkis D (2021) Evaluating Local and Directional Resolution of Cryo-EM Density Maps. Methods Mol Biol 2215:161–187. https://doi.org/10.1007/978-1-0716-0966-8_8
Zhong ED, Bepler T, Berger B, Davis JH (2021) CryoDRGN: reconstruction of heterogeneous cryo-EM structures using neural networks. Nat Methods 18:176–185. https://doi.org/10.1038/s41592-020-01049-4
Flocco MM, Mowbray SL (1994) Planar stacking interactions of arginine and aromatic side-chains in proteins. J Mol Biol 235:709–717. https://doi.org/10.1006/jmbi.1994.1022
D'Imprima E et al (2019) Protein denaturation at the air-water interface and how to prevent it. Elife 8. https://doi.org/10.7554/eLife.42747
Glaeser RM et al (2016) Factors that Influence the Formation and Stability of Thin, Cryo-EM Specimens. Biophys J 110:749–755. https://doi.org/10.1016/j.bpj.2015.07.050
Glaeser RM, Han BG (2017) Opinion: hazards faced by macromolecules when confined to thin aqueous films. Biophys Rep 3:1–7. https://doi.org/10.1007/s41048-016-0026-3
Glaeser RM, PROTEINS, INTERFACES, AND, CRYO-EM GRIDS (2018) Curr Opin Colloid Interface Sci 34, 1–8 https://doi.org/10.1016/j.cocis.2017.12.009
Xia C et al (2011) Conformational changes of NADPH-cytochrome P450 oxidoreductase are essential for catalysis and cofactor binding. J Biol Chem 286:16246–16260. https://doi.org/10.1074/jbc.M111.230532
Suloway C et al (2005) Automated molecular microscopy: the new Leginon system. J Struct Biol 151:41–60. https://doi.org/S1047-8477( 05)00072-9 [pii] 10.1016/j.jsb.2005.03.010
Zheng SQ et al (2017) MotionCor2: anisotropic correction of beam-induced motion for improved cryo-electron microscopy. Nat Methods 14:331–332. https://doi.org/10.1038/nmeth.4193
Scheres SH (2012) RELION: implementation of a Bayesian approach to cryo-EM structure determination. J Struct Biol 180:519–530. https://doi.org/10.1016/j.jsb.2012.09.006
Rohou A, Grigorieff N (2015) CTFFIND4: Fast and accurate defocus estimation from electron micrographs. J Struct Biol 192:216–221
Punjani A, Zhang H, Fleet DJ (2020) Non-uniform refinement: adaptive regularization improves single-particle cryo-EM reconstruction. Nat Methods 17:1214–1221. https://doi.org/10.1038/s41592-020-00990-8
Sanchez-Garcia R et al (2021) DeepEMhancer: a deep learning solution for cryo-EM volume post-processing. Commun Biol 4:874. https://doi.org/10.1038/s42003-021-02399-1
Bepler T et al (2019) Positive-unlabeled convolutional neural networks for particle picking in cryo-electron micrographs. Nat Methods 16:1153–1160. https://doi.org/10.1038/s41592-019-0575-8
Punjani A, Fleet D (2023) 3D Flexible Refinement: Determining Structure and Motion of Flexible Proteins from Cryo-EM. Microsc Microanal 29:1024. https://doi.org/10.1093/micmic/ozad067.518
Pettersen EF et al (2004) UCSF Chimera - A visualization system for exploratory research and analysis. J Comput Chem 25:1605–1612. https://doi.org/10.1002/jcc.20084
Jumper J et al (2021) Highly accurate protein structure prediction with AlphaFold. Nature 596:583–589. https://doi.org/10.1038/s41586-021-03819-2
Roney JP, Ovchinnikov S (2022) State-of-the-Art Estimation of Protein Model Accuracy Using AlphaFold. Phys Rev Lett 129:238101. https://doi.org/10.1103/PhysRevLett.129.238101
Adams PD et al (2010) PHENIX: a comprehensive Python-based system for macromolecular structure solution. Acta Crystallogr D Biol Crystallogr 66:213–221. https://doi.org/S0907444909052925 [pii] 10.1107/S0907444909052925
Emsley P, Lohkamp B, Scott WG, Cowtan K (2010) Features and development of Coot. Acta Crystallogr D Biol Crystallogr 66:486–501. https://doi.org/S0907444910007493 [pii] 10.1107/S0907444910007493

There is NO Competing Interest.

VideoS1.mp4
Supplemental Video 1
supplementalNC.docx

Download PDF

Version 1

posted

You are reading this latest preprint version

Structure of dimerized assimilatory NADPH-dependent sulfite reductase reveals the minimal interface for diflavin reductase binding

Status:

Version 1

Abstract

Figures

Introduction

Results

The SiRHP-SiRFP interaction is highly sensitive to cryogenic TEM (cryo-EM) preparation

SiRFP-SiRHP binding

The SiRHP interaction with SiRFP differs from SIR interaction with ferredoxin

The SiRHP active site loop is locked in its closed conformation

SiRFP is highly mobile

Discussion

Conclusions

Materials and methods

Protein Expression and Purification

Cryo-EM sample preparation

Declarations

Acknowledgements and funding

Author contributions

References

Additional Declarations

Supplementary Files

Status:

Version 1