X-ray and EPR Characterization of the Auxiliary Fe-S Clusters in the Radical SAM Enzyme PqqE

Abstract

The Radical SAM (RS) enzyme PqqE catalyzes the first step in the biosynthesis of the bacterial cofactor pyrroloquinoline quinone, forming a new carbon-carbon bond between two sidechains within the ribosomally synthesized peptide substrate PqqA. In addition to the active site RS 4Fe-4S cluster, PqqE is predicted to have two auxiliary Fe-S clusters, like the other members of the SPASM domain family. Here we identify these sites and examine their structure using a combination of X-ray crystallography and Mössbauer and electron paramagnetic resonance (EPR) spectroscopies. X-ray crystallography allows us to identify the ligands to each of the two auxiliary clusters at the C-terminal region of the protein. The auxiliary cluster nearest the RS site (AuxI) is in the form of a 2Fe-2S cluster ligated by four cysteines, an Fe-S center not seen previously in other SPASM domain proteins; this assignment is further supported by Mössbauer and EPR spectroscopies. The second, more remote cluster (AuxII) is a 4Fe-4S center that is ligated by three cysteine and one-aspartate residue. In addition, we examined the roles these ligands play in catalysis by the RS and AuxII clusters using site-directed mutagenesis coupled with EPR spectroscopy. Lastly, we discuss the possible functional consequences that these unique AuxI and AuxII clusters may play in catalysis for PqqE and how these may extend to additional RS enzymes catalyzing the post-translational modification of ribosomally encoded peptides.

Graphical Abstract

INTRODUCTION

Pyrroloquinoline quinone (PQQ) is a redox active quinocofactor produced by bacteria that is used primarily in the metabolism of simple alcohols and sugars by methanol or glucose dehydrogenases¹. The pqq operon encodes the genes required to synthesize PQQ, which contains—at a minimum—pqqA, pqqB, pqqC, pqqD, and pqqE, in a highly conserved order. PqqA is a short peptide of 20–30 amino acids, depending on the species, and is part of a class of peptides known as RiPPs (Ribosomally-produced and Post-translationally modified Peptides) that are translated by the ribosome but later modified with a series of enzymes during their maturation. The biosynthesis of PQQ begins with cyclization of PqqA in a reaction catalyzed by the radical SAM enzyme PqqE². This reaction takes place with PqqA bound to the peptide binding protein PqqD³, and involves formation of a new carbon bond between glutamate and tyrosine (See Figure 1A). PqqE, a member of the RS superfamily of enzymes, catalyzes the carbon-carbon bond formation. Most members of the RS superfamily of enzymes bind S-adenosyl-L-methionine (SAM) to a [4Fe-4S]^+1/+2 cluster (FeS), which upon reduction from the resting +2 state to the active +1 state, generates methionine and a 5′-deoxyadenosyl (5′dA) radical that initiates the chemistry by abstracting a hydrogen atom from the target substrate (Figure 1A).^4,5. The RS cluster in PqqE is located at the N-terminus of the protein, while the remaining two clusters are predicted to bind to the C-terminal SPASM domain, performing unknown functions that have, nonetheless, been shown in several cases to be required for activity⁴. Characterized members of the SPASM family act predominantly via the modification of ribosomally translated peptides, and have been coined “SPASM” for the founding members: AlbA, PqqE, AnSME, and MftC that participate in subtilosin A and PQQ formation and anaerobic sulfatase and mycofactocin maturation, respectively^4,5,6. A C-terminal FeS binding motif CX₂CX₅CX₃C is nearly invariant for members of the SPASM subfamily, including AnSME from C. perfringens that acts on a sulfatase protein precursor. A direct role for the auxiliary clusters found in the SPASM domain containing proteins remains to be defined; however, one proposal is that they act as a conduit for the controlled transfer of an unpaired electron during the oxidation of the substrate radical intermediate⁷, or in the binding and positioning of substrate close to the RS center; as in MoaA⁸ or AlbA ⁹.

Figure 1.

A, proposed mechanism for carbon-carbon bond formation in the biosynthesis of PQQ, and its relation to the rest of the biosynthetic pathway. The final product of the PqqE reaction is boxed. B, ClustalW alignment of the C-terminal region of SPASM domain family members. The cysteines ligated to the innermost auxiliary cluster (Aux I) in AnSME are identified by the blue bracket, and the outermost (AuxII) are bracketed in green. In PqqE, D319 is placed exactly where, in the other three examples, there is a conserved cysteine; the latter is seen to ligate to Aux II in the AnSME X-ray structure.

Recent work has suggested, based on mutagenesis and spectroscopic data, that PqqE binds both a [2Fe-2S] cluster and a carboxylate-liganded [4Fe-4S] cluster within its SPASM domain¹⁰. In working toward a structural basis that can be linked to such diverse functional roles, we present an X-ray structural model of PqqE from M. extorquens, as well as EPR spectroscopic characterization of the auxiliary FeS centers in PqqE. These data unambiguously identify the ligands to the FeS clusters of PqqE’s SPASM domain and highlight key structural differences between PqqE and other SPASM domain containing enzymes. We also show that these auxiliary clusters are necessary for the formation of the initial carbon-carbon bond in PqqA.

MATERIALS AND METHODS

M. extorquens PqqE was purified and reconstituted as described previously^2,3. Samples prepared in this way were used for all of the described biophysical probes with the exception of the X-ray studies. We have found it extremely difficult to obtain diffractable crystals of PqqE. The fortuitous observation of a crystalline form and structure from unreconstituted PqqE, while absent the RS site, has allowed observation of the two Aux centers (Figure 2 below). Site-directed mutagenesis of MexPqqE was accomplished by the following primers and their reverse complements using the QuickChange mutagenesis kit.: D319A (GGCGCGAGAAGGCTTGGGGCGGGTGCC); C323S/C325S (GGGGCGGGTCCCGCTCCCAGGCGC); -RS (CGCGCCCCGCTGCGCGCCCCATACGCCTCGAACC); -AUXII (GGAGCCCGCCCGCTCCGCCGACCGGCGC). The PqqE +RS mutant was made by Genescript, Inc. Kinetics of 5′dA formation were determined using 50 μM PqqE in the presence of 1 mM sodium dithionite and 500 μM SAM. Aliquots were removed at designated time and quenched with 1% formic acid with subsequent analysis by HPLC as described previously¹¹. Iron and sulfide quantification was obtained as previously reported². Modification of PqqA was detected using LC-MS, according to the published protocol².

Figure 2.

Structure of MePqqE. (A) The Radical SAM (RS) domain (lavender) contains the RS (β/α)₆ partial TIM barrel. The RS cluster and canonical motif are missing from the structure, but the location of the cluster has been approximated using a superposition of the AnSMEcpe structure as a model. The SPASM domain (green) harbors two auxiliary clusters. (B) Stereo view of the MePqqE SPASM domain, which contains a [2Fe-2S] cluster ligated by four cysteines and one [4Fe-4S] cluster ligated by three cysteines and one conserved aspartate. The Fe-edge (9.686 keV) anomalous difference electron density map (gold mesh) is contoured at 4.0 RMSD. (C) Surface representation of sequence conservation as calculated by the ConSurf server [http://conseq.bioinfo.tau.ac.il/] between MePqqE and 1061 members of the PqqE family (SFLD, http://sfld.rbvi.ucsf.edu/django/) using pairwise alignment. Strictly conserved residues are shown in orange, neutral substitutions are shown in tan, and variable regions are shown in white. The view is rotated 90° about the horizontal axis (middle) and 180° about the vertical axis (bottom).

EPR Spectroscopy

The X-band (9.43 GHz) continuous-wave (CW) EPR spectra were recorded on a Bruker (Billerica, MA) Biospin EleXsys E500 spectrometer equipped with a super-high Q resonator (ER4122SHQE). Cryogenic temperatures were achieved and controlled using an ESR900 liquid helium cryostat in conjunction with a temperature controller (Oxford Instruments ITC503) and gas flow controller. CW EPR data were collected under slow-passage, non-saturating conditions at all temperatures. The spectrometer settings were as follows: modulation amplitude = 0.2 mT, and modulation frequency = 100 kHz; conversion time = 88 ms. Spectra were simulated using the EasySpin Matlab package.

Mössbauer Spectroscopy

Zero-field ⁵⁷Fe Mössbauer spectra were recorded in a constant acceleration spectrometer (See Co., Edina, MN) at 4.2 K using a Janis Research Co. cryostat (Willmington, MA). Spectra were analyzed with the WMOSS software package (See Co., Edina, MN). Isomer shifts are reported relative to α-iron (27 μm foil) at room temperature. PqqE samples were prepared by freezing solutions in a Teflon sample holder (thickness 0.2 inch) under an Argon/Nitrogen atmosphere.

Crystallography

Diffraction quality crystals of unreconstituted MePqqE were obtained by sitting-drop vapor diffusion at 20 °C in an anaerobic chamber maintained at < 0.1 ppm oxygen (MBraun). Crystals appeared in 70 days, after combining 0.5 μL protein solution (20 mg mL⁻¹ of MePqqE in 50 mM Tris, 100 mM NaCl, 1 mM TCEP pH 7.4) with 0.5 μL reservoir solution (0.2 M magnesium formate, 20% polyethylene glycol 3,350). Crystals were mounted on nylon loops and flash-cooled in liquid nitrogen inside the anaerobic chamber, without added cryoprotectant, and stored in liquid nitrogen prior to data collection. X-ray diffraction data were collected at 100 K on the beamline X29A (National Synchrotron Light Source, Brookhaven National Laboratory, Upton, NY) at a wavelength of 1.280 Å (9.686 keV), and processed and scaled with HKL3000¹². Diffraction from these crystals is consistent with the trigonal space group P3₁21, with unit cell parameters a=b=97.57 Å, c=86.42 Å, with one molecule per asymmetric unit, and solvent content of 57%. Exploiting the intrinsic iron-sulfur cluster, phases were determined by SAD with autoSHARP¹³, and an initial poly-alanine model was built with ARP/wARP¹⁴. Subsequent rounds of automated model building were performed by AutoBuild in PHENIX¹⁵, interspersed with manual model building and refinement using Coot¹⁶, phenix.refine¹⁷, and Refmac5¹⁸. All figures were generated with PyMOL¹⁹. The final model consists of 312 residues, with gaps occurring from residues 1-13, 29-43, 127-140, 202-204, 349-384, and includes 6 iron ions, 6 sulfide ions, and 13 water molecules. Data collection and refinement statistics are shown in Table 1.

Table 1.

Data collection and refinement statistics

PDB ID 6C8V	MePqqE
Data collection statistics
Space group	P3121
Unit-cell parameters (Å)	a = b = 97.57, c = 86.42
Resolution (Å)	19.75–3.20 (3.46–3.20)
Rmerge	0.227 (0.834)
No. of unique reflections	79485 (16073)
Multiplicity	9.8 (9.7)
Completeness (%)	99.6 (100)
Mean I/σ(I)	9.4 (2.8)
CC1/2	0.990 (0.741)
Wilson B factor (Å2)	64.03
Refinement statistics
Resolution (Å)	19.75–3.20 (3.315–3.20)
Rwork (%)	0.1967 (0.2687)
Rfree (%) (5% of data)	0.2514 (0.3115)
No. of residues	312
No. of waters	13
Total No. of atoms	2374
B factor (Å2)
Protein	63.60
Waters	44.68
Ligand	50.49
R.m.s.d., bond lengths (Å)	0.0039
R.m.s.d., angles (°)	0.894
Ramachandran outliers (%)	0.0
Side-chain outliers (%)	1.2

Highest-resolution shell data are shown in parentheses

RESULTS

Sequence Comparison of MePqqE to members of the SPASM domain family reveals unique features

Figure 1B shows the multiple sequence alignment of four well-characterized proteins containing SPASM domains. PqqE, AnSME and MftC are founding members of the SPASM domain family, while StrB was recently shown to carry out a reaction similar to PqqE, the formation of a carbon-carbon bond within a small peptide⁷. Among the defining members of the RS-SPASM family, only a single X-ray crystal structure of AnSME, (PDB 4K37)²⁰ had been available over many years, serving as the structural surrogate for the entire family. Very recently two additional X-ray structures became available: one for the peptide modifying enzyme CteB from Clostridium thermocellum, involved in carbon-sulfur cross-linking²¹, and another for the enzyme SuiB, a homolog of StrB that forms a carbon-carbon bond between a lysine and tryptophan side chains in the peptide substrate SuiA ²². Significantly, anSME, SuiB, and CteB show highly conserved features at all three iron sulfur sites: cysteine coordination of [4Fe-4S] clusters in the AuxI and AuxII sites, with the major difference being that CteB has an open coordination sphere within its AuxI site²¹. With the exception of PqqE, all enzymes characterized thus far indicate 5 conserved Cys residues at the C-terminus, four of which are assigned to the AuxII site based on the AnSME structure. Using the side chain numbering for PqqE in this region (310, 313, 319, and 341), PqqE indicates instead an aspartate at position 319, implicating CX₂CX₅DX₃C as the motif for AuxII. Consistent with this feature, PqqE contains only 3 conserved cysteines at its C-terminus. As shown in Figure 1B, the spacing of cysteine ligands for AuxI is much more variable than for AuxII, making it difficult to predict ligands to AuxI based on sequence data alone.

Crystal structure of MePqqE shows the auxiliary clusters and their ligation environment

The final model of MePqqE from Methylobacterium extorquens was refined to 3.20 Å (R=19.67%, R_free=25.14%) and spanned residues 13 to 348 out of 384 residues. Domain analysis of the structure revealed a partial [β/α₆] TIM barrel (amino acids 22-179), followed by the C-terminal SPASM domain (amino acids 268-348, TIGR04085)⁵. Residues 29 to 43 contain the canonical radical SAM motif (CX₃CX₂C), however this region was not observed due either to a high degree of structural disorder and/or an incompletely formed RS site.

MePqqE SPASM domain

The two auxiliary clusters of PqqE are easily visualized from the structure (Figure 2A, 2B) as a 2Fe-2S in the AuxI site and a [4Fe-4S] in the II site. The AuxI 2Fe-2S cluster is ligated by residues C248, C268, C323, and C325. The AuxII 4Fe-4S cluster is located ~12 Å from AuxI and is ligated by residues C310, C313, D319, and C341. Notably, the participation of D319 in AuxII ligation—implied from the sequence alignment—(Figure 1) is corroborated by the crystal structure (Figure 2B). This ligand pattern for AuxII in the SPASM domain of MePqqE appears to be highly conserved in all annotated sequences of PqqE: a pair-wise alignment²³ using 1061 sequences²⁴ showed D319 to be invariant.

Comparison of MePqqE to AnSME

Though the cofactor within the RS-domain of PqqE cannot be visualized, the RS domain of AnSME (amino acids 10-181) aligns well with MePqqE (amino acids 22-179), with an RMSD of 2.69 Å over 217 aligned C-alpha carbons (sequence identity 29%). The C-terminal SPASM domain of MePqqE (amino acids 268-348) diverges more significantly from AnSME (RMSD 3.03 Å, Figure 3) with the aforementioned CX₂CX₅DX₃C motif for PqqE contrasting with the predicted motif of CX₂CX₅CX₃C motif. Furthermore, the last cysteine in this motif, C325, ligates Aux I in MePqqE, whereas the equivalent cysteine (C333) in anSME is unbound. PqqE has a unique ligand set for AuxI, with both C323 and C325 liganded to the same iron of a [2Fe-2S] cluster (Figure 3A, 3B). This is an unusual ligation motif (CXC) found predominantly in certain [2Fe-2S] clusters^25,26, and only rarely in [4Fe-4S] clusters²⁷, making it less likely that we are observing a degraded [4Fe-4S] cluster instead of a native [2Fe-2S] cluster. For comparative purposes, a structural overlay of MePqqE and MoaA is given in Fig. S1.

Figure 3.

Structural similarity of MePqqE and AnSMEcpe. Secondary structure and topology of MePqqE (A) and AnSMEcpe (B) show the ligation of auxiliary clusters. AUX I and AUX II are labeled as I and II. Ligating residues are represented by circles and corresponding residue numbers are shown to the right of each topological rendering. Asp 319 is shown by number 5, yellow highlighting. Panel (C) Shows the structural alignment of the SPASM domains of MePqqE in green (Fe is orange and S is yellow) and AnSMEcpe in pink (Fe and S are grey; RMSD 3.03 Å over 78 C-alphas, sequence identity 19.23%).

Impact of conserved residues in PqqE on Fe-S cluster formation and activity of PqqE

In order to determine the effects of removing an Fe-S cluster, we pursued mutagenesis to alter ligating residues within the three unique Fe-S clusters of PqqE. While our goal was to eliminate the RS center, AuxI and AuxII in a step-wise fashion, we were never able to isolate a stable variant that was disrupted solely at AuxI. On the other hand, proteins for which either the RS or AuxII was disrupted were successfully prepared. The iron binding and activity of the resulting variants are summarized in Table 2. Wild type PqqE purifies with ca. 7 irons and sulfides when expressed recombinantly in E. coli. Upon chemical reconstitution up to 13 irons and 12 sulfurs per peptide have been found previously². The variant in which the RS site was disrupted (C32A, C35A termed –RS cluster) causes a decrease in the iron loading of PqqE, and, as expected, abolishes both the reductive cleavage of SAM and cross-link formation on PqqA (Table 2). The crystal structure shows that C310 and C313 bind to auxiliary AuxII, similar to what is found in anSME, CteB, and SuiB.^12–14 The C310A and C313A variant (termed –AuxII protein) likewise causes a decrease in iron content. In this instance, the protein is able to reductively cleave SAM, but is incapable of forming the cross-linked PqqA product. The variant MePqqE that contained only cysteines 28, 32, and 35 (termed +RS protein) was highly unstable and capable of only producing low amounts of 5′dA for each molecule of protein without detectable cross-linking of PqqA. The MePqqE D319A variant led to a decrease in iron loading (~9 Fe per protein) and, importantly, was also unable to cross-link PqqA. This suggests that correct ligation of AuxII is essential for function of PqqE. Several reviewers noted the lack of exact correspondence between the titers for Fe in reconstituted proteins and the Fe-S structures inferred from the combined application of X-ray, EPR and Mossbauer methodologies. It is well known that RS protein preparations can contain significant adventitious Fe that is bound to either contaminating proteins (contributing broad backgrounds to Mossbauer spectra) or possibly to His-tags incorporated to facilitate protein purification (as the case for PqqE). For each of the protein variants described in Table 2, the presented Fe-protein ratios are included as representative of the preparations and cannot be used to draw precise conclusions regarding structures at the individual Fe-S sites²⁸.

Table 2.

Mutants of M. extorquens PqqE studied. Each was reconstituted, assayed for SAM cleavage, Fe content and ability to form cross-linked PqqA. Assays were carried out using established procedures (Materials and Methods).

Name	Mutations of Cys within RS and 2 Aux Sites	SAM Cleavage	Fe/Protein	Modification of PqqAa
WT	None	0.031/min	13.0 +/− 0.1	Yes
−RS	C32A,C35A	None Observed	8.7 +/− 0.8	No
−AuxII	C310A,C313A	0.0045/min	7.3 +/− 0.6	No
+RS	All to Ser except C28,32,35	0.00043/min	4.6 +/− 0.7	No
D319A	D319A	0.0037/min	8.9 +/− 0.9	No

^aWe note the low yields for cross-linked PqqA, optimally 10% for WT enzyme: A failure to observe cross-linked product for the variants described is, thus, “within the limits of detection.”

Distinct Spectroscopic Properties at each of the three Fe-S clusters of PqqE

Low temperature EPR

We first compared the 10 K EPR behavior of the dithionite-reduced forms of WT, -RS and –AuxII PqqE. All samples show a continuous wave (CW) EPR spectrum with multiple components centered near g = 2 (Figure 4). These are distinct from samples not treated with dithionite that typically show signals consistent with a variable amount of [3Fe-4S]⁺ clusters. By raising the temperature to 60 K, only the signal corresponding to the [2Fe-2S]⁺ cluster persists, owing to its unique relaxation properties (see below)^29,30. At the lower temperature (10 K) the g₁-values for two [4Fe-4S]⁺ signals are also evident (g₁ = 2.055 and g₁ = 2.035). Examination of a 10 K EPR difference spectrum of the reduced WT PqqE and the –RS PqqE samples indicates loss of the small g = 2.035 feature, which we have assigned to RS cluster cluster (typical RS cluster g =2.037, 1.92, 1.89). The signal at g₁ = 2.055 is attributed to an alternate [4Fe-4S]⁺ center, possibly AuxII. Consistent with the X-ray structure that indicates D319 as the fourth ligand to AuxII, mutation of this residue to Ala leads to a spectrum very similar to the –AuxII variant indicating that D319 is required for proper formation of the 4Fe-4S cluster in AuxII (Figure 4 bottom). The inset in Figure 4, which shows the region surrounding g₁ = 2.055, may indicate the persistence of a small signal in both the –AuxII variant and D319A; assignment of this feature will require further study.

Figure 4.

Continuous-wave EPR of dithionite reduced reconstituted wild-type PqqE (100 μM), PqqE -RS (200 μM), PqqE –AuxII (200 μM), and PqqE D319A (95 μM) at 10 K. We see three major EPR signals contributing to the spectrum of reduced WT PqqE: (1) a weak signal with g₁ = 2.037 which corresponds to the radical SAM cluster (see simulation in Figure S2), (2) a signal with g₁ = 2.055, and (3) a signal with g₁ = 2.005, g₂ = 1.959, g3 = 1.905, which corresponds to the auxiliary clusters. All spectra have been scaled according to the concentration of PqqE, with the PqqE –AuxII spectrum further divided by a factor of 3 for ease of comparison. Note that disruption of the AuxII cluster induces a dramatic increase in the amount of [2Fe–2S]⁺ signal relative to the [4Fe–4S]⁺ cluster signals in most samples. Inset focuses on the g₁-region for the [4Fe–4S]⁺ cluster signals.

Temperature-dependent EPR

The dithionite reduced WT PqqE was further analyzed using CW EPR spectroscopy at varying temperatures. The resulting spectra show a distinctive transition between 10 K and 60 K (Figure 5A) at which only a single paramagnetic species is observed that is characterized by the g-values = [2.005, 1.959, 1.905]. At the elevated temperatures, EPR signals for most [4Fe-4S]⁺ clusters relax too fast to be observed; by contrast, [2Fe-2S]⁺ clusters retain strong EPR signals at 60 K. The average g-value of the high temperature signal of WT PqqE (g_av = 1.971) is at the high end of the range of values for ferredoxins coordinated by four cysteine sidechains (typically g_av =1.96.) (Table S1).

Figure 5.

Temperature dependence of EPR spectra of DTH-reduced wild-type PqqE (A) and –AuxII PqqE (B). As the temperature is elevated above 10 K, the [4Fe-4S]⁺ signals that are dominated by AuxII (Figure 4) completely disappear, due to fast relaxation that is a signature of [4Fe-4S]⁺ clusters. At 60 K, all that remains is a signal with g-values = [2.0049, 1.958, 1.906]. The average g-value = 1.9563 for the remaining signal is consistent with that measured for several all cysteinyl-coordinated, ferredoxin-type [2Fe-2S]⁺ clusters (Table S1).

Of considerable interest, this high temperature feature dominates the spectra for –AuxII PqqE at all temperatures (Figure 5B). This finding argues against the generation of new degradation products in PqqE –AuxII and supports the assignment of the high temperature spectra of WT enzyme to a 2Fe-2S center within the AuxI cluster, because no new spectral features appear in the PqqE –AuxII mutant that do not already exist in the wild-type. For the reduced WT PqqE sample at 10 K, the [2Fe-2S]⁺ cluster signal can be estimated to account for ca.15% of the total spins detected. We note that this result is variable, with occasional preparations of enzyme failing to show strong signals at the elevated temperatures.

Mössbauer spectroscopy is consistent with the presence of a single 2Fe-2S cluster and two 4Fe-4S clusters

The PqqE variant C32A/C35A (PqqE -RS) was grown in the presence of ⁵⁷Fe and chemically reconstituted. Figure 6, bottom, shows the zero-field Mössbauer spectrum of PqqE -RS, as compared with the spectrum of WT from ref. (2) (top). Wild-type PqqE shows two doublets: a [4Fe-4S]²⁺ signal (δ = 0.478, ΔEq = 1.172), comprising 80% of the total iron, and a [2Fe-2S]²⁺ signal (δ = 0.324, ΔEq = 0.558), comprising 20% of the total iron. In the -RS mutant, the contribution of the 2Fe-2S signal (δ = 0.327, ΔEq = 0.521) increases significantly, accounting for ca 55% of all iron in the sample, whereas the [4Fe-4S] signal (δ = 0.523, ΔEq = 1.264) accounts for the remaining 45%. While we see less 4Fe-4S signal than expected, these data provide strong support for the presence of a [2Fe-2S] cluster within the SPASM domain of PqqE.

Figure 6.

Zero-field, 4 K Mossbauer spectra of reconstituted PqqE WT and PqqE -RS knockout show loss of 4Fe-4S clusters and presence of 2Fe-2S cluster. Data (circles) were fit to two doublets, representing [4Fe-4S] (solid line) and [2Fe-2S] (dotted line). WT spectrum from Barr et al., 2016.

DISCUSSION

From mechanistic considerations alone, it is not clear why an enzyme such as PqqE would require auxiliary clusters in order to function. Many RS family enzymes catalyze complex reactions that require both oxidation and reduction of radical intermediates using only a single RS 4Fe-4S cluster, such as NirJ,³¹ DesII³², RlmN³³, Cfr³⁴, and QueE³⁵: i.e. all RS enzymes that utilized SAM as a cosubstrate rather than a cofactor. For others, the auxiliary clusters function in substrate ligation (MoaA⁸) or as a source of sulfur or iron (BioB³⁶ and HydG¹¹). Prior to the determination of a structure for the RS SPASM proteins, the most likely assumption was that an auxiliary cluster housed in the SPASM domain would contain at least one open coordination site that enabled the binding of their peptide or protein substrate³⁷. This can no longer be assumed, and focus has shifted to their potential role in electron transfer. What is clear is that in cases where the auxiliary clusters of a SPASM enzyme have either been removed by mutagenesis or incorrectly formed, activity is abolished, as shown in Table 2 and Refs. 11, 18, 19. This remains the most puzzling aspect of the RS SPASM enzymes, and one that experimental results have just begun to address.

Whatever the explanation, an acceptable hypothesis for the unusual Fe-S clusters of PqqE must center on the peculiarities of PqqE’s mechanism, or the biological context in which it operates. The RS enzyme SuiB carries out a reaction very similar to PqqE^22,38. In this reaction, a carbon-carbon bond is formed between the β-carbon of a lysine residue and the aromatic carbon-7 of tryptophan in the peptide substrate SuiA, where the reactive side chains are separated by three residues. In PqqE’s reaction, cross-linking occurs between the β-carbon of a glutamic acid and the aromatic carbon-3 of tyrosine, also separated by three residues. SuiB additionally has, like many peptide modifying enzymes, a domain with homology to PqqD, the binding partner of PqqA that also associates with PqqE. While these features alone may have predicted auxillary FeS clusters in PqqE that are highly similar to SuiB, the crystal structure of SuiB shows a SPASM domain with topology much closer to anSME or CteB and neither a 2Fe-2S cluster in AuxI nor an aspartate ligand to AuxII, highlighting how far PqqE has diverged from the other characterized members of the SPASM family.

Asp-ligation and redox tuning

The two notable features of PqqE are the presence of the 2Fe-2S cluster in the Aux I site of the SPASM domain and the participation of aspartate as a ligand for the 4Fe-4S cluster in Aux II. The latter observation may be of functional importance, as the reduction potentials of iron-sulfur clusters are significantly influenced by the nature of the ligating functionalities^39–41. Similar to Aux II from MePqqE, Pyrococcus furiosus ferredoxin (PfFd) contains a single 4Fe-4S cluster ligated by 3 Cys residues and 1 Asp residue^39,40. The effects of ligand environment of the PfFd were extensively studied by Brereton and co-workers⁴⁰, who reported that the reduction potential of the wild-type protein was −368 mV, but drops to −427 mV in the Asp to Cys variant. Substituting the Asp ligand to Ser further reduces the redox potential to −505 mV. Because PqqE catalyzes the two-electron oxidation of its substrate, a more positive reduction potential for the unique Asp-ligation in the Aux II cluster may facilitate oxidation of the substrate-derived radical intermediate by providing a thermodynamically favorable electron sink (see Fig. 1A).

Presence of [2Fe-2S] cluster in AuxI site

While the presence of a [2Fe-2S] center in PqqE has been previously proposed,^10,42 there is always the potential of degradation within an originally formed 4Fe-4S cluster site. Therefore, characterization of 2Fe-2S clusters in the presence of 4Fe-4S clusters is often difficult. Our inability to produce a soluble AuxI knock-out to directly interrogate the remaining clusters complicated the analysis; the latter would have been expected to show the presence of only 4Fe-4S clusters in the RS and AuxII sites. In the only example of an AuxI variant that could be characterized thus far for PqqE¹¹, the C323S variant has been reported to be capable of binding both 2Fe-2S and 4Fe-4S clusters. Despite some remaining experimental ambiguity, the X-ray, EPR and Mössbauer characterizations presented herein, together with the recent biochemical and Mössbauer studies of PqqE variants¹¹, argue for the presence of a [2Fe-2S] center within AuxI.

Mechanistic Implications

The presence of a [2Fe-2S] cluster in PqqE is unusual as the other known SPASM domain enzymes studied are shown to exclusively bind [4Fe-4S]. The substitution of a four cysteine-liganded [2Fe-2S] cluster in this position in PqqE, where other SPASM enzymes show a [4Fe-4S], further suggests to us that this cluster may be used for electron transfer; electron transfer is one of the demonstrated functions of Fe-S clusters for which [4Fe-4S] and [2Fe-2S] are commonly interchangeable, e.g. in [4Fe-4S] and [2Fe-2S] ferredoxins⁴³. [2Fe-2S] clusters generally have higher redox potentials than [4Fe-4S], commonly in the −400 to −300 mV range. While redox potentials have not yet been measured for any of the clusters of PqqE, this will be an important next step in relating function to structure.

Several reasons for the unusual FeS clusters of PqqE can be devised, and a second plausible explanation centers around the fact that PqqE must operate in the presence of molecular oxygen, which is required for other steps in the biosynthetic pathway ⁴⁴. As 2Fe-2S clusters have been observed to be less sensitive to oxygen than 4Fe-4S clusters, the 2Fe-2S cluster might represent an evolutionary strategy for a SPASM domain enzyme operating in an oxygen-rich environment. For PqqE this stability has been shown experimentally⁴² but only for resting, oxidized protein. Another possible effect of having two auxiliary clusters with unusually high redox potentials might be to “dissipate” the energy of the electron being removed from the substrate in the final oxidation step. In the presence of oxygen, an electron exiting the active site by means of a 4Fe-4S cluster with a potential around −500 mV is far more likely to produce a reactive oxygen species than one where the potential has been altered to around −350 mV.

For SPASM-domain containing enzymes, including PqqE, a great deal of work has been done to determine the reaction catalyzed and the number and character of the auxiliary clusters. What is missing is perhaps the most important yet difficult aspect of these enzymes, a mechanistic description of the reaction itself. From the current work, and from previous studies, it is clear that the auxiliary clusters of these enzymes are essential to whatever mechanism is used, and that a complete description of the action of PqqE and its relatives will need to account for the correct structural relationship within a complex containing the peptide substrate, its chaperone (as a separate protein or N-terminally fused domain), SAM, the RS enzyme and a reductase to initiate the reaction. The current work is a necessary first step in that direction, giving us for the first time a three dimensional picture of the SPASM domain of PqqE and the nature of the FeS clusters therein.

ABBREVIATIONS

PQQ

pyrolloquinoline quinone

RS

Radical SAM

SAM

S-adenosyl methionine

AnSME

anaerobic sulfatase maturing enzyme