Intermolecular electron transfer in radical SAM enzymes as a new paradigm for reductive activation

Abstract

Radical S-adenosyl-L-methionine (rSAM) enzymes bind one or more Fe-S clusters and catalyze transformations that produce complex and structurally diverse natural products. One of the clusters, a 4Fe-4S cluster, binds and reductively cleaves SAM to generate the 5'-deoxyadenosyl radical, which initiates the catalytic cycle by H-atom transfer from the substrate. The role(s) of the additional auxiliary Fe-S clusters (ACs) remains largely enigmatic. The rSAM enzyme PapB catalyzes the formation of thioether cross-links between the β-carbon of an Asp and a Cys thiolate found in the PapA peptide. One of the two ACs in the protein binds to the substrate thiol where, upon formation of a thioether bond, one reducing equivalent is returned to the protein. However, for the next catalytic cycle to occur, the protein must undergo an electronic state isomerization, returning the electron to the SAM-binding cluster. Using a series of iron-sulfur cluster deletion mutants, our data support a model whereby the isomerization is an obligatorily intermolecular electron transfer event that can be mediated by redox active proteins or small molecules, likely via the second AC in PapB. Surprisingly, a mixture of FMN and NADPH is sufficient to support both the reductive and the isomerization steps. These findings lead to a new paradigm involving intermolecular electron transfer steps in the activation of rSAM enzymes that require multiple iron-sulfur clusters for turnover. The implications of these results for the biological activation of rSAM enzymes are discussed.

Keywords: S-adenosyl-L-methionine (SAM); electron transfer; enzymatic activation; enzyme mechanism; iron–sulfur protein; radical SAM.

Figure 1

Robetta-predicted (41) structural model of PapB and the thioether formation mechanism.A, the conserved Cys residues in the sequence alignment (Fig. S2) are predicted to occur in three distinct regions. The Cys residues that ligate the three 4Fe-4S clusters in PapB are shown in red (RS), yellow (AC1), and purple (AC2). B, upon reduction of the RS cluster, the homolytic cleavage of SAM produces a dAdo⋅. C, the dAdo⋅ abstracts an H-atom from the position α to the carboxylate in the peptide, generating a substrate radical. Thioether bond formation is coupled to disengagement from AC1 (24).

Figure 2

Dependence of the activity of PapB on components of the reducing system. PapB is fully active in the buffer solutions that contain FldA, FPR, SAM, and NADPH (see Fig. S3), with quantitative cross-linking of msPapA observed in 15 min. In a charge state of +3, a cross-link corresponds to a shift of −0.6719, which is a Δ2 Da (m/z = 843.4478, M = 2530.3434). A, control reactions with no SAM or (B) NADPH show only trace quantities of cross-linked peptide. C, by contrast, ∼40% completion is observed in the absence of FldA. D, nearly quantitative conversion to the cross-linked species is observed in the absence of FPR. The red dotted line signifies the unmodified monoisotopic mass of the msPapA peptide (m/z = 844.1197, M = 2532.3591).

Figure 3

PapB is active in the presence of FMN and NADPH.A, in the absence of PapB, no cross-linking is observed (top row). Full cross-linking of the peptide substrate is seen after 3 h when FMN and NADPH are present (bottom row). B, when NADPH is excluded, PapB still cross-links msPapA, albeit inefficiently. C, by contrast to (A and B) where FMN is present, removing FMN from the assay mixture leads to the formation of <1% cross-linked msPapA. The red dashed lines denote the calculated unmodified monoisotopic mass of msPapA (m/z = 844.1197).

Figure 4

FldA and FMN are reduced in the presence of NADPH. A solution containing (A) 55 μM FldA or (B) FMN was incubated with 2 mM NADPH and spectra were obtained over a period of 180 min. While the flavin is reduced in both cases, the spectral changes in (A) suggest that FldA is reduced to the semiquinone, likely from equilibration of oxidized and two-electron reduced. FMN is reduced to the hydroquinone form. Each experiment was carried out in triplicate, but (A and B) show a representative example.

Figure 5

Prereduced PapB can cross-link Y17W msPapA and is more efficient in the presence of either oxidized FldA or oxidized FMN. The assays contained 450 μM Y17W msPapA, 1.9 μM prereduced PapB, 2 mM DTT, and 2.4 mM SAM. The assays with electron mediators contained 25 μM of either oxidized FMN (green square) or FldA (red triangle). Each experiment was carried out in triplicate and all replicates are shown. For mass spectra of sample withdrawn at 2, 4, 6, and 18 h, see Fig. S5 for prereduced PapB, Fig. S6 for prereduced PapB + FldA(ox), and Fig. S7 for prereduced PapB + FMN(ox).

Figure 6

Activation of QueE with prereduced PapB.A, Photodiode array (PDA)-detected chromatograms of reaction mixtures containing QueE, CPH₄, and msPapA. Control reactions lacking any source of reducing equivalents show CPH₄ and msPapA, eluting at ∼7.9 and ∼24 min, respectively. A sample of the flow-through (50 μl) from the PapB concentration step was added to control for the presence of any excess reductant. The identities of (B) CPH₄ or CDG and (C) msPapA are confirmed by mass spectral analysis of the corresponding species in the mass spectrometric analyzer. CPH₄ exhibits an m/z of 212.077 (expected monoisotopic mass = 212.0778, ppm error = −0.47). The absence of any CDG indicates no reducing equivalents sufficient for activation of QueE are present in the assay. As a positive control, the same experiment was carried out in the presence of NaDT (middle row). Unlike in the negative control shown in (A), the presence of NaDT leads to complete conversion of CPH₄ to CDG, which elutes at 8.9 min and exhibits an m/z of 195.0510 (expected monoisotopic mass = 195.0513, ppm error = −1.54). As expected, since no PapB is present, no cross-links are observed in msPapA. In the presence of prereduced PapB, complete conversion of CPH₄ to CDG is observed. CDG has the same retention time and m/z as the product made in the NaDT control (middle column, middle row versus middle column, bottom row). In addition, cross-linking of msPapA is observed, as evidenced by a 2 Da shift during the 18-h assay. The arrow in (C) shows the position of the monoisotopic peak of the cross-linked msPapB. Experimental conditions shown here are a representative spectrum of the t = 18 h timepoint. All of the assays contained 200 μM Y17W msPapA, 12 μM oxidized QueE, 2.43 mM CPH₄, 2 mM MgSO₄, 2 mM DTT, and 2.4 mM SAM. Additionally, the (+) Reductant assay (second row) contained 2 mM NaDT and the prereduced PapB assay (third row) contained 1.9 μM prereduced PapB and no NaDT.

Figure 7

Intermolecular electron transfer between prereduced PapB Fe-S cluster variants and oxidized QueE. An aliquot of each assay containing oxidized QueE and (A) flow-through from the PapB variant, (B) NaDT, or (C) prereduced PapB was analyzed by LC-MS with in-line UV-visible (at 299 nm) and mass spectrometric detection. In all the cases, a small amount of methylthioadenosine is observed resulting from nonenzymatic hydrolysis of SAM. In the presence of NaDT, CDG is observed, consistent with the ability of NaDT to fully activate QueE. In the presence of prereduced PapB, CDG is formed in all samples. However, the amount of CDG produced is reduced in the ΔRS and ΔAC2 samples relative to the ΔAC1 samples. The comparison of the mass spectra between the observed and expected monoisotopic masses and ppm error for each peak in a prereduced WT PapB + oxidized QueE chromatogram are shown in Fig.S13. See Table S2 for a comparison of the relative CDG formed in each KO variant relative to the WT.

Figure 8

Redox states and cycling in PapB. When purified, PapB (A) is fully oxidized. Upon reductive activation of the RS cluster to (B), the RS cluster catalyzes the cleavage of SAM to dAdo⋅, which initiates the chemistry by H-atom transfer from the substrate. A reducing equivalent is then transferred to AC1 concomitant with thioether bond formation to generate the postcatalysis state (C). Isomerization of the enzyme to the catalytically active RS reduced state (B) requires an intramolecular electron transfer from AC1 to AC2 forming (D), followed by an intermolecular transfer involving a redox active species to form the reduced RS state (B). The iso-mechanism involves conversion of (C) to (B) via (D). The enzyme only requires a single priming reductive activation step, with all subsequent cycles relying on the electronic isomerization to reactivate the protein.