Functional AdoMet Isosteres Resistant to Classical AdoMet Degradation Pathways

Abstract

S-adenosyl-l-methionine (AdoMet) is an essential enzyme cosubstrate in fundamental biology with an expanding range of biocatalytic and therapeutic applications. We report the design, synthesis, and evaluation of stable, functional AdoMet isosteres that are resistant to the primary contributors to AdoMet degradation (depurination, intramolecular cyclization, and sulfonium epimerization). Corresponding biochemical and structural studies demonstrate the AdoMet surrogates to serve as competent enzyme cosubstrates and to bind a prototypical class I model methyltransferase (DnrK) in a manner nearly identical to AdoMet. Given this conservation in function and molecular recognition, the isosteres presented are anticipated to serve as useful surrogates in other AdoMet-dependent processes and may also be resistant to, and/or potentially even inhibit, other therapeutically relevant AdoMet-dependent metabolic transformations (such as the validated drug target AdoMet decarboxylase). This work also highlights the ability of the prototypical class I model methyltransferase DnrK to accept non-native surrogate acceptors as an enabling feature of a new high-throughput methyltransferase assay.

Graphical abstract

Methyltransferase (MT)-catalyzed S-adenosyl-l-methionine (AdoMet)-dependent methylation is essential to all walks of life where it contributes to modulating the function of a vast array of biomolecules ranging from small metabolites^1–5 to macromolecules.^5–15 Consistent with this, alterations in methylation-dependent processes are associated with cancer,¹⁶ neurodegenerative/neuropsychiatric disorders,^17–19 inflammation,^20,21 metabolic disorders,²² fundamental development/regenerative medicine,^23,24 susceptibility to disease/adverse drug response,^25,26 and drug resistance.^{13–15,27,28} While there have been great advances in methylation-dependent bioinformatics and disease-associated biomarkers,^29–31 the study of intracellular MT spatial/temporal resolution, specificity, and/or function remains a challenge.^{9,14,15,25–28,32–36} Toward this end, the pioneering demonstration of AdoMet analogs, bearing alternative alkyl donor substituents, as cosubstrates for DNA³⁷ or natural product (NP)³⁸ MTs has inspired new tools and strategies to study NP,^39–42 protein,^43–51 and nucleic acid^6,52–57 methylation where the recent development of enzyme-based strategies for the synthesis of differentially alkylated AdoMet analogs has simplified access to these unique cosubstrates.^{42,49,50,57–59} However, the stability of AdoMet or its corresponding differentially alkylated analogs under physiological conditions limits their utility as reagents or therapeutics by virtue of two fundamental degradative processes: intramolecular cyclization to homoserine lactone and 5′-deoxy-5′-(alkylthio)adenosine (Figure 1, pathway a) and depurination (Figure 1, pathway b).^50,60–62 To address this deficiency, herein we report the chemoenzymatic synthesis of AdoMet isosteres designed to circumvent the major AdoMet degradative pathways and demonstrate, via both biochemical and structural studies, these analogs to serve as productive MT cosubstrates for the prototypical class I MT DnrK. This work highlights a unique conceptual strategy to modulate the fundamental properties of AdoMet and corresponding analogs as reagents in epigenetic, proteomic, biocatalytic, and/or therapeutic applications¹³ and also demonstrates a high-throughput colorimetric assay for DnrK based upon the non-native surrogate substrate 2-chloro-4-nitrophenol (ClNP).

Figure 1.

Utilization and degradation of AdoMet. General scheme highlighting methyltransferase (MT)-catalyzed use of AdoMet in methylation reactions (upper) and major chemical degradation pathways of AdoMet (lower). AdoMet, S-adenosyl-l-methionine (also referred to as SAM); AdoHcy, S-adenosyl-l-homocysteine (also referred to as SAH).

RESULTS AND DISCUSSION

Synthesis of AdoMet Carboxylate Isosteres

Adenine isosteres, such as 7-deazaadenine (^7dzA), have been successfully employed to reduce the propensity for nucleotide depurination (Figure 1, pathway b).^63–65 Thus, we focused our attention on addressing the AdoMet degradative intramolecular cyclization to homoserine lactone and 5′-deoxy-5′-(alkylthio)adenosine (Figure 1, pathway a) by replacing the AdoMet carboxylate with a sterically constrained and/or less nucleophilic isosteric functional group. As potential carboxylate isosteres,^66,67 tetrazolates display a similar pKa,⁶⁸ are less nucleophilic,⁶⁹ and offer improved membrane permeability.⁷⁰ In addition, MM2 calculations predict the formation of a sterically strained tetrazole-containing bicyclic ring structure via intramolecular cyclization (Figure 1, pathway a) as energetically unfavorable (see Supporting Information). To test this concept, the tetrazole-l-methionine isostere (l^tMet) was synthesized from commercially available N-Fmoc-l-methionine in four simple steps (Figure 2, 41% overall yield) via a slight modification of previously reported methods^71–78 and subsequently evaluated as a substrate for the permissive methionine adenosyltransferase hMAT2A.⁴² Importantly, this study revealed the k_cat of hMAT2A-catalyzed reactions using native substrate l-methionine (lMet) or isosteric l^tMet as nearly identical (lMet, 8.0 ± 0.1 min⁻¹; l^tMet, 6.1 ± 0.6 min⁻¹; Table 1, Figure S2) with an 18-fold observed reduction in K_m (lMet, 397 ± 27 μM; l^tMet, 7,212 ± 1,250 μM) and enabled ~50% conversion to desired Ado^tMet in 20 h in a standard nonoptimized small scale assay (10 mM l^tMet, 2 mM ATP, 10 μM hMAT2A, 25 mM Tris·HCl, pH 8.0, 10 mM MgCl₂, 50 mM KCl, 37 °C). Using these parameters as a guide, a subsequent large scale (100 mL) chemoenzymatic reaction followed by preparative HPLC provided 19.9 mg of Ado^tMet (0.047 mmol, ~25% recovered yield based on the limiting reagent ATP), the purity and composition of which was confirmed via HPLC, NMR, and HRMS (see Supporting Information, Figures S1 and S6).

Figure 2.

Synthesis of AdoMet analogs and corresponding relevant degradation pathways. (A) Synthesis of l^tMet: (i) (Boc)₂O, pyridine, NH₄HCO₃, rt, 5 h, 97%; (ii) (TFA)₂O/pyridine (1:1), THF, 0 °C, 3 h, 80%; (iii) NaN₃, ZnBr₂, H₂O/2-propanol (2:1), 80 °C, 16 h, 65%; (iv) Et₂NH, CH₂Cl₂, rt, 0.5 h, 82%. (B) General scheme for hMAT2A-catalyzed synthesis of AdoMet, Ado^tMet, ^7dzAdoMet, and ^7dzAdo^tMet. (C) Degradative pathways for Ado^tMet, ^7dzAdoMet, and ^7dzAdo^tMet. Fmoc, fluorenylmethyloxycarbonyl.

Table 1.

Kinetic Parameters and Stability Measurements

enzyme	substrate	kcat min−1	Km μmol L−1	kcat/Km relative	t1/2 mina
hMAT2A	l-Metb	8.0 ± 0.1	397 ± 29	1	-c
	l-tMetb	6.1 ± 0.6	7,212 ± 1,250	0.04	-c
DnrK	ClNPd	0.019 ± 0.001	106 ± 10	N.A.e	-c
	AdoMetf	0.017 ± 0.001	138 ± 7	1	942 ± 7
	AdotMetf	0.031 ± 0.001	335 ± 40	0.8	5007 ± 6
	7dzAdoMet	– c	– c	– c	1243 ± 1
	7dzAdotMet	– c	– c	– c	no degradationg

^aStability studies conducted at 37 °C, pH 8.0.

^bhMAT2A substrates were assayed in the presence of saturating ATP (2 mM).

^cNot determined.

^dDnrK substrate ClNP was assayed in the presence of saturating AdoMet (1.6 mM).

^eNot applicable.

^fDnrK substrates AdoMet and Ado^tMet were assayed in the presence of saturating ClNP (1.5 mM).

^gNo detectable degradation after 3300 min ± standard error (SEM) at 95% confidence interval (CI).

Evaluation of AdoMet Carboxylate Isosteres

The subsequent comparison of AdoMet and Ado^tMet stability in 100 mM Tris·HCl, pH 8.0, at 37 °C revealed Ado^tMet as >5-fold improved in overall stability (Ado^tMet t_1/2 = 5,007 ± 6 min; AdoMet t_1/2 = 942 ± 7 min; Table 1, Figures S1 and S5). Consistent with our hypothesis, similar levels of AdoMet and Ado^tMet depurination (Figure 1, pathway b) were observed while Ado^tMet enabled complete inhibition of the observed AdoMet degradative intramolecular cyclization reaction (Figure 1, pathway a). It is important to note that while the decay of AdoMet has been reported to follow zero order⁵⁰ or first order⁶¹ kinetics, our results are consistent with the former as is our determined AdoMet t_1/2 value consistent with those previously reported.^50,56,61 To assess the utility of the stabilized Ado^tMet as a cosubstrate for MTs, we selected the prototypical class I MT DnrK as a model. DnrK is the carminomycin 4-OMT that catalyzes a culminating step in the biosynthesis of the anticancer agent daunorubicin in Streptomyces peucetius and has also been reported to methylate various modified anthracyclines and flavonoids as surrogate substrates.^79–84 Capitalizing on the permissive nature of DnrK and inspired by the recent application of ClNP as a surrogate acceptor for glycosyltransferases,⁸⁵ we developed a real-time colorimetric assay for DnrK-catalyzed alkylation using ClNP as a surrogate acceptor substrate where alkylation abolishes the classical nitrophenolate color (A₄₁₀) observed under basic conditions (Figure 3A). While the kinetic parameters for the native DnrK substrate carminomycin have not been reported, the determined ClNP kinetic parameters (k_cat = 0.019 ± 0.001 min⁻¹, K_m = 106 ± 10 μM with AdoMet as the alkyl-donating cosubstrate) are consistent with ClNP as a sufficient non-native surrogate where clear concentration dependency was observed in realtime assays (Figure 3B). A comparison of the DnrK kinetic parameters for AdoMet (k_cat = 0.017 ± 0.001 min⁻¹, K_m = 138 ± 7 μM) and Ado^tMet (k_cat = 0.031 ± 0.001 min⁻¹, K_m = 335 ± 40 μM) revealed nearly equivalent specificity constants (k_cat/ K_m; Table 1, Figure S4). As a simple robust high-throughput colorimetric assay for a prototypical class I model MT, this newly validated assay also presents an enabling platform for future class I MT engineering, biochemical studies, and/or tool development.

Figure 3.

DnrK colorimetric assay. (A) General scheme for a high-throughput DnrK colorimetric assay enabled by the surrogate acceptor 2-chloro-4-nitrophenol (ClNP). (B) Representative assay results demonstrating DnrK-catalyzed alkylation of ClNP diminishes classical ClNP color (A₄₁₀) over time [30 μM DnrK; 500 μM ClNP; AdoMet: (i) 100 μM, (ii) 500 μM, (iii) 800 μM, (iv) 1,000 μM, (v) 1,600 μM, and (vi) 2,000 μM; 25 mM Tris· HCl; pH 8.0].

Elucidation of DnrK/Ado^tMet/ClNP Ligand-Bound Structures

To better understand the molecular recognition of surrogate substrates Ado^tMet and ClNP by DnrK, the three-dimensional structures of the DnrK/S-adenosyl-tetrazole-lhomocysteine (Ado^tHcy, the byproduct after methyltransfer) and DnrK-ClNP-AdoHcy were determined at 2.25 and 1.82 Å resolution in space groups C2 and P2₁, respectively. The average real-space correlation coefficient of all ligands built into these two DnrK structures is over 0.96, and the electron density maps can be found in Figure S7. In both crystal forms, DnrK is identified as a tightly packed dimer with a buried surface area of ~3100 Ų, which is consistent with previously reported DnrK structures.⁸⁴ As expected, the DnrK-Ado^tHcy structure reveals Ado^tHcy and AdoHcy to bind in a nearly identical manner where the Ado^tHcy tetrazolate forms extra hydrogen bonds to water molecules and the corresponding sulfur resides slightly closer to the donor nucleophile (Figure 4A). This observation is also in agreement with reports of reduced donor–acceptor distances in O-MTs correlating with higher turnover rates,⁸⁶ as the k_cat of DnrK with Ado^tMet is nearly double the k_cat of DnrK with AdoMet. Consistent with the biochemical data, the orientation of one molecule of CINP in the DnrK-ClNP ligandbound structure is poised for alkylation reminiscent of that of the native acceptor carminomycin. Surprisingly, five additional ordered CINPs were also observed in the same structure (Figure 4B), suggesting a wide array of unproductive ClNP binding modes by virtue of key hydrogen bonds and π-stacking interactions.

Figure 4.

DnrK ligand-bound structures. (A) DnrK-Ado^tHcy (gray) and DnrK-AdoHcy-carminomycin (green) ligand-bound structures. Polar contacts between Ado^tHcy and nearby residues (stick models) and water molecules (spheres) are highlighted by dashed lines. In this model, the distance between carminomycin O-4 and the AdoHcy or Ado^tHcy sulfur is 4.2 and 3.8 Å, respectively. (B) DnrK-AdoHcy-ClNP (purple) and DnrK-AdoHcy-carminomycin (green) ligand-bound structures with ligands represented as stick models. In this model, the distance between carminomycin O-4 or ClNP O-1 and the AdoHcy sulfur is the same (4.2 Å).

Synthesis and Evaluation of AdoMet Surrogates Containing Base Isosteres

To confirm that base isosteres prohibit AdoMet depurination and to assess the potential for synergism with carboxylate isosteric replacement, the turnover of ^7dzATP by hMAT2A in the presence of lMet and l^tMet was subsequently assessed. In hMAT2A-catalyzed 24 h end-point assays, ^7dzATP (1 mM) and lMet (10 mM) gave 92% conversion to ^7dzAdoMet compared to 57% conversion with ^7dzATP (1 mM) and l^tMet (10 mM; Figure S3). Stability studies with the purified ^7dzAdoMet and ^7dzAdo^tMet confirmed the corresponding ^7dzAdoMet as resistant to depurination and ^7dzAdo^tMet as completely resistant to depurination and intramolecular cyclization (^7dzAdoMet t_1/2 = 1243 ± 1 min; ^7dzAdo^tMet t_1/2 = no detectable degradation over 3300 min; Table 1, Figures S1 and S5). Finally, simple coupled ClNPbased hMAT2A-DnrK assays (10 μM hMAT2A, 30 μM DnrK, 2 mM ATP or ^7dzATP, 10 mM lMet or l^tMet, 1 mM ClNP, 25 mM Tris·HCl, pH 8.0, 10 mM MgCl₂, 50 mM KCl, 37 °C, 24 h) revealed all stabilized AdoMet surrogates to serve as suitable MT cosubstrates (ATP/lMet, 52%; ATP/l^tMet, 19%; ^7dzATP/lMet, 21%; ^7dzATP/l^tMet, 14%; Figure S1 and Table S1).

Conclusions

In summary, this study highlights the efficient chemoenzymatic synthesis of enzymatically competent AdoMet isosteres that are considerably less prone to the two primary AdoMet physicochemical degradation pathways (intramolecular cyclization and depurination, Figure 1), where the corresponding tetrazole isosteric replacement also restricts the potential for AdoMet carboxylate participation in sulfonium epimerization.^15,60 Based on the DnrK-Ado^tMet biochemical and structural studies presented, Ado^tMet is able to substitute for AdoMet as a functional enzyme cosubstrate and implicates the isosteric modification as potentially advantageous in the context of other AdoMet-dependent enzymatic processes, as exemplified by AdoMet decarboxylase (AdoMetDC)-catalyzed AdoMet decarboxylation. Notably, AdoMetDC, a central catalyst in polyamine biosynthesis, is a validated anticancer and antiprotozoal drug target where substrate surrogates lacking the requisite AdoMet carboxylate (reminiscent of Ado^tMet) serve as potent inhibitors.^63,87 Thus, the new conceptual framework for engineering AdoMet chemical and metabolic stability put forth presents unique opportunities to advance AdoMet biocatalytic and therapeutic applications including, but not limited to, those involving MTs.

METHODS

Gene Cloning, Protein Expression, and Purification

N-His₆-hMAT2A was produced and purified as previously described.⁴² The dnrK gene encoding carminomycin 4-O-methyltransferase from Streptomyces peucetius was amplified from pWHM903⁷⁹ using primers (forward, 5′-AGAGCAGTCATATGACCGCTGAACCGACCGTCGCGGCCCGGCCGCAGCAG-3′; reverse, 5′-TACAGTGAATTCTCAGGCGCCGGTGGCCGCGGGGGCAAGGAC-3′) containing NdeI (forward) and EcoRI (reverse) restriction sites. The PCR product was digested and ligated into pET28a to provide an expression plasmid to enable N-His₆-DnrK production. The corresponding E. coli BL21(DE3) host was grown in the presence of 50 μg mL⁻¹ of kanamycin at 37 °C to an OD₆₀₀ of ~0.6, at which point the temperature was lowered to 25 °C and gene expression induced with 0.5 mM IPTG. Cells were allowed to continue to grow at 25 °C for approximately 18 h at 220 rpm. Cells were harvested by centrifugation (30 min, 5000 rpm), resuspended in buffer A (20 mM NaH₂PO₄, 300 mM NaCl, 10 mM imidazole, pH 7.8), and lysed via incubation with 1 mg mL⁻¹ lysozyme (~50 000 U mg^–1) for 30 min on ice followed by sonication (VirSonic 475; Virtis, Gardiner, NY; 100 W, 10 × 15 s pulses, ~ 1 min between pulses) on ice. N-His₆-DnrK was purified via affnity chromatography (5 mL HiTrap Ni-NTA chelating column, GE Healthcare, Piscataway, NJ) using a standard linear gradient (50 mM NaH₂PO₄, 300 mM NaCl, pH 8.0 with a linear gradient of imidazole of 10–500 mM) using an AKTA Purifier 10 (GE Healthcare). Concentration of the pooled fractions containing the purified protein was accomplished using an Amicon Ultracel centrifugal filter unit (30 kDa molecular weight cutoff; Merck Millipore Ltd., Tullagreen, Carrigtwohill, Co. Cork, IRL). Buffer exchange of combined and concentrated fractions containing the purified protein was accomplished using a PD-10 column (GE Healthcare) eluted with 25 mM Tris-HCl, pH 8.0. Protein concentrations were determined by Bradford assay (Bio-Rad, Hercules, CA, USA) using BSA as a standard. For this study, all proteins retained the N-terminal-His₆ affnity tag.

MAT Assay

To determine the K_m and k_cat of hMAT2A with varying concentrations of lMet or l^tMet, in vitro hMAT2A reactions were conducted in a volume of 50 μL with saturating ATP (2 mM), varied concentrations of lMet or l^tMet (20, 50, 100, 200, 500, 1000, 2000, 4000, 6000, 8000, and 10 000 μM), 10 μM purified hMAT2A in 25 mM Tris, pH 8.0, 10 mM MgCl₂, and 50 mM KCl. Reactions were incubated for 10 min at 37 °C and subsequently quenched by adding an equal volume of MeOH followed by centrifugation (10 000g, 20 min) to remove the precipitated protein. Product formation for each reaction was subsequently analyzed by RP-HPLC using method A (see Supporting Information, Figure S1). For each reaction, product (AdoMet or Ado^tMet) concentration was based upon the integration of species at 260 nm and calculated by multiplying the initial concentration of ATP by the quotient of the integrated product (AdoMet plus MTA or Ado^tMet) HPLC peak area (mAU*sec) over the sum of the integrated peak areas for product and remaining ATP. MTA derives from AdoMet and was thereby also considered as contributing to the total product concentration. Assays were repeated in triplicate where the Michaelis–Menten plots in Figure S2 represent an average of replicates. Controls lacking hMAT2A, lMet, l^tMet, and/or ATP led to no product.

To determine if ^7dzATP (TriLink Biotechnologies, San Diego, CA) was a viable cosubstrate for hMAT2A in combination with lMet or l^tMet, hMAT2A assays containing 10 000 μM lMet or l^tMet and variable ^7dzATP (500 μM, 1000 μM, 2000 μM, and 4000 μM) were conducted in a similar fashion. Aliquots recovered at defined incubation times (10 min, 30 min, 60 min, 1500 min, and 2940 min) were analyzed via HPLC following an identical protocol as above, and the results are summarized in Figure S3. Controls lacking hMAT2A, lMet, l^tMet, and/or ^7dzATP led to no product.

DnrK Assay

DnrK reactions were conducted in a volume of 100 μL containing 30 μM purified DnrK in 100 mM Tris, pH 8.0. Acceptor (2-chloro-4-nitrophenol, ClNP) concentrations were saturating (1500 μM) or variable (7.8125, 31.25, 125, 500, 1000, 1500, 1750, or 2000 μM), and donor (AdoMet or Ado^tMet) concentrations were saturating (1600 μM) or variable (25, 50, 100, 200, 400, 800, 1600, and 3200 μM). Reactions were incubated for 120 min at 37 °C and subsequently quenched and analyzed via HPLC as previously described for the hMAT2A assays using the HPLC method A with detection at 317 nm (Figure S1). Assays were repeated in triplicate where the Michaelis– Menten plots in Figure S4 represent an average of replicates. Corresponding identical plate-based DnrK assays (see Supporting Information) were conducted in clear-bottom 96-well plates where the plate reader limit of quantification required assay dilution at ClNP concentrations >500 μM. The change in absorbance (ΔA₄₁₀, representing disappearance of ClNP) over time was used as an indirect measure of ClNP methylation where product concentration was inferred by multiplying the percentage change in A₄₁₀ at t = 0 min and A₄₁₀ at t = 120 min by the initial ClNP concentration (Figure S4).⁸⁵ Controls lacking DnrK, AdoMet, Ado^tMet, and/or ClNP led to no product.

Coupled hMAT2A-DnrK Assay

Coupled hMAT2A-DnrK reactions were conducted in a total volume of 50 μL containing 25 mM Tris·HCl, pH 8.0, 50 mM KCl, 10 mM MgCl₂ with 10 μM hMAT2A, 30 μM DnrK, 1000 μM ClNP (saturating), and one of the following combinations of hMAT2A cosubstrates per reaction: lMet (10 000 μM) and ATP (2000 μM), l^tMet (10 000 μM) and ATP (2000 μM), lMet (10 000 μM) and ^7dzATP (2000 μM), or l^tMet (10 000 μM) and ^7dzATP (2000 μM). Reactions were incubated at 37 °C in PCR tubes on a thermocycler with a heated lid for 24 h and subsequently quenched and analyzed via HPLC as previously described for DnrK assays using HPLC method A (Figure S1 and Table S1). Controls lacking hMAT2A, DnrK, ClNP, lMet, l^tMet, ATP, and/or ^7dzATP led to no product.

Chemoenzymatic Synthesis and Purification of Ado^tMet, ^7dzAdoMet, and ^7dzAdo^tMet

Preparative MAT reactions to produce Ado^tMet (10 mM l^tMet, 1 mM ^7dzATP), ^7dzAdoMet (10 mM lMet, 1 mM ^7dzATP), or ^7dzAdo^tMet (10 mM l^tMet, 1 mM ^7dzATP) were conducted in a volume of 100 mL, 0.8 mL, or 0.8 mL, respectively with 10 μM hMAT2A in 25 mM Tris·HCl, pH 8.0, 50 mM KCl, and 10 mM MgCl₂.

The preparative Ado^tMet reaction was conducted in 2 × 50 mL Falcon tubes incubated at 37 °C for 48 h in a hybridizing oven (Techne Hybridizer HB-1D, Bibby Scientific, Staffordshire, United Kingdom) with rotation. After 48 h, the reactions were combined in a 500 mL round-bottom flask, flash-frozen, and lyophilized to dryness. The mixture was purified via C₁₈ column (3.5 cm × 18 cm) washed with 800 mL of ddH₂O (wash) followed by elution with 600 mL of 5:95 CH₃CN/ddH₂O, 600 mL of 10:90 CH₃CN/ddH₂O, and 800 mL of 15:85 CH₃CN/ddH₂O (40 mL fractions collected throughout elution). Analytical HPLC (Method A) revealed fractions 18–50 to contain the desired pure Ado^tMet, and these product-containing fractions were combined, concentrated in vacuo, flash frozen, and lyophilized to yield 19.9 mg of a brown-orange amorphous solid. Aqueous stock solutions were prepared in 10% EtOH and 5 mM H₂SO₄ aqueous solution (identical to commercial AdoMet stock preparations) where corresponding concentrations were based upon the molar extinction coefficient for AdoMet (15 400 L mol⁻¹ cm⁻¹ at 254 nm)⁶⁰ and verified via analytical HPLC (method A) comparison to commercial AdoMet standards.

The preparative ^7dzAdoMet/^7dzAdo^tMet reactions were conducted in several PCR tubes (8 × 100 μL for each reaction) and incubated in a thermocycler (with a heated top to prevent evaporation) at 37 °C for 1.5 h (lMet and ^7dzATP) or 24 h (l^tMet and ^7dzATP). Recombinant shrimp alkaline phosphatase (rSAP; New England BioLabs, Ipswich, Massachusetts, USA) was subsequently added to a final concentration of 0.1 U μL⁻¹ and the reactions incubated for an additional 20 min (to convert unreacted ^7dzATP to ^7dzadenosine). For each target, the cumulative reactions were combined, quenched with an equal volume of MeOH, and centrifuged (10 000g, 20 min, 4 °C) to remove precipitated protein and the supernatant concentrated to ~250 μL in vacuo. The desired product in each case was recovered via HPLC (method B), concentrated, flash frozen, and lyophilized to dryness. Aqueous stock solutions were prepared in 10% EtOH and 5 mM H₂SO₄ aqueous solution (identical to commercial AdoMet stock preparations) where corresponding concentrations were based upon the molar extinction coefficient for 7-deaza-adenosine (9100 L mol⁻¹ cm⁻¹ at 260 nm)⁸⁸ and verified via analytical HPLC (method A) comparison to commercial ^7dzATP and AdoMet standards.

Determination of t_1/2 of AdoMet, Ado^tMet, ^7dzAdoMet, and ^7dzAdo^tMet

AdoMet, Ado^tMet, ^7dzAdoMet, or ^7dzAdo^tMet (1 mM 10% EtOH, 5 mM H₂SO₄ stock) was diluted to 100 μM in 100 mM Tris buffer at a pH of 8.0 and incubated at 37 °C. Aliquots of each solution were taken at t = 0 min (and at various time points thereafter) and analyzed via HPLC (method A, A₂₅₄). The concentrations of AdoMet species and corresponding degradation products were determined and graphed on an XY-scatterplot, and linear regression was performed on each data set to obtain degradation rates (Figure S5) as previously described.^50,89 The t_1/2 of each AdoMet species was determined from the line slope considering zeroth order kinetics, as in the following eq 1:

$$t_{1∕2}=\frac{{\left[\text{AdoMet}\right]}_0}{2k}$$ (1)

where t_(1/2) is the time (in min) that it takes for one-half of a chemical species (in this example, AdoMet) to degrade, [AdoMet]₀ is the starting % concentration of a chemical species, and k is the zero-order rate constant of degradation of a chemical species (in % min⁻¹).