Mechanistic and Structural Insights into a Divergent PLP-Dependent l-Enduracididine Cyclase from a Toxic Cyanobacterium

Cyclic arginine noncanonical amino acids (ncAAs) are found in several actinobacterial peptide natural products with therapeutically useful antibacterial properties. The preparation of ncAAs like enduracididine and capreomycidine currently takes multiple biosynthetic or chemosynthetic steps, thus limiting the commercial availability and applicability of these cyclic guanidine-containing amino acids. We recently discovered and characterized the biosynthetic pathway of guanitoxin, a potent freshwater cyanobacterial neurotoxin, that contains an arginine-derived cyclic guanidine phosphate within its highly polar structure. The ncAA l-enduracididine is an early intermediate in guanitoxin biosynthesis and is produced by GntC, a unique pyridoxal-5′-phosphate (PLP)-dependent enzyme. GntC catalyzes a cyclodehydration from a stereoselectively γ-hydroxylated l-arginine precursor via a reaction that functionally and mechanistically diverges from previously established actinobacterial cyclic arginine ncAA pathways. Herein, we interrogate l-enduracididine biosynthesis from the cyanobacterium Sphaerospermopsis torques-reginae ITEP-024 using spectroscopy, stable isotope labeling techniques, and X-ray crystallography structure-guided site-directed mutagenesis. GntC initially facilitates the reversible deprotonations of the α- and β-positions of its substrate before catalyzing an irreversible diastereoselective dehydration and subsequent intramolecular cyclization. The comparison of holo- and substrate-bound GntC structures and activity assays on site-specific mutants further identified amino acid residues that contribute to the overall catalytic mechanism. These interdisciplinary efforts at structurally and functionally characterizing GntC enable an improved understanding of how nature divergently produces cyclic arginine ncAAs and generate additional tools for their biocatalytic production and downstream biological applications.

14 Fig. S5. In vitro GntC assays with N-or C-terminally His6-tagged enzyme.  Table S1. NCBI accession codes for sequences used in this study 36 Table S2. Data collection and refinement statistics of GntC 37 Table S3. Residues modeled in each chain of GntC (residues 1-370) 38 Table S4. GntC  GntC heterologous protein expression A general method from a literature reference 1 was followed for GntC and each GntC mutant. After sequence confirmation, purified plasmid was transformed into E. coli DH10b and BL21(DE3) cells. From these cultures, glycerol stocks were prepared, and a colony was picked from the E. coli BL21(DE3) plate for protein expression starter cultures for the GntC mutants. Starter cultures contained 15 mL LB media, 50 mg/mL kanamycin, and a colony of the appropriate GntC mutant, and were incubated overnight at 37 °C at 200 rpm. After overnight incubation, 10 mL of the starter culture was inoculated into 1 L of Terrific Broth media containing 50 mg/mL kanamycin and was incubated at 37 °C and 200 rpm until the culture reached an OD600 of 0.8. Afterword, the 1 L culture was cooled to 18 °C with 200 rpm for 1 hour prior to the addition of 100 µM isopropyl-b-Dthiogalactopyranoside (IPTG) to induce protein expression. Cultures were incubated overnight at 18 °C and 200 rpm prior to cell harvesting. Cultures were harvested by centrifugation at 2500 x g at 4 °C for 30 minutes, then resuspended in 30 mL cell lysis buffer (1 M NaCl, 20 mM Tris-HCl pH 8.0, 20 mM imidazole, 10% glycerol) and stored at -80 °C until protein purification.

GntC protein purification
Protein purification conditions adapted from a literature reference. 1 E. coli BL21 (DE3) cell pellets containing (His)6-gntC variants were thawed, then sonicated on ice for a total of 6 minutes at 40% amplitude with 15 seconds on, and 45 seconds off (FisherBrand Model 505 Sonic Dismembrator, 3.2 mm microtip). The lysate was clarified by centrifugation at 15,000 x g for 30 minutes at 4 °C. Proteins were purified at 4 °C using an AKTAGo FPLC system with a HisTrap FF (5 mL) column (GE Healthcare Life Sciences) preequilibrated with at least 25 mL Buffer A (30 mM Tris pH 8.0, 30 mM imidazole, 300 mM NaCl) at a maximum flow rate of 2 mL/minute. Buffers used for FPLC purification were filtered through a 0.22 µM nitrocellulose membrane prior to usage. After loading the clarified cell lysate onto the column, the column was rinsed with Buffer A until the UV absorbance returned to baseline, then the column was washed with 10% Buffer B (20 mM Tris pH 8.0, 1 M NaCl, and 250 imidazole) to remove non-specifically bound proteins using either 25 mL Buffer B or until the UV absorbance returned to baseline. His6-GntC was eluted with a linear gradient to 100% B over 60 mL in 30 minutes, while collecting 5 mL fractions. The eluate fractions were assessed for purity using a 10% SDS-PAGE gel, and fractions with over 90% GntC, were combined and concentrated to less than 2.5 mL using by Amicon ultra centrifugal filters (30 kDa molecular weight cut-off (MWCO), EMD-Millipore). Proteins were buffer exchanged into GF Buffer (50 mM HEPES pH 8.0, 300 mM KCl) using a preequilibrated PD-10 column (GE Healthcare Life Sciences). Protein concentrations were estimated using the Bradford assay with a bovine serum albumin standard and, if necessary, protein was further concentrated with the 30 kDa Amicon ultra centrifugal filter. After obtaining purified protein, the protein was aliquoted into 50 µL aliquots and stored at -80 °C.
*During FPLC elution, protein eluted with the absence of PLP and appeared to aggregate in a concentrationdependent manner, in which 500 µL of 2.00 mM PLP was added to each fraction to stabilize the enzyme. After elution, aggregate was removed via centrifugation at 20,000 x g at 4 °C for 10 minutes to remove aggregates.

Protein oligomerization assessment
To assess oligomerization of site-directed GntC variants, a Superdex 200 size exclusion column (GE Healthcare Life Sciences) was equilibrated with 40 mL of GF buffer. Protein samples were diluted to 4 mg/mL, and 60 µL was directly injected to the FPLC. Using a flow rate of 0.20 mL/minute, fractions were eluted, and the main dimer peak was collected.
GntC buffer exchange for D2O assays Enzyme buffer exchange procedures were adapted from a literature reference. 2 A 500 µL aliquot of GntC was thawed on ice from storage at -80 °C. The aliquot would be pipetted into a 30 kDa MWCO Amicon centrifugal filter with 50 mM K2HPO4 at pH 8.0 (5 mL) in D2O and centrifuged at 3,400 x g for 20 minutes. The GntC aliquot would be washed with two additional K2HPO4 D2O washes prior to Bradford assay to determine enzyme concentration.

GntC apoenzyme preparation
GntC coelutes with the cofactor PLP during protein purification. To obtain apo-GntC, a 1 mL aliquot of GntC was thawed on ice from storage at -80 °C. GntC was added to GF buffer and 5 mM hydroxylamine to a volume of 10 mL. The solution was incubated on ice for 4 hours prior to adding to a 30 kDa MWCO Amicon centrifugal filter with GF buffer and centrifuged at 3,400 x g for 20 minutes. The solution was centrifuged with two additional GF buffer washes prior to Bradford assay to determine enzyme concentration.

Enzyme Assay Methods H2O enzyme assays
Wild-type (WT) and mutant GntC assays were conducted in a similar manner to previously reported. 1 Assays were conducted in 50 mM K2HPO4 (pH 8.0), using 1 mM substrate, 100 µM PLP, and 50 µM purified GntC enzyme. Total reaction volumes were brought to a total volume of 50 µL using MilliQ water and incubated at room temperature overnight. After incubation, reactions were derivatized using the detailed protocol above for Marfey's derivatization, then subjected to LCMS analysis.

Apoenzyme GntC assay
GntC PLP-stripped assays were conducted in 50 mM K2HPO4 (pH 8.0) using 1 mM substrate, and 50 µM apoenzyme GntC. Reaction volumes were brought to 50 µL with MilliQ water and incubated overnight at room temperature. Reactions with derivatized with Marfey's reagent (protocol detailed above) prior to LCMS analysis.

D2O enzyme assays
GntC deuterium incorporation assays were conducted in 50 mM K2HPO4 (pH 8.0) using 1 mM substrate, 100 µM PLP, and 38 µM of purified GntC enzyme that was buffer exchanged into D2O and 50 mM K2HPO4. Reaction volumes were brought to their total volume of 50 µL with D2O. Assays were incubated at room temperature overnight, then derivatized with Marfey's reagent prior to LCMS analysis.

H2 18 O enzyme assays
To help establish the irreversibility of a step of the GntC mechanism, GntC assays were conducted in 50% H2 18 O (Cambridge Isotope Laboratories) by using 50 mM K2HPO4 (pH, 8.0), 1 mM substrate, 100 µM PLP, and 50 µM GntC. All reagents were stored in MilliQ water, and enzyme assays were diluted to a final volume of 50 µL using H2 18 O. After 16 hours of incubation at room temperature, the reactions were derivatized with Marfey's reagent for LCMS analysis.

NMR enzyme assays
GntC NMR assays were conducted in 50 mM K2HPO4 (pH 8.0), 2.6 mM substrate (unless otherwise stated), 100 µM PLP, and 30 µM purified WT GntC. GntC enzyme was buffer exchanged into D2O using the protocol above. Reaction volumes were brought to 500 µL with D2O. Reactions placed into a clean, oven-dried NMR tube and analyzed using 1 H NMR obtained on a Bruker 500 MHz NMR instrument using the default settings except for using 256 scans rather than 16 scans.

Spectroscopy assays
GntC UV-Vis assays were adapted from a literature reference. 3 Assays were conducted in 50 mM K2HPO4 (pH 8.0), 1 mM substrate, and 40 µM purified WT GntC enzyme. No additional PLP was added to the reaction, and reaction volumes were brought to 500 µL using MilliQ water. Assays were analyzed in a 500 µL, 10 mm quartz cuvette cell (VWR International, LLC) using an Eppendorf BioSpectrometer between 300 and 550 nm.

Standard curve preparation
To establish a calibration curve for GntC kinetic assays, standards of 2 were prepared in 50 µL aliquots in triplicate. All conditions included 50 mM KPi (pH 8) and 100 µM PLP, and the range of concentrations for 2 were 0.75 µM, 1 µM, 5 µM, and 50 µM. Prior to derivatization, 2.75 µL of 10 mM glycine was added with 2.25 µL of MilliQ water to use the 500 µM glycine internal standard, and have a final volume of 55 µL. The standards were derivatized using the detailed protocol above and the supernatant was subjected to UPLC-MS analysis.

Enzyme kinetics assay
GntC kinetics assays were set up in 50 mM K2HPO4 (pH 8.0), 100 µM PLP, 10 µM GntC, and varying concentrations of substrate and brought to a total volume of 50 µL with MilliQ water. The following substrate concentrations were used in triplicate: 100 µM, 150 µM, 250 µM, 625 µM, 1,250 µM, 2,500 µM, 5,000 µM. The concentration conditions were divided into two groups to ensure accurate reaction times. Reactions incubated at room temperature (24 °C) for 30 minutes, prior to the addition of 20 µL saturated aqueous sodium bicarbonate, 2.25 µL MilliQ water, and 2.75 µL of 10 mM glycine. Derivatization took place immediately afterwards with the addition of 100 µL 1% w/v Marfey's reagent in acetone, following the detailed protocol above. The supernatant was subjected to UPLC-MS analysis.

Protein Crystallography Methods Protein expression and purification for crystallography experiments
For crystallography studies, GntC was expressed and purified in a similar manner above with the following changes: A single colony was inoculated into a 10 mL of LB starter culture, grown at 37 °C and 220 rpm shaking overnight. The starter culture was then inoculated into 1 L LB media, grown at 37 °C and 220 rpm shaking until OD600 reaches 1.0. An hour after the temperature adjustment, the protein expression was induced by IPTG at 1 mM final concentration. Cells were harvested by centrifugation (8,000 x g, 4 °C, 10 min) 16 hours after induction.
For protein purification the following changes were made from the protocol above: The column was further equilibrated with 40 mL of Buffer A (25 mM Tris pH 8.0, 300 mM NaCl, 10 mM imidazole, 100 µM PLP). The column was eluted with a linear gradient to 100% Buffer B (25 mM Tris pH 8.0, 300 mM NaCl, and 250 mM imidazole) over 40 mL volume while collecting 5 mL fractions. The protein was further purified on a Superdex 75 16/60 size exclusion column (GE Healthcare Life Sciences) in the gel filtration buffer (100 mM MOPS pH 6.7 and 20 μM PLP). The dimeric peak from gel filtration was collected and concentrated by Amicon ultra centrifugal filters (30 kD molecular weight cut-off) to a final concentration of 10.6 mg/mL. The protein was flash-frozen in 25 μL aliquots and stored in -80 °C freezer.

Data collection and processing
A data set of GntC was collected at Advanced Light Source (Berkeley, California, USA) on beamline 8.2.1 using a ADSC Q315R detector at a temperature of 100 K. Resolution cutoffs were chosen based on I/s limit of 2. Data were indexed, integrated, and scaled in AutoProc. 4 A data set of GntC with 1 bound was collected at Stanford Synchrotron Radiation Lightsource (Menlo Park, California, USA) on beamline 12-2 using a Pilatus 6M PAD detector at a temperature of 100 K. Resolution cutoffs were chosen as CC1/2 ~ 0.7. Data were indexed, integrated, and scaled in XDS. 5 The crystal of GntC with 1 bound was determined by Phenix 6 Xtriage to have merohedral twinning, with a twin law of h, -k, -l. The twin fraction was refined in Phenix Refine to 48%, and the structure was refined to account for twinning.

Structure determination and refinement
The structure of GntC was determined to 2.10-Å resolution by molecular replacement (MR) using MRage 7 implemented in Phenix. A PLP-dependent aminotransferase (ZP_03625122.1) from Streptococcus suis 89/1591 (PDB ID: 3OP7), which shared 28.9% identity with GntC, was trimmed by Sculptor 8 and used as a model for molecular replacement. The MR gives one solution containing two GntC dimers per asymmetric unit (ASU) with a log-likelihood gain (LLG) value of 655 and a final translational function Z score (TFZ) of 20.6. The MR solution was used as an input for AutoBuild 9 implemented in Phenix for initial model building. After the AutoBuild run, the atomic coordinates and B-factors were iteratively refined in Phenix Refine with model building and manual adjustment of model in Coot. 10 TLS parameters were refined in the last few rounds of refinement in which each GntC monomer is a TLS group. Water molecules were added manually throughout real space refinements using Fo-Fc electron density contoured to 3.0s as criteria. Four-fold non-crystallographic symmetry (NCS) restraints were used throughout refinement. Restraints for Lys-PLP adduct were generated from Grade Web Server (Global Phasing).
The structure of GntC with 1 bound was determined to 2.04-Å resolution by rigid-body refinement from 2.10-Å resolution GntC structure. The atomic coordinates and B-factors was iteratively refined in Phenix Refine with model building and manual adjustment of model in Coot. Water molecules were added manually throughout real space refinements using Fo-Fc electron density contoured to 3.0σ as criteria. Four-fold non-crystallographic symmetry (NCS) restraints were used throughout refinement. Restraints for the PLP-1 adduct were generated from Grade Web Server (Global Phasing). A composite-omit electron density map calculated by Phenix Composite_omit_map was used to verify the model. Restraints for Lys-PLP adduct were generated from Grade Web Server (Global Phasing).

Supplementary figures and tables:
Figure S1. Previously reported in vivo guanitoxin stable isotope labeling results. 11 Heavy isotope labeled 1-d6 was successfully incorporated into guanitoxin when fed to toxin-producing Anabaena flos-aquae cyanobacterium. Only three of the original deuterium atoms were retained during this biotransformation as depicted in the guanitoxin-d3 structure above. Based on our previous biochemical characterization of GntD, GntE, and GntG 1 and GntC mechanistic results from this study, we propose the following biosynthetic intermediates (beginning with 2-d3), which would rationalize the in vivo re-introduction of the two hydrogen atoms adjacent to the dimethylamine moiety within isolated guanitoxin-d3.  Figure S2. In vitro GntC PLP dependency assay. To remove the co-purifying PLP cofactor and generate the apoenzyme, GntC was treated with hydroxylamine following the procedures above. Enzyme assays with apo-GntC (top trace), no enzyme (middle trace), and apo-GntC with 100 µM PLP exogenously added (bottom trace) were incubated at room temperature for 16 hours prior to Marfey's derivatization and UPLC-MS analysis. Relative intensities of positive mode extracted ion chromatograms from UPLC-MS traces were extracted for derivatized product 2 ([M+H] + 425.15 ± 0.50 m/z).  Figure S3. In vitro GntC pH preference assay. In vitro GntC assays were set up as previously described for 14 hours under buffered potassium phosphate at pHs 6 -10. An internal standard of 500 µM glycine was added prior to Marfey's derivatization to correct for any discrepancies during derivatization and for relative product conversion measurements via LCMS analysis. Relative intensities of positive mode extracted ion chromatograms from UPLC-MS traces were extracted for derivatized 2 and glycine ([M+H] + 425.15, 328.08 ± 0.50 m/z respectively). The 2:glycine peak areas ratios at the 5 pHs tested were compared and identified pH 8 as optimal for 2 production.  Figure S4. In vitro GntC divalent metal cation preference assay. In vitro GntC assays were set up as previously described for 16 hours with the addition of 1 mM EDTA or divalent metal salt (MgCl2 • 6H2O, CaCl2, MnSO4 • H2O, FeSO4 • 7H2O, CoBr2, CuSO4 • 5H2O, ZnCl2) following adaption from literature resources. 12 An internal standard of 1 mM glycine was added prior to Marfey's derivatization to correct for any discrepancies during derivatization and for relative product conversion measurements via LCMS analysis. Relative intensities of positive mode extracted ion chromatograms from UPLC-MS traces were extracted for derivatized 2 and glycine ([M+H] + 425.15, 328.08 ± 0.50 m/z respectively). The 2:glycine peak area ratios were compared to no added metal and showed negligible enhancement of 2 production by GntC.   The product peak area was adjusted using the peak area of the glycine internal standard and plotted to produce a standard curve to use for quantifying in vitro enzyme kinetics assays. Figure S7. In vitro kinetics assay of GntC Michaelis-Menten curve. In vitro enzyme assays were conducted in triplicate at room temperature for 30 minutes prior to Marfey's derivatization for UPLC-MS analysis. A 500 µM glycine internal standard was added to each condition prior to derivatization to quantify for product conversion.   Figure S9. UV-Vis spectra of GntC assays with product 2. In vitro enzyme assays were incubated at room temperature for up to 16 hours with 1 mM 2, 40 µM GntC and 50 mM K2HPO4 buffer (pH 8.0) and no additional PLP. UV-Vis absorption spectra were taken immediately after the addition of substrate 2 and then at the time points listed.    Figure S11. In vitro 1 H NMR GntC assay with substrate diastereomer 1′ in D2O. In vitro enzyme assays were set up in a ~50:1 substrate diastereomer 1′:GntC molar ratio in buffered D2O and analyzed using 500 MHz 1 H NMR. Following a 20 hour incubation of substrate diastereomer 1′ without (top trace) and with GntC (bottom trace), a substantial decrease in a-and b-hydrogen signals for 1′ were observed without any obvious cyclic or alternative products.  Figure S12. In vitro GntC assay with product L-enduracididine (2) in D2O. In vitro incubation of 2 for 16 hours without (top) and with GntC (bottom) in the presence of in D2O conditions. Following Marfey's derivatization and UPLC-MS analysis, up to three 3 deuterium incorporations occur in the product (2-d3) based on the mass spectrometry envelope.  Figure S13. In vitro GntC assay to assess overall reaction reversibility. In vitro enzyme assays were set up with product 2 and incubated for 14 Figure S14. In vitro GntC assay in 50% H2 18 O to assess dehydration reversibility. In vitro enzyme assays were set up as previously described in 50% H2 18 O buffered conditions. Following Marfey's derivatization and UPLC-MS analysis, no difference in isotopic mass distribution was observed for starting material 1 when comparing the no enzyme (top left) and with GntC (bottom left) experiments. This indicates that the dehydration step is not reversible under in vitro GntC assay conditions.  Figure S15. Sequence alignment of GntC to PLP-dependent capreomycidine cyclases. Enzymes listed are GntC (PDB ID: 8FFT), VioD, and OrfR (PDB ID: 4M2M, Chain A). Alignment was generated using Clustal Omega and viewed using ESPript 3.0. Sequence numbering follows the GntC sequence, and highlighted residues are orange with a triangle underneath them. Conserved residues include a glutamate residue towards the N-terminus of the sequences (GntC E9), a serine residue conserved amongst GntC, OrfR, and VioD (GntC S25), and a lysine residue (GntC K219). NCBI accession numbers for all sequences are listed in Table S1.     Figure S16. Sequence alignment of GntC to PLP-dependent intermolecular g-substitution enzymes and aspartate aminotransferase AspC. The sequences were aligned in Clustal Omega and visualized in ESPript 3.0.

Relative extracted ion intensity
Residues of interest are highlighted in orange, and the sequence numbering follows GntC. GntC shares minimal sequence homology to other g-substitution enzymes, such as PLP-specific residues D186 which allows for deprotonation of the nitrogen in the pyridoxal ring of PLP, and K219, which keeps PLP in the enzyme active site. NCBI accession numbers for all sequences are listed in Table S1.        Figure S18. Active site comparison of PLP-1 GntC to aspartate aminotransferase. Structure alignment of GntC (green) to aspartate aminotransferase (gray, PDB ID: 1ARS). Conserved residues between GntC and an aminotransferase include the canonical lysine residue needed for the internal aldimine (K219 in GntC). Some differences include a tryptophan residue that resides below the PLP in aspartate aminotransferase (W140) versus phenylalanine (GntC F108).  Figure S19. Active site comparison of PLP-1 GntC and VioD AlphaFold model. Structure alignments of GntC (green, PDB ID: 8FFU) and modeled VioD (gray, structure generated with AlphaFold 2.0). Structural alignments indicate a lysine residue is conserved (GntC K219) for external aldimine formation of the two PLP-dependent enzymes. Both structures also conserve two arginine residues (GntC R346 and GntC R227), and a histidine residue (GntC H108) that sits farther away from the pyridoxal ring of PLP. Intriguingly, GntC utilizes F108 to p stack with the PLP cofactor, whereas this residue is replaced with a tyrosine residue (Y121) in VioD. A glutamate residue (GntC E9) is conserved between both structures, which is also conserved in PLP-dependent OrfR and revealed in the sequence alignments ( Figure S12). An aromatic residue is also present which is a tyrosine residue (Y189) in GntC, and a phenylalanine residue (F202) in VioD. GntC Y189 is 3 Å from the hydroxyl group of PLP in the PLP-1 adduct and is 2.3 Å from the same hydroxyl group when PLP is present as an internal aldimine.  Figure S20. Active site comparison of PLP-1 GntC and Fgm3 AlphaFold model. Structure alignments of GntC (green, PDB ID: 8FFU) and Fgm3 (gray, generated using AlphaFold 2.0). 14 Structural alignments indicate a lysine residue (GntC K219) for internal aldimine formation, threonine residue (GntC T84) for coordinating to the substrate guanidine group, and histidine residue (GntC H109) are conserved. Although both enzymes utilize 1 as a substrate, the Fgm3 structure contains a tyrosine residue (Y225) that could coordinate with the g-hydroxy group of the substrate that GntC does not contain to putatively participate in retro-aldolase activity.  Figure S21. Analytical size exclusion chromatograms of wild-type GntC and mutants. GntC and variants were expressed and purified using the protocol detailed in the Methods and Materials section. Samples were diluted to 4 mg/mL before injection onto an analytical size exclusion column. No obvious retention time difference was observed between GntC and its site-directed mutants, suggesting similar homodimeric forms.     The procedure to synthesize SI-1 was adapted from literature references. 1,22 Boc-L-Asp-OtBu (1.002 g, 3.46 mmol) was dried and concentrated with toluene (15 mL) in vacuo to remove residual water. Nitromethane (20 mL) and 1,1′-carbonyldiimidazole (0.582 g, 3.59 mmol) were added sequentially and stirred at room temperature for 45 minutes under Ar gas. Potassium tert-butoxide (0.783 g, 6.98 mmol) was added and stirred at room temperature for an additional 3 hours. The reaction mixture was quenched by the addition of 50% glacial acetic acid (50 mL), then extracted using ethyl acetate (3 x 50 mL). Pooled organic layers were washed with water (50 mL), saturated sodium bicarbonate (50 mL), water again (50 mL), then brine (50 mL). The organic layers were dried over MgSO4, filtered, then concentrated in vacuo. The crude product was dried twice with toluene coevaporations, then resuspended in methanol (10 mL). Sodium borodeuteride (0.1499 g, 3.58 mmol) was added portion wise at 0 °C and let reach room temperature over the course of 25 hours. The reaction was quenched with 1 N HCl until the pH was 3-4. The reaction was concentrated in vacuo, resuspended in water (20 mL), then extracted with ethyl acetate (3 x 25 mL). The organic layers were pooled, washed with brine (20 mL), dried over MgSO4, filtered, then concentrated in vacuo. The crude product was purified via a silica flash column with a 4-6% diethyl ether:dichloromethane eluent system to separate the product diastereomers. A second column with a 90:5:5 toluene:THF:ethyl acetate eluent system was used to separate the diastereomer products from impurities that co-eluted from the initial column to give a white solid (0.343 g, 30%

SI-1
This reaction was adapted from a literature reference. 1 An aliquot of SI-1 (0.327 g, 0.98 mmol) was suspended in methanol (15 mL) and acetic acid (0.056 mL, 0.976 mmol). This solution had 10% Pd/C (0.117 g, 1.10 mmol) added to it, and was flushed with Ar gas, then was left under a H2 atmosphere for 16 hours. The reaction mixture was filtered through a Celite pad then concentrated in vacuo. The crude mixture was resuspended in toluene (20 mL) and N,N′-di-Boc-1H-pyrazole-1-carboxamidine (0.310 g, 1.00 mmol) and triethylamine (0.68 mL, 4.88 mmol) added sequentially. This solution was heated to 55 °C and stirred for 20 hours. The reaction was quenched with saturated aqueous NH4Cl (25 mL), and the organic components were extracted with ethyl acetate (3 x 25 mL), then dried over magnesium sulfate, and filtered prior to concentrating in vacuo. The crude reaction was purified with silica flash chromatography using an eluent of 4:1 hexanes:ethyl acetate + 0.1% triethylamine. Fractions were pooled, then concentrated in vacuo to yield the desired product as a white solid (0.297 g, 56% A solution of 1 N HCl (10 mL) was added to SI-2 (0.107 g, 0.195 mmol). The reaction was stirred for 2 hours, then concentrated in vacuo. The crude reaction mixture was resuspended in water (10 mL) and NaOH (0.039 g, 0.98 mmol), then 1 N HCl was added dropwise to the reaction until the pH reached 7, then lyophilized overnight to give a white solid (0.112 g, quantitative). 1