**1. Introduction**

Microcystins (MCs), such as MC-LR (**2**), are cyclic heptapeptide hepatotoxins (Figure 1) produced primarily in cyanobacterial genera such as *Microcystis*, *Dolichospermum (Anabaena)*, *Nostoc* and *Planktothrix (Oscillatoria)* [1]. The common feature of MCs is their cyclic structure and possession of several rare, highly conserved amino acids moieties [2]. One of the unusual structural features of MCs is the β-amino acid (2*S*,3*S*,8*S*,9*S*)-3-amino-9-methoxy-2,6,8-trimethyl-10-phenyldeca-4(*E*),6(*E*)-dienoic acid (Adda) at position 5 (Figure 1) [3]. Adda5 and Glu<sup>6</sup> appear to be primarily responsible for the characteristic biological activity of MCs [2,3]. Protein phosphatase inhibition is directly related to the toxins' mechanism of action and animal studies have demonstrated that MCs are potent tumor promoters [1]. To date, the number of identified MCs continues to increase and more than 250 analogues have been characterized [4]. However, due to a lack of standards for these analogues, very few studies have adequately assessed their distribution in natural waters.

**Figure 1.** Structures of the microcystins (MCs) mentioned in the text, showing their *m*/*z* values and characteristic mass spectral fragment ions in positive (*m*/*z* 135.0804) and negative (*m*/*z* 128.0353) ionization modes. Numbers in circles indicate the amino acid residue number, while atoms are numbered starting from the carboxyl carbon of each amino acid.

As part of feasibility studies for a cyanobacterial matrix reference material [5], a survey of cyanobacterial cultures from Canada was conducted using LC with UV and MS detection. Among the samples analyzed, two *Microcystis aeruginosa* cultures from Saskatchewan and Alberta, CPCC-464 and CPCC-299, showed the presence of a new microcystin tentatively identified as [D-Leu1]MC-LY (**1**) [5] together with the previously reported [6,7] and well-characterized [8] [D-Leu1]MC-LR (**3**). [D-Leu1]MC-LY (**1**) was also tentatively identified recently by LC–HRMS/MS in a cyanobacterial bloom sample from southwestern Ontario, Canada [9], indicating that it may be a significant component of natural cyanobacterial blooms in this and other parts of the world. It is therefore necessary to verify the structure, and to evaluate its toxicity relative to other MCs because limited data are available on the toxicological consequences of varying the amino acid at position 1 and 2.

#### **2. Results and Discussion**

The analysis of *M. aeruginosa* culture CPCC-464 by LC–UV–MS/MS is shown in Figure 2. A very similar profile was observed with culture CPCC-299, with the only differences being in relative peak areas. The LC–UV chromatogram of CPCC-464 (Figure 2a) showed two major peaks due to **1** and **3**. In the same experiment, the MS was operated with a precursor scan using the *m*/*z* 135 product ion for Adda, which is characteristic of most MCs [10]. Examination of all peaks in the total ion current chromatogram (Figure 2b) revealed three major MCs of interest, **1**, **3** and **6**, with [M+H]<sup>+</sup> ions at *m*/*z* 1044.5, 1037.5 and 1071.5, respectively. These ions are plotted as extracted ion chromatograms in Figure 2c. LC–HRMS chromatograms with and without mercaptoethanol derivatization are shown in Figures S1 and S2. Compounds **1** and **3** were hypothesized to be [D-Leu1]MC-LY and [D-Leu1]MC-LR, respectively [5], while the MS characteristics of **6** are consistent with [D-Leu1]MC-M(O)R, tentatively identified recently by LC–HRMS/MS in cyanobacterial blooms in Canada and the USA [9,11].

**Figure 2.** LC–UV–MS analysis of an extract of CPCC-464: (**a**) UV absorbance at 238 nm; (**b**) total ion current chromatogram from *m*/*z* 135 precursor scan; and (**c**) extracted ion chromatograms from *m*/*z* 135 precursor scan.

Large scale culturing of CPCC-464 followed by centrifugation provided 188 g of biomass for purification of **1**. This material was extracted with 70% MeOH–H2O, then taken through a preparative isolation procedure consisting of a hexane partitioning, C18 LC, LH-20 gel permeation, C18-flash chromatography, and semi-preparative HPLC. The total yield of **1** was 28.7 mg containing a small amount of [D-Leu1,D-Glu(OMe)6]MC-LY and a trace of what is believed to be [D-Leu1,(6*Z*)-Adda5]MC-LY (observed by LC–MS selected reaction monitoring modes).

The structure of **1** was elucidated from NMR spectra acquired in CD3OH in order to observe the exchangeable amide protons. The proton NMR spectrum had six resonances in the amide region with a profile similar to that of a peptide. Individual spin systems from each amide resonance were identified and assigned using 2D 1H-1H DIPSI-2 and 1H-1H COSY correlations (Table 1). Detailed spectra are provided in Figures S3–S10 and an overlay of chemical shifts on the proposed 2-dimensional chemical structure of **1** is shown in Figure S11.


**Table 1.** NMR spectroscopic data for [D-Leu1]MC-LY (**1**) in CD3OH*<sup>a</sup>*.


**Table 1.** *Cont*.

*<sup>a</sup>*s, singlet; d, doublet; t, triplet; q, quartet; m, multiplet; br, broad; ND, not detected.

Carbon assignments were determined indirectly using 1H-13C HSQC and 1H-13C HMBC 2D NMR spectra. One carbon resonance was not assigned for Leu<sup>1</sup> (C1) due to spectral overlap. The Adda unit was assembled with the aid of the HMBC data, which determined the positions of the methyl groups. *Trans*-configuration of the 4,5-double bond is indicated by the large coupling constant between Adda-H4 and -H5 (15.5 Hz) and by the observation of a ROESY correlation between Adda-H4 and Adda-6-Me, and is consistent with the absence of a ROESY correlation between Adda-H4 to Adda-H5. The second double bond was also *trans* as a ROESY correlation was observed between Adda-H5 and Adda-H7 (Figure 3).

**Figure 3.** Three-dimensional structure of [D-Leu1]MC-LY (**1**), modeled from the solution structure of MC-LR (**2**) [12], showing NoE correlations observed in the ROESY NMR spectrum of **1**.

The relative stereochemistry of C2 and C3 of Adda was determined from the observation of a ROESY correlation between Adda-H3 and both Adda-H5 and Adda-2-Me, while Adda-H2 showed correlations to Adda-NH, Glu-NH, Adda-2-Me and Adda-H4, indicating that H2 and H3 are on the opposite faces of the Adda plane (Figure 3). This is consistent with the ca. 9.7 Hz coupling constant between Adda-H2 and -H3. The glutamic acid unit (Glu), *N*-methyldehydroalanine (Mdha) and *erthyo*-β-methylaspartic acid (Masp) were identified in a similar manner, and their proton and carbon resonances were very similar to those previously published for **3** [8]. Two leucine units were identified by the similarities of their 1H and 13C resonances to those in the BMRB database (http://www.bmrb.wisc.edu/ref\_info/; accessed September 2011) and those published for **3** [10]. The relative stereochemistry for H2 and H3 of the Masp unit was determined from the absence of Masp-H2 to Masp-3-Me correlation and the presence of a ROESY correlation between Masp-H2 and -H3, which places H2 and H3 on the same side of the plane. The tyrosine unit (Tyr) was assigned from the presence of the two doublets at 6.99 and 6.62 ppm, characteristic of a para-substituted phenyl ring. The aromatic protons, Tyr-H5 and -H9 correlated to a carbon at 36.3 ppm, characteristic of an aromatic amino acid. The accurate mass from LC–HRMS/MS indicated that the substituent on the phenyl ring was a hydroxyl group (Table 2), establishing its identity as Tyr.

The amino acid subunits assigned in the 1H-1H DIPSI-2, 1H-1H COSY and 1H-13C HMBC spectra were linked through correlations observed in the ROESY, NOESY (Figure 3) and HMBC NMR spectra. In the HMBC spectra, a correlation between the *N*-methyl of Mdha and the carbonyl of Glu, and between the Masp-NH and the leucine carbonyl at 174.5 ppm, linked Glu6 to Mdha7 and Masp<sup>3</sup> to Leu2. Additionally, ROESY correlations were observed between Leu2-NH and both Leu1-NH and Masp3-NH. Furthermore, the Tyr-NH showed ROESY or NOESY correlations to Masp3-H3 and Adda5-NH, Adda5-NH showed correlations to Adda-H4 and Adda-H2, and Adda5-H2 showed correlations to Adda-H4, Adda-2-Me and Glu6-NH. These correlations show **1** to contain Leu-Leu-Masp-Tyr-Adda-Glu-Mdha, and the molecular formula established from LC–HRMS requires an amide linkage between Mdha7 and Leu<sup>1</sup> moieties. That this linkage is present is demonstrated by the presence of numerous product ions in the HRMS/MS spectrum that are attributable to fragments containing both Leu1 and Mdha7, such as those at *m*/*z* 169.1334, 197.1283 and 488.2745 (Table 2). The ROESY correlations observed for **1** (Figure 3), especially those between the amide protons, were consistent with those expected based on the established 3-dimensional solution structure for MC-LR [12], which is reported to be very similar to that of [D-Leu1]MC-LR (**3**) [8]. Thus, **1** has the same relative stereochemistry as **2** and **3**. This is also supported by the close similarity of the 13C NMR chemical shifts of **1** to those reported for [D-Leu1]MC-LR (**3**) in the same solvent (Table S1). The fact both **1** and **3** are biosynthesized together by the MC synthetase of *M. aeruginosa* strain CPCC-464, and that **1** was subsequently found to have similar inhibitory potency to MC-LR (**2**) against protein phosphatase 2A (PP2A) (Figure 4), both indicate that **1** has the same absolute stereochemistry as **2** and **3** and that **1** is therefore [D-Leu1]MC-LY (Figure 1).

**Table 2.** Reported product ions for MC-LA (**4**), with their exact *m*/*z* values and amino acid origins, and the corresponding product ions and their mass di fferencesrelative to the reported values for MC-LA, observed during LC–HRMS/MS analysis of MC-LA (**4**), MC-LY (**5**), and [D-Leu1]MC-LY (**1**) in the positive ionization mode*a*.


*a*Reported fragments, origins *m*/*z* 2014; Mayumi 2006; 2013; Stewart et al. 2018 [13–16]; black squares indicate amino acids contributing to each product ion; gray squares indicate neutral loss of 134.0732 Da from Adda5; Diff., difference from calculated *m*/*z* for the corresponding product ion in MC-LA; ND, not detected.

**Figure 4.** Protein phosphatase 2A (PP2A) inhibition curves for **2** and **1** fitted to a four-parameter logistic curve. The estimated IC50 values were 0.62 ng/mL (standard error 0.05) for **2**, and 0.80 ng/mL (SE 0.11) for **1** (0.62 and 0.76 nM, respectively).

The positive and negative LC–HRMS spectra of **1** were consistent with a molecular formula of C55H77O13N7 which, together with the presence of a prominent [M + Na]<sup>+</sup> adduct ion, a weak neutral loss of *m*/*z* 134.0727, and the late retention time, were consistent with a non-Arg-containing MC. This indicated that **1** contained an extra C3H6 (42.0470 Da) relative to MC-LY (**5**), and C9H10O (134.0732 Da) relative to MC-LA (**4**), which differ from **1** only in their amino acids at position 1 (**5**), and at position 1 and 4 (**4**) (Table 2). The negative ion MS/MS spectrum obtained from the FS/DIA (full scan/data independent acquisition) LC–HRMS of **1** showed a prominent product ion at *m*/*z* 128.0355, consistent with the presence of a MC containing Glu at position 6, and a neutral loss of *m*/*z* 112.0190, consistent with the presence of Masp at position 3 [17]. Careful comparison of the positive ion targeted LC–HRMS/MS spectrum of **1** with those of standards of **4** and **5** (Table 2) showed that product ions in **1** that contained amino acid-1 were consistently heavier by 42.047 Da (Leu vs. Ala) than the corresponding product ions from **5**, and 42.047 (Leu<sup>1</sup> vs Ala1), 92.026 (Tyr4 vs Ala4) or 134.073 Da (Leu1 and Tyr<sup>4</sup> vs. Ala1 and Ala4) heavier than the corresponding product ions from **4** that contained amino acid-2, amino acid-4, or both amino acid-1 and -4, respectively (Table 2, Figures S12–S18). Furthermore, the UV spectrum of **1** obtained during LC–UV analysis was identical to that of **5**, and differed from that of **2** (Figure S19), suggesting the presence of Tyr in **1** and **5** in addition to the UV-absorbing chromophores also present in **2** (i.e., Adda5 and Mdha7). The LC–MS/MS and LC–UV results are therefore entirely consistent with **1** being [D-Leu1]MC-LY.A portion of the purified **1** was used to prepare a stock solution. This was quantitated using qNMR [18] and LC with chemiluminescence nitrogen detection (CLND) [19], then accurately diluted with 1:1 MeOH–H2O to prepare a reference material (RM) (~7.7 μM). LC–UV analysis of this RM showed the relative concentration of [D-Leu1,D-Glu(OMe)6]MC-LY to be 3.1%. The putative [D-Leu1,(6*Z*)-Adda5]MC-LY was below the limit of quantitation in LC–UV, but the relative concentration was estimated be to below 0.5% using HRMS/MS. Because MCs containing (6*Z*)-Adda<sup>5</sup> or D-Glu(OMe)<sup>6</sup> do not inhibit protein phosphatases [20], the RM of **1** was used for the PP2A inhibition assay without correcting for impurities. In the PP2A assay, the IC50 for a certified RM (CRM) of MC-LR (**2**) was 0.62 nM (0.62 ng/mL), while that for the RM of **1** was 0.76 nM (0.80 ng/mL) (Figure 4). Matthiensen et al. [7] reported that MC-LR (**2**) and [D-Leu1]MC-LR (**3**) had similar toxicities to mice

when injected intraperitoneally, and that the IC50 values of **2** and **3** in a PP1 assay were 3.1 and 4.4 nM, respectively. Similarly, Park et al. [6] independently found that **2** and **3** both had the same IC50 value of 0.3 nM in their PP1 assay. Ikehara et al. [21] found that the IC50 of MC-LF(**9**), which differs from MC-LY (**5**) only by the absence of a phenolic hydroxyl group on residue-2, was 3-fold higher than that of MC-LR (**2**) (0.096 vs. 0.032 nM) in their PP2A assay. Taken together with the data presented here, these results suggest that the replacement of D-Ala with D-Leu at position 1 in the MC structure has only a minor effect on the toxicity of MCs or on their inhibitory effects on PP1 and PP2A.

Authentic **3** in a cyanobacterial bloom extract from Poplar Island, MD, USA, whose structure has been verified as [D-Leu1]MC-LR by purification and NMR analysis [11], had identical retention time and product ion spectra to the peak for **3** in CPCC-464 when analyzed by LC–HRMS/MS, thus verifying its identity as proposed by Hollingdale et al. [5]. LC–HRMS/MS also showed the presence of a minor microcystin with [M + H]<sup>+</sup> at *m*/*z* 1071.5556 (C51H79O13N10S<sup>+</sup>, Δ 1.2 ppm) consistent with [D-Leu1]MC-M(O)R (**6**) tentatively identified in bloom samples from south-western Ontario, Canada [9] and Maryland, USA [11], by untargeted LC–HRMS/MS methods and selective chemical oxidation. The identity of this compound in the culture was further verified by targeted LC–HRMS/MS, together with oxidation to its sulfone (**8**) using Oxone [22] (see Figures S20 and S21). In LC–HRMS/MS, **6** showed the neutral loss of methylsulfenic acid (63.9983 Da, CH3SOH) characteristic of methyl sulfoxides, and was slowly oxidized to its sulfone by Oxone. The [M + H]<sup>+</sup> ion for both sulfone-**8**, and sulfoxide-**6** after neutral loss of CH3SOH, showed the expected characteristic series of microcystin product ions. Further examination of the LC–HRMS/MS chromatograms revealed the presence of a peak corresponding to [D-Leu1]MC-MR (**7**) ([M + H]<sup>+</sup> *m*/*z* 1055.5608, C51H79O12N10S<sup>+</sup>, Δ 1.3 ppm) which showed the expected product ions and which was rapidly converted to **6** by oxidation with Oxone, thus verifying the structures of both **6** and **7**. Trace amounts of the corresponding sulfone (**8**) ([M + H]<sup>+</sup> *m*/*z* 1087.5491, C51H79O14N10S<sup>+</sup>, <sup>Δ</sup> <sup>−</sup>0.1 ppm) were also detected in the culture extract, something that was recently also reported by Foss et al. [11] in a cyanobacterial bloom sample, together with **6** and **7**. Microcystins **6–8** from this sample and from *M. aeruginosa* CPCC-464 showed identical retention times and mass spectral characteristics. Methionine sulfoxide analogues of MCs appear to be formed by autoxidation [22], and it appears that the same process can also lead to formation of the corresponding sulfones. The stereochemistry of **6**–**8** cannot be verified by LC–MS methods. However, because **7** is presumably biosynthesized in the culture by the same synthetase that produces **1** and **3**, and that **6** and **8** are autoxidation products of **7**, **6**–**8** can therefore be assumed to have the same stereochemistry as **1** and **3** (Figure 1).

A careful non-targeted LC–MS analysis of a field sample by Foss et al. [11] recently reported more than 20 Leu1-containing MCs in a cyanobacterial bloom, with [D-Leu1]MC-LR (**3**) as the major component, but no **1** was detected. Including the present study, **1** now appears to have been detected in samples originating from three different locations in Canada [5,9], but so far, nowhere else in the world. Geographical differences in the distribution of microcystins are being reported [23,24]. Leu1-containing MCs have been implicated in bird deaths in both Canada and the USA [6,11] and have been reported in samples from cyanobacterial blooms in Brazil and Argentina [7,25,26] as well as in lichens from Argentina, USA, China, Japan, Norway, Sweden, and Finland [25,27]. Leu1 variants may be more common and widespread than these studies indicate, as many analyses for MCs are conducted using highly targeted LC–MS/MS methods, and the Leu1-containing variants are heavier by 42 Da than the more common (and more commonly targeted) Ala1-containing MCs. Both types of variants would be readily detected if they were targeted in the LC–MS/MS method, or if untargeted LC–MS methods were used. Protein phosphatase inhibition assays, or immunoassays with appropriate cross-reactivities [28,29], can also be expected to detect both D-Ala1- and D-Leu1-containing MCs although they cannot indicate which type of variant is present.

### **3. Conclusions**

[D-Leu1]MC-LR (**3**) has been reported previously and its structure confirmed by NMR spectroscopy [8]. [D-Leu1]MC-LY (**1**) and [D-Leu1]MC-M(O)R (**6**) have been tentatively identified by LC–HRMS/MS in a Canadian cyanobacterial bloom sample and in cultures [5,9]. The results presented here firmly establish the identity of **1** and show that it has similar inhibitory potency towards PP2A as MC-LR (**2**). A calibration reference material has been prepared that can be used to identify and quantitate **1** in field samples and cultures. Microcystins containing D-Leu at position 1 may be fairly common in the Americas, and the data presented here and elsewhere suggest these to be only slightly less toxic than their more common D-Ala1-containing congeners. It is therefore important to consider the possible presence of a range of D-Leu1-containing MCs when analyzing bloom samples.

#### **4. Materials and Methods**

#### *4.1. General Experimental Procedures*

Purified **1** (250 μg) was dissolved in 30 μL of CD3OH for NMR spectroscopy. NMR spectra were acquired on a Bruker Avance III 600 MHz spectrometer (Bruker Biospin Ltd., Billerica, MA, USA) operating at a 1H frequency of 600.28 MHz and 13C frequency of 150.94 MHz using TOPSPIN 2.1 acquisition software with a 1.7 mm TXI gradient probe at 277 K. Standard Bruker pulse sequences were used for structure elucidation: one dimensional 1H spectrum with composite pulse pre-saturation of water, double quantum filtered 1H-1H COSY, 1H-1H DIPSI-2 (mixing time 120 ms), 1H-13C HSQC, 1H-13C HMBC (60 and 90 ms delay for long range coupling evolution), 1H-1H ROESY (mixing time 400 ms) and 1H-1H NOESY (mixing time 200 ms). A second sample of **1** (1 mg in 210 μL) was prepared in a 3 mm NMR tube for acquisition of the 13C spectrum on a Bruker Avance III 700 MHz spectrometer operating at 1H frequency of 700.15 MHz and a 13C frequency of 176.07 MHz equipped with a 5 mm cryogenically cooled probe at 278 K. The residual 1H resonance of CD3OH was referenced to 3.31 ppm and the carbon to 48.0 ppm. Data were processed with TOPSPIN 2.1 and analyzed using the NMR assignment software Sparky (T. D. Goddard and D. G. Kneller, SPARKY 3, University of California, San Francisco: https://www.cgl.ucsf.edu/home/sparky/).

The initial survey of cultures for the presence of MCs was performed by LC–UV–MS using an Agilent (Mississauga, ON, Canada) 1200 LC coupled with a SCIEX (Concord, ON, Canada) API 4000 Q-Trap mass spectrometer with UV monitoring at 238 nm and positive electrospray ionization MS, with full scans, *m*/*z* 135 precursor scans, product ion scans, and selected reaction monitoring. The LC column (50 × 2.1 mm; Agilent) was packed with 1.8 μm Zorbax SB-C18 and maintained at 40 ◦C. The flow rate was 0.3 mL/min, with a gradient of 10%–80% B over 30 min. Solvent A was water and B was 95% acetonitrile, each with 50 mM formic acid and 2 mM ammonium formate.

LC–UV spectra of **1**, **2** and **5** were acquired using an Agilent 1260 LC System with diode array detector (DAD) with UV monitoring at 238 and 210 nm. Separations were on an Acquity HSS T3 1.8 μm column (100 × 2.1 mm; Waters, Milford, MA, USA) held at 40 ◦C. Isocratic elution was performed with 50% MeOH-H2O (0.1% *v*/*v* trifluoroacetic acid) at 0.25 mL min−<sup>1</sup> with injection volumes of 5 μL.

LC–HRMS was conducted with a Q Exactive-HF Orbitrap mass spectrometer equipped with a HESI-II heated electrospray ionization interface (ThermoFisher Scientific, Waltham, MA, USA) with an Agilent 1200 G1312B binary pump, G1367C autosampler, and G1316B column oven. Analyses were performed with a 3.5 μm Symmetry Shield C18 column (100 × 2.1 mm; Waters) held at 40 ◦C with mobile phases A and B of H2O and CH3CN, respectively, each of which contained formic acid (0.1% *v*/*v*). A linear gradient (0.3 mL min−1) was used from 20% to 90% B over 18 min, then to 100% B over 0.1 min, followed by a hold at 100% B (2.9 min), then returned to 20% B over 0.1 min with a hold at 20% B (3.9 min) to equilibrate the column. Injection volume was typically 1–5 μL. In positive ion mode the mass spectrometer was calibrated from *m*/*z* 74–1622, the spray voltage was 3.7 kV, the capillary temperature was 350 ◦C, and the sheath and auxiliary gas flow rates were 25 and 8 units, respectively, with MS data acquired from 2 to 20 min. Mass spectral data were collected using a

combined FS/DIA method. FS data were collected from *m*/*z* 500–1400 using the 60000 resolution setting, an AGC target of 1 <sup>×</sup> 106 and a max IT of 100 ms. DIA data were collected using the 15000 resolution setting, an AGC target of 2 <sup>×</sup> <sup>10</sup>5, max IT set to 'auto' and a stepped collision energy of 30, 60 and 80 V. Precursor isolation windows were 62 *m*/*z* wide and centered at *m*/*z* 530, 590, 650, 710, 770, 830, 890, 950, 1010, 1070, 1130, 1190, 1250, 1310, and 1370. DIA chromatograms were extracted for product ions at *m*/*z* 121.1011, 121.0647, 135.0804, 135.1168, 375.1915, 389.2072, 361.1758, 213.0870, 426.2096, 440.2252, 454.2409, 412.1939, 393.2020, 379.1864, 585.3395, 599.3552, and 613.3709. Putative MCs detected using the above FS/DIA method were further probed in a targeted manner using the PRM scan mode with a 0.7 *m*/*z* precursor isolation window, typically using the 30,000 resolution setting, an AGC target of 5 <sup>×</sup> 10<sup>5</sup> and a max IT of 400 ms. Typical collision energies were: stepped CE at 30 and 35 eV for MCs with no Arg, and stepped CE at 60, 65 and 70 for MCs with one Arg. LC–MS/MS/MS spectra of *m*/*z* 1071.5 <sup>→</sup> 1007 for [D-Leu1]MC-M(O)R (**6**) were obtained from an LC–HRMS/MS chromatogram run in PRM mode at *m*/*z* 1007.5 with in-source fragmentation energy set at 100 eV. In negative mode, the mass spectrometer was calibrated from *m*/*z* 69–1780 and the spray voltage was −3.7 kV, while the capillary temperature, sheath and auxiliary gas flow rates were the same as for positive mode. Mass spectral data were collected in FS/DIA scan mode as above using a scan range of *m*/*z* 750–1400, a resolution setting of 60000, AGC target of 1 <sup>×</sup> 106 and max IT of 100 ms. DIA data were collected using a resolution setting of 15,000, AGC target of 2 <sup>×</sup> 105, max IT set to 'auto', and stepped collision energy 65 and 100 V. Isolation windows were 45 *m*/*z* wide and centered at *m*/*z* 772, 815, 858, 902, 945, 988, 1032, 1075, 1118, 1162, 1205, 1248, 1294, 1335, and 1378. DIA chromatograms were extracted for product ions at *m*/*z* 128.0353.

The Oxone oxidations of **6** and **7**, based on Miles et al. [22] were performed by adding 1 μL of Oxone (10 mg/mL) in water to 19 μL of 1:1 MeOH–H2O, then 20 μL of culture extract was added with vortex mixing. The reaction was monitored by LC–HRMS/MS after 15 min and then periodically thereafter.
