Leucinostatins from Ophiocordyceps spp. and Purpureocillium spp. Demonstrate Selective Antiproliferative Effects in Cells Representing the Luminal Androgen Receptor Subtype of Triple Negative Breast Cancer

Abstract

The structures of four leucinostatin analogs (1-4) from Ophiocordyceps spp. and Purpureocillium spp. were determined together with six known leucinostatins: [leucinostatins B (5), A (6), B2 (7), A2 (8), F (9), and D (10)]. The structures of the metabolites were established using a combination of analytical methods including HRMS and MS/MS experiments, 1D and 2D NMR spectroscopy, chiral HPLC, advanced Marfey’s analysis of the acid hydrolysate, as well as additional empirical and chemical methods. Compounds 1-10 were evaluated for their biological effects on triple negative breast cancer (TNBC) cells. Leucinostatins 1-10 showed selective cytostatic activities in MDA-MB-453 and SUM185PE cells representing the luminal androgen receptor subtype of TNBC. This selective activity motivated further investigation into the mechanism of action of leucinostatin B (5). The results demonstrate that this peptidic fungal metabolite rapidly inhibits mTORC1 signaling in leucinostatin sensitive TNBC cell lines, but not in leucinostatin resistant cells. Leucinostatins have been shown to repress mitochondrial respiration through inhibition of the ATP synthase and we demonstrated that both the mTORC1 signaling and LAR selective activities of 5 were recapitulated by oligomycin. Thus, inhibition of the ATP synthase with either leucinostatin B or oligomycin is sufficient to selectively impede mTORC1 signaling and inhibit the growth of LAR subtype cells.

It is estimated that over 275,000 new cases of breast cancer will be diagnosed in the United States in 2020 alone.¹ Approximately 15–20% of these cases will be triple negative breast cancers (TNBC), thus named because they do not express estrogen/progesterone receptors (ER/PR) or have amplification of the human epidermal growth factor receptor 2 (HER2) gene.^2–4 In recent years, disease-free survival rates for ER/PR+ and HER2+ breast cancers have steadily improved due to the development and use of targeted therapies. Unfortunately, there are currently no molecularly targeted therapies available for the treatment of TNBC,⁵ which is an aggressive form of breast cancer with an overall poorer prognosis than receptor-positive breast cancers.^2,6,7 TNBC represents a significant unmet medical need and effective targeted therapies are required to improve the survival of TNBC patients.

Genomic analyses show that TNBCs are a complex, heterogeneous group of breast cancers.^8–11 A single treatment strategy for TNBCs with a “one target − one drug” approach will not suffice. Fortunately, genomic investigations have illuminated new paths to success by identifying molecularly discrete subgroups of TNBC. In 2011, Lehmann and Bauer identified six molecularly defined TNBC subtypes based on genomic evaluations of patient tumors, and importantly, assigned corresponding cell line models for each subtype.⁹ These models were later consolidated into 4 subtypes, which included the mesenchymal, luminal androgen receptor (LAR), and two basal-like subtypes: basal-like 1 and 2.¹⁰ Other analyses of TNBC tumors confirmed the LAR subtype in diverse populations of TNBC patients.^8,12 The LAR subtype is noted as the most resistant to chemotherapy, and accounts for 10–23% of all TNBC cases.^10,12,13

A majority of the anticancer drugs used clinically are natural products or are derived from compounds found in nature.^14–16 Natural products remain an important resource for the identification of novel therapeutic leads, and provide avenues for the discovery of new molecular targets. Our team’s collaborative goal to discover TNBC-subtype-selective compounds led us to evaluate leucinostatins in cells representing the different TNBC subtypes. Leucinostatins A (6) and B (5) were the first metabolites to be described from a series of non-ribosomal-peptide-synthetase-derived natural products and have subsequently been obtained from taxonomically diverse fungal species.^17–22 In general, these metabolites are nonapeptides and contain a rather distinctive set of structural moieties including, but not limited to 2-amino-6-hydroxy-4-methyl-8-oxodecanoic acid (AHMOD), cis-4-methyl-l-proline (MePro), l-threo-β-hydroxyleucine (HyLeu), α-aminoisobutyric acid (Aib), l-leucine (Leu), β-alanine (β-Ala), as well as an (4S,2E)-4-methylhex-2-enoic acid (MeHA) at the N-terminus, and (2S)-N¹-methylpropane-1,2-diamine (MPD) or (2S)-N¹,N¹-dimethylpropane-1,2-diamine (DPD) at the C-terminus. To date, evidence supporting the existence of >20 leucinostatins has been reported;^{19–21,23–31} however, the data substantiating those claims have varied in scientific rigor ranging from total synthesis and thorough NMR-focused structure determination efforts to less exhaustive studies utilizing only MS-derived data. Additionally, controversy has emerged concerning the stereochemical properties of the leucinostatins with the C-6 configuration of the AHMOD residue still contested (Figure S1, Supporting Information).^32,33

A variety of biological activities have been ascribed to the leucinostatins, which has generated further interest in exploring their applications in human health and agriculture. The leucinostatins are noted to exhibit antimicrobial activities against several types of bacteria and fungi,^{20,21,25,27–29,34,35} as well as trypanosomes.³⁶ Leucinostatin A inhibited the growth of DU-145 human prostate cancer cells in vitro and in vivo only when co-cultured with prostate stromal cells through a proposed mechanism of reduced insulin-like growth factor signaling.³⁵ The cytotoxicity of leucinostatin A, but not leucinostatin Y, in human pancreatic cell lines under nutrient-deprived conditions has also been reported and correlates with inhibition of mitochondrial respiration.^26,35 The effects of leucinostatins against multiple subtypes of TNBC have not been reported, making them an intriguing starting point for uncovering new mechanistic insights. By way of this report, we have attempted to expand the understanding of this intriguing class of fungal metabolites through chemical analyses of leucinostatin analogs, as well as provided evidence supporting their promising activity in a discrete subset of TNBC, namely, in cells representing the LAR subtype of TNBC through inhibition of mTORC1 signaling.

RESULT AND DISCUSSION

Our team encountered leucinostatins as bioactive compounds in investigations of infectious disease targets. Those observations were instrumental in leading us to initiate an exploration of the chemodiversity within this metabolite family. Utilizing a library of fungi obtained through the University of Oklahoma, Citizen Science Soil Collection Program,³⁷ we used LC-PDA-MS data to help steer our focus toward five isolates (fungal isolate codes: OK1086 GZ-1, CA6630 RBM-4, TN2181 PBA-1, IN01344 TV8–2, and IL6653 TV8–3) that (1) appeared to contain leucinostatin-like compounds, and (2) originated from geographically disparate regions of the United States. Scale-up purification of leucinostatin metabolites from OK1086 GZ-1 (Ophiocordyceps sp.) yielded leucinostatins B (5),²⁴ A (6),²⁴ B2 (7),²⁴ A2 (8),²⁴ F (9),²⁹ and D (10),²⁸ along with trace amounts of four additional analogs (1-4) that could not be straightforwardly dereplicated or identified. Since all the fungal isolates used in this study made 1-4 in relatively low yields, the decision was made to pool fractions containing the metabolites in preparation for their final purification. As efforts to ascertain their structures progressed, it became evident to us that some of the leucinostatin analogs we obtained might be similar or identical to molecules whose structures had been proposed based solely on MS data.^30,31 Unfortunately, the absence of additional spectroscopic or chemical evidence from these reports prohibited us from decisively matching our metabolites to those reported leucinostatin analogs. Thus, we embarked on an effort to ascertain the structures of the low yield leucinostatin analogs using an orthogonal series of spectrometric, spectroscopic, and chemical methods. Given our uncertainty as to the relationships of our metabolites to the partially described leucinostatins, as well as the potential confusion resulting from the irregular naming practices used by others to identify compounds in this family (i.e., a combination of discontinuous alphanumeric codes and Roman numerals), we chose to adopt a distinctive set of trivial names [leucinostatins NPDG A-D (1-4, respectively)] for the fungal peptides described in this report.

Leucinostatin NPDG A (1) was obtained as a colorless, amorphous solid, and its molecular formula was determined to be C₆₁H₁₀₉N₁₁O₁₂ based on an [M + H]⁺ ion observed at m/z 1188.8344 by HRESIMS. Considering this assignment, as well as the observation that the ¹H and ¹³C NMR data for 1 and 5 were similar, we hypothesized that the two metabolites were structural analogs. Specifically, metabolite 1 lacked one oxygen atom compared to 5, which was traced to a change in the structure of the AHMOD residue. Whereas the H-6/C-6 resonances for the AHMOD moiety appeared at δ_H 4.46 (m)/δ_C 65.6 in 5 (Table S1, Supporting Information), metabolite 1 exhibited signals at δ_H 1.56 and 1.66 (m)/δ_C 21.79 (Table 1) indicative of a 2-amino-4-methyl-8-oxodecanoic acid (AMOD) group. This change from an AHMOD to an AMOD group was further substantiated by MS/MS experiments in which the b1 fragment ion [m/z 419 (Figure 1)] provided evidence for the missing oxygen atom (Figure S2, Supporting Information). To further validate the preliminary bond-line structure of 1, ¹H-¹H COSY, ¹H-¹H ROESY, ¹H-¹³C HSQC, and ¹H-¹³C HMBC experiments were performed (Figure 2) and the results provided strong evidence in support of the proposed structure. It was determined that 1 matched the bond-line structure presented for leucinostatin V;^30,31 however, in the absence of additional data from the preceding publications substantiating that proposal (e.g., MS/MS data only; no spectroscopic or other chemical evidence presented) we were unable to confirm if the molecules were identical or stereoisomers.

Table 1.

¹H (500 MHz) and ¹³C (100 MHz) NMR Spectroscopic Data for Compound 1 in Pyridine-d₅^a

residue	position	δC, type	δH (J, Hz)	residue	position	δC, type	δH (J, Hz)
MeHA	1	166.9, C	-	Leu1	1	176.8, C	-
	2	121.4, CH	6.39, d (15.2)		2	56.6, CH	4.51, m
	3	152.0, CH	7.15, dd (15.2, 7.8)		3	41.1, CH2	2.12, 2.44, m
	4	39.1, CH	2.36, m		4	25.5, CH	2.33, m
	5	29.8, CH2	1.44, 1.64, m		5	22.5, CH3	0.99b
	6	12.6, CH3	1.03b		6	23.5, CH3	1.12, d (7.3)
	7	19.8, CH3	1.10, d (7.2)		CONH	-	8.10, br d (5.0)
MePro	1	176.3, C	-	Leu2	1	175.7, C	-
	2	64.6, CH	4.55, m		2	57.3, CH	4.44, m
	3	39.9, CH2	1.76, 2.44, m		3	40.7, CH2	2.16, 2.26, m
	4	34.3, CH	2.13, m		4	25.4, CH	2.12, m
	5	55.2, CH2	3.71, 3.89, m		5	23.1, CH3	0.95, d (6.5)
	6	17.1, CH3	0.97b		6	23.2, CH3	1.00b
AMOD	1	177.6. C	-		CONH	-	8.81, br d (4.7)
	2	56.3, CH	4.64, m	Aib2	1	176.6, C	-
	3	37.5, CH2	1.69, 2.20, m		2	57.0, C	-
	4	30.2, CH	1.94, m		3	23.6, CH3	1.76, s
	5	37.5, CH2	1.18, 1.35, m		4	28.0, CH3	1.86, s
	6	21.79, CH2	1.56, 1.66, m		CONH	-	8.76c
	7	43.1, CH2	2.33, m	Aib3	1	178.0, C	-
	8	211.0, C	-		2	57.5, C	-
	9	36.1, CH2	2.34, m		3	23.3, CH3	1.68, s
	10	8.5, CH3	1.02b		4	28.4, CH3	1.87, s
	11	18.9, CH3	0.91, d (6.6)		CONH	-	8.17, s
	CONH	-	10.25, br d (6.8)	β-Ala	1	174.8, C	-
β-HyLeu	1	174.0, C	-		2	38.5, CH2	2.40, 3.29, m
	2	64.0, CH	4.55, m		3	38.3, CH2	3.34, 4.27, m
	3	73.9, CH	4.55, m		CONH	-	8.66, br dd (9.0, 4.3)
	4	30.9, CH	2.25, m	MPD	1	56.2, CH2	3.27, 3.77, m
	5	15.7, CH3	1.43b		2	43.2, CH	4.68, m
	6	21.75, CH3	1.44b		3	18.3, CH3	1.27, d (6.7)
	CONH	-	9.65, br s		4	34.8, CH3	3.00, s
Aib1	1	177.7, C	-		CONH	-	7.75, br d (8.6)
	2	57.2, C	-		NH-2	-	n.d.d
	3	23.6, CH3	1.79, s
	4	27.4, CH3	1.95, s
	CONH	-	9.26, s

^aAbbreviations: MeHA, 4-methyl-2-hexenoic acid; MePro, 4-methylproline; AMOD, 2-amino-4-methyl-8-oxodecanoic acid; β-HyLeu, β-hydroxyleucine; Aib, 2-aminobutyric acid; Leu, leucine; β-Ala, β-alanine; MPD, N¹-methylpropane-1,2-diamine.

^bMultiplicity was not determined due to overlapping signal.

^COverlapped with solvent signal; chemical shift was extracted from the HMBC correlation data.

^dNot detected.

Figure 1.

MSⁿ fragmentation of compounds 1-4 (observed for m/z 1189, 1203, 1191, and 1205 [M + H]⁺, respectively). The dashed lines through the structures show the “b” fragments obtained in LC-MS/MS experiment. The depicted numbers indicate the corresponding m/z values.

Figure 2.

Key COSY, ROESY, and HMBC correlations in compounds 1-4.

HRESIMS data were obtained for leucinostatin NPDG B (2) that indicated the metabolite had the molecular formula C₆₂H₁₁₁N₁₁O₁₂. Combining this information with its ¹H and ¹³C NMR data (Table 2) led to the conclusion that 2 contained an additional methyl group compared to 1. Further interpretation of the NMR data revealed that 2 possessed two isochronous N-methyl groups at δ_H 3.04 (s)/δ_C 44.8 (C-4 and C-5 of the DPD moiety) versus the single N-methyl group in the MDP residue of 1. The MS/MS fragmentation pattern of 2 provided additional substantiation for this structural change with b-fragment ions (b1 to b7, as well as b9-H₂O, Figure 1) that supported the presence of a second methyl group on the metabolite’s N-terminus. The bond-line structure for the compound was established using data from a combination of ¹H-¹H COSY, ¹H-¹H ROESY, ¹H-¹³C HSQC, and ¹H-¹³C HMBC experiments (Figure 2). Metabolite 2 was determined to match the bond-line structure reported for leucinostatin R,^30,31 but in the absence of additional supporting data from the original publications, we could not confirm if the compounds were identical or stereoisomers.

Table 2.

¹H (500 MHz) and ¹³C (125 MHz) NMR Spectroscopic Data for Compound 2 in Pyridine-d₅^a

residue	position	δC, type	δH (J, Hz)	residue	position	δC, type	δH (J, Hz)
MeHA	1	166.9, C	-	Leu1	1	176.8, C	-
	2	121.7, CH	6.39, d (15.1)		2	56.6, CH	4.50, m
	3	152.0, CH	7.15, dd (15.1, 7.8)		3	41.0, CH2	2.13, 2.43, m
	4	39.1, CH	2.35, m		4	25.5, CH	2.32, m
	5	29.8, CH2	1.44, 1.64, m		5	22.5, CH3	0.98b
	6	12.6, CH3	1.02b		6	23.49, CH3	1.12, d (6.6)
	7	19.8, CH3	1.10, d (6.8)		CONH	-	8.11, br d (5.0)
MePro	1	176.3, C	-	Leu2	1	175.7, C	-
	2	64.6, CH	4.55, m		2	57.19, CH	4.42, m
	3	39.9, CH2	1.76, 2.45, m		3	40.6, CH2	2.12, 2.25, m
	4	34.3, CH	2.14, m		4	25.4, CH	2.09, m
	5	55.2, CH2	3.71, m 3.88, d (10.0, 7.4)		5	23.0, CH3	0.94, d (6.4)
	6	17.1, CH3	0.97b		6	23.2, CH3	0.99b
AMOD	1	177.6. C	-		CONH	-	8.81, br d (4.7)
	2	56.3, CH	4.63, m	Aib2	1	176.7, C	-
	3	37.5, CH2	1.68, 2.20, m		2	57.0, C	-
	4	30.2, CH	1.92, m		3	23.54, CH3	1.74, s
	5	37.5, CH2	1.17, 1.34, m		4	27.9, CH3	1.85, s
	6	21.78, CH2	1.55, 1.66, m		CONH	-	8.77, s
	7	43.1, CH2	2.32, m	Aib3	1	178.0, C	-
	8	210.9, C	-		2	57.4, C	-
	9	36.1, CH2	2.34, m		3	23.4, CH3	1.66, s
	10	8.5, CH3	1.02b		4	28.3, CH3	1.86, s
	11	18.9, CH3	0.91, d (6.6)		CONH	-	8.19, s
	CONH	-	10.25, br d (6.8)	β-Ala	1	174.3, C	-
β-HyLeu	1	174.0, C	-		2	38.3, CH2	2.45, 3.28, m
	2	64.0, CH	4.54, m		3	38.4, CH2	3.41, 4.19, m
	3	73.9, CH	4.54, m		CONH	-	8.66, br dd (8.6, 4.2)
	4	30.9, CH	2.24, m	DPD	1	63.7, CH2	3.04, 3.76, m
	5	15.7, CH3	1.42b		2	41.3, CH	4.72, m
	6	21.73, CH3	1.43b		3	19.1, CH3	1.23, d (6.7)
	CONH	-	9.65, br d (2.6)		4, 5	44.8, CH3c	3.04, s
Aib1	1	177.8, C	-		CONH	-	7.70, br d (9.3)
	2	57.23, C	-
	3	23.54, CH3	1.78, s
	4	27.3, CH3	1.94, s
	CONH	-	9.27, s

^aAbbreviations: MeHA, 4-methyl-2-hexenoic acid; MePro, 4-methylproline; AMOD, 2-amino-4-methyl-8-oxodecanoic acid; β-HyLeu, β-hydroxyleucine; Aib, 2-aminobutyric acid; Leu, leucine; β-Ala, β-alanine; DPD, N¹,N¹-dimethylpropane-1,2-diamine.

^bMultiplicity was not determined due to overlapping signal.

^CChemical shift was extracted from the HSQC correlation data.

The molecular formula for leucinostatin NPDG C (3) was established to be C₆₁H₁₁₁N₁₁O₁₂ based on the HRESIMS data. An inspection of the ¹³C NMR data for 3 (Table 3) drew our attention to a missing ketone resonance, which is a distinctive feature of the AHMOD moiety in 5. Comparative analyses of the 1D and 2D NMR representing those portions of compounds 3 and 5 led to the observation that the AHMOD moiety had been replaced by a hydroxylated 2-amino-4-methyldecanoic acid (AHMD) residue. Analysis of the ¹H-¹³C HMBC data pointed to the alcohol being located at the C-6 position of the residue. While the 2D NMR data provided compelling support for a 2-amino-6-hydroxy-4-methyldecanoic acid in 3, the significant signal overlap in the high-field region of the NMR spectra combined with the precedence set by a previous report of a near-identical metabolite containing a 2-amino-8-hydroxy-4-methyldecanoic acid residue (leucinostatin IV)³¹ demanded that we proceed with caution for this assignment. Upon further investigation, two factors emerged strengthening our proposal that 3 contained a C-6-hydroxy versus a C-8-hydroxy in the AHMD group. First, we could find no justification offered in the previous report to substantiate the claim for a C-8 alcohol in the putative 2-amino-8-hydroxy-4-methyldecanoic acid residue of leucinostatin IV (i.e., the MS/MS data presented in the report were incapable of distinguishing between C-6 versus C-8 alcohols). Second, the ¹³C NMR data generated by our team favored a C-6 alcohol since the γ-effect³⁸ caused by a C-8 alcohol would require a substantial upfield shift of the terminal methyl group (δ_C ~10 ppm), which instead appeared at δ_C 14.9 ppm in 3 and was consistent with an alcohol that was >4 bonds away. In the absence of additional data, our concern as to the validity of the structure originally proposed for leucinostatin IV was irresolvable, but we believe it is prudent to treat this molecule’s structure as suspect. Thus, it is not certain if natural product 3 is a new compound or if it should be treated as a structure revision of leucinostatin IV.

Table 3.

¹H (500 MHz) and ¹³C (100 MHz) NMR Spectroscopic Data for Compound 3 in Pyridine-d₅^a

residue	position	δC, type	δH (J, Hz)	residue	position	δC, type	δH (J, Hz)
MeHA	1	166.9, C	-	Leu1	1	176.8, C	-
	2	121.4, CH	6.38, d (15.2)		2	56.6, CH	4.50, m
	3	152.2, CH	7.14, dd (15.2, 7.8)		3	41.1, CH2	2.12, 2.42, m
	4	39.05, CH	2.33, m		4	25.51, CH	2.31, m
	5	29.8, CH2	1.42, 1.61, m		5	22.5, CH3	0.97b
	6	12.6, CH3	1.00b		6	23.5, CH3	1.11, d (6.4)
	7	19.8, CH3	1.08, d (6.7)		CONH	-	8.13, br d (4.9)
MePro	1	176.2, C	-	Leu2	1	175.7, C	-
	2	64.7, CH	4.52, m		2	57.2, CH	4.44, m
	3	39.9, CH2	1.74, 2.42, m		3	40.7, CH2	2.23, m
	4	34.3, CH	2.12, m		4	25.46, CH	2.11, m
	5	55.2, CH2	3.68, 3.88, m		5	23.1, CH3	1.00b
	6	17.2, CH3	0.93b		6	23.25, CH3	0.95b
AHMD	1	177.6. C	-		CONH	-	8.76c
	2	56.4, CH	4.70, m	Aib2	1	176.6, C	-
	3	36.5, CH2	1.62, 2.40, m		2	57.0, C	-
	4	27.6, CH	2.30, m		3	23.7, CH3	1.75, s
	5	46.5, CH2	1.40, 1.60, m		4	27.9, CH3	1.85, s
	6	68.1, CH	3.85, m		CONH	-	8.70, s
	7	39.07, CH2	1.54, 1.61, m	Aib3	1	178.0, C	-
	8	29.0, CH2	1.50, 1.69, m		2	57.5, C	-
	9	23.8, CH2	1.35, m		3	23.29, CH3	1.68, s
	10	14.9, CH3	0.91b		4	28.3, CH3	1.86, s
	11	20.7, CH3	1.05, d (6.6)		CONH	-	8.16, s
	CONH	-	10.09, br d (6.5)	β-Ala	1	174.8, C	-
β-HyLeu	1	174.0, C	-		2	38.5, CH2	2.39, 3.27, m
	2	63.8, CH	4.59, dd (9.0, 4.6)		3	38.3, CH2	3.33, 4.26, m
	3	74.1, CH	4.51, m		CONH	-	8.63, br dd (9.1, 4.2)
	4	30.9, CH	2.24, m	MPD	1	56.2, CH2	3.27, 3.73, m
	5	15.8, CH3	1.41b		2	43.2, CH	4.68, m
	6	21.7, CH3	1.42b		3	18.3, CH3	1.26b
	CONH	-	9.47, br d (4.6)		4	34.8, CH3	2.99, s
Aib1	1	177.8, C	-		CONH	-	7.77, br d (8.0)
	2	57.3, C	-		NH-2	-	n.d.d
	3	23.6, CH3	1.79, s
	4	27.3, CH3	1.94, s
	CONH	-	9.19, s

^aAbbreviations: MeHA, 4-methyl-2-hexenoic acid; MePro, 4-methylproline; AHMD, 2-amino-6-hydroxy-4-methyldecanoic acid; β-HyLeu, β-hydroxyleucine; Aib, 2-aminobutyric acid; Leu, leucine; β-Ala, β-alanine; MPD, N¹-methylpropane-1,2-diamine.

^bMultiplicity was not determined due to overlapping signal.

^cOverlapped with solvent signal; chemical shift was extracted from the COSY and HMBC correlation data.

^dNot detected.

HRESIMS data obtained for leucinostatin NPDG D (4) revealed that this metabolite possessed the molecular formula C₆₂H₁₁₃N₁₁O₁₂. When combined with the ¹H and ¹³C NMR data for 4 (Table 4), it was evident that this compound was similar to metabolite 3, but it contained an additional methyl group. Further investigation showed that 4 possessed two isochronous N-methyl groups at δ_H 2.68 (br s)/δ_C 45.3 representing CH₃-4 and CH₃-5 of a DPD residue. The structure of the remaining portion of the compound was confirmed by ¹H-¹H COSY, ¹H-¹H ROESY, ¹H-¹³C HSQC, and ¹H-¹³C HMBC experiments resulting in the bond-line structure shown for 4. Consistent with the concerns raised about the structure proposed for leucinostatin IV (vide supra), we reasoned that 4 might represent the correct structure for what had been published as leucinostatin S.^30,31 However, in the absence of additional supporting data from the original publications, we can only speculate as to the relationship of our metabolite to the reported compound, and conjecture as to the veracity of the previous structural proposal.

Table 4.

¹H (500 MHz) and ¹³C (125 MHz) NMR Spectroscopic Data for Compound 4 in Pyridine-d₅^a,b

residue	position	δC, type	δH (J, Hz)	residue	position	δC, type	δH (J, Hz)
MeHA	1	167.0, C	-	Leu1	1	176.4, C	-
	2	121.2, CH	6.34, d (15.2)		2	56.4, CH	4.52, m
	3	152.4, CH	7.13, dd (15.2, 7.8)		3	41.0, CH2	2.10, 2.41, m
	4	39.0, CH	2.30, m		4	25.7, CH	2.30, m
	5	29.7, CH2	1.39, 1.58, m		5	22.3, CH3	0.98c
	6	12.5, CH3	0.98c		6	23.55, CH3	1.11, d (6.5)
	7	19.7, CH3	1.07c		CONH	-	8.10, br d (5.0)
MePro	1	176.0, C	-	Leu2	1	175.7, C	-
	2	64.7, CH	4.54, m		2	56.8, CH	4.48, m
	3	39.7, CH2	1.75, 2.44, m		3	40.6, CH2	2.18, 2.26, m
	4	34.2, CH	2.13, m		4	25.5, CH	2.12, m
	5	55.1, CH2	3.65, 3.88, m		5	23.0, CH3	0.93c
	6	17.1, CH3	0.95c		6	23.2, CH3	1.01, d (6.4)
AHMD	1	n.d.d	-		CONH	-	8.66, br d (4.8)
	2	56.5, CH	4.67, m	Aib2	1	176.1, C	-
	3	36.6, CH2	1.67, 2.36, m		2	57.3, C	-
	4	27.9, CH	2.28, m		3	24.01, CH3	1.78c
	5	46.0, CH2	1.40, 1.60, m		4	27.7, CH3	1.86, s
	6	68.1, CH	3.86, m		CONH	-	8.67, s
	7	39.1, CH2	1.54, 1.61, m	Aib3	1	177.4, C	-
	8	29.0, CH2	1.49, 1.68, m		2	57.6, C	-
	9	23.7, CH2	1.37, m		3	24.03, CH3	1.78, s
	10	14.8, CH3	0.92c		4	28.0, CH3	1.89, s
	11	20.7, CH3	1.05c		CONH	-	8.02, s
	CONH	-	10.03, br d (6.1)	β-Ala	1	n.d.d	-
β-HyLeu	1	174.0, C	-		2	38.1, CH2	2.61, 3.11, m
	2	63.4, CH	4.60, dd (8.2, 5.0)		3	38.0, CH2	3.63, 4.08, m
	3	74.4, CH	4.46, m		CONH	-	8.46, m
	4	30.8, CH	2.23, m	DPD	1	64.5, CH2	3.10, 4.07, m
	5	16.3, CH3	1.40, d (6.6)		2	42.5, CH	4.57, m
	6	21.5, CH3	1.37, d (6.8)		3	19.4, CH3	1.26, d (6.7)
	CONH	-	9.32, br s		4, 5	45.3, CH3	2.68, br s
	OH-3	-	6.11, br s		CONH	-	7.87, br d (8.6)
Aib1	1	177.7, C	-
	2	57.4, C	-
	3	23.63, CH3	1.80, s
	4	27.4, CH3	1.93, s
	CONH	-	9.06, s

^aAbbreviations: MeHA, 4-methyl-2-hexenoic acid; MePro, 4-methylproline; AHMD, 2-amino-6-hydroxy-4-methyldecanoic acid; β-HyLeu, β-hydroxyleucine; Aib, 2-aminobutyric acid; Leu, leucine; β-Ala, β-alanine; DPD, N¹,N¹-dimethylpropane-1,2-diamine.

^bChemical shifts were extracted from the COSY, ROESY, HSQC, and HMBC correlation data.

^cMultiplicity was not determined due to overlapping signal.

^dNot detected.

With the bond-line structures of 1-4 secured, we proceeded to address the absolute configurations of the metabolites. To more securely link the results of our analyses of 1-4 with other known leucinostatins, most experiments were also carried out in parallel on metabolites 5 and 6 and those data are provided in the Supporting Information. The advanced Marfey’s method^39–43 was used to analyze the acid hydrolysates (6 N HCl, 110 °C, 16 h) of 1-6. For these experiments, combinations of the l-FDLA (1-fluoro-2,4-dinitrophenyl-5-leucinamide) and dl-FDLA (a 1:1 mixture of d- and l-FDLA) derivatives were prepared as an efficient means to secure the retention-time characteristics of several residues for which authentic standards were not readily available.³⁹ All Leu residues in 1-6 were confirmed as l-configured based on comparisons with authentic standards (Figures S4–S6, Supporting information).^39,44 The β-HyLeu residues derived from 1-6 were analyzed next via comparisons with (2S,3R)-β-HyLeu, as well as (2R/S,3R)-β-HyLeu (Figures S7–S9, Supporting Information).^43,45 Those experiments established the absolute configurations of the β-HyLeu residues in 1-6 as 2S,3R, which was in agreement with expectation based on prior assignments made for leucinostatins B (5) and A (6).²⁵ Next, the configuration of the MePro residues in 1-6 were secured as cis-l-MePro based on a combination of data derived from Marfey’s analyses (Figures S10 and S11, Supporting Information) and ROE correlations (H-2/H-4 of MePro), as had been previously used to establish the absolute configuration of leucinostatin A (6).²⁶

Turning to an examination of the MPD and DPD residues, we noted that few options seemingly existed for the configurational analyses of these residues beyond direct comparisons with synthetic chiral standards. However, the detection of putative FDLA derivatives of MPD and DPD in the LC-MS traces of material derived from the Marfey’s reaction products inspired us to investigate if this method could be systematically applied to the configurational assessment of the MPD and DPD residues in 1-6. While MPD and DPD are not amino acids, which are the targets for most applications of Marfey’s method,^39,44,46 we were aware of prior efforts to use this strategy on amine-containing compounds.⁴⁶ Authentic samples of (2S)- and (2R/S)-DPD were derivatized with both L-FDLA and DL-FDLA and the reaction products analyzed by LC-MS, which provided the dataset needed for further comparisons to the expected amine-containing compounds (m/z 397 [M + H]⁺) (Figure S12, Supporting Information). The Marfey’s reaction products prepared from the acid hydrolysates of 2, 4, and 6 were likewise tested yielding analytes that corresponded with the standard prepared from (2S)-DPD (Figure S13, Supporting Information). Moving on to MPD-containing compounds 1, 3, and 5, we expected two potential products consisting of the mono- and di-FDLA derivatives [we rationalized that the primary amine of (S)-N¹-methylpropane-1,2-diamine would be more reactive, thus negating consideration of the alternative mono-FDLA products that would result from reactions exclusively or preferentially with the secondary amine]. A prominent peak was observed in the selected ion trace [m/z 383 [M + H]⁺] of the LC-MS chromatogram made from authentic 5 that corresponded to the mono-FDLA product of MDP and matched the retention times and masses for the reaction products made from the hydrolysates of 1 and 3 (Figure S14, Supporting Information). Taking these observations into consideration and bearing in mind their shared biogenic origins, the MPD and DPD residues in 1-6 were proposed to be 2S-configured.

Next, we endeavored to address the absolute configurations of the stereogenic carbon atoms in the AMOD (i.e., compounds 1 and 2) and AHMD (i.e., compounds 3 and 4) residues of the leucinostatins. Beginning with the α-carbon, we utilized an empirical rule set demonstrating that l-FDLA derivatives of amino acids consistently elute from reverse-phase sorbents before their corresponding d-FDLA derivatives.^39–41,43 Out of an abundance of caution, we tested this method on l-2-aminodecanoic acid and confirmed the expected elution order (Figure S15, Supporting Information). Whereas the AMOD residues were found without difficulty from 1 and 2 and their elution characteristics were indicative of possessing l configurations (Figure S16, Supporting Information), we were unsuccessful in obtaining incontrovertible evidence supporting the formation of FDLA-derivatives of AHMD made from 3 and 4 (we suspect that dehydration of the C-6 alcohol occurred leading to the formation of multiple alkene-containing products) (Figure S17, Supporting Information). In line with expectations, both compounds 5 and 6 yielded identical cyclization products from hydrolysis, 4-methyl-6-(2-oxobutyl)-2-piperidinecarboxylic acid (MOPA),^19,23,24,26 which were determined to have l configurations based on Marfey’s analysis (Figure S18, Supporting Information). Further evidence in support of the L configurations of these residues came from an examination of the ¹H and ¹³C NMR chemical shifts for C-2 and C-3, which were found to be extremely similar to data for the AMOD residues of 1-2, AHMD residues of 3-4, and AHMOD residues of 5-6 (Figure S19, Supporting Information). Considering the cooccurrences of 3 and 4 in fungi that that also yielded l-configured α-carbons in the AMOD residues of 1-2, the AHMOD residues in 5-6, and the AMODE residues in 7 and 8, we reasonably speculate that the AHMD residues in 3 and 4 are likely to also have l configurations.

The configurations of the C-6 hydroxy groups in the AHMD residues of 3 and 4 proved problematic to probe in a direct manner due to their chemical instability under hydrolysis conditions (vide supra). Furthermore, attempts to carry out Mosher ester analyses both on intact 3 and 4, as well as their hydrolysis products proved unsuccessful. However, we did find that natural products 5 and 6 were readily converted to their Mosher ester products providing strong evidence supporting 6S configurations for these metabolites (Figure 3). The record of controversy surrounding the absolute configuration of C-6 in the AHMOD residue of leucinostatins^{32,33,47–49} was noted by our team, but considering the unequivocal results from our Mosher-ester-formation experiments, 6S configurations in 5 and 6 were determined with reasonable surety. By analogy to the results obtained for 5 and 6, we speculate based on biogenic considerations that the AHMD residues in 3 and 4 bear 6S configurations as well, but this assumption should be further probed in the future by alternative direct methods.

Figure 3.

Δδ (δ_S − δ_R) values (ppm) for (S)- and (R)-MTPA esters of compound 6.

Focusing on the absolute configuration of C-4 in the AMOD and AHMD moieties of compounds 1-2 and 3-4, respectively, we found ourselves compelled to utilize a strategy based on the likely conservation of biosynthetic outcomes between known natural products 5-6 and metabolites 1-4. Looking at the Mosher ester data generated for 5 and 6, we noted that the diastereotopic protons on C-5 exhibited distinctive changes in their chemical shifts as compared to those of the underivatized natural products. (Figure S20, Supporting Information). We noted that Takada and colleagues had fortuitously described the same phenomenon upon comparing the chemical shifts for the Mosher ester derivatives of natural products containing 1-hydroxy-3-methyl moieties.⁵⁰ Specifically, it was reported that in cases involving S*,R* 1-hydroxy-3-methyl moieties, the chemical shifts of the methylene protons (i.e., H-2a and H-2b) for the Mosher ester derivatives converged (Δδ_Ha-Hb ~0.1 ppm), but in the corresponding systems involving S*,S* isomers, the resonances attributed to those protons diverged (Δδ_Ha-Hb >0.4 ppm). Applying this rule to the 1-hydroxy-3-methyl moieties in the AHMOD residues in 5 and 6, we observed the near isosynchronicity of H-5a/H-5b (Δδ_Ha-Hb <0.1 ppm) (Figure S20B and S20C, Supporting Information). Bearing in mind that the Mosher ester analysis supported 6S configurations in 5 and 6, as well as the fact that the presence of a C-8 ketone “flipped” the priorities for C-6 stereocenter relative to the 1-hydroxy-3-methyl moieties examples cited by Takada,⁵⁰ we proposed that the AHMOD residues had 4S absolute configurations. This conclusion was in agreement with previous reports.^19,23 By extension of biosynthetic logic, as well as results from comparative analyses of the ¹H and ¹³C NMR data (Figure S19, Supporting Information), we speculated that the C-4 stereocenter in the AMOD (i.e., 1 and 2) and AHMD (i.e., 3 and 4) residues likely share the same 4S configurations.

The absolute configurations of the 4-methyl-2-hexanoic acid (MeHA) residues in the metabolites were also investigated. The olefinic bonds in 1-6 were determined to possess E-geometries because they all exhibited large coupling constants ranging from ³J_H-H = 15.1–15.2 Hz (Tables 1–4 and Tables S1 and S2, Supporting Information). Since the absolute configuration of the MeHA residue in 6 has been reported and was uncontested,^23,24,32 we subjected the metabolite to acid hydrolysis to generate an authentic reference standard of (S,E)-4-methylhex-2-enoic acid (S,E-11). In parallel, racemic 11 was prepared via Knoevenagel condensation of 2-methylbutanal and malonic acid.^51,52 Both sets of products were subsequently treated with α-bromo-2-acetonaphthone to generate their corresponding 2-naphthacyl ester derivatives, which were readily detectable by LC-PDA-MS. Similarly, the hydrolysates prepared from compounds 1, 3, 5, 7, and 9, as well as 6 were treated with α-bromo-2-acetonaphthone and the products analyzed by C₁₈ LC-PDA-MS and chiral HPLC. The derivatized hydrolysis products from the leucinostatins matched the elution characteristics of the 2-naphthacyl ester of S,E-11 (i.e., S,E-12) (Figure 4 and Figure S21, Supporting Information). This confirmed a 4S absolute configuration for the MeHA residue, which agreed with previous reports for leucinostatins.^23,24,32

Figure 4.

(A) Preparation of 2-naphthacyl esters of 4-methylhex-2-enoic acid (MeHA) (B) Chiral HPLC analysis (Cellulose-3; CH₃CN/H₂O containing 0.1% HCOOH, 11:9) of (B1) compounds 12, (B2) 12a, (B3) 12b, (B4) the corresponding product from leucinostatin A (6), and (B5) co-injection of 12 and the corresponding product.

The leucinostatins were evaluated for antiproliferative and cytotoxic effects in a panel of molecularly defined TNBC cell lines. The results showed that 1-10 have selective antiproliferative activities at concentrations of 1–100 nM in MDA-MB-453 and SUM185PE cells that represent the LAR subtype of TNBC (Figure 5). In contrast, the leucinostatins have little to no antiproliferative effects at concentrations up to 100 nM in HCC1937, HCC1806, BT-549 and MDA-MB-231 cell lines representing other subtypes of TNBC (Figure 5). The GI₅₀ values of 1-10 in the sensitive LAR subtype cell lines ranged from 1.4 – 8.0 nM, while the GI₅₀ values for all compounds were >100 nM in the resistant MDA-MB-231 cells (Table 5). Growth inhibitory activities were observed at higher concentrations in each of the TNBC cell lines with an average 92-fold selectivity for antiproliferative potency for MDA-MB-453 cells as compared to four non-LAR subtype TNBC cell lines. Leucinostatins 1-10 were cytotoxic to all cell lines at concentrations greater than 1 μM (Supporting Information, Figure S23). The selective sensitivity of the LAR-subtype cells to the cytostatic actions of the leucinostatins in the low nanomolar range was consistent with the activity of rapamycin, a known mTORC1 inhibitor.⁵³ These data prompted us to further probe the mechanism of action of leucinostatin B (5).

Figure 5.

Concentration-response curves for compounds 1-10 in a panel of six triple negative breast cancer cell lines. n = 3 ± SEM.

Table 5.

GI₅₀ Values for Compounds 1–10.^a

compound	cell line
	MDA-MB-453	SUM-185PE	MDA-MB-231
1	5.4 ± 0.9	3.9 ± 0.8	>100
2	3.5 ± 0.4	2.1 ± 0.8	>100
3	3.3 ± 2.2	2.1 ± 0.5	>100
4	3.7 ± 0.3	2.8 ± 0.6	>100
5	3.6 ± 0.5	8.0 ± 5.1	>100
6	2.5 ± 0.6	1.5 ± 0.1	>100
7	4.0 ± 0.7	2.2 ± 0.7	>100
8	7.9 ± 1.4	4.4 ± 0.7	>100
9	2.6 ± 0.6	2.0 ± 0.2	>100
10	2.3 ± 0.3	1.4 ± 0.2	>100
Rapamycin	0.05 ± 0.04	0.04 ± 0.01	>100

^aConcentrations (nM) that inhibit growth by 50%, n = 3 ± SD.

Our previous studies with deguelin showed that LAR-subtype cells are sensitive to compounds that inhibit the mTORC1 pathway.⁵³ Thus, the effects of 5 on canonical mTORC1 signaling were evaluated in sensitive, MDA-MB-453 cells, and the resistant, MDA-MB-231 cells. The effects of 5 were evaluated at 2 and 6 h to identify initial, early effects and at 18 h for later downstream effects. A 100 nM concentration was used because it has maximal cytostatic effects in MDA-MB-453 cells, but did not affect the growth of the resistant MDA-MB-231 cells. The effects of 5 on mTOR and downstream mTORC1 signaling, as measured by the phosphorylation (activation) status of (i) mTOR, (ii) ribosomal protein S6 kinase (S6K), (iii) ribosomal protein S6 (RPS6), and (iv) eukaryotic translation initiation factor 4E binding protein (4E-BP1) were evaluated (Figure 6). The results showed that 5 completely inhibited S6K phosphorylation within 2 h in the sensitive MDA-MB-453 cell line. The phosphorylation of RPS6 was also attenuated at 2 h with complete loss by 6 h and activation of 4E-BP1 was also reduced 2 h after treatment. Inhibition of mTORC1 can lead to rebound phosphorylation of AKT at serine 473 and our results showed this occurred within 18 h after the addition of 5 in the MDA-MB-453 cells. These effects were not observed in the resistant MDA-MB-231 cells treated with 5 (Figure 6) and this response is different from that obtained with the positive control rapamycin, which inhibited mTOR signaling in both cell lines (Figure 7B). These data suggested that while both rapamycin and 5 inhibit mTOR signaling and selectively inhibit the growth of LAR subtype TNBC cells (Figure 5, Figure 7A), they do so through distinct mechanisms.

Figure 6.

The effects of compound 5 on the mTORC1 signaling pathway in sensitive and resistant TNBC cells. The sensitive, MDA-MB-453 or the resistant, MDA-MB-231 cells were treated for 2, 6, or 18 h with vehicle (DMSO) or 100 nM of 5. Cell lysates were made and probed with the antibodies indicated. Data are representative of two independent experiments.

Figure 7.

Effects of rapamycin and oligomycin on the inhibition of proliferation and mTOR-dependent RPS6 phosphorylation of LAR subtype cells as compared to other TNBC cell lines (A) Concentration-response curves of rapamycin in LAR subtype MDA-MB-453 and SUM185PE cells as compared to MDA-MB-231 cells. n = 3 ± SEM. (B) Immunoblotting of whole cell lysates from MDA-MB-453 and MDA-MB-231 cells treated for 2, 6, or 18 h with vehicle (DMSO) or 10 nM rapamycin. Representative of two independent experiments. (C) Concentration-response curves of oligomycin in LAR subtype MDA-MB-453 and SUM185PE cells lines as compared to MDA-MB-231, HCC1806 and HCC70 cells. n = 3 ± SEM. (D) Immunoblotting of whole cell lysates from MDA-MB-453 and MDA-MB-231 cells treated for 2, 6, or 18 h with vehicle (DMSO) or 100 nM oligomycin. Representative of two independent experiments.

The leucinostatins have been reported to inhibit the mitochondrial F₁F₀ ATP synthase and uncouple oxidative phosphorylation in a similar manner to oligomycin, particularly at low concentrations.^26,54 We therefore evaluated whether the oligomycin-like activity of 5 might be responsible for the LAR-subtype selective activities of the leucinostatins. Indeed, we found that oligomycin selectively inhibited the growth of LAR cells as compared to other TNBC cell lines (Figure 7C). Additionally, oligomycin selectively inhibited mTOR activity in the sensitive MDA-MB-453 cells, but not in the resistant MDA-MB-231 cells, as evidenced by inhibition of RPS6 and S6K phosphorylation (Figure 7D, Figure S22, Supporting Information). These results demonstrated that the oligomycin-like uncoupling of oxidative phosphorylation by 5 was sufficient to selectively inhibit mTOR signaling in TNBC cells of the LAR subtype. Considered as a whole, these results suggest that the selective antiproliferative effects of compound 5 against TNBC cells exhibiting mTOR-dependent functions is a significant feature associated with the bioactivity traits of leucinostatins, and that further studies to probe the precise molecular mechanisms of these metabolites may afford new avenues for the targeted control the LAR-subtype.⁹

EXPERIMENTAL SECTION

General Experimental Procedures.

Data for specific rotation measurements were obtained on a Rudolph Research Autopol III automatic polarimeter. Electronic circular dichroism (ECD) measurements were performed using a JASCO J-715 spectropolarimeter. The 1D and 2D NMR data were recorded on Varian Unity Inova 400, 500, and 600 MHz FT-NMR instruments, and the data were processed using MestReNova 10.0 software (Mestrelab Research SL, Santiago de Compostela, Spain). Accurate mass spectrometry data were obtained on a Waters SYNAPT G2-Si Q-TOF/MS system coupled to a Waters M-Class chromatography system. The MS/MS experiments were performed on a Thermo LTQ XL mass spectrometer coupled to a Thermo Vanquish UHPLC System with a Thermo Accucore Vanquish C₁₈+ UHPLC -column (100 × 2.1 mm, 1.5 μm, 0.4 mL/min). The LC-PDA-MS data were obtained on a Shimadzu LC-MS 2020 system (ESI quadrupole) coupled to a photodiode array detector, with a Phenomenex Kintex 2.6 μm C₁₈ column (100 Å, 75 × 3.0 mm, 0.4 mL/min). Silica gel (60 Å, 40–60 μm, VWR International, Radnor, PA, USA) and HP-20ss (Sorbent Technologies, Norcross, GA, USA) were used for flash column chromatography. HPLC separations were performed on a Shimadzu system using a SCL-10A VP pump and system controller with a Phenomenex Gemini 5 μm C₁₈ column (110 Å, 250 × 21.2 mm, 10 mL/min) for preparative HPLC, and on a Waters system using a Waters 1525 binary pump and Waters 2998 photodiode array detector with Phenomenex Gemini 5 μm C₁₈ (110 Å, 250 × 10.0 mm, 4 mL/min) and Phenomenex Kinetex 5 μm biphenyl (100 Å, 250 × 10.0 mm, 4 mL/min) columns for semipreparative HPLC. Chiral HPLC analyses were conducted using a Phenomenex Lux 5 μm Cellulose-3 column (250 × 4.6 mm, 1 mL/min) on a Waters system with a Waters 1525 binary pump and Waters 2998 photodiode array detector. All solvents were of ACS grade or better.

Procurement of Fungal Isolates and Identification.

Two fungal isolates (internal strain designation codes: OK1086 GZ-1 and CA6630 RBM-4) were obtained from soil samples submitted to the University of Oklahoma Citizen Science Soil Collection Program (sample numbers: 11086 and 16630, respectively). Both isolates were identified as probable Ophiocordyceps spp. (most closely associated to Ophiocordyceps heteropoda) based on comparative BLAST analyses of their ribosomal internal transcribed spacer (ITS) sequence data (GenBank accession number MN097074 and MN097106, respectively) with data contained in the NCBI database (Tables S3 and S4, Supporting Information). Three other fungal isolates (internal strain designation codes: TN2181 PBA-1, IN01344 TV8–2, and IL6653 TV8–3) were also obtained from soil samples (sample numbers: 12181, 101344 and 16653, respectively) and identified as probable Purpureocillium spp. by the same procedure (GenBank accession number for ITS sequence data: MN097128, MN097129, and MN104660, respectively, Tables S5–S7, Supporting Information).

Extraction and Isolation.

The five fungal isolates were grown for four weeks on Cheerios® breakfast cereal in three large mycobags (Unicorn Bags, Plano, TX, USA) prior to being homogenized and extracted with EtOAc (×3). Details of the culture conditions and extraction methods have been previously described by our team.⁵⁵ The EtOAc extracts were fractionated by silica gel vacuum column chromatography and eluted using a step gradient of hexanes, hexanes/DCM (1:1), DCM, DCM/MeOH (10:1), and MeOH. The MeOH fractions were subjected to HP-20ss flash column chromatography using MeOH/H₂O mixtures (1:1 and then 1:0) as the eluents and washed with DCM/MeOH (1:1). The MeOH/H₂O (1:1) fractions containing minor leucinostatin analogs were combined and subjected to gradient elution by C₁₈ preparative HPLC (CH₃CN/H₂O containing 0.1% HCOOH, 2:3 → 1:0) followed by isocratic C₁₈ semipreparative HPLC eluted with CH₃CN/H₂O containing 0.1% CF₃COOH (9:11, 19:21, or 2:3), to afford 2 (5 mg), 3 (8 mg), 4 (2 mg), 5 (18 mg), 6 (9 mg), 7 (2 mg), 8 (2 mg), and 9 (4 mg) together with a mixture of 1 and 10. Compounds 1 (8 mg) and 10 (5 mg) were obtained in pure form from semipreparative HPLC using a biphenyl column (CH₃CN/H₂O containing 0.1% CF₃COOH, 21:19). Notably, compounds 7 and 8 were consistently observed in samples containing metabolites 5 and 6, respectively. We speculated that acid-catalyzed dehydration of 5 and 6 led to their formation. Experiments in which 5 and 6 were incubated in HPLC solvents containing 0.1% CF₃COOH under conditions used for rotary evaporation showed that analogs 7 and 8, respectively, could be generated in short periods under these rather mild conditions. Thus, compound 7 and 8 should be regarded with caution as possible isolation artifacts.

Leucinostatin NPDG A (1):

Colorless, amorphous solid; [α]²²_D −8 (c 0.1, MeOH); ECD (MeOH) λ_max (Δε) 207 (−13.7) nm; ¹H and ¹³C NMR data, Table 1; HRESIMS m/z 1188.8344 [M + H]⁺ (calcd for C₆₁H₁₁₀N₁₁O₁₂, 1188.8330).

Leucinostatin NPDG B (2):

Colorless, amorphous solid; [α]²²_D −10 (c 0.1, MeOH); ECD (MeOH) λ_max (Δε) 208 (−13.8) nm; ¹H and ¹³C NMR data, Table 2; HRESIMS m/z 1202.8477 [M + H]⁺ (calcd for C₆₂H₁₁₂N₁₁O₁₂, 1202.8486).

Leucinostatin NPDG C (3):

Colorless, amorphous solid; [α]²²_D −12 (c 0.1, MeOH); ECD (MeOH) 208 (−12.9) nm; ¹H and ¹³C NMR data, Table 3; HRESIMS m/z 1190.8458 [M + H]⁺ (calcd for C₆₁H₁₁₂N₁₁O₁₂, 1190.8486).

Leucinostatin NPDG D (4):

Colorless, amorphous solid; [α]²²_D −8 (c 0.1, MeOH); ECD (MeOH) λ_max (Δε) 208 (−14.5) nm; ¹H and ¹³C NMR data, Table 4; HRESIMS m/z 1204.8610 [M + H]⁺ (calcd for C₆₂H₁₁₄N₁₁O₁₂, 1204.8643).

Advanced Marfey’s Analysis.

l- and d-FDLA (1-fluoro-2,4-dinitrophenyl-5-leucinamide) were purchased from TCI chemicals (Tokyo, Japan). dl-FDLA was prepared as a 1:1 mixture of l- and d-FDLA. Approximately 0.3 mg of compounds 1-10 were hydrolyzed using 6 N HCl (500 μL) at 110 °C for 16 h, and the resultant solutions were neutralized by addition of NaHCO₃. Aliquots (50 μL) of each sample were treated with 1N NaHCO₃ (20 μL) and 1% w/v acetone solution of l- or dl-FDLA (20 μL), and the reaction mixtures were stirred for 1 h at 40 °C. The resulting mixtures were cooled to room temperature and neutralized with 1N HCl (20 μL). The resulting l- and dl-FDLA derivatives were diluted with CH₃CN and subjected to LC-PDA-MS analysis over a C₁₈ column (75 × 3.0 mm) under gradient elution conditions [CH₃CN/H₂O containing 0.1% HCOOH (1:9 → 3:2) for 65 min at a flow rate of 0.4 mL/min]. An authentic sample of (2S,3R)-β-HyLeu was purchased (Santa Cruz Biotechnology, Inc., Dallas, TX, USA), while the diastereomer, (2RS,3R)-β-HyLeu, was prepared using a reported method.⁴⁵ Briefly, (2S,3R)-β-HyLeu (5 mg) was dissolved in acetic acid (98%, 500 μL) by addition of salicylaldehyde (2 μL), and then the reaction mixture was stirred at 100 °C for 1h, after which the remaining solvent was removed in vacuo. Authentic samples of (2RS)-DPD (Matrix Scientific, Columbia, SC, USA) and (2S)-DPD (Enamine, Monmouth Junction, NJ, USA) were purchased. The retention times (min) of the FDLA derivatives of the authentic samples were determined to be as follows: l-Leu (40.5), D-Leu, (48.3), (2S,3R)-β-HyLeu (32.0), (2S,3S)-β-HyLeu (32.5), (2R,3R)-β-HyLeu (39.2), (2R,3S)-β-HyLeu (42.1), (2S)-DPD (18.5), and (2R)-DPD (25.1). The retention times (min) of FDLA derivatives corresponding to MePro and AHMD were observed as follows: cis-l-MePro (35.4), cis-d-MePro (40.0), (2S)-AMOD (43.8), and (2R)-AMOD (50.6). An authentic sample of l-2-aminodecanoic acid was purchased (AK Scientific, Inc., Union City, CA, USA) and its l- and d-FDLA derivatives were found to elute at 56.3 and 64.6 min, respectively.

Mosher’s Ester Analysis of Leucinostatin A (6).

For the modified Mosher’s ester reaction, compound 6 (1 mg) was dissolved in pyridine-d₅ and split equally among two oven-dried NMR tubes together with 2 μL of (R)- or (S)-MTPA-Cl and catalytic amounts of DMAP to afford the (S)- or (R)-MTPA-ester products. The sealed NMR tubes were incubated at 40 °C for 4 h with intermittent monitoring by ¹H NMR. The COSY correlation data were used to assist in the assignments of the ¹H NMR signals and the molecular weights of the products were later confirmed by LC-MS.

Synthesis of 2-(Naphthalen-2-yl)-2-oxoethyl (E)-4-Methylhex-2-enoate (12) and Chiral HPLC Separation of 12a and 12b.

(E)-4-Methylhexenoic acid (11) was prepared according to a published protocol.⁵² Esterification of compound 11 (20 mg) was carried out at room temperature in 3 h with the addition of α-bromo-2-acetonaphthone (42 mg), Et₃N (23 μL), and EtOAc (3 mL),⁵⁶ and the resulting products purified by C₁₈ preparative HPLC (CH₃CN/H₂O containing 0.1% HCOOH, 4:1) to afford 12 (t_R 16.4 min), which was subjected to chiral HPLC using a Cellulose-3 column (CH₃CN/H₂O containing 0.1% HCOOH, 11:9) to afford 12a (t_R 22.1 min) and 12b (t_R 24.5 min).

(E)-4-Methylhex-2-enoic acid (11):

Colorless oil; ¹H NMR (600 MHz, methanol-d₄) δ 6.82 (1H, dd, J = 15.7, 8.0 Hz), 5.75 (1H, d, J = 15.7 Hz), 2.22 (1H, m), 1.41 (2H, dq, J = 7.4, 7.4 Hz), 1.04 (3H, d, J = 6.7 Hz), 0.89 (3H, t, J = 7.5 Hz); ¹³C NMR (150 MHz, methanol-d₄) δ 168.8, 154.7, 119.6, 37.9, 28.5, 18.0, 10.5.

2-(Naphthalen-2-yl)-2-oxoethyl (E)-4-Methylhex-2-enoate (12):

Colorless oil; ¹H NMR (500 MHz, methanol-d₄) δ 8.59 (1H, br s), 8.04 (1H, br d, J = 8.1 Hz), 7.97 (2H, m), 7.93 (1H, br d, J = 8.2 Hz), 7.65 (1H, ddd, J = 8.2, 6.9, 1.3 Hz), 7.59 (1H, ddd, J = 8.2, 6.9, 1.3 Hz), 6.98 (1H, dd, J = 15.7, 7.9 Hz), 5.97 (1H, dd, J = 15.7, 1.3 Hz), 5.62 (2H, s), 2.29 (1H, m), 1.45 (2H, dq, J = 7.4, 7.4 Hz), 1.09 (3H, d, J = 6.7 Hz), 0.92 (3H, t, J = 7.4 Hz); ¹³C NMR (125 MHz, methanol-d₄) δ 193.1, 166.3, 155.8, 135.9, 132.5, 131.5, 129.5, 129.3, 128.6, 128.4, 127.5, 126.7, 122.8, 118.6, 66.0, 38.1, 28.5, 17.9, 10.6; ESIMS [M + H]⁺ m/z 297, [M − H]⁻ m/z 295.

(−)-2-(Naphthalen-2-yl)-2-oxoethyl (R,E)-4-methylhex-2-enoate (12a):

Colorless oil; [α]²²_D −10 (c 0.1, CH₂Cl₂), −20 (c 0.1, MeOH); ¹H NMR (500 MHz, methanol-d₄) δ 8.62 (1H, br s), 8.08 (1H, br d, J = 7.9 Hz), 8.00 (2H, m), 7.96 (1H, br d, J = 8.4 Hz), 7.67 (1H, ddd, J = 8.2, 6.9, 1.3 Hz), 7.62 (1H, ddd, J = 8.2, 6.9, 1.3 Hz), 7.00 (1H, dd, J = 15.8, 8.0 Hz), 5.99 (1H, dd, J = 15.8, 1.3 Hz), 5.64 (2H, s), 2.32 (1H, m), 1.47 (2H, dq, J = 7.4, 7.4 Hz), 1.11 (3H, d, J = 6.7 Hz), 0.94 (3H, t, J = 7.4 Hz); HRESIMS m/z 319.1316 [M + Na]⁺ (calcd for C₁₉H₂₀NaO₃, 319.1305).

(+)-2-(Naphthalen-2-yl)-2-oxoethyl (S,E)-4-methylhex-2-enoate (12b):

Colorless oil; [α]²²_D +8 (c 0.1, CH₂Cl₂), +12 (c 0.1, MeOH) ¹H NMR (500 MHz, methanol-d₄) δ 8.62 (1H, br s), 8.08 (1H, br d, J = 8.0 Hz), 8.00 (2H, m), 7.96 (1H, br d, J = 8.0 Hz), 7.67 (1H, ddd, J = 8.2, 6.9, 1.3 Hz), 7.62 (1H, ddd, J = 8.2, 6.9, 1.3 Hz), 7.00 (1H, dd, J = 15.7, 8.0 Hz), 5.99 (1H, dd, J = 15.7, 1.3 Hz), 5.64 (2H, s), 2.32 (1H, m), 1.47 (2H, dq, J = 7.4, 7.4 Hz), 1.11 (3H, d, J = 6.7 Hz), 0.94 (3H, t, J = 7.4 Hz); HRESIMS m/z 319.1309 [M + Na]⁺ (calcd for C₁₉H₂₀NaO₃, 319.1305).

Acid Hydrolysis and 2-Naphthacyl Esterification of Leucinostatins for Chiral HPLC Analysis.

Acid hydrolysis of compound 6 (0.2 mg) was performed using 6 N HCl (200 μL) at 110 °C for 2h. The distilled H₂O (1 mL) was added to the resultant solution, and further partitioned with EtOAc (1 mL × 3). α-Bromo-2-acetonaphthone (0.5 mg) and Et₃N (50 μL) were applied to the collected organic phase without evaporation of EtOAc. The reaction mixture was stirred at room temperature for 3 h. After removing solvent from reaction mixture in vacuo, the products were purified by C₁₈ semipreparative HPLC (CH₃CN/H₂O containing 0.1% HCOOH, 7:3). Compounds 1, 3, 5, 7, and 9 were processed in the same manner to get the corresponding 2-naphthacyl esters. Chiral HPLC analyses were performed on a Cellulose-3 column (CH₃CN/H₂O containing 0.1% HCOOH, 11:9) with authentic standards (t_R 12a: 22.1 min, 12b: 24.5 min). Sample identities were confirmed by co-injection with the racemic mixture 12.

Cell Culture.

The BT-549 cell line was acquired from the Lombardi Cancer Center at George Washington University and authenticated by STR-based profiling (Promega). SUM185PE cells were purchased from Asterand Bioscience (Detroit, MI, USA). All other cell lines were obtained from the American Type Culture Collection (Manassas, VA, USA) and their identities authenticated (Genetica DNA Laboratories, Burlington, NC, USA). The HCC1806, HCC1937, HCC70 and BT-549 cells were grown in RPMI-1640 supplemented with 5% fetal bovine serum (FBS) and 0.5% gentamicin. MDA-MB-453, MDA-MB-231, and SUM185PE cells were cultured in IMEM with 5% FBS and 0.3% gentamicin. All cellular assays were conducted within 4 months of retrieval of cells from liquid nitrogen stocks.

Sulforhodamine B Assay.

The sulforhodamine B (SRB) assay was used to evaluate antiproliferative and cytotoxic effects of compounds as previously described.⁵⁷ Briefly, cells were seeded in 96-well plates at predetermined densities, allowed to adhere to the plate surface and treated with compounds 1 – 10, rapamycin (LC Labs, Woburn MA, USA), oligomycin (Cayman Chemical, Ann Arbor, MI, USA), or vehicle (DMSO) for 48 h. Following treatment, cells were fixed and proteins stained with SRB dye, and the absorbance measured at 560 nm. Concentration-response curves were generated and GI₅₀ values (concentrations that caused 50% inhibition of proliferation as compared to vehicle controls) determined by non-linear regression using GraphPad Prism 6. All data are representative of three independent experiments with each experiment conducted in triplicate.

Immunoblotting.

Cells were treated with 100 nM compound 5, 10 nM rapamycin, 100 nM oligomycin or DMSO vehicle for 2, 6, or 18 h. Cells were harvested by scraping and whole-cell lysates were prepared using a commercial cell extraction buffer (Invitrogen, Carlsbad CA, USA) containing PSMF, Na₃VO₄, and protease inhibitor cocktail. Quantitative protein assays were performed for each sample and equal amounts of protein (10 μg) were applied to SDS-PAGE gels with 10% Bolt Bis-Tris (Invitrogen) for separation. The proteins were transferred to PVDF membranes (MilliporeSigma, Burlington, MA, USA) and the membranes probed for P-S473-AKT, AKT, P-S2448-mTOR, mTOR, P-T389-S6K, S6K, P-S235/236-RPS6, RPS6, P-T37/46–4E-BP1, 4E-BP1, or GAPDH (Table S9, Supporting Information) diluted 1:1000 in 30% Odyssey blocking buffer (LI-COR Biosciences, Lincoln Ne, USA) with TBST. P-S235/236-RPS6 and RPS6 were visualized on the same membranes using 2-color imaging, all other proteins were imaged individually. Immunoblots were visualized using an Odyssey Fc (LI-COR Biosciences) imaging system after incubation with suitable IRDye 680 or IRDye 800 secondary antibodies (LI-COR Biosciences).

Scheme 1.

Structures of leucinostatins (1-10) and annotations noting the abbreviations used for the amino acid residues described in this report.