Cytotoxic dimeric epipolythiodiketopiperazines from the ascomycetous fungus Preussia typharum.

Two new dimeric epipolythiodiketopiperazines, preussiadins A (1) and B (2), together with two known diastereomers, leptosins C (6) and A (7), were obtained from the mycelia of a Preussia typharum isolate. The structures of the new compounds were established by spectroscopic methods, and the absolute configurations of 1 and 2 were assigned by chemical transformations and comparisons of quantum chemical ECD and VCD calculations to experimental data. Compound 1 exhibited potent cytotoxic activity in the NCI-60 cell line panel with an average LC50 value of 251 nM. Further studies demonstrated that 1 circumvents P-glycoprotein-mediated drug resistance, yet had no significant antitumor activity in a xenograft UACC-62 melanoma model.

The epipolythiodiketopiperazine (ETP) alkaloids represent a large family of fungal secondary metabolites with potent cytotoxic activities.[1,2] Recently, ETP alkaloids have been reported to inhibit a variety of cellular targets and signaling processes including topoisomerases I and II, Akt, and protein–protein interactions critical for cancer cell growth including the interaction between HIF1-α and p300.[3−15] Some of the ETPs were also found to inhibit the growth of tumor xenografts in mice.[4−7,10,12] One of the most widely studied ETPs, chaetocin, has displayed a wide spectrum of in vivo antitumor activities against SK-OV-3 human ovarian cancer,[4] U937 human leukemia,[5] RPMI 8226 human myeloma,[6] and mouse and human hepatomas.[10] These observations suggest the ETPs may have utility as lead molecules for drug development and/or molecular tools. In this report, we describe the isolation, structural determination, absolute configuration assignments, and conformational analyses of two new ETPs, preussiadins A (1) and B (2), together with two known diastereomers, leptosins C (6) and A (7), from Preussia typharum. We also investigated the in vitro cytotoxicity of these compounds, as well as the in vivo antitumor efficacy of 1.

Results and Discussion

Preussiadin A (1) was obtained as a white, amorphous powder from a P. typharum isolate that originated from a bottomland-forest-derived soil sample.[16] Its molecular formula was established as C32H32N6O7S4 by HRESIMS, indicating 20 degrees of unsaturation. The IR spectrum confirmed the presence of amide (νmax 1673) and hydroxyl (νmax 3399) groups. The 1H NMR spectrum displayed characteristic resonances attributable to two ABCD aromatic systems (δH 5.71–7.70, 8H) (Table 1). Analysis of the 13C NMR and 1H–13C HSQC spectra enabled the assignment of the 32 carbon resonances to four methyl groups including two N-methyls (δC 27.5 and 28.0), 13 methines, one oxygenated methylene (δC 60.7), four carbonyls (δC 167.6, 160.7, 165.5, and 163.1), and 10 additional quaternary carbons (Table 1). Further analysis of the 1D (1H and 13C) and 2D (HMBC) NMR data (Table S1) enabled the determination that 1 was a new dimeric ETP, which was a diastereomer of leptosin C (6).[17] The 13C NMR data for 1 and 6 were nearly superimposable for the lower monomeric unit (C-1–C-16); however, significant variation occurred (>2 ppm changes) among the chemical shifts assigned to the C-1′, C-3′, C-4′, C-5a′, and C-11′ resonances for the upper monomeric portion. We surmised that this was due to a change in the absolute configuration of the diketopiperazine system in the upper half of the molecule. To investigate this in greater detail, 1 was subjected to cleavage of the disulfide bond followed by S-methylation.[17] In addition to the expected product 3, two minor N-methylated products, 4 and 5, were also obtained. The planar structure of 3 was confirmed by analysis of its HRESIMS and 1D (1H and 13C NMR, Tables 1 and 2) and 2D (HSQC and HMBC, Table S3) NMR data. The N-methylation patterns of 4 and 5 were determined by comparisons of the 1H and 13C (Table 1) NMR data with those of 3. Significant downfield shifts were observed for C-5a′ or C-6a′ and C-5a or C-6a due to the N-methylation of their respective adjacent nitrogen atoms in 4 and 5, respectively. The relative configuration of 3 was established as shown in Figure 1 based on ROESY correlation data (H-11 to S-Me-12, H-10, H-11′; H-10 to H-10′, OH-11′; OH-11′ to H-10′, S-Me-12′; OH-11 to H-11′, H-5a′; H-5a′ to H-5a, Me-16). Thus, the configuration assignments for 3 enabled us to infer the inversion of C-3′ and C-12′ in 1 relative to the remaining portions of the molecule. This confirmed that metabolite 1 was a diastereomer of leptosin C (6).

Table 1

1H NMR Data of 1–5 (400 MHz, TMS, δ ppm)

	1a		2a		3a		4a		5a
no.	δC	δH	δC	δH	δC	δH	δC	δH	δC	δH
1	167.6		167.4		165.8		165.9		165.9
3	80.4		80.7		79.2		79.1		79.3
4	160.7		160.6		164.7		164.7		164.6
5a	79.0	5.96, s	80.4	5.77, s	79.7	5.76, d (3.0)	79.9	5.61, d (3.1)	86.1	5.62, s
6		5.46, s		4.64, s		4.45, d (2.9)		4.44, d (3.1)
6a	148.2		150.0		151.1		151.2		153.5
7	110.4	6.62, d (7.6)	110.0	6.42, d (7.6)	109.1	6.35, d (8.3)	109.1	6.35, d (7.4)	107.0	6.19, d (7.8)
8	130.4	7.08, t (7.0)	130.5	7.18, t (7.6)	129.9	7.14, td (7.7, 1.2)	129.9	7.15, td (7.7, 1.2)	130.0	7.20, td (7.8, 1.2)
9	119.9	6.45, t (7.7)	119.2	6.88, t (7.5)	118.4	6.82, td (7.5, 0.9)	118.4	6.80, td (7.5, 0.9)	117.3	6.76, td (7.4, 0.8)
10	125.3	5.71, d (7.7)	126.3	7.55, d (7.3)	125.6	7.72, d (7.5)	125.6	7.67, d (7.5)	125.2	7.65, d (7.4)
10a	126.3		127.1		128.4		128.5		128.7
10b	65.4		64.6		63.1		63.0		62.2
11	81.9	5.01, s	82.7	5.25, s	81.1	5.17, d (2.8)	81.1	5.14, d (3.0)	80.8	5.15, d (3.2)
12	75.2		75.5		72.5		72.6		72.9
13	28.0	3.08, s	27.8	3.04, s	30.0	3.04, s	30.0	3.04, s	30.1	3.04, s
14	32.4	2.69, heptet (6.8)	32.4	2.68, heptet (6.8)	37.6	2.45, heptet (7.0)	37.5	2.45, heptet (7.0)	37.4	2.46, heptet (7.0)
15	18.1	1.44, br s	18.2	1.43, d (6.8)	18.2	1.23, d (7.0)	18.2	1.27, d (7.0)	18.2	1.24, d (6.8)
16	18.7	1.44, br s	18.7	1.43, d (6.8)	18.2	1.03, d (7.0)	18.5	1.09, d (7.0)	18.5	1.04, d (7.0)
11-OH		6.82, s		5.58, s		3.39, d (3.0)		3.37, d (2.9)		3.20, d (3.2)
3-SMe					13.8	2.11, s	13.8	2.11, s	13.9	2.11, s
6-NMe									33.1	2.53, s
12-SMe					16.5	1.96, s	16.4	1.96, s	16.5	1.95, s
1′	165.5		170.3		165.8		165.9		165.9
3′	75.4		78.5		71.0		71.6		71.0
4′	163.1		167.8		163.8		163.4		163.8
5a′	77.0	5.67, s	79.1	6.56, d (2.0)	78.1	6.03, d (2.3)	83.9	6.10, s	78.0	6.00, d (2.6)
6′		4.48, s		4.84, s		4.83, d (2.3)
6a′	151.0		149.8		150.1		152.2		150.0
7′	110.2	6.52, d (7.3)	107.9	6.41, d (7.8)	108.8	6.56, d (7.8)	105.7	6.37, d (7.5)	108.7	6.59, d (7.8)
8′	130.2	7.26, t (7.3)	129.7	6.98, t (7.6)	129.6	7.07, td (7.7, 1.8)	129.8	7.10, td (7.5, 1.8)	129.7	7.03, td (7.7, 1.2)
9′	119.6	6.98, t (7.6)	118.1	6.28, t (7.4)	118.2	6.35, t (7.7)	118.4	6.27, td (7.5, 1.0)	118.1	6.34, td (7.6, 1.0)
10′	128.1	7.70, d (7.0)	128.9	5.85, d (7.6)	128.4	5.78, d (8.0)	128.1	5.70, dd (7.6, 1.0)	128.5	5.70, d (7.6)
10a′	124.0		122.8		123.0		123.3		123.0
10b′	64.4		60.0		61.0		60.0		61.1
11′	75.8	5.72, s	79.9	5.63, d (3.4)	79.6	5.41, d (11.4)	80.0	5.33, d (10.2)	79.6	5.38, d (11.4)
12′	77.4		78.5		73.7		73.6		73.7
13′	27.5	3.15, s	28.7	3.11, s	28.7	3.15, s	28.7	3.15, s	28.7	3.14, s
14′a	60.7	4.23, dd (7.9, 10.2)	62.9	4.04, d (12.4)	64.2	3.89, d (12.1)	64.1	3.87, d (12.0)	64.3	3.89, d (12.0)
14′b		4.36, dd (4.3, 10.2)		4.38, d (12.4)		4.34, d (12.1)		4.35, d (12.0)		4.34, d (12.0)
11′-OH		5.14, s		3.59, d (3.5)		3.15, d (11.4)		3.09, d (10.2)		3.12, d (11.5)
14′-OH		3.45, dd (4.3, 7.9)				5.03, s		5.01, d (1.2)
3′-SMe					14.0	2.25, s	14.0	2.25, s	14.0	2.25, s
6′-NMe							32.1	3.05, s
12′-SMe					16.1	1.98, s	16.1	1.99, s	16.0	1.98, s

Recorded in CDCl3 at 25 °C.

Table 2

Inhibition of in Vitro Cancer Cell Lines by Preussiadin A (1)a

cell line	GI50 (nM)b	TGI (nM)c	LC50 (nM)d
Leukemia
CCRF-CEM	9.3	38.0	>50 000
HL-60 (TB)	26.3	195.0	>50 000
K-562	14.8	114.8	>50 000
MOLT-4	13.2	67.6	>50 000
RPMI-8226	12.3	34.7	>50 000
SR	11.2	31.6	>50 000
Non-Small Cell Line Cancer
A549/ATCC	26.3	93.3	263.0
EKVX	28.2	257.0	2041.7
HOP-62	11.7	30.9	109.6
HOP-92	11.0	28.8	112.2
NCI-H226	10.0	24.0	72.4
NCI-H23	16.2	56.2	234.4
NCI-H322M	20.0	102.3	489.8
NCI-H460	20.9	75.9	245.5
NCI-H522	8.5	17.8	38.0
Colon Cancer
COLO 205	14.8	49.0	154.9
HCC-2998	53.7	114.8	245.5
HCT-116	12.3	51.3	162.2
HCT-15	17.0	89.1	478.6
HT29	5.6	13.5	32.4
KM12	19.5	87.1	295.1
SW-620	8.3	31.6	128.8
CNS Cancer
SF-268	22.4	107.2	416.9
SF-295	57.5	154.9	416.9
SF-539	15.1	58.9	199.5
SNB-19	112.2	512.9	1659.6
SNB-75	13.5	50.1	158.5
U251	17.8	61.7	177.8
Melanoma
LOX IMVI	8.3	21.4	46.8
MALME-3M	7.8	18.6	45.7
M14	13.5	53.7	182.0
MDA-MB-435	10.7	28.8	120.2
SK-MEL-2	9.8	20.4	43.7
SK-MEL-28	12.6	38.0	134.9
SL-MEL-5	14.8	57.5	151.4
UACC-257	8.7	19.5	43.7
UACC-62	18.2	72.4	295.1
Ovarian Cancer
IGROV1	24.5	100.0	338.8
OVCAR-3	10.7	26.3	85.1
OVCAR-4	5.9	12.9	27.5
OVCAR-5	52.5	138.0	363.1
OVCAR-8	9.5	19.1	38.0
NCI/ADR-RES	33.9	120.2	354.8
SK-OV-3	11.5	33.9	190.5
Renal Cancer
786-0	11.7	26.3	70.8
A498	9.8	28.2	104.7
ACHN	15.5	60.3	234.4
CAKI-1	8.7	22.4	85.1
RXF 393	7.8	14.8	28.2
SN12C	22.9	524.8	1778.3
TK-10	11.7	24.5	57.5
UO-31	9.8	26.3	100.0
Prostate Cancer
PC-3	15.5	60.3	208.9
DU-145	14.8	53.7	199.5
Breast Cancer
MCF7	10.0	55.0	166.0
MDA-MB-231/ATCC	18.6	70.8	323.6
HS 578T	26.3	446.7	NTe
BT-549	14.8	66.1	208.9
T-47D	9.1	26.3	102.3

Data obtained from the NCI-60 cell line panel.

GI50 is the molar concentration causing 50% growth inhibition of tumor cells.

TGI is the molar concentration giving total growth inhibition.

LC50 is the molar concentration leading to 50% net cell death.

Not tested.

Figure 1

Key ROESY correlations for 3.

The independent assignment of the absolute configuration of 1 was initially hindered when all attempts to crystallize the compound and its derivatives proved unsuccessful. Alternatively, quantum chemical computational calculations were carried out to generate theoretical VCD and ECD spectra, as well as a specific rotation value for 1. A conformational search was carried out with Spartan’10 at the molecular mechanics level (MMFF)[18] based on the proposed absolute configuration 3S*, 5aR*, 10bS*, 11S*, 12S*, 3′R*, 5′aS*, 10′bR*, 11′S*, 12′R*. Geometry, frequency, specific rotation, and IR and VCD intensity calculations of the 20 lowest energy conformers (within 5 kcal/mol) resulted from a conformational search carried out at the DFT level (B3LYP functional/6-31+G** basis set) with Gaussian 09.[19] The calculated frequencies were scaled by 0.975, and the IR and VCD intensities were converted to Lorentzian bands with a 6 cm–1 half-width for comparisons to experimental values.

Gaussian calculations resulted in six low-energy conformers within 2.7 kcal/mol of the lowest energy conformer, whereas all other conformers were more than 4.4 kcal/mol higher in energy. The IR and VCD spectra and specific rotations of the six conformers were summed based on a Boltzmann statistical weighting (Figure 2A). TD-DFT calculations (B3LYP functional/6-31+G** basis set) provided individual UV and CD spectra for the six conformers, which were then combined using SpecDis 1.60 (using a sigma value of 0.2 eV).[20] After applying a UV-shift correction of −50 nm, the theoretical CD spectra were compared with the experimental data (Figure 2B). In both cases, the calculated VCD (Figure 2A) and ECD (Figure 2B) spectra matched well with the experimental data. In addition, a positive specific rotation value was calculated for 1 ([α]D 53), which was in agreement with the experimental observations ([α]24D 66). Therefore, the absolute configuration of 1 was determined to be 3S, 5aR, 10bS, 11S, 12S, 3′R, 5′aS, 10′bR, 11′S, 12′R.

Figure 2

Comparison of experimental and calculated VCD (A) and ECD (B) spectra of 1.

Preussiadin B (2) was assigned a molecular formula of C32H32N6O7S6 by HRESIMS, indicating the presence of two additional sulfur atoms in the molecule compared to 1. The 1H and 13C NMR data (Tables 1 and 2) as well as HMBC correlations (Table S2) enabled us to confirm that 1 and 2 shared a majority of the same structural features. Further scrutiny of the NMR data revealed that the lower portions of molecules 1 and 2 were identical; however, the upper parts of the compounds exhibited some small, but perceptible changes that led us to conclude the two additional sulfur atoms formed a tetra-sulfide bridge between C-3′ and C-12′. The proposed structure and absolute configuration assignment for 2 was determined by subjecting the metabolite to the previously described S-methylation reaction. This provided three products (3–5) that were confirmed as being identical to the products of the cleavage/methylation reaction of 1. Thus, we were able to confirm that both compounds 1 and 2 shared the same absolute configuration.

Two previously described dimeric ETPs, leptosins A (7) and C (6),[21] were also purified from the fungal extract. Fortuitously, the 1H NMR spectrum of 6 exhibited some unusual features that ultimately led to further insights concerning the axial chirality of these compounds. In CDCl3 at 25 °C, we observed broad proton resonances for 6 (Figure 3A) that we interpreted as arising from the interconversion of conformers. Using variable-temperature (VT) NMR experiments at −20, −40, and −60 °C, we succeeded in generating two distinct sets of sharpened proton signals (Figure 3). However, upon switching to acetone-d6 as the solvent, we were able to generate a single major stereoisomer that was stabilized at low temperature (Figure 3B). This contrasted with data for 1–5 and 7, in which only a single major stereoisomer was observed in CDCl3. These results prompted us to further probe the axial chirality features of this class of compounds. ROESY experiments enabled us to directly deduce that 1 favored a P configuration, whereas 2 and 3 favored an M configuration under these experimental conditions (Tables S1 and S2 and Figures 1 and 4A). Whereas the assignments for 4–7 could not be directly extrapolated from the ROESY data, our results did generate an additional useful pattern that was evident in the 1H NMR data. Namely, the 1H NMR spectrum of 1 showed that H-10 was shifted upfield to δH 5.71 relative to H-10′ (δH 7.70) due to the shielding effect of phenyl ring B (Figures 4A and 4B). In contrast, H-10′ was observed in upfield positions for 2 (δH 5.85) and 3 (δH 5.78) due to shielding of phenyl group A (Figures 4A and 4B). Applying these observations to the remaining compounds in this series, we deduced that 4–6 adopted an M configuration, whereas 7 favored a P configuration (Figure 4).

Figure 3

Variable-temperature 1H NMR experiments of 6 in CDCl3 (A) and acetone-d6 (B).

Figure 4

Key ROESY correlations for 1 (P-isomer), 2, and 3 (M-isomer) (A) and selected 1H NMR data for 1–7 (B).

The cytotoxic activities of 1–7 were initially tested in a human pancreatic cancer cell line (MIA PaCa-2). Compounds 1 and 2 were found to be the most potent, with IC50 values of 6.6 and 9.1 nM, respectively. Diastereomers 6 and 7 were determined to be nearly five times less potent, with IC50 values of 33.6 and 43.1 nM, respectively. Compounds 3–5 exhibited no cytotoxic effects against the MIA PaCa-2 cells at concentrations of up to 30 μM. These results further support the importance of the polythioether bridge for triggering mammalian cell toxicity.

To further elaborate the scope of activity for compound 1, it was evaluated in the NCI-60 in vitro drug screening program. Compound 1 exhibited potent antiproliferative activity against each of the 60 cell lines with an average GI50 value of 14.8 nM (Table 2). Compound 1 also exhibited cytotoxic effects against all of the solid tumor cell lines with an average LC50 of 251 nM; however, it was not cytotoxic against the six leukemia cell lines, indicating that 1 exhibited some degree of selectivity.

A common mechanism of multidrug resistance is the expression of the P-glycoprotein (Pgp) transporter, the product of the MDR1 gene. Efflux of drug molecules through this transporter contributes to development of innate and acquired drug resistance, including resistance to many structurally complex natural products.[22] The ability of 1 to circumvent Pgp-mediated drug resistance was studied using an isogenic cell line pair, the parental SK-OV-3 line and the Pgp-expressing clone SK-OV-3/MDR-1-M6/6 (M6/6).[23] The dose–response curves for 1 and paclitaxel, a known Pgp substrate, are shown in Figure 5A. The relative resistance values, indicating the effects of Pgp expression, were calculated by dividing the GI50 of 1 in the Pgp-expressing M6/6 line by the GI50 obtained in the parental SK-OV-3 line.[24] As expected, paclitaxel was significantly less potent against the Pgp-expressing cell line and had a relative resistance value of 127 (Figure 5). In contrast, 1 had a relative resistance of 1.7, indicating that it was a poor Pgp substrate (Figure 5).[25] These results are consistent with the NCI-60 data, where 1 was effective against the Pgp-expressing NCI/ADR-RES cell line.[26−28] These results indicate that 1 overcomes Pgp-mediated drug resistance.

Figure 5

(A) Dose–response curves for growth inhibition of SK-OV-3 and SK-OV-3-M6/6 cells following treatment with 1 or paclitaxel (PTX). Data represent mean ± SE for n = 1–4 independent experiments. When not visible, error bars are contained within the boundaries of the data points, except for PTX treatment at 30 and 250 nM (n = 1). (B) GI50 values for 1 and PTX in SK-OV-3 and SK-OV-3-M6/6 cells and their respective relative resistance (Rr) values. Results represent n = 3 or 4 independent experiments.

Cellular persistence evaluates the long-term effects of a short exposure of cells to a drug. Although it is unclear whether high cellular persistence is a desired property of anticancer agents, recent studies have demonstrated that it can be a predictor of high in vivo efficacy.[29,30] The cellular persistence of 1 was evaluated in a clonogenic assay using UACC-62 melanoma cells, which was one of the most sensitive cell lines evaluated in the NCI-60 panel, as well as HeLa cells. The cells were exposed to different concentrations of 1 for 4 h, washed to remove residual compound, and allowed to form colonies for 14 days. A 4 h exposure of cells to 1 caused a dose-dependent inhibition of colony formation, with 5 nM 1 reducing colony formation by 38% as compared to vehicle-treated cells ( 0.0286) (Figure 6). Similar results were also obtained in experiments using HeLa cells (data not shown). A previous study by our group using a similar experimental design demonstrated that cellular persistence of a compound after 4 h of treatment can vary dramatically for different cytotoxic agents, indicating that this effect is specific to 1 and not a general effect of all cytotoxins.[31] These results suggest that the cellular effects of 1 are highly persistent and that short exposure times are sufficient to induce cytotoxicity.

Figure 6

(A) Representative images of crystal violet-stained UACC-62 colonies 14 days after 4 h of treatment with vehicle or various concentrations of 1. (B) Quantification of UACC-62 colony number. Results represent n = 2–4 independent experiments, with each concentration tested in duplicate or triplicate. *p < 0.05 compared to vehicle determined by one-way ANOVA and Dunnet’s post hoc test.

The COMPARE algorithm[32] was used to evaluate the potential mechanism of action of 1. The results indicated that 1 might share a mechanism of action with the mTORC1 inhibitor rapamycin and other mTORC inhibitors. To determine whether 1 is capable of inhibiting cellular mTORC1 signaling, whole-cell lysates of vehicle and 1-treated HeLa cells (HeLa GI50 = 20 nM) were analyzed by Western blotting to determine the relative levels of phosphorylation of S6 kinase (S6K), a major downstream effector of mTORC1. While the known mTORC1 inhibitor rapamycin (100 nM) completely inhibited phosphorylation of the T389 residue of S6K, 1 had no effect on this phosphorylation at concentrations up to 100 nM (Figure S22). This finding indicates that 1 does not inhibit signaling through the mTORC1 pathway at cytotoxic concentrations to HeLa and other cancer cell lines. This suggests that mTORC1 inhibition is not involved in the mechanism of action of 1.

It has previously been reported that chaetocin, a compound with structural similarity to 1, inhibits the di- and trimethylation of histone H3 lysine 9 (H3K9) by targeting the lysine methyltransferase SU(VAR)3-9.[33] Another recent report suggested that chaetocin may nonspecifically inhibit lysine methyltransferases.[34] The effects of 1 on H3K9 methylation were studied in HeLa cells treated with vehicle or 200 nM 1 for 1–24 h. The relative levels of trimethylated H3K9 (H3K9me3) were evaluated by immunoblotting. The results indicated that 1 did not cause a decrease of H3K9me3 levels, but actually caused higher expression of H3K9me3 (Figure S23). This increase is likely an indirect result of suppression of transcription as cells begin to die.

The antitumor effects of 1 were evaluated in a UACC-62 xenograft mouse model. On the basis of the high level of cellular persistence of 1, it was expected that this compound could be administered infrequently while retaining efficacy. Initial dose–tolerance testing was conducted with 10 mg/kg of 1, but this dose caused rapid and persistent weight loss of 5–10%. Further studies showed that two or more doses of 1 at 2 mg/kg also caused rapid and persistent weight loss. A dose of 1.5 mg/kg every 4 days was determined to be the maximum tolerated dose and was used for antitumor efficacy studies. With this dosing schedule, 1 slightly inhibited tumor growth (T/C% = 67%) and reduced final tumor volume relative to untreated control mice, but this effect was not statistically significant (p = 0.2896; Figure S24). The results of this study indicate that 1 may have antitumor efficacy in vivo, but it has a very narrow therapeutic window.

Experimental Section

General Experimental Procedures

Optical rotations were measured on a Rudolph Research Autopol III automatic polarimeter. UV data were measured with a Hewlett-Packard 8452A diode array spectrophotometer. ECD spectra were recorded on a model 202-01 AVIV circular dichroism spectrometer. IR spectra were measured on a Bruker Vector 22 FT-IR spectrometer. NMR data were obtained on a Varian VNMR spectrometer (400 MHz for 1H and 2D NMR, 100 MHz for 13C NMR) with a broad band resonance probe. Electrospray-ionization mass spectrometry data were collected on an Agilent 6538 high-mass-resolution QTOF mass spectrometer. HPLC separations were performed on a Shimadzu system using an SCL-10A VP system controller and a Gemini 5 μm C18 column (110 Å, 250 × 21.2 mm and 250 × 10.0 mm) with flow rates of 10 or 4 mL/min. All solvents were of ACS grade or better.

Fermentation, Extraction, and Isolation

The identification of P. typharum (Sacc.) Cain (GenBank accession JX143871), its fermentation on Cheerios breakfast cereal, and extraction procedures were described previously.[16] The crude extract (100 g) was separated into nine fractions by silica column chromatography (eluted with a gradient of hexanes–dichloromethane and dichloromethane–MeOH). Fraction 7 was further separated into three subfractions by C18 vacuum column chromatography. Fraction 7-2 was subjected to repeated Sephadex LH20 column chromatography (eluted with dichloromethane–MeOH, 1:1) and silica preparative TLC (hexanes–EtOAc) to yield compounds 6 (66 mg, 0.066% yield) and 7 (10 mg, 0.010% yield). Subfraction 7-2-3-4 was further subjected to semipreparative HPLC (Luna 5 μm, C8, 110 Å, 250 × 10.00 mm, MeCN–H2O, 55:45, 4.0 mL/min) to yield 1 (10 mg, 0.010% yield) and 2 (3.6 mg, 0.0036% yield).

Preussiadin A (1):

white powder; [α]24D 66 (c 0.1, CHCl3); UV (EtOH) λmax (log ε) 208 (4.55), 300 (3.59); IR (film) νmax 3399, 2966, 2935, 1673, 1482, 1463, 1351, 1315, 1248, 1217, 1140, 1061 cm–1; CD (EtOH) λmax (Δε) 219 (−5.3), 236 (16.8), 262 (2.3), 313 (1.2); 1H and 13C NMR data in Tables 1 and 2, respectively; HRESIMS m/z 763.1102, [M + Na]+ (calcd for C32H32N6O7S4Na, 763.1108).

Preussiadin B (2):

white powder; [α]24D 34 (c 0.1, CHCl3); UV (EtOH) λmax (log ε) 210 (4.79), 300 (3.69); IR (film) νmax 3399, 2962, 2928, 1664, 1474, 1455, 1418, 1373, 1313, 1261, 1216, 1096, 1066 cm–1; CD (EtOH) λmax (Δε) 212 (20.6), 234 (18.2), 268 (−8.5), 335 (1.2); 1H and 13C NMR data in Tables 1 and 2, respectively; HRESIMS m/z 827.0556, [M + Na]+ (calcd for C32H32N6O7S6Na, 827.0549).

Computational Details

The conformational analysis was carried out using Spartan’10. Geometry, frequency, IR and VCD intensity, UV and ECD spectra, and specific rotation calculations were applied at the DFT level with Gaussian 09. ComputeVOA v0.1 was used to sum IR or VCD spectra after a Boltzmann statistical weighting. UV and ECD spectra were added based on Boltzmann distribution using SpecDis 1.60. The assignment based on VCD comparison was evaluated using the CompareVOA program.[35] The confidence level of the assignment for the proposed absolute configuration of 1 is 95% based on its current database, which includes 105 previous correct assignments for different chiral structures.

Formation of the Tetrakis(methylsulfanyl) Derivatives 3–5

Preussiadin A (1) (13 mg) was dissolved in 0.26 mL of pyridine–MeOH (5:8). MeI (1 mL) and NaBH4 (5 mg) were added, and the mixture was stirred for 20 min at room temperature. The reaction mixture was then diluted with water and extracted with EtOAc. The solvent was evaporated under reduced pressure, and the residue was separated on a preparative silica TLC plate with hexanes–EtOAc (7:4) as an eluent. The products were further purified by semipreparative HPLC using MeOH–H2O (4:1) to afford 3 (2.6 mg), 4 (2.1 mg), and 5 (0.5 mg). The same reaction with 2 (2.0 mg) gave 3 (0.7 mg), as well as 4 and 5 in trace amounts.

Preussiadin A1 (3):

white powder; [α]24D −14.2 (c 0.14, CHCl3); UV (EtOH) λmax (log ε) 208 (4.55), 244 (sh, 3.98), 306 (3.56); IR (film) νmax 3378, 2966, 2935, 1655, 1484, 1457, 1420, 1383, 1317, 1259, 1205, 1088 cm–1; CD (EtOH) λmax (Δε) 212 (16.0), 237 (sh, 1.3), 254 (−10.5), 305 (−0.3), 324 (−0.8); 1H and 13C NMR data in Tables 1 and 2, respectively; HRESIMS m/z 823.2053, [M + Na]+ (calcd for C36H44N6O7S4Na, 823.2047).

Preussiadin A2 (4):

white powder; [α]24D −30 (c 0.12, CHCl3); UV (EtOH) λmax (log ε) 210 (4.60), 248 (sh, 4.03), 308 (3.58); IR (film) νmax 3396, 2967, 2924, 1654, 1488, 1419, 1382, 1308, 1254, 1196, 1123, 1096, 1047 cm–1; CD (EtOH) λmax (Δε) 212 (28.0), 239 (sh, 3.0), 258 (−10.0), 301 (2.5), 326 (−2.3); 1H and 13C NMR data in Tables 1 and 2, respectively; HRESIMS m/z 837.2213, [M + Na]+ (calcd for C37H46N6O7S4Na, 837.2203).

Preussiadin A3 (5):

white powder; [α]24D −32 (c 0.03, CHCl3); UV (EtOH) λmax (log ε) 210 (4.65), 248 (sh, 4.10), 308 (3.65); IR (film) νmax 3408, 2966, 2927, 1655, 1464, 1422, 1382, 1306, 1260, 1199, 1133, 1083, 1045 cm–1; CD (EtOH) λmax (Δε) 208 (6.6), 219 (−2.3), 227 (2.2), 235 (−1.0), 243 (1.9), 257 (−11.8), 292 (−2.9), 326 (2.4); 1H and 13C NMR data in Tables 1 and 2, respectively; HRESIMS m/z 837.2212, [M + Na]+ (calcd for C37H46N6O7S4Na, 837.2203).

Cell Culture

MIA PaCa-2, SK-OV-3, and HeLa cell lines were obtained from ATCC and cultured as previously described.[24,36] UACC-62 cells were obtained from AddexBio and cultured in RPMI-1640 media (Sigma-Aldrich) supplemented with 10% fetal bovine serum and 50 μg/mL gentamicin reagent (Gibco). The SK-OV-3/MDR-1-6/6 cell line is derived from a single cell clone from the SK-OV-3/MDR1 cell line obtained from S. Kane (Division of Molecular Medicine, Beckman Research Institute of the City of Hope, Duarte, CA, USA) and cultured as previously described.[23]

In Vitro Antiproliferative/Cytotoxicity Assays

The MTT assay with MIA PaCa-2 cells and the calculation of IC50 values were performed as previously described.[36] The in vitro antiproliferative and cytotoxic effects in SK-OV-3 and SK-OV-3/MDR-1-6/6 cells were tested using the sulforhodamine B (SRB) assay.[37] The GI50, TGI, and LC50 values reported were calculated from the dose–response curves for percent cell growth using GraphPad Prism 6. Percent cell growth was calculated as previously described by Boyd et al.[37] Results represent the mean ± standard error.

In Vitro Clonogenic Assays

UACC-62 cells (600) or HeLa cells (300) were allowed to adhere in 60 mm3 tissue culture dishes and subsequently treated with vehicle (DMSO) or 1. After 4 h of treatment, cells were washed with Dulbecco’s phosphate-buffered saline (DPBS) and fresh growth medium was added. After 14 days, cells were fixed and stained with 0.5% crystal violet in 10% methanol. Colonies were counted using GeneSnap software (PerkinElmer). Data were analyzed by one-way ANOVA with Dunnett’s posthoc test using GraphPad Prism 6.

Immunoblotting/Cell Lysates

HeLa cells in log-phase growth were harvested and lysed with cell extraction buffer (Invitrogen) containing protease inhibitors. Total protein concentrations were measured, and equal amounts of protein were separated by SDS-PAGE and transferred to a PVDF membrane. Membranes were probed for actin (Sigma-Aldrich), S6 kinase (Cell Signaling), P-T389-S6 kinase (Cell Signaling), or H3K9me3 (Abcam). Signals were visualized with Amersham ECL Plus (GE Health Care) in a Geliance (PerkinElmer) imaging system.

Xenograft Studies

Female athymic nude mice were obtained from Jackson Laboratories and housed in an AAALAC approved facility at the University of Texas Health Science Center at San Antonio. Mice were transplanted bilaterally with 2 × 106 UACC-62 cells injected sc into each flank. They were randomized (n = 5 mice/group) after tumors reached a median volume of 150 mm3. Compound 1 was dissolved in DMSO–Cremophor (1:1) and administered ip after diluting in 200 μL of DPBS (maximum final concentration of DMSO and Cremophor was 2.5%). Mice were weighed and examined daily for tumor growth. Tumors were measured twice weekly using calipers, and tumor volume was calculated using the following formula: volume (mm3) = length (mm) × width (mm) × height (mm). Data for final tumor volume after 14 days postimplantation were analyzed by a two-tailed t-test using GraphPad Prism 6. T/C% was calculated using the formula [mean final tumor volume (treated) – mean initial tumor volume (treated)] ÷ [mean final tumor volume (untreated) – mean initial tumor volume (untreated)] × 100.