Inhibition of Tumor Cells Interacting with Stromal Cells by Xanthones Isolated from a Costa Rican Penicillium sp.

Abstract

CR1642D, an endophytic isolate of Penicillium sp. collected from a Costa Rican rainforest, was identified through a high-throughput approach to identify natural products with enhanced anti-tumor activity in the context of tumor-stromal interactions. Bioassay-guided separation led to the identification of five xanthones (1-5) from CR1642D. The structures of the xanthone dimer penexanthone A (1) and monomer penexanthone B (2) were elucidated on the basis of spectroscopic analyses, including 2D NMR experiments. All of the compounds were tested against a panel of tumor cell lines in the presence and absence of bone marrow stromal cells. Compound 3 was the most active, with IC₅₀ values of 1~17 μM, and its activity was enhanced two-fold against tumor cell line RPMI8226 in the presence of stromal cells (IC₅₀ 1.2 μM, but 2.4 μM without stromal cells).

Identification of agents for the treatment of cancer typically involves target-based screens or phenotypic screens against isolated cancer cell lines. However, tumor cells in patients interact with diverse types of non-malignant cells, which can stimulate tumor cell survival, proliferation and resistance to diverse classes of established anticancer drugs and investigational agents. Recently, we have also shown that the presence of non-malignant accessory cells, such as bone marrow stromal cells (BMSCs), can sensitize tumor cells to other novel agents.¹ Therefore, in vitro screening for novel anticancer agents has to be performed in cocultures of tumor cells with non-malignant accessory cells, in order to detect whether an agent is subject to microenvironment-dependent drug resistance or sensitization. In our continuing search for biologically active natural products from Costa Rican tropical rainforests as part of an International Cooperative Biodiversity Group (ICBG) program,² we screened endophytic fungal extracts from Costa Rica at Harvard Medical School’s high throughput screening facility (ICCB-L) against a panel of cancer cell lines with and without BMSCs (Scheme 1). One extract, designed as CR1642D and identified as being from Penicillium sp., had enhanced anti-tumor activity against MM.1S with 40.2% viability in the presence of BMSCs, but 66.8% without BMSCs at 1 μg/mL. Bioassay-guided fractionation using C₁₈ open column and phenyl-hexyl HPLC yielded five xanthones (1-5). Herein we report the structure elucidation of compounds 1 and 2 and the anti-tumor properties of compounds 1-5.

Scheme 1.

High-throughput screening to identify compounds active against tumor cells interacting in vitro with bone marrow stromal cells.

Compound 1 had the molecular formula C₃₆H₃₆O₁₅ by HRESIMS. The UV spectrum of 1 was very similar to that of phomoxanthone B, a xanthone dimer.³ The IR spectrum of 1 supported the presence of carbonyl groups (1744 and 1605 cm⁻¹). The proton NMR spectrum of 1 showed two aromatic AB spin systems, six protons bound to oxygenated carbons, two methines, two methylenes, three acetyl methyl groups, and two chelated phenolic hydroxyl groups. In ring A of monomer I, H-4 exhibited ³J HMBC correlations to C-2 and C-9a. Peri to a carbonyl group, the C-1 hydroxyl group correlated to C-1, C-2 and C-9a. H-3 was ortho coupled with H-4, and showed ³J HMBC correlations to C-1, C-4a, and C-4′. Therefore ring A was determined as a 2,3,6-trisubstituted phenol (2,4a,9a-trisubstituted phenol). H-2′ and H-3′ showed ³J HMBC correlations to C-4′ and C-9a’, and C-1′ and C-4a’, respectively, hence ring A’ in monomer II was determined as a 2,3,4-trisubstituted phenol (4′,4a’,9a’-trisubstituted phenol), and rings A and A’ were connected through the C-2-C-4′ bond. Since 1-OH was chelated and C-4a was oxygenated, it could be deduced that A/A’ joined C/C’ to form a 4-chromanone.

The remaining parts of the molecule to be determined are rings B and B’. The methyl group at δ_H 1.07 (H₃-11) showed HMBC correlations to C-5 (δ_C 72.1), C-6 (28.0), and C-7 (33.7). The oxygenated methylene at δ_H 3.94 and 3.75 (H₂-12) correlated to C-10a (δ_C 80.7), C-5, and C-8a (101.5), while the oxygenated methine at δ_H 5.55 (H-5) exhibited correlations to C-10a, C-8a, C-6, C-7, C-12 (δ_C 65.6), C-11 (17.9), and 5-OCOCH₃ (170.9). Therefore, this ring was determined to be 4-acetoxyl-3-hydroxymethyl-5-methyl cyclohexenol with the other two substituents at 2,3-positions of this fragment, which could be ring B or ring B’. Similarly, the last ring was determined to 4-acetoxyl-3-acetoxymethyl-5-methyl cyclohexenol with the other two substituents at 2,3-positions of this fragment, which could be ring B’ or B. The only difference between ring B and ring B’ is the substituent at the oxygenated methylene - one is a hydroxyl while another is an acetoxy. A ROESY experiment showed that the oxygenated methylene (12-position) was on one face of the hemi-chair ring while both the methyl (11-position) and the acetoxyl group (5-OAc) were on the other face (Figure 2). Molecular modeling with Chem3D Ultra (9.0) suggested that the substituent at the 10a-position was far away (> 5 Å) from the aromatic protons of ring A. In the NOESY spectrum of compound 1, both 1–OH (δ_H 11.62) and H-3 (δ_H 7.16) correlated with the signal at δ_H 1.83 (12′-OCOCH₃), and H-4 (δ_H 6.43) showed correlation with the signal at δ_H 2.11 (5-OCOCH₃). Since the signals at δ_H 1.83 (12′-OCOCH₃) and 2.05 (5′-OCOCH₃) were in the same ring system, the signal at δ_H 2.11 must be the 5-acetoxyl methyl. Hence, the structure of compound 1 including its relative stereochemistry was determined as shown. Both monomers were assigned the relative stereochemistry (R*,R*,R*) based on the assumption that they share the same biosynthetic pathway. All the ¹³C NMR chemical shifts except C-9 and C-9′ were obtained from the HSQC and HMBC spectra due to insufficient sample for a good ¹³C NMR spectrum. The planar structure of 1 was reported in a Korean patent, but with an unclean ¹H NMR spectrum, which most likely resulted from an impure sample.⁵

Figure 2.

key NOSEY correlations of 1

The molecular weight of compound 2 was 376.1158 as suggested by HRESIMS, which was 1 unit more than that of monomer II of compound 1. The proton NMR data of compound 2 were similar to those of monomer II of compound 1, but with an extra aromatic proton signal. The planar structure and relative stereochemistry of compound 2 were determined by 2D NMR experiments as shown. A structure identical to compound 2 was used for CD calculation, but never isolated.⁶

Compounds 3-5 were identified as dicerandrols B, A, and C, respectively, by comparison of physical and spectroscopic data (UV, IR, ¹H NMR, [α]_D, and MS) with literature values.⁴ⁱ These were previously purified from the endophytic fungus Phomopsis longicolla isolated from an endangered mint from Florida.⁴ⁱ Compounds 4 and 5 are homodimers with a 2—2′ connectivity. The heterodimer 3 has the same two monomers as 1, but they are dimerized through a 2—2′ linkage, while the two different monomeric units of 1 are coupled through the 2—4′ positions. Many such xanthone dimers with a methyl at the 6-position and an oxygenated methylene at the 10a-position have been reported, but most of them are symmetrical homodimers as in 4 and 5.⁴ Usually, the unsymmetrical xanthone heterodimers are different only at either ring A/A’ like phomoxanthone B,³ or ring C/C’ as in 3. Curiously the monomeric units in these xanthone dimers (1, and 3-5) had not been previously isolated from natural sources.

All five pure compounds were tested against a panel of tumor cell lines (from myeloma, lymphoma, leukemia, as well as breast and prostate cancer) in the presence and absence of BMSCs. As expected, the activity was either decreased or enhanced in the context of tumor-stromal interactions. As can be seen in Table 2, the cytotoxic activity of these compounds increases upon dimerization and acetylation of 12′-OH (4) to 12′-OAc (3). The C-12 and C-12′ diacetate (5) exhibits less activity than 3, and the 2-4′ linked monoacetate (1) is much less active than the 2-2′ linked monoacetate (3). The most active compound, 3, exhibits moderate activity against Dox40, Farage, H929, HT, OPM2 and RPMI8226 in the presence of stromal cells with IC₅₀ values of 2.3, 1.3, 3.4, 1.3, 1.5, and 1.2 μM, respectively. The activity of 3 against cancer cell lines RPMI8226 and H929 is doubled or tripled in the presence of stromal cells - results that would not have been identified through a traditional cell based screen. On the other hand, compound 3 is much less toxic against human immortalized non-malignant cells, such as HS-5 bone marrow stromal cells, HOBIT osteoblast-like cells, THLE-3 hepatocytes, and SVGp12 astrocytes, with IC₅₀ values of 13.0, 9.2, 10.0, and 13.7 μM, respectively (Figure 2S). The relative selectivity of these compounds, especially compound 3, warrants further investigation.

Table 2.

Assay results of compounds l~5^a

		1			2			3			4			5
	bIC50 (μM)	cIC50 (μM)	dDiff	bIC50 (μM)	cIC50 (μM)	dDiff	bIC50 (μM)	cIC50 (μM)	dDiff	bIC50 (μM)	cIC50 (μM)	dDiff	bIC50 (μM)	cIC50 (μM)	dDiff
Dox40e	15	21.6	6.7	11.6	24.2	12.6	2.2	2.3	0.1	22.8	36.9	14.1	10	13.9	3.9
						-
Faragef	12.4	10.5	−1.9	199.8	38.7	161.1	1.6	1.3	−0.3	15.6	14.8	−0.8	3.5	3.3	−0.3
H929e	54.5	35.2	−19.3	187.5	94.8	−92.7	10.2	3.4	−6.8	42.5	22.9	−19.6	10.8	5.4	−5.4
HTf	10.8	9.9	−0.9	100.0	34.9	−65.2	1.2	1.3	0.1	ND	ND		2.1	4.0	1.9
KMS34e	22.6	55.6	33	152.9	199.1	46.1	5.5	9.3	3.8	ND	ND		ND	ND
KU812Fg	14.6	24.2	9.5	44.9	95.9	51.0	2.7	3.6	0.9	24.6	46.6	22	7.1	14.3	7.2
						-
L363e	66.0	79.4	13.4	464.2	198.6	265.6	16.7	9.4	−7.3	ND	ND		ND	ND
MDA-MB- 231h	43.4	34.2	−9.3	163.9	118.5	−45.4	8.5	5.5	−2.9	43.3	38.4	−4.9	18.1	13.3	−4.8
MM1Se	37.1	47.1	10.0	168.5	185.6	17.1	7.3	8.0	0.7	28.9	33.3	4.4	14.7	20.8	6.1
OCILY17Rf	31.1	47.4	16.3	1616	293.4	−1323	4.6	5.2	0.6	ND	ND		ND	ND
OCIMY5e	18.8	47.7	28.8	158.0	383.9	225.8	2.8	5.7	2.9	ND	ND		ND	ND
OPM2e	12.7	10.5	−2.2	42.8	44.8	1.9	1.9	1.5	−0.3	ND	ND		4.8	7.5	2.6
						-
PC3i	122.3	66.3	−56.0	1382.	216.0	1166.	34	14.1	−19.9	100.	71.7	−28.4	121	45.2	−75.4
RPMI8226e	12.9	8.9	−4	71.6	34	−37.6	2.4	1.2	−1.2	ND	ND		4.8	2.8	−2.0

^aActivity of five compounds from the PeniciUium expansion extract across various tumor types in the presence and absence of bone marrow stromal cells. Tumor cells of various types were cultured in the presence and absence of HS-5 stromal cells treated with increasing concentrations of compounds. IC₅₀ was calculated both in the presence and absence of stroma.

^bNo stroma.

^cPlus Stroma.

^dDifference in IC₅₀ values.

^eMyeloma.

^fLymphoma.

^gLeukemia.

^hBreast.

ⁱProstate.

EXPERIMENTAL SECTION

General Experimental Procedures

All NMR experiments were carried out on a Varian INOVA 600 MHz spectrometer. IR spectra were measured on a Bruker Alpha-P spectrometer and UV spectra on an Amersham Biosciences Ultrospec 5300 pro spectrophotometer. All the compounds (1-5) were purified from CR1642D on an Agilent 1100 series HPLC (Agilent Technologies) using a semi-preparative Phenomenex Luna Phenyl-hexyl column (Luna, 25 cm × 10 mm, 5 μm particle size) and a Phenomenex Luna C₁₈ HPLC column (250 × 10 mm, 5 μm particle size). Optical rotations were obtained using a Jasco Polarimeter.

Culturing

The isolated strain CR1642D is deposited at INBio, Costa Rica. Agar plugs of CR1642D were initially grown at 25 °C on yeast malt agar plates supplemented with 30 μg/mL streptomycin and 12 μg/mL chlortetracycline. After one week, 3 macerated agar plugs from these plates were placed in 75 mL of rich seed medium [tryptone peptone (5 g/L), dextrose (10 g/L), yeast extract (3 g/L), and malt extract (10 g/L)] in a 125 mL erlenmeyer (×2) with a pH value of 6.2. It was grown at 25 °C and 150 rpm for 6 days. 150 mL of 0.66% (w/v) malt extract and 5 g HP-20 resin were then added to each Erlenmeyer (250 mL each, ×8), which was inoculated with 15 mL of the rich seed media, and the fungi were cultured under the same conditions for 16 days. The fungal cultures were then held at 25 °C without shaking for 5 days.

Sequencing and species identification

For identification by internal transcribed spacer (ITS) sequencing, CR1642D was cultured in the above-mentioned rich seed medium (see the “Culturing” section) for 6 days. Mycelia were then retrieved by filtration and ground to a fine powder in liquid N₂. Genomic DNA was extracted using the SurePrep™ RNA/DNA/Protein Purification Kit (Fisher Bioreagents), and large subunit rDNA was amplified by PCR using primers LR5 (5′-TCCTGAGGGAAACTTCG-3′) and LROR (5′-ACCCGCTGAACTTAAGC-3′). PCR products were sequenced at Genewiz (http://www.genewiz.com/). The DNA sequence data obtained from the fungal strain CR1642D have been deposited at GenBank with accession number JQ778844.

Extraction and separation

The cultures were filtered, the mycelial mat and HP-20 were extracted with 90% EtOH three times, and the extract was concentrated under vacuum. The crude extract (1.2 g) of CR1642D in 90% H₂O-MeOH was passed through a C₁₈ SPE and then washed with MeOH. The MeOH wash was evaporated to dryness on a rotary evaporator, redissolved and fractionated using a phenyl-hexyl column (2 mL/min; 70% CH₃CN for 20 min then to 100% CH₃CN in 10 min). Fractions 2, 4, 5, and 7 yielded compounds 4 (SC2-30-2, t_R 13.5 min, 1.0 mg/L), 3 (SC2-30-4, t_R 19 min, 19 mg/L), 1 (SC2-30-5, t_R 23.5 min, 1.3 mg/L), and 5 (SC2-30-7, t_R 27 min, 0.8 mg/L), respectively. Further purification of the most polar fraction, fraction 1, using a Phenomenex Luna C₁₈ HPLC column (2 mL/min; 60% CH₃CN for 30 min then to 100% CH₃CN in 10 min) yielded compound 2 (SC2-32-3, t_R 22.5 min, 0.4 mg/L).

Penexanthone A (1)

yellow powder; [α]²³_D −36 (c 0.1, MeOH); UV (MeOH) λ_max (log ε) 212 (4.23), 242 (sh), 337 (4.30) nm; IR (film) ν_max 3419, 2966, 2934, 1744, 1605, 1562, 1435, 1410, 1373, 1324, 1229, 1048 cm⁻¹; ¹H NMR (600 MHz, CD₃CN): δ 7.30 (1H, d, J = 7.8 Hz, H-3′), 7.18 (1H, d, J = 8.4 Hz), H-3, 6.50 (1H, d, J = 7.8 Hz, H-2′), 6.40 (1H, d, J = 8.4 Hz, H-4), 5.65 (1H, s, H-5), 5.35 (1H, s, H-5′), 4.53 (1H, d, J = 13.2 Hz, H-12α’), 3.99 (1H, d, J = 12.6 Hz, H-12α), 3.90 (1H, d, J = 12.6 Hz, H-12β’), 3.60 (1H, d, J = 13.2 Hz, H-12β), 2.30-2.50 (6H, m, H-6, H-6′, H₂-7, H₂-7′), 2.06 (3H, s, 5-OCOCH₃), 2.03 (3H, s, 5′-OCOCH₃), 1.76 (3H, s, 12′-OCOCH₃), 1.03 (3H, d, J = 6.6 Hz, CH₃-11), 0.95 (3H, d, J = 6.6 Hz, CH₃-11′); ¹H NMR (600 MHz, CDCl ) and ¹³C NMR (150 MHz, CDCl₃) see Table 1; HRESIMS m/z 709.2126 ([M+H]⁺ calcd for C₃₆H₃₇O₁₅, 709.2132).

Table 1.

¹H^a and ¹³C^b NMR Data of Compounds 1 and 2

		1			2
#	δC, type (I)	δC, type (II)	δH (J in Hz) (I)	δH (J in Hz) (II)	δC, type	δH (J in Hz)
1, 1’	160.2, C	161.9, C			162.0, C
2, 2’	118.6, C	110.3, CH		6.58, d (8.4)	108.1, CH	6.35, d (8.4)
3, 3’	139.4, CH	139.4, CH	7.16, d (8.4)	7.32, d (8.4)	139.4, CH	7.39, t (8.4)
4, 4’	107.6, CH	117.4, C	6.43, d (8.4)		109.7, CH	6.47, d (8.4)
4a, 4a’	157.6, C	155.0, C			158.2, C
5, 5’	72.1, CH	69.8, CH	5.55, br s	5.30, br s	70.6, CH	5.90, br s
6, 6’	28.0, CH	27.9, CH	2.47, m	2.28, m	27.5, CH	2.57, m
7, 7’	33.7, CH2	33.3, CH2	2.30~2.50, m	2.30~2.50, m	33.4, CH2	2.57, m; 2.37, m
8, 8’	177.9, C	177.9, C			179.5, C
8a, 8a’	101.5, C	100.8, C			100.3, C
9, 9’	−, C	−, C			−, C
9a, 9a’	106.8, C	106.7, C			106.5, C
10a, 10a’	80.7, C	80.3, C			80.8, C
5,5’- OCOCH3	170.9, C 3	170.8, C			169.8, C
5,5’- OCOCH3	21.3, CH3	21.3, CH3	2.11, s	2.05, s	20.3, CH3	2.03, s
11, 11’	17.9, CH3	17.6, CH3	1.07, d (6.0)	0.97, d (6.0)	17.0, CH3	1.03, d (6.0)
12, 12’	65.6, CH2	64.2, CH2	3.94, m	4.52, d (12.0)	65.0, CH2	4.51, d (12.0)
			3.75, m	3.86, d (12.0)		4.27, d (12.0)
12,12’- OCOCH3		170.1, C			169.5, C
12,12’- OCOCH3		21.1, CH3		1.83, s	20.1, CH3	1.99, s
1,1’-OH			11.62, s	11.20, s

^a† (ppm) 600 MHz; multiplicities; J values (Hz) in parentheses.

^b† (ppm) 150 MHz; chemical shifts from gHSQC and gHMBC.

Penexanthone B (2)

yellow powder; [α]²³_D +2.5 (c 0.08, MeOH); UV (MeOH) λ_max (log ε) 208 (3.89), 222 (sh), 267 (sh), 276 (3.28), 332 (3.90) nm; IR (film) ν_max 3420, 1741, 1602, 1441, 1230, 1044 cm⁻¹; ¹H NMR (600 MHz, acetone-d₆) and ¹³C NMR (150 MHz, acetone-d₆) see Table 1; HRESIMS m/z 377.1243 ([M+H]⁺ calcd for C₁₉H₂₁O₈, 377.1236).

Biological Assay

As previously described,¹ stromal cells were plated and allowed to seed overnight for non-adherent tumor experiments. The following day, tumor cells were overlaid and treated with natural product extracts. Following 48 hrs of incubation, luciferin substrate was added, cultures incubated for 30 min and bioluminescence signal read on an Envision or Luminoskan luminometer. For adherent tumors, stromal cells were counted and plated with tumor cells, both allowed to seed overnight and treated with extracts the following day. Cultures were again incubated for 48 h in the presence of extracts and bioluminescence signal read on a luminometer following addition of luciferin substrate. Conditions were normalized to each respective non-treated control and each condition was performed in quadruplicate. High-throughput screening of natural product extract libraries was performed by screening MM.1S multiple myeloma (MM) tumor cells in the presence and absence of HS-5 bone marrow stromal cells (BMSCs). Natural product hits were identified as those extracts which resulted in enhanced anti-tumor killing in the presence of BMSCs compared to their absence, with minimal cytotoxic activity against the HS-5 BMSCs alone. Tumor cells were co-cultured with stromal cells for 48 h unless otherwise stated.

Figure 1.

Key HMBC correlations in monomer I of 1

Chart 1.

Structures of compounds 1-5