Chlorizidine, A Cytotoxic 5H-Pyrrolo[2,1-a]isoindol-5-one-Containing Alkaloid from a Marine Streptomyces sp

Abstract

Cultivation of an obligate marine Streptomyces strain has provided the cytotoxic natural product chlorizidine. X-Ray crystallographic analysis revealed that the metabolite is composed of a chlorinated 2,3-dihydropyrrolizine ring attached to a chlorinated 5H-pyrrolo[2,1-a]isoindol-5-one. The carbon stereocenter in the dihydropyrrolizine is S-configured. Remarkably, the 5H-pyrrolo[2,1-a]isoindol-5-one moiety has no precedence in the field of natural products. The presence of this ring system, which was demonstrated to undergo facile nucleophilic substitution reactions at the activated carbonyl group, is essential to the molecule’s cytotoxicity against HCT-116 human colon cancer cells.

There are a substantial number of chemotherapeutic drugs on the market that are based on the scaffolds of actinomycete-derived natural products.¹ For instance, chemical studies of terrestrial Streptomyces bacteria led to the discovery of actinomycin D,^2a doxorubicin,^2b bleomycin,^2c carzinophilin,^2d chromomycin A3,^2e mithramycin,^2f mitomycin C,^2g sarkomycin,^2h and streptozocin,²ⁱ and neocarzinostatin.^2f No truly novel actinomycete-derived natural product structure has led to the development of a drug in recent years,¹ which at least suggests that terrestrial actinomycetes are no longer a viable source for new lead compounds. Studies of marine actinomycete bacteria, however, continue to yield unique chemical structures with anticancer activity.³ For example, salinosporamide A from Salinispora tropica is poised to enter phase II clinical trials.⁴

In an effort to identify new chemotypes for therapeutic development, Streptomyces sp. strain CNH-287 was cultivated in a seawater-based medium (20 × 1 L).⁵ Notably, the strain required seawater for growth. Amberlite resin (XAD-18) was added after the first day of cultivation. The resin was filtered and extracted with acetone after seven days, and the crude material was then fractionated on silica gel. One fraction displayed significant cytotoxicity against HCT-116 human colon cancer. A cytotoxic metabolite with a prominent UV/vis profile was isolated from this fraction using C18 reversed-phase HPLC.

It proved difficult to obtain pure compound sufficient for full spectroscopic analysis. During concentration from either organic or aqueous solutions, extensive degradation occurred. Much less decomposition was observed when solutions were kept cool and dried under a stream of nitrogen. In addition, the metabolite could be stored in dilute solutions away from light and air.

A nonzero optical rotation, [α]_D −35 (c 0.50, CH₃CN), indicated that the natural product was optically active, and a strong IR stretch at 1721 cm⁻¹ revealed the presence of a carbonyl group. Mass spectrometry data for the natural product [HRESI-FT-MS m/z (M+H)⁺ = 442.9511, 444.9481, 446.9452, 448.9422] showed a molecular ion cluster consistent with molecular formulae that include Cl₂Br or Cl₄. With 13 degrees of unsaturation, however, only C₁₈H₁₀Cl₄N₂O₃ agreed with proton and carbon NMR data.

We initially attempted to solve the structure of the natural product using 1D and 2D NMR (COSY, HSQC, HMBC) experiments (Table 1). The numbering of the molecule is shown in Figure 1. Low-field signals at δ_H 6.55 (δ_C 101.9) and 6.42 (δ_C 108.2) were conspicuous, in addition to two overlapping signals at δ_H 5.80 (δ_C 99.3 and δ_C 53.0). A spin system including a proton at δ_H 5.80 and the remaining upfield methylene protons at δ_H 3.08, 2.90, 2.84, and 2.54 was apparent in the ¹H-¹H COSY spectrum. Interestingly, the upfield proton signals from δ 2.54-3.08 exhibited complex splitting patterns due to the flexibility in the molecule (vide infra). The paucity of hydrogen atoms and the plethora of quaternary carbon atoms made complete structural elucidation by NMR problematic.

Table 1.

¹H, ¹³C, and HMBC NMR spectral data for chlorizidine A (1) (CD₃CN)

C no.	δCa	δH, mult. (J, Hz)b	HMBCb
2,3,5a,14,15c
1	101.9	6.55, br s
5	163.0
6	163.9
7	113.3d
8	157.5
9	108.2	6.42, s	5,7,9a,9b
9a	135.5e
9b	132.7e
10	53.0	5.80, ddf	6-8,11
11	32.5	2.84, m	7,12,12a
		2.54, m	6-8,10,12,12a
12	25.4	3.08, m	10,11,12a,13
		2.90, m	10,11,12a,13
12a	136.8
13	99.3	5.80, sf	12a,14

^a75 MHz.

^b500 MHz.

^cδ_C = 116.9, 113.0^d, 109.7, 106.0, 105.9.

^{d, e}Signals may be switched.

^fOverlapping signals.

Figure 1.

(−)-(S)-Chlorizidine A (1).

The structure of chlorizidine A (1) was finally determined using X-ray crystallographic techniques (Figure 2). Slow evaporation of a concentrated solution of 1 in benzene provided X-ray quality crystals. A molecule of benzene was incorporated into the crystal lattice.⁶ Additionally, the lone tertiary carbon stereocenter was assigned an S-configuration [Flack parameter −0.02(2)].

Figure 2.

ORTEP plot of 1 with benzene (co-crystallized).

Chlorizidine A (1) displays an unprecedented structure involving a nitrogen-containing carbon skeleton. The discovery of a naturally-occurring 5H-pyrrolo[2,1-a]isoindol-5-one ring system has not been previously described in the literature. Various synthetic accounts of the heterocycle, however, are well documented.⁷ The pyrroloisoindolone is connected to a dichlorinated 2,3-dihydro-1H-pyrrolizine at C-7.⁸ In the solid state, the congested region about the sp²-sp³ bond between the two ring systems forces the molecule to adopt a twisted conformation. In solution, though, 1 does not appear to exhibit atropisomerism.⁹

The phenolic substituents at C-6 and C-8 of chlorizidine A (1) could be readily functionalized. Treatment of 1 with acetic anhydride/triethylamine gave 2, and methylation with dimethyl sulfate provided 3 (Scheme 1). Acetate 2 was a stable chemical entity much less prone to degradation than 1. Interestingly, like the natural product, its proton NMR spectrum showed evidence of slow C-7/C-10 bond rotation relative to the NMR time scale. The well-resolved proton signals at C-1, C-9, C-10, and C-13 in 1 were now “doubled” in 2. The diacetate structure was confirmed using X-ray crystallography. The crystal was composed of two low-energy “twisted” conformers (see Supporting Information).

Scheme 1.

Acylation and methylation of the phenolic groups in chlorizidine A (1)

The semisynthesis of bulkier phenolic esters—isobutyrate 4, pivalate 5, and benzoate 6—was undertaken in an attempt to produce an atropiosmeric mixture, but this approach was not successful (see Scheme 1).⁶ Analysis of these derivatives was much more complex, as the ester functionalities revealed their own conformational preferences. The proton NMR spectra showed the simultaneous presence of several confomers with distinct chemical shifts (see Supporting Information).

A second metabolite that was prone to degradation, chlorizidine B (7), was isolated from culture extracts of CNH-287 (Scheme 2). A stable diacetate adduct 8 was constructed using acetic anhydride. An additional aromatic NMR signal corresponding to C-7 was noticeable. Presumably, pyrrole 7 arises from hydrolysis and decarboxylation of chlorizidine A (1). The notion that 7 is an artifact of the culturing process was substantiated by treating 1 with a pH 10 buffer composed of K₂B₄O₇, K₂CO₃, and KOH.¹⁰ Under these conditions, the C-7 carboxylic acid was observed [LRESI-MS m/z (M–H)⁻ = 459, 461, 463].¹¹ Notably, many other compounds are formed from the degradation of chlorizidine (1) under basic conditions (see Supporting Information).

Scheme 2.

Reactivity of the 5H-pyrrolo[2,1-a]isoindol-5-one in chlorizidine A (1)

The central electrophilic carbonyl group of the 5H-pyrrolo[2,1-a]isoindol-5-one moiety engages sulfur-, oxygen-, and amine-containing nucleophiles in a substitution reaction, whereby the electron-poor dichloropyrrole functions as leaving group (see Scheme 2).^5g When subjected to N-acetylcysteamine and potassium carbonate, thioester 9 was produced. Likewise, treatment of 1 with benzylamine gave amide 10 and treatment with potassium carbonate in methanol afforded ester 11. Unlike 9 and 10, ester 11 was not suitably stable for complete purification and analysis. Peracetylation of 9-11 furnished stable derivatives 12-14. A more efficient method for the synthesis of 14 was accomplished by treating diacetate 2 with aqueous NaOH in CH₃OH followed by acetylation. Certainly, the isolation of pure chlorizidine A (1) is hindered, in part, by the lability of the pyrroloisoindolone moeity toward nucleophiles.

Chlorizidine A (1) and acylated derivatives 2, 4, and 5 exhibit noteworthy activity in a colon cancer cytotoxicity bioassay (Table 2).¹² Against the HCT-116 adenocarcinoma cell line,¹³ 1 showed an IC₅₀ of 3.2-4.9 μM. Compounds 2, 4, and 5 showed similar activity. Irreversible methylation of the phenolic functionality in 1, yielding 3, rendered the compound completely inactive. Any of the series of derivatives lacking the key pyrroloisoindolone ring system (7-14) had no measurable activity, strongly suggesting that this moiety is a crucial part of the metabolite’s pharmacophore.

Table 2.

Cytotoxicity of 1-14 (μM)

	HCT116 (IC50)a
1	3.2-4.9
2	0.6-3.0
3	NSA
4	1.0-1.2
5	2.1-3.7
6-14	NSA

^aHCT-116 is a human colon cancer cell line. Positive control: etoposide (IC₅₀ = 0.49-4.9 μM).

NSA = no significant activity.

Chlorizidine A diacetate (2) was tested against the NCI’s panel of 60 tumor cell lines.¹⁴ It was modestly selective in terms of its cytotoxicity (See Supporting Information). However, against SK-MEL-5 and SK-MEL-2 melanoma cancer cells, 2 showed a pronounced LC₅₀ of 3.6 μM and 11 μM, respectively. Against MDA-MB-231/ATCC breast cancer cells, an LC₅₀ of 11 μM was also determined.

Chlorizidine A (1) has obvious structural similarity to marinopyrrole A (15), a secondary metabolite from marine-derived Streptomyces sp. CNQ-418 (Scheme 3).¹⁵ Biosynthetic precursor 16 is derived from a mixed NRPS-PKS pathway, whereby proline is loaded onto a peptidyl carrier protein, oxidized, chlorinated by an FADH₂-dependent halogenase, and subsequently extended by the PKS machinery.¹⁶ Cyclization/aromatization provides monodeoxypyoluteorin (17).¹⁷ 1,3’-Bipyrrole 15 is then formed via a novel atroposelective N,C-pyrrole coupling reaction.¹⁸ An alternative route using 16 that includes N-acylation and reduction could yield chlorizidine A (1). That 16, despite its simplicity, may be utilized to make two diverse, complex structure types is remarkable.

Scheme 3.

Proposed biosynthetic relationship of 1 to marinopyrrole A (15)

Derived from what appears to represent a new, obligate marine Streptomyces sp., chlorizidine is the first example of a natural product containing a 5H-pyrrolo[2,1-a]isoindol-5-one ring. Studies are now in progress to examine the biosynthesis and mechanism of action of these novel metabolites.