Histone Deacetylase Inhibitors as a Tool to Up-regulate New Fungal Biosynthetic Products - Isolation of EGM-556, a Cyclodepsipeptide, from Microascus sp.

Abstract

The histone deacetylase (HDAC) inhibitor suberoylanilide hydroxamic acid (SAHA) was used to turn on the biosynthesis of EGM-556, a new cyclodepsipeptide of hybrid biosynthetic origin, isolated from the Floridian marine sediment-derived fungus Microascus sp. The absolute configurations of 3 chiral centers were determined by Marfey’s derivatization. EGM-556 represents one of the few examples in which silent biosynthetic genes, encoding a new secondary metabolite, were activated by means of epigenetic manipulation of the fungal metabolome.

Marine-derived fungi¹ continue to be a source of novel bioactive metabolites, even in genera as well studied as Aspergillus² and Penicillium.³ Continuing attention on these taxa is, in part, being motivated by advances in genome sequencing,⁴ especially through genome mining.⁵ A core intent of these studies usually involves identifying biosynthetic gene clusters producing novel structures^6,7 to test the hypothesis that such approaches will allow expression of ‘silent’ metabolite pathways.^8,6a Another related strategy involves unraveling the regulatory mechanisms controlling the biosynthetic genes⁹ using insights gained from chemical genetics¹⁰ and epigenetics¹¹ experiments.

The use of small molecule HDAC (Histone Deacetylase) inhibitors¹² to perturb the fungal secondary biosynthetic machinery^10c has been rewarding and represents a proof-of-principle of ideas summarized above. Commercially available compounds such as 5-azacytidine (Aza)¹³ and suberoylanilide hydroxamic acid (SAHA),^12a, explored by several labs, have provided valuable outcomes. For example, the culturing of terrestrial Aspergillus niger in the presence of SAHA yielded the novel secondary metabolite nygerone, carrying a highly unusual 1-phenylpyridin-4(1H)-one core.^11c New oxylipins plus the unique pigments, the cladochromes, were produced by C. cladosporioides cultured in the presence of either Aza or SAHA. As another parallel result, the lunalides, octadeca-dienoic acids, were produced by cultures of Diatrype sp. grown in the presence of Aza.^11d

We used a test bed of 12 marine sediment-derived fungi (strain numbers and properties shown in Table S1, Supporting Information) to assess changes in their biosynthetic products induced by adding 10⁻⁴ M SAHA to the saltwater cultures. All fungi were isolated from shallow water sediments collected in Florida, and cultured for 18 days to reach the stationary growth phase. The group of 24 ethyl acetate crude extracts (from broths + or − SAHA), were profiled by analytical RP C18 LC-ELSD-MS, and significant changes were observed in the scans between the matched pairs. One member of this set, Microascus sp. (strain number 098059A, identified by morphological and molecular methods¹⁴), showed noteworthy changes and was chosen for further study. Shown for this sample in Figure 1, are the primary data consisting of LC-ELSD-MS traces for control (C) and SAHA-spiked (SP) cultures. Significantly, the SP sample contained a new peak at retention time 21.5 min with an m/z of 557 ([M+H]⁺) that was not present in the control culture. In addition, the ethyl acetate crude extract yields were very different between the two samples: SP = 114.6 mg vs. C = 30.6 mg. Of further note was our literature search that revealed that only ~10 secondary metabolites have been reported for Microascus and its asexual anamorph Scopulariopsis species, described as human pathogens in immuno-compromised patients.¹⁵

Figure 1.

RP LC-ELSD-MS traces showing the absence (a) and presence (b) of EGM-556 (1) in cultures of Microascus sp.

Isolation of the RT = 21.5 min substance using RP HPLC (see General Experimental Procedures, Supporting Information), was straightforward and yielded 3.5 mg of a white solid. This compound, named EGM-556 (1) with [α]²³_D = +2.60 (c 0.60, MeOH), possessed a molecular formula of C₃₀H₄₄N₄O₆, (11 unsaturation units) assigned by HR ESITOFMS data for the [M+H]⁺ ion at m/z 557.33064 (calcd. 557.33336, Δ = −4.8 ppm, see General Experimental Procedures, Supporting Information). The DEPT NMR spectrum (Table 1, and Figs S2, S3, Table S2, Supporting Information) showed a partial molecular formula of C₃₀H₄₀ and the remaining H₄N₄O₆ atoms were indirectly accounted for by four amides (NH protons at δ_H 7.60, 7.74, 8.10, 8.56, see Fig S1, Supporting Information) and one ester (C=O)-Z (see five peaks δ_C 168.7 - 173.0). These functionalities comprised five unsaturation units with five more from an mono-substituted benzene ring (δ_H 7.44 (2H, d), 7.36 (2H, dd), δ 7.28 (1H, t)) and an E-disubstituted conjugated double bond (δ_H 6.28 (dd, J = 15.9, 7.1 Hz, 1H), δ_H 6.68 (d, J = 15.9 Hz, 1H)). Collectively, these observations meant that an additional ring must also be present as a macrocyclic depsipeptide. Using all of the preceding structural constraints and formulas as input, extensive dereplication searches were carried out but yielded no structural matches.

Table 1.

NMR Assignments for EGM-556 (1) in DMSO-d₆

structural unit	position	13Ca δ, typeb	1H δ (m, J(Hz), int)c
Val-1	CO	168.7, qC
	α	58.2, CH	4.24 (dd, 9.6, 6.0, 1H)
	β	28.5, CH	2.25 (ds, 6.0, 6.6, 1H )
	β-CH3	18.9, CH3	0.89 (d, 6.6, 3H)
	β-CH3	19.3, CH3	0.89 (d, 6.6, 3H)
	NH		7.60 (d, 9.6, 1H)
Ile-2	CO	170.4, qC
	α	57.2, CH	4.26 (d, 9.6, 1H)
	β	35.6, CH	2.05 (m, 1H)
	γ	23.6, CH2	1.26 (m, 2H)
	β-CH3	15.5, CH3	0.81 (d, 6.6, 3H)
	γ-CH3	11.9 CH3	0.81 (t, 7.5, 3H)
	NH		8.10 (d, 9.6, 1H)
Ala-3	CO	173.0, qC
	α	48.8, CH	4.32 (dq, 6.3, 6.4, 1H)
	α-CH3	16.3, CH3	1.18 (d, 7.2, 3H)
	NH		8.56 (d, 6.3, 1H)
Val-4	CO	172.3, qC
	α	57.5, CH	4.14 (dd, 8.7, 7.2, 1H)
	β	29.8, CH	1.96 (ds, 6.9, 6.8, 1H)
	β-CH3	19.6, CH3	0.89 (d, 6.6, 3H)
	β-CH3	18.2, CH3	0.89 (d, 6.6, 3H)
	NH		7.74 (d, 8.7, 1H)
Hyppa-5	1	168.7, qC
	2	40.1, CH2*	2.42 (dd. 14.4, 1.8, 1H)
			2.88 (dd, 14.4, 10.5, 1H)
	3	72.9, CH	5.50 (ddd, 10.5, 7.1, 1.8, 1H)
	4	126.7, CH	6.28 (dd, 15.9, 7.1, 1H)
	5	132.4, CH	6.68 (d, 15.9, 1H)
	6	135.7, qC
	7	126.5, CH	7.44 (d, 7.2, 2H)
	8	128.7, CH	7.36 (dd, 7.5, 7.2, 2H)
	9	128.1, CH	7.28 (t, 7.5, 1H)

^aMeasured at 150 MHz.

^bCarbon type established from ¹³C DEPT and/or gHMQC experiments.

^cMeasured at 600 MHz.

^*Overlapping with DMSO-d₆

The 2D COSY NMR results clearly showed five separate proton spin systems consistent with the substructures I-V drawn in Scheme 1. The presence of seven methyl groups plus the ¹H-¹H COSY NMR data accounted for two valines (Val-1, Val-4), alanine (Ala-3), and isoleucine (Ile-2). The remaining substructure V, 3-hydroxy-5-phenyl-4(E)-pentenoic acid, Hyppa, was visualized from the 2D NMR correlations shown in Scheme 1. The inter-substructural gHMBC correlations, also shown in Scheme 1, unambiguously indicated the sequence Val-Ile-Ala-Val. The insertion of the Hyppa (V) substructure was complicated by the accidental isochrony of two carbonyl groups at δ_C 168.7 (due to Val and Hyppa arrays, indicated in red and blue in Scheme 1). The gHMBC correlations between protons at δ_H 4.24, 2.25 (Val-1), 4.14, 7.74 (Val-4), 5.50, 2.88 and 2.42 (Hyppa-5) and these carbonyls were consistent with two possible ways to dock V into the tetrapeptide as shown by structural candidates 1 and 1′ (Scheme 1). The gray carbonyl group in substructure IV is common to a carbonyl group in V, either in red, resulting in alternative structure 1 or in blue resulting in alternative structure 1′. Fortunately, the latter structure was quickly ruled out because it contains a carbamate and a ketone whose diagnostic ¹³C shifts of δ_C ~157 and δ_C ~200 respectively, were not observed.

Scheme 1.

Substructures, alternative final structures, and key 2D NMR correlations. Accidentally isochronous carbonyl groups occurred in 1 at δ_C 168.7, indicated in red, gray and blue.

The pathway to further confirm the structure proposed of 1 along with efforts to resolv the configurations at the six chiral centers was approached in the following way. Hydrolysis of 1 (1.5 mg, 1N NaOH) followed by esterification with diazomethane¹⁶ afforded the acyclic methyl ester 2, [α]²³_D = −1.36 (c 1.30, MeOH) shown in Scheme 1. Its molecular formula of C₃₁H₄₈N₄O₇ was confirmed by HR ESITOFMS, based on the [M+Na]+ ion m/z 611.34268 (calcd. 611.34152, Δ = 1.9 ppm, see General Experimental Procedures, Supporting Information) and as expected was one unsaturation unit less than that of 1. The NMR data for 2 (See Table S4, Supporting Information) were consistent with the expected sequence and all five carbonyl resonances were anisochronous, which made it possible to unequivocally confirm the juxtaposition of Hyppa-5 and Val-4 through an amide bond.

Acid mediated hydrolysis of 1 (1 mg, 6M HCl) followed by derivatization with Marfey’s reagent¹⁷ and subsequent C18 LC-ELSD-MS analysis of the adducts, was the approach taken to deduce the absolute configurations of I-IV. The overall results indicated the constituent amino acids were: (R)-Val, (S)-Val, (2S, 3S)-Ile, and (S)-Ala. Retention times of the standard amino acids under our analysis conditions were as follows: (R)-Ala 59.7 min., (S)-Ala 64.4 min., (S)-Val 68.9 min., (R)-Val 75.4 min., (2S, 3S)-Ile 74.7 min., (2R, 3R)-Ile 81.3 min (See also Fig S13, Supplemental Information). Since both (R)-Val and (S)-Val were observed, EGM-556 (1) is one of four possible structures: cyclic ((R)-Hyppa-5)-((R)-Val-4)-((S)-Ala-3)-((2S, 3S)-Ile-2)-((S)-Val-1), cyclic ((R)-Hyppa-5)-((S)-Val-4)-((S)-Ala-3)-((2S,3S)-Ile-2)-((R)-Val-1), cyclic ((S)-Hyppa-5)-((R)-Val-4)-((S)-Ala-3)-((2S, 3S)-Ile-2)-((S)-Val-1) or cyclic ((S)-Hyppa-5)-((S)-Val-4)-((S)-Ala-3)-((2S, 3S)-Ile-2)-((R)-Val-1).

Reverse phase chromatography did not differentiate between (2S, 3S)-isoleucine and (2S, 3R)-isoleucine and these amino acids were separated using a chiral column (Chirex 3010, Phenomenex, 4.1 × 150 mm).¹⁸ Solvent A was 0.1 M aqueous ammonium acetate, pH = 3 with TFA and solvent B acetonitrile. The gradient used was 20 – 75 % solvent B in A over 50 min, 1 mL/min). Retention times of amino acids under these conditions were: (2S, 3S)-Ile 23.0 min., (2R, 3R)-Ile 24.3 min., (2S, 3R)-Ile 22.3 min., (2R, 3S)-Ile 23.5 min. (See also Fig. S14, Supporting Information). Our results showed that an (2S, 3S)-Ile is present.

Another approach considered, to assign the unresolved configuration of 1 at C3 of Hyppa-5, involved attempts to make a Mosher ester derivative¹⁹ of the alcohol moiety of 2. However, all of our experiments were unsuccessful. The configuration is therefore left unassigned since there is precedent for both (R) and (S) in similar compounds in the literature, as discussed below.

As an important additional note, we utilized MSⁿ data to also address the subunit sequence for both 1 and 2. The complex array of fragmentations observed for 1 was consistent with principles outlined in the literature for cyclic depsipeptides.²⁰ By contrast, the pattern observed for 2 was quite distinct. These contrasting outcomes will be fully described elsewhere.

The EGM-556 (1) substructural units are likely derived from a HPN (Hybrid PKS/NRPS) biosynthetic pathway. Moreover, the 16 atom peptolide core of EGM-556 (1) is rare; the only other examples are: the antimicrobial unnarmicins (3) from a marine-derived Photobacterium sp. MBIC06485²¹ and the histone deacetylase inhibitory/antitumor active FK228 (FR901228, 4) from Chromobacterium violaceum No. 968.²² Side-by-side comparison of 1, 3, and 4 (see Figure 2) reveals some additional similarities and contrasts. The ketide portions of all three contain β-hydroxy amides; and, analogous to the situation of 1, FK228 (4) possess (R)-Val and (S)-Val units in close proximity to the PKS side chain. The mono substituted benzene ring-containing side chain of 1 is analogous to similar functionality present in the cryptophycins, especially -26, isolated from Nostoc strains of cyanobacteria.²³ Rigorous biosynthetic analysis²⁴ has shown the phenyl-octenoic polyketide unit of cryptophycin-26 is likely derived from a phenylalanine (via trans-cinnamic acid), which is clearly present in 1. Finally, the 3(R)-hydroxy-5-phenyl-4(E)-pentenoic acid (5) coded as Hyppa in 1, as shown in Figure 3, has one precedent. The free acid was first isolated from the Chinese plant Avicennia marina together with five new iridoid glucosides, and one marinoid E (6) also contained the (3R)-Hyppa unit.²⁵

Figure 2.

Unnarmicin C (3) from marine Photobacterium sp.²¹ and FK228 (4) from C. violaceum.²²

Figure 3.

3(R)-hydroxy-5-phenyl-4(E)-pentenoic acid (Hyppa) (5) and marinoid E (6) isolated from A. marina.²⁵

Our successful initial experiments with the 12 marine-derived fungi explored here, extend previous parallel results where spiking cultures with the potent actin inhibitor Jasplakinolide^8b upregulated different biosynthetic products. The discovery of EGM-556 (1) represents an exciting outcome of expressing hitherto silent HPN biosynthetic genes. In the future, we will report on parallel outcomes based on additional HDAC spiking experiments to extend the biosynthetic products from well-studied fungi including Aspergillus and Penicillium, but isolated from marine sources.