Antimicrobial isoflavans and other metabolites of Dalea nana

Abstract

The phytochemical investigation of extracts from Dalea nana roots and aerial parts led to the isolation of thirteen phenolic compounds. Three previously undescribed isoflavans, named verdeans A-C (1, 3, and 7), were characterized. Two additional isoflavans (2 and 5) were previously undescribed enantiomers of known compounds. A previously undescribed isoflavone (verdean D, 10) was found, and the known specialized metabolites, isoflavans 4, 6, 8, and 9, isoflavone 11, flavone 12, and a 2-arylbenzofuran 13, were also isolated. All but one (7) of the isoflavans were prenylated. The structures of the previously undescribed compounds were deduced by NMR spectroscopy, supported by HRESI mass spectrometry. The absolute configurations of 1-3, 5, and 7-9 were determined by ECD. Compounds 1, 3, 4, 6, and 8 exhibited in vitro antimicrobial activities, causing complete growth inhibition (MIC) at concentrations between 6.7–37.0 µM against Cryptococcus neoformans and between 8.9–25.0 µM against methicillin resistant Staphylococcus aureus (MRSA). The most broadly active previously undescribed compound was verdean A (1), with MIC values of 6.7 and 12.9 µM toward C. neoformans and MRSA, respectively, and an MIC of 10.0 µM against the often-intractable C. albicans.

Graphical Abstract

1. Introduction

Dalea nana Torr. ex A. Gray (Fabaceae) (Fig. 1S), the dwarf prairie clover (Kearney and Peebles, 1960), has not been studied previously for its chemical content. A recent report from our laboratory on metabolites of D. jamesii (Torr.) Torr. & A. Gray (Belofsky et al., 2023) revealed a suite of previously undescribed isoflavans and two unique 2-arylbenzofurans. D. nana, D. jamesii, and a third species, D. wrightii A. Gray, are taxonomically closely related and said to be “sympatric in western Texas and adjoining states” (Barneby, 1978). One may speculate that speciation of these occurred in a realtively recent timeframe in a shared geographic space. D. nana differs from D. jamesii in that it possesses pinnately, rather than ternately compound leaves. It has a measurably shorter calyx tube (2–2.5 mm long) and, while both have yellow flowers, those of D. nana age to a more reddish color compared to D. jamesii (Ackerfield, 2015). These fine distinctions do not diminish the close taxonomic relationship of the two plants, and both produce predominantly isoflavonoid metabolites (Belofsky et al., 2023).

Isoflavans are relatively rare, even within the subclass of isoflavonoids. A recent review stated that “in contrast to other groups of flavonoids, the occurrence of isoflavonoids in the plant kingdom is relatively sparse,” and from 2012 to 2017 for example, isoflavans comprised only 4.4% of the 391 previously undescribed isoflavonoids reported in the literature (Al-Maharik, 2019).

Our laboratory has reported antimicrobial activity for a variety of isoflavans, but compounds from other classes of flavonoids have shown comparable low-micromolar in vitro activities (as IC50 or MIC values) against microbial pathogens (Fahmy et al., 2018). Structure activity relationship (SAR) questions are raised by this observation, but reliable trends are difficult to elucidate. One seeking to determine the influence of the C-4 carbonyl group of isoflavones, absent in isoflavans, or simply the effect of the position of ring B at C-2 (flavans) compared to C-3 (isoflavans) will encounter limited success. Trends in SAR vary dramatically due to differences in pathogen strains and across testing laboratories, making comparisons challenging. While much more work in this area is needed, there has, for example, been some detailed work (Kiyama, 2023) on two well-known members of this class, equol and in particular, glabridin (Simmler et al., 2013).

This report describes the phytochemical investigation of D. nana leading to the identification of four previously undescribed metabolites, along with two compounds that are previously undescribed by virtue of their absolute configurations. Five of the thirteen isolated metabolites exhibited moderate (MIC ≤25.0 µM) antimicrobial activity against methicillin-resistant S. aureus, with additional activity observed toward biomedically relevant yeast pathogens.

2. Results and discussion

2.1. Structure determination

Open-column chromatography of the D. nana methanolic extracts led to compounds 1-13. The initial separation by vacuum liquid chromatography over silica gel was followed by Sephadex LH-20 chromatography, and successive stages of linear- and step-gradient silica gel chromatography (See section 4.3).

Compound 1 had the molecular formula C₂₅H₃₀O₄ by HRESIMS (obsvd. [M+H]⁺ m/z 395.2215, calcd. For C₂₅H₃₁O₄, 395.2217). The ¹H, ¹³C, and DEPT NMR spectra revealed patterns of signals characteristic of a di-prenylated isoflavan possessing 25 carbons (Table 1). These included signals for the oxygenated C-2 methylene (δ_C 70.8), a C-3 methine (δ_C 32.9) and a C-4 methylene (δ_C 31.9). The attached protons at these positions (H-2_ax, H-2_eq, H-3, H-4_ax, and H-4_eq; Table 1) were established by HSQC correlations, and their mutual couplings in COSY NMR (Fig. 6S) confirmed these observations. The two 3-methylbut-2-en-1-yl (prenyl) groups were recognized by their characteristic methylene and methine signals with, for example, H₂-1′′ (δ_H 3.23, d) and H-2′′ (δ_H 5.31, br t) assigned for one of these two units. HMBC correlations from these same protons to the carbons, C-3′′ (δ_C 131.7), C-3′′-Me (δ_C 17.9), and C-4′′ (δ_C 26.0) verified the structure of this prenyl group (Table 1). The ¹H and ¹³C NMR signals for the second prenyl group were very similar to those of the first. The methine protons of both were distinctly seen, side-by-side (at δ_H 5.25, br t and (δ_H 5.31 br t) in the ¹H NMR spectrum (Fig. 2S), with no other proton signals within 1 ppm in either direction. Having established these key features of 1, correlations in HSQC, HMBC, and COSY NMR spectra were examined, with the goal of establishing connections between what are essentially five separate proton spin systems. Figure 2 depicts the key HMBC correlations that aided this, with emphasis (asterisks) on 2-and 3-bond proton-to-carbon correlations from one substructure to another. Complete assignment of protons and carbons within each substructure was established in standard fashion using ¹H, ¹³C, DEPT, COSY, HSQC, and HMBC NMR methods.

Table 1.

¹H (400 MHz) and ¹³C (100 MHz) NMR Spectroscopic Data for Compounds 1, 3, 7, and 10 (δ in ppm)

position	1		3		7		10
	δC	δH (J in Hz)	δC	δH (J in Hz)	δC	δH (J in Hz)	δC	δH (J in Hz)
2ax	70.8	3.89, t (10.2)	70.9	4.01, t (10.2)	70.8	4.01, t (10.0)	153.4	8.24, s
2eq		4.16, br dt (10.2, 2.5)		4.29, br d (10.2)		4.33, dt (10.0, 2.6)
3	32.8	3.46, m	33.0	3.45, m	32.6	3.47, m	124.7
4ax	31.9	2.86, dd (15.6, 10.9)	31.9	2.99, dd (15.5, 11.2)	31.4	2.98, dd (15.6, 11.1)	176.1
4eq		2.74, dd (15.6, 4.1)		2.79, dd (15.5, 4.1)		2.81, dd (15.6, 4.2)
5	131.1	6.75, s	127.8	6.72, d (8.2)	124.8	6.66, d (8.3)	124.2	7.84, s
6	121.1		108.4	6.39, d (8.2)	108.6	6.39, d (8.3)	128.1
7	154.7		154.7		149.6		158.0
7-OH		7.97, s		7.86, s		7.53, br s		8.25, s
8	103.6	6.29, s	116.2		136.6		116.6
8-OMe					60.7	3.77, s
9	154.1	153.9			148.9		155.1
10	114.3		114.6		115.9		118.9
1′	120.9		119.2		119.6		125.9
2′	154.3		154.9		156.9		131.1	7.57, d (8.7)
2′-OH		7.12, s		8.15, s		8.11, br sa
3′	116.4		105.0	6.45, s	103.7	6.46, d (2.3)	114.1	6.97, d (8.7)
4′	155.4		155.6		158.1		160.5
4′-OH		8.06, s		7.75, s		8.39, br sa
4′-OMe							55.7	3.83, s
5′	108.4	6.43, d (8.4)	126.1		107.9	6.34, dd (8.3, 2.3)	114.1	6.97, d (8.7)
6′	125.3	6.80, d (8.4)	127.0	6.99, s	128.9	6.96, d (8.3)	131.1	7.57, d (8.7)
1′′	28.5	3.23, d (7.2)	23.1	3.32, d (6.9)			29.3	3.47, d (7.2)
2′′	124.7	5.31, br t (7.2)	124.6	5.25, br t (6.9)			122.45	5.39, m
3′′	131.7		130.6				134.4
3′′-Me	17.9	1.70, s	18.0	1.74, s			18.0	1.73, s
4′′	26.0	1.70, s	26.0	1.63, s			26.0	1.77, s
1′′′	23.4	3.45, d (6.7)	27.7a	1.42, s			23.1	3.64, d (6.9)
2′′′	124.0	5.25, br t (6.7)	40.8				122.52	5.25, m
2′′′-Me			27.6a	1.42, s
3′′′	132.0		149.4	6.25, dd (17.5, 10.7)			133.2
3′′′-Me	18.1	1.77, s					18.1	1.84, s
4′′′	26.0	1.66, s	110.4	4.98, d (17.5) 4.93, d (10.7)			25.9	1.67, s

^aAssignments may be interchanged.

Fig. 2.

HMBC correlations (arrows) that established key connectivities and substitution patterns in the structure determination of verdean A (1). Asterisks indicate important ring and side-chain connections. Dashed lines indicate connected sub-structures. A complete list of observed HMBC correlations is provided in Table 1S. Similar patterns of HMBC data and associated interpretations were used for the determination of verdeans B and C (3, 7), and compounds 2, 4-6, 8, and 9.

Having determined the overall structure of 1, ¹H NMR coupling constants were important for assigning the chemical shifts of H-2_ax (δ_H 3.89), H-2_eq (δ_H 4.16), H-4_ax (δ_H 2.89), and H-4_eq (δ_H 2.74). Ring C of isoflavan 1 adopts a half-chair conformation where the nearly anti-periplanar H-2_ax and H-3 are coupled with the same magnitude as the H-2_ax to H-2_eq geminal coupling, both 10.2 Hz. Ring B is equatorial and H-4_ax exhibited coupling to H-3 of similar strong magnitude (10.9 Hz), with large geminal coupling also observed between H-4_ax and H-4_eq (15.6 Hz). The half-chair conformation, and the equatorial ring B, verified in this way, were consistent with the related phenomena observed in the electronic circular dichroism (ECD) spectrum of 1 (Fig. 10S) that exhibits M-helicity, with the key features of a positive Cotton effect between 230 nm and 250 nm (¹L_a band) and a negative Cotton effect between 280 nm and 300 nm (¹L_b band, π – π* transition). Taken together, these observations indicated the S configuration at C-3 in accordance with well-established precedent for isoflavans (Slade et al., 2005; Kim et al., 2009). Compound 1 (Fig. 1) was therefore identified as (S)-7,2′,4′-trihydroxy-6,3′-(3-methylbut-2-en-1-yl)-isoflavan, which we have summarily named verdean A.

Fig. 1.

Structures of compounds 1-13 isolated from D. nana.

Compound 2 had the molecular formula C₂₀H₂₂O₄ by HRESIMS (obsvd. [M+H]⁺ m/z 327.1591, calcd. for C₂₀H₂₃O₄, 327.1591), and was one of five compounds in this study that afforded an exact mass to four decimal places. The ¹H NMR spectrum of 2 was very similar to 1, but with the absence of the characteristic set of signals for one of the 3-methylbut-2-en-1-yl groups relative to 1. The molecular formula supported this, revealing five fewer carbons, eight fewer protons, and one less unsaturation by comparison. The absence of a prenyl group at C-6 was evident by the appearance of the ortho-coupled (8.2 Hz) C-5 and C-6 protons (δ_H 6.88, d, and (δ_H 6.36, dd, respectively). These signals, and the meta-coupling (2.4 Hz) observed between C-6 and C-8 (δ_H 6.27, d) verified the ABX spin system of ring A. The remaining ¹H and ¹³C NMR signals of 2 were in very close alignment with 1 (Table 1); nearly identical signals were observed for all of the atoms of ring B. The overall structure of 2, and its spectroscopic assignments were verified by HSQC, HMBC, and COSY methods, with connections between subunits established in the same manner as for 1 (Fig. 2). The ECD spectrum of 2 (Fig. 19S) showed the same combination of a positive Cotton effect between 230 nm and 250 nm, and negative Cotton effect between 280 nm and 300 nm, likewise indicating an S configuration at C-3. Compound 2 (Fig. 1) was therefore identified as (S)-7,2′,4′-trihydroxy-3′-(3-methylbut-2-en-1-yl)-isoflavan. This is the previously undescribed enantiomer of the known compound phaseollidin isoflavan, which was originally reported as a synthetic derivative (Biggs and Fielder, 1994). It was later found as a natural product in the R configuration, having exhibited a negative Cotton effect at 233 nm by ECD (Tanaka et al., 2010).

Compound 3 had a molecular formula identical to 1, C₂₅H₃₀O₄, as provided by HRESIMS. ¹³C and DEPT NMR spectroscopic data revealed the same characteristic signals for the core-ring C of an isoflavan, with C-2 (δ_C 70.9), C-3 (δ_C 33.0), and C-4 (δ_C 31.9). Chemical shifts and coupling constants of the attached protons H₂-2, H-3, and H₂-4, as determined by HSQC, (Table 1) were consistent with this. In fact, compounds 1-9 of this study had this spin-system in common that for each confirmed the same relative configuration, a half-chair with an equatorial ring B, as described for 1.

The differing substitution pattern of rings A and B of 3 were readily determined. The ortho-coupled (8.2 Hz) protons H-5 (δ_H 6.72, d) and H-6 (δ_H 6.39, d) were comparable to those of 2. The position of H-5 is distinctive for its HMBC correlation to C-4 (δ_C 31.9), itself a distinctive position that is the site of a ketone in many flavonoids. Oxygenation of C-7 is also a common feature of flavonoids, as is the case for 1-12 here. The three phenolic protons of 3 were narrow peaks in the ¹H NMR spectrum that therefore provided a valuable set of HMBC correlations to their attached carbons and to both adjacent carbons in each case (Tables 1 and 3S). The placement of the 3-methylbut-2-en-1-yl group at C-8 was confirmed by HMBC correlations to C-8 (δ_C 116.2) from H₂-1′′ and by a long-range correlation from H-6 (δ_H 6.39, d) to C-1′′ (δ_C 23.1). Ring B was substituted with a 2-methylbut-3-en-2-yl side chain, another ubiquitous prenyl group. The placement of this group at C-5′ was established by HMBC correlations between protons and carbons in the region of the point of attachment. One such correlation, for example, was observed between H-6′ (δ_H 6.99, s) and the distinctive, quaternary C-2′′′ (δ_C 40.8) of this type of prenyl group. The 2-methylbut-3-en-2-yl side chain and similar observed HMBC correlations were common to 3-6 and 13. Ideally placed, H-6′ was consistent in affording useful HMBC correlations that verified multiple assignments and the substitution patterns of rings B and C for compounds 1-12. The structure of 3 (Fig. 1) was established, with ECD spectroscopic features identical to 1 and 2, as (S)-7,2′,4′-trihydroxy-8-(3-methylbut-2-en-1-yl)-5′-(2-methylbut-3-en-2-yl)-isoflavan, mercifully named verdean B.

Compound 5 (C₂₀H₂₂O₄ by HRESIMS, Fig. 43S) also possessed an isoflavan core structure. As was seen for 1-4, the carbons C-2 (δ_C 70.8), C-3 (δ_C 33.2), and C-4 (δ_C 31.3) were present. Ring A of 5 was similar to that of 2, with H-5, H-6, and H-8 comprising an ABX spin system, and the positions of each established with HSQC and HMBC correlations. The substitution pattern of ring B of 5 was also similar to that of 3 and 4. Key HMBC correlations from H-6′ to C-3 (δ_C 33.2) and C-2′′ (δ_C 40.9) unambiguously placed the 2-methylbut-3-en-2-yl side chain at C-5′. H-6′ also showed strong 3-bond HMBC correlations to the oxygenated carbons C-2′ (δ_C 154.9) and C-4′ (δ_C 155.7). H-3′ (δ_H 6.45, s) was clearly para to H-6′, both singlets, and while H-3′ correlates strongly to C-1′ (δ_C 119.0) and to C-5′ (δ_C 126.2), it shows only a weak, long-range coupling to H-3. Such long-range correlations are common in the HMBC spectra of flavonoids and are overwhelmingly consistent with the primary connectivity established by 2- and 3-bond correlations. Compound 5 was therefore established (Fig. 1) and its overall structure corresponded to the known compound manuifolin K. Compound 5 ([α]d +13.5) was, however, determined to be in the S configuration by ECD (Fig. 44S) and is therefore the previously undescribed enantiomer of the reported (R)-manuifolin K ([α]d −22.71, Zeng et al., 1998; Keßberg, et al. 2018).

The structure of compound 7 (C₁₆H₁₆O₅ by HRESIMS, Fig 60S) was deduced using the same methods as were used for 1-6. Key features included the characteristically coupled H₂-2-to-H-3-to-H₂-4 proton spin system of the isoflavan ring C, and the ortho-coupled (8.3 Hz) H-5 and H-6 of ring A. In the case of 7, the dihydroxy-substituted ring B revealed an ABX pattern for protons H-6′ (δ_H 6.96, d, 8.3 Hz), H-5′ (δ_H 6.34, dd, 8.3 and 2.3 Hz), and H-3′ (δ_H 6.46, d, 2.3 Hz). The useful H-6′ exhibited strong HMBC correlations to C-3, C-2′, and C-4′. Compound 7 was the only non-prenylated isoflavan encountered in this study. The distinctive NMR signals for the C-8 methoxy group (δ_H 3.77, s, δ_C 60.7) were observed, all assignments verified by 2D NMR spectroscopic methods, and the configuration at C-3 tentatively established by ECD; a positive Cotton effect was seen at 233 nm (Fig 61S). The very weak ECD signal and low specific rotation of 7 ([α]d –1.75) may suggest the presence of the R enantiomer. Compound 7, or verdean C, was found to be the previously undescribed 7,2′,4′-trihydroxy-8-methoxy-isoflavan.

Compound 10 (C₂₆H₂₈O₄, HRESIMS, Fig. 87S) contained an α,β-unsaturated ketone moiety, with C-2 (δ_C 153.4), C-3 (δ_C 124.7), and a C-4 carbonyl (δ_C 176.1) characteristic of an isoflavone structure (Belofsky et al., 2014). The para-disubstituted ring B possessed a methoxy group (δ_H 3.83, s, δ_C 55.7) at C-4′. The equivalent H-2′/6′ protons showed strong 3-bond HMBC correlations to C-3 (δ_C 124.7), verifying the ring B-to-ring C juncture. Ring A was the site of two 3-methylbut-2-en-1-yl side chains located at C-6 and C-8 by HMBC correlations into the ring from the methylene and methine protons of each. H-5 (δ_H 7.84, s) provided an anchor point, exhibiting strong HMBC correlations to the carbonyl C-4, to the prenyl group point of attachment C-1′′, and to both oxygen-bearing ring carbons C-7 and C-9. Verdean D (10) was determined to be a previously undescribed compound, with the most similar known compound being eryvarin S (Tanaka et al., 2005).

Compounds 4, 6, 8, 9, and 11-13 were characterized using ¹H, ¹³C, DEPT, HSQC, and HMBC NMR spectroscopic methods, and by chiroptical methods where relevant. Their molecular formulae were verified by HRESIMS (see all spectra and tabulated NMR spectroscopic data in Supplementary material). The previously described compounds 4, 6, and 8 were identified as the D. jamesii metabolites ormegans C, A, and E, respectively (Belofsky et al., 2023). Compound 9, (S)-glabridin, exhibited a specific rotation ([α]d −9.25) opposite in sign to that reported for (R)-glabridin ([α]d +9.5; Kinoshita et al., 1996). Importantly, the absolute configurations of synthetic glabridin enantiomers were determined in a prior study by chiral HPLC, and ECD (Kim et al., 2009). The (S)-glabridin found in this study is the first report of this enantiomer of glabridin as an isolated natural product. Compound 11 was found to be the known isoflavone 3′-O-methylorobol (Almeida and Gottleib, 1974; Tahara et al., 1984) and Compound 12 was identified as sideroxylin (Park et al., 2010). Compound 13, ormegafuran A, was also a metabolite of D. jamesii (Belofsky et al., 2023).

2.2. Antimicrobial susceptibility

The isolated D. nana metabolites were evaluated for in vitro antimicrobial activity against a panel of microorganisms (See section 4.4). The antimicrobial growth inhibition curves for the active compounds, crude plant extracts, and positive controls may be seen in the Supplementary material (Figs. 109S–113S), along with error bars and R² coefficients of determination for each. None of the compounds were active against the gram-negative bacteria Escherichia coli, Acinetobacter baumannii, and Klebsiella pneumoniae. Compounds 4, 6, 8, and 13, recently found to be active (Belofsky et al., 2023) toward vancomycin-resistant Enterococcus faecalis (VRE) and C. neoformans, were not retested here. Retesting of these four in this study (Table 2) resulted in MICs matching those found previously for each against MRSA.

Table 2.

Antimicrobial susceptibility testing of D. nana pure compounds. Results for pure compounds are reported as minimum inhibitory concentrations (MIC, µM), while extracts are in µg/mL.

Compound	Pathogen
	Methicillin resistant S. aureus	C. neoformans b	C. albicans b
1	12.9	6.7	10.0
2	>40	---	>100
3	14.1	9.8	31.5
4 a	8.9	---	---
5	>40	37.0	---
6 a	23.2	---	---
7	>40	---	---
8 a	25.0	---	---
9	>40	7.8	---
tetracycline	0.79	---	---
amphotericin B	---	0.12	0.03
aerial parts extract	>100	---	---
root extract	7.1	---	---

^aThese compounds were tested previously against C. neoformans and C. albicans (Belofsky et al., 2023).

^bCompounds not previously tested against these pathogens were selected for testing based on pre-screening. MIC determinations were made on seven serial dilutions and represent the smallest concentration that completely retards bacterial growth (Gompertz model; Lambert and Pearson, 2000).

Verdean A (1) was the most active previously undescribed D. nana compound against MRSA, with an MIC of 12.9 µM (5.1 µg/mL; Table 2), and it was broadly active in a similar way to ormegans A (6) and C (4). Verdean A (1) had an MIC of 6.7 µM against C. neoformans. It had an MIC of 10.0 µM toward the human fungal pathogen C. albicans, only moderately active when compared to that of the positive control amphotericin B (0.03 µM).

Prior activity toward MRSA was reported for the R enantiomer of compound 2, phaseollidin isoflavan (Tanaka et al., 2010). The MIC₉₀ was reported to be 12.5 µM, suggesting a difference from the result of our assays (MIC >40 µM) for the S isomer, although susceptibilities due to strain and laboratory differences can also be significant. The relationship between absolute configuration and bioactivity is an important topic, and an examination of the phenomenon for prenylated isoflavans may be of interest for future studies. On this note, a large body of work on the varied biological activities of glabridin was done on the R enantiomer (Simmler et al., 2013), but the S-isomer was isolated here (9). Our results showing verdean D (10) to be inactive against MRSA are generally in agreement with a prior result for its C-4′ hydroxy analog (Tanaka et al., 2005).

In all, nineteen phenolic metabolites of the taxonomically close D. nana and D. jamesii were evaluated in the same assays and laboratory, and some observations may be made regarding structure activity trends. Except for the D. jamesii compound ormegafuran B, a dimeric version of ormegafuran A (13), the isoflavans were more active than any other chemotype (isoflavone, flavone, chroman, arylbenzofuran). Isoflavans are fully saturated at C-4, while flavonoids having a ketone at this position are far more commonly reported. The ketone-possessing compounds 10-12 in this work, and two such compounds from D. jamesii, were inactive in all assays. These were the only inactive compounds isolated, suggesting a negating effect of the ketone or, perhaps more specifically an α,β-unsaturated ketone.

The presence and number of prenyl substituents had a consistent effect on bioactivity in these assays. This phenomenon has been linked to the improvement of bioavailability with the increase in lipophilicity, and has been the subject of review (Mukai, 2018, Chen et al., 2014). All compounds lacking a prenyl group were inactive, as seen for 7, 11, and 12 in this study and for a chroman metabolite of D. jamesii. Of the compounds possessing one prenyl group (2, 5, 8, and 13), only ormegan E (8) was active, with MICs of 25.0 µM (this study) and 4.6 µM (Belofsky et al. 2023) against MRSA and C. neoformans, respectively. Interestingly, 8 possessed a prenyl group on ring A (C-8), and ormegans E and F from D. jamesii were also mono-prenylated at C-8 and exhibited weak activities. The inactive 2, by comparison, has the same type of prenyl group located on ring B and a nearly identical structure otherwise, lacking only a methoxy at C-7. Glabridin (9) possesses a single, cyclized prenyl group, is otherwise similar to 8, and is active only against C. neoformans (MIC 7.8 µM) in these assays. These observations suggest that the site of prenylation and whether cyclization has occurred may be important.

Di-prenylation had dramatic effects on the observed bioactivities of the isoflavan metabolites. The most active and broadly active compounds were di-prenylated. Ormegan B, from D. jamesii, was an exception that might be explained by the fact that its prenyl group is an unusual 7-(3-methylbut-2-en-1-yl)oxy group and perhaps susceptible to facile cleavage. Di-prenylation appears to have its limits however and was unable to overcome the presence of a ketone in compound 10, and in a flavone of D. jamesii, both of which were inactive. Verdeans A (1), B (3), and ormegans A (6) and C (4) were all di-prenylated and were the most active and broadly active isoflavans. With the exception of 1, all possess a 2-methylbut-3-en-2-yl at C-5′. In light of these observations, the activity of the D. jamesii compound ormegafuran B (MICs ≤ 5.1 µM against MRSA, VRE, and C. neoformans), might best be attributed to possessing two 2-methylbut-3-en-2-yl groups.

3. Concluding remarks

The potential importance of isoflavans as a bioactive subgroup of flavonoids was reinforced in this study, as was the importance of prenylation for antimicrobial activity. The presence of a C-4 ketone was found to be detrimental here, perhaps suggesting a reevaluation of the continued and widespread research focus on ubiquitous flavonoids having this feature. Four previously undescribed isoflavonoids (1, 3, 7, and 10) were found, adding to the prolific variety of metabolites and activities of Dalea spp.

Prior observation (Belofsky et al., 2014, 2023) that antimicrobial activity is localized in the roots of Dalea spp. was supported by the results of this study. The extract of the aerial parts was inactive in initial screening while that of the roots expressed strong activity against MRSA (MIC 7.1 µg/mL). Compounds 7, 11, and 12, all inactive, were found only in the plant aerial portion extract. Compounds 4 and 5 were found in both root and aerial portions, but only 4 exhibited antimicrobial activity. All other metabolites reported were from the roots.

A recent study (Kumarihamy, 2021) of another Daleoid, Psorothamnus schottii (Torr.) Barneby (syn: Dalea schottii) revealed weak in vitro activity against MRSA for a previously undescribed prenylated isoflavone schottiin, likewise isolated from the plant roots. There were further observed synergistic effects of schottiin in combination with methicillin that led to an 8-fold decrease in MIC against MRSA. These results contribute to a growing body of knowledge of similar phenomena observed for other suites of metabolites of Dalea spp. (Belofsky, 2004, 2013). In this context of synergy as a chemical-ecological phenomenon, it would be of interest to conduct field and laboratory studies to characterize the microbial and invertebrate populations of the root environments of Dalea spp., and to investigate the potential effects of these compounds on associated organisms.

In the absence of much needed DNA-based phylogenetic analyses for the majority of Dalea spp., the close relationship of D. nana and D. jamesii has been based only on morphological features. This study reveals a similar metabolite profile, with four major metabolites in common for these two species, providing a synapomorphy in the form of chemotaxonomic evidence. However, further phylogenetic studies would enable the prioritization of Dalea spp. for investigation based on anticipated similar, or perhaps hoped-for different, chemical content.

4. Experimental

4.1. General experimental procedures

Optical rotations were recorded on a PerkinElmer 341 polarimeter (Na lamp, 589 nm). Specific rotations for all compounds were measured at a concentration of 0.002 g/mL in CHCl₃. V spectra were recorded on an HP-Agilent 8453 photodiode array instrument. ECD spectra were obtained on an Olis DSM-10 circular dichroism spectrometer using dry MeOH as the solvent and a 0.5 cm path length quartz cuvette. IR spectra were recorded on a Bruker alpha spectrometer with universal Alpha T attachment or ATR attachment. NMR spectra were obtained on a Bruker Avance 400 MHz system with Topspin 1.3 software. HR-ESIMS data were recorded on a Waters Q-TOF Premier hybrid mass spectrometer. A Waters Acquity UPLC was used to inject samples in 1:1 MeCN−H₂O using flow injection analysis (100 μL/min) with no intervening column. Positive mode ESIMS was used to generate [M+H]⁺ ions. A Molecular Devices Spectramax i3 plate reader was used for measuring the optical density of microbial pathogens.

All D. nana compounds were purified by open-column chromatography. Chromatography with Sephadex LH-20 was performed in 2.5 cm (i.d.) columns of varying gel height, slurry-packed in MeOH and equilibrated in 3:1:1 hexanes-toluene-MeOH until there was no further shrinking. Samples were dissolved in minimal amounts of the same solvent mixture and eluted (flow rate < 0.5 mL/min.), using an automated fraction collector, into test tubes (~8 mL/tube). Near the end of the Sephadex column elution the hexanes-toluene-MeOH solvent system was changed back to 100% MeOH. Preparative linear gradient chromatography employed a custom two-chamber apparatus creating a continuous gradient, with gravity flow (∼20 mL/min), over 70−230 mesh silica gel. This was connected to glass columns of varying sizes sealed with PTFE end-fittings. Samples were adsorbed onto silica gel in solution and evaporated to dryness prior to loading onto the column. Eluting solvent percentages given (See section 4.3) represent estimates, for the sake of reproducibility, of the solvent compositions entering the column. Collection of 20 fractions (50 mL each) for example, with a linear gradient of EtOAc (0→100%) in hexanes, with 500 mL of solvent per chamber, will result in fraction 10 having a composition of ∼50% EtOAc. Step-gradient columns of various sizes employed silica gel (70–230 mesh) and were run with adsorbed samples. TLC plates (Sigma-Aldrich; silica gel 60, F254) were eluted with mixtures of CH₂Cl₂:MeOH or hexane:EtOAc of increasing polarity and visualized with UV (254 nm) and with the spray reagent vanillin−H₂SO₄ (1 g/100 mL w/v) followed by gentle heating.

4.2. Plant material

Whole plants of Dalea nana (Fig. 1S) were collected by G. Belofsky, H. Wolhart, and N. Hansen on August 31, 2015, alongside railroad tracks intersecting FR318, approximately 150 meters north of the Verde River, Yavapai County, Arizona, GPS coordinates N 34°53.722′, W 112°12.622′, elevation 1189 meters. A voucher specimen was authenticated by David Giblin, University of Washington, Seattle, WA and was deposited in the Burke Herbarium of the same institution, accession no. WTU-407594. Aerial parts and roots (washed with deionized H₂O) were air-dried for ~48 h and stored (−20 °C) separately prior to extraction.

4.3. Extraction and isolation

See Supplementary Material File

D. nana roots (143.0 g) were chopped-up and extracted by blending in a Waring blender with MeOH (1.5 L), filtered, and rotary evaporated to afford 15.4 g of extract. This material was re-dissolved in MeOH and adsorbed to silica gel (70–230 mesh) by rotary evaporation. The adsorbed sample was then loaded onto a column bed measuring 10 cm (i.d.) × 5 cm (h) packed with silica gel (230–400 mesh) and fractionated by vacuum liquid chromatography (VLC) column eluted with 1 L of hexane, followed by mixtures of 500 mL each of EtOAc in hexane: 20%, 40%, 60%, 80%, and 100%, successively. Elution continued with mixtures of 500 mL each of MeOH in CH₂Cl₂: 2%, 5%, 8%, 10%, and 30% in succession, resulting in eleven VLC fractions. Fractions 3 and 4 were combined (2.0 g) and further fractionated by Sephadex LH-20 column chromatography (65 × 2.5 cm) (See section 4.1). Collection tubes were analyzed by TLC and similar materials combined, resulting in 29 fractions. Compounds 1-6, 9, 10, and 13 all resulted from further purification of fractions from this column.

The combined fractions 2 and 3 (22.2 mg) were further purified using linear gradient column chromatography (10 × 1 cm) over silica gel (See section 4.1), eluted with MeOH in CH₂Cl₂ (0→5%). The obtained fractions 4–7, eluted with ~1% MeOH, afforded compound 1 (16.4 mg).

Fraction 19 (96.0 mg) was further purified using linear gradient column chromatography (13 × 2.5 cm) over silica gel, eluted with mixtures of EtOAc:hexane (0→50%); fractions 5–7, eluted with ~25% EtOAc, afforded compound 3 (24.6 mg).

Fraction 23 (24.5 mg) was purified using a step-gradient column (10 × 1.5 cm), using mixtures of EtOAc in hexane (0→50%). The combined fractions 1–4 from this column, eluted with ~30% EtOAc, contained compound 2 (4.4 mg).

Moreover, the combined fractions 11 and 12 (410.1 mg) from the Sephadex LH 20 column were purified by another Sephadex LH-20 column (62 × 2.5 cm), using first hexane-toluene-MeOH, 6:2:1, and changing to 3:1:1 after ~18 h (flow rate 0.5 ml/min.). Fractions 12 and 13, eluted with the 3:1:1 solvent mixture, were combined (267.1 mg) and further purified by linear gradient column chromatography (21 × 2.5 cm) over silica gel, eluted with MeOH in CH₂Cl₂ (0→5%). Fraction 1 from this column, eluted with ~3–4% MeOH contained pure compound 4 (10.4 mg). Additional amounts (189.2 mg; >90% purity) of this major compound continued to be eluted in the eight subsequent fractions.

Fraction 21 (41.8 mg) from the initial Sephadex LH 20 column was purified by linear gradient column chromatography (10 × 2.5 cm) over silica gel, eluted with EtOAc in hexane (0→50%). Fraction 2 from this column, eluted with ~30% EtOAc, afforded compound 5 (19.3 mg).

Fraction 13 (30.7 mg) from the initial Sephadex LH 20 column was purified using a step-gradient column (10 × 2 cm) over silica gel, eluted with MeOH in CH₂Cl₂ (0→10%). Fractions 1–3 from this column, eluted with ~1–3% MeOH, were combined (26.4 mg) and further purified by linear gradient column chromatography (10 × 1 cm) over silica gel, eluted with EtOAc in hexane (0→30%). The combined fractions 3–5 from this column, eluted with ~10% EtOAc, afforded compound 6 (14.5 mg).

Fraction 17 (158.1 mg) from the initial Sephadex column was purified by linear gradient column (17 × 2.5 cm) chromatography over silica gel, eluted with EtOAc in hexanes (0→50%). The combined fractions 7–9 from this column, eluted with ~30% EtOAc, yielded compound 8 (93.5 mg).

Fraction 18 (40.5 mg) from the initial Sephadex column was purified by linear gradient column (10 × 2.5 cm) chromatography over silica gel, eluted with EtOAc in hexanes (0→30%). Fractions 1–4 from this column (19.7 mg) were passed through a second Sephadex LH-20 column without improvement in purity, recombined, and successfully purified using a step-gradient column over silica gel, eluted with MeOH in CH₂Cl₂ (0→10%). The combined fractions 3 and 4 from this column, eluted with ~3% MeOH, afforded compound 9 (9.3 mg).

Fraction 8 (34.9 mg) from the initial Sephadex column was purified using step-gradient column (12 × 2 cm) chromatography over silica gel, eluted with MeOH in CH₂Cl₂ (0→10%). Fraction 3 from this column, eluted with ~3–5% MeOH, yielded compound 10 (17.2 mg).

Fraction 26 (89.4 mg) from the initial Sephadex column was purified by linear gradient column (14 × 2.5 cm) chromatography over silica gel, eluted with EtOAc in hexanes (0→50%). The combined fractions 2–4 from this column, eluted with ~30% EtOAc, afforded compound 13 (36.4 mg).

D. nana aerial parts (612 g) were blended in a Waring blender, stirred for 24 h with an overhead mixer, filtered, and rotary evaporated to afford 33.3 g of extract. This material was re-dissolved in MeOH and adsorbed to silica gel (70–230 mesh) by rotary evaporation. The adsorbed sample was then loaded onto a column bed measuring 10 cm (i.d.) × 6 cm (h) packed with silica gel (230–400 mesh). This VLC column was eluted with 2 L of hexane, followed by 1 L each of mixtures of EtOAc in hexane: 20%, 40%, 60%, 80%, and 100% successively. Elution continued with mixtures of 1 L each of MeOH in CH₂Cl₂: 2%, 5%, 8%, 10%, and 30% in succession, resulting in eleven VLC fractions. Fractions 4 and 5 were combined (2.13 g) and further fractionated by Sephadex LH-20 column (38 × 2.5 cm) chromatography. Collection tubes were analyzed by TLC, and similar materials combined, resulting in 48 fractions. Compounds 7, 11, 12, and additional amounts of compounds 4 and 5, all resulted from further purification of fractions from this Sephadex column.

The combined fractions 36 and 37 (27.4 mg) were further purified using a step-gradient column (7 × 1.5 cm) over silica gel, eluted with MeOH in CH₂Cl₂ (0→6%). Fraction 3 from this column (10.9 mg), eluted with ~5% MeOH, was further purified using a step-gradient column (7.5 × 1.5 cm) over silica gel, eluted with EtOAc in hexane (0→100%). The combined fractions 1 and 2 from this column, eluted with ~60% EtOAc, yielded compound 7 (9.6 mg).

Fraction 32 (115.3 mg) from the Sephadex column was further purified by linear gradient column (13 × 2.5 cm) chromatography over silica gel, eluted with EtOAc in hexane (0→50%). Fractions 4 and 5 from this column (43.2 mg), eluted with ~50% EtOAc, were combined, evaporated, and triturated with CHCl₃ (2 mL). This resulted in a suspended solid that was removed by filtration through a cotton plug, re-dissolved in CH₂Cl₂-MeOH, 2:1, and evaporated to yield compound 11 (24.4 mg).

The combined fractions 13–15 (220.2 mg) from the Sephadex column were purified by linear gradient column (12 × 2.5 cm) chromatography over silica gel, eluted with MeOH in CH₂Cl₂ (0→10%). Fraction 3 from this column (138.8 mg), eluted with ~4–5% MeOH, was evaporated and triturated with acetone (2 mL). The suspended solid was removed by filtration through a cotton plug and the resulting filtrate was evaporated to afford compound 12 (42.5 mg).

Fraction 11 (65.2 mg) from the Sephadex column was purified by linear gradient column (13 × 2.5 cm) chromatography over silica gel, eluted with EtOAc in hexane (0→25%). Fraction 1 from this column (16.7 mg), eluted with ~20–25% EtOAc was further purified by step-gradient column (7 × 1.5 cm) chromatography over silica gel, eluted with EtOAc in hexanes (0→50%). The combined fractions 7 and 8 from this column, eluted with ~25–30% EtOAc, provided additional compound 4 (4.2 mg) to that isolated from the root extract.

The combined fractions 30 and 31 (136.5 mg) from the Sephadex column were purified by linear gradient column (11 × 2.5 cm) chromatography over silica gel, eluted with EtOAc in hexane (0→50%). Fraction 2 from this column (60.2 mg), eluted with ~45% EtOAc, was further purified by linear gradient column (11 × 1.5 cm) chromatography over silica gel, eluted with MeOH in CH₂Cl₂ (0→6%). The combined fractions 2 and 3 from this column, eluted with ~3% MeOH, yielded additional compound 5 (47.3 mg) to that isolated from the root extract.

4.3.1. Verdean A (1)

White solid; [α]d −2.5 (c 0.002 g/mL, CHCl₃); ECD (c 1.3 mM, MeOH), λ_max (θ) 234 (114.0), 289 (−23.1) nm; UV (MeOH) λ_max (log ε) 207 (4.84), 227 (sh) (4.41), 285 (3.94), 294 (sh) (3.84) nm; IR (KBr) ν_max 3408 (br OH), 2922, 1619, 1497, 1451, 1288, 1164, 1114, 1032, 668 cm⁻¹; ¹H and ¹³C NMR (acetone-d₆), see Table 1; Complete HMBC correlations, see Table 1S; HRESIMS m/z 395.2215 [M+H]⁺ (calcd for C₂₅H₃₁O₄, 395.2217).

4.3.2. Phaseollidin isoflavan (2)

White solid; [α]d −0.75 (c 0.002 g/mL, CHCl₃); ECD (c 1.5 mM, MeOH), λ_max (θ) 231 (131.2), 285 (−20.0) nm; UV (MeOH) λ_max (log ε) 206 (4.72), 228 (sh) (4.26), 283 (3.83), 290 (sh) (3.74) nm; IR (KBr) ν_max 3383 (br OH), 2922, 1618, 1601, 1508, 1453, 1292, 1274, 1228, 1155, 1116, 1030 cm⁻¹; ¹H, ¹³C, and complete HMBC NMR correlations (acetone-d₆), see Table 2S; HRESIMS m/z 327.1591 [M+H]⁺ (calcd for C₂₀H₂₃O₄, 327.1591).

4.3.3. Verdean B (3)

Yellow oil; [α]D –8.0 (c 0.002 g/mL, CHCl₃); ECD (c 1.3 mM, MeOH), λ_max (θ) 235 (68.2), 285 (−9.1) nm; UV (MeOH) λ_max (log ε) 207 (4.85), 229 (sh) (4.43), 286 (4.09) nm; IR (KBr) ν_max 3408 (br OH), 2966, 2922, 1613, 1504, 1452, 1432, 1278, 1167, 1084, 1058 cm⁻¹; ¹H and ¹³C NMR (acetone-d₆), see Table 1; Complete HMBC correlations, see Table 3S; HRESIMS m/z 395.2217 [M+H]⁺ (calcd for C₂₅H₃₁O₄, 395.2217).

4.3.4. Manuifolin K (5)

Yellow oil; [α]d +13.5 (c 0.002 g/mL, CHCl₃); ECD (c 1.5 mM, MeOH), λ_max (θ) 234 (371.0), 288 (−85) nm; UV (MeOH) λ_max (log ε) 206 (4.76), 225 (sh) (4.28), 285 (3.97), 290 (sh) (3.94) nm; IR (KBr) ν_max 3404 (br OH), 2965, 2927, 1623, 1599, 1507, 1469, 1274, 1155, 1115, 1027, 841 cm⁻¹; ¹H, ¹³C, and complete HMBC NMR correlations (acetone-d₆), see Table 5S; HRESIMS m/z 327.1591 [M+H]⁺ (calcd for C₂₀H₂₃O₄, 327.1591).

4.3.5. Verdean C (7)

Yellow oil; [α]d –1.75 (c 0.002 g/mL, CHCl₃); ECD (c 1.7 mM, MeOH), λ_max (θ) 234 (53.4), 286 (~0) nm; UV (MeOH) λ_max (log ε) 207 (4.72), 227 (sh) (4.30), 281 (3.81), 286 (sh) (3.77) nm; IR (KBr) ν_max 3359 (br OH), 2924, 1619, 1498, 1473, 1460, 1297, 1211, 1171, 1085, 1049, 975 cm⁻¹; ¹H and ¹³C NMR (acetone-d₆), see Table 1; Complete HMBC correlations, see Table 7S; HRESIMS m/z 289.1069 [M+H]⁺ (calcd for C₁₆H₁₇O₅, 289.1071).

4.3.6. Verdean D (10)

Clear oil; UV (MeOH) λ_max (log ε) 202 (4.59), 247 (sh) (4.40), 255 (4.43), 309 (4.00) nm; IR (KBr) ν_max 3416 (br OH), 2924, 1604, 1510, 1441, 1373, 1288, 1244, 1176, 1104, 1028 cm⁻¹; ¹H and ¹³C NMR (acetone-d₆), see Table 1; Complete HMBC correlations, see Table 10S; HRESIMS m/z 405.2050 [M+H]⁺ (calcd for C₂₆H₂₉O₄, 405.2060).

4.4. Antimicrobial assays

4.4.1. Microbial pathogens and culture conditions

All pathogens were purchased as ‘Kwik stick’ cultures from Microbiologics, St. Cloud, MN, and catalog numbers are indicated after the ATCC designations. All compounds and extracts were tested against methicillin-resistant Staphylococcus aureus (MRSA) ATCC 43300 (0852P), vancomycin-resistant Enterococcus faecalis (VRE) ATCC 51299 (0959P), Escherichia coli ATCC 25922 (0335A), Acinetobacter baumannii ATCC 19606 (0357P), Klebsiella pneumoniae ATCC 13883 (0351P), Cryptococcus neoformans ATCC 66031 (0985P) and Candida albicans ATCC 10231 (0443P) using a modified broth dilution assay (Clinical and Laboratory Standards Institute, 2017, in References). All materials tested were of an approximate purity of >95% by cursory analysis of ¹H NMR spectroscopic data. The strains of S. aureus, E. coli and K. pneumoniae were grown at 37 °C on Tryptic Soy media (TSA, TSB; BD Biosciences,). A. baumannii was cultured at 37 °C on nutrient medium (BD Biosciences). Brain heart infusion (BHI; BD Biosciences) was used for the cultivation of VRE at 37 °C. Yeast malt (YM; BD Biosciences) media was used for cultivating C. albicans at 30 °C. C. neoformans was grown at 30 °C on Sabouraud dextrose media.

4.4.2. MIC determination for Gram-positive and -negative bacteria and fungi

Compounds and extracts were dissolved in DMSO at 4 mg/mL and stored at −20 °C. Bacteria and C. albicans were grown to mid-log phase, diluted with fresh medium to an optical density at 600 nm (OD₆₀₀) of 0.030–0.060 and then diluted again 1:10. C. neoformans was grown to mid-log phase, diluted to an OD₆₀₀ of 0.030–0.060 and used directly without further dilution. Cell suspensions (195 µL) were added to wells in a 96 well microtiter plate (Sarstedt) and 5 µL of compound dissolved in DMSO was added to give a final concentration of 100–0.05 µg/mL at 2.5% DMSO by volume. A DMSO negative control and standard antibiotic positive controls were included in each plate. Tetracycline (Sigma; 10 µg/mL in DMSO) was used as positive control against S. aureus, B. subtilis, E. coli, A. baumannii and K. pneumoniae. Penicillin G (Sigma; 10 µg/mL) served as the positive control against VRE. Amphoterocin (Sigma; 10 µg/mL) was used as the positive control for C. albicans and C. neoformans. All compounds were tested in duplicate for each concentration. Plates were sealed with parafilm, placed in a Ziploc bag to prevent evaporation, and incubated at 30 °C (fungi) or 37 °C (bacteria) for 16–20 hours (48 hours for C. neoformans). The OD₆₀₀ values for each well were determined with a plate reader (Biotek, EL800) and the data were standardized to the DMSO control wells, after subtracting the background from the blank media wells, and divided by the average DMSO treated cells control value to determine percent viability. MIC determinations were calculated with Graphpad Prism using the Gompertz function for curve fitting (Lambert and Pearson, 2000).

Supplementary Material

Appendix A. Supplementary material

Supplementary data to this article can be found online at https://doi.org/10.1016/j.phytochem.2024.114224.

Highlights.

• Four new prenylated isoflavanoids (1, 3, 7, and 10) were isolated from Dalea nana.

• Compounds 1, 3, 4, 6, and 8 exhibited antimicrobial activity to human pathogens.

• Prenylation augmented bioactivity, and activity was concentrated in plant roots.

Data availability

Data will be made available by request.