Force-enhanced sensitive and specific detection of DNA-intercalative agents directly from microorganisms at single-molecule level

Abstract

Microorganisms can produce a vast array of bioactive secondary metabolites, including DNA-intercalating agents like actinomycin D, doxorubicin, which hold great potential for cancer chemotherapy. However, discovering novel DNA-intercalating compounds remains challenging due to the limited sensitivity and specificity of conventional activity assays, which require large-scale fermentation and purification. Here, we introduced the single-molecule stretching assay (SMSA) directly to microbial cultures or extracts for discovering DNA-intercalating agents, even in trace amounts of microbial cultures (5 μl). We showed that the unique changes of dsDNA in contour length and overstretching transition enable the specific detection of intercalators from complex samples without the need for extensive purification. Applying force to dsDNA also enhanced the sensitivity by increasing both the binding affinity K_a and the quantity of ligands intercalation, thus allowing the detection of weak intercalators, which are often overlooked using traditional methods. We demonstrated the effectiveness of SMSA, identified two DNA intercalator-producing strains: Streptomyces tanashiensis and Talaromyces funiculosus, and isolated three DNA intercalators: medermycin, kalafungin and ligustrone B. Interestingly, both medermycin and kalafungin, classified as weak DNA intercalators (K_a ∼10³ M^–1), exhibited potent anti-cancer activity against HCT-116 cancer cells, with IC₅₀ values of 52 ± 6 and 70 ± 7 nM, respectively.

Graphical Abstract

Graphical Abstract.

Introduction

DNA has long been a target for anticancer small-molecule medicines (1,2). Chemotherapeutic drugs that act as double-strand DNA (dsDNA)-binding ligands, including alkylating agents, intercalators and groove binders, remain a crucial pillar of many cancer treatment regimens (1). Approximately 17% (55 out of 321) of the therapeutic anticancer drugs approved between 1949 and 2019 are DNA-interacting agents (3), with around half of these agents derived from natural products (3,4). Many notable antibiotics and/or anticancer natural products are DNA intercalators, including actinomycin D, doxorubicin, idarubicin and camptothecin (5). These intercalators bind DNA by inserting aromatic moieties between adjacent DNA base pairs, and subsequently disrupt fundamental processes like DNA replication and RNA transcription (6,7). Understanding genome structure and gene function has also created new opportunities for DNA intercalators targeting cellular DNA or associated proteins, including topoisomerase (8), RNA polymerase (9), FACT (facilitates chromatin transcription factor) (10), and targeting non-canonical DNA structures like G-quadruplexes (11,12).

Given the escalating issue of drug resistance, there is a pressing need for novel DNA interacting agents. Additionally, considering the diverse clinical efficacy observed among drugs from various chemical families targeting the same molecule, novel DNA-binding agents hold promise for combating different types of tumors. Microorganisms, which produce diverse secondary metabolites to gain a competitive advantage, represent a rich source of bioactive DNA-binding agents (13,14). Bioactive DNA-binding agents-producing microorganisms can develop various self-resistance mechanisms including efflux pumps, chemical modifications, and metabolic dormancy, to survive the effects of their own products (4,15). However, the ability to produce DNA-interacting compounds is not universal among organisms, instead, such compounds are restricted to specific species under particular environmental conditions (14). This highlights the need for a more accessible and rapid screening method to identify microorganisms that produce DNA-intercalating compounds with chemical diversity.

The conventional approach to identifying microorganisms that produce bioactive natural products primarily relies on their biological activity in crude extracts or purified compounds (16). However, this approach may overlook potentially valuable medicinal compounds with low yields or low binding affinities. Moreover, predicting the targets of these compounds before chemical purification and structural elucidation remains challenging. For the detection of DNA-binding compounds, the most commonly used and sensitive detection methods currently include surface plasmon resonance (17), affinity mass spectrometry (18), fluorescence-based assays, ultraviolet spectroscopy (UV) (19) or circular dichroism (CD) methods. Although each of these techniques has specific advantages and limitations, they generally require purified reagents, especially for the analysis of non-covalent DNA-ligands interactions (binding affinity K_a ∼10⁵ to 10⁷ M⁻¹). However, the separation and purification process of small molecules necessitate a substantial culture volume (>10 l of liquid culture or 1 kg of solid culture) to isolate more than a microgram of reagents (20).

Direct observation and manipulation of DNA at single-molecule level using optical tweezers (21), magnetic tweezers (22), atomic force microscopy (22), and tethered particle motion assay (23) have enabled the measurement of its mechanical properties (24,25) and expanded our understanding of its interactions with proteins or small ligands (26–29). The mechanical properties of DNA polymer are described by worm-like chain model using two parameters: the contour length (L₀ = 0.34 nm) and the persistence length (A = 50 nm) to describe the force-extension curves of dsDNA (25). Previous studies have shown that each intercalator inserted into dsDNA leads to an increase in the contour length L₀ of DNA base pairs by 0.34 nm per mono-intercalator (30). Moreover, the binding of intercalators and major groove binders can significantly alter the behavior of the DNA overstretching transition occurring at 65 pN (31–33). While previous single-molecule assays have primarily focus on purified small molecules in well-defined buffers like HEPES, phosphate or Tris buffer (27–31), the contents of microbial cultures are far more complex, harboring a vast array of unknown substrates. Therefore, the application of single-molecule manipulations to the field of natural product chemistry for the direct analysis of complex mixtures remains unexplored. It is not yet clear whether the SMSA is applicable in the presence of complex sample systems, such as microbial cultures containing a wide variety of metabolites or proteins.

Here, we introduce a pioneering application of SMSA directly in microbial cultures or extracts to screen microorganisms that produce DNA-intercalating agents (Figure 1A). We demonstrated that a notable elongation of dsDNA (>15% of the extension of pure dsDNA) under high forces (>10 pN) serves as a distinctive feature of DNA intercalators. Importantly, this elongation remains unaffected by the presence of primary metabolites involved in the basic growth of microorganisms, such as carbohydrates, lipids, and amino acids. Moreover, the application of external force on dsDNA can enhance the binding affinity and number of intercalations, thereby increasing the sensitivity of detection. The high sensitivity and specificity of the method enable the identification of trace amounts of DNA intercalative agents from just 5 μl of microbial culture, eliminating the need for purification steps. It is noteworthy that the integration of magnetic tweezers with a microfluidic flow chamber allowed for buffer exchange, rendering it well-suited for measuring multiple samples in the same chamber. The ability of SMSA to effectively process complex samples, including those potentially containing DNA intercalators with genotoxicity risks, underscores its suitability for environmental monitoring applications (23).

Figure 1.

Experimental design. (A) Single-molecule stretching assay for microbial culture using magnetic tweezers. (B) A 6618 bp dsDNA tethered between a paramagnetic bead and a coverslip was used as a sensor to detect the presence of DNA intercalating agents. The chemical structure of daunorubicin (compound 1). (C) Typical bead height-force curves measured in the presence of various daunorubicin concentrations. The increase in bead height indicates the elongation of dsDNA due to intercalation by daunorubicin. (D) Titration curves of dsDNA elongation measured at different forces. Data points were expressed as means ± standard deviations. Lines were fitted to the data by the Hill function. (E) Force-dependent binding constants were approximated to the Arrhenius-Bell mode.

Materials and methods

Reagents

Daunorubicin hydrochloride was purchased from Shanghai Yuanye Bio-Technology Co., Ltd (China). The chemical solvents including petroleum ether, ethyl acetate, dichloromethane and methanol used for extraction and column chromatography were obtained from Hubei Shenshi Chemical Technology Co., Ltd (China). The HPLC-grade solvents (methanol and acetonitrile) used for LC and LC–MS analysis were procured from Fisher Chemical/Merck (USA/GER). Trifluoroacetic acid (TFA, HPLC-grade) was acquired from Energy Chemical (China). Deuterated solvents, including methanol-d₄ and chloroform-d, were sourced from Cambridge Isotope Laboratories, Inc. (UK).

Strains and culture conditions

The strain of S. coeruleorubidus CICC11043 was acquired from China Center of Industrial Culture Collection. S. tanashiensis DSM 731, Serratia symbiotica DSM 23270, S. fonticola DSM 4576, S. nematodiphila DSM 21420, Xenorhabdus szentirmaii DSM 16338 and X. ishibashii DSM 22670 were obtained from Leibniz Institute DSMZ-German Collection of Microorganisms and Cell Cultures GmbH. Streptomyces sp. Gö66 was generously provided by Prof. Dr Axel Zeeck from the University of Göttingen, Germany. S. coelicolor A3(2), S. lividans K4-114, S. albus J1074 and Escherichia coli BL21 (DE3) are general hosts maintained in our lab. Talaromyces pinophilus R5SSF1, T. funiculosus R5SSF2, T. flavus var. flavus R5HwF7, T. radicus BLYF1 and BLYF1 were isolated from Lycium barbarum at 105°97′E, 38°28′N, Hui Autonomous Region, Ningxia, China. Aspergillus sp. CarHF15 was obtained from Curcuma aromatica Salisb at 116°25′E, 39°47′N, Nanning, Guangxi, China.

All fungal strains were individually inoculated onto potato dextrose agar (PDA) plates and incubated at 28°C for 7 days. Subsequently, the fungus-growing PDA agar was cut into small pieces and transferred into sterilized rice media composed of 120 g of rice and 180 mL of double-distilled water (dd H₂O) at 28°C for 20 days. For Gram-positive bacteria, cells were initially cultivated on GYM agar plates (1 l: 4 g of d-glucose, 4 g of yeast extract, 10 g of malt extract, 2 g of CaCO₃, pH 7.2 and 20 g of agar) at 30°C for 5 days for sporulation. Subsequently, the spore suspension of cells was transferred into M2 media (1 l: 4 g of d-glucose, 4 g of yeast extract, 10 g of malt extract, pH 7.0) at 30°C, 200 rpm for 7 days, except for S. tanashiensis DSM 731, which was inoculated into R4 medium (1 l: 10 g of d-glucose, 10 g of MgCl₂·6H₂O, 5.6 g of 2-[tris(hydroxymethyl)methylamino]-1-ethanesulfonic acid, 4 g of CaCl₂·2H₂O, 3 g of l-proline, 1 g of yeast extract, 0.2 g of K₂SO₄, 0.1 g of casamino acids, 2 ml of trace element solution, pH 7.2). The trace element solution was prepared by dissolving 40 mg of ZnCl₂, 200 mg of FeCl₃·6H₂O, 10 mg of CuCl₂·2H₂O, 10 mg of MnCl₂·4H₂O, 10 mg of Na₂B₄O₇·10H₂O and 10 mg of (NH₄)₆Mo₇O₂₄·4H₂O in 1 l of dd H₂O. Six Gram-negative bacteria were separately cultured on Luria-Bertani (LB) agar plates (1 l: 10 g of tryptone, 5 g of yeast extract, 10 g of NaCl, pH 7.2, and 20 g of agar) and subsequently inoculated into trypticase soya broth (TSB) media for fermentation (1 l: 30 g of TSB powder dissolved in 1 l dd H₂O) at 30°C, 200 rpm for 3 days.

Single-molecule stretching assay (SMSA)

The SMSA used in this study employed single-molecule magnetic tweezers for the analysis (BioPSI, Singapore). The measurements were conducted with a 6618 bp PCR product amplified from bacteriophage DNA (λDNA) (Thermo Fisher Scientific, USA) using a pair of primers labeled with biotin and thiol group at the respective 5′ end (Sangon Biotech Co., Ltd, China). A flow-channel was prepared as previously described (34). The 5′-thiol end of dsDNA was attached to the coverslip surface via a sulfosuccinimidyl 4-(N-maleimidomethyl) cyclohexane-1-carboxylate (sulfo-SMCC) crosslinker (Huateng Pharma Co., Ltd, China), while the 5′-biotin-end of dsDNA was attached to 2.8 μm-diameter streptavidin-coated paramagnetic beads Dynal M280 (Thermo Fisher Scientific, USA). The force was applied to the magnetic beads by a pair of permanent magnets positioned above sample. The bead-height was measured from the diffraction patterns of the beads using magnetic tweezers. The bead-height force curves were obtained through a force-jump procedure, and at each force, the magnet was held for 5 s to measure the bead-heights. All single-molecule experiments were conducted at room temperature of 20–23°C.

Bacterial samples preparation for single-molecule stretching assay

A 0.3 ml of bacterial liquid culture was centrifugated at 12 000 rpm for 5 min and the supernatant was heated at 95°C for 5 min to inactivate DNase before SMSA. For agar cultures, the bacterial colony was picked from the agar plate and extracted with 1 ml of ddH₂O and centrifugated at 12 000 rpm for 5 min to collect the supernatant for SMSA. For crude extracts from fungi, 50 g rice culture was extracted with 50 ml of ethylacetate (EtOAc) and the solvent was evaporated under vacuum to obtain the crude extract. The crude extract from each microorganism was accurately weighed and dissolved in methanol to ensure consistent concentration (1 mg/ml). The dissolved crude extract was diluted 100-fold into 1× PBS buffer (pH 7.3) before SMSA. Finally, 5 μl of the analyte was added to the flow chamber.

HPLC guided fractioning

The crude extract of S. coeruleorubidus was dissolved in 500 μl of MeOH and subsequently analyzed using HPLC. The analysis was performed employing an Agilent Technologies 1260 Infinity system equipped with an Agilent Proshell 120 EC-C₁₈ column (2.7 μm, 150 × 3.0 mm). Liquid chromatography was conducted with acetonitrile containing 0.01% TFA (v/v) as solution A and MilliQ H₂O containing 0.01% TFA (v/v) as solution B. A linear gradient from 10% to 45% A was applied over 0 to 25 min, followed by a rapid gradient from 45 to 100% A from 25 to 30 min. The column was washed for 10 min with 100% A and then equilibrated with an isocratic flow at 10% A from 40 to 45 min. Chromatography of compounds was continuously monitored by measuring absorbance at 210 and 480 nm. To determine the presence of daunorubicin, the crude extract was fractionated into 4 fractions at a flow rate of 0.6 ml/min, with each fraction collected for 10 min per tube. All tubes containing HPLC fractions were dried using a vacuum centrifugal concentrator (Beijing JM Technology, China). After drying, 100 μl of methanol was added to each fraction and then diluted 100 times with 1 × PBS solution. The prepared samples (100 μl) were added to the channel for SMSA.

LC–MS analysis

The filtered samples were analyzed by an Agilent & 1290 Infinity II/6545 QTOF LC/MS using an Agilent Zorbax Eclipse Plus C18 RRHD column (1.8 μm, 50 × 2.1 mm) with a linear gradient of 5% to 100% solvent A (solvent A: 0.1% HCOOH in CH₃CN; solvent B: 0.1% HCOOH in H₂O) over 17 min at a flow rate of 0.4 ml/min.

Phylogenetic and gene cluster analysis

The sequences used in this study were obtained from the National Center for Biotechnology Information (NCBI, www.ncbi.nlm.nih.gov/) and analyzed using the BLAST algorithm. The phylogenetic tree was constructed via the maximum likelihood algorithm implemented in MEGA 7, based on multiple sequence alignment by MUSCLE. Bootstrap values were calculated after 1000 replications. The accession numbers of selected sequences were listed in Supplementary Table S1. The nucleotide comparison of gene cluster generated using Easyfig 2.2.5 (35).

Purification and structural elucidation of DNA intercalators

S. tanashiensis DSM 731 was grown in 5 l of R4 media for 7 days at 30°C with shaking at 200 rpm. The culture was then extracted with EtOAc four times at room temperature and the solvent was evaporated under vacuum to obtain 586.0 mg of crude extract. This extract was subjected to silica gel column chromatography (100–200 and 200–300 mesh, Qingdao Marine Chemical Inc., Qingdao, China) and eluted with a gradient of petroleum ether–ethyl acetate (v/v 10:1, 5:1, 1:1, 1:5, 1:10, 0:1) and CH₂Cl₂–MeOH (v/v 20:1, 10:1, 1:1, 0:1) to afford eight fractions (fractions A–H). Fraction D (51.2 mg) was further purified through semipreparative HPLC (equipped with a Cosmosil 5C₁₈-MS-II column, 5 μm, 250 × 10 mm) using 50% acetonitrile in H₂O as the mobile phase, resulting in the isolation of compound 3 (7.6 mg, t_R 34.280 min). Fraction G (129.8 mg) was separated by Sephadex LH-20 column chromatography (Cytiva, CH₂Cl₂: MeOH 1:1), followed by purification using semipreparative HPLC with isocratic elution (18% acetonitrile in H₂O) to yield compound 2 (4.5 mg, t_R 32.084 min) (Supplementary Figure S1).

T. funiculosus R5SSF2 was grown on 480 g of rice media for 20 days at 28°C. The fermented rice was then extracted with EtOAc three times at room temperature, and the solvent was evaporated under vacuum to obtain 10.2 g of crude extract. This extract was chromatographed on a silica gel column, eluted with a gradient of PE–EtOAc (v/v 10:1, 5:1, 1:1, 1:5, 1:10, 0:1) to give six fractions (fractions A–F). Fraction C (1.7 g) was separated by Sephadex LH-20 column chromatography CH₂Cl₂: MeOH 1:1) to afford four fractions (C1–C4). Fraction C2 (0.6 g) was fractionated by silica gel column chromatography (eluted with a gradient of PE–EtOAc from 10:1 to 1:10, v/v) to yield five fractions (C2-1–C2-5). Fraction C2-4 (51.2 mg) was further purified by semipreparative HPLC with 55% acetonitrile in H₂O as the mobile phase to obtain compound 4 (7 mg, t_R 29.578 min) (Supplementary Figure S1).

All nuclear magnetic resonance (NMR) data were recorded on a Bruker AM-400 spectrometer with TMS as the internal standard. The structures of compounds 2–4 were confirmed by comparison of their NMR and HRESIMS data with literature values (NMR and HRESIMS) and were corroborated with literature values (Supplementary Figures S2–S18).

Cytotoxicity assay

The HCT-116 human colon cancer cell line was a generous gift from Prof. Shunichi Takeda. The HCT-116 cells were initially cultured in high-glucose DMEM (Wuhan Servicebio Technology Co., Ltd) supplemented with 10% of fetal bovine serum (FBS, Pan-Biotech) and 100 U/ml penicillin/streptomycin (Beyotime Biotechnology, China). Cells were seeded at a density of 3000 cells/well in a 96-well plate and incubated for 24 h. Subsequently, cells were treated with the test compounds at different concentrations. After 48 h incubation at 37°C, MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide, BioFroxx, Germany) in 1 × PBS was added to each well at a final concentration of 0.25 mg/ml. The plate was further incubated for 2 h, followed by replacement of the medium with 100 μl of DMSO to solubilize the formazan products. The absorbance of the wells at 570 nm was measured using a Synergy H1 microplate reader (BioTek Instruments, Inc., USA). The IC₅₀ value was determined by fitting the relative cell viability to .

Results

External forces increase the sensitivity and specificity for detecting DNA intercalating agents

For proof of principle, we employed a single-molecule magnetic tweezer to analyze a purified DNA intercalator, daunorubicin that is a widely used anthracycline antibiotic to treat various types of cancer (36). The experimental design using magnetic tweezers was shown in Figure 1B, where a 6618 bp dsDNA was tethered between a paramagnetic bead and a cover glass slide, serving as a sensor to detect the DNA interacting agents (11,27). By monitoring the extension of dsDNA using magnetic tweezers and applying forces ranging from 1 to 75 pN, the extension changes induced by DNA binding small molecule can be measured. The elongation of dsDNA measured in the presence of different concentration of daunorubicin was shown in Figure 1C. Dissociation constants K_d(F) for daunorubicin at each applied force were fitted using the Hill equation (37), , where is the elongation of dsDNA at ligand concentration c and force F, is the saturate dsDNA elongation n is the Hill coefficient (Figure 1D). The dissociation constant obtained at each applied force, denoted as K_d(F), is the inverse of the binding affinity K_a(F).

We observed a significant enhancement in the sensitivity for detecting DNA intercalators in response to applied force, which could be attributed to several factors. Firstly, DNA is resistant to being straightened in solution due to thermal fluctuations (25), which causes challenges in detecting intercalator-induced elongations. According to the worm-like chain model, a force exceeding k_BT/A (>0.1 pN) is needed to extend the dsDNA, where k_B is Boltzmann constants and T is the absolute temperature. The high spatial resolution of our magnetic tweezers (∼2 nm at forces >5 pN) allowed the detection of few molecules binding (0.34 nm/intercalation), thereby achieving sensitive detection. Secondly, the ligand-induced elongation is sensitive under high forces as force can increase the number of intercalators bound to dsDNA (30). For instance, in the presence of 20 nM daunorubicin, insignificant elongation (<50 nm) was observed at low forces (<10 pN), whereas substantial elongation (329 ± 26 nm) occurred at 60 pN, representing approximately to 14 ± 1% (fractional elongation) of the extension of pure dsDNA (Figure 1D). Thirdly, external forces applied to the dsDNA significantly enhance the binding affinity of ligands to dsDNA at force F, as determined by zero-force binding affinity K_a(0) using the Arrhenius–Bell model (29):, where is the transition distance. By the fitting of K_a measured at different forces to this equation, we obtained zero-force K_a (0) = (0.71 ± 0.13) ×10⁶ M^–1, and Δx = 0.21 ± 0.01 nm. The K_a increased ∼10-fold to (7.6 ± 2.1) × 10⁶ M^–1 at 60 pN (Figure 1E). The increase of DNA binding affinity under external force is a general property for dsDNA intercalators (29), thus this approach can be applied to wide range of unknown DNA intercalators, enabling the detection of low-abundance DNA intercalating agents from microorganisms.

In addition to lengthening dsDNA, the binding of ligands to dsDNA also altered the behavior of the DNA overstretching transition, providing a sensitive indicator for the presence of DNA interacting reagents (31–33). At a critical force of ∼65 pN, pure dsDNA showed sudden 70% increase in extension to 0.58 nm per bp (Figure 1B, black curve), indicating the transition of B-form DNA to S-DNA (stretched DNA). At 2 nM daunorubicin, the overstretching transition force increased, suggesting that 2 nM daunorubicin binds and stabilizes B-form DNA. The detection limit for strong DNA intercalator like daunorubicin is ∼20 nM, as the overstretching transition at ∼65 pN disappears in the presence of 20 nM daunorubicin. Given the small sample volume (5 μl) required for detection, SMSA can detect ligand–DNA interactions using only 0.052 ng of daunorubicin.

Importantly, the application of force also enhances the specificity for detecting DNA intercalators. This is because the lengthening of dsDNA at high force (>10 pN) is a specific characteristic of DNA intercalators (26). Solution conditions, primarily ionic strength, mainly influence the persistence length of dsDNA but do not significantly increase its contour length (38). Therefore, changes in ionic strength do not lead to substantial elongation (<10% of the extension of pure dsDNA) at high force. Most DNA-binding proteins change the dsDNA extension by stiffening (increasing persistence length), bending, looping or wrapping dsDNA, thereby reducing the extension at low force (<10 pN). However, they do not cause lengthening of dsDNA at high force (22). To our knowledge, only RecA and single-stranded DNA (ssDNA) form a filament that binds to dsDNA and can result in the lengthening of dsDNA (39). Therefore, we speculate that SMSA can directly detect DNA-intercalating agents from microbial culture. We also conducted buffer-exchange experiments to confirm that the non-covalently bound daunorubicin could be effectively removed through washing. The flow-chamber design facilitates multiple measurements using a single DNA tether, enhancing the experimental efficiency (Supplementary Figure S19).

Validating the suitability of SMSA for detecting DNA intercalators in microbial culture using S. coeruleorubidus

To verify the suitability of SMSA for directly analyzing microbial culture, we investigated the liquid culture of the known daunorubicin-producing bacterium S. coeruleorubidus. Bacteria typically generate primary metabolites in the early stage and significant quantities of secondary metabolites during the late-exponential or stationary phases. By monitoring the absorbance at 480 nm, a characteristic wavelength for anthracyclines, we confirmed the presence of daunorubicin or its derivatives in the bacterial supernatants collected at various stages of culture growth (Figure 2A). The dsDNA stretching curve with supernatants from early-phase bacterium culture (cultivated for 70 h) (Figure 2B, red) and M2 medium (Supplementary Figure S20, blue) overlapped with that of pure DNA (black), indicating no significant elongation. However, supernatants from late-exponential phase culture (cultivated for 178 h) induced significant dsDNA elongation (>450 nm) (Figure 2C), suggesting the presence of DNA-intercalating agents. The significant elongation observed at forces ranging from 10 to 60 pN, accompanied by the disappearance of the B-to-S phase transition at 65 pN, is a characteristic of dsDNA intercalators. This result suggests that the dsDNA elongation at high forces (>10 pN) is highly specific to DNA intercalators, remaining unaffected by primary metabolites involved in the basic bacteria growth, such as carbohydrates, lipids, and amino acids. To broaden the scope of the applicability to various type of microorganism cultures, we also investigated single bacterium colonies from the agar plates cultured with S. coeruleorubidus (Figure 2D). Each colony, along with the agar (weighing 135 ± 11 mg) was immersed in 1 ml of water and subjected to extraction using an ultrasonic bath for 30 minutes. Subsequently, 5 μl of the resulting supernatants were introduced into the flow chamber for SMSA. The bead height-force curve measured at force ranging from 10 to 60 pN exhibited significant elongation (ranging from 215 ± 77 to 499 ± 27 nm), indicating the presence of DNA-intercalating agents in the bacterial colonies.

Figure 2.

Validation of SMSA for analyzing culture of Streptomyces coeruleorubidus. (A) Production of daunorubicin and its derivatives was quantified based on the absorption at 480 nm. The data represent the mean values ± standard deviations obtained from three independent culture supernatants collected at 12-hour intervals. Additionally, photographs of culture flasks captured at 70 and 178 h were included for references. (B, C) SMSA results for the addition of 5 μl of bacterial supernatants from cultures cultivated for 70 h (B) and 178 h (C) were shown. The curves were color-coded: black, pure DNA; red, supernatants. (D) SMSA for bacterial colony. A single colony of S. coeruleorubidus grown on an M2 agar plate was selected and extracted with 1 mL of ddH₂O, then 5 μl of the supernatants were subjected to the assay. (E) HPLC fragmentations of the crude extracts derived from S. coeruleorubidus. UV spectra of crude extracts at 210 nm (upper panel), 480 nm (middle panel), and daunorubicin standard (lower panel). (F) Extracted ion chromatogram (EIC) of (528.1884, [M + H]⁺) from daunorubicin standard, crude extracts and Fraction III.

Subsequently, high performance liquid chromatography (HPLC) was employed to fractionate the crude extracts (1 mg) of S. coeruleorubidus culture into four fractions at 10-minute intervals (Figure 2E), enabling the separation of polar and nonpolar compounds. The resulting four fractions were subjected to the SMSA, with fraction III displayed the highest average elongation compared to the other fractions (Supplementary Figure S21). This suggests the presence of DNA binding reagents in fraction III, consistent with the retention time of the daunorubicin standard. Liquid chromatography-high resolution mass spectrometry (LC-HRMS) analysis further confirmed the presence of the molecular ion peak at m/z 528.1884 [M + H]⁺ for daunorubicin in both the crude extract and fraction III (Figure 2F). Additionally, we observed the presence of DNA intercalating compounds in fraction II as well. This likely stems from daunorubicin derivatives produced by S. coeruleorubidus, as the biosynthetic pathway for daunorubicin also produces a variety of anthracycline derivatives (4). Taken together, our findings demonstrate that the SMSA can facilitate the identification of DNA intercalators from trace amounts of bacterial culture during chromatography purification without the need for isolating individual compounds. This is particularly valuable for unknown DNA-interacting agents lacking specific absorbance and molecular weight information.

Identification of microorganisms producing DNA interacting agents using SMSA

It is noteworthy that the DNA-interacting agents are not universally present in secondary metabolites of all organisms. To assess the prevalence of microorganisms capable of producing DNA intercalators across a diverse range, we employed the SMSA to analyze microbial cultures from 17 distinct species, including representatives of Gram-positive bacteria, Gram-negative bacteria and fungi (Figure 3A, Supplementary Figure S22 and S23). The strains were classified based on the sequences of the Internal Transcribed Spacer (ITS) in fungi and 16S rRNA in bacteria (Supplementary Table S1). Following cultivation of the strains, the supernatants from bacterial cultures and the EtOAc crude extracts from fungi (1 mg dissolved in 0.1 ml DMSO or MeOH) were individually subjected to the same SMSA. The results revealed that the supernatants of S. tanashiensis DSM 731 exhibited the significant lengthening of dsDNA at forces >30 pN, suggesting the presence of dsDNA intercalators in the culture (Figure 3B). In contrast, all the analyzed Gram-negative bacterial cultures showed less than 10% fractional elongation at 60 pN (Supplementary Figure S23). Additionally, the extension of dsDNA measured in the presence of crude extracts produced by fungus T. funiculosus R5SSF2 containing DNA intercalators, evidenced by a significant elongation (943 ± 101 nm, fractional elongation = 42 ± 4%) of DNA at 60 pN (Figure 3C), suggests that this species may also hold the potential for producing DNA intercalators.

Figure 3.

Screening of microorganisms that produce DNA-intercalating agents. (A) Phylogenetic analysis of the microorganisms used in this study. The culture supernatants of bacteria and the fungal crude extracts (0.1 mg/μl) were used for SMSA. Fractional elongation was estimated based on the elongation measured at 60 pN. Error represents standard deviations calculated from three independent measurements. The strains highlighted in red were used for subsequent analysis. (B, C) SMSA for the supernatants of S. tanashiensis DSM 731 (B) and the crude extract of T. funiculosus R5SSF2 (C). (D) The nucleotide comparison between the daunorubicin gene cluster (dnr) from S. coeruleorubidus and medermycin gene cluster (med) from S. tanashiensis generated using Easyfig 2.2.5. Grey shading indicates nucleotide similarity between two gene clusters (67–100%). Functional regions of gene clusters were indicated: polyketide synthase (PKS): red, deoxysugar (Sugar): orange, post-PKS tailoring (Tailoring): blue, Regulation/Export: purple. (E) Extracted ion chromatogram (EIC) of medermycin (458.1858, [M + H]⁺) from the crude extract of S. tanashiensis DSM 731.

Natural products are synthesized by specific gene clusters within the genomes of their respective organisms. To preliminarily identify the types of DNA-interacting agents produced by S. tanashiensis DSM 731, we analyzed its complete genome sequence (CP084204) using antiSMASH (The antibiotics and Secondary Metabolites Analysis Shell) (40). The analysis of S. tanashiensis DSM 731 revealed the presence of medermycin biosynthetic gene cluster (med), indicating its ability to producing medermycin (41). Both medermycin and daunorubicin are classified as anthracycline polyketides, which shares a similar anthracycline ring structure. Daunorubicin is biosynthesized by a type II polyketide synthase (PKS) gene cluster known as dnr (42). Comparison of the dnr and med clusters showed a similarity of at least 67%. These gene clusters can be categorized into four parts: PKS related genes for anthracycline core (PKS), deoxysugar genes (Sugar), the tailoring genes for methylation and oxidation (Tailoring), and regulatory genes (Regulation/Export) (Figure 3D). Based on these findings, we hypothesized that medermycin and its derivatives are the intercalating agents presented in S. tanashiensis DSM 731 bacteria supernatants, causing the observed elongation in SMSA measurements. To verify this hypothesis, we analyzed crude extracts of S. tanashiensis DSM 731 using LC-HRMS. The extracted ion chromatogram (EIC) showed a single prominent peak with an m/z value of 458.1858 ([M + H]⁺) (Figure 3E), suggesting that S. tanashiensis DSM 731 produces medermycin. However, we were unable to perform biosynthetic gene cluster analysis (genome mining) for the fungus T. funiculosus due to the unavailability of its genome.

Purification and characterization of DNA intercalators from the Gram-positive bacterium and the fungus

To further characterize the DNA intercalating compounds from the cultures of the Gram-positive bacterium S. tanashiensis DSM 731 and the fungus T. funiculosus, we conducted large-scale fermentation for these two strains. Guided by SMSA, we purified the target DNA intercalators, resulting in the isolation of compounds 2–4, respectively. Subsequent analysis using ¹H and ¹³C NMR spectroscopic data, as well as high-resolution electrospray ionisation mass spectrometry (HRESIMS) data revealed that compounds 2–3 isolated from S. tanashiensis DSM731 were identified as medermycin and kalafungin (43), originally discovered from Streptomyces sp. K73 in 1976. Compounds 2–3 belong to the family of pyranonaphthoquinone antibiotics, known for their antibacterial and anticancer bioactivity. Compound 4, isolated from fungi T. funiculosus was structurally characterized as the known compound ligustrone B. The intercalation of medermycin has been analyzed using the ethidium bromide displacement assay (44), while the intercalative binding abilities for kalafungin and ligustron B have not been previously reported.

Using SMSA, we further individually analyzed the interactions of purified compounds 2–4 (Figure 4A). The force-extension curves of dsDNA at serial concentrations of compounds 2–4 showed the concentration-dependent elongation from 1 to 60 pN (Figure 4B). Notably, the overstretching transitions were significantly influenced by compounds 2–4 at concentrations beyond 10 μM. The elongation−concentration curves of compounds 2–4 were fitted with the Arrhenius-Bell model (Supplementary Figure S24). This allowed us to obtain the zero-force binding constants K_a of compounds 2–4 when intercalating into dsDNA: (3.5 ± 0.3) × 10³ M^–1 (compound 2), (1.8 ± 0.2) × 10³ M^–1 (compound 3) and (10.5 ± 1.7) × 10³ M^–1 (compound 4), respectively. It is noteworthy that stretching curves under low force reveal distinct intercalating features among these compounds. Compared to compound 3, compound 2 showed a more significant increase in dsDNA elongation at forces below 15 pN, suggesting a higher degree of ligand intercalation at low forces. This implies that the electrophilic side chain of compound 2 facilitates its insertion between dsDNA bases at low forces.

Figure 4.

Characterization of purified DNA-interacting agents. (A) The chemical structures of medermycin (2), kalafungin (3), ligustrone B (4). (B) Bead height-force curves measured in the presence of various concentrations of compounds 2–4. (C) IC₅₀ value of compounds 2–4 measured in HCT-116 cells. The y-axis represents relative cell viability compared to non-treated cells (control). The error bars represent the standard deviations from three independent replicates.

We also assessed the cytotoxic potential of compounds 2–4 against human colon cancer HCT-116 cells using MTT assay to measure cell proliferation. Medermycin and kalafungin exhibited remarkable antiproliferative effects against HCT-116 cells, with half-maximal inhibitory concentration (IC₅₀) of 52 ± 6 and 70 ± 7 nM, respectively. In contrast, compound 4 showed only very weak activity against HCT-116 cells with IC₅₀ greater than 2 × 10⁴ nM (Figure 4C). The significant activity of compound 2 is consistent with previous reports, emphasizing its therapeutic potential in cancer treatment (45). The low activity of compound 4 is likely attributed to its poor solubility, which may impede its cellular uptake and target engagement.

Discussion

In this study, we present a highly sensitive and specific single-molecule assay that enables the direct detection of DNA-intercalating ligands from minute amounts of microorganism cultures. This mechanical assay is highly specific for DNA intercalators and is unaffected by the presence of various metabolites, eliminating the need for complex sample separation or purification. Our assay requires only nanogram quantities of DNA interacting reagents, which is <1% of the amount needed for any other methods, thus facilitating the rapid screening of microorganisms producing DNA-intercalating reagents with low abundance. Analyzing microbial culture at the single-molecule level provides valuable insights into their binding mechanisms, further accelerating the discovery of bioactive compounds.

A potential limitation of SMSA is the competition from groove binders such as netropsin (46), and DNA binding proteins for intercalation sites (47). This competition could reduce the assay's sensitivity by outcompeting the test compound for binding. We employed SMSA to confirm that neither netropsin nor the DNA binding peptides (KWKWKKA) induced dsDNA lengthening (Supplementary Figures S25 and S26). This suggests that dsDNA lengthening remains a reliable indicator for DNA intercalators. Furthermore, we investigated daunorubicin binding using SMSA in the presence of netropsin and DNA binding peptides using SMSA. Although the presence of these competitors did decrease the assay sensitivity, as evidenced by reduced dsDNA elongation with 200 nM daunorubicin, complete inhibition of daunorubicin was not observed even at 10 μM of netropsin and peptides.

One of the major challenges in conventional natural product discovery is the laborious workflow involved in fermenting and isolating natural products from microbes, which often results in the rediscovery of known compounds. Advances in high-throughput sequencing and mass spectroscopy technologies have made massive data on microbial genomes, transcriptomes and proteomes readily accessible. While bioinformatic tools excel at identifying conserved biosynthetic gene clusters responsible for the production of compounds, such as polyketides, non-ribosomal peptides, terpenes, etc. (40), they fall short in predicting the actual DNA binding properties of secondary metabolites. In this context, integrating SMSA with genome mining strategies and mass spectroscopy analysis can streamline the identification process for new chemical scaffolds with DNA-binding properties.

In our study, the isolated compounds medermycin and kalafungin showed modest DNA intercalating activity but significant inhibition of cancer cell proliferation at sub-micromolar concentrations (Figure 4). This implies that the force-enhanced binding affinity facilitated by SMSA enables the identification of weak DNA intercalators from microorganism cultures while still exhibiting potent bioactivity. This may be attributed to the fact that many DNA-interacting drugs also interfere with DNA-associated proteins such as topoisomerase (8), RNA polymerase and impact chromatin biology (9,10). For instance, etoposide, a clinically used camptothecin derivative acting as a topoisomerase I poison, has been demonstrated to intercalate into bare DNA (48). Unlike typical DNA intercalators like ethidium bromide, topotecan is a weak dsDNA intercalator (48) but interacts at the interfaces between DNA and topoisomerase I, forming a ternary complex (8). The discovery of novel DNA intercalators may contribute to the identification of novel non-camptothecin topoisomerase I inhibitors. Additionally, RNA intercalators also have a wide range of applications, exemplified by the first approved small molecule treatment for Spinal Muscular Atrophy (SMA), which binds to RNA hairpins (49). By substituting the dsDNA tether with dsRNA or RNA/DNA hybrid tethers, the current approach can also be adapted to detect RNA-interacting small molecules from microorganisms (50).

Advanced single-molecule techniques, such as magnetic tweezers, optical tweezers, and nanopore sensors, hold substantial promise for natural product discovery (21,51). These methodologies allow for precise and targeted measurements of various interactions, including those between ligands and DNA or proteins (52). While it is regrettable that we did not discover any DNA-interacting agents with new chemical scaffolds due to the limited numbers of microorganisms screened, our findings still underscore the effectiveness of SMSA in specifically detecting DNA-interacting agents. Our flow-cell enabled continuous measurements of different fungal crude extracts using a single dsDNA molecule (Supplementary Figure S23), however, dsDNA scission during the measurements limited the maximum number of extracts analyzed per flow-cell. By refining microfluidic systems, implementing automation and high-throughput magnetic tweezers (53), the throughput of SMSA can be significantly improved, thus expanding its potential applications in drug discovery endeavors.

Data availability

The data underlying this article are available in the article and in its online supplementary material.

Supplementary data

Supplementary Data are available at NAR Online.

Funding

National Natural Science Foundation of China [32171225, 32371496, 31972852, 81903527]; Key project at central government level: The ability establishment of sustainable use for valuable Chinese medicine resources (2060302). Funding for open access charge: Key project at central government level: The ability establishment of sustainable use for valuable Chinese medicine resources (2060302).

Conflict of interest statement. The authors submitted a patent application for the SMSA method used in this work.