Spermidine is a common component of the eDNA-dependent biofilm matrix

Abstract

Biofilms are structured multicellular bacterial communities encased within an extracellular matrix comprised of exopolysaccharides, proteins, extracellular DNA (eDNA), and other biopolymers that provide protection against environmental stressors. We and others have shown that eDNA serves as a fundamental structural element common to even multispecies biofilms. During biofilm maturation, ubiquitous DNABII proteins facilitate the conversion and stabilization of eDNA into the rare and rigid Z-DNA conformation, thereby enhancing matrix integrity and rendering the underlying eDNA resistant to nucleases. We have previously shown that the removal of positively charged molecules, such as DNABII proteins, results in rapid, significant disruption of diverse biofilms. Here, we identify the polyamine spermidine as another essential positively charged molecule that, together with DNABII proteins, contributes to the development and maintenance of the eDNA-dependent extracellular matrix. We also provide evidence that SPD is present within the biofilm matrix alongside DNABII proteins in multiple bacterial pathogens. Our findings indicate that SPD and DNABII proteins cooperate to promote Z-DNA formation. Depletion of SPD and DNABII using cation exchanger P11-phosphocellulose or inhibition of SPD synthesis via dicyclohexylamine impaired biofilm formation and destabilized preformed biofilms. These results suggest that polyamine synthesis or accumulation represents a potential target for biofilm disruption and control.

Introduction

Bacteria in the biofilm state can exist either as aggregated cells or as surface-attached communities on both biotic and abiotic surfaces. Bacteria in biofilms reside within a self-produced extracellular matrix or extracellular polymeric substance (EPS), which consists of many components, including extracellular DNA (eDNA), proteins, lipids, polysaccharides, biopolymers, and divalent cations. This matrix provides a protective barrier against harsh environments, antimicrobials, and host immune effectors^1–3, making biofilms more resistant to host immune responses and antibiotics^4–7. As a result, pathogens in the biofilm state are difficult to eradicate and pose a substantial problem to medicine^4,8. According to the National Institutes of Health (NIH), up to 80% of human bacterial infections require the biofilm state for pathogenicity⁹. Moreover, biofilm-associated diseases impose a significant economic burden, with global healthcare costs estimated to be around $500 billion annually¹⁰.

While biofilm stability is influenced by species-specific matrix components, a key factor that contributes to the structural integrity of mature biofilms is the conversion of eDNA from the DNase-sensitive B-DNA form to the DNase-resistant Z-DNA form, which accumulates due to the activity of positively charged DNABII proteins such as integration host factor (IHF) and histone-like protein (HU)¹¹. Removal of these proteins by use of either specific antibodies^1,12–14 or a cation exchanger such as phosphocellulose P11 can collapse the biofilm matrix¹⁵. Similarly, depletion of divalent cations from biofilms via the use of EDTA increases the susceptibility of bacteria, such as the otopathogen nontypeable Haemophilus influenzae (NTHI), to antibiotic treatment¹⁶. These observations suggest the importance of positively charged extracellular molecular species in the maintenance of biofilms.

Among the important positively charged molecular species are polyamines (PAs), commonly found in all living organisms, where their intracellular concentrations can reach up to millimolar levels^17–20. Similar to DNABII proteins, PAs such as spermine (SPM) and spermidine (SPD) promote the conversion of B-DNA to Z-DNA^21,22. Their DNA-binding affinity correlates with net positive charge, following the order SPM⁺⁴ > SPD⁺³ > putrescine (PUT)⁺²/cadaverine (CAD)⁺²²⁰. Although SPM is more potent than SPD in the conversion of B- to Z-DNA²¹, it is absent in most bacteria^23,24.

Due to their cationic nature, PAs interact with negatively charged macromolecules, including DNA, RNA, and proteins, thereby modulating their structure and function^21,22,25. By neutralizing negative charges on the DNA, PAs help to promote its condensation into higher-order structures, thereby regulating DNA accessibility to transcription factors and RNA polymerases, contributing to the gene regulation and protection against oxidative damage^26–28. Major cellular PAs such as SPD, PUT, and SPM are abundant and extensively studied^18,29 whereas CAD, ornithine (ORN), and norspermidine (NSPD) are less abundant due to rapid turnover¹⁷. Collectively, PAs participate in diverse biological processes, including cell proliferation, apoptosis, gene regulation, and oxidative stress protection^{19,20,30–33}.

Most bacteria synthesize PAs, with SPD being essential for planktonic growth in multiple species^34–36 and critical for biofilm formation^37–39. While PAs promote biofilm formation at physiological levels, high exogenous concentrations of SPD, SPM, or CAD can inhibit biofilm formation by Pseudomonas aeruginosa and Staphylococcus aureus^40,41. Inhibiting PA synthesis also affects bacterial growth; for example, dicyclohexylamine (DCHA), a spermidine synthase inhibitor, has shown strong antibacterial activity, causing complete growth arrest of Streptococcus uberis and partial inhibition of S. aureus, Escherichia coli, and P. aeruginosa^42,43. This growth inhibition effect can be rescued by exogenous SPD; however, the impact of DCHA on biofilm development and viability remains poorly characterized.

In this study, we show the presence of the polyamine SPD in the matrix of the biofilms formed by diverse pathogenic bacteria. As the biofilms mature, there is an accumulation of SPD, which is concomitant with the accumulation of DNABII proteins and DNase-resistant Z-DNA. We also provide evidence that SPD and HU are required to shift the equilibrium of eDNA into the Z-DNA form, an essential step in the maintenance of biofilms. We show that the mechanism by which the cation exchanger P11 phosphocellulose prevents biofilm formation, as well as disrupts biofilms, relies on the ability of this resin to deplete specific cationic molecules, namely SPD and DNABII proteins. Indeed, the biofilm disruption effect was rescued by supplying exogenous SPD and HU, which suggested the importance of these components in the formation and maintenance of the biofilms. DCHA, an inhibitor of the SPD biosynthesis pathway, also showed an inhibitory effect on biofilm formation. These results highlight the importance of polyamines in the biofilm lifecycle and generate potential new targets to control biofilm-associated infections.

Methods

Bacterial strains

Clinically isolated NTHI strain 86-028NP⁴⁴, S. mutans strain UA159⁴⁵, UPEC strain UTI89⁴⁶, and K. pneumoniae strain TOP52⁴⁷; and P. aeruginosa strain ATCC 27853, were used for this study.

Bacterial growth

NTHI was grown in Brain Heart Infusion (BHI) broth supplemented with hemin (2 µg/ml) and β-NAD (2 µg/ml) (e.g., sBHI). Uropathogenic Escherichia coli (UPEC) and K. pneumoniae were grown in Luria Broth (LB), P. aeruginosa was grown in Tryptic Soy Broth (TSB), and S. mutans was grown in Todd Hewitt Broth (THB). All cultures were grown at 37 °C in a 5% CO₂ humidified incubator.

In vitro biofilms

Biofilms were allowed to form in a LabTek^TM 8-well glass-bottom chambered slide (Fisher Scientific 155411) and maintained with replenishment of the respective media at 16 and 24 h until the end of incubation. NTHI was grown as indicated above, then diluted to 2.5 × 10⁵ CFU/ml in 200 µl sBHI. For UPEC and K. pneumoniae, overnight cultures were grown on LB agar, suspended in LB broth to an OD₄₉₀ of 0.65, diluted (1:12 in LB), and incubated statically for 2.5 h. The cultures were then further diluted to 2.5 × 10⁵ CFU/ml in 200 µl LB and inoculated into an 8-well glass chamber slide. P. aeruginosa biofilms were grown as above, but without an additional 2.5 h of incubation. After 40 h, the pellicle at the air-liquid interface was collected onto a glass slide. S. mutans grown overnight in THB was diluted 1:25 in THB broth and incubated at 37 °C in 5% CO₂ to an OD₄₉₀ of 0.65. The culture was then further diluted to an OD₄₉₀ of 0.05 and seeded into an 8-well glass chamber slide (Cellvis, C8-1.5H-N).

Antibodies and reagents

The following primary antibodies were utilized: mouse anti-dsDNA (B-DNA; Abcam, ab27156), mouse IgG2a (Invitrogen, 02-6200), IgG2b (Invitrogen, 02-6300); rabbit anti-Z-DNA (Absolute Antibodies, Ab00783-23.0), mouse anti-Z-DNA (Ab00783-3.0), rabbit IgG isotype control (Abcam, ab172730), rabbit anti-putrescine (Cloud-Clone, PAX051Ge01), in-house polyclonal guinea pig anti HU_NTHI, rabbit polyclonal anti-putrescine (Cloud-Clone Corp., PAX051Ge01), rabbit polyclonal anti-cadaverine (GeneTex, GTX 13202), rabbit polyclonal anti-spermidine (Abcam, Ab7318). Secondary antibodies included, goat anti-rabbit IgG conjugated to Alexa Fluor® 488 (Invitrogen, A11008), goat anti-mouse IgG conjugated to Alexa Fluor® 405 (Invitrogen, A31553), goat anti-rabbit IgG conjugated to Alexa Fluor® 488 (Invitrogen, A31556), goat anti-guinea pig IgG conjugated to Alexa Fluor® 488 (Invitrogen, A11073). Rabbit IgG was purified from naïve serum as previously described⁴⁸. This isotype naïve serum served as a control for nonspecific staining.

Induction of Z-DNA using polyamines

Genomic DNA (gDNA) from NTHI and poly(dGdC) (InvivoGen) were brominated using 4 M NaCl and bromine water, as described previously⁴⁹, with a few modifications as follows. gDNA substrate (200 µg) was incubated overnight in 10.5 ml 20 mM Na citrate buffer with 4 M NaCl supplemented with 500 µl bromine water (RICCA, 1195-16) or with SPD (Sigma, S2626-1G) (500 µM) or HU_NTHI (200 nM) or with the combination of both, at room temperature for 2 h. Bromination reduces the energetic threshold for B-DNA conversion into Z-DNA but is not a requirement for Z-DNA formation under physiological conditions⁵⁰. Z-DNA formation was determined by the absorbance ratio of 260–295 nm as described previously^51,52. Absorbance values were obtained by a Biotek Synergy H1 spectrophotometer at 260 and 295 nm wavelengths. Buffers without DNA substrates served as reference blanks.

Pulmozyme protection assay

Brominated gDNA (2 µg) was treated with SPD, PUT (0–100 µM), or HU_NTHI (50, 100, 1000 nM) alone or in combination and incubated with and without 0.5 units Pulmozyme® (Genentech, dornase alfa, NDC Code 50242-100-40) for 10 min at 37ᵒ C. EDTA (5 mM) was added to stop the reaction, and samples were resolved by agarose gel electrophoresis and visualized using a Bio-Rad Gel Doc system.

Cadaverine-mediated biofilm inhibition assay

NTHI and S. mutans biofilms were initiated in the presence of 0, 1, 3, or 30 mM CAD (prevention), or 24-hour-old preformed biofilms were treated with 0, 1, or 3 mM CAD (Sigma, D22606) (disruption). Biofilms were stained with LIVE/DEAD® stain, fixed, and visualized using CLSM. COMSTAT software was used for biomass and thickness quantification⁵³. To assess the effect of SPD and HU_NTHI on the CAD effect, co-treatments were applied with 3 mM CAD and 1 mM SPD and/or 500 nM HU_NTHI.

P11 phosphocellulose activation and biofilm assays

P11 resin was sequentially washed with Milli-Q water, 0.5 N NaOH, and 0.5 N HCl, then equilibrated in 20 mM sodium phosphate buffer. For biofilm assays, NTHI biofilms were initiated with or without P11 (0.5%, 1%, or 1.5% w/v; prevention assay) and incubated for 16 h or 24 h old, preformed NTHI biofilms were incubated with P11 (0.5%, 1%, or 1.5% w/v; disruption assay) for 16 h. For transwell assays, biofilms were initiated in the basolateral chamber of the transwell system (Costar, 3470) for 16 h with P11 in the apical chamber (0.5%, 1%, or 1.5% w/v; prevention assay) or 24 h, and then P11 was added to the apical chamber (0.5%, 1%, or 1.5% w/v; disruption assay) for 2 h. To investigate the time of action of P11, biofilms were treated with P11 for 10/30/60/120 min in the disruption assay as mentioned above. To assess the counteracting effect of SPD/PUT (MP Biomedicals, 100450) or HU_NTHI on the effect of P11, SPD/PUT (0.5/1/3 mM), or HU_NTHI (250/500/2500 nM) or SPD (1 mM) with or without HU_NTHI (500 nM) were added in the basal chamber at the initiation of the biofilms. Biofilms were either analyzed via Live/Dead staining or for investigation of SPD/ HU_NTHI. The average biomass and thickness of the biofilms were calculated as described above.

Visualization of biofilm-associated components

Treated biofilms were washed two times with PBS and stained with primary antibodies against B-DNA (1 μg; mouse monoclonal [35I9 DNA] to ds DNA; Abcam, ab27156), Z-DNA (1 μg; rabbit monoclonal α-Z-DNA [Z22]; CedarLane Labs, Ab00783-23.0), SPD (Abcam, ab7318), HU_NTHI (rabbit polyclonal), isotype controls (mouse IgG2a isotype Control; Invitrogen, 02-6200/ rabbit monoclonal isotype control; Abcam, ab172730) (1:200 dilution), in PBS that contained 5% (w/v) BSA for 2 h at room temperature. After incubation, biofilms were washed two times with PBS and were incubated with goat α-rabbit IgG conjugated to AlexaFluor® 405 (Invitrogen, A11032), goat α-mouse IgG conjugated to AlexaFluor® 594 (Invitrogen, A11001) secondary antibodies (1:200) and counterstained with FM^TM 4–64 (Invitrogen, T3166) 5 µg/ml in PBS that contained 5% BSA. Biofilms were then washed once with PBS and imaged using a ×63 oil objective on a Zeiss LSM 800 microscope.

Co-localization of SPD and HU

Biofilms were stained to visualize SPD and HU as mentioned above and were imaged using a Zeiss LSM 800 confocal microscope (Carl Zeiss Inc.). Co-localization analysis was carried out using NIS-Elements AR (Nikon) with the workflow created to calculate the Pearson co-localization coefficient (PCC). The PCC values were plotted using GraphPad Prism. A PCC > 0.5 shows strong, between 0.3 and 0.5 shows moderate, and 0–0.3 shows weak co-localization. A value of 0 or below shows a negative co-localization.

Growth curves of planktonic NTHI and UPEC

Planktonic growth curves of NTHI were initiated at OD490 of 0.02 in sBHI supplemented with or without P11 (1.5% w/v) or SPD/CAD/PUT/SPM (3 mM) or dicyclohexylamine (DCHA, 50 µM) in a 96-well plate and grown for 16–24 h at 37 ᵒC using a Biotek Synergy H1 multi-mode plate reader.

Dicyclohexylamine treatment of UPEC biofilms

UPEC biofilms were initiated with or without DCHA (Sigma, 185841-5G) (50 µM) for 16 h in Lysogeny Broth (LB). To test the effect of SPD on the effect of DCHA, SPD (1 mM) was added with or without DCHA. The treated biofilms were either analyzed for Live/Dead staining or for investigation of SPD using immunofluorescence. The average biomass and thickness of the biofilms were calculated as mentioned above.

Quantification and statistical analysis

Statistical significance was determined by either one-way ANOVA or paired t-test with Tukey’s multiple comparisons or Kruskal–Wallis nonparametric test with Dunn’s multiple comparison, using GraphPad Prism (version 10.3.1). Data are presented as mean ± SEM and significance is shown as *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. All statistical details are provided in the respective figure legends and include the statistical tests used and the exact value of n. Each biological replicate included three technical replicates. Wherever applicable, the data from each experiment were normalized to the control within the corresponding biological replicate and are presented as a percentage of the control.

Results

Phosphocellulose-mediated sequestration of positively charged molecular species inhibited biofilm formation and promoted biofilm disruption

Positively charged molecules are critical for all biological processes and, in particular, mediate the structural integrity of DNA, in part, by binding to the negatively charged phosphates of DNA nucleotides. Indeed, DNA-binding proteins utilize basic (positively charged) amino acids to facilitate binding to DNA via its negatively charged phosphates through electrostatic interactions. When this charge neutralization occurs on one side of the DNA, bending ensues. This is a common mechanism for DNA-bending proteins, including the DNABII proteins HU and IHF⁵⁴.

Previous studies have shown that depletion of divalent cations, such as Mg²⁺, from biofilms increases NTHI susceptibility to antibiotic treatment¹⁶, while targeted depletion of DNABII proteins with specific antibodies results in rapid, significant biofilm collapse^1,12–14. Similarly, we found that when NTHI biofilms were grown in the basolateral chamber of a transwell system with a cation exchange resin (P11 phosphocellulose) in the apical chamber, biofilms underwent rapid destabilization¹⁵. Herein, we first expanded upon these observations to further characterize the molecular mechanisms involved.

To investigate the mechanism underlying this effect, we employed P11, a well-characterized cation-exchange resin commonly used as a DNA mimetic and for affinity purification of DNABII proteins^55,56. To evaluate the inhibitory potential of P11 on biofilm formation, NTHI biofilms were initiated and maintained in the presence of P11 (0%, 0.5%, 1%, and 1.5% w/v) for 16 h (Fig. 1a). To assess biofilm disruption, 24-h NTHI biofilms were treated with the same concentrations of P11 for an additional 16 h (Fig. 1b). Following treatment, the biofilms were imaged using confocal laser scanning microscopy (CLSM) and average thickness and biomass were calculated using COMSTAT. As indicated in Fig. 1a and b, P11 significantly inhibited biofilm development and effectively disrupted established biofilms in a dose-dependent manner. In addition, we also assessed the potential of P11 as a surface-coating agent for biofilm prevention. Chamber slides were pre-coated with increasing concentrations of P11 (0–1.5% w/v), and NTHI biofilms were cultured on the coated surfaces for 16 h. Coating with P11 significantly reduced biofilm formation in a concentration-dependent manner (Supplementary Fig. 1a). Importantly, P11 did not inhibit bacterial growth at the 1.5% w/v concentration, indicating that its antibiofilm effect is not due to direct bacteriostatic or bactericidal activities (Supplementary Fig. 1b).

Fig. 1. P11 phosphocellulose sequesters SPD and HU and induces biofilm collapse.

a NTHI biofilms were initiated in the presence of 0–1.5% (w/v) P11 and grown for 16 h (Prevention). b Mature 24 h biofilms were treated with 0–1.5% (w/v) P11 for 2 h (Disruption). For both a and b, biofilms were stained with a live/dead^TM viability dye and imaged by CLSM. Biomass and average thickness were quantified relative to untreated controls. Data represent mean ± SEM (n = 4). c Schematic of the transwell system. P11 is placed in the apical chamber, while biofilms were grown in the basolateral chamber. The chambers were separated by a 0.4 μm membrane that allows diffusion of small biomolecules (double-headed arrows) but prevents passage of P11 resin and bacterial cells. d 24 h NTHI biofilms grown in the transwell system were exposed to 0–1.5% w/v P11 in the apical chamber for 2 h. The average % thickness (relative to control) was quantified. Data represent mean ± SEM (n = 8). Scale bar, 30 µm. e NTHI biofilms were grown for 16 h in the basolateral chamber in the presence or absence of 1.5% (w/v) P11 in the apical chamber. Biofilms were immunostained for HU (green, upper panel) or spermidine (SPD; pink, lower panel), and bacterial cells were stained with FM^TM 4–64 (shown in gray). Naïve sera served as the negative control. Representative CLSM images are shown. Fluorescence intensities were quantified in ImageJ, normalized to the untreated control, and plotted beside the images (n = 6). 24 h old biofilms of NTHI, S. mutans and UPEC were grown under the conditions described in e and supplemented in the basolateral chamber with 1 mM SPD or 500 nM HU (recombinant protein based on respective organism), or both. Biofilms were stained and imaged using CLSM. Representative images from NTHI biofilms are shown in (f). Scale bar, 30 µm. g Average thickness was quantified relative to untreated media controls and is shown as a percentage of the control. Data represent mean ± SEM (n = 4). Statistical significance was determined by one-way ANOVA followed by Kruskal–Wallis nonparametric t-test (a, b, d and g) and for e the Mann–Whitney t-test was used (ns = p > 0.05; *p < 0.05; **p < 0.005; ***p < 0.0005; ****p < 0.0001). P11 disrupts biofilms by depleting essential positively charged species, specifically HU and spermidine. Restoration of biofilm structure occurs only when both HU and spermidine are supplied together.

Because P11 functions by sequestering positively charged molecules, we next asked whether its antibiofilm activity required direct contact with the biofilm. To test this, we utilized a transwell system in which NTHI biofilms were cultured in the basal chamber, while P11 was added to the apical chamber separated by a 0.4 μm pore membrane (Fig. 1c), impermeable to the P11 resin. This arrangement permits diffusion of small molecules, such as DNABII proteins, polyamines (PAs), and divalent cations, while preventing the passage of P11 resin and bacterial cells. The apical chamber was supplemented with 0% (sBHI control), 0.5%, 1%, and 1.5% (w/v) P11 and incubated for 2 h, after which biofilms were analyzed for relative biomass and average thickness. As shown in Fig. 1d, incubation with P11 significantly reduced biofilm thickness, confirming that the antibiofilm effect of P11 does not require direct physical interaction with the biofilm. To assess the kinetics of P11 action, 24-hour NTHI biofilms were treated with P11 for 10, 30, 60, or 120 min in this transwell setup. Remarkably, exposure for as little as 10 min was sufficient to trigger biofilm disruption (Supplementary Fig. 1c).

Given the established role of DNABII proteins in stabilizing extracellular DNA and promoting the conversion of B-DNA to Z-DNA¹¹, we hypothesized that P11 disrupts biofilms by depleting DNABII proteins from the matrix. Immunofluorescence staining for HU confirmed a marked reduction in HU within P11-treated biofilms (Fig. 1e). The importance of DNABII proteins in the maintenance of biofilms is due to their ability to stabilize the structural integrity of eDNA and drive the DNase-sensitive B-DNA to DNase-resistant Z-DNA¹¹. To test whether restoring DNABII proteins (e.g., HU isolated from NTHI; HU_NTHI) could rescue biofilms from P11-mediated disruption, we supplemented cultures with increasing concentrations of HU (250–2500 nM) while simultaneously exposing them to 1.5% P11 in the apical chamber. Supplementation with exogenous HU partially restored biofilm stability, with recovery observed only at the highest concentration tested (550 nM) (Supplementary Fig. 1d). This finding suggested that other positively charged molecules might also be depleted by P11 treatment. Given that the P11 is a cation exchanger, we also tested whether the addition of cations such as Ca²⁺ and Mg²⁺ could rescue the biofilms from the disruptive effect of P11. However, Addition of these molecules did not restore biofilms in the presence of P11 (Supplementary Fig. 1e). Given this outcome, as well as prior evidence that certain polyamines, such as the SPD, can facilitate the conversion of B-DNA to Z-DNA^21,22, SPD became a stronger potential candidate for biofilm restorative capacity following P11 treatment.

Consistent with this hypothesis, immunofluorescence staining revealed that P11 treatment significantly reduced SPD levels within NTHI biofilms (Fig. 1f). Supplementation with SPD across a range of concentrations (0–3 mM) demonstrated that only the highest concentration (3 mM) effectively restored biofilm in the presence of P11 (Supplementary Fig. 1f). In contrast, supplementation with PUT, a polyamine that cannot induce Z-DNA, did not rescue the biofilm (Supplementary Fig. 1g). This result was consistent with the ability of SPD and not PUT to induce Z-DNA.

Whereas when used at lower concentrations, neither SPD nor HU was individually capable of fully restoring the biofilm state in the presence of P11 (Supplementary Fig. 1d and f), we investigated whether co-supplementation of SPD and HU might be able to synergistically counteract P11-mediated biofilm prevention. NTHI, uropathogenic E. coli (UPEC) and Streptococcus mutans (S. mutans) biofilms were initiated in the basal chamber of the transwell system, which contained either 1 mM SPD or 550 nM HU alone or a cocktail of both (HU from respective organism was used), while either 0% or 1.5% P11 w/v was added to the apical chamber, then grown for 16 h. Only co-treatment with both SPD and HU significantly restored biofilm formation (Fig. 1f and g).

Together, these findings demonstrated that phosphocellulose P11 disrupts NTHI biofilms by sequestering essential positively charged components—specifically DNABII proteins and the polyamine SPD—thereby impairing Z-DNA formation and destabilizing the extracellular DNA scaffold. Importantly, this disruption occurs without direct physical interaction with the biofilm, indicating that depletion of these cationic factors alone is sufficient to compromise biofilm integrity.

Polyamines are present in in-vitro biofilms

Given the apparent requirement for SPD in NTHI biofilms, we used immunofluorescence to investigate the presence of SPD within the biofilms formed by additional diverse and clinically relevant Gram-positive and Gram-negative bacterial pathogens. 40 h old biofilms formed by NTHI, UPEC, P. aeruginosa, S. mutans, or Klebsiella pneumoniae were probed with anti-spermidine (SPD) antibodies to detect the presence of SPD. Additionally, bacterial cells within the biofilms were membrane-stained (FM^TM 4–64), and the DNA-binding DNABII protein HU was immunofluorescently labeled^11,48. These images were analyzed to calculate the Pearson co-localization coefficient (PCC). A PCC of 1 indicates complete co-localization; values between 0.5 and 1 indicate strong co-localization; 0.3–0.5 indicates moderate co-localization; 0–0.3 indicates weak co-localization; 0 or below indicates negative co-localization; and −1 indicates a perfect inverse correlation (exclusion)⁵⁷. All biofilms tested positive for the presence of SPD, which co-localized with HU (Fig. 2a, Supplementary Fig. 2).

Fig. 2. Presence of polyamine spermidine in biofilms grown in vitro.

a 40 h old biofilm of NTHI, UPEC, P.aeruginosa, S. mutans, and K. pneumoniae was immunolabeled to detect spermidine (SPD, pink), DNABII protein HU (green), and cells (FM^TM 4–64, white). Use of naïve serum as a negative control is shown in the upper panel for each organism. Biofilms were visualized using CLSM and images were reconstructed with AxioVision 3.0. Representative images out of 3 independent biological replicates (n) are shown for each organism. b NTHI and UPEC biofilms grown for 24, 40, and 90 h were immunolabeled for SPD and HU. 90-h-old biofilms are also stained with naïve sera. Representative CLSM images out of 3 independent biological replicates (n) are shown for each organism. Scale bar 30 µm. Spermidine and HU are conserved matrix components across multiple bacterial biofilms, and their abundance changes dynamically as biofilms mature.

Further analysis of NTHI and UPEC biofilms revealed that as these biofilms matured over 24, 40, and 90 h, there was a progressive accumulation of SPD, which paralleled the increased accumulation of HU (Fig. 2b). This trend was concomitant with prior findings, which demonstrated a similar temporal accumulation of Z-DNA in mature biofilms¹¹.

Collectively, the findings presented herein demonstrate that SPD accumulates within the biofilm matrix in association with HU protein, suggesting that PAs are integral components of the EPS of bacterial biofilms and contribute to biofilm structural maintenance.

SPD and HU facilitate Z-DNA formation

The DNA-binding DNABII protein HU plays a crucial role in Z-DNA formation, which contributes to the structural integrity of the biofilm matrix and enhances eDNA resistance to DNase-mediated degradation¹¹. We wondered if this B- to Z-DNA conversion could be recapitulated in vitro by admixing only DNA and HU, and or SPD. Indeed, confocal imaging analysis revealed that incubating SPD with genomic DNA (gDNA) isolated from NTHI resulted in Z-DNA formation at levels comparable to those observed with HU alone (Fig. 3a). Moreover, the combination of HU and SPD further enhanced Z-DNA formation, implying the two work in harmony to convert and stabilize the Z-DNA structure (Fig. 3a).

Fig. 3. HU- and SPD-mediated Z-DNA formation in the EPS mimetic model.

a Genomic DNA (gDNA) isolated from NTHI and brominated was incubated for 16 h in the presence or absence of SPD, recombinant HU_NTHI, or both, then immunolabeled with antibodies specific for B-DNA (cyan) and Z-DNA (yellow); naïve serum served as a negative control, and DNA was counterstained with DAPI. Representative CLSM images from three independent replicates are shown. b The A₂₆₀/A₂₉₅ absorbance ratio was used to assess DNA conformation following incubation of brominated gDNA with SPD and HU or NaCl (positive control). The ratio of A₂₆₀/A₂₉₅ is plotted for each condition, where a ratio >8.6 indicates the presence of B-DNA, while a ratio of ~3.2 suggests the formation of Z-DNA^11,58. The data are presented as mean ± SEM of n = 9 independent experiments. c Brominated gDNA was incubated with increasing concentrations of SPD and HU, individually or in combination, for 1.5 h at 37 °C, followed by treatment with Pulmozyme® for 20 min. DNA degradation was assessed using agarose gel electrophoresis. Statistical analysis was performed using one-way ANOVA followed by Tukey’s multiple comparisons test, with significance levels indicated as follows: ns = p > 0.05, **p < 0.005, ***p < 0.0005. Co-incubation of HU and SPD promotes Z-DNA formation and thereby enhances the resistance of DNA to DNase digestion.

To quantify this structural shift, we measured the optical density (OD) ratio at 260/295 nm, where ratios ~4 indicate a predominance of Z-DNA and ratios ~11 reflect B-DNA primarily, while the ratio between the two suggests the presence of transitionary or mixed population^11,49. Consistent with the imaging results, both SPD and HU, when added together, lowered the OD ratio below 4, indicating Z-DNA formation; 4 M NaCl, a known inducer of Z-DNA, served as a positive control (Fig. 3b). We also tested whether the conditioned media collected from the biofilms could induce Z-DNA. Conditioned media collected from 24 h old NTHI biofilms were incubated with gDNA with or without SPD or HU. Z-DNA formation was observed only when HU was supplemented with conditioned media (Supplementary Fig. 3).

Since Z-DNA is known to be resistant to DNase degradation^11,58, we tested whether the Z-DNA induced by SPD and HU alone or in combination protected gDNA from DNase I degradation. Increasing concentrations of SPD or HU individually conferred partial DNase resistance; however, their combination achieved significant protection at lower concentrations (Fig. 3c). PUT cannot convert B-DNA to Z-DNA formation^21,59; therefore, we used it as a negative control. Unlike SPD, PUT did not confer significant DNase resistance (Supplementary Fig. 4), highlighting a specific role for SPD in stabilizing Z-DNA within biofilms.

Cadaverine inhibited biofilm formation and disrupted existing biofilms by inhibiting Z-DNA accumulation

PAs exhibit varied impacts on bacterial biofilms. In P. aeruginosa, exogenous supplementation of CAD promotes planktonic growth while inhibiting biofilm accumulation and also disrupts biofilms⁴⁰. To evaluate whether this effect extends to other species, we used NTHI, a Gram-negative and S. mutans a Gram-positive pathogens. Biofilms were initiated in the presence of CAD at varying concentrations (1, 3, and 30 mM). After 16 h, biofilms were analyzed for biomass and average thickness. In S. mutans, 1 mM CAD significantly reduced biofilm formation (Fig. 4a, b). In contrast, NTHI biofilms were unaffected at 1 mM CAD; however, both 3 and 30 mM CAD resulted in marked decreases in biomass and thickness (Fig. 4b). To assess CAD-mediated disruption of established biofilms, 24-h S. mutans and NTHI biofilms were treated with CAD for an additional 16 h. CAD at 1 and 3 mM significantly disrupted biofilms of both species (Fig. 4a, b). Importantly, 3 mM CAD, as well as other tested polyamines (e.g., SPD, SPM, PUT), did not affect planktonic growth of NTHI (Supplementary Fig. 5a).

Fig. 4. Effect of Cadaverine on biofilms.

a NTHI or b S. mutans biofilms were initiated with or without 1-, 3-, or 30-mM CAD and grown for 16 h (prevention) or 24 h old biofilms were treated with 1- or 3-mM CAD for 16 h (disruption). Biofilms were stained with live–dead stain and imaged using CLSM, and biomass and average thickness relative to % of control were plotted. The data are presented as mean ± SEM of n = 6 independent experiments. c NTHI biofilms were initiated in the presence or absence of 1- or 3-mM CAD (prevention), or d 24-h biofilms were treated with or without 1 or 3 mM CAD (disruption). Biofilms were stained to identify Z-DNA (red) and B-DNA (cyan) using specific antibodies, and bacterial cells were stained with FM^TM 4–64 (gray). The corresponding fluorescence intensity ratios (B-DNA/Z-DNA; B-DNA/Cells) were quantified and shown below the confocal images. Scale bar 30 µm. e NTHI biofilms were initiated in sBHI media alone or supplemented with 3 mM CAD with or without 1 mM SPD, 500 nM HU_NTHI, or a combination of 1 mM SPD and 500 nM HU_NTHI. The ratios of B-DNA/Z-DNA, B-DNA/Cells, and Z-DNA/Cells were plotted as mentioned above. P values (ns = p > 0.05, *p < 0.05, **p < 0.005, ***p < 0.0005, ****p  < 0.0001) using one-way ANOVA followed by Tukey’s multiple t-test (a, b) and Kruskal–Wallis nonparametric t-test (c–e). CAD prevents and disrupts biofilm by reducing Z-DNA accumulation, an effect that can only be reversed by the combined addition of HU and SPD.

Multiple mechanisms have been proposed to explain the inhibitory effects of CAD on biofilms⁴⁰. CAD does not promote the B-to-Z-DNA transition, although it can bind B-DNA^21,60. Given that HU and SPD act together to promote Z-DNA formation (Fig. 3), and Z-DNA formation is crucial for the biofilm integrity and presence¹¹, we hypothesized that the addition of CAD interferes with HU–SPD-mediated Z-DNA formation of eDNA. To test this hypothesis, NTHI biofilms were initiated or treated post-formation with 1 or 3 mM CAD. Immunofluorescence analysis demonstrated a significant reduction in Z-DNA accumulation at 3 mM CAD in both prevention and disruption conditions (Fig. 4c, d).

Further, CAD treatment (3 mM) significantly reduced the steady-state accumulation of both HU and SPD within NTHI biofilms. This reduction was reversed only when HU and SPD were supplied together, but not individually (Supplementary Fig. 5b). Similarly, co-treatment with the SPD and HU restored the levels of Z-DNA (Fig. 4e) and biofilm biomass (Supplementary Fig. 5c), but not with either of them alone. We also tested whether CAD can inhibit the Z-DNA formation in the in vitro assay. CAD at the concentration of 30 mM was able to inhibit SPD and HU-mediated Z-DNA formation, while CAD at 1 or 3 mM did not show a similar effect (Supplementary Fig. 5d).

Together, these findings demonstrate that CAD impairs biofilm formation and maintenance by reducing HU and SPD availability in the biofilm matrix, thereby preventing Z-DNA formation. Restoration of Z-DNA and biofilm architecture only when HU and SPD are supplied together supports the model that CAD disrupts biofilm integrity by interfering with the cooperative HU–SPD-dependent stabilization of Z-eDNA.

Inhibition of spermidine biosynthesis by dicyclohexylamine suppressed biofilm formation

As we observed the importance of the SPD in biofilm formation and maintenance, we next tested the effect of an inhibitor of the SPD biosynthesis pathway, DCHA. Effects of the spermidine synthase inhibitor DCHA on E. coli growth have been well characterized⁴²; therefore, we used UPEC strain UTI89 to study the effect of DCHA on biofilms. UPEC biofilms were grown in the presence or absence of DCHA alone or DCHA + SPD and analyzed using CLSM. There was a significant reduction in the average thickness of biofilms when treated with DCHA (Fig. 5a). This effect was nullified upon the supplementation of exogenous SPD (Fig. 5a). We also found a significant reduction in SPD immunofluorescence when treated with DCHA, which was restored to that of media control when SPD was added exogenously (Fig. 5b and c). DCHA at 50 μM did not have any effect on UTI89 growth (Fig. 5d), which indicated that the effect was due to the reduction of SPD synthesis, which did not have an effect on growth rate.

Fig. 5. Spermidine synthase inhibitor inhibited biofilm formation by UPEC.

a UPEC biofilms were initiated in the presence or absence of DCHA, SPD (1 mM), or a combination of both for 16 h. Biofilms were stained with live-dead stain and imaged using CLSM, and biomass and average thickness relative to the % of control were plotted. The data are presented as mean ± SEM of n = 4 independent experiments. b UPEC biofilms were grown as for NTHI. These biofilms were stained to identify SPD (pink), and cells were stained with FM-464 (gray). Naïve serum is used as a negative control. Scale bar 30 µm. c The Fluorescence Intensities (FI) for SPD and cells from the CLSM images were calculated using ImageJ, and the normalized FI of SPD is shown. d Growth of UPEC in the presence or absence of DCHA was monitored for 16 h and plotted as a measure of OD at 490. The data are presented as mean ± SEM of n = 4 independent experiments. P values (ns = p > 0.05, *p < 0.05, **p < 0.005) using one-way ANOVA followed by Kruskal–Wallis nonparametric t-test. DCHA reduces SPD accumulation and biofilm formation in UPEC without affecting planktonic growth, and SPD supplementation counteracts the effect of DCHA.

Discussion

The biofilm matrix is maintained by the interactions of various macromolecules, but particularly eDNA. These molecular interactions govern the structural integrity of biofilm through the formation and maintenance of an underlying Z-DNA-dependent structure. Towards that goal, various positively charged biomolecules facilitate interactions between adjacent negatively charged phosphate groups along DNA strands, thereby stabilizing the DNA structures. Well-known amongst these biomolecules are the basic DNABII proteins, HU and IHF^48,61, that interact with DNA in part through electrostatic interactions. This stabilizing interaction is consistent with previous studies that demonstrate increased biofilm susceptibility to antibiotics following magnesium (Mg²⁺) depletion¹⁶, or targeted DNABII protein removal^1,12–14. In a previous study, we showed that the cation exchanger resin, phosphocellulose P11, a DNA mimetic, disrupts biofilms, releasing antibiotic-susceptible newly released bacteria¹⁵. In this study, we found that P11 treatment can also inhibit the formation of biofilms. We also showed that P11 acts rapidly and can disrupt biofilms in just 10 min. The effectiveness of P11 may rely on the rapid shift in the equilibrium of the positively charged molecules away from the biofilms, inducing the biofilm collapse.

We argue that P11 sequestered positively charged molecules, e.g., DNABII proteins and PAs, specifically SPD, that are necessary and sufficient to form and maintain biofilms. PA-capturing resins like P11 represent a promising strategy for the treatment and prevention of biofilm-associated infections through sequestration of essential cationic molecules, including polyamines and DNABII proteins. By depleting extracellular SPD and related polyamines required for biofilm maturation and matrix stabilization, P11 impairs biofilm formation and can reduce antibiotic tolerance. Polyamine depletion has been shown to sensitize bacteria to conventional antimicrobials, including cationic peptides and aminoglycosides⁶². In this context, P11-based surface coatings may be particularly attractive for indwelling medical devices, where prevention of biofilm establishment is critical.

However, several limitations must be considered. The role of polyamines in biofilm biology is species-specific. While polyamine depletion inhibits biofilm formation in certain pathogens (e.g., E. coli³⁷ and Yersinia pestis⁶³), others may react differently. For example, the reduction of SPD increases biofilm formation in Synechocystis sp. Strain PCC 6803⁶⁴, and removal of SPM can potentiate biofilm formation by Neisseria gonorrhoeae⁶⁵. Thus, therapeutic efficacy may vary depending on the organism and environmental context. In addition, practical considerations may limit clinical translation. Long-term stability of PA-capturing materials remains a concern, as protein adsorption or biofouling could mask active binding sites and diminish sequestration efficiency over time. At present, the use of P11 may be most suitable for topical or localized applications, such as device coatings, rather than systemic or oral administration. Furthermore, there is limited information regarding the effects of P11 on host tissues, underscoring the need for future studies evaluating biocompatibility and host responses. Collectively, while PA-capturing resins offer a mechanistically compelling antibiofilm strategy, their species-dependent effects and translational limitations warrant careful consideration.

PAs are ubiquitous small cationic molecules found in all living organisms. In bacteria, the most common PAs are SPD, CAD, and PUT, which are essential to numerous cellular processes⁶⁶. Although universally present, their concentrations and compositions vary across species. For instance, E. coli exhibits high PUT levels, while in many other bacteria and eukaryotes, SPD is one of the most common PAs^18,67. Polyamine detection is generally carried out using various methods, such as the enzyme-linked immunosorbent assay (ELISA), a combination of mass spectrometry (MS) with gas chromatography (GC) or liquid chromatography (LC), high-performance liquid chromatography (HPLC), and nuclear magnetic resonance (NMR). In parallel to these methodologies, fluorescence-based polyamine detection systems have also been developed⁶⁸. In the present study, we employed immunofluorescence to successfully detect and localize PAs within biofilms. Notably, we revealed a previously uncharacterized spatial association between PAs and DNABII proteins.

In a previous study¹¹, we demonstrated that DNABII proteins, particularly HU, mediate the transition of extracellular DNA (eDNA) from the canonical right-handed B-DNA conformation to the left-handed Z-DNA conformation. This structural transition is critical for biofilm maturation, as Z-DNA accumulation enhances matrix stability and confers resistance of the eDNA-dependent biofilm scaffold to nuclease degradation. Moreover, B- to Z-DNA conversion promotes a shift toward a biofilm-dominant state over planktonic growth¹¹. As biofilms mature, Z-DNA progressively accumulates and, through stabilization by DNABII proteins, reinforces the structural integrity of the matrix¹¹. In this study, we observed similar accumulation patterns for HU and SPD during biofilm maturation. Notably, SPD exhibited a non-uniform spatial distribution in early-stage in vitro biofilms, with enrichment at the liquid/solid interface. This pattern was most apparent in younger biofilms and in species that form comparatively weaker biofilms, whereas more mature biofilms displayed a more homogeneous distribution. This interface enrichment may reflect early biofilm establishment processes, where, through initial surface attachment and early matrix assembly at the liquid/solid interface, structural organization is initiated. Progressive B- to Z-DNA conversion is thought to begin in these regions and to contribute to stable surface anchoring and matrix consolidation. The co-localization of SPD with HU and its preferential enrichment at this interface are consistent with a role for SPD in facilitating B- to Z-DNA transition, potentially in concert with DNABII proteins. Although our data are limited and do not establish causality, our data are consistent with SPD being at the highest in situ concentration at the regions of the biofilms where B-DNA is converting to the Z-form.

Independent studies have also implicated PAs, including SPD and SPM, in facilitating this B- to Z-DNA transition²¹. Spermidine, a polyamine with a +3 charge, plays a dual role in DNA condensation and the B- to Z-DNA transition by neutralizing the negative charges of the DNA backbone. This electrostatic interaction allows DNA to overcome the natural repulsion between its phosphate groups, enabling it to condense into more compact structures^21,22,69. Our current findings are consistent with SPD, in conjunction with HU, being capable of promoting Z-DNA formation within NTHI biofilms. Beyond its role in DNA conformational dynamics, SPD is part of a broader class of polyamines that participate in numerous cellular processes. Intracellularly, PAs act as key regulators and structural stabilizers by interacting extensively with nucleic acids. These interactions help prevent DNA denaturation and contribute to the stabilization of nucleic acid structures⁷⁰. Among cellular nucleic acids, RNA in particular plays a central role in regulating polyamine homeostasis by serving as a major binding reservoir for these cations⁷¹. The binding of polyamines to RNA effectively buffers intracellular polyamine levels, thereby influencing the steady state levels of free polyamines available for other processes, including structural transitions such as the B- to Z-DNA conversion⁷¹. Although beyond the scope of this study, it remains to be seen if a limited extracellular RNA reservoir is in part responsible for the facilitation of the B- to Z-DNA transition.

Furthermore, PAs are essential for several key processes: they participate in catalysis, control the biosynthesis of nucleic acids (for example, SPD stimulates DNA synthesis)⁷², and are responsible for translation, stress response, and cell cycle regulation^33,71. In addition to these intracellular roles, polyamines are increasingly recognized as important extracellular factors in microbial communities. In both Gram-negative and Gram-positive bacteria, polyamines influence or are required for biofilm formation and stability⁷³. Taken together, these observations support a central role for SPD in biofilm biology. Our results further extend this concept by demonstrating that SPD contributes to the development and maintenance of the biofilms examined in this study, likely through its cooperative interaction with HU and its capacity to modulate the structural state of extracellular DNA.

Though PAs have an important role in biofilm formation, they can have contrasting effects; depending on the species under consideration, some facilitate biofilm formation and progression, while others have a negative effect on biofilm formation or existing biofilms^37,38,74,75. Similarly, extracellular addition of CAD inhibits P. aeruginosa biofilm accumulation. This inhibition by CAD may partially result from its impact on the levels of the second messenger bis-(3’-5’)-cyclic dimeric guanosine monophosphate (c-di-GMP) or its ability to modify the expression of adhesion proteins⁴⁰. Consequently, CAD can affect biofilm matrix components, leading to weaker biofilm attachment and reduced biofilm accumulation⁴⁰. We also observed such a detrimental effect of CAD on S. mutans and NTHI biofilms. Extracellular addition of CAD inhibited biofilm formation and disrupted preformed biofilms. As biofilms mature, a greater proportion of the extracellular B-DNA gets converted to Z-DNA¹¹; however, the addition of CAD to NTHI biofilms inhibited this conversion. The steady state levels of B-DNA remained elevated in the CAD-treated biofilms, while the amount of Z-DNA and overall biomass both decreased significantly, which suggested interference of CAD in the natural process of conversion of B-DNA to Z-DNA.

The capacity of PAs to bind and condense DNA is directly correlated in part with their positive charge density. Longer polyamines with higher positive charges bind more strongly to DNA and are more effective at inducing the B- to Z-DNA transition^76,77. The higher charge-to-mass ratio of PAs such as SPM (4+) and SPD (3+) is therefore significantly more effective in B to Z DNA conversion than their less-charged counterparts, PUT or CAD (2+)^60,76–79. However, a high concentration of the divalent PAs (PUT or CAD) can disrupt this process. At high concentration, these 2+ charged ions compete with the more highly charged SPD for the negatively charged phosphate-binding sites of DNA. This competition disrupts the specific, strong interactions required for SPD to effectively drive the B- to Z transition⁸⁰. CAD readily binds to B-DNA but cannot induce the B- to Z-DNA transition^21,60. This binding likely allows CAD to act as a competitive inhibitor, interfering with the B- to Z transition mediated by SPD. Given the observed interaction between SPD and HU in promoting Z-DNA, we hypothesized that CAD might interfere with this interaction. Indeed, CAD treatment significantly reduced the accumulation of both SPD and HU within the biofilm matrix and at the higher concentration (30 mM) inhibited SPD and HU-mediated Z-DNA conversion. This inhibitory effect in the biofilm was reversible, and co-supplementation with SPD and HU successfully counteracted the CAD-induced interference, fully restoring Z-DNA formation to stabilize the overall biofilm structure. This finding reinforces the critical and interactive function of SPD and HU in managing DNA dynamics within the biofilm environment.

As virtually all eubacteria express and/or utilize PAs, these molecules likely serve as a key component of the inclusive EPS along with DNA and DNABII proteins^11,81. SPD is one of the most prevalent polyamines found across numerous species⁶⁷ and is synthesized by the addition of an aminopropyl group to putrescine by spermidine synthase enzyme^82,83. Disruption of the spermidine synthase gene (speE) causes a severe defect in biofilm formation in E. coli³⁷; therefore, to directly assess the functional significance of SPD, we employed the SPD synthase inhibitor, DCHA. This inhibitor is known to effectively block SPD synthesis in several species, including S. uberis, S. aureus, E. coli, and P. aeruginosa⁴². Inhibition of SPD biosynthesis significantly impaired SPD accumulation and biofilm formation without affecting bacterial viability, thereby affirming SPD’s critical role in biofilm maintenance rather than simply in bacterial survival. While we successfully used DCHA to reduce SPD synthesis, a complementary approach to reducing its functional pool would involve enhancing its catabolism. For instance, the rate-limiting enzyme in polyamine catabolism, spermidine/spermine N1-acetyltransferase (SAT1)^84,85, efficiently catalyzes the acetylation of SPD, marking it for rapid degradation. Other enzymes that could achieve a similar inhibitory effect by actively reducing the functional pool of SPD include spermidine oxidase or an exogenous polyamine-specific deaminase^85–87. This targeted manipulation of the polyamine pool—whether through the inhibition of synthesis by DCHA or the enhancement of catabolism by such enzymes—could be a powerful tool to manipulate SPD steady-state level, which appears to be a key metabolic checkpoint controlling the EPS composition and the structural integrity of the bacterial biofilm.

In summary, our findings establish a critical role for SPD in the formation and structural stabilization of Z-DNA within biofilms. SPD facilitates this by working in conjunction with DNABII proteins, particularly HU, to drive the formation and stabilization of Z-DNA, which is essential for biofilm structural integrity. While SPD is a necessary structural agonist, our work identifies CAD as a potent antagonist. CAD exerts its inhibitory effect by disrupting the critical PA-HU interactions, thereby blocking the B- to Z-DNA transition and compromising biofilm stability. These insights provide a molecular framework for understanding how polyamine–protein interactions drive biofilm architecture. While agents such as P11 and DCHA demonstrate proof-of-principle efficacy by depleting or displacing key positively charged biofilm components, they establish yet another target for biofilm prevention and or disruption.

Author contributions

K.R.M. and S.D.G. designed the experiments. K.R.M. performed experiments and wrote and modified the initial manuscript draft. A.D., J.R.B., and S.B. performed preliminary experiments. L.O.B. and S.D.G. supervised the project. L.O.B. and S.D.G. conceptualized the study, reviewed, and edited the manuscript. All the authors read and approved the final manuscript.

Data availability

All relevant data are contained within the manuscript and supplementary files; additional data, if any, are available from the authors on appropriate request.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41522-026-01004-9.