Two ss-oligos were purchased from Beijing Tsingke Biotech Co., Ltd., including a 60-nt ss-oligo-L (TAATAGCTAATCATAGAATCTCTTTTAAATATAAAGTCATATCAAATTAAGAGAGAATTG) with and without FAM-labeling at the 5′ end, and a 21-nt ss-oligo (CTTACGCTGAGTACTTCGATT) with and without TAMRA-labeling at the 5′ end. The unlabeled ss-oligo samples were dissolved in deionized water to obtain a stock solution of 6 mg/mL, then mixed with 6.0 × 10⁻³ mg/mL Fam-labeled ss-oligo-L or TAMRA-labeled ss-oligo to track their distribution during coacervate droplet formation and separation. Q-dextran and TMB (≥99%) were purchased from Aladdin (Shanghai, China) and dissolved in deionized water. Hydrochloric acid (HCl, 36.0%–38.0%), hydrogen peroxide (H₂O₂, 30%), sodium hydroxide (NaOH), sodium acetate (NaOAc), acetic acid (AcOH), disodium hydrogen phosphate (Na₂HPO₄), glycine, and sodium citrate (C₆H₅Na₃O₇) were all purchased from Sinopharm Chemical Reagent Co., Ltd. All chemicals were used as received without further purification. Deionized water was used throughout the experiments.

Recent studies have demonstrated that FeS₂ exhibits a peroxidase-like (POD-like) activity (Cao et al., 2024; Meng et al., 2021), prompting our investigation into how the POD-like activity of pyrite affects droplet dynamics. Bulk pyrite was obtained from Taobao, a popular e-commerce platform in China. Pyrite microparticles (about 13 μm) were obtained by grinding and passing through a 1,000-mesh sieve. Pyrite purity was checked using X-ray diffraction (XRD) performed on a Rigaku-D/MAX-PC 2500 instrument, equipped with Cu Kα radiation and a graphite monochromator, operating at 40 kV and 200 mA. Samples were scanned over a 2-theta range of 3–70 with a step size of 0.02 and a counting time of 0.12 s per step.

The POD-like activity of pyrite was measured using substrate TMB in the presence of H₂O₂ under different reaction conditions. All experiments were performed at 23 °C. First, experiments were carried out in acetic acid–sodium acetate (AcOH–NaOAc) buffer (200 mM, pH 3.6), containing 1.0 mg/mL TMB, 0.10 mg/mL pyrite, and/or 0.50 mM H₂O₂. After the reaction had proceeded for 10 min, the absorbance of the colored oxidation products of TMB was monitored across 300–1,000 nm using a multi-mode microplate reader (BioTek Synergy™ H4, USA). Although TMBox exhibits stronger absorbance peaks at shorter wavelengths, the 652 nm peak was chosen for subsequent kinetic analyses because it provides the most stable and interference-free signal.

To evaluate the influence of different buffer compositions on pyrite activity, experiments were conducted in 160 mM/80 mM/40 mM AcOH–NaOAc, glycine–HCl, or citric acid–disodium hydrogen phosphate (CAPS) buffer at pH 3.6. After 10 min, the absorbance at 652 nm was measured, confirming the highest activity in 40 mM glycine–HCl. To determine the optimal reaction pH, the glycine–HCl buffer was adjusted using HCl/NaOH to pH 1.50, 2.33, 3.10, 3.60, 4.07, 4.70, 5.31, 6.31, 7.61, 8.53, and 10.1. The optimal pH for pyrite activity was found to be 4.07. Thus, all subsequent experiments were conducted under the conditions of 40 mM glycine–HCl at pH 4.07. To assess the effect of pyrite concentration, experiments were conducted using pyrite concentrations of 0.15 mg/mL, 0.10 mg/mL, 0.080 mg/mL, 0.040 mg/mL, and 0.020 mg/mL.

For kinetic parameter determination, all experiments were carried out in glycine–HCl buffer (40 mM, pH 4.07). The pyrite-only group contained 2.0 mg/mL TMB, 0.15 mg/mL pyrite and a concentration gradient of H₂O₂ from 0 to 5.0 mM (0, 0.078, 0.16, 0.31, 0.62, 1.2, 2.5, and 5.0 mM). The droplet-encapsulated pyrite group contained 3.0 mg/mL unlabeled ss-oligo (21-nt), 3.6 mg/mL Q-dextran, 0.15 mg/mL pyrite, and 2.0 mg/mL TMB. After incubating for 5.0 min, the same concentration gradient of H₂O₂ was added to initiate the reaction. The kinetic parameters were measured at 652 nm, 10 min after H₂O₂ addition. The Michaelis–Menten constants were obtained by fitting the data to the Michaelis–Menten equation using GraphPad Prism (10.1.1):

where v is the initial reaction velocity, [S] is the substrate concentration, V_max is the maximal reaction rate observed at the saturating substrate concentration, and K_M is the substrate concentration at which the reaction rate is half of V_max. v is calculated using the following Equation 2:

where ΔA is the change of absorbance value, Δt is the reaction time interval for the change (s), ε is the molar absorption coefficient of the colorimetric substrate, which is typically 39,000 M⁻¹ cm⁻¹ at 652 nm for oxidized TMB, and l is the path length of light traveling in the cuvette (0.2 cm in our experiments). The kinetic constants K_M and V_max were calculated in GraphPad Prism 10.1.1 software through fitting the v versus [S] curve using the Michaelis–Menten equation (Equation 1).

The catalytic constant (k_cat) is defined as the maximum number of substrate molecules converted to product per unit of time and is calculated by the following Equation 3:

where [E] is the concentration of pyrite (M).

k_cat/K_M, which characterizes the catalytic ability of pyrite to the substrate (TMB), reflects the catalytic efficiency of pyrite.

Unlabeled ss-oligo (21-nt, 3.0 mg/mL), Q-dextran (3.6 mg/mL), pyrite (0.30 mg/mL), TMB (0.30 mg/mL), and 0.10 mM H₂O₂ were incubated in glycine–HCl buffer (40 mM, pH 4.07), in a total volume of 10 μL. In the fluorescence observation experiments, an additional 6.0 × 10⁻³ mg/mL of FAM-labeled ss-oligo-L or TAMRA-labeled ss-oligo was added to facilitate the observation of DNA distribution. After incubation for 5.0 min at 23 °C to allow for formation of condensates, samples were visualized by brightfield microscopy and fluorescent microscopy at 100× magnification using a Zeiss Axio Observer Z1 Inverted Fluorescence Microscope (Germany).

Oxidized TMB exhibits absorbance across a broad wavelength range, which interferes with fluorescence imaging. Consequently, droplets with higher TMBox concentrations appeared darker under fluorescence microscope. To mitigate the color interference of TMBox on FAM-labeled ss-oligo-L/TAMRA-labeled ss-oligo, GSH (glutathione) was used to reduce TMBox to TMB. The final working concentrations of pyrite microparticle, TMB and H₂O₂ were 0.10 mg/mL, 0.10 mg/mL and 0.50 mM, respectively. GSH (0.40 mM) was added either at 0 or 6.0 min after the reaction had begun, and the change of peak intensity (652 nm) was monitored. The group without GSH was supplemented with deionized water.

To investigate the mobility of DNA across and within droplets, FRAP experiments were conducted on both single-phase and multiphase droplets. The standard assay (10 μL) contained 3.0 mg/mL unlabeled ss-oligo, 6.0 × 10⁻³ mg/mL TAMRA-labeled ss-oligo, 3.6 mg/mL Q-dextran, 1.0 mM H₂O₂ in glycine–HCl buffer (40 mM, pH 4.07). After incubation for 5.0 min at 23 °C to allow condensate formation, samples were imaged under the Super Resolution Microscope N-SIM S (Nikon A1 N-SIM S, Japan) using a 100 × oil-immersion objective. Condensates with a clear red fluorescent signal were selected for photobleaching (“automatic focus” was activated, “pixel binning” was set at 2 × 2, and “exposure time” was set at 300 ms). The partial region was photobleached at 561 nm (Laser Power: 17.6), followed by a fluorescence recovery period of 50.4 s. The fluorescence intensity of the photobleached area was normalized relative to that of the unbleached area.

XRD patterns of pyrite matched the standard pyrite pattern [FeS₂ (PDF 42-1340)], confirming its crystalline structure (Figure 1A). The POD-like catalytic activity was then investigated using TMB oxidation. In the sample containing pyrite, TMB and H₂O₂, a full scan absorption spectrum of TMBox displayed peaks at 380 nm, 450 nm, and 652 nm. No significant absorbance was detected when TMB was mixed with H₂O₂ without pyrite (Figure 1B). However, when pyrite was mixed with TMB without H₂O₂, a weak absorbance peak corresponding to TMBox was observed (Figure 1B). This is consistent with previous reports (Cohn et al., 2006; Gil-Lozano et al., 2017; He et al., 2023), indicating that when pyrite is exposed to water, a small amount of H₂O₂ is spontaneously generated. The resulting hydroxyl radicals subsequently react with TMB, forming trace amount of TMBox. Although the absorption peak at 652 nm is not the highest in intensity, it provides the most stable and interference-free signal for kinetic quantification. Therefore, 652 nm was selected as the detection wavelength.

To evaluate the POD-like activity of pyrite microparticles, we first determined the optimal buffer system for the reaction. Pyrite exhibited POD-like activity in both acetic acid–sodium acetate (AcOH–NaOAc) and glycine–HCl buffer systems, while no activity was detected in citric acid-disodium hydrogen phosphate (CAPS) buffer (Figure 1C), possibly due to adsorption of citric acid onto pyrite surface (Zhang et al., 2017). Furthermore, the catalytic activity was influenced by buffer concentration, with a relatively high activity around 40 mM. Considering that different buffer components may introduce additional effects, the catalytic activity of pyrite was subsequently evaluated in 40 mM glycine buffer, with the pH adjusted between 1.50 and 10.07 using HCl or NaOH, revealing the highest activity at pH 4.07 (Figure 1D).

According to a previous study (Song et al., 2020), the catalytic process of pyrite microparticles can be described in two steps. First, hydroxyl radicals (·OH) are produced from the interaction of pyrite microparticles with H₂O₂. Subsequently, these radicals oxidize TMB, resulting in the formation of oxidized TMB (TMBox). It is anticipated that a higher dosage of pyrite microparticles would generate more ·OH, thereby enhancing its catalytic efficiency. This hypothesis is supported by our data (Figure 2A), where the activity is proportional to pyrite concentration. The steady-state kinetic properties were examined by varying the concentration of H₂O₂ in a constant concentration of 0.15 mg/mL pyrite-based catalytic system. The oxidation reaction catalyzed by pyrite exhibited a typical Michaelis–Menten curve (Figure 2B). Based on the fitting, the K_M value for the reaction is 1.73 × 10⁻⁴ M, and the V_max is 5.15 ± 0.50 μM/s (Table 1).

The POD-like activity of pyrite highlights its potential as an early catalyst in the origin of life. In prebiotic chemistry studies, this concept is important because the interaction of pyrite with early Earth environments might have played a role in facilitating primitive metabolic reactions. To explore whether mineral catalytic activity could occur within primitive cells, we investigated the interactions between pyrite and protocells. As expected, after mixing pyrite, ss-oligo and Q-dextran coacervate droplets with TMB (as chromogenic agent) for 5 min, the pyrite microparticles became efficiently encapsulated within the droplets (Figure 2C). However, because the number of droplets exceeded the number of pyrite microparticles, many droplets remained without pyrite.

To evaluate the catalytic activity of encapsulated pyrite, varying concentrations of H₂O₂ were added to the samples, and the formation of TMBox was monitored through colorimetric detection. Indeed, the oxidation reaction catalyzed by droplet-encapsulated pyrite also exhibited a typical Michaelis–Menten curve and can be fitted to the Michaelis–Menten equation (Figure 2D). The K_M value of pyrite microparticles inside droplets is much higher than that of un-encapsulated pyrite (K_M: 3.75 × 10⁻⁴ M vs. 1.73 × 10⁻⁴ M, Table 1), indicating a weakened affinity for H₂O₂. However, the V_max of droplet-encapsulated pyrite was approximately 3-fold higher than that of pyrite alone (V_max: 15.5 ± 1.4 μM/s vs. 5.15 ± 0.50 μM/s), resulting in a higher catalytic efficiency (K_cat/K_M: 33.1 ± 3.0 vs. 23.8 ± 2.3).

These results suggest that, although encapsulation within coacervate droplets introduces physical barriers that may reduce the accessibility of H₂O₂ to the pyrite surface and restrain the diffusion of ·OH, the local enrichment of TMB within the droplets compensates for these effects and ultimately enhances the overall catalytic efficiency.

To investigate the impact of mineral catalytic activity on droplets, we observed the color changes and dynamic behavior of droplets under a microscope. Our observations confirmed that, consistent with the catalytic activity assays, the blue TMBox was not observed in the presence of either pyrite or H₂O₂ alone (Figure 3A; Supplementary Figure S1), suggesting that the low amount of TMBox formed in these samples (Figure 1B) was insufficient to produce a visible color change. In contrast, in the presence of both pyrite and H₂O₂, the droplets turned blue, indicating the formation of TMBox. Notably, the smaller droplets exhibited a darker blue color, suggesting that the reaction proceeded more rapidly in these droplets. Under all experimental conditions, droplet fusions were observed. Interestingly, in the absence of H₂O₂, two droplets fused to form a homogeneous droplet (Figure 3B; Supplementary Video S1). However, in the presence of both pyrite and H₂O₂, fusion resulted in multiphase droplets, with smaller droplets remaining stably within the larger ones (Figures 3A–C; Supplementary Video S2).

To further reveal the formation mechanism of the multiphase droplets, as well as the distribution of components within the droplets, we initially introduced 6.0 × 10⁻³ mg/mL of FAM-labeled ss-oligo-L (which is excited at 492 nm and emits light at 517 nm) into the droplet system. Ss-oligo-L is about three times longer than that of the ss-oligo used in the droplet system, allowing us to explore how the multiphase structure affects the distribution of exogenous nucleotides. Since TMBox (oxidized TMB) exhibits absorbance across full wavelengths, smaller droplets with higher concentrations of TMBox appeared black under fluorescence microscopy (Figure 4A). To mitigate the color interference of TMBox on FAM-labeled ss-oligo-L, GSH (glutathione) was used to reduce TMBox to TMB (Figure 4B). In the pyrite-H₂O₂ system, an increase in absorbance from TMBox production was observed. Introducing GSH at the beginning of the reaction prevented any color development, indicating that GSH inhibited the oxidation of TMB. Also, addition of GSH after 6.0 min of the reaction resulted in a rapid decrease in absorbance, confirming that TMBox was reduced to TMB. Following the reduction of TMBox to TMB by GSH, the multiphase system maintained its stability, and as anticipated, the fluorescence of FAM-labeled ss-oligo-L was clearly observable and was predominantly enriched in the smaller droplets (Figure 4C). The multiphase droplets, as visualized by FAM-labeled ss-oilgo-L, are similar to those by TAMRA-labeled ss-oligo (Figure 3A), indicating that the multiple phase formation is driven by ss-oligo and independent of ss-oligo length.

To investigate the mobility of DNA in single-phase and multiphase droplets, fluorescence recovery after photobleaching (FRAP) experiments were conducted using TAMRA-labeled ss-oligo, which is identical to the ss-oligo used in the droplet system, eliminating any potential differences in mobility due to variation of sequence length. Before photobleaching, an image was captured as a reference for fluorescence intensity, and 0 s was marked after photobleaching (Figures 4D,E). The results showed that after photobleaching, the fluorescence intensity of TAMRA-labeled ss-oligo decreased to 60%–70% of its original value. In single-phase droplets, because of high DNA mobility, TAMRA-labeled ss-oligo quickly recovered fluorescence to >96% after 50.4 s. In contrast, the mobility of TAMRA-labeled ss-oligo in the multiphase droplets was significantly reduced, with fluorescence recovery reaching only 71% after the same amount of time. This suggests that DNA mobility decreases after the formation of the multiphase system, implying that mineral catalysis may promote DNA accumulation and preservation within protocells.