Enhancing aquatic ecosystem environmental DNA capture, preservation and analysis with magnetic spiky silica nanoparticles

Organisms in aquatic environments continuously release genetic material, including nuclear and mitochondrial DNA, into the surrounding water, referred to as environmental DNA (eDNA), which can be enriched and analyzed using molecular approaches to enable the detection of aquatic species^1,2,3,4. This method has become a powerful tool for ecological monitoring due to its high sensitivity, cost-effectiveness, and scalability, proving particularly valuable for tracking non-indigenous species (NIS), assessing endangered populations, and conducting comprehensive biodiversity assessments while minimizing ecosystem disturbance^5,6,7,8.

However, once released into aquatic systems, eDNA undergoes rapid degradation and dilution through horizontal and vertical dispersion, making efficient and timely concentration critical for downstream analysis^9,10. Filtration remains the most widely used method for eDNA concentration, where water samples are passed through filter membranes (e.g., nitrocellulose, mixed cellulose esters, or polyethersulfone (PES)) to capture DNA¹¹. Despite its effectiveness, this approach faces practical limitations: membrane clogging reduces sampling efficiency, and processing large volumes (>1 L) typically requires vacuum equipment and long processing time, hindering rapid, on-site eDNA concentration^12,13.

Nanoparticles have emerged as powerful tools for nucleic acid capture due to their unique physicochemical properties, particularly their high surface area-to-volume ratios, which enhance binding capacity and efficiency^14,15,16. The combination of porous architectures and versatile surface modification strategies enables selective and high-performance DNA/RNA isolation^17,18. Among these, magnetic beads coated with mesoporous silica (MSN) have been widely adopted to increase the accessible surface area for DNA binding. They also exhibit exceptional stability under extreme conditions, including high temperature, pressure, and exposure to strong acids or bases, while maintaining low toxicity and excellent biocompatibility, making them ideal for environmental and biomedical applications^19,20. However, the intrinsic architecture of conventional MSN, featuring inward-oriented, tortuous pore channels, introduces fundamental limitations when handling complex environmental samples^21,22. These internal mesopores are easily clogged by suspended particulates and biomolecular aggregates, substantially reducing the effective surface area and slowing mass transfer, ultimately constraining both adsorption and desorption kinetics.

These limitations prompted a conceptual shift: rather than further optimizing internal pore channels, we reasoned that inverting the structure from an inward-pore system to an outward-facing, radially protruding surface could fundamentally overcome the bottlenecks of MSN-based magnetic beads. An outwardly spiky architecture would maximize externally exposed binding sites, minimize transport resistance, and avoid pore blockage while enabling rapid interaction with DNA molecules. However, constructing such a well-defined spiky silica morphology on a magnetic core is synthetically challenging, as it requires precise control over silica nucleation, anisotropic growth, and interface stability during shell formation.

Significant progress has been made in engineering spiky silica architectures to increase external surface area and enhance mass transfer. Template-based strategies can generate well-defined spiky morphologies. However, these methods require high concentrations of cationic surfactants and organic structure-directing agents, which introduce biological safety concerns, complicate purification, and often leave residual contaminants that interfere with downstream nucleic-acid analyses²³. In contrast, template-free approaches have emerged as a more straightforward and potentially safer alternative. Recent advances, including our previous work, demonstrating that manipulating silica precursor chemistry and condensation kinetics can drive spontaneous formation of outwardly protruding silica spikes without the need for surfactants^24,25. These methods avoid template-related contamination, simplify synthesis and purification, and are inherently more compatible with biomolecular applications. Nevertheless, existing template-free systems remain limited by challenges in achieving uniform spike morphology, integrating magnetic components, and producing fully functional nanostructures optimized for rapid nucleic-acid capture in natural waters.

Here, we address these synthetic and functional limitations by developing core–shell magnetic spiky silica nanoparticles (MSSNs) that integrate a superparamagnetic Fe₃O₄ core with a fully inorganic, outwardly spiky silica shell. Moreover, the MSSN architecture can be finely tuned by adjusting the temporal sequence of TEOS (tetraethyl orthosilicate) and resin addition, the amine-mediated hydrolysis rate, and the water and resin contents, allowing precise control over spike length and morphology, silica core size, and shell porosity. Our MSSNs combine the advantages of rapid magnetic separation with maintained high-efficiency nucleic acid adsorption performance, enabling enhanced collection efficiency from complex matrices while improving handling through magnetic manipulation. This advancement significantly improves nucleic acid capture methodologies, particularly for applications demanding high sensitivity coupled with practical field utility, such as aquatic eDNA. The streamlined MSSNs-based eDNA recovery pipeline demonstrated superior performance compared to commercial eDNA extraction kits, with the additional advantage of maintaining nucleic acid integrity at ambient temperatures. This platform seamlessly integrates with downstream molecular analyses, including metabarcoding and metagenomic sequencing, offering significant advantages for field-based environmental monitoring.

MSSNs: synthesis and characterization

To realize a silica architecture capable of providing abundant, fully accessible binding sites, we designed a multi-stage synthesis route to construct MSSNs comprising a Fe₃O₄ core, a compact silica interlayer, and an outward-oriented spiky silica shell (Fig. 1a). The morphology of MSSNs is highly tunable, particularly in terms of particle size and spiky-shell architecture²⁵. The spiky-shell features mainly include the degree of spike development, spike length, and overall surface roughness. These morphological characteristics can be precisely regulated by adjusting key synthesis parameters, such as the TEOS dosage, ammonia concentration, ethanol-to-water ratio, reaction temperature, and the co-assembly conditions between the organic components and the silica source.

The morphology and physicochemical properties of the MSSNs are systematically characterized in Fig. 1b-g and Figs. S1–2. TEM images (Fig. 1b, c) show that the MSSNs exhibit a uniform spherical morphology with a clear core–shell contrast, where a lighter spiky silica shell homogeneously encapsulates the dark Fe₃O₄ core; the radially oriented “spike-like” protrusions confirm the successful construction of a rough, high-surface-area outer layer. Nitrogen adsorption–desorption isotherms (Fig. 1d) show a typical type-IV curve with a distinct hysteresis loop, indicating the presence of mesoporous structures in the spiky silica shell. A sharp increase in adsorption at relative pressures of 0.4–0.7 is attributed to capillary condensation within mesoporous channels. The Brunauer-Emmett-Teller (BET) surface area of the MSSNs was determined to be 216.6 m²·g⁻¹, with a total pore volume of 0.31 cm³·g⁻¹, demonstrating the highly porous nature of the particles (Fig. 1d, Table S1). The BJH pore-size distribution (Fig. 1d) reveals a relatively narrow mesopore distribution with a peak centered at 6.5 nm, further confirming the formation of well-defined mesoporous structures in the spiky silica shell. FTIR spectra (Fig. 1e) reveal strong bands at ~1080, ~800 and ~460 cm⁻¹ assigned to Si-O-Si stretching and bending vibrations of the silica network. After APTES grafting, a weaker band appearing in the 1550–1650 cm⁻¹ region corresponds to N-H vibrations, confirming the successful introduction of amino groups. In addition, a band near ~580 cm⁻¹ can be attributed to Fe-O stretching, evidencing the presence of the magnetic core. Zeta-potential measurements (Fig. 1f) show that bare spiky silica nanoparticles possess a negative surface charge originating from deprotonated silanol groups, whereas amino-functionalized MSSN-NH₂ exhibits a markedly increased positive potential due to protonated -NH₂moieties, providing strong electrostatic affinity toward negatively charged DNA. Finally, the magnetic hysteresis loop recorded by VSM (Fig. 1g) shows an S-shaped curve with negligible coercivity and remanence, indicative of superparamagnetic behavior, and a saturation magnetization of ~20 emu g⁻¹, demonstrating that the Fe₃O₄ cores retain strong magnetic responsiveness after formation of the spiky silica shell, which is advantageous for rapid magnetic manipulation and separation in aqueous environments.

Streamlined eDNA recovery from aquatic samples using MSSNs

Environmental DNA (eDNA) from aquatic organisms is ubiquitously distributed in aquatic systems, yet existing extraction methods often compromise between speed, cost, and simplicity^26,27. To overcome these limitations, we developed an MSSNs-based extraction protocol that streamlines both field sampling and laboratory processing. Our field-to-lab workflow enables researchers to: (1) collect water samples directly in 50 mL conical tubes,(2) incubate water samples with MSSNs for immediate eDNA binding and stabilization without pretreatment steps, and(3) either process samples on-site or transport them to the laboratory for eDNA extraction and purification (Fig. 2a).

The magnetic core of MSSNs enables a dual-mode separation process, allowing samples to be processed using either magnetic separation (via a magnetic rack for rapid supernatant removal) or conventional centrifugation. Following separation, a simple elution protocol with pH adjustment efficiently releases eDNA. This approach leverages the strong affinity between MSSNs and nucleic acids in water samples, eliminating the need for complex instrumentation during field collection. Notably, the method facilitates convenient sample transport: after magnetic separation, the aqueous phase can be discarded, and only the MSSNs–eDNA complexes need to be retained in the original collection tubes. This significantly reduces shipping volume and weight while ensuring sample stability.

To determine the ideal nanoparticle concentration, elution volume ratio, and recovery conditions for downstream applications, we optimized the protocol using standardized DNA samples. Initial eDNA extracted from seawater samples from the sea squirt aquaculture tank via the phenol-chloroform-isopropanol method exhibited a bimodal size distribution, with distinct fragments >5000 bp and <1000 bp on 1% agarose gel electrophoresis (Fig. S3a). This fragmentation pattern mirrored genomic DNA isolated from Ciona savignyi using the same method, as further confirmed by Bio-fragment analyzer (Fig. S3b). For standardization purposes, test samples (TS) were prepared by spiking 1 μg of C. savignyi genomic DNA into 50 mL artificial seawater (ASW) to approximate natural eDNA concentrations in aquatic environments.

To establish optimal elution parameters, we tested various elution volume ratios to 200 μL of MSSNs (2 mg) suspension, which had been incubated with TS. Unless otherwise stated, MSSNs were harvested by centrifugation in the following experiments to ensure maximal DNA recovery under laboratory conditions. We evaluated three elute volume to MSSNs ratios (1:1, 3:2, and 2:1) and found no statistically significant differences in DNA recovery efficiency (Fig. 2b, S4a). The 1:1 ratio was selected for downstream applications due to its operational simplicity and reagent economy. Subsequent dose-response experiments demonstrated saturable binding kinetics, with DNA recovery plateauing at 200 μL MSSNs volume (yielding 0.22 μg [22.09%] recovery from TS). No significant yield improvement was observed when increasing to 250 μL (Fig. 2c, S4b, S5), establishing 200 μL for a 50 mL water sample as the optimal working concentration where MSSNs become saturated. Further characterization of the system’s capacity revealed that 200 μL MSSNs consistently achieved approximately 20% recovery efficiency across a broad concentration range (0.1-3 μg genomic DNA in 50 mL ASW). However, we observed a decrease in recovery rate when the DNA concentration is 5 μg in 50 mL ASW (Fig. 2d, S4c), indicating the maximum binding capacity of this MSSNs system falls within the 3-5 μg range for 50 mL samples. Importantly, this threshold (3–5 μg/50 mL) exceeds typical eDNA concentrations encountered in environmental samples, as confirmed in multiple field-collected water specimens (Fig. 4c, e). Validation experiments demonstrated that the optimized 200 μL MSSNs suspension remained effective across diverse aquatic matrices. When applied to both natural seawater and freshwater samples, the system maintained a consistent recovery pattern comparable to those obtained with standardized TS conditions (Fig. S4d). Finally, the time-dependent adsorption of DNA onto MSSNs in TS solution was characterized. MSSNs adsorbed a significant amount of DNA within 5 min, with adsorption increasing over time and approaching saturation after 20 min (Fig. 2e, S4e). Although magnetic collection offers convenient and feasible field operation, we observed that magnetic guidance yielded only one-fifth the eDNA amount compared to centrifugal collection when using the same 200 μL MSSNs suspension. However, this recovery rate improved to one-half when the MSSNs quantity was doubled in the magnetic guidance approach (Fig. S6).

Binding mechanism of eDNA to MSSNs

To elucidate how MSSNs achieve rapid and high-capacity nucleic-acid capture, we investigated the spatial distribution, interaction mode, and thermodynamic characteristics of DNA binding on the spiky silica surface. Elemental mapping revealed the emergence of additional nitrogen (N) and phosphorus (P) signals—signature elements of nucleic acids—predominantly enriched at the outer region of the spiky silica shell. Both N and P exhibited a surface-enriched distribution compared to silicon (Si) and oxygen (O), indicating that eDNA localizes on the externally exposed binding sites rather than infiltrating into the silica framework. This spatial distribution is consistent with the molecular composition of DNA, which contains nitrogenous bases and a phosphate backbone. Together, the spatially correlated signals of Fe (core), Si/ O (shell), and N/P (adsorbate) delineate a hierarchical “core–shell–adsorbate” architecture, confirming efficient surface loading of eDNA and supporting the structural advantages of the outward-facing spikes for rapid and high-capacity nucleic-acid capture (Fig. 2f). Confocal laser scanning microscopy (CLSM) further verified uniform and dense DNA coverage across the spiky shell, stained with DAPI or SYBR Green I – two fluorescent dyes that both bind to the minor groove of double-stranded DNA, with the latter also capable of intercalating between DNA base pairs. Their maximum emission wavelengths are approximately 461 nm and 520 nm, respectively, confirming efficient capture facilitated by the outward-facing morphology (Fig. 2g, S7).

The thermodynamic interaction between DNA (50 µM) and MSSNs (16.1 nM) was analyzed by isothermal titration calorimetry (ITC) (Fig. 2h, i, S8a, b). The resulting binding isotherm, presented as the molar ratio of DNA to MSSNs during the titration, exhibits a biphasic profile indicative of two distinct, stepwise binding processes: (i) an initial exotherm-dominated regime, followed by (ii) a subsequent endotherm-dominated region (Fig. 2i, Table S2). The initial exothermic phase reflects an enthalpy-driven binding process arising primarily from strong electrostatic attraction between positively charged surface amine groups (–NH₂) on the MSSNs and the negatively charged phosphate backbone of DNA. Such favorable enthalpic contributions are characteristic of short-range Coulombic interactions and have been widely observed for nucleic-acid adsorption onto amine-functionalized mesoporous silica surfaces²⁸. Upon further DNA addition, the binding enters a second regime characterized predominantly by endothermic heat uptake. This phase is thermodynamically unfavorable in enthalpy but driven by a net gain in entropy, which is attributed mainly to the release of confined hydration water molecules and counterions from both DNA and MSSNs’ surface upon binding. Similar entropy-driven endothermic signatures have been reported for nucleic acids interacting with functionalized nanoporous silica, where desolvation and configurational freedom outweigh enthalpic penalties²⁸. Together, this dual-phase thermodynamic profile of an initial enthalpy-driven electrostatic association followed by an entropy-driven desolvation process confirms a two-step binding mechanism that the spiky shell not only increases surface area but also accelerates mass transport, enhances local binding affinity, and provides structurally favorable microenvironments for strong yet reversible nucleic-acid adsorption.

Together, these results establish a mechanistic model in which the externally oriented silica spikes drive rapid, high-capacity, and clog-free DNA capture, mechanistically differentiating MSSNs from traditional mesoporous silica materials.

No significant fragment size bias in DNA capture using MSSNs

Natural eDNA samples typically contain genomic DNA, mitochondrial DNA, and partially sheared DNA fragments, which commonly appear as fragments >5000 bp band and a smear below on the 1% agarose gel electrophoresis. To quantitatively assess the potential fragment size bias in DNA recovery using MSSNs, we amplified a 6073 bp mitochondrial gene fragment from C. savignyi along with primer dimers (<100 bp) to represent large and small DNA fragments in eDNA. We then spiked 1 μg of these products into 50 mL of artificial seawater (ASW) to evaluate MSSNs capture efficiency across different fragment sizes. We demonstrated that MSSNs efficiently recovered both the 6073 bp fragment and primer dimers without significant size bias (Fig. 3a, b). Additionally, MSSNs successfully captured eDNA from natural seawater and fresh samples (eDNA Samples 1 and 2), recovering a broad range of fragment sizes (Fig. S9a). Overall, MSSNs exhibit no significant fragment size bias for either artificially spiked DNA or naturally sourced eDNA, demonstrating their exceptional potential for unbiased, high-efficiency eDNA capture in downstream applications.

DNA preservation post‑capture by MSSNs

A key challenge in field-based eDNA analysis is preventing rapid degradation of DNA after sample collection. While approaches such as cold storage, UV protection, and chemical additives (e.g., CTAB) can stabilize eDNA, these are often impractical to maintain during routine fieldwork^29,30,31. We demonstrate that DNA captured by MSSNs remains remarkably stable at ambient temperature for at least 7 days with minimal degradation.

In experiments using 200 μL of MSSNs to capture DNA from standardized TS, the eluted DNA quality showed no significant deterioration when measured immediately, or after 1, 3, and 7 days of storage at room temperature. Only a slight, non-significant decrease in DNA yield was observed: from 224.67 ± 16.04 ng on Day 1, 223.33 ± 6.11 ng on Day 3 and 214.53 ± 19.43 ng on Day 7. In contrast, unprotected DNA in the TS degraded rapidly under the same conditions, with 47.10% degraded by Day 1, 55.49% degraded by Day 3, and only 17.86% remaining by Day 7 (Fig. 3c, S9b). Bio‑fragment analyzer results further confirmed the protective effect of MSSNs on DNA integrity (Fig. S10). Furthermore, we showed that MSSNs-bound DNA is effectively shielded from enzymatic degradation. When exposed to DNase I, unprotected DNA was degraded by approximately 98.07%. In contrast, 88.36% of the DNA was retained when it was pre‑adsorbed onto MSSNs prior to DNase treatment (Fig. 3d, S9c). The preservation capability of MSSNs offers a substantial practical advantage for field sampling. Researchers can magnetically recover MSSNs on-site and store the particle-bound eDNA at ambient temperature for several days without cold-chain logistics, simplifying transport and enabling delayed laboratory processing.

MSSNs-based workflow outperformed commercial DNA extraction kits

To assess the practical efficacy of our method, the performance of the MSSNs-based extraction workflow was rigorously benchmarked against five commercially available DNA extraction kits (Table S3). Using standardized TS, the MSSNs recovered 232.67 ± 21.20 ng of DNA (mean ± SD) with optimal A260/A280 ratios (1.8-2.0), representing a 634.67% increase over the second-best performing commercial kit (Fig. 4a, b). This comparative evaluation was extended to environmental samples, including both marine and freshwater ecosystems. From 50 mL seawater samples collected from controlled aquaculture tanks, the MSSNs yielded 745.33 ± 124.13 ng of eDNA, outperforming the nearest commercial alternative by 511.93% (Fig. 4c). Purity analysis showed that only the MSSNs method and Kit 3 consistently maintained A260/A280 values within the ideal range (1.8–2.0), while other kits showed greater variability (Fig. 4d). Freshwater samples from campus pond ecosystems produced comparable results, with the MSSNs method again demonstrating superior performance (about 725.33 ng eDNA/50 mL sample) (Fig. 4e), while the A260/A280 ratio (mean 1.76) slightly fell short of the ideal range in these samples (Fig. 4f). Fragment integrity analysis revealed more significant advantages. The Qsep100 Biofragment Analyzer showed the MSSNs-extracted eDNA from seawater samples maintained a defined fragment distribution, with fragments present both below 1000 bp and above 5000 bp. In contrast, all commercial kits produced evenly distributed smears lacking the <1000 bp fragments preserved by our method, and some kits also lacked distinct > 5000 bp fragments, largely due to low DNA yields falling below the recommended lower detection limit of the Qsep100 Biofragment Analyzer (Fig. S11). These findings collectively demonstrate that the MSSNs approach provides superior eDNA recovery while better maintaining nucleic acid integrity compared to conventional extraction methods.

Compatibility of MSSNs-extracted eDNA with amplicon and metagenomic sequencing

To evaluate the compatibility of MSSNs-extracted eDNA with downstream molecular analyses, we conducted controlled metabarcoding and shotgun metagenomic sequencing experiments. Using an indoor culture system containing seawater tanks (20 L each) with precisely controlled populations of C. savignyi (0, 5, 10, and 20 individuals per tank), we extracted eDNA from 50 mL water samples per tank (Fig. 5a). The resulting libraries generated 263.59 million reads (mean) for each amplicon library targeting 16S, COX1, and CYTB. Taxonomic assignment revealed 386158-30335979 reads specifically mapped to C. savignyi (Table S4). Importantly, we observed strong linear correlations between: (1)initial animal abundance and eDNA yield (R² = 0.99), and(2)animal abundance and target gene (16S, COX1, and CYTB) read counts across all three markers (R² = 0.77, 0.83, 0.83) (Fig. 5b-e). These results demonstrate that our MSSNs-based extraction method preserves the quantitative relationship between source organism abundance and sequencing results while maintaining excellent compatibility with the amplicon sequencing workflow.

Shotgun metagenomic analysis of both seawater (aquaculture tank) and freshwater ecosystem (pond on campus) revealed distinct microbial profiles. We identified 18 phyla and 189 genera in total, with Pseudomonadota, Actinomycetota, and Bacteroidota dominating freshwater communities, while seawater samples showed high abundances of Pseudomonadota, Actinomycetota, and Campylobacterota (Fig. 6a). At the genus level, freshwater systems were characterized by SYFN01, Limnohabitans A, and Rhodoferax C, contrasting with seawater’s Vibrio, Cognaticolwellia, and Pseudoalteromonas dominance (Fig. 6b).

Our analysis of seawater successfully characterized the aquatic community in an indoor ascidian aquaculture tank. Metagenomic sequencing reads assigned to ascidians were overwhelmingly dominated by the intentionally introduced species, C. savignyi (98.08%). Despite this dominance, we also detected trace eDNA from 22 other ascidian species (Fig. 6c, d). Notably, most of these additional species—including Ciona robusta, Botryllus schlosseri, and Styela clava—are prevalent in local coastal harvesting sites (Fig. S12), suggesting their eDNA or the organisms themselves were inadvertently introduced into the tank system. Beyond ascidians, the MSSNs-based extraction captured eDNA from a broad spectrum of marine invertebrates, identifying nine phyla. Ranked by relative abundance, the dominant phyla were Chordata (dominated by ascidians; 79.50%), Arthropoda (13.94%), Cnidaria (2.97%), Bryozoa (1.99%), and Mollusca (1.26%); other phyla constituted 0.35% of reads (Fig. 6e). The detection of this diverse invertebrate signal, including taxa at very low abundance, underscores the effectiveness and high sensitivity of MSSNs for capturing trace eDNA from seawater.

eDNA-based aquatic ecosystem monitoring faces pressing challenges in field sample processing, particularly regarding fast and efficient nucleic acid enrichment from water samples^26,27. While filtration remains the widely used method, its requirement for large water volumes, vacuum equipment, and lengthy processing times severely limits its rapid field applications and sample transportation^32,33. To overcome these limitations, we innovatively engineered MSSNs – hybrid nanomaterials that combine a magnetic Fe₃O₄ core with a uniquely structured silicon-based coating featuring three-dimensional radial silica spikes. This design represents a substantial advancement over conventional magnetic mesoporous silica materials, with the externally oriented silica spikes enabling rapid, high-capacity, and clog-free DNA capture, addressing challenges associated with limited surface-active sites and inefficient mass transfer kinetics.

Our investigation revealed a dual-phase thermodynamic profile for DNA binding to MSSNs: an initial enthalpy-driven electrostatic association, followed by an entropy-driven desolvation process. Given that the MSSNs exhibit a mesopore size centered at approximately 6.5 nm, we speculate that DNA adsorption does not require the entire molecule to be fully inserted into the interspike spaces. Instead, the interaction likely involves partial insertion into accessible interspike spaces, with multipoint contact mediating local attachment, wrapping, or bridging of longer DNA fragments along the rough three-dimensional surface. We acknowledge that the binding kinetics and molecular-level binding states require further investigation using higher-resolution techniques in the future.

The practical utility of MSSNs enables a highly efficient and streamlined workflow for eDNA extraction from diverse aquatic environments. Using an optimized protocol, we consistently recovered sufficient eDNA from both marine and freshwater samples for downstream molecular applications. Notably, MSSNs exhibited exceptional eDNA preservation capability. Once bound, eDNA remained stable at ambient temperature for days without significant degradation. This breakthrough allows for secure long-distance sample transport without the need for cold chain logistics, as samples maintain integrity during regular shipping.

The separation process of MSSNs offers dual-mode flexibility: samples can be processed using either magnetic separation or conventional centrifugation. Magnetic collection leverages the strong affinity between MSSNs and nucleic acids, eliminating the need for complex instrumentation during field collection. Although magnetic collection is more convenient for field operation, it recovers less eDNA than centrifugation, though this recovery improves when the MSSNs volume is increased. This performance gap primarily stems from the suboptimal magnetic responsiveness of the current MSSNs formulation. Increasing the nanoparticle dosage can partially compensate for the lower magnetic separation efficiency, but at the cost of higher material consumption. Residual MSSNs remaining in solution after magnetic separation likely account for a significant portion of uncollected eDNA, highlighting the need for further optimization of the nanoparticles’ magnetic properties to achieve performance comparable to centrifugation while maintaining field applicability.

Notably, the initial workflow did not incorporate a cell or tissue lysis step because ionic detergents such as CTAB and SDS are incompatible with MSSNs-mediated eDNA capture. This restricts the protocol to capturing only free eDNA molecules in water samples. Nevertheless, in its current setting, MSSNs already capture adequate eDNA for downstream analysis from less than 100 mL of water sample. Importantly, our latest research has found that non-ionic detergents are compatible with the MSSNs workflow and can significantly increase DNA yield (Fig. S13). Therefore, a lysis step will be incorporated into the MSSNs workflow for future practical applications.

The downstream compatibility of MSSNs-extracted eDNA was rigorously validated through comprehensive amplicon and metagenomic sequencing. Our results demonstrate that the method preserves partial yet significant quantitative biological information, as evidenced by the linear correlations observed between source organism abundance and eDNA yields/sequencing outcomes. Shotgun metagenomic sequencing successfully identified distinct microbial community signatures in both freshwater and marine samples, confirming the method’s effectiveness for comprehensive biodiversity profiling. Furthermore, the consistent detection of dominant ascidian species in controlled culture systems provided additional validation of the method’s reliability for eDNA-based organism detection.

In conclusion, the MSSNs-based eDNA workflow successfully bridged the gap between technical performance and field practicality. By simultaneously addressing three major challenges in environmental DNA analysis – efficient capture, ambient-temperature preservation, and downstream analytical compatibility – it represents a significant progress in aquatic ecosystem monitoring, for the capacity for sensitive, scalable, and standardized biodiversity assessment – a critical need in the face of accelerating global environmental change.

Synthesis and characterization of MSSNs

MSSNs were synthesized through a sequential procedure involving the preparation of citrate-stabilized Fe₃O₄ magnetic cores, silica coating, spiky mesoporous silica shell growth, and subsequent surface amination.

Synthesis of Fe₃O₄ magnetic cores

Fe₃O₄ nanoparticles were first synthesized via a solvothermal method. Briefly, FeCl₃·6H₂O (1.35 g) and trisodium citrate (0.40 g) were dissolved in 40 mL of ethylene glycol under continuous stirring to obtain a homogeneous solution. Sodium acetate (3.6 g) was then added, and the mixture was stirred for 30 min to ensure complete dissolution. The resulting solution was transferred into a 50 mL Teflon-lined stainless-steel autoclave and heated at 200 °C for 10 h. After naturally cooling to room temperature, the black precipitate was collected by magnetic separation and washed several times with ethanol and deionized water to remove residual reactants and byproducts. The purified Fe₃O₄ nanoparticles were finally redispersed in ethanol for subsequent silica coating.

Silica interlayer coating

To construct the silica interlayer and provide nucleation sites for the subsequent spiky growth, the Fe₃O₄ nanoparticles (42 mg) were dispersed in an ethanol (40 mL)-water (10 mL) mixture containing aqueous ammonia (1.56 mL) and ethylenediamine (0.225 mL). The dispersion was mechanically stirred at 600 rpm using an overhead stirrer equipped with a stirring paddle and heated to 60 °C. Once the temperature stabilized, tetraethyl orthosilicate (TEOS) (1.74 mL) was added dropwise to initiate the hydrolysis and condensation of silica precursors. Under alkaline conditions, ammonia and ethylenediamine catalyzed TEOS hydrolysis and the subsequent silanol condensation, yielding a uniform Fe₃O₄@SiO₂ core–shell structure. This intermediate silica layer served as a stable interface and nucleation platform for the formation of the outer spiky architecture.

Growth of the spiky mesoporous silica shell

The spiky mesoporous silica shell was subsequently generated through a cooperative assembly process involving the simultaneous condensation of TEOS and the in situ polymerization of aminophenol-formaldehyde resin. Specifically, after the formation of the initial silica layer, an additional portion of TEOS (1.74 mL) was introduced into the reaction system under vigorous stirring. Due to the limited solubility of TEOS in the ethanol-water system, the precursor was added slowly to avoid local supersaturation and ensure homogeneous hydrolysis. The suspension was stirred for 5 min to allow uniform distribution and initial nucleation of silica species.

Subsequently, 3-aminophenol (0.4125 g, 3.8 mmol) was added to the reaction mixture, followed by the addition of formaldehyde solution (0.9 mL). Under alkaline conditions, 3-aminophenol and formaldehyde underwent in situ polycondensation to form phenolic oligomers. These oligomers gradually deposited on the surface of the growing silica particles, while the simultaneous condensation of TEOS generated a hybrid silica-resin shell. The cooperative assembly between silica condensation and resin polymerization produced a rough organic-inorganic precursor layer composed of densely packed rod-like protrusions, ultimately forming a virus-mimetic spiky surface morphology. The reaction was maintained at 60 °C for 5 h under continuous mechanical stirring at 600 rpm in a fume hood.

After completion of the reaction, the solid products were collected by magnetic separation and washed repeatedly with ethanol and deionized water. The obtained particles were dried at 60 °C overnight and subsequently calcined in air at 500 °C for 5 h with a heating rate of 1 °C min⁻¹ to remove the phenolic resin template and generate open mesoporous channels within the silica shell. The resulting material was designated as magnetic spiky mesoporous silica nanoparticles (MSSNs).

Surface amination of MSSNs

To introduce amino functional groups onto the particle surface, the as-prepared MSSNs were further modified with (3-aminopropyl)triethoxysilane (APTES). In a typical procedure, 50 mg of MSSNs were dispersed in 40 mL of anhydrous ethanol by ultrasonication for 10 min to obtain a homogeneous suspension. APTES (200 μL) was then added dropwise under continuous mechanical stirring, and the reaction was allowed to proceed at room temperature for 12 h. During this process, the ethoxy groups of APTES underwent hydrolysis and subsequent condensation with surface silanol groups (Si-OH) on the MSSN, leading to the covalent grafting of aminopropyl groups onto the silica surface.

After the reaction, the amino-functionalized nanoparticles were collected by magnetic separation and washed three times with ethanol to remove unreacted silane molecules. The purified particles were then dried in a vacuum oven at 60 °C for 12 h to obtain amino-terminated MSSNs. The introduction of surface amino groups provides positively charged binding sites, enabling strong electrostatic interactions with negatively charged nucleic acids and facilitating subsequent nucleic-acid adsorption experiments.

Characterization methods

The morphology and microstructure of the MSSNs were examined using transmission electron microscopy (TEM, JEOL JEM-2100PLUS, 200 kV). Nitrogen adsorption–desorption isotherms were measured at 77 K using a MicrotracBEL BELSORP-max instrument. Fourier-transform infrared (FTIR) spectra were recorded on a Bruker TENSOR 27 spectrometer using the KBr pellet method. Zeta-potential measurements were performed using a Malvern Zetasizer Nano ZS90 instrument. Magnetic properties were characterized using a vibrating-sample magnetometer (VSM, LakeShore 7404) at room temperature (300 K) with an applied field sweep from –20 to +20 kOe. The saturation magnetization was determined from the high-field plateau of the hysteresis loop.

DNA extraction from C. savignyi tissue and mitochondrial DNA amplification

DNA was isolated from dissected siphon muscles of C. savignyi using a standard phenol-chloroform-isopropanol protocol^34,35. Following precipitation with ethanol and sodium acetate, the DNA pellet was resuspended in deionized water for storage. A 6073 bp fragment of the mitochondrial genome, spanning the ND5, COX2, and CYTB genes, was amplified by PCR using ApexHF HS DNA Polymerase CL (AG) and Mito6073-F, Mito6073-R primers (Table S5). PCR amplification was performed in a total reaction volume of 50 μL, containing 1 μL of DNA polymerase, 25 μL of 2× ApexHF CL Buffer, 1 μL each of forward and reverse primers (10 μM), and 22 μL of DNA template. Primer sequences are listed in Table S5. The PCR amplification program was set as follows: 94 °C for 60 s, 98 °C for 10 s, 57 °C for 15 s, 68 °C for 183 s,with 35 cycles, and a final extension at 68 °C for 300 s.

DNA fluorescence staining and confocal microscopy

C. savignyi genomic DNA was separately stained with DAPI or SYBR Green I. For DAPI staining, 10 μL of DAPI solution (10 μg/mL) was added to the DNA solution (1 μg in 50 μL) and incubated at room temperature for 20 min. For SYBR Green I staining, 10 μL of 100× SYBR Green I solution was added to the DNA sample and incubated at room temperature for 20 min. The stained DNA was subsequently incubated with 50 μL of MSSNs (10 mg/ml) at room temperature for 30 min to allow for binding. The resulting DNA–MSSNs complexes were recovered by centrifugation and washed multiple times with nuclease-free water, and then mounted on glass slides using an anti-fade mounting medium to prevent fluorescence quenching. Images were captured using a Zeiss LSM 980 confocal laser scanning microscope to visualize the localization of fluorescent DNA on the MSSNs surfaces.

Isothermal titration calorimetry (ITC)

Isothermal titration calorimetry (ITC) experiments were performed in Malvern MICROCAL PEAQ-ITC at 25 °C. A suspension of amine-functionalized nanoparticles was made in nuclease-free water by ultrasonication for 10 min. The DNA (Mito6073-F, 51 bp oligo, synthesized and purified by high-affinity purification) solutions were diluted using nuclease-free water and injected into the nanoparticle solution in a 280 μL ITC sample cell using the injection syringe. The reference cell was filled with degassed DIUF water, which was replaced every three days with fresh water. The DNA solution (50 µM) was injected into the ITC cell containing the MSSNs suspension (16.1 nM), while the reference cell was filled with DIUF water. The titration consisted of an initial injection of 0.4 μL, followed by subsequent injections of 1 μL at 150 s intervals with a stirring rate of 350 rpm. The molecular weight of the MSSNs was characterized by dynamic light scattering (DLS) using a WTATT DynaPro NanoStar instrument, which was used to calculate the molar concentration of the MSSNs solution. Finally, the thermodynamic parameters of the isothermal titration calorimetry (ITC) experiment were calculated using the MicroCal PEAQ-ITC Analysis Software. Dissociation constant (K_d) and enthalpy change (ΔH) were determined by non-linear least squares regression. ΔG and ΔS were estimated using the thermodynamic relation presented in ΔG = RTln(K_d) = ΔH − TΔS, where T is the isothermal temperature of the experiments³⁶.

Water sample collection and preprocessing

Seawater samples were collected from an indoor ascidian aquaculture tank (Fig. S14a). The water temperature in the culture tanks was maintained around 18 °C, with salinity around 29.6 ppt. The tanks seasonally cultured four species, including Ciona savignyi, Ciona robusta, Styela clava, and Halocynthia roretzi, all collected from the Yellow Sea. Freshwater samples were collected from a pond on the campus of Ocean University of China (Fig. S14b). The pond water temperature was approximately 29 °C, and the salinity was 0.32 ppt. A standardized test sample (TS) was prepared by spiking 1 μg of C. savignyi genomic DNA into 50 mL of artificial seawater (ASW).

eDNA quantification and quality assessment

For all extracts, DNA concentration was quantified using a Qubit fluorometer. Purity was assessed via A260/A280 ratios measured on a Nanodrop spectrophotometer, and fragment size distribution was profiled using a Qsep100 bio-fragment analyzer.

Standardized workflow of eDNA enrichment using MSSNs

A 10 mg/mL MSSNs suspension was prepared via dispersion of 10 mg MSSNs in 1 mL H₂O by sonication, followed by thorough vortexing immediately before use. For eDNA adsorption, 200 µL of this MSSNs suspension was added to a 50 mL water sample and incubated with shaking at room temperature for 30 min. MSSNs were then recovered by centrifugation at 4200 rpm for 5 min, with the supernatant discarded to collect the precipitate; note that magnetic collection using a DynaMag rack could be substituted during field work, though with a trade-off in capture efficiency (Fig. S6b). The precipitate was transferred to a 1.5 mL tube, washed twice with H₂O, and collected on a magnetic stand for 2 min. For elution, a 1:1 elution volume-to-MSSNs ratio was used, incubating the precipitate with 200 µL of alkaline release buffer (1 mM Tris, pH 13) at room temperature for 30 min. Finally, the tube was placed on a magnetic stand for 2 min to separate the supernatant, which was then pH-adjusted to neutral (PH = 7.0) using dilute hydrochloric acid (2.28 M) for downstream applications.

Comparison of MSSNs-based and commercial kit eDNA extraction efficiency

eDNA was extracted from 50 mL water samples using both the MSSNs-based method and five commercial kits (Kit 1-Kit 5; details in Table S3). The MSSNs extraction was executed using the standardized workflow described above. In contrast, extraction with commercial kits required a membrane enrichment step before eDNA extraction. Each 50 mL sample was first vacuum-filtered through a 0.2 μm PCTE membrane using a sterile apparatus (pump, funnel, and flask). The membrane was then aseptically removed, cut into fragments with sterilized tools, and transferred to a 2 mL tube for processing. To prevent cross-contamination, all filtration equipment and tools were rigorously decontaminated with an overnight sodium hypochlorite soak, followed by repeated rinsing with distilled water and a final flush with sample water. Subsequent DNA extraction steps were performed strictly according to each manufacturer’s instructions.

Amplicon sequencing

The eDNA captured and purified using MSSNs was directly subjected to amplicon sequencing without any additional purification steps. Amplicon sequencing of ascidian eDNA was performed using a proprietary detection method and kit (Chinese Patent ZL 2023 1 1864239.9)³⁷. The assay simultaneously targets the mitochondrial 16S rRNA, COI, and CYTB genes with three pairs of degenerate primers with sample barcodes (Table S5). The resulting PCR products were pooled in equimolar ratios and sequenced on an Illumina NovaSeq 6000 platform with a 150 bp paired-end read configuration.

Metagenomic sequencing

The eDNA captured and purified using MSSNs was directly subjected to metagenomic sequencing without any additional purification steps. Library construction for metagenomic sequencing was performed using a Tn5 transposase-based tagmentation method³⁸. The procedure consisted of three main stages: (1) DNA fragmentation and tagmentation, (2) adapter ligation and PCR amplification, and (3) size selection. The tagmentation reaction was carried out in a mixture containing 12.5 μL of 4× HMW buffer, 12.5 μL of DMF, 35.5 μL of DNA sample, and 2 μL of Tn5 transposase, followed by incubation at 55 °C for 20 min. The tagmented DNA was purified and used for PCR amplification in a 50 μL reaction system containing 25 μL of 2X M5 HiPer plus Taq HiFi PCR mix, 23 μL of purified product, and 1 μL each of forward and reverse primers. The PCR program consisted of an initial step at 72 °C for 5 min; 94 °C for 30 s; 13 cycles of 94 °C for 30 s, 60 °C for 30 s, and 72°C for 30 s; and a final extension at 72 °C for 5 min. PCR products were size-selected and purified using magnetic beads. Final libraries were sequenced on an Illumina NovaSeq 6000 platform with 150 bp paired-end reads.

Bioinformatics analysis

All bioinformatics analyses were performed on a Linux server. For amplicon sequencing data, paired-end reads were merged using PANDAseq³⁹. Dereplication, chimera removal, and generation of operational taxonomic units (OTUs) were performed using VSEARCH⁴⁰. The processed sequences were clustered into OTUs with a threshold of 97%. Taxonomic assignment to the species level was carried out using BLAST+⁴¹. against the NCBI NT database (https://www.ncbi.nlm.nih.gov), with a minimum similarity threshold of 97%. For metagenomic data, reads were de novo assembled into contigs using MEGAHIT⁴². Microbial community composition was profiled using Kraken2 for taxonomic classification⁴³, and Bracken⁴⁴ was used for abundance estimation, with the GTDB database as the reference ⁴⁵. To specifically identify ascidian species, a custom mitochondrial gene database was constructed using genomic and mitochondrial sequences (including COX1, ND1, CYTB, and 12 other mitochondrial genes) from 260 Ascidiacea species, sourced from NCBI⁴⁶, BOLD⁴⁷, and MIDORI2⁴⁸. Species-level identification was performed using BLAST+, with stringent thresholds set at >99% identity, >90% query coverage, and an e-value < 1 × 10⁻¹⁰. Marine invertebrate composition at the phylum level was assessed by re-analyzing metagenomic data with Kraken2 and Bracken against the NCBI NT database, followed by manual curation to exclude non-marine metazoan taxa.

Statistical analysis

Statistical analyses were performed using GraphPad Prism (v8.0)⁴⁹ and R (v4.4.1)⁵⁰. The results are presented as mean ± standard deviation (x ± SD) from three independent experiments. Differences between groups were assessed using one-way ANOVA or two-way ANOVA, followed by post-hoc multiple comparisons with Bonferroni correction, with a significance threshold of P < 0.05. Significance levels are denoted as follows: * P < 0.05, ** P < 0.01, and *** P < 0.001.