A Biotechnological Approach to Enzyme‐Based Fertilisers: Immobilisation of Acid Phosphatases

ABSTRACT

We explore enzyme‐based technologies as sustainable alternatives to conventional chemical fertilisers, addressing the challenges associated with using enzymes in free or immobilised form for agricultural applications. We use the metagenome‐derived Class A acid phosphatase M2‐32, selected for its high activity, broad pH tolerance and thermophilic properties, and evaluated its immobilisation on clay minerals to enhance stability and applicability in soils. Several clays were tested as immobilisation supports. Bentonite caused complete enzyme inactivation, while kaolin formed aggregates and was unsuitable. In contrast, palygorskite, sepiolite and agrozeolite adsorbed more than 99% of the added enzyme. However, only a fraction of the immobilised enzyme retained catalytic activity, with optimal performance observed at moderate protein loading (40–80 μg protein). Among the tested supports, palygorskite consistently provided the highest specific activity (22,000 ± 2200 U/mg), followed by sepiolite (11,000 ± 730 U/mg), whereas agrozeolite (2250 ± 40 U/mg) showed comparatively low activity. ATR‐FTIR spectroscopy confirmed successful enzyme immobilisation without significant alteration of the clay structures. Immobilised M2‐32 preserved a broad pH range (between 4 and 8.5) and thermophilic behaviour similar to the free enzyme, remaining active up to 50°C. Immobilisation increased substrate affinity while reducing V_max relative to the free enzyme. To assess environmental compatibility, the effects of free and palygorskite‐immobilised M2‐32 on soil microbial communities were evaluated using corn rhizosphere microcosms with different organic matter contents. Metabarcoding high‐throughput sequencing revealed that microbial diversity and community structure were primarily shaped by soil type, plant presence and incubation time. Enzyme application, whether free or immobilised, did not significantly alter microbial diversity or composition. Overall, these results support palygorskite‐immobilised M2‐32 as a promising, environmentally compatible candidate for enzyme‐based fertiliser development.

The immobilization performance of M2‐32 varied with clay type, with palygorskite providing the highest catalytic reaction and stability. After immobilization, the enzyme preserved its broad pH and temperature tolerance. Neither the free nor the immobilized protein produced measurable changes in the rhizosphere microbial diversity. Instead soil type, plant presence and incubation time were the main determinants of microbial structure. These findings identify palygorskite‐immobilized M2‐32 as a robust, and environmentally compatible biocatlyst suitable for further application.

1. Introduction

Enzymes are highly efficient biological catalysts that participate in a wide range of biochemical reactions, thereby shaping metabolic networks, regulating metabolite fluxes, and enabling both intracellular and intercellular communication (Al Zubaidi et al. 2025; Robinson 2015). In living organisms, enzymatic activities are finely regulated and optimised rather than maximised, a feature that ensures metabolic balance but challenges their direct biotechnological applications (Duque et al. 2018). Therefore, successful deployment of enzymes outside their native biological context requires careful optimisation to ensure efficiency and appropriate mode of application. A key consideration is whether enzymes should be used in their free form in solution or immobilised onto a suitable carrier (Wahab 2025).

The development of sustainable alternatives to conventional chemical fertilisers is an urgent priority, driven by the environmental costs associated with intensive use of nitrogen, phosphorus and potassium (NPK) inputs (Shahzad et al. 2025). Excessive application of chemical fertilisers contributes to soil degradation, eutrophication of aquatic ecosystems, greenhouse gas emission and long‐term loss of microbial diversity and soil fertility (Timmis and Ramos 2021). In this context, enzyme‐based biological fertilisers represent a promising alternative, as they enhance nutrient mobilisation directly within soil ecosystems and reduce reliance on chemical inputs (Ramos 2022; Beltran‐Medina et al. 2023; Iqbal et al. 2024).

Our research aims to develop enzyme‐based fertilisers that promote the mobilisation of organically bound nitrogen and phosphorus in soils. We hypothesise that soil organic matter, derived from the decomposition of microbes, animals and plants, is naturally enriched in organophosphorus compounds, while plant seeds contain high amounts of phytate, their primary phosphorus reserve (Helfenstein et al. 2024). Furthermore, approximately 15% of the dry weight of organic matter consists of nitrogen, predominantly present in proteins and other nitrogen‐rich molecules (Cotrufo et al. 2013; Lehmann and Kleber 2015). We hypothesise that targeted enzymatic hydrolysis of these reservoirs can increase the bioavailability of essential nutrients in the rhizosphere, thereby enhancing plant nutrition in a manner aligned with the UN Sustainable Development Goals (SDGs) (Crowther et al. 2024).

Free enzymes are widely used in fermenter‐based biotechnological processes, including, for instance, amylases for starch breakdown, enzymatic cocktails for lignocellulose deconstruction, or proteases for the production of L‐amino acids from proteins (Álvarez et al. 2016; Chukwuma et al. 2020; Contato et al. 2025; Godoy et al. 2021; Kuddus et al. 2024; Liu and Kokare 2017; Reyes‐Sosa et al. 2017). However, for industrial applications in complex and heterogeneous environments such as soils, enzyme immobilisation offers distinct advantages (Hassan et al. 2025; Sheldon and van Pelt 2013). Immobilised enzymes generally exhibit enhanced stability, the potential for reusability and protection against harsh environmental conditions (Bolivar et al. 2022; Huang et al. 2009; Rosas et al. 2008; Shirvani et al. 2020; Zhu et al. 2010). These properties are relevant for agricultural applications, where enzymes must remain active under fluctuating pH, temperature and moisture conditions.

Within our nitrogen and phosphorus (NP) enzyme‐based fertiliser framework, we focus on thermostable enzymes capable of functioning in broad pH and temperature ranges (Recio et al. 2024, 2025). As an initial step, we have concentrated on enzymes involved in the phosphorus cycle. Through metagenomic screenings, we identified two acid phosphatases with broad substrate specificity and maximal activity between 20°C and 55°C. One enzyme, designated FS6, is a decameric Class C acid phosphatase (Recio et al. 2024); while the second, M2‐32, belongs to Class A acid phosphatases and crystallises as a dimer of dimers (Recio et al. 2025). Although both acid phosphatases are relatively thermotolerant, M2‐32 exhibits the remarkable ability to refold after thermal denaturation, a property not observed for FS6. Moreover, M2‐32 protein can be purified to homogeneity at high concentrations and exhibits a high specific activity comparable to that of industrially used enzymes. This combination of biochemical and biophysical properties led us to prioritise M2‐32 for immobilisation studies.

The selection of an appropriate immobilisation support is crucial for both industrial processes and environmental compatibility. Among the materials commonly used, functionalised and non‐functionalised plastic‐based carriers represent an important class due to their versatility and proven effectiveness in industrial settings (Hernández‐Arriaga et al. 2022; Jesionowski et al. 2014; Zdarta et al. 2018). While effective in industrial settings, they are poorly suited for agricultural applications due to their persistence in the environment and their potential to generate micro‐ and nano‐plastic pollutants that can propagate through terrestrial and aquatic food webs (Cusworth et al. 2024; Jin et al. 2022). In contrast, clay minerals represent an attractive alternative because of their natural abundance, biocompatibility and high surface area (Chen and Arai 2023). Furthermore, under natural soil conditions, phosphatases are often adsorbed or immobilised on clay minerals and other mineral surfaces (Burns et al. 2013; Quiquampoix and Burns 2007). Therefore, we have explored clay‐derived carriers, which have been used for immobilisation of various industrial enzymes, including phytases in animal feed (An et al. 2015; Basso and Serban 2019; Maghraby et al. 2023).

In this study, we assessed the immobilisation performance of several clay minerals and identified palygorskite as the most effective support for M2‐32. The performance of immobilised M2‐32 was tested in maize rhizosphere assays to assess its impact on microbial communities. Statistical analyses of grouped parameters revealed that supplementation with the M2‐32 enzyme had no significant effects on microbial communities, supporting the feasibility of enzyme‐based biofertilisers in sustainable agriculture. Overall, this work provides a proof‐of‐concept for the development of environmentally compatible, non‐chemical alternatives to conventional NPK fertilisers based on immobilised enzymes.

2. Materials and Methods

2.1. Enzyme Purification

Escherichia coli BL21 (DE3) cells were grown in 2 L conical flasks with 400 mL LB medium under continuous shaking at 200 rpm. Induction of M2‐32 expression was done as described before by Recio et al. (2025). Cells were harvested by centrifugation, and pellets were frozen at −20°C until use for protein purification. About 1.5 g cell pellet was suspended in 25 mL of Buffer A (40 mM MES, 40 mM acetic acid, 40 mM HEPES, pH 7.0, 500 mM NaCl, 1 mM dithiothreitol (DTT), 10% (v/v) glycerol and EDTA‐free protease inhibitor cocktail). Cells were lysed by three passes through a French Press at 1000 psi. The lysate was then centrifuged at 5000 g for 30 min, and the supernatant filtered through a sterile 0.22 μm filter. The clarified lysate was then loaded onto a 5 mL His‐Trap chelating column placed in an FPLC (GE Healthcare, St. Gibes, UK). Bounded proteins were eluted using a linear gradient of 0–500 mM imidazole in Buffer B (Buffer A but with 500 mM imidazole). To verify the purity and presence of the target protein, fractions were analysed by SDS‐PAGE 12% (w/v) gels. Fractions containing homogeneous M2‐32 protein were pooled and dialyzed overnight against Buffer A but with 150 mM NaCl instead of 500 mM NaCl.

2.2. Protein Immobilisation

For enzyme immobilisation, 8 U of M2‐32 enzyme in 5 mL immobilisation buffer (10 mM MES, 10 mM HEPES at pH 7.0) was mixed with 0.5 g of clay and gently stirred at room temperature. The pH of the resulting clay suspensions was approximately 7.5 for all clays, except for sodium bentonite, which showed a higher pH of around 8.5 (Table S2).

During immobilisation, aliquots were collected at defined time intervals and centrifuged (10,000 g, 10 min) to separate the supernatant from the clay. The residual enzymatic activity in the supernatant was determined using the standard assay (see below). After completion of the process, the suspension was filtered through Whatman Grade 1 filter paper and washed thoroughly with distilled water to remove unbound enzyme. The recovered clay was weighed using a precision balance.

Loading efficiency (%) was defined as the ratio between the amount of enzyme bound to the clay and the initial total amount of enzyme added, according to the following equation:

$$\text{Loading efficiency}\left(\%\right)=\frac{\text{Immobilised enzyme protein content}\left(\frac{\mathrm{\mu g}}{\mathrm{mL}}\right)}{\text{Free enzyme protein content}\left(\frac{\mathrm{\mu g}}{\mathrm{mL}}\right)}\times 100$$

For these assays, initial protein amounts were between 40 and 240 μg protein/0.5 g clay. Protein concentration before immobilisation and in the supernatant was determined using a NanoDrop One C (ThermoFisher, WI, USA) operated at a wavelength of 240 nm.

2.3. Enzymatic Assays

Enzymatic activity was measured following the method described by Recio et al. (2025). All assays were done in triplicate using at least three independently immobilised enzyme preparations. Immobilised enzyme (0.5 g clay) was resuspended in 5 mL immobilisation buffer, and 100 μL of the suspension was used per assay. Reactions were started by adding 20 μL of 0.115 M p‐nitrophenyl phosphate (pNPP) as substrate at 25°C and pH 5.5. After 15 min incubation, reactions were terminated by adding 100 μL of 0.5 M NaOH, followed by the addition of 800 μL of distilled water to obtain a final total volume of 1.02 mL. Samples were centrifuged for 1 min at 10,000 g, and the absorbance of the reaction product (ρ‐nitrophenol) in the supernatant was measured at 405 nm using a Tecan Sunrise plate reader (Tecan Austria GmbH). The concentration of p‐nitrophenol was calculated using an extinction coefficient at 405 nm of Ʃ = 18,000 M⁻¹ cm⁻¹. One unit (U) of acid phosphatase activity was defined as the amount of enzyme producing 1 μmol p‐nitrophenol per minute under the assay condition. Specific activity was defined as the number of enzyme units per milligram of total protein (U mg⁻¹).

To evaluate the effects of temperature and pH on enzymatic activity, standard assays were carried out at temperatures of 20°C, 30°C, 40°C, 50°C, 60°C, 70°C and 80°C, and at pH values of 4, 5.5, 7 and 8.5.

For kinetic assays, pNPP was added at increasing concentrations (0–0.22 mM). The initial reaction rates (v₀) were plotted as a function of substrate concentration, and kinetic parameters (V_max and K_m) were determined by fitting the data to the Michaelis–Menten equation model and by Lineweaver–Burk linearisation (R ² > 0.9).

2.4. ATR‐FTIR Spectroscopy

To characterise the functional groups of the different clays and to confirm enzyme immobilisation, Attenuated Total Reflectance Fourier Transform Infrared (ATR‐FTIR) spectroscopy was performed using a JASCO FT/IR‐6200 type A spectrophotometer equipped with an ATR PRO ONE accessory (45° incident angle). Spectra were recorded in the range of 400–4000 cm⁻¹ at a resolution of 2 cm⁻¹, with 60 accumulations per sample. Samples were analysed directly as dry powders placed on the ATR crystal, ensuring uniform contact. Spectra were analysed using Spectra Manager software (JASCO), and final graphs were generated with Origin 2025.

2.5. Rhizosphere Assays With M2‐32

A cambisol soil was collected from El Padul, Granada, Spain (37° 00″ 55.1′ N, 3° 38″ 50.5′ W). The soil was sieved through a 2‐mm mesh to remove stones, large particles and plant debris. This soil was then mixed with 20% (w/w) sand and divided into two portions: one was left unamended (Soil A), while the other was supplemented with compost to a final concentration of 1% (wt/wt) (Soil B). Three independent samples of 1 g were collected to determine the initial microbiota composition.

Pots containing 1 kg of Soil A or Soil B were then prepared. One seed, pre‐germinated on moist filter paper, was sown per pot. For each of the three treatments 5 pots were used. The treatments consisted of: (i) control (unamended soil), (ii) soil amended with free M2‐32 enzyme (1 mg/mL) and (iii) soil amended with M2‐32 immobilised on palygorskite clay. Samples were collected 15 days and 30 days after sowing.

The samples were analysed as follows (Table S1): Group 1 (G1): Soil A before seeding; Group 2 (G2): Soil B before seeding; Group 3 (G3): Soil A control after 15 days; Group 4 (G4): Soil A with free enzyme after 15 days; Group 5 (G5): Soil A with enzyme in palygorskite after 15 days; Group 6 (G6): soil B control after 15 days; Group 7 (G7): Soil B with free enzyme after 15 days; Group 8 (G8): soil B with enzyme in palygorskite after 15 days; Group 9 (G9); soil A control after 30 days; Group 10 (G10): soil A with free enzyme after 30 days; Group 11 (G11): soil A with enzyme in palygorskite after 30 days; Group 12 (G12): soil B control after 30 days; Group 13 (G13): soil B with free enzyme after 30 days and Group 14 (G14): soil B with enzyme in palygorskite after 30 days.

2.6. Total DNA Isolation, Amplicon Sequencing and Bioinformatics Analysis

Total DNA was extracted from the rhizosphere using the Fast DNA Spin Kit for Soil (MP Biomedicals LLC, OH, USA), following the manufacturers' recommendations. DNA quality and concentration were evaluated using a Nanodrop One C (ThermoFisher, WI, USA) spectrophotometer. Amplicon sequencing of the V3–V4 region of the bacterial 16S rRNA gene was performed by Novogene Co. Ltd. (Cambridge, UK) using Illumina paired‐end technology (470 bp) with the primers 5′ CCTAYGGGRBGCASCAG 3′ and 5′ GGACTACNNGGGTATCTAAT 3′. Raw sequencing data were obtained as forward and reverse reads with barcodes after removing primer sequences. Subsequent bioinformatic analyses were conducted using QIIME2 (version 2022.2; https://qiime2.org) and graphical outputs were created using Origin 2025.

3. Results and Discussion

3.1. Immobilisation of an Acid Phosphatase Enzyme on Clays

Based on its physicochemical properties, the Class A M2‐32 acid phosphatase, rescued from a metagenomic library, was purified to homogeneity as described (Recio et al. 2025), and subsequently used to study its immobilisation on different clay carriers. The clays evaluated included palygorskite, sepiolite, bentonite, kaolin and agrozeolite. All clays were suspended in the immobilisation buffer described above and showed no intrinsic acid phosphatase activity. These clays possess high specific surface areas (Table S2) (Abbaslou et al. 2020; Bahranowski et al. 2021; Cataldo et al. 2021; Galan 1996; Galan and Carretero 1999; Grifasi et al. 2024; Karimi and Salem 2011; Liu et al. 2013; Madejová 2003; Meirelles et al. 2023), which favour interactions with organic molecules and proteins (Chen and Arai 2023). With the exception of sodium bentonite, that displays a basic pH (Table S2) the other clays present neutral pH values when suspended in water (Table S2). Given that M2‐32 is highly stable over a pH range between 4 and 8.5 (Recio et al. 2025), both enzyme binding and retention of catalytic activity were measured. For the initial tests, approximately 8 U of enzyme in 5 mL of immobilisation buffer were incubated with 0.5 g of clay, as described in Materials and Methods. The enzyme activity associated with the clay fraction and the residual activity in the supernatant were monitored over a 2 h incubation period. At the end of the incubation, no activity was detected in the supernatant (not shown), indicating complete adsorption of the enzyme onto the tested clays. At different time points, the clays were recovered and the activity of the immobilised acid phosphatase was determined (Figure 1). After 120 min, no activity was detected in the bentonite‐bound enzyme, whilst the other clays retained between 50% and 99% of the initial acid phosphatase activity (Figure 1).

FIGURE 1.

Immobilisation of M2‐32 acid phosphatase on different clays over time. Immobilisation was carried out as described in Materials and Methods. Aliquots were collected at the indicated time points, and the percentage of activity recovered on the clay is shown. Data points represent the mean and the standard deviation of three independent experiments, each done in triplicate using three different protein preparations. Grey squares, palygorskite; blue triangles, sepiolite; magenta diamonds, agrozeolite; red circles, kaolin and inverted green triangles, bentonite.

The complete loss of activity observed with bentonite was attributed to enzyme inactivation upon adsorption, and this clay was excluded from further assays. Kaolin was also discarded, despite its moderate activity recovery, because it tended to form aggregates that often resulted in non‐homogeneous suspensions. Therefore, subsequent analyses were carried out using palygorskite, sepiolite and agrozeolite.

Binding was further evaluated at different protein loadings, ranging from 40 to 240 μg per 0.5 g of clay. As assessed by loading efficiency (%), nearly all M2‐32 was adsorbed onto each of the three clays under study, independently of the initial protein load (not shown). We then calculated the specific activity (U/mg protein) retained on the clay, normalised to the amount of enzyme added during the immobilisation. For M2‐32, lower protein loadings (40–80 μg/0.5 g) resulted in higher specific activities than higher loadings (120–240 μg/0.5 g) for all three clays (Figure S1). Thus, although essentially all of the enzyme added during the immobilisation process was adsorbed, only a fraction retained catalytic activity, particularly at higher protein loadings (Figure S1), consistent with previous reports (Shirvani et al. 2020).

Since the highest specific activity was recorded at a protein loading of 80 μg, this loading was chosen for subsequent optimisation. Then, specific activity was determined using 80 μg of protein immobilised on each clay, following the standard protocol described in Materials and Methods. Specific activity ranged from 22,000 ± 2200 U/mg in the case of palygorskite, 11,000 ± 730 U/mg for sepiolite and 2250 ± 40 U/mg for agrozeolite (Figure 2).

FIGURE 2.

Specific activity of M2‐32 acid phosphatase adsorbed onto different clays. Specific activity is expressed as U/mg protein and was measured at 25°C and pH 5.5. The given values represent the mean and standard deviation of three independent assays, each done in triplicate.

To confirm true immobilisation, Attenuated Total Reflectance Fourier Transform Infrared (ATR‐FTIR) spectroscopy was performed on the three selected clays. The IR spectrum of the free protein in buffer displayed characteristic vibration bands at 1044 and 1430 cm⁻¹ that were absent in the buffer control (Figure S2). The band near 1430 cm⁻¹ can be attributed to CH bending vibrations or symmetric stretching of carboxylic groups, while the signal around 1044 cm⁻¹ is consistent with C–O stretching vibrations associated with functional groups present in protein side chains (Yang et al. 2015) (Figure S2). Figure 3 shows the FTIR spectra of each clay alone (green traces) and after protein loading at 500 μg and 1000 μg (blue and red traces, respectively). Increasing protein loading resulted in a progressive increase in spectral signal, suggesting successful enzyme immobilisation (Figure 3). Upon immobilisation onto palygorskite (Figure 3A) (Yan et al. 2012), sepiolite (Figure 3B) (Alkan et al. 2005) and agrozeolite (Figure 3C) (Król et al. 2012), the characteristic Si–O–Si bands in the 508–900 cm⁻¹ range were not significantly altered (right panels). In contrast, the appearance of amide I and II bands at 1641/1657 and 1409/1415 cm⁻¹ confirmed binding of a protein rich in alpha‐helical structure on the clay surfaces. Additional bands between 977 and 1000 cm⁻¹ were assigned to C–N and C–O vibrational modes associated with the immobilised enzyme.

FIGURE 3.

ATR‐FTIR spectra of the different clays tested in the absence and in the presence of immobilised protein. (A) palygorskite, (B) sepiolite and (C) agrozeolite. Right panels show the spectra in the 400 and 1200 cm⁻¹ region, while left panels show spectra between 1200 and 2000 cm⁻¹. Spectra of clays alone (black line) are compared with those of clays containing immobilised protein loadings of 500 μg (red line) and 1000 μg (blue line). Relevant absorption bands are indicated.

The free M2‐32 enzyme exhibits maximal activity between 30°C and 55°C and remains functional in the pH range between 4.0 and 8.5 (Recio et al. 2025). Based on these properties, we evaluated the activity of immobilised M2‐32 at different pH values and temperatures. It should be noted that M2‐32 immobilised on agrozeolite consistently exhibited low activity across the tested pH and temperature ranges (Figure 4A,B). In contrast, M2‐32 immobilised on palygorskite and sepiolite retained high catalytic activity. For both clays, the immobilised enzyme remained active up to 50°C (Figure 4A), although specific activity (measured as U/mg protein) was consistently higher on palygorskite than on sepiolite (i.e., approx. 22,000 ± 2000 vs. 12,000 ± 730 U/mg protein). Activity declined sharply at 60°C (Figure 4A). The immobilised enzyme also retained high activity in the pH range between 4.0 and 8.5 (Figure 4B). Overall, these results mirror the behaviour of the free enzyme in solution and confirm the thermophilic potential and broad pH tolerance of M2‐32 following immobilisation.

FIGURE 4.

Acid phosphatase activity of M2‐32 immobilised on different clays under varying temperature and pH conditions. Assays were performed as described in the Materials and Methods section. (A) Enzyme activity measured at temperatures ranging from 20°C to 80°C with a reaction time of 15 min. (B) Enzyme activity measured at different pH values using Buffer A, assays were conducted for 15 min. Black line: Palygorskite; red line: Sepiolite; and blue line: Agrozeolite. The pH was maintained constant throughout each assay. Values represent the mean and standard deviation of three independent assays done in triplicate.

A distinctive property of M2‐32 in solution is its ability to refold and regain activity after heating for 60 min at 80°C (Recio et al. 2025). We have examined if the immobilised M2‐32 retained this property. Heating immobilised M2‐32 for 60 min at 80°C resulted in complete loss of enzymatic activity, with no recovery observed upon cooling at room temperature for 2 h. These results indicate that immobilisation inhibits the intrinsic refolding capability of M2‐32, consistent with previous observations for immobilised enzymes (Datta et al. 2017; Huang et al. 2009).

We also evaluated the storage stability of immobilised M2‐32 on the different clays over 24 to 192 h at 4°C, room temperature (15°C–20°C) and 30°C (Table 1). The highest stability was observed at 4°C, consistent with the free enzyme. After 48 h, M2‐32 immobilised on agrozeolite and palygorskite retained 50%–60% of their initial activity, while only about 30% activity was retained with sepiolite (Table 1). With palygorskite, activity remained (> 20%) after 192 h storage (Table 1). At room temperature (15°C–20°C) agrozeolite showed the highest relative stability; however, activity dropped by 70% after 96 h and by 85% after 192 h (Table 1). Storage at 30°C resulted in significant rapid loss of activity in all clays (not shown).

TABLE 1.

Storage stability of M2‐32 immobilised on the different clays: 80 μg of M2‐32 were immobilised on palygorskite (P), sepiolite (S) and agrozeolite (A).

Time (h)	4°C			RT
	P	S	A	P	S	A
0	100%	100%	100%	100%	100%	100%
24	72%	53%	85%	30%	17%	59%
48	47%	27%	61%	24%	10%	40%
96	32%	19%	59%	15%	7%	30%
192	21%	12%	18%	7%	5%	16%

Note: The initial activity of the immobilised enzyme for each clay was defined as 100%. Samples were then stored at 4°C and room temperature (RT), and the residual activity was measured at the indicated time points.

After 24 h of storage at 4°C, the kinetic parameters of immobilised M2‐32 were determined using pNPP as substrate. The apparent K_m (K_{m app}) values ranged from 0.099 ± 0.05 to 0.28 ± 0.1 mM (Table 2), which are slightly lower than the K_m determined for the free enzyme in solution (0.49 ± 0.12 mM). The V_max of the free enzyme at pH 5.5 and 25°C was 3.416 × 10⁻⁷ ± 6.64 × 10⁻⁸ M s⁻¹ (Recio et al. 2025). However, the enzyme immobilised on clays exhibited lower V_{max app} than the free enzyme. M2‐32 immobilised on palygorskite and agrozeolite showed higher V_{max app} (4.55 ± 1.55–8.43 ± 2.73 × 10⁻⁸ M s⁻¹) than the enzyme immobilised on sepiolite (5.79 ± 2.32 × 10⁻⁹ M s⁻¹) (Table 2). Therefore, immobilisation reduced V_{max app} but increased apparent substrate affinity, consistent with restricted enzyme mobility and diffusion effects commonly observed for immobilised enzymes. Values of catalytic efficiency (V_{max app}/K_{m app}) are also reported in Table 2.

TABLE 2.

Summary of M2‐32 immobilisation efficiency and activity parameters for the different clays.

Protein	Clays	Loading efficiency	Supernatant activity	Clay activity	Stability	Temperature profile	pH profile	Cycles	Kinetic parameters
M2‐32	Palygorskite	99.9%	None	Major	Optimal storage at 4°C, activity is lost with rising temperature	Similar to free protein	Best pH 5.5	> 4 cycles	K m (mM) 0.28 ± 0.1 V max (M/s) 4.55E‐08 ± 1.55E‐08 kcat (s −1 ) 2.31E‐04 ± 1.12E‐04
	Sepiolite	99.9%	None	Intermediate	Optimal storage at 4°C, activity is lost with rising temperature	Optimal at 20°C–30°C	Not remarkable pH influence	> 4 cycles	K m (mM) 0.099 ± 0.05 V max (M/s) 5.79E‐09 ± 2.32E‐09 kcat (s −1 ) 6.36E‐05 ± 1.75E‐05
	Agrozeolite	99.9%	None	Minor	Optimal storage at 4°C, activity is lost with rising temperature	Similar to free protein	Not remarkable pH influence	> 4 cycles	K m (mM) 0.18 ± 0.05 V max (M/s) 8.43E‐08 ± 2.73E‐09 kcat (s −1 ) 4.68E‐04 ± 9.23E‐05
	None	—	—	—	Optimal storage at 4°C	Maximum from 30°C–55°C	Not remarkable pH influence	—	K m (mM) 0.48 ± 0.11 V max (M/s) 3.16E‐07 ± 6.44E‐08 kcat (s −1 ) 0.32 ± 0.06

The observed differences in catalytic behaviour among the immobilised systems indicate that the physicochemical characteristics of the supports play a critical role in modulating the balance between substrate diffusion and enzyme orientation. In particular, the uniformly low activity exhibited by the enzyme immobilised on agrozeolite across all tested pH and temperature conditions suggests that mass‐transfer limitations, likely caused by its microporous structure, as well as unfavourable multipoint adsorption, may restrict substrate access to the enzyme's active site, effectively masking its intrinsic catalytic performance. In contrast, the higher V_{max app} values observed with fibrous supports, especially palygorskite, imply improved external accessibility and a more favourable enzyme orientation that preserves catalytic function. Sepiolite, while also fibrous, showed comparatively lower V_max values, which may be attributed to stronger enzyme‐support interactions or steric hindrance effects that limit substrate access or reduce the dynamic flexibility of the enzyme. The overall reduction in V_{max app} upon immobilisation, combined with a decrease in apparent K_m is consistent with reduced enzyme mobility at high substrate concentrations and possible substrate partitioning at the support surface. These findings collectively highlight that support architecture and surface chemistry, rather than intrinsic enzymatic properties, are the dominant factors determining catalytic efficiency in these immobilised systems. Optimising surface properties—such as porosity, surface charge, hydrophilicity and functional group availability—is therefore essential to improving biocatalytic performance (Santos et al. 2015).

The apparent kinetic parameters indicate that enzyme performance is strongly influenced by the properties of the immobilisation support. Comparison of apparent catalytic efficiencies (V_max/K_m) reveals higher values for agrozeolite (4.68E‐04 ± 9.23E‐05), followed by palygorskite (2.31E‐04 ± 1.12E‐04) and sepiolite (6.36E‐05 ± 1.75E‐05). However, it is important to note that these values reflect a combination of intrinsic catalytic activity and transport phenomena inherent to immobilised systems. The high apparent efficiency observed for agrozeolite likely represents the activity of a localised fraction of enzyme molecules with favourable positioning within accessible surface regions, while the overall system remains constrained by micropore‐induced mass transfer limitations. In contrast, palygorskite exhibited a more balanced kinetic profile, consistent with enhanced substrate accessibility and a more favourable enzyme orientation on the external surface. Sepiolite, despite exhibiting a relatively low apparent K_m, displayed the lowest catalytic efficiency due to a markedly reduced V_max, suggesting that diffusional resistance or steric hindrance may limit turnover in the system. Overall, the observed decrease in apparent V_max accompanied by reduced K_m; values across all assays supports a model in which diffusion constraints limit maximal catalytic rates, while substrate partitioning at the support interface leads to enhanced apparent substrate affinity.

A key advantage of enzyme immobilisation is the enhanced stability it confers, allowing enzymes to be reused across multiple industrial cycles (Guisan et al. 2022; Homaei et al. 2013). Although immobilised enzymes cannot be recovered once applied to soil, we examined the reusability of M2‐32 immobilised on each of the three selected clays by monitoring the enzymatic activity over successive reaction cycles. Following a standard activity assay, the immobilised enzyme was recovered by filtration through Whatman grade 1 filter paper. The recovered clay‐enzyme complexes were then re‐used for subsequent determination of activity. Remarkably, we observed that after 4 consecutive assay cycles, nearly 100% of the initial activity was retained (Table 2), suggesting minimal loss of catalytic performance. These results support that immobilised M2‐32 can be readily recovered and reused across multiple catalytic cycles while maintaining high activity, supporting its suitability for repeated‐use applications.

Table 2 summarises the immobilisation performance of M2‐32 on different clays. More than 99% of the protein added to each clay was adsorbed; however, palygorskite consistently yielded the highest specific activity. In addition, M2‐32 immobilised on palygorskite showed good storage stability at 4°C and showed no significant changes in its pH or temperature activity profiles with respect to the free enzyme in solution. Although V_max values were reduced for all immobilised forms, substrate affinity was higher than that of the free enzyme. This set of results provides a solid foundation for subsequent assays aimed to evaluate the effects of immobilised M2‐32 on microbial communities in rhizosphere soils.

3.2. Minimal Impact of Free or Immobilised Enzymes on Rhizosphere Microbial Communities

As a subsequent step in characterising the behaviour of the M2‐32 immobilised on palygorskite, we evaluated if the application of the enzyme, either in free form or immobilised on this clay, affects soil microbial community structure. For these analyses, microcosm assays were established using a cambisol soil either unamended (Soil A) or amended with 1% (wt/wt) compost (Soil B). Soil samples were taken at the beginning of the assay and after sowing germinated corn seeds. Rhizosphere samples were collected after 15 and 30 days for total DNA extraction, and microbial community composition was analysed by 16S RNA gene metabarcoding, as described in Materials and Methods.

Rarefaction curves (Figure 5) revealed that sequencing depth and quality data were adequate for robust community profiling. Observed Amplicon Sequence Variant (ASV) curves (Figure 5A) approached a plateau at the achieved sequencing depths, indicating that most of the resolvable richness at the ASV level was captured. For the majority of samples, curve slopes flattened at around 15,000 reads, with minimal variation among replicates. Samples collected at later time points and those corresponding to soil B exhibited higher ASV richness, consistent with the higher organic matter content in this soil.

FIGURE 5.

Alpha diversity metrics for all samples analysed in this study. Samples were grouped into 14 experimental groups (G1‐G14), As described in Materials and Methods. Panels show: (A) Observed Amplicon Sequence Variants (ASV); (B) Shannon diversity index; (C) Good's coverage index. Each point represents the mean value calculated from 5 subsamples per group.

Shannon diversity curves (Figure 5B), which integrate both richness and evenness, showed consistent trajectories among replicates and rapidly stabilised with increasing sequencing depth, reaching a plateau at approximately 2500 reads. This stabilisation indicates that Shannon diversity estimates were robust and largely independent of sampling depth. Finally, Good's coverage values (Figure 5C) rapidly reached high values (> 95%–100% in most samples), indicating that nearly all detectable taxa were captured.

Alpha diversity was further assessed to evaluate microbial richness and evenness of the samples. Specifically, the Chao1 index was used as an estimator of species richness, the Shannon index to account for both richness and evenness, and Pielou's evenness index to assess the distribution of species abundance (Figure 6). Together, these complementary metrics provide an integrated view of the microbial community structure of the experimental groups G1–G14.

FIGURE 6.

Boxplot representation of Alpha diversity indices across samples groups (G1–G14). Panels show: (A) Shannon diversity index (community richness and evenness); (B) Pielou's evenness index (uniformity of species distribution); (C) Chao1 index (estimated species richness). Boxes represent the interquartile range (IQR), the central line indicates the median, whiskers represent the standard deviation, and individual points denote outliers where present.

The Shannon diversity index (Figure 6A) showed significant overall differences among groups according to the Kruskal–Wallis test (p = 0.026). However, pairwise post hoc comparisons for soils A and B—performed using Dunn's and Conover's tests—revealed no significant differences among individual treatments. This pattern suggests that the effect was due to the presence of seeds rather than to enzyme application.

The Pielou evenness index (Figure 6B) showed a slight decrease in community evenness over time in both types of soils, with values varying from 0.90 to 0.92, reflecting modest shifts in dominance of specific taxa. Although the Kruskal–Wallis test detected global differences (p = 0.002), no significant pairwise differences were identified, suggesting that changes in evenness were not associated with specific treatments.

The Chao1 diversity index (Figure 6C) increased after sowing the seeds in soil A, whereas this increase was less pronounced in soil B. These results suggest that compost amendment increased initial microbial diversity, rendering the organically rich soil richer than the soil without amendment. However, in both soils, corn root exudates influenced microbial richness, as taxa initially undetected were identified after sowing. The Kruskal–Wallis test (p = 0.018) also revealed significant global differences among groups.

Beta diversity analysis based on Bray‐Curtis distances revealed clear differences in microbial community composition among samples (Figure 7). Control soils (G1 and G2) were clearly separated from planted and treated samples, as expected due to the absence of plants and treatment. Principal Coordinates analysis (PCoA) showed a strong separation along PC1 (31.51%), which was mainly associated with soil type, indicating that soil characteristics were the primary determinant of community structure. In soil A, samples further separated according to sampling time (15 and 30 days), reflecting temporal shifts in microbial composition; but this pattern was not observed in soil B, where samples from both times clustered together.

FIGURE 7.

Principal Coordinates Analysis (PCoA). PCoA based on Bray–Curtis dissimilarity illustrating beta diversity of microbial communities across treatments, time points and soil types (groups G1–G14). Each point represents an individual sample, and distances between points reflect differences in community composition. Colour coding and shading are indicated on the right side of the figure.

Within each soil type, samples clustered closely regardless of treatment, indicating that enzyme application—either in the free form or immobilised—had only a minor or secondary influence on microbial community structure. In fact, in soil A, samples from the three planted conditions (seed only (G3), seed plus free enzyme (G4) and seed plus immobilised enzyme (G5)) clustered closely together, demonstrating that the enzyme treatment did not significantly alter the microbial community structure. Similar patterns were observed for soil B at both 15 days (G6, G7 and G8) and 30 days (G12, G13 and G14). Overall, these results indicate that enzyme application did not significantly alter rhizosphere microbial community composition.

Following alpha and beta diversity analyses, the relative abundance of microbial taxa was analysed at different taxonomic levels (phylum, family and the 30 most abundant genus), considering soil type, incubation time and treatment. Each soil was analysed separately and the results are shown in Figures 8, 9, 10. At the phylum level in soil A (Figure 8, left panel), marked shifts in microbial composition were observed 15 days after sowing (G3, G4 and G5), Actinobacteriota abundance substantially decreased from 60% to 25%, accompanied by increases in Proteobacteria (from 10% to 30%), Acidobactariota (from 5% to 15%) and Gemmatimonadota (from 10% to 25%). This trend was less pronounced in group G5, where the enzyme was supplied in immobilised form. Thirty days post‐sowing (G9, G10 and G11), Actinobacteriota partially recovered to around 55%, while Proteobacteria (22%), Acidobacteriota (5%) and Gemmatimonadota (5%) returned to levels close to the initial proportions. Proteobacteria remained enriched in the rhizosphere compared to bulk soil (G9, G10 and G11 vs. G1). No significant differences were detected among soils sown with corn seeds regardless of the addition of enzyme (G3 vs. G4 and G5; G9 vs. G10 and G11). Other phyla, including Bacteroidota (2%), Firmicutes (1%), Verrumicrobiota (1%), Planctomycetota (1%) remained relatively stable (Figure 8, left panel). In contrast, in soil B (Figure 8, right panel), sowing seeds with and without enzyme addition caused only modest shifts at the phylum level, although slight increases were observed in Actinobacteriota (from 50% to 55%), Proteobacteria (from 17% to 24%) and Crenarchaeota (from 1% to 5%), accompanied by decreases in Acidobacteriota (from 8% to 3%) and Gemmatimonadota (from 10% to 3%). In general, these shifts in soil B were less pronounced than those observed in soil A, indicating that organic matter amendment buffered microbial community responses at the phylum level.

FIGURE 8.

Relative abundance of the dominant bacterial phyla in the different experimental assays. Grouped bar plots show the percentage distribution of major bacterial phyla in soil A (left panel) and soil B (right panel).

FIGURE 9.

Relative abundance of the dominant bacterial families in the different experimental assays. Grouped bar plots represent the relative proportion of bacterial families in soil A (left panel) and soil B (right panel).

FIGURE 10.

Relative abundance of the dominant bacterial genera in the different experimental assays. Clustered bar plots show the relative distribution of major genera in soil A (left panel) and soil B (right panel).

At the family level, moderate differences associated with soil type and sampling time were observed (Figure 9). In Soil A (Figure 9, left panel) the dominant families were 67–14 (12%), Solirubrobacteraceae (10%), Gemmatimonadaceae (7%) and Sphigomonadaceae (3%). Shortly after sowing, Sphingomonadaceae increased from 3% to 10% and Gemmatimonadaceae from 7% to 10%, before gradually returned to the initial levels at later sampling times (G3, G4 and G5 vs. G9, G10 and G11). With longer incubation times, Micromonosporaceae (3%) and Gemmatimonadaceae (3%) became relatively more abundant, particularly in treatments where the enzyme was supplied immobilised on clay, suggesting a possible benefit from enzyme‐mediated changes in substrate availability for members of these families (see G5 and G11). In soil B (Figure 9, right panel), family‐level composition remained stable over time. Seed sowing caused a decrease in Sphingomonadaceae (from 6% to 3%) and Gemmatimonadaceae (from 8% to 2%) families, accompanied by slight increases in the Solirubrobacteraceae (from 8% to 10%) and Geodermatophilaceae (from 2% to 6%). These trends were similar to those observed in soil A, but were less pronounced, further supporting the stabilising effect of organic matter supplementation. Consistent with phylum level observations, the addition of enzyme did not provoke significant changes in family‐level composition.

Finally, microbial composition was examined at the genus level (Figure 10). In Soil A (Figure 10, left panel), temporal shifts were observed. For instance, 10 days after sowing the relative abundance of Sphingomonas increased from 2% to 10%, with a concomitant decrease in Solirubrobacter (from 7% to 3%) and Gaiella (from 4% to 1%). By 30 days, increases in Blastococcus (from 2% to 6%) and Rubrobacter (from 1% to 4%) were observed. At this later time point, the 30 most abundant genera together accounted for approximately 50% of the total community, indicating a greater contribution of less dominant taxa. This pattern is consistent with the increase in Shannon diversity described above, and it happened irrespective of enzyme treatment, indicating that it was primarily driven by plant presence and incubation time.

In Soil B (Figure 10, right panel), genus‐level shifts were more moderate but followed similar trends. A decrease in Sphingomonas (from 4% to 1%) and an increase in Blastococus (from 2% to 5%) and Rubrobacter (from 1% to 4%) were observed among the most abundant genus. As in Soil A, the relative contribution of the 30 most abundant genera increased over time, from approximately 30% to nearly 40% of the total community. As in Soil A, these changes were likewise independent of enzyme addition, suggesting that temporal dynamics and plant‐associated effects outweighed any treatment‐related influence at the genus level.

3.3. Limitations and Future Prospects

Immobilisation should be viewed not as a strategy for recyclability but as an approach to improve formulation and spatial retention. In this context, immobilisation onto natural clay carriers enhances handling, enables targeted delivery to the rhizosphere, and offers partial protection against dilution, adsorption, or proteolytic degradation in soil microenvironments. Although immobilisation reduced the apparent catalytic rate compared to the free enzyme, it preserved activity across a broad pH and temperature range, supporting its practical relevance for field applications. Among the tested clays, clay type, palygorskite, showed superior performance in terms of catalytic retention and stability. Despite these advantages, several limitations still remain. A key challenge lies in scaling up from laboratory conditions to industrial production. Industrial deployment requires not only robust enzymatic performance but also high production yields, stable formulations and cost‐efficient manufacturing processes (Adrio and Demain 2014). In agricultural applications, purification and downstream processing can account for a large fraction of total production costs, while optimisation of expression systems and fermentation conditions is often time‐consuming and resource‐intensive. This constraints highly the need for further technological development before this enzyme can be realistically applied as a commercial biofertiliser.

Additionally, current evaluations were conducted under controlled pot and growth‐chamber conditions, which, although useful for initial evaluation, do not fully replicate the complexity of natural agricultural systems (Ge et al. 2024; Belt et al. 2026). Under field conditions, enzyme performance may be affected by adsorption to soil particles, degradation by proteases, UV exposure and interactions with native microbial communities (Gomez et al. 2020; Jindo et al. 2023; Richardson and Simpson 2011). Therefore, field trials with different crops, soil types and climate conditions are essential to validate efficacy, refine application strategies and assess long‐term agronomic impacts.

From a broader perspective, the extensive diversity of acid phosphatases in bacteria and fungi (Udaondo et al. 2020; Gómez‐Gallego et al. 2026) suggests potential for identifying enzymes with enhanced catalytic properties. Moreover, advances in delivery methods, such as seed coatings, irrigation‐based systems, or slow‐release formulations, as well as the development of improved carrier materials may further enhance stability and field performance.

Overall, while important challenges remain, this study provides a solid proof of concept and highlights the potential of immobilised enzymes as innovative tools for sustainable nutrient management in agricultural soils.

Author Contributions

María‐Isabel Recio: conceptualisation, methodology, investigation, formal analysis, writing – original draft, writing – review and editing. Jesús de la Torre: methodology, investigation, writing – review and editing. Javier Rocha‐Martin: methodology, supervision, writing – original draft, writing – review and editing. Isabel de la Mata: methodology, supervision, writing – original draft, writing – review and editing. Juan Luis Ramos: conceptualisation, investigation, formal analysis, supervision, funding acquisition, project administration, writing – original draft, writing – review and editing.

Funding

This study was supported by Junta de Andalucía Grant PREDOC_01447 and the Plan Nacional PID2021‐123469OB‐IOO funded by MICIU/AEI/10.13039/501100011033 and by ‘ERDF A way of making Europe’, by ‘ERDF/EU.’

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Figure S1: Percentage of activity recovered (%) for different M2‐32 loadings. M2‐32 was immobilised on palygorskite (grey bars), sepiolite (blue bars) and agrozeolite (red bars). Although all of the added enzyme adsorbed onto the clays, as no activity was detected in the supernatants, higher activity was obtained at lower protein loadings for all three clays.

Figure S2: ATR‐FTIR of immobilisation buffer in the absence and in the presence of 500 μg of M2‐32. The grey line corresponds to the buffer alone, while the green line represents the buffer containing protein. Characteristic protein absorption bands at 1044 and 1430 cm−1 are indicated.

Table S1: Experimental design and distribution of soil type, treatment and sampling time across groups.

Table S2: Summary of the main physicochemical properties of the clays used in this study. The pH values correspond to measurements of the pH of the immobilisation buffer or water, after addition of 0.5 g clay, and were recorded once equilibrium was reached.

Data Availability Statement

The original contributions presented in this study are included in the article and the Supporting Information.