Study on antibacterial properties of heated oyster shell particle against Bacillus subtilis spores in rainwater by response surface methodology based on central composite design

Oysters, a nutrient-rich bivalve mollusk (Chakraborty et al. 2016; Zhu et al. 2018; Maurya 2021), can be found in many parts of the world. They inhabit salty or brackish coastal waters, congregating on old shells, rocks, piers, or hard, submerged surfaces. In Taiwan, oyster cultivation accounts for more than one-sixth of the total aquaculture area, generating more than 2 billion USD in annual revenue (Vaschenko et al. 2013). Along the western coast of Taiwan, oyster fields stretch from Xiangshan in Hsinchu City in the north to Dapeng Bay in Pingtung County in the south (Liu et al. 2015; Ueng et al. 2020). Although oyster farming has substantial economic potential, the environmental impact from its by-products/wastes is unavoidable. Oyster shells (OSs) abandoned along the coast can cause a variety of health and environmental issues, involving noxious odors (Chilakala et al. 2019; Sadeghi et al. 2019) from the decomposition of remaining attached flesh, natural water contamination, soil pH increase, and marine ecosystem modification (Mohamed et al. 2012b; Li et al. 2015; Thenepalli et al. 2017).

According to the Council of Agriculture, an estimated 160,000 metric tons of OSs are generated annually as marine industrial waste in Taiwan, placing a significant burden on the fishing community in waste management and disposal (Hwang & Weng 2017). Many efforts toward sustainable recycling of OS waste into value-added products have been made (Hellen et al. 2019; Jovic et al. 2019; Andrade et al. 2020; Bonnard et al. 2020). With the main components being aragonite and calcite (Chilakala Ramakrishna & Whan 2018) (CaCO₃ > 95 wt.%), OSs can be a renewable natural calcium resource for well-known applications, especially as an alkali activator in the production of unfired fly ash bricks (Li et al. 2015; Mo et al. 2018; Thomas et al. 2020) or as filler in composites (Lee et al. 2014; AlBadr et al. 2020), bio-concrete production (Hong et al. 2021), agricultural supplements (Lee et al. 2008; Mako et al. 2017; Hellen et al. 2019), pollutant remediation (Chiou et al. 2014; Xu et al. 2019; Khirul et al. 2020; Xu et al. 2021; Zheng et al. 2022), soil quality improvement (Ok et al. 2011; Torres-Quiroz et al. 2021; Wu et al. 2022), and high-tech polymer synthesis (Li et al. 2017). When calcined at 550 °C and above, the calcite structure of OSs transforms into calcium oxide (CaO) (Kwon et al. 2004; Alidoust et al. 2015; Huh et al. 2016), a promising versatile raw material. OS-derived CaO powder can be utilized in medical preparations to prevent osteoporosis or treat diseases related to calcium metabolism (Bonnard et al. 2020), and in industries as a catalyst, material synthesis (Rujitanapanich et al. 2014; Khan et al. 2018), or a standard reagent (Ferraz et al. 2019). Calcined OS is also a potential antibacterial agent with strong antimicrobial properties, diverse bactericidal mechanisms, and biocompatibility (Sadeghi et al. 2019).

Taiwan receives more than 2.6 times the global average rainfall (Hsu & Hung 2019). However, water scarcity on the island has reached alarming levels in recent years due to topographical factors and the rapid expansion of industries. Semiconductor manufacturing and agriculture are the two main economic drivers of the island, both heavily dependent on water. On this account, Taiwan has pushed for strategic measures to promote the sustainable development of water resources, which requires initiatives such as rainwater harvesting in urban and rural settlements. A fundamental problem often encountered with this approach is that the water extracted through rainwater harvesting systems is exposed to contaminants from various sources (Schets et al. 2010; Hamilton et al. 2017).

Harvested rainwater's overall physical, chemical, and microbiological qualities degrade rapidly during storage without proper treatment, resulting in change in odor, color, and taste, making it unsuitable for direct or indirect use. Therefore, it is important to obtain a cost-effective way to inactivate microorganisms in stored rainwater. Bacillus produces spores during their stationary phase of development, which are capable of long dormancy and are also highly resistant to heat, radiation, and toxic chemicals (Sunde et al. 2009). Bacillus spore inactivation can be utilized as a standard for the decontamination of biological agents from drinking water infrastructure due to their disinfection resistance (Szabo et al. 2017). A simple yet effective solution being explored is processed seashells, a promising alternative material for low-cost water treatment and disinfection. The antibacterial activity of calcined seashells is related to CaO by two mechanisms in which the primary mechanism is the alkalinity of the medium and the secondary mechanism is the generation of Ca²⁺ and reactive oxygen species (ROS) on the surface of CaO, leading to cell wall rupture. Unlike other shells, OSs are explored less for this application. Therefore, for the practical application of heated oyster shell particles (HOSP) as a disinfectant, it is necessary to understand the effect of different processing and activation parameters contributing to its sterilization potency.

This study investigated the sporicidal activity of HOSP against Bacillus subtilis spores. The relative effects of four operating parameters – (i) OS particle heating temperature, (ii) OS particle heating duration, (iii) HOSP concentration, and (iv) treatment time – on disinfection efficacy were assessed by employing response surface methodology (RSM) based on a five-level, four-factorial central composite design (CCD). Besides, a kinetic analysis of the bactericidal action of HOSP was also determined.

Nutrient medium for the growth/cultivation of B. subtilis was prepared with 5 g of peptone and 3 g of beef extract for a 1 L solution. The pH of the medium was adjusted to 7.0 with phosphate buffer. For agar medium, bacteriological agar was added at 2% to the aforementioned preparation.

B. subtilis (ATCC 6051) endospores were cultured based on a modified protocol described by Morris (2012). Harvested endospore stock was serially diluted using 0.001 M phosphate-buffered saline solution (PBS) and spread on nutrient agar plates to check their quality and quantity. An average of 9.2 × 10⁸CFU/mL was observed in the harvested stock.

Fresh rainwater was harvested locally from the rooftop, funneled, and collected directly into sterile glass containers. Harvested rainwater was then pasteurized at 63 °C for 30 min before storing aseptically for further use. The characteristics of pasteurized rainwater are summarized in Table 1.

To evaluate the influence of process parameters on the antibacterial activity of HOSP, a combination of RSM and CCD developed by Design-Expert^® software version 12.0.7.0 (Stat-Ease, Inc., USA) was adopted. A four-factor, five-level CCD with 30 experimental runs consisting of 16 factorial points, 8 axial points, and 6 replicates at the central point (0, 0, 0, 0) was performed. The inactivation rates of B. subtilis spores were chosen as the dependent variable, whereas the independent variables were OSP heating temperature (X₁), OSP heating duration (X₂), HOSP concentration (X₃), and treatment time (X₄). Ranges and values of the independent variables in Table 2 were estimated based on the relevant scientific literature and our preliminary studies.

This study identified an appropriate kinetic model for the inactivation rate of B. subtilis endospores against HOSP in rainwater. Commonly used deceleration kinetic models, i.e., the modified Chick–Watsons model, Homs model, and Selleck–Collins model, were applied and tested to find a suitable fit.

The Chick–Watson model is the most commonly used in water treatment as a classical kinetic equation. Although the model is widely used to describe the inactivation kinetics of chemicals, it has limitations in practical applications where the sterilization rate does not change with time. In such cases, other models are used to explain deviations from the Chick–Watson first-order kinetics (Azzellino et al. 2011; Idris et al. 2017).

The model that yields the highest coefficient of determination (adjusted R²) was the best fit for the purpose. To determine the inactivation kinetics and estimate the model parameter, nonlinear multivariate regression analysis was performed using R software for Macintosh, version 3.5.3.

The antibacterial effect of HOSP against B. subtilis endospores, expressed as log inactivation, was investigated under various conditions in accordance with the experimental runs indicated by Design-Expert^® software version 12.0.7.0. Table 3 summarizes the results of all 30 runs in the CCD matrix for response surface modeling. The observed inactivation rates for B. subtilis endospores ranged from −0.02 to 4.86 log. The model predictions were consistent with observed log inactivation.

In CCD studies, model adequacy check is an integral part to guarantee that it gives an adequate estimation for real analysis systems (Arslan-Alaton et al. 2009; Ivanescu et al. 2016; Arslan & Kara 2017). Approximating model functions would yield inaccurate or misleading results if the fit were poor (Körbahti 2007; Arslan & Kara 2017). The quality of the proposed model was evaluated by constructing a diagnostic plot between the predicted and actual log inactivation values. As shown in Supplementary Material, Figure S1(a), no significant difference was found since the data points were very close to the diagonal line. In other words, the predicted values from the developed model agreed well with the observed values in the experimental data. The plot of residuals versus predicted responses is indicated in Supplementary Material, Figure S1(b). The distributions of residuals were random without any trends. Furthermore, a one-way analysis of variance (ANOVA) was tested to check the adequacy of the suggested model, and the results are described in Table 4. The model's F-value of 31.62 reveals that the model is statistically fit. A p-value less than 0.05 means that the model parameters are highly significant, whereas values greater than 0.1 indicate insignificant terms. In this case, X₁, X₂, X₃, X₄, X₁X₄ (interaction between X₁ and X₄), X₁² (second-order effect of X₁), X₃² (second-order effect of X₃), and X₄² (second-order effect of X₄) are important model terms, and X₂² (second-order effect of X₂) is negligible. The large R² of 0.9343 shows a high degree of consistency between the observed and calculated results. The quality of fit estimates from the coefficients R² represents a reasonable agreement between the predicted R² and the adjusted R² (the difference is less than 0.2), implying that the model provides an adequate fit for the data. A signal-to-noise ratio of 20.7114 provides an adequate signal, demonstrating that the chosen model can be used to navigate the design space. Consequently, the proposed reduced quadratic model represents the best approach for the relationship between variables and responses and is sufficient to assess the effects of variables.

Once high dependability was established, predictions about the antibacterial efficacy of HOSP disinfectant could be obtained through the reduced second-order polynomial equations with coded and actual variables as Equations (6) and (7), respectively.

The regression model provides the preferred effects of the selected variables as X₄ (treatment time) > X₃ (HOSP concentration) > X₁ (OSP heating temperature) > X₂ (OSP heating duration).

Figure 2 shows the effect of HOSP dosage and treatment time on the inactivation rate at constant shell heating time and temperature of 30 min and 800 °C, respectively. The inactivation rate was predicted to increase steadily with both parameters. Specifically, the bactericidal effects of HOSP rapidly increased from 0.677 log₁₀ to 4.080 log₁₀ with an increase in HOSP dosage from 0.1 to 1.7 mg/mL at 144 h treatment time. The more HOSPs used, the more CaO is generated, resulting in more ROS generation. This assists in a higher inactivation rate of B. subtilis endospores. A similar trend was also observed with the increasing treatment time in the system from 0 to 288 h. The inactivation rate increased from 0.132 log₁₀ to 3.905 log₁₀ at a HOSP concentration of 0.9 mg/mL.

In rainwater medium, the coefficient of determination calculated for the chosen models are presented in Table 5. Here, Hom's empirical model is defined to better describe the experimental data based on R² best fit.

The relative effects between OS heat activation and treatment conditions on the sporicidal efficacy were determined using a central composited design of the response surface model. The model indicated that the predicted and experimental values were not significantly different. The ANOVA showed that the effects of OS heating temperature, HOSP dosage in the treatment medium, and the treatment time contributed significantly to the sporicidal effects (p < 0.05). Polynomial models fitting the predicted response, i.e., log inactivation of the test organism (B. subtilis), suggested that the yield of active compound (CaO) in calcined OSs and the resulting microbicidal effect increased with the increase in OS heating temperature and HOSP concentration and contact time in rainwater medium. OS heating duration did not significantly increase the active component yield at a given heating temperature. In addition, the inactivation kinetics for HOSP in aqueous media under stagnant storage conditions are in good agreement with Hom's model. The results from this study may be applied to rainwater harvesting tank construction. Incorporating HOSP in rainwater harvesting tank materials, such as submerged bricks or tank inner lining, may prevent water from fouling during long-term storage. Despite their simplicity and low cost, CaO-based disinfectants have some limitations due to alkaline pH and Ca²⁺ residues in treated rainwater, which presents a major challenge.

This study was supported by the National Science and Technology, Taiwan (Project 109-2221-E-002-079-MY3) and National Taiwan University Core Consortiums (NTUCCP-111L895102 and 112L893902) within the framework of the Higher Education SPROUT Project by the Ministry of Education (MOE) in Taiwan.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.