Biovalorization of raw agro-industrial waste through a bioprocess development platform for boosting alkaline phosphatase production by Lysinibacillus sp. strain APSO

Isolation and identification of ALP-producing bacteria

In the preliminary exploration, the growth of the bacterial microflora of a slime sample from Alexandria paper and pulp factory was enriched in a modified Pikovskaya (PVK) broth¹¹ medium supplemented with eggshell powder as a phosphate source. The pure isolates were screened to produce extracellular ALP using chromogenic agar medium, such as methyl green (MG)-phenolphthalein diphosphate (PDP) agar and rich medium supplemented with ρ-nitrophenyl phosphate (ρNPP). The MG-PDP method has a unique advantage over other procedures, through which the contamination opportunity is avoided and the distinction between excreted and cellular phosphatase was more obvious. These media had previously been used to pick out the most potent ALP-producing bacteria, as reported by Patel Falguni et al.¹². Approximately 30 isolates showed notable ALP activity on MG-PDP agar. Based on the colony color intensity, 4 of 30 isolates exhibited a high capability of ALP productivity. Among these, the colony of the APSO isolate manifested the highest obviousness of green and yellow color intensity, as shown in Fig. 2a, and showed a maximal ALP activity of 145 U L⁻¹ min⁻¹ compared to the other three isolates that showed enzyme activities of 76.99, 35.93, and 14.6 U L⁻¹ min⁻¹ according to the quantitative screening program. For this reason, the APSO isolate was picked out for a further detailed investigation. The representative isolate that displayed the greatest activity was identified using molecular identification. Sanger sequencing of the dideoxynucleotide of the amplified 16S rRNA gene (S.1.) has resulted in a 1211 bp nucleotide sequence. Analysis based on genome-derived 16S rRNA gene sequences in this study showed that the APSO isolate has a homology of > 99.92% relative to the genus Lysinibacillus. The 16S rRNA gene sequences have been submitted to the national center for biotechnology information (NCBI) GenBank with accession number MT975241. The 16S rRNA gene sequences of the reference taxa were retrieved from the NCBI GenBank and aligned using MEGA 7.0 for dendrogram construction. The resultant tree topologies were evaluated by bootstrap analysis based on 500 replicates. The results of this analysis (Fig. 2b) indicated that APSO formed a coherent branch within the genus Lysinibacillus, with a bootstrap value of 100%. It also formed a robust cluster with strains of the genus Arenimonas and shared the highest sequence similarity with Lysinibacillus sp. strain KDP-SUK-M5 (99.92%) and a query cover of 100%, so it could be identified as Lysinibacillus sp. strain APSO. Lysinibacillus sp. strain APSO was characterized by a rod shape (0.72 wide × 3.45 μm long), Gram-positive staining, and spherical endospore, as shown in Fig. 2c,d. Several earlier studies reported that the Bacillus genus is one of the most authoritative ALP producers^4,13.

Effects of physical parameters on ALP productivity

One of the factors that control ALP production processes is the physical parameter. Temperature is one of the most critical environmental parameters for controlling the physiological and biological activities of microorganisms; hence, it influences all bioprocess operations¹⁴. To evaluate the optimal temperature for enhancing ALP productivity, the bacterial culture was incubated at different temperatures (37 °C, 40 °C, and 45 °C) using a modified PVK broth medium¹¹ with eggshell powder. The results elucidated that ALP productivity gradually increased from 37 °C (29 U L⁻¹) passing by 40 °C (41.5 U L⁻¹) until it reached the optimal activity peak at 45 °C with 209.61 U L⁻¹. This finding was congruent with earlier observations of Behera et al.^13,15, who recorded that the maximal phosphatase production by Serratia and Alcaligenes faecalis was accomplished at 45 °C. Recently, Abdelgalil et al.¹⁶ reported that the peak of ALP production from Bacillus paralicheniformis strainAPSO was achieved at 45 °C.

ALP production depends greatly on the initial pH of the culture medium, as it can influence many enzymatic processes and nutrient transport across the cytoplasmic membrane. This study revealed that the optimal ALP production was recorded at pH 6.0 and 7.0 (149.4 and 231.6 U L⁻¹), and the gradual decline in enzyme productivity (147.8, 106.3, and 63.2 U L⁻¹) was obtained at pH 8.0, 9.0, and 10.0, respectively. The results were in accordance with Jena et al.¹⁷ and Dhaked et al.¹⁸, who found that neutral pH (7.0) is the most suited for high ALP production by Bacillus sp. In contrast, Abdelgalil et al.¹⁶ revealed that B. paralicheniformis strain APSO exhibited maximal ALP production at pH 9.0, incompatible with this finding.

To optimize the inoculum size for improving ALP productivity, bacterial cells were grown on a series of activated precultured inoculum sizes (1%, 2%, 5%, and 10%). The results elucidated that a gradual decline in enzyme production was observed as the inoculum size increased. Maximum ALP productivity was obtained using 1% activated precultured inoculum size (343.4 U L⁻¹), whereas a gradual decrease in productivity was noted as the inoculum size increased by twofold (166.2 U L⁻¹), 2.5-fold (137.1 U L⁻¹), and 3.5-fold (94.1 U L⁻¹) at 2%, 5%, and 10% inoculum sizes. This finding matched Qureshi et al.¹⁹, who utilized 1% inoculum size for ALP production by Escherichia coli EFRL 13 in submerged fermentation. In contrast, the optimal inoculum size for the highest ALP throughout was recorded at 5% of the preactivated culture of B. paralicheniformis strain APSO¹⁶.

Statistical optimization of ALP production by Lysinibacillus sp. strain APSO

Medium optimization has reached new horizons with the development of statistical technologies, as these strategies increase the productivity of the process, minimize production time and labor expense, and contribute to improving the overall economy of the process. To optimize the ALP production process, the following experimental design strategies are highly recommended: First, Plackett–Burman design (PBD) was employed to evaluate the relative significance of cultivation variables of the respective ALP production system (Screening). Table 1 shows the randomized PBD matrix selected to screen the most significant variables for ALP production along with the corresponding experimental and predicted response (ALP productivity). Table 1 shows a noticeably broad variation in ALP productivity throughout different trials ranging from 14.8 to 959.4 U L⁻¹, reflecting the significance of medium optimization for improving ALP productivity. The highest ALP activity was achieved at trial 17 followed by trials 16 and 7 that contained molasses and MgCl₂·6H₂O at the highest level. This inferred that molasses and MgCl₂·6H₂O are the most significant variables for ALP stimulation by Lysinibacillus sp. strain APSO. In contrast, trial 5 that contained the lowest level of molasses and MgCl₂·6H₂O showed unremarkable activity (14.8 U L⁻¹), demonstrating the significance of molasses and MgCl₂·6H₂O on the production process. Analysis of variance (ANOVA) results provided the weight of each independent variable for the investigated response, as shown in Table 2.

The probability (p) value was utilized to conclude the main nutrimental factors at a confidence level of > 95%, as summarized graphically in Fig. 3. The lower probability value (p < 0.05) and the larger t-value indicated that the process variable significantly affects ALP yield, and p-values > 0.05 indicate that the model terms are nonsignificant. Based on ANOVA, the most significant factors that positively influence ALP production were molasses (X₁) with a contribution ratio of 99.99 and p = 0.00000063, followed by MgCl₂·6H₂O (X₆) with a contribution ratio of 96.383 and p = 0.036. The third significant variable in the production process was NaNO₃ (X₂) with a contribution ratio of 82.5 and p = 0.174 (Table 2). The insignificant factors with a low contribution ratio and p > 0.05, which negatively influenced the production process, including eggshell powder, CuSO₄·5H₂O, (NH₄)₃SO₄, NaCl, CoCl₂·6H₂O, and ZnSO₄·H₂O, were neglected in a further study. A standardized Pareto chart (Fig. 3b) was employed to determine the influence of the most important parameters. It was constructed by bars with a length proportional to the absolute value of the estimated effects divided by the standard error. The bars are presented in order of the effect size, with the largest effects on top.

In this study, ANOVA of PBD demonstrated that the model was highly significant, as evident from a Fisher’s F-test value of 12.91 with a very low probability value (p = 0.0000399; Table 2). This implied that there was a statistically significant relationship between the variables at the 99.999999% confidence level. Statistical R² showed that the fitted model explained 90% of ALP activity variation. Furthermore, the highest value of the adjusted R² statistic (82%; Table 2) indicated that the model was highly significant, and the correlation between experimental variables and ALP productivity was more precise and reliable. The first-order polynomial model describing the correlation between the nine factors and ALP productivity could be expressed as follows:

The important graphical method to detect and explain whether the obtained data set is normal or deviates from normality is the normal probability plot²⁰. In this study, residuals were plotted against the expected normal values of the model, as illustrated in Fig. 3c. The normal probability chart clearly showed that residuals were distributed normally; the points were distributed nearly to the diagonal line. This indicated that the expected ALP productivity was well fitted with experimental results. To assess the accuracy of PBD, a confirmation experiment was performed. The culture conditions predicted to be optimum for maximal ALP production from Lysinibacillus sp. APSO were pH 6.45, temperature 45 °C, inoculum size 1% v v⁻¹, agitation speed 200 rpm, incubation time 48 h, and medium of the following composition: molasses, 10 g L⁻¹; NaNO₃, 0.5 g L⁻¹; and MgCl₂·6H₂O, 0.2 g L⁻¹. The maximal ALP activity obtained under these optimal conditions was 3208.333 U L⁻¹, higher than ALP activity obtained before application of PBD (343 U L⁻¹) by ~ 9.3-fold.

There are very few reports on the optimization of bacterial ALP production using statistical and experimental PBD. Pandey et al.²¹ reported the optimization of a physical parameter only for enhancing ALP production by B. licheniformis. Glucose, yeast extract, and agitation were reported to enhance maximal ALP production from Geobacillus thermodenitrificans I2 isolate, whereas NaCl, ZnSO_4, and CuSO₄ contributed negatively; this finding matched that in this study²². The other published articles depended on the one-factor-at-a-time (OFAT) method to optimize bacterial ALP production.

The most published articles used a synthetic medium and applied the OFAT method for bacterial ALP production. In addition, the use of an agro-industrial waste-based medium for ALP production has been poorly investigated. To the authors’ knowledge, Abdelgalil et al.¹⁶ were the first investigators to utilize an agro-industrial waste to boost ALP production from B. paralicheniformis strain APSO. They demonstrated that the most significant variables that directly enhanced ALP production by PBD are molasses, (NH₄)₂SO₄, and KCl; this finding was consistent with this study. Adopting this trend, this study directed to the ultimate beneficial utilization of agro-industrial and food waste and low-cost nutrimental variables for enhancing bacterial ALP output on laboratory-scale production to reduce the cost of the production process through complicated statistical optimization strategies.

Compared to the results obtained by other investigators, Qureshi et al.¹⁹ reported that 2% sodium nitrate had the strongest influence on ALP from E. coli EFRL 13 among other variables used (tryptone, ammonium chloride, and potassium nitrate). Furthermore, molasses (15 g L⁻¹) and sodium nitrate were utilized for improving ALP production from rhizospheric bacterial isolates²³. In contrast to this study, Parhamfar et al.¹¹ documented that maximal ALP production from Bacillus spp. was achieved by glucose (10 g L⁻¹) and ammonium sulfate (5 g L⁻¹) as the carbon and nitrogen source. Additionally, Jatoth et al.⁴ found that starch (5 g L⁻¹) and egg albumin (5 g L⁻¹) were suited and acted as enhancers of ALP production from B. subtilis.

Response surface methodology (RSM)

Based on the fit of empirical models to the experimental data in relation to the experimental design, collections of mathematical and statistical approaches expressed as RSM are utilized. To maintain high-efficiency and profitable bioprocesses, microbial industry sponsors are recommended to utilize RSM approaches during production²⁴. To determine the optimal response region of ALP productivity, the most significant independent variables that positively affect ALP production as identified by PBD design results (molasses, NaNO₃, and MgCl₂·6H₂O with contribution percentages of 99.99%, 82.52%, and 96.38%, respectively) were explored more via the application of central composite design (CCD) with uniform precision by five levels for each parameter. Consequently, the intercorrelations among independent variables and their optimal levels were evaluated via regression analysis of this approach.

Table 3 illustrates the array of the design matrix for a total number of 20 experiential exploratories of CCD-uniform precision with different combinations of the three independent factors along with experimental and theoretically predicted results of ALP production with residuals. In contrast, the other variable according to the PBD results exerted a negative effect on the ALP production process; hence, these insignificant variables were omitted from further optimization by uniform precision-CCD to improve enzyme production and diminish the final cost of the production process. The six axial points, eight factorial points, and six center points formed a precision random array for CCD in 20 empirical probationary for optimizing the picked variables. To assess the experimental errors, the six center point replicates were conducted. The findings indicated considerable variation in the potency of ALP activity depending on the three independent factor variations. ALP activities ranged from 935 to 5316.66 U L⁻¹ based on the observed data. In Table 3, center point experimental trials (numbers 1, 4, 7–9, and 14) that comprise zero-level molasses, NaNO₃, and MgCl₂·6H₂O concentrations (30, 0.6, and 0.45 g L⁻¹, respectively) were recorded to have the highest ALP yield with an approximate activity of 5306.47 to 5316.66 U L⁻¹. In contrast, the lowest ALP titer was achieved at the highest molasses (50 g L⁻¹) concentration and middle NaNO₃ (0.6 g L⁻¹) and MgCl₂·6H₂O (0.45 g L⁻¹) concentrations with an ALP activity of 935 U L⁻¹. This resulted from the suppression effect of molasses, as reported by Qureshi et al.¹⁹. The theoretically forecasted ALP productivity was in line with the experimentally achieved ALP activity.

Multiple regression analysis and ANOVA

Arithmetic operators of multiple regression statistical analysis and ANOVA computations, a critical approach to evaluating the significance and adequacy of the quadratic regression model, were employed to analyze and interpret the CCD-uniform precision experimental data results as enumerated in Table 4. In Table 4, the determination coefficient R²-value of this regression model was 0.964, indicating that 96.40% of the disparity in ALP productivity was attributed to the independent variables and only 3.60% of the total disparity created by variables could not be explained by the model and could not explain ALP activity. This regression model is trustworthy to reflect reality correlations between independent nutrimental variables that influence ALP throughput.

Additionally, the model’s precision to calculate the predicted ALP output and a significant correlation between the observed and fitted values for ALP production was confirmed by the coefficient of multiple correlations (multiple R) of 0.982 and the adjusted determination coefficient value (adjusted R²) of 0.933 and predicted R²-value of 0.709. Furthermore, the model’s significance, fitness, and reliability determined by the low percentage of the coefficient of variation value (16.89%) and the root mean square error and mean values were 477 and 2825, respectively. Adequate precision measured the signal-to-noise ratio; a precision value of 14.008 indicated an adequate signal, and this model could be used to navigate the design space. ANOVA revealed the chosen model to be fabulously relevant, as obvious from the low probability p of 0.00000415 and the Fisher’s F-test of 30.4660, computed as the ratio of mean square regression and mean square residual. The higher the F-value is, the higher is the connotation of the model. p-values, F-values, and t-test were utilized for data interpretation, verification of the significance for each variable, and recognition of the mutual interaction dynamics between variables under investigation as illustrated in the fit summary results (Table 4). Furthermore, the interrelation between two variables can be either positive or negative to the response depending on the signs of the variable coefficients.

A negative sign implies an antagonistic effect, but the synergistic or complementary effect of two variables on the response is demonstrated by a positive sign. Accordingly, the antagonistic or oppositional effect (negative coefficients) was apparent in the mutual interaction between molasses (X₁) and sodium nitrate (X₂) together with the mutual interaction between sodium nitrate (X₂) and MgCl₂·6H₂O (X₃), with F-values (7.42 and 3.44, respectively), p-values (0.0214 and 0.0932, respectively), and t-values (− 2.7250 and − 1.8555, respectively) with confidence levels (97.8% and 90.6%, respectively). Based on the negative sign of a coefficient, these mutual interaction effects are nonsignificant model terms and do not improve the response. In contrast, mutual interaction between molasses (X₁) and MgCl₂·6H₂O (X₃) had a synergistic and significant effect on ALP productivity, obviously from the F-value of 2.09 and t-test value of 1.4449. Although its p-value and contribution percentage were 0.179 and 82%, respectively, it was a positive coefficient sign, contributing to the elevation and enhancement of ALP production. Another point worth mentioning is that the linear coefficients of sodium nitrate (X₂) had highly significant effects on ALP productivity, as manifested from p-value (0.0182), F-value (7.94), and t-value (2.817) with confidence level (98.17%). Although the linear coefficients of molasses (X₁) and MgCl₂·6H₂O (X₃) together with the quadratic effects of molasses (X₁²), NaNO₃ (X₂²), and MgCl₂·6H₂O (X₃²) were characteristic of the negative sign of coefficients, they had not significantly contributed to the enhancement of ALP production by the strain under study. Arithmetic indications for their significance degrees were computed as follows: p-value (0.02, 0.022, 0.00000039, 0.0000046, and 0.0000079, respectively), F-value (7.63, 7.2, 134.7, 79, and 69.9, respectively), and t-test (− 2.76, − 2.68, − 11.6, − 8.89, and − 8.36, respectively) with confidence level (97.99%, 97.7%, 99.99%, 99.99%, and 99.99%, respectively). The second-order polynomial function fitted to the experimental results of ALP productivity was employed for predicting the optimal point within experimental constraints as in the following equation:

where Y is the predicted response (ALP activity), and X₁, X₂, and X₃ are the coded levels of the independent variables of molasses, NaNO₃, and MgCl₂·6H₂O, respectively.

Model adequacy checking

Some empirical statistics were conducted to ensure that the model was appropriate to meet the assumptions of the analysis. To illustrate whether the residuals follow a normal distribution or deviated from it, the distribution residuals were drawn against a theoretical normal distribution through the normal probability plot. As shown in the normal probability plot of the studentized residuals in Fig. 4a, the plotted points are located nearly to a straight line without linearity, indicating that the model has been well fitted with the experimental results. In contrast, most of the findings clustered in the center and the values distant from the general mean dwindled on both sides of the center peak equally. Extreme residual values are not preferred on both sides. Additionally, the residual versus predicted response chart was constructed and is depicted in Fig. 4b, through which randomly scattered residuals around the horizontal zero-line reference were observed and no particular patterns could be identified. In essence, this implied that the residuals are not correlated and distributed independently and thus have constant variance, meaning that the model is adequately compliant with the study’s postulations. This indicated a good fit of the model for ALP production by Lysinibacillus sp. APSO. Moreover, plotting the predicted values against the actual data sets (Fig. 4c) revealed that the data sets were divided evenly down the 45° parallel. This could help detect if the model cannot smoothly predict some value(s); however, all data points are measurable, guaranteeing model aptness. Figure 4d clarifies the Box-Cox graph. The best λ-value (− 0.39) was represented by the green line, and this transformation (λ = 1) was denoted by the blue line, whereas the red lines symbolized the lowest and highest values of confidence intervals between − 1.09 and 1.7. Because the λ best value was between the values of confidence intervals, no recommendation for data transformation was required due to the blue line between the two red lines. This revealed that the model was well appropriate to the obtained experimental data.

Contour and three-dimensional (3D) plots

After the preceding assurance of model appropriateness, the interactive effects of the three measured variables were defined. Also, the optimal values of each independent variable needed for maximal ALP production were calculated through the construction and visualization of 3D surface and outline contour chart by plotting ALP activity on the Z-axis against two of the variables while keeping the third variable at the center point, as shown in Fig. 5.

The 3D surface and contour plot of Fig. 5a shows ALP productivity efficacy as a function of molasses (X₁) and NaNO₃ (X₂) concentrations with maintained MgCl₂·6H₂O concentration at zero levels (0.45 g L⁻¹). The highest point for ALP activity was situated around the center points of molasses, out of this range; rather, low ALP productivity was noticed. Further, ALP activity was enhanced with increments of NaNO₃ concentrations until reaching its optimum, but the higher level supporting low ALP activity was remarked. The maximal predicted pinpoint of ALP output was estimated through the software program of the point prediction option. The maximal predicted ALP activity of 5363.26 U L⁻¹ was attained at the optimal predicted molasses (28.23 g L⁻¹) and NaNO₃ (0.645 g L⁻¹) concentrations at a MgCl₂·6H₂O concentration of 0.45 g L⁻¹. The same trend of ALP productiveness was noticed for the pairwise combination of molasses (X₁) and MgCl₂·6H₂O (X₃) with zero levels of NaNO₃ (X₂) 0.6 g L⁻¹. In Fig. 5b, the highest ALP throughput was recorded at a low MgCl₂·6H₂O (0.4 g L⁻¹) concentration near the middle point; the molasses concentration increased, resulting in a declining trend of ALP productivity. Beyond the optimal center point of X₁ and X₃, their lower and higher levels caused a relative increase in ALP yield. Therefore, by solving Eq. (3) and analyzing Fig. 5b, using a zero-level NaNO₃ (0.6 g L⁻¹) concentration at the predicted optimal molasses (28.23 g L⁻¹) and MgCl₂·6H₂O (0.41 g L⁻¹) concentrations contributed to the utmost ALP predicted productivity of 5354.8 U L⁻¹. The 3D plot and the corresponding contour plot (Fig. 5c) highlighted the effects of NaNO₃ (X₂) and MgCl₂·6H₂O (X₃) concentrations on ALP production at 30 g L⁻¹ of zero-level molasses (X₁) concentration. ALP activity increased progressively at a low NaNO₃ concentration, accompanied by increasing MgCl₂·6H₂O concentrations to its optimum; then, ALP production declined. The greatest ALP throughput was clearly supported at the center level of MgCl₂·6H₂O (X₃), passing the optimal points of its concentration; whether an increase or decrease, it will lead to a reduction in productivity. By solving Eq. (3) and analyzing Fig. 5c, the maximal predicted ALP activity of 5368.31 U L⁻¹ was attained at the optimal predicted NaNO₃ (0.64 g L⁻¹) and MgCl₂·6H₂O (0.41 g L⁻¹) concentrations at a molasses concentration of 30 g L⁻¹.

Verification of the experimental model

To estimate the optimal combinations of the variables under investigation, which ameliorate ALP productivity, response optimization was carried out to augment the efficacy of Lysinibacillus sp. APSO ALP. The optimal predicted molasses (28.23 g L⁻¹), NaNO₃ (0.64 g L⁻¹), and MgCl₂·6H₂O (0.41 g L⁻¹) concentrations for boosting the efficacy of ALP production in the fermentation medium to 5413.91 U L⁻¹ were predictable and theoretically based on the equation of the quadratic model. Laboratory validation via bench-scale experiments was utilized to confirm the theoretical data results of the optimization process, revealing that the practical ALP activity of 5683 U L⁻¹ was near that obtained theoretically by the regression equation of the model (5413.91 U L⁻¹) with 104.97% accuracy, confirming the model’s authenticity. Hence, the ultimate culture conditions and medium composition were molasses, 28.23 g L⁻¹; NaNO₃, 0.64 g L⁻¹; and MgCl₂·6H₂O, 0.41 g L⁻¹, and the pH was adjusted to 7.0 by 1 M NaOH with 1% old activated preculture inoculum. The inoculated medium was incubated at 45 °C and 200 rpm for 24 h.

Compared to the results cited by other investigators, very few investigators dealt with statistical optimization strategies for improving ALP productivity. Pandey et al.^21,25 were the only researchers who exploited CCD to enhance the production process of ALP from B. licheniformis. In 2010²¹, they documented the optimization strategies of physical variables that influence ALP productivity. Their study revealed that the maximal ALP yield of 792.043 U mL⁻¹ was achieved at pH 8.0, temperature of 36.7 °C, fermentation time of 78 h, and orbital speed of 165 rpm. In 2012²⁵, the same team recorded that the predicted glucose (2.39%), peptone (1.35%), yeast extract (0.15%) concentrations were obtained from the theoretical calculation of the regression model of CCD for enhancing the B. licheniformis ALP output. They documented that the correlation coefficient (R²) was found to be 0.932 (93.2%). Recently, through the regression model of rotatable-CCD by Abdelgalil et al.¹⁶, the highest predicted ALP productivity (3753.27 U L⁻¹) produced from B. paralicheniformis strain APSO could be attained using molasses, (NH₄)₂NO₃, and KCl concentrations of 30, 0.79, and 1.2 g L⁻¹, respectively.

Scale-up fermentation strategies for Lysinibacillus sp. APSO ALP

The improvement of ALP productivity is needed, and the biomass necessary for sufficient yield is economically limited unless the phosphatase-specific activity is augmented. Therefore, the opportunity for optimizing growth conditions, which give the highest specific phosphatase activity, was taken²⁶.

Cell growth kinetics and ALP production in shake-flask under batch conditions

Before moving forward to bioreactor cultivations, this study intended to determine the correlations between the specific growth rate of the culture and the ALP production rate via a cell growth kinetics study in a shake-flask cultivation system, which could likely be referred to for the bioreactor strategy for batch cultivation. Table 5 illustrates the cell growth kinetics and ALP production parameters by Lysinibacillus sp. strain APSO affected by different cultivation strategies. Batch growth kinetics of Lysinibacillus sp. strain APSO followed a typical growth pattern, and ALP was produced simultaneously with cell growth, as shown in Fig. 6a. This finding was analogous to Butler et al.²⁷ and Abdelgalil et al.¹⁶, who found that phosphatase production increased during exponential growth of a Citrobacter sp. and B. paralicheniformis strain APSO, respectively.

Initial lag-phase cells experienced exponential growth with a growth rate of 0.044 (g L⁻¹ h⁻¹) and a specific growth rate (µ) of 0.115 h⁻¹. The culmination of biomass production (0.906 g L⁻¹) was achieved at 28 h cultivation period with an obvious yield coefficient Y_x/s (0.07 g of cells/g of the substrate), surpassing this point, driving cell growth to stationary phase. The volumetric ALP production and protein content were concurrent with the cell growth pattern. Both increased gradually until reaching their zenith (6365.74 U L⁻¹ and 4.92 g L⁻¹, respectively) with production rate Q_p (588.703 U L⁻¹ h⁻¹), specific productivity P_max,specific (1292 U g⁻¹), and yield coefficient Y_p/s (965.85 U g⁻¹) at 20 h incubation time. A gradual decline in ALP activity and protein content was observed beyond this time. Meanwhile, due to cell growth, proliferation, and the essential role of ALP in phosphate transportation and metabolism²⁵, a notable decrease in total carbohydrate and total phosphatase concentrations was found from their initial concentrations of 14.71 and 0.027 g L⁻¹ to reach 5.03 and 0.0003 g L⁻¹, with a consumption rate of − 0.26068 and 0.0004169 g L⁻¹ h⁻¹, respectively. From the pH profile throughout the fermentation process (Fig. 6a), at the first 10 h incubation period, considered a lag-phase period, a gradual elevation in the initial pH of the culture medium from 6.4 to 7.6 occurred but decreased to 7.0 at 18 h. After that time, the pH increased again with up-down fluctuations to reach 8.0 at 30 h cultivation time. This pH change was accompanied by a gradual increase of bacterial growth, ALP productivity, and protein content to reach their own climax. This resulted from the influence of pH on bacterial growth or pH-dependent control of the gene expression for enzyme synthesis¹⁹.

There was an obvious need for metal ions to promote ALP productivity. Chaudhuri et al.² reported that the regulatory control was mainly carried out by metal ion chelation necessary for enzyme production or stabilization. As noticed from the heavy metal survey throughout the cultivation period via atomic absorption spectrometry (AAS) analysis, the concentration of some metal ions, such as Cu⁺, Pb⁺, Fe⁺, Zn⁺, Mg⁺, Sn⁺, and Ni⁺, was synchronous with ALP productivity. In contrast, the apex of their concentrations (4.04, 20.36, 51.1, 5.835, 1530, 287.9, and 5.25 mg L⁻¹) was achieved when the peak of ALP production was obtained, which was 20 h incubation time. After transcending this point, their concentrations decreased until the end of the fermentation period, as shown in Table 6. Li⁺ reached its peak (1.27 mg L⁻¹) at the end of cultivation time after up-down fluctuations throughout the fermentation period, there was no change observed in Co⁺, Cr⁺, and Mn⁺ concentrations throughout the fermentation period, and Cd²⁺ ions were consumed early to < 0.0335 mg L⁻¹ during the growth course. This finding was in agreement with that recorded by Abdelgalil et al.¹⁶, who noticed that the growth of B. paralicheniformis strain APSO was associated with an increment in Sn⁺, Zn⁺, Mg⁺, and Mn⁺ ion concentrations, reaching their peak of 283, 22, 652.5, and 4.27 mg L⁻¹ at 26 h, at the peak of ALP output. There was no change in Co⁺ and Cr⁺ concentrations throughout the fermentation period.

Cell growth kinetics and ALP production in the bioreactor under uncontrolled pH batch conditions

The scale-up strategy is required to reproduce the culture characteristics and kinetic parameters from shake-flasks to fermenters. Therefore, the cultivation kinetics of Lysinibacillus sp. strain APSO was monitored in a 7 L stirred-tank bench-top bioreactor (Bioflo 310; New Brunswick Scientific, Edison, NJ, USA) under uncontrolled pH for further development and optimization. Knowledge of the fermentation process kinetics may be valuable for improving the performance of batch processes. Finally, product yield and substrate conversion are the most significant criteria toward productivity²⁸. The results showed great similarity in the growth and ALP production pattern to those achieved in the shake-flask counterpart (Fig. 6b).

A significant improvement was seen in cell growth and enzyme production under uncontrolled pH batch cultivation conditions. The peak of cell biomass production (1.0322 g L⁻¹) was achieved at 18 h incubation time, earlier than those in shake-flasks by ~ 10 h, whereas the stationary phase started beyond this point at 20 h incubation time. Cells growing exponentially with a growth rate of 0.068482 g L⁻¹ h⁻¹ and a specific growth rate (µ) of 0.18875 h⁻¹ showed a yield coefficient Y_x/s of 0.11 g cells/g substrate), considerably higher than those achieved by shake-flask cultivation. Further, ALP production and protein content were simultaneous with the growth pattern. The peak volumetric ALP productivity was accomplished at 15 h cultivation with an activity of 7119.44 U L⁻¹ and a production rate Q_p of 605.55 U L⁻¹ h⁻¹, which were about 111.83% and 102.86%, respectively, higher and earlier than those in shake-flask cultivation by ~ 5 h. Moreover, considerable elevation in ALP yield coefficient Y_p/s of 1123.62 U g⁻¹ was noticed, about 1.16 times higher than the obtained counterpart. It is noteworthy that growth and ALP production patterns were accompanied by substrate consumption patterns. With cell growth and enzyme productivity increased progressively, substrate consumption also increased, leading to a trend of total carbohydrate and total phosphate concentrations decreasing from their initial concentrations (14.71 and 0.027 g L⁻¹, respectively) to reach their minimal concentrations (1.42 and 0.00038 g L⁻¹, respectively) at the end of the cultivation period. This was accompanied by an increasing consumption rate of total carbohydrate and total phosphate (− 0.98 and − 0.0017 g L⁻¹ h⁻¹, respectively) by > 3.79- and 4.1-fold, respectively, than those in shake-flask cultivation. Compared to the findings of Abdelgalil et al.¹⁶, ALP throughput of 6650.9 U L⁻¹ and µ of 0.0943 h⁻¹ were achieved at 8 h incubation time of B. paralicheniformis strain APSO under uncontrolled pH batch cultivation conditions in 7 L stirred-tank bioreactor.

Gupta et al.²⁹ reported that changes in oxygen availability might lead to drastic effects on fermentation kinetics. Therefore, in this investigation, dissolved oxygen (DO) availability in the culture medium was monitored throughout the cultivation period. Figure 6c shows that the DO percentage at lag phase times of 2 and 4 h was 100% and 96%, respectively. Once bacterial cells moved toward the exponential phase, the DO percentage declined sharply, reaching 0.7% at the mid-exponential phase (10 h) due to actively growing cells. After this point, it was at a steady state from 12 to 20% until 15 h. This was synchronous with ALP peak production. At the late log-phase and the beginning of the stationary phase, it increased again in obvious increments, reaching 88.9% at the end of the fermentation period. Additionally, the tracking of the culture medium pH showed the same pattern as those in the shake-flask counterpart. A noticeable decline in pH values from 6.45 to 5.96 was noticed at the initial lag-phase period from incubation time. After 6 h, a gradual increase in the culture medium pH was observed, reaching 8.07 at the end of the fermentation period. This matched that of previous shake-flask cultivation.

Based on the above results, a > 111.83% improvement of peak volumetric productivity was achieved, higher than in shake-flask cultivation (6365.74 U L⁻¹ at 20 h). These enhancements in efficiency resulted from good cultivation of oxygenation and involvement conditions and the vessel’s bioreactor capacity. However, maximal volumetric productivity increased by 148.18% from that in shake-flask cultivation (4488.266 U L⁻¹ at 26 h) by B. paralicheniformis strain APSO¹⁵.

Cell growth kinetics and ALP production in the bioreactor under controlled pH batch conditions

Based on previous results, uncontrolled pH batch cultivation showed the best conditions for improving and enhancing volumetric and specific ALP productivity. In this study, Lysinibacillus sp. strain APSO was forced to grow under controlled pH culture conditions (pH 6.45) through a batch cultivation system in a 7 L stirred-tank bench-top bioreactor (Bioflo 310) to evaluate the influence of constant pH value on the growth and productivity characteristic features. Figure 7a shows the relations between cell growth, enzyme production, and substrate consumption as a function of time for batch fermentation of Lysinibacillus sp. strain APSO under controlled pH conditions. In Fig. 7a, the general patterns show an analogy with those in shake-flask cultivation. As pointed out in previous cultivation conditions, the ascending increment of cell growth was noticed at 6 h cultivation period with a growth rate of 0.035143 g L⁻¹ h⁻¹ until reaching its peak (X_max = 0.702 g L⁻¹) at 24 h, about 22.51% and 31.9% lower than that in shake-flask and uncontrolled pH batch cultivation, respectively. Furthermore, the lowest yield coefficient Y_x/s (0.04 g cells/g substrate) was obtained compared to previous culture conditions.

Additionally, volumetric enzyme productivity and protein content patterns were parallel to cell growth patterns, with a lesser production rate Q_p of 387.764 U L⁻¹ h⁻¹ than those in previous batch cultivations by 34.13% and 35.96%, respectively. The targeted productivity peak (5795.37 U L⁻¹) was achieved at 20 h after such a point the production curve started to decrease, about 8.95% and 18.59% lesser than those achieved in shake-flask and uncontrolled pH batch cultivation, respectively. However, a 33.7% to 43.04% decline in yield coefficient Y_p/s (639.93 U g⁻¹) than those in other counterparts was noticed. Also, protein content decreased rapidly after overstepping its peak point (4.93363 g L⁻¹) at 18 h. The correlation of substrate consumption with the growth and enzyme production pattern led to a decline in total carbohydrate and total phosphate concentrations throughout the fermentation period, reaching 5.292 and 0.001 g L⁻¹ at the end of cultivation (26 h) with a consumption rate of − 0.559 and − 0.0008 g L⁻¹ h⁻¹, respectively. The consumption rate indicated the lower substrate consumption than consumed during the uncontrolled pH batch cultivation system. This resulted from a reduction in cell growth and enzyme productivity. Moreover, a slight elevation in DO concentration (100%) was observed during the first 4 h. With cell growth, DO consumption increased, leading to a decline in DO concentration to 19.2% at 16 h. Then, it increased again until the end of the cultivation time to reach 85% (Fig. 7b).

With bacterial cells growing and surviving under controlled pH conditions, unfavorable results were obtained because a steady pH hindered cell growth and adaption. Under these conditions, Lysinibacillus sp. strain APSO lost its ability to produce ALP with efficiency. Therefore, the uncontrolled pH culture condition represents the most suitable and favored condition for supporting and enhancing ALP productivity. This finding was congruent with Abdelgalil et al.¹⁶, who noticed that controlled pH conditions neither supported cell growth nor enhanced ALP productivity.

The sequential optimization strategy led to a systematic improvement of Lysinibacillus sp. strain APSO ALP productivity, which started with PBD (3208.33 U L⁻¹), followed by CCD-uniform precision (5683 U L⁻¹), and ended with uncontrolled pH batch cultivation strategy designs (7119.44 U L⁻¹). Throughout the process, productivity was enhanced by 9.3-, 16.5-, and 20.75-fold, respectively, compared to basal medium (343.44 U L⁻¹).

Due to the scarcity of specific literature, a comparison of results to others is difficult. This study is the first to report the scale-up production of extracellular ALP from Lysinibacillus sp. strain APSO on a bench-top bioreactor scale using sugarcane molasses as a nutrient source for stimulating ALP production.