You are viewing the site in preview mode

Skip to main content
Download PDF

Artificial Neural Network - Multi-Objective Genetic Algorithm based optimization for the enhanced pigment accumulation in Synechocystis sp. PCC 6803

BMC Biotechnology volume 25, Article number: 23 (2025) Cite this article

Abstract

Background

Natural colorants produced by the cyanobacterium include carotenoids, chlorophyll a and phycocyanin. The current study used the Synechocystis sp. PCC 6803 to examine how abiotic stress conditions, such as low temperature as well as high light intensity, affect the pigment accumulations in comparison to the control conditions. Additionally, using the response surface methodology (RSM) and artificial neural network - multi-objective genetic algorithm (ANN-MOGA), the impact of several nitrogen sources such as urea, ammonium chloride, and sodium nitrate as nutritional stress on the pigment accumulations in the Synechocystis sp. PCC 6803 was examined.

Results

The results showed that the pigment accumulation was more pronounced when urea and ammonium chloride was used in combination with nitrate, respectively, as nitrogen source. With the help of our prediction model that used ANN-MOGA, we were able to enhance the synthesis of chlorophyll a, carotenoids, and phycocyanin by 21.93 µg/mL, 9.78 µg/mL, and 0.05 µg/mL, respectively compared to control with 6.37, 3.88 and 0.008 µg/mL. The significant scavenging activity of pigment was showed with 7.66 ± 0.001 values of IC50. Additionally, a very good correlation of coefficient (R2) value 0.99, 0.99 and 0.92 was obtained for APX, CAT and GPX enzyme activity, respectively.

Conclusions

The findings lays the groundwork for future attempts to turn cyanobacteria into a commercially viable source of natural pigments by demonstrating the benefits of using the RSM and machine learning techniques like ANN-MOGA to optimise the production of cyanobacterial pigments. The significant scavenging and antioxidant activities like CAT, GPX and APX were also shown by the pigments of the Synechocystis sp. PCC 6803. Furthermore, these machine learning tools can be used as a model to improve and optimize the yields for other metabolites production.

Graphical abstract

Peer Review reports

Background

Cyanobacteria are prokaryotic photosynthetic organisms which plays a key role in the global CO2 sequestration and generation of several high-value metabolites. These are one of the oldest life forms on earth, with a 3.5-billion-year evolutionary history. Cyanobacteria have unique characteristics, like their rapid growth ability, ubiquitous presence, and efficient nitrogen-fixing potential [1]. Other than plants, the cyanobacteria can also survive in diversified environments, ranging from high-saltwater channels to arid fields, overcoming agricultural challenges using sunlight and atmospheric carbon dioxide [2]. Thus, these unique characteristics inhibit competition for farmland and make the cyanobacteria a sustainable solution for fuel and commodity chemicals production [2, 3]. Cyanobacteria have applications in various industries such as biofertilizers, pigments, proteins, fatty acids (biofuel), carbohydrates and others [4,5,6,7,8]. The high-value products of cyanobacteria are explored for their use in cosmetics, pharmaceuticals, nutraceuticals, chemical industries, and biomedical health applications [1, 9]. Moreover, cyanobacteria has huge application in removal of heavy metals from domestic and industrial wastewater treatment [10, 11].

Pigment production from cyanobacteria is an emerging research area with a potential for abundant applications in various industries, such as, food, cosmetics, nutraceuticals, and pharmaceuticals [2, 12,13,14,15,16]. Cyanobacterial pigments majorly include chlorophyll a, carotenoids, and phycocyanin. Chlorophyll a being the optically active centre and most abundant pigment, plays a vital role in photosynthesis in sunlight absorption, energy transference, and electron transport [17]. On the other hand, phycobiliproteins are water-soluble and coloured proteins involved in light harvesting and energy-transferring photosynthetic apparatus to chlorophyll. They are broadly classified into phycocyanin (PC), allophycocyanin (APC), and phycoerythrin (PE) [18]. These are bioactive compounds with antioxidants and anti-inflammatory, anticancer, and antiviral properties. They are also utilized as analytical reagents to create fluorescent probes that can be customized for immunohistochemistry, flow cytometry, and confocal laser microscopy [5]. Other than chlorophyll a and phycobiliproteins, cyanobacteria accumulate carotenoids. These are natural lipophilic non-enzymatic antioxidant pigments consist of unbranched hydrocarbon chains and are responsible for the red, yellow, and orange colours [4], found in all photoautotrophic organisms as well as in few non-photosynthetic bacteria and yeasts [19, 20]. Carotenoids are enclosed in cellular membranes that help in the assembly and function of photosynthetic apparatus [8]. Moreover, these are also involved in photosynthesis by serving as an accessory light-harvesting pigments providing photoprotection against stress conditions [19, 20].

The cyanobacterial bioproducts such as pigments could be enhanced by optimizing the process variables by employing the heuristic computational mathematical models [21]. Computational empirical optimization is one of the essential experimental perspective because it reduces production time and costs by opting for the best criterion among the set of factors that utilize the different statistical methodologies. The statistical methods such as Plackett-Burman, and central composite randomised design (CCRD) can be used to enhance the production of cyanobacterial pigments in a short stretch of time. These statistical approaches are an immense multi-practical apparatus for building experimental models that can be utilized in almost all organisms to optimize the recovery of bioproducts [22]. Many researchers uses the RSM which helps to obtain the maximum optimized level of factors by screening or by factorial designs [23]. By optimizing the independent variables as well as predicting the response accurately, RSM lowers the experimental runs. Still, it fails to describe the non-linear regression equation, which is the major drawback of using the RSM [24]. To overcome this problem, artificial neural network (ANN) models are employed in this study to solve non-linear issues and explain the non-linear regression equation as well [25, 26]. ANN model illustrates the non-linear relations between input and output factors, which led to its outspread applications in modelling biological systems and tuning up multiple variables [26, 27]. In order to simulate the biomass, total lipids as well as oleic acid concentrations in Chlorella vulgaris, the ANN model was established where all exhibiting high accuracy (for training R2 > 0.97) with a concentration of 2663.34 mg/L, 1266.33 mg/L, 745.21 mg/L respectively, according to Liyanaarachchi et al. [27]. Moreover, Saini et al., [21] optimized the production of cell biomass and phycobiliproteins (PBPs) using multi-objective hybrid machine learning approach in Nostoc sp. CCC-403 and found this approach is highly efficient in improving the yield of total PBPs (61.76% increase) and in cell biomass (90% increase). Furthermore, the process model, RSM and ANN were used for anaerobic digestion of the poultry litter together with wheat straw in addition to biochar for methane gas production [28]. However, when RSM is integrated with various modelling tools (viz., machine learning, AI, simulation and economic analysis models), it robustness and accurate predicting ability are improved. These model integrations could be helpful in designing and exploring new techniques and algorithms, which can generate understanding in process dynamics, searching for key variables and interpreting of interactive effects [29]. Additionally, it has the potential to discover innovative nanotechnology and sensor technologies for treating emerging contaminants resulting in resource recovery [30]. Moreover, it provides the idea of incorporating circular economy approaches and the sustainable criteria optimization process into the livestock wastewater treatment process [31, 32].

To best of our knowledge, this study represents the most appropriate BG-110 medium containing three different nitrogen sources (sodium nitrate, ammonium chloride and urea), for Synechocystis sp. PCC 6803 which employs RSM-AI-based ANN-MOGA model as the machine learning optimization tool. The use of different nitrogen sources at various concentration of sodium nitrate (1–18 mM), ammonium chloride (0.50-3 mM) and urea (0.50-3 mM) in BG-110 medium shortens the cultivation period while improving the pigment accumulation in Synechocystis sp. PCC 6803 thus making the process economically viable. The hypothesis behind this study was to enhance the accumulation of cyanobacterial pigments (chlorophyll a, carotenoids and phycocyanin) in the Synechocystis sp. PCC 6803, a unicellular and freshwater species, using conventional (OFAT) and RSM-AI based ANN-MOGA optimization tool. Based on the experimental data, the CCRD model was established which was obtained from OFAT studies conducted under different nitrogen sources (nitrate, urea, and ammonium chloride), two different temperatures, and light intensities. Furthermore, optimization model - CCRD and ANN-MOGA were employed sequentially to study the interaction between the above abiotic factors and identify the optimum concentration of the considerable factors for higher pigments production. The integrated ANN-MOGA model was developed for the optimization and enhancement of pigment production in Synechocystis sp. PCC 6803 under three different nitrogen sources.

Methods

Cyanobacteria culture conditions

Cyanobacterium Synechocystis sp. PCC 6803 (Accession no. PRJNA821690) was procured from the Central University of Hyderabad (UoH) [Courtesy: Prof. JSS Prakash, Department of Biotechnology and Bioinformatics, UoH]. The axenic culture was photo-autotrophically cultivated in the 250 mL Erlenmeyer flask containing 50 mL of BG-110 medium (pH 8) at 30 ± 2 ℃ under continuous illumination of 50 µmol photons /m2/s.

Estimation of pigments

The estimation of chlorophyll a and carotenoids were performed according to protocol described by Tanweer et al. [33]. Briefly, 1 mL of culture was aseptically harvested and centrifuged at room temperature (25 ℃) at 8000×g for 7 min. The obtained pellets were homogenised in absolute methanol and kept overnight at 4 ℃. After overnight incubation, the sample mixture was centrifuged at 11,519×g for 7 min at 4 ℃. The absorbance of the supernatant at optical density (OD) of 470 nm and 720 nm for carotenoids, 665, and 720 nm for chlorophyll a was measured using microplate reader (FlexA-200, Genetix Biotech Asia Pvt. Ltd.). The concentration of the chlorophyll a as well as carotenoids were calculated using the following [Eqs. 1 & 2] [34].

$$\:\text{C}\text{h}\text{l}\text{a}\:[{\mu\:}\text{g}/\text{m}\text{L}]=12.9447(\text{A}665-\:\text{A}720)$$
(1)
$$\:\text{C}\text{a}\text{r}\text{o}\text{t}\text{e}\text{n}\text{o}\text{i}\text{d}\text{s}\:[{\mu\:}\text{g}/\text{m}\text{L}]=\frac{\left[\text{1,000}\:\right(\text{A}470\:-\:\text{A}720)\:-\:2.86\:(\text{C}\text{h}\text{l}\text{a}\:[{\mu\:}\text{g}/\text{m}\text{L}]\left)\right]}{221}$$
(2)

Phycocyanin content in the Synechocystis sp. PCC 6803 culture was determined by taking an absorbance at OD 620, and 650 nm. The concentration of phycocyanin (PC) was evaluated according to following [Eq. 3] [5].

$$\:\text{P}\text{C}({\mu\:}\text{g}/\text{m}\text{L})=\frac{(\text{A}620-0.47\times\:\text{A}650)}{5.34}\:\:$$
(3)

Optimization of pigments production through one factor at a time approach (OFAT)

Cyanobacterial pigments viz., chlorophyll a, carotenoid, and phycocyanin production were optimized using OFAT approach. The process variables such as incubation temperatures (28 ℃ and 16 ℃), continuous illumination at light intensities (~ 50 and ~ 150 µmol photons/m2/s) and three different nitrogen sources (sodium nitrate, ammonium chloride and urea) were used to optimize the chlorophyll a, carotenoid and phycocyanin production. Initially, the experiments were setup in two distinct stress conditions: (i) control condition (28 ℃ and ~ 50 µmol photons/m2/s) and (ii) cold (16 ℃) and high light (~ 150 µmol photons/m2/s) condition. The erlenmeyer flasks of 250 mL containing 50 mL each of BG-110 medium supplemented by (i) sodium nitrate (17.6 mM), (ii) ammonium chloride (5 mM), urea (1 mM), and (iii) combination of nitrate (17.6 mM) with ammonium chloride (5 mM) and nitrate (17.6 mM) with urea (1 mM) were used for cultivation. All the flasks were inoculated with an initial inoculum size of 0.1 OD at 680 nm and kept on a rotary shaker at 150 rpm for incubation up to 12 days in all the experiment sets under control (28 ℃ and ~ 50 µmol photons/m2/s) and stress condition (16 ℃ and high light of ~ 150 µmol photons/m2/s). The samples were routinely harvested and processed for growth study and pigment production.

Experimental design

The software namely, design expert version 13.0 (Stat-Ease, Inc., Minneapolis, MN, USA) was opted for statistically optimizing the production of chlorophyll a, carotenoids, and phycocyanin. Based on OFAT observation, different nitrogen sources were used for further optimization for analysing the best fitted process variables for obtaining the desired outcomes. In this study, a three-factor, four-level central composite randomised design (CCRD) was performed to evaluate statistically the effects of three essential input variables, namely: A, urea (0.5 to 3.0 mM); B, ammonium chloride (0.5 to 3.0 mM), and C, sodium nitrate (1.0 to 18.0 mM) on the pigment’s production. The nonlinear quadratic models are provided as the equations for the different responses, after exploring the various factors. Along with a midpoint an entire set of twenty runs with three reproducible sets was designed which permits the evaluation of the pure error sum of squares. In each set, the resulting data was an average of a triplicate runs.

Development of ANN-MOGA model

ANN is based on the human neural system which is a more advanced approach than RSM for optimizing the process or product [35]. Multilayer perceptron model was employed in the ANN methodology, which was established by nntool in MATLAB Version 9.3 (R2017b). Herein, three-layered feed-forward ANN system employed as input layer, hidden layer (by using Sigmoid function) together with output layer (by using Purlin function), including 10 neurons. ANN combined with Multi-Objective Genetic Algorithm (MOGA) further utilised for the better optimization of process parameters, which could provide a significant and best fitted optimized variables based on provided fitness function. Briefly, from ANN model a.mat file was obtained having input variables viz., sodium nitrate, urea and ammonium chloride, with output responses chlorophyll a, carotenoid and phycocyanin was loaded upon fitness and at last to produce the final optimised variables for the enhanced response, these files are fed to MOGA. The following parameters such as crossover fraction (0.8), and constraint-independent were used for the ANN-MOGA execution. By setting 100 as the population size, number of iteration 100, and elite count 1 as optimization parameters for MOGA tool. An ideal design was selected on the basis of architecture having the minimum root mean square error (RMSE) value as well as the superlative regression coefficient by ANN prediction model. The RMSE value was estimated by using [Eq. 4] [36].

$$\:\text{R}\text{M}\text{S}\text{E}\:=\sqrt{{\sum}_{i=1}^{\text{N}}\frac{1}{\text{N}}{\left({Y}_{p}-{Y}_{e}\right)}^{2}}$$
(4)

Where, Ye is an actual experimental value, Yp is a predicted experimental value, moreover N is number of test samples.

2, 2-diphenyl-1-picrylhydrazyl (DPPH) assay

Free radical scavenging activity of the extracted pigment of Synechocystis sp. PCC 6803 was determined using 2, 2-diphenyl-1-picrylhydrazyl (DPPH) assay [20]. The ANN-MOGA optimized pigment extracts were obtained from cell pellets after the overnight incubation with methanol at 4℃. Further, the DPPH of 1000 mM concentration stock was prepared in ethanol and methanol was used to dilute the sample, ranging in concentration from 10 to 40 µg/mL. The 60 µL DPPH solution was added to the 60 µL of pigment extract and incubated for 30 min at room temperature. After incubation, the absorbance at OD 517 nm was measured by a microplate reader (Genetix Biotech Asia Pvt. Ltd, India) using methanol as a blank. The ascorbic acid was used as standard (positive control) for preparing the calibration curve. The following equation [Eq. 5] was used, for the determination of the antioxidant activity [20]:

$$\:\text{D}\text{P}\text{P}\text{H}\:\text{r}\text{a}\text{d}\text{i}\text{c}\text{a}\text{l}\:\text{s}\text{c}\text{a}\text{v}\text{e}\text{n}\text{g}\text{i}\text{n}\text{g}\:\text{a}\text{c}\text{t}\text{i}\text{v}\text{i}\text{t}\text{y}\:\left(\%\right)=\left[\frac{{A}_{0}-{A}_{1}}{{A}_{0}}\right]\times 100$$
(5)

Preparation of enzyme extract for the determination of antioxidant activities

To extract the enzymes from Synechocystis sp. PCC 6803, 1 mL of ANN-MOGA optimized culture was centrifuged at 8000 rpm at 4℃ for 10 min. Cell pellets were homogenized in 1 mL of extraction buffer containing 50 mM K3PO4 and 0.1 mM EDTA at pH 7. The homogenized mixture was centrifuged for at 4˚C for 20 min at 8000 rpm, and the supernatant was used as the enzyme extract to analyse the enzyme activity.

GPX-activity assay

The guaiacol peroxidase activity was assayed using a protocol described by Senousy et al. [37] with some modifications. The 1 mL reaction mixture contains 920 µL of 100 mM potassium phosphate buffer of pH 7 with 20 mM guaiacol mixed with 65 µL of enzyme extract. To start the reaction, 15 µL of 0.6% H2O2 was added to this mixture, and absorbance change was measured by measuring the OD at 470 nm in every 10 s for 3 min using microplate reader (Genetix Biotech Asia Pvt. Ltd., India). The guaiacol activity was calculated using 26.6 mM/cm as an extinction coefficient.

CAT-activity assay

The catalase activity of the extracts was evaluated by measuring the change in the OD at 240 nm and recorded in every 10 s for 3 min using microplate reader (Genetix Biotech Asia Pvt. Ltd). Here, 1 mL of reaction mixture was prepared using 920 µL of potassium phosphate buffer, 65 µL of supernatant extract, and 15 µL of 750 mM H2O2. The catalase activity was calculated using 39.4 mM/cm as an extinction coefficient [38].

APX-activity assay

The ascorbic peroxidase activity of the extracts was measured at OD of 290 nm. The absorbance was recorded in every 10 s for 3 min using microplate reader (Genetix Biotech Asia Pvt. Ltd). 1 mL of the reaction mixture was prepared by mixing 15 µL of ascorbic acid (0.5 mM) in 920 µL of phosphate buffer (50 mM), which contains 5 mM EDTA. The addition of 15 µL of H2O2 (1 mM) initiated the reaction in the reaction mix. The extinction coefficient for ascorbate, which is 2.8 mM/cm, was used to compute the ascorbic activity [39].

Statistical analysis

Each experiment was performed in triplicates. One-way analysis of variance (ANOVA) used for calculating the significant differences (p < 0.05). RSM experiment was statistically studied by using the statistical design expert 13.0 statistical software package.

Results

One-factor-at-a-time (OFAT) optimization

The chlorophyll a, carotenoid and phycocyanin production were optimized using a single factor-at-a-time approach. The single factor experiment study was performed under control conditions (28℃ and ~ 50 µmol photons/m2/s) and abiotic stress conditions, i.e., cold at (16℃ and high light ~ 150 µmol photons/m2/s), withdifferent nitrogen sources (nitrate, urea, ammonium chloride, nitrate with urea and ammonium chloride) on Synechocystis sp. PCC 6803 as tabulated in Table 1. The Synechocystis sp. PCC 6803 was grown for twelve days under control and stressed conditions, which allowed us to compare growth profile with pigment yields (Figs. 1 and 2). The pigment accumulation consistently increased from day one to day twelve in all the control conditions (50 µmol photons/m2/s light intensity at 28℃) except for the BG-110 media supplemented with ammonium chloride at 5 mM concentration. As a result, the best sequence found for pigment accumulation was nitrate (17.6 mM) and ammonium chloride (5 mM) > nitrate (17.6 mM) and urea (1 mM) > nitrate (17.6 mM) > urea (1 mM) under control condition (50 µmol photons/m2/s light intensity at 28℃). At twelfth day of cultivation, production of chlorophyll a, carotenoid and phycocyanin was 13.53, 9.25, and 0.019 µg/mL, respectively which was seen highest in the medium containing both nitrate (17.6 mM) and ammonium chloride (5 mM) (Fig. 1b, c & d). Furthermore, under nitrate (17.6 mM) and urea (1 mM) combination, the chlorophyll a content of 12.90 µg/mL, carotenoid 6.00 µg/mL and phycocyanin 0.016 µg/mL was the second highest vs. the control condition (28℃ and ~ 50 µmol photons/m2/s) with chlorophyll a 6.37 µg/mL, carotenoid 3.88 µg/mL and phycocyanin 0.008 µg/mL. Synechocystis sp. PCC 6803 grown in BG-110 media containing 17.6 mM nitrate showed the highest chlorophyll a (1.15 µg/mL), carotenoid (1.06 µg/mL) and phycocyanin (0.001 µg/mL) under stress condition (~ 150 µmol photons/m2/s high light at 16℃) (Table 1).

Fig. 1
figure 1

Growth and pigment production analysis in Synechocystis sp. PCC 6803 under control photoautotrophic conditions. a) Growth of Synechocystis sp. PCC 6803 for 24 h with control photoautotrophic conditions (50 µmol photons/m2/s light at 28℃) in different nitrogen sources, b), c), and d) chlorophyll a, carotenoids, and phycocyanin content of Synechocystis sp. PCC 6803 in different nitrogen sources- nitrate (N), urea (U) and ammonium chloride (A) under photoautotrophic condition. Data was represented as the means of error bars which indicates the standard deviations (± SD) of biological and experimental triplicates

Full size image
Fig. 2
figure 2

Growth and pigment production analysis in Synechocystis sp. PCC 6803 under abiotic stress conditions. a) Growth of the Synechocystis sp. PCC 6803 under high light (~ 150 µmol photon/m2/s at 16℃) in different nitrogen sources, b), c), and d) chlorophyll a, carotenoids and phycocyanin content of the Synechocystis sp. PCC 6803 in different nitrogen sources- nitrate (N), urea (U) and ammonium chloride (A) under high light (~ 150 µmol photon/m2/s at 16℃). Data was represented as the means of error bars which indicates the standard deviations (± SD) of biological and experimental triplicates

Full size image
Table 1 OFAT approach for cyanobacterial pigment optimization in the Synechocystis sp. PCC 6803 under two different conditions. The variables used in a single factor at a time approach for an optimization of cyanobacterial chlorophyll a, carotenoid and phycocyanin production from Synechocystis sp. PCC 6803 with different nitrogen sources cultivated for 24 h illumination (I) control photoautotrophic conditions (50 mmol photons/m2/s light intensity at 28℃ ) and (II) abiotic stress conditions (~ 150 mmol photons/m2/s high light at16℃)
Full size table

Statistical optimization by CCRD model of RSM

All set of an experiment design were carried out under laboratory-controlled environment (28℃ and ~ 50 µmol photons/m2/s). A set of 20 tests were generated and all were set in biological triplicates with chlorophyll a, carotenoid and phycocyanin concentration as the output responses. The experiment run numbers i.e., one and twelve yielded highest chlorophyll a content of 22.00 µg/mL, than the other sets. The highest carotenoid concentration of 11.63 µg/mL was produced by experiment run number fifteen (refer to Table 2). The results of this experiment showed that the contents of phycocyanin and chlorophyll a were 0.010 and 20.50 µg/mL, respectively, which is slightly lower than experiment run number one. On the other hand, the highest phycocyanin was recorded as 0.026 µg/mL, but the values for carotenoid and chlorophyll a were 9.65 µg/mL and 13.70 µg/mL found in the experiment run number two (Table 2).

Table 2 Central composite randomized design (CCRD) experimental matrix for media optimization for the growth of Synechocystis sp. PCC 6803. Experimental matrix of CCRD for optimization of urea, ammonium chloride and sodium nitrate concentrations in the cultivation media of the Synechocystis sp. PCC 6803. The bold highlights represent the maximum response obtained
Full size table

In addition, for statistical analysis the model was performed using the analysis of variance (ANOVA). The model was evaluated using coefficient of determination (R2) and also its significance was analysed with the help of p-value and f-value (Table 3). The study found that model terms with lower p-values (< 0.0001) were significant, in our study C, A², B², C² for chlorophyll a, C, AC, A², C² for carotenoid, and A, B, C, AB, A², B², C² for phycocyanin. Second-order polynomial equation was used to correlate with an independent factor achieving the high chlorophyll a, carotenoids and phycocyanin production, as shown below.

Table 3 ANOVA analysis for pigment optimization in Synechocystis sp. PCC 6803. ANOVA analysis of central composite randomized design (CCRD) model for the optimization of chlorophyll A, carotenoid and phycocyanin production from the Synechocystis sp. PCC 6803
Full size table

Chlorophyll a (µg/mL) = 20.2488 + -0.0464823 * A + 0.81797 * B + 4.56193 * C + -0.471665 * AB + -1.02154 * AC + 1.25349 * BC + -4.76921 * A2 + -1.78231 * B2 + -2.50667 * C2.

Carotenoid (µg/mL) = + 11.0205 + -0.167231 * A + 0.115313 * B + 1.83205 * C + 0.05275 * AB + -1.39275 * AC + 0.22725 * BC + -2.06397 * A2 + -0.305046 * B2 + -1.18363 * C2.

Phycocyanin (µg/mL) = 0.0169327 + 0.00139847 * A + -0.00114919 * B + -0.00283519 * C + 0.00238358 * AB + 0.000436642 * AC + -2.88702e-05 * BC + -0.00262057 * A2 + 0.00221869 * B2 + -0.0026564 * C2.

Where, A = urea, B = ammonium chloride, and C = sodium nitrate.

The mutual effect of the experimental factors (variables) on the response value (chlorophyll a, carotenoid, and phycocyanin yield) was revealed by the 3D response surface graph as shown in Figs. 3 and 4I. The interaction between ammonium chloride and urea, sodium nitrate and ammonium chloride, sodium nitrate and urea were found to be the strongest (Fig. 3) in case of chlorophyll a and carotenoid. On the other hand, the interaction effect of model terms on phycocyanin production was not found appropriate (Fig. 4I). The highest chlorophyll a of 21.89 µg/mL and carotenoid of 9.46 µg/mL were achieved, whereas the low phycocyanin concentration of 0.048 µg/mL, was observed under control condition (28℃ and ~ 50 µmol photons/m2/s). The plot between the actual and predicted values is shown in Fig. 4II. The intended RSM model was validated by a close conformance between these values along the line.

Fig. 3
figure 3

The 3-dimensional response surface plot for effects. (a) ammonium chloride and urea, (b) sodium nitrate and ammonium chloride, (c) sodium nitrate and urea concentration as well as their interaction effect on (I) chlorophyll a and (II) carotenoid production by Synechocystis sp. PCC 6803. Other variables were set at zero level

Full size image
Fig. 4
figure 4

The 3-dimensional response surface plot for effects. (a) ammonium chloride and urea, (b) sodium nitrate and ammonium chloride, (c) sodium nitrate and urea concentration as well as their interaction effect on (I) phycocyanin production by Synechocystis sp. PCC 6803. (II) Predicted vs. actual value for (a) chlorophyll a (b) carotenoid (c) phycocyanin production from Synechocystis sp. PCC 6803

Full size image

Optimisation in multi-objective genetic algorithm (MOGA) and artificial neural network (ANN) tool

The Feed Forward Back Propagation (FFBP) algorithm was often run to minimize the mean square error value. A tangent sigmoid activation function in a regression plot was generated from ANN-MOGA has been shown in Fig. 5. The R2 values of training, testing, and validation were obtained with value of 0.99, 0.98 and 0.98, respectively. The overall result displayed the 0.98 R2 value, which corresponds model accuracy to 98% (Fig. 5).

Fig. 5
figure 5

A regression plot obtained from the Levenberg-Marquardt back propagation algorithm depicts the ANN model’s training, test validation and overall performance

Full size image

The favourable network configuration for the generation of pigments was found at an epoch of 2. The mean squared error (MSE) of the network and the performance of the ANN for the computation of pigment production throughout training, validation, and testing were shown in Fig. 6(b). The minimum value of MSE (1.45) for validation was acquired at this point, after which MSE began to increase. The error value was initially higher but during the training, testing, and validation phases when more epochs were added, the error dropped from its initial value (Fig. 6a and b). At epoch 2, the model’s training was completed. Errors between the target and output responses were illustrated by an error histogram plot that resembles how predicted values were deviated from target values (Fig. 6c). The error histogram plot is separated into 20 tiny bins, also the zero-error line is on the zero region at the X axis, which is composed of testing, training and validation model representing optimal result making the ANN model considerable. The pigments production efficiency error histogram (Fig. 6c) showed that errors primarily lie between 25 and 26. Zero error is on a vertical line parallel to the ordinate at 23 instances during training. Figure 6d displays the Pareto front which is the graph of the output of three objectives: chlorophyll a (µg/mL), carotenoids (µg/mL), and phycocyanin (µg/mL). The Pareto front offers all possible chlorophyll a, carotenoid, and phycocyanin production choices.

Fig. 6
figure 6

The validation plot (a) of the ANN model through gradient along with validation checks at epoch 6, (b) The plot for validation performance of mean square error vs. epoch for neural networks where outstanding validation performance was 1.45 achieved at 2nd epoch, (c) An error histogram plot for the network illustrating the error convergence during data testing in which the minimal difference is shown among output and predicted response depicts by zero error (d) A Pareto front is a graph illustrating an average mean square value among the different output responses for chlorophyll a, carotenoid and phycocyanin

Full size image

Experimental validation of ANN-MOGA model

The ANN-MOGA obtained optimized solution was experimentally validated for pigments production (chlorophyll a, carotenoid and phycocyanin). In virtue of ANN-MOGA validation experiment, the elevated amount for carotenoid, chlorophyll a, and phycocyanin were 9.78 µg/mL, 21.93 µg/mL and 0.05 µg/mL, respectively in BG-110 medium supplemented with urea (2.25 mM), ammonium chloride (2.29 mM) and sodium nitrate (17.99 mM).

Determination of antioxidant activities

In the present study, antioxidant activity was examined in the pigment extract of Synechocystis sp. PCC 6803. Since methanolic extracts of the pigments expressed radical scavenging ability towards DPPH radicals and reducing power, the achieved data indicates the scavenging activity. The scavenging activity was determined by the changes in the absorbance value of the pigment extract solution when it has reduced DPPH free radicals. In the present study, 10–60 µg/mL of pigment extract revealed a scavenging percentage of 82.48–17.49% in response to DPPH. Figure 7a showed that the pigment showed the highest inhibition efficiency of 82.48%, and ascorbic acid exhibited 99.37% of free radical scavenging activity at 10 µg/mL. The results of the IC50 value of pigment extracts and standard ascorbic acid against DPPH free radicals are shown in Fig. 7b. As depicted in Fig. 7b, the IC50 value of 7.66 ± 0.001 µg/mL was obtained in the case of pigment extract of ANN-MOGA compared to standard with IC50 value of 6.15 ± 0.005 µg/mL.

Fig. 7
figure 7

(a) DPPH radical scavenging activities and (b) IC50 of various concentrations of pigment extracts by Synechocystis sp. PCC 6803, where ascorbic acid was used as standard. Mean ± SD, n = 3

Full size image

Apart from DPPH radical scavenging activity, antioxidant enzyme activities CAT, GPX and APX were also evaluated in the supernatant of the Synechocystis sp. PCC 6803. It was observed that Synechocystis sp. PCC 6803 exhibited significant enzyme activity of 4.0409 unit/min/mL with R2 value of 0.9985 for APX (Fig. 8a), while enzyme activity was 0.39438 unit/min/mL with R2 value of 0.9959 for CAT (Fig. 8b). The GPX enzyme activity was 0.01497 unit/min/mL with a 0.9231 R2 value (Fig. 8c).

Fig. 8
figure 8

(a) APX, (b) CAT and (c) GPX radical scavenging activities of various concentrations of carotenoid extracts by Synechocystis sp. PCC 6803 where the R2 ≥ 0. 998, 0.995 and 0.923. Mean ± SD, n = 3

Full size image

Discussion

The cultivation of Synechocystis sp. PCC 6803 was assessed for its growth and pigment production for twelve days under control and stressed conditions. Under the control condition (28℃ and ~ 50 µmol photons/m2/s), the growth of the Synechocystis sp. PCC 6803 cells were found to be consistently increasing from the day of inoculation to twelfth day in all nitrogen sources (refer Table 1 for concentration) except ammonium chloride. All cultures exhibited rapid growth from the beginning of the cultivation with a brief lag phase vs. the cultures grown solely on BG-110 medium supplemented with ammonium chloride. In case of ammonium chloride (5 mM), we observed slow growth with a slow lag phase of the cyanobacterial cells at the third day, which further not increased till twelfth day. In BG-110 media supplemented with urea (1 mM), nitrate (17.6 mM), combined nitrate (17.6 mM) and urea (1 mM), and combined nitrate (17.6 mM) and ammonium chloride (5 mM), the Synechocystis sp. PCC 6803 cultures reached the stationary phase on the 4th, 6th, 8th, and 12th day of cultivation respectively, as shown in Fig. 1a. This confirmed that the longest exponential phase taken by the cyanobacterial cells was in cultures grown in BG-110 medium supplemented with nitrate in combination with ammonium chloride (twelfth days) and nitrate in combination with urea (eight days), respectively as nitrogen sources.

The sequence of accumulation of pigments was best found in the order, nitrate and ammonium chloride > nitrate and urea > nitrate > urea under control condition (50 µmol photons/m2/s light at 28℃). The accumulation of cyanobacterial pigments was seen consistently increasing from day one to day twelfth in all control condition except in case of ammonium chloride growth media. In ammonium chloride based BG-110 medium, cyanobacterial cells lost pigments during the cultivation process. The production of chlorophyll a, carotenoid and phycocyanin (13.53, 9.25, and 0.019 µg/mL), respectively was assessed to be highest on the twelfth day of cultivation in the BG-110 media supplemented with combination of nitrate and ammonium chloride (17.6 mM and 5 mM) compared to other four nitrogen sources such as nitrate and urea (17.6 mM and 1mM), nitrate (17.6 mM), urea (1mM) and ammonium chloride (5 mM) under control condition (28℃ and ~ 50 µmol photons/m2/s) (Fig. 1b, c & d). This may be due to the reason that in combination media, the concentration of ammonium was so less, that makes it non-toxic for the cyanobacterial cells. However, when ammonium chloride was solely used as nitrogen source, it caused toxicity to the cells [40]. Inabe et al. [41] and Deschoenmaeker et al. [42] also observed similar phenomenon, where they found that limited amount of ammonium concentrations to be non-toxic to cyanobacterial cells instead it allows faster growth as well as their bloom development. The chlorophyll a content of 12.90 µg/mL, carotenoid 6.00 µg/mL and phycocyanin 0.016 µg/mL was obtained second highest under BG-110 medium supplemented with nitrate and urea combination (Table 1) which was similar to the study reported by Morocho-Jácome et al. [43] in Arthrospira platensis with chlorophyll a of 372.2 ± 7.7 mg/L/day. Similarly, in another study [44] chlorophyll a 0.603 mg/g, carotenoid 0.0012 mg/g and phycocyanin 0.011 mg/g was reported in Anabaena sphaerica MBDU 105.

Many researchers have reported the enhancement of pigment accumulation in algae in response to abiotic stress conditions, for example, high light and cold conditions [45, 46]. According to Dong et al. [47], cell growth performance was also found to be influenced by several abiotic stress factors. To overcome these stresses, cells regulate their cellular mechanisms and accumulate stress metabolites that support the growth of cyanobacteria. Taking these facts into consideration, we performed the Synechocystis sp. PCC 6803 cultivation in combination of high light (~ 150 µmol photons/m2/s) and cold temperature (16℃) to evaluate the effect of the different stress conditions on growth and accumulation of chlorophyll a, carotenoids and phycocyanin content. The results indicated that Synechocystis sp. PCC 6803 in all nitrogen based BG-110 media underwent abnormally sluggish growth at the beginning of the culture under cold and high light stress conditions, and the lag phase lasted for an excessively long time until the sixth day after cultivation. Herein, cultures supplemented with all different nitrogen sources and nitrate combinations (refer Table 1) in BG-110 media enter the log phase on 8th day of cultivation, respectively (Fig. 2a). The production of chlorophyll a, carotenoid, and phycocyanin in all BG-110 medium supplemented with nitrogen source (refer Table 1) began to increase on the sixth day of cultivation; nevertheless, the yield was determined to be lower compared to control condition at 28℃ and ~ 50 µmol photons/m2/s (Fig. 2).

The sequence of accumulation of chlorophyll a, carotenoid, and phycocyanin was best found in the order nitrate > nitrate and ammonium > nitrate and urea > urea > ammonium chloride in the stress condition (~ 150 µmol photons/m2/s high light at 16℃). Synechocystis sp. PCC 6803 grown in 17.6 mM nitrate supplemented with BG-110 medium showed the highest chlorophyll a (1.15 µg/mL), carotenoid (1.06 µg/mL) and phycocyanin (0.001 µg/mL) compared to the cultures grown in other nitrogen sources in case of stress condition (Table 1). Also, cultures cultivated in a BG-110 medium containing 5 mM of ammonium chloride showed no growth. Ammonium chloride showed a repressive effect on the growth and nutrient uptake rate of cyanobacteria. The results confirmed that combination of cold temperature and high light stress in presence of ammonium chloride supplemented with BG-110 medium may lower the metabolism of the culture which slowed down the growth as well as the synthesis of pigments. Guo et al. [48], reported that when organisms encounter low-temperature stress, it impacts their energy production, metabolic activity, cell growth, and growth rate. The speculated reason behind the slower cell growth could be the presence of lower carotenoids content in cyanobacteria [49]. Carotenoids are essential in energy dissipation in cyanobacteria and these pigments also protect against the photo-oxidative damage [48, 49]. Therefore, due to the less accumulation of carotenoid, the cyanobacterial cells were unable to protect themselves from the oxidative stress. The study by Dong et al. [47], found that cold stress conditions leads to decrease in chlorophyll a, carotenoids, and phycocyanin content in Arthrospira. Similarly, in Synechocystis sp. PCC 6803, Kłodawska et al. [50] found a reduction in the total carotenoid pool in response to extreme temperature stress condition. Furthermore, Daddy et al. [51] reported that under high-intensity light exposure a detrimental impact on cell development, photosynthetic efficiency, and the survivability of Synechocystis sp. PCC 6803 was observed.

In many studies, RSM was used to evaluate the data to achieve the highest glycerol utilisation in a mixotrophic system and to produce high biomass and lipid outputs [52]. Similarly, CCD was employed to access the effect of chemical modulators on lipids, biomass lipid productivity and fatty acid [53]. Using CCRD model chlorophyll a, carotenoid as well as phycocyanin concentration was improved in the present study. Similar to this study, RSM approach was previously used to improve the production of zeaxanthin in microalgae Dunaliella tertiolecta (NIOT-141) [22]. Further, researchers optimized the De Walne’s medium for microalgae cultivation in turn to maximise the zeaxanthin production (20.2 ± 1.29 mg/L) [22]. In Anabaena variabilis, RSM methodology was utilized by researchers to boost biomass and total carbohydrate yield by around 1.5 and 2.4 times, respectively [54]. In another study, a maximum glycerol consumption rate of 770.2 mg/L/d, was identified in C. vulgaris using RSM approach [52].

In order to enhance the pigments production, in addition to RSM, generated ANN model was incorporated into a multi-objective genetic algorithm to provide an optimal variable with maximum targeted values. To prevail over the restraint of RSM, ANN tool successively optimizes the biological multi-response variables. An experimental design matrix with 20 experimental sets of input factors for each variable range along with three output factors for chlorophyll a, carotenoid, and phycocyanin was derived from the CCRD analysis, which was further used as an experimental datasets generation to train the ANN model. Out of 20 experimental sets, fourteen were used to train the model, three were used to test its performance, and remaining three were used to validate it. The feed forward back propagation (FFBP) algorithm was often run to minimize the mean square error value. To train, test and validate the data an ANN approach was employed with a three input and a three output factors. Into an input layer a data was imported, discerned by ten hidden layers in order to produce a required response as similar to Saini et al. [21]. To make an ANN model as a noteworthy, the regression coefficient result depicts the strong relation between the input as well as the output variables. The R2 values of training, testing along with the validation were nearly to one, R2 value of 0.99 was produced by the training model, the 0.98 R2 value was produced by validation model, whereas R2 value for the testing model was 0.98. In a study by Zhang et al. [55], they observed the R2 value of training and testing model as 0.86 and 0.87, respectively. Whereas Rahman et al. [56] observed the R2 value of 0.99 of training, validation and testing model.

The best-optimized solution was generated and selected from ANN-MOGA and was experimentally validated for chlorophyll a, carotenoid and phycocyanin production. The experiments for the validation were conducted in triplicates with three input factors to prove the predicted model. Through ANN-MOGA validation experiment, the highest carotenoid, chlorophyll a, and phycocyanin contents were found to be 9.78 µg/mL, 21.93 µg/mL and 0.05 µg/mL, respectively in BG-110 medium supplemented with urea concentration of 2.25mM, ammonium chloride 2.29mM and sodium nitrate 17.99mM under control conditions (28℃ and ~ 50 µmol photons/m2/s). However, through RSM, the highest chlorophyll a, and phycocyanin production were found 21.89 µg/mL and 0.04 µg/mL, respectively and carotenoid production was 9.46 µg/mL under control conditions (28℃ and ~ 50 µmol photons/m2/s). In this study, the experimental results were coherent to the predicted outcome produced by the hybrid machine learning tool so the accuracy of the expected outcome was validated. Therefore, it can be concluded that in order to determine the well-suited optimal medium formulation for the simultaneous generation of chlorophyll a, carotenoid, and phycocyanin from Synechocystis sp. PCC 6803, the machine learning multi-objective hybrid optimization approach, ANN-MOGA was effectively employed. In the study, to find out the optimum values for energy efficiency and biomass production, which were 0.043% and 4.61 × 10− 5 mg/mL, respectively for Synechococcus HS-9, Rahman et al. [56] used a process called multi-objective optimization. In another study showed simultaneous improvement in the cell biomass as well as cyanobacterial PBPs production in Nostoc sp. CCC-403 by opting for a multi-objective hybrid optimization approach CNNGA [21]. They achieved an increment in total PBP production of 61.76% and cell biomass of 90% by employing a central composite design (CCD) [21]. In another work, Ma et al. [57] used model-free deep reinforcement learning along with ANN strategy with an aim to develop the bioreactor conditioning for cyanobacterial-phycocyanin (C-PC) production from the Plectonema sp. UTEX 1541. From the experimental validation, they found a 52.1% and 20.1% increase in the product yield and phycocyanin concentration, respectively, compared to a control group. Overall, ANN-MOGA method was found beneficial for regulating product production in microbes. Different nutrients and metabolites substantially have an impact on cyanobacterial metabolic processes which leads to variation in product generation, so by finding optimized nutrient conditions through ANN-MOGA, cyanobacterial pigment production process was improved. Previous studies have been conducted for pigment optimisation using artificial intelligence, which is tabulated in Table 4.

Table 4 Comparative analysis of nutrient optimization in media by employing AI tool for pigment production in various cyanobacteria
Full size table

Additionally, antioxidant activity was also performed where the reduction in absorbance was observed when the DPPH radicals were scavenged. The lower IC50 values are linked to the stronger DPPH radical scavenging activity [58]. Initial DPPH concentration (IC50) was reduce to 50% at 59.67 µg/mL of estimated antioxidant concentration in carotenoids by Metwally et al. [59].

Conclusions

In the present study, ANN-MOGA was employed to produce successive optimisation of pigment accretion in the Synechocystis sp. PCC 6803 cells. The production of chlorophyll a, carotenoid, and phycocyanin was significantly increased by 3.44, 2.52, and 6.25-fold, respectively as compared to control. Furthermore, the model has higher accuracy is confirmed by the validation R2 of 0.98. Therefore, RSM-ANN-MOGA hybrid optimisation approaches were found to be significantly useful model for enhancing the pigment production from Synechocystis sp. PCC 6803. In addition, the obtained pigments showed significant antioxidant and enzyme activities. The outcome from this study provides a basis for the future research on the use of RSM in combination with ANN-MOGA for enhanced accumulation of the industrially significant pigment product in cyanobacteria. The machine learning AI model could be employed for product improvement in most organisms with reduced experimental runs in a short period, which could be commercially helpful for other metabolites production.

Data availability

All the analyzed and generated data are included in this study.

Abbreviations

AI:

Artificial Intelligence

APC:

Allophycocyanin

ANN-MOGA:

Artificial neural network - multi-objective genetic algorithm

ANOVA:

One-way analysis of variance

APX:

Ascorbic peroxidase activity

CAT:

Catalase activity

CCD:

Central composite design

CCRD:

Central composite randomized design

CNN:

Connected neural network

CNNGA:

Connected neural network-genetic-algorithm

DPPH 2:

2-diphenyl-1-picrylhydrazyl

EDTA:

Ethylenediaminetetraacetic acid

H2O2 :

Hydrogen peroxide

FFBP:

Feed-forward back propagation

GA:

Genetic Algorithm

GPX:

Guaiacol peroxidase activity

K3PO4 :

Tripotassium phosphate

MOGA:

Multi-objective genetic algorithm

MSE:

Mean squared error

OD:

Optical density

OFAT:

One-factor-at-a-time

PBPs:

Phycobiliproteins

PC:

Phycocyanin

PE:

Phycoerythrin

RMSE:

Root mean square error

RSM:

Response surface methodology

References

  1. Garlapati D, Chandrasekaran M, Devanesan AA, Mathimani T, Pugazhendhi A. Role of cyanobacteria in agricultural and industrial sectors: an outlook on economically important byproducts. Appl Microbiol Biotechnol. 2019;103:4709–21.

    Article  CAS  PubMed  Google Scholar 

  2. Bibi S, Saadaoui I, Bibi A, Al-Ghouti M, Abu-Dieyeh MH. Applications, advancements, and challenges of cyanobacteria-based biofertilizers for sustainable agro and ecosystems in arid climates. Bioresour Technol Rep. 2024;25:101789.

    Article  CAS  Google Scholar 

  3. Barbosa MJ, Janssen M, Südfeld C, D’Adamo S, Wijffels RH. Hypes, hopes, and the way forward for microalgal biotechnology. Trends Biotechnol. 2023;41:452–71.

    Article  CAS  PubMed  Google Scholar 

  4. Marzorati S, Schievano A, Idà A, Verotta L. Carotenoids, chlorophylls and phycocyanin from Spirulina: supercritical CO2 and water extraction methods for added value products cascade. Green Chem. 2020;22:187–96.

    Article  CAS  Google Scholar 

  5. Noh Y, Lee H, Kim M, Hong SJ, Lee H, Kim DM, et al. Enhanced production of photosynthetic pigments and various metabolites and lipids in the cyanobacteria Synechocystis sp. PCC 7338 culture in the presence of exogenous glucose. Biomolecules. 2021;11:1–17.

    Article  Google Scholar 

  6. Liu D, Liberton M, Hendry JI, Aminian-Dehkordi J, Maranas CD, Pakrasi HB. Engineering biology approaches for food and nutrient production by cyanobacteria. Curr Opin Biotechnol. 2021;67:1–6.

    Article  CAS  PubMed  Google Scholar 

  7. Li W, Su HN, Pu Y, Chen J, Liu LN, Liu Q, et al. Phycobiliproteins: molecular structure, production, applications, and prospects. Biotechnol Adv. 2019;37:340–53.

    Article  CAS  PubMed  Google Scholar 

  8. Żbik P, Kłodawska K, Malec P. The effect of solvent on the optical properties of myxoxanthophyll from Synechocystis sp. PCC 6803. J Mol Liq. 2023;375.

  9. Lucakova S, Branyikova I, Hayes M. Microalgal proteins and bioactives for food, feed, and other applications. Appl Sci. 2022;12.

  10. Cui J, Xie Y, Sun T, Chen L, Zhang W. Deciphering and engineering photosynthetic cyanobacteria for heavy metal bioremediation. Sci Total Environ. 2021;761:144111.

    Article  CAS  PubMed  Google Scholar 

  11. Asih DR, Handayani W, Ananda AZ, Yuniati R, Hirota R, Summerfield TC, et al. An insight into potential phosphate bioremediation and renewable energy from agricultural waste via integrated wastewater treatment systems in Indonesia. Environ Res Commun. 2024;6.

  12. Singh T, Pandey VK, Dash KK, Zanwar S, Singh R. Natural bio-colorant and pigments: sources and applications in food processing. J Agric Food Res. 2023;12:100628.

    CAS  Google Scholar 

  13. Liu Y, Cruz-Morales P, Zargar A, Belcher MS, Pang B, Englund E, et al. Biofuels for a sustainable future. Cell. 2021;184:1636–47.

    Article  CAS  PubMed  Google Scholar 

  14. Mohamed Hatha AA, Sumayya NS. Antioxidants from marine cyanobacteria. In: Marine Antioxidants: Preparations, Syntheses, and Applications. 2022. pp. 119–31.

  15. Jodlbauer J, Rohr T, Spadiut O, Mihovilovic MD, Rudroff F. Biocatalysis in green and Blue: Cyanobacteria. Trends Biotechnol. 2021;39:875–89.

    Article  CAS  PubMed  Google Scholar 

  16. Venil C, Lakshmanperumalsamy P. An insightful overview on microbial pigment, prodigiosin. E J Bio. 2009;5:49–61.

    Google Scholar 

  17. Martins M, Albuquerque CM, Pereira CF, Coutinho JAP, Neves MGPMS, Pinto DCGA, et al. Recovery of chlorophyll a derivative from Spirulina maxima: its purification and photosensitizing potential. ACS Sustain Chem Eng. 2021;9:1772–80.

    Article  CAS  Google Scholar 

  18. Pagels F, Guedes AC, Amaro HM, Kijjoa A, Vasconcelos V. Phycobiliproteins from cyanobacteria: chemistry and biotechnological applications. Biotechnol Adv. 2019;37:422–43.

    Article  CAS  PubMed  Google Scholar 

  19. Lehmann M, Vamvaka E, Torrado A, Jahns P, Dann M, Rosenhammer L, et al. Introduction of the carotenoid biosynthesis α-Branch into Synechocystis sp. PCC 6803 for lutein production. Front Plant Sci. 2021;12:1–13.

    Article  Google Scholar 

  20. Natesungnoen M, Pongrakhananon V, Lindblad P, Jantaro S. Overexpressing carotenoid biosynthetic genes in Synechocystis sp. PCC 6803 improved intracellular pigments and antioxidant activity, which can decrease the viability and proliferation of lung cancer cells in vitro. Int J Mol Sci. 2023;24.

  21. Saini DK, Yadav D, Pabbi S, Chhabra D, Shukla P. Phycobiliproteins from Anabaena variabilis CCC421 and its production enhancement strategies using combinatory evolutionary algorithm approach. Bioresour Technol. 2020;309:123347.

    Article  Google Scholar 

  22. Priyanka S, Kirubagaran R, Leema JTM. Optimization of algal culture medium for Zeaxanthin production by Dunaliella tertiolecta: an RSM based approach. Curr Sci. 2020;119:1997–2005.

    Article  CAS  Google Scholar 

  23. González-Vega RI, Cárdenas-López JL, López-Elías JA, Ruiz-Cruz S, Reyes-Díaz A, Perez-Perez LM, et al. Optimization of growing conditions for pigments production from microalga Navicula incerta using response surface methodology and its antioxidant capacity. Saudi J Biol Sci. 2021;28:1401–16.

    Article  PubMed  Google Scholar 

  24. Srivastava G, Paul AK, Goud VV. Optimization of non-catalytic transesterification of microalgae oil to biodiesel under supercritical methanol condition. Energy Convers Manag. 2018;156:269–78.

    Article  CAS  Google Scholar 

  25. Liyanaarachchi VC, Nishshanka GKSH, Nimarshana PHV, Ariyadasa TU, Attalage RA. Development of an artificial neural network model to simulate the growth of microalga Chlorella vulgaris incorporating the effect of micronutrients. J Biotechnol. 2020;312:44–55.

    Article  CAS  PubMed  Google Scholar 

  26. Sakiewicz P, Piotrowski K, Ober J, Karwot J. Innovative artificial neural network approach for integrated biogas – wastewater treatment system modelling: effect of plant operating parameters on process intensification. Renew Sustain Energy Rev. 2020;124.

  27. Liyanaarachchi VC, Nishshanka GKSH, Sakarika M, Nimarshana PHV, Ariyadasa TU, Kornaros M. Artificial neural network (ANN) approach to optimize cultivation conditions of microalga Chlorella vulgaris in view of biodiesel production. Biochem Eng J. 2021;173:108072.

    Article  CAS  Google Scholar 

  28. Zhan Y, Zhu J. Response surface methodology and artificial neural network-genetic algorithm for modeling and optimization of bioenergy production from biochar-improved anaerobic digestion. Appl Energy. 2024;355:122336.

    Article  CAS  Google Scholar 

  29. Yu A, Liu Y, Li X, Yang Y, Zhou Z, Liu H. Modeling and optimizing of NH4+ removal from stormwater by coal-based granular activated carbon using RSM and ANN coupled with GA. Water. 2021;13:608.

    Article  CAS  Google Scholar 

  30. Reza A, Chen L, Mao X. Response surface methodology for process optimization in livestock wastewater treatment: A review. Heliyon. 2024;10:e30326.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  31. Lin J, Liu S, Han Z, Ma R, Cui C, Sun S. Scaled-up microwave pyrolysis of sludge for hydrogen-rich biogas and life cycle assessment: parameters synergistic optimization, carbon footprint analysis and technology upgrade. Chem Eng J. 2023;452:139551.

    Article  CAS  Google Scholar 

  32. Nowrouzi M, Abyar H. A framework for the design and optimization of integrated fixed-film activated sludge-membrane bioreactor configuration by focusing on cost-coupled life cycle assessment. J Clean Prod. 2021;296:126557.

    Article  CAS  Google Scholar 

  33. Tanweer S, Dash K, Panda B. Enhanced biomass production of Synechocystis sp. PCC 6803 by two associated bacteria Paenibacillus camelliae and Curtobacterium ammoniigenes. Arch Microbiol. 2022;204:1–14.

    Article  Google Scholar 

  34. Zavřel T, Sinetova MA, Červený J. Measurement of chlorophyll a and carotenoids concentration in Cyanobacteria. Bio-Protocol. 2015;5:1–5.

    Article  Google Scholar 

  35. Selvalakshmi S, Immanual R, Priyadharshini B, Sathya J. Artificial neural network (ANN) modelling for the thermal performance of bio fluids. Mater Today Proc. 2022;66:1289–94.

  36. Yu T, Fan F, Huang L, Wang W, Wan M, Li Y. Artificial neural networks prediction and optimization based on four light regions for light utilization from Synechocystis sp. PCC 6803. Bioresour Technol. 2024;394:130166.

    Article  CAS  PubMed  Google Scholar 

  37. Senousy HH, Abd Ellatif S, Ali S. Assessment of the antioxidant and anticancer potential of different isolated strains of cyanobacteria and microalgae from soil and agriculture drain water. Environ Sci Pollut Res. 2020;27:18463–74.

    Article  CAS  Google Scholar 

  38. Muhetaer G, Jayasanka SMDH, Fujino T. Oxidative stress and antioxidant responses of Phormidium ambiguum and Microcystis aeruginosa under diurnally varying light conditions. Microorganisms. 2020;8:1–14.

    Article  Google Scholar 

  39. Nakano Y, Asada K. Hydrogen peroxide is scavenged by ascorbate-specific peroxidase in spinach chloroplasts. Plant Cell Physiol. 1981;22:867–80.

    CAS  Google Scholar 

  40. Sassano CEN, Gioielli LA, Almeida KA, Sato S, Perego P, Converti A, et al. Cultivation of Spirulina platensis by continuous process using ammonium chloride as nitrogen source. Biomass Bioenergy. 2007;31:593–8.

    Article  CAS  Google Scholar 

  41. Inabe K, Miichi A, Matsuda M, Yoshida T, Kato Y, Hidese R, et al. Nitrogen availability affects the metabolic profile in cyanobacteria. Metabolites. 2021;11.

  42. Deschoenmaeker F, Bayon-Vicente G, Sachdeva N, Depraetere O, Cabrera Pino JC, Leroy B, et al. Impact of different nitrogen sources on the growth of Arthrospira sp. PCC 8005 under batch and continuous cultivation – A biochemical, transcriptomic and proteomic profile. 2017.

  43. Morocho-Jácome AL, Sato S, de Lara Capurro Guimarães L, Knysak Camargo de Jesus C, Monteiro de Carvalho JC. Simultaneous use of sodium nitrate and Urea as nitrogen sources improves biomass composition of Arthrospira platensis cultivated in a tubular photobioreactor. Eng Life Sci. 2016;16:338–47.

    Article  Google Scholar 

  44. Anahas AMP, Prasannabalaji N, Muralitharan G. Enhancing biodiesel production in Anabaena sphaerica MBDU 105: exploring photo-, hetero-, and mixotrophic cultivation for biomass, lipid, and fuel properties. Biomass Convers Biorefinery. 2024. https://doiorg.publicaciones.saludcastillayleon.es/10.1007/s13399-024-05640-z.

    Article  Google Scholar 

  45. Supakorn P, Chonlada Y, Anchalee S. Pigment production under cold stress in the green microalga Chlamydomonas reinhardtii. Agric. 2021;11.

  46. Maltsev Y, Maltseva K, Kulikovskiy M, Maltseva S. Influence of light conditions on microalgae growth and content of lipids, carotenoids, and fatty acid composition. Biology. 2021;10:1–24.

    Article  Google Scholar 

  47. Dong S, Jiang Y, Bu Y, Wang S, Zhang H, Wang R. Growth, photosynthetic pigment proteins responses and transcriptome analysis provide insights into survival strategies against short-term cold stress in the blue-green algae. Arthrospira Aquac Rep. 2022;27:101403.

    Google Scholar 

  48. Guo R, Liu T, Guo C, Chen G, Fan J, Zhang Q. Carotenoid biosynthesis is associated with low-temperature adaptation in Rhodosporidium Kratochvilovae. BMC Microbiol. 2022;22:1–9.

    Article  Google Scholar 

  49. Pagels F, Vasconcelos V, Guedes AC. Carotenoids from cyanobacteria: biotechnological potential and optimization strategies. Biomolecules. 2021;11.

  50. Kłodawska K, Kovács L, Várkonyi Z, Kis M, Sozer Ö, Laczkó-Dobos H, et al. Elevated growth temperature can enhance photosystem I trimer formation and affects xanthophyll biosynthesis in Cyanobacterium Synechocystis sp. PCC 6803 cells. Plant Cell Physiol. 2015;56:558–71.

    Article  PubMed  Google Scholar 

  51. Daddy S, Zhan J, Jantaro S, He C, He Q, Wang Q. A novel high light-inducible carotenoid-binding protein complex in the thylakoid membranes of Synechocystis PCC 6803. Sci Rep. 2015;5:1–8.

    Article  Google Scholar 

  52. Rincon SM, Beyenal H, Romero HM. A response surface methodology study for Chlorella vulgaris mixotrophic culture optimization. Microorganisms. 2024;12.

  53. Anahas AMP, Muralitharan G. Central composite design (CCD) optimization of phytohormones supplementation for enhanced cyanobacterial biodiesel production. Renew Energy. 2019;130:749–61.

    Article  CAS  Google Scholar 

  54. Deb D, Mallick N, Bhadoria PBS. Engineering culture medium for enhanced carbohydrate accumulation in Anabaena variabilis to stimulate production of bioethanol and other high-value co-products under cyanobacterial refinery approach. Renew Energy. 2021;163:1786–801.

    Article  CAS  Google Scholar 

  55. Zhang ZX, Xu LW, Xu YS, Li J, Ma W, Sun XM, et al. Integration of genetic engineering and multi-factor fermentation optimization for co-production of carotenoid and DHA in Schizochytrium sp. Bioresour Technol. 2024;394:130250.

    Article  CAS  PubMed  Google Scholar 

  56. Rahman A, Nastiti KD, Prihantini NB, Oktaufik MAM, Indrijarso S, Levison AK, et al. Multi-objective genetic algorithm optimization of energy efficiency and biomass concentration of Synechococcus HS-9 cultivation for third-generation biodiesel feedstock. Case Stud Chem Environ Eng. 2024;9:100614.

    Article  CAS  Google Scholar 

  57. Ma Y, Noreña-Caro DA, Adams AJ, Brentzel TB, Romagnoli JA, Benton MG. Machine-learning-based simulation and fed-batch control of cyanobacterial-phycocyanin production in Plectonema by artificial neural network and deep reinforcement learning. Comput Chem Eng. 2020;142:107016.

    Article  CAS  Google Scholar 

  58. Ereifej KI, Feng H, Rababah TM, Tashtoush SH, Al-U’datt MH, Gammoh S, et al. Effect of extractant and temperature on phenolic compounds and antioxidant activity of selected spices. Food Nutr Sci. 2016;07:362–70.

    CAS  Google Scholar 

  59. Metwally RA, El-Sersy NA, El Sikaily A, Sabry SA, Ghozlan HA. Optimization and multiple in vitro activity potentials of carotenoids from marine Kocuria sp. RAM1. Sci Rep. 2022;12:1–19.

    Article  Google Scholar 

  60. Pyo JC, Park LJ, Pachepsky Y, Baek SS, Kim K, Cho KH. Using convolutional neural network for predicting cyanobacteria concentrations in river water. Water Res. 2020;186:116349.

    Article  CAS  PubMed  Google Scholar 

  61. Wang C, Yang Y, Hou J, Wang P, Miao L, Wang X, et al. Optimization of cyanobacterial harvesting and extracellular organic matter removal utilizing magnetic nanoparticles and response surface methodology: A comparative study. Algal Res. 2020;45:101756.

    Article  Google Scholar 

  62. Sintra TE, Bagagem SS, Ghazizadeh Ahsaie F, Fernandes A, Martins M, Macário IPE, et al. Sequential recovery of C-phycocyanin and chlorophylls from Anabaena cylindrica. Sep Purif Technol. 2021;255:117538.

    Article  CAS  Google Scholar 

  63. Guendouzi S, Benmati M, Bounabi H, Vicente Carbajosa J. Application of response surface methodology coupled with artificial neural network and genetic algorithm to model and optimize symbiotic interactions between Chlorella vulgaris and Stutzerimonas stutzeri strain J3BG for chlorophyll accumulation. Bioresour Technol. 2024;394:130148.

    Article  CAS  PubMed  Google Scholar 

  64. Mehra A, Jutur PP. Application of response surface methodology (RSM) for optimizing biomass production in Nannochloropsis oculata UTEX 2164. J Appl Phycol. 2022;34:1893–907.

    Article  CAS  Google Scholar 

Download references

Acknowledgements

NB acknowledges Non-NET Fellowship (Grant No. R/Dev./Sch./UGC-Non-NET Fello./2022-23/52516) for financial support. PS acknowledges the Science & Engineering Research Board (SERB) Department of Science and Technology, GoI (Grant No. CRG/2021/001206) and the ’Faculty Incentive Grant’ by Institute of Eminence (IoE) Scheme by BHU, Varanasi (Letter No R/Dev/ D/IoE/Seed & Incentive/2022-23/50024). The authors acknowledge the Raja Jwala Prasad Post-Doctoral Fellowship under Institute of Eminence Scheme of Banaras Hindu University, Varanasi, India (No SRICC/RJP-PDF/2023-23/5081). Authors duly acknowledge Department of Science and Technology, Ministry of Science and Technology, New Delhi, Govt. of India, through FIST Grant -Level B, (Grant No. SR/FST/LS-I/2024/1375) for providing infrastructural facilities. NB also acknowledges valuable minor editing and suggestions by Dr. Megha Sailwal during preparation of the manuscript.

Funding

Not applicable.

Author information

Authors and Affiliations

  1. Enzyme Technology and Protein Bioinformatics Laboratory, School of Biotechnology, Institute of Science, Banaras Hindu University, Varanasi, 221005, India

    Namrata Bhagat, Guddu Kumar Gupta & Pratyoosh Shukla

  2. TERI Deakin Nanobiotechnology Centre, Sustainable Agriculture Division, The Energy and Resources Institute, New Delhi, India

    Amritpreet Kaur Minhas

  3. Department of Mechanical Engineering, University Institute of Engineering and Technology (UIET), Maharshi Dayanand University, Rohtak, Haryana, 124001, India

    Deepak Chhabra

Authors
  1. Namrata Bhagat
    View author publications

    You can also search for this author inPubMed Google Scholar

  2. Guddu Kumar Gupta
    View author publications

    You can also search for this author inPubMed Google Scholar

  3. Amritpreet Kaur Minhas
    View author publications

    You can also search for this author inPubMed Google Scholar

  4. Deepak Chhabra
    View author publications

    You can also search for this author inPubMed Google Scholar

  5. Pratyoosh Shukla
    View author publications

    You can also search for this author inPubMed Google Scholar

Contributions

PS conceptualized the manuscript. NB and GKG wrote the first draft of the manuscript. PS, DC and AKM reviewed and edited the manuscript. All the authors read and approved the final manuscript.

Corresponding author

Correspondence to Pratyoosh Shukla.

Ethics declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Additional information

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Bhagat, N., Gupta, G.K., Minhas, A.K. et al. Artificial Neural Network - Multi-Objective Genetic Algorithm based optimization for the enhanced pigment accumulation in Synechocystis sp. PCC 6803. BMC Biotechnol 25, 23 (2025). https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s12896-025-00955-9

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s12896-025-00955-9

Keywords

BMC Biotechnology

ISSN: 1472-6750

Contact us