**Preface to "Advances in Microbial Fermentation Processes"**

Fermentation processes are under the spotlight of scientific research in order to improve the quantitative and qualitative properties of the final products. In the food industry, microbial-based fermentation has traditionally been used to obtain edible foods and beverages denoted by extended shelf life and relevant nutritional properties. Furthermore, numerous helpful microorganisms are able to prevent pathogens/spoilers growing and to inactivate undesirable compounds, such as biogenic amines and mycotoxins. Fermented foods can enhance human health by interactions with live microbes (probiotic effect) or indirectly, thanks to the ingestion of microbial metabolites of fermentative origin (biogenic effect).

An incessant investigation concerning microbial diversity is underway, in order to describe and exploit innovative microbial-based biotechnological approaches for the utilization of novel foodstuff to address the current worldwide food crisis. Moreover, numerous micro-organisms have been suggested as cell factories for the synthesis of different desired compounds such as antimicrobial, antioxidants, vitamins and other bioactive molecules, and for use as initial substrate different agro-industrial wastes.

This book, "Advances in Microbial Fermentation Processes", collects the accounts of different investigations concerning the study and the application of new fermentation approaches mediated by microorganisms of industrial interest. In particular, the chapters include innovative studies about the microbial production of valuable compounds: penicillin production by Penicillium chrysogenum under different physiological conditions; the synthesis of GABA using purified recombinant GAD from L. plantarum; the antibiotic biosynthesis in S. clavuligerus strains; and medium-chain fatty acids by using both pure cultures and mixed microbial communities.

Different studies are also reported that investigate the roles of volatile compounds associated with ascomycete/bacteria interaction in fighting plant pathogens, and improving bread and wine quality. Novelties in the microbial-mediated production of a fermented milk-derived food to promote growth in stunted children and of traditional meat-derived foods are also described here. Two interesting chapters show innovative results obtained by the assessment of novel protocols for the production of Saccharomyces cerevisiae and Kluyveromyces marxianus biomasses. The assessment of biodiversity of microbial communities in whole-plant corn silages is included in two different chapters.

This volume contributes to the development of knowledge regarding microbial fermentation processes, by describing the newest applications for the exploitation of microorganism biodiversity in different biotechnological fields.

> **Maria Tufariello and Francesco Grieco** *Editors*

### *Editorial* **Advances in Microbial Fermentation Processes**

**Maria Tufariello \* and Francesco Grieco \***

National Research Council—Institute of Sciences of Food Production (ISPA), Via Prov.le, Lecce-Monteroni, 73100 Lecce, Italy

**\*** Correspondence: maria.tufariello@ispa.cnr.it (M.T.); francesco.grieco@ispa.cnr.it (F.G.); Tel.: +39-083-2422612 (M.T.)

In the food sector, fermentation processes have been the object of great interest in regard to enhancing the yield, the quality, and the safety of the final product. Microbial fermentation has been traditionally used to produce foods denoted by a prolonged shelf life and digestibility. The benefits extended to human health by fermented foods are expressed either directly through the interactions of ingested live microorganisms with the host (probiotic effect) or indirectly as the result of the ingestion of microbial metabolites synthesized during fermentation (biogenic effect). Moreover, several beneficial microbes can inhibit pathogens/spoilers growth and degrade toxins. Several novel microbial-based biotechnological solutions have been recorded and continuous explorations of microbial diversity are being carried out worldwide. In addition, most recently, fermentation has been considered a sustainable approach for maximizing the utilization of bio-resources to address the recent global food crisis. For example, several microbial-based bioconversions have been proposed for the production of enzymes, vitamins, antioxidants, biofuels, feeds, antimicrobial molecules, and other bioactive chemicals, also exploiting agro-industrial wastes [1–6].

The Special Issue "*Advances in Microbial Fermentation Processes*" covers eleven contributes: eleven original research papers and two reviews. As guest editors, we briefly report an overview of these contributions.

Wang et al. [7] investigated, through quantitative metabolomics and a stoichiometric analysis, the role of the trehalose metabolism in the *Penicillium chrysogenum* strain. The authors showed the key role of the intact trehalose metabolism in ensuring penicillin production in the *P. chrysogenum* strain under both steady state and dynamic conditions.

Helmyati and collaborators [8] described an innovation food-based approach to address the stunting problem. They evaluated the ability of a symbiotic milk enriched with iron and zinc and fermented with *Lactobacillus plantarum* to promote growth in stunted children, obtaining excellent results. The investigation of Yogeswara et al. [9] demonstrated a significant increase in the enzymatic synthesis of GABA using purified recombinant GAD from *L. plantarum* FNCC 260. In their original paper, Li and coworkers [10] focused on the activity of volatile compounds produced by Bacillus velezensis CT32 on *Verticillium dahlia* and *Fusarium oxysporum* responsible for strawberry vascular wilt. This study highlighted the key role of some volatile compounds as a biofumigant for the management of vascular wilt pathogens.

The novel cell-level Fed-Batch (FBC) technology for the high-cell-density cultivation of *Saccharomyces cerevisiae* was proposed by Malairuang et al. [11] with a clear illustration of the principle of operation, the potential dextrin substrate, and the mechanism of substrate utilization to regulate FBC, FBC kinetics and material balances through a bioreactor design and scale-up. Yepes-García and coworkers [12] provided an important contribution to the knowledge of antibiotic biosynthesis in the *Streptomyces* genus by studying the relationship between *S. clavuligerus* ATCC 27064 morphology and CA biosynthesis. An interesting contribution of the influence of different smoking techniques on the development of polycyclic aromatic hydrocarbons (PAH) in traditional dry sausage was presented by

**Citation:** Tufariello, M.; Grieco, F. Advances in Microbial Fermentation Processes. *Processes* **2021**, *9*, 1371. https://doi.org/10.3390/pr9081371

Received: 3 August 2021 Accepted: 4 August 2021 Published: 5 August 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

Mastanjevi´c et al. [13]. These authors showed the crucial role of the smoking method in the formation of PAHs revealing that collagen samples presented significantly lower values than samples created with traditional gut. Malairuang et al. [14] selected a *Kluyveromyces marxianus* strain for single-step ethanol fermentation, also to establish a practical approach to produce a high-cell-density yeast biomass by an intensive multiple sequential batch simultaneous saccharification and cultivation.

An interesting evaluation of the differences in microbial communities, metabolites, and the aerobic stability between whole-plant corn silages from different areas of Inner Mongolia in North China has been assessed by Wang et al. [15]. Moreover, the same authors investigated the variation of bacterial dynamics during the fermentation process in whole-plant corn silages processed in Heilongjiang, Inner Mongolia and Shanxi of North China [16].

Concerning the review papers, both contributors focused on two interesting topics.

Stamatopoulou et al. summarized the state-of-the-art-concerning Medium-Chain Fatty Acids (MCFA) by using both pure cultures and mixed microbial communities, highlighting future perspectives to improve MCFA production from complex feedstocks [17]. Pati and collaborators [18] reviewed the aspects related to the quantitative analysis of volatile compounds in wines application by HS-SPME-GC/MS, in particular discussing the optimization approaches in the method development stage and the critical aspects related to quantification methods.

This collection contributed to improve the knowledge on the microbial-based fermentation approaches and on the latest innovative application for promoting and monitoring bacterial action in different biotechnological fields.

**Author Contributions:** Writing—original draft preparation, review and editing, M.T. and F.G. Both authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** None.

**Acknowledgments:** This work was partially supported by the Apulia Region projects: "Innovazione nella tradizione: tecnologie innovative per esaltare le qualità dei vini autoctoni spumante della murgia barese—INVISPUBA" (P.S.R. Puglia 2014/2020 -Misura 16.2); "Birra: dal campo al boccale— BEˆ2R" (P.S.R. Puglia 2014/2020—↑Misura 16.2). We would like to thank Giovanni Colella, Domenico Genchi, and Vittorio Falco of the Institute of Sciences of Food Production—CNR—for their skilled technical support provided during the realization of this work.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


## *Article* **Impact of Altered Trehalose Metabolism on Physiological Response of** *Penicillium chrysogenum* **Chemostat Cultures during Industrially Relevant Rapid Feast/Famine Conditions**

**Xinxin Wang, Jiachen Zhao, Jianye Xia \*, Guan Wang \*, Ju Chu and Yingping Zhuang**

State Key Laboratory of Bioreactor Engineering, East China University of Science and Technology (ECUST), Shanghai 200237, China; m18217787017@163.com (X.W.); jczhao@mail.ecust.edu.cn (J.Z.); juchu@ecust.edu.cn (J.C.); ypzhuang@ecust.edu.cn (Y.Z.)

**\*** Correspondence: jyxia@ecust.edu.cn (J.X.); guanwang@ecust.edu.cn (G.W.); Tel.: +86-021-64250719 (G.W.)

**Abstract:** Due to insufficient mass transfer and mixing issues, cells in the industrial-scale bioreactor are often forced to experience glucose feast/famine cycles, mostly resulting in reduced commercial metrics (titer, yield and productivity). Trehalose cycling has been confirmed as a double-edged sword in the *Penicillium chrysogenum* strain, which facilitates the maintenance of a metabolically balanced state, but it consumes extra amounts of the ATP responsible for the repeated breakdown and formation of trehalose molecules in response to extracellular glucose perturbations. This loss of ATP would be in competition with the high ATP-demanding penicillin biosynthesis. In this work, the role of trehalose metabolism was further explored under industrially relevant conditions by cultivating a high-yielding *Penicillium chrysogenum* strain, and the derived trehalose-null strains in the glucose-limited chemostat system where the glucose feast/famine condition was imposed. This dynamic feast/famine regime with a block-wise feed/no feed regime (36 s on, 324 s off) allows one to generate repetitive cycles of moderate changes in glucose availability. The results obtained using quantitative metabolomics and stoichiometric analysis revealed that the intact trehalose metabolism is vitally important for maintaining penicillin production capacity in the *Penicillium chrysogenum* strain under both steady state and dynamic conditions. Additionally, cells lacking such a key metabolic regulator would become more sensitive to industrially relevant conditions, and are more able to sustain metabolic rearrangements, which manifests in the shrinkage of the central metabolite pool size and the formation of ATP-consuming futile cycles.

**Keywords:** feast/famine conditions; industrial-scale bioreactor; metabolomics; metabolic response; penicillin; *Penicillium chrysogenum*; scale-down

#### **1. Introduction**

The filamentous fungus *Penicillium chrysogenum* has long been explored for its production of β-lactam antibiotics (e.g., penicillin G, cephalosporin C), and the potential of the biosynthetic gene clusters in *Penicillium* species has recently been revisited, revealing that it would be a promising cell factory for the production of a series of secondary metabolites and natural products [1,2]. Bioprocess scale-up is the critical step for the commercialization of biotech innovations. Nonetheless, a loss of production capacity in terms of either titer, yield or productivity, or combinations thereof, has often been observed during the process scale-up. Although the interplay between cell systems and their production conditions can be measured, the underlying mechanism is partially unknown, which is, however, seldom accounted for during lab-scale research and development [3,4]. Representatively, the nonideal mixing and mass transfer limitations at the large scale in most cases are not rigorously considered in lab-scale designs, and thus the outcome of the environmental impacts cannot match the reality at the large scale [5,6]. In industrial settings, the environmental gradients, such as those of substrate, dissolved oxygen and pH, caused by insufficient mixing and

**Citation:** Wang, X.; Zhao, J.; Xia, J.; Wang, G.; Chu, J.; Zhuang, Y. Impact of Altered Trehalose Metabolism on Physiological Response of *Penicillium chrysogenum* Chemostat Cultures during Industrially Relevant Rapid Feast/Famine Conditions. *Processes* **2021**, *9*, 118. https://doi.org/ 10.3390/pr9010118

Received: 17 December 2020 Accepted: 5 January 2021 Published: 7 January 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

mass transfer restrictions, and of the shear force caused by the impellers, often exert a negative impact on the resulting commercial indicators (i.e., titer, yield and productivity) and thus the economic benefits [7–9].

Although there are many metabolic models being developed to characterize, predict, and evaluate the growth and production dynamics of cell factories in the biological system, a huge gap between laboratory-scale research and industrial applications still exists because of the limited knowledge about the intracellular dynamics under large-scale production limitations [4]. Increasing evidence has shown that steady-state data at the laboratory scale do not suffice to extract transient dynamics and potential regulators in response to a change in conditions, which are, however, very likely actionable at the production scale [10]. To address this, scale-down studies that take into account the environmental conditions experienced by the cells at the large scale have been carried out to evaluate the process performance and elucidate the mechanisms regulating the cellular metabolic flux [8,9,11–15]. As an example, in large-scale penicillin production by *Penicillium chrysogenum*, the cells are often repeatedly forced to experience high/low substrate concentrations (feast-famine cycles), which may partly account for the productivity loss. In an attempt to explain the mechanism behind this, de Jonge et al. (2011) [16] and Wang et al. (2018) [8] have adopted block-wise feeding schemes to simulate the substrate heterogeneity experienced by a highyielding *Penicillium chrysogenum* strain in glucose-limited chemostat cultures. The results revealed that under feast–famine cycles, the penicillin production capacity was halved, and the intracellular metabolite pools displayed fast dynamics. This efficient and robust control of intracellular metabolite concentration very likely indicated a potential key for the cell homeostasis. For instance, the turnover of intracellular carbohydrates, especially reduced sugars such as mannitol, arabitol, erythritol, as well as trehalose and glycogen, was drastically enhanced during rapid feast/famine conditions in *Penicillium chrysogenum*, which, according to metabolic flux analysis, accounts for about 52% of the gap in the ATP balance, and might partly explain the productivity loss in this scenario [17,18].

Among these ATP-consuming cycles, there has been a focus on the physiological role of transient trehalose futile cycling in the cellular phenotype because trehalose has been proven to harbor a multitude of functions, such as energy and carbon reserves, structural components, and dynamic regulators [19,20]. It has long been known that a functional trehalose pathway is vitally important for *Saccharomyces cerevisiae* grown on rapidly fermentable sugars [21], e.g., glucose, and a sudden shift from glucose-limiting to glucose-excess conditions would lead to growth arrest, which has been linked to the autocatalytic design of the pathway. Additionally, either the unregulated influx of glucose or insufficient phosphate availability have accounted for the appearance of this state [22]. Very recently, van Heerden et al. (2014) reported that trehalose metabolism constitutes a futile cycling that would facilitate metabolic balance via maintaining a proper inorganic phosphate level inside the model eukaryote *Saccharomyces cerevisiae* cell, in order to cope with a sudden glucose availability [23]. In our previous study, as shown in Figure 1, the trehalose synthetic pathway was (partly) blocked by knocking out either the gene *tps1* (encoding trehalose-6-phosphate synthase) or the gene *tps2* (encoding trehalose-6 phosphate phosphatase), and we have concluded that in steady-state glucose-limited chemostat cultures, the intact trehalose pathway plays an important role in metabolic regulation and is instrumental in maintaining a higher penicillin production capacity [24]. Nonetheless, as noted above, it is much more pertinent to evaluate the effect of an altered trehalose metabolism on process performance under industrially relevant conditions, e.g., feast/famine conditions in representative scale-down simulators.

**Figure 1.** An overview of trehalose cycling in *Penicillium chrysogenum*. Tre6P is formed through the transfer of a glucosyl residue from uridine-diphospho-glucose (UDP-glucose) to G6P, which is catalyzed by Tre6P synthase (encoded by *tps1*). Trehalose is then produced through the dephosphorylation of Tre6P, which is catalyzed by Tre6P phosphatase (encoded by *tps2*). Trehalose can be exported outside the cell, and it can be degraded into glucose via extracellular trehalase. Tre: trehalose; Glc\_ext: extracellular glucose; Tre\_ext: extracellular trehalose; G6P: glucose-6-phosphate; Tre6P: trehalose-6-phosphate.

In this study, a high-producing industrial strain of *P. chrysogenum* Wisconsin 54-1255, and the derived strains, *P. chrysogenum* Δ*tps1* (lacking trehalose-6-phosphate synthase) and *P. chrysogenum* Δ*tps2* (lacking trehalose-6-phosphate phosphatase), grown in glucoselimited chemostats at the average dilution rate of 0.05 hr−1, were assessed under two substrate availability conditions by performing either a continuous or a block-wise feeding scheme. A systems approach using fluxomics (stoichiometry) and metabolomics for both conditions was for the first time carried out. Overall, we aim at the systematic metabolic characterization and the identification of the potential metabolic role of the trehalose metabolism under industrially relevant conditions.

#### **2. Materials and Methods**

#### *2.1. Strains*

The *Penicillium chrysogenum* Wisconsin 54-1255 was purchased from ATCC, while the *P. chrysogenum-*Δ*tps1* and *P. chrysogenum-*Δ*tps2* were constructed using the *Agrobacterium* transformation method in our laboratory. For construction details, please refer to Wang et al. (2019) [24]. Fungal spores were prepared on potato dextrose agar (PDA) medium and a concentrated spore suspension was aliquoted and conserved in physiological salt solution (0.9% (*w*/*v*) NaCl in demineralized water) at −80 ◦C. A spore suspension inoculation method was used for all experiments and the spore suspension was prepared to ensure the final spore concentration in the bioreactor at about 1 × <sup>10</sup>6/mL after the inoculation, as described previously [24].

#### *2.2. Media*

The medium for the batch phase and chemostat cultivation contained the same components (per kg of medium): 16.5 g C6H12O6·H2O, 5 g (NH4)2SO4, 1 g KH2PO4, 0.5 g MgSO4·7H2O, 2 mL trace elements, 1ml antifoaming agent. The trace element composition (per kg of deionized water) was 75 g Na2EDTA·2H2O, 10 g ZnSO4·7H2O,10 g MnSO4·1H2O, 20 g FeSO4·7H2O, 2.5 g CaCl2·2H2O, 2.5 g CuSO4·5H2O. The phenylacetic acid (PAA) concentrations in the batch and the chemostat media were supplied at 0.4085 g/kg and 0.68 g/kg, respectively, which were neither limiting nor toxic for cell growth throughout the cultivation process [25]. The preparation and sterilization of the cultivation medium have been described previously [26]. Briefly, the glucose solution and the PAA-containing salt solution were prepared separately and autoclaved at 110 ◦C and 121 ◦C for 40 min and 30 min, respectively. The PAA was dissolved in a KOH solution, with a PAA:KOH molar ratio of 1:1.2.

#### *2.3. Chemostat Cultivation*

The chemostat cultivation was performed the same as described previously [24]. Briefly, the chemostat system was based ona5L bioreactor (Shanghai Guoqiang Bioengineering Equipment Co., Ltd., Shanghai, China) with a 3 L working volume, followed by the depletion of the initial glucose in the batch phase. As shown in Figure 2, steady state conditions were ensured as the standard reference condition where the medium was continuously fed to the cultivation system; fast feast/famine cycles were initiated via an intermittent feeding regime, which was imposed on the culture through the cultivation. An on/off feeding was applied with a cycle time of 360 s, and the feed pump was precisely controlled by a timer with an algorithm, switching it on every first 36 s of the cycle. During the feeding interval, the pump speed was set 10 times higher than that under reference conditions to keep the average glucose feeding rate of the intermittently fed cultures the same as that of the control chemostats.

All aerobic chemostat cultivations were controlled at the average dilution rate of 0.05 h−1, 2 L/min, and 0.5 bar overpressure. Effluent was removed discontinuously into an effluent vessel if the broth weight exceeded 3 kg by means of a weight-controlled sensor controlling (on/off) a peristaltic pump. Dissolved oxygen tension (DOT) was monitored in situ with an oxygen probe (Mettler-Toledo, Greifensee, Switzerland). The pH of the culture system was monitored using a sterilizable pH probe (Mettler-Toledo, Greifensee, Switzerland) mounted in the bioreactor and was automatically maintained at 6.5 by adding 4M NaOH. The offgas oxygen and carbon dioxide levels were real-time monitored by offgas mass spectrometry (MAX300-LG, Extrel, Pittsburgh, PA, USA). Each experiment was carried out at least in duplicate for biological relevance.

**Figure 2.** Scheme of the experimental setup. (**A**) Control chemostat cultures, where the culture is continuously fed and (**B**) Experimental chemostat culture, where an intermittent feeding regime (36 s feed, 324 s no feed) is applied.

#### *2.4. Cell Dry Weight*

The cell dry weight (CDW) was measured through the weight difference between empty and dried glass fiber filters (47 mm in diameter, 1 μm pore size, type A/E; Pall Corporation, East Hills, NY, USA) with biomass. An amount of 15 mL of broth was withdrawn and divided into three portions for the measurement of the CDW. For each CDW sample, 5 mL of the broth was filtered, and the cell cake was washed twice with 10 mL of demineralized water and dried at 70 ◦C for 24 h. The biomass-containing filters were cooled to room temperature in a desiccator before weighing.

#### *2.5. Rapid Sampling, Quenching and Metabolite Extraction*

Under the reference steady state conditions, samples were taken at each residence time. Under fast feast/famine conditions, the continuous rapid sampling of the broth for the measurement of intracellular metabolites was carried out after 100 h of intermittent feeding at 0, 8, 16, 26, 36, 50, 70, 90, 110, 145, 180, 200, 220, 240, 260, 280, 320 and 350 s within a complete 360 s feeding cycle.

For extracellular sampling, the cold steel-bead method combined with liquid nitrogen was used for the fast filtration and quenching of extracellular enzyme activities, as described previously [27]. For intracellular sampling, about 1 mL of broth was taken from the bioreactor into a tube containing the quenching solution (−27.5 ◦C, 40% *V*/*V* aqueous methanol). To ensure fast sampling and rapid quenching within a second to circumvent as much as possible the change of intracellular metabolites, a customized fast sampling device was developed. The sampling device consists of three normally closed electric pinch valves controlled by a programmed single-chip microcomputer (SCM). The sequential opening and closing of each valve and the combinations thereof can allow sampling with 1 mL of broth within half a second. For detailed sampling procedures, please refer to Li et al. (2018) [28]. Fast filtration and a modified cold washing method were used for the rapid and effective removal of all compounds present outside the cells. In this study, a previously well-established rapid sampling, quenching as well as the follow-up metabolite extraction protocol was followed, as described previously [10].

#### *2.6. Analytical Procedures*

Samples for intracellular amino acids, sugar phosphates, organic acids and sugar alcohols quantifications were analyzed using gas chromatography–mass spectrometry (GC–MS) (7890A GC coupled to 5975C MSD; Agilent Technologies, Santa Clara, CA, USA). The analytical procedure undertaken as per de Jonge et al. [16] with some minor modifications in the column and temperature gradients. For more specific settings, please refer to Liu et al. [29].

Concentrations of PAA, Penicillin G and other byproducts in the penicillin biosynthetic pathway were measured with an isocratic reversed-phase high performance liquid chromatography (HPLC) method, which was equipped with an Agilent Zorbax SB-C18 reversed-phase column (150 mm× 4.6 mm ID, 5 μm). The mobile phase consisted of 0.44 g KH2PO4 per liter in the acetonitrile/water solution (65/35, *V*/*V*). The sample injection volume, the detection wave length, the flow rate and the column temperature were 5 μL, 214 nm, 1.5 mL/min and 25 ◦C, respectively [27].

#### *2.7. Calculation Methods and Data Reconciliation*

The specific rates, such as *qs*, *qCO*<sup>2</sup> , *qO*<sup>2</sup> and μ, were calculated using respective mass balances [9], and then were reconciled using the approach of Verheijen based on elemental, charge and degree of reduction balances [30]. The elemental biomass composition and molar weight of 28.05 gDW·Cmol−<sup>1</sup> were taken from de Jonge et al. [16]. They were assumed to be constant for all conditions.

#### **3. Results and Discussion**

#### *3.1. General Observations*

To investigate the effect of the altered trehalose metabolism on the physiological response of *Penicillium chrysogenum* under industrially relevant conditions, i.e., dynamic substrate availability in this study, the strains were cultivated in the glucose-limited chemostat mode where rapid feast/famine conditions were imposed via a block-wise feeding strategy. All chemostat cultivations were preceded by a batch phase wherein the strains grew exponentially until all glucose was depleted. The end of batch was determined when the online oxygen uptake rate (OUR) and carbon evolution rate (CER) went down while the dissolved oxygen (DO) level and pH went up. Due to the carbon catabolite repression, no penicillin was produced in the batch phase. After about five residence times (~100 h), all chemostat cultures reached a dynamic steady state (Figure 3). Additionally, the measured time series of the concentrations of biomass, PAA, penicillin-G (PenG) (Figure 3), the main byproducts (Figure 4) associated with penicillin biosynthesis, and the biomass specific rates (Figure 5) show that the chemostat cultivations were well reproducible.

**Figure 3.** Measured concentrations of (**A**) biomass, (**B**) PAA and (**C**) PenG during rapid feast/famine conditions throughout the chemostat phase. The symbols in black, red and green denote the results obtained from *Penicillium chrysogenum* Wisconsin 54-1255, *Penicillium chrysogenum* Δ*tps1* and *Penicillium chrysogenum* Δ*tps2*, respectively. Time 0 represents the start-up of chemostat cultivation. Abbreviation: PAA, phenylacetic acid; PenG, penicillin-G.

**Figure 4.** Measured concentrations of (**A**) 6-APA, (**B**) OPC and (**C**) o-OH-PAA during rapid feast/famine conditions throughout the chemostat phase. *Penicillium chrysogenum* Δ*tps1* and *Penicillium chrysogenum* Δ*tps2*, respectively. Time 0 represents the start-up of chemostat cultivation. Abbreviation: 6-APA, 6-aminopenicillanic acid; OPC, 6-oxopipeidine-2 carboxylic acid; o-OH-PAA, ortho-hydroxyphenyl acetic acid.

**Figure 5.** The biomass specific penicillin production rate during rapid feast/famine conditions throughout the chemostat phase. The symbols in black, red and green denote the results obtained from *Penicillium chrysogenum* Wisconsin 54-1255, *Penicillium chrysogenum* Δ*tps1* and *Penicillium chrysogenum* Δ*tps2*, respectively. Time 0 represents the start-up of chemostat cultivation.

In addition, Table 1 shows that the carbon and redox balances closed within 5% for all chemostat cultures, which suggests that marginal unknown compounds are formed during the cultivation.

**Table 1.** Comparison of the reconciled biomass-specific rates and relevant yields of different *Penicillium chrysogenum* strains on glucose obtained from feast/famine cultures at the average dilution rate of 0.05 h−<sup>1</sup> in the time range of 100–200 h of cultivation. Measurements are given as average ± standard deviation of at least two individual experiments. Significant test was conducted \* *p* < 0.05 vs. WT, by Student's *t*-test. Numbers marked with superscript a and b means significantly increased and decreased, respectively.


#### *3.2. Penicillin Production*

A schematic overview of the penicillin synthesis pathway, including the byproduct mentioned in this study, can be seen in Figure 6. It is apparent that the biomass-specific rates for penicillin production (*qPenG*) obtained for all cultivations reached a maximum value at about 100 h after the start-up of the chemostat phase, which suggests the typical behavior of the induction of the penicillin pathway enzymes because of the repression of the genes encoding these enzymes at high glucose concentrations in the batch phase [26]. Strikingly, the penicillin production capacity in terms of *qPenG* was reduced by about 30%, 74% and 78% for *P. chrysogenum* Wisconsin 54-1255, *P. chrysogenum*-Δ*tps1* and *P. chrysogenum*- Δ*tps2* strains under feast/famine conditions, respectively, relative to those under steady state conditions [24]. This can be anticipated because the cells under rapid feast/famine conditions will use multi-layered metabolic regulation mechanisms to maintain a balanced metabolic state at different omics levels (e.g., short-term stringent regulation, long-term repeatedly switched on/off of related genes, protein turnover, etc.), which is often at the expenditure of extra ATP and/or reducing equivalents [31–34]. Meanwhile, it was found that the absence of intact trehalose metabolism can aggravate the loss of the penicillin production capacity, which can be manifested by an almost 40% greater reduction in penicillin productivity for *P. chrysogenum*-Δ*tps1* and *P. chrysogenum*-Δ*tps2* strains under industrially relevant conditions as compared to the steady state conditions. Combining this with the findings from the steady state conditions as we reported elsewhere [24], it can therefore be concluded that trehalose plays an essential role in regulating penicillin production under both non-perturbed and perturbed conditions.

**Figure 6.** A schematic view of the penicillin biosynthesis pathway and the derived byproducts mentioned in this study. Precursors are L-α-amino adipic acid (αAAA), L-cysteine (Cys) and L-valine (Val). ACVS: L-α-(δ-aminoadipyl)-L-α-cysteinyl-D-α-valine synthase; IPNS: isopenicillin N synthase; AT: acyl-CoA:isopenicillin acyltransferase.

In spite of the reduced production performance under rapid feast/famine conditions, an interesting observation showed that the cells appeared to maintain their production capacity throughout the cultivation process (Figure 5), which was, however, not the case for the strains under steady state conditions, which unavoidably undergo strain degeneration (i.e., loss of production capacity) followed by the onset of the maximum *qPenG* value [16,35]. This result is very consistent with previous reports by de Jonge et al. [16] and Wang et al. [8], and also suggests that there must be an unknown regulatory mechanism contributing to this non-declining phenomenon.

#### *3.3. Stoichiometry*

The biomass-specific rates and relevant yields are provided in Table 1, and the results shows that the absence of intact trehalose metabolism gives rise to significant changes in strain process performance. Notably, during the chemostat phase, the biomass-specific carbon dioxide evolution rates (*qCO*<sup>2</sup> ) were increased by 55% and 48%, respectively, in *P. chrysogenum*-Δ*tps1* and *P. chrysogenum*-Δ*tps2* strains compared to the wild-type *P. chrysogenum* Wisconsin strain. This considerable increase in *qCO*<sup>2</sup> is mainly ascribed to the significant increase in specific glucose consumption rates (~90%) and the increase in PAA consumption rates (~6–7%). In the Wisconsin family strains, the reduced oxidative activity of a phenylacetate-oxidizing cytochrome P450 makes more PAA available for the production of PenG [36]. However, the PAA recoveries were not closed in the chemostat phase, leaving 90% or even more PAA catabolized towards the formation of o-OH-PAA, and further, to acetoacetate and fumarate, through the homogentisate pathway. In the engineered strains, more PAA was consumed while much less PenG and o-OH-PAA accumulated, which also pointed to the increased catabolism of PAA as the second carbon source (Table 1). In contrast with this, the PAA recoveries were very well closed during the early phase of the batch process before the initiation of the chemostat mode (Figure 7). In the batch phase, no penicillin was produced, and the summation of the residual PAA concentration and the formed o-OH-PAA concentration was almost equal to the initial PAA concentration in the bioreactor. Nonetheless, the PAA concentration was gradually decreased during the late phase of the batch fermentation. Meanwhile, we observed that the glucose was rapidly consumed and reached a threshold value at the late stage, which then coincided with the PAA consumption. Combining this, the results indicated that the catabolism of PAA as the carbon source can only be activated under carbon-limited conditions rather than carbon-rich scenarios, which was proven to be controlled via carbon catabolite repression.

**Figure 7.** Measured concentrations of PAA, o-OH-PAA and glucose during the batch process. The dash line represents the initial PAA concentration in the bioreactor. The symbols in black, red and green denote the results obtained from *Penicillium chrysogenum* Wisconsin 54-1255, *Penicillium chrysogenum*Δ*tps1* and *Penicillium chrysogenum* Δ*tps2*, respectively. Time 0 represents the start-up of chemostat cultivation.

The other byproduct associated with penicillin production, such as 6APA and OPC, were also measured in this work. The results showed that as compared to the wild-type strain, the specific formation rates for 6APA and OPC were significantly increased and decreased, respectively. The formation of OPC was because of the spontaneous cyclization of one of the precursor amino acids, α-aminoadipic acid (αAAA), and thus the decreased excretion rate of OPC might be ascribed to the lower precursor demand for penicillin production, since the αAAA was not actually consumed during the PenG production. The five-fold lower *qPenG* values in the engineered strains seem to corroborate this (Table 1). It has been reported that PenG can be formed via both isopenicillin acyltransferase (IAT) and acyl-CoA: 6APA acyltransferase (AAT), the two activities of the key enzyme, acyl-CoA: isopenicillin acyltransferase (AT), in the PenG biosynthesis. Besides this, the AT also has the activity of isopenicillin amidohydrolase (IAH), which converts isopenicillin N (IPN) into 6APA [37]. Recently, Deshmukh et al. estimated the enzyme capacity ratio of IAH:IAT:AAT, which was 250:1:6, and the results showed that the IAH-AAT route is a high-capacity but low-affinity route, while the IAT route is a low-capacity but high-affinity route [38]. In the present study, it was observed that the excretion of 6APA was about four-fold higher outside the cells in the engineered strains (Table 1), which makes it reasonable that IPN (not measured) and 6APA were the main end products when the productivity of PenG was reduced by 80%. In addition, this result also indicated that the IAH activity was less affected, while the IAT and the AAT activities become much more sensitive to rapid feast/famine conditions in the engineered strains than in the Wisconsin 54-1255 strain.

The relevant yields, i.e., *YX*/*<sup>S</sup>* and *YO*/*S*, were also calculated in Table 1. Consistent with previous studies [7], the biomass yields under feast/famine conditions were reduced

by over 20% in the engineered strain as compared to the steady state values [24]. However, the biomass yield in the Wisconsin strain was not significantly varied between the conditions, which is consistent with the results of a high-yielding strain derived by de Jonge et al. [16]. This result showed that although trehalose takes up a small amount of the cell composition [24], it may be involved in maintaining the constant biomass yield under industrially relevant conditions. Meanwhile, the *YO*/*<sup>S</sup>* value indicates the efficiency of glucose combustion by the cells. We have observed that as compared to the Wisconsin strain, the absence of trehalose metabolism in the *Penicillium chrysogenum* cells slightly increases the energy efficiency under the steady state conditions, which was reflected by the lower *YO*/*<sup>S</sup>* values [24]. Contrary to the results under the steady state conditions, the *YO*/*<sup>S</sup>* values for the *P. chrysogenum-*Δ*tps1* and *P. chrysogenum-*Δ*tps2* strains became 46% and 37% higher under dynamic conditions. However, the energy efficiency of the Wisconsin strain did not significantly change between the conditions. The substrate consumption by the *P. chrysogenum* strain can be described by the well-known Herbert–Pirt relation, i.e., *qs* = *αμ* + *βqp* + *ms*, where the parameters α and β are the inverse values of the maximum biomass yield and product yield, respectively [39]. Under rapid feast/famine conditions, the *qs* values for the *P. chrysogenum*-Δ*tps1* and *P. chrysogenum*-Δ*tps2* strains became 30% and 25% higher as compared to the Wisconsin strain, while the *qp* values were much lowered. This result clearly indicated that the maintenance requirements for the engineered strains became 25–30% higher when the substrate distributed towards penicillin production was neglected (less than 0.5%). This increased maintenance for the engineered strain is very likely to be associated with the loss of the physiological role of trehalose. It has well been documented that trehalose is a multifunctional molecule, which can act as a carbon/energy reserve, as a stabilizer and protectant against adverse substances/environments, as a sensing compound and/or growth regulator, and as a structural component of the cell envelope [19]. More importantly, the flux branching off towards trehalose lies at the glucose-6-phosphate node, which is very close to the gate of glucose uptake, and hence the formation/mobilization of this compound can react immediately to the environmental perturbations, e.g., feast/famine conditions. For example, in a high-producing *P. chrysogenum* strain, de Jonge et al. showed that storage turnover is increased under dynamic cultivation conditions, and upon one cycle of repetitive glucose pulses, about 18% of the carbon entering the cell was recycled in the trehalose node [17]. Although this periodic formation and degradation of trehalose will generate extra ATP costs, the functional role of trehalose cycling can buffer the extracellular nutrient dynamics, and thus contribute to maintaining a balanced cellular state [23]. Based on this, we can hypothesize that the absence of trehalose could expose the cell to more severe intracellular dynamics, which often leads to forming futile cycles at the expense of extra ATP, and also could result in an increased protein turnover rate because the loss-of-function proteins would become higher, which may lead to one of the highest maintenance costs in trehalose-null *P. chrysogenum* strains [33].

#### *3.4. Intracellular Metabolites*

In addition to the above-mentioned comparison of the altered trehalose metabolism upon the physiological response of *Penicillium chrysogenum* strains from a macroscopic view, we also carried out a quantitative metabolomics study to investigate the change in the intracellular metabolite pools under industrially relevant conditions, i.e., the feast/famine setup. Moreover, the results from dynamic conditions were compared with those obtained under steady state conditions. As shown in Figures 8 and 9, sugar phosphates, organic acids, sugar alcohols as well as amino acids are measured. Obviously, the majority of the metabolites were significantly decreased under feast/famine conditions relative to steady state conditions. This phenomenon has already been observed in the high-yielding *P. chrysogenum* strain, DS 17690 [8,10], which is derived from the *P. chrysogenum* Wisconsin 54-1255 strain used in this study. This may be associated with metabolic adaptation, where the shrinkage of the metabolite pools can allow the cells to cope with rapid metabolite, flux and growth rate responses to changes in substrate availability. For instance, in prolonged aerobic, glucose-limited chemostat cultures of *Saccharomyces cerevisiae*, the reduced metabolite pools and a partial loss of capacity (e.g., ATP regeneration, glycolytic enzymes) have been reported [40,41]. Such a decrease in the overcapacities in cells can be accompanied by the faster turnover of the metabolite pools, which results in the optimized metabolic productivity at the cost of metabolic flexibility.

**Figure 8.** Intracellular amount of sugar phosphates, organic acids and sugar alcohols in *Penicillium chrysogenum* Wisconsin 54-1255, *Penicillium chrysogenum* Δ*tps1* and *Penicillium chrysogenum* Δ*tps2* strains under both steady state conditions and feast/famine conditions. Under steady state conditions, samples were taken at each residence time and datasets were collected from triplicate experiments, while under rapid feast/famine conditions, samples were rapidly taken within a feeding cycle time of 6 min twice for each experiment. The datasets are shown as average value ± standard deviation. The numbers of data points for the steady state conditions and feast/famine conditions are 30 and 32, respectively. A two-tail Student's *t*-test was conducted \* *p* < 0.05 (significant) and \*\*\*\* *p* < 0.01 (extremely significant) versus WT using GraphPad Prism 8. N.A., not available. ❚ WT steady state conditions; ❚ WT feast/famine conditions; ❚ Δtps1 steady state conditions; ❚ Δ*tps1* feast/famine conditions ❚ Δ*tps2* steady state conditions; ❚ Δ*tps2* feast/famine conditions. The results from steady state conditions were obtained from Wang et al. (2019) [24]. Abbreviations: G6P, glucose-6-phosphate; F6P, fructose-6-phosphate; 6PG, 6-phosphogluconate; S7P, sedoheptulose-7-phosphate; EMP, Embden–Meyerhof–Parnas; PPP, Pentose phosphate; TCA, tricarboxylic acid.

The above results indicated that under feast/famine conditions there might be a large increase in maintenance requirements in the engineered strains, which suggests the rearrangement of metabolism. Evidence for such a rearrangement can be derived by comparing the mass action ratios (MARs) of near equilibrium reactions in the central metabolism. In this study, the mass action ratios for F6P/G6P and malate/fumarate were calculated for *P. chrysogenum* Wisconsin 54-1255, *P. chrysogenum*-Δ*tps1* and *P. chrysogenum*-Δ*tps2* strains (Table 2), which were (0.25, 1.56), (0.25, 3.07) and (0.33,1.51) for steady state conditions, while they were (0.68, 5.18), (1.34, 4.95) and (1.02, 4.93) for feast/famine conditions, respectively.

**Figure 9.** Intracellular amounts of amino acids in *Penicillium chrysogenum* Wisconsin 54-1255, *Penicillium chrysogenum* Δ*tps1* and *Penicillium chrysogenum* Δ*tps2* strains under both steady state conditions and feast/famine conditions. Amino acids were classified into six families, which are derived from the same precursors in the central carbon metabolism. Under steady state conditions, samples were taken at each residence time and datasets were collected from triplicate experiments, while under rapid feast/famine conditions, samples were rapidly taken within a feeding cycle time of 6 min twice for each experiment. The datasets are shown as average value ± standard deviation. The numbers of data points for steady state conditions and feast/famine conditions are 30 and 36, respectively. A two-tailed Student's *t*-test was conducted \* *p* < 0.05 (significant) and \*\*\*\* *p* < 0.01 (extremely significant) vs. WT using GraphPad Prism 8. N.S., not significant; N.A., not available. ❚ WT steady state conditions; ❚ WT feast/famine conditions; ❚ Δ*tps1* steady state conditions; ❚ Δ*tps1* feast/famine conditions ❚ Δ*tps2* steady state conditions; ❚ Δ*tps2* feast/famine conditions. The results from steady state conditions were obtained from Wang et al. (2019) [24].

**Table 2.** Comparison of mass action ratios for phosphoglucose isomerase (PGI) and fumarase of different *Penicillium chrysogenum* strains obtained from both steady state conditions and feast/famine cultures at the average dilution rate of 0.05 h−<sup>1</sup> in the time range of 100–200 h of cultivation. Measurements are given as average <sup>±</sup> standard deviation of at least two individual experiments.


The significant increase in these MARs indicated that the flux through these key nodes is reduced or even reversed under feast/famine conditions. This has also been reported in our previous work [8], which shows that the intracellular metabolite pool was first increased and then decreased during the feast/famine cycle. The MARs can be increased above their equilibrium values during the famine phase [8]. Nonetheless, under feast/famine conditions, the *qs* values did not decrease but increased in the engineered strains (Table 1). Combining the above results, this indicated that there might be triggered an increased ATP-consuming futile cycling under feast/famine conditions through forming a cycling flux between the pentose phosphate (PP) pathway and the reversed Embden– Meyerhof–Parnas (EMP) pathway [9]. To corroborate this, the intracellular amounts of 6PG were significantly increased under feast/famine conditions, as compared to the steady state scenario. Furthermore, this futile cycling might become much more enhanced for the engineered *P. chrysogenum* strains lacking an intact trehalose storage pool and possessing a reduced pool size of other stored carbohydrates (Figure 8), e.g., mannitol, arabitol and erythritol, because upon the repetitive glucose pulses given by block-wise feeding there is no extra stored carbon pool to alleviate the perturbation of the whole metabolic system. As a result, more carbon and thus energy can be wasted in this futile cycling, which would lead to increased maintenance requirements and thus reduced biomass yield and productivity. In the future, this metabolic rearrangement should be confirmed by 13C metabolic flux analysis.

#### **4. Conclusions**

In the present study, the physiological role of the trehalose metabolism has been explored by cultivating *P. chrysogenum* Wisconsin 54-1255 (wild type), *P. chrysogenum*-Δ*tps1* and *P. chrysogenum*-Δ*tps2* strains in the chemostat cultivation where the glucose feast/famine cycles were imposed. The results clearly showed that the absence of intact trehalose metabolism gave rise to a loss of penicillin production capacity under dynamic conditions, which might be caused by the enhanced maintenance requirement. Additionally, this increased maintenance cost may be caused by triggering a reversed EMP–PPP futile cycle. Meanwhile, the shrinkage of the metabolite pools and the partial loss of enzyme capacity under rapid feast/famine cycles might contribute as a generic mechanism to the metabolic productivity. Furthermore, we could thus conclude that trehalose is indispensable to maintaining the balanced metabolic state and thus high penicillin production capacity under both steady state and feast/famine conditions. However, one should be aware that the detailed physiological role of trehalose in this industrial-relevant *P. chrysogenum* strain should be further investigated via more cell physiology studies, e.g., in more sophisticated and representative scale-down simulators.

**Author Contributions:** Conceptualization, G.W.; methodology, G.W., J.X; formal analysis, G.W., and X.W.; investigation, X.W., J.Z., G.W.; writing—original draft preparation, G.W.; writing—review and editing, G.W.; visualization, G.W., X.W; supervision, G.W., J.C., Y.Z.; project administration, G.W., Y.Z., J.X.; funding acquisition, G.W., Y.Z. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the National Natural Science Foundation of China grant number [31900073, 21978085], the Science and Technology Commission of Shanghai Municipality grant number [19ZR1413600] and the 111 Project grant number [B18022].

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** The data presented in this study are available on request from the corresponding author.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


## *Article* **Synbiotic Fermented Milk with Double Fortification (Fe-Zn) as a Strategy to Address Stunting: A Randomized Controlled Trial among Children under Five in Yogyakarta, Indonesia**

**Siti Helmyati 1,2,\*, Karina Muthia Shanti 3, Fahmi Tiara Sari 3, Martha Puspita Sari 3, Dominikus Raditya Atmaka 4, Rio Aditya Pratama 5, Maria Wigati 2, Setyo Utami Wisnusanti 1,2, Fatma Zuhrotun Nisa' 1,2 and Endang Sutriswati Rahayu <sup>6</sup>**


**Abstract:** Stunting is one of the public health problems that has yet to be solved in Indonesia. This study developed synbiotic fermented milk with iron and zinc fortification that was then tested in a clinical setting. The product was made from skimmed milk and fructooligosaccharides (FOS) and fermented with *Lactobacillus plantarum*. A sample of 94 stunted children under five years old were randomly assigned to intervention or control groups. The intervention group received doublefortified synbiotic milk, while the control group drank non-fortified milk. After three months, the number of normal children in both groups, according to weight- or height-for-age z-score category, was found to be increasing. However, the difference between the two groups was not significant (*p* > 0.05). The study suggests that fermented milk may have a good effect on child growth. Further research is needed to deepen the potency of synbiotic fermented milk for stunted children.

**Keywords:** children; double fortification; fermented milk; iron and zinc; stunting; synbiotic

#### **1. Introduction**

Developing countries are known for their complex public health problems, including stunting. Globally, it is estimated that 171 million children under five are stunted. Although this number lowered to 149 million children in 2019, stunting remains one of the serious global health challenges that need to be solved. This problem is also happening in Indonesia as one of the developing countries [1–3]. The prevalence of stunting among Indonesian children under five reached 30.8% in 2018. This situation is highly varied in each of the 34 provinces—for example, in East Nusa Tenggara, the stunting prevalence has reached more than 40% [4]. This alarming number emphasizes the need for innovative approaches to combat stunting [5,6].

Stunting is marked by diminished nutritional status and quality of life. Prendergast and Humphrey [7] mentioned that stunting in children is associated with morbidity and mortality, low physical and economic capacity, and an increased risk of metabolic disease in

**Citation:** Helmyati, S.; Shanti, K.M.; Sari, F.T.; Sari, M.P.; Atmaka, D.R.; Pratama, R.A.; Wigati, M.; Wisnusanti, S.U.; Nisa', F.Z.; Rahayu, E.S. Synbiotic Fermented Milk with Double Fortification (Fe-Zn) as a Strategy to Address Stunting: A Randomized Controlled Trial among Children under Five in Yogyakarta, Indonesia. *Processes* **2021**, *9*, 543. https://doi.org/10.3390/pr9030543

Academic Editors: Maria Tufariello and Francesco Grieco

Received: 29 January 2021 Accepted: 25 February 2021 Published: 19 March 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

adulthood. This is supported by a study in Indonesia that showed stunting was associated with disease incidence, family income, and parental education [8]. It is estimated that stunting in developing countries costs 13.5% of Gross Domestic Product (GDP) per capita. A one-point increase of child stunting led to a 0.4% reduction of GDP per capita [9].

Stunting is closely related to inadequate nutritional intake. According to Blaney et al. [10], more than 50% of Indonesian children have low energy, likely due to poor iron, zinc, and calcium intake per AKG (*Angka Kecukupan Gizi* or Indonesian Nutrition Adequacy Rate). These are essential minerals that play key roles in human growth [11–13]. The combination of iron and zinc can increase height by 1.1 cm above the standard growth of children [14]. This is in accordance with a meta-analysis by Liu et al. [15] which provided evidence of the potency of zinc supplementation to prevent stunting. Zinc supplementation given after birth can increase height by 0.23 cm and when given to children above 2 years old can increase their height by 1.37 cm. On the other hand, iron supplementation is known to prevent iron deficiency anemia. Larson et al. [16] suggested a strong association between iron supplementation and increased hemoglobin levels and mental development among children. Research on the benefits of zinc and iron combination intake on children's growth is scarce. It is strategic to conduct a clinical trial on the effect of iron and zinc consumption on stunted children in the form of food fortification.

Food fortification is a cost-effective effort to tackle the problem of micronutrient deficiencies [17,18]. Cost-effectiveness is generally defined as the affordable cost to achieve a certain outcome. In this article, it means the cost to avert one case of stunting among children under five. It can also be calculated as the cost per disability-adjusted life-year (DALY) saved [18]. Cost-effectiveness is related to other types of interventions. An analysis in 48 countries showed that food fortification would cost between USD 1 and USD 134 per DALY saved [19].

We planned to develop functional foods by fortifying fermented milk with iron and zinc. Fermented milk was chosen as the vehicle to increase its nutritional value. A metaanalysis by Matsuyama et al. [20] revealed that fortified milk could lower the risk of anemia but could not increase height. On the other hand, several studies about fermented milk development showed its positive effects on health [21–24]. A clinical trial with 494 children in Indonesia demonstrated beneficial effects in that the consumption of probiotics modestly increased weight and height [25]. Onubi et al. [26] suggested that probiotic consumption is beneficial to improving child growth in developing countries. The fermentation process in milk is also essential since it will be fortified with iron and zinc. A higher intake of iron is associated with various side effects including diarrhea, increasing pathogens, and other inflammatory diseases in the gut [27,28]. Lin et al. [29] mentioned that consumption of probiotic and prebiotic could ameliorate those side effects. The presence of pre- and probiotics in diets, called synbiotic foods, which are good to the gut microbiota balance, can reduce the likelihood of micronutrient utilization by pathogenic bacteria in the colon and increase nutrients absorption [30].

Fermented milk with double fortification has potential as a functional food to promote children's growth. Bearing the complex causes of stunting among children under five, the innovation of food development is essential to support stunting management programs in the country. We carried out this research and aimed to determine the effect of synbiotic fermented milk with double fortification on the height and nutritional status of stunted children in Indonesia. We expected this research to give a novel evidence-based intervention for Indonesia and the other low- and middle-income countries which face the same problem, stunting.

#### **2. Materials and Methods**

#### *2.1. Subjects and Design of the Study*

The study was conducted between May 2017 and October 2018 among under-5-yearolds in Seyegan District, Yogyakarta Province, Indonesia. Informed consent was obtained from the parents or guardians of the children. The Medical and Health Ethics Committee of

the Faculty of Medicine, Public Health and Nursing, Universitas Gadjah Mada gave ethical permission with the reference no. KE/FK/0640/EC/2017. The study is also registered in ClinicalTrial.gov no NCT03495401.

The study used a double-blind, randomized, controlled trial design. There were three main parts: screening, intervention, and a follow-up phase. Screening was conducted with inclusion and exclusion criteria (n = 212). The inclusion criteria were children aged between 2 and 5 years old and categorized as stunted, according to the height-for-age z-score (HAZ < −2 standard deviation). Exclusion criteria were one or more congenital abnormalities, chronic disease, presence of edema or weight-for-height z-score (WHZ) below −3SD, and allergy to milk, iron, or zinc supplements. There were 7 children who refused to be measured, 101 who did not meet the inclusion criteria, and 10 who refused to participate. The subjects were then randomly divided into the intervention (n = 47) and control groups (n = 47). In the intervention phase, the researchers gave synbiotic fermented milk with double fortification (Fe-Zn) to the intervention group and the same milk without fortification to the control group for three months. In the middle of the study, nine subjects from the intervention group and four from the control group dropped out. Anthropometric measurements were conducted twice, namely before and after intervention. Dietary intake was assessed eight times throughout the intervention phase. The number of subjects who remained through the entire study were 38 children from the intervention group and 43 from the control group. The baseline characteristics can be seen in Table 1.


**Table 1.** Baseline sociodemographic characteristics.


N/A, not available; <sup>a</sup> chi-squared test; <sup>b</sup> Fisher's exact test; significant if *p* < 0.05; intervention group, subjects given synbiotic fermented milk fortified with iron and zinc; control, subjects given synbiotic fermented milk without fortification.

#### *2.2. Production of Synbiotic Fermented Milk with Double Fortification (Fe-Zn)*

The researchers collaborated with CV Violla Foods to produce synbiotic fermented milk with and without fortification. Ingredients consisted of skim milk, sugar, probiotic *Lactobacillus plantarum* (*L. plantarum*), prebiotic fructooligosaccharides (FOS), iron fortificant (ferrous sulfate, Merck, Branchburg, NJ, USA), and zinc fortificant (zinc acetate, Merck). The skim milk (Lactona, Yogyakarta, Indonesia) and sugar (Gulaku, Lampung, Indonesia) were procured from the local public market in Yogyakarta, the probiotic was obtained from the Food and Nutrition Culture Collection at Universitas Gadjah Mada, and the prebiotic was from Beneo Orafti, Indonesia. Skim milk is commonly used as an ingredient for making fermentation products. Before we carried out this research, we conducted trials by comparing skim milk, whole milk, and ultra-high-temperature milk as the main ingredient. We concluded that using skim milk will have better consistency and sensory characteristics that the other two. In line with our trials, several studies mentioned the advantages of using skim milk including lower price and a positive relation with probiotic viability and fermented milk flavor [31–33].

The steps in the process to make synbiotic fermented milk were as follows: (1) dissolve sugar, skim milk, and add FOS and fortificants into water; (2) sterilize; (3) inoculate *L. plantarum*; and (4) incubate. CV Violla Foods had the authority to encode the milk, and the code was not known to the research team. The code was revealed only after the research ended. All types of milk were packed in 100 mL bottles and had the same color, flavor, and taste so that participants could not differentiate the products. Before the study began, we made sure that all subjects had received deworming medication to minimize the potential effects of worm infection. The participants consumed synbiotic fermented milk every day for three months. Considering the study subjects are children, we did not give rigid dietary rules but advised the parents to not change the dietary habits of their children during the study. Children were permitted to consume the milk at any meal each day. Each bottle of synbiotic fermented milk with double fortification contained 79.93 kcal energy, 2.26 g protein, 1.95 g fat, 13.67 g carbohydrate, 1.27 g crude fiber, 90 mg calcium, 2.26 mg iron, 1.22 mg zinc, and 3.23 × 108 CFU/mL *L. plantarum*. Synbiotic fermented milk without fortification contained 85.75 kcal energy, 2.05 g protein, 1.51 g fat, 16.44 g carbohydrate, 0.74 g crude fiber, 88 mg calcium, 0.73 mg iron, 0.18 mg zinc, and 3.19 × 108 CFU/mL *L. plantarum*.

#### *2.3. Socio-Demographic Characteristics Assessment*

Socio-demographic characteristics were collected from the parents or guardians of the subjects using questionnaires before the intervention began. The variables assessed were birth weight, birth length, sex, number of siblings, parents' educational levels, household income, illness frequency in the last three months, and illness duration. The variability of these factors between the two groups was assessed.

#### *2.4. Anthropometry Measurements*

Body weight and height were measured before and after the intervention phase using a standardized procedure [34]. The weights of the children were assessed using a digital weight scale (GEA Medical, Jakarta, Indonesia) with an accuracy of 0.1 kg; the height was measured using a microtoise with an accuracy of 0.1 cm. We assessed HAZ and WAZ to categorize the nutritional status of the subjects.

#### *2.5. Dietary Intake Assessment*

Dietary intake information was collected using 24-h food recall from the parents or guardians of the children. Trained enumerators conducted eight recalls in non-consecutive time with a range of 1–2 weeks for each assessment. The enumerators used a food model book to help with data collection. NutriSurvey software was used to analyze dietary data consumed by the subjects as energy, carbohydrates, protein, fat, iron, and zinc.

#### *2.6. Statistical Analysis*

Data were analyzed using STATA 13. Baseline characteristics were written as categorical data, then assessed using a chi-squared test or Fisher's exact test. A chi-square test was used to compare categorical data between the intervention and control groups; if the chi-square assumption could not be fulfilled (there is an expected value < 5 in the cell), then we used Fisher's Exact test [35]. Normality of the data was determined using the Shapiro–Wilk Test. Normally distributed data are presented as a mean ± standard deviation (SD) while abnormally distributed data are presented as a median (interquartile range). For normal data, the mean difference between the intervention and control groups was measured using an independent *t*-test. The mean difference between pre- and post-intervention was assessed using the paired *t*-test. For data that were not normally distributed, the Mann–Whitney U test was used instead of the independent *t*-test, and the Friedman test instead of the paired *t*-test. The differences in the nutritional status categories between the intervention and control groups were evaluated using the chi-squared test

or Fisher's exact test. Multiple linear regression was also used to assess the influence of several nutrients on subjects' nutritional status according to HAZ.

#### **3. Results**

Weight-for-age z-score (WAZ)

Ninety-four (n = 94) children under five years of age were divided into two groups: intervention and control. Throughout the research, four subjects dropped out from the control group and nine dropped out from the intervention group. Subject characteristics were evenly distributed between the two groups. In the current study, the age of the subjects ranged from 24 to 59 months with the number of boys and girls nearly equal. More than 50% of the subjects had birth lengths of 48 cm and birth weights of more than 2500 g. The complete results can be seen in Table 1.

After drinking synbiotic fermented milk for three months, the heights and weights of the subjects remained nearly the same. Table 2 shows that there was no difference in nutritional status between the intervention and control groups (*p* > 0.05), according to their HAZ and WAZ data.


**Table 2.** Height-for-age z-score (HAZ) and Weight-for-age z-score (WAZ) category after intervention.

Data presented in frequency n (%); <sup>a</sup> chi-squared Test; \* significant at *p* < 0.05; intervention group, given synbiotic fermented milk fortified with iron and zinc; control, given synbiotic fermented milk without fortification.

Post-intervention Underweight 15 (39) 26 (60) 0.059 <sup>a</sup>

Normal 23 (61) 17 (40)

In the preliminary phase of the intervention group, all 38 subjects were categorized as stunted. When the subjects were measured again three months after intervention, nine were categorized as normal according to height-for-age (*p* < 0.05). Six months after the intervention, 11 subjects had dropped out. Of these children, four were categorized as stunted in the past two measurements, and two as severely stunted. Two more were severely stunted before the intervention but improved to stunted after the intervention. Three subjects who were stunted before the intervention were re-categorized as normal afterwards. Although four subjects in the control group were also categorized as normal after the intervention, it was not considered statistically significant.

Table 3 shows there was a significant difference of energy, carbohydrates, protein, fat, and iron intake between the intervention and control groups. The intake was higher in the control group than in the intervention group.

**Table 3.** Comparison of dietary intake between intervention and control group.


<sup>1</sup> Data presented in median (IQR); <sup>2</sup> data presented in mean <sup>±</sup> standard deviation (SD); <sup>a</sup> Mann–Whitney U test; <sup>b</sup> independent *<sup>t</sup>*-test; significant if *<sup>p</sup>* < 0.05.

#### **4. Discussion**

The research subjects were stunted young children aged 2–5 years. After the subjects were randomized, there were some who dropped out during the study. The drop-out rate in the intervention group was 19.1%, while in the control group, it was 8.5%. According to Dumville et al. [36], the stunting level of the dropouts did not bias the results of the study. Statistical test results also showed no significant difference between subjects in the control group and the intervention group.

Synbiotic fermented milk in this study was fortified with as much as 2.26 mg (28% AKG) of iron and 1.22 mg (30% AKG) of zinc per 100 mL serving per day. In this study, the amount of fortification complied with the WHO Recommended Dietary Allowance (RDA), which mentions requirements for dairy products and their preparations in the range of 15–30% [18]. A study by Sazawal et al. [37] and El Menchawy et al. [38] also used 30% RDA for the amount of fortification added to milk, which proved to have a positive impact on growth. Antagonistic interactions between iron and zinc in research can be minimized with a ratio of iron to zinc of 1.8:1. Another study noted that iron and zinc fortification in milk formula, with a ratio of 1.3:1, had no effect on reducing zinc absorption [39]. Iron and zinc content of 1:1, however, can cause interactions that reduce the concentration of iron in the plasma, while a ≥ 2:1 ratio inhibits the absorption of zinc in the intestine [40].

Three months after being given the product, the bodyweight of subjects in the intervention group increased by 0.7 kg, while height increased by 2.58 cm. Weight and height gains in the control group were 0.6 kg and 2.5 cm. Based on the WHO Child Growth Standards, the increase in body weight and height of children aged 2–5 years over the course of three months ranged from 0.5 to 0.6 kg and 1.5 to 2.5 cm, respectively [41]. This shows that the results in both groups equaled or exceeded normal growth. Growth velocity of body weight and height in stunted children exceeded the speed of growth of children with normal height after the phase of growth restriction [42]. Every 1 cm in height was associated with a lower growth rate of 0.03 cm, and each one-unit increase in HAZ was associated with a decrease in growth speed of 0.08 standard deviation (SD) per year [43].

Nutrition intake, health status, and biological systems interact with each other [44]. The mechanisms underlying this interaction occur locally as well as systemically. The phase of stunted growth causes a decrease in the speed of cell proliferation and molecular changes. Increased levels of pro-inflammatory cytokines in the body can inhibit growth-promoting protein [45]. After this phase is completed, there is an increase in the speed of proliferation at the growth plate, or non-skeletal organs, beyond the normal speed according to age, because the organ feels a growth that is not in accordance with age [46].

Comparisons between weight, height, HAZ, and WAZ of the intervention and control groups did not show a significant difference. Various studies show inconsistent results related to the effect of micronutrient fortification on growth and growth indicators. This was not in accordance with the research of [47], which showed a significant difference in all growth indicators—body weight, height, WAZ, WHZ, and HAZ—in the larger group of children aged 1–4 years, given milk fortified with iron, zinc, and several other micronutrients for one year.

The provision of micronutrients showed mixed results on the growth of children. A meta-analysis by Ramakrishnan et al. [48] noted that iron and zinc supplements affected young children's height and weight gains, though the effect size was minor. Lind et al. [49] said that in his group, giving iron and zinc supplements had no effect on the HAZ after 12 months of intervention. Fahmida et al. [14] found a significant increase in HAZ and height compared to those given placebos, as well as stunted children given zinc after just 10 mg iron and 10 mg zinc supplements for four months. Iron and zinc produce varying effects on children's growth, especially those aged < 24 months who are not iron-deficient but, on the contrary, experience iron repletion. Iron repletion is negatively related to growth because iron in this condition can be a pro-oxidant that increases the number of pathogenic microorganisms [50].

Sufficient energy intake (≥80% RDA) was found to affect the HAZ. Energy plays a role in the synthesis of new tissue to form normal body composition, including adipose tissue, lean tissue, and skeletal tissue. The existence of a positive energy balance in the body can increase weight and height [51]. Micronutrients alone cannot support growth; the adequacy of macronutrients in the body also plays an important role. The lack of significant differences in growth between the intervention and control groups could be caused by the synbiotic milk given to both. Synbiotic milk contains energy of around 80 kcal/serving/day, and protein of around 2 g/serving/day that can support growth.

The addition of synbiotics to milk influences growth. A study by Sazawal et al. [47] showed a significant difference in the speed of weight gain that was greater in children who were given the synbiotic milk *Bifidobacterium lactis* HN019 and oligosaccharides compared to children who were given normal milk. Agustina et al. [25] stated that children who were given milk supplemented with *Lactobacillus reuteri* DSM 17938 for six months experienced significantly faster growth in body weight and height, and a significantly higher WAZ than those without probiotic supplementation. Another strain, *Lactobacillus plantarum* FNCC 260 from Indonesia, also has microbial activity against pathogens that is beneficial for gut health [52].

This study had several limitations. First, the duration was only three months. Although it is appropriate to measure height changes due to an intervention, it is difficult to conclude whether the effects of the intervention will continue. Secondly, the study group only consisted of the intervention and control groups. Both groups were given synbiotic fermented milk; the only difference was the presence of fortified iron and zinc. This design left the researchers without an exact conclusion regarding whether the synbiotic milk has a truly beneficial effect on growth, since the two groups' results were almost identical.

This study introduces an innovation to address stunting using a food-based approach. This research serves as the beginning of further studies in Indonesia, where many local resources can be utilized to address the country's nutritional problems. As a contribution to society, the development of the product in this study has been following the Innovator Innovation Indonesia Expo (*Inovator Inovasi Indonesia Expo*) in 2018 and is in the process of commercialization with the name of "Forty Milk".

#### **5. Conclusions**

There were no significant differences in height between the intervention and the control groups after they were given synbiotic milk, with or without double fortification (Fe-Zn). Nutritional status improvement according to HAZ and WAZ tended to be higher in the intervention group than in the control. However, statistical analysis revealed no significant difference observed, which suggests that the consumption of synbiotic milk alone had a good effect on the nutritional status of the children. Further study to include a third group who are not given fermented milk is needed for better comparison.

**Author Contributions:** Conceptualization, S.H. and E.S.R.; Supervision, S.H. and F.Z.N.; investigation, K.M.S., F.T.S., and M.P.S.; formal analysis, K.M.S., F.T.S., M.P.S., S.H., and M.W.; project administration and validation, D.R.A., R.A.P., and S.U.W.; writing—original draft K.M.S. and M.W. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the Indonesian Ministry of Research and Higher Education throughout *Hibah Penelitian Terapan Unggulan Perguruan Tinggi* on behalf of SH, grant number 1964/UN1/DITLIT/DIT-LIT/LT/2018.

**Institutional Review Board Statement:** The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the the Medical and Health Ethics Committee of the Faculty of Medicine, Public Health and Nursing, Universitas Gadjah Mada (protocol code KE/FK/0640/EC/2017 in 6 June 2017 and was amended in 9 February 2018). The study is also registered in ClinicalTrial.gov no NCT03495401.

**Informed Consent Statement:** Informed consent was obtained from all subjects involved in the study.

**Data Availability Statement:** Data available on request due to privacy issues.

**Acknowledgments:** The authors want to acknowledge all participants who followed this study and Beneo Orafti who gave some materials needed to make the product. They also acknowledge Bapak Jumarko, a local community nutritionist; Dita, a local resident; and all enumerators who helped to coordinate with participants and conduct this research.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


## *Article* **Microbial Production and Enzymatic Biosynthesis of** γ**-Aminobutyric Acid (GABA) Using** *Lactobacillus plantarum* **FNCC 260 Isolated from Indonesian Fermented Foods**

**Ida Bagus Agung Yogeswara 1,2,\*,†, Suwapat Kittibunchakul 1,3,†, Endang Sutriswati Rahayu 4,5, Konrad J. Domig 6, Dietmar Haltrich <sup>1</sup> and Thu Ha Nguyen 1,\***


**Abstract:** In the present study, we isolated and screened thirty strains of GABA (γ-aminobutyric acid) producing lactic acid bacteria (LAB) from traditional Indonesian fermented foods. Two strains were able to convert monosodium glutamate (MSG) to GABA after 24 h of cultivation at 37 ◦C based on thin layer chromatography (TLC) screening. Proteomic identification and 16S rDNA sequencing using MALDI-TOF MS identified the strain as *Lactobacillus plantarum* designated as *L. plantarum* FNCC 260 and FNCC 343. The highest yield of GABA production obtained from the fermentation of *L. plantarum* FNCC 260 was 809.2 mg/L of culture medium after 60 h of cultivation. The supplementation of 0.6 mM pyridoxal 5'-phosphate (PLP) and 0.1 mM pyridoxine led to the increase in GABA production to 945.3 mg/L and 969.5 mg/L, respectively. The highest GABA production of 1226.5 mg/L of the culture medium was obtained with 100 mM initial concentration of MSG added in the cultivation medium. The open reading frame (ORF) of 1410 bp of the *gad*B gene from *L. plantarum* FNCC 260 encodes 469 amino acids with a calculated molecular mass of 53.57 kDa. The production of GABA via enzymatic conversion of monosodium glutamate (MSG) using purified recombinant glutamate decarboxylase (GAD) from *L. plantarum* FNCC 260 expressed in *Escherichia coli* was found to be more efficient (5-fold higher within 6 h) than the production obtained from fermentation. *L. plantarum* FNCC 260 could be of interest for the synthesis of GABA.

**Keywords:** GABA; Indonesian fermented foods; glutamate decarboxylase; lactic acid bacteria; *L. plantarum*

#### **1. Introduction**

γ-aminobutyric acid (GABA), which is a non-protein amino acid and plays a major role as a suppressive neurotransmitter, is widely present in plants, microorganisms, and the mammalian brain [1–3]. GABA has been extensively studied due to its physiological and pharmacological effects including anti-depressant, hypotensive activity, anti-diabetic in humans [4–6]. Recently, GABA administration in fluoride-exposed mice showed protective effects against hypothyroidism and maintained lipid and glucose levels in vivo [4]. Furthermore, GABA-enriched foods have been developed [5–11]. GABA-rich chlorella

**Citation:** Yogeswara, I.B.A.; Kittibunchakul, S.; Rahayu, E.S.; Domig, K.J.; Haltrich, D.; Nguyen, T.H. Microbial Production and Enzymatic Biosynthesis of γ-Aminobutyric Acid (GABA) Using *Lactobacillus plantarum* FNCC 260 Isolated from Indonesian Fermented Foods. *Processes* **2021**, *9*, 22. https://dx.doi.org/ 10.3390/pr9010022

Received: 30 November 2020 Accepted: 21 December 2020 Published: 24 December 2020

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/ licenses/by/4.0/).

has been shown to significantly lower high blood pressure in hypertensive subjects [12]. A number of GABA-enriched foods such as soymilk [13], fermented milk [14], natto [15], green tea [16], and cheese [5] have been reported to suppress the elevation of blood pressure in spontaneously hypertensive rats (SHR) and hypertensive subjects.

GABA can be synthesized using chemical or biochemical means, of which the latter involves enzymatic conversion, whole-cell biocatalysts, or microbial fermentation. The chemical synthesis is considered hazardous due to corrosive nature of used reagents [17–19], hence the application in the food industry is limited. Moreover, the supplementation of synthetic GABA to food system is considered unnatural and unsafe [14]. Therefore, it is important to develop a natural and safe method to increase GABA in foods, since there are no side effects of natural GABA supplementation [16]. Recent studies showed that several strains of lactic acid bacteria (LAB) are promising candidates as GABA-producing bacteria [1–3,5] due to their GRAS (generally recognized as safe) status. A number of GABA-producing bacteria has been isolated from fermented foods such as *L. brevis* (from kimchi) [20], *L. rhamnosus* (from fermented pickles) [21], *L. plantarum* (from fermented dairy products) [22], *L. helveticus* (from koumis fermented milk) [23], *L. buchneri* (from kimchi) [24], *L. otakiensis* (from Pico cheese) [25], *L. namurensis* (from fermented green papaya) [11], and *L. paracasei* (from Italian cheese) [3]. These reports have shown that fermented foods are promising sources of GABA-producing bacteria. In addition, screening GABA-producing LAB from various fermented foods might open the possibilities to obtain newly isolated strains for the use as functional starter cultures in the food industry.

Several fermented foods from Indonesia namely *gatot, growol, tape ubi, bekasam,* and *tempoyak* are spontaneously fermented by LAB, which mainly involve the strains of the genera *Lactobacillus, Pediococcus,* and *Streptococcus* [26,27]. However, the potential of these LAB strains from Indonesian fermented foods to be used as GABA-producing bacteria as well as their relevant enzymes have not yet been studied. Therefore, the development of GABAenriched foods using suitable LAB is a promising strategy, to bring new functional foods to the market. In addition, biosynthesis of GABA using LAB also provides advantageous effects including probiotic activity and extension of the shelf-life of food products [9].

The biosynthesis of GABA involves irreversible decarboxylation reaction of glutamate to GABA and carbon dioxide catalyzed by glutamate decarboxylase (GAD). GAD (EC 4.1.1.15) is a major enzyme for GABA synthesis and it requires pyridoxal 5'-phosphate (PLP) as a cofactor [28–30]. The GAD genes from various sources have been cloned, expressed and their biochemical properties have been characterized [20,31–34]. The use of purified GAD for the biosynthesis of GABA is also of interest because only simple downstream purification of GABA is required and yet, the process could overcome the limitation of microbial fermentation (i.e., GABA catabolism). In the present study, we describe the screening of GABA-producing LAB from Indonesian fermented foods (fermented soybeans, *growol, gatot, tempeh*, and *bekasam*) and GABA productions using microbial fermentation of the isolated strain and the purified GAD of this strain for the conversion of glutamate to GABA.

#### **2. Materials and Methods**

#### *2.1. Screening of GABA-Producing LAB*

Thirty isolates of *Lactobacillus* spp. were previously isolated from Indonesian fermented foods such as fermented soybeans, *growol*, *gatot*, *tempeh,* and *bekasam* (fermented fish) [35]. *Lactobacillus* spp. were the predominant genus according to cell morphology, Gram reactions and catalase tests. All strains were obtained and stored in the Food and Nutrition Culture Collection, Universitas Gadjah Mada (Yogyakarta, Indonesia). Prior to screening, all strains were grown in MRS broth containing 118 mM monosodium glutamate (MSG) (Ajinomoto, Tokyo, Japan) for 24–48 h at 37 ◦C under microaerophilic conditions. The cultures broth was then centrifuged at 8000× *g* for 5 min at 4 ◦C. GABA formation in the supernatant was analyzed using thin layer chromatography (TLC). Briefly, 0.5–1.0 μL of supernatants were spotted onto TLC plates Silica gel 60 F254 (Merck, Darmstadt, Germany).

The mobile phase consists of a mixture of 1-butanol: acetic acid: distilled water (5:2:2). Subsequently, the plates were sprayed with 0.5% ninhydrin and heated at 105 ◦C for 5 min to visualize the spots. GABA (Sigma Aldrich, St. Louis, MO, USA) was used as a standard, and the Rf values were calculated. LAB cultures showing the same Rf values as GABA standard were considered as positive GABA-producers. Positive GABA-producing strains were identified using proteomic and genotype techniques. Furthermore, the amount of GABA produced was determined by the GABase assay [36]. All chemicals were of the highest grade.

#### *2.2. Identification of GABA-Producing LAB*

Proteomic and genotype techniques were performed to identified GABA-producing LAB. Genomic DNA of GABA-producing LAB was extracted using peqGOLD Bacterial DNA Mini Kit (PeqLab, Erlangen, Germany) according to manufacturer's instructions. The extracted DNA was used as a template for partial 16S rDNA amplification. The amplifications of 16S rDNA were performed using forward primer bak4 (5 -AGGAGGTCATCCARC CGCA-3 ) and reverse primer bak11w (5 -AGTTTGATCMTGGCTCAG-3 ) [37], The PCR reaction mixtures consisted of 10× PCR buffer (Dynazyme buffer 10× Thermo scientific, Waltham, MA, USA), 10 nmol/μL dNTP mix (GE Healthcare Buckinghamshire, UK), 2 U/μL DNA polymerase (Dynazime II, Thermo scientific) and high-quality sterile water to a total volume of 25 μL. The conditions for PCR amplification were as follows: initial denaturation at 95 ◦C for 3 min, followed by 30 cycles of denaturation at 95 ◦C for 30 s, annealing at 56 ◦C for 30 s, extension 72 ◦C for 2 min, and a final extension at 72 ◦C for 7 min. After PCR amplification, the amplified products were visualized by gel electrophoresis. The gel was stained with GelRed Nucleic Acid (Biotium, Hayward, CA, USA) and subsequently visualized with an ultraviolet transilluminator (BioRad, Hercules, CA, USA). The PCR products were purified using QIAquick PCR purification Kit (Qiagen, Venlo, The Netherlands) and sent for sequencing (Eurofins MWG Operon, Ebersberg, Germany). Subsequently, the partial 16S rDNA sequence was compared with the National Center for Biotechnology Information (NCBI) sequence database using Basic Local Alignment Search Tool (BLAST) program.

Proteomic identification was performed using matrix-assisted laser desorption/ionizing time-of-flight mass spectrometry (MALDI-TOF MS). GABA-producing bacteria were identified by the extended direct transfer method. A single colony was directly spread onto a MALDI target plate. The spot was overlayed with 1 μL of 70% formic acid and allowed to dry at room temperature. Furthermore, 1 μL of 10 mg/μL HCCA (α-cyano-4 hydroxycinnamic acid) solution was then added to the spot and allowed to dry at room temperature. The target plate was immediately applied to MALDI-TOF MS and analyzed using Microflex LT bench-top mass spectrometer (Bruker Daltonics, Bremen, Germany) equipped with the FlexControl 3.4 software. A mass spectrum was processed using Bio-Typer software (version 3.0, Bruker Daltonics, Bremen, Germany). MALDI-TOF MS profiles were obtained from bacteria isolates and matched with a database containing 8223 reference MALDI-TOF MS profiles.

#### *2.3. Determination of GABA Production and GAD Assay*

The GABAse method was performed to determine GABA concentration in culture supernatants. Briefly, the culture broth was centrifuged at 8000× *g* for 5 min at 4 ◦C. 10 μL of supernatants were mixed with 140 μL of 100 mM K4P2O7 buffer (pH 8.6), 30 μL of 4 mM NADP+, 10 μL of 1 U/mL GABase (Sigma-Aldrich, St. Louis, MO, USA). The mixtures were dispensed into each well of 96-well plate. The initial absorbance was read at 340 nm in PerkinElmer plate reader (PerkinElmer, Buckinghamshire, UK). After the initial reading, 10 μL of 20 mM α-ketoglutarate were added and the mixtures were incubated for 1 h. The final absorbance was read after 1 h at the same wavelength. GABA concentrations were determined based on the difference of A340 values and the standard curve of GABA.

The GAD assay was carried out using colorimetric method [38]. The reaction mixtures consist of 200 mM Na2HPO4-citric acid buffer (pH 5.0), 20 mM L-MSG, 0.2 mM PLP, and 20 μL of purified GAD. The mixtures were thoroughly mixed and incubated at 37 ◦C for 1 h and then deactivated by boiling for 5 min. The reaction mixtures were used to determined GAD activity using the Berthelot reaction method, which was composed of 100 μL of reaction sample, 250 μL of H2O, 50 μL of 200 mM sodium borate (pH 9.0), 250 μL of 6% phenol and 200 μL of 5% (*w*/*v*) sodium hypochlorite. Subsequently, the reaction mixtures were thoroughly mixed and boiled for 10 min until the blue color developed, then immediately placed on ice for 15 min. The mixtures were analyzed colorimetric at 630 nm to determine the absorption value. One unit of GAD activity was defined as the amount of enzyme that liberates 1 μmol of GABA per minute under activity assay conditions.

The concentrations of GABA formed in supernatants and after enzymatic conversions were confirmed using Ultra Performance Liquid Chromatography (UPLC Acquity H- Class, Waters Corporation, Milford, MA, USA) equipped with a PDA detector and an AccQ. Tag Ultra C18 column (1.7 μm particle, 2.1 × 100 mm). The samples were hydrolyzed using 6 N HCL and followed by derivatization of the samples and the GABA standard using AccQ-Tag ultra-derivatization kit (Waters, Milford, MA, USA) according to the manufacturer's instructions. For UPLC analysis, the derivatized samples were injected to Acquity UPLC H class [39]. The system was operated at a flow rate of 0.5 mL/min at 49 ◦C with a wavelength of 260 nm. The mobile phase used were AccQ. Tag Ultra Eluent A 100%; Accq. Tag Ultra Eluent B (Aquabides 90:10); Aquabides Eluent C; AccQ. Tag Ultra Eluent B 100%.

#### *2.4. GABA Production*

The MRS medium was inoculated with 5% inoculum of GABA-producing LAB and incubated at 37 ◦C for 108 h. The optical densities (OD600) of the cultures were measured every 12 h. A concentration of MSG (25–100 mM) (Sigma Aldrich, St. Louis, MO, USA), pyridoxal 5-phosphate (PLP, 0.2 and 0.6 mM) and pyridoxine (vitamin B6, 0.1–0.3 mM) were added to the MRS medium and GABA production under these conditions was investigated subsequently.

#### *2.5. Cloning of gadB Gene*

The glutamate decarboxylase gene (*gad*) from *L. plantarum* FNCC260 was amplified using degenerated primers *gad*\_FwdNdeI (5 -CATATGATGGCAATGTTRTAYGGTAAAC-3 ) and *gad*\_RevEcoRI (5 -GAATTCCAGTGTGTGAATMSGTATTTC-3 ), which were designed based on the sequences of the *gad* genes of *Lactobacillus* spp. available in GenBank (Accession numbers JN248358.1, KU214639.1, AB986192.1, CP029349.1, AL935263.1, CP018209.1, CP028977.1, GU987102.1, JX545343.1).

The primers were supplied by VBC-Biotech Service (Vienna, Austria) and the appropriate endonuclease restriction sites were introduced in the forward and reverse primers (underlined sequences). The conditions for PCR reactions were as follows: initial denaturation at 98 ◦C for 20 s; 30 cycles of denaturation at 98 ◦C for 20 s, annealing at 58 ◦C for 20 s, extension at 72 ◦C for 1 min 45 s, and final extension at 72 ◦C for 2 min. The amplified PCR products were purified using the Monarch DNA Gel Extraction Kit (New England Biolabs, Ipswich, MA, USA), digested with *Nde*I and *EcoR*I and cloned into the pET 21(+a) vector (Novagen, Merck KGaA, Darmstadt, Germany) resulting in the plasmid pET21GAD. *E. coli* NEB5α was used as a host for obtaining the plasmids in sufficient amounts. The sequence of the insert was confirmed by DNA sequencing performed by a commercial provider (Microsynth, Vienna, Austria). The alignment tool (BLAST) from the National Center for Biotechnology Information BLAST website was used for the alignment of the nucleotide sequence of the *gad* gene from *L. plantarum* FNCC 260 with the available *gad* sequences from LAB. The comparison of glutamate decaroboxylases (GAD) from different LAB species was carried out using the program Clustal Omega (https://www.ebi.ac.uk/Tools/msa/clustalo/) [31,33].

#### *2.6. Overexpression of GADLbFNCC260 in E. coli and Protein Purification*

The expression vector of pET21GAD harboring the *gad* gene from *L. plantarum* FNCC 260 was transformed into *E. coli* T7 Express GRO carrying the plasmid pGRO7, which encodes the chaperones GroEL and GroES (Takara, Shiga, Japan). Subsequently, *E. coli* T7 Express GRO carrying the plasmid pET21GAD was cultivated in LB broth medium supplemented with 100 μg/mL ampicillin, 20 μg/mL chloramphenicol, and 1 mg/mL arabinose until OD600nm of 0.6 was reached. Thereafter, the isopropyl β-D-1-thiogalactopyranoside (IPTG) was added to a final concentration of 0.5 mM for induction. The culture was further incubated at 18 ◦C for 20 h with shaking at 180 rpm. The cells were harvested, washed twice with 50 mM sodium phosphate buffer (pH 6.5), and resuspended in buffer A (50 mM NaH2PO4, 300 mM NaCl, and 10 mM imidazole, pH 7.0). The resuspended cells were disrupted using a French press (Aminco, Silver Spring, MD, USA) and centrifuged at 15,000× *g* for 20 min at 4 ◦C. The cell-free extracts were collected and loaded to a prepacked 1 mL HisTrap HP Ni-immobilized metal ion affinity chromatography (IMAC) column (GE Healthcare, Uppsala, Sweden) that was pre-equilibrated with buffer A (50 mM NaH2PO4, 300 mM NaCl, and 10 mM imidazole, pH 7.0). The His-tagged protein was eluted at a rate of 1 mL/min with a 15 mL linear gradient from 0 to 100% buffer B (20 mM NaH2PO, 500 mM imidazole, 500 mM NaCl, pH 6.5). Active fractions were pooled, desalted, and concentrated by ultrafiltration using an Amicon Ultra centrifugal filter unit with a 30 kDa cut-off membrane (Millipore, Burlington, MA, USA). The purified enzyme was stored in 50 mM citrate-phosphate buffer (pH 5.0) for further characterization and enzymatic conversion. The molecular masses of purified GAD were determined by SDS-PAGE and Native PAGE. Protein bands were visualized by staining with Bio-safe Coomassie (Bio-Rad). The determination of protein mass was carried out using Unstained Precision plus Protein Standard (Bio-Rad, Hercules, CA, USA).

The size of the protein was also confirmed by LC-ESI-MS analysis. The proteins were S-alkylated with iodoacetamide and digested with Trypsin (Promega, Madison, WI, USA). The digested proteins were directly injected to LC-ESI-MS (LC: Dionex Ultimate 3000 LC). A gradient from 10 to 80% acetonitrile in 0.05% trifluoroacetic acid (using a Thermo ProSwiftTM RP-4H column (0.2 × 250 mm) at a flow rate of 8 μL/min was applied (30 min gradient time). Detection was performed with a Q-TOF instrument (Bruker maxis 4G, Billerica, MA, USA) equipped with standard ESI source in positive ion, MS mode (range: 400–3000 Da). Instrument calibration was performed using ESI calibration mixture (Agilent, Santa Clara, CA, USA). Data were processed using Data analysis 4.0 (Bruker) and the spectrum was deconvoluted by MaxEnt.

#### *2.7. Enzymatic Synthesis of GABA*

Batch conversion reactions were carried out in 2 mL scale with 0.64 U/mL purified GAD using 100 mM MSG in 50 mM citrate-phosphate buffer (pH 4.5) containing 0.2 mM of PLP as cofactor. Decarboxylation reactions were performed at 30 ◦C with 300 rpm agitation using a Thermomixer (Eppendorf, Hamburg, Germany). The samples were withdrawn at time intervals and the enzyme GAD was inactivated at 100 ◦C for 5 min. The samples were stored at −20 ◦C for subsequent analysis. GABA content in the reaction mixtures was determined using the GABase assay and confirmed with UPLC analysis as described in Section 2.3.

#### **3. Results and Discussion**

#### *3.1. Screening and Identification of GABA-Producing LAB*

Thirty isolates of *Lactobacillus* spp. from Indonesian fermented foods were screened for the formation of GABA in the culture medium using the TLC method and only two isolates showed clear spots on TLC plate (Figure 1), which have similar Rf value (0.78) as the GABA standard. These two isolates were FNCC 260 and FNCC 343 isolated from fermented cassava and fermented fish, respectively. The two strains FNCC 260 and FNCC 343 were cultivated in MRS broth containing 118 mM MSG for 48 h to determine the GABA

production in the culture medium, which was analyzed to be 352 mg and 328 mg of GABA per liter of culture medium, respectively. Based on morphological observation, these two strains were Gram positive, rod-shape, microaerophilic, and catalase negative.

**Figure 1.** Thin layer chromatography (TLC) screening of GABA–producing LAB. Lane S: GABA standard, Lane C: MRS with 118 mM monosodium glutamate (MSG), Lane 1–8: the strains FNCC 245, FNCC 344, FNCC 343, FNCC 253, FNCC 283, FNCC 235, FNCC 257, and FNCC 260, respectively. All strains were cultivated in MRS broth supplemented with 118 mM MSG and incubated at 37 ◦C for 48 h.

Subsequently, these two GABA-producing LAB were identified using 16S rDNA and MALDI-TOF MS (Bruker Biotyper). Proteomic identification by MALDI-TOF MS was performed since this technique is very effective in identifying species and subspecies of LAB. A number of species and subspecies of LAB have been successfully identified using proteomic-based identification technique [40–43]. Based on MALDI-TOF MS identification, both strains FNCC 260 (2.15 log score) and FNCC 343 (2.12 log score) were identified to be *L. plantarum*, which matches with the strain in a reference database. The log scores also indicate the accuracy and reliability of MALDI-TOF MS identification. A log score between 2.00 and 2.30 indicates accurate identification to the genus and species level [44,45]. Subsequently, GABA-producing LAB were subjected to 16S rDNA sequencing.

Based on partial 16S rDNA sequencing (~1400 bp), both strains belong to the species *L. plantarum*, with 99.81% sequence identity with *L. plantarum* strain CIP 103,151 (accession number, NR\_104573.1) for the strain FNCC 260, and 99.81% sequence identity with *L. plantarum* strain NBRC 15,891 (accession number, NR\_113338.1) for the strain FNCC 343, respectively. The results confirmed that both strains FNCC 260 and FNCC 343 are indeed *L. plantarum*.

#### *3.2. Time-Course of GABA Production by L. plantarum FNCC 260*

*L. plantarum* FNCC 260 was cultivated in MRS medium supplemented with 118 mM MSG at 37 ◦C. The time courses of GABA production, the pH value and the growth profile of *L. plantarum* FNCC 260 are shown in Figure 2. GABA production started when cell growth reached the stationary phase after 12 h of cultivation. A slightly higher GABA production was obtained after 48 h (450 mg/L) compared to the GABA production mentioned above in the screening experiment (352 mg/L). This was due to different MSG used in these two experiments (see Materials and Methods). The highest GABA production was 809.2 mg/L of cultivation medium after 60 h of cultivation, at which cell growth is still in the stationary phase. This observation agrees with previous reports in the literature. The maximum GABA production of *L. brevis* L-32 was observed between 36 to 72 h of cultivation and GABA was mainly produced during the stationary growth phase [46–48]. However, GABA production decreased when the cultivation time was prolonged further. This might be due to the activity of the enzyme GABA transaminase (GABA-T), which degrades GABA. This enzyme catalyzes GABA degradation to succinic semialdehyde by using either pyruvate or α-ketoglutarate as the amino acceptors and succinic semialdehyde is irreversibly oxidized to succinate by succinic semialdehyde dehydrogenase [46,49,50]. Interestingly, we observed that the cell growth did not show a decreasing trend when the cultivation time was extended up to 108 h since GABA is utilized as a nutrient during prolonged cultivation [49]. Ko et al. (2013) reported a similar observation of cell growth of *L. brevis* FPA 3709 during GABA synthesis when GABA production decreased [49].

**Figure 2.** Time course of γ-aminobutyric acid (GABA) production, bacterial growth, and change in pH during cultivation of *L. plantarum* FNCC 260 strain in MRS broth supplemented with 118 mM MSG at 37 ◦C. Data shown as mean ±SD with SD less than 5%. The experiments were conducted at least in duplicates.

During GABA production, a decrease in pH of MRS medium was observed. The pH of the cultivation medium rapidly decreased from the initial pH 6.5 to pH 4.1 after 12 h of cultivation. The decrease in pH was due to lactic acid and acetic acid formation during the cultivation of the organism [50]. GABA production involves cytoplasmic decarboxylation, which results in extracellular proton consumption after the uptake of glutamate by its specific transporter [28]. This may lead to the removal of hydrogen ions and an increase of pH in the cytoplasm [51]. Apparently, we observed that the decrease in pH of the cultivation medium (to below pH 4.0) correlated with the increase in GABA production during cultivation of *L. plamtarum* FNCC 260. Similarly, the maximum GABA production of *L. buchnerii* was achieved when the pH of the cultivation medium decreased to pH 5.0 [24]. In this study, although GABA production started to decrease after 60 h of cultivation, which might be due to the activity of the enzyme GABA transaminase, the pH of the cultivation medium kept decreasing until 84 h of cultivation indicating that decarboxylation of glutamate still occurred.

#### *3.3. The Effect of Cofactors on GABA Production and Cell Growth*

Glutamate decarboxylase is a pyridoxal 5'-phosphate (PLP) dependent enzyme. Theoretically, the addition of PLP to the medium could increase GAD activity and GABA production [2,21,52,53]. PLP and pyridoxine were added into the medium at various

concentrations. As expected, the addition of 0.2 mM and 0.6 mM PLP led to the increase in GABA production, reaching 903.0 mg/L and 945.3 mg/L after 108 h of cultivation, respectively (Figure 3a). Furthermore, GABA was still produced when cultivation time was prolonged to 108 h. In contrast, GABA production in a medium without PLP rapidly decreased after 72 h. it was clear that the strain was able to utilize PLP to produced GABA. The addition of PLP to the medium did not inhibit cell growth during the cultivations (Figure 3b). Previous studies by Komatsuzaki et al. [2] and Yang et al. [53] reported that the addition of 0.1 mM and 0.02 mM PLP significantly enhanced GABA production and GAD activity in the culture media of *L. paracasei* NFRI 7451 and *S. thermophillus* Y2, respectively.

**Figure 3.** Effect of pyridoxal 5'-phosphate (PLP) on (**a**) GABA production and (**b**) cell growth during the cultivation of *L. plantarum* FNCC 260. The MRS medium were supplemented with 118 mM MSG and PLP at concentrations 0, 0.2 mM and 0.6 mM. The strain was incubated at 37 ◦C for 108 h. Data expressed as means ±SD with SD less than 5%. The experiments were conducted at least in duplicates.

Pyridoxine (vitamin B6) is a water-soluble vitamin that is ubiquitously found in nature. Pyridoxine can be taken by the cells at the plasma membrane and is subsequently phosphorylated to form PLP within the cytoplasm [52]. The utilization of pyridoxine could be an alternative to replace PLP since PLP is significantly more expensive with low availability. Therefore, we hypothesized that the addition of pyridoxine would improve GABA production. It is shown that the addition of 0.1 mM pyridoxine had a better enhancement on GABA production compared to higher concentrations of pyridoxine tested. The addition of 0.1 mM pyridoxine enhanced GABA production reaching 839.5 mg/L and 969.5 mg/L after 60 h and 96 h of cultivation, respectively (Figure 4a), which are higher compared to the cultivation without pyridoxine (754 mg/L and 606 mg/L, respectively). This observation

suggests that pyridoxine can be taken up by the cells and is phosphorylated to form PLP, which is essential for GAD activity. However, GABA production decreased to 878.8 mg/L after 108 h of cultivation (Figure 4a). This could be due to the degradation of pyridoxine during cultivation and hence it lost its activity as a cofactor [53]. The addition of pyridoxine did not have notable effects on cell growth during the cultivation of *L. plantarum* FNCC 260 (Figure 4b). In both cases of PLP and pyridoxine additions, we observed that the growth reached OD600 ~ 8 within 12 h and then maintained at OD ~ 6–7 during the entire cultivation time up to 108 h. Li et al. reported that the addition of cofactor did not improve or inhibit the cell growth of *L. brevis* NCL912 [28].

**Figure 4.** Effect of pyridoxine on (**a**) GABA production and (**b**) cell growth during the cultivation of *L. plantarum* FNCC 260. The MRS medium was supplemented with 118 mM MSG and various concentrations of pyridoxine 0, 0.1 mM, 0.2 mM, 0.3 mM. The strain was incubated at 37 ◦C for 108 h. Data expressed as mean ± SD with SD less than 5%. The experiments were conducted at least in duplicates.

#### *3.4. The Effect of Monosodium Glutamate on GABA Production*

The presence of MSG is a key factor in producing GABA. The optimal culture conditions for GABA production were determined by measuring the GABA content in the cultivation medium of *L. plantarum* FNCC 260 with different initial MSG concentrations in the MRS medium. As shown in Figure 5 increasing MSG concentrations increased GABA production (Figure 5a) and maintained cell viability (Figure 5b). The maximum GABA production was achieved at 1226 mg/L at 96 h in an MRS medium containing 100 mM MSG. It appears that a prolonged incubation time did not increase GABA productivity of the strain. A possible reason led to such circumstances was due to GABA catabolism, resulting from GABA transaminase activity. The activity of GABA transaminase could decrease GABA production by converting GABA to succinic semi-aldehyde (SSA).

**Figure 5.** Effect of various MSG concentrations on (**a**) GABA production and (**b**) the growth of *L. plantarum* FNCC 260 cultivated in MRS medium at 37 ◦C. Data expressed as mean ± SD with SD less than 5%. The experiments were conducted at least in duplicates.

Similarly, high concentrations of MSG resulted in decreased GABA production of the strains *L brevis* CRL 1942, *S. thermophillus* Y2 and *L. paracasei* NFRI 7415 [2,54,55]. It was suggested that high glutamate concentrations become more toxic to some strains of LAB and suppressed the expression of *gad*B genes [56]. In this study, we observed that MSG concentration up to 100 mM did not have effects on bacterial growth (Figure 5b) indicating that GABA is consumed by the cells to maintain its viability during the cultivation period. However, as it was shown in Figure 2, when the initial concentration of MSG in cultivation medium was 118 mM, the production of GABA was significantly lower compared to the production obtained with 100 mM MSG. This suggests that the observations from previous studies [2,28,53] about the negative effects of high glutamate concentrations on bacterial growth of some LAB strains and the expression of *gad*B genes could be an explanation for a similar observation in our study.

#### *3.5. Cloning, Expression of Glutamate Decarboxylase from L. plantarum FNCC 260 in E. coli and Purification of the Enzyme*

The *gad*B gene from *L. plantarum* FNCC 260 was cloned and its complete open reading frame consists of 1410 bp, encoding 469 amino acids. The predicted molecular mass of GAD is 53. 57 kDa and the theoretical isoelectric point (pI) is 5.62 as calculated using ExPASy program (www.expasy.org). The GadB sequence from *L. plantarum* FNCC 260 shared 98% homology with the GadB from *L. futsaii* CS3 (accession number AB839950), *L. plantarum* Taj-Apis362 (accession number AHG59384) and *L. plantarum* WCFS1(accession number CCC80401.1).

The *gad*B gene was cloned into the expression vector pET-21a(+) (Novagen, Merck KGaA, Darmstadt, Germany). The resulting vector pET21Gad was subsequently transformed into *E. coli* T7 Express carrying the plasmid pGRO 7 for the enhanced expression of the chaperones GroEL/GroES (*E. coli* T7 Express GRO). *E. coli* cells were cultivated in LB medium and induced with 0.5 mM IPTG as described in Section 2. The obtained expression yield was 1.38 kU/L fermentation medium with a specific activity of 0.24 U/mg. The recombinant GAD was purified with a single-step purification using the His-trap HP column and the specific activity of the purified enzyme was 1.12 U/mg with a purification factor of 4.5. The apparent molecular mass as judged by SDS-PAGE and native PAGE was estimated to be ~51 kDa and ~140 kDa, respectively (Figure 6a,b). The size of the protein was also confirmed by LC-ESI-MS and it was determined to be 51.79 kDa (data not shown). Several bands were found in native PAGE with the largest band was ~140 kDa. The LC-ESI-MS analysis revealed that these bands on native PAGE contain components of the subunit. The first band represent the intact dimeric enzyme and the other bands with lower molecular masses result from degradation of the intact protein. It also suggested that GadB from *L. plantarum* FNCC 260 is a homodimeric enzyme. GadB from *L. plantarum* ATCC 14917 was also reported as a homodimer [56].

**Figure 6.** (**a**) SDS-PAGE analysis of purified recombinant GAD from *L. plantarum* FNCC 260 expressed in *E. coli.* Lane M; protein marker, 1; purified GAD. The arrow indicated GAD with molecular masses of approximately 53 kDa. (**b**) Native PAGE analysis of purified recombinant GAD. The molecular masses of GAD were estimated to be 140 kDa, Lane M; protein marker, 1; purified GAD.

> The deduced amino acid sequence of *L. plantarum* FNCC 260 GAD contains a highly conserved catalytic domain that belongs to the PLP-dependent decarboxylase superfamily (Figure 7). A lysine residue (K280) is considered as the PLP-binding site for most bacterial GADs [51,56]. Lysine residue is also found in plant GADs since high homology between bacterial and plant GADs has been revealed [57,58]. In addition, the two residues T215 and D247 are crucial to promote decarboxylation [50]. Furthermore, the consensus sequence HVDAASGG is highly conserved in many bacterial GADs (Figure 7) and is also found in several GADs from Lactobacillus spp including *L. futsaii* CS3, *L. brevis* HYE1, *L. zymae*, and *L. paracasei* NFRI 7415 [19,34,36,51].

**Figure 7.** Alignment of amino acid sequences of GAD from *L. plantarum* FNCC 260 and six other GADs from LAB. The consensus sequence HVDAASGGF indicated by a smaller red box is highly conserved in GAD sequences. The sequence SINASGHKYGLVYPGVGWVVWR in the bigger red box is the PLP-binding domain [49]. GadB sequences shown are from *L. delbrueckii* (Ldelbrueckii), *Lactococcus lactis* (Lclactis), *S. thermophilus* (St. thermophilus), *L. plantarum* FNCC 260 (Lp260), *L. futsaii* (Lfutsaii), *L. herbarum* (Lherbarum), and *L. mudajiangensi* (Lmudanjiangensi).

#### *3.6. GABA Synthesis by Recombinant Glutamate Decarboxylase from L. plantarum FNCC 260*

For enzymatic GABA synthesis, we performed the conversion of MSG using 0.64 U/mL purified recombinant GAD in a 2-mL scale of reaction mixtures. Most GADs from *Lactobacillus* spp. have optimum activities at acidic pH values [33,36,57,58], and the recombinant GAD from *L. plantarum* FNCC 260 showed an optimum pH at pH 4.5 as expected (data not shown). MSG (100 mM) in 50 mM citrate-phosphate buffer (pH 4.5) containing 0.2 mM PLP was used as substrate and the reaction was performed at 30 ◦C. GABA synthesis reached its highest yield at 6450 mg/L (63 mM) within 6 h of reaction (Figure 8), and the enzyme retained 73% of its initial activity after 6 h of reaction. It was clear that the use of purified GAD was more efficient in terms of both GABA production and conversion time. Enzymatic synthesis of GABA using purified recombinant GAD from *L. plantarum* FNCC 260 showed 5 to 7-fold higher product concentrations than microbial fermentations in a significantly shorter time. This suggests that the use of purified GAD is crucial to overcome the limitations in GABA production due to GABA-degrading enzymes in the cells, slow reaction rate, and low production yield [28,53,59,60]. Furthermore, UPLC analysis was performed to confirm GABA production in both microbial fermentation and enzymatic conversion with a retention time of GABA at 8.5 min (Figure 9).

(**a**)

**Figure 9.** *Cont.*

**Figure 9.** UPLC analysis of GABA in the culture medium of microbial fermentation (**a**) and in the reaction mixture of enzymatic conversion of MSG (**b**). The strain was cultivated in MRS broth containing 118 mM MSG and incubated at 37 ◦C for 48 h. The culture broth was centrifuged and the supernatans were collected (**a**). Enzymatic conversion was performed with 0.64 U/mL purified GAD using 100 mM MSG in 50 mM citrate-phosphate buffer (pH 4.5) containing 0.2 mM of PLP. The reactions were carried out at 30 ◦C with 300 rpm agitation (**b**). All samples (microbial fermentations and enzymatic conversion) were hydrolyzed and derivatized prior to UPLC analysis.

#### **4. Conclusions**

In the present study, we compared GABA synthesis between microbial fermentation of *L. plantarum* FNCC 260, which was isolated from Indonesian fermented cassava, and enzymatic conversion of glutamate using recombinant glutamate decarboxylase (GAD) from *L. plantarum* FNCC 260 expressed in *E. coli*. MSG, PLP, and pyridoxine were shown to positively affect GABA production during the cultivations of *L. plantarum* FNCC 260. Enzymatic synthesis of GABA using purified recombinant GAD from *L. plantarum* FNCC 260 showed at least 5-fold higher GABA titres than microbial fermentations in a significantly shorter time. The newly isolated GABA-producing LAB is of great interest to extend the area of applications. *L. plantarum* FNCC 260 should be considered as a potential candidate for GABA production via both fermentation and enzymatic synthesis and can be also developed as functional starter culture.

**Author Contributions:** Conceptualization, I.B.A.Y., E.S.R., D.H. and T.H.N.; methodology, I.B.A.Y., S.K., T.H.N. and D.H.; software, I.B.A.Y.; validation, I.B.A.Y., S.K. and T.H.N.; formal analysis, I.B.A.Y. and S.K.; investigation, I.B.A.Y., S.K. and T.H.N.; resources, I.B.A.Y.; data curation, I.B.A.Y. and S.K.; writing—original draft preparation, I.B.A.Y.; writing—review and editing, I.B.A.Y. and T.H.N., D.H., K.J.D.; visualization, I.B.A.Y.; supervision, T.H.N., D.H., K.J.D. and E.S.R.; project administration, I.B.A.Y.; funding acquisition, D.H. and T.H.N. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Informed Consent Statement:** This study did not involve any human/animal interventions.

**Acknowledgments:** This study was part of the project titled "GABA-producing lactic acid bacteria from Indonesian fermented foods and its development for functional foods ingredients and starter cultures". We are also grateful for the Indonesian Endowment Fund for Education (LPDP) under Beasiswa Unggulan Dosen Indonesia-Luar Negeri (BUDI-LN) batch I 2016 for funding this study (grant number PRJ-3607/LPDP.3/2016). S.K. is thankful for the Ernst Mach—ASEA Uninet scholarship granted by the OeAD—Austrian Agency for International Cooperation in Education & Research and financed by the Austrian Federal Ministry of Science, Research and Economy. We would like to thank Clemens Grünwald-Gruber (Department of Chemistry, University of Natural Resources and Life Sciences-BOKU Wien) for his assistance in LC-ESI-MS analysis. The authors thank the EQ BOKU VIBT GmbH – Core facility for food and bio processing for providing the MALDI-TOF MS (Bruker Biotyper) for the identification of lactic acid bacteria.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**

