Chickpeas’ and Lentils’ Soaking and Cooking Wastewaters Repurposed for Growing Lactic Acid Bacteria

Legumes processing involves large amounts of water to remove anti-nutrients, reduce uncomfortable effects, and improve organoleptic characteristics. This procedure generates waste and high levels of environmental pollution. This work aims to evaluate the galacto-oligosaccharide (GOS) and general carbohydrate composition of legume wastewaters and assess their potential for growing lactic acid bacteria. Legume wastewater extracts were produced by soaking and/or cooking the dry seeds of chickpeas and lentils in distilled water and analysed using high-performance liquid chromatography with refractive index detection. GOS were present in all extracts, which was also confirmed by Fourier transform infrared spectroscopy (FTIR). C-BW extract, produced by cooking chickpeas without soaking, provided the highest extraction yield of 3% (g/100 g dry seeds). Lentil extracts were the richest source of GOS with degree of polymerization ≥ 5 (0.4%). Lactiplantibacillus plantarum CIDCA 83114 was able to grow in de Man, Rogosa, and Sharpe (MRS) broth prepared by replacing the glucose naturally present in the medium with chickpeas’ and lentils’ extracts. Bacteria were able to consume the mono and disaccharides present in the media with extracts, as demonstrated by HPLC and FTIR. These results provide support for the revalorisation of chickpeas’ and lentils’ wastewater, being also a sustainable way to purify GOS by removing mono and disaccharides from the mixtures.


Introduction
Pulses are an important component of the human diet and contain proteins, oligosaccharides, dietary fibre, minerals, and antioxidant polyphenols [1]. Although their consumption has important health benefits (e.g., the reduction of cholesterol and the prevention of the development of diabetes and cancer [1,2]), certain innocuous but uncomfortable associated effects (e.g., flatulence) can make their ingestion undesirable. Therefore, before consumption, soaking and cooking treatments are used to reduce these effects, as well as to enhance the bioavailability of important compounds and improve organoleptic characteristics, such as texture and flavour [3].
According to the latest definition, prebiotics are substrates selectively utilised by the host microorganisms conferring a health benefit [4]. Pulses contain various types of oligosaccharides with prebiotic effects, including galacto-oligosaccharides (GOS) and fructo-oligosaccharides (FOS), that are not absorbed or hydrolysed in the upper part of the gastrointestinal tract. GOS are composed of a varying number of galactose (Gal) units and a terminal glucose (Glu) or sucrose residue, resulting in different degrees of polymerization (DP). The type of GOS present in pulses, specifically α-GOS, belong to the raffinose-family ( Figure 1). A 1:5 (w/v, seeds: distilled water) ratio was used for the three treatments in both legumes.
Raw dried seeds were soaked for 8 h at 20 • C. After that, the seeds were strained, obtaining the waters from the soaking chickpeas (C-SW) and lentils (L-SW). The soaked chickpeas were then cooked in a pressure cooker (C-CW) and the soaked lentils in a pot (L-CW), both for 30 min. Thirdly, extracts were obtained by boiling dry (i.e., without soaking) chickpeas (C-BW) and lentils (L-BW) for 30 min in a pressure cooker and a pot, respectively. After the thermal treatments, the water extracts were cooled to 20-25 • C and centrifuged (15 min, 4000× g) (Heraeus Instruments, Hanau, Germany). The supernatants were filtered using filter funnels with a porosity of 4 (10-16 µm pores). Afterwards, the samples were freeze-dried in a Martin Christ Gefriertrocknungsanlagen GmbH freeze-dryer (Alpha 1-2 LD Plus, Osterode, Germany).

HPLC-RI Analysis
The extracts obtained in Section 2.2 were filtered with 0.45 µm of cellulose acetate filters (Frilabo, Maia, Portugal), and the carbohydrates were determined using high-performance liquid chromatography with refractive index (HPLC-RI) detection (UltiMate 3000, Dionex, Sunnyvale, CA, USA). A ReproGel-Na column (Dr. Maisch, Ammerbuch, Germany) of 250 × 8 mm and a particle size of 9 µm was used with a CARBOSep CHO 411 pre-column (Concise Separations, San Jose, CA, USA) at 80 • C. The RI detector (Shodex RI-101) was maintained at 50 • C. Degasified filtered ultrapure water was used as mobile phase with a flow rate of 500 µL/min. The obtained chromatograms were analysed with Chromeleon 6.80 software (Dionex Corporation, Sunnyvale, CA, USA).
Because of the absence of GOS standards in the market, different concentrations of Vivinal ® GOS syrup were used as standards, to determine the retention times of the sugars of interest and to calibrate the analytical method used. The syrup was composed of Gal, Glu, lactose (Lac), and short chain β-GOS (up to DP = 7) as shown in Table 1 (Friesland Campina DOMO, Amersfoort, The Netherlands, 2017) [19], alongside the retention times determined for each carbohydrate species. A calibration curve for fructose (Fru), whose retention time is 11.4 min, was obtained from a previous work [20]. The HPLC-RI calibration is detailed in Supplementary Material S1. From the determined calibration curves, the sugar composition of the samples was expressed as g/100 g of fresh extract. a GOS species with DP = 7, 6, and 5 were treated as one due to poor resolution of the obtained peaks. b Lac present in Vivinal ® GOS syrup was used as a reference for other similar DP = 2 carbohydrates (i.e., disaccharides) with the same retention time. c Total GOS indicates the sum of GOS DP = 3, DP = 4, DP = 5, DP = 6, and DP = 7 s contents.

Extraction Yield
The extraction yields were calculated based on the HPLC-RI analysis and the concentration (g/100 g of fresh extract) determined for each carbohydrate detected. The sugar content on a dry basis (g/100 g of dry extracts) and their extraction yield (g/100 g of dry seeds) were calculated by Equations (1) and (2), respectively, after performing • Brix measurements (RX-100, Atago digital refractometer) and determining the volume of fresh extract obtained. The yields for GOS-DP ≥ 5, GOS-DP = 4, GOS-DP = 3, Lac, Glu, Gal, and Fru's extraction were calculated separately from each calibration curve.

FTIR Analysis
The freeze-dried samples obtained in Section 2.2 were analysed by a Fourier transform infrared spectrometer (Spectrum Two, Perkin Elmer, Waltham, MA, USA) with attenuated total reflectance (ATR) equipment with a diamond crystal (UATR Two, Perkin Elmer, Waltham, MA, USA). The spectra were registered in the 4000-400 cm −1 range by co-adding 32 scans with 4 cm −1 spectral resolution at 20 • C. The spectra were analysed using spectrum software (Perkin Elmer, Waltham, MA, USA). The Vivinal ® GOS syrup was also analysed as a reference material for a complex carbohydrate mixture.

Microbiological Assays with Oligosaccharide Mixtures
Lactiplantibacillus plantarum CIDCA 83114 was isolated from kefir grains [21] and maintained frozen at −80 • C in 120 g/L of non-fat milk solids. Cultures were grown in an MRS broth [22] at 37 • C overnight in aerobic conditions to obtain approximately 10 10 −10 11 of CFU/mL (stationary phase). Then, they were harvested by centrifugation at 10,000× g for 10 min at 4 • C, and the pellets were washed twice with a phosphate buffered saline (PBS) solution (K 2 HPO 4 0.144 g/L; NaCl 9.00 g/L; Na 2 HPO 4 0.795 g/L, pH 7) and used to evaluate the microbial growth potential of the water extracts derived from chickpeas and lentils (Section 2.2).
To that aim, the microorganisms were inoculated (1% v/v) in MRS (5 mL) broth formulated without glucose (composition in Supplementary Material S2) and supplemented at 0.3% w/v, either with the extracts obtained in Section 2.2 or with Gal, Glu, Lac, or Vivinal ® GOS syrup. All the solutions were sterilized by filtration (0.45 µm pore filter diameter, Frilabo, Maia, Portugal). Blank controls were carried out by inoculating the strain in MRS without the addition of extracts or sugars. The samples were incubated at 37 • C for 24 h and then serially diluted in PBS, plated on MRS agar, and incubated at 37 • C for 48 h in aerobic conditions. The results were expressed as colony-forming units (CFU) per millilitre (CFU/mL). To calculate the growth potential of the samples, the CFU/mL values for the samples were referred to those of the blank control (MRS without sugars) (Equation (3)).
log CFU/mL = log CFU/mL Sugar − CFU/mL Blank The consumption of carbohydrates during fermentation was evaluated by HPLC-RI and FTIR, following similar protocols as those explained in Sections 2.3 and 2.5. Figure 1 shows a summary of the experimental procedures. A final microbiological experiment was performed by growing L. plantarum CIDCA 83114 at 37 • C for 24 h in a C-BW extract prepared in distilled water at 51 g/L (MRS concentration). After growing, the culture was centrifuged and the supernatant was filtered and analysed in the HPLC-RI for comparison with the initial sample. The FTIR spectra were also recorded for the supernatant after lyophilization and compared with the C-BW extract's spectrum obtained in Section 2.5.

Statistical Analysis
The obtained results were evaluated by one-way analysis of variance (ANOVA), using InfoStat software. When this analysis expressed statistical differences (p < 0.05), intragroup comparisons were tested using the Tukey test.

Statistical Analysis
The obtained results were evaluated by one-way analysis of variance (ANOVA), using InfoStat software. When this analysis expressed statistical differences (p < 0.05), intragroup comparisons were tested using the Tukey test.

HPLC-RI Results and Extraction Yield
The chromatograms obtained from the fresh extracts of chickpeas and lentils exhibited similar features, especially in the boiling water extracts ( Figure 2) and resembled those of Vivinal ® GOS syrup (S1). Considering that the same seeds-to-water weight proportion was used in all procedures, the results indicate that chickpeas are a richer source of saccharides than lentils, because the chromatograms showcase a higher number of peaks, and those in common have higher intensities. GOS were found in higher quantities in the chickpea extracts, and lentils provided much fewer mono-and disaccharides. Fru was only detected in the lentil extracts; however, the juxtaposition of its peak (11.4 min) with that of Gal (11.2 min) may cause its eclipsing in chickpea waters. In all the chromatograms peaks found between 7 and 8 min and at around 9 min have not yet been identified. The prominent peak present in all samples at around 5 min can be attributed to a Glu-composed polysaccharide, most likely starch. This inference is supported by the observation of starch in the presence of iodine during an assay conducted on the legumes' wastewaters from the present study. The chromatograms' resemblance to that of Vivinal (S1) highlights the relevance of its use for HPLC-RI calibration, in view of the absence of GOS standards in the market.

HPLC-RI Results and Extraction Yield
The chromatograms obtained from the fresh extracts of chickpeas and lentils exhibited similar features, especially in the boiling water extracts ( Figure 2) and resembled those of Vivinal ® GOS syrup (S1). Considering that the same seeds-to-water weight proportion was used in all procedures, the results indicate that chickpeas are a richer source of saccharides than lentils, because the chromatograms showcase a higher number of peaks, and those in common have higher intensities. GOS were found in higher quantities in the chickpea extracts, and lentils provided much fewer mono-and disaccharides. Fru was only detected in the lentil extracts; however, the juxtaposition of its peak (11.4 min) with that of Gal (11.2 min) may cause its eclipsing in chickpea waters. In all the chromatograms peaks found between 7 and 8 min and at around 9 min have not yet been identified. The prominent peak present in all samples at around 5 min can be attributed to a Glu-composed polysaccharide, most likely starch. This inference is supported by the observation of starch in the presence of iodine during an assay conducted on the legumes' wastewaters from the present study. The chromatograms' resemblance to that of Vivinal (S1) highlights the relevance of its use for HPLC-RI calibration, in view of the absence of GOS standards in the market. Tables 2-4 show the GOS, DP2 sugars, Glu, Gal, and Fru contents in g/100 g of fresh and dry extract and the extraction yield in g/100 g of dry seeds used for the soaking and cooking, respectively. Fresh extract represents the extract after the centrifugation and filtration steps, which is useful to analyse the sample for direct utilization. Dry extract is expressed on a dry basis and facilitates the manipulation and storage steps.
When evaluating the soaking and cooking treatments, soaking provided the lowest extraction yields, but the obtained mixtures were purer, since the measured ºBrix values (total dissolved solids) of the fresh extracts were closer to the calculated total quantifiable sugar content. C-BW was the fresh extract with the highest total GOS, total mono-and disaccharides, and total saccharides, when compared with the rest of treatments and with lentil extracts. The soaking methods are effortless, and all the extracted material is highly hydrophilic and water soluble. Contrarily, the cooking (C-CW) and boiling (C-BW and L-BW) processes enable the extraction of compounds with lower solubility [9], resulting in a significantly lower concentration of total GOS in the dry extracts (Table 3). It is worth noting that the combined contents of GOS-DP ≥ 5 and DP = 4 are the same for all chickpea extracts (16 g/100 g of dry extract); however, C-SW possesses almost 10% more GOS-DP = 3, twice the amount of DP2 sugars, and thrice the amount of Glu and Gal than C-CW and C-BW. This trend was not observed for the lentil-derived extracts, with the total saccharide content being approximately 36-40% in all the dry extracts. L-CW was the lentil extract with the lowest total mono-and disaccharides; however, Fru was detected but was not quantified because its area was below the limit of quantification. According to the USDA [23], raw lentil possesses up to 0.27 g of Fru/100 g, and L-BW extraction produced 0.04 g of Fru/100 g of lentils; for chickpea, no Fru values are reported by the USDA [23], and no extraction of this sugar was observed. When comparing the dried extracts of both legumes, it can be noted that the total GOS, total mono-and disaccharides, and total saccharides content were significantly lower in lentils than in chickpeas.  Tables 2-4 show the GOS, DP2 sugars, Glu, Gal, and Fru contents in g/100 g of fresh and dry extract and the extraction yield in g/100 g of dry seeds used for the soaking and cooking, respectively. Fresh extract represents the extract after the centrifugation and filtration steps, which is useful to analyse the sample for direct utilization. Dry extract is expressed on a dry basis and facilitates the manipulation and storage steps.
When evaluating the soaking and cooking treatments, soaking provided the lowest extraction yields, but the obtained mixtures were purer, since the measured ºBrix values (total dissolved solids) of the fresh extracts were closer to the calculated total quantifiable sugar content. C-BW was the fresh extract with the highest total GOS, total mono-and According to the USDA [23], raw lentil possesses 0.27 g of Fru/100 g, and L-BW extraction produced 0.04 g of Fru/100 g of lentils, while no values are reported for chickpea.
In terms of the total GOS production, as seen in Table 4, the C-BW was the most interesting extract, with a 3% extraction yield (i.e., 3 g of the total GOS (of which half are DP = 3 sugars) were produced out of 100 g of dry chickpea seeds used), and without the need for soaking the legume. Similarly, the L-BW extract stands out as the most efficient source for GOS-DP ≥ 5's extraction, specifically, producing 0.4 g out of 100 g of dry legume seeds. This results in less utilization of resources and generates fewer waste products. In terms of total mono-and disaccharides, chickpeas' extract obtained significantly higher values than lentils'. This is also true for the total GOS content, except for L-BW.
Serventi [24] evaluated the cooking water of legumes (after soaking), finding that this water possesses high amounts of oligosaccharides. In the present study, it was observed that treatments that involve the use of heat (cooking and boiling) led to a higher extraction of GOS than soaking, which can be correlated with the results of Liu and Serventi [25], who showed that the process of cooking after soaking can lead to a loss of 60-85% of oligosaccharides in legumes, compared to soaking (50-75%).     Table 4. Chickpeas' and lentils' extraction yields expressed as g of sugar/100 g of dried seeds *.

C-SW C-CW C-BW L-SW L-CW L-BW
GOS-DP ≥ 5 0.03 ± 0.00 0.10 ± 0.00 0.14 ± 0.01 0.12 ± 0.00 0. 25 Figure 3 shows the FTIR spectra obtained for the dried chickpeas' and lentils' extracts, and for the reference material, Vivinal ® GOS syrup. The spectra of all the extracts were very similar and superimposable. This indicates that the soaking process has no noteworthy qualitative effect on the biomolecules detected by a spectral analysis, meaning that both soaking and cooking processes enable the extraction of similar compounds. The spectrum obtained for the Vivinal ® GOS syrup, which corresponds to a mixture of β-GOS, Lac, Glu, and Gal, was also similar to those of the extracts. The bands shared between the samples and the Vivinal ® GOS syrup included the broad one at 3500-3000 cm −1 (OH stretching) and those in the fingerprint region (1200-800 cm −1 ), arising from the C-O-C glycosidic linkage, the COH bending, and the C-C stretching vibrational modes, that collectively provide a characteristic pattern for each carbohydrate [26,27]. The differences observed between the Vivinal ® GOS syrup and the extracts in this region may be related to the fact that the former does not contain polysaccharides. As stated before, we hypothesize that the peak at 5 min observed in the HPLC chromatograms of the extracts corresponds to starch. This conclusion is further supported by the FTIR analysis. Romano et al. [28] also observed starch-related bands in the 1250-800 cm −1 region when evaluating the FTIR spectra of quinoa flour. The main differences between the samples' and the Vivinal ® GOS syrup's spectra were related to the relative intensities of the bands observed in the stillundiscussed double-bond stretching and local symmetry regions, around 1800-1500 cm −1 and 1500-1200 cm −1 , respectively. The bands detected in the former can be attributed to the presence of unsaturated bonds (e.g., in the C=O groups found in carbohydrates) and unspecific CH 2 bending vibrations, whereas those found in the latter can be ascribed to vibrations arising from C-O groups, also observed in carbohydrates [29]. Figure 4 shows the results of the 24 h bacterial growth assays obtained for all the extracts, the standards, and the Vivinal ® GOS syrup used as a reference material. All the sample extracts were capable of promoting the growth of L. plantarum CIDCA 83114, showcasing a growth potential comparable to that of the standard sugars assayed. The chickpea extracts were the most successful in this regard, leading to bacterial counts close to those obtained for Glu (the sugar present in the standard MRS medium composition) and for Vivinal ® GOS syrup, which is composed of β-GOS, unlike α-GOS present in the extracts. L-BW was the best extract from the lentils' counterpart.

Microbiological Assay with Oligosaccharide Mixtures
the FTIR spectra of quinoa flour. The main differences between the samples' and the Vivinal ® GOS syrup's spectra were related to the relative intensities of the bands observed in the still-undiscussed double-bond stretching and local symmetry regions, around 1800-1500 cm −1 and 1500-1200 cm −1 , respectively. The bands detected in the former can be attributed to the presence of unsaturated bonds (e.g., in the C=O groups found in carbohydrates) and unspecific CH2 bending vibrations, whereas those found in the latter can be ascribed to vibrations arising from C-O groups, also observed in carbohydrates [29].  The joint content of mono-and disaccharides does not explain the observed results, where extracts such as C-BW-which only presents 20% of these sugars-showcase bacterial counts close to those obtained for Glu (100%), at 8.2 and 8.7 log CFU/mL, respectively. This indicates that other sugars present, namely GOS, are being utilized by the microorganisms. However, when comparing the total sugar content (total quantifiable sugars) of C-BW (60%) with that of the Vivinal ® GOS syrup (100%) this correlation falls short. There are two possible explanations. The first is that L. plantarum CIDCA 83114 is capable of more efficiently utilising α-GOS (such as those present in the extracts), than β-GOS (such as those found in the Vivinal ® GOS syrup) for its growth, which has been previously reported for some probiotic cultures. Oh et al. [30] studied the growth effect of different αand β-GOS on several non-probiotic and probiotic bacterial strains and determined that the oligosaccharide structure influences their growth, even amongst different strains of the same species. The second, and most likely reason, is that the unknown polysaccharide found in the extracts (Figure 2) is also being metabolized by L. plantarum CIDCA 83114. extracts, the standards, and the Vivinal ® GOS syrup used as a reference material. All the sample extracts were capable of promoting the growth of L. plantarum CIDCA 83114, showcasing a growth potential comparable to that of the standard sugars assayed. The chickpea extracts were the most successful in this regard, leading to bacterial counts close to those obtained for Glu (the sugar present in the standard MRS medium composition) and for Vivinal ® GOS syrup, which is composed of β-GOS, unlike α-GOS present in the extracts. L-BW was the best extract from the lentils' counterpart. The joint content of mono-and disaccharides does not explain the observed results, where extracts such as C-BW-which only presents 20% of these sugars-showcase bacterial counts close to those obtained for Glu (100%), at 8.2 and 8.7 log CFU/mL, respectively. This indicates that other sugars present, namely GOS, are being utilized by the microorganisms. However, when comparing the total sugar content (total quantifiable sugars) of C-BW (60%) with that of the Vivinal ® GOS syrup (100%) this correlation falls short. There are two possible explanations. The first is that L. plantarum CIDCA 83114 is capable of more efficiently utilising α-GOS (such as those present in the extracts), than β-GOS (such as those found in the Vivinal ® GOS syrup) for its growth, which has been previously reported for some probiotic cultures. Oh et al. [30] studied the growth effect of different α-and β-GOS on several non-probiotic and probiotic bacterial strains and determined that the oligosaccharide structure influences their growth, even amongst different strains of the same species. The second, and most likely reason, is that the unknown polysaccharide found in the extracts ( Figure 2) is also being metabolized by L. plantarum CIDCA 83114.
To answer this question, L. plantarum CIDCA 83114 was grown at 37 °C for 24 h in C-BW (51 g/L) and the final medium was analysed in the HPLC-RI ( Figure 5). It was observed that Glu (10.5 min) and Gal (11.4 min) peaks entirely disappeared, while that of DP = 2 sugars (8 min) greatly decreased its intensity after fermentation. The intensity of To answer this question, L. plantarum CIDCA 83114 was grown at 37 • C for 24 h in C-BW (51 g/L) and the final medium was analysed in the HPLC-RI ( Figure 5). It was observed that Glu (10.5 min) and Gal (11.4 min) peaks entirely disappeared, while that of DP = 2 sugars (8 min) greatly decreased its intensity after fermentation. The intensity of GOS peaks at around 6-7 min was not altered. The increase in the GOS-DP = 4 peak (6.3 min) can be explained considering the formation of L-lactic acid, whose retention time is similar to that of GOS-DP = 4. After fermentation, the C-BW extract's chromatogram had 97% of the total area of the initial C-BW sample before fermentation, and a 51 g/L C-BW sample had 18.7 g/L total GOS. As GOS were unaffected by the bacterial metabolic activity, they are expected to still be present at this concentration after fermentation. Interestingly, the polysaccharide peak at around 5 min decreased its intensity. Therefore, it can be concluded that L. plantarum CIDCA 83114 preferably utilises sugars where Glu is present. As previously mentioned, the polysaccharide is very likely starch, a molecule composed of Glu units, whereas GOS-in particular RFO-at most contain only one Glu residue. This explains the values of CFU/mL for extracts such as C-CW, C-BW, and L-BW (Figure 4), that showed low concentrations of mono-and disaccharides (as a whole) but whose chromatograms ( Figure 2) exhibited a very prominent polysaccharide peak. Future investigations will delve further into this matter.
After fermentation, the C-BW sample (sans the bacteria) was freeze-dried and analysed using FTIR (Figure 6, top). Although the spectrum is pretty much similar to that of the initial C-BW extract (Figure 3), it exhibited some new bands, which are denoted in Figure 6. The most noticeable change was the shoulder occurring at 1723 cm −1 , probably arising from a C=O stretching vibrational mode. Such a functional group is possibly due to the formation of L-lactic acid, as previously identified in the HPLC-RI analysis ( Figure 5). The presence of L-lactic acid was further confirmed by spiking the initial C-BW extract powder with a small amount of L-lactic acid and analysing the mixture using FTIR (Figure 6, middle spectrum). As expected, an increase in some bands (corresponding to L-lactic acid) was observed and such bands corresponded to those new bands previously observed for C-BW fermented with L. plantarum CIDCA 83114. These bands were definitely confirmed as belonging to L-lactic acid by analysing the spectrum of the pure compound ( Figure 6, bottom), which resembled that found in the National Institute of Standard and Technology database [31].
These results show that legume wastewaters, such as C-BW alone, have the nutritional requirements for growing L. plantarum CIDCA 83114, and GOS content does not change according to the fermentation process for this strain and GOS are still available at the end of the process. The fermentation of legume extracts with bacterial strains can therefore be an efficient purification method to remove mono-and disaccharides from α-GOS, before their employment in other industries and applications. If legume extracts were to be incorporated in food products, the functionality of legume extracts could be improved by the addition of probiotic bacteria with beneficial effects for human health.
GOS peaks at around 6-7 min was not altered. The increase in the GOS-DP = 4 peak (6.3 min) can be explained considering the formation of L-lactic acid, whose retention time is similar to that of GOS-DP = 4. After fermentation, the C-BW extract's chromatogram had 97% of the total area of the initial C-BW sample before fermentation, and a 51 g/L C-BW sample had 18.7 g/L total GOS. As GOS were unaffected by the bacterial metabolic activity, they are expected to still be present at this concentration after fermentation. Interestingly, the polysaccharide peak at around 5 min decreased its intensity. Therefore, it can be concluded that L. plantarum CIDCA 83114 preferably utilises sugars where Glu is present. As previously mentioned, the polysaccharide is very likely starch, a molecule composed of Glu units, whereas GOS-in particular RFO-at most contain only one Glu residue. This explains the values of CFU/mL for extracts such as C-CW, C-BW, and L-BW (Figure 4), that showed low concentrations of mono-and disaccharides (as a whole) but whose chromatograms ( Figure 2) exhibited a very prominent polysaccharide peak. Future investigations will delve further into this matter. After fermentation, the C-BW sample (sans the bacteria) was freeze-dried and analysed using FTIR (Figure 6, top). Although the spectrum is pretty much similar to that of the initial C-BW extract (Figure 3), it exhibited some new bands, which are denoted in Figure 6. The most noticeable change was the shoulder occurring at 1723 cm −1 , probably arising from a C=O stretching vibrational mode. Such a functional group is possibly due to the formation of L-lactic acid, as previously identified in the HPLC-RI analysis ( Figure  5). The presence of L-lactic acid was further confirmed by spiking the initial C-BW extract powder with a small amount of L-lactic acid and analysing the mixture using FTIR ( Figure  6, middle spectrum). As expected, an increase in some bands (corresponding to L-lactic acid) was observed and such bands corresponded to those new bands previously observed for C-BW fermented with L. plantarum CIDCA 83114. These bands were definitely confirmed as belonging to L-lactic acid by analysing the spectrum of the pure compound ( Figure 6, bottom), which resembled that found in the National Institute of Standard and Technology database [31].
These results show that legume wastewaters, such as C-BW alone, have the nutritional requirements for growing L. plantarum CIDCA 83114, and GOS content does not change according to the fermentation process for this strain and GOS are still available at the end of the process. The fermentation of legume extracts with bacterial strains can therefore be an efficient purification method to remove mono-and disaccharides from α-GOS, before their employment in other industries and applications. If legume extracts were to be incorporated in food products, the functionality of legume extracts could be improved by the addition of probiotic bacteria with beneficial effects for human health.

Conclusions
Legumes are a rich source of prebiotic compounds, such as GOS. Utilising the wastewaters produced during the soaking and cooking of chickpeas and lentils proved to be a cost-effective and efficient method for their extraction and recovery. Cooking

Conclusions
Legumes are a rich source of prebiotic compounds, such as GOS. Utilising the wastewaters produced during the soaking and cooking of chickpeas and lentils proved to be a cost-effective and efficient method for their extraction and recovery. Cooking chickpeas provided the highest GOS extraction yields, while lentils' cooking waters were the richest source of GOS-DP ≥ 5 compounds, with low concentrations of mono-and disaccharides. Microbial experiments carried out with L. plantarum CIDCA 83114 showed that legume wastewaters can be valorised and effectively repurposed as microbiological growth media for lactic acid bacteria, leading to less waste disposal. Considering that this strain consumed the monosaccharides and greatly diminished the DP = 2 carbohydrates present in the C-BW extract, its use could be further explored as a potential purification method for GOS mixtures.
The obtained results allow for the elucidation of the initial aspects of the use of wastewaters from the processing of legumes, in order to avoid environmental contamination and obtain compounds with prebiotic activity, mainly at the laboratory level. The simplicity of this approach and its ease of scalability make it highly suitable for future implementation by industrial stakeholders.