*3.1. Clean Water Generation, Spray Drying of Concentrated Ultrafiltered Whey and Characterization of the Obtained Fractions*

The first part of the project regarded the evaluation of a simple membrane-based process based on sequential UF and NF steps for the reduction of the organic load (BOD and COD) of whey that represents a substantial pollution source. The use of filtration is easily applicable in small/medium-sized companies that can not only reduce costs due to the disposal of numerous tons of discarded whey produced daily, but also reuse it for the separation of fractions enriched in lactose, proteins and peptides, and for the recovery of clean water. Results of the NF process are reported in Table 1a,b.

**Table 1.** Downstream processing of whey (**a**) Nanofiltration on 150–200 Da membranes of buffalo whey previously ultrafiltered on 20 kDa membranes. TMP, transmembrane pressure; LMH, L/m<sup>2</sup> ·h; (**b**) Composition change of buffalo whey during ultrafiltration and nanofiltration. \* Indicates the volume of permeate used for the nanofiltration experiments in the present study.


Thirteen liters of UF\_Ret20 were spray-dried in 6 h and resulted in the recovery of about 525 g of powder (Table 2). The residual water present in the sample resulted equal to 3.50 ± 0.50%. The powder was of thin and palpable grain size and contained prevalently lactose (Table 2). When suspended at 20 g/L, pH in bidistilled water was equal to 5.15 ± 0.05 and a conductivity of 3.18 ± 0.10 mS/cm was measured. Spray drying of the volume used in this work was affected by the void volume within the equipment, thus the yield was lower than 70%. However, the treatment of greater volumes on an industrial scale typically improves process yields, as the amount of solids lost in the spray dryer remains constant once a steady-state is achieved, and only the very fine powder that cannot be separated in the cyclone defines the actual process yield on solids.

**Table 2.** Characterization of spray-dried powder UF\_Ret20Pow. Lac, lactose; Gal, galactose; Glu, glucose; LA, lactic acid. Ins. Solids, insoluble solids.


*3.2. Evaluation of Ultrafiltered Spray-Dried Whey as Substrate for the Growth of L. fermentum*

The use of whey proteins for the growth of biotechnologically interesting microorganisms is well established [16]. For example, the probiotic strain *L. casei* is a well-known case study for the production of biomass [11] and other antimicrobial products such as nisin and bacteriocins [12,13]. Since lactic acid bacteria present specific and critical nutritional requirements, often supplementation with growth factors, vitamins and amino acids is necessary. *L. fermentum* DSM 20,049 was previously grown on whey with the addition of hydrolyzed lupin flour as an auxiliary nitrogen source in flask experiments, showing a shorter lag phase and a 70% higher biomass yield as compared to growth on MRS in the same conditions [24].

In the present study, buffalo milk-derived whey was evaluated as a substrate for the growth of an *L. fermentum* strain isolated from buffalo milk. In particular, the powder obtained from the spray-dried ultrafiltered retentate (UF\_Ret20Pow) was used in fermentation experiments to evaluate its potential as a one-pot medium; this would in fact strongly

simplify cultivation medium preparation and overall upstream processes. Since this fraction was not diafiltered it contained a large amount of sugars, in particular 44%, 8% and 6% of lactose, glucose and galactose, respectively, and about 6% of protein, a necessary nitrogen source for bacterial growth. Strain viability, sugars consumed and metabolic products produced (lactic acid, acetic acid and ethanol) were initially evaluated in bottle experiments. *L. fermentum* was cultivated on semi defined SGSL medium supplemented with different amounts of UF\_Ret20Pow. SGLS supplemented with glucose, the carbon source most efficiently used by *L. fermentum* [25], or lactose, the main sugar present in whey powder, were used as controls. As shown in Figure 2, a higher concentration of viable cells and related metabolic products (e.g., lactic acid, ethanol) were observed in relation to higher initial concentrations of ultrafiltered whey; in particular, in the presence of 4% UF\_Ret20Pow, the final average concentration of viable cells (8.85 ± 0.17 Log10 CFU/mL) was comparable to that obtained in control experiments on glucose (8.97 ± 0.14 Log10 CFU/mL), whereas it was significantly higher compared to results obtained in control experiments with lactose (8.43 ± 0.17 Log10 CFU/mL). Additional experiments on UF\_Ret20Pow dissolved in water (in the absence of SGLS medium components) showed that growth was still supported by the organic compounds present in the ultrafiltered whey fraction, although a significantly lower sugar consumption and biomass concentration were achieved (Figure 2), probably due to the lower amount of nitrogen source compared to that present in SGSL medium [25]. metabolic products produced (lactic acid, acetic acid and ethanol) were initially evaluated in bottle experiments. *L. fermentum* was cultivated on semi defined SGSL medium supplemented with different amounts of UF\_Ret20Pow. SGLS supplemented with glucose, the carbon source most efficiently used by *L. fermentum* [25], or lactose, the main sugar present in whey powder, were used as controls. As shown in Figure 2, a higher concentration of viable cells and related metabolic products (e.g., lactic acid, ethanol) were observed in relation to higher initial concentrations of ultrafiltered whey; in particular, in the presence of 4% UF\_Ret20Pow, the final average concentration of viable cells (8.85 ± 0.17 Log10 CFU/mL) was comparable to that obtained in control experiments on glucose (8.97 ± 0.14 Log10 CFU/mL), whereas it was significantly higher compared to results obtained in control experiments with lactose (8.43 ± 0.17 Log10 CFU/mL). Additional experiments on UF\_Ret20Pow dissolved in water (in the absence of SGLS medium components) showed that growth was still supported by the organic compounds present in the ultrafiltered whey fraction, although a significantly lower sugar consumption and biomass concentration were achieved (Figure 2), probably due to the lower amount of nitrogen source compared to that present in SGSL medium [25].

*Fermentation* **2021**, *7*, x FOR PEER REVIEW 7 of 11

MRS in the same conditions [24].

addition of hydrolyzed lupin flour as an auxiliary nitrogen source in flask experiments, showing a shorter lag phase and a 70% higher biomass yield as compared to growth on

growth of an *L. fermentum* strain isolated from buffalo milk. In particular, the powder obtained from the spray-dried ultrafiltered retentate (UF\_Ret20Pow) was used in fermentation experiments to evaluate its potential as a one-pot medium; this would in fact strongly simplify cultivation medium preparation and overall upstream processes. Since this fraction was not diafiltered it contained a large amount of sugars, in particular 44%, 8% and 6% of lactose, glucose and galactose, respectively, and about 6% of protein, a necessary nitrogen source for bacterial growth. Strain viability, sugars consumed and

In the present study, buffalo milk-derived whey was evaluated as a substrate for the

**Figure 2.** Small-scale experiments performed in 100 mL bottles at 37 °C and 150 rpm. Gal, galactose; Lac, lactose; Glc, glucose; LA, lactic acid; AA, acetic acid; EtOH, ethanol. Sup indicates media in which UF\_Ret20Pow was reconstituted in SGLS medium. Data were analyzed by two-tailed non homoscedastic Student's *t*-test. \* indicates *p* < 0.05 compared to results obtained on SGLS Glu; \*\* indicates *p* < 0.01 compared to results obtained on SGLS Glu; § indicates *p* < 0.05 compared to results obtained on SGLS Lac; §§ indicates *p* < 0.01 compared to results obtained on SGLS Lac. **Figure 2.** Small-scale experiments performed in 100 mL bottles at 37 ◦C and 150 rpm. Gal, galactose; Lac, lactose; Glc, glucose; LA, lactic acid; AA, acetic acid; EtOH, ethanol. Sup indicates media in which UF\_Ret20Pow was reconstituted in SGLS medium. Data were analyzed by two-tailed non homoscedastic Student's *t*-test. \* indicates *p* < 0.05 compared to results obtained on SGLS Glu; \*\* indicates *p* < 0.01 compared to results obtained on SGLS Glu; § indicates *p* < 0.05 compared to results obtained on SGLS Lac; §§ indicates *p* < 0.01 compared to results obtained on SGLS Lac.

With the aim of simplifying upstream and downstream procedures on an industrial scale by using buffalo milk waste as the only component of fermentation media, bioreactor experiments in controlled conditions were performed. *L. fermentum* probiotic biomass production was therefore investigated on UF\_Ret20 as a one-pot medium (powder reconstituted in water) and on UF\_Ret20 supplemented with SGSL salts, yeast extract and soy peptone (1/5th of that present in SGLS). Each experiment was performed at least in duplicate. Table 3 shows the results obtained. Controlled pH and constant air sparging improved viability that reached 8.1 ± 0.2 Log10 CFU/mL on the medium containing concentrated whey only. The addition of salts and of low amounts of complex N sources (soy peptone and yeast extract), yielded similar results, indicating an impact only on sugar consumption and LA production which increased (Table 3); LA, in particular, showed a 3.3 fold titer increase with a final concentration of about 10.0 ± 0.3 g/L and a final yield of 0.61 ± 0.03 g/g. Apparently, due to the lower amount of nitrogen source, the sugars were addressed to acid instead of biomass production [26].

**Table 3.** Data obtained by growing *L. fermentum* in batch in a Biostat CT plus (3 L) bioreactor reported as mean ± s.d. UF\_Ret20Pow indicates that the fermentation medium was obtained by reconstituting the UF\_Ret20Pow spray-dried powder in water; UF\_Ret20Pow sup indicates the additional presence of SGLS salts, yeast extract and soy peptone (2 g/L) in the medium. Lac, initial lactose; Gal, initial galactose; Glu, initial glucose; LA, lactic acid; AA, acetic acid; EtOH, ethanol. Data were analyzed by two-tailed non homoscedastic Student's *t*-test: \* *p* < 0.05; \*\* *p* < 0.01.

