The strains B. animalis subsp. lactis BI1 and B. breve BBR8 were a gift of the Centro Sperimentale del Latte S.r.l. (Lodi, Italy), the B. pseudocatenulatum M115 was from the Dairy Research Institute of Asturias (IPLA, Villaviciosa, Spain, Department of Science and Food Technology of the Spanish National Research Council, CSIC) and B. animalis subsp. lactis BB12 was previously isolated and included in the ISPA-CNR bacterial collection.

Fresh microbial cultures of bifidobacteria strains from frozen cultures (-80°C) were routinely grown in MRS (MRS Agar ISO Formulation, Biolife Italiana srl, Milan, Italy) amended with 0.5 g L^-1 of L-cysteine (MRSC) for 48h at 37°C under anaerobic conditions (ANAEROGEN, AN0025, Oxoid S.p.A., Milan, Italy).

Bifidobacteria strains were evaluated for their ability to grow in whey-based media (WBM). Whey protein isolate (WPI, Mirabol^® Whey Protein Natural 97, Volchem s.r.l., Grossa di Gazzo, Italy), containing ca. 97 g of whey proteins on 100 g of powder, was diluted in distilled water in order to obtain two WBM at 10 and 20 g L^-1 of total protein (WBM10 and WBM20). Based on Mirabol^®’s nutritional fact, the raw composition per liter of WBM10 included 5 g of β-lactoglobulin, 1.7 g of α-lactalbumin, 1.8 g of other whey proteins, 0.02 g of simple sugars, 0.04 g of lipids; the protein concentration in WBM10 allowed to estimate a content 2.3 and 0.5 g L^-1 of branched chain and sulfur-containing amino acids, respectively. In the case of WBM20 the concentration of nutrients doubled. In addition, WBM were supplemented with 0.5 g L^-1 of L-cysteine and 10 g L^-1 of lactose. Fresh bifidobacteria cultures were diluted in sterile saline solution in order to read an absorbance at 600 nm of 0.4 ± 0.05 (ca. 8 log cfu mL^-1), further diluted 1000 times in WBM10 and WBM20 and incubated as above described; MRSC was used as positive control. Then, bifidobacteria were enumerated on Bifidus Selective Medium (BSM) agar plates (Sigma–Aldrich SRL, Milan, Italy) recording pink-purple colored colonies. WBMs were also enriched with prebiotic fiber by the addition of 10 g L^-1 of chicory inulin with a degree of polymerization > 23 (Orafti HP, BENEO-Orafti, Tienen, Belgium) or 10 g L^-1 of retrograded RS (Novelose^® 330, Ingredion Incorporated, Westchester, IL, USA) and then pasteurized at 110°C × 5 min. In order to reduce whey protein precipitation, thermal treatment parameters were set up by preliminary assays; however, the efficacy of pasteurization was preserved. Prebiotic WBMs were cooled at room temperature and then inoculated with fresh cultures of bifidobacteria, as described above. Strains were incubated under anaerobic conditions at 37°C up to 96 h. Viable cell evaluations were carried out in triplicate every 24 h, as described above.

Two synbiotic whey protein fortified beverages enriched with prebiotic fiber (500 ml) were obtained, in triplicate, by supplementing WBM, at the protein concentration selected in the section “Growth of Bifidobacterium Strains in Whey-based Medium (WBM),” with inulin or with RS (1%). Glass bottles were filled with WPF beverages leaving less than 5 ml empty space on the top of the fluid, under the cork. After pasteurization and cooling, WPF beverages were inoculated with the bifidobacteria strains M115 or BBR8 (Figure 1, Step 3) at approximately 4 log cfu mL^-1 by using a fresh bifidobacteria culture.

WPF beverages were incubated at 37°C under anaerobic conditions in order to obtain a concentration of bifidobacteria of at least 8 log cfu mL^-1; depending on the results achieved in the fermentation assays, the incubation ranged from 48 to 72 h. After fermentation, WPF beverages were refrigerated at 4°C for 30 days. WPF beverages without bifidobacteria were also prepared as negative controls.

The concentration of viable cells in samples was calculated as the average number of colonies found for each decimal dilution, corrected by the dilution factor and expressed as the log cfu mL^-1 ± standard deviation. Bifidobacteria viable cell counts, RS and inulin content were analyzed by one-way analysis of variance (ANOVA) carried out using IBM SPSS Statistics 22 (IBM Corporation, Armonk, NY, USA). Tukey’s test and Fisher LSD post hoc test were applied in order to evaluate significant differences (P ≤ 0.05) among means of viable cell counts and dietary fibers, respectively. In order to compare inoculum levels as well as viable cell load recorded after the same period of incubation among different experiments, the Mann–Whitney non-parametric test was applied.

The independent effects and interactions of the main factors (times of storage, beverages and sensory attributes) on sensory scores were evaluated applying a three- and two-way ANOVA (P ≤ 0.05); multiple comparisons among individual means of the same strain were made by Fisher’s LSD post hoc test after rejecting the homogeneity of their variances with the Levene’s test (P ≤ 0.05).

The viable cell counts of Bifidobacteria, grown for 48 h at 37°C, in MRSC and in WBM supplemented with different amount of whey proteins (10 or 20 g L^-1; WBM10, WBM20), were reported in Table 1. WBM20 cultures of M115, BBR8, and BI1 showed cell counts significantly higher (P < 0.05) than those retrieved in WBM10, although all cultures reached the highest load values in MRSC (9.08 log cfu mL^-1 on average). This result was in accordance with preliminary experiments carried out using different amounts of whey protein concentrate at 80% of proteins (Baruzzi et al., 2014). Furthermore, BB12 cell counts (7.75 log cfu mL^-1, on average) in WBM10 and WBM20 were consistent with results of Matijević et al. (2008), who also reported no significant difference in cultures of this strain grown in medium containing 1.5 or 3% of whey protein concentrate. On the basis of these results, the 20 g L^-1 concentration was chosen for the subsequent analyses.

Aimed at promoting bifidobacteria growth at the similar levels of MRSC, used as a reference medium, within the same period of incubation, WBM enriched with total whey proteins to 20 g L^-1 (WBM20) was amended with additional carbon sources different from lactose.

Therefore, the WBM20 was supplemented with 10 g L^-1 of inulin (WBM20-I) or RS (WBM20-RS), as already assayed by Crittenden R. et al. (2001) and Rossi et al. (2005). Inulin and RS were selected as indigestible for humans but favor bifidobacteria growth. Figure 2 showed that the addition of 10 g L^-1 of inulin or the addition of 10 g L^-1 of RS to WBM20 (containing 10 g L^-1 of lactose) allowed to obtain microbial growth values similar to those occurred in MRS containing 20 g L^-1 of glucose only when time of incubation was extended.

However, bifidobacteria growth kinetics demonstrated to be different after WBM20 was supplemented with inulin or RS; in particular, BI1 was negatively affected by both probiotic fibers, whilst M115 was favored by RS. Inulin caused a significant and faster growth increase of both M115 and BBR8 strains. Mann–Whitney analysis also showed the viable cell concentrations of all strains at T0 and T48 throughout the experiments of steps 2 and 3 were similar (P > 0.05) at experimental steps 2 and 3 (Table 1 and Figure 1).

In accordance with previous studies, our results showed the bifidobacteria growth in presence of glucose (herein included in MRS) or other monosaccharides was improved in presence of complex carbohydrates (Palframan et al., 2003; Rossi et al., 2005; Rada et al., 2008).

Since amylolytic activity is present in approximately 60% of bifidobacteria species (Crittenden R. et al., 2001) only BB12 and M115 strains increased their viable cell loads after the addition of 10 g L^-1 of RS to WBM20. This finding agrees with Wronkowska et al. (2006) demonstrating either a good growth and an acidifying activity of bifidobacteria in fermented heat-treated starch. With regards to inulin supplementation, the higher cell loads were found only for the M115 and BBR8 strains in WBM20-I, compared to WBM20, and results were in accordance with Selak et al. (2016) reporting the heterogeneously inulin hydrolysis spread among Bifidobacteria species.

Moreover, in this work the growth of B. pseudocatenulatum M115 was favored by inulin-type fructans at each time of sampling over 72 h incubation suggesting the presence and expression of β-fructofuranosidase genes as already reported (Janer et al., 2004; Omori et al., 2010). In addition, M115 was the only strain able to grow better also in presence of RS; this behavior could be attributed to the occurrence of glycosyl transferase, starch, and glycoside hydrolase genes generally found in the genome of bifidobacteria (Liu et al., 2015) and particularly in the genome of the probiotic strain B. pseudocatenulatum IPLA 36007 (Alegría et al., 2014). These results increase the knowledge about the B. pseudocatenulatum M115 already proposed as a powerful probiotic strain thanks to its interesting pro-healthy features (Delgado et al., 2008; Losurdo et al., 2013).

In the light of previous results, the strain BI1 was excluded as both fibers reduced its viable cell load, conversely, the BBR8 was considered for beverage production only when WBM20 was supplemented with inulin as this fiber promoted its growth within 48 h.

As concerns BB12 and M115, the viable cell loads of both strains increased in WBM20-RS in comparison with WBM20 after 48 and 72 h, respectively. However, due to higher value reached by M115, this strain was selected and BB12 was excluded.

Thus, two synbiotic beverages fortified with whey proteins (at 20 g L^-1; WPF) were manufactured: the first beverage contained inulin and the strain BBR8, whilst the second one contained RS and it was fermented with the strain M115. After fermentation (48 h for the BBR8 and 72 h for the M115), WPF beverages were stored at 4°C evaluating viable cell count and pH (Figure 3) until day 30.

Confirming the results of preliminary fermentation assays, bifidobacteria reached a load by means of 8.5 log cfu mL^-1; then a reduction of ca. 2 log cycles was found in beverages within 15 days of cold storage. Subsequently, cell concentration remained stable for B. pseudocatenulatum M115, whereas a further reduction was observed in B. breve BBR8 reaching ca. 4.9 log cfu mL^-1 (Figure 3).

The fiber concentration in WPF beverages during 30 days of cold storage is reported in Table 2.

In the case of inulin, after its supplementation in WPF beverage, its concentration was 6.77 ± 0.36 g L^-1, far below the estimated amount reported on the product label (10 g L^-1 of Orafti^® HP, corresponding to ca. 9.9 g of pure inulin); this difference could be attributed to the different methods used to determine fructans, AOAC 997.08, as stated by the BENEO-Orafti, and the AOAC 999.03 of the Megazyme kit used in this work. At the end of cold storage inulin concentration halved (Table 2).

At contrary, the amount of RS remained quite stable (ca. 6 g L^-1) during both the manufacturing process and the first 15 days of cold storage (Table 2). The large standard deviation found in the concentration of RS could be ascribed to their low solubility in water phase; in addition, as reported by the K-RSTAR kit manufacturers, higher errors are expected for samples with RS content lower than 2%.

Since the concentration of RS did not change during preparation, fermentation, and cold storage, it is possible to argue that the growth improvement recorded for M115 (Figure 2) could be due to non-resistant polysaccharides supplied with Novelose^® 330.

At the end of fermentation, the total protein content of BBR8- or M115-fermented beverages, did not change significantly (P < 0.05) remaining stable at their initial average value of 4.19 ± 0.13 mg mL^-1. Conversely, the concentrations of small peptides and amino acids dropped from 161.6 ± 4.8 (control samples) to 11.99 ± 0.36 and 8.58 ± 0.26 μg mL^-1 for BBR8 and M115 strains, respectively. No changes in total protein content as well as in small peptide and amino acid contents were found in both WPF beverages during cold storage (data not shown).

As concerns the reduction of viability of BBR8 and MM115 strains during refrigerate period, our result agreed with those of Donkor et al. (2006) and Jayamanne and Adams (2009) reporting the reduction in viable cells of B. lactis LAFTI^® B94 or of B. longum and B. animalis ssp. lactis strains in yogurt over cold storage, respectively. Part of this decrease could be also attributed to the absence of L-cysteine that did not protect bifidobacteria cells against deleterious effects of cold storage, as demonstrated in refrigerated milk by Bolduc et al. (2006). We removed L-cysteine to avoid a negative influence on flavor. Refrigeration temperature hampered the increase in microbial load; in addition, cell viability was negatively affected by the post-acidification phenomenon as recently reported for six bifidobacteria cold stored for 21 days in milk (Djelled et al., 2016).

In our beverages the survival of bifidobacteria reached the same levels reported by Kailasapathy (2006) after cell microencapsulation, and was higher than that found by Antunes et al. (2005) in a probiotic yogurt containing a B. longum strain and enriched with both whey protein concentrate and skim milk powder. It is possible to argue that the occurrence of fibers in WPF beverages helped the survival of bifidobacteria as demonstrated by Crittenden R.G. et al. (2001) in yogurt supplemented with both RS and inulin, throughout 5 weeks of cold storage period. The inulin supplementation (40 mg g^-1 of reconstituted skim milk) has been already reported by de Souza Oliveira et al. (2011, 2012), to cause an increase in the viable cell load of a B. animalis subsp. lactis strain or in the biomass of lactic acid bacteria; however, no data were produced about potential inulin consumption. The decrease in the inulin content found during 30 days of cold storage (to about 40% of the initial amount) suggests that inulin was consumed by the strain. The reduction in the inulin content together with the drop in pH values could be consistent with the production of SCFA from this fiber as already demonstrated either by single strains (Marx et al., 2000) and fecal cultures (Pompei et al., 2008; De Vuyst and Leroy, 2011).

The bifidogenic effect of RS has been widely reported (Wang et al., 2002; Bouhnik et al., 2004; Lesmes et al., 2008); however, to the best of our knowledge, no studies were reported about the effect of RS on the growth of bifidobacteria monocolture and their fate in milk-based beverage under the experimental conditions reported in this work. It is interesting to note that the M115 strain reached a viable cell load in WPF amended with RS close to that found in different milk based foods and considered sufficient to provide therapeutic benefits (Shin et al., 2000; Pérez-Conesa et al., 2005). Differently from the WPF beverage amended with inulin, the stable content of RS during storage suggests that this WPF beverage could contribute to the adult dietary fiber daily intake.

As concerns protein content, the results of this work confirm the well known low proteolytic activity of Bifidobacterium strains, in particular when compared with that of lactic acid bacteria (Klaver et al., 1993), that in our case assimilated preferentially free amino acids for their metabolisms during fermentation.

Sensory evaluation showed the overall acceptability decreased from 7.8 to 4.6 and from 6.8 to 4.2 for WPF with inulin and RS, respectively. As concerns sensory descriptors, the three-way interaction between beverage, time of storage and sensory attributes was not statistically significant (p = 0.657); thus, the scores of attributes changed only in relation to the beverage and throughout the time of cold storage. Therefore, in order to examine the effect of the attributes in relations to the time on sensory score only two-way ANOVA analyses were statistically allowed (P < 0.00001) for each beverage. Indeed, the simple main effect analysis showed the sensory scores of the beverage M115 were significantly (P < 0.00001) higher than those registered by BBR8 only in relation to the mouthfeel notes (data not shown). By contrast, both beverages showed, on average, a highly significant (P < 0.01) increase in the taste score only after 30 days of cold storage (Figure 4).

The overall acceptability of WPF beverages, at the early stage of cold storage, showed values similar to those previously found in a probiotic yogurt containing skim milk powder and WPC (Antunes et al., 2005). The worsening in sensory attributes was previously found in a yogurt containing microencapsulated bifidobacteria cells (Kailasapathy, 2006). The sensory scores registered for the WPF beverages including RS and fermented with the M115 strain could be partially attributed to the positive influence of RS in accordance with results of Kailasapathy (2006) suggested as positive effect of capsulant and filler materials (alginate and Hi-Maize^TM Starch) in masking the grittiness (mouthfeel) of the yogurt fermented with bifidobacteria.