Elsevier

LWT

Volume 133, November 2020, 110119
LWT

Towards a starter culture of Lactobacillus plantarum AFS13: Assessment of more relevant effects for in vitro production and preservation thereof, via fractional factorial design methodology

Highlights

•

A native strain from brined table olives manufactured in Douro region was selected.

•

Biomass growth and preservation conditions were optimized toward a starter culture.

•

Optimum growth was under anaerobiosis, at higher temperature, and in the absence of NaCl.

•

Uncontrolled pH, and lowest stirring rate and glucose concentration were more economical.

•

Lyophilized cell viability was essentially unaffected up to 4 mo of storage.

Abstract

Spontaneous fermentation by native strains of lactic acid bacteria has traditionally been used to preserve table olives, by taking advantage of a proactive and competitive reduction of growth of unwanted microorganisms. Starter cultures have proven effective in stabilizing fermentation processes at large. Despite the many starter cultures available commercially, few have originated on the native microflora – and table olives have seldom been processed via deliberate inoculation with them. This study screened variables relevant for the growth of Lactobacillus plantarum AFS13 originally obtained from brines in Alto Douro (Portugal); and for preservation thereof as the starter, should it prove appropriate. A 2IV6−2 factorial design was used to assess six parameters (and their interactions): 19 fermentation batches were carried out to ascertain growth rates. Temperature and aeration degree were the most significant processing parameters. Maximum growth rate, 0.356 h−1, was attained under anaerobic conditions, at 37 °C, and in the absence of NaCl; ca. 9 log10 CFU mL−1 was reached by 6 h. Glucose concentration, pH, and stirring rate proved not significant, so economic considerations should fix their optimum levels. The starter culture could be preserved for up to 4 mo after freeze-drying in sucrose, with an acceptable reduction in cell viability.

Keywords

Bioreactor
Biomass
Storage
Freeze-drying
Lactic acid bacteria

1. Introduction

The adventitious microflora in spontaneous fermentation is often poorly efficient and hardly predictable; and labile to subsequent technological treatments. A few recent studies have focused on the possible use of wild strains, previously isolated from traditional products, in the manufacture of actual starter cultures – in attempts to improve technological and functional features of the final product (Todorov & Holzapfel, 2015).

To be useful in food fermentation, the strains selected are to satisfy several characteristics (Bonatsou, Tassou, Panagou, & Nychas, 2017), namely intrinsic features and technological characteristics of the microorganism, and safety and benefits for the human being. The final fermented product is indeed dependent on the quality and quantity of the microbiota provided, in the first place. However, promising strains – as indicated by their (optimal) performance at the laboratory scale, may perform differently in actual fermentations, due to the mode of inoculum preparation. Therefore, it is more attractive to use starter cultures already acclimatized to the specific fermentation conditions, as this will more likely lead to a successful product (Perricone, Corbo, Sinigaglia, & Bevilacqua, 2013).

The bioreactor is the core element in any bioprocess aimed at producing starter cultures to a large scale; therefore, optimization of its operation is in order. The most common type of bioreactor is “in-line” – suitable for monitoring (and thereby controlling) pH, stirring speed, airflow rate, temperature, pressure, dissolved oxygen (DO), dissolved carbon dioxide, weight/level, and degree of foaming (Stanbury, Whitaker, & Hall, 2017). Quantifying cell concentration in the culture medium is, in turn, crucial to determine both kinetics and stoichiometry of microbial growth. The methods used thereto can be classified in two categories: direct – which resort to an optical density (OD), packed cell volume, dry weight, or direct cell counting; and indirect – which permit an inference of biomass levels based on substrate consumption and/or product formation (Liu, 2002).

Design of fermentation media aimed at improving product yield, decreasing medium cost, and preventing the growth of unwanted microorganisms is a sine qua non for a fermented product to retain its competitive edge in the marketplace (Stanbury et al., 2017). However, most media used at industrial scale often lead to considerable batch-to-batch variation, and thus unpredictable biomass and/or product yields. This is so because they are normally complex formulations of poorly defined ingredients, selected chiefly for being non-expensive. Chemically defined media have meanwhile become quite popular, despite being more expensive and demanding longer developing periods. In fact, pure substrates lead to more predictable yields among batches and are less demanding in recovery, purification, and effluent treatment (Stanbury et al., 2017).

Storage of starter cultures should assure their viability and activity while eliminating genetic change and protecting against contamination (Stanbury et al., 2017). It is widely known that a high water activity plays a major role in the degradation of stored material, as it favors the autolysis or growth of spoilage microorganisms. Therefore, it is necessary to reduce (or somehow immobilize) the water inventory of stored materials, with the ultimate goal of stabilizing the culture (Wolkers & Oldenhof, 2015).

The most widely applicable method to keep microorganisms for long periods is cryopreservation (Wolkers & Oldenhof, 2015); however, storing and transporting samples of frozen material is costly. Therefore, cultures are preferably kept in dehydrated form – obtained upon drying (using hot air) or freeze-drying (using low-temperature sublimation) (Stanbury et al., 2017; Wolkers & Oldenhof, 2015). The latter resorts to vacuum desiccation – and produces a dry, active, shelf-stable, and readily suspendable product, yet containing a combination of dead, viable, and sublethally injured cells to variable proportions (Stanbury et al., 2017). Due to lack of structure of the departing growth medium, the resulting powder entails low-cost distribution at suprazero temperatures – and enables long-term storage; in addition, viability is promptly restored following rehydration. Protection from injury during freezing, and subsequent sublimation can be achieved via previous addition of sugars, either individually or in combination with other solutes; this is a consequence of the replacement of water-forming hydrogen bonds by specific biomolecules.

Starter cultures are far from being routinely utilized in table olive fermentations (Corsetti, Perpetuini, Schirone, Tofalo, & Suzzi, 2012); nevertheless, tailor-designed starter cultures may bring about several technological advantages. Lactic fermentation appears to be the most important step in table olive fermentation – because it debitters olives, decreases brine pH, and improves consistency and sensory treats able to meet consumers’ expectations (Bonatsou et al., 2017). The strains playing the more relevant role are those able to attach onto the olive epidermis and survive (or even grow) under increasing salt concentrations and decreasing pH (Bonatsou et al., 2017). The most prominent strains in this regard are Lactobacillus plantarum and Lactobacillus pentosus – both belonging to the lactic acid bacteria (LAB) group (Comunian et al., 2017); they are effective at viable numbers above 105–106 CFU mL−1 (Bleve et al., 2015).

This study aimed at finding relevant processing variables, using (higher) maximum specific growth rate as objective function – and resorting to L. plantarum strain ASF13 as model (adventitious) strain. A batch in-line bioreactor was accordingly employed, and the preservation of a posteriori in powdered form was tested.

2. Materials and methods

2.1. Microorganism

Lactobacillus plantarum strain ASF13, isolated by our group from table olive brines during artisanal manufacture in Alto Douro region, was selected owing to its technological and probiotic potential (unpublished results: Malheiro, 2018 and Portilha-Cunha, 2019; Research Project Reports, Bioengineering Master Degree). The strain was identified by comparing its 16S rRNA gene sequence with those of type strains available in the EzBioCloud database (https://www.ezbiocloud.net/). Pure cultures were maintained at −80 °C in 15% glycerol and routinely propagated on de Man, Rogosa, and Sharpe (MRS) agar plates.

2.2. Production of biomass

2.2.1. Experimental design

A two-level, fractional factorial design in 6 variables (A – F) of resolution IV (2IV6−2), with generators E = ABC and F=BCD, was followed in the screening experiments to assess the effect of selected fermentation parameters upon biomass production: A – glucose concentration (15 or 30 g.L−1); B – NaCl concentration (0 or 20 g.L−1); C – pH (4.6 or 6.0); D – temperature (30 or 37 °C); E − stirring rate (100 or 300 rpm); and F – degree of aeration (0 or 100% pO2). The above levels were based on literature information pertaining to fermentation media and growth parameters of L. plantarum (see Table 1, Table 2, respectively). A meaningful fraction of the full 26 factorial design, consisting of 16 runs, was accordingly pursued; 3 center points were added, to test for response curvature (see Table 3). The response used to assess biomass growth was maximum specific growth rate (h−1) – taken as the slope of the log10(OD) vs. time (h) plot, throughout the exponential growth phase.

Table 1. Modified Man, Rogosa and Sharpe broth formulations used to grow Lactobacillus plantarum strains.

Components (g.L−1)MRS brothModified MRS broth
Glucose20201010/2015 or 30
Enzymatic digest of casein10
Meat extract10
Yeast extract510405/10520
K2HPO420.211.325
Sodium acetate5
Triammonium citrate222
MgSO4·7H2O0.20.010.20.20.20.2
MnSO4·H2O0.050.050.050.050.05
Tween 80 (mL.L−1)1.08111
Lactose75
Tryptone5
Peptone10/15
Sodium glutamate2
KH2PO40.20.24.6
FeSO4·7H2O0.01
FeCl30.05
NaCl0 or 20
Vitamin solution (mL.L−1)1
ReferenceMerck Millipore (2020)Noori et al. (2016)Hwang, Chen, Huang, and Mao (2011)Siaterlis, Deepika, and Charalampopoulos (2009)Zannini, Santarelli, Osimani, dell’Aquila, and Clementi (2005)Present study

Table 2. Operational parameters used to grow Lactobacillus plantarum strains.

Parameter values
Seed culture200 mL MRS
37 °C
48 h		MRS +0.05%
l-ysteine
16 h		37 °C
140 rpm
12–18 h
Working volume (L)530.11.31
Total volume (L)2150.52
Inoculum (%(v/v))1051101
Growth time (h)24241210 to 15812
Stirring rate (rpm)40250150100200
Temperature (°C)37–40306–30a303037373535
pH7.266–4b6.56.256.5
NaCl (%(w/v))4.50–6cnot controlled
initial: 5.8
AerationAnaerobic0.1 vvm (air)0.5 vvm (air)Anaerobic or 0.15 vvm (air)Anaerobic or 0.2 vvm (air)
ReferenceNoori et al. (2016)de Angelis et al. (2015)Patent No. EP2901867A1 (2015)Wouters et al. (2013)Zotta et al. (2012)Hwang et al. (2011)Siaterlis et al. (2009)Zannini et al. (2005)Zotta et al. (2013)
a

Seasonal temperature patterns.

b

To adapt to conditions of acidity.

c

To adapt to conditions of salinity.

Table 3. 2IV6−2 Fractional factorial design used to ascertain the effect of six fermentation parameters, A – F, upon biomass growth of Lactobacillus plantarum AFS13.

Factor CodeABCDEFResponse
Run #Design OrderGlucose concentration (g.L−1)NaCl concentration (g.L−1)pH (units)Temperature (°C)Stirring rate (rpm)Degree of aeration (% pO2)Maximum specific growth rate (h−1)
1611504.63010000.302
523004.63030000.268
1315204.6303001000.244
7430204.6301001000.245
1351506.0303001000.237
1463006.0301001000.219
6715206.03010000.299
17830206.03030000.276
1291504.6371001000.296
3103004.6373001000.301
91115204.63730000.338
21230204.63710000.331
18131506.03730000.334
4143006.03710000.356
101515206.0371001000.277
191630206.0373001000.262
81722.5105.333.5200500.296
151822.5105.333.5200500.289
111922.5105.333.5200500.283

2.2.2. Fermentation medium and inoculum

A modified MRS (mMRS) broth was employed – formulated as indicated in Table 1; silicone antifoam 30% in water (100 μL.L−1) was added. Overnight cultures in mMRS, with the same composition as that being tested in the corresponding experimental run, were inoculated (10%(v/v)) in fresh mMRS, and left to grow at 30 °C up to an OD600 of ca. 0.5 (Shimadzu UV-1800 240 V IVDD spectrophotometer).

2.2.3. Fermentation

The fermentation batches were carried out in a 7 L-bioreactor (FMT series, Fermentec), previously autoclaved with pH and DO sensors assembled (Mettler-Toledo); 2.7 L of sterilized medium (121 °C, 20 min) was aseptically poured thereinto. The pH was kept constant via addition of NaOH (4 mol.L−1) or HCl (2 mol.L−1), as appropriate (Table 3). Temperature, stirring rate, degree of aeration, and pH were controlled via the integrated software (HMI Touch PC Type Lab Scale Fermenter system, Fermentec). The bioreactor was inoculated with 10%(v/v) inoculum, under aseptic conditions, so as to reach an initial OD600 of ca. 0.05 in a working volume of 3 L. Each batch was carried out for ca. 6 h – as the best compromise between large amounts of biomass and prevention of reaching the stationary phase. Samples were taken on an hourly basis, to ascertain OD600. By 6 h, the spread plate technique on MRS agar was performed with culture aliquots; Colony-Forming Units (CFUs) were counted following plate incubation for 48 h at 30 °C.

The 3,5-dinitrosalicylic acid (DNS) method (dos Santos et al., 2017) was applied to samples of the original medium (prior to inoculation), and of the batch by the end of fermentation – to ascertain the level of glucose available and consumed, respectively. Samples were centrifuged to remove the suspended biomass in pellet form – and the supernatant was diluted so that the final concentration would lay within the range of validity of the calibration curve encompassing absorbance (at 540 nm) versus glucose concentration.

2.3. Preservation of biomass

After each fermentation batch had been finished, the broth culture was withdrawn and centrifuged (Centrifuge 5810 R, Eppendorf) at 4000 rpm and 4 °C, for 15 min. The cell biomass was washed once with sterile saline solution (8.5 g.L−1); it was then centrifuged at 4000 rpm and 4 °C, for 25 min. The pellets harvested were re-suspended in protective medium, i.e. mMRS (15 g.L−1 glucose), and added with either 50 or 100 g.L−1 sucrose. These samples (bearing a concentration twenty-fold that of the departing biomass broth culture) were transferred to wide-opening sterile flasks, and kept at −80 °C (at least) overnight, before vacuum-drying. Samples were then kept for 2 d at −80 °C, under a pressure of 0.01 Pa in the freeze-drier (FreeZone 2.5 Plus, Labconco). The contents of the flasks from each run were pooled together in another sterile flask, previously tared to allow calculation of the weight of the starter to be obtained (in powdered form) – and eventually stored in a dark, dry place at room temperature. Protective media, of both concentrations, were freeze-dried following the same procedure – to further determine the dried cell biomass in the starter.

Two viability comparisons were performed. The first entailed viability changes during biomass broth handling, between the end of fermentation (t = 6 h) and the beginning of storage (t = 0 d). The second pertained to viability change throughout a 4 mo-storage period (t = 4 mo). A given weight of starter was accordingly re-hydrated in 0.03 L sterile distilled water, either immediately after removal from the freeze-dryer or by 4 mo of storage – to produce a bacterial suspension, with biomass concentration equivalent to that prevailing by the end of the fermentation batch. The calculation considered the biomass yield for the corresponding run, obtained by filtration of samples at the end of fermentation; all values were converted to CFU.g−1biomass.

2.4. Statistical analysis

The normality of the data was assessed via a Shapiro-Wilk test; all statistical analyses were performed at a significance level of 5%. Regarding the responses as per the fractional factorial experimental design, confirmation of effects preselected as relevant, as well as validation thereof and of fit of coded and actual model equations were performed with the aid of software Design-Expert (Stat-Ease Inc. v. 11). Hypothesis tests, using the SPSS package (IBM SPSS Statistics for Windows, Version 26.0), allowed comparison of changes in biomass viability, in log10(CFU.g−1), during culture handling prior to and during storage proper.

3. Results and discussion

3.1. Relevant parameters for biomass production

Starting with glucose as medium component, one intended to confirm whether a higher concentration had a negative effect (due to growth inhibition) or a positive effect; and whether less substrate (thus leading to a cheaper medium) would still support an acceptable performance. Preliminary tests indicated that the AFS13 strain entered the death phase earlier when grown in 10 g.L−1 glucose; and that 15 and 30 g.L−1 glucose led to essentially identical growth for at least 6 h. Hence, these were chosen as levels for scrutiny, and also because the reference MRS broth concentration (20 g.L−1) appears in between.

It has been claimed that bacteria may grow faster throughout olive fermentation owing to previous adaptation to salty media. Although the selected strain grew better in the absence of NaCl, preliminary tests revealed some growth in media up to 70 g.L−1 NaCl – but not at 100 g.L−1 (unpublished results: Portilha-Cunha, 2019; Research Project Report, Bioengineering Master Degree). Therefore, it was decided to investigate whether growth would be affected by the median value of 20 g.L−1, relative to the absence of NaCl at all.

Regarding pH, the autoclaved mMRS medium showed a pH of 6.0; and preliminary tests indicated that the AFS13 strain took considerably longer to reach the same growth level at pH 4.6 than at pH 6.0. Therefore, those two levels were elected, in attempts to ascertain the effect of pH upon growth.

The strain selected is mesophilic (Reimer et al., 2019) – and preliminary tests confirmed growth at 30 and 37 °C. Despite higher temperatures being expected to enhance growth, the later was chosen as upper level to better ascertain interactions with other variables.

Although preliminary tests unfolded little differences in bacterium growth at 100 rpm or without agitation, some stirring is needed to ensure medium homogenization – and accordingly enhance substrate diffusion, gas transfer, and heat transfer. A stirring rate of 100 rpm was thus chosen to assess whether the presence of stirring would favor growth, or have a negative effect instead.

Finally, it is known that the effects of respiration upon L. plantarum growth are strain-specific (Zotta, Guidone, Ianniello, Parente, & Ricciardi, 2013). Specific strains do not consume O2, although being able to grow in its presence (Götz, Elstner, Sedewitz, & Lengfelder, 1980); whereas a higher cell yield was found during aerobic growth compared to anaerobiosis. Therefore, one decided to test growth under anaerobic conditions versus presence of a high concentration of DO – to evaluate how significant such effect would prove.

The first step in analyzing the experimental results was to find the statistically significant effects, i.e. those larger than noise – and so hardly generated by pure chance. For a 2-level factorial design, the normal plot of effects is currently used with the response chosen (see Fig. 1); the goal is to pinpoint the effects that fall off the line, starting at the extremes. To check for the inclusion of dubious effects, the Pareto chart proves a useful tool: effects above the t-value threshold are possibly important, and should be added if they make sense to the experimenter. Such a chart for the response chosen is conveyed by Fig. 2. The Shapiro-Wilk test further unfolded no significant deviation from the assumption of normality (p = 0.378) for the effects not selected – so these can be safely considered as indistinguishable from noise. Therefore, the effects selected for the model – i.e. degree of aeration and temperature (as main effects), and salt content × temperature (as second-order interaction), appeared to have been correctly chosen.

Fig. 1
  1. Download : Download high-res image (299KB)
  2. Download : Download full-size image

Fig. 1. Normal plot of (standardized) maximum specific growth rate (h−1). Significant effects, not explained by pure chance, were: D – temperature, F – degree of aeration, and BD – interaction NaCl concentration × temperature.

Fig. 2
  1. Download : Download high-res image (371KB)
  2. Download : Download full-size image

Fig. 2. Pareto chart for maximum specific growth rate (h−1). Significant effects, not explained by pure chance, were: D – temperature, F – degree of aeration, and BD – interaction NaCl concentration × temperature.

The addition of experiments run as center points supported a curvature check. Since the p-value for that term was 0.7121, it was considered negligible – thus validating the goodness-of-fit of the (simpler) planar two-level factorial model (see Fig. 3).

Fig. 3
  1. Download : Download high-res image (503KB)
  2. Download : Download full-size image

Fig. 3. Surface plot for maximum specific growth rate (h−1) as a function of aeration degree and temperature, with remaining operational parameters set at: A – 15 g.L−1 glucose, B – 0 g.L−1 NaCl, C – pH 4.6, and E − 100 rpm as stirring rate.

Two types of models were fitted to our data: coded (see Eq. (1)) and actual (see Eq. (2)).(Eq. 1)μ(h−1)=0.2870−0.0027×B+0.0253×D−0.0264×F−0.0074×BD(Eq. 2)μ(h−1)=0.0029−0.0681×NaCl(g.L−1)+0.0093×Temperature(˚C)−0.0005×aeration(%pO2)−0.0211×NaCl(g.L−1)×Temperature(˚C)

Although they produce equivalent predictions and a similar R2 (0.9262), interpretation of actual model effects is more difficult in the latter due to lack of centering and differing units of measure; furthermore, the coefficients of the coded model are proportional to the observed coded effect.

The higher factor coefficients are associated to degree of aeration (F) and temperature (D), so these parameters have the strongest impact upon the response (see Fig. 3). Salt (B) entails a small coefficient, yet it was added to the model due to the relevance of the BD interaction. Remember that an interaction occurs when the response is different from a linear combination of the corresponding factors – as emphasized by the lack of parallelism of the two (straight) lines in Fig. 4.

Fig. 4
  1. Download : Download high-res image (277KB)
  2. Download : Download full-size image

Fig. 4. Interaction plot between D – temperature and B – NaCl concentration (▀ – 0 g.L−1 or ▲– 20 g.L−1), in terms of maximum specific growth rate (h−1), with remaining operational parameters set at: A – 15 g.L−1 glucose, C – pH 4.6, E − 100 rpm as stirring rate, and F – 0% aeration degree, with the indication of deviation bars (I).

According to our analysis of variance, all relevant effects found to possess a p-value lower than 0.05 – meaning that they can explain a significant portion of the variance. The model was also considered significant (F = 43.90, p < 0.0001), with negligible lack of fit (F = 2.93, p = 0.2825). Descriptive statistics were used as a secondary check: the predicted and adjusted R2 are in reasonable agreement with each other since they differ by 0.0526 < 0.2; and adequate precision is 20.405 > 4. Therefore, the models underlying Eqs. (Eq. 1), (Eq. 2)) can be reliably used to interpolate, within the given levels of each factor; and possess a strong enough signal to be suitable for optimization.

Due to their poor influence, values for glucose concentration (A), pH (C), and stirring rate (E) in Fig. 3 were set to those found to be more economical – i.e. lower glucose concentration, pH 4.6, and lower stirring rate. Salt concentration (B) was set to its lower level, so as to display the highest possible response values. L. plantarum ASF13 may be claimed as tolerant to the presence of high oxygen concentrations in the medium – since growth was observed in all experimental runs; however, it grows better under anaerobiosis, as per the higher response values under 0% pO2 (see Fig. 3 and Table 3). Temperature, in turn, shows a significant improvement in response when set to its highest level (37 °C) – thus confirming the initial assumptions. This finding also agrees with data by Manzoor, Qazi, Haq, Mukhtar, and Rasool (2017) – who worked with a L. plantarum strain that grew better when temperature increased up to 40 °C. However, presence of NaCl seems to affect that response, as illustrated by Fig. 4. Although not statistically significant at 30 °C, the maximum specific growth rate did not increase as much as in its absence at 37 °C. This interaction indicates that NaCl may have a negative impact on strain growth as the temperature is increased.

With regard to the remaining factors, the lack of significance of glucose concentration upon the response selected reveals that the values tested were neither high enough to inhibit growth nor low enough to be limiting – at least during a 6 h-fermentation. This can be confirmed by the results obtained from the reduced sugar assay: on average, the glucose consumption was ca. 5.9 g.L−1, with values ranging from 0.6 to 12.1 g.L−1. Based on these findings, one presumes that lowering glucose concentration would help reduce substrate costs, without compromising growth – up to some degree that might be interesting to find. Nonetheless, Noori, Ebrahimi, and Jafari (2016) claimed that glucose was an influential factor toward L. plantarum growth and that 26 g.L−1 (lying within the 15–30 g.L−1 range) would maximize OD600. Similarly, Manzoor et al. (2017) reported that increasing glucose concentration increased dry cell mass (within the 10–20 g.L−1 range). However, the latter study used 15 g.L−1 for optimized formulation – and its authors pointed out that the importance of glucose depends on other medium factors. Concerning stirring, 100 rpm appears adequate to provide sufficient homogenization of medium and diffusion of substrate; increasing the stirring rate will not likely bring about any advantage. This finding supports a further reduction of the operation costs via reduction in power consumption. Finally, it seems that maintaining pH at 4.6 or 6.0 has a similar impact upon strain growth. This observation contradicts reports by Hou et al. (2016) – who found that a L. plantarum strain attained higher viable cell counts at pH 5.5, compared to other controlled pH values, and to fermentation without pH control. Noori et al. (2016) further reported that pH interacted with glucose and yeast extract – and proposed 7.26 as the optimal pH value (within the 4 to 8 range). Our study suggests that pH 4.6 may be the best option, at least for the 6 h-fermentation tested – and within the remaining factor ranges, which would minimize the need for pH neutralizers.

3.2. Loss of viability during handling and storage

Since the difference in viability between end of fermentation and beginning of storage was found to be normally distributed variables (the corresponding p-value for the Shapiro-Wilk test was indeed p = 0.072), the parametric paired-samples t-test could be safely applied. The 2-tailed p-value was equal to 0.000 – associated with a 95% confidence interval, and a significant positive mean difference between paired observations. Since log growth by 6 h > log growth by 0 d, there is a statistically significant reduction in biomass viability during handling of biomass broth culture; on average, this reduction was 0.8 log (see Table 4). This result was somehow expected because biomass loss cannot be avoided during the separation process, and some volume is wasted after pellet washing. Furthermore, some degree of bacterial death may also have occurred throughout handling, possibly because of inadequate choice of freeze protector. As biomass viability is a major factor toward economic feasibility of starter production, a focus should be put on the improvement of those handling steps.

Table 4. Biomass viabilities (log CFU.g−1 biomass dry weight; mean ± standard deviation) of Lactobacillus plantarum AFS13 by the end of fermentation (6 h) and after freeze-drying (0 days and 4 months).

FermentationPreservation
Run #6 h0 d4 mo
112.13 ± 0.0010.84 ± 0.508.94 ± 0.06
211.80 ± 0.0111.33 ± 0.509.68 ± 0.06
312.38 ± 0.0211.15 ± 0.288.19 ± 0.86
412.03 ± 0.0711.86 ± 0.1310.30 ± 0.01
511.84 ± 0.0710.88 ± 0.508.51 ± 0.15
611.77 ± 0.0111.34 ± 0.129.67 ± 0 .12
711.93 ± 0.0311.73 ± 0.058.84 ± 0.14
811.82 ± 0.0311.34 ± 0.0610.13 ± 0.06
912.11 ± 0.0510.60 ± 0.018.24 ± 0.34
1011.80 ± 0.1610.98 ± 0.139.20 ± 0.02
1111.76 ± 0.1110.38 ± 0.0710.57 ± 0.09
1211.96 ± 0.0610.84 ± 0.058.17 ± 0.12
1311.99 ± 0.0710.65 ± 0.059.47 ± 0.06
1411.88 ± 0.0511.24 ± 0.0410.26 ± 0.07
1511.70 ± 0.1011.20 ± 0.019.64 ± 0.14
1611.95 ± 0.0110.84 ± 0.339.09 ± 0.08
1711.90 ± 0.0611.64 ± 0.269.84 ± 0.05
1811.94 ± 0.0411.62 ± 0.1610.79 ± 0.09
1911.92 ± 0.0211.48 ± 0.0410.11 ± 0.01

The difference in biomass viabilities (expressed as log CFU) between the two times tested, about storage, appeared normally distributed (p = 0.488). The two-tailed p-value pertaining to their difference was equal to 0.000, for a 95% confidence interval, and a significant positive mean difference; the observed reduction in biomass viability was 1.7 log. Therefore, viability values were slightly affected by the specific storage conditions chosen, over 4 mo-period. This is a promising result toward the definition of the best storage conditions for biomass – a critical issue for starters to appropriately perform their unique functions upon reconstitution. Dried powder viability over long storage periods depends on several factors, including oxygen content, temperature, pH, and water activity (Hou et al., 2016). Hence, further improvement of storage conditions may be possible – e.g. use of a vacuum bag, which is successful in preventing rehydration. Recently, Li, Liu, Tian, Luo, and Liu (2019) proposed encapsulating probiotic bacterial cells as a strategy to achieve long-term storage stability; a marginal loss of viability was observed during freeze-drying, and the rate of release could be controlled.

Freeze protectants are known to strongly affect both viabilities throughout storage and performance upon reconstitution; Yeo et al. (2018) further claimed their strain-dependence – which explains the very many studies developed so far, focused on optimization of their use (Hou et al., 2016; Li et al., 2019; Shu, Zhang, Hui, Chen, & Wan, 2017). However, the optimal selection of such compound was not the major thrust of our work.

4. Conclusions

The 2IV6−2 fractional factorial design permitted simultaneous testing of a relatively large number of parameters, at the expense of a relatively low number of experimental trials. Our results demonstrated that the maximum specific growth rate is a good response indicator when screening for variables affecting biomass growth, and in attempts to find their optimum values – at least in the case of bulk production of L. plantarum ASF13 biomass. Degree of aeration, temperature, and temperature × salt concentration interaction were found significant within the range tested; our model (adventitious) strain grows better under anaerobic conditions, at higher temperatures, and in the absence of NaCl. On the other hand, economic criteria may be used to set the best values for the remaining (statistically not significant) parameters: pH 4.6, minimum stirring rate, and lower glucose concentrations. Finally, it was not possible to draw conclusions on the quantitative effect of the freeze protectant upon preservation performance – owing to the (acceptable) small loss of biomass viability over the 4 mo-storage period tested.

CRediT authorship contribution statement

M. Francisca Portilha-Cunha: Conceptualization, Methodology, Validation, Formal analysis, Investigation, Data curation, Writing - original draft, Visualization. F. Xavier Malcata: Resources, Writing - review & editing, Visualization, Supervision, Project administration, Funding acquisition. Patrícia J.M. Reis: Conceptualization, Methodology, Validation, Investigation, Data curation, Writing - review & editing, Visualization. Angela C. Macedo: Conceptualization, Methodology, Validation, Formal analysis, Writing - review & editing, Visualization, Supervision, Funding acquisition.

Declarations of competing interest

None.

Acknowledgments

This work was financially supported by Project PROMETHEUS– POCI-01-0145-FEDER-029284, funded by FEDER funds through COMPETE2020 – Programa Operacional Competitividade e Internacionalização (POCI) and by national funds (PIDDAC) through FCT/MCTES; and by Base Funding - UIDB/00511/2020 of the Laboratory for Process Engineering, Environment, Biotechnology and Energy – LEPABE - funded by national funds through the FCT/MCTES (PIDDAC). The authors thank Ms. Joana Rocha for laboratory assistance; and Mr. Rocha Fernandes, c/o Direção Regional de Agricultura e Pescas do Norte, for support in selecting olive producers for screening, and in transporting raw materials to our laboratory premises.

References

View Abstract
© 2020 Elsevier Ltd. All rights reserved.