Production of potential probiotic Spanish-style green table olives at
pilot plant scale using multifunctional starters
F. Rodríguez-Gómez1, V. Romero-Gil1, J. Bautista-Gallego1,2, P. García-García1, A.
Garrido-Fernández1 & F.N. Arroyo-López1,*
1
Department of Food Biotechnology. Instituto de la Grasa (CSIC). Avda. Padre García
Tejero 4, 41012, Seville, Spain.
2
DIVAPRA, Agricultural Microbiology and Food Technology Sector, Faculty of
Agriculture, University of Turin. Via Leonardo da Vinci 44, 10095 Grugliasco, Torino,
Italy.
Running title: Production of functional olives
*Corresponding author: Francisco Noé Arroyo-López, Ph.D. Tel: +34 954 692 516
ext 115. Fax: +34 954 691 262. e-mail address: fnarroyo@cica.es.
1
1
Abstract
2
This work evaluates the use of two multifunctional starters of Lactobacillus pentosus
3
species (TOMC LAB2 and TOMC LAB4) during elaboration of Manzanilla olive fruits
4
processed according to the Spanish-style. Data show that the use of inocula at the onset
5
of fermentation led to a proper acidification and sugar consumption of brines compared
6
to the spontaneous process, obtaining in a shorter period of time the maximum
7
population for lactic acid bacteria. Both inoculated L. pentosus strains were recovered at
8
high frequencies at the end of fermentation on the olive surface, which was corroborated
9
by RAPD-PCR analysis. In situ observation of olive epidermis slices by scanning
10
electron
11
microorganisms, which reached population levels of approximately 6 and 7 log10
12
cfu/cm2 for yeasts and lactic acid bacteria, respectively. Enterobacteriaceae on the olive
13
surface were also found at the onset of fermentation (~9 log10 cfu/cm2), but they
14
declined during the process and were below the detection limit at the end of
15
fermentation. Results obtained in this study show the advantage of using
16
multifunctional starters with the ability to adhere to the olive epidermis because,
17
ultimately, the fruits are the food ingested by consumers.
microscopy
revealed
a
strong
aggregation
and
adhesion
between
18
19
Keywords: Biofilm; fermentation; starter culture; functional olives; Lactobacillus
20
pentosus; probiotic.
2
21
1. Introduction
22
According to the International Olive Oil Council statistics, the last recognized
23
production of table olives (2010/2011 season) was 2,563,000 tons (IOC, 2013). It is
24
estimated that approximately 60% of this production was processed as Spanish-style
25
green table olives, which implies a lye treatment followed by typical lactic acid
26
fermentation after brining (Garrido Fernández et al., 1997). Although the main objective
27
of the elaboration of this fermented vegetable is the preservation of the product by
28
acidification and salting, the preservation of its texture and the development of pleasant
29
flavor have allowed its worldwide implementation.
30
Spanish green table olive fermentation is a process typically dominated by lactic
31
acid bacteria (LAB). These microorganisms may have potential benefits on human
32
health, because, among others, the following individual or combined favorable effects
33
have been described for some species: improvement of lactose assimilation, food
34
digestibility, hypercholesterolemia, immune response, and the prevention of intestinal
35
infections, cancer, food allergies and constipation (Champagne and Gardner, 2005).
36
Moreover, table olives might also be considered as a functional food because of their
37
high content in dietary fiber, antioxidant compounds, vitamins and anticancer
38
compounds (Garrido Fernández et al., 2001).
39
Ranadhera et al. (2010) consider that the type of food carrier plays an essential
40
role in buffering the probiotic throughout the gastrointestinal tract, regulating their
41
colonization or interacting with the probiotic to alter functionality. Lavermicocca et al.
42
(2005) used table olives as a vehicle to incorporate probiotic bacteria species into the
43
human body. Particularly, one strain of Lactobacillus rhamnosus remained invariant
44
and showed a good recovery (about 6 log10 cfu/g) after 30 days of its incorporation to
3
45
fermented table olives. Lactobacillus paracaseis IMPC2.1 successfully colonized the
46
olive surface, dominating the natural LAB population until the end of the fermentation
47
(De Bellis et al., 2010), making the product a suitable carrier for delivering probiotic
48
bacteria to humans. According to these authors, the high survival rates observed for
49
probiotic strains on olives implies that the consumption of about 80 g of olives allowed
50
the ingestion of more than one billion L. paracasei or Lactobacillus plantarum live cells
51
(Lavermicocca et al., 2010). Recently, it has been demonstrated that diverse L.
52
plantarum and Lactobacillus pentosus strains establish polymicrobial communities on
53
the surface of green Spanish-style Gordal (Domínguez Manzano et al., 20012) or
54
Manzanilla (Arroyo López et al., 2012) cultivars. In both cases, in situ observation of
55
the olive epidermis by scanning electron microscopy (SEM) showed strong aggregation
56
and adhesion between bacteria and yeasts by the formation of a complex matrix which
57
embedded the microorganisms.
58
Nowadays, a great diversity of bacterial populations are found during Spanish-
59
style green table olive processing (Doulgeraki et al., 2013; Hurtado et al., 2012). Such
60
diversity may be caused by the empirical production process (Botta and Cocolin, 2012).
61
To improve the fermentation profile, the use of starter cultures for the initiation of the
62
process and to control the microbiological population in the brines has been proposed
63
(Sánchez et al., 2001; De Castro et al. 2002; Peres et al., 2008). A recent study showed
64
that the inoculation with a starter culture led to higher LAB and yeast populations, to
65
decrease in the Enterobacteriaceae populations and to faster acidification of the brines,
66
but did not assure per se the presence of the added strains in the brine solutions at the
67
end of fermentation (Rodríguez-Gómez et al., 2013). Most of the above mentioned
68
works did not control the survival and imposition of the specific inoculated strain and
4
69
the favourable effects on fermentation could not be attributed with total certainty to the
70
inoculum activity. On the contrary, Argyri et al. (2014) and Blana et al. (2014) have
71
evaluated the use of potential probiotic LAB strains originally isolated from olive
72
fermentation as starters with promising results, paying particular attention to their
73
imposition and presence at the end of fermentation. Therefore, a proper selection of the
74
starter strain and manipulation of the fermentation process is essential for succeeding in
75
the production of functional olives and the imposition of selected strains.
76
With the present study we aim to determine the performance, at pilot plant scale,
77
of two preselected LAB strains (L. pentosus TOMC LAB2 and TOMC LAB4) for the
78
fermentation and production of functional table olives. The research was based on a
79
multidisciplinary approach using molecular biology, analytical chemistry, modeling,
80
scanning electron microscopy and food microbiology techniques to determine the
81
microbial growth, acidification kinetics, imposition of the inoculated strains in the
82
brines and on the olive surface as well as their ability to form biofilm, which is an
83
essential characteristic to turn table olives into an appropriate bacteria food carrier.
84
2. Material and methods
85
2.1. Olive processing
86
The fruits used in the present study were of the Manzanilla variety (Olea
87
europaea pomiformis), picked by hand at the green maturation stage during the
88
2010/2011 season and supplied by JOLCA S.A. (Huevar del Aljarafe, Seville, Spain).
89
Six cylindrical PVC fermentation vessels with a total volume of 100 L (Ø 0.4 m
90
x 0.8 m high), provided with a reduction in the top (Ø 0.15 m x 0.15 m), were filled
91
with 64 kg of olives. For debittering, fruits were lye-treated with a 2.2% NaOH (40 L)
92
solution for 5 h (until the lye penetrated 2/3 of the flesh), followed by immersion in tap
5
93
water for 20 h to remove excess alkali. Then, a brine solution with 11% (w/v) NaCl and
94
35 ml of HCl 37% was added to partial neutralize of the remaining NaOH. After 2 days,
95
CO2 was bubbled through the fermentation vessels (olives and brine) to reach a pH of
96
nearly 7.5 units. Anaerobic conditions were achieved by using a floating closing device
97
placed on the top of the vessels.
98
2.2. Inoculation and fermentation
99
After pH adjustment, the fermentation vessels were inoculated with overnight
100
cultures (MRS at 37º for 18h) of strains L. pentosus TOMC-LAB2 and TOMC-LAB4,
101
selected from previous experiments because of their potential probiotic characteristics
102
(Bautista Gallego et al., 2013), ability of adhesion to olive epidermis (Arroyo-López et
103
al., 2012) and good performance in previous trials carried out at laboratory scale
104
(Rodríguez-Gómez et al., 2013). These cultures were grown until early stationary phase
105
and then an aliquot of the suspension was added to the fermentation vessels in a
106
proportion of about 0.03% to reach an initial inoculum level of approximately 6 log10
107
cfu/mL in the brines. The experimental design consisted of: F1, spontaneous and un-
108
inoculated treatment; F2, treatment inoculated with LAB2 strain; and F3, treatment
109
inoculated with LAB4 strain. Each treatment was carried out in duplicate and monitored
110
for 135 days.
111
The fermentation vessels were kept during the entire process at the Instituto de la
112
Grasa pilot plant (CSIC, Seville, Spain), where the room temperature decreased
113
progressively from 28 ºC (October) to 14 ºC (January), which was maintained untill the
114
end of the experiments (February). After 18 days of fermentations, 2 L of brine from the
115
bottom of the vessels were removed and substituted with the same volume of fresh brine
116
containing 5% NaCl and 15% glucose (to reach a final concentration in the brines of 7.5
6
117
g/L of glucose). On the 54th day of fermentation, the brine was again supplemented with
118
a 2.8 g/L glucose solution. This practice is common during Spanish-style olive
119
processing to achieve adequate final pH values (<4.2) and ensure the safe storage of the
120
fermented olives (Garrido-Fernández et al., 1997; Chorianopoulos et al., 2005).
121
2.3. Physicochemical analyses of the brines and modelling
122
Analysis of pH and titratable acidity of the fermentation brines was carried out
123
using the methodology described by Garrido-Fernández et al. (1997). Sugars (glucose,
124
fructose, sucrose and mannitol), organic acids (lactic and acetic) and ethanol were
125
determined by HPLC according to the protocols described by Rodríguez Gómez et al.
126
(2012). The evolution of these parameters through fermentation was modeled using the
127
following equations:
128
129
i) Exponential decay function (for pH and total sugar concentration):
Y=D+S*e-(K*t)
130
where Y is the dependent variable, t is the time (days), D is the minimum
131
asymptotic value when t→∞, S is the estimated value of change, and K is the kinetic
132
constant of change (days-1).
133
134
135
ii) Reparameterized Gompertz function (Zwietering et al., 1990) (for lactic acid,
acetic acid and titratable acidity):
Y=A*exp{-exp[((µmax*e)/A)*(λ-t))+1]}
136
where Y is the dependent variable, A is the maximum asymptotic value reached
137
when t→∞, µ is the maximum rate of production (days-1), and λ is the period of time
138
without production (days).
7
139
Model parameters were obtained by a non-linear regression procedure,
140
minimizing the sum of squares of the difference between the experimental data and the
141
fitted model, i.e., loss function (observed-predicted)2. This task was accomplished using
142
the non-linear module of the Statistica 7.1 software package (StatSoft Inc, Tulsa, OK,
143
USA) and its Quasi-Newton option. Fit adequacy was checked by the proportion of
144
variance explained by the model (R2) with respect to the experimental data.
145
Surface color of olives was measured at the end of the fermentation process
146
using a BYK Gardner Model 9000 Color-view spectrophotometer. Interference by stray
147
light was minimized by covering the samples with a box having a matte black interior.
148
Color was expressed in terms of the CIE L* a* b* parameters and as color index (Ci),
149
calculated according to Sánchez et al. (1985) as follows:
150
151
152
Ci=[(-2*R560+R590+4*R635)/3]
where Rs are the reflectance values at 560, 590 and 635 nm, respectively. The
data of each measurement were the average of twenty olives.
153
The firmness of olives was measured at the end of the fermentation process
154
using a Kramer shear compression cell coupled to an Instron Universal Machine
155
(Canton, MA, USA). The crosshead speed was 200 mm/min. The firmness, expressed as
156
N/100 g flesh, was the mean of ten replicate measurements, each of which was
157
performed on three pitted olives.
158
2.4. Microbiological analyses of the brines and modelling
159
Brine samples or their decimal dilutions were plated using a Spiral Plating
160
System model dwScientific (Don Whitley Sci. Ltd., Shipley, U.K) on the media
161
described below. Plates were counted using a CounterMat v.3.10 (IUL, Barcelona,
8
162
Spain) image analysis system, and the results expressed as log10 cfu/mL.
163
Enterobacteriaceae were counted on VRBD (Crystal-violet Neutral-Red bile glucose)-
164
agar (Merck, Darmstadt, Germany), LAB on MRS (de Man, Rogosa and Sharpe)-agar
165
(Oxoid) supplemented with 0.02% (w/v) sodium azide (Sigma, St. Louis, USA), and
166
yeasts on YM (yeast-malt-peptone-glucose) agar (DifcoTM, Becton and Dickinson
167
Company, Sparks, MD, USA) supplemented with oxytetracycline and gentamicin
168
sulphate as selective agents for yeasts. Plates were incubated at 37 ºC for 24 h
169
(Enterobacteriaceae) or 30ºC for 48 h (yeasts and LAB).
170
171
Changes in the microbial populations versus time in the brines were modelled
using:
172
i) the Two-term Gompertz equation proposed by Bello & Sánchez-Fuertes
173
(1995) when microbial growth and decay was observed. It has the following expression:
174
log Nt=log(N0)+k1*exp[−exp(−k2(t−k3))]−k4*exp[−exp(−k5(t−k6))]
175
where Nt is the population (log10 cfu/mL) at time t (days); N0 is the initial
176
population (log10 cfu/mL); k1 is the increase in microorganisms from the initial level to
177
the maximum (log10 cfu/mL); k2 is the relative growth rate (days-1); k3 is the time at
178
which growth rate is maximum (days); k4 is the decrease from the maximum to a
179
minimum level (log10 cfu/mL); k5 is the relative death rate (days-1) and k6 is the time
180
(days) at which death rate is maximum.
181
182
183
ii) the model of Pruitt & Kamau (1993) in the case of a first and rapid decrease
of the inoculum followed by a further growth. It has the following expression:
Nt = (Nmax/[1+exp(-μ(t-τ))]+Nd*exp(-γ*t)
9
184
where Nt is the population (log10 cfu/mL) at time t (days), Nmax is the maximum
185
asymptotic population (log10 cfu/mL), µ is the maximum growth rate (days-1), τ is the
186
time (days) for Nmax/2, Nd is the damage population (log10 cfu/mL) and γ is the
187
maximum death rate (days-1).
188
The diverse growth/death parameters were obtained by a non-linear regression
189
procedure using the Statistica 7.1 software package.
190
2.5. Microbiological analyses of the olive surface
191
To determine the number of microorganisms adhered to the olive epidermis, the
192
protocol developed by Böckelmann et al. (2003) was slightly adapted to the specific
193
characteristics of table olives. Briefly, two fruits from each fermentation vessel were
194
randomly taken at different sampling times and washed for 1 h with 250 mL of a sterile
195
PBS buffer solution (8.0 g/L NaCl, 0.2 g/L KCl, 1.44 g/L Na2HPO4, 0.24 g/L KH2PO4,
196
pH finally adjusted to 4.7 with HCl 1M) to remove non-adhering cells. Then, olives
197
were transferred to 50 mL of a PBS solution added of the following enzymes: 14.8
198
mg/L lipase (L3126), 12.8 mg/L β-galactosidasa (G-5160) and 21 µL/L α-glucosidasa
199
(G-0660) (Sigma-Aldrich, St. Louis, USA). To achieve biofilm disintegration and
200
removal of the adhered cells, the fruits were incubated at 30ºC in this enzyme cocktail
201
with slight shaking (150 rpm). After 12 h, the olives were removed and the resulting
202
suspension was centrifuged at 9,000 x g for 10 min at 4ºC. Finally, the pellet was re-
203
suspended in 2 mL of PBS and spread onto the different culture media described above.
204
Olive microbial counts were expressed as log10 cfu/cm2, using the formula of a prolate
205
spheroid for the calculus of olive surface from the longitudinal and transverse axes of
206
fruits (Weisstein, 2013). For the Manzanilla fruits used in the present study, the average
207
area was 10.99±1.01 cm2.
10
208
Changes in the microbial populations vs. time on the olive surface were assessed
209
by estimating the area under the corresponding growth/decline curves. Areas were
210
calculated by integration using OriginPro 7.5 software (OriginLab Corporation,
211
Northampton, USA). This parameter has proven to be a good indicator of the overall
212
microbial growth due to its relationship with the biological growth parameters
213
maximum specific growth rate, lag phase and maximum population level (Bautista-
214
Gallego et al., 2008; Arroyo-López et al., 2009).
215
For “in situ” observations of the microbiota adhered to olive epidermis, scanning
216
electron microscopy (SEM) was used with the method developed by Kubota et al.
217
(2008). Olives were taken from each fermentation vessel at the end of fermentation and
218
washed twice for 1 h with a 100 mM phosphate buffer (pH 7.0) to remove non-adhering
219
cells. Then, the fruits were placed for 2 h in the same phosphate buffer with 5%
220
glutaraldehyde and then washed several times. Slices (0.5 cm2) of the olive epidermis
221
were dehydrated in increasing concentrations of ethanol (50, 70, 80, 90, 95 and 100%)
222
and fixed onto glass slides. Finally, samples were sputtered with gold using a Scancoat
223
Six SEM sputter coater (Edwards, Gat, Israel) for 180 s and observed with a SEM
224
model JSM-6460LV (Jeol Ltd, Tokyo, Japan).
225
2.6. Characterization of the lactic acid bacteria population
226
For characterization of the lactobacilli population, a RAPD-PCR analysis with
227
primer OPL5 was followed according to the protocol described by Rossi et al. (1998).
228
This methodology was used to determine the imposition of the inoculated strains over
229
the native LAB microbiota. It was also previously used in table olive fermentations by
230
Dominguez-Manzano et al. (2012) and Rodriguez-Gómez et al. (2013). A total of 60
231
isolates obtained from the fermentation brines and olive surface were randomly picked
11
232
when maximum population was reached (~10 days), and also from the olive surface
233
(other 60 isolates) when fermentation was completed (~135 days). They were named
234
with the name of treatment (F1, F2 or F3), with A or B (for the first or second
235
fermentation vessel of each treatment, respectively) and with B or O (if they were
236
isolated from the brines or olives, respectively). Then, their pattern profiles of bands
237
(from 100 up to 4,000 bp) were compared with the strain used to inoculate the
238
treatment. For this purpose, PCR products were electrophoresed in a 2% agarose gel and
239
visualized under ultraviolet light by staining with ethidium bromide. The resulting
240
fingerprints were digitally captured and analyzed with the BioNumerics 6.6 software
241
package (Applied Maths, Kortrijk, Belgium). The similarity among digitalized profiles
242
was calculated using the Pearson product-moment correlation coefficient. Dendrograms
243
were obtained by means of the Unweighted Pair Group Method using Arithmetic
244
Average (UPGMA) clustering algorithm.
245
2.7. Statistical analysis
246
Analysis of variance was performed by means of the one-way ANOVA module
247
of Statistica 7.1 software to check for significant differences among treatments. For this
248
purpose, a post-hoc comparison was applied by means of the Scheffé test, which is
249
considered to be one of the most conservative post-hoc tests (Winer, 1962).
250
3. Results
251
3.1. Evolution of the physicochemical characteristics
252
The evolution of pH, titratable acidity, lactic acid, acetic acid and total sugar
253
consumption in the brines during pilot plant fermentations could be properly modeled
254
by the use of both exponential decay and reparameterized Gompertz functions. An
255
example of both fits is shown in Figure 1 for pH (upper panel) and lactic acid
12
256
production (lower panel). The quality of the fit for all physicochemical characteristics
257
was in general good (R2 ranged from 0.83 to 0.98). Table 1 shows the model parameters
258
obtained after the fit for all treatments. Although no significant differences were
259
obtained among treatments according to a Scheffé post-hoc comparison test (p<0.05),
260
some interesting tendencies related to the use of inocula were noticed.
261
The changes of pH in the spontaneous treatment showed a tendency to have
262
higher K values than in the two inoculated processes. This is indicative of a faster decay
263
of pH in the case of both inoculated fermentation vessels in comparison with the
264
spontaneous. When fermentation finished, the final pH obtained ranged from 3.67 (F1,
265
spontaneous) to 3.82 (F2, inoculated with LAB2 strain), which is a guarantee of
266
obtaining a stable and safe product in all cases. Evolution of pH was in agreement with
267
data obtained for the production of lactic acid. Therefore, the inoculated processes with
268
LAB2 and LAB4 strains (F2 and F3, respectively) had a higher kinetic of production of
269
lactic acid (1.16 and 1.35 d-1) compared to the spontaneous (0.99 d-1), albeit the final
270
value of lactic acid produced, which ranged from 15.17 to 17.40 g/L, was not influenced
271
by the use of the inocula (Table 1).
272
The total sugar consumption kinetic was also higher in both inoculated
273
fermentation systems, showing a low final residual sugar (0.16 and 0.04 g/L for F2 and
274
F3 treatments, respectively). Production of acetic acid was reduced and ranged from
275
1.83 to 1.95 g/L, with a slower formation in treatment inoculated with LAB4 strain
276
(0.045 d-1). Finally, when the acid content was evaluated as titratable acidity (as usually
277
made by the industry), the fit was similar to lactic acid production; the modeling results
278
showed the highest asymptotic value for the spontaneous process (0.93%) and the
279
lowest for the treatment inoculated with LAB2 strain (0.81%) (Table 1). The content of
13
280
ethanol during most of the process was low and ranged from 0.3 to 0.4 g/L, although
281
this parameter could not be properly modeled.
282
With respect to olive surface color at the end of the fermentation, there was no
283
significant difference (p<0.05) among the treatments according to the Scheffé post-hoc
284
comparison test, except for b color parameter (yellowness) which was lower in F3
285
treatment (Table 2). Thereby, lightness (L*) ranged from 53.1 to 54.5, greenness (a*)
286
ranged from 2.42 to 2.54, yellowness (b*) ranged from 33.6 to 35.7, and Color Index
287
(Ci) ranged from 26.3 to 27.8. The firmness found at the end of the process was slightly
288
lower for the treatments inoculated with LAB2 strain (F2) in comparison with the
289
spontaneous (F1) and inoculated with LAB4 (F3) processes (Table 2).
290
3.2. Evolution of the microbial populations
291
Modeling of the evolution of the different groups of microorganisms in the
292
brines (Table 3) was carried out by using the Bello and Sanchez-Fuertes model, for the
293
case of growth and decline, or the Pruitt and Kamau equation for the pattern of a first
294
decline followed by a further growth. Different examples of these fits are shown in
295
Figure 2, which always had R2 values above 0.94 (data not shown).
296
As usual in green Spanish-style olive fermentations, there was an initial growth
297
of Enterobacteriaceae population during the first days, which reached its maximum
298
(between 4.19 and 5.69 log10 cfu/mL) on approximately the 3rd day of fermentation.
299
Then, the population declined and disappeared from the brines from the 7th day onwards
300
(Figure 2a). According to the Scheffé post hoc comparison test, the only model
301
parameters which showed significant higher values among treatments were k1 (increase
302
from the inoculum level up to maximum) and k5 (decline rate), both belonging to the
14
303
inoculated F2 treatment. Thereby, LAB2 strain apparently produced a faster
304
disappearance of the Enterobacteriace population.
305
Regarding to the response of yeasts in the brines, only the first phase of growth
306
was modeled because this group of organisms did not show decline during the
307
monitored period (Figure 2b). No significant differences were noticed among treatments
308
for the model parameters k1, k2 or k3, with maximum population levels ranging from
309
4.63 to 4.79 log10 cfu/mL (Table 3), which were very similar to those reached by
310
Enterobacteriaceae. The time to reach the maximum growth rate (k3) ranged from 5.00
311
to 6.60 days, and it was statistically higher than for the Enterobacteriaceae population.
312
Finally LAB population, in the case of inoculation, had an initial decrease
313
followed by a fast increase in the number of cells which reached the maximum (around
314
8.5 log10 cfu/mL) in a short period of time (Figure 2c). The changes in the inoculated
315
treatments were modeled by the Pruitt and Kamau model, without significant
316
differences (p<0.05) among treatments (Table 3). No LAB cells were detected during
317
the first days of fermentation in F1 (spontaneous) treatment, which showed a lag phase
318
of approximately 3 days followed by a fast growth. The maximum population was
319
observed with a slight delay of 5 days with respect to the inoculated treatments (Figure
320
2c). The rapid colonization of the fermentation brines by LAB in the case of inoculation
321
caused the rapid consumption of sugars, production of lactic acid and pH decrease
322
mentioned above.
323
The presence of microorganisms on the olive surface was unable to be modeled.
324
Therefore, the growth of the different groups of microorganisms was evaluated by the
325
comparison
326
Enterobacteriaceae were found on the olive surface at the onset of fermentation at
of
the
area
under
their
15
growth/decline
curves
(Figure
3).
327
approximately ~9 log10 cfu/cm2, but then they declined and were not detected at the end
328
of the process (Figure 3a). Yeasts and LAB reached lower population levels, around 6
329
log10 and 7 log10 cfu/cm,2 on the 10th day of fermentation, respectively, but their
330
population levels were practically maintained until the end of the fermentation process
331
(Figure 3b and 3c). Table 4 shows the area values obtained for the diverse groups of
332
microorganisms. The highest area value was obtained for the LAB population, followed
333
by yeasts and finally by the Enterobacteriaceae population, with significant differences
334
among them (data not shown). However, no significant differences within each specific
335
group of microorganisms were noticed among treatments.
336
3.3. Imposition of the inoculated strains
337
As commented above, the olive epidermis was mainly colonized by LAB and
338
yeasts, which were able to survive until the end of the process; initially, there were also
339
Enterobacteriaceae, albeit they declined as fermentation progressed. To obtain an
340
evaluation of the imposition of the inoculated microorganisms over the native LAB
341
microbiota at the moment of maximum population and also at the end of fermentation,
342
LAB characterization was performed by means of RAPD-PCR analysis with primer
343
OPL5.
344
Figure 4 shows the dendrogram generated at the moment of maximum
345
population (~10 days) using the patterns profile of the sixty LAB isolates randomly
346
obtained from olive epidermis (30) and fermentation brines (30) plus the two inoculated
347
strains. The cluster analysis showed that the isolates obtained from both the F3 and F1
348
treatments formed a group clearly differentiated from the rest of the lactobacilli, sharing
349
only 42% similarity in their banding profile with the isolates obtained from F2
350
treatment. Within the F3 and F1 cluster, the strain used to inoculate the F3, LAB4, was
16
351
also included, sharing a 91% similarity with all isolates obtained from F3. The LAB4
352
inoculum also shared a 79% similarity with lactobacilli obtained from the spontaneous
353
fermentation. On the contrary, isolates obtained from F2 treatment formed another
354
cluster clearly differentiated from F1 and F3, sharing a 79% similarity among them and
355
with the strain used to inoculate F2 treatment (LAB2). It must be emphasized that
356
isolates obtained from the fermentation brines and olives of the same treatment were
357
very similar among them, but they formed different sub-clusters.
358
Figure 5 shows the dendrogram generated at the end of fermentation (~135 days)
359
using the patterns profile of the sixty LAB isolates randomly obtained from olive
360
epidermis plus the two inoculated strains. The cluster analysis showed that many (12 of
361
20) of the isolates obtained from F2 formed a group clearly differentiated from the rest
362
of the lactobacilli, sharing 88% similarity in their banding profile and with the strain
363
used to inoculate the treatment (LAB2). On the contrary, many (16/30) of the isolates
364
obtained from F3 shared 84% similarity among them and with the strain used to
365
inoculate the treatment (LAB4). The rest of isolates obtained from F3, F2 and
366
spontaneous process (F1) were grouped in different sub-clusters (a total of seven
367
considering 80% similarity), which is indicative that these isolates may belong to
368
different lactobacilli strains.
369
3.4. Formation of biofilms
370
At the end of fermentation (~4 months), olive epidermis from all fermentation
371
systems was analyzed by SEM to prove the in situ formation of microbial biofilms.
372
Figure 6 shows, as an example, the formation of biofilms in one fermentation vessel
373
inoculated with LAB2 strain. This picture shows clearly that microbial cells were
374
strongly embedded in an exopolysaccharide matrix, with some bacteria apparently
17
375
"trying to leave" the biofilm. Similar micrographs were also taken from the rest of the
376
fermentation vessels. These observations are in agreement with the high values obtained
377
from the plate counts from the olive surface, especially for yeasts and LAB.
378
4. Discussion
379
In the present study, in both inoculated and spontaneous green olive
380
fermentations, it was reported by plate count the presence of both yeasts and LAB on
381
olive epidermis at the end of the process. Apparently, the biofilm formation in table
382
olive processing may be a generalized process, regardless of inoculation or not. Our
383
data are consistent with those obtained by Arroyo-López et al. (2012) and Dominguez-
384
Manzano et al. (2012), who previously reported the formation of biofilms during
385
Spanish-style green table olive fermentations using Manzanilla and Gordal fruits, and
386
with those obtained by Nychas et al. (2002) with fermented Greek black olives, who
387
also reported the presence of a high number of yeasts and bacteria adhered to olive
388
epidermis. Although it was also noticed that Enterobacteriaceae was present on the
389
olive epidermis at the onset of fermentation and during a certain period of fermentation
390
(possibly protected by the biofilm), only yeast and LAB were able to survive and reach
391
high population levels on the surface of fermented olives (above 6-7 log10 cfu/cm2) at
392
the end of the process. Because the surface of the Manzanilla fruits used in the present
393
study had an average value of around 11 cm2, a total of 107 yeasts and 108 LAB cells
394
could be ingested by consumers who eat only one olive. The study of Lavermicocca et
395
al. (2005) showed that the olive surface could also be colonized by exogenous
396
microflora not isolated originally from table olive fermentations. These authors added
397
high inoculum levels of different probiotic strains to several table olive elaborations,
18
398
obtaining high counts from the olive surface, especially for L. paracasei IMPC2.1, a
399
human-origin isolate.
400
Recently, Rodríguez-Gómez et al. (2013) used multifunctional starters of L.
401
pentosus species to ferment at laboratory scale Manzanilla fruits processed according to
402
the Spanish style. LAB starters for producing functional olives must possess appropriate
403
technological characteristics such as adequate growth rate, rapid and high lactic acid
404
production, sugar consumption and tolerance or even synergy with other components of
405
the starter (lactobacilli strains or yeasts), in addition to their probiotic characteristics
406
(Ammor and Mayo, 2007). Green table olives are a traditional lactic acid fermented
407
food which, when using an appropriate starter culture selection, may be transformed
408
into a probiotic functional vegetable product. For this reason, different authors have
409
recently screened table olive fermentation microflora to isolate LAB strains with
410
promising probiotic characteristics (Argyri et al., 2013; Bautista-Gallego et al., 2013),
411
and evaluated, at laboratory scale, their application as starter cultures during olive
412
processing (Rodríguez-Gómez et al., 2013; Argyri et al., 2014; Blana et al., 2014).
413
Although the use of inoculation to control olive fermentation is frequently found
414
in the literature (Sánchez et al., 2001; De Castro et al., 2002; Vega Leal-Sánchez et al.,
415
2003), studies on their impositions are scarce. Rodriguez-Gómez et al. (2013) reported
416
that diverse genetic profiles different to the inoculated strains were found among the
417
LAB population at the end of Spanish green table olive fermentations. Hence,
418
inoculation did not assure per se the imposition of the selected strain in the brines.
419
However, in the present study, the use of molecular techniques has permitted the
420
conclusion that the inoculated strains were able to dominate over native LAB
421
populations and other microbial groups present at the onset of the fermentation process,
19
422
in both fermentation brines and on the olive surface. Argyri et al. (2014) and Blana et al.
423
(2014) also reported the imposition and dominance of the inoculated strains at the end
424
of fermentation. The fermentation process begins in the brines with the addition of
425
starter cultures, and it is in this medium where the microorganisms produce lactic acid,
426
enzymes and other compounds that determine the sensorial profile of fermentation, but
427
only the microorganisms adhered to the olive epidermis will be finally ingested by
428
consumers. For this reason, we also studied the imposition of the inoculated strains on
429
the olive surface when fermentation was completed. Data show that even after ~4
430
months of fermentation, both inocula were recovered with frequencies of 60% (12/20
431
isolates) and 53% (16/30 isolates) for LAB2 and LAB4, respectively. However, these
432
values were lower compared with the frequencies obtained at the moment of maximum
433
population (100% for both strains), which is indicative that other lactobacilli strains
434
displaced to the inocula as fermentation progressed. Any case, high population levels of
435
the inoculated strains can be obtained from olive epidermis at the end of fermentation.
436
In this work, the use of two selected L. pentosus strains as starters originated a
437
good acidification rate and consumption of all fermentable substrates and produced the
438
corresponding lactic acid, which is in agreement with the favorable effects found with
439
the use of other starter cultures in table olives (Sánchez et al., 2001; De Castro et al.,
440
2002; Skandamis and Nychas, 2003; Rodríguez-Gómez et al., 2013). However, they did
441
not lead to olives with a lower final pH than those following spontaneous processes,
442
possibly because of the high buffer capacity of the brines (Garrido-Fernández et al.,
443
1997) and the limitations of the LAB themselves (sensibility to NaCl, pH and titratable
444
acidity). Apart from the initial acetic acid produced during the lye treatment (Rodríguez
445
de la Borbolla y Alcalá et al., 1952), there was a progressive formation of this organic
20
446
acid and ethanol associated with microbial growth. This means that, possibly, the LAB
447
strains used during the assays could have a certain hetero-fermentative activity under
448
the assayed conditions. However, most of the acetic acid and ethanol could also have
449
been produced by yeasts (Garrido-Fernández et al., 1997).
450
5. Conclusions
451
The use of appropriate multifunctional LAB starters for processing green
452
Spanish-style table olives may lead to proper sugar consumption and lactic acid
453
production (acidification) in the fermentation brines, as well as a predominance of the
454
LAB strains used as inoculum over the native microbial populations. A proper selection
455
of the starter strain is essential for succeeding in the production of functional table
456
olives. Apart from their probiotic characteristics, the selected strains must be able to
457
colonize olive epidermis because, ultimately, olives are the food ingested by consumers.
458
Therefore, the present work, performed at pilot scale, opens the possibility of the
459
production of functional table olives with multifunctional starters of L. pentosus species
460
at industrial scale. Furthermore, because olives have diverse compounds with functional
461
effects, the use of probiotic microorganisms as starters during olive processing could
462
make of this fermented vegetable a synbiotic food.
463
Acknowledgements
464
The research leading to these results has received funding from the EU's Seventh
465
Framework Programme ([FP7/2007-2013] under grant agreement n° 243471
466
(PROBIOLIVES). We also thank the Spanish Government for financial support
467
(projects AGL2009-07436/ALI, and AGL2010-15529/ALI partially financed by
468
European regional development funds, ERDF), and the Junta de Andalucía (through
469
financial support to group AGR-125). Thanks to JOLCA and ASEMESA for supplying
21
470
the fruits and their own expertise for the development of this work. F.N. Arroyo-López
471
and J. Bautista-Gallego wish to express thanks for their Ramón y Cajal (Spanish
472
government) postdoctoral research and Assegno di Ricerca (UNITO, Italy) contracts,
473
respectively.
474
References
475
Ammor, M.S., Mayo, B., 2007. Selection criteria for lactic acid bacteria to be used as
476
functional starter cultures in dry sausage production: An update. Meat Science 76,
477
138–146.
478
Argyri, A.A., Zoumpopoulou, G., Karatzas, K.A.G., Tsakalidou, E., Nychas, G.J.E.,
479
Panagou, E.Z., Tassou, C.C., 2013. Selection of potential probiotic lactic acid
480
bacteria from fermented olives by in vitro tests. Food Microbiology 33, 282-291.
481
Argyri, A.A., Nisiotou, A.A., Malauchos, A., Panagou, E.Z., Tassou, C.C., 2014.
482
Performance of two potential probiotic Lactobacillus strains from the olive
483
microbiota as starters in the fermentation of heat shocked green olives. International
484
Journal of Food Microbiology 171, 68-76.
485
Arroyo-López, F.N., Querol, A., Barrio, E., 2009. Application of a substrate inhibition
486
model to estimate the effect of fructose concentration on the growth of diverse
487
Saccharomyces cerevisiae strains. Journal of Industrial Microbiology and
488
Biotechnology 36, 663-669.
489
Arroyo López, F.N., Bautista Gallego, J., Domínguez Manzano, J., Romero Gil, V.,
490
Rodríguez Gómez, F., García García, P., Garrido Fernández, A., Jiménez Díaz, R.,
491
2012. Formation of lactic acid bacteria-yeasts communities on the olive surface
492
during Spanish-style Manzanilla fermentations. Food Microbiology 32, 295-301.
22
493
Bautista-Gallego, J., Arroyo-López, F.N., Durán-Quintana, M. C., Garrido-Fernández,
494
A.,
495
chloride salts on Lactobacillus pentosus and Saccharomyces cerevisiae growth.
496
Journal of Food Protection 71, 1412-1421.
2008. Individual effects of sodium, potassium, calcium, and magnesium
497
Bautista-Gallego, J., Arroyo López, F.N., Kantsiou, K., Jiménez Díaz, R., Garrido
498
Fernández, A., Cocolin, L., 2013. Screening of lactic acid bacteria isolated from
499
fermented table olives with probiotic potential. Food Research International 50,
500
135-142.
501
Blana, V.A., Grounta, A., Tassou, C.C., Nychas, G.J., Panagou, E.Z., 2014. Inoculated
502
fermentation of green olives with potential probiotic Lactobacillus pentosus and
503
Lactobacillus plantarum starter cultures isolated from industrially fermented olives.
504
Food Microbiology 38, 208-218.
505
Böckelmann, U., Szewzyk, U., Grohmann, E., 2003. A new enzymatic method for the
506
detachment of particle associated soil bacteria. Journal of Microbiological Methods
507
55, 201-211.
508
Botta, C., Cocolin, C., 2012. Microbial dynamics and biodiversity in table olive
509
fermentation: culture-dependent and –independent approaches. Frontiers in
510
Microbiology 3, 245, 1-9.
511
Bello, J., Sánchez Fuertes, M.A., 1995. Application of a mathematical model to
512
describe the behaviour of the Lactobacillus spp. during the ripening of a Spanish
513
sausage (chorizo). International Journal of Food Microbiology 27, 215-227.
514
515
Champagne, C.P., Gardner, N.J., 2005. Challenges in the addition of probiotic cultures
to foods. Critical Reviews in Food Science and Nutrition 45, 61-84.
23
516
Chorianopoulos, N.G., Boziaris, I.S., Stamatiou, A. Nychas, G.-J.E., 2005. Microbial
517
association and acidity development of unheated and pasteurized green-table olives
518
fermented using glucose or sucrose supplements at various levels. Food
519
Microbiology 22, 117-124.
520
De Bellis, P., Valerio, F., Sisto, A., Lonigro, S.L., Lavermicocca, P., 2010. Probiotic
521
table olives: microbial populations adhering on the olive surface in fermentation sets
522
inoculated with the probiotic strains Lactobacillus paracasei IMPC2.1 in an
523
industrial plant. International Journal of Food Microbiology 140, 6-13.
524
De Castro, A., Montaño, A. Casado, F.J., Sánchez, A.H., Rejano, L., 2002. Utilization
525
of Enterococcus casseliflavus and Lactobacillus pentosus as starter cultures for
526
Spanish-style green olive fermentation. Food Microbiology 19, 637-644.
527
Domínguez Manzano, J., Olmo-Ruiz, C., Bautista Gallego, J., Arroyo López. F.N.,
528
Garrido Fernández, A., Jiménez Díaz, R. 2012. Biofilm formation on abiotic and
529
biotic surfaces during Spanish style green table olive fermentation. International
530
Journal of Food Microbiology 157, 230-238.
531
Doulgeraki, A.I., Promateftaki, P., Argyri, A.A., Nychas, G-J., E., Tassou, C., Panagou,
532
E.Z., 2013. Molecular characterization of lactic acid bacteria isolated from
533
industrially fermented Greek table olives. LWT-Food Science and Technology 50,
534
353-356.
535
536
Garrido Fernández, A., Fernández Díez, M.J., Adams, R.M., 1997. Table olives.
Production and processing. Chapman & Halls. London.
537
Garrido Fernández, A., Romero Barranco, C., García García, P., Brenes Balbuena, M.,
538
2001. Aceitunas de mesa. Composición y características nutricionales, in “Aceite de
24
539
oliva virgen: muestro patrimonio alimentario (Mataix Verdú. J., ed). Fundación
540
Puleva e Instituto Omega-3, Granada (Spain)
541
542
543
Hurtado, A., Reguant, C., Bordons, A., Rozès, N., 2012. Lactic acid bacteria from
fermented table olives. Food Microbiology 31, 1-8.
IOC (International Olive Council). 2013. Key figures on the work marked for table
100th
544
olives.
545
http://www.internationaloliveoil.org/estaticos/view/132-world-table-olive-
546
figures.Kubota, H., Senda, S., Nomura, N., Tokuda, H., Uchiyama, H., 2008.
547
Biofilm formation by lactic acid bacteria and resistance to environmental stress.
548
Journal of Bioscience and Bioengineering 106, 381-386.
session.
19-23
November
1012.
Madrid,
Spain.
549
Lavermicocca, P., Valerio, F., Lonigro, S.L., De Angelis, M., Morelli, L., Callegari,
550
M.L., Rizzello, C.G., Visconti, A., 2005. Study of the adhesion and survival of
551
Lactobacilly and Bifidobacteria on table olives with the aim of formulating a new
552
probiotic food. Applied and Environmental Microbiology 71, 4242-4240.
553
Lavermicocca, P., Rossi, M., Russo, F., Srirajaskanthan, R., 2010. Table olives: A
554
carrier for Delivering Probiotic Bacteria to Humans. In Olives and olive oil in
555
Health and Disease Prevention (Preedy, V. R., Watson, R.R., editors). Academic
556
Press, Oxford, 2010. ISBN: 978-0-12-374420-3. Pg 735- 743.Nychas, G.-J.E.,
557
Panagou, E.Z., Parker, M.L., Waldron, K.W., Tassou, C.C., 2002. Microbial
558
colonization of naturally black olives during fermentation and associated
559
biochemical activities in the cover brine. Letters in Applied Microbiology 34, 173-
560
177.
25
561
Peres, C. Catulo, L., Brito, D., Pintado, C., 2008. Lactobacillus pentosus DSM 16366
562
starter added to brine as freeze–dried and as culture in the nutritive media for green
563
Spanish-style olive production. Grasas y Aceites 59, 234-238.
564
Pruitt, K.M., Kamau, D.N., 1993. Mathematical model of bacterial growth, inhibition
565
and death under combined stress conditions. Journal of Industrial Microbiology 12,
566
221-231
567
568
Ranadhera, R.D., Baines, C.S., Adams, M.C., 2010. Importance of food in probiotic
efficacy. Food Research International 43, 1-7.
569
Rodríguez de la Borbolla y Alcalá, J.M., Gómez Herera, C., Gúzman, R., Vazquez, R.,
570
1952. Estudio sobre el aderezo de aceitunas verdes VII. Efecto del tratamiento con
571
lejía. Anales Real Sociedad Española de Física y Química 48, 427-430.
572
Rodríguez-Gómez, F., Bautista-Gallego, J., Romero-Gil, V., Arroyo-López, F.N.,
573
Garrido-Fernández, A., García-García, P., 2012. Effects of salt mixtures on Spanish
574
green table olive fermentation performance. LWT. Food Science and Technology
575
46, 56-63.
576
Rodríguez Gómez, F., Bautista Gallego, J., Arroyo-Lopez, F,N., Romero Gil, V.,
577
Jiménez Díaz, R., Garrido Fernández, A., García García, P., 2013. Table olive
578
fermentation with multifunctional Lactobacillus pentosus strains. Food Control 34,
579
96-105.
580
Rossi, F., Torriani, S., Dellaglio, F., 1998. Identification and clustering of dairy
581
propionibacteria by RAPD-PCR and CGE-REA methods. Journal of Applied
582
Microbiology 85, 956-964.
26
583
584
Sánchez, A.H., Rejano, L., Montaño, A., 1985. Determinación del color en las aceitunas
verdes aderezadas de la variedad Manzanilla. Grasas y Aceites 36, 258–261.
585
Sánchez, A.H., Rejano, L., Montaño, A., de Castro, A., 2001. Utilization at high pH of
586
starter cultures of lactobacilli for Spanish-style green olive fermentation.
587
International Journal of Food Microbiology 67, 115-122.
588
Skandamis, P.N., Nychas, G.J.E., 2003. Modelling the microbial interaction and death
589
of Escherichia coli O157:H7 during the fermentation of Spanish-style green table
590
olives. Journal of Food Protection 66, 1166-1175.
591
Vega Leal-Sánchez, M., Ruiz-Barba, J.L., Sánchez, A.H., Rejano, L., Jiménez-Díaz, R.,
592
Garrido, A., 2003. Fermentation profile and optimization of green table olive
593
fermentation using Lactobacillus plantarun LPC10 as starter culture. Food
594
Microbiology 20, 421-430.
595
Weisstein, E.W., 2013. "Prolate Spheroid." From MathWorld - A Wolfram Web
596
Resource. http://mathworld.wolfram.com/ProlateSpheroid.html. Last access: 11
597
June 2013.
598
599
Winer, B. J., 1962. Statistical principles in experimental design. New York: McGrawHill.
600
Zwietering, M.H., Jongerburger, I., Rombouts, F.M., Van’t Riet, K., 1990. Modeling of
601
the bacteria growth curve. Applied and Environmental Microbiology 56, 1875-
602
1881.
27
603
Figure Legends
604
Figure 1. Two examples of different model fits for the physicochemical parameters.
605
Upper panel, fit of the pH values obtained from one spontaneous treatment (F1) to an
606
exponential decay function. Lower panel, fit of the lactic acid production data obtained
607
from one treatment inoculated with LAB2 strain (F2) to the reparameterized Gompertz
608
equation.
609
Figure 2. Different examples of the model fit for microbial populations: a)
610
Enterobacteriaceae population in one of the fermentation vessels inoculated with LAB4
611
strain (F3), b) yeast population in one of the fermentation vessels (F1), and c) LAB
612
population in one of the fermentation vessels inoculated with LAB2 strain (F2) and the
613
other spontaneous (F1).
614
Figure 3. Enterobacteriaceae (a), yeasts (b) and lactic acid bacteria (c) changes (log10
615
cfu/cm2) on the olive epidermis of the different treatments assayed in the present study.
616
F1, F2 and F3 stand for spontaneous, inoculated with LAB2 and LAB4 strains,
617
respectively.
618
Figure 4. Dendrogram generated after cluster analysis of the digitalized PCR
619
fingerprints with primer OPL5 at the moment of maximum lactobacilli population (~10
620
days). Reference for treatments is from F1 to F3 (with A or B for each duplicate of the
621
fermentation system), while 1-5 is the reference for the isolates obtained within each
622
fermentation vessel. B or O stands for isolates obtained from the fermentation brines or
623
olive surfaces, while LAB2 and LAB4 are the potential probiotic Lactobacillus
624
pentosus strains used to inoculate F2 and F3 treatments, respectively.
28
625
Figure 5. Dendrogram generated after cluster analysis of the digitalized PCR
626
fingerprints of lactobacilli isolates with primer OPL5 at the end of fermentation (~135
627
days). Reference for treatments is from F1 to F3 (with A or B for each duplicate of the
628
fermentation system), while 1-15 is the reference for the isolates obtained within each
629
fermentation vessel. O stands for isolates obtained from the olive surfaces, while LAB2
630
and LAB4 are the potential probiotic Lactobacillus pentosus strains used to inoculate F2
631
and F3 treatments, respectively.
632
Figure 6. Bacteria adhering to the olive surface in one of the fermentation vessels
633
inoculated with LAB2 strain. Picture was taken by scanning electronic microscopy and
634
show at the bottom, from left to right, the voltage used, the scale, the distance in mm
635
from microscopy to the sample and the spot size.
29