Quality assurance criteria for probiotic bacteria1–4
Elina Tuomola, Ross Crittenden, Martin Playne, Erika Isolauri, and Seppo Salminen
ABSTRACT Acid and bile stability and intestinal mucosal
adhesion properties are among the criteria used to select probiotic
microbes. The quality control of probiotic cultures in foods tradi-
tionally has relied solely on tests to ensure that an adequate num-
ber of viable bacteria are present in the products throughout their
shelf lives. Viability is an important factor, but not the only crite-
rion for quality assurance. To be effective, probiotic strains must
retain the functional health characteristics for which they were
originally selected. Such characteristics include the ability to sur-
vive transit through the stomach and small intestine and to colo-
nize the human gastrointestinal tract. In vitro test protocols can be
readily adopted to examine the maintenance of a strain’s ability to
tolerate acidic conditions, survive and grow in the presence of bile,
and metabolize selective substrates. Molecular techniques are also
available to examine strain stability. Adhesion characterization
may be an important quality-control method for assessing gut bar-
rier effects. Adhesion has been related to shortening the duration
of diarrhea, immunogenic effects, competitive exclusion, and
other health effects. Adhesion properties should be carefully mon-
itored, including adhesion to intestinal cells (eg, Caco-2) and
human intestinal mucus. This article outlines the types of in vitro
testing that can be used to ensure quality control of functional pro-
biotic strains. Am J Clin Nutr 2001;73(suppl):393S–8S.
KEY WORDS Probiotics, quality control, adhesion, acid
stability, viability
INTRODUCTION
Probiotics are viable bacteria that beneficially influence the
health of the host (1, 2). Probiotic bacteria selected for com-
mercial use in foods and in therapeutics must retain the char-
acteristics for which they were originally selected (1–3). These
include characteristics for growth and survival during manu-
facture and, after consumption, during transit through the stom-
ach and small intestine. Importantly, probiotics must retain the
characteristics that give rise to their health effects. Conse-
quently, it is necessary to test the stability of these characteris-
tics during manufacture and storage and to ensure that they are
retained in different types of foods (3, 4). The initial screening
and selection of probiotics includes testing of the following
important criteria: phenotype and genotype stability, including
plasmid stability; carbohydrate and protein utilization patterns;
acid and bile tolerance and survival and growth; bile metabo-
lism; intestinal epithelial adhesion properties; production of
antimicrobial substances; antibiotic resistance patterns; ability
Am J Clin Nutr 2001;73(suppl):393S–8S. Printed in USA. © 2001 American Society for Clinical Nutrition
to inhibit known gut pathogens, spoilage organisms, or both;
and immunogenicity.
Examples of how each of these criteria can be unstable are
most abundant in the area of acid stability, which can vary con-
siderably by strain depending on how the strains are used in dif-
ferent food products. This illustrates the necessity for ongoing
quality control of probiotic bacteria during manufacture and use
and for continual monitoring of the effectiveness of probiotics in
humans. It also indicates the need for selection of more stable
Downloaded from www.ajcn.org by on January 23, 2006
probiotic strains for commercial use.
THE QUALITY OF METHODOLOGY USED FOR
SELECTION
Many in vitro tests are performed when screening for potential
probiotic strains. The ability to adhere to the intestinal mucosa is
one of the more important selection criteria for probiotics
because adhesion to the intestinal mucosa is considered to be a
prerequisite for colonization (5). As substrata, enterocyte-like
Caco-2 tissue culture cells and intestinal mucus are currently
used. However, these represent only a distinct part of the intesti-
nal mucosa. In this respect, mucus-secreting HT29-MTX tissue
culture cells would come closer to the true situation in the intes-
tine. In addition to these models, human ileostomy glycoproteins
have been used to study adhesion to the small-intestinal mucosa
(6). All of these in vitro systems provide valuable information on
the ability of probiotics to adhere and colonize the intestine.
Adhesion to colonic or intestinal biopsy samples, if possible,
should be considered as a final in vitro adhesion test that would
be most like the in vivo situation. Not only would this be a bet-
ter approximation of the in vivo situation, it would allow for the
study of adhesion to different parts of the intestine. This is espe-
cially important regarding immune stimulation by oral adminis-
tration of probiotics.
Adhesion is also considered important for stimulation of the
immune system. Adhesion to M cells or Peyer’s patches may
1
From the Departments of Biochemistry and Food Chemistry and Pedi-
atrics, University of Turku, Turku, Finland, and Food Science Australia,
Melbourne Laboratory, Highett, Australia.
2
Presented at the symposium Probiotics and Prebiotics, held in Kiel, Ger-
many, June 11–12, 1998.
3
Supported by The Academy of Finland.
4
Address reprint requests to S Salminen, Department of Biochemistry and
Food Chemistry, University of Turku, 20014 Turku, Finland. E-mail:
seppo.salminen@utu.fi.
393S
394S TUOMOLA ET AL
therefore be an important determinant of possible immune-
stimulating properties of probiotic microorganisms.
Fecal samples have been used in most colonization studies with
probiotic bacteria. These, however, reflect only the bacteriologic
situation in the fecal material and do not give an accurate picture
of the different parts of the gastrointestinal tract or the mucosal
layer of the gut. The use of biopsies from the intestinal mucosa is
a more accurate means of determining colonization. Lactobacillus
strains were found to adhere to rectal mucosa obtained from vol-
unteers who had consumed a fermented oatmeal soup (7).
When tested in vitro, probiotics are usually grown in labora-
tory media. With many probiotics, the aim is at least transient
colonization, in which case the probiotics may need to grow or
at least metabolize in the intestine. The adhesive properties,
metabolism, and morphology of probiotics grown in intestinal
contents or intestinal mucus have been shown to be different
from those of probiotics grown in laboratory media. These dif-
ferences may affect the health effects of the probiotics. By using
culture media more resembling the nutrients available in the
intestine, one may obtain a more accurate understanding of the
properties of probiotics in vivo.
One of the selection criteria for probiotics is the production of
antimicrobial substances, and many probiotic bacteria have been
shown to produce them (8). Among these substances are not only
growth-inhibiting metabolites, eg, organic acids and hydrogen
peroxide, but also bacteriocins, adhesion inhibitors, and a range
of small antimicrobial substances. These substances have been
shown to be produced in laboratory media but their production
and efficacy in vivo remain uncertain (8). It has not been tested
whether administration of purified bacteriocins alone has effects,
eg, on diarrheal disease. Nor has it been tested whether bacteri-
ocins are produced in vivo. If bacteriocins are produced and
active in vivo, it may be necessary to assess their effects on the
indigenous microflora. There is the potential risk that beneficial
strains in the indigenous microflora are also affected by the pres-
ence of a bacteriocin and that the bacteriocins may thereby alter
the natural resistance of the indigenous microflora to coloniza-
tion. Because the production of antimicrobial substances is
regarded as an important selection criterion for probiotics, it
must be confirmed whether these substances are indeed pro-
duced in vivo and exert a beneficial effect. Intestinal or fecal
microflora studies are needed to confirm these properties.
CRITERIA FOR FUNCTIONAL QUALITY ASSURANCE
The criteria currently used to select probiotics define the opti-
mal quality control of probiotic strains in industrial practice. Pro-
biotics are often used in fermented foods and fermentation acts to
retain and optimize microbial viability and productivity while
simultaneously preserving the probiotic’s properties. During fer-
mentation, several metabolic products appear in the food product,
including acetic acid, lactic acid, and possibly bacteriocins, and
the pH of the product is lowered. Such changes may affect the sta-
bility of probiotic bacteria and may alter the bacteria’s functional
properties. Also, long-term industrial use of the starter culture for
production purposes may influence both viability and functional
properties. Important quality-control properties that must be con-
stantly controlled and optimized are the following: adhesive prop-
erties; bile and acid stability; viability and survival throughout the
manufacturing process; effects on carbohydrate, protein, and fat
utilization; and, especially, colonization properties and immuno-
genicity. Most of these properties are related to the physiologic
properties of the strain, but long-term industrial processing and
storage conditions may influence probiotic properties. Thus, in
addition to technologic properties, functional properties should
be considered in quality-control measures.
Acid and bile stability
To survive passage through the stomach and small intestine,
probiotic strains must tolerate the acidic and protease-rich condi-
tions of the stomach, and survive and grow in the presence of bile
acids. Acid tolerance is also important for the probiotics’ survival
in food (3). The dominant food vehicles for probiotics remain to
be yogurts and fermented milks, both of which provide a rela-
tively low-pH environment in which the probiotic bacteria must
survive. Hence, acid tolerance is one of the first properties
screened for when selecting probiotic strains. Simple in vitro tests
can be used to assess acid tolerance. Such tests have been applied
to lactic acid bacteria and Bifidobacterium strains used in the
dairy industry and proposed as probiotics. As shown in Figure 1,
the results of these tests can predict the ability of the strains to
survive in acidic products. In vivo validation of survival through
the human stomach is more difficult to obtain. In vitro assays
Downloaded from www.ajcn.org by on January 23, 2006
examining the inhibitory effect of bile acids on the growth of pro-
biotic strains are also relatively simple to perform, although
again, quantitative extrapolation to probiotic performance in vivo
is difficult. Intraspecies variation in the ability to grow in the
presence of bile is often observed between potential probiotic
strains (Figure 2), and in vitro tests can be used to select the best
strains on a relative basis.
These in vitro tests for selection of acid- and bile-tolerant
strains can readily be applied to ensuring the quality of probiotic
cultures during manufacture and storage and throughout the shelf
life of the product. Both short-term environmental factors affect-
ing gene regulation (eg, culture growth phase and stress leading
to the production of shock proteins), and selection of variants
through long-term subculturing may produce changes in culture
performance (3, 9). The former of these is the most likely to have
the greatest influence on probiotic performance (potentially pos-
itively) because appropriate culture maintenance procedures lim-
iting the number of passages should prevent the selection of
genetic variants in industrial processes. Acid tolerance is likely to
be a relatively intrinsic property of bacteria and acidification of
culture broth during fermentation would also make selection of
less acid-tolerant variants unlikely if serial subculturing was prac-
ticed. However, data showing the long-term stability of acid and
bile tolerance in probiotics during subculturing are lacking.
Adhesion stability
Adhesion characterization may be an important quality-control
method for assessing the surface structure of probiotic bacteria and
related gut barrier effects. In several studies, adhesion was related
to a shortening of duration of diarrhea, immunogenic effects, com-
petitive exclusion, and other health effects (2, 5, 10–12).
Adhesion of probiotic strains is variable. Adhesion in differ-
ent in vitro models varies even within the same strain and dif-
ferences between strains can be significant (13–15). Adhesion
of some common probiotic strains was studied by using a
human colon carcinoma cell line (Caco-2) and human ileostomy
glycoproteins as in vitro models for intestinal epithelium and
mucus, respectively (Figure 3). Of 6 probiotic strains tested,
only Lactobacillus johnsonii LJ-1 and Lactobacillus GG were
 QUALITY ASSURANCE OF PROBIOTICS 395S

Downloaded from www.ajcn.org by on January 23, 2006

FIGURE 1. In vitro tests of acid tolerance (A) can be predictive of survival of probiotic strains in yogurt (B). In A, the viable counts of 3 strains
of bifidobacteria were measured at time 0 () and after incubation for 105 min at 37 C in 0.2 mol HCl-KCl/L buffer, pH 2.0, plus 0.1% peptone ().
This correlated well with the ability of the strains to survive in yogurt, as shown in B. The data for B were supplied by L Tran and M Harvey from
Gist-brocades Australia, Moorebank, Sydney, Australia.

adhesive in both models. The most adhesive strain to Caco-2
cells [Lactobacillus casei (Fyos)] adhered poorly to ileostomy
glycoproteins, indicating that the surface properties needed for
adhesion to epithelial cells and mucus may be different. There-
fore, possible changes in adhesion stability should be examined
by using more than one model.
Some reports on the stability of adhesion properties are avail-
able in the literature. Elo et al (16) tested the stability of Lacto-
bacillus GG from different production lots and products by com-
paring the original strain with cultures used for a longer period in
industrial processes. Only slight variation in adhesion properties
was observed. However, a more significant drop was reported in
the adhesion properties of a culture that had been maintained in
MRS broth for 3.5 y with a weekly transfer. Lactobacillus GG
isolated from the fecal samples of subjects consuming a fer-
mented whey drink containing Lactobacillus GG had adherence
properties equal to those of the original strain (16; Table 1).
With use of the human intestinal mucus glycoprotein adhesion
model, the adhesion properties of Lactobacillus GG were stud-
ied by using different production lots as well as the strain recov-
ered from feces. The adhesion tests were conducted as described
earlier (6). The production lots included the original Lactobacil-
lus GG (ATCC 53103; a gift from SL Gorbach, Tufts University,
Boston), a Lactobacillus GG pharmaceutical grade starter cul-
ture (Valio Ltd, Helsinki), a Lactobacillus GG starter culture
provided by Valio Ltd, and frozen samples of freeze-dried pro-
duction lots 1089 (Valio Ltd, 1987) and WA IV (Valio Ltd,
1987). In addition, a fecal isolate of Lactobacillus GG recovered

FIGURE 2. The effect of bile on the growth of Lactobacillus amylovorus CSCC 5442 (A) and Lactobacillus amylovorus CSCC 5197 (B). , con-
trol (MRS broth); , treatment (MRS broth containing 0.3% ox bile).
396S TUOMOLA ET AL

FIGURE 3. Adhesion of commercial probiotic strains in 2 in vitro models of the intestinal mucosa. , adhesion to a differentiated Caco-2 cell
monolayer; , adhesion to intestinal mucins.

Downloaded from www.ajcn.org by on January 23, 2006

after consumption of a whey drink containing Lactobacillus GG
was studied.
A less adhesive lot of Lactobacillus GG was found in Boston
in the course of clinical studies (B Goldin, unpublished
observations, 1996). This isolate was also found to colonize
human subjects less frequently than the original culture when
assessed according to fecal counts. Later, the adhesion properties
of 3 different production lots of Lactobacillus GG were studied in
humans by observing the persistence of the strain in fecal samples
(16). It was observed that at an intake of 1  10 10 colony-
forming units/d, differences between production lots were
reported, some of which could be related back to the adhesion
study with the Caco-2 cell line. At higher intakes the colonizing
properties of different production lots were similar (17).
Some production lots were tested for their ability to adhere to
human ileal cells (B Goldin, unpublished observations, 1990).
When compared with the original Lactobacillus GG isolate,
adhesion of Lactobacillus GG from lyophilized powder found to
colonize human subjects poorly was reduced. Binding of lot
1089 to human ileal cells was lower than the binding of the orig-
inal Lactobacillus GG although they both adhered similarly to
intestinal mucus glycoproteins, indicating that the same lots
express different adhesion characteristics in different models.
In another test, we studied a probiotic strain from a commer-
cial French Lactobacillus acidophilus yogurt; the first isolate
was obtained in 1995 and the second was isolated in 1996. These
isolates were analyzed both in the Food Science Australia labo-
ratories (Highett, Australia) and at the University of Turku labo-
ratories in Finland. In these studies, the strains isolated 1 y apart
had significantly different adhesion properties. The first samples
adhered well in both the Caco-2 model and the mucus model.
However, the second sample taken 1 y later adhered significantly
poorer in both models (Figure 4). The identities of both isolates
were compared by using the sugar fermentation method (API,

TABLE 1
Adhesion properties of Lactobacillus rhamnosus strain GG (LGG) from different sources1
Origin of sample strains Type of process condition Relative adhesion
Original strain2 — +++
Broth-maintained strain Weekly transfer for 3.5 y ++
Gefilus fermented whey drink strain Dairy process +++
LGG production batch no. 1789 Freeze-dried3 +++
LGG production batch no. 1789 Freeze-dried4 +++
LGG production batch no. 4788 Freeze-dried3 ++(+)
LGG isolate 1 Isolated from a human stool sample +++
LGG isolate 2 Isolated from a human stool sample +++
LGG isolate 3 Isolated from a human stool sample +++
Escherichia coli B44 Negative control, bovine enterotoxigenic strain 
E. coli H 10407 Positive control, human enterotoxigenic strain +++++
Lactobacillus acidophilus Dairy control strain, yogurt sample 
1
LGG is ATCC 53103. Data from reference 16.
2
From SL Gorbach, Tufts University, Boston.
3
Sucrose used as a cryoprotectant.
4
Monosodium glutamate used as a cryoprotectant.
 QUALITY ASSURANCE OF PROBIOTICS
FIGURE 4. Adhesion of a commercial Lactobacillus strain isolated at different times (A and B) from the same product line. , adhesion to a dif-
ferentiated Caco-2 cell monolayer; , adhesion to intestinal mucins.
Biomerieux, France) and the Riboprint method (DuPont Quali-
con, Wilmington, DE) and the strains were identical in both tests
systems. Thus, changes in the adhesion capacity apparently
occurred during industrial production.
If adhesion is modified during industrial processes, other
probiotic traits may also be altered. Adhesion properties,
including adhesion to intestinal cells (eg, Caco-2) and human
intestinal mucus preparations, should be monitored carefully.
These control measurements form the basis of human intestinal
colonization, health effects, and future monitoring of produc-
tion procedures.
Viability and properties during processing and storage
Consumption of probiotics may aid lactose digestion, control
intestinal infections, and balance the intestinal mucosal barrier.
However, most such studies were conducted with viable bacter-
ial preparations, and the definition of a probiotic includes viabil-
ity as an important factor (1, 18). The viability of several strains
in fermented milks is dependent on both the production method
and the strain. In one study, 5 strains of L. acidophilus and Lac-
tobacillus GG (ATCC 53103) were tested to determine the effect
of refrigeration on the viability of the strains in cultured butter-
milk and yogurt (19). In cultured buttermilk, 3 of the strains
showed no significant loss of viability during storage, but 2 strains
had significantly decreased viability. Results were similar in
yogurt. It is possible that cultures producing organic acids,
diacetyl, or other inhibitory compounds in the fermented milk
may influence the survival of some probiotic cultures. The bace-
riocins produced by different dairy cultures were reviewed by
Ouwehand and Salminen (20). L. casei GG showed no loss of
viability during storage of any of the cultured products. Thus, the
results indicate that the production method for fermented milk
needs to be carefully evaluated to offer consumers the right
amount of viable cultures to obtain the reported health effects.
Studies of a defined probiotic preparation for the prevention
of antibiotic-associated diarrhea produced conflicting results. In
2 studies, Clements et al (21) reported that 1 of 2 batches of a
lyophilized lactobacillus preparation reduced the volume and
duration of neomycin-associated diarrhea. A second batch had
no effect, although the question of differences in viability
397S
between the preparations may be raised. It is important to take
Downloaded from www.ajcn.org by on January 23, 2006
viability into account because many strains exert effects in the
nonviable form as well (21, 22). Further studies on viability and
health effects are clearly needed.
CONCLUSIONS
It is important to ensure that the specific properties of a pro-
biotic used originally as selection criteria are also targets for
quality assurance. This is an important prerequisite for any
health claims and should be continuously controlled if the health
claims are approved for use. Thus, functional food regulations
should take into account strain properties and their stability dur-
ing industrial processing and use. On the other hand, not all
selection criteria are always necessary for in vivo functional
effects. Each important strain property and its influence on
health should be assessed in relation to the clinical effects
observed in various studies. In vitro test protocols can be readily
adopted to examine the maintenance of a strain’s ability to toler-
ate acidic conditions, survive and grow in the presence of bile,
and metabolize selective substrates.
Adhesion characterization may be an important quality-con-
trol method for assessing the surface structure of probiotic bac-
teria and related gut barrier effects. Adhesion has been related to
immunogenic effects, shortening the duration of diarrhea, and
other health effects. If adhesion is modified during industrial
processes, probiotic effects may consequently also be altered.
Adhesion properties and their dependence on processes and
process changes should be monitored. Suitable models include
adhesion to intestinal cells, such as Caco-2 cells, and binding to
human intestinal mucus preparations. At least 2 models should
be used for routine quality control of adhesion of probiotic
microbes because this would allow different adhesion character-
istics to be measured.
REFERENCES
1. Salminen S, Bouley C, Boutron-Ruault MC, et al. Functional food
science and gastrointestinal physiology and function. Br J Nutr
1998; 80(suppl):S147–71.
398S TUOMOLA ET AL
2. Salminen S, Isolauri E, Salminen E. Clinical uses of probiotics for sta-
bilizing the gut mucosal barrier: successful strains and future chal-
lenges. Antonie Van Leeuwenhoek 1996;70:347–58.
3. Lee YK, Salminen S. The coming of age of probiotics. Trends Food
Sci Technol 1995;6:241–5.
4. Huis In’t Veld JHJ, Shortt C. Selection criteria for probiotic microorgan-
isms. London: The Royal Society of Medicine Press Ltd, International
Congress and Symposium Series, 1996;219:27–36.
5. Salminen S, Laine M, von Wright A, Vuopio-Varkila J, Korhonen T,
Mattila-Sandholm T. Development of selection criteria for probiotic
strains to assess their potential in functional foods: a Nordic and Euro-
pean approach. Biosci Microflora 1996;15:61–7.
6. Ruseler van Ebmden JHG, van Lieshout L, Gosselink M, Marteau P.
Inability of Lactobacillus casei strain GG, Lactobacillus acidophilus,
and Bifidobacterium bifidum to degrade intestinal mucosa glycopro-
teins; clearing the way for mucosa-protective therapy. Scand J Gas-
troenterol 1995;30:675–80.
7. Alander M, Korpela R, Saxelin M, Vilpponen-Salmila T, Mattila-
Sandholm T, von Wright A. Recovery of Lactobacillus rhamnosus GG
from human colonic biopsies. Lett Appl Microbiol 1997;24:361–4.
8. Ouwehand AC. Antimicrobial components from lactic acid bacteria.
In: Salminen S, von Wright A, eds. Lactic acid bacteria, microbiology
and functional aspects. New York: Marcel Dekker Inc, 1998:139–60.
9. Lee YK, Wong S-F. Stability of lactic acid bacteria in fermented milk.
In: Salminen S, von Wright A, eds. Lactic acid bacteria, microbiology
and functional aspects. New York: Marcel Dekker Inc, 1998:103–14.
10. Malin M, Verronen P, Korhonen H, et al. Dietary therapy with Lacto-
bacillus GG, bovine colostrum or bovine immune colostrum in
patients with juvenile chronic arthritis: evaluation of effect on gut
defence mechanisms. Inflammopharmacology 1997;5:219–36.
11. Saavedra JM, Bauman N, Oung I, Perman J, Yolken R. Feeding of Bifi-
dobacterium bifidum and Streptococcus thermophilus to infants in hos-
pital for prevention of diarrhea and shedding of rotavirus. Lancet 1994;
344:1046–9.
12. Isolauri E, Juntunen M, Rautanen T, Sillanaukee P, Koivula T. A
human Lactobacillus strain (Lactobacillus GG) promotes recovery
from acute diarrhea in children. Pediatrics 1991;88:90–7.
13. Lehto E, Salminen S. Adhesion of two Lactobacillus strains, one Lac-
tococcus strain and one Propionibacterium strain to cultured human
intestinal Caco-2 cell line. Biosci Microflora 1997;16:13–7.
14. Lehto E, Salminen S. Adhesion of twelve different Lactobacillus
strains to Caco-2 cell cultures. Nutr Today 1996;31:49–50.
15. Tuomola E, Salminen S. Adhesion of some probiotic and dairy Lacto-
bacillus strains to Caco-2 cell cultures. Int J Food Microbiol 1998;
41:45–51.
16. Elo S, Saxelin M, Salminen S. Attachment of Lactobacillus casei
strain GG to human colon carcinoma cell line Caco-2: comparison
with other strains of dairy starter bacteria. Lett Appl Microbiol
1991;13:154–6.
17. Saxelin M, Elo S, Salminen S, Vapaatalo H. Dose response in faecal
colonization with Lactobacillus GG. Microbiol Ecol Health Dis 1991;
4:209–14.
18. Fuller R. Probiotics in man and animals. J Appl Bacteriol 1989;66:
365–78.
19. Nighswonger BD, Brashears MM, Gilliland SE. Viability of Lacto-
bacillus acidophilus and Lactobacillus casei in fermented milk prod-
Downloaded from www.ajcn.org by on January 23, 2006
ucts during refrigerated storage. J Dairy Sci 1996;79:212–9.
20. Ouwehand AC, Salminen SJ. The health effects of viable and non-
viable cultured milks. Int Dairy J 1998;8:749–58.
21. Clements ML, Levine MM, Ristaiano PA, Daya VE, Hughes TP.
Exogenous lactobacilli fed to man—their fate and ability to prevent
diarrheal disease. Prog Food Nutr Sci 1983;7:29–37.
22. Salminen S, Ouwehand A, Benno Y, Lee Y-K. Probiotics: how should
they be defined? Trends Food Sci Technol 1999;10:1–4.