There is a well-established link between stress, mood disorders and gastrointestinal (GI) disease(Reference Forsythe, Sudo and Dinan1). While the organism is generally capable of adapting to stressors, chronic overload can result in GI and mood disorders(Reference Mayer2–Reference O'Mahony, Marchesi and Scully4). Indeed, several studies have indicated that stressful events are associated with the onset of chronic GI disturbances(Reference Lutgendorff, Akkermans and Söderholm5), functional ones(Reference Drossman, Sandler and McKee6–Reference Gwee8), inflammatory bowel disease(Reference Duffy, Zielezny and Marshall9–Reference Bennett, Tennant and Piesse12) or peptic ulcers(Reference Levenstein, Kaplan and Smith13–Reference Aoyama, Kinoshita, Fujimoto and Himeno16), as well as anxiety and depression depending on the genetic background(Reference Finlay-Jones and Brown17–Reference Kendler, Prescott and Myers19). Since depression reduces the capacity of coping with stress(Reference Wheatley20), GI disorders may be accelerated or exacerbated. For example, inflammatory bowel disease was associated with mood disorders in more than 50 % of patients, but with more pronounced psychological disturbances during periods of active intestinal distress(Reference Whitehead, Palsson and Jones21, Reference Graff, Walker and Bernstein22). Depression is sometimes the primary culprit, as demonstrated by the successful use of antidepressants in treating inflammatory bowel disease(Reference Neufeld and Foster23).
There is emerging evidence from preclinical studies of a role for gut microbiota on the central nervous system function(Reference Forsythe, Sudo and Dinan1, Reference Neufeld and Foster23). GI bacterial infection induced anxiety-like behaviour in mice, probably due to the stimulation of brain areas implicated in integrating viscerosensory information and mood via the vagus nerve, such as paraventricular hypothalamus, amygdala and bed nucleus of the stria terminalis(Reference Goehler, Park and Opitz24). Moreover, germ-free mice have an increased responsiveness of the hypothalamic–pituitary–adrenal (HPA) axis and modified serotonin and noradrenaline levels compared with specific pathogen-free mice(Reference Sudo, Chida and Aiba25). Mono-association with probiotics in these germ-free mice before 6 weeks of age reversed HPA hyper-reactivity. Neonatal maternal separation predisposed adult rats to intestinal mucosal dysfunction in response to stress(Reference Söderholm, Yates and Gareau26) and the development of visceral hyperalgesia(Reference Coutinho, Plotsky and Sablad27). Probiotics restored gut physiology in this stress model by regulating the interaction between mucosa and bacteria and reducing HPA hyperreactivity(Reference Gareau, Jury and MacQueen28). Moreover, probiotics reversed apoptosis markers in the limbic system following myocardial infarction in rats(Reference Girard, Bah and Kaloustian29). Monkeys exposed to stress during setting up of the intestinal microflora had an altered gut colonisation(Reference Bailey, Lubach and Coe30).
The adult human GI tract is the natural habitat of a large and dynamic population of micro-organisms thriving in the relationship between external and internal environments. It comprises at least 160 of different bacterial species per individual from the pool of 1000 and 1150 prevalent species of bacteria(Reference Qin, Li and Raes31), among which probiotics administered in adequate amounts are proposed to confer a health benefit(Reference Pineiro and Stanton32) and as a novel therapeutic strategy(Reference Arvanitoyannis and Van Houwelingen-Koukaliaroglou33), particularly for mood disorders(Reference Forsythe, Sudo and Dinan1, Reference Neufeld and Foster23).
One probiotic formulation (PF), a combination of Lactobacillus helveticus R0052 and Bifidobacterium longum R0175, showed beneficial effects on GI symptoms in patients subjected to chronic stress(Reference Diop, Guillou and Durand34). When administered separately, these two strains have also showed beneficial effects(Reference Estrada, Drew and Van Kessel35, Reference Wine, Gareau and Johnson-Henry36). For example, several strains of Lactobacillus displayed anti-inflammatory properties in vitro in human intestinal epithelial cells(Reference Wallace, Bradley and Buckley37), while oral treatment with B. longum R0175 showed beneficial properties in human subjects with ulcerative colitis(Reference Haskey and Dahl38). Moreover, L. helveticus had favourable actions on sleep efficiency in elderly subjects(Reference Yamamura, Morishima and Kumano-go39). To evaluate the role of PF on anxiety, we first assessed its effects in the conditioned defensive burying test in the rat(Reference Treit, Pinel and Fibiger40, Reference Messaoudi, Lalonde and Schroeder41), in which rats exposed to a probe associated with a single footshock show anxiety-related probe burying, head stretchings and approaches/escape sequences towards the probe. The potential anxiolytic effects of PF were then assessed on human distress, anxiety and depression evaluated with the Hopkins Symptom Checklist (HSCL-90)(Reference Derogatis, Lipman and Rickels42, Reference Derogatis, Lipman and Rickels43), the Hospital Anxiety and Depression Scale (HADS)(Reference Zigmond and Snaith44), the Perceived Stress Scale (PSS)(Reference Cohen, Williamson, Spacapan and Oskamp45) and the coping checklist (CCL)(Reference Vitaliano46, Reference Paulhan, Nuissier and Quintard47). Moreover, 24 h urinary free cortisol was assayed as a physiological index of stress level(Reference Mason, Giller and Kosten48, Reference Martin, Rezzi and Peré-Trepat49).
Subjects and methods
Thirty-six male Wistar rats (HsdBrlHan, Harlan, The Netherlands) weighing 200 g were housed three per cage inside polycarbonate cages measuring 48 × 27 × 20 cm (U.A.R., Epinay-Sur-Orge, France) in a regulated environment (temperature 22 ± 2°C; humidity 50 ± 10 %; lights on from 21.00 to 09.00 hours). After a 7 d adaptation period and tail marking, the rats were weighed and randomly distributed into three groups (n 12): probiotic preparation (PF), placebo (0·5 % methylcellulose solution) and diazepam (Valium® 1 %; Roche, Neuilly-sur-Seine, France) as the reference substance. The rats had free access to food pellets (Teklad diet no. 2016; Harlan Teklad, Oxon, UK) and tap water until the day before anxiety testing, when deprived of food at 06.00 hours until the following day (day 14). The present experiment adhered to the guidelines provided by the ASAB Ethical Committee for the use of animals in behavioural research(50) and by the Canadian Council on Animal Care(51). All procedures complied with the European Communities Council Directive of 24 November 1986 (86/609/EEC).
The behaviours were recorded by experimenters unaware of the administered products. The rats were placed under a dim red light inside a clear Plexiglas chamber (44 × 28 × 18 cm), whose floor was evenly covered with a 5 cm high bedding of wooden sawdust. On days 12 and 13, the rats were familiarised with the chamber for 20 min/d. At the centre of one wall, 2 cm above the bedding material, a shock probe (7 × 2 × 0·5 cm) overlaid with a copper wire-integrated circuit connected to a two-pole shock generator (Intellibio, Nancy, France) was inserted on day 14, facing away from where the rat was placed. When the animal touched the probe with its forepaws, a single 2 mA electric shock was delivered, and its behaviour was recorded for 5 min, evaluated from videotapes (Sony™ video camera and recorder) by a trained observer: duration of probe burying (piling sawdust with forelimbs in the direction of the probe); head stretchings towards the probe; approaches towards the probe and retreats away from the probe. The percentage of approaches followed by escapes was then calculated (escapes/approaches × 100), followed by a global stress/anxiety score by adding the ranks of duration of probe burying, head stretchings and percentage of approaches/escapes(Reference Messaoudi, Lalonde and Schroeder41).
The test product is a proprietary PF from Institut Rosell-Lallemand, Blagnac, France, containing a mixture of freeze-dried lactic acid bacteria and excipients. The lactic acid bacteria strains are L. helveticus R0052 (strain number I-1722 in the French National Collection of Cultures of Microorganisms (CNCM), Institut Pasteur, Paris, France) and B. longum R0175 (CNCM strain number I-3470). Excipients are xylitol, maltodextrin, flavour and malic acid. PF contains three billion colony-forming units/1·5 g sachet. The genetic identification of L. helveticus R0052 has been described previously(Reference Naser, Hagen and Vancanneyt52). Strain R0175 was identified as a B. longum by 16S rRNA and tuf gene sequencing. To be brief, genomic DNA was extracted from an overnight broth culture of R0175 described previously for R0052(Reference Hagen, Tramp and Altermann53). Extracted DNA was diluted one-twentieth for PCR to a final concentration of 100 ng/μl. DNA from R0175 was used as a template in PCR to amplify approximately 1370 nucleotides of the 16S rRNA gene using the primers P0 and P6 as described by Ventura et al. (Reference Ventura, Callegari and Morelli54). The tuf genes, approximately 970 nucleotides, were amplified with BIF-1 and BIF-2(Reference Ventura, Canchaya and Meylan55). PCR products were sent to Genome Quebec (Montreal, QC, Canada) according to the guidelines of the DNA Sequencing Platform. Nucleotide sequences for the 16S rDNA and tuf genes of strains R0175 were compared with the BLASTN database available on GenBank(Reference Altschul, Madden and Schaffer56) and were deposited under accession numbers (HM009032 and HM009033, respectively).
All products were freshly prepared every day and administered by gavage at a volume of 5 ml/kg. PF was dissolved in a 0·9 % NaCl solution and stirred until homogenisation just before its administration at a dose of 250 mg/rat per d (109 colony-forming units/d). Diazepam was suspended in a 0·5 % methylcellulose solution and administered at 3 mg/kg 60 min before the test session on day 14. The placebo (PL) group received the 0·5 % methylcellulose vehicle from days 1 to 14.
Comparisons between treated groups and controls were performed by the Kruskal–Wallis test and the Mann–Whitney U test (MWT). The results are expressed as medians with inferior and superior quartile values. Differences were considered to be significant at the P < 0·05 level. All statistical analyses were carried out with the StatView®5 statistical package (SAS Institute, Inc., Cary, NC, USA).
After written informed consent was obtained from all subjects, healthy Caucasian men and women (age and sex distribution of the sample on initial examination are summarised in Table 1) were recruited from the general population from a database of former research participants (Biofortis Clinical Investigation Center) and from a variety of sources including Internet, newspaper and radio advertisements. The formalities were performed in accordance with the rules of Good Clinical Practices (Guidelines GCP ICH), the Helsinki Declaration and French government guidelines ‘Code de la Santé Publique, titre II du livre premier’ relating to biomedical research. The protocol was favourably received by the following ethics committee: ‘Comité de Protection des Personnes (CPP) Ouest IV-Nantes’ on 14 November 2008.
Sixty-six subjects were included from the pool of ninety-nine subjects based on standard biological safety parameters and a score of ≤ 12 in the HADS-anxiety subscale (HADS-A) and in the HADS-depression subscale (HADS-D) and equal to or less than 20 in the HADS total score on initial examination (see Results section and Table 3). Fifty-five of them participated and finished the clinical trial. Subjects were excluded when suffering from neurological, psychiatric, renal, hepatic, cardiovascular and respiratory diseases, or food allergy, or when taking psychotropic drugs during the previous month, stimulating nutritional supplements (vitamin C), ginger, guarana, ginseng, dehydroepiandrosterone, melatonin, antioxidants, anxiolytics, antidepressants, selenium, narcotics, replacement hormones, more than 5 cups of coffee or tea/d, 0·2 litres of cola, 30–40 g of chocolate, three glasses of wine, or two fermented dairy products, or else when smoking more than twenty cigarettes. Pregnant women and subjects who had participated in another clinical study over the past 2 months were also excluded.
Calculation of the sample size, randomisation and blinding
Calculation of the sample size is based on the anxiety dimension of the HSCL-90 scale. A difference from 1·5 to 2 points in the score in this dimension is considered as significant. The results of published studies indicate a mean of 3 (sd 2). To detect such a treatment effect with 80 % power at a 5 % level of statistical significance, it seems necessary to include twenty-eight subjects in each of both groups. Thus, the protocol plans the participation of fifty-six to sixty subjects distributed into two groups: probiotic formulation (PF) and placebo (PL).
After eligibility determination, subjects were then randomised based on age and sex according to a computer-generated randomisation list in sealed, opaque envelopes into two groups: PF and PL groups. The randomisation list was generated and kept by a project nurse not involved in the clinical trial. Subjects and clinical staff involved in the trial experiments were blinded to the treatment group assigned. The codes for the treatment groups were revealed only after the completion of the whole study and statistical analysis.
The clinical trial was designed as a double-blind, controlled, randomised, parallel study lasting 30 d. There were three visits to the Biofortis Clinical Investigation Center: preliminary examination; baseline (14 d later); follow-up (30 d after baseline). During preliminary medical examination, to be included, the volunteers were subjected to blood sampling in order to verify whether their safety biological parameters were within normal ranges, and the HADS. Two subjects did not participate in a satisfactory way and were discarded. The sample of fifty-five subjects was divided into two groups: test product (PF) or PL. At 2 weeks after the preliminary examination at baseline, the subjects completed the HSCL-90, the PSS and the CCL. Each participant then received thirty sticks of the probiotic preparation or placebo for 30 d. At follow-up, the subjects received a second medical examination and completed the rest of the tests. In addition, the day before baseline and follow-up, the subjects collected their urine samples over a period of 24 h to dose the urinary free cortisol.
Products (for the probiotic product characterisation, see above)
During or just after breakfast, all volunteers took one stick of 1·5 g/d of PF (Probio'Stick®: batch no. 6533308; Institut Rosell-Lallemand, Blagnac, France) containing L. helveticus R0052 and B. longum R0175 (3 × 109 colony-forming units/stick) or placebo (xylitol, maltodextrin, plum flavour and malic acid) of identical taste and appearance for 30 d. The first treatment was taken in the evening of baseline, and the last in the morning of the final test period. Study compliance was assessed by counting the number of sticks returned by participants to the study coordinator.
Hopkins Symptom Checklist-90
The HSCL-90 is a 90-item self-reported multidimensional questionnaire(Reference Derogatis, Lipman and Rickels42, Reference Derogatis, Lipman and Rickels43) screening a broad range of psychopathological disorders. The HSCL-90 measures nine primary symptom dimensions (somatisation, obsessive–compulsive, interpersonal sensitivity, depression, anxiety, anger–hostility, phobic anxiety, paranoid ideation and psychoticism). Each item is rated on a five-point scale, ranging from ‘not at all’ to ‘extremely’. The subject's overall psychological distress was evaluated by the global severity index.
Hospital Anxiety and Depression Scale
The HADS, a four-point scale(Reference Zigmond and Snaith44) that ranges from 0 (never) to 4 (very often), is a fourteen-item self-assessment instrument, often applied and convenient for measuring psychological distress in subjects with somatic or psychosomatic disorders(Reference Bjelland, Dahl and Haug57). Three subscores were obtained: HADS global score, HADS-A and HADS-D.
Perceived Stress Scale
The PSS is a fourteen-item self-reported questionnaire(Reference Cohen, Williamson, Spacapan and Oskamp45) assessing the degree to which recent life situations are appraised as stressful. Respondents indicate how often they have felt or thought a certain way over the past month on a five-point scale that ranges from 0 (never) to 4 (very often). Responses are then summed to indicate the level of perceived stress.
The CCL, derived from the ‘Ways of Coping Check-list’ of Lazarus & Folkman(Reference Lazarus and Folkman58), is a validated twenty-nine-item questionnaire(Reference Paulhan, Nuissier and Quintard47, Reference Mason, Giller and Kosten48) measuring five types of coping strategies when confronting an adverse event: ‘problem solving’; ‘avoidance with wishful thinking’; ‘seeks social support’; ‘positive re-evaluation’; ‘self-blamed’. Coping is currently defined as ‘the various cognitive or behavioural efforts intended to master or tolerate the internal or external demands which threaten or go beyond the resources of a subject’(Reference Lazarus and Folkman58).
Urinary free cortisol
Free cortisol in urine represents a direct filtration fraction of blood-free cortisol and tends to parallel the cortisol production rate. Cortisol is usually referred to as the ‘stress hormone’ as it is involved in response to stress and anxiety(Reference Mason, Giller and Kosten48, Reference Martin, Rezzi and Peré-Trepat49). The 24 h collection time reflects the amount of cortisol that is released over a complete circadian cycle. This measure is insensitive to over- or underestimates obtained on moment-to-moment sampling that can occur due to transient fluctuations. R&D Systems’ Cortisol Immunoassay (Ref KGE008) is a competitive enzyme immunoassay designed to measure cortisol in urine. This assay is based on competitive binding of cortisol with a fixed amount of horseradish peroxidase-labelled cortisol for sites on a mouse monoclonal antibody. During the incubation, the monoclonal antibody becomes bound to the goat anti-mouse antibody coated on the microplate. Following a wash to remove excess conjugate and the unbound sample, a substrate solution is added to the wells to determine the bound enzyme activity. The colour development is stopped, and the absorbance is read at 450 nm.
Per protocol evaluations were carried out for all efficacy parameters. Raw data from subjects completing all tests on both sessions were delivered on case report form paper and entered with double data entry. Data are expressed as the means and standard deviations (age) and as medians with interquartile ranges with inferior and superior quartile values. The parameters included the HSCL-90 global severity index and subscores of the HSCL-90, the HADS global score and HADS-A and HADS-D subscores, the PSS score, the CCL score and the cortisol level. As the assumptions required for parametric tests were not met, the comparisons between groups and repeated measures in each group were performed with the non-parametric MWT and Wilcoxon test (WT), respectively. Differences were considered significant at P < 0·05. The SPSS statistical software package version 11.0 (SPSS, Inc., Chicago, IL, USA) was used for statistical analyses.
Conditioned defensive burying
The Kruskal–Wallis test shows a group difference in the stress/anxiety score (H(df = 2) = 13·76; P = 0·001), which was lower in rats treated with PF (47·50 (36·25–68·75)) and diazepam (33·50 (33·50–41·13)) than with vehicle (62·25 (53·00–84·75)) (MWT: U = 36; P = 0·04 and U = 9·5; P = 0·0004, respectively).
Hopkins Symptom Checklist-90
As shown in Table 2, the percentage change in the global severity index after 30 d between baseline and follow-up was higher in the PF-treated subjects than in the PL-treated subjects (MWT: z = 1·98; P < 0·05), particularly due to improved somatisation, depression and anger–hostility subscales (MWT: z = 2·16; P = 0·03, z = 1·96; P < 0·05 and z = 2·41; P = 0·02, respectively).
* Median values were significantly different when PF is compared with PL: P < 0·05 (Mann–Whitney U test).
Hospital Anxiety and Depression Scale
The percentage changes in HADS and HADS-A scores were higher in the PF-treated subjects (MWT: z = 2·19; P = 0·03 and z = 1·92; P = 0·06, respectively) with baseline scores being equivalent (Table 3). No significant differences were observed for HADS-D scores between the two groups at baseline (MWT: z = 1·66; P < 0·10) and over time (MWT: z = 0·02; P = 1). However, the HADS-D subscore of PF-treated subjects decreased between the two sessions (WT: z = 2·65; P = 0·008), whereas that of the control subjects remained stable (WT: z = 0·60; P = 0·55) (Table 3).
* Median values tended to be different when PF is compared with PL: P < 0·10 (Mann–Whitney U test).
** Median values were significantly different when PF is compared with PL: P < 0·05 (Mann–Whitney U test).
*** Median values were significantly different when BL is compared with FU in each group: P < 0·01 (Wilcoxon test).
Perceived Stress Scale
As indicated in Table 3, no group differences were observed for PSS scores at baseline (MWT: z = 0·36; P = 0·72) and over time (MWT: z = 0·36; P = 0·72).
PL subjects increased their positive re-evaluation score between baseline and follow-up (WT: z = 2·79; P = 0·005), borderline for their problem solving score (WT: z = 1·91; P = 0·06), while PF subjects decreased their self-blame score and displayed a higher problem solving score between the two test sessions (WT: z = 2·50; P = 0·01 and z = 2·05; P = 0·04, respectively) (Table 4).
WT, Wilcoxon test.
* Median values tended to be different when PF is compared with PL: P < 0·10 (Mann–Whitney U test).
** Median values were significantly different when BL is compared with FU in each group: P < 0·05 (WT).
*** Median values were significantly different when BL is compared with FU in each group: P < 0·01 (WT).
Urinary free cortisol
Four PF subjects and two PL ones were discarded from analysis for not collecting their urines during 24 h. No significant differences were observed between the cortisol levels at baseline and over time (MWT: z = 1·11; P = 0·27 and z = 0·01; P = 1, respectively). However, the median urinary free cortisol level in ng/ml of PF-treated subjects decreased between baseline (50·5 (39·8–68)) and follow-up (43·7 (29·2–56·6) (WT: z = 2·03; P = 0·04)), whereas that of controls did not (47·4 (33·1–57·7) and 44·2 (31·7–52·7), respectively; WT: z = 1·08; P = 0·28).
In the conditioned defensive burying test of anxiety, in which rats pile bedding on the source of perceived stress(Reference Treit, Pinel and Fibiger40, Reference Messaoudi, Lalonde and Schroeder41), PF was better than PL, and similar to diazepam as the standard reference substance. These results favour the hypothesis of anxiolytic properties for this compound. It remains to be determined whether other anxiety tests will be equally sensitive, as one test does not necessarily generalise to others.
We next assessed whether a daily dose of L. helveticus R0052 and B. longum R0175 taken in combination over 30 d influenced the psychological impact of everyday life events in normal volunteers. PF-treated subjects had a lower global severity index of the HSCL-90 over time than PL-treated controls, due to lower values for somatisation, depression and anger–hostility. The potential usefulness of PF as an anti-stress/anti-anxiety agent is further supported by diminished HADS global scores over time, due to a lower HADS-A subscore. Taken together, PF appears to show a beneficial effect on general signs of anxiety and depression, which did not generalise to the PSS, although all three tests comprise self-reported measures. It remains to be determined whether the PSS is sensitive during a longer treatment period. The CCL provides an assessment of coping strategies used to counter the stress of daily life. The two groups differed in emotional reactivity, with subjects administered PF reducing their self-blame score, while controls increasing their positive re-evaluation score. Moreover, PF-treated volunteers reported being more focused on the problem solving dimension than controls. In addition, cortisol values of PF-treated subjects decreased over time, while that of controls remained stable. Diop et al. (Reference Diop, Guillou and Durand34) reported beneficial effects of the same mixture administered for 3 weeks on self-reported stress-related GI disturbances. But unlike the present results, they observed no effect of treatment on psychological symptoms. This discrepancy may be due to the duration of the period of administration of the preparation and/or to the use of a different questionnaire on stress-induced symptoms at the beginning and the end of the trial(Reference Diop, Guillou and Durand34).
Other probiotics provide favourable results on behaviour. L. helveticus was demonstrated to favour sleep in elderly subjects(Reference Yamamura, Morishima and Kumano-go39). The Lactobacillus casei Shirota strain improved mood scores in normal subjects(Reference Benton, Williams and Brown59) and decreased anxiety in patients with chronic fatigue syndrome(Reference Rao, Bested and Beaulne60). It is interesting to note that the latter treatment increased the GI content of Lactobacillus and bifidobacteria.
The beneficial effects of probiotics on anxiety and depression may be explained by competitive exclusion of deleterious gut pathogens, decreases in pro-inflammatory cytokines and communication with the central nervous system via vagal sensory fibres, leading to changes in neurotransmitter levels or function(Reference Forsythe, Sudo and Dinan1, Reference Yan and Polk61–Reference Ramiah, van Reenen and Dicks63). As for the first explanation, marked alterations of the GI microflora occur in autism, including increases in various Clostridium spp., competitively displaced as other potentially pathogenic gut bacteria by Lactobacillus (Reference Ramiah, van Reenen and Dicks63, Reference Parracho, McCartney and Gibson64). It has been shown that the addition of B. longum R0175, one of the strains used here, increased the number of bifidobacteria in the GI content of pigs(Reference Estrada, Drew and Van Kessel35).
Clostridium and Bacteroides spp. produce propionic acid, a SCFA increasing anxiety and aggression in animals(Reference Hanstock, Clayton and Li65), as well as increasing social isolation and stereotypes while decreasing play(Reference Shultz, MacFabe and Ossenkopp66). While L. helveticus R0052 had never been tested in competition with Clostridium, this strain was recently demonstrated to protect GI microflora against the invasion of pathogenic bacteria(Reference Wine, Gareau and Johnson-Henry36).
The role of inflammatory processes on emotion is indicated by findings of a link between depression and elevated levels of IL-6, TNF and C-reactive protein(Reference Alesci, Martinez and Kelkar67). Systemically injected cytokines induce depressive symptoms(Reference Capuron, Neurauter and Musselman68, Reference Hauser, Khosla and Aurora69), prevented by antidepressants(Reference Musselman, Lawson and Gumnick70). It has been suggested that antidepressants act in part via generation of perhaps the most potent immunoregulatory cytokine, IL-10, thereby suppressing inflammation and depressive mood(Reference Maes71). Lactobacillus and Bifidobacterium strains attenuated inflammatory responses or else induced IL-10 production in rodents(Reference Desbonnet, Garrett and Clarke72–Reference Karimi, Inman and Bienenstock74). In accordance with this finding, both L. helveticus R0052 and B. longum R0175 showed anti-inflammatory properties in human cell lines(Reference Wallace, Bradley and Buckley37). Thus, bacteria may be used to influence mood in patients with elevated inflammatory chemicals(Reference Logan and Katzman75).
The normal activity of the HPA axis is regulated by diurnal excitatory inputs, stress-induced stimulation and various negative feedback loops, mediated by corticotrophin-releasing hormone, adrenocorticotrophin hormone and to a large extent by cortisol(Reference Munck, Guyre and Holbrook76). However, the ability of cortisol to regulate its own production may be impaired during chronic stress, resulting in sustained increase in its plasma level(Reference Ottenweller, Natelson and Pitman77). In the present study, the daily administration of PF for 30 d significantly decreased urinary free cortisol levels in subjects under daily life events as a source of stress. The administration of bacteria may support resilience and positively alter stress-related emotional behaviour in stressed animals(Reference Lowry, Hollis and de Vries78). To our knowledge, no clinical study has yet reported on measurements of cortisol evolution following oral subchronic treatment with probiotics. However, in preclinical studies, corticosterone levels decreased in rat pups in response to lactobacilli strains(Reference Gareau, Jury and MacQueen28). Likewise, germ-free mice had an increased responsiveness of the HPA axis compared with specific pathogen-free mice, reversed with a probiotic treatment before 6 weeks of age(Reference Sudo, Chida and Aiba25). Enterochromaffin cells, the source of serotonin in the bowel, may be involved, since these are affected by enteric flora and release neuroendocrine mediators activating afferents to the HPA axis as well as the paraventricular hypothalamus, amygdala and bed nucleus of the stria terminalis controlling stress responses and mood(Reference Goehler, Park and Opitz24).
Consumption of the PF containing L. helveticus R0052 and B. longum R0175 in combination mitigated psychological distress in three tests without displaying any adverse event. These results provide further evidence that gut microflora play a role in stress, anxiety and depression, perhaps via the enteric nervous system as well as centrally. Subject to the confirmation of these results, probiotics might offer a useful novel therapeutic approach to neuropathological disorders and/or as adjunct therapies in psychiatric disorders(Reference Logan and Katzman75). Though these data are preliminary, preclinical and clinical investigations should be extended to examine specific gut microbes and physiological markers associated with psychological distress.
The authors thank the Rosell-Lallemand Group for supplying the PF samples. This clinical trial was funded by Rosell-Lallemand, Blagnac, France. M. M. and D. D. contributed to the planning of the clinical trial, conducted all data collection and analysis and prepared the first draft of the manuscript. R. L. and C. R. contributed to the data interpretation and manuscript writing. N. V., H. J., A. N. and J.-F. B. contributed to the data management and provided intellectual input into the preparation of the manuscript. M. P., J.-M. C. and M. C. provided infrastructure (BIOFORTIS), contributed to the planning of the clinical trial, supervised data collection and provided intellectual input into the preparation of the manuscript. All authors participated in the concept and design of the study, critically reviewed the manuscript and approved the final version submitted to the British Journal of Nutrition. None of the authors has any financial relationship with the funding sponsor, and there were no conflicts of interest.