Hostname: page-component-586b7cd67f-2plfb Total loading time: 0 Render date: 2024-11-30T21:41:29.199Z Has data issue: false hasContentIssue false

Maternal consumption of fructo-oligosaccharide diminishes the severity of skin inflammation in offspring of NC/Nga mice

Published online by Cambridge University Press:  27 October 2009

Reiko Fujiwara
Affiliation:
Laboratory of Gastrointestinal Physiology, Graduate School of Life Science, Hokkaido University, Sapporo060-8589, Japan
Naoki Takemura
Affiliation:
Laboratory of Gastrointestinal Physiology, Graduate School of Life Science, Hokkaido University, Sapporo060-8589, Japan
Jun Watanabe
Affiliation:
Creative Research Institute ‘Sousei’, Hokkaido University, Sapporo001-0021, Japan
Kei Sonoyama*
Affiliation:
Division of Applied Bioscience, Laboratory of Food Biochemistry, Research Faculty of Agriculture, Hokkaido University, Sapporo060-8589, Japan
*
*Corresponding author: Dr Kei Sonoyama, fax +81 11 706 2496, email [email protected]
Rights & Permissions [Opens in a new window]

Abstract

Strategies to manipulate the gut microbiota in infancy have been considered to prevent the development of allergic diseases later in life. We aimed to elucidate the effects of maternal dietary supplementation with a prebiotic oligosaccharide on gut microbiota and spontaneously developing atopic dermatitis-like skin lesions in the offspring of NC/Nga mice. Female NC/Nga mice were fed diets either with or without fructo-oligosaccharide supplementation during pregnancy and lactation. After weaning, offspring were fed the diets supplemented with or without fructo-oligosaccharide for 11 weeks in an air-uncontrolled conventional room. Changes in gut microbiota were assessed by denaturing gradient gel electrophoresis of the PCR-amplified 16S rRNA gene. Skin lesions were evaluated by a clinical score and scratching behaviour. Serum antibody levels were measured by ELISA, and expression levels of cytokines and chemokines in lesional tissue were evaluated by quantitative RT-PCR. Maternal supplementation with fructo-oligosaccharide modulated the gut microbiota in sucklings. Although maternal supplementation with fructo-oligosaccharide suppressed the increase in clinical skin severity score and scratching behaviour in offspring, dietary fructo-oligosaccharide after weaning was less effective. The diminution of skin lesions was accompanied by lower serum concentrations of total IgG1 and lower expression levels of TNF-α in the lesional tissue. These data suggest that maternal consumption of fructo-oligosaccharide diminishes the severity of atopic dermatitis-like skin lesions in the offspring of NC/Nga mice.

Type
Full Papers
Copyright
Copyright © The Authors 2009

The gut microbiota play an important role in maintaining host health by preventing the colonisation of pathogens, degrading dietary compounds and maintaining normal mucosal immunity. Particularly, the composition of the gut microbiota early in life profoundly influences later immune responses(Reference Cebra1Reference Kelly, King and Aminov4). Therefore, strategies to manipulate the microbiota in infancy have been considered in preventing the onset of allergic diseases. This idea is supported by epidemiological data demonstrating that differences in the composition of gut microbiota in infancy precede the development of atopic dermatitis (AD)(Reference Bjorksten, Sepp and Julge5Reference Penders, Thijs and van den Brandt8). In addition, clinical trials showed that maternal administration of Lactobacillus rhamnosus GG (i.e. probiotics) during pregnancy and lactation was beneficial in preventing the development of AD in at-risk children during the first 4 years of life(Reference Kalliomaki, Salminen and Arvilommi9, Reference Kalliomaki, Salminen and Poussa10). However, L. rhamnosus GG administration did not benefit adolescents suffering from birch pollen allergy(Reference Helin, Haahtela and Haahtela11). These findings suggest that the time point for exposure to probiotics is important in preventing the development of allergic diseases.

Indigestible oligosaccharides are regarded as prebiotics that affect the host by selectively stimulating the growth and/or activity of beneficial bacteria such as bifidobacteria and lactobacilli in the intestinal tract, and thus improving host health(Reference Gibson, Beatty and Wang12Reference Macfarlane, Macfarlane and Cummings15). Indeed, a mixture of long-chain fructo-oligosaccharide (FOS) and short-chain galacto-oligosaccharide reportedly reduced the incidence of AD in formula-fed high-risk infants(Reference Moro, Arslanoglu and Stahl16, Reference Arslanoglu, Moro and Schmitt17). Our animal studies demonstrated that dietary raffinose and α-linked galacto-oligosaccharide reduced allergic airway inflammation in ovalbumin (OVA)-sensitised Brown Norway rats(Reference Watanabe, Sonoyama and Watanabe18, Reference Sonoyama, Watanabe and Watanabe19). We also showed that dietary short-chain FOS reduced 2,4-dinitrofluorobenzene-induced contact hypersensitivity in BALB/c mice(Reference Watanabe, Sasajima and Aramaki20). In addition, Fujitani et al. showed that dietary FOS reduced the infiltration of inflammatory cells and oedema formation in duodenal mucosa using an OVA-induced food allergy model in NC/jic mice(Reference Fujitani, Ueno and Kamiya21). Furthermore, Vos et al. reported that consumption of a mixture of long-chain FOS and short-chain galacto-oligosaccharide suppressed allergic airway inflammation in an OVA-induced allergic asthma model of BALB/c mice(Reference Vos, van Esch and Stahl22). These findings suggest that administration of indigestible oligosaccharides is effective in the prevention of allergic diseases through modulation of gut microbiota. Given that the time point for administration of probiotics, namely the early phase of life, is important in preventing allergic diseases, modulation of gut microbiota by prebiotics in infancy may produce desirable effects later in life.

We recently demonstrated that supplementation with FOS in female BALB/c mice during pregnancy and lactation altered the composition of gut microbiota in their sucklings(Reference Fujiwara, Watanabe and Sonoyama23). This finding provides a model that enables the examination of whether the modulation of gut microbiota by prebiotics in infancy influences the onset of allergic diseases later in life. Matsuda and colleagues have demonstrated that NC/Nga mice provide an excellent animal model for human AD(Reference Kawamoto, Matsuda and Chan24). When NC/Nga mice are kept in air-uncontrolled conventional surroundings, they develop skin lesions. In contrast, when kept in a specific pathogen-free room, they exhibit no clinical signs. Therefore, environmental allergens such as mite antigens are thought to contribute to the development of skin lesions(Reference Sasakawa, Higashi and Sakuma25). Clinical signs begin with scratching behaviour and, starting at the age of 8 weeks, IgE elevation, followed by the onset of eczematous conditions along with the infiltration of various inflammatory cells in the skin lesions(Reference Kawamoto, Matsuda and Chan24). Elevated expression of Th2 cytokines and chemokines is observed in lesional skin areas(Reference Kawamoto, Matsuda and Chan24). The present study examined whether the administration of FOS in female NC/Nga mice during pregnancy and lactation influences the development of dermatitis in their offspring.

Materials and methods

Animals and diets

The following study was approved by the Hokkaido University Animal Use Committee, and animals were maintained in accordance with the guidelines for the care and use of laboratory animals of Hokkaido University.

Four male (aged 9 weeks) and twelve female (aged 8 weeks) NC/Nga mice were purchased from Japan SLC (Hamamatsu, Japan) and housed in standard plastic cages in a temperature-controlled (23 ± 2°C) room with a dark period from 20.00 to 08.00 hours. The room was not equipped with air purifiers such as a HEPA (high efficiency particulate air) filter. Mice were allowed ad libitum access to food and water, and were fed either a synthetic diet prepared according to AIN-93G guidelines (Table 1) (Reference Reeves, Nielsen and Fahey26) or the same diet supplemented with FOS (Meioligo P, donated by Meiji Food Materia Co., Tokyo, Japan). These diets were referred to as FOS ( − ) and FOS (+), respectively. The FOS (+) was prepared by adding (50 g/kg diet) FOS to the FOS ( − ) at the expense of α-maize starch. According to the manufacturer, FOS is composed of d-glucose and d-fructose (1·3 %), sucrose (2·5 %), 1-kestose (37·3 %), nystose (49·1 %) and fructosylnystose (9·8 %).

Table 1 Composition of test diets

FOS ( − ), diet without fructo-oligosaccharide supplementation; FOS (+), diet with fructo-oligosaccharide supplementation; FOS, fructo-oligosaccharide.

* Amylalpha CL (Chuo-Shokuryou, Aichi, Japan).

ALACID (New Zealand Dairy Board, Wellington, New Zealand).

Gifted from Nippon Beet Sugar Mfg. Co. (Obihiro, Japan).

§ Wako Pure Chemical Industries, Ltd (Osaka, Japan).

Cellulose powder type D (Advantec Toyo Kaisa Ltd, Tokyo, Japan).

The mineral mixture and vitamin mixture (Nihon Nosan Kogyo Co., Yokohama, Japan) are identical to AIN-93G-MX and AIN-93G-VX as reported by Reeves et al. (Reference Reeves, Nielsen and Fahey26), respectively.

** FOS (Meioligo P) was donated by Meiji Food Materia Co. (Tokyo, Japan). According to the manufacturer, FOS is composed of D-glucose and D-fructose (1·3 %), sucrose (2·5 %), 1-kestose (37·3 %), nystose (49·1 %) and fructosylnystose (9·8 %).

Experimental design

Mice were divided into two groups: FOS ( − ) and FOS (+) (two male and six female mice in each group). Three females were mated to one male in one cage. Pregnant mice were housed individually and then monitored daily until delivery. The day of birth was referred to as day 0 of neonatal life. Offspring were separated from dams on day 21. We used female offspring in the following experiment, because male mice occasionally develop fighting-related skin lesions. Offspring were then fed either FOS ( − ) or FOS (+) for 11 weeks. FOS ( − )-fed offspring whose dam was fed FOS ( − ) were referred to as FOS ( − )( − ) (n 5). Thus, offspring were divided into four groups: FOS ( − )( − ), FOS ( − )(+), FOS (+)( − ) and FOS (+)(+) (five, six, six and six animals per group, respectively). The experimental design is summarised in Fig. 1. In order to induce stable dermatitis, two or three offspring in each group were housed during the experimental period in one cage together with two or three 13-week-old female NC/Nga mice with severe skin lesions, according to Takano et al. (Reference Takano, Arai and Kurachi27). Fresh faeces were collected on days 14, 35 and 98 after birth for analysis of intestinal microbiota as described below. At age 98 d, mice were anaesthetised by inhalation of diethyl ether. Whole blood was drawn from the carotid artery, and serum samples were stored at − 80°C for antibody measurements as described below. The right ear auricle of each mouse was excised, snap-frozen in liquid N2, and stored at − 80°C for isolation and analysis of RNA as described below.

Fig. 1 Schematic presentation of the experimental design. Female NC/Nga mice were fed a diet supplemented with fructo-oligosaccharide (FOS (+); ■) or without fructo-oligosaccharide (FOS ( − ); □) during pregnancy and lactation. After weaning at age 21 d, offspring were fed either FOS (+) or FOS ( − ) for 11 weeks. FOS ( − )-fed offspring whose dam was fed FOS ( − ) were referred to as FOS ( − )( − ). Thus, offspring were divided into four groups: FOS ( − )( − ), FOS ( − )(+), FOS (+)( − ) and FOS (+)(+). Faecal samples at ages 14, 35 and 98 d were subjected to PCR–denaturing gradient gel electrophoresis analysis to assess changes in gut microbiota.

Evaluation of severity of atopic dermatitis-like skin lesions

The severity of AD-like skin lesions in mice was scored weekly for the presence of (1) flare haemorrhage, (2) oedema, (3) excoriation and erosion, and (4) incrustation and xerosis(Reference Ohmura, Konomi and Satoh28). For each of these signs, a score was assigned as follows: 0, no sign; 1, mild; 2, moderate; or 3, severe. The sum of the individual scores was taken as the dermatitis score. In addition, the frequency of scratching behaviour, such as scratching of the nose, ears and dorsal skin with the hind paws, was measured during a 10 min period in each mouse at age 91 d. A series of scratching behaviours generally observed for about 1 s was counted as one incident of scratching according to Kuraishi et al. (Reference Kuraishi, Nagasawa and Hayashi29). Observers who were blinded to the experimental groups performed these observations.

Antibody ELISA

Ninety-six-well microtitre plates (Nunc, Roskilde, Denmark) were coated overnight at 4°C with rat anti-mouse IgE (100-fold dilutions, clone ME-01-DE; Yamasa, Choshi, Japan) diluted with carbonate buffer (pH 9·6). After blocking with PBS containing 1 % bovine serum albumin and 0·02 % Tween-20 (PBS-BT) for 1 h at 37°C, serial dilutions of serum with PBS-BT were added and incubated for 2 h at 37°C. Thereafter, biotinylated anti-mouse IgE (1000-fold dilutions, clone ME-02-B; Yamasa) diluted with PBS-BT was added and incubated for 2 h at room temperature. Horseradish peroxide-conjugated streptavidin (1000-fold dilutions; GE Healthcare Life Sciences, Tokyo, Japan) diluted with PBS-BT was added and incubated for 15 min at room temperature. Between each step, wells were washed five times with PBS containing 0·02 % Tween-20. Plates were developed at room temperature after the addition of commercially available substrate reagent (TMB One Component HRP Microwell Substrate; BioFX Laboratories, MD, USA). Finally, 1 m-H2SO4 was added, and absorbance at 450 nm was measured using a microplate reader (Genios plus; Tecan, Salzburg, Austria). For IgG1, anti-mouse IgG1 (200-fold dilutions, clone 15H6; Southern Biotech, Birmingham, AL, USA) and horseradish peroxide-conjugated rabbit anti-mouse IgG1 (80 000-fold dilutions; Zymed Laboratories, San Francisco, CA, USA) were used as capture and detection antibodies, respectively. For IgG2a, anti-mouse IgG2a (200-fold dilutions, clone HOPC-1; Southern Biotech) and horseradish peroxide-conjugated rabbit anti-mouse IgG1 (5000-fold dilutions; Zymed Laboratories) were used as capture and detection antibodies, respectively. Correlation coefficient values (r 2) of standard curves for IgE, IgG1 and IgG2a were 0·99, 0·91 and 0·91, respectively. Detection limits for IgE, IgG1 and IgG2a were 0·78 μg/ml, 0·39 mg/ml and 3·9 mg/ml, respectively. The average absorbance in substrate blank wells, to which three times the standard deviation was added, provided the detection limit. Inter-assay and/or intra-assay CV were less than 10 %.

Quantitative RT-PCR for cytokine gene expression in mouse tissue

Because the external ear is one of the constantly lesioned tissues, we used it for assessment of cytokines and chemokines. Total RNA was isolated from the ear auricle as previously described(Reference Fujiwara, Watanabe and Sonoyama23). In order to remove any genomic DNA, RQ1 RNase-free DNase (Promega, Madison, WI, USA) was added to the total RNA sample and then incubated at 37°C for 45 min. After phenol–chloroform extraction, total RNA was precipitated with ethanol and then re-suspended in 8 μl double-distilled water. Approximately 10 ng total RNA was annealed with Oligo (dT)12–18 primer (Invitrogen, Carlsbad, CA, USA) at 70°C for 10 min, and first-strand cDNA was then synthesised using Moloney murine leukemia virus RT (M-MLV RT; Invitrogen). In detail, a 20 μl solution composed of 1 × first-strand buffer, 0·5 mm each deoxynucleoside triphosphate (dNTP), 10 mm-dithiothreitol, 40 U of RNase OUT and 200 U of M-MLV RT was incubated for 50 min at 42°C, followed by RNA digestion with DNase-free RNase H (Invitrogen).

Real-time PCR was performed using a Thermal Cycler Dice TP800 (Takara, Ohtsu, Japan). Primer sequences were as follows: TNF-α (forward, 5′ CAT CTT CTC AAA ATT CGA GTG ACA A; reverse, 5′ TGG GAG TAG ACA AGG TAC AAC CC); IL-6 (forward, 5′ GAG GAT ACC ACT CCC AAC AGA CC; reverse, 5′ AAG TGC ATC ATC GTT GTT CAT ACA); IL-8 (forward, 5′ ATG GCT GGG ATT CAC CTC AA; reverse, 5′ AAG CCT CGC GAC CAT TCT T); thymus and activation-regulated chemokine (TARC) (forward, 5′ GAG CTG GTA TAA GAC CTC AGT GGA G; reverse, 5′ TGG CCT TCT TCA CAT GTT TGT C); glyceraldehyde-3-phosphate dehydrogenase (forward, 5′ TCA CCA CCA TGG AGA AGG C; reverse, 5′ GCT AAG CAG TTG GTG CA)(Reference Giulietti, Overbergh and Valckx30, Reference Heishi, Imai and Katayama31). Amplification was carried out in a 25 μl reaction volume containing 12·5 μl 1 × SYBR Premix Ex Taq (Takara), 200 nm of each primer and 1 μl of template cDNA. The reaction condition was: 95°C for 10 s, followed by forty cycles at 95°C for 5 s and 60°C for 30 s, with dissociation curve at 95°C for 15 s, 60°C for 30 s and 95°C for 15 s. Relative gene expression levels for each sample were normalised to the levels for glyceraldehyde-3-phosphate dehydrogenase.

Profile analysis of faecal microbiota by PCR–denaturing gradient gel electrophoresis

DNA was extracted from fresh faeces using a faecal DNA isolation kit (MO Bio Laboratories, Carlsbad, CA, USA) according to the manufacturer's instructions. DNA samples were used as a template to amplify the fragments of 16S rRNA gene with universal primers, U968-GC (CGC CCG GGG CGC GCC CCG GGC GGG GCG GGG GCA CGG GGG GAA CGC GAA GAA CCT TAC) and L1401 (CGG TGT GTA CAA GAC CC)(Reference Zoetendal, Akkermans and de Vos32), and denaturing gradient gel electrophoresis (DGGE) analysis of the amplicon was carried out as previously described(Reference Fujiwara, Watanabe and Sonoyama23). Quantity One software (version 4.6.0; Bio-Rad, Hercules, CA, USA) was used for band identification and normalisation of band patterns from DGGE gels. The Dice similarity coefficient was used for computing sample similarity based on band position and intensity, and the unweighted pair group method using the arithmetic average (UPGMA) algorithm was used to construct a dendrogram of DGGE band profiles as previously described(Reference Fujiwara, Watanabe and Sonoyama23).

Statistical analysis

Results are presented as mean values with their standard errors. The Tukey–Kramer test (for frequency of scratching behaviour, antibodies and cytokines) or the Mann–Whitney U test (for clinical skin severity score) following two-way ANOVA was used to analyse the differences among the means of the experimental groups. StatView for Macintosh (version 5.0; SAS Institute, Inc., Cary, NC, USA) was used for the analysis. Differences were considered significant at P < 0·05.

Results

Effect of maternal supplementation with fructo-oligosaccharide on gut microbiota in offspring

As shown in Fig. 1, female NC/Nga mice were fed either a diet supplemented with or without FOS during pregnancy and lactation. PCR–DGGE analysis of the 16S rRNA gene in the faeces of lactating mice showed that dietary FOS influenced the composition of the gut microbiota (data not shown), which is consistent with our previous study(Reference Fujiwara, Watanabe and Sonoyama23). In suckling pups at age 14 d, PCR–DGGE band profiles are divided into two large clusters: one comprising pups whose dams were fed FOS ( − ), and the other comprising pups whose dams were fed FOS (+) (Fig. 2(A) and (B)). At age 21 d, all female offspring were separated from dams and then fed either FOS ( − ) or FOS (+). At 2 weeks after weaning, no distinct cluster is shown in PCR–DGGE band profiles (Fig. 2(C) and (D)), suggesting the cessation of influence of maternal supplementation with FOS. Thereafter, offspring at age 98 d showed a new cluster pattern with two large clusters comprised of offspring fed FOS ( − ) and FOS (+) (Fig. 2(E) and (F)).

Fig. 2 PCR–denaturing gradient gel electrophoresis (DGGE) analysis of faecal microbiota based on 16S rRNA gene sequences in offspring of NC/Nga mice. DGGE gel images (A, C and E) and respective dendrograms (B, D and F) of DGGE band profiles on days 14 (A and B), 35 (C and D) and 98 (E and F) are shown. Each lane in the gel images and each line in the dendrograms represent individual mice. In (A) and (B), FOS ( − ) and FOS (+) represent sucklings of dams fed diets without and with fructo-oligosaccharide supplementation, respectively. For explanation of the offspring groups in (C) to (F), see Fig. 1. Distance is measured in arbitrary units. M, size marker.

Effect of maternal supplementation with fructo-oligosaccharide on atopic dermatitis-like skin lesions in offspring

In order to induce stable dermatitis, weaned offspring were housed together with 13-week-old female NC/Nga mice with severe skin lesions under air-uncontrolled conventional conditions(Reference Takano, Arai and Kurachi27). Skin lesions were first observed at age 49 d, and the clinical skin severity score continued to increase up to the end of the experiment, i.e. age 98 d, in all the mice (Fig. 3(A)). Two-way ANOVA showed that maternal feeding with FOS significantly affected the clinical skin severity score in offspring at ages 70, 77, 84, 91 and 98 d. In contrast, consumption of FOS in offspring showed no significant influence. Thus, the clinical skin severity score was significantly lower in the FOS (+)( − ) and FOS (+)(+) groups than in the FOS ( − )( − ) group from age 70 d to age 98 d. The score in the FOS ( − )(+) group did not significantly differ from that in the FOS ( − )( − ) group and from that in the FOS (+)( − ) and FOS (+)(+) groups throughout the experimental period.

Fig. 3 Severity of spontaneously developing atopic dermatitis-like skin lesions in offspring of NC/Nga mice. (A) Changes in clinical skin severity score in offspring fed a post-weaning diet without fructo-oligosaccharide (FOS) whose dams were not supplemented with FOS (FOS ( − )( − ); ), in offspring fed a post-weaning diet supplemented with FOS whose dams were not supplemented with FOS (FOS ( − )(+); ), in offspring fed a post-weaning diet without FOS whose dams were supplemented with FOS (FOS (+)( − ); ) and in offspring fed a post-weaning diet supplemented with FOS whose dams were supplemented with FOS (FOS (+)(+); □). Values are means, with standard errors represented by vertical bars. P values at ages 70, 77, 84, 91 and 98 d were P = 0·0022, P = 0·0025, P = 0·0074, P = 0·0107 and P = 0·0119, respectively, for FOS in dams, P = 0·4006, P = 0·0860, P = 0·0838, P = 0·1914 and P = 0·4415, respectively, for FOS in offspring, and P = 0·2618, P = 0·1979, P = 0·2017, P = 0·1914 and P = 0·2142, respectively, for FOS in dams × FOS in offspring (two-way ANOVA). a,b Mean values with unlike letters on each day were significantly different (P < 0·05; Mann–Whitney U test). (B) Frequency of scratching behaviour for 10 min at age 91 d. Values are means, with standard errors represented by vertical bars. P values were P = 0·0165, P = 0·7377 and P = 0·7931 for FOS in dams, FOS in offspring and their interaction, respectively (two-way ANOVA). a,b Mean values with unlike letters were significantly different (P < 0·05; Tukey–Kramer test).

AD-like skin lesions in NC/Nga mice are accompanied by a higher frequency of scratching behaviour, such as scratching of the nose, ears and dorsal skin with the hind paws(Reference Kawamoto, Matsuda and Chan24). Fig. 3(B) shows the frequency of scratching behaviour in offspring at age 91 d. Two-way ANOVA showed that maternal feeding with FOS significantly affected the frequency of scratching. However, dietary FOS in offspring after weaning showed no significant influence. The frequency of scratching behaviour was significantly lower in the FOS (+)( − ) and FOS (+)(+) groups than in the FOS ( − )( − ) groups. The value in the FOS (+)( − ) group was intermediate. Thus, data on the frequency of scratching behaviour were consistent with the clinical skin severity score in offspring.

Effect of maternal supplementation with fructo-oligosaccharide on serum antibodies and tissue cytokines and chemokines in offspring

AD-like skin lesions in NC/Nga mice are accompanied by higher concentrations of serum IgE(Reference Kawamoto, Matsuda and Chan24). In the present study, although there was no significant difference in the serum concentration of total IgE in offspring aged 98 d among the groups (Fig. 4(A)), the concentration of total IgG1, i.e. Th2 response in mice(Reference Snapper, Finkelman and Paul33), was significantly lower in the FOS (+)( − ) and FOS (+)(+) groups than in the FOS ( − )( − ) and FOS ( − )(+) groups (Fig. 4(B)). Two-way ANOVA showed that the concentration of total IgG1 was significantly affected by maternal, but not post-weaning, supplementation with FOS. No consistent change was observed in the concentration of total IgG2a, i.e. Th1 response in mice (Fig. 4(C)).

Fig. 4 Serum concentrations of total IgE (A), IgG1 (B) and IgG2a (C) in offspring of NC/Nga mice at age 98 d. FOS, fructo-oligosaccharide; FOS ( − )( − ), offspring fed a post-weaning diet without FOS, dams not supplemented with FOS; FOS ( − )(+), offspring fed a post-weaning diet supplemented with FOS, dams not supplemented with FOS; FOS (+)( − ), offspring fed a post-weaning diet without FOS, dams supplemented with FOS; FOS (+)(+), offspring fed a post-weaning diet supplemented with FOS, dams supplemented with FOS. Values are means, with standard errors represented by vertical bars. For total IgE, P values were P = 0·2020, P = 0·7181 and P = 0·6887 for FOS in dams, FOS in offspring and their interaction, respectively (two-way ANOVA). Similarly, P values for total IgG1 were P = 0·0011, P = 0·9248 and P = 0·5928 for FOS in dams, FOS in offspring and their interaction, respectively. P values for IgG2a were P = 0·9068, P = 0·3836 and P = 0·3068 for FOS in dams, FOS in offspring and their interaction, respectively. a,b Mean values with unlike letters were significantly different (P < 0·05; Tukey–Kramer test).

Expression of inflammation-associated cytokines and chemokines in ear auricles was evaluated by quantitative RT-PCR in offspring at age 98 d. Two-way ANOVA showed that the level of TNF-α mRNA was significantly affected by maternal, but not post-weaning, supplementation with FOS (Fig. 5). There was no significant difference in the level of IL-6, IL-8, and thymus and activation-regulated chemokine (TARC) among the groups.

Fig. 5 Ear auricle expression levels of IL-6, IL-8, thymus and activation-regulated chemokine (TARC) and TNF-α evaluated by quantitative RT-PCR in offspring of NC/Nga mice at age 98 d. (■), Offspring fed a post-weaning diet without fructo-oligosaccharide (FOS), dams not supplemented with FOS (FOS ( − )( − )); (), offspring fed a post-weaning diet supplemented with FOS, dams not supplemented with FOS (FOS ( − )(+)); (), offspring fed a post-weaning diet without FOS, dams supplemented with FOS (FOS (+)( − )); (□), offspring fed a post-weaning diet supplemented with FOS, dams supplemented with FOS (FOS (+)(+)). The levels for FOS ( − )( − ) are expressed relative to the control values, which are taken as 1. Values are means, with standard errors represented by vertical bars. For IL-6, P values were P = 0·1251, P = 0·7063 and P = 0·5766 for FOS in dams, FOS in offspring and their interaction, respectively (two-way ANOVA). Similarly, P values for IL-8 were P = 0·0882, P = 0·7435 and P = 0·2326 for FOS in dams, FOS in offspring and their interaction, respectively. P values for TARC were P = 0·2636, P = 0·6687 and P = 0·1408 for FOS in dams, FOS in offspring and their interaction, respectively. P values for TNF-α were P = 0·0275, P = 0·9225 and P = 0·9813 for FOS in dams, FOS in offspring and their interaction, respectively. a,b Mean values with unlike letters were significantly different (P < 0·05; Tukey–Kramer test).

Discussion

To our knowledge, the present study is the first demonstrating that modulation of the gut microbiota in infancy by maternal consumption with indigestible oligosaccharide diminished the severity of allergic skin inflammation. Our recent study showed that supplementation with FOS in female BALB/c mice during pregnancy and lactation altered the composition of the gut microbiota in their sucklings(Reference Fujiwara, Watanabe and Sonoyama23). This finding provided the impetus for the present study examining whether modulation of gut microbiota in infancy affects the development of AD-like skin lesions in NC/Nga mice. In the present study, reduced progress in spontaneous skin lesions in offspring, under air-uncontrolled conventional conditions where the room was not equipped with air purifiers such as a HEPA (high efficiency particulate air) filter, was associated with the consumption of FOS in their dams during pregnancy and lactation. The diminution of skin lesions was accompanied by a lower frequency of scratching behaviour, lower serum concentrations of IgG1 and lower expression levels of TNF-α mRNA in the lesional tissue. However, further investigation is required to elucidate the involvement of TNF-α in the diminution of skin lesions, because expression of mRNA is not always equivalent to expression of the corresponding protein. Supplementation with FOS in offspring after weaning was shown to be less effective as compared with maternal supplementation with FOS. PCR–DGGE analysis of 16S rRNA gene profiles in the faeces of offspring suggested that maternal supplementation with FOS modulated the gut microbiota in sucklings, and that the gut microbiota in offspring after weaning were controlled by the diet they were consuming. Therefore, the present study supports the idea that the time point for modulation of gut microbiota is important in preventing the development of allergic diseases.

It remains unclear what kind of gut bacterial species in infant NC/Nga mice was involved in diminishing the severity of AD-like skin lesions later in life. Previous epidemiological studies showed that differences in the composition of gut microbiota in infancy precede the development of AD. According to Bjorksten et al. babies who developed allergy were less often colonised with enterococci during the first month of life and with bifidobacteria during the first year of life when compared with healthy infants(Reference Bjorksten, Sepp and Julge5). In addition, allergic infants had higher counts of clostridia at 3 months, and the prevalence of colonisation with Staphylococcus aureus was also higher at 6 months, whereas the counts of Bacteroides were lower at 12 months(Reference Bjorksten, Sepp and Julge5). Kalliomaki et al. reported that atopic children had more clostridia at age 3 weeks(Reference Kalliomaki, Kirjavainen and Eerola6). Furthermore, Escherichia coli and Clostridium difficile were reportedly associated with infants who developed AD(Reference Penders, Stobberingh and Thijs7, Reference Penders, Thijs and van den Brandt8). Therefore, it might be possible that changes in the population of these bacterial species in infant NC/Nga mice are associated with the development of AD-like skin lesions later in life. In our preliminary experiments, quantitative RT-PCR of the 16S rRNA gene showed that bifidobacteria in the faeces of NC/Nga mice were below detection limits (106 copies/g faeces; R Fujiwara, J Watanabe and K Sonoyama, unpublished results). Further investigations are thus needed to elucidate the composition of infant gut microbiota beneficial in preventing allergic diseases later in life.

The mechanisms by which the gut microbiota in infancy control the development of allergic diseases later in life remain to be elucidated. Sawada et al. described that dietary supplementation with heat-killed L. rhamnosus GG in NC/Nga mice during pregnancy and lactation suppressed the spontaneous development of AD-like skin lesions in offspring(Reference Sawada, Morita and Tanaka34). In addition, Inoue et al. demonstrated that oral administration of L. johnsonii NCC533 in NC/Nga mice around the weaning period (i.e. age 20–22 d) prevented the development of allergic skin lesions induced by topical application of mite antigen from age 6 weeks(Reference Inoue, Nishio and Fukushima35). Furthermore, Blümer et al. reported that administration of L. rhamnosus GG in female BALB/c mice during pregnancy and lactation suppressed the OVA-induced allergic airway inflammation in their offspring(Reference Blümer, Sel and Virna36). These findings, as well as the present study, provide experimental models enabling the investigation of how the gut microbiota in infancy influence the development of the immune system and subsequently control the development of allergic diseases later in life.

Alternatively, the possibility that mediators, transferred from the mother to its offspring, are involved in the suppression of allergic diseases cannot be excluded. Indeed, Verhasselt et al. showed that breast milk-mediated transfer of OVA from mouse dams to their sucklings resulted in the induction of oral immune tolerance, which led to the prevention of allergic airway inflammation in OVA-immunised offspring(Reference Verhasselt, Milcent and Cazareth37). In addition, Polte et al. reported that the tolerance induced with oral OVA in pregnant BALB/c mice was protective against allergic airway inflammation in OVA-immunised offspring through breast milk-mediated mechanisms(Reference Polte and Hansen38). Furthermore, Blümer et al. demonstrated that prenatal L. rhamnosus GG administration in BALB/c mice increased TNF-α gene expression in placenta, which was associated with a reduction in OVA-induced airway inflammation in offspring(Reference Blümer, Sel and Virna36). In this context, modulation of the gut microbiota in pregnant and lactating NC/Nga mice by dietary FOS might influence the transfer of mediators from dams to offspring, which in turn results in the diminution of AD-like skin lesions later in life.

Because FOS is an indigestible oligosaccharide, dietary supplementation with FOS could lead to reduction of metabolisable energy intake. However, the present study did not measure body weight and food intake in dams and their offspring. Additionally, consumption of indigestible oligosaccharides such as FOS increases the production of SCFA, predominant fermentation products of bacteria, in the distal gut(Reference Roberfroid39). Therefore, we could not rule out the possibility that the diet-related difference in metabolisable energy intake and/or SCFA production contributed to the difference in the severity of skin lesions in the offspring.

In conclusion, our present study suggests that maternal consumption of FOS diminishes the severity of AD-like skin lesions in the offspring of NC/Nga mice. We therefore propose that the present findings support the idea that the modulation of gut microbiota by prebiotic oligosaccharides early in life may be beneficial in preventing dermatitis, although further investigation is required to reveal the underlying mechanisms involved. The present study provides an experimental model enabling the investigation of how gut microbiota in infancy influence immune system development and subsequently control the development of allergic diseases later in life.

Acknowledgements

The present study was partly supported by Special Coordination Funds for Promoting Science and Technology, by a Grant-in-Aid for Scientific Research from The Ministry of Education, Science, Sports and Culture of Japan (no. 19380070), and by the Mishima Kaiun Memorial Foundation.

R. F. and N. T. were involved in the animal experiments, ELISA and RT-PCR. J. W. performed the molecular biological analyses of gut microbiota. K. S. was involved in designing the study and writing the manuscript.

None of the authors has a conflict of interest.

References

1Cebra, JJ (1999) Influences of microbiota on intestinal immune system development. Am J Clin Nutr 69, 10461051.CrossRefGoogle ScholarPubMed
2Wills-Karp, M, Santeliz, J & Karp, CL (2004) The germless theory of allergic disease: revisiting the hygiene hypothesis. Nat Rev Immunol 1, 6975.CrossRefGoogle Scholar
3Macpherson, AJ & Harris, NL (2004) Interactions between commensal intestinal bacteria and the immune system. Nat Rev Immunol 4, 478485.CrossRefGoogle ScholarPubMed
4Kelly, D, King, T & Aminov, R (2007) Importance of microbial colonization of the gut in early life to the development of immunity. Mutat Res 622, 5869.CrossRefGoogle Scholar
5Bjorksten, B, Sepp, E, Julge, K, et al. (2001) Allergy development and the intestinal microflora during the first year of life. J Allergy Clin Immunol 108, 516520.CrossRefGoogle ScholarPubMed
6Kalliomaki, M, Kirjavainen, P, Eerola, E, et al. (2001) Distinct patterns of neonatal gut microflora in infants in whom atopy was and was not developing. J Allergy Clin Immunol 107, 129134.CrossRefGoogle Scholar
7Penders, J, Stobberingh, EE, Thijs, C, et al. (2006) Molecular fingerprinting of the intestinal microbiota of infants in whom atopic eczema was or was not developing. Clin Exp Allergy 36, 16021608.CrossRefGoogle Scholar
8Penders, J, Thijs, C, van den Brandt, PA, et al. (2007) Gut microbiota composition and development of atopic manifestations in infancy: the KOALA Birth Cohort Study. Gut 56, 661667.CrossRefGoogle ScholarPubMed
9Kalliomaki, M, Salminen, S, Arvilommi, H, et al. (2001) Probiotics in primary prevention of atopic disease: a randomised placebo-controlled trial. Lancet 357, 10761079.CrossRefGoogle ScholarPubMed
10Kalliomaki, M, Salminen, S, Poussa, T, et al. (2003) Probiotics and prevention of atopic disease: 4-year follow-up of a randomised placebo-controlled trial. Lancet 361, 18691871.CrossRefGoogle ScholarPubMed
11Helin, T, Haahtela, S & Haahtela, T (2002) No effect of oral treatment with an intestinal bacterial strain, Lactobacillus rhamnosus (ATCC 53103), on birch-pollen allergy: a placebo-controlled double-blind study. Allergy 57, 243246.CrossRefGoogle ScholarPubMed
12Gibson, GR, Beatty, ER, Wang, X, et al. (1995) Selective stimulation of bifidobacteria in the human colon by oligofructose and inulin. Gastroenterology 108, 975982.CrossRefGoogle ScholarPubMed
13Blaut, M (2002) Relationship of prebiotics and food to intestinal microflora. Eur J Nutr 41, 1116.CrossRefGoogle ScholarPubMed
14Sartor, RB (2004) Therapeutic manipulation of the enteric microflora in inflammatory bowel diseases: antibiotics, probiotics, and prebiotics. Gastroenterology 126, 620633.CrossRefGoogle ScholarPubMed
15Macfarlane, S, Macfarlane, GT & Cummings, JH (2006) Review Article: prebiotics in the gastrointestinal tract. Aliment Pharmacol Ther 24, 701714.CrossRefGoogle ScholarPubMed
16Moro, G, Arslanoglu, S, Stahl, B, et al. (2006) A mixture of prebiotic oligosaccharides reduces the incidence of atopic dermatitis during the first six months of age. Arch Dis Child 91, 814819.CrossRefGoogle ScholarPubMed
17Arslanoglu, S, Moro, GE, Schmitt, J, et al. (2008) Early dietary intervention with a mixture of prebiotic oligosaccharides reduces the incidence of allergic manifestations and infections during the first two years of life. J Nutr 138, 10911095.CrossRefGoogle ScholarPubMed
18Watanabe, H, Sonoyama, K, Watanabe, J, et al. (2004) Reduction of allergic airway eosinophilia by dietary raffinose in Brown Norway rats. Br J Nutr 92, 247255.CrossRefGoogle ScholarPubMed
19Sonoyama, K, Watanabe, H, Watanabe, J, et al. (2005) Allergic airway eosinophilia is suppressed in ovalbumin-sensitized Brown Norway rats fed raffinose and α-linked galactooligosaccharide. J Nutr 135, 538543.CrossRefGoogle ScholarPubMed
20Watanabe, J, Sasajima, N, Aramaki, A, et al. (2008) Consumption of fructo-oligosaccharide reduces 2,4-dinitrofluorobenzene-induced contact hypersensitivity in mice. Br J Nutr 100, 339346.CrossRefGoogle ScholarPubMed
21Fujitani, S, Ueno, K, Kamiya, T, et al. (2007) Increased number of CCR4-positive cells in the duodenum of ovalbumin-induced food allergy model Nc/jic mice and antiallergic activity of fructooligosaccharides. Allergol Int 56, 131138.CrossRefGoogle ScholarPubMed
22Vos, AP, van Esch, BC, Stahl, B, et al. (2007) Dietary supplementation with specific oligosaccharide mixtures decreases parameters of allergic asthma in mice. Int Immunopharmacol 7, 15821587.CrossRefGoogle ScholarPubMed
23Fujiwara, R, Watanabe, J & Sonoyama, K (2008) Assessing changes in composition of intestinal microbiota in neonatal BALB/c mice through cluster analysis of molecular markers. Br J Nutr 99, 11741177.CrossRefGoogle ScholarPubMed
24Kawamoto, K & Matsuda, H (2003) Spontaneous mouse model of atopic dermatitis in NC/Nga mice. In Animal Models of Human Inflammatory Skin Diseases, pp. 372384 [Chan, LS, editor]. San Diego, CA: CRC Press.Google Scholar
25Sasakawa, T, Higashi, Y, Sakuma, S, et al. (2001) Atopic dermatitis-like skin lesions induced by topical application of mite antigens in NC/Nga mice. Int Arch Allergy Immunol 126, 239247.CrossRefGoogle ScholarPubMed
26Reeves, PG, Nielsen, FH & Fahey, GC Jr (1993) AIN-93 purified diets for laboratory rodents: final report of the American Institute of Nutrition ad hoc writing committee on the reformulation of the AIN-76A rodent diet. J Nutr 123, 19391951.CrossRefGoogle Scholar
27Takano, N, Arai, I & Kurachi, M (2006) A method to induce stable atopic dermatitis-like symptoms in NC/Nga mice housed with skin-lesioned mice. Br J Dermatol 154, 426430.CrossRefGoogle ScholarPubMed
28Ohmura, T, Konomi, A, Satoh, Y, et al. (2004) Suppression of atopic-like dermatitis by treatment with antibody to lymphocyte function-associated antigen-1 in NC/Nga mouse. Eur J Pharmacol 504, 113117.CrossRefGoogle ScholarPubMed
29Kuraishi, Y, Nagasawa, T, Hayashi, K, et al. (1995) Scratching behavior induced by pruritogenic but not algesiogenic agents in mice. Eur J Pharmacol 275, 229233.CrossRefGoogle Scholar
30Giulietti, A, Overbergh, L, Valckx, D, et al. (2001) An overview of real-time quantitative PCR: applications to quantify cytokine gene expression. Methods 25, 386401.CrossRefGoogle ScholarPubMed
31Heishi, M, Imai, Y, Katayama, H, et al. (2003) Gene expression analysis of atopic dermatitis-like skin lesions induced in NC/Nga mice by mite antigen stimulation under specific pathogen-free conditions. Int Arch Allergy Immunol 132, 355363.CrossRefGoogle ScholarPubMed
32Zoetendal, EG, Akkermans, AD & de Vos, WM (1998) Temperature gradient gel electrophoresis analysis of 16S rRNA from human fecal samples reveals stable and host-specific communities of active bacteria. Appl Environ Microbiol 64, 38543859.CrossRefGoogle ScholarPubMed
33Snapper, CM, Finkelman, FD & Paul, WE (1988) Regulation of IgG1 and IgE production by interleukin 4. Immunol Rev 102, 5175.CrossRefGoogle ScholarPubMed
34Sawada, J, Morita, H, Tanaka, A, et al. (2007) Ingestion of heat-treated Lactobacillus rhamnosus GG prevents development of atopic dermatitis in NC/Nga mice. Clin Exp Allergy 37, 296303.CrossRefGoogle ScholarPubMed
35Inoue, R, Nishio, A, Fukushima, Y, et al. (2007) Oral treatment with probiotic Lactobacillus johnsonii NCC533 (La1) for a specific part of the weaning period prevents the development of atopic dermatitis induced after maturation in model mice, NC/Nga. Br J Dermatol 156, 499509.CrossRefGoogle ScholarPubMed
36Blümer, N, Sel, S, Virna, S, et al. (2007) Perinatal maternal application of Lactobacillus rhamnosus GG suppresses allergic airway inflammation in mouse offspring. Clin Exp Allergy 37, 348357.CrossRefGoogle ScholarPubMed
37Verhasselt, V, Milcent, V, Cazareth, J, et al. (2008) Breast milk-mediated transfer of an antigen induces tolerance and protection from allergic asthma. Nat Med 14, 170175.CrossRefGoogle ScholarPubMed
38Polte, T & Hansen, G (2008) Maternal tolerance achieved during pregnancy is transferred to the offspring via breast milk and persistently protects the offspring from allergic asthma. Clin Exp Allergy 38, 19501958.CrossRefGoogle Scholar
39Roberfroid, MB (2005) Introducing inulin-type fructans. Br J Nutr 93, Suppl. 1, S13S25.CrossRefGoogle ScholarPubMed
Figure 0

Table 1 Composition of test diets

Figure 1

Fig. 1 Schematic presentation of the experimental design. Female NC/Nga mice were fed a diet supplemented with fructo-oligosaccharide (FOS (+); ■) or without fructo-oligosaccharide (FOS ( − ); □) during pregnancy and lactation. After weaning at age 21 d, offspring were fed either FOS (+) or FOS ( − ) for 11 weeks. FOS ( − )-fed offspring whose dam was fed FOS ( − ) were referred to as FOS ( − )( − ). Thus, offspring were divided into four groups: FOS ( − )( − ), FOS ( − )(+), FOS (+)( − ) and FOS (+)(+). Faecal samples at ages 14, 35 and 98 d were subjected to PCR–denaturing gradient gel electrophoresis analysis to assess changes in gut microbiota.

Figure 2

Fig. 2 PCR–denaturing gradient gel electrophoresis (DGGE) analysis of faecal microbiota based on 16S rRNA gene sequences in offspring of NC/Nga mice. DGGE gel images (A, C and E) and respective dendrograms (B, D and F) of DGGE band profiles on days 14 (A and B), 35 (C and D) and 98 (E and F) are shown. Each lane in the gel images and each line in the dendrograms represent individual mice. In (A) and (B), FOS ( − ) and FOS (+) represent sucklings of dams fed diets without and with fructo-oligosaccharide supplementation, respectively. For explanation of the offspring groups in (C) to (F), see Fig. 1. Distance is measured in arbitrary units. M, size marker.

Figure 3

Fig. 3 Severity of spontaneously developing atopic dermatitis-like skin lesions in offspring of NC/Nga mice. (A) Changes in clinical skin severity score in offspring fed a post-weaning diet without fructo-oligosaccharide (FOS) whose dams were not supplemented with FOS (FOS ( − )( − ); ), in offspring fed a post-weaning diet supplemented with FOS whose dams were not supplemented with FOS (FOS ( − )(+); ), in offspring fed a post-weaning diet without FOS whose dams were supplemented with FOS (FOS (+)( − ); ) and in offspring fed a post-weaning diet supplemented with FOS whose dams were supplemented with FOS (FOS (+)(+); □). Values are means, with standard errors represented by vertical bars. P values at ages 70, 77, 84, 91 and 98 d were P = 0·0022, P = 0·0025, P = 0·0074, P = 0·0107 and P = 0·0119, respectively, for FOS in dams, P = 0·4006, P = 0·0860, P = 0·0838, P = 0·1914 and P = 0·4415, respectively, for FOS in offspring, and P = 0·2618, P = 0·1979, P = 0·2017, P = 0·1914 and P = 0·2142, respectively, for FOS in dams × FOS in offspring (two-way ANOVA). a,b Mean values with unlike letters on each day were significantly different (P < 0·05; Mann–Whitney U test). (B) Frequency of scratching behaviour for 10 min at age 91 d. Values are means, with standard errors represented by vertical bars. P values were P = 0·0165, P = 0·7377 and P = 0·7931 for FOS in dams, FOS in offspring and their interaction, respectively (two-way ANOVA). a,b Mean values with unlike letters were significantly different (P < 0·05; Tukey–Kramer test).

Figure 4

Fig. 4 Serum concentrations of total IgE (A), IgG1 (B) and IgG2a (C) in offspring of NC/Nga mice at age 98 d. FOS, fructo-oligosaccharide; FOS ( − )( − ), offspring fed a post-weaning diet without FOS, dams not supplemented with FOS; FOS ( − )(+), offspring fed a post-weaning diet supplemented with FOS, dams not supplemented with FOS; FOS (+)( − ), offspring fed a post-weaning diet without FOS, dams supplemented with FOS; FOS (+)(+), offspring fed a post-weaning diet supplemented with FOS, dams supplemented with FOS. Values are means, with standard errors represented by vertical bars. For total IgE, P values were P = 0·2020, P = 0·7181 and P = 0·6887 for FOS in dams, FOS in offspring and their interaction, respectively (two-way ANOVA). Similarly, P values for total IgG1 were P = 0·0011, P = 0·9248 and P = 0·5928 for FOS in dams, FOS in offspring and their interaction, respectively. P values for IgG2a were P = 0·9068, P = 0·3836 and P = 0·3068 for FOS in dams, FOS in offspring and their interaction, respectively. a,b Mean values with unlike letters were significantly different (P < 0·05; Tukey–Kramer test).

Figure 5

Fig. 5 Ear auricle expression levels of IL-6, IL-8, thymus and activation-regulated chemokine (TARC) and TNF-α evaluated by quantitative RT-PCR in offspring of NC/Nga mice at age 98 d. (■), Offspring fed a post-weaning diet without fructo-oligosaccharide (FOS), dams not supplemented with FOS (FOS ( − )( − )); (), offspring fed a post-weaning diet supplemented with FOS, dams not supplemented with FOS (FOS ( − )(+)); (), offspring fed a post-weaning diet without FOS, dams supplemented with FOS (FOS (+)( − )); (□), offspring fed a post-weaning diet supplemented with FOS, dams supplemented with FOS (FOS (+)(+)). The levels for FOS ( − )( − ) are expressed relative to the control values, which are taken as 1. Values are means, with standard errors represented by vertical bars. For IL-6, P values were P = 0·1251, P = 0·7063 and P = 0·5766 for FOS in dams, FOS in offspring and their interaction, respectively (two-way ANOVA). Similarly, P values for IL-8 were P = 0·0882, P = 0·7435 and P = 0·2326 for FOS in dams, FOS in offspring and their interaction, respectively. P values for TARC were P = 0·2636, P = 0·6687 and P = 0·1408 for FOS in dams, FOS in offspring and their interaction, respectively. P values for TNF-α were P = 0·0275, P = 0·9225 and P = 0·9813 for FOS in dams, FOS in offspring and their interaction, respectively. a,b Mean values with unlike letters were significantly different (P < 0·05; Tukey–Kramer test).