Microbial composition and fecal fermentation end products from colicky infants – a probiotic supplementation pilot

Abstract

Objectives: The etiology of infantile colic has remained unknown. The present study aimed to identify possible differences in the fecal microbial composition and metabolic end products between colicky and non-colicky infants and to examine whether an orally administered probiotic supplementation has any effect on the microbiota and relief in colicky symptoms. Subjects and methods: The study population consisted of nine colicky and nine non-colicky infants. Five colicky infants received probiotic supplementation (Lactobacillus rhamnosus GG, Lactobacillus rhamnosus LC705, Bifidobacterium breve Bbi99, and Propionibacterium freudenreichii ssp. shermanii JS) and four colicky infants received placebo for 2 weeks. Fecal microbiota of all infants was studied by culture, cellular fatty acid (CFA) analysis, and gas and short chain fatty acid (SCFA) production in 48 h fermentation. Microbial parameters and crying profiles were compared between the cases and controls at baseline and between the probiotic and placebo supplementation after 2 weeks. Results: Although SCFA and gas production and bacterial total counts were similar, the prevalence of indole-producing coliforms was significantly higher in colicky infants compared with controls (89% vs 33%), while many aerobic genera present in controls were not detected in colicky infants. CFA composition reflected the differences, since >2% of the CFAs were present exclusively in the colicky group. After probiotic supplementation, the total counts of anaerobic bacteria, especially lactobacilli and bifidobacteria, increased significantly. Fermentation parameters were not extensively affected; however, acetic and lactic acid production tended to increase and hydrogen production tended to decrease. No significant difference was observed in crying patterns between the probiotic and placebo group. Conclusion: The composition of the intestinal microbiota differs between colicky and non-colicky infants. Although no apparent relief in colicky symptoms was achieved, the probiotic supplementation seems to increase the bacterial diversity and strengthen the succession towards a balanced commensal gut microbiota.

• infantile colic
• fecal bacteria
• probiotics
• colonization
• SCFA

Introduction

Infantile colic is a common problem; the prevalence rate varies between 9% and 17%, depending on the criteria and methodology used [1]. The etiology of colic has remained unknown despite the fact that a wide range of behavioral, gastrointestinal, and functional factors have been connected with it [2]. The role of the composition and activities of the intestinal microbiota has not been fully determined yet. No single bacterial by-product, pathogen, or species has been found to be the immediate cause of infantile colic.

The commensal microbiota of neonates develops gradually; in general, the very first colonizers are facultative enterobacteria and enterococci, followed by bifidobacteria and other anaerobic species [3]. A number of factors, such as the mode of delivery, microbes in the immediate environment, and form of feeding, influence the composition of the early intestinal microbiota [4], [5]. Birth by cesarean section delays infants’ colonization by lactobacilli, bifidobacteria, and the B. fragilis group organisms early in life [4], [5].

In addition to intrinsic factors, the intestinal microbiota can be modified with bacterial therapy. Probiotic strains of Lactobacillus, Bifidobacterium, and Propionibacterium are known to provide several beneficial effects on the host [6], [7]. Long-term consumption of infant formula containing Bifidobacterium lactis and Streptococcus thermophilus was well tolerated and reduced colic and irritability as well as antibiotic usage in infants aged 3–24 months [8]. A significantly greater reduction in colicky cry was reported with Lactobacillus reuteri compared with simethicone [9]. A newborn baby is, at least in theory, an ideal object for bacterial therapy because his/her indigenous microbiota is still immature and, therefore, may be easily manipulated [10].

Fermentation capacity measured as production of gas and short chain fatty acids (SCFAs) has been used to evaluate the microecology of the gut and functionality of the microbiota [11], [12]. These parameters and cellular fatty acid (CFA) composition [13] also reflect the composition of the microbiota. Bacterial metabolites are used as energy source by the host and other bacteria [6]. On the other hand, when accumulated some end products of especially protein breakdown and dissimilatory amino acid metabolism (indoles, amines, phenols) are bioactive and may become harmful [11]. If the basis of colic lies in the disturbance of the intestinal microbiota, probiotic supplementation might be a proper means to balance the microbial composition, metabolites, and activities in the gut.

The aims of the present study were to determine whether the composition of the fecal microbiota, studied by culture and culture-independent CFA analysis, or microbial metabolites (SCFAs, gases) differ between colicky and non-colicky infants aged 1–6 weeks and whether an orally administered probiotic supplementation (Lactobacillus rhamnosus GG, L. rhamnosus LC705, Bifidobacterium breve Bbi99, and Propionibacterium freudenreichii ssp. shermanii JS) has any effect on the monitored symptoms or microbial parameters. In this pilot study a wide variety of laborious methodology was used for a relatively small but carefully chosen group of infants.

Subjects and methods

Participants and study design

Infants in the Helsinki area were recruited in maternity clinics, maternity hospitals, and child health care centers. Their mothers gave a written informed consent, and the City of Helsinki Ethical Committee of Health Care approved the protocol. The mothers were interviewed and instructed on keeping a daily diary of sleeping, eating, and crying habits of the baby, including the type (mild, moderate, severe) and duration of crying. The inclusion criteria for infants were: age 3–6 weeks and colicky symptoms as described by Wessel et al. [14], i.e. 3 h of crying per day for 3 days per week for 3 weeks. The exclusion criteria were: congenital malformations or abnormalities, bowel diseases, prematurity (<38 gestational weeks or birth weight <2500 g), antimicrobial therapy during 2 preceding weeks, flatulence associated with feeding or consumption of other probiotic-containing products. Altogether 18 breast-fed infants were recruited in the case group and 9 infants in the control group. However, after the first study week (week 0) nine infants were excluded from the case group because the amount of crying did not strictly fulfill the inclusion criteria. Demographic characteristics of the 18 infants (9 cases and 9 controls) who fulfilled the inclusion criteria are shown in Table I.

Table I.  Demographic characteristics of the infants at enrolment.

Parameter	Colicky	Control
Gender
Males	3	3
Females	6	6
Age (weeks)
Mean	3	4
Range	2–5	1–6
Mode of delivery
Vaginal	8	6
Cesarean	1	3
Birth weight (kg)
Mean	4.2	4.2
Range	3.3–5.0	3.7–4.6
Type of feeding
Breast-fed	9	8
Formula-fed	0	1*
Colicky cry (h/week)
Mean	16	−†
Range	9–28	−†

*Fed with both breast milk and formula.

†Sporadic, infrequent.

Capsules each containing a mixture of probiotic bacteria (L. rhamnosus GG, ATCC 53103 (LGG), 5×10⁹ cfu; L. rhamnosus LC705, 5×10⁹ cfu; B. breve Bbi99, 2×10⁸ cfu; and P. freudenreichii ssp. shermanii JS, 2×10⁹ cfu), and microcrystalline cellulose as a filling agent (Valio Ltd, Helsinki, Finland) were given to five of nine cases, whereas the remaining four infants received placebo capsules (microcrystalline cellulose) once a day for 2 weeks in a randomized, double-blind manner. The contents of the capsules were suspended in water or breast milk by the mothers.

Microbiology

Fecal samples were collected at baseline and after 2 weeks of the probiotic or placebo supplementation. The samples were transported anaerobically in containers sealed in gas-impermeable plastic pouches (AnaerocultA, Merck, Darmstadt, Germany) within 4 h of defecation and processed immediately. After homogenization the pH of the feces was measured (Benchtop 420 pH Meter, Orion Inc., Beverly, ME, USA), and the homogenate was serially diluted (10⁻¹−10⁻⁷) in prereduced PY (peptone-yeast extract) broth (pH 7.0). Aliquots of 10 µl or 100 µl of the homogenates or appropriate dilutions were plated onto several non-selective and selective agar media and incubated at 36°C as follows: blood (5% sheep blood) agar for total aerobes in 5% CO₂ atmosphere for 2–4 days; cystine lactose electrolyte-deficient (CLED) agar for coliforms, bile esculin (BE) agar for enterococci, and Sabouraud agar with chloramphenicol for yeasts in ambient air for 2–4 days; brucella agar supplemented with hemin and vitamin K₁ for total anaerobes, bacteroides bile esculin (BBE) agar for bile-resistant Bacteroides spp., tomato juice-based agar supplemented with hemin and maltose for bifidobacteria, deMann Rogosa Sharpe (MRS) agar for lactobacilli, cycloserine cefoxitin fructose egg yolk (CCFA) agar for Clostridium difficile, and neomycin egg-yolk (NEYA) agar for clostridia in anaerobic jars filled with gas mixture (90% N₂, 5% CO₂, 5% H₂) for 5–7 days.

The total counts and main groups of anaerobic and aerobic bacteria and yeasts were enumerated from applicable plates (detection limit 10² cfu/g wet weight). Different colony morphotypes were isolated under a stereomicroscope and identified using established methods, including aerotolerance testing, Gram stain, various biochemical tests, metabolic end product analysis by gas-liquid chromatography, and enzyme profiling [15], [16]. Coliforms were further identified with Api 20E and lactobacilli with Api 50CHL (bioMérieux, Marcy l'Etoile, France). The presence of C. difficile toxins A and B in feces was determined with commercial kits (Premier Toxins A&B, Meridian Bioscience Inc., Cincinnati, OH, USA, and C. difficile toxin A test, Oxoid, Hampshire, England) according to the manufacturers’ instructions. The species and clonal identity of LGG isolates were confirmed by arbitrarily primed PCR [17]. Other probiotics were identified by conventional methods including metabolic end product analysis.

Cellular fatty acids (CFAs)

For the analysis, 100 mg of feces was suspended in 5 ml PY (peptone-yeast extract) broth, allowed to sediment at 4°C for 2 h, remixed, and allowed to sediment for 15 min. Supernatant was removed and centrifuged for 15 min (1000 g). Fatty acids were extracted from the pellets as described in the operating manual of the Microbial Identification System (MIS) software package by MIDI (Microbial ID, Inc., Newark, DE, USA). The procedure includes saponification (cell lysis), methylation (formation of methyl esters to increase volatility for GC), extraction, and wash. CFA analysis was performed with an HP 5890 series II gas chromatograph equipped with an HP U2 cross-linked 5% phenyl methyl silicone fused capillary column with a flame ionization detector (Hewlett-Packard, Palo Alto, CA, USA). Hydrogen was used as carrier gas. Injection temperature was 250°C, that of detector 300°C, and a cycle of 170°C, 260°C, and 310°C for the oven.

Fermentation

Fermentations were performed in triplicate in 20 ml glass head-space vials containing 10 ml of either PY broth or PY with 1% glucose (PYG) inoculated in an anaerobic chamber with fresh fecal homogenate (final concentration 0.2% w/v) and sealed with rubber stoppers and aluminium seals. The vials were incubated at 36°C in a shaker (100 rpm/min) and after 48 h transferred to 4°C until analyzed within few hours.

Percentage of carbon dioxide (CO₂), hydrogen (H₂), methane (CH₄), and air of the bottles head-space gas volume were measured at 0 h and 48 h by HP gas chromatograph model 5890 equipped with Porapak N column and molecular sieve (Hewlett-Packard). The amount of oxygen (air) was monitored to control the atmospheric conditions of fermentation. Before the measurement, the syringe, valve, and needle were flushed with helium to remove air. Helium was used as the carrier gas (total flow 20 ml/min). The temperature of the injector was 150°C, that of oven 45°C, and detector 200°C. The baseline (0 h) values were subtracted from the 48 h values. As only traces or no methanethiol (CH₃SH) or hydrogen sulfide (H₂S) were expected from breast-fed infants [18] they were not monitored.

Both volatile and non-volatile metabolic fatty acids were analyzed. Before analysis, the fermentation suspension was acidified with sulfuric acid and centrifuged (Labofuge GL, Heraeus, Hannover, Germany) for 30 min (5000 g). SCFAs were extracted with methyl tert-butyl ether (16 Jousimies-Somer) and analyzed with an HP Agilent Plus GC 6890 equipped with an HP Innowax column (Hewlett-Packard). The carrier gas was hydrogen (constant flow 15 ml/min). Injection and FID-detector temperatures were 250°C and oven temperature was 200°C (initial 2 min 100°C).

The gas production of selected coliform isolates from baseline samples was analyzed using pure cultures by adding 200 µl bacterial suspension (1:10 dilution of Mcfarland 4 in NaCl) similarly in head-space vials containing 10 ml PY broth with 5% lactose (PYL). Only the total volume (ml) of gas produced after 5 h incubation at 36°C in a shaker (100 rpm/min) was measured using a gas-tight syringe and allowing the piston to be driven up by the pressure generated in the vial during fermentation.

Statistics

Statistical analyses were performed using SPSS14 (SPSS, Chicago, IL, USA) and Epi Info 6 Statcalc (CDC, Atlanta, GA, USA) software for Windows. Related samples t test and single table test were used to compare the concentrations and prevalence between the groups, and paired samples t test for comparison before and after the probiotic or placebo supplementation. A p value <0.05 was considered statistically significant.

Results

Microbial findings in colicky and non-colicky infants at baseline

The number of different isolates per sample was similar between the colicky and control infants (9.3 vs 9.7) and the fecal pH values were also similar (5.5 vs 5.3). Table II presents the prevalence and concentration of microbial taxa detected in the cases and controls. The predominant anaerobic findings were bifidobacteria followed by the Bacteroides fragilis group organisms and lactobacilli, and predominant aerobes were indole-positive coliforms, coagulase-negative staphylococci, and enterococci. The prevalence of indole-positive coliforms (Escherichia coli and Klebsiella oxytoca) was significantly higher in the colicky infants than in the controls: 89% (8/9) vs 33% (3/9), p = 0.016. The total load measured as the proportion of indole-positive coliforms of the total count varied but was slightly higher in colicky than control infants: 8.7% (SD 16.9) vs 2.3% (SD 6.2) (NS). When grouped together, gram-negative anaerobes covered a slightly higher proportion of the microbiota in the colicky group than in controls: 19.4% (SD 13.3) vs 10.9% (SD 15.7) of the total bacterial count (NS). In contrast, the colicky infants lacked several aerobic genera; Streptococcus, Micrococcus, Corynebacterium, Bacillus spp. and yeasts were cultivable, although in low prevalence, from the controls only. The ratio anaerobes:aerobes varied drastically between infants in both groups (range 0.7–2000) with similar median values, 12.2 in the colicky and 14.3 in the control group. The ratio lactic acid bacteria:total bacterial count, with lactic acid bacteria including lactobacilli, bifidobacteria, streptococci, and enterococci, varied less and was similar in the colicky and in the control group (0.41 vs 0.53, NS). Also the species distribution of lactobacilli was similar between the groups with Lactobacillus paracasei ssp. paracasei and L. rhamnosus dominating in both groups. Only two infants were colonized with more than one Lactobacillus species, although several infants harbored two subtypes of the same species with clearly different colony morphologies. One control infant carried a strain identical to LGG (log₁₀ 9.2 cfu/g feces). All fecal samples were negative for C. difficile toxins A and B and no C. difficile was isolated at baseline. None of the four infants delivered by cesarean section carried E. coli, and the B. fragilis group organisms and bifidobacteria were absent from three infants and lactobacilli from two infants.

Table II.  Microbial findings (log cfu/g wet weight) in nine colicky and nine control infants at baseline.

	Control			Colicky
Parameter	n	Mean	Range	n	Mean	Range
Total anaerobic growth	9	10.2	8.5–10.7	9	10.3	8.8–10.7
Total lactobacilli	6	9.0	5.0–9.5	7	9.5	3.4–10.3
Non-LGG lactobacilli	6	8.9	5.0–9.5	7	9.5	3.4–10.3
L. rhamnosus GG	1	8.2	0.0–9.2	0
Bifidobacteria	8	10.0	8.3–10.7	8	9.9	8.6–10.5
Clostridium spp.	2	7.8	8.5	4	7.5	6.5–8.3
Gram-positive rods†	4	8.8	6.0–9.6	2	8.8	8.0–9.8
Bacteroides fragilis group	6	9.3	5.3–10.0	7	9.6	5.0–10.3
Gram-negative rods‡	1	3.0	0.0–4.0	2	7.3	3.0–8.3
Gram-positive cocci	3	8.6	8.3–9.5	1	7.0	0.0.–8.0
Gram-negative cocci	1	8.1	0.0–9.0	3	9.0	8.0–10.0
Total aerobic growth	9	9.1	8.0–9.4	9	9.5	7.0–10.1
Coliform, indole-positive	3	9.0	8.0–9.4	8*	9.3	4.0–10.0
Coliform, indole-negative	2	7.9	2.0–8.9	1	7.5	0.0–6.6
Pseudomonas spp.	3	2.6	2.7–3.5	4	5.6	2.2–6.5
Staphylococcus aureus	5	6.5	2.6–7.3	4	5.9	1.3–6.8
Coagulase-negative staphylococci	7	8.0	6.7–8.5	9	8.2	5.3–8.7
Enterococci	5	7.5	7.0–8.0	2	8.1	8.7–8.8
Streptococci	2	7.2	7.8–8.0	0
Micrococcus spp.	1	5.0	0.0–6.0	0
Corynebacteria	3	8.0	4.0–8.9	0
Bacillus spp.	3	8.5	4.0–9.3	0
Yeasts	2	2.3	2.0–3.2	0

*p < 0.05.

†Includes Actinomyces spp., Eubacterium-like organisms, and Propionibacterium spp.

‡Includes Bilophila spp., Fusobacterium spp., and Prevotella spp.

Microbe-derived fecal cellular fatty acids (CFAs)

A total of 45 CFAs were identified and their percentage of the total fatty acid content was calculated (percentage of the total peak area in chromatogram). The four most abundant fatty acids – 16:0 (palmitic), 18:0 (stearic), 14:0 (myristic), and 18:1 cis 9 – comprised 75% (each 4–34%) of the total CFA content. Altogether 44 CFAs (mean 20 acids per infant) were detected among colicky infants and 34 CFAs (mean 21) among control infants. Acids comprising 0.5% or more are presented in Figure 1. Of the individual fatty acids, significant differences between colicky and control infants were found in the prevalence of saturated 17:0 and hydroxy 18:0 12OH, both being more common among controls. Grouped together, unsaturated and branched fatty acids were slightly more prevalent in colicky infants than controls: unsaturated 20.3% (SD 4.6) vs 12.5% (SD 3.1) of the total CFAs, p = 0.049 and branched 7.1% (SD 1.4) vs 2.2% (SD 0.4), p = 0.048, respectively, while saturated fatty acids and dimethyl acetals were equally common. The 11 fatty acids that were present only among colicky infants were dispersed among 6 infants and included cyclic, straight, branched, saturated and unsaturated acids, aldehydes, and dimethyl acetals, with varying carbon chain length. Their proportions of the total fatty acids remained low (mean 0.19%, range 0.01–0.70) as well as the number of infants carrying them (mean 1 infant, range 1–3). When grouped together they comprised 2.3% of the total CFAs.

Figure 1.  Proportions of bacterial cellular fatty acids (CFAs) of total fatty acids (mean area percentage and SD) and number of infants carrying them among nine colicky and nine control infants at baseline. Only CFAs comprising at least 0.5% of the total CFAs are presented. *Difference between groups was significant (p < 0.05). Dma, dimethyl acetal.

Short chain fatty acids (SCFAs) and gas as metabolic fermentation end products

A total of nine SCFAs were identified and their concentrations measured. The individual SCFA concentration varied considerably (between 14.5 and 928.6 mmol/l) and the number of different fatty acids (variety) between three and seven. Variation between the three parallel (triplicate) samples was subtle. The profile of the fermentation products differed as expected between carbohydrate-rich PYG and peptone-rich PY media (lacking fermentable carbohydrate) (Figure 2). Despite individual variation the overall distribution and mean concentrations of SCFAs were similar in colicky and control infants in both fermentation media.

Figure 2.  Fecal volatile and nonvolatile short chain fatty acids (SCFAs) (mean mmol/l and SD) and hydrogen and carbon dioxide gases (percentage of total gas volume, SD) produced at baseline in 48 h fermentation in peptone-yeast extract broth (PY) and peptone-yeast extract-glucose broth (PYG) by nine colicky and nine control infants. Also minor amounts (1–5 mmol/l) of isovaleric, isobutyric, isocapronic, and phenylacetic acids were produced from PY and pronionic acid from PYG (not shown). No capronic or valeric acids were detected. *Difference between the groups is in the limit of being significant, p = 0.051.

Variation in the fecal gas production between the study subjects was also substantial; however, no significant differences between the groups were found. CO₂ had systematically the highest percentages (percentage of the bottle's head-space gas volume) (see Table IV). In PY the relation CO₂:H₂ was slightly smaller in the colicky (3.3) than in the control group (5.0); however, no significant difference between the groups was seen in the proportion of CO₂ (11.6% vs 10.0%, NS), while there was a trend of a higher proportion of H₂ in the colicky group compared with the control group (3.5% vs 1.9%, p = 0.051). In PYG fermentation no trend was seen. The highest levels of H₂ were measured among infants harboring coliforms, especially Klebsiella and/or Clostridium perfringens. Infants harboring lactobacilli produced slightly less H₂ in PYG (2.6% vs 8.8%, NS). No methane was detected.

Gas production by coliform isolates

Variation in volumes of the total gas production between the isolates after 5 h in pure culture was dependent on the species, being either zero (Enterobacter spp., Serratia spp.) or ranging between 7.1 and 12.0 ml (E. coli, K. oxytoca, K. terrigena). There was no difference in gas volumes between the isolates from the colic and control group or between the most prevalent coliforms (E. coli, K. oxytoca, and K. terrigena). Altogether 15 coliform strains originating from baseline samples were analyzed.

Microbial findings in colicky infants after the probiotic or placebo supplementation

After the probiotic supplementation, the total counts of anaerobic bacteria increased significantly compared with the baseline, whereas after the placebo supplementation the total counts of both anaerobic and aerobic bacteria were similar to those at baseline (Table III). This was due to a considerable increase of bifidobacteria and LGG. The number of different isolates per sample increased slightly with both supplementations but tended to be higher after the probiotic than after the placebo supplementation (seven vs six anaerobic findings, and five vs four aerobic findings per sample, respectively). The infants randomized for the placebo supplementation happened to have higher numbers of coliforms (the most numerous aerobic finding) at baseline than did the infants randomized for the probiotic supplementation. The relative proportion of coliforms of the total microbial concentration in infants receiving probiotics was 2.4% (range 0–5.2%) at baseline and 1.2% (range 0.05–3.3%) after the probiotic supplementation and in infants receiving placebo it was 18% (range 0–51%) at baseline and 13% (range 0.3–25%) post supplementation.

Table III.  Microbial key findings (log cfu/g wet weight) in colicky infants before and after supplementation with the probiotic (five infants) or placebo (four infants).

	Probiotic supplementation						Placebo supplementation
	Before			After			Before			After
Bacteria	n	Mean	Range	n	Mean	Range	n	Mean	Range	n	Mean	Range
Total anaerobes	5	10.3	8.8–10.6	5	10.7*	10.2–10.8	4	10.4	9.9–10.7	4	10.5	9.8–10.9
Bifidobacteria	5	9.9	8.6–10.5	5	10.4*	10.0–10.5	3	9.8	9.8–10.0	3	10.2	8.8–10.7
L. rhamnosus GG	0	0		5	8.6*	7.5–8.9	0	0		1	3.6	0.0–4.2
Non-LGG lactobacilli	4	9.0	3.4–9.7	3	9.6	2.0–10.3	4	9.7	8.6–10.3	3	8.4	3.1–9.0
Total lactobacilli	4	9.0	3.4–9.7	5	9.6	7.5–10.3	4	9.7	8.6–10.3	4	8.4	3.1–9.0
Enterococci	0	0		4	7.9*	5.0–8.5	2	8.5	8.7–8.8	3	8.2	7.9–8.5
Total aerobes	5	9.5	7.7–10.0	5	9.7	8.5–10.2	4	9.5	7.0–10.1	4	9.3	8.3–9.7

*p<0.05.

After the probiotic supplementation, LGG appeared in all five infants in high concentrations (mean log₁₀ 8.6 cfu/g), and the counts of bifidobacteria increased significantly (Table III). Both lactobacilli and bifidobacteria counts exceeded the corresponding counts in the placebo group. L. rhamnosus LC705 was isolated in one (log₁₀ 8.0 cfu/g) but Propionibacterium JS in none of the infants. In addition, the prevalence of enterococci increased significantly. In general, the prevalence and mean concentration of anaerobic bacteria increased, with the exception of gram-negative cocci. Aerobic bacteria either increased or decreased, depending on the species and infant. The proportion of lactic acid bacteria of the total bacterial count increased after probiotic supplementation from 47% to 65%.

After the placebo supplementation, the counts of lactobacilli and bifidobacteria were similar to those at baseline (Table III); similarly the proportion of lactic acid bacteria of the total bacterial count was unaffected by placebo supplementation (34% before and 36% after). The only statistically significant change in the placebo group was the decrease in the prevalence and mean concentration of coagulase-negative staphylococci. A strain identical to LGG was isolated from one infant (log₁₀ 4.2 cfu/g) after the placebo supplementation.

All fecal samples were negative for C. difficile toxins A and B. C. difficile was found (log₁₀ 3 cfu/g) in one infant as a new finding after the probiotic supplementation. In pure culture, the strain produced toxin A but the carrier infant had no symptoms other than colicky cry.

After the probiotic or placebo supplementation, there was no difference in fecal pH values (5.7 vs 5.3) between the groups. No side effects related to the supplementation were recorded.

SCFAs and gas production after supplementation

Probiotic supplementation increased the production of acetic acid from PYG markedly in four of five infants and the production of lactic acid in three of five infants (Table IV). Acetic acid production from PY increased markedly in two infants in the placebo group. Otherwise SCFA concentrations remained fairly similar to the baseline.

Table IV.  Fecal volatile and nonvolatile short chain fatty acids (SCFA mmol/l) and hydrogen and carbon dioxide gases (% of gas volume) produced in 48 h fermentation in peptone-yeast extract broth (PY) and peptone-yeast extract-glucose broth (PYG) before and after probiotic (five infants) or placebo (four infants) supplementation.

	Probiotic supplementation						Placebo supplementation
	Before			After			Before			After
Products	n	Mean	Range	n	Mean	Range	n	Mean	Range	n	Mean	Range
PY
Acetic	5	119	33–232	5	97	15–231	4	63	11–127	4	160	102–201
Propionic	5	35	22–66	5	30	8.0–74	4	18	3.6–47	4	31	19–52
Isobutyric	1	0.3	0–1.4	1	0.7	0–3.4	1	1.5	0–6.1	2	0.4	0–1.6
Butyric	4	13	0–33	5	13	0.8–32	3	7.7	0–21	4	24	0–42
Isovaleric	2	0.7	0–2.4	2	1.4	0–7.0	2	1.2	0–3.5	1	1.3	0–5.2
Valeric	0	0		0	0		1	<0.1	0–0.01	0	0
Isocapronic	0	0		0	0		1	1.5	0–5.9	1	1.5	0–5.9
Lactic	1	0.9	0–4.4	0	0		0	0		0	0
Succinic	2	3.3	0–8.9	0	0		2	3.3	0–6.9	2	4.5	0–7.9
Phenylacetic	0	0		1	0.8	0–3.9	1	0.5	0–1.8	1	0.5	0–1.8
H2	5	3.5	0.4–5.6	3	0.6	0–4.5	4	4.3	2.1–6.2	4	3.4	3.4–4.6
CO2	5	11	10–12	5	11	9.1–13	4	12	9.0–14	4	9.7	8.0–14
PYG
Acetic	5	148	6.6–462	5	436	53–962	4	74	17–206	4	80	19–181
Butyric	1	1.6	0–8.1	1	3.6	0–18	0	0		1	23	0–90
Lactic	5	159	89–247	5	196	185–211	4	159	73–278	4	157	99–233
Succinic	4	1.4	0–3.2	5	4.8	1.0–15	4	2.5	0.8–4.8	4	8.6	0.1–3.3
H2	5	1.5	0.2–5.0	5	1.8	0.2–4.2	4	5.6	2.8–9.3	4	4.6	1.2–7.5
CO2	5	5.6	2.5–8.6	5	5.3	1.1–11	4	10.9	6.8–15	4	9.5	6.3–13

Note that statistics were not performed.

In both supplementation groups the amounts of CO₂ were unaffected and it remained the major fermentation gas. After probiotic supplementation fecal H₂ production from PY decreased in four of five infants and remained the same in one infant, while in the placebo group and in PYG no trends were seen (Table IV). In PY the mean proportion of H₂ was 0.6% after the probiotic and 3.4% after the placebo supplementation (percentage of the bottle's head-space gas volume), and there was a 10-fold difference in the relation CO₂:H₂ (20.7 vs 2.9, respectively) between the supplementations.

Cellular fatty acids (CFAs) after supplementation

Altogether, 33 fatty acids were detected post supplementation, of which 30 were detected after the probiotic and 28 after the placebo supplementation. Probiotic supplementation had no systematic effect on the major fatty acids. Some minor acids were replaced individually with both supplementations.

Crying patterns

The crying patterns of the colicky infants during the week preceding the baseline sample collection and those during the second week of the probiotic or placebo supplementation were compared. At baseline, the total crying time turned out to be longer in the placebo group (Table V). The mean duration of total crying time per week decreased markedly after both supplementations, as did the colicky cry, including moderate and severe crying. No statistical differences in the crying patterns between the probiotic and placebo group were found. The mode or amount of crying could not be linked to bacterial findings.

Table V.  Mean crying times (SD) (h/week) preceding the baseline (week 0) and at the second week of the probiotic or placebo supplementation (week 2).

	Week 0		Week 2		Difference
Type of crying	Probiotic	Placebo	Probiotic	Placebo	Probiotic	Placebo
Moderate/severe	16.1 (2.8)	16.2 (7.3)	9.6 (6.5)	6.8 (3.7)	−6.5 (40%)	−9.4 (58%)
Mild	11.6 (6.6)	22.0 (10.9)	13.7 (5.3)	23.9 (16.3)	2.1 (19%)	1.9 (9%)
Total	27.7 (7.9)	38.3 (13.9)	23.4 (4.8)	30.7 (19.2)	−4.3 (16%)	−7.6 (20%)

Discussion

In the present study wide-ranging methods were chosen to gain an extensive description of the gut microbiota as a functional organ, as well as to give an insight to the associations between the microbiota, microbial by-products, and colicky symptoms. Indeed, colicky infants had a significantly higher prevalence of indole-producing coliforms in their feces than did their controls. In contrast, various aerobic genera were absent from colicky infants. Minor differences were recorded also among fecal cellular fatty acids, as over 2% of the CFAs in the colicky group were absent from the control group. CFA composition shifts and reflects all bacteria, including those that are nonviable and uncultivable. In addition to the bacterial composition, the patterns of bacterial fermentation products are affected by growth conditions. Still, no consistent difference in the major fermentation end products was detected. Moore et al. [19] speculated that the immaturity of the gut might affect motility, absorption, fermentation conditions, and handling of colonic gas or other bioactive compounds. Our findings suggest that colic is not connected to bacterial growth or fermentation conditions but more likely to factors influencing microbial colonization patterns and bacterial composition and/or factors affecting the host reactions to bacteria and their metabolites.

We found no previous reports on elevated levels of coliforms in colicky infants; however, coliform-derived indole, having physiologic effects in mammals, may have relevance. Higher urinary levels of 5-OH indole acetic acid, a serotonin metabolite, have been reported in colicky than control infants [20]. Fecal indole levels were not monitored in the present study. We tested the total gas production capacity of the coliform isolates but found no difference between isolates from the colicky and control groups.

Previously, Lehtonen et al. [21] reported C. difficile more frequently in stool samples from colicky infants than their age-matched controls and, in addition, differences in saturated 15:0 and branched 17:0 iso and 17:0 anteiso fatty acids between infants with severe colic and controls. We found C. difficile in one colicky infant and differences between the groups in hydroxy acid 18:0 12OH and branched acids grouped together, but unlike Lehtonen et al. [21], we found branched CFAs more often in the colicky group. Palmitic and stearic fatty acids, which are also the major CFAs in most microbial cell walls, predominated. Hydroxy acids typical of gram-negative bacteria were more prevalent in the control group, contradicting the proposal that gram-negative bacteria made up a slightly higher proportion of the total bacterial count in the colicky group. The amount of dimethyl acetals and aldehydes common in anaerobic bacteria remained relatively low. This observation agrees with the culture results showing only mild dominance of anaerobes. Generally anaerobes are 100–1000-fold more numerous than aerobes, while in these infants the relation was 1:13. Our culture results agree with Savino et al. [22] in reporting higher yield of gram-negative anaerobes in colicky than control infants but disagree with the reported decreased lactobacilli. Later Savino et al. reported an altered Lactobacillus species composition in colicky infants but this time no difference in lactobacillar counts [23]. Although no difference was seen among Lactobacillus species or counts in the present study, the proportion of lactic acid bacteria in the total bacterial count tended to be lower in the colicky group. The colonization patterns of lactobacilli in our study are similar to those reported in healthy infants using PCR-DGGE, with some infants already carrying several lactobacilli at 2 months [24], [25].

In our study the ranking order of fecal SCFAs produced was fairly uniform but the individual concentrations differed significantly. All infants produced detectable amounts of acetic and propionic acid and often also butyric acid. Samples yielding high butyric acid concentration yielded butyrate-producing genera such as Clostridium or Porphyromonas. Fecal fermentation in carbohydrate-poor PY media, chosen to resemble the substrate-poor distal colon, yielded SCFAs in a molar ratio 73:18:8 for acetate:propionate:butyrate. Corresponding colonic values for adults are approximately 60:20:18 [26].

Breath hydrogen production has been reported to be significantly more frequent by colicky than non-colicky infants [19]. In contrast, Belson et al. [27] failed to show any association between hydrogen concentration and colic but instead an association between low levels of methane at 6 months of age and colic history. We detected no significant difference in hydrogen production and no methane in either the colic or control group at <2 months of age. This may be reasonable, as hydrogen production begins immediately after intestinal colonization, while methane production is individual and more age-related [27]. Intestinal gases are produced but simultaneously also consumed by numerous colonic bacteria [18]. Like fecal culture, in vitro fermentation may be a reflection rather than an exact description of the microbiota and its activities in the upper gut. However, due to practical problems in direct examination, fermentation offers a feasible approach to studying the intestinal microbiota.

Probiotic supplementation aims to balance the gut microbiota and its function and to ease symptoms originating from the gut. Probiotics are generally well tolerated, as was also the case in the present study, where none of the infants had diarrhea or other side effects during the period of supplementation. Supplemented lactobacilli and bifidobacteria effectively colonized the infant's gut, being isolated in high concentrations from all infants in the probiotic group. B. breve is known to colonize infants effectively [28]. The ubiquitous colonization of LGG demonstrated in the present study was higher than previously reported: 67% in healthy newborns [10], 90% in premature infants receiving antibiotics [29], 47% in infants with low birth-weight [30], and 83% in healthy newborns delivered vaginally or by cesarean section whose mothers were colonized with LGG during pregnancy [31]. Restrictions in the methodology and sample material probably led to lower recovery rate of Propionibacterium JS and Lactobacillus LC705 than expected, since fecal bacterial composition differs markedly from that of the upper gut where many probiotics are aimed to act. In general, no major changes or effects on colonization or growth patterns of other bacteria have been reported after probiotic supplementation [29–31]. In our study, there was a tendency towards the dominance of anaerobes and vast species diversity in the probiotic group. Probiotics influenced the fermentation patterns by increasing acetic and lactic acid production in glucose-rich PYG media and by decreasing the release of hydrogen in glucose-poor PY, even though the effect was mild. Changes in individual infants may also reflect the natural succession progress of the immature microbiota as seen in the placebo group. Hydrogen release in PYG tended to be lower at baseline in infants harboring lactobacilli. In theory, proliferation of anaerobic lactic acid bacteria accompanied by increased acid production may strengthen the development of colonization resistance upheld by commensal microbiota.

In the present study, the counts of total anaerobes tended to increase slightly in infants receiving the placebo supplementation, but there were no clear trends at the species level. Despite the strict exclusion criteria, two infants not receiving probiotics harbored an isolate similar to LGG (one in the control group at baseline and one in the placebo group after supplementation). LGG, which was originally isolated from the human microbiota, may have been residing in the microbiota of their mothers, since various LGG-containing products are on the market and widely used in Finland. These two cases may be explained by salivary transmission between the mother and her child.

The overall changes in crying patterns between the probiotic and placebo group were comparable, and the infants thrived well with both supplementations. Colicky cry decreased dramatically in both groups, as colic normally does by the age of 4 months. In the present study, mean total crying time and colicky cry corresponded well with those presented in previous studies [9], [32]. At baseline, the total crying time happened to be longer among the infants receiving placebo; thus, a more marked reduction was expected and recorded in that group. Consistently, the total crying time remained longer in the placebo group. As to colicky cry, however, the groups did not differ markedly from each other at any time of measurement.

Conclusions

A shift towards less lactic acid bacteria and more indole-producing coliforms differentiated colicky infants from non-colicky infants. To thoroughly assess the observed differences in coliform populations, indole production, CFA profiles and fermentation parameters, larger study groups, and selected molecular technology are required. Succession towards a balanced commensal microbiota, i.e. increased counts of anaerobic bacteria, bifidobacteria, and lactobacilli, and towards lactic acid metabolism and improved gas utilization, may be achieved by probiotic supplementation.

Acknowledgements

The authors thank Maija Saxelin PhD for constructive comments on the manuscript, and Soile Tynkkynen PhD for her help on identification of LGG isolates. The late Professor Hannele Jousimies-Somer is acknowledged for her contributions to the study.