Influence of microbial species on small intestinal myoelectric activity and transit in germ-free rats

Einar Husebye, Per M. Hellström, Frank Sundler, Jie Chen, Tore Midtvedt


The effect of an intestinal microflora consisting of selected microbial species on myoelectric activity of small intestine was studied using germ-free rat models, with recording before and after specific intestinal colonization, in the unanesthetized state. Intestinal transit, neuropeptides in blood (RIA), and neuromessengers in the intestinal wall were determined.Clostridium tabificum vp 04 promoted regular spike burst activity, shown by a reduction of the migrating myoelectric complex (MMC) period from 30.5 ± 3.9 min in the germ-free state to 21.2 ± 0.14 min (P < 0.01). Lactobacillus acidophilus A10 and Bifidobacterium bifidum B11 reduced the MMC period from 27.9 ± 4.5 to 21.5 ± 2.1 min (P < 0.02) and accelerated small intestinal transit (P < 0.05). Micrococcus luteus showed an inhibitory effect, with an MMC period of 35.9 ± 9.3 min compared with 27.7 ± 6.3 min in germ-free rats (P < 0.01). Inhibition was indicated also for Escherichia coli X7gnotobiotic rats. No consistent changes in slow wave frequency were observed. The concentration of neuropeptide Y in blood decreased after introduction of conventional intestinal microflora, suggesting reduced inhibitory control. Intestinal bacteria promote or suppress the initiation and aboral migration of the MMC depending on the species involved. Bacteria with primitive fermenting metabolism (anaerobes) emerge as important promoters of regular spike burst activity in small intestine.

  • bacteria
  • migrating myoelectric complex
  • gastrointestinal motility
  • neuropeptides
  • neurotransmitters

humans and domestic animals spend ∼1–2% of their life span in the germ-free state while in utero, and at birth microbes rapidly colonize the gastrointestinal tract. The number of bacteria in the gastrointestinal tract of adult humans amounts to 1015 (31), and 400–500 bacterial species are indicated from detailed cultures of feces (37). Most of these species are anaerobic, and the predominant average-sized bacterium is lactobacillus, accounting for ∼50% of the weight of all bacteria (31).

The motility of the gastrointestinal tract is important for absorption, transport, and clearance. Absorption is promoted by slow transit because of prolonged contact time (9) and clearance by rapid transit. Gut transit is slow in the absence of the intestinal microflora (2). The enlarged cecum of germ-free mammals accounts for this only in part, and stimulation of transit by the intestinal microflora has been demonstrated in several mammalian species (14) at different levels of the gastrointestinal (GI) tract (2) and both during fasting (1) and after feeding (2).

Recently, the influence of the conventional intestinal microflora on intestinal myoelectric activity was studied in vivo (6, 22,23). Myoelectric recordings in germ-free rats showed longer intervals between activity fronts of the migrating myoelectric complex (MMC) (phase III) compared with a group of conventional rats (6). When germ-free rats were recorded first in the germ-free state, then exposed to conventional intestinal microflora, and recorded again 7–10 days later, the MMC intervals decreased by 56% to the level of conventional rats and phase III migrated further aborally (22, 23). This shift in myoelectric activity was associated with accelerated transit through small intestine in the fasting state (23). The intestinal myoelectric response to food remained unchanged apart from an earlier return of activity fronts of MMC (phase III), reflecting the increased recurrence rate of phase III (23). Changes of spike burst activity were responsible for modulation of phase III, whereas the slow wave frequency remained unchanged (6, 23), although there were indications of a response depending on developmental maturity (23).

The present series of experiments studied how the composition of the intestinal microflora influences microbial modulation of enteric neuromuscular function. Extracellular myoelectric activity was recorded from unanesthetized germ-free rats before and again after inoculation with one or more pure cultures of bacteria. Bacterial species with different metabolic profiles, relationship to oxygen, and morphological characteristics were tested, and additional studies of neuroendocrine mediators, neuromessengers, and microbial metabolic products were performed to elucidate the potential cell messengers involved.


Experimental Animals

Germ-free caesarean section was performed on conventional Sprague-Dawley rats (ALAB, Stockholm, Sweden), and the offspring were bred and reared under germ-free conditions (15). The animals were kept on a steam-sterilized pelleted diet (R6; Ewos, Södertälje, Sweden), and the germ-free status was controlled weekly as described elsewhere (15). The experiments were conducted according to the ethical guidelines for animal experiments at Karolinska Institute after written approval.

A total of 95 Sprague-Dawley rats, males unless otherwise stated, were studied. For each experiment the total number of rats included and the number of rats with complete data sets are given, because failure of electrodes, intestinal obstruction, bleeding, etc. resulted in some missing data points.

Experimental Protocols

Experiments in germ-free state repeated after specific contamination.

The procedure for these experiments can be summarized as follows:1) surgical implantation of electrodes within the germ-free isolators; 2) recording of intestinal myoelectric activity within the isolators after 5–7 days of recovery; 3) transport under sterile conditions to a second isolator previously contaminated with the test microbe(s) using visitor rats; 4) on arrival at the second isolator, oral and rectal administration in study rats of 1 ml of fresh fecal contents, dissolved in 0.9% saline, from visitor rats to ensure immediate intestinal colonization (rats were then denoted gnotobiotic); 5) recording in the gnotobiotic state performed within the second isolator 7–10 days later, when the intestinal tract was colonized.


Chr. Hansen (Hønsholm, Denmark) provided the bacterial cultures ofLactobacillus acidophilus A10(Lactobacillus a.) and bifidobacterium bifidum B11(Bifidobacterium b.). Ten germ-free rats (7 males and 3 females), aged 120–150 days, were included, and a complete set of data was obtained in each of seven rats. The transit study was performed in eight rats.


Lactobacillus a. was provided by Chr. Hansen. Eight germ-free rats aged 90–120 days were included, and a complete set of data was obtained in each of six rats. After the experiment in the monocontaminated state, four of the rats were further contaminated withBifidobacterium b. and a third myoelectric recording was performed.


Clostridium tabificum n. sp. VP 04 (Clostridium t.) was provided by the late Prof. Z. Banhini. Eight germ-free rats aged 90–120 days were included, and a complete set of data was obtained in each of seven rats. The transit study was performed in six rats.


E. coli X7 (E. coli) was derived from rat feces by T. Midtvedt (18). Eight rats aged 90–120 days were included, and complete sets of data were obtained in all these rats. After the experiment all rats were killed for culture to determine the segmental distribution of bacteria in the small intestine (seeMicrobiological methods).

Experiments in gnotobiotic state repeated after introduction of conventional intestinal microflora.

The animals were operated on within their specific-contaminated isolators, and myoelectric recording was performed 1 wk later. The rats were then transferred to a conventional environment, exposed to feces from conventional rats as previously described (22), and recorded again after 1 wk.


The rats were accidentally monocontaminated with Micrococcus luteus (Micrococcus l.) within their isolator before surgery. Eight rats aged 120–150 days were included, and a complete set of data was obtained in each of six rats. Because of missing data after conventionalization, estimates for phase IIIs present at a site 5 cm from the duodenojejunal junction (J5; Jn represents a site n centimeters from duodenojejunal junction) that migrated to J25 [J5(J25)] had to be replaced by estimates for phase IIIs at J5 that migrated to J15 [J5(J15)].

Experiments in gnotobiotic state.

The animals were operated on within the specific-contaminated isolator, and myoelectric recording was performed 1 wk later.


The rats were accidentally monocontaminated with Micrococcussp. within their isolator before surgery. Six rats aged 90–120 days were included. Recordings were obtained in five rats, two of which were partial. The transit study was performed in five rats.

Studies of intestinal transit.

gnotobiotic state.

The transit study was performed as indicated above for the particular models.


The transit study was performed in eight germ-free rats, as reported elsewhere (23), and in two groups of conventional rats. In 10 conventional rats (conventional I), the implanted electrodes were sutured to the abdominal wall in an attempt to avoid twisting of intestinal loops. In eight conventional rats (conventional II), surgery was performed in the same way as in the germ-free and gnotobiotic groups.

Studies of cell messengers.

The tissue distribution of neuromessengers and endocrine cell markers and plasma concentrations of gastrointestinal neuroendocrine peptides were studied in a separate contamination experiment.


Fourteen germ-free rats ∼90 days old were studied. On the first day four rats were transferred to a conventional environment and 1 ml of fresh fecal contents from conventional Sprague-Dawley rats, dissolved in saline 0.9%, was administered orally and rectally (ex-germ-free 25 days). Fifteen days later, the same procedure for conventionalization was performed for another five germ-free rats (ex-germ-free 10 days). At day 25 all rats were killed, providing three randomized groups: germ-free, ex-germ-free 10 days, and ex-germ-free 25 days. Tubular whole tissue specimens in duplicate were obtained: 4 mm of gastric antrum, 5 mm of duodenum 10 cm distal to pylorus, 5 mm of jejunum 10 cm distal to the duodenojejunal junction, 5 mm of ileum 10 cm proximal to cecum, 5 mm of colon 10 cm distal to cecum. All specimens were flushed carefully with saline, opened along the mesenteric border, pinned flat, and fixed by a mixture of 2% formaldehyde and 15% saturated aqueous picric acid solution in 0.1 M phosphate buffer (pH 7.2). Specimens were transported to the laboratory for further processing the same day for subsequent immunocytochemical demonstration of neuropeptides, other neuronal messengers, and markers.

For radioimmunoassay of gastrointestinal peptides blood was taken by heart puncture and collected in heparinized tubes. Samples were immediately centrifuged at 2,500 rpm for 10 min and then stored at −70oC for subsequent RIA.


Surgical procedures.

Germ-free rats were operated on within their sterile isolators, whereas conventional rats were operated on under clean although not sterile conditions. The animals were anesthetized with pentobarbital sodium (25–50 mg/kg ip, Pentotal; Apoteksbolaget, Umeå, Sweden). After laparotomy, three or four bipolar electrodes of Teflon-coated stainless steel wire with diameter 0.250 mm (SS-5T, Clark Electromedical Instruments, Reading, UK) were implanted into the muscular wall of the small intestine at the antimesenteric border at 5, 15, 25, and 35 (if 4 electrodes) cm distal to the duodenojejunal junction, defined by the plica duodenocolica. The notation Jn, with nindicating the distance from the duodenojejunal junction in centimeters, refers to the recording sites. The electrodes were covered by polyethylene tube (PE-50; Clay-Adams, Parsippany, NJ). The catheter for infusion of radioactive marker was implanted 1 cm distal to J5. Electrodes and catheters were tunneled to the interscapular region, and catheters were flushed with 0.9% sterile saline and closed.

Myoelectric recording.

After 14–20 h of fasting with free access to water, recording was performed between 8 AM and 3 PM with the rats awake in a restriction cage. All rats were trained in the cage before experiments. A Grass polygraph (Grass Instruments, Quincy, MA) with preamplifiers (7P5B, Grass Instruments) and driver amplifiers (7B, Grass Instruments) was applied in a germ-free environment as previously described (22). Myoelectric recording in the selectively contaminated state was performed as in the germ-free environment within the same type of isolator (15) contaminated with the bacterium(a) under study. Slow waves were recorded for at least 1 min during phase I with a time constant of 0.45 s, an upper cut-off frequency of 3 Hz, and a paper speed of 2.5 mm/s. Subsequently, six consecutive phase IIIs, migrating at least from J5 to J25, were recorded with a time constant of 0.015 s, an upper cut-off frequency of 35 Hz, and a paper speed of 10 mm/min. The sensitivity was set to 0.25 mV/cm.

Phase III was defined as clearly distinguishable periods of regular spiking activity at the maximum frequency and amplitude observed at each electrode, indicated by at least doubling of baseline myoelectric activity (22). A minimum duration of 1 min was required, and the interval between phase IIIs was denoted the MMC period.1) MMC period J5 was based on all phase IIIs present atJ5, regardless of length of aboral migration, to indicate the rate of initiation for phase III of proximal origin. 2) MMC period J5(J25) was based on all phase IIIs present at J5 that migrated at least to J25, to indicate changes in both initiation and aboral migration of phase III. 3) The MMC migration ratio was calculated by the ratio between MMC period J5(J25) and MMC period J5 as previously defined (23). When six consecutive phase IIIs migrating from J5 to J25 are recorded, this ratio is estimated by 5/(N J5 − 1), whereN J5 is the total number of phase IIIs at J5 regardless of their aboral migration (23). An increase of the migration ratio reflects extended aboral migration of phase III.

Small intestinal transit.

A bolus of 0.4 ml of isotonic (300 mosmol/kg H2O; pH 7.0) test solution with 85 μl of Na2 51CrO4 (0.31 MBq), 40 μl of polyethylene glycol 400, and 275 μl of distilled water was injected slowly through the transit catheter 2 min after termination of a phase III at J5. Sixty minutes later, the rats were killed by cervical dislocation. Germ-free and gnotobiotic rats were immediately taken out of their isolators, and the abdominal cavity was opened. The gastroduodenal and ileocecal junctions were ligated, and the small intestine was removed in toto and divided into 10-cm segments for detection of radioactivity. Cecal contents were collected. Radioactivity was counted for 10 min per sample in a gamma scintillator (Packard Auto Gamma, Downers Grove, IL).

The percentage of total radioactivity in each segment was derived, and then the leading peak of radioactivity (23), geometric center (36), and proximal retention, defined as percentage of radioactivity within the proximal three segments, were calculated. The distance between electrodes was measured and adjusted for post mortem elongation.

Microbiological methods.

All strains were stored in pure cultures in the laboratory. Intestinal colonization was confirmed by culture of fecal samples collected 3–5 days after contamination using a standard culturing technique.

Furthermore, samples from different segments of the gut were collected in rats selected at random as follows. After laparotomy, the intestinal wall was opened by diathermy. Aliquots of 0.1 μl of intestinal contents were added to two tubes with 10 ml of Todd-Hewitt broth, one of which was prereduced for anaerobic growth. Anaerobiosis was secured by the pyrogallol method (16). Series of 10-fold dilutions were incubated at 37°C for 72 h in liquid media and plated on standard media appropriate for the selected microbes.

After myoelectric recordings in E. coli gnotobiotic rats, bacterial density in small intestine was studied in detail. After overnight fasting, samples were collected from four segments of small intestine at 0 (pylorus)-20 cm, 20–40 cm, and 40–60 cm and a 20-cm segment ending 5 cm proximal to cecum. Samples were inoculated onto liquid meat broth, Cled agar (4 rats), and lactose-bromthymol blue agar (3 rats). Triple series of 10-fold dilutions were used, and the mean value was calculated. A high degree of agreement was found for the three culture media. Values from agar plates are presented here.

Radioimmunoassay of neuroendocrine peptides.

vasoactive intestinal peptide.

Vasoactive intestinal peptide (VIP)-like immunoreactivity (VIP-LI) was analyzed using antiserum VIP2 raised against conjugated natural porcine VIP (50). The detection limit of the assay was 3 pmol/l. The coefficient of variation was 9%.


Neurokinin A (NKA)-like immunoreactivity (NKA-LI) was analyzed using antiserum K12, which reacts with NKA but not with substance P (SP) (45). The detection limit of the assay was 12 pmol/l, and the coefficient of variation was 7%.


Neurotensin (NT)-like immunoreactivity (NT-LI) was analyzed using antiserum H, which reacts with NT but not with NH2-terminal fragments of NT (46). The detection limit of the assay was 8 pmol/l, and the coefficient of variation was 8%.


Calcitonin gene-related peptide (CGRP)-like immunoreactivity (CGRP-LI) was analyzed using antiserum CGRPR8 raised in a rabbit against conjugated rat CGRP (19). The detection limit of the assay for rat CGRP was 9 pmol/l, and the coefficient of variation was 9%.


Somatostatin was analyzed as described by Grill et al. (13). The detection limit of the assay was 2 pmol/l. The coefficient of variation was 7%.


Neuropeptide Y (NPY)-like immunoreactivity (NPY-LI) was analyzed using antiserum (45). The detection limit of the assay was 11 pmol/l, and the coefficient of variation was 7%.


SP-like immunoreactivity (SP-LI) was analyzed using antiserum SP2 (5). The detection limit of the assay was 10 pmol/l, and the coefficient of variation was 7%.

Immunocytochemistry of neuromessengers.

Fixed tissue specimens were cryosectioned (10-μm thickness). Sections glued to slides by chromealum were then subjected to indirect immunofluorescence for the demonstration of VIP, NPY, gastrin-releasing peptide (GRP), CGRP, SP, galanin, serotonin, chromogranin A, and neuronal-type nitric oxide (NO) synthase, using well-characterized antibodies (32). Neuronal elements and enterochromaffin cells were examined.


Student's t-test for matched pairs and independent samples was applied. For each experiment, data in the germ-free and gnotobiotic states are summarized by mean and SD values unless otherwise stated. Changes from germ-free to gnotobiotic state were tested by factorial ANOVA. The level of statistical significance was P < 0.05. Standard correlation techniques were used.


Bacterial Colonization in Gnotobiotic Models

Segmental bacterial colonization of the gastrointestinal tract of gnotobiotic rats is shown in Table 1.

View this table:
Table 1.

Microbial density along the GI tract of ex-germ-free rats mono- and bicontaminated with selected microbial species

Slow Wave Frequency

Introduction of Clostridium t. resulted in a statistically significant increase of the slow wave frequency at J25 (P < 0.05), from 32.8 ± 2.7 to 35.9 ± 2.8 cycles/min, but not at J5 and J15. Within the other study groups, no changes of slow wave frequency were observed. In the germ-free and the gnotobiotic state, group mean ranges were 37.4–38.7 (J5), 35.2–36.8 (J15), 32.8–35.2 (J25), and 36.8–39.6 (J5), 33.8–37.3 (J15), 31.2–36.8 (J25) cycles/min, respectively. The slow wave frequency was within the same range for the conventionalized rats.

For the three groups with recording in both the germ-free and gnotobiotic state, the changes in slow wave frequency at J25 wereLactobacilli a. −6%; E. coli −3%; andClostridium t. +10%. A weak statistically significant difference was found between the groups (ANOVA; F = 6.9,P < 0.05).

MMC of Small Intestine

Conversion from germ-free to gnotobiotic state.

The small intestinal spike burst response to mono(bi)bacterial colonization of the gastrointestinal tract differed considerably for the species tested. The most marked changes were observed for the rate of recurrence of phase III of MMC, but there were also changes in the length of aboral migration of phase III (Table2). The germ-free recording of each rat served as control.

View this table:
Table 2.

Intestinal microflora of specific bacteria: Influence on initiation and migration of activity fronts of MMC in Sprague-Dawley rats


The rate of initiation increased, reflected by a statistically significant reduction of the MMC period J5, in Lactobacillus a.combined with Bifidobacterium b. (P < 0.01) (Fig. 1) and Clostridium t. (P < 0.05) gnotobiotic rats (Table 2; see Fig. 2). No change was found for Lactobacillus a. and E. colignotobiotic rats.

Fig. 1.

Lactobacilli acidophilus and Bifidobacterium bifidum stimulate intestinal myoelectric activity. Two myoelectric recordings from proximal jejunum of the same Sprague-Dawley rat are shown. The first recording (top tracings) was performed in the germ-free state and the second recording (bottom tracings) after bicontamination with Lactobacilli acidophilus and Bifidobacterium bifidum. J5, J15 and J25, electrodes located 5, 15, and 25 cm distal to the duodenojejunal junction, respectively.

Fig. 2.

Intestinal migrating myoelectric complex (MMC) period before and after introduction of specific intestinal microflora. The change in MMC period J5(J25) from the germ-free state to the gnotobiotic state is shown. Mean values with 95% confidence limits (based on all data points) and P-values for the paired comparisons are indicated. A: rats bicontaminated with Lactobacilli acidophilus and Bifidobacterium bifidum. B: rats monocontaminated with Clostridium tabificum. C: rats monocontaminated with Escherichia coli.


Increased length of aboral migration of phase III, reflected by an increased migration ratio, was found in Lactobacillus a.gnotobiotic rats (P < 0.05). For the other experimental groups, no statistically significant changes were observed. In Clostridium t. gnotobiotic rats, a borderline change was observed (P = 0.08; Table 2).


Lactobacillus a. (P < 0.05), Lactobacillus a. combined with Bifidobacterium b.(P < 0.02), and Clostridium t.(P < 0.01) gnotobiotic rats all showed a statistically significant and marked reduction of the MMC period J5(J25) (Table 2; Figs. 1 and2), reflecting the combined effect on initiation and length of aboral migration. No statistically significant change was found for E. coli gnotobiotic rats, but there was a trend toward prolonged MMC period J5(J25) (P = 0.08) (Fig. 2, Table 2).

The MMC period J5(J25) in the gnotobiotic state was expressed as a fraction of the value in the germ-free state. The following group estimates were obtained: Lactobacilli a. 0.76 ± 0.23; Lactobacillus a. and Bifidobacterium b.0.78 ± 0.15; E. coli 1.18 ± 0.08; andClostridium t. 0.69 ± 0.05. There were statistically significant differences between the groups for this estimate (ANOVA; F = 10.9, P < 0.001). The estimate for the E. coli experiment differed from those of the other experiments (P < 0.05, Fisher's protected least significant difference test, SchefféF-test), stratifying the microbial species into two groups, one group of bacteria that promoted the MMC, includingLactobacilli a., Lactobacillus a. combined withBifidobacterium b., and Clostridium t., and one group of bacteria without effect or inhibition of the MMC, including E. coli.


No statistically significant changes in duration or migration velocity of phase III of MMC were found for the recording segment J5–J25 in Lactobacillus a., Lactobacillus a. combined withBifidobacterium b., E. coli, and Clostridium t.gnotobiotic rats.

Gnotobiotic state with conversion to conventional intestinal microflora.

Germ-free rats from the experiments with prior recording in the germ-free state served as controls (n = 25).


Micrococcus l. gnotobiotic rats showed a reduced rate of initiation of phase III, reflected by a prolonged MMC period (Fig.3). The MMC period J5 was 24.0 ± 7.1 in the gnotobiotic rats (Table 2) compared with 18.8 ± 4.7 min (P < 0.05) in the germ-free controls. Later, when normal intestinal microflora was introduced to the Micrococcus l. gnotobiotic rats, the myoelectric pattern changed considerably. Phase III was initiated more frequently in the proximal small intestine, as reflected by a reduction of the MMC period J5 to 14.4 ± 3.6 min (P < 0.05) (Table 2).

Fig. 3.

Micrococcus luteus inhibits intestinal myoelectric activity. Two myoelectric recordings from proximal jejunum of the same Sprague-Dawley rat are shown. The first recording (top tracings) was performed when this rat was monocontaminated withMicrococcus luteus and the second recording (bottom tracings) after introduction of conventional intestinal microflora.


The MMC migration ratio in Micrococcus l. gnotobiotic rats was similar to that of germ-free controls (0.68 ± 0.15), and introduction of conventional intestinal microflora did not change the migration ratio statistically (Table 2).


The combined effect was shown by the prolonged MMC period J5(J25) of 35.9 ± 9.3 min in the gnotobiotic rats (Table 2) compared with 27.7 ± 6.3 min (P < 0.01) in the germ-free control group. Conventionalization of Micrococcus l.gnotobiotic rats reduced the MMC period J15 (P < 0.02) and the MMC period J15(J25) (P < 0.05).


The duration of phase III was similar in Micrococcus l.gnotobiotic rats and germ-free controls and did not change statistically after introduction of conventional microflora.

The migration velocity of phase III in Micrococcus l.gnotobiotic rats (1.03 ± 0.27 cm/min) was slower than in germ-free controls (1.28 ± 0.26 cm/min; P < 0.05) for the segment J5–J25, and also for the segment J5–J15 (P < 0.02). After conventionalization, the migration velocity J15–J25 increased significantly (P < 0.05).

Gnotobiotic state.

In Micrococcus sp. gnotobiotic rats complete myoelectric recordings were obtained in only three rats and partial recordings in two, precluding statistical comparisons. The MMC period J5(J25) (26.2 ± 10.4 min) was similar to that found in germ-free rats (27.7 ± 6.3 min) and in E. coli gnotobiotic rats (Table 2). The MMC migration ratio of 0.68 ± 0.23 also compared with that found in germ-free rats (0.68 ± 0.15) and in E. coli and Micrococcus l. gnotobiotic rats (Table 2).

MMC and segmental density of bacterial colonization.

The relationship between segmental bacterial density in small intestine and the MMC was studied in the E. coli experiment (Table 1).

A statistically significant correlation was found between the MMC period J5(J25) and the density of E. coli within the most proximal segment of small intestine (r = 0.81,P = 0.05). High counts in this 20-cm segment were found in the gnotobiotic rats with MMC period exceeding 30 min. No correlation was found for the other segments, and MMC migration ratio, duration, and migration velocity were not related to microbial density.

In the experiment with Lactobacilli a. andBifidobacterium b., statistically significant correlations were not found, but the number of animals studied was low and a corresponding relationship was indicated between bacterial density in proximal jejunum and MMC period J5 (r = 0.70,P = 0.08).

Small Intestinal Transit

In all rats a distinct leading peak was detected in small intestine, and <1% of radioactivity was found in the cecum samples (Fig. 4). Small intestinal transit estimated by leading peak differed significantly between the groups examined (F = 11.4, P < 0.001). The leading peak was located at 66% of small intestine in germ-free rats compared with 82% in rats with a conventional intestinal microflora (P < 0.05), reflecting the promotive effect of the intestinal microflora (Table 3).

Fig. 4.

Intestinal transit and composition of the intestinal microflora. The distribution of the radioactive marker (Na2 51CrO4) along the small intestine of Sprague-Dawley rats 60 min after intraluminal instillation of marker 15 cm distal to pylorus is shown. Solid line, a germ-free rat; gray line, a rat bicontaminated with Lactobacilli acidophilus and Bifidobacterium bifidum; dashed line, a conventional rat. The percentage of total radioactivity is given for 1/10 segments of small intestine. Note the proximal peak in the germ-free rat showing partial retention of radioactivity (see text).

View this table:
Table 3.

Intestinal microflora and interdigestive small intestinal transit

Rats monocontaminated with Clostridium t. andMicrococcus sp. showed leading peaks similar to their germ-free counterparts. In rats bicontaminated with Lactobacilli a. and Bifidobacterium b., an intermediate value of 71% was found (Fig. 4; Table 3). In all study groups with mono- or bicontamination, the leading peak was detected more proximally than in conventional rats (P < 0.05) (Table 3), demonstrating a slower rate of transit.

Geometric center of radioactivity showed the differences between groups less clearly (F = 3.9, P < 0.05). In the conventional I rats a bimodal distribution of radioactivity in small intestine was detected (Fig. 4), with 46% located within the proximal three segments compared with only 14% for the conventional II rats with “free” electrodes (P < 0.001). Accordingly, the geometric center was located more proximally in conventional I rats with fixed electrodes than in conventional II rats (5.0 vs. 6.3, P < 0.05) (Table 3). Leading peak, however, was the same in these conventional control groups (Table 3) (83% vs. 81%, not significant), showing that partial retention of marker because of the electrodes did not interfere with leading peak, which invariably was found aboral to the distal electrode.

Rats bicontaminated with Lactobacilli a. andBifidobacterium b. showed significantly faster transit than germ-free rats (P < 0.05) as estimated by geometric center (Table 3).

Small intestinal transit and MMC.

All rats with transit and MMC data were pooled into one group (n = 26) to study the relationship between MMC of proximal small intestine and fasting transit through the small intestine. This included the experiments with Lactobacillus a.and Bifidobacterium b., Clostridium t., and conventional control rats. Leading peak of radioactivity, MMC period J5, the MMC migration ratio, and MMC period J5(J25) were analyzed.

Leading peak was inversely correlated to MMC period J5(J25) (r = 0.60, P < 0.01), showing that a high frequency of aborally migrating phase IIIs in proximal small intestine predicted rapid aboral transit.

Separate analyses within each experimental study group did not show statistically significant correlation, but a similar pattern emerged in all rats harboring an intestinal microflora, regardless of the composition.

To control for an effect of partial retention by fixed electrodes on the MMC, proximal retention was compared with MMC period J5(J25) for the respective experiments. No statistically significant or borderline correlation was found.

Small intestinal transit and intestinal microbial colonization.

No statistically significant correlation was found between leading peak and density of microbial colonization (r = 0.69,P = 0.13).

Circulating Neuroendocrine Peptides


In germ-free rats, 10 days ex-germ-free rats, and 25 days ex-germ-free rats with conventional intestinal microflora plasma somatostatin was 83.4 ± 17.5 (mean ± SE), 47.0 ± 11.5, and 53.8 ± 13 pmol/l, respectively. There was a trend toward lower levels after conventionalization, regardless of time lapse from introduction of intestinal microflora, but the difference did not reach statistical significance (P = 0.09).


In the germ-free, 10 days ex-germ-free, and 25 days ex-germ-free rats plasma NPY was 664 ± 110, 525 ± 137, and 380 ± 83 pmol/l, respectively, showing a trend toward lower levels of NPY after conventionalization. One clearly outlying measurement (high value) occurred in the ex-germ-free 10 days group. Without this observation, which was reasonable to exclude, the difference between germ-free and all ex-germ-free rats was statistically significant (P< 0.05).

Other neuroendocrine peptides.

For VIP, CGRP, SP, NKA, and NT, plasma concentrations were below detection limits of 3.9, 7.8, 7.8, 7.8, and 7.8 pmol/l, respectively, except for one NT value of 11 pmol/l (ex-germ-free 10 days rat).

Gastrointestinal Neuromessengers and Endocrine Cell Markers

Density and distribution of nerve fibers and nerve cell bodies containing VIP, NPY, GRP, CGRP, SP, galanin, and neuronal-type NO synthase were similar in germ-free rats and 10 and 25 days ex-germ-free rats colonized by conventional intestinal microflora. Similarly, screening for endocrine cells using antibodies against serotonin and chromogranin A did not reveal differences between the groups.


This study shows how colonization of the gastrointestinal tract of a mammal by a single bacterial species influences fasting myoelectric activity of small intestine.

Clostridium t. alone and Lactobacillus a. andBifidobacterium b. in combination increased the rate of initiation of phase III in the proximal small intestine. Extended aboral migration was found in Lactobacillus a. gnotobiotic rats and was indicated in rats monocontaminated by Clostridium t. Increased migration velocity for phase III was observed after introduction of conventional microflora to Micrococcus l.gnotobiotic rats. These changes of fasting intestinal myoelectric activity resemble those previously reported after the introduction of conventional intestinal microflora to germ-free rats (23).

In contrast, introduction of E. coli, which colonized the small intestine with similar density, showed no statistically significant effect. Rather, this species tended to suppress the occurrence of phase III and slow down the rate of aboral migration. With increased MMC period for six of eight rats (Fig. 2 C) after introduction of E. coli, failure to detect a statistically significant suppressive effect may be a type II error. Marked suppression of the MMC was seen in Micrococcus l.gnotobiotic rats, demonstrating that microbial modulation of intestinal myoelectric activity is diverse and species dependent. The ANOVA comparing the relative changes in MMC period J5(J25) between groups showed that these effects exceeded random variation, indicating that intestinal bacteria can be stratified with regard to their effect on MMC as promotive, ineffective, or suppressive.

The promotive influence of the conventional intestinal microflora on MMC (6, 22, 23) thus reflects the net effect of bacterial species with partly opposite effects. The fact that strict anaerobic bacteria constitute >90% of the indigenous intestinal microflora (37) concurs with the observation of a promotive effect in particular for anaerobic bacteria in the present study.

The unvarying recording in the germ-free state before bacterial contamination could result in bias due to the fixed order of the experiments. There is, however, no difference between MMC variables of the first and the second recording in this model (23), and the diverse effects of the microbial species tested also argue against an effect of order.

Recording was confined to the fasting state, because it was shown previously that the intestinal microflora modulates fasting rather than postprandial intestinal myoelectric activity (23) and because the fasting motility pattern still serves as the best indicator of motility-dependent clearance of small intestine (21).

Phase III of MMC and Transit

The influence on the MMC was evidenced, in particular, by the changes in MMC period J5(J25), reflecting both initiation and aboral migration of phase III (23). The expectation in germ-free rats was 28 min in the present study (n = 25), according well with the 31-min period previously reported (23). Thus the intestinal microflora promotes the initiation and aboral migration of phase III activities in proximal small intestine of rats on the order of 60%, using 17.5 min as the estimate in the conventional state (23). This implies that the intestinal microflora is responsible for ∼40% of the stimulatory drive for physiological MMC activity (60/160 × 100).

The distinction between rate of initiation (MMC period J5) and degree of aboral migration for phase III (migration ratio) may be of physiological relevance. Although our data are not unambiguous at this point, the highest migration ratio, close to that of conventional rats, was observed in Lactobacillus a. and Lactobacillus a.combined with Bifidobacterium b. gnotobiotic rats. The latter experimental group was the only group with evidence for accelerated intestinal transit. Accordingly, a less promotive effect ofClostridium t. on aboral migration of phase III may be one reason why transit was not accelerated in that experiment, despite the considerably increased rate of initiation of phase III. Reduced rate of intestinal transit (49) despite increased initiation of phase III in proximal small intestine after subcutaneous octreotide and retroperistalic contractions during duodenal phase III in humans (4) are findings demonstrating that properties other than the temporal occurrence of phase III predict changes of intestinal transport and clearance (21).

The complexity of the relationship between MMC phase III activity and transit was further indicated by the correlation analysis. Only 36% of the variability of intestinal transit could be attributed to changes in the MMC in the present study (r 2 × 100), and in a previous study the conventional microflora changed the initiation and aboral migration of phase III twice as much as the acceleration of intestinal transit (23). Changes in phase II activity, which normally account for ∼50% of fasting transit, are likely to explain this in part (29). The present model, however, did not allow for detailed and quantitative analysis of phase II spike burst activity.

Failure of the transit technique to detect microflora-associated changes could also be involved. Although the present technique eliminates the effects of gastric emptying and accounts for MMC phase dependence at marker instillation (23), it also suffers from limitations. In some rats an additional proximal peak of radioactivity was observed, and suturing of the electrodes to the parietal peritoneum induced this peak. At post mortem laparotomy electrodes were sometimes adherent to the parietal peritoneum, although free by implantation, which may explain the occasional occurrence of a proximal peak. It follows that implanted electrodes can bias geometric center, otherwise considered the most reliable and valid parameter for this transit test (36). Proximal retention was negligible in Lactobacillus a. combined with Bifidobacterium b. gnotobiotic rats, which is why this parameter could be applied in that experiment (cf. Table 3). Proximal retention of marker did not affect leading peak of radioactivity or the MMC period J5(J25), confirming the validity of the present myoelectric and transit experiments. Localization of the majority of radioactivity in one distinct distal peak (Fig. 4) further favors transit measurement by leading peak in the presence of implanted electrodes.

This part of the analysis shows that implanted electrodes in general may interfere with transport. In the present study the MMC parameters and leading peak of the radioactive marker were not biased, indicating that this interference is limited and that it can be accounted for by appropriate choice of parameters. Degree of proximal retention serves as an index of electrode-mediated interference with intestinal transit, and a transit study can thus be used to assess experimental quality in models with implanted intestinal electrodes.

Slow Wave Frequency

Increased slow wave frequency was indicated after introduction ofClostridium t. Because the slow wave frequency was within the range of germ-free rats for all the other study groups, and for conventionalized rats in particular, this finding must be interpreted with care. The normal exposure to the intestinal microflora at birth may result in an increase of slow wave frequency in rats (23), which is not seen when germ-free rats are conventionalized after suckling (6) and at the age of 2–3 mo (23). Clostridium t. was introduced at the age of 3 mo, and the minor change confined to J25 may thus be due to chance.

Gnotobiotic Models: Role of Bacterial Species

Gnotobiotic models have previously been used to study intestinal transit and cecum size.

Gnotobiotic mice associated with E. coli,Streptococcus faecalis, lactobacilli, and severalBacteroides sp. showed small intestinal transit rate only slightly higher than that in germ-free mice (26). Further colonization with a clostridium strain and 12 strains of strictly anaerobic species was required to obtain a transit rate similar to that of conventional mice (26). Gnotobiotic rats that harbored four strictly anaerobic species, able to reduce cecal size, showed increased rate of intestinal transit compared with germ-free rats (48). In accordance with these findings, monocontamination of rats in the present study did not change small intestinal transit significantly. The strongest indication of stimulated transit was seen in Lactobacillus a. and Bifidobacterium b.gnotobiotic rats.

Cecum of germ-free mice was reduced to conventional size by monocontamination with Clostridium difficile or with two strains of Bacteroides, neither of which were effective alone (43). Aerobic bacterial isolates from cecal contents of conventional mice, however, had no effect on cecal size (43), indicating less importance for facultative bacteria. In germ-free rats, C. difficile (17) and aStreptococcus sp. (Enterococcus sp.) (20) isolated from feces of conventional rats partially reduced the size of cecum, whereas Lactobacillus acidophilusalone had no effect (30). Reduction of cecum size in germ-free rats has also been induced by monocontamination withProteus vulgaris and E. coli (31). Moreover, adding procaine penicillin to the diet of conventional rats, which reduced S. faecalis counts and virtually eliminatedClostridium perfringens, increased the size of cecum close to the germ-free level (12). Although there are conflicting reports regarding the effect of E. colimonocontamination on cecum size (31, 43), E. coli X7, used in the present study, does not reduce the size of cecum (18).

Anaerobic bacteria thus accelerate gastrointestinal transit (26,48) and reduce the cecum of germ-free mice (43) and rats (12).

Metabolic Properties of Bacteria

The species-dependent diversity in myoelectric response shown in the present study could be related to the phylogenesis of prokaryotes that predicts their main metabolic pathways. Prokaryotes with anaerobic fermenting metabolisms are considered an early stage of cellular life.

The further development of prokaryotes added new metabolic pathways in the following temporal order: respiratory metabolism, utilization of light, fixation of CO2, and later the utilization of oxygen. Clostridium t., Lactobacillus a., andBifidobacterium b. are prokaryotes with anaerobic fermenting metabolisms, the former two aerotolerant, whereas E. coliand micrococci with aerobic metabolic pathways represent the late stage in the evolution of prokaryotes.

Bifidobacteria, belonging to the family Lactobacteriacea, are difficult to establish as monoflora in germ-free rats because of their sensitivity to oxygen, and the cocolonization with a lactobacillus was applied to reduce the redox potential sufficiently for their growth. Being obligate fermentative, lactobacilli display a primitive type of metabolism, and the ability to utilize lactose reflects later adaptation to the mammalian digestive tract. The idea that lactate could be of importance for luminal control of MMC is opposed by the fact that E. coli and most enterobacteriacea also have lactate as a metabolic end product.

Clostridium t. is sufficiently aerotolerant to colonize the GI tract of germ-free rats as monoflora. Like lactobacillacea, clostridia are fermentative and normally do not exhibit cytochromes and catalase, resulting in the production of short-chain fatty acids (SCFA), CO2, H2, and sometimes NH3as metabolic end-products. CO2 and H2 are common end-products of bacterial metabolism also for E. coli, suggesting a role of SCFA as mediators.

Physiological composition and concentrations of SCFA induce propulsive motility when infused in ileum of the dog, probably through a local neural reflex involving opioid and prostanoid pathways (27). In the rat, ileal infusion of a similar solution accelerated mouth-to-cecum transit time, a response that was conversely proportional to chain length (40). In humans, infusion of SCFA in proximal small intestine did not alter motility patterns or orocecal transit time (33). Stimulation of colonic but not cecal muscle strips in vitro further indicates a modulating role of SCFA with segmental difference (44).

Germ-free rats monocontaminated with E. coli X7 have both acetic and propionic acid in cecal contents, but in markedly lower concentrations than conventional rats (18). Moreover,i-butyrate and in particular n-butyrate, typical end products of the metabolism of clostridia, could not be detected (18). We later analyzed, by established techniques (34), feces of germ-free mice monocontaminated withClostridium t. (Table 4; unpublished data). Both i-butyrate and i-valerate were detected, whereas acetic acid and propionic acid were found in concentrations four times higher than in rats monocontaminated byMicrococcus sp. Monocontamination of mice withLactobacillus a. resulted in the production of similar types and concentrations of SCFA, but the response varied among the animals (Table 4). Further studies are thus needed to clarify the role of SCFA. Pilot experiments with infusion of acetic, propionic, and butyric acid into different segments of the small intestine of germ-free and conventional rats did not show the type of myoelectric changes reported in the present study (unpublished data). SCFA may thus serve as markers of a metabolically active anaerobic microflora rather than being responsible mediators.

View this table:
Table 4.

Markers of microbial metabolism in fecal contents of gnotobiotic mice

A range of bacteria belonging to genera that are mainly anaerobic (8, 16) metabolizes bile acids. Accelerated small intestinal transit and reduced cecum size in gnotobiotic rats harboring an intestinal microflora of four strictly anaerobic species unable to metabolize the main bile salts, however, indicate that microbially transformed bile acids are not required (48). Moreover, significant amounts of urobilin in feces were not detected in mice later monocontaminated by Clostridium t., Micrococcus sp., or Lactobacillus a. (Table 4).

Absence of a stimulatory effect of E. coli on MMC, as earlier shown also for intestinal transit (26), is important because of the wide metabolic spectrum of this bacteria and the bioactive lipopolysaccharide (LPS) expressed by the cell membrane. It is notable that E. coli and micrococci with aerobic metabolic pathways, dating their origin to a later stage in the evolution of bacteria, were without stimulatory influence. The species-dependent MMC response thus seems to reflect the stage of prokaryotic phylogenesis to which the respective bacteria belong.

The stimulatory effect of the intestinal microflora on MMC is rapidly established within a few days (23) and sustained without evidence for the development of tolerance (23,48). The classical cell messenger-receptor interaction to a sustained stimulus is downregulation of the receptors, resulting in a decreased response by time. Reset of the expression of cell and luminal messengers by introduction of intestinal microflora thus may not be sufficient to explain the MMC response.

Even germ-free animals are exposed to microbial products, because the autoclaved food contains dead bacteria (35) with mediators like LPS not inactivated by the sterilization. Presence of live bacteria may thus be a prerequisite. Microbial inactivation of host-derived bioactive substances, or the activation of a precursor, should therefore be considered.

A musculoactive substance (MAS) in cecum of germ-free rats that relaxes smooth muscle has been detected, the inactivation of which is seen after conventionalization (11). It has been suggested that the MAS, not yet chemically identified, is derived from the mucus of cecum. Although attractive to explain the changes in cecal tone, microbial degradation of the MAS does not necessarily explain changes of intestinal MMC activity and transit. Interestingly,Clostridium t. was the only one of the bacteria later tested in germ-free mice that showed evidence of microbial degradation of mucin and cholesterol (Table 4).

The present study thus provides evidence that anaerobic bacterial metabolism is of importance, with certain SCFA and components of mucus emerging as potential mediators, whereas the role of microbially derived or modified LPS, lactate, H2, and bile acids remains elusive.

Cell Messengers

Evidence for changes in neurohumoral control mechanisms of intestinal smooth muscle function was elucidated by the experiment on cell messengers.

The role of serotonergic transmission in control of MMC in rats was shown when low intravenous doses of 5-hydroxytryptophan promoted the recurrence of phase III, whereas high doses induced irregular spiking activity (42). The conventional intestinal microflora increases the volume of serotonin immune-reactive cells in colon and ileum and in the gastric fundus but not in the small intestine (47). The immunocytochemistry data in the present study accord with this. Although Clostridium t. was selected for its ability to produce biogenic amines in vitro (28) it is speculative to assign the MMC response to luminal 5-hydroxytryptamine of microbial origin. Serotonin is rapidly degraded in lumen, and the segmental distribution of the microflora suggests distal effects rather than effects in the proximal small intestine, where the density of bacteria is low and variable (cf. Table 1).

NO is also involved in the control of MMC, with increased recurrence of MMC when the NO synthase is inhibited byN G-nitro-l-arginine (41). We did not find changes in neuronal-type NO synthase by immunocytochemistry. This does not rule out modulation of the kinetics of NO or the influence on other types of NOS in the intestinal epithelium.

Hormonal mechanisms were also explored. Of a series of gastrointestinal peptides, NPY was the only one with a change in blood concentration (reduction) after introduction of conventional intestinal microflora. A similar effect was suggested for somatostatin. Reduced blood concentration of these predominantly inhibitory peptides accords with reduced tissue concentration of somatostatin in small intestine of conventional rats to ∼30% of that found in germ-free rats (47).

Enteric NPY fibers are dense in all layers of small intestine, and there is one population supplying mainly the mucosa and the smooth muscle. The colocalization with VIP (7) and indications of an inhibitory effect on ileal motor activity (10) concur with the present finding of reduced blood levels of NPY after conventionalization, because phase III-like activity is promoted when neural inhibition of smooth muscle is reduced (41).

Somatostatin administered into the lateral ventricles of the rat increases the frequency of MMC, whereas high doses intravenously reduce it (3). Reduced somatostatin level in blood may contribute to accelerated transit, but the effect on MMC is dose dependent and more complex.

In conclusion, this part of the present study shows evidence for a role of NPY, possibly liberated from enteric nerves, in microbial modulation of MMC. For the other cell messengers the data were negative.

Bacteria and Motility

The detailed bacteriological sampling in the E. coliexperiment showed a positive correlation between bacterial density in the most proximal segment and the MMC periodJ5(J25) in the gnotobiotic state. The same correlation was indicated in the Lactobacillus a. andBifidobacterium b. gnotobiotic model and was recently reported in conventional rats (38).

The MMC data for the E. coli gnotobiotic rats allow inferences to be made regarding the issue of cause and effect for the interaction between intestinal bacteria and motility. Because the rats with the longest MMC periods in the gnotobiotic state also exhibited the longest MMC periods in the germ-free state (Fig. 2 C), the density of E. coli in the proximal small intestine can partly be explained as the result of the inherent recurrence rate of MMC, as evidenced in the germ-free state. Thus rats with inherent long MMC periods will have a higher density of bacteria after colonization. It follows that the rate of recurrence of MMC in conventional rats is the result of both the inherent recurrence rate and the modulating influence of the intestinal microflora. The finely tuned MMC activity then controls the segmental density of the microflora along the small intestine as shown in Table 1. Although modest MMC-related differences in bacterial density can be seen among rats within the proximal small intestine, Table 1 shows that the average density of bacteria throughout the small and large intestines is roughly the same for the bacteria tested and similar to that found for conventional microflora in rats of this stain (23).

According to the relationship between E. coli and the MMC in the proximal small intestine in the present study, the density of Gram-negative bacilli in the upper gut of humans is closely related to the MMC activity of proximal small intestine (24) when there is no blind loop or fistula present. Reduced incidence and migration velocity of phase III predicted Gram-negative bacilli in the upper gut (24). The density of Gram-positive microflora, however, can be high also in the presence of normal MMC activity, when hypochlorhydria provides a gastric reservoir for multiplication of swallowed oropharyngeal bacteria (24, 25).

The present findings suggest that in bacterial overgrowth, for example, the colonizing bacterial species may be of importance for the alterations in small intestinal function and symptoms. Some bacteria my further impede peristalsis, whereas others may promote transport and clearance. It remains to be clarified whether the types of overgrowth flora only reflect the underlying pathophysiology (24,25), or whether (and how) the overgrowth flora itself may contribute to the problem.

Although the composition of the microflora in healthy individuals is rather stable over years, there is also a gradual shift by time, because the continuous supply of live microbes by the oral route is a prerequisite for maintaining a complex intestinal microflora (39). Considering the alterations in bowel habits often resulting from major changes of microbial environment, it is possible that even subtle temporal changes of the intestinal microflora during life may contribute to the generation of yet unexplained GI symptoms. A better understanding of the role that different intestinal microbes play in the control of intestinal neuromuscular function is thus needed.

The intestinal microflora permanently promotes regular spike burst activity and transit of the small intestine during fasting, being responsible for ∼40% of the stimulatory drive for phase III of MMC. The slow wave frequency remains unchanged.

Modulation is likely to involve multiple pathways with different messengers, some of which have been indicated by the present study, e.g., certain SCFA and NPY, in addition to degradation products of mucin. The present data argue against the idea that the promoting effect is dependent on metabolites (e.g., lactate, hydrogen, acetic and propionic acid) and bioactive substances (e.g., LPS) associated with the colonization of facultative aerobic bacteria.

The promoters identified appear to be bacteria with primitive fermenting metabolism (anaerobes), of which clostridia may be of particular importance. Bifidobacteria also exhibit this effect, whereas the facultative or aerobic bacteria like E. coli andMicrococcus l. and Micrococcus sp. seem to be indifferent or suppressive. The phylogenetic differences among bacteria may thus explain the diversity in intestinal myoelectric response.


We thank Dr. Odd O. Aalen, University of Oslo, for statistical advice.


  • This study was financially supported by the Norwegian Medical Research Council (no. 191), the Swedish Medical Research Council (nos. 7916, 4499, and 6852), Ekhage Foundation, Påhlson Foundation, Åke Wiberg's Foundation, funds of the Karolinska Institute, and Løvens Chemical and by the Research Forum of Ullevaal Hospital.

  • Address for reprint requests and other correspondence: E. Husebye, Clinic of Medicine, Ullevaal Univ. Hospital, N-0407 Oslo, Norway (E-mail: einarhu{at}

  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.


View Abstract