Protective effects of Lactobacillus rhamnosus GG against human rotavirus-induced diarrhoea in a neonatal mouse model

Abstract

Group A human rotaviruses (RV) are a leading cause of severe dehydration and gastroenteritis in infants and young children. A large body of evidence suggests that Lactobacillus rhamnosus GG (LGG) has an effect on the incidence and severity of acute RV-induced diarrhoea; however, the timing and dosage of LGG treatment remains controversial. In the present study, a neonatal mouse model with human RV-induced diarrhoea was set up and the pathophysiological characteristics of the animals were examined. Our results indicated that RV-infected mice developed diarrhoea, accompanied by increased secretion of intestinal mucosa sIgA and serum interferon (IFN)-γ, tumour necrosis factor (TNF)-α, as well as decreased serum IgA. In addition, epithelium vacuolation was noticed in the jejunum microvillus of RV-infected mice. After intragastric administration of low (2 × 10 ⁵ CFU), middle (2 × 10 ⁷ CFU) or high (2 × 10 ⁹ CFU) levels of LGG for four consecutive days before or after RV infection respectively, the RV-infected mice showed a shortened duration of diarrhoea and decreased epithelium vacuolation in the jejunum. Administration of a high dose of LGG before the RV infection was found to have better protective effects against RV infection than other regimens. This study demonstrates that the protective effects of LGG against RV-induced diarrhoea are highly correlated with the timing and dosage of LGG administration in neonatal mice.

diarrhoea, dosage-dependent, human rotavirus, Lactobacillus rhamnosus GG , timing-dependent

Introduction

Group A human rotavirus (RV) is the leading cause of diarrhoea in infants and young children worldwide, and more than 450 000 children under the age of 5 die of RV-infected disease each year (Tate et al. , 2012 ). Oral rehydration therapy (ORT) and intravenous infusions are considered a useful method for the treatment of RV-induced diarrhoea by reducing dehydration and adjusting acid-base disturbance; however, this treatment does not shorten the duration of RV-induced diarrhoea or reduce severe loss of fluid (Ozuah et al. , 2002 ; O'Ryan et al. , 2010 ).

Lactobacillus rhamnosus GG (LGG) possesses antimicrobial activities that contribute to the host gastrointestinal defense system, and many specific beneficial effects of LGG have been documented in a large number of clinical trials (Guarino et al. , 2009 ; Grandy et al. , 2010 ; Ferrie & Daley, 2011 ). These include shortening the duration of rotavirus diarrhoea, reducing the number of diarrhetic episodes, lessening rotavirus shedding, normalizing gut permeability, and increasing the production of rotavirus-specific antibodies (Isolauri et al. , 1991 ; Pant et al. , 1996 , 2007 ; Guandalini et al. , 2000 ; Szajewska et al. , 2007 ; Ritchie et al. , 2010 ). However, quite a few studies have queried the protective effects of LGG against RV-induced diarrhoea. Costa-Ribeiro et al . found that LGG can neither shorten the duration of diarrhoea nor reduce stool output in RV-induced diarrhoea (Costa-Ribeiro et al. , 2003 ). Mastretta's investigation found that LGG may be ineffective in preventing nosocomial RV infection or protecting against the onset of diarrhoeal symptoms in infected patients (Mastretta et al. , 2002 ). Szajevska and coworkers observed that LGG did not prove to be effective in the prevention of rotavirus infections; however, it did prevent the symptoms of diarrhoea (Szajevska et al. , 2001 ). These inconsistent results may be due partly to different timing of administration and dosage of LGG was used in the investigations.

Isolauri et al .'s ( 1995 ) study indicated that the immunogenicity of oral rotavirus vaccines can be increased by simultaneous administration of viable LGG. Pant suggested that the combination of LGG with specific bovine colostrum-derived immunoglobulins could be used as an effective prophylactic method against RV-infected diarrhoea in the mouse model (Pant et al. , 2007 ). These two studies also mentioned that administration of LGG may be more effective before than after RV infection. Moreover, although many studies have investigated the dose-dependent effects of Lactobacillus (Mattar et al. , 2001 ; Szajewska & Mrukowicz, 2005 ), the appropriate doses of LGG against RV-induced diarrhoea remain largely undefined. Therefore, an RV-infected neonatal mouse model was used here to evaluate the efficacy of different timing and dosage of administration of LGG against RV-induced diarrhoea using multiple research techniques including pathophysiology, histopathology, immunohistochemistry and immunology.

Materials and methods

Cells and viruses

The virulent Wa strains RV (G1P1A) were purchased from the Virus Research Institute of the Chinese Academy of Preventive Medicine and propagated as described previously (Takahashi et al. , 2001 ). MA104 cells (African rhesus monkey kidney cells) were used for the multiplication of RV strains. Viral titre was determined by a standard plaque assay (Warfield et al. , 2006 ). The virus was multiplied until the viral titre reached 10 ⁷ plaque-forming units (PFU) in the following experiments.

Probiotic bacterial strains

Lactobacillus rhamnosus GG strain (ATCC 53103) was purchased from The Guangzhou Baiqing Biology Science and Technology Corporation in China. LGG was propagated in lactobacilli MRS broth overnight at 37 °C in sealed anaerobic jars. The bacterial counts were expressed as colony forming units per millilitre (CFU mL ⁻¹ ).

Animals

Fourteen-day pregnant Kunming mice were purchased from the Experimental Animal Center of Wuhan University; all pregnant mice had negative RV-antigen in the faecal samples by ELISA. The mice were returned to the dam and allowed to suckle after each administration of RV and LGG. All procedures were performed in strict adherence to the Tsinghua University Guide for the Care and Use of Laboratory Animals.

Experimental design

A total of 120 neonatal mice were divided into eight groups of 15 mice each (Table 1 ). One group of animals were RV-infected without LGG treatment (RV-infected group). Three groups of mice were pretreated with low (1 × 10 ⁶ CFU mL ⁻¹ ), medium (1 × 10 ⁸ CFU mL ⁻¹ ) or high (1 × 10 ¹⁰ CFU mL ⁻¹ ) doses of LGG before RV infection, respectively (L-LGG-pretreated group, M-LGG-pretreated group, H-LGG-pretreated group). Three groups of mice were treated with the above concentrations of LGG after RV infection (L-LGG-treated group, M-LGG-treated group, H-LGG-treated group). There was one normal control group. The animals in the RV-infected group were intragastically administered 50 μL RV fluid (3 × 10 ⁶ PFU mL ⁻¹ ) two times on the 4th day after birth, as described previously (Boshuizen et al. , 2003 ; Guerrero et al. , 2010 ). The L-, M-, and H-LGG-pretreated groups were given 50 μL 1 × 10 ⁶ CFU mL ⁻¹ , 1 × 10 ⁸ CFU mL ⁻¹ , and 1 × 10 ¹⁰ CFU mL ⁻¹ doses, respectively, of LGG four times per day from the 1st to 4th day after birth, and were given 50 μL RV fluid (3 × 10 ⁶ PFU mL ⁻¹ ) two times on the 4th day. The L-, M-, and H-LGG-treated groups were administered 50 μL RV fluid (3 × 10 ⁶ PFU mL ⁻¹ ) two times on the 4th day, followed by 50 μL 1 × 10 ⁶ CFU mL ⁻¹ , 1 × 10 ⁸ CFU mL ⁻¹ and 1 × 10 ¹⁰ CFU mL ⁻¹ doses of LGG four times daily from the 5th to 8th day, respectively. The animals in the normal control group were given 50 μL medium without antibiotics two times on the 4th day .

Table 1

Animal groups

	Intragastical administration of LGG
Group	Day 1	Day 2	Day 3	Day 4	Day 5	Day 6	Day 7	Day 8
Normal control group	–	–	–	C	–	–	–	–
RV-infected group	–	–	–	R	–	–	–	–
L-LGG-pretreated group	L	L	L	L+R	–	–	–	–
M-LGG-pretreated group	M	M	M	M+R	–	–	–	–
H-LGG-pretreated group	H	H	H	H+R	–	–	–	–
L-LGG-treated group	–	–	–	R	L	L	L	L
M-LGG-treated group	–	–	–	R	M	M	M	M
H-LGG-treated group	–	–	–	R	H	H	H	H

–, untreated; C, intragastically administered with MEM medium; R, intragastically administered with RV (3 × 10 ⁶ PFU mL ⁻¹ ); L, intragastically administered with a low dose of LGG (2 × 10 ⁵ CFU); M, intragastically administered with medium dose of LGG (2 × 10 ⁷ CFU); H, intragastrically administered with a high level of LGG (2 × 10 ⁹ CFU).

Diarrhoea index

The diarrhoea of mice was observed twice a day (morning and evening) from the 5th to 8th day after administration of RV. Faecal samples were collected daily, and a six-point rating system was used to describe the Diarrhoea index (DI) (Knipping et al. , 2011 ): 1, no stool; 2, normal brown formed stool; 3, soft brown stool; 4, soft-mucous brown-yellow stool; 5, muddy-mucous yellow; 6, liquid yellow stool. DI 1, 2 and 3 were considered to be no diarrhoea; 4 and 5 light diarrhoea; and 6 heavy diarrhoea.

Viral shedding in faeces

Levels of RV antigen in faeces were measured using a commercial EIA kit (Shanghai Senxiong Science and Technology Industry Corporation). Stool specimens at dilutions of 1 : 5, 1 : 10, 1 : 50, and 1 : 100, respectively, in phosphate-buffered saline (PBS) were collected to measure the amount of RV antigen. The reference concentration of RV antigen in stock solutions of 300 ng mL ⁻¹ was used as the standard, and concentrations more than 150 ng mL ⁻¹ were regarded as RV antigen-positive.

Detection of intestinal mucosa sIgA, and serum RV-specific IgA, IFN-γ and TNF-α

On the morning of the 9th day, blood samples were obtained under anaesthesia by cardiac puncture, mice were sacrificed by cervical dislocation, and small intestine specimens were collected. A 5–10 cm section of small intestine from the pylorus to the ileocaecal junction was cut off and cleaned with cold sterilized PBS. The top of small intestinal mucosa was collected and stored at −70 °C for later analysis. Intestinal mucosa sIgA was measured using an ELISA assay (Sigma), and the standard curve was established from the standard samples. Serum samples were centrifuged at 400 g for 10 min, and the RV-specific IgA, IFN-γ and TNF-α were also detected using an ELISA assay (Shanghai Senxiong Science and Technology Industry Corporation).

Viral load of RV-VP7 in intestine

Tissue samples were excised from small intestine and total cellular RNA was extracted. Real-time RT-PCR was performed to measure the expression of RV VP7 gene. The bars represent the geometric mean of the viral load after normalization with the housekeeping GAPDH gene. The copy number of RV-VP7 was measured by real-time quantitative PCR (Invitrogen) with an identical reaction without reverse transcriptase. The copies of VP7 gene were calculated according to the standard curve established by a series of pUC18T-VP7 plasmid vector concentrations. PCR amplification was performed with a 7300 Real-Time PCR System (Applied Biosystems). Fewer than 10 copies mL ⁻¹ virus VP7 RNA was defined as infection clearance.

Histopathological analysis

Tissues of jejunum and ileum were excised and perfused with formalin on day 9. The intestinal tissues were kept immersed in formalin (10%) for < 1 day and were transferred into 70% ethanol. The samples were embedded in paraffin and sections were stained with haematoxylin and eosin (HE) using standard protocols. The ultrathin sections of jejunum and ileum were stained with uranyl acetate and lead acetate, and were examined with a transmission electron microscope (TEM, JEOL-1200EX).

Statistical analysis

All data were analyzed using spss 15.0. When differences between the groups were detected, the same test was used in a pairwise fashion to identify the nature of the differences. One-way analysis of variance ( anova general linear model), followed by Duncan's multiple range test, was used to compare the mean duration of virus shedding, total faecal viral antigen shedding, intestine sIgA and serum RV-specific IgA, IFN-γ and TNF-α for the animals receiving different doses of LGG. Correlation coefficients were calculated based on a simple linear model. All statistical significance was assessed at P < 0.05.

Results

Diarrhoea and duration of RV in faeces

No signs of diarrhoea or death were observed in the normal control group in the experiment. The RV-infected mice had diarrhoea from days 5 to 9 the daily viral shedding of faeces reached more than 270 ng mL ⁻¹ after RV administration, seven animals died before day 9 (mortality rate of 46.67%), and four and five animals had light and heavy diarrhoea, respectively. Compared with the RV-infected group, the L-, M-, and H-LGG-pretreated groups showed reduced diarrhoea and shortened diarrhoea duration; the mortality rates were 20, 6.67 and 0%, respectively. The differences in the pretreated and treated groups in the amount of diarrhoea and death, and duration of positive RV antigen were negatively correlated with the dosage of LGG (Table 2 ).

Table 2

Amount of diarrhoea, diarrhoea duration and RV antigen in faeces (mean ± SD)

	Diarrhoea		Death
Group	Light ( n )	Heavy ( n )	n	Rate (%)	Diarrhoea duration (days)	RV antigen-positive (days)
Normal control	0	0	0	0	0	0
RV-infected	4	11	7	46.67	4.20 ± 0.37	3.88 ± 0.54
L-LGG-pretreated	2	4	3	20.00	3.29 ± 0.38 *	2.85 ± 0.46 **
M-LGG-pretreated	2	1	1	6.67	2.61 ± 0.26 *	2.09 ± 0.19 *
H-LGG-pretreated	2	0	0	0	1.74 ± 0.38 *	1.15 ± 0.26 *
L-LGG-treated	6	7	6	40.00	4.09 ± 0.51	3.72 ± 0.62
M-LGG-treated	2	6	5	33.33	3.53 ± 0.53 **	3.12 ± 0.48 **
H-LGG-treated	2	2	2	6.67	2.17 ± 0.32 *	1.32 ± 0.38 *

Diarrhea index (DI) of light diarrhoea is 4 or 5, heavy diarrhoea is 6.

Compared with the RV-infected group: * P < 0.01 and ** P < 0.05.

Viral shedding of RV in faeces

Viral shedding of RV in the faeces of all groups was determined by measuring the amount of virus antigen shed at days 5, 6, 7 and 8. In the normal control group, RV was rare, whereas RV shedding increased gradually from days 5 to 8 in the RV-infected group, reaching 887 ng mL ⁻¹ on day 8. Compared with the RV-infected group, RV shedding of LGG-pretreated and -treated groups decreased gradually with increasing LGG dosage. Overall, the RV shedding of the H-LGG pretreated group was the lowest (122 ng mL ⁻¹ ) at day 8, with the exception of the normal control group (Fig. 1 ).

Figure 1

Viral shedding of rotaviruse (RV) in faeces of mice. Detection of RV antigen in faeces (ng mL ⁻¹ , n = 8). Results are expressed as means ± SD, ^aP < 0.01 and ^bP < 0.05 compared with the normal control group; ^cP < 0.01 and ^dP < 0.05 compared with the RV-infected group; ^eP < 0.01 and ^fP < 0.05 compared with the treated group at equivalent doses of LGG.

Real-time RT-PCR detection of viral load in the small intestinal mucosa

In the normal control group, RV was rarely detected, in contrast to all the other groups, in which the viral load in infected group reached as high as 2806 copies mL ⁻¹ . In the treated groups, the lowest viral load in the small intestine was 400 copies mL ⁻¹ in H-LGG-treated group. Administration of 2 × 10 ⁹ CFU LGG before RV infection resulted in a significant reduction of viral load (242 copies mL ⁻¹ ) (Fig. 2 ).

Figure 2

Real-time RT-PCR detection of virus load in the small intestinal mucosa. Results of virus load in the intestine are expressed as means ± SD, ^aP < 0.01 and ^bP < 0.05 compared with the normal control group; ^cP < 0.01 and ^dP < 0.05 compared with the RV-infected group; ^eP < 0.01 and ^fP < 0.05 compared with the treated group at equivalent doses of LGG.

Detection of intestine sIgA and serum RV-specific IgA, IFN-γ andTNF-α

Compared with the immune responses from the normal control group, the secretion of intestinal mucosa sIgA, serum IFN-γ and TNF-α in the RV-infected group was increased, with the exception of serum IgA. After different concentrations of LGG were given to the RV-infected animals, the secretion of intestinal mucosa sIgA, serum IgA and IFN-γ increased gradually with increasing LGG doses, whereas the amount of TNF-α decreased gradually with increasing concentrations of LGG (Fig. 3 ).

Figure 3

Detection of intestinal mucosa sIgA, serum RV-specific antibody IgA, IFN-γ, TNF-α. (a) Detection of sIgA titres in the small intestinal mucosa ( n = 8). (b) Detection of blood serum RV-specific IgA antibody titres ( n = 8). (c) Detection of blood serum IFN-γ and TNF-α ( n = 8). Results are expressed as means ± SD, ^aP < 0.01 and ^bP < 0.05 compared with the normal control group; ^cP < 0.01 and ^dP < 0.05 compared with the RV-infected group; ^eP < 0.01 and ^fP < 0.05 compared with the treated group at equivalent doses of LGG.

Histopathological alterations of the jejunum

Transmission electron micrograph of jejunum

Compared with the normal control group, the jejunum microvilli of the RV infected group were sparse and had partly dropped off, and the smooth endoplasmic reticulum presented a typical vacuolization (Fig. 4b ). The acidophilic lymphocytes were found in the jejunal lamina propria of RV-infected mice (Fig. 4c ). By comparison, the jejunum microvilli of low, middle and high-LGG pretreated groups were almost regular, with no obvious pathological changes (data not shown). Plasma cells were found in the jejunal lamina propria of the H-LGG-pretreated group (Fig. 4d ).

Figure 4

Transmission electron micrograph of jejunum. (a) In the normal jejunum mucosa of the normal control group, the long and regularly distributed microvilli, dense cytoplasm and normal mitochondria are indicated by arrows. Magnification, ×15 000. (b) RV-infected group shows microvilli and mitochondrial swelling, endoplasmic reticulum oedema ( arrow ) and expended perinuclear space in the cell of small intestine. Magnification, ×45 000. (c) Acidophilic lymphocytes were found in the lamina propria of the intestinal wall of the jejunum in RV-infected mice ( arrow ). Magnification, ×45 000. (d) The plasma cells were found in the lamina propria of the jejunum mucosa of the H-LGG-pretreated group ( arrow ). Magnification, ×45 000.

Discussion

Human studies have suggested the clinical benefit of LGG against RV-infected diarrhoea (Isolauri et al. , 1991 ; Pant et al. , 1996 ; Guandalini et al. , 2000 ; Szajewska et al. , 2007 ; Ritchie et al. , 2010 ).

The LGG administration protocols of the previous studies in human were 1 × 10 ¹⁰ and 1 × 10 ⁹ CFU per day, respectively (Mastretta et al. , 2002 ; Costa-Ribeiro et al. , 2003 ). In the present study, compared with the doses of 2 × 10 ⁵ CFU (low) and 2 × 10 ⁷ CFU (medium) LGG four consecutive days after, or even before, RV infection, a high dose of 2 × 10 ⁹ CFU LGG before RV infection in RV-infected mice significantly shortened the duration of diarrhoea, decreased viral shedding of RV in faeces, increased the secretion of intestinal mucosa sIgA, raised production of serum IFN-γ, IgA, decreased serum TNF-α, and reduced epithelial vacuolation in the jejunum. Interestingly, the vacuolization of jejunum microvilli in RV-infected mice decreased gradually with increasing doses of LGG; in particular, at a dose of 2 × 10 ⁹ CFU before RV infection, no notable vacuolization or degeneration was observed in the jejunum (Fig. 5 ).

Figure 5

Histopathological changes of the jejunum of the normal control group (a), RV-infected group (b) and H-pretreated group (c). Haematoxylin/eosin staining of jejunum. (b) RV-infected group shows typical swollen and vacuolated, unstainable villus tips (arrows). (c) No apparent pathological changes can be observed in the jejunum of the H-pretreated group. There was no significant difference between the normal control group and H-pretreated group (a, c). Magnification, ×200.

Petschow's human study mentioned the therapeutic effect of the consecutive administration of different doses (1 × 10 ⁸ , 1 × 10 ⁹ and 1 × 10 ¹⁰ CFU per day) of LGG (Petschow et al. , 2005 ). Fang et al . ( 2009 ) suggested an effective dose of LGG against RV-induced diarrhoea of over 6 × 10 ⁸ CFU. Concerning the administration of LGG in clinical practices, the concept of ‘the higher the dose, the clearer the effect’ (Van Niel et al. , 2002 ) of this product is debatable. In the present study, of several administration models, the dose of 2 × 10 ⁹ CFU LGG was the most effective against RV-induced diarrhoea. It must be noted here that no significant difference was found in this study between the results from the doses of 2 × 10 ⁹ CFU and 2 × 10 ¹¹ CFU (data not shown).

To our knowledge, few studies have focused on the effects of different administration times of LGG against human RV-induced diarrhoea. Costa-Ribeiro indicated the beneficial effects of LGG would not occur until 2–3 days after initial administration (Costa-Ribeiro et al. , 2003 ), which is consistent with the present study. Compared with administration of LGG after RV infection, we found that a consecutive 4-day usage of LGG before RV infection may be the most effective against human RV-induced diarrhoea by stimulating the secretion of sIgA, raising production of serum IFN-γ, IgA, and decreasing serum TNF-α.

In this study, the pathological changes of the RV-infected group were quite different from those of the normal control group. The jejunum epithelial cells of RV-infected mice presented widespread cellular vacuolization and degeneration, consistent with previous studies (Majerowicz et al. , 1994 ; Danan et al. , 1998 ). The protective alterations of histopathology and immunohistochemistry may be closely related with these immune reactions due to the fact that LGG can increase mucosa sIgA and prevent intestinal epithelial surface from bacterial or viral adhesion, inhibit virus assembly and release, and reduce viral replication in the mucosa (Kaila et al. , 1995 ; Cunningham-Rundles, 2001 ; Isolauri, 2004 ; Feng et al. , 2006 ; Wen et al. , 2012 ). LGG can enhance the production of RV-specific antibody IgA (Isolauri et al. , 1995 ; Kaila et al. , 1995 ; Williams et al. , 1998 ), increase the function of antiviral activity and immune regulation of IFN-γ, stimulate CTL and Thl cells to produce IFN-γ (Sugata et al. , 2008 ), and may decrease the production of TNF-α directly or indirectly by means of suppressing a variety of pro-inflammatory cytokines (such as IL-6, IL-8) (Weninger & Andrian, 2003 ). Furthermore, LGG has been found to be capable of suppressing the activities of TNF-α (Zhang et al. , 2005 ; Donato et al. , 2010 ). Taken together, LGG may preserve the intestinal mucosa by suppressing the secretion and activities of TNF-α. In the present study, LGG administration 4 days before RV infection was found to be capable of initiating or activating the immune system of the host, which is in agreement with Costa-Ribeiro's observations (Costa-Ribeiro et al. , 2003 ). Finally, the serological and pathological results above showed that LGG administration before RV infection was more effective than treatment starting when symptoms appeared. In clinical practice, treatment before the appearance of symptoms is very difficult. Fortunately, the results from this study also confirmed that LGG treatment after symptoms of RV-induced diarrhoea was almost as effective as prophylactic treatment.

In summary, the protective effects of administering LGG are better before than after RV infection, and the dose of 2 × 10 ⁹ CFU for four consecutive days before RV infection extensively protected the neonatal mice against human RV-induced diarrhoea. Further clinical studies are needed to determine the timing and dosage of administration of LGG against human RV-induced diarrhoea in children.

Acknowledgements

We would like to thank Dr Xiaosong Song (Graduate School at Shenzhen, Tsinghua University) for help in revising the manuscript and giving some useful suggestions in this study. We also thank the Shenzhen Bureau of Science Technology & Information (grant JC200903180532A to ZY), RFDP (20090002120055 to ZY), Nanshan district Bureau of Science Technology, and Wuhan Bureau of Science Technology foundation (grant: 201050231036) for the financial funding. We declare that we have no conflicting interests.