Rotavirus-specific T-cell responses in young prospectively followed-up children

Abstract

Rotavirus is a major cause of gastroenteritis in young children. Antibodies seem to protect against rotavirus infection but cell-mediated immune responses are probably also important for protection. We evaluated the development of T-cell responses to rotavirus in follow-up samples from 20 healthy children with an increased genetic risk for type 1 diabetes. Blood samples from 16 healthy adults were also available for the study. T-cell proliferation was analysed at 3–6 month intervals from the age of 3 months to the age of 4–5 years using the Wa strain of human rotavirus and the NCDV strain of bovine rotavirus as antigens. IgG and IgA antibodies to rotavirus were studied from simultaneously drawn plasma samples with EIA method using NCDV as an antigen. A total of 24 infections were revealed by antibody analysis. Sixteen children showed diagnostic increases in both IgG and IgA antibodies to rotavirus, while 5 children showed increases in IgA antibodies only and 3 in IgG only. Antibody rises were accompanied by T-cell responses to rotavirus (SI > 3) in 9 of the 24 cases. T-cell responses to purified or lysed human rotavirus were stronger after a rise in rotavirus antibodies than the responses before infection (P = 0·017 and 0·027, respectively). There was a correlation between T-cell responses to purified and lysed human rotavirus and NCDV. Strong T-cell responses to rotavirus were transient and the ability to respond usually disappeared in one year, but in all adults T-cell responses to rotavirus were strong implicating that several infections are needed to develop consistent, strong T-cell responsiveness.

INTRODUCTION

Group A rotavirus infections are common worldwide among infants and young children. Re-infections occur in persons of all ages, but the symptoms of rotavirus infection are less severe in adults and subclinical infections are common [1].

Rotavirus replicates in mature villus enterocytes in the small intestine leading to diarrhoea, vomiting and dehydration. In immunocompetent persons, rotavirus infections are limited in the gut. Nevertheless, rotavirus also induces both local and systemic immune responses. The data from adult purposefully infected volunteers [2,3] and subjects suffering from natural infection [4,5] indicate that serum antibodies are valid markers of protection from rotavirus infection. Primary rotavirus infection induces production of mainly serotype-specific antibodies, but reinfections create a broader immune response including production of cross-reactive heterotypic antibodies [6–8].

In 1988, Totterdell et al. [9] were the first to utilize a lymphoproliferative assay to test T-cell responses to rotavirus, when they observed cellular responses to rotavirus in various groups of adults. Offit et al. [10] then evaluated rotavirus specific T-cell responses in children as well as in adults confirming that most children had rotavirus-specific lymphoproliferative activity against at least 1 of the 3 studied rotavirus strains by age of 2 years. The number of recognized serotypes increased with age. Using a series of 3 consecutive samples, Offit et al. [11] also demonstrated rotavirus-specific T-cell responses in 7 of 8 young children during convalescence after a primary infection. Also Johansen et al. [12] have reported T-cell responses to recombinant NSP4 antigen of SA11 rotavirus in 4/10 healthy Swedish adults.

We have now examined the development of rotavirus specific T-cell responses and IgA and IgG antibodies in repeated samples from 20 prospectively followed-up children. Blood samples were obtained at 3 or 6 month intervals between the ages of 3 months and 4–5 years. We specifically wanted to determine how tightly lymphoproliferative responses against rotavirus antigens occur in association with an increase in rotavirus antibody concentration.

MATERIALS AND METHODS

Study design and subjects

Twenty healthy children (15 boys) with increased genetic risk for type 1 diabetes (HLA-DQB1*02/*0302) were recruited from the Diabetes Prediction and Prevention (DIPP) study [13]. Blood samples were collected at 3 month intervals from the age of 3 months to the age of 2 years and subsequently at 6 month intervals. For comparison, single blood samples were obtained from 16 healthy laboratory workers. All children were vaccinated according to the standard Finnish vaccination protocol in which BCG immunization is given to infants at a few days of age and DPT vaccination (Diphtheria, Pertussis and Tetanus) at ages 3, 4, 5 and 20–24 months. All plasma samples were negative for islet cell antibodies.

Antigens

The Nebraska Calf Diarrhoea (NCD) strain (G serotype 6, P serotype 6) of bovine rotavirus was propagated in LLC-MK2 cells as reported [14]. The Wa strain (G serotype 1 A, P serotype 1) of human rotavirus was kindly donated by Dr L. Maunula (University of Helsinki, Helsinki, Finland). For virus propagation, the Wa strain was pretreated with trypsin (10 µg/ml) and inoculated in fetal green monkey kidney cells (MA104) in MEM medium supplemented with 10 mm HEPES, glutamin and gentamycin (10 µg/ml). After adsorption, basal medium containing trypsin (1 µg/ml) was added. The cells were harvested when the virus had caused an advanced cytopathic effect, and the cells and supernatant fluid were frozen and thawed three times to release virus particles. Control antigen was prepared identically from uninfected MA104 cells. The protein concentrations of antigens were determined with Pierce BCA protein assay reagent (Pierce, Rockford, USA).

The preparation of purified Wa rotavirus was made by sucrose gradient centrifugation with a method modified from that described by Rott and coworkers [15]. The infected cells were first centrifugated at 10 000 g for 20 min. Polyethyl glycol 6000 and NaCl were added to 7% and 2·2% concentration, respectively, and the mixture was stirred at 4°C overnight. The precipitated virus was then retrieved by centrifugation (11 000 g for 20 min) and suspended in R buffer (10 mm Tris-hydrochloride, 0·2 m NaCl, 50 mm MgCl₂, 10% glycerol). Sodium deoxylate and Nonidet P-40 were added to 0·3% and 0·6%, respectively, and after 30 min at 4°C, the suspension was centrifugated at 1 200 g for 5 min. The virus was then ultracentifugated through a 30% sucrose cushion at 180 000 g during 2 h. The bands were collected, diluted in R buffer and the virus was pelleted by centrifugation at 120 000 g for 2·5 h. Virus pellets were resuspended in phosphate-buffered saline (PBS) and stored at − 70°C.

Lymphocyte proliferation assay

Peripheral blood mononuclear cells (PBMC) from followed-up children and adult controls were isolated from heparinized blood by Ficoll-Paque gradients (Pharmacia, Uppsala, Sweden). The cells were washed twice with RPMI 1640 supplemented with gentamycin (10 µg/ml) and resuspended in RPMI 1640 medium containing 7·5% human AB serum (Finnish Red Cross, Helsinki, Finland), glutamine, HEPES and gentamycin (10 µg/ml). They were stored at − 135°C in 10% dimethyl sulphoxide (DMSO) (Merck, Darmstadt, Germany). The cells collected from each child at different time points were always cultured in the same experiment. The cells (5 × 10⁴ cells in a final volume of 200 µl of complete medium) were incubated in quadruplicate with antigens in 96-well round-bottomed microtitre plates. After 6 days incubation at 37°C in 5% CO₂, the cells were labelled with tritiated thymidine (2 µCi/ml, Amersham, UK) for 18 h. The cells were harvested on glass fibre filters with Tomtec 93 Mach Manual Harvester (Tomtec, Orange, USA) and the incorporated radioactivity was measured with Micro-Beta scintillation counter (Wallac, Turku, Finland). Stimulation indices (SI) were calculated by dividing median experimental counts per minute (cpm) by the median control cpm. SIs ≥ 3 were regarded as positive.

To optimize the lymphoproliferative assay, lymphocytes were incubated with rotavirus antigens in concentrations of 0·1, 1 and 10 µg/ml. The concentrations giving highest stimulation indices were used to test rotavirus-specific T-cell responses. Purified human rotavirus was used at 0·1 and 1 µg/ml concentrations, human rotavirus lysate and uninfected MA104 cell lysate at 1 µg/ml concentration and bovine rotavirus and uninfected LLC cell lysates at 10 µg/ml concentration. Tetanus toxoid (TT; 1 µg/ml; National Public Health Institute, Helsinki, Finland) and PPD (2,5 µg/ml; Staatens Seruminstitut, Copenhagen, Denmark) were used as control antigens. Phytohaemagglutinin (100 µg/ml; Difco, Detroit, USA) was used as a mitogen control.

Rotavirus antibodies

Microtitre plates (Nunc Immunoplate, Nunc, Roskilde, Denmark) were coated by incubating them with NCD virus (1·0 µg/well) in a carbonate buffer (1·6% Na₂CO₃, 2·9% NaHCO₃, 0·2% NaN₃) at room temperature overnight. For detection of rotavirus-specific IgG antibodies, the plasma samples were diluted 1 : 200 in PBS +0·5% BSA +0·5% Tween 20 and incubated for 2 h at +37°C. After washes, peroxidase conjugated anti-human IgG antibody (Dako, Copenhagen, Denmark) was added in 1 : 1000 dilution and the plates were incubated for 1 h. O-phenylenediamine tablets were used as substrate. Each specimen was tested in duplicate. Positive and negative standards were included on each plate in four different concentrations. For detection of rotavirus-specific IgA antibodies, the plates were first residual-coated with PBS +1% BSA. The samples were diluted 1 : 10 in PBS + 0·2% BSA +0·05% Tween 20 and incubated for 2 h at +37°C. The plates were washed and biotinylated anti-human IgA antibody (Vector Laboratories, Burlingame, USA) was added in 1 : 100 dilution. After an 90 min incubation and washes, alkaline phosphatase-streptavidin (Zymed, San Francisco, USA) was used as a secondary conjugate in 1 : 1000 dilution. The plates were incubated for 1 h, washed and P-nitrophenyl phosphate substrate tablets (Sigma, St Louis, USA) were added in a carbonate buffer (0·05 m NaHCO₃, 0·05 m Na₂CO₃, 0·02% MgCl₂, 0·02% NaN₃). The colour intensities were measured with spectrophometer. An antibody response was defined as a ≥2-fold increase in absorbance between consecutive specimens. Also a ≥3-fold absorbance compared to negative control specimen was considered positive. All positive samples had to exceed the cut-off level of seropositivity (0·150 optical density units).

Statistical analysis

Mann–Whitney U-test was used in comparison of two groups. Differences between paired results were compared with Wilcoxon test. Frequencies of positive T-cell responses between two groups were compared with Fisher's exact test. Correlations between different parameters were calculated with Spearman correlation test. P-values <0·05 were considered significant.

Ethics

Approval for the study was obtained from the Ethics Committee of Turku University Hospital. Informed consent was obtained from the parents of each study subject before participation.

RESULTS

All children had received the standard Finnish vaccination program including BCG at birth and tetanus toxoid at ages 3, 4, 5, and 20–24 months. Responsiveness to either PPD or tetanus was used as a quality control of cryo-preserved cells and the cultures without apparent (SI ≥ 3) responses to these antigens were excluded from the analysis. T-cell responses to uninfected cell lysates were never observed. A correlation was observed between T-cell responses to the human Wa strain and the bovine NCDV strain of rotavirus. (Fig. 1).

Fig. 1.

T-cell responses to different rotavirus antigens. Nebraska calf diarrhoea virus (NCDV), human Wa rotavirus lysate (RV) (a) and purified human Wa rotavirus (PRV)(b). T-cell responses to human Wa rotavirus lysate (P < 0·0001, r_s = 0·52, Spearman correlation test) and purified Wa rotavirus (P < 0·0001, r_s = 0·56) correlated with responses to bovine NCD virus (number of samples analysed in all three assays = 193) although the responses to human rotavirus were stronger than to bovine rotavirus.

The study subjects (n = 20) were selected because they had available lymphocyte samples and preliminary antibody screening revealed clear patterns associated with rotavirus infections. Maternal rotavirus IgG antibodies were detected in 12 of the 16 available samples taken at 3 months of age. The antibody titres then decreased reaching a nadir at 6 months of age. Two children had rotavirus IgA antibodies at 3 months and increasing IgG levels between 3 and 6 months of age, indicating a recent rotavirus infection. Of the children selected for the study, one had no serologically documented rotavirus infections during the follow-up, 14 experienced one rotavirus infection and 5 children had two or more rotavirus infections during the follow-up. Sixteen children showed diagnostic increases in both IgG and IgA antibodies while 5 children showed increases in IgA antibodies only. Interestingly, 4 of 5 infections followed by an IgA antibody response only occurred in the children before the age of one year. Three apparent re-infections were characterized by a response in IgG only. Eleven of the 12 control adults from whom plasma samples were available had IgG antibodies to rotavirus (i.e. ≥3-fold absorbance compared to negative control sample) but only 6 of them had IgA antibodies when the same criteria for antibody positivity were used.

T-cell responses to cell lysate antigens and purified virus were minimal or absent in children with low rotavirus antibody titres, but when antibody titres had shown an increase, also the T-cell responses became markedly stronger (P = 0·017 and 0·027, respectively, Wilcoxon test) (Fig. 2). The difference in T-cell responses to NCDV was nonsignificant (P = 0·11). The increases in antibody concentrations were accompanied by T-cell responses to rotavirus (SI ≥ 3) in 9 of the 24 cases. However, one child had two strong rotavirus specific T-cell response peaks without simultaneous increases in rotavirus antibodies suggesting that two reinfections may have occurred (Fig. 3c) and the antibody assay has not been sensitive enough to detect it. It has been demonstrated that e.g. copro IgA assay is more sensitive to detect reinfections than serological assays [16]. With increasing age, T-cell responses became slightly more common (n.s., Fisher's exact test) (Fig. 4), and also reinfections (n = 5) tended to create a T-cell response more often than primary infections (n = 19) (n.s) (Fig. 2).

Fig. 2.

T-cell responses to rotavirus lysate before and after serologically demonstrated (a) primary or (b) secondary rotavirus infections. T-cell responses to rotavirus lysate were stronger after an increase in rotavirus antibodies than before the antibody increase, but only a part of infections were associated with a T-cell response to rotavirus.

Fig. 3.

Serological and cellular responses to rotavirus in three children. (a, b) This child had concomitant rises in rotavirus IgG and IgA antibodies and he also showed a cellular response to rotavirus antigens between the ages of 15–18 months. (c, d) This child had maternal IgG and a rise in IgA antibodies at the age of 6 months without an associated cellular response. However, a rise in rotavirus antibodies at the age of 27 months was accompanied by a cellular response to rotavirus. (e, f) IgG antibody titres increased in this child at the age of 12 months and remained elevated thereafter. Cellular responses suggested that the child had experienced several rotavirus infections, at the ages of 12, 18–21 and 48 months. (a, c, e) Rotavirus ○ IgG, • IgA; (b, d, f) ○ NCDV, • RV, □ PRV4, ▪ PPD.

Fig. 4.

T-cell responses to rotavirus antigens. (a) Nebraska calf diarrhoea virus (NCDV), (b) human rotavirus lysate (RV), (c) purified human rotavirus (PRV) and (d) positive control antigen PPD in different age groups of children (3–6 months of age, n = 23; 9–12, n = 26; 15–24, n = 65; 27–36, n = 38; 39–48, n = 31; 51–60, n = 11) and in control adults (n = 16). Median value in each group is shown with the horizontal line in the box. The boxes delineate values between the 25th and 75th percentiles, while values shown with the open circles are values outside this range. T-cell responses to rotavirus were in general weak in the children, while the adult control subjects had stronger T-cell responses to NCDV (P = 0·0001–0·0067, Mann–Whitney U-test), human rotavirus lysate (P = 0·0008–0·011) and purified human rotavirus (P = 0·0044–0·083) than any age group of children. The tendency of weaker proliferation responses in the age group of 51–60 months is probably due to the small sample size in this age group. T-cell responses to PPD were strong also in children.

Titres of rotavirus-specific IgG and IgA antibodies remained elevated during the whole follow-up time after a rotavirus infection. However, T-cell responses to rotavirus began to decline shortly after infection and were normally detectable less than 12 months after a primary infection (mean 5 months) (Fig. 3). T-cell responses to rotavirus were in general weaker in children than in the adult control subjects (Fig. 4). The adults had markedly stronger T-cell responses to NCDV (P = 0·0001–0·0067, Mann–Whitney U-test), human rotavirus lysate (P = 0·0008–0·011) and purified human rotavirus (P = 0·0044–0·083) than any age group of children, but the differences between T-cell responses to PPD were nonsignificant in the different age groups (P = 0·53–0·91).

DISCUSSION

This study shows that rotavirus–specific lymphoproliferative responses occur in tight association with increases in rotavirus antibody titre in prospectively followed-up children. We used the Wa strain of human rotavirus and the NCD strain of bovine rotavirus as antigens in the T-cell proliferation assay. The Wa strain belongs to the most prevalent human rotavirus serotype, G1, which causes over 50% of rotavirus infections throughout the world [1], while the NCD virus belongs to the serotype G6 viruses which rarely infect humans. Amino acid sequences of different serotypes vary markedly between the strains, but heterotypic antibodies are commonly detected after a rotavirus infection [6–8]. In a mouse model, the cross-reactive T-helper cell epitopes have been mapped to the rotavirus VP6 protein [17,18]. In our study, serologically confirmed rotavirus infections augmented T-cell responses to the human Wa rotavirus and the NCD virus. Despite differences in the amino acid sequences in the major antigenic epitope, there was a correlation between T-cell responses to these two viruses. Yasukawa et al. [19] previously observed cross-reactive lymphoproliferative responses to human Wa rotavirus and bovine NCD virus in two healthy adults, while Offit et al. [10] found that most adults and children older than 5 years of age had lymphoproliferative responses against all 3 rotavirus strains tested. Interestingly, in our study strong T-cell responses to NCD virus were often observed also in young children.

We found that rotavirus-specific T-cell responses declined briefly after a serologically confirmed rotavirus infection. This pattern was observed also after secondary infections. In a previous study [11] rotavirus-specific T-cell responses were detected 2–8 weeks after a rotavirus infection in 7 of 8 children. All four children who were re-examined 3–5 months later had persistent T-cell responses to rotavirus. The tight follow-up schedule in our study allowed the analysis of responses at short time intervals, as previous data about the persistence of rotavirus specific T-cell responses are sparse. As most of rotavirus specific lymphocytes express gut associated α₄β₇ homing receptor [15], it is possible that resting rotavirus-specific memory cells rarely circulate but are mostly homed in the gut. Nevertheless, we found strong rotavirus specific lymphoproliferative responses in healthy adults suggesting that numerous reinfections may lead to the development of persisting cell-mediated rotavirus response. In agreement with our results, Jaimes et al. [20] have found rotavirus-specific interferon gamma-secreting T-cells in symptomatically infected adults and rotavirus-exposed laboratory workers but not in children with rotavirus diarrhoea.

All children in our study carry either the HLA-DQB1*02 or *0302 alleles associated with type 1 diabetes risk and no one have protective DQB1*0301 or *0602 alleles. If responsiveness to rotavirus is strongly affected by certain HLA alleles this might have an effect to the results. The study subjects still have a large panel of various HLA-haplotypes and the results are probably applicable to general population as well. Viral infections have been implicated as triggers of type 1 diabetes [21] and the possible role of rotavirus infections in development of diabetes is of interest [22]. The present study is focused on cell-mediated immune responses to rotavirus in general in children and will serve as a background for further studies elucidating possible association of cellular rotavirus responses and diabetes-related autoimmunity.

In summary, we investigated rotavirus-specific lymphoproliferative activity in a series of follow-up samples drawn from 20 young children and in single samples drawn from healthy adult controls. Lymphoproliferative responses to rotavirus occurred simultaneously with increases in antibody titres in 9 out of 24 cases. In young children, T-cell responses to rotavirus were weak and declined shortly after the infection. In adults, strong T-cell responses were observed, probably due to several previous reinfections. Our data suggest that rotavirus infections induce systemic lymphoproliferative responses in young children. However, strong and consistent cellular responsiveness to rotavirus develops only after several rotavirus infections.