CLINICAL AND DIAGNOSTIC LABORATORY IMMUNOLOGY, Nov. 1998, p. 897 901 Vol. 5, No. 6 1071-412X/98/$04.00 0 Copyright 1998, American Society for Microbiology. All Rights Reserved. Longitudinal Studies of Neutralizing Antibody Responses to Rotavirus in Stools and Sera of Children following Severe Rotavirus Gastroenteritis BARBARA S. COULSON* Department of Gastroenterology and Clinical Nutrition, The Royal Children s Hospital, Parkville 3052, Victoria, Australia Received 23 April 1998/Returned for modification 21 July 1998/Accepted 10 August 1998 Rotavirus-neutralizing antibody responses in sera and stools of children hospitalized with rotavirus gastroenteritis and then monitored longitudinally were optimally detected by using local rotavirus strains. Stool responses were highest on days 5 to 8 after the onset of diarrhea. Longitudinal monitoring suggested that serum neutralizing antibody responses were a more useful measure of severely symptomatic rotavirus infection than stool responses but that stool antibody responses may be a useful measure of rotavirus immunity. Rotaviruses are recognized as the major cause of severe acute infantile gastroenteritis. Based on the viral outer capsid proteins VP4 and VP7, which independently elicit virus-neutralizing, protective antibodies, a dual serotyping system for rotavirus has been adopted. VP4 genotypes, which were identified on the basis of sequence differences and which, when tested, correlate with serotypes, have also been designated. Human rotaviruses contain at least eight VP4 (P) genotypes and at least nine VP7 (G) serotypes, the most common of which are P[4], [6], and [8] and G1 to 4. Immunity to rotavirus infection in children has been shown to correlate with serum (15) and intestinal or stool antibodies (5) to rotavirus, but titers of serotype-specific, heterotypic, and neutralizing serum antibodies and isotype-specific antibodies in serum and intestine or stools cannot be used reliably as markers of protection against subsequent illness (15). The contribution of neutralizing coproantibodies (fecal antibodies) to immunity in children requires more study, particularly as serological immune correlates of protection have not been identified for design and evaluation of effective rotavirus vaccines, and intestinal antibody responses have not yet been measured during vaccine trials (16). Intestinal immunoglobulin A (IgA) to rotavirus has been shown to be the most-sensitive marker of rotavirus infection (6), and fecal antirotaviral IgA levels can be used to predict the presence of duodenal IgA (14). Fecal IgA coproconversions correlate with fecal rotavirus-neutralizing antibody conversions (8). Coproconversions in rotavirus-neutralizing IgA are more-sensitive indicators of rotavirus infection and reinfection than seroconversion in IgG, IgM, IgA, or neutralizing antibodies, and persistent elevations in stool rotavirus-neutralizing IgA (termed coproiga plateaus) correlate with protection against reinfection and symptomatic illness in young children (5). In a small number of children, the serotype specificity of the stool rotavirus-neutralizing IgA responses has been studied (6, 8). However, it is not known whether the P or G serotype specificity of these responses parallels the specificity of the rotavirus-neutralizing responses in serum following severe rotavirus gastroenteritis and rotavirus reinfection. The duration * Mailing address: Dept. of Microbiology and Immunology, the University of Melbourne, Parkville 3052, Victoria, Australia. Phone: 61 3 9344 8823. Fax: 61 3 9347 1540. E-mail: b.coulson@microbiology.unimelb.edu.au. 897 of neutralizing coproantibody excretion in stools following rotavirus infection is not known either. The aim of this study was to compare the nature and duration of rotavirus-neutralizing antibody responses in sera and stools of children during the acute and convalescent phases of severe rotavirus gastroenteritis and during at least 5 months of longitudinal monitoring thereafter. The children studied were admitted to the infectious diseases ward of the Royal Children s Hospital, Melbourne, Australia, between April 1984 and September 1985 with acute rotavirus gastroenteritis diagnosed on clinical grounds and in the laboratory by the presence of rotavirus by electron microscopic examination of stool extracts and/or by the presence of viral antigen in stools detected by enzyme immunoassay (EIA). The 15 children studied, 2 to 39 months old at recruitment, were a subset of the 44 children recruited at this time for longitudinal study of rotavirus infection and immune responses. This subset was selected from the first 24 children from whom complete sets of samples were obtained and was chosen to contain similar numbers of children infected with G1 and G4 rotavirus. The clinical, demographic, and laboratory findings for these 44 children have already been described (5, 6, 14). Prior to enrollment, parents were provided with a detailed explanation of the study (including the need to obtain blood samples from the infants), and they gave their signed consent. The study was approved by the Human Ethics Committee of the Royal Children s Hospital. Titers of neutralizing antibody were measured in sera collected in the acute and convalescent phases and at 4-month intervals post-onset of diarrhea, in fecal specimens collected daily while the child was in the hospital, and in stools collected at 7- to 10-day intervals for 219 to 721 days from the onset of severe rotavirus gastroenteritis. Stools collected by parents at home were stored frozen at 4 C for up to 1 month before transport to the Royal Children s Hospital (14). Feces and sera were stored at 70 C until tested. Rotavirus-neutralizing antibodies were measured by fluorescent focus reduction neutralization assay (FFN) with MA104 cells as described previously (6, 8). Samples were titrated against cell culture-adapted human rotavirus strains RV-4, Wa and Ku (P[8], G1), (), (), ST-3 (), and VA70 (P[8], G4). RV-4 and were isolated from stools of Melbourne children with rotavirus gastroenteritis, whereas was obtained from an asymptomatically infected Melbourne neonate (). Strains Wa, Ku, and VA70 were obtained from chil-
898 NOTES CLIN. DIAGN. LAB. IMMUNOL. TABLE 1. Neutralizing antibody responses to human rotaviruses (G types 1 to 4) in acute-phase and convalescent-phase sera and stools in 15 children with acute severe gastroenteritis caused by known virus serotypes and monotypes Infecting serotype and monotype a Sample Patient no. b (age in mo) Day(s) stools collected c P[8], G1a RV-4 P[8], G1a/c Wa FFN response to indicated virus P[8], G1c Ku ST-3 G1a P[8] Serum 21 d (2) c e f c c Stools 9, 16, 23 c c G1c Serum 12 (3) g ND h Stools 8, 11 G1c Serum 14 (10) c c c Stools 5, 9, 11 G1a Serum 23 (11) c ND Stools 8, 12, 15 G1c Serum 31 (16) c c ND Stools 4, 5, 6 c c c G1c Serum 1 (21) c c c c c c ND Stools 9, 15, 22 G1c P[8] Serum 33 (23) c c c c ND Stools 3, 6, 7, 9 c c c G1a Serum 9 d (27) c c c c ND Stools 2, 5 9 c c ND c c c ND G1c Serum 15 (39) c c c c c ND Stools 4 G4 P[8] Serum 27 (12) c c c c Stools 7, 8, 15 ND ND G4 Serum 2 (20) c c c c c c Stools 15, 22, 28 ND G4 Serum 17 (21) c c c c c c c Stools 9 c c c c ND G4 P[8] Serum 24 (24) c c c c c c c Stools 4 8 c c c c c c c G4 P[8] Serum 4 d (24) c c c c c Stools 13, 20 c G4 Serum 36 (24) c Stools 4, 5 c P[8], G4 VA70 a Monotypes are indicated with the letters a and c. b Patients and their numbers correspond to those studied previously (4). c All stools collected in the first 28 days after onset of symptoms were tested. Only those with positive FFN titers are shown, except when no stool showed an FFN-positive result. d All acute-phase serum FFN titers were negative ( 200). e c, FFN antibody conversion (increase in FFN titers by at least fourfold). f, FFN titer of 200. g, FFN antibody present at stationary or reducing titers or FFN titer between 200 and 800. h ND, not done. dren hospitalized with gastroenteritis in the United States, Japan, and Italy, respectively. ST-3 was isolated from an asymptomatically infected neonate in the United Kingdom. The origins and sources of these rotaviruses have been reported previously (3, 7, 9). All virus strains were propagated in MA104 cells in the presence of trypsin (9). Fourfold dilutions of sera, starting at 1:100, were incubated with each trypsinactivated, cultivated rotavirus strain. The neutralization titer of each sample was expressed as the reciprocal dilution giving 50% reduction in the number of fluorescing cells. Fecal extracts (n 40) containing no antirotaviral IgA, IgM, or IgG by EIA all gave reciprocal titers of 200 by FFN. Stool samples with titers of 200 were therefore considered to be negative for neutralizing antibody to the rotavirus strain tested. An immune conversion in neutralizing antibody in sera or stools was considered to be a fourfold increase in reciprocal titer to at least 400 (6). Stool rotavirus antigen, VP7 serotypes, and VP7 monotypes (classified on the basis of antigenic variation within a VP7 serotype) were determined with rotavirus-specific monoclonal
VOL. 5, 1998 NOTES 899 TABLE 2. Geometric mean FFN titers to rotavirus in acute-phase and convalescent-phase sera and stools collected from children infected with G1 or G4 rotavirus Sample type and patient group a GMT to indicated virus (no. of samples tested if more than one) RV-4 Wa Ku ST-3 VA70 b G1 serum A 274 c 397 d 247 e 112 123 188 c 100 C 2,576 1,125 c 1,175 251 283 585 f 1,026 G1 stool A 795 (9) 189 (9) 203 (9) 153 (9) 378 (9) 133 (9) 197 (9) C 133 (9) 100 (8) 100 (7) 100 (8) 133 (8) 100 (8) 100 (8) G4 serum A 148 c 119 d 100 e 166 144 329 c 548 C 2,377 757 c 865 316 314 1,714 f 1,897 G4 stool A 827 (6) 245 (6) 383 (6) 233 (6) 790 (6) 284 (5) 195 (5) C 224 (4) 100 (4) 136 (4) 100 (4) 316 (4) 126 (3) 134 (3) a A, acute phase; C, convalescent phase. Acute-phase and convalescent-phase stools were collected days 2 to 17 and 17 to 28 post onset of diarrhea, respectively. b Titers were not statistically analyzed for VA70, as stools from only two patients in the G1 group were tested. For all other viruses, all nine patients in the G1 group and all six patients in the G4 group were tested. c No significant differences in serum titer to the given rotavirus strain between the G1 and G4 groups of children (P 0.05 by Student s t test). d P is 0.025 by Student s t test. e P is 0.005 by Student s t test. f P is 0.05 by Student s t test. antibodies in antigen-capture EIAs as described previously (4, 10). For a few patients, sufficient stool sample was available for determination of VP4 genotype by reverse transcription-pcr amplification of viral RNA with nested primers (11). VP4 (P) genotypes were determined to confirm that they corresponded to the genotype expected of G1 and G4 rotaviruses causing gastroenteritis in children, i.e., P[8]. All genotypes were P[8] and are listed in Table 1. All children infected with either G1 or G4 rotavirus showed serum neutralizing antibody responses to G1 and G4 rotaviruses (Table 1). Compared with serum responses, neutralizing antibody was found less often in stools, and fewer children showed antibody conversions to fewer rotavirus strains. All of the children with any detectable stool antibody response (80%) produced neutralizing antibody to G1 rotavirus RV-4. A broadening of these responses to include additional virus serotypes and strains was observed with increased age of the children at the time of illness. For example, only one child aged 12 months or less had serum neutralizing antibody to the P[4], G2 virus, and none had antibodies neutralizing the P[6], G3 virus. Nine of the ten children aged 16 to 39 months had serum antibody to at least one of these two heterotypic viruses. Most children seroconverted and coproconverted (or showed positive coproantibody levels) to P[8], G1 rotavirus(es). Many showed seroconversion to P[6] or [8], G4 rotaviruses (80%) and coproconversion to virus (67%) and P[6] or P[8], G4 viruses (47%). Fewer children seroconverted to the and strains (33%) or coproconverted to virus (13%). As neutralizing antibody titers of children infected with G1a and G1c virus strains were similar, comparisons of the geometric mean titers (GMT) of serum and fecal neutralizing antibody in children infected with G1 or G4 rotavirus could be made (Table 2). The GMT to G1 and G4 viruses in convalescent-phase sera were at least twice the GMT to and viruses. Titers to G1 viruses Wa and Ku were significantly higher in acute-phase sera in G1 than G4 rotavirus infections, whereas G4 infections stimulated significantly higher levels of neutralizing antibody to ST-3 than did G1 infections. The highest GMT of serum neutralizing antibody were directed to local P[8], G1 virus RV-4, followed by P[8], G1 viruses Ku and Wa, in the case of P[8], G1-infected children, and by P[8], G4 virus in the case of P[8], G4 virusinfected children. The children infected with G4 rotavirus showed slightly higher GMT of neutralizing coproantibodies to all strains tested than did the G1 virus-infected children, although these differences were not significant. The highest titers observed were to the local P[8], G1 virus RV-4 and the local virus,, irrespective of the infecting rotavirus serotype. These results show that during the acute and convalescent phases of severe primary G1 or G4 rotavirus gastroenteritis, neutralizing antibody responses were directed to viruses of the infecting G type in both serum and stools. However, significant responses were also directed to G1 viruses, in the case of G4-infected children, and G4 viruses, in the case of G1-infected children, as has been reported previously for serum responses alone (12). Thus, the responses in serum and stools during and following primary P[8], G1 and P[8], G4 rotavirus gastroenteritis were heterotypic for P and G serotypes and G monotypes and similar between sites. The duration of excretion of virus-neutralizing coproantibodies after severe rotavirus gastroenteritis was determined for the first time in this study. As shown in Fig. 1, stools collected between 5 and 8 days after the onset of diarrhea were the most likely to contain neutralizing coproantibodies (70%). In contrast, only 31 (56%) of those collected on days 9 to 16 contained neutralizing coproantibody and 6 (26%) of stools collected on days 17 to 28 contained this antibody. Thus, in vaccine trials and further studies of natural infection, more limited acute- and convalescent-phase stool sampling than was done here is still likely to be informative. The FFN coproantibody and serum antibody profiles of the children followed from 120 days to at least 391 days post-onset of severe rotavirus gastroenteritis are presented in Tables 3
900 NOTES CLIN. DIAGN. LAB. IMMUNOL. FIG. 1. Frequency of detection of rotavirus-neutralizing antibodies in stools relative to the timing of stool collection after the onset of rotavirus gastroenteritis. Patient no. a (G type of infecting virus) and 4. As assessed by seroconversion in antirotaviral IgG, IgA, and IgM and neutralizing antibodies, coproconversion in IgA and neutralizing antibodies, and rotaviral antigen excretion in stools, these children had one to four rotavirus reinfections during this period (5, 6). All sera collected 120 days after onset contained neutralizing antibodies to P[8], G1 virus Wa, and most contained antibodies able to neutralize,, and rotaviruses (Table 4). In the children originally infected with G1 rotavirus, a higher percentage of sera (78%) than stools (10%) contained G4-neutralizing antibody. Children with P[8], G1 or P[8], G4 virus as the cause of their severe gastroenteritis showed similar percentages of stools containing P[8], G1,, and -neutralizing antibodies. However, children with severe gastroenteritis caused by P[8], G4 rotavirus showed a significantly higher proportion of stools (74%) containing (ST-3)-neutralizing antibody (P 0.01) than did children initially enrolled with P[8], G1 virus-associated gastroenteritis (10%). The prevailing rotavirus serotypes during this period in this population were G1 (common) and G4 (rare) in the first year and G1 in the second year (2). Thus, although it was not possible to type the reinfecting virus, the children monitored longitudinally were likely to have been reinfected with G1 rotavirus. This suggests that the concept of original antigenic sin, whereby prior exposure to one influenza virus strain is able to divert the antibody response to a second challenging virus strain to focus on the shared (cross-reactive) epitopes (13), may apply to the rotavirus-neutralizing coproantibody responses of these children. Further studies of these antibodies during natural infection and in vaccine trials are needed to resolve this question. As was seen in stools collected within 28 days of the onset of severe gastroenteritis (Table 1), stools collected at least 120 TABLE 3. Frequency of detection of rotavirus-neutralizing antibodies in stools collected from the study children 120 days after the onset of severe rotavirus gastroenteritis No. of stools tested Period studied (days postonset) RV-4 P[8], G1a Stools containing neutralizing antibody to indicated virus (%) Wa P[8], G1a/c Ku P[8], G1c ST-3 b 21 (1) 11 464 698 100 46 27 0 91 27 ND 12 (1) 27 181 583 59 26 19 7 63 4 ND 14 (1) 26 124 618 31 4 0 4 39 8 0 31 (1) 39 121 645 78 51 49 21 74 10 ND 33 (1) 41 127 721 56 59 54 10 51 7 ND 37 (1) 20 125 219 35 5 0 5 45 20 ND VA70 P[8], G4 Total (1) 164 All c 58 35 30 10 59 10 ND 27 (4) 12 121 503 83 25 42 25 100 91 82 24 (4) 14 126 661 14 86 79 21 43 93 100 4 (4) 14 196 474 93 43 57 43 93 71 71 36 (4) 13 124 391 31 8 8 0 85 39 23 Total (4) 53 All 55 42 47 23 79 74 70 a Patients ages at onset are listed in Table 1. Patient 37 was infected with G serotype 1b rotavirus at 15 months of age. b P is 0.01 by t test, for proportions of stools containing ST-3 neutralizing antibodies, in G1 and G4 virus-infected children. c All, summation of all the periods studied.
VOL. 5, 1998 NOTES 901 TABLE 4. Serological neutralizing antibody responses to rotavirus 120 days after onset of severe gastroenteritis Patient no. (G serotype of infecting virus) No. of serum samples tested Period studied (days postonset) Sera containing neutralizing antibody to indicated virus (%) a ST-3 21 (1) 3 241 409 0 0 0 12 (1) 3 245 499 33 33 66 14 (1) 4 178 430 75 50 100 31 (1) 5 191 642 100 100 100 33 (1) 3 238 460 100 100 100 4 (4) 3 239 488 100 66 100 a One hundred percent of sera for the patient numbers and G serotypes indicated contained neutralizing antibody to the Wa P[8], G1 virus. days after the onset of severe rotavirus diarrhea contained neutralizing antibody to the local neonatal G3P[6] rotavirus as often as antibody directed to the homologous serotype (G1 or G4) (Table 3). In contrast, the most frequently detected antibodies in sera collected at least 120 days after the onset of severe rotavirus diarrhea were directed to virus of the infecting serotype (Table 4). In a previous study of children in this population, we showed nonspecific neutralization of by a maximum of 7% of stool extracts tested (8), so it is likely that at least 93% of responses to in this study were specific. These children may have been infected with this virus as neonates (2), made no lasting serum rotavirus-neutralizing antibody response to, and remained susceptible to severe disease. Neonatal infection with -like rotaviruses has been shown to be protective against later severe rotavirus gastroenteritis but not against reinfection (1). However, neutralizing antirotaviral antibodies were not measured in that study. Alternatively, the neutralizing coproantibody responses in these children may have been directed to epitopes on VP7 or VP4 which are shared with (9), whereas serum FFN responses might not include these epitopes. Differences like this in FFN responses to rotavirus between stools and sera may help explain serotype-specific immunity in the absence of typespecific neutralizing antibody in serum (16) and need further study. FFN responses in sera and stools to the local virus RV-4 were of a higher level and were more frequent than those to the U.S. strain, Wa, and the Japanese isolate, Ku, irrespective of monotype of infecting or test virus. Similarly, as discussed above, stool FFN responses to local virus were frequent and of a high level. Thus, local rotavirus isolates may be needed to give the most-sensitive measures of rotavirus-neutralizing antibodies in both sera and secretions. Overall, this study shows that serum neutralizing antibody responses appear to be a more useful measure of severely symptomatic rotavirus infection than stool responses, but these serum responses did not correlate with immunity to rotavirus in previous studies with nonlocal virus strains. Use of local strains in neutralization assays may help identify immune correlates of protection in vaccine trials. I thank Ruth Bishop for her help in selection of the children s samples for testing, Paul Mascendycz and Leanne Unicomb for their assistance with virus neutralization assays, Simone Richardson and Rebecca Gorrell for performing P genotype analysis, Ian H. Holmes for helpful discussion, and Jane Lee for her help in typing the manuscript. This study was supported by project grants 840422, 860298, and 890347 from the National Health and Medical Research Council of Australia and by the Royal Children s Hospital Research Foundation. REFERENCES 1. Bishop, R. F., G. L. Barnes, E. Cipriani, and J. S. Lund. 1983. Clinical immunity after neonatal rotavirus infection. A prospective longitudinal study in young children. N. Engl. J. Med. 309:72 76. 2. Bishop, R. F., L. E. Unicomb, and G. L. Barnes. 1991. Epidemiology of rotavirus serotypes in Melbourne, Australia, from 1973 to 1989. J. Clin. Microbiol. 29:862 868. 3. Coulson, B. S. 1993. Typing of human rotavirus VP4 by an enzyme immunoassay using monoclonal antibodies. J. Clin. Microbiol. 31:1 8. 4. Coulson, B. S. 1987. Variation in neutralization epitopes of human rotaviruses in relation to genomic RNA polymorphism. Virology 159:209 216. 5. Coulson, B. S., K. Grimwood, I. L. Hudson, G. L. Barnes, and R. F. Bishop. 1992. Role of coproantibody in clinical protection of children during reinfection with rotavirus. J. Clin. Microbiol. 30:1678 1684. 6. Coulson, B. S., K. Grimwood, P. J. Masendycz, J. S. Lund, N. Mermelstein, R. F. Bishop, and G. L. Barnes. 1990. Comparison of rotavirus immunoglobulin A coproconversion with other indices of rotavirus infection in a longitudinal study in childhood. J. Clin. Microbiol. 28:1367 1374. 7. Coulson, B. S., and C. Kirkwood. 1991. Relation of VP7 amino acid sequence to monoclonal antibody neutralization of rotavirus and rotavirus monotype. J. Virol. 65:5968 5974. 8. Coulson, B. S., and P. J. Masendycz. 1990. Measurement of rotavirusneutralizing coproantibody in children by fluorescent focus reduction assay. J. Clin. Microbiol. 28:1652 1654. 9. Coulson, B. S., J. M. Tursi, W. J. McAdam, and R. F. Bishop. 1986. Derivation of neutralizing monoclonal antibodies to human rotaviruses and evidence that an immunodominant neutralization site is shared between serotypes 1 and 3. Virology 154:302 312. 10. Coulson, B. S., L. E. Unicomb, G. A. Pitson, and R. F. Bishop. 1987. Simple and specific enzyme immunoassay using monoclonal antibodies for serotyping human rotaviruses. J. Clin. Microbiol. 25:509 515. 11. Gentsch, J. R., R. I. Glass, P. Woods, V. Gouvea, M. Gorziglia, J. Flores, B. K. Das, and M. K. Bhan. 1992. Identification of group A rotavirus gene 4 types by polymerase chain reaction. J. Clin. Microbiol. 30:1365 1373. 12. Gerna, G., A. Sarasini, M. Torsellini, D. Torre, M. Parea, and M. Battaglia. 1990. Group- and type-specific serologic response in infants and children with primary rotavirus infections and gastroenteritis caused by a strain of known serotype. J. Infect. Dis. 161:1105 1111. 13. Good, M. F., Y. Zevering, J. Currier, and J. Bilsborough. 1993. Original antigenic sin, T cell memory, and malaria sporozoite immunity: an hypothesis for immune evasion. Parasite Immunol. 15:187 193. 14. Grimwood, K., J. C. Lund, B. S. Coulson, I. L. Hudson, R. F. Bishop, and G. L. Barnes. 1988. Comparison of serum and mucosal antibody responses following severe acute rotavirus gastroenteritis in young children. J. Clin. Microbiol. 26:732 738. 15. Ward, R. L. 1996. Mechanisms of protection against rotavirus in humans and mice. J. Infect. Dis. 174(Suppl. 1):S51 S58. 16. Ward, R. L., D. R. Knowlton, E. T. Zito, B. L. Davidson, R. Rappaport, and M. E. Mack, for the U.S. Rotavirus Vaccine Efficacy Group. 1997. Serologic correlates of immunity in a tetravalent reassortant rotavirus vaccine trial. J. Infect. Dis. 176:570 577.