Monoreassortant Rotaviruses of Multiple G Types Are Differentially Neutralized by Sera From Infants Vaccinated With ROTARIX and RotaTeq

Abstract

Background

Rotavirus is a leading cause of pediatric diarrheal mortality. The rotavirus outer capsid consists of VP7 and VP4 proteins, which, respectively, determine viral G and P type and are primary targets of neutralizing antibodies.

Methods

To elucidate VP7-specific neutralizing antibody responses, we engineered monoreassortant rotaviruses each containing a human VP7 segment from a sequenced clinical specimen or a vaccine strain in an identical genetic background. We quantified replication and neutralization of engineered viruses using sera from infants vaccinated with monovalent ROTARIX or multivalent RotaTeq vaccines.

Results

Immunization with RotaTeq induced broader neutralizing antibody responses than ROTARIX. Inclusion of a single dose of RotaTeq in the schedule enhanced G-type neutralization breadth of vaccinated infant sera. Cell type-specific differences in infectivity, replication, and neutralization were detected for some monoreassortant viruses.

Conclusions

These findings suggest that rotavirus VP7, independent of VP4, can contribute to cell tropism and the breadth of vaccine-elicited neutralizing antibody responses.

Monoreassortant rotaviruses were engineered and used to define roles of outer-capsid protein VP7 in replication and neutralization. VP7 independently contributes to cell tropism, and vaccination with multivalent RotaTeq induces broader neutralizing antibody responses in infants than monovalent ROTARIX.

For rotavirus, a leading cause of diarrheal mortality for children under 5 years of age, understanding individual outer-capsid antigen functions and vaccine engineering have been hampered by narrow tropism and a lack of genetic tractability [1]. The outer capsid of nonenveloped rotavirus virions contains 260 VP7 glycoprotein trimers forming the shell and 60 VP4 trimers projecting from the surface [2, 3]. VP7 and VP4, respectively, determine viral G and P type and are primary targets of neutralizing antibodies (NAbs). Human rotaviruses replicate poorly in continuous cell lines unless adapted [4]. Rotavirus attachment protein VP4 primarily dictates tropism, but VP7 can interact with integrin coreceptors to mediate internalization [5–10]. After 1 or more natural rotavirus infections, children have progressively lower risk of subsequent rotavirus infection and disease [11]. Animal models suggest that B cells are critical for long-term protection against rotavirus disease, although T cells may help clear virus and establish memory responses [12]. Initial exposures to rotavirus vaccination or infection elicit primarily homotypic VP7-directed antibodies, whereas cross-reactive VP7 antibodies are elicited after subsequent exposures [13–15]. An entirely plasmid-based reverse genetics system may enable focused assessment of outer-capsid antigen functions in the context of intact virus [16]. Specifically, reassortant rotaviruses containing mixtures of animal and human rotavirus segments may alleviate the poor replication of human rotaviruses in cell lines while permitting directed study of human rotavirus VP7.

Although vaccines have decreased global rotavirus disease burden, the breadth and specificity of NAb responses they elicit are incompletely understood [17]. ROTARIX (RV1; GlaxoSmithKline), 1 of 2 domestically approved vaccines, is an attenuated human G1P[8] isolate that is administered at 2 and 4 months of age [18–20]. RotaTeq (RV5; Merck) consists of 5 monoreassortant virus strains, each containing a human VP7 (G1–G4) or VP4 (P[8]) segment in a bovine rotavirus background and is administered at 2, 4, and 6 months of age [21, 22]. In low mortality countries, RV1 vaccine effectiveness (VE) is 86% (95% confidence interval [CI], 81–90), and RV5 VE is 86% (95% CI, 76–92) against severe rotavirus disease in children up to 12 months [23]. In high-mortality countries, RV1 VE is 63% (95% CI, 54–70), and RV5 VE is 66% (95% CI, 51–76) in children up to 12 months. Enhanced RV5 protection, compared with that of the background bovine virus, suggests that the human outer-capsid antigens specifically promote immune-mediated protection [24]. RV1 and RV5 were derived using human rotavirus strains that circulated in the 1980s, when G1–G4, P[4], and P[8] predominated [19, 25]. In recent decades, G1 prevalence declined, while rotaviruses containing G8, G9, and G12 VP7, which are absent from RV1 and RV5, emerged and gained global epidemiologic relevance [26–29]. For represented genotypes, VP7 amino acid sequences of circulating rotaviruses may form antibody-binding epitopes differing from those of vaccine strains [30, 31]. Seroconversion and NAb titer typically have been quantified using human rotavirus strains containing frequently detected combinations of G and P types (eg, G1P[8] or G2P[4]) [2, 32]. Thus, NAb responses to individual outer-capsid antigens elicited by vaccines are incompletely characterized.

Rotavirus neutralization assays historically have been conducted using monkey kidney epithelial (MA104) cells. Neutralization properties of VP4 can be different based on its antigenic context [33] or cell type, with some monoclonal antibodies selectively neutralizing rotavirus on human intestinal epithelial (HT29) cells but not MA104 cells [34]. Total rotavirus neutralization titers (NTs) in adult and infant serum also are higher on HT29 than MA104 cells. Although much of this difference may be attributable to VP4, cell type-specific neutralization differences also may extend to VP7.

To elucidate VP7-specific NAb responses, we engineered rotaviruses in an identical genetic background that encode VP7 proteins derived from RV1, RV5, or circulating strains representing multiple G types. We quantified infectivity, replication, and the capacity of sera from differentially vaccinated infants to neutralize the viruses in MA104 and HT29 cells. The data indicate that VP7 independently contributes to rotavirus cell tropism and that RV5 vaccination elicits a broader VP7-specific serum NAb response than RV1 vaccination. Furthermore, inclusion of a single dose of RV5 in the vaccination schedule can enhance G-type neutralization breadth. HT29 cells may provide more consistent results than the MA104 cells historically used to conduct neutralization assays. These findings suggest that broad VP7 exposure leads to broader cross-NAb responses and support the use of reverse genetics to discern individual rotavirus outer-capsid antigen functions.

METHODS

Cells

MA104 and Vero cells were grown in Eagle’s minimum essential medium (MEM) with Earle’s salts and L-glutamine (Corning) plus 5% fetal bovine serum (FBS) (Gibco). HT29 cells were grown in McCoy’s 5A with L-glutamine plus 10% FBS. Baby hamster kidney fibroblasts expressing T7 ribonucleic acid (RNA) polymerase (BHK-T7 cells) were grown in Dulbecco’s MEM (Corning) and L-glutamine plus 5% FBS, with 1mg/mL Geneticin (Gibco) in alternate passages. Cells were maintained at 37°C with 5% CO₂. Media were additionally supplemented with 100 units/mL penicillin/100 µg/mL streptomycin (Corning) and 25ng/mL amphotericin B (Corning).

Viruses

SA11 (G3P[2]) and monoreassortant rotaviruses were engineered using reverse genetics [16]. For monoreasortants, a pT7-g9 plasmid encoding a human VP7 segment replaced SA11 pT7-g9. VP7 sequences were derived from RV1, RV5, or clinical specimens collected at Vanderbilt University Medical Center (VUMC) from 2011 to 2013 [35]. GenBank accession numbers are as follows: JN849114 (G1 RV1), KT919122 (G1 VUMC), GU565068 (G2 RV5), KT918847 (G2 VUMC), GU565079 (G3 RV5), KT919067 (G3 VUMC), GU565090 (G4 RV5), KT919508 (G9 VUMC), and KT919794 (G12 VUMC). G4 strains were not detected in Nashville from 2011 to 2013 [35]. Sequences were synthesized by Genewiz, LLC then introduced into pT7-g9 using restriction enzyme digestion. Cloning of G1 RV5 (GU565057) was unsuccessful. Monoreassortant virus stocks were made by 1 round of amplification in MA104 cells, and virus titer was quantified by plaque assay [36]. G2 VUMC was amplified 1 to 3 additional times. Ribonucleic acid extraction and Sanger sequencing revealed no polymorphisms in G2 VUMC segment 9 at passage 4.

Antibodies and Sera

Commercial antibodies include sheep α-rotavirus polyclonal (Invitrogen) and Alexa Fluor 488-labeled α-sheep immunoglobulin G (Invitrogen). Rabbit polyclonal rotavirus antisera were made by Bioqual, Inc. In brief, RV1 (~4 × 10⁶ infectious units/mL) or RV5 (~1 × 10⁷ infectious units/mL) was administered intramuscularly on days 0 and 14 with TiterMax Gold adjuvant. Test bleeds were taken days 0, 14, and 28, with final serum harvest day 49. Preimmunization sera initially failed to react with rotavirus laboratory strains by immunoblotting, but a putative VP7 band was later detected for SA11 after overexposure (Supplementary Figure 2). Vaccinated infant sera were collected at the Vanderbilt Vaccine Research Program and other sites [32]. With parental informed consent, infants aged 6 to 15 weeks were enrolled and administered RV1 (2 doses), RV5 (3 doses), or 1 of 3 RV1 and RV5 combinations (3 doses: RV5-RV1-RV1, RV5-RV5-RV1, or RV1-RV5-RV5). Sera were collected 1 month after final dose administration, with permission to retain residual serum for future studies. This study was approved by the Vanderbilt Institutional Review Board (IRB no. 170756).

Fluorescent Focus Assay

HT29 (~2 × 10⁵/well) or MA104 (~1.5 × 10⁵/well) cells were seeded in black-walled, clear-bottom, 96-well plates and incubated at 37°C overnight. Virus was activated with 1 μg/mL trypsin (no. LS003708, Worthington) for 1 hour at 37°C and serially diluted 1:4 in serum-free medium. After 2 washes, cells were adsorbed with virus dilutions for 1 hour at 37°C with centrifugation at 1000 ×g. Cells were washed twice and incubated at 37°C for 16–18 hours before methanol fixation. Cells were stained to detect nuclei using DAPI (Invitrogen) and rotavirus proteins before imaging and quantification using an ImageXpress Micro XL Widefield High-Content Analysis System (Applied Biosystems). Virus titer was quantified from total and infected cells quantified in 4 fields of view/well in duplicate wells.

Fluorescent Focus Reduction Neutralization Assay

HT29 (~2 × 10⁵/well) or MA104 (~1.5 × 10⁵/well) cells were seeded in black-walled, clear-bottom, 96-well plates and incubated at 37°C overnight. Enough rotavirus to infect ~200 cells/field in HT29 and ~100 cells/field in MA104 cells was activated with 1 μg/mL trypsin for 1 hour at 37°C before incubation with 1:2 serially diluted serum (starting at 1:10) from a rabbit or infant immunized with RV1, RV5, or a combination thereof for 1 hour at 37°C. Cells were washed twice and adsorbed in duplicate with virus-serum mixtures for 1 hour at 37°C with centrifugation at 1000 ×g before 2 additional washes and incubation at 37°C for 16–18 hours. Cells were fixed, stained, imaged, and quantified as for fluorescent focus assay (FFA). NT was the reciprocal of the highest serum dilution yielding at least 60% decrease in infectivity.

Replication Time Course

HT29 or MA104 cells (~1.7 × 10⁵/well) were seeded in 24-well plates and incubated at 37°C until confluent. Rotaviruses were activated with 1 µg/mL trypsin for 1 hour at 37°C. Cells were washed twice and adsorbed with activated viruses for 1 hour at 37°C at a multiplicity of infection of 0.01 focus-forming units (FFU)/cell, quantified on the corresponding cell line. Cells were washed to remove unbound virus and incubated with serum-free medium plus 0.5 µg/mL trypsin at 37°C for 0 or 48 hours. Plates were frozen at −80°C and thawed 3 times before titering virus by FFA on MA104 cells.

Statistical Analyses

Statistical analyses were conducted using GraphPad Prism 8. To determine whether VP7 identity alters SA11 replication, we used one-way analysis of variance followed by individual pairwise comparisons with Student’s t test to determine whether virus titer differs from that of SA11. To correct for multiple comparisons, we used Bonferroni. Thus, P < .0056 were considered significant. To determine whether VP7 identity alters neutralization of SA11 for a vaccination condition or whether the vaccine administered affects NT for a virus, we used Kruskal-Wallis followed by individual pairwise comparisons using a Wilcoxon rank-sum (Mann-Whitney) test. Due to relatively small sample size and the discrete nature of NT data, large differences were required to detect significant difference between data sets. Thus, we did not correct for multiple comparisons.

RESULTS

VP7 Influences Infectivity and Replication Efficiency in Simian and Human Cell Lines

To elucidate VP7 functions in replication and neutralization, we used reverse genetics to engineer a panel of monoreassortant rotaviruses. Each virus contains a human VP7 segment in the genetic background of simian rotavirus strain SA11 (Figure 1A and Supplementary Figure 1). Human VP7 sequences were derived from RV1, RV5, or human rotaviruses that circulated in Nashville, Tennessee between 2011 and 2013, and they represent 6 different G types (G1–G4, G9, and G12) [35]. The VP7 monoreassortants, which made plaques similar to those of recombinant SA11 (Figure 1A), were titered for infectivity on MA104 and HT29 cells. Titers of G1 RV1 and G1 VUMC, respectively, were more than 100-fold and 5-fold lower in HT29 than MA104 cells (Figure 1B). In contrast, stock titers for G3 VUMC and G4 RV5, respectively, were approximately 50-fold and 15-fold higher in HT29 than MA104 cells. We quantified the influence of VP7 on replication in MA104 and HT29 cells by adsorbing monolayers with 0.01 FFU/cell of viruses titered in the corresponding cell type and quantified virus titer in cell lysates after 48 hours. Almost all viruses replicated less efficiently than SA11 in both cell lines (Figure 1C). In accord with infectivity, both G1 monoreassortants replicated to significantly higher titer in MA104 than HT29 cells. G3 VUMC, G4 RV5, and G12 VUMC replicated to significantly higher titer in HT29 than MA104 cells. These findings suggest that most engineered SA11 rotaviruses incorporating a human VP7 segment replicate efficiently, and VP7 can contribute to replication efficiency and cell tropism.

Figure 1.

Monoreassortant rotavirus stock titer and replication in simian and human cells. (A) Schematics of rotavirus virions showing genome segments as lines and outer capsid protein VP7 as a hexagon. Light gray coloring represents SA11. The source of the human VP7 sequence is indicated by color and a text label. Plaques in CV-1 monkey kidney cells stained with crystal violet are shown to the right of each virus schematic and below SA11. (B) Monoreassortant rotavirus stocks were titered on MA104 or HT29 cell monolayers. Shown are average titers from 4 quantified fields of view per virus for a single virus stock. Bars represent standard error of the mean. Dashed line, limit of detection. (C) Monolayers of MA104 or HT29 cells were adsorbed with 0.01 focus-forming units (FFU)/cell of the indicated virus. After 48 hours, cells were lysed, and virus titer in lysates was quantified by fluorescent focus assay. Mean virus titer and standard deviation (error bars) are shown. n = 3 independently diluted inocula from 1 quantified virus stock. ∗, P < .05; ∗∗, P < .01; ∗∗∗, P < .001 by Student’s t test. RV1, ROTARIX vaccine; RV5, RotaTeq vaccine; VUMC, human clinical specimen collected at Vanderbilt University Medical Center 2011–2013.

RV5 Elicits a Broader VP7 Neutralizing Response Than RV1

To gain insight into NAb responses elicited by RV1 and RV5 against VP7, we quantified neutralization of monoreassortant rotaviruses by sera from immunized rabbits and vaccinated infants using a fluorescent focus reduction neutralization (FFRN) assay in MA104 cells. To establish assay conditions, serum from a single rabbit per immunization condition was tested 3 times in duplicate, with preimmune serum providing an assay baseline. To compare antibody responses elicited by vaccines, sera from 10 infants per vaccination condition were each tested in duplicate. In a rabbit and in infants, RV1 elicited significantly higher titers of NAbs against G1 RV1, G1 VUMC, G2 VUMC, G3 RV5, and G3 VUMC than against SA11 (Figure 2A). However, rabbit preimmune serum significantly neutralized G2 VUMC, G3 RV5, and G3 VUMC. Although rabbit preimmune serum failed to detect the conserved rotavirus VP6 protein by immunoblot, a faint band at the molecular weight of VP7 suggests a possible prior rotavirus exposure (Supplementary Figure 2). In a rabbit, RV5 elicited significantly higher titers of NAbs against G2 VUMC, G3 RV5, and G4 RV5, and G9 VUMC than SA11, although low neuralization titers were detected for all tested viruses (Figure 2B). Preimmune rabbit serum significantly neutralized G2 VUMC. In infants, RV5 elicited NAbs against all tested viruses, including G9 VUMC and G12 VUMC, G types not included in RV5, although none were significantly higher than for SA11. Sera from RV5-RV5-RV5- but not RV1-RV1-vaccinated infants efficiently neutralized RV5 (Supplementary Figure 3). These data, taken together, suggest that multivalent RV5-elicited NAbs are cross-reactive, whereas monovalent RV1-elicited NAbs recognize a narrower breadth of G-type antigens.

Figure 2.

Neutralization of VP7 monoreassortant viruses by immunized rabbit or infant sera. A dilution of rotavirus stock yielding ~100 infected MA104 cells per field was activated with trypsin. Activated rotavirus was incubated with serially diluted serum from rabbits primed and boosted with RV1, preimmune serum from the RV1-immunized rabbit, or infants vaccinated with 2 doses of RV1 (A) or was incubated with serially diluted serum from rabbits primed and boosted with RV5, preimmune serum from the RV5-immunized rabbit, or infants vaccinated with 3 doses of RV5 (B). Confluent MA104 monolayers were adsorbed with virus-serum mixtures before washing and incubation. Cells were fixed, stained to detect nuclei and rotavirus proteins, and quantified using an ImageXpress Micro XL Widefield High-Content Analysis System. Neutralization titer is the reciprocal of the highest serum dilution at which infectivity was reduced by at least 60%. Boxes extend from 25th to 75th percentile. Bars extend from minimum to maximum. The line in the middle of the box is the median. For rabbit sera, 3 independent experiments were conducted in duplicate. For infant sera, 10 samples were tested in duplicate. Individual data points are shown. ∗, P < .05; ∗∗, P < .01; ∗∗∗, P < .001 compared with SA11 by Wilcoxon rank-sum test (Mann-Whitney). VUMC, Vanderbilt University Medical Center.

A Single Dose of RV5 Enhances the Breadth of VP7 Neutralization

To determine how mixed RV1 and RV5 dosing affect VP7 serum NAb specificity, we conducted FFRN assays in MA104 cells using VP7 monoreassortants and sera from infants vaccinated with 1 of 3 different 3-dose combinations of RV1 and RV5. Mean NTs for SA11 were significantly lower for RV1-RV1 and all mixed dosing strategies compared with those of RV5-RV5-RV5 (Table 1). Mean NTs for G2 RV5 and G12 VUMC for some dosing strategies and for G4 RV5 and G9 VUMC for all dosing strategies were significantly higher than those for infants vaccinated with RV1-RV1. Although all mixed-dosing schedules included 3 doses compared with the 2-dose standard RV1 schedule, inclusion of a single dose of RV5 led to significant serum neutralization of G4 RV5 and G9 VUMC, and 2 doses of RV5 at the end of the dosing schedule (RV1-RV5-RV5) resulted in significant neutralization of most chimeric viruses compared with SA11 (Figures 2B and 3). These results suggest that mixed vaccine dosing may enhance the breadth and specificity of the serum NAb response to human rotavirus VP7.

Table 1.

Mean Neutralization Titers for Monoreassortant Viruses Based on Vaccine Dosing Strategy

Virus strain	RV1-RV1a	RV5-RV1-RV1	RV5-RV5-RV1	RV1-RV5-RV5	RV5-RV5-RV5
SA11	7 (−2.1 to 16.1)	9 (2.8–15.2)	6 (2.5–9.5)∗	13.5 (4.7–22.3),†	27 (18.0–36.0)†
G1 RV1b	34 (20.2–47.8)	53 (9.2–96.8)	32 (8.5–55.5)	39.5 (18.6–60.4)	59.5 (16.3–102.8)
G1 VUMC	79 (33.5–124.5)	60 (22.6–97.4)	50 (10.5–89.5)	54.5 (19.6–89.4)	85.5 (15.6–155.5)
G2 RV5	4 (1.2–6.8)∗	19 (6.8–31.2)	5.5 (−0.3 to 11.3)∗	23 (0.8–45.2)	26.5 (9.5–43.5)†
G2 VUMC	47.5 (9.6–85.4)	241 (57.6–424.4)	50.5 (15.1–85.9)	303 (16.9–589.1)	243 (52.9–433.1)
G3 RV5	261 (40.7–381.3)	151 (4.8–297.2)	110.5 (27.1–193.9)	185 (88.5–281.5)	370.5 (74.6–666.4)
G3 VUMC	87 (10.4–163.6)	19 (1.0–37.0)	83 (9.0–175.0)	67.5 (15.2–119.8)	24 (12.5–35.5)
G4 RV5	10 (2.3–17.7)∗	28 (13.6–42.4)∗,†	36 (14.2–57.8)∗,†	103.5 (15.3–191.7)†	83.5 (33.6–133.4)†
G9 VUMC	1 (−0.4 to 2.4)∗	35 (7.9–62.1)†	112.5 (−22.1 to 247.1)†	137.5 (16.8–258.2)†	35 (13.9–56.1)†
G12 VUMC	4 (0.8–7.2)∗	17 (4.7–29.3) ∗	15 (5.3–24.7)∗,†	35 (16.0–54.0)†	37 (15.3–58.7)†

Abbreviations: VUMC, Vanderbilt University Medical Center.

^aMean neutralization titer followed by 95% confidence interval in parentheses for the indicated virus and vaccination condition are shown. Neutralization titers are shown graphically in Figures 2 and 3 but duplicated in the table for comparison of a given virus treated with sera from 10 infants per vaccination condition.

^bG1 RV1, G1 VUMC, G2 VUMC, G3 RV5, and G3 VUMC do not differ significantly by Kruskal-Wallis and were not further statistically analyzed.

∗P < .05 compared with RV5-RV5-RV5 by Wilcoxon rank-sum test (Mann-Whitney).

†P < .05 compared with RV1-RV1 by Wilcoxon rank-sum test (Mann-Whitney).

Figure 3.

Neutralization of VP7 monoreassortant viruses by mixed-dose vaccinated infant sera. Neutralization assays were conducted as in Figure 2 using sera from infants vaccinated with 3-dose combinations of RV1 and RV5, as indicated. Neutralization titer is the reciprocal of the highest serum dilution at which infectivity was reduced by at least 60%. Boxes extend from 25th to 75th percentile. Bars extend from minimum to maximum. The line in the middle of the box is the median. For each virus, 10 samples were tested in duplicate. Individual data points are shown. ∗, P < .05; ∗∗, P < .01; ∗∗∗, P < .001 compared with SA11 by Wilcoxon rank-sum test (Mann-Whitney). VUMC, Vanderbilt University Medical Center.

VP7 Neutralizing Responses Are Similar in Human and Simian Cell Lines

To identify differences in VP7-specific rotavirus serum neutralization between simian and human cells, we quantified serum neutralization of monoreassortant rotaviruses in HT29 cells. Variability in NT in HT29 cells appeared to be reduced compared with that in MA104 cells, but overall neutralization patterns were largely similar (Figures 2 and 4). In a rabbit, RV1 elicited significantly higher NAb titers against all viruses except G4 RV5 and G9 VUMC than against SA11, although significant neutralization of G1 VUMC, G2 VUMC, and G3 VUMC was detected for preimmune serum (Figure 4A). In infants, RV1 elicited significantly higher NAb titers against all viruses except G3 VUMC, G4 RV5, G9 VUMC, and G12 VUMC than against SA11. NTs for RV1-RV1-vaccinated infant sera were significantly higher for G2 VUMC and lower for both G3 viruses in HT29 than MA104 cells (Table 2). In a rabbit, RV5 elicited significantly higher NAb titers against all viruses except G1 RV1 and G1 VUMC than against SA11, although significant neutralization of G2 VUMC was detected for preimmune serum (Figure 4B). In infants, RV5 elicited NAbs against all tested viruses with few serum samples failing to exhibit at least low levels of neutralization. NTs for RV5-RV5-RV5-vaccinated infant sera were significantly higher for G2 VUMC in HT29 cells compared with MA104 cells (Table 2). These findings suggest that VP7 may exert cell type-specific influences on neutralization assay outcomes.

Figure 4.

Neutralization of VP7 monoreassortant viruses by RV1- or RV5-immunized rabbit or infant sera in human cells. A dilution of rotavirus stock that would yield ~200 infected HT29 cells per field was activated with trypsin. Activated rotavirus was incubated with serially diluted serum from rabbits primed and boosted with RV1, preimmune serum from the RV1-immunized rabbit, or infants vaccinated with 2 doses of RV1 (A) or was incubated with serially diluted serum from rabbits primed and boosted with RV5, preimmune serum from the RV5-immunized rabbit, or infants vaccinated with 3 doses of RV5 (B). Confluent monolayers of HT29 cells were adsorbed with virus-serum mixtures before washing and incubation. Cells were fixed, stained, and quantified as in Figure 2. Neutralization titer is the reciprocal of the highest serum dilution at which infectivity was reduced by at least 60%. Boxes extend from 25th to 75th percentile. Bars extend from minimum to maximum. The line in the middle of the box is the median. For rabbit sera, 3 independent experiments were conducted in duplicate. For infant sera, 10 samples were tested in duplicate. Individual data points are shown. ∗, P < .05; ∗∗, P < .01; ∗∗∗, P < .001; ∗∗∗∗, P < .0001 compared with SA11 by Wilcoxon rank-sum test (Mann-Whitney). VUMC, Vanderbilt University Medical Center.

Table 2.

Mean Neutralization Titers for Monoreassortant Viruses Based on Cell Type

Virus strain	Mean NT in MA104 (95% CI)a	Mean NT in HT29 (95% CI)
RV1-RV1
SA11	7 (−2.1 to 16.1)	1 (−0.4 to 2.4)
G1 RV1	34 (20.2–47.8)	50 (35.3–64.7)
G1 VUMC	79 (33.5–124.5)	58 (36.6–79.5)
G2 RV5	4 (1.2–6.8)	7 (3.3–10.8)
G2 VUMC	47.5 (9.6–85.4)	156.5 (82.2–230.8)∗
G3 RV5	261 (40.7–381.3)	14 (5.2–22.8)†
G3 VUMC	87 (10.4–163.6)	2.5 (−0.1 to 5.1)‡
G4 RV5	10 (2.3–17.7)	7 (0–14.0)
G9 VUMC	1 (−0.4 to 2.4)	2.5 (0.4–4.6)
G12 VUMC	4 (0.8–7.2)	3.5 (0.8–6.2)
RV5-RV5-RV5
SA11	27 (18.0–36.0)	25 (17.6–32.4)
G1 RV1	59.5 (16.3–102.8)	38 (25.3–50.7)
G1 VUMC	85.5 (15.6–155.5)	33 (20.2–45.8)
G2 RV5	26.5 (9.5–43.5)	37 (22.9–51.1)
G2 VUMC	243 (52.9–433.1)	254 (165.7–342.3)‡
G3 RV5	370.5 (74.6–666.4)	79.5 (51.9–107.1)
G3 VUMC	24 (12.5–35.5)	29.5 (16.4–42.6)
G4 RV5	83.5 (33.6–133.4)	77.5 (57.7–97.3)
G9 VUMC	35 (13.9–56.1)	29 (21.0–37.0)
G12 VUMC	37 (15.3–58.7)	41 (26.4–55.6)

Abbreviations: CI, confidence interval; NT, neutralization titer; VUMC, Vanderbilt University Medical Center.

^aMean NT followed by 95% CI in parentheses for the indicated virus and vaccination condition are shown. NT is shown graphically in Figures 2 and 4 but duplicated in the table for comparison between cell lines.

∗P < .001,

†P < .01, and

‡P < .05 compared with MA104 by Wilcoxon rank-sum test (Mann-Whitney).

DISCUSSION

Using a panel of engineered monoreassortant rotaviruses, we found that VP7 independently contributes to infectivity and replication. Consistent with a role for VP4 in rotavirus tropism, engineered monoreassortants containing human VP7 tend to replicate well in cultured cells, whereas those containing nonadapted human VP4 replicate poorly [37–39] (Figure 1). However, differences in infectivity and replication relative to SA11 suggest that VP7 also contributes to rotavirus cell tropism (Figure 1). VP7 interacts with α_xβ₂ and α_vβ₃ integrins, which may function as coreceptors, and VP7 may interact with yet unidentified molecules to mediate entry [5–8]. Accordingly, after serial passage of clinical rotavirus specimens in cell lines, changes arose in VP7 as well as VP4 [40]. Consistent with a published human G2 monoreassortant, G2 VUMC exhibited limited infectivity and modest replication [38] (Figure 1B and C). Reduced G2 VUMC infectivity may have increased neutralization sensitivity, artificially inflating NTs (Figures 2 and 4). Because human VP7 is readily incorporated into SA11, often yielding viruses that replicate efficiently, reverse genetics may permit introduction of chosen VP7 antigens into an animal rotavirus background for vaccine engineering [22, 41].

Our findings suggest that RV5 elicits broader VP7 NAb responses than RV1 and that including multiple G-type antigens in at least 1 vaccine dose can elicit cross-NAbs. Background NTs in RV1-immunized rabbits detected using preimmune serum suggest that most VP7-specific NAbs in the immunized animal were directed against G1 RV1 and G1 VUMC, whose exposed antigenic surfaces are almost identical (Figures 2, 4, and 5). Background neutralization for G2 and G3 monoreassortants in RV5-immunized rabbits suggests that we overestimated NTs for these viruses, particularly G2 VUMC, but fails to account for most neutralization breadth, even after only 2 RV5 immunizations (Figures 2 and 4). Accordingly, few amino acids are unique to the antigenic surfaces of G2 VUMC and G3 VUMC when aligned with all RV5 VP7 components (Figure 5). Low NTs for G1 RV1 and G1 VUMC may reflect failure of the individual rabbit to seroconvert to all RV5 vaccine components (Figures 2 and 4). Infants were tested in greater numbers than rabbits, providing more robust data, but received nonequal numbers of RV1 or RV5 immunizations. Because cross-reactive VP7 antibodies are elicited after subsequent rather than initial exposures [13–15], antibody cross-reactivity may vary based on immunization number, and we cannot exclude the possibility that an additional dose of RV1 could expand neutralization breadth. We lack prevaccination serum for infants, but in both MA104 and HT29 cells, VP7-specific NAb responses elicited by multivalent RV5 were more broadly directed than those elicited by monovalent RV1 (Figures 2 and 4 and Supplementary Figure 3). Almost every combination of vaccine doses elicited significantly higher NTs for G types not represented in the vaccine than did RV1 vaccination alone (Table 1). These results support and extend previous observations, in which seropositivity was determined using G1–G4, and G9 VP7 presented in the context of varied P-type antigens [32].

Figure 5.

Amino acid variation between VP7 proteins in monoreassortant viruses and vaccine strains. In the top left, a rotavirus virion (strain RRV, G3P[3]) is shown as a surface representation with VP4 colored red and VP7 colored yellow (PDB 4V7Q) [3]. The boxed area indicates an enlarged VP7 trimer colored light gray, with antigenic epitopes colored red (7–1a), khaki (7–1b), and purple (7–2) (PDB 3FMG) [49, 50]. Positions at which the amino acid sequence of a VUMC VP7 protein differs from that of G1 RV1 VP7, a single indicated human RV5 VP7 vaccine component, or the 4 human (G1, G2, G3, G4) and 1 bovine (G6) VP7 vaccine components of RV5 are colored cyan. VUMC, Vanderbilt University Medical Center.

Vaccine-mediated protection from rotavirus disease is complex. Despite eliciting limited NAb breadth compared with RV5, RV1 is an effective vaccine [19, 23, 32, 42]. Although some studies support correlation between rotavirus NAb titer and disease protection, in many cases protection exceeds vaccine seroresponse rate, especially for RV1 [19, 43, 44]. Monovalent, human G9P[11] vaccine ROTAVAC exhibited similar VE to RV1 and RV5 and efficacy against various commonly circulating genotypes in India [45]. Monovalent antigens can elicit heterotypic NAbs [15]. However, attenuated human rotavirus vaccines, like RV1 and ROTAVAC, also may present T-cell epitopes or elicit NAbs against other viral proteins, including VP6 [46, 47]. Because monovalent, human rotavirus vaccines are effective, increasing their VP7 antigenic breadth may be of little practical value. However, for vaccines that use a poorly protective animal rotavirus as a platform to display human VP7 and VP4, NAb breadth may contribute substantially to protection and should be a design consideration [24, 48].

Although we observed similar overall neutralization profiles, the more tightly clustered data points with smaller confidence intervals and decreased G3 NTs for RV1-RV1-vaccinated infant sera suggest that HT29 cells may be useful for rotavirus neutralization studies (Figures 2 and 4, Table 2). We anticipated detecting higher NTs in HT29 cells and genotype-specific NT differences [34] (Figure 1). However, we observed few significant NT differences between cell types, which correlated poorly with viral infectivity (Figure 1 and Table 2). Input virus dilutions typically differed only by up to 2-fold between cell lines. Thus, it is difficult to draw specific conclusions about the role of cell type in VP7 neutralization. Monoreassortant rotaviruses may help to discern biological differences that mediate assay outcomes and determine whether they relate to viral attachment and entry.

CONCLUSIONS

In the current study, we identified differences in the capacity of serum antibodies to neutralize rotavirus based on VP7 identity and vaccine administered. This information enhances our understanding of differences in homotypic and heterotypic vaccine-elicited antibody responses in infants. These studies reveal the utility of reverse genetics for neutralization studies and may inform vaccine evaluation and design approaches.

Supplementary Data

Supplementary materials are available at The Journal of Infectious Diseases online. Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author.

Notes

Acknowledgments. We thank Kathryn Edwards, the Vaccine and Treatment Evaluation Unit Rotavirus Vaccine Study Work Group, and the Division of Microbiology and Infectious Diseases at the National Institutes of Health for vaccinated infant serum samples. We acknowledge the Vanderbilt High Throughput Screening Facility for assistance with infectivity and neutralization assays. We thank Dr. James C. Slaughter from the Department of Biostatistics at Vanderbilt University Medical Center (VUMC) for statistical methods consultation. M. H. C. currently attends medical school at Wayne State University School of Medicine. S. V. S. currently attends graduate school at the University of Minnesota.

Financial support. This work was supported by a Turner-Hazinski award from the Department of Pediatrics at Vanderbilt University Medical Center and by the National Institutes of Health [R21 AI146698 to K.M.O.]. Vaccinated infant serum samples used in this study were generated from clinical studies where funds were provided by the National Institute of Allergy and Infectious Diseases, National Institutes of Health, and United States Department of Health and Human Services under contracts HHSN27220080000C (Vanderbilt University); HHSN272200800057C (University of Maryland, Baltimore); HHSN272200800006C (Cincinnati Children’s Hospital, Cincinnati, OH); HHSN27220800008C (University of Iowa); HHSN272200800004C (Group Health Cooperative); HHSN272200800013C (Emmes Corporation).

Potential conflicts of interest. All authors: No reported conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest.

Presented in part: 2019 SACNAS, October 31–November 2, 2019, Honolulu, HI, USA; American Society for Virology 40th Annual Meeting, July 19–23, 2021, virtual.