The Association between Symptomatic Rotavirus Infection and Histo-Blood Group Antigens in Young Children with Diarrhea in Pretoria, South Africa

Objectives: Recently, histo-blood group antigens (HBGAs) have been identified as receptors or attachment factors of several viral pathogens. Among rotaviruses, HBGAs interact with the outer viral protein, VP4, which has been identified as a potential susceptibility factor, although the findings are inconsistent throughout populations due to HBGA polymorphisms. We investigated the association between HBGA phenotypes and rotavirus infection in children with acute gastroenteritis in northern Pretoria, South Africa. Methods: Paired diarrheal stool and saliva samples were collected from children aged ≤ 59 months (n = 342) with acute moderate to severe diarrhea, attending two health care facilities. Rotaviruses in the stool samples were detected by commercial EIA and the rotavirus strains were characterized by RT-PCR targeting the outer capsid VP7 (G-type) and VP4 (P-type) antigens for genotyping. Saliva-based ELISAs were performed to determine A, B, H, and Lewis antigens for blood group typing. Results: Blood type O was the most common blood group (62.5%) in this population, followed by groups A (26.0%), B (9.3%), and AB (2.2%). The H1-based secretors were common (82.7%) compared to the non-secretors (17.3%), and the Lewis antigen positive phenotypes (Le(a+b+)) were predominant (54.5%). Blood type A children were more likely to be infected by rotavirus (38.8%) than any other blood types. P[4] rotaviruses (21/49; 42.9%) infected only secretor individuals, whereas P[6] rotaviruses (3/49; 6.1%) only infected Le(a−b−), although the numbers were very low. On the contrary, P[8] rotaviruses infected children with a wide range of blood group phenotypes, including Le(a−b−) and non-secretors. Conclusions: Our findings demonstrated that Lewis antigens, or the lack thereof, may serve as susceptibility factors to rotaviral infection by specific VP4 genotypes as observed elsewhere. Potentially, the P[8] strains remain the predominant human VP4 genotype due to their ability to bind to a variety of HBGA phenotypes.


Introduction
Globally, rotavirus is the leading cause of severe gastroenteritis in children under 5 years of age. Although two vaccines, Rotarix ® (GSK Biologicals, Rixensart, Belgium) and RotaTeq ® (Merck & Co, White River, PA, USA) have been recommended for use since 2009 by the World Health Organization (WHO) and are utilized in over 100 countries, rotavirus is still responsible for an estimated 128,500 deaths globally every year [1]. The majority of rotavirus deaths are in low-income countries in Asia and sub-Saharan Africa [1]. Since the initial WHO recommendation of the two vaccines, two new rotavirus vaccines developed We therefore decided to examine the association between HBGA profiles of children with rotavirus gastroenteritis (RVGE) in a cross-sectional study in northern Pretoria, South Africa.

Sample Population
Diarrheal surveillance in infants and young children < 5 years of age, has been ongoing in South Africa for over 30 years. Initially established to monitor rotavirus disease burden, and post-rotavirus vaccine introduction in 2009, to monitor the impact of the vaccine programme. More recently, this national surveillance program has focused on both rotavirus and other etiologies associated with acute watery diarrhea in young children attending health care facilities for treatment.
The children enrolled in this study constituted part of this on-going diarrhea surveillance in the Diarrheal Pathogens Research Unit (DPRU) based at Sefako Makgatho Health Sciences University and Dr. George Mukhari Academic Hospital complex. Dr. George Mukhari Academic Hospital (DGMAH) is a tertiary-care 1700 bed hospital based 30 kms northwest of Pretoria in South Africa. It has 23 associated health clinics and District Level Hospitals. Oukasie Primary Health Care Clinic (OPHC) is an outpatient health-based clinic 23.6 kilometers from DGMAH and refers patients requiring admission to DGMAH.
This study utilized the ongoing diarrhea surveillance to investigate the association of HBGAs and rotavirus infection. Parents of children with diarrhea were offered participation in this study and gave informed consent. Between June 2015 and November 2017, paired stool and saliva samples were collected in a total of 342 children less than 5 years of age who were seeking medical assistance for acute diarrhea at either OPHC or DGMAH. Acute diarrheal disease as reported by the parents was a presenting symptom for inclusion in the study. Diarrhea was defined as three or more episodes within 24 h.

Ethical Consideration
Ethical approval to conduct the study was obtained from Sefako Makgatho Health Sciences University Research Ethics Committee (SMUREC/P219/2015) and the management of OPHC and DGMAH. Paired diarrheal stools and saliva samples were collected from the children (under 59 months of age) whose parents/guardians gave consent. All the study methods were carried out guided by appropriate ethical guidelines and regulations.

Sample Collection
Diarrheal stool samples were collected from children on the day of enrolment. The stool samples were collected into 5 mL stool collection containers (B & M Scientific, Cape Town, South Africa) and stored immediately at 2-8 • C in refrigerators on site for later transport to the laboratory for testing. Stool samples at OPHC were transferred to the laboratory weekly, whereas samples at the Pediatric Department of DGMAH were transferred daily to the laboratory. EIAs for detection of rotavirus antigen are conducted monthly in the laboratory at the Diarrheal Pathogens Research Unit (DPRU) based at Sefako Makgatho Health Sciences University.
For saliva collection, a plain sterile oral swab (AEC-Amersham soc Ltd., Johannesburg, South Africa) was used to rub the inside of the child's mouth on both cheeks and was placed under the tongue for 10 s to absorb saliva. Immediately thereafter, the swab was placed into a 4 mL cryovial tube (Carl Roth GmbH & co, Karlsruhe, Germany) containing 1 mL of distilled water and transported to the DPRU on ice where they were kept at −80 • C until testing.

Rotavirus Detection and Strain Characterization in Stool
The detection of rotavirus in stool was performed following the WHO manual of rotavirus detection and characterization [25]. Briefly, group A rotavirus antigens in stool samples were tested using the commercially available ProSpecT TM rotavirus enzyme im-Viruses 2022, 14, 2735 4 of 14 munoassay (EIA) kit (Oxoid, Basingstoke, UK). Rotavirus dsRNA was extracted using Qiagen ® Viral RNA extraction kit (Qiagen, Hilden, Germany). Gene segment 4 that codes for VP4 and gene segment 9 which codes for VP7 were reverse transcribed using Avian Myeloblastosis Virus (AMV) reverse transcriptase. Two well-recognized sets of primers, Con2/Con3 and sBeg/End9" specific to VP4 (VP8* region) and VP7 were used to amplify the genes, respectively [25][26][27][28]. These amplicons were utilized to characterize rotavirus strains into G and P types. The following rotavirus genotypes were targeted: G1-G4, G8, G9, G10, G12 including P [4], P [6], P [8] and P [14], representing the most common human G and P types circulating in Sub-Saharan Africa [28].
The reactions were observed by adding HRP substrate reagents (optEIA, BD Bioscience, San Diego, CA, USA) and were read by spectra Max 340pc (Molecular Devices, San Jose, CA, USA). Four well-characterized saliva samples from our lab stock containing all interested HBGA types were included in each plate as internal controls. The cut-off of a positive signal was OD 450 = 0.2. Based on HBGA biosynthesis pathway, different HBGA categories are defined as follows: (i) secretors are defined as those who have one or more H1 antigenrelated HBGA types, including A, B, H1, and/or Le b antigen; (ii) non-secretors are defined as those who lack the H1 antigen-related HBGA types and are positive with Le a antigen; (iii) Le (a−b+) individuals are Le a negative but Le b positive; (iv) Le (a+b+) are those with both Le a and Le b antigens; (v) Le (a+b−) are those with Le a positive, but Le b negative, and (vi) Le (a−b−) individuals are those lacking Le a and Le b antigens.

Statistical Analysis
Data entry was carried out using MS Excel software (2019, Seattle, WA, USA). All statistical analysis were conducted using R software version 4.2.1 (RStudio, 2022, Boston, MA, USA). Descriptive statistics were used to summarize baseline characteristics. The absence or presence of rotavirus infection was considered the outcome of interest. Cross tabulations of characteristics and predictor variables with the dependent variable were constructed. Contingency tables were evaluated by Pearson's chi-squared test for qualitative variables above 5 and Fisher's exact test for qualitative variables below 5.
After descriptive analysis, simple and multiple logistic regression models were explored. Predictor variables with a p-value less than 0.20 in the simple logistic regression models were initially considered for multivariable logistic regression model building based on Hosmer-Lemeshow test. To measure the strength of association, we obtained the crude odds ratio (cOR), using simple logistic regression models, and adjusted odds ratio (aOR), with 95% CI using multiple logistic regression models, for controlling possible confounding variables. The Hosmer_Lemeshow was used to test the goodness of fit for the multiple logistic regression models at a significance level of 0.05.

Demographics
The 342 recruited children were predominantly males (51.5%, 176/342), aged between 0 and 59 months (mean age of 13.2 months, standard deviation of 10.1 months), and belonged to the Black ethnic group (Table 1). Of these children, 101 were admitted overnight to DGMAH pediatric wards and 241 attended OPHC as outpatients. Clinically, these children exhibited acute diarrhea, accompanied with fever, vomiting, and refusal to eat. Based on the information retrieved from their Road to Health Card (RTHC), which provides information on the child's health parameters and immunization records at the time of study recruitment, only 1.2% (4/342) of the children were unvaccinated with Rotarix vaccine; 6.7% (23/342) had received only one dose; and 82.7% (283/342) had received the full two doses of the vaccine. Immunization information was unknown for 9.4% (32/342) of the children overall. Routine immunization in South Africa is targeted at 6, 10 and 14 weeks of age, with rotavirus vaccine recommended at 6 and 14 weeks of age.

Rotavirus Infection in the Sample Population
Rotavirus was detected in 14.3% (49/342) of the diarrheal stool samples from children with diarrhea overall. The proportion of rotavirus-positive cases mirrored the age distribution of the study sample-thus, in children less than 12 months of age, encompassing 60.5% of the sample size (207/342), rotavirus was identified in 30 cases (61.2% of the study sample). Forty of the forty-nine rotavirus-positive cases were detected in children < 18 months of age (81.6%) ( Table 1). Older children above 24 months of age were hardly infected.
Only four children were documented to have not been immunized with rotavirus vaccine. None of these four children were positive for rotavirus diarrhea. The majority of children infected with rotavirus had received at least one dose of Rotarix vaccination (77.6% two doses; 10.2% single dose), although the vaccine record for six (12.2%) was unavailable. These six children were 6-, 12-, 13-, 18-, 37-, and 56-months old. In comparison, the vaccine record was not available for 26 (8.9%) of the 293-rotavirus negative children. Rotavirus cases were detected only in June-July 2015, July-August 2016, and June-September 2017, which are mid-winter and early spring months in South Africa.
Other clinical symptoms experienced by the children with diarrhea included fever, vomiting, and refusal to eat. Of these symptoms, fever and vomiting were statistically significant to rotavirus infection (p-value of <0.05). To substantiate further, rotavirus infection was stratified by site as shown in supplementary Table S1. Although more children overall (70.5%, 241/342) were admitted at OPHC with diarrheal disease, more children with rotavirus (59.2%; 29/49 versus 40.8%; 20/49) were admitted in DGMAH requiring hospitalization (Table S1). Furthermore, the children admitted to the wards at DGMAH presented with fever, vomiting, and acute diarrhea of a shorter duration.

Distribution of ABO Blood Groups, Lewis Antigens and Secretor Status in Rotavirus Infected Children
Based on the HBGA characterization noted above, 19 saliva samples were inconsistent in their classification and were excluded from analyses. Thus, a total of 323 samples were analyzed for both rotavirus and HBGA types. The distribution of blood groups in the population demonstrated that blood type O was most common in this population (62.5%; 202/323), followed by A (26.0%; 84/323), B (9.3%; 30/323), and then AB (2.2%; 7/323). As shown in Table 2  Furthermore, most children infected with rotavirus were secretors bearing the Lewis antigens Le (a+b+) , followed by Le (a−b+) ; Lewis-negative children ( Le(a−b−) were the least infected. None of the Le (a+b−) were infected with rotavirus. Multivariate logistic regression indicated there was no significant association between rotavirus infection to any of the HBGA predictors (Table 3).   Non-sec/le-(Le (a−b−) ) 2 (8.3%) 1 (33.3%) P [6] strains infected three Lewis-negative (Le (a−b−) ) children aged 5, 7, and 13 months. Two of the children were secretors possessing the blood type A and B while one was a non-secretor with blood type O.

Discussion
Globally, various studies have reported a putative association between the HBGAs and susceptibility to rotavirus infection, which may be driven by specific rotavirus strains as characterized by their VP4 genotype. Most have concluded that the presence of at least one of the genes (FUT2 and FUT3) encoding HBGA has been associated with greater susceptibility to rotavirus infection [17,29,30]. This was confirmed in this study with greater susceptibility noted amongst the blood group A and O children. Furthermore, several studies have examined the relationship between rotavirus vaccine response and the observed lower efficacy in sub-Saharan Africa and South Asia in populations categorized by their HBGA [22,23,[31][32][33][34]. Nevertheless, it remains unclear whether the HBGAs have a dramatic impact on rotavirus infection susceptibility or response to rotavirus vaccines. Recent data demonstrating specific binding domains within the H-1 type antigen and the P [8] strains may provide a mechanistic event by which this occurs [24], and this should be explored further in new studies. In the present study, we conducted the analysis of the H1-related blood group phenotypes in our local population to provide an early assessment of whether symptomatic rotavirus infection in South African children may be linked to blood group antigens.
South Africa was the first African country to introduce the rotavirus vaccine [35]. As observed globally, the introduction of the monovalent vaccine Rotarix had a great impact on the mortality and morbidity of rotavirus infection in South Africa [36,37]. During the period of this study, to examine the etiology of acute gastroenteritis disease in Northern Pretoria, South Africa post introduction (June 2015-December 2017), rotavirus accounted for 14.3% of diarrheal cases seeking medical attention or hospitalization. Most of the rotavirus positive children were hospitalized in DGMAH compared to outpatient treatment at OPHC. This may be an indication of more severe diarrheal disease caused by rotavirus requiring hospitalization and was corroborated by the observed higher rates of fever and vomiting in these hospitalized children. Interestingly, almost all rotavirus infected children had been vaccinated against rotavirus, having received at least one dose of Rotarix. Despite this, the effectiveness of the vaccine can be seen when comparing our findings to the 2003-2006 surveillance study pre-introduction, which reported 22.8% of rotavirus infections in children with gastroenteritis at DGMAH alone [38].
The introduction of the vaccine has had an impact not only on the reduction of rotavirus burden of disease, but also on circulating genotypes globally, including South Africa [39]. Characterization of the infecting genotypes revealed a diversity of strains, including those known to be commonly circulating globally such as G9P [8], G9P [6], G2P [4], and G1P [8], mixed strains (G3/G1, G3/G9), untypeable strains, and atypical strains such as G8P [4] and G3P [4]. Additionally, the genotype distribution displayed yearly fluctuations in this study, and this is a feature of rotavirus disease that has been reported previously in this region. Interestingly, the genotypes differed between the two sites which are located 23.6 km from each other. OPHC is an outpatient clinic and refers severe diarrhea cases requiring hospitalization to DGMAH, similar to other primary district hospitals and primary health care clinics in the catchment area of the tertiary care DGMAH. This wide catchment area which includes more rural areas which could have influenced the difference in the circulating genotypes at the two sites. Even so, year to year strain fluctuations, mixed infections, and untypeable and atypical rotavirus strains are common in Africa regardless of vaccination status of the population [28,40,41].
Rotaviruses are known to recognize HBGAs in a strain specific manner, suggesting that we may be able to map rotavirus genotype distribution according to HBGA profiles globally. If rotavirus infection is dependent on α1,2-linked fucose residue present in secretors as indicated in other studies [15,42], non-secretors should be somewhat resistant to some rotavirus strains and infection. In this study, the rotavirus infected children were mostly secretors (Lewis-positive secretors 79.6% and Lewis-negative secretors 14.3%) although this correlation was not absolute; 6.1% of Lewis-negative non-secretor children were infected by rotavirus. The secretor status may have played a role here.
Previous observational studies have reported conflicting results on the association between rotavirus and HBGA. Especially, the association of infecting rotavirus P-genotypes to the blood group antigens. For instance, in France [15], Nicaragua, and Burkina Faso [17], rotavirus P [8] exclusively infected secretors, which led to the conclusion that non-secretors may be resistant to rotavirus infection with P [8] strains. Recently, other studies have also observed P [8] infection in non-secretors, such as in South Africa [43], Spain [44], and Tunisia [21], although statistically, infections were more likely to be in secretors. Our study reports similar results in which P [8] infected both secretors and non-secretors but mostly secretors. Previously, rotavirus P [4] has been reported to share the same host preference with P[8] [9]. However, this was not observed in our study participants as P [4] infected almost exclusively Lewis-positive secretors; this has been reported elsewhere [45].
Furthermore, although the numbers were very low (n = 3), the association of P[6] strains with Lewis-negative subjects was noted, confirming previous results. Initially, it was suggested that rotavirus P [6] strains infect only Lewis-negative individuals, which was postulated as a reason for their prevalence among circulating strains in Africa where the Lewis-negative phenotype is common [9,29]. Unfortunately, we only identified three P [6] strains, and although we found that P [6] exclusively infected Lewis negative children in a population comprising of 11.2% Lewis negative profile, the numbers are too small to draw any real conclusion. Previously, P [6] rotaviruses were detected at approximately 9% in South Africa during the period 2006-2014 and were the third most common P-type [39]. This figure was 25% of circulating strains in northern Pretoria where this study was conducted, in 2003-2006 [38].
However, surveillance studies in other parts of the world have shown P [6] to infect Lewis-positive individuals too [17]. This was illustrated in in vitro binding assays as well, where P [6] weakly bound Lewis b antigens [16,45]. Altogether, this implies that P [4] and P [8] infection is significantly higher in Lewis-positive secretors and P [6] in Lewis-negative individuals, regardless of secretor status.
Interestingly, none of the 45 children characterized as Le (a+b−) were infected by rotavirus. It was previously shown in Nicaragua that children characterized as Le (a+b−) did not seroconvert when vaccinated with either Rotarix or RotaTeq [31]. Similar results were also observed in South African children, 7 days post-vaccination [46]. The Le (a+b−) phenotype lacks the fucose added by the FUT2 enzyme in the alpha 1,2 linkage which is present in secretors [47]. This proposes that the lack of alpha 1,2 fucose in this position is restrictive to vaccine replication [30] and may be protective of natural rotavirus infection.
Most children in this study (54.5%; 176/323) were phenotyped as Le (a+b+) and a similar proportion (53.1%; 26/49) were shedding rotavirus. This HBGA phenotype is considered uncommon globally, although it has been reported in 51.9% of Black Brazilians of African descent [48], supporting that Le (a+b+) may be common among African populations, as seen in our study.
Our study had few limitations. Firstly, relatively few cases of rotavirus diarrhea were detected as a result of the impact of rotavirus vaccination in South Africa which has significantly reduced diarrheal hospitalization and death [36,37]. In addition, the HBGA binding profiles of recombinant VP4/VP8* proteins of corresponding rotaviruses to saliva samples and genotyping of FUT2 and FUT3 genes of the saliva samples were not conducted. Previous studies that corroborated saliva binding assay results with genotyping showed compatibility within the results [31]. Disappointingly, the numbers of rotavirus genotypes with a VP4 P [6] were low (n = 3), which did not enable an evaluation of the putative association between P [6] and HBGAs as reported elsewhere.

Conclusions
In conclusion, this supplements the few studies conducted in African countries on the association of rotavirus to HBGAs. The present study shows that the presence (secretors) and absence (non-secretors) of HBGAs does not affect rotavirus infection but may impact the viral genotype diversity causing infection. Rotaviruses and HBGA are ubiquitous in different populations, although both show differences in frequency. Their interaction is complicated and even more so is their impact on geographical strain diversity and vaccine effectiveness. More studies are necessary to investigate the exact role of HBGAs as ligands for rotaviruses or even other ligands that could serve as rotavirus receptors.
Supplementary Materials: The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/v14122735/s1, Table S1: Distribution of demographic characteristics in rotavirus positive and negative by site.
Author Contributions: L.M.S. and K.R. conceptualized the study. K.R., I.P. and M.G. conducted the laboratory methods on the stool samples. P.H. and M.T. conducted the blood group antigen laboratory testing. A.D.S. and K.R. reviewed, analyzed, and interpreted the data. A.D.S., L.M.S. and K.R. drafted the manuscript. All authors have read and agreed to the published version of the manuscript. Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.
Data Availability Statement: Not applicable.