Detection of Salmonella genes in stool samples of children aged 5 years and younger in urban and rural areas of Bangladesh

Introduction: Typhoid incidence in children is higher in urban areas than in rural areas of Bangladesh. This study examined whether healthy urban children harboured higher levels of Salmonella genes than healthy rural children. Methodology: Stool samples from 140 children were studied: 70 from rural areas and 70 from urban metropolitan areas. Results: The stool samples of urban children contained more Salmonella genes (median 4, IQR 3-4) than those of rural children (median 3, IQR 3-4). This suggests that urban Bangladeshi children have more Salmonella genes in their guts than rural children. Especially, in those under 12 months of age, the Salmonella gene prevalence in urban children was unique. They had more Salmonella genes (median 4, IQR 4-5) than rural children in the same age group (median 3, IQR 2.5-4). We also found more Salmonella genes in urban children who drank tap water (median 4, IQR 3-5) than in rural children whose water source was tube well water (median 3, IQR 2-4) and boiled pond water (median 3, IQR 3-3.5). However, there was no significant difference of Salmonella genes between urban children who drank tap-water and children whose water source was a tube well (median 4, IQR 3-4). Conclusions: These data suggest that the urban environment, including the drinking water supply system, increases the likelihood of healthy children in urban areas harbouring more potentially pathogenic Salmonella organisms in their gut than found in rural healthy children.


Introduction
Typhoid fever is a critical health problem in many developing countries, including Bangladesh. An estimate for 2017 showed 14.3 million cases of typhoid and paratyphoid infections globally [1]. Also, nontyphoidal salmonellosis, which is related with gastroenteritis and sepsis, was estimated at 3.4 million cases in 2015 [2].
Bangladesh is one of the most densely populated countries in the world with an estimated population of 162.7 million and a population density of 1,103 persons per square kilometre in 2017 [3]. It is a riverine country and routinely faces disasters from monsoon floods and riverbank erosion that cause homelessness and landlessness and compel migration to cities in search of work. In one study, about 10% of the respondents migrated because of a natural disaster and 70% of those migrants remained permanently in urban areas [4]. This in-migration creates huge infrastructure challenges for cities. Around 37% of the Bangladeshi population lived in urban areas in 2018 [5].
Bangladesh has a high incidence of salmonellosis and typhoid fever, which disproportionately affect younger children, with the highest incidence in children < 5 years old [3,6]. The incidence of invasive salmonellosis is high among residents of the densely populated urban communities in Dhaka and in areas adjacent to the Dhaka metropolitan area [7]. This study reported 2.0 typhoid fever episodes (95% confidence interval [CI] 1.5-2.8)/1000 person-years in 2010 using blood culture system. They also found that the incidence in children < 5 years old was 12-fold higher than the incidence among children ≥ 5 years of age.
In Bangladesh, the incidence of nontyphoidal salmonellosis has historically been higher in metropolitan areas than in rural villages. A study conducted in 1977-1979 found that of 214 isolates from blood and stool from metropolitan Dhaka, 0.3% of 66,341 cultures were positive compared with 12 (0.04%) of 27,265 positive cultures of isolates from rural areas [8][9][10]. A study of multidrug-resistant pathogenic bacteria in the gut of healthy young children in Bangladesh concluded that the gut of young children, below the age of 5 years, was an important reservoir for pathogenic bacteria [9]. Another study reviewed 19,265 blood cultures from an urban paediatric hospital in Dhaka. Of these, 855 (4.4%) were positive by culture for ST/PTF. The same study found 25 (0.2%) of 15,455 blood cultures from a rural hospital in Mirzapur were positive for Salmonella [10]. A 2005 community-based study in an urban slum in Bangladesh reported an overall incidence of salmonellosis of 3.9/1000 personyears and a higher rate in children aged 0-4 years (18.7/1000 person-years) [11]. Another study found a higher incidence in children aged < 5 years (10.5/1000 person-years) with an overall incidence rate of 2.0/1000 person-years [12].
Previous studies established that the pathogenesis of enteric fever depends on several factors, including the infecting species and infectious dose. These organisms, when ingested in high doses, survive exposure to gastric acid before gaining access to the small bowel, where they penetrate the epithelium, enter the lymphoid tissue, and disseminate via the lymphatic or hematogenous route [13][14][15].
Several studies showed a higher incidence of typhoid fever in urban children than in rural children. These studies attributed this higher incidence of typhoid fever in urban children to several factors, including high population density in slum areas, poor sanitary conditions, drinking unsafe water, and ingestion of contaminated food. No study, however, has examined whether the presence of Salmonella genes in the guts of children play a role in typhoid fever (i.e., whether the presence of higher numbers of Salmonella genes in the intestines of healthy urban children is a predisposing factor for higher incidence of typhoid fever in urban children compared with rural children).
Most studies agree that both environment and infrastructure are related to a high incidence of typhoid fever in young urban Bangladeshi children. However, to date, no studies have assessed whether the intestines of urban children harbour more Salmonella, even when they are healthy (i.e., not suffering from salmonellosis), than the guts of healthy rural children.
The hypothesis of this study was that in Bangladesh healthy urban children harbour more Salmonella genes in their gut as commensal bacteria than rural children. This reasoning was supported by prior studies [8,16], including one that showed children in urban areas were more likely to carry Escherichia coli [16].
To assess our hypothesis, we examined stool samples of healthy urban and rural children using molecular detection of Salmonella genes by two-step nested PCR to determine the mean number of Salmonella genes present in the intestines of the sample children.

Stool sample collection
In accord with this study's objective, we collected samples in two settings: the urban district of Narayanganj and the rural district of Sirajganj. The 2011 population densities of the two districts were 4,308 and 1,290 per square kilometre, respectively [17]. In order to ensure representative sampling, we collected samples in five separate administrative areas within Narayanganj (Fatulla, Godenail, Kadamtoli, Matuail, and Siddhirganj) and eight areas of Sirajganj (Belkutchi, Kamarkhand, Kazipur, Rayganj, Sadar, Shahjadpur, Tarash, and Ullapara). Narayanganj is adjacent to Dhaka, a metropolitan city, and has similar sanitary and water-supply infrastructure as well as overcrowding. The study enrolled a total of 140 healthy children aged < 5 years (70 urban and 70 rural). Household selection was random, and samples were accepted only from children who could fulfil all study criteria. We also collected information from each child's parents on the child's eating habits, personal hygiene practices, and the source of the family's drinking water ( Figure 1). We began by collecting samples in the rural district of Sirajganj (in May 2016), which was followed by sample collection in the urban district of Narayanganj (in June 2016). During sample collection, healthy children lived at home and did not have any disease/symptoms of gastroenteritis (e.g., vomiting, diarrhoea, fever) or history of taking antibiotics during the prior 6 months. Health workers visited the children at home, obtained informed and written consent from their parents, and collected demographic information such as age, sex, personal hygiene, and dietary habits the day before obtaining samples. The parents collected the samples at home and put them in a refrigerator or in the supplied icebox (if the household lacked refrigeration) at 4°C. The next day, the health workers collected stool samples in the containers provided to the parents, placed them in an icebox, and immediately transported the samples to a laboratory where samples were stored at -20℃ until processing for DNA extraction.

DNA extraction
DNA was extracted from 180 to 200 mg of stool samples according to the manufacturer's instructions using NucleoSpin Stool DNA extraction kit (Macherey-Nagel, Düren, Germany). The amount of eluted DNA was quantitatively measured by Nanodrop 2000 Spectrophotometer (Thermo Fisher Scientific, Waltham, MA) and stored at -20℃. The integrity of DNA was estimated by using 2% (w/v) agarose gel electrophoresis in 0.5X TAE buffer. We mixed 1 μL of Midori Green with 10 μL of extracted DNA, then loaded on the gel. These DNA samples were used for PCR.

Molecular detection by nested PCR
We used nested PCR with a thermal cycler (Veriti 96-well thermal cycler, Applied Biosystems, Foster City, CA) to screen for the presence of Salmonella genes in healthy children's stool specimens. In primary PCR, six Salmonella gene targets were used for amplification (Table 1). Salmonella gene targets included 16S rRNA [18], Salmonella pathogenicity island I gene (hilA) [19], Salmonella enterotoxin gene (stn) [20], invA gene [21], Fur-regulated gene (iroB) [22], and histidine transport operon (hisJ) [23]. For primary PCR, we used TopSimple nTaq premix (Enzymomics, Daejeon, Korea) for amplification. Total reaction volume was 20 μL (10 μL of pre-mix and 10 μL of master mix) (forward primer, 1 μL; reverse primer, 1 μL; stool DNA, 1 μL [10 ng/μL]; and ddH2O, 7 μL). All primers were amplified for 35 cycles (Table  1). For secondary PCR, 1 μL of product from primary PCR was used to amplify the target band with different sets of primers (Table 2). PCR conditions were as follows: initial denaturation at 95℃ for 5 minutes, denaturation at 95℃ for 30 seconds, annealing at 58℃ for 30seconds, extension at 72℃ for 45seconds, and final extension at 72℃ for 10 minutes. Amplification products and sizes were determined by electrophoresis using 2% agarose gels. PCR amplification was considered positive if the amplicon matched the anticipated size and negative if no amplicon was detected ( Figure 2). Salmonella gene counts were scored according to the number of positive Salmonella gene bands shown in the nested PCR using multiple replications of single stool DNA samples and each of the Salmonella gene primer pairs. The primer set used for 16s rRNA detection failed to amplify genes of some Salmonella organisms but was also negative for many genetically related intestinal bacteria (e.g., Citrobacter, Klebsiella, Proteus, and Escherichia species). Some genes including hisJ, iroB, and invA are well-conserved in non-Salmonella species but are not detected by Salmonella-specific primers. Primer pairs for other factors were not tested for non-Salmonella species in other studies [24].

Statistical Analysis
We used the Shapiro-Wilk test to determine whether Salmonella gene values were normally distributed. Multivariable logistic regression analysis was used for comparing the effects of multiple factors to the binary-categorized Salmonella gene number (detected genes ≤ 3 or detected genes ≥ 4) in rural and urban children (SPSS software version 12.0; SPSS, Chicago, IL, USA). Statistical significance for continuous variables among and between groups was assessed by the Kruskal-Wallis test. The Chi-square test or Fisher's exact test was used to compare the proportion of each factor in rural and urban areas (GraphPad Prism software for Windows, version 5.01). Differences were considered significant at p < 0.05.

Ethics Statement
The Department of Biomedical Sciences, Seoul National University (SNU) College of Medicine, Republic of Korea, and the Department of Microbiology, Bangladesh University of Health Sciences (BUHS), Dhaka, jointly carried out this study from January 2016 to May 2018. Because the studychildren were minors, we obtained written informed consent from their parents, who voluntarily agreed to participate in the study after the study details were explained to them. All personal information has been kept confidential -as per ethics requirements. This study was approved by the ethical review committee of BUHS, Dhaka.

Salmonellosis in urban and rural children
We first tested five stool DNA samples from Salmonella-culture-positive patients and all samples produced the expected band sizes (16S rRNA, 215 bp; iroB, 201 bp; hilA, 281 bp; hisJ, 231 bp; invA, 197 bp;   Stn, 132 bp) for every Salmonella gene assessed by nested PCR (Figure 2A). Of the 140 stool samples analysed, seven (3 from rural areas [Sirajganj] and 4 from urban areas [Narayanganj]) were negative for 16s rRNA.
We then used Salmonella gene counts (0-6) as determined by positive Salmonella gene bands in nested PCR to score all samples (Supplementary Table 1). We found significantly more amplified Salmonella genes in urban samples than in rural samples ( Figure 3A and Figure 3B). In this study, the stool samples of urban children contained more Salmonella genes (mean, 3.92 per sample) than found in rural children (mean, 3.13 per sample).

Drinking-water source and Salmonellosis in children
As water is an important medium of Salmonellosis transmission, we examined the findings for each drinking-water source. Of the rural children studied, 38 (54.3%) drank water from tube wells and 32 (45.7%) consumed boiled pond water. Among the urban children, 55 (78.6%) consumed tap water and 15 (21.4%) water from tube wells ( Table 3).
Children who drank tap water had the highest number of Salmonella genes (mean, 3.95 per sample), followed by children who drank water from tube wells (urban and rural tube wells, respectively, mean 3.80 and 3.24 per sample) and boiled pond water (mean 3.03 per sample) ( Figure 4). These differences suggest that water source may affect the prevalence of Salmonella genes. However, in the multivariable analysis, water supply system did not significantly affect the Salmonella gene number when other factors were adjusted constantly (Supplementary Table 2).

Personal hygiene and Salmonellosis in children
As Salmonellosis is also spread by food, personal hygiene habits may have an impact of Salmonellosis in children. We, therefore, also sought information about each child's hand-washing habits and whether the household used a pit latrine with or without a slab. Of the rural children, 33 (47.1%) used pit latrines without a slab and did not wash their hands after visiting the  latrine or before eating food and 37 (52.9%) used pit latrines with slabs and washed their hands after defaecation and before eating food; among the urban children, 15 (21.4%) used latrines with flush water and washed their hands after visiting the latrine and before taking food and 55 (78.6%) used modern toilets and washed their hands after defaecation and before eating food. However, hygiene habits did not have a significant impact on the number of salmonella genes detected (Table 4).

Salmonellosis in children: Does it vary with age and sex?
Ages of the children studied by locality (rural and urban) were as follows: 0-12 months, 16 rural ( We analysed the data by age group (Figure 5). Of interest, children in the under 12-month urban group had significantly higher Salmonella gene scores than found in rural samples (p < 0.001) ( Figure 5A) or in other urban age groups. Urban children under 12 months of age had more Salmonella genes (mean, 4.29) than rural children in the same age group (mean, 2.94). By gender subgroup in rural or urban areas, there was no significant difference in numbers of children harbouring Salmonella genes (Supplementary Table 2).

Discussion
We collected stool samples in both rural and urban settings to compare the composition of gut microbiota of healthy children in two different living settings. We selected Sirajganj, a rural district in north-eastern Bangladesh, a typical Bangladeshi district (zilla), composed of a small town as its headquarters and several sub-districts (upazillas). Some of the villages have civic facilities such as tube well water and better access to health services; others still rely on traditional water sources (i.e., open ponds). These areas lack municipal sewerage systems and water supplies.
For our urban samples, we selected Narayanganj in the outskirts of the Bangladeshi capital, Dhaka. Some of Narayanganj is on the bank of Shitalakshya River, a tributary of the Brahmaputra River. Samples were collected from children in the Narayanganj City Corporation area, which is relatively well-serviced; however, it is an industrial area and is densely populated. Civic facilities are available, but due to congestion and the mostly unplanned and haphazard development, water lines occasionally are connected accidentally to the sewerage system and supply polluted tap water to most households. As this is an industrial area, many women work, mostly in factories, precluding their ability to breastfeed their children during their long working hours. Instead, they depend on non-breast milk, such as milk from cows or powdered milk. They maintain relatively good levels of personal hygiene and most have access to tap water for drinking.  . Urban children whose tap water harbours more Salmonella genes.
Many women in rural Sirajganj breastfeed their children at least until the age of 24 months. Rural children receive little protein compared to their urban counterparts and primarily eat carbohydrates and vegetables. A study in 2013 found that 80.6% of the respondent mothers in a Bangladeshi village were housewives [25]. The study also found that 70.7% of those mothers breastfed their children and 75.9% fed colostrum to their babies. Most mothers (92.6%) continued to breastfeed their children even if the child was ill. The study concluded that more rural women breastfed their children than the corresponding national average.
Although our study did not include examining a direct relationship between breast-feeding and less incidence of Salmonellosis, we have examined the existing literature on the same and found that breastfeeding indeed positively affect the incidence of Salmonellosis in children. A 1980 study noted vigorous responses of colostral and breast milk cells against Salmonella spp. [26]. The study concluded that colostral and breast milk cells were demonstrated to be more active against Salmonella than blood neutrophils. Another 2004 study found a strong association between having a liquid diet other than breast milk only and sporadic infant salmonellosis, which suggests that breast-feeding prevents infant salmonellosis [27].
Since Bangladesh is in the tropics, we also considered the possibility that seasonal variation might affect our results. We collected the rural samples in May 2016 and the urban samples in June 2016, thus minimising the collection time gap between the two sample groups. A study found two seasonal peaks of Typhoidal Salmonella (Salmonella enterica serovar Typhi and Paratyphi) in urban Dhaka in January and February and from August to November. According to the same study, rural Matlab had a single seasonal peak from August to November, but those seasonal peaks were insignificant [28]. Another study show there was no significant seasonal difference between May and June in Asia [29].
This study relied on PCR findings of extracted DNA from stools of healthy children. Isolation of live Salmonella from the samples collected would have been desired, but this was not possible as half of our stool samples were collected in remote rural areas, where we lacked laboratory facilities that would have enabled us to isolate live organisms from the samples. This was a limitation of this study.
The study's sample size was also relatively small, which may have been due to the strict sample criterion (clinically healthy children with no history of gastroenteritis and/or antibiotic use in the 6 months prior to the sample collection) that we followed in this study. Thus, finding the right samples was extraordinarily difficult. We also had budgetary limitations in part caused by working in two locations: Bangladesh for sample collection and DNA extraction, and South Korea for PCR.
Finally, we were unable to conclusively establish a direct correlation between the higher presence of Salmonella organisms in the intestines of urban children, who also had a higher incidence of typhoid fever, compared with rural children. We do anticipate that the study results may be useful in future studies in areas with high prevalence of ST/PTF (Salmonella Typhi/Paratyphi). Fast diagnosis of ST/PTF through a non-culture-based method (PCR) would aid the development of preventative measures, including routine screening that can be followed by early treatment -as opposed to time-consuming diagnosis of ST/PTF through traditional culture methods. This, in turn, should have a positive impact on reducing typhoid fever incidence in both urban and rural young children in Bangladesh and in countries with similar milieu.
Culture of organisms (TCM, the traditional culture method) remains the reference standard for identification of bacteria. Although culture remains the preferred diagnostic method for detection of bacteria, it is time-consuming and may fail to isolate a particular organism due to external factors, such as absence of high specificity or sensitivity. This study also states that only 40-60% of blood cultures from patients in the early phase of infection are positive [30]. Molecular detection methods, on the other hand, are suitable for rapidly identifying pathogens in human excreta as these methods are highly sensitive. PCR can detect minute quantities of the DNA of specific pathogens through amplification of a defined DNA segment and by discriminating on one reaction between different organisms even if they are closely related [31,32].
We hypothesized that urban children in Bangladesh may harbour more Salmonella genes in their intestines than children in rural areas and that this higher level of Salmonella might be the reason that urban Bangladeshi children have a higher incidence of typhoid fever than rural children. Our results agree with earlier findings [8,11]: Urban children in Narayanganj have a higher incidence of salmonellosis than rural children in Sirajganj. We also found that the young urban children of Narayanganj (aged 0-12 months) harbour more Salmonella genes (p < 0.05) than rural Sirajganj children in the same age group. Although urban children aged 13-60 months also harbour more Salmonella genes than rural children in the same age group, the difference was significant only for the youngest age group. This finding agrees with earlier studies that found typhoid fever disproportionately affects younger children, particularly children < 5 years old [11,12].
When we looked at children < 5 years old in urban Narayanganj and rural Sirajganj, the difference was not sufficiently strong to reach a definitive conclusion regarding a correlation of greater number of Salmonella genes in the intestines of healthy urban children as a predisposing factor for the higher incidence of typhoid fever. Even though we could not conclusively establish a correlation between greater numbers of Salmonella in the intestines of healthy urban children and the higher incidence of typhoid fever in urban than rural children, our findings may be a reference point for future studies and enable a definitive affirmative or negative conclusion to this hypothesis.
Of interest, our study found that the urban children in Narayanganj who drank tap water harboured more Salmonella genes than the urban children who drank tube well water and their rural counterparts in Sirajganj who drank water from tube wells and ponds (boiled water). Of note, the rural people of Bangladesh now usually drink boiled pond water due to years of awareness campaigns. Of the children studied, 38 of the rural Sirajganj children drank tube well water and 32 boiled pond water. Among the urban children, 55 consumed tap water and 15 tube well water.
In this study, children who drank tap water had the highest number of Salmonella genes (average, 3.95 per sample). Urban children who drank tube well water had a mean of 3.80. Rural children who drank tube well water and boiled pond water had means of 3.24 and 3.03 Salmonella genes, respectively. These findings are similar to earlier findings [33][34][35] and reinforce the notion that the lack of safe drinking water remains a major factor in the transmission and distribution of salmonellosis in urban and rural populations, particularly among young children.
A previous study suggested that both contaminated surface water and piped water could amplify the likelihood of waterborne infections among people living in the Dhaka metropolitan area, particularly of salmonellosis [33]. The authors also found that in highrisk areas (overcrowding, poor housing conditions, inadequate sanitation), 72.7% of residents had low quality-of-life (QOL), 18.2% had medium QOL, and only 9.08% had high QOL. The study found that unplanned urbanization, higher population density, and lack of critical urban infrastructure in the Dhaka metropolitan area had a considerable impact on the transmission and distribution of salmonellosis.
Another study, based on HPC (Heterotrophic Plate Count), found that all tap water sources of Dhaka city restaurants were highly contaminated [35]. The HPC of the Dhaka Water and Sewerage Authority was between 1.2 and 5.4 × 10 4 cfu/mL compared with World Health Organization recommended levels of 100-500 cfu/mL [34]. Some 90% of the samples were contaminated with one or more of the three potential pathogenic species-Vibrio, Shigella, and Salmonella.
Overcrowding and unsanitary urban conditions, particularly lack of access to safe drinking water, have been linked to the high incidence of salmonellosis in metropolitan children. One study found that 62.5% of school-age children in Dhaka city who consumed piped water without boiling were positive for Salmonella as were 20.8% of those who drank piped water after boiling and 16.7% who drank tube well water [35].
Consumption of contaminated water, vegetables, and fast-food from restaurants also have been identified as major sources of Salmonella infection in Dhaka city residents [36]. This study reported that availability of junk food in urban areas, which is not usually found in rural Bangladesh, also contributes to the high incidence of salmonellosis in urban children. The study found that 79.2% of children who ate both home-cooked and junk food were positive for Salmonella against 20.8% who were accustomed only to food prepared at home. Another study found that prevalence of typhoid fever was higher among children of school-age groups (66.7%), children who drank unsafe drinking water (58.3%), and those who ate junk food (72.9%) [35].

Conclusions
As a consequence of environmental, infrastructural, and other reasons, the healthy urban children of Narayanganj in the < 5 years age group harboured more Salmonella genes in their intestines than the rural children of Sirajganj of identical ages. Further research in this area should focus on why the relationship between prior presence of Salmonella and the incidence of typhoid fever differs in urban and rural healthy children. Moreover, there is a need for studies to concentrate on developing preventive measures such as routine screening and early treatment of salmonellosis to reduce mortality and morbidity of young children. Such measures would be particularly effective in developing countries that lack significant investment in public health infrastructure.  /pit latrine, wash  13-24  Breast milk and soft food  Tube well  Matuail  Male  +  +  +  -+  -4  1   Urban  19  Flush/pit latrine, wash  13-24  Breast milk and soft food  Tube well  Matuail  Male  +  +  +  -+  -4  1   Urban  20  Flush/pit latrine, wash  0-12  Breast milk and soft food  Tube well  Matuail  Female  +  -+  -+  -3  0   Urban  21  Modern WC, wash  13-24  Breast milk and soft food  Tap water  Kadamtoli  Male  +  -+  +  +  -4  1   Urban  22  Modern WC, wash  0-12  Breast milk  Tap water  Kadamtoli  Female  +  -+  +  +  +  5  1   Urban  23  Modern WC, wash  0-12  Breast milk  Tap water  Kadamtoli  Male  +  -+  +  +  +  5  +; gene-specific band positive, -; gene-specific band negative. Pit latrine w/o slab, w/o wash indicates children use a pit latrine without a slab and do not wash hands afterwards; flush/pit latrine, wash indicates children use a flush or a pit latrine and wash their hands after visiting latrine and before eating food; modern WC, wash indicates children use modern toilet, wash hands after defecation and before eating food. Under Category, children were categorized by the numbers of genes detected: 0, 1-3 genes detected; 1, 4-6.