Application of miniaturized most probable number method for bacterial detection in water samples: detection of multi-drug-resistant Ralstonia insidiosa in drinking water

Water, a crucial component necessary for sustaining life, can be sourced from various consumables. The current array of available beverages provides options for obtaining drinking water. Ensuring the cleanliness and hygiene of water containers is crucial in averting the transmission of foodborne diseases, with over 200 ailments attributed to the ingestion of contaminated food and beverages, containing bacteria, viruses, parasites, or harmful chemicals like heavy metals (WHO 2022). The escalating public health concern not only strains healthcare systems and impacts tourism and trade but also significantly contributes to mortality rates (WHO 2022).

Foodborne disease, originating from the contamination of food and beverages, can manifest at any stage of the production delivery, and consumption chain. Environmental factors, such as water and soil pollution, coupled with unsanitary food storage and processing practices, can lead to various illnesses, ranging from diarrhea to severe conditions like cancer. In particular, gastrointestinal issues, often caused by diarrhea, pose a significant challenge globally, especially in low and middle-income countries and among children under the age of 5 (WHO 2022). Klebsiella pneumoniae is one of the important bacteria in foodborne diseases. K. pneumoniae is also included as a coliform bacteria that is used as an indicator for contamination in food, milk, and water. K. pneumoniae is the predominant coliform bacteria associated with foodborne diseases (Niyoyitungiye et al. 2020).

In Indonesia, the challenge of accessing clean water persists, primarily relying on sources such as rivers, lakes, reservoirs, and wells (Puspitasari 2012). Disturbingly, data from the Ministry of Environment in West Java reveal that 46 villages experienced water pollution between 2020 and 2022. The consumption of polluted water can have severe health implications, exemplified by an elevated microbial count leading to conditions like diarrhea. Notable bacterial contaminants include Staphylococcus aureus, Coliform, and Escherichia coli, serving as indicators of fecal contamination and adverse water quality conditions. A study by Genter et al. (2022) discovered alarming rates of E. coli in household drinking water sources in Bekasi and Metro Lampung cities, mirroring national data indicating fecal contamination in 7 out of 10 household drinking water samples.

The detection of pathogenic microbes in clean water plays a pivotal role in water management to prevent disease outbreaks. Standard tests, such as most probable number (MPN) and total plate count, often require intricate laboratory procedures, incurring high costs, and are consequently avoided by consumers and distributors. Existing alternative methods, like test strips and H₂S strip test, which utilize the properties of enzymes to detect coliform contamination with enzyme β-galactosidase production and E. coli with enzyme β-glucosidase production (Kontryana & Kikuchi 2021) always come with limitations, such as cost and the inability to determine bacterial quantities accurately. Another fast method on the market is MicroSnapTM Coliform which uses a bioluminogenic reaction which produces light when enzymes produced by Coliform bacteria react with a special substrate and the resulting light is measured with a luminometer. The weakness of this method is that the tool is expensive. As a response to these challenges, this study explores the application of a miniature MPN method for microbial quality testing of drinking water. Miniature MPNs, utilizing microplates, present a practical alternative to conventional MPNs, offering efficiency in time and media usage (Colla et al. 2014).

Based on the weaknesses of alternative methods for microbial quality testing of drinking water above, further study on exploring alternative methods is needed. MPN is more often used for drinking water testing due to its being easier for water samples. The disadvantage of MPN is that it requires many tools and it is not practical to have many samples in one go (Colla et al. 2014). Miniature MPNs are known to be a practical alternative to manual MPNs. Miniature MPN uses microplates and can count microorganisms in possible numbers, cutting down on time and media. Colla et al. (2014) developed a mini MPN to count Salmonella using modified semi-solid Rapapport Vassiliadis (MSRV) media followed by selection on chromogenic media (Colla et al. 2014). Coliforms are bacteria which can cause foodborne disease, especially in bottled and unbottled drinking water. This is based on Indonesian National Standard No. SNI 01-3553-2006 of the Indonesian Ministry of Industry and Trade that the maximum limit for total coliform bacteria numbers is < 2 in 100 ml of drinking water. The study carried out by these authors showed that the mini MPN method compared to the TPC technique was not significantly different (P > 0.05). The experiment was carried out with three repetitions. Statistical analysis used the non-parametric U Mann–Whitney t-test. The bacterial suspensions tested were 10 and 100 CFU/ml, with results of P 0.5 and P 4.0 rejecting H0. The experiments used parameters appropriate for laboratory conditions. However, the development plan for this kit is that it can be used anywhere outside laboratory conditions. Thus, in its development, it requires optimization in temperature, such as room temperature (RT) compared to 37° C, and also using plastic pipettes compared to laboratory micropipettes. Other parameters outside laboratory conditions such as sample inoculation outside the biosafety cabinet (BSC), and media storage time will also be investigated in the development of this portable tool.

This study was conducted from April to September 2023 in the Bacteriology Laboratory, Department of Medical Laboratory Technology, STIKes Mitra Keluarga, Indonesia. The research design employed an experimental study with variations in medium storage time, the type of pipette (automated micropipette versus plastic pipette), and the inoculation place (inside versus outside the BSC). The BSC utilized in this study was the JSCB-900SB BSC (JSR, Gongju, Republic of Korea), with incubation temperature as the independent variable. Bacterial count in CFU/ml served as the dependent variable. The interpretation of the MPN method adhered to the MPN table (Benson 2002).

K. pneumoniae isolates obtained from sputum in a private hospital were utilized in this study. Identification of the K. pneumoniae isolate was performed using the Automated Identification and Susceptibility Testing VITEK 2 COMPACT (Biomerieux, Paris, France). The isolate's confirmation involved growth on an endo agar medium. A liquid suspension of the isolate with concentrations of 10, 100, and 1,000 CFU/ml was employed for testing. To prepare the inoculum, colonies grown on endo agar were streaked into 3 ml of Luria broth medium in 10 ml sterile tubes. This culture was incubated with a shaker at approximately 28 °C for 24 h. Subsequently, 1 ml of the culture was transferred to a sterile Eppendorf tube, which was then centrifuged at 5,000 RPM for 2 min, resulting in the formation of a supernatant and bacterial pellet. The supernatant was removed, and 500 μl of sterile distilled water was aliquoted into the bacterial pellet, followed by resuspension. The bacterial suspension's optical density was measured using a spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA) at 600 nm. This suspension was adjusted to an OD = 1 at 600 nm. An OD = 1 corresponds to 10⁹ CFU/ml. Subsequently, this solution was serially diluted to achieve concentrations ranging from 10 to 1,000 CFU/ml (Ilsan et al. 2023).

We used commercial drinking water in a dispenser machine for this trial. The method followed the previous instruction in ‘Methods for mini MPN’. After incubation, the wells that were grown by bacteria in LB were streaked on Endo agar medium for confirmation. A single colony of bacteria from Endo agar then was tested for Gram staining and antimicrobial susceptibility toward several antibiotics using the diffusion disk method. The bacterial isolate was tested using Vitek 2 Compact automated identification and antimicrobial susceptibility testing.

Bacterial isolate was streaked on Mueller–Hinton agar (MHA) for 24 h incubation. The grown colony was then swabbed with a sterile cotton bud and suspended in 3 ml sterile 0.9% NaCl. The bacterial suspension was adjusted to 0.5 Mac Farland standard. Adjusted bacterial suspension was then swabbed by sterile cotton bud onto MHA. Antibiotic discs were placed on MHA containing the bacterial test. Antibiotic discs used were cefixime, ceftriaxone, meropenem, chloramphenicol, ciprofloxacin, and colistin sulfate. The inhibition zone was measured after 24 h of incubation. Susceptibility criteria interpretation followed CLSI (2024).

Bacterial isolate was grown on nutrient agar (NA), and then the colony was resuspended into sterile 3 ml 0.9% NaCl. Bacterial suspension was adjusted into 0.5 Mac Farland using a turbidimeter. Gram staining showed that the bacteria was Gram-negative (GN). GN identification card and AST antimicrobial susceptibility card for GN bacteria were used for identification and susceptibility testing using Vitek 2 Compact (bioMerieux, Paris, France). The result will show after 18 h of incubation.

Virulence screening test using O. fuscidentalis larvae following Ilsan et al. (2024) and Ilsan et al. (2023). Bacterial isolate was streaked onto MHA for 24 h of incubation. The bacterial colony was then sub-cultured into lactose broth, and incubated with shaking for 16 h. The broth culture was centrifuged at 12.000 rpm for 2 min, then the supernatant was removed. As much as 500 μl of sterile 0.9% NaCl was resuspended into a bacterial pellet. The bacterial suspension was adjusted into OD600 = 1 equal to 109 CFU/ml using a spectrophotometer. The bacterial suspension was then serially diluted to reach 106 CFU/ml. As many as 10 μl of 106 CFU/ml bacterial suspension was injected so the final total bacteria was 106 CFU/larvae. Ten larvae were injected for each isolate as a replicate. The injection site was the last left proleg of larvae. K. pneumoniae strain 122 from hospital patient sputum was used as a comparative strain. 0.9% NaCl was used as a negative control. Survival percentage was evaluated at 4, 8, and 24 h after injection.

The utilization of a simple and portable kit for detecting bacterial contamination in water samples, particularly drinking water, presents a promising solution for ensuring water sustainability. One significant aspect of bacterial contamination is the presence of the coliform group. Among these, K. pneumoniae is an important genus capable of causing foodborne diseases.

In this study, we aim to develop a prototype kit for the simple and portable detection of bacterial contamination. To facilitate mass application, several variables need to be thoroughly tested and validated. These include medium storage time (comparing 3-month storage to fresh), the type of pipette (comparing automated micropipette to plastic pipette), inoculation place (comparing inside to outside the BSC), and incubation temperature (comparing RT to 37 °C). The successful validation of these variables is crucial for the effectiveness and reliability of the proposed kit in practical applications.

In the beginning, we conducted tests on the mini MPN method for comparison with the standard total plate count (TPC) method. This testing was carried out in three replicates, each with two different concentrations, namely 10 and 100 CFU/ml. The results indicated no significant difference between the mini MPN and TPC methods at concentrations of 10 and 100 CFU/ml, with p-values of 0.5 and 4.0, respectively (P > 0.05) (Figure 2(a)).

Next, the 3-month storage of medium in mini MPN method was tested against fresh medium with a liquid mixture of 10 CFU/ml K. pneumoniae for evaluating the impact on the result. This result showed there is no significant difference with P 0.499 (>0.05) (Figure 2(b)). The standard laboratory incubation, i.e. 37 °C, was compared to RT incubation in mini MPN method with liquid 100 CFU/ml of K. pneumoniae. Incubation in RT is easier to do without the incubator instrument. The result showed there is no significant difference between these two groups with P 0.423 (>0.05) (Figure 2(c)). For the field mass application, the use of an automated micropipette will be more difficult, so we try to use a boiled-plastic pipette as an easier analogue for an automated micropipette. The mini MPN method of using an automated micropipette against a sterile-plastic pipette for inoculation showed no significant difference in result with P 0.603 (>0.05) (Figure 2(d)). Lastly, we compared the inoculation placed inside the BSC and outside the BSC with flame. The result showed there is no significant difference with P 0.11 (>0.05) (Figure 2(e)).

Drinking water is an essential substance for all living things, particularly humans. Drinking water, a vital resource for all living organisms, holds particular significance for human health. The presence of microbial contamination in drinking water can pose a serious threat to human well-being. Foodborne diseases, often transmitted through microbial contamination in both food and water, are a major concern. Coliform, a GN, non-spore-forming Bacillus capable of producing β-galactosidase, plays a crucial role in foodborne diseases. Traditionally used as an indicator for contamination in food, milk, and water, the predominant coliform bacteria associated with foodborne diseases include Klebsiella, Escherichia, and Enterobacter genera (Niyoyitungiye et al. 2020).

In India, a study revealed that Klebsiella spp. was identified in 116 drinking water samples, with a 15% occurrence rate (Kumar et al. 2013). Meanwhile, in Jambi City, Indonesia, the Klebsiella genus was detected in 4.8% of samples. All these studies employed biochemical tests for bacterial identification. Notably, K. pneumoniae, a member of the Klebsiella genus, has been found resistant to various antibiotics and carries virulence genes contributing to severe symptoms that are challenging to treat. Research in Singapore (Hartantyo et al. 2020) demonstrated the presence of K. pneumoniae in 21% of raw and ready-to-eat foods. Among the K. pneumoniae strains in this study, 8% contained at least one of four virulence factors, such as the wcaG gene, K1, K2, or K54. The wcaG gene enhances virulence by converting sugar, evading the phagocytic activity of macrophages (Yu et al. 2007). K54 is associated with a capsule exhibiting extended-spectrum β-lactamase activity (Jenney et al. 2006), while K1 and K2 are linked to aerobactin production (Yu et al. 2007; Siu et al. 2012). Intriguingly, K. pneumoniae from ready-to-eat poultry dishes was reported to be resistant to ampicillin, amoxicillin-clavulanic acid, tetracycline, chloramphenicol, and trimethoprim/sulfamethoxazole (Hartantyo et al. 2020).

The need for a simple and low-cost method for detecting harmful bacteria is crucial, especially for mass usage without the requirement of special instruments and skills. Several bacterial detection kits and methods have been developed thus far. Conventional methods such as MPN, total plate count (TPC), and membrane filtration necessitate numerous glass instruments and involve labor-intensive work. On the other hand, enzymatic, immunological, and nucleic acid-based methods require higher costs (Tambi et al. 2023). Some potential challenges will be faced when this prototype is applied on a larger scale. Firstly, we feel the user will find it difficult to use the low-cost plastic pipette. Every person has their own power when they press the plastic pipette for inoculation. The adjustable plastic pipette seems interesting to solve this problem. Secondly, although this study suggests users should use flame for aseptic inoculation, the inoculation place seems to play an important role in asepticity. We suggest using cardboard boxes as a low-cost alternative, to minimize aerial contamination.

In the application of this modified mini MPN, we analyzed one of the commercial drinking waters, including bacterial enumeration, isolation, biochemical identification, antimicrobial susceptibility test, and virulence screening test (Figure 1). The number of bacteria in this drinking water sample was 87 CFU/ml using the TPC method on Plate Count Agar (PCA), and 70,3 MPN/ml using the mini MPN method. Bacteria which was grown in the MPN plate was then streaked onto selective and differential endo agar medium. After 24 h of incubation, only one morphology colony was grown, thus encoded with D isolate. Bacterial automated identification using Vitek 2 Compact showed D isolate was Ralstonia insidiosa. Ralstonia genera is a GN bacterium that was first introduced as Burkholderia genera in 1973. In 1995, Yabuuchi et al. (1995) proposed to separate Ralstonia genera. Ralstonia can survive in limited nutrition such as water environment, and also has been found in municipal water and medical-related water purifications (Vincenti et al. 2014; Chen et al. 2017). In China, R. insidiosa has been reported to cause meningitis following lumbar infections. This infection exhibited severe symptoms, such as headache, fever, and dizziness (Liao et al. 2023). A case in Turkey reported an outbreak of bloodstream infections caused by R. insidiosa from contaminated heparinized syringes (Tuzemen et al. 2022). Recently, R. insidiosa has been reported to cause bronchial pneumonia in a 2-year-old child in China. The specimen containing R. insidiosa has been detected in bronchoscopy and bronchoalveolar lavage fluid culture (Lin et al. 2023).

In this study, R. insidiosa strain D was resistant to at least four classes of antibiotic including aminoglycosides (Amikacin and Tobramycin), Penicillins (Ampicillin-Sulbactam), Cephalosporin (Cefepime, Cefuroxime, Ceftriaxone, and Cefixime), Carbapenem (Doripenem, Imipenem, and Meropenem) (Table 1). Bacteria are categorized as multi-drug resistant (MDR) if not susceptible to at least three classes of antimicrobial agents (Magiorakos et al. 2012). R. insidiosa strain D in this study was susceptible to Fluoroquinolones (Ciprofloxacin, Ofloxacin), Glycylcycline (Tigecycline), Phenicols (Chloramphenicol), and Polymyxin B (Colistin sulfate). The previous study showed from 15 R. insidiosa from the environment, quinolones and folate pathway inhibitor class were the most promising. These R. insidiosa isolates are mostly isolated from various industrial purified water systems including Millipore laboratory in Ireland. R. insidiosa from cell culture contamination also was resistant to many antibiotics, including beta-lactamase inhibitors, cephalosporin, carbapenem, and aminoglycosides. This isolate was only susceptible to quinolone class (Nurjadi et al. 2020). In another study, Ralstonia from cystic fibrosis has high resistance levels. Interestingly, the genomic analysis showed it lacked transposable elements despite it having many resistance genes (Fluit et al. 2021). It seems resistance genes in Ralstonia hardly disseminate to other bacteria.

We concluded that this mini MPN can be an alternative to the TPC method for enumerating K. pneumoniae in drinking water. For simple and portable usage of this prototype, some variables can be considered, such as the use of a simple plastic pipette, sample inoculation which can be performed outside of BSC with flame, the use of 3-month storage medium, and RT for incubation. We have tried this modified mini MPN with commercialized drinking water as a sample. In this sample, we found MDR R. insidiosa which was resistant to at least four antimicrobial classes, including aminoglycosides, penicillins, cephalosporin, and carbapenem. Vitek 2 Compact was performed for bacterial identification and antimicrobial susceptibility testing, then confirmed by the disk diffusion method. A virulence screening test in O. fuscidentalis larvae showed R. insidiosa strain D had a low virulence.

This study was supported by Badan Riset Inovasi Nasional, Indonesia with program Riset Inovasi Indonesia Maju (RIIM) with grant number 164/IV/KS/11/2023.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.