Assessment of the presence of Vibrio cholera and detection of toxigenic Vibrio cholerae in river sources within the Vhembe District Municipality Limpopo province of South Africa

Vibrio cholerae (V. cholerae) is a Gram-negative, rod-shaped, non-invasive aquatic bacterium that is transmitted via faecal-oral routes because of consuming contaminated food or drinking contaminated water (Rashid et al. 2017; Lemaitre et al. 2019). The most common notable human Vibrio spp includes Vibrio cholerae, Vibrio parahaemolyticus, Vibrio vulnificus, and Vibrio fluvialis (Chakraborty et al. 2006), with Vibrio cholerae serogroups O1 and O139 recognised as the principal etiological agents that cause cholera (Lemaitre et al. 2019).

The most typical symptom of cholera is sudden, severe loose stool, which might also include some vomiting (Nadri et al. 2018). If left untreated, these symptoms can result in dehydration and even death (Dutta et al. 2011). Qasem & Rabbani (2023) have also reported that dehydration as a result of vomiting and severe watery stools can lead to shock, hypokalaemia, pulmonary edema, and renal failure. Oral and intravenous rehydration therapy with a salty solution are the usual treatment therapies (Clemens et al. 2017).

A study done by Falconer et al. (2022) reported that globally, an estimated billion of people are at risk of cholera transmission and further estimated that there are two million cases of cholera each year, with an annual fatality rate exceeding 100,000, with many of these deaths affecting children under the age of five years. Caribbean America, Southeast Asia, and sub-Saharan Africa are some of the areas with the highest incidence rates of cholera (Ali et al. 2015). According to Legros (2018) and Spiegel et al. (2016), cholera outbreaks have been reported throughout the past 10 years in several sub-Saharan nations, including Malawi, Tanzania, Zimbabwe, and South Africa, as well as a handful from the Middle East (Yemen, Iran, and Iraq) and the Caribbean (Haiti). The primary causes of the cholera outbreak in these areas were attributed to poverty, poor hygiene, and a lack of access to clean water and sanitation (Ali et al. 2015; D'Mello-Guyett et al. 2020). The ongoing outbreak of cholera infections are also made worse by elements such as inadequate infrastructure, natural disasters (flooding, mud slides, etc.), and inadequate integrated water and sanitation management policies (Kazaji 2015; Kativhu et al. 2021).

According to Küstner et al. (1981), the first cholera outbreak in South Africa occurred more than 50 years ago. Generally, South Africa is not regarded as a cholera endemic area because most outbreaks are caused by importation from neighbouring cholera-ravaged nations like Mozambique, Zimbabwe, and Malawi (Le Roux et al. 2020). This was demonstrated by the outbreak that occurred in Mpumalanga, Limpopo, and KwaZulu-Natal between 2010 and 2014 (Okoh 2018). These outbreaks totally caught South Africa off guard because the country was not ready to deal with the extent of such at that particular period. The first cases in this outbreak appeared in Mozambique, Malawi, and Zimbabwe and then spread to South Africa through the Musina border post as people travelled with the disease (Smith et al. 2021). Other periodic outbreaks have been documented in South Africa alone, particularly in heavily populated and remote locations where access to clean water and proper sanitation are difficult to come by (Hemson & Dube 2004; Bazaanah & Mothapo 2023). The continuous pollution of the existing freshwater resources and catchments since 2015 is the main cause of the recent cholera outbreak in June 2023 in Hammanskraal, Pretoria, that claimed over 30 lives and had dozens admitted to various hospitals and clinics around the community (Herbig 2023). This outbreak was due to poor sanitation, poor quality of drinking water, and discharge of partially and untreated wastewater from Rooiwal WWTP directly downstream into the Apies River (Herbig 2023). According to South African Government News Agency (2023), the problem was exacerbated by the non-compliance of the effluent discharged by the two dysfunctional facilities (Temba Water Treatment Plants (WTP) and Rooival Wastewater Treatment Plant (WWTP) respectively as par South African water quality standards.

The fast detection and identification of cholera are crucial for managing the outbreak and stopping the spread of the disease (Ganesan et al. 2020). However, in peri-urban and rural areas frequent water quality testing is not done, and river sources do get contaminated due to indiscriminate use of water by various activities such as vehicle washing, swimming, bathing, laundry, and dumping of solid wastes by the surrounding communities into the catchment (Davison et al. 2005; Traoré et al. 2016). The primary goal of this study was therefore to determine whether toxicogenic V. cholerae was present in the river water sources within the Vhembe municipality district, which could be harmful to the communities relying on these resources for domestic, agricultural, and recreational purposes.

The Vhembe District Municipality is situated in the northern region of the province of Limpopo in South Africa. It borders the districts of Mopani and Capricorn to the west and east, respectively (Vhembe District Municipality 2020). The district is also bordered by Kruger National Park to the northeast, Zimbabwe to the north, Botswana to the northwest, and Mozambique to the southeast. Vhembe (2013) states that the district has a total population of 1,393,949 and occupies 21,407 square kilometres of territory. The average annual temperature and rainfall, respectively, ranged between 20 and 26 °C and from 450 mm on the low-lying plains to more than 2,300 mm in the mountains (Murei et al. 2023). Few of the rivers in the Vhembe area are seasonal, which include Mphongolo and the Little Letaba where sampling did not take place because they were dry, with the majority being perennial. The permanent rivers (n = 12) that were sampled for this study are shown in Table 1 along with their corresponding coordinates.

Water samples were collected once off in 2-litre sterile plastic bottles for three months between August and October 2023 using aseptic techniques. The sample sites were collected from an upstream and a downstream point in each river. The sampling sites (upstream and downstream) were decided with reference to the settlements or agricultural or other activities such as brick making, car washing etc. in between the sampling sites of each river. These references were also to assist in establishing whether they are negatively compromising the quality of water after (downstream), or the quality of water was already compromised before the chosen reference point (upstream) for each river. The collected water samples were transported to the laboratory at the University of Venda in a cooler box containing ice and analysed within 8 hours of collection (Rice et al. 2012).

The physicochemical parameters used in this study included total dissolved solids (TDS), temperature, electrical conductivity (EC), pH, and dissolved oxygen (DO) which were determined immediately in the laboratory using a pH probe (Hach Intellical™ PHC101) and DO probe (Hach LDO^® Model 2), respectively. The analysis of turbidity was also determined on site using a Lovibond TB 211 turbidimeter (Lovibond. IR, Germany) in the laboratory.

Three colonies of presumptive yellow/green-yellow obtained after culturing for 24 h was suspended in sterile saline water (0.85% NaCl, NutriSelect^® Plus, Sigma Aldrich, SA) for inoculation of an API 20E strip (Biomerieux Industries, South Africa) according to the instructions of the manufacturer and incubated at 37 °C for 24 h. Each isolate was interpreted using the API 20E analytical profile index (Version 5, Biomerieux Industries, France).

The DNA was extracted from the purified cultured colonies using the ZymoBIOMICS^TM DNA Miniprep Kit (Zymo Research, CA, USA) according to the manufacturer's instructions. The quantity of the extracted DNA was determined using the NanoDropTM 2000 spectrophotometer (Thermo Scientific, Johannesburg, South Africa). The PCR reactions were performed in a total volume of 25 μL, including 12.5 μL of mastermix (1.5 mM MgCl₂, 50 mM KCl, 10 mM Tris-HCl (pH 9.0), 0.1% Triton X-100, 200 μM dNTPs each (Fermentas), 0.5 μL of each V. cholerae-specific primer, 2.5 μL of template DNA, and the rest was RNA free pure water. Amplification reactions were carried out using a qPCR (Eppendorf Master Cycler 5330, Eppendorf-Nethel-Hinz GmbH, Hamburg, Germany) in the following order: heat denaturation at 94 °C for 2 min, followed by 35 cycles of heat denaturation at 94 °C for 1 min, primer annealing at 58 °C for 1 min, and DNA extension at 72 °C for 1 min (Momtaz et al. 2013). After the last cycle, the samples were kept at 72 °C for 2 min to complete the synthesis of all strands before cooling at 4 °C.

The oligonucleotide primers used in the study are shown in Table 2. The OmpW primer gene used was specific for coding a protein that plays an important role as osmoregulatory-sensitivity for harsh conditions such as high salt environments and provides channels for transportation of substances for V. cholera in the host (human epithelial cells) (Nandi et al. 2000). The screening of toxigenic V. cholerae was done using a specific primer (ctxA) gene (Mehrabadi et al. 2012), which also codes for the toxins in V. cholerae.

Table 3 lists the various activities that were observed taking place around the river sites during the collection of water samples. These activities include agricultural activities, washing laundry, car washing, littering, open defecation by community members, and domestic solid waste disposal besides human settlements.

The physicochemical analysis results are shown in Table 4. The pH counts of the rivers were between 5.0 and 8.5, which were still within the acceptable South African water quality standards/guidelines of 6.5–8.5 (Department of Water Affairs & Forestry 1996). The average temperatures of these rivers ranged between 16 and 26 °C, while their respective electrical conductivities (EC) ranged between 22.5 and 391 μS/cm. Most of the rivers (Mvudi, Livhuvhu, Dzindi, Nzhelele, Mutale, Shingwedzi, Tshinane, and Nwedi) reported EC counts that were above the acceptable South Africa water quality standards/guidelines of 0–70 μS/cm (Department of Water Affairs and Forestry 1996). The total dissolved solids (TDS) of all the rivers were also found to be within the acceptable South African water quality standards/guidelines of 0–450 mg/L (Department of Water Affairs and Forestry 1996). The turbidity counts of most of the rivers sampled in this study were within the acceptable set standards limit of between 10–20 NTU for river water containing sediments and untreated aquatic water (Department of Water Affairs & Forestry 1996). The values of the dissolved oxygen (DO) of most of the rivers were within the acceptable limit of between 6.5–8.0 ppm with the exception of the Mvudi, Shingwedzi, and the Nwedi rivers, which were above the acceptable limit, and the Madanzhe river, whose DO counts was below the acceptable water quality standards/guidelines.

Table 5 contains results of API where on the upstream sites of the Mvudi River, there were presumptive colonies for V. cholerae together with Proteus spp, and V. flavialis and Aeromonas spp were identified on the downstream site by the API test (Biomerieux Industries, South Africa). On the Livuvhu River, there was no Vibrio spp identified on its upstream site, but presumptive Vibrio spp, Serratia spp and Aeromonas spp were identified using the API test on the downstream site (Table 5). On the Dzindi River, Burkholderia cepacie was identified by the API test on the downstream site. There was no Vibrio spp identified on the upstream site of the Dzindi river. There was presumptive Vibrio spp identified on the downstream site, and presumptive Vibrio spp together with Providencia spp on the upstream site of Madanzhe River (Table 5). For the Nzhelele River, presumptive Vibrio spp was identified in both upstream and downstream sites respectively however, Serratia spp was also reported on the downstream of the river. No Vibrio spp was identified upstream of the Phiphidi River together with V. parahaemolyticus, Proteus vulgaris group and only Klebsiella spp was reported on the downstream site (Table 5). In the Mutale River, presumptive Vibrio spp was identified in the downstream site of the river and the API test further identified Serratia spp and Aeromonas hydrophila gr. 2 on the upstream site of the river (Table 5). There was no Vibrio spp identified in the upstream site of the Sambandou River, but presumptive Vibrio spp was identified at the downstream site of the river. In the Shingwedzi River, there was no Vibrio spp detected in the river, but other species such as Klebsiella spp and Serratia spp were identified in both the upstream and downstream sites of the river respectively. In the Nwedi River, presumptive Vibrio spp, Aeromonas spp and Serratia spp was identified on the downstream site, and Klebsiella oxytoca and Serratia spp respectively identified on the upstream site (Table 5). In the Mukhase River, there was no Vibrio spp identified however, Providencia stuartii and Serratia spp were reported from the downstream site with Aeromonas hydrophila detected on the upstream site of the river. Species such as Erwina spp and Enterobacter spp, Aeromonas hydrophila gr. 2, Serratia spp, Proteus vulgaris species were identified at the upstream and downstream sites, respectively, on the Tshinane River (Table 5).

The PCR assessment of all the presumptive colonies isolated from the river samples using the OmpW primer/probes, revealed that there was no Vibrio spp detected in the Mvudi, Dzindi and Shingwedzi Rivers (Table 6). There were no Vibrio spp detected on the upstream sites of the Livuvhu, Nzhelele, Sambandou, Mukhase, and Tshinane rivers, respectively. The Mutale and Madanzhe rivers were the only rivers that had Vibrio spp on their upstream sites (Table 6). In addition, Vibrio spp detected on both upstream and downstream sites of the Nwedi and Phiphidi rivers, respectively. From 12 rivers, V. cholerae was initially detected in nine rivers (upstream/downstream/both). The presence of pathogenicity in Vibrio spp was further analysed from these rivers, which initially established the presence of V. cholera using the ctxA primer. This primer codes for the toxigenic V. cholera and, four river samples, (upstream of Mutale, downstream of Livuvhu, Nwedi, and Mukhase, respectively) tested positive for toxigenic V. cholera as reported in Table 6.

This study set out to determine the presence of toxigenic V. cholerae in river water sources within the Vhembe municipality district in the province of Limpopo, South Africa. Rural areas and neighbouring countries bordering South Africa bear the brunt of perennial waterborne outbreaks associated with waterborne transmission diseases such as Vibrio cholerae due to human movement, inadequate sanitation infrastructures, and polluted water used for domestic, agricultural, and recreational purposes (Du Preez et al. 2003; Mukandavire et al. 2011; Ismail et al. 2013). The authors further reported that, a decade ago, there was a huge outbreak of cholera in Zimbabwe where more than 70,000 cases and approximately 5,000 deaths. In addition, Mintz & Guerrant (2009) further opined the extent of the outbreak of cholera cases spreading into neighbouring countries such as Zambia and South Africa. In South Africa alone, a total of 12,000 cases of cholera were reported with the majority of these cases from Mpumalanga and Limpopo provinces (Ismail et al. 2013).

The monitoring of physicochemical parameters such as temperature, turbidity, TDS, pH, DO, and EC of the water is essential because they can adversely influence the densities and survival/adaptation of V. cholerae (Huq et al. 2012). Furthermore, there is a significant correlation between rainfall, moisture, and conductivity and the overall temperature of water in relation to the outbreak of cholera at a specific time (National Centre for Infectious Diseases US 1994). Vibrio cholerae are a group of organisms that are phenotypically heterogenous, making them able to thrive in a wide range of temperatures (Huq et al. 2012). Most of the EC counts measured for the rivers in this study (Mvudi, Livuvhu, Dzindi, Nzhelele, Mutale, Shingwedzi, Tshinane, and Nwedi rivers) were above the acceptable South African water quality guidelines/standard limit (Department of Water Affairs and Forestry 1996) of 0–70 μS/cm (Table 2). The high values of EC are directly related to the presence of ionic compounds (inorganic) in these water sources, which include carbonate, bicarbonate, chloride, sulphate, nitrate, sodium, potassium, calcium, and magnesium, all of which can carry or transfer an electrical charge, unlike organic compounds that do not affect the EC (Holmes 1996). V. cholerae are also able to survive these harsh conditions because of their ability to produce chitinase, especially in salinity and estuarine environments (Huq et al. 2012).

The presence of sediments, especially in rivers that are originated for agricultural and brick-making activities because of water runoff plays an important ecosystem for the microorganisms such as V. cholera in aquatic environments in providing both biotic and abiotic surfaces for their development (Ntema et al. 2014). The turbidity of the water also influences the survival of these bacteria, especially in the presence of biofilms or high suspended solids (Steadmon et al. 2023). Fong et al. (2010) reported that V. cholerae can form biofilm as a survival strategy in both the environment and within the host. The attachment of V. cholera to the biotic and abiotic surfaces of the river could also drive their presence in these rivers (Ntema et al. 2014). The values of the DO of most of the rivers in this study were within the acceptable South African water quality guidelines/standard limit (Department of Water Affairs and Forestry 1996) of 6.5–8.0 ppm, except for the Mvudi, Shingwedzi and the Nwedi rivers (Table 2). This might be due to higher temperatures (Table 2) that drive faster depletion of the available oxygen or high pollution of the rivers (Holmes 1996).

The study further reported that there was a high count (CFU/mL) of Vibrio spp obtained from the downstream sites of some of the tested rivers (Mvudi, Livuvhu, Dzindi, and Madandze rivers), which ranged between 4.0 × 10³ and 7.0 × 10⁴ CFU/mL. This might be due to the activities that were observed around the rivers shown in Table 3. The higher CFU/mL counts obtained in this study were contrary to the results that were obtained by Akoachere et al. (2013) which reported CFU/mL counts of Vibrio spp ranging between 0 and 8.0 × 10³ CFU/mL through their study done in Douala, Cameroon.

Biochemical tests such as API are vital for the identification and differentiation of various families of Enterobacteriaceae, which form the bulk of Gram-negative bacteria. According to Le Roux (2006), API tests assist in distinguishing V. cholerae spp from other closely related species such as Aeromonas spp and Proteus spp using various tests that include arginine dihydrolase, lysine decarboxylase, ornithine decarboxylase, and the utilisation of amygdalin and arabinose. Farmer & Hickman-Brenner (1992) further reported that Vibrio spp and Aeromonas spp can easily be confused because they are phenotypically related. A typical example is Proteus spp which is a Gram-negative rod-shaped and facultative anaerobic bacterium that can ferment sucrose to produce yellow colonies as V. cholerae and V. alginolyticus and causes urinary infection together with Klebsiella spp, Providencia spp and Serratia spp (Yeung & Thorsen 2016). The presence of bacteria such as Klebsiella spp, Providencia spp, Serratia spp and Proteus spp which was detected with the API test could also be due to some of the activities such as swimming and bathing that were taking place in these rivers by community members who might be suffering from urinary tract infection and urinating directly into the river during these activities as further reported by Yeung & Thorsen (2016). Many of the activities seen and reported in Table 3 are seen as potential factors that are compromising the quality of various river waters through contamination (Murei et al. 2023; Rajgire 2013). Murei et al. (2023) reported further that heavy rains place river sources at risk of flooding, together with faecal matter entering the rivers from the surrounding areas.

The presence of Vibrio spp detected in these rivers was analysed using a specific OmpW (outer membrane protein) primer (Nandi et al. 2000). This gene codes for a protein that plays an important role as osmoregulatory-sensitivity for harsh conditions such as a high saline environment and provides channels for the transportation of substances (both hydrophobic and iron molecules across the cell) in most Vibrio spp (Fu et al. 2018). The PCR results revealed that no Vibrio spp were detected on Mvudi, Dzindi and Shingwedzi Rivers however, in rivers such as the Livuvhu, Nzhelele, Sambandou, Mukhase, and Tshinane rivers, Vibrio spp were detected in their upstream sites (Table 6).

CtxAB and tcpA (Mehrabadi et al. 2012) are considered important genes for the detection of toxigenic V. cholerae (Taylor et al. 1987; Fields et al. 1992), and for this study, ctxA primer was used to target for cholera toxin which is the main Vibrio cholerae virulence factor unlike tcpA primer that targets the toxin-coregulated pili (TCP) – a colonization factor that also acts as a receptor for bacteriophage CTXϕ (National Centre for Infectious Diseases 1994; Le Roux 2006). Toxigenic Vibrio spp affects the small intestines by activating the G protein that is responsible for stimulating the adenylate cyclase, which results in excessive loss of chloride and other vital electrolytes (Na⁺, K⁺, H⁺, and ⁠) through diarrhea and vomiting (Mondiale de la Santé & WHO 2017). Toxigenic V. cholerae are also responsible for causing cholera, a highly epidemic diarrhoeal disease that continues to devastate many developing countries where socio-economic conditions are poor, sanitary systems and public hygiene are rudimentary, and safe drinking water is not available (Du Preez et al. 2003). According to Fykse et al. (2007), toxigenic V. cholerae is responsible for the increase in intestinal secretion of electrolytes and water into the colon when it invades its host (human). Hoshino et al. (1998) also suggested that the presence of ctxA does not always lead to the production of active Vibrio toxin, as truncation, point mutation, and misreading in the ctxA gene sequence can result in the positive ctxA or non-toxin producing strain. In PCR analysis of toxigenic V. cholerae, the ctxA gene encoding the A subunit of cholera toxin is the preferred target (Fields et al. 1992; Keasler & Hall 1993; Singh et al. 2002). In this study, the prevalence of the toxigenic V. cholerae was assessed, and four out of the 11 river water samples tested positive for the toxigenic V. cholerae gene. Most were from the downstream of the following rivers: Livuvhu, Nwedi, and Mukhase (Table 6). The presence of this toxigenic V. cholera in four rivers might be exacerbated by the activities that were taking place along and inside the rivers, such as open defecation by the surrounding community members, making bricks, washing cars, swimming or bathing, and farm workers (Table 3). These pathogenic V. cholerae obtained normally cause infection in human hosts through the release of enterotoxins, as reported by Okoh (2018). Okoh (2018) further suggested that the majority of the samples were negative for pathogenic V. cholerae, and this implied that any diarrheal disease caused may be less severe. Hasan et al. (2013) and Li et al. (2014) also reported that sometimes ctxA genes produced by toxin cholera can also be present in V. cholerae non-O1/non-O139 isolates. In a study done in Karachi city, Pakistan by Rasheed et al. (2019), there were a high number of toxigenic V. cholerae found in the water around the city due to poor sanitation infrastructure. The normal V. cholerae is also capable of acquiring toxigenic genes and virulence factors, which are encoded through the transfer of genetic elements that are moving from the surrounding environment (Sakib et al. 2018). Chakraborty et al. (2000) also reported that toxigenic V. cholerae normally exhibits a synergistic effect on numerous genes including non-toxicogenic ones that are dispersed among environmental strains produced at a given time to develop a strong virulence gene pool in aquatic environments. According to Khouadja et al. (2014), V. cholerae toxigenic serogroups O1 and O139 are the serogroups that are associated with epidemic and pandemic cholera. However, there are also a few cases of non-O1 and non-O139 septicaemia or lesions that are associated with diarrhoea (Chowdhury et al. 2016). These non-serogroups are different from those of toxigenic species because they lack the ctx gene cassettes, however, they possess heat-stable enterotoxin (hstn) (Ramamurthy & Nair 2014). The National Centre for Infectious Diseases (1994) also reported that the production of cholera toxin (CT) is one of the essential virulence factors for toxigenic strains of the O1 serogroups. Therefore, the rivers that tested negative for toxigenic V. cholera are not potentially able to cause enterotoxigenic epidemics due to lack of the ctxA gene (Le Roux 2006).

To mitigate this challenge, an integrated surveillance of river waters and wastewater quality into a timely warning system for cholera outbreaks in this district and a rapid surveillance approach are needed to detect not only pathogenic V. cholera but also provide an extent of cholera-associated risk factors in a rapid, reliable, and sensitive manner. This will further prove to be a vital tool for timely monitoring of river catchments or sources of pollution (point and nonpoint).

It is possible that there are reservoirs discharging illegal raw sewage or poorly treated final effluents from dysfunctional wastewater treatment plants that could be causing the deterioration of river water quality and were not included in this study. This will assist in providing comprehensive details about the sources of V. cholerae that are polluting and flowing through river water bodies. Additionally, this study was only conducted for a period of three months, and more thorough long-term and seasonal studies are needed to monitor the presence of V cholerae spp and the potential health risk to vulnerable communities.

The study reported that the electric conductivity counts measured for the following rivers – Mvudi Livuvhu, Dzindi, Nzhelele, Mutale, Shingwedzi, Tshinane, and Nwedi rivers) were above the acceptable South African water quality guidelines/standard limits. In addition, the DO of most of the rivers in this study were within the acceptable South African water quality guidelines/standard limits. The study further reported that there was a high count (CFU/mL) of Vibrio spp obtained from the downstream sites of some of the tested rivers (Mvudi, Livuvhu, Dzindi, and Madandze rivers), which were between 4.0 × 10³ and 7.0 × 10⁴ CFU/ml. In terms of the assessment of the prevalence of toxigenic V. cholerae a total of four out of 12 rivers tested positive, and most were from the downstream of the following rivers: Livuvhu, Nwedi, and Mukhase. The findings showed that the water quality in most of these river catchments, which are utilised by rural communities for agriculture, recreation, and drinking, has potential health risks to consumers if not treated before use. To mitigate the threat, a solid plan for regularly evaluating the quality of the water and the sources of pollution must be developed. The results indicated the importance of implementing river basin management measures to guarantee their long-term sustainability. This also serves as a strong motivation for the importance of ongoing monitoring of these freshwater bodies in the province.

The authors would like to thank Mr Damien Jacobs, Mr Ceryl and Mr Maphaisa for dedicating their time during sampling period and always available to assist throughout the study.

L.K., A.N.T and N.P. were involved in conception, planning, writing an original draft, review, and editing. M.P., A.N.T and L.K were involved in sample collection and data analysis and L.K and N.P were involved in supervision, planning, critical review, and editing of the article.

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

The authors declare there is no conflict.