Diversity of fungi in bottled water in Jeddah, Saudi Arabia

Fungi are ubiquitous in nature and are able to survive and grow in water sources, including drinking water. Fungi were observed to survive through the drinking water disinfection process in the 1980s (Niemi et al. 1982). Recently, potentially pathogenic species have frequently been isolated from drinking water systems (Paterson & Lima 2005; Paterson et al. 2009; Hageskal et al. 2011; Oliveira et al. 2013; Babič et al. 2016; Hurtado-McCormick et al. 2016). More than half (66%) of the fungal species identified in different drinking water sources in Brazil were considered potential pathogens (Oliveira et al. 2013). An emerging pathogen, Aspergillus calidoustus, has been frequently isolated from Norwegian water systems (Hageskal et al. 2011). Several fungal species found in drinking waters are known to cause infectious diseases (Paterson & Lima 2005; Paterson et al. 2009) but no report about any acute disease caused by fungal contamination in purified drinking water was found in a recent review (Hageskal et al. 2009). However, health effects are not fully understood and several articles have regarded fungal contamination as a possibly underestimated problem in drinking water distribution systems (Hageskal et al. 2006; 2009; 2012; Kanzler et al. 2008; Pereira et al. 2009; 2010; Siqueira et al. 2011; Al-gabr et al. 2014; Skaar & Hageskal 2015). Also a recent review of Babič et al. (2017) concludes that harmful health effects of pathogenic fungi are possible especially for immunocompromised people. In addition to health effects, fungal contamination may also be responsible for the mycotoxins that possibly cause organoleptic defects and allergenic reactions (Mata et al. 2015; Skaar & Hageskal 2015; Bai et al. 2017).

The use of bottled water, as a safe substitute for tap water, has increased in the past few decades, but the possible microbial contamination of bottled water has been studied very little. It has been reported that bottled water may be contaminated with bacteria and fungi (Cabral & Pinto 2002; Criado et al. 2005; Yamaguchi et al. 2007). However, only a few studies have reported the contamination of bottled water at the species level, although diseases, mycotoxins, pigment and odor formation have been associated more with the individual species than with the genus (Siqueira et al. 2011; Oliveira et al. 2013). Cladosporium cladosporioides, Penicillium sp. and Alternaria alternata were found in bottled water in Buenos Aires, Argentina (Cabral & Pinto 2002; Criado et al. 2005). In Brazil, 20% of the bottled water samples were contaminated by fungi; three species of the genus Candida were found (Yamaguchi et al. 2007). The review of Babič et al. (2017) reports ten fungal species identified from bottled water during 30 years. These species were Aspergillus fumigatus, A. versicolor, Aureobasidium pullulans, Debaryomyces hansenii, Exophiala spinifera, Penicillium chrysogenum, P. glabrum, Talaromyces rugulosus, Trichoderma longibrachiatum, and Filobasidium magnum. In addition they report three genera, namely Cladosporium, Fusarium and Paecilomyces.

We aim to fill the knowledge gap in fungal contamination of bottled waters and analyze the occurrence and diversity of fungi as contaminants in waters. Purity of water is especially important in places where people drink mainly bottled water, for instance, in Saudi Arabia where the land is poor in natural water resources. We collected forty bottled waters, of international trade marks, and analyzed them using both classical and molecular techniques. The results give information on the occurrence of potentially harmful pathogenic fungi in bottled drinking water.

Forty unopened water bottles were randomly collected from different markets in Jeddah, Saudi Arabia in 2012–2013. The origin and the water processing information are provided in Table 1. Nine of the bottles were described as being ozone treated (ozone) while 31 bottles had no mention of the treatment (no-ozone). The details of the ozone treatments are not known. The t-test was used to study the difference between the ozone and no-ozone treated bottles.

The bottles were stored unopened at room temperature (28 ± 2°C) until studied (0 d). All work after opening the bottles was performed with aseptic techniques under sterile conditions and all possible contamination outside the bottles was avoided. The bottles were sterilized outside with ethanol before entering the sterile environment and the lips were sterilized with ethanol after opening the bottle. No growth was observed in the blanks inoculated with sterile water. Five of the bottles were randomly chosen for further analysis after they had been stored for 180 d and 365 d at room temperature in order to find out the potential of fungi to reproduce in bottled water. Three replicate analyses were performed, the mean colony forming units (CFU) was calculated and the species results were combined to represent the trademark.

The membrane filtering technique was used (Pereira et al. 2010). An aliquot of 100 ml of water was filtered through a 0.45 μm membrane. The membrane was placed on the surface of sterilized petri dishes containing autoclaved potato dextrose agar medium (PDA). The plates were incubated at 28 ± 2°C for one week. The number of colonies was counted and the grown fungal mycelia were collected for identification. The isolated fungi were maintained on PDA slants at 4°C. All the media used in this study were obtained from HiMedia, Mumbai, India.

For the classic morphological identification, the fungi were sub-cultured in suitable agar media (Dichloran Rose Bengal Chloramphenicol agar (DRBC), Czapek Yeast extract agar (CYA), Czapek Dox agar (CZ), Synthetic Nutrient Medium (SNM)) according to Samson & Frisvad (2004). Slide preparations were stained with lactofuchsin, with or without alcohol, lactic acid or in double distilled water. The fungi were phenotypically identified to the genus level, or to species level when possible, under a light microscope (Barnett & Hunter 1972; De Hoog et al. 2000; Klich 2002). The identifications were checked for consistency with the latest diagnoses. In total, 35 fungal isolates were identified either morphologically or using molecular techniques (13 isolates).

For the molecular identification, an aliquot of 2 ml of potato dextrose broth (PDB) was poured into the PDA slants containing well grown fungi and shaken thoroughly. The PDB containing spores of each fungi were poured individually into flasks containing 100 ml of sterilized PDB. The flasks were incubated at room temperature without shaking for a week. The grown mycelium in the broth culture was collected aseptically by filtration and ground in liquid nitrogen in a sterile mortar to obtain a mycelium powder. Further, the DNA was extracted from 20 mg of mycelium powder using a cycle-sequencing kit (Applied Biosystems, Darmstadt, Germany).

The internal transcribed spacer (ITS) region of the ribosomal DNA (rDNA) was amplified by polymerase chain reaction (PCR) with the primers ITS1-F(CTTGGTCATTTAGAGGAAGTAA) and ITS4 (TCCTCCGCTTATTGATATGC) (White et al. 1990; Gardes & Bruns 1993). PCR amplifications were performed in a final volume of 50 μl by mixing 2 μl of DNA with 0.5 μM of each primer, 150 μM of dNTP, 6 U of Taq DNA polymerase and PCR reaction buffer. Amplification was conducted in a thermal cycler with an initial denaturation of 3 min at 94°C, followed by 35 cycles of 1 min at 94°C, 1 min at 50°C, 1 min at 72°C, and a final extension of 10 min at 72°C. Aliquots of PCR products were checked by electrophoresis on agarose gel (1%) revealed with ethidium bromide and visualized by UV trans-illumination. The PCR products were purified by ExoSAP-IT (USB Corporation, under license from GE Healthcare) based on the manufacturer's instructions. The purified products were sequenced using an automated DNA sequencer (ABI PRISM 3700) using the BigDye Deoxy Terminator cycle-sequencing kit (Applied Biosystems) following the manufacturer's instructions. Sequences were submitted to GenBank on the NCBI website (http://www.ncbi.nlm.nih.gov). They will be deposited in the World Data Centre for Microorganisms (http://new.wfcc.info/ccinfo/index.php/collection/by_id/907).

Sequences obtained in this study were compared with the GenBank database using the BLAST software on the NCBI website (http://www.ncbi.nlm.nih.gov/BLAST/). DNA sequences were first aligned with Clustal X₂ for Windows (version 1.3b), which was used to construct a neighbor-joining tree using the Jukes-Cantor model.

Fungal contamination was found in 58% of the bottled water samples; 23 out of 40 bottles were contaminated with fungi. The contamination frequency was high compared to previous studies where 20–33% of the bottles had been contaminated in Brazil (Yamaguchi et al. 2007) and Argentina (Cabral & Pinto 2002). The reason for the higher contamination frequency in Saudi Arabia than in Brazil or Argentina is not evident. In the first place, all samples represented different international trademarks. Thus, many more trademarks were studied in Saudi Arabia than in Brazil or Argentina. Although the membrane filtration technique was used in all studies, the different media used for fungal enumeration may explain the results.

The bottles described as being treated with ozone (ozone, n = 9) did not differ significantly from the non-ozonated bottles (no-ozone, n = 31). Up to three species were identified in both treatment groups (Table 2). The mean CFU was lower, although not significantly (t-test), in the ozone bottles (0.5 ± 5.7 CFU in 100 ml, mean ± SD) than in the no-ozone bottles (2.6 ± 5.7). The variation was high and the counts started from 0 CFU in both groups. The maximum CFU counted was lower (3.5) in the ozone group than in the non-ozone group (20). The ozone bottles were less frequently (44%) contaminated than the no-ozone bottles (61%). It seems possible that the ozone treatments used had reduced fungal contamination. An ozone treatment has been observed to be the most effective treatment against fungi in general (Hageskal et al. 2012). However, the efficiency of the treatment has depended on the dose of ozone used (Hageskal et al. 2012). We had no information on the ozone dose used in our bottles and thus, we cannot verify its effect reliably. The susceptibility of species to ozone treatment varied a lot in the study of Hageskal et al. (2012), who also found different species to those we did. In summary, there was an indication that the ozone treatment may reduce fungal contamination in bottled waters. However, this conclusion is highly speculative because more detailed information on the treatments used as well as a more balanced study design would have been needed in order to confirm this.

The total CFUs observed were relatively low. Up to 20 CFU were detected in 100 ml water (Table 3). The level of CFU was about the same as in bottled waters in the study of Cabral & Pinto (2002). Much higher levels, up to 3,000 CFU, have been observed in drinking water systems (Hurtado-McCormick et al. 2016; Oliveira et al. 2016). However, comparison between the studies is difficult because of the slight differences in the methods and the sensitivity of microbial growth to growing conditions. A more reliable comparison is presented in Yamaguchi et al. (2007), where the same conditions were used for tap water and bottled water. The authors conclude that the tap water samples had a clearly lower fungal count and contamination frequency because bottled waters are mostly unique natural products that cannot be treated, nor can any exogenous elements be added to them. As a summary, the variation in fungal contamination can be assessed to be high throughout the world.

Our focus was on the identification and diversity of species found in bottled drinking waters. Different drinking water resources and drinking water systems have been reported to be contaminated with a high variety of fungal genera and species, reviewed by Hageskal et al. (2009). The species composition seems to be determined by the concentrations of inorganic ions, such as calcium, magnesium and nitrate in water (Babič et al. 2016).

The genera isolated most frequently are Penicillium and Aspergillus both in drinking water systems and in bottled waters (Hageskal et al. 2009; Oliveira et al. 2013; 2016; Babič et al. 2016; Fish et al. 2016). In our study, the 35 fungal isolates found belonged to 11 fungal genera (Table 3). The 14 identified species were Aspergillus niger, A. flavus, A. terreus, A. fumigatus, A. caespitosus, A. tubingensis, A. chevalieri, Cladophialophora bantiana, C. sphaerospermum, Exophiala cancerae, Gliomastix murorum, Penicillium crustosum, Rhizopus nigricans and Sarocladiumim plicatum. In addition, Mycelium sterilium, a fungal strain that cannot be identified, and three unidentified species from the genera Geotrichum, Periconia and Phialocephala were observed. A recent review reports ten fungal species and three genera identified in bottled waters during 30 years (Babič et al. 2017). Compared to that result, we identified a large variety of species in our sampling. This may be first of all explained with the techniques developed to identify the species.

Rhizopus nigricans was the most frequently found species occurring in 14 samples, which is 61% of the contaminated samples. The total counts of R. nigricans, however, were relatively low, a maximum of 0.5 CFU in 100 ml. The highest total counts, over 10 CFU in 100 ml, were observed for G. murorum, C. bantiana, E. cancerae and Phialocephala sp. They, however, occurred only in one or two samples each. The genus Aspergillus occurred in six samples. Aspergillus niger occurred in two samples and six different Aspergillus species occurred each in one sample.

Several species found in drinking waters have been reported as emerging pathogens. Hageskal et al. (2012), reported Aspergillus calidoustus, Penicillium spinulosum, Trichoderma viride and Fusarium solani as potential pathogens and common drinking water system contaminants in Norway. According to their experiment on possible pathogenicity, Oliveira et al. (2013) classified several Penicillium and Trichoderma species as potential pathogens, but we did not identify exactly the same species in our samples. Aspergillus niger, which we identified, was classified as non-pathogenic by Oliveira et al. (2013). We identified an Aspergillus species in 26% of the contaminated samples. The genus is a common drinking water contaminant; it has been reported in several studies (Anaissie et al. 2003; Hageskal et al. 2006; 2007; 2009; Kennedy & Williams 2007; Kanzler et al. 2008; Pires-Gonçalves et al. 2008; Gashgari et al. 2013; Oliveira et al. 2016; Ma 2017).

In addition, many of these studies have reported the genus Penicillium to occur in drinking water samples. We observed it only in one sample. As a summary, we observed several species that, however, seemed to occur at low frequencies and are probably mostly non-pathogenic. Moreover, most of these potentially pathogenic fungi are only pathogenic after inhalation as opposed to ingestion (De Hoog et al. 2000; Zhou et al. 2007). Therefore, it seems that the fungal contamination in bottled water is not a great health risk to humans. However, a recent finding that the resistance to disinfection of Penicillium and Aspergillus species could facilitate their survival in drinking water systems (Ma 2017), raises a need for further studies about fungal contamination also in bottled waters.

The ITS rDNA sequences of the isolated fungal strains (submitted to the European Molecular Biology Laboratory (EMBL) were compared with published sequences at GenBank (Table 4). The GenBank accession numbers and the closest relatives of the isolates are listed in Table 4. The phylogenetic tree was established (Figure 1). The detected fungal strains were classified as members of the subphylum Pezizomycotina and Saccharomycotina (phylum Ascomycota). All detected fungal strains were placed in six orders, Eurotiales (A. tubingensis, A. chevalieri, A. caespitous, A. terreus, A. flavus and P. crustosum), Chaetothyriales (Exophiala cancerae and Cladophialophora bantiana), Helotiales (Phialocephala sp.), Hypocreales (Sarocladium implicatum and Gliomastix murorum), Pleosporales (Periconia sp.), and Saccharomycetales (Geotrichum sp.) (Figure 1). Most fungal contaminants belonged to the Ascomycetes. This is in accordance with the previous finding of Cabral & Pinto (2002), who associated the contamination of eight different commercial brands of bottled water in Argentina mainly with Ascomycetes. More recently, Gashgari et al. (2013) reported that most mycobiota in four different drinking water distribution points in Jeddah City (Saudi Arabia) belonged to the Ascomycetes.

Five samples were selected to study the effect of storage on the growth of fungi. Two of the samples remained negative for fungal growth during the storage. In one sample, all three species identified at the beginning (0 days) survived for 180 d (Table 5). However, the fungi were not able to reproduce effectively, most likely due to the lack of nutrients. The CFU of the two species, A. tubingensis and A. chevalieri, and the unidentifiable species M. sterilium, decreased from 3.5 at the beginning (day 0) to 0.66 in 100 ml water after 180 d storage. After 365 d storage, the total count of M. sterilium was 0.66 in 100 ml water. Unidentified species (M. sterilium) were found in three out of the five 365 d stored bottled waters. Our results are only tentative because of the low number of replicates, but they support the interpretation of Morais & Da Costa (1990) and Ferreira et al. (1994) that microbial growth and species might change during storage due to the presence of oxygen and increasing surface area (mass of microbes) inside the packaging. However, fungi seem not to be able to grow in bottled water to any great extent. Thus, we suggest that even long storage bottled water does not increase the health risks for humans.

This is one of few studies about fungal contamination in bottled waters. The samples were of international trademarks and bought in Saudi Arabian markets. The diversity of fungi (18 species belonging to 11 fungal genera) occurring in bottled water seems to be relatively high. The species that frequently contaminated bottles were Rhizopus nigricans and seven different species of Aspergillus. Penicillium sp. were found in one sample. Although some species are known as pathogens, most of the species seemed to be non-pathogenic and thus, we conclude that the fungal contamination in bottled water seems to be a low risk to human health. Harmful health effects seem to be possible, mostly for immunocompromised people. However, the link to health effects is still not fully understood. Because we observed fungal contamination in more than half of the bottled waters studied, the fungi should be taken into account in the bottled water purification processes and in the quality control assessment. The ozone treatment may reduce fungal contamination in bottled waters. Future studies should focus on the mycotoxins the fungi are producing in water.

FA, AA and RG designed the experiment. FA performed the laboratory analyses and conducted data analyses. FA drafted the manuscript. All authors revised and approved the final manuscript and agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

The authors extend their thanks to the Deanship of Scientific Research at King Saud University for funding this work through research group NO (RGP-1438-029).