Relationships between the abundance of potential geosmin- and 2-methylisoborneol-producing (2-MIB) cyanobacteria and bacteria, and concentrations of the two taste and odour compounds (T&Os) were examined in a 7 day incubation of natural water from the surface and bottom of three reservoirs in southeast Queensland, Australia. Only a single known T&O-producing cyanobacterium (Geitlerinema spp.) was detected by microscopy at low density, and only in one reservoir. Densities of potential T&O-producing Streptomyces (determined by quantitative polymerase chain reaction (qPCR) assay) were highest in the bottom water and varied from 0.7 × 103 to 775 × 103 cells L−1. Geosmin ranged from 6 to 59 ng L−1 (with the highest concentrations in the bottom water), while 2-MIB varied from 6 to 47 ng L−1 (with the highest concentration in surface water). Concentrations of both compounds declined during the incubation under both light and dark conditions. Presence of the geosmin synthase gene, geoA, in cyanobacteria and Streptomyces was examined by different PCR approaches. Cloning of PCR products from amplification of geoA showed a high similarity to geoA in cyanobacteria, but not to streptomycetes. Our results demonstrate that more research on the ecology and molecular biology of T&O producers is required to better understand the dynamics of T&Os and to monitor emerging T&O episodes.
Freshwater reservoirs provide drinking water in many urban regions, but taste and odour compounds (T&Os), produced by microorganisms in the reservoirs, may spoil the quality of the drinking water. Typical T&Os in the reservoirs are geosmin (which gives an earthy flavour) and 2-methylisoborneol (2-MIB; which gives a mildewed flavour) (Juttner & Watson 2007). Although non-toxic, tainting by geosmin and 2-MIB makes the water non-palatable to many consumers and introduces fear of bacterial contamination of the water (Antonopoulou et al. 2014).
T&Os are produced by certain species of cyanobacteria and various non-phototrophic bacteria, especially Streptomyces species, but recent research suggests that bacteria belonging to the Myxococcales order may also produce T&Os in freshwater (Auffret et al. 2013). Concentrations and seasonal fluctuations of T&Os, as well as potential T&O-producing organisms, have been studied in several reservoirs and natural waters, yet there is limited knowledge on processes controlling production of the T&Os (Srinivasan & Sorial 2011). For example, the influence of specific environmental conditions, such as light, inorganic and organic matter and temperature, on the abundance of potential T&O-producing organisms has only been studied in a few cases (Jüttner & Watson 2007; Matsushita et al. 2015).
Since removal or reduction of T&Os from drinking water requires expensive and time-consuming procedures, e.g., treatment by particulate activated charcoal or advanced oxidation procedures (Bruce et al. 2002; Antonopoulou et al. 2014), there is demand for alternative approaches to ensure lower levels of T&Os in drinking water. One such approach is the ability to predict upcoming T&O episodes by detection of the densities of geosmin- and/or 2-MIB-producing organisms in the water. Recently, molecular methods for detection of essential genes in the synthesis of geosmin and 2-MIB, i.e., geosmin synthase (geoA) or MIB synthase (MIBS) genes, have been published (Ludwig et al. 2007; Kakimoto et al. 2014; Kutovaya & Watson 2014). For monitoring of emerging T&O problems in drinking water reservoirs, e.g., caused by cyanobacteria, the density of geosmin-producing cyanobacteria can be estimated from the number of geoA genes (Tsao et al. 2014) at regular intervals during periods of expected high geosmin concentrations. This may allow water treatment operators to change to alternative sources of water.
In this study, we wished to test a practical application, e.g., by public water quality laboratories, of recent polymerase chain reaction (PCR) methods for detection and quantification of geosmin-producing phytoplankton and bacteria by amplification of the geoA gene. For detection of geoA in Streptomyces and cyanobacteria in water reservoirs in southeast Queensland, Australia, the general primers by Ludwig et al. (2007) were applied, while two geoA-specific primer sets for quantitative PCR, one set targeting geoA in selected Streptomyces (Auffret et al. 2011), the other targeting geoA in certain species of cyanobacteria and actinomycetes, e.g., Streptomyces (Kutovaya & Watson 2014), were applied. Water from the surface and bottom of three reservoirs with known T&O problems was incubated for a week at in situ temperatures to monitor changes in geosmin and 2-MIB relative to the abundance of potential T&O producers.
MATERIALS AND METHODS
Subsamples were collected from the plastic bags at the commencement of incubation and for 5 more days over a 7 day period for geosmin and 2-MIB analyses, enumeration of algae, cyanobacteria and bacteria, and PCR analysis for Streptomyces spp. and abundance of the geoA gene in microorganisms in the water. Glass vials were filled to the brim with water samples for geosmin and 2-MIB analyses, and chilled until analysed. Samples for counting of algae and cyanobacteria were fixed in Lugol's solution and counted using Sedgewick Rafter cells under a compound microscope. Densities of algae and cyanobacteria were converted to biovolumes according to Hillebrand et al. (1999) using biovolumes previously determined for species in reservoirs in the region (Ann Chuang, unpublished data). Densities of bacteria were determined by epifluorescence microscopy at 100 × magnification after staining with SYBR Gold according to Shibata et al. (2006). Cell numbers in 10 randomly selected areas were counted on each filter.
For molecular studies, water samples of 500 mL were filtered onto 0.2 μm 47 mm diameter membrane filters and frozen until analysed. Each filter was cut into three equal pieces; the pieces were treated by three-fold 20 min freeze–thaw cycles (−70 °C–65 °C) to improve cell wall destruction and improve extraction of DNA. DNA was extracted using the PowerWater kit by Mobio Laboratories Inc. (USA). Unfortunately, the DNA samples from February 6th and 13th were destroyed during extraction. Instead, DNA was extracted from samples taken on February 7th and 11th. Densities of streptomycete bacteria were determined by quantitative PCR (qPCR), combining TaqMan probe and primer technology (www.appliedbiosystems.com), according to Rintala & Nevalainen (2006) and Lylloff et al. (2012). Filters from the start and end of the incubation period were selected for analysis.
qPCR assays for quantification of the geoA gene were adapted from Auffret et al. (2011) and Kutovaya & Watson (2014). The primer sets of AMgeo-F (GAGTACATCGAGATGCGCCGCAA) AMgeo-R (GAGAAGAGGTCGTTGCGCAGGTG) and geo_cya543F (ATCGAATACATYGARATGCG) geo_cya728R (ACTTCTCTYTGRTAGGA) were applied to target geoA of Streptomyces and cyanobacteria, respectively. The qPCR reactions were performed with a 20 μL reaction mixture containing the final concentration of 1 × RealQ PCR master mix with green dye (Ampliqon, Denmark), 200 nM of each primer (AMgeo-F&R for Streptomyces and geo_cya543F&728R for cyanobacteria), 0.2 mg BSA mL−1 and 2 μL DNA template. The cycling conditions were as follows: heat activation at 95 °C for 15 min, 40 cycles of denaturation at 95 °C for 30 sec, annealing at 61 °C (Streptomyces) and 55 °C (cyanobacteria) for 30 sec and extension at 72 °C for 60 sec, followed by melting curve analysis of the amplified product. Template DNA of Streptomyces coelicolor A3(2) and Oscillatoria spp. 327/2 (the strain mentioned in Suurnakki et al. (2015)) were used as positive controls for the PCR amplification. DNA extracted from conidia in two Streptomyces species was used for construction of the calibration curve (gene copy number in Streptomyces vs. Ct value) as in Lylloff et al. (2012).
For detection of the geoA gene in the water samples, the primer set CycFW (TGGTAYGTITGGGTITTYTTYTTYGAYGAYCAYTT) and CycRW (CATRTGCCAYTCRTGICCICCISWYTGCCARTCYTG), targeting both cyanobacteria and Streptomyces and designed by Ludwig et al. (2007), was applied. In brief, the PCR protocol was: 50 μL reaction mixture with 1.5 mM MgCl2, 400 μM dNTPs, 600 nM of each primer, 20 μg BSA, 1 μL of 100% DMSO, 2.5 U of Taq DNA polymerase (vwr.com) and 2 μL of DNA template. The PCR cycling conditions included heat activation at 95 °C for 2 min, followed by 35 cycles of denaturation at 95 °C for 30 sec, annealing at 55 °C for 45 sec and extension at 72 °C for 60 sec, and finished with extension for 5 min at 72 °C.
The amplified PCR products were purified using a Qiagen gel extraction kit (qiagen.com), cloned into pGEM®-T Easy Vector (promega.com) and transformed into Escherichia coli (JM109 High Efficiency Competent Cells; Promega) in accordance with the manufacturers’ guidelines. Positive clones were sequenced by GATC-Biotech (Germany). The sequences were submitted to NCBI and given accession numbers KX250357 to KX250366.
Concentrations of geosmin and 2-MIB were determined using a Teledyne Tekmar Stratum Purge and Trap Concentrator (Teledyne Tekmar, USA) connected to an Agilent 7890B GC System with an Agilent 5977A MS detector (Agilent Technologies, Australia) with a detection limit of 1 ng/L for both compounds. Nutrient concentrations were measured following standard colorimetric methods (APHA 2005). Total nutrients were initially digested in persulfate prior to the colorimetric analysis (Hosomi & Sudo 1986). The detection limit for dissolved inorganic nutrients was 0.002 mg L−1 and for total nutrients was 0.005 mg L−1.
Geosmin and 2-MIB
Phytoplankton and bacteria
In Wyaralong reservoir, the cyanobacteria were dominated by Planktolyngbya minor and Gloeothece spp. (making up 55 and 20% of all cells), while Planktolyngbya limnetica and Gloeothece spp. co-dominated in Little Nerang reservoir (Table S2, available with the online version of this paper). In the LHD reservoir, Woronichinia spp. made up about half of the cells. Among known T&O-producing organisms, only Geitlerinema spp. was detected in Wyaralong surface water (1.2 × 103 cells L−1), while the common T&O producers, such as Dolichospermum (formerly Anabaena) and Oscillatoria, were not present in measurable densities.
PCR amplification of the geoA gene
Initial tests of the qPCR assay for the geoA gene in DNA extracted from cultures of Streptomyces and Oscillatoria demonstrated that the approach produced the expected amplification curves and Ct values, indicating that the PCR assay amplified geoA in the two groups of organisms. However, the qPCR assay did not form a successful product for geoA for DNA extracted from surface and bottom water samples of the three reservoirs. Incorrect melting point curves were found, implying a non-target amplification in the reaction.
In order to examine for the presence of the geoA gene in organisms in the reservoirs, a non-QPCR approach (using primer set CycFW and CycRW) was applied to DNA extracted from the bottom water of the Little Nerang reservoir, which had the highest geosmin content. As controls, DNA from cultures of Streptomyces and Oscillatoria were used. The PCR reaction generated positive products with the expected size from all samples. Taxonomic affiliation of the amplicons was examined by cloning into a vector, sequencing of 10 positive clones (sequences given in Table S3, available with the online version of this paper), and creation of a neighbour-joining phylogenetic tree of the cloned sequences. All sequences were identical to geoA in cyanobacterial strains and formed three different clusters. NCBI BLAST suggests that the sequences from one cluster were 88–89% identical to Phormidium sp. P2r sesquiterpene synthase (GenBank Identity (GI) 156707951), 85–86% identical to Oscillatoria sp. 327/2 geosmin synthase (GI: 730042929) and 85–86% identical to Aphanizomenon gracile WH-1 putative geosmin synthase (GI: 783305622). The sequences from the second cluster were 90–93% identical to Phormidium sp. P2r sesquiterpene synthase (GI: 156707951). The sequence in the third cluster was 71% identical to Nostoc sp. UK1 geosmin synthase (730042927). Sesquiterpene synthases include a large number of enzymes that form sesquiterpenes from farnesyl diphosphate. The amino acid sequence of geosmin synthase resembles that in some sesquiterpene synthases (Jiang et al. 2007), and the sesquiterpene synthase found in Phormidium sp. and other cyanobacteria might actually be geosmin synthase.
Analysis of results from the reservoirs showed three key results: (1) major difference in which T&O compound was most abundant in the reservoirs (prevalence of geosmin in Little Nerang and 2-MIB in Wyaralong) with the absence of the 2-MIB producing cyanobacteria; (2) unsuccessful quantification and detection of geosmin-producers by application of recent molecular approaches; (3) reduction of both geosmin and 2-MIB over the incubation period although both compounds are assumed to be relatively recalcitrant.
T&O concentrations vs. densities of cyanobacteria and Streptomyces
The difference in the dominance of either geosmin or 2-MIB as the major T&O compound between Little Nerang and Wyaralong suggests that different organisms produced T&Os in the two reservoirs. Cyanobacteria can produce both geosmin and 2-MIB, or just one of the compounds, while all non-photosynthetic microorganisms that produce 2-MIB are also reported to produce geosmin (Jüttner & Watson 2007). In the light-exposed surface water of Wyaralong (but not in the 35 m deep bottom water of the reservoir) the high 2-MIB concentration of 47 ng L−1 might originate from cyanobacteria. However, only a single potential T&O-producing cyanobacterium, Geitlerinema spp., was identified in the surface water, and at low abundance. Further, in the Geitlerinema genus, only one species, G. splendidum, is known to produce T&O, and only geosmin, not 2-MIB (Jüttner & Watson 2007). Thus, cyanobacteria most likely did not produce the measured 2-MIB. Similarly, in the Little Nerang reservoir, where no known geosmin-producing cyanobacteria were identified in the water, the relatively high geosmin concentrations in the surface (19.8 ng L−1) and bottom water (58.6 ng L−1) most likely did not originate from cyanobacteria.
Other potential T&O producers in the water were streptomycete bacteria. The abundance of Streptomyces in the reservoirs (0.7 × 103 to 775 × 103 cells L−1) fall within the range previously found for freshwater in southeast Queensland (Lylloff et al. 2012); but were streptomycete bacteria sufficiently abundant to produce the measured concentrations of geosmin and 2-MIB? Production of geosmin and 2-MIB by Streptomyces spp. in the water can be estimated, assuming that production rates from laboratory experiments with selected species of aquatic Streptomyces by Klausen et al. (2005) reflect conditions in the present reservoirs. Applying the highest cell-specific production rates by Klausen et al. (2005) (the observed range was 0.1–35 and 0.2–15 ag bacterium−1 h−1 for geosmin and 2-MIB, respectively) to the Wyaralong bottom water (water with the maximum density of Streptomyces), the time needed to produce the observed concentrations was 23 days for geosmin (15 ng L−1) and 54 days for 2-MIB (50 ng L−1). Thus, species of Streptomyces might have produced the measured geosmin and 2-MIB in the bottom water in Wyaralong. At other locations in the three reservoirs, the density of Streptomyces was significantly lower, and these bacteria were probably not the dominant producers of geosmin and 2-MIB. Also, only about 70% of known streptomycete species produce T&Os (geosmin and/or 2-MIB) (Jüttner & Watson 2007). This indicates that other T&O-producing organisms were present in the reservoirs, speculatively bacteria belonging to order Myxococcales as suggested by Auffret et al. (2013), but more studies are needed to confirm this.
In the dark incubation of water from the Wyaralong and Little Nerang reservoirs, a significant growth of especially the cyanobacteria Planktolyngbya limnetica and Gloeothece spp. occurred. Mixotrophic growth may explain the cell growth, at least for Gloeothece, since species within this genus are reported to use organic compounds for biosynthesis (Bandyopadhyay et al. 2010). The relatively high concentrations of total P and N and ammonium in the bottom water might, together with the DOC, have promoted the growth.
Detection of the geoA gene
The unsuccessful qPCR amplification of the geosmin synthase gene in cyanobacteria and Streptomyces with the applied quantitative assays could be caused by too few target cells in the water, or non-matching primers. The present qPCR assay for geoA was adopted from Kutovaya & Watson (2014), who designed functional PCR primers targeting geoA in isolated strains of cyanobacteria and actinomycetes (i.e., filamentous Actinobacteria such as Streptomyces), using separate primer sets. Since only one T&O-producing cyanobacterium (within the Geitlerinema genus) was identified by microscopy in the reservoirs, a low number of cyanobacterial geoA genes was expected. However, the lack of detection of geoA in Streptomyces by the qPCR assay was surprising, since these bacteria were abundant in the reservoirs. Kutovaya & Watson (2014) designed their primer sequences for actinomycetes on information from Auffret et al. (2011), who successfully detected geoA in isolated Streptomyces species and in environmental samples (biofilters from aquaculture systems). Although the qPCR assay by Kutovaya & Watson efficiently targeted geoA in their laboratory strains of actinomycetes, the authors did not detect the gene in the water or sediment of Canadian lakes. The failure to detect geoA in the environmental samples, as also observed in our study of Australian reservoirs, may reflect an undocumented genetic variability of the geoA gene in different geosmin-producing organisms (Ludwig et al. 2007).
In contrast to the qPCR assay, the non-QPCR assay using the degenerate primer set by Ludwig et al. (2007) indicated that geoA and related sesquiterpene genes actually did occur in microorganisms in the Little Nerang reservoir. As mentioned previously, the amino acid sequence of geosmin synthase and some sesquiterpene synthases are similar, and geoA might in fact be a sesquiterpene synthase (Ludwig et al. 2007). Surprisingly, the cloned sequences could only be related to known geoA genes (or sesquiterpenes genes) in cyanobacteria (in species belonging to the Phormidium, Oscillatoria and Aphanizomenon genera), and none of these cyanobacteria were identified in the Little Nerang reservoir. There were no matches to Geitlerinema due to lack of information about the geoA sequence for this genus. Within the Phormidium and Aphanisomenon genera, several species are known to produce geosmin, while species of Oscillatoria appear to produce 2-MIB rather than geosmin (Jüttner & Watson 2007). Among the three cyanobacteria identified in the clones, only Aphanizomenon sp. was found in Wyaralong surface water, but not in Little Nerang (Table S2). Sequences of the selected clones had no matches to known geoA sequences in streptomycetes, although Streptomyces occurred in the reservoirs.
The rather unsuccessful detection of geoA, using published primer sets for cyanobacteria and Streptomyces, suggests that existing information on molecular genetics of the geoA gene(s) is inadequate. Consequently, the design of suitable PCR primers, e.g., for monitoring of potential geosmin-producing organisms in water reservoirs, is correspondingly problematic. Recent studies on biosynthesis and molecular biology of geosmin and 2-MIB synthesis in selected species of cyanobacteria (Tsao et al. 2014; Suurnakki et al. 2015; Wang et al. 2015) may help to develop universal molecular approaches for targeting the geoA as well as the MIBS genes.
Degradation of T&O
Biological degradation of geosmin and 2-MIB is reported to be a slow process (Saito et al. 1999; Klausen et al. 2005), yet there was relatively fast reduction of both compounds during the week-long incubation in this study. In bioreactors with selected bacteria and in aquaculture waste treatment, stimulated biodegradation of geosmin and 2-MIB was measured (Ho et al. 2007; McDowall et al. 2009; Hsieh et al. 2010; Jiang et al. 2007), indicating that natural microbial populations include organisms with the capacity to degrade these compounds. Among specific bacterial species with a high catabolic capacity for geosmin and 2-MIB is Bacillus subtilis (Ma et al. 2015; Ma et al. 2016), and some species within the phylum Proteobacteria (Ho et al. 2007). The reduction in geosmin and 2-MIB in the reservoirs might indicate the presence of similar T&O-degrading organisms, but this was not studied. Loss of geosmin and 2-MIB by volatilization probably also occurred, but studies of the kinetics of volatilization of the two compounds at various temperatures and concentrations suggest that less than 10% was lost from the water as volatile compounds during the week-long incubation period (Li et al. 2012).
The purpose of this study was to test the application of available PCR-based methods for detection of geosmin-producing microorganisms in relation to actual abundance to geosmin in the water. Unfortunately, our results did not provide clues to specific geosmin producers in the reservoirs, and emphasize the need for more research on biology and molecular mechanisms that control production of T&O. Newly published information on the biosynthesis and molecular biology of geosmin and 2-MIB synthesis in selected cyanobacteria may help in designing universal PCR primers for both geoA and MIBS genes. If abundance of specific geosmin and 2-MIB producers can be detected by reliable PCR assays, a rapid prediction of emerging T&O episodes in water reservoirs can be achieved.
We wish to thank Seqwater for collecting the water samples, and analysis of T&O, Ann Chuang (Griffith University) for algal and cyanobacterial identification and enumeration, and U. Rasmussen (University of Copenhagen) for undertaking the QPCR of streptomycete bacteria.