The Development and Application of DDPCR Technology on Quantification of Total Coliforms in Water

: In this research, the detection method for absolute quantification of total coliforms was established based on Droplet Digital Polymerase Chain Reaction (DDPCR) technology using lac Z as the target gene for coliform group detection. The experimental conditions (e.g. primer and probe concentrations, annealing temperatures, etc ) were well optimized. Besides, the linear range, precision and limit of quantification (LOQ) of this method were investigated and evaluated. The results illustrated that the optimal primer concentration was 0.2 µmol/L, whereas the optimal probe concentration was 0.5 µmol/L. The optimal annealing temperature was 56°C. The linear relationship between the total coliform genome DNA concentrations derived from DDPCR and DNA fluorometer was quite good (R 2 = 0.999). The linear range was 3.95 ~ 7.80 × 10 4 copies/20 µL DDPCR reaction system. The LOQ for total coliforms was single copy per reaction system. Practical applications using real water samples collected from water supply system in Macao illustrated that this innovative method possessed high efficiencies and capabilities. This is probably the first research using DDPCR technology to absolutely qualify and quantify total coliforms and successfully applied it in Macao water supply system. The achievements from this research could provide with significant values for setting-up the emergency mechanism of water pollution in early stage.


Introduction
Rapid economic development and its corresponding population explosion are putting incredible strains on our environment. Once water-borne pathogens enter water supply system, it may cause threats to public health and even lives [1]. Therefore, water quality and consequent safety problems have always been our top concerns. Total coliforms, as microbial indicators, have been selected and widely applied to suggest the presence of pathogens and thus to assess the microbiological quality of water [2][3][4][5][6].
Internationally recognized total coliforms detection methods are mainly membrane filtration [7,8], multiple-tube fermentation [9,10], immunomagnetic separation [11,12], etc. Even though these traditional methods have been regarded as the "Golden Standards" worldwide, they pose disadvantages such as longer detection period (several days or a dozen of days), tedious operation procedures (requirement for verification experiments), failure to detect damaged microbes, lower sensitivity and potential biohazards, etc [13][14][15][16]. Therefore, these methods cannot meet the requirement for rapid diagnosis on water quality in water supply system, especially when the water quality suddenly changes, or customer complains. The selection and setup of a safer, faster and simpler detection method with higher specificity and sensitivity is of most significance and urgency. It has been the innovation "hotspot" among water industries [17][18][19][20].
Molecular methods, especially polymerase chain reaction (PCR) technologies are assuming a dominant place in modern diagnostic microbiology because of its rapid detection and higher specificity and sensitivity. Thus, PCR-based technologies have been applied in a wide range of areas, such as clinics, pharmacy, environment, food, agriculture, etc [21][22][23][24]. So far, water quality monitoring and particularly microbial source identification rely heavily on real-time PCR (qPCR) method since the detection targets could be quantified [25][26][27]. In qPCR, quantification is accomplished through comparing the quantification cycle (Ct) results from a sample with unknown detection targets. Despite its wide acceptance, the accuracy and precision of this technology are highly affected due to its intrinsic constraints, such as requirement for standards of known concentration, inhibition of PCR amplification due to substances within water samples, etc [28,29].
Droplet digital PCR (DDPCR) is the most advanced generation PCR without the need for calibration curves. Specifically, due to "water-in-oil" technology, each assay is partitioned into approximately 20,000 highly uniform nanoliter droplets, such that each droplet in the emulsion is an independent reaction and all the droplets are assorted in random fashion. After PCR amplification, the fluorescence of each droplet is individually measured and defined as either positive (presence of detecting target) or negative (absence of detecting target) events. The quantification results can then be obtained directly from the ratio of positive events to total partitions, based on binominal Poisson statistics. This technology can not only shorten the detection period (several hours), but also enhance the sensitivity and absolutely quantify the pathogens. Thus, DDPCR could introduce new level of advanced water-borne pathogen detections in both qualification and quantification [30][31][32][33][34][35][36]. However, seldom researches apply the DDPCR technology in pathogen detections in water environment.

Objectives
In this research, the DDPCR detection method for absolute quantification of total coliforms in Macao water supply system was established by selection of lacZ as the target gene for coliform group detection. By rapid diagnosis of water-borne pathogens (e.g. total coliforms), water treatment plant could take measures to improve the water quality at priority, thus avoiding unnecessary and inestimable lost. The main objectives are: (1) to setup the DDPCR detection method by optimizing the experimental parameters, such as reagent concentrations, annealing temperatures, etc., (2) to verify the capabilities and efficiencies of optimized DDPCR detection method using practical water samples from local water supply system, (3) to work as a demonstration role in setting up the emergency mechanism on microbiological water pollution in Macao.

Materials and Reagents
Primers and Taqman-MGB probes were synthesized by Integrated DNA Technologies (Coralville, IA, USA) unless otherwise mentioned. PowerSoil® DNA Isolation Kit (MOBIO, USA) was used for coliforms DNA extraction. The cellulose esters membrane which had a pore size of 0.45 µm (Merck Millipore, Germany) was used for sample filtration. All the reagents and consumables related to DDPCR were purchased from Bio-Rad (Pleasanton, CA, USA).

Instruments
The Qubit® 3.0 fluorometer (Life Technologies, USA) was used to measure the DNA concentrations after extraction. QX200 TM DDPCR system which includes manual droplet generator, C1000 touch thermal cycler, PX1 PCR plate sealer, droplet reader and QuantaSoft v1.7.4 software was purchased from Bio-Rad (Pleasanton, CA, USA). Rainin® E4-200XLS single-channel electronic pipettes were purchased and applied through the research (Rainin, USA).

Selection of Primer Set and Probe
Total coliforms contain a group of coliform species which includes the genera Citrobacter, Enterobacter, Escherichia, Hafnia, Klebsiella, Serratia and Yersinia [36]. Therefore, the design of primer set becomes the main difficulties when nucleic acid targets quantification technology is applied for multiple coliforms detections. The characteristics of primer set should not only have the capabilities to detect all the required coliform species among different genus, but also exclude those similar species of unrequired ones. In our research, the designed primer set and probe [37] was applied according to the gene sequence of lacZ in GenBank. The sequences of primer set and probe were listed in Table 1.

Setup of DDPCR Detection Procedure
An aliquot of 100 mL water samples was filtered by membrane filtration method and total coliforms were concentrated on a cellulose esters membrane (0.45 µm pore size). Then total coliforms genome DNA was extracted by PowerSoil® DNA Isolation Kit according to manufacturer's instruction and used as the DNA template in DDPCR reaction system. The concentration of the DNA template was then measured by fluorometer immediately. In DDPCR reaction system, each reaction system contained 10 µL of DDPCR TM Supermix, 1 µL of total coliform lacZ genome DNA template, primer set and probe (concentration to be optimized), the final volume was reached to 20 µL by sterilized distilled water.
The aliquot of 20 µL reaction system were placed into QX200 TM droplet generator, which utilized droplet generation oil and microfluidics to partition the samples into 20,000 nanoliter-sized uniform droplets. Almost 40 µL of emulsified samples were pipetted to 96-well PCR plate and then the plate was sealed with the PX1 TM PCR plate sealer.
The template DNA was amplified using C1000 touch thermal cycler with the following cycling conditions: 10 minutes at 95°C initial denaturation, 40 cycles each consisting of a 30 seconds denaturation at 94°C, followed by 60 seconds under the annealing temperature (to be optimized), and 10 minutes at 98°C for a final extension.
After amplification, the 96-well PCR plate was loaded into the QX200 TM droplet reader, then raw fluorescence data from each well were exported from the software (QuantaSoft v1.7.4). Droplets were automatically classified as positive and negative based on Poisson theory. Then, ambiguous results or potentially spurious events were filtered out followed a custom algorithm.
The template DNA with the following cycling conditions: 10 minutes at 95°C initial denaturation, 40 cycles each consisting of a 30 seconds denaturation at 94°C followed by 60 seconds at 56°C annealing temperature, and a final extension at 98°C for 10 minutes. After cycling, droplets were analyzed immediately. All the other conditions were same as in section 3.3.2.
iii. Optimization of Annealing temperatures In DDPCR reaction system, 0.5 µmol/L of the probe concentration and 0.2 µmol/L of respective forward and reverse primer concentrations were applied. Annealing temperatures gradient at 45°C, 46°C, 47.8°C, 50.5°C, 53.6°C, 56.3°C, 58.1°C and 59°C were tested for optimization, respectively. All the other conditions were same as in section 3.3.2.
In DDPCR, each reaction system contained 10 µL of DDPCR TM Supermix, 1 µL of total coliform lacZ genome DNA template, 0.2 µmol/L of primer set and 0.5 µmol/L of probe, the final volume was reached to 20 µL by sterilized distilled water.
The template DNA with the following cycling conditions: 10 minutes at 95°C initial denaturation, 40 cycles each consisting of a 30 seconds denaturation at 94°C followed by 56°C annealing temperature for 60 seconds, and a final extension at 98°C for 10 minutes. After cycling, droplets were analyzed immediately.
Four replicates were applied in each concentration for linear range confirmation.

Evaluation of Precision
The method precision was evaluated based on the relative standard deviation (RSD) derived from DDPCR results. The relationship between the total coliforms lacZ genome concentrations derived from Qubit® 3.0 fluorometer and DDPCR was investigated. Four replicates were applied for method precision investigation.

Validation Study
A total of 50 water samples collected from Macao water supply system were used to verify the capability and efficiency of this innovative DDPCR detection method in the research. The water samples included raw water, treated water and water from networks ( Figure 1). The quantification detection of total coliforms in subsequent samples were processed by DDPCR as described above.

Data Analysis
DDPCR data was initially analyzed in QuantaSoft v1.7.4 software following the manufacturer's recommendation. For the wells which had the accepted droplet less than 10,000 were excluded from analysis. Wells with less than 3 positive droplets were regarded as negative. One standard deviation above negative droplets was set as the fluorescence threshold.
Dose responses and regressions were plotted in Excel 2016. Goodness of fit was confirmed by R 2 values. The limit of quantification (LOQ) which used to compare total coliforms DNA genome concentrations were determined by a t-test. All the significant confidence levels were quoted at 95% (p ≤ 0.05).

Amplification Performance
The interpretation of DDPCR results is based on the accumulated fluorescence of all droplets by the end of measurement. Thus, the accuracy of this detection method is dependent on the intensity of accumulated fluorescence after amplification. In this research, the DDPCR amplification performance on total coliforms genome DNA detection were shown in Figure 2.  One-dimensional scatter plot and histogram figures showed an optimized DDPCR assay with a clear distinction between positive and negative partitions. This good DDPCR amplification performance provided with capabilities for target DNA quantification after optimization.
Regardless of probe concentrations, fluorescent amplitudes of each droplet generated both a positive and negative cluster.
The positive fluorescent amplitude increased as the probe concentration increased, thus more lacZ genome DNA was amplified by DDPCR method. The fluorescent amplitude was the highest when the probe concentration was 2 µmol/L. When the probe concentration was 0.5 µmol/L, it provided with the clearest separation of the negative and positive droplets. Therefore, considering the intensity of positive fluorescent amplitude, stability and the corresponding clearness of droplets separation, the optimized probe concentration was confirmed at 0.5 µmol/L.

Confirmation of Optimized Primer Set Concentration
The primer set concentration was optimized after the probe concentration confirmed at 0.5 µmol/L. The DDPCR amplification performance which was in the form of fluorescent amplitude at different primer concentrations was shown in Figure 4. When the primer concentration was 0.2 µmol/L, it provided with the clearest separation of the negative and positive droplets and the positive cluster was more convergently distributed. As the primer concentration further increased from 0.25 µmol/L to 1 µmol/L, there was no significant difference between positive fluorescent amplitudes (p=0.68). The positive fluorescent amplitude was inhibited when the primer concentration further increased to 1.8 µmol/L. Considering the stability, the corresponding clearness of droplets separation and reagent cost, the optimized primer concentration was confirmed at 0.2 µmol/L.

Confirmation of Optimized Annealing Temperature
Annealing temperatures of DDPCR reaction system were optimized followed by the confirmation of operational primer set and probe concentrations. The DDPCR amplification performance which was in the form of fluorescent amplitude at a serial of annealing gradients was shown in Figure 5. As annealing temperature increased, the positive fluorescent amplitude increased. When the annealing temperature increased to 56.3°C, it showed the highest fluorescent amplitude and the positive cluster was more convergent. As the annealing temperature further increased, the fluorescent amplitude decreased which indicated that higher temperature inhibited amplification. Considering the stability and corresponding clearness of droplets separation, the optimized annealing temperature was 56.3°C.

Prerequisites for Quantification
The average event (number of droplets) generated from DDPCR reactions (n=32) was 15,307±1,528. The average event of accepted droplets at each diluted concentration was above 14,000 which provided the one of the prerequisites for accuracy DDPCR quantification ( Figure 6).
As the concentration of lacZ genome DNA increased (from S7 to S1), the event of positive droplets increased whereas the event of negative droplets decreased. In sample S1, there was no significant difference in events between accepted droplets and positive ones (p=0.52) which illustrated that most of the droplets in sample S1 (the concentration of 10 -1 dilution) were positive droplets. All the No Template Control (NTC) samples without any lacZ genome DNA (worked as negative control) were tested by optimized DDPCR method. There was no single positive droplet generated in all NTC samples (Figure 7). This fluorescent amplitude results demonstrated that all the amplification systems were not contaminated and/or there was no non-specific amplification generated. Thus, this innovative DDPCR method was regarded as high specificity.

Linear Range for Quantification
The dose response relationship between lacZ genome DNA concentrations derived by both fluorometer and optimized DDPCR assay was linear (Figure 8). In this research, the linear range of innovative DDPCR method was 3.95 ~ 7.80 × 10 4 copies/20 µL DDPCR reaction system (R 2 =0.999).
The LOQ of the method refers to the lowest lacZ genome DNA concentration corresponded to a significant increase in the fluorescent amplitude (p ≤0.05) relative to NTC. The lacZ genome DNA concentration (0.2 copies/µL DDPCR reaction system) in sample S5 was the lowest concentration for accurate quantification. Thus, the LOQ of DDPCR method was 4 copies/20µL DDPCR reaction system.

Confirmation of Method Precision
Dose response relationship between four replicates of lacZ genome DNA concentrations derived by optimized DDPCR method and its corresponding RSD was fitted well with the resultant sigmoidal models (Sigmoidal, 4 parameters) (R 2 =0.89). As lacZ genome DNA concentration increased, its corresponding RSD decreased, thus the detection precision increased (Figure 9). When lacZ genome DNA concentration was 4 copies/20µL DDPCR reaction system (LOQ), the RSD of was 65%. When lacZ genome DNA concentration within the range of 57.5 ~ 77,950 copies/20µL DDPCR reaction system, the RSD was less than 5%. Therefore, better precision and accuracy of the method would be obtained if the operational lacZ genome DNA concentration maintained within this range.

Validation from Practical Samples
A total of 50 water samples were collected from both up-streams (3 water reservoirs in Zhuhai, China: marked A, B and C) and Macao local water supply systems (Macao SAR, China). The Macao local water samples included 5 samples from Macao water reservoirs (marked D ~ H), 24 raw water samples from treatment plants, 4 samples from treated water, and 14 samples from networks.
The results derived from DDPCR method after lacZ genome DNA extraction for raw water were shown below ( Table 2 and Table 3).  The results showed that lacZ genome DNA was detected by innovative DDPCR method which indicated that total coliforms existed in all raw water samples.
In Table 2, for reservoir water samples, the lacZ genome DNA concentration of "sample A" from up-streams in Zhuhai was the highest (301.0 ± 21.2 copies/20µL DDPCR reaction system), whilst the lacZ genome DNA concentration of "sample D" from Macao was the lowest (1.5 ± 0.1 copies/20µL DDPCR reaction system).
In Table 3, for raw water samples from water treatment plants, there was no significant difference between lacZ genome DNA concentration derived from DDPCR method (p = 0.65).
DDPCR method was also applied to treated water samples and samples from networks after DNA extraction. The concentration of lacZ genome DNA concentration was below LOQ which proved that no coliforms were detected from those samples.

Conclusions
Even though water-borne pathogen detections using conventional culture-based methods has become the "Golden Standards", it normally takes at least 4~6 days for total coliforms detections from sample-to-result time including verification experiment. The LOQ of this method is 1 Colony Forming Unit (CFU)/100mL. Compared to culture-based methods, qPCR provides faster sample-to-result time, higher specificity and sensitivity, thus it is a widely accepted and applied method for water-borne pathogen detections. Especially, qPCR for Legionella detection has become an internationally recognized method [37]. However, it is still a relative quantification method based on Ct and the absolute concentration of target samples remains unknown until calibrated with standard samples. Therefore, the reliability and consistency of required standards have tremendous impact on the accuracy of quantification. Besides, qPCR is susceptible to inhibitors which might naturally exist in environmental samples and these inhibitors are normally complex and often contain substances that interfere with qPCR amplification [38].
Digital PCR, especially DDPCR, has the potential to overcome the limitations of qPCR based on its partitioning process and Poisson statistics. Thus, the target DNA copies can be estimated by counting the frequency of positive partitions and no external standards are needed for quantification of unknown samples. Most of studies regarding DDPCR have focused on food, clinical and pharmaceutical areas, little is known for DDPCR applications in environmental samples, especially in water supply system.
The interpretation of DDPCR result is based on the accumulated fluorescence of all droplets by the end of measurement. Thus, the accuracy of this detection method is dependent on the intensity of accumulated fluorescence after amplification. Very few DDPCR publications detail any information regarding PCR assay optimization. Here the optimal experimental conditions, such as reagent concentrations and annealing temperatures, were well confirmed to further increase the accuracy and consistency of the assay. In addition, DDPCR partitions are made up of water-in-oil emulsion droplets which are quite fragile. Therefore, single-channel electronic pipettes were used through all the experiments to avoid the loss of droplet events. Furthermore, the sample-to-result time was only 5 hours using the innovative DDPCR method in this research and the LOQ of this method was only single copy/20µL DDPCR reaction system which was in accordance with other studies in pathogen quantifications [39,40]. Practical applications using real water samples collected from water supply system in Macao illustrated that this innovative method possessed high efficiencies and capabilities.
In this research, a safer, faster and simpler detection method with higher specificity and sensitivity for absolute total coliforms qualification and quantification was setup based on DDPCR technology. This is probably the first research successfully applied DDPCR technology in Macao water supply system. The achievements from this research could provide with significant values and work as a demonstration role for setting-up the emergency mechanism of water pollution in early stage. Future field case studies are needed to further evaluate its full spectrum of other water-borne pathogens and potential limitations to be concerned.