Simple Polydisperse Droplet Emulsion PCR with ... - ACS Publications

ACS Paragon Plus Environment. Analytical Chemistry. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13 ... The polydisperse droplet system allows for accurate ...
2 downloads 0 Views 906KB Size
Subscriber access provided by Kaohsiung Medical University

Article

Simple Polydisperse Droplet Emulsion PCR with Statistical Volumetric Correction is Comparable with Microfluidic ddPCR Samantha Byrnes, Tim C Chang, Toan Huynh, Anna Astashkina, Bernhard Weigl, and Kevin Paul Nichols Anal. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.analchem.8b01988 • Publication Date (Web): 09 Jul 2018 Downloaded from http://pubs.acs.org on July 10, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 10 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

Simple Polydisperse Droplet Emulsion PCR with Statistical Volumetric Correction is Comparable with Microfluidic ddPCR Samantha A. Byrnes, Tim C. Chang, Toan Huynh, Anna Astashkina, Bernhard H. Weigl, Kevin P. Nichols* Intellectual Ventures Laboratory, Bellevue, WA, USA Corresponding Author *Kevin P. Nichols: [email protected] Intellectual Ventures Laboratory, 14360 SE Eastgate Way, Bellevue, WA, 98007, USA

ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 10

ABSTRACT: Nucleic acid amplification technology, such as PCR, has enabled highly sensitive and specific disease detection and quantification, leading to more accurate diagnosis and treatment regimens. Lab-on-a-chip applications have developed methods to partition single biomolecules, such as DNA and RNA, into picoliter-sized droplets. These individual reaction vessels lead to digitization of PCR enabling improved time-to-detection and direct quantification of nucleic acids without a standard curve, therefore simplifying assay analysis. Though impactful, these improvements have generally been restricted to centralized laboratories with trained personnel and expensive equipment. To address these limitations and make this technology more applicable for a variety of settings, we have developed a statistical framework to apply to droplet PCR performed in polydisperse droplets prepared without any specialized equipment. The polydisperse droplet system allows for accurate quantification of ddPCR and RT-ddPCR that is comparable to commercially available systems such as BioRad’s ddPCR. Additionally, this approach is compatible with a range of input sample volumes, extending the assay dynamic range beyond that of commercial ddPCR systems. In this work, we show that these ddPCR assays can reduce overall assay time while still providing quantitative results. We also report a multiplexed ddPCR assay and demonstrate proof-of-concept methods for rapid droplet preparation in multiple samples simultaneously. Our simple polydisperse droplet preparation and statistical framework can be extended to a variety of settings for the quantification of nucleic acids in complex samples.

Quantification of biomolecules, such as DNA and RNA, can improve disease diagnosis and aid in monitoring of disease progression and treatment(1). Traditional polymerase chain reaction (PCR) allows for the qualitative detection of specific nucleic acid sequences. Quantitative PCR (qPCR) expanded this capability to allow for quantitation of input targets based on a standard curve, assuming the samples and standards have the same amplification efficiency(2,3). There have also been reports of qPCR achieving single-molecule detection(4). However, the required logarithmic standard curve can impact absolute quantification when there is variation between replicates, especially if samples and standards are in different matrices(5–7). Additionally, qPCR is prone to inhibition when using complex samples, which can reduce assay efficiency and therefore quantitation(8,9). Digital PCR has been developed in part to overcome the limitations of qPCR, where bulk reaction volumes are partitioned into thousands or millions of small reaction vessels, or droplets, each containing zero or one target molecule based on the concept of limiting dilutions(10). This approach digitizes the output signal because each positive droplet will correspond to the amplification of just one target. The number of positive droplets can be counted resulting in absolute target quantification without the need for a standard curve(6,8,9). Additionally, the picoliter-sized (pL-sized) reaction volumes of droplet digital PCR (ddPCR) reduces the effects of diffusion resulting in faster reactions and higher signal to noise ratios (SNRs)(11). Initial work comparing qPCR and ddPCR shows agreement for assay efficiency and quantification across these two methods, often with lower limits of detection and higher sensitivities reported for ddPCR(6,9,12,13). There have also been reports that ddPCR has as well as reduced sensitivity to amplification inhibitors compared to traditional qPCR(14). Many implementations of ddPCR rely on the creation of monodisperse droplets(15). A common and effective way to create uniform droplets is by using microfluidic chips(15–17). Beer et al reported the first demonstration of ddPCR in a microfluidic chip(18); since then, there have been multiple reports of custom chips developed to perform droplet generation and ddPCR(16,17,19,20). Zhu et al reviewed much of the early development and applications of ddPCR technology(15). In recent years, several commercial ddPCR systems have become available. One of the most common is the BioRad ddPCR system which has been demonstrated with complex samples(13,21) and reported to detect low levels of pathogen-

specific nucleic acids within single cells(12,22). Dropletbased amplification also enables highly multiplexed reactions. Reports have demonstrated multiplexed ddPCR reactions using the commercially available BioRad ddPCR system(13,21,23). In addition to ddPCR, RNA-based amplification in droplets (RT-ddPCR) has been reported. Albayrak et al described a partial RT-ddPCR reaction by first performing the RT step in a bulk solution followed by droplet formation and ddPCR(22). There have also been reports of custom microfluidic chips for performing RT-ddPCR using agarose droplets(24) and indroplet cell lysis followed by RNA amplification(19). Although these demonstrations have been effective and achieved very low limits of detection, they all rely on expensive commercial systems or specialized microfluidic chips that require trained technicians to operate. One of the major limiting steps in this process is the formation of monodisperse droplets(15,25). Earlier work theorized that digital assays could be performed as multivolume systems when reaction volumes are separated into discrete wells(26,27). Sidore et al demonstrated single cell genome amplification in polydisperse droplets for sequencing applications, but found the polydispersity introduced bias in the sequencing coverage(28). Building on these efforts, we reported using polydisperse droplet emulsions with a statistical framework for highly sensitive cell culture with encapsulated bacteria(29). The statistical correction overcomes any bias as a result of polydisperse droplet populations, and the simplicity of a shaken emulsion preparation removes the need for expensive microfluidic equipment and trained personnel for droplet formation. In this work, we compare our results to the commercial BioRad ddPCR system and show excellent agreement in quantitation. Our approach can also potentially process significantly higher sample volumes (mL) and does not result in any sample loss. The BioRad system is limited to 25 µL reaction volumes and removes droplets that fall outside of its set size limitation, resulting in data loss and a lack of truly “absolute” quantitation. Additionally, we’ve demonstrated droplet formation in a 96-well plate which takes seconds to prepare instead of nearly an hour for the same number of samples using the BioRad automated droplet generation system. EXPERIMENTAL SECTION

ACS Paragon Plus Environment

Page 3 of 10 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

Reagents. All primers and probes for ddPCR were purchased from IDT (Skokie, IL, USA). Primers and probes for RTddPCR were purchased from ThermoFisher Scientific (Waltham, MA, USA). Supermix for ddPCR and RT-ddPCR, droplet generation oil (cat # 1863005), and droplet analysis oil were purchased from BioRad (Hercules, CA, USA). Countess cell counting chambers for droplet analysis using a microscope were purchased from Fisher Scientific (Hampton, NH, USA). N. gonorrhoeae gDNA was purchased from ATCC (cat # 700825DQ, Manassas, VA, USA) and E. coli gDNA was extracted from E. coli bacteria (strain BL21) grown in our lab. Cell lysis and DNA purification reagents were purchased from Sigma-aldrich (St. Louis, MO, USA) and ThermoFisher Scientific (Waltham, MA, USA). E. coli gDNA extraction and purification. E. coli gDNA was extracted and purified from bacterial cells using a standard bead beating and ethanol precipitation protocol. Briefly, cells from an overnight culture grown in LB were harvested by spinning at 13,000 xg for 3 minutes. The supernatant was discarded and the pellet was resuspended in 10 mM Tris, pH 8. Cells were lysed by bead beating for 3 one-minute cycles with one minute of rest between each. Cell lysate was treated with 0.1x volumes of 3M sodium acetate, 2x volumes of ice cold absolute ethanol, and 0.01x volumes of 20 mg/mL glycogen followed by incubation at -20 ˚C for 15 minutes and centrifugation at 21,000 xg for 15 minutes to pellet DNA. After pelleting, the supernatant was discarded and the pellet was washed with 10x volumes of 70% absolute ethanol followed by centrifugation at 21,000 xg for 5 minutes. The supernatant was discarded and the pellet was allowed to air dry for 10 minutes at room temperature. Finally, the pellet was resuspended in 1x volume of molecular biology grade water and heated to 37 ˚C for 10 minutes. A nanodrop was used to quantify the concentration of the purified E. coli gDNA. Droplet preparation. Droplets were prepared using either the BioRad automated droplet generator (BioRad, Hercules, CA, USA) when comparison experiments with the BioRad system were being performed, or our simplified in-house method to create a polydisperse droplet distribution. Briefly, the BioRad automated droplet generator uses a microfluidic chip to prepare ~100 µm diameter droplets (~520 pL) from 25 µL aqueous reactions and an oil/surfactant mixture(30). For this work, our polydisperse method used 100 µL of the same aqueous reaction solution used in the BioRad system and 200 µL of the BioRad oil/surfactant mixture. These immiscible fluids were vortexed at maximum speed for 10-20 seconds to create a population of polydisperse droplets ranging in diameter from ~1.5 to 13,117 µm with a medium diameter of ~56 µm (90 pL), Figure S1. We used Fisher Scientific (Hampton, NH, USA) 1.7 mL Eppendorf tubes for droplet generation and a VWR analog vortex mixer (cat. # 10153-838; Radnor, PA, USA). After formation, emulsions were allowed to settle at room temperature for 5 minutes followed by droplet transfer via pipette to 0.2 mL PCR tubes (BioRad, Hercules, CA, USA) for amplification. Singleplexed PCR in droplets. DNA targets were amplified using the BioRad Supermix for ddPCR. Final primer and probe concentrations were 500 nM and 250 nM, respectively.

Primer and probe sequences for the E. coli rodA(31) and N. gonorrheae porA(32) genes are available in the Supporting Information. The droplet-based reactions were run in either a C100 or C1000 thermocycler (BioRad, Hercules, CA, USA) using the following protocol: 95 ˚C hold for 10 minutes, 20-40 cycles of 95 ˚C for 30 seconds and 60 ˚C for 1 minute. Fluorescent data were collected using either the BioRad ddPCR reader for BioRad samples (BioRad, Hercules, CA, USA) or a Nikon Ti2 fluorescent microscope for the polydisperse samples (Minato, Tokyo, Japan). Multiplexed PCR in droplets. Multiplexed reactions were run with primers and probes for both the E. coli rodA gene and N. gonorrhoeae porA gene as described above. In the BioRad ddPCR system, the E. coli probe was labeled with a Hex fluorophore and the N. gonorrhoeae probe was labeled with a FAM fluorophore (IDT, Skokie, IL, USA). In our in-house system, the E. coli probe was labeled with a Texas Red fluorophore (IDT, Skokie, IL, USA) and the N. gonorrhoeae probe was labeled with a FAM fluorophore (IDT, Skokie, IL, USA). All other reactions conditions remained unchanged. In the BioRad ddPCR system, multiplexed reactions were run with a maximum input of 104 copies/reaction, as recommended by the manufacturer(30). In our in-house samples, multiplexed reactions were run with a maximum of 105 copies/reaction due to the increased dynamic range our approach enables. One-step RT-ddPCR in droplets. RNA targets were amplified using the BioRad one-step RT-ddPCR advanced kit for probes with a 20x primer/probe reaction mix for human GAPDH gene (ThermoFisher Scientific, Waltham, MA, USA). An additional 500 nM (final concentration) of each primer was added to the reaction. The droplet-based reactions were run in either a C100 or C1000 thermocycler (BioRad, Hercules, CA, USA) using the following protocol: 50 ˚C hold for 60 minutes, 95 ˚C hold for 10 minutes, 40 cycles of 95 ˚C for 30 seconds and 60 ˚C for 1 minute. Fluorescent data were collected using either the BioRad ddPCR reader (BioRad, Hercules, CA, USA) or a Nikon Ti2 fluorescent microscope (Minato, Tokyo, Japan). Droplet analysis: BioRad droplets. Droplets prepared and run with the BioRad ddPCR system were analyzed using their QuantaSoft Analysis Pro software (BioRad, Hercules, CA, USA). The software analyzes all droplets in a sample and only reads those that meet the size standard. The software then reports the number of droplets that are positive for one or both available fluorophores (Hex and FAM). Image acquisition for polydisperse droplets. Droplets were imaged in a countess cell counting chamber slide using a Nikon Eclipse Ti2 inverted microscope (Nikon Instruments, Melville, NY, USA) with epifluorescence illumination. An automated x-y stage and a 20x objective were used to acquire images. Large stitched images were generated with a 15% overlap using 26 x 13 individual fluorescence images. Droplets in the countess slide were focused manually prior to imaging. All images were acquired with a 14-bit Nikon DS-Qi2

ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 10 

CMOS camera and Nikon NIS-Elements AR software to control image acquisition.

1 

Image processing, droplet analysis, and statistical corrections for polydisperse droplets. ImageJ(33) image analysis software was used to automate the quantification of positive droplets in each image. Briefly, images were cropped to remove edges with uneven illumination, converted to an 8-bit format, and thresholded before positive droplets were identified. The same image processing and analysis was applied to each image. For multiplexed images, the two fluorescent channels were merged prior to analysis. The MATLAB Image Processing Toolbox (Mathworks, Natick, MA, USA) was used for droplet size identification. Full image analysis details are provided in the Supporting Information. Multiplexed droplet statistical analysis. Using a modified Poisson corrected number of positive droplets for each target (see the next section for details), we were able to statistically determine the expected number of double positive droplets and compare the predicted values to experimental data. RESULTS AND DISCUSSION Droplet size distribution and statistical correction for polydisperse droplets. In this work, we have developed a statistical model to apply to ddPCR performed in polydisperse droplets. This method results in highly sensitive and quantitative digital amplification without the need for any specialized equipment for droplet preparation. A larger droplet is more likely to contain a target molecule than a smaller droplet; therefore, at assay conclusion, larger droplets have a higher probability of turning “on”, i.e. becoming a positive droplet. In our system with a polydisperse droplet distribution, there are a range of probabilities that a droplet turns on based on its size. Previously, we developed a statistical procedure to calculate the expected number of positive droplets in a system with a polydisperse droplet size distribution(29), and have applied the same statistical framework here. We derived a method to calculate the concentration of the target, . According to Poisson statistics, for a droplet with volume , the probability to turn on is 1    . For a droplet in a sample of droplets with volume probability distribution function  , the probability to turn on is  1       1       Using volume values measured in a separate experiment ( ,  , … ,  ) to estimate  , we have 

 1   

1     

From the experiment, the estimate of  is ⁄, where  is the number of positive droplets out  droplet in total. Therefore, we numerically solve for  using the following.

1       

Singleplexed PCR in droplets. The first demonstration of the polydisperse droplet system determined a limit of detection (LoD) for a singleplexed PCR reaction in droplets. We compared amplification of E. coli DNA in our polydisperse droplet system to the commercially available BioRad ddPCR system. Our results align well with the BioRad results, and show a linear relationship between the expected number of positive droplets based on the input copy number, Figure 1. Additionally, our system had a larger dynamic range than the BioRad system which is limited by an input volume of 25 µL/reaction. We used a three-parameter logistic curve to statistically compare the LoDs and 95% confidence intervals (CIs) of each method, similar to approaches used by other groups(37). The LoD of our method was 0.682 copies/µL (95% CI 0.526 – 0.665 copies/µL) compared to the BioRad LoD of 2.4 copies/µL (95% CI 1.85 – 2.95 copies/µL). Both methods result in similar LoDs with our system being slightly more sensitive due to the cleaner no template controls (negative control samples for PCR where no DNA is added to the reaction).

Figure 1. Singleplexed ddPCR in our polydisperse droplet system compared to the commercially available BioRad ddPCR system. Both systems show the expected number of positive droplets based on known input concentrations, but the polydisperse system has a larger dynamic range than the BioRad system. Averages of at least N=3 for polydisperse droplets and at least N=4 for BioRad droplets are reported with error bars representing ± one standard deviation. The zeroes are represented by straight lines and shaded areas for the standard deviations. The LoD of each system was calculated using a three-parameter logistic curve. The 1-to-1line shows that the data is approximately linear across the range of inputs tested. The BioRad system has a few practical limitations as a result of the restricted input volume per reaction(30). This volume limits the total positive counts per reaction to a maximum on the order of 104 copies input resulting in a reduced dynamic range(12). To process more concentrated samples or larger

ACS Paragon Plus Environment

Page 5 of 10 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

volumes with the BioRad system, inputs must be diluted and/or split into multiple reaction wells – increasing both complexity and potential cost. The polydisperse system is not limited by sample volume and can create droplets from small (100 µL) or large volumes (10 mL) in seconds, effectively extending the dynamic range(36). We have also demonstrated our droplet forming method in a 96-well plate to enable rapid, parallel droplet formation of 96 samples in only seconds, Figure S3. The BioRad system also has strict droplet size requirements, and droplets that fall outside of the narrow range are removed from the system resulting in data loss and lack of truly “absolute” quantitation, Figure S4. Our polydisperse droplet method, on the other hand, analyzes droplets regardless of size. We also evaluated the time-to-detection for the droplet-based reactions by varying the number of amplification cycles. Both the polydisperse and BioRad systems performed similarly with detectable signals from positive droplets appearing after 25

cycles. Both amplifications were complete by 30 cycles for these assays, Figure 2A. These results also indicate that ddPCR assay times can be reduced by 25% (from 40 to 30 cycles) without compromising assay efficiency or quantification. Previous work has demonstrated this concept in monodisperse droplet systems(18), like the BioRad ddPCR system, and we have here shown that this concept holds true for polydisperse droplet distributions. In our polydisperse droplet system, we explored the relationship between time-to-detection and droplet size. We expected smaller droplets with target to be detectable sooner than larger droplets with target because smaller droplets have less volume for signal diffusion and therefore more concentrated signal. Our data aligns with this expectation; as cycle number increases, the size distribution of positive droplets expands to include larger droplets, Figure 2B. This may be useful as an additional validation method for predicted concentrations.

Figure 2. Time-to-detection analysis for ddPCR experiments. A. The BioRad and polydisperse ddPCR methods show similar time-to-detection with signal from positive droplets detectable by 25 cycles, and fully developed by 30 cycles. These results indicate that assay times can be reduced by 25% (from 40 to 30 cycles). Averages of at least N=8 for both methods are reported with error bars representing ± one standard deviation. B. Relationship between droplet size and time-to-detection in the polydisperse droplet system. On average, smaller droplets show positive signals sooner than larger droplets due to smaller volumes for diffusion and therefore a higher concentration of fluorophore. This is also apparent as the number and size of positive large droplets increase with increasing cycle number. ** Note, the distribution for 35 and 40 cycles continues to larger droplet sizes (maximum ~500 pL), but was cropped to more directly show the shift from 20 – 40 cycles.

ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

In addition to the effects on SNR ratio, varying droplet size may also influence reaction rates. Smaller droplets will have less space for reagent diffusion, and therefore require less time for components to find each other and react. For example, primers may take longer to find their target in a larger droplet potentially resulting in a slower reaction. Therefore, assays run in polydisperse systems could help optimize reaction speeds by comparing signal generation across different droplet sizes. Multiplexed PCR in droplets. A major advantage to sequence-specific amplification, like PCR, is the ability to multiplex reactions to simultaneously identify multiple targets. The BioRad ddPCR system is capable of multiplexing, which has been demonstrated in complex samples(13,21). To further expand the capabilities of our polydisperse droplet system, we have developed a multiplexed proof-of-principle assay targeting two different pathogens within one sample. To compare the polydisperse and BioRad multiplexed ddPCR assays, we used the percent of single positive droplets to predict the expected number double positives. For example, if there were 100 total droplets with 10 positive for each target (10% positive), then there should be 1 double positive droplet (10% positive target 1 * 10% positive target 2 = 1% double positive). We can then align this expected percent of double positive droplets with the actual (Poisson corrected) counts

Page 6 of 10

obtained from both the BioRad and polydisperse systems. Additionally, this method of comparison normalizes the differences between the systems in terms of total number of droplets. Both ddPCR methods aligned well with the expected number of double positive droplets, indicating that our simplified, polydisperse method works as well as gold-standard commercial systems, Figure 3A. Additionally, our polydisperse system showed a tighter distribution (95% CI) compared to the BioRad system, potentially because our system does not eliminate any droplet, regardless of size. The BioRad system, on the other hand, has a strict droplet size requirement and removes droplets that fall outside of this range resulting in data loss. The number of removed droplets varies between samples, Figure S4, which may account for the increase in variation seen in the multiplexed reaction. Double positive droplets in the polydisperse system were identified by overlapping fluorescent signals and appeared yellow in images, Figure 3B. This co-localization of color within a single droplet has been previously shown to identify colocalization of nucleic acids and target cells(19). To the best of our knowledge, we have shown the first published demonstration of multiplexed ddPCR in a polydisperse droplet system. Additionally, our methods do not require the expensive, specialized equipment for droplet preparation.

ACS Paragon Plus Environment

Page 7 of 10 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

Figure 3. Multiplexing in ddPCR. A. Comparing the expected to the actual number of double positive droplets for both the BioRad and our polydisperse droplet systems. Both 95% confidence intervals align well with the expected outcome (1-to-1 line in black), but the BioRad system is more variable than our polydisperse method. This variability may result from an unknown amount of data loss in the BioRad system when droplet sizes fall outside of the strict size range. Our analysis of the polydisperse system, on the other hand, is independent of size allowing for more data to be collected and more accurate results. Averages of N=6 for polydisperse droplets and N=4 for BioRad droplets are reported with error bars representing ± one standard deviation and solid polygons representing the 95% confidence intervals of the linear regressions. B. An example image of a multiplexed reaction from our polydisperse droplet system. Red droplets are positive for E. coli, green droplets are positive for N. gonorrhoeae, and yellow droplets are positive for both targets. The larger black spots are air bubbles introduced when loading the sample into the microscope slide. Scale bar = 1mm. One-step RT-ddPCR. In addition to multiplexed ddPCR, the polydisperse droplet system successfully performed an RTddPCR assay in a one-step reaction. Our results align well with the BioRad system, but tend to over-estimate droplet counts at lower concentrations of input RNA, Figure 4. The LoD of the polydisperse system was 269 copies/µL (95% CI 150 – 388 copies/µL) compared to the BioRad LoD of 8.9 copies/µL (95% CI 5.0 – 12.8 copies/µL), see Figure S5 for LoD analysis curves. The BioRad system is more sensitive, likely due to the cleaner no template controls (negative control samples for PCR where no DNA is added to the reaction) and larger variation between samples within the polydisperse system. Similar to the ddPCR work, the polydisperse method can quantify higher droplet counts than the BioRad system due to differences in number of partitions and limitation in volumes processed by BioRad.

Figure 4. RT-ddPCR in the polydisperse droplet system compared to the commercially available BioRad ddPCR system. Averages of at least N=4 for each system are reported with error bars representing ± one standard deviation. The zeroes are represented by straight lines and shaded areas for the standard deviations. The LoD of each system was calculated using a three-parameter logistic curve. The 1-to-1line shows that the data is approximately linear across the range of inputs tested. For the polydisperse system, a potential source of error may come from variation of the RT-step, which has been noted by previous work performed in uniform droplets(19,24,36); this variation may be exacerbated in a sample with polydisperse droplet sizes. Variation may result from different diffusion rates in smaller v. larger droplets therefore leading to different

ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

amounts of generated cDNA. To address this potential issue, future work can focus on optimization of the reverse transcriptase concentration and reaction conditions. CONCLUSIONS Here we have demonstrated a polydisperse droplet emulsions method with a statistical correction that is comparable with microfluidic ddPCR. First, the polydisperse method was shown to accurately quantify input target DNA with a similar LoD and a larger dynamic range than a commercially available system. Also, our method requires significantly less equipment and time to execute compared to droplet-forming microchips (seconds to prepare droplets for 96-samples v. an hour with the BioRad chip-based system). Additionally, the polydisperse droplet system matched the improved time-todetection of the BioRad system, resulting in a 25% reduction in required assay time. This polydisperse method and statistical correction overcome a few practical limitations of the BioRad system including data loss due to droplet size. The polydisperse nature of our system enables the measurement of droplets regardless of size. We have also successfully demonstrated a multiplexed ddPCR assay and an RT-ddPCR assay in an emulsion with a polydisperse droplet size distribution. Importantly, this statistical approach also enables the quantification of nucleic acids in a sample without having to measure the actual distribution of droplet sizes, even in very polydisperse samples. The combination of rapid droplet preparation and this statistical framework results in significantly reduced complexity for an end user which can lead to faster adaption in a variety of locations including limited resource settings. The assays in this work were characterized using a traditional laboratory microscope; but for future applications, a reader will be used in conjunction with the statistical method to quantify samples. Future work will aim to demonstrate this improved and affordable reader technology with realistic, complex samples.

SAB – designed experiments and collected data for all assays, developed analysis protocols, wrote and edited the manuscript TC – collected data for single and multiplexed ddPCR assays, developed analysis protocols, wrote and edited the manuscript TH – developed the statistical model, wrote and edited the manuscript AA – collected data for the RT-ddPCR assay, wrote and edited the manuscript BHW – designed experiments, wrote and edited the manuscript KPN – designed experiments, wrote and edited the manuscript

ACKNOWLEDGMENT Funding provided by The Global Good Fund I, LLC (www.globalgood.com). The authors also gratefully acknowledge Bill Gates and Nathan Myhrvold for their support and supervision. We thank David Bell and Akos Somoskovi from Global Good, David Gasperino for help with the LoD analysis, and Taylor Moehling from Global Good’s Virtual Research Analyst Network for background information gathering.

REFERENCES 1. 2.

3. 4.

5.

6.

7.

ASSOCIATED CONTENT

8.

Supporting Information The Supporting Information is available free of charge on the ACS Publications website. Supporting information includes 5 figures and 1 table.

There is no safety information to declare and no unexpected or significant hazards associated with this work.

9.

10.

11.

AUTHOR INFORMATION

12.

Corresponding Author *Kevin P. Nichols: [email protected] Intellectual Ventures Laboratory, 14360 SE Eastgate Way, Bellevue, WA, 98007, USA.

13.

Author Contributions The manuscript was written through contributions of all authors and all authors have given approval to the final version of the manuscript.

Page 8 of 10

14.

15.

Sunney Xie X, Yu J, Yang WY. Living Cells as Test Tubes. Science (80- ). 2006;312(5771):228–30. Dhanasekaran S, Doherty TM, Kenneth J. Comparison of different standards for real-time PCR-based absolute quantification. J Immunol Methods. 2010;354(1–2):34–9. Bar T, Kubista M, Tichopad A. Validation of kinetics similarity in qPCR. Nucleic Acids Res. 2012;40(4):1395–406. Kralik P, Ricchi M. A basic guide to real time PCR in microbial diagnostics: Definitions, parameters, and everything. Front Microbiol. 2017;8(FEB):1–9. Brankatschk R, Bodenhausen N, Zeyer J, Burgmann H. Simple absolute quantification method correcting for quantitative PCR efficiency variations for microbial community samples. Appl Environ Microbiol. 2012;78(12):4481–9. Hayden RT, Gu Z, Ingersoll J, Abdul-Ali D, Shi L, Pounds S, Caliendo AM. Comparison of droplet digital PCR to real-time PCR for quantitative detection of cytomegalovirus. J Clin Microbiol. 2013;51(2):540–6. Karlen Y, McNair A, Perseguers S, Mazza C, Mermod N. Statistical significance of quantitative PCR. BMC Bioinformatics. 2007;8:1–16. Zhao Y, Xia Q, Yin Y, Wang Z. Comparison of Droplet Digital PCR and Quantitative PCR Assays for Quantitative Detection of Xanthomonas citri Subsp. citri. PLoS One. 2016;11(7):e0159004. Verhaegen B, De Reu K, De Zutter L, Verstraete K, Heyndrickx M, Van Coillie E. Comparison of droplet digital PCR and qPCR for the quantification of shiga toxin-producing Escherichia coli in bovine feces. Toxins (Basel). 2016;8(5):1–11. Burdukiewicz M, Rödiger S, Sobczyk P, Menschikowski M, Schierack P, Mackiewicz P. Methods for comparing multiple digital PCR experiments. Biomol Detect Quantif. 2016;9:14–9. Sakakihara S, Araki S, Noji H. A single-molecule enzymatic assay in a directly accessible femtoliter droplet array †. Lab Chip. 2010;10:3355–62. Henrich TJ, Gallien S, Li JZ, Pereyra F, Kuritzkes DR. Lowlevel detection and quantitation of cellular HIV-1 DNA and 2LTR circles using droplet digital PCR. J Virol Methods [Internet]. 2012;186(1–2):68–72. Available from: http://dx.doi.org/10.1016/j.jviromet.2012.08.019 Shehata HR, Li J, Chen S, Redda H, Cheng S, Tabujara N, Li H, Warriner K, Hanner R. Droplet digital polymerase chain reaction (ddPCR) assays integrated with an internal control for quantification of bovine, porcine, chicken and turkey species in food and feed. PLoS One. 2017;12(8):1–17. Sedlak RH, Kuypers J, Jerome KR. A multiplexed droplet digital PCR assay performs better than qPCR on inhibition prone samples. Diagn Microbiol Infect Dis. 2014;80(4):285–6. Zhu Z, Jenkins G, Zhang W, Zhang M, Guan Z, Yang CJ.

ACS Paragon Plus Environment

Page 9 of 10 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

Analytical Chemistry Single-molecule emulsion PCR in microfluidic droplets. Anal Bioanal Chem. 2012;403(8):2127–43. Chen J, Luo Z, Li L, He J, Li L, Zhu J, Wu P, He L. Capillarybased integrated digital PCR in picoliter droplets. Lab Chip [Internet]. 2018;18(d):412–21. Available from: http://xlink.rsc.org/?DOI=C7LC01160A Tanaka H, Yamamoto S, Nakamura A, Nakashoji Y, Okura N, Nakamoto N, Tsukagoshi K, Hashimoto M. Hands-off preparation of monodisperse emulsion droplets using a poly(dimethylsiloxane) microfluidic chip for droplet digital PCR. Anal Chem. 2015;87(8):4134–43. Beer NR, Hindson BJ, Wheeler EK, Hall SB, Rose KA, Kennedy IM, Colston BW. On-chip, Real-Time, Single-Copy Polymerase Chain Reaction in Picoliter Droplets. Anal Chem. 2007;79(22):8471–5. Kim SC, Clark IC, Shahi P, Abate AR. Single-Cell RT-PCR in Microfluidic Droplets with Integrated Chemical Lysis. Anal Chem. 2018;90(2):1273–9. Zhang H, Jenkins G, Zou Y, Zhu Z, Yang CJ. Massively parallel single-molecule and single-cell emulsion reverse transcription polymerase chain reaction using agarose droplet microfluidics. Anal Chem. 2012;84(8):3599–606. Cai Y, He Y, Lv R, Chen H, Wang Q, Pan L. Detection and quantification of beef and pork materials in meat products by duplex droplet digital PCR. PLoS One. 2017;12(8):1–12. Albayrak C, Jordi CA, Zechner C, Lin J, Bichsel CA, Khammash M, Tay S. Digital Quantification of Proteins and mRNA in Single Mammalian Cells. Mol Cell. 2016;61(6):914– 24. Whale AS, Huggett JF, Tzonev S. Biomolecular Detection and Quantification Fundamentals of multiplexing with digital PCR. Biomol Detect Quantif [Internet]. 2016;10:15–23. Available from: http://dx.doi.org/10.1016/j.bdq.2016.05.002 Zhang H, Li XF, Le, XC. Binding-induced DNA assembly and its application to yoctomole detection of proteins. Analytical Chemistry. 2012: 84, 877-884. Chen Z, Liao P, Zhang F, Jiang M, Zhu Y, Huang Y. Centrifugal micro-channel array droplet generation for highly parallel digital PCR. Lab Chip [Internet]. 2017;17:235–40. Available from: http://dx.doi.org/10.1039/C6LC01305H Kreutz JE, Munson T, Huynh T, Shen F, Du W, Ismagilov RF. Multiplexed quantification of nucleic acids with large dynamic range using multivolume digital RT-PCR on a rotational

27.

28.

29.

30.

31.

32.

33. 34.

35.

36.

37.

SlipChip tested with HIV and hepatitis C viral load. Anal Chem. 2011;83:8158–68. Shen F, Sun B, Kreutz JE, Davydova EK, Du W, Reddy PL, Joseph LJ, Ismagilov RF. Multiplexed quantification of nucleic acids with large dynamic range using multivolume digital RTPCR on a rotational SlipChip tested with HIV and Hepatitis C viral load. J Am Chem Soc. 2011;133(44):17705–12. Sidore AM, Lan F, Lim SW, Abate AR. Enhanced sequencing coverage with digital droplet multiple displacement amplification. Nucleic Acids Res. 2016;44(7):e66. Byrnes SA, Phillips EA, Huynh T, Weigl BH, Nichols KP. Polydisperse emulsion digital assay to enchance time to detect and extend dynamic range in bacterial cultures enables by a statistical framework. Analyst. 2018;143:2828–36. BioRad. QX200 Droplet Digital PCR System [Internet]. Available from: http://www.bio-rad.com/en-us/product/qx200droplet-digital-pcr-system?ID=MPOQQE4VY Chern EC, Siefring S, Paar J, Doolittle M, Haugland R a. Comparison of quantitative PCR assays for Escherichia coli targeting ribosomal RNA and single copy genes. Lett Appl Microbiol. 2011;52:298–306. Hjelmevoll SO, Olsen ME, Sollid JUE, Haaheim H, Unemo M, Skogen V. A fast real-time polymerase chain reaction method for sensitive and specific detection of the Neisseria gonorrhoeae porA pseudogene. J Mol Diagn. 2006;8(5):574–81. Abràmoff MD, Hospitals I, Magalhães PJ, Abràmoff M. Image Processing with ImageJ. Biophotonics Int. 2004;11(7):36–42. Lievens A, Jacchia S, Kagkil D, Savini C, Querci M. Measuring digital PCR quality: performance parameters and their optimization. PLoS One. 2016;e0153317. Majumdar N, Banerjee S, Pallas M, Wessel T, Hegerich P. Poisson plus quantification for digital PCR systems. Sci Rep. 2017;7:9617. Huggett JF, Foy CA, Benes V, Emslie K, Garson JA, Haynes R, Hellemans J, Kubista M, Mueller RD, Nolan T, et al. The digital MIQE guidelines: Minimum information for publication of quantitative digital PCR experiments. Clin Chem. 2013;59(6):892–902. Holstein CA, Griffin M, Hong J, Sampson PD. Statistical Method for Determining and Comparing Limits of Detection of Bioassays. Anal Chem. 2015;87(19):9795–801.

TOC Figure

ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Paragon Plus Environment

Page 10 of 10

10