Development of 3D printed mesofluidic devices to elute and concentrate DNA

Abstract

To understand structural variation for personal genomics, an extensive ensemble of large DNA molecules will be required to span large structural variations. Nanocoding, a whole-genome analysis platform, can analyze large DNA molecules for the construction of physical restriction maps of entire genomes. However, handling of large DNA is difficult and a system is needed to concentrate large DNA molecules, while keeping the molecules intact. Insert technology was developed to protect large DNA molecules during routine cell lysis and molecular biology techniques. However, eluting and concentrating DNA molecules has been difficult in the past. Utilizing 3D printed mesofluidic device, a proof of principle system was developed to elute and concentrate lambda DNA molecules at the interface between a solution and a poly-acrylamide roadblock. The matrix allowed buffer solution to move through the pores in the matrix; however, it slowed down the progression of DNA in the matrix, since the molecules were so large and the pore size was small. Using fluorescence intensity of the insert, 84% of DNA was eluted from the insert and 45% of DNA was recovered in solution from the eluted DNA. DNA recovered was digested with a restriction enzyme to determine that the DNA molecules remained full length during the elution and concentration of DNA.

1. Introduction

Structural variation in the human genome accounts for a larger number of bases varied than single nucleotide polymorphisms [1, 2] and has been linked to various diseases [3–5]. These variations are at least 1 kb in size or greater [6], but larger structural variations (> 50 kb) are difficult to determine due to their large size compared to single nucleotide polymorphisms or small insertion and deletions. Physical mapping platforms (Optical Mapping [7–9] and Nanocoding [10–13]) are able to span these large structural variations (5 kb – hundreds of kilobases in size to megabase in size). However, in order to span extremely large insertions or inversions, molecules need to be long enough to span the variation with enough information on each side to allow for alignment.

In Nanocoding, fluorescently labelled DNA with sequence-specific information highlighted with fluorescently labelled nucleotides are driven into nanoslits with an electric field and then the electric field is turned off to form a DNA dumbbell. A dumbbell occurs when the molecule traverses the nanoslit, electric field is turned off, so both ends of the molecule reside in opposing microchannels [10, 12–14]. In order to analyze large genomes, a population of large DNA is required to analyze the genome and find large structural variations in the genome. In order to discern large structural variations, molecules that span the variation with enough unique information on either end are required to understand complex genomes, such as cancer. With Nanocoding’s unique method of presentation, which allows DNA to be uniformly stretched at specific locations within the device (nanoslits), large molecules are required with a minimum size to form a dumbbell in order to present a complete genomic scaffold to determine structural variations.

Large DNA molecules, needed to span large structural variations, are fragile and during routine molecular biology manipulations will break. Therefore, Schwartz et al. developed insert technology to protect DNA during cell lysis and other routine molecular biology manipulations when they developed pulsed-field gel electrophoresis [15, 16]. However, getting large DNA out of the insert and into solution is difficult due to the fragility of the molecules and Nanocoding requires a very low ionic strength solution to form fully stretched DNA dumbbells. Therefore, a system is required to elute and then concentrate large DNA without breaking them for Nanocoding or other sequencing platforms.

Amicon filters and NucleoSpin gDNA are able to readily concentrate small DNA and are the go-to method of concentrating PCR products, proteins, and plasmids, but are limited to smaller DNA sizes [17, 18]. The filters are optimized for PCR products and plasmids, which are smaller than the size required for Nanocoding. Other methods use magnetic particles to enrich DNA concentration (75, 100,140 bp), but would be difficult to scale up and implement this system for extremely large DNA molecules needed for Nanocoding or other sequencing platforms aiming for molecules that are hundreds of kilobase pairs [19].

We developed a proof of principle system to elute and concentrate DNA using a polyacrylamide “roadblock” to concentrate DNA at the gel-solution interface. We eluted 84% of the original DNA in the insert and concentrated and recovered 45% of the eluted DNA. We found DNA, when recovered from the elution-concentration device, was intact and full length.

2. Materials and methods

2.1. Fabrication of 3D printed meso-fluidic devices

3D printed mesofluidic devices were designed in AutoCAD and exported to Cura (Ultimaker software) to determine the print speed, fill-in, adhesion, etc. (Figure 1). Devices were then printed on an Ultimaker³2 or Ultimaker3 3D printer (Figure 1A) using polylactic acid (PLA) material. Device dimensions are located in SI Figure 1. After the print was completed, double-sided tape was placed on the bottom of the device, minus the channel area, and adhered to 2 × 3 in. glass slide. Caulk was applied to the peripheral edge of the device to seal the device to the glass slide.

Figure 1. Device Design and Loading Scheme.

The production of each device started with a design in AutoCAD. Files were then converted, using Cura software (A), and printed with an Ultimaker2 or Ultimaker3 printer (B). PLA devices were attached with double-sided tape to a glass slide and sealed with caulk around the edges to perform each experiment (C). Gel inserts with DNA were dyed with YOYO-1 and placed into the device (D and E). A given voltage (26.1-V) was applied to the device for a given period of time. Fluorescent images were taken throughout the experiment to capture the movement of the DNA.

2.2. Polymerization of polyacrylamide gels in PLA channels

A matrix was set up of different amounts of acrylamide/bis solution, TEMED, and APS to determine how quickly the polyacrylamide gel would polymerize in our PLA devices. (Information about materials is located in the SI Methods section 1.1.) A 1X acrylamide/bis and 1X TEMED/APS solution contained 111 μL 30% 29:1 or 19:1 acrylamide/bis solution, 889-μL H₂O, 7.5-μL 10% APS, and 0.8-μL TEMED. For each different concentration, the acrylamide/bis solution or TEMED/APS solution was increased. 1X was used as a starting point for the matrix, since it was utilized in Dimalanta et al. [9]. PLA channels were 3D printed using an Ultimaker2 or Ultimaker3 printer. The channels either had a PLA bottom or were open (SI Figure 2). The channels that were open were affixed to the glass slide using double-sided tape and caulk. Once the channels were completed, the acrylamide solution was prepared by adding acrylamide/bis and water to an Eppendorf tube and the solution was degassed for 15-min. Next, APS and TEMED were added to the solution, respectively, and the solution was vortexed for 15-sec. The solution was then pipetted into PLA channels and the gel was checked with a glass hook by poking and then lifting the gel up to determine if the polyacrylamide gel had polymerized (Figure 2).

Figure 2. Determination of polymerization time of acrylamide-bis gel in PLA devices.

A 3D printed polylactic acid (PLA) channel was printed and attached to a glass slide. (SI Figure 2B). 30% 19:1 Acrylamide-Bis was mixed with water, TEMED, and APS in an Eppendorf tube. 1X 19:1 acrylamide-bis solution and 1X TEMED/APS was added to the PLA device and the time it took for the solution to polymerize was measured by poking/lifting the gel to determine if the gel had polymerized. Depending on the curing time, a color was assigned for each time.

2.3. Preparation of polyacrylamide gel roadblock

A 4X acrylamide-bis solution and 2X TEMED/APS was utilized to make the roadblocks due to the curing time and pore size. To make a dam to hold the acrylamide gel, PDMS pieces were wedged into the channel to keep the acrylamide in a specific region. In order to make these pieces, PDMS was mixed with a 10:1 prepolymer to catalysis ratio, poured into an empty petri dish and cured at 60°C overnight. The pieces were cut from the petri dish to be slightly larger than the channel width. 1 – 2-μl of TEMED was placed on the bottom of the device and the PDMS pieces were wedged into the device to create a dam, so the acrylamide solution would not leak out. A 4X acrylamide-bis solution and 2X TEMED/APS solution was made and loaded into the concentration region to cure. The device was placed in a humidified box at 4°C overnight. The following day, excess polyacrylamide gel was removed from the device with a needle adaptor so only the specified area for the roadblock remained. Alternatively, earlier versions of the roadblock had the acrylamide solution poured into the whole device and then the gel was cut out with a needle. Both methods were used in this work.

2.4. Elution and concentration of DNA, imaging and analysis

A fluorescently stained DNA insert was placed in the loading region within the 3D printed device (SI Method 1.2). If the device had an acrylamide roadblock, that step would have been completed the previous day. The channel was then loaded with 1X TE buffer, placed on top of a blue light transilluminator, platinum electrodes were added to each well, and an orange filter was placed in between the device and the vertically mounted Canon EOS Rebel T3I camera. Once the device was set up, a voltage was applied (26.1 V) using a across the device to move DNA molecules from the insert to the roadblock. Images were taken at discrete time points. Finally, images were analyzed with ImageJ to determine the amount of DNA leaving an insert and concentrating at the roadblock [20].

3. Results and discussion

3.1. Developing 3D printed mesofluidic devices

In order to develop a mesofluidic device to elute and concentrate DNA, rapid prototyping and a device with a transparent bottom were required. Through many iterations, the most consistent method to create a device with a transparent, non-fluorescent bottom and fast prototyping was attaching a 3D printed device to a glass slide with double-sided tape, minus the channel region (Figure 1). The device was designed in AutoCAD and then loaded on Cura to convert it into a format for our 3D printers and printed.

First, the channel width of the device was varied to determine the optimal channel width, so the insert stayed in the original location and eluted the most DNA. The 3D device was assembled with a glass bottom and lambda DNA insert stained with YOYO-1 was placed into the loading region of a device and the device was filed with 1X TE. Platinum wires were placed into the reservoirs and attached to a power supply and 26.1-V was applied to electrokinetically migrate DNA through the insert into the channel (Figure 1 D and E). A variety of different voltages were applied and 26.1 V was the most ideal voltage for out experiments due to the amount of time needed to run an experiment, but did not require a cooling apparatus. A montage (images collected over 30-min. and placed sequentially in the same image) is seen in SI Figure 3 and 4. In SI Figure 4, the channel image is cropped and placed next to each image to show elution of DNA over the course of the experiment and to see, in greater detail, DNA eluted from the insert. Channel width was varied from 3.0 to 5.0 mm. For a width of 3.0 mm, DNA elution was minimal and a width of 5.0 mm caused the insert to slip during the course of the experiment, so it rotated and moved into the channel. The movement of the insert may be due to electroosmotic forces. A width of 3.5 mm allowed DNA to migrate into the channel, while the insert stayed in the original location, and was chosen for all future devices.

3.2. 3D printed devices to concentrate DNA

A variety of different devices were designed to concentrate DNA by a 3D printed concentration region or a matrix that decreased the movement of DNA. An insert with stained DNA was loaded into a device, as well as the gel matrix if it was used to concentrate DNA (Kelcogel, agarose, sodium alginate). Next, 1X TE was added, so the device was full and platinum electrodes were placed in the reservoirs. 26.1V was applied and images were taken over a period of time to determine the amount of DNA that was eluted from the insert and the amount of DNA that was concentrated. In SI Figure 1, the device dimensions for all devices are listed. In SI Figure 5 (graph, yellow circles), for each device, DNA eluted. In SI Figure 5, a funnel, perpendicular channel, triangle and bow tie were created to trap DNA. However, minimal amount of DNA was concentrated. The perpendicular channel (highlighted by the blue and green box) and triangle were able to trap DNA in the concentration region for several minutes, but eventually DNA migrated from the concentration region towards the electrode. Therefore, a concentration region required a physical barrier to inhibit or slow down the progression of DNA.

Next, a physical barrier was implemented to determine if a matrix could stop or slow down the progression of DNA in a channel. The following matrices were tried: 2% agarose, 0.5% agarose/0.25% sodium alginate, and 0.75% Kelcogel LT100/0.25% agarose. To impart a negative charge on the gel matrix, sodium alginate and Kelcogel LT100 were tried to induce an electroosmotic flow to concentrate DNA at a given area. The 2% agarose allowed DNA to move through the matrix without stopping (not pictured). Alginate and Kelcogel allowed DNA to concentrate for ~20-min. before DNA migrated through the matrix (SI Figure 6). Using these matrices, DNA concentrated for a brief period of time, but DNA eventually migrated through the matrix.

3.3. Polymerization of polyacrylamide in 3D printed devices

The pore size of agarose was too large to inhibit the migration of DNA [21], so we theorized that smaller pore sizes of polyacrylamide would stop the progression of DNA to create a roadblock. Dimalanta et al. used a polyacrylamide gel that was used as the basis for the original 1X solution [9]. However, we found the original 1X concentration of polyacrylamide and 1X TEMED/APS would not fully cure within a PLA device, so a matrix of different concentrations of acrylamide-bis solutions (29:1 or 19:1; acrylamide: bisacrylamide), TEMED and APS were varied to determine a set of solutions that would polymerize in our PLA devices. In order to determine how quickly the gel polymerized, each solution was added to a 3D printed channel (SI Figure 2 A and B) with either the bottom made out of PLA or glass. To test if the gel polymerized, the gel was poked with a sealed-end glass pipette at each time point (Figure 2 and SI Figure 2). As the APS/TEMED mixture increased, the polymerization time decreased. As the acrylamide bis-acrylamide mixture increased, the polymerization time decreased, but not to the same extent that changing the APS/TEMED mixture did. Formation of the polymer was initiated by APS, which formed free radicals and TEMED acted to stabilize free radicals and promoted polymerization. We choose 4X acrylamide-bis solution and 2X APS/TEMED, as the polymerization times allowed us to add the solution to the device without it curing too quickly.

Additionally, the qualitative effect of PLA was also tested to determine what role PLA had on the polymerization of acrylamide in channels with a PLA bottom versus glass bottom. PLA must slightly inhibit the formation of polyacrylamide at the surface of the plastic due to the longer polymerization times. Additionally, the original 1X acrylamide-bis and 1X APS/TEMED did not fully polymerize after 24 hrs. Further research would be needed to determine why PLA slightly inhibited the formation of polyacrylamide.

3.4. Creating an acrylamide gel roadblock

Utilizing a 4X acrylamide-bis and 2X APS/TEMED solution, a roadblock was created inside a 3D printed mesofluidic device by using PDMS dams that fit on either side of the concentration region to hold the acrylamide gel in that location, while it cured (Figure 3A). In order to prevent the acrylamide mixture from leaking out under the bottom of the PDMS, 1 – 2-μl of TEMED was added to the bottom, so the gel would polymerize quickly if it started to leak. Earlier versions of the acrylamide roadblock had the whole device full of the cured polyacrylamide gel and then sections were removed with a needle. Acrylamide solution was mixed and added between the PDMS pieces then cured overnight in a humidifier box at 4°C. A YOYO-1 stained DNA insert was loaded into the loading region (LR) and the rest of the channel was loaded with 1X TE buffer. Platinum wires were added to the reservoirs and connected to the power supply. A montage of time lapsed images of the loading region (LR), channel, (Ch), and roadblock (RB) are shown in Figure 3. At 0-min., DNA was in the loading region, but after 20-min., most of the DNA was concentrated at the roadblock region (dotted line). The pore size of the acrylamide is ~100–200-nm in size based on the %T and %C using data from Stellwagen et al.[21] DNA may have concentrated at the interface between the gel and the solution due to a decrease of DNA mobility between the gel and the solution. Also, DNA threading into the pore will take some time, further slowing down the progression of DNA. Some DNA embeds in the acrylamide gel, where other DNA might be partially in the gel with part of the DNA in solution. To allow for DNA that is partially inserted into the gel to unravel, the voltage is turned off for 20-min. to allow for DNA relaxation into solution before removing the solution. We found with agarose inserts, DNA migrated fairly quickly through the agarose gel matrix, but with acrylamide, DNA that entered the gel only migrated into the first few mm of the gel and stopped either through tangling around obstacles within the matrix or due to the amount of time we ran the experiment. The fluorescence intensity was plotted against time to show that as time increased, the fluorescence intensity of the loading region decreased due to DNA leaving the agarose insert and as time increased for the roadblock region, the fluorescence intensity increased as DNA concentrated at the roadblock region.

Figure 3. Time-lapse imaging of DNA eluted and concentrated with an acrylamide roadblock.

(A) An image of a 3D printed device attached to a glass slide with double-sided tape and caulk. (B) A 3D printed device was produced to hold a polyacrylamide gel matrix in the roadblock section (RB) (SI Figure 1B). Lambda DNA was embedded within a 0.5% agarose insert and placed into the device in the loading region (LR). The electrodes were placed in the wells (circles) and an electric field was applied (26.1-V) to concentrate DNA at the interface between the solution and gel. A montage of images was produced with ImageJ to illustrate the migration and concentration of lambda DNA on the 4X acrylamide roadblock in the 3D printed device. As illustrated, the acrylamide gel was able to concentrate most of the DNA at the acrylamide roadblock. (C) Cartoon image of the device and electrodes with an arrow pointing in the direction of DNA migration. DNA insert was placed in the loading region (LR) and DNA was eluted into the channel (Ch) and then concentrated at the roadblock (RB). (D) The intensity of DNA in an insert versus time. (E) The intensity of the DNA concentrated at the interface between the solution and roadblock over time.

In a separate set of experiments, 5 devices were run for 1 hr. to determine the elution, concentration, and recovery rate. To determine the amount of DNA eluted from an insert and concentrated at a roadblock, each insert had 2-μl of 585-ng/μl lambda DNA added (SI Figure 7) and was stained with YOYO-1 for a 1:4(dye to DNA bp) ratio, overnight. The DNA insert was placed into a device with a roadblock and immersed in 1X TE buffer and imaged every 15-min. The 1X TE was chosen, as it had one of the highest free solution mobilities of DNA (SI Methods 1.4, SI Figure 8), similar to what was found in Lallman et al. [22]. Once the electric field was turned off, DNA in the solution stayed and DNA that had one end in the gel and the other end in solution slowly relaxed into the solution due to entropy. The mobility difference and the size of the acrylamide pores slowed down DNA from entering into the gel. This difference allowed DNA to concentrate between the acrylamide and solution. The fluorescence intensity of DNA was compared at 0 and 60 min. to determine the amount of DNA eluted from the insert (86 % ± 7% or 1010 ± 80 ng) (Figure 4A). Given the data in Figure 4, the elution of DNA was reproducible with a relative standard deviation (RSD) of 8%. Additionally, the amount of DNA remaining in an insert after the completion of the experiment was determined by comparing the insert intensity to known amounts of DNA within an insert (84%±11% or 1000±100 ng). The amount of DNA concentrated at the roadblock was measured by fluorescence intensity (83%±11% or 970±130 ng) (Figure 4B) and compared to the original intensity at 0-min. to determine the amount at the roadblock. The amount recovered was measured with a Nanophotometer, and compared to the amount of eluted DNA also determined by fluorescence intensity, was 45±14% (RSD = 31%). The rest of the DNA was embedded in the acrylamide. Our recovery was better than Amicon filters in Garvin et al.; average recovery rate of Amicon filters at 800g for 6 samples was 38±11% (RSD = 28%) [17]. The relative standard deviation of our device is similar to the Amicon filter data published by Garvin et al [17] and our data was reproducible. However, there are several factors that can be optimized to improve the recovery of DNA for our concentration device such as concentration time and modifying the acrylamide gel. Once DNA is concentrated in our device, it could be displayed in nanoslit or on a surface for Nanocoding.

Figure 4. Amount of DNA that was eluted and concentrated.

(A) Over the course of 5-experiments, the amount of DNA eluted from the insert was measured using two different methods. (Black) The fluorescent intensity of DNA was measured prior to the voltage (0-min.) and at 60-min. and background was subtracted to determine the ratio of DNA left in the insert after the completion of the experiment. The ratio was converted to the amount of DNA eluted based on the initial fluorescent intensity at 0 min. (Gray) After the experiment was completed; the insert was saved and compared to a standard calibration curve of DNA inserts with a dynamic range of DNA. (B) (Diagonal Strip) For those 5 experiments, the amount of DNA that was concentrated at the interface between a solution and acrylamide gel was measured with fluorescence intensity and converted to the amount of DNA concentrated. (Blue) The amount of DNA recovered was measured with a Nanophotometer.

In order to determine if DNA was sheared during the course of the experiments, DNA was digested with HindIII, loaded, run on an agarose gel and imaged (SI Methods 1.3). In Lane-1, YOYO-1 DNA concentrated from the device and Lane-2 was a stock solution of YOYO-1 stained lambda DNA and both samples were digested with HindIII. If the recovered DNA were damaged during concentration, then the resulting bands in Figure 5 would have smeared. The bands from the control and the sample were identical indicating DNA was full length after concentration at the roadblock. The slight mobility difference between lane-1 and lane-2 was the amount of YOYO-1 intercalated. Our concentrated sample did not have as much YOYO-1, which is why the mobility of the bands was faster [23] than the control and why lane-2 was brighter.

Figure 5. Restriction digest of DNA concentrated at the acrylamide roadblock.

A restriction digest experiment with HindIII was performed to determine if the recovered lambda DNA was full length. Lane-1 contained the recovered fluorescently stained DNA from our 3D printed device and Lane-2 contained stock lambda DNA stained with YOYO-1 (Control) and both were digested with HindIII.

4. Concluding remarks

We developed a 3D printed mesofluidic device with a polyacrylamide roadblock to concentrate DNA electrokinetically at the interface between a polyacrylamide roadblock and solution. We were able to elute 84% of DNA from DNA inserts in 1 hr. and recovered 45% of the eluted full-length lambda DNA. DNA eluted from this device can be used for Nanocoding or other sequencing platforms that require extremely large DNA for genome analysis. As DNA size increases, DNA mobility inside the acrylamide should decrease thus improving our ability to concentrate DNA molecules in solution.

Abbreviations:

YOYO-1

1,1-[1,3-propanediylbis[(dimethyliminio)-3,1-propanediyl] bis[4-[(3-methyl 2(3H)benzoxazolylidene)methyl]-quinolinium iodide

TE

Tris EDTA

PLA

polylactic acid

TEMED

Tetramethylethylenediamine

APS

ammonium persulfate

RB

roadblock

LR

loading region

Ch

channel