Towards automatic gene synthesis with bioinformatics software, novel one step real time PCR assembly, and lab chip gene synthesis 2

In order to combine the simplicity and cost-effectiveness of the one-step process, with the assembly efficiency of the two-step process in the synthesis of relatively long genes, and to

CHAPTER V AUTOMATIC TOUCHDOWN ONE-STEP GENE SYNTHESIS

This chapter describes an analytical model of PCR gene synthesis based on the thermodynamics and kinetics of the assembly and amplification processes The kinetics difference between standard PCR amplification and one-step PCR gene synthesis has been analyzed using this model, and validated using real-time gene synthesis In addition, a cost-effective Automatic TouchDown (ATD) gene synthesis method is introduced that enables the synthesis of long DNA of up to 1.5 kbp with just one PCR process The ability of this ATD method has been demonstrated in the design and synthesis of human protein kinase B-2 (PKB2) (1446 bp) and the promoter of human calcium-binding protein A4 (S100A4) (752 bp) with oligonucleotides concentration of as low as 1

nM The Automatic TouchDown gene synthesis approach is co-developed by me and my colleague Dr Wai Chye Cheong I have initialized the one-step gene synthesis while Dr Wai Chye Cheong has further optimized this method systematically

The existence of various PCR gene synthesis methods suggests that there is lack of a standard or universal method [94] Depending on the complexity of target genes, the synthetic genes are often constructed with a one-step or two-step overlapping process The one-step process is preferred for short DNAs (< 500 bp), while two-step process is more suitable for long DNAs synthesis Different PCR conditions should be applied to optimize the assembly and amplification processes separately In order to combine the simplicity and cost-effectiveness of the one-step process, with the assembly efficiency of the two-step process in the synthesis of relatively long genes, and to develop a universal PCR based gene synthesis method, TopDown gene synthesis is introduced in Chapter VI This method minimizes the competition between PCA and PCR by differentiating the melting temperature of inner oligos and outer primers and running these two processes in one step with two different annealing temperatures However, this method has a limitation where the cycle

different lengths and sequence contents For relatively short DNA (< 500 bp), the emerge of length DNA occurred at around 6 to 11 cycles during PCA stage, and excess thermal cycles would result in the formation of non-specific long DNA with random sequence [103] While long DNA might need more than 15 cycles Insufficient thermal cycles would lead to the production of various truncate DNA instead of full- length DNA In order to improve the TopDown one-step PCR gene synthesis and develop a more universal method, a novel approach – automatic TouchDown one-step PCR is developed

full-5.1.1 Principle of Automatic TouchDown one-step gene synthesis

Similar to TopDown PCR, automatic TouchDown (ATD) gene synthesis combines the simplicity and cost-effectiveness of the one-step process, with the assembly efficiency of the two-step process in the synthesis of relatively long genes This method utilizes a software program (TmPrime) to design primers with two distinct melting temperatures not only to minimize the competition between PCA and PCR amplification in the one-step gene synthesis, but also to maximize the emerging full-length amplification

Figure 5.1: Schematic illustration of Automatic TouchDown (ATD) one-step gene synthesis

combining PCR assembly and amplification into a single stage The melting temperatures of inner oligonucleotides (Tmo) and outer primers (Tp1 and Tp2) are designed with the conditions

of Tp2 ≥ 72°C and Tmo - Tp1 ≥ 5°C to minimize potential assembly-amplification interference and maximize the full-length amplification during PCR

Figure 5.1 shows the concept of the proposed ATD one-step gene assembly method The outer primers are designed with two melting temperatures (Tp1 and Tp2) where Tp1 is lower than the melting temperature of assembly oligonucleotides (Tmo), and Tp2 is ≥ 72°C The overlapping gene synthesis is conducted in one PCR mixture with annealing temperature matched to Tmo The outer primers are subjected to an elevated annealing condition (Tmo – Tp1 ≥ 5°C) during assembly, which prevents mis-pairing among primers and oligonucleotides When the full-length template emerges, the outer primers would initially create full-length DNA with flanked tails, causing the melting temperature of outer primer-flanked template to shift to elevated Tp2 (≥ 72°C) This cascade of reactions enhances the annealing possibility of outer primer with flanked template, and boosts the corresponding amplification of full-length template This approach provides a unique benefit with its automatic switch in favor of full-length amplification as the reaction process progressed This key feature is demonstrated in synthesizing a relatively long gene (1446 bp) with single PCR from a pool of 62 oligonucleotides of 1 nM

5.1.2 Mechanisms of PCR synthesis process

In order to further understand the mechanisms of PCR synthesis process, the underlining kinetics

of PCR gene synthesis and standard PCR amplification are investigated in this study Typically, a standard PCR mixture contains only two different types of DNAs with excess outer primers (~0.4 µM) and a very small quantity of template (< 106 copies) [88] The outer primer-template annealing temperature remains unchanged during the reaction process, which simplfies the conditions of mathematical modeling Nevertheless, in gene synthesis reaction, the synthesis pool could consist

of 20–60 distinct oligonucleoitdes of 10 nM with an excess of outer primers (~0.4 µM) As the synthesis process progressed, various intermediate DNAs would be generated from the original short oligonucleotides, which would complicate the annealing process of the mixture Both extendable and unextendable pairings could occur These factors would dramatically complicate the analysis modeling From a molecular stand point, oligonucleotides in assembly mixture would require more time to find their complementary DNAs in a complex mixture, which would be reflected in the half-time constant of the hybridization reaction [110, 111]

In this study ATD one-step gene synthesis and real-time PCR are combined together to investigate the gene synthesis process for the promoter of human calcium-binding protein A4 (S100A4; 752 bp) and protein kinase B-2 (PKB2; 1446 bp) Gel electrophoresis results are compared with the real-time fluorescence signals to study the effects of the oligonucleotide concentration, stringency of annealing temperature, duration of PCR extension (which would reflect the half-time constant of the hybridizaiton reaction), and the dNTPs concentration

5.2.1 Design of oligonucleotides for gene synthesis

Gene sequence for the promoter of human calcium-binding protein A4 (S100A4, 752 bp; chr1:1503312036-1503311284) (6) and E coli codon-optimized human protein kinase B-2 (PKB2,

1446 bp) [26] are selected for synthesis via assembly PCR Oligonucleotides are derived by the TmPrime program (prime.ibn.a-star.edu.sg) Two sets of oligonucleotides (SA100A4-1 and S100A4-2) with different melting temperature uniformities (∆Tm: 2.3°C and 9.1°C) were designed

to investigate the effect of melting temperature on the assembly efficiency The oligonucleotide sets designed for the selected genes are summarized in Table 5.1, and their detailed information is provided in Appendix III Table S1–S3

Table 5.1: Data of oligonucleotide set

Overlap length (nt)

Oligo length (nt)

5.2.2 Automatic TouchDown one-step real-time gene synthesis

Automatic TouchDown (ATD) one-step process was optimized using real-time PCR conducted with Roche’s LightCycler 1.5 real-time thermal cycling machine with a temperature transition of 20°C/s Real-time gene synthesis was conducted with 20 µl of reaction mixture containing 1× PCR buffer (Novagen), 2× LCGreen I (Idaho Technology Inc.), 4 mM of MgSO4, 1 mM each of

dNTP (Stratagene), 500 µg/ml of bovine serum albumin (BSA), 1–40 nM of oligonucleotides,

400 nM of forward and reverse primers, and 1 U of KOD Hot Start (Novagen) The PCRs were conducted with: 2 min of initial denaturation at 95°C; 30 cycles of 95°C for 5 s, 58–70°C for 30 s, 72°C for 90 s; and final extension at 72°C for 10 min Desalted oligonucleotides were purchased from Sigma-Aldrich without additional purification The outer primers are summarized in Table 5.2 with predicted melting temperatures calculated using IDT SciTools [85] according to the assembly buffer condition

5.2.3 Gel electrophoresis

The synthesized products were analyzed by 1.5% agarose gel (NuSieve® GTG®, Cambrex Corporation), stained with ethidium bromide (Bio-Rad Laboratories) or SYBR Green (Invitrogen), and visualized using Typhoon 9410 variable imager (Amersham Biosciences) Gel electrophoreses were performed at 100 V for 45 min with 100 bp ladder (New England) and 5 μL

of DNA samples

Table 5.2: Summary of primers for conventional one-step, and ATD one-step gene syntheses

All PCR assemblies are performed with an annealing temperature of 70°C

5.3 Theoretical analysis of DNA hybridization kinetics

The assembly efficiency of PCR and LCR gene synthesis relies on the effectiveness of hybridization reaction of assembly oligonucleotides at the annealing temperature The hybridization effectiveness, expressed as the half-time constant of the hybridization reaction of a single-stranded DNA (ssDNA) in a mixture, is a function of the number of unique oligonucleotides and the oligonucleotide concentration [110]

The DNA hybridization reaction starts when the portion of two complementary ssDNA strands collides and forms a nucleation site; then the rest of the sequence rapidly zippers to form a dsDNA It has been shown that the nucleation step is the reaction limitation, and the hybridization reaction rate constant of a ssDNA in a mixture is given by [110]:

N

L k

k= N S

where L S is the length of the shorter strand participated, k Nis a nucleation rate constant, and N is the complexity of the mixture, which is the number of unique oligonucleotide in the gene assembly mixture, or the primer length for standard PCR amplification

For standard PCR amplification whereby the mixture contains only excess primers and template DNA, the hybridization reaction can be described by a pseudo-first order reaction with a half-time constant of:

o

kC

t1/2 = ln2

5.2 where C o is the total nucleotide concentration Under the typical PCR amplification conditions (k N≈ 5 ×104 /M⋅s) with a primer of 20 base long (L S = N = 20) and a primer concentration (C) of

1 µM (C o = C × N), the annealing half-time is ~ 3 s

For gene assembly where the DNA is constructed from a pool of oligonucleotides with equal concentration, the hybridization reactions can be described by second-order kinetics with a half-time constant of:

For overlapping PCR assembly, the average DNA length is getting longer with each PCR cycle, while the total number of strands does not change As the reaction proceeds, various intermediate DNAs are generated from the original short oligonucleotides Hence, the complexity (N) and <Ls> will increase while concentration of each DNA fragment (C) will gradually decrease Both extendable and unextendable pairings could occur Duplex annealed in the 3’ recessed configuration can be extended, while dsDNA annealed with 3’ ends protruded will not

be extended Unlike the exponential nature of PCR amplification, the average DNA length is most likely to increase linearly while the complexity (N) may increase more rapidly as intermediate DNAs are generated The unextendable annealing could further complicate the assembly Accounting for these factors, the half-time constant may increase as reaction proceeds

The Lightcycler has an ultrafast temperature transition (20°C/s) For a typical thermocycler, the ramp rate is normally ≤ 4°C/s (DNA Engine PTC-200, Bio-Rad) With this thermocycler, the ramp time from 95°C to 60°C (annealing temperature) can take ~ 8.75 s, which would be sufficient for the annealing reaction to be completed in normal PCR amplification In addition, KOD polymerase has a very fast elongation rate (~ 120 bases/s) [112] The required extension time is shorter than 10s for 1 kbp extension, which roots out the potential reaction limitation contributed by polymerase enzyme

In summary, it is important to realize that the complexity of the assembly mixture will increase the half-life in gene assembly The outer primer and assembly oligonucleotide have

different annealing half-times that depend on their concentrations Reducing the oligonucleotide concentration may only slightly affect its melting temperature, but it can profoundly affect the annealing kinetics The same derivation may be applied to the ligase chain reaction (LCR) gene synthesis, which has similar underlying annealing reaction

For normal PCR amplifications, their half-time constant could be as short as a few seconds, dependent on the outer primer concentration However, this constant can be significantly increased to hundreds to thousands of seconds due to the low oligonucleotide concentration (usually 10–40 nM), and the complex assembly mixture containing several tens of oligonucleotides

5.4.1 Effect of varying extension time during ATD one-step gene synthesis

The key mechanism of reaction half-time was demonstrated by synthesizing the S100A4 (752 bp) and PKB2 (1446 bp) using a rapid thermal cycler with a temperature transition of 20°C/s ATD one-step gene synthesis was performed using the empirically optimized real-time gene synthesis protocol (Chapter IV), with either 20 s or 120 s of combined annealing (70°C) and extension (72°C), 2× LCGreen I, 4 mM dNTPs, and 4 mM Mg2+ ion Results clearly indicated that insufficient hybridization (20-s reaction) could cause the assembly efficiency to degrade, resulting

in incomplete products with DNA length of ~ 200–300 bp (Figure 5.2)

The effect of reaction time was further studied by varying the extension time from 30 s to

120 s for S100A4, assembled with 10 nM and 1 nM oligonucleotide, respectively For assembly with 10 nM oligonucleotide, the reaction time was less critical Fairly high assembly efficiency was observed where the fluorescence intensity increased as the assembly process progressed (Figures 5.3a,c) The normal 30-s extension was sufficient to generate the full-length products, whereas prolonged extension (≥ 90 s) promoted the reaction so that the assembly process reached the plateau faster (in ~ 25 cycles) In contrast, the assembly from 1 nM oligonucleotide has very low assembly efficiency (Figures 5.3b,d), with a fluorescence curve like the single molecular

DNA amplification [88] The gel results clearly indicated that prolonged hybridization (≥ 90 s) was essential for ssDNA to be effectively annealed at such a low oligonucleotide concentration

(d)

Figure 5.2: Effect of hybridization reaction time Top: Agarose gel results of (a) S100A4-1, (b)

S100A4-2, and (c) PKB2 synthesized with: (1) 10-s annealing (70°C) plus 10-s extension (72°C), and (2) 30-s annealing (70°C) plus 90-s extension (72°C) Bottom: The corresponding fluorescent curves for S100A4-1 (□: 20 s, ■: 120 s), S100A4-2 (Δ: 20 s, ▲: 120 s), and PKB2 (○: 20 s, ●: 120 s) The concentrations of oligonucleotides and outer primers are 10 nM and 400

nM, respectively

5.4.2 Effect of varying initial oligonucleotides concentration

The gene synthesis took place in several phases, as revealed by the variation in slopes with the number of PCR cycles (Figure 5.4) The overlapping assembly was a parallel process (see Appendix IV for the derivation) Theoretically, 5 PCR cycles would be sufficient for assembling S100A4 (752 bp) from a pool of 32 oligonucleotides Hence, relatively few PCR cycles were needed to create a full-length dsDNA This was clearly indicated by the slope change in the fluorescent curve in the early cycles (< 10 cycles) The slope became steeper as the full-length

template emerged and became amplified, taking advantage of the exponential nature of PCR amplification This phenomenon was remarkable with an oligonucleotide concentration of 5–20

nM No obvious full-length gene product was obtained with 1 nM oligonucleotide within 30 PCR cycles, since the amplification stage was delayed due to its low assembly efficiency

Figure 5.3: The synthesis yield is dependent on the extension time S100A4-2 (752 bp) is

synthesized with various extension time from 30 s to 120 s at an annealing temperature of 70°C (30 s) with oligonucleotide concentration of (a,c) 10 nM and (b,d) 1 nM (a, b) Fluorescence as

a function of extension time of 30 s (◊), 60 s (▲), 90 s (♦), and 120 s (□) (c, d) The corresponding agarose gel electrophoresis results The synthesis from 10 nM oligonucleotides reaches the plateau within 30 cycles, while the reaction from 1 nM oligonucleotides only enters the amplification phase after 30 cycles

For gene synthesis with ≥ 20 nM of oligonucleotides, the PCR process reached the plateau within 15–20 cycles Additional cycles would favor non-specific PCR, and lead to the build up of high molecular weight products (7,10-12) and the generation of spurious bands as shown in Figure 5.4b (indicated by the arrow) The gel results and real-time PCR curves

suggested that the optimal oligonucleotide concentration was 5–15 nM for ATD gene synthesis, which coincided with that of the conventional one-step (13,14), TopDown one-step [89] and two-step [93] processes

(a)

(b)

Figure 5.4: The effect of oligonucleotide concentration on the successful gene synthesis

S100A4-2 (752 bp) is synthesized with various oligonucleotide concentrations ranging from 1

nM to 40 nM All PCR are conducted with 30-s annealing at 70°C and 90-s extension at 72°C (a) Fluorescence as a function of PCR cycle number for oligonucleotide concentrations of 1 nM (□), 5 nM (∆), 10 nM (▲), 15 nM (○), 20 nM (●), and 40 nM (◊) The change in the slopes of fluorescence increment indicates the emergence of full-length template (b) The corresponding agarose gel electrophoresis results The arrow indicates the undesired DNA with 2× length of full-length template, generated from non-specified full-length amplification of excess PCR

5.4.3 Effect of varying annealing temperature

The effect of varying the annealing temperature (from 58°C to 70°C) was further investigated also (Figure 5.5) The fluorescence intensity curves were indiscriminant to the annealing temperatures during the assembly phase (first 10 cycles), and began to deviate presumably only after the full-length template emerged Interestingly, a higher yield of the desired DNA was obtained with a stringent annealing temperature (70°C) higher than the average Tm of oligonucleotides (66°C); this was consistent with the TopDown one-step process (Chapter IV) Performing gene synthesis

at stringent annealing temperature would increase the specialization of oligonucleotide hybridization, and minimize the potential mishybridization that might occur during the gene synthesis process (see Appendix IV Tables S4 and S5 of the potential hybridization for S100A4 and PKB2)

(a) (b)

Figure 5.5: (a,c) S100A4-2 (752 bp) and (b,d) PKB2 (1446 bp) synthesized with various

annealing temperatures ranging from 58°C to 70°C (30 s) and 90-s extension at 72°C (a,b) Fluorescence as a function of PCR cycle number for annealing temperatures of 58°C (◊), 60°C (∆), 62°C (□), 65°C (♦), 67°C (○), and 70°C (▲) (c,d) The corresponding agarose gel electrophoresis results Higher synthesis yield is obtained with a stringent assembly annealing temperature (70°C) The slope changes in fluorescence intensity indicate the automatic switch feature in the assembly and amplification processes

5.4.4 Synthesis of long gene by ATD process

The applicability of the ATD one-step process was demonstrated by synthesizing the relatively long gene, PKB2 (1446 bp), which could not be achieved by the conventional one-step gene synthesis [26] Surprisingly, the PKB2 has higher assembly efficiency than that of S100A4, even although the PKB2 is ~ 2× longer than S100A4 The fluorescent signal indicated the S100A4 and PKB2 syntheses reached the plateau at ~ 28 and ~ 22 cycles, respectively Indeed, the ATD one-step process has fairly high assembly efficiency for oligonucleotide concentrations of ≥ 10 nM Relatively few PCR cycles (~ 10 cycles) were needed to create a full-length dsDNA, as suggested

by the slope changes in fluorescent intensity in Figures 5.5a,b This discovery matched well with

the theoretically derivation (see Appendix IV), which predicted that 5 and 6 PCR cycles were sufficient for assembling S100A4 (752 bp) and PKB2 (1446 bp) from a pool of 32 and 62 oligonucleotides, respectively

5.4.5 Effect of varying dNTP concentration

In the one-step gene synthesis process, the dNTPs could deplete and cease the PCR reaction [85, 95]due to the assembly-amplification interference, and the generation of a large portion of intermediate DNA products This dNTPs depletion was critical for DNA with high GC content or length (7,10) Therefore, to determine the dNTPs effects, we have used the optimized synthesis condition determined in previous experiments, and conducted the gene synthesis with dNTPs of 4

mM (4 mM Mg2+) and 0.8 mM (1.5 mM Mg2+) with Mg2+ ion (MgSO4) concentration adjusted to compensate the dNTPs–Mg2+ chelation, which would affect the polymerase activity (18,19)

Successful gene syntheses were achieved in both of conventional step and ATD step gene synthesis for all three genes, except the case of PKB2 with 0.8 mM dNTPs (see Figure 5.6) The dNTPs concentration became more critical for relative long PKB2 where more intermediate products could be generated The gel results and fluorescence curves (see Figure 5.7) indicated that the conventional one-step process has comparable assembly efficiency with the ATD one-step for S100A4 synthesized with the optimized conditions No obvious difference was observed for relatively short S100A4 assembled with 4 mM or 0.8 mM dNTPs in both gel results and fluorescence curves To make the ATD a universal synthesis method for various gene lengths,

one-4 mM dNTPs should be used

5.4.6 Effect of melting temperature uniformity of partitioned oligonucleotides

Another factor that could affect the assembly efficiency was melting temperature uniformity of assembly oligonucleotides Two oligonucleotide sets, S100A4-1 (∆Tm = 9.1°C) and S100A4-2 (∆Tm = 2.03°C), with different Tm uniformity were synthesized with 10 nM and 1 nM oligonucleotide (Figure 5.8) Indeed, S100A4-2 has a higher assembly efficiency than the S100A4-1 It reached the plateau within 28 cycles, whereas S100A4-1 was still in the

amplification stage after 28 cycles (see Figure 5.7a, b) However, for synthesis with ultralow oligonucleotide (1 nM), the Tm uniformity requirement became more essential Only the assembly from S100A4-2 with highly uniform Tm was success With this finding, successful gene synthesis was demonstrated for PKB2 (∆Tm = 1.9°C) with 1 nM oligonucleotide The results suggested that the uniformity of melting temperature would be critical for ultralow oligonucleotide assembly, which has very low assembly efficiency To my knowledge, this is the first time that the successful gene synthesis has been achieved with an ultralow concentration of oligonucleotides of

1 nM

(c)

Figure 5.6: Agarose gel electrophoresis results of conventional 1-step and ATD one-step

(30-cycle) gene syntheses with dNTPs concentrations of 4 mM and 0.8 mM for (a) S100A4-1 (752 bp), (b) S100A4-2 (752 bp) and (c) PKB2 (1446 bp) All PCRs are conducted with 30-s annealing at 70°C and 90-s extension at 72°C The concentrations of oligonucleotides and outer primers are 10 nM and 400 nM, respectively

Figure 5.8: Agarose gel electrophoresis results of S100A4-1 (lanes 1 and 3) and S100A4-2

(lanes 2 and 4) with oligonucleotide concentrations of 10 nM and 1 nM, and PKB2 (lane 5) with 1 nM oligonucleotides The arrow indicates the full-length DNA Syntheses are performed with 30 and 36 cycles, respectively, for 10 nM and 1 nM oligonucleotides, with 30-s annealing

at 70°C and 90-s extension at 72°C

5.5 Discussion

In this chapter, the ATD gene synthesis method have been presented, which offered a simple, rapid and low-cost approach for synthesizing long DNA (1446 bp) with only one PCR step and concentration of oligonucleotides as low as 1 nM The experiments have demonstrated that ATD one-step gene synthesis was fairly efficient The assembly process automatically switched to preferential full-length amplification as the full-length template emerged The ATD process improved the previously discussed TopDown process (Chapter 4) by having the PCR amplification tailored to follow the emergence of full-length DNA to avoid excess PCR

It was found that the quality and quantity of PCR-based gene synthesis were influenced

by several factors, including annealing time, annealing temperature, concentration of oligonucleotides, concentration of dNTPs monomers, and number of PCR cycles The hybridization mechanisms of normal PCR amplification and PCR gene synthesis were demonstrated to be different by using a rapid thermocycler Prolonged annealing (≥ 90 s) was essential for the assembly of ultralow concentration of oligonucleotides (≤ 1 nM), especially for long gene synthesis The annealing duration was less critical for commonly reported gene synthesis with a DNA length of ≤ 500 bp and 10 nM oligonucleotides In addition, the typical thermocycler has a slow ramp rate of ≤ 4°C/s (DNA Engine PTC-200), which could contribute additional annealing time for temperature ramping from 95°C to 60°C With the help of the established model, insights into the optimization of gene synthesis conditions were attained It was expected that the minimum concentration of oligonucleotides could be further reduced to 0.1

nM, which would facilitate gene synthesis using the oligonucleotides from DNA microarray (20,21)

The fluorescence signals indicated that an oligonucleotide concentration of 5–15 nM provided optimal assembly efficiency with a high quantity and quality of full-length products The number of PCR cycle might have to be optimized according to sequence content and the oligonucleotide concentration to minimize the formation of abnormal products generated by excess PCR cycle (see Figure 5.4) The abnormal products with incorrect DNA sequences would

potentially complicate the enzymatic cleavage or the consensus shuffling error correction process (22-24) Predicting the optimal PCR cycle number would be difficult, as it could rely on several factors including the complexity and length of DNA sequence, oligonucleotide concentration, annealing temperature, and Tm uniformity The real-time gene synthesis with fluorescence monitoring described herein would help by providing instant feedback, terminating the process in time as it reached the plateau

The experiment data showed that performing the assembly with an annealing temperature slightly higher than the average Tm of oligonucleotides would increase the specialization of oligonucleotides hybridization as in Touchdown PCR [106], and reduce the possibility of potential mis-pairing among oligonucleotides, preventing the generation of faulty sequence The experiemnt data also suggested that the dNTPs could be depleted for relatively long genes (≥ 1.5 kbp), and 4 mM dNTPs should be used for universal gene synthesis The melting temperature uniformity of assembly oligonucleotides turned out to be critical for the assembly of ultralow concentration of oligonucleotides Therefore, it would be desirable to design the oligonucleotide sets using a bioinformatic program such as TmPrime

CHAPTER VI INTEGRATED TWO-STEP GENE SYNTHESIS ON CHIP

The final goal of this project is to develop a miniaturized automatic gene synthesis system In Chapter III, IV and V, bioinformatics software, TopDown and Automatic TouchDown one-step gene synthesis methods are presented respectively The development of these methods is aimed to provide a most suitable gene synthesis protocol for the integrated gene synthesis In this chapter,

an integrated microfluidic device is presented, which is capable of performing two-step gene synthesis to assemble a pool of oligonucleotides into genes with the desired coding sequence

The synthesis method used here is PCR-based overlapping DNA assembly method Figure 6.1 shows its concept of creating a synthetic gene A pool of short oligonucleotides is first assembled into long double-stranded DNA (called template) with the desired length and sequence information using the polymerase cycling assembly (PCA) [22].The quantity of the assembled template DNA is then amplified by the PCR step The Polymerase cycling assembly utilizes the DNA polymerase to extend the oligonucleotides During the thermal cycling, the mixed oligonucleotides are hybridized at their overlap regions where their 3’ ends are extended to generate longer double-stranded products This process is repeated until the full-length gene is obtained Synthesis via PCR can be performed either as a one-step process, combining assembly PCR and amplification PCR into a single stage, or as a two-step process with separate stages for assembly and amplification

Herein an integrated microfluidic device integrating PCRs with in situ hydrogel valves and micromixer for constructing short oligonucleotides into a long DNA sequence is reported The baseline protocols on the concentrations of oligonucleotide and primer were first developed

in PCR tube, and then applied to microfluidic syntheses Synthesis in microfluidic environments was successfully demonstrated in constructing a 760 bp GFPuv gene segment from a pool of 39 oligonucleotides using both the one-step and two-step synthesis processes The accuracies of

microfludic gene synthesis were determined by DNA sequencing, and compared along with control experiments performed in standard PCR tubes within a commercial thermal cycler

Along with the integrated gene syntheis chip, a microfluidic design to purify the synthesis product and prepare buffer solution for downstream application is described Silica-coated magnetic beads were employed for the solid-phase PCR purification The DNA extraction efficiency was tested A short heat shock was applied to enhance the extraction efficiency

Figure 6.1: Schematic illustration of PCR-based gene synthesis One-step synthesis combines

PCA and PCR amplification into a single stage The two-step synthesis is performed with separate stages for assembly and amplification

6.2.1 Microfluidic device fabrication

The fabrication process flow of the microfluidic device is shown in Figures 6.2a and b Instead of using SU-8-based lithography process to create the polydimethylsiloxane (PDMS) casting mold, a three-dimensional (3D) rapid prototype method that printed 3D structure using photosensitive resin was adopted [113] The 3D structure was designed in SolidWorks and transferred to the Eden

350 (Objet Geometries), which printed photopolymer material (FullCure 720) and support material (FullCure 705) layer by layer The photopolymer layer was cured by UV light immediately after it was printed Upon completion, the fabricated structure was soaked in 25% tetramethylammonium hydroxide (TMAH) solution for 3 h to remove the support material designed for supporting the printed geometries The microfluidic mold was soaked in water for 1

h to wash away TMAH This method provided a resolution of 42 µm in the x-axis and y-axis, and

a resolution of 16 µm in the z axis, well-suited for generating thick and multilevel structures without lithographic process Conversely, other rapid prototype methods such as liquid phase photopolymerization [114] and contact liquid photolithographic polymerization [115] utilized photomasks to facilitate construction of structures with superior resoultion As shown in Figure 6.2a, two-level mold was designed with different heights for connection channels (height: 0.2 mm; width: 0.2 mm) and chambers (height: 0.5 mm) to minimize the dead volume of connection channels

Poly(dimethylsiloxane) (PDMS) precursor was prepared by mixing Sylgard 184 base and Sylgard 184 curing agent in a 10:1 volume ratio The precursor was poured into the mold, degassed in vacuum chamber for 30 min, and cured in a convection oven at 75°C for 3 h The 3-

mm thick PDMS slab was then peeled off from the mold, and connection holes were pierced The microfluidic device was assembled by bonding PDMS and silicon substrate (500 µm-thick) Both the PDMS and silicon substrate were treated with electrical discharges (Model BD-10AV, Electro-Technic Products) for 30 s, and the two surfaces were brought together immediately after treatment [116] Finally, the device was cured in an oven at 75 °C for 2 h to ensure irreversible bonding between PDMS and the silicon substrate

To prevent sample evaporation, the bonded device was deposited with a 2 μm-thick Parylene C using the PDS 2010 Parylene Deposition System (SCS, USA) The vapor deposited Parylene C created a barrier to control the water vapor diffusion Parylene also passivated the inner surface of the device in preventing unwanted protein absorption [117, 118]

6.2.2 Preparation of hydrogel valves

Thermosensitive hydrogel valves were selected for fluidic regulation and confining PCR reaction Hydrogel was synthesized following the method suggested by H J van der Linden et al [119] Temperature-sensitive monomer N-isopropylacrylamide (NIPAAM, 286 mg), N,N’-methylene bisacrylamide (BIS, 7.88 mg) crosslinker and 2,2’-dimethoxy-2-phenyl acetophenone (DMPAP, 18.86 mg) photoinitiator were mixed in 500 μL of dimethylsulfoxide (DMSO), generating a

precursor solution containing 2% BIS crosslinker The precursor was purged with nitrogen to remove oxygen, and wrapped with aluminum foil to avoid unwanted photopolymerization All the chemicals were purchased from Sigma-Aldrich (Singapore)

As illustrated in Figure 6.2b, the mixture was then injected into the fabricated chip using

a 1-ml syringe through the access holes, and photopolymerized in situ at 32°C with a chromium mask defining the exposed area The sample was then irradiated at a wavelength of 365 nm (dose:

252 mJ/cm2) using OmniCure Series 2000 UV illumination system (EXFO, Canada) After ultraviolet exposure, the device was placed on a hotplate at 60°C to keep hydrogel valves open, and the unpolymerized precursor was removed with de-ionized water at a flow rate of 500 μL/ min for 40 min using a syringe pump (74900 Series, Cole-Parmer Instrument Company) Finally, the device was baked at 75°C for 3 h in an oven to dry its inner surface and hydrogel valves

The NIPAAm-based hydrogel is thermosensitive with a lower critical solution temperature (LCST) of 32°C [119] The hydrogel would swell at temperatures below 32°C, blocking the fluidic channel At a temperature above 32°C, the polymer chains become hydrophobic, causing the hydrogel to shrink and allowing fluid flowing through The opening and closing of valves are controlled by varying the temperature between 4°C and 60°C Figure 6.2c shows the fabricated two-step gene synthesis chip with solid-phase PCR purification module (65 mm × 50 mm)

(a) (b)

(c)

Figure 6.2: (a) Fabrication process of microfluidic chip (b) Fabrication process of hydrogel

valve The PCR reactions and hydrogel valves are controlled by two separate thermoelectric heaters (TE 1 and TE 2) The insertion shows a closed hydrogel valve (c) Photograph of a two-step gene synthesis chip with solid-phase PCR purification (65 mm × 50 mm)

6.2.3 PCR thermal cycling

The PCR was performed by using a home-made thermal cycler, which included a fan, a thermoelectric (TE) heater/cooler (9501/127/030, FerroTec) and a thermoelectric control kit (FerroTec, USA) consisting of FTA600 H-bridge amplifier, FTC100 temperature controller and FTC control software The thermoelectric heater was powered by the FTA600 amplifier, which was controlled by the FTC100 temperature controller A T-type thermocouple (5TC-TT-T-40-36, OMEGA Engineering) was mounted on the TE heater to measure the temperature, and used as a feedback to the FTC100 temperature controller The temperature difference between the thermoelectric heater and actual temperature inside the PCR chamber was calibrated using a calibration chip, which has identical dimensions as the actual device but a thermocouple embedded inside the PCR chamber filled with PCR mixture The temperature drop between the heater surface and inside the chamber was noted in the FTC control software and compensated during the operation The desired temperature profile was programmed into a computer through the FTC control software, which controlled the FTC100 temperature controller using a PID (proportional-integrative-derivative) algorithm to optimize the temperature response time

6.3.1 Gene assembly and amplification

Published sequence of GFPuv gene segment with a total length of 760 bp (sequence 261-1020 with T357C, T811A and C812G base substitutions) was selected for the synthesis experiment The gene segment was assembled using 37 of 40-mer and 2 of 20-mer oligonucleotides with 20

bp overlap [86] (Appendix I Table S2) The PCR synthesis reactions were performed both within the microfluidic devices and in the standard 0.2-ml PCR tubes with a commercial thermal cycler (DNA Engine PTC-200, Bio-Rad) for comparison of the synthesis performance Synthesis via PCR was performed either as a one-step process, combining assembly PCR and amplification PCR into a single stage, or as a two-step process with separate stages for assembly and amplification The one-step process in PCR tube was conducted with 50 µl of reaction mixture

including 1× PCR buffer (Novagen), 1 mM of MgSO4, 0.25 µM each of dNTP (Stratagene), 5-25

nM of oligonucleotides, 0.1–0.4 µM of forward and reverse primers, and 1 U KOD Hot Start (Novagen) The PCR was performed under the following conditions: 2 min of initial denaturation

at 95 °C; 30 cycles of 95 °C for 30 sec, annealing at 50 °C for 30 s, 72 °C for 30 sec, and last extension at 72 °C for 10 min The PCR protocol of the two-step process was essentially the same

as that for the one-step process For PCR assembly, 5–25 nM of oligonucleotides were used without the forward and reversed primers For gene amplification, the assembled product was 2× diluted with fresh amplification reaction mixture containing a final primers concentration of 0.4

µM Microfluidic syntheses were conducted with the same PCR conditions in adjusted chamber volume All processes were performed with desalted oligonucleotides from Research Biolabs (Singapore) without additional purification

6.3.2 Solid-phase buffer exchange

Solid-phase buffer exchange was conducted using the magnetic beads based PCR purification method (ChargeSwitch PCR clean-up kit, Invitrogen) on microfluidic devices, and in standard 0.2-mL PCR tubes (as control) with the synthesized PCR product and 100 bp DNA ladder (New England, 170 ng/ μL) as control

For the control experiment performed in PCR tube, the 100 bp DNA ladder or synthesized product (7 μL) was mixed with 5 μL of beads and 11 μL of purification buffer (Invitrogen), and incubated for 1 min The beads were then captured by a magnet to remove the supernatant We washed the beads with 150 μL of washing buffer (Invitrogen) three times, and loaded 7 μL of elution buffer (10 mM of Tris-HCl, pH 8.5) to the washed beads The elution buffer and beads were incubated at different conditions (25–80°C for 2–3 min) to optimize the elution efficiency of bound DNA The concentrations of the original and eluted DNA samples were measured and compared by UV-Vis spectrophotometer (ND-100, NanoDrop Technologies)

PCR-Similar process was conducted on the two-step microfluidic device (Figure 6.2c) The 100

bp ladder or PCR product (7 μL) and magnetic beads (5 μL) in purification buffer (11 μL) were first loaded into M3 and M4, respectively, with a volume defined by the meter chambers These

two solutions were mixed using an external syringe pump (Cavro XLP 6000), pushed to the bead chamber (C3), and incubated for 1 min The impurities in the bead chamber were then washed with washing buffer introduced from A5 at a flow rate of 200 µL/min for 15 min with beads captured by a permanent magnet (M1219-5, Assemtech) After washing, elution buffer (10 mM of Tris-HCl, pH 8.5, 5 μL) was introduced into the bead chamber, and incubated with the beads at 25–80°C for 2–3 min to release the bound DNA The magnetic beads were actively mixed at a rate of 0.5 Hz before elution using external electromagnets with the setup shown in Figure 6.4b A permanent magnet was mounted on a flexible and suspended metal arm located between two electromagnets (GMHX, Magnet-Schultz Ltd) Alternate magnetic forces were applied to the metal arm when 180° out-of-phase voltages were supplied to the electromagnets, which swung the metal arm and the mounted magnet The electromagnets were powered through the solid-state relays (ODCM-5, Tyco) and a DC power supply (HY3003, Digimess) Electromagnetic forces were regulated through the relays using an analog voltage output board (PCI-6713, National Instruments), and a computer with a LabVIEW program (National Instruments)

6.3.3 Agarose gel electrophoresis

Synthesized products were analyzed by 1.5% agarose gel (NuSieve® GTG®, Cambrex Corporation), stained with ethidium bromide (Bio-Rad Laboratories), and visualized using Typhoon 9410 variable imager (Amersham Biosciences) Gel electrophoreses were performed at

100 V for 45 min with 100 bp ladder (New England) and 5 μL of DNA samples collected from commercial thermal cycler and devices

6.3.4 DNA sequencing

One-step and two-step overlapping synthesis products were sequenced to check the error rate GFPuv gene synthesis products (without further PCR purification) were cloned into vector pCR®2.1-TOPO® (Invitrogen) and transformed into chemically competent TOP10 cells After overnight growth on 1× Luria-Bertani (LB) agar plate (with 100 µg/ml of ampicillin), individual colonies were picked and grown in 1× LB media (with 100 µg/ml of ampicillin) The plasmid

DNA was extracted by using QIAprep Spin Miniprep Kit (QIAGEN), and sequenced by Research Biolabs (Singapore) In total, 150 individual samples were sequenced using M13 forward and reverse sequencing primers (for one-step process: 96 from microfluidic device and 48 from 0.2-ml PCR tube; For two-step process: 54 from microfluidic device and 48 from 0.2-ml PCR tube) All sequence results were analyzed using sequence analysis tool Vector NIT, and the errors were verified by visual confirmation of the electrophoregrams of ABI PRISM® 3100-Avant Genetic Analyzer

6.4.1 Device operation

A precision syringe pump with multi-position valve (Cavro XLP6000, Tecan Systems) was used

to manipulate reagents inside the microfluidic device This syringe pump is capable of withdrawing and dispensing reagents with a volume resolution of better than 10 µL, as controlled

by a LabVIEW program (National Instruments) To control the hydrogel valves and thermal cycling simultaneously and separately, two TE modules with individual temperature controllers were used (Figure 6.2 (c)) One TE module (TE 1) was located under the PCR chambers to perform the temperature cycling, and the other TE module (TE 2) was located under the hydrogel valves to control their action

The overall device operation of the gene synthesis device was illustrated in Figure 6.3(A) with the process duration of each step The totoal operation time is ~ 3 hr 7 min Figure 6.3(B) shows the detail operation of each step with volume defined by each chamber Oligonucleotide and PCR mixture was first loaded into the PCA chamber through the inlet port (A1) The solution was then sealed by the hydrogel valves (V1 and V2) and thermally cycled with the thermoelectric heater to assemble oligonucleotides After PCA, the hydrogel valves (V1 and V2) were opened, and the solution was pumped into meter chamber M1, and simultaneously mixed with an equal volume of fresh PCR mixture containing outer primers from meter chamber M2 To enhance mixing, this mixture was shuttled between two mixing chambers (C1 and C2) five times (flow rate = 120 µL/min) with the precision syringe pump at inlet port B2, and then moved to PCR

chamber with hydrogel valves (V3 and V4) kept open After PCR amplification, the hydrogel valves (V3 and V4) was open again, and the solution was moved to meter chamber M3 (through inlet port A3), and simultaneously mixed with the magnetic beads solution defined by meter chamber M4 in the beads chamber (C3) With the DNA-absorbed magnetic beads captured by a permanent magnet, the impurities solution was washed out Finally, the elution buffer was loaded and mixed with the magnetic beads; the DNA was then released into the elution buffer To control the flow direction, unused inlets and outlets were plugged with metal pins For example, to direct PCA mixture to PCR chamber, the inlets (A4–A7) for solid-phase PCR purification were plugged

Two micromixers were developed to effectively mix the PCA product with fresh PCR mixture for PCR amplification, and mix the magnetic beads with DNA solution and elution buffer for solid-phase PCR purification The gene synthesis chip was to be developed as a bench-top instrument to perform automatic gene synthesis To control the cost and simplify the fabrication process of these disposable chips, mixing approaches utilizing simple fluidic structures and methods were desired Figure 6.4 shows our approaches using shuttle mixing and electromagnetic mixing In shuttle mixing, solution was shuttled between two chambers connected by a narrow channel This narrow channel reduced the diffusion distance of two mixing reagents, and the abrupt opening at channel-chamber junctions created chaotic advection at the junctions and recirculated the flow [120] Both of these features were reported to enhance mixing [65] Figure 6.4a demonstrates the performance of the shuttling micromixer Two colored food dyes (blue and red) were well mixed after shuttled three times between two chambers at a flow rate of 120 µL/min, pumped by a precision syringe pump This method was effective with compact and simple fluidic structures as compared to other reported methods [65] Mixing was completed within 1 min in our application with a fluid volume of 19 µL No visible air bubbles were trapped inside the solution

(B)

Figure 6.3: (A) Device operation diagram with process time of each step (B) Detailed schematic

diagrams of each step: (a) Oligonucleotides and PCR mixture were loaded into PCA chamber (highlighted in red) from A1 PCA was then conducted (b) PCA-assembled solution (pumped through B1) was mixed with fresh PCR mixture containing outer primers (pumped through A2) The mixed PCR precursor was illustrated in green (c) Mixed PCR precursor (green color) was positioned in PCR chamber, and the PCR amplification was performed (d) PCR-synthesized product (highlighted in green) and ChargeSwitch reagent (illustrated in yellow with black dots) were pumped and loaded into beads chamber After mixing and incubation the magnetic beads were captured by a magnet (e) Magnetic beads were washed by washing buffer pumped from A5 (f) Elution buffer was loaded and mixed with magnetic beads, after incubation the magnet was applied to fix the beads Synthesis product was eluted into elution buffer and collected through A7 (highlighted in green)

Permanent neodymium rare earth magnet was utilized to capture magnetic beads in the microfluidic device [72, 121, 122], as it provided a strong magnetic force However, this strong magnet

could also cause the aggregation of beads [123], and hinder the beads from full contact with the desired biomolecules in solution To make sure that the beads were well mixed with solution, we have developed an approach to agitate the solution inside the chamber (Figure 6.4b) A permanent magnet was mounted on a flexible metal arm that was sandwiched by two electromagnets When out-of-phase voltages were applied to the electromagnets, alternating magnetic forces were generated, which swung the metal arm and the permanent magnet simultaneously The swinging magnet dragged the magnetic beads and agitated the solution This simple approach was employed to mix the elution buffer with DNA-bound magnetic beads in the final step of PCR purification at a mixing rate of 0.5 Hz

Figure 6.4: (a) Photographs of micromixer Colored dyes (blue and red) were well mixed after

being shuttled three times between two chambers (b) Schematic illustration of the experimental arrangement with a syringe pump, electromagnetic mixer, thermoelectric heaters and data acquisition

6.4.2 In situ hydrogel valve

Microvalves are critical to the successful integration of PCR process into a microfluidic device, which has to be able to withstand at least 6.8 psi to ensure successful sealing of the PCR mixture within the chamber During the PCR process, the air solubility variation from 4 °C to 94 °C could create a pressure of ~3.1 psi [124], and potential trapped air bubbles would contribute to an additional pressure of 3.7 psi at 94 °C [73]

Figure 6.5: The thermal response of in situ photopolymerized hydrogel valve The valve

functions were highly repeatable The insets showed the transitions of valve functions

The hydrogel valves were tested prior to use on the single chamber device (Figure 6.2b) with a liquid flow meter (SLG1430, Sensirion) connected between a constant pressurized water reservoir (8 psi) and the device The flow rate variation was monitored as the valve was subjected

to repetitive cooling and heating by a thermoelectric heater underneath the device Figure 6.5 shows the valve’s temperature and the flow rate as functions of time The valve dimensions were 1.5 mm × 1.5 mm × 0.5 mm At a temperature below the hydrogel’s LCST (32 °C), the thermally responsive hydrogel swelled and blocked the valve, indicated by the desrease in flow rate When the temperature was above the hydrogel’s LCST, the hydrogel shrunk in volume and allowed for fluid flow through the channel As indicated in Figure 6.5, the valve functions were highly

repeatable with valve’s opening and closing times of ~5 sec and ~20 sec respectively (see inset in Figure 6.5), limited by the ramping rate of the underneath heater, and the water diffusion rate of the hydrogel swelling/de-swelling process [125] The closed valve exhibited no leakage (zero flow rate) at 8 psi, showing that it was strong enough to seal the PCR chamber Yu et al [126] reported that in situ photopolymerized NIPAAm-based valve could withstand a pressure of up to 200 psi Wang et al [74, 127] also described the successful integration of chemically polymerized NIPAAm hydrogel valve with PCR by manual insertion of pre-synthesized hydrogel in the flow paths

6.4.3 PCR thermal cycling

The gene synthesis process was integrated into a chip composed of a PDMS fluidic structure on a silicon substrate Although PDMS has a number of interesting material properties that make it superior for constructing highly integrated biological microsystems, its non-specific protein adsorption [33, 128] and permeability to water vapour [129] could pose problems in performing PCR

in microfluidic environment, which has a small volume and a high surface-to-volume ratio To address these problems, the fabricated devices have been coated with 2 µm-thick parylene, which created a barrier to against water vapor diffusion and improved the surface compatility with PCR mixture [117]

A thermoelectric module with heat sinks and fan was utilized for thermal cycling Figure 6.6 showed the temperature profiles of the thermal cycler obtained from a calibration chip, which has identical dimensions as the actual device, but has a thermocouple embedded within the PCR chamber Temperatures at the heater surface and within the PCR chamber were measured The temperature difference between these two locations indicated that the 500 µm-thick silicon substrate could cause a temperature drop of > 5 °C, which was compensated during the operation

of thermal cycling The heating and cooling rates estimated from Figure 6.6 were 2.4 °C/sec and 4.3 °C/sec, respectively, which were faster than those in commercial thermal cycler (DNA Engine PTC-200)

To generate enough quantity of synthesized products for further process, the PCR chamber was designed with a volume of 7 µL Gene synthesized by PCR methods contained both

full-length DNA and intermediaries with shorter lengths After synthesis, gel electrophoresis was usually conducted to confirm the success of the synthesis, and to separate the full-length product, which was then extracted from the gel by using gel extraction kits Some DNA could be lost due

to these steps and the pipetting process The PCR mixture was introduced into the PCR chamber through the hydrogel valves that were kept opened by thermoelectric heater at 60 °C Once the PCR chamber was filled with the solution, the hydrogel valves was cooled to 4 °C, sealing the chamber Since silicon with high thermal conductivity was used as the device substrate, the PCR chamber and hydrogel valves were positioned apart to minimize thermal interference between the PCR thermal cycling and the valves’ operation The hydrogel valve has to be kept below the transition temperature to seal the PCR chamber during thermal cycling, which could reach a temperature as high as 95 °C One way to suppress the thermal interference and reduce the dead volume between the PCR chamber and valves was to use a polymer substrate (such as polycarbonate) [130] or an isolation trench to suppress the lateral heat flow along the substrate, as reported by Wang et al [74] and Yang et al [131]

Figure 6.6: Thermal cycling profiles of the custom-built PCR thermal cycler A thermocouple

mounted on the heater was used in the temperature feedback control (heater temperature) for thermal cycling The temperature difference between the heater surface and within the PCR chamber (chamber temperature) was compensated using a LabVIEW program

6.4.4 Comparison of one-step and two-step gene syntheses

The thermal cycler’s requirement for PCR assembly was the same as the standard PCR amplification However, the number of oligonucleotides involved in PCR assembly was much larger than in the standard PCR amplification Full-length DNA was constructed from a pool of solution containing tens of oligonucleotides with various melting temperatures The efficiency of successful gene synthesis relied on several important factors including the polymerase, concentrations of assembly oligonucleotides and amplification primers, and structure and properties of oligonucleotides [37, 92]

To identify the baseline of oligonucleotide and primer concentrations, a segment of GFPuv (760 bp) was synthesized from a pool of short oligonucleotids (40 bases) using two-step PCR process by varying oligonucleotide concentration from 5 to 25 nM, and primer concentration from 0.1 to 0.4 μM; this was conducted on the commercial thermal cycler Desired full-length product was first assembled from oligonucleotides without outer primers (PCA assembly), and then amplified by adding these primers at the second PCR (PCR amplification) To match the microfluidic device design (Figure 6.2c), the PCR amplification was performed with the PCA product diluted with an equal volume of fresh amplification reaction mixture Gel electrophoresis results for PCA assembly (Figure 6.7a) and PCR amplification (Figure 6.7b) were illustrated for the indicated oligonucleotide and primer concentrations The PCA has smearing gel results, indicating that the assembled product contained a spectrum of DNAs, the majority of which possessed lower molecular weights than the desired target (760 bp) For products assembled from oligonucleotide concentrations of < 10 nM, the quantity of full-length DNA (760 bp) was very low and invisible in the PCA gel images, but this was effectively boosted with PCR amplification PCR gel images showed samples 1-1 to 1-3 synthesized with an oligonucleotide concentration of

5 nM and a primer concentration of 0.1 μM, 0.2 μM and 0.4 μM, respectively Other samples were the same as samples 1-1 to 1-3, except that the oligonucleotide concentrations used were 10

nM, 15 nM and 25 nM Syntheses with an oligonucleotide concentration of > 10 nM and a primer concentration of 0.1 μM and 0.2 μM failed to provide the desired full-length (760 bp) product In

DNA Of the four samples, sample 2-3 with an oligonucleotide concentration of 10 nM and a primer concentration of 0.4 μM produced the most full-length product, even though samples 3-3 (15 nM, 0.4 μM) and 4-3 (25 nM, 0.4 μM) have more full-length DNA generated initially from the PCA step These results corresponded well with those reported by Kong et al [34] and Wu et al

[92]

for one-step gene synthesis process Based on these findings, an oligonucleotide concentration

of 10 nM and a primer concentration of 0.4 μM were selected for gene synthesis on a microfluidic device

Figure 6.7: Agarose gel (1.5%) electrophoresis showing the synthesis yields with

oligonucleotide concentrations of 5–25 nM and outer primer concentrations of 0.1–0.4 μM for the two-step process Syntheses were conducted using a commercial thermal cycler (a) PCA results (b) PCR amplification results

With the optimized oligonucleotide and primer concentrations, GFPuv (760 bp) was successfully synthesized from a pool of short oligonucleotids (40 bases) by using either one-step

(single-chamber chip) or two-step microfluidic devices Strong, dominant band of the desired products were obtained in the gel images (Figure 6.8) The visually estimated yields of microfluidic devices were ~ 50% of the controls performed in PCR tubes with a commercial thermal cycler These were limited by the dead volume (2.87 µL) in the channels between the PCR chamber (7 µL) and the valves The oligonucleotides mixture within the dead volume did not assemble, but contributed to ~ 30% of the eluted solution The gel results also demonstrated that parylene was compatible with PCR reaction mixture, and effectively blocked the reagents against evaporation from the water vapor-permeable PDMS

Figure 6.8: Agarose gel (1.5%) electrophoresis comparing the synthesis results conducted

within commercial thermal cycler (machine) and microfluidic device (a) One-step process (device: single-chamber chip) and (b) two-step process (device: two-step chip) conducted with

an oligonucleotide concentration of 10 nM and a primer concentration of 0.4 µM

Compared to the one-step process, the two-step process generated much more full-length product from the same amount of initial oligonucleotides In the one-step process, the assembly and amplification were conducted simultaneously, which competed for the fixed amount oligonucleotides and monomers (dNTPs), and rendered intermediary products with lower molecular weights (Figure 6.8a), which went alone well with the previous analysis mentioned in chapter IV and V The process competition was minimized in the two-step process, resulting in more full-length product [79].The two-step process was reported to be more reliable than the one-

step process, which sometimes failed to generate full-length DNA [8, 26] Gene synthesis with the two-step process also allowed for different annealing temperatures to optimize the assembly and amplification processes separately It should be addressed that other than two-step gene synthesis, TopDown and Automatic TouchDown one-step gene synthesis method are also suitable approaches for chip based DNA assembly

The assembled sequence was identified by DNA sequencing Synthesized products from the microfluidic devices and PCR tubes were cloned directly without further purification using PCR®2.1-TOPO® cloning vector (Invitrogen) Full-length target along with intermediary products were all cloned to reflect the real composition of the synthesized products Table 6.1 showed the sequencing results The error rates per kilobase (kb) calculated from full-length clones were 3.45 in device and 4.36 in PCR tube for the one-step process, and 4.01 in device and 4.10 in PCR tube for the two-step process These values were within the range of the error rates reported (1.8–6 per kb) [8, 14, 20, 81] Most errors (> 85%) were associated with single-base insertion, deletion and mutation The indifference in error rates implied that they were independent of the synthesis methods (device versus PCR tube) and processes (one-step versus two-step) [14, 81] Hoover et al

[14]

and Tian et al [81] pointed out that the greatest errors were attributed to the quality of synthetic oligonucleotides, not from the fidelity of polymerase enzyme Oligonucleotides were chemically synthesized base-by-base with a step yield of ~98.5% [132].The overall yield of full-length oligonucleotides decreased as the oligonucleotide length increased For example, only 54.6% of oligonucleotides was full-length in a targeted 40 base-long synthesis product The building blocks

of synthetic oligonucleotides containing both perfect match sequence and impurities with mismatch (single base and multiple bases) could all have participated in the PCR process and generated products of incorrect sequence In contrast, the DNA polymerase has a replication error rate of ~10-6 base/duplication [59], which was 3–4 orders lower than the error rate of synthetic gene products Performing gene synthesis in a microfluidic device might not improve the accuracy of synthesis products, as demonstrated by Kong et al in microPCR one-step gene synthesis [34] However, it would reduce the handling time and reagents costs, and eliminate human process factors

PCR product cloning and DNA sequencing were required to ensure that an accurate synthesis product was obtained These processes involved substantial laboratory efforts To obtain

an error-free gene, many randomly selected clones were sequenced [60, 86], which might contain either undesired truncated DNAs or the desired full-length DNA The greater full-length yield of the two-step process increased the possibility in obtaining effective full-length clones, and in achieving an error-free gene About three out of four clones (35/47 in PCR tube) produced by the two-step process contained full-length products, which was greater than that produced in the one-step process (about one out of three clones (16/47 in PCR tube)) (Table 6.1) Therefore, the two-step process would be preferred in minimizing the number of colony sequencing required to obtain an error-free gene, and the effort of cloning and DNA sequencing, especially for long DNAs

Table 6.1: Errors and efficiencies in the synthesis of GFPuv using one-step and two-step

processes in the microfluidic device vs standard PCR tube (machine)

6.4.5 Thermally enhanced solid-phase PCR purification

For applications such as cell-free protein synthesis [133, 134] (which directly use synthetic genes for protein expression) and integration of enzymatic error filtering methods [60, 86, 104, 135] on chip to reduce the error rate of synthesized products, a solid-phase buffer exchange process was integrated with the two-step microfluidic device utilizing magnetic beads based PCR purification method (ChargeSwitch PCR clean-up Kits, Invitrogen) This process was intended to purify the assembled product from short primers and dNTPs, and to prepare the buffer solution for downstream application Silica-coated magnetic beads could help simplify the device integration

[72, 136]

as compared to other nucleic acid extraction methods reported by Jemere et al.,[52] West et

al [137] and Breadmore et al [49]

ChargeSwitch utilized the same approach as other reported methods [49, 137] DNA was first adsorbed onto the silica surface under high ionic strength conditions The unbound impurities were washed away, and then the adsorbed DNA was released into solution with a higher pH (10

mM of Tris-HCI, pH 8.5) The ChargeSwitch Kit was first optimized in standard PCR tubes using

100 bp DNA ladder with a known DNA quantity (1.19 µg) as the control following the approach and protocol suggested by manufacturer The reagents volume was modified to match the design

of microfluidic device After the baseline protocol was established using PCR tube and 100 bp ladder, the procedure was applied to microfluidic device for 100 bp DNA ladder and PCR synthesized product The total amount of 100 bp DNA ladder (1.19 µg) or PCR product (1.98 µg) was less than the binding capability of the ChargeSwitch beads loaded Based on the manufacturer’s protocol, the ChargeSwitch beads would bind double-stranded DNAs with lengths

of > 90 bp; thus, 100 bp DNA ladder was selected as the control The DNA extraction included three steps – DNA capture, impurities wash, and DNA elution The DNA elution conditions (time and temperature) were investigated to increase the extraction efficiency

Figure 6.9: The effect of elution temperature and incubation time on DNA extraction

conducted within microfluidic device (■: 3 min) and standard PCR tube (□: 3 min; ◊: 2 min)