Joseph Jacobson
Associate Professor
Program in Media Arts and Sciences
School of Architecture and Planning
Department of Mechanical Engineering
Thesis Advisor
Peter Carr
Research Scientist
Program in Media Arts and Sciences
School of Architecture and Planning
Thesis Reader
Scott Manalis
Associate Professor
Department of Biological Engineering
Massachusetts Institute of Technology
Thesis Reader
Franco Cerrina
Lynn H. Matthias Professor in Engineering
Department of Electrical and Computer Engineering
University of Wisconsin-Madison
Thesis Reader
1

The ability to synthesize custom de novo DNA constructs rapidly, accurately, and inexpensively is highly desired by researchers, as synthetic genes and longer DNA constructs are enabling to numerous powerful applications in both traditional molecular biology and the emerging field of synthetic biology. However, the current cost of de novo synthesis—driven largely by reagent and handling costs—is a significant barrier to the widespread availability of such technology. The use of microfluidic technology greatly reduces reaction volumes and corresponding reagent and handling costs.
Additionally, microfluidic technology enables large numbers of complex reactions to be performed in parallel. While microfluidic devices have been used to miniaturize a variety of chemical and biological processes, the benefits of such devices have yet to be realized in the area of de novo DNA synthesis. In this thesis, I aim to employ microfluidic technology to gene synthesis, demonstrating firstly the synthesis of custom DNA molecules in a microfluidic environment, followed by development of technology for applications such as the hierarchical synthesis of long DNA molecules, integration of gene synthesis with on-chip protein synthesis, and direct synthesis of genes from hybrid microfluidic-DNA microarray devices.
2

It has long been recognized that the capacity to design and synthesize genes and longer DNA constructs can be enabling to a broad cross section of applications within molecular biology
1
including the study of large sets of single genes
2
, the design of genetic circuitry
3
, the engineering of entire metabolic pathways for target molecule manufacture 4 , and even the construction and re-engineering of viral and bacterial genomes
5-7
. Currently, the availability of such synthetic DNA molecules is limited due to the significant time and cost associated with their synthesis. Gene-length constructs require 2-4 weeks to synthesize and cost approximately $0.65 to $1.10 per base (i.e. $650 to $1100 for a 1 kilobase gene). In comparison, shorter synthetic oligonucleotides, typically tens of bases in length, are widely available at a cost of ~$0.10 per base and can be ordered for arrival in a day. The goal of this thesis is to develop technology to help reduce the cost and time required for the synthesis of gene-length, and longer, DNA constructs, by employing microfluidic technology.
Microfluidic technology is a research area currently undergoing a rapid period of development. Fabricated via lithographic techniques developed in the semiconductor industry, microfluidic devices featuring dense arrays of reactors and active components for complex fluid manipulation, all with feature sizes in the tens to hundreds of microns
(10
-6
m), have shown promise to revolutionize biological and chemical processes just as integrated circuits revolutionized computing. The primary benefits of such technology are the significant reduction in sample volumes (often several orders of magnitude as compared to traditional bench-top processes), and thus associated reagent costs, high throughput facilitated by the parallelization of tens to thousands of simultaneous
3

reactions or more, the automation and integration of a variety of processes on a single chip, and also the study of novel physical phenomena occurring at such length-scales. By applying microfluidic technology to gene synthesis, associated reagent and samplehandling costs could be dramatically reduced, thus increasing the availability of such synthetic constructs and enabling the broad set of research applications mentioned above.
Additionally, the benefits of automation and process integration enabled by microfluidic technology will be explored in applications such as the automated, hierarchical synthesis of long DNA molecules, the integration of gene synthesis with protein synthesis, and also the integration of DNA microarrays to microfluidic devices for the direct synthesis of genes from inexpensive microarray oligos.
This section will discuss in greater detail the two major research areas of this thesis, namely gene synthesis and microfluidic technology, respectively.
The core technology for custom DNA synthesis centers on the assembly of pools of oligonucleotides (oligos), typically less than 50 nucleotides in length, into increasingly larger DNA molecules. These oligos, hereafter referred to as “construction oligos,” are synthesized by variations of phosphoramidite chemistry
8
, and are the building blocks for the different gene synthesis techniques developed thus far. The most widely reported methods for building long DNA molecules involve variations of the polymerase-mediated
4

assembly technique shown in Figure 1, collectively termed Polymerase Construction and
Amplification (PCA) 9-10 . Here, much like in the more conventional Polymerase Chain
Reaction (PCR), three temperature steps are employed to denature, anneal, and elongate the various overlapping construction oligos until, after multiple rounds of thermocycling, the desired full length DNA construct is obtained. Furthermore, assembly and amplification can be performed in a single reaction with the introduction of amplifying primers 11 . Thus, once a minute quantity of full length product is assembled, this product is amplified as per PCR. Using such polymerase-mediated techniques, researchers have successfully synthesized DNA constructs as large as 12 kb
12
and 15 kb
11
. A PCA process was also employed as the first step in generating a 32 kb DNA construct by Santi and coworkers
13
. In addition, significant progress has been made in correcting synthesis errors,
Figure 1. Schematic for gene synthesis by Polymerase Construction and Amplification
(PCA). Multiple rounds of oligo annealing and extension by DNA polymerase generate successively longer DNA assemblies from a starting pool of construction oligos, typically < 50 nt, until the full-length gene is produced. The pool of heterogeneous
DNA products is enriched for the full-length species by amplification in a separate subsequent reaction, or in the same reaction by including amplifying primers in the reaction mixture.
5

which originate primarily from the phosphoramidite synthesis of initial oligonucleotide building blocks. The use of protein-mediated error correction has been effective in increasing the accuracy of synthetic DNA
14-16
, with error rates as low as 1 per 10,000 base pairs reported
14
.
Despite these promising results significant challenges remain, most significantly the cost and time of synthesizing long constructs. Currently, while conventionally synthesized oligos are available at a cost on the order of $0.1 dollars per nucleotide, the cost for custom gene synthesis services is significantly higher, on the order of $0.65-
$1.10 dollars per base pair, with the major expenditure components for such long syntheses being attributable to reagent and sample handling. Microfluidic technology provides an elegant means to overcome these limitations. By scaling reactions down to volumes of less than a microliter, reagent costs can be substantially reduced 17 .
Furthermore, microfluidic technology enables highly parallelized synthesis along with the potential for automated sample handling and process integration.
We are currently in a period of rapid microfluidic technology development as an increasing variety of chemical and biological processes are being miniaturized and integrated on-chip. Many parallels can be drawn between the current efforts to miniaturize biology and the revolutionary effects miniaturization had on computing. Just as integrated circuits utilized miniaturized transistors to automate computation, similarly microfluidic chips employing miniaturized valves and higher order components seek to automate biology. Current biological automation is comparable to the vacuum tube era
6

of computing, as large fluid-handling robots fill entire rooms. However, increasingly complex (in terms of valves/cm 2 ) micro- and nano-fluidic chips capable of higher degrees of process integration are becoming gradually more common, thus defining a trajectory toward fully automated fluidic biological systems.
Microfluidic devices have been fabricated utilizing a wide range of materials and techniques for a similarly wide variety of applications. The first device that could be reasonably characterized as ‘microfluidic’-handling either microliters of fluid or featuring micron feature sizes-was a silicon based gas chromatograph fabricated at Stanford
University in 1975
18
. Research in this area was sparse until the 1990s, when a wider array of applications emerged, such as on-chip capillary electrophoresis, flow cytometry, and PCR. The development of polymer materials for microfluidic applications, most notably the invention of ‘soft lithography’ by George Whitesides’ group at Harvard 19
(1998), along with the first demonstration of an elastomeric valve by Stephen Quake’s group at CalTech
20
(2000), led to an explosion of research activity that continues today.
Perhaps the most prominent approach for fabricating polymer microfluidic devices involves the molding of a transparent silicone based elastomer, poly(dimethylsiloxane), or PDMS. PDMS, most commonly used for applications such as water-proofing or electrical insulation (and also as the major component of silly putty), has proven over the past ten years to be the most widely used material for fabricating microfluidic devices. Fabricating devices from PDMS is inexpensive and relatively straight forward. Utilizing traditional optical lithography with cheap (<$100) transparency masks, fluidic designs can be transferred to photoresist on a silicon wafer, which can then act as a mold from which liquid PDMS can be polymerized and cast, thus
7

transferring the desired pattern to the PDMS. Once cast, the PDMS can be bonded to a variety of substrates, commonly glass, but also to other layers of PDMS for constructing multiple layer structures.
A large variety of biological processes as diverse as polymerase chain reaction
(PCR)
17 and protein crystallization
21 have been miniaturized and automated with, in some cases, substantial improvements over conventional bench-top methods. Highly automated systems capable of performing parallel reactions with integrated, multiple functions have also been demonstrated
22
. Despite these advances, the microfluidic de novo synthesis of DNA constructs remains an unexplored area of research. Large numbers of research laboratories commit tremendous resources (labor, money, and time) to manipulating, modifying, and sometimes synthesizing, desired DNA molecules. If user-specified DNA molecules were available rapidly and at low cost, these same researchers would be able to perform many more experiments with the same resources. If larger, more complex DNA constructs were attainable, then new areas of research would be made possible, employing the direct synthesis of biochemical pathways, gene networks, or even the genomes of simple organisms.
In this thesis, strategies to apply microfluidic technology to gene synthesis will be explored. While initial experiments will focused on demonstrating basic benefits of microfluidic technology such as reduced reagent consumption and smaller reactor volumes, additional advantages such as increased automation and higher degrees of process integration will also be pursued.
8

The major goals of this thesis will be to demonstrate, firstly, the fabrication of a microfluidic device capable of performing gene synthesis in parallel reactions at volumes at least an order of magnitude smaller than those conventionally utilized, and secondly, the extension of this technology to any of three proposed areas: (1) a device capable of synthesizing long (>1 kb) DNA molecules via an automated, hierarchical synthesis approach; (2) a hybrid device combining a microfluidic device with a DNA microarray; or (3), an integrated device capable of on-chip gene and protein synthesis, respectively.
The benefits and strategies for each approach will be discussed in the sections below.
The first major research goal will be to demonstrate the synthesis of genes, in parallel, in a microfluidic environment. Because conventional gene synthesis reactions are conducted in volumes on the order of tens of microliters, successfully scaling down synthesis reaction volumes to those typically utilized in microfluidic technology (< 1 microliter) would result in a significant reduction in reagent usage and subsequent costs, and also a reduction in waste. Furthermore, because microfluidic technology is highly scalable, the initial demonstration of limited parallel synthesis leads quite readily to the expansion of the number of parallel reactions to the maximum achievable density for a typical substrate, such as a microscope slide, which can be on the order of tens of thousands of unique reactions.
9

In order to achieve this foundational goal, several important engineering factors must be considered. Firstly, material selection in designing microfluidic devices is crucial. Because microfluidic devices have a surface area to volume ratio orders of magnitude higher than that of conventional bench top reaction vessels, the frequency of surface interactions between reactor walls and the various biomolecules likewise increases. These interactions can often be undesirable, as depending upon the surface characteristics of the reactor walls and the biomolecules, non-specific adsorption can occur, thus altering reagent concentrations and hindering the reaction. Secondly, technology for driving active, on-chip components such as valves and mixers must be chosen.
Initially, microfluidic devices will be fabricated from PDMS, which as described in section 3.2 has proven to be relatively simple to machine (with the appropriate molds, fabrication time is ~3-4 hours; from conception of an idea, fabrication time is 1-2 days), and has elastomeric mechanical properties that enable low-pressure pneumatic valves.
An example of such a device fabricated by the author can be seen in Figure 2. Here, four reactors, indicated in blue and green, are defined, each with a volume of 500 nanoliters
(10 -9 liters). The device shown is composed of three-layers of PDMS: a ‘flow’ layer through which reaction mixtures are loaded (indicated by blue and green features), another network of channels atop the ‘flow’ layer that act as a ‘control’ layer for valves
(indicated in red), and finally a featureless PDMS-layer on the bottom. With this type of device geometry, the intersection of ‘control’ and ‘flow’ layer channels defines a microfluidic valve. By applying positive pressure in ‘control’ layer channels, the thin membrane of PDMS in between the two channels deflects downwards, thus sealing the
10

‘flow’ channel. An example of such a valve is shown in the right inset of Figure 2. In this device, two control lines (red) are utilized to address the inlet and outlets of each of the four reactors. Additionally, because PDMS is porous and thus prone to evaporation, a network of fluid channels (yellow) is placed above the reactors to help reduce sample
evaporation.
Figure 2. Optical images of a microfluidic device capable of conducting four parallel
500 nL reactions with various features emphasized with food coloring. Left inset: gene synthesis chamber (blue and green) and water jacket (yellow) layers. Right inset: fluid inlet channel (blue) overlaid with red valve channel (red). Scale bars correspond to 200 μm.
Upon demonstrating a set of materials and technology capable of conducting gene synthesis in a microfluidic environment, next, materials and valving technology will be
11

explored for developing devices capable of more complex, automated processes such as the hierarchical synthesis of long (>1 kb) DNA molecules. Ultimately, there are limits to the number of unique construction oligos present that can be successfully utilized as the building blocks for a given synthesis reaction. If the complexity of the oligo pool exceeds a certain amount, the synthesis reaction will not proceed to completion and will instead yield a heterogeneous mixture of incomplete species. DNA molecules longer than ~1 kb are difficult to synthesize in a single reaction. Thus, in order to build truly long constructs (some of the most exciting applications call for the de novo synthesis of bacterial chromosomes on the order of 10
6
bp), a hierarchical approach is required. Such an approach is shown schematically in Figure 3. Here, upon synthesizing four initial
DNA segments A-D, sets of reactions are conducted pair-wise to synthesize longer,
Figure 3. Work flow for a hierarchical fluidic gene synthesis device. A hierarchical assembly begins with parallel syntheses in four separate chambers generating DNA fragments A-D. These fragments can then be combined pair-wise in two parallel reactions generating the intermediates, E and F, which can finally be assembled into one large construct several kilobases in length. In this situation 7 total fluidic chambers with two hierarchical levels are employed to synthesize the full-length product.
12

intermediate molecules E and F, which can be assembled into the desired construct in a final reaction. Such a process has been utilized by Church et al., for example, to synthesize a ~15 kb construct
11
.
The advantages of automation enabled by microfluidic technology apply well to the problem of hierarchical synthesis. A CAD design file of a device capable of a singlestep hierarchical synthesis is shown in Figure 4; fluid lines are indicated in blue, and control lines for valving are indicated in red. Here, reaction mixtures capable of synthesizing two DNA fragments A and B are loaded into their respective reactors, followed by an initial parallel synthesis reaction. Upon completion of this reaction, volumes containing synthesized A and B segments are then loaded into the central rectangular chamber, whereby a series of valves can be utilized as a peristaltic pump to mix the two DNA segments along with a fresh mixture of reagents necessary for the final synthesis stage. The device shown in Figure 4 has been demonstrated to mix three separate solutions of food coloring—two 30 nL volumes representing the two gene segments and a 240 nL volume for fresh reagents, giving a 300 nL total mixer volume-into a homogeneous mixture in under 30 seconds. Upon completion of mixing, the final synthesis reaction can be conducted to assemble the desired >1 kb construct.
13

Figure 4. A CAD design file for a microfluidic device capable of a single-stage hierarchical assembly. Two gene fragments will be synthesized by PCA in reactors A and B, followed by the mixing of the two assembly products in the rectangular mixer along with a fresh supply of reagents for subsequent assembly. Fluid channels are indicated in blue, with control lines indicated in red. Such a device should be capable of synthesizing a gene of greater than 1000 bp.
Upon completion of this demonstration of single-stage hierarchical synthesis, extension of this process to larger, higher order assemblies is readily realizable.
Additional materials, including modified versions of PDMS, will be investigated for the hierarchical assembler proposed in this section, as initial experiments have shown that contaminant molecules generated from the PDMS impede hierarchical synthesis.
Thus, rigid materials such as glass, and coatings compatible with PDMS (e.g. CYTOP, a fluropolymer), will be explored as well.
Recent experiments have demonstrated the first use of oligos from DNA microarrays as the building blocks for gene synthesis
11,23-24
. Because in such arrays large
14

numbers of distinct oligos--for a single high-density array, 10
4
-10
5
or more--are synthesized in parallel, an oligo on the array can cost as little as 1x10 -5 to 1x10 -3 dollars per base, depending upon the array, which typically cost between a few hundred to a few thousand dollars (e.g. $489 for a 224,000 spot Agilent microarray). These numbers are orders of magnitude less than the cost of conventionally synthesized oligos, which are commercially available on the order of 1x10
-1
dollars per base (list price Invitrogen,
IDT). The per-base cost for companies that offer custom gene synthesis services is even higher, as much as $0.65-$1.10 dollars per base. Thus, the current significant contribution of oligo costs to the overall cost of gene synthesis could be reduced to an almost trivial amount if the wealth of raw building material provided by microarrays could be successfully accessed. Unfortunately, the bulk sample handling of oligos harvested from microarrays has not yet approached this potential, as at most several hundred oligos have been used to make 21 genes
11
.
A number of factors currently limit the utility of microarray oligos for gene synthesis. As described above, PCA can become limited as the complexity of the oligo pool increases. Just as multiplex PCR (simultaneous amplification of several DNA templates in the same reactor) suffers from inconsistencies such that each template may not be equally amplified, similarly the simultaneous amplification and assembly of 10
5
or more sequences is unlikely to proceed evenly. For gene synthesis, this is a serious problem; if the pool becomes dominated by a few DNA species the required pool diversity would be lost, rendering assembly impossible. The absence of a single oligo prevents the assembly of its corresponding gene, so that losses even as low as 0.1% could interfere with the production of dozens or hundreds of genes.
15

Microfluidic technology provides a potentially elegant solution to the above problems. By bonding a fluidic device fabricated with chambers, microchannels, and other functional components directly to an oligo microarray substrate, gene synthesis could be confined to relatively small volumes isolating specific sequences for assembly.
Thus, as shown in Figure 5, each chamber of the fluidic would enclose oligo spots designed to assemble into a desired gene or gene segment. Solutions containing the reagents necessary for single-step PCA and cleavage of oligos from the microarray surface would be introduced into each chamber, whereupon, in the case of restriction enzyme digestion, enzymes would cleave the anchored microarray oligos, releasing them in solution for subsequent gene assembly. This process is shown schematically in Figure
6 for a restriction enzyme requiring a double-stranded recognition sequence for cleavage.
Microfluidic devices mounted on glass slides have been fabricated with sufficient density and resolution to readily enclose the necessary microarray chamber areas, which would
Figure 5 . Schematic hybrid microfluidic-microarray device. Each chamber isolates a distinct set of oligonucleotides (orange spots) to be cleaved and assembled into a different gene.
16

be on the order of 10 5  m 2 for a chamber containing 16 oligo spots, sufficient to build a modest size gene, as laid out on a commercially available DNA microarray. Successful direct synthesis of genes from microarray oligos would represent a significant advancement in the goal of facilitating widely available, inexpensive synthetic DNA constructs.
Figure 6. An enzymatic cleavage method to release construction oligos from the surface of the microarray. Initially (stage 1) the oligos are anchored by their 3’ ends
(arrowheads) to the array surface. Enzymes (blue circles) and helper oligos (short orange arrows) are also in solution. Helper oligos hybridize to complementary regions, indicated in orange, on the anchored oligos (stage 2). The restriction enzyme binds to the recognition site (stage 3) and cleaves the oligo (stage 4).
In all of the devices described thus far, upon synthesizing genes and other DNA constructs, samples would normally be retrieved from the device for subsequent bench top processes such as protein synthesis. However, just as the concept of integration proved revolutionary in the field of computing, similarly, the integration of multiple biological processes in a discrete device is a goal highly sought after by researchers. In this proposed application, upon completion of gene synthesis, a microfluidic module capable of protein synthesis would be located downstream, thus describing a device
17

capable of integrated gene and protein synthesis. The architecture for such a device is shown in Figure 7; fluid channels are shown in light blue and valves are shown in dark blue. Here, gene synthesis could be conducted in a series of initial chambers with inlets labeled 1-3 (orange spots represent distinct oligo sequences on the surface of a microarray; for initial experiments, reactor chambers similar to those described in section
4.1 and Figure 2 would be sufficient), followed by the loading of each gene product into a subsequent chamber where a commercially available protein expression mixture is introduced through inlet 4. Upon completion of mixing (initiated by a series of three valves), incubation at the appropriate temperature would occur thus initiating the expression of the desired protein. For the schematic device shown in Figure 7, three distinct fluorescent proteins (red, yellow, and green) are expressed.
Figure 7. Architecture for device capable of conducting three parallel gene syntheses followed by protein expression. Orange spots represent oligos to be cleaved from the microarray surface for gene synthesis and blue rectangles represent valves and pumps.
Numbered circles represent inlets. Protein expression in the incubation chambers gives rise to observable fluorescence for different proteins DsRed (red), YFP (yellow) and
GFP (green).
Such a result would not only demonstrate integration of distinct biological processes, but for this specific application would start to simulate the activity of a
18

biological cell, as given the input of oligo fragments and associated reagents that might be found in cytoplasm (individual nucleotides, enzymes, etc.), the output is expressed protein.
Results for the proposed experimental areas will be evaluated primarily via standard molecular biology processes. For example, the success or failure of the microfluidic devices described in sections 4.1 to 4.3 will be determined by techniques such as polyacrylamide gel electrophoresis (PAGE) and DNA sequencing, which verify the length and sequence information of synthesized DNA molecules, respectively.
Because the length and sequence of the desired synthesis products are known prior to conducting reactions in the microfluidic device, techniques such as PAGE and DNA sequencing will be central to evaluating the output of the various proposed devices.
For example, to date the parallel synthesis of various gene products has already been conducted, and the synthesis products of the microfluidic device described in section 4.1 have been analyzed via both PAGE and DNA sequencing. Figure 8 shows two PAGE gels demonstrating the synthesis of DNA molecules of the appropriate length, along with negative controls, while the tabulation and comparison of synthesis errors in a microfluidic device and a standard PCR tube, as determined by DNA sequencing, are shown in Table I. These results are published in Kong et al
25
.
The integrated gene and protein synthesis device proposed in section 4.4 will require more instrumentation. Depending upon the proteins synthesized, reactions could be again evaluated via gel electrophoresis of protein, or, in the case of fluorescent
19

proteins, additionally via microscopy. For example, the device architecture proposed in figure 7 could be monitored using microscopy techniques, with successful synthesis of the desired proteins yielding red, yellow, and green fluorescence, respectively.
Figure 8. Polyacrylamide gel electrophoresis (PAGE) showing successful parallel synthesis of genes along with negative controls. In the presence of construction oligos, DNA constructs GFP and dsRed (993 and 733 bp respectively, figure 8A) and
OR128-1 and ble (942 and 461 bp respectively, figure 8B) are synthesized and amplified. Without construction oligos, no product bands are generated. Molecular weight markers are shown (M) with 500, 750, and 1000 bp positions indicated.
Table I. Summary of errors for synthesis of gene DsRed in a microfluidic device as compared to in a standard PCR-tube.
Error Type
Microfluidic
Device
PCR tube
Deletion Single-base 19 16
Deletion Multiple-base 5 5
Transition G/C to A/T 3 6
Transition A/T to G/C 0 2
Transversion G/C to C/G
Transversion G/C to T/A
Transversion A/T to C/G
Transversion A/T to T/A
Total Errors:
Bases Sequenced:
Error Rate (per base):
0
1
1
0
29
16,250
0.0018
0
1
0
0
30
13,389
0.0022
20

The device and goals described in Section 4.1 (parallel gene synthesis in a microfluidic device) have already been achieved, and are published in Kong et al
25
. The remaining extensions of this foundational work are currently being pursued in parallel, with the hopeful achievement of either single-stage hierarchical synthesis, direct synthesis of a modest sized gene from a DNA microarray, or the integrated on-chip synthesis of gene then protein, by Spring ’07 or Summer ’07.
Resources required for this work include equipment and laboratory supplies currently available in the labs of Joseph Jacobson, Scott Manalis, Shuguang Zhang, the
BioMicro Center at MIT, and the Microtechnology Labs (MTL) at MIT.
21

David Sun Kong is the son of two Chinese immigrants. He was born and raised in the Boston area by a loving and supportive family.
22

1.
Khorana, H.G. (1968) Nucleic acid synthesis in the study of the genetic code, in
Nobel Lectures: Physiology or Medicine (1963-1970). Elsevier Science Ltd,
Amsterdam, pp. 341-369.
2.
The MGC Project Team (2004) The status, quality, and expansion of the NIH full-length cDNA project: the mammalian gene collection (MGC). Genome Res.,
14 , 2121-2127.
3.
Elowitz, M.B. and Leibler, S. (2000) A synthetic oscillatory network of transcriptional regulators. Nature , 403 , 335-338.
4.
Martin, V.J., Pietera, D.J., Withers, S.T., Newman, J.D. and Keasling, J.D. (2003)
Engineering a mevalonate pathway in Escherichia coli for production of terpenoids. Nat. Biotechnol.
, 21 , 796-802.
5.
Cello, J., Paul, A.V., and Wimmer, E. (2002) Chemical synthesis of poliovirus cDNA: generation of infectious virus in the absence of natural template .
Science ,
297 , 1016-1018.
6.
Smith, H.O., Hutchison, C.A., 3rd, Pfannkoch, C., and Venter, J.C. (2003)
Generating a synthetic genome by whole genome assembly: phiX174 bacteriophage from synthetic oligonucleotides .
Proc Natl Acad Sci U S A , 100 ,
15440-15445.
7.
Hutchison, C. A., III, Peterson, S. N., Gill, S. R., Cline, R. T., White, O., Fraser,
C. M., Smith, H. O. and Venter, J. C. (1999) Global Transposon Mutagenesis and a Minimal Mycoplasma Genome. Science 286, 2165–2169.
8.
Caruthers, M.H. (1985) Gene synthesis machines: DNA chemistry and its uses.
Science , 230 , 281-285.
9.
Mullis, K., Faloona, F., Scharf, S., Saiki, R., Horn, G., and Erlich, H. (1986)
Specific enzymatic amplification of DNA in vitro: the polymerase chain reaction.
Cold Spring Harb Symp Quant Biol , 51, 263-273.
10.
Stemmer, W.P., Crameri, A., Ha,K.D., Brennan, T.M. and Heyneker, H.L. (1995)
Single-step assembly of a gene and entire plasmid from large numbers of oligonucleotides. Gene , 164 , 49-53.
23

11.
Tian, J., Gong, H., Sheng, N., Zhou, X., Gulari, E., Gao, X., and Church, G.
(2004) Accurate multiplex gene synthesis from programmable DNA microchips.
Nature , 432 , 1050-1054.
12.
Xiong, A.S., Yao, Q.H., Peng, R.H., Duan, H., Li, X., Fan, H.Q., Cheng, Z.M., and Li, Y.
(2006 ) PCR-based accurate synthesis of long DNA sequences. Nature
Protocols , 1 , 791-797.
13.
Kodumal, S.J., Patel, K.G., Reid, R., Menzella, H.G., Welch, M., and Santi, D.V.
(2004) Total synthesis of long DNA sequences: synthesis of a contiguous 32-kb polyketide synthase gene cluster .
Proc Natl Acad Sci USA, 101, 15573-15578.
14.
Carr, P.A., Park, J.S., Lee, Y.J., Yu, T., Zhang, S. and Jacobson, J.M. (2004)
Protein-mediated error correction for de novo DNA synthesis. Nucleic Acids Res.,
32 , e162.
15.
Binkowski, B.F., Richmond, K.E., Kaysen, J., Sussman, M.R., and Belshaw, P.J.
(2005) Correcting errors in synthetic DNA through consensus shuffling. Nucleic
Acids Res.
, 33 , e55.
16.
Fuhrmann, M., Oertel, W., Berthold, P., and Hegemann, P. (2005) Removal of mismatched bases from synthetic genes by enzymatic mismatch cleavage.
Nucleic Acids Res.
, 33 , e58.
17.
Liu, J., Hansen, C., and Quake, S.R. (2003) Solving the “world-to-chip” interface problem with a microfluidic matrix. Anal. Chem.
, 75 , 4718-4723.
18.
Terry SC, et al. (1979) A gas chromatographic air analyzer fabricated on silicon wafer. IEEE Trans Electron Devices , 26 , 1880-6.
19.
Xia, Y., and Whitesides, G.M. (1998) Soft Lithography. Ang Chem Int Ed, 37 ,
550-575.
20.
Unger, M.A., Chou, H.P., Thorsen, T., Scherer, A., and Quake, S.R. (2000)
Monolithic microfabricated valves and pumps by multilayer soft lithography.
Science. 288 , 113-6.
21.
Hansen, C.L., Skordalakes, E., Berger, J.M., and Quake, S.R. (2002) A robust and scalable microfluidic metering method that allows protein crystal growth by free interface diffusion. PNAS , 99 , 16531-16536.
22.
Hong, J.W., Studer, V., Hang, G., Anderson, W.F., and Quake, S.R. (2004) A nanoliter-scale nucleic acid processor with parallel architecture. Nat Biotechnol ,
22 , 435-439.
24

23.
Richmond, K.E., Li, M.H., Rodesch, M.J., Patel, M., Lowe, A.M., Kim, C., Chu,
L.L., Venkataramaian, N., Flickinger, S.F., Kaysen, J., Belshaw, P.J., Sussman,
M.R., and Cerrina, F. (2004) Amplification and assembly of chip-eluted DNA
(AACED): a method for high-throughput gene synthesis. Nucleic Acids Res., 32 ,
5011-5018.
24.
Zhou, X., Cai, S., Hong, A., You, Q., Yu, P., Sheng, N., Srivannavit, O.,
Muranjan, S., Rouillard, J.M., Xia, Y., Zhang, X., Xiang, Q., Ganesh, R., Zhu, Q.,
Matejko, A., Gulari, E., and Gao, X. (2004) Microfluidic PicoArray synthesis of oligodeoxynucleotides and simultaneous assembling of multiple DNA sequences.
Nucleic Acids Res ., 32 , 5409-5417.
25.
Kong, D.S., Carr, P.A., Chen, L., and Jacobson, J.M. (2007) Parallel gene synthesis in a microfluidic device. Nucleic Acids Res., in press.
25