PAPER

University of California, Berkeley, College o

Effect of PDMS-based Microdevice Channel Width on Plasmid DNA

Transformation Efficiency

Albert Peng,a Simrunn Girn,

a Regine Labog,

Submitted 9th December 2010

The effect of PDMS-based microdevice channel width on GFP plasmid transformation efficiency 5

in E. coli was studied in this project. Four different device designs consisting of 50 µm, 100 µm,

250 µm, and 500 µm channel widths were used in conjunction with s

soft lithography fabrication techniques to create PDMS microdevices. Multiple chemical

transformation trials using optimal macroscale heat shock parameters

devices, and data was collected from agar plate cultures and subsequently analyzed by the ImageJ 10

software package. Although we have successfully demonstrated chemical transformation in a

microscale environment, our data suggests that variability in transformation efficiency introduced

by experimental error is large enough such that any potential influence channel width may have on

transformation efficiency is masked.

Introduction 15

Plasmid DNA transformation is a key molecular biology

concept of introducing new functionality to existing bacteria

strains by importing desired DNA molecules into cells. DNA

transformation in E. coli is generally accomplished by

chemical and electrical means, and various studies have been 20

performed to maximize transformation efficiency for both

methods. While there are advantages and disadvantages to

both techniques, chemical transformation is cheaper and more

accessible than electroporation and is the mai

study. 25

Although heat shock chemical transformation is widely

used and accepted,6 it is relatively unclear how it functions.

During chemical transformation, it is theorized that

in a ����� and E. coli solution envelop the cell membrane

thus producing a net positive charge on the surface, attracting 30

the negatively charged plasmid DNA.1,3,4 A heat shock step

then opens pores on the cell surface and facilitates passage

through the cell membrane due to the close proximity of the

plasmid DNA to the cell. An ice incubation step is thought to

reduce the thermal motion of the DNA and allow further 35

binding to the cell membrane.1 Finally a warm

incubation in rich LB media allows the cells to recover from

the previous disturbances to cellular processes and promotes

survivability of the culture. In addition, this incubation step

could allow further uptake of plasmid into the cell as sort of a 40

second heat shock step.1

Traditional transformation optimization stud

always been done at macroscale.5 Applications such as

genomic and cDNA library construction typically require

transformation with low DNA copy, so it is necessary to find 45

parameters that maximize transformation efficiency.

transformation has been shown to be possible at microscale,

the influence of channel width on transformation efficiency

has never been studied. Since the exact mechanism of plasmid

DNA uptake in E. coli during chemical transformation is 50

unknown, it is important to study all the possible parameters

University of Calfornia, Berkeley

of Engineering 2010

based Microdevice Channel Width on Plasmid DNA

Efficiency in E. coli

Regine Labog,a and Yiqing Zhao

a

based microdevice channel width on GFP plasmid transformation efficiency

was studied in this project. Four different device designs consisting of 50 µm, 100 µm,

250 µm, and 500 µm channel widths were used in conjunction with standard photolithography and

soft lithography fabrication techniques to create PDMS microdevices. Multiple chemical

transformation trials using optimal macroscale heat shock parameters1 were performed with these

ate cultures and subsequently analyzed by the ImageJ

software package. Although we have successfully demonstrated chemical transformation in a

microscale environment, our data suggests that variability in transformation efficiency introduced

l error is large enough such that any potential influence channel width may have on

Plasmid DNA transformation is a key molecular biology

concept of introducing new functionality to existing bacteria

strains by importing desired DNA molecules into cells. DNA

is generally accomplished by

ans, and various studies have been

performed to maximize transformation efficiency for both

methods. While there are advantages and disadvantages to

both techniques, chemical transformation is cheaper and more

accessible than electroporation and is the main focus of this

Although heat shock chemical transformation is widely

it is relatively unclear how it functions.

During chemical transformation, it is theorized that ���� ions

solution envelop the cell membrane

thus producing a net positive charge on the surface, attracting

A heat shock step

then opens pores on the cell surface and facilitates passage

close proximity of the

plasmid DNA to the cell. An ice incubation step is thought to

reduce the thermal motion of the DNA and allow further

Finally a warm 37�

incubation in rich LB media allows the cells to recover from

previous disturbances to cellular processes and promotes

survivability of the culture. In addition, this incubation step

could allow further uptake of plasmid into the cell as sort of a

Traditional transformation optimization studies have almost

Applications such as

genomic and cDNA library construction typically require

transformation with low DNA copy, so it is necessary to find

parameters that maximize transformation efficiency.1 While

tion has been shown to be possible at microscale,2

the influence of channel width on transformation efficiency

has never been studied. Since the exact mechanism of plasmid

during chemical transformation is

study all the possible parameters

that may affect transformation efficiency, such as the channel

size in which transformation occurs. For our study a standard 60

chemical transformation procedure is used with chemically

competent E. coli cells and GFP. While

are specific to the strain of E. coli and GFP plasmid used in

these experiments, the results gathered could be useful for

future work with other strains of bacteria and plasmids.65

Materials and Methods 60

Device Fabrication

The design of our 50 µm, 100 µm, 250 µm, and 500 µm 75

channel width devices was drawn using AutoCAD and sent to

an external manufacturer to produce mylar masks for

photolithography (Figure 1). When designing our device we

needed to incorporate three functions: an inlet fo

GFP loading, a heat shock chamber with the required channel 80

dimensions, and an outlet to collect the pool. S

chosen for the transformation chamber to maximize volume

while maintaining the designated channel widths. In addition,

the devices were designed to have identical volumes of 3.4 µL

each in order to make channel width the only varying factor 85

between devices. This resulted in fewer S

channel width devices compared with the smaller channel

width devices.

Fig. 1 A 50 µm channel width device design

University of Calfornia, Berkeley | BioE 121L

Bioengineering | 1

based Microdevice Channel Width on Plasmid DNA

that may affect transformation efficiency, such as the channel

size in which transformation occurs. For our study a standard

chemical transformation procedure is used with chemically

cells and GFP. While the results we obtain

and GFP plasmid used in

these experiments, the results gathered could be useful for

future work with other strains of bacteria and plasmids.

our 50 µm, 100 µm, 250 µm, and 500 µm

channel width devices was drawn using AutoCAD and sent to

an external manufacturer to produce mylar masks for

photolithography (Figure 1). When designing our device we

needed to incorporate three functions: an inlet for E. coli and

GFP loading, a heat shock chamber with the required channel

dimensions, and an outlet to collect the pool. S-curves were

chosen for the transformation chamber to maximize volume

while maintaining the designated channel widths. In addition,

e devices were designed to have identical volumes of 3.4 µL

each in order to make channel width the only varying factor

between devices. This resulted in fewer S-curves for the larger

channel width devices compared with the smaller channel

A 50 µm channel width device design

2 | Bioengineering

Fig. 3 Experimental procedure used for transformation

A standard contact photolithography procedure with

negative SU-8 2035 photoresist was then done using the

previously created mylar mask and a 4” silicon wafer. Contact

photolithography was used to keep production costs low while 5

maintaining high resolution of features. Spin coating

parameters were chosen to create a single final photore

height of 50 µm for all devices (Figure 2), and the appropriate

UV exposure times were chosen to accommodate the contact

aligner measured UV intensity, which is variable with the age 10

and quality of the UV bulb. After the necessary heat,

developing and cleaning treatments, the wafer is then placed

into a vacuum chamber for silanizing. Silanizing the wafer

allows cured PDMS to be more readily removable from the

surface of the wafer and is essential for soft lithography.15

Fig. 2 Close-up channel dimensions of a 50 µm device

The silanized wafer with all the device features was then

used as a mold for PDMS soft lithography. A 10:1 ratio of

base to curing agent was weighed out and thoroughly mixed,

resulting in a 50 g base: 5 g curing agent mixture. This 20

solution was degassed by vacuum and then poured over the

clean wafer, and allowed to cure overnight on a

plate. Once cured, the PDMS layer was carefully peeled off

the wafer. Individual devices were cut out from the PDMS

sheet and 1 mm holes were punched at the inlet and outlet. 25

Devices and microscope glass slides were then both tape

cleaned and chemically cleaned by acetone, IPA and DI water,

and subjected to UVO treatment to modify the surface

chemistry to facilitate bonding. The PDMS devices and sli

were then bonded together to produce a final useable 30

microdevice.

Experimental Procedure

After our devices were created, we began running chemical

transformation trials (Figure 3). 20 µL of chemically

University of California, Berkeley, College

Experimental procedure used for transformation

A standard contact photolithography procedure with

was then done using the

previously created mylar mask and a 4” silicon wafer. Contact

photolithography was used to keep production costs low while

maintaining high resolution of features. Spin coating

parameters were chosen to create a single final photoresist

, and the appropriate

UV exposure times were chosen to accommodate the contact

aligner measured UV intensity, which is variable with the age

and quality of the UV bulb. After the necessary heat,

cleaning treatments, the wafer is then placed

into a vacuum chamber for silanizing. Silanizing the wafer

allows cured PDMS to be more readily removable from the

surface of the wafer and is essential for soft lithography.

s of a 50 µm device

The silanized wafer with all the device features was then

used as a mold for PDMS soft lithography. A 10:1 ratio of

base to curing agent was weighed out and thoroughly mixed,

resulting in a 50 g base: 5 g curing agent mixture. This

ution was degassed by vacuum and then poured over the

clean wafer, and allowed to cure overnight on a 55� hot

plate. Once cured, the PDMS layer was carefully peeled off

the wafer. Individual devices were cut out from the PDMS

hed at the inlet and outlet.

Devices and microscope glass slides were then both tape

cleaned and chemically cleaned by acetone, IPA and DI water,

and subjected to UVO treatment to modify the surface

chemistry to facilitate bonding. The PDMS devices and slides

were then bonded together to produce a final useable

After our devices were created, we began running chemical

. 20 µL of chemically

competent E. coli was thawed on ice for 30 minutes50

which 2 µL GFP plasmid obtained through miniprep was

added and mixed by gentle tapping. After incubating on ice

for 30 minutes, 5 µL of this solution was vacuum loaded into

each device. Vacuum loading was done on ice until the entire

device was loaded. The devices were then placed on a hot 55

plate set at 42� for 30 seconds as monitored by a

thermocouple, and then placed on ice for 2 minutes. A syringe

was placed at the inlet of each device and used air pressure to

evacuate the device of bacteria, and pool was collected at the

outlet and incubated in 50 µL LB-Amp media for 1 hour. The 60

appropriate dilutions were made and the culture was plated on

agar-Amp plates and allowed to grow overnight. Pictures were

taken the following day and colonies were counte

ImageJ software.

Results and Discussion

Prior to performing any transformation experiments, we

attempted to vacuum load our devices with

that vacuum loading is a viable technique to introduce

solutions into microdevices. A picture using phase contrast 65

microscopy was taken demonstrating successful vacuum

loading (Figure 4). A total of 28 devices were then

successfully used in transformation runs and had enough

colonies on their corresponding plates to be counted.

Transformation efficiency is determined quantitatively as the 70

total colony count on each plate, with higher counts equating

to higher transformation efficiencies.

Fig. 4 Phase microscopy image of E. coli loaded in a 50 µm device

The first set of experiments we tried to perform included: 3 65

x 50 µm, 3 x 100 µm, 3 x 250 µm, and 3 x 500 µm channel

e of Engineering 2010

was thawed on ice for 30 minutes, after

which 2 µL GFP plasmid obtained through miniprep was

added and mixed by gentle tapping. After incubating on ice

for 30 minutes, 5 µL of this solution was vacuum loaded into

each device. Vacuum loading was done on ice until the entire

ded. The devices were then placed on a hot

for 30 seconds as monitored by a

thermocouple, and then placed on ice for 2 minutes. A syringe

was placed at the inlet of each device and used air pressure to

pool was collected at the

Amp media for 1 hour. The

appropriate dilutions were made and the culture was plated on

Amp plates and allowed to grow overnight. Pictures were

taken the following day and colonies were counted by the

Prior to performing any transformation experiments, we

attempted to vacuum load our devices with E. coli to show

that vacuum loading is a viable technique to introduce

solutions into microdevices. A picture using phase contrast

microscopy was taken demonstrating successful vacuum

). A total of 28 devices were then

ormation runs and had enough

colonies on their corresponding plates to be counted.

Transformation efficiency is determined quantitatively as the

total colony count on each plate, with higher counts equating

loaded in a 50 µm device

The first set of experiments we tried to perform included: 3

x 50 µm, 3 x 100 µm, 3 x 250 µm, and 3 x 500 µm channel

University of California, Berkeley, College of Engineering 2010 Bioengineering | 3

width devices. We wanted to do three runs of each channel

width in order to average the data from all three and generate

more reliable results. Out of these runs only: 1 x 50 µm, 2 x

100 µm, 2 x 250 µm, and 2 x 500 µm devices were able to

generate any measurable data (Figure 5). Some devices were 5

not able to load completely in a reasonable amount of time

and had to be discarded. In addition, our initial batch of 50

µm channel width devices were not bonded very well to the

glass slides, and popped off when we attempted to use air

pressure to empty the device of E. coli. 10

Fig. 5 Colony count data gathered from the first set of transformations

The data generated using these devices shows that colony

count decreases as channel width increases, since the 100 µm

devices had an average of 900 colonies while the 250 µm and

500 µm devices had an average of 800 and 600 colonies, 15

respectively. This suggests that smaller channel widths

coincide with higher transformation efficiency. However, due

to the low number of successful trials for each device, we

decided to do more transformations in order to confirm our

findings. 20

For the second set of transformation runs we wanted to see

if there was a legitimate difference in transformation

efficiency between smaller and larger channel widths. Since

our data from the first set of runs was relatively sparse due to

experimental error, we decided that we should only focus on 25

two channel widths and make sure that we believe our results.

We ran trials with 4 x 100 µm and 4 x 250 µm devices in the

same fashion as the first set of runs and gathered the colony

data (Figure 6). The data shows that the average colony

number from the 100 µm and 200 µm devices are 1000 and 30

1200 respectively, which is in direct contradiction of the trend

observed in the first set of runs. This new data suggests that

there is relatively little difference between the transformation

efficiency of the 100 µm and 200 µm channel width devices.

Judging from the extreme variability of the individual trials in 35

the second run (1500 colonies in trial 1 and 400 colonies in

trial 4 of the 250 µm set), it appeared that our experimental

methods were still unable to generate consistent results.

Fig. 6 Colony count data gathered from the second set of 40

transformations

In a last attempt to obtain coherent data, we performed a

third and final set of transformations. For these trials we used:

4 x 50 µm, 4 x 100 µm, 4 x 250 µm, and 4 x 500 µm channel

width devices. The 50 µm and 500 µm channel widths 45

performed the best at an average of 250 and 300 colonies

respectively, while the 100 µm and 250 µm channel widths

had 100 and 180 colonies each (Figure 7). Unfortunately this

data still does not agree with our previous runs, and we must

end this project with inconclusive results. 50

Fig. 7 Colony count data gathered from the third set of transformations

Different dilution factors were used for each run prior to

plating, so the colony counts between runs are very different

in our data. However, only the relative difference in colony

counts between individual devices within runs matters, and 55

from the three sets of runs that we performed, there was no

clear trend indicating the effect of channel width on

transformation efficiency. One reason for this could be due to

experimental error. The transformation has been shown to be

very robust even at a 10x dilution factor across all device 60

widths, indicating that slight errors in experimental procedure

such as inexact transfer volumes can result in high variability

4 | Bioengineering University of California, Berkeley, College of Engineering 2010

in colony counts. For example trial 1 of the 50 µm device in

run 2 had 600 colonies while trial 2 of the same device in the

same run had only 100 colonies, even though they both

experienced a 10x dilution before plating. Any effect that

channel width may have had on these colony counts would 5

have been masked by the extreme variability introduced by

experimental error.

Conclusion

Colony count data collected from three separate runs of

multiple transformation trials did not reveal a clear trend 10

between microdevice channel width and transformation

efficiency. Transformation was robust amongst all devices

even at high dilution factors, suggesting that the effect of

channel width is small compared to the inherently high

transformation efficiency. Variability in colony counts 15

introduced due to experimental error also contributed to the

inability to generate consistent data. Due to limitations in our

original device design and time constraints we must end this

project with inconclusive results. Future work can be done to

improve both device design and the experimental procedure 20

by performing everything on-chip, to minimize compounding

errors due to inexact off-chip activities such as E. coli

evacuation from the device, dilution factors, and inconsistent

plating technique.

References 25

a College of Engineering, Bioengineering Department, University of

California, Berkeley,CA, 94704, USA.

1 Mahipal Singh, Arpita Yadav, Xiaoling Ma and Eugene Amoah.

Plasmid DNA Transformation in Escherichia Coli: Effect of Heat

Shock Temperature, Duration, and Cold Incubation of CaCl2 Treated 30

Cells. International Journal of Biotechnology and Biochemistry,

Volume 6 Number 4 (2010) pp. 561–568.

2 Sha Li, L. Meadow Anderson, Jui-Ming Yanga, Liwei Lin, Haw

Yang. DNA transformation via local heat shock. APPLIED

PHYSICS LETTERS 91, 2007. 35

3 W. Edward Swords. Chemical Transformation of E. coli. Methods in

Molecular Biology, 2003, Volume 235, 49-53, DOI: 10.1385/1-

59259-409-3:49.

4 Dagert M, Ehrlich SD. Prolonged incubation in calcium chloride

improves the competence of escherichia coli cells. Gene. 1979 40

May;6(1):23-8.

5 Huff JP, Grant BJ, Penning CA, Sullivan KF. Optimization of routine

transformation of escherichia coli with plasmid DNA.

BioTechniques. 1990 Nov;9(5):570,2, 574, 576-7.

6 Bergmans HE, van Die IM, Hoekstra WP. Transformation in 45

escherichia coli: Stages in the process. J Bacteriol. 1981

May;146(2):564-70.