Design and Optimization of Fluorophore- Tagged

Design and Optimization of Fluorophore-

Tagged Oligonucleotides for FRET Detection in

Optofluidic ARROW Waveguides

Melissa Eberle Bioengineering Senior Thesis

2

Table of Contents:

ABSTRACT ........................................................................................................... 4

1 INTRODUCTION ............................................................................................. 5

2 BACKGROUND............................................................................................... 5

2.1 Optofluidic Devices Improve The Monitoring of Molecular Dynamics ......................... 5

2.2 FRET and Fluorophores ..................................................................................................... 6

2.3 Liquid Core ARROW .......................................................................................................... 8

3 THE DETECTION CHIP AND SYSTEM ............................................ 10

3.1 Liquid Core ARROW Characterization .......................................................................... 10

3.2 Detection System Setup ..................................................................................................... 11

4 BIOMOLECULE RESEARCH ................................................................ 14

4.1 A Biomolecule with Controlled Fluorophore Radial Distance ...................................... 14

4.2 Fluorophore Specifications ............................................................................................... 15

4.3 Oligonucleotide and Fluorophore Design ........................................................................ 15

4.4 Improving The Oligonucleotides Annealing Efficiency ................................................. 18

4.5 Improve Observed FRET Efficiency of Oligonucleotides .............................................. 20

5 CONCLUSION ............................................................................................... 24

ACKNOWLEDGMENTS ................................................................................ 26

REFERENCES ................................................................................................... 26

APPENDIX ........................................................................................................... 26

A1: Oligonucleotide Company ................................................................................................ 26

3

A2: Oligonucleotide Annealing Protocol ............................................................................... 26

A3: Exonuclease I Protocol ..................................................................................................... 27

A4: Purification Protocol ........................................................................................................ 28

4

Abstract

Optofluidics are key in analyzing single molecular dynamics. The Applied Optics lab is

working towards developing an optofluidic device integrated with the liquid core Anti-

Resonant Reflecting Optical Waveguide (ARROW) to quantify the Fluorescence Resonance

Energy Transfer (FRET) of single biomolecules. I helped design and optimize a biomolecule

with an efficient FRET output that would be compatible with and testable in the ARROW, to

help develop and test the optical detection system that would monitor FRET of single

biomolecules. I used DNA oligonucleotides as the biomolecule because they are inexpensive,

robust and can be ordered with the fluorophores attached. Two different fluorophore pairs were

chosen, Fluorescein/ Cy3 and Cy3/ Cy5, and oligonucleotide complementary strands were

designed for each. The melting temperatures for the two fluorophore-tagged oligonucleotides

were 54.6oC for the Fluor./ Cy3 oligonucleotide and 54.8

oC for the Cy3/ Cy5 oligonucleotide.

The melting temperature should be around 50-60oC for helical stability. Based on the 1nm

radial distance between fluorophores that the DNA helix creates, the expected FRET efficiency

was ~99%. FRET testing started with the annealed Fluor./ Cy3 oligonucleotide. Unfortunately,

do to Fluorescein’s degradation, we were unable to obtain any FRET data. We then switched to

the Cy3/ Cy5 pair and measured a FRET efficiency of 60%. The efficiencies were calculated

using the photobleaching effect of the acceptor fluorophore. To improve the efficiency, I used

the Exonuclease I from E. Coli to removed the ssDNA from the sample because the free

fluorophores were contributing to the donor photon count. With the Exonuclease I treated Cy3/

Cy5 sample, we were able to obtain a FRET efficiency of 91%, which is only 8% below the

ideal expected efficiency. Not only did I work on the oligo. efficiency, but I also worked on

ARROW characterization, determining the waveguides’ liquid and solid core throughputs and

calculating its transmittances, ensuring the waveguide had a high enough signal to noise ratio

for data collection.

5

1 Introduction

As biology advances, the need to better understand the molecular dynamics of

individual biomolecules becomes increasingly important. Biological tools have been created to

help monitor these molecular dynamics. One tool is Fluorescence Resonance Energy Transfer

(FRET), which can be used to monitor protein or enzyme conformational changes by being

dependent on the distance two fluorophores are labeled. Because of the need to monitor the

dynamics of single-molecules, the need for small volume devices that can measure FRET

grows. There are many different devices known as “labs-on-a-chip” that handle micro-liter

volumes, but only a few can measure fluorescence. These devices are called optofluidic

devices, which fully integrate optics, electronics, and micro-fluidics on the plane of a single

chip.

The Applied Optics Lab, under Holger Schmidt, is working toward creating an

optofluidic device using a liquid core Anti-Resonant Reflecting Optical Waveguide (ARROW)

chip that can monitor single-molecule dynamics. My work in this project was to help design

and optimize a biomolecule, with an efficient FRET output that would be compatible with and

testable in the ARROW as a positive control to help develop and test the optical detection

system that would monitor FRET of single biomolecules. I used DNA oligonucleotides, which

are short nucleic acid polymers with twenty or fewer bases, as the biomolecule because they

are inexpensive, robust and can be ordered with the fluorophores attached. Also

oligonucleotides were used because once annealed, they can bring and hold two fluorophores

within FRET distance (1-10nm). Fluorophores with a high energy transfer and high quantum

yields were used. I aimed to perfect the oligonucleotide helical preparation to create a FRET

efficiency as close to the expected 99% that the detection system would allow to prove our

system can accurately detected FRET.

2 Background

2.1 Optofluidic Devices Improve The Monitoring of Molecular Dynamics

Optofluidics and its implementation with chip-scale devices have enabled the control

and manipulation of small volumes of liquids reliably and rapidly [3]. Volumes for a single

experiment can be in the nano to picoliter range enabling the analysis of single-molecules. This

6

also reduces process time and the amount of reagents necessary, substantially reducing costs

[4]. Optofluidics holds great promise for novel devices for many biomedical instrumentation

applications that deal with biomolecules. The ribosome, for example, which has been

successfully tagged with fluorophores allowing for FRET detection of conformational changes,

can be used in these devices and the dynamics of the changes can be quantified (Fig. 1) [2].

Figure 1: 70S ribosome tagged with two fluorophores, Cy3 and Cy5 [2]. When the ribosome is in a nonrotated conformation the two fluorophores are within FRET distance and the energy is transferred from the donor (Cy3) to the acceptor (Cy5), as can be seen in the green and red intensity graph. When the ribosome rotates into the hybrid state the fluorophores become too far apart for FRET and the energy transfer is lost, as can be seen in the blue FRET graph. Figure copied from Cornish [2].

Thus, optofluidic devices are key to monitoring molecular mechanics on a chip because

they are able to analyze molecules at the ultimate sensitivity limit that has previously required

bulky and costly microscopy [3] and can monitor the mechanics at a single-molecule resolution

due to the small volume level.

2.2 FRET and Fluorophores

A powerful tool used to study the dynamics of biomolecules is Fluorescence Resonance

Energy Transfer (FRET). FRET is the energy transfer between two fluorophores: a donor and

an acceptor. A donor fluorophore has an emission spectrum that overlaps with the acceptor

fluorophore’s excitation spectrum (Fig. 2). When the donor is excited and is within a 1-10nm

7

range of the acceptor, the energy from the donor’s emission is transferred to the acceptor and

acts as the acceptor’s exciter. The acceptor’s emission wavelength is then observed.

Figure 2: The above graph visually represents the spectral overlap of the donor and the acceptor

fluorophore. The green shaded area represents the spectrum where energy transfer would occur, which is the crossing of the Cy3 red emission line and the Cy5 blue excitation line. The yellow block and the red block represent the excitation and emission wavelengths respectively that are filtered in order to detect FRET. Figure copied from www.biotek.com.

The efficiency of energy transfer, E, is given as E = (1 + (R/Ro) 6)

-1, where R is the

inter-dye distance and Ro is the Förster radius at which E = 0.5 (Fig. 3) [8].

Figure 3: Graphical representation of FRET efficiency, E, as a function of inter-dye distance (R)

for Ro = 50Å. At R = Ro, E = 0.5, or 50% efficiency, but at smaller distances, it is >0.5 and at larger distances, it is <0.5, as can be seen by the blue line function [8]. Figure copied from Roy [8].

The radial distance of the two fluorophores must be close enough for efficient FRET, thus the

biomolecule used must allow for close proximity fluorophore labeling and be able to maintain

this close distance, like in the ribosome.

8

When performing FRET experiments, a phenomenon known as photobleaching must be

taken into consideration. Photobleaching is the destruction of a fluorophore due to

overexposure from its excitation wavelength. The power of the excitation light must be

monitored to reduce the destruction of the fluorophores. However, photobleaching can be used

to calculated FRET efficiency. This is done in two steps. First, the normal donor and acceptor

intensities are collected and then the sample is photobleached at the acceptors’ excitation

wavelength, destroying the acceptor fluorophores. The new donor and acceptor intensities are

collected and the efficiency is calculated using the following formulas:

IA: acceptor intensity, ID: donor intensity

The changes in the intensities of the acceptor and donor fluorophores before and after

photobleaching show the efficiency with which the fluorophores are coupled and thus the

FRET efficiency [8].

2.3 Liquid Core ARROW

The liquid core ARROW is a hollow-core optical waveguide, which allows for the

integration of optics with microfluidics on a single chip with planar optical and fluidic

architecture. Figure 4 is a cartoon representation of the waveguide. The LC-ARROW has two

cores: a liquid core and a solid core that are perpendicular to each other. The excitation

wavelength is propagated through the solid core and the fluorophore labeled molecule is

detected at the intersection of the two cores (Fig. 6) [3]. The right angle acts as a filter to

separate out the excitation wavelength (solid core) from the emission wavelengths (liquid core)

by creating a 90º change from the excitation light direction to where the emission light is

collected. The emission wavelengths are detected from the liquid core after the copropagating

excitation wavelength is filtered out [3].

9

Figure 4: Cartoon representation of a waveguide. The liquid core can be seen filled with blue

liquid and the solid core is the all gray core perpendicular to the liquid core. The cylinders are the wells that are attached to hold the liquid. Figure copied from Hawkins [3]

Figure 5 shows the S-shape LC-ARROW wafer and how the liquid core terminates in

two fluidic reservoirs that are attached to the sample surface. The reservoirs serve the dual

purpose of providing easy introduction and modification of the sample material and are used to

fill the channel with solution [3].

Figure 5: S-shaped LC-ARROW with wells attached. Two solid cores (vertical) and liquid core (horizontal) can be seen. The liquid core starting in the wells can also be seen. The dimensions of the ARROW wafer are about 1cm x 1cm.

Figure 6: The intersection where the molecule is detected. Excitation light (blue arrow) is coupled into the solid core and the emission light (red arrow) is detected out the liquid core. Figure copied from Hawkins [3].

Figure 7: The ARROW liquid core with the dimensions of the outer core and the layer types [4]. Figure copied from Hawkins [4].

Figure 8: The Solid core of a waveguide with dimensions of ~25μm x 7μm [4]. Figure copied

from Hawkins [4].

10

The LC-ARROW liquid core’s dimensions are 5μm x 12μm (Fig. 7) [3]. The volume of

the inner core of the liquid core intersection with the solid core, where the molecule is

detected, is 9μm x 3μm x 9μm. The solid core is made of SiO2 and its dimensions can be seen

in Figure 8. The samples we used in our experimentation were single overcoat waveguides,

which have a significant increase in sensitivity by increasing transmission efficiency [9].

3 The Detection Chip and System

3.1 Liquid Core ARROW Characterization

The first step in preparing the detection system is to characterize the ARROW sample

by measuring the cores transmittance, which is the ratio of the intensity (power) of the light

that passed through the sample to the intensity (power) of the light when it entered the sample

(T = Pout / Pin) [3]. The transmittance represents the loss of light due to imperfect coupling at

various waveguide connections and the waveguide itself, and is an effective way at measuring

whether the waveguide has low enough loss to be used in biomolecule testing. Both the solid

core and liquid core are tested and are tested with the liquid core filled with water to reduce

scattering.

The transmittance is taken using a Helium-neon laser (HeNe) with an operational

wavelength of 633nm. The HeNe is used for this initial testing because it is often used in the

detection system testing. Before the laser is aligned, the maximum power is measured for Pin.

The HeNe laser fiber is then aligned to the solid core and maximum throughput is obtained by

perfecting the alignment in the x, y, and z direction. Pout is then collected from the waveguide.

The laser powers are measured using an ILX Lightwave OMM – 6810B Optical Multimeter.

Once the solid core Pin and Pout are collected, the waveguide is rotated and the laser is again

aligned with the liquid core and a new Pout is measured (Fig. 9).

11

Figure 9: A picture of the waveguide throughput measuring setup. Starting from right to left, the HeNe laser is labeled and is pointing toward the waveguide. The laser would be coupled to the waveguide and then focused into the power meter with the lens. The maximum throughput would then be obtained by adjusting the laser in the x, y, and z direction to obtain a maximum throughput.

An example of data collected for the LC-ARROW W16-26 is:

Table 1: The LC-ARROW throughput measurements and transmittance calculations.

S.C. L.C.

Pin 1.05μW 1.04μW

Pout 162nW 11.27nW

T 15.4% 1.07%

Pin = power in, Pout = power out, T =transmittance, S.C. =solid core, L.C. =liquid core, W =watts

The percent transmittance that is ideal, indicating low loss, is 10-20% for the solid core

and 0.5-1% for the liquid core. Thus, the sample above had high transmittance and therefore

had a low enough loss to use in testing.

3.2 Detection System Setup

The detection system is the optical and electrical setup that allows for FRET testing in

ARROW waveguides on biomolecules. The detection system consists of lasers, filters,

photodiodes, and a microcontroller. Figure 10 is a block diagram representing the general

setup used in the FRET experiments with ARROW chips.

Laser

Waveguide

Power meter

Lens

12

Figure 10: Block diagram of testing setup. The excitation laser light propagates through the waveguide at the excitation wavelength. The power meter is opposite the waveguide to ensure maximum throughput is obtained. The SPAD is a Single Photon Avalanche photoDiode. Two SPADs are used for dual detection of donor and acceptor emissions. Filters are used to separate the two signals. The photobleaching laser is used to photobleach the acceptor fluorophore and is the acceptor’s excitation wavelength. The color arrows represent the approximate color the light would be at each stage of the experiment, noting the decrease in energy level of the light.

The excitation laser is the wavelength needed for the donor fluorophore being used in

the FRET experiment and an Optical Parametric Oscillator (OPO), which is a tunable laser, is

used for the photobleaching experiment and is tuned to the wavelength of the acceptor’s

excitation. For FRET detection, a 4-channel SPAD router is used to enable four SPADs to be

used at once. We only used two for our experiment, one for the donor photons and one for the

acceptor photons. We used two SPAD filters, the donor filter is a Band Pass (BP) 565/40nm

and the acceptor filter is a Discrimination Filter (DF) 670/40nm. These filters are used to

discriminate between emission light from the donor fluorophore and emission light from the

acceptor fluorophores. They also filter out the excitation light. A PicoQuant TimeHarp 200 is

used in conjunction with the photodiodes as the microcontroller.

The detection system is used after the waveguide is characterized. The ARROW is

placed in the center and the excitation laser is aligned to the solid core of the waveguide using

the power meter. The liquid core is aligned with the photodiodes (Fig. 11). The maximum

alignments are necessary to reduce loss and increase the signal to noise ratio.

Computer

Waveguide

Excitation

Laser

SPAD

Power Meter

Photobleaching

Laser

13

Figure 11: Picture of the testing setup. The excitation laser can be seen at the bottom and is labeled blue representing the wavelength of the laser used. The photobleaching laser is on the right in red representing the wavelength of the laser used. The waveguide holder is where the stand that holds the waveguide fits in. The left is the direction of the SPAD and where the light is detected from the waveguide. The waveguide and SPAD are coupled using the lens.

A background noise measurement is taken first with the liquid core filled with water to

ensure the background of the waveguide is low enough, below 400 counts per second, for

FRET detection. The background is taken at different laser pump power levels (Fig. 12).

Water Background

-50

50

150

250

350

0 50 100 150 200 250

Time(Seconds)

Co

un

ts p

er

Seco

nd Acceptor Donor

Figure 12: This graph shows the background counts at different laser pump power levels. Starting from right to left each drop represents a drop in power starting at 15μW, then 10μW, 5μW,and finally 1μW.

Once the background is taken, the FRET labeled biomolecules are loaded into a well

and pulled into the liquid core using a vacuum attached to the opposite well. This initiates the

flow, but the biomolecules freely flow through the liquid core when data collection is in

Excitation Laser

Photobleaching Laser

SPAD

Waveguide Holder

14

process reducing the possibility of photobleaching. The excitation laser is then turned on and

data is collected. The background is always subtracted from the resulting photon counts when

FRET efficiency is calculated.

4 Biomolecule Research

4.1 A Biomolecule with Controlled Fluorophore Radial Distance

We needed a biomolecule that could be used to test our optical detection system on that

would have a repeatable and stable FRET output. We chose DNA oligonucleotides, which are

short nucleic acid polymers with twenty or fewer bases, as our biomolecule because they are

inexpensive, robust and can be ordered custom synthesized with the fluorophores attached (See

Appendix A1). One strand can be tagged with the donor fluorophore and a complementary

strand can be tagged with the acceptor fluorophore. The two strands can then be annealed to

create a biomolecule with two fluorophores within FRET distance. Because there are only four

DNA base pairs, there is a large problem of spontaneous annealing at locations that match the

DNA sequence but is not the desired annealing location Due to the oligonucleotide’s short

length, spontaneous annealing decreases, increasing the probability for efficient FRET

coupling between the two fluorophores.

There are four specifications that need to be addressed when engineering an

oligonucleotide and its complementary strand with fluorophores attached. The first is to ensure

that the melting temperature (Tm), which is the temperature at which half of the DNA strands

are in the double-helical state and half are not, is above room temperature, indicating the

duplex’s stability. Melting temperatures around 50ºC - 60ºC are ideal [5]. The second is to

create a varying pattern of nucleotides with no area where a complementary strand could bind

to more than one location by more than four nucleotides, which would lead to secondary

structures [5]. The third is the sequence should be longer than ten base pairs, the length a PCR

primer requires to remain annealed to its target for extension to occur, and thus is a good

minimum threshold to consider if making dsDNA [5]. The fourth is to engineer a sequence that

will reduce nucleotide fluorescent quenching. When designing the oligonucleotide, the

nucleotide adjacent to the fluorophore must not be a guanine, which quenches fluorophores,

significantly reducing the quantum yield [1].

15

4.2 Fluorophore Specifications

The acceptor and donor fluorophores must comply with certain specifications for

efficient FRET. The ideal fluorophore must have a high molar extinction coefficient, which is

how strongly a chemical absorbs light at a given wavelength and depends on the structure of

the chemical. The molar extinction coefficient (ε) must be greater than 50,000 M-1

cm-1

at

maximum absorbance [8]. The pair must also have high energy coupling. Energy coupling is

the effectiveness with which the donor’s energy transfers to the acceptor. To choose an

effective pair where optimal energy is transferred, the emission wavelengths of the donor must

considerably cross the excitation wavelengths of the acceptor and this must be a high energy to

lower energy transfer, or low to high wavelength. Thus, the more overlap the two spectrums

have the better the fluorophore pair. Last, the pair must have large spectral separation between

donor and acceptor emissions and high quantum yields [8].

4.3 Oligonucleotide and Fluorophore Design

Based on the criteria stated above for fluorophores and oligonucleotide selection, I

designed, with help from Dmitri Ermolenko from Harry Noller’s lab, two different

fluorophore-tagged oligonucleotides. The first oligonucleotide we created was

The Fluorescein/ Cy3 pair was selected because they have good spectral overlap, have high

quantum yields, extinction coefficients of 75,000 and 136,000 respectively, and have emission

peaks that are distinctly different as can be seen in Fig. 13.

5’- /Fluor/ TGC TGA ACT CGC TGC A –3’

3’-/Cy3/ ACG ACT TGA GCG ACG T –5’

16

Figure 13: Fluorescein isothiocyanate (FTIC) has an emission that can be seen in green

overlapping with the Cy3 excitation also in green. The overlapping area is where FRET occurs, the larger the overlap the better the FRET. As can be observed from this graph the wavelengths have a high to low energy transfer. The two emission peaks can be seen to have distinctly different peaks. The blue line is FTIC’s excitation spectra and the red line is Cy3’s emission spectra Figure copied from www.biotek.com.

Dmitri helped design the basic sequence, ensuring that there would be low spontaneous

annealing and no hairpins by insisting on the shortness of the strand and the variation of the

pattern of nucleotides. The Tm was 54.6ºC showing duplex stability and there were no

indications of secondary structure (see appendix A1). However, I ultimately added the thymine

after the Fluorescein due to guanine’s ability to quench the fluorophore [1].

The second fluorophore-tagged oligonucleotide that we designed was

The Cy3/ Cy5 pair was selected because they have good spectral overlap and distinct emission

spectra (Fig. 14). They were also chosen because Cyanine dyes have been engineered to have

high quantum yields and high photostability, which means the dyes are unchanged when

exposed to light, decreasing the photobleaching effect. The extinction coefficients for the

Cy3/Cy5 pair are 136,000 and 250,000 respectively.

5’-/ Cy5/ AT GCT GAA CTC GCT GCA -3’

3’-/Cy3/ CGA CTT GAG CGA CGT -5’

17

Figure 14: The two emissions spectrums are in red and can be seen to have two distinct peaks. The Cy3 emission and the Cy5 excitation overlap in the green shaded area showing the large spectral overlap between the two fluorophores. Figure copied from ww.biotek.com.

The oliognucleotide was designed with an overhang due to research that found that Cy3

has a higher quantum yield on ssDNA. The overhang was to create more distance between the

dyes and hopefully help Cy3 retain its high quantum yield [6]. The oligonucleotide was also

designed with the specifications stated in section 4.1, had a Tm of 54.8ºC, and had no

indication of secondary structures (see Appendix A1).

The expected FRET efficiency for the oligonucleotides was estimated based on the

radial distance between the two fluorophores. The fluorophores were within 1nm and thus the

efficiency should come out around 99% (Fig. 15).

0%

20%

40%

60%

80%

100%

1 3 5 7 9 11

R(nm)

% E

ffic

ien

cy

Figure 15: Graph of % efficiency vs. the radial distance between fluorophores. Calculated using the formula from section 2.2: E = (1 + (R/Ro) 6) –1, where Ro = 5nm at 50% efficiency [8]. This graph shows the radial distance required for efficient FRET between fluorophores.

18

This expected FRET efficiency is what we used to compare our experimental extracted

efficiency against. We knew that if the DNA is properly annealed then the fluorophores should

be within a radial distance of 1nm and thus have ~99% FRET efficiency.

4.4 Improving The Oligonucleotides Annealing Efficiency

Together, Dr. Aiqing Chen and I, started testing with the Fluor./ Cy3 labeled

oligonucleotides. We initially used ribosome buffer (See Appendix A2) for our annealing

buffer because it was the buffer we ultimately wanted to use in the ARROW with ribosomes.

However, after using the ribosome buffer, we found that it was not conducive to DNA

annealing and clogged the liquid channel due to its high salt concentration and Nikkol. As a

result we were unable to collect any data.

I researched DNA annealing protocols and buffers and found a replacement buffer,

STE buffer, and a better annealing protocol to increase annealing efficiency (See Appendix

A2). One key aspect to increasing annealing efficiency was insuring that the concentrations of

both ssDNA were equal when combined. We used a NanoDrop spectrophotometer 1000 to

determine the concentration of our DNA by measuring the absorbance of the DNA at 260nm

and using the equation:

Concentration n = A/ (εL)

A = Absorbance, ε = extinction coefficient, and L = 1mm for the NanoDrop

It was also important to view the absorbance of the fluorophores at their excitation

wavelengths to ensure there was close to a 1:1 concentration ratio of DNA to fluorophore,

ensuring the fluorophore was well tagged and had not degraded. The concentration ratio was

calculated using two separate calculations: the extinction coefficient of the DNA and the

DNA’s absorbance and the extinction coefficient of the dye and the dye’s absorbance. The two

concentrations were then compared.

I discovered, however, that our Fluorescein-tagged oligonucleotides were either not

well tagged, had extremely degraded, or were being quenched by guanine for the fluorophore’s

absorbance had significantly decreased when tested in the NanoDrop (Fig. 16).

19

Figure 16: NanoDrop absorbance of Fluorescein tagged oligonucleotides. The absorbance of

DNA at 260nm is 0.142 with a concentration of 9μM, however, the fluorophore absorbance at 467nm is significantly degraded, compared with the DNA, at 0.019 with a concentration of 1.3μM. There is obviously not a 1:1 ratio of DNA to dye concentration.

Figure 17: NanoDrop absorbance of Cy3 tagged oligonucleotides. The absorbance of DNA at 260nm is 0.345 with a concentration of 20μM and the fluorophore absorbance at 549nm is 0.318 with a concentration of 13μM, which is almost a 1:1 concentration ratio of DNA to fluorophore.

However, Cy3 did not have any sign of degradation (Fig. 17) and had approximately

equal concentrations to the DNA. Due to the high degradation of Fluorescein, we decided to

then use Cy3/ Cy5 tagged oligonucleotides.

20

4.5 Improve Observed FRET Efficiency of Oligonucleotides

We began testing the Cy3/Cy5 oligo. using the new annealing protocol and STE buffer

(see Appendix A2). The NanoDrop absorbency data were:

Figure 18: NanoDrop absorbance of Cy3 tagged oligonucleotides. The absorbance of DNA at

260nm is 0.967 with a concentration of 65μM and the fluorophore absorbance at 549nm is 0.543 with a concentration 63μM, which is almost a 1:1 concentration ratio of DNA to fluorophore.

Figure 19: NanoDrop absorbance of Cy5 tagged oligonucleotides. The absorbance of DNA at

260nm is 0.372 with a concentration of 22μM and the fluorophore absorbance at 646nm is 0.5 with a concentration of 20μM, which is almost a 1:1 concentration ratio of DNA to fluorophore.

The absorbencies for each fluorophore were around a 1:1 ratio of DNA to fluorophore. Since it

was apparent that the fluorophores were well bound and had not degraded, the absorbencies

21

where then used to calculate the concentrations for each oligo. The oligos. were then annealed

(see Appendix A2) and were tested in the ARROW on the testing set-up and the following data

was collected:

0

2000

4000

6000

8000

9 11 13 15 17 19

Time (Seconds)

Co

un

ts p

er

Se

co

nd

Acceptor Donor

Figure 20: Graph of FRET data collected for Cy3/Cy5 oligo. showing acceptor and donor photon counts. The red line is the acceptor photon count (~8000). The green line is the donor photon count (~3500). The graph represents the transfer of donor energy to acceptor energy demonstrating energy transfer from donor to acceptor because the acceptor count is higher then the donor count.

After the above data was collected, we then used the photobleaching technique to photobleach

the acceptor fluorophore and got the following data:

0

1000

2000

3000

4000

5000

6000

5 15 25 35 45 55

Time (Seconds)

Co

un

ts p

er

Sec

on

ds

Acceptor Donor

Figure 21: Graph of FRET data collected for Cy3/Cy5 oligo. after photobleaching showing acceptor and donor photon counts. The red line is the acceptor photon count (~2500). The green line is the donor photon count (~500). The graph represents the transfer of donor energy to acceptor energy

22

demonstrating the decrease of energy transfer from donor to acceptor because the donor count is now higher then the acceptor count.

It is apparent in Figure 21 that the donor count increased, compared to Fig. 20, due to

the destruction of acceptor fluorophore and thus its ability to transfer energy with the donor.

The FRET efficiency was then calculated using the photobleaching equations and the changes

in donor and acceptor intensities. The oligo. sample had a FRET efficiency of 60%. However,

60% is low for a 1nm radial distance between fluorophores compared to our expected

efficiency of ~99%.

We needed to improve the efficiency and I speculated that the reason for the low

efficiency was because of free ssDNA in the sample solution, for the free donor fluorophore,

unbound to an acceptor, would contribute to the donor photon count and thus bring down the

coupling efficiency. To improve the efficiency, the ssDNA needed to be removed from the

sample without interfering with the dsDNA.

With help from John Kim in Dr. Nader Pourmand’s lab, we decided to use Exonulease I

from E. Coli to remove the ssDNA since it is an enzyme that catalyzes the removal of

nucleotides from ssDNA in the 3’- 5’ direction and doesn’t interfere with helical DNA [2]. We

first tested whether the enzyme would interfere with our helical DNA and ran the following

agarose gel:

PCR+, PCR-, Helix+, Helix-, Sing+, Sing-, ladder

23

Figure 22: Agarose gel of Exonuclease I experiment. Starting left to right the first two bands are a known PCR helix, used as the control, to prove Exonuclease I dose not interfere with known helical DNA. The + and – mean with and without Exonuclease I respectively. The next set of two bands is the annealed oligos sample. The gel shows that the Exonuclease I did not interfere with my annealed sample because the two bands are the same intensity and are the same distance down the gel. The next set of two bands is my ssDNA with and without Exonuclease I. The first band is the treated sample and it is apparent that the Exonuclease I did have an effect on the sample because of the difference between the treated and non-treated bands. The intensities of the two bands are significantly different. The non-treated sample has a high concentration and thus a bright band. The treated band is dull in comparison and thus is not ssDNA. There is still a faint band and this could be due to free fluorophore or partially digested ssDNA. Because of the faint band I decided we needed to purify the sample after Exonuclease I treatment.

For the experiment, we tested the Exonuclease I on our annealed oligo. sample, on our

single stranded sample and we also tested the Exonuclease I on a known PCR helix as a control

to prove that it did not interfere with the dsDNA. As can be seen in Fig. 22, the Exonuclease I

did not interfere with the helical DNA and did digest the ssDNA. I therefore decided to test

whether the Exonuclease I would improve our FRET efficiency by removing the ssDNA.

I modified our oligo. preparation to include the Exonuclease I protocol. I used the

existing annealing protocol (see appendix A2). Once annealed, the oligo. sample was treated

with Exonuclease I (see appendix A3) and after treatment they were purified (see appendix

A4). After preparation, the following FRET data was collected:

0

500

1000

1500

2000

2500

3000

9 14 19 24 29 34 39

Time (Seconds)

Co

un

ts p

er

Sec.

Acceptor Donor

Figure 23: Graph of FRET data collected for Cy3/ Cy5 oligo. showing acceptor and donor photon counts. The red line is the acceptor photon count (~2500). The green line is the donor photon count (~500). The graph represents the transfer of donor energy to acceptor energy demonstrating energy transfer from donor to acceptor because the acceptor count is higher then the donor count.

24

After the initial data was collected, we used the photobleaching technique to photobleach the

acceptor fluorophores and got the following data:

Figure 24: Graph of FRET data collected for Cy3/ Cy5 oligo. after photobleaching showing acceptor and donor photon counts. The red line is the acceptor photon count (~1000). The green line is the donor photon count (~2000). The graph represents the transfer of donor energy to acceptor energy demonstrating the decrease of energy transfer from donor to acceptor because the donor count is now higher then the acceptor count.

In Fig. 24, compared to Fig. 23, the donor count increased and the acceptor count

decreased due to acceptor photobleaching. FRET efficiency was then calculated using the

photobleaching equations and the changes in donor and acceptor intensities. The Exonuclease I

treated oligo. sample had a FRET efficiency of 91%.

This was a 30% increase in efficiency from the untreated sample, concluding that the

free ssDNA in the solution was a cause of the low FRET efficiency and that using Exonuclease

I to remove the ssDNA from the sample improved the FRET efficiency. This was also only 8%

lower then the expected efficiency of ~99%. This 8% can partially be accounted for in the loss

due to heat, radiation, and the photostability of the fluorophore.

5 Conclusion

In creating an optofluidic FRET detection system, we needed a biomolecule that could

be used as a positive control with which to test in our system that would efficiently produce

fluorescence. We chose DNA oligonucleotides because they are cheap, robust and can be

ordered with fluorophores attached. They were also chosen because once annealed they

produce and maintain a ~1nm distance between fluorophores, which creates an optimal FRET

25

efficiency of ~99%. We created two oligonucleotide designs, one with fluorophore pair

Fluorescein/ Cy3 and the other with fluorophore pair Cy3/ Cy5. The Cy3/ Cy5

oligonucleotides were the only ones that produced data because the Fluorescein on the Fluor./

Cy3 either quickly degraded or was quenched, but could not efficiency FRET with Cy3. We

were able to obtain 60% efficiency with the Cy3/ Cy5 tagged oligonucleotides. To improve the

FRET efficiency we used Exonuclease I from E. Coli to remove the ssDNA from our sample.

The Exonuclease I treated Cy3/ Cy5 oligonucleotides produced 91% efficiency, concluding

that ssDNA was contributing to the low efficiency and we were able to get an experimental

efficiency that was close the predicted efficiency of ~99%.

After the FRET efficiency was calculated for the treated sample, I calculated the

percent efficiency vs. power level (Fig. 25) to determine if the efficiency and thus the

concentration of the sample is stable in the ARROW waveguide.

FRET Efficiency v. Power Level

72%

74%

76%

78%

80%

82%

84%

86%

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Power Level (uW)

% E

ffic

ienc

y

Figure 25: Graph of calculated FRET efficiencies using the photobleaching efficiency equation and the gamma value minus the background vs. the power level correlating to the percent efficiency. This graph should flatten as the power level increases showing the saturation of the fluorophores. However, there is an increase, as the power increases showing that the concentration of the sample is not stable and is increasing as well.

Unfortunately, the percent efficiencies are not stable with the increase of power leading to the

speculation that the concentration of the sample in the ARROW is not constant and is

increasing with the power level as time increases. This is one limitation that needs to be

overcome if the ARROW and the detection system are ever going to be used for single-

molecule detection. The microfluidics need to be controllable for single-molecule

measurements and I think this is the next step in the ARROW development.

26

Acknowledgments I would like to thank my advisor, Dr. Holger Schmidt, for giving me this opportunity,

Dr. Aiqing Chen for working with me on the optics of this project, John Kim in Dr. Nader

Pourmand’s lab for teaching, providing materials and helping me with the wet lab procedure

for the Exonuclease I research, and Dmitri Ermolenko from Dr. Harry Noller’s lab for assisting

with the DNA oligonucleotide nucleic acid and fluorophore design.

References [1] M. Behlke, et al., “Fluorescence and Fluorescence Applications,” Integrated DNA

Technologies, pp. 1-13, 2005. [Online]. Available: IDT,

http://www.idtdna.com/TechVault/TechVault.aspx. [Feb. 2010].

[2] P. Cornish, D. N. Ermolenko, H. F. Noller, and T. Ha, “Spontaneous Intersubunit Rotation

in Single Ribosome,” Molecular Cell. vol. 30, pp. 578-588, June 2008.

[3] A. Hawkins, H. Schmidt, “Optofluidic waveguides: II. Fabrication and structures,”

Microfluid Nanofluid. vol. 4, pp. 3-16, 2008.

[4] A. Hawkins, H. Schmidt, “Optofluidic waveguides: II. Fabrication and structures,”

Microfluid Nanofluid. vol. 4, pp. 17-32, 2008.

[5] Integrated DNA Technologies, “Frequently Asked Questions,” Integrated DNA

Technologies, IDT, 2003. [Online]. Available: http://www.idtdna.com. [Accessed: Feb.

2010].

[6] M. Massey, W.R. Alger, U.J. Krull, “FRET for DNA biosensors: FRET pairs and Forster

distance for various dye-DNA conjugates,” Analytica Chimica Acta. Vol. 568, pp. 181-

189, 2006.

[7] D. Mijatovic, J. C. T. Eijkel and A. van den Berg , “Technologies for nanofluidic systems:

tob-down vs. bottom-up- a review,” The Royal Society of Chemistry, Lab Chip, vol. 5,

pp. 492-500, 2005.

[8] R. Roy, S. Hohng, and T. Ha, “A practical guide to single-molecule FRET,” Nature. vol.5,

no. 6, pp.507-516, June 2008.

[9] E. Lunt, et. Al, “Hollow waveguide optimization for fluorescence based detection,” Proc.

of SPIE. Vol. 6883 68830H-1, 2008.

Appendix A1: Oligonucleotide Company

Company that fluorophore tagged oligonucleotides can be ordered from and that has a

Tm and secondary structure analyzer: www.idtdna.com

A2: Oligonucleotide Annealing Protocol

- Ribosome Buffer

50mM Hepes-KOH (pH 7.5)

4mM MgCl2

400mM NH4Cl

6mM β- mercaptoethanol

.1% Nikkol

27

- STE Buffer

10 mM Tris pH 8.0

50 mM NaCl,

1 mM EDTA

Annealing Protocol from www.idtdna.com

ssDNA Prep.:

- Thaw DNA at room temperature

Keep DNA covered at all times to reduce degradation

- Once thawed, vortex and then quickly spin down to ensure all sample is at

the bottom

- Use the NanoDrop to measure the absorbance at 260nm for the DNA and at

the absorbance of the fluorophore used to get the concentrations

- Find the concentration for both ssDNA

Calculate the concentration using the equation stated in section 4.4

- Dilute using the equation C1V1=C2V2 to ensure both ssDNA are in equal

concentration

Make sure the concentrations are high, 15mM or higher, for efficient

annealing

- Once equal concentrations are obtained combine equal volumes of each

ssDNA

Remember that the concentration of dsDNA is half the concentration of

ssDNA

Annealing:

- Preheat a water bath on a heat block to ~70oC

Make sure the temp it higher than 60oC but less than 80

oC

- Once the temperature is obtained place the sample in the water bath and turn

off and unplug the heat block

- Wait ~1 hour for annealing to occur as temperature cools

- Then place in freezer till use

A3: Exonuclease I Protocol

Protocol From Nader Pourmand’s lab:

- Annealed DNA sample

- PCR Buffer:

500 mM KCl 100 mM Tris-HCl (pH 8.3)

15 mM MgCl2

- MgCl

- Exonuclease I

Exonuclease and DNA prep:

- The best results were obtained when using the following amounts of

reagents

- 8μL DNA

- 1μL PCR buffer

28

- 1μLMgCl

- 0.2μL Exonuclease

A higher volume can be used just use the ratios stated above

- Once combined in a PCR tube, place in a PCR thermocycler

- The cycle should be 37oC for 30min. for the Exonuclease to degrade the

ssDNA and then 80oC for 20min. to inactivate the enzyme

A4: Purification Protocol

Protocol From Nader Pourmand’s lab:

Component Ratio Example Amt.

Sample X X=10μL 100μL

2x Binding

Washing Buffer

0.3*X 30μL

99.5% EtOH 2.7*X 270μL

75% EtOH 200μL

Carboxy Beads 10μL 10μL

Resuspension Vol. Y Greater than 15μL

1. Add binding washing buffer into Lo-Bind tube containing sample

2. Add 99.5% EtOH

3. Add Carboxy beads

4. Mix by vortexing briefly, spin down quickly, incubate at room temp. for 15 min

- Be sure to keep the beads moving during this 15 min either by hand or

on a slow mixer

5. Magnetize tube for at least 1 min, make sure beads are all collected

6. Discard supernatent

7. Wash with 100μL of th 75% EtOH and mix beads by pipetting up and down

8. Magnetize for 1min

9. Discard supernatent

10. Repeat steps 7-9

11. Allow beads to dry for ~ 5min to remove excess ethanol. Do not allow beads to

dry completely - Try to dry by blowing air with pipett but be careful not to get any beads on

pipett

12. Resuspend in Y μL of H2O

13. Magnetize tube

14. Transfer supernatent to clean tube