abstract - Cornell University

thelemicbathΒιοτεχνολογία

20 Φεβ 2013 (πριν από 4 χρόνια και 5 μήνες)

189 εμφανίσεις



ELECTROSPUN NANO
-
FIBERS AS BIOSEPARATORS: CHARGE BASED
SEPARATION OF
ESCHERICHIA COLI

FROM LIQUID SAMPLES








A thesis

Presented

to the
Faculty

of the Graduate School

of Cornell University

in
Partial

Fulfillment

of the
Requirements

for

the degree of

Master of Engineering







by

Marissa Agustin

January 2011
















© Marissa Agustin 2011



ABSTRACT

Biosensors for pathogen detection have the potential to prevent and identify

health hazards in environmental

samples
and food samples. However, i
n order to detect
pathogens in
complex

samples, concentration and separation of the bacteria from the
sample

matrix is necessary.
Obtaining rapid accurate results is essential for detecting
foodborne pathogens
,

as traditional methods can take several days
to produce a result
.
Sample preparation is time consuming and usually performed off
-
chip.
New technologies
such as
lab
-
on
-
a
-
chip
platforms

aim to

integrate

several laboratory operations on
to

a
single chip. Electrospun nanofibers can be incorporated into mi
croflu
i
dic channels and
can be functionalized with various bio
-
recognition elements.
The e
lectrospinning
process
uses an electrical charge to draw out thin continuous polymer threads with diameters
ranging from 10
-
1000nm.
These nanofibers

are therefore ide
al for use
in

filtration
because they provide a higher surface area as compared to traditional membranes.

Functionalized n
anofibers carrying
positive

and
negative

charges were spun by
the Frey Group at Cornell University
. The fibers were investigated for
their ability to
separate
E.coli

from
buffer samples

and
actual
apple juice samples. This separation is
based on the knowledge that
E.coli

and gram negative bacteria have a strong net negative
charge at neutral pH and a net positive charge at low pH caused by
deproto
nation
and
proto
nation of carboxyl and ammonium groups on the cell surface
1
.
Samples buffers at
pH7

and
pH4

were used in this s
tudy because the pH of apple juice is
typically
around 3
or 4.

Filtration was evaluated using two methods 1)
by
collecting cells at the channel
outlet, plating on
Trypto Soy Agar

(TSA)


and
counting
colonies the next day and

2)
by
staining the cells

with
S
yto9

green fluorescent dye

and
capturing digital images using RS
CoolSNAP camera and RS Image software. The images were then analyzed
in

ImageJ

to
measure
fluorescence intensity in the channel
, high fluorescence correlating to
the
presence of
E.coli

on the fibers
.


In the first method, p
ositive fibers were shown to retain 98 ± 1.6% of
E.coli

that
passed through the microfluidic channels whereas negative fibers only retained 35 ±
1.1%.

Unexpectedly, i
n the empty channel only 53
±

4.4
% of

cells were re
covered at the
outlet
.
Since there are no fibers in the channel, close to 100% recovery would be
expected however, c
ell losses could be
attributed to cells adhering to the hydrophobic
channel and tubing surfaces or
cells

becoming trapped at the inlet or ou
tlet holes.
Coating

the channel and tubing with 1% BSA prior to experiments

increased the recovery at the
outlet to 83 ±15 % of the cells.

In the second method, e
pifluorescence measurements provide
d

a less quantitative
result but allow
ed

for visualization

of
E.coli

on the fibers.

E.coli

in
pH7

buffer
was

filtered onto positive fibers

producing a strong fluorescence. The cells
were
eluted by
shifting pH

conditions

by introducing a
pH4

washing buffer

producing
an

84% drop in
fluorescence intensity
.
E.coli

in
pH4

buffer attached better to negative fibers
producing
ave
rage fluorescence intensity 1.6 times

higher than the channels with positive fibers
.
The results were repeated with apple juice spiked with
E.coli

producing similar
results.

T
he negative channe
ls prod
uced fluorescence intensity 2 times

higher than in the positive
channels
.

These results show the potential for electrospun nanof
ibers to be used on
-
chip as
bio
separators.
The filtration investigated here is non
-
specific with the purpose of isolatin
g
bacteria from a sample matrix which might be an environmental sample or a food sample.
Once the bacteria are immobilized on the fibers, they can either be detected using
fluorescently tagged antibodies or eluted to be detected further downstream.
Due to
time
constraints

and availability of fibers
, some experiments in this study were not repeated in
triplicate. Further studies should be done to confirm the results.
In order to design
nanofiber mats
for
different filtration applications, further experiments

need to be done to
correlate

spinning time
to

fiber pore size and
fiber mat
density.




ACKNOWLEDGEMENTS


I am grateful for all the guidance

that my advisor, Professor Antje Baeumner, has
given me in this project
.

This thesis would not have been possible without her knowledge
and expertise.
I would also like to thank her for
the kindnes
s and patience she has shown
me

as my undergraduate advisor, research advisor and teacher

throughout

the four and
half years that I have been at Cornell.


I would like to thank Lauren Matlock
-
Colangelo for
the long hours she spe
nt in
the CNF making electrodes,
for
teaching me how to use the lab equipment

and how to
fabricate
PMMA
devices.
I would like to

thank

Christian

Willrodt

and Lauren
for
their
help and
b
eing there to
bounce ideas off
.

I would like to

thank
Dr.
Da
ehwan Cho and
Lauren for spinning the nano
fibers

I needed for my experiments
.

I would like to thank the Department of Biological and Enviro
nmental
Engineering
staff, faculty
and my professors

who have been a tremendous help and from
whom I have learned so much.


I would like to give a big hug
to

all m
y friends who stayed at Cornell for MEng

or
PhD

degrees

and toughed
the semester

out with me
.
Thanks to

Coll
e
getown Bagels for
all
the coffee, warmth, and
pastries
and for
allowing me to sit
in your store for hours on end
analyzing data

and writing this thesis
.
Finally
,

I

would like to thank my parents, my
brother, Michael and my
sisters,

Angela,

Rebecca and Rosenda,
for

all

the

love and
support

they have shown me
.

BIOGRAPHY


Marissa Agustin is from Scarsdale, New York. She has received both a Master of
Engineering and a Bachelor of Science in Biological Engineering from Cornell
Univers
ity. Her
interests are in micro
fluidics, biosensors, biological
process engineering
,
and food safety.

She is certified as an Intern Engineer by the State of New York

having
passed the Fundamentals of Engineering
E
xam in April 2010.
She is a member of the
Cornell Io
ta Beta Chapter of Alpha Epsilon, the national honors society of Agricultural,
Food and Biological Engineering.
Her hobbies include

playing
flute, Filipino folk
dancing

and making
her own
fermented food products.



LIST OF FIGURES


Figure 1: Top view of
device
channel design

Figure 2:
Microfluidic Device
Fabrication Scheme

Figure 3: Electrospinning Setup

Figure 4: Size distribution of particles in apple juice

Figure 5:
Volume Fraction

of particles

in apple juice
according to diameter


Figure 6:
Cell
counting
:

p
reliminary results of nanofiber filtration.

Figure 7:
Cell counting
:

n
anofiber filtration
of
E.coli


Figure 8:
Fluorescent measurements for filtration
of
pH7

samples


Figure 9:


Fluorescent images for positive fibers washed with
pH4

buffer

Figure 10:

Fluorescent images for positive fibers washed with
pH7

buffer

Figure 11:

Comparison of filtration of
pH4

buffer samples and apple juice samples


LIST OF ABREVIATIONS


ATCC


American Type Culture Collection

BSA



Bovine Serum Albumin

CDC


Center

for Disease Control

CFU


Colony forming units

CNF


Cornell NanoScale Science and Technology Facility

DI


Deionized

DLS


Dynamic Light Scattering

DMSO


Dimethyl sulfoxide

DNA


Deoxyribonucleic acid

IMS


Immuno
-
magnetic separation

NASBA

Nucleic acid
sequence based amplification

PBS



Phosphate Buffered Saline

PMMA

Poly
-
methyl methacrylate

Polybrene


H
exadimethrine bromide

PolyMA


P
olymethyl vi
nyl ether
-
alt
-
maleic anhydride

PVA


Polyvinyl
Alcohol

RT
-
PCR

Real time polymerase chain reaction



TSA


Trypticase soy

Agar

TSB


Trypticase
soy Broth

UF


Ultrafilt
ration

UV


Ultraviolet

UV/O
3`

Ultraviolet ozone treatment



TABLE OF CONTENTS


ABSTRACT

................................
................................
................................
..............................

3

LIST OF FIGURES

................................
................................
................................
..................

7

LIST OF ABREVI
ATIONS

................................
................................
................................
.....

8

1.0 INTRODUCTION

................................
................................
................................
............

11

1.1 Motivation

................................
................................
................................
.....................

11

1.2 Separation Methods

................................
................................
................................
......

12

1.3 Electrospun Nanofibers

................................
................................
................................
.

13

1.4 Microfluidic Devices

................................
................................
................................
....

13

1.5 Sample Matrix

................................
................................
................................
...............

14

2.0 DESIGN

................................
................................
................................
............................

16

3.0 MATERIALS AND METHODS

................................
................................
......................

18

3.1 Materials

................................
................................
................................
.......................

18

3.2 Bacterial Culture

................................
................................
................................
...........

18

3.3 Dynamic Light Scattering

................................
................................
.............................

19

3.4 Device Fabrication

................................
................................
................................
........

19

3.5 Electro
-
spinning

................................
................................
................................
............

21

3.6 Method 1: Quantitative Cell Counts

................................
................................
.............

22

3.7 Method

2: Epifluorescent Microscopy

................................
................................
..........

23


4.0 RESULTS

................................
................................
................................
.........................

24

4.1 Apple Juice Dynamic Light Scattering

................................
................................
.........

24

4.2 Results: Quantitative Cell Counts

................................
................................
.................

25

4.3 Results: Epifluorescent Microscopy

................................
................................
.............

28

5.0 CONCLUSION AND FUTURE WORK

................................
................................
.........

32





1.0
INTRODUCTION


1.1
Motivation

The most recent statistics on
food borne illness

published in 2011 by the CDC
estimate
that there

are 47.8 million
cases
, 127,839 hospitalizations, and 3,037 deaths per
year in the United States caused by foodborne illnesses
2
,
3
. These numbers are lower than
the statistics reported in
1999
4
;

however
,

because of differences in
methodology

the two
studies
cannot

be compared. T
herefore it
also
cannot be concluded that food quality and
safety in the US has improved. In fact, the new data
indicates

that cases of foodborne
illness are still quite high.
To ensure that health and safety standards are being met, t
here
is a need for rapid

pathogen

dete
ction
methods that are

both sensitive and specific.

However, r
esearch efforts aimed at detection of food
-
borne pathogens often focus
solely
on identification of
the
bacteria and overlook the preliminary sample preparation.
The processing steps used for sam
ple concentration and separation are important because
food matrices are both diverse and complex. Food samples usually contain very low
levels of pathogenic organisms and may contain a wide range of background microflora.
In addition, particulates in food

may shield microorganisms or inhibit reactions necessary
for DNA based detection. In the food industry, the gold standard for pathogen detection is
enrichment culture. This method relies on availability of media that will selectively
enhance growth of a t
arget organism while inhibiting growth of other bacteria that may
be present in the sample. Typically, amplification of cell numbers to levels dete
ctable by
standard plate count

can take over 24 hours. Consequently, producers experience
economic losses bec
ause food products need to be held until lab tests are completed.
Furthermore, this method does not give direct quantitative information about the original
level of bacteria in the food sample. To address these problems, rapid tests and sensors
that use
re
al time polymerase chain reaction (RT
-
PCR)
,
nucleic acid sequence based
amplification (NASBA),

immunoassays or amperometric methods have been developed
for detection of pathogens

5
,
6
. Molecular based detection methods are fast, accurate and
sensitive;
however, they can only process sample volumes of a few µL. This is a problem
because standard food sampling usually requires greater than 25g or 10mL per sample
because of the low levels of pathogenic bacteria in foods. Compounds present in the food
sample

or enrichment media may also interfere with enzymes and antibodies used in
molecular based detection methods. In order for these more advanced technologies to be
adopted for quality control in the food industry, methods must be developed that are
capable
of both reducing sample volumes and separating the target analyte from the food
matrix.


1.2 Separation Methods

Existing separation methods for isolation of bacteria from food samples can be
grouped into two categories: physical methods and adsorption meth
ods. Summaries of
different types of physical and adsorption methods used in the food industry have been
published in microbiology and food science journals. The more commonly used methods
include centrifugation, filtration, and
immune
-
magnetic separation
(IMS
)

7
,
8
,
9
.

Centrifugation and filtration are able to effectively concentrate bacteria but also end up
concentrating food particles along with the target analyte. The removal of food debris
requires multiple wash and centrifugation steps and, in the case of filtratio
n, the debris
will eventually foul the filtration membrane and decrease flux rates. IMS is an attractive
method because it is capable of concentrating samples, removing impurities and isolating
specific bacterial strains from food matrices. However, IMS is

expensive because of the
cost of monoclonal antibodies and because the method can only process relatively small
sample volumes. Food debris may become trapped between the magnetic beads and
particles in the food may interfere with binding events. Therefor
e, IMS may require
multiple wash and
magnetic collection

steps which is time consuming and hard to
integrate into an on
-
line system.

1.3 Electrospun Nanofibers

Recently, there has been increased interest in electrospun nanofibers which
exhibit properties
attractive for filtration, as well as a wide range of other applications in
the food, biotechnology and biomedical industries
10
. Nanofibers with small diameters,
10
-
1000nm, can be spun uniformly into non woven mats and across microfluidic
channels. These fi
bers have high surface area to volume ratios, porosity and mechanical
strength
7
. Research has also shown that it is possible to functionalize these nanofibers by
incorporating compounds into the electrospinning solution
11
. Functionalization allows for
selec
tivity based on chemical properties of the target analyte in addition to selectivity
based on size. The pores formed by these nanofiber mats can range from 0.16 to 8.03µm
depending on fiber diameters
12
. In comparison
E.coli

cells are about 0.5µm in diameter

and
1
-
2µm long. Integration of membranes into microfluidic devices has substantial
applications for sample preparation, separation, purification, biosensing, etc.
13
.


1.4 Microfluidic Devices

Fabrication methods, well established in the semiconductor
industry utilize glass
and silicon as substrates. For biological applications, glass has useful properties but tends
to be expensive. P
olymer chips are increasingly being used for microfluidic devices
because they are transparent,
disposable and relatively

low cost
14
.

The f
abrication
processes for microfluidic devices made using polymer substrates

have been described
previously
15
,
16
.

Microfluidic channels were

fabricated in
poly
-
methyl methacrylate
(PMMA)

using hot embossing with a copper master. PMMA
require
s surface
modification with
UV
-
ozone (
UV/O
3
)

treatment

to allow for successful low temperature
thermal bonding.

UV/O
3

exposure inserts oxygen containing functional groups into the
polymer surface decreasing hydrophobicity
17
.



1.
5

Sample
Matrix

Commercial juices are a feasible starting point for separation and concentration of
actual food samples because they contain few solid particles and are not very viscous.
Recently, attention has been focused on monitoring the processing of fruit juices.
Co
ntamination with
Escherichia coli

O157:H7 is usually associated with ground beef and
dairy products, however, between 2002 and 2005, six juice
-
associated outbreaks of
illness were reported by the CDC and five out of the six were linked to apple juice
18
. The

source of
E.coli

contamination is usually feces and manure which can be introduced to
fruits when they fall on the ground. Fruits pressed with the skins on can then lead to
contamination of the juice. To my knowledge, nanofibers spun into microfluidic cha
nnels
have not yet been applied to sample concentration for microbial te
sting of beverages. One
study
investigated electrospun nanofiber membranes for clarification of apple juice
19
. In
that study, a dead
-
end filtration was
performed

using a large, 63mm dia
meter, poly
(ethylene terephthalate) nanofiber mat. The nanofiber filtration produced similar quality
apple juice to product obtained using a traditional cellulose
UF

membrane. In addition,
the composition of dissolved solutes (phenolic compounds, sugars a
nd organic acids) in
the nanofiber filtered apple juice was almost the same as the composition of the un
-
clarified control juice. Phenolic compounds are known to interfere with PCR detection of
E
.

coli

0157:H7 in juice samples
20
. Therefore the fact that the
se compounds passed
through the nanofiber mat is promising for sample concentration. Ideally, the nanofiber
mat should retain the target bacteria while allowing impurities that would interfere with
pathogen detection to pass through the fibers.

Apple juice

composition will vary with variety of the fruit, climate, processing
conditions and storage. Typically the pH is between 3.0
-
3.8. Sugars make up about 7
-
14% of soluble components in apple juice, malic acid is present at a concentration of
0.5% and soluble

proteins, phenolic compounds and pectin make up 120
-
500ppm
weight/volume of apple juice. Other constituents include minerals, potassium, calcium,
and esters
21
. The natural microflora of apple juice is comprised of mostly yeasts and
molds. However, contamin
ation with
E.coli

and Salmonella has been associated with
unpasteurized apple juice.



2.0
DESIGN


Previous work done in the Baeumner lab
has

shown that
electrospun nanofibers
can be successfully
incorporated into
the
microfluidic channels
of PMMA devices
using
thermal bonding.
Polyvinyl alcohol

(PVA)
nanofibers, provided by the Frey group
in the
department of Fiber Science and Apparel Design
at Cornell University, were
tested

for
their
effectiveness for

charged based separation.

Negatively
charged f
luorescent
liposomes

have been found
to adher
e to positively charged fibers
. The liposomes

can be
observ
ed

by placing the PMMA devices under an epifluorescence microscope and
measuring
fluorescence intensity
over
time.

In this study functionalized nanofibers were tested for a different application.
E.coli

and gram negative bacteria are
also
known to have a negative charge at neutral
pH
1

but bacterial cells are much larger than liposomes
. The objective
s

of this study
were

to 1)
provide a proof of concept that
E.coli

can be
capture
d by functionalized nanofibers
2)

visualize
E.coli

on the fibers using fluorescent sta
ining 3) determine the conditions to
elute

attached

bacteria from
the
fibers and 4
)
evaluate

the effectiveness of nanofiber
filtration of actual apple juice samples spiked with
E.coli
.


The filtration was evaluated using two methods 1) by collecting cells at the
channel outlet, plating on TSA and counting colonies the next day and 2) by staining the
cells with Styo9 and using epifluorescent microscopy and an imaging software to
measure fl
uorescence intensity in the channel.
Fluorescence measurements offer a
qualitative measurement of binding however,
E.coli

can be quantified directly using
standard plate count
s
.

In method 1, the number of cells collected at the outlet can give a
quantitati
ve measure of how many cells pass through the fibers and because the initial
concentration of cells is known, how many cells were retained by the fibers.
E.coli

samples in
pH7

buffer were filtered through channels with no fibers, positive fibers and
negati
ve fibers.

In method 2,
t
he filtration effectiveness of
channels
was determined by the
fluorescence intensity in the channel. Channels with

positive
fibers
and negative fibers
were

compared for
E.coli

spiked samples in
pH7

buffer and
pH4

buffer
. The experiments
done in
pH4

buffer were
also
compared to experiments with actual
apple juice

samples

(ph
3.
9
) spiked with
E.coli
.
After pumping the sample through the channel for 20
minutes, the syringe was switched and the fibers were washed with a
pH4

wash buffer, a
pH7

wash buffer or apple juice depending on the experiment.


Originally the microfluidic device design consisted of a single channel 30mm in
length, 1mm wide and 50μm deep. The second chip design consisted of 4 parallel
channels with the sam
e dimensions as in the original chip. Tubing was cut to a length
such that the wash buffer would hit the channel exactly 20 minutes after the syringe was
switched if the flow rate was kept at 1μL/min.


3.0
MATERIALS AND METHODS


3.1 Materials

Syto 9

® Green Fluorescent Nucleic Acid Stain 5mM solution in
dimethyl sulfoxide
(
DMSO
)

was

purchased from Invit
rogen
(Eugene , OR)
.
Trypto

soy agar (TSA
) and
Trypto

Soy Broth (TSB) were purchased from BD Biosciences (
Sparks, MD)
.

Apple
juice (Tops Brand) was purchase from the local supermarket (Ingredients: apple juice
concentrate, water, ascorbic acid).

Stainless
steel
blunt needles with Luer polypropylene
hub 22 Gauge

were purchased from VWR.

Tygon Micro Bore PVC tubing
with
inn
er
diameter 0.02 in

and o
uter diameter 0.06

in were purchased from VWR.


3.2 Bacterial Cultur
e

Non pathogenic
E.coli

K
-
12 strain ATCC 25922 was obtained from
Prof. Randy Worobo
at the Dept. of Food Science and Technology, Geneva, NY.

The culture was
s
treaked for
isolation
on

plates containing
40g/L

Trypto

Soy Agar (TSA)

pH 7.3 and incubated at
37°C for 20 hours
.
A 40g sample of TSA contains approximately 15g pancreatic digest of
casein, 5 g enzymatic digest of soybean meal, 5g sodium chloride, and 15g
agar. The
media was prepared in a 1L Erlenmeyer flask, 20g TSA powder was dissolved in 0.5L DI
water and autoclaved at 121°C for 35 minutes. The media was cooled to 60°C and poured
into petri plates. Plates were left on the bench overnight to cool and then

stored at 5°C for
later use. Liquid c
ultures were grown in liquid media containing
3
0g/L Trypto Soy Broth
(TSB), final pH 7.3.
A 30g sample of TSB contains approximately 17g of pancreatic
digest of casein, 3g enzymatic digest of soybean meal, 2.5g dextrose, 5g sodium chloride,
and 2.5 g of dipotassium phosphate. The liquid media was prepared
by dissolving 6g of
TSB in 200mL of DI
water. 10mL aliquots
of TSB
were di
spensed into test tubes and
then

autoclaved

at 121°C for 35

minutes.

To prepare samples for experiments, a single colony of
E.coli

was

transferred

into
a test
tube with
10mL
TSB

(
3
0g/L)

and incubated for 20 hours

in a La
b
-
Line Orbit Environ
Shaker (Melrose, IL) at 150rpm.

To determine the initial concentration of cells in the
broth,

100µL samples
of the

culture
dilutions 10
4
-

10
6

in PBS
were
pipetted

and spread
on
to

TSA
plates
and incubated overnight
at 37°C
.
Colonies
were counted the following
day and the initial concentration in CFU/mL was calculated from this number. CFU =
#colonies x 10
n

where n = the dilution factor. It was found that consistently, a
fter 20
hours, t
he culture

contained

a concentration of approximat
ely
10
9

CFU/mL
.

The culture
was
then
serially diluted
in phosphate buffered saline (PBS)
to desired concentration
s for
experiments
.


3.3

Dynamic Light Scattering

DLS analysis was
performed

using a 90 Plus Nanoparticle Size analyzer with a Peltier
temperature control system

(Brookhaven Instruments, Holtsville, NY)
. 40mL samples of
20%, 50% and 100% juice solutions in UF water were prepared.
Measurements were
taken at a 90° angle.
A cuvette w
as filled with the sample and inserted into the machine
for a reading. The temperature was set to 20°C., dust cut
-
off was set at 95% to increase
the signal retention because of preliminary measurements that showed particles in the μm
range.
Apple juice v
i
scosity was assumed to be that of water, 1.002 cP

and particles were
assumed to be
perfect spheres
. Refractive index for each of the samples was measured
using a

digital
refractometer

(Misco Products Division, Cleveland, OH) and inputted into
the BIC softw
are.


3.
4

Device Fabrication

Microflu
i
dic devices were fabricated in PMMA using hot embossing and UV assisted
thermal bonding. Pieces of PMMA approximately 50mm x 50 mm were cut using a band
saw (Ryobi Tools).



Figure 1: Top view of channel design


The copper master template was fabricated
by Lauren Matlock
-
Colangelo in the CNF
using photolithography,
wet etching
and electroplating
to produce the desired raised
pattern
12
. Microfluidic channels were embossed
into the PMMA
by sandwiching the

PMMA chip

between the copper master and another copper plate
. The PMMA copper
sandwich was warmed on the hot
press (Fred S. Carver Inc., Summit NJ)
at 120
°
C for 5
minutes and then 5000lb of pressure was applied for 10 minutes at 120
°
C.

Inlet and
outlet holes with a diameter of 0.8mm were drilled using the drill press (Ryobi Tools).
PMMA
chips with embossed channels

were
sonicated
(Aquasonic 75D VWR Scientific)
at room temperature
for 10

min in 2
-
propanol
.


The PMMA chips

with the emboss
ed channels and the PMMA chips with the
nanofiber mats

were
then
exposed to
UV/O
3

treatment

to activate
the polymer surface

and lower

the glass transition temperature
of the PMMA
.

The UVO Cleaner

(Jelight
Company Inc, Mason Irvine, California) was turned
on and warmed up for 10min on and
the oxygen flow rate set to 1.0L/min.

The oxygen flow rate was lowered to 0.5L/min for
the treatments. PMMA c
hips were placed channel or fiber side up on the tray in the UVO
Cleaner
and PMMA chips with channels were treate
d for

10 minutes
and
PMMA chips
with nanofibers were treated for
6 minutes
.

Exhaust time was set to 2 minutes.

The two PMMA chips were then sandwiched between two copper plates and
bonded for 10 minutes at
80°
C
under

4000lb of pressure
.

Tubing was cut
,su
per
-
glued
into the inlet and outlet holes and allowed to dry overnight.



Figure

2
: Fabrication Scheme (a) PMMA

chip

sandwiched between the copper master
templa
te and another copper plate (b) Sandwich warmed on hot plate for 5 minutes at
120°C then placed under
5000lb pressure at 120
°
C for 10

minutes (c) PMMA
chip
with

embossed channels (d) nanofiber mat

spun onto another PMMA chip

(e)
PMMA chip
s

with
nanofiber m
at and
embossed channels
placed between two copper plates and
bonded
at 80°C for 10 minutes

under 4000lb of pressure

(e)
final
device with nanofibers
enclosed

in microflu
idic channels


3.5
Electro
-
spinning

Electrospun nanofibers

were produced by Dr. Daehwan Cho from the Frey Group,
Cornell University Fiber Science Department and by Lauren Matlock
-
Colangelo from the
Baeumner Group, Cornell University Biological and Environmental Engineering
Department.
Spinning dope solutions were

prepared by
dissolving

10 wt% PVA polymer
in DI water. Fu
nctional polymers
,

P
olybrene and

PolyMA,
were added to the spinning
dope
for positive and negatively charged nanofibers respectively.

A 5 mL plastic
syringe
with an 18 gauge needle
connected to a pu
mp was used to eject the spinning dope
.

A
volta
ge
of 12
-
15kV was applied to the
needle
tip
and fibers were collected on grounded
PMMA chips as shown in Figure 3
.

Figure 3: Electrospinning Setup and PMMA collecting chip with gold electrode array
.



The
distance between the collector and the n
eedle tip was
10
-
15 cm
. Electrospinning was
performed

at room temperature.

Gold electrode arrays on the PMMA collecting chip
were necessary
to direct the fibers to land on the chip and for

the fibers to attach to the
PMMA
surface. The electrode arrays were
produced by Lauren Matlock
-
Colangelo in the
CNF.
The

process for patterning Au electrodes on PMMA using gold
-
thiol chemistry has
been described previousl
y
11
.


3.
6
Method 1:
Quantitative
Cell Counts

E.coli

was grown overnight to a concentration of 10
9

CFU/ml and then diluted to 10
3

CFU/mL in PBS at
pH7
. A 1
00

µ
L sample
containing approximately 100 CFU
was
pumped into the device at 1μL/min. Effluent was collected at the outlet and plated on

TSA pl
ates to determine the number of cells that passed through the device and
indirectly, how many cells were retained by the fibers
. For blocking experiments, 1%

BSA
was pumped through the channels at 1μL/min for 3

hours and then the solution was
flushe
d out by p
umping air through the device.



3.
7
Method 2: Epi
fluorescent Microscopy

E.coli

was grown overnight to a concentration of 10
9

CFU/ml

and then stained with
Syto9

green fluorescent nucleic acid dye with excitation 485 nm and emission 498 nm
.

The
E.coli

cells were centrifuged and washed to remove the nutrient broth. A 1mL
aliquot of culture was centrifuged at 10,000rpm for 1 minute, the TSB was discarded and
the cells re
-
suspended in 1mL of
the sample solution
. The cells were centrifuged again,
the

supernatant discarded and the cells re
-
suspended in 1mL
of the sample solution
.
Sample solutions tested were
pH7

PBS buffer,
pH4

acetate buffer and apple juice.
Cells
were stained with Syto9 nucleic acid stain by adding 1
µL

Syto 9 for every 300
µL

of
undiluted cell culture. A 1:10 dilution
in
of the stained cells was used to provide a
concentration high enoug
h to saturate the fibers but prevent

clog
ging

the channel inlet.

For the fluorescent measurements,
a FITC 31001 filter (Chroma Technology) wit
h
excitation 480

nm
, emission 535

nm

was used.

Image acquisition took 3 seconds. T
he
microfluidic devices were placed underneath the microscope
(Leica Microsystems,
Weltzar, Germany)
for the duration of the experiment. Images were captured using RS
Image s
oftware

every 5 minutes
.

The sample spiked with stained
E
.coli

cells was pumped
through the device at 1μL/min for 20 minutes. The syringe was then switched to the wash
buffer, either a
pH4

buffer,
pH7

buffer, or apple juice. The wash buffer was pumped into
the device at 1μL/min. The tubing was cut such that the wash buffer would reach the
channel at around the 40 minutes from the start of the experiment. Images were analyzed
in ImageJ software. The mea
n gray value was used to determine fluorescence intensity.


4.0
RESULTS


4.1
Apple Juice Dynamic Light Scattering

Commercial apple juice is filtered to remove sediment however protein
-
polyphenol hazes
can develop in clarified fruit juices especially
in
juice where pectin content is high.
22

Particles larger than 10μm suspended in apple juice may clog the tubing, channels or
fibers

in microfluidic devices
. Therefore, a

DLS

analysis was performed on pa
steurized
clarified apple juice
to determine whether filt
ration or centrifugation of the apple juice
would be needed

DLS

measures fluctuations in the intensity of
scattered

light caused by
random Brownian
motion o
f

particles in a solution
.

Figure 4 shows that there are 2
groups
of particles present in apple ju
ice. There are small particles less than 1μm in
diameter and large particles that range between 10
-
50μm in diameter.


Figure 4:
Size distribution of particles in apple juice
.




0
20
40
60
80
100
120
1
10
100
1000
10000
Intensity

Diameter, nm

T
he
volume fraction of particles

vs
.

particle
diameter

in the apple juice sample

was
plotted
in Figure

5
.
Number corresponds to the volume fraction which is the solids
volume divided by the total volume. The graph shows that the volume fraction of
particles with diameters ≤ 1nm was 80%. while the volume fract
ion of particles with
diameters between 10
-
50μm was very small, approximately 1%, t
he results confirm that
before testing actual apple juice samples, a pre
-
filtration or centrifugation
would be
needed

in case there are large particles present in the sample
.


Figure 5:
Graph of number (
volume of solids/ total volume)

of particles

according to
particle
diameter in apple juice

samples.
The bars represent the volume fraction of
particles at that diameter.
For example, the selected bin (green bar) corresponds to the
volume fraction of particles with 1nm diameters. Particles with diameters between 10
-
50μm make up approximately 1% volume fraction of the sample.
The red line is a graph
of cumulative volume frac
tion vs. particle diameter.
The cumulative number
for particles
with diameters ≤ 1nm
is represented on the graph as the green horizontal line and
indicates that approximately 80% o
f particles in the solution are ≤ 1nm in diameter.


4.2
Results
:
Quantitativ
e Cell Counts

E.coli

has a net negative charge at neutral pH
1

therefore
,

in theory at these conditions
bacterial

cells should adhere to positively charged fibers and pass through neutral fibers
.

PBS buffer at
pH7

was spiked with
E.coli

and filtered thro
ugh
channels with no fibers,
positively charged fibers and
neutral fibers
. A

flow rate of 1μL/min

was chosen
to
prevent the fibers
from being

washed out of the channel and because no noticeable
deterioration of the fibers was observed at this flow rate.
Th
e

preliminary
results in
Figure 6 show that almost all
of
the cells were
captured by

the positive fibers
. Only 55
cells were recovered at the outlet of the channel with positive fibers
whereas

there were
close to 1000 cells
recovered at the outlet of
the empty channel and
at the outlet of
the
channel with
the
neutral fibers
.




Figure 6:

Preliminary results of nanofiber filtration. C
ell
s
were
collected at the outle
t and
plated on TSA. The sample volume was 1000μL and contained approximately 10
3

CFU
of
E.coli
. A flow rate of 1μL/min was used.


When negative fibers became available, the experiment was repeated and the
results recorded in Figure 7. To shorten the run time, the sample volume was reduced

to
100 μL
which also lowered the initial conce
ntration of cells.
Positive fibers were shown
to retain 98 ± 1.6% of
E.coli

that passed through the microfluidic channels whereas
negative fibers only retained 35 ± 1.1%. In the channels with no fibers only 53 ± 4.4% of
cells were recovered at the outlet.
Cell losses could be attributed to cells adhering to the
hydrophobic channel and tubing surfaces or cells becoming trapped at the inlet or outlet
0
200
400
600
800
1000
1200
Initial Cell
Concentration
No Fibers
Neutral
Positive
CFU

holes.
Studies have shown that the isoelectric point of various bacteria range from 1.7


3.8 depending on the

cell surface balance of carboxyl and ammonium groups and
that
the
iso
e
lectric point affects adhesion to hydrophobic polymer surfaces
23
. A 1% BSA solution
was used to pre
-
coat the
plastic
tubing and
the
channel and the experiment was repeated.
Coating the
channel and tubing with 1% BSA prior to experiments increased the recovery
at the outlet to 83 ±15 % of the cells.
BSA has an isoelectric point at 5.4 and is negatively
charged at neutral pH
24
.
When the
tubing and channel with positive fibers was coated
with 1% BSA, any positive fibers in the channel
were
also saturated with BSA,
preventing
E.coli

from binding to the fibers during subsequent filtration
. T
herefore a high
number of cells at the ou
t
let
were recovered

when 1% BSA was used to coat channels
wit
h positive fibers as shown in Figure 7. Cell losses were not observed in
the channels
with neutral and negative fibers possibly because the fibers fill the channel and prevent
the bacterial cells from coming in contact or settling on the channel surfaces.



Figure 7:
Nanofiber filtration of
E.coli

in channels with i) no fibers ii) no fibers with 1%
BSA coating iii)
negative fibers iv)
positive
fiber
s
v)

positive fibers with 1% BSA
coating. All experiments performed in triplicate except for the positive fib
er channel with
BSA coating.


0
20
40
60
80
100
120
Initial
Number of
Cells
No fibers
No Fibers,
1% BSA
coating
Negative
Fibers
Positive
fibers
Positive
Fibers, 1%
BSA
coating
CFU

4.3
Results:
Epifluorescent
Microscopy

While fluorescent measurements are qualitative, the benefit of this assay
was

that it
took only an hour

to complete whereas t
he

cell count assays on TSA plates
took

2 days
.
E.coli

cells
were stained with Syto9, a
cell
-
permeant green fluorescent
nucleic acid
stain
.

In order for the cells to be visible on the fibers and for the fibers to become saturated, a
high concentration of cells, approximately 7
-
8 logs was required. The
microfluidic
devices were placed underneath the microscope
and i
mages were captured
every 5
minutes
using
the
RS Image software.

In each
experiment
, t
he initial sample was injected
at a flow rate of 1 μL/min. The syringe was switched to the wash buffer at
20 minutes
,
the
wash buffer reached the channel at 40 minute

and was allowed to run for another 20
minutes
.

The result
s of the experiments testing
p
H
7
samples
were recorded in Figure 8.



Figure 8
:
Graph of mean gray value over time. The initial sample is
pH7

PBS buffer
spiked with approximately 10
7
-
10
8

CFU/mL
E.coli

stained with Syto9.
The syringe was
switched to
a
wash buffer

(either pH7 of pH4)

at 20
min

and the buffer

reached

the
channel at 40 mi
n
. The

data shown in blue is an average of two replicates.
The red and
orange data are single replicates.

0
2
4
6
8
10
12
14
16
0
10
20
30
40
50
60
70
Mean Gray Value

( + ) Fibers ph4 wash
( + ) Fibers ph7 wash
( - ) Fibers ph7 wash
A
s
explained previously
,
E.coli

have a net negative charge at neutral pH. In

the

experiments
shown in Figure 8
,
the negatively charged bacteria accumulated on the
positive fibers within the first few minutes.
The fluorescence intensity was maintained
when the fibers (red squares) were washed with a
pH7

buffer for 20 minutes. The fibers
(blue diamonds) that were wash
ed with a
pH4

buffer for 20 minutes caused the
fluorescence to drop by 84% indicating that a majority of the bacteria were eluted from
the positive fibers by washing with
pH4

buffer.
In the case of negative fibers (orange
circles
), the negatively charged b
acteria were repelled and bacteria were easily flushed
from the channel with a
pH7

buffer shown by the drop in fluorescence intensity.
The
fluorescent images taken at 0, 20, 40 and 60 minutes for the blue
diamond data

shown in
Figure 8 are shown in Figure
9
.











(a)





(b)








(c)






(d)

Figure 9
:

Epifluorescence images corresponding to
blue data shown in Figure 8
.
E.coli

was filtered out of a
pH7

PHS buffer using positive fibers and then washed with a
pH4

buffer to elute the
bacteria
. The images correspond to (a) 0 minutes, (b) 20 minutes
(c) 40 minutes and (d) 60 minutes. The wash buffer reaches the channel at 40 minutes.

After the 20 minutes of
washing with a
pH4

buffer there is a visible reduction in
fluorescence intensity in the channel (Figure 9d vs. 9c).
The fluorescent images taken at 0,
20, 40 and 60 minutes for the
red

square
data shown in Figure 8 are shown in Figure
10
.

After 20 minutes o
f washing with a
pH7

buffer there is no noticeable change in the
fluorescence in the channel (Figure 10d vs. 10c).








(
a)







(b)




(
c
)






(d
)

Figure 10
: Epifluorescence images corresponding to
red data shown in Figure 8
.
E.coli

was filtered out of a
pH7

PHS buffer using positive f
ibers and then washed with a
pH7

buffer
. The images correspond to (a) 0 minutes, (b) 20 minutes (c) 40 minutes and (d) 60
minutes. The wash buffer reaches the channel at 40 minutes.

In the

second group of filtration experiments
E.coli

samples
were prepared in
pH4

buffer
, filtered through negative and positive fibers and then fibers were washed with a
pH4 buffer
. In these expe
riments,
E.coli

were expected to have a net positive charge
because of the low pH of the sample buffer. These filtration results were then compared
to filtration of apple juice samples spiked with
E.coli

(
Figure 11
)
. From the DLS analysis
of apple juice it

was confirmed that there are large particles greater than 10μm in
diameter present in apple juice. For that reason, the apple juice used in the experiments
was filtered through Whatman grade 1 filter paper prior to addition of the
E.coli.


E.coli

in
pH4

buffer (squares) attached better to negative fibers (solid squares)
producing average fluorescence intensity 1.6 times higher than the channels with positive
fibers (open squares). The results were repeated with apple juice spiked with
E.coli

(triangles)
producing similar results. The negative channels (solid triangles) produced
fluorescence intensity 2.0 times higher than in the positive channels (open triangles). The
higher overall fluorescence in the apple juice samples may be due to the Syto9 dye
inter
acting with compounds in the apple juice.


Figure 11
: Images were analyzed in ImageJ. Mean gray value was plotted over time.
The sample buffers and wash buffer were both

pH4
.
T
he
syringe was switched to the
wash buffer at 20 minutes and the
buffer
reached

the channel at 40 minutes. Apple juice
was filtered using 100
W
h
atman

filter paper to remove any large particles.
Experiments
with
pH4

buffer sample filtered through negative fibers (solid square) were performed in
triplicate,
pH4

buffer sample filtered through positive fibers (open square)
was

an
average of 2 replicates. The apple juice experiments
were

both single replicates.


-5
0
5
10
15
20
25
30
35
40
45
0
10
20
30
40
50
60
70
ph4 buffer ( - ) fiber
ph4 buffer ( + ) fiber
Apple juice ( - ) fiber
apple juice ( + ) fiber
5.0
CONCLUSION AND FUTURE WORK


The results of both the cell counting and
epi fluorescence

experiments s
how that
E.coli

in a neutral pH sample can be filtered onto positively charged fibers and
are
repelled by negatively charged fibers.
E.coli

in neutral pH samples that are filtered onto
positive fibers can be eluted by washing with a buffer with a lower pH.
E.coli

in
pH4

buffer sample can be filtered onto negatively charged fibers

and are repelled from
positively charged fibers
.
The samples wi
th actual
apple juice did not
negatively
affect
the charge interaction and filtration. However, apple juice was found to contain large
particles that could potentially clog microfluidic channels. Therefore a pre
-
filtration step
to remove large particles be

performed.

These results show the potential for electrospun nanofibers to be used on
-
chip as bio
-
separators. Due to time constraints, some experiments in this study were not repeated in
triplicate. Further studies should be done to confirm the results. Th
e design of the
microfluidic device can be improved by adding a second inlet hole or using a T
-
junction
syringe for introducing the wash buffer into the channel. This would prevent drastic
changes in pressure and backflow when the syringes are switched. If

cell losses are
apparent, coating the PMMA surface with 1% BSA may be necessary. Coating the
channels by pumping 1% BSA through the device as done in this study is time
consuming. Therefore coating the PMMA surface prior to bonding is another method that
could be investigated. The quality of electrospun nanofibers can vary greatly due to small
changes in electrospinning parameters

and environmental conditions
. In order to design
nanofibers for filtration applications, it
may be useful

to determine
a correl
ation

between
spinning time and fiber pore size and
fiber mat
density.

REFERENCES





1

Poortinga AT, Bos R, Norde W, and HJ Busscher
,

Electric double layer interactions in

bacterial adhesion to surfaces
,

Surf

Sci Rep
,

2002, 47:1
-
32.


2

Scallan E, Hoekstra RM, Angulo FJ, Tauxe RV, Widdowson MA, Roy SL,
et al
,

Foodborne illness acquired in the United States
-
major pathogens
,

Emerg Infect
Dis
,
2011, 17: 7
-
15.


3

Scallan E, Griffin PM, Angulo FJ, Tauxe RV, Hoekstra RM
,

Foodborne illness acquired

in the United States
-
unspecified agents
,

Emerg Infect Dis
,

2011, 17: 16
-
22.


4

Mead PS, Slutsker L, Dietz V, McCaig LF, Bresee JS, Shapiro C,
et al
,

Food
-
related

illness

and death in the United States
,

Emerg Infect Dis
, 1999, 5: 607
-
24.


5

Cook

N
, T
he use of NASBA for the dete
ction of microbial pathogens in
food and

environmental samples
,

Journal of Microbiological Methods
,

2003,
53
: 165
-
174
.


6

Baeumner AJ
,

Biosensors
for environmental pollutants and food contaminants
,

Anal


Bioanal Chem
,

2003, 377: 434
-
445
.


7

Brehm
-
Stecher B, Young C, Jaykus L and ML Tortorello
,

Sample Prepa
ration: The
forgotten beginning
,

Journal of Food Protection
, 2009, 72(
8
)
: 1774
-
1789.

8

Benoit PW and DW Donahue
,

Methods for rapid separation and concentration of
bacteria in food that bypass time
-
consuming cultural enrichment
,

Journal of Food
Protection
,

2003, 66(
10
)
: 1935
-
1948.

9

Stevens KA and L Jaykus.
,

Bacterial separation and concentr
ation from complex sample
matrices: A review
,

Critical Reviews in Microbiology
,
2004, 30.1: 7
-
24.

10

Burger C, Hsiao BS and B Chu
,

Nanofibrous materials and their applications
,

Annual
Review of Materials Research
,
2006, 36: 333
-
368.

11

Li D, Fr
ey MW, Vynias
D and AJ Baeumner
,
Availability of biotin incorporated in
electrospun PLA fibers for streptavidin binding
,

Polymer
,

2007, 48: 6340
-
6347.

12

Li D, Frey MW, YL Joo
,

Characterization of nanofibrous membranes with capillary
flow porometry
,

Journal of Membrane S
cience
,

2006, 286: 104
-
114.

13

de Jong, J, Lammertink RGH and M Wessling
,

Membranes and microfluidics: a
review
,

Lab on a chip
,

2006
,

6: 1125
-
1139.

14

Becker H and LE Locascio
,

Polymer microfluidic devices
,

Talanta
, 2002, 56(
2
)
: 267
-

287.








15

Nugen S,
Asiello P, Baeumner A
,

Design and fabrication of a microfluidic device for

near
-
single cell mRNA isolation using a copper hot embossing master
,

Microsystems Technology
,

2008
,

15
:

477
-
483.


16

Nugen S, Asiello P, Connelly J, Baeumner A
,

PMMA biosensor for nucleic acids with

integrated mixer and electrochemical detection
,

Biosensors and Bioelectronics
,

2008
, 24: 2428
-
2433.


17

Tsao CW, Hromada L, Liu J, Kumar P and DL DeVoe
,

Low temperature bonding

of PMMA and COC microfluidic substrates

using UV/ozone surface treatment
,

Lab Chip
,

2007, 7: 499
-
505
.


18

Vojdan
i JD, Beuchat, LR and RV Tauxe
,

Juice
-
associated outbreaks of human illness
in the United States, 1995 through 2005
,

Journal of Food Protection
, 2008,
71(
2
):

356
-
364.

19

Veleirinho B and JA Lopes
-
da
-
Silva
,
Application of electrospun poly (ethylene
terephthalate) nanofiber mat to apple juice clarification
,

Process Biochemistry
,

2009,

44
:

353
-
356.

20

Ogunjimi AA and PV Choudary
,

Adsorption of endogenous polyphenols relieves the
inhibition by fruit juices and fresh produce of immuno
-
PCR detection of
Escherichia coli

0157:H7,
REMS Immunology and Medical Microbiology
,

1999
,

23
:

213
-
220
.

21

Shahidi
F,

“Non
-
carbonated fruit juices and
fruit beverages” in
Phenolics in food and
nutraceuticals
, CRC Press LLC,
2004.

22

Siebert KJ, Carrasco A

and PY Lynn
,

Formation of protein
-
polyphenol haze in

B
everages
,

J Agric

Food Chem
,

1996
,
44

(8):1997
-
2005.


23

Rijnaarts
et al
,

The isoelectric

point of bacteria as an indicator for the presence of cell

surface polymers that inhibit adhesion
,

Colloids Surfaces B: Biointerfaces
,
1995
,

4: 191
-
197
.


24

Shi Q, Zhou Y, and Y Sun
,

Influence of pH and Ionic Strength on the Steric

Mass
-
Action Model Para
meters around the Isoelectric Point of Protein
,

Biotechnol

Prog
,

2005, 21: 516
-
523.