Biophysical studies of passive transport across synthetic lipid bilayers

PDF

Download

Open document

Flip pages

Copy asset link

Request this asset

Transcript (if available)

Content

Biophysical
Studies
of
Passive

Transport
Across
Synthetic
Lipid

Bilayers

Kristina
Runas

Committee:

Dr.
Noah
Malmstadt
(Committee
Chair)

Dr.
Muhammad
Sahimi

Dr.
Michelle
Povinelli

Doctoral
Dissertation

Mork
Department
of
Chemical
Engineering
and
Materials
Science

University
of
Southern
California,
Los
Angeles,
CA

August
2015

2

Table
of
Contents

1
Background
........................................................................................................................
4

1.1
The
biological
role
of
plasma
membrane
permeability
............................................
4

1.2
Advantages
of
biomimetic
membranes
for
measuring
transport
.........................
5

1.3
Theoretical
models
of
lipid
bilayer
transport
..............................................................
6

1.3.1
Overton’s
Rule
and
the
solution-‐diffusion
model
..............................................................
6

1.3.2
Molecular
dynamics
simulations:
complexity
of
membrane
structure
....................
8

1.4
Experimental
techniques
for
measuring
membrane
permeability
......................
8

1.4.1
The
unstirred
layer:
artifacts
and
effects
on
membrane
transport
...........................
9

1.4.2
Parallel
Artificial
Membrane
Assay
(PAMPA)
...................................................................
11

1.4.3
Steady
state
methods
for
measuring
permeability
........................................................
12

1.4.4
Transient
methods
for
measuring
permeability
.............................................................
13

1.5
Advantages
and
disadvantages
of
choosing
a
measurement
technique
...........
14

2
Low
levels
of
lipid
oxidation
radically
increase
the
passive
permeability
of

lipid
bilayers
..........................................................................................................................
15

2.1
Motivation
..............................................................................................................................
15

2.1.1
Importance
of
oxidation
in
cells
.............................................................................................
15

2.1.2
Effects
of
lipid
oxidation
on
membrane
properties
.......................................................
16

2.2
Materials
and
Methods
......................................................................................................
18

2.2.1
Materials
...........................................................................................................................................
18

2.2.2
GUV
composition
and
formation
............................................................................................
19

2.2.3
Preparation
of
test
molecule
....................................................................................................
20

2.2.4
Microfluidic
channel
preparation
and
design
...................................................................
22

2.2.5
Transport
experiment
protocol
..............................................................................................
23

2.2.6
Examination
of
pore
formation
...............................................................................................
24

2.3
Data
Analysis
.........................................................................................................................
24

2.3.1
Background
subtraction
and
flat-‐fielding
..........................................................................
24

2.3.2
Removal
of
pinhole
crosstalk
contributions
.....................................................................
25

2.3.3
Finite
difference
model
and
membrane
permeability
..................................................
33

2.4
Results
and
Discussion
......................................................................................................
37

2.4.1
Comparison
of
capture
methods:
mechanical
vs.
chemical
trapping
.....................
37

2.4.2
Permeability
of
vesicles
with
low
levels
of
oxidation
...................................................
38

2.4.3
Pore
formation
in
vesicles
with
high
levels
of
oxidation
.............................................
41

2.4.4
Relationship
between
permeability
and
vesicle
diameter
..........................................
43

2.4.5
Impact
of
biotin-‐DPPE
on
measured
permeability
.........................................................
45

2.4.6
Discussion
........................................................................................................................................
46

2.5
Conclusion
..............................................................................................................................
49

3
The
addition
of
cleaved
tail
fragments
during
lipid
oxidation
stabilizes

membrane
permeability
behavior
.................................................................................
50

3.1
Motivation
..............................................................................................................................
50

3.2
Methods
..................................................................................................................................
54

3.2.1
Materials
...........................................................................................................................................
54

3.2.2
Vesicle
composition
and
formation
.......................................................................................
55

3.2.3
Preparation
of
test
molecule
....................................................................................................
56

3

3.2.4
Microfluidic
channel
preparation
and
design
...................................................................
56

3.2.5
Transport
experiment
protocol
..............................................................................................
57

3.2.6
Pore
formation
protocol
.............................................................................................................
58

3.2.7
Measurement
of
bilayer
thickness
.........................................................................................
58

3.3
Data
Analysis
.........................................................................................................................
59

3.3.1
Filtering
capacitance
measurements
using
a
Fourier
transform
approach
........
59

3.4
Results
and
discussion
.......................................................................................................
62

3.5
Conclusion
..............................................................................................................................
72

4
Crossing
into
the
liquid-‐liquid
immiscibility
region
causes
a
drastic
change

in
the
permeability
of
lipid
bilayers
..............................................................................
73

4.1
Motivation
..............................................................................................................................
73

4.2
Materials
and
methods
......................................................................................................
76

4.2.1
Materials
...........................................................................................................................................
76

4.2.2
GUV
composition
and
formation
............................................................................................
77

4.2.3
Preparation
of
test
molecule
....................................................................................................
79

4.2.4
Microfluidic
channel
preparation
and
design
...................................................................
80

4.2.5
Transport
experiment
protocol
..............................................................................................
80

4.3
Data
Analysis
.........................................................................................................................
81

4.3.1
Using
map
projection
techniques
to
measure
surface
area
and
perimeter
of
the

liquid
disordered
phase
.............................................................................................................................
81

4.4
Results
.....................................................................................................................................
83

4.4.1
Measured
permeability
for
vesicles
with
no
microscopic
phase
separation
......
83

4.4.2
Permeability
of
binary
vesicle
compositions
....................................................................
85

4.4.3
Permeability
of
vesicles
with
microscopic
phase
separation
....................................
86

4.4.4
Examining
the
relationship
between
membrane
permeability
and
disordered

phase
surface
area
........................................................................................................................................
89

4.4.5
Examining
the
relationship
between
membrane
permeability
and
disordered

phase
perimeter
............................................................................................................................................
91

4.5
Discussion
..............................................................................................................................
93

4.6
Conclusions
............................................................................................................................
95

5
Summary
and
Outlook
.................................................................................................
96

6
Acknowledgements
.......................................................................................................
98

7
References
........................................................................................................................
99

4

1 Background

While
there
are
numerous
active
mechanisms
controlling
which
molecules
cross
a

cell’s
plasma
membrane
to
enter
the
cytoplasm,
these
are
not
the
only
routes
by

which
molecules
can
enter
cells.
In
fact,
transport
by
passive
diffusion
across
the

lipid
bilayer
of
the
plasma
membrane
represents
a
nearly
universal
mechanism
of

molecular
entry
of
drugs
and
environmental
toxins.
Understanding
the
barrier

properties
of
the
lipid
bilayer
and
what
molecular
characteristics
control
its

permeability
is
therefore
of
fundamental
interest
to
toxicology
and
drug

development.
Today,
sophisticated
biomimetic
cell
membranes,
coupled
with

advanced
analytical
tools
and
computer
modeling,
allow
new
insight
into
this

important
biological
property.

1.1 The
biological
role
of
plasma
membrane
permeability

For
an
orally
delivered
drug
to
reach
systemic
circulation,
it
must
first
diffuse
across

the
mucus
gel
layer,
intestinal
epithelial
cells,
the
lamina
propria,
and
finally,
the

capillary
endothelium.
1

The
route
that
orally
administered
drugs
generally
take

from
the
gut
into
circulation
involves
passive
diffusion
through
epithelial
cell

membranes
rather
than
interaction
with
proteins
or
other
active
modes
of

transport.
2

Environmental
toxins
are
also
capable
of
permeating
the
cell

membrane.
3

Species
capable
of
passive
transport
across
the
plasma
membrane

include
heavy
metals,
4

cyclic
hydrocarbons,
5

polymer
nanoparticles,
6,7

fullerenes,
8

semiconductor
quantum
dots,
9

and
gold
nanoparticles.
10

Understanding
the

characteristics
and
mechanisms
by
which
toxins
bypass
the
cell
membrane
can
lead

5

to
a
more
complete
understanding
of
environmental
toxicity
of
chemicals
and

materials.

1.2 Advantages
of
biomimetic
membranes
for
measuring
transport

Despite
the
multiple
interfaces
that
drugs
must
cross
during
adsorption,
a
single

layer
of
intestinal
epithelial
presents
the
most
significant
barrier
to
passive

diffusion.
11

As
such,
one
of
the
most
common
techniques
for
assaying
the
potential

of
a
drug
to
be
absorbed
is
by
measuring
its
permeability
though
a
monolayer
of

Caco-‐2
cells.
Caco-‐2
is
a
colorectal
adenocarcinoma-‐derived
cell
line
that
can
be

cultured
such
that
the
cells
differentiate
to
resemble
the
epithelium
of
the
small

intestine.
12

In
the
Caco-‐2
permeability
assay,
a
cell
monolayer
is
cultured
between
a

donor
chamber
and
a
receptor
chamber,
with
the
solute
of
interest
(i.e.
the
drug

candidate)
added
to
the
donor
chamber.
13

Samples
are
removed
from
each
chamber

at
regular
intervals
for
analysis
to
develop
a
time
course
of
permeation,
and
solute

concentration
is
measured
via
radiolabeling,
14

UV
spectrophotometry,
15

or

HPLC.
1,16,17

Cell
based
systems
such
as
the
Caco-‐2
assay
have
significant
shortcomings
as
tools

to
understand
mechanisms
of
transmembrane
permeability.
Caco-‐2
cell
membranes

are
significantly
less
controlled
than
biomimetic
membranes
formed
from
artificial

bilayers:
permeation
pathways
can
include
active
and
paracellular
transport
as
well

as
passive
transport.
13

The
heterogeneity
of
Caco-‐2
cells
leads
to
variation
in
cell

shape,
size,
and
multilayer
formation.
18

For
the
purpose
of
rapid
screening
drug

candidates,
Caco-‐2
and
other
such
cell
assays
are
impractical,
and
as
such,
it
is

6

necessary
to
find
a
low
cost,
high-‐throughput
biomimetic
cell
membrane
system
to

support
rapid
solute
permeation
analysis.
13

1.3 Theoretical
models
of
lipid
bilayer
transport

Theoretical
constructs
describing
passive
transport
are
important
both
in
analyzing

data
and
in
making
the
ultimate
mechanistic
connections
between
molecular

structure
and
membrane
permeability.
Descriptive
theories
of
passive
transport

include
transport
modeling
of
the
permeation-‐diffusion
process,
simulations
of

molecular
interactions
between
permeating
molecules
and
the
bilayer,
and

analytical
treatments
of
the
permeation
mechanism.

1.3.1 Overton’s
Rule
and
the
solution-‐diffusion
model

Early
theoretical
exploration
of
membrane
permeation
was
driven
by
the

widespread
acceptance
of
Overton’s
rule,
which
indicates
that
the
membrane

permeability
of
a
drug-‐like
molecule
scales
with
its
lipophilicity.
19-‐21

The

permeability
coefficient
(P)
is
a
constant
of
proportionality
between
the
flux
of
a

molecular
species
through
the
membrane
and
the
concentration
gradient
of
that

species
across
the
membrane.
The
key
consequence
of
Overton’s
rule
is
that
more

lipophilic
molecules
are
better
drug
candidates,
as
they
have
increased
oral

bioavailability.
One
way
of
representing
the
lipophilicity
of
a
molecule
is
through
the

oil/water
partition
coefficient
(K).
The
partition
coefficient
of
a
solute
is
expressed

as
the
ratio
of
its
solubility
in
an
oil
to
its
solubility
in
water.
21

The
most
common
oil

used
to
estimate
solubility
in
lipid
systems
is
octanol,
and
as
such,
the
octanol/water

partition
coefficient
is
often
used
as
a
predictor
of
permeability.
Increased
octanol

7

solubility
corresponds
to
increased
partitioning
into
the
lipid
phase.
Typically,

Overton’s
rule
is
interpreted
as
a
relationship
between
oil/water
partition

coefficient
and
permeability
coefficient.

The
simplest
theoretical
interpretation
of
the
observations
summarized
by

Overton’s
rule
is
the
solution-‐diffusion
model,
which
treats
the
membrane
as
a

homogenous
hydrocarbon
slab.
22

The
solution-‐diffusion
model
assumes
all

transport
is
diffusive
and
that
there
is
a
negligible
concentration
gradient
on
either

side
of
the
membrane.
The
molar
flux
across
the
membrane
can
then
be
expressed

using
Fick’s
Law.
23

Molar
flux
J
through
the
membrane
can
be
expressed
in
terms
of

the
permeability
P
and
the
concentration
gradient,
where
co
is
the
concentration

outside
the
membrane,
and
ci
is
the
concentration
inside
the
membrane.

𝐽=𝑃 𝑐
!
−𝑐
!

(1)

Equilibrium
partitioning
of
the
species
between
solution
and
the
membrane
is

described
by
the
partition
coefficient
K:
the
equilibrium
ratio
of
concentration

dissolved
in
solution
to
concentration
dissolved
in
the
membrane.
Overton’s
rule

then
expresses
the
permeability
P
is
as:

𝑃=
𝐾𝐷
𝑙

(2)

Since
K
tends
to
vary
on
a
per-‐species
basis
much
more
than
D,
and
l
is
roughly

constant
for
all
plasma
membranes,
the
partition
coefficient
controls
and
can
be

8

used
to
predict
membrane
permeability.
22,24-‐27

This
is
the
crux
of
Overton’s
rule:

strong
lipophilic
partitioning
leads
to
high
permeability.

1.3.2 Molecular
dynamics
simulations:
complexity
of
membrane
structure

Although
Overton’s
rule
may
be
capable
of
predicting
permeability
of
many
small

molecules,
the
simplification
of
the
cell
membrane
into
a
single
solid
slab
does
not

accurately
represent
the
membrane’s
complex
structure
and
heterogeneity.
In

particular,
treating
the
membrane
as
a
uniform
hydrocarbon
ignores
the
charge

state
of
the
lipid
head
groups,
the
presence
of
membrane
proteins,
lateral
phase

segregation
in
the
membrane,
and
variations
of
membrane
fluidity.
For
many

molecules
of
varying
charge
state,
size,
or
hydrogen
bonding
capacity,
the
solution-‐
diffusion
model
may
not
be
adequate
to
accurately
estimate
membrane

permeability.
Molecular
dynamics
simulations
have
been
used
to
model
the
bilayer

as
a
more
complex
system
by
dividing
it
into
different
regions
of
transport.
28-‐35

By

treating
the
bilayer
as
a
more
complex
system
than
single
component
slab,
it
is

possible
to
more
accurately
model
the
transport
process.
30

1.4 Experimental
techniques
for
measuring
membrane
permeability

Isolating
and
understanding
the
molecular
processes
that
underlie
passive
transport

require
well-‐controlled
experimental
systems.
Precise
control
of
lipid
bilayer

structure
and
composition,
for
instance,
allows
for
the
lipid-‐driven
processes
in

passive
transport
to
be
isolated,
something
that
is
impossible
in
more
complex

systems
such
as
the
Caco-‐2
assay
described
in
Section
1.2.
Synthetic
bilayer
models

such
as
liposomes
and
planar
lipid
bilayers
were
first
introduced
in
the
1960s.
36,37

9

There
are
numerous
advantages
to
measuring
transport
in
biomimetic
lipid
bilayer

systems,
such
as
the
availability
of
a
variety
of
methods
for
measuring
the
solute

flux.
These
techniques
can
usually
be
divided
into
two
categories:
steady-‐state

measurements,
and
transient
measurements.
For
steady-‐state
experiments,
the

donor
side
of
the
membrane
is
treated
as
an
infinite
source
of
solute
at
fixed

concentration
and
the
receptor
is
treated
as
an
infinite
sink.
This
means
that
the

concentration
profile
across
the
membrane
does
not
change
with
respect
to
time,

and
steady-‐state
assumptions
and
analytical
equations
can
be
used
to
determine

permeability.
In
a
transient
experiment,
the
concentration
profile
varies
with

respect
to
time,
allowing
for
permeability
analysis
based
on
a
concentration
vs.
time

curve.
Each
of
these
categories
has
positive
and
negative
attributes.
The
ideal

experimental
technique
would
allow
for
high-‐throughput
data
collection
of
highly

controlled
systems
without
artifacts.

1.4.1 The
unstirred
layer:
artifacts
and
effects
on
membrane
transport

Prior
to
discussing
the
various
techniques
for
measuring
membrane
permeability,
it

is
important
to
understand
a
major
feature
they
have
in
common.
These
approaches

all
must
cope
with
the
existence
of
an
unstirred
layer
(USL),
or
a
region
of
static

fluid
close
to
the
membrane
in
which
the
diffusion-‐driven
transport
of
the
permeant

leads
to
a
non-‐uniform
concentration.
The
USL,
indicated
in
gray
in
Figure
1,
is

defined
as
the
region
in
which
the
concentration
of
the
solute
cannot
be
assumed

equal
to
the
bulk
concentration.
While
it
is
easy
to
measure
the
bulk
concentrations

on
either
side
of
the
membrane
(co
and
ci
in
Figure
1),
it
is
the
concentrations
closest

to
the
membrane
(c
‘
Mo
and
c
‘
Mi
in
Figure
1)
that
determine
the
driving
force
for

10

membrane
transport.
Thus,
it
is
necessary
to
be
able
to
model
the
USL
in
order
to

accurately
predict
membrane
permeability
from
bulk
concentration

measurements.
38

Figure
1:
Concentration
profile
across
a
membrane,
with
the
unstirred
layer
shown
in
gray.
Note
the

solute
depletion
in
this
region,
making
it
difficult
to
measure
the
driving
force
across
the
membrane.

The
potential
problems
introduced
by
the
USL
can
be
understood
by
treating
the

entire
system
as
a
sum
of
resistances.
If
the
USL
is
assumed
to
be
symmetric
on

either
side
of
the
membrane,
the
measured
permeability
Pmeas
can
be
calculated

based
on
the
permeability
of
the
solute
through
the
USL,
PUSL,
and
the
permeability

of
the
membrane,
P.

If
permeability
is
taken
as
the
reciprocal
of
resistance,
the

resistance
through
the
total
system
can
be
defined
as
the
sum
of
the
resistances

from
the
membrane
and
the
USL:
24

1
𝑃
!"#$
=
1
𝑃
+
1
𝑃
!"#

(3)

11

If
the
permeability
of
the
solute
in
the
USL
is
estimated
as
DUSL,
the
diffusivity
of
the

solute,
divided
by
δ,
the
length
of
the
USL,
this
relationship
gives:

𝑃
!"#$
=
1
!
!
+
!
!
!"#

(4)

As
discussed
by
Barry
and
Diamond,
determining
P
accurately
becomes
difficult
if

the
USL
is
large.
38

In
this
case,
the
transport
through
the
system
is
dominated
by

diffusion
through
the
USL,
rather
than
permeation
across
the
membrane.

1.4.2 Parallel
Artificial
Membrane
Assay
(PAMPA)

The
current
dominant
technique
for
high-‐throughput
screening
of
drug

permeability
is
the
parallel
artificial
membrane
permeability
assay
(PAMPA).

PAMPA
experiments
use
porous
filter
membranes
impregnated
with
organic

phospholipid
solutions
as
mimics
of
the
plasma
membrane.
39

PAMPA
has
numerous

advantages
relative
to
Caco-‐2
assays,
including
lower
cost,
simpler
implementation,

and
allowance
for
a
larger
pH
range
to
better
mimic
the
gastrointestinal
tract.
39,40

The
PAMPA
method
is
based
on
multi-‐well
microtiter
plates
with
filter
inserts
(e.g.

Corning
Transwell
®

membrane
filters).
The
filters
are
impregnated
with
a
lipid

solution
in
an
organic
solvent;
the
lipid
composition
can
be
varied
to
simulate

membranes
of
various
tissues.
The
wells
are
filled
with
an
aqueous
buffer
solution

(the
acceptor
solution)
and
the
filters
are
inserted.
The
solute
of
interest
(in
the

donor
solution)
is
added
to
the
top
chambers
of
the
filter
wells
and
transport
rates

12

are
determined
by
observing
the
change
in
concentration
of
the
solute
of
interest
in

the
acceptor
solution.
41

PAMPA
readout
is
by
simultaneous
UV
measurements
of
the

96-‐well
plate,
allowing
for
high
throughput
drug
screening.
41

While
the
PAMPA
assay
is
capable
of
producing
high-‐throughput
results,
42

the
filter

system
used
in
PAMPA
is
far
from
a
perfect
model
of
a
plasma
membrane
lipid

bilayer.
For
example,
it
is
not
clear
that
impregnation
of
filters
with
lipid
solution

leads
to
the
reliable
formation
of
bilayers
in
the
filter
pores.
The
formation
of
a

bilayer
in
every
pore
in
the
filter
membrane
is
not
guaranteed;
some
might
be

empty
and
others
might
be
clogged
with
lipid
or
solvent.

1.4.3 Steady
state
methods
for
measuring
permeability

The
most
common
steady
state
method
for
measuring
permeability
involves
the
use

of
planar
lipid
membranes.
In
this
method,
lipids
dissolved
in
a
hydrophobic
solvent

are
brushed
into
an
aperture
positioned
between
two
chambers.
After
the
solvent

has
dried,
the
chambers
are
filled
with
an
aqueous
solution,
and
the
aperture
is

painted
with
a
lipid
solution
for
a
second
time.
36

This
causes
a
bilayer
to

spontaneously
form
over
the
aperture.
Transport
of
various
solutes
can
then
be

measured
across
the
bilayer.

To
measure
the
permeation
of
species
with
a
charged
state,
microelectrodes
can
be

used
for
detection.
This
method
involves
positioning
microelectrodes
at
varying

positions
near
the
membrane
to
accurately
measure
the
pH
or
ion
concentration

profile
through
the
USL.
43,44

At
steady
state,
the
donor
side
of
the
membrane
is

13

treated
as
an
infinite
source
of
solute
at
fixed
concentration
and
the
receptor
is

treated
as
an
infinite
sink.
Upon
the
establishment
of
constant-‐flux
conditions,
the

system
will
take
on
a
steady-‐state
concentration
profile
that
can
be
characterized
by

the
microelectrode.
Steady-‐state
transport
equations
are
then
used
to
quantify
the

flux
across
and
permeability
of
the
membrane.
This
technique
has
been
used
to

measure
a
variety
of
properties,
such
as
the
effect
of
membrane
charge
on

permeability.
45

Additional
work
measured
the
permeability
of
sodium
ions,
46

proton

transport,
47-‐50

weak
acid
transport,
51-‐53

and
even
water
and
ammonia
transport.
54,55

1.4.4 Transient
methods
for
measuring
permeability

Although
there
are
numerous
techniques
for
transient
measurements
of
membrane

permeability,
we
will
only
discuss
a
few
notable
methods.
Stopped-‐flow
techniques

are
one
of
the
most
popular
ways
of
measuring
solute
transport
into
or
out
of

liposomes.
The
stopped-‐flow
technique
involves
adding
liquids
to
a
mixing
chamber

via
syringe.
The
influx
of
liquid
forces
the
old
contents
of
the
cell
into
a
stop
syringe.

When
the
cell
volume
has
been
completely
replaced
with
new
reagents,
the
stop

syringe
will
engage
a
trigger
switch
to
stop
the
flow
and
begin
acquisition.
The
time

from
initial
insertion
of
liquid
to
stopping
the
flow
is
considered
the
“dead
time”
of

the
setup,
and
is
dependent
on
the
design
of
the
apparatus.
This
dead
time

represents
the
time
the
two
inlet
liquids
have
been
in
contact
before
observation.
56

Water
transport
across
bilayers
can
be
measured
by
stopped-‐flow
using
an
osmotic

gradient
across
liposomes;
as
liposomes
shrink
or
swell
in
response
to
osmotically

driven
transport
of
water,
their
scattering
properties
will
change.
56-‐58

USLs
are
a

significant
concern
with
stopped-‐flow
techniques.
59

Some
studies
suggest
that
the

14

USL
in
stopped-‐flow
transport
experiments
can
be
up
to
5.5
μm
thick
(many
times

larger
than
the
liposomes
themselves).
56

Recent
advances
in
imaging
techniques
allow
for
spatially
resolved
transient

measurements
of
the
transport
process
in
through
the
membranes
of
giant

unilamellar
lipid
vesicles
(GUVs).
In
2010,
Li
et
al.
published
a
method
to
measure

the
transport
of
fluorescently
labeled
short-‐chain
poly(ethylene
glycol)
molecules

into
GUVs
with
spinning
disc
confocal
microscopy.
60

Vesicle
imaging
can
also
be

combined
with
rapid
microfluidic
buffer
exchange,
as
demonstrated
by
Li
et
al.
in

2011.
61

This
technique
allows
for
the
resolution
of
the
time
evolution
of
the
USL
by

visualizing
the
concentration
profile
as
a
function
of
time.
One
potential
setback
of

this
technique
is
that
it
requires
a
fluorescent
species
for
analysis.

1.5 Advantages
and
disadvantages
of
choosing
a
measurement

technique

Selecting
a
method
for
measuring
passive
transport
across
a
biomimetic
bilayer
is
a

matter
of
balancing
the
characteristics
of
the
various
techniques
described
above.

While
an
approach
that
can
directly
resolve
the
USL
is
of
obvious
benefit,
such

approaches
are
generally
limited
in
terms
of
the
types
of
permeant
species
they
can

analyze
(e.g.
fluorophores
or
charged
species).
Methods
that
are
amenable
to

general
chemical
species
as
permeants
are
inherently
more
prone
to
USL
artifacts

and
thus
are
less
accurate.
Despite
their
biological
relevance,
fast
permeating

species
are
a
difficult
class
of
solutes,
as
many
techniques
are
not
capable
of

resolving
relatively
fast
transport
processes.
High-‐throughput
approaches,
which

15

are
necessary
for
realistically
scaled
drug
discovery
efforts,
generally
suffer
from

both
USL
artifacts
and
a
lack
of
a
realistic
biomimetic
membrane.
The
ideal

transport
measurement
technique
would
be
capable
of
forming
consistent
bilayers

of
both
symmetric
and
asymmetric
compositions
and
analyzing
high-‐throughput

data
without
suffering
from
USL
artifacts.

2 Low
levels
of
lipid
oxidation
radically
increase
the

passive
permeability
of
lipid
bilayers

2.1 Motivation

The
oxidation
of
phospholipids
has
become
a
recent
topic
of
interest
within
the
field

of
membrane
biophysics.
The
process
of
lipid
oxidation
has
been
associated
with

tubule
formation
and
membrane
budding,
62

increases
in
membrane
surface

area,
63,64

decreases
in
membrane
fluidity,
65

and
the
promotion
of
phase

separation.
66,67

In
medical
research,
the
oxidation
process
has
been
linked
with

physiological
conditions
such
as
atherosclerosis
68

and
aging,
69

as
well
as
being
used

with
photodynamic
therapy
for
tumor
treatment.
67

2.1.1 Importance
of
oxidation
in
cells

Oxidation
in
human
tissue
has
been
associated
with
numerous
physiological

conditions.
Both
the
link
between
lipid
oxidation
and
apoptosis,
70-‐72

and
that

between
lipid
oxidation
and
cytotoxicity
73-‐76

have
been
studied
extensively.

Oxidation
of
both
phospholipids
and
lipoproteins
have
been
associated
with
the

pathogenesis
of
atherosclerosis
calcification
and
osteoporosis.
77

Atherosclerosis
is
a

disease
specifically
linked
with
increases
in
artery
wall
thicknesses
due
to
the

16

accumulation
of
calcium
and
other
types
of
fatty
material.
78

The
presence
of

oxidized
phospholipids
has
been
studied
in
relevance
to
not
only
prediction
of

atherosclerosis,
but
also
in
the
formation
of
atherosclerotic
legions.
79-‐83

2.1.2 Effects
of
lipid
oxidation
on
membrane
properties

One
phenomenon
that
has
not
been
examined
in
detail
is
the
effect
of
phospholipid

oxidation
on
the
permeability
of
the
membrane.
Passive
diffusion,
while
not
the
only

mechanism
by
which
molecules
can
cross
a
membrane,
represents
a
generic

pathway
by
which
drugs
and
environmental
toxins
can
enter
a
cell.
2,4

As
such,

understanding
and
measuring
how
oxidation
of
the
lipid
bilayer
affects
permeability

is
key
to
understanding
how
oxidation
alters
the
barrier
properties
of
the

membrane,
potentially
leading
to
cell
damage.

To
measure
the
effect
of
lipid
oxidation
on
bilayer
permeability,
we
directly

measured
the
permeation
of
a
test
species
across
biomimetic
membranes.
Our
test

system
is
based
on
fast
confocal
imaging
combined
with
a
microfluidic
approach
to

analyze
transport
across
the
membranes
of
giant
unilamellar
vesicles
(GUVs).
61

By

immobilizing
GUVs
in
a
microfluidic
channel,
then
performing
a
buffer
exchange,
it

is
possible
to
image
the
transport
of
a
fluorescent
solute
across
the
membrane.

GUVs
were
formed
in
which
the
molar
ratio
of
unsaturated
PLinPC
(1-‐palmitoyl-‐2-‐
linoleoyl-‐sn-‐glycero-‐3-‐phosphocholine)
to
its
corresponding
oxidized
product—
POxnoPC
(1-‐palmitoyl-‐2-‐(9'-‐oxo-‐nonanoyl)-‐sn-‐glycero-‐3-‐phosphocholine)—was

varied.
84,85

This
approach
was
designed
to
mimic
different
compositions
as
more

17

and
more
of
the
unsaturated
lipid
is
oxidized.
We
considered
a
simplified
version
of

the
total
oxidative
pathway
that
would
be
expected
as
a
reactive
oxygen
species

attacks
the
double
bonds
in
an
unsaturated
lipid
tail
group.
The
oxidized
product

used
in
this
work
would
be
the
result
of
three
oxidative
steps.
85

First,
a
reactive

oxygen
species
causes
the
formation
of
a
phospholipid
radical
species
by
abstracting

a
hydrogen
adjacent
to
the
double
bond.
Next,
these
carbon-‐centered
radicals

rapidly
react
with
free
oxygen
to
form
peroxyl
radicals.
The
final
step
is
the

termination,
where
the
cleaved
tail
group
is
no
longer
reactive.
POxnoPC
is
a
major

termination
product
of
PLinPC
oxidation.

In
synthetic
bilayer
systems,
the
presence
of
reactive
oxygen
species
has
been

linked
to
both
the
formation
of
singlet
oxygen
from
photo-‐induced
oxidation
86

as

well
as
a
process
known
as
autoxidation.
Autoxidation
occurs
when
unsaturated

lipids,
stored
in
water
at
physiological
temperature,
show
the
effects
of
oxidation
in

as
little
as
72
hours.
67

Biological
systems
have
a
more
diverse
set
of
pathways
to

introduce
reactive
oxygen
species:
the
presence
of
oxidants
has
been
linked
with

cell
respiration
and
metabolism;
87

signaling
processes;
88

the
inhalation
of
air

pollutants;
85

and
exposure
to
radiation.
85

High
levels
of
lipid
oxidation
lead
to
membrane
degradation.
The
amount
of

oxidation
that
a
lipid
bilayer
vesicle
can
withstand
before
collapse
depends
on
the

composition
of
the
membrane.
For
example,
membranes
consisting
of
saturated

DPPC
(1,2-‐dipalmitoyl-‐sn-‐glycero-‐3-‐phosphocholine)
and
oxidized
linoleoyl-‐

18

palmitoyl-‐lecithin
could
not
form
when
there
was
more
oxidized
species
than

saturated
species.
89

20%
POxnoPC
in
POPC
(1-‐palmitoyl-‐2-‐oleoyl-‐sn-‐glycero-‐3-‐
phosphocholine)
vesicles
compromises
the
membrane
via
sustained
pore

formation.
90

In
order
to
measure
passive
diffusion
permeability
rather
than
pore

transport,
total
oxidized
species
concentration
was
kept
below
this
20%
limit
in
our

experiments.

This
study
evaluated
the
effect
of
oxidation
of
unsaturated
lipids
on
passive

membrane
permeability
of
the
membrane
to
a
fluorescent
short
chain
poly(ethylene

glycol)
molecule
(PEG12-‐NBD).
The
permeability
of
this
molecule
was
characterized

Li
et
al.
in
2010
with
spinning
disk
confocal
microscopy
(SDCM).
60

The
PEG-‐NBD

species
was
chosen
for
this
experiment
because
it
is
hydrophilic,
but
not
a
charged

species.
As
charged
species
are
more
sensitive
to
changes
in
membrane

permeability
and
interactions
with
charged
lipid
headgroups,
91

choosing
an

uncharged
species
simplifies
the
transport
model.
Additionally,
the
fluorescent
label

makes
imaging
the
transport
process
simple
with
SDCM.

2.2 Materials
and
Methods

2.2.1 Materials

1-‐palmitoyl-‐2-‐linoleoyl-‐sn-‐glycero-‐3-‐phosphocholine

(PLinPC),
1,2-‐dimyristoyl-‐sn-‐
glycero-‐3-‐phosphocholine
(DMPC),
1-‐palmitoyl-‐2-‐(9'-‐oxo-‐nonanoyl)-‐sn-‐glycero-‐3-‐
phosphocholine
(POxnoPC),
cholesterol,
1,2-‐dihexadecanoyl-‐sn-‐glycero-‐3-‐
phosphoethanolamine-‐N-‐(cap
biotinyl)
(biotin-‐DPPE),
and
1,2-‐dipalmitoyl-‐sn-‐
glycero-‐3-‐phosphoethanolamine-‐N-‐(lissamine
rhodamine
B
sulfonyl)
(rhodamine-‐

19

DPPE)
were
purchased
from
Avanti
Polar
Lipids.
Avidin
and
succinimidyl
6-‐(N-‐(7-‐
nitrobenz-‐2-‐oxa-‐1,3-‐diazol-‐4-‐yl)amino)Hexanoate
(NBD
NHS-‐ester)
were
obtained

from
Invitrogen.
Amino-‐dPEG12-‐alcohol
was
purchased
from
Quanta
Biodesign.

Poly(dimethylsiloxane)
(PDMS)
was
purchased
from
Dow
Chemical,
and
the
indium-‐
tin-‐oxide
(ITO)
coated
glass
was
obtained
from
Delta
Technologies.
α-‐tocopherol

and
all
other
chemicals
were
purchased
from
Sigma-‐Aldrich.

2.2.2 GUV
composition
and
formation

GUVs
were
prepared
with
the
electroformation
technique
introduced
by
Angelova
et

al.

92,93

Vesicle
composition
prior
to
oxidation
was
chosen
to
be
a
42.5:42.5:15
molar

ratio
of
DMPC:PLinPC:Chol.
Rhodamine-‐DPPE
was
added
at
0.01
mol%
to
visualize

the
membrane.
PLinPC
was
stored
with
0.5
mol%
α-‐tocopherol
to
act
as
an
oxygen

scavenger
in
the
membrane
and
prevent
additional
oxidation
during

electroformation
and
imaging.
94

6
mol%
biotin-‐DPPE
was
used
to
attach
the
GUVs
to

the
avidin-‐treated
coverslip.
The
lipids
were
dissolved
in
chloroform,
then
spread

into
a
thin
film
on
the
surface
of
the
ITO
coated
glass.
The
film
was
dried
in
a

vacuum
for
at
least
two
hours
before
rehydration
in
pH
7.0
buffer
with
200
mM

sucrose
and
4
mM
HEPES.
Electroformation
was
performed
at
room
temperature

with
a
1
V
signal
oscillating
at
10
Hz
for
two
hours.

GUV
composition
was
varied
to
simulate
the
oxidation
process.
The
total

concentration
of
oxidized
species
was
varied
up
to
18
total
mol%,
to
remain
below

the
previously
characterized
poration
limit.
90

To
mimic
the
oxidation
of
unsaturated

20

lipids,
GUVs
were
fabricated
with
various
concentrations
of
POxnoPC
replacing

PLinPC
(shown
in
Table
1).

Table
1:
Vesicle
compositions
studied.

Mol%
DMPC
Mol%
PLinPC
Mol%
POxnoPC
Mol%
Chol

42.5
42.5
0
15

42.5
40
2.5
15

42.5
37.5
5
15

42.5
32.5
10
15

42.5
30
12.5
15

42.5
27.5
15
15

42.5
24.5
18
15

After
electroformation,
GUVs
were
immediately
transferred
to
the
microfluidic

channel.
The
GUVs
not
captured
in
the
channel
were
gently
flushed
from
the
system

using
a
syringe
pump
and
glucose
buffer.

2.2.3 Preparation
of
test
molecule

To
prepare
the
PEG12-‐NBD,
the
amine-‐terminated
poly(ethylene
glycol)
alcohol
was

reacted
with
the
NBD
NHS-‐ester.
60

A
reaction
schematic
is
shown
in
Figure
2.
A
1:1

molar
ratio
of
poly(ethylene
glycol)
to
NBD
NHS-‐ester
was
reacted
in
chloroform.

The
reaction
was
run
at
45°C
for
two
hours,
then
at
room
temperature
overnight.

The
reaction
products
were
separated
using
high
performance
liquid

chromatography.

21

Figure
2:
Reaction
scheme
for
the
preparation
of
the
PEG12-‐NBD
test
molecule.
The
amine-‐group
of

the
poly(ethylene
oxide)
alcohol
reacted
with
the
NHS-‐ester
of
the
fluorescent
NBD
to
form
the

fluorescent
PEG
species,
PEG12-‐NBD.

The
product,
PEG12-‐NBD,
was
confirmed
using
nuclear
magnetic
resonance
(NMR)

spectroscopy.
Following
the
reaction,
the
peaks
corresponding
to
the
amide
bond

were
apparent
at
6.5
and
6.9
ppm.
The
NMR
spectrum
of
the
PEG12-‐NBD
molecule

is
shown
in
Figure
3.

Figure
3:
NMR
spectra
confirming
the
product
PEG12-‐NBD.
Peaks
are
shown
with
indicators
for

significant
structures
in
the
molecule.
The
polymer
appears
at
3.25-‐3.75
PPM,
with
the
amide

occurring
at
6.5
and
6.9
PPM.
The
connection
between
the
PEG
chain
and
the
NBD
structure
is
found

at
1.5-‐2
PPM,
and
the
single
bonds
of
the
NBD
benzene
ring
occur
at
6.25
and
8.5
PPM.

22

2.2.4 Microfluidic
channel
preparation
and
design

Coverslips
were
sonicated
in
MilliQ
water
(Millipore)
at
80°C
for
30
minutes,
then

submerged
in
sulfuric
acid
with
NoChromix
(Sigma-‐Aldrich)
overnight.
The
acid
was

rinsed
away
before
sonicating
the
coverslips
for
30
minutes
in
water.
A
microfluidic

channel
was
fabricated
from
PDMS
using
standard
polymer
molding
techniques.
95

For
irreversible
bonding,
oxidation
of
the
PDMS
and
a
#1
coverslip
was
performed

with
a
corona
treatment
(BD-‐20AC,
Electro-‐Technic
Products).
96

As
shown
in
Figure

4,
two
types
of
device
were
used
for
vesicle
capture.
The
first
is
a
three
inlet,
one

outlet
channel.
97

The
three
inlets
converge
in
a
widened
channel
area
(3
mm
width

by
4
mm
length)
containing
pairs
of
50
μm
diameter
posts.
98

By
varying
the
distance

between
the
posts,
vesicles
of
different
sizes
could
be
trapped
in
the
post
array.
This

device
type
was
suitable
for
capture
and
imaging
of
vesicles
larger
than
25
μm
in

diameter.
To
prevent
interaction
between
the
PDMS
and
the
PEG-‐NBD,
the
channel

was
treated
with
a
solution
of
1
mg/mL
BSA
by
flushing
for
thirty
minutes.
The

channel
was
then
flushed
thoroughly
with
200
mM
glucose
buffer
with
4
mM
HEPES

at
pH
7.0
prior
to
adding
vesicles.
The
second
channel
was
a
Y-‐junction,
with
two

inlets
and
one
outlet.
The
channel
had
a
depth
of
100
μm,
a
width
of
1
mm,
and
a

length
of
1
cm.
Vesicles
containing
6
mol%
biotin-‐DPPE
were
captured
using
a

biotin-‐avidin
interaction
at
the
glass
surface.
The
coverslips
were
treated
with
a
1

mg/mL
solution
of
avidin
in
water
for
30
minutes,
then
flushed
with
200
mM

glucose
and
4
mM
HEPES
buffer
at
pH
7.0
prior
to
addition
of
GUVs.
Avidin
was

nonspecifically
adsorbed
to
the
glass
surface.

23

Figure
4:
Schematic
of
the
two
microfluidic
devices.
a)
Device
with
post-‐based
traps.
The
100
μm

height
channel
has
three
inlets:
one
for
buffer,
one
for
PEG12-‐NBD
solution,
and
one
for
loading
the

channel
with
vesicles.
The
main
chamber
has
an
array
of
posts.
The
solution
flow
pushes
vesicles
into

the
posts,
capturing
them
mechanically
while
buffer
flows
through
the
device.
The
inset
panel
shows

a
GUV
captured
in
between
a
pair
of
posts,
with
a
scale
bar
of
10
μm.
b)
A
simple
Y-‐channel
design,

with
two
inlets
and
a
single
outlet.
The
first
inlet
is
for
buffer,
the
second
is
for
buffer
with
PEG12-‐
NBD.
The
glass
surface
was
functionalized
with
a
1
mg/mL
avidin
solution
prior
to
the
experiment.

As
shown
in
the
inset,
vesicles
are
captured
in
this
device
using
a
biotin-‐avidin
interaction.

2.2.5 Transport
experiment
protocol

After
flushing
the
channel
of
all
unbound
vesicles,
a
single
unilamellar
vesicle
of

diameter
greater
than
10
μm
was
selected
for
observation.
A
syringe
pump
was

used
to
add
a
solution
of
5
μM
PEG12-‐NBD
in
200
mM
glucose
and
4
mM
HEPES

a)

b)

24

buffer
at
pH
7.0.
A
flow
rate
of
1.0
mL/h
was
chosen
as
the
fastest
flow
that

consistently
did
not
damage
or
detach
the
GUVs.
Transport
was
observed
using

spinning
disk
confocal
microscopy,
performed
with
a
Yokogawa
CSUX
confocal
head

on
a
Nikon
TI-‐E
inverted
microscope.
50
mW
solid-‐state
lasers
at
491
or
561
nm

were
used
as
the
illumination
sources.
Transport
of
PEG12-‐NBD
was
imaged
with

excitation
at
491
nm
and
emission
centered
at
525
nm.
Rhodamine-‐DPPE
was

excited
at
561
nm
and
emission
at
centered
595
nm.
Images
were
collected
at
a

regular
interval
during
buffer
exchange
as
the
fluorescent
species
crossed
the
GUV

membrane.

2.2.6 Examination
of
pore
formation

To
examine
the
formation
of
pores
in
the
membrane,
GUVs
were
added
to
a
solution

of
1
mg/mL
fluorescein-‐dextran
in
glucose
buffer.
To
estimate
the
size
of
the
pores,

GUVs
were
imaged
with
two
sizes
of
fluorescein-‐dextran:
40
kDa
and
2000
kDa.

2.3 Data
Analysis

2.3.1 Background
subtraction
and
flat-‐fielding

Image
analysis
was
performed
using
Matlab
(The
MathWorks,
Inc.).
First,
reference

images
with
no
fluorophore
in
the
field
were
used
to
subtract
the
background
from

each
image
in
the
sequence.
The
average
intensity
value
for
each
pixel
was

calculated
for
the
series
of
background
reference
images.
This
matrix
of
background

intensities
was
then
subtracted
from
each
of
the
series
of
transport
images.
Next,
to

correct
for
illumination
heterogeneity
across
the
field
of
observation,
a
flat-‐fielding

technique
was
applied
to
each
image.
A
reference
image
taken
of
5
μM
PEG12-‐NBD,

25

with
no
GUVs
in
the
field,
was
used
to
determine
a
multiplicative
factor
for
each

pixel
to
make
it
equal
to
the
mean
intensity
value
of
the
reference
image.
This
was

repeated
for
five
reference
images,
and
the
resulting
multiplicative
values
were

averaged
at
each
pixel
position.
This
matrix
of
factors
was
applied
to
each
image
in

the
experimental
sequence
to
flat-‐field
the
data
set.

2.3.2 Removal
of
pinhole
crosstalk
contributions

One
concern
with
confocal
microscopy
is
its
imperfect
exclusion
of
light
from

outside
the
focal
plane.
99

As
shown
in
Figure
5,
the
intensity
inside
the
vesicle
prior

to
permeation
is
not
flat.
This
curved
profile
indicates
the
presence
of
pinhole

crosstalk,
causing
an
increase
in
measured
intensity
inside
the
vesicle.
Analysis
of

the
out
of
plane
light
contribution
for
a
vesicle
with
varying
concentrations
showed

that
the
relationship
can
be
considered
to
be
linear,
and
corrected
accordingly.

26

Figure
5:
Intensity
profile
across
a
vesicle
immersed
in
5
μM
PEG12-‐NBD
prior
to
permeation.
This

image
has
been
corrected
for
uneven
illumination
by
flat-‐fielding.
As
the
fluorescein-‐dextran
cannot

permeate
the
membrane,
the
intensity
inside
of
the
vesicle
should
be
0,
and
the
profile
across
the

vesicle
should
be
flat.
The
presence
of
a
non-‐flat
profile
indicates
that
there
is
pinhole
crosstalk

present
in
the
experiments.
This
contribution
to
the
intensity
inside
of
the
vesicle
was
accounted
for

during
data
analysis.

Pinhole
crosstalk
is
a
characteristic
spinning
disk
confocal
microscopy
(SDCM)
that

limits
the
ability
of
the
spinning
disk
apparatus
from
regions
outside
of
the
focal

plane.
Previous
work
has
been
performed
to
establish
the
contribution
of
out-‐of-‐
plane
fluorescent
sources
to
the
apparent
in-‐plane
fluorescence.
99

The
results

showed
that
the
pinhole
crosstalk
contribution
from
a
non-‐focal
plane
scales
with

the
inverse
of
the
distance
between
the
non-‐focal
and
the
focal
plane.

27

During
a
permeability
experiment
with
a
fluorescent
molecule,
such
as
PEG12-‐NBD,

the
fluorescence
inside
the
vesicle
changes
with
concentration.
If
the
background

fluorescence
is
assumed
to
be
constant,
then
the
out
of
plane
contribution
for
area

outside
of
the
vesicle
can
also
be
assumed
to
be
constant.
However,
the
significant

out-‐of-‐plane
contribution
is
from
the
fluorophores
inside
the
vesicle.
This
means

that
we
need
to
develop
a
relationship
between
out-‐of-‐plane
light
contributions
and

fluorophore
concentration
inside
the
vesicle.

To
evaluate
the
out-‐of-‐plane
contribution
for
a
given
concentration
of
fluorophore

inside
the
vesicle,
GUVs
were
created
with
encapsulated
40
kDa
fluorescein-‐dextran

in
200
mM
sucrose
buffer
at
pH
7.0.
40
kDa
fluorescein-‐dextran
was
chosen
as
it
will

not
cross
the
membrane.
Encapsulation
concentrations
were
chosen
between
0.1

μM
and
0.5
μM
to
show
intensities
in
the
same
range
as
the
permeability

experiments.
GUVs
were
created
with
a
1:1:1
DPPC:PLinPC:Chol
molar
ratio,
with

0.01
mol%
rhodamine-‐DPPE.

After
electroformation,
GUVs
were
transferred
a
Sykes-‐Moore
chamber
containing

200
mM
glucose
buffer
at
pH
7.0.
Buffer
exchange
was
performed
until
the

background
surrounding
the
vesicles
showed
no
fluorescence
intensity
due
to

excess
fluorescein
dextran.
Between
75
and
100
GUVs
up
to
100
μm
in
diameter

were
chosen,
such
that
for
a
given
encapsulation
concentration,
the
relationship

between
vesicle
intensity
and
diameter
could
be
determined.

28

Prior
to
analysis,
each
image
was
flat-‐fielded.
As
the
laser
alignment
in
a
confocal

microscope
causes
the
center
of
the
image
to
be
brighter
than
the
edges
of
the

image,
flat-‐fielding
will
remove
the
dependence
of
GUV
location
on
its
intensity.
To

obtain
reference
images,
a
Sykes-‐Moore
chamber
was
filled
with
200
mM
glucose

buffer
containing
the
concentration
of
40
kDa
fluorescein-‐dextran.
Five
background

images
were
collected.
For
this
series
of
background
images,
the
value
of
each
pixel

was
averaged
to
decrease
the
effect
of
noise.
A
factor
was
calculated
that,
when

multiplied,
made
each
pixel
in
the
averaged
background
image
equal
to
the
mean

value
across
the
entire
image.
For
each
GUV
containing
the
reference
fluorescein-‐
dextran
concentration,
the
image
was
multiplied
by
this
matrix
of
factors
to

generate
a
flat-‐fielded
image.

After
performing
the
flat-‐fielding
technique,
the
intensity
and
diameter
of
each
GUV

was
measured
using
the
image
processing
toolbox
in
Matlab
(MathWorks).
For
each

concentration
of
encapsulated
fluorescein-‐dextran,
the
relationship
between
vesicle

intensity
and
diameter
appears
to
be
linear,
as
shown
in
Figure
6.

29

Figure
6:
Linear
relationship
between
vesicle
interior
intensity
and
diameter
for
a)
0.2
μM
and
b)
0.5

μM
encapsulated
40
kDa
fluorescein-‐dextran.

a)

b)

30

Assuming
the
y-‐intercept
of
the
line
is
the
“true”
intensity
of
the
fluorescein-‐dextran,

then
it
is
simple
to
show
that
for
a
given
diameter,
the
pinhole
crosstalk

contribution
is
given
by
the
difference
between
the
actual
intensity
and
the
“true”

intensity,
as
shown
in
Figure
7.

Figure
7:
Vesicle
interior
intensity
vs.
“true”
intensity
of
0.2
μM
40
kDa
fluorescein-‐dextran.

Thus,
the
amount
of
out
of
plane
light
has
a
dependence
on
both
vesicle
diameter

and
concentration.
Experiments
were
repeated
for
concentrations
of
0.1,
0.2,
0.3,

0.4,
and
0.5
μM
fluorescein-‐dextran,
and
a
similar
analysis
was
performed.
The

resulting
linear
relationships
are
shown
in
Figure
8.

31

Figure
8:
Linear
relationships
for
five
concentrations
of
encapsulated
40
kDa
fluorescein-‐dextran

Next,
the
change
in
out
of
plane
with
respect
to
concentration
was
evaluated
for
a

given
diameter
of
25
μm.
As
shown
in
Figure
9,
the
relationship
between
the
out
of

plane
contribution
and
encapsulated
concentration
appears
to
be
linear.

32

Figure
9:
Linear
fit
for
pinhole
crosstalk
contribution
vs.
concentration
for
25
μm
vesicle
diameter

The
analysis
was
repeated
for
diameters
between
10
μm
and
50
μm,
the
typical

range
for
vesicles
used
in
permeability
experiments.
For
each
diameter,
the

relationship
between
out
of
plane
light
and
concentration
is
linear.
The
results
for

this
diameter
range
are
shown
in
Figure
10.

33

Figure
10:
Linear
relationships
for
pinhole
crosstalk
contribution
vs.
concentration
for
diameters

between
10
and
50
μm.

From
this
analysis,
it
is
clear
that
the
out
of
plane
light
correction
is
dependent
on

both
the
diameter
of
the
vesicle
and
the
concentration
encapsulated.
However,
for
a

constant
diameter,
the
relationship
with
respect
to
concentration
is
consistently

linear.
When
analyzing
the
permeability
data,
the
diameter
of
the
GUV
is
constant.

This
means
that
to
correct
for
out
of
plane
light
as
the
concentration
inside
the

vesicle
changes,
the
correction
should
be
linear
with
respect
to
concentration.
We

corrected
for
pinhole
crosstalk
by
subtracting
the
crosstalk
contribution
predicted

by
the
relationships
shown
in
Figure
10.

2.3.3 Finite
difference
model
and
membrane
permeability

To
calculate
the
permeability
from
the
intensity
vs.
time
images,
a
finite
difference

model
was
based
on
Fick’s
Second
Law
in
spherical
coordinates.
Assuming
no

34

reaction,
symmetry
in
both
angular
dimensions,
a
spherical
vesicle,
and
that

diffusion
is
a
significantly
higher
contributor
than
convection
to
the
permeation

process,
the
governing
equation
simplifies
to
that
shown
in
Equation
5.
The
variable

cA
represents
the
concentration
of
species
A,
t
is
time,
r
is
the
radial
coordinate
(0
at

the
vesicle
center),
and
DA
is
the
diffusivity
of
species
A
in
the
buffer
solution.

(5)

This
equation
was
solved
in
space
and
time
using
a
finite
difference
approach.
To

simplify
calculations,
the
method
described
by
Somersalo
et
al.
was
used
to

decouple
the
time
and
space
dimensions.
100

This
technique
allowed
for
a
simple

time
determination
of
concentration
at
each
discretized
space
step.

The
advantage
of
their
technique
was
in
the
decoupling
of
the
time
and
space

dimensions,
and
solving
the
system
as
a
matrix
equation.
This
allowed
for
fast

convergence
of
the
model.
An
example
result
from
the
finite
difference
model
is

shown
in
Figure
11.
The
model
takes
the
background
intensity
as
the
concentration

boundary
condition
at
the
vesicle
exterior.
At
each
time
point,
the
model
then

calculates
the
concentration
along
a
series
of
space
steps.
For
the
space
step

representing
the
membrane,
a
flux
boundary
condition
incorporates
the
membrane

permeability
into
the
model.
The
concentration
is
then
calculated
for
the
space
steps

contained
inside
the
vesicle
membrane;
at
the
vesicle
center,
a
no-‐flux
boundary

condition
is
applied.
This
process
is
repeated
for
each
time
step
until
equilibrium
is

∂c
A
∂t
= D
A
1
r
2
∂
∂r
r
2
∂c
A
∂r
"
#
$
%
&
'
(
)
*
+
,
-

35

reached
or
the
data
set
is
complete.
For
Figure
11,
the
center
of
the
vesicle
is
located

at
0
μm
along
the
x-‐axis.
The
radius
of
the
vesicle
is
15
μm,
so
the
membrane
is

located
at
this
point
on
the
x-‐axis.
The
y-‐axis
represents
the
intensity
at
each
special

location,
and
the
color
gradient
shows
the
time
series.

Figure
11:
Results
from
a
finite
difference
simulation
of
0%
POxnoPC.
The
vesicle
center
is
located
at

0
along
the
x-‐axis,
with
the
membrane
location
at
15
μm.
The
y-‐axis
indicates
the
percentage
of
the

background
concentration,
and
the
color
gradient
indicates
different
time
points
during
the

permeation
process.
In
this
manner,
the
model
can
show
both
spatial
and
time
variation
of

concentration.
The
insets
show
images
from
the
transport
experiment
corresponding
to
certain
time

points.
The
scale
bar
is
10
μm.

Boundary
conditions
for
the
model
were
set
at
three
points:
an
exterior
point,
the

membrane
itself,
and
the
vesicle
center.
The
exterior
boundary
was
located
at
a

point
from
the
membrane
equal
to
the
vesicle
radius.
At
this
boundary,

concentration
was
fixed
at
the
experimentally
observed
fluorophore
concentration.

Note
that
this
concentration
changes
with
time
as
the
buffer
is
exchanged.
At
the

membrane,
flux
J
was
determined
by
membrane
permeability
such
that:

36

(6)

Permeability
P
was
the
only
free
parameter
in
the
model.
At
the
center
of
the
GUV,
a

no-‐flux
boundary
condition
was
applied.

To
fit
the
experimental
data
to
model
results,
experimentally
observed
average

fluorescence
in
a
circular
region
around
the
center
of
the
vesicle
was
compared
to

modeled
concentration
in
the
analogous
spatial
region.
Permeability
P
was

determined
by
performing
a
χ
2

minimization
between
the
model
output
and
the

measured
intensity
values.
The
resulting
concentration
vs.
time
curve
for
the
data

set
and
the
best
fit
model
results
is
shown
in
Figure
14
(Section
2.4.2).

To
determine
the
best
fit,
the
χ
2

value
for
model
vs.
experimental
intensity
curves

was
calculated
at
a
range
of
model
permeabilities.
χ
2

is
minimized
using
the

fminsearch
routine
in
Matlab.
The
χ
2

landscape
with
respect
to
permeability
was

used
to
determine
the
uncertainties
of
the
model
result,
with
a
p-‐value
of
0.05.
An

example
chi-‐squared
landscape
is
shown
in
Figure
12.
These
results
correspond
to

the
data
set
shown
in
Figure
11.

J =PΔc

37

Figure
12:
χ
2

landscape
for
a
data
set
of
0%
POxnoPC.
The
minimum
χ
2

value
corresponds
to
the

permeability
from
the
best
fit
of
the
model.
The
inset
panel
shows
a
close
up
view
of
the
minimum
χ
2

value.
A
chosen
critical
value
of
p
=
0.05
corresponds
to
a
χ
2

greater
than
this
minimum;
at
this
χ
2
,
the

uncertainty
of
the
permeability
can
be
calculated.
The
critical
value
indicates
the
+/-‐
of
the

permeability
for
that
degree
of
certainty
(shown
in
purple
on
the
inset).

2.4
Results
and
Discussion

2.4.1 Comparison
of
capture
methods:
mechanical
vs.
chemical
trapping

Prior
to
capturing
experimental
data,
we
wanted
to
compare
the
permeability

values
measured
using
the
two
trapping
methods
discussed
in
Section
2.2.4.
Vesicles

with
a
molar
ratio
of
42.5:42.5:15
DMPC:PLinPC:Chol
were
examined
using
the

mechanical
trapping
system
and
the
biotin-‐avidin
capture.
Results
for
permeability

of
each
capture
method
are
shown
in
Figure
13.

χ
2

Value

38

Figure
13:
Comparison
of
permeability
for
vesicles
captured
using
a
mechanical
trap
vs.
a
biotin-‐
avidin
chemical
trap.
The
permeability
for
vesicles
captured
using
the
biotin-‐avidin
interaction
is

slightly
higher
than
that
of
vesicles
captured
using
the
mechanical
trap.

While
there
are
advantages
to
using
the
mechanical
trap,
such
as
larger
vesicle
sizes

and
smaller
uncertainty
values,
this
type
of
trapping
could
not
capture
the
smaller

vesicles
formed
at
higher
levels
of
POxnoPC.
As
the
biotin-‐avidin
trapping
technique

was
capable
of
capturing
all
vesicle
sizes,
it
was
the
preferred
method
for

permeability
experiments.

2.4.2 Permeability
of
vesicles
with
low
levels
of
oxidation

For
concentrations
of
0-‐10%
POxnoPC,
GUVs
were
unilamellar
with
diameters

between
10-‐50
μm.
Representative
permeation
curves
for
0
mol%
and
2.5
mol%

POxnoPC
are
shown
in
Figure
14.
The
data
shown
include
the
fluorescence

intensities
at
the
exterior
boundary
point
and
the
vesicle
center
as
functions
of
time.

The
finite
difference
model
results
are
shown
in
red.
After
addition
of
POxnoPC,
the

39

vesicles
were
too
small
to
be
efficiently
captured
by
the
mechanical
trapping
device.

For
these
compositions,
the
biotin-‐avidin
interaction
was
used
to
capture
vesicles

for
analysis.
As
shown
in
Figure
14a,
the
permeability
of
the
0%
POxnoPC
vesicle
is

2.80
x10
-‐6

cm/s.
This
is
one
order
of
magnitude
less
than
the
permeability
of
the

2.5%
POxnoPC
vesicle
in
Figure
14b
(2.85x10
-‐5

cm/s).

Figure
15
shows
the
trend
between
measured
permeabilities
vs.
percent
oxidation

over
several
trials
for
each
composition.
Two
permeability
regimes
were
clearly

observed.
The
first
regime,
observed
for
membranes
with
no
POxnoPC,
corresponds

to
slow
passive
transport
across
the
membranes,
with
permeability
on
the
order
of

1.5x10
-‐6

cm/s.
The
second
regime,
seen
for
concentrations
of
POxnoPC,
between
2.5

and
10%
of
the
total
composition,
corresponds
to
fast
transport
with
permeability

on
the
order
of
1.5x10
-‐5

cm/s.
Even
small
quantities
of
oxidation
cause
a
large

increase
in
the
permeability
of
the
membrane.
The
addition
of
the
POxnoPC
also

caused
an
overall
decrease
in
vesicle
size
(details
in
Section
2.4.4).

40

Figure
14:
Intensity
data
and
model
fit
for
a
data
set
with
a)
0%
POxnoPC
and
b)
2.5%
POxnoPC.
The

green
data
points
indicate
the
normalized
outside
intensity
data
(measured
at
one
vesicle
radius

from
the
GUV
membrane),
the
blue
indicate
normalized
vesicle
interior
intensity
data.
The
red
line

indicates
the
results
of
the
finite
difference
model
for
the
best-‐fit
permeability.
Image
scale
bars
are

10
μm.
The
permeability
of
the
0%
POxnoPC
data
in
a)
shows
a
permeability
of
2.80x10
-‐6

cm/s,

which
is
almost
two
orders
of
magnitude
lower
than
the
permeability
of
the
2.5%
POxnoPC

membrane
(P
=
2.85x10
-‐5

cm/s)
shown
in
b).

a)

b)

41

Figure
15:
Calculated
permeability
vs.
percent
oxidized
species
for
four
vesicle
compositions.
Note

that
there
appear
to
be
two
regimes
of
permeability:
a
slow
regime
for
no
POxnoPC
present,
and
a

fast
regime
for
low
levels
(2.5-‐10%)
POxnoPC
present.
The
difference
between
these
two
regimes
is

almost
two
orders
of
magnitude.

2.4.3 Pore
formation
in
vesicles
with
high
levels
of
oxidation

For
compositions
of
12.5-‐18%
POxnoPC,
the
standard
electroformation
protocol

produced
a
low
yield
of
high-‐quality
unilamellar
vesicles.
Most
vesicles
displayed

characteristics
such
as
tubule
formation,
irregular
shape,
or
decreased
overall
size.

Representative
images
of
vesicle
formation
for
each
composition
are
shown
in

Figure
16.

42

Figure
16:
SDCM
images
showing
vesicle
formation
for
a)
12.5%
POxnoPC,
b)
15%
POxnoPC,
and
c)

18%
POxnoPC.
For
each
composition,
in
addition
to
spherical
vesicles,
the
lipids
appear
to
form

tubules
and
nonspherical
structures.
Spherical
vesicles
are
typically
quite
small
(<10
μm
in

diameter).
Scale
bars
are
10
μm.

These
vesicles
all
displayed
extremely
rapid
transport
properties
inconsistent
with

passive
diffusive
transport
across
the
bilayer.
For
all
compositions
in
this
range,

both
40
kDa
and
2000
kDa
fluorescein-‐dextran
molecules
were
able
to
cross
the

membrane.
At
POxnoPC
compositions
less
than
12%,
the
membrane
was

impermeable
to
either
dextran
species.
These
results
indicate
that
macromolecule-‐
admitting
pores
are
formed
in
the
membrane.
Since
the
2000
kDa
fluorescein-‐
dextran
was
able
to
travel
through
these
pores,
we
can
estimate
the
pore
size
from

the
hydrodynamic
diameter
of
the
fluorescein-‐dextran.
Previous
studies
have

established
the
diameter
of
2000
kDa
fluorescein-‐dextran
as
545
Å,
101

so
we
can

estimate
that
pore
diameter
must
be
greater
than
this
value.
Figure
17
shows

images
of
GUVs
in
fluorescein-‐dextran
solution
at
various
compositions.
Dextran

permeability
is
apparent
at
the
higher
concentrations
of
oxidized
species.

a)

b)

c)

43

%POxnoPC

561

(40
kDa)

491

(40
kDa)

561

(2000
kDa)

491

(2000
kDa)

10%

12.5%

15%

18%

Figure
17:
SDCM
images
showing
pore
formation
for
vesicles
with
12.5-‐18%
POxnoPC.
The
10%

POxnoPC
images
are
shown
as
a
reference
for
vesicles
without
pore
formation,
as
neither
species
of

fluorescein-‐dextran
could
enter
the
vesicle.
For
higher
concentrations
of
oxidized
species,
both
the

40
kDa
and
the
2000
kDa
fluorescein-‐dextran
was
able
to
enter
the
vesicle.
As
both
species
are
too

large
to
enter
the
vesicle
through
passive
transport
pathways,
this
indicates
that
nanoscale
pores
are

present
in
the
membrane
for
these
concentrations
of
oxidized
lipid.
Scale
bar
is
20
μm,
and
all
images

are
the
same
resolution.

2.4.4 Relationship
between
permeability
and
vesicle
diameter

The
addition
of
the
POxnoPC
caused
an
overall
decrease
in
vesicle
size.
Figure
18

shows
the
average
diameter
measured
for
each
of
the
compositions
analyzed
in
the

main
text.

From
the
analyzed
data
points,
the
diameter
is
shown
to
decrease
with

increased
amounts
of
oxidation.
This
decrease
in
vesicle
size
is
not
unexpected:

previous
research
has
demonstrated
the
changes
in
membrane
curvature
and

44

budding
due
to
lipid
oxidation.
62

The
cleaved
tail
group
has
been
associated
with

decreased
lipid
area
and
corresponding
decreases
in
membrane
area.
64

Figure
18:
Correlation
between
vesicle
diameter
and
percent
oxidation.
The
average
vesicle

diameter
was
calculated
for
each
of
the
four
compositions
measured
for
permeability
(shown
in

Figure
15).
As
oxidation
increases,
overall
vesicle
size
decreases.
Despite
this
trend,
there
does
not

appear
to
be
a
correlation
between
measured
permeability
and
vesicle
diameter,
as
shown
in
Figure

19.

It
is
important
to
note,
however,
that
the
permeability
does
not
appear
to
be

correlated
with
the
diameter.
Figure
19
plots
permeability
for
all
vesicles
observed

as
a
function
of
vesicle
diameter.
While
the
trend
shown
in
Figure
18
clearly
results

in
clustering—0%
POxnoPC
vesicles
are
big
and
relatively
impermeable
in

comparison
to
other
compositions—there
is
no
discernable
dependence
of

permeability
of
vesicle
size
within
the
two
clusters.

45

Figure
19:
Measured
permeability
vs.
vesicle
diameter
for
four
levels
of
lipid
oxidation.
There
does

not
appear
to
be
a
correlation
between
measured
permeability
and
vesicle
diameter.
The

permeability
was
plotted
vs.
diameter
for
vesicles
containing
various
amounts
POxnoPC.
There
does

not
appear
to
be
a
trend
between
these
two
parameters.

2.4.5 Impact
of
biotin-‐DPPE
on
measured
permeability

Since
biotin-‐DPPE
was
used
to
immobilize
the
vesicles
to
the
surface
of
the

microfluidic
channel,
we
examined
the
effect
of
biotin-‐DPPE
concentration
on
the

measured
permeability.
To
test
if
the
presence
of
biotin-‐DPPE
had
an
effect
on

membrane
permeability,
we
varied
the
amount
of
biotin-‐DPPE
present
in
the

membrane
then
measured
the
permeability.
We
chose
two
levels
of
oxidation
(0

mol%
POxnoPC
and
10
mol%
POxnoPC)
to
examine.
For
each
of
these
oxidation

points,
we
decreased
the
biotin-‐DPPE
concentration
from
6
mol%
to
either
4
mol%

or
2
mol%.
Results
for
these
tests
are
shown
in
Figure
20.
There
is
no
apparent

effect
of
biotin-‐DPPE
on
the
membrane
permeability.

46

Figure
20:
Measured
permeability
vs.
mol
%
biotin-‐DPPE
for
two
levels
of
oxidation:
0
mol%
and
10

mol%
POxnoPC.
For
each
composition,
permeability
does
not
appear
to
depend
on
biotin-‐DPPE

concentration.

2.4.6 Discussion

The
replacement
of
an
unsaturated
lipid
in
a
bilayer
with
its
oxidization
product

increases
the
permeability
of
the
bilayer
to
a
test
species.
As
the
concentration
of

this
oxidized
product
increases,
it
is
possible
to
observe
three
distinct
permeability

regimes.
The
first
regime
consists
of
vesicles
containing
no
POxnoPC
where

transport
is
slow.
The
second
regime
is
for
low
to
mid
levels
of
oxidation
(2.5-‐10%

POxnoPC),
where
transport
is
about
one
order
of
magnitude
faster
than
the
slow

regime.
In
this
regime,
the
passive
transport
process
can
still
be
measured
and
is

consistent
with
permeation-‐diffusion
across
the
membrane.
23

The
third
regime

consists
of
vesicles
containing
high
amounts
of
oxidation
(12.5-‐18%
POxnoPC).

These
membranes
show
characteristics
of
pore
formation,
allowing
molecules
as

large
as
2000
kDa
fluorescein-‐dextran
to
cross
the
membrane.
This
is
lower
than
the

47

20%
limit
of
POxnoPC
in
a
POPC
membrane
at
which
defects
have
been
previously

observed
by
Volinsky
and
coworkers.
90

The
discrepancy
is
likely
explained
by

different
membrane
compositions:
the
same
study
confirms
a
significantly
lower

poration
limit
for
a
different
oxidation
product
of
POPC.
90

A
second
study

demonstrated
that
destabilizing
pore
formation
in
membranes
consisting
of
DPPC

and
oxidized
linoleoyl-‐palmitoyl-‐lecithin
did
not
occur
below
50%
oxidized

species.
89

Taken
together,
these
results
indicate
that
both
the
membrane

composition
and
the
identity
of
the
oxidized
species
determine
how
much
oxidation

causes
membrane
destabilization.

A
second
important
result
is
that
a
small
amount
of
oxidation
(2.5%
POxnoPC)
can

cause
a
drastic
increase
membrane
permeability.
This
increase,
corresponding
to

one
order
of
magnitude
in
permeability,
represents
a
significant
compromise
of
the

barrier
function
of
the
lipid
bilayer.
This
would
indicate
that
even
low
levels
of

oxidation
can
have
significant
impact
on
membrane
function.
However,
notably,

there
was
no
discernible
difference
between
the
measured
permeability
between

2.5%
and
10%
oxidation,
indicating
that
low
levels
of
oxidation
are
as
damaging
as

significantly
higher
levels
to
the
barrier
properties
of
the
membrane.
Until
pore

formation
began
to
occur,
additional
oxidation
of
the
vesicles
did
not
continue
to

increase
the
membrane
permeability.

This
result
is
surprising,
especially
given
early
work
on
the
oxidation/permeability

relationship.
In
1980,
Mandal
and
Chatterjee
examined
the
effect
of
ultraviolet

48

radiation
on
the
oxidation
process
and
permeability
of
liposomes.
102

Vesicle
leakage

of
chromate
anions
was
measured
during
the
irradiation
process,
demonstrating
a

linear
increase
in
leakage
with
the
amount
of
oxidation.
The
amount
of
oxidation
per

liposome
was
measured
via
optical
absorbance
of
conjugated
dienes.
While
this

technique
is
a
simple
method
for
measuring
a
single
oxidation
species,
it
has
been

suggested
that
this
species
accounts
for
less
than
5%
of
the
total
oxidation

products.
103

Both
the
importance
of
composition
on
oxidation
effects
and
the

presence
of
other
potential
oxidation
products
in
the
liposomes
could
account
for

the
differences
in
the
reported
trends
between
permeability
and
lipid
oxidation.

The
compositions
in
the
pore-‐formation
regime
are
also
notable.
While
these

membranes
showed
pore
formation,
the
membranes
themselves
were
stable
over

several
hours.
Previous
studies
have
shown
that
oxidation
can
cause
rapidly

opening
and
closing
pores
in
phospholipid
membranes.
64

While
it
is
not
clear
if
the

pores
formed
in
the
third
regime
are
transient
or
stable,
the
existence
of
pores
is

consistent
with
previous
studies.
104

One
particular
concern
in
obtaining
GUVs
with
precise
levels
of
oxidation
is
the

potential
for
lipid
oxidation
during
electroformation.
Although
electroformation
is

known
to
cause
both
oxidation
and
variations
in
concentration,
the
presence
of
α-‐
tocopherol
(as
used
in
this
study)
has
been
consistently
shown
to
prevent

oxidation.
105

The
difference
in
composition
between
vesicles
formed
during
the

electroformation
process
has
been
studied
previously.
106

By
examining
the

49

difference
in
fluorescence
intensity
of
the
membrane,
it
was
shown
that
the
overall

composition
varied
by
less
than
2
mol%
between
vesicles
in
one
electroformation

batch.

2.5 Conclusion

Our
results
contribute
additional
understanding
to
the
field
of
changes
in
bilayer

properties
during
phospholipid
oxidation.
Previous
work
has
examined
the
effect
of

lipid
oxidation
on
numerous
membrane
properties.
Shape
changes
such
as
the

formation
of
membrane
buds
and
tubules
have
been
examined
extensively.
62

These

shape
changes
are
attributed
to
an
initial
increase,
followed
by
a
decrease,
in

membrane
curvature
with
the
oxidation
process.
107

The
variation
in
curvature
is

also
associated
with
an
increase
in
membrane
surface
area.
63

Additional
work
has

shown
decreases
in
membrane
tension,
108

fluidity,
65

and
increases
in
lipid
flip-‐flop

rate.
90

Phase
separation
is
promoted
by
lipid
oxidation
in
both
liquid
ordered
–

liquid
disordered
systems,
66,67,86

as
well
as
membranes
containing
a
liquid-‐extended

phase.
109

Some
studies
have
shown
complete
membrane
disintegration
at
high

levels
of
lipid
oxidation.
108

Despite
significant
research
into
the
impact
of
oxidation

on
membrane
properties,
the
effect
of
membrane
damage
from
oxidation
on

permeability
has
not
been
characterized
in
detail.
Previous
studies
of
the
impact
of

lipid
oxidation
on
membrane
characteristics
have
shown
in
general
that
decreasing

bilayer
thickness
84

and
increasing
lipid
mobility
110
—both
demonstrated
results
of

oxidation—would
lead
to
increased
permeability.
While
we
observed
no
phase

separation
in
any
of
the
vesicles
examined,
it
is
also
possible
that
phase
separation

at
a
scale
below
the
resolution
of
optical
microscopy
could
lead
to
changes
in

50

membrane
permeability.
Here
we
show
conclusive
evidence
that
permeability
does

increase
with
oxidation,
leading
to
three
distinct
regimes
of
permeability
for
zero,

low,
and
high
levels
of
lipid
oxidation.
The
low
levels
of
oxidation
in
this
study
are

particularly
of
interest,
as
concentrations
as
low
as
1.3x10
-‐4

%
of
oxidized
to
total

phospholipid
species
have
been
associated
with
the
formation
of
atherosclerotic

lesions
in
humans.
111

Further
experimentation
of
low
levels
of
phospholipid

oxidation
is
necessary
to
fully
understand
the
role
of
phospholipid
oxidation
in
cell

degradation
and
death.

3 The
addition
of
cleaved
tail
fragments
during
lipid

oxidation
stabilizes
membrane
permeability
behavior

3.1 Motivation

Lipid
oxidation
is
associated
with
numerous
physiological
conditions
including

atherosclerosis
68

and
aging.
69

A
key
set
of
lipid
oxidation
reactions
includes
those

involving
an
unsaturated
lipid
tail
group
and
a
reactive
oxygen
species.
The

presence
of
reactive
oxygen
species
in
biological
systems
have
been
linked
to

exposure
to
radiation;
85

air
pollutant
inhalation;
85

signaling
processes;
88

and
cell

respiration
and
metabolism.
87

Significant
membrane
effects
such
as
increases
in

membrane
surface
area,
63,64

the
formation
of
tubules
and
membrane
budding,
62

the

promotion
of
phase
separation,
66,67

and
decreased
membrane
fluidity
65

have
also

been
noted.

However,
the
impact
of
lipid
oxidation
on
passive
membrane

permeability
has
not
been
extensively
studied.
As
passive
transport
is
a
generic

pathway
by
which
drugs
and
environmental
toxins
can
enter
a
cell,
2,4

the
impact
of

51

lipid
oxidation
on
passive
permeability
is
significant
to
understanding
how
this

process
affects
membrane
barrier
function.

We
previously
showed
that
by
mimicking
lipid
oxidation
with
the
replacement
of
an

unsaturated
lipid
with
its
corresponding
oxidation
product,
a
radical
increase
in

membrane
permeability
occurs
with
only
a
2.5%
change
in
oxidation.
112

The

previous
study
used
a
known
pair
of
unsaturated
lipid—PLinPC
(1-‐palmitoyl-‐2-‐
linoleoyl-‐sn-‐glycero-‐3-‐phosphocholine)—and
its
corresponding
oxidized
product—
POxnoPC
(1-‐palmitoyl-‐2-‐(9'-‐oxo-‐nonanoyl)-‐sn-‐glycero-‐3-‐phosphocholine).
84,85

This

technique
is
an
effective
method
of
controlling
membrane
composition
to
correlate

oxidation
to
permeability.
The
oxidation
pathway
of
PLinPC
has
been
shown
to

result
in
lipid
tail
scission
yielding
POxnoPC
as
well
as
a
short
chain
aldehyde
as
the

cleaved
tail
group.
113

The
aldehyde
can
be
further
oxidized
to
a
carboxylic
acid.
A

representative
example
schematic
of
the
reaction
process
is
shown
in
Figure
21.

Both
of
the
cleaved
tail
fragments
are
expected
to
be
surface-‐active.
As
the
presence

of
surfactants
in
model
membranes
has
been
associated
with
the
formation
of
both

transient
and
stable
pores,
114

phase
destabilization
115

or
transition,
116

and
the

formation
of
micelles,
115

the
introduction
of
these
cleaved
tail
groups
are
integral
to

understanding
the
effects
of
lipid
oxidation
on
membrane
permeability.
In
our

previous
work
examining
the
effects
of
oxidation
on
bilayer
permeability,
only
the

POxnoPC
product
was
included;
112

here,
we
examine
the
effect
on
permeability
of

including
the
oxidation
products
corresponding
to
the
cleaved
tail
group.

52

It
is
expected
that
bilayer
permeability
should
be
related
to
bilayer
thickness.

Oxidation
has
been
linked
with
both
decreased
bilayer
thickness
and
increased
lipid

tail
group
interdigitation.
110,117,118

The
difference
between
a
non-‐interdigitated

bilayer
and
an
interdigitated
one
can
be
examined
by
looking
at
the
amount
of

bilayer
thickness
corresponding
to
the
tail
groups
of
each
leaflet.
119

For
a
non-‐
interdigitated
bilayer,
the
tails
of
each
leaflet
are
half
of
the
bilayer
thickness,
with

no
overlap
between
leaflets.
In
interdigitated
bilayers,
the
tail
groups
from
each

leaflet
overlap,
where
the
amount
of
overlap
corresponds
to
the
amount
of

interdigitation.
Previous
studies
have
shown
that
increasing
bilayer
thickness
yields

a
linear
decrease
in
permeability.
120

The
relationship
between
level
of

interdigitation
and
permeation
has
also
been
examined,
and
demonstrated
that

increasing
interdigitation
in
gel-‐phase
membranes
leads
to
increased

permeability.
121

In
the
present
report,
we
used
an
electrophysiological
technique
to

observe
changes
in
bilayer
thickness
upon
oxidation.
122-‐124

Giant
unilamellar
vesicles
(GUVs)
were
used
as
model
membranes
to
measure

passive
transport.
The
test
system
is
based
on
the
use
of
spinning
disk
confocal

microscopy
(SDCM)
to
measure
permeation
of
a
fluorescent
test
solute
in
a

microfluidic
channel.
60,61,112

The
GUVs
were
immobilized
in
a
simple
microfluidic
Y-‐
channel
using
a
biotin-‐avidin
interaction,
and
the
test
species
was
added
to
the

channel
using
a
rapid
microfluidic
buffer
exchange.
The
transport
process
was
then

imaged
until
it
reached
equilibrium.

53

The
oxidation
pathway
was
mimicked
by
varying
the
molar
ratio
of
PLinPC
to
its

oxidation
products
(POxnoPC
and
one
of
the
two
tail
fragments,
hexanal
or
hexanoic

acid)
in
GUVs.
113

The
molecular
structures
are
shown
in
Figure
21.
POxnoPC
and
the

tail
fragment
were
added
in
equimolar
amounts.
By
varying
the
ratio
of
the

unsaturated
lipid
to
the
oxidation
products,
we
can
simulate
different
points
along

the
oxidation
pathway.
It
has
previously
been
demonstrated
that
POxnoPC
is
a

major
termination
product
of
PLinPC
after
three
oxidative
steps.
85

The
first
two

steps
involve
abstracting
a
hydrogen
next
to
the
double
bond,
then
a
reaction

between
the
carbon-‐centered
radicals
with
singlet
oxygen
to
form
peroxyl
radicals.

The
third
and
final
step
results
in
termination
of
the
tail
group,
releasing
the
tail

fragment.

Figure
21:
Reaction
schematic
of
the
oxidation
process
of
(1)
PLinPC
into
(2)
POxnoPC
and
(3)

hexanal,
then
the
second
oxidation
step
of
(3)
hexanal
into
(4)
hexanoic
acid.

54

This
study
focuses
on
evaluating
the
effect
of
lipid
oxidation
on
passive
transport
of

a
fluorescent
short
chain
poly(ethylene
glycol)
molecule
(PEG12-‐NBD).
PEG12-‐NBD

is
an
uncharged,
hydrophilic
molecule
which
was
previously
characterized
by
Li
et

al.
in
2010
using
SDCM
to
measure
its
octanol/water
partition
coefficient
(0.20
±

0.02),
molecular
weight
(821.4
g/mol),
and
diffusivity
(3.6x10
-‐6

cm
2
/s).
60

Choosing

an
uncharged
species
simplifies
the
transport
process,
as
a
charged
molecule
can

interact
with
lipid
headgroups
and
demonstrate
a
pH-‐
or
charge-‐dependent

permeability.
91

3.2 Methods

3.2.1 Materials

1,2-‐dimyristoyl-‐sn-‐glycero-‐3-‐phosphocholine
(DMPC),
1-‐palmitoyl-‐2-‐linoleoyl-‐sn-‐
glycero-‐3-‐phosphocholine
(PLinPC),
cholesterol
(Chol),
1-‐palmitoyl-‐2-‐(9’-‐oxo-‐
nonanoyl)-‐sn-‐glycero-‐3-‐phosphocholine
(POxnoPC),
1,2-‐dihexadecanoyl-‐sn-‐glycero-‐
3-‐phosphoethanolamine-‐N-‐(cap
biotinyl)
(biotin-‐DPPE),
and
1,2-‐dipalmitoyl-‐sn-‐
glycero-‐3-‐phosphoethanolamine-‐N-‐(lissamine
rhodamine
B
sulfonyl)
(rhodamine-‐
DPPE)
were
purchased
from
Avanti
Polar
Lipids.
Amino-‐dPEG12-‐alcohol
was

purchased
from
Quanta
Biodesign,
and
poly(dimethylsiloxane)
(PDMS)
was

purchased
from
Dow
Chemical.
Succinimidyl
6-‐(N-‐(7-‐nitrobenz-‐3-‐oxa-‐1,3-‐diazol-‐4-‐
yl)amino)hexanoate
(NBD
NHS-‐ester)
and
avidin
were
obtained
from
Invitrogen.
All

other
chemicals
were
purchased
from
Sigma-‐Aldrich.

55

3.2.2 Vesicle
composition
and
formation

The
electroformation
technique
introduced
by
Angelova
et
al.
was
used
to
prepare

GUVs.
92,93

Prior
to
oxidation,
vesicle
composition
was
chosen
to
be
42.5:42.5:15

molar
ratio
of
DMPC:PLinPC:Chol.
0.5
mol%
α-‐tocopherol
was
added
to
the
stock

solution
of
PLinPC
to
act
as
an
oxygen
scavenger
and
prevent
additional
oxidation

during
electroformation
and
imaging.
0.01
mol%
rhodamine-‐DPPE
was
added
to

visualize
the
membrane.
To
attach
the
vesicles
to
the
coverslip,
we
added
6
mol%

biotin-‐DPPE.
The
lipids
were
dissolved
in
chloroform,
then
a
thin
film
was
spread
on

the
ITO-‐coated
surface
of
the
glass.
The
film
was
dried
in
a
vacuum
for
at
least
two

hours
prior
to
rehydration
in
200
mM
sucrose,
4
mM
HEPES
buffer
at
pH
7.0.

Electroformation
was
performed
with
an
oscillating
signal
of
1
V
at
10
Hz
for
two

hours
at
room
temperature.
Vesicle
compositions
are
shown
in
Table
2.

Table
2:
Vesicle
compositions
studied,
in
a
molar
ratio.
Tail
group
is
either
hexanal
or
hexanoic
acid.

Mol

DMPC

Mol

PLinPC

Mol

POxnoPC

Mol

Chol

Mol

Tail

42.5
42.5
0
15
0

42.5
40
2.5
15
2.5

42.5
37.5
5
15
5

42.5
35
7.5
15
7.5

42.5
32.5
10
15
10

42.5
30
12.5
15
12.5

42.5
27.5
15
15
15

42.5
24.5
18
15
18

GUVs
were
immediately
transferred
to
the
microfluidic
channel
after

electroformation.
Un-‐captured
vesicles
were
gently
flushed
from
the
system
using

200
mM
glucose,
4
mM
HEPES
buffer
at
pH
7.0
using
a
syringe
pump.

56

3.2.3 Preparation
of
test
molecule

PEG12-‐NBD
was
used
as
the
permeating
species.
Further
discussion
can
be
found
in

Section
2.2.3.

3.2.4 Microfluidic
channel
preparation
and
design

#1
coverslips
were
sonicated
in
three
solvents,
MilliQ
water
(Millipore),
isopropyl

alcohol,
and
methyl
alcohol,
each
for
30
minutes
at
45°C.
The
coverslips
were
then

dried
in
an
oven
at
45°C.
A
microfluidic
Y-‐channel
mold
was
formed
using
3D

printing
of
acrylonitrile
butadiene
styrene.

The
channel
had
a
depth
of
1
mm,
a

width
of
1
mm,
and
a
length
of
1
cm.
PDMS
was
then
cast
using
the
3D
printed
mold.

The
PDMS
and
a
clean
coverslip
were
oxidized
with
corona
treatment
(BD-‐20AC,

Electro-‐Technic
Products)
for
irreversible
bonding.
96

Vesicles
were
captured
and

immobilized
in
a
Y-‐junction
channel,
with
two
inlets
and
one
outlet.
A
schematic
of

the
channel
can
be
found
in
Figure
22.

Figure
22:
Schematic
of
the
microfluidic
device,
a
simple
Y-‐channel
design
with
two
inlets
and
a

single
outlet.
The
first
inlet
is
for
plain
buffer
and
the
second
is
for
buffer
containing
the
transport

solute.
The
glass
surface
was
functionalized
with
biotinylated-‐PEG-‐silane
solution
prior
to

experimentation.
After
channel
preparation,
a
1
mg/mL
avidin
solution
was
added
to
the
channel.
As

shown
in
the
inset,
vesicles
are
captured
in
this
device
using
a
biotin-‐avidin
interaction.

57

A
biotin-‐avidin
interaction
was
used
to
capture
the
vesicles.
After
bonding
the

channel,
the
glass
was
silanized
using
silane-‐PEG-‐biotin.
First,
the
silane-‐PEG-‐biotin

was
hydrolyzed
in
a
95:5
(v/v)
solution
of
ethanol
and
MilliQ
water.
After
five

minutes
of
hydrolysis,
50
μL
of
the
solution
was
added
to
the
channel.
The
solution

was
allowed
to
rest
for
10
minutes
in
the
channel,
then
was
flushed
from
the

channel
with
5x
excess
95:5
(v/v)
ethanol:water
solution.
The
silane
layer
was
then

cured
by
allowing
the
device
to
sit
at
room
temperature
for
24
hours.
The
coverslips

were
treated
with
a
1
mg/mL
solution
of
avidin
in
water
for
30
minutes,
then

flushed
with
plain
buffer
prior
to
addition
of
GUVs.
Vesicles
containing
6
mol%

biotin-‐DPPE
were
then
captured
at
the
glass
surface.

3.2.5 Transport
experiment
protocol

After
all
unbound
vesicles
were
flushed
from
the
channel,
a
single
unilamellar

vesicle
of
diameter
greater
than
10
μm
was
selected
for
observation.
A
buffer

solution
of
200
mM
glucose,
4
mM
HEPES,
and
5
μM
PEG12-‐NBD
at
pH
7.0
was

added
to
the
channel
using
a
syringe
pump.
A
flow
rate
of
5
mL/h
was
chosen
as
the

maximum
flow
rate
that
consistently
did
not
detach
or
damage
the
GUVs.
Spinning

disk
confocal
microscopy
with
a
Yokogawa
CSUX
confocal
head
on
a
Nikon
TI-‐E

inverted
microscope
was
used
to
observe
the
transport
process.
Illumination

sources
were
50
mW
solid-‐state
lasers
at
either
491
or
561
nm.
To
select
a

unilamellar
GUV
for
observation,
the
rhodamine-‐DPPE
was
excited
at
561
nm
with

emission
centered
at
595
nm.
PEG12-‐NBD
transport
was
imaged
with
excitation
at

491
nm
and
emission
centered
at
525
nm.
While
imaging
the
transport
process,
the

samples
were
only
illuminated
with
the
491
nm
laser.
Images
were
collected
at
a

58

regular
interval
during
the
buffer
exchange
to
visualize
the
fluorescent
species

crossing
the
GUV
membrane.

3.2.6 Pore
formation
protocol

The
presence
of
membrane
pores
was
determined
using
the
protocol
described
in

Section
2.2.6.

3.2.7 Measurement
of
bilayer
thickness

Lipid
bilayer
membranes
were
formed
on
clear
acrylic
chips
containing
three

collinear
wells,
as
described
previously.
125

The
outer
wells
were
connected
to
each

other
through
a
channel
on
the
bottom
of
the
chip,
and
the
center
well
was

connected
to
this
channel
through
a
150-‐250
μm
circular
aperture
in
a
75
μm
thick

sheet
of
delrin
(McMaster-‐Carr).
The
chip
was
designed
such
that
the
aperture
in
the

delrin
sheet,
and
therefore
the
lipid
bilayer,
were
oriented
horizontally
and
easily

imaged
through
the
clear
acrylic
with
an
inverted
microscope.

All
lipids
were
prepared
at
20
mg/mL
in
decane
and
allowed
to
shake
vigorously
for

1
hour
prior
to
use.
1
M
KCl,
10
mM
TRIS-‐HCl,
pH
8.0
was
used
as
the
aqueous
phase

in
all
experiments.
Lipid
bilayer
formation
was
adapted
from
Mueller
et
al.
36

Lipid
in

decane
was
applied
to
the
aperture
before
filling
the
chambers
with
aqueous

solution.
The
organic
solvent
coating
is
then
physically
manipulated
with
a
glass
rod

until
a
bilayer
is
formed,
as
observed
optically
or
electrically.

59

Ag/AgCl
electrodes
were
placed
in
one
outer
well
and
the
center
well
and
connected

to
a
custom
current
to
voltage
amplifier,
126

made
from
an
OPA111
op-‐amp
(Texas

Instruments)
and
a
1
GΩ
feedback
resistor.
The
output
of
this
circuit
was
captured

by
a
LabVIEW
(National
Instruments)
computer
data
acquisition
system.
Bilayer

capacitance
was
determined
by
applying
a
40
mV
8
Hz
triangle
wave
to
the

electrodes
while
measuring
the
resultant
current.

The
entire
apparatus
was
placed

in
a
small
aluminum
box
with
4
inch
square
central
cutouts
in
the
top
and
bottom

planes
and
mounted
on
a
Leica
DMIRBE
inverting
microscope.
The
box
was
used
as

a
Faraday
cage,
to
eliminate
background
noise.
Images
were
recorded
with
an

Optronics
TEC-‐470
CCD
camera
and
bilayer
area
was
quantified
through
image

analysis
using
ImageJ.

In
the
case
of
higher
levels
of
oxidation,
the
bilayers
were
only
stable
for
a
short

period
of
time.
In
these
cases,
the
Faraday
cage
could
not
be
replaced
prior
to

bilayer
rupture.
For
these
samples,
filtering
the
background
noise
was
necessary
to

analyze
membrane
capacitance.
Further
discussion
of
the
filtering
techniques
used

can
be
found
in
the
Section
3.3.1.

3.3 Data
Analysis

A
discussion
of
data
analysis
techniques
can
be
found
in
Section
2.3.

3.3.1 Filtering
capacitance
measurements
using
a
Fourier
transform

approach

When
analyzing
bilayer
capacitance,
a
Faraday
cage
was
placed
over
the
samples
to

eliminate
background
noise.
As
the
bilayer
can
be
modeled
as
a
capacitor,
applying
a

60

triangle
wave
voltage
across
the
membrane
should
yield
a
square
wave
current

response.
Bilayers
containing
higher
percentages
of
the
oxidized
products

(POxnoPC,
hexanal,
and
hexanoic
acid)
were
significantly
less
stable
than
less

oxidized
bilayers.
Between
5-‐7.5%
oxidation,
some
bilayers
were
only
stable
on
the

order
of
seconds,
making
collection
of
electrical
measurements
difficult.
In
these

cases,
the
bilayers
would
fuse
before
the
Faraday
cage
could
be
replaced
over
the

sample.
For
these
compositions,
the
measured
voltages
had
contributions
from
both

the
membrane
and
the
background
noise.

To
eliminate
the
background
noise,
we
took
a
Fourier
transform
of
the
voltage
vs.

time
signal,
to
convert
the
data
from
the
time
domain
to
the
frequency
domain.
Once

in
the
frequency
domain,
the
extra
peak
corresponding
to
the
60
Hz
background

noise
is
evident.
The
time-‐domain
and
frequency-‐domain
data
for
a
signal
with

background
noise
is
shown
in
Figure
23.
The
triangle
wave,
shown
in
blue,
was
the

voltage
across
the
membrane,
and
the
membrane
response
is
shown
in
red.

61

Figure
23:
Raw
voltage
data
collected
without
a
Faraday
cage.
For
bilayers
of
higher
oxidation
levels

(5-‐7.5%
oxidation),
the
bilayers
were
not
stable
for
long
enough
to
replace
the
cage,
leading
to

background
noise.
The
raw
data
in
the
time-‐domain
does
not
show
the
typical
square
wave
response

expected
of
a
capacitor.
The
inset
panel
shows
the
data
in
the
frequency-‐domain,
with
the

background
noise
peak
located
at
~60
Hz.

The
background
noise
peak
in
the
frequency-‐domain
is
located
at
approximately
60

Hz.
If
we
filter
the
data
by
eliminating
the
noise
peak,
we
achieve
the
frequency-‐
domain
spectra
shown
in
Figure
24
(inset).
With
an
inverse
Fourier
transform,
we

can
regain
our
original
signal
(Figure
24).
The
applied
voltage
triangle
wave
is

shown
in
blue,
and
the
membrane
response
is
shown
in
green.

62

Figure
24:
The
raw
data
from
Figure
23
after
filtering.
The
filter
was
applied
in
the
frequency

domain
(shown
in
the
inset
panel)
by
eliminating
the
noise
peak
at
~60
Hz.
An
inverse
Fourier

transform
was
then
used
to
convert
the
filtered
data
from
the
frequency-‐domain
to
the
time-‐domain.

After
filtering,
the
membrane
response
is
the
expected
square
wave
shape.

3.4 Results
and
discussion

As
established
in
Runas
and
Malmstadt,
vesicle
composition
was
chosen
to
prevent

phase
separation
prior
to
and
after
the
addition
of
the
oxidized
species.
112

PLinPC

and
POxnoPC
were
chosen
as
a
commercially
available
unsaturated
and
oxidized

pair.
DMPC
was
added
to
provide
a
non-‐oxidizing
background,
and
cholesterol
was

added
to
decrease
membrane
permeability
such
that
the
transport
process
was

sufficiently
slow
to
measure
precisely.
The
cholesterol
content
was
tailored
such

that
vesicles
did
not
show
phase
separation
after
addition
of
POxnoPC.
Membrane

63

structure
and
lack
of
phase
separation
for
each
GUV
was
confirmed
prior
to
each

transport
experiment.

GUV
composition
was
varied
to
simulate
the
oxidation
process
between
0
and
18

mol%
total
oxidation,
to
remain
below
the
previously
established
poration
limit.
90

To
simulate
lipid
oxidation,
portions
of
the
PLinPC
was
replaced
with
its

corresponding
oxidized
product
(POxnoPC),
and
one
of
the
cleaved
tail
groups

(hexanal
or
hexanoic
acid)
in
equimolar
amounts.

There
were
two
distinct
regimes
of
permeability
for
0-‐18
mol%
oxidation.
The
first

region
consists
of
vesicles
between
0-‐10
mol%
oxidation,
where
passive

permeability
could
be
explicitly
measured.
The
second
region
is
vesicles
with
12.5-‐
18
mol%
oxidation,
in
which
there
is
pore
formation
in
the
membrane.
Our
previous

work
on
vesicles
containing
PLinPC
with
POxnoPC
as
the
sole
oxidation
product

showed
three
distinct
regimes.
First,
vesicles
containing
0%
POxnoPC
had
a

permeability
on
the
order
of
1.5
x
10
-‐6

cm/s,
and
vesicles
with
2.5-‐10
mol%

POxnoPC
had
a
permeability
on
the
order
of
1.5
x
10
-‐5

cm/s.
112

Vesicles
with
12.5-‐18

mol%
POxnoPC
demonstrated
pore
formation.
112

Here,
we
demonstrate
that
the

addition
of
the
cleaved
tail
groups
caused
a
significant
decrease
in
measured

permeability
for
2.5-‐10
mol%
oxidation.

Figure
25a
shows
the
trend
between
measured
permeabilities
vs.
percent
oxidation

with
POxnoPC
and
hexanal.
There
are
three
main
conclusions
to
be
drawn
from
the

64

addition
of
the
tail
fragments:
(1)
the
cleaved
tail
fragments
drastically
change
the

membrane
response
to
small
amounts
of
oxidation,
(2)
membrane
permeability

varies
with
amount
of
oxidation
when
tail
fragments
are
present,
and
(3)
membrane

permeability
is
not
dependent
on
which
tail
fragment
is
present.
First,
with
the
tail

fragments
incorporated
into
the
membrane,
there
was
no
measurable
difference
in

permeability
between
0
and
2.5
mol%
oxidation.
Without
these
fragments,
there

was
an
order
of
magnitude
increase
with
that
slight
change
in
oxidation.
112

By

adding
the
tail
fragments
to
the
membrane,
we
can
negate
the
effects
of
small

amounts
of
lipid
oxidation.
Second,
when
oxidation
was
mimicked
by
replacing

PLinPC
with
only
POxnoPC,
membrane
permeability
was
characterized
by
two

distinct
regimes:
there
was
no
measurable
difference
between
permeability
at
2.5%

oxidation
vs.
10%
oxidation.
112

With
the
addition
of
the
tail
fragments,
however,
we

see
that
the
permeability
varies
with
percent
oxidation,
increasing
between
2.5-‐7.5

mol%
oxidation,
then
decreasing
slightly
at
10
mol%
oxidation
for
either
tail

fragment.
The
tail
fragment
also
decreases
the
membrane
permeability
when

compared
to
the
data
without
the
tail
fragment.
Similar
trends
are
demonstrated
for

the
measured
permeability
vs.
percent
oxidation
for
products
POxnoPC
and

hexanoic
acid,
as
shown
in
Figure
25b.

65

Figure
25:
Permeability
vs.
percent
oxidation
for
(a)
POxnoPC
and
hexanal
and
(b)
POxnoPC
and

hexanoic
acid
as
the
two
oxidation
products
of
PLinPC.
For
both
cleaved
tail
products,
there
is
no

difference
between
0
and
2.5
mol%
oxidation.
However,
there
is
an
increase
in
permeability
between

2.5-‐7.5
mol%
oxidation,
with
a
slight
decrease
in
permeability
at
10
mol%
oxidation
for
both
cleaved

tail
species.

a)

b)

66

These
three
trends
in
comparison
to
membranes
containing
only
POxnoPC
as
the

oxidized
product
are
shown
in
Figure
26.
Each
data
point
was
calculated
by
taking

the
weighted
average
and
weighted
uncertainty
of
the
data
presented
in
Figure
4.

None
of
the
compositions
examined
had
a
permeability
close
to
the
1.5
x
10
-‐5

cm/s

value
reported
for
vesicles
with
2.5-‐10%
POxnoPC.
112

Data
with
POxnoPC
as
the

only
oxidation
product
are
reproduced
from
Runas
and
Malmstadt.
112

Figure
26:
Average
permeability
for
vesicles
that
are
oxidized
using
POxnoPC
only
(squares),

POxnoPC
with
hexanal
(circles),
and
POxnoPC
with
hexanoic
acid
(triangles)

One
possible
explanation
for
this
behavior
is
that
the
presence
of
the
cleaved
tail

alters
the
bilayer
structure
and
thickness.
While
there
are
other
techniques
for

measuring
bilayer
thickness,
electrical
measurements
have
the
unique
characteristic

in
that
they
are
relatively
simple
to
set
up
and
analyze.
A
simple
brightfield

microscope
is
necessary
for
measuring
lipid
area.
Then,
a
triangle
wave
voltage
is

67

applied
to
the
membrane,
which
can
be
modeled
as
a
capacitor.
The
capacitive

current
should
then
be
a
square
wave.
By
finding
the
capacitance
from
the

measured
current
and
combining
this
with
the
microscopically
measured
area,
it
is

simple
to
determine
specific
capacitance
(capacitance/area),
which
is
inversely

proportional
to
bilayer
thickness.
Results
of
specific
capacitance
vs.
percent

oxidation
are
shown
in
Figure
27.
Compositions
containing
hexanal,
and
those
with

10
mol%
oxidation
products
did
not
form
stable
planar
bilayers
in
our
apparatus.

Figure
27:
Specific
capacitance
vs.
percent
oxidation
for
bilayers
containing
POxnoPC
(squares)
or

POxnoPC
and
hexanoic
acid
(triangles)
as
the
oxidation
product.

Our
data
demonstrate
that
the
presence
of
the
cleaved
tail
as
an
oxidation
product

leads
to
a
decrease
in
bilayer
thickness,
while
bilayers
that
have
only
POxnoPC
as

68

the
oxidation
product
have
a
thicker
bilayer
than
unoxidized
membranes.
While
the

behavior
of
oxidized
membranes
containing
only
POxnoPC
is
inconsistent
with

previous
studies
demonstrating
a
decrease
in
thickness
with
oxidation,
the

discrepancy
can
likely
be
explained
by
membrane
composition.
Membranes

containing
POxnoPC
have
been
demonstrated
to
show
an
increased
area
per
lipid,
as

the
POxnoPC
is
capable
of
reorienting
itself
inside
the
membrane.
127

With
the

inclusion
of
cholesterol,
a
bulky
species,
and
the
cylindrical
saturated
DMPC,
it
is
not

unreasonable
to
expect
that
the
reorientation
of
POxnoPC
will
cause
an
increase
in

bilayer
thickness.
Similarly,
this
reorientation
could
cause
the
increase
in

permeability
demonstrated
in
bilayers
containing
POxnoPC
as
the
oxidation

product.

Conversely,
the
addition
of
the
cleaved
tail
product
hexanoic
acid
caused
a
small

decrease
in
bilayer
thickness
for
2.5-‐5%
oxidation,
and
a
slight
increase
in
thickness

for
7.5%
oxidation.
This
behavior
is
consistent
with
previous
studies
suggesting
that

bilayer
thickness
decreases
with
oxidation,
confirming
that
the
inclusion
of
the

cleaved
tail
species
is
integral
to
mimicking
the
oxidation
pathway
in
a
bilayer.
It
is

worth
noting
that
bilayer
thickness
alone
is
not
an
adequate
predictor
of

permeability:
a
plot
of
permeability
vs.
capacitance
per
area
did
not
yield
any

obvious
trends
(Figure
28).

69

Figure
28:
Permeability
vs.
specific
capacitance
for
vesicles
with
0-‐7.5%
oxidation.
Neither
type
of

oxidation
product
shows
a
linear
trend
between
permeability
and
capacitance
per
area.

It
is
worth
noting
that
decane
has
been
demonstrated
to
incorporate
into
optically

black
lipid
membranes.
Decane
incorporates
in
these
membranes,
increasing
their

thickness.
128

For
these
experiments,
we
have
assumed
that
the
effect
of
decane
on

bilayer
thickness
is
consistent
regardless
of
varying
membrane
composition.
While

this
means
that
we
cannot
determine
the
thickness
of
the
membranes
used
to

measure
permeability
(which
contained
no
decane),
we
can
examine
the

relationships
between
trends
in
specific
capacitance
and
permeability.

At
higher
levels
of
oxidation,
both
oxidation
products
cause
pore
formation.
As

shown
in
Figure
29,
at
compositions
with
12.5-‐18
mol%
oxidation,
the
standard

70

electroformation
techniques
produce
a
low
yield
of
high
quality
vesicles.
Instead

most
vesicles
display
characteristics
such
as
small
vesicle
size,
tubule
formation,
or

irregular
shape.

Oxidation

Product

12.5%
15%
18%

Hexanal

Hexanoic
Acid

Figure
29:
SDCM
images
showing
vesicle
formation
for
12.5-‐18%
oxidation.
In
addition
to
small

spherical
vesicles
(<10
μm
in
diameter),
electroformation
yields
tubules
or
non-‐spherical
structures.

Scale
bar
is
20
μm,
and
all
images
are
the
same
resolution.

For
oxidation
amounts
less
than
12.5%,
the
two
fluorescein-‐dextran
species
could

not
permeate
the
membrane.
The
results
in
Figure
30
demonstrate
that

macromolecule-‐admitting
pores
are
formed
in
the
membrane
for
higher
levels
of

lipid
oxidation.
Since
the
2000
kDa
fluorescein-‐dextran
is
able
to
cross
the

membrane,
we
can
estimate
the
pore
size
from
the
hydrodynamic
radius
of
the

molecule.
It
has
previously
been
established
that
2000
kDa
fluorescein-‐dextran
has

a
radius
of
545
Å,
so
the
pore
radius
must
be
greater
than
this
value.
101

71

%

Hexanal

561

(40
kDa)

491

(40
kDa)

561

(2000
kDa)

491

(2000
kDa)

10%

12.5%

15%

18%

a)

72

%

Hexanoic

Acid

561

(40
kDa)

491

(40
kDa)

561

(2000
kDa)

491

(2000
kDa)

10%

12.5%

15%

18%

Figure
30:
SDCM
images
showing
pore
formation
for
vesicles
with
12.5-‐18%
(a)
hexanal
and
(b)

hexanoic
acid.
The
10%
oxidation
images
are
shown
as
a
reference
for
vesicles
without
pore

formation,
where
neither
species
of
fluorescein-‐dextran
could
enter
the
vesicle.
For
higher
amounts

of
oxidation,
both
the
40
kDa
and
the
2000
kDa
fluorescein-‐dextran
were
able
to
enter
the
vesicle.

Given
the
size
of
the
2000
fluorescein-‐dextran,
this
would
indicate
that
nanoscale
pores
are
present

in
the
membrane.
Scale
bar
is
20
μm,
and
all
images
are
the
same
resolution.

3.5 Conclusion

Our
results
demonstrate
the
significance
of
the
cleaved
tail
product
when
mimicking

the
oxidation
process
in
synthetic
lipid
bilayers.
While
this
work
and
the
work

published
by
Runas
and
Malmstadt
in
2015
are
unique
for
mimicking
the
oxidation

pathway
by
varying
the
ratio
of
unsaturated
to
oxidized
product
during
vesicle

b)

73

formation,
we
demonstrate
here
that
the
choice
of
oxidation
product
is
critical
for
a

true
model
of
the
oxidative
path.
112

The
tail
fragments
have
both
a
drastic
impact
on

permeability
and
bilayer
thickness,
as
demonstrated
here
by
measurements
of

specific
capacitance.
Both
results
were
surprising,
as
the
tail
fragments
are

surfactants,
which
are
commonly
known
for
their
membrane
destabilizing

properties.
114

When
added
to
membranes
also
containing
the
oxidized
POxnoPC,

however,
we
see
an
initial
decrease
in
permeability.
The
specific
capacitance
data

suggest
that
the
inclusion
of
these
tail
fragments
decreases
membrane
thickness.

The
increase
in
permeability
is
therefore
likely
the
result
of
altering
the
membrane

properties
in
some
other
way,
possibly
via
tighter
lipid
packing
in
the
bilayer.
As

with
oxidized
membranes
without
tail
groups,
the
tail
group-‐inclusive
membranes

formed
pores
at
12.5
mol%
oxidation
and
above,
suggesting
that
the
inclusion
of
the

tail
fragments
has
no
impact
on
pore
formation.
These
considerations
are
integral

when
choosing
a
composition
to
accurately
model
the
oxidation
pathway
in
a
lipid

bilayer.

4 Crossing
into
the
liquid-‐liquid
immiscibility
region

causes
a
drastic
change
in
the
permeability
of
lipid

bilayers

4.1 Motivation

Since
it
was
first
hypothesized
in
the
late
1990s,
129

the
topic
of
physically
relevant

lateral
lipid
segregation
in
biological
systems
has
been
a
source
of
controversy.
130,131

This
“lipid
raft
hypothesis”
holds
that
the
separation
of
lipids
in
the
plasma

74

membrane
into
liquid
ordered
(lo)
and
liquid
disordered
(ld)
phases
corresponds
to

biologically
relevant
phenomena.
129

The
ordered
phase
is
concentrated
in
saturated

lipids
and
cholesterol,
while
the
disordered
phase
contains
primarily
unsaturated

lipids.
132

Liquid
ordered
rafts
are
thought
to
associate
proteins
such
as
GPI-‐
anchored
proteins
and
certain
transmembrane
proteins.
133-‐137

Signal
transduction
is

hypothesized
to
be
linked
to
rafts,
as
several
signaling
molecules
are
believed
to

partition
into
the
lo
phase.
137-‐139

While
the
topic
of
lipid
liquid-‐liquid
phase

separation
was
sporadically
examined
throughout
the
1980s,
direct
observation
of

lipid
rafts
wasn’t
demonstrated
in
model
membranes
with
three
or
more

components
until
the
early
2000s.
140,141

This
type
of
direct
demonstration
of
lipid

rafts
is
difficult
to
apply
in
biological
systems,
142

making
model
membranes
the

optimal
platform
for
studying
lipid
rafts.

In
biomimetic
membranes,
preferential

partitioning
of
commonly
used
fluorescent
probes
into
one
of
the
phases
is
a
simple

method
for
identifying
phase
separation;
143

as
such,
the
presence
of
lo
/
ld

macroscopic
phase
separation
in
ternary
systems
of
synthetic
or
purified
lipids
has

been
examined
in
great
detail
using
fluorescence
microscopy.
140,141,144

An
example

of
a
vesicle,
with
rhodamine-‐DPPE
partitioned
into
the
disordered
phase,
is
shown

in
Figure
31.
Förster
resonance
energy
transfer
(FRET)
measurements
have
been

used
to
determine
the
presence
of
nanoscopic
domain
segregation
in

liposomes.
145,146

By
tagging
both
the
lo
and
the
ld

phases
with
a
FRET
pair
of
dyes,
a

change
in
the
ratio
of
donor
to
acceptor
fluorescent
signals
can
indicate
domain

formation
that
cannot
be
resolved
by
traditional
microscopy
techniques.
145,147

75

Figure
31:
SDCM
image
showing
hemispherical
phase
separation
in
a
GUV.
The
dye,
rhodamine-‐
DPPE,
partitions
into
the
disordered
phase.
Note
that
the
ordered
phase
is
difficult
to
see,
as
no
dye

has
partitioned
into
it.
The
scale
bar
is
10
μm.

Nuclear
magnetic
resonance
spectroscopy
is
another
frequently
used
technique
for

examining
phase
separation
in
model
membranes.
106,132,145,148

For
example,
the

biological
role
of
liquid-‐liquid
phase
separation
has
also
been
approached
through

examination
of
critical
fluctuations
using
NMR.
149

At
a
critical
point
composition,

when
temperature
is
varied
a
transition
occurs
and
the
mixture
transitions
from
a

co-‐existing
liquid
phases
to
a
single
liquid
phase.
150

These
critical
points
typically

exist
on
the
edge
of
the
miscibility
boundary.
151,152

It
has
been
suggested
that

biological
systems
have
been
evolutionarily
tuned
to
exist
at
a
critical
composition,

as
membrane
organization
at
a
critical
point
requires
the
minimal
amount
of
energy

to
maintain.
153

The
impact
of
this
phase
separation
could
be
relevant
for
all

biological
systems
tuned
to
the
critical
point.

76

Little
is
known
about
the
nature
and
properties
of
the
physical
boundary
between

phase-‐separated
domains.
At
this
location,
there
is
a
hydrophobic
mismatch

between
the
two
phases.
154

The
lipids
at
this
boundary
will
likely
deform,
possibly

by
changing
their
tilt
and
splay,
in
order
to
minimize
the
line
tension.
155

The

deformation
of
lipids
at
the
boundary
could
potentially
lead
to
defects.
If
this
is
the

case,
domain
boundaries
could
represent
regions
of
increased
permeability.
Here,

we
measure
the
permeability
of
phase-‐separated
synthetic
lipid
bilayers
to

determine
what
impact
the
presence
of
phase
boundaries
might
have
on
membrane

permeability.

We
evaluate
the
impact
of
phase
separation
on
the
permeability
of
two
small

molecule
species:
a
fluorescent
short
chain
poly(ethylene
glycol)
molecule
(PEG12-‐
NBD),
and
sodium
fluorescein
(Na-‐Fl).
Both
molecules
have
been
previously
used
as

test
solutes
in
lipid
bilayer
permeation
experiments.
156-‐159

They
were
chosen

because
they
are
small,
hydrophilic,
but
uncharged.
159

Choosing
an
uncharged

species
simplifies
the
transport
model,
as
charged
species
can
interact
with
the
lipid

headgroups,
and
have
ionic
strength-‐
or
pH-‐dependent
permeability.
91

The

fluorescent
nature
of
these
two
solutes
makes
imaging
the
transport
process
simple

with
spinning
disk
confocal
microscopy,
as
we
have
demonstrated
previously.
60,61,112

4.2 Materials
and
methods

4.2.1 Materials

1,2-‐dipalmitoyl-‐sn-‐glycero-‐3-‐phosphocholine
(DPPC),
1,2-‐dioleoyl-‐sn-‐glycero-‐3-‐
phosphocholine
(DOPC),
cholesterol,
1,2-‐dihexadecanoyl-‐sn-‐glycero-‐3-‐

77

phosphoethanolamine-‐N-‐(cap
biotinyl)
(biotin-‐DPPE),
and
1,2-‐dipalmitoyl-‐sn-‐
glycero-‐3-‐phosphoethanolamine-‐N-‐(lissamine
rhodamine
B
sulfonyl)
(rhodamine-‐
DPPE)
were
purchased
from
Avanti
Polar
Lipids.
Avidin
and
succinimidyl
6-‐(N-‐(7-‐
nitrobenz-‐2-‐oxa-‐1,3-‐diazol-‐4-‐yl)amino)Hexanoate
(NBD
NHS-‐ester)
were
obtained

from
Invitrogen.
Amino-‐dPEG12-‐alcohol
was
purchased
from
Quanta
Biodesign.

Poly(dimethylsiloxane)
(PDMS)
was
purchased
from
Dow
Chemical,
and
the
indium-‐
tin-‐oxide
(ITO)
coated
glass
was
obtained
from
Delta
Technologies.
α-‐tocopherol

and
all
other
chemicals
were
purchased
from
Sigma-‐Aldrich.

4.2.2 GUV
composition
and
formation

GUVs
were
prepared
with
the
electroformation
technique
introduced
by
Angelova
et

al.

92,93

Vesicle
composition
was
chosen
for
hemispherical
phase
separation,
liquid

ordered,
and
liquid
disordered
phase
compositions
based
on
the
ternary
phase

diagram
at
25°C
(reproduced
from
Veatch
et
al.,
2004,
shown
in
Figure
32).
106

78

Figure
32:
Phase
diagram
replicated
from
Veatch
et
al.
in
2004.
106

The
gray
area
is
the
two-‐phase

region,
with
the
thick
black
dashed
indicating
a
tie
line.
The
thinner
dashed
lines
indicate
the

uncertainty
on
the
tie
line.
The
colored
points
represent
the
percent
distance
along
the
tie
line,
with

the
pure
disordered
phase
composition
located
at
0,
and
pure
ordered
phase
located
at
100.
The

membrane
composition
at
each
point
is
discussed
further
in
Table
3.

Rhodamine-‐DPPE
was
added
at
0.01
mol%
to
visualize
the
membrane.
PLinPC
was

stored
with
0.5
mol%
α-‐tocopherol
to
act
as
an
oxygen
scavenger
and
prevent

additional
oxidation
during
electroformation
and
imaging.
94

6
mol%
biotin-‐DPPE

was
used
to
attach
the
GUVs
to
the
avidin-‐treated
coverslip.
The
lipids
were

dissolved
in
chloroform,
then
spread
into
a
thin
film
on
the
surface
of
the
ITO
coated

79

glass.
The
film
was
dried
in
a
vacuum
for
at
least
two
hours
before
rehydration
in

pH
7.0
buffer
with
200
mM
sucrose
and
4
mM
HEPES.
Electroformation
was

performed
at
room
temperature
with
a
1
V
signal
at
10
Hz
for
two
hours.

Table
3:
Vesicle
compositions
studied.

%
Along

Tie
Line

Mol%

DPPC

Mol%

DOPC

Mol%

Chol

0
15
65
20

5
17
62
21

15
21
56
23

25
25
50
25

50
35
35
30

75
43
20
37

95
51
8
41

100
53
5
42

After
electroformation,
GUVs
were
immediately
transferred
to
the
microfluidic

channel.
The
GUVs
not
captured
in
the
channel
were
gently
flushed
from
the
system

using
a
syringe
pump
and
glucose
buffer.

4.2.3 Preparation
of
test
molecule

The
technique
for
forming
the
test
molecule,
PEG12-‐NBD,
can
be
found
in
Section

2.2.4.
Sodium
fluorescein
(Na-‐Fl)
is
available
in
powder
form
from
Sigma-‐Aldrich.
A

stock
solution
of
8
μM
Na-‐Fl
in
4
mM
HEPES,
pH
7.4
buffer
was
used
to
measure

transport
across
the
membrane.
The
diffusivity
of
Na-‐Fl
in
water
at
300K
has
been

experimentally
measured
as
3.9x10
-‐6

cm
2
/s,
160

while
molecular
dynamics

simulations
predict
a
diffusivity
of
4.2x10
-‐6

cm
2
/s.
161

For
the
purpose
of
these

experiments,
the
experimental
value
of
3.9x10
-‐6

cm
2
/s
was
the
estimated
diffusivity

of
Na-‐Fl
at
25°C.

80

4.2.4 Microfluidic
channel
preparation
and
design

Details
on
the
Y-‐channel
design
with
a
silanized
glass
surface
for
vesicle
capture
can

be
found
in
Section
3.2.4.

4.2.5 Transport
experiment
protocol

After
all
unbound
vesicles
have
been
removed,
a
single
GUV
was
chosen
for

observation.
The
selected
vesicle
was
greater
than
10
μm
in
diameter,
and
for
some

compositions,
displayed
phase
separation.
For
each
of
the
solutes,
two
buffers
were

exchanged
for
the
transport
process.
The
compositions
of
these
buffers
can
be
found

in
Table
4.
The
plain
buffer
(a
buffer
not
containing
the
transport
solute
PEG12-‐NBD

or
Na-‐Fl)
was
used
to
flush
the
system
prior
to
measuring
permeation.
Next,
the

buffer
containing
the
small
molecule
of
interest
was
added
at
a
flow
rate
of
5
mL/h.

This
flow
rate
was
chosen
as
the
highest
flow
rate
that
would
not
damage
or
detach

the
GUVs
from
the
coverslip.

As
discussed
previously,
as
the
Na-‐Fl
formed
a

complex
with
the
sugars
present
in
the
buffer,
experiments
to
measure
Na-‐Fl

transport
used
buffers
with
no
glucose
or
sucrose.

Table
4:
Buffer
compositions
used
for
transport
experiments.
Each
test
solute
(PEG12-‐NBD
and
Na-‐
Fl)
had
a
“plain”
buffer
solution,
and
a
transport
buffer
solution
that
contained
the
solute
of
interest.

Buffer
mM

Glucose

mM

HEPES

μM

PEG12-‐NBD

μM

Na-‐Fl

pH

Plain

(PEG12-‐NBD)

200
4
0
0
7.0

PEG12-‐NBD
200
4
5
0
7.0

Plain

(Na-‐Fl)

0
4
0
0
7.4

Na-‐Fl
0
4
0
8
7.4

81

A
Nikon
TI-‐E
inverted
microscope
with
a
Yokogawa
CSUX
confocal
head
was
used
to

perform
spinning
disk
confocal
microscopy
to
observe
the
transport
process.
50

mW
solid-‐state
lasers
of
either
491
or
561
nm
were
used
to
illuminate
the
samples.

The
561
nm
laser
with
emission
centered
at
595
nm
was
used
to
excite
the

rhodamine-‐DPPE.
This
channel
was
exclusively
used
to
select
a
GUV
for
analysis.
For

vesicles
showing
phase
separation,
the
561
nm
channel
was
used
to
create
a
3D
map

of
the
phase
structure.
After
choosing
a
vesicle,
solute
permeation
was
imaged
with

excitation
at
491
nm
and
emission
centered
at
525
nm.
Only
the
491
nm
laser
was

used
to
illuminate
the
sample
during
the
course
of
the
permeation
experiment.

Images
were
collected
at
regular
intervals
as
the
fluorescent
species
permeated
the

membrane.

4.3 Data
Analysis

A
discussion
of
data
analysis
techniques
can
be
found
in
Section
2.3.

4.3.1 Using
map
projection
techniques
to
measure
surface
area
and

perimeter
of
the
liquid
disordered
phase

A
3D
z-‐stack
series
of
images
was
acquired
for
each
vesicle
prior
to
measuring

membrane
permeability.
These
z-‐stacks
of
images
(shown
in
Figure
33
as
mean

value
projections)
were
then
used
to
identify
two
relevant
parameters
for
phase-‐
separated
vesicles:
percent
disordered
phase
surface
area,
and
normalized

perimeter
of
the
disordered
phase.
In
order
to
measure
these
parameters
from
the

z-‐stack,
a
map
projection
technique
was
implemented
to
turn
the
three-‐dimensional

sphere
surface
into
a
two-‐dimensional
map
of
the
two
phases.

82

Figure
33:
Ternary
phase
diagram
with
two
phase
region
indicated
in
gray.
Tie
line
is
indicated
by

the
thick
dashed
line,
with
uncertainty
of
the
tie
line
in
the
thinner
dashed
line.
Phase
diagram

reproduced
from
Veatch
et
al.
2004.
106

Three
inset
images
show
representative
z-‐stacks
of
(a)
25%,

(b)
50%,
and
(c)
95%
distance
along
the
tie
line.
At
these
three
compositions,
vesicles
demonstrated

clear
phase
separation.
Scale
bars
are
10
μm.

Gall-‐Peters
equal-‐area
projection
equations
were
used
to
maintain
the
area

measurements
for
each
phase.
An
array
of
pixel
values
corresponding
to
the
z-‐stack

was
analyzed
using
Matlab.
The
globe
surface
was
defined
using
the
measured

radius
(r)
of
the
vesicle.
Each
set
of
xyz
coordinates
was
converted
to
longitude
(λ)

and
latitude
(φ)
in
radians.
Each
coordinate
set
was
then
analyzed
using
the
Gall-‐
Peters
equations,
shown
as
Equation
7
and
Equation
8.
The
surface
was
projected

onto
a
2D
map
surface.
From
this
map,
a
built-‐in
Matlab
edge
detection
algorithm

was
then
used
to
identify
the
liquid
disordered
and
liquid
ordered
phases.

𝑥= 𝑟𝜆

(7)

83

𝑦= 2𝑟𝑠𝑖𝑛 𝜑
(8)

Next,
the
percent
surface
area
of
the
liquid
disordered
phase
could
be
calculated

based
on
the
number
of
bright
pixels
to
the
total
number
of
pixels
on
the
vesicle

surface.
Finally,
the
number
of
edge
pixels
was
used
to
identify
the
perimeter
value

of
the
disordered
phase.
This
perimeter
was
normalized
to
the
vesicle
diameter,
to

make
the
value
dimensionless.

4.4 Results

Phase
separation
in
a
ternary
system
has
two
notable
consequences
for
membrane

permeability.
First,
compositions
near
the
border
of
the
two-‐phase
region
without

visible
phase
separation
show
drastic
changes
in
permeability
for
small

compositional
variance.
Second,
vesicles
showing
phase
separation
have
a
large

variation
in
permeability,
potentially
indicating
the
presence
of
bilayer
disruption

along
the
phase
boundary.

4.4.1 Measured
permeability
for
vesicles
with
no
microscopic
phase

separation

The
first
significant
result
focuses
on
membrane
behavior
when
crossing
the
phase

boundary.
Vesicles
at
these
compositions
(0,
5,
and
100%
distance
along
the
tie-‐
line)
did
not
show
microscopically
observable
phase
separation.
As
shown
in
Figure

34,
a
5%
change
in
tie-‐line
position
across
the
two-‐phase
boundary
from
0-‐5%

results
in
a
drastic
change
in
permeability.
For
the
solute
PEG12-‐NBD
(Figure
34a),

the
variation
in
permeability
across
the
boundary
is
over
one
order
of
magnitude.

For
the
solute
Na-‐Fl
(Figure
34b)
the
permeability
variation
is
approximately
4x.

84

Figure
34:
Permeability
vs.
distance
along
tie
line
for
(a)
PEG12-‐NBD
and
(b)
Na-‐Fl
as
the
solute
of

interest.
Inset
panel
shows
the
location
of
the
four
points
on
the
phase
diagram.
The
5%
change
in

(a)

(b)

85

distance
along
the
tie-‐line
shows
a
drastic
change
in
permeability
across
the
phase
boundary
for
both

solutes.

It
has
been
commonly
recognized
that
increased
cholesterol
concentration

decreases
membrane
permeability;
162

however,
it
is
important
to
note
that
the

sharp
change
in
permeability
at
the
edge
of
the
two-‐phase
region
is
too
drastic
to
be

caused
by
the
change
in
cholesterol
concentration.
As
shown
in
Table
3,
a
5%

change
in
tie-‐line
position
corresponds
to
a
1
mol%
change
in
cholesterol.

4.4.2 Permeability
of
binary
vesicle
compositions

To
test
the
impact
of
a
slight
change
in
cholesterol,
we
measured
the
permeability
of

a
binary
system
containing
DOPC
and
cholesterol.
By
examining
a
binary
system,
we

were
able
to
avoid
any
effects
of
phase
separation
on
measured
transport
rates.
All

vesicles
contained
6
mol%
biotin-‐DPPE,
and
0.01
mol%
rhodamine-‐DPPE.

The

compositions
for
these
experiments
are
shown
in
Table
5.

Table
5:
Binary
vesicle
compositions
studied.

Mol%
DOPC
Mol%
Chol

100
0

90
10

80
20

70
30

The
total
amount
of
cholesterol
was
kept
below
the
solubility
limit
to
prevent

crystallization
of
cholesterol.
163

Permeability
of
these
membranes
was
measured

using
PEG12-‐NBD,
as
discussed
previously.
The
results
of
permeability
vs.
percent

cholesterol
are
shown
in
Figure
35.
By
comparing
the
results
in
Figure
35
with

Figure
34a,
we
can
see
that
a
30%
change
in
cholesterol
is
necessary
for
the
order
of

magnitude
change
demonstrated
across
the
phase
boundary
for
PEG12-‐NBD.

86

Figure
35:
Permeability
vs.
percent
cholesterol
for
binary
mixtures
of
DOPC
and
cholesterol.

Permeation
was
measured
using
PEG12-‐NBD.
These
results
can
be
compared
to
the
permeability

measured
in
Figure
34a,
where
a
1%
change
in
cholesterol
between
0-‐5%
distance
along
the
tie-‐line

yields
an
order
of
magnitude
change
in
measured
permeability.
To
achieve
an
order
of
magnitude

decrease
in
permeability
of
the
binary
system,
a
30%
change
in
cholesterol
content
is
necessary.

4.4.3 Permeability
of
vesicles
with
microscopic
phase
separation

The
second
significant
result
is
the
impact
of
visible
phase
separation
on
measured

membrane
permeability.
For
the
compositions
measured,
positions
between
25-‐
95%
position
along
the
tie-‐line
showed
microscopically
observable
domain

formation.
For
these
compositions,
the
measured
permeability
varied
by
as
much
as

an
order
of
magnitude.
This
variation
was
consistent
regardless
of
transport
solute:

as
shown
in
Figure
36,
both
PEG12-‐NBD
and
Na-‐Fl
had
widely
variable
permeability

measurements
for
vesicles
demonstrating
phase
separation.

87

While
this
inconsistency
appears
to
be
connected
to
the
appearance
of
phase

separated
domains,
it
is
important
to
note
that
there
were
no
observable
trends

between
permeability
and
surface
area
of
the
domains,
or
between
permeability
and

perimeter
between
the
domains
(further
discussion
in
Section
4.4.4
and
4.4.5).
This

would
indicate
that,
while
the
existence
of
phase
separation
causes
some
sort
of

membrane
disruption
leading
to
inconsistent
permeability
measurements,
the

relationship
between
permeability
and
amount
of
phase
separation
is
more
complex

than
simply
surface
area
or
perimeter.

88

Figure
36:
Permeability
vs.
distance
along
tie
line
for
(a)
PEG12-‐NBD
and
(b)
Na-‐Fl
as
the
solute
of

interest.
Inset
panel
shows
the
location
of
the
four
points
on
the
phase
diagram.
For
vesicle

compositions
showing
visible
phase
separation,
there
was
significant
variation
in
the
measured

(a)

(b)

89

permeability
values.
This
would
indicate
that
the
presence
of
phase
separation
causes
some
sort
of

disruption
along
the
phase
boundary.

4.4.4 Examining
the
relationship
between
membrane
permeability
and

disordered
phase
surface
area

For
each
of
the
two
test
solutes,
PEG12-‐NBD
and
Na-‐Fl,
the
percent
disordered

phase
surface
area
was
calculated
for
each
vesicle
demonstrating
phase
separation.

The
compositions
with
visible
phase
segregation
were
25%,
50%,
and
95%
distance

along
the
tie
line,
as
shown
in
Figure
36.
After
calculating
the
percent
disordered

phase,
we
could
examine
the
relationship
with
permeability.
Figure
37
shows

permeability
vs.
percent
disordered
phase
surface
area
for
25%,
50%,
and
95%

distance
along
the
tie
line
for
(a)
PEG12-‐NBD
and
(b)
Na-‐Fl.

90

Figure
37:
Permeability
of
(a)
PEG12-‐NBD
and
(b)
Na-‐Fl
vs.
percent
disordered
phase
surface
area.

The
compositions
shown
had
visible
phase
separation:
25%
distance
along
the
tie-‐line
(squares),

50%
(triangles),
and
95%
(circles).
Note
that
25%
distance
data
was
not
collected
for
Na-‐Fl.
There
is

no
apparent
trend
for
either
solute.

(a)

(b)

91

No
trends
were
apparent
for
either
solute
in
any
individual
composition,
or
for
all

three
compositions.
While
it
is
likely
that
there
is
some
relationship
between

permeability
and
phase
behavior,
given
the
variation
in
measured
permeabilities
for

vesicles
showing
visible
phase
separation,
the
correlation
is
not
as
straightforward

as
permeability
and
disordered
phase
surface
area.

4.4.5 Examining
the
relationship
between
membrane
permeability
and

disordered
phase
perimeter

A
second
possibility
is
that
the
permeability
variation
is
dependent
on
the
perimeter

of
the
phase
boundary.
After
calculating
the
perimeter
of
the
disordered
phase
for

each
vesicle,
then
normalizing
it
to
the
vesicle
diameter,
we
plotted
permeability
vs.

normalized
perimeter
for
the
same
three
compositions
shown
in
Figure
36.
Figure

38
shows
the
permeability
vs.
normalized
perimeter
for
25%,
50%,
and
95%

distance
along
the
tie
line
for
(a)
PEG12-‐NBD
and
(b)
Na-‐Fl.

92

Figure
38:
Permeability
of
(a)
PEG12-‐NBD
and
(b)
Na-‐Fl
vs.
normalized
phase
perimeter.
The

compositions
shown
had
visible
phase
separation:
25%
distance
along
the
tie-‐line
(squares),
50%

(a)

(b)

93

(triangles),
and
95%
(circles).
Note
that
25%
distance
data
was
not
collected
for
Na-‐Fl.
There
is
no

apparent
trend
for
either
solute.

Again,
there
is
no
apparent
trend
between
permeability
and
normalized
perimeter.

While
it
is
likely
that
the
permeability
does
vary
with
phase
boundary,
given
the

existence
of
hydrophobic
mismatch
154

and
lipid
deformation
to
minimize
line

tension
155
,
the
correlation
was
not
a
simple
one.
It
is
possible
that
permeability

varies
with
both
surface
area
and
perimeter
of
the
phase
boundary.

4.5 Discussion

The
first
result
demonstrates
the
importance
of
compositional
changes
at
the
edge

of
the
miscibility
region.
Near
these
boundaries,
a
minor
change
in
composition

causes
a
change
in
permeability
by
a
factor
of
ten
(PEG12-‐NBD)
or
a
factor
of
four

(Na-‐Fl)
without
visible
phase
separation.
The
variance
between
the
compositions
of

0
and
5
mol%
ordered
phase
is
minimal,
and
could
not
account
for
the
drastic

change
in
permeability.
One
possible
explanation
for
this
dramatic
variation
is
the

presence
of
nanodomains
too
small
to
resolve
using
microscopy
techniques.
While

these
nanodomains
could
not
be
directly
imaged,
it
is
possible
that
their
presence

could
cause
a
shift
in
permeability
different
than
that
caused
by
macroscopic

domain
formation.

The
second
result
indicates
some
degree
of
bilayer
destabilization
caused
by
phase

separation,
as
the
presence
of
visible
domain
segregation
leads
to
unpredictably

varying
permeability.
One
possible
explanation
is
the
idea
of
critical
behavior
at
the

two-‐phase
region
boundary.
Critical
points
in
lipid
mixtures
have
been
associated

94

with
decreased
line
tension,
150

reduced
interfacial
energy,
164

and
decreased
lateral

lipid
diffusivity.
153

It
has
been
previously
demonstrated
that
the
hydrophobic

mismatch
at
the
phase
boundary
154

could
cause
lipid
deformation
in
order
to

minimize
line
tension.
155

This
deformation
instead
appears
to
cause
varying
degrees

of
bilayer
disruption,
which
in
turn
cause
a
variable
permeability
measurement
for
a

given
composition
for
vesicles
presenting
with
phase
separation.
However,
our

results
indicate
that
there
is
not
a
simple
relationship
between
permeability
and

domain
surface
area,
or
permeability
and
perimeter
of
the
domain
(see
Supporting

Information).
While
it
is
evident
that
the
presence
of
micrometer-‐sized
domains

leads
to
widely
varying
permeability
measurements
for
a
single
composition,
it
is

not
what
the
source
of
this
variation
is.

Figure
39
summarizes
the
results
for
each
of
the
seven
compositions
evaluated
for

PEG12-‐NBD
permeation.
The
largest
change
in
permeability
occurs
between
the
0-‐
5%
distance
along
the
tie-‐line,
when
moving
from
left
to
right.
For
vesicles
with

visible
phase
separation,
the
measured
permeability
values
were
significantly
more

variable
than
for
the
non-‐phase
separated
points.
Identical
trends
were

demonstrated
for
a
second
test
solute,
Na-‐Fl.

95

Figure
39:
Permeability
of
PEG12-‐NBD
at
seven
points
along
the
tie
line
(0,
5,
25,
50,
75,
95,
100%

distance
along
the
tie-‐line).
Square
points
show
compositions
without
phase
separation,
triangle

points
show
points
with
visible
phase
separation.
The
color
scale
bar
indicates
the
range
of

permeability
for
each
measured
composition.
For
compositions
with
visible
phase
separation,
the

variation
in
permeability
is
much
greater
than
for
compositions
without
phase
separation.

4.6 Conclusions

Our
results
contribute
additional
understanding
of
how
phase
coexistence
could

alter
cell
membrane
properties.
Most
notably,
a
small
compositional
change
near
the

phase
boundary
resulted
in
a
drastic
change
in
permeability
without
visible
phase

separation,
far
beyond
what
could
be
explained
by
the
change
in
cholesterol

concentration.
A
similar
transition
from
single
phase
to
two
immiscible
phases
can

be
reached
through
either
temperature
152,153

or
composition
fluctuation
106

at
the

critical
point.
As
it
has
been
suggested
that
biological
membranes
have
been

evolutionarily
tuned
to
critical
compositions,
153

the
relationship
between
this

96

transition
and
permeability
could
have
significant
impact
on
transport
in
biological

systems.
Additionally,
our
results
support
the
idea
that
lipid
deformation
at
the

phase
boundary
causes
bilayer
disruption,
which
results
in
irregular
permeability

measurements
for
vesicles
demonstrating
phase
segregation.
However,
the

relationship
between
lipid
raft
area,
perimeter,
and
permeability
is
not
simple.

Further
studies
of
phase
separated
systems
are
necessary
to
fully
understand
the

role
of
lipid
rafts
on
membrane
permeability.

5 Summary
and
Outlook

The
significance
of
plasma
membrane
permeability
cannot
be
understated,
as

passive
diffusion
represents
a
universal
pathway
for
small
molecules
such
as
drugs

and
toxins.
The
primary
contribution
of
this
work
is
a
furthered
understanding
of

how
membrane
characteristics
impact
the
transport
of
small
molecule
species

through
the
development
of
an
assay
combining
spinning
disk
confocal
microscopy

and
microfluidics.
Development
of
a
high-‐throughput
passive
transport
assay
of

highly
controllable
cellular
membranes
has
been
a
primary
goal
of
the

pharmaceutical
industry,
as
it
is
the
key
to
rapid
screening
potential
drug

candidates.
Here,
we
present
a
novel
technique
capable
of
high-‐throughput

screening
while
achieving
carefully
controlled
compositions
with
minimal

assumptions.

We
demonstrated
the
significant
impact
that
membrane
composition
can
have
on

permeability.
By
creating
vesicles
with
varying
levels
of
oxidation
for
two
different

97

types
of
oxidation
pathway
models,
we
can
identify
the
significance
of
the
pathway

when
measuring
permeability
and
other
characteristics.
These
results
give
new

insight
into
the
importance
of
oxidative
stress
on
membrane
function
and
barrier

behavior.
Additional
research
demonstrated
that
phase
separation
has
a
measurable

impact
on
the
transport
properties
of
bilayers.
This
is
particularly
significant,
as
it

has
been
suggested
that
biological
membranes
have
been
evolutionarily
tuned
to

compositions
displaying
phase
segregation.
Understanding
the
relationship

between
biologically
relevant
characteristics
and
permeability
has
implications
for

both
the
membrane
biophysics
and
the
drug
delivery
communities.

While
this
work
contributes
to
our
understanding
of
how
membrane
composition

and
behavior
affect
passive
transport,
it
is
also
the
potential
foundation
for
two

fields
of
study:
the
creation
of
a
physiologically
relevant
synthetic
cell
structure,
and

the
mechanism
of
passive
transport.
First,
while
the
lipid
bilayers
presented
here

are
reasonable
mimics
of
the
plasma
membrane,
future
research
can
further
the

model
to
incorporate
other
complexities,
such
as
proteins.
The
inclusion
of
such

molecules
can
provide
more
insight
to
cellular
transport.
For
a
new
technique
to

overtake
the
traditional
PAMPA
method,
the
cells
used
must
be
able
to
reliably

model
the
complexity
of
an
actual
cellular
structure.
Second,
while
we
have
made

strides
in
the
development
of
a
new
high-‐throughput
screening
technology
for

permeability,
the
assay
is
not
complete.
Future
studies
could
focus
on
adapting
this

method
for
use
with
non-‐fluorescent
species,
or
analyzing
assays
of
vesicles

simultaneously.
Additionally,
this
research
has
focused
primarily
on
how
membrane

98

characteristics
impact
permeability.
Future
work
could
examine
the
role
of
the

solute
in
transport
by
examining
properties
such
as
hydrogen
bond
potential.
With

the
goal
of
furthering
the
knowledge
of
how
both
small
molecules
and
bilayer

characteristics
alter
passive
transport,
this
work
is
the
backbone
for
a
breadth
of

study
in
the
fields
of
lipid
biophysics,
drug
delivery,
and
biomedical
research.

6 Acknowledgements

The
author
would
like
to
thank
Dr.
Su
Li
for
her
help
establishing
the
transport

experiment
protocol,
and
Dr.
Carson
Riche
for
his
help
with
the
design
and

fabrication
of
the
microfluidic
devices.
We
would
also
like
to
thank
Dr.
Rich
Roberts,

Dr.
Terry
Takahashi,
and
Farzad
Jalali-‐Yazdi
for
their
assistance
with
HPLC.
This

work
was
supported
by
the
Office
of
Naval
Research
(Young
Investigator
Award

N00014-‐12-‐1-‐0620),
the
National
Science
Foundation
(Award
CBET-‐1067021),
the

National
Institutes
of
Health
(Award
1R01GM093279),
and
a
Mork
Family
Doctoral

Fellowship
award.

99

7 References

1.

Artursson,
P.,
J.
Karlsson.
Correlation
between
oral
drug
absorption
in
humans
and
apparent
drug

permeability
coefficients
in
human
intestinal
epithelial
(Caco-‐2)
cells.
Biochemical
and

Biophysical
Research
Communications.
175
(3):
880-‐885,
1991.

2.

Veber,
D.F.,
S.R.
Johnson,
H.-‐Y.
Cheng,
et
al.
Molecular
properties
that
influence
the
oral

bioavailability
of
drug
candidates.
Journal
of
Medicinal
Chemistry.
45
(12):
2615-‐2623,
2002.

3.

Lipnick,
R.L.
Base-‐line
toxicity
predicted
by
quantitative
structure-‐activity
relationships
as
a
probe

for
molecular
mechanism
of
toxicity.
Probing
Bioactive
Mechanisms.
413:
366-‐389,
1989.

4.

Foulkes,
E.
Transport
of
toxic
heavy
metals
across
cell
membranes.
Proceedings
of
the
Society
for

Experimental
Biology
and
Medicine.
223
(3):
234-‐240,
2000.

5.

Sikkema,
J.,
J.
De
Bont,
B.
Poolman.
Mechanisms
of
membrane
toxicity
of
hydrocarbons.

Microbiological
Reviews.
59
(2):
201-‐222,
1995.

6.

Li,
S.,
N.
Malmstadt.
Deformation
and
poration
of
lipid
bilayer
membranes
by
cationic

nanoparticles.
Soft
Matter.
9
(20):
4969-‐4976,
2013.

7.

Yacobi,
N.R.,
N.
Malmstadt,
F.
Fazolallahi,
et
al.
Mechanisms
of
alveolar
epithelial
translocation
of
a

defined
population
of
nanoparticles.
American
Journal
of
Respiratory
Cell
and
Molecular
Biology.

42
(5):
604-‐614,
2010.

8.

Bedrov,
D.,
G.D.
Smith,
H.
Davande,
et
al.
Passive
transport
of
C-‐60
fullerenes
through
a
lipid

membrane:
A
molecular
dynamics
simulation
study.
Journal
of
Physical
Chemistry
B.
112
(7):

2078-‐2084,
2008.

9.

Wang,
T.,
J.
Bai,
X.
Jiang,
et
al.
Cellular
uptake
of
nanoparticles
by
membrane
penetration:
a
study

combining
confocal
microscopy
with
FTIR
spectroelectrochemistry.
ACS
Nano.
6
(2):
1251-‐
1259,
2012.

10.

Verma,
A.,
O.
Uzun,
Y.H.
Hu,
et
al.
Surface-‐structure-‐regulated
cell-‐membrane
penetration
by

monolayer-‐protected
nanoparticles.
Nature
Materials.
7
(7):
588-‐595,
2008.

11.

Jackson,
M.
Drug
transport
across
gastrointestinal
epithelia.
Physiology
of
the
Gastrointestinal

Tract.
2:
1597,
1987.

12.

Yee,
S.
In
vitro
permeability
across
Caco-‐2
cells
(colonic)
can
predict
in
vivo
(small
intestinal)

absorption
in
man—fact
or
myth.
Pharmaceutical
Research.
14
(6):
763-‐766,
1997.

13.

Kerns,
E.H.,
L.
Di,
S.
Petusky,
et
al.
Combined
application
of
parallel
artificial
membrane

permeability
assay
and
Caco-‐2
permeability
assays
in
drug
discovery.
Journal
of
Pharmaceutical

Sciences.
93
(6):
1440-‐1453,
2004.

14.

Wilson,
G.,
I.
Hassan,
C.
Dix,
et
al.
Transport
and
permeability
properties
of
human
Caco-‐2
cells:

an
in
vitro
model
of
the
intestinal
epithelial
cell
barrier.
Journal
of
Controlled
Release.
11
(1):
25-‐
40,
1990.

15.

Pade,
V.,
S.
Stavchansky.
Link
between
drug
absorption
solubility
and
permeability

measurements
in
Caco ‐2
cells.
Journal
of
Pharmaceutical
Sciences.
87
(12):
1604-‐1607,
1998.

16.

Karlsson,
J.,
P.
Artursson.
A
method
for
the
determination
of
cellular
permeability
coefficients

and
aqueous
boundary
layer
thickness
in
monolayers
of
intestinal
epithelial
(Caco-‐2)
cells

grown
in
permeable
filter
chambers.
International
Journal
of
Pharmaceutics.
71:
55-‐64,
1991.

17.

Karlsson,
J.,
P.
Artursson.
A
new
diffusion
chamber
system
for
the
determination
of
drug

permeability
coefficients
across
the
human
intestinal
epithelium
that
are
independent
of
the

unstirred
water
layer.
Biochimica
et
Biophysica
Acta.
1111:
204-‐210,
1992.

18.

Behrens,
I.,
W.
Kamm,
A.H.
Dantzig,
et
al.
Variation
of
peptide
transporter
(PepT1
and
HPT1)

expression
in
Caco ‐2
cells
as
a
function
of
cell
origin.
Journal
of
Pharmaceutical
Sciences.
93

(7):
1743-‐1754,
2004.

19.

Kleinzeller,
A.
Ernest
Overton's
contribution
to
the
cell
membrane
concept:
a
centennial

appreciation.
Physiology.
12:
4,
1997.

20.

Missner,
A.,
P.
Pohl.
110
years
of
the
Meyer-‐Overton
rule:
predicting
membrane
permeability
of

gases
and
other
small
compounds.
ChemPhysChem.
10
(9-‐10):
1405-‐1414,
2009.

21.

Overton,
C.E.
Studies
of
Narcosis,
(ed.
and
trans.
by
R.
L.
Lipnick).
New
York:
Chapman
and
Hall

Ltd.
and
The
Wood
Library-‐Museum
of
Anesthesiology,
1991.
(Original
Published
in
1901).

100

22.

Finkelstein,
A.
Water
and
nonelectrolyte
permeability
of
lipid
bilayer
membranes.
Journal
of

General
Physiology.
68
(2):
127-‐135,
1976.

23.

Seader,
J.D.,
E.J.
Henley,
D.K.
Roper.
Separation
Process
Principles
(3
ed.).
New
York:
Wiley,
2010

24.

Wolosin,
J.M.,
H.
Ginsburg.
Permeation
of
organic-‐acids
through
lecithin
bilayers
resemblance
to

diffusion
in
polymers.
Biochimica
et
Biophysica
Acta.
389
(1):
20-‐33,
1975.

25.

Walter,
A.,
J.
Gutknecht.
Permeability
of
small
nonelectrolyes
through
lipid
bilayer
membranes.

Journal
of
Membrane
Biology.
90:
207-‐217,
1986.

26.

Lipinsky,
C.A.,
F.
Lobardo,
B.W.
Dominy,
et
al.
Experimental
and
computational
approaches
to

estimate
solubility
and
permeability
in
drug
discovery
and
development
settings.
Advanced

Drug
Delivery
Reviews.
46:
3-‐26,
2001.

27.

Wohnsland,
F.,
B.
Faller.
High-‐throughput
permeability
pH
profile
and
high-‐throughput

alkane/water
log
P
with
artificial
membranes.
Journal
of
Medicinal
Chemistry.
44:
923-‐930,

2001.

28.

Bassolino-‐Klimas,
D.,
H.E.
Alper,
T.R.
Stouch.
Solute
diffusion
in
lipid
bilayer
membranes:
an

atomic
level
study
by
molecular
dynamics
simulation.
Biochemistry.
32
(12624-‐12637),
1993.

29.

Marrink,
S.J.,
H.J.C.
Berendsen.
Simulation
of
water
transport
through
a
lipid
membrane.
Journal

of
Physical
Chemistry.
98:
4155-‐4186,
1994.

30.

Marrink,
S.J.,
H.J.C.
Berendsen.
Permeation
process
of
small
molecules
across
lipid
membranes

studied
by
molecular
dyanmics
simulations.
Journal
of
Physical
Chemistry.
100:
16729-‐16738,

1996.

31.

Bemporad,
D.,
C.
Luttmann,
J.W.
Essex.
Computer
simulation
of
small
molecule
permeation
across

a
lipid
bilayer:
dependence
on
bilayer
properties
and
solute
volume,
size,
and
cross-‐sectional

area.
Biophysical
Journal.
87
(1):
1-‐13,
2004.

32.

Bemporad,
D.,
J.W.
Essex,
C.
Luttmann.
Permeation
of
small
molecules
through
a
lipid
bilayer:
a

computer
simulation
study.
Journal
of
Physical
Chemistry
B.
108:
4975-‐4884,
2004.

33.

Orsi,
M.,
W.E.
Sanderson,
J.W.
Essex.
Permeability
of
small
molecules
through
a
lipid
bilayer:
a

multiscale
simulation
study.
Journal
of
Physical
Chemistry
B.
113:
12019-‐12029,
2009.

34.

Orsi,
M.,
J.W.
Essex.
Permeability
of
drugs
and
hormones
through
a
lipid
bilayer:
insights
from

dual-‐resolution
molecular
dynamics.
Soft
Matter.
6
(16):
3797,
2010.

35.

Vacha,
R.,
M.L.
Berkowitz,
P.
Jungwirth.
Molecular
model
of
a
cell
plasma
membrane
with
an

asymmetric
multicomponent
composition:
water
permeation
and
ion
effects.
Biophysical

Journal.
96
(11):
4493-‐4501,
2009.

36.

Mueller,
P.,
D.O.
Rudin,
H.
Ti
Tien,
et
al.
Reconstitution
of
cell
membrane
structure
in
vitro
and
its

transformation
into
an
excitable
system.
Nature.
194:
979-‐980,
1962.

37.

Bangham,
A.,
M.
Standish,
J.
Watkins.
Diffusion
of
univalent
ions
across
the
lamellae
of
swollen

phospholipids.
Journal
of
Molecular
Biology.
13
(1):
238-‐IN227,
1965.

38.

Barry,
P.H.,
J.M.
Diamond.
Effects
of
unstirred
layers
on
membrane
phenomena.
The
American

Physiological
Society.
64
(3):
763-‐872,
1984.

39.

Escuder-‐Gilabert,
L.,
J.J.
Martı́nez-‐Pla,
S.
Sagrado,
et
al.
Biopartitioning
micellar
separation

methods:
modelling
drug
absorption.
Journal
of
Chromatography
B.
797
(1-‐2):
21-‐35,
2003.

40.

Zhu,
C.,
L.
Jiang,
T.-‐M.
Chen,
et
al.
A
comparative
study
of
artificial
membrane
permeability
assay

for
high
throughput
profiling
of
drug
absorption
potential.
European
Journal
of
Medicinal

Chemistry.
37
(5):
399-‐407,
2002.

41.

Kansy,
M.,
F.
Senner,
K.
Gubernator.
Physicochemical
high
throughput
screening:
parallel

artificial
membrane
permeation
assay
in
the
description
of
passive
absorption
processes.

Journal
of
Medicinal
Chemistry.
41
(7):
1007-‐1010,
1998.

42.

Leung,
S.S.,
J.
Mijalkovic,
K.
Borrelli,
et
al.
Testing
physical
models
of
passive
membrane

permeation.
Journal
of
Chemical
Information
and
Modeling.
52
(6):
1621-‐1636,
2012.

43.

Antonenko,
Y.N.,
A.A.
Bulychev.
Measurements
of
local
pH
changes
near
bilayer
lipid
membrane

by
means
of
a
pH
microelectrode
and
a
protonophore-‐dependent
membrane
potential.

Comparison
of
the
methods.
Biochimica
et
Biophysica
Acta
-‐
Biomembranes.
1070
(1):
279-‐282,

1991.

44.

Dzekunov,
S.M.,
Y.N.
Antonenko.
Dynamics
of
formation
and
dissipation
of
local
pH
gradients
in

the
unstirred
layers
near
bilayer
lipid
membranes.
Bioelectrochemistry
and
Bioenergetics.
41:

187-‐190,
1996.

101

45.

Ohki,
S.
Membrane
Potential
and
ion
permeability
of
lipid
bilayer
membranes.

Bioelectrochemistry
and
Bioenergetics.
7:
487-‐501,
1980.

46.

Sandeaux,
R.,
J.
Sandeaux,
C.
Gavach,
et
al.
Transport
of
Na+
by
monensin
across
biomolecular

lipid
membranes.
Biochimica
et
Biophysica
Acta.
684:
127-‐132,
1982.

47.

Gutknecht,
J.,
A.
Walter.
Transport
of
protons
and
hydrochloric
acid
through
lipid
bilayer

membranes.
Biochimica
et
Biophysica
Acta.
641:
183-‐188,
1981.

48.

Gutknecht,
J.,
A.
Walter.
Hydroxyl
ion
permeability
of
lipid
bilayer
membranes.
Biochimica
et

Biophysica
Acta.
645:
161-‐162,
1981.

49.

Gutknecht,
J.,
A.
Walter.
Hydrofluoric
and
nitric
acid
transport
through
lipid
bilayer
membranes.

Biochimica
et
Biophysica
Acta
-‐
Biomembranes.
644
(1):
153-‐156,
1981.

50.

Walter,
A.,
J.
Gutknecht.
Monocarboxylic
acid
permeation
through
lipid
bilayer
membranes.

Journal
of
Membrane
Biology.
77
(3):
255-‐264,
1984.

51.

Antonenko,
Y.N.,
G.A.
Denisov,
P.
Pohl.
Weak
acid
transport
across
bilayer
lipid
membrane
in
the

presence
of
buffers.
Biophysical
Journal.
64:
1701-‐1710,
1993.

52.

Pohl,
P.,
E.
Rosenfelt,
R.
Millner.
Effects
of
ultrasound
on
the
steady
state
transmembrane
pH

gradient
and
permeability
of
acetic
acid
through
bilayer
lipid
membranes.
Biochimica
et

Biophysica
Acta.
1145:
279-‐283,
1993.

53.

Antonenko,
Y.N.,
P.
Pohl.
Steady-‐state
nonmonotonic
concentration
profiles
in
the
unstirred

layers
of
bilayer
lipid
membranes.
Biochimica
et
Biophysica
Acta
-‐
Biomembranes.
1235
(1):
57-‐
61,
1995.

54.

Krylov,
A.V.,
P.
Pohl,
M.L.
Zeidel,
et
al.
Water
permeability
of
asymmetric
planar
lipid
bilayers:

leaflets
of
different
composition
offer
independent
and
additive
resistances
to
permeation.

Journal
of
General
Physiology.
118:
333-‐339,
2001.

55.

Antonenko,
Y.N.,
P.
Pohl,
G.A.
Denisov.
Permeation
of
ammonia
across
bilayer
lipid
membranes

studied
by
ammonium
ion
selective
microelectrodes.
Biophysical
Journal.
72:
2187-‐2195,
1997.

56.

Sha'Afi,
R.,
G.
Rich,
V.W.
Sidel,
et
al.
The
effect
of
the
unstirred
layer
on
human
red
cell
water

permeability.
Journal
of
General
Physiology.
50
(5):
1377-‐1399,
1967.

57.

Lawaczeck,
R.
Water
permeability
through
biological
membranes
by
isotopic
effects
of

fluorescence
and
light
scattering.
Biophysical
Journal.
45
(3):
491-‐494,
1984.

58.

Mlekoday,
H.J.,
R.
Moore,
D.G.
Levitt.
Osmotic
water
permeability
of
the
human
red
cell.

Dependence
on
direction
of
water
flow
and
cell
volume.
Journal
of
General
Physiology.
81
(2):

213-‐220,
1983.

59.

Holland,
R.,
H.
Shibata,
P.
Scheid,
et
al.
Kinetics
of
O2
uptake
and
release
by
red
cells
in
stopped-‐
flow
apparatus:
effects
of
unstirred
layer.
Respiration
Physiology.
59
(1):
71-‐91,
1985.

60.

Li,
S.,
P.
Hu,
N.
Malmstadt.
Confocal
imaging
to
quantify
passive
transport
across
biomimetic
lipid

membranes.
Analytical
Chemistry.
82
(18):
7766-‐7771,
2010.

61.

Li,
S.,
P.C.
Hu,
N.
Malmstadt.
Imaging
Molecular
Transport
across
Lipid
Bilayers.
Biophysical

Journal.
101
(3):
700-‐708,
2011.

62.

Heuvingh,
J.,
S.
Bonneau.
Asymmetric
oxidation
of
giant
vesicles
triggers
curvature-‐associated

shape
transition
and
permeabilization.
Biophysical
Journal.
97
(11):
2904-‐2912,
2009.

63.

Riske,
K.A.,
T.P.
Sudbrack,
N.L.
Archilha,
et
al.
Giant
vesicles
under
oxidative
stress
induced
by
a

membrane-‐anchored
photosensitizer.
Biophysical
Journal.
97
(5):
1362-‐1370,
2009.

64.

Sankhagowit,
S.,
R.
Biswas,
S.-‐H.
Wu,
et
al.
The
dynamics
of
giant
unilamellar
vesicle
oxidation

probed
by
morphological
transitions.
Biochimica
et
Biophysica
Acta.
1828
(10):
2615-‐2624,

2014.

65.

Borst,
J.W.,
N.V.
Visser,
O.
Kouptsova,
et
al.
Oxidation
of
unsaturated
phospholipids
in
membrane

bilayer
mixtures
is
accompanied
by
membrane
fluidity
changes.
Biochimica
et
Biophysica
Acta
-‐

Molecular
and
Cell
Biology
of
Lipids.
1487
(1):
61-‐73,
2000.

66.

Volinsky,
R.,
R.
Paananen,
P.K.
Kinnunen.
Oxidized
phosphatidylcholines
promote
phase

separation
of
cholesterol-‐sphingomyelin
domains.
Biophysical
Journal.
103
(2):
247-‐254,
2012.

67.

Jacob,
R.F.,
R.P.
Mason.
Lipid
peroxidation
induces
cholesterol
domain
formation
in
model

membranes.
Journal
of
Biological
Chemistry.
280
(47):
39380-‐39387,
2005.

68.

Ross,
R.
The
pathogenesis
of
atherosclerosis:
a
perspective
for
the
1990s.
Nature.
362:
801-‐809,

1993.

102

69.

Cadenas,
E.,
K.J.
Davies.
Mitochondrial
free
radical
generation,
oxidative
stress,
and
aging.
Free

Radical
Biology
and
Medicine.
29
(3):
222-‐230,
2000.

70.

Tyurina,
Y.,
A.
Shvedova,
K.
Kawai,
et
al.
Phospholipid
signaling
in
apoptosis:
peroxidation
and

externalization
of
phosphatidylserine.
Toxicology.
148
(2):
93-‐101,
2000.

71.

Kagan,
V.E.,
B.
Gleiss,
Y.Y.
Tyurina,
et
al.
A
role
for
oxidative
stress
in
apoptosis:
oxidation
and

externalization
of
phosphatidylserine
is
required
for
macrophage
clearance
of
cells
undergoing

Fas-‐mediated
apoptosis.
The
Journal
of
Immunology.
169
(1):
487-‐499,
2002.

72.

Fabisiak,
J.P.,
Y.Y.
Tyurina,
V.A.
Tyurin,
et
al.
Random
versus
selective
membrane
phospholipid

oxidation
in
apoptosis:
role
of
phosphatidylserine.
Biochemistry.
37
(39):
13781-‐13790,
1998.

73.

Radi,
R.,
J.S.
Beckman,
K.M.
Bush,
et
al.
Peroxynitrite-‐induced
membrane
lipid
peroxidation:
the

cytotoxic
potential
of
superoxide
and
nitric
oxide.
Archives
of
Biochemistry
and
Biophysics.
288

(2):
481-‐487,
1991.

74.

Esterbauer,
H.
Cytotoxicity
and
genotoxicity
of
lipid-‐oxidation
products.
The
American
Journal
of

Clinical
Nutrition.
57
(5):
779S-‐785S,
1993.

75.

Chen,
R.,
L.
Yang,
T.M.
McIntyre.
Cytotoxic
phospholipid
oxidation
products
cell
death
from

mitochondrial
damage
and
the
intrinsic
caspase
cascade.
Journal
of
Biological
Chemistry.
282

(34):
24842-‐24850,
2007.

76.

Morel,
D.W.,
J.R.
Hessler,
G.M.
Chisolm.
Low
density
lipoprotein
cytotoxicity
induced
by
free

radical
peroxidation
of
lipid.
Journal
of
Lipid
Research.
24
(8):
1070-‐1076,
1983.

77.

Parhami,
F.,
A.D.
Morrow,
J.
Balucan,
et
al.
Lipid
oxidation
products
have
opposite
effects
on

calcifying
vascular
cell
and
bone
cell
differentiation
A
possible
explanation
for
the
paradox
of

arterial
calcification
in
osteoporotic
patients.
Arteriosclerosis,
Thrombosis,
and
Vascular
Biology.

17
(4):
680-‐687,
1997.

78.

Epstein,
F.H.,
R.
Ross.
Atherosclerosis—an
inflammatory
disease.
New
England
Journal
of

Medicine.
340
(2):
115-‐126,
1999.

79.

Witztum,
J.L.
The
oxidation
hypothesis
of
atherosclerosis.
The
Lancet.
344
(8925):
793-‐795,

1994.

80.

Stocker,
R.,
J.F.
Keaney.
Role
of
oxidative
modifications
in
atherosclerosis.
Physiological
Reviews.

84
(4):
1381-‐1478,
2004.

81.

Berliner,
J.A.,
J.W.
Heinecke.
The
role
of
oxidized
lipoproteins
in
atherogenesis.
Free
Radical

Biology
and
Medicine.
20
(5):
707-‐727,
1996.

82.

Harrison,
D.,
K.K.
Griendling,
U.
Landmesser,
et
al.
Role
of
oxidative
stress
in
atherosclerosis.
The

American
Journal
of
Cardiology.
91
(3):
7-‐11,
2003.

83.

Witztum,
J.L.,
D.
Steinberg.
Role
of
oxidized
low
density
lipoprotein
in
atherogenesis.
Journal
of

Clinical
Investigation.
88
(6):
1785,
1991.

84.

Wong-‐Ekkabut,
J.,
Z.
Xu,
W.
Triampo,
et
al.
Effect
of
lipid
peroxidation
on
the
properties
of
lipid

bilayers:
a
molecular
dynamics
study.
Biophysical
Journal.
93
(12):
4225-‐4236,
2007.

85.

Reis,
A.,
C.M.
Spickett.
Chemistry
of
phospholipid
oxidation.
Biochimica
et
Biophysica
Acta
-‐

Biomembranes.
1818
(10):
2374-‐2387,
2012.

86.

Haluska,
C.K.,
M.S.
Baptista,
A.U.
Fernandes,
et
al.
Photo-‐activated
phase
separation
in
giant

vesicles
made
from
different
lipid
mixtures.
Biochimica
et
Biophysica
Acta
-‐
Biomembranes.

1818
(3):
666-‐672,
2012.

87.

Dröge,
W.
Free
radicals
in
the
physiological
control
of
cell
function.
Physiological
Reviews.
82
(1):

47-‐95,
2002.

88.

Lambeth,
J.D.
Nox/Duox
family
of
nicotinamide
adenine
dinucleotide
(phosphate)
oxidases.

Current
Opinion
in
Hematology.
9
(1):
11-‐17,
2002.

89.

Megli,
F.M.,
L.
Russo,
K.
Sabatini.
Oxidized
phospholipids
induce
phase
separation
in
lipid

vesicles.
FEBS
Letters.
579
(21):
4577-‐4584,
2005.

90.

Volinsky,
R.,
L.
Cwiklik,
P.
Jurkiewicz,
et
al.
Oxidized
phosphatidylcholines
facilitate
phospholipid

flip-‐flop
in
liposomes.
Biophysical
Journal.
101
(6):
1376-‐1384,
2011.

91.

Seo,
P.R.,
Z.S.
Teksin,
J.P.
Kao,
et
al.
Lipid
composition
effect
on
permeability
across
PAMPA.

European
Journal
of
Pharmaceutical
Sciences.
29
(3-‐4):
259-‐268,
2006.

92.

Angelova,
M.I.,
D.S.
Dimitrov.
Liposome
electroformation.
Faraday
Discussions
of
the
Chemical

Society.
81:
303-‐311,
1986.

103

93.

Angelova,
M.,
S.
Soleau,
P.
Méléard,
et
al.
Preparation
of
giant
vesicles
by
external
AC
electric

fields.
Kinetics
and
applications.
Progress
in
Colloid
and
Polymer
Science.
89:
127-‐131,
1992.

94.

Buettner,
G.R.
The
pecking
order
of
free
radicals
and
antioxidants:
lipid
peroxidation,
α-‐
tocopherol,
and
ascorbate.
Archives
of
Biochemistry
and
Biophysics.
300
(2):
535-‐543,
1993.

95.

Kersten,
G.F.,
D.J.
Crommelin.
Liposomes
and
ISCOMS
as
vaccine
formulations.
Biochimica
et

Biophysica
Acta
-‐
Biomembranes.
1241
(2):
117-‐138,
1995.

96.

Gregoriadis,
G.,
P.
Leathwood,
B.E.
Ryman.
Enzyme
entrapment
in
liposomes.
FEBS
Letters.
14

(2):
95-‐99,
1971.

97.

Anderson,
J.R.,
D.T.
Chiu,
H.
Wu,
et
al.
Fabrication
of
microfluidic
systems
in
poly

(dimethylsiloxane).
Electrophoresis.
21:
27-‐40,
2000.

98.

Robinson,
T.,
P.
Kuhn,
K.
Eyer,
et
al.
Microfluidic
trapping
of
giant
unilamellar
vesicles
to
study

transport
through
a
membrane
pore.
Biomicrofluidics.
7
(4):
044105,
2013.

99.

Sandison,
D.R.,
W.W.
Webb.
Background
rejection
and
signal-‐to-‐noise
optimization
in
confocal

and
alternative
fluorescence
microscopes.
Applied
Optics.
33
(4):
603-‐615,
1994.

100.

Somersalo,
E.,
R.
Occhipinti,
W.F.
Boron,
et
al.
A
reaction–diffusion
model
of
CO
2
influx
into
an

oocyte.
Journal
of
Theoretical
Biology.
309:
185-‐203,
2012.

101.

Venturoli,
D.,
B.
Rippe.
Ficoll
and
dextran
vs.
globular
proteins
as
probes
for
testing
glomerular

permselectivity:
effects
of
molecular
size,
shape,
charge,
and
deformability.
American
Journal
of

Physiology
-‐
Renal
Physiology.
288
(4):
F605-‐F613,
2005.

102.

Mandal,
T.,
S.
Chatterjee.
Ultraviolet-‐and
sunlight-‐induced
lipid
peroxidation
in
liposomal

membrane.
Radiation
Research.
83
(2):
290-‐302,
1980.

103.

Janero,
D.R.
Malondialdehyde
and
thiobarbituric
acid-‐reactivity
as
diagnostic
indices
of
lipid

peroxidation
and
peroxidative
tissue
injury.
Free
Radical
Biology
and
Medicine.
9
(6):
515-‐540,

1990.

104.

Mertins,
O.,
I.O.
Bacellar,
F.
Thalmann,
et
al.
Physical
damage
on
giant
vesicles
membrane
as
a

result
of
methylene
blue
photoirradiation.
Biophys
J.
106
(1):
162-‐171,
2014.

105.

Chatterjee,
S.,
S.
Agarwal.
Liposomes
as
membrane
model
for
study
of
lipid
peroxidation.
Free

Radical
Biol
Med.
4
(1):
51-‐72,
1988.

106.

Veatch,
S.,
I.
Polozov,
K.
Gawrisch,
et
al.
Liquid
domains
in
vesicles
investigated
by
NMR
and

fluorescence
microscopy.
Biophysical
Journal.
86
(5):
2910-‐2922,
2004.

107.

Kerdous,
R.,
J.
Heuvingh,
S.
Bonneau.
Photo-‐dynamic
induction
of
oxidative
stress
within

cholesterol-‐containing
membranes:
shape
transitions
and
permeabilization.
Biochimica
et

Biophysica
Acta
-‐
Biomembranes.
1808
(12):
2965-‐2972,
2011.

108.

Caetano,
W.,
P.S.
Haddad,
R.
Itri,
et
al.
Photo-‐induced
destruction
of
giant
vesicles
in
methylene

blue
solutions.
Langmuir.
23
(3):
1307-‐1314,
2007.

109.

Sabatini,
K.,
J.-‐P.
Mattila,
F.M.
Megli,
et
al.
Characterization
of
two
oxidatively
modified

phospholipids
in
mixed
monolayers
with
DPPC.
Biophysical
Journal.
90
(12):
4488-‐4499,
2006.

110.

Beranova,
L.,
L.
Cwiklik,
P.
Jurkiewicz,
et
al.
Oxidation
changes
physical
properties
of

phospholipid
bilayers:
fluorescence
spectroscopy
and
molecular
simulations.
Langmuir.
26
(9):

6140-‐6144,
2010.

111.

Praticò,
D.,
L.
Iuliano,
A.
Mauriello,
et
al.
Localization
of
distinct
F2-‐isoprostanes
in
human

atherosclerotic
lesions.
Journal
of
Clinical
Investigation.
100
(8):
2028,
1997.

112.

Runas,
K.A.,
N.
Malmstadt.
Low
levels
of
lipid
oxidation
radically
increase
the
passive

permeability
of
lipid
bilayers.
Soft
Matter.
11
(3):
499-‐505,
2015.

113.

Gardner,
H.,
R.
Plattner.
Linoleate
hydroperoxides
are
cleaved
heterolytically
into
aldehydes
by

a
Lewis
acid
in
aprotic
solvent.
Lipids.
19
(4):
294-‐299,
1984.

114.

Groot,
R.,
K.
Rabone.
Mesoscopic
simulation
of
cell
membrane
damage,
morphology
change
and

rupture
by
nonionic
surfactants.
Biophysical
Journal.
81
(2):
725-‐736,
2001.

115.

Almgren,
M.
Mixed
micelles
and
other
structures
in
the
solubilization
of
bilayer
lipid

membranes
by
surfactants.
Biochimica
et
Biophysica
Acta
-‐
Biomembranes.
1508
(1):
146-‐163,

2000.

116.

Inoue,
T.,
K.
Miyakawa,
R.
Shimozawa.
Interaction
of
surfactants
with
vesicle
membrane
of

dipalmitoylphosphatidylcholine.
Effect
on
gel-‐to-‐liquid-‐crystalline
phase
transition
of
lipid

bilayer.
Chemistry
and
Physics
of
Lipids.
42
(4):
261-‐270,
1986.

104

117.

Mason,
R.P.,
M.F.
Walter,
P.E.
Mason.
Effect
of
oxidative
stress
on
membrane
structure:
small-‐
angle
X-‐ray
diffraction
analysis.
Free
Radical
Biology
and
Medicine.
23
(3):
419-‐425,
1997.

118.

Wong-‐Ekkabut,
J.,
Z.
Xu,
W.
Triampo,
et
al.
Effect
of
lipid
peroxidation
on
the
properties
of
lipid

bilayers:
a
molecular
dynamics
study.
Biophysical
Journal.
93
(12):
4225-‐4236,
2007.

119.

Slater,
J.L.,
C.-‐H.
Huang.
Interdigitated
bilayer
membranes.
Progress
in
Lipid
Research.
27
(4):

325-‐359,
1988.

120.

Paula,
S.,
A.
Volkov,
A.
Van
Hoek,
et
al.
Permeation
of
protons,
potassium
ions,
and
small
polar

molecules
through
phospholipid
bilayers
as
a
function
of
membrane
thickness.
Biophysical

Journal.
70
(1):
339,
1996.

121.

Zeng,
J.,
K.E.
Smith,
P.
Chong.
Effects
of
alcohol-‐induced
lipid
interdigitation
on
proton

permeability
in
L-‐alpha-‐dipalmitoylphosphatidylcholine
vesicles.
Biophysical
Journal.
65
(4):

1404,
1993.

122.

White,
S.H.
A
study
of
lipid
bilayer
membrane
stability
using
precise
measurements
of
specific

capacitance.
Biophysical
Journal.
10
(12):
1127,
1970.

123.

White,
S.H.,
T.
Thompson.
Capacitance,
area,
and
thickness
variations
in
thin
lipid
films.

Biochimica
et
Biophysica
Acta
-‐
Biomembranes.
323
(1):
7-‐22,
1973.

124.

Fettiplace,
R.,
D.
Andrews,
D.
Haydon.
The
thickness,
composition
and
structure
of
some
lipid

bilayers
and
natural
membranes.
Journal
of
Membrane
Biology.
5
(3):
277-‐296,
1971.

125.

Lu,
B.,
G.
Kocharyan,
J.J.
Schmidt.
Lipid
bilayer
arrays:
Cyclically
formed
and
measured.

Biotechnology
Journal.
9
(3):
446-‐451,
2014.

126.

Miller,
C.
Ion
channel
reconstitution:
Springer
Science
&
Business
Media,
1986

127.

Khandelia,
H.,
O.G.
Mouritsen.
Lipid
gymnastics:
evidence
of
complete
acyl
chain
reversal
in

oxidized
phospholipids
from
molecular
simulations.
Biophysical
Journal.
96
(7):
2734-‐2743,

2009.

128.

Hladky,
S.,
D.
Gruen.
Thickness
fluctuations
in
black
lipid
membranes.
Biophysical
Journal.
38

(3):
251,
1982.

129.

Simons,
K.,
E.
Ikonen.
Functional
rafts
in
cell
membranes.
Nature.
387
(6633):
569-‐572,
1997.

130.

Munro,
S.
Lipid
rafts:
elusive
or
illusive?
Cell.
115
(4):
377-‐388,
2003.

131.

Simons,
K.,
M.J.
Gerl.
Revitalizing
membrane
rafts:
new
tools
and
insights.
Nature
Reviews
-‐

Molecular
Cell
Biology.
11
(10):
688-‐699,
2010.

132.

Veatch,
S.L.,
S.L.
Keller.
Separation
of
liquid
phases
in
giant
vesicles
of
ternary
mixtures
of

phospholipids
and
cholesterol.
Biophysical
Journal.
85
(5):
3074-‐3083,
2003.

133.

Danielsen,
E.M.,
B.
Van
Deurs.
A
transferrin-‐like
GPI-‐linked
iron-‐binding
protein
in
detergent-‐
insoluble
noncaveolar
microdomains
at
the
apical
surface
of
fetal
intestinal
epithelial
cells.
The

Journal
of
Cell
Biology.
131
(4):
939-‐950,
1995.

134.

Danielsen,
E.M.
Involvement
of
detergent-‐insoluble
complexes
in
the
intracellular
transport
of

intestinal
brush
border
enzymes.
Biochemistry.
34
(5):
1596-‐1605,
1995.

135.

Brown,
D.A.,
J.K.
Rose.
Sorting
of
GPI-‐anchored
proteins
to
glycolipid-‐enriched
membrane

subdomains
during
transport
to
the
apical
cell
surface.
Cell.
68
(3):
533-‐544,
1992.

136.

Skibbens,
J.E.,
M.G.
Roth,
K.S.
Matlin.
Differential
extractability
of
influenza
virus
hemagglutinin

during
intracellular
transport
in
polarized
epithelial
cells
and
nonpolar
fibroblasts.
The
Journal

of
Cell
Biology.
108
(3):
821-‐832,
1989.

137.

Sargiacomo,
M.,
M.
Sudol,
Z.
Tang,
et
al.
Signal
transducing
molecules
and
glycosyl-‐
phosphatidylinositol-‐linked
proteins
form
a
caveolin-‐rich
insoluble
complex
in
MDCK
cells.
The

Journal
of
Cell
Biology.
122
(4):
789-‐807,
1993.

138.

Simons,
K.,
D.
Toomre.
Lipid
rafts
and
signal
transduction.
Nature
Reviews
-‐
Molecular
Cell

Biology.
1
(1):
31-‐39,
2000.

139.

Anderson,
R.
Caveolae:
where
incoming
and
outgoing
messengers
meet.
Proceedings
of
the

National
Academy
of
Sciences.
90
(23):
10909-‐10913,
1993.

140.

Dietrich,
C.,
L.
Bagatolli,
Z.
Volovyk,
et
al.
Lipid
rafts
reconstituted
in
model
membranes.

Biophysical
Journal.
80
(3):
1417-‐1428,
2001.

141.

Samsonov,
A.V.,
I.
Mihalyov,
F.S.
Cohen.
Characterization
of
cholesterol-‐sphingomyelin
domains

and
their
dynamics
in
bilayer
membranes.
Biophysical
Journal.
81
(3):
1486-‐1500,
2001.

142.

London,
E.,
D.A.
Brown,
X.
Xu.
Fluorescence
quenching
assay
of
sphingolipid/phospholipid

phase
separation
in
model
membranes.
Methods
in
Enzymology.
312:
272-‐290,
1999.

105

143.

Baumgart,
T.,
G.
Hunt,
E.R.
Farkas,
et
al.
Fluorescence
probe
partitioning
between
Lo/Ld
phases

in
lipid
membranes.
Biochimica
et
Biophysica
Acta
-‐
Biomembranes.
1768
(9):
2182-‐2194,
2007.

144.

Baumgart,
T.,
S.T.
Hess,
W.W.
Webb.
Imaging
coexisting
fluid
domains
in
biomembrane
models

coupling
curvature
and
line
tension.
Nature.
425
(6960):
821-‐824,
2003.

145.

Feigenson,
G.W.,
J.T.
Buboltz.
Ternary
phase
diagram
of
dipalmitoyl-‐PC/dilauroyl-‐
PC/cholesterol:
nanoscopic
domain
formation
driven
by
cholesterol.
Biophysical
Journal.
80
(6):

2775-‐2788,
2001.

146.

Feigenson,
G.W.
Phase
diagrams
and
lipid
domains
in
multicomponent
lipid
bilayer
mixtures.

Biochimica
et
Biophysica
Acta
-‐
Biomembranes.
1788
(1):
47-‐52,
2009.

147.

Heberle,
F.A.,
J.
Wu,
S.L.
Goh,
et
al.
Comparison
of
three
ternary
lipid
bilayer
mixtures:
FRET
and

ESR
reveal
nanodomains.
Biophysical
Journal.
99
(10):
3309-‐3318,
2010.

148.

de
Almeida,
R.F.,
A.
Fedorov,
M.
Prieto.
Sphingomyelin/phosphatidylcholine/cholesterol
phase

diagram:
boundaries
and
composition
of
lipid
rafts.
Biophysical
Journal.
85
(4):
2406-‐2416,

2003.

149.

Veatch,
S.L.,
S.L.
Keller.
Seeing
spots:
complex
phase
behavior
in
simple
membranes.
Biochimica

et
Biophysica
Acta
-‐
Molecular
Cell
Research.
1746
(3):
172-‐185,
2005.

150.

Honerkamp-‐Smith,
A.R.,
P.
Cicuta,
M.D.
Collins,
et
al.
Line
tensions,
correlation
lengths,
and

critical
exponents
in
lipid
membranes
near
critical
points.
Biophysical
Journal.
95
(1):
236-‐246,

2008.

151.

Davis,
J.H.,
J.J.
Clair,
J.
Juhasz.
Phase
equilibria
in
DOPC/DPPC-‐d
62/cholesterol
mixtures.

Biophysical
Journal.
96
(2):
521-‐539,
2009.

152.

Veatch,
S.L.,
O.
Soubias,
S.L.
Keller,
et
al.
Critical
fluctuations
in
domain-‐forming
lipid
mixtures.

Proceedings
of
the
National
Academy
of
Sciences.
104
(45):
17650-‐17655,
2007.

153.

Veatch,
S.L.,
P.
Cicuta,
P.
Sengupta,
et
al.
Critical
fluctuations
in
plasma
membrane
vesicles.
ACS

Chemical
Biology.
3
(5):
287-‐293,
2008.

154.

García-‐Sáez,
A.J.,
S.
Chiantia,
P.
Schwille.
Effect
of
line
tension
on
the
lateral
organization
of
lipid

membranes.
Journal
of
Biological
Chemistry.
282
(46):
33537-‐33544,
2007.

155.

Kuzmin,
P.I.,
S.A.
Akimov,
Y.A.
Chizmadzhev,
et
al.
Line
tension
and
interaction
energies
of

membrane
rafts
calculated
from
lipid
splay
and
tilt.
Biophysical
Journal.
88
(2):
1120-‐1133,

2005.

156.

Ambati,
J.,
C.S.
Canakis,
J.W.
Miller,
et
al.
Diffusion
of
high
molecular
weight
compounds
through

sclera.
Investigative
Ophthalmology
and
Visual
Science.
41
(5):
1181-‐1185,
2000.

157.

Tsuboi,
S.,
T.
Fujimoto,
Y.
Uchihori,
et
al.
Measurement
of
retinal
permeability
to
sodium

fluorescein
in
vitro.
Investigative
ophthalmology
&
visual
science.
25
(10):
1146-‐1150,
1984.

158.

Nugent,
L.J.,
R.K.
Jain.
Extravascular
diffusion
in
normal
and
neoplastic
tissues.
Cancer
Research.

44
(1):
238-‐244,
1984.

159.

Flaten,
G.E.,
A.B.
Dhanikula,
K.
Luthman,
et
al.
Drug
permeability
across
a
phospholipid
vesicle

based
barrier:
A
novel
approach
for
studying
passive
diffusion.
European
Journal
of

Pharmaceutical
Sciences.
27
(1):
80-‐90,
2006.

160.

Perale,
G.,
F.
Rossi,
M.
Santoro,
et
al.
Drug
release
from
hydrogel:
A
new
understanding
of

transport
phenomena.
Journal
of
Biomedical
Nanotechnology.
7
(3):
476-‐481,
2011.

161.

Casalini,
T.,
M.
Salvalaglio,
G.
Perale,
et
al.
Diffusion
and
aggregation
of
sodium
fluorescein
in

aqueous
solutions.
The
Journal
of
Physical
Chemistry
B.
115
(44):
12896-‐12904,
2011.

162.

Zocher,
F.,
D.
van
der
Spoel,
P.
Pohl,
et
al.
Local
Partition
Coefficients
Govern
Solute
Permeability

of
Cholesterol-‐Containing
Membranes.
Biophysical
Journal.
105
(12):
2760-‐2770,
2013.

163.

Huang,
J.,
J.T.
Buboltz,
G.W.
Feigenson.
Maximum
solubility
of
cholesterol
in
phosphatidylcholine

and
phosphatidylethanolamine
bilayers.
Biochimica
et
Biophysica
Acta
-‐
Biomembranes.
1417

(1):
89-‐100,
1999.

164.

Widom,
B.
Surface
tension
and
molecular
correlations
near
the
critical
point.
The
Journal
of

Chemical
Physics.
43
(11):
3892-‐3897,
1965.

Asset Metadata

Creator Runas, Kristina A. (author)

Core Title Biophysical studies of passive transport across synthetic lipid bilayers

Contributor Electronically uploaded by the author (provenance)

School Andrew and Erna Viterbi School of Engineering

Degree Doctor of Philosophy

Degree Program Chemical Engineering

Publication Date 01/21/2016

Defense Date 06/19/2015

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag biomimetic membranes,giant unilamellar vesicles,lipid bilayer,lipid oxidation,membrane phase separation,microfluidics,OAI-PMH Harvest,passive diffusion,passive transport,permeability,plasma membrane,spinning disk confocal microscopy

Format application/pdf (imt)

Language English

Advisor Malmstadt, Noah (committee chair), Povinelli, Michelle L. (committee member), Sahimi, Muhammad (committee member)

Creator Email kristina.runas@gmail.com,runas@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-601663

Unique identifier UC11299452

Identifier etd-RunasKrist-3659.pdf (filename),usctheses-c3-601663 (legacy record id)

Legacy Identifier etd-RunasKrist-3659.pdf

Dmrecord 601663

Document Type Dissertation

Format application/pdf (imt)

Rights Runas, Kristina A.

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA

Abstract (if available)

Abstract Transport by passive diffusion acro ss the lipid bilayer of the plasma membrane represents a nearly universal mechanism of molecular entry of drugs and environmental toxins into the cell. Understanding the barrier properties of the lipid bilayer and what molecular characteristics control its permeability is therefore of fundamental interest to toxicology and drug development. Biomimetic membranes formed from commercially available synthetic lipids have been studied extensively as a method for mimicking the plasma membrane. Development of assays to rapidly measure bilayer permeability has been focused on optimization of three areas: maximizing high-throughput capabilities, minimizing analysis assumptions and artifacts, and maintaining a highly controllable membrane composition. This work presents a novel technique capable of all three areas of interest. The technique involves combining spinning disk confocal microscopy with a microfluidic platform to examine the permeability of synthetic lipid bilayers, which was then used to examine how varying membrane composition to reflect disease states can impact passive diffusion.