S
1
Supporting Information
fo
r
Cohesive Living Bacterial Films with Tunable Mechanical
Properties from Cell Surface Protein Display
Hanwei Liu
†,
Priya K. Chittur
†, Julia A. Kornfield, David A. Tirrell*
Division of Chemistry and Chemical Engineering, California Institute of Technology,
Pasadena, California 91125, United States
† These authors contributed equally:
Hanwei
Liu
,
Priya Chittur
*Corresponding author. Email:
tirrell@caltech.edu
This PDF file includes:
Figs. S
1
to S
19
Tables S
1
to S
3
Notes S
1
to S
6
References
S
2
Supplementary Note 1:
Reagents and s
uppliers
Restriction enzymes, ligase, and Q5 DNA polymerase were purchased from New England Biolabs
(Beverly, MA). DNA oligos and G
-
blocks were purchased from Integrated DNA Technologies
(Coralville, IA).
Supplementary Note 2:
Plasmid subcloning
Recombinant fusion proteins were produced by standard recombinant DNA technology.
E. coli
strain DH10B was
used for all cloning steps and material preparation.
Genes encoding the autotransporter protein along with elastin
solubility/stability tags have been
previously cloned by our group into modified pQE
-
80L plasmids.
1
All plasmids used in this work
were derived from
the
pAT
-
ST plasmid
1
, which encodes a SpyTag peptide at
the
N
-
terminus fused
with
the
transmembrane domain
of EhaA
at
the
C
-
terminus. In such construct
s
, the N
-
terminal
peptide will be displayed at
the
cell surface and
the
C
-
terminal domain
is
inserted in
to
the outer
membrane of
E coli
.
Plasmid
pX
-
E6
1
,
which encodes a 150
-
amino acid
elastin
-
like
-
protein, was digested
with BamHI
and XhoI and inserted into a similarly digested pAT
-
ST vector to
yield
pAT
-
E6. The T5 promoter
of
pAT
-
E6
drives constitutive expression of protein E6
-
AT.
pAT
-
ST
1
was mutated to
encode
a cysteine
residue
after
the
6xHis tag and before
the
SpyTag
peptide.
The resulting plasmid was designated
pAT
-
Cys
-
ST
.
pX
-
E6
,
encoding
six 25
-
residue
elastin
-
like
repeats
flanked by
5’ BamHI and 3’ XhoI
sites
,
was digested
with the
corresponding
enzymes and inserted into a
pAT
-
Cys
-
ST plasmid
digested with 5’ BamHI and 3’ XhoI
to
yield
pAT
-
CE6. The T5 promoter of the plasmid drives constitutive expression of protein CE6
-
AT.
S
equence
s
of all constructs were confirmed by Laragen Inc
(Culver City
, CA)
.
Supplementary Note 3: Buffer recipe
The HEPES buffer
used
in this work contains 20 mM HEPES, 115 mM NaCl, and 1.2 mM MgCl
2
buffered at pH 7.0
2
.
The PBS buffer used in this work was purchased from ThermoFisher Scientific
and
contain
s
155
mM NaCl, 1 mM KH
2
PO
4
and 3 mM Na
2
HPO
4
,
buffered at pH 7.4.
S
3
Table S1: Plasmids used in this study
Name
Backbone/origin/promoter
Purpose
pQE
-
Empty
pQE80l/colE1/T5
Empty plasmid for cloning and maintaining
ampicillin resistance
pAT
-
E6
pQE80l/colE1/T5
Constitutive expression of E6
-
AT protein on
cell surface
pAT
-
CE6
pQE80l/colE1/T5
Constitutive expression of CE6
-
AT protein on
cell surface
pX
-
E6
1
pQE80l/colE1/T5
Cloning of pAT
-
E6 and pAT
-
CE6
pAT
-
ST
1
pQE80l/colE1/T5
Cloning of pAT
-
E6 and pAT
-
CE6
pAT
-
Cys
-
ST
pQE80l/colE1/T5
Cloning of
pAT
-
CE6
pKPY680
1
pBAD33/p15a/pJ23100
Constitutive expression of mWasabi
pKPY681
1
pBAD33/p15a/pJ23100
Constitutive expression of mCherry
S
4
Table S2: Protein sequences
Protein:
Sequence
(N
-
terminal
amino acid
first)
E6
-
AT
MKYLLPTAAAGLLLLAAQPA
MAMRGS
HHHHHH
GSVD
VPGA
GVPGAGVPGEGVPGAGVPGAGVPGAGVPGAGVPGEGVPGAG
VPGAGVPGAGVPGAGVPGEGVPGAGVPGAGLDVPGAGVPGA
GVPGEGVPGAGVPGAGVPGAGVPGAGVPGEGVPGAGVPGAG
VPGAGVPGAGVPGEGVPGAGVPGAG
LE
TPTPGPDLNVDNDLR
PEAGSYIANLAAANTMFTTRLHERLGNTYYTDMVTGEQKQTT
MWMRHEGGHNKWRDGSGQLKTQSNRYVLQLGGDVAQWSQ
NGSDRWHVGVMAGYGNSDSKTISSRTGYRAKASVNGYSTGL
YATWYADDESRNGAYLDSWAQYSWFDNTVKGDDLQSESYK
SKGFTASLEAGYKHKLAEFNGSQGTRNEWYVQPQAQVTWMG
VKADKHRESNGTLVHSNGDGNVQTRLGVKTWLKSHHKM
DD
GKSREFQPFVEVNWLHNSKDFSTSMDGVSVTQDGARNIAEIKT
GVEGQLNANLNVWGNVGVQVADRGYNDTSAMVGIKWQF
CE6
-
AT
MKYLLPTAAAGLLLLAAQPA
MAMRGS
HHHHHH
C
GSVD
VPG
AGVPGAGVPGEGVPGAGVPGAGVPGAGVPGAGVPGEGVPGA
GVPGAGVPGAGVPGAGVPGEGVPGAGVPGAGLDVPGAGVPG
AGVPGEGVPGAGVPGAGVPGAGVPGAGVPGEGVPGAGVPGA
GVPGAGVPGAGVPGEGVPGAGVPGAG
LE
TPTPGPDLNVDNDL
RPEAGSYIANLAAANTMFTTRLHERLGNTYYTDMVTGEQKQT
TMWMRHEGGHNKWRDGSGQLKTQSNRYVLQLGGDVAQWS
QNGSDRWHVGVMAGYGNSDSKTISSRTGYRAKASVNGYSTG
LYATWYADDESRNGAYLDSWAQYSWFDNTVKGDDLQSESY
KSKGFTASLEAGYKHKLAEFNGSQGTRNEWYVQPQAQVTWM
GVKADKHRESNGTLVHSNGDGNVQTRLGVKTWLKSHHKM
D
DGKSREFQPFVEVNWLHNSKDFSTSMDGVSVTQDGARNIAEI
KTGVEGQLNANLNVWGNVGVQV
ADRGYNDTSAMVGIKWQF
Highlight
Color Reference
(starting from N
-
terminus)
:
Red: PelB lea
d
er peptide
Yellow: 6x His tag
Magenta: Cysteine
Blue: E6
Green: EhaA
autotransporter
S
5
Figure S1. Expression construct for surface displayed adhesive proteins.
Figure S
2
.
A
ntibody staining of E6
-
AT films
.
a
, Microtome sections of E6
-
AT films were stained with anti
-
His tag antibody
conjugated with Dylight 488. Expression of E6
-
AT protein across
the full
thickness
is apparent
. Scale bar, 100 μm.
b
, A
n
E6
-
AT
film was engineered to express mCherry and stained
with
anti
-
His tag antibody conjugated with Dylight 488.
The
mCherry channel
show
s
cell pack
ing
in the bacterial film
; the
Dylight 488 channel show
s
expression of E6
-
AT at
the
cell surface. Scale bar, 2 μm.
S
6
Figure S
3
. Flow cytometry of
cells derived from
E6
-
AT films
. Flow cytometry enables relative quantification of protein
expression.
C
ontrol and E6
-
AT films were sampled and stained with anti
-
His tag antibody conjugated to Dylight 488 and
analyzed
by
flow cytomet
ry
. X
-
axis values correspond to
antibody
labeling intensity
;
Y
-
axis values correspond to percentage of population.
E6
-
AT
exhibits
stronger labeling
than
the control. CE6
-
AT
could not
be analyzed
by
flow cytometry
because cells could not be
adequately dispersed
.
Figure S
4
. Erosion assay results for control, CE6
-
AT and E6
-
AT films.
OD
600
of PBS buffer was measured after 1 h and 24 h
of
rocking (15° angle, 15 cycles per minute)
.
S
7
F
igure
S
5
. Schematic of
reservoirs used to drive the
bulge test.
(A)
The ports to the
reservoirs shown in Figure 2A are connected
to syringes
that serve as fluid reservoirs
.
(B)
As fluid
is added to Reservoir 2, the hydrostatic pressure below the bacterial film
increases
. (C) The pressure difference across the bacterial film
causes it to bulge upward
through
the
central aperture
(illustration
in gray rectangle, right)
.
In cyclic loading experiments, the pressure difference is reduced by adding fluid to Reservoir 1 until the
fluid levels are matched, then repeating the cycle by adding fluid t
o Reservoir 2.
In the present experiments,
6
0
-
m
L
syringes were
used
(5.61
cm
2
internal cross section)
,
such that adding 858 μ
L
of PBS increases the reservoir level by 1.53 mm, which increases
the hydrostatic pressure on the corresponding face of the sample by 15.0 Pa
.
S
8
Figure S6. The only CE6
-
AT film that exhibited failure.
This CE6
-
AT film was accidentally soaked in PBS for 4 h prior to
testing. During testing, when the nominal stress reached approximately 4500
Pa, failure occurred abruptly.
Regions of
(A)
the
stress
-
strain curve correspond to physical features observable in
(B) OCT cross
-
section images of
the biofilm:
the
short region of
decreasing slope
of stress vs strain from 0.15 to 0.18
(A, orange
rectangle
) corresponds to
a transition in film
shape from
a
spherical
cap to
having stronger curvature near the apex
(B, top), followed by failure within seven frames
acquired at 16
ms intervals
(A
,
green rectangle, elapsed time less than 0.2
s; representative images are in the middle row of B). The bottom row of
OCT images
were acquired
after rupture, when strain could no longer be measured and the pressure difference was lost as fluid from Reservoir
2 flowed through the gaps in the film
). Scale
bar
500 μm.
Unlike E6
-
AT films, which yield initially and then fail,
the one
CE6
-
AT
film that failed
exhibit
ed
brittle failure
: a sharp crack rapidly propagated
around the bulge
.
S
9
Figure S
7
.
Water content of
E6
-
AT and CE6
-
AT films.
Both
films contain roughly
75% water
by mass, similar to the water
content of
E. coli
cell
s
.
3
Number of replicates: 3.
Figure S
8
. Colony forming units
(CFU) per unit mass
of E6
-
AT and CE6
-
AT films.
Both films
contain approximately
3*10
7
CFU
per mg of film. Number of replicates: 3.
S
10
Supplementary Note 4: Calculation of CE6
-
AT
copy
-
number per cell using TCEP
r
eduction
Using Beer’s law,
4
we measured the absorbance versus concentration for Dylight 633
-
maleimide
at 633 nm
to prepare a
calibration curve.
A =
ε
lc
A is absorbance;
ε
is molar extinction coefficient of the molecule; l is light path length; c is
concentration of the molecule.
Figure S
9
. Calibration curve for Dylight 633
-
maleimide at 633 nm.
The calibration curve yields a
molar extinction coefficient
for
Dylight 633 maleimide
of
0.1627
μM
-
1
cm
-
1
. We assume
that
the difference
in labeling intensity
Δ
between CE6
-
AT TCEP + and
E6
-
AT TCEP +
films
is a result of
reduction and labeling of
CE6
-
AT by Dylight 633 maleimide.
The calibration curve allows determination of
the concentration of CE6
-
AT protein.
A
ssuming the
mass of one
E. coli
cell to be 1 pg (10
9
cells per mg of bacterial film)
the
number of
CE6
-
AT
proteins per cell
3,5,6
is given by:
# of proteins per cell = (
Δ
/
ε
l
)(Dilution
factor
)/10
9
This method yields a value of
2.5
x
10
5
proteins per cell
.
S
11
Supplementary Note
5: Calculation of CE6
-
AT and E6
-
AT
cop
ies
per cell using
q
uantitative
w
estern
b
lot
ting
CE6
-
AT and E6
-
AT films of known mass
were
lysed in 4% SDS
,
1xPBS pH 7.4 at 100 ºC for 30
min on a thermo shaker (VWR Scientific) at 900 rpm. E6
-
AT protein
was
expressed in BL21 strain
and
purified under denaturing conditions (8 M urea)
on
an Anti
-
His tag resin (Qiagen) and eluted
with 20 mM imidazole. A BCA assay kit (Thermo Fisher) was used to measure the concentration
of purified E6
-
AT protein in 8 M urea, Tris buffer
,
pH 8.0. The concentration of protein was
measured to be 0.76 mg/m
L
.
Buffer exchange for
denatured, purified E6
-
AT
in
to a 10 mM
ammonium acetate solution was accomplished with Amicon Ultra diafiltration units (3 kDa
MWCO) by repeat centrifugation and wash steps. Th
e
protein solution was mixed 1:1 with super
-
DHB matrix
which consists of a 9:1 (w/w) mixture of 2,5
-
dihydroxybenzoic acid and 2
-
hydroxy
-
5
-
methoxybenzoic acid
and analyzed by MALDI
-
TOF
; the molecular weight was determined
to
be
54182
.
This result suggested
that
the pelB leader peptide
was
not cleaved
from the protein,
consistent
with the observation that
the
expression host
(
BL21
)
cells
did not
aggregate
in
planktonic culture
when induced with 0.1 mM IPTG
.
We then
subjected purified
E6
-
AT protein
solutions of
known concentration
to SDS PAGE along with
lysate
s
of E6
-
AT and CE6
-
AT films.
The gel was transferred to
an
iBlot protein transfer apparatus (Invitrogen) and
the membrane
was
blocked with 5% milk in 0.1% tween
-
20 in 1x PBS for 1.5 h. Dylight 650
-
labeled
anti
-
6x His tag
antibody at a concentration of 0.1 μg/m
L
was used to stain the blo
t, which
was imaged
on a
Typhoon Gel Scanner (General Electric).
The mass difference caused by
cleavage of the
pelB
leader peptide
in the
cell lysate
s
is not clear
ly
resolved on
the
blot
.
The image was analyzed by
ImageJ software with
the
E6
-
AT
protein used
for calibration. The calibration curve
i
s shown in
Figure S1
1
b
.
The i
ntensit
ies
of
the
bands for
the
E6
-
AT and CE6
-
AT lysate
s
w
ere
also quantified
by ImageJ
;
comparison with
the calibration curve
allows the
protein
copy
-
number per cell
to
be
calculated. For CE6
-
AT, we
found
2.5
x
10
5
± 2.6
x
10
4
protein
copies
per cell;
for
E6
-
AT 2.6
x
10
5
±
1.3
x
10
4
protein
copies
per cell.
S
12
Figure S1
0
. Quantitative
w
estern blot analysis. a,
Western
b
lot of 3 replicates of E6
-
AT
films
, 3 replicates of CE6
-
AT films and
calibration loading of purified E6
-
AT
at
known concentration
s
.
b
, Calibration curve
for
E6
-
AT protein.
b
, M
ALDI
-
TOF m
ass
spectr
um
of purified E6
-
AT.
d
, Number of proteins per cell estimat
ed
for CE6
-
AT and E6
-
AT
. Both CE6
-
AT and E6
-
AT are
expressed at a level of 2.5
x
10
5
proteins per cell. Number of replicates: 3.
S
13
Figure S1
1
. Stress v
s strain curves for first 6 loading and unloading cycles of E6
-
AT and CE6
-
AT films in oscillatory bulge
experiments.
CE6 films displayed an elastic response over multiple cycles. E6 films appeared to have progressively less energy
lost as hysteresis as the cycles progressed, and an initial plastic deformation in the first cycle.
Figure S1
2
. Sample holder for
in situ
tracking of bacterial film healing.
A
gar plate
s
that held healing bacterial films were
placed in the circular
indent
. The marks around the circular holder
were aligned to similar marks on the petri dishes holding the
bacterial films, and
we
re used to
en
sure
that
the agar plate
was
placed in
approximately
the same orientation during each imaging
session.
S
14
Figure S
1
3
.
OCT scans of CE6
-
AT control on 2YT plate
. Top row
–
Faint, regular patterning
is
due to software, not biofilm.
Scale: top, 4 x 4 mm scan box; bottom, 1 x 1 x 1 mm scan box. Insets are OCT camera images, manually cropped to region being
scanned. Red rectangular outline represents scan box (automatic, from ThorImage OCT software).
Table S3. Sample statistics for healed biofilms
Total
number of
samples
Successfully
peeled from agar
and loaded
(% of total)
Survived initial
filling/pressure
equilibration
(% of total)
Failed within
imposed pressures
during bulge test
(% of
tested
)
Original
(Day 0)
4
4 (100)
4 (100)
0 (0)
Controls
(16 hours)
4
4 (100)
4 (100)
2 (50)
Healed
(6 hours)
4
3 (75)
0 (0)
-
Healed
(12 hours)
6
5 (83)
4 (66)
4 (100)
Healed
(16 hours)
7
6 (86)
3 (43)
3 (100)
Healed
(24 hours)
10
8 (80)
0 (0)
-
S
15
Figure S
1
4
.
OCT scans of CE6
-
AT films with defect and healed.
Original, day 7 CE6
-
AT films (A) do not fail within the limits
of our test. Pictured maximum stress tested: 6.
8
kPa. Cut films (B) immediately after injury cannot be tested, as the fluid freely
flows through the tear (right). *50% of control films failed; pictured film (C) showed failure at stress: 4.
7
kPa. All healed films
failed (D); pictured maximum stress tolerated 3.3 kPa. Scale 200 μm.
Figure S
1
5
.
Confocal microscopy scanning of mWasabi CE6
-
AT films during healing process.
Images
recorded at 0, 3, 6, 18
,
and 24 hours. Dimensions of 3D rendering: 1000 μm x 1000 μm x 170 μm
.
S
16
Supplementary Note
6
. Failure of healed CE6
-
AT films
We
assessed the extent of recovery of mechanical properties using the 16 h mark as our time point,
comparing the original uncut film on day 0, the control (uninjured film grown on the healing plate),
and healed films (Figure 5C
-
E).
We
observe
d
a few clear differences among the films tested: first,
the original (uninjured, day 7) films did not fail at the maximum pressure imposed by our device.
The actual
maximum
stress depended on the sample thickness and properties, but the highest
observed stress tested o
n a CE6
-
AT sample of 89 μm thickness was 7.21 kPa. While there was
some variation in the range of applied stresses and observed strains during our test, we observed
that CE6
-
AT films on day 7 consistently tolerated stresses greater than 6.5 kPa (
Fig
ure
S
1
4
A). In
contrast, two out of four of the control films (16 h) failed during the experiment. One failed at a
stress of 4.
7
kPa
(
Fig
ure
S
1
4
C
)
, while the other film failed at a stress of 3.
3
kPa (
Fig
ure
S
1
4
D
).
All of the healed films that could be loaded and tested failed within the range of stresses applied
in the test. To probe the extent of healing of the defect, we used CE6
-
AT films expressing the
fluorescent protein mWasabi, and observed the healing proc
ess using confocal microscopy (which
has higher resolution than OCT). Confocal microscopy images (Figure
S
1
5
) revealed visible
defects until 16 h of healing.
Figure S1
6
.
CE6mW (CE6 mWasabi) exhibited modulus (A) and day 7 thickness (B) similar to those of CE6. Modulus: 44.0 ±
5.63 kPa (CE6), 45.6 ± 6.7 kPa (CE6mW). Thickness: 89.4 ± 5.07 μm (CE6), 93.6 ± 3.53 μm (CE6mW). Day 7 films did not fail
within the pressure range
imposed by our test.
S
17
Figure S1
7
.
E6mC (E6 mCherry) can also heal
after
injury
. OCT images taken immediately (left) and after 16 h (right) show
growth of biofilm in previously cut region. See main text for protocols. Scale
bar
100 μm.
Figure S
1
8
.
(
A) Spherical cap approximation for stress assumes that deformed film is part of a larger spherical pressure vessel and
the stress in the walls of the film balances the applied pressure. Strain is estimated as a difference in the arc length (red
arrow)
comp
ared to the original (flat) length of the biofilm.
(
B) Image processing scripts binarize and clean up OCT images and detect the
top and bottom surfaces of the film over thousands of images (bottom right).