Supporting Information
Subramanian et al. 10.1073/pnas.1718810115
SI Materials and Methods
Chemostat Growth Conditions.
S. oneidensis
MR-1 cells were
grown in continuous flow bioreactors (BioFlo 110; New Brunswick
Scientific) with an operating liquid volume of 1 L, as previ-
ously described (1
–
3). A total of 5 mL of a stationary-phase
overnight LB culture was injected into the bioreactor containing
1 L of a defined medium (3), while dissolved oxygen tension
(DOT) was maintained at 20% by adjusting the ratio of N
2
/air
mixture entering the reactor (using automatic mode). After 20 h,
and as the culture reached stationary phase, continuous flow
of the defined medium (3) was started with a dilution rate of
0.05 h
−
1
while DOT was still maintained at 20%. After 48 h of
aerobic growth under continuous flow conditions, the DOT was
manually reduced to 0% by adjusting the N
2
/air mixture entering
the reactor. O
2
served as the sole terminal electron acceptor
throughout the experiment. pH was maintained at 7.0, temper-
ature at 30 °C, and agitation at 200 rpm to minimize mechanical
shear forces [in previous studies (1, 2), chemostat cultures used
for
Shewanella
nanowire conductivity measurements were oper-
ated at 50 rpm agitation]. Forty hours after DOT reached 0%,
samples were taken from the chemostat for TEM imaging.
Batch Culture Growth Conditions.
S. oneidensis
MR-1 was grown
overnight in LB broth at 30 °C up to an OD
600
of 2.4
–
2.8. A total
of 200
μ
L of this overnight culture was added to each of two
duplicates, sealed, and autoclaved in 100-mL serum bottles
containing 60 mL of a defined medium (3). One of the two
bottles acted as a control and was not used for imaging. To the
control bottle, 5
μ
M resazurin was added to indicate the O
2
levels in the medium. The bottles were then placed in an in-
cubator at 30 °C, with shaking at 150 rpm until the color due to
resazurin in the control bottle completely faded, indicating an-
aerobic conditions. We then took 200
μ
L of sample from the
bottle that did not contain resazurin for TEM imaging.
Negative Stain TEM.
Chemostat.
Samples taken from the chemostat were immediately
fixed with 2.5% glutaraldehyde and stored at 4 °C. A total of 2
μ
L
of this fixed sample was spotted on a glow-discharged, X-thick
carbon-coated, R2/2, Au NH2 London finder Quantifoil EM
grid (Quantifoil Micro Tools) and allowed to sit for 2 min. Any
remaining liquid was blotted off gently using a kimwipe and the
grid was stained with 1% uranyl acetate for 2 min before gently
blotting the remaining stain. The grid was then allowed to dry for
1 d at room temperature before imaging in TEM.
Perfusion flow imaging platform.
The fixed grid from the perfusion
flow imaging platform was blotted to remove excess liquid and
stained with 1% uranyl acetate fo
r 2 min before gently blotting the
remainingstain.ToobtainanfLMimageforcorrelationwithEM,
the grid was rehydrated with deionized water and reimaged on the
inverted fluorescent microscop
e. The grid was then blotted again
and allowed to dry at room temperature before imaging in TEM.
Imaging of samples was performed on an FEI 120-keV electron
microscope.
ECT Data Analysis and Structure Overlay of MtrC and MtrA on EM
Map.
The IMOD software package was used for labeling of
model points and generating the 3D views presented in Figs. 6 and
7. For Fig. 6
C
–
F
(
Movie S11
and similar analyses on vesicles
from 13 other tomograms), a meshed view of the membrane
surface was generated by labeling the membrane as a separate
contour in every fifth tomographic slice. This allowed the surface
area of the meshed membrane surface to be determined in
IMOD. Red and green model points were generated by picking
all observed exterior and interior densities, respectively, in every
slice of the tomogram. Particles were labeled as cytochromes if
they were within
∼
5 nm inside or outside of the membrane.
Neighboring particles were labeled as separate points only if they
were separated by more than
∼
5 nm. Particle density was cal-
culated by dividing the number of labeled particles by the
membrane surface area. For Fig. 7
C
and
Movie S12
, the mem-
brane was modeled using a smoothed isosurface model object in
IMOD, and all observed densities on the membrane interior and
exterior were manually labeled as model points. Density coor-
dinates were extracted and used for distance calculations. The
number of model points in Fig. 7
C
is greater than the densities
observed in Fig. 7
A
since Fig. 7
C
is a 3D representation of the
densities seen in the entire tomogram whereas Fig. 7
A
is a slice
from the tomogram that does not contain all of the densities
(
Movie S12
). For Fig. 7
B
, the crystal structure of MtrC was vi-
sualized and oriented using UCSF Chimera (4). Adobe Photo-
shop was then used to overlay this crystal structure onto densities
on the EM map. The surface view of MtrA was visualized using
the PyMOL software (The PyMOL Molecular Graphics System,
Version 1.7.2.0; Schrodinger LLC) and overlaid on EM densities
using Adobe Photoshop.
Direct Tunneling Limit Calculation.
To find whether direct tunneling
is possible between the observed densities in Fig. 7
A
, we used the
available structures of MtrC (5) and MtrA (6) to calculate the
largest center-to-center intermolecular distance that would allow
direct tunneling. From the crystal structure, the dimensions of
MtrC are
∼
9
×
6
×
4 nm. Since the orientation of MtrC on the
OM is unknown, we took the largest dimension (9 nm) to calcu-
late the direct tunneling limit. The known direct tunneling limit
for the distance between the donor and acceptor redox sites in
biological ET is
∼
2nm(7).Thissetsthelimitofdirecttun-
neling between MtrCs at 11 nm (9 nm
+
2 nm) center-to-center
intermolecular distance, assuming redox sites are located at the
edges of the molecule. In the case of MtrA, the known mo-
lecular dimensions are
∼
10
×
5
×
2.5 nm. From the EM map, it
appears that MtrA is oriented with its long axis (10 nm) per-
pendicular to the outer membrane (Fig. 7
B
). Therefore, it is
likely that any intermolecula
r electron tunneling between
MtrAs occurs along the shorter axes (5 nm and 2.5 nm). Taking the
larger of these two dimensions (5 nm), we calculate the limit of
direct tunneling between MtrAs to be at 7 nm (5 nm
+
2nm)
center-to-center intermolecular distance.
Cytochrome Diffusion and Electron Hopping Rate Calculations.
Fol-
lowing the Blauch
–
Saveant (8) approach for modeling ET in an
assembly of redox carriers in two dimensions, the diffusion coef-
ficients for electron hopping between redox carriers (
D
e
) and the
physical motion of redox carriers (
D
phys
)are(8)
D
e
=
δ
2
4
t
e
,
D
phys
=
δ
2
4
t
p
,
[S1]
where
δ
is the center-to-center distance of closest approach be-
tween two redox carriers (approximate diameter of one redox
carrier), and
t
e
and
t
p
are time constants of electron hopping and
physical motion, respectively.
D
phys
has been measured for in-
tegral membrane proteins similar in size to MtrC to be
∼
3
μ
m
2
/s
(9). Assuming
δ
to be the average size of MtrC [6.33 nm (5)], we
Subramanian et al.
www.pnas.org/cgi/content/short/1718810115
1of16
calculate
t
p
to be
∼
3
×
10
−
6
s.
D
e
can be calculated using
t
e
, which
is estimated by the electron residence time in OM decaheme
cytochromes (
∼
10
−
4
s) (10, 11). In the limit where
t
p
<
t
e
, we use
the mean-field expression (Eq.
1
) for the apparent diffusion
coefficient (
D
ap
) (8). To calculate the electron flux (
J
)inanOM
extension, we assume a linear drop in the concentration of re-
duced redox carriers along the OM extension length (
L
) and that
cytochromes are completely reduced at one end and completely
oxidized at the other end of the extension:
J
=
D
ap
∂
C
∂
x
≈
D
ap
C
ð
x
=
0
Þ
−
C
ð
x
=
L
Þ
L
=
D
ap
C
0
L
.
[S2]
Here,
C
0
and
C
represent the total and reduced redox carrier
concentrations, respectively. Using Eq.
1
for
D
ap
and assuming a
cylindrical OM extension with length
L
and diameter
d
, we can
find the overall ET rate (
I
),
I
=
J
π
d
=
D
ap
C
0
L
π
d
=
C
0
L
π
d
D
phys
ð
1
−
X
Þ
f
c
+
D
e
X
,
[S3]
where the fractional loading (
X
) is defined as the ratio of cyto-
chrome density to the maximum possible density,
C
0
/
C
Max
. As-
suming spherical redox carriers,
C
Max
can be estimated by the
concentration of circles in a hexagonal lattice,
C
Hex
:
C
Max
≈
C
Hex
=
0.91
π
δ
2
2
.
[S4]
Therefore, we can replace
C
0
in Eq.
S3
by
XC
Hex
. In addition, the
correlation factor
f
c
in Eq.
1
and Eq.
S3
can be written as (8)
f
c
=
ð
2
−
X
Þ
f
c
,
X
=
1
2
f
c
,
X
=
1
+
1
−
2
f
c
,
X
=
1
X
,
[S5]
where the correlation factor at full fractional loading,
f
c
,
X
=
1
is
0.466942 for diffusion in 2D space (8).
Finally, the overall ET rate as a function of redox carrier
fractional loading can be written as
I
ð
X
Þ
=
C
Hex
L
π
dX
D
phys
ð
1
−
X
Þ
f
c
+
D
e
X
.
[S6]
Fig. 8 shows
I
(
X
) for MtrC-mediated ET in a cylindrical OM
extension with
L
=
1
μ
m and
d
=
100 nm.
1. Gorby YA, et al. (2006) Electrically conductive bacterial nanowires produced by
Shewanella oneidensis
strain MR-1 and other microorganisms.
Proc Natl Acad Sci USA
103:11358
–
11363.
2. El-Naggar MY, et al. (2010) Electrical transport along bacterial nanowires from
Shewanella oneidensis
MR-1.
Proc Natl Acad Sci USA
107:18127
–
18131.
3. Pirbadian S, et al. (2014)
Shewanella oneidensis
MR-1 nanowires are outer membrane
and periplasmic extensions of the extracellular electron transport components.
Proc
Natl Acad Sci USA
111:12883
–
12888.
4. Pettersen EF, et al. (2004) UCSF Chimera
–
A visualization system for exploratory re-
search and analysis.
J Comput Chem
25:1605
–
1612.
5. Edwards MJ, et al. (2015) Redox linked flavin sites in extracellular decaheme proteins
involved in microbe-mineral electron transfer.
Sci Rep
5:11677.
6. Firer-Sherwood MA, Ando N, Drennan CL, Elliott SJ (2011) Solution-based structural
analysis of the decaheme cytochrome, MtrA, by small-angle X-ray scattering and
analytical ultracentrifugation.
J Phys Chem B
115:11208
–
11214.
7. Gray HB, Winkler JR (2003) Electron tunneling through proteins.
Q Rev Biophys
36:
341
–
372.
8. Blauch D, Saveant J (1992) Dynamics of electron hopping in assemblies of redox
centers. Percolation and diffusion.
J Am Chem Soc
114:3323
–
3332.
9. Ramadurai S, et al. (2009) Lateral diffusion of membrane proteins.
J Am Chem Soc
131:12650
–
12656.
10. White GF, et al. (2013) Rapid electron exchange between surface-exposed bacterial
cytochromes and Fe(III) minerals.
Proc Natl Acad Sci USA
110:6346
–
6351.
11. Breuer M, Rosso KM, Blumberger J (2014) Electron flow in multiheme bacterial cy-
tochromes is a balancing act between heme electronic interaction and redox po-
tentials.
Proc Natl Acad Sci USA
111:611
–
616.
Subramanian et al.
www.pnas.org/cgi/content/short/1718810115
2of16
Fig. S1.
Summary of the sample preparation methods attempted in this work for transmission electron microscopy studies of
S. oneidensis
OM extensions.
Different experimental methods are listed in the sequence in which they were attempted (
Left
to
Right
) based on the result from the previous column.
“
Liquid
Cultures
”
(
Left
) refers to growing cells in liquid cultures using a batch culture or a chemostat, whereas
“
Solid Surface
”
(
Right
) indicates OM extensions
produced by cells on an EM grid in the perfusion flow imaging platform. Within each column, a specific set of experimental methodologies, the resultin
g
observation, and our interpretation of the results are marked by solid green, dashed green, and dashed orange arrows, respectively.
Subramanian et al.
www.pnas.org/cgi/content/short/1718810115
3of16
Fig. S2.
TEM images of negative stained, glutaraldehyde-fixed
S. oneidensis
cells from the chemostat method of growth (corresponds to column 3 in Fig. S1,
Left
). (
A
–
C
) Images show the three OM extensions detected from the thousands of cells screened, indicating that OM extensions are rare in liquid cultures. (
D
and
E
) Single vesicles, which could be precursors to OM extensions, were seen budding from the cell or broken off from the cell. (Scale bars: 200 nm.)
Subramanian et al.
www.pnas.org/cgi/content/short/1718810115
4of16
Fig. S3.
ECT images of glutaraldehyde-fixed
S. oneidensis
cells grown using the serum bottle method frequently showed membrane blebs (arrows) (corre-
sponds to column 2 in Fig. S1,
Left
). These membrane blebs, however, did not assemble into the typically long OMV chain morphology, and it was not clear
whether they were artifacts of fixation or other sample preparation steps. (Scale bars: 100 nm.)
Subramanian et al.
www.pnas.org/cgi/content/short/1718810115
5of16
Fig. S4.
TEM images of negative stained, formaldehyde-fixed
S. oneidensis
cells from the perfusion flow imaging platform (corresponds to column 2 in Fig. S1,
Right
). (
A
–
C
) Presumed disrupted OM extensions (arrows) that were not sufficiently preserved with formaldehyde fixation. (
D
and
E
) While formaldehyde
fixation did not preserve all OM extensions, some of them were preserved (arrows). (Scale bars in
A
–
E
:0.5
μ
m.)
Fig. S5.
(
A
and
B
) Two unfixed OM extensions (arrows) from the perfusion flow imaging platform that remained intact until ECT imaging. These images are
from a
S. oneidensis
Δ
crp
mutant (lacking the cAMP receptor protein). The OMV chain morphology seen here confirms that this morphology is not an artifact of
fixation (corresponds to column 1 in Fig. S1,
Right
). (Scale bars: 100 nm.)
Subramanian et al.
www.pnas.org/cgi/content/short/1718810115
6of16
Fig. S6.
Correlative fLM and negative stained, glutaraldehyde-fixed TEM images of
S. oneidensis
OM extensions from the perfusion flow imaging platform
(corresponds to column 3 in Fig. S1,
Right
).
A
and
B
are
S. oneidensis
wild type and
S. oneidensis
Δ
crp
and correspond to Movies S1 and S2, respectively. (
A
and
B
,
Left
) Overlay of the fLM and TEM images. (
A
and
B
,
Center
and
Right
) Enlarged fLM and TEM views of the areas in the white dashed boxes, respectively.
White arrows indicate the OM extensions.
Fig. S7.
Cryo-EM projection image of a smooth membrane filament of
S. oneidensis
from the perfusion flow imaging platform. Next to the filament are
flagella that have been marked for comparison.
Inset
is an enlarged view of the boxed region. (
Inset
scale bar: 500 nm.)
Subramanian et al.
www.pnas.org/cgi/content/short/1718810115
7of16
Fig. S8.
Enlarged views of constriction points from the unfixed near-native OM extension in Fig. 4
C
. Dashed black boxes highlight the presence of densities at
four different junctions that connect vesicles along the OMV chain. (Scale bars: 50 nm.)
Fig. S9.
(
A
–
E
) Branching of OMV chains from the perfusion flow imaging platform shown in (
A
and
B
) cryo-EM projection images and (
C
–
E
) ECT images. (Scale
bars in
A
and
B
, 500 nm; in
C
–
E
, 100 nm.) Arrows indicate the branching points.
A
–
D
are
S. oneidensis
wild type and
E
is
S. oneidensis
Δ
crp
mutant. Movie S13
corresponds to
D
.
Fig. S10.
(
A
–
C
) Three examples of Cryo-EM projection images of
S. oneidensis
OM extensions from the perfusion flow imaging platform showing they are
flexible, increasing the likelihood of contacting terminal EAs. Black arrows point to the OM extensions and white arrows indicate the end of the OM ex-
tensions. (Scale bars: 500 nm.)
Subramanian et al.
www.pnas.org/cgi/content/short/1718810115
8of16
Table S1. Strains used in this study
Strain
Relevant genotype
Ref.
S. oneidensis
MR-1
Wild type
(1)
S. oneidensis
,
Δ
Mtr/
Δ
mtrB
/
Δ
mtrE
Δ
mtrB
/
Δ
mtrE
/
Δ
mtrC
/
Δ
omcA
/
Δ
mtrF
/
Δ
mtrA
/
Δ
mtrD
/
Δ
dmsE
/
Δ
SO4360
/
Δ
cctA
/
Δ
recA
(2)
S. oneidensis
Δ
flg
Lacking flagellin genes SO_3237 and SO_3238
(3)
S. oneidensis
Δ
crp
Lacking the cAMP receptor protein (CRP)
(4)
1. Myers CR, Nealson KH (1988) Bacterial manganese reduction and growth with manganese oxide as the sole electron acceptor.
Science
240:1319
–
1321.
2. Coursolle D, Gralnick JA (2012) Reconstruction of extracellular respiratory pathways for iron(III) reduction in
Shewanella oneidensis
strain MR-1.
Front Microbiol
3:56.
3. Bouhenni R, et al. (2010) The role of
Shewanella oneidensis
MR-1 outer surface structures in extracellular electron transfer.
Electroanalysis
22:856
–
864.
4. Charania MA, et al. (2009) Involvement of a
membrane-bound class III adenylate cycla
se in regulation of anaerobic respiration in
Shewanella oneidensis
MR-1.
JBacteriol
191:4298
–
4306.
Movie S1.
Correlative fLM and negative stained, glutaraldehyde-fixed TEM of
S. oneidensis
OM extensions from the perfusion flow imaging platform
(corresponds to Fig. S6
A
). fLM and TEM images of a field of view with cells are shown first, followed by an overlay of the two images highlighting their
correlation. A specific cell with OM extensions is then enlarged and subsequent correlative TEM reveals the OMV chain architecture of OM extensions n
ot
visible in fLM. Red fluorescence is the result of membrane staining by FM 4-64FX.
Movie S1
Subramanian et al.
www.pnas.org/cgi/content/short/1718810115
9of16
Movie S2.
Correlative fLM and negative stained, glutaraldehyde-fixed TEM of
S. oneidensis
Δ
crp
OM extensions (corresponds to Fig. S6
B
). fLM and TEM
images of a field of view with cells are shown first, followed by an overlay of the two images highlighting their correlation. A specific cell with an OM e
xtension
is then enlarged and subsequent correlative TEM reveals the OMV chain architecture of OM extensions not visible in fLM. Red fluorescence is the result
of
membrane staining by FM 4-64FX.
Movie S2
Movie S3.
Time-lapse movie of
S. oneidensis
OM extension growth on an EM grid in the perfusion flow imaging platform (corresponds to Fig. 2).
Inset
is an
overlap of red fluorescence and reflective brightfield channels, revealing both the holey carbon film coating the EM grid and the fluorescently labe
led cells
attached to it. (Scale bar: 5
μ
m.) Red fluorescence is the result of membrane staining by FM 4-64FX. The interval between two consecutive frames is 5 min, but is
reduced to 333 ms (three frames per second). Real time is shown at the top in hours and minutes (h:m).
Movie S3
Subramanian et al.
www.pnas.org/cgi/content/short/1718810115
10 of 16
Movie S4.
The 3D reconstruction of an OM extension, a flagellum, and a pilus from
S. oneidensis
, allowing direct comparison of their sizes and morphologies
(corresponds to Fig. 4
D
). The 3D reconstruction (tomogram) is shown slice by slice.
Movie S4
Subramanian et al.
www.pnas.org/cgi/content/short/1718810115
11 of 16
Movie S5.
The 3D reconstruction of a thin OM extension and a flagellum from
S. oneidensis
(corresponds to Fig. 4
F
) allowing direct comparison of their
dimensions. The 3D reconstruction (tomogram) is shown slice by slice.
Movie S5
Subramanian et al.
www.pnas.org/cgi/content/short/1718810115
12 of 16
Movie S7.
Time-lapse movie of OM extension growth and retraction by
S. oneidensis
Δ
flg
. OMV chains grow from a cell and are later retracted to form single
large vesicles at the growth initiation points (corresponds to Fig. 5
B
). (Scale bar: 5
μ
m.) Red fluorescence is the result of membrane staining by FM 4-64FX. The
interval between two consecutive frames is 5 min, but is reduced to 333 ms (three frames per second). Real time is shown at the top in hours and minutes (h:
m).
Movie S7
Movie S6.
The 3D reconstruction of a
S. oneidensis
OM extension, highlighting the
“
junction densities
”
(corresponds to Fig. 5
A
). Arrowheads point to
constriction points along the OMV chain. The junction densities at three enlarged constriction points are circled in red. These densities are propos
ed to fa-
cilitate the constriction of the membrane on the OM extension interior. The 3D reconstruction (tomogram) is shown slice by slice.
Movie S6
Subramanian et al.
www.pnas.org/cgi/content/short/1718810115
13 of 16
Movie S9.
Time-lapse movie of
S. oneidensis
wild-type OM extension growth and transformation. An OM extension grows from a cell as an OMV chain
composed of large vesicles and is later transformed into a smoother form (corresponds to Fig. 5
C
). (Scale bar: 2
μ
m.) Red fluorescence is the result of membrane
staining by FM 4-64FX. The interval between two consecutive frames is 5 min, but is reduced to 333 ms (three frames per second). Real time is shown at the t
op
in hours and minutes (h:m).
Movie S9
Movie S8.
Time-lapse movie of OM extension growth and retraction by
S. oneidensis
wild-type. An OMV chain grows from a cell and is later retracted to form
a single large vesicle at the growth initiation point. (Scale bar: 5
μ
m.) Red fluorescence is the result of membrane staining by FM 4
–
64FX. The interval between
two consecutive frames is 5 min, but is reduced to 333 ms (3 frames per second). Real time is shown at the top in hours and minutes (h:m).
Movie S8
Subramanian et al.
www.pnas.org/cgi/content/short/1718810115
14 of 16
Movie S11.
The 3D reconstruction of the vesicle shown in Fig. 6
C
. Membrane (cyan line), interior particles (green circles), and exterior particles (red circles) are
labeled as model points. Meshed view of the membrane is generated and all of the observed interior and exterior densities are shown as model points in 3D
.
The 3D reconstruction (tomogram) is shown slice by slice. (Scale bar: 50 nm.)
Movie S11
Movie S10.
Time-lapse movie of
S. oneidensis
wild-type OM extension growth and transformation. An OM extension alternates between smooth OMV chain
and large vesicle-chain forms. (Scale bar: 5
μ
m.) Red fluorescence is the result of membrane staining by FM 4-64FX. The interval between two consecutive
frames is 5 min, but is reduced to 333 ms (three frames per second). Real time is shown at the top in hours and minutes (h:m).
Movie S10
Subramanian et al.
www.pnas.org/cgi/content/short/1718810115
15 of 16
Movie S13.
OMV chain branching demonstrated by ECT of
S. oneidensis
(corresponds to Fig. S9
D
). The 3D reconstruction (tomogram) is shown slice by slice.
Movie S13
Movie S12.
The 3D reconstruction of an unfixed OM extension from
S. oneidensis
wild type with all of the observed interior (putative MtrA) and exterior
(putative MtrC) densities marked as model points in green and red, respectively (corresponds to Fig. 7
A
). Also shown is the 3D isosurface view of the OM
extension with the putative EET proteins (corresponds to Fig. 7
C
). The 3D reconstruction (tomogram) is shown slice by slice.
Movie S12
Subramanian et al.
www.pnas.org/cgi/content/short/1718810115
16 of 16