Light-powering Escherichia coli with proteorhodopsin
Jessica M. Walter*†, Derek Greenfield*‡, Carlos Bustamante*†‡§¶?, and Jan Liphardt*†‡**
Departments of *Physics,§Chemistry, and¶Molecular and Cell Biology,?Howard Hughes Medical Institute, and‡Biophysics Graduate Group, University
of California, Berkeley, CA 94720; and†Physical Biosciences Division, Lawrence Berkeley National Laboratory, Berkeley, CA 94720
Contributed by Carlos Bustamante, December 13, 2006 (sent for review December 1, 2006)
community sequencing of ocean samples. Previous studies have
but its role in powering cells and participation in ocean energy
fluxes remains unclear. Here, we show that when cellular respira-
tion is inhibited by depleting oxygen or by the respiratory poison
azide, Escherichia coli cells expressing PR become light-powered.
Illumination of these cells with light coinciding with PR’s absorp-
tion spectrum creates a proton motive force (pmf) that turns the
flagellar motor, yielding cells that swim when illuminated with
green light. By measuring the pmf of individual illuminated cells,
we quantify the coupling between light-driven and respiratory
proton currents, estimate the Michaelis–Menten constant (Km) of
PR (103photons per second/nm2), and show that light-driven
pumping by PR can fully replace respiration as a cellular energy
source in some environmental conditions. Moreover, sunlight-
illuminated PR?cells are less sensitive to azide than PR?cells,
consistent with PR?cells possessing an alternative means of
maintaining cellular pmf and, thus, viability. Proteorhodopsin
challenges by harvesting light energy.
light-driven proton pumps ? oceanic bacteria
photosynthesis (1, 2). Other light-harvesting mechanisms include
the light-driven proton pump bacteriorhodopsin, used by halobac-
teria living in salt ponds to supplement respiration (3). In 2000, a
novel light-driven proton pump, proteorhodopsin (PR), was dis-
covered (4). The world’s oceans contain an estimated 1028PR-
expressing bacteria, placing them among the most prevalent organ-
isms on Earth (5, 6). Proteorhodopsins are widely distributed and
spectrally tuned to their oceanic environments (5, 7–12). PR is
distinguished from bacteriorhodopsin by its high sequence homol-
ogy to sensory rhodopsins (postulated to have evolved from a
different origin than bacteriorhodopsin; ref. 4), its presence in
members of Eubacteria (as opposed to Archaea), and significant
out of the cell (4).
Abundant evidence exists for PR’s function as a transmem-
brane proton pump, including light-mediated transport of pro-
of protons by illuminated PR?Escherichia coli cells (4, 7, 13).
Despite the ability of PR to pump protons, green light illumi-
nation did not increase SAR11 growth rates or cell yield in cell
culture experiments (14), leaving proteorhodopsin’s in vivo
contribution unclear. Moreover, the extent to which light-
harvesting by PR can complement or replace other cellular
energy sources remains to be quantified.
of protons across the membrane, maintained under aerobic
conditions by oxidative phosphorylation. Bacteria use the pmf to
synthesize ATP, drive chemiosmotic reactions, and power the
rotary flagellar motor (15, 16) In 1974, Racker and Stoeckenius
(17) demonstrated that the pmf generated by light-driven proton
pumping by bacteriorhodopsin could be used to produce ATP.
We reasoned that measuring the pmf of illuminated PR?cells
could identify a role of PR in cellular energy fluxes. The speed
quatic ecosystems play a major role in the conversion of light
energy into chemical energy, principally via chlorophyll-based
of the E. coli flagellar motor is proportional to the pmf over a
wide range of speeds (18, 19), therefore, proton extrusion by PR
should increase the flagellar motor’s rotation rate when the cells
are illuminated. We used E. coli as the heterologous host in our
PR experiments because it is the primary model organism for
studying Gram-negative bacteria and techniques for pmf mea-
surement and PR expression in E. coli have been refined (4, 5,
7–11, 13, 18, 19).
Results and Discussion
We tracked swimming PR?E. coli in two dimensions by using
dark-field microscopy, periodically illuminating the cells with
green light at PR’s absorption maximum. We observed single
cells to characterize rapid responses of the cellular pmf to light.
No detectable increase in cell swimming velocity occurred upon
illumination with green light. We surmised that light-driven
proton pumping may benefit the cell only under certain envi-
ronmental conditions, as suggested by Giovannoni et al. (14).
To test the possibility that light-driven proton pumping is
most beneficial to aerobically grown cells when their ability to
respire is suddenly impaired, we energy-depleted the cells.
Because E. coli is difficult to energy-deplete by nutrient
limitation because of its endogenous energy stores (20, 21), we
additionally used the respiratory poison azide, which has
multiple cellular effects (22, 23) but primarily inhibits cyto-
chrome oxidase and, thus, proton extrusion by the respiratory
chain, stopping the flagellar motor (18).
Strikingly, with respiration inhibited by azide, PR?cells
responded to green light. PR?cells in 30 mM azide swam slowly
in red illumination, but they showed a marked velocity increase
with green illumination (Fig. 1b). Upon removal of the green
light, they slowed to their previous velocity. To increase the
accuracy of our flagellar rotation measurements, we subse-
quently used a tethered cell geometry (Fig. 2a), permitting
extended observation of the same bacterium in different illumi-
nation conditions. PR?cells were allowed to stick to the
coverslip via a flagellum, and we then monitored the angular
rotation rate of cells (Fig. 2a). A typical tethered cell rotated at
a mean rate of 0.2–1 Hz, depending on its length and the position
of the stuck flagellum along its body. To facilitate data inter-
pretation, we deleted the cheY gene (24, 25), yielding smooth-
swimming mutants whose flagellar motors do not reverse.
As expected, there was no effect of green light on the cell’s
rotation rate in the absence of azide (Fig. 2b). However, as we
inhibited respiration by adding azide, the cells again became
light-responsive. PR?cells sped up upon illumination with green
light [Fig. 2 a and c, see also supporting information (SI) Movie
1]. The PR?cells were converting light energy into an electro-
Author contributions: J.M.W. designed research; J.M.W. performed research; J.M.W., D.G.,
and J.L. wrote the paper; D.G. analyzed data; D.G. performed model simulations; and C.B.
and J.L. conceived the experiment.
The authors declare no conflict of interest.
Abbreviations: pmf, proton motive force; PR, proteorhodopsin.
**To whom correspondence should be addressed. E-mail: firstname.lastname@example.org.
This article contains supporting information online at www.pnas.org/cgi/content/full/
© 2007 by The National Academy of Sciences of the USA
February 13, 2007 ?
vol. 104 ?
chemical potential used to do mechanical work. PR?cells
stopped moving or slowed considerably when the green illumi-
nation light was removed. For example, at low concentrations of
azide (5–15 mM), angular velocity dropped by one-fourth when
the green light was removed. At higher azide concentrations
(80–110 mM), angular velocity dropped further (50–60%) be-
cause more of the cells’ cytochromes were bound by azide (26).
To determine the extent to which light can replace respiration
as an energy source, we varied the azide concentration. Increas-
ing the azide concentration from 0 to 110 mM caused cellular
pmf levels, and therefore average cellular angular velocity, to
decrease (SI Fig. 4a). The pmf dropped to 50% of its original
value at an azide concentration of 55 mM, in agreement with
inhibition studies of isolated bo cytochrome (26). When the cells
were illuminated with intense green light, their rotation rate was
restored to the speed of cells with an unimpaired respiratory
system. As proton extrusion by the respiratory system dropped
with increasing azide, PR provided an ever larger fraction of the
proton efflux needed to maintain the cellular pmf at the original,
no-azide value (Fig. 3a). At the highest azide concentration
studied, the average cell velocity increased 70% upon green light
illumination. Cells lacking PR exhibited no increasing angular
velocity in green light, indicating that PR expression is required.
We subsequently varied the intensity of the green light illu-
mination. The rotation rate was clearly stimulated even at the
lowest light intensity studied (?5 mW/cm2) (Fig. 3b). The rate
increased rapidly with intensity up to 10 mW/cm2(15 mM azide)
or 20 mW/cm2(60 mM azide) (Fig. 3b), where the effect
saturated. At ?50 mW/cm2, there was no detectable benefit of
increased illumination. The energy absorbed by PR through our
energy absorbed by PR from the sun at sea level, taking into
account the absorption cross-section of PR and the solar irra-
diation spectrum (detailed information is published as SI Text).
It was now essential to confirm the postulated mechanism of
action. Reduced proton pumping by the respiratory system
causes the pmf to drop. Light-based proton pumping by prote-
orhodopsin can then increase the pmf. Removal of oxygen from
the cell culture also should lead to light-responsive bacteria with
none of the possible confounding effects of azide. We needed to
reduce O2levels substantially, because at 0.3 ppm, the E. coli
respiratory system still generates ?50% of the normal pmf (27).
We gently bubbled nitrogen through the cell culture for 15 min
and then used a nitrogen-filled glovebox to prepare sealed
imaging chambers containing 2 ?l of cells. Just as we had seen
upon addition of azide, PR?cells became light-responsive upon
oxygen depletion. Illumination increased the cells’ angular ve-
locity by 45 ? 25% (Fig. 2c; P ? 0.005, one-sided t test). Addition
of the protonophore carbonyl cyanide 3-chlorophenylhydrazone
(100 ?M) to low oxygen cultures eliminated all bacterial motility,
and movement was not restored with 50 seconds of green
illumination (n ? 300), demonstrating that cells unable to
maintain a pmf cannot be revived.
To clarify the relationships between pmf, azide, and light, we
constructed a highly simplified model of E. coli membrane
fluxes. Our cells have multiple proton pumps that can contribute
to the pmf (28), including PR, the respiratory chain, and the
ATPase (Fig. 3c). Pmf is consumed by the flagellar motor and
numerous transporters. In addition, the bacterial membrane has
a basal permeability to protons (29). In a simpler system,
Rotterdam et al. (30) showed that lipid vesicles with only one
kind of ion pump reach a steady-state pmf whose magnitude is
well approximated by a simple RC circuit.
Hypothesizing that an analogous circuit might be able to
capture the functional relationship between pmf, azide, and
Absorption (Monterey PR)
40 x 40 µm
green and red bars indicating illumination wavelengths. (b) Single PR?bacterium swims faster when illuminated with green light. The cell’s position is recorded
at a rate of 0.5 Hz via constant red dark-field illumination throughout the entire track. Illumination with green light at the absorption peak is periodic (20 s on,
20 s off) and occurs only during green circles of the path. Velocity during periods of green illumination (10.6 ? 0.9 ?m/s) is 96% higher than during periods of
red illumination alone (5.4 ? 0.4 ?m/s; P ? 0.005, one-tailed t test). Respiration has been inhibited with 30 mM azide in motility buffer. (Inset) Raw tiled movie
frames of swimming cell.
Light-powering of PR?E. coli. (a) Overview of spectra and spectral overlaps. The PR absorption spectrum [redrawn from Beja (4)] is shown in black, with
Walter et al.
February 13, 2007 ?
vol. 104 ?
no. 7 ?
light, we constructed the model shown in Fig. 3d (see also SI Fig.
5) and parameterized it by fitting to the data shown in Fig. 3 a
and b and in SI Fig. 4. This model describes in vivo time-
dependent dynamics between light-driven proton pumping and
respiration and is described in detail in SI Text. Despite the
model’s simplicity, it suggests why no effect of PR on growth
rates has been reported. The model indicates that the maximum
potential PR can generate by using the free energy from photon
absorption (VPR) is similar to the potential generated by E. coli
respiration. Thus, in E. coli grown at neutral pH in rich or
minimal media, or in E. coli respiring aerobically by using
endogenous energy stores, PR cannot pump protons. Only when
the pmf falls below the maximum potential (VPR) during respi-
PR increases as the pmf falls. PR is able to maintain E. coli
cellular pmf near this maximum potential (VPR? ?0.2V) with
sufficiently bright illumination (KM? 60mW/cm2).
The cell motility studies together with the results of the pmf
modeling raised the possibility that PR could pump sufficient
protons to increase cell viability in addition to powering the
Angular velocity (Hz)
0 50 100150200250300350400
Angular velocity (Hz)
050 100150200250300350 400
0100 200300 400
Angular position (radians)
tethered cell in red light and green light (full movie is available at SI Movie 1), a plot of the angular position versus time of a cell in 60 mM azide, and a schematic
of tethered cell geometry. In red light, the cell rotationally diffuses about its attachment point. Green light (shaded area) leads to counter clockwise rotation
of the cell. (b and c) Cells without (b) and with (c) PR. Solid, dotted, and dashed lines show three representative cells per condition. Note the considerable
cell-to-cell variation in absolute rotation rates due to variation of cell length and tethering geometry. With or without azide, cells lacking PR show no response
with green light.
Flagellar response to green light measured in PR?and PR?bacteria observed at different levels of respiratory inhibition. (a) Movie frames showing a
www.pnas.org?cgi?doi?10.1073?pnas.0611035104Walter et al.
flagellar motor. To clarify a possible relationship between
viability and proton pumping by PR, we plated cell cultures after
their exposure to 30 mM azide for 30 min in sunlight. The cells
lacking retinal were most sensitive to azide; no colonies were
recovered after plating the azide exposed cells. The cells lacking
PR but having retinal were slightly more azide resistant; 1% of
cells survived (SI Table 1). The cells with both PR and retinal
were significantly more azide resistant than in all three other
conditions (11% of cells survived; P ? 0.005, Mann–Whitney U
test). Consistent with the motility studies and the pmf model, PR
is able to sustain cellular pmf at a level that increases viability.
These findings directly illustrate the survival benefit of PR-based
proton pumping under natural illumination conditions.
Our data demonstrate that during respiratory challenges,
light-driven proton pumping by PR can augment the cellular pmf
to the extent that it powers cell motility and increases cell
oxygen depletion via light harvesting. Conservation of the basic
features of energy metabolism in proteobacteria such as SAR86
and SAR11 makes it likely that PR expression will confer similar
benefits to other members of this class and, perhaps, even to
for a variety of purposes via PR expression combined with
modulation of the cell’s natural pmf-generating mechanisms.
Materials and Methods
Cell Culture. We expressed the SAR86 ?-proteobacterial PR-
variant (Geneart, Toronto, ON, Canada) in E. coli cells (RP437
DE3 ?cheY CmR) by using a T7-based expression system
(pET200, KanR; Invitrogen, Carlsbad, CA). Cells were grown in
T broth (1% tryptone/0.5% NaCl) supplemented with kanamy-
1 mM isopropyl ?-D-thiogalactoside and the medium was sup-
plemented with 10 ?M ethanolic all-trans-retinal. Cells were
collected in late log phase by gentle centrifugation (4,500 ? g for
5min) and carefully resuspended in motility medium (1 mM PBS
(Ambion, Foster City, CA)/0.1 mM EDTA, pH 7.4). Throughout
the article, PR?denotes cells expressing proteorhodopsin, and
PR?denotes cells (RP437 DE3 ?cheY CmR) without the PR
plasmid. PR?cells, unless otherwise noted, also were induced
with isopropyl ?-D-thiogalactoside, supplemented with all-trans-
retinal and grown in T broth with chloramphenicol (25 ?g/ml).
Unless otherwise noted, all experiments were done in glucose-
free motility medium at room temperature.
Instrumentation. Throughout the article, power density values for
‘‘green light’’ refer to the power density passed by a D540/25?
filter (Chroma, Rockingham, VT) and originating at a 175 W
Xenon bulb (Lambda light source; Sutter Instruments, Novato,
CA), or, for ‘‘red light,’’ to the power density passed by a
HQ620/60? filter and originating at a 100 W Quartz Halogen
bulb (Nikon). To visualize the cells, the sample chamber was
illuminated continuously with faint red light (0.09 mW/cm2) at
the tail of PR’s absorption spectrum (4) (Fig. 1a). We periodi-
mW/cm2) coinciding with the maximum of PR’s absorption
spectrum, ?525 nm (4). Cells were imaged at a framerate of 5
Green light intensity (mW/cm2)
140120 20 160 100
NADH + H+
O2 + H+
H+ Influx (sinks)
H+ Efflux (sources)
Azide concentration (mM)
Normalized velocity increase
020 80100 120
of the respiratory system. The difference between angular velocity in green light and red light (?G–?R) becomes pronounced in PR?bacteria (filled circles) as
respiration is inhibited by low oxygen or sodium azide. PR?cells (open circles) show no change between red and green illumination. To facilitate comparison
between cells, the angular velocities are normalized by each bacterium’s maximum velocity. n ? 5–14 cells per condition. Green line, fit to model described in
d. (b) The rotation speed of PR?cells depends on the intensity of green illumination. Individual PR?spinner cells were exposed to six intensities of green light.
The mean angular velocity at each intensity is plotted (n ? 5–6 cells for each intensity), normalized by the velocity at maximum illumination. Dashed lines, fits
to model described in d. (c) Overview of transmembrane fluxes and proton pumping in PR?E. coli. Sources of proton motive force include respiration and PR.
motor), and the membrane capacitance. The variable resistors RRand RPRmodel the effect of azide and light on proton extrusion by respiration and PR,
respectively. The voltmeter (top-most circuit element) measures the potential difference across the membrane (equivalent to the pmf).
Walter et al.
February 13, 2007 ?
vol. 104 ?
no. 7 ?
Hz by using an Andor (South Windsor, CT) iXon camera
mounted to a Nikon TE2000 microscope. The Nikon microscope
was modified for dark-field work by attaching a Zeiss 1.2–1.4
N.A. oil immersion dark-field condenser to the Nikon condenser
turret by using a custom adapter. Custom software written in
C?? was used to control the Andor camera, and Matlab was
used to process the images. All errors and error bars represent
standard errors of the mean.
Cell Viability Experiments. PR?and PR?cells were grown in LB
plus Kanamycin (PR?) or LB plus Chloramphenicol (PR?) to an
OD600 of 0.25–0.3, induced with 1 mM isopropyl ?-D-
thiogalactoside and supplemented with 10 ?M all-trans-retinal
(if retinal?), grown to an OD600of 0.5–0.6, spun down (4,500 ?
g for 5 min), resuspended in motility buffer (1 mM PBS/0.1 mM
EDTA) to an OD600of 0.1–0.2. One to two milliliters of each
culture was placed in a clear plastic culture tube, sodium azide
was added to a concentration of 30 mM, and tubes were placed
outside in the sun. Samples were taken before addition of azide
of each sample was plated at a dilution of 1/104(0 min) or 1/103
(30 min) onto LB with the appropriate antibiotic. Cell density at
0 min (before the addition of azide) was ?3 ? 106colonies in 50
?l, a density that matches well with the measured OD600 of
0.1–0.2. Colonies were counted manually after overnight incu-
the null hypothesis that the ability to pump protons does not
increase cell survival.
We thank Howard Berg (Harvard University) for suggesting the oxygen
depletion experiment, Ed Delong (Massachusetts Institute of Technol-
ogy) for encouragement and feedback, Sydney Kustu (University of
California, Berkeley, CA) for her keen microbiological eye, and Andy
Spakowitz (Stanford University). We thank the Zusman laboratory
(University of California, Berkeley) for the gift of strain RP437. We also
thank Ann McEvoy, Adam Politzer, and Elizabeth Kremen. J.M.W. was
supported by the National Science Foundation for Graduate Research
Support. This work was supported by the University of California,
Berkeley (J.L.), the Hellman Faculty Fund (J.L.), the Sloan and Searle
foundations (J.L.), and the Department of Energy Office of Science,
Energy Biosciences Program (J.L. and C.B.).
1. Behrenfeld MJ, Falkowski PG (1997) Limnol Oceanogr 42:1–20.
2. del Giorgio PA, Duarte CM (2002) Nature 420:379–384.
3. Danon A, Stoeckenius W (1974) Proc Natl Acad Sci USA 71:1234–1238.
4. Beja O, Aravind L, Koonin EV, Suzuki MT, Hadd A, Nguyen LP, Jovanovich
SB, Gates CM, Feldman RA, Spudich JL, et al. (2000) Science 289:1902–1906.
5. Venter JC, Remington K, Heidelberg JF, Halpern AL, Rusch D, Eisen JA, Wu
D, Paulsen I, Nelson KE, Nelson W, et al. (2004) Science 304:66–74.
6. Morris RM, Rappe ´, MS, Connon SA, Vergin KL, Siebold WA, Carlson CA,
Giovannoni SJ (2002) Nature 420:806.
7. Beja O, Spudich EN, Spudich JL, Leclerc M, DeLong EF (2001) Nature
8. Sabehi G, Beja O, Suzuki MT, Preston CM, DeLong EF (2004) Environ
9. Sabehi G, Massana R, Bielawski JP, Rosenberg M, Delong EF, Beja O (2003)
Environ Microbiol 5:842–849.
10. Man-Aharonovich D, Sabehi G, Sineshchekov OA, Spudich EN, Spudich JL,
Beja O (2004) Photochem Photobiol Sci 3:459–462.
11. de la Torre JR, Christianson LM, Beja O, Suzuki MT, Karl DM, Heidelberg
J, DeLong EF (2003) Proc Natl Acad Sci USA 100:12830–12835.
12. Sabehi G, Loy A, Jung KH, Partha R, Spudich JL, Isaacson T, Hirschberg J,
Wagner M, Beja O (2005) PLoS Biol 3:1409–1417.
13. Wang WW, Sineshchekov OA, Spudich EN, Spudich JL (2003) J Biol Chem
14. Giovannoni SJ, Bibbs L, Cho JC, Stapels MD, Desiderio R, Vergin KL, Rappe
MS, Laney S, Wilhelm LJ, Tripp HJ, et al. (2005) Nature 438:82–85.
15. Macnab RM (2005) in Escherichia coli and Salmonella: Cellular and Molec-
ular Biology, ed Neidhardt FC (Am Soc Microbiol, Washington, DC), 2nd Ed,
16. Schuster SC, Khan S (1994) Annu Rev Biophys Biomol Struct 23:509–539.
17. Racker E, Stoeckenius W (1974) J Biol Chem 249:662–663.
18. Gabel CV, Berg HC (2003) Proc Natl Acad Sci USA 100:8748–8751.
19. Fung DC, Berg HC (1995) Nature 375:809–812.
20. Berg HC, Tedesco PM (1975) Proc Natl Acad Sci USA 72:3235–3239.
21. Adler J, Templeton B (1967) J Gen Microbiol 46:175–184.
22. Noumi T, Maeda M, Futai M (1987) FEBS Lett 213:381–384.
23. Dioumaev AK, Wang JM, Balint Z, Varo G, Lanyi JK (2003) Biochemistry
24. Datsenko KA, Wanner BL (2000) Proc Natl Acad Sci USA 97:6640–6645.
25. Parkinson JS, Houts SE (1982) J Bacteriol 151:106–113.
26. Kita K, Konishi K, Anraku Y (1984) J Biol Chem 259:3375–3381.
27. Setty OH, Hendler RW, Shrager RI (1983) Biophys J 43:371–381.
Molecular Biology, ed Neidhardt FC (Am Soc Microbiol, Washington, DC), 2nd
Ed, Chap 19.
29. Maloney PC (1979) J Bacteriol 140:197–205.
30. van Rotterdam BJ, Crielaard W, van Stokkum IH, Hellingwerf KJ, Westerhoff
HV (2002) FEBS Lett 510:105–107.
31. Conover WJ (1998) Practical Nonparametric Statistics (Wiley, New York),
www.pnas.org?cgi?doi?10.1073?pnas.0611035104Walter et al.