Bioluminescence Microscopy
New Avenues in Live Cell Imaging
Dr. Christoph R. Bauer,
Bioimaging Center,
University of Geneva, Switzerland
In contrast to fluorescence methods, bioluminescence microscopy does not need excitation by
light. As photon emission results from a chemical
reaction, results are highly specific and quantifiable. Until recently bioluminescence microscopy
was difficult to approach as a result of rather dim
signal intensities. Due to better probes and especially thanks to better and more specific instrumentation this technique has now become much
more accessible and can in many situations outperform fluorescent approaches.
Bioluminescence Microscopy:
Advantages and Challenges
In bioluminescence microscopy we detect light that is produced due to a chemical reaction of an enzyme (luciferase)
with its substrate (luciferin). Similar to
the much better known fluorescence approaches, bioluminescence is a technique
that can be used for non-invasive analysis of molecular functions in living cells
and tissues (fig. 1). In contrast to fluorescence where the light to be produced is
generated by the absorption of photons,
bioluminescence does not need excitation light. A second important difference
is energy-dependency (fig. 2). Both these
characteristics give bioluminescence microscopy several conceptual advantages:
In a traditional fluorescence microscope the light path is generally more
complex, as specially designed filter sets
(plus other optical elements, see below)
are used to separate emission from excitation light. This filtering is challenging as
the excitation light is by principal much
stronger than the corresponding emission
light. This problem is non-existent in bioluminescence. A second challenge in flu-
Fig. 1: Bioluminescence time lapse microscopy of NIH3T3 fibroblasts stably expressing Bmal1-luciferase [6] (only selected images of the full time-lapse are shown). Images were taken with the LV 200
bioluminescence microscope with 30 minutes exposure times.
orescence comes from autofluorescence:
Components other than fluorochromes in
and outside of cells emit fluorescent light
in a non-specific manner. Again autofluorescence is totally absent in bioluminescence leading to extremely low levels of
background light. Although fluorescence
signals might be often stronger in absolute values (see below), the signal to noise
ratio can thus still be as good or better in
bioluminescence. Consequently bioluminescence microscopy is ideal to quantify
expression levels by observing small signal changes close to background levels.
Another advantage for bioluminescence comes from the aforementioned
energy- (ATP-) dependency: Only physiologically active and intact cells will produce light. In fluorescence dead or dying
cells do actually express a very high
autofluorescent signal.
The third important issue is phototoxicity and photobleaching. In fluorescence
microscopy photobleaching decreases
signals over time while phototoxicity
leads to damaged or dead cells due to
light-induced generation of free radicals.
Both these phenomena are again nonexistent in bioluminescence.
G.I.T. Imaging & Microscopy 4/2013, pp 32–34, WILEY-VCH Verlag GmbH & Co. KGaA, GIT VERLAG, Weinheim, Germany
What are the challenges of bioluminescence microscopy? The main challenge
was and to some extent still is the fact
that bioluminescence signals are generally much weaker than fluorescence.
Weak signals result in a) poor time resolution and/or b) poor spatial resolution.
Fortunately, optimized microscopical
setups and better probes have improved
this situation (see next chapter). Other
challenges with less impact are variability of the half-live of genetically encoded
luciferases and variability in their enzymatic activity.
Evolution of Probes and Instrumentation
As stated, a big challenge in bioluminescence microscopy is signal intensity. To
overcome this, advances have been undertaken on two fronts namely a) the
development of better probes and b) the
development of better instrumentation.
For probes efficient light emission
is obviously the most important factor.
Other factors to consider are signal stability and expression efficiency. Spectral
properties can additionally be of interest
for multi-channel approaches.
Luciferases have been isolated from
several organisms (fig. 3). Besides the
classical firefly (Photinus pyralis) luciferase, luciferase genes have been isolated from copepods (Gaussia princeps),
from renillidae (sea pansy = Renilla reniformis) [1], from oplophoridae (deep sea
shrimp = Oplophorus gracilirostris) [2]
and from Elateridae (click beetle). Compared to firefly luciferase, the luciferases
from Gaussia and Oplophorus are significantly brighter (ca. 100 fold for Gaussia,
ca. 150 fold for Oplophorus) [2,3], giving them a clear advantage. To their disadvantage is that Gaussia and Oplophorus have both their emission peaks at
470 nm (in contrast to the 560 nm mission light of firefly luciferase). This 470
nm emission wavelength could be a disadvantage for tissue imaging as light
scattering is increased towards shorter
wavelengths. Eluc is a modified firefly luciferase reported to be up to ten times
brighter than wild type luciferase [4].
Instrumentation improvements have
been realized by a) redesigning the microscope setup and b) by improved detection sensitivity. The first instruments
for bioluminescence microscopy used
traditional microscope setups. Parts of
traditional setups are extra magnification lenses, filters, mirrors and other optical elements that are built into the path
between the specimen and the camera.
Each of these elements increases the
minimal distance between specimen,
plus eats up light intensity. None of these
elements is needed for bioluminescence.
The only commercial setup, where
dedicated beam-path optimizations have
been undertaken so far is the LV (Luminoview) 200 from Olympus. In this instrument the light path containing only one
tube lens, is shortened to less than half
compared to traditional microscopes. This
beam-path improvement leads to an over
10x increase in light collection efficiency
when compared to traditional microscope
setups. Additional (comfort-) improvements come from a light-tight enclosure
that protects camera and specimen from
outside light. In traditional microscope
setups great care had to be taken to design an extremely dark room in order to
do bioluminescence microscopy.
Similarly important as the microscope
setup, is the camera. Highest quantum efficiencies are currently realized with electron multiplying CCD (EMCCD) cameras.
The advantage of this chip construction
lays in the fact that the electron multiplication is happening before the read-out.
This means that charge (photon induced
but also unwanted thermally generated
charge) from each pixel is multiplied directly on the sensor. The thermally generated electrons are responsible for dark
current noise, the most important noise in
these cameras. Fortunately this noise is
temperature dependent [5]. By cooling the
camera chip it can be kept small. In our
setup (Hamamatsu EMCCD C9100-13) we
use water- plus Peltier-cooling and work
with a camera chip cooled down to -92°C.
Application Examples
We have previously shown that cultured
cells contain autonomous and self-sustained clocks using long-time fluorescence
live cell imaging [6]. In later experiments
we wanted to test robustness of circadian
rhythms against changes in temperature
or global transcription rates. For these experiments fluorescence microscopy could
not be used: As temperature changes and
drug treatments were needed, cells did not
tolerate any additional phototoxic stress
for more than a few hours. In contrast by
using bioluminescence time-lapse microscopy of NIH3T3 fibroblasts expressing luciferase under the control of a circadian
Fig. 2: Light production scheme in bioluminescence and fluorescence microscopy.
promoter (Bmal1-luc cells [7]), circadian
gene expression could be monitored and
quantified over several days.
In a related project, bioluminescence
microscopy with neuronal precursor
cells helped to demonstrate that circadian gene expression is already apparent
during early stages of development [8].
In a very different subject bioluminescence microscopy showed to be equally
useful to precisely quantify short-time
events. In both prokaryotes and eukaryotes, transcription has been described
as being temporally discontinuous, most
genes being active mainly during short
activity windows interspersed by silent
Fig. 3: Colors of bioluminescence: NanoLuc (deep sea Oplophorus gracilirostri luciferase, Promega [2]),
Renilla (Rluc from renilla reniliformis), Eluc (Enhanced Beetle Luciferase [4]), CBG (click beetle green
luciferase [11]), SLG (green-emitting luciferases, Toyobo), Luc2 (synthetic firefly luciferase Promega), SLO
(orange-emitting luciferase [12]), CBR (red-emitting click beetle luciferase, Promega), SLR(red-emitting
beetle luciferase from Phrixothrix hirtus, Toyobo). Image kindly provided by Olympus Corporation, Tokyo.
periods. To characterize this in more
detail transcription rates needed to be
monitored at higher temporal resolution. This was done by establishing various cell lines expressing a short-lived luciferase protein from an unstable mRNA.
Using high camera binning (4x4 pixels)
and photon counting mode we could record and quantify transcription levels
for up to 72 hours with a time resolution
of five minutes. This allowed to characterize transcriptional kinetics of endogenous mammalian genes and let to the
conclusion that mammalian genes are
transcribed with widely different bursting rates having each its characteristic
kinetics signature [9].
In a recent paper we describe circadian gene expression in pancreatic human islets. For this project we needed
additional channels besides bioluminescence. To do this we added a CoolLed
light source to our bioluminescence microscope. As there is no room for dichroic
mirrors in the system, we used highly selective emission filters inserted in a filter wheel between specimen and camera.
With this setup we could do time-lapse
experiments combining bioluminescence
with fluorescence and transmission channels [10] (fig. 5).
Fig. 4: Dexamethasone-induced cytosol to nucleus translocation of NanoLuc-glucocorticoid receptor
(GR) fusion proteins measured in HeLa cells. Images were taken with the LV 200 bioluminescence
microscope with 2 seconds exposure times (reprinted with permission from [2]).
Summary and Outlook
Bioluminescence microscopy offers new
avenues in live cell imaging and can
replace fluorescence microscopy approaches if phototoxicity becomes critical. This is frequently the case in longterm recording experiments and/or if the
experimental setup implements unavoidable stress. In contrast to fluorescence
approaches, bioluminescence microscopy
does not need excitation by light (with potentially phototoxic effects). In addition,
bioluminescence results are highly specific and quantifiable. In distinction to fluorescence microscopy where autofluorescence of specimens as well as reflections
or contamination from the excitation
light can contribute to signal intensities,
bioluminescence signals correspond in a
one-to-one fashion to molecular events.
Until recently bioluminescence microscopy was a difficult task. Improvements in probes and better-designed instrumentation have made the technique
a more accessible and highly intriguing tool opening new avenues to the biomedical research community.
Fig. 5: Combined bioluminescence and fluorescence image of individual human pancreatic islets. a)
bioluminescence channel (expression of circadian luciferase reporter in islet cells), b) corresponding
transmitted light image, c) fluorescent channel image (td Tomato fluorescent protein under the control
of insulin promoter specifically expressed in human beta cells), d) overlay of bioluminescence and
fluorescence channels [10].
[1] Loening A.M. et al.: Nat Methods 7, 5–6 (2010)
[2] Hall M.P. et al.: ACS Chem Biol 7, 1848–1857
[3] Tannous B. A. et al.: Mol Ther 11, 435–443
[4] Nakajima Y. et al.: PLoS ONE 5, e10011 (2010)
[5] Dussault D. and Hoess P.: Proc. SPIE5563,
195–204 (2004)
[6] Nagoshi E. et al.: Cell 119, 693–705 (2004)
[7] Dibner C. et al.: EMBO J 28, 123–134 (2009)
[8] Kowalska E. et al.: J Biol Rhythms 25, 442–
We would like to thank U. Schibler for his
continuous support.
Please find the complete list of references online at “”.
449 (2010)
[9] Suter D.M. et al.: Science 332, 472–474 (2011)
[10]Pulimeno P. et al.: Diabetologia 56, 497–507
Dr. Christoph R. Bauer
Jerome Bosset
Bioimaging Center
University of Geneva
Geneva, Switzerland
Dr. Charna Dibner
Division of Endocrinology, Diabetes and Nutrition
University Hospital of Geneva
Geneva, Switzerland
Download PDF