In C. elegans, assessment of ATP levels has until now been carried out in vitro [7-9,23],
a method which is accurate but implies the destruction of worms, limits
the scalability of the analysis and cannot be carried out in real-time.
We report on the construction of a luminescent transgenic C. elegans strain and have tested the hypothesis that light emitted by living C. elegans is
a reflection of its ATP pools. We used two approaches: i) exposure to
and recovery from sodium azide, a specific mitochondrial inhibitor, and
ii) RNAi towards genes that are essential for mitochondrial function.
Azide inhibits complex IV of the mitochondrial respiratory chain by
binding reversibly to cytochrome c oxidase ,
this arrests the flow of electrons and leads to a decrease in ATP
synthesis. Because the inhibition is reversible azide has been widely
used as a C. elegans anaesthetic .
Changes in bioluminescence upon exposure to, and recovery from,
sublethal concentrations of azide were consistent with reversible ATP
depletion caused by azide. Recovery in bioluminescence occurs within 30
min of removing worms from azide. Similarly, recovery of ATP depletion
resulting from anoxia occurs within 45 min of reversal of anoxia .
Luciferase expression and activity levels stayed constant in the azide
experiments and therefore did not contribute to observed changes in
The substrate luciferin has to be present intracellularly for light emission. It is not known how luciferin enters C. elegans.
It is possible that it passes through the permeabilised cuticle.
Alternatively, it may be ingested through the pharynx. If an active
pharynx was a requirement for luciferin entry, then substrate
availability could be a limiting factor for luminescence under
experimental conditions that alter the pumping rate, such as toxin
However, in this work, although exposure to azide stopped pharynx
activity, the worms were able to emit light when provided with the
substrate, indicating that luciferin was able to cross the
permeabilised cuticle. Luciferin has been deemed to be poorly taken up
by cells ,
however, empirical data suggest that luciferin uptake occurs readily.
Luciferin crossed the blood-brain barrier in mice easily, passing
through endothelial cells that are amongst the least permeable cells in
the mouse . Studies in other organisms such as Drosophila , zebra fish  and Arabidopsis 
also indicated that diffusion and permeability of luciferin did not
limit bioluminescence. Furthermore, we have observed that luciferin was
capable of crossing the shell of unhatched C. elegans embryos,
considered poorly permeable to chemicals, resulting in bioluminescence.
We have also captured images of widespread luminescence in the worm's
tissues, contrary to what would be expected if luciferin was poorly
taken up by cells.
Inhibition of respiratory chain components by RNAi provided a means
of depleting ATP levels without exogenous chemicals. The response
measured by bioluminescence is in agreement with in vitro data. In vitro ATP was reduced to 20–40% when cyc-1 or atp-3 were knocked down, and to 40–60% for cco-1 RNAi . All experimental data were consistent with the hypothesis that C. elegans luminescence
reflects its ATP pools. Additionally, the RNAi experiments illustrate
bioluminescence as a phenotype that could be the basis for genetic
Wildtype firefly luciferase is targeted to peroxisomes  and has a 3 h half-life [27,33]. In this study, the nematodes were transformed with a modified firefly luciferase gene luc+ fused to gfp,
which is not targeted to the peroxisomes. Luc+ is expressed in the
cytosol and has a half-life of 10 h in human breast cancer cells . The half-life of GFP is 26 h 
and therefore should not contribute to instability of the fusion
protein. The greater stability of Luc+ is an advantage for studies
where ATP changes may be tracked over time, as opposed to the
requirement for a short half-life when luciferase reports on gene
One critical aspect that will affect the luminescence readings is
the levels of firefly luciferase present, this may vary for example
between different strains and developmental stages. Hence, the exact
relationship between light output and ATP concentrations will depend on
the experimental conditions, ruling out precise determination of ATP
content. We propose this strain as a relative sensor of ATP levels
which can be applied to interrogate mitochondrial function and
metabolism of living worms in a non-destructive, real-time and scalable
manner. This will have broad appeal as sublethal physiological
parameters are often difficult to quantify, especially on a large
scale. Perhaps the most exciting developments will be the
identification of novel genes and pathways underlying physiological
response, as well as a better understanding of classical pathways. The
integrated luc+:: gfp fusion we described can be crossed into
available strains carrying gene deletions or into new mutants generated
by mutagenesis. Alternatively, any of the worm's genes can be targeted
for inactivation by RNAi and bioluminescence will provide an easily
quantifiable metabolic phenotype. The luciferase gene may also be
placed under promoters that will drive its expression in specific
tissues and allow for relative assessment of ATP levels in that tissue.
Ballistic transformation methods will enable expression of the
transgene in the germline where required. The transgenic strains
described here offer a unique opportunity to explore the links between
physiology and genetics of C. elegans and many other organisms with which it shares homologue genes, including humans .