ORIGINAL ARTICLE
Non-invasive in vivo optical imaging of the lacZ and
luc gene expression in mice
V Josserand
1,2,3
, I Texier-Nogues
4
, P Huber
2,5,6
, M-C Favrot
1,2
and J-L Coll
1,2
1
INSERM U823, Institut Albert Bonniot, La Tronche Cedex, France;
2
Universite´ Joseph Fourier, Grenoble Cedex, France;
3
ANIMAGE,
CERMEP, 59 boulevard Pinel, Lyon, France;
4
LETI-DTBS, CEA-Grenoble, 17 rue des Martyrs, Grenoble Cedex, France;
5
INSERM
U882, CEA-Grenoble, Grenoble, France and
6
CEA iRTSV-APV, CEA-Grenoble, Grenoble, France
The bacterial lacZ gene encoding for b-galactosidase (b-gal)
is a common reporter gene used in transgenic mice.
Nonetheless, the absence of fluorigenic substrates usable
in live animals greatly hampered the non-invasive follow-up
of this reporter gene expression. We used far-red fluores-
cence for imaging b-Gal expression in live cells in vitro or in
vivo. The 9H-(1,3-dichloro-9,9-dimethylacridin- 2-one-7-yl)
b-D-galactopyranoside substrate was used to monitor b-Gal
expression as a reporter of tumor growth, or of the
physiological levels of an endogenous gene or of gene
transfer in lung. A quantitative evaluation of this method
as well as a comparison of its sensitivity with Firefly
Luciferase-based bioluminescence was also performed.
In vivo measurements showed that 10
3
b-Gal tumor cells
located under the skin were detectable. In deeper organs like
lung, as little as 5 ng of b-Gal or Luciferase enzymes per mg
of proteins were measured, confirming that both techniques
reached similar sensibilities. Nonetheless, quantitative com-
parison of b-Gal levels measured with far-red imaging or with
a standardized enzymatic evaluation after killing revealed
that the 2D-fluorescent reflectance imaging method is
submitted to a color-dependent disparity of the organs and
cannot supply quantitative measurements but that a simple
correction can be applied.
Gene Therapy (2007) 14, 1587–1593; doi:10.1038/
sj.gt.3303028; published online 20 September 2007
Keywords: optical imaging; fluorescence; b-Gal; reporter gene; transgenic mice
Introduction
Reporter genes, such as b-Galactosidase (b-Gal), Chloram-
phenicol Acetyltransferase, Luciferase and Green Fluore-
scent Protein, are widely used in biological systems.
1
Since most of these reporters are detected in vitro
on tissue sections, biopsies or postmortem samples, these
methods are not suitable for a long-term follow-up of gene
expression and regulation.
2
Several studies recently
established that reporters such as Luciferases,
3
Thymidine
Kinases
4,5
and Green Fluorescent Protein
6
could be
detected non-invasively. Adapted imaging modalities were
developed, with variable resolutions, sensitivities and cost.
In addition, most of these methods are semi-quantitative or
qualitative only.
7
The bacterial lacZ gene, which encodes the b-Gal
enzyme, is the most commonly used reporter gene,
because of its historical availability and the existence of a
large panel of chromogenic
8,9
and fluorogenic
10–13
sub-
strates. Nonetheless, the use of these products required
the killing of the animals, until the description of a non-
invasive visualization of b-Gal expression using a
gadolinium-based probe and magnetic resonance ima-
ging.
14
Latter on, a radioiodinated competitive inhibitor
of the b-Gal enzyme as well as a b-Gal substrate with
far-red fluorogenic properties were successfully used for
in vivo imaging.
15,16
Meanwhile, innovative advances in optical imaging
technologies and especially the development of highly
sensitive photon-detection cameras, allowed cell biolo-
gists to carry out quantitative examination of whole-cell
structure and function with high spatial and temporal
resolution. The rapid adaptation of these tools as well as
the development of dedicated equipments for imaging
deep tissues in live animals is currently influencing how
researchers can approach molecular processes in vivo.
17
Light in the visible-wavelength spectrum is routinely
used for conventional and intravital microscopy.
18
Although extremely powerful, methods like confocal
and multiphoton microscopy allow imaging of super-
ficial locations down to a few hundred micrometers only,
due to optical constraints inherent to the physics of these
instruments. In addition, most of the fluorophores
currently in use are working in the 450–600 nm spectral
region. This will also limit the path of light through the
sample because biological tissues are strongly absorbing
in the visible range. Deeper events, typically ranging
from 1 to 10 millimeters under the skin surface, can be
visualized using macroscopic epi-illumination imaging
(called 2D-fluorescence reflectance imaging (FRI)). This
also require the use of far-red (600–650 nm) or near-infra-
red (650–900 nm) fluorophores to reduce tissue absorption
and to avoid most of the problems of autofluorescence
coming mainly from the constituents of the extracellular
matrix (collagens, elastin and many more) especially in
the skin.
Received 22 January 2007; revised 20 June 2007; accepted 14 August
2007; published online 20 September 2007
Correspondence: Dr J-L Coll, INSERM U823, Institut Albert
Bonniot, Grenoble, La Tronche Cedex 38706, France.
E-mail: jean-luc.coll@ujf-grenoble.fr
Gene Therapy (2007) 14, 1587–1593
& 2007 Nature Publishing Group All rights reserved 0969-7128/07 $30.00
www.nature.com/gt

In this study, we used 2D-FRI far-red fluorescence
for direct and non-invasive imaging of the b-gal repor-
ter gene expression in live mice. For this purpose,
the 9H-(1,3-dichloro-9,9-dimethylacridin-2-one-7-yl) b-D-
galactopyranoside (DDAOG)
19
was administered by an
intracardiac injection, to visualize the pattern of expres-
sion of b-Gal in the lung of mice after polyethylenimine
(PEI)-mediated non-viral delivery of a pCMV-lacZ
plasmid, or in a transgenic strain of mice expressing
physiological levels of b-Gal. A comparison of the
sensitivity and limitations of this method as compared
to bioluminescence detection of Luciferase in vivo is also
presented.
Results
Quantitative imaging of b-Gal activity in vitro
A b-Gal stable clone derived from the murine breast
carcinoma Ts/Apc cells was established. Using a
standardized chemiluminescent in vitro b-Gal assay, we
measured 250 nU cell

1
of b-Gal enzymatic activity,
corresponding to 63 mU of b-Gal produced per mg of
protein.
Imaging of increasing quantities of these cells seeded
in 24 wells with the fluorogenic DDAOG b-Gal substrate
gives a linear relation between the detected fluorescence
(relative light units per pixel (RLU per pixel)) and the
number of cells (Figure 1). As a rule, we set up the cut-off
at 4500 RLU per pixel, which represents 1.5-fold the
background level. Accordingly, as little as 2.10
4
cells per
well can be detected, and this measure was linear until at
least 2 10
5
cells per well.
Quantitative imaging of b-Gal activity produced by
tumor cells in vivo in mice
Mice bearing subcutaneous matrigel plugs containing
increasing numbers of b-Gal Ts/Apc cells were imaged
before and immediately after intracardiac injection of
DDAOG.
As indicated in Figure 2a, the background level of
fluorescence measured beside the mouse (noise)
remained constant at 1884 RLU per pixel during the
experiment. Before DDAOG injection (T¼ 0), the fluor-
escence level was 4761 RLU per pixel in the normal skin
and 5542 RLU per pixel in the plug, representing a
measure of the autofluorescence. Fifteen minutes after
DDAOG injection, the signal in the skin rose to 7230 RLU
per pixel and stabilized later on at a baseline level
around 6400 RLU per pixel. This level in the skin
represented the sum of the noise, plus autofluorescence,
plus nonspecific signal coming from a possible degrada-
tion of DDAOG in normal tissues. In the plug, fluores-
cence increased during the first 60 min after substrate
injection, reached a plateau around 13 000 RLU per pixel
between 60 and 120 min and then slowly decreased
(Figure 2a). As shown in Figure 2b, the plug imaged in
these conditions was very clearly visualized on the whole
animal.
Inoculation of increasing quantities of b-Gal Ts/Apc
cells in the plugs allowed us to demonstrate that as little
as 10
3
b-Gal cells can be specifically detected. As a
standard, we decided that the signal had to be more
elevated than 1.5 times the normal skin value before
being considered as positive. Plugs containing 100 cells
were not discernible from the normal skin (61407100 vs
5609794 RLU per pixel). Positivity was reached when as
little as 103 cells were included in the plug, with a plug/
skin ratio of 1.5, which increased to 4.6 when 106 cells
were engrafted in the plug (Figure 2c).
In vivo imaging of gene transfer in mice
Mice lung were transfected by an intravenous injection of
PEI–DNA complexes as described previously.
20
Fifty
micrograms of pCMV-lacZ or pCMV-luc plasmids was
mixed with PEI and injected intravenously. Twenty-four
hours later, the level of expression of each reporter gene
was measured using either DDAOG and fluorescence or
luciferin and bioluminescence in anesthetized mice.
As expected, both methods revealed the presence of
the reporter enzymes in the lung (Figure 3). In b-Gal-
transfected mice, fluorescence was observed in the lung
60 min after DDAOG injection in live animals or after
killing and/or extraction of the lung. No false b-Gal-
positive signal was detected in the lung of pCMV-luc-
transfected (Figures 3a and b) or of not transfected
control animals (data not shown). The stomach was
strongly positive, both before and after the injection due
to the autofluorescence of the food. The presence of b-Gal
in the lung of pCMV-lacZ-transfected animals was then
confirmed by the presence of the expected blue color in
the alveoli of X-gal-stained lung (Figure 3b).
The intensities of fluorescence or bioluminescence
emissions were measured in vivo from the lung of several
animals expressing different levels of each reporter gene
after PEI-mediated transfection (n¼ 8 and 7 per group,
respectively). The animals were then killed, the lung
extracted, chopped and the b-Gal or Luc activities
measured on the protein extracts using standard, quanti-
tative methods in vitro (Figure 3c). In both cases, very
good linear relations were observed between the in vitro
assays and the in vivo detection (R
2
¼ 0.98 for both).
The sensibility was also found to be very similar, since as
little as 5 ng of enzyme per mg of proteins were
detectable non-invasively in the lung.
In vivo imaging of physiological levels of lacZ gene
expression in mice
To demonstrate the interest of DDAOG-based fluores-
cence to follow the expression of a physiological level of
Figure 1 Fluorescence emission of b-Galactosidase Ts/Apc cells in
vitro. Measurement of the b-Gal activity was performed 60 min after
the addition of the 9H-(1,3-dichloro-9,9-dimethylacridin- 2-one-7-yl)
b-D-galactopyranoside substrate into the culture plate containing
known amount of lacZ
+
cells. Exposure time¼ 200 ms.
Reporter gene fluorescence imaging
V Josserand et al
1588
Gene Therapy