**4. Bioluminescent imaging (BLI) and sensitivity**

As outlined in the previous section, cells and animal models can be engineered to express luciferase in a variety of ways. However, as implied in the examples above these cells must be exposed to a substrate in order for BL to be produced. Thus although BLI is non-invasive, in vivo luminescence is generated only following intraperitoneal (IP), subcutaneous (SQ), intratumoral (IT), oral (PO) or intravenous (IV) injections of the substrate luciferin. Following injection, up to 15 minutes are required for sufficient distribution of the substrate to sites where luciferase-expressing cells are located and to achieve optimal signal intensity prior to imaging. This step represents the most significant limitation of BLI in models of cancer.

It should also be noted that BL is a weak phenomenon producing low-intensity light that cannot be observed using conventional cameras. Therefore specially designed low light imaging cameras are required. Several commercially available systems are capable of detecting such low levels of light and are listed in Table 3. In general, the components of each system include a light-tight imaging chamber and a super-cooled charged coupled device (CCD) camera. Detected photons emitted from within the body of the animal are converted to a pseudo-colour image representing light intensity (from blue for least intense to red for most intense) and superimposed over the grayscale digital image as shown in Figure 3A. The spatial resolution of BLI, however, is relatively low (1–2 mm), when compared to CT, PET and SPECT. It has been suggested that the poor resolution of BLI is the result of scattering and diffraction of light due to changes in the refractive index at cell membranes and organelles. However, because BLI is associated with little to no background noise, the anatomical origin of photons can be determined, to approximately 1 mm (Edinger,

Zhang 2010(Zhang, Triplett

et al. 2010)

Hypoxic stress Goldman 2011(Goldman,

Angiogenesis Angst 2011(Angst, Chen et

al. 2010)

Hamner et al. 2007)

Chen et al. 2011)

and St-Pierre 2011)

Pagel et al. 2011)

Spontaneous mammary

(p38DN/AP-1-Luc) Cell Cycle and apoptosis Dickinson 2011(Dickson,

Transcriptase (hTERT-Luc) Telomerase activity Jia 2011(Jia, Wang et al. 2011)

(MMP-9-Luc) Invasion and metastasis Biron-Pains 2011(Biron-Pain

(Egr-1-Luc) Growth factor signaling Dussman 2011(Dussmann,

Table 2. Luciferase based transgenic mouse models with applications in preclinical cancer

As outlined in the previous section, cells and animal models can be engineered to express luciferase in a variety of ways. However, as implied in the examples above these cells must be exposed to a substrate in order for BL to be produced. Thus although BLI is non-invasive, in vivo luminescence is generated only following intraperitoneal (IP), subcutaneous (SQ), intratumoral (IT), oral (PO) or intravenous (IV) injections of the substrate luciferin. Following injection, up to 15 minutes are required for sufficient distribution of the substrate to sites where luciferase-expressing cells are located and to achieve optimal signal intensity prior to imaging. This step represents the most significant limitation of BLI in models of

It should also be noted that BL is a weak phenomenon producing low-intensity light that cannot be observed using conventional cameras. Therefore specially designed low light imaging cameras are required. Several commercially available systems are capable of detecting such low levels of light and are listed in Table 3. In general, the components of each system include a light-tight imaging chamber and a super-cooled charged coupled device (CCD) camera. Detected photons emitted from within the body of the animal are converted to a pseudo-colour image representing light intensity (from blue for least intense to red for most intense) and superimposed over the grayscale digital image as shown in Figure 3A. The spatial resolution of BLI, however, is relatively low (1–2 mm), when compared to CT, PET and SPECT. It has been suggested that the poor resolution of BLI is the result of scattering and diffraction of light due to changes in the refractive index at cell membranes and organelles. However, because BLI is associated with little to no background noise, the anatomical origin of photons can be determined, to approximately 1 mm (Edinger,

**Transgene Research focus Reference** 

tumors

**4. Bioluminescent imaging (BLI) and sensitivity** 

A triple transgenic strain (MMTV-Cre, CAG-betageo-

A double transgenic strain

HIF-1α oxygen-dependent degradation domain (ODD-

Human Reverse Telomerase

Vascular Endothelial Growth Factor Receptor (VEGFR2- Luc or VEGFR2-Luc)

Matrix metalloproteinase 9

Early growth response 1

tTA, TetO-Luc )

Luc)

research

cancer.

Cao et al. 2002). Using topographical scanning it is now possible to construct a three dimensional image of the animal at the same time BL data is being collected. This combined imaging may help to provide better resolution and signal localization. The IVIS 200 system is able to create a 3 dimensional image of the animal where a scanning laser positioned in the horizontal plane is used to make a measurement of surface topography as shown in Figure 3B. This image is converted into a digital reconstruction of the animal that can be superimposed onto an animal atlas and used to localize the depth of signal as seen in Figure 3C. Newer BL imaging systems such as the Spectrum CT by Caliper are able to simultaneously create a Computed Tomography scan for the purpose of constructing a three dimensional image of the animal (Figure 3D). Multi-modal imaging can aid in localization, and an assessment of the depth of signal within body cavities with more precision than BLI alone.


Table 3. Commercially available BL imaging systems (Baert 2008)

Fig. 3. 2D and 3D detection of F-Luc labeled cells using the IVIS system from Caliper Life Sciences. A grayscale digital image is taken at the same time as photon capture, subsequently BL is superimposed onto the digital image (A). Surface topography is constructed using horizontal laser scanning (B), and used to create 3D rendering of the animal which can be overlain on an organ atlas (C). Figure C shows red dots where photon intensity is highest, indicating the depth and localization of the signal. Newer systems incorporate a CT scan with BL (D) in order to gain higher precision in signal localization. (Reproduced by permission from Caliper Life Sciences, Hopkinton, MA, USA.)

Bioluminescence Applications in Preclinical Oncology Research 149

cancers or gene expression studies, the detection of smaller signals will depend on the presence of larger signals located close by as signal intensity from one region can attenuate

Three main aspects of cell physiology have been shown to affect BLI. First, Czupryna et al suggested that F-Luc activity can be substantially altered in studies where reactive oxygen species are elevated (Czupryna and Tsourkas 2011). This poses a very relevant problem in studies of tumor biology as oxidative stress can occur within a tumor or as a result of therapy. Second, hypoxic regions within tumors may also affect signal intensity. BL is dependent on oxygen and a number of studies have found that the amount of light emitted from luciferase-labeled cells is reduced as the oxygen concentration decreases (Cecic, Chan et al. 2007) (Moriyama, Niedre et al. 2008). Lastly, the expression level of ABC transporters can affect BLI intensity. Huang et al did a comparative study looking at the effects of different ATP-binding cassette (ABC) transporters on BLI readout when Click Beetle, Firefly, Renilla or Gaussia substrates were used in vitro. They show that ABCG2/BCRP is able to pump D-luciferin out of cells. Some groups have begun looking into increasing the stability of luciferin in vivo. For example, to improve the stability of and provide a continuous and prolonged delivery of the substrate D-luciferin for BLI, Kheirolomoom et al created a liposomal formulation of luciferin which had a prolonged release over 24 hours compared

In oncology research, BL was first used in vitro and in vivo to assess cellular metabolism and indirectly to measure viability in an experiment known as the ATP assay. In this assay, luciferin and luciferase are added to media and in the presence of ATP, luciferin is oxidized giving off light that can be measured using a luminometer or a CCD camera. These assays are a direct measure of cell viability and can be used to assess cell proliferation and cytotoxicity in vitro (Garewal, Ahmann et al. 1986; Kuzmits, Aiginger et al. 1986; Kuzmits, Rumpold et al. 1986; Ahmann, Garewal et al. 1987; Sevin, Peng et al. 1988; Petru, Sevin et al. 1990; Crouch, Kozlowski et al. 1993). Using the ATP assay Garewal et al were able to distinguish between cytostatic and cytotoxic effects of therapeutics on a colon cancer cell

In the late 80s and early 90's Mueller-Klieser and Walenta et al began mapping metabolites in excised tumor tissues using similar principles to the ATP assay. Their method allowed for the assessment of glucose, lactate, and ATP distributions in sections of tumors and normal tissue from cryobiopsies. Briefly, the procedure involves application of a solution containing gelatin and enzymes linked to luciferase. Upon tissue thawing, the enzyme solution diffuses into the tissue section initiating the BL reaction wherever the substrate of interest is found. The photons emitted are visualized in tumor sections using a CCD equipped microscope (Mueller-Klieser, Walenta et al. 1988; Mueller-Klieser, Kroeger et al. 1991; Walenta, Schroeder et al. 2002). Figure 4 shows a bright-field image (A) and BLI of ATP (B) from a cryosection of melanoma from the Syrian golden hamster. The BLI clearly indicates high concentrations of ATP in viable cell regions of the periphery. Furthermore, the studies show that high levels of glucose are found in the tumor periphery while necrotic tumor centers

less intense signals from other regions.

to the free form (Kheirolomoom, Kruse et al. 2010).

**5. Use of BLI in oncology research** 

line in vitro (Garewal, Ahmann et al. 1986).

exhibited high lactate levels (Walenta, Dellian et al. 1992).

Although some of the limitations of this imaging method are summarized above, these limitations are offset by the fact that BLI can detect low levels of gene expression as well as small numbers of cells in animal tumor models. This method can capture minimal disease and/or micro-metastases at distant sites before the appearance of palpable nodules or clinical symptoms. Indeed, signals can be readily visualized immediately following inoculation of tumor cells (Kalra, Anantha et al. 2011). To illustrate the sensitivity of BLI, Lipshutz et al used adenoviral vector carrying the luciferase gene to deliver luciferase to day 15 fetal mice through intraperitoneal delivery. This group was able to image a single luciferase positive liver cell in one million using BLI (Lipshutz, Titre et al. 2003). Sweeney et al were able to use BLI to detect as few as 100-2500 luciferase positive cervical cancer cells in the peritoneal cavity (Sweeney, Mailander et al. 1999). It is also worth noting that in addition to excellent sensitivity, BLI is amenable to medium throughput screening where 6 animals can be imaged at once. Once optimized, the image acquisition times can be less than 30 seconds, and the image acquisition and follow-up analysis are user-friendly.

There are additional advantages associated with BLI in comparison to other imaging methods. The equipment and reagent costs tend to be relatively low compared to PET and/or MRI. BLI is also a non-radioactive imaging modality, in contrast to other modalities such as PET and SPECT. Furthermore, BL light is emitted directly by the specimen without the need to add excitation light, which is required fluorescence imaging. Also photobleaching and phototoxic effects are not a concern. Mammalian tissues and chemical agents such as chemotherapeutics do not normally emit BL, but can generate autofluorescence that can interfere with fluorescent based imaging methods. BLI has low background and any photon emission would be the result of engineered BL cells or genes. Finally, luciferase and its substrate, luciferin, are not toxic to mammalian cells, and no functional differences have been reported between cells expressing luciferase when compared with parental cell lines (Sweeney, Mailander et al. 1999; Edinger, Cao et al. 2002; Tiffen, Bailey et al. 2010; Kalra, Anantha et al. 2011).

When designing animal studies that employ BLI, it is important to consider some factors that can influence the detection of BL photons; factors that make this approach to imaging semi-quantitative at best. These factors include 1) the distance signals must travel through tissues, 2) the nature of overlying structures and 3) cell physiology. For example, luciferase positive cells located deep within the body will appear less bright than an equivalent number of cells located near the surface of the skin (El-Deiry, Sigman et al. 2006; O'Neill, Lyons et al. 2010); as tissues overlying the target cells can attenuate photon emission. Melanin is a pigment that is meant to scatter light for the purpose of protection against harmful radiation, and by a similar mechanism will attenuate light that arises from within the animal. Thus skin, fur and hair color may interfere with BL output and influence sensitivity of imaging. Studies have shown that the light emission from dark-colored mice such as the Rag2M strain, is significantly reduced when compared with white or hairless mice. For this reason, albino nude animals are often used in BL studies (Edinger, Cao et al. 2002). Curtis et al showed that even local depilation can cause pigment changes which interfere with BLI (Curtis, Calabro et al. 2010). Hemoglobin is another pigment that quenches light, thus highly vascularized organs tend to have lower levels of photon emission compared with less vascularized tissues. Finally, in the context of metastatic

Although some of the limitations of this imaging method are summarized above, these limitations are offset by the fact that BLI can detect low levels of gene expression as well as small numbers of cells in animal tumor models. This method can capture minimal disease and/or micro-metastases at distant sites before the appearance of palpable nodules or clinical symptoms. Indeed, signals can be readily visualized immediately following inoculation of tumor cells (Kalra, Anantha et al. 2011). To illustrate the sensitivity of BLI, Lipshutz et al used adenoviral vector carrying the luciferase gene to deliver luciferase to day 15 fetal mice through intraperitoneal delivery. This group was able to image a single luciferase positive liver cell in one million using BLI (Lipshutz, Titre et al. 2003). Sweeney et al were able to use BLI to detect as few as 100-2500 luciferase positive cervical cancer cells in the peritoneal cavity (Sweeney, Mailander et al. 1999). It is also worth noting that in addition to excellent sensitivity, BLI is amenable to medium throughput screening where 6 animals can be imaged at once. Once optimized, the image acquisition times can be less than 30

There are additional advantages associated with BLI in comparison to other imaging methods. The equipment and reagent costs tend to be relatively low compared to PET and/or MRI. BLI is also a non-radioactive imaging modality, in contrast to other modalities such as PET and SPECT. Furthermore, BL light is emitted directly by the specimen without the need to add excitation light, which is required fluorescence imaging. Also photobleaching and phototoxic effects are not a concern. Mammalian tissues and chemical agents such as chemotherapeutics do not normally emit BL, but can generate autofluorescence that can interfere with fluorescent based imaging methods. BLI has low background and any photon emission would be the result of engineered BL cells or genes. Finally, luciferase and its substrate, luciferin, are not toxic to mammalian cells, and no functional differences have been reported between cells expressing luciferase when compared with parental cell lines (Sweeney, Mailander et al. 1999; Edinger, Cao et al. 2002;

When designing animal studies that employ BLI, it is important to consider some factors that can influence the detection of BL photons; factors that make this approach to imaging semi-quantitative at best. These factors include 1) the distance signals must travel through tissues, 2) the nature of overlying structures and 3) cell physiology. For example, luciferase positive cells located deep within the body will appear less bright than an equivalent number of cells located near the surface of the skin (El-Deiry, Sigman et al. 2006; O'Neill, Lyons et al. 2010); as tissues overlying the target cells can attenuate photon emission. Melanin is a pigment that is meant to scatter light for the purpose of protection against harmful radiation, and by a similar mechanism will attenuate light that arises from within the animal. Thus skin, fur and hair color may interfere with BL output and influence sensitivity of imaging. Studies have shown that the light emission from dark-colored mice such as the Rag2M strain, is significantly reduced when compared with white or hairless mice. For this reason, albino nude animals are often used in BL studies (Edinger, Cao et al. 2002). Curtis et al showed that even local depilation can cause pigment changes which interfere with BLI (Curtis, Calabro et al. 2010). Hemoglobin is another pigment that quenches light, thus highly vascularized organs tend to have lower levels of photon emission compared with less vascularized tissues. Finally, in the context of metastatic

seconds, and the image acquisition and follow-up analysis are user-friendly.

Tiffen, Bailey et al. 2010; Kalra, Anantha et al. 2011).

cancers or gene expression studies, the detection of smaller signals will depend on the presence of larger signals located close by as signal intensity from one region can attenuate less intense signals from other regions.

Three main aspects of cell physiology have been shown to affect BLI. First, Czupryna et al suggested that F-Luc activity can be substantially altered in studies where reactive oxygen species are elevated (Czupryna and Tsourkas 2011). This poses a very relevant problem in studies of tumor biology as oxidative stress can occur within a tumor or as a result of therapy. Second, hypoxic regions within tumors may also affect signal intensity. BL is dependent on oxygen and a number of studies have found that the amount of light emitted from luciferase-labeled cells is reduced as the oxygen concentration decreases (Cecic, Chan et al. 2007) (Moriyama, Niedre et al. 2008). Lastly, the expression level of ABC transporters can affect BLI intensity. Huang et al did a comparative study looking at the effects of different ATP-binding cassette (ABC) transporters on BLI readout when Click Beetle, Firefly, Renilla or Gaussia substrates were used in vitro. They show that ABCG2/BCRP is able to pump D-luciferin out of cells. Some groups have begun looking into increasing the stability of luciferin in vivo. For example, to improve the stability of and provide a continuous and prolonged delivery of the substrate D-luciferin for BLI, Kheirolomoom et al created a liposomal formulation of luciferin which had a prolonged release over 24 hours compared to the free form (Kheirolomoom, Kruse et al. 2010).
