In vivo optical imaging Reporters

There is a plethora of choice in reporter tags for non-invasive in vivo optical imaging that are currently available from various commercial suppliers. This generates tremendous flexibility on one hand, but dilemma of choice on the other. This manuscript reviews the determining factors and will facilitate decision making using an interactive flow chart.

Optical Imaging Modalities: Bioluminescence versus Fluorescence

In vivo optical imaging is a molecular imaging modality. It is important to distinguish the two major modalities of optical imaging: Bioluminescence versus Fluorescence. For bioluminescence, interaction of an enzyme with a substrate results in emission of light. For fluorescence, it is an external excitation light that excites the fluorophore to a higher energy state, followed by emission of photons at a longer wavelength. Thus, for in vivo imaging of bioluminescence, one routinely injects a substrate into the animal, entailing substrate kinetics dependency. For fluorescence imaging, the instrument needs to be equipped with an excitation light source and excitation and emission filters.

Tissue Optics

Imaging in deep tissue is fairly complex due to the variable factors affecting its optical physics properties. Tissue is a turbid medium that scatters and absorbs light. The double bonds in the porphyrin ring structure of the heme molecule strongly quench light of wavelengths less than 600nm, while water absorption quenches wavelengths above 900nm. Therefore, the primary rule in the election of a reporter for deep tissue imaging is a preferential emission wavelength of >600nm and <900nm. This phenomenon is easily illustrated by pressing a flashlight against your hand. Only the deep red light passes through, bone is very transparent, but blood vessels contrast.

Fluorescence imaging as opposed to bioluminescence is significantly jeopardized by spontaneous tissue autofluorescence. Endogenous chromophores, including elastin, collagen, tryptophan, NADH, porphyrins, and flavins present in animal tissue and in particular in the skin strongly autofluoresce upon excitation in the lower wavelengths <600nm and significantly tapering off at 800nm (Troy et al., 2004). Furthermore, the alfalfa/lucerne (chlorophyll) in mouse food is known to be strongly fluorescent between 650 and 750nm. It is recommended to place the animals on an alfalfa free diet 2 weeks prior to fluorescence imaging.
All these autofluorescent proteins are non-specifically excited by the excitation light source during imaging, creating significant spontaneous autofluorescent background signal. This spontaneous autofluorescence results in a high threshold for detectability of underlying fluorescent sources in deep tissue. At 500nm, autofluorescence is 5 orders of magnitude stronger compared to the faint autoluminescence (yes, we all glow in the dark).
Based upon this fundamental difference in signal to noise ratio, bioluminescence reporters remain to date more sensitive than fluorescence reporters in deep tissue imaging applications. Indeed, for fluorescence the emission intensity is directly proportional to excitation light intensity, however the autofluorescence intensity is many orders of magnitude larger than autoluminescence.

Transgene Reporter or Exogenous Probe

The next level of categorization is whether the reporter is a transgenic - genetically encoded DNA tag or an exogenous fluorescent probe. The advantage of a genetic tag is the fact that the cell constitutively emits the same amount of light (at least if expression is driven by a constitutive promoter), independent of mitosis. On the contrary, fluorescent probes suffer from serial dilution upon cell mitosis resulting in a decreasing signal, rendering this method non-quantitative over time.
 

DNA tag

In order to have cells constitutively express a DNA tag one needs to introduce and transfect the cell with the transgene. This can be a lengthy, labor intensive process with potential harm to the cell. Traditional lipid transfection methods suffer from low efficiency and poor stable integration. Viral transduction methods, especially the use of lentiviral particles have however greatly facilitated the process and success rate (Kim et al., 2004). Primary cells can be transduced at >95% expressing the reporter gene without the need of cell division. Both retroviruses and lenti’s have RT in the viral particles. However, in contrast to oncoretroviruses, lenti’s can productively infect non-dividing cells because the preintegration complex is actively transported through the nucleopore.

Luciferases

Highest sensitivity for deep tissue non-invasive optical imaging is still obtained with firefly luciferase constructs such as luc2/pGL4 from Promega. Recently, Dr Rabinovich et al. also published visualization of fewer than 10 T-cells with an enhanced firefly luciferase (Rabinovich et al., 2008). Modified firefly luciferases express in the cytoplasm as opposed to native peroxisomal expression. The half life of F-luc is short: 3 hrs, allowing for real time imaging as opposed to e.g. GFP which is robust and has a half life of 48hrs. D-Luciferin is the substrate for Firefly luciferase and Click beetle luciferase. It is a small molecule which freely diffuses across membranes and the BBB. It is not metabolized and excreted via the kidneys. It is routinely inject intraperitoneally with a 2hr excretion profile. Signal peaks around 15 min post injection followed by a plateau phase of 15-20min after which the signal steadily drops off. Alternative routes of administration are intravenous tail vein or retroorbital injection, subcutaneous injection in the scruff of the neck, inhalation, or osmotic pumps. Bacterial luciferase synthesizes its own bacterial luciferin and therefore a substrate does not need to be administered exogenously. Renilla and Gaussia luciferase utilize Coelenterazine as a substrate. Different synthetic derivatives are being evaluated for stability, wavelength and solubility. Due to their flash kinetics as opposed to the glow kinetics of firefly luciferase, intravenous administration of coelenterazine is recommended.
Besides constitutive expression of DNA tag reporters, elegant inducible reporters have been published to assess molecular events (Lyons et al., 2005). Simple constructs utilize a promoter from the gene of interest, driving expression of luciferase. E.g. for angiogenesis -VEGFR2; for inflammation - NFkB-RE; for hypoxia – HIF1alphaRE; etc. More complex models utilize split luciferases to assess protein-protein interactions or fusion proteins to e.g. evaluate MAP kinase activity.
Of course one can and must question whether the random integration of the transgene affects phenotype of the cell, might affect cell metabolism, is immunogenic, etc. These are all valid concerns which need to be carefully controlled for during validation of novel model development. E.g. For firefly luciferase, this molecule is highly identical to endogenous enzymes of mammalian cells and is an oxygen scavenger (originally expressed peroxisomally). Thanks to the similarity to endogenous molecules firefly luciferase is poorly immunogenic. However, a tendency to more aggressive behavior (due to the redox properties) of F-luc transfected cell lines vs. parentals has been observed. On the other hand GFP and some novel RFP’s appear to possess a potential toxicity to the cells, slowing tumor growth at high expression levels. Therefore, it is important to compare the phenotype of transfected clones to parental lines prior to in vivo grafting.

Red shifted proteins

Only recently, red shifted proteins that emit just above 600nm have been cloned. E.g. TdTomato, mCherry, mPlum, Katushka, mKate2. Even though tissue transparency is much better at 600nm, we are still facing high autofluorescence, and lower quantum yield of the longer wavelength, lower energy proteins. Ideally, one would have a fluorescent protein that emits at 800nm, where tissue transparency is great and autofluorescence is low. There is only a slim chance to find one of these proteins in nature since the near infrared wavelengths are outside the visible spectrum and would therefore be non-functional in nature. However, major efforts are undertaken in several laboratories around the world generating synthetic derivative proteins emitting in the far red. Dr. Roger Tsien who won the Nobel prize for his discovery of GFP, recently engineered an infrared fluorescent protein emitting at 708nm. This IFP is based upon the bacteriophytochrome from Deinococcus radiodurans, and incorporates biliverdin as the chromophore (Shu et al., 2009).

With DNA tags, one obviously has to be able to tag the cells prior to grafting. Not all small animal preclinical models are however based on grafts. Spontaneous tumor models are commonly used in both rats and mice. These can be chemically or genetically induced or spontaneous mutations. Elegant models of reporter mice have been designed that can be crossbred to existing non-reporting models. Per example, TET-ON/TET-OFF, Cre-lox, E2F1, TRAMP, all these are available in luciferase reporter settings. Cross breeding, strain differences, lag time of tumor development are a few of the reasons these are time consumptive solutions. In that case it may be advantageous to pursue fluorescent probes.
 

Fluorescent Probes

Fluorescent Probes - can be categorized in five major groups:

1) Targeting Probes - a target specific moiety (e.g. antibody, small molecule, peptide, siRNA or biomarker) directed against disease specific extracellular receptors is attached to a fluorescent dye molecule. Examples are BoneTag (Ca++ chelating), OsteoSense (hydroxyapatite), IntegriSense (αvβ3), RGDProbe (αvβ3), EGFProbe (epidermal growth factor), 2-DG Probe (2-deoxy-D-glucose). 2) Activatable Probes - generate high levels of fluorescence through enzyme-mediated (extracellular proteases) release of their fluorochrome. Examples are: ProSense (cathepsin) and MMPSense (metalloproteinase). 3) Cell Stains – lipophylic dyes which intercalate in the cell membrane such as CellVue and DiR. Quantum dots are also used to label cells. 4) Vascular Probes – are used to monitor and assess vascular perfusion and leakage or blood pooling in tumors and inflammation. These can be fluorescent dyes bound to e.g. albumin or transferrin, or can be fluorescent particles in a variety of sizes: PEGylated qdots, microspheres or nanoparticles. 5) Labeling Probes – fluorescent dyes, qdots, nanoparticles or self illuminating qdots that can be chemically linked to an antibody, peptide, siRNA, microspheres, liposome, etc. allowing non-invasive imaging of biodistribution of these molecules. Per example, a good start is 50ug of labeled antibody per animal with a degree of labeling of 3 dye molecules per antibody.

Specificity and sensitivity will be the determining factors of success of fluorescent probes. Factors to experimentally control for are: free dye versus bound dye, dye metabolism and clearance, dye-moiety stability, probe-target bonding, backbone quenching, dye kinetics, lipophilicity etc. will determine and affect probe functionality. It is important to assess biodistribution and kinetics of free dye. A blank dye control group is of fundamental essence. On the other hand, the in vivo environment - pH, the protein environment, and the binding state of the conjugated label - can change the spectral and decay characteristics of fluorescent labels. These variables need to be carefully controlled for during model validation. Strategic positioning of tumor implantation can significantly determine success of a study. Keep in mind that free dye will be metabolized in e.g. the liver and eliminated via the kidneys and bladder or intestines. Lymphatic drainage to the spleen has also been observed. Know your dye, know your disease model. Stay remote from sites of non-specific dye accumulation. Lipophilicity will determine if the probe can pass the cell membrane or the blood brain barrier. If one wants to assess cell biodistribution in a short time frame with non-dividing cells, labeling the cells with dye or quantum dots may be faster than DNA tag transfections.

A couple of notes on specific fluorescent labeling agents: Quantum dots have the great advantage of a large stokes shift, facilitating spectral unmixing from autofluorescence and/or other reporters. However, in circulation they tend to be scavenged by the RES unless PEGylated. Self-illuminated quantum dots rely on the BRET interaction of Renilla luciferase exciting the qdot – enhancing the signal intensity thus sensitivity. ICG – indocyanine green emits at 800nm and is the only dye approved by the FDA.
Keep in mind that excitation and emission peaks of fluorophores defined in vitro, may shift in vivo due to the fact that tissue transparency changes at different wavelengths. It is adviced to optimize and determine filter settings by running a sequential combination of excitation and emission filters. As a rule of thumb, due to the inherent absorption and scattering of light, a bandwidth separation of 50nm between multiple reporters in one animal is recommended; unless reporters can be discriminated alternatively, e.g. by sequential substrate injection.

Instrumentation

Instrumentation is a determining factor in reporter choice. What excitation and emission wavelengths are available? What is the quantum efficiency of the CCD for a particular wavelength? What level of sensitivity can be detected with the instrument – picomolar levels of fluorescent dye? – Single cells? Is the CCD super cooled (-90^oC) to maximally suppress dark current/noise allowing for sensitivity to detect minute amounts of bioluminescent light? Recently, it was reported that a single cell could be detected subcutaneously in a live animal with an optimized firefly luciferase reporter.
If the reporter serves dual functionality for e.g. in vivo non-invasive imaging as well as in vitro microscopy and flow cytometry, compatibility with all equipment needs to be assessed. Alternatively, fusion reporter genes for multimodality imaging can be considered.
 

Conclusion

In conclusion, many factors determine reporter choice. Last but not least, time to development, cost and availability of product at in vivo imaging quantities are major factors to be considered. The following link utilizes a flow chart wizard to help determine your best fit optical reporter. New reporters are being developed at an exponential rate opening opportunities to translational medicine and applicability in the clinic.