Dead mice tell too few tales. Conventional animal imaging requires sacrificing multiple animals at numerous time points, then slicing and staining tissue to identify the location and state of particular molecules at a point in time. However, it would be easier and more instructive if researchers could image the whole body to follow the biological processes within a live animal, ideally with no need for surgery or other invasive techniques. There are many established technologies for human imaging that work in animals, but advances in optical imaging in particular are providing fresh opportunities to see inside smaller creatures.

By combining modalities, researchers can pinpoint where processes are taking place. Credit: CALIPER

Such advances are allowing researchers to follow disease progression and drug response more precisely than they can from dissecting organs, says David Piwnica-Worms, who directs the molecular imaging centre at Washington University in Saint Louis, Missouri. “Because you can use each animal as its own control, something that would have been lost in the noise becomes very, very clear.” And whole-body imaging means not just better data, but also new types of data. For example, disease processes often start well before symptoms become evident. Last year, using new optical imaging technologies, researchers led by Stanley Prusiner at the University of California, San Francisco, detected the onset of a mouse equivalent of neurodegenerative conditions such as Creutzfeldt–Jakob disease nearly two months before behaviour was affected1. Rather than comparing brains of animals killed at various stages of disease, the researchers could watch the prion disease spread through the brains of individual animals. Researchers can also monitor other processes: T cells travelling to inflamed sites, tumours spreading or shrinking, even enzymes catalysing reactions within cells.

A diverse range of imaging techniques, or modalities, is now available (see Table). Magnetic resonance imaging (MRI), positron electron tomography (PET), computed tomography (CT) and single-photon emission computed tomography (SPECT) are all used routinely to scan patients in hospitals; smaller, less expensive versions have been produced for animal research. These techniques can penetrate deep into tissue, and sources of distortion are relatively few and largely understood. They are often used to survey whole bodies for disease and to do cross-sectional imaging, particularly for research on deep-seated organs.

Table 1 Pros and cons of imaging modalities
Full size table
David Piwnica-Worms has developed new bioluminescent techniques as well as ways to combine bioluminescence with magnetic resonance imaging. Credit: WASHINGTON UNIV., ST LOUIS

Optical techniques are used to probe more-local processes and can readily make use of multiple labels, or signal-emitting tags that are attached to a molecule of interest, such as a fluorescent dye on an antibody. The downside is that light is scattered and absorbed quickly within the body, complicating attempts to quantify signals and limiting imaging ability to a couple of centimetres or so below the skin surface. “No single modality has the lock on being the best molecular strategy for whole-animal imaging under all circumstances. Different biological queries require different strategies,” says Piwnica-Worms. “The different modalities have strengths that apply in one niche but not in another.”

As a field, whole-animal imaging has come a long way since the 1990s, when vendors began offering mouse-sized versions of their instruments to lure customers from research hospitals, and when those selling systems to read labelled gels started to adapt their systems to image mice.

Fast, cheap and multicoloured

Optical techniques have probably been the fastest growing way to image small animals. At the US National Cancer Institute's Small Animal Imaging Resource Program, twice as many funding recipients bought optical equipment as bought MRI, PET or SPECT machines for imaging small animals from 1999 to 2007. Programme director Barbara Croft says that much of this can be explained by the fact that optical instruments are perhaps a quarter of the cost of the other imaging technologies, and are also easy to use. “People who don't know anything else about small-animal imaging are more likely to buy optical equipment than anything else,” she says.

There are two main approaches to optical imaging. In fluorescent techniques, labels introduced into an animal give off light of one wavelength when excited by light of another wavelength, so light must travel into the animal and back out again, getting scattered and absorbed in both directions. Bioluminescent techniques, however, rely on chemical reactions that produce light from within the animal, so light needs to travel in only one direction. However, almost all techniques rely on transgenic cells that express the enzyme luciferase, and the light produced by bioluminescent reactions tends to be in shorter wavelengths that are more readily scattered by tissue. In a typical experiment, mice with impaired immune systems are implanted with pathogens or cancer cells that have been genetically engineered to express luciferase, then injected with the appropriate substrate, resulting in a chemical reaction that emits light.

Researchers can then create cohorts of animals or administer treatments according to how much a tumour has grown or an infection has spread. “Investigators like this because it allows you to eliminate some of the animal-to animal variation,” says Stephen McAndrew of Taconic in Cranbury, New Jersey, which recently acquired the business of transgenic animals, cancer-cell lines and bacterial lines from Caliper Life Sciences in Hopkinton, Massachusetts. This can be a powerful way to show the efficacy of certain drug compounds and is regularly used in applications for the approval of investigational new drugs, he says. “You can get a lot more data points per group of animals using these technologies. My sense is that the Food and Drug Administration is very receptive.”

Taconic offers nearly four dozen light-producing mice, and plans to greatly expand the types of light-emitting cancer lines it offers. And bioluminescence can be used to follow not just engineered cells but also the molecular processes within them by means of proteins called split reporters. These are used to ensure that luciferase fragments come together only when certain proteins interact or when a connecting peptide is dephosphorylated; they can also be used to monitor cell signalling or kinase activity within a living animal.

Like bioluminescent techniques, fluorescent techniques can be used for molecular imaging, such as tracking enzyme activity, or marking cells expressing a particular receptor. In this case, rather than genetically engineering cells to express a luciferase, researchers attach fluorescing dyes or nanoparticles to appropriate ligands in the laboratory, and the whole complex is then injected into the animal. One disadvantage of this method (as well as with similar labels in MRI, PET and SPECT) is that the imaging agents will also show up not just in the target site (for example, a tumour containing a particular surface receptor), but also in blood and in healthy organs. Such spurious signals can make it difficult to distinguish between, for instance, a highly vascularized site and the target. Moreover, strategies to avoid this problem can introduce other issues: increasing the dose of the imaging agent can be toxic, and waiting for the agent to clear the bloodstream and aggregate in the target zone decreases the time available for experiments.

Caliper Life Sciences' IVIS Spectrum performs fluorescent and bioluminescent imaging in two and three dimensions. Credit: CALIPER

Hisataka Kobayashi at the US National Cancer Institute in Bethesda, Maryland, grew frustrated with this problem of 'always on' labels when working with MRI and PET. So he decided to work on new probes, designing 'activatable fluorophores', which emit signals only at the desired target. Activatable fluorophores can be made in many ways, but the goal is the same, says Piwnica-Worms, who has developed similar technologies. “The signal is silent while the fluorophore is circulating around the body. You build up signal at the target and only at the target. That idea has been out there for a while, but putting it into whole-animal models is what's coming along now.”

One of the most common strategies for fine-tuning the signal is to combine a fluorophore with a transporter tag and a quencher molecule. The transporter means that the combined molecule will be taken up only by certain types of cells. The quencher prevents fluorescence until the fluorophore binds to a cell or interacts with intercellular proteases, allowing fluorescence to occur. Kobayashi recently published work on the fluorophore indocyanine green (or ICG, which, despite its name, gives off infrared light when it fluoresces). Although ICG has been clinically approved for years, it stops fluorescing when it conjugates (or covalently binds) to protein, making chemists reluctant to use it. Kobayashi reasoned that this silencing might be an advantage: he attached ICG to an antibody that both silenced the fluorescence and carried the fluorophore to the target of interest. Once the antibody was bound to the target and brought into the cell, ICG detached, becoming fluorescent2. The fact that this advantage was overlooked, says Kobayashi, is an example of how chemists' thinking can hold back development of activatable fluorophores; chemists work with such labels in solution and aren't trained to think about how fluorophores will act inside an animal's body, he says. “We pick up a lot of trash.”

Probes are the workhorses of the optical imaging system, and researchers such as Kobayashi and Piwnica-Worms are coming up with increasingly creative ways to get them into the right places and to detect them once they are there. Such advances in technology promise that researchers will soon be able to image more processes in more types of cells (see 'Probe progress'). Partly as a result, expectations for optical approaches have been rising steadily, says William McLaughlin of Carestream Molecular Imaging in Woodbridge, Connecticut, which produces instruments for optical imaging. “Initially, people were seeing things they hadn't seen before, so they just wanted to see them,” he recalls. Now they want to go beyond pretty pictures and actually get quantitative data, he says. Instrument-makers have to calibrate everything: from how lamps change with age, to how light criss-crosses the machine — anything that could affect the measurement.

Stephen Oldfield is the senior director of imaging marketing at Caliper, which sells the IVIS Spectrum, among other instruments. He thinks that three-dimensional (3D) imaging will be the next advance to become routine in optical imaging. “Early feedback was that 3D data just didn't smell right, but we can now show data that look like what you would expect from the anatomy,” he says. “I think that's going to become a standard. It offers more in-depth quantification of biological events, and the ability to co-register or integrate data with other clinical 3D modalities.”

Multimodalities

Bioluminescent T cells can be used to study wayward immune reactions in disorders such as severe combined immunodeficiency disease4.

“Combining modalities is certainly where people have been looking most recently,” says Croft. Companies such as Siemens in Berlin and Philips in Amsterdam now provide machines that combine CT with PET or SPECT. Researchers are also combining PET with MRI.

Bringing modalities together can be far from straightforward. The magnetic fields of MRI interfere with the radioactivity of PET and SPECT, for instance. Even though bioluminescence and fluorescence techniques both work by detecting light emitted from labels within the animal, fluorescent labels must first be activated by light and that must be accounted for. Moreover, the cameras need to be optimized for the different wavelengths emitted by the various labels.

Samuel Achilefu, a radiologist at Washington University in Saint Louis, is creating imaging agents designed for multimodal studies. He and his group have created an imaging probe3 that binds readily to 64Cu — a radionuclide used in PET — and that contains a dye that fluoresces only after part of the probe has been cleaved by the enzyme caspase-3. This approach allows researchers to use PET to collect quantitative information about a disease tissue, then use the activatable fluorescent signal to detect a biomolecular event, says Achilefu, who is working to create monomolecular imaging agents that combine optical probes with probes for MRI, CT and ultrasound.

An alternative way to combine modalities is to create a tray, or gantry, that enables an animal to be moved between instruments without repositioning it. Systems that perform both radioisotope and X-ray imaging, for example, have trays that slide sedated animals from one detector to another. Other approaches include moulds or tubes that keep the animal still and can be moved between machines depending on what data are desired.

LI-COR's Pearl Imager can be used to visualize how probes target specific tissues or receptors. Credit: LI-COR

Furthermore, time is a factor: X-rays and optical imaging can be completed in about a minute, so both can be done in one session relatively easily. An MRI scan, however, can take hours. Combining the modalities can therefore lower throughput and extend the time that animals need to be kept under anaesthesia or are subjected to radiation, potentially harming the animal or altering the disease process.

Despite these issues, combining modalities can bring levels of clarity that had not been anticipated, says McLaughlin. Just a few years ago, researchers had to overlay optical images over an outline of the animal. False-colour splotches in a mouse-shaped shadow give only a general idea of where the signals are coming from, but overlaying this information with X-rays leads to a richer interpretation of the images. “It just makes it so much easier when you've got that anatomical background to be able to see where the signals are actually coming from,” he says. McLaughlin recalls the reaction of researchers getting their first look at optical scans combined with X-rays about two years ago: “People would point at the images and say 'yes, yes, I understand'.”

Future images

New probes and technologies allow for more sophisticated, high-definition imaging in real time. But another trend needs to be taken into account: the influx of biologists into the field. “They don't know much about imaging, but they need images,” says Michael Olive, vice-president of science and technology at LI-COR Biosciences in Lincoln, Nebraska, which sells optical instruments. This is a trend that Carestream has noticed, too: in the past, the company supplied machines to laboratories with radiologists or to other imaging experts. But recently, says McLaughlin, it has started to focus on “crossing the chasm”— reaching basic-research labs in which the experience with imaging has been in microscopy rather than whole-body techniques. “We're looking at pulling some of the choices out of our system,” he says. The goal is to allow researchers to study biology without needing to know what a steradian is. “It's really moving into the masses,” McLaughlin says.