Fluorescence lifetime imaging microscopy (FLIM) enables researchers in the life sciences to get information from live specimens about interactions on the molecular scale. The technique captures the differences in the excited state decay rate from a fluorescent sample, rather than relying on the concentration of a fluorophore. Since imaging does not derive from the intensity of a signal, the technique lessens the impact of photon scattering in thick layers of sample and is generally considered more robust than intensity-based methods. This can open doors for both researchers and clinicians. Breast tumor specimen with FLIM-derived heat map (left) shows regions highly suspicious for cancer in red. A white light image of the specimen (center) overlaid with the pathologist’s analysis of the corresponding histology section (right). The overlay of the pathologist’s analysis shows cancer outlined in red, adipose outlined in black and fibrous tissue outlined in aqua. Courtesy of Jakob Unger from the Marcu Lab. To gain a deeper understanding of what FLIM has to offer, we asked a panel of academic researchers and industry experts to talk about the strengths, weaknesses and future of this powerful imaging tool. FLIM has a range of applications — from tracking the effects of drugs in real time and guiding measurements during surgery to detecting energy transfer from FRET (Förster or fluorescence resonance energy transfer). Our Contributors Are: Stephen Boppart is the head of the Biophotonics Imaging Laboratory at the Beckman Institute for Advanced Science and Technology and is a professor at the University of Illinois at Urbana-Champaign. He is also a member of the BioPhotonics editorial advisory board. Johan Herz is the sales engineer at Lambert Instruments responsible for the company’s FLIM products. Gerhard Holst is the head of research at PCO AG in Kelheim, Germany. Michael Z. Lin is an assistant professor in neurobiology and bioengineering at Stanford University, where his laboratory develops technologies for optical sensing and control of biology. James Lopez is the manager of the Life Sciences Applications Group at Olympus Scientific Solutions Americas in Waltham, Mass. Sandra Orthaus-Müller is a technical sales specialist for microscopy at PicoQuant who concentrates on applying FLIM, FRET and FCS (fluorescence correlation spectroscopy) for biological applications. Jennifer Phipps is a project scientist in the biomedical engineering department at the University of California, Davis, working in the laboratory of professor Laura Marcu. Q: What applications of fluorescence lifetime imaging do you think are flourishing and why? If you are a researcher, how has FLIM particularly suited your research needs? Lopez: Most of what we see these days is one of three FLIM applications: FRET FLIM, observing changes in the intracellular environment and spectral unmixing. FRET FLIM is the change in fluorescence lifetimes due to the presence of energy transfer from FRET. In the presence of FRET, you see shortened donor molecule lifetimes as energy is donated to the acceptor. This provides a quantitative method of verifying that FRET is occurring. FRET FLIM is powerful because it provides data independent of dye or protein concentration, photobleaching and excitation light intensity — that is, laser power. Light intensity images (top row, colorized). Corresponding fluorescence lifetime images (bottom row, colorized). The average fluorescence lifetime of the cells increases over time, as shown in the graph on the right. Courtesy of Lambert Instruments. Quantitative FLIM studies are sometimes done to record changes in the intracellular environment around the fluorescent molecules of interest. If an environment becomes more acidic, or if the researcher fixes the cell, the lifetimes change. FLIM allows the scientist to watch the physiology of the cell in terms of pH or redox state changes or other perturbations of the cell. The third application is spectral or channel unmixing. We often separate fluorescence molecules based on spectrum (bandwidth). But if there is a lot of spectral overlap, true spectral methods’ utility breaks down. FLIM can provide a useful alternative, for instance, when a researcher is trying to view RFP [red fluoresecent protein] or GFP [green fluoresecent protein] and a lot of autofluorescence is present. In these cases, which often occur especially in brain and plant tissue, FLIM allows the researcher to separate unwanted autofluorescence from desired fluorescence based on lifetimes, in order to measure only the desired fluorescence. Holst: As a researcher, FLIM has helped me a lot in the field of optical chemical sensing since all calibrations on fluorescence intensity require an incredibly stable lightfield. Any change in intensity, including those that have nothing to do with a change of the analyte concentration, would be nevertheless interpreted as a change of the analyte concentration, while the fluorescence lifetime is a much more stable parameter. The field of FRET will benefit a lot, because FLIM also allows faster access and measurement for experiments and investigations, which use FRET as the method to investigate cell metabolism, inner cell reactions and so on. Further, FLIM can be potentially used for tissue differentiation without staining. Lots of biological tissues have their own fluorescing molecules which might be used for FLIM. Boppart: Increasingly, FLIM is being used to differentiate tissue autofluorescence between healthy and diseased states, and Dr. Laura Marcu’s work for intraoperative FLIM is showing much promise. For my own research, FLIM is being used to track drugs and the metabolic changes and effects from these drugs in real-time, label-free, and even in human clinical studies. We are also using autofluorescence-based FLIM to track cell-death processes such as apoptosis and necrosis, all label-free and now in vivo. Autofluorescence FLIM is an excellent label-free method to provide added contrast about the metabolic state and microenvironment of cells and tissues. Two-photon autofluorescence (gray-scale) and corresponding fluorescence lifetime imaging microscopy (FLIM) images (color-scale) of unstained rat testis tissue captured with fast-acquisition FLIM. The color-scale variations in the FLIM images show cell-to-cell variations in fluorescence lifetime, indicative of different metabolic activity or microenvironmental changes. Courtesy of Biophotonics Imaging Laboratory, Beckman Institute for Advanced Science and Technology, University of Illinois at Urbana-Champaign. Phipps: FLIM is flourishing in biomedical applications due to its ability to clearly distinguish the biochemical differences between diseased and normal tissue in a manner that can be implemented intraoperatively and nondestructively. In the Marcu Laboratory at the University of California, Davis, we are using fluorescence lifetime imaging to study human disease, in particular cancer and cardiovascular disease. We are incorporating FLIM with a Da Vinci surgical robot to acquire measurements during head and neck cancer surgeries. The goal of this project is to provide surgeons with a tool to guide surgical resection of tumors. We are also conducting several studies that involve imaging tumors removed from patients with breast and prostate cancer. Our goal is to better understand these cancers to design an intraoperative FLIM tool for surgeons treating these diseases as well. Additionally, we have incorporated FLIM into an intravascular ultrasound catheter for investigating coronary arteries. We have performed in vivo studies with this catheter in pigs as well as ex vivo studies of human coronary arteries. Orthaus-Müller: FLIM is now a well-established imaging method and has been routinely used both in life and materials science for more than a decade. The method has proven to be very powerful and versatile in elucidating molecular interactions and in probing the local environment of fluorescent probes in biological imaging. Up to now, several thousand papers dealing with fluorescence lifetime imaging have been published, and this number is constantly growing. Fluorescence lifetime measurements were first used in physical approaches to characterize materials and follow protein folding on the single molecule level. Meanwhile, its use has spread out into life science and has become a very popular method, thanks to improved analysis tools, such as fast FLIM and pattern matching. In biology, is typically used to monitor the local environment within cells — for example, pH, ion concentration, amount of lipids and water, and even temperature. This approach requires specific sensors. The most popular approach for FLIM in life sciences is to study molecular interactions by using FRET. This approach provides spatial as well as temporal information about where and when molecules such as proteins, DNA or RNA interact. More importantly, this allows quantifying these interactions. Also, intramolecular distances can be monitored, which allows following processes such as protein folding or enzyme activity. Therefore, molecules of interest have to be labeled with appropriate (donor and acceptor) fluorophores. Another important application of FLIM is the characterization of the autofluorescence of cells to discriminate, for example, healthy from cancerous tissue since the NAD+/NADH ratio varies depending on the metabolic state. Fluorescence imaging in plant biology also benefits strongly from FLIM, since it allows discriminating in a very easy manner the fluorescence label from the (often troublesome) autofluorescence (background) signal. In this way, one obtains artifact-free fluorescence images from the labeled structures. Lin: FLIM has become an essential method for imaging FRET biosensors, due to its quantitative nature and its multiplexing abilities. The requirement for only one fluorophore to measure FRET efficiency allows a FLIM-FRET sensor to be used together with another marker, or most recently even another FLIM-FRET sensor. FLIM is also the only way to reliably quantify FRET from bimolecular sensors, where the donor and acceptor are on different molecules and thus in different stoichiometries in each cell. A mixture of HEK cells depicting the results from a FLIM-FRET experiment. The blue area shows an average fluorescence lifetime of 1.8 ns, while the green area shows an average fluorescence lifetime of 2.3 ns. This means that the blue cells show more FRET than the green cells. Courtesy of Prof. Fred S. Wouters and Dr. Gertrude Bunt, University Medicine, Göttingen and Gerhard Holst, PCO. Herz: The biggest application in cell biology is FRET-FLIM. It is an easy and reliable approach for performing FRET experiments. It is more expensive to set up (in comparison to sensitized emission FRET) because you need a FLIM (camera-based or laser scanning) system, but as more and more imaging facilities and big research groups are getting FLIM systems, it will become more available. FRET-FLIM, for instance, is easier than sensitized emission because it requires less controls. Q: What improvements to the technology do you see on the horizon? Similarly, are there any obstacles to wider use or better performance of FLIM that will remain in place for the foreseeable future? Lin: As with many leading-edge imaging techniques, FLIM needs to become cheaper and easier and widely supported by commercial microscope manufacturers before it can be widely used. Most published uses of FLIM come from labs with expertise in optics or microscope development. While there are a few commercial vendors of FLIM equipment, the combination of expense and complex or custom software has made FLIM a hard sell. FLIM may be clearly the better way to do FRET, but the less quantitative FRET measurement method of ratiometric imaging requires only a standard fluorescence setup and is more intuitive to the novice. For FLIM to become popular, it thus has to become cheaper and easier to implement, and preferably training has to be supplied by the big microscope distributors. The last condition is difficult to achieve as there [are] a limited number of labs that truly need FLIM. Orthaus-Müller: An interesting development lies in the combination of FLIM with other characterization methods such as spectral, dynamic or even topological information from atomic force microscopy (AFM). Such combined measurement methods open promising prospects by accessing different information types from the same sample area in a single experiment. For example, acquiring topological information from AFM and molecular behavior as detected by FLIM was previously limited to correlative experiments, requiring large amounts of statistics especially for heterogeneous biological samples. With a combined FLIM-AFM setup, as can be realized by interfacing the MicroTime 200 microscope to an AFM, the data are acquired in a simultaneous and correlated manner. Extending the spectral range of FLIM into the deep UV is also of great interest since many biological molecules feature naturally occurring chromophoric groups such as tyrosine or tryptophane. By using UV lasers and adequate optical elements, one can exploit this native fluorescence, reducing the need to label such molecules with dyes. In contrast to intensity imaging, FLIM enables [differentiation of] the (auto) fluorescence in UV spectral range based on its different contributions. Increasing the optical spatial resolution is always welcome, as it permits a more precise localization of molecular species or events. The resolution of FLIM can be improved by applying the stimulated emission depletion (STED) method (resulting in a method called FLIM-STED), which permits reaching lateral resolutions below 50 nm. Last, but not least, speeding up FLIM data acquisition in the time domain is highly desirable for studying processes in living cells. Optimized hardware components allow using much higher detection count rates, while reducing artifacts due to the pile-up effect. Herz: FLIM can be divided into two domains: the frequency domain and the time domain. For the time domain, the improvement on the horizon is faster detection due to scalability of parallel detection; on the frequency domain, solid-state sensors. Where in the past an image intensifier was needed for the high-frequency modulation, nowadays image sensors can directly be modulated, replacing the old image intensifier. This gives higher spatial resolution, and it becomes a more affordable solution. Boppart: With improvements, our focus has been on faster methods for real-time FLIM. With these capabilities, we are able to capture FLIM over larger areas of tissues in real time. This helps in a number of drug-delivery and efficacy applications where real-time feedback is essential, and where dynamic changes are happening quickly in the cells and tissues. With fast FLIM, we are also able to capture very fast dynamic events over a limited spatial extent, such as the changes occurring in single neurons or neural circuits. As for obstacles, there will always be a need to capture FLIM images and data at faster rates and at higher resolutions, similar to other optical imaging challenges. Innovative methods continue to emerge, such as from the work of professor Liang Gao at the University of Illinois at Urbana-Champaign leveraging his high-speed imaging techniques for FLIM as well. This image was taken using dual-color FLIM experiments, showing real-time imaging of two biosensors in brain tissue. Schematics of CaMKII-alpha and RhoA-CyRM FLIM sensors (left). Simultaneous fluorescence lifetime images of RhoA and CaMKII activation in a dendritic segment of a CA1 pyramidal neuron in an organotypic hippocampal slice, acquired with 2-photon FLIM (right). Courtesy of Tal Laviv et al. Simultaneous dual-color fluorescence lifetime imaging with novel red-shifted fluorescent proteins. Nature Methods, 12. (October 2016). Holst: Developments and improvements in CMOS image sensor design will help. And there is great interest in the use of time-of-flight CMOS image sensors, which are primarily made for distance measurements to help in the improvement of the existing technologies by enabling higher frequency modulations or better image quality. Another focus will be on a properly designed and user friendly software, to ease the application. This also involves all fields of research, such as the development of optical chemical sensors with more stable responses or longer lifetimes. But any application of FLIM, whether it is in time domain or in frequency domain, requires a certain fundamental understanding of the underlying photo physics, otherwise there are good chances of misinterpretation of the results. Lopez: One of the most important improvements is that FLIM integration is getting better, so today, more companies incorporate FLIM technology into their hardware and software solutions. Ease of use also is improving over time, and the software packages available for FLIM now are streamlining analysis. These changes make the systems more accessible, and you no longer have to be a FLIM expert to have the benefit of using FLIM technology. Phipps: FLIM is a mature technique that has been widely explored in laboratory settings for multiple areas, such as fundamental biomedical research. The major limitation often associated with fluorescence lifetime measurements refers to the complexity, size and cost of the instrumentation. This limitation may still hamper the widespread application of this technique, particularly in clinical settings, where compact and user-friendly devices are required. However, technological advances in recent years have permitted the development of lower- cost alternatives to the traditionally expensive and bulky fluorescence lifetime instrumentation. It is expected that maturation of this instrumentation will promote wider deployment of FLIM devices including for clinical applications. Aside from making FLIM technology cheaper and more portable, another advance that we think would be advantageous would be multiphoton-compatible fibers that could be used for FLIM measurements. These fibers could allow for nonlinear FLIM — the advantage of nonlinear FLIM would be the ability to image deeper into tissues. Currently, FLIM measurements originate from several hundred microns into tissue with our fibers. A higher penetration depth could increase biomedical applications. Lastly, FLIM data is quite complex and requires experts to process and analyze. This challenge to wider dissemination of FLIM as a research and diagnostic tool is being addressed by machine learning algorithms that can be incorporated into easy-to-use software to allow clinicians or other nonexpert FLIM users to interpret FLIM data. Q: What is the next frontier FLIM? What potential remains unfulfilled? Holst: Certainly there is room left for measurements in the picosecond time range with a high frame rate. With scanning systems, it is possible but not so fast, but with imaging systems it is a bit to the limit. First papers have been published with extremely fast CMOS image sensors, and it’ll be interesting to see the performance of these sensors. Sensitivity also could be improved as well as resolution in combination with a decent frame rate. Once FLIM is more widely applied, new applications will follow, I am sure. Phipps: Besides the use of FLIM intraoperatively for head and neck cancer surgeries such as the procedures conducted at our lab, there are many other surgical procedures that could benefit from the guidance of FLIM measurements through flexible fiber optic probes. FLIM can also be added to existing clinical imaging systems to make them more powerful and specific to biochemical composition of tissue. Bi-modal imaging is proving to be very promising for improved detection of diseases since it allows for multiple aspects of a disease to be studied in a single measurement. Clinical application of these bi-modal imaging techniques could provide very unique tools for surgical or diagnostic value. Orthaus-Müller: A very fascinating development is anti-bunching imaging, which allows quantifying the number of emitters per pixel, and in turn makes it possible to obtain information about fluorophore concentration in a biological sample through fluorescence. Another emerging field is the development of FLIM cameras, which are an interesting way to increase acquisition speed. A FLIM camera allows recording a full FLIM image at once instead of scanning the sample area point by point as in conventional FLIM. Lopez: In the past, FLIM’s temporal resolution was slow. Today, it’s much faster; you can now do FLIM at several images per second, which has opened up live cell applications. Of course, FLIM resolution drops as you go to faster image capture rates. So the development of even faster FLIM systems with higher resolution would expand research horizons even farther. FLIM is powerful but it remains a niche application, used in a relatively small number of laboratories overall. The holy grail of FLIM systems would be a low-cost, easy-to-use system that provides readily interpretable results. Science and technology still have a way to go to get there. Boppart: For application areas, FLIM will continue to emerge as a more robust and reliable imaging modality that adds metabolic and molecular contrast to structure-based images. FLIM will be used increasingly to differentiate between health and diseased or dying tissue, and at cellular-level resolution. We will see and understand more about the spatial heterogeneity of effects that occur from cell to cell from drugs or diseases like cancer. All these benefits may also converge. For example, FLIM can begin to distinguish the metabolic differences between individual (and adjacent) cells in a tumor, which we know to be very heterogeneous. The effects of chemotherapy will similarly have various heterogeneous effects between cells throughout the tumor. Finally, FLIM can be used to identify cell-death processes (both natural and from the chemotherapy) in vivo, to improve our fundamental understanding of cancer and develop better drugs for therapy. Herz: More and more FRET sensors are being developed, with a certain purpose. For instance, detection of calcium or glucose. One can imagine that certain sick cells (maybe even cancerous cells) that express a certain unique signal path or have lack of a certain signal path, can be identified with these FRET sensors. Using FLIM to determine FRET, in particular with a camera-based FLIM detector, makes it a fast and less invasive technique for diagnostics.