Nearly two decades ago, Winfried Denk and Karel Svoboda argued in the journal Neuron that “multiphoton imaging is more than a gimmick”1. At the time, confocal microscopy — combined with advancements in fluorescent reporter systems and an increase in desktop computing power — had set a high standard for 3D imaging of biological specimens. But Denk, a German physicist at Bell Laboratories in Murray Hill, N.J., who developed two-photon imaging in 1990 when he was a postdoctoral fellow at Cornell University in N.Y.2, and Svoboda, a physicist-turned-neuroscientist who was then a postdoctoral fellow at Bell Labs, recognized the limitations of confocal imaging. It lead to photobleaching and photodamage of the object under observation and was a “wasteful use of excitation.” The pair concluded: “Virtually all of these problems can be solved by the use of multiphoton optical absorption to mediate excitation in laser scanning microscopy.” Two-photon microscopy has since surpassed confocal imaging for the investigation of intact tissue in live animals. The technique has also taken advantage of the tabletop-sized, ultrafast lasers that have come of age over the past 20 years. In essence, two or more low-energy, long-wavelength photons generated off a Ti:sapphire or ytterbium fiber femtosecond laser are absorbed simultaneously by fluorophores expressed in a tissue of interest. The technique allows for wide imaging windows, deep specimen penetration, reduced photobleaching and photodamage, and pinpoint light focusing. Background signal and tissue autofluorescence are also suppressed without requiring complicated optical setups or spatial filters, as are required in confocal microscopy. Two > one Chris Xu, a professor of applied and engineering physics at Cornell University, was involved in the early development of multiphoton microscopy and worked closely with Denk at Bell Labs after Denk had developed the two-photon technique at Cornell. “The biggest advantage of two-photon imaging is in scattering tissue. The excitation is very localized in three-dimensional space. One photon simply cannot do it,” Xu said, noting that the advantage is most readily apparent in neuroscience. “Unlike other tissues, where you can get quite a lot of information histologically, [in neuroscience] the brain has to be alive. Memory formation, for example, can only occur in the living brain.” He sees two-photon microscopy playing a critical role in our understating of cancer biology and immunology as well. According to Rick Ayer, the senior product manager for multiphoton microscopes at Novato, Calif.-based Sutter Instrument Co., their two-photon Movable Objective Microscope (MOM) is based on Denk’s initial design. “Sutter wasn’t really in the microscope business until we were in the two-photon microscope business,” he said. “The difference from other existing two-photon scopes at the time was that the MOM was effectively the first device that allowed in vivo two-photon, specifically toward imaging in the brain with live animals.” The objective can be moved at an angle to the subject, he added. “The other way had experimenters rotating the animal, which the animal didn’t necessarily appreciate.” Much has changed in microscopy since the initial two-photon setups. Ayer sites galvanometer and resonant scanning, which use mirrors to alter the path of the incoming laser pulse, greatly increasing image acquisition rates to up to 60 fps. This is crucial to monitoring the movements of calcium ions in neurons. “You can give the animal some stimulus — whether visual, olfactory, or auditory — and then watch cells in the brain respond in real time,” Ayer said. He also sites whole-field imaging (in which hundreds of cells are visualized at the same time) as benefitting from multiphoton techniques. Kishan Dholakia — head of the Optical Manipulation Group at the University of St Andrews in Fife, Scotland, a professor in its school of physics and astronomy, and the 2018 recipient of the SPIE Dennis Gabor Award in diffractive optics — and his research group have been working to extend and improve upon multiphoton microscopy. They have combined whole-field, two-photon imaging with temporal focusing and single-pixel detection in a technique they have termed TRAFIX (TempoRAl Focusing microscopy with single-pIXel detection)3. “We stretch the pulse out in time. Then we send the pulse into the tissue, and then we focus it,” Dholakia said. “It’s like survival of the fittest in terms of photons.” The Dholakia group projected Hadamard patterns onto a fluorescent sample — in this case, a smiley face image — through a light-scattering medium of fixed rat brain tissue of up to 400 µm in thickness, and then measured the reflected intensity (Figure 1). A spatial light modulator generated the patterns off of an 800-nm wavelength pulse from a femtosecond Ti:sapphire laser. Figure 1. Kishan Dholakia of the University of St Andrews has combined whole-field, two-photon imaging with temporal focusing and single-pixel detection in a technique called TRAFIX (TempoRAl Focusing microscopy with single-pIXel detection). Courtesy of Kishan Dholakia/University of St Andrews. “We’re excited by this technique because it breaks a barrier in terms of imaging at depth, which is one of the last pillars that we haven’t already tackled properly,” Dholakia said. “These are images you could not take with standard two-photon microscopy. You can have no knowledge of the tissue, its thickness or anything, and still reconstruct an image.” Three > two Xu demonstrated a three-photon microscopy technique while he was a graduate student at Cornell, prior to his stint at Bell Labs. Similar to the skepticism for the two-photon techniques that came before it, there were doubts about the need for three-photon imaging. “At that time, the logic wasn’t clear,” Xu said. “But by the early 2000s, we could see the limitations of two-photon imaging. With three photons you can use much longer wavelengths, as the energy of each photon is even lower.” The imaging wavelength for most two-photon excitation is below 1.1 µm, but with three-photon imaging, the window widens to 1.3 and 1.7 µm, allowing for deeper penetration and better confinement, while maintaining the same overall microscope setup. Sutter’s Ayer said transforming a two-photon microscope into a three-photon one was as easy as swapping the optics. “Most of the objectives in the garden-variety MOM or other two-photon setups begin to lose transmission in the range of 1.1 to 1.3 µm,” he said. Xu concurs: “The laser is different, but the microscope is mostly the same.” The three-photon technique requires a laser with higher peak powers than two-photon excitation. “Initially, it took us a while to develop this laser,” he said, “but now they’re commercially available from a number of companies.” In 2018, Xu’s team and colleagues demonstrated that three-photon excitation could peer beneath the intact mouse skull, which is about 150 µm thick, to reveal brain vasculature and calcium-labeled neurons to a depth of about 500 µm (Figure 2). He said the same experiments could not be done with two-photon excitation, even at a long wavelength4. Figure 2. Chris Xu of Cornell University uses three-photon imaging to image live mouse brain through intact skull. 3PM: three-photon microscopy; RFP: red fluorescent protein. Courtesy of Tianyu Wang and Chris Xu/Cornell University. Typically, to image live brain in a mouse, researchers carve a cranial window from the animal, removing a section of skull and replacing it with a transparent, silicone-based polymer such as polydimethylsiloxane. This surgery introduces sources of potential infection and can diminish the animal’s life span, interfering with possible experiments. According to Xu, his three-photon results would also benefit scientists who work with other tissues in live animals — such as cuticles in flies, skulls in adult zebra fish, and the opaque white matter that covers spinal cords — in which optically aberrated layers can cause degradation of image contrast. The disadvantage, however, is that by increasing the order of excitation, the cross section of the fluorophores decreases. In November 2018, the University of St Andrews’ Dholakia reported an imaging technique that combines three-photon with light sheet fluorescence microscopy (LSFM)5 (Figure 3). While many microscopy techniques — including confocal and standard multiphoton imaging — rely on epifluorescence (in which a light source illuminates an object of interest through an objective and light is returned through the same optic), in LSFM, the light source illuminates the object perpendicularly to the viewing optics. This geometry allows for a lower photon dose to the sample, as well as faster acquisition times. It has been previously operated with one- and two-photon excitation. Figure 3. Kishan Dholakia’s research group at the University of St Andrews has reported the first three-photon light sheet imaging using a propagation-invariant Bessel beam. A bright-field image of a HEK spheroid with a diameter of 450 µm (a). Single XY and YZ near-surface planes (b, c). Blue rectangle in (a) imaged with 3P-LSFM. Courtesy of Kishan Dholakia/University of St Andrews. Although the group used a shorter wavelength of three-photon excitation — at about 1 µm — than Xu typically employs, they were able to three-dimensionally image a cluster of human embryonic kidney cells with a laser light source from one direction. Dholakia’s technique deployed a Bessel beam — a propagation-invariant beam shape that does not spread and “appears to disobey diffraction,” he said. “This beam shape with this new [light] geometry will add a lot of value. If we’re looking at the whole brain project, for example, we can happily keep samples biologically functioning for long periods of time without adverse effects.” Dholakia plans to employ this combination of imaging techniques on live rodent brain with three photons at 1.3 µm, hoping to penetrate samples up to 1 cm. He believes that new optical methods may displace some areas of electron microscopy and high-frequency ultrasound, just as two-photon excitation has displaced confocal imagining for intact tissue in live animals. “In a way,” he said, “physical techniques are driving biological discovery.” Optogenetics Another field that has benefitted from multiphoton excitation is optogenetics, which uses light to manipulate cellular physiology, typically in neurons. In optogenetics, scientists express light-gated ion channels, such as channelrhodopsin (ChR), in cells of interest. Laser pulses activate the channels, which respond to specific wavelengths of light — similar to fluorescent reporter systems. In 2010, the journal Science called optogenetics the breakthrough of the decade. The method has since been making inroads into regenerative medicine and cardiology research as well. In November 2017, physicist Valentina Emiliani of the French National Centre for Scientific Research, and Massachusetts Institute of Technology neurotechnology professor Edward Boyden, a pioneer in optogenetics, reported using a two-photon excitation technique to control, with millisecond precision, individual neurons in intact mouse brain circuits6 (Figure 4). Figure 4. Two-photon holographic stimulation of a soma-targeted channelrhodopsin (soCoChR) enables mapping of functional connectivity in brain slices. The volume around a patched opsin-positive cell (yellow spot) expressing CoChR-GFP (a, top), or soCoChR-GFP (a, bottom) is used to sequentially position a 10- to 14-μm holographic spot (red spots) on nearby opsin-expressing cells. GFP: green fluorescent protein. Courtesy of Or Shemesh/Massachusetts Institute of Technology. Emiliani’s lab developed the two-photon excitation method, which used computer-generated holography and temporal focusing to “sculpt” light onto a subset of specific neurons. “Today, multiphoton excitation allows for the best optical resolution,” she said. “Combined with targeted opsins, holographic multiphoton excitation enables precise manipulation of neuronal circuits, which would be otherwise impossible with conventional optical or electrophysiological approaches.” Boyden’s group created a molecule called soCoChR that directed the expression of the channelrhodopsin CoChR into the soma or cell body of neurons. These were then illuminated by Emiliani’s excitation technique. Boyden said it was important to achieve single-cell resolution in the brain. “Even adjacent neurons can have completely different patterns of activity.” “Up until now, most studies in optogenetics have activated or silenced a population of neurons together,” he said. “By stimulating neurons independently, we enable realistic patterns of neural activity.” According to Emiliani, who directs the photonics department at the Vision Institute in Paris, recent progress in the development of amplified, low-repetition-rate lasers will allow experimenters to reduce excitation power, further minimizing thermal damage to tissue. www.linkedin.com/in/farooqtheahmed Acknowledgments The author would like to thank Chris Xu, Cornell University; Rick Ayer, Sutter Instrument Co.; Kishan Dholakia, the University of St Andrews; Edward Boyden, Massachusetts Institute of Technology; Valentina Emiliani, French National Centre for Scientific Research. References 1. W. Denk et al. (1997). Photon upmanship: why multiphoton imaging is more than a gimmick. Neuron, Vol. 18, Issue 3, pp. 351-357. 2. W. Denk et al. (1990). Two-photon laser scanning fluorescence microscopy. Science, Vol. 248, Issue 4951, pp. 73-76. 3. A. Escobet-Montalbán et al. (2018). Wide-field multiphoton imaging through scattering media without correction. Sci Adv, Vol. 4, Issue 10. 4. T. Wang et al. (2018). Three-photon imaging of mouse brain structure and function through the intact skull. Nat Methods, Vol. 15, Issue 10, pp. 789-792. 5. A. Escobet-Montalbán et al. (2018). Three-photon light-sheet fluorescence microscopy. Opt Lett, Vol. 43, Issue 21, pp. 5484-5487. 6. O. Shemesh et al. (2017). Temporally precise single-cell-resolution optogenetics. Nat Neurosci, Vol. 20, pp. 1796-1806.