Microscopy School Lesson 8 - Lifting the Cellular Fog with Total Internal Reflection Fluorescence (TIRF) Microscopy

Hello, this is John Oreopoulos speaking. I'm a microscopy system specialist at Andor Technology, and today I'm going to be introducing you to the technique of total internal reflection fluorescence, or TIRF microscopy. Whenever I introduce this topic, I always like to start with a picture of a mountain. Often weather conditions and wind movement patterns cause clouds and fog to cover the mountain tops and hide the underlying detail of the mountain base or the mountain face itself.

An analogous situation occurs when we look at cellular cultures with the widefield epifluorescence microscope. Let's imagine that we have a culture of cells growing in a dish, and we go to observe these cells using our traditional fluorescence microscope. In these types of setups, a mercury-based or Xenon-based light source is used to eliminate the sample and effectively we ended up blasting all of the specimen with light that passes through the entire cell volume.

Any fluorescent objects in the cells represented here in this diagram by these green circles will then light up and emit fluorescence that we can view after passing through appropriate emission filters. Well, here's a real-life example of that situation. In this micrograph, we're looking at a monolayer of epithelial cells that have been transfected with a fluorescent version of the protein clathrin, which plays a key role in endocytic processes.

In some parts of this image, we can see the tiny membrane endosome uptake sites as small, bright pinpoint diffraction-limited spots. But in many places, our view of these spots is hindered by the out-of-focus hays of cytoplasmic proteins inside the cell, and further away from the basal surface in contact with the microscope coverslip. This cellular fog is much like the clouds hiding the details of the mountain top I just showed in the first slide of this presentation. Not hindered view of the endosomes is rather unfortunate because it prevents us from observing this multifaceted process that in fact involves many proteins that interact with clathrin.

Other biophysical methods like cryo-electron microscopy have shown that clathrin forms tri-skeleton structure cages around endosomes. But the dynamic nature of this cage formation and the chain of events that lead to membrane uptake by the cell is still poorly understood. Many researchers are trying to tease out the kinetic details of clathrin-mediated endocytosis using multi-colored fluorescence microscopy, coupled with protein gene modification techniques. But again, the cellular fog generated by widefield epi-illumination makes us a difficult task.

There's a good quote by this researcher, Derek Toomre at Yale University, which nicely sums up the problem for this, another biological phenomenon. It reads, ''There are entire sub-disciplines of biology whereby if one could merely see the process with high enough spatial resolution, models of that process that had been debated for decades could be directly proven.'' And this explains why so many years of study have been spent, trying to develop new ways to avoid or remove out-of-focus, fluorescence blur using various optical sectioning techniques.

Most people are of course familiar with confocal microscopy and perhaps deconvolution microscopy as well, which both achieve some degree of optical sectioning. Here, I'm going to explain in detail how a lesser-known third method called TIRF microscopy does the same thing, but with a few key differences compared to the other two techniques. So how does it work? Let's start with a slab of glass and then look at it from its side.

Most of us at some point were enrolled in an introductory physics course at school or university. And from there, you might recall a law of optics called Snell's Law. Snell's Law is a simple mathematical equality, which predicts the angles at which a beam of light approaches indicated by theta one and departs from theta two, an interface between two different optical media. The equality depends on the so-called index of refraction associated with both mediums. For our slab of glass, the associated index of refraction has a value of about 1.51. For the air that surrounds the glass lab, the index of refraction has a value of one. Now let's imagine that we pass a laser beam through the interface first at an angle that is normal or perpendicular to the interface. In this case, we see the laser beam passes straight through with no angular deviation.

If we pass the laser beam through the slab of glass at a small angle with respect to the interface normal thereby increasing theta one, we see that a small amount of laser light reflects at the same angle back into the glass slab and a larger portion of the laser beam refracts out of the glass at a larger angle theta two, again, obeying Snell's Law. This trend continues as the angle of incidence of the laser beam is increased further. The trend continues until a special angle is achieved, which we call the critical angle theta C whereby the emergent beam from the glass-air interface refracts at an angle of theta two equal to 90 degrees and skims among the interface surface. Notice that the critical angle can be calculated since it only depends on the ratio of the index of refractions for the two optical mediums.

And then if the angle of incidence in the glass slab is increased just a bit more past the critical angle, something very interesting happens. Instead of refracting out of the glass, the laser beam now completely reflects back into the slab. This optical effect is called Total Internal Reflection, and this phenomenon explains why diamonds are so brilliant and sparkly. And this can be understood by examining the side profile of a traditional diamond. Here, you can easily see why the angular cut is so important and can highly impact the cost of a diamond. Diamond with an ideal cut doubly exhibits Total Internal Reflection, causing it to appear very bright and dazzling. Whereas a diamond with a cut that is too deep or shallow will appear darkened and cloudy respectively. So make sure your fiance knows about Snell's Law when they go to purchase an engagement ring.

Total Internal Reflection also explains how and why optical fibers are able to transmit light over great distances with little to no power losses. But what you learned in basic high school or university physics is actually a very simplified geometric optics view of what happens when light internally reflects on an interface. What occurs at the point of contact where the instant laser beam strikes the interface and subsequently completely reflects back into the high index optical medium? The answer to that question can be found in any physical optics textbook like this one by Hecht.

Without going into full details, a simple explanation is that most physics phenomenon are continuous at a boundary. That is to say, there can't be a discontinuous change for a physical phenomenon and an interface. Using Maxwell's equations of electromagnetism, it's possible to derive the existence of a special electric field called the evanescent wave or an evanescent field, which has an intensity distribution that exhibits an exponentially decaying characteristic.

The decay constant of the evanescent field called the penetration depth, DP, depends on the refractive indices of the optical media, the light beam angle of incidents, and the wavelength. Using typical values, the penetration depth works out to be about a fifth of the wavelength of light being used. As stated by Hecht in his textbook in many ways, the evanescent wave is the photon analog of electron quantum tunneling through a dielectric interface.

So let's now consider the situation where the second optical medium is water, which has an index of refraction and two equal to approximately 1.33. And we'll inject a blue 48-nanometer laser beam at the glass water interface with an angle of incidence of approximately 65 degrees. In that situation, the penetration depth of the evanescent field is about 100 nanometers. We'll place a small fluorescent object represented by this green sphere in the water above the interface. Our fluorescent object won't emit detectable levels of green fluorescence until it diffuses into and enters the evanescent field.

Well, this means that some amount of energy actually did transfer into the water medium to cause light absorption and subsequent fluorescence emission. In this case, the word total in total internal reflection is a bit of a misnomer. The energy transfer process only occurs if there's something there in the second medium that can absorb the light. And this process is termed frustrated total internal reflection.

You can actually observe a non-fluorescence version of this effect the next time you hold a cold glass of water with a bit of moisture on the outside with your hands. Looking through the water, you'll very clearly be able to see the skin patterns, lines, and curves that make up your fingerprints. So what's the application of all this? Let's again, consider our slab of glass from a side view profile.

In the early 1980s, Dan Axelrod, who's now an emeritus professor of the University of Michigan Ann Arbor had the bright idea to culture some living cells on a slab of glass, pass a laser beam through the glass at a large oblique angle, and then put the entire setup under a microscope. Axelrod had labelled the cell membranes of these cells with the fluorescent lipophilic carbocyanine probe, DiI-C 16, using evanescent illumination and focusing the microscope through the specimen and onto the basal surface of the cells. Axelrod was then able to obtain an ultra clear picture of the cellular membrane regions in contact with the glass interface. Essentially, he was able to see the cells' fingerprint profile because of frustrated total internal reflection. This gets to the optical sectioning concept of seeing more by illuminating less.

And here are the associated micrographs of that pioneering study by Axelrod originally presented in ''The Journal of Cell Biology.'' On the left, we see the regular widefield epifluorescence image of this kind of specimen and on the right, the corresponding TIRF image of the sample is shown. In wide-field epifluorescence, the cells mostly appear solid with little to no contrast against the dark background, but under TIR illumination, the non-even contact regions of the cell become readily apparent.

And here again, is our wide-field epifluorescence image of epithelial cells expressing clathrin GFP. Let's look at this same field of view using TIR illumination. The corresponding TIRF image of this sample presents a clear view of every clathrin-coated pit in the cells without any cytoplasmic protein out-of-focus blur or background haze. Note that the nuclear shadow of each cell is now invisible, another indication that we are seeing a very small and shallow optical section of the cell membrane against the glass coverslip. In fact, in this image, it's now nearly impossible to determine where the nucleus is for each cell in the field of view.

And since TIRF microscopy is a wide-field technique where the images are acquired with a fast sensitive camera detector, it's possible to acquire movies of these living cells to observe dynamic protein behavior. This is a zoomed-in view of one of the cells from the previous slide showing several minutes of acquisition sped up the video rate to demonstrate the different types of protein interactions with the membrane.

TIRF movies like this one can be processed using various automated image analysis routines that quantify protein movement, intensity changes, or even cluster or spot size distributions. All of these image parameters can be used to formulate statistical data about the biological process under examination. So what did Axelrod's experimental apparatus to acquire TIRF images of cellular specimens look like? He used a series of optomechanical parts to direct and focus a free space launched laser beam towards a custom sample chamber mounted on an inverted microscope. The laser light was coupled into the sample substrate interface at a steep oblique angle using a glass cube optically contacted to the chamber coverslip using a drop of immersion oil. Axelrod called this the prism-based TIRF apparatus configuration.

Here's a closeup schematic of the prism-based configuration sample chamber. Cells were cultured upside down on the glass coverslip and the chamber was sealed allowing for continuous fluid exchange. Upright microscope versions of the same configuration have also been presented by Axelrod's research group. Although the prism-based TIRF microscope design is simple and low cost, it does come with a number of disadvantages as well.

First, the alignment is tricky and difficult to reproduce. The specimen must be mounted in a very thin chamber and the illuminated surface must be imaged through the entire volume of the specimen, and so optical aberrations need to be considered. Lastly, this configuration uses an open laser beam path. So laser and eye safety concerns abound. So the question is, is there an alternative way that the same illumination conditions can be achieved without these disadvantages of the early TIRF microscope designs? The answer to this question is yes, and the solution was presented several years later by Andrea Stout and Dan Axelrod using an ingenious optical trick with high numerical aperture objective lenses.

On the left, we see a modern-day high NA oil immersion objective lens. These lenses are very expensive due to the large number of optical elements within the objective barrel consisting of exotic glasses, which when shaped and used in the correct combination lead to high spatial resolution aberration-free images with high signal levels. Often, these optical elements are still polished and assembled into the objective lens barrel by the hand of a skilled technician. On the right is shown an optical model of the glass elements inside such an objective lens composed using a lens glass and shaped prescription in an optical Ray tracing program like Zemax. Zemax allows one to see how light propagates through the objective lens.

Here, I've used Zemax to launch an expanded and collimated laser beam into the back aperture of a high NA objective lens. In this situation, we see how individual rays associated with the laser beam eventually form a tight focus at the top of the objective lens and subsequently diverge from the sample plane. This is the condition in which laser light enters the objective lens for single-point laser scanning confocal microscopy

Before continuing with this ray tracing exercise, I want to remind everyone objective lenses, not only have an associated front focal plane at the top of the objective, but also an associated back focal plane at or near the bottom of the objective lens. You can prove this fact for yourself by holding up an objective lens to a faraway target object and viewing that object through the back end of the objective lens. When you do this, you'll see a small demagnified and inverted image of the target somewhere inside or close to the back aperture of the objective lens. What you're looking at is the objective lens back focal plane under this condition.

So coming back to the ray-tracing model, if I now insert another simple lens into the optical pathway of the laser beam such that the light focuses down onto the objective lens back focal plane, instead of a tight focus emerging from the objective, I now achieve a thin tightly collimated beam pointing straight out of the top of the objective lens. If I laterally offset the position of the simple focusing lens from the optic axis, such that the beam moves transversely and radially along the objective lens back aperture, we see that the collimated beam begins to emerge from the top of the objective lens at an angle with respect to the optic axis.

I can continue to increase the lens's lateral offset to further increase the angle of the collimated beam emerging from the top of the objective lens. I can keep doing this until a point is reached where the emerging collimated beam exits the objective lens at an angle of 90 degrees with respect to the optic axis and skims across the glass coverslip surface. This effect can be visualized by eye simply by creating an analogous optical setup on a microscope and filling a dish with a bit of a fluorescent aqueous media. Finally then, if I translate the focus laser beam, latterly towards the edge of the objective lens aperture, a bit further beyond this point as you might expect, all of the collimated laser light totally internally reflects at the glass water interface at the top. All of this is to demonstrate that there exists a relationship between the radial position of the laser beam in the back focal plane and the illumination angle of incidence at the front focal plane. Longer radial positions of the focus laser beam correspond to more oblique angles of illumination, which lead to shallower penetration depths.

Looking through the back of the objective lens again, we can see that total internal reflection occurs only when the laser beam is focused in a very thin annular region indicated here by the color red of the objective lens back focal plane. Let's now take a closer look at what's happening at the front focal plane of the objective lens. Let's imagine again, that we have a monolayer of fluorescent cells growing on the glass coverslip. In this zoomed-in view of the objective lens model, we can see that the collimated laser beam approaches and internally reflects from the front focus of the objective lens with a well-defined angle of incidence. And look at that, we've achieved the conditions for total internal reflection illumination using a high numerical aperture objective lens and any fluorescent light generated by the sample and within the cone that defines the objective lens numerical aperture is captured and directed to the camera after passing through appropriate emission filters.

The experimental apparatus that achieves this illumination scheme is referred to as the through-the-objective TIRF microscope configuration, and indeed many commercial TIRF microscopes use this scheme instead of the prism-based design because of its simplicity and ability to simultaneously collect as much fluorescent light from the sample as possible. Usually, the laser light enters through the backport of an epifluorescence microscope with some adjustment mechanism to transversely offset the laser beam and the emission light is directed to a camera mounted on one of the microscopes imaging ports.

Let's just take a moment to again, consider the regions of the back focal plane of the objective lens that permit TIR illumination to occur. This diagram shows how the annular regions available for TIR illumination in the back focal plane shaded here, red, vary with its numerical, aperture, and magnification. The lens on the left has a very small annular region available for TIRF, and it can be difficult to focus and align the beam properly for this case. The lens in the middle, however, is the same as the one on the left, except with a higher numerical aperture. Note how the shaded region of this lens's back focal plane has a thickness that is about double than that of the one on the left.

Then on the right, we can see that holding the numerical aperture constant and increasing the magnification from 60X to 100X has the effect of, again, shrinking the annular region where TIRF illumination can be achieved. The objective in the middle can achieve smaller penetration depth than the objective on the left. But the objective on the right can achieve the same minimum penetration depth as the objective in the middle, albeit that the laser alignment must be more precise because of the smaller shaded annular region.

The thicknesses of the annular region stated on this slide assume that the sample refractive index is equal to that for water. But in reality, the sample refractive index may be a bit higher depending on the specimen characteristics and subsequently reduce the thicknesses of these annular rings. It should be noted that even larger numerical aperture TIRF objective lenses exist that can increase the thickness of the annular region in the back focal plane. These objective lenses allow the user to access even shallower penetration depths that lead to thinner optical sections at the glass water interface. For example, 100X objective lenses with a numerical aperture of 1.65 are available. However, it should also be noted that objective lenses like these require volatile and toxic immersion oil that can be corrosive. And the coverslips are expensive because they must be made out of sapphire, which has an index of refraction equal to 1.78.

TIRF microscopy can complement other imaging techniques. The Andor Dragonfly uniquely combines widefield epifluorescence, high-speed 3D confocal imaging, and TIRF microscopy into a single instrumental platform that connects to any research-grade microscope. The top use schematic on the right of this slide shows the light paths for these three different illumination methods and how they're merged and injected into the microscope side port. The fluorescence emission pathway is common for all three imaging modes and they share the same camera, which means that we can automatically register the images of these three imaging modes as well. And since the images of these three imaging modes are registered with each other in the Dragonfly platform, it allows us to directly compare the optical sectioning capability of these different illumination techniques

Here again, we can see a cell expressing fluorescent clathrin proteins. The confocal image of the specimen in the middle clearly does a better job at rejecting out-of-focus light that is seen in the epifluorescence image on the left, but the TIRF image on the right exhibits, even better contrast and out-of-focus blur avoidance than the confocal image. Here's another example demonstrating the same thing. This time, we're looking at a cell expressing a fluorescent version of the focal adhesion protein zyxin. The nuclear shadow of the cell is visible in both the epifluorescence and confocal image of the cell, but not in the TIRF image. The TIRF image truly shows the footprint regions of the cell that contact the glass coverslip, and it's much easier to discern the small, bright adhesion points that accumulate zyxin proteins.

Here's the important point you need to keep in mind, TIRF microscopy's greatest strength is also its greatest weakness. It really is a double-edged sword. Among the various super-resolution light microscopy techniques available today TIRF easily achieves one of the highest axial resolutions, but TIRF imaging is limited to the sample substrate interface. Three-dimensional optical sectioning of the sample as achieved with confocal microscopy is simply not possible.

This table directly contrasts some properties of TIRF microscopy against confocal microscopy. The lateral resolution of both techniques are about the same but notice again that the axial resolution of TIRF is approximately an order of magnitude better compared to confocal microscopy when imaging a sample at the glass water interface. The other big advantages of TIRF microscopy over confocal microscopy are the detector sensitivity, the very fast imaging speeds, and the typically lower cost.

I would like to now focus on some specific applications of TIRF microscopy. Here's a list of the main cellular processes and structures that TIRF excels at observing. This includes secretory granule and vesicle tracking, endocytosis and exocytosis, extracellular matrix structure and assembly, looking at lipid and membrane protein biophysics, single-molecule protein motion and dynamics, cell adhesion, migration, motility, and mechanics, or even membrane trafficking dynamics. We can look at calcium sparks and ion channels, membrane and cytoskeleton remodeling, surface or interfacial chemistry, signal transduction, viral assembly and dynamics, and even super resolution imaging.

Let's look at a few examples from the past scientific literature and a few modern case studies as well. Here's a very early example from the research group of Kai Simmons published back in the year 2000. In this study, TIRF was used to monitor vesicle trafficking from the golgi apparatus to the cell membrane. To do this, the researchers doubly labeled the golgi and the vesicles with two different colored fluorescent probes. One probe was imaged with widefield epifluorescence, which of course allowed them to see the entire cell volume all at once. And the other probe was then observed in a TIFR channel, which allowed them to restrict the view to just the cell membrane.

Rapid switching between these two imaging modes proved to be an effective way to track the movement of vesicles from higher up in the cell volume, down to the cell membrane, right up to the point of vesicle fusion. The movies acquired of living cells in this early study really are quite stunning to watch and they reveal the highway-like transport vesicles throughout the cell. But the most interesting moments captured in this study are those a vesicle fusion with the cell membrane. Here's a zoomed-in view of the cell. Each bright green flash represents a single vesicle fusion event. The researchers noticed that the motion of the vesicles near the membrane could be categorized into one of three types, including vesicles that fuse, vesicles moving close to the membrane and then receiving backup into the cell without fusing and also vesicles that remain stationary, hovering close to the membrane for some extended period of time before eventually fusing.

This was one of the first studies to categorize and analyze a large number of such events. And this allowed the researchers to develop the so-called kiss and run model of vesicle dynamics. This figure from the paper shows how the researchers used region of interest intensity image analysis methods to build up statistics of the different vesicle behaviors. Another early groundbreaking study was presented by Funatsu and coworkers in ''Nature Magazine'' in 1995. This was the first study to demonstrate video-rate room temperature imaging of single fluorescent molecules and was made possible by stringent sample preparation techniques combined with the usage of TIRF microscopy to achieve ultra low background signal levels on an intensified CCD camera.

The researchers could be sure that they were observing and restricting their analysis to only single molecules by tracking the stepwise photobleaching spot intensity behavior to background noise levels of the camera. One of the amazing results of this paper was that the researchers were able to determine through image analysis alone, the ATP energy molecule dissociation rate from single myosin molecules bound to the coverslip substrate using linear regression of the observed lifetime of molecular turnover events.

Here's a more recent example where TIRF microscopy was used to get a better understanding of viral assembly within living cells. This 2008 study presented in ''Nature Magazine'' used the sensitive imaging characteristics of TIRF to track the dynamic formation of individual HIV virions at the cell membrane surface. Statistical intensity analysis of individual virus assembly events again allowed determination of three distinct types of virion assembly pathways, something that wouldn't be possible by traditional biochemical means.

Molecular motors are important cell machines, which facilitate vesicle transport throughout the cell. Over the years, structural biology studies and methods had done the hard work to understand how these motors bind to cytoskeletal microtubules. But the dynamic mechanism of transport was lacking. In this study, Ahmet Yildiz and Paul Selvin developed a technique using TIRF microscopy to track and localize the position of single mysoin molecular motors with very high precision down to one nanometer using the concept of sub-pixel localization.

This level of precision allowed the researchers to determine which of the two potential models of myosin movement was indeed the one that exists in nature, a hand-over-hand walk or an inchworm walk, each with different predictable step sizes ranging in tens of nanometers. Single molecular TIRF imaging experiments combined with localization analysis, allowed them to conclude that the first model of motion, the hand over hand walking motion was the correct mechanism. And because of this, we now have very accurate molecular models of this vesicle transport process and the implications for various muscular degenerative diseases are much clearer.

Sometimes cellular imaging studies require the fast sensitivity of TIRF and the 3D imaging capability of confocal. Tokunaga and co-workers showed in this 2008 nature methods publication that a compromise in imaging quality can be struck by tuning the TIRF illumination angle to be at or slightly above the critical angle. This leads to a highly inclined and laminated oblique beam of light that passes through the sample in a manner that is similar to light sheet microscopy.

This imaging strategy, which is abbreviated as HILO imaging can be very effective for single-molecule studies that occur internally within the cell and beyond the cell membrane. It turns out that TIRF and HILO imaging are also extremely useful for so-called localization-based super-resolution microscopy methods, which draw upon the same sub pixel localization strategy that was demonstrated with molecular motors. DNA paint is one such super-resolution technique that uses TIRF or HILO imaging and exploits the on-off blinking behavior of fluorophores that are linked to complementary single-strand DNA sequences. In solution, these DNA probe molecules transiently bind to secondary antibodies with DNA that matches the probe molecules and which are attached to a cellular structure of interest. Rapidly acquiring thousands of TIRF or HILO images of a specimen labeled in this manner and subsequently localizing the sub pixel position of each fluorescent on-off event allows one to compose a super resolution image of the sample.

Here's a super-resolution rendering of another cell adhesion complex protein, talin, by one of our Andor Dragonfly customers based out of the University of Delaware. And this is a closeup view of the same cell showing the comparison between regular diffraction-limited TIRF imaging of the specimen and the super-resolution rendering of the same field of view. The improvement in image resolution allows the researchers to probe and ask deeper questions about the molecular mechanisms associated with this protein and the other proteins that it interacts with within the adhesion complexes.

Finally, let me end this presentation with a few practical tips for anyone wanting to try out and use the TIRF imaging technique. First under true TIRF conditions and by that, I mean, when the laser beam is set to a supercritical angle of elimination, you should only be able to focus on one plane within the specimen, the cover slip glass water interface. TIRF uses a coherent laser source, and so interference fringes are difficult to eliminate. It's essential to keep all optical surfaces as dust-free as possible, especially any microscope auto-focused dichroics and clean them as necessary.

Finally, you can always sanity check the optical alignment using fluorescent beads and or a highlight or die immersed in water. The images here show that a dish of fluorescent water allows one to determine if the laser beam is focused to the objective lens back focal plane, and determine if the laser beam resides at a sub or super-critical angle of incidence.

This is what you should see when imaging a solution of fluorescent beads. When the angle of incidence is set to a subcritical angle of illumination, the field of view should be flooded with fluorescent beads that rapidly diffuse and go in and out of focus. In contrast, when the angle of incidence is set beyond the critical angle, you should only observe fluorescent beads adherent to the glass coverslip and the image background should become very low. Non-adherence fluorescent beads that diffuse into the evanescent field should appear to twinkle or rapidly flash as they move up and down.

And here I leave you with a few references of interest related to TRIF microscopy. All of the review papers by Dan Axelrod should be considered essential reading. The book chapter by Parker in ''Methods in Enzymology'' provides some very practical advice for learning to work with optomechanical prototyping parts, and even building your own TIRF microscope.

And to get up-to-date information on the latest developments for this imaging technique, I would recommend the published works of the three researchers listed at the bottom of this slide. So we started with a picture like this at the beginning of the presentation, noting how the clouds mast our view of the mountain. I said that these clouds were like the cellular fog or out-of-focus blur that hinders our view of the cell when using epifluorescence microscopy. TIRF avoids the generation of this fog, lifting the clouds away and allowing us to see the fine details that lie at the bottom of the mountain. And with that, I'd like to thank you for joining me today.