Thinned-skull cranial window technique for long-term imaging of the cortex in live mice

Technical limitations and controls for imaging through thinned-skull windows

The limitations of the thinned-skull technique are mainly related to the fact that skull thickness is critical for image quality and achieving the optimal thickness in a consistent way requires significant surgical practice. Although a thinned-skull window approach has been used in various studies, such as intrinsic optical imaging of functional organization within rat barrel cortex²¹, TPLSM imaging of the cerebral vasculature⁸ and amyloid plaques¹⁹, these studies were focused on imaging much larger structures than synaptic connections (dendritic spines, filopodia and axonal boutons) and did not require the skull to be exceedingly thin^8,19. A non-uniform skull thickness may cause significant spherical aberrations, resulting in decreased two-photon excitation and distortion of fluorescent structures located deep inside the cortex²². We have found that the skull thickness should be < 25 μm in order to obtain high-resolution images of synapses^2,7,10. Conversely, over-thinning the skull beyond 15 μm carries the risk of mild cortical trauma resulting from pushing the skull downwards with the drill bit or microsurgical blade, leading to mild cortical inflammation. In such cases, the thinning process generally leads to the activation of microglia, as manifested by their amoeboid morphologies, and is often associated with mild neuronal injury, as manifested by axonal and dendritic blebbing, and the eventual disappearance of fluorescent structures. Injury of this type can be prevented by carefully using the microsurgical blade at an angle during the thinning procedure and avoiding pushing the skull against the cortical surface. Furthermore, although multiple imaging sessions within the first few days after the initial surgery can be easily carried out and require only minimal removal of debris over the thinned cranium, repeated imaging over intervals greater than 2–3 d requires skull re-thinning, which could be challenging for an inexperienced operator. It is often difficult to carry out more than five imaging sessions in the same animal, as the optical properties of the skull may gradually deteriorate with repeated thinning.

To control for the quality of images acquired through thinned-skull windows, it is important to measure the thickness of the skull by imaging it with TPLSM to ensure that the thinned region is ~20 μm and relatively even in thickness. It is also important to ensure that no surgery-induced cortical injury has occurred, as indicated by axonal and dendritic blebbing or glial activation, in the superficial cortical layer under the thinned-skull regions. Supplementary Movie 1 shows an imaging stack from a successful preparation covering the thinned skull (~20 μm), meningeal layers and neuronal structures within the first 100 μm under the pial surface. The finest structures, such as dendritic filopodia, should be clearly seen under the thinned-skull window. Images of the superficial cortical layer (the first 50 μm under the pial membrane) obtained with a thinned-skull preparation should have nearly the same signal to noise ratio as those obtained with the same laser power and imaging settings using a craniotomy approach².

Comparison of thinned-skull versus open-skull windows for in vivo imaging

In addition to the thinned-skull preparation, the cortex can also be visualized through open-skull windows after performing a craniotomy and replacement of the skull by a cover glass with or without a layer of agarose on top of the dural layer^23–31. In Table 1, we compared and contrasted the results from the studies of postsynaptic dendritic spine plasticity using thinned-skull and open-skull windows in the past several years. It is clear that these two methods have generated drastically different results when measuring the turnover rates of dendritic spines in the cortical excitatory pyramidal neurons^{7,10,11,23,24,26,28}. This discrepancy has led to contradictory views on the degree of structural synaptic plasticity in the adult brain and how long-term information might be stored in synaptic circuits^2,26,32–34. Although the precise reasons responsible for the differential degree of spine plasticity measured using these two preparations remain to be determined, it is evident that skull removal can induce a significant inflammatory reaction and could, thus, confound efforts to elucidate changes of neurons and glia in the intact brain^7,20,31. In contrast, the thinned-skull window approach not only causes minimal perturbation but also allows imaging of the cortex immediately after surgery as opposed to the open-skull method, in which optimal imaging quality is achieved many days after craniotomy^{2,10,11,24,26}. Therefore, we believe that the thinned-skull window approach should be used for a majority of experiments requiring in vivo TPLSM imaging of the living cortex, whereas the open-skull window approach may be used in those experiments in which it is necessary to image a large portion of the cortical surface or bathe the cortex with pharmacological agents.

Applications of thin-skull windows for in vivo imaging

In the past several years, TPLSM imaging through a thinned-skull window has become an important tool for studying structural and functional changes in the living cortex^1–14. With the availability of a large number of fluorescent reporters to probe the structure and function of the brain^35,36, we envision that the transcranial TPLSM imaging will greatly expand our present understandings of how the neural circuits are assembled and modified throughout life and how glial and other cells function in the living brain.

REAGENTS

• Experimental animals: Transgenic mice expressing fluorescent proteins in the cytoplasm of cortical neurons (e.g., thy1-YFP line) or CNS resident microglia (e.g., CX3CR1-EGFP line)^37,38 ! CAUTION Animal breeding and handling should comply with relevant institutional and national animal care guidelines.

• Ketamine (Fort Dodge)

• Xylazine (Shenandoah)

• Sterile alcohol prep pad (Fisherbrand, cat. no. 06-669-62)

• Sterile lubricant eye ointment (DEL Pharmaceuticals)

• Cyanoacrylate glue (Fisher Scientific, cat. no. 11-999-24)

• 6-0 Silk suture (LOOK, cat. no. M552830)

• NaCl (Sigma, cat. no. S9888)

• NaHCO₃ (Sigma, cat. no. 401676)

• KCl (Sigma, cat. no. P9333)

• NaH₂PO₄ (Sigma, cat. no. S8282)

• MgCl₂ (Sigma, cat. no. 449172)

• Glucose (Sigma, cat. no. G7528)

• CaCl₂ (FisherBiotech, cat. no. 996862-36)

• Dextran Rhodamine B MW 70,000 (Molecular Probes, cat. no. D1841) 10 mg ml^{− 1} in sterile saline.

EQUIPMENT

• Head immobilization device: Custom built plate and skull holder (Fig. 1a and Supplementary Fig. 1)

  • A 14 cm × 10 cm × 0.1 cm aluminum plate.
  • Two 18 mm × 18 mm × 18 mm aluminum blocks.
  • Two ¼ inch screws and two spacers.
  • Three or four conventional double-edge shaving blades.
  • Custom built plate: Glue the two aluminum blocks to the plate. The blocks are placed about 2 cm from one of the short sides of the plate and 3 cm from each other. A hole with internal thread was drilled on each block to accommodate the ¼ inch screw.
  • Skull holder: Glue 3 or 4 shaving blades to each other.

• High-speed micro drill (Fine Science Tools, cat. no. 18000-17)

  ▲ CRITICAL High-speed operation of the drill is critical for consistent results. Slow drill speeds may result in nicking of the skull or uneven thinning. Make sure that the drill is adequately charged and in good working order.

• Micro drill steel burrs (Fine Science Tools, cat. no. 19007-05 or 19007-07)

  ▲ CRITICAL The drill bur should be frequently replaced and should not be used if it becomes heavily contaminated with debris or if it is performing poorly.

• Double-edge shaving blades (CAMB Machine Knives International LLC, cat. no. CMK169S)

• Microsurgical blades (Surgistar, cat. no. 6900) ▲ CRITICAL The size, dimensions and quality of the microsurgical blade are important for achieving maximum thinning without depression of the skull. Use a new blade each time for optimal skull thinning.

• Dissecting stereomicroscope (Olympus, BX50WI)

• CCD camera (Olympus, U-CMAD3): although the CCD camera is attached to a stereomicroscope in our setup, it is advantageous to have it directly equipped on the TPLSM microscope platform if possible.

• TPLSM microscope: We have used either a Bio-Rad 2001 multi-photon microscope or a custom-built multi-photon microscope equipped with a mode-locked laser system. For both systems, the laser system (Tsunami and Millenia Xs, Spectra Physics) is tunable from 690 to 1000 nm wavelength with 80 MHz pulse repeat and < 100 fs pulse width. The laser-scanning system is coupled to an upright fluorescence microscope. For the custom-built unit, an Olympus laser scanning system is used and detectors (photomultiplier tubes) are placed close to the objective to facilitate the signal detection. It is worth mentioning that standardized microscope objective holders generally do not provide a positioning accuracy sufficient for co-centering different objective lenses at micron precision. However, some microscope producers provide objective holders including spring loaded adjustable micropositioners, providing the means to co-center the objective lenses. Such objective holders are useful when switching from one objective to another for imaging.

• Water-immersion objective (e.g., Olympus ×60; NA 1.1, Olympus)

  ▲ CRITICAL Although many ×40/×60 water immersion long working distance objectives are suitable for TPLSM imaging, the decisive factor for resolution purposes is the numerical aperture. Any objective used for TPLSM should be assembled from lenses with special coatings, which do not absorb or reflect a large fraction of the incident laser light. The users should choose the objectives based on their experimental needs and check with microscope producers for their newest development of lenses for TPLSM imaging. For example, Olympus renders a ×25-water immersion lens at 1.05 NA, which is specially developed for the purpose of TPLSM and useful for in vivo imaging. We generally use Olympus ×60, NA 1.1 objective for in vivo imaging of dendritic spines.! CAUTION Keep the objectives immersed in water after imaging and clean them periodically with 100% ethanol.

Image acquisition software (Fluoview 5.0 or Lasersharp 2000)

REAGENT SETUP

Thinned-skull preparation ● TIMING 30–45 min

• 1| Anesthetize mice by intraperitoneal injection (5–6 μl g^{− 1} body weight) of KX. Wait for 5–10 min. Place the mouse on a cotton pad after a surgical level of anesthesia has been reached. A heating pad may be inserted under the cotton pad to maintain a body temperature of ~37 °C.

  ! CAUTION All animal experiments related to surgery and imaging should comply with relevant institutional and national animal care guidelines.

  ▲ CRITICAL STEP Continuously monitor the depth of anesthesia by testing the animal’s reflexes (e.g., pinching the animal’s foot with a blunt pair of forceps and checking for the absence of the eye blinking reflex) during the surgery and inject more KX when necessary.

• 2| Thoroughly shave the hair over most of the scalp with a double edge razor blade. Remove the residual hair and clean the scalp with sterile alcohol prep pad, then perform a midline scalp incision, which extends approximately from the neck region (between the ears) to the frontal portion of the head (between the eyes). Carefully disrupt the fascia located between the scalp and the underlying muscle and skull with a pair of spring scissors, taking care not to sever blood vessels.

• 3| Lubricate both eyes with a drop of eye ointment. If microglial imaging is being carried out, then inject 100 μl of rhodamine dextran solution retro-orbitally for vessel labeling before the application of eye ointment or alternatively inject it through the tail vein.

  ! CAUTION Dehydration of eye tissue can cause permanent damage to the eyes.

• 4| Localize the brain area to be imaged based on stereotactic coordinates and mark it with a pen.

• 5| Remove the connective tissue attached to the skull over the area to be imaged by gently scraping the skull with a blunt microsurgical blade. Local anesthetics may be added to the skull surface to minimize pain associated with drilling of the periostium.

  ! CAUTION If local anesthetics are added, it is important to examine the effect of anesthetics on the brain structure (e.g., spine and filopodia dynamics), as it may confound the results of experiments.

• 6| Place a small amount of cyanoacrylate glue around the edges of the internal opening of the skull holder and press the holder against the skull. Make sure that the area to be imaged is exposed in the center of the internal opening (Fig. 1a).

  ▲ CRITICAL STEP Skull immobilization, which is in part determined by the quality of the skull holder to skull bonding, is very important to reduce movement artifacts during imaging.

• 7| Gently pull the loose skin upto the edges of the internal opening of the skull holder and apply a small amount of glue to the edge of skin to adhere it to the skull.

  ▲ CRITICAL STEP Gluing the skin around the internal opening of the skull holder and the skull helps to hold the ACSF in place during imaging when using a water immersion lens.

• 8| Wait for ~5 min until the skull holder is well glued to the skull. Attach the skull holder to the skull immobilization device by gently inserting the lateral edges of the skull holder between the aluminum blocks and washers/screws of the skull immobilization device. Tighten the screws to completely immobilize the skull holder (Fig. 1a). Wash the opening of the skull holder a few times with ACSF to remove the remnants of non-polymerized glue. This helps prevent the glue from contaminating the microscope objectives.

• 9| Immerse the skull with a drop of ACSF (close to body temperature, 36–37 °C). Use a high-speed micro-drill to thin a circular area of skull (typically ~0.5–1 mm in diameter; Fig. 1b) over the region of interest under a dissecting microscope. Carry out drilling intermittently during the thinning procedure to avoid friction induced overheating. ACSF helps soften the bone and absorbs heat. Replace the ACSF periodically and wash away the bone debris.

  ▲ CRITICAL STEP Do not thin a large region (>1.5 mm) to a thin layer (< 50 μm) as it is difficult to exercise fine control of the drilling process and may cause damage to the underlying cortex.

• 10| The mouse skull consists of two thin layers of compact bone, sandwiching a thick layer of spongy bone. The spongy bone contains tiny cavities arranged in concentric circles and multiple canaliculi that carry blood vessels. Remove the external layer of compact bone and most of the spongy bone with the drill. Some bleeding from the blood vessels running through the spongy bone may occur during the thinning process. This bleeding will usually stop spontaneously within a few minutes.

• 11| After removing the majority of the spongy bone, remaining concentric cavities within the bone can usually be seen under the dissecting microscope, indicating that drilling is approaching the internal compact bone layer. At this stage, skull thickness should still be more than 50 μm. Continue skull-thinning with a microsurgical blade to obtain a very thin (~20 μm) and smooth preparation (~200 μm in diameter) (Fig. 1b). During the thinning process, repeatedly examine the preparation with a conventional fluorescence microscope until the dendrites and spines in the area of interest can be clearly visualized (Supplementary Fig. 2).

  ▲ CRITICAL STEP Hold the microsurgical blade at ~45° during thinning and take great care not to push the skull downwards against the brain surface or to break through the bone, as minor brain trauma or bleeding may potentially cause inflammation and disruption of neuronal structures. The thickness of the skull can be directly measured by imaging the skull auto-fluorescence with the TPLSM microscope. Periodic measurement of the skull thickness during thinning may help the novice user in preventing skull over-thinning.

  ! CAUTION It is important not to thin a large area (>300 μm in diameter) to < 15 μm in thickness, as it may cause disruption of neuronal structures and activation of immune cells as indicated by neurite blebbing, microglia process retraction and growth of epidural connective tissues, which greatly reduce the quality of subsequent images over days and may confound experimental results.

Mapping the imaging area for future relocation ● TIMING 5 min

• 12| In order to identify the same imaged area at a later time point, take a high quality picture of the brain vasculature with a CCD camera attached to a stereo dissecting microscope (Fig. 2a) or directly to the TPLSM setup.

• 13| Carefully move the prepared mouse with the entire skull immobilization device to the TPLSM microscope stage. Select a properly thinned area for imaging under a fluorescence microscope and carefully mark the selected area on the CCD brain vasculature map by observing the pattern of blood vessels adjacent to it (Fig. 2a).

TPLSM imaging of neuronal or glial structures ● TIMING 15 – 45 min

• 14| Tune the TPLSM to the appropriate wavelength (e.g., 920 nm for YFP and 890 nm for GFP). When possible, use high numerical aperture water-immersion objectives (e.g., ×60, 1.1 NA) to acquire images.

  ! CAUTION Mouse ACSF should be used at all times during imaging for objective immersion. If there is a sudden deterioration of imaging quality, check that the lens remains fully immersed in ACSF. Gradual leakage of ACSF is usually due to inadequate gluing of the scalp around the skull holder openings (Step 6).

• 15| For neuronal imaging, obtain a low magnification stack of fluorescently labeled neuronal processes (e.g., ×60 objective; 200 μm × 200 μm; 512 × 512 pixel; 2 μm step), which serves as a higher resolution map for accurate relocation of the same region at later time points in conjunction with the CCD brain vasculature map (Fig. 2b). If microglial imaging is being performed, then take a low-magnification (×10 air objective) stack of the rhodamine–dextran labeled vasculature and mark the selected area before switching to ×40/×60 objective, and make sure that objectives (×10 and ×40/×60) are co-centered.

  ? TROUBLESHOOTING

• 16| Without changing the position of the stage, take high-magnification images (e.g., 66.7 μm × 66.7 μm; 512 pixel × 512 pixel; 0.75 μm step: Fig. 2c) from the same area. The stack is typically taken within ~100 μm below the pial surface for spine imaging (Supplementary Movie 1) and within ~200 μm for microglia imaging.

  ! CAUTION We typically use laser intensities in the range of 10 to 30 mW (measured at the sample) to minimize phototoxicity.

  ? TROUBLESHOOTING

Recovery ● TIMING >1 h

• 17| Following imaging, detach the head immobilization device from the skull by gently pushing the skull away from the skull holder. When removed, the head immobilization device should take with it any remaining glue and be separated from the skin. If this does not occur, gently remove any remaining dried glue on the skull and/or detach the skin from the skull holder with a pair of forceps. Suture the scalp with 6-0 silk and leave the mouse in a separate cage until it is fully mobile. After recovery, return the mouse to its original housing cage until the next viewing.

  ■ PAUSE POINT Depending on the design of the experiment, re-imaging can be obtained hours to years after the first view.

Re-imaging ● TIMING 1 – 2 h

• 18| Repeat Steps 1–8. Find the thinned region based on the brain vasculature map (CCD for dendrite imaging and/or ×10 two-photon angiography for microglia). If the second view is within 1 week after the first view, then carefully remove the connective tissue that has re-grown on top of the thinned region using a microsurgical blade and check the image quality with the TPLSM microscope. If the image quality is poor (blurry features, high background fluorescence or significantly reduced depth of penetration), use a microsurgical blade to carefully shave the skull (as in Step 11) until a clear image can be obtained. If the second view is more than a week after the first view, repeat Steps 9–11.

  ▲ CRITICAL STEP The bone in the thinned area can re-grow substantially if the time interval between imaging sessions is longer than a week. The newly grown bone layer is optically less transparent for TPLSM imaging, so it is generally necessary to shave the bone slightly thinner than the previous imaging session.

  ? TROUBLESHOOTING

• 19| Find the previously imaged region under the fluorescence microscope. Align the region according to the low-magnification map under TPLSM, and then zoom in to high magnification to further align it. After the region is precisely aligned with the first view, take images as in Step 16.

Multiple-session imaging ● TIMING 1 – 2 h

• 20| For multiple imaging sessions, repeat Steps 18 and 19.

  ▲ CRITICAL STEP Although we have successfully imaged the same region with high optical clarity up to five times, the number and quality of imaging sessions is dependent upon the quality of the initial surgery, the intra-session interval and the experience of the operator. Although thinning the skull to the extremes (~15 μm) during the first surgery may provide excellent clarity in the initial imaging session, it will make repeated imaging sessions more difficult because of an increased propensity for compensatory bone re-growth. We recommend minimizing the perturbation to the system by limiting the number of imaging sessions and by thinning the skull only as much as necessary to gain an image clear enough to observe the desired structures.

Concluding remarks

Studies in the past several years have demonstrated that the transcranial TPLSM imaging approach is minimally invasive and technically reliable, allowing studies of detailed changes of neurons and glia over time in the living intact cortex. On the other hand, open-skull preparations involve skull removal and implantation of a glass window, which can cause significant neuroinflammation and induce artifactual changes in the cortex^7,20,31. Indeed, it has been found that large open-skull preparations (~5 mm in diameter) induce significant activation of microglia and astrocytes, initial dendritic spine loss after surgery and a dramatic increase in spine turnover⁷. These findings have been replicated by a more recent study (see Table 1 for details) even though the authors of this study claimed the opposite and suggested that open-skull preparation is excellent for chronic imaging if surgeries are done properly³¹. It is worth mentioning that potent immunomodulators such as dexamethasone are frequently used following open-skull preparations to reduce inflammation and achieve optimal imaging properties^{23,24,26–30}. These substances may have a significant impact on brain physiology, further confounding the studies of brain structure and function. For these reasons, we think that the claim that open-skull preparations are excellent for chronic imaging of cortical structures is assumptive and whether the results generated by such a preparation represent physiological conditions remains to be determined. We suggest that transcranial TPLSM imaging should be the method of choice for studying cortical structure and function in the living brain except for experiments that cannot be easily performed without removing the skull. In such cases, it is important to interpret the data carefully in the context of the confounding factors associated with open-skull preparations.