Registration and Alignment Between in vivo Functional and Cytoarchitectonic Maps of Mouse Visual Cortex

This protocol describes a method for registration of in vivo cortical retinotopic map with cytochrome c oxidase (CO) labeled architectonic maps of the same mouse brain through the alignment of vascular fiducials. By recording surface blood vessel pattern and sequential alignment at each step, this method overcomes the challenge imposed by tissue distortion during perfusion, mounting, sectioning and histology procedures. This method can also be generalized to register and align other types of in vivo functional maps like ocular dominance map and spatial/temporal frequency tuning map with various anatomical maps of mouse cortex.


Background
The mouse visual cortex can be segregated into functionally distinct visual areas by in vivo retinotopic mapping (Marshel et al., 2011;Garrett et al., 2014;Zhuang et al., 2017) or by neuronal track-tracing techniques aided by architectonic structures (Olavarria and Montero, 1989; Wang and Burkhalter, 2007). These different visual areas have distinct response properties and corticocortical connectivity (Andermann et al., 2011;Marshel et al., 2011;Roth et al., 2012;Wang et al., 2011 and. These results suggest that mouse visual areas form segregated visual streams processing different types of visual information (Murakami et al., 2017;Smith et al., 2017). Studying the mouse visual system in the context of visual area maps is essential to understanding the organization of visual cortex. However, although the functional maps and structure maps are broadly similar, the two maps have been shown not matching perfectly . For example, the primary visual cortex (V1) appears as an upward triangle in both maps, but the lateral edge of V1 in retinotopic map can be up to 300 micrometers more medial than that in anatomical map .
Since the smallest visual areas in mouse cortex are only a few hundred micrometers wide, ignoring this mismatch will potentially bias our interpretation of visual area functions.
This article is distributed under the terms of the Creative Commons Attribution License (CC BY 4.0). Furthermore, both types of maps vary significantly across different individuals. Therefore, to study the functions of identified visual areas, it is important to be able to reliably generate and compare functional and anatomical maps in the same animal. However, the tissue distortion during perfusion, mounting, sectioning and histological procedure makes it difficult to directly compare functional maps recorded in vivo with anatomical maps

11.
Peristaltic pump (Harvard Apparatus, model: MA1-55-7766) Name it image B. Image A and image B should be perfectly co-registered by nature given the imaging optical axis is perpendicular to the cranial window ( Figure 2A/B).

B.
Perfusion and cortex flattening (modified from Wang and Burkhalter, 2007)

a.
Saline wash 10 ml/min for 100 ml.

c.
Wait for 5 min for DyLight lectin to adsorb to tissue.

2.
With brain within the skull, acquire fluorescence image of cranial window (filter setting: 655/670 nm). The fiducial marks made in Procedure A should be visible. Name it image C ( Figure 2C).

3.
Collect brain tissue. Since the animal was perfused by 1% PFA, the brain tissue will be relatively soft for cortex flattening. Be careful not to make any damage.

4.
Optional: Acquire bright field and fluorescence images (filter setting: 655/670 nm) of surface vasculature of the whole brain using a dissecting scope. The major surface blood vessels should be visible in the bright field image (can be registered with image A) and the DyLight labeled blood vessels should be visible in the fluorescence image (can be registered with image D).

5.
Isolate the cortical sheet of windowed hemisphere (the procedure can be done in a Petri dish sitting on ice). Carefully keep track of the orientation of cortical sheet. For video guidance, please see this EJN video protocol (made by Hoey Sarah, Universität Zürich): http:// www.ejnnews.org/video-protocol-isolation-adult-mouse-hippocampi/.

a.
Separate the two hemispheres of the brain with a razor blade. Keep the windowed hemisphere and discard the other hemisphere.

b.
Cut off olfactory bulb with a razor blade.

c.
Cut off brain tissue posterior to the neocortex (this include cerebellum, posterior midbrain and hind brain) with a razor blade.

d.
From the medial side, gently pull out the thalamus, septum and striatum by using a spatula. Cut off these subcortical tissue.

e.
Gently flip the hippocampal formation out and then separate it from cortex using a spatula.

6.
Flatten the isolated cortical sheet on a slide glass with the pia surface against the glass. Cover the other side of the cortical sheet with a piece of sponge. Cover the sponge with another piece of the slide glass. Space the two slides with two coins (we used United State dimes with thickness of 1.35 mm). Clip the both sides of slides ( Figure 1).

7.
Immerse the 'sandwich' made in Step B6 in 1% PFA overnight (in a Petri dish in 4 °C fridge). Make sure the whole 'sandwich' is fully submerged.

10.
Remove the clips and remove the cortical sheet. Cut the outer edge of the sheet so that it is in an asymmetric shape and the orientation of the cortex (anterior, posterior, medial and lateral) is easy to identify.

11.
Take a fluorescence vasculature image (filter setting: 655/670 nm) of the flattened and cut cortex sheet before sectioning. Name it image D ( Figure 2D).

C. Tangential sectioning of flattened cortical sheet
Note: This is the crucial step and do it with extra caution.

1.
Sufficiently cool the platform with dry ice before mounting (~10 min) the tissue and keep the platform frozen (dry ice always presented in the wells at the both end of the platform) for the whole sectioning process.

2.
Embed flattened cortex sheet in OCT with cortical surface facing up on microtome platform.

3.
Quickly put one glass slide on top of the cortex sheet before it freezes. Apply gentle pressure with fingers on the slides so that it flattens the tissue surface until it freezes.

6.
Take brightfield images of the sections (image series E, for example of a section across layer 4 in this series see Figure 2E, showing architectonic labeling of primary sensory cortices and retrosplenial cortex).

1.
Adjust the contrast and pixel resolution of images A, C, D, E so that the vasculature and cytoarchitectonic features are prominent and they all have roughly same pixel size.

2.
Image B should go through same transformations as image A, so that they remain co-registered (Figure 2A/B).

Use in vivo images (image A/B) as reference and align other images
progressively. Align image C to image A/B → align image D to image A/B/C → align image series E to image A/B/C/D.

a.
Use non-linear transformation function (inside the TrakEM2 plugin) to align vasculature fiducials between adjacent image layers.

b.
Use surface vasculature to align images A, C, D ( Figure 2F).

c.
Use the section outline and ascending/descending vessel cross sections to align image D and image series E ( Figure 2G).

5.
Once all images are co-registered, hide all the intermediate image layers and superimpose image B and the image showing the most prominent cytoarchitectonic features in image series E. The overlay image allows a direct comparison between the in vivo functional map and the CO labeled architectonic map ( Figure 2H).

1.
The duration of Steps B2-B6 (after perfusion to flattening) should be as short as possible, longer delays may cause the brain to harden and affect the result of flattening.

2.
In images C (recorded in Step B2) and D (recorded in Step B11), only a subset of the cortical surface vasculature in image A (recorded in Procedure A) will be labeled.

3.
In image C recorded in Step B11, same cortical surface vasculature as that in image B should be visible.

4.
When rinsing the sections during CO staining, the rotation speed of the shaker should be less than 20 rpm to avoid displacing sections from the slide.

7.
The resolution of image D, image E and image series F should be high enough that the cross sections of ascending/descending vessels are visible (we used ~3 µm/pixel).

8.
Sometimes inverting the contrast of some images during image alignment may help visualize the fiducials across images.

9.
For image alignment, any image analysis software allowing the use of independent layers and nonlinear/warping transformations may be used; however, a suitable and widely available software is the TrakEM2 function