Brain atlas of the Mongolian gerbil (Meriones unguiculatus) in CT/MRI-aided stereotaxic coordinates

A new stereotaxic brain atlas of the Mongolian gerbil (Meriones unguiculatus), an important animal model in neurosciences, is presented. It combines high-quality histological material for identification of brain structures with reliable stereotaxic coordinates. The atlas consists of high-resolution images of frontal sections alternately stained for cell bodies (Nissl) and myelinated fibers (Gallyas) of 62 rostro-caudal levels at intervals of 350 μm. Brain structures were named according to the Paxinos nomenclature for rodents. The accuracy of the stereotaxic coordinate system was improved substantially by comparing and matching the series of histological sections to in vivo brain images of the gerbil obtained by magnetic resonance imaging (MRI). The skull outlines corresponding to the MR images were acquired using X-ray computerized tomography (CT) and were used to establish the relationship between coordinates of brain structures and skull. Landmarks such as lambda, bregma, ear canals and occipital crest can be used to line up skull and brain in standard atlas coordinates. An easily reproducible protocol allows sectioning of experimental brains in the standard frontal plane of the atlas.


Introduction
During the last decades, the Mongolian gerbil (Meriones unguiculatus, Thomas 1908) has emerged as an important animal model in neuroscience. It is a versatile and advantageous laboratory animal because of its robustness, its ease of handling and its reliable breeding under laboratory conditions. Virtually all sensory systems, especially the auditory system, are being intensively studied in gerbils, involving a wide range of neuroanatomical and neurophysiological approaches. Topics include development and plasticity as well as effects of aging. Research in the motor system and investigations of behavioral mechanisms, learning and memory and of transmitter systems use gerbils as model organism as well. Due to a peculiarity of the cerebral arteries (circle of Willis) in Mongolian gerbils, cerebral infarction can be induced in a controllable way and has made it a widely used model for cerebral ischemia. It is also a model animal for inherited epilepsy, hippocampal seizure and pathogenesis of CNS infections.
Despite a large body of literature related to the investigation of the gerbil brain, the availability of brain atlases published for this animal species is limited. To date, there are two stereotaxic atlases of the gerbil's brain. The 'Stereotaxic Atlas of the Mongolian Gerbil Brain' (Loskota et al. 1974) includes photographic montages of corresponding hemispheres of adjacent sections stained for myelinated fibers (Weil) and cell bodies (Nissl). Brain structures are outlined and labeled separately, while the neocortex is represented without subdivisions. The heavy shrinkage of the brain caused by the celloidin embedding technique was not corrected in the stereotaxic coordinates.
The brains used for the 'Stereotaxic Atlas of the Gerbil Brain' by Thiessen and Yahr (1977) were frozen and cut in a cryostat, which causes only little shrinkage and thus more reliably reproduces stereotaxic coordinates. This atlas incorporates the earlier 'Stereotaxic Atlas of the Hypothalamus' by Thiessen and Goar (1970). The atlas presents only schematic outlines of structures and does not provide illustrative material of the underlying Nissl-stained histological sections. In addition, the sectioning plane deviates from the conventional frontal plane in rodents perpendicular to the axis of the brain stem in both atlases.
Thus, the need for a new stereotaxic atlas of the gerbil brain that combines high-quality histological material to identify brain structures with reliable stereotaxic coordinates is evident. Brain sections are inevitably subject to distortions during tissue fixation and subsequent histological procedures (embedding, sectioning, staining and section mounting). Here, we improved the accuracy of the stereotaxic coordinate system substantially by comparing and matching the series of histological sections to in vivo brain images of the gerbil obtained by magnetic resonance imaging (MRI). Moreover, X-ray computerized tomography (CT) yielded the outlines of the skull corresponding to the MR images, which helped to establish the relationship between coordinates of brain structures and skull coordinates. This is essential for any stereotaxic procedure using landmarks on the skull to reliably target brain structures for recording, imaging, tracer or virus applications. The atlas can also be used effectively as a common reference base to collect and compare positional data from any kind of research in the gerbil brain.

Methods Animals
Twenty-one young adult male Mongolian gerbils (Meriones unguiculatus) at the age of 4 months and weighing between 80 and 100 g were used for this study. Out of them, brains of seven animals were processed for cyto-and myeloarchitectonic features. Six other brains were additionally processed for chemo-and immunoarchitecture to support identification of anatomical structures. This mate-rial is not included in the atlas and will be published separately. Overall 10 CT scans of skulls and a total of 13 MR brain scans were performed in various combinations.
All experiments were in agreement with the NIH Guide for the Care and Use of Laboratory Animals (2011) and the guidelines of the European Communities Council Directive (86/609/EEC) and approved by the animal care committee of Sachsen-Anhalt, Germany.

MR imaging
Animals were anesthetized with isoflurane (1.0-1.5 % in 1:1 O 2 :N 2 volume ratio) and fixed with bite bars in a headholder to reduce motion artifacts. MR scans were performed on a Bruker Biospec 47/20 scanner (Bruker Biospin GmbH, Rheinstetten, Germany) at 4.7 T (free bore of 20 cm) equipped with a BGA 09 (400 mT/m) gradient system. A 35 mm Litzcage small animal imaging system (DotyScientific Inc., Colombus, SC, USA) was used for radio frequency (RF) excitation and signal reception. Two days before MRI measurements, animals were injected subcutaneously with an aqueous solution containing 1 lmol/g MnCl 2 (manganese enhanced MRI: ME-MRI). A data set of T1-weighted images was obtained using a 3D MDEFT (modified driven equilibrium Fourier transform) pulse sequence with the following parameters: repetition time 13.6 ms; echo time 4.3 ms; flip angle 20°; field of view 30 9 30 mm 2 ; matrix 256 9 256 (yielding a nominal in plane resolution of 117 9 117 lm 2 ); standard frontal orientation; slice thickness 350 lm; 20 averages. Images were reconstructed using Bruker ParaVision 4.0 (Bruker Biospin GmbH, Rheinstetten, Germany) and exported as raw images as well as in DICOM format. The open source program AMIDE (amide.exe 1.0.4, Ó Andreas Loening, http://amide.sourceforge.net/) was used to align CT and MR scans and to extract images shown in the atlas.

Histology
Animals were anesthetized with a lethal dose of ketamine (40 mg/100 g body weight, i.p.) and xylazine (2 mg/100 g body weight, i.p.). When a deep anesthetic state marked by a complete loss of the flexor reflex at all limbs was reached, animals were perfused transcardially with 20 mL of phosphate buffered saline (0.1 M PBS, pH 7.4) supplemented with 0.1 % heparin followed by 200 mL of 4 % PFA (in 0.05 M PBS, pH 7.4). The brains were postfixed in the skull with 4 % PFA (in 0.05 M PBS, pH 7.4) at 4°C for at least 7 days before removal to best preserve the brain shape.
Brains were cryo-protected in 22.5 % sucrose in PBS (0.05 M, pH 7.4) overnight and cut in a cryostat (LEICA CM 3050S) into four series of 40 lm thick frontal sections. The sections were directly mounted on gelatine-coated slides and dried overnight. Alternating section series were stained on-slide either for cells (Nissl) or for myelin (Gallyas 1979). The brains additionally processed for chemo-and immunoarchitecture were stained for cytochrome oxidase, acetylcholine-esterase (AChE), NADPHdiaphorase, calcium-binding proteins (parvalbumin, calbindin and calretinin) and neurofilament protein (SMI-32) in various combinations. Sections were imaged with a virtual slide microscope (VS120 S1, Olympus BX61VST, Olympus-Deutschland, Hamburg, Germany) at 109 magnification using the proprietary software dotSlide Ò (Olympus).

Atlas coordinate system
The coordinate system of the brain atlas is based on the conventional definition of anatomical sectioning planes in rodents. Frontal sections are perpendicular to the brainstem axis, which in the Mongolian gerbil is also parallel to the plane defined by the most dorsal points of cerebrum and cerebellum ( Fig. 1). This plane is therefore chosen as origin for the dorsoventral dimension with negative values in ventral direction. The lateral dimension is zeroed to the midsagittal plane with negative values towards the right and positive values towards the left side. The anterior to posterior coordinates of the atlas are given for different origins (bregma, lambda, interaural line and occipital crest as skull landmarks) and are valid for the skull in standard atlas orientation.
The frontal sectioning plane was implemented by a standardized embedding procedure using an acrylic glass box (Fig. 1). Each brain was oriented within the box so that the brainstem axis ( Fig. 1bs) was parallel to the base of the box and the midsagittal plane lined up with the long axis of the box. Note that in this orientation the plane through the highest point of cerebellum and cerebrum (Fig. 1cc) is parallel to the base of the box and can therefore also be used to align the brain. The brain was stabilized in this orientation by adjustable supporting needles protruding from the bottom and from a bracket on top of the box. The volume around the brain was filled with embedding medium, namely a freshly prepared mixture of gelatin-albumin-glutaraldehyde. After 2-3 min, this mixture had hardened and the block was taken out of the box. Subsequently, the block was shock frozen in dry ice and mounted with its hind surface on the cutting platform of the cryostat. Due to the prior orientation within the box, the sectioning plane was now perpendicular to the long axis of the block and therefore also perpendicular to the brainstem axis and the horizontal plane through the highest cerebellar and cerebral points.

Stereotaxic reference system
In rats and mice, the connecting line through lambda and bregma coincides with that through lambda and occipital crest and is used as a horizontal guideline to align the in vivo brain in the classical planes (Paxinos and Watson Fig. 1 View of fixed gerbil brain positioned for embedding. In the lower part of the figure, the brain is shown in the acrylic glass box used for embedding (rectangular block volume indicated by fine dotted lines). The brain is positioned on three pins (one is hidden by the left front pin) protruding from the base so that the plane defined by the most dorsal elevation of cerebrum and cerebellum (cc) as well as the brainstem axis (bs) are aligned parallel to the base. The anterior and posterior surfaces of the embedding block define the frontal sectioning plane (sp) perpendicular to cc and bs. A pin protruding from a bracket over the side walls of the box (only partly shown) prevents the brain from being washed off when the embedding medium is poured into the box 2007; Paxinos and Franklin 2001). In the Mongolian gerbil, the line linking lambda and bregma deviates from that linking lambda and occipital crest (Fig. 2, lower panel) and should, therefore, not be used as horizontal guideline to position the gerbil skull and brain in the atlas coordinate system. A horizontal adjustment of the skull along the line between lambda and occipital crest (Fig. 2, horizontal solid line) results in the best approximation to the atlas orientation (Fig. 2, dotted line) and is recommended as standard orientation.

Selection of atlas series
The atlas series of histological sections was selected according to the following criteria: • the entire series, alternately stained for cell bodies (Nissl) and myelin (Gallyas), had to show good staining quality and tissue preservation • the atlas series had to match the MR scan of an average-sized brain, and relative distances of indicative structures of the brain had to show congruency with the distances in the available MR scans.
The following structures that could clearly be determined both in histological sections and in MR slices were used as 'indicative structures' (Fig. 3): the rostral beginning of neocortex (1), the crossing of the anterior commissure (2), the distinct appearance of the medial habenular nucleus (3), the end of the superior colliculus concurrent with the middle of the inferior colliculus (4) and the end of the cerebellum (5). To judge brain size and to probe the consistency of individual histology series, the distances between indicative brain structures and the rostral pole of neocortex were evaluated and compared to the corresponding median distances in 13 MR scans ( Fig. 3; Table 1).
The MR series that corresponded best to the median values was chosen as 'ATLAS MRI'. The same distance measurements were performed in seven high-quality histological series. The series that corresponded best to the atlas MRI median values was designated as 'ATLAS histology series'. Table 1 shows the conformance of the atlas histology series with the atlas MR scan and the median values of MRI series.
CT scans of the skull provide the interface to the brain coordinate system in vivo. Therefore, the available CT scans were overlaid to the atlas MRI. The CT scan matching best was chosen as 'ATLAS CT' series. For all CT scans the distances between bregma and the skull landmarks lambda, interaural line and occipital crest were calculated ( Table 2). The comparison across animals corresponded well to the values of the atlas CT scan.

Preparation of images and plates
For each 350 lm thick slice of the atlas MR series a corresponding Nissl-stained section of the atlas series was selected and grouped with the adjacent myelin-stained section to represent one of the 62 rostro-caudal levels ( Fig. 4). Usually, every forth Nissl-stained section fitted best to the subsequent MR slice, which corresponded to a distance of 320 lm between the matching Nissl-stained sections. The 30 lm difference between the MR slices and the Nissl-stained sections can be explained by the shrinkage of the atlas brain due to histological processing, mainly fixation. This shrinkage is in the range of 8-10 % generally observed for cryo-protected frozen-cut brains with PFA fixation (4 %). Contrast and brightness of the images of the sections were corrected with Photoshop (CS6, Adobe Systems, San Jose, CA, USA), and distortions due to histological processing were compensated by slightly transforming the sections to optimize the congruency of anatomical structures between histological sections and MR images. Images were arranged in the atlas coordinate frame using CorelDraw graphics suite version X6 or X7 (Corel Corporation, Ottawa, ON, Canada). MR and CT

S4
Brain Struct Funct (2016) 221 (Suppl 1):S1-S272 Fig. 3 Indicative structures in histological and MRI brain series. The following structures were used (from rostral to caudal): beginning of neocortex (1), midline crossing of anterior commissure (2), distinct appearance of medial habenular nucleus (3), end of the superior colliculus (concurrent with the middle of the inferior colliculus) and (4) end of the cerebellum (5). Montages combine CT and MR scans and half of the corresponding Nissl-stained section. The anteriorposterior location of the corresponding atlas plates is indicated by dotted lines and respective numbers in the central brain image Distances of skull landmarks lambda, interaural line and occipital crest are evaluated relative to bregma for the atlas CT scan (column: ATLAS CT) and as median distance values across all CT scans (columns: All CT). The range of values around the median is indicated by the minimum and maximum distance values taken from ten CT scans images were adjusted according to the definition of the atlas coordinate system in 62 plates and reflect the in vivo orientation of the brain and skull. The images of cell-and myelin-stained sections were inserted in line with the corresponding MR image. The anterior-posterior coordinates of the plates are indicated relative to bregma, lambda, interaural line and the occipital crest. All outlines were drawn in CorelDraw on the base of the Nissl-stained section of each atlas plate. The structural boundaries seen in the corresponding myelin-stained section generally correlate well with these outlines.

Anatomical structures, nomenclature and abbreviations
Anatomical structures were identified on the basis of cytoand myeloarchitecture and their relative location. For comparison we mainly used the published atlases of the Mongolian gerbil brain (Loskota et al. 1974;Thiessen and Yahr 1977), the atlases and books for rat brain of Paxinos, Swanson and Zilles (Paxinos 1995(Paxinos , 2004Paxinos and Watson 2007;Paxinos et al. 2009;Swanson 1992Swanson , 2004Zilles 1985) and for mouse brain (Paxinos and Franklin 2001;Dong 2008;Franklin and Paxinos 2008;Watson and Paxinos 2010;Watson et al. 2012). Brain series stained for chemoarchitectonic markers were consulted to support the structural identification. Unfortunately, no unified neuroanatomical nomenclature exists to date (Swanson 2015). Therefore, we decided to use the widely accepted Paxinos nomenclature and abbreviations for naming structures.
Auditory midbrain and brainstem nuclei for which gerbil specific terms were already established (Budinger et al. 2000Mylius et al. 2013;Radtke-Schuller et al. 2015) were labeled according to these studies.

Practical hints
Sectioning in atlas coordinates: It is also possible to section the brain in the standard atlas plane without the above described embedding procedure. In this case, the brain is positioned upside down on a flat surface so that it is seated with the cerebellum and cerebrum on the base. Then, part of the brain is cut off perpendicular to the base to create a surface for mounting the brain's portion of interest on the cryostat platform. By subsequent sectioning of the brain parallel to this cutting surface the resulting sections correspond best to the frontal plane of the atlas. Stereotaxic procedure: In addition to traditional landmarks and reference points such as lambda, bregma and interaural line, we recommend the occipital crest ( Fig. 2) for anterior-posterior reference and adjustment of the skull in vivo. The traditional landmarks are often difficult to discern, show individual variations and cannot be accessed in some experimental approaches (e.g., interaural coordinates in auditory research where ear bars are avoided). In general, a higher precision of in vivo positioning of the skull can be achieved by using the specific pattern of skull profiles instead of single reference points [for profile oriented stereotaxic procedure see Schuller et al. (1986)].

Index of Structures
The structures are listed in alphabetical order followed by their abbreviation and the plate number(s) of occurrence

Index of Abbreviations
The abbreviations are listed in alphabetical order followed by the name of the structure and the plate number(s) of occurrence.