Neurochemical atlas of the cat spinal cord

Abstract

The spinal cord is a complex heterogeneous structure, which provides multiple vital functions. The precise surgical access to the spinal regions of interest requires precise schemes for the spinal cord structure and the spatial relation between the spinal cord and the vertebrae. One way to obtain such information is a combined anatomical and morphological spinal cord atlas. One of the widely used models for the investigation of spinal cord functions is a cat. We create a single cell-resolution spinal cord atlas of the cat using a variety of neurochemical markers [antibodies to NeuN, choline acetyltransferase, calbindin 28 kDa, calretinin, parvalbumin, and non-phosphorylated heavy-chain neurofilaments (SMI-32 antibody)] allowing to visualize several spinal neuronal populations. In parallel, we present a map of the spatial relation between the spinal cord and the vertebrae for the entire length of the spinal cord.

Introduction

The spinal cord is the most ancient part of the central nervous system, which is responsible for the autonomic control of the visceral organs and for the sensorimotor control of the muscle activity (Watson and Kayalioglu, 2009). Spinal cord injuries and diseases can be accompanied by corresponding violations or shutdowns of these functions (McDonald and Sadowsky, 2002). Treatments for the rehabilitation of these disorders are based on complex experimental data from neuromorphological, neurochemical, neurophysiological, and genetic studies.

Currently, rodents are the main experimental object in neurosciences (Ellenbroek and Youn, 2016). However, some features of rodents (e.g., small size) may be a possible source for particular limitation in using this model in neurophysiological studies. One of the limitations is increased requirement for fine manipulations during surgical procedures, which may lead to an additional increase in the sample of experimental objects. In this case, larger animals (e.g., cats, dogs, and mini-pigs) are more suitable for these experiments. Since the seminal studies of C.S. Sherrington and T.G. Brown, who elaborated the hypothesis for the spinal central pattern generators (Sherrington, 1910; Brown, 1911, 1913), the cat has become a widely used model for the investigation of spinal cord functions especially the locomotor control. Data obtained using this animal model allowed to discover a number of mechanisms of spinal cord and brainstem control for locomotion (Shik et al., 1969) and identify special triggering zones in spinal cord, epidural or subdural, stimulation of which induces coordinated locomotion (Iwahara et al., 1992; Musienko et al., 2009). Later, these data were incorporated into the neurorehabilitation approaches of spinal function recovery (Gerasimenko et al., 2018; Gill et al., 2018). The large size and high recoverability (Kuhtz-Buschbeck et al., 1996; Martinez et al., 2011) of the cat allows to use many invasive (e.g., implanted electromyographic and extracellular recording sensors) and non-invasive sensors and stimulating devices, and perform fine surgical procedures for the acute and chronic studies on both locomotor and visceral control (Musienko et al., 2012; Merkulyeva et al., 2019; Afanasenkau et al., 2020). The precise neurochemical mapping for spinal neuronal populations, combining with stereotaxic coordinates can be important for the design of a neurophysiological experiment and subsequent data interpretation. Currently, such spinal cord atlases were created for rodents (Watson et al., 2009, 2021; Sengul et al., 2012) and primates (Tokuno et al., 2011; Sengul et al., 2012; Watson et al., 2021) but not cats.

Two major classes of neurons can be distinguished within the spinal cord gray matter, i.e., motoneurones and interneurons. They can be visualized using specific neurochemical markers.

• (1)

  Choline acetyltransferase (ChAT) is a specific marker for motoneuronal pools and neurons responsible for visceral control; ChAT catalyzes the biosynthesis of the neurotransmitter acetylcholine (Wu and Hersh, 2008) and is detected in the entire population of motoneurons (small, medium, and large-sized), in their fibers, and neuropil (Barber et al., 1984); ChAT is also found in structures responsible for the visceral control, i.e., intermediolateral nucleus (IML), intercalated nuclei (IC), intermediomedial nucleus (IMM), sacral parasympathetic nucleus (SPN), and dorsal gray commissure (DGC) (Barber et al., 1984; Phelps et al., 1984).

• (2)

  Ca²⁺-binding proteins calbindin 28 kDa (CB), calretinin (CR), and parvalbumin (PV) have been used to label predominantly interneuronal subpopulations but also visceral neurons (Ren and Ruda, 1994). These proteins are more well-known members of Ca²⁺-binding proteins, a heterogeneous group of proteins that participate in numerous cellular functions (e.g., Ca²⁺ homeostasis and Ca²⁺ signaling pathways) (Yáñez et al., 2012).

  Calbindin labels Renshaw cells (Sanna et al., 1993; Alvarez et al., 1999; Stepien et al., 2010) and several visceral structures, e.g., IML (Zhang et al., 1990; Grkovic and Anderson, 1997; Moiseev et al., 2019), IMM (Zhang et al., 1990), IC (Porseva, 2015), DGC (Ren and Ruda, 1994; Grundy et al., 2019), SPN (Menétrey et al., 1992), and Onuf’s nucleus (Ince et al., 1993; Alexianu et al., 1994).

  It has been proposed that CR labels the population of excitatory interneurons located in the dorsal horns (Smith et al., 2019; Peirs et al., 2021). At the same time, CR is detected in all spinal laminae without clear preference and labels subpopulation of the small interneurons of the Clarke’s nuclei (CN) (Veshchitskii et al., 2021). CR also labels visceral neurons of the IML (Edwards et al., 1996) and DGC (Ren and Ruda, 1994; Merkulyeva et al., 2019).

  Parvalbumin is predominantly associated with the spinal proprioceptive system marking large-caliber afferent fibers and CN (Zhang et al., 1990; Clowry et al., 1997). PV is expressed in V1 interneuron population including Renshaw cells and Ia inhibitory interneurons (Alvarez et al., 2005; Siembab et al., 2010).

• (3)

  SMI-32 antibody is a marker for non-phosphorylated heavy-chain neurofilaments (Sternberger and Sternberger, 1983). SMI-32 labels neurons with large soma and fast large-caliber axons (Bickford et al., 1998); in the spinal cord, these features are mainly inherent to motoneurons (Tsang et al., 2000) but also to some ascending tract neurons, e.g., large cells of the CN (the main origin of the dorsal spinocerebellar tract) (Tsang et al., 2000; Clowry et al., 2005; Merkul’eva et al., 2017).

To reveal the total population of the spinal neurons, a NeuN antibody was used. The main advantage of this protein is its absence in non-nervous tissues and in non-neuronal cells of the nervous tissue (Gusel’nikova and Korzhevskiy, 2015).

Therefore, the main aim of this study is to create a high-resolution neurochemical atlas of the cat spinal cord depicting multiple cellular populations throughout the entire spinal cord. This atlas can be used for the setting up of new neuroanatomical and neurophysiological experiments and for the interpretation of the already obtained neurophysiological data.

Materials and methods

Subjects

Experiments were performed in accordance with requirements of Council Directive 2010/63EU of the European Parliament on protection of animals used in experimental and other scientific purposes, and with the approval of the Ethics Commission of the Pavlov Institute of Physiology (Protocol #30/01/2020).

Spinal cord of the one intact adult domestic female cat (weighing 3.5 kg) was used for the creation of the images for the cat spinal cord atlas. Preliminary we verified all histological and immunohistochemical procedures with spinal cord, and all staining patterns using multiple animals (Merkulyeva et al., 2016, 2019; Shkorbatova et al., 2019; Veshchitskii et al., 2021). Immunohistochemical patterns are completely coincide with the present data.

According to the “3R” rule, the brain of the animal taken for the atlas was used for other studies dedicated to visual system development (Merkulyeva et al., 2018, 2020; Mikhalkin and Merkulyeva, 2021).

Perfusion and dissection

The animal was deeply anesthetized with a mixture of Zoletil (Virbac, France; 20 mg/kg) and Xyla (Interchemie werken “De Adelaar” BV, Netherlands; 2 mg/kg), intramuscularly. Heparin (Endopharm, Russia; 0.5 ml/kg) was injected intramuscularly 10 min before the start of perfusion. Then the animal was transcardially perfused with 0.9% NaCl (2 liters, with Heparin, 0.5 ml/l) followed by 4% paraformaldehyde (2 L). Then the animal was placed in a prone position, the spinal cord was exposed by removing the vertebral arches and the dorsal part of the dura mater. The distances between the caudal-most parts of the dorsal rootlet attachment zones, connected to the neighboring dorsal root ganglia were measured as spinal cord segments (Shkorbatova et al., 2019). The lengths of all vertebrae and the positions of the segments in relation to the vertebrae were also thoroughly documented. Then, the spinal cord was cut to segments and each of them was replaced in 10, 20, and 30% sucrose, for cryoprotection, and thereafter cut into 50 μm transverse slices on a freezing microtome (Reichert, Austria). Storage of the slices was in 0.1 M PBS with 0.1% NaN₃ at +4°C.

Algorithm of choosing the slices for the atlas

The spinal cord is subdivided into regions (i.e., C, T, L, S, and Co) and corresponding segments; however, some peculiarities of structural organization can be observed even within the segment (Vanderhorst and Holstege, 1997). Therefore, for all segments, 2–3 regions of interest were used for the atlas. For the more homogeneous segments (T2–L3), two regions of interest were used, i.e., from the rostral (interrootlet zone) and caudal (rootlet zone) parts. For highly variable segments (C1–T1 and L4–Co2) with weakly expressed or absent interrootlet zone, three equally spaced regions of interest were used (i.e., rostral, middle, and caudal). Six slices were chosen for immunohistochemical processing (Figures 1B,C).

FIGURE 1

Immunohistochemical processing

Slices were processed as free floating. Between all procedures, the slices were washed in 0.01M PBS. Antigens were unmasked in 1% NaBH₄, an endogenous peroxidase activity was blocked by 0.3% H₂O₂, unspecific immunoreactivity was lowered by incubation in 5% normal goat serum (NGS, Vector Labs). Thereafter slices were incubated for 70 h in primary antibody (Table 1) with 0.1% NaN₃. Then the slices were incubated for 1 day in a biotinylated secondary antibody (Table 1) with 0.1% NaN₃. Thereafter slices were subsequently processed using an avidin-biotin horseradish-peroxidase complex (ABC Elite system, Vector Laboratories) and diaminobenzidine (DAB)-NiCl-Í₂Î₂ reaction. After washing in distH₂O, slices were mounted, dehydrated, cleared and placed under coverslips in Bio Mount HM (Bio-Optica Milano, Italy). To control the specificity of antibodies, we previously verified it using Western blot (Merkulyeva et al., 2022).

Antibody characterization

The antibody against NeuN (Sigma-Aldrich, Cat.# MAB377, RRID: AB_2298772) is a mouse monoclonal IgG1 recognizing 46 and 48 kDa bands on Western blots of a cat brain (Merkulyeva et al., 2022). The immunostaining of sections through the cat spinal cord (used at 1:5000) produced a pattern of NeuN labeling that was identical to previous descriptions (Veshchitskii et al., 2022), the total population of exclusively neuronal cells.

Calbindin antibody (Sigma, Cat.# C9848, RRID: AB_476894) is mouse monoclonal IgG1, recognizing s single 28 kDa band on Western blots of a cat brain (Merkulyeva et al., 2022). The immunostaining of sections through the spinal cord (used at 1:3000) produced a pattern of CB labeling that was identical to previous descriptions on cat spinal cord (Merkulyeva et al., 2016), specific neuronal bands in the dorsal horns and clusters of cells in intermediate gray matter of spinal enlargements. The general pattern of immunopositive neurons corresponds to the data on other species with using of other antibodies to calbindin (Ren and Ruda, 1994).

Calretinin antibody (Sigma-Aldrich, Cat.# AB5054, RRID: AB_2068506) is a rabbit polyclonal IgG recognizing 29 kDa band on Western blots of a cat brain (Merkulyeva et al., 2022). In the cat spinal cord CR is present by many immunopositive neurons throughout the gray matter in the same structures as in previous our works with this antibody (Veshchitskii et al., 2021) and works of other authors with other calretinin antibody (Anelli and Heckman, 2005).

Parvalbumin antibody (Abcam, Cat.# ab11427, RRID: AB_298032) is a rabbit polyclonal IgG recognizing 12 kDa band on Western blots of a cat brain (Merkulyeva et al., 2022). In the spinal cord as in other works PV is present in specific pre-motor neural populations and fibers related the proprioceptive system (Ren and Ruda, 1994; Anelli and Heckman, 2005).

ChAT antibody (Sigma-Aldrich, Cat.# AB144P, RRID: AB_2079751) is a goat polyclonal IgG recognizing 70/74 kDa band on Western blots of a mouse brain (manufacturer’s datasheet). No papers concerning these antibodies on cats were obtained. But data obtained using rodents illustrate the same results, with specific staining for motoneurons and visceral neurons (Mesnage et al., 2011).

SMI-32 antibody (BioLegend, Cat.# SMI-32P, RRID: AB_2564642) is a mouse monoclonal IgG1 to non-phosphorylated heavy-chain neurofilaments recognizing ∼200 kDa band on Western blots of a cat brain (Burnat et al., 2012). These neurofilaments are labeled in neurons with large soma (for example, motoneurons) (Tsang et al., 2000) and fast large-caliber axons (Bickford et al., 1998).

Image processing

Digital images of the slices with the identified antigens were obtained using a computer setup equipped with an Olympus CX31 light microscope (Olympus Corporation, Japan; 4 × and 10 × objectives), digiCamControl software package distributed as an open source, and Nikon camera (D3200, Nikon Corporation, Japan). Image background processing was performed using the FlatBFv2 plugin in a free Fiji software package (Schindelin et al., 2012).

Morphometry

The morphometric characteristics of the spinal cord were measured only in unstained wet slices, that lead to the only one step of tissue shrinkage—due to the fixation (Watson et al., 2009). As for the immunohistochemical processing, it led to the additional “shrinkage.” Therefore, it was not possible to obtain spinal cord statistical parameters that would be similar to those of an alive animal. Thus, for each region of interest for all segments, 5 adjacent transverse sections were used in morphometric analysis (a total of 420 slices). All slices were photographed at 4× magnification. Using the Fiji software package, the following morphometric parameters were calculated in these images, i.e., general area of transverse slice, area of gray matter with central canal, central canal area, area of white matter folds (formed as a result of different levels of “shrinkage” of white and gray matter during fixation), height and width of the spinal cord, distances from the central canal to the dorsal, as well as ventral and lateral borders of the white matter (Figure 1D). Based on these parameters, the area of gray and white matter was calculated according to the formulas presented in Figure 1D. The average areas of gray and white matter ± standard deviation (SD) and SD of distances from the central canal to the dorsal, ventral, and lateral borders of the white matter were added to the scheme of each segment.

Results

Spatial relationship between the spinal segments and vertebrae and morphometry

The first objective was to obtain a spatial relationship between the vertebrae (V) as the reference point and the spinal segment localized below it. This spatial relationship was obtained for all cervical (C), thoracic (T), lumbar (L), and sacral (S) segments and for the first and second coccygeal (Co) segments. The total length of the segments, their interrootlet and rootlet zones, and the vertebrae were also assessed. The obtained results and a schematic representation of the spatial relationship between the segments and the vertebrae for the cat used in the atlas are shown in Figure 2A and in Table 2.

FIGURE 2

It is well-known that the positions of the spinal cord segments inside the vertebral canal do not correspond to the location of eponymous vertebrae owing to different rate of their growth during ontogeny (Maierl and Liebich, 1998), which is known as the “spinal cord ascension” (Streeter, 1919). However, ascension is a complicated process, accompanied by descension of some spinal cord regions during the formation of enlargements (Maierl and Liebich, 1998; Shkorbatova et al., 2022). Thus, in the cat, the C1 segment is not localized strictly in VC1 but extends rostrally to the occipital bone. The caudal part of C1 and entire C2 segments occupy VC1 and the rostral part of VC2 (the region of the odontoid process and atlantoaxial joint) (Figure 2A). The long segment C3 occupies almost all length of VC2 and a rostral third of VC3. Such shift (in one third) between segments and vertebrae position is maintained up to the C7 segment, which, owing to its short size, completely fits into the VC6. The last cervical vertebra (VC7) almost completely includes two segments at once, i.e., C8 and partly T1.

The VT1 contains the entire segment T2. Beginning with VT2, the following vertebrae and segments have similar sizes, and a stable shift of segments rostrally in relation to the eponymous vertebra by a half or a whole segment appears for mid-thoracic division. The gradual return of the segment to the eponymous vertebra as a result of an increase in the length of the segment relative to the length of the vertebra begins in T10 segment. The T11 segment is already mostly localized in VT11, and this pattern of localization is observed up to the L3 segment. Whereas the caudal parts of T12–L1 segments shift caudally in relation to their eponymous vertebrae, demonstrating a “descension”; L2 and L3 segments are localized in eponymous vertebrae. For the following segments, the rostral shift of the segment in relation to the eponymous vertebra appears, and it becomes stronger the more caudally the segment is located. Thus, the spinal cord of the cat ends in the lumbar vertebra. The L4 segment and the rostral half of the L5 segment are located in VL4; the caudal half of L5 segment, segments L6, L7, and the most rostral part of S1 segment share the length of VL5; the main part of S1 segment and the segments S2 and S3 are located in VL6, and the coccygeal segments are located in the VL7.

For all segments of the cat used in the analysis, several general morphometric parameters (i.e., general area of transverse slice, separate area of gray and white matter, height and width of the slice) were calculated (see details in Section “Morphometry”) and are shown in Table 3 and Figure 2B.

Creation of the atlas page

All data obtained (immunohistochemical, morphometric, spatial relationship between the spinal segments and vertebrae) were grouped at two pages (Figure 3). Two-three regions of interest for every segment were illustrated.

FIGURE 3

Each page includes the following elements:

• 1.

  The spinal cord region (C—cervical, T—thoracic, L—lumbar, S—sacral, Co—coccygeal).

• 2.

  The serial number of the segment.

• 3.

  The part of the segment (rostral, middle, caudal).

• 4.

  Part of the scheme of the interrelation between the spatial position of the segment and vertebra, the slices of which are presented on the page, and the vertebrae in which this segment is localized, as well as the segments and vertebrae adjacent to it.

• 5.

  Stereotaxic grid and a schematic representation of the transverse slice and laminar and nuclear borders. The borders were defined on the basis of features of all markers used (see details in Section “Definition of laminae and nuclei in the grey matter of the cat spinal cord”). Region of interest is shown as the vertical grid.

• 6.

  Unstained—unstained slice.

• 7.

  NeuN—NeuN stained slice.

• 8.

  ChAT—choline acetyltransferase stained slice.

• 9.

  Calbindin—calbindin 28 kDa stained slice.

• 10.

  Calretinin—calretinin stained slice.

• 11.

  Parvalbumin—parvalbumin stained slice.

• 12.

  SMI-32—SMI-32 stained slice.

• 13.

  Scale bar corresponds to 500 μm.

Definition of laminae and nuclei in the gray matter of the cat spinal cord

The gray matter of the spinal cord is divided into laminae and nuclei using several cytoarchitectonic features (Rexed, 1954). We modified the Rexed’s scheme for spinal laminae and nuclei definition using not only the total cytoarchitectonic features of the gray matter but also an additional neurochemical criteria originating from several immunohistochemical reactions. The key histological and neurochemical features distinguishing the laminae are presented below (methods that do not allow to see interlaminae or nuclear borders are not discussed). Summary data of neurochemical markers of the spinal cord laminae and nuclei in all segments is presented at Table 4 by color. All pages of the atlas are presented at Supplementary material.