Axon diversity of lamina I local-circuit neurons in the lumbar spinal cord

Abstract

Spinal lamina I is a key area for relaying and integrating information from nociceptive primary afferents with various other sources of inputs. Although lamina I projection neurons have been intensively studied, much less attention has been given to local-circuit neurons (LCNs), which form the majority of the lamina I neuronal population. In this work the infrared light-emitting diode oblique illumination technique was used to visualize and label LCNs, allowing reconstruction and analysis of their dendritic and extensive axonal trees. We show that the majority of lamina I neurons with locally branching axons fall into the multipolar (with ventrally protruding dendrites) and flattened (dendrites limited to lamina I) somatodendritic categories. Analysis of their axons revealed that the initial myelinated part gives rise to several unmyelinated small-diameter branches that have a high number of densely packed, large varicosities and an extensive rostrocaudal (two or three segments), mediolateral, and dorsoventral (reaching laminae III-IV) distribution. The extent of the axon and the occasional presence of long, solitary branches suggest that LCNs may also form short and long propriospinal connections. We also found that the distribution of axon varicosities and terminal field locations show substantial heterogeneity and that a substantial portion of LCNs is inhibitory. Our observations indicate that LCNs of lamina I form intersegmental as well as interlaminar connections and may govern large numbers of neurons, providing anatomical substrate for rostrocaudal "processing units" in the dorsal horn.

Figure 1

Somatodendritic features of LCNs. A: Images of large (left pair) and smaller (right pair) LCNs during the process of cell labeling. B: Photomicrograph of the soma, dendrites, and axon branches of a typical flattened LCN in a sagittal spinal cord section. C: Photomicrograph of a typical multipolar LCN in a sagittal section. D: Another multipolar LCN in a transverse section. Dashed line indicates the rough borders of lamina II. The section is slightly distorted because of detachment from the supporting agar during processing. E: Sagittal (left) and transverse (right) rotated views of a 3-D reconstructed flattened LCN (reconstruction from sagittal sections; axon omitted for clarity). F: Sagittal (left) and transverse (right) rotated views of a 3-D reconstructed multipolar LCN (reconstruction from sagittal sections; axon omitted for clarity). G: Histograms of soma area distribution (left) and number of stem dendrites (right) in LCNs (shaded bars) and ALT-PNs (open bars). The data for the ALT-PNs are from Szucs et al. (2010). H: 3-D Sholl analysis of the number of dendritic branch points in 50-μm-thick shells. C, caudal; M, medial; D, dorsal; R, rostral (throughout the figures). Bin sizes in G = 50 μm² in the soma area histogram; 1 in the stem dendrite histogram. Scale bars = 20 μm in A; 50 μm in B–D.

Figure 2

Typical multipolar LCN reconstructed from transverse serial sections. A: Overlaid image of 15 transverse, 100-μm-thick, serial sections. Dendrites (black) and axon (dark gray) of the LCN (cell ID: L316_E6) occupy laminae I–II and protrude into lamina III. An axon collateral descends ventrally beyond the neck of the dorsal horn, and some dendrites are located in the DLF. For clarity, borders of the white and gray matter (continuous light gray line; same in B,C) as well as of lamina II (dashed light gray line; same in B,C) are shown only for the section containing the cell body. B: Reconstructions showing the location of dendritic pieces (black) in the individual transverse serial sections. C: Reconstructions indicating the location of axonal pieces (black) in the individual transverse serial sections. soma, Section with the cell body; r, rostral; c, caudal. Scale bar = 500 μm.

Figure 3

Comparison of basic axon parameters of 3-D reconstructed LCNs and ALT-PNs. A: 3-D reconstruction of two LCNs (cell ID: L292_E1 in red; cell ID: L292_E5 in green) and a mixed-collateral-type (MCT) ALT-PN (cell ID: L292_E4 in blue), filled in the same spinal cord. Rostrocaudal, mediolateral, and dorsoventral dimensions of the bounding box (light gray) are indicated along the corresponding axis. B: Length of different lumbar spinal cord segments (L1–L6) at postnatal days 14 (P14) and 21 (P21). Columns in each case show the average and standard mean error of measurements in three different animals. C–E: Total axon length (i.e., sum of the lengths of all reconstructed axon segments), total axon volume (i.e., sum of all reconstructed axon segment volumes, calculated from segment length and local diameter), and number of branch points along the axon of individual cells (red, green, blue, and open columns) along with the mean value for 3-D reconstructed LCNs (gray column; n = 7).

Figure 4

Sagittal (A), horizontal (B), transverse (C), and perspective (D) view of a 3-D reconstructed but not fully connected LCN (cell ID: Zs079_E13). The majority of the axon (yellow) in this case is located medially, considerably remote from the lateral cell body and dendrites (orange). The neuron has a solitary branch running caudally in the dorsolateral funiculus on the ipsilateral side. The medial part of the axonal tree, because of distortion of some sections, could not be connected to form a single axon. White and gray matter borders are indicated with gray, and central canal is shown in green. Scale bar = 1 mm.

Figure 5

General morphological features of the axon of LCNs. Main axon (asterisk) originating from the soma (A) and from a primary dendrite (B) of two LCNs. Arrows point at the axon origin. C: Characteristic appearance of LCN axons in the vicinity of the cell body. The main axon (asterisk) and primary and secondary branches (arrowheads) are intermingled with fine terminal branches enriched with varicosities (t). D: The main axon (asterisk) of an LCN giving rise to primary branches (arrowheads) in an alternating manner. E: Perpendicular side branch from a solitary axon branch running in the dorsolateral funiculus. F: Straight, thick, solitary axon with swellings together with a more undulating, thin, varicose, also solitary branch next to the surface of the preparation, most likely in Lissauer's tract. G: Thick, myelinated-appearing solitary branch of an LCN, running in the dorsal funiculus. A–E,G: Extended focal images; F: single focal plane image. Scale bars = 25 μm.

Figure 6

Organization of the primary axon branches of LCNs. A: Sagittal (top), transverse (middle left), and horizontal (bottom) view of the 3-D reconstruction of an LCN with five primary branches (cell ID: L420_E2) where the first and second branches (orange and yellow) dominate the tree, showing partial overlap with the rest of the branches. The end of the main axon turns caudally and finishes abruptly close to the surface, most likely damaged when pia mater was removed during preparation. B: Main axon and the origination points of each primary branch. Note the dorsoventral loop of the main axon. C: Another LCN (cell ID: L292_E5) with seven primary branches, from which the last two (light blue and dark blue) cover the largest area. Similarly to the previous cell, the main axon forms a loop before it finishes abruptly close to the surface with a visible swelling at the end, probably caused by truncation. The first primary branch (orange) of this axon traverses the dorsal horn medially and target similar regions than the last branch (dark blue). D: Sagittal view of 3-D reconstructions of four other LCNs. Cell IDs are indicated at left in the corresponding reconstruction. Colored boxes at bottom indicate the color codes of the sequence of primary axon branches. S/d, soma and dendrites; main, main axon; 1st–7th, order of primary branches from the main axon. Scale bars = 500 μm in A,C; 1 mm in D.

Figure 7

Varicosity distribution along the axon of LCNs. Spatially dependent visual representation of axon varicosity distribution from sagittal (A) and horizontal (B) views, as indicated on the schematic drawings above each column. Axon varicosities along the 3-D reconstructed LCN axons were counted in predetermined space units (voxels). The maximum varicosity number in the predefined voxels (100 μm × 100 μm × 100 μm) is indicated as the maximum value (red) on the scale bar next to the particular neuron. The actual number of varicosities in a voxel is used as a color and opacity value for the cube representing that voxel. Cell IDs are indicated at left. 3-D scale bars (D, dorsal, green; L, lateral, blue; R, rostral, red) = 250 μm.

Figure 8

Physical and path distance of axon varicosities in 3-D reconstructed LCN axons. Sholl analysis was performed using 100-μm shells. Bin size in path distance histograms is 100 μm. A: Three cells showing a major accumulation of varicosities in the vicinity of, but not centered on, the soma, with one peak both in the Sholl analysis and in the path distance histogram. B: Other LCNs, with lower overall number of varicosities, showing additional local accumulations, evidenced by multiple peaks in the Sholl analysis, and at the same time wider distribution of the path distance histogram. Note the different scaling of the Y axis between A and B and also the difference on the X axis in cell ID: L292_E1 in B.

Figure 9

Fine-structural difference between myelinated and unmyelinated parts of an LCN axon. A: 2-D reconstruction of a representative LCN axon (cell ID: L279_E2). B: 2-D reconstruction of a single section of the same axon. Light gray processes are dendrites (d); the axon is in black. Continuous and dotted gray lines indicate the border of the section and border between the gray and white matter, respectively. Boxes indicate regions of interest with a single myelinated-appearing axon branch (C,D) and several varicose branches (E,F). C–F show the reconstruction of the corresponding region and a photomicrograph of the same region from the surface of the block used for preparing the electron microscopic sections. Black parts of the reconstructions in C,E show axon pieces still present on the block surface, whereas gray indicates parts that were already cut. G,H: Electron microscopic images of axon profiles from region C,D with several concentric layers of myelin. I: Diameter histogram of axon profiles (n = 24) from the same region measured with (open bars) and without myelin (shaded bars). J: Electron microscopic image of an intervaricosity segment and a varicosity (K) from region E,F. L: Diameter histogram of axon profiles (n = 157) from the same region. Bins = 50 nm in I,L. Scale bars = 1 mm in A; 50 μm in B; 10 μm in C–F; 500 nm in G,H,J,K.

Figure 10

Action potential (AP) propagation time maps of LCN axons. A: Sagittal view (see schematic at upper right) of a cell (ID: L395_E2) with a more compact tree and two cells (IDs: Zs022_E8-1 and L292_E1) with distant terminal branching areas. The whole axon is considered unmyelinated. B: AP propagation time maps of two cells (IDs: Zs022_E8-1 and L292_E1) replotted assuming that axon pieces with a diameter above 0.35 μm are myelinated. The maximum propagation time value in the map is indicated by red, whereas 0 msec is blue on the scale bar next to the particular neuron. Arrows point to regions where the propagation time map shows visible differences. C: Propagation time histogram of the axon varicosities of the three LCNs shown in A. Bins = 1 msec. Shaded bars, total axon unmyelinated; open bars, axon pieces with a diameter above 0.35 μm are myelinated.

Figure 11

Example of an inhibitory LCN. A: Single section of a biocytin-filled axon (magenta) of an LCN (cell ID: L571_E20-3). Arrows with numbers indicate the VGAT-immunoreactive axon varicosities (green) shown in detail in the insets at right. Arrows in insets point to the actual varicosity. The main image is projected from 36 optical sections at 0.5-μm Z spacing, whereas images in the insets are single optical sections. B: Primary branch organization of the same LCN axon shows relatively little overlap among the branches. C,D: The number of axon varicosities in this axon is moderate, and varicosity distribution shows more than one, relatively dispersed accumulation. E: Temporal dispersion of action potential propagation times in the same LCN axon did not show a noticeable difference when using the diameter threshold for myelinated axons, indicating the lack of long myelinated parts in the axon. Bins = 100 μm in D; 1 msec in E. Scale bars = 10 μm in A; 5 μm in insets; 500 μm in B.