Three-dimensional architecture and linearized mapping of vibrissa follicle afferents

Abstract

Understanding vibrissal transduction has advanced by serial sectioning and identified afferent recordings, but afferent mapping onto the complex, encapsulated follicle remains unclear. Here, we reveal male rat C2 vibrissa follicle innervation through synchrotron X-ray phase contrast tomograms. Morphological analysis identified 5% superficial, ~32 % unmyelinated and 63% myelinated deep vibrissal nerve axons. Myelinated afferents consist of each one third Merkel and club-like, and one sixth Ruffini-like and lanceolate endings. Unsupervised clustering of afferent properties aligns with classic morphological categories and revealed previously unrecognized club-like afferent subtypes distinct in axon diameter and Ranvier internode distance. Myelination and axon diameters indicate a proximal-to-distal axon-velocity gradient along the follicle. Axons innervate preferentially dorso-caudally to the vibrissa, presumably to sample contacts from vibrissa protraction. Afferents organize in axon-arms innervating discrete angular territories. The radial axon-arm arrangement around the vibrissa maps into a linear representation of axon-arm bands in the nerve. Such follicle linearization presumably instructs downstream linear brainstem barrelettes. Synchrotron imaging provides a synopsis of afferents and mechanotransductory machinery.

Fig. 1. Dense vibrissa follicle reconstruction.

a Rat face photograph showing characteristic mystacial vibrissae. b Schematic of X-ray phase contrast imaging at the GINIX parallel beam setup (DESY, voxel size of 650 nm). c 3D rendering of the obtained C2 follicle dataset. d Virtual 2D section through the follicle shows detailed anatomical structures (yellow inset = axonal innervation). Inset shows myelinated axons of the deep vibrissal nerve at high magnification. e 3D rendering of densely reconstructed follicle anatomy, including capsule (transparent grey), cavernous sinus (red), ring sinus (red), ringwulst (yellow), Merkel cell region (pink), inner conical body (mint), sebaceous gland (green) and axonal innervation (random color assignment). f High magnification 3D rendering of 174 myelinated and ≥58 unmyelinated deep vibrissal nerve axons (lower) and 14 circumferential lanceolate axons (upper) supplied by the superficial vibrissal nerve.

Fig. 2. Three follicle axon classes and axonal velocity gradient of deep vibrissal afferents.

a Myelinated (blue) and unmyelinated (green) innervation supplied by the deep vibrissal nerve and circumferential innervation (orange) supplied by the superficial nerve. Ringwulst shown in transparent grey. b Circumferential innervation supplied by the superficial nerve wraps around the upper region of the inner conical body (ICB, transparent grey). c 14 superficial axons enter in four axonal fascicles. d Unmyelinated innervation supplied by the deep vibrissal nerve targets the root sheath (transparent grey) in straight vertical trajectories. e >58 unmyelinated axons (lower bound estimate) are polarized to the dorso-caudal circumference of the vibrissa shaft (see also Fig. 3). f Fiber area plotted against terminal height (relative to nerve entrance into the follicle) show a gradient of increasing axon thickness. Myelinated data is shown in blue and unmyelinated in green. g Estimated axonal conduction velocity plotted against terminal height. Conduction velocity was estimated according to Waxmann and Bennett (r = 0.93, p = 1.67 × 10⁻⁵⁰; refers to linear regression). h Fiber diameter of afferents terminating below 850 µm (transparent grey) and above 850 µm (white with black outlines) relative to the nerve entrance into the follicle (see Fig. 1e). l lateral, d dorsal, c caudal.

Fig. 3. Myelinated afferent types and endings.

a Whole mount myelinated axonal innervation (n = 149 axons) colored by afferent type (see color legend) and ringwulst, shaft and outer root sheath in transparent grey. b Left: Ring sinus level Merkel (n = 51) and lanceolate afferents (n = 21). Merkel afferents are distinct by a kink at the axon terminal from piercing the glassy membrane (white arrow). Right: High magnification of a completely reconstructed Merkel afferent including Merkel feet (MF, blue) and Merkel cells (MC, light blue) terminating behind the glassy membrane and lanceolate ending terminating outside the glassy membrane. Virtual 2D sections of the respective endings are shown above. c Left: Ringwulst level club-like afferent (n = 52) population. Right: Side view of club-like afferents terminating on the glassy membrane of the root sheath directly adjacent to the ringwulst. White arrows indicate pinching of the afferent terminals upon shaft deflection. d Left: Cavernous sinus level Ruffini-like afferents (n = 25). Right: Top view of a single Ruffini-like afferent terminating on the glassy membrane below the ringwulst. Virtual 2D section of the ending is shown above. Ruffini-like endings are distinct by non-straight trajectories targeting the root sheath. e Ending characteristic overview of 149 classified afferents. Black ticks indicate the presence and white ticks indicate the absence of the morphological feature as displayed on the left that were scored for each afferent. Twenty two afferents remain unclassified due to incompleteness. Multibranch refers to >3 branches and shaft trajectory to almost horizontal ending trajectories (as in 4 d). f Fiber area (displayed as median) of classified afferents. Asterix’ indicate significant differences (Kruskal Wallis: H = 38.85, p = 1.8 × 10⁻⁸; Significance according to Dunn’s Bonferroni corrected pairwise comparison with alpha = 0.05). Boxplots display the 25th to 75th percentile range as the box and the median as center line. Boxplot whiskers extend by the inter quartile range. Outliers are plotted individually. Note the thickness gradient from upper (left) to lower (right) terminal heights. g Histogram of branches per afferent for afferents that branched, colored by afferent type. h Left: Estimated proportions of vibrissal innervation composition (unmyelinated axons are lower bound estimate). Right: Afferent type proportions of the myelinated innervation subpopulation (n = 149 classified axons).

Fig. 4. Unsupervised hierarchical clustering supports classical afferent categories and reveals club like afferent subtypes.

a A factor analysis of mixed data of myelinated afferents based on intrinsic parameter colored by morphologically identified afferent type (Merkel = blue, lanceolate = orange, club like = green, Ruffini = red). b Left: Gower dissimilarity matrix of afferents based axonal parameter (ringwulst termination, shaft entry, vertical terminal height, fiber diameter, median internode length and branch count). Right: same as left but based on morphological parameter (same as used for morphological scoring in Fig. 3). Both matrices show distinct afferent groups with high similarity, which closely resemble each other. Color bar applies to both plots. c Hierarchical clustering (Wards method) dendrogram from Gower dissimilarity indices based on intrinsic parameter. Biggest cluster match morphologically identified afferent classes (Ruffini, red and lanceolate, orange). Morphologically identified club-like afferent types both separate into two further sub-cluster (club like 1 and club like 2 in dark and light green). d Histogram of club like afferents fiber area and Ranvier internode length colored by club like subclass. Asterisk’ refer to significance according to two-sided t-test, comparing both distributions (Fiber area: p = 1.97 × 10⁻¹², Median internode length: ip = 2.35 × 10⁻⁶). e 3D rendering of club like subtypes and ringwulst. f Histogram of club like subclass positions along the vibrissal circumference colored by club like subclass.

Fig. 5. Polarized innervation caudal to the vibrissa.

a Schematic rat head during whisking. Blue square indicates the C2 vibrissa follicle. b MicroCT 3D rendering of a C2 follicle and musculature (red) with movement schematic. Contraction of the sling-like muscles induces vibrissa protraction (red arrow). c Volume rendering showing asymmetric axonal innervation and ringwulst (top view; red arrow = protraction direction) of the C2 vibrissa. Data of the left C2 vibrissa was mirrored to follow right = forward conventions. N = 232 axons (174 myelinated axons in light blue and 58 unmyelinated axons in dark blue). d Angular distribution of afferent terminal positions. Innervation is significantly denser caudal (90–270°) than rostral. Myelinated axons contribute weakly (light blue) and unmyelinated axons strongly (dark blue) to the polarization effect. e To assess the statistical significance of axonal polarization bias we applied a two-sided binomial test, comparing the afferent distribution of opposing semi-circles for myelinated (light blue), unmyelinated (dark blue) and total innervation (blue). The x-axis represents a cut that separates the semi-circles, which shifts in 20° increments clockwise (as indicated above). Asterisk indicates the axis in which distribution is significantly different from chance for both myelinated and unmyelinated afferents (myelinated: p = 0.006, unmyelinated: p = 1.11 × 10⁻¹⁶, total: p = 9.68 × 10⁻¹⁰). f Raster plot of afferent types sorted by angular position.

Fig. 6. Axon arms as angular units and radial to linear transformation of axon arms in the vibrissal nerve.

a Volume rendering showing two axon arms (yellow and red). Inter-arm axon crossing is rare, while intra-arm axon crossing occurs more frequent. b Histogram of axon crossing index distribution. Negative values indicate more inter-arm and positive values more intra-arm crossings (see a) per axon. c Top, terminal axon positions color coded by axon arm. Bottom, angle territories of axon arms. Boxplots display the 25th to 75th percentile range as the box and the median as center line. Boxplot whiskers extend by the inter quartile range. Outliers are plotted individually. d 3D rendering of axonal innervation (color coded by axon arm). White boxes indicate sections shown in (e, f). e Follicle arm section (see d), showing radially arranged axon arms and the cut of the radial representation introduced in the nerve (see e). f Nerve section (see d). Note the linearization of the radial axon-arm representation seen in (e). Axon diameter in (e, f) are proportional but not on scale. l lateral, r rostral, d dorsal. g Idealized models of afferent terminal angular position layouts in the vibrissal nerve. Left: Terminal angular positions are represented as a condensed angular (radial) arrangement in the nerve cross section. Right: Terminal angular positions are represented linearly along the nerve cross section. h Predicted afferent terminal angular position by the radially-condensed model (grey) and the linearly-unwrapped model (black). The y-axis corresponds to the actual terminal afferent position. R- and p-value refers to linear regression for the linearly unwrapped model (black).