Ultrastructural observations of synaptic connections of vibrissa afferent terminals in cat principal sensory nucleus and morphometry of related synaptic elements

Synaptic connectivity of urinary bladder afferents in the rat superficial dorsal horn and spinal parasympathetic nucleus

That visceral sensory afferents are functionally distinct from their somatic analogues has been known for a long time but the detailed knowledge of their synaptic connections and neurotransmitters at the first relay nucleus in the spinal cord has been limited. To provide information on these topics, we investigated the synapses and neurotransmitters of identified afferents from the urinary bladder to the superficial laminae of the rat spinal dorsal horn (DH) and the spinal parasympathetic nucleus (SPN) by tracing with horseradish peroxidase, quantitative electron microscopical analysis, and immunogold staining for GABA and glycine. In the DH, most bladder afferent boutons formed synapses with 1-2 postsynaptic dendrites, whereas in the SPN, close to a half of them formed synapses with 3-8 postsynaptic dendrites. The number of postsynaptic dendrites and dendritic spines per bladder afferent bouton, both measures of synaptic divergence and of potential for synaptic plasticity at a single bouton level, were significantly higher in the SPN than in the DH. Bladder afferent boutons frequently received inhibitory axoaxonic synapses from presynaptic endings in the DH but rarely in the SPN. The presynaptic endings were GABA- and/or glycine-immunopositive. The bouton volume, mitochondrial volume, and active zone area, all determinants of synaptic strength, of the bladder afferent boutons were positively correlated with the number of postsynaptic dendrites. These findings suggest that visceral sensory information conveyed via the urinary bladder afferents is processed differently in the DH than in the SPN, and differently from the way somatosensory information is processed in the spinal cord.

Changes in the axon terminals of primary afferents from a single vibrissa in the rat trigeminal nuclei after active touch deprivation or exposure to an enriched environment

Lasting modifications of sensory input induce structural and functional changes in the brain, but the involvement of primary sensory neurons in this plasticity has been practically ignored. Here, we examine qualitatively and quantitatively the central axonal terminations of a population of trigeminal ganglion neurons, whose peripheral axons innervate a single mystacial vibrissa. Vibrissa follicles are heavily innervated by myelinated and unmyelinated fibers that exit the follicle mainly through a single deep vibrissal nerve. We made intraneural injections of a mixture of cholera-toxin B (CTB) and isolectin B4, tracers for myelinated and unmyelinated fibers, respectively, in three groups of young adult rats: controls, animals subjected to chronic haptic touch deprivation by unilateral whisker trimming, and rats exposed for 2 months to environmental enrichment. The regional and laminar pattern of terminal arborizations in the trigeminal nuclei of the brain stem did not show gross changes after sensory input modification. However, there were significant and widespread increases in the number and size of CTB-labeled varicosities in the enriched condition, and a prominent expansion in both parameters in laminae III-IV of the caudal division of the spinal nucleus in the whisker trimming condition. No obvious changes were detected in IB4-labeled terminals in laminae I-II. These results show that a prolonged exposure to changes in sensory input without any neural damage is capable of inducing structural changes in terminals of primary afferents in mature animals, and highlight the importance of peripheral structures as the presumed earliest players in sensory experience-dependent plasticity.

Distribution of gastrin-releasing peptide in the rat trigeminal and spinal somatosensory systems

Gastrin-releasing peptide (GRP) has recently been identified as an itch-specific neuropeptide in the spinal sensory system in mice, but there are no reports of the expression and distribution of GRP in the trigeminal sensory system in mammals. We characterized and compared GRP-immunoreactive (ir) neurons in the trigeminal ganglion (TG) with those in the rat spinal dorsal root ganglion (DRG). GRP immunoreactivity was expressed in 12% of TG and 6% of DRG neurons and was restricted to the small- and medium-sized type cells. In both the TG and DRG, many GRP-ir neurons also expressed substance P and calcitonin gene-related peptide, but not isolectin B4 . The different proportions of GRP and transient receptor potential vanilloid 1 double-positive neurons in the TG and DRG imply that itch sensations via the TG and DRG pathways are transmitted through distinct mechanisms. The distribution of the axon terminals of GRP-ir primary afferents and their synaptic connectivity with the rat trigeminal sensory nuclei and spinal dorsal horn were investigated by using light and electron microscopic histochemistry. Although GRP-ir fibers were rarely observed in the trigeminal sensory nucleus principalis, oralis, and interpolaris, they were predominant in the superficial layers of the trigeminal sensory nucleus caudalis (Vc), similar to the spinal dorsal horn. Ultrastructural analysis revealed that GRP-ir terminals contained clear microvesicles and large dense-cored vesicles, and formed asymmetric synaptic contacts with a few dendrites in the Vc and spinal dorsal horn. These results suggest that GRP-dependent orofacial and spinal pruriceptive inputs are processed mainly in the superficial laminae of the Vc and spinal dorsal horn.

Whisker-related circuitry in the trigeminal nucleus principalis: ultrastructure

Trigeminal (V) nucleus principalis (PrV) is the requisite brainstem nucleus in the whisker-to-barrel cortex model system that is widely used to reveal mechanisms of map formation and information processing. Yet, little is known of the actual PrV circuitry. In the ventral "barrelette" portion of the adult mouse PrV, relationships between V primary afferent terminals, thalamic-projecting PrV neurons, and gamma-aminobutyric acid (GABA)-ergic terminals were analyzed in the electron microscope. Primary afferents, thalamic-projecting cells, and GABAergic terminals were labeled, respectively, by Neurobiotin injections in the V ganglion, horseradish peroxidase injections in the thalamus, and postembedding immunogold histochemistry. Primary afferent terminals (Neurobiotin- and glutamate-immunoreactive) display asymmetric and multiple synapses predominantly upon the distal dendrites and spines of PrV cells that project to the thalamus. Primary afferents also synapse upon GABAergic terminals. GABAergic terminals display symmetric synapses onto primary afferent terminals, the somata and dendrites (distal, mostly) of thalamic-projecting neurons, and GABAergic dendrites. Thus, primary afferent inputs through the PrV are subject to pre- and postsynaptic GABAergic influences. As such, circuitry exists in PrV "barrelettes" for primary afferents to directly activate thalamic-projecting and inhibitory local circuit cells. The latter are synaptically associated with themselves, the primary afferents, and with the thalamic-projecting neurons. Thus, whisker-related primary afferent inputs through PrV projection neurons are pre- and postsynaptically modulated by local circuits.

Ultrastructural analysis of low-threshold mechanoreceptive vibrissa afferent boutons in the cat trigeminal caudal nucleus

Ultrastructural parameters related to synaptic release and their correlation with synaptic connectivity were analyzed in the low-threshold mechanoreceptive vibrissa afferent boutons in laminae III and IV of the trigeminal caudal nucleus (Vc). Rapidly adapting vibrissa afferents were intra-axonally labeled, and quantitative ultrastructural analyses with serial sections were performed on the labeled boutons and their presynaptic endings (p-endings). The volume of the labeled boutons was widely distributed from small to large ones (0.8~12.3 µm³), whereas the p-endings were small and uniform in size. The volume of the labeled boutons was positively correlated with the ultrastructural parameters such as mitochondrial volume (correlation coefficient, r=0.96), active zone area (r=0.82) and apposed surface area (r=0.79). Vesicle density (r=-0.18) showed little correlation to the volume of labeled boutons, suggesting that the total vesicle number of a bouton is proportional to its volume. In addition, the bouton volume was positively correlated with the number of p-endings (r=0.52) and with the number of dendrites postsynaptic to the labeled bouton (r=0.83). These findings suggest that low-threshold mechanoreception conveyed through vibrissa afferents is processed in a bouton size-dependent manner in the Vc, which may contribute to the sensory-motor function of laminae III/IV in Vc.

GABA- and glycine-like immunoreactivity in axonal endings presynaptic to the vibrissa afferents in the cat trigeminal interpolar nucleus

The goal of this study was to analyze the synaptic interaction of primary afferents with GABA- and/or glycine-immunopositive presynaptic endings in the cat trigeminal interpolar nucleus (Vi). Fast adapting vibrissa afferents were labeled by intra-axonal injections of horseradish peroxidase. Postembedding immunogold labeling on serially cut ultrathin sections and quantitative ultrastructural analysis of the labeled boutons and their presynaptic endings (p-endings) in the Vi were performed. The majority of p-endings presynaptic to labeled boutons (83%) were immunopositive for both GABA and glycine and 8% were immunopositive for glycine alone. A small fraction of p-endings were immunopositive for GABA alone (4%) or immunonegative for both GABA and glycine (4%). Ultrastructural parameters related to synaptic release, i.e. bouton volume, mitochondrial volume, and active zone area, were significantly larger in the labeled boutons of primary afferents than in the p-endings. The volume of labeled boutons was positively correlated with the number of the postsynaptic dendrites and p-endings. In addition, fairly large-sized labeled boutons and p-endings were frequently observed in the Vi. These results reveal that large majority of vibrissa afferents in the Vi are presynaptically modulated by interneurons immunopositive for both GABA and glycine, and suggest that the Vi plays a distinct role in the processing of orofacial sensory information, different from that of other trigeminal sensory nuclei.

The synaptic microcircuitry associated with primary afferent terminals in the interpolaris and caudalis of trigeminal sensory nuclear complex

Previous ultrastructural studies indicating a higher number of axoaxonic contacts on individual low-threshold mechanoreceptive afferents in the principalis (Vp) than in the oralis (Vo) of cat trigeminal sensory nuclear complex (TSNC) suggest that the synaptic microcircuitry associated with primary afferents manifests unique differences across the sensory nuclei of TSNC. To address this issue, we analyzed synaptic microcircuits associated with fast adapting vibrissa afferent terminals in the interpolaris (Vi) and caudalis (Vc, laminae III/IV) by using intraaxonal injections of horseradish peroxidase (HRP) in cats. Forty-two and 65 HRP-labeled boutons were analyzed in the Vi and Vc, respectively. The labeled boutons contained clear, spherical vesicles. They most frequently formed asymmetric axodendritic synapses and were commonly postsynaptic to unlabeled axon terminals containing pleomorphic vesicles (p-endings) with symmetric junctions. The examination of synaptic contacts over the entire surface of individual boutons indicated that the afferent boutons made contacts with an average of two postsynaptic targets in the Vi and Vc. In contrast, axoaxonic contacts, and labeled boutons participating in synaptic triads, where p-endings contacted both the boutons and their postsynaptic targets, were, on average, higher in the Vi than in the Vc. These results suggest that the output of sensory information conveyed through low-threshold mechanoreceptive afferents is more strongly controlled at the level of the first synapse by presynaptic and postsynaptic mechanisms in the Vi responsible for sensory discriminative functions than in the Vc for sensorimotor reflexive functions.

Neural mechanisms controlling jaw-jerk reflex in the cat

Signal substances of axon terminals presynaptic to jaw spindle Ia afferents and their ultrastructural features were examined using a combination of intra-axonal horseradish peroxidase injection and postembedding immunogold-labeling techniques in cats. A total of 35 axon terminals presynaptic to 22 horseradish peroxidase-labeled Ia boutons were examined. Of the 35 presynaptic axon terminals, 14 (40%) were immunoreactive for both gamma-aminobutyric acid and glycine, 9 (26%) for gamma-aminobutyric acid alone and 9 (26%) for glycine alone. The bouton volume, mitochondrial volume, active zone area, and apposed surface area were larger for Ia boutons than for presynaptic axon terminals, while each of the values is similar among the three types of presynaptic axon terminals. These results suggest that gamma-aminobutyric acid and glycine play an important role for modulating the jaw-jerk reflex presynaptically and that the smaller size of presynaptic axon terminals is important to prevent action potential generation from Ia afferents.

Quantitative analysis of tooth pulp afferent terminals in the rat brain stem

This study analyzed quantitatively the ultrastructural features of tooth pulp afferent terminals and their presynaptic axonal endings (p-endings) in the trigeminal principal (Vp), dorsomedial oral (Vdm), and caudal nuclei (Vc). Mitochondrial volume, active zone area, apposed surface area, and vesicle number were highly correlated with afferent bouton volume. The afferent bouton volume varied widely in Vp, compared to that in Vdm and Vc. The values of all parameters of p-endings were within a narrow range, and were smaller than those of afferent boutons. The afferent bouton volume correlated with the number of postsynaptic dendrites and p-endings. These results suggest that pulpal afferent information is regulated in a unique manner in the each trigeminal sensory nucleus.

GABA and glycine in synaptic microcircuits associated with physiologically characterized primary afferents of cat trigeminal principal nucleus

Previous studies suggest that sensory information conveyed through trigeminal afferents is more strongly controlled at the level of the first synapse by GABA-mediated presynaptic mechanisms in the trigeminal principal sensory nucleus (Vp) than other sensory nuclei. However, it is unknown if such a mechanism is common to functionally different classes of primary afferent in the same nucleus or across the nuclei. To address these issues, the present study focused on synaptic microcircuits associated with slowly adapting (SA) mechanosensory afferents innervating the periodontal ligaments in the cat Vp and attempted to examine GABA, glycine, and glutamate immunoreactivity in axon terminals involved in the circuits. Afferents were physiologically characterized before injection of horseradish peroxidase (HRP) and preparation for electron microscopy. HRP-labeled afferent boutons were serially sectioned and immunostained with antibodies against GABA, glycine, and glutamate using a postembedding immunogold method. All the afferent boutons examined contacted non-primary dendrites and they were frequently postsynaptic to unlabeled axons (p-endings). Axodendritic and axoaxonic contacts per afferent bouton were 1.3 (46/35) and 2.0 (70/35), respectively. Most p-endings were immunoreactive for GABA (63/70) and also glycine was co-stained in the majority of the p-endings (49/63). Thirty percent of p-endings with the colocalization of GABA and glycine participated in synaptic triads where a p-ending formed a synapse with the same dendrite as the afferent bouton. None of the p-endings was immunoreactive for glutamate. Most afferent boutons were enriched with glutamate but were immunonegative for GABA and glycine. This study provides evidence suggesting that transmission from SA afferents is strongly controlled presynaptically by GABAergic interneurons with colocalized glycine, and that a proportion of these interneurons, involved in synaptic triads, may also have postsynaptic inhibitory actions on target neurons of the SA afferents.

Synaptic interactions between the terminals of slow-adapting type II mechanoreceptor afferents and neurones expressing gamma-aminobutyric acid- and glycine-like immunoreactivity in the rat spinal cord

The object of this study was to analyse the synaptic interactions of slow-adapting type II (SAII) afferent terminals in laminae III-V of the rat spinal cord. The axons of SAII afferents were physiologically characterized by intracellular recording before injection with neurobiotin and preparation for electron microscopy. Axon terminals were serially sectioned and immunolabelled with antibodies against gamma-aminobutyric acid (GABA) or glycine by using a postembedding immunogold procedure. Computer-aided reconstruction was used to reveal the relative distribution of different types of synapses on terminal and en passant synaptic boutons. Eighty-nine percent of boutons received axoaxonic synaptic contacts, the mean number of contacts per bouton being 3.5. Fifty-nine percent of presynaptic axons were immunoreactive for both GABA and glycine and 45% for GABA alone. Most boutons (95%) made axodendritic contacts, and the mean number of dendrites contacted was 1.6. More than half of the postsynaptic dendrites were greater than 1 microm in diameter. Twenty-three percent were immunoreactive for glycine, and 71% were not immunoreactive for either antibody. Synaptic triads in which an axon presynaptic to the afferent was also in contact with a dendrite postsynaptic to the afferent were seen at 63% of boutons. These results are discussed in the light of similar studies of other low-threshold mechanoreceptive afferent terminals in the rat and cat and in the context of what is known of the sensory interneurones carrying information from slow- and rapid-adapting mechanoreceptors.

Expression of vanilloid receptor TRPV1 in the rat trigeminal sensory nuclei

Little is known about the central projection patterns of trigeminal afferent neurons expressing the vanilloid receptor TRPV1 and their coexpression of neuromodulatory peptides. To address these issues, we examined the distribution of TRPV1-positive neurons in the trigeminal ganglion (TG) and trigeminal sensory nuclei principalis (Vp), oralis (Vo), interpolaris (Vi), and caudalis (Vc) in the rat via light and electron microscopy. In addition, we studied the colocalization of TRPV1-positive neurons with substance P (SP) and calcitonin gene-related peptide (CGRP) via confocal microscopy. In TG, only small and medium-sized neurons were immunopositive for TRPV1. The staining for TRPV1 was found in axon collaterals in the dorsal parts of Vp, Vo, and Vi and in terminals and fibers throughout lamina I and the outer zone of lamina II (IIo) of Vc. With electron microscopy, TRPV1-positive fibers in the ascending and descending trigeminal tracts were found to be unmyelinated. Almost all TRPV1-positive terminals in Vc contained numerous large dense-core vesicles and formed synaptic contacts with single small dendrites. Multiple immunofluorescence revealed a high degree of colocalization of TRPV1 with SP and CGRP in TG neurons as well as in fibers and terminals confined to laminae I and IIo of Vc. These results suggest that the central projections of unmyelinated (C) afferents sensitive to noxious heat and capsaicin are organized differently between Vc and the rostral trigeminal nuclei and that Vc may play a role in the development of hyperalgesia.

Synaptic organization of tooth pulp afferent terminals in the rat trigeminal sensory nuclei

Previous studies provide evidence that a structure/function correlation exists in the distinct zones of the trigeminal sensory nuclei. To evaluate this relationship, we examined the ultrastructure of afferent terminals from the tooth pulp in the rat trigeminal sensory nuclei: the principalis (Vp), the dorsomedial part of oral nucleus (Vdm), and the superficial layers of caudalis (Vc), by using transganglionic transport of wheat germ agglutinin conjugated to horseradish peroxidase (WGA-HRP). A total of 93 labeled boutons were serially sectioned, in which some sections were incubated with gamma-aminobutyric acid (GABA) antiserum. Almost all labeled boutons formed asymmetric contact with nonprimary dendrites, in which more than half of labeled boutons in the Vc made synapses with their spines. The labeled boutons could be divided into two types on the basis of numbers of dense-cored vesicles (DCVs) in a boutons: S-type and DCV-type. Almost all labeled boutons in the Vp and Vdm were S-type, whereas two types were distributed evenly in the Vc. In contrast to DCV-type boutons, the S-type was frequently postsynaptic to unlabeled axon terminals containing a mixture of round, oval, and flattened vesicles (p-endings) and forming symmetrical synapses. Most p-endings examined were immunoreactive to GABA. The frequency of axoaxonic contacts was higher for labeled boutons in the Vp than in the Vdm and Vc. These results suggest that the three structures of trigeminal sensory nuclei serve distinct functions in nociceptive processing.

Synaptic relationships between hair follicle afferents and neurones expressing GABA and glycine-like immunoreactivity in the spinal cord of the rat

gamma-Aminobutyric acid (GABA) and glycine have been implicated in the inhibition of sensory pathways in the dorsal horn of the spinal cord. The object of this study is to investigate the interactions between neurones immunoreactive for GABA and/or glycine and hair follicle afferent terminals labelled by intracellular injection with neurobiotin. GABA and glycine-like immunoreactivity in axons and dendrites in synaptic contact with the afferent terminals was demonstrated by using a postembedding immunogold method, and serial section reconstruction was used to show the distribution and nature of these interactions in lamina III of the dorsal horn. Most afferent boutons (94%) were postsynaptic at axo-axonic synapses: 67% of presynaptic boutons presynaptic to the afferent terminals were immunoreactive for GABA and glycine, 24% for GABA alone, and 7% for glycine alone. Only a small percentage of dendrites postsynaptic to afferent boutons appeared to belong to inhibitory interneurones: 3% were immunoreactive for GABA and glycine, 10% for glycine alone, but 87% were immunoreactive for neither antibody. Many afferent boutons were the central terminals of what appeared to be type IIb glomeruli and were involved triadic synaptic arrangements at which boutons presynaptic to an afferent terminal also made axodendritic contacts with dendrites postsynaptic to the afferent. Many of the presynaptic boutons involved in the triads were immunoreactive for GABA and glycine. Because afferent terminals do not themselves express glycine receptors (Mitchell et al. [1993] J. Neurosci. 13:2371-2381), glycine may therefore act on dendrites postsynaptic to hair follicle afferent terminals at these triads.

Quantitative ultrastructure of slowly adapting lingual afferent terminals in the principal and oral nuclei in the cat

Previous studies provide evidence that a structure/function correlation exists in the cytoarchitectonically different zones of the trigeminal sensory nuclei. To extend this relationship, we examined the ultrastructural features of trigeminal primary afferent neurons in the cat dorsal principal nucleus (Vpd) and the rostrodorsomedial oral nucleus (Vo.r) using intra-axonal labeling with horseradish peroxidase and morphometric analyses. All labeled boutons contained round synaptic vesicles. Eighty-two percent of the boutons in the Vo.r and 99% of the boutons in the Vpd were presynaptic to nonprimary dendrites. The remaining boutons in the Vo.r were presynaptic to somata (8%) or primary dendrites (10%). The average number of postsynaptic profiles per labeled bouton did not differ in the Vpd and Vo.r. Most labeled boutons in the two nuclei were postsynaptic to unlabeled axon terminals with pleomorphic vesicles (p-ending). The number of p-endings per labeled bouton was higher in the Vpd than Vo.r A morphometric analysis indicated that labeled bouton volume and apposed surface area were larger in the Vpd than Vo.r while active zone area and vesicle number did not differ. All these parameters were larger than those of p-endings in each nucleus. In both labeled boutons and p-endings, the parameters were positively correlated with bouton size. These results suggest that sensory information conveyed through trigeminal afferents is more strongly controlled at the level of the first synapse by presynaptic mechanisms in the Vpd than in the Vo.r, while the efficacy of transmission at primary afferent synapses does not differ.

Morphological differences between fast and slowly adapting lingual afferent terminations in the principal and oral nuclei in the cat

Previous studies indicated that fast-adapting (FA) and slowly adapting (SA) mechanoreceptive afferents innervating the facial or intraoral structures give rise to morphologically distinct terminal arbors in the individual subdivisions of the trigeminal sensory nuclear complex. The present study examined the collateral morphologies of lingual afferents in the nuclei principalis (Vp) and oralis (Vo) of the cat. Seven FA and six SA lingual afferents were physiologically characterized and stained by the intra-axonal horseradish peroxidase (HRP) injection technique. The two types of afferents established terminal arbors in the dorsomedial subdivision (Vpd) of the Vp, and the rostrodorsomedial (Vo.r) and dorsomedial subdivisions (Vo.dm) of the Vo, but the collateral morphologies are different between the two types. The FA afferents gave rise to mediolaterally extended oblong arbors in each subdivision, but the arbors were better developed in the Vo.r than in the Vpd and Vo.dm. The number of collaterals, intercollateral distance, number of boutons per collateral, and bouton size were also different among the subdivisions. The SA afferents were divided into two subtypes; one had a preferential projection into the Vpd or the Vo.r and Vo.dm, and others lacked a selected projection. Although the shape of their arbors varied from a stringy form to a roundish form, the general profile was denser, better developed, and rounder than that of FA afferents in each subdivision. The intercollateral distance and bouton size were different among the subdivisions. The number of boutons per collateral, bouton density, and bouton size were larger in SA than FA afferents in each subdivision. The present study demonstrated that two functionally distinct lingual afferents manifest unique morphological differences in the Vpd and Vo.

Identification of gamma-aminobutyric acid-immunoreactive axon endings associated with mesencephalic periodontal afferent terminals and morphometry of the two types of terminals in the cat supratrigeminal nucleus

A previous study has shown that mesencephalic periodontal afferent terminals receive contacts more frequently from axonal endings containing pleomorphic, synaptic vesicles (P-endings) in the supratrigeminal nucleus (Vsup) than in the trigeminal motor nucleus, suggesting that interneurons in Vsup play an important role in modulating the jaw-closing reflex. The present study was attempted to identify neurotransmitters in P-endings associated with mesencephalic periodontal afferents in cat Vsup through the use of intracellular staining of horseradish peroxidase combined with the postembedding immunogold methods. A morphometric analysis was carried out to compare the ultrastructural features of these two types of terminals. Serial sections of 31 labeled boutons and of their associated 38 P-endings were examined. They were processed for postembedding immunogold labeling with antibodies to the neurotransmitter gamma-aminobutyric acid (GABA). The 38 P-endings presynaptic to periodontal afferents showed GABA-like immunoreactivity, but the afferent terminals were free from the labeling. The morphometric analysis indicated that bouton volume, apposed surface area, total active zone size, and mitochondrial volume were smaller in GABA-immunoreactive P-endings than in periodontal afferents, but the pooled data of the two types of terminals showed that each synaptic parameter was highly correlated in a positive, linear manner with bouton volume. These observations provide evidence that P-endings presynaptic to mesencephalic periodontal afferents contain the neurotransmitter GABA and that their axoaxonic synapses are organized in accordance with the ultrastructural "size principle" proposed by Pierce and Mendell (Pierce and Mendell [1993] J. Neurosci. 13:4748-4763) on Ia-motoneuron synapses.