Regulating muscle spindle and Golgi tendon organ proprioceptor phenotypes

Abstract

Proprioception is an essential part of motor control. The main sensory subclasses that underlie this feedback control system - muscle spindle and Golgi tendon organ afferents - have been extensively characterized at a morphological and physiological level. More recent studies are beginning to reveal the molecular foundation for distinct proprioceptor subtypes, offering new insights into their developmental ontogeny and phenotypic diversity. This review intends to highlight some of these new findings.

INTRODUCTION.

Skeletal muscles, joints and skin are endowed with specialized mechanoreceptive organs that are activated by changes in muscle length or tension, joint angle or skin stretch through self-generated or passive movements^1,2. Proprioceptive afferents that innervate these receptors continuously provide the central nervous system with real-time updates on the position and movement of body and limbs^1,2. This information enables motor centers in the brain to compute an internal body plan that serves as a reference for motor planning. In addition, at spinal levels, proprioceptive feedback merges with descending commands to adjust motor output during unanticipated environmental circumstances (e.g. a sudden incline in the road).

Proprioceptive feedback from skeletal muscle originates from muscle spindle (MS) and Golgi tendon organ (GTO) receptors (Figure 1). Many decades of work have revealed considerable phenotypic diversity among MS and GTO receptors and their cognate sensory afferents^3,4. More recent advances have enabled the selective visualization and manipulation of muscle proprioceptors in dorsal root ganglia (DRG), spinal cord and muscle. Along with next generation sequencing techniques, these studies have uncovered novel insights into the developmental mechanisms that underlie the specification of distinct proprioceptor subtypes in the mouse, and are the subject of this review. Focusing on MS and GTO afferents innervating muscles of the trunk and limbs, we will discuss the emergence of a generic proprioceptor fate, the molecular basis of MS and GTO afferent subtypes, and the developmental mechanisms that contribute to their muscle type identity. For insights into other classes of mammalian proprioceptors (e.g. oculomotor palisade endings, joint receptors, jaw proprioceptors in the mesencephalic trigeminal nucleus), or on proprioceptive neurons in other animal models we refer to other reviews^5–8.

Figure 1. Proprioceptive muscle afferent subtypes.

A. Schematic indicating the main proprioceptor afferent subtypes as identified based on their association with MS or GTO afferent receptor organs, muscle target, and spinal projection pattern. Mature proprioceptors project to a single muscle in the periphery, and upon entering the spinal cord, bifurcate into a rostral and caudal projecting branch. Collaterals from the main axonal branches project to interneurons or projection neurons at intermediate spinal levels (MS and GTO afferents), and to ventrally located motor neurons (MS afferents). MS afferents form monosynaptic connections with homonymous and synergistically acting motor neurons, but avoid contacts with motor neurons that control antagonistic muscle. Afferent information is also relayed to supraspinal targets, either directly (at forelimb levels), or through projection neurons of the spinocerebellar tracts (hind limb levels).

Abbreviations: DRG (dorsal root ganglion), MS (muscle spindle), GTO (Golgi tendon organ), IN (interneuron), MN (motor neuron), SCT (Spinocerebellar tract).

B. Distribution of MS and GTO sensory terminals in a p2 neonatal Gluteus muscle. Sensory terminals are genetically marked by a Cre-inducible GFP reporter (Tau:mGFP-iNLZ; activated through a PV:Cre driver)⁴². MS group Ia and II afferents cannot yet be distinguished at this point.

C. Groups Ia and II MS sensory endings in an adult adductor muscle, genetically marked by tdTomato (tdT) expression, using an intersectional genetic approach based on PV:Cre, Runx3:FlpO, and the Cre- and Flp-recombinase dependent (Ai65) tdT reporter²³. Scale 20 μm.

D. Schematic of the composition of a typical MS. MS consists of several intrafusal muscle fibers (nuclear Bag1, Bag2, and chain)³. Group Ia afferents innervate the equatorial region of the spindle and associate with nuclear bag 1 and 2 fibers as well as intrafusal nuclear chain fibers¹⁷. Group II terminals are positioned (bi)lateral to the group Ia terminals (only one group II afferent is shown), and associate with chain fibers and bag 2 fibers (less often), but almost never with bag 1 fibers.

In addition to their different intra-spindle termination, group Ia afferents can be distinguished from group II afferents physiologically based on their dynamic sensitivity, their lower activation threshold, and their faster conduction velocity⁴. The polar ends of the intrafusal myofibers are contacted by gamma MNs, which regulate the threshold sensitivity of the sensory terminals for muscle stretch (only one set of motor endings is shown). Depending on their muscle of origin, spindles may consist of different complements of intrafusal fibers.

E. Group Ib afferent sensory ending in an adult adductor muscle, marked by tdT using the intersectional genetic approach described in C. Scale 20 μm.

F. Sensory terminals of group Ib afferents intertwine between the collagen fibers that attach the extrafusal muscle fibers to tendons or aponeuroses.

DEVELOPMENT OF A GENERIC PROPRIOCEPTOR SENSORY IDENTITY.

Spinal somatosensory neurons derive from the trunk neural crest, a transient progenitor population that delaminates from the dorsal aspect of the neural tube and coalesces into DRG⁹. Proprioceptors are amongst the earliest sensory neurons to emerge within the ganglia, and shortly after their generation, they project a peripheral axon to their muscle target and a central collateral to their recipient spinal targets^10,11 (Figure 2). The subsequent specification and patterning along the rostro-caudal axis, differentiation into distinct proprioceptor subtypes, and incorporation into dedicated sensory-motor circuits appear to result from a complex interplay between intrinsic genetic determinants and extrinsic signals.

Figure 2. Mechanisms of development of proprioceptor phenotypes.

A. Early developmental events leading to the acquisition of a generic proprioceptive identity. Shortly after cell cycle exit (e10), Runx3 drives bipotent TrkC⁺/TrkB⁺ progenitors to commit to a proprioceptive fate by maintaining expression of TrkC while repressing Shox2, which is required for TrkB expression and cutaneous mechanoreceptor fate^12–14,51. At the same time, NT3 expressed in the surroundings of the growing peripheral axons signals through TrkC receptors to promote their survival and outgrowth⁵².

B. Control of rostro-caudal organization of proprioceptive muscle subtype identity. Hox gene networks coordinate the organization of stretch reflex circuits at thoracic and limb levels directly through specification of Ia afferent muscle subtype identity - axial and limb muscle respectively - and indirectly by specifying motor neuron identities and expression of factors controlling sensory-motor connectivity^32,43,46,53.

C. Regulation of fine-grained muscle type identity. At a single muscle level proprioceptor identity appears to be regulated by yet to be defined extrinsic factors: in the distal hindlimb, dorsally and ventrally connected Ia afferents are characterized by specific molecular signatures (dorsal: cdh13; sema5a; ventral: crtac1; vstm2b) and induced by the limb mesenchyme³⁸.

D. MS and GTO afferent identities are marked by selective expression of Heg1 and Pou4f3, as revealed by single cell transcriptional analysis²³. Failure to detect these molecular markers in prior transcriptomic studies may be attributed to differences in sequencing depth and/or diversity of the neuronal population included in the analysis. Expression of Heg1 and Pou4f3 can be detected from early developmental stages, however maturation and diversification into further subsets continues into postnatal development supporting the existence of multiple molecularly and functionally distinct classes of proprioceptive afferents²³. Potential sources of the signals that promote proprioceptor diversification may include the mesenchyme, muscle, or the nascent sensory receptor organ.

The initial phase of this process, leading to the acquisition of a generic proprioceptor identity and organization along the rostro-caudal neuraxis, is largely controlled by Neurotrophin 3 (NT3), its receptor TrkC, and the runt domain transcription factor Runx3^{10, 12–14} (Figure 2). In accordance with the neurotrophic hypothesis, limiting amounts of peripheral NT3 regulate proprioceptor number to accommodate the different anatomical requirements of limb and trunk musculature. However, a recent study suggests that proprioceptor competitiveness for NT3 is not a stochastic process but is biased by an intrinsic molecular signature, including elevated levels of Runx3¹⁵. Early Runx3 expression levels - before embryonic day (e) 12.5 - positively correlate with proprioceptor survival, and appear to be controlled not through NT3, but by retinoic acid (RA) signaling from the paraxial mesoderm^12,15. These observations suggest that variations in RA concentrations along the neuraxis may differentially scale proprioceptor number in thoracic and limb level DRG¹⁵. Interestingly, RA also plays important roles in motor neuron and skeletal muscle generation and could represent a signaling hub for coordinating early events in the development of sensory-motor circuits.

DIFFERENTIATION OF MS AND GTO AFFERENT PROPRIOCEPTOR SUBTYPES.

After acquiring their generic identity, proprioceptors differentiate along two concurrent developmental trajectories, corresponding to the receptor organ they associate with and the muscle they innervate. With respect to the former, proprioceptors can be distinguished on the basis of their innervation of MS or GTO receptor organs. MSs consist of specialized intrafusal muscle fibers and are typically innervated by one primary (group Ia) and several secondary (group II) proprioceptive neurons^3,16,17 (Figure 1). Both types of afferents respond to stretch of the intrafusal fibers, such that changes in limb position result in increased or decreased firing rates¹⁸. Despite supplying the same sensory receptor, group Ia and group II afferents exhibit distinct intra-spindle innervation patterns, activation thresholds, and conduction velocities - features that appear to render them biased to qualitatively different information of muscle stretch (Figure 1). In addition, group Ia afferents typically possess a high dynamic sensitivity, which not only enables them to signal static changes in limb position (contributing to the sense of position), but also the speed at which the changes occur (contributing to the sense of movement)¹⁹. Based on morphological and electrophysiological observations, it appears however, that the classification of group Ia and II afferents is somewhat ambiguous, as these afferents can exhibit a rather large degree of overlap in their presumptive defining features⁴ (see Figure 1 for details). GTOs are located at the myotendinous junction and are innervated by a single group Ib afferent²⁰ (Figure 1). Group Ib terminals are particularly sensitive to contraction of the motor units that are associated with the tendon organ they innervate²¹. The physiological properties of group Ib neurons are generally aligned with those of group Ia (low threshold, fast conduction velocity, dynamic sensitivity), but as with MS afferents, the response properties of group Ib afferents can vary widely. As of yet it is unclear whether this variability in proprioceptor phenotypes just reflects developmental noise or may be a property indicative of a divergent physiological function²².

The molecular identity of MS and GTO afferents remained largely unclear until recently. Single cell RNA sequencing of adult proprioceptors has revealed specific molecular signatures that delineate different neuronal classes²³ (Figure 2). In particular, expression of the transmembrane protein Heg1 and the transcription factor Pou4f3 appear to define MS and GTO innervating subtypes, respectively. Interestingly, Heg1⁺ proprioceptors further segregate into four molecularly distinct subsets. One of these clusters is marked by the expression of Calretinin (CR), while the remaining three are characterized by the expression of the synaptic molecule Neurexophilin1 (Nxph1). Based on morphological criteria, CR⁺ neurons appear to represent group Ia afferents, possibly indicating that the Nxph1⁺ clusters correspond to subsets of group II afferents. Transcriptional analysis of earlier developmental stages shows that the molecular distinctions between GTO and MS afferents are first manifest between e14.5 and birth - after muscle target innervation²³. Differences between group Ia and presumptive group II neurons appear in the postnatal period, and the full molecular maturation of MS and GTO innervating subtypes continues until at least two weeks after birth²³. These data extend and support previous transcriptomic studies indicating that sensory subtype identity is acquired relatively late in development²⁴.

The observation that molecular distinctions between MS and GTO afferents emerge after target innervation supports a possible role for reciprocal inductive interactions between afferents and their receptors. Contact between proprioceptors and nascent intrafusal fibers induces spindle differentiation through Neuregulin 1 (Nrg1) and Erb2 signaling²⁵. Intrafusal muscle fibers, in turn, may provide retrograde signals to support the specialization of MS afferents. Indeed, some MS afferent markers are no longer expressed in Egr3 mutant animals - in which spindle development is perturbed - thus supporting this idea^26,27. The nature of these putative retrograde signals remains unclear, but developing spindles are known to secrete NT3 and GDNF. While expression of NT3 is preserved in the rudimentary spindles of Egr3 mutants, GDNF is no longer present, nor are gamma motor neurons²⁸, suggesting that either could have a role in specifying groups Ia and II neurons. The new insights into the molecular distinctions between MS and GTO afferents offer a means to test these ideas and advance our understanding of the developmental mechanisms through which these neuronal subtypes emerge.

SPECIFICATION OF PROPRIOCEPTOR MUSCLE TYPE IDENTITY.

MS and GTO afferents can be further partitioned into distinct subsets based on the type and biomechanical properties of the muscle they innervate (here referred to as “muscle type identity”). Analyses of muscle type identity have concentrated on group Ia afferents which form the sensory component of the monosynaptic stretch reflex. In this circuit, Ia sensory afferents make selective connections with motor neurons that innervate the same muscle or muscles with synergistic function, while avoiding those controlling the activity of antagonist muscles. As such, the Ia stretch reflex offers an accessible conduit to examine afferent muscle-type identity²⁹. The exquisite selectivity in the pattern of sensory-motor connectivity is established at late embryonic stages, with no significant changes occurring during postnatal development aside from the contribution of neural activity in shaping the weight of synaptic input amongst synergist motor neurons^30,31. In addition, afferents appear endowed with a coordinate system that directs their axons to terminate into defined dorso-ventral spinal domains, independently of the presence or identity of motor neurons^32–34. These observations suggest that muscle type identity is the manifestation of an intrinsic molecular program that oversees the orderly wiring of sensory-motor circuits.

What are the determinants specifying proprioceptor muscle type identity, how are these molecular signatures acquired, and how do they control central connectivity of Ia afferents? While each of these questions remains largely unanswered, recent studies have begun to offer some clues. Single cell RNA sequencing efforts have revealed the molecular underpinnings of the acquisition of the generic, as well as MS and GTO, proprioceptive fates, yet these studies have thus far failed to capture the molecular traits of muscle type identities^{14,23,24,35–37}. However, transcriptional profiling of selected proprioceptors based on their muscle connectivity, using retrograde labeling techniques, identified molecular markers that specify subtypes innervating either distal dorsal (cdh13; sema5a) or ventral (crtac1; vstm2b) hindlimb muscles³⁸ (Figure 2). These observations indicate that in order to fully resolve proprioceptor muscle type identity, it may be necessary to exhaustively sample proprioceptors based on their peripheral target connectivity.

These studies also provide evidence that proprioceptor muscle type identity may depend on extrinsic factors. Using selective genetic manipulations, the authors showed that the source of the signals that specify distal dorsal or ventral hindlimb muscle type identity was shown to be from the developing limb mesenchyme³⁸. Other studies have implicated the muscle as an alternative source of instructive signals. Embryonic muscles express different levels of NT3, which, at this developmental stage, are associated with correspondingly graded expression levels of Runx3 and Etv1 - two transcription factors known to regulate the extension of afferent collaterals into the ventral spinal cord^{10,11,39–42}. In addition, conditional inactivation of Runx3 after muscle innervation disrupts sensory-motor connectivity without affecting neuronal survival¹¹. Interestingly, afferents that innervate the biceps muscle are disproportionally affected when compared to triceps, indicating that the requirement for Runx3 in regulating central connectivity may vary depending on target-derived extrinsic factors¹¹. Thus, these data support the idea that muscle type identity is at least partially controlled by extrinsic factors. In contrast, evidence supporting a role for an intrinsic genetic program comes from a recent study focusing on the Hox family of transcription factors. As previously shown for various spinal neurons subtypes, Hox expression defines subsets of proprioceptors along the rostro-caudal axis of the spinal cord^43–46 (Figure 2). In particular, Hoxc8 is expressed in afferents innervating flexor muscles of the distal forelimb, and a sensory-specific deletion of Hoxc8 disrupts specificity in sensory-motor connectivity, without affecting afferent survival or selective muscle targeting⁴⁶. Moreover, Hoxc8 expression in proprioceptors is maintained in the absence of either mesenchyme or muscle. Together, these data raise the hypothesis that at a segmental level of specification, broad classes of proprioceptor muscle type identities (e.g. proximal or distal limb, axial), may be regulated by an intrinsic genetic program, while at a finer-grain level, target-derived extrinsic factors act to superimpose a muscle specific fate.

CONCLUSIONS/FUTURE PERSPECTIVES.

A major challenge in understanding the emergence of proprioceptor identities is that their embryonic development is temporally compressed and takes place in a dynamic and spatially diverse environment. As highlighted above, recent studies have begun to resolve the molecular identities of proprioceptor muscle and MS/GTO subtypes. These advances are likely to spur the generation of new genetic tools, enabling deeper molecular and epigenetic profiling of individual subtypes across their development. These studies should help reveal to what extent the developmental trajectories for MS/GTO and muscle type identity intersect, and the various extrinsic signals that influence the maturation of distinct proprioceptor phenotypes. In addition, the ability to permanently label proprioceptor subtypes based on their molecular signatures will enable exploring whether plastic changes in the identity of these afferents – both with respect to peripheral and central targets – can occur under conditions of injury or disease.

Genetic access to specific classes of proprioceptors will, for the first time, also enable exploration of the physiological function of individual subtypes in motor control. Mapping proprioceptor connectivity patterns in the spinal cord and brainstem, along with in vivo recordings and functional imaging during motor behavior^47–50, should provide new insights into the organization and function of groups Ia, II, or Ib proprioceptive feedback circuits in relation to the biomechanical properties of individual muscles. Finally, while group Ia afferents have long enjoyed the limelight, the availability of new genetic resources will permit analysis of all muscle proprioceptors with equal scrutiny. Clearly, we have exciting times ahead of us.

Abbreviations:

DRG

dorsal root ganglion/ganglia

MS

muscle spindle

GTO

Golgi tendon organ

NT3

Neurotrophin 3

RA

Retinoic acid