Etv1 inactivation reveals proprioceptor subclasses that reflect the level of NT3 expression in muscle targets

Abstract

The organization of spinal reflex circuits relies on the specification of distinct classes of proprioceptive sensory neurons (pSN), but the factors that drive such diversity remain unclear. We report here that pSNs supplying distinct skeletal muscles differ in their dependence on the ETS transcription factor Etv1 for their survival and differentiation. The status of Etv1-dependence is linked to the location of proprioceptor muscle targets: pSNs innervating hypaxial and axial muscles depend critically on Etv1 for survival, whereas those innervating certain limb muscles are resistant to Etv1 inactivation. The level of NT3 expression in individual muscles correlates with Etv1-dependence and the loss of pSNs triggered by Etv1 inactivation can be prevented by elevating the level of muscle-derived NT3-revealing a TrkC-activated Etv1-bypass pathway. Our findings support a model in which the specification of aspects of pSN subtype character is controlled by variation in the level of muscle NT3 expression and signaling.

Figure 1. Proprioceptive Sensory Neurons Express TrkC, Rx3, Pv, and Etv1

(A) Schematic illustration of proprioceptor subtype identities. According to subtype identity, proprioceptors exhibit distinct intraspinal collateral trajectories, terminating in medial (m), lateral (l), intermediate (i), or ventral (v) spinal cord. (B) Expression of TrkC, Rx3, and Pv in wild-type rostral lumbar DRG at p0. Arrowheads indicate TrkC⁺Rx3^offPv^offneurons. Asterisk shows a TrkC⁺Rx3⁺Pv^off neuron. Enlarged (boxed) area shows marker expression individually. (C) Percentage of TrkC⁺ sensory neurons expressing Rx3 or Pv in rostral lumbar DRG at p0. (D) Expression of Pv, Rx3 and Etv1 in wild-type rostral lumbar DRG at p0. Arrow head indicates a Pv⁺Rx3^offEtv1^off neuron. Boxed area is enlarged to indicate Rx3⁺Pv^offEtv1 ^offneurons (asterisk). (E) Percentage of Rx3⁺ sensory neurons expressing TrkC, Pv, or Etv1 in rostral lumbar DRG at p0. (F) Percentage of Pv⁺ sensory neurons expressing TrkC, Rx3, or Etv1 in rostral lumbar DRG at p0. (G) Expression of TrkC and GFP in TrkC:GFP reporter mice in rostral lumbar DRG at p0. Inset (boxed area) shows that expression of GFP is restricted to TrkC⁺Rx3⁺ neurons. (H) Expression of Pv in Rx3⁺GFP⁺ neurons in TrkC:GFP reporter mice in rostral DRG at p0. (I) Peripheral endings in Rx3:Cre^ER/Tau:lsl:mGFP-nLZ (Rx3:mGFP-nLZ) mice following a single tamoxifen injection at e17.5. Expression of mGFP is observed in a mosaic pattern in vGluT1⁺ MS and GTO endings in p5 gluteus muscle. (J) mGFP⁺ MS (arrowheads) and GTO (asterisks) sensory endings in p5 gluteus muscle in Pv:Cre/Tau:lsl:mGFP-nLZ (Pv:mGFP-nLZ) mice. Numbers of mGFP⁺ MSs and GTOs in gluteus muscle of Pv:mGFP-nLZ mice are similar to the numbers of vGluT1⁺ MSs and GTOs in WT gluteus. (K) Summary of Pv, Rx3, Etv1, TrkC:GFP, and TrkC expression profiles in proprioceptive and cutaneous mechanoreceptive afferents in DRG. (+) denotes “expression,” (−) denotes “no expression,” (+/−) denotes “expression in a subset of neurons.” Error bars represent SEM. Scale bars represent 20 µm(H), 50 µm (B, D, G, and I), and 200 µm (J). See also Figures S1, S2, S3, and Table S1.

Figure 2. pSN Viability Depends on Etv1

(A) Diagram illustrating the regional identities of rostral lumbar (L1–L2; axial and hypaxial) and caudal lumbar (L3–L5; limb) pSNs. (B) Expression of Rx3 and Etv1 in L2 and L5 DRG of p0 wild-type (Etv1^nLZ/+) and Etv1^−/− mice. Arrows indicate pSNs coexpressing Rx3 and Etv1. Expression of Etv1 is visualized using a nuclear β-galactosidase reporter expressed from the Etv1 locus. (C) Percentage of pSNs (Rx3⁺βGal⁺) in p0 Etv1^−/−DRG in comparison to wild-type. Average number for wild-type L2: 234.5 pSNs (n = 8 DRG), average number for wild-type L5: 564 pSNs (n = 6). The number of neurons in both Etv1^−/− L2 (n = 4) and L5 (n = 4) is significantly reduced compared to wild-type (L2 p < 0.001; L5 p < 0.001; Student’s t test), but the loss of pSNs is more severe in L2 than in L5 DRG (p < 0.001; Student’s t test). (D) Expression of Rx3, TrkC:GFP, and Etv1 in p1–3 rostral lumbar(L1–L2) DRG of Bax1^−/− (i and ii) and Etv1^−/−;Bax1^−/− (iii) mice. Images in (i and ii) are taken of the identical region of the section. Arrows indicate coexpression of Rx3 and TrkC:GFP (i) or Rx3 and βGal (ii and iii). Boxed area in (i and ii) illustrates the small-sized Rx3⁺GFP⁺ neurons that lack Etv1 in Bax1^−/− and Etv1^−/−;Bax1^−/− DRG. (E) Number of pSNs (Rx3⁺nLZ⁺)/section in p1–3 rostral lumbar (L1-L2) DRG of Etv1^−/−, Bax1^−/−, and Etv1^−/−;Bax1^−/− mice compared to wild-type (100% = 7.58 pSNs/section; 19 sections). The percentage of pSNs/section in Etv1^−/− is significantly different compared to both wild-type and Etv1^−/−;Bax1^−/− mice (p < 0.001), whereas the percentage of pSN/section in Bax1^−/− and Etv1^−/−;Bax1^−/− is equivalent (p = 0.15; Mann-Whitney U test). (F) Expression of Rx3 in p6 L2 DRG in wild-type (Etv1^flx/flx or Etv1^flx/+) and Ht-PA:Cre;Etv1^flx/flx animals. (G) Rx3⁺ neurons in L2 DRG in constitutive (Etv1^−/−, n = 3) and sensory neuron-specific (Ht-PA:Cre;Etv1^flx/flx, n = 6) Etv 1 mutants (p = 0.077, Student’s t test). (H) Schematic illustrates proprioceptor segregation into Etv1-dependent (Etv1^D) and Etv1-independent (Etv1^I) subsets. Error bars represent SEM (C, G), or maximum and minimum values (E). Scale bars represent 20 µm (F), 25 µm (D), and 50 µm (B). See also Figure S4 and Table S1.

Figure 3. Etv1-Dependence Does Not Segregate with MS pSNs

(A) Genetic strategy to label MS-innervating pSNs in Egr3:WGA transgenic mice. WGA expressed in the intrafusal muscle fibers of the muscle spindle is taken up by spindle pSN sensory endings and transported to their (Rx3⁺Etv1⁺) cell bodies in DRG. (B) Expression of WGA in p5 gluteus muscle and p7 rostral lumbar DRG in Egr3:WGA transgenic mice. Expression of WGA is readily observed in vGluT1⁺ MS sensory endings, but not in GTO endings. In DRG, accumulation of WGA in Rx3⁺ neurons implies they correspond to MS afferents. WGA is also observed in some Rx3^off neurons (asterisk), reflecting expression of the human Egr3 promoter element in a subset of nonproprioceptive peripheral sensory endings. (C) Expression of Rx3, Etv1 (βGal) and WGA in p7 L2 DRG of Etv1^nLZ/+;Egr3:WGA (WT) and Etv1^−/−;EGR3:WGA mice (images with Rx3/WGA expression are from identical area as images with Etv1 (βGal)/WGA expression). In wild-type, coexpression of Rx3, Etv1, and the accumulation of WGA defines MS-afferent cell bodies (arrow heads). Asterisk indicates a WGAo^ffRx3⁺Etv1⁺ pSN presumed to correspond to a GTO-afferent. Rx3⁺Etv1 ^offcutaneous mechanoreceptors (arrow) also lack accumulation of WGA. In DRG of Etv1^−/− mice, WGA⁺ as well as WGA^off pSNs can be observed. (D) Number of WGA⁺ pSNs (defined by Rx3 and Etv1^nLZ coexpression; red bars) in comparison to total numbers of pSNs in wild-type (black bars) and Etv1 ^−/−(gray bars) L2 at p7–10(n = 6 DRG for wild-type L2 and L5 DRG; for Etv1^−/− DRG, n = 6 for L2 and n = 4 for L5).The number of WGA⁺ pSNs in L2 and L5 DRG of Etv1 mutants is significantly reduced compared to wild-type (L2 p < 0.001, L5 p < 0.001; Student’s t test). (E) Analysis of WGA expression in MS intrafusal fibers in p5 Etv1^−/−;EGR3:WGA Soleus muscle. The number of vGluT1⁺ MS sensory endings that colocalizes with WGA expression is reduced in Etv1 mutant muscle (sample includes: soleus, medial gastrocnemius, extensor digitorum longus, tibialis anterior, plantaris) when compared to wild-type muscle. Error bars represent SEM. Scale bars represent 20 µm (C), 50 µm (B; DRG, E) and 100 µm (B; muscle). See also Figure S5 and Table S3.

Figure 4. Selective Loss of pSN Innervation in Etv1 Mutant Mice

(A) Analysis of vGluT1⁺ spindle-associated sensory endings (SSEs) in axial, hypaxial (body wall), and limb muscle (soleus), in wild-type and Etv1^−/− mice at p3–5. Insets in soleus panels compare the levels of vGluT1 expression in wild-type and Etv1^−/− SSEs. Schematic insets indicate plane of image. (B) Analysis of MSs in hindlimb muscle from wild-type and Etv1^−/− mice using an Etv4^nLZ reporter allele expressing βGal in the nuclei of intrafusal muscle fibers. Boxed area shows reduced levels of βGal activity in Etv1^−/− EDL MS, implying a reduced level of Etv4 expression. (C) Number of vGluT1⁺ SSEs in axial, hypaxial, and limb muscle targets in wild-type and Etv1^−/−mice. Individual points represent individual SSE counts in defined area/animal (see Supplemental Experimental Procedures). Counts for limb muscles are the average number of SSEs observed for a set of muscles (described in D)/animal. Median values are represented by a red dotted line. The number of SSEs in Etv1^−/− muscle was significantly reduced compared to wild-type (Mann-Whitney U test). (D) Percentage of remaining vGluT1⁺ SSEs in Etv1^−/− mice when compared to wild-type. Muscles analyzed were gluteus (Gl), biceps femoris (BF), quadriceps (Q; rectus femoris, vasti), semitendinosus (St), semimembranosus (Sm), gastrocnemius (G; lateral and medial), soleus (Sol), plantaris (Pl), tibialis anterior (TA), extensor digitorum longus (EDL), and peroneus longus (PL) and brevis (PB). Error bars represent SEM. Many muscles showed a significant reduction in the number of SSEs when compared to wild-type, but others did not (Mann-Whitney U test). (E) Initial formation of SSEs at e15.5 in hypaxial muscles of wild-type and Etv1^−/− mice as analyzed by expression of endogenous TrkC protein (wild-type and Etv1^−/−) and the TrkC:GFP reporter (Etv1^−/−). Scale bars represent 20 µm (E) and 100 µm (A). See also Figures S6 and S7.

Figure 5. Differential pSN Etv1-Dependence Is Reflected in Central Projection Phenotypes

(A) Choleratoxin-B (CTB) labeling of (ChaT⁺) motor neurons innervating intercostal, body wall, and distal hindlimb muscle, which are targeted by collaterals from pSNs located in T9, L2, and L5 DRG, respectively. (B) Collateral trajectories of pSNs residing in T9, L2 and L5 DRG in wild-type at p5–6, visualized using RhD dorsal root fills. Boxed areas indicate the medial (M) and lateral (L) ventral horn compartments used in quantitative analysis comparing the percentage of medial and lateral projections. (C) Analysis of collateral density of RhD⁺ T9, L2, and L5 medial and lateral trajectories in Etv1^−/−and Etv1^−/−;mlc1NT3⁺ mice at p5–6, in comparison to wild-type. Collateral densities were expressed as mean pixel intensity (mpi), calculated using ImageJ analysis software; see Experimental Procedures). Values for M and L cannot be compared across levels as regions are defined differently for each segmental level (see B). Error bars represent SEM. Scale bars represent 50 µm.

Figure 6. pSN Etv1-Dependence Correlates with Muscle Target NT3 Levels

(A) Expression of NT3, βGal (expressed from the NT3:lacZ allele), and Rx1 in lumbar DRG at p0. NT3 and βGal are virtually absent from L1–2 DRG, but are abundantly expressed in L4–L5 DRG. Boxed area is enlarged in (B). (B) Expression of βGal in L4/5 DRG of NT3:lacZ mice can be observed in Rx1⁺ cutaneous neurons but is excluded from Rx3⁺ pSNs. (C) Percentage of pSNs (Rx3⁺Pv⁺) that remains in L5 DRG of Etv1^−/−(n = 4), and Etv1^−/−;Ht-PA:Cre; NT3^flx/flx mice at p0–5 is similar (p = 0.707; Student’s t test). (D) βGal-activity levels in e15.5 embryonic hindlimb muscles of NT3:lacZ mice. Muscles indicated are rectus femoris (RF), soleus (Sol), extensor digitorum longus (EDL), Tibialis anterior (TB), and biceps femoris (BF). (E) Relative quantity of NT3 mRNA detected by qRT-PCR in embryonic (e15–e16) body wall (BW), tibialis anterior (TA), and soleus (Sol) muscle. Data was obtained from four independent experiments and represent relative quantity normalized to MyoD. (F) Muscle NT3 expression levels correlate inversely with pSN Etv1-dependence. Error bars represent SEM. Scale bars represent 50 µm (A).

Figure 7. Elevated Levels of Muscle NT3 Ameliorates Etv1 Mutant Phenotypes

(A) Expression of Rx3 in wild-type, mlc1NT3⁺, Etv1^−/−, and Etv1^−/−;mlc1NT3⁺ rostral lumbar (L1-L2) DRG at p0. (B) Percent change in the number of pSNs in L2 and L5 DRG in Etv1^−/− (L2: n = 4; L5: n = 4), mlc1NT3⁺ (L2: n = 6; L5: n = 2) and Etv1^−/−;mlc1NT3⁺ (L2: n = 4; L5: n = 4 DRG) mice in comparison to wild-type L2 (n = 8) and L5 (n = 6) DRG at p2–8. The increase in pSNs was not different between mlc1NT3⁺ and Etv1^−/−;mlc1NT3⁺ DRG (Student’s t test). Error bars represent SEM. (C) Percentage of pSNs in L2 and L5 NT3^+/− DRG at p0–2 compared to WT. (D) vGluT1⁺ SSEs in body wall muscle in wild-type and Etv1^ETS/ETS;mlc1NT3 mice at p7. Scale bar represents 200 µm. See also Table S2.

Figure 8. Activities of NT3, Etv1, and Rx3 in pSN Differentiation

(A) Loss of Etv1 uncovers the existence of an alternate (Alt.) NT3-activated signaling pathway that is progressively activated with increased NT3/TrkC signaling. The relative weight of both pathways is a function of the ambient levels of muscle-derived NT3. Both Etv1 and Alt. pathways may control pSN survival and/or subtype specialization, with afferents from hypaxial, axial and certain limb muscles (limb^s) primarily dependent on Etv1 signaling, whereas afferents from other limb muscles show an intermediate requirement (limbⁱ) for Etv1 or are refractory (limb^r) to the loss of Etv1. (B) Intersection of TrkC, Rx3, and Etv1 signaling pathways in pSN differentiation. At early stages (e12.5) Rx3 solidifies pSN identity by repressing TrkB and promoting TrkC expression, whereas later (>e13.5) Rx3 activity levels control the extent of pSN collateral projections into the ventral horn (Chen et al., 2006). Varying levels of Etv1-dependency reflect the differential requirement for Etv1 in the development of distinct, muscle-specific, pSN subtypes.