Sox11 regulates survival and axonal growth of embryonic sensory neurons

Authors

  • L. Lin,

    1. Department of Developmental Biology, University of Texas Southwestern Medical Center, Dallas, Texas
    Search for more papers by this author
    • Drs. Lin, Lee, and Wang contributed equally to this work.

  • V.M. Lee,

    1. Division of Developmental Biology, Department of Pediatrics, Medical College of Wisconsin, Milwaukee, Wisconsin
    Search for more papers by this author
    • Drs. Lin, Lee, and Wang contributed equally to this work.

  • Y. Wang,

    1. Department of Biology and Center for Excellence in the Neurosciences, University of New England, Biddeford, Maine
    Search for more papers by this author
    • Drs. Lin, Lee, and Wang contributed equally to this work.

  • J.S. Lin,

    1. Department of Neurosurgery, Medical College of Wisconsin, Milwaukee, Wisconsin
    Search for more papers by this author
  • E. Sock,

    1. Institut für Biochemie, Emil-Fischer-Zentrum, Friedrich-Alexander-Universität, Erlangen-Nürnberg, Erlangen, Germany
    Search for more papers by this author
  • M. Wegner,

    1. Institut für Biochemie, Emil-Fischer-Zentrum, Friedrich-Alexander-Universität, Erlangen-Nürnberg, Erlangen, Germany
    Search for more papers by this author
  • L. Lei

    Corresponding author
    1. Department of Biology and Center for Excellence in the Neurosciences, University of New England, Biddeford, Maine
    • Department of Biology and Center for Excellence in the Neurosciences, University of New England, Biddeford, ME 04005
    Search for more papers by this author

Abstract

Sensory neurons transduce various stimuli including temperature, pain, and touch from the periphery to the central nervous system. Sensory neuron development is governed by a combination of extracellular cues and specific gene expression. We demonstrated that the transcription factor Sox11 was highly expressed in the developing sensory neurons. To test the function of Sox11, we used a knockin mouse model where the entire coding region of Sox11 was replaced by a LacZ reporter. The ablation of Sox11 caused severe reduction in sensory neuron survival in the trigeminal and dorsal root ganglia, although it did not affect migration of neural crest cells or acquisition of major sensory neuron subtypes. We further demonstrated that ablating Sox11 caused an arrest of axonal outgrowth in vivo and in vitro. This defect could not be fully rescued by blocking cell death. Our data suggest that Sox11 is a key regulator of sensory neuron development. Developmental Dynamics, 2011. © 2010 Wiley-Liss, Inc.

INTRODUCTION

Sensory ganglia are neural crest-derived clusters of primary neurons in the peripheral nervous system (PNS). Sensory information is relayed through sensory neurons in the PNS to the central nervous system (CNS) for further processing. Somatosensory information from the trunk and limbs is transduced to the spinal cord through sensory neurons residing in the dorsal root ganglia (DRG; Scott, 1992), while facial sensory information is relayed to the brain stem by trigeminal ganglion (TG) sensory neurons (Waite, 1995).

Several transcription factors have been shown to play important cell-intrinsic roles in various stages of sensory neuron development (Marmigere and Ernfors, 2007). The basic helix–loop–helix (bHLH) transcription factors Neurogenin 1 (Ngn1) and Neurogenin 2 (Ngn2) are required for neurogenesis and specification of peripheral sensory neurons (Fode et al., 1998; Ma et al., 1998, 1999). Kruppel-like zinc finger transcription factor Klf7 and POU domain transcription factor Brn3a are required for the survival of sensory neurons through regulating the expression of Trk receptors (Huang et al., 1999; Eng et al., 2001; Ma et al., 2003; Lei et al., 2005; Lei et al., 2006). Brn3a also regulates axonal growth and target innervations of sensory neurons independent of Trks (Eng et al., 2001). Runx family transcription factors play important roles in sensory neuron subtype specification and diversification. Runx1 is required specifically for development of TrkA-positive nociceptive neurons, while Runx3 is important for development of TrkC-positive proprioceptive neurons (Levanon et al., 2002; Chen et al., 2006; Kramer et al., 2006). In addition, ETS-domain transcription factors ER81 and PEA3, paired homeodomain transcription factor DRG11, and Runx3 control the connection between sensory neurons and their central targets (Lin et al., 1998; Chen et al., 2001; Chen et al., 2006).

The Sry-related box (Sox) genes encode a group of transcription factors with a high mobility group (HMG) -type DNA binding domain (Bowles et al., 2000; Kamachi et al., 2000; Schepers et al., 2002). Multiple members of the Sox family transcription factors are involved in the specification and maintenance of neural progenitor identity in the CNS and PNS (Pevny and Placzek, 2005; Wegner and Stolt, 2005). Among them, the SoxE group of factors (Sox8, Sox9, and Sox10) have specific expression and functions in neural crest cell development (Cheung and Briscoe, 2003; Kim et al., 2003). For example, Sox9 is required for initiating neural crest cell development and specification of cranial neural crest into chondrocytes (Cheung and Briscoe, 2003). Sox10 maintains neural crest cell multipotency and inhibits or delays their differentiation toward neuronal lineage (Kim et al., 2003).

SoxC, another group of Sox transcription factors which comprise of Sox4, Sox11, and Sox12, are expressed in both CNS and PNS starting from early embryonic stages (Hargrave et al., 1997; Kuhlbrodt et al., 1998). Although they have overlapping expression and transactivation, targeted deletion of each SoxC factor revealed different phenotypes, suggesting that they each have distinct functions that cannot be compensated. Sox4 and Sox11 have been demonstrated to be important for nervous system development. In the CNS, overexpression of Sox11 promotes neuronal maturation in the developing chick neural tube (Bergsland et al., 2006). In the PNS, a recent report showed that Sox4 and Sox11 are required for sympathetic neuron development (Potzner et al., 2010). We are particularly interested in Sox11, whose expression is the most robust in the nervous system. Expression of Sox11 is abundant in embryonic but at very low levels in adult sensory ganglia (Tanabe et al., 2003). After peripheral nerve injury, Sox11 mRNA is up-regulated in adult DRG (Tanabe et al., 2003). In vitro culture suggests that Sox11 is necessary for survival and axonal growth of adult sensory neurons (Jankowski et al., 2006). Using RNA interference, Sox11 has also been shown to modulate peripheral nerve regeneration in adult mice (Jankowski et al., 2009). However, the in vivo function of Sox11 during the embryonic development of sensory neurons has not been explored.

In this study, we investigate the role of Sox11 in embryonic sensory neuron development using a transgenic mouse model where the exon encoding the entire Sox11 open reading frame is replaced by a nuclear localized β-galactosidase (lacZ) reporter gene (Sock et al., 2004), essentially creating a Sox11 null animal. We demonstrate that Sox11 is expressed in embryonic sensory neurons in both dorsal root and trigeminal ganglia. Deletion of Sox11 increases neuronal cell death and causes reduction of all sensory neuron subtypes in dorsal root and trigeminal ganglia. The absence of Sox11 also leads to defects in axonal growth of sensory neurons that is largely independent of cell death. Our data demonstrate that Sox11 is essential for the survival of sensory neurons as well as neurite outgrowth.

RESULTS

Sox11 Is Expressed in Embryonic Sensory Neurons

Previous reports showed that Sox11 mRNA is widely expressed in the developing CNS and PNS (Hargrave et al., 1997; Kuhlbrodt et al., 1998) and the expression of Sox11 is evident in dorsal root and trigeminal ganglia by E11.5 (Hargrave et al., 1997). To specifically localize Sox11 expression in the newly formed sensory ganglia and the CNS, we performed Sox11 in situ hybridization on head and trunk sections in embryonic day (E) 10.5 embryos. Our results demonstrated that Sox11 is prominently expressed in trigeminal ganglion, eye, and the neuroepithelium of the brain (Fig. 1A–D). To test if trigeminal neurons express Sox11, we immunostained sections with a neuronal marker, Tuj1, after Sox11 in situ hybridization. We found that many neurons co-express Sox11 and Tuj1 (Fig. 1B–D). In the trunk, Sox11 is expressed throughout the neuroepithelium of the neural tube and dorsal root ganglia at E10.5 (Fig. 1E). Double labeling with Tuj1 showed that dorsal root ganglia neurons express Sox11 (Fig. 1F–H).

Figure 1.

Sox11 is expressed in the developing nervous system. A: RNA in situ hybridization performed on wild-type embryonic day (E) 10.5 sections showed that Sox11 was expressed in the neuroepithelium of the forebrain (FB), midbrain (MB), trigeminal ganglion (Tg), and the eye. B–D: Higher magnification of the trigeminal ganglion. B: Single channel Sox11 in situ signal. C: Tuj1 immunostaining on the same section as B identified neurons in the trigeminal ganglion. D: Merged image of B and C demonstrated that Sox11 was expressed in trigeminal neurons. E: In the trunk, Sox11 was expressed throughout the neuroepithelium of the neural tube (NT) and dorsal root ganglia (DRG). F–H: Higher magnification of the boxed region in E. F: Single channel Sox11 in situ signal in the dorsal root ganglion. G: Tuj1 immunofluorescence identified neurons in the dorsal root ganglion. H: Merged image of F and G revealed that Sox11 transcript could be detected in dorsal root ganglion neurons. DRG, dorsal root ganglion; FB, forebrain; FL, forelimb; NT, neural tube; Tg, trigeminal ganglion. Scale bar = 100 μm in A,E, B–D = 50 μm in B–D, 50 μm in F–H.

Because sensory neurons in the trigeminal and dorsal root ganglia appeared to have high levels of Sox11, we sought to further characterize its expression in relation to neuronal differentiation. We took advantage of the Sox11+/− embryos, in which the complete open reading frame of Sox11 is replaced with a β-galactosidase (lacZ) reporter gene such that lacZ expression reflects the endogenous Sox11 expression patterns (Sock et al., 2004). To examine whether Sox11 expression in sensory ganglia is restricted to neurons, we performed double labeling experiments on dorsal root and trigeminal ganglia sections from Sox11+/− embryos and found that all lacZ-positive cells in dorsal root and trigeminal ganglia co-express Tuj1 at E11.5 (Fig. 2A–F). We also observed a small number of Tuj1-positive neurons that were lacZ negative, suggesting that not all sensory neurons express Sox11 (arrowheads in insets of Fig. 2). Similar expression patterns were observed in DRG and TG from E12.5 and E13.5 animals (data not shown). Taken together, these data demonstrate that Sox11 is expressed in embryonic sensory neurons.

Figure 2.

Sox11 is expressed in sensory neurons. A–F: Tuj1 (green) and lacZ (red) double labeling of dorsal root ganglia (DRG; A–C) and trigeminal ganglion (TG; D–F) sections from embryonic day (E) 11.5 Sox11+/lacZ embryos showed that the majority of lacZ-positive cells were also Tuj1 positive (C,F), thus are neurons. Some Tuj1-positive neurons were negative for lacZ (arrowheads in C and F, inset). G–L: BrdU (green) and lacZ (red) double labeling of E11.5 DRG (G–I) and TG (J–L) sections showed that lacZ-positive cells were not positive for bromodeoxyuridine (BrdU), thus are not proliferating progenitors. Insets depict higher magnification images, showing the colocalization of lacZ and Tuj1 signals in most cells (insets in C and F), and the nonoverlapping lacZ and BrdU signals (insets in I and L). Scale bars = 100 μm in G–L, 50 μm in A–F, 10 μm in insets.

Neurogenesis in sensory ganglia occurs in distinctive waves both during neural progenitor cell migration and after the initial formation of ganglia (Marmigere and Ernfors, 2007). To test if Sox11 is expressed in progenitor cells in the sensory ganglia, we pulsed pregnant females with the thymidine analog bromodeoxyuridine (BrdU) at E11.5 when neurogenesis in TG and DRG is readily detectable. We did not detect any LacZ+ cells that were also BrdU+ in TG and DRG (Fig. 2G–L), indicating that Sox11 is not expressed in dividing precursor cells within these sensory ganglia during the peak of sensory neurogenesis.

All Subtypes of Sensory Neurons Are Depleted in Sox11−/− Mice

To investigate the impact of the ablation of Sox11 on sensory neuron development, we first performed Nissl staining and quantified the number of sensory neurons at P0, when Sox11−/− mice die of major congenital heart and lung defects (Sock et al., 2004). We found that the number of trigeminal neurons in Sox11−/− mice is dramatically decreased compared with controls (Fig. 3I–K). To examine whether this cell loss affected a particular class of neurons in the ganglia, we performed immunostaining using antibodies against the neurotrophin receptors TrkA, TrkB, TrkC, and Ret (Fig. 3A–H). A similar percentage decrease was observed in all four subtypes of sensory neurons in Sox11−/− TG compared with controls (Fig. 3K), suggesting that the loss of Sox11 leads to increased cell death of all sensory neuron subtypes. Significant loss of DRG neurons was also present in E18.5 Sox11−/− embryos at several cervical levels compared with controls (Fig. 4I–K). Similar to the trigeminal ganglia, all subpopulations of DRG neurons (based on the expression of Trk receptors and Ret) were reduced at a comparable proportion (Fig. 4A–H). Thus, these data suggest that Sox11 is required for the survival of all subtypes of sensory neurons in vivo.

Figure 3.

Different subtypes of sensory neurons are lost in postnatal day (P) 0 Sox11−/− trigeminal ganglion (TG). Nissl staining revealed that at P0, the total neuronal number decreased by approximately 60% in Sox11−/− TG (J) compared with control (I, K, N = 4 for each genotype). A–H: The neuronal loss in Sox11−/− TG was not restricted to any subtype as shown by immunostaining using antibodies for neurotrophic receptors: TrkA (A,B), TrkB (C,D), TrkC (E,F), and Ret (G,H). The number of each type of sensory neurons in Sox11−/− TG all decreased at a similar proportion (K, N = 2 for each genotype). Student's t-test. Data represents mean ± SEM. *P < 0.05. Scale bars = 50 μm in A–H, 100 μm in I–L.

Figure 4.

Loss of different subtypes of sensory neurons in embryonic day (E) 18.5 Sox11−/− dorsal root ganglia (DRG). I–K: Nissl staining revealed that at E18.5, a significant decrease of neuronal number was observed in Sox11−/− DRG (J) compared with control (I) at several levels (K, cervical 1–3, N = 2 for each genotype at each cervical level examined). A–H: Immunofluorescent staining for different markers revealed that at this time point, all subtypes of sensory neurons at the cervical level were reduced in Sox11−/− DRG compared with control: 67.6% for TrkA (A,B), 78.8% for TrkB (C,D), 79.2% for TrkC (E,F), and 70.2% for Ret (G,H), respectively (≥11 sections were counted for each marker). Percentage was calculated by dividing the number of specific marker-positive cells in Sox11−/− by that in the control. Student's t-test. Data represent mean ± SEM. *P < 0.05. Scale bars = 100 μm.

To ask if the initial formation of sensory neurons is dependent on Sox11 activity, we examined the number of neurons in sensory ganglia at earlier time points. No significant difference was observed in E12.5 Sox11−/− TG (Fig. 5K–M) or DRG (Fig. 5N) compared with controls as revealed by Nissl staining. Moreover, the percentage of different subtypes of sensory neurons (TrkA, TrkB, TrkC, and Ret positive, respectively) was not significantly different in Sox11−/− TG compared with controls (Fig. 5A–J), suggesting that the loss of Sox11 does not affect the formation of major sensory neuron subtypes during early development.

Figure 5.

The loss of Sox11 does not affect the acquisition of major sensory neuron subtypes revealed by neurotrophic receptor expression in embryonic day (E) 12.5 sensory ganglia. K–M: Nissl staining revealed that at E12.5, the total neuronal number did not change in Sox11−/− trigeminal ganglion (TG; L) compared with control (K and M, N = 2 for each genotype). The number of DRG neurons in Sox11−/− was also comparable to controls. N: There was no significant change of neuronal number in Bax−/− or Sox11−/−;Bax−/− DRG compared with control or Sox11−/− (N, N = 2 for Bax−/−, N = 6 for the other three genotypes). A–J: Adjacent sections from control (A,C,E,G,I) and Sox11−/− (B,D,F,H,J) E12.5 TG were subjected to immunostaining using antibodies for TrkA (A,B), TrkB (C,D), TrkC (E,F), Ret (G,H), and Tuj1 (I,J). The percentage of different sub-types of neurons was comparable between control and Sox11−/− (TrkA+: 80.8% vs. 81.7%; TrkB+: 14.9% vs. 12.3%; TrkC+: 25.5% vs. 20.6% and Ret+: 9.6% vs. 8.5%). The numbers of specific marker-positive cells were counted, and the percentages were calculated by dividing that by the total number of Tuj1-positive cells within the same section. A total of 1458 and 2010 Tuj1-positive cells were counted in control and Sox11−/−, respectively. Student's t-tests were performed. Data represent mean ± SEM. Scale bars = 50 μm in A–J, 100 μm in K,L.

Loss of Sox11 Does Not Affect Neural Crest Cell Migration or Cell Proliferation in Sensory Ganglia

To understand the mechanism of cell loss in the sensory ganglia in the absence of Sox11, we first compared the migration of neural crest cells in control and Sox11−/− embryos by performing whole-mount in situ hybridization with a neural crest marker, Sox10. We saw no obvious difference in the pattern of neural crest cell migration (Fig. 6A–C). This suggests that deleting Sox11 does not affect initial formation or migration of neural crest precursors.

Figure 6.

The loss of Sox11 does not affect pattern of neural crest migration or cell proliferation in sensory ganglia. A–C:Sox10 whole-mount in situ hybridization demonstrated that the overall pattern of neural crest migration appeared normal in Sox11−/− embryos (C), compared with wild-type (A) and heterozygous (B) littermates. D–G: Dorsal root ganglia (DRG; D,F) and trigeminal ganglion (TG; E,G) sections from control (D,E) and Sox11−/− (F,G) embryonic day (E) 11.5 embryos were subjected to bromodeoxyuridine (BrdU) immunostaining (red, BrdU; green, DAPI). H: The number of BrdU-positive cells did not change in Sox11−/− DRG or TG compared with controls (N = 2 for each genotype), indicating that proliferation of neural crest cell progenitors at this time point is not affected by the loss of Sox11. Student's t-tests were performed. Data represent mean ± SEM. Scale bars = 100 μm in E,G, 50 μm in D,F.

We also examined whether cell proliferation in sensory ganglia is impaired in the absence of Sox11. The number of BrdU-positive cells in E11.5 DRG and TG were quantified but no significant difference was detected between Sox11−/− and controls (Fig. 6D–H), consistent with the finding that lacZ+ and BrdU+ cells did not colocalize in sensory ganglia at this time point (Fig. 2G–L). Therefore, Sox11 is not required for proliferation of neural progenitors in sensory ganglia during the peak of neurogenesis.

Loss of Sox11 Causes Increased Cell Death in Sensory Ganglia

Next, we examined whether deleting Sox11 might alter survival of sensory neurons by using an antibody against activated caspase-3 (AC3), a marker for apoptotic cells. At E12.5, a significant increase of AC3(+) cells was detected in Sox11−/− DRG (Fig. 7B) and TG (Fig. 7F) compared with controls (Fig. 7A and E, respectively). At E13.5, increased cell death was observed in Sox11−/− DRG (P = 0.06), while there was no difference in cell death between Sox11−/− and control TG (Fig. 7C,G). This suggests that Sox11 supports the survival of sensory neurons at critical developmental stages.

Figure 7.

Increased cell death in Sox11−/− sensory ganglia. A,B,E,F: Dorsal root ganglia (DRG; A,B) and trigeminal ganglion (TG; E,F) sections from control (A,E), Sox11−/− (B,F), Bax−/− and Sox11−/−;Bax−/− embryos were subjected to cleaved caspase 3 staining. At E12.5, there was a significant increase of caspase 3-positive cells in Sox11−/− DRG and TG (B and F, respectively) compared with controls (A and E, respectively). C,G: At E13.5, there was also an increase of caspase 3-positive cells in Sox11−/− DRG (C, P = 0.06), but no difference was observed in TG at this time point (G). D,H: At E12.5, significant decrease of caspase 3-positive cells was observed in both DRG and TG (D and H, respectively) from Sox11−/−;Bax−/− embryos compared with Sox11−/− and controls, suggesting that the increased cell death in Sox11−/− sensory ganglia can be rescued by the additional loss of Bax. N = 4 for each genotype at each time point. Student's t-test. Data represent mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.005. Scale bars = 100 μm.

By double labeling AC3 and β-galactosidase (LacZ) in Sox11−/− embryos, we found that many dying cells in the mutant were LacZ (Sox11)-positive (Fig. 8B); we also observed AC3+/Sox11+ cells in control DRG (Fig. 8A). This supports the notion that Sox11 is expressed in DRG neurons when they are most sensitive to apoptosis, and is compatible with at least a partially cell-autonomous role of Sox11 in survival. It is also worth noting that there were AC3+ cells that were negative for Sox11 or β-gal in control or Sox11−/− DRG, respectively, consistent with the naturally occurring cell death in sensory ganglia at this developmental stage (White et al., 1998).

Figure 8.

Sox11-positive cells undergo apoptosis in the mutant dorsal root ganglia (DRG). A,B: Transverse sections of embryonic day (E) 12.5 embryos from wild-type (A) and Sox11−/− (B) were subjected to immunofluorescence using antibodies against activated caspase-3 (AC3, green), Sox11 (in A, red), or β-gal (in B, red). DAPI (4′,6-diamidine-2-phenylidole-dihydrochloride; blue) was used to label the nuclei. Arrows indicate Sox11-positive or β-gal–positive cells that are undergoing apoptosis. N = 4 embryos of each genotype were examined. Scale bar = 50 μm.

To examine whether the increased cell death in the absence of Sox11 depends on the pro-apoptotic gene Bax (White et al., 1998), we crossed Bax+/− with Sox11+/− mice and examined the number of activated caspase 3-positive cells in DRG and TG from Sox11−/−;Bax−/− embryos. Our results revealed that compared with Sox11−/−, DRG (Fig. 7D) and TG (Fig. 7H) from Sox11−/−;Bax−/− embryos had significantly fewer number of activated caspase 3-positive cells, suggesting that the loss of Bax rescues Sox11−/− sensory neurons from apoptosis in vivo.

Sox11−/− DRG Neurons Exhibit Decreased Nerve Growth Factor–Dependent Survival In Vitro

Next, we cultured DRG neurons in dissociation and examined their survival in vitro. In the presence of nerve growth factor (NGF; 10 ng/ml), DRG neurons isolated from E13.5 Sox11−/− embryos showed a significant decrease in survival when compared with controls (Fig. 9A). This finding supports the notion that Sox11 is required for the survival of sensory neurons in a cell-autonomous manner.

Figure 9.

In vitro survival and axonal growth defects of Sox11−/− DRG neurons. A: Embryonic day (E) 13.5 dissociated dorsal root ganglia (DRG) neuronal culture in the presence of nerve growth factor (NGF; 10 ng/ml) revealed a significant decrease of neuronal numbers in Sox11−/− cultures compared with controls at 48 hr and 72 hr after plating. All data were normalized to controls for each time point (Duplicate wells each from two embryos for each genotype were used for each time point). B: DRG explants were taken from E13.5 control and Sox11−/− embryos, cultured in NGF containing media for 30 hr and then subjected to 2H3 immunostaining. C: Quantification of the average axonal length of DRG explants as shown in B. Sox11+/+ (N = 16) and Sox11+/− (N = 29) explants exhibited comparable average axonal length, while significant decrease of average axonal length was observed in Sox11−/− DRG explants (N = 9). D: Even in the presence of apoptosis inhibitor BAF, Sox11−/− DRG explants still had shorter axons compared with controls (Sox11+/+ and Sox11+/−), suggesting that the axonal growth defect of Sox11−/− DRG neurons in vitro is at least partially independent of cell death. Student's t-test. *P < 0.05, **P < 0.01, ***P < 0.005. Data represent mean ± SEM. Scale bar = 200 μm.

Sox11−/− DRG Neurons Exhibit Decreased Axonal Growth In Vitro

To examine whether Sox11 is required in DRG for regulating axonal growth, we cultured E13.5 DRG explants in the presence of NGF (10 ng/ml) and examined their axonal growth properties in vitro. Explant cultures revealed a significant decrease of the average axonal length in Sox11−/− DRG compared with controls (Fig. 9B,C). The defect in axonal growth was maintained in Sox11−/− DRG explants even in the presence of the apoptosis inhibitor BAF (50 μM; Fig. 9D). Therefore, these data demonstrate that Sox11 is required for axonal growth of sensory neurons in vitro in a cell-death-independent manner.

Reduced Axonal Growth of Sensory Neurons in Sox11−/− Embryos

To examine whether Sox11 also regulates axonal growth of sensory neurons in vivo, we performed whole-mount neurofilament staining in E12.5 embryos. Projections from both trigeminal and DRG neurons are complex. We focused our analyses on cutaneous projections of the ophthalmic branch (ONB) of the trigeminal nerve and the lateral cutaneous branch (LCB) of the spinal nerve as these nerves are well developed in wild-type embryos at this developmental stage and have reached their targets (Lonze et al., 2002). Compared with controls (Fig. 10A,E), these projections were shorter and less branched in Sox11−/− mice (Fig. 10B,F). A large number of E12.5 embryos were examined, and the axonal growth defects were fully penetrant.

Figure 10.

Cell death-independent axonal growth defects in Sox11−/− embryos. A–H: Embryonic day (E) 12.5 embryos of control (A,E; n = 63), Sox11−/− (B,F; n = 15), Bax−/− (C,G; n = 9), and Sox11−/−;Bax−/− (D,H; n = 4) were subjected to whole-mount neurofilament staining. The number of embryos examined was also listed on each panel. Cutaneous projections of ophthalmic branches (ONB) of the trigeminal nerve (arrowheads in A) and projections of lateral cutaneous branch (LCB) of the spinal nerve into the hind limb (arrowhead in E) were indicated. Compared with controls (A,E), fewer and shorter nerve branches at these locations were observed in Sox11−/− embryos (B,F). The additional loss of Bax did not completely rescue these axonal growth defects (C,D; and G,H), indicating that the axonal growth defect in Sox11−/− embryos is at least partially independent of cell death. Scale bar = 1 mm.

To test whether the axonal growth defect in Sox11−/− embryos in vivo is affected by neuronal apoptosis, we examined the Sox11−/−;Bax−/− embryos in which cell death of sensory neurons was abolished. As shown in Figure 5N, the number of DRG neurons in E12.5 Bax−/− and Sox11−/−;Bax−/− embryos was not significantly different from control or Sox11−/− embryos. However, the axons in the Bax−/−;Sox11−/− embryos were still less extensive than in control or Bax−/− embryos (Fig. 10C,D,G–H). Although these axons in the Sox11−/−; Bax−/− embryos were more extensive than in Sox11−/− embryos, suggesting that the loss of Bax might have partially rescued the axonal growth defects, the rescue was not complete. Taken together, these data indicate that Sox11 is required for in vivo axonal growth of sensory neurons at least partially in a cell death-independent manner.

DISCUSSION

Several classes of transcription factors have been shown to regulate sensory neuron development (Marmigere and Ernfors, 2007). Sox11, a SoxC group transcription factor, is expressed in sensory neurons at a high level during embryonic development but down-regulated in adult DRG. Its expression is up-regulated in adult DRG after peripheral nerve injury (Tanabe et al., 2003). In vitro experiments using RNA interference (RNAi) suggested that Sox11 might be required for survival and axonal growth of adult DRG neurons (Jankowski et al., 2006). In this study, we apply a genetic approach and use a transgenic mouse that has complete disruption of Sox11, thus providing an in vivo model to examine the physiological functions of this gene in sensory neurons.

Our results demonstrate that Sox11 is not required for neural crest cell migration, initial sensory ganglia formation, or proliferation of sensory neuron progenitors in Sox11 deficient mice. Moreover, Sox11 is not required for the specification of the major sensory neuron subtypes as revealed by the normal expression of the neurotrophin receptors in E12.5 Sox11−/− TG. However, Sox11 is necessary for the survival of embryonic sensory neurons, and the ablation of this gene leads to the reduction of all subtypes of sensory neurons. Sox11 is also necessary for normal axonal growth in a cell death-independent manner both in vitro and in vivo. Therefore, our data establish Sox11 as a transcription factor that controls several key steps of sensory neuron development both in vivo and in vitro.

SoxC proteins are also important for development of the sympathetic nervous system (Potzner et al., 2010). In the sympathetic ganglia, Sox11 is predominantly expressed in actively dividing neural progenitors and not so much in postmitotic cells; while in DRG and TG, Sox11 is predominantly expressed in postmitotic neurons but not in dividing cells. Sox4 and Sox11 were shown to regulate both neuronal proliferation and survival in the sympathetic ganglia, although the survival function segregated more with Sox4 than Sox11. To the contrary, our data here demonstrate that Sox11 is important for the survival but not proliferation of sensory neurons. These comparisons suggest that the effect on survival may map to different SoxC proteins in different types of peripheral neurons. The effect of SoxC proteins on axonal growth of sympathetic neurons remains to be examined.

In E11.5 sensory ganglia, the expression of Sox11 was restricted to neurons but not in proliferating cells as revealed by Tuj1/lacZ and BrdU/lacZ double labeling. Furthermore, cell proliferation in both DRG and TG was not affected by the loss of Sox11, suggesting that Sox11 may mainly exert its function at later stages of sensory neuron development. Developmentally regulated apoptosis takes places within sensory ganglia to selectively eliminate some neurons generated during early embryonic stages (Vogelbaum et al., 1998). NGF, BDNF, NT3, and their receptors are necessary for the survival of specific populations of embryonic sensory neurons (Klein et al., 1993; Crowley et al., 1994; Farinas et al., 1994; Jones et al., 1994; Klein et al., 1994; Smeyne et al., 1994; Tessarollo et al., 1994; Liebl et al., 1997). Increased cell death in DRG of Ngf−/− and TrkA−/− mice starts at E13.5 (White et al., 1996) but cell death is significant in DRG of Nt3−/− and TrkC−/− mice as early as E11.5. It is consistent with the notion that early born sensory neurons express TrkC and require NT3 for survival (Farinas et al., 1996, 1998). Significant neuronal loss is also detectable in BDNF−/−, NT3−/−, and TrkC−/− DRG as early as E12.5 (Liebl et al., 1997). Our data demonstrated that, at E12.5, increased cell death is already apparent in Sox11−/− sensory ganglia. However, at this time the expression of TrkA, TrkB, TrkC, and Ret is not altered in Sox11−/− TG compared with controls (Fig. 5), suggesting that the effect of Sox11 on the survival of these neurons may not simply be through regulating the expression of neurotrophin receptors. At later time points such as around birth, all subtypes of sensory neurons were lost at a similar proportion in Sox11−/− sensory ganglia, supporting a broad role of Sox11 in regulating the survival of sensory neurons.

The proapoptotic gene Bax is required for naturally occurring cell death in embryonic sensory ganglia and the loss of Bax can rescue sensory neuron death in TrkA−/−, Ngf−/−, and Nt3−/− mice (Patel et al., 2000, 2003; White et al., 1998). The loss of sensory neurons in Creb−/−, Brn3a−/−, and Klf7−/− embryos can also be rescued by the loss of Bax (Lonze et al., 2002; Ma et al., 2003; Lei et al., 2005). Our current study demonstrated that the increased cell death in Sox11−/− DRG and TG was dependent on Bax, suggesting a common link between these transcription factors in mediating neurotrophin-dependent neuronal survival. It is interesting to note that the expression of various apoptosis-promoting or apoptosis-inhibiting genes was affected when the level of Sox11 was knocked down by short interference RNA (siRNA) in vitro (Jankowski et al., 2006). This suggests possible regulatory functions of Sox11 on apoptosis-associated genes.

Although neuronal numbers were not affected in E12.5 Sox11−/− sensory ganglia, there were significant axonal growth defects in Sox11−/− embryos as revealed by whole-mount neurofilament (2H3) staining. To examine directly whether this defect is a result of increased cell death, Bax−/− mice were crossed into the Sox11 null background to block apoptosis. Our data demonstrated that Sox11−/−;Bax−/− embryos still had axonal growth defects compared with Bax−/− and control embryos, suggesting that Sox11 is required for in vivo axonal growth at least partially through a cell death-independent manner.

It was previously reported that the treatment of primary DRG neurons in culture with Sox11 siRNA causes a significant decrease in neurite outgrowth and survival (Jankowski et al., 2006). Our culture of explants and dissociated neurons from Sox11−/− and control DRG demonstrated that Sox11 is indeed required for the survival and axonal growth of sensory neurons in a cell-autonomous manner. This is consistent with the specific expression of Sox11 in postmitotic neurons in the sensory ganglia in vivo.

The molecular mechanisms of how Sox11 regulates sensory neuron survival and axon growth remain to be elucidated. Although Sox11 is not required for the expression of Trk receptors and Ret at E12.5, it remains plausible that Sox11 may be required for maintaining the expression of these receptors during later developmental stages. It is also possible that Sox11 may regulate the expression of other components of the neurotrophin signaling pathways, the dysfunction of which can cause sensory neuron defects as well. In the future, gene expression profiling will be needed to identify potential targets of Sox11 that are important for sensory neuron survival and/or axonal growth.

EXPERIMENTAL PROCEDURES

Breeding and Genotyping of Sox11−/− and Sox11−/−;Bax−/− Mice

Mice were housed and maintained according to the IACUC guidelines of the University of Texas Southwestern Medical Center, the Medical College of Wisconsin, and the University of New England. Sox11+/lacZ knock-in mice, which were referred to as Sox11+/− mice, were mated to generate Sox11−/− embryos (Sock et al., 2004). Sox11+/− mice were crossed with Bax+/− mice to generate Sox11+/−;Bax+/− mice, and then Sox11+/−;Bax+/− mice were mated to generate Sox11−/−;Bax−/− embryos. Timed pregnant female mice were killed to collect embryos at different developmental stages. Sox11+/− and Sox11+/+ animals were phenotypically indistinguishable from each other and thus were pooled as controls unless indicated otherwise. Genotyping of the Sox11 and Bax mutant mice was accomplished by PCR as previously described (Ma et al., 2003; Sock et al., 2004).

Whole-Mount In Situ Hybridization

Mouse embryos were processed for whole-mount in situ hybridization as previously described (Wilkinson, 1992). Following linearization of DNA templates, digoxigenin-labeled probes were generated using appropriate RNA polymerases (Promega) and color reactions were developed with BM Purple (Roche).

Section In Situ Hybridization and Immunohistochemistry

For in situ hybridization on sections, mouse embryos were fixed with modified Carnoy's solution, processed for paraffin sectioning, and in situ hybridization as described (Etchevers et al., 2001). For post-in situ immunohistochemistry, slides were incubated in Tuj-1 (Covance/Babco, Lee et al., 1990) overnight at 4°C followed by three washes in phosphate buffer saline with 0.1% Tween 20. Fluorochrome-conjugated secondary antibodies (Jackson ImmunoResearch) were applied for 2 hr at room temperature in dark, and then slides were rinsed in phosphate buffer saline with 0.1% Tween 20 before coverslipping. Brightfield in situ hybridization and fluorescence immunostaining signals on sections were photographed using a Zeiss Z1 microscope equipped with a color HRc AxioCam camera.

Nissl Staining and Cell Counts

For embryonic day 18.5 (E18.5) and postnatal day (P) 0 cell counts, tissues were fixed in 4% paraformaldehyde (PFA) overnight, protected in 30% sucrose and then embedded in Optimal Cutting Temperature (OCT) compound. For E18.5 DRG, trunk transverse sections were cut continuously at 14 μm using a cryostat (Leica), collected directly on Superfrost (+) slides (Fisher) and kept in a −80°C freezer before use. For P0 TG, head sagittal sections were cut serially at 14 μm using a cryostat, collected onto Superfrost plus slides, and kept in a −80°C freezer before use. For E12.5 DRG and TG counts, tissues were fixed in 4% paraformaldehyde, processed, and embedded in paraffin, and 5-μm sagittal sections were made. Sections were stained with 0.5% cresyl violet and cell counting was performed as described previously (Lei et al., 2005). Briefly, pictures were taken every 140 μm for P0 TG, every 14 μm for E18.5 DRG, every 150 μm for E12.5 TG, and every 30 μm for E12.5 cervical DRG, respectively, and cells with visible nucleoli were counted as neurons. For P0 TG, the total number of neurons was calculated by multiplying the sum by 10. For E18.5 DRG, the sum was reported as the total number of neurons. For E12.5 TG and DRG, the total number of neurons was calculated by multiplying the sum by 30 or 6, respectively.

Immunohistochemistry

Whole-mount neurofilament immunostaining was performed using the 2H3 hybridoma supernatant (Developmental Studies Hybridoma Bank, 1:70) as previously described (Lonze et al., 2002). For DRG explant immunostaining, the 2H3 hybridoma supernatant was diluted 1:10 in blocking solution (5% normal donkey serum in phosphate buffered saline + 0.3% Triton X-100) and incubated overnight at 4°C. Frozen sections were processed for immunofluorescent staining as previously described (Lei et al., 2006). Primary antibodies included TrkA (rabbit, L.F. Reichardt, 1:500), TrkB (chick, L.F. Reichardt, 1:500), TrkC (goat, L.F. Reichardt, 1:500), Ret (rabbit, IBL, 1:50), and Tuj1 (mouse, Covance, 1:500). Immunofluorescent staining on paraffin sections was carried out by incubating with primary antibodies in 2% normal goat serum overnight at 4°C after antigen retrieval. Primary antibodies included β-gal (rabbit, ICN, 1:500), β-gal (goat, Bio Trend, Cologne, Germany, 1:500), cleaved caspase 3 (rabbit, Cell Signaling, 1:200), Sox11 (guinea pig, 1: 200; Hoser et al., 2008), Tuj1 (mouse, Covance, 1:500), and BrdU (mouse, Dako, 1:100). Immunofluorescence was visualized after 1 hour of incubation with appropriate Cy2-, Cy3-, or Cy5-conjugated secondary antibodies (Jackson Labs, 1:200) at room temperature. DAPI (4′,6-diamidine-2-phenylidole-dihydrochloride; Fluka, 1 μg/ml) was added together with the secondary antibodies to stain nuclei. The immunostaining was visualized using a Nikon Eclipse TE2000-U fluorescence microscope or a Zeiss 510 confocal microscope. Digital images were captured using a Photometrics CoolSnap CCD camera (Roper Scientific) and analyzed using the Metamorph software (Universal Imaging).

Cell Proliferation

Pregnant female mice were injected intraperitoneally (IP) with 100 μg/g of body weight 2-bromo-5′-deoxyuridine (BrdU, Sigma) 2 hr before being killed. Embryos were collected and 14-μm transverse cryostat sections were made. Anti-BrdU staining was performed using a standard procedure with anti-BrdU-Alexa488 (mouse, Molecular Probes, 1:500; Lei et al., 2005). The number of BrdU-positive cells in TG (every tenth section) and C1/C2 DRG (every section) were counted.

Apoptosis

Paraffin sections were deparaffinized and rehydrated. After antigen retrieval, sections were incubated with an antibody against cleaved caspase 3 (rabbit, Cell Signaling Technology, 1:200). The visualization of primary antibodies was performed with a horseradish peroxidase system (Vectastain ABC kit, Vector). Digital images were captured with a CCD camera under a light microscope (Olympus). Cleaved caspase 3-positive cells were counted in at least 10 sections for each TG or DRG.

Dissociated DRG Neuron Culture

Dissociated DRG neuron culture was carried out as described previously (Lei et al., 2005). Briefly, DRG were isolated from E13.5 embryos and dissociated by trypsin. Neurons were cultured in Neurobasal media (Gibco) containing NGF (NGF-7S, Sigma, 10 ng/ml) on poly-D-ornithine and laminin (Sigma) -coated chamber slides (Nalge Nunc). Cells were fixed with 4% PFA at 6 hr, 24 hr, 48 hr, and 72 hr, after plating and staining with 2H3 antibody (1:50). The percentage of neuronal survival was calculated by dividing the neuron number at 24 hr, 48 hr, and 72 hr, respectively, by the numbers at 6 hr. The values for the mutants were normalized against the values for the wild-type neurons, which were set at 100%.

DRG Explants Culture

DRG explants culture was performed as described previously (Lonze et al., 2002). Briefly, DRG explants were removed from E13.5 embryos and grown on chamber slides coated with poly-D-ornithine and laminin in Neurobasal medium containing NGF (10 ng/ml) with or without the caspase inhibitor BOC-Asp(OMe)-FMK (BAF, Enzyme Systems, 50 μM). After 30 hr, ganglia were fixed and immunostained with 2H3 antibody (1:50). Axon lengths were quantified as described previously (Fitzli et al., 2000). For each ganglion, the 20 longest axons were measured using Metamorph software (Universal Imaging). Only ganglia with visible axonal processes after 1 day in vitro (DIV) were scored.

Statistics

All cell counts and measurements were performed in a genotype-blind manner. Statistical analysis was conducted using Student's t-test with P values ≤ 0.05 as being statistically significant. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001. Error bar, SEM.

Acknowledgements

We thank Dr. L.F. Reichardt (UCSF) for generous gifts of Trk antibodies. V.L. would like to dedicate this manuscript to the memory of J.L., whom we lost to a tragic car accident on February 6, 2010.

Ancillary