Adult Neurogenesis in Peripheral Nervous System

Although postnatal neurogenesis has been discovered in some regions of the peripheral nervous system (PNS), only indirect evidences indicated that some progenitors in the adult sciatic nerve and dorsal root ganglion (DRG) serve as a source of newly born sensory neurons. Here, we report the discovery of neurons and neuronal stem cells in the adult rat sciatic nerve. Lineage tracing detected a population of sciatic nerve neurons as progeny of adult neuronal stem cells. With the further finding of labeled DRG neurons in adult transgenic rats with local sciatic nerve staining, we propose a model of adult neurogenesis in the sciatic nerve in which neuronal stem cells in sciatic nerve mature as sensory neurons in adults along the sciatic nerve to DRG. This hypothesis provides a new way to understand sensory formation in adults. Those neuronal stem cells in the sciatic nerve may help to therapy of nerve trauma and disease in the future.


Introduction 20
Neurogenesis occur in specific regions of the mammalian brain throughout life with 21 critical roles in brain plasticity such as learning, memory and mood regulation (Berg et 22 al., 2019). Postnatal peripheral nervous system neurogenesis has been discovered in 23 mammalian parasympathetic ganglia of the head (Dyachuk et al., 2014;Espinosa-Medina 24 et al., 2014) and the gut (Uesaka et al., 2015). This finding suggests that neurogenesis 25 might also occur in the adult PNS, such as in the sciatic nerve and dorsal root ganglion 26 (DRG). 27 However, confirming the in vivo existence of neuronal stem cells in these regions is 28 challenging. In contrast to the extensive research of adult neurogenesis in the mammalian 29 brain, we know very little about the adult neuronal progenitors in the PNS. Recent 30 studies have revealed stem-like populations in DRG that displayed sphere-forming 31 potential and multipotency in vitro, yet the in vivo presence of neuronal stem cells in the 32 DRG has not been documented in any ultrastructural studies in adult mammals (Li et al., 33 2007;Nagoshi et al., 2008;Vidal et al., 2015). The sensory DRG neurons are derived 34 from the thoracolumbar region of the trunk neural crest. The cervical region of those 35 neural crest differentiate into large diameter neurons at first (Lawson and Biscoe, 1979). 36 Late emigrating trunk neural crest give rise to boundary cap neural crest stem cells, a 37 source of multipotent sensory specified stem cells (Radomska and Topilko, 2017). As a 38 transient population, the embryonic neural crest quickly transfer from multipotent to 39 restricted progenitors with limited capacity to self-renew before birth (Bronner and 40 Simoes-Costa, 2016). In mammalian fetal and adult peripheral nerves and skin, neural 41 2 crest derivatives give rise to multiple derivatives in vitro (Gresset et al., 2015;Morrison 42 et al., 1999;Wong et al., 2006). This finding suggests that a subset of the neural crest 43 population in the sciatic nerve and skin maintain multipotency after embryonic 44 development. But the identity of precursors to adult sciatic nerve and DRG neurons and 45 how they maintain their multipotency during development from embryonic to adult 46 mammals are unknown. 47 One major obstacle to studying the adult neurogenesis in PNS is a lack of methods 48 to the identification of neuronal stem cells in vivo. Many studies focus on cell isolation 49 or in vitro culture of adult sciatic nerve and DRG because of the lack of a more specific 50 in vivo tool (Baggiolini et al., 2015;Morrison et al., 1999). We established a sciatic nerve 51 crush model in adult rats. By whole-mount staining and optical imaging of the crushed 52 sciatic nerve tissue for stathmin 2 (Stmn2, or Scg10) (Shin et al., 2014), we observed 53 neurons and neuronal stem cells in adult rat sciatic nerve. As an intermediate filament 54 protein in neuroepithelial precursor cells, Nestin is considered a hallmark of neural 55 stem/progenitor cells (Dubois et al., 2006;Hockfield and McKay, 1985;Lendahl et al., 56 1990;Wiese et al., 2004). We characterized neuronal stem cells labeled by the 57 Nestin-CreER T2 rat line and Nestin-Cre rat line in the adult sciatic nerve but not in DRG 58 with clonal lineage-tracing. In adult rats stained for the neuron marker Stmn2 and 59 Peripherin (Escurat et al., 1990), the lineage-tracing neuronal stem cell and its progeny 60 were temporality and spatiality distributed along the sciatic nerve from the dermal nerve 61 ending to the DRG, suggesting that adult neurogenesis in the DRG does not occur in situ 62 but, rather, new cells migrate along the sciatic nerve. During embryonic development, 63 the neural crest migrate from the neural tube to the DRG as sensory neurons and to the 64 sciatic nerve and dermal nerve ending as multipotent cells (Baggiolini et al., 2015;65 Gresset et al., 2015;Morrison et al., 1999). Our study provides a new perspective that 66 those multipotent cells will mature in the sciatic nerve and migrate from sciatic nerve to 67 DRG as sensory neurons in adult. 68

69
The Adult Sciatic Nerve Contains Neuronal Cell Bodies

70
To enable us to image deep within PNS structures, we used a clearing reagent called 71 ScaleS that renders the rat DRG and sciatic nerve transparent, but completely preserves 72 fluorescent signals from labeled cells (Hama et al., 2015). Optical clearing of tissue 73 allowed us to identify neuronal cell bodies in the sciatic nerve. Three days after creating 74 a 1mm sciatic nerve lesion in twenty adult rats for 30s crush by forceps, we used an 75 Stmn2 antibody to identify regeneration projections of the damaged DRG neurons. 76 Surprisingly, we observed neuron-like cells at the distal end of six rats sciatic nerve, 77 though 14 other rats did not contain these cells (Fig. 1a). As mature Schwann cells 78 generate Schwann spheres and pigment cells in crushed sciatic nerve in vitro and in vivo, 79 the existence of the neuron-like cells may have been induced by injury (Takagi et al., 80 2011). To assess whether the neuron-like cells exist in undamaged nerves, we used the 81 same optical clearing method on intact sciatic nerves of adult rats. We found that the 82 neuron-like cells were also present in intact sciatic nerves in five of 25 control rats (Fig. 83 3 1b).

84
The unpredictable existence and location of those neuron-like cells in the sciatic 85 nerve make those cells difficult to trace and identify. Delivery of adeno-associated virus 86 (AAV) provides a noninvasive method for broad gene delivery to the nervous 87 system (Foust et al., 2009). To label the neuron-like cells in vivo, we infected the sciatic 88 nerves with engineered AAV2/9 and the hSYN and hEF1a promoter, to elicit stable 89 expression of green fluorescent protein (GFP) in the cells. We detected the neuron 90 projections in these infected neuron-like cells in vivo (Fig. 1c). Forty-six of 140 rats had 91 these neuron-like cells in their sciatic nerves. Although only some of the neuron-like 92 cells were stained, the expression of the neuron-specific marker NeuN(Mullen et al., 93 1992) indicated that part of those cells were indeed neurons (Fig. 1c). Together, this 94 evidence indicates that there are neurons in adult rat sciatic nerve. in vitro method to culture the sciatic nerve in a defined serum-free medium. We infected 111 the adult rat sciatic nerve with AAV2/9 containing the hSYN and hEF1a promoter in 112 vitro so that those cells stably expressed GFP, and observed labeled cells in sciatic nerves 113 four to seven days later (Fig. 2a). For two weeks culture, sciatic nerves of 40 rats 114 contained labeled cells while sciatic nerves of 170 rats were without. We then identified 115 those cells by staining for peripherin, a marker of peripheral neurons (Fig. 2b). 116 Next, to prospectively identify the labeled cells enables us to directly examine their 117 properties at the molecular level. We conducted single-cell sequencing after 2 weeks of 118 in vitro culturing (Fig.2c). 114 cells were dissected from GFP-positive sciatic nerves in 119 vitro and in vivo (9 cells were from in vivo nerves). 10 DRG neurons were used as a 120 positive control. Unsupervised clustering analysis assessed the separation of DRG 121 neurons, Stmn2 + cells, and Stmn2 − cells (Blondel et al., 2008). Cells were separated into 122 four clusters by further unsupervised clustering analysis. Cluster 1 and 2 cells showed 123 high expression of transcripts for Egr2 (Krox20), nestin, and Mpz (protein 0), which are 124 markers of migrating neural crest cells during the early fetal period(Dupin and Sommer, 125 2012). Cluster 3 and 4 cells showed high expression of Tac1 and Tubb3, which encode 126 markers of mature neurons (Hokfelt et al., 2001;Jiang and Oblinger, 1992) (Fig.2d). As 127 many cell-cycle genes were commonly expressed to varying degrees, this finding suggest 128 that the labeled cells include neurons (some cells in cluster 4) and quiescent neuronal 129 stem cells(some cells in cluster 1) that expressed low levels of cell-cycle genes, active 130 neuronal stem cells(some cells in cluster 2,3) that expressed high levels of cell-cycle 131 genes in the adult sciatic nerve. 132 We used a set of immunobiological markers and morphological criteria to identify 133 and quantify the different cell types labeled by AAV2/9 in the sciatic nerve in vivo and in 134 vitro. In both contexts, some of the cells exhibited radial glia-like morphology and 135 expressed Nestin . 136 Together, these findings suggest that neuronal stem cells which we called 137 multipotential sciatic nerve neural crest stem cells(snNCSCs) exist in the adult sciatic 138 nerve in vitro and in vivo. To examine the cell fate of Nestin-positive cells in the adult rat sciatic nerve, we 161 constructed a Nestin-CreER T2 rat for sequential observation (Dubois et al., 2006). We 162 analyzed recombination in situ using reporter rats carrying an Tdtomato transgene whose 163 expression was dependent on Cre-mediated recombination. 164 At different time points after tamoxifen and EdU intraperitoneal injection, We used 165 stmn2 to identify and quantify cells labeled with Tdtomato in the sciatic nerves (Fig3a). 166 The first two or three weeks, Tdtomato + cells were not co-labeled with stmn2 but in the 167 segments consisted of the stmn2 + cells. Most of those cells are set aside in quiescence 168 without co-labeled with EdU(Fig3b). At four or more weeks, part of Tdtomato + cells 169 were co-labeled with stmn2 and EdU. Some of the Tdtomato + stmn2cells and 170 Tdtomato -stmn2 + cells both co-labeled with EdU indicated those cells not only able to 171 be differentiation but also self-renew (Fig3c). It consist with our single cell sequencing 172 result and indicated most of those cells are quiescent neuronal stem cells and progenitors. 173 For a long term tamoxifen injection, all the stmn2 + cells we detected are co-labeled with 174 Tdtomato in 8 monthes old rats. This indicated those stmn2 + cells are derived from the 175 Tdtomato + cells.

176
All those together, we concluded that the snNCSCs will set aside in quiescence and 177 gradually differentiate into progenitors and neurons in the adult sciatic nerves in vivo. rats. We dissected the L4 or L5 DRG and sciatic nerve of those rats. Surprisingly, NeuN 195 staining revealed mCherry + neurons in DRG of 3 rats (Fig. 4a). Next, we sacrificed rats 7 196 weeks after tail vein injection of AAV-PHP.S virus. Nineteen of 20 successfully injected 197 rats contained mCherry + neurons in their DRG (Fig. 4b). 198 To specifically assess the cell fate of the marked DRG cells, we injected the sciatic 199 nerve of Nestin-Cre rats with hSYN-GFP DIO (FLEx switch under the induction of Cre) 200 AAV-PHP.S virus, and sacrificed ten rats 3 weeks later. In 2 rats, GFP + cells were 201 observed in the DRG and sciatic nerve ipsilateral to injection with virus, whereas no 202 GFP-positive cells were observed on the contralateral side. We further characterized 203 traced cells with markers for different categories of mature DRG neurons, including 204 peptidergic sensory neurons, such as Plexin C1 and Calca-positive neurons (Usoskin et al.,205 2015) ( Fig. 4c). In addition, co-labeled with Ssea1(Sieber-Blum, 1989) but not Nestin or 206 Egr2 indicated the limited differentiation potential of mCherry + cells and newly born 207 neurons in the DRG of mCherry DIO AAV-PHP.S-injected Nestin-Cre rats (Fig. 4d). 208 Together, these findings indicate the possibility that neurons from the sciatic nerve 209 migrate into the DRG as sensory neurons. The spatial distribution of the labeled neurons 210 indicated a wave of sensory neuron migration from the sciatic nerve to the DRG in adult 211 rats. and Simoes-Costa, 2016; Parfejevs et al., 2018). Our discovery of neurons and neuronal 232 stem cells in the adult mammalian sciatic nerve indicates that the nerve contains 233 multipotent stem cells that differentiate into neurons. Our findings moreover indicate that 234 sciatic nerve neurons will migrate into the DRG in adult.

235
As the existence and location of those cells are unpredictable for now in the adult 236 sciatic nerve, we hypothesize that neural crest cells at the neural tube migrate to the 237 sciatic nerve and dermis nerve ending in the skin, where they give rise to different neural 238 crest derivatives and self-renewing cells before birth (Bronner and Simoes-Costa, 2016;239 Gresset et al., 2015;Morrison et al., 1999) (Fig. 5a); those self-renewing cells, snNCSCs, 240 mature in the sciatic nerve and eventually migrate to the DRG in adult (Fig. 5b). In the crush model (Fig. 1a), the sciatic nerve presents a snapshot of adult 249 neurogenesis process outlined in our hypothesis. Along the sciatic nerve from a location 250 distal to the DRG, snNCSCs arranged in a segmental chain of cell groups develop larger 251 cell bodies and undergo a reduction in number. Those segmental form and migration 252 pattern are with same character as the embryonic neural crest (Szabo and Mayor, 2018).

253
In Nestin-CreER T2 rats, those segmental chain comprise neuronal stem cells, 254 progenitors and neurons. Most of those stem cells and progenitors are in quiescence and 255 gradually development along the sciatic nerve. 256 11 In Nestin-Cre transgenic rats labeled with the virus via local sciatic nerve injection, 257 we found labeled neurons in the DRG. Without finding of nestin + progenitor cells 258 co-labeled with traced marker in the DRG of those virus injected rats further confirmed 259 our hypothesis that sciatic nerve neurons migrate to the DRG in adult.

260
Our findings raise two major questions that require further study. First, we must 261 determine the source of snNCSCs. NCSCs in fetal sciatic nerve were shown to have 262 self-renewing ability in vivo and to maintain multipotency in vitro (Morrison et al., 1999). programmed organizational process that occurs during specific periods of adulthood.

279
Although it must be regarded cautiously until verified, we hypothesize that neuronal 280 stem cells in the sciatic nerve are derived from self-renewing cells, which themselves are 281 derived from the neural crest and persist in dermis nerve endings or a branch of the 282 sciatic nerve. The maturation process of those cells is triggered by changes in the 283 environment. These changes are similar to those brought about by an immunogen during 284 the secondary immune response. Once an environmental change (such as the arrival of an 285 immunogen) stimulates the dermis nerve ending or the branch of the sciatic nerve, 286 snNCSCs are activated and stored in the trunk of the sciatic nerve. As mature neurons 287 and snNCSCs in quiescent condition are both in the adult sciatic nerve as EdU staining 288 experiment and single-cell sequencing indicated, the same environmental change will 289 initiate a rapid sensory response. This hypothesis may help explain mammalian 290 responses to environmental change and even drive a new understanding of the evolution 291 of the mammalian sensory system. 292 Adult neural stem cells are attracting increased interest as potential candidates for 293 cell transplantation therapy for nerve trauma and disease because they are present in 294 tissue that can be harvested from the patient (Parfejevs et al., 2018;Radomska and 295 Topilko, 2017). Moreover, skin stem cells contribute to skin regeneration and wound 296 repair through cellular programs that can be hijacked by cancer cells (Ge and Fuchs, 2018;297 Mantyh, 2006;Nakada et al., 2011) Our snNCSCs migration model may therefore 298 provide clues to cancer cell migration along the sciatic nerve, expanding knowledge 299 about their role in hijacking the hematopoietic system via blood vessels and lymphatic 300 vessels (Crane et al., 2017). In the future, we hope this work will also facilitate 301 transplantation of adult neuronal stem cells in the sciatic nerve using a method that 302 simulates typical adult sensory reconstruction processes so that we can eventually realize 303 functional sensory reconstruction. Furthermore, our observation of newly born sensory 304 neurons may help elucidate the mechanisms of pain, touch, and other senses, and may 305 one day enable adult sensory reconstruction and help overcome barriers to limb 306 reconstruction. 307

308
Animals 309 Three rat lines were used for this study: Sprague-Dawley (SD), a transgenic Nestin 310 reporter SD rat line that expresses Cre from the endogenous H11 locus, a transgenic 311 Nestin reporter SD rat line that expresses CreER T2 from the endogenous H11 locus and 312 pCAG-loxP-3XSTOP-loxP-Tdtomato-WPRE-bGHpA from the endogenous ROSA26 locus.

313
Male experimental and control rats were littermates housed together before the 314 experiment. We produced Nestin-Cre and Nestin-CreER T2 knock-in rats via the 315 CRISPR/Cas9 system. First, a single guide RNA (sgRNA) targeting the H11 locus, the SD 316 Nestin promoter-Cre-PA-Nestin Enhancer fragment, was inserted into the H11 locus of 317 rats using CRISPR/Cas9 technology. The rat H11 locus (which is positioned between the 318 Eif4enif1 and Drg1 genes) is ubiquitous, allowing the use of an exogenous promoter to 319 drive higher expression when inserted at the locus. Second, Cas9, sgRNA, and the donor 320 vector were co-injected into zygotes. We transferred the injected zygotes into the 321 oviduct of pseudopregnant SD females. F0 rats were birthed 21-23 days after 322 transplantation, and were identified by PCR and sequencing of tail DNA. Positive F0 rats 323 were genotyped. Lastly, we crossbred positive F0 rats with SD rats to generate 324 heterozygous rats. All animal procedures were performed in accordance with 325 Institutional Animal Care guideline of Nantong University, and were ethically approved 326 by the Administration Committee of Experimental Animals, Jiangsu Province, China. 327

371
We cultured sciatic nerves from adult 6-to 8-week-old SD rats in 10 cm plates (Corning) 372 coated with poly-D-lysine (Sigma) and laminin (Sigma). The sciatic nerve was cut from 373 below the DRG (omitting all DRG tissue) to the nerve ending. We carefully peeled away 374 the epineural sheath in cold PBS. The collected sciatic nerves were plated as a line at a 375 density of approximately eight sciatic nerves per dish and kept for 20 min at 37°C (make 376 sure it fixed on the plates), and a Neurobasal medium (Invitrogen) supplemented with 2% 377 (vol/vol) B27 (Invitrogen) and 25 ng/mL nerve growth factor (Sigma) was added. 378 Cultured sciatic nerves were maintained for 1 day prior to injection. The culture medium 379 was discarded before the injection. The virus (20 μl/8 nerves, with virus titers of 380 approximately 2.50E+13) was dropped slowly and uniformly onto the sciatic nerve and 381 then incubated for 2 h before adding the culture medium(carefully ensuring that the 382 14 sciatic nerve did not dry out). Then, the sciatic nerve was cultured for 1-2 weeks before 383 observation to allow for GFP expression from the sciatic nerve, where the cell bodies 384 were placed. For more than one month cultures, the medium was changed every 5 days. 385 We examined neurons using a fluorescence microscope. 386

387
For the sciatic nerve lesion experiment, the adult rats (8-10 weeks old) were 388 anesthetized by intraperitoneal injection with 85 mg trichloroacetaldehyde 389 monohydrate, 42 mg magnesium sulfate, and 17 mg sodium pentobarbital. We exposed 390 sciatic nerve at the sciatic notch by making a small incision. The nerve was then crushed 391 at the same position for 30s under the same pressure by a ultra-fine hemostatic forceps, 392 and the crush site was marked with a size 10-0 nylon epineural suture. For the control 393 rats (8-10 weeks old), the sciatic nerve was exposed but left uninjured. After surgery, 394 the wound was closed, and the rats were allowed to recover for two hour.

395
For the sciatic nerve AAV injection, we anesthetized the adult rats (6-8 weeks old, 396 normal rats and 6 weeks old, Nestin-Cre rats) with an intraperitoneal injection of 397 complex narcotics (85 mg trichloroacetaldehyde monohydrate, 42 mg magnesium 398 sulfate, and 17 mg sodium pentobarbital) and carefully opened the skin and muscle and 399 to expose the sciatic nerve. Then 10 cm capillary glass tubes (Sutter Instrument, Novato, 400 CA) were pulled using a micropipette puller (model 720, David KOPF Instruments, 401 Tujunga, CA). The tips of the pulled tubes were pinched with forceps to create pipettes 402 with an external diameter of approximately 10 μm. A 2.5 μl volume of AAV2/9 (virus 403 titers were approximately 2.50E+13) was gradually injected into the sciatic nerve with 404 one pump of a microsyringe pump at a rate of 1 μl/min (Stoelting Instruments). The 405 needle tip was inserted into the epineural sheath, and the drops caused it to plump up.

406
After three injections at three different sites along the sciatic nerve, the wound was 407 closed, and the rats were allowed to recover for two hour. 408 For the tail AAV injection, we placed the 3-week-old rats in a restraint device. The tail 409 was stabilized between the investigator's thumb and forefinger. To soften the skin, the 410 tail was prepared in 40°C water for 5 min and then sterilized by 70% ethanol. We used the Vazyme method, followed by cDNA amplification as described below. 436 Whole transcriptome amplification was performed using the Discover-scTM WTA Kit V2 437 (Vazyme, N711). First, 124 active cells were isolated and transferred into a lysis buffer. 438 Then, mRNA was copied into first-strand cDNA using Discover-sc Reverse Transcriptase 439 and oligo dT primer. At the same time, we added a special adapter sequence to the 3' 440 end of the first-strand cDNA. Full-length cDNA enrichment was performed by PCR, and 441 the products were purified by VAHTSTM DNA Clean Beads (Vazyme, N411). Next, we 442 performed quality control using the WTA cDNA. The cDNA concentration was measured 443 using a Qubit DNA Assay Kit in a Qubit 3.0 Fluorometer (Life Technologies, CA, USA).

444
DNA fragment size was tested using an Agilent Bioanalyzer 2100 system (Agilent 445 Technologies, CA, USA). A total of 1 ng of qualified WTA cDNA product per sample was 446 used as input material for the library preparation.

447
We generated sequencing libraries using the TruePrep DNA Library Prep Kit V2 for 448 Illumina (Vazyme, TD503), following the manufacturer's recommendations. First, cDNA 449 was randomly fragmented by the Tn5 transposome at 55°C for 10 min at the same time 450 as a sequencing adapter was added to the 3' adenosine on the fragment. After 451 tagmentation, the stop buffer was added directly into the reaction to end tagmentation.

452
PCR was performed, and the products were purified with VAHTSTM DNA Clean Beads 453 (Vazyme, N411We conducted preliminary quantification of the library concentration 454 using a Qubit DNA Assay Kit in Qubit 3.0. Insert size was assessed using the Agilent 455 Bioanalyzer 2100 system, and if the insert size was consistent with expectations, it was 456 more accurately quantified using qPCR with the Step One Plus Real-Time PCR system 457 (ABI, USA).

458
We identified the neuron-like cells via Stmn2 expression. We identified 94 Stmn2 + cells 459 in vitro and six Stmn2 + cells in vivo using qPCR before sequencing. 14 Stmn2cells were 460 identified as negative control. 10 DRG neurons were identified as positive control. 461 Clustering of the index-coded samples was performed on a cBot Cluster Generation 462 System (Illumina) according to the manufacturer's instructions. After cluster generation, 463 the library preparations were sequenced on an Illumina Hiseq X Ten platform with a 150 464 bp paired-end module. 465

466
Samples were then normalized by down sampling to a minimum number of 124 467 transcripts per cell for the clustering analyses or a minimum of 100 transcripts per cell 468 for differential gene expression analyses. Cells with fewer transcripts were excluded 469 from the analyses. The modularity optimization technique SLM was used for 470 unsupervised cell clustering. We used t-SNE to place cells with similar local 471 neighborhoods in high-dimensional space together (McDavid et al., 2013). 472