Two factor-based reprogramming of rodent and human fibroblasts into Schwann cells

Abstract

Schwann cells (SCs) generate the myelin wrapping of peripheral nerve axons and are promising candidates for cell therapy. However, to date a renewable source of SCs is lacking. In this study, we show the conversion of skin fibroblasts into induced Schwann cells (iSCs) by driving the expression of two transcription factors, Sox10 and Egr2. iSCs resembled primary SCs in global gene expression profiling and PNS identity. In vitro, iSCs wrapped axons generating compact myelin sheaths with regular nodal structures. Conversely, iSCs from Twitcher mice showed a severe loss in their myelinogenic potential, demonstrating that iSCs can be an attractive system for in vitro modelling of PNS diseases. The same two factors were sufficient to convert human fibroblasts into iSCs as defined by distinctive molecular and functional traits. Generating iSCs through direct conversion of somatic cells offers opportunities for in vitro disease modelling and regenerative therapies.

ARTICLE

Received 29 Jun 2016 | Accepted 18 Nov 2016 | Published 7 Feb 2017

Pietro Giuseppe Mazzara1,2, Luca Massimino1, Marta Pellegatta3, Giulia Ronchi4, Alessandra Ricca5,Angelo Iannielli1, Serena Gea Giannelli1, Marco Cursi6, Cinzia Cancellieri1, Alessandro Sessa1, Ubaldo Del Carro6, Angelo Quattrini7, Stefano Geuna4, Angela Gritti5, Carla Taveggia3 & Vania Broccoli1,8

Schwann cells (SCs) generate the myelin wrapping of peripheral nerve axons and are promising candidates for cell therapy. However, to date a renewable source of SCs is lacking. In this study, we show the conversion of skin broblasts into induced Schwann cells (iSCs) by driving the expression of two transcription factors, Sox10 and Egr2. iSCs resembled primary SCs in global gene expression proling and PNS identity. In vitro, iSCs wrapped axons generating compact myelin sheaths with regular nodal structures. Conversely, iSCs from Twitcher mice showed a severe loss in their myelinogenic potential, demonstrating that iSCs can be an attractive system for in vitro modelling of PNS diseases. The same two factors were sufcient to convert human broblasts into iSCs as dened by distinctive molecular and functional traits. Generating iSCs through direct conversion of somatic cells offers opportunities for in vitro disease modelling and regenerative therapies.

DOI: 10.1038/ncomms14088 OPEN

Two factor-based reprogramming of rodent and human broblasts into Schwann cells

1 Division of Neuroscience, San Raffaele Scientic Institute, 20132 Milan, Italy. 2 University of Milano-Bicocca, 20126 Milan, Italy. 3 Division of Neuroscience, Axo-glia Interactions Unit-INSPE, San Raffaele Institute, 20132 Milan, Italy. 4 Department of Clinical and Biological Sciences, and Cavalieri Ottolenghi Neuroscience Institute, University of Turin, Ospedale San Luigi, 10043 Orbassano, Turin, Italy. 5 SR-Tiget, Division of Regenerative Medicine, Stem Cells and Gene Therapy, San Raffaele Scientic Institute, 20132 Milan, Italy. 6 Department of Neurology, Neurophysiology and Neurorehabilitation Unit-INSPE, San Raffaele Scientic Institute, 20132 Milan, Italy. 7 Department of Neurology, Neuropathology Unit-INSPE, San Raffaele Scientic Institute, Via Olgettina 48, 20132 Milan, Italy. 8 National Research Council (CNR), Institute of Neuroscience, 20129 Milan, Italy. Correspondence and requests for materials should be addressed to V.B. (email: mailto:[email protected]

Web End [email protected] ).

NATURE COMMUNICATIONS | 8:14088 | DOI: 10.1038/ncomms14088 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 1

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14088

Schwann cells (SCs) are neural crest-derived cells able to produce the myelin sheaths, wrapping neuronal axons in the peripheral nervous system (PNS). When peripheral

nerves are injured, SCs adaptively respond by supporting and stimulating tissue regeneration1. Nevertheless, after severe nerve injuries or in genetic and metabolic myelin disorders, the loss of myelin ensheathing axons cannot be replaced, leading to disabling sensory defects and motor dysfunctions2,3. A valuable therapeutic option for the treatment of peripheral nerve insults is represented by the transplantation of SCs, alone or in combination with the nerve guide4,5. However, this therapeutic approach is strongly limited by the current lack of a renewable source of SCs in humans.

Isolation of primary cultures of myelin-competent SCs works very poorly in mice and humans6 and methods currently available for differentiating SCs from pluripotent stem cells are time-consuming, technically complex and generate SC precursors with unproven myelination potential7. Generation of SCs has been recently obtained through differentiation of somatic progenitor cells8,9. Nonetheless, these approaches are limited by the need of isolating rare progenitor cells in tissues. Moreover, most of these methods are laborious and generate SCs with low myelination efciency that strongly limits the development of cell-based therapies and in vitro disease-modelling studies. To overcome these limitations, we speculated that a direct cell conversion approach to convert skin broblasts into SCs would offer a more straightforward and convenient procedure. Supra-physiological expression of dened sets of developmental neural transcription factors (TFs) is sufcient to impose a neural identity to somatic cells in a rapid and single-step procedure, generating induced neurons and glial cells with mature morphological and functional properties1014. In particular, TF-mediated reprogramming can be applied to generate induced oligodendrocyte precursor cells that express appropriate OPC markers, produce myelin sheaths in vitro and sustain myelin regeneration in mouse brains with genetic dysmyelination15,16. Importantly, induced oligodendrocyte precursor cells were shown to lack Myelin protein zero (MPZ) protein, a specic SC marker, and myelinated multiple axons conrming their central, and not peripheral, glial cell identity15,16. We, therefore, sought to determine whether SCs could be generated by direct lineage conversion from readily available somatic lineages such as broblasts. We identied two factors sufcient to convert rodent broblasts into SCs with molecular PNS identity and competent to generate compact and functional myelin sheets. The same factor combination could be used to promote conversion of human post-natal broblasts into SCs with comparable properties and functions.

ResultsTwo TF-based reprogramming of broblasts into SCs. Over the last decade, an intertwined regulatory network has been shown to have a critical role in promoting PNS myelination and its maintenance17,18. We selected Sox10, Pou3f1 (known also as Oct6), Egr2 (known also as Krox20) and Brn2 for their cardinal role during SC myelination as good candidates for cell lineage reprogramming19. To this end, the factors were individually cloned in doxycycline (dox)-inducible lentiviral vectors and E15.5 mouse embryonic broblasts were infected with one or more lentiviruses and cultured in a SC culture medium supplemented with Neuregulin-1 (NRG1) and forskolin (Fsk) (Fig. 1a)20. At rst, we realized that primary cultures of embryonic and adult skin broblasts often contain a fraction of CD271 cells with neural crest stem cell features and can give rise to SC precursors (Supplementary Fig. 1a,b)21. Thus, before

each reprogramming experiment, primary broblast cultures were selected against CD271 cells by ow-cytometry with a stringent gating selection (Supplementary Fig. 1c). To evaluate the SC lineage conversion, we monitored for the concomitant activation of S100 and O4, both highly expressed in SCs, likely in an immature state, while undetectable in broblasts. Fourteen days in vitro (DIV) after viral-mediated gene transfer the amount of double S100/O4 positive cells was scored with the different TF combinations (Supplementary Fig. 1d,e). Interestingly, we found that Sox10 combined with Egr2 expression produced the highest amount of S100/O4 cells(12.33.1% of total population, n 8 independent experiments.

Values are meanss.d.) (Fig. 1b,c and Supplementary Fig. 1d,e). The eventual addition of either Pou3f1 or Brn2 as a third factor did not further increase the yield of S100/O4 expressing cells (Supplementary Fig. 1e). Conversely, tTA infected or not infected (control) broblasts did not show detectable expression of either SC marker (Fig. 1b, inset). To better examine the reprogrammed cell fraction, we isolated the O4 cell population by ow-cytometry (Fig. 1d) and expanded the culture for the following analysis. Candidate gene expression analysis of the Sox10/Egr2 reprogrammed cell progeny revealed a strong activation of several SC markers, such as Erbb2, Erbb3, Cnx32, Pmp22 and Mpz (also known as P0) (Fig. 1e)22. Likewise, Mpz protein was readily detectable by immunostaining in the great majority of the S100 /O4 cellular fraction as it is normally found in somatic peripheral SCs, but not in

CNS oligodendrocytes (Fig. 1gi)22. Conversely, sorted cells did not express oligodendrocytes-specic transcription factors and markers such as Olig1 and Olig2, Mog, Myrf (Mrf), Ng2 and Nkx2.2 (Fig. 1f) conrming their PNS regional identity. In addition, endogenous Sox10 and Egr2 were expressed in transduced broblasts indicating that crucial components of the SC gene regulatory network were reactivated in the Sox10/Egr2, but not tTA, infected cells. (Fig. 1e). In accordance, known direct target genes of Sox10 and Egr2 during SC development and myelination such as Prx, Erbb3 and Mpz, were strongly activated as conrmed by quantitative reverse transcription PCR (qRTPCRs) (Supplementary Fig. 1f)23. Importantly, silencing of the two lentiviral transgenes by dox withdrawal for 2 weeks did not alter the expression of S100 and O4, indicating a stable expression of these SC markers independently by the reprogramming process (Supplementary Fig. 1g). Thus, given the strong activation of the crucial factors distinctive for the SC lineage and their subsequent maintenance even after transgene expression silencing, we named the Sox10/Egr2 reprogrammed broblasts as induced Schwann cells (iSCs). The two factors were also effective in reprogramming neonatal and adult murine skin broblasts generating S100 /O4 iSCs with an efciency of8.53.1% and 18.95.3% of total population, respectively (n 4) (Supplementary Fig. 2). Despite the driving force of the

reprogramming factors, the culture conditions proved to be critical for the subsequent survival of the reprogrammed cells. In particular the depletion of both NRG1 and Fsk, but not Fsk alone, caused a strong reduction in the number of S100/O4/Mpz iSCs due to a signicant increase in cell death (Supplementary Fig. 3).

Transcriptional proling of iSCs. Given that the establishment of pure and highly proliferating SC cultures is often obtained starting from neonatal rats, we wondered whether our method could be applied in other species rather than the mouse. Rat neonatal broblasts were converted into iSCs by the two-factor expression with similar efciency and dynamics compared to mouse cells (Supplementary Fig. 4a,b). Fluorescence-activated cell sorting-isolated rat O4 iSCs showed an elongated bipolar

2 NATURE COMMUNICATIONS | 8:14088 | DOI: 10.1038/ncomms14088 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14088 ARTICLE

a

Lv.tTA

Lv.TetO.TFs iSCs

Fibroblasts

21 Analysis

DIV 0

LV +DOX

1 3

+NRG1/Fsk

14 DOX

40 S100

b

c

SE

S100/O4

30

20

% Positive cells/total

***

tTA

10

0 tTA

Fib Mpz

S100/O4

SE

d

e f

256

tTA

Mbp Mag Erbb2 Erbb3 Cnx32 Pmp22 Sox10e

Egr2e 18S

80 110

82.5

55

27.5

0

256 SE

192

128

R3

R3

iSCs SN Fib

Olig1 Olig2

Mog Myrf

Ng2 Nkx2.2

18S

iSCs SN SpC

0.07%

6.1%

SSC

SSC

64

64

0

O4 FITC Log O4 FITC Log

0 00 101 102 103 104 105 00 101 102 103 104 105

g h i

S100 O4 MPZ

O4 MPZ

S100/O4 S100/MPZ

% Positive cells/

total

60

40

20

0

% Positive cells/

S100+

Figure 1 | Direct conversion of mouse skin broblasts into iSCs. (a) Schematic representation of the reprogramming strategy for generating iSCs using the dox-inducible lentiviral transgenes expressing tTA, combinations of transcription factors (TFs) together with the small molecules NRG1 and Fsk.(b) S100 (green) and O4 (red) immunouorescence staining of Sox10/Egr2 (SE) or tTA-infected (inset) MEFs. (c) Quantication of S100 and

S100 /O4 cells expressed as percentage over the total number of cells in experimental (SE) and control (tTA) conditions at 14 DIV (Means.d., n 8

independent experiments, 10 randomly selected 20 elds per sample were examined). (d) Fluorescence-activated cell sorting-based purication of the

O4 iSC population. (e) Expression of SC cardinal markers by RTPCR in broblasts (Fib), induced SCs and adult sciatic nerve (SN, positive control). Sox10e, endogenous Sox10, Egr2e, endogenous Egr2. (f) Expression of Oligodendrocyte-specic markers by RTPCR in broblasts (Fib), induced SCs, sciatic nerve (SN) and P20 spinal cord (SpC, positive control). (g) S100/O4 and S100/MPZ immunouorescence staining in iSCs. (h,i) Quantication of 14 DIV iSCs expressing S100, O4 and MPZ alone or in combination (Means.d., n 6 independent experiments, three coverslips/experiment/antigen,

10 randomly selected 20 elds per sample were examined). ***Po0,001; Students t-test. Scale bars, 50 mm.

shape and resulted immunodecorated for S100 and Mpz in most cases (Supplementary Fig. 4ce). Thus, in vitro cultures of puried rat O4 iSCs, their respective broblasts of origin and somatic SCs were proled for genome-wide expression by next-generation sequencing analysis of total RNA (RNA-seq).

Remarkably reprogrammed and somatic SCs, but not bro-blasts, display strong correlation as assessed by unsupervised hierarchical clustering (UHC), principal component analysis and correlation of gene expression patterns (Fig. 2a,b and Supplementary Fig. 5a). Differential gene expression analysis

showed about 4,000 transcripts differentially expressed between either iSCs or SCs versus broblasts. Of note, a large set of the differentially expressed genes (DEGs) was shared by the two SC populations (Fig. 2c). Moreover, gene ontology analysis revealed functional enrichment for multiple biological pathways invaluable for SC biology (Fig. 2d). Furthermore, by considering only those genes/microRNAs known to be pivotal for SC development, UHC clustered together iSCs and somatic SCs, thus conrming the similarities between the two cell types (Fig. 2e,f). Accordingly, qRTPCR analysis conrmed that both

NATURE COMMUNICATIONS | 8:14088 | DOI: 10.1038/ncomms14088 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 3

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14088

Color key and histogram

a

b

c

Count

500

0

2 1 0

Row Z-score

1 2

Fibro iSchwann Schwann

Schwann/Fibroblast iSchwann/Fibroblast

Schwann/Fibroblast iSchwann/Fibroblast

1 222

3

494

1 22

3

1 222222

3

d

KEGG_WNT_SIGNALING_PATHWAY KEGG_NOTCH_SIGNALING_PATHWAY KEGG_HEDGEHOG_SIGNALING_PATHWAY

KEGG_MTOR_SIGNALING_PATHWAY

KEGG_ERBB_SIGNALING_PATHWAY KEGG_CELL_CYCLE

KEGG_ADHERENS_JUNCTION

3,608 1,069

KEGG_DORSO_VENTRAL_AXIS_FORMATION

Fibroblast

Fibroblast

Fibroblast

Schwann

iSchwann

Schwann

Schwann

iSchwann

iSchwann

0 5 10 15 20 25 30 35

Percentage of genes

e f

g

SC Fib

SC

SC-specific miRNAs

Fib-specific genes

log10 p Schwann/Fibroblast

log10 p Schwann/Fibroblast

8.0

6.0

4.0

2.0

log10 p iSchwann/Fibroblast

log10 p iSchwann/Fibroblast

10.0

6.0

4.0

2.0

0

4% 13%

Fib

8.0

84%

6%

8%

86%

15% 4%

SC-specific genes

i

0

20.0 10.0 0 10.0 20 20.0 10.0 0 10.0 20

Log2 Schwann/Fibroblast

20.0 10.0 0 10.0 20

Log2 Schwann/Fibroblast

Log2 iSchwann/Fibroblast

20.0 10.0 0 10.0 20

Log2 iSchwann/Fibroblast

h

8.0

6.0

4.0

2.0

0

10.0

6.0

4.0

2.0

0

NRG1/Fsk

29%

NRG1/Fsk

41%

8.0

12%

14%

Relative

Row min Row max

Figure 2 | Comparative transcriptomic analysis of iSCs, primary SCs and broblasts of origin. (a) Heat-map of the unsupervised hierarchical clustering (UHC) of 2,500 random genes for broblasts and somatic and induced Schwann cells. The correlation between samples is shown as a dendrogram on top of the picture. (b) Principal component analysis showing the rst three components, explaining about 75% of the total variance. (c) Venn diagram showing the overlap between the differentially expressed genes (DEGs) comparing Schwann and iSchwann cells versus broblasts. (d) Functional enrichment analysis of the DEGs. (e,f) UHC of somatic SC-specic genes/microRNAs in the three cell populations. (g) Volcano plots showing the distribution of SC- and broblast-specic genes in the two conditions. (Chi-square test SC-specic genes P 0.94; Chi-square test broblast-specic genes P 0.81)

(h) Volcano plots showing the distribution of the genes perturbed upon exposure to NRG1/Fsk in the two comparisons (Chi-square test P 0.82).

(i) UHC of broblast-specic genes in the three cell populations.

4 NATURE COMMUNICATIONS | 8:14088 | DOI: 10.1038/ncomms14088 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14088 ARTICLE

SC derivatives showed comparable levels of many SC-enriched transcripts, including Cnp, Erbb3, GalC, Mal, Mapk3, Mbp, Mpz, Pmp22 and Pou3f1 (Supplementary Fig. 5b). To corroborate the similarities between the two SC types, we ltered for those genes physiologically relevant in SCs during development24 and subdivided SC- and broblast-specic DEGs according to fold change and statistical signicance. Noteworthy, analysis of the DEG frequency distribution did not show any deviation of the observed frequencies (iSCs) from the expected frequencies (SCs) neither in SC- and broblast-specic DEGs, nor in SC developing genes (Fig. 2g and Supplementary Fig. 5d). Similarly, analysis of the DEG frequency distribution of those genes perturbed upon exposure to the differentiating factors NRG1 and Fsk did not show any deviation of the DEG frequencies (Fig. 2h)25. UHC ltering for genes belonging to biological pathways relevant for SC biology revealed an extensive clustering between iSCs and SCs (Supplementary Fig. 5ej). Conversely, iSCs showed a consistent silencing of the genes specically enriched in the broblasts. In fact, UHC analysis of a representative array of broblast-specic genes revealed that the majority of these were either strongly downregulated or even completely silenced in iSCs (Fig. 2i). These results were corroborated by qRTPCR assays showing that somatic and induced SCs exhibited comparable levels of expression for randomly selected genes such as Col2a1, Col5a2, Ecm1 and Prrx1 (Supplementary Fig. 5c). Of note, only a minimal fraction of this class of genes (o5%), normally very poorly detectable in somatic SCs, were still expressed in 3-weeks reprogrammed iSCs although to a reduced extent compared with broblasts (Supplementary Fig. 5c). These results show that iSCs do not represent a hybrid state of SCs and the donor cell type, but instead exhibit a global transcriptional reconguration towards the reprogrammed cell type with minimal signs of memory at gene expression level of the cells of origin.

bFGF is a potent mitogen for iSCs. We, next, asked whether additional molecules added during reprogramming could further improve iSCs properties. We selected VEGF, basic broblast growth factor (bFGF) and TGFb for their known role during

Schwann cell commitment and maturation26. Whereas with VEGF and TGFb we did not observe any evident effect, in the presence of bFGF, iSCs displayed a signicant increase in cell proliferation as shown by an increased number of Ki67 cells within the O4 cell population (Fig. 3a,b). In addition, bFGF treatment promoted a sustained proliferation rate over time in iSC cultures (Supplementary Fig. 6). Importantly, bFGF-treated cells strongly upregulated GFAP, a molecular marker of immature SCs while signicantly repressing the myelinating-associated protein Mbp (Fig. 3ce). In accordance, bFGF sustained a global change in iSC morphology from mature (with extensive lamellipodia) to immature (thin bipolar or tripolar cells) cellular traits (Fig. 3fi). Thus, the presence of bFGF during the phase of cell expansion fostered a substantial increase in the number of puried iSCs with an immature morphology, Gfap expression and high proliferation rate providing a convenient condition to expand iSCs for multiple passages (tested up to P7) in vitro.

iSCs are efcient in generating myelin in vitro. To determine the myelinogenic potential of iSCs in vitro, we adapted a co-culture system using mouse dorsal root ganglion (DRG) neurons27,28. DRGs were allowed to develop an extensive axonal network before seeding iSCs pre-treated with or without bFGF. iSCs and DRG co-cultures were maintained for 4 weeks without dox. Intriguingly, in all conditions iSCs, but not broblasts, were able to generate Mbp myelin segments surrounding neuronal

axons (Fig. 4a and Supplementary Fig. 7a). However, bFGF-stimulated iSCs produced a three-fold increase in the overall amount of Mbp myelin segments reaching levels close to those displayed by somatic SCs (Fig. 4b). Likewise also internodal length was comparable between bFGF-stimulated iSCs and primary SCs (Supplementary Fig. 7b). In all conditions, iSC-derived myelin developed mature and well-organized internodes as showed by the correct localization of the paranodal Caspr protein and nodal voltage-gated sodium channels (NaV)29 (Fig. 4cf). Electron microscopy imaging revealed the formation of multiple and compact myelin sheaths around the axons conrming the correct structure of the iSC-derived myelin (Fig. 4g,h). Electron microscopy analyses showed also the presence of non-myelinating iSCs and pre-myelinating iSCs (Supplementary Fig. 7c) indicating that these cells can recapitulate the full behaviour of SCs during myelination. Altogether, these results demonstrate that iSCs acquired a remarkable myelination potential, in particular when previously exposed to exogenous bFGF, generating myelin segments with a correct morphology and organization.

iSCs from Twicher mice are impaired in myelin formation. Setting a fast and straightforward method for the generation of myelinating SCs is particularly attractive for in vitro disease modelling, since their isolation from mouse tissues and subsequent culture and differentiation is particularly laborious and poorly efcient. Thus, to determine whether iSC reprogramming could represent a useful platform to model PNS pathology in vitro, we focused on the well-characterized Twitcher (Twi) mouse model30,31, which carries a spontaneous mutation inactivating the beta-Galactocerebrosidase (GALC) gene function and closely recapitulates the disease manifestations of the human globoid cell leukodystrophy (GLD, or Krabbes disease), a severe neurodegenerative lysosomal storage disorder. In GLD, lack of functional GALC results in the accumulation of galactosylceramide (GALCER) and the toxic lysolipid galactosylsphingosine (psychosine) mainly in CNS and PNS myelinating cells, causing progressive demyelination and neurodegeneration32,33. Nonetheless, models of in vitro myelination are not yet available for this disease. Although Twi primary skin broblasts displayed some pathological features such as the expansion of the lysosomal compartment and GALCER and psychosine accumulation (Supplementary Fig. 8), they proliferated normally and could be reprogrammed into iSCs at comparable efciency with their wild-type (WT) counterparts (Supplementary Fig. 9a,b). In addition, upon bFGF treatment Twi and WT iSCs acquired a comparable morphology and growth expansion (Fig. 5a,b and Supplementary Fig. 9c). However, only Twi iSCs displayed signicant psychosine accumulation when compared with parental broblasts, and evident cytoplasmic deposition of GALCER was observed in Twi iSC as compared with WT counterparts (Fig. 5ce and Supplementary Fig. 9d,e). Intriguingly, Twi iSCs showed a remarkable immature morphology with a higher number of Ki67 proliferating cells compared with WT counterparts (Fig. 5h and Supplementary Fig. 9f,g).

Likewise, a higher number of Twi iSCs upregulated the immature SC marker GFAP while repressing Mbp expression (Fig. 5f,g and Supplementary Fig. 9h,i). These results are in line with a previous study showing that in Twi mice SCs displayed immature features and altered myelin sheath morphology34. Next, we assessed the myelinogenic potential of Twi and WT iSCs in organotypic cultures with explanted Twi and WT DRGs in different genotype combinations (Fig. 5i). Intriguingly, co-cultures of both Twi DRG and iSCs revealed a striking loss of MBP internodes and those remaining exhibited a consistent reduction in average

NATURE COMMUNICATIONS | 8:14088 | DOI: 10.1038/ncomms14088 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 5

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14088

a

Control

+bFGF

b

50 Cntr

**

**

***

***

+bFGF

Cntr +bFGF

Cntr +bFGF

% Positive

cells/O4+

O4/Ki67

S100/O4

S100/MPZGFAP

40

30

28

21

14

7

30

22.5

15

7.5

90

67.5

45

22.5

130

122.5

115

107.5

100

20 Ki67

0 GFAP

0 MBP

c

Control

+bFGF

d

% Positive

cells/total

e

MBP

% Positive

cells/total

f

Mature SC-like morphology Immature SC-like morphology

g

IM MM

S100 O4 MPZ

% Positive

cells/S100+

0 Control

+bFGF

h

Control +bFGF

i

**

*

% Positive cells/

total/control

+bFGF

Figure 3 | bFGF maintains iSCs in an immature and highly proliferative state. Immunouorescence (a) and quantication (b) for Ki67 staining in O4 iSCs with or without bFGF stimulation 1 week after dox withdrawal (Means.d., n 5 independent experiments, three coverslips for each, 10

randomly selected 20 elds per sample were examined). Immunouorescence (c) and quantication (d,e) for GFAP and MBP staining in the presence or

not of bFGF (Means.d., n 3 independent experiments, two coverslips/experiment/antigen, 10 randomly selected 20 elds per sample were

examined). (f) Schematic view of a single SC with a mature (left) or immature (right) morphology. (g) Quantication of S100 iSCs with mature (MM) and immature (IM) morphology in the two bFGF conditions (Means.d., n 3 independent experiments, three coverslips/experiment/antigen, 10

randomly selected 20 elds per sample were examined). Immunouorescence (h) and quantication (i) for S100/O4 and S100/MPZ stainings in iSCs

at week 1 after dox withdrawal in the absence (left) or presence (right) of bFGF (Means.d., n 3 independent experiments, 10 randomly selected

20 elds per sample were examined). *Po0,05; **Po0,01; ***Po0,001; Students t-test (b,d,e); One-way ANOVA (g); Multiple t-test (i).

Scale bars, 50 mm.

length compared with their counterparts in WT co-cultures at 4 weeks (Fig. 5i, bottom right and top left, Fig. 5j,k and Supplementary Fig. 10). In addition, clustering of sodium channels was impaired suggesting alterations in the nodal structures in Twi iSC-derived myelin (Fig. 5l, bottom right and top left, Fig. 5m). Conversely, in association with Twi DRG, WT iSCs did not reveal any alteration indicating that the loss of

GALC enzyme in axons is not affecting the morphological features of the surrounding myelin (Fig. 5i top right and Fig. 5j,k). To determine whether the observed functional defects in Twi SCs could be eventually cross-corrected by WT cells, we co-cultured Twi iSCs with WT DRGs and scored for the presence of myelination defects. Remarkably, in the presence of WT DRGs, the number of Mbp internodes and their length were

6 NATURE COMMUNICATIONS | 8:14088 | DOI: 10.1038/ncomms14088 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14088 ARTICLE

FibiSCs bFGF iSCs SCs

** ***

***

a

Fib

iSCs bFGF iSCs SCs

b

NF/MBP

Internodes per field

400

300

200

100

0 MBP

***

c e

Caspr/MBP NaV/MBP

d f

g h

Figure 4 | In vitro myelination properties of iSCs. (a) Myelin generated by broblasts (Fib), iSCs with or without bFGF retreatment and somaticSCs plated on mouse DRG axons as visualized by immunouorescence for MBP (internodes, red), and Neurolament (NF, green) at 2 weeks of co-culture. (b) Quantication of the myelin internodes generated by the four different cells types after 2 weeks of co-culture with DRG neurons (Means.d., n 4

independent experiments, 48 coverslip/experiment, 10 randomly selected 20 elds per coverslip were examined). (c,d) Representative

immunouorescence staining for MPB internodes together with Caspr (c) or NaV (d) to visualize the paranodal (arrowsheads) and nodal (arrowheads) regions, respectively, at 4 weeks of iSCs with DRG neurons co-cultures. (e,f) High-magnication elds of the boxed area in c,d, respectively.

(g,h) Representative electron microscopy pictures of myelinated iSCs at 4 weeks of co-culture. (h) Is a high-magnication image of the boxed area in e. **Po0,01; ***Po0,001; One-way ANOVA with Bonferroni correction. Scale bars, 500 mm (a), 50 mm (c,d), 1 mm (e), 100 nm (f).

signicantly normalized and maintained over time in the Twi iSC-derived myelin (Fig. 5i, bottom left, Fig. 5j,k, and Supplementary Fig. 10). Similarly, correct localization of sodium channels at the nodes was strongly ameliorated in these co-cultures (Fig. 5l, bottom left, and Fig. 5m). These results suggest that Twi iSC-related dysfunctions can be largely reverted by WT DRGs. It is well known that lysosomal enzymes are normally released in the extracellular space and can be re-uptaken by neighbour cells (cross-correction)31,35. Our ndings suggest that in our experimental setting the functional GALC enzyme produced by WT DRGs is available for metabolic cross-correction of Twi iSCs. Indeed, previous studies showed that metabolic cross-correction and reconstitution of enzymatic activity mediated by WT NSCs was effective in rescuing critical functions in Twi NSCs and mature progeny in vitro and in vivo36,37. Altogether, these results provide evidence that the iSC cellular platform is a powerful and exible system where

investigating pathological mechanisms affecting SC biology and myelination processes.

Sox10 and Egr2 drive reprogramming into human iSCs. Next, we asked whether human broblasts of non-neural crest origin could be reprogrammed into SCs. To this end, human postnatal broblasts were transduced with Sox10 and Egr2 expressing lentiviruses and cultured for 2 weeks in SC inducing medium. In these conditions, 50.6% of the total cell population (n 3

independent experiments. Values are meanss.d.) acquired a strong expression of the two SC markers S100 and O4 (Fig. 6a,b). Conversely, no S100 /O4 cells were detectable in tTA infected or not infected broblast cultures (Fig. 6a, inset). We, then isolated the O4 iSC population by ow-cytometry (Fig. 6c and

Supplementary Fig. 11a,b) and grew the cells for 1 week without doxycycline for the following analysis. Fluorescence-activated cell

NATURE COMMUNICATIONS | 8:14088 | DOI: 10.1038/ncomms14088 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 7

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14088

sorting-puried human iSCs (hiSCs) activated several SC markers such as MBP, ERBB3, MAG, EGR2, SOX10 and, most importantly, MPZ (Fig. 6df; Supplementary Fig. 11c). Surprisingly, differently from rodent iSCs, hiSCs exhibited a strong expression of both GFAP and MBP (Fig. 6g). As expected, sorted cells did not express oligodendrocyte-specic transcription factors and markers such as MOG, OLIG1 and OLIG2, NG2 and NKX2.2

(Supplementary Fig. 11d), conrming also in this case the PNS regional identity of iSCs.

To functionally characterize hiSCs we adapted the pseudopod assay to study the paracrine effects of iSCs on DRG and vice versa38. In this indirect co-culture setting SCs were placed over a lter on the top of a Boyden chamber while the DRGs seeded at the bottom of the multiwell (Supplementary

100 80

60

40

20

0

a b

Cntr +bFGF

% Positive cells/

total

**

80

75

50

25

0 S100 O4 MPZ

60

40

20

0

c

WT

Twi

d e

WT Twi

WT Twi

Lamp1/GALCER

S100/O4 S100/MPZ

*** *

***

3

Lamp1 area/

nuclei/WT

4 9

6

3

0 Psy

2

Pmol mg1 protein

1

0

f

WT Twi

g

h

36 **

WT Twi

% Positive cells/

total

18

% Positive cells/

S100+

27

GFAP

100

75

50

25

0

9

0 GFAP

IM

**

WT Twi

i l

WT DRG

Twi DRG WT DRG Twi DRG

MBP

WT iSCs Twi iSCs

NaV/MBP

WT iSCs Twi iSCs

j k m

WT DRG

WT DRG

WT DRG

Twi DRG

Twi DRG

Twi DRG

** *** ******

***

***

*

MBP+ internodes per

field fold change to WT

0.825

0.55

0.275

0 WT iSCs Twi iSCs WT iSCs Twi iSCs WT iSCs Twi iSCs

1.1 40

30

20

10

0

Internode length (um)

NaV+ nodes per field

(%)

90

67.5

45

22.5

0

***

8 NATURE COMMUNICATIONS | 8:14088 | DOI: 10.1038/ncomms14088 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14088 ARTICLE

erasure of the donor cell epigenetic memory40,41. iSCs showed a robust in vitro myelinogenic capacity generating with time correctly structured and patterned myelin sheaths when co-cultured with DRG neurons. To our knowledge, this is the rst time that MPZ SCs with an extremely high myelinating potential are obtained by cell reprogramming. The identication of bFGF as a potent mitogen for iSCs is relevant for enabling the effective expansion of a pure population of iSCs after isolation by immune-purication. Interestingly, initial bFGF treatment also strengthens subsequent iSC differentiation potential and its associated myelin generating capability. This action of bFGF on iSCs is in line with its crucial action in promoting the commitment of immature SCs and maintaining their expansion in vitro25. Thus, this condition uncovers the opportunity for a large-scale production of iSCs to generate sufcient number of functional iSCs that can be exploited for autologous cell replacement therapies.

The fast and straightforward process to generate iSCs should facilitate in vitro disease modelling of peripheral demyelination disorders and axonal damage. In fact, these studies have been hampered by the intrinsic challenge to isolate these cells in high number and sufcient purity, in particular from mice, in which the majority of the genetic models for human diseases are developed. Herein, we provided a strong proof of concept that Twi iSCs retained the cellular and metabolic alterations described in peripheral nerve cells in vivo, which result in signicant reduced myelination capability and severe alterations in myelin organization42,43. This setting enabled us to validate a signicant rescue of the pathological defects of iSCs by the co-culture with WT DRGs. Further, we provided a direct evidence of a non-cell autonomous mechanism of disease correction likely relying on efcient uptake by Twi iSCs of a functional GALC enzyme produced and secreted by WT DRGs (cross-correction). This phenomenon provides the rationale and is thought to underlie the remarkable benecial effects of cell and gene therapeutic approaches for lysosomal storage disorders in mice and humans31,44. However, its direct consequences were not yet corroborated in a physiological in vitro model of myelination. Thus, iSCs might be exploited as a useful platform to decipher the molecular mechanisms controlling lysosomal enzymatic release and re-uptake in the contest of normal and pathological SC biology and axonal myelination. More in general, this system can result informative for a broad spectrum of disease conditions.

The iSC direct reprogramming disclosed the opportunity to establish the rst strategy of autologous cell transplantation, thereby circumventing immunogenicity, for peripheral nerve injury.

Figure 5 | Disease related features of Twitcher (Twi) iSCs and derived myelin. (a) Immunouorescence analysis for S100/O4 and S100/MPZ in O4 immune-puried O4 Twi iSC population reprogrammed in the presence of bFGF. (b) Bar graphs showing percentages of S100, O4 and MPZ positive iSCs cells after bFGF treatment or in naive conditions (Means.d., n 3 independent experiments, three coverslips/experiment/antigen, 10 randomly selected

20 elds per sample were examined). (c) Representative immunouorescence pictures for Lamp1 (green) and the lipid metabolite GALCER (red)

staining in WT and Twi iSCs. (d) Bar graph showing the Lamp1 positive area relative to the total nuclei in Twi iSCs, expressed as fold change (Means.d., WT and Twi n 2 independent lines, three coverslips/line, 10 randomly selected 20 elds per coverslip were examined). (e) Mass-spectrometry-based

quantication of psychosine (Psy) in WT and Twi iSCs (Means.d., WT and Twi n 2 independent lines, three samples per line were examined).

(f) Representative immunouorescence images for GFAP in WT and Twi iSCs. (g) Quantication of GFAP cells in WT and Twi iSCs (Means.d., n 3

independent experiments, three coverslips/line, 10 randomly selected 20 elds per sample were examined). (h) Quantication of S100 WT and Twi

iSCs with immature (IM) morphological traits (Means.d., n 3 independent experiments, 10 randomly selected 20 elds per sample were examined).

(i) Representative low-magnication pictures of MBP immunostaining to label myelin internodes of iSCs co-cultured with DRG neurons in different combinations between WTand Twi genotypes. (j,k) Quantication of the number (j) and length (k) of MBP internodes in the four different combinations of iSCs and DRG neurons from WT or Twi genotypes (Means.d., n 2 independent experiments, 48 coverslip/experiment, 10 randomly selected

20 elds per coverslip were examined). (l) Representative immunouorescence pictures and quantication (m) of the NaV myelin nodes generated

after 4 weeks of by co-culture of iSCs with DRG neurons from either WT or Twi genotypes (Means.d., n 2 independent experiments, four coverslip/

experiment, 10 randomly selected 20 elds per coverslip were examined). *Po0,05; **Po0,01; ***Po0,001; Students t-test (b,d,e,h,g); One-way

ANOVA with Bonferroni correction (j,k,m). Scale bars, 50 mm (a,c,f); 500 mm (i); 20 mm (l).

Fig. 12a). Three DIV after co-culture, we evaluated the DRG axonal projection area and the pseudopod formation on the internal surface of the Boyden chamber. Interestingly, DRGs, grown in the presence of the hiSCs and primary SC, exhibited signicantly longer axons compared with DRGs maintained alone or with control broblasts (Fig. 6h, top lane and Fig. 6i), indicating a dual positive paracrine effect between iSCs and DRGs. Furthermore, alike primary SCs and different from broblasts, hiSCs failed to extend pseudopods on the internal surface of the Boyden chamber as expected by the distance separating SCs and DRGs which precluded to convey the contact-mediated juxtacrine signals (Fig. 6h, bottom lane and Fig. 6j)39. Rat iSCs showed a comparable response in these assays as both hiSC and somatic mouse SCs (Supplementary Fig. 12bd). Finally, to test the behaviour of the hiSCs to juxtacrine signals, we established direct co-cultures between hiSCs and mouse DRGs. Unfortunately, a reliable protocol for the in vitro myelination of hSCs with primary neurons has not yet been developed to date. For these reasons we focused on the early stages of myelination, in which SCs form pseudopods to take direct contact with axons. Interestingly, after 5 DIV of co-culture, electron microscope imaging showed multiple examples of iSCs contacting several axons and emanating pseudopods to surround them, a behaviour normally observed with rodent SCs in the early phase of myelination (Fig. 6k and Supplementary Fig. 12e).

These data clearly indicated that the combination of Sox10 and Egr2 transcription factors is also effective in reprogramming human broblasts into induced SCs with an expression pattern and cellular behaviour comparable to their somatic counterpart.

DiscussionHerein, we identied Sox10 and Egr2 as a unique combination of factors necessary for the effective generation of myelinogenic iSCs by the direct conversion of adult rodent broblasts. Remarkably, transcriptome studies revealed a substantial equivalence in the SC-specic molecular signature between somatic and reprogrammed SC progenies with a prevailing epigenetic silencing of the donor cell gene expression prole. Only a minor fraction of specic broblast-specic genes (o5%) were found still expressed in iSCs, although at reduced levels compared with the donor cells. This might indicate that 3 weeks of reprogramming in these conditions might not be sufcient to fully reshape the activity of these gene loci. This nding is consistent with previous observations in induced pluripotent stem and broblastto-neuron cell reprogramming, and suggests that the time in culture after reprogramming is a critical factor for the pervasive

NATURE COMMUNICATIONS | 8:14088 | DOI: 10.1038/ncomms14088 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 9

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14088

a

SE

b c

S100 256

192

128

64

0 00 101 102 103 104 105 O4 FITC Log

10

S100/O4

***

SE

% Positive cells/

total

R32.61%

7.5

5

2.5

SSC

tTA

0 tTA SE

S100/O4

S100/O4 GFAP

d

e

g

S100 O4 MPZ

O4 MPZ

100

110

95

80

65

50

% Positive cells/

total

75

50

25

0

f

% Positive cells/

S100+

S100/MPZ

NF/DAPI

MBP

h i

Empty Fib iSC SC

Fib iSC SC

***

**

**

**

Total area of neuritis/

total area of the

explant body

26

19.5 136.5 0

Empty Fib iSC SC

j

***

***

Pseudopods

area %

40 30 20 10

Phalloidin

0 Empty Fib iSC SC

k

Figure 6 | Direct conversion of human skin broblasts into iSCs. (a) S100 (green) and O4 (red) immunouorescence staining of Sox10/Egr2 (SE) or tTA-infected (inset) postnatal broblasts. (b) Quantication of S100 and S100/O4 cells expressed as percentage over the total number of cells in experimental (SE) and control (tTA) conditions at 14 DIV (Means.d., n 4 independent experiments, 10 randomly selected 20 elds per sample were

examined). (c) Fluorescence-activated cell sorting-based purication of the O4 iSC population. (d) S100/O4 and S100/MPZ immunouorescence staining in iSCs. (e,f) Quantication of 14 DIV iSCs expressing S100, O4 and MPZ alone or in combination (Means.d., n 3 independent experiments,

three coverslips/experiment/antigen, 10 randomly selected 20 elds per sample were examined). (g) Immunouorescence for GFAP and MBP staining.

(hj) Pseudopod assay: sample images and quantication of DRG axonal growth (h, top lane, and i) and pseudopod formation (h, bottom lane, and j) after 5 days of indirect co-culture (Means.d., n 3 independent experiments, 23 coverslip/experiment, 10 randomly selected 20 elds per coverslip

were examined). (k) Representative electron microscopy pictures of one iSC or SC and their pseudopods (brown) contacting multiple axons (green) at 7 days of direct co-culture. **Po0,01; ***Po0,001; Students t-test (b); Multiple t-test (i,j). Scale bars, 50 mm (a,d,h bottom); 500 mm (h, top); 1 mm (k).

Importantly, the same minimal combination of two factors triggered the conversion of postnatal human broblasts into hiSCs. Despite the low reprogramming efciency, hiSCs could be

puried for O4 expression to establish enriched hiSC cultures. Thus, this approach provides a fast and easy protocol to produce hiSCs for biomedical studies. The hurdles experienced in the

10 NATURE COMMUNICATIONS | 8:14088 | DOI: 10.1038/ncomms14088 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14088 ARTICLE

past in obtaining hSCs have hindered the development of protocols for in vitro myelination and in vivo transplantation. hiSC generation opens new opportunities to advance experimental methodologies using human cells. This is the rst evidence of the generation of reprogrammed SCs endowed with a robust myelinogenic potential suitable for informative in vitro studies on physiological and pathological mechanisms of myelin formation/maintenance in peripheral nerves. Therefore, our ndings represent a crucial step for the potential future application of iSCs in human PNS disease biology and therapy.

Methods

Cell culture. MEFs were isolated from E13.5 wild-type and GALC-decient Twitcher mice (mixed C57BL6/FVB background; referred to as Twi mice in this study) mice embryos. Head, vertebral column, dorsal root ganglia and all internal organs were removed and discarded and the remaining embryonic tissue was manually minced and incubated in 0.25% trypsin (Sigma) for 1015 min for cell dissociation. Cells from each embryo were plated onto a 15-cm tissue culture dish in MEF media (Dulbeccos modied Eagle medium (Invitrogen) containing 10% fetal bovine serum (FBS; Hyclone), non-essential amino acids (Invitrogen),

L-glutamine (Invitrogen), sodium pyruvate and penicillin/streptomycin (Invitrogen). In all experiments cells were used only within the second passage and not split more than four times. Mouse neonatal and adult broblasts and rat neonatal broblasts were isolated from tail tip samples. Tails were peeled, minced into 1 mm pieces, placed on culture dishes, and incubated in MEF media for 5 days.

Primary rat SCs were prepared as described45 and maintained in SC media (Dulbeccos modied Eagle medium (Invitrogen), 10% FBS (Hyclone), L-glutamine (Invitrogen), 2.5 mM Forskolin (Fsk, Sigma Aldrich), 2.5 ng ml 1 recombinant human NRG1b1 (NRG1, EGF domain, R&D), until used.

MRC5 broblast cell line was acquired from ATCC (ATCC CCL-171) and maintained in MEF media.

Mouse DRG was isolated from E14.5 wild-type and Twitcher embryos, and established on collagen-coated glass coverslips as described46. Explants were cycled with the anti-mitotic reagent 5-uoro-2-deoxy-uridine to eliminate all nonneuronal cells. Neuronal media was supplemented with 50 ng ml 1 NGF (Harlan,

Bioproducts for Science) and, in some cases, 25 ng ml 1 BDNF (PeproTech) and 10 ng ml 1 NT3 (Austral Biologicals). Rat SCs, Fibroblasts and iSCs (200,000 cells/coverslip) were added to established explant cultures of

DRG neurons, and myelination was initiated by supplementing media with50 mg ml 1 ascorbic acid (Sigma-Aldrich). Primary human SCs were obtained by Dr K. Haastert-Talini47. Testing for Mycoplasma cell contamination was routinely performed.

Molecular cloning and viral infection. Complementary DNAs (cDNAs) for reprogramming transcription factors were cloned into lentiviral vectors under the control of the tetracycline operator. Replication-incompetent, VSVg-coated lentiviral particles were packaged in 293T cells. MEFs, neonatal and adult mouse and neonatal rat broblasts were infected in MEF media. Sixteen to twenty hours after infection, cells were switched into fresh MEF media containing doxycycline (2 mg ml 1; Sigma). After 48 h medium was replaced with Schwann cell media containing doxycycline. The medium was changed every 23 days for a further 1222 days.

Immunocytochemical assays. For immunocytochemical analysis, 5 104 bro-

blasts were plated on matrigel-coated glass coverslips the day before the infection. Ten to twenty-eight days after viral infection cells were xed for 20 min at room temperature (2025 C) in 4% paraformaldehyde in PBS, permeabilized for 30 min in PBS containing 0.1% Triton X-100 and 10% normal goat serum, and incubated overnight at 4 C in PBS containing 10% normal goat serum and primary antibodies. Then cells were washed three times with PBS and incubated for2 h at room temperature with secondary antibodies. Co-cultures were permeabilized in methanol and stained as described previously48. A list of antibodies used can be found in Supplementary Table 1. The percentages of the cells that were positive to specic markers were determined from at least three independent experiments for each condition. Images for quantication were selected randomly. First, the number of 4,6-diamidino-2-phenylindole-positive nuclei was counted, followed by counting the number of cells expressing the markers of interest in at least ve elds per condition. To quantify the immunopositive area (expressed in pixels) pictures in each coverslip/sample were taken and the immunopositive area was calculated using the ImageJ software and normalized on nuclear counts. Cells from WT mice were used to set the signal threshold in the comparison between WT and Twitcher samples.

RTPCR analysis. RNA was extracted and puried with the RNeasy Mini kit (QIAGEN). Two micrograms of RNA were reverse transcribed Transcriptor High Fidelity cDNA Synthesis Kit (Roche). One twentieth of the reverse transcribed cDNA was amplied in a 25 microlitres of reaction mixture containing Taq polymerase buffer (Fisher BioReagents), 0.2 mM dNTPs (Finnzymes OY, Espoo,

Finland), 0.4 micromolar each primer, 1 U Taq polymerase (Fisher BioReagents). Primers for gene expression analysis are listed in Supplementary Table 2. The thermal prole consisted of a rst denaturing step for 5 min at 95 C, 30 cycles of 30 s at 95 C, 30 s at the specic annealing temperature (see Supplementary Table 2) and 40 s at 72 C, followed by a nal extension of 10 min at 95 C.

RNA sequencing and bioinformatics analysis. Total RNA isolation was performed with the RNeasy Mini Kit (QIAGEN). Libraries were prepared with the TruSeq RNA Library Preparation Kit v2 (Illumina) and sequenced with the Genome Analyser IIx (Illumina). Bioinformatics analysis was performed with the GeneProf web-based software49,50. Sequence quality control and alignment to the Rn5 reference genome were performed with TopHat 2.1.0 (ref. 51). Estimates of gene expression levels were calculated with the Quantitate Gene Expression module. Differential gene expression was performed with DESeq v2.6 (ref. 52). Statistical analysis was performed with SPSS Statistics (IBM) software and GENE-E (Broad Institute) web-based application. Functional enrichment analysis was performed with FunRich53.

Electron microscopy. Electron microscopy analyses on Schwann cellDRG neurons myelinating co-cultures were performed as described previously54. Cells were xed in a solution containing 2.5% glutaraldehyde and 0.5% sucrose in 0.1 M Srensen phosphate buffer (pH 7.2) for 46 h. The specimens were then washed in a solution containing 1.5% sucrose in 0.1 M Srensen phosphate buffer for a period of 612 h, post-xed in 2% osmium tetroxide, dehydrated and embedded in Glauerts embedding mixture. Semi-thin sections perpendicular were cut on an Ultracut UCT ultramicrotome (Leica, Wetzlar, Germany), and stained with toluidine blue. From the same tissue blocks, ultra-thin sections (5070 nm) were also cut using the same ultramicrotome and placed on copper grids. Grids were then stained with uranyl acetate and lead citrate and observed on a JEM-1010 transmission electron microscope (JEOL, Tokyo, Japan) operating at 80 kV and equipped with a Mega-View-III digital camera and a Soft-Imaging-System(SIS, Mnster, Germany) for the computerized acquisition of the images.

Determination of GALC activity. Cells were harvested, washed in PBS, lysed for 1 h in 10 mmol l 1 sodium phosphate buffer, pH 6.0, containing 0.1% (vol/vol)

Nonidet NP-40, and subjected to sonication. These steps were performed at 4 C. Protein content was measured using the Bradford Protein Assay kit with bovine serum albumin as the reference standard. GALC activity is measured as previously described55. Briey, the assay is performed using 27.5 mg of sample proteins with the articial uorogenic substrate 4-methylumbelliferone-b-galactopyranoside(1.5 mM) resuspended in 0.1/0.2 mol l 1 citrate/phosphate buffer, pH 4.0, and AgNO3 (11 mM) at 37 C. The enzymatic reaction is stopped by adding 0.2 M pH 10.6 Glycine/NaOH. Fluorescence of liberated 4-methylumbelliferone is measured on a spectrouorometer (l ex 360 nm, l em 446 nm).

Psychosine dosage. Psychosine quantication was performed on frozen cellular pellets by Dr W. Kulik (AMC, Amsterdam), as previously described56. Briey, dimethylpsychosine was added to cell-homogenates (0.20.5 mg of protein) as internal standard. After deproteinization and extraction by methanol:chloroform and evaporation to dryness of the organic layer, the sample extract was reconstituted in acidic methanol. After injection and elution over a C8 analytical column, psychosine and its internal standard were detected by positive ESI in the MRM mode on a TSQ Quantum AM mass spectrometer. Pychosine concentrations were established by the use of calibration standards.

Pseudopod assay. Pseudopod assay was performed as previously described38. with few modications: mouse DRG neurons were plated on the bottom ofa 24-well plate. Cells (1.5 105) were starved for 12 h, and then trypsinized with

0.5% TrypsinEDTA, pelleted, resuspended and plated into the upper compartment of the Boyden chamber insert. We used 0.33 cm2 polycarbonate transwell permeable inserts with 3 mm pores (Corning #3415) coated with matrigel.

Cells were maintained in Schwann cell media without doxycycline for 4 h at 37 C to attach and spread on the upper surface of the insert. Then, inserts were washed with PBS and transferred to a 24-well plate containing DRGs in a medium without ascorbic acid. Pseudopods were allowed to grow at 37 C. After 3 days, the upper surface of each membrane was cleaned with a cotton swab, and the pseudopods that migrated into the lower surface were xed in 4% paraformaldehyde and stained with phalloidin (Alexa Fluor 594 phalloidin, Invitrogen). Filters were, next, excised from the insert with microscissors and mounted on a slide for subsequent imaging analysis.

Animals. Mice were maintained at San Raffaele Scientic Institute Institutional mouse facility (Milan, Italy) in micro-isolators under sterile conditions and supplied with autoclaved food and water. Sprague Dawley rats were housed in the animal facility at the Neuroscience Institute Cavalieri Ottolenghi, c/o Ospedale San Luigi, Orbassano, Turin, Italy. All procedures were performed according to protocols approved by the internal IACUC and reported to the Italian Ministry of Health according to the European Communities Council Directive 2010/63/EU.

NATURE COMMUNICATIONS | 8:14088 | DOI: 10.1038/ncomms14088 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 11

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14088

Statistical analysis. Statistical signicance between two samples was determined with Students t-test (GraphPad Prism 5; GraphPad). The statistical signicance among multiple samples was determined with one-way analysis of variance with Bonferronis multiple comparison tests. The data were considered statistically signicant for Po0.05 and are shown as the means.d. or s.e.m as indicated. All data were acquired from at least three independent experiments.

Data availability. Microarray data have been deposited in the NCBI Gene Expression Omnibus database under accession code GSE78712. The authors declare that all data supporting the ndings of this study are available within the article and its Supplementary Information les or from the corresponding author upon reasonable request.

References

1. Jessen, K. R. & Mirsky, R. The origin and development of glial cells in peripheral nerves. Nat. Rev. Neurosci. 6, 671682 (2005).

2. Fancy, S. P., Chan, J. R., Baranzini, S. E., Franklin, R. J. & Rowitch, D. H. Myelin regeneration: a recapitulation of development? Annu. Rev. Neurosci. 34, 2143 (2011).

3. Jessen, K. R., Mirsky, R. & Arthur-Farraj, P. The role of cell plasticity in tissue repair: adaptive cellular reprogramming. Dev. Cell 34, 613620 (2015).

4. Geuna, S., Raimondo, S., Fregnan, F., Haastert-Talini, K. & Grothe, C. In vitro models for peripheral nerve regeneration. Eur. J. Neurosci. 43, 287296

2016:

5. Lehmann, H. C. & Hke, A. Use of engineered Schwann cells in peripheral neuropathy: hopes and hazards. Brain Res. 1638, 97104 (2016).

6. Tao, Y. Isolation and culture of Schwann cells. Methods Mol. Biol. 1018, 93104 (2013).

7. Lee, G. et al. Isolation and directed differentiation of neural crest stem cells derived from human embryonic stem cells. Nat Biotechnol. 25, 14681475 (2007).

8. Sakaue, M. & Sieber-Blum, M. Human epidermal neural crest stem cells as a source of Schwann cells. Development 142, 31883197 (2015).

9. Krause, M. P. et al. Direct genesis of functional rodent and humanschwann cells from skin mesenchymal precursors. Stem Cell Reports 3, 85100 (2014).

10. Vierbuchen, T. et al. Direct conversion of broblasts to functional neurons by dened factors. Nature 463, 10351041 (2010).

11. Pang, Z. P. et al. Induction of human neuronal cells by dened transcription factors. Nature 476, 220223 (2011).

12. Caiazzo, M. et al. Direct conversion of broblasts into functional astrocytes by dened transcription factors. Stem Cell Reports 4, 2536 (2015).

13. Liu, M. L., Zang, T. & Zhang, C. L. Direct lineage reprogramming reveals disease-specic phenotypes of motor neurons from human ALS patients. Cell Rep. 14, 115128 (2016).

14. Heinrich, C. et al. Sox2-mediated conversion of NG2 glia into induced neurons in the injured adult cerebral cortex. Stem Cell Reports 3, 10001014 (2014).

15. Yang, N. et al. Generation of oligodendroglial cells by direct lineage conversion. Nat. Biotechnol. 31, 434439 (2013).

16. Najm, F. J. et al. Transcription factor-mediated reprogramming of broblasts to expandable, myelinogenic oligodendrocyte progenitor cells. Nat. Biotechnol. 31, 426433 (2013).

17. Svaren, J. & Meijer, D. The molecular machinery of myelin gene transcription in Schwann cells. Glia 56, 15411551 (2008).

18. Pereira, J. A., Lebrun-Julien, F. & Suter, U. Molecular mechanisms regulating myelination in the peripheral nervous system. Trends Neurosci. 35, 123134 (2012).

19. Ghislain, J. & Charnay, P. Control of myelination in Schwann cells: a Krox20 cis-regulatory element integrates Oct6, Brn2 and Sox10 activities. EMBO Rep. 7, 5258 (2006).

20. Monk, K. R., Feltri, M. L. & Taveggia, C. New insights on Schwann cell development. Glia 63, 13761393 (2015).

21. Etxaniz, U. et al. Neural-competent cells of adult human dermis belong to the Schwann lineage. Stem Cell Reports 3, 774788 (2014).

22. Nave, K. A. & Werner, H. B. Myelination of the nervous system: mechanisms and functions. Annu. Rev. Cell Dev. Biol. 30, 503533 (2014).

23. Srinivasan, R. et al. Genome-wide analysis of EGR2/SOX10 binding in myelinating peripheral nerve. Nucleic Acids Res. 40, 64496460 (2012).

24. Buchstaller, J. et al. Efcient isolation and gene expression proling of small numbers of neural crest stem cells and developing Schwann cells. J. Neurosci. 24, 23572365 (2004).

25. Ishii, M. et al. A stable cranial neural crest cell line from mouse. Stem Cells Dev. 21, 30693080 (2012).

26. Haastert, K. et al. Differentially promoted peripheral nerve regeneration by grafted Schwann cells over-expressing different FGF-2 isoforms. Neurobiol. Dis. 21, 138153 (2006).

27. Taveggia, C. et al. Neuregulin-1 type III determines the ensheathment fate of axons. Neuron 47, 681694 (2005).

28. La Marca, R. et al. TACE (ADAM17) inhibits Schwann cell myelination. Nat. Neurosci. 7, 857865 (2011).

29. Eshed-Eisenbach, Y. & Peles, E. The making of a node: A co-production of neurons and glia. Curr. Opin. Neurobiol. 23, 10491056 (2013).

30. Lattanzi, A. et al. Therapeutic benet of lentiviral-mediated neonatal intracerebral gene therapy in a mouse model of globoid cell leukodystrophy. Hum. Mol. Genet. 23, 32503268 (2014).

31. Ricca, A. et al. Combined gene/cell therapies provide long-term and pervasive rescue of multiple pathological symptoms in a murine model of globoid cell leukodystrophy. Hum. Mol. Genet. 24, 33723389 (2015).

32. Sakai, N. Pathogenesis of leukodystrophy for Krabbe disease: Molecular mechanism and clinical treatment. Brain Dev. 31, 485487 (2009).

33. Graziano, A. C. E. & Cardile, V. History, genetic, and recent advances on Krabbe disease. Gene 555, 213 (2015).

34. Kagitani-Shimono, K., Mohri, I., Yagi, T., Taniike, M. & Suzuki, K. Peripheral neuropathy in the twitcher mouse: accumulation of extracellular matrix in the endoneurium and aberrant expression of ion channels. Acta Neuropathol. 115, 577587 (2008).

35. Schultz, M. L., Tecedor, L., Chang, M. & Davidson, B. L. Clarifying lysosomal storage diseases. Trends Neurosci. 34, 401410 (2011).

36. Neri, M. et al. Neural stem cell gene therapy ameliorates pathology and function in a mouse model of globoid cell leukodystrophy. Stem Cells 29, 15591571 (2011).

37. Santambrogio, S. et al. The galactocerebrosidase enzyme contributes to maintain a functional neurogenic niche during early post-natal CNS development. Hum. Mol. Genet. 21, 47324750 (2012).

38. Poitelon, Y. et al. Spatial mapping of juxtacrine axo-glial interactions identies novel molecules in peripheral myelination. Nat. Commun. 6, 8303 (2015).39. Muratori, L. et al. Can regenerated nerve bers return to normal size? A long-term post-traumatic study of the rat median nerve crush injury model. Microsurgery 32, 383387 (2012).

40. Kim, K. et al. Epigenetic memory in induced pluripotent stem cells. Nature 467, 285290 (2010).

41. Marro, S. et al. Direct lineage conversion of terminally differentiated hepatocytes to functional neurons. Cell Stem Cell 9, 374982 (2011).

42. Dolcetta, D. et al. Myelin deterioration in Twitcher mice: motor evoked potentials and magnetic resonance imaging as in vivo monitoring tools.J. Neurosci. Res. 81, 597604 (2005).43. Smith, B. et al. Peripheral neuropathy in the Twitcher mouse involves the activation of axonal caspase 3. ASN Neuro. 3, e00066 (2011).

44. Bif, A. et al. Lentiviral hematopoietic stem cell gene therapy benets metachromatic leukodystrophy. Science 341, 1233158 (2013).

45. Einheber, S. et al. The axonal membrane protein Caspr, a homologue of neurexin IV, is a component of the septate-like paranodal junctions that assemble during myelination. J. Cell Biol. 139, 14951506 (1997).

46. Zanazzi, G. et al. Glial growth factor/neuregulin inhibits Schwann cell myelination and induces demyelination. J. Cell Biol. 152, 12891299 (2001).

47. Haastert, K., Mauritz, C., Chaturvedi, S. & Grothe, C. Human and rat adult Schwann cell cultures: fast and efcient enrichment and highly effective non-viral transfection protocol. Nat. Protoc. 2, 299104 (2007).

48. Taveggia, C. et al. Neuregulin-1 type III determines the ensheathment fate of axons. Neuron 47, 681694 (2005).

49. Halbritter, F., Vaidya, H. J. & Tomlinson, S. R. GeneProf: analysis of high-throughput sequencing experiments. Nat. Methods 9, 78 (2012).

50. Halbritter, F., Kousa, A. I. & Tomlinson, S. R. GeneProf data: a resource of curated, integrated and reusable high-throughput genomics experiments. Nucleic Acids Res. 42, D851D858 (2014).

51. Trapnell, C., Pachter, L. & Salzberg, S. L. TopHat: discovering splice junctions with RNA-Seq. Bioinformatics 25, 11051111 (2009).

52. Anders, S. & Huber, W. Differential expression analysis for sequence count data. Genome Biol. 11, R106 (2010).

53. Pathan, M. et al. FunRich: An open access standalone functional enrichment and interaction network analysis tool. Proteomics 15, 25972601 (2015).

54. Bolis, A. et al. Dlg1, Sec8, and Mtmr2 regulate membrane homeostasis in Schwann cell myelination. J. Neurosci. 29, 88588870 (2009).

55. Martino, S. et al. Specic determination of beta-galactocerebrosidase activity via competitive inhibition of beta-galactosidase. Clin. Chem. 55, 541548 (2009).

56. Santambrogio, S. et al. The galactocerebrosidase enzyme contributes to maintain a functional neurogenic niche during early post-natal CNS development. Hum. Mol. Genet. 21, 47324750 (2012).

Acknowledgements

We are thankful to G. Lazzari, D. Meijer, A. Bolino and S. Previtali for sharing of reagents and all members of the Broccolis lab for helpful discussion, P. Podini for electron microscopy imaging, W. Kulik (AMC, Univ of Amsterdam) for the psychosine dosage, F.

12 NATURE COMMUNICATIONS | 8:14088 | DOI: 10.1038/ncomms14088 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14088 ARTICLE

Morena and S. Martino (Dipartimento di Medicina Sperimentale e Scienze Biochimiche, Sezione di Biochimica e Biologia Molecolare, Universit degli Studi di Perugia) for determination of GALC activity and K. Haastert-Talini for providing primary human SCs. We acknowledge the FRACTAL core facility for expert supervision in ow-cytometry. This work was supported by the European Research Council (AdERC #340527).

Author contributions

V.B. and P.G.M. conceived and planned the experiments; P.G.M. performed the experiments and analysed data; L.M. elaborated transcriptomic data; P.G.M., M.P. and C.T. performed in vitro co-cultures; G.R. and S.G. carried out in vivo isolation and characterization of rat broblasts; A.R. and A.G. provided Twitcher samples and analysed lipid metabolites; A.Q. performed and analysed electron microscopy imaging data; A.I. and S.G.G. contributed to gene cloning and lentiviral particles production; A.S. and C.C. supported cell culture; M.C. and U.D.C. contributed to immunouorescence staining; V.B. supervised and supported the project and wrote the paper with P.G.M., S.G., C.T and A.G.

Additional information

Supplementary Information accompanies this paper at http://www.nature.com/naturecommunications

Web End =http://www.nature.com/ http://www.nature.com/naturecommunications

Web End =naturecommunications

Competing nancial interests: The authors declare no competing nancial interests.

Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions/

Web End =http://npg.nature.com/ http://npg.nature.com/reprintsandpermissions/

Web End =reprintsandpermissions/

How to cite this article: Mazzara, P. G. et al. Two factor-based reprogramming of rodent and human broblasts into Schwann cells. Nat. Commun. 8, 14088doi: 10.1038/ncomms14088 (2017).

Publishers note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional afliations.

This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the articles Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

Web End =http://creativecommons.org/licenses/by/4.0/

r The Author(s) 2017

NATURE COMMUNICATIONS | 8:14088 | DOI: 10.1038/ncomms14088 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 13