A r t i c l e s

1262 VOLUME 18 | NUMBER 8 | AUGUST 2012 nAture medicine

Skeletal muscle and skin have exceptional regenerative capacity and can undergo many rounds of repair in response to injury1,2. Although parenchymal regeneration is ensured by tissue stem cells, efficient tissue repair requires activation of the surrounding stromal micro­environment, mainly α­smooth muscle actin–positive (α­SMA+) myofibroblasts, to produce growth and survival factors, proinflam­matory chemokines and components of the extracellular matrix, such as collagens. Inappropriate activation of the stromal compartment, however, leads to persistence of α­SMA+ collagen­overproducing cells (also called profibrotic cells), extracellular matrix accumulation and overgrowth of fibrous or adipogenic tissue, thereby compromis­ing organ recovery and impairing function3. In the skeletal muscle, overgrowth of fibrous tissue has been suggested to originate from muscle stem cells exposed to environmental modifications associated with aging or injury4,5. In addition, resident platelet­derived growth factor receptor­α–positive (PDGFR­α+) mesenchymal cells distinct from muscle stem cells proliferate early upon muscle damage to pro­mote tissue regeneration and show fibrogenic­adipogenic bipotential in vitro6,7. Accordingly, chronic activation of PDGFR­α leads to wide­spread organ fibrosis in mice8.

Myofibroblasts are a functionally heterogeneous population of cells and, at least in vitro, can be generated from several cell types, including circulating bone marrow–derived CD45+CD34+ fibrocytes, fibroblasts, epithelial cells, mesenchymal stem cells, perivascular cells and nerve­associated cells9–13. Genetic cell­fate mapping would allow the identification of relevant precursors to profibrotic cells in vivo, in the context of acute muscle and dermis injury, with minimal

perturbation of the biological system being investigated14. However, such genetic approaches require specific markers that have yet to be reported for profibrotic cells.

ADAM12 is a membrane­anchored metalloprotease that is implicated in the signaling of epidermal growth factor, insulin­like growth factor and transforming growth factor­β (TGF­β)15–17. ADAM12 is transiently expressed during embryonic morphogenesis of skeletal muscles and visceral organs18 and is re­expressed in several human diseases with a fibrotic component, such as liver and muscle injuries, scleroderma and desmoplastic tumors including aggressive fibromatosis19–23. Such a restricted and specific expression pattern prompted us to investigate the identity and role of ADAM12+ cells during injury. Using models of acute injury in the muscle and the dermis, we demonstrate here that ADAM12+ cells are programmed during vascular wall development and are re­induced by injury as a distinct subset of PDGFR­α+ perivascular progenitors with a specific profibrotic fate and function. Accordingly, such cells can be specifically targeted to limit the generation of profi­brotic cells and collagen accumulation during scarring.

RESULTSInjury induces the development of ADAM12+ PDGFR-a+ cellsTo label ADAM12+ cells, we generated transgenic mice, termed M12­CIG, expressing GFP and Cre recombinase under control of the Adam12 locus on a bacterial artificial chromosome (BAC; Fig. 1a). We have shown earlier that acute tissue injury induced by complete Freund’s adjuvant (CFA) injection into the ear dermis induces, within days, a massive generation of proinflammatory gp38+ myofibroblasts

1Institut Pasteur, Lymphoid Tissue Development Unit, Paris, France. 2Centre National de la Recherche Scientifique (CNRS), Unité de Recherche Associée (URA)1961, Paris, France. 3Institut Pasteur, Centre d’Ingénierie Génétique Murine, Paris, France. Correspondence should be addressed to L.P. (lucie.peduto@pasteur.fr).

Received 24 January; accepted 26 May; published online 29 July 2012; doi:10.1038/nm.2848

Lineage tracing and genetic ablation of ADAM12+ perivascular cells identify a major source of profibrotic cells during acute tissue injurySophie Dulauroy1,2, Selene E Di Carlo1,2, Francina Langa3, Gérard Eberl1,2 & Lucie Peduto1,2

Profibrotic cells that develop upon injury generate permanent scar tissue and impair organ recovery, though their origin and fate are unclear. Here we show that transient expression of ADAM12 (a disintegrin and metalloprotease 12) identifies a distinct proinflammatory subset of platelet-derived growth factor receptor-a–positive stromal cells that are activated upon acute injury in the muscle and dermis. By inducible genetic fate mapping, we demonstrate in vivo that injury-induced ADAM12+ cells are specific progenitors of a major fraction of collagen-overproducing cells generated during scarring, which are progressively eliminated during healing. Genetic ablation of ADAM12+ cells, or knockdown of ADAM12, is sufficient to limit generation of profibrotic cells and interstitial collagen accumulation. ADAM12+ cells induced upon injury are developmentally distinct from muscle and skin lineage cells and are derived from fetal ADAM12+ cells programmed during vascular wall development. Thus, our data identify injury-activated profibrotic progenitors residing in the perivascular space that can be targeted through ADAM12 to limit tissue scarring.

npg

© 2

012

Nat

ure

Am

eric

a, In

c. A

ll rig

hts

rese

rved

.

A r t i c l e s

nAture medicine VOLUME 18 | NUMBER 8 | AUGUST 2012 1263

expressing low to medium levels of α­SMA (α­SMAlow/med) and genes essential for leucocyte recruitment and survival24. In M12­CIG mice, we detected Adam12­expressing GFP+ cells in a subset of CFA­induced gp38+ stromal cells that developed around muscle fibers close to the ear cartilage (Fig. 1a) and infiltrated the dermis during the following week (Supplementary Fig. 1a). A similar process was observed in skeletal muscle after cardiotoxin­induced injury (Fig. 1a), a model for acute muscle injury and regeneration25. At 4 d after injury, ADAM12+ cells represented <1% of total nonhematopoietic (CD45−), nonendothelial (CD31−) stromal cells (Fig. 1b). Gp38+ stro­mal cells expanded markedly during the first few days after injury and then decreased progressively over 1 month (Fig. 1c). Other stromal populations, such as PDGFR­α+ cells and PDGFR­β+ cells, expanded similarly during the first week and partially overlapped with gp38+ cells (Supplementary Fig. 1b). The highest number of ADAM12+ cells occurred 4 d after injury (representing 4.6% of total gp38+ cells in the muscle); however, their numbers decreased rapidly thereafter in contrast to gp38+ stromal cells (Fig. 1c). A similar kinetic profile has been reported for muscle­resident PDGFR­α+Sca­1+ mesenchymal cells proliferating upon muscle damage; this cell population shows fibrogenic­adipogenic bipotential, with roles in tissue regeneration and ectopic fat formation6,7. Accordingly, injury­induced ADAM12+ cells all expressed PDGFR­α and Sca­1 (Fig. 1c and Supplementary Fig. 1c,d), although they represented only a minor fraction of total PDGFR­α+ mesenchymal cells.

At 4 d after CFA­ or cardiotoxin­induced injury, ADAM12+ cells expressed transcripts for nerve growth factor receptor (Ngfr, also known as p75NTR or p75), neural cell adhesion molecule 1 (Ncam1) and glial fibrillary acidic protein (Gfap; Fig. 1d). This phenotype is reminiscent of Schwann cells dedifferentiating upon nerve injury26. Such a phenotype has also been reported for particular stromal cell populations, such as hepatic stellate cells, which are the main collagen­ expressing cells upon liver injury27. Notably, p75 is required for the differentiation of stellate cells into α­SMA+ collagen­producing myofibroblasts28. In the first 2 weeks after injury, similar to our ear­lier findings on gp38+ cells in inflamed tissues24, ADAM12+ cells expressed proinflammatory chemokines and cytokines (Fig.  1d). These cells also expressed transcripts for connective tissue growth factor (Ctgf) and Tgfb1, both of which are involved in the differentia­tion of myofibroblasts and Pdgfrb, the latter of which has an essential role in pericyte recruitment29.

The potential of injury­induced ADAM12+ to differentiate into α­SMA+ myofibroblasts was confirmed in vitro. ADAM12+ stromal cells isolated 4 d after injury were mostly spherical and poorly adher­ent. When grown in TGF­β–supplemented medium, the majority of cells (>85%) differentiated into adherent α­SMA+ myofibroblasts with prominent stress fibers (Fig. 1e). However, in contrast to the case with ADAM12− cells, we observed only minor differentiation of ADAM12+ cells into adipocytes when they were grown in adipocyte differentiation medium. Consistent with a transient progenitor state, ADAM12+ cells

a

Day 4 15 4 15

CFA-dermis

+ + – –4 15

Adam12NgfrNcam1GfapIl1bCcl27Cxcl12Cxcl13Il6Cxcl1Ccl19PdgfrbFgf1Tgfb1Ctgf

4 15

CX-muscle

+ + – –GFP

c

e

b

d f

Injury

Gp38 GFP DAPI

CF

A-d

erm

isC

X-m

uscl

e

– 4 d

M12-CIGTGpAdam12 Cre Ires Egfp pA

CD45–

CD31–

WT littermate

0 0.42

0 0.62

MI12-CIG

CF

A-d

erm

isC

X-m

uscl

eG

FP

FSC

CX

Gp38

4.9

0.9 4.6 2 1.2 0

32.6 28.1 8.3

PDGFR-α

4 d 15 d 30 d

CD45–

CD31–

WT littermate

–

FS

CG

FP

Growth mediumMyo�broblast

medium (TGF-β)

α-SMA DAPI

Adipocytemedium

Oil Red O

AD

AM

12+

AD

AM

12–

100

Cel

ls (

%)

α-SMA+

80

60

40

20

0

ADAM12

+

ADAM12

–

Oil Red O+

100

ADAM12

+

ADAM12

–

Are

a de

nsity

(%

)

80

60

40

20

0

ADAM12+ cells

0 5

Time of culture + TGF-β (d)

0 5

Adam12 Acta22.5 0.5

Rel

ativ

e ge

ne e

xpre

ssio

n

2.0

1.5

1.0

0.5

0

0.4

0.3

0.2

0.1

0

Figure 1 ADAM12 expression identifies a proinflammatory subset of gp38+PDGFR-α+ stromal cells transiently induced upon dermis and muscle injury. (a) Immunofluorescence staining of ADAM12 (GFP) and gp38 in stromal cells developing 4 d after injection of CFA in ear dermis (top) and cardiotoxin (CX) in tibialis anterior muscle (bottom) of M12-CIG mice. A schematic diagram of the BAC transgene in M12-CIG mice is shown at the top. pA, poly(A). Insets show gp38+GFP+ cells at higher magnification (×400). (b) Fluorescence-activated cell sorting (FACS) plots and percentage of nonhematopoietic (CD45−) and nonendothelial (CD31−) stromal cells expressing GFP in M12-CIG mice at 4 d after injury. WT, wild type. FSC, forward scatter. (c) Top, FACS plots and percentage of CD45−CD31− stromal cells expressing gp38 before injury (−) and at 4 d, 15 d and 30 d after cardiotoxin-induced muscle injury in M12-CIG mice. Bottom, percentage of gp38+ stromal cells expressing GFP and PDGFR-α. (d) Gene expression analysis in ADAM12+ and ADAM12− stromal cells isolated from dermis or muscle of M12-CIG mice at 4 d and 15 d after injury. Transcripts were measured by quantitative RT-PCR (qRT-PCR). Data are mean of three to five independent experiments. (e) Cell differentiation of ADAM12+ cells and ADAM12− stromal cells isolated by FACS at 4 d after cardiotoxin injury and grown for 5 d in TGF-β-supplemented myofibroblast medium or adipogenic differentiation medium. Percentage of myofibroblasts (α-SMA+) and adipocytes (Oil Red O+) was assessed in ten fields from three independent experiments. Insets show higher magnification of α-SMA+ stress fibers (top) and Oil Red O+ cells (bottom) (×400). Data are mean and s.d. (f) Gene expression of Adam12 and Acta2, as measured by qRT-PCR, in freshly isolated ADAM12+ cells (4 d after cardiotoxin injury) and after 5 d of differentiation in TGF-β–supplemented medium. In a–c, f, data are representative of two to four independent experiments. Scale bars, 100 µm.

npg

© 2

012

Nat

ure

Am

eric

a, In

c. A

ll rig

hts

rese

rved

.

A r t i c l e s

1264 VOLUME 18 | NUMBER 8 | AUGUST 2012 nAture medicine

downregulated expression of Adam12 during differentiation to myofi­broblasts while upregulating the transcript for Acta2 (encoding α­SMA; Fig. 1f ). Together, these data indicate that a subset of gp38+PDGFR­α+Sca­1+ cells, transiently expressing ADAM12 upon injury, preferen­tially differentiates into proinflammatory myofibroblasts.

Injury-induced myofibroblasts are progeny of ADAM12+ cellsADAM12+ cells were most prominent during the first week after injury and cell numbers declined thereafter (Fig. 1c). To determine the fate of ADAM12+ cells in vivo, we mapped their cellular progeny by crossing M12­CIG mice to Cre­dependent Rosa26floxSTOP­RFP reporter mice30 (M12cre/RFP, Fig. 2a). In such mice, all the prog­eny of ADAM12+ cells was permanently labeled with RFP, whereas ADAM12+ cells expressed both GFP and RFP. RFP+ cells, present before injury albeit at low numbers, markedly expanded upon injury (Fig.  2b) to represent a major fraction of gp38+ and PDGFR­α+ stromal cells within a few days to 2 weeks after injury (Fig.  2c and Supplementary  Fig.  2a, respectively) and then progressively decreased in number as the tissue healed. The expansion of RFP+ cells followed extensive proliferation of ADAM12+ cells in the first days after injury, whereas proliferation of RFP+GFP− cells was less marked but still substantial 15 d after injury (Fig. 2d). At 20 d after injury, 60–70% of α­SMA+ myofibroblasts were progeny of ADAM12+ cells (Fig. 2e). In contrast, no muscle fibers or skin epi­dermal cells expressed RFP (Supplementary Fig. 2b), indicating that ADAM12+ cells are developmentally distinct from muscle and skin tissue stem cells activated upon injury. Finally, a parabiosis experi­ment, in which an M12­CIG mouse and a wild­type partner were surgically sutured together, confirmed that ADAM12+ cells were resident and not recruited from the bloodstream (Fig. 2f ). Together,

these results indicate that a majority of gp38+, PDGFR­α+ and α­SMA+ myofibroblasts generated in vivo upon muscle and dermis acute injury is derived from tissue­resident ADAM12+ cells.

ADAM12+ profibrotic cells are induced de novo by injuryAs ADAM12 is expressed during ontogeny18, and cells derived from ADAM12+ progenitors were detected before injury (Fig. 2b, time 0), we investigated whether myofibroblasts were generated from ADAM12+ cells induced de novo by injury. To that end, we carried out an inducible, tetracycline transactivator–based cell fate mapping experiment. We generated M12tTA/LC1/YFP mice that express tetra­cycline transactivator under control of the Adam12 locus on a BAC (M12­tTATG), Cre under control of the tetracycline transactivator (LC­1; ref. 31) and the conditional reporter Rosa26floxSTOP­YFP locus32 (Fig. 3a). In these triple­transgenic mice, YFP labeling of the prog­eny of ADAM12+ cells is controlled in time by the administration of doxycycline, which blocks the tetracycline transactivator–mediated labeling cascade, allowing for the distinct fate mapping of fetal and adult injury­induced ADAM12+ cells (Fig. 3b).

First, we administered doxycycline continuously from concep­tion until shortly before tissue injury (embryonic day 0 to postnatal day 23, E0–P23) to map the fate of adult ADAM12+ cells induced de novo by injury. In that setting, and 10 d after injury, 50% (in the dermis) and 60% (in the tibialis anterior muscle) of the injured areas were infiltrated with YFP+ cells (Fig. 3c, doxycyline E0–P23) with a CD45−gp38+PDGFR­αmed/high phenotype (Fig. 3d), demonstrat­ing that a major fraction of the injury­induced stroma is generated from ADAM12+ cells induced de novo by injury. In these conditions, up to 53% of total gp38+ cells in the whole tibialis anterior muscle were progeny of ADAM12+ cells at 20 d after injury; this percentage

a

c

e f

d

bM12cre/RFP mice

Injury

CF

A-d

erm

isC

X-m

uscl

e

– 15 d100 GFP

RFP

GFPRFP

Per

cent

age

of E

du+ c

ells

GF

P a

ndR

FP

are

a de

nsity

(%

)

pAdam12

pRosa26 STOP Rfp

Cre

Cre Ires Egfp pAM12-CIGTG

Rosa26floxSTOP-RFP

×

75

50

100

RFP EDU

50

25

00 4

1 2 3 4 8 15

15 30Time afterinjury (d)

Time after injury (d)

Injury

Injury

Injury 20 d

M12

-CIG

WT

litte

rmat

e

CX

CX

CX Parabiosis

GFP

M12-GFP

WT littermate

CFA

Injury

CF

A-d

erm

is

CF

A-d

erm

is

Per

cent

age

of R

FP

amon

g α-

SM

A c

ells

M12

-CIG

M12

cre/

RF

P

CX

-mus

cle

CX

-mus

cle

RF

P

RFP+ α-SMA

– 4 d

53.1 86 84.7 60.6

9.6

FSC

80.2

Merge

68.7 33

15 d 30 dCD45–

CD31–

Gp38+40

30

20

10

0

75

50

25

0

RFP GFP DAPI

EDU GFPEDU GFP

Figure 2 The majority of myofibroblasts developing in vivo upon injury are progeny of ADAM12+ cells. (a) Strategy for fate mapping of ADAM12+ cells. (b) Immunofluorescence staining of ADAM12 (GFP, white arrowheads) and cellular progeny of ADAM12+ cells (RFP) in M12cre/RFP mice before injury and 15 d after injection of CFA in ear dermis (top) or of cardiotoxin in tibialis anterior muscle (bottom). Proportion of GFP+ cells (green) and RFP+ cells (red) among total cells (blue) was measured by pixel quantification at the indicated time after injury. Values are mean and s.d. of CFA and cardiotoxin data; n = 5–10 sections from three independent experiments, with representative pictures. Insets show RFP+GFP+ cells at higher magnification (×250). (c) FACS analysis of stromal cells before injury (−) and at 4 d, 15 d and 30 d after CFA (top) or cardiotoxin (bottom) induced injury in M12cre/RFP mice. Numbers indicate the percentage of gp38+ cells expressing RFP. (d) Proliferation rate of GFP+ and GFP−RFP+ cells in CFA-injured dermis of M12-CIG and M12cre/RFP mice, respectively (n = 10 sections from two independent experiments). Representative pictures of Edu immunostaining in GFP+ cells (3 d after injury) and in RFP+ cells (15 d after injury) are shown on the left. DAPI (blue) stains the cell nuclei. Scale bar, 50 µm. (e) Immunofluorescence staining of α-SMA expression (cyan) by RFP+ cells (red) in M12cre/RFP mice at 20 d after injury. Quantification was done in ten sections from three independent experiments (representative pictures). (f) Immunofluorescence staining (right) of GFP in sections of injured muscle of a parabiotic pair between M12-CIG mice and wild-type littermate mice as illustrated (left) at 4 d after cardiotoxin injection. In c,f, data are representative of two to four independent experiments. In d,e, values are mean ± s.d. Scale bars, 100 µm (b,e,f).

npg

© 2

012

Nat

ure

Am

eric

a, In

c. A

ll rig

hts

rese

rved

.

A r t i c l e s

nAture medicine VOLUME 18 | NUMBER 8 | AUGUST 2012 1265

gradually decreased as the tissue healed (Fig. 3e). Most YFP+ cells expressed α­SMAlow/med, consistent with progressive differentiation into α­SMA+ myofibroblasts (Fig. 3f ). Furthermore, most progeny of injury­induced ADAM12+ cells colocalized with interstitial colla­gen deposits, both during muscle regeneration and when healing was mostly achieved at 40 d after injury (Fig. 3g), showing that de novo–induced ADAM12+ cells generated a subset of stromal cells targeted to fibrotic regions of the tissue.

The presence of a subset of α­SMA+Col1A+YFP− cells (20–30% at 10 d after injury, Supplementary Fig. 3a) indicated that the effi­ciency of the tetracycline transactivator–based labeling system is not 100%, or that subsets of myofibroblasts arise from distinct origins and potentially carry different functions. At 40 d after injury, inter­stitial mesenchymal cells in regenerated normal muscle tissue were not progeny of ADAM12+ cells (Supplementary Fig. 3b).

Next, we administered doxycyline continuously from birth (P1–P50) to map the fate of fetal ADAM12+ cells. In the ear dermis, the progeny of fetal ADAM12+ cells was already substantial before injury and further expanded to include most stromal cells after injury (Fig. 3c, P1–P50), indicating that ADAM12 is expressed during ontogeny by early mesenchymal progenitors (such as cranial neural crest cells, which generate all mesenchyme of the head33). In skel­etal muscle, the progeny of fetal ADAM12+ cells, which was limited before injury, expanded to 40% of injured areas (Fig. 3c, P1–P50).

In both ear dermis and muscle, the progeny of fetal ADAM12+ cells included α­SMA+ cells (Fig. 3f) and other cell types (such as nerves, Supplementary  Fig.  3c), in contrast to the progeny of de  novo–induced ADAM12+ cells. Thus, the progeny of fetal ADAM12+ cells included both fibrogenic and nonfibrogenic lineages, but only cells of the fibrogenic lineage reexpressed ADAM12 upon acute injury (schematic diagram, Supplementary Fig. 4).

Ablation of injury-induced ADAM12+ cells decreases fibrosisTo better define the role of ADAM12+ cells and their cellular progeny in scarring, we sorted ADAM12+PDGFR­α+, ADAM12−PDGFR­α+ and ADAM12−PDGFR­α− stromal cells from ADAM12­reporting mice (GFP+ cells) at 4 d or 10 d after injury, as well as the progeny of ADAM12+ cells (YFP+ cells) from M12tTA/LC1/YFP mice at 20 d after injury. In the first weeks after injury, ADAM12+ cells expressed Col1a1 transcripts more highly than ADAM12−PDGFR­α+ cells, although the former had lower or similar expression of Acta2 at 4 d and 10 d, respectively (Fig. 4a). At 20 d after injury, Adam12 expres­sion became undetectable in the progeny of ADAM12+ cells, whereas the PDGFR­αmed subset of this progeny most highly expressed Col1a1 and Acta2. These data demonstrate that ADAM12+ cells and their cellular progeny are a subset of PDGFR­αhigh/med cells that produce excess amounts of type I collagen during differentiation toward the myofibroblast phenotype.

d eDox E0–P23 Gated YFP+ CX –

0 0 4.7 26.9 5.3 48 2.4 26.7

6010.9406.744.823.691.88.200.19

88.910.9

10 d 20 d 40 dCD45–

CD31–

Gp38+

Gp3

8

CD45 PDGFR-α PDGFR-α

YF

P

YF

P

pAdam12 tTA

Luc (TRE)7-pminhCMV Cre

STOPpRosa26 Yfp

Cre

tTA Dox

Ires Dtr pA

a bM12tTA/LC1/YFP mice

Dox

E0–P23

YFP

YFPP1–P50

E0–P50

E0 Birth P20 P30 P40 P50

AnalysisInjury

×

×

M12-tTATG

LC-1

Rosa26floxSTOP-YFP

cDox:CX or CFA

E0–P23

10 d 10 d 10 d––

E0–P23

P1–P50

P1–P50

E0–P50

E0–P50

CFA-dermis

CF

A-d

erm

is

CX-muscle

CX

-mus

cle

10080604020

0

YF

P+ a

rea

dens

ity (

%)

10080604020

00 010 10 10 Time (d)

YFP DAPI

f g

Col1a

Col1a

CD

M

M

Merge

Mergeα-SMA

Merge

CD

Dox E0–P23

YFP

YFP

YFP

CX

20

dC

X 4

0 d

Col1a

Dox: E0–P23 P1–P50

CF

A-d

erm

isC

X-m

uscl

e

YFP α-SMA DAPI

Figure 3 ADAM12+ profibrotic progenitors are induced de novo by tissue injury. (a) Strategy for inducible fate mapping and ablation of ADAM12+ cells. tTA, tetracycline transactivator; Dtr, diphtheria toxin receptor; Luc, luciferase; TRE, tet-responsive element; hCMV, human cytomegalovirus; Dox, doxycycline. (b) Experimental design for fate mapping of fetal and adult ADAM12+ cells. E0, conception; P, postnatal day. (c) Immunofluorescence staining of YFP in ear dermis (top) and tibialis anterior muscle (bottom) of M12tTA/LC1/YFP mice treated as indicated. Quantification of YFP+ cells (green) among total cells (blue) was assessed by pixel quantification in injured regions versus normal tissue. Values are mean ± s.d.; n = 5–10 sections from three independent experiments, with representative images. (d) FACS plots and percentage of YFP+ cells expressing gp38 and PDGFR-α (right) in injured tibialis anterior muscles of M12tTA/LC1/YFP mice at 10 d after injecting cardiotoxin. (e) FACS plots and percentage of total gp38+ stromal cells expressing YFP and PDGFR-α in tibialis anterior muscle of M12tTA/LC1/YFP mice before injury, and at 10 d, 20 d and 40 d after injecting cardiotoxin. (f) Immunofluorescence analysis of α-SMA expression (red) by fetal (right) and adult (left) progeny of ADAM12+ cells (YFP) in M12tTA/LC1/YFP mice at 10 d after injury. (g) Immunofluorescence staining of type I collagen (red) and α-SMA (blue) by YFP+ cells at indicated time after cardiotoxin-induced injury in M12tTA/LC1/YFP mice. M, muscle tissue; CD, collagen deposit. In d,e,g, we treated M12tTA/LC1/YFP mice with doxycycline from conception until 1 week before injury. In d–g, representative data from three to four independent experiments are shown. Nuclei are stained with DAPI. Scale bars, 100 µm.

npg

© 2

012

Nat

ure

Am

eric

a, In

c. A

ll rig

hts

rese

rved

.

A r t i c l e s

1266 VOLUME 18 | NUMBER 8 | AUGUST 2012 nAture medicine

To formally demonstrate that ADAM12+ cells and their progeny have a causal role in tissue fibrosis during acute injury, we injected M12tTA/LC1/YFP mice, which also express the human diphthe­ria toxin receptor under control of the Adam12 locus (Fig. 3a), with diphtheria toxin subunit A (DTA) to induce the ablation of ADAM12+ cells. As we expected, injection of DTA induced a marked reduction of the progeny of ADAM12+ cells (YFP+ cells) (Fig. 4b) and of α­SMA+ stromal cells (Supplementary Fig. 5). Furthermore, consistent with the proinflammatory nature of ADAM12+ cells (Fig. 1d), leucocyte infiltration decreased upon DTA treatment (Fig. 4c). Because of the loss of ADAM12+ cells and their progeny, interstitial collagen was significantly reduced in injured tissues (Fig. 4d).

Next, to test whether ADAM12 itself has a role in fibrosis, we injected injured muscle of M12­CIG mice with siRNA directed against Adam12. At 3 weeks after injury, interstitial collagen within injured muscles was significantly reduced in treated mice (Fig. 4e), as was the expression of transcripts for stromal growth factors Ctgf and Fgf1. In contrast, growth factors essential for muscle regenera­tion such as Igf1 and Igf2 were increased when Adam12 was silenced (Fig. 4f). As TGF­β has a central role in wound healing and fibrosis

by stimulating expression of collagens34, we assessed whether ADAM12 might be involved in this process. Consistent with a role of ADAM12 in TGF­β signaling, forced expression of Adam12 in mouse fibroblastic cells in vitro increased Col1a1 expression in response to TGF­β, and silencing of endogenous Adam12 decreased Col1a1 expression upon TGF­β stimulation (Supplementary  Fig.  6a,b). Notably, Adam12 itself is rapidly, although transiently, induced by TGF­β in vitro (Supplementary Fig. 6c).

Injury reactivates ADAM12+ cells in nerves and vascular wall To clarify the identity of fetal and adult ADAM12+ progenitors, we mapped ADAM12 expression from ontogeny until before injury. We detected ADAM12 expression in p75+ neural crest cells (NCCs) at the outer border of the neural tube, starting at E9.5, when NCCs undergo an epithelial­to­mesenchymal transition and delaminate from the neural tube (Fig. 5a). NCCs subsequently migrate and give rise to multiple neural and non­neural cell types, including cells of the peripheral nervous system and craniofacial mesenchyme35. ADAM12 was downregulated in NCCs after delamination and re­expressed at E12.5 in NCC­derived p75+ Schwann cell precursors (SCPs), as defined by the lack of detectable expression of the Schwann cell marker S100

ADAM12+ cells (GFP)

a b

d

Progeny of ADAM12+ cells(YFP)

–DTA

Fib

rotic

sco

re

5

4

3

2 **

1

0

–DTA

+DTA

Col1a YFP DAPI

+DTA

CD45–

Rel

ativ

ege

ne e

xpre

ssio

n

Rel

ativ

ege

ne e

xpre

ssio

n

Rel

ativ

ege

ne e

xpre

ssio

n

CD31–

1

2

CD45–

CD31–

4.5 0.9

4.5

3.0

1.5

0

0.6

0.3

0

Adam12 Acta2

1 2

3.0

******

***

***

****** ***

***

*

1.5

0PDGFR-α

YFP+ + +

++ Hi Me +

+ +–––

–– –

4 d 10 d 20 d

–+GFPPDGFR-α

YFP+ + +

++ Hi Me +

+ +–

n.s.21

Col1a1

21

–––– –

4 d 10 d 20 d

–+ YFP+ + +

++ Hi Me +

+ +–––

–– –

4 d 10 d 20 d

–+GFPPDGFR-α

GFP

0.5

33.158.9

3.5 15.1

14.356

GF

PY

FP

PDGFR-α

PDGFR-α

e

Fib

rotic

sco

re

5

4

3

2

*

1

0

siRNA

cont

siRNA

Adam12

siR

NA

con

tsi

RN

A Adam

12

f

50

Adam12

Ctgf

Fgf1lgf1

lgf2

0

–50

Per

cent

age

of e

xpre

ssio

n

–100

c –DTA +DTA

125

100

*75

50

25

0

–DTA

+DTA

13.2 8

CD

45+ c

ells

per

fiel

d

FSC

–DTA +DTA

CD45 YFP

CD

45

–DTA +DTA

14.3 0.4

14.1

16.2

13.4

16.4

15.5

0.5

CD45–

CD31–

YF

PY

FP

PDGFR-α

Gp38

Figure 4 Ablation of ADAM12+ cells induced de novo by injury reduces inflammation and tissue fibrosis. (a) Expression of Adam12, Acta2 and Col1a1, as measured by qRT-PCR, in the indicated cell populations isolated by FACS (gating strategy is shown on the left; numbers indicate the percentage of cells per gated type of cells in one representative experiment) from tibialis anterior muscle of m12-CIG mice (1) or M12tTA/LC1/YFP mice (2) at the indicated times after cardiotoxin injection. Hi, high; Me, medium. Data are mean of two to four independent experiments. (b) FACS plots and percentage of CD45−CD31− stromal cells expressing YFP and PDGFR-α (top) or Gp38 (bottom) in injured tibialis anterior muscle of M12tTA/LC1/YFP mice, 20 d after cardiotoxin injection, that were injected with DTA (right) or vehicle (PBS, left; n = 3 independent experiments). (c) FACS plots and percentage of CD45+ cells (top) and immunofluorescence staining of CD45 and YFP (bottom) in muscles treated as in b. Number of CD45+ cells per injured area was counted in 10–15 sections from three independent experiments. (d) Immunofluorescence staining of type I collagen and YFP (left) and picrosirius staining (right) of sections from cardiotoxin-injured muscles treated as in b. (e) Picrosirius staining of cardiotoxin-injured tibialis anterior muscles of M12-CIG mice injected with siRNA targeting Adam12, compared with siRNA control. In d,e, we quantified fibrosis in picrosirius-stained sections at 3 weeks (n = 25 fields from three to five independent experiments, representative images). Error bars, s.d.; *P < 0.05, **P < 0.005, ***P < 0.0005, unpaired Student’s t-test. (f) Expression of genes coding for growth factors were measured by qRT-PCR in cardiotoxin-injured muscles treated as in e (n = 4 individual experiments, analysis of variance test). In a–d, M12tTA/LC1/YFP mice were given doxycycline from conception until 1 week before injury. Scale bars, 100 µm.

npg

© 2

012

Nat

ure

Am

eric

a, In

c. A

ll rig

hts

rese

rved

.

A r t i c l e s

nAture medicine VOLUME 18 | NUMBER 8 | AUGUST 2012 1267

calcium­binding protein­β (S100) and close association with PGP9.5+ axons36 (Fig.  5a and Supplementary  Fig. 7a). These find­ings are in accordance with earlier micro­array data showing highest expression of ADAM12 in NCCs at E9.5 and in SCPs at E12.5 (ref. 37). From E10 to E14, ADAM12 was also expressed in p75+ NCCs that migrated to the gut to generate the enteric nervous system (Supplementary  Fig. 7a). In addition, at about E12.5, a population of ADAM12+PDGFR­β+p75−α­SMA−CD31−pax7− mesenchymal cells was detected surrounding CD31+ endothelium in close proximity to somites (Fig. 5a and Supplementary Fig. 7a). As PDGFR­β–mediated signaling induces a pericyte fate in undifferentiated mesenchymal cells recruited to stabilize nascent vessels29, these data suggest that fetal ADAM12+ mesenchymal cells include PDGFR­β+ pericyte pro­genitors. During postnatal growth (before weaning age), ADAM12 was transiently reexpressed in nonmyelinating S100+ Schwann cells capping the axon terminals of the skeletal neuromuscular junc­tion (NMJ), also called terminal Schwann cells (Fig.  5b), and in PDGFR­β+ perivascular cells, in both skeletal muscle and dermis (Fig. 5b and Supplementary Fig. 7b; the NMJ were visualized by staining with α­bungarotoxin (α­Bgtx), which binds the nicotinic acetylcholine receptors at the NMJ).

Together, these results show that ADAM12 is transiently expressed in distinct subsets of NCC­derived cells as well as in mesenchymal cells in close proximity to vessels, during nerve and vascular wall development respectively. Their progeny in adult muscle and dermis therefore includes nerves and glial cells on one hand (as described for NCCs33; Supplementary  Fig.  3c) and mesenchymal perivas­cular cells on the other (Supplementary  Fig. 8a), which might generate myofibroblasts upon injury through adult ADAM12+ pro­genitor cells. We thus characterized adult ADAM12+ cells induced

within hours of injury. Reminiscent of ADAM12 expression pattern during postnatal growth, ADAM12 was first reexpressed in S100+ terminal Schwann cells capping the neuromuscular junctions and in PDGFR­β+ perivascular cells of injured regions (Fig. 5c). At 2–3 d after injury, ADAM12+p75+/lowS100−stromal cells not associated with the neuromuscular junction or blood vessels gradually infiltrated the interstitial stroma of injured regions (Fig. 5c). These results indicate that tissue injury reactivates ADAM12+ cells programmed during nerve and vascular wall development.

Perivascular ADAM12+ cells are profibrotic progenitorsWe found ADAM12+ cells among NCC­derived Schwann cells and perivascular cells. As both cells of the NCC lineage and perivas­cular cells have the potential to generate myofibroblasts11,12,38, we mapped the fate of Wnt1+ cells, as Wnt1 is expressed by NCCs39 but not by perivascular cells. In Wnt1­Cre mice (expressing Cre under the control of the Wnt1 promoter) crossed to Rosa26floxSTOP­YFP mice, nerves and S100+ Schwann cells (including terminal Schwann cells) in adult muscle and dermis were YFP+, and so were progeny of Wnt1+ cells (Supplementary Fig. 8b). In addition, myofibroblasts developing upon ear injury expressed YFP, as mesenchyme of the head derives from NCCs35 (Fig.  6a). However, in contrast to the progeny of ADAM12+ cells (RFP+ cells in Fig. 2b), no S100− stromal

Ontogeny

Postnatal

Tissue injury0 h

NC

C li

neag

eM

C li

neag

eN

CC

line

age

MC

line

age

NC

C li

neag

eM

C li

neag

e

E10.5NT

NCC

p75 GFP DAPI PGP9.5 GFP α-SMA DAPI

E12.5

NCC

MC

N

N

N

N N

N

MCMC

BV

BV

E12.5 SCP

NDRG

SCP SCP

a

b

c

p75 GFP α-SMA DAPI

α-Bgtx GFP DAPI

α-Bgtx GFP DAPI

α-Bgtx

α-Bgtx GFP DAPI

PDGFR-β GFP CD31

α-Bgtx GFP DAPI

tSC

NMJ

BV

tSC

Merge

Merge

PDGFR-β GFP

PDGFR-β GFP CD31

PDGFR-β

CD31

CD31 CD31 GFP

GFP

GFPS100

p75 GFP

NMJ

BV

PGP9.5

P1

P1 P12

NMJNMJ

tSC

BV

P6 P45

P45

p75 GFP α-SMA DAPI

CD31 GFP p75

S100 GFP α-SMA DAPI

PDGFR-βGFP CD31

100

Per

cent

age

ofG

FP

+ tS

Cs

Per

cent

age

of G

FP

+

periv

ascu

lar

cells

24 h

GFP

S100

48 h 72 h

75

50

25

0

100

75

50

25

0

1 6 12 21 45

1 6 12 21 45

Age (d)

Age (d)

PDGFR-β

Figure 5 Tissue injury reactivates ADAM12+ cells programmed during nerve and vascular wall development. (a) Immunofluorescence staining of ADAM12 (GFP) in sections from M12-CIG mouse embryos at the indicated embryonic age (days after conception), stained as indicated. White arrowheads, ADAM12+PDGFR-β+ perivascular cells in E12.5 embryos. Scale bars, 100 µm. (b) Immunofluorescence staining of ADAM12 (GFP) in sections from tibialis anterior muscles of M12-CIG mice at the indicated age (postnatal days), stained as indicated. White arrowhead, ADAM12+PDGFR-β+ perivascular cells in muscles of 12-day-old pups. Scale bars, 50 µm. (c) Immunofluorescence staining of ADAM12 (GFP) in sections from cardiotoxin-injured tibialis anterior muscle of M12-CIG adult mice at indicated time after cardiotoxin injection. White arrowheads, ADAM12+PDGFR-β+ perivascular cells. Similar results were obtained in CFA-injured dermis and cardiotoxin-injured skeletal muscle. Representative pictures of three independent experiments. Scale bar, 50 µm. NCC, neural crest cells; MC, mesenchymal cells; NT, neural tube; DRG, dorsal root ganglia; N, nerve; SCP, Schwann cell precursor; BV, blood vessel; tSC, terminal Schwann cell; NMJ, neuromuscular junction.

npg

© 2

012

Nat

ure

Am

eric

a, In

c. A

ll rig

hts

rese

rved

.

A r t i c l e s

1268 VOLUME 18 | NUMBER 8 | AUGUST 2012 nAture medicine

progeny of Wnt1+ cells was found upon muscle injury (Fig.  6a), demonstrating that ADAM12+ profibrotic progenitors developing in cardiotoxin­injured tibialis anterior muscles are not derived from the NCC lineage and ADAM12+ Schwann cells, but rather from perivas­cular ADAM12+ cells.

Consistent with this hypothesis, ADAM12+ cells expressed PDGFR­β and NG2, which are also expressed by pericytes (Fig. 6b,c), as well as CD29, CD44 (Fig. 6b), PDGFR­α and Sca­1 (Fig. 1c and Supplementary Fig. 1c), markers commonly used to identify multi­potent mesenchymal progenitor cells in mice40,41. However, con­firming our in  vitro results (Fig.  1e), the progeny of ADAM12+ cells induced in vivo upon injury was restricted to profibrotic cells and did not include other mesenchymal derivatives such as adipo­cytes, even in conditions favoring adipocyte differentiation in vivo (Supplementary Fig. 8c). ADAM12+ cells induced upon injury were not embedded in the vascular basement membrane (the location of pericytes at steady state), in contrast to ADAM12+ cells during post­natal growth (Fig. 6c). These data indicate that ADAM12 expression identifies a profibrotic subset of mesenchymal progenitors within the perivascular space.

DISCUSSIONWe found by inducible fate mapping in vivo that a majority of profi­brotic cells developing upon acute muscle and dermis injury are gen­erated from a distinct subset of PDGFR­α+ proinflammatory stromal cells that are identified by transient expression of ADAM12 upon injury. Accordingly, genetic ablation of ADAM12+ cells decreased tissue fibrosis during scarring. We further found that the profibrotic fate of ADAM12+ cells is irreversible, as interstitial mesenchymal cells repopulating the healed tissue were not derived from ADAM12+ cells. The progeny of ADAM12+ cells was gradually eliminated as the tissue recovered, indicating that healing is associated with elimination of collagen­overproducing cells, rather than reversion to a normal, more quiescent mesenchymal phenotype. This is reminiscent of observa­tions in the liver, in which resolution of fibrosis is associated with apoptosis of hepatic stellate cells, the primary collagen­producing cells in the liver42. Consistent with the proinflammatory nature of ADAM12+ profibrotic progenitors, inflammation generally precedes and contributes to the development of fibrotic tissue, even though accumulation of fibrotic cells often outlasts inflammation3.

On the basis of in vitro differentiation assays, muscle stem cells have been suggested to have an alternative mesenchymal fate and generate fibrogenic cells4,5. Our in vivo lineage tracing data show that ADAM12+ profibrotic progenitors do not contribute to parenchymal regeneration, indicating that they are distinct from muscle and skin stem cells. A detailed analysis shows that injury­induced ADAM12+ cells include terminal Schwann cells capping the neuromuscular junc­tion and PDGFR­β+ perivascular cells, two cell types endowed with high plasticity and roles in repair mechanisms of peripheral nerves and blood vessels, respectively29,43. We further demonstrate that ADAM12+ cells of the mesenchymal lineage, but not of the Schwann cell lineage, are profibrotic progenitors in acute injured skeletal muscle. This finding is in agreement with recent reports showing that the perivascular niche of multiple organs contains pluripotent mesenchymal progenitors, which are suggested to be pericytes11,44. However, the cellular progeny of ADAM12+ cells induced de novo by acute injury in the cardiotoxin and CFA models is restricted to profibrotic cells and does not include other mesenchymal derivatives such as adipocytes, suggesting that distinct subsets of mesenchymal progenitors, or progenitors at different stages of differentiation, are present in adult perivascular niches. In agreement with our findings, PDGFR­β signaling opposes adipocyte differentiation in pericytes while activating immune genes45. In contrast to ADAM12+ cells during postnatal growth, injury­induced ADAM12+ cells are not embedded within the vascular basement membrane, suggesting that they are activated pericytes detaching upon injury or mesenchymal cells in close proximity to vessels. The expression of pericyte mark­ers such as NG2 and PDGFR­β would favor the first hypothesis. The detachment of pericytes from endothelium has been reported dur­ing vascular remodeling occurring in chronic pathologies, such as tumorigenesis, diabetic retinopathy and kidney disease46,47. These results are consistent with recent work in kidney fibrosis and spinal cord injury showing that subsets of pericytes are the major source of myofibroblasts and scar tissue48,49.

We further found that the Adam12 molecule has a role in tissue fibrosis. Notably, transcripts for growth factors with essential roles in muscle regeneration, such as Igf1 and Igf2, were upregulated when Adam12 was silenced in vivo during injury, suggesting that target­ing Adam12 alters stromal growth but favors tissue regeneration. These results are consistent with a report showing that transgenic

NG2 GFP CD31

Pos

tnat

alIn

jury

CollV GFP CD31c

Injury

CX

-mus

cle

CF

A-d

erm

is

Wnt1-cre/YFP

15 d–a

S100 YFP DAPI

CX

-mus

cle

cell

coun

tsC

FA

-der

mis

cell

coun

ts

CD44

CD45–

CD31–

ADAM12–

ADAM12+

CD29 PDGFR-β

b

Figure 6 Mesenchymal perivascular ADAM12+ cells are profibrotic progenitors. (a) Immunofluorescence staining of YFP (progeny of Wnt1+ cells, green) and S100 (Schwann cells, red) at 15 d after injection of CFA in ear dermis (top) and cardiotoxin in tibialis anterior muscle (bottom) in Wnt1-cre mice crossed to Rosa26floxSTOP-YFP mice. Nuclei are stained with DAPI (blue). Scale bars, 100 µm. (b) Flow cytometric analysis of CD29, CD44 and PDGFR-β expression by ADAM12+ and ADAM12− cells isolated from CFA-injured ear dermis (top) and cardiotoxin-injured muscle (bottom) of M12-CIG mice at 72 h after injury. (c) Immunofluorescence staining of ADAM12 (GFP) in postnatal (top, P12) and cardiotoxin-injured (bottom, 24 h after injury) tibialis anterior muscle of M12-CIG mice. The basement membrane (collagen IV in red, white arrows) of microvessels (blue) is at left; pericyte staining with NG2-specific antibodies is at right. Nuclei are stained with DAPI (light gray). Representative pictures of three independent experiments. Scale bar, 25 µm.

npg

© 2

012

Nat

ure

Am

eric

a, In

c. A

ll rig

hts

rese

rved

.

A r t i c l e s

nAture medicine VOLUME 18 | NUMBER 8 | AUGUST 2012 1269

overexpression of Adam12 under control of the muscle creatine kinase promoter both aggravates fibrosis and decreases muscle regeneration after injury, a phenotype further aggravated in the dystrophic back­ground of mdx (X­linked muscular dystrophy) mice50.

ADAMs comprise several different protein modules51, including a catalytic domain, and disintegrin­like cysteine­rich domains which have been implicated in cell­cell interactions. Potential mechanisms underlying the profibrotic role of ADAM12 could therefore depend on the proteolytic release of membrane proteins (known substrates of ADAM12 include EGFR ligands15) or extracellular matrix proteins, or on cell­cell or cell­matrix interactions via its disintegrin domain. ADAM12 also enhances TGF­β–induced transcriptional responses by modulating TGF­β receptor trafficking17, consistent with our data showing that ADAM12 potentiates TGF­β–mediated type I collagen synthesis in fibroblastic cells. As Adam12 is rapidly, though tran­siently, induced by TGF­β in vitro (our data and refs. 19,52), this sug­gests the hypothesis that ADAM12 has a role in the positive feedback loop of TGF­β signaling. Nevertheless, further studies are required to determine whether this mechanism is relevant in vivo during injury. It also remains to be assessed whether ADAM12 has a direct role in the fate choice leading to the generation of α­SMA+ myofibroblasts rather than adipocytes.

Increasing evidence suggests that mesenchymal progenitors resid­ing in adult organs have multiple developmental origins. Accordingly, an initial transient wave of mesenchymal stem cells in the postnatal bone derives from the neuroepithelial and neural crest cells53, whereas all mesenchymal cells of the head derive from the cranial neural crest54. Here we show that fetal ADAM12+ cells programmed during vascular wall development contribute to the generation of mesen­chymal perivascular cells in the adult skeletal muscle. We propose that reactivation of ADAM12+ cells upon injury recapitulates an ontogenic program aimed at restoring vascular integrity, which, in the context of tissue damage and inflammation, leads to an inap­propriate fibrotic outcome. The association of microvascular dam­age, inflammation and fibrosis characterizes several chronic human diseases including scleroderma (systemic sclerosis), a chronic con­nective tissue disease inducing extensive fibrosis of the skin and internal organs. Notably, ADAM12 is overexpressed in fibroblasts isolated from patients with scleroderma21. Future studies manipulat­ing ADAM12+ cells in vivo are needed to assess whether persistence of such cells in scleroderma and other chronic fibrotic diseases, such as liver fibrosis and aggressive fibromatosis19,21,22, has a causal role in the continuous generation of collagen­overproducing cells and fibrosis. Our work emphasizes that elimination versus persistence of functionally distinct subsets of mesenchymal cells might be key to restoring a healthy balance in extracellular matrix production during scarring, with implications for fibrotic diseases. As ADAM12+ cells are rare in adult noninjured organs, we suggest that targeting injury­induced ADAM12+ cells, or inhibiting ADAM12, might represent a new therapeutic approach to specifically limit the generation of collagen­overproducing cells.

METHODSMethods and any associated references are available in the online version of the paper.

Note: Supplementary information is available in the online version of the paper.

ACknowLEDGmEntSThis work was funded by the Institut Pasteur, grants from the Mairie de Paris, the Agence Nationale de la Recherche and an Excellence grant from the European

Commission. S.E.D.C. is funded by La Ligue contre le Cancer. We thank H.­J. Fehling (University Clinics Ulm, Ulm, Germany) for providing the Rosa26floxSTOP­RFP mice, H. Bujard (ZMBH, Heidelberg, Germany) for providing the LC­1 mice and A. Farr (University of Washington, Seattle) for gp38­specific antibody. We thank L. Polomack and the team of the Centre d’Ingénierie Génétique Murine for technical assistance and members of the Development of Lymphoid Tissue Unit for discussions.

AUtHoR ContRIBUtIonSS.D. and S.E.D.C. carried out experiments and analyzed data. F.L. carried out BAC­mice transgenesis. G.E. participated in experiment design and data analysis, and wrote the manuscript. L.P. designed and directed the project, carried out experiments, analyzed data and wrote the manuscript.

ComPEtInG FInAnCIAL IntEREStSThe authors declare no competing financial interests.

Published online at http://www.nature.com/doifinder/10.1038/nm.2848. Reprints and permissions information is available online at http://www.nature.com/reprints/index.html.

1. Collins, C.A. et al. Stem cell function, self-renewal, and behavioral heterogeneity of cells from the adult muscle satellite cell niche. Cell 122, 289–301 (2005).

2. Blanpain, C. & Fuchs, E. Epidermal homeostasis: a balancing act of stem cells in the skin. Nat. Rev. Mol. Cell Biol. 10, 207–217 (2009).

3. Wynn, T.A. Cellular and molecular mechanisms of fibrosis. J. Pathol. 214, 199–210 (2008).

4. Brack, A.S. et al. Increased Wnt signaling during aging alters muscle stem cell fate and increases fibrosis. Science 317, 807–810 (2007).

5. Li, Y. et al. Transforming growth factor-β1 induces the differentiation of myogenic cells into fibrotic cells in injured skeletal muscle: a key event in muscle fibrogenesis. Am. J. Pathol. 164, 1007–1019 (2004).

6. Uezumi, A., Fukada, S., Yamamoto, N., Takeda, S. & Tsuchida, K. Mesenchymal progenitors distinct from satellite cells contribute to ectopic fat cell formation in skeletal muscle. Nat. Cell Biol. 12, 143–152 (2010).

7. Joe, A.W. et al. Muscle injury activates resident fibro/adipogenic progenitors that facilitate myogenesis. Nat. Cell Biol. 12, 153–163 (2010).

8. Olson, L.E. & Soriano, P. Increased PDGFRα activation disrupts connective tissue development and drives systemic fibrosis. Dev. Cell 16, 303–313 (2009).

9. Herzog, E.L. & Bucala, R. Fibrocytes in health and disease. Exp. Hematol. 38, 548–556 (2010).

10. Hinz, B. et al. The myofibroblast: one function, multiple origins. Am. J. Pathol. 170, 1807–1816 (2007).

11. Crisan, M. et al. A perivascular origin for mesenchymal stem cells in multiple human organs. Cell Stem Cell 3, 301–313 (2008).

12. Real, C., Glavieux-Pardanaud, C., Vaigot, P., Le-Douarin, N. & Dupin, E. The instability of the neural crest phenotypes: Schwann cells can differentiate into myofibroblasts. Int. J. Dev. Biol. 49, 151–159 (2005).

13. Thiery, J.P., Acloque, H., Huang, R.Y. & Nieto, M.A. Epithelial-mesenchymal transitions in development and disease. Cell 139, 871–890 (2009).

14. Joyner, A.L. & Zervas, M. Genetic inducible fate mapping in mouse: establishing genetic lineages and defining genetic neuroanatomy in the nervous system. Dev. Dyn. 235, 2376–2385 (2006).

15. Horiuchi, K. et al. Substrate selectivity of epidermal growth factor-receptor ligand sheddases and their regulation by phorbol esters and calcium influx. Mol. Biol. Cell 18, 176–188 (2007).

16. Shi, Z., Xu, W., Loechel, F., Wewer, U.M. & Murphy, L.J. ADAM 12, a disintegrin metalloprotease, interacts with insulin-like growth factor–binding protein-3. J. Biol. Chem. 275, 18574–18580 (2000).

17. Atfi, A. et al. The disintegrin and metalloproteinase ADAM12 contributes to TGF-β signaling through interaction with the type II receptor. J. Cell Biol. 178, 201–208 (2007).

18. Kurisaki, T., Masuda, A., Osumi, N., Nabeshima, Y. & Fujisawa-Sehara, A. Spatially- and temporally-restricted expression of meltrin α (ADAM12) and β (ADAM19) in mouse embryo. Mech. Dev. 73, 211–215 (1998).

19. Le Pabic, H. et al. ADAM12 in human liver cancers: TGF-β–regulated expression in stellate cells is associated with matrix remodeling. Hepatology 37, 1056–1066 (2003).

20. Borneman, A., Kuschel, R. & Fujisawa-Sehara, A. Analysis for transcript expression of meltrin α in normal, regenerating, and denervated rat muscle. J. Muscle Res. Cell Motil. 21, 475–480 (2000).

21. Shi-Wen, X. et al. Endogenous endothelin-1 signaling contributes to type I collagen and CCN2 overexpression in fibrotic fibroblasts. Matrix Biol. 26, 625–632 (2007).

22. Skubitz, K.M. & Skubitz, A.P. Gene expression in aggressive fibromatosis. J. Lab. Clin. Med. 143, 89–98 (2004).

23. Peduto, L. et al. ADAM12 is highly expressed in carcinoma-associated stroma and is required for mouse prostate tumor progression. Oncogene 25, 5462–5466 (2006).

npg

© 2

012

Nat

ure

Am

eric

a, In

c. A

ll rig

hts

rese

rved

.

A r t i c l e s

1270 VOLUME 18 | NUMBER 8 | AUGUST 2012 nAture medicine

24. Peduto, L. et al. Inflammation recapitulates the ontogeny of lymphoid stromal cells. J. Immunol. 182, 5789–5799 (2009).

25. Harris, J.B. Myotoxic phospholipases A2 and the regeneration of skeletal muscles. Toxicon 42, 933–945 (2003).

26. Jessen, K.R. & Mirsky, R. Negative regulation of myelination: relevance for development, injury, and demyelinating disease. Glia 56, 1552–1565 (2008).

27. Hernandez-Gea, V. & Friedman, S.L. Pathogenesis of liver fibrosis. Annu. Rev. Pathol. 6, 425–456 (2011).

28. Passino, M.A., Adams, R.A., Sikorski, S.L. & Akassoglou, K. Regulation of hepatic stellate cell differentiation by the neurotrophin receptor p75NTR. Science 315, 1853–1856 (2007).

29. Gaengel, K., Genove, G., Armulik, A. & Betsholtz, C. Endothelial-mural cell signaling in vascular development and angiogenesis. Arterioscler. Thromb. Vasc. Biol. 29, 630–638 (2009).

30. Luche, H., Weber, O., Nageswara Rao, T., Blum, C. & Fehling, H.J. Faithful activation of an extra-bright red fluorescent protein in “knock-in” Cre-reporter mice ideally suited for lineage tracing studies. Eur. J. Immunol. 37, 43–53 (2007).

31. Schönig, K., Schwenk, F., Rajewsky, K. & Bujard, H. Stringent doxycycline dependent control of CRE recombinase in vivo. Nucleic Acids Res. 30, e134 (2002).

32. Srinivas, S. et al. Cre reporter strains produced by targeted insertion of EYFP and ECFP into the ROSA26 locus. BMC Dev. Biol. 1, 4 (2001).

33. Le Douarin, N.M., Calloni, G.W. & Dupin, E. The stem cells of the neural crest. Cell Cycle 7, 1013–1019 (2008).

34. Verrecchia, F. & Mauviel, A. Transforming growth factor-β and fibrosis. World J. Gastroenterol. 13, 3056–3062 (2007).

35. Betancur, P., Bronner-Fraser, M. & Sauka-Spengler, T. Assembling neural crest regulatory circuits into a gene regulatory network. Annu. Rev. Cell Dev. Biol. 26, 581–603 (2010).

36. Jessen, K.R. & Mirsky, R. The origin and development of glial cells in peripheral nerves. Nat. Rev. Neurosci. 6, 671–682 (2005).

37. Buchstaller, J. et al. Efficient isolation and gene expression profiling of small numbers of neural crest stem cells and developing Schwann cells. J. Neurosci. 24, 2357–2365 (2004).

38. Joseph, N.M. et al. Neural crest stem cells undergo multilineage differentiation in developing peripheral nerves to generate endoneurial fibroblasts in addition to Schwann cells. Development 131, 5599–5612 (2004).

39. Danielian, P.S., Muccino, D., Rowitch, D.H., Michael, S.K. & McMahon, A.P. Modification of gene activity in mouse embryos in utero by a tamoxifen-inducible form of Cre recombinase. Curr. Biol. 8, 1323–1326 (1998).

40. Pittenger, M.F. et al. Multilineage potential of adult human mesenchymal stem cells. Science 284, 143–147 (1999).

41. Morikawa, S. et al. Prospective identification, isolation, and systemic transplantation of multipotent mesenchymal stem cells in murine bone marrow. J. Exp. Med. 206, 2483–2496 (2009).

42. Iredale, J.P. Models of liver fibrosis: exploring the dynamic nature of inflammation and repair in a solid organ. J. Clin. Invest. 117, 539–548 (2007).

43. Griffin, J.W. & Thompson, W.J. Biology and pathology of nonmyelinating Schwann cells. Glia 56, 1518–1531 (2008).

44. da Silva Meirelles, L., Caplan, A.I. & Nardi, N.B. In search of the in vivo identity of mesenchymal stem cells. Stem Cells 26, 2287–2299 (2008).

45. Olson, L.E. & Soriano, P. PDGFRβ signaling regulates mural cell plasticity and inhibits fat development. Dev. Cell 20, 815–826 (2011).

46. Chung, A.S., Lee, J. & Ferrara, N. Targeting the tumour vasculature: insights from physiological angiogenesis. Nat. Rev. Cancer 10, 505–514 (2010).

47. Lin, S.L., Kisseleva, T., Brenner, D.A. & Duffield, J.S. Pericytes and perivascular fibroblasts are the primary source of collagen-producing cells in obstructive fibrosis of the kidney. Am. J. Pathol. 173, 1617–1627 (2008).

48. Humphreys, B.D. et al. Fate tracing reveals the pericyte and not epithelial origin of myofibroblasts in kidney fibrosis. Am. J. Pathol. 176, 85–97 (2010).

49. Göritz, C. et al. A pericyte origin of spinal cord scar tissue. Science 333, 238–242 (2011).

50. Jørgensen, L.H., Jensen, C.H., Wewer, U.M. & Schroder, H.D. Transgenic overexpression of ADAM12 suppresses muscle regeneration and aggravates dystrophy in aged mdx mice. Am. J. Pathol. 171, 1599–1607 (2007).

51. Blobel, C.P. ADAMs: key components in EGFR signalling and development. Nat. Rev. Mol. Cell Biol. 6, 32–43 (2005).

52. Solomon, E., Li, H., Muggy, S.D., Syta, E. & Zolkiewska, A. The role of SnoN in transforming growth factor β1-induced expression of metalloprotease-disintegrin ADAM12. J. Biol. Chem. 285, 21969–21977 (2010).

53. Takashima, Y. et al. Neuroepithelial cells supply an initial transient wave of MSC differentiation. Cell 129, 1377–1388 (2007).

54. Le Douarin, N.M., Brito, J.M. & Creuzet, S. Role of the neural crest in face and brain development. Brain Res. Rev. 55, 237–247 (2007).

npg

© 2

012

Nat

ure

Am

eric

a, In

c. A

ll rig

hts

rese

rved

.

nAture medicinedoi:10.1038/nm.2848

ONLINE METHODSMice. To generate BAC­transgenic Adam12­cre-ires-egfpTG mice (M12­CIG), we generated the coding sequence for cre-ires-egfp, including a poly(A) sequence after egfp, by overlap PCR and inserted into exon 1 of Adam12 in place of the endogenous ATG translation start codon, on a 200­kb BAC (Invitrogen) carrying 80 kb of sequence upstream of the Adam12 translation start site. The transgene insertion was done by homologous recombination as described55. M12­CIG mice were crossed with Rosa26floxSTOP­RFP reporter mice30 for fate­mapping experiments. Parabiosis between M12­CIG and nontransgenic littermates was done as described24. To generate BAC­transgenic Adam12-tTA-ires-DtrTG mice (M12­tTA), we generated by overlap PCR the coding sequence for tTA-ires-Dtr (human), including a poly(A) sequence after Dtr, and inserted into exon 1 of Adam12 in place of the endogenous ATG translation start codon, on the same 200­kb BAC as described above. We carried out inducible fate mapping experi­ments in triple­transgenic mice M12tTA/LC1/YFP obtained by crossing M12­tTA mice with tet­inducible LC1 mice31, and then further with Rosa26floxSTOP­YFP mice32. In both BAC­transgenic models, GFP and YFP expression was consist­ent with reported ADAM12 expression data during development (cranial and trunk NCCs and SCPs18,37) and tumorigenesis (prostate tumor in the TRAMP model23). We obtained Rosa26floxSTOP­YFP mice and Wnt1­cre mice from Jackson Laboratory. All mice were kept in specific pathogen­free conditions, and mouse experiments were approved by the committee on animal experimentation of the Institut Pasteur and by the French Ministry of Agriculture.

Mice treatment. We injected anesthetized mice (xylazine­ketamine) with 25 µl CFA (Sigma) in the ear dermis or with 50 µl of 10 µM cardiotoxin (Latoxan) in tibialis anterior muscles. At different time points after treatment, we processed CFA­injected ears and cardiotoxin­injected muscles for histology, FACS analysis or RNA extraction and qRT­PCR analysis. To induce fatty muscle degeneration, we injected 50 µl glycerol (50% (v/v)) in tibialis anterior muscles, as glycerol induces destabilization of the cytoplasmic membrane followed by cell death and fatty degeneration of the injected muscle56. To stop fate­mapping labeling of ADAM12+ cells, we treated M12tTA/LCI/YFP triple­transgenic mice with doxycycline (Sigma­Aldrich) at 1 mg ml−1 in drinking water containing 5% sucrose. To delete ADAM12+ cells, we injected M12tTA/LCI/YFP mice intra­peritoneally (i.p.) with 100 ng of DTA every day after cardiotoxin injection for 5 d, and then every other day until analysis. This strategy is based on the fact that mouse cells, unlike cells carrying a primate diphtheria toxin receptor, are insensitive to killing by DTA57.

Cells isolation and FACS. To isolate stromal cells, we processed 1­mm pieces of injured tissues at 37 °C for 30 min in a digestion solution composed of DMEM (Gibco), 1 mg ml−1 collagenase D (Roche), blendzyme III (Roche) and 1 U ml−1 DNase 1 (Invitrogen). After 30 min, we collected and washed the digested frac­tion of the mix in DMEM 10% FCS. We carried out one or two additional cycles of digestion for the remaining tissue, collected and filtered through a 100 µm mesh. For FACS staining, we first incubated cells with monoclonal antibody 2.4G2 to block Fcγ receptors, and then with the indicated antibodies for 40 min in a total volume of 100 µl of PBS containing 2 mM EDTA and 2% bovine serum (PBS­F), followed by appropriate secondary antibodies for 30 min when neces­sary, centrifuged in 2 ml PBS­F and dissolved in 200 µl of PBS­F for analysis. Cells were incubated for 1 min with DAPI (Sigma) before analysis to exclude dead cells. Cell doublets were systematically excluded during analysis. Cells were analyzed with Fortessa (BD Biosciences) or Cyan ADP (Beckman Coulter), and Flowjo software (Tristar). Fluorescence intensity is expressed in arbitrary units on a logarithmic scale or on a linear scale for forward scatter. Cells were sorted with FACS Aria 3 (BD Biosciences) to 95–98% purity.

Cell proliferation. To quantify cell proliferation in vivo, we injected 50 µl of 10 mM 5­ethynyl­2′­deoxyuridine (EdU; Invitrogen) solution i.p. into cardio­toxin or CFA­treated M12­CIG and M12cre/RFP mice 1 d before analysis. Incorporation of EdU into cells was detected by histology with the Click­iT EdU Assay Kit­Alexa Fluor 647 (Invitrogen), according to the manufacturer’s instruc­tions. We stained sections with GFP­specific antibody (A11122, 1:1,000) and Alexa Fluor 488 anti­rabbit antibody (Invitrogen) to enhance GFP visualization of paraformaldehyde­fixed cells.

Knockdown of ADAM12 by RNA interference. To knock down Adam12 in vivo, we injected cardiotoxin­injured muscles with stealth­RNAi siRNAs targeting mouse Adam12 (Invitrogen). Briefly, we mixed in vivo–quality purified siRNAs with Invivofectamine (Invitrogen) to obtain a final concentration of 20 mg ml−1 siRNA duplex. We diluted the lipid complex with 15 ml of 5% glucose and centri­fuged for 1 h in Amicon Ultra­15 centrifugal device with Ultracel­50 membrane (Millipore). After collecting and filtering the Invivofectamine–siRNA complex through a 0.22­µm filter, we injected 15 µl of 20 mg ml−1 siRNA solution in tibialis anterior muscles at 2 d, 7 d, 12 d and 17 d after cardiotoxin injection. As control, we injected scrambled siRNAs duplex with similar G+C content in the contralateral tibialis anterior muscle. We analyzed muscle tissue for fibrosis (in picrosirius­stained sections) and gene expression (by qRT­PCR) 3 weeks after cardiotoxin injection. We calculated the percentage of gene expression as the transcript expression in muscles injected with siRNA Adam12 compared with the expression obtained in the controlateral muscle injected with siRNA control. Graphs representing gene expression after ADAM12 silencing were generated and analyzed using the analysis of variance test.

Antibodies. Antibodies to PDGFR­β (APB5), to CD45.2 (104), to PDGFR­α (APA5), to CD29 (HMb1­1), to CD31 (390) and to CD44 (IM7) were purchased from eBiosciences and used at 1:100 dilution. Cy3­conjugated and FITC­ conjugated α­SMA­specific antibody (1A4, 1:500) and streptavidins were pur­chased from Sigma. Purified rabbit GFP­specific antibody (A11122), Alexa Fluor 555–conjugated α­bungarotoxin (B35451), Alexa Fluor 488–conjugated anti­rabbit IgG (H+L) and anti­chicken IgG (H+L), Alexa Fluor 555– conjugated anti­rat IgG (H+L) and Alexa Fluor 647–conjugated anti­syrian hamster were purchased from Invitrogen, and used at 1:1,000. Rabbit polyclonal antibody to p75 (G3231; 1:1,000) was purchased from Promega. Rabbit polyclo­nal antibody to PGP9.5 (7863­0504), type I collagen (2150­1410) and type IV collagen (2150­1470) were purchased from Serotec, and used at 1:1,000. Rabbit polyclonal antibody to S100b (Z0311; 1:1,000) was purchased from Dako. GFP­specific polyclonal chicken antibody (ab13970; 1:1,500) was purchased from Abcam. Cy3­conjugated anti­syrian hamster was purchased from Jackson Immunoresearch. Syrian hamster antibody to gp38 (1:50 dilution) was a gift from A. Farr (University of Washington, Seattle).

Histology. Tissue processing and staining procedures for immunofluorescence have been described24. Briefly, tissues were fixed overnight at 4 °C in 4% para­formaldehyde (Sigma), washed overnight in PBS, incubated in a solution of 30% sucrose (Sigma) until the samples sank, embedded in OCT compound 4583 (Sakura Finetek), frozen in a bath of isopentane cooled with liquid nitrogen and stored at −80 °C. We processed 8­µm­thick sections for antibody staining: after blocking with 10% bovine serum in PBS containing 1% Triton (PBS­XG) for 1 h at room temperature, we incubated slides with primary antibodies in PBS­XG overnight at 4 °C, washed three times for 5 min with PBS­XG, incu­bated with secondary antibodies or streptavidin for 1 h at room temperature, washed once, incubated with DAPI (Sigma) for 5 min, washed three times for 5 min and mounted with Fluoromount­G (Southern Biotechnology Associates). We examined slides with an AxioImager M1 fluorescence microscope (Zeiss) equipped with a CCD camera and processed images with AxioVision software (Zeiss). For picrosirius staining (stains collagen), frozen sections were fixed in chilled acetone for 10 min, hydrated through grades of alcohol, stained with a solution of picrosirius red (Sigma) for 15 min, rinsed with 0.5% acetic acid, dehydrated, cleared and mounted with Histomount (Zymed). For quantifica­tion of fibrosis, we scored picrosirius­stained muscle sections as follows: 0 = no staining; 1 = myofibers surrounded by a noncontinuous staining; 2 = myofibers surrounded by a thin but continuous staining; 3 = myofibers surrounded by a thick continuous staining; 4 = thick endomysium staining in continuity with perimysium; 5 = intense staining replacing myofibers.

RNA  isolation  and  quantitative  RT-PCR. We FACS­sorted cells into vials containing RLT buffer (Qiagen) supplemented with β­mercaptoethanol, and extracted total RNA using RNeasy Micro Kit (Qiagen). We assessed the quality of total RNA using the 2100 Bioanalyzer system (Agilent Technologies), and used 250–500 pg high­quality total RNA for one linear mRNA amplification cycle using the MessageBooster Kit for qRT­PCR (Epicentre Biotechnologies).

npg

© 2

012

Nat

ure

Am

eric

a, In

c. A

ll rig

hts

rese

rved

.

nAture medicine doi:10.1038/nm.2848

Amplified mRNA (50–100 ng) was transcribed into cDNA using Superscript III reverse transcriptase (Invitrogen). To purify total muscle RNA, we extracted total RNA from 2 mm2 frozen tissue using RNeasy Mini kit (Qiagen). All procedures were done according to the manufacturer’s protocols. We carried out qRT­PCR using RT2 qPCR primer sets (SABiosciences) and RT2 SYBR­Green master mix (SABiosciences) on a PTC­200 thermocycler equipped with a Chromo4 detector (Bio­Rad Laboratories), and analyzed data using Opticon Monitor soft­ware (Bio­Rad Laboratories). We normalized Ct values to the mean Ct values obtained for the housekeeping genes Hsp90 and Gapdh. Heatmaps were made using JavaTreeView version 1.1.3. In the heatmap, data are further normalized and centered horizontally so that the mean of increases (red) and decreases (blue) equals 1.

In vitro  cell  differentiation. ADAM12+ and ADAM12− cells were FACS­sorted from cardiotoxin­injured muscle of M12­CIG mice at 4 d after cardio­toxin injection. Cell differentiation was assessed by growing cells in conditions favoring myofibroblast differentiation (DMEM 2% horse serum supplemented with 2 ng ml−1 TGF­β), or adipogenic differentiation (StemXVivo mouse MSC adipogenesis media, R&D Systems). After 5 d, cells were washed, fixed

in 4% paraformaldehyde and stained for α­SMA (myofibroblasts) or with 0.5% Oil Red O (fat droplets in adipocytes; Sigma).

Cell transfections. For overexpression of ADAM12 in vitro, we transfected an expression vector containing mouse Adam12 cDNA, or the empty vector, into NIH3T3 cells (ATCC) using Lipofectamine 2000 according to the manufacturer’s protocol (Invitrogen). To knock down Adam12 in vitro, we transfected stealth siRNAs targeting mouse Adam12 (Invitrogen), or nontargeting stealth siRNAs with similar G+C content as control, in NIH3T3 using lipofectamine RNAiMAX according to manfacturer’s instructions (Invitrogen). In both cases, we added TGF­β (2 ng ml−1) after 2 d for 40 h and analyzed gene expression by qRT­PCR as described above.

55. Sparwasser, T., Gong, S., Li, J.Y.H. & Eberl, G. General method for the modification of different BAC types and the rapid generation of BAC transgenic mice. Genesis 38, 39–50 (2004).

56. Arsic, N. et al. Vascular endothelial growth factor stimulates skeletal muscle regeneration in vivo. Mol. Ther. 10, 844–854 (2004).

57. Pappenheimer, A.M. Jr., Harper, A.A., Moynihan, M. & Brockes, J.P. Diphtheria toxin and related proteins: effect of route of injection on toxicity and the determination of cytotoxicity for various cultured cells. J. Infect. Dis. 145, 94–102 (1982).

npg

© 2

012

Nat

ure

Am

eric

a, In

c. A

ll rig

hts

rese

rved

.