functions of rhomboid family protease rhbdl2 and thrombomodulin in wound healing

Functions of Rhomboid Family Protease RHBDL2 andThrombomodulin in Wound HealingTsung-Lin Cheng1,2, Yu-Ting Wu1, Hung-Yu Lin1, Fu-Chih Hsu2, Shi-Kai Liu1, Bi-Ing Chang2,Wei-Sheng Chen1, Chao-Han Lai2, Guey-Yueh Shi1,2 and Hua-Lin Wu1,2

The expression of thrombomodulin (TM), a membrane glycoprotein, is upregulated in neoepidermis duringcutaneous wound healing. Rhomboid-like-2 (RHBDL2), an intramembrane serine protease, specifically cleavesTM at the transmembrane domain and causes the release of soluble TM (sTM). However, the physiologicalfunctions of TM and RHBDL2 in wound healing remain unclear. We demonstrated that both TM and RHBDL2 areupregulated in HaCaT cells stimulated by scratch wounds; furthermore, increased sTM was found in culturemedium. Conversely, inhibition of RHBDL2 by 3,4-dichloroisocoumarin (DCI) or short hairpin RNA significantlyinhibited wound-induced TM ectodomain shedding and wound healing. Both conditioned media frommultiple-scratch-wounded HaCaT and recombinant sTM accelerated wound healing in HaCaT cells; such effectswere abrogated by anti-TM antibodies. The RNA released from injured cells is involved in the induction ofTM and RHBDL2. RHBDL2 and sTM were upregulated in ex vivo tissue culture of the injured skin. Furthermore,DCI inhibited sTM production and wound healing; this was reversed by recombinant sTM in mice. Thus,RHBDL2 and TM have important roles in wound healing via the release of sTM from keratinocytes; this mayfunction as an autocrine/paracrine signal promoting wound healing.

Journal of Investigative Dermatology (2011) 131, 2486–2494; doi:10.1038/jid.2011.230; published online 11 August 2011

INTRODUCTIONThrombomodulin (TM) is a type-I transmembrane glycoproteinthat is expressed on the surfaces of endothelial cells andepidermal keratinocytes (Esmon, 1995; Lager et al., 1995; Weilerand Isermann, 2003). It is composed of five distinct domains: anNH2-terminal lectin-like region (TM domain 1, TMD1), anepidermal growth factor-like domain (TMD2) consisting of sixepidermal growth factor-like repeats, an O-glycosylation site-rich domain (TMD3), a transmembrane domain (TMD4), and acytoplasmic tail (TMD5) (Figure 1a). The various biologicalfunctions of TM domains have been demonstrated in differentphysiological and pathological processes (Esmon, 1995; Healyet al., 1995; Zhang et al., 1998; Huang et al., 2003; Shi et al.,2005, 2008). Until now, the function of TM in epidermaldifferentiation and wound healing was largely unknown,although the expression of TM is tightly regulated duringkeratinocyte differentiation (Mizutani et al., 1994; Raife et al.,1994, 1996, 1998; Lager et al., 1995; Senet et al., 1997).

Soluble TM (sTM) in plasma and urine is found in healthysubjects (Ishii and Majerus, 1985). It is proposed that variousproteases are responsible for the shedding and release of sTMfrom the cell membrane, and that elevated plasma sTM servesas a marker of endothelial damage (Blann and Lip, 1998). Itwas recently demonstrated that TM is the specific trans-membrane protein substrate of rhomboid-like-2 (RHBDL2), amember of rhomboid serine protease family, in mammaliancells (Lohi et al., 2004). Previous report indicates that theexpression of TM in keratinocytes is upregulated duringcutaneous wound healing (Peterson et al., 1999). However,overall functions of the TM and RHBDL2 system in cutaneouswound healing remain unknown.

As recombinant TM is a potent mitogenic factor (Shi et al.,2005), we hypothesize that TM is cleaved by RHBDL2 at itstransmembrane domain and released from the neoepidermisinto the wound matrix. Consequently, sTM may then functionas a mitogenic factor in the healing process. To test thishypothesis, we measured the induction of TM and RHBDL2expression in injured keratinocytes in vitro and in the marginof wound openings in mice. Effects of the specific knock-down of RHBDL2 by RNA interference and an inhibitor onwound healing were investigated. A mechanism for TM andRHBDL2 induction in wound healing is also proposed.

RESULTSRHBDL2 mediates the ectodomain shedding of TM fromkeratinocytes

Whether the shedding of TM ectodomain occurred in thekeratinocyte remains unknown. First, we detected the TM

2486 Journal of Investigative Dermatology (2011), Volume 131 & 2011 The Society for Investigative Dermatology

ORIGINAL ARTICLE

Received 18 December 2011; revised 2 June 2011; accepted 16 June 2011;published online 11 August 2011

1Department of Biochemistry and Molecular Biology, College of Medicine,National Cheng Kung University, Tainan, Taiwan, ROC and 2CardiovascularResearch Center, National Cheng Kung University, Tainan, Taiwan, ROC

Correspondence: Hua-Lin Wu or Guey-Yueh Shi, Department ofBiochemistry and Molecular Biology, College of Medicine, National ChengKung University, Tainan 701, ROC. E-mail: [email protected] [email protected]

Abbreviations: CM, conditioned medium; DCI, 3,4-dichloroisocoumarin;GFP, green fluorescent protein; RHBDL2, rhomboid-like-2; shRNA, shorthairpin RNA; sTM, soluble TM; TM, thrombomodulin; TMD, TM domain

http://dx.doi.org/10.1038/jid.2011.230

mailto:[email protected]

mailto:[email protected]

domains of keratinocytes in the conditioned medium (CM).Results showed that one major band of sTM that had amolecular mass of B90 kDa, similar to the recombinantectodomain of TM (rTMD123), was detected in the CM.The intact TM had a molecular mass 495 kDa (Figure 1d).The molecular mass of rTMD23 was B64 kDa (Figure 1band c). Our result indicated that sTM might represent theectodomain of intact TM. The amount of the sTM found in theCM increased in a time-dependent manner (Figure 1d).Moreover, rhomboid serine protease inhibitor (3,4-dichlor-oisocoumarin, DCI) inhibited the production of sTM in adose-dependent manner (Figure 1e). The amount of sTMwas significantly decreased by the RHBDL2 short hairpin(sh)RNA (shRHBDL2) transfection and was increased by

RHBDL2-green fluorescent protein (GFP) overexpression(Figure 1f). Overexpression of RHBDL2-GFP could restorethe release of sTM under conditions of DCI treatment (5 mM;Supplementary Figure S1 online). The results imply thatRHBDL2 is involved in the shedding of TM ectodomain inthe keratinocytes.

RHBDL2 expression and sTM production are increased in thecell culture after wounding

RHBDL2 expression and TM shedding in keratinocytes afterdamage by scratch wounding have never been studiedbefore. The confluent HaCaT were subjected to differentnumbers of scratch wounds, and samples of CM and celllysates were harvested 1 day after injury. Western blot

Thrombomodulin TMD5

(kDa)

130957255

43

33

(kDa)

130957255

43

33

Mar

ker

rTM

D23

rTM

D123

rTM

D23

rTM

D123

TMD4

TMD3TMD2

1 2 3 4 5 6

1 2 3 4 5 6

1 2 3 4 5 6

100 aa

sTM

GST

**

******

***

12010080604020

00 1.25 2.5

DCI (μM)

sTMGST

800 *** sTM

*

600

400

200

0

HaCaT

GFP

RHBDL2-G

FP

shRHBDL2

Rel

ativ

e in

tens

ity(%

)

5 10 20Rel

ativ

e in

tens

ity (

%)

(kDa)1309572

TM in

lysa

te

rTM

D123

Incubation time (hours)

0 4 8 12 24

sTM

GST

CM of HaCat120

100

80

60

40

20

00 4 8 12 24

Time (hours)

Rel

ativ

e in

tens

ity (

%)

TMD1

rTMD123

rTMD23

Figure 1. Thrombomodulin (TM) is cleaved by rhomboid-like-2 (RHBDL2) to release soluble TM (sTM) from keratinocytes. (a) Illustration of the protein

domain structure of TM and recombinant TM (rTM). Arrow indicates the predicted cutting site of RHBDL2. Dotted line indicates the antibody region against

TM (D-3). Purified rTMD23 and rTMD123 were analyzed by Coomassie blue staining (b) and western blotting (c). (d) Western blot analysis of TM in HaCaT

lysates, purified rTMD123 (20 ng), and sTM in the serum-free conditioned medium (CM) collected at the indicated time points. (e) Western blotting and

quantitative representation of the sTM after treatment with 3,4-dichloroisocoumarin (DCI) for 24 hours. (f) Western blotting and quantitative representation of the

sTM from HaCaT and stable transfected cells (green fluorescent protein (GFP), RHBDL2-GFP, and shRHBDL2). *Po0.05, **Po0.01, ***Po0.001.

aa, amino acid; GST, glutathione S-transferase.

www.jidonline.org 2487

T-L Cheng et al.RHBDL2-Released sTM Mediates Wound Healing

http://www.jidonline.org

analysis revealed a positive correlation among the numberof scratch wounds and expression of RHBDL2 and sTM(Figure 2a). After the cells were subjected to 5 or 10 lines ofscratch wounds across the culture dish, they demonstrated anapproximately 2-fold increase in both RHBDL2 and sTMcompared with the unwounded control (Figure 2b).

The process by which scratch wounds are created in cellcultures may lead to a significant amount of cell death, aswell as the release of cellular contents into the medium.These contents may stimulate TM and RHBDL2 expression innon-damaged cells. To test this hypothesis, UV-irradiatedkeratinocytes (UVR cells) were used to evaluate the effects ofthe substances released from dead cells. Results show that

the TM and RHBDL2 protein expression levels in normalHaCaT were increased when stimulated by coculturing withUVR cells. The effects of UVR cells were abolished whenpretreated with RNase but not DNase (Figure 2c and e). Theupregulation of TM and RHBDL2 by scratch wounds was alsosuppressed by incubation with RNase (Figure 2d and f). Theresults imply that UVR cells and scratch wounds can induceTM and RHBDL2 expression via similar regulatory mechan-isms; in addition, RNA release from damaged cells may havea pivotal role in the process. However, UVR cells and scratchwounds did not induce significant changes in the expressionof other RHBDL2 substrates, ephrin B2 and B3 (Figure 2c–f).

RHBDL2 is involved in wound-induced TM shedding fromkeratinocytes

We examined whether RHBDL2 participates in the modula-tion of TM in response to wound stimulation. The releaseof sTM from keratinocytes promoted by scratch woundswas inhibited by DCI, as shown by western blot analysis(Figure 3a). Transient transfected RHBDL2 shRNA signifi-cantly reduced RHBDL2 expression and sTM production(Figure 3b). These results indicate that the inhibition ofrhomboid activity or the knockdowns of RHBDL2 can down-regulate the release of sTM from scratch-wounded HaCaT.The results suggest that wound-stimulated upregulation ofsTM production depends on RHBDL2 activity in the cellculture.

Number of scratch wounds

0

RHBDL2

β-Actin

Lys

300 250 sTM

****** ***200

150

100

50

00 1 2 5 10

Rel

ativ

e in

tens

ity (

%)

Rel

ativ

e in

tens

ity (

%)

Rel

ativ

e in

tens

ity (

%)

RHBDL2 ** **250

200

150

50

00

Number of scratch wounds Number of scratch wounds

UVR cells – + ++

+

+––

–– –

–DNaseRNase

UVR cells – + ++

+ – + + +

+––

–– –

–DNaseRNase

Wound+

+––

–– –

–DNaseRNase

Wound – + ++

+

+––

–– –

–DNaseRNase

TM

RHBDL2

Ephrin B2

Ephrin B3

β-Actin

TM(kDa)

95

34

3455

43

RHBDL2

Ephrin B2

Ephrin B3

250

200

150

100

50

0

250 TMEphrinB2

***

******

****

RHBDL2EphrinB3

TMEphrinB2

RHBDL2EphrinB3

200

150

100

50

0

β-Actin

1 2 5 10

Rel

ativ

e in

tens

ity (

%)

100

1 2 5 10 (kDa)

sTM

GST

Med

43

34

43

Number of scratch wounds

0 1 2 5 10 (kDa)

95

26

Figure 2. Injury-induced rhomboid-like-2 (RHBDL2) upregulation and

thrombomodulin (TM) ectodomain shedding in HaCaT cells were inhibited

by RNase. (a) Western blot analysis of RHBDL2 expression in cell lysates (left)

and soluble TM (sTM, right) in the conditioned medium of HaCaT after

scratch wounding. (b) Quantitative representation of the results of a. Western

blot analysis of normal HaCaT cocultured with UV-irradiated keratinocytes

(UVR cells) (c) and scratch-wounded cells (d) that were pretreated with

DNase (4 U ml�1) or RNase (20 mg ml�1). (e, f) Quantitative representation of

the results of c and d. The number of scratch wounds is five. Data are from

three independent experiments. *Po0.05, **Po0.01, ***Po0.001 versus

untreated controls. GST, glutathione S-transferase; Lys, lysate; Med, medium.

Scratch –– –

–++

Controlvector

Scratch

sTM

GST

RHBDL2

– + +–

RHBDL2shRNA

Controlvector

RHBDL2shRNA

++

DCI (5 μM)

Scratch Scratch–

– –

–

+

+ – + – +

+

+

DCI (5 μM)

sTM

GST

Med

Lys

Med

Lys

sTM

RHBDL2

RHBDL2

250160

140

120

100

80

60

40

20

0

***

***

***

200

150

100

50

0

Rel

ativ

e in

tens

ity (

%)

Rel

ativ

e in

tens

ity (

%)

β-Actin β-Actin

Figure 3. Scratch-wound-induced thrombomodulin (TM) ectodomain

shedding in HaCaT cells inhibited by serine protease inhibitor

3,4-dichloroisocoumarin (DCI) and rhomboid-like-2 (RHBDL2) short

hairpin RNA (shRNA). (a) The soluble TM (sTM) in concentrated conditioned

medium and RHBDL2 in cell lysates of HaCaT cells after scratch wounding

and treatment with DCI (left panel) or transient transfection with RHBDL2

shRNA (right panel) were analyzed by western blotting. (b) Quantitative

representation of the results of a. The number of scratches is five. Data are

from three independent experiments. ***Po0.001. For all western blot

analyses, one representative experiment of three is shown. GST, glutathione

S-transferase; Lys, lysate; Med, medium.

2488 Journal of Investigative Dermatology (2011), Volume 131


RHBDL2 is required for wound recovery, cell migration, andproliferation

The stable RHBDL2-knockdown cell line was established toevaluate the physiological effects of RHBDL2 on woundrecovery, cell migration, and proliferation. Results show thatRHBDL2 was significantly silenced in shRHBDL2-transfectedcells, and sTM becomes undetectable (Figure 4a). The rate ofwound healing was measured using a time-lapse recordingsystem when cells were cultured in complete medium. Theresults demonstrate that the control cells transfected with GFPwere almost completely healed within 24 hours. However,the shRHBDL2 cells were only about 50% recovered at24 hours (Figure 4b and c). Cell migration and proliferationare the two important factors that contribute to the rate ofwound healing. The results shown by the transwell migrationassay indicated that shRHBDL2 cells migrated slower thanHaCaT and GFP cells within 24 hours (Figure 4d). Moreover,the cellular proliferation of shRHBDL2 cells was significantlyinhibited after 2 days of culture (Figure 4e). Thus, RHBDL2has a significant impact on cell migration and proliferation,and therefore is involved in the wound-healing process inkeratinocytes.

sTM and recombinant human TM domains promote epithelialwound healingThe recombinant human TM domain (rTMD23) is amitogenic factor (Tohda et al., 1998; Shi et al., 2005).The rTMD123 is also mitogenic for HaCaT (SupplementaryFigure S2 online). Therefore, RHBDL2 may promote cuta-neous wound healing by producing sTM in situ. Serum-freeCM from the wound-stimulated HaCaT promoted the healingof keratinocytes, and pretreatment with purified anti-TMantibody (400 ng ml�1, 37 1C, 30 minutes) abolished thestimulatory effect. Moreover, the CM from shRHBDL2 cellshad no effect on wound healing. Furthermore, purifiedrTMD123 (20 nM) also had a similar effect as that of the CMfrom wounded HaCaT (Figure 5). The results suggest thatsTM released by RHBDL2 has a significant role in woundhealing in HaCaT.

The roles of RHBDL2 and sTM in cutaneous wound healingin vivo

To test the roles of RHBDL2 and sTM in mouse cutaneouswound healing, we first tested how wound stimulationregulates their expression. The results show that TM expres-sion was elevated 1–5 days after transdermal incision surgery;this is consistent with the results of the immunohistochemicalstudy (Peterson et al., 1999). RHBDL2 was also upregulatedat 3–5 days after incision (Figure 6a). Furthermore, RHBDL2activity and TM shedding were observed in murine skinorgan cultures. Treatment with DCI reduced sTM secretionin a dose-dependent manner, and complete inhibition wasobserved at higher doses (i.e., 1 mM; Figure 6b). These resultsimply that RHBDL2 is involved in sTM release from injuredskin. The effects of RHBDL2 inhibitor and sTM on cutaneouswound healing were also evaluated. Two days after incision,DCI (1 mM) or vehicle (Lipovenoes 10%) was injected

subcutaneously into the wound margin once a day for2 days. Results indicate that the inhibitory effects of DCI onwound healing are significantly abolished by treatment withrTMD123 proteins (Figure 6c and d). In addition, Masson’strichrome and proliferating cell nuclear antigen stainingresults indicated that expression of hyperproliferative epithe-lial cells was reduced under DCI treatment, and this changecould be abolished by rTMD123 (Supplementary Figure S3online). On the basis of these results, we propose that theexpressions of RHBDL2 and TM are upregulated, leading tothe release of sTM, which most likely has an important role incutaneous wound healing.

DISCUSSIONRHBDL2 has the potential to catalyze the proteolyticcleavage of membrane-bound substrates and release solubleprotein ectodomains, including TM, ephrin B2, and B3 (Lohiet al., 2004; Pascall and Brown, 2004). It has been shown thatTM can be efficiently and specifically cleaved by RHBDL2(Lohi et al., 2004). As TM knockdown in mice is embryoniclethal and no RHBDL2 knockout mice were generated,we propose that TM shedding by RHBDL2 conceals someunreported physiological functions. Upregulation of TMexpression was demonstrated in damaged tissues, includingthe brain (Pindon et al., 2000), endothelium, and epidermis.Until now, the mechanism of TM release and the physio-logical function of sTM in tissue repair remained mostlyunknown. Cutaneous wounds induced prominent upregula-tion of TM in the neoepidermis; this is an interesting modelfor investigating the function of TM and RHBDL2 in tissuedamage. Results of our study provide previously unreportedevidence that the expressions of TM and RHBDL2 areupregulated in keratinocytes after damage, both in vivo andin vitro. The RNA released from damaged tissue mayrepresent one of the critical factors in the induction of TMand RHBDL2 expression. We also demonstrated that sTMrelease from culture cells or tissues is RHBDL2 dependentand that it has a critical role in cutaneous wound healing.These results are consistent with those of previous reports thatindicate that rTM is a mitogenic factor (Tohda et al., 1998;Shi et al., 2005).

Seven rhomboid homologs (RHBDL1–RHBDL7) have beenidentified; only RHBDL2 cleaves Drosophila Spitz (Lee et al.,2001; Urban et al., 2001). However, little is known about theregulatory mechanisms and physiological significance ofRHBDL2. Our results indicate that the generation of sTMin vitro is significantly inhibited by the RHBDL2 inhibitor.Moreover, a more specific shRHBDL2 was applied to confirmthat wound-induced sTM production is due to the catalyticactivity of RHBDL2 in keratinocytes. In addition, woundingmay evoke a large amount of necrosis or apoptosis. RNA thatreleased from damaged cells can induce tumor necrosisfactor-a production in keratinocytes by activation of Toll-likereceptor 3 (Kono and Rock, 2008; Lai et al., 2009). The CM ofdamaged cells becomes less effective in inducing TM andRHBDL2 after treatment with RNase. Our preliminary resultsshowed that pretreatment with Toll-like receptor 3 shRNAinhibited UVR-induced upregulation of TM and RHBDL2




(data not shown). On the basis of these results, we furtherpropose that the wound-induced upregulation of RHBDL2and TM is regulated by RNA/Toll-like receptor 3 signalingpathways.

The shRNA-induced downregulation of RHBDL2 inhibitedcell proliferation and motility, subsequently resulting indelayed wound recovery in keratinocytes. These findingshighlight the pivotal role of RHBDL2 in regulating the growth

GFP

sTM

120a cGFP

shRHBDL2

***

100

80

60

40

20

00 6

Time (hours)

Time (hours)

0

GF

Psh

RH

BD

L2

120

12

10

8

6

4

2

00 1 2

Days

3 4

100

80

60

40

20

0HaCaT GFP shRHBDL2

shRHBDL2

GFP

HaCat

Stable line

***

***

***

***

Tran

swel

l mig

ratio

n (%

)

Cel

l num

ber/

wel

l in

6-w

ell p

late

(× 1

05 )

6 12 24

12 24

Wou

nd r

ecov

ery

(%)

GST

Med

Lys

RHBDL2

β-Actin

shRHBDL2

Figure 4. Short hairpin RNA rhomboid-like-2 (shRHBDL2)-derived downregulation of RHBDL2 caused the inhibition of thrombomodulin (TM) shedding,

wound recovery, cell migration, and proliferation in HaCaT. (a) Western blot analysis of TM shedding and RHBDL2 expression in stably transfected cells

(green fluorescent protein (GFP) and shRHBDL2). (b) Scratch-wound healing assay of stably transfected cells cultured in complete medium and observed by

time-lapse recording. (c) Quantitative representation of the results of b. Data were calculated on the basis of the experiments with three different stable clones.

(d) Transwell migration assays for HaCaT, GFP, and shRHBDL2 stable clones. (e) Cell growth was monitored by cell counts using a hemocytometer.

Bar¼200 mm. Data are from three independent experiments. ***Po0.001. GST, glutathione S-transferase; Lys, lysate; Med, medium; sTM, soluble TM.



and migration of epithelial cells, probably through the releaseof mitogenic factors, such as sTM, or a reduction of cell–celladhesion as TM is also involved in cell–cell adhesion.

Soluble TM fragments are present in plasma, urine, andsynovial fluid (Jackson et al., 1994). Our study indicates that90-kDa sTM, the same size as human rTMD123, may bereleased from keratinocytes. The molecular mass of sTM isconsistent with the hypothesis that TM shedding is due to itsproteolytic cleavage by RHBDL2 at the N-terminal regionof the transmembrane domain and the fact that the entireectodomain of TM was identified in the culture medium(Lohi et al., 2004).

Our results indicate that TM antibodies and DCI couldsignificantly retard wound healing in cultured keratinocytesand mouse skin. These observations suggest that sTMoriginates from the neoepidermis at the edge of the woundand serves as an autocrine/paracrine growth factor thatpromotes wound healing. In previous studies, variousTM-deficient mice were used to evaluate the impact of TMduring cutaneous wound healing; however, no significantdifferences with respect to the rate or extent of re-epitheliali-zation were observed (Peterson et al., 1999). In fact, theseTM-deficient, heterozygous TM-deficient (TMþ /�), com-pound heterozygous (TMpro/�), and TM�/� chimeric mice

were not completely null for TM. Sufficient sTM may bereleased from these TM-deficient mice under conditions suchas wounding. Therefore, skin-specific TM-knockout mice orRHBDL2-knockout mice are warranted for validating theeffects of TM or sTM in wound healing.

Wound healing is an intricate process, and many EGFRligands have important roles in the repair process (Schultzet al., 1987; Greenhalgh, 1996; Steed, 1998). Interestingly,most EGFR ligands are synthesized as membrane-anchoredforms that are subjected to proteolytic processes to generatebioactive soluble forms (Massague and Pandiella, 1993).The ectodomain shedding of EGFR ligands is essentialfor keratinocyte migration during cutaneous wound healing(Tokumaru et al., 2000). In this study, we demonstratedthat RHBDL2 participates in the wound repair processby specifically releasing sTM during the early proliferativephase. sTM may activate cell proliferation and migrationby binding to its specific receptors, which should beidentified in the future.

Inflammation is an essential protective mechanism thatoccurs after injury. Toll-like receptor 3 is required forcommensal bacteria to regulate inflammation after skin injury(Lai et al., 2009). It is interesting to note that RNA releasedfrom the damaged tissue can stimulate upregulation of

Controlmedium

0

a

b

24

100

***

Control medium WoundedHaCaT CM

WoundedHaCaT CM +

TM Ab

WoundedshRHBDL2 CM

rTMD123

*** ***

***80

60

40

20

0Wou

nd r

ecov

ery

(%)

Tim

e (h

ours

)

WoundedHaCaT CM

WoundedHaCaT CM

+ TM Ab

WoundedshRHBDL2

CM rTMD123

Figure 5. Role of rhomboid-like-2 (RHBDL2) in thrombomodulin (TM) shedding and wound healing. Conditioned medium (CM) from scratch-wounded

HaCaT cells or human rTMD123 enhances scratch-wound healing. (a) The CM from scratch-wounded HaCaT enhanced wound healing (left 2). The anti-TM

antibody (Ab) blocked the effect of the CM (left 3). The CM from scratch-wounded short hairpin RNA rhomboid-like-2 (shRHBDL2) cells did not promote

wound healing (left 4). Purified rTMD123 (20 nM) could promote wound healing in HaCaT cells (right). Bar¼200 mm. (b) Quantitative representation of a.

Data are from three independent experiments. ***Po0.001.




TM and RHBDL2, leading to the release of sTM. Recombi-nant TMD1 can neutralize Gram-negative bacteria-inducedinflammatory responses by binding to specific Lewis Y anti-gens (Shi et al., 2008). Therefore, upregulation of TM andsTM may provide another mechanism for modulatinginflammatory response in wound healing.

In conclusion, our work clearly indicates that RHBDL2expression and TM shedding have important roles in thecutaneous wound healing process (Figure 6e). This study also

provides a foundation for exploring the potential therapeuticeffects of sTM in treating chronic wounds caused by diabetes,chemotherapy, radiotherapy, or steroid hormones.

MATERIALS AND METHODSMaterialsThe HaCaT human keratinocyte cell line was grown with 10% fetal

bovine serum (Invitrogen, Carlsbad, CA; Huang et al., 2003). The

RHBDL2 shRNA expression vector, HuSH-29, was purchased from

Days after wounding

0

RHBDL2

sTM

0

Wound + DCI (μM)

50 100 1,000Unw

ound

ed

GST

TM

Vehicle

1

5

7

9

Day

s af

ter

wou

ndin

g

DCIDCI +

rTMD123100

DCI + rTMD123

DCI

Vehicle

Inject

**

90

80

70

60

50

40

30

20

10

01 2 3 4 5 6 7 8 9

Days after wounding

Wound stimulation(RNA)

RHBDL2

TMTM Ab

Keratinocytes

Migration,proliferation,wound healing

DCI

RHBDL2 shRNA

Wou

nd r

ecov

ery

(%)

β-Actin

1 3 5 7 9

Figure 6. Cutaneous wounding in mice induces rhomboid-like-2 (RHBDL2) expression and soluble thrombomodulin (sTM) release, which is involved in

wound healing. (a) The lysates of wound-edge cells were harvested at the indicated time points and subjected to western blot analysis. (b) The conditioned

medium from wounded culture skin was harvested for western blot analysis after 2 days. (c) Macroscopic observation of wound healing in mice that were

subcutaneously injected with vehicle (Lipovenoes), 3,4-dichloroisocoumarin (DCI, 1 mM), or DCI plus rTMD123 at the margin of the wound. Bar¼4 mm.

(d) Quantitative representation of c. rTMD123 treatment group compared with DCI group; *Po0.05 (n¼ 5). (e) Schematic model of RHBDL2 and sTM

promotion of wound healing. Ab, antibody; GST, glutathione S-transferase; TM, thrombomodulin.



OriGene Technologies (Rockville, MD). GFP expression vector,

pEGFP-N1, was purchased from Clontech (Palo Alto, CA). Human

RHBDL2 cDNA (GeneCopoeia, Rockville, MD) was constructed into

pEGFP-N1 to generate GFP-tagged RHBDL2 expression vector,

pRHBDL2-GFP. RHBDL2 antibody was purchased from Aviva

Systems Biology (San Diego, CA). TM (D-3) antibody, b-actin

antibody, and glutathione S-transferase antibody were purchased

from Santa Cruz Biotechnology (Santa Cruz, CA). The neutralizing

antibody against human TM was generated by immunizing mouse

with rTMD23 (Shi et al., 2005), and purification by protein-G column.

Expression and purification of rTMD proteinsThe pCR3 vector (Invitrogen) was used for the expression and

secretion of rTMD23 and rTMD123 in the HEK293 protein expres-

sion system. DNA fragments encoding rTMD23 and rTMD123 were

obtained as described previously (Han et al., 2000). Expressed

rTMD23 or rTMD123 was applied to a nickel-chelating Sepharose

column (Amersham Phamacia Biotech, Piscataway, NJ), and rTMD-

containing fractions were eluted.

Scratch-wound stimulation

Confluent cells in six-well plate were synchronized with serum-

free DMEM for 12 hours before scratch-wound stimulation, and

then treated with different numbers of scratch wounds (0, 1, 2, 5, 10)

using a 10-ml pipette tip, and maintained under serum-free condi-

tions. After 24 hours of incubation, cell-free media were collected

and concentrated using Centricon tubes (Amicon, Beverly, MA), and

glutathione S-transferase was added (20 mg per sample) as an internal

control. The concentrated samples and cell lysates were then

subjected to western blot analysis.

UVR

Cultured human keratinocytes were irradiated by a single 30-mJ dose

of UV radiation (UV-Stratalinker 1800; Stratagene, La Jolla, CA).

After 24 hours, the UV-irradiated cells (UVR cells) were collected

and added to untreated normal keratinocytes.

Cell growth assays

HaCaT and selected stable cells were plated in six-well plates

(3� 105 cells per well) with culture medium containing 10%

fetal bovine serum. To assess cell growth, the cells were

washed with phosphate-buffered saline, trypsinized, and counted

by a hemocytometer.

In vitro wound healing assayThe confluent cells in six-well plate were scratched using a 10-ml

pipette tip to generate a cell-free zone (0.6–0.8 mm wide). After

extensive washing with DMEM, the cells were incubated for

24 hours at 37 1C with serum-free DMEM, indicated CM or

rTMD123. The cells were photographed using a time-lapse video

microscopy system (Olympus, Tokyo, Japan) immediately after

wounding.

Automated time-lapse video tracking

Time-lapse image acquisition was performed by serial phase-

contrast imaging every hour for 24 hours using an automated

motorized stage. A � 10 objective lens was used on an inverted

Olympus IX81 microscope linked to a DP30 monochrome camera

running analySIS LS Professional software (Olympus). Cells were

only exposed to light during image acquisition. All instrumenta-

tion was enclosed in a chamber maintained at 37 1C and 5% CO2.

Stacked images were superimposed and concatenated into separated

images (.jpg file format) using Metamorph software (Universal

Imaging Corporation, Downingtown, PA). The ratio of wound

recovery was quantitatively determined with ImageJ software

(National Institutes of Health, Bethesda, MD).

shRNA-mediated RHBDL2 gene silencing

The RHBDL2 shRNA-expressing or control vector was transfected

into cells using a Lipofectamine 2000 transfection protocol (Invitro-

gen). The transfection efficiency is about 70%. Transfected cells

were then harvested for western blot analysis or for selecting stable

cell lines with puromycin 48 hours after transfection.

Western blot analysis

Total protein (30 mg) was separated using a 10% SDS-PAGE and

transferred onto a polyvinylidene difluoride membrane. The

membrane was blocked with 5% nonfat dry milk for 1 hour at room

temperature. After being probed with a primary antibody overnight

at 4 1C, the membrane was incubated with a secondary antibody

for 1 hour at room temperature. The signal was detected using

an enhanced chemiluminescence reagent (Amersham Phamacia

Biotech). The band intensity was quantitatively determined using

ImageJ software.

Transwell cell migration assay

Cell migration was analyzed using a modified Boyden chamber

assay in 24-well plates (Transwell; Costar, New York, NY). Cells

were starved by 0.1% fetal bovine serum for 18–20 hours, and placed

into the upper compartment of the chambers (1� 104 cells per

chamber). Culture medium containing 10% fetal bovine serum was

added to the lower chambers. After 24 hours incubation, cells on the

upper face of the membrane were scraped off, and cells on the

lower face were fixed and stained with Liu’s solution (Handsel

Technologies, Taipei, Taiwan). The number of migrated cells on the

filters was counted in five separate fields under � 100 magnification.

Assays were performed in triplicate.

Murine skin organ culture

Punch biopsies (4 mm) were prepared under sterile conditions from

the back skin of adolescent C57BL/6JNarl (B6) mice (Jackson

Laboratories, Bar Harbor, ME) according to previously described

basic organ culture protocols (Li et al., 1992; Paus et al., 1994;

Spiekstra et al., 2007). The culture media were renewed two times

per week. Skin substitutes were ready for experimentation after 3

weeks of culture. After wounding and addition of DCI (0, 50, 100, or

1000mM; Sigma-Aldrich) for 48 hours, the CM was then harvested for

analysis.

In vivo cutaneous wound-healing assay

This assay was modified from Royji and Daniel (Tsuboi and Rifkin,

1990). A full-thickness 20-mm linear wound was made with a

scalpel blade on the back skin over the paravertebral area at the mid-

dorsum of each mouse. The skin surrounding the wound was treated

with injections of sterile Lipovenoes 10% (Fresenius-Kabi, Bad

Homburg, Germany) as a vehicle control, 1 mM DCI, or DCI plus




rTMD123 (125 mg kg�1). This treatment was applied to the wounds

daily for 2 days at the indicated time points. The cut edge of each

wound was photographed on days 1–9 after surgery for the analysis

of wound closure by using ImageJ software. The percentage of daily

wound closure was compared with the original wound area on day 1

(Greenhalgh et al., 1990). All animal experiments were approved by

the Institutional Animal Care and Use Committee of National Cheng

Kung University, Tainan, Taiwan.

Statistical analysis

Data are expressed as mean±SEM. Statistical significance was

analyzed using one-way or two-way analysis of variance. P-values of

o0.05 were considered significant.

CONFLICT OF INTERESTThe authors state no conflict of interest.

ACKNOWLEDGMENTSThis work was supported by a grant from the National Science Council,Executive Yuan, Taiwan (NSC98-2752-B-006-002-PAE to H.-L. Wu).

SUPPLEMENTARY MATERIAL

Supplementary material is linked to the online version of the paper at http://www.nature.com/jid

REFERENCES

Blann AD, Lip GY (1998) The endothelium in atherothrombotic disease:assessment of function, mechanisms and clinical implications. BloodCoagual Fibrin 9:297–306

Esmon CT (1995) Thrombomodulin as a model of molecular mechanisms thatmodulate protease specificity and function at the vessel surface. FASEB J9:946–55

Greenhalgh DG (1996) The role of growth factors in wound healing. J Trauma41:159–67

Greenhalgh DG, Sprugel KH, Murray MJ et al. (1990) PDGF and FGFstimulate wound healing in the genetically diabetic mouse. Am J Pathol136:1235–46

Han HS, Wu HL, Lin BT et al. (2000) Effect of thrombomodulin onplasminogen activation. Fibrinol Proteol 14:221–8

Healy AM, Rayburn HB, Rosenberg RD et al. (1995) Absence of the blood-clotting regulator thrombomodulin causes embryonic lethality in micebefore development of a functional cardiovascular system. Proc NatlAcad Sci USA 92:850–4

Huang HC, Shi GY, Jiang SJ et al. (2003) Thrombomodulin-mediated celladhesion: involvement of its lectin-like domain. J Biol Chem 278:46750–9

Ishii H, Majerus PW (1985) Thrombomodulin is present in human plasma andurine. J Clin Invest 76:2178–81

Jackson DE, Tetaz TJ, Salem HH et al. (1994) Purification and characterizationof two forms of soluble thrombomodulin from human urine. Eur JBiochem 221:1079–87

Kono H, Rock KL (2008) How dying cells alert the immune system to danger.Nat Rev Immunol 8:279–89

Lager DJ, Callaghan EJ, Worth SF et al. (1995) Cellular localization ofthrombomodulin in human epithelium and squamous malignancies.Am J Pathol 146:933–43

Lai Y, Di Nardo A, Nakatsuji T et al. (2009) Commensal bacteria regulate Toll-like receptor 3-dependent inflammation after skin injury. Nat Med15:1377–82

Lee JR, Urban S, Garvey CF et al. (2001) Regulated intracellular ligandtransport and proteolysis control EGF signal activation in Drosophila.Cell 107:161–71

Li L, Paus R, Slominski A et al. (1992) Skin histoculture assay for studying thehair cycle. In Vitro Cell Dev Biol 28A:695–8

Lohi O, Urban S, Freeman M (2004) Diverse substrate recognitionmechanisms for rhomboids; thrombomodulin is cleaved by Mammalianrhomboids. Curr Biol 14:236–41

Massague J, Pandiella A (1993) Membrane-anchored growth factors. AnnuRev Biochem 62:515–41

Mizutani H, Hayashi T, Nouchi N et al. (1994) Functional and immuno-reactive thrombomodulin expressed by keratinocytes. J Invest Dermatol103:825–8

Pascall JC, Brown KD (2004) Intramembrane cleavage of ephrinB3 bythe human rhomboid family protease, RHBDL2. Biochem Biophys ResCommun 317:244–52

Paus R, Luftl M, Czarnetzki BM (1994) Nerve growth factor modulateskeratinocyte proliferation in murine skin organ culture. Br J Dermatol130:174–80

Peterson JJ, Rayburn HB, Lager DJ et al. (1999) Expression of thrombomodulinand consequences of thrombomodulin deficiency during healing ofcutaneous wounds. Am J Pathol 155:1569–75

Pindon A, Berry M, Hantai D (2000) Thrombomodulin as a new marker oflesion-induced astrogliosis: involvement of thrombin through theG-protein-coupled protease-activated receptor-1. J Neurosci 20:2543–50

Raife TJ, Demetroulis EM, Lentz SR (1996) Regulation of thrombomodulinexpression by all-trans retinoic acid and tumor necrosis factor-alpha:differential responses in keratinocytes and endothelial cells. Blood88:2043–9

Raife TJ, Lager DJ, Madison KC et al. (1994) Thrombomodulin expression byhuman keratinocytes. Induction of cofactor activity during epidermaldifferentiation. J Clin Invest 93:1846–51

Raife TJ, Lager DJ, Peterson JJ et al. (1998) Keratinocyte-specific expression ofhuman thrombomodulin in transgenic mice: effects on epidermaldifferentiation and cutaneous wound healing. J Investig Med 46:127–33

Schultz GS, White M, Mitchell R et al. (1987) Epithelial wound healingenhanced by transforming growth factor-alpha and vaccinia growthfactor. Science 235:350–2

Senet P, Peyri N, Berard M et al. (1997) Thrombomodulin, a functionalsurface protein on human keratinocytes, is regulated by retinoic acid.Arch Dermatol Res 289:151–7

Shi CS, Shi GY, Chang YS et al. (2005) Evidence of human thrombomodulindomain as a novel angiogenic factor. Circulation 111:1627–36

Shi CS, Shi GY, Hsiao SM et al. (2008) Lectin-like domain of thrombomodulinbinds to its specific ligand Lewis Y antigen and neutralizes lipopoly-saccharide-induced inflammatory response. Blood 112:3661–70

Spiekstra SW, Breetveld M, Rustemeyer T et al. (2007) Wound-healing factorssecreted by epidermal keratinocytes and dermal fibroblasts in skinsubstitutes. Wound Rep Reg 15:708–17

Steed DL (1998) Modifying the wound healing response with exogenousgrowth factors. Clin Plast Surg 25:397–405

Tohda G, Oida K, Okada Y et al. (1998) Expression of thrombomodulin inatherosclerotic lesions and mitogenic activity of recombinant thrombo-modulin in vascular smooth muscle cells. Arterioscler Thromb Vasc Biol18:1861–9

Tokumaru S, Higashiyama S, Endo T et al. (2000) Ectodomain shedding ofepidermal growth factor receptor ligands is required for keratinocytemigration in cutaneous wound healing. J Cell Biol 151:209–20

Tsuboi R, Rifkin DB (1990) Recombinant basic fibroblast growth factorstimulates wound healing in healing-impaired db/db mice. J Exp Med172:245–51

Urban S, Lee JR, Freeman M (2001) Drosophila rhomboid-1 defines a familyof putative intramembrane serine proteases. Cell 107:173–82

Weiler H, Isermann BH (2003) Thrombomodulin. J Thromb Haemost1:1515–24

Zhang Y, Weiler-Guettler H, Chen J et al. (1998) Thrombomodulin modulatesgrowth of tumor cells independent of its anticoagulant activity. J ClinInvest 101:1301–9



http://www.nature.com/jid

http://www.nature.com/jid

functions of rhomboid family protease rhbdl2 and thrombomodulin in wound healing

Documents