original article© The American Society of Gene & Cell Therapy

Molecular Therapy vol. 17 no. 11, 1959–1966 nov. 2009 1959

Degenerative disc disease (DDD) is a common disease which affects millions of people. Autograft of the bone marrow derived mesenchymal stem cells (BMSCs) have been shown to have the ability to arrest degeneration in rabbit and canine intervertebral discs. In this study, we have used the mouse model to investigate the mecha-nism of degeneration arrest. BMSC from Egfp transgenic mice were injected into the degenerated murine inter-vertebral discs induced by annular puncture. We found that BMSC could arrest the progressive degeneration of the discs with significant regeneration of the nucleus pulposus (NP). In the regeneration, expression of pro-teoglycan genes were upregulated and extracellular matrix (ECM) progressively accumulated in the NP after BMSC injection. Combined in situ hybridization and immunohistochemistry revealed that BMSC underwent chondrocytic differentiation in the regeneration process. Interestingly, BMSC-induced an increase of endogenous notochordal cells in NP and expression of chondrocytic markers. In this study, we have firstly shown that the BMSC could arrest the degeneration of the murine noto-chordal NP and contribute to the augmentation of the ECM in the NP by both autonomous differentiation and stimulatory action on endogenous cells.

Received 31 December 2008; accepted 2 June 2009; published online 7 July 2009. doi:10.1038/mt.2009.146

IntroductIonDegenerative disc disease (DDD) is a common disease which is a major cause of the pain-related disabilities. Pain-related costs aggravate the economic burden of the whole industrial society.1 Compared with the normal disc, the degenerated disc has morpho-logical changes such as the irregular lamellas in the annulus fibrosus (AF) and the diminished celluarity in the nucleus pulposus (NP). There are also biochemical changes in the degenerated disc, includ-ing loss of hydration and decrease of osmotic pressure, quantitative

and distributive changes of collagen, decreased proteoglycan content and downregulated proteoglycan expression.2

Various kinds of biological therapeutic methods have been studied in the hope of regenerating the dysfunctional discs for DDD.3 Among these methods increasing the number of functional cells in NP can be a key to improve the function of intervertebral discs.4 Previous studies have suggested that bone marrow derived mesenchymal stem cells (BMSCs) could dif-ferentiate toward a NP-like phenotype in vitro.5 Coculture of human BMSC and NP cells may also induce BMSC to differenti-ate and express NP-like markers.6,7 Mochida et al. have studied the regenerative effect of autologous BMSC on disc degeneration and demonstrated BMSC may induce recovery of disc height and proteoglcyan level in both rabbit and canine model.8–11 The injected BMSC could survive up to 48 weeks in the rabbit disc12 and express some markers like hypoxia-inducible factor 1 and keratin sulfate,12 suggesting the differentiation of BMSC in the discs.

Compared with human and other large animals mouse has more notochordal cellular components in the NP. Notochordal cells have been suggested to play an important role in regulating the metabolism of the NP13 and the loss of the notochordal cells in NP is closely related to the onset of the disc degeneration.13 Although previous studies showed that BMSC implantation can be a strategy to treat disc degeneration; whether BMSC could stimulate the notochordal NP in vivo and the differentiation status of BMSC in the regeneration process are not very clear. Studying disc degeneration in mouse model would therefore provide a valu-able opportunity to study the differentiation of stem cells in noto-chordal NP and explore the possible roles of notochordal cells in the regeneration process.

In order to test the hypothesis that BMSC could arrest the degeneration of notochordal NP, we established a mouse injured disc model14 to investigate the effects of BMSC and mechanism of degeneration arrest. We found that BMSC could arrest the progressive degeneration of the notochordal NP and the effect was associated with the re-establishment of type II collagen and

Correspondence: Kenneth MC Cheung, Department of Orthopaedics and Traumatology, 21 Sassoon Road, Li Ka Shing Faculty of Medicine, The University of Hong Kong, Hong Kong SAR, People’s Republic of China. E-mail: cheungmc@hkucc.hku.hk

Mesenchymal Stem Cells Arrest Intervertebral Disc Degeneration Through Chondrocytic Differentiation and Stimulation of Endogenous CellsFan Yang1,2, Victor YL Leung1,3, Keith DK Luk1, Danny Chan3 and Kenneth MC Cheung1

1Department of Orthopaedics and Traumatology, Li Ka Shing Faculty of Medicine, The University of Hong Kong, Hong Kong SAR, People’s Republic of China; 2Shenzhen Institute of Advanced Technology, Chinese Academy of Science, Key Lab for Biomedical Informatics and Health Engineering, Institute of Biomedical and Health Engineering, Shenzhen, People’s Republic of China; 3Department of Biochemistry, Li Ka Shing Faculty of Medicine, The University of Hong Kong, Hong Kong SAR, People’s Republic of China

1960 www.moleculartherapy.org vol. 17 no. 11 nov. 2009

© The American Society of Gene & Cell TherapyA Mouse Model Study

proteoglycans-enriched extracellular matrix (ECM) in the NP. Differentiation of BMSC into a chondrocytic phenotype and an increased number of Col2a1-expressing endogenous cells in the NP might have contributed to the re-established ECM. Our study has provided new insights into how BMSC behave and contribute to the regenerative process in notochordal NP and hence ways to improve stem cell therapy for DDD in future.

resultsProgressive degeneration of the intervertebral disc induced by the annulus punctureSafranin O staining was performed to follow the progressive degenerative changes of the disc structure and glycosaminoglycan (GAG) distributive pattern after the puncture. The original noto-chordal phenotype of the murine discs gradually changed from pre-op to 12 weeks (Figure 1a–d). In the pre-op disc the lamellas in AF are parallel and NP contained many vacuolated notochordal cells (Figure 1a). At 2 weeks after the puncture the annulus became infolded and disorganized with safranin O-positive matrix concentrated in the NP (Figure 1b). At 6 weeks, clusters of small unvacuolated cells were observed in the NP (Figure 1c); Finally, at 12 weeks after the puncture the boundary between the AF and NP disappeared and GAG distribution became diffused in the disc, although some clefts could be observed in both AF and NP region (Figure 1d).

A degeneration grading system modified from the rabbits (Supplementary Table S1), ranging the score from 0 to 8 was used to reflect the degree of degeneration after the puncture. The mean histological scores of punctured disc increased from pre-op to 2 weeks (P < 0.05), then increased significantly from 2 weeks to 12 weeks (P < 0.05) (Figure 1e). Disc height index (DHI) decreased significantly at 2 and 6 weeks after the puncture com-pared with the normal control (P < 0.05). From 6 to 12 weeks, the DHI decreased further and was significantly lower than that of normal control (P < 0.01) (Figure 1f).

Expression analysis of major proteoglycans and related genes showed that the expression of Col2a1, Aggrecan, and Sox9 were all downregulated continuously from 2 to 12 weeks compared with the pre-op control group (P < 0.05) (Figure 1g). GAG and DNA contents were also quantified to reflect proteoglycan content changes (Figure 1h). GAG/DNA decreased significantly at 6 week (P < 0.05) and 12 week (P < 0.05) compared with the pre-op control group.

Morphology and characterization of BMsc in vitro cultureBMSCs were isolated from the bone marrow of green fluorescent protein (GFP) mice.15 The freshly harvested bone marrow derived cells have varied sizes and heterogeneous level of green fluores-cence (data not shown); After the first passage the cells grew into

pre-op

2 Weeks

6 Weeks

12 Weeks

a

pre-op 2Weeks

6Weeks

Time

12Weeks

* *98765

His

tolo

gica

l sco

re

43210

e

05060708090

% D

HI 100

110120

2 4 6Weeks after puncture

8 10 12

PunctureControl

* ***

f

4

3

2

GA

G/D

NA

(µg

/µg)

1

0pre-op 2

Weeks6

Weeks

Time

12Weeks

* *

h

b

c

d

COL2a1 Aggrecangene

SOX9

1.25

* *

*

*

**

* **

1.00

0.75

Gen

e ex

pres

sion

0.50

0.25

0.00

g pre-op2 Weeks6 Weeks12 Weeks

Figure 1 Annulus puncture could induce the progressive degeneration of murine intervertebral disc. (a–d) Safranin O staining of the interver-tebral discs from pre-op to 12 weeks after the puncture. Progressive degenerative changes were observed at (a) pre-op, (b) 2 weeks, (c) 6 weeks, and (d) 12 weeks after the puncture. Bar = 200 µm (e) the histological score increased from pre-op to 2 weeks (P < 0.05), then increased from 2 to 12 weeks (P < 0.05) after the puncture. (f) Disc height index (%DHI) decreased continuously at 2, 6 weeks (P < 0.05) and 12 weeks (P < 0.01) after the puncture. (g) The expression of Col2a1, Aggrecan, and Sox9 were significantly downregulated from pre-op to 12 weeks after the puncture (P < 0.01). (h) Continuous decrease of GAG/DNA from pre-op to 12 weeks after the puncture. The GAG content decreased significantly at 6 and 12 weeks were compared with the pre-op control group (P < 0.05).GAG, glycosaminoglycan.

Molecular Therapy vol. 17 no. 11 nov. 2009 1961

© The American Society of Gene & Cell TherapyA Mouse Model Study

confluence and GFP signal intensity appeared more uniform among the cells (Figure 2a,b). Immunostaining was performed to characterize the surface markers of the cultured BMSC. Most of the cultured BMSC expressed the standard mesenchymal stem cell surface markers16 Stro-1 (Figure 2c), and CD73 (Figure 2d), whereas they did not express the hematopoietic markers CD34 (Figure 2e) and CD45 (Figure 2f). Gene expression analysis further showed that articular chondrocytes expressed two chon-drocytic markers: Sox9 and Col2a1, which BMSC could not or expressed at very low level (Figure 2g).

BMsc could arrest the progressive degeneration of the intervertebral discsBMSC was injected into the intervertebral discs 2 weeks after the degeneration was induced by annular puncture (Figure 2h). In normal group, histology revealed that the AF composed of reg-ularly aligned lamellas and NP appeared as an organized struc-ture containing abundant cellular components (Figure 3a–c). In degeneration group, the lamellas in AF became irregular and NP became disorganized and decreased in cellularity with clusters of cells formed at 4 weeks (Figure 3e). At 24 weeks, the cellularity diminished further in NP and the extent of safranin O staining significantly reduced (Figure 3f). In the treatment group such progressive degenerative changes were arrested. The alignment of lamellas in AF remained regular and parallel and the cellular components in NP were still abundant at 4 weeks postinjection (Figure 3h). At 24 weeks, the structure of AF was still relatively intact; NP became dominantly positive for safranin O with increased matrix to cell ratio (Figure 3i).

DHI remained relatively stable in normal group whereas decreased continuously in degeneration group from 0 to 24 weeks postinjection. The DHI of treatment group is signifi-cantly higher than degeneration group at 24 weeks (P < 0.05)

(Figure 3j). Histological grading (Figure 3k) showed that at 4 weeks postinjection the mean score of treatment group (2.2) was significantly lower than degeneration group (P < 0.05). At 24 weeks postinjection the mean score of treatment group (3.8) was significantly lower than the degeneration group (P < 0.05), reflect-ing the degenerative process was arrested in the treatment group. From 4 to 24 weeks there is no significant change of histological score in the treatment group (P > 0.05).

ECM gene expression was upregulated in the discs after BMsc injectionReverse transcription–PCR analysis showed that there is a decrease in the expression of Col2a1, Aggrecan, and Sox9 in the degenera-tion group whereas the reduction of expression was significantly arrested (P < 0.05) in treatment group at 24 weeks postinjection (Figure 4a). GAG/DNA content in the treatment group was sig-nificantly higher compared with the degeneration group (P < 0.01) at 24 weeks postinjection (Figure 4b).

In situ hybridization with Col2a1 probe showed that in normal group, Col2a1-expressing cells were concentrated in the inner annulus (AF) and the endplate (Figure 4c–e). In degeneration

a b

c d

e f

Sox9

BMSC

AC

Col2a1g

MSC1

2

C57 mouse GFP mouse

Induce taildisc degeneration

h

Figure 2 Bone marrow derived mesenchymal stem cells (BMscs) isolated from GFP mice. (a,b) BMSC were cultured in Dulbecco’s modi-fied Eagle medium at P1. The cells observed in the (a) bright field can emit green fluorescence under (b) fluorescence microscope. Bar = 100 µm. Immunostaining of (c) Stro-1, (d) CD73, (e) CD34, (f) CD45 of the cul-tured BMSC. BMSC could express Stro-1 and CD73, whereas could not express CD34 and CD45. Bar = 40 µm (c) reverse transcription–PCR analysis of Sox9 and Col2a1 in articular chondrocytes and BMSC. (h) Experimental design for transplantation procedure of BMSC in the degenerated disc. GFP, green fluorescent protein.

pre-op

a b c

d e f

g h i

4 Weeks 24 Weeks

His

tolo

gica

l sco

re

0

1

2

3

4

5

6

7

8

9*

*

4 Weeks 24 Weeks

k NormalDegenerationTreatment120

110100

9080

% D

HI

70605040

0 4 8 12

*

Time (weeks)16 20 24

j NormalDegenerationTreatment

Figure 3 the progressive degeneration of murine disc was arrested by BMsc. (a–i) Safranin O staining of intervertebral discs isolated from normal group (a,b,c), degeneration group (d,e,f) and treatment group (g,h,i) at pre-op (a,d,g), 4 weeks (b,e,h) and 24 weeks (c,f,i) after injection. In normal disc (a–c), lamellas in annulus fibrosus (AF) align parallel and nucleus pulposus (NP) is a highly organized structure; degenerative changes were observed in degeneration group (e,f): the lamellas in AF became disorganized, clusters of cells formed in NP at 4 weeks (e) and further decreased at 24 weeks (f); regenerative changes were observed in treatment group (h,i): the lamellas in AF maintained regular at 4 week (h) and 24 week (i), cellular components maintained in NP at 4 weeks (h) and changed into glycosaminoglycan rich pheno-type at 24 weeks (i). Bar = 200 µm. (j) Change of DHI in three different groups: At 24 weeks postinjection %DHI in treatment group was signifi-cantly higher than the degeneration group (P < 0.05). (k) Histological score of normal, degeneration and treatment group at 4 weeks and 24 weeks after the injection. Histological score in D group was signifi-cantly higher than T group at both 4 weeks (P < 0.05) and 24 weeks (P < 0.05). DHI, disc height index.

1962 www.moleculartherapy.org vol. 17 no. 11 nov. 2009

© The American Society of Gene & Cell TherapyA Mouse Model Study

group (Figure 4f–h) few signals could be detected in the NP region at 24 weeks (Figure 4h, arrow). In treatment group (Figure 4i–k), Col2a1-expressing cells increased in the NP region at 4 weeks (Figure 4j, arrow); at 24 weeks there were more Col2a1-expressing cells in the treatment group than the normal and degeneration group. Some cells formed clusters within the entire NP (Figure 4k, arrows). Immunostaining showed that in normal group, type II collagen was distributed regularly at the peripheral region of NP from 4 to 24 weeks (Figure 4l,m). In degeneration group, little collagen II signal was detected in NP except trace signals around the cleft at 24 weeks (Figure 4o). In treatment group, accumula-tion of collagen II was observed in the NP from 4 to 24 weeks (Figure 4p,q).

chondrogenic differentiation of the BMsc in the discGFP immunostaining showed a decreased number of BMSC by 24 weeks after the injection (Figure 5a). At 4 weeks, most of the BMSC were spindle shape (Figure 5b,d arrow). At 24 weeks postinjection, some BMSC attained a round chondrocyte-like morphology (Figure 5c,e arrow).

Co-staining GFP and chondrocytic markers (Sox9/Col2a1) revealed that some GFP-positive BMSC cells expressed Col2a1 (Figure 5f) or Sox9 (Figure 5g) at 4 weeks postinjection. The per-centage of Col2a1 and Sox9 positive cells in NP was quantified to

follow the differentiation status of BMSC after injection (Figure 5h). We found that the number of Col2a1+/GFP+ cells increased from 4 weeks (42.5 ± 5.26%) to 24 weeks (61.92 ± 6.33%) postinjec-tion (P < 0.05). Similarly, the number of Sox9+/GFP+ cells also increased significantly (47.6 ± 5.30% to 67.0 ± 5.41%, P < 0.05). This suggests a progressive chondrogenic differentiation of BMSC in NP after injection.

Increased population and expression of chondrocytic markers of endogenous cells after BMsc injectionThe double-staining experiments also allowed us to trace and examine the GFP-negative endogenous notochordal cells. At 4 weeks postinjection the cell clusters formed in NP of treatment group comprised mostly endogenous vacuolated cells (Figure 6b, arrow) and was relatively more abundant than the degeneration group (Figure 6a, arrow). At 24 weeks, the amount of endogenous cell clusters in treatment group (Figure 6d, arrow) was still more abundant than the degeneration group (Figure 6c, arrow). By quantification, the number of GFP-negative endogenous cells in treatment group was significantly higher than degeneration group at both 4 weeks (P < 0.01) and 24 weeks (P < 0.05) (Figure 6e).

Double staining with Col2a1 probe and GFP antibody revealed four types of cells located in NP in treatment group: Col2a1 posi-tive (Figure 6f) and negative (Figure 6g) endogenous cells, as well

4 Weeks 24 Weeksl m

n o

p q

pre-op 4 Weeks 24 Weeks

c d e

f g h

i j k

4

3

**

2

GA

G/D

NA

(µg

/µg)

1

0Normal Degeneration Treatment

b

Col2a10.0

0.5

1.0

Gen

e ex

pres

sion

1.5

*

*

*

Aggrecangene

Sox9

aNormal

Degeneration

Treatment

Figure 4 upregulated ECM gene expression and increased extracellular matrix in the regenerated discs. (a) Gene expression analysis of Col2a1, Aggrecan, and Sox9 in normal, degeneration and treatment group at 24 weeks postinjection. The expression level in treatment group was significantly higher than degeneration group (P < 0.05). (b) Change of GAG contents in different groups: GAG increased significantly in treatment group com-pared with degeneration group (P < 0.01). (c–k) Different expression pattern of Col2a1 were observed in normal (c,d,e), degeneration (f,g,h) and treatment group (i,j,k) at pre-op (c,f,i), 4 weeks (d,g,j) and 24 weeks (e,h,k) postinjection. Positive signals in degeneration group decreased from 4 weeks (g, arrow) to 24 weeks (h, arrow). Positive signals continuously increased in treatment group from 4 weeks (j, arrow) to 24 weeks (k, arrow). Bar = 200 µm. (l–q) Immunostaining of collagen II of the normal (l,m), degeneration (n,o) or treatment group (p,q) from 4 weeks to 24 weeks. Collagen II accumulated around the cell clusters from 4 to 24 weeks (p,q). Bar = 50 µm. GAG, glycosaminoglycan.

Molecular Therapy vol. 17 no. 11 nov. 2009 1963

© The American Society of Gene & Cell TherapyA Mouse Model Study

as Col2a1 positive (Figure 6h) and negative (Figure 6i) BMSC. In treatment group, the number of Col2a1+/GFP− cells, which indicate the Col2a1-expressing endogenous cells, was significantly higher than the degeneration group at both 4 weeks (P < 0.05) and 24 weeks (P < 0.01) (Figure 6j), suggesting that the endo-genous cells in notochordal NP transformed into a chondrocyte-like phenotype under the influence of BMSC.

dIscussIonIn this study we used an injured mouse tail disc model14 to explore the effect and role of BMSC in the degenerative murine discs. By tracking the fate of both GFP-tagged BMSC and endogenous cells in NP, we firstly showed that a proportion of BMSC could sur-vive up to 24 weeks in the degenerated murine intervertebral discs and arrest the progressive degeneration of the notochordal NP through self-differentiation and stimulatory effect on endogenous notochordal cells.

The survival rate of the allogenic cells is much dependent on the immune reactions.17 In our study, we did not find infiltrating

cells like neutrophils and lymphocytes after BMSC injection. This is in line with the notion that the intervertebral disc is being immune privileged owing to its avascular nature.18 The decreased number of BMSC observed at 24 weeks postinjection suggests that the degenerative environment may be detrimental to the survival of the BMSC. We postulate that an alteration of nutrient diffu-sion in degenerated NP could be a potential factor that limit the activity and survival of BMSC in long term.18

Our data showed that BMSC is beneficial in arresting the degenerative process of murine disc as the loss of disc height can be arrested from 4 to 24 weeks after BMSC injection. Histological assessment confirmed that the BMSC had regenerated the struc-ture of the degenerated intervertebral disc, as indicated by repara-tion of both lamella organization in AF and partial recovery of cellular components in NP. The regenerative effect is also sup-ported by the lowered degeneration grade observed in the BMSC treatment group.

The upregulated Aggrecan expression and the increased GAG content in the BMSC-treated discs suggested that the anabolic function of the intervertebral discs had been partially restored after the injection of BMSC. Increase of ECM composing colla-gen II from 4 to 24 weeks suggests the NP of the BMSC-treated discs has become cartilagenous from 4 to 24 weeks. The con-tinuous accumulation of the ECM in NP could also explain the structural modulation and the disc height recovery of the BMSC-treated discs. The coexpression of GFP and Col2a1/Sox9 suggested chondrogenic differentiation of the BMSC in the notochordal NP. The increase of Col2a1 and Sox9 positive cells in the regeneration process from 4 to 24 weeks indicates that the chondrogenic dif-ferentiation of BMSC in NP is in parallel with the accumulation of cartilagenous matrix after BMSC injection. We postulate that the differentiated BMSC might have a role in contributing to the elevated Aggrecan expression and GAG content in NP which lead to the formation of Safranin O-positive matrix.

Then we investigated the in vivo interaction between BMSC and endogenous notochordal cells. Our data have demonstrated that BMSC have a stimulatory effect on the endogenous noto-chordal cells. BMSC injection induced an increase in the endo-genous cell population in vivo, suggesting that BMSC can promote notochordal cell survival and proliferation, or prevent them from apoptosis. The BMSC could also stimulate the endogenous cells in notochordal NP to transform to chondrocytic phenotype through expressing chondrocytic markers, contributing to the deposition of cartilagenous ECM in NP. Previous studies have suggested that notochordal cells could maintain the metabolic balance of NP through signaling to surrounding cells19 or stimulate cells in NP to synthesize ECM composed of proteoglycan.20 Here our data further suggested that the anabolic function of notochordal cells could be stimulated by BMSC, thus contributing to the functional recovery of the NP. Endogenous cells played an important role in assisting the BMSC-based regeneration of the notochordal NP and maintaining the normal function of the intervertebral disc.

The balance of synthesis and degradation of the ECM deter-mines the functional status of the intervertebral discs. Stem cells have been reported to regenerate the tissues and organs through multiple mechanisms.21 In our study we hypothesize that two mechanisms co-exist in the BMSC-induced arrest of degeneration.

f

g

b c

d e50

25

4 Weeks 24 Weeks

Time

GF

P+ c

ell n

umbe

r (%

)

0

100

*

a

75

Per

cent

age

(%)

4 Weeks 24 Weeks

Time

0

50

100

Sox9Col2a1

*

*

h

Figure 5 In vivo chondrocytic differentiation of bone marrow derived mesenchymal stem cell (BMsc) in the regenerated discs. (a) Quanti-fication of GFP-positive cells in the murine discs. GFP-positive cells decreased significantly from 4 to 24 weeks (P < 0.05), the number of GFP-positive cells at 4 weeks was set as 100%. (b–e) GFP staining (b,c) and safranin O staining (d,e) of the BMSC. BMSC changed from spindle shape (b,d, arrow) into the round chondrocytic phenotype (c,e, arrow). Bar = 10 µm. (f,g) Double staining of the injected stem cells using GFP antibody (red) with Col2a1 probe (f, green) or Sox9 antibody (g, green) at 4 weeks postinjection. 4′,6-diamidino-2-phenylindole was used to stain the cell nucleus (blue) and the colors were labeled in the image. Bar = 50 µm. (h) Ratio changes of Col2a1 and Sox9 positive cells at 4 weeks and 24 weeks postinjection. The ratio increased significantly from 4 weeks to 24 weeks for Col2a1 (P < 0.05) and Sox9 (P < 0.05).GFP, green fluorescent protein.

1964 www.moleculartherapy.org vol. 17 no. 11 nov. 2009

© The American Society of Gene & Cell TherapyA Mouse Model Study

On one hand, BMSC could directly differentiate into chondrocytic phenotype in the regeneration process possibly due to local niche in the NP; on the other hand, BMSC could also stimulate the endogenous notochordal cells or NP progenitors to proliferate or prevent them from undergoing apoptosis through secreting signaling molecules or direct cell–cell contact.22 These stimulated endogenous cells, together with the differentiated BMSC, secreted cartilagenous ECM which outbalance the ECM degradation in NP and maintained the structure of the intervertebral disc.

Future investigation using transgenic mouse models may help further understanding the mechanism underlying the BMSC-based IVD regeneration, such as the mode of BMSC action and the origin of the de novo NP cells being induced. In addition, in order to correlate the increased GAG and ECM production in NP with the functional recovery of the intervertebral disc, the bio-mechanical property of the disc also needs to be investigated in future. As notochordal cells no longer exist in human degenera-tive disc, the clinical relevance or translation of the current study might be limited due to the presence of residue notochordal cells in the injured murine discs. Nevertheless, our findings indicate a role of notochordal cells in the stem cell-based regeneration pro-cess and hence provide insights in the design of new regenerative methods, such as cotransplantation of the notochordal cells and mesenchymal stem cell, for the treatment of DDD.

In conclusion, we have shown that allogenic BMSC can be viable in the degenerated murine intervertebral discs and arrest the degeneration through partial regeneration of the disc struc-ture. BMSC can undergo chondrogenic differentiation and

simultaneously stimulate the endogenous NP cells, either the notochordal cells or other progenitor cells, to re-establish a carti-laginous NP. Further understanding the molecular mechanisms of the regeneration, as well as tests in larger animal models shall substantiate the use of bone marrow derived stem cells in treating DDD in future.

MAterIAls And MethodsEstablishment of injured mouse disc model by annular puncture. All experimental protocols were approved by the Committee on the Use of Live Animals in Teaching and Research of the University of Hong Kong. A total of 108 mice were used for the study. 45 C57 mice aged 10 weeks were used for the characterization of the degeneration process after the annu-lar puncture. The other 63 mice were evenly divided into three groups: control normal group, degeneration group, and treatment group. Disc degeneration was induced in the tail discs in degeneration and treatment groups at caudal 4–5 level by annular puncture. Briefly, we anesthetized the mice by intraperitoneal injection of Hypnorm (0.5 ml/kg) and dormicum (0.5 ml/ kg; Janssen, Oxford, UK).23 After locating the operative disc from radiographs, the target disc was punctured through the annulus by a 31G needle at a controlled depth of needle bevel under microscopic guidance. After the operation no leakage of disc tissue was found around the wound, then the subcutaneous muscle and skin were sutured and the mice were allowed of free activity in the cages.

BMSC culture and delivery into degenerated discs. Five GFP transgenic mice15 aged 10 weeks were used for isolation of BMSC using the protocol adapted by Makino et al.24 The purified cells were plated in the 75 cm2 flask and cultured with Dulbecco’s modified Eagle medium supplemented with 20% fetal calf serum and penicillin (100 U/ml)/streptomycin (100 μg/ ml). The culture medium was refreshed every 3 days and the nonadherent

a b

c d

f g

h i10µm 10µm

10µm 10µm

pre-op 4 Weeks 24 WeeksTime

4

e

3**

*2

NP

cel

l num

ber

(103 )

1

0

NormalDegenerationTreatment

pre-op 4 Weeks 24 WeeksTime

**

*

1.0

0.8

0.6

0.4

0.2

0.0

j

Col

2a1+ N

P c

ell n

umbe

r (1

03 )

NormalDegenerationTreatment

Figure 6 endogenous nP cells increased and differentiated into the chondrocytic phenotype. (a–d) GFP staining of the nucleus pulposus in degeneration group (a,c) and treatment group (b,d) at 4 weeks (a,b) and 24 weeks (c,d). The endogenous NP cells which were GFP negative could be observed at both time points (arrow). Bar = 50 µm (e) comparison of the number of endogenous NP cells between normal group and treatment group. The cell number in treatment group was significantly higher than degeneration group at 4 week (P < 0.01) and 24 weeks (P < 0.05). (f–i) Different molecular phenotypes of cells were observed from the double staining, including Col2a1-expressing (f) and nonexpressing (g) GFP-negative endogenous cells as well as Col2a1 expressing (h) and nonexpressing (i) GFP-positive bone marrow derived mesenchymal stem cell. (j) The number of Col2a1 positive endogenous NP cell changes from 4 weeks to 24 weeks. The positive NP cells significantly increased in treatment group compared with degeneration group from 4 to 24 weeks (P < 0.05). GFP, green fluorescent protein; NP, nucleus pulposus.

Molecular Therapy vol. 17 no. 11 nov. 2009 1965

© The American Society of Gene & Cell TherapyA Mouse Model Study

hematopoietic stem cells and other blood cells were removed after the second change of the medium. When the adherent cells became conflu-ent, they were passaged and expanded. The articular chondrocytes were also isolated from the articular cartilage of the GFP mice as previous described.25 Immunostaining and reverse transcription–PCR analysis were conducted for characterizing the surface markers and gene expression pro-file of the cultured BMSC.

At 2 weeks after the annular puncture, ~1 × 103 P1 BMSC in Dulbecco’s modified Eagle medium (106 cells/ml in 1µl) were injected into the discs in the treatment group with a 33G micro-injector (Hamilton, Reno, NV); Only Dulbecco’s modified Eagle medium was injected into the discs in degeneration group. After the operation the subcutaneous muscle and skin were sutured and the mice were allowed of free activity in the cages.

Radiographic analysis of disc height. Anterior–posterior radiographs of the mouse tail were taken at pre-op, 4 weeks and 24 weeks after BMSC injection using cabinet X-ray system (model 43855a; Faxitron, IL) with an exposure time of 40 second and penetration power of 20 kv. The DHI of the disc was calculated as previously described.26 The change of the DHI was expressed as %DHI (post-op DHI/pre-op DHI). Two examiners par-ticipated in the image analysis for the DHI assessment. The inter observer error and the within-subject standard deviation were calculated.

Histological staining and grading. The discs and adjacent vertebral bodies were isolated at pre-op, 4 week and 24 weeks after BMSC injection. Tissue processing and sectioning was carried out as previously described.27 Tissue sections were stained with safranin O for glycosaminogycan. A modified grading system (Supplementary Table S1) was used to assess the degree of degeneration based on the established rabbit disc grading system.28 A grade score ranged from 1 to 4 was assigned to both AF and NP structure of the disc which were summated to obtain an overall score to determine the degree of degeneration.

Immunostaining. For detecting the surface markers of the GFP-expressing BMSC, BMSC at passage 1 was fixed with paraformaldehyde, then the mouse antibodies to Stro-1, CD73, CD34, CD45 (Invitrogen, Carlsbad, CA) at 5 μg/ml were used to characterize the surface markers of BMSC. For the tissues, the midsagittal sections of the discs were dewaxed, rehydrated, and blocked by goat serum. Then the sections were incubated with rabbit anti-GFP polyclonal antibody at 5 μg/ml, rabbit antimouse collagen II antibody or rabbit antimouse Sox9 antibody (Chemicon, Temecula, CA) at 10 μg/ml for 60 minutes. The slides containing cells or tissue sections were washed three times and then incubated with biotinylated secondary antimouse or antirabbit antibody (Vector, Burlingame, CA) for 30 minutes. The sec-tions were washed again and treated with avidin-biotin complex reagents followed by 3, 3-diaminobenzendine reagents (Vector, Burlingame, CA). Sections were counterstained with Harris hematoxylin.

In situ hybridization. In situ hybridization using Col2a1 probe was per-formed on disc sections as previously described.29 Briefly, the sections were hybridized with hybridization solution containing digoxygenin-UTP-labeled antisense riboprobes at 60 °C overnight. The sections were then treated with fluorescin-conjugated anti-digoxigenin antibody and color was developed. For double staining, immunostaining was performed after the in situ hybridization. Finally, the sections were counterstained with 4′,6-diamidino-2-phenylindole to stain the nuclei and then mounted.

Double staining and cell counting. Three discs from each of the pre-op, 4 weeks and 24 weeks postinjection group were analyzed to quantify the Col2a1+/GFP+, Sox9+/GFP+, Col2a1+/GFP−, Col2a1−/GFP− in NP region. In each section five rectangular regions of the same size in NP region were selected for analysis; the number of positive cells was sepa-rately counted and summed up. For each disc the number of positive cells was represented as the amount of positive cells summed up from 10 sagittal

sections. The number of Col2a1+/GFP+, Sox9+/GFP+ cells was expressed as a percentage of the total GFP-positive cells.

Gene expression analysis using reverse transcription–PCR. Total RNA were separately extracted from three whole intervertebral discs ( including endplates) isolated from each of the normal, degeneration and treatment groups at 24 weeks using Trizol method. RNA was also extracted from cultured BMSC and articular chondrocytes. Primers were chosen from the gene database (Supplementary Table S2). PCR was performed by denaturation at 94 °C for 30 seconds, annealing at 55 °C for 30 seconds and extension at 68 °C for 45 seconds for a total of 35 cycles. The gels were scanned under UV light with a densitograph system (Chinetek Scientific, Hong Kong, P.R.China). The band density was semiquantified and nor-malized to Gapdh gene using the Image J software (version, 1.36, National Institutes of Health, Bethesda, MD).

Biochemical analysis. Three whole discs were isolated from each of the nor-mal, degeneration, and treatment groups at 24 weeks after BMSC injection, then the disc tissues were digested overnight at 55 °C in 500 µl papain solu-tion (200 μg/ml papain, 50 mmol/l EDTA, 5 mmol/l cystein). GAG content was analyzed by using microplate reader (SPECTRA MAX 340, Molecular Devices, Sunnyvale, CA) as previously described.12 The DNA content of the discs was measured using Hoechst 33258 method.12 The data was expressed as micrograms of GAG per micrograms of DNA (GAG/DNA).

Statistical analysis. The significance of differences among the means on radiograph measurement and number of GFP or Col2a1/Sox9 positive cells in different groups was tested by analysis of variance and Bonferroni’s multiple comparison test was used as a post hoc test. The gene expres-sion level and GAG/DNA level among different groups was tested by one way analysis of variance and Dunnett’s multiple comparison test was used as a post hoc test. Mann–Whitney test was used to analyze the histo-logical grading scores at different time points. All the statistical tests were performed using the Statview (version 5.0, SPSS, Chicago, IL) program package. A P value <0.05 was determined as statistical significance.

suPPleMentAry MAterIAlTable S1. Definition of the histological grading scale.Table S2. Primers used for RT-PCR analysis.

AcknowledGMentThis work is supported by the CRCG grant from the University of Hong Kong.

reFerences1. Katz, JN (2006). Lumbar disc disorders and low-back pain: socioeconomic factors and

consequences. J Bone Joint Surg Am 88 Suppl 2: 21–24.2. Urban, JP and Roberts, S (2003). Degeneration of the intervertebral disc. Arthritis Res

Ther 5: 120–130.3. Evans, C (2006). Potential biologic therapies for the intervertebral disc. J Bone Joint

Surg Am 88 Suppl 2: 95–98.4. Leung, VY, Chan, D and Cheung, KM (2006). Regeneration of intervertebral disc by

mesenchymal stem cells: potentials, limitations, and future direction. Eur Spine J 15 Suppl 3: S406–S413.

5. Steck, E, Bertram, H, Abel, R, Chen, B, Winter, A and Richter, W (2005). Induction of intervertebral disc-like cells from adult mesenchymal stem cells. Stem Cells 23: 403–411.

6. Yang, SH, Wu, CC, Shih, TT, Sun, YH and Lin, FH (2008). In vitro study on interaction between human nucleus pulposus cells and mesenchymal stem cells through paracrine stimulation. Spine 33: 1951–1957.

7. Richardson, SM, Walker, RV, Parker, S, Rhodes, NP, Hunt, JA, Freemont, AJ et al. (2006). Intervertebral disc cell-mediated mesenchymal stem cell differentiation. Stem Cells 24: 707–716.

8. Sakai, D, Mochida, J, Yamamoto, Y, Nomura, T, Okuma, M, Nishimura, K et al. (2003). Transplantation of mesenchymal stem cells embedded in Atelocollagen gel to the intervertebral disc: a potential therapeutic model for disc degeneration. Biomaterials 24: 3531–3541.

9. Sakai, D, Mochida, J, Iwashina, T, Hiyama, A, Omi, H, Imai, M et al. (2006). Regenerative effects of transplanting mesenchymal stem cells embedded in atelocollagen to the degenerated intervertebral disc. Biomaterials 27: 335–345.

10. Ho, G, Leung, VY, Cheung, KM and Chan, D (2008). Effect of severity of intervertebral disc injury on mesenchymal stem cell-based regeneration. Connect Tissue Res 49: 15–21.

1966 www.moleculartherapy.org vol. 17 no. 11 nov. 2009

© The American Society of Gene & Cell TherapyA Mouse Model Study

11. Hiyama, A, Mochida, J, Iwashina, T, Omi, H, Watanabe, T, Serigano, K et al. (2008). Transplantation of mesenchymal stem cells in a canine disc degeneration model. J Orthop Res 26: 589–600.

12. Sakai, D, Mochida, J, Iwashina, T, Watanabe, T, Nakai, T, Ando, K et al. (2005). Differentiation of mesenchymal stem cells transplanted to a rabbit degenerative disc model: potential and limitations for stem cell therapy in disc regeneration. Spine 30: 2379–2387.

13. Hunter, CJ, Matyas, JR and Duncan, NA (2003). The notochordal cell in the nucleus pulposus: a review in the context of tissue engineering. Tissue Eng 9: 667–677.

14. Yang, F, Leung, VY, Luk, KD, Chan, D and Cheung, KM (2009). Injury-induced sequential transformation of notochordal nucleus pulposus to chondrogenic and fibrocartilaginous phenotype in the mouse. J Pathol 218: 113–121.

15. Novak, A, Guo, C, Yang, W, Nagy, A and Lobe, CG (2000). Z/EG, a double reporter mouse line that expresses enhanced green fluorescent protein upon Cre-mediated excision. Genesis 28: 147–155.

16. Chamberlain, G, Fox, J, Ashton, B and Middleton, J (2007). Concise review: mesenchymal stem cells: their phenotype, differentiation capacity, immunological features, and potential for homing. Stem Cells 25: 2739–2749.

17. Olszewski, WL (2005). Innate immunity processes in organ allografting--their contribution to acute and chronic rejection. Ann Transplant 10: 5–9.

18. Grunhagen, T, Wilde, G, Soukane, DM, Shirazi-Adl, SA and Urban, JP (2006). Nutrient supply and intervertebral disc metabolism. J Bone Joint Surg Am 88 Suppl 2: 30–35.

19. Hunter, CJ, Matyas, JR and Duncan, NA (2004). The functional significance of cell clusters in the notochordal nucleus pulposus: survival and signaling in the canine intervertebral disc. Spine 29: 1099–1104.

20. Aguiar, DJ, Johnson, SL and Oegema, TR (1999). Notochordal cells interact with nucleus pulposus cells: regulation of proteoglycan synthesis. Exp Cell Res 246: 129–137.

21. Lee, JP, Jeyakumar, M, Gonzalez, R, Takahashi, H, Lee, PJ, Baek, RC et al. (2007). Stem cells act through multiple mechanisms to benefit mice with neurodegenerative metabolic disease. Nat Med 13: 439–447.

22. Umeda, M, Kushida, T, Sasai, K, Asada, T, Oe, K, Sakai, D et al. (2009). Activation of rat nucleus pulposus cells by coculture with whole bone marrow cells collected by the perfusion method. J Orthop Res 27: 222–228.

23. Clowry, GJ and Flecknell, PA (2000). The successful use of fentanyl/fluanisone (‘Hypnorm’) as an anaesthetic for intracranial surgery in neonatal rats. Lab Anim 34: 260–264.

24. Makino, S, Fukuda, K, Miyoshi, S, Konishi, F, Kodama, H, Pan, J et al. (1999). Cardiomyocytes can be generated from marrow stromal cells in vitro. J Clin Invest 103: 697–705.

25. Gosset, M, Berenbaum, F, Thirion, S and Jacques, C (2008). Primary culture and phenotyping of murine chondrocytes. Nat Protoc 3: 1253–1260.

26. Masuda, K, Aota, Y, Muehleman, C, Imai, Y, Okuma, M, Thonar, EJ et al. (2005). A novel rabbit model of mild, reproducible disc degeneration by an anulus needle puncture: correlation between the degree of disc injury and radiological and histological appearances of disc degeneration. Spine 30: 5–14.

27. Walsh, AJ, Bradford, DS and Lotz, JC (2004). In vivo growth factor treatment of degenerated intervertebral discs. Spine 29: 156–163.

28. Nomura, T, Mochida, J, Okuma, M, Nishimura, K and Sakabe, K (2001). Nucleus pulposus allograft retards intervertebral disc degeneration. Clin Orthop Relat Res: 94–101.

29. Aszódi, A, Chan, D, Hunziker, E, Bateman, JF and Fässler, R (1998). Collagen II is essential for the removal of the notochord and the formation of intervertebral discs. J Cell Biol 143: 1399–1412.