Contact with Existing Adipose Tissue Is Inductive

for Adipogenesis in Matrigel

JOHN L. KELLY, M.B., B.Ch., B.A.O.,1,2* MICHAEL W. FINDLAY, M.B.B.S., B.Sc.,1,2*KENNETHR.KNIGHT, Ph.D.,1,2 ANTHONYPENINGTON,M.B.B.S.,1,2 ERIKW.THOMPSON, Ph.D.,1,2

AURORA MESSINA, Ph.D.,1,2 and WAYNE A. MORRISON, M.B.B.S., M.D.1,2

ABSTRACT

The effect of adipose tissue on inductive adipogenesis within Matrigel (BD Biosciences) was assessed byusing a murine chamber model containing a vascular pedicle. Three-chamber configurations that variedin the access to an adipose tissue source were used, including sealed- and open-chamber groups that hadno access and limited access, respectively, to the surrounding adipose tissue, and a sealed-chamber groupin which adipose tissue was placed as an autograft. All groups showed neovascularization, but varied inthe amount of adipogenesis seen in direct relation to their access to preexisting adipose tissue: openchambers showed strong adipogenesis, whereas the sealed chambers had little or no adipose tissue;adipogenesis was restored in the autograft chamber group that contained 2- to 5-mg fat autografts. Theseshowed significantly more adipogenesis than the sealed chambers with no autograft ( p< 0.01). Autograftswith 1mg of fat were capable of producing adipogenesis but did so less consistently than the largerautografts. These findings have important implications for adipose tissue engineering strategies and forunderstanding de novo production of adipose tissue.

INTRODUCTION

ADIPOSE TISSUE GRAFTS are widely used in reconstructive

and cosmetic surgery. For reconstruction of large de-

fects after cancer or trauma, such as breast reconstruction

after mastectomy, larger volumes of adipose tissue are re-

quired than can be provided by simple fat grafting. There

has, therefore, recently been increasing interest in the pos-

sibility of creating new fat tissue by tissue engineering

techniques.1–11

Kawaguchi et al.1 demonstrated that if Matrigel, a base-

ment membrane–rich extracellular matrix derived from a

mouse sarcoma,12 is injected subcutaneously in mice with

the proangiogenic growth factor basic fibroblast growth

factor (bFGF), the gel will be replaced by mature adipose

tissue over 3–10 weeks. If bFGF is not added, adipose tissue

does not form; as a result, the authors suggested a critical

role for angiogenesis in the process of adipose tissue for-

mation. Subsequent studies have suggested that the me-

chanism for the new tissue formation is inductive, with the

surrounding tissues having an important role in providing

stem cells for the de novo adipogenesis.13 We recently de-

scribed an in vivo murine model of tissue engineering for

vascularized adipose tissue.2 This involves a subcutaneously

implanted 5-mm-long, 0.05-mL-volume silicone chamber

that ensleeves the superficial epigastric vessels. This induces

spontaneous angiogenesis and adipogenesis, and the result-

ing product is based on a vascular pedicle.

The development of a human equivalent to Matrigel for

adipose tissue engineering requires a better understanding of

the mechanisms involved in such inductive adipose tissue

engineering, particularly the influence of neovascularization

and the role of the surrounding tissue. To date, the relative

roles of the surrounding tissue and neovascularization in

1Bernard O’Brien Institute of Microsurgery, Melbourne, Australia.2Department of Surgery, University of Melbourne, St. Vincent’s Hospital, Melbourne, Australia.

*These authors contributed equally to the work.

TISSUE ENGINEERINGVolume 12, Number 7, 2006# Mary Ann Liebert, Inc.

2041

inductive adipogenesis remain unclear. Our murine ‘‘cham-

ber model,’’ comprising a perivascular silicone tubing

chamber and modified from the method first described by

Cronin et al.,2 is ideal for investigating these factors because

it provides an isolated environment for such studies. Silicone

was retained as the chamber material for several reasons,

including its proven use in the existing model, its ready

availability, its pliable handling, its minimal tissue reaction,

and its established use in other animal models for purposes

such as nerve entubulation repair.14

METHODS

Chamber insertion and subcutaneous Matrigel plugs

Tissue engineering chambers were inserted bilaterally in a

method modified from Cronin et al.,2 as shown in Fig. 1.

Eight-week-old male C57/Bl6 mice were anesthetized with

4% intraperitoneal chloral hydrate (Merial, Paramatta, Aus-

tralia) and given 5 mg/kg of Carprofen (Pfizer, West Ryde,

Australia) subcutaneously for analgesia. The groin regions

were depilated. A longitudinal incision was made in the

groin, and the superficial epigastric vessels were isolated,

then stripped meticulously of adipose tissue for 7–8 mm and

wrapped in a 5-mm length of 3.35-mm-internal-diameter

silicone tubing (Dow Corning, Midland, MI), which had

been slit along its length. The tubing was allowed to close

around the vessel and was anchored to adjacent muscle

fascia with a 10/0 nylon suture (Davis & Geck, St. Louis,

MO). One end was sealed with bone wax (Ethicon, Somer-

ville, NJ), taking care not to occlude the emerging vessels,

and the chamber was filled with growth factor reduced

Matrigel (BD Biosciences, Bedford, MA) containing 1 mg/

mL of human recombinant bFGF (Sigma-Aldrich, Sydney,

Australia) and 80 U/mL of unfractionated sodium heparin

(APP, Schaumburg, IL). The steps to complete the chamber

insertion were then varied from this point to give 3 different

chamber configurations (Fig. 2) and a no-chamber control.

There was a minimum of 5 animals in each group. In group 1

the second chamber end was also sealed with wax, except

where the vascular pedicle exited the chamber, forming a

completely sealed chamber. In group 2 the chamber was

partially sealed with a wax plug with a 1-mm defect forming

an ‘‘open’’ chamber. This opening abutted the surrounding

epigastric fat pad, providing some contact with adipose tis-

sue. Group 3 was identical to group 1 except that a piece of

autologous adipose tissue from the wound bed (epididymal

fat pad) was inserted into the chamber as a free graft before

sealing, as in group 1 (Fig. 1B). Two different amounts of fat

were inserted: in group 3A, 2–5 mg was used (approximately

4–10% of the chamber volume), and in group 3B, 1 mg was

used (mean weight, 1.03 mg� 0.05 mg [SE], approximately

2% of the chamber volume). In group 4, 300 mL of the same

Matrigel mixture as used in the chamber groups was injected

subcutaneously in the flank of the animal as a no-chamber

control. The skin was closed with 9-mm Auto clips (Becton

Dickinson, Sparks, MD).

The St. Vincent’s Hospital institutional ethics committee

approved all experiments in advance, and animal manage-

ment complied with the Australian Code of Practice for the

care and use of animals for scientific purposes.

Chamber removal and assessment

All animals were anesthetized without recovery at 6

weeks after chamber insertion, and the chambers were ex-

posed surgically (Fig. 1C). Vessel patency within the cham-

bers was assessed macroscopically by looking for bleeding

from the cut ends of the pedicle as it exited either end of the

FIG. 1. (A) Chamber insertion showing depilated right groin with

silicone chamber (short arrow) around superficial inferior epigastric

pedicle (long arrow) before application of bone wax as chamber

sealant. (B) two- to 5-mg adipose tissue autograft (long arrow)

within chamber with Matrigel and the vascular pedicle (short arrow)

running through it. (C) Chamber depicted end-on at harvest, with

new tissue (long arrow) supplied by the vascular pedicle exiting the

chamber (short arrow). (D) Transillumination of a chamber speci-

men showing neovascularization in specimen (arrow). (Color

images available online at www.liebertpub.com/ten.)

FIG. 2. Diagrams of chamber configurations. (A) Sealed. (B)

Open. (C) Sealed with adipose tissue autograft (arrow). 1, wax

plug; 2, superficial inferior epigastric pedicle; 3, silicone chamber.

2042 KELLY ET AL.

chamber. Contents of the chambers and subcutaneous plugs

were removed and weighed, and the volume was determined

by weighing the tissue while suspending it in a 0.9% sodium

chloride solution (normal saline) at room temperature.15

Tissue was routinely fixed in 10% neutral buffered formalin

(Australian Biostain, Traralgon, Australia) and then em-

bedded in paraffin. Serial histologic sections were stained

routinely with hematoxylin and eosin and Masson’s tri-

chrome. The percentages of adipose tissue, residual Matrigel,

fibrous granulation tissue, and vessels were determined by

point-counting every 20th 5-mm section of each specimen by

using digital video imaging ( JVC, TK C1480E) and an au-

tomated, systematic point-counting system (CAST system,

Olympus, Albertslund, Denmark). Only living adipose tis-

sue was counted; any nonviable fat, including graft remnants

where seen, was not included in the count. Nonviable fat was

distinguished from viable fat morphologically because it

displays decreased intensity of staining, loss of nucleus,

breakdown of cell membrane integrity, interstitial mono-

nuclear cell infiltration, and formation of lipid cysts. Lipid

cysts have a typical appearance of a space surrounded by a

cyst wall containing multiple fibroblastoid cells with or

without an inflammatory infiltrate.

It was not possible to distinguish between engrafted and

newly produced fat in all specimens; therefore, percentage

and volume of total adipose tissue were determined and

compared with the volume of adipose tissue at insertion.

Vessels were identified morphologically by their typical

endothelial lining and by the presence of erythrocytes within

their lumen. Results from such vessel counting have corre-

lated well with point-counting of vessels labeled by using

methods, such as lectin staining or India ink perfusion, ap-

plied in previous studies (results not shown). The proportion

of each tissue type was then multiplied by the total volume of

the chamber specimens to determine the volume of each

tissue. Shrinkage from histologic processing was assumed to

be constant across the 4 tissue types.

Statistical analysis

Mean tissue percentages and volumes were compared for

statistical significance by using 1-way analysis of variance

with post hoc analysis, with Dunnett’s test as appropriate

(GraphPad Prism Software, San Diego, CA). Statistical

significance was set at p< 0.05.

RESULTS

Macroscopic appearance and vessel patency

The vessel patency rate of all chambers was over 90%

and independent of chamber type. Those chambers in

which the vascular pedicle had thrombosed were excluded

from the study; however, no tissue grew in these chambers

regardless of experimental group. Comparison of the spe-

cimens from the different groups, using transillumination

and low magnification (�12.5), showed 2 distinct patterns.

Vascularization of Matrigel occurred from the vascular

pedicle and extended to the periphery of the specimen,

spreading out over the surface of the specimen in all cases

(Fig. 1D). Where fat had been left in contact with Matrigel,

the specimens were opaque. Because Matrigel is transpar-

ent, opacity signified tissue/adipogenic replacement.

Histology

The same patterns of neovascularization as seen macro-

scopically were observed microscopically in all chamber

groups, with new vessels branching from the main pedicle

and extending through Matrigel to the periphery of the

specimen (Fig. 3). Most specimens had a fine fibrous cap-

sule containing additional vessels. Fundamental differences

were evident in the relative proportions of Matrigel and

adipose tissue, and the distribution of these, in the different

chamber configurations.

Sealed-chamber group. There was the same new vessel

growth as seen in the other groups, and fibroblastoid cells,

but few adipocytes were evenly distributed within the

Matrigel (Fig. 3A). Any adipocytes seen were usually close

to the vascular pedicle and may be due to residual adipose

tissue left on the vascular pedicle at chamber insertion.

Autograft chamber group. These showed adipocytes

evenly distributed throughout the Matrigel (Fig. 3C and D).

Occasionally, small clusters of lipid cysts with typical

multicellular, dense connective tissue cyst walls, with or

without a surrounding inflammatory infiltrate, were present

at the site of the original autograft. Adipogenesis within the

1-mg autograft chamber group showed 2 patterns. Ap-

proximately 60% of chambers showed good adipogenesis

with the chamber specimens being indistinguishable from

the 2- to 5-mg autograft chamber specimens. The remainder

showed little or no adipogenesis despite the autograft rem-

nants being visible, similar to the sealed-chamber specimens

without an autograft.

Open-chamber group. There was marked adipogenesis

within Matrigel in proximity to the open end of the cham-

ber, and little or no adipogenesis seen at the opposite end of

the chamber, where the chamber was completely sealed

(Fig. 3B).

Matrigel plug group. Subcutaneous injection of 300 mL

of growth factor reduced Matrigel, together with 1 mg/mL

of bFGF (Fig. 4) and heparin, resulted in angiogenic in-

vasion of the plug and almost complete replacement by fat

by 6 weeks, similar to that reported by Kawaguchi et al.1

These specimens formed a no-chamber positive control of

in vivo adipogenesis.

Findings of the point-counting are summarized in Figs. 5

and 6. The chambers seeded with adipose tissue grafts had

ADIPOGENESIS IN MATRIGEL 2043

significantly more adipose tissue than the sealed- or open-

chamber groups (Fig. 5). The difference in the percentage of

adipose tissue between chamber groups was maximal and

most consistent when a 2- to 5-mg autograft was included in

the chamber ( p< 0.01 compared with sealed chamber). The

1-mg autograft group showed greater variability in percen-

tage of adipose tissue, with most specimens having a per-

centage of adipose tissue similar to that of the 2- to 5-mg

autograft group and others having very little adipogenesis.

Because of this variability in response, no statistically sig-

nificant difference in the mean adipose tissue percentage or

volume was detected when we compared the 1-mg autograft

group to the sealed-chamber specimens. Sealed chambers,

with minimal or no access to adipose tissue, contained the

least adipose tissue (<5%). Open chambers with intermediate

access to preexisting adipose tissue developed intermedi-

ate amounts of adipose tissue, but this was not statis-

tically significant when compared with the sealed-chamber

specimens.

Vessels extended from the vascular pedicle to the per-

iphery of the specimens in all groups. Neovascularization

was similar across all chamber configurations, making up

10% of the chamber volume.

Granulation tissue was also measured. This was pre-

dominantly found at entrance and exit points for the pedicle

and around the periphery of the specimen, where it formed a

fine fibrous capsule. The volume of fibrous tissue produced

did not differ between the groups (approximately 20–25% of

the chamber volume in each case).

A significant residual volume of Matrigel was still evident

at 6 weeks, although this was less in the specimens in which

significant adipose tissue had formed ( p< 0.01 when com-

paring the sealed-chamber and 2- to 5-mg autograft groups).

DISCUSSION

The development of an injectable substance for human

use that induces its own replacement by the de novo growth

of adipose tissue would be of great benefit to many patients

across multiple medical specialties. Such inductive adipo-

genesis has the benefit that it does not require the addi-

tion of stem cells harvested from the patient. Matrigel is a

murine prototype of such a substance. When injected sub-

cutaneously in mice, it induces angiogenesis followed by

adipogenesis until its volume is replaced by adipose tissue.1

FIG. 3. Histologic characteristics of representative chamber

specimens by chamber group (stained with hematoxylin and eosin).

Section of whole specimen on left, higher-power view of re-

presentative area on right. (A) Sealed-chamber specimen showing

good neovascularization but few adipocytes. (B) Open chamber

with open end to left, where strong adipogenesis is seen. (C) two- to

5-mg autograft chamber specimen showing a strong adipogenic

response throughout the specimen. (D) one-mg autograft chamber

specimen showing a strong adipogenic response throughout.

FIG. 4. Transverse section of subcutaneous Matrigel plug

(�100) showing replacement of Matrigel (light blue) with adipose

tissue (Masson’s trichrome stain). (Color images available online

at www.liebertpub.com/ten.)

2044 KELLY ET AL.

Matrigel is derived from a sarcoma in mice and therefore

cannot be used for tissue engineering in humans.12 It is

hoped that a better understanding of the mechanism by

which adipogenesis occurs within Matrigel will facilitate

the development of an analog suitable for human use.

The mechanism whereby Matrigel induces de novo adi-

pogenesis remains uncertain. Previous research has fo-

cussed on the role of Matrigel when combined with bFGF

in promoting angiogenesis, which, in turn, appears to sup-

port adipogenesis, possibly via recruitment of precursors

from the surrounding tissues.13 The potential role of an-

giogenesis in facilitating adipose tissue production has been

studied in vitro by the co-culture of endothelial cells, adi-

pocytes, and preadipocytes in various combinations,16–18

but to date it has not been possible to isolate each of these

components in vivo.

In a simple mouse model of tissue engineering we have

examined the interaction between existing vessels and tis-

sue in the inductive formation of new adipose tissue within

Matrigel. We have shown that Matrigel with bFGF, when

excluded from an adipose tissue source, will become vas-

cularized but will not develop significant adipose tissue.

When a free graft of adipose tissue is added to the chamber,

or when one end of the chamber is open for access to

Adipose

05

101520253035404550556065

Matrigel Fibrous Vasc/Ped

Tissue Type

Per

cent

age

Tiss

ueby

Poi

nt C

ount

ing

Percentage Tissue by Chamber Type

Sealed

Open

2–5-mg Autograft

1-mg Autograft

FIG. 5. Tissue percentage by chamber type–adipose tissue percentages compared with analysis of variance; the differences were

significant [F(3,22) ¼ 76.9; p¼ 0.003] when compared with the sealed-chamber group. On the post hoc test, only the 2- to 5-mg

autograft group differed significantly from the sealed-chamber group ( p< 0.01).

Adipose Matrigel Fibrous Vasc/Ped

Tissue Type

Vol

ume

Dis

plac

emen

t (µL

)

Tissue Volume by Chamber Type

Sealed

Open

2–5-mg Autograft

1-mg Autograft

0

1

2

3

4

5

6

7

FIG. 6. Tissue volume by chamber type–adipose tissue volumes compared with analysis of variance; the differences were significant

[F(3,22) ¼ 5.9; p¼ 0.0057] when compared with the sealed-chamber group. On the post hoc test, only the 2- to 5-mg autograft group

differed significantly from the sealed-chamber group ( p< 0.01).

ADIPOGENESIS IN MATRIGEL 2045

surrounding fat, new adipose tissue forms within the Ma-

trigel adjacent to this preexisting tissue. The degree of

volume change in the autograft specimens could not be

explained by adipocyte hypertrophy because adipocyte size

in the chamber specimens at harvest was similar to the

graft adipocyte size at insertion.

It is evident in our in vivo model that a suitable matrix, a

patent vascular pedicle, and bFGF are not sufficient by

themselves to generate fat. A tissue source in addition to a

vascular pedicle is essential for the inductive growth of new

adipose tissue. Such definition has not been possible in

subcutaneous models but is afforded by the isolative nature

of our sealed chamber. We found that adipose tissue ap-

peared to grow best in close proximity to an existing or

grafted fat source. Toriyama et al.,13 using the subcutaneous

Matrigel model in which fat abuts the Matrigel plug, ob-

served by electron microscopy that multiple layers of cells at

the periphery differentiate into adipocytes or endothelial

cells, and progressively invade the Matrigel. They observed

that the new fat derives at the point where existing fat con-

tacts the Matrigel, consistent with our observations. The

absence of significant adipogenesis in our sealed-chamber

model supports the assertion that access to surrounding tis-

sues is important for inductive adipogenesis within Matrigel.

No statistically significant differences were detected in

the vascularity of the chamber specimens across the differ-

ent chamber configurations. This finding suggests that an-

giogenesis in Matrigel only requires access to an existing

vasculature or that the influence of existing tissue (e.g., the

autograft) on angiogenesis was negligible. The development

of neovascularization throughout Matrigel within sealed

chambers did not itself induce an adipogenic response. The

few adipocytes seen in the sealed-chamber specimens may

be residual adipocytes left on the vascular pedicle at

chamber insertion. Alternatively, they may represent adi-

pogenesis produced by recruitment of mesenchymal stem

cells via the circulation in the absence of preexisting tissue.

Unlike neovascularization, neoadipogenesis in Matrigel /

bFGF requires more than access to an existing vasculature.

The factor or factors provided by existing fat that promote

neoadipogenesis are not yet determined. Mesenchymal stem

cells have been identified in adipose tissue,19 and thus the

adipose tissue in or near our chambers may be acting as a

source of preadipocytes. Stem cells make up only a very

small percentage of the cells within each tissue,20 and this is

consistent with the variable adipogenic response found with

inclusion of a 1-mg autograft within the chamber when

compared with the more consistent adipogenesis seen with

2- to 5-mg autografts. If stem cells were the major con-

tribution of the preexisting adipose tissue (either host or

grafted) in our system, this would suggest that angiogenesis

in a sealed chamber leads to minimal or no recruitment of

mesenchymal stem cells via the circulation. Also, recent

studies in our laboratory using human adipose tissue xeno-

grafts show that the new adipose tissue is predominantly

murine, suggesting that the graft is inductive of preadipocyte

recruitment rather than providing stem cells directly.21

Other factors, such as the release of extracellular matrix

components,15,16 paracrine factors,17 or chemokines from

the autograft, may also play a role in fostering the neoadi-

pogenesis by stimulating adipogenesis from precursors re-

cruited from the circulation. Experiments designed to isolate

each of these pathways to examine their role in tissue-

induced adipogenesis are in progress.

Increasing evidence suggests close interactions between

angiogenesis and adipogenesis. The precise nature of these

interactions remains unclear. A paracrine-based interaction

between endothelial cells and preadipocytes has been

shown in vivo to involve vascular endothelial growth factor

(VEGF) and VEGF receptor-2 signaling within the en-

dothelial cells.22 Aoki et al.18 showed that co-culture of

endothelial cells and mature adipocytes (in vitro) actively

promotes immature preadipocyte development. These pre-

adipocytes appeared predominantly at the adhesion sites

between the endothelial and mature adipocytes. This effect

was not reproduced by endothelial cell–conditioned media

or by certain endothelial cell–produced cytokines, sug-

gesting that endothelial cells are involved in the enlarge-

ment mechanism of adipose tissue mass through their direct

adhesion to mature adipocytes. Antiangiogenesic agents

have been shown to cause vascular remodeling and reduce

adipose tissue mass in murine models of obesity in a dose-

dependent fashion, confirming a regulation of adipose tis-

sue by its vasculature.23 Hypoxia has been shown to inhibit

adipogenesis via the hypoxia-inducible factor-1–regulated

gene DEC1/Stra13, which causes repression of peroxisome

proliferator-activated receptor-g.24 These findings are con-

sistent with our observation that no fat formed when the

epigastric vessel pedicle was thrombosed, but the absence

of adipogenesis in the sealed chamber, despite considerable

angiogenesis, indicates that angiogenesis is necessary but

not sufficient in the Matrigel/bFGF environment to cause

neoadipogenesis.

The current clinical strategies for the use of adipose

tissue in surgical reconstruction and augmentation rely on

engraftment and revascularization of the adipose tissue

graft, rather than on de novo adipogenesis. Our data in-

dicate an important inductive role of preexisting adipose

tissue, whether vital or isolated, in generating new adipose

tissue with Matrigel with bFGF. Greater understanding of

the mechanism of adipogenesis in response to Matrigel in

mice takes us closer to developing a human equivalent for

human use and has important implications for chamber

design for the production of larger volumes of tissue.

ACKNOWLEDGMENTS

We gratefully acknowledge the surgical assistance from

Sue McKay, Liliana Pepe, Anna Deftereos, and Amanda

Rixon from the Experimental Medical Surgical Unit,

St. Vincent’s Hospital Melbourne. Grant support from the

2046 KELLY ET AL.

National Health and Medical Research Council and Na-

tional Australia Bank is gratefully acknowledged.

REFERENCES

1. Kawaguchi, N., Toriyama, K., Nicodemou-Lena, E., Inou, K.,

Torii, S., and Kitagawa, Y. De novo adipogenesis in mice

at the site of injection of basement membrane and basic fi-

broblast growth factor. Proc Natl Acad Sci U S A 95, 1062,

1998.

2. Cronin, K.J., Messina, A., Knight, K.R., Cooper-White, J.J.,

Stevens, G.W., Penington, A.J., and Morrison, W.A. New

murine model of spontaneous autologous tissue engineering,

combining an arteriovenous pedicle with matrix materials.

Plast Reconstr Surg 113, 260, 2004.

3. Green, H., and Kehinde, O. Formation of normally differ-

entiated subcutaneous fat pads by an established preadipose

cell line. J Cell Physiol 101, 169, 1979.

4. Mandrup, S., Loftus, T.M., MacDougald, O.A., Kuhajda, F.P.,

and Lane, M.D. Obese gene expression at in vivo levels by fat

pads derived from s.c. implanted 3T3-F442A preadipocytes.

Proc Natl Acad Sci U S A 94, 4242, 1997.

5. Patrick, C.W., Chauvin, P.B., Hobley, J., and Reece, G.P.

Preadipocyte seeded PLGA scaffolds for adipose tissue en-

gineering. Tissue Eng 5, 139, 1999.

6. von Heimburg, D., Zachariah, S., Low, A., and Pallua, N.

Influence of different biodegradable carriers on the in vivo

behavior of human adipose precursor cells. Plast Reconstr

Surg 108, 411, 2001.

7. von Heimburg, D., Zachariah, S., Heschel, I., Kuhling, H.,

Schoof, H., Hafemann, B., and Pallua, N. Human pre-

adipocytes seeded on freeze-dried collagen scaffolds in-

vestigated in vitro and in vivo. Biomaterials 22, 429, 2001.

8. Kimura, Y., Ozeki, M., Inamoto, T., and Tabata, Y. Adipose

tissue engineering based on human preadipocytes combined

with gelatin microspheres containing basic fibroblast growth

factor. Biomaterials 24, 2513, 2003.

9. Patrick, C.W., Jr., Zheng, B., Johnston, C., and Reece, G.P.

Long-term implantation of preadipocyte-seeded PLGA scaf-

folds. Tissue Eng 8, 283, 2002.

10. Selye, H., Jean, P., Cantain, M., and Lemire, Y. Induction of

adipose tissue development by mechanical means. Plast Re-

constr Surg 24, 250, 1959.

11. Xaymardan, M., Gibbins, J., and Zoellner, H. Adipogenic

healing in adult mice by implantation of hollow devices in

muscle. Anat Rec 267, 28, 2002.

12. Kleinman, H.K., McGarvey, M.L., Liotta, L.A., Robey, P.G.,

Tryggvason, K., and Martin, G.R. Isolation and character-

ization of type IV procollagen, laminin, and heparan sulfate

proteoglycan from the EHS sarcoma. Biochemistry 21, 6188,

1982.

13. Toriyama, K., Kawaguchi, N., Kitoh, J., Tajima, R., Inou, K.,

Kitagawa, Y., and Torii, S. Endogenous adipocyte precursor

cells for regenerative soft-tissue engineering. Tissue Eng 8,

157, 2002.

14. Odaka, M., Uchiyama, Y., Oka, Y., and Tamaki, T. Muscle

Evaluation of morphological and functional regeneration of

rat nerve-muscle units after temporary and permanent tubu-

lization. Nerve 28, 194, 2003.

15. Weibel, E.R. Stereological Methods, Vol. 1: Practical Meth-

ods for Biological Morphometry. London, Academic Press,

1979.

16. Shillabeer, G., Forden, J.M., and Lau, D.C.W. Induction of

preadipocyte differentiation by mature fat cells in the rat. J Clin

Invest 84, 381, 1989.

17. Varzaneh, F.E., Shillabeer, G., and Wong, K.L., Lau, D.C.W.

Extracellular matrix components secreted by microvascular

endothelial cells stimulate preadipocyte differentiation in vi-

tro. Metabolism 43, 906, 1994.

18. Aoki, S., Toda, S., Sakemi, T., and Sugihara, H. Coculture of

endothelial cells and mature adipocytes actively promotes

immature preadipocyte development in vitro. Cell Struct

Funct 28, 55, 2003.

19. Zuk, P.A., Zhu, M., Ashjian, P., De Ugarte, D.A., Huang, J.I.,

Mizuno, H., Alfonso, Z.C., Fraser, J.K., Benhaim, P., and

Hedrick, M.H. Human adipose tissue is a source of multi-

potent stem cells. Mol Biol Cell 13, 4279, 2002.

20. Asakura, A., and Rudnicki, M.A. Side population cells from

diverse adult tissues are capable of in vitro hematopoietic

differentiation. Exp Hematol 30, 1339, 2002.

21. Stillaert, F., Findlay, M.W., Palmer, J., Cheang, S., Messina,

A., Abberton, K., Morrison, W.A., and Thompson, E.W.

(Manuscript in preparation).

22. Fukumura, D., Ushiyama, A., Duda, D.G., Xu, L., Tam, J.,

Krishna, V., Chatterjee, K., Garkavtsev, I., and Jain, R.K.

Paracrine regulation of angiogenesis and adipocyte differ-

entiation during in vivo adipogenesis. Circ Res 93, e88, 2003.

23. Rupnick, M.A., Panigrahy, D., Zhang, C.Y., Dallabrida, S.M.,

Lowell, B.B., Langer, R., and Folkman, M.J. Adipose tissue

mass can be regulated through the vasculature. Proc Natl

Acad Sci U S A 99, 10730, 2002.

24. Yun, Z., Maecker, H.L., Johnson, R.S., and Giaccia, A.J.

Inhibition of PPAR gamma 2 gene expression by the HIF-1-

regulated gene DEC1/Stra13: a mechanism for regulation of

adipogenesis by hypoxia. Dev Cell 2, 31, 2002.

Address reprint requests to:

Wayne A. Morrison, M.D.

Hugh Devine Professor of Surgery

University of Melbourne Department of Surgery

St. Vincent’s Hospital

Melbourne, Australia 3065

E-mail: morriswa@svhm.org.au

ADIPOGENESIS IN MATRIGEL 2047