Content uploaded by Brian Cunningham
Author content
All content in this area was uploaded by Brian Cunningham on Feb 15, 2016
Content may be subject to copyright.
105
TISSUE ENGINEERING
Volume 6, Number 2, 2000
Mary Ann Liebert, Inc.
Original Articles
Silicon Micromachining to Tissue Engineer Branched
Vascular Channels for Liver Fabrication
SATOSHI KAIHARA, M.D.,
1,3
JEFFREY BORENSTEIN, Ph.D.,
2,3
RAHUL KOKA, B.A.,
1,3
SONAL LALAN, B.A.,
1,3
ERIN R. OCHOA, M.D.,
3,4
MICHAEL RAVENS,
1,3
HOMER PIEN, Ph.D.,
2,3
BRIAN CUNNINGHAM, Ph.D.,
2,3
and JOSEPH P. VACANTI, M.D.
1,3
ABSTRACT
To date, many approaches to engineering new tissue have emerged and they have all relied
on vascularization from the host to provide permanent engraftment and mass transfer of
oxygen and nutrients. Although this approach has been useful in many tissues, it has not
been as successful in thick, complex tissues, particularly those comprising the large vital or-
gans such as the liver, kidney, and heart. In this study, we report preliminary results using
micromachining technologies on silicon and Pyrex surfaces to generate complete vascular
systems that may be integrated with engineered tissue before implantation. Using standard
photolithography techniques, trench patterns reminiscent of branched architecture of vas-
cular and capillary networks were etched onto silicon and Pyrex surfaces to serve as tem-
plates. Hepatocytes and endothelial cells were cultured and subsequently lifted as single-cell
monolayers from these two-dimensional molds. Both cell types were viable and proliferative
on these surfaces. In addition, hepatocytes maintained albumin production. The lifted mono-
layers were then folded into compact three-dimensional tissues. Thus, with the use micro-
fabrication technology in tissue engineering, it now seems feasible to consider lifting en-
dothelial cells as branched vascular networks from two-dimensional templates that may
ultimately be combined with layers of parenchymal tissue, such as hepatocytes, to form three-
dimensional conformations of living vascularized tissue for implantation.
INTRODUCTION
T
HE FIEL D OF T IS SU E EN GINE ERING is now maturing and undergoing explosive growth.
1– 4
Virtually every
tissue and organ of the body has been studied. Many tissue-engineering technologies are becoming
1
Department of Surgery, Harvard Medical School and the Massachusetts General Hospital, Boston, Massachusetts.
2
The Charles Stark Draper Laboratories, Cambridge, Massachusetts.
3
The Center For Innovative Minimally Invasive Therapy, Boston, Massachusetts.
4
Department of Pathology, Harvard Medical School and Massachusetts General Hospital, Boston, Massachusetts.
available for human use.
5–13
Over time, several techniques to engineer new living tissue have been stud-
ied. Technologie s include the use of growth factors to stimulate wound repair and regeneration, techniques
of guided tissue regeneration using nonliving matrices to guide new tissue development, cell transplanta-
tion, and cell transplantation on matrices.
14
More recently, new understanding in stem cell biology has led
to studies of populations of primordial cells, stem cells, or embryonic stem cells to use in tissue engineer-
ing approaches.
15,16
To date, all approaches in tissue engineering have relied on the in-growth of blood ves-
sels into tissue-engineered devices to achieve permanent vascularization. This strategy has worked well for
many tissues. However, it falls short for thick, complex tissues such as large vital organs, including liver,
kidney, and heart. Techniques using three-dimensional printing technology to achieve ordered arrays of
channels have been described to begin to solve this problem.
17,18
In parallel to these advances, the rapidly emerging field of MicroElectroMechanical Systems (MEMS)
has penetrated a wide array of applications, in areas as diverse as automotives, inertial guidance and navi-
gation, microoptics, chemical and biological sensing, and, most recently, biomedical engineering.
19,20
Mi-
crofabrication methods for MEMS represent an extension of semiconductor wafer process technology orig-
inally developed for the integrated circuit (IC) industry. Control of features down to the submicron level is
routinely achieved in IC processing of electrical circuit elements; MEMS technology translates this level
of control into mechanical structures at length scales stretching from , 1 m m to . 1 cm. Standard bulk mi-
cromachining enables patterns of arbitrary geometry to be imprinted into wafers using a series of subtrac-
tive etching methods. Three-dimensional structures can be realized by superposition of these process steps
using precise alignment techniques.
Several groups have used these highly precise silicon arrays to control cell behavior and study gene ex-
pression and cell-surface interactions.
21,22,23
However, this approach is essentially a two-dimensional tech-
nology, and it has not been apparent that it might be adapted to the generation of thick, three-dimensional
tissues. We now report early studies using microfabrication in silicon and Pyrex to generate hepatic tissue
and vascular tissue that may be able to be combined to form thick, three-dimensional vascularized tissues.
MATERIALS AND METHODS
Micromachining techniques
Templates for the formation of sheets of living vascularized tissue were fabricated utilizing microma-
chining technology. For the present work, a single-level etch was used to transfer a vascular network pat-
tern into an array of connected trenches in the surface of both silicon and Pyrex wafers.
In this prototype, a simple geometry was selected for patterning the vascular network. Near the edge of
each wafer, a single inlet or outlet was positioned, with a width of 500 m m. After a short length, the inlet
and outlet branched into three smaller channels of width 250 m m; each of these branched again into three
125-m m channels, and finally down to three 50-m m channels. From the 50-m m channels extends the cap-
illary network, which comprises the bulk of the layout. In between these inlet and outlet networks lies a
tiled pattern of diamonds and hexagons forming a capillary bed and filling the entire space between the in-
let and outlet. In one configuration, the capillary width was set at 25 m m, whereas the other capillaries were
fixed at 10 m m. This geometry was selected because of its simplicity as well as its rough approximation to
the size scales of the branching architecture of the liver. Layout of this network was accomplished using
CADENCE software (Cadence, Chelmsford, MA) on a Silicon Graphics workstation. A file with the lay-
out was generated and sent electronically to Align-Rite (Burbank, California), where glass plates with elec-
tron beam-generated patterns replicating the layout geometry were produced and returned for lithographic
processing.
Starting materials for tissue engineering template fabrication were standard semiconductor-grade silicon
wafers (Virginia Semiconductor, Powhatan, Virginia), and standard Pyrex wafers (Bullen Ultrasonics, Eaton,
Ohio) suitable for MEM S processing. Silicon wafers were 100 mm in diameter and 525 microns thick, with
primary and secondary flats cut into the wafers to signal crystal orientation. Crystal orientation was , 100. ,
and wafers were doped with boron to a resistivity of approxim ately 5 W-cm. The front surface was pol-
ished to an optical finish and the back surface ground to a matte finish. Pyrex wafers were of composition
KAIHARA ET AL.
106
identical to Corning 7740 (Corning Glass Works, Corning NY), and were also 100 mm in diameter, but
had a thickness of 775 microns. Both front and back surfaces were polished to an optical finish. Prior to
micromachining, both wafer types were cleaned in a mixture of 1 part H
2
SO
4
to 1 part H
2
O
2
for 20 min at
140°C, rinsed eight times in deionized water with a resistivity of 18 MW, and dried in a stream of hot N
2
gas.
For silicon and Pyrex wafers, standard photolithograp hy was employed as the etch mask for trench for-
mation. Etching of Pyrex wafers requires deposition of an intermediate layer for pattern transfer that is im-
pervious to the etch chemistry. A layer of polysilicon of thickness 0.65 m m over the Pyrex was utilized for
this purpose. This layer was deposited using Low Pressure Chemical Vapor Deposition (LPCVD) at 570°C
and 500 mTorr via the standard silane decomposition method. In the case of silicon, photoresist alone could
withstand limited exposure to two of the three etch chemistries employed. For the third chemistry, a 1.0-
m m layer of silicon dioxide was thermally deposited at 1100°C in hydrogen and oxygen.
Once the wafers were cleaned and prepared for processing, images of the prototype branching architec-
ture were translated onto the wafer surfaces using standard MEMS lithographic techniques. A single layer
of photoresist (Shipley 1822, MicroChem Corp., Newton, MA) was spun onto the wafer surfaces at 4000
rpm, providing a film thickness of approximately 2.4 m m. After baking at 90°C for 30 min, the layer of
photoresist was exposed to UV light using a Karl Suss MA6 (Suss America, Waterbury, VT) mask aligner.
Light was passed through the lithographic plate described earlier, which was in physical contact with the
coated wafer. This method replicates the pattern on the plate to an accuracy of 0.1 m m. Following expo-
sure, wafers were developed in Shipley 319 Developer (MicroChem Corp., Newton, MA), and rinsed and
dried in deionized water. Finally, wafers were baked at 110°C for 30 min to harden the resist, and exposed
to an oxygen plasma with 80 Watts of power for 42 s to remove traces of resist from open areas.
Silicon wafers were etched using three different chemistries, whereas Pyrex wafers were processed us-
ing only one technique. For Pyrex, the lithographic pattern applied to the polysilicon intermediate layer was
transferred using a brief (
,
1 min) exposure to SF
6
in a reactive ion-etching plasma system (Surface Tech-
nology Systems [STS] Newport, UK). Photoresist was removed, and the pattern imprinted into the poly-
silicon layer was transferred into trenches in the silicon using a mixture of two parts HNO
3
to one part HF
at room temperature. With an etch rate of 1.7 microns per minute, 20-micron-deep trenches were etched
into the Pyrex wafers in approximately 12 min. Because the chemistry is isotropic, as the trenches are etched
they become wider. Processing with the layout pattern with 25-m m-wide capillary trenches tended to result
in merging of the channels, while the use of 10-m m-wide trenches avoided this phenomenon. Interfero-
metric analysis of the channels after etching showed that surface roughness was less than 0.25 m m. Once
channel etching of Pyrex wafers was completed, polysilicon was removed with a mixture of 10 parts HNO
3
to one part HF at room temperature, and wafers were recleaned in one part H
2
SO
4
to one part HF.
Three different chemistries were employed to etch silicon and investigate the interaction between chan-
nel geometry and cell behavior. First, a standard anisotropic plasma etch chemistry, using a mixture of SF
6
and C
4
F
8
in a switched process plasma system from STS,
24
was used to produce rectangular trenches in
silicon. An example of this process is shown in Fig. 1; note that narrower trenches are shallower than deep
trenches due to a phenomenon know n as RIE lag. A second process utilized a different plasma system from
STS, which produces isotropic trenches with a U-shaped profile, shown schematically in Fig. 2. Although
the process is isotropic, widening of the trenches is not as severe as is experienced in the isotropic Pyrex
etching process described earlier. In both of these plasma-etching cases, trenches were etched to a nominal
depth of 20 m m. For the third process, anisotropic etching in KOH (45% wt/wt in H
2
O at 88°C), the in-
termediate silicon dioxide layer mentioned above was employed. First, the silicon dioxide layer was pat-
terned using HF etching at room temperature. The KOH process produces angled sidewalls rather than the
rectangular profile or U-shaped profile produced by the first two recipes, respectively. Crystal planes in the
, 111. orientation are revealed along the angled sidewalls, due to anisotropic properties of the KOH etch
process as a function of crystal orientation.
25
Due to the self-limiting nature of the channels produced by
this process, trench depth was limited to 10 m m. After completion of the silicon wafer etching, all layers
of photoresist and silicon dioxide were removed, and wafers were cleaned in one part H
2
SO
4
to one part
H
2
O
2
at 140°C, followed by rinsing in deionized water and drying in nitrogen gas.
For this set of experiments, no attempt was made to alter the surface chemistry of the silicon and Pyrex
MICROFABRICATION IN TISSUE ENGINEERING
107
wafers. Prior to processing, silicon wafers were uniformly hydrophobic, whereas Pyrex wafers were equally
hydrophilic, as determined by observations of liquid sheeting and sessile drop formation. After processing,
unetched surfaces appeared to retain these characteristics, but the surface chemistry within the channels was
not determined.
Animals
Adult male Lewis rats (Charles River Laboratories, Wilmington, MA), weighing 150–200 g, were used
as cell donors. Animals were housed in the Animal Facility of Massachusetts General Hospital in accor-
dance with National Institutes of Health (NIH) guide lines for the care of laboratory animals. They were
allowed rat chow and water
ad libitum
and maintained in 12-h light and dark cycle.
Cell isolations
Male Lewis rats were used as hepatic cell donors. Hepatocytes were isolated using a modified two-step
collagenase perfusion procedure as described in previous reports.
26,27
Briefly, the animals were anesthetized
with nembutal sodium solution (Abbott Laboratories, North Chicago, IL), 50 mg/kg, and the abdomen was
prepared in sterile fashion. A midline abdominal incision was made and the infrahepatic inferior vena cava
was cannulated with a 16-gauge angiocatheter (Becton Dickinson). The portal vein was incised to allow
retrograde efflux and the suprahepatic inferior vena cava was ligated. The perfusion was perform ed at a
KAIHARA ET AL.
108
FIG. 2. Test structure pattern etched using inductively-couple d (IPC) system described in the text.
FIG. 1. Process for fabricating U-shaped trenches in silicon wafers.
flow rate of 29 mL/min initially with a calcium-free buffer solution for 5–6 min, then with a buffer con-
taining collagenase type 2 (Worthington Biomedical Corp., Freehold, NJ) at 37°C. The liver was excised
after adequate digestion of the extracellular matrix and mechanically agitated in William’s E medium (Sigma,
St. Louis, MO) with supplements to produce a single-cell suspension. The suspension was filtered through
a 300-m m mesh and separated into two fractions by centrifugation at 50 g for 2 min at 4°C. The pellet con-
taining the viable hepatocyte fraction was resuspended in William’s E medium and further purified by an
isodensity Percoll centrifugation. The resulting pellet was then resuspended in hepatocyte growth medium,
and cell counts and viabilities of HCs were determined using the trypan blue exclusion test.
The endothelial cells were derived from rat lung microvessels and they were purchased directly from the
vendor (Vascular Endothelial Cell Technologie s, Rensellaer, NY).
Hepatocyte culture medium
William’s E medium supplemented with 1g of sodium pyruvate (Sigma, St. Louis, MO) and 1% gluta-
mine-penicillin-streptomycin (Gibco BRL, Gaithersburg, MD) were used during the cell isolation process.
The plating medium was Dulbecco’s modified Eagle medium (DMEM; Gibco BRL) supplemented with
10% fetal bovine serum (FBS), 1% penicillin-streptomycin, 44 mM sodium-bic arbonate, 20 mM HEPES,
10 mM niacinamide, 30 m g/mL L-proline, 1 mM ascorbic acid-2-phospate, 0.1 m M dexamethasone (Sigma),
insulin-transferrin-sodium selenite (5 mg/L-5 mg/L-5 m g/L, Roche Molecular Biomedicals, Indianapolis,
IN), and 20 ng/mL epidermal growth factor (Collaborative Biomedical Products, Bedford, MA).
Endothelial cell culture medium
We used a modified version of a previously used endothelial cell culture medium.
28
DMEM (Gibco BRL)
was supplemented with 10% FBS, 1% penicillin-streptomycin, 25 mg of ascorbic acid (Sigma), 10 mg of
L-alanine (Sigma), 25 mg of L -proline (Sigm a), 1.5 m g of cupric sulfate (Sigma), glycine (Sigma), and 1 M
HEPES buffer solution (Gibco BRL). The media was supplemented with 8 mg of ascorbic acid every day.
Cell attachment and lifting from nonetched silicon and Pyrex wafers
Silicon and Pyrex were both tested as possible substrates for the culture and lifting of endothelial cells
and hepatocytes. Prior to cell seeding, the Pyrex wafers were sterilized with 70% ethanol (Fisher, Pittsburg,
PA) overnight and washed three times with sterile phosphate-buffered saline (PBS; Gibco BRL). Silicon
wafers were first soaked in acetone for 1 h, followed a methanol rinse for 15 min, and overnight steriliza-
tion in 100% isopropyl alcohol. Rat lung microvascular endothelial cells was cultured on noncoated Pyrex
and silicon surfaces, as well as wafers coated with vitrogen (30 m g/mL), Matrigel® (1%), or Gelatin (10
mg/mL). Once isolated, the cells were resuspended in endothelial cell culture medium, seeded uniform ly
onto the wafer at a density of 26.7 3 10
3
cells/cm,
2
and cultured at 5% CO
2
and 37°C. After reaching con-
fluence, we tested the ability for the monolayer of endothelial cells to lift from the wafers using a cell scrap-
per to promote detachment.
The rat hepatocytes were also cultured on noncoated Pyrex and silicon, as well as wafers coated with a
thin and thick layers of vitrogen (30 m g/mL and 3 m g/mL) and Matrigel (1%) to determine the optimal
methods for lifting hepatocyte sheets. Once isolated, the hepatocytes were resuspended in hepatocyte growth
media, seeded onto the wafer at a density of 111.3 3 10
3
cells/cm
2
, and cultured at 5% CO
2
and 37°C. Cell
attachment and growth was observed daily using microscopy and cell lifting occurred spontaneously.
After determining which method for culturing was best for lifting the hepatocytes and endothelial cells
in an intact layer, both membranes were fixed in 10% buffered formalin for 1 h and harvested for histo-
logical study. The hepatocytes were stained immunohistochemically.
Immunohistochemical staining
The hepatocyte cell monolayer membrane was fixed in 10% buffered formalin and processed for hema-
toxylin-eosin and immunohistoch emical staining using a labeled streptavidin biotin method (LSAB2 kit for
rat specimen, DAKO, Carpinteria, CA). The primary antibody was rabbit anti-albumin (ICN, Costa Mesa,
MICROFABRICATION IN TISSUE ENGINEERING
109
CA). Three-micron sections were prepared and deparaffinized. The specimens were treated with per-
oxidase blocking buffer (DAKO) to prevent the nonspecific staining. Sections were stained with albu-
min diluted with PBS, follow ed by biotinyla ted anti-rabbit antibody and horseradish peroxida se (HRP)-
conjugated streptavidin. Sections were treated with diaminobenzidine (DAB) as substrate and were
counterstained with hematoxylin .
Albumin production
To assess hepatocyte function, albumin concentration in the culture medium was measured every 24 h
for 5 days pre-cell detachment using an enzyme-linked immunosorbent assay (ELISA) (
n
5 5).
29
In brief,
a 96-well microplate was coated with anti-rat albumin antibody (ICN). After blocking nonspecific responses
with a 1% gelatin solution, each sample was seeded onto the plate and incubated for 1 h. This was followed
by another 1-h incubation with peroxidase conjugated anti-rat albumin antibody (ICN). Finally, the sub-
strate was added and extinction was measured with a microplate reader at 410 nm.
R
2
of the standard curve
was . 0.99.
Statistical analysis
All data were expressed as mean 6 SD. Statistical analysis was performed with a paired
t
-test. When the
p
value of each test was less than 0.05, we judged it to be statistically significant.
Cell attachment to etched silicon and Pyrex wafers
Endothelial cells and hepatocytes were also seeded onto etched silicon and Pyrex wafers. Prior to cell
seeding, the Pyrex wafers were sterilized with 70% ethanol (Fisher) overnight and washed three times with
sterile PBS (Gibco BRL). Silicon wafers were first soaked in acetone for 1 h, followed a methanol rinse
for 15 min, and overnight sterilization in 100% isopropyl alcohol. Onto these wafers were seeded rat lung
microvascular endothelial cells at a density of 26.7 3 10
3
cells/cm
2
, or rat hepatocytes at a density of 111.3 3
10
3
cells/cm
2
. These cells were cultured at 5% CO
2
and 37°C, and their attachment and growth observed
daily using microscopy.
Implantation of hepatocyte sheets into the rat omentum
Hepatocytes were cultured on silicon wafers coated with a thin layer of vitrogen (30 m g/ml), and lifted
in sheets. Retrorsine is a drug known to inhibit the regeneration of the normal liver by producing a block
in the hepatocyte cell cycle with an accumulation of cells in late S and/or G
2
phase.
30
This drug was ad-
ministered into the peritoneal cavity of two rats at a dose of 3 mg/mL per 100 g on day 0, and after 2 weeks.
Three weeks later, a portacaval shunt was created, and the following week a hepatocyte sheet, lifted after
4 days culture on vitrogen-coated silicon (30 m g/mL), was implanted onto the microvasculature of the rat
omentum and rolled into a three-dimensional cylinder, and a 60% hepatectomy was performed. The rolled
omentum with hepatocytes was harvested at 4 weeks and at 3 months after implantation and analyzed us-
ing histology.
RESULTS
Micromachining
A schematic of the vascular branching network design used as a template for micromachining is shown
in Fig. 3A. This pattern was transferred to silicon and Pyrex wafers using the processes described in the
Materials and Methods section. Typical trench depths of 20 microns on silicon and 10 microns on glass
were achieved utilizing these processes. An optical micrograph of a portion of the capillary network etched
into a silicon wafer is shown in Fig. 3B. In Fig. 3C, a scanning electron micrograph cross section of an an-
gled trench etched using the anisotropic etching process described earlier is shown. This process resulted
in excellent adhesion and enhanced lifting of living tissue.
KAIHARA ET AL.
110
Growth and lifting of cells from unpatterned silicon and Pyrex wafers
We studied the adhesion and growth of endothelial cells and hepatocytes on several different substrate
surfaces. On all Pyrex wafers, coated or noncoated, the endothelial cells proliferated and grew to conflu-
ence within 4 days. These cells did not lift spontaneously, and, when scraped, did not lift as a single sheet.
In addition, when the noncoated silicon wafers were seeded with endothelial cells, the cell sheet fragmented
MICROFABRICATION IN TISSUE ENGINEERING
111
FIG. 3. (A) Vascular branching network pattern used for silicon and Pyrex wafer micromachinin g. (B) Optical mi-
crograph or portion of capillary network etched into silicon wafer using the process shown in Fig. 1. (C) Scanning elec-
ton micrograph of anisotropic etching process used to form angled sidewall trenches (N. Gerrish and J. Ricker, Draper
Laboratory).
C
A
B
upon lifting. On the other hand, endothelial cells seeded onto silicon surfaces coated with vitrogen (30
m g/mL), Matrigel (1%), and gelatin (10 mg/mL) did lift with a cell scraper, and provided an intact mono-
layer sheet of endothelial cells. Upon microscopic observation, there were no significant differences in the
effects of the three coatings on the detached cell sheets.
Hepatocytes also attached and spread well on all coated and noncoated Pyrex wafers, and did not lift
spontaneously or in sheets when scraped after several days of growth. However, when seeded onto silicon
wafers, they lifted spontaneously on all the noncoated and coated wafers (Fig. 4). The hepatocyte sheets
lifted from the noncoated wafers after 3 days, but were very fragile and fragmented easily. The monolay-
ers that lifted from the thin and thickly coated vitrogen substrates (30 m g/mL and 3 m g/mL) lifted after 4
days in culture to form an intact hepatocyte layer. Cells lifted from the Matrigel coated (1%) silicon wafers
after 5 days in culture. There were no significant differences in appearance between the cell sheets lifted
from the vitrogen and Matrigel-coated wafers.
Histological assessment of the detached cell monolayers of both hepatocytes and endothelial cells man-
ifested prom ising results. Hemotoxylin and eosin (H&E) staining of both showed that all cells were viable
and that most were undergoing mitoses. The endothelial cells were observed to be primarily attenuated and
to form a single-celled alignment (Fig. 5A). The monolayer of hepatocytes was of a spheroid configuration
with eosinophilic floculent cytoplasm and a large nucleus with a bright red nucleolus, similar to that seen
in the native liver. Moreover, cellular attachments were less attenuated than the endothelial cells (Fig. 5B).
Thus, these results are reminiscent of each of the cell types’ specific functions. In biological systems, the
endothelium functions as a barrier with a thin, smooth outer surface and as a transport channel and so it is
understandable that these cells are observed here to be primarily attenuated and in a single-celled array.
The hepatocytes have more of a tendency to form tissue and so we see less of a single-celled array and
more of a rounded multi-layered array.
KAIHARA ET AL.
112
FIG. 4. Hepatocyte sheet lifted after 4 days in culture from a nonetched silicon surface coated with dilute vitrogen,
30
m
g/mL (Magnification: 3 8).
Values for albumin secretion into the hepatocyte culture medium at days 2, 3, 4, and 5 were 165.96 6
29.87, 164.44 6 17.22, 154.33 6 18.46, and 115.47 6 18.09 (m g/day, Graph 1), respectively. Although
there was a statistical significant difference between days 4 and 5, no significant differences were observed
between days 2, 3, and 4 (
p
, 0.05 by the paired
t
-test). Hence, this data shows the cells cultured on the
silicon wafer were able to maintain a fairly constant rate of albumin production until day 4.
Moreover, through immunohistoch emical staining of the detached hepatocyte monolayers, many cells
were stained positive for albumin, indicating further that hepatocyte function was maintained while on the
silicon wafers.
Cell growth on etched surfaces of silicon and Pyrex wafers
On all Pyrex and silicon-etched wafers, coated and noncoated, the endothelial cells proliferated and grew
to confluence within 4 days inside and outside the channels. These cells did not lift spontaneously, but they
did cover the walls of the channels. Hepatocytes also attached and spread on the etched Pyrex wafers, coated
and noncoated, covering the walls of the channels. On the silicon wafers, the hepatocytes attached and
spread, but did not lift spontaneously as with the nonetched wafers on the noncoated or vitrogen-coated (30
m g/mL) wafers (Fig. 6).
Implantation of hepatocyte sheet into the rat omentum
H&E staining of hepatocyte sheets implanted into rat omentum demonstrated that all cells were viable
and showed proliferation at 4 weeks and 3 months. The implanted hepatocyte monolayer sheets, when har-
vested, were over five cell layers thick in most areas (Fig. 7).
DISCUSSION
This study demonstrates that silicon microfabrication technology can be used to form large sheets of liv-
ing tissue. It also demonstrates the feasibility of etching ordered branching arrays of channels that allow
living endothelial cells to line the luminal surface of the channels. In addition, we have shown that orga-
nized sheets of engineered hepatocyte tissue and endothelial tissue can be lifted from the surface of silicon
or Pyrex wafers and can be folded into a compact three-dimensional configuration. We have then taken the
hepatocyte sheets and placed them into rats on the highly vascular surface of the omentum. That structure
has then been rolled into a three-dimensional cylinder as a model for an engineered vasculature. We show
the formation of vascularized hepatic tissue as a permanent graft. These steps are preliminary first results
MICROFABRICATION IN TISSUE ENGINEERING
113
FIG. 5. (A) H&E staining for detached monolayers of hepatocytes. (B) H&E staining for detached monolayers of en-
dothelial cells.
A B
toward a novel approach to the problem of generating complex, vascularized, thick tissues using the prin-
ciples of tissue engineering.
The field of tissue engineering is now maturing as both a scientific and a clinical discipline. The
concept of replacing living tissue with living tissue designed and then fabricated is less than 30 years
old, although tumor cells were implanted into animals encased in a polymer membrane in 1933.
31
To
date, all engineered tissue devices have been built as systems without a blood supply and have relied
on vascularization from the host to provide permanent engraftment and mass transfer of oxyge n and
nutrients. These systems range from nonliving implants as forms of guided regeneration devices, to
devices composed of living cells combined with some form of scaffolding.
14
This funda mental ap-
proach has shown broad utility in many different tissues of the body. However, an avascular approach
to large vascular organs such as the liver, kidne y, or heart has substantial shortcomings and has not
been reliable in animal models. Because of this difficulty, we began to study possible approaches to
provide a complete vascular system to the engineered structure before implantation. To date, three-di-
mensional printin g has been used as a technolog y to provide an ordered array of branching channe ls
to a polym er before cell seeding.
17 ,3 2
This report is based on a novel concept that involves making a complete branching vascular circula-
tion in two dimensions on the surface of silicon using the tools of microfabrication, and then lifting it
from the silicon mold and folding or rolling it into a compact three-dimensional structure. The study
demonstrates feasibility in several of the important components. Microfabrication technology has been
used in important studie s in cell and developmental biology to understand complex biologic signaling
events occurring at the cell membrane-surface interface.
23
It has also been used in tissue engineering to
guide cell behavior and the formation of small units of tissue.
21
To our knowledge, this is the first re-
port of adapting the technology to make a coherent structure over a broad range of scale. The channels
begin as a single channel with a diameter of 500 m m, branch through four generations following a geo-
metric scaling law that halves the channel width for each successive generation, form an array of cap-
illary channels 10 microns in diameter, and then sequentially branch back to a single outflow vein. We
have demonstrated that not only can we form the channels in silicon and Pyre x, but that living en-
dothelial cells will line the channels. In other experiments, we have demonstrated that cells on surfaces
of silicon and Pyrex will lay down matrix and form sheets of tissue of the cell type of origin, either he-
patic or endothelial. We have also demonstrated the ability to peel these sheets from the surface and
form three-dimensional units of tissue. In effect, the wafer of silicon or Pyrex has acted as a mold for
the formation of tissue.
Microfabrication technology has been adapted to suit the needs of forming living tissue. Its pote n-
tial power lies in its control of form over extremely small distances. The resolution is on the order of
0.1 microns from point to point. This level of precision adds new levels of control in our ability to de-
sign and guide new tissue formation. For instance, surfaces can be imprinted with submicron grooves
or scallops, and corners can be made rounde d, angled, or sharp with this same level of submicron pre-
cision. Geometric control at this scale can have a powerful impact on cell adhesion through mecha-
nisms such as contact guidance.
33
The funda mental barrier to applic ation has been that the technology
for the purpose of tissue engine ering was limited to very thin structures because designs were con-
strained to the surface of the silicon wafers. By using the wafer as a temporary mold and lifting the
tissue from it, we have overcome this limitation and folded the tissue into three-dimensional space. It
now seems feasible to consider lifting the channels formed from endothelial cells and combining them
with layers of parenchym al tissue such as hepatocytes. After combining and folding into three dimen-
sions, if flow can be initiated, vascularized tissue will have been fabricated. Those studies are currently
underway. These preliminary studies can be expanded to studies that include specific surface chem-
istry alterations that can help with cell adhesion, cell proliferation, and matrix production. Advances
in polym er chemistry can aid in the mechanical tasks of lifting and folding as well as the biologic tasks
of adhesion and gene expression. Microfabrication appears to be a powerful new tool for the tissue en-
gineer to not only unde rstand the biology of tissue development but also to aid in the fabrication of
large tissue and organ structures specifically designed for human therapy.
MICROFABRICATION IN TISSUE ENGINEERING
115
ACKNOWLEDGMENTS
This work was supported by a generous grant from The Center For Innovative Minimally Invasive Ther-
apy and the Department of Defense Grant #:DAMD17-99-2-9001. The authors gratefully acknowledge Ms.
Miranda Kelly for her assistance in the preparation of the manuscript, and the students and technical staff
of the Draper Laboratories who helped in the design and fabrication of the silicon and Pyrex wafers.
REFERENCES
1. Vacanti, J.P., and Langer, R. Tissue engineering: the design and fabrication of living replacement devices for sur-
gical reconstruction and transplantation. Lancet 354, 32, 1999.
2. Langer, R., and Vacanti, J. Tissue engineering. Science 260, 920, 1993.
3. Rennie, J. ed. Special report: The promise of tissue engineering. Sci. Am. 280, 37, 1999.
4. Lysaght, M.J., Nguy, N.A.P., and Sullivan, K. An econom ic survey of the emerging tissue engineering industry.
Tiss. Eng. 4, 231, 1998.
5. Bell, E., Ehrlich, P., Buttle, D.J., and Nakatsuji, T. Living tissue formed in vitro and accepted as skin-equivalent
of full thickness. Science 221, 1052, 1981.
6. Burke, J.F., Yannas, I.V., Quimby, W.C., Bondoc, C.C., and Jung, W.K. Successful use of a physiologically ac-
ceptable artificial skin in the treatm ent of extensive burn injury. Ann. Surg. 194, 413, 1981.
7. Compton, C., Gill, J., Bradford, D., Regauer, S., Gallico, G., and O’Connor, N. Skin regenerated from cultured ep-
ithelial autografts on full-thickness burn wounds from 6 days to 5 years after grafting. Lab. Invest. 60, 600, 1989.
8. Parenteau, N., Nolte, C., Bilbo, P., Rosenburg, M., Wilkins, L., Johnson, E., Watson, S., Mason, V., and Bell, E.
Epidermis generated in vitro: practical considerations and applications. J. Cell. Biochem. 45, 24, 1991.
9. Parenteau, N., Sabolinski, M., Prosky, S, Nolte, C., Oleson, M., and Kriwet, K. Biological and physical factors in
influencing the successful engraftment of a cultured human skin substitute. Biotechnol. Bioeng. 52, 3, 1996.
10. Purdue, G.F., Hunt, J.L., et al. A multicenter clinical trial of a biosynthetic skin replacement, Dermagraft-T C, com-
pared with cryopreserved human cadaver skin for temporary coverage of excised burn wounds. J. Burn Care Re-
hab. 18, 52, 1997.
11. Hansbrough, J.F., and Franco, E.S. Skin replacem ents. Clin. Plastic Surg. 25, 407, 1998.
12. Vacanti, C.A., Cima, L.G., Ratkowski, D., Upton, J., and Vacanti, J.P. Tissue engineered growth of new cartilage
in the shape of a human ear using synthetic polymers seeded with chondrocytes. In: Cima LG, Ron ES, eds. Tis-
sue-inducing biomaterials, Materials Research Society Symposium Proceedings, Pittsburgh: Materials Research So-
ciety 252, 367, 1992.
13. Isogai, N., Landis, W., Kim, T.H., Gerstenfeld, L.C., Upton, J., and Vacanti, J.P. Tissue engineering of a phalangeal
joint for application in reconstructive hand surgery. J. Bone Joint Surg. Am. 81, 306, 1999.
14. Lanza, R., Langer, R., Chick, W. eds. Principles of Tissue Engineering. Austin, TX: Academ ic Press, 1997.
15. Pedersen, R. Embryonic stem cells for medicine. Sci. Am. 44, 68, 1999.
16. Thomson, J.A., Itskovitz-Eldor, J., Shapiro, S.S., et al. Embryonic stem cell lines derived from human blastocysts.
Science 282, 1145, 1998.
17. Griffith, L.G., Wu, B., Cima, M.J., Powers, M.J., Chaignaud, B., and Vacanti, J.P. In vitro organogenesis of liver
tissue. Ann. NY Acad. Sci. 831, 382, 1997.
18. Langer, R., and Vacanti, J.P. Tissue engineering: the challenges ahead. Sci. Am. 280, 62, 1999.
19. McWhorter, P.J., Frazier, A.B., and Rai-Choudhury , P. Micromachining and trends for the twenty-first century. In:
P. Rai-Choudhur y, ed. Handbook of Microlithography, Micromachini ng and Microfabrication, Bellingham, WA:
SPIE Press, 1997.
20. Kourepenis, A., Borenstein, J., Connelly, J., Elliott, R., Ward, P., and Weinberg, M. Perform ance of MEMS iner-
tial sensors. Proc. AIAA GN&C Conference, Boston, MA, 1998.
21. Griffith, L., Powers, M., and Tannenbaum , S. Microfabricated structures for control of 3D co-cultures of liver cells.
Ann. Biomed. Eng. 26, 1998.
22. Folch, A., and Toner, M. Cellular micropatterns on biocom patible materials. Biotechnology Progr. 14, 388, 1998.
23. Kane, R.S., Takayama, S., Ostuni, E., Ingber, D.E., and Whitesides, G.M. Patterning proteins and cells using soft
lithography. Biomaterials 20, 2363, 1999.
24. Ayon, A.A., Chen, K.S., Lohner, K.A., Spearing, S.M., Sawin, H.H., and Schmidt, M.A. Deep reactive ion etch-
ing of silicon. Mat. Res. Soc. Symp. Proc. 546, 51, 1999.
KAIHARA ET AL.
116
25. Kendall, D.L., Malloy, K.J., and Fledderm ann, C.B. Critical technologies for the micromachinin g of silicon. In:
K.T. Faber and K.J. Malloy, eds. Semiconductor s and Metals. New York: Academic Press, 1992.
26. Aiken, J., Cima, L., Schloo, B., et al. Studies in rat liver perfusion for optimal harvest of hepatocytes. J. Pediatr.
Surg. 25, 140, 1990.
27. Seglen, P.O. Preparation of isolated rat liver cells. Methods Cell Biol. 13, 29, 1976.
28. Niklason, L.E., Gao, J., Abbott, W.M., et al. Functional arteries grow n in vivo. Science 284, 489, 1999.
29. Schwere, B., Bach, M., and Bernheimer, H. ELISA for determination of albumin in the nanogram range: assay in
cerebrospinal fluid and comparison with radial immunodiffusion. Clin. Chem. Acta 163, 237, 1987.
30. Peterson, J.E. Effects of the pyrrolizidine alkaloid lasiocarpine-N-o xide on nuclear and cell division in the liver of
rats. J. Pathol. Bacteriol. 89, 153, 1965.
31. Bisceglie, V. Uber die antineoplastische immunitat; heterologe Einpflanzung von Tumoren in Huhner-embry onen.
Ztschr f Krebsforsch 40, 122, 1933.
32. Kim, S.S., Utsunomiya, H., Koski, J.A., Wu, B.M., Cima, M.J., Sohn, J., Mukai, K., Griffith, L.G., and Vacanti,
J.P. Survival and function of hepatocytes on a novel three dimensional synthetic biodegradable polym er scaffold
with an intrinsic network of channels. Ann. Surg. 228, 8, 1998.
33. Den Braber, E.T., de Ruijter, J.E., Ginsel, L.A., von Recum, A.F., and Jansen, J.A. Orientation of ECM protein
deposition, fibroblast cytoskeleton, and attachment complex com ponents on silicon microgrooved surfaces, J. Bio-
med. Mater. Res. 40, 291, 1998.
Address correspondence to:
Joseph P. Vacanti, M.D.
John Homans Professor of Surgery
Harvard Medical School and Massachusetts General Hospital
55 Fruit Street, Warren 1157
Boston, MA 02114
E-mail:
jvacanti@partners.org
MICROFABRICATION IN TISSUE ENGINEERING
117
