This document summarizes Victoria Virador's presentation on engineering functional 3-D tissue models. It discusses using 3D tissue models to improve drug development and testing. Some key points include: (1) The drug development pipeline is slow, taking 9+ years and $1-2 billion; (2) 3D tissue models, like spheroids and organoids, can better model human physiology compared to traditional 2D cell cultures; (3) These models are being used for disease modeling, predictive toxicology screening, and angiogenesis research to aid drug development. The document provides several examples of 3D tissue models, including for skin, cancer, and validating model responses to toxic insults.

1. Engineering functional 3-D
tissue models. October
28-29, 2013
Victoria M. Virador, Ph.D.

2. Adapted from Anna Barker, Bioconference Live, 2013
Drug
Discovery
Pre-clinical
Clinical trial
FDA
5000-10000
compounds
250 compounds
5 compounds
1 compound
Phase I
Phase II
Phase III
3-6 years
1 yr
2 yr
3 yr
0.5-2 years
20-100
volunteers
100-500
volunteers
1000-5000
volunteers
It is estimated that it takes 9 years and costs between 0.8 and
1.7 billion dollars to bring a drug through clinical trials.
The translational drug pipeline is slow

3. DesRochers, et al (2013) PLoS ONE, 8 (3), art. no. e59219, .
Cell-based assays are expanding into the realm
of tissue analysis. Three-dimensional (3-D)
micro-organoid systems will play an increasing
role in drug testing and therapeutics over the
next decade.
I will highlight recent breakthroughs in tissue
bioengineering aimed at enhancing the success
of drug candidates through preclinical
optimization.
The translational drug pipeline is slow:
Biomarkers, assay models

4. Building a 3D model of murine skin
3D models and HTS
Models of normal tissues
Models of cancer
The challenge of angiogenesis
Validating 3D models
Conclusions

5. Building a 3D model of murine skin (1)
Bellas et al, Macromolecular Bioscience, 2012
Collagen gel-based co-culture models including cancer cells and fibroblasts.
Che et al BBRC, 2006
Human skin
reconstructs
Murine skin
reconstructs
Epidermis formed by
keratinocytes
Dermis formed by
collagen embeded
fibroblasts

6. Wt hair follicle buds
EGFR (+/-) hair follicle buds
EGFR (-/-) hair follicle buds
Model of EGFR -/- mice hair follicle: Genetic deletion of the epidermal growth factor receptor
in mice results in early postnatal death and hypoplastic epidermis.
Dermal equivalent
consisting of
collagen with
embedded Swiss
3T3 J2 fibroblasts,
raft cultures
(Cheng et al, (1995)
Genes Dev. 9, 2335–
2349
Wt 15d skin
EGFR (-/-) 15d skin
Building a 3D model of murine skin (2)
EGFR -/- mice

7. Why is it that we could not make murine
skin reconstructs?
Kimlin and Virador, in Skin Stem Cells, Methods and Protocols, 2012
G. Casagrande

8. Find skin epidermal progenitor
population
Crigler et al, Faseb J, 2007
Total epidermal
preparation
Epi
EPIDERMIS
300 rpm
Dermal
Hair
Follicles
(DHF)
DERMIS
1400 rpm
Dermal
fibroblasts
(fD)
1000 rpm
Dermal
fibroblasts
(fE)
A combination of gradient centrifugation and selective adhesion to
separate skin subpopulations, then put them in simple model

9. Epi and fE looked alike
under the same culture
conditions
Epidermal cells alone did not survive,
but a mixture with dermal (fE) did
Fraction A with dermis, air
exposure with Hi Ca
Fraction E, air exposure
with 0.24mM Ca2+
On dermis (MU)

10. So fE may contain a skin
progenitor population
We hypothesized that skin
progenitor populations would
form an epithelium under
conditions requiring wound
healing.
Our very simple model for
wound healing consisted of a
succession of submerged and
air exposed conditions in
which progenitor cells would
differentiate and expand
epidermal precursors.
submerged
Air-exposed
Air-exposed on dermis
Gamma irradiated fb
on underside

11. fE did not require underlying dermis
and expanded epidermal precursors
K10/K14
K10/K14
K14
K14
2air/4sub/7air
2 air
2 air
6air
9 air
Mu
LoCa+KGF, 1 week
submerged, then air
exposed
Involucrin/K14

12. Characterization of fractions
Fraction B Fraction E
0.7%
Fraction D
0.4%
UV blue
UVred
UV blue UV blue
UVred
UVred
3%
CD49f-FITC
CD34-PE
CD34-PE
Fraction A
Fraction E
% CD117=2.7
CD49f-FITC
% CD117=6.2
A. Side population was not a specific marker to distinguish
dermal sub-fractions
B. CD117+ dermal melanocyte precursors, not from epidermis
B. Taylor

13. Fractions % K10 % K14 % Pou3f2
Epi 88 33 93
DHF 12 48 7
D 0 8 0
E 0 11 0
Relative levels in fractions
p63-epidermal stem
cells
K15-hair follicle
bulge
Epidermal precursors in fE are
not from the hair follicle bulge

14. MRP8 uniquely differentiates skin
subpopulations
Epi DHF D E
10%
90%
Skin
MRP8 (S100 family) is found in endothelial cells and
keratinocytes

15. In vivo engraftment demonstrates fE
progenitor potential
A
B
Silicon graft
Lichti U, Anders J, Yuspa SH.
Nat Protoc. 2008;3(5):799-810.
Subcutaneous
injection

16. Epi
E
DHF
D
Epi
DHF
D
E
Center of graft
Center of graft
FM
stain
In silicon chamber grafts, conducive to epidermal
differentiation, melanocytes not associated with hair follicles,
were abundant in the dermis
In vivo engraftment demonstrates fE
progenitor potential-epidermal

17. desmin
Dermal fE gave rise to structures that stained positive for
smooth muscle actin and desmin.
Epi
DHF
desmin
SMA
fE
1 week
2 week
FM stain
H&E
H&E
H&E
E
In vivo engraftment demonstrates fE
progenitor potential-subcutaneous
D

18. Conclusions
Using 3-dimensional cultures of murine skin under stress
conditions in which only reserve epithelial cells are
expected to survive and expand, we demonstrate that a
mesenchymal population resident in neonatal murine
dermis has the unique potential to develop an epidermis in
vitro.
The multipotential cells can be isolated from neonatal murine
dermis by differential centrifugation/adhesion.
Results suggest that progenitors capable of epidermal
differentiation exist in the mesenchymal compartment of
an abundant tissue source and may have a function in
mesenchymal-epithelial transition upon insult.
Crigler et al, Faseb J, 2007

19. Building a 3D model of murine skin
3D models and HTS
Models of normal tissues
Models of cancer
The challenge of angiogenesis
Validating 3D models
Conclusions

20. 3D models with potential for high
throughput screening
1. Spheroids
Organoids-tissue slices
2. Cell sheet stacking
3. Lithography models

21. 1. Spheroids
Lee, G. Y., et al. (2007). Nat
Methods 4(4): 359-65
Clusters of cells suspended in medium in order to mimic in
vivo tissues including extracellular matrices.

22. Spheroids in orthotopic models
3D Spheroids
Spheroid models for drug screening
Spheroids in Bioreactors
Spheroids/organoids embeded in
synthetic biomaterials/scaffolds
Kimlin et al, Molecular carcinogenesis 2013;52(3):167-82

23. Self organizing 3D tissues: bioink
Bioprinted organoids are continuous biological structures
formed by similar mechanisms to those of early
morphogenesis in which tissue- or organ-specific ECM is
developed, and with it biomechanical and biochemical
conditions compatible with implantation.
Tasoglu S, et al., Trends Biotechnol
2013
http://
www.explainingthefuture.com/
bioprinting.html
Example of fabrication of
bioprinted tissue

24. ü Basic parameters such as cell viability and spheroid
volume can be automatically analyzed.
u Mechanistic assays are not readily available.
ü Control of spheroid size can be used in anticancer
drug screening.
ü Spheroids can be used to predict gradients of
oxygen that determine cellular responses.
u Assay development and analysis of large quantities
of data are two important challenges for the future
of spheroid HTS.
Advantages and disadvantages of
spheroid cultures for drug screening
Fennema et al., Trends in Biotechnology
February 2013

25. Spheroids vs organoids vs tissue slices
for proof of concept and validation
Organoids generated
in vivo from single-
cell suspensions of
primary human
mammary cells
Eirew et al, Nature Medicine 2008
Zhao et al., Am J Pathol. 2010
Tissue slice from
fresh primary
prostatic
adenocarcinomas
grafted under
renal capsule of
immunodeficient
mice.
Precision-cut tissue
slices as a tool to
predict metabolism
of novel drugs.
Graaf IA, et al.
Expert Opin Drug
Metab Toxicol. 2007
a
b
c

26. Building a 3D model of murine skin
3D models and HTS
Models of normal tissues
Models of cancer
The challenge of angiogenesis
Validating 3D models
Conclusions

27. Functional 3-D tissues by stacking cell
sheets in vitro
Haraguchi et al, Nature Protocols, 2012

28. Successes and challenges in tissue
regeneration
Cultured autologous oral mucosal epithelial
cell sheet transplantation for the treatment
of corneal deficiency
Organ-specific scaffolds for in vitro
expansion, differentiation, and
organization of primary lung cells
Anterior cruciate ligament
regeneration using mesenchymal
stem cells and silk scaffold (pig)
Vascularized and functional human liver from
an iPSC-derived organ bud transplant
Burillon et al, Investigative
Ophthalmology & Visual Science, 2012
Shamis et al TISSUE ENGINEERING:
Part C., 2011
Fan et al, Biomaterials 30 (2009)
Takebe et al, doi:10.1038/nature12271

29. Miniaturized tissue models should be useful for drug testing
if appropriate phenotypic assays are developed.
Limitations to cell sheet engineering:
• the thickness of viable tissue is limited by the time period
for in vivo vascular maturation and its connection with
host blood vessels.
• New techniques for media-perfusable lumens.
Tissue models should be cross-validated in vivo or in organ-
tissue slices.
NORMAL TISSUES FOR
TRANSPLANTATION (AND HTS?)

30. Building a 3D model of murine skin
3D models and HTS
Models of normal tissues
Models of cancer
The challenge of angiogenesis
Validating 3D models
Conclusions

31. Human Ovarian
3D models of cancers
Normal murine lung cells are
dissociated, placed on top of Gelfoam
sponge with mammary tumor cell line
R221A-GFP in a tissue culture dish
containing media (a). Confocal image of
GFP and brightfield overlay. Sponge
matrix (white arrow), R221A-GFP cells
(black arrow)
Murine Lung
Martin et al, Clin Exp Metastasis, 2008
3D model of human
omentum:
• primary human
fibroblasts and
mesothelial cells from
human omentum.
Fibroblasts embedded
in ECM.
• confluent layer of
primary human
omental mesothelial
cells plated on top.
• Labeled ovarian
cancer cells or normal
ovarian surface
epithelial cells added
Kenny et al, International J. of Cancer, 2007

32. ü Which cancer processes.
ü Cancer in which tissue.
ü Decide relevant cells and 3D format.
ü Use the appropriate cell lines. Use cocultures.
ü Preclinical testing of newly established models in parallel
with traditional 2D cultures.
ü Primary cells from patients if the model is relevant
In vitro 3D models of cancer
Kimlin et al, Molecular carcinogenesis 2013

33. Building a 3D model of murine skin
3D models and HTS
Models of normal tissues
Models of cancer
The challenge of angiogenesis
Validating 3D models
Conclusions

34. Angiogenesis and vasculogenesis
Angiogenesis is required for
cancer development and
metastasis because tumors cannot
enlarge beyond 1 to 2 millimeters
in diameter unless they are
vascularized.
Vascular architecture is needed
for stable grafts and for survival
of fabricated microtissues.
Knowledge of angiogenesis
signaling is translatable to
vasculogenesis.

35. 3D responsive angiogenic implanted network (rain)-droplet assay
Angiogenesis in 3D models
Zeitlin et al, Lab Invest 2012

36. Building a 3D model of murine skin
3D models and HTS
Models of normal tissues
Models of cancer
The challenge of angiogenesis
Validating 3D models
Conclusions

37. VALIDATING 3D MODELS:
Cells exposed to cytotoxic insult
respond in various ways:
• If the insult is lethal, cells undergo
necrosis or other pathways of cell
death, such as apoptosis or
autophagy.
• Cells exposed to a sublethal insult
may stop actively growing and
dividing (a decrease in cell
proliferation).
Responses can be measured individually
or in multiplex
• induction of superoxide.
• depletion of glutathione.
• secrease of mitochondrial membrane
potential.
• reduction in overall viability.
Bioluminescence marker
Mehta, et al., J. Control. Release
(2012)
Viability and toxicity assays

38. The kidney is a major site of drug–induced toxicity.
• About 7% of drug candidates fail due to nephrotoxicity in pre-clinical
testing;
• 30–50% cases of severe acute renal failure in patients due to drug–induced
nephrotoxicity.
• 3D kidney models include mouse models for proof of concept such as
Xinaris et al, J Am Soc Nephrol. 2012, and human models with
immortalized cell lines such as DesRochers et al, PlosOne, 2013
Liver toxicity is a common source of drug withdrawals.
• Fey and colleagues produced 3D spheroids using immortalized human
hepatocyte line. (Fey et al, Toxicol Sci 2012)
Skin models can be used to assess toxicity.
• Canton and colleagues modeled paracrine interactions between
keratinocytes and fibroblasts using SDS and Formi as irritants at subtoxic
concentrations. NFkB activity in fibroblast detects inflammatory stress
linked to keratinocyte-triggered activation (Canton et al, Biotechnol
Bioeng, 2010)
Viability and toxicity- tissue models

39. Verbridge, 2010
Viability and toxicity are necessary
but not sufficient for a valid model
Tissue engineering provides
tools to recreate cell-
microenvironment
interactions in vitro

40. Drug uptake and diffusion. Ability
to modulate and measure the
increase or decrease of a known
property of the tissue with a drug
such as decrease endothelial
sprouting.
Intra and extravasation in a
model of metastasis in a chip
Zeitlin et al, Lab Invest 2012
Shin et al, Lab Chip 2011
Examples of functional assays

41. Functional Endpoints based on tissue morphology

42. K. Kim et al. / Biomaterials 33 (2012) 1406e1413
Albumin secretion and urea synthesis in
culture medium at the indicated time points.
Co-cultures are useful for modeling interaction and signaling between
different cell types
Functional Endpoints based on tissue physiology
Virador, unpublished
Human melanoma and carcinoma
lines on inert agar plugs -
communication between cell types
demonstrated by transfer of
melanosomes to keratinocytes
Monolayer hepatocytes
hepatocytes stratified
with endothelial cell
sheet
hepatocytes stratified
with endothelial cell
sheet
Monolayer hepatocytes

43. Lei et al, Anal. Biochem, 2002
Standardization

44. ü Many tissue-mimicking methods hold potential to their
specific area of study
ü Optimizing for each system requires deemphasizing
universality of application
ü Validation, standardization, collaboration
In vitro 3D models hold great promise
for clinical testing
Kunz-Schughart, 2004 Journal of Biomolecular Screening
Kunz-Schughart, 2004 Journal of Biomolecular Screening

45. Melanocyte and Keratinocyte cocultures
T. Lie, Jackie Muller, Jeannette Ridge, Vincent
Hearing. FDA, NCI
Murine skin 3D model
Lauren Crigler, Amita Kazhanie, Tae-Jin Yoon,
Julia Zakhari, Joanna Anders and Barbara
Taylor, with outstanding support from Stuart
Yuspa, Ulrike Lichti, and Luowei Li. NCI
3D Models of Cancer
Lauren Kimlin and Giovanna Casagrande. NCI
3D in vitro tissue models
Lauren Kimlin and Jareer Kassis
Acknowledgements