A Novel Network Profiling Analysis Reveals System
Changes in Epithelial-Mesenchymal Transition
Teppei Shimamura1*, Seiya Imoto1, Yukako Shimada2, Yasuyuki Hosono2, Atsushi Niida1, Masao
Nagasaki1, Rui Yamaguchi1, Takashi Takahashi2, Satoru Miyano1
1Human Genome Center, Institute of Medical Science, University of Tokyo, Minato-ku, Tokyo, Japan, 2Nagoya University Graduate School of Medicine, Showa-ku,
Patient-specific analysis of molecular networks is a promising strategy for making individual risk predictions and treatment
decisions in cancer therapy. Although systems biology allows the gene network of a cell to be reconstructed from clinical
gene expression data, traditional methods, such as Bayesian networks, only provide an averaged network for all samples.
Therefore, these methods cannot reveal patient-specific differences in molecular networks during cancer progression. In this
study, we developed a novel statistical method called NetworkProfiler, which infers patient-specific gene regulatory
networks for a specific clinical characteristic, such as cancer progression, from gene expression data of cancer patients. We
applied NetworkProfiler to microarray gene expression data from 762 cancer cell lines and extracted the system changes
that were related to the epithelial-mesenchymal transition (EMT). Out of 1732 possible regulators of E-cadherin, a cell
adhesion molecule that modulates the EMT, NetworkProfiler, identified 25 candidate regulators, of which about half have
been experimentally verified in the literature. In addition, we used NetworkProfiler to predict EMT-dependent master
regulators that enhanced cell adhesion, migration, invasion, and metastasis. In order to further evaluate the performance of
NetworkProfiler, we selected Krueppel-like factor 5 (KLF5) from a list of the remaining candidate regulators of E-cadherin
and conducted in vitro validation experiments. As a result, we found that knockdown of KLF5 by siRNA significantly
decreased E-cadherin expression and induced morphological changes characteristic of EMT. In addition, in vitro experiments
of a novel candidate EMT-related microRNA, miR-100, confirmed the involvement of miR-100 in several EMT-related aspects,
which was consistent with the predictions obtained by NetworkProfiler.
Citation: Shimamura T, Imoto S, Shimada Y, Hosono Y, Niida A, et al. (2011) A Novel Network Profiling Analysis Reveals System Changes in Epithelial-
Mesenchymal Transition. PLoS ONE 6(6): e20804. doi:10.1371/journal.pone.0020804
Editor: Eric J. Bernhard, National Cancer Institute, United States of America
Received November 2, 2010; Accepted May 13, 2011; Published June 7, 2011
Copyright: ? 2011 Shimamura et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits
unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This research was supported by "Research and Development of the Next-Generation Integrated Simulation of Living Matter" (a part of the
Development and Use of the Next-Generation Supercomputer Project of MEXT) and "Integrative Systems Understanding of Cancer for Advanced Diagnosis,
Therapy and Prevention" (Grant-in-Aid for Scientific Research on Innovative Areas from MEXT, Japan). The funders had no role in study design, data collection and
analysis, decision to publish, or preparation of the manuscript.
Competing Interests: The authors have declared that no competing interests exist.
* E-mail: firstname.lastname@example.org
Currently, several large-scale omics projects, such as the National
Cancer Institute’s Cancer Genome Atlas (http://cancergenome.nih.
gov/) and the Sanger Institute’s Cancer Genome Project (http://
www.sanger.ac.uk/genetics/CGP/), produce large amounts of data,
including genomic, epigenomic, and transcriptomic information,
about cancer patients or cell lines. Two challenges in omics are to
construct and analyze patient-specific molecular networks to develop
a comprehensive understanding of the molecular mechanisms of
tumorigenesis and to identify molecules that are critical for tumor
proliferation and progression . If these challenges can be
overcome, it may be possible to personalize cancer therapy, improve
its efficacy, and reduce its toxicity and cost [2,3].
Systems biology integrates various types of omics data and
computational tools to represent and analyze complex biological
systems. For example, gene network estimation that is based on
Bayesian networks or mutual information networks can reconstruct
biological systems from gene expression data . However, most
traditional gene network estimation methods construct a static
network by using gene expression data from different cellular
conditions. As a result, these methods only produce an averaged
network for all patients and cannot reveal patient-specific molecular
specific gene network from only a few gene expression profiles of the
patient without making any assumptions about the network.
In this study, we developed a novel statistical method called
NetworkProfiler, which infers patient-specific gene regulatory
networks from a dataset of cancer gene expression profiles.
NetworkProfiler is based on a statistical graphical model with varying
coefficients and a kernel-based data integration method with elastic
net regularization for parameter estimation. A key feature of
NetworkProfiler is that the strengths of the relationships between
cancer progression, metastasis, disease-free survival, and drug
sensitivity. NetworkProfiler groups samples according to the specific
cancer characteristics so that neighboring samples have common
gene regulatory systems. Then, by integrating the gene expression
profiles of neighboring samples with a kernel method, NetworkPro-
filer produces a gene regulatory network for each sample. Finally, we
analyzed 2 post-analysis to discover upstream regulatory genes and
downstream target genes for specific cancer characteristics. Network-
PLoS ONE | www.plosone.org1June 2011 | Volume 6 | Issue 6 | e20804
Profiler is the first algorithm for constructing patient-specific gene
regulatory networks from clinical cancer gene expression data to
elucidate cancer heterogeneity.
We applied NetworkProfiler to gene expression microarray data
from 762 cancer cell lines to determine system changes related to the
epithelial-mesenchymal transition (EMT). The epithelial-mesenchy-
mal transition (EMT) is a process that changes proliferating cellsfrom
an aplanetic state to a motile state , which allows cancer cells to
leave the primary tumor and metastasize. The loss of E-cadherin, a
cell adhesion molecule, is a biomarker of EMT . NetworkProfiler
identified 25 key regulators of E-cadherin, of which half have been
previously described and the other half were novel candidates.
NetworkProfiler also revealed regulatory changes in miR-141, ZEB1,
and E-cadherin. Specifically, our results suggested that decreased
expression of miR-141 in mesenchymal cells disrupts the negative
feedback loop between miR-141 and ZEB1, which would allow ZEB1
to decrease the expression of E-cadherin during the EMT. In
addition, we predicted 45 EMT-dependent putative master regula-
tors that control sets of genes involved in cell adhesion, migration,
invasion and metastasis, namely, 17 of which are downstream targets
of TGFB1, a master switch of the EMT. To further validate the
performance of NetworkProfiler, we experimentally evaluated in silico
predictions obtained by NetworkProfiler. We consequently found
that knockdown of KLF5, a new candidate regulator of E-cadherin,
characteristic of EMT. In addition, the functional involvement of
miR-100 was validated in some EMT-related aspects, which was
consistent with the predictions obtained by Network Profiler.
Overview of NetworkProfiler
Here, we provide an overview of NetworkProfiler; please refer to
the Methods section for a complete description. NetworkProfiler is a
modulator-dependent graphical model because it includes a
modulator (M) variable in addition to regulator (R) and target (T)
variables (genes). R controls the transcription of T and M is a
cofactor that modulates the interaction between R and T. In this
study, we defined M as a biological or a clinical feature that is related
to cancer, such as drug response, survival risk, or a molecule or
pathway that is related to cancer initiation, progression, or metastasis.
The relationships between R, T, and M are illustrated in Figure 1a.
T varies depending on the value of M. Thus, M does not affect R
and T directly; instead, it influences the strength of the relationship
between R and T. In contrast, existing graphical models, such as
Bayesian networks and mutual information networks , do not
consider the effect of M (Figure 1c),sothe strength of the relationship
between R and T remains constant for all values of M (Figure 1d).
In addition, NetworkProfiler can infer the relationships between
R and T, given a value of M. As a result, we could use
NetworkProfiler to construct patient-specific networks with varying
R-T relationships that reflect changes in the feature of interest in
cancer patients. A simple example with synthetic data for R, T, and
M is shown in Figure 2a. In this example, we assume that R
regulates T only with a high value of M (Figure 2b). In this case,
most existing methods that only consider R and T in all of the
samples (Figure 2c) and ignore M would conclude that R does not
regulate T. In contrast, NetworkProfiler attempts to quantify the
strength of the relationship between R and T for a specific value m
ofM byreweightingthedataaccordingtothevalueofM toidentify
the neighborhood of samples with values of M that are close to m.
Then, NetworkProfiler measures the dependency between R and T
on the basis of these neighboring samples. The optimization of the
size of the neighborhood is explained in the Method section.
A schematic representation of the entire analytical process of
NetworkProfiler is shown in Figure 3. NetworkProfiler used 2
inputs: (1) gene expression data and (2) the modulator for each
sample (Figure 3a). The gene expression data was represented as a
p|n matrix, where p is the number of genes and n is the number
Figure 1. The relationships between a regulator (R), a target(T), and a modulator (M) in NetworkProfiler and existing graphical
models. (a). The relationships between R, T and M in NetworkProfiler. The directed solid-line edge from R to T represents ‘‘R regulates the
transcript of T’’. The directed dot-line edge from M to the edge between R and T describes ‘‘M controls the strength of the relationship between R
and T’’. (b). The strength of the relationship between R and T in NetworkProfiler that varies depending on the value of M. (c). The relationships
between R and T in existing graphical models that do not consider the effect of M. (d). The strength of the relationship between R and T in existing
graphical models that remains constant for all values of M.
Network Profiling Analysis
PLoS ONE | www.plosone.org2June 2011 | Volume 6 | Issue 6 | e20804
of samples (patients). If the modulator was an observable variable,
then we directly applied NetworkProfiler to these inputs. However,
if the modulator was a variable that is difficult to observe, then we
used a signature-based hidden modulator extraction algorithm to
estimate the value of the modulator. The output of NetworkPro-
filer is a set of gene networks for every value of M (i.e., sample-
specific gene networks) shown in Figure 3b.
Afterwards, we used 2 post-analysis techniques to extract
biological information from the networks. The first technique
identified upstream regulators of a target gene of interest in the
constructed modulator-dependent gene networks. To evaluate the
modulator-dependent strength of a regulator for the target gene,
we created a measure called the regulatory effect. The regulatory
effect profiles of the upstream regulators for specific target genes
are shown in Figure 3c. The second technique discovered putative
master regulators that control downstream target gene sets with
previously curated functions. To evaluate the enrichment of the
target genes on a functional gene set, we created measure called
the enrichment score. The resulting regulator-function matrix
(Figure 3d) illustrates the candidate regulators (rows) of functions
(columns) that are enhanced in the target genes.
Identification of system changes in the epithelial-
To identify system changes during the EMT, we applied
NetworkProfiler to gene expression profiles of 762 cancer cell lines
from the Sanger Cell Line Project (http://www.broadinstitute.
org/cgi-bin/cancer/datasets.cgi). This dataset included the ex-
pression profiles of 22,777 probes, which correspond to 13,006
mRNAs in these cancer cell lines from the Affymetrix GeneChip
Figure 2. A regulatory change between a regulator (R) and a target (T) depending on the value of a modulator M. (a). A simple
example with synthetic data from 1000 samples for R, T, and M where x-, y-, and z-axises correspond to the expressions of R and T, and the values
of M, respectively. (b). The 3 scatter plots of R and T that are conditioned on the value of M. The left, middle, and right figures represent the scatter
plots from 1-st sample to 333-th sample, from 334-th sample to 666-th sample, and from 667-th sample to 1000-th sample in order of ascending M,
respectively. (c). The scatter plot of R and T that are not conditioned on the value of M.
Network Profiling Analysis
PLoS ONE | www.plosone.org3June 2011 | Volume 6 | Issue 6 | e20804
The supercomputing resource was provided by Human Genome Center
(University of Tokyo).
Conceived and designed the experiments: TT. Performed the experiments:
YS YH. Analyzed the data: TS AN. Wrote the paper: TS. Organized the
project: SM. Provided statistical expertise: SI RY. Provided computational
expertize: AN MN. Provided experimental expertise: TT. Provided
manuscript review: SI AN RY TT.
1. Wang E (2010) Cancer systems biology CRC Press.
2. Schisky RL (2010) Personalized medicine in oncology: the future is now. Nat
Rev Drug Discov 9(5): 363–6.
3. Gonzalez-Angulo AM, Hennessy BT, Mills GB (2010) Future of personalized
medicine in oncology: a systems biology approach. J Clin Oncol 28(16):
4. Bansal M, Belcastro V, Ambesi-Impiombato A, di Bernardo D (2007) How to
infer gene networks from expression profiles. Mol Syst Biol 3: 78.
5. Thiery JP, Acloque H, Huang RY, Nieto MA (2009) Epithelial-mesenchymal
transitions in development and disease. Cell 139(5): 871–90.
6. Subramanian A, Tamayo P, Mootha VK, Mukherjee S, Ebert BL, et al. (2005)
Gene set enrichment analysis: a knowledge-based approach for interpreting
genome-wide expression profiles. Proc Natl Acad Sci U S A 102(43): 15545–50.
7. Niida A, Smith AD, Imoto S, Aburatani H, Zhang MQ, et al. (2009) Gene set-
based module discovery in the breast cancer transcriptome. BMC Bioinformatics
8. Gregory PA, Bert AG, Paterson EL, Barry SC, Tsykin A, et al. (2008) The miR-
200 family and miR-205 regulate epithelial to mesenchymal transition by
targeting ZEB1 and SIP1. Nat Cell Biol 10(5): 593–601.
9. Comijn J, Berx G, Vermassen P, Verschueren K, van Grunsven L, et al. (2001)
The two-handed E box binding zinc finger protein SIP1 downregulates E-
cadherin and induces invasion. Mol Cell 7(6): 1267–78.
10. Yori JL, Johnson E, Zhou G, Jain MK, Keri RA (2010) Kruppel-like factor 4
inhibits epithelial-tomesenchymal transition through regulation of E-cadherin
gene expression. J Biol Chem 285(22): 16854–63.
11. Song Y, Washington MK, Crawford HC (2010) Loss of FOXA1/2 is essential
for the epithelialto-mesenchymal transition in pancreatic cancer. Cancer Res
12. Sobrado VR, Moreno-Bueno G, Cubillo E, Holt LJ, Nieto MA, et al. (2009) The
class I bHLH factors E2-2A and E2-2B regulate EMT. J Cell Sci 122(Pt 7):
13. Kato M, Zhang J, Wang M, Lanting L, Yuan H, et al. (2007) MicroRNA-192 in
diabetic kidney glomeruli and its function in TGF-beta-induced collagen
expression via inhibition of E-box repressors. Proc Natl Acad Sci U S A 104(9):
14. Wang B, Herman-Edelstein M, Koh P, Burns W, Jandeleit-Dahm K, et al.
(2010) E-cadherin expression is regulated by miR-192/215 by a mechanism that
is independent of the profibrotic effects of transforming growth factor-beta.
Diabetes 59(7): 1794–802.
15. Eger A, Aigner K, Sonderegger S, Dampier B, Oehler S, et al. (2005) DeltaEF1
is a transcriptional repressor of E-cadherin and regulates epithelial plasticity in
breast cancer cells. Oncogene 24(14): 2375–85.
16. Hajra KM, Chen DY, Fearon ER (2002) The SLUG zinc-finger protein
represses E-cadherin in breast cancer. Cancer Res 62(6): 1613–8.
17. Yang J, Mani SA, Donaher JL, Ramaswamy S, Itzykson RA, et al. (2004) Twist,
a master regulator of morphogenesis, plays an essential role in tumor metastasis.
Cell 117(7): 927–39.
18. Perez-Moreno MA, Locascio A, Rodrigo I, Dhondt G, Portillo F, et al. (2001) A
new role for E12/E47 in the repression of E-cadherin expression and epithelial-
mesenchymal transitions. J Biol Chem 276(29): 27424–31.
19. Almeida MS, Bray SJ (2005) Regulation of post-embryonic neuroblasts by
Drosophila Grainyhead. Mech Dev 122(12): 1282–93.
20. Cowger JJ, Zhao Q, Isovic M, Torchia J (2007) Biochemical characterization of
the zinc-finger protein 217 transcriptional repressor complex: identification of a
ZNF217 consensus recognition sequence. Oncogene 26(23): 3378–86.
21. Yang Y, Goldstein BG, Chao HH, Katz JP (2005) KLF4 and KLF5 regulate
proliferation, apoptosis and invasion in esophageal cancer cells. Cancer Biol
Ther 4(11): 1216–21.
22. Zou H, Hastie T (2005) Regularization and variable selection via the elastic net.
J Roy Statist Soc Ser B 67: 301–20.
23. Bracken CP, Gregory PA, Kolesnikoff N, Bert AG, Wang J, et al. (2008) A
double-negative feedback loop between ZEB1-SIP1 and the microRNA-200
family regulates epithelial-mesenchymal transition. Cancer Res 68(19): 7846–54.
24. Willis BC, Borok Z (2007) TGF-beta-induced EMT: mechanisms and
implications for fibrotic lung disease. Am J Physiol Lung Cell Mol Physiol
25. Chen H, Zhu G, Li Y, Padia RN, Dong Z, et al. (2009) Extracellular signal-
regulated kinase signaling pathway regulates breast cancer cell migration by
maintaining slug expression. Cancer Res 69(24): 9228–35.
26. Adiseshaiah P, Lindner DJ, Kalvakolanu DV, Reddy SP (2007) FRA-1 proto-
oncogene induces lung epithelial cell invasion and anchorage-independent
growth in vitro, but is insufficient to promote tumor growth in vivo. Cancer Res
27. Cowden Dahl KD, Robertson SE, Weaver VM, Simon MC (2005) Hypoxia-
inducible Factor Regulates alphavbeta3 Integrin Cell Surface Expression. Mol
Biol Cell 16(4): 1901–12.
28. Imtiyaz HZ, Williams EP, Hickey MM, Patel SA, Durham AC, et al. (2010)
Hypoxia-inducible factor 2alpha regulates macrophage function in mouse
models of acute and tumor inflammation. J Clin Invest 120(8): 2699–714.
29. Yang Y, Tetreault MP, Yermolina YA, Goldstein BG, Katz JP (2008) Kruppel-
like Factor 5Controls Keratinocyte Migration via the Integrin-linked Kinase.
J Biol Chem 283(27): 18812–20.
30. Marlowe JL, Puga A (2005) Aryl hydrocarbon receptor, cell cycle regulation,
toxicity, and tumorigenesis. J Cell Biochem 96(6): 1174–84.
31. Nakagawa H, Liyanarachchi S, Davuluri RV, Auer H, Martin EW, et al. (2004)
Role of cancerassociated stromal fibroblasts in metastatic colon cancer to the
liver and their expression profiles. Oncogene 23(44): 7366–77.
32. Malin D, Kim IM, Boetticher E, Kalin TV, Ramakrishna S, et al. (2007)
Forkhead box f1 is essential for migration of mesenchymal cells and directly
induces integrin-beta3 expression. Mol Cell Biol 27(7): 2486–98.
33. Buchwalter G, Gross C, Wasylyk B (2005) The Ternary Complex Factor Net
Regulates Cell Migration through Inhibition of PAI-1 Expression. Mol Cell Biol
34. Hayes SA, Huang X, Kambhampati S, Platanias LC, Bergan RC (2003) p38
MAP kinase modulatesSmad-dependent changes in human prostate cell
adhesion. Oncogene 22(31): 4841–50.
35. Matsuzaki K, Kitano C, Murata M, Sekimoto G, Yoshida K, et al. (2009)
Smad2 and Smad3 phosphorylated at both linker and COOH-terminal regions
transmit malignant TGF-beta signal in later stages of human colorectal cancer.
Cancer Res 69(13): 5321–30.
36. Sekimoto G, Matsuzaki K, Yoshida K, Mori S, Murata M, et al. (2007)
Reversible Smad-Dependent Signaling between Tumor Suppression and
Oncogenesis. Cancer Res 67(11): 5090–6.
37. Sato M, Muragaki Y, Saika S, Roberts AB, Ooshima A (2003) Targeted
disruption of TGFbeta1/Smad3 signaling protects against renal tubulointerstitial
fibrosis induced by unilateral ureteral obstruction. J Clin Invest 112(10):
38. Chan SW, Lim CJ, Guo K, Ng CP, Lee I, et al. (2008) A role for TAZ in
migration, invasion, and tumorigenesis of breast cancer cells. Cancer Res 68(8):
39. Sachdeva M, Mo YY (2010) MicroRNA-145 suppresses cell invasion and
metastasis by directly targeting mucin 1. Cancer Res 70(1): 378–87.
40. Matsuya M, Sasaki H, Aoto H, Mitaka T, Nagura K, et al. (1998) Cell adhesion
kinase beta forms a complex with a new member, Hic-5, of proteins localized at
focal adhesions. J Biol Chem 273(2): 1003–14.
41. Lu X, Yan CH, Yuan M, Wei Y, Hu G, et al. (2010) In vivo dynamics and
distinct functions of hypoxia in primary tumor growth and organotropic
metastasis of breast cancer. Cancer Res 70(10): 3905–14.
42. Okuyama H, Krishnamachary B, Zhou YF, Nagasawa H, Bosch-Marce M,
et al. (2006) Expressionof vascular endothelial growth factor receptor 1 in bone
marrow-derived mesenchymal cells is dependent on hypoxia-inducible factor 1.
J Biol Chem 281(22): 15554–63.
43. Kim KS, Sengupta S, Berk M, Kwak YG, Escobar PF, et al. (2006) Hypoxia
Enhances Lysophosphatidic Acid Responsiveness in Ovarian Cancer Cells and
Lysophosphatidic Acid Induces Ovarian Tumor Metastasis In vivo. Cancer Res
44. Schedin PJ, Eckel-Mahan KL, McDaniel SM, Prescott JD, Brodsky KS, et al.
(2004) ESX induces transformation and functional epithelial to mesenchymal
transition in MCF-12A mammary epithelial cells. Oncogene 23(9): 1766–79.
45. Panda DK, Miao D, Lefebvre V, Hendy GN, Goltzman D (2001) The
transcription factor SOX9 regulates cell cycle and differentiation genes in
chondrocytic CFK2 cells. J Biol Chem 276(44): 41229–36.
46. Mori-Akiyama Y, Akiyama H, Rowitch DH, de Crombrugghe B (2003) Sox9 is
required for determination of the chondrogenic cell lineage in the cranial neural
crest. Proc Natl Acad Sci U S A 100(16): 9360–5.
47. Tomioka N, Osumi N, Sato Y, Inoue T, Nakamura S, et al. (2000) Neocortical
origin and tangential migration of guidepost neurons in the lateral olfactory
tract. J Neurosci 20(15): 5802–12.
48. Jean C, Blanc A, Prade-Houdellier N, Ysebaert L, Hernandez-Pigeon H, et al.
(2009) Epidermal growth factor receptor/beta-catenin/T-cell factor 4/matrix
metalloproteinase 1: a new pathway for regulating keratinocyte invasiveness after
UVA irradiation. Cancer Res 69(8): 3291–9.
49. Minami T, Miura M, AirdWC, Kodama T (2006) Thrombin-induced
Autoinhibitory Factor, Down Syndrome Critical Region-1, Attenuates NFAT-
dependent Vascular Cell Adhesion Molecule-1 Expression and Inflammation in
the Endothelium. J Biol Chem 281(29): 20503–20.
Network Profiling Analysis
PLoS ONE | www.plosone.org16 June 2011 | Volume 6 | Issue 6 | e20804
50. Huang S, Pettaway CA, Uehara H, Bucana CD, Fidler IJ (2001) Blockade of
NF-kappaB activity in human prostate cancer cells is associated with suppression
of angiogenesis, invasion, and metastasis. Oncogene 20(31): 4188–97.
51. Shair KH, Schnegg CI, Raab-Traub N (2008) EBV Latent Membrane Protein 1
Effects on Plakoglobin, Cell Growth, and Migration. Cancer Res 68(17):
52. Gakidis MA, Cullere X, Olson T, Wilsbacher JL, Zhang B, et al. (2004) Vav
GEFs are required for beta2 integrin-dependent functions of neutrophils. J Cell
Biol 166(2): 273–82.
53. Schymeinsky J, Sindrilaru A, Frommhold D, Sperandio M, Gerstl R, et al.
(2006) The Vav binding site of the non-receptor tyrosine kinase Syk at Tyr 348 is
critical for beta2 integrin (CD11/CD18)- mediated neutrophil migration. Blood
54. Hong IK, Jin YJ, Byun HJ, Jeoung DI, Kim YM, et al. (2006) Homophilic
Interactions of Tetraspanin CD151 Up-regulate Motility and Matrix Metallo-
proteinase-9 Expression of Human Melanoma Cells through Adhesion-
dependent c-Jun Activation Signaling Pathways. J Biol Chem 281(34):
55. Janulis M, Silberman S, Ambegaokar A, Gutkind JS, Schultz RM (1999) Role of
mitogen-activated protein kinases and c-Jun/AP-1 trans-activating activity in the
regulation of protease mRNAs and the malignant phenotype in NIH 3T3
fibroblasts. J Biol Chem 274(2): 801–13.
56. Cai C, Hsieh CL, Omwancha J, Zheng Z, Chen SY, et al. (2007) ETV1 Is a
Novel Androgen Receptor-Regulated Gene that Mediates Prostate Cancer Cell
Invasion. Mol Endocrinol 21(8): 1835–46.
57. Bauer K, Kratzer M, Otte M, de Quintana KL, Hagmann J, et al. (2000)
Human CLP36, a PDZdomain and LIM-domain protein, binds to alpha-
actinin-1 and associates with actin filaments and stress fibers in activated
platelets and endothelial cells. Blood 96(13): 4236–45.
58. Zent CS, Mathieu C, Claxton DF, Zhang DE, Tenen DG, et al. (1996) The
chimeric genes AML1/MDS1 and AML1/EAP inhibit AML1B activation at
the CSF1R promoter, but only AML1/MDS1 has tumor-promoter properties.
Proc Natl Acad Sci U S A 93(3): 1044–8.
59. Perry C, Sklan EH, Birikh K, Shapira M, Trejo L, et al. (2002) Complex
regulation of acetylcholinesterase gene expression in human brain tumors.
Oncogene 21(55): 8428–41.
60. Zhang X, Milton CC, Humbert PO, Harvey KF (2009) Transcriptional output
of the Salvador/ warts/hippo pathway is controlled in distinct fashions in
Drosophila melanogaster and mammalian cell lines. Cancer Res 69(15):
61. Aung CS, Hill MM, Bastiani M, Parton RG, Parat MO (2010) PTRF-cavin-1
expression decreases the migration of PC3 prostate cancer cells: role of matrix
metalloprotease 9. Eur J Cell Biol, 2010 Aug 21. [Epub ahead of print].
62. Hurst DR, Edmonds MD, Scott GK, Benz CC, Vaidya KS, et al. (2009) Breast
cancer metastasis suppressor 1 up-regulates miR-146, which suppresses breast
cancer metastasis. Cancer Res 69(4): 1279–83.
63. Cordes KR, Sheehy NT, White MP, Berry EC, Morton SU, et al. (2009) miR-
145 and miR-143 regulate smooth muscle cell fate and plasticity. Nature
64. Kasai H, Allen JT, Mason RM, Kamimura T, Zhang Z (2005) TGF-b1 induces
human alveolar epithelial to mesenchymal cell transition (EMT). Respir Res 6:
65. Taguchi A, Yanagisawa K, Tanaka M, Cao K, Matsuyama Y, et al. (2008)
Identification of hypoxiainducible factor-1 alpha as a novel target for miR-17-92
microRNA cluster. Cancer Res 68(14): 5540–5.
66. Kozaki K, Miyaishi O, Tsukamoto T, Tatematsu Y, Hida T, et al. (2000)
Establishment and characterization of a human lung cancer cell line NCI-H460-
LNM35 with consistent lymphogenous metastasis via both subcutaneous and
orthotopic propagation. Cancer Res 60(9): 2535–40.
67. Tokumaru S, Suzuki M, Yamada H, Nagino M, Takahashi T (2008) let-7
regulates Dicer expression and constitutes a negative feedback loop. Carcino-
genesis 29(11): 2073–7.
68. Tibshirani R (1996) Regression shrinkage and selection via the lasso. J Royal
Statist Soc B 58(1): 267–88.
69. Hoerl AE, Kennard R (1970) Ridge regression: biased estimation for
nonorthogonal problems. Technometrics 12: 55–67.
70. Shimamura T, Imoto S, Yamaguchi R, Fujita A, Nagasaki M, et al. (2009)
Recursive regularization for inferring gene networks from time-course gene
expression profiles. BMC Syst Biol 3: 41.
71. Shimamura T, Imoto S, Yamaguchi R, Nagasaki M, Miyano S (2010) Inferring
dynamic gene networks under varying conditions for transcriptomic network
comparison. Bioinformatics 26(8): 1064–72.
72. Mardia K, Kent J, Bibby J (1979) Multivariate Analysis Academic Press.
73. Benjamini Y, Hochberg Y (1995) Controlling the false discovery rate: a practical
and powerful approach to multiple testing. J Roy Statist Soc Ser B 57(1):
74. Hartung J, Knapp G, Sinha BK (2008) Statistical meta-analysis with
Network Profiling Analysis
PLoS ONE | www.plosone.org17 June 2011 | Volume 6 | Issue 6 | e20804