LOXL2 in colon cancer prognosis - Digital CSICdigital.csic.es/bitstream/10261/128643/1/LOXL2-Casal.pdf · 2020. 11. 16. · in the stratification of high- and low-risk colon cancer

LOXL2 in colon cancer prognosis

1

LOXL2 IS HIGHLY EXPRESSED IN CANCER-ASSOCIATED FIBROBLASTS AND ASSOCIATES TO POOR COLON CANCER SURVIVAL

Sofía Torres1*, Irene Garcia-Palmero1, Mercedes Herrera2, Rubén A. Bartolomé1, Cristina Peña2, Mª Jesús Fernandez-Aceñero3, Guillermo Padilla4, Alberto Peláez-García1, María Lopez-Lucendo4, Rufo Rodriguez-Merlo5, Antonio García de Herreros6, Félix Bonilla2 and J. Ignacio Casal1*

1. Department of Cellular and Molecular Medicine. Centro de Investigaciones Biológicas (CIB-CSIC). 28040 Madrid, Spain.

2. Department of Medical Oncology. Hospital Puerta de Hierro Majadahonda. Madrid. Spain

3. Department of Pathology. Fundación Jiménez Díaz. Madrid. Spain

4. Core facilities. CIB-CSIC. Madrid. Spain

5. Department of Pathology. Hospital Virgen de la Salud. Toledo. Spain

6. IMIM-Hospital del Mar. Barcelona. Spain

RUNNING TITLE: LOXL2 in colon cancer prognosis

KEYWORDS: LOXL2; prognostic biomarker; cancer-associated fibroblasts; colon cancer

Word count: 5107 Figures: 5 Tables: 1

CONFLICT OF INTEREST: The authors declare no conflict of interest.

*Corresponding author: J. Ignacio Casal Department of Cellular and Molecular Medicine Centro de Investigaciones Biológicas (CIB-CSIC) Ramiro de Maeztu, 9 28040 Madrid. Spain Phone: +34 918373112 Fax: +34 91 5360432 e-mail: [email protected]

mailto:[email protected]


2

Translational relevance

There is an urgent need to find adequate biomarkers for correct risk stratification of

colon cancer patients, particularly at stage II and III. Cancer-associated fibroblasts

(CAFs) have been associated to recurrence and survival. After proteomic

characterization of purified colon cancer-associated fibroblasts we identified a number

of stromal-associated proteins that can be used as biomarkers. The validity of this model

was supported by the similar protein profile obtained after TGFβ activation of normal

fibroblasts. Among the selected stromal biomarkers, LOXL2 and TAGLN exhibited

strong predictive value as molecular classifiers. High expression of LOXL2 showed a

clear association with poor prognosis, either disease-free or overall survival. In

addition, it displayed a significant prognostic value in the identification of high-risk

stage II and III patients. LOXL2 has the potential to be used in clinical practice for risk

classification, which could lead to a more aggressive therapy and better outcome of the

patients.


3

ABSTRACT

Purpose: Cancer associated-fibroblasts (CAFs) are major mediators in tumor

microenvironment. We investigated the changes in protein expression in colon cancer

associated-fibroblasts compared to normal fibroblasts (NFs) in the context of searching

for prognostic biomarkers, particularly for stage II patients.

Experimental design: CAFs and NFs isolated from colon cancer patients were used to

identify differentially-expressed proteins using quantitative proteomics. Stromal

expression of deregulated proteins was analyzed by immunohistochemistry. Prognostic

impact was studied using external gene expression datasets for training, then

quantitative PCR and immunohistochemistry for validation in different cohorts of

patients. Combined datasets were used for prediction of risk assessment at stage II and

III.

Results: A desmoplastic signature composed of 32 proteins, highly specific for stromal

components in colon cancer, was identified. These proteins were enriched for

extracellular matrix organization components, TGFβ signaling pathway, fibrosis and

wound healing proteins. The expression in CAFs of 11 up-regulated proteins and 4

down-regulated proteins, selected for biomarker validation, was verified by orthogonal

techniques. LOXL2 displayed a high prognostic impact by using external independent

datasets and further validation in two different cohorts of patients. High expression of

LOXL2 was associated with higher recurrence P = 0.001 HR: 5.38 (1.70-17.01) and

overall survival P = 0.001 HR: 8.52 (1.90-38.29). Immunohistochemistry analysis

revealed a prognostic value for LOXL2 in stage II patients.


4

Conclusions: We identified LOXL2 to be associated with the outcome of colon cancer

patients. Furthermore, it can be used to stratify patients at stages II and III for further

therapeutic decisions.


5

INTRODUCTION

Cancer-associated fibroblasts (CAFs) play an important role in the development and

progression of tumors (1, 2). In many solid tumors, cancer progression is accompanied

by the activation of stromal fibroblasts into myofibroblasts or cancer-associated

fibroblasts (CAFs), characterized by “de novo” expression of α-smooth muscle actin (α-

SMA) (3). Fibroblasts are involved in tissue remodeling and repair. CAFs are also

responsible for the increased deposition of extracellular matrix (ECM) components in

the tumor invasion front, known as desmoplastic reaction (4). CAFs acquire a

phenotype similar to wound healing-activated myofibroblasts. While this activation is

temporal in a normal wound; in cancer, the fibroblasts remain endlessly activated. They

synthesize multiple factors that stimulate and accelerate the healing process,

angiogenesis and immune cell infiltration, all contributing to create an appropriate

microenvironment for tumor growth and invasion.

A tumor cannot develop without the parallel expansion of a tumor stroma, as

described for breast (5), lung (6), prostate (7) and colon cancer (8), among others. TGFβ

regulates fibrosis and is the most important factor affecting CAF activation (9),

promoting the expression of factors involved in paracrine signaling, ECM production

and remodeling (10). Stromal changes have been associated with tumor progression and

prognostic impact. Unfortunately, only a few selective biomarkers of myofibroblasts are

well established: α-SMA, fibroblast activation protein (FAP) and the fibroblast specific

protein (FSP1), also known as S100A4, are the most commonly and widely used CAF

markers in solid cancers. However, these proteins are not only present in CAFs but also

in pericytes or endothelial cells (11). Moreover, CAFs show a variable expression of

these markers in different tumors.


6

The epithelial-stromal ratio of colon carcinomas has been used as a predictor of

survival, independent from lymph node status and tumor stage (8, 12). Some studies

have correlated the number of stromal myofibroblasts (13), vimentin expression (14)

and FAP expression (15) with colon cancer prognosis. In other exploratory studies for

stromal biomarkers, Nakagawa et al. (16) compared metastatic colon cancer fibroblasts

with liver and skin fibroblasts to get a gene expression profile that included many

adhesion molecules and ECM-remodeling genes. Recently, a comparative

transcriptomic analysis of paired CAFs and normal fibroblasts (NFs) identified 108

genes, mostly down-regulated (17). In a proteomic study, colon cancer CAFs were

compared with bone-marrow precursors to identify tenascin C, ED-A fibronectin and

SDF-1 as deregulated in CAFs (18). Recently, we used a murine model of sporadic

colon cancer based on azoxymethane/ dextran sodium sulfate to optimize fibroblast

purification and characterize alterations in murine CAFs (19).

An “in depth” proteomic analysis of human CAFs should give us a more

complete picture of stromal biomarkers for prognostic classification in order to help us

in the stratification of high- and low-risk colon cancer patients. This is particularly

relevant for stage II and III patients, where chemotherapy choices have to be made.

Proteomics displays several advantages over transcriptomics for a rapid clinical

translation. Quantified proteins are easier to adapt to standard laboratory techniques.

Direct protein identification avoids problems of translational stability and poor

correlation between mRNA and protein abundance (20).

Here, we investigated the differences in the protein expression profile between

colon cancer CAFs and NFs by using iTRAQ (isobaric Tags for Relative and Absolute

Quantification), which allows the concurrent identification and relative quantification of

the proteome from different biological samples in a single experiment (21). A problem


7

associated to human CAFs is the heterogeneity and relatively large variability between

patients. To overcome this problem, we used pools of samples for the proteomic study.

Pool results were then confirmed in individual patient analysis. A large collection of

proteins deregulated in colon cancer stroma as compared to healthy tissue were

identified and validated by different approaches, including TGFβ-activation of NFs. We

obtained a panel of stromal markers with a prognostic impact using different datasets.

LOXL2 exhibits a great promise in colon cancer prognosis and survival prediction,

including stage II-only patients.


8

MATERIALS AND METHODS

Patients and tissue samples

For proteomic analysis, surgically resected biopsies of colon cancer and paired non-

cancerous tissues (collected 10 cm apart from the tumor) were collected from 12

patients at stages II and III from Hospital Virgen de la Salud (Supplementary Table

S1A). For LOXL2 validation, two sample collections from patients followed for more

than five years were obtained from the Hospital Fundación Jimenez Diaz, 70 for

quantitative PCR and 121 for immunohistochemistry analysis in tissue microarray

format (Supplementary Table S1B, C, respectively). Informed written consent was

obtained from all participants, as required and approved by the Research Ethics

Committees of Virgen de la Salud Hospital (Toledo) and the Hospital Fundación

Jimenez Diaz (Madrid), respectively. The histologic diagnosis for each sample was

reconfirmed using microscopic examination of hematoxylin/eosin-stained sections of

each tissue block.

External cohorts of validation

We used three independent external cohorts in the prognostic study. For biomarker

training, we used the GSE17538 superserie, which contains a cohort of 232 patients of

colon cancer (Moffitt samples and VMC samples). For risk stratification at stage II and

III, we used the Amsterdam cohort GSE33113 with 90 patients at stage II and the Berlin

cohort GSE12945, which contains data for 21 patients at stage III (22).

Fibroblast isolation and cell culture

CAFs and NFs were obtained from colonic tissues by the explant technique as described

(19, 23, 24). See Supplementary Methods for additional information. Colon cancer cell


9

lines SW480, SW620, KM12C and KM12SM were cultured as BJ-hTERT cells.

KM12C and KM12SM human colon cancer cells were obtained from I. Fidler´s lab

(MD Anderson Cancer Center. Houston, TX, USA). Other cell lines were obtained

directly from the American Type Culture Collection (ATCC). All these cell lines were

authenticated by short tandem repeat analysis or characterized by karyotype analysis.

These cell lines were passaged fewer than 6 months after purchase for all the

experiments

ITRAQ proteomic analysis

A detailed description of sample preparation and mass spectrometry analysis is given in

Supplementary Methods.

Immunohistochemistry analysis

Each sample was deparaffinized for antigen retrieval using sodium citrate (pH 6.0) for

20 min and subsequent incubation with the respective primary antibody: ACAN (1:100;

MAB1220; R&D system), αSMA (Clone 1A4; DAKO), COL14A1 (1:100;

HPA023781; Sigma), DKK3 (1:100; sc-25518; Santa Cruz Biotechnology), CDH13

(1:100; HPA001380; Sigma), TAGLN (1:200; AF7886; R&D system), TGM2 (1:250;

ab73170; Abcam) or LOXL2 (1:200; NBP1-32954; Novus Biologicals). The reaction

was developed using diaminobenzidine (DAB) as chromogen and hematoxylin for

counterstaining. In all cases, sections from normal colonic mucosa distant from the

tumor site were used as negative controls. In some cases, Human Protein Atlas (HPA)

was used for meta-analysis of expression of deregulated proteins in colorectal tissues

(25).

Quantitative PCR


10

Total RNA was isolated from cellular cultures using the mirVana isolation kit

(Ambion). RNA from formalin-fixed paraffin-embedded (FFPE) tissue was isolated

from 3x10µm-sections using the NucleoSpin totalRNA FFPE XS kit (Macherey-Nagel).

cDNA was subsequently obtained using SuperScript II First Strand Synthesis System

with random hexamers (Invitrogen). Real-time quantitative PCR (QPCR) was

performed using the FastStart Master Mix (Roche) with probes from the Universal

Probe Library Set (Roche). Amplifications were run in a 7900 HT-Fast Real-Time PCR

System (Applied Biosystems). Each value was adjusted using GAPDH, β-actin and 18S

RNA levels as reference. Primer sequences and probes used in this study can be found

in Supplementary Methods. Expression of the selected genes was considered positive

when the tumor/normal ratio showed a fold-change ≥2.

Prognostic analyses using public datasets

To remove technical variations between datasets, expression levels for all probes within

each sample (patient) were transformed to z-score by subtracting the overall average

probe intensity from each expression value and dividing the difference by the standard

deviation of all the intensities. Subsequently, z-score ratios were obtained by subtracting

the average z-score for the cohort from each individual z-score and dividing the result

by the standard deviation of the z-scores for the cohort (26). Positive (z score > 0) and

negative (z score < 0) ratios indicate a significantly higher or lower expression level of

genes in the module, respectively. We derived Kaplan-Meier survival curves for

patients, dividing the cohort into two risk groups, for each selected markers. The log-

rank test p value was determined for differences in time to recurrence (stage I-III

patients) and death (stage I-IV patients) between the two risk groups. Univariant Cox

proportional hazard regression models were fitted for estimation of independent hazard

ratios (HR) for each marker. Kaplan-Meier survival curves were plotted with IBM


11

SPSS Statistics 20. Log-rank tests and Cox regressions were performed with the SAS®

software version 9.3 (SAS). A multivariable Cox regression analysis, adjusted for age,

gender, AJCC stage and differentiation grade was used to calculate hazard ratios (HR).


12

RESULTS

Differentially-expressed proteins in colon CAFs using iTRAQ quantification

Biopsies corresponding to 12 colon cancer patients (Supplementary Table S1A) were

cut in small fragments and fibroblasts were isolated using the explant technique. To

study differences in protein expression between CAFs and NFs, we prepared pools from

complete cell extracts and conditioned medium collected from 6 paired fibroblast

cultures at 95-100% confluence. Protein extracts were trypsin-digested. Then, peptides

were iTRAQ-labeled and fractionated with OFFGEL (pH 3-10). To avoid biases in

peptide labeling, we performed biological replicates using the 114 and 116 iTRAQ tags

for NFs and 115 and 117 for CAFs (Supplementary Fig. S1). Each fraction was

analyzed by duplicate on a LTQ-Orbitrap Velos. In total, 1780 and 305 proteins were

identified in the whole cellular extract and conditioned medium, respectively, using

Proteome Discoverer v1.4 and MASCOT (Supplementary Fig. S2). Total quantified

proteins are listed in Supplementary Table S2 (whole cell extract and conditioned

medium). After applying a fold-change ≥ 1.5, we found 57 and 43 proteins deregulated

in complete cell lysates and conditioned medium, respectively (Supplementary Table

S3). Sixty percent of the proteins were up-regulated, including extracellular matrix

(ECM) proteoglycans like aggrecan (ACAN), biglycan (BGN) or chondroitin sulfate

proteoglycan 4 (CSPG4); proteins involved in organization and modification of ECM

such as lysyl oxidase-like 2 protein (LOXL2) and procollagen-lysine, 2-oxoglutarate 5-

dioxygenase 2 (PLOD2); proteins implicated in cytoskeleton regulation and

organization such as tropomyosin 1 (TPM1), palladin (PALLD), myosin regulatory

light polypeptide 9 (MYL9) or transgelin (TAGLN); proteins related with wound

healing like transglutaminase 2 (TGM2), myosin light chain kinase (MYLK) or

thrombospondin 1 (THBS1) and other proteins like CDH13 or DKK3. Among the


13

down-regulated proteins there were also ECM proteins like decorin (DCN), collagen

type XIV (COL14A1) or periostin (POSTN) and cytokines like colony-stimulating

factor 1 (CSF1).

After literature and data-mining search for novel proteins, not previously

associated to colon cancer, we selected 15 proteins (9 from whole cellular extract and 6

from secretome) for validation by Western Blot and semi-quantitative PCR. Western

blot results, either in pools or individual samples of CAFs and NFs, were consistent

with the iTRAQ quantification data. CDH13, TAGLN and TGM2 proteins were up-

regulated, whereas COL14A1 was down-regulated in the cell lysate. ACAN, DKK3 and

LOXL2 were up-regulated and POSTN down-regulated in the secretome (Fig. 1A). In

addition, we used semi-quantitative PCR to verify the changes in expression of these 15

proteins (Fig. 1B). In general, mRNA ratios of expression followed a similar trend to

those observed by iTRAQ, confirming the alterations in expression for these proteins.

Functional annotation and protein-protein interaction networks in colon CAFs

We carried out functional annotation of colon CAFs-deregulated proteins based on GO

analysis using Genomatix software (Supplementary Table S4). Regarding “biological

processes” the top rank was for “ECM organization” (27 genes, p-value 3.09E-23)

(Supplementary Fig. S3A), followed by “platelet activation” (17 genes, p-value 7.43E-

14), “cell adhesion” (28 genes, p-value 8.79E-12) or “wound healing” (23 proteins, p-

value 7.77E-12) (Supplementary Fig. S3B). Regarding signal transduction pathways,

the TGFβ/SMAD signaling pathway (p-value 7.58E-12) with 26 deregulated proteins

(16 up-regulated) was the most represented (Supplementary Fig. S3C), followed by

“Focal adhesion kinase” or “Integrin-linked kinase”. Finally, regarding

“Overrepresented diseases”, “Fibrosis” (26 proteins, p-value 1.54E-18)


14

(Supplementary Fig. S3D) and “Wounds and Injuries” (27 proteins, p-value 6.97E-18)

were the top alterations. “Neoplasm metastasis” was also highly represented with 27

proteins but a lower p-value 4.89E-10. In summary, GO analysis confirmed ECM

organization, TGFβ/SMAD signaling pathway and fibrosis and wound healing-related

proteins as the most represented functionalities exhibited by the deregulated proteins in

CAFs.

TGFβ activation of fibroblasts displays a protein profile similar to colon CAFs

The activation of the TGF-β pathway plays a major role in fibroblast activation during

wound healing, organ fibrosis and cancer as confirmed by GO analysis of our data

(Supplementary Fig. S3C) (27). Consequently, we expected that the activation of

normal fibroblasts with exogenous TGFβ would yield an expression pattern similar to

that observed in CAFs. Therefore, BJh-TERT fibroblasts were treated with TGFβ for 24

h and candidate CAFs biomarkers were analyzed by quantitative PCR and western blot.

Except for POSTN and ACAN, which did not get amplified after RT-PCR, most of the

up-regulated proteins in CAFs (BGN, CNN3, CSPG4, DKK3, LOXL2, MYL9, PLOD2,

TAGLN and TGM2) and down-regulated (COL14A1 and CSF1) were also

differentially expressed in TGFβ-activated BJh-TERT fibroblasts by QPCR (Fig. 1C)

and western blot (Fig. 1D). These findings confirmed that activated colon CAFs display

an expression profile similar to TGFβ-activated fibroblasts.

Stromal specificity of deregulated proteins in colon CAFs

To analyze cellular specificity of altered CAF proteins, we used two strategies. First,

we used the GSE39396 dataset corresponding to the expression profile of four sorter-

isolated cell types (endothelial cells: CD31+, epithelial cells: EPCAM+, inflammatory

cells: CD45+ and CAFs: FAP+) from six colorectal cancer patients (28). Data were


15

transformed in z-score values and analyzed using color-coded heat map. Positive and

negative z-scores indicated a significantly higher or lower expression level,

respectively. 28 out of 60 up-regulated proteins (Fig. 2A) and 24 out of 40 down-

regulated proteins (Fig. 2B) corresponded to genes highly-expressed in the fibroblast

compartment. Second, we analyzed the gene expression for these proteins in KM12C,

KM12SM, SW480 and SW620 epithelial colorectal cancer cells using QPCR. Ten

proteins (ACAN, BGN, CDH13, CSPG4, DKK3, LOXL2, MYL9, PLOD2, TAGLN,

and TGM2) were specific or overexpressed in fibroblasts (Fig. 2C). Together, these

results confirm the value of these proteins as specific colon cancer stromal markers.

Desmoplastic signature associated to CAFs biomarkers

Activated fibroblasts contribute greatly to desmoplastic activity. Desmoplastic changes

are associated to fibrosis and correlate with highly-expressed smooth muscle proteins.

Therefore, we investigated for each protein whether or not smooth muscle expression

was among the five top expression sites according to Bio-GPS and HPA. Many up-

regulated (45 out of 60, 75%) and down-regulated (24 out of 40, 60%) CAFs proteins

displayed preferential smooth muscle expression (Supplementary Table S5). Some

representative up-regulated proteins were tested for stromal staining in colon cancer

using immunohistochemical analysis (Fig. 3) or the HPA database (Supplementary

Fig. S4). The HPA showed a preferential expression in cancer stroma of 36 up-

regulated CAFs proteins. We found that ACAN, αSMA, CDH13, DKK3, TAGLN,

TGM2 and LOXL2 were preferentially expressed in cancer stroma, with variable or no

expression in epithelium or in normal colon mucosa stroma (Fig. 3). In agreement with

expression data (Fig. 2A), LOXL2 also exhibited some expression in endothelial cells,

as previously reported (29). ACAN and CDH13 were only present in the leading edge


16

of the tumor. In contrast, COL14A1 was highly expressed in normal crypt epithelial

cells, with very weak expression in normal stroma and no expression in tumoral tissues.

To study if desmoplastic changes found in the primary colon cancer lesion

would take place also in the metastatic lesions, we performed an expression analysis in

liver metastasis (Fig. 3). αSMA, TAGLN, TGM2 and LOXL2 were highly expressed at

the metastatic site, whereas ACAN, CDH13 and DKK3 showed reduced levels respect

to the primary tumors. This result suggests minor differences between the desmoplastic

site at the primary tumor and the metastatic site. In summary, 32 out of the 36 tested

proteins showed myofibroblast-like staining, with preferential stromal expression.

These 32 proteins would constitute a desmoplastic signature for colon CAFs.

Prognostic value of the individual genes according to external datasets

We selected 15 confirmed stromal markers to identify a molecular signature in order to

predict disease-specific survival and recurrence. First, we used the pooled cohort

GSE17538 dataset (232 patients with colon cancer). Patients were divided into “low

expression” with negative score and “high expression” with positive score. The

individual prognostic value for each gene was analyzed using Kaplan-Meier survival

curves and its significance was tested with log rank tests and univariate Cox

proportional hazards models. In the training set consisting of 232 patients, five genes,

LOXL2 (log-rank P < 0.0001) HR: 2.63 (CI, 1.68-4.12), TAGLN (log-rank P = 0.025)

HR:1.61 (CI, 1.06-2.45), MYL9 (log-rank P = 0.022) HR: 1.63 (1.07-2.49), POSTN

(log-rank P=0.008) HR:1.80 (CI, 1.16-2.81) and CSF1 (log-rank P=0.015) HR:0.60 (CI,

0.39-0.91) showed the strongest prognostic relevance (Supplementary Fig. S5).

We carried out a similar analysis for the prediction of disease-free survival in the

same cohort using 173 patients at stages I-III. The six genes showing better disease-free


17

prediction were LOXL2 log-rank P = 0.0002 HR: 4.18 (1.82-9.59), BGN log-rank P <

0.0001 HR: 5.13 (2.13-12.38), MYL9 log-rank P = 0.002 HR: 3.22 (1.46-7.10), POSTN

log-rank P = 0.0001 HR: 5.97 (2.11-16.94), PLOD2 log-rank P = 0.008 HR: 2.81(1.27-

6.21) and TAGLN log-rank P < 0.0001 HR: 4.86 (2.02-11.73) (Supplementary Fig.

S6). From these results, we selected LOXL2 and TAGLN for further validation, as they

were consistently reliable in overall and disease-free survival prognosis.

Validation of LOXL2 and TAGLN prognostic value in a different patient cohort

The prognostic value of LOXL2 and TAGLN was initially validated in a different

cohort of 70 human colon cancer samples followed for more than five years. We did

not include rectal adenocarcinomas as they are managed with neoadjuvant therapy prior

to surgery. We purified mRNA from the 70 paired FFPE tissues for real-time

quantitative PCR. We observed an increased expression for LOXL2 in 44% of the

tumoral samples respect to normal tissue. High expression of LOXL2 was associated

with overall survival P = 0.001 HR: 8.52 (1.90-38.29) (Fig. 4A) and higher recurrence

P = 0.001 HR: 5.38 (1.70-17.01) (Fig. 4B). Whereas high expression of TAGLN

correlated with overall survival log-rank P = 0.004 HR: 4.10 (1.45-11.62), it did not

show association with disease-free survival by QPCR (Fig. 4A, B). After adjustment for

age, sex, stage and grade, multivariate analysis confirmed that LOXL2 expression was

associated with poor prognosis and an independent prognostic factor for disease-free

and overall survival using GSE17538 (P = 0.0037, P <0.0001, respectively) and internal

mRNA data set (P = 0.0006, P = 0.0035, respectively) (Table 1). We did not find

association between LOXL2 expression and other clinicopathological characteristics

(Supplementary Table S6). Subsequently, we used another sample cohort of 121

patients for immunohistochemistry analysis. About 32% of the patients exhibited strong

LOXL2 expression (>50% of the sample stained positively) and 23% showed moderate


18

expression (20-50%) (Supplementary Table S1C). LOXL2 staining was mainly

nuclear in the stroma (Fig. 3 and Supplementary Fig S7). Stromal (nuclear) positive

staining correlated with overall survival P= 0.033 HR: 4.19 (1.22-14.41) and disease-

free survival P= 0.015 HR: 2.20 (1.14-4.25). Therefore, after testing and validation

using different cohorts, the expression of LOXL2 showed a significant association to

poor clinical outcome and higher recurrence in human colon cancer. Combinations of

LOXL2 with other markers did not improve its prognostic power.

Predictive prognostic value of LOXL2 for stage II and stage III patients

To explore the value of LOXL2 for risk assessment in stage II and III patients, we

analyzed recurrence of stage II patients from combined datasets GSE17538 (70 patients

at stage II) and GSE33113 (89 patients at stage II, with disease-free information) or

survival of patients with stage III from combined GSE17538 (76 patients at stage III)

and GSE12945 (21 patients at stage III). The prognostic value was validated by QPCR

and immunohistochemistry. Stage II patients showed shorter time to recurrence when

LOXL2 was highly expressed P value = 0.011 HR: 2.74 (1.22-6.20) (Fig. 5A). The 3-

year DFS was 73.9% (64.4-84.9%) for LOXL2-high versus 89.6% (82.6-97.2%) for

LOXL2-low patients. In line with these results, high LOXL2 expression was associated

with shorter disease-free using QPCR in 47 patients at stages II and III P = 0.001 HR:

5.69 (1.79-18.07). Using the immunohistochemistry values of the 55 patients at stage II

from the previous cohort (Supplementary Table S1C), the disease-free survival was

P= 0.005 HR: 4.35 (1.43-13.25) (Fig. 5A). Regarding overall survival for patients with

stage III, we found association between high expression of LOXL2 and overall survival

P = 0.004 HR: 3.40 (1.40-8.24) (Fig. 5B), confirmed by QPCR in patients at stage II

and III P = 0.001 (Fig. 5B). In summary, these results suggest that LOXL2 is a

suitable marker for poor prognosis, recurrence and patient stratification at stage II. This


19

promising value for risk-stratification will require further validation in a larger

independent cohort.


20

DISCUSSION

We used primary cultures of purified fibroblast populations isolated from colon cancer

patients to identify stromally-expressed LOXL2 as a protein displaying a significant

prognostic value. We demonstrated the prognostic power of LOXL2 using combined

external colon cancer datasets for biomarker training and, then, validation by Q-PCR

and immunohistochemistry on different cohorts of patients. Since twenty-five percent of

stage II colon cancer patients will have recurrence of disease within 5 years (8), we

were interested in testing the prognostic value in this group of patients. Using public

databases, LOXL2 identified a high-risk group within stage II patients that could benefit

from adjuvant chemotherapy. This prognostic power was confirmed by

immunohistochemistry. Therefore, LOXL2 could complement clinical guidelines for

high-risk definition (number of resected specimens, tumor differentiation, vascular

invasion, etc) and correct patient classification. A similar prognostic ability of LOXL2

was found for stage III patients. Although these patients are usually given adjuvant

treatment, some groups (elderly and frail patients) do not always receive treatment (30).

In this case, LOXL2 might help in clinical decisions for treating patients older than 75

years.

In addition, we identified a desmoplastic signature for colon cancer that

contained 32 CAFs-derived proteins showing cancer stromal staining. Proteins

contributing to desmoplastic lesions are characteristic of tumor type. We observed a

limited coincidence with mouse fibroblasts coming from a chemically-induced colon

cancer (19), among other coincidences we found: FN1, TPM2, TNC, COL1A2,

IGFBP7, FSTL1, THBS1, COL4A2, BGN and CSF1. We also observed CDH13, a

mesenchymal cadherin similar to CDH11, previously reported in mouse cancer

fibroblasts (19). However, affected biological functions were equivalent between mice


21

and human: TGFβ activation, ECM, cell adhesion and wound-healing. In any case, the

relatively low level of coincidence with mouse fibroblasts could be explained because

the mouse model does not recapitulate metastasis.

A protein profile similar to CAFs was obtained after activation of NFs with

TGFβ. Remarkably, many stromal proteins (BGN, TAGLN, TGM2, LOXL2) exhibit a

strong prognostic impact alone or in combination with other markers and deserve

further investigation. TAGLN contributes to the cross-linking and polymerization of

actin (31) and phenotype differentiation (32), showing a good predictive power that

would require larger cohorts for validation. BGN and TGM2 are associated to the

extracellular matrix.

Recently, LOXL2 has been shown to be critical in metastatic niche formation in

hepatocellular carcinoma (33) and a marker of poor prognosis in breast, gastric cancer

and squamous cell carcinomas (34-36), mostly associated to metastasis. In colon

cancer, LOXL2 was associated to less differentiated tumors (37), but no effect on

prognosis was reported. Here, LOXL2 alone showed to be an excellent molecular

classifier by two different approaches, QPCR and immunohistochemistry analysis.

Prognostic value was associated to the nuclear staining of stromal cells, although some

increase in cytoplasmic expression was observed in epithelial cells in the metastasis.

Therefore, the nuclear activity of LOXL2 seems essential for the association with poor

prognosis. However, it is remarkable that LOXL2 expression in complete tumors leads

to prognostic value, even if the differential expression was identified in fibroblasts of

the stromal compartment. This finding should facilitate the incorporation of LOXL2

expression analysis to clinical routine. The identification of potential molecular targets

that define prognosis can become even more important if these molecules can be

inhibited with some of the drugs in development. For LOXL2, a specific inhibitory


22

monoclonal antibody has been developed (38), which is currently in clinical trials for a

number of fibrotic conditions and cancers.

LOXL2 is a member of the lysyl oxidase gene family that catalyzes the

crosslinking of collagens and elastin in the extracellular matrix. This provokes an

increase of ECM stiffness and leads to increased activation of the PI3K pathway (39).

The cross-linking facilitates the recruitment of bone marrow-derived cells that release

cytokines and growth factors, which facilitate the cancer cell colonization in distant

organs (40). Moreover, LOXL2 interacts with Snail1 transcriptional repressor and

represses E-cadherin and cell-polarity genes by Snail-dependent and independent

mechanism (34, 41). In addition, a nuclear role for LOXL2 has been described (42).

LOXL2 works as a transcriptional co-repressor through its activity in histone H3

deamination (43). Furthermore, LOXL2 could oxidize H3 to remove the trymethylated

amino group located in the ε-position of Lysine 4 (44), which is usually related with

transcriptional control (45). The association of preferential nuclear staining in the

stroma with poor survival could be related with the role of LOXL2 as transcriptional

regulator in stromal cells.

In summary, our results support the relevance of CAFs-derived proteins, such as

BGN, TAGLN or LOXL2, in colon cancer desmoplastic analysis, prognosis and

survival. We identified a large number of colon cancer stroma-specific proteins and

LOXL2 as a single and highly predictive prognostic factor for overall survival and

relapse in colon cancer patients using different cohorts for training and validation.

Currently, lymph node involvement is admitted as the main prognostic factor and

determines the administration of adjuvant systemic therapy. However, in stage II other

tumour characteristics influence the indication of therapy. In this context, a biomarker


23

as LOXL2 could be useful in the decision to treat patients more aggressively, especially

the subgroup of stage IIA patients with LOXL2 overexpressed in tumour tissue.


24

GRANT SUPPORT

S. Torres was a recipient of a Juan de la Cierva programme. R.A. Bartolomé was

supported by a grant to established research groups of the Asociación Española Contra

el Cáncer (AECC). María Lopez-Lucendo was a recipient of a ProteoRed contract. A.

Peláez was a FPI fellow from the MINECO. This research was supported by grants to

established research groups of the “Asociación Española Contra el Cancer (AECC)”,

BIO2012-31023 from the Spanish Ministry of Economy and Competitiveness,

S2010/BMD-2344/ Colomics2 from the Comunidad de Madrid and the Grant PRB2

(IPT13/0001 - ISCIII-SGEFI / FEDER).


25

REFERENCES

1. Orimo A, Weinberg RA. Stromal fibroblasts in cancer: a novel tumor-promoting

cell type. Cell Cycle. 2006;5:1597-601.

2. Kalluri R, Zeisberg M. Fibroblasts in cancer. Nat Rev Cancer. 2006;6:392-401.

3. Navab R, Strumpf D, Bandarchi B, Zhu CQ, Pintilie M, Ramnarine VR, et al.

Prognostic gene-expression signature of carcinoma-associated fibroblasts in non-small

cell lung cancer. Proc Natl Acad Sci U S A. 2011;108:7160-5.

4. Xing F, Saidou J, Watabe K. Cancer associated fibroblasts (CAFs) in tumor

microenvironment. Front Biosci (Landmark Ed). 2010;15:166-79.

5. Kim JB, Stein R, O'Hare MJ. Tumour-stromal interactions in breast cancer: the

role of stroma in tumourigenesis. Tumour Biol. 2005;26:173-85.

6. Bremnes RM, Donnem T, Al-Saad S, Al-Shibli K, Andersen S, Sirera R, et al.

The role of tumor stroma in cancer progression and prognosis: emphasis on carcinoma-

associated fibroblasts and non-small cell lung cancer. J Thorac Oncol. 2011;6:209-17.

7. Josson S, Matsuoka Y, Chung LW, Zhau HE, Wang R. Tumor-stroma co-

evolution in prostate cancer progression and metastasis. Semin Cell Dev Biol.

2010;21:26-32.

8. Huijbers A, Tollenaar RA, v Pelt GW, Zeestraten EC, Dutton S, McConkey CC,

et al. The proportion of tumor-stroma as a strong prognosticator for stage II and III

colon cancer patients: validation in the VICTOR trial. Ann Oncol. 2013;24:179-85.

9. Lieubeau B, Garrigue L, Barbieux I, Meflah K, Gregoire M. The role of

transforming growth factor beta 1 in the fibroblastic reaction associated with rat

colorectal tumor development. Cancer Res. 1994;54:6526-32.

10. Pickup M, Novitskiy S, Moses HL. The roles of TGFbeta in the tumour

microenvironment. Nat Rev Cancer. 2013;13:788-99.

11. Ohlund D, Elyada E, Tuveson D. Fibroblast heterogeneity in the cancer wound.

J Exp Med. 2014;211:1503-23.

12. Mesker WE, Junggeburt JM, Szuhai K, de Heer P, Morreau H, Tanke HJ, et al.

The carcinoma-stromal ratio of colon carcinoma is an independent factor for survival

compared to lymph node status and tumor stage. Cell Oncol. 2007;29:387-98.

13. Tsujino T, Seshimo I, Yamamoto H, Ngan CY, Ezumi K, Takemasa I, et al.

Stromal myofibroblasts predict disease recurrence for colorectal cancer. Clin Cancer

Res. 2007;13:2082-90.


26

14. Ngan CY, Yamamoto H, Seshimo I, Tsujino T, Man-i M, Ikeda JI, et al.

Quantitative evaluation of vimentin expression in tumour stroma of colorectal cancer.

Br J Cancer. 2007;96:986-92.

15. Henry LR, Lee HO, Lee JS, Klein-Szanto A, Watts P, Ross EA, et al. Clinical

implications of fibroblast activation protein in patients with colon cancer. Clin Cancer

Res. 2007;13:1736-41.

16. Nakagawa H, Liyanarachchi S, Davuluri RV, Auer H, Martin EW, Jr., de la

Chapelle A, et al. Role of cancer-associated stromal fibroblasts in metastatic colon

cancer to the liver and their expression profiles. Oncogene. 2004;23:7366-77.

17. Berdiel-Acer M, Sanz-Pamplona R, Calon A, Cuadras D, Berenguer A, Sanjuan

X, et al. Differences between CAFs and their paired NCF from adjacent colonic mucosa

reveal functional heterogeneity of CAFs, providing prognostic information. Mol Oncol.

2014;8:1290-305.

18. De Boeck A, Hendrix A, Maynard D, Van Bockstal M, Daniels A, Pauwels P, et

al. Differential secretome analysis of cancer-associated fibroblasts and bone marrow-

derived precursors to identify microenvironmental regulators of colon cancer

progression. Proteomics. 2013;13:379-88.

19. Torres S, Bartolome RA, Mendes M, Barderas R, Fernandez-Acenero MJ,

Pelaez-Garcia A, et al. Proteome profiling of cancer-associated fibroblasts identifies

novel proinflammatory signatures and prognostic markers for colorectal cancer. Clin

Cancer Res. 2013;19:6006-19.

20. Maier T, Guell M, Serrano L. Correlation of mRNA and protein in complex

biological samples. FEBS Lett. 2009;583:3966-73.

21. Ross PL, Huang YN, Marchese JN, Williamson B, Parker K, Hattan S, et al.

Multiplexed protein quantitation in Saccharomyces cerevisiae using amine-reactive

isobaric tagging reagents. Mol Cell Proteomics. 2004;3:1154-69.

22. Smith JJ, Deane NG, Wu F, Merchant NB, Zhang B, Jiang A, et al.

Experimentally derived metastasis gene expression profile predicts recurrence and death

in patients with colon cancer. Gastroenterology. 2010;138:958-68.

23. Dolznig H, Rupp C, Puri C, Haslinger C, Schweifer N, Wieser E, et al. Modeling

colon adenocarcinomas in vitro a 3D co-culture system induces cancer-relevant

pathways upon tumor cell and stromal fibroblast interaction. Am J Pathol.

2011;179:487-501.


27

24. Beddy D, Mulsow J, Watson RW, Fitzpatrick JM, O'Connell PR. Expression and

regulation of connective tissue growth factor by transforming growth factor beta and

tumour necrosis factor alpha in fibroblasts isolated from strictures in patients with

Crohn's disease. Br J Surg. 2006;93:1290-6.

25. Ponten F, Jirstrom K, Uhlen M. The Human Protein Atlas--a tool for pathology.

J Pathol. 2008;216:387-93.

26. Cheadle C, Vawter MP, Freed WJ, Becker KG. Analysis of microarray data

using Z score transformation. J Mol Diagn. 2003;5:73-81.

27. Siegel PM, Massague J. Cytostatic and apoptotic actions of TGF-beta in

homeostasis and cancer. Nat Rev Cancer. 2003;3:807-21.

28. Calon A, Espinet E, Palomo-Ponce S, Tauriello DV, Iglesias M, Cespedes MV,

et al. Dependency of colorectal cancer on a TGF-beta-driven program in stromal cells

for metastasis initiation. Cancer Cell. 2012;22:571-84.

29. Bignon M, Pichol-Thievend C, Hardouin J, Malbouyres M, Brechot N, Nasciutti

L, et al. Lysyl oxidase-like protein-2 regulates sprouting angiogenesis and type IV

collagen assembly in the endothelial basement membrane. Blood. 2011;118:3979-89.

30. Sveen A, Nesbakken A, Agesen TH, Guren MG, Tveit KM, Skotheim RI, et al.

Anticipating the clinical use of prognostic gene expression-based tests for colon cancer

stage II and III: is Godot finally arriving? Clin Cancer Res. 2013;19:6669-77.

31. Kobayashi R, Kubota T, Hidaka H. Purification, characterization, and partial

sequence analysis of a new 25-kDa actin-binding protein from bovine aorta: a SM22

homolog. Biochem Biophys Res Commun. 1994;198:1275-80.

32. Untergasser G, Gander R, Lilg C, Lepperdinger G, Plas E, Berger P. Profiling

molecular targets of TGF-beta1 in prostate fibroblast-to-myofibroblast

transdifferentiation. Mech Ageing Dev. 2005;126:59-69.

33. Wong CC, Tse AP, Huang YP, Zhu YT, Chiu DK, Lai RK, et al. Lysyl oxidase-

like 2 is critical to tumor microenvironment and metastatic niche formation in

hepatocellular carcinoma. Hepatology. 2014;60:1645-58.

34. Moreno-Bueno G, Salvador F, Martin A, Floristan A, Cuevas EP, Santos V, et

al. Lysyl oxidase-like 2 (LOXL2), a new regulator of cell polarity required for

metastatic dissemination of basal-like breast carcinomas. EMBO Mol Med. 2011;3:528-

44.


28

35. Peinado H, Moreno-Bueno G, Hardisson D, Perez-Gomez E, Santos V,

Mendiola M, et al. Lysyl oxidase-like 2 as a new poor prognosis marker of squamous

cell carcinomas. Cancer Res. 2008;68:4541-50.

36. Peng L, Ran YL, Hu H, Yu L, Liu Q, Zhou Z, et al. Secreted LOXL2 is a novel

therapeutic target that promotes gastric cancer metastasis via the Src/FAK pathway.

Carcinogenesis. 2009;30:1660-9.

37. Fong SF, Dietzsch E, Fong KS, Hollosi P, Asuncion L, He Q, et al. Lysyl

oxidase-like 2 expression is increased in colon and esophageal tumors and associated

with less differentiated colon tumors. Genes Chromosomes Cancer. 2007;46:644-55.

38. Barry-Hamilton V, Spangler R, Marshall D, McCauley S, Rodriguez HM, Oyasu

M, et al. Allosteric inhibition of lysyl oxidase-like-2 impedes the development of a

pathologic microenvironment. Nat Med. 2010;16:1009-17.

39. Levental KR, Yu H, Kass L, Lakins JN, Egeblad M, Erler JT, et al. Matrix

crosslinking forces tumor progression by enhancing integrin signaling. Cell.

2009;139:891-906.

40. Erler JT, Bennewith KL, Cox TR, Lang G, Bird D, Koong A, et al. Hypoxia-

induced lysyl oxidase is a critical mediator of bone marrow cell recruitment to form the

premetastatic niche. Cancer Cell. 2009;15:35-44.

41. Peinado H, Del Carmen Iglesias-de la Cruz M, Olmeda D, Csiszar K, Fong KS,

Vega S, et al. A molecular role for lysyl oxidase-like 2 enzyme in snail regulation and

tumor progression. EMBO J. 2005;24:3446-58.

42. Cano A, Santamaria PG, Moreno-Bueno G. LOXL2 in epithelial cell plasticity

and tumor progression. Future Oncol. 2012;8:1095-108.

43. Herranz N, Dave N, Millanes-Romero A, Morey L, Diaz VM, Lorenz-Fonfria V,

et al. Lysyl oxidase-like 2 deaminates lysine 4 in histone H3. Mol Cell. 2012;46:369-76.

44. Iturbide A, Garcia de Herreros A, Peiro S. A new role for LOX and LOXL2

proteins in transcription regulation. FEBS J. 2015;282:1768-73.

45. Campos EI, Reinberg D. Histones: annotating chromatin. Annu Rev Genet.

2009;43:559-99.


29

Table 1: Univariate and multivariable analysis for LOXL2 in disease-free and overall survival in colon cancer patients

TRAINING COHORT (GSE17538) VALIDATION COHORT (RNA_ DATA)

Disease-free survival Overall survival Disease-free survival Overall survival

HR (95% CI) P-value n

(events) HR (95% CI) P-value n

(events) HR (95% CI) P-

value n

(events) HR (95% CI) P-value n

(events) Univariate

LOXL2 negative 1 86(7) 1 113(27) 1 33(4) 1 39 (2) LOXL2 positive 4.18 (1.82-9.59) 0.0002 87(28) 2.63 (1.68-4.12) <0.0001 119(66) 5.38 (1.70-17.01) 0.004 22(11) 8.52 (1.90-38.29) 0.0052 31(13)

Multivariable LOXL2 positive 3.49 (1.50-8.09) 0.0037 3.19 (2.00-5.07) <0.0001 8.69 (2.51-30.07) 0.0006 11.41 (2.23-58.32) 0.0035

Sex 0.71 (0.34-1.48) 0.3656 1.09 (0.69-1.71) 0.7086 0.88 (0.28-2.76) 0.8283 0.80 (0.22-2.85) 0.7323 Age 0.99 (0.97-1.07) 0.4951 1.02 (1.00-1.04) 0.0175 1.04 (0.99-1.08) 0.1017 0.99 (0.93-1.05) 0.6980

Stage III vs I/II 2.41 (1.18-4.93) 0.0162 1.78 (0.99-3.21) 0.0557 3.29 (1.00-10.81) 0.0499 9.64 (1.53-60.63) 0.0157 IV vs I/II 9.21(5.23-16.2) <0.0001 13.07 (2.14-79.81) 0.0054

Grade

mod. dif. vs well dif 0.96 (0.29-3.21) 0.9497 0.94 (0.37-2.37) 0.8938 4.38 (0.48-39.96) 0.1899 0.48 (0.07-3.61) 0.4791

poorly. dif. vs well dif

1.31 (0.32-5.40) 0.7059 1.46 (0.53-4.08) 0.4670 18.96(1.31-274.9) 0.0310 0.38 (0.04-3.62) 0.3996

undif. vs well dif 1.48 (0.48-4.52) 0.4908 1.985 (0.13-30.02) 0.6329 Multivariable analysis adjusted for age, gender, AJCC stage (I, II, III, IV) and grade of differentiation (well differentiated, moderately differentiated, poorly differentiated and undifferentiated). Abbreviations: HR = Hazard ratio; CI = confidence interval.


30

Legend to Figures

Figure 1. Validation of CAFs deregulated proteins. (A) Whole cellular extracts or

concentrated protein samples from conditioned medium of CAFs and NFs were

separated on SDS-PAGE gels, transferred to nitrocellulose membranes and probed with

the indicated antibodies. Tubulin was used as a loading control. Protein abundance was

quantified by densitometry to compare the expression with the iTRAQ ratios. (B)

cDNAs synthesized from total RNA obtained from NFs and CAFs were subjected to

QRT-PCR using specific primers for the selected genes and 18S rRNA for

normalization. BJh-TERT cells were starved in serum-free medium, treated with TGFβ

(5 ng/mL) for 24 h and lysed. The extracts were analyzed by QPCR (C) and Western

blot (D) using specific primers and the indicated antibodies. Data represent the mean ±

SD of three experiments. (*, p < 0.05, **, p < 0.01; ***, p < 0.001).

Figure 2. Stromal specificity of CAFs proteins. Differentially-expressed genes in

CAFs vs NFs (60 up-regulated (A) and 40 down-regulated (B)) are represented by heat-

maps of scaled gene-expression levels according to a dataset for: epithelial cells:

EPCAM+, leukocytes: CD45+, fibroblasts: FAP+ and endothelial cells: CD31+ (28).

Color scale represents the value of the z-score calculated in the dataset. Red indicates a

higher z-score value, whereas blue indicates a lower value. (C) cDNAs synthesized

from total RNA from NFs, CAFs and epithelial colon cancer cell lines (KM12SM,

KM12C, SW480 and SW620) were subjected to QPCR as in Figure 1. Data represent

the mean ± SD of three experiments.

Figure 3. Expression of stromal biomarkers in paired tissues of colon cancer

patients. Immunohistochemical staining for ACAN, αSMA, COL14A1, CDH13,


31

DKK3, TAGLN, TGM2 and LOXL2 in normal mucosa, primary colon cancer and liver

metastasis. Pictures were taken at x200 magnification.

Figure 4. Prognostic value of LOXL2 and TAGLN in colon cancer. Biomarker

validation was carried out in two steps: 1. By QPCR in a different cohort of 70 colon

cancer patients: (A) overall survival and (B) disease-free. 2. By immunohistochemistry

analysis using 121 samples in a tissue microarray slide. Positive expression was

associated to nuclear staining of the stromal fibroblasts. (C) Overall survival and (D)

disease-free survival. Kaplan-Meier curves were performed using the log-rank test. The

Hazard Ratio was based on the Cox model.

Figure 5. Risk stratification of colon cancer patients at stages II and III using

LOXL2 expression. (A) Tumor recurrence-free survival was determined using:

GSE17538+GSE31331 datasets of colon patients at stage II, QPCR with a different

cohort of 48 patients at stage II-III or immunohistochemistry (IHC) with 55 samples

from patients at stage II. (B) Overall survival was determined using

GSE17538+GSE12945 datasets of patients at stage III or QPCR with samples from 48

patients at stage II-III. Kaplan-Meier curves were performed using the log-rank test. The

Hazard Ratio was based on the Cox model.

63

180

28

75

17

180

35

75

100

63

A

POSTN 1.0 0.3

1.0 4.9

DKK3

1.0 2.4 LOXL2

ACAN 1.0 2.4

1.0 5.3 TGM2

CDH13 1.0 3.5

TUB

COL14A1 1.0 0.3

1.0 8.0 TAGLN

1.0 0.6 0.5 0.2 0.3 0.1

1.0 1.3 0.9 1.9 0.7 2.1

1.0 2.9 1.4 2.7 0.7 1.8

1.0 2.1 11 1.9 1.1 2.0

1.0 0.1 0.5 0.0 0.3 0.0

1.0 2.4 0.9 6.9 2.0 3.4

1.0 3.6 2.6 4.5 3.1 7.5

1.0 6.0 0.9 2.4 0.6 6.9

***

0.03

0.06

0.13

0.25

0.50

1.00

2.00

4.00

8.00

16.00

CAF

s/N

Fs m

RN

A

expr

essi

on l

evel

s (a

.u.)

*** ***

*** ***

*** ***

* * * *

* *

* *

B Pool

Figure 1

TGFβ:

TUB

1.0 1.5

D C

63

180

28

75

17

180

35

75

100

63

Conditioned m

edium

LOXL2

POSTN 1.0 0.2

1.0 12.0 75

DKK3 1.0 6.1

35

100

63

α-SMA

S100A4

COL14A1

Cellular extract

TAGLN

TGM2

VIM

1.0 70.3

1.0 0.7

1.0 15.8

1.0 3.5

1.0 100

48

17

180

28

75

48

35 17

KDa KDa

KDa

Individual samples

Conditioned m

edium

Cellular extract

0.06

0.13

0.25

0.50

1.00

2.00

4.00

8.00

16.00

32.00

64.00

TGFβ

/veh

mR

NA

ex

pres

sion

leve

ls (a

.u.)

*** ***

***

***

*** *** *** ** *

*** **

**

Figure 2

C

0

4

8

12

16

20

mR

NA

exp

ress

ion

le

vels

(a.u

.)

CAFs

KM12C

KM12SM

SW480

SW620

B

Dow

n-re

gula

ted

prot

eins

A

Up-r

egul

ated

pro

tein

s

4 -4

High Low

Figure 3

C

olon

Can

cer

(prim

ary)

N

orm

al

Col

on

CDH13 ACAN COL14A1

TGM2 TAGLN DKK3

α-SMA

LOXL2

Li

ver

Met

asta

sis

C

olon

Can

cer

(prim

ary)

N

orm

al

Col

on

Li

ver

Met

asta

sis

Dise

ase

free

sur

viva

l 1.0

0.8

0.6

0.4

0.2

0.0

0

Time (months) 20 80 40 60

P = 0.001

HR = 5.38 (1.70-17.01)

n = 33

n = 22

Dise

ase

free

sur

viva

l 1.0

0.8

0.6

0.4

0.2

0.0

0


P = 0.428

HR = 1.57 (0.53-4.58)

n = 41

n = 14

86.9 (75.7-99.7)

52.5 (34.9-79.1)

Ove

rall

surv

ival

1.0

0.8

0.6

0.4

0.2

0.0

Time (months) 0 25 100 125 50 75

n = 31

n = 39

P = 0.001

HR = 8.52 (1.90-38.29)

Ove

rall

surv

ival

1.0

0.8

0.6

0.4

0.2

0.0

Time (months) 0 25 100 125 50 75

n = 20

n = 50

P = 0.004

HR = 4.10 (1.45-11.62)

97.3 (92.2-100.0)

72.5 (57.8-90.8)

Figure 4

mRNA from paired FFPE tissues

A Negative Positive

B Negative Positive

TAGLN LOXL2

C 1.0

0.8

0.6

0.4

0.2

0.0

Ove

rall

surv

ival

n = 67

n = 54

Time (months) 0 20 80 40 60 120 100

1.0

0.8

0.6

0.4

0.2

0.0

Dise

ase

free

sur

viva

l

n = 48

n = 45

0

Time (months) 20 80 40 60 120 100

D

P = 0.033

HR = 4.19 (1.22-14.41)

P = 0.015

HR = 2.20 (1.14-4.25)

Negative Positive

Negative Positive

LOXL2

IHC analysis

A 1.0

0.8

0.6

0.4

0.2

0.0

Dis

ease

fre

e su

rviv

al

n = 82

n = 77

0

Time (months) 20 100 120 40 60 80

P = 0.011

HR = 2.74 (1.22-6.20)

Stage II

GSE33113/GSE17538

Low High

89.6 (82.6-97.2)

73.9 (64.4-84.9)

Dis

ease

fre

e su

rviv

al

1.0

0.8

0.6

0.4

0.2

0.0

0


P = 0.001

HR = 5.69 (1.79-18.07)

n = 28

n = 19


Negative Positive

Stage II + III

1.0

0.8

0.6

0.4

0.2

0.0

Dis

ease

fre

e su

rviv

al

n = 28

0


n = 27

120 100

Stage II

P = 0.005

HR = 4.35 (1.43-13.25)

IHC analysis

Negative Positive

Figure 5

Ove

rall

surv

ival

1.0

0.8

0.6

0.4

0.2

0.0

Time (months) 0 25 100 125 50 75

n = 20

n = 28

P = 0.001

Negative Positive


Stage II + III B

1.0

0.8

0.6

0.4

0.2

0.0

Ove

rall

surv

ival

n = 40

0

Time (months) 20 100 120 40 60 80

P = 0.004

n = 57

HR = 3.40 (1.40-8.24)

Stage III

GSE12945/GSE17538

Low High

81.7 (69.1-96.4)

51.6 (38.9-68.4)

Documents

LOXL2 in colon cancer prognosis - Digital CSICdigital.csic.es/bitstream/10261/128643/1/LOXL2-Casal.pdf · 2020. 11. 16. · in the stratification of high- and low-risk colon cancer