Author Manuscript Published OnlineFirst on May 29, 2018; DOI: 10.1158/1078-0432.CCR-17-3533 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.
1
Title:
2
Glycosyltransferase gene expression identifies a poor prognostic colorectal cancer subtype
3
associated with mismatch repair deficiency and incomplete glycan synthesis
4 5
Authors and affiliations:
6
Masaru Noda1,2, Hirokazu Okayama1, Kazunoshin Tachibana2, Wataru Sakamoto1, Katsuharu Saito1,
7
Aung Kyi Thar Min1, Mai Ashizawa1, Takahiro Nakajima1, Keita Aoto1, Tomoyuki Momma1, Kyoko
8
Katakura3, Shinji Ohki1, and Koji Kono1
9 10
1
11
Gastroenterology, Fukushima Medical University School of Medicine, Japan
Department of Gastrointestinal Tract Surgery,
2
Departmet of Breast Surgery,
3
Department of
12 13
Running title:
14
Prognostic CRC subtypes based on glycosyltransferase profile
15 16
Keywords:
17
Colorectal cancer, glycosyltransferase, cancer-associated glycans, gene expression profiling, molecular
18
diagnosis and prognosis
19 20
Additional information:
21
This work was supported by JSPS KAKENHI Grant Numbers 15K10143 and 25870582. Hirokazu
22
Okayama and Masaru Noda were supported by Takeda Science Foundation.
23 24
Correspondence: Hirokazu Okayama, M. D., Ph. D.
25
Department of Gastrointestinal-tract Surgery, Fukushima Medical University School of Medicine
26
1 Hikarigaoka, Fukushima city, Fukushima, 960-1295, Japan.
27
TEL: +81-24-547-1259
28
FAX: +81-24-547-1980
29
E-mail:
[email protected]
30 31
The authors declare no conflicts of interest associated with this manuscript.
1
Downloaded from clincancerres.aacrjournals.org on June 14, 2018. © 2018 American Association for Cancer Research.
Author Manuscript Published OnlineFirst on May 29, 2018; DOI: 10.1158/1078-0432.CCR-17-3533 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.
32 33
Translational Relevance
34
Here we report the identification and validation of a poor prognostic subgroup, displaying
35
mismatch repair deficiency (dMMR) and decreased GALNT6 levels, based upon glycosyltransferase
36
expression and methylation profiles in multiple cohorts containing a total of 4223 samples. We show that
37
downregulation
38
precancerous/preinvasive neoplasia to invasive carcinoma in a certain subset of tumors that frequently
39
exhibit dMMR. Those transcriptional analyses were robustly recapitulated by immunohistochemistry on
40
403 specimens, where tumors lacking GALNT6 protein was associated with dMMR and poor patient
41
outcomes. Strikingly, loss of GALNT6 protein expression and decreased GALNT6 mRNA expression
42
each discriminated postoperative stage III patients with poor survival. Our study highlights the
43
possibility of GALNT6 as a novel prognostic biomarker for CRC and suggests its contribution to
44
colorectal carcinogenesis through incomplete glycan synthesis.
of
GALNT6
via
epigenetic
silencing
occurs
during
transition
from
45
2
Downloaded from clincancerres.aacrjournals.org on June 14, 2018. © 2018 American Association for Cancer Research.
Author Manuscript Published OnlineFirst on May 29, 2018; DOI: 10.1158/1078-0432.CCR-17-3533 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.
46
Abstract
47
Purpose: We aimed to discover glycosyltransferase gene (glycogene)-derived molecular subtypes of
48
colorectal cancer (CRC) associated with patient outcomes.
49
Experimental Design: Transcriptomic and epigenomic datasets of non-tumor, pre-cancerous, cancerous
50
tissues and cell lines with somatic mutations, mismatch repair status, clinicopathological and survival
51
information, were assembled (n=4223) and glycogene profiles were analyzed. Immunohistochemistry
52
for a glycogene, GALNT6, was conducted in adenoma and carcinoma specimens (n=403). The
53
functional role and cell surface glycan profiles were further investigated by in vitro loss-of-function
54
assays and lectin microarray analysis.
55
Results: We initially developed and validated a 15-glycogene signature that can identify a
56
poor-prognostic subtype, which closely related to deficient mismatch repair (dMMR) and GALNT6
57
downregulation. The association of decreased GALNT6 with dMMR was confirmed in multiple datasets
58
of tumors and cell lines, and was further recapitulated by immunohistochemistry, where approximately
59
15% tumors exhibited loss of GALNT6 protein. GALNT6 mRNA and protein was expressed in
60
premalignant/preinvasive lesions but was subsequently downregulated in a subset of carcinomas,
61
possibly through epigenetic silencing. Decreased GALNT6 was independently associated with poor
62
prognosis in the immunohistochemistry cohort and an additional microarray meta-cohort, by
63
multivariate analyses, and its discriminative power of survival was particularly remarkable in stage III
64
patients. GALNT6 silencing in SW480 cells promoted invasion, migration, chemoresistance and
65
increased cell surface expression of a cancer-associated truncated O-glycan, Tn-antigen.
66
Conclusions: The 15-glycogene signature and the expression levels of GALNT6 mRNA and protein
67
each serve as a novel prognostic biomarker, highlighting the role of dysregulated glycogenes in
68
cancer-associated glycan synthesis and poor prognosis.
69 70 71
Introduction
72
Despite major advances in diagnosis and treatment, colorectal cancer (CRC) remains one of
73
the leading causes of cancer-death worldwide (1, 2). CRC is commonly grouped into two categories:
74
tumors with microsatellite instability (MSI), caused by defective function of the DNA mismatch repair
75
(MMR) system, and tumors that are microsatellite stable but exhibiting chromosomal instability (CIN)
76
(3-5). The majority of CRC (~85%) follows the CIN pathway, often accompanied by KRAS mutations
77
and TP53 inactivation. Approximately 15% of CRCs that exhibit deficient MMR (dMMR) frequently
78
carry BRAF mutations (3, 5). Clinical trials implicated MMR status as a potential therapeutic classifier
3
Downloaded from clincancerres.aacrjournals.org on June 14, 2018. © 2018 American Association for Cancer Research.
Author Manuscript Published OnlineFirst on May 29, 2018; DOI: 10.1158/1078-0432.CCR-17-3533 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.
79
for stage II patients in the adjuvant setting (6-8). In the metastatic setting, KRAS and BRAF mutations are
80
used for predicting unresponsiveness to EGFR-targeted therapies (4). Despite those increasing
81
knowledge, clinicopathological staging system remains the only prognostic classification currently used
82
in clinical practice. However, clinicopathologically similar tumors can strikingly differ in clinical
83
behaviors that likely reflect molecular heterogeneity. Although it is recommended that stage III patients
84
receive postoperative chemotherapy, approximately 30-40% of patients develop recurrence even after
85
standard treatment (9-12).
86
Glycosylation is a common post-translational modification that involves sequential addition of
87
single sugar residues to target structures, resulting in glycan elongation. Further chemical modifications
88
and branching can finally form a vast array of glycan structures (13). Those procedures are regulated by
89
the multienzymatic reaction of glycosyltransferases, whose encoding genes, namely “glycogenes”, are
90
equivalent to 1% of human genome. Cell surface glycans undergo changes during malignant
91
transformation and tumor progression accompanied by distinct biological functions and unique tumor
92
phenotypes, thereby making glycans as potential cancer biomarkers (13, 14). For instance, a
93
cancer-associated glycan epitope, CA19-9, called sialyl Lewis A (sLea), is routinely utilized as a serum
94
tumor marker (15). CA19-9 and several other cancer-associated glycans, including sialyl Lewis X (sLex),
95
sialyl Tn, Tn and T antigens, are associated with tumorigenesis and poor prognosis of CRC (13, 16).
96
Such glycans can be attributed to transcriptional dysregulation of glycosyltransferases that has been
97
postulated as two principal mechanisms, “incomplete synthesis” and “neo-synthesis” (13, 16-18). Some
98
glycogenes are repressed by epigenetic silencing during early stages of tumorigenesis, which lead to the
99
biosynthesis of truncated structures, such as Tn and STn expression, called incomplete synthesis.
100
Conversely, in the neo-synthesis process, transcriptionally induced glycogenes can result in the de novo
101
expression of cancer-antigens, such as sLea and sLex.
102
In the present study, with the aim to discover distinct classes of CRC on the basis of the
103
expression of glycosyltransferases, we compiled an extensive number of transcriptomic profiles obtained
104
from multiple cohorts by integrating other available data sources, including mutations, MMR status,
105
methylation, protein expression as well as non-tumor, pre-cancerous, preinvasive and cancerous samples.
106
We initially described a novel subtype based upon clustering analysis of genome-wide “glycogene”
107
expression patterns, and this led us to identify a glycogene, GALNT6 as a promising biomarker for
108
disease prognosis. Moreover, we found the functional characteristics of GALNT6 involved in tumor
109
progression and glycosylation, suggesting the contribution of epigenetic silencing of GALNT6 to
110
colorectal carcinogenesis, through the incomplete synthesis of cell surface glycans.
111
4
Downloaded from clincancerres.aacrjournals.org on June 14, 2018. © 2018 American Association for Cancer Research.
Author Manuscript Published OnlineFirst on May 29, 2018; DOI: 10.1158/1078-0432.CCR-17-3533 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.
112
Materials and Methods
113
Microarray data analysis, hierarchical clustering and assembly of the TCGA dataset
114
All microarray and methylation array data are publicly available in the Gene Expression
115
Omnibus (GEO) database (http://www.ncbi.nlm.nih.gov/geo) as shown in Supplementary Table S1. We
116
utilized the normalized expression values obtained from each dataset. If a gene is represented by
117
multiple probes, they were averaged. To generate a list of glycogenes, official gene symbols and Entrez
118
Gene
119
http://acgg.asia/ggdb2/).
120
Affymetrix_3PRIME_IVT_ID
121
(http://david.abcc.ncifcrf.gov/home.jsp) as shown in Supplementary Table S2.
IDs
for
190
glycogenes Among
were
190
obtained
glycogenes,
using
from
185
DAVID
GGDB
unique
(GlycoGene
genes
were
Bioinfomatics
DataBase;
converted
to
Resources6.7
122
Hierarchical clustering was initially performed using an Affymetrix dataset, GSE17536,
123
consisted of 177 CRC patients with survival information. Expression levels of 185 glycogenes were
124
median-centered, and then genes and samples were subjected to an unsupervised clustering by the
125
centroid linkage method using the Cluster3.0 and the Java Treeview program (19). Among 39
126
differentially expressed genes between two major clusters (Cluster A vs B, p