MiRNA-mediated TUSC3 Deficiency Enhances UPR

ARTICLE
DOI: 10.1038/s41467-018-07561-8 OPEN
miRNA-mediated TUSC3 deficiency enhances UPR

and ERAD to promote metastatic potential of
NSCLC
Young-Jun Jeon1,2, Taewan Kim 1, Dongju Park1, Gerard J. Nuovo1, Siyeon Rhee 3, Pooja Joshi1,
Bum-Kyu Lee4, Johan Jeong5, Sung-suk Suh6, Jeff E. Grotzke7, Sung-Hak Kim8,9, Jieun Song1, Hosung Sim1,
Yonghwan Kim 10, Yong Peng1,11, Youngtae Jeong2, Michela Garofalo1,12, Nicola Zanesi13, Jonghwan Kim4,
1234567890():,;
Guang Liang14, Ichiro Nakano15, Peter Cresswell16, Patrick Nana-Sinkam17, Ri Cui1,14 & Carlo M. Croce1,13
Non-small cell lung carcinoma (NSCLC) is leading cause of cancer-related deaths in the
world. The Tumor Suppressor Candidate 3 (TUSC3) at chromosome 8p22 known to be
frequently deleted in cancer is often found to be deleted in advanced stage of solid tumors.
However, the role of TUSC3 still remains controversial in lung cancer and context-dependent
in several cancers. Here we propose that miR-224/-520c-dependent TUSC3 deficiency
enhances the metastatic potential of NSCLC through the alteration of three unfolded protein
response pathways and HRD1-dependent ERAD. ATF6α-dependent UPR is enhanced
whereas the affinity of HRD1 to its substrates, PERK, IRE1α and p53 is weakened. Conse-
quently, the alteration of UPRs and the suppressed p53-NM23H1/2 pathway by TUSC3
deficiency is ultimately responsible for enhancing metastatic potential of lung cancer. These
findings provide mechanistic insight of unrecognized roles of TUSC3 in cancer progression
and the oncogenic role of HRD1-dependent ERAD in cancer metastasis.
1 Comprehensive Cancer Center, The Ohio State University, Columbus, OH 43210, USA. 2 Stanford Cancer Institute, Stanford University School of Medicine,
Stanford, CA 94305, USA. 3 Department of Biology, Stanford University, Stanford, CA 94305, USA. 4 Institute for Cellular and Molecular Biology, Center for
Systems and Synthetic Biology, The University of Texas at Austin, Austin, TX 78712, USA. 5 Department of Pathology, Stanford University, Stanford, CA
94305, USA. 6 Department of Biosciences, Mokpo National University, Muan 58554, South Korea. 7 Departments of Immunobiology, Yale University School
of Medicine, New Haven, CT 06520, USA. 8 Department of Animal Science, College of Agriculture and Life Sciences, Chonnam National University, Gwangju
61186, Korea. 9 Gwangju Center, Korea Basic Science Institute, Gwangju 61186, Korea. 10 Department of Life System, Sookmyung Woman’s University, Seoul
140-742, Republic of Korea. 11 Department of Thoracic Surgery, State Key Laboratory of Biotherapy, West China Hospital, Sichuan University, and
Collaborative Innovation Center for Biotherapy, 610041 Chengdu, China. 12 Transcriptional Networks in Lung Cancer Group, Cancer Research United Kingdom
Manchester Institute, University of Manchester, Manchester M20 4BX, United Kingdom. 13 Department of Cancer Biology and Genetics, The Ohio State
University, Columbus, OH 43210, USA. 14 School of Pharmaceutical Sciences, Wenzhou Medical University, Wenzhou, 325035 Zhejiang, China.
15 Department of Neurosurgery UAB Comprehensive Cancer Center, University of Alabama at Birmingham, Birmingham, AL 35294, USA. 16 Departments of
Immunobiology, Howard Hughes Medical Institute, Yale University School of Medicine, New Haven, CT 06520, USA. 17 Division of Pulmonary, Allergy,
Critical Care and Sleep Medicine, Medical Oncology, The Ohio State University, Columbus, OH 43210, USA. These authors contributed equally: Young-Jun
Jeon, Taewan Kim. Correspondence and requests for materials should be addressed to R.C. (email: [email protected])
or to C.M.C. (email: [email protected])
NATURE COMMUNICATIONS | (2018)9:5110 | DOI: 10.1038/s41467-018-07561-8 | www.nature.com/naturecommunications 1

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-07561-8
L
ung cancer is leading cause of cancer deaths in the world. Results
About 80% of all lung cancer are determined as a Non-Small TUSC3 deficiency by miR-224/-520c in lung cancer. To analyze
Cell Lung Carcinoma (NSCLC) type. The reasons for the TUSC3 expression in lung cancer, we employed 20 normal, 14
poor overall 5 year survival in lung cancer are multifactorial primary and 21 lymph node metastasized lung tumor tissues.
including late clinical presentation of disease, occult metastatic TUSC3 mRNA expression using qRT-PCR analysis were found to
disease and few targeted therapeutics. Cancer metastasis is a be reduced in metastatic cancer patient tissues compared to pri-
complex, multistep process based in reciprocal interactions mary and/or normal lung tissues (Supplementary Fig. 1a). The
between tumor cells and their microenvironment. Although there downregulation of TUSC3 protein was further confirmed by
have been massively studied on this process, the sequence of immunohistochemistry using 50 sets of the primary and its cor-
critical events and molecular mechanisms for cancer metastasis responding metastasized tumor samples (Fig. 1a and Supple-
remain poorly understood1–3. mentary Fig. 1b). Moreover, TUSC3 downmodulation was
Cells have Endoplasmic Reticulum (ER) quality control associated with poor cancer survival rate (n = 223, Fig. 1b),
machineries to monitor the proper folding status of a polypeptide. suggesting that TUSC3 could play an important role in NSCLC
The major response to the accumulation of unfolded and/or metastasis.
misfolded proteins referred to as ER stress is an activation of A target search using in silico tools predicted that miR-224 and
Unfolded Protein Response (UPR) pathway. There are three miR-520c were likely to target TUSC3 at 102-108bp and 22-28bp
commonly described UPR pathways in ER-stressed cells. The downstream of the 3’UTR, respectively (Supplementary Fig. 1c).
contribution of these pathways on tumorigenesis and cancer The luciferase activity using TUSC3 3’UTR was decreased by
metastasis are context-dependent and vary based on the duration either miR-224 or miR-520c overexpression but rescued in
and strength of the ER stress4,5. To date, PERK-eIF2α axis and luciferase activity with 3’UTR deleted in the binding sites (Fig. 1c).
the IRE1α pathway have been described to enhance and suppress Consistently, miR-224 and -520c overexpression suppressed
cancer progression in different contexts4–6. Additionally, ATF6α- TUSC3 mRNA and protein levels, whereas the suppression of
dependent UPR has shown cytoprotective effects leading to those miRNAs enforced by anti-miRNA inhibitors enhanced
oncogenic roles in tumorigenesis. Consistently, several in vitro TUSC3 expression (Fig. 1d, e and Supplementary Fig. 1d). We
and in vivo experiments suggest that ER molecular chaperons next sought to determine if there is an inverse correlation
induced by active ATF6α play critical roles in cancer progres- between the miRNAs and TUSC3 gene expression in vivo. We
sion7–10. Endoplasmic Reticulum Associated Degradation performed an in situ hybridization analysis with 5’-dig-labeled
(ERAD) is a constitutive protein degradation pathway and is LNA probes for the miRNAs and found that both miR-224 and
highly activated upon ER stress induction. HRD1 protein is an E3 miR-520c expression was significantly enhanced in metastasized
ubiquitin ligase in mammalian ERAD machinery11. Although tumor tissues (Fig. 1f and Supplementary Fig. 1e). Moreover, co-
HRD1-specific substrates have been identified, the exact roles of expression analyses using IHC assay followed by in situ
HRD1 in cancer progression are unclear. HRD1 degraded the hybridized samples revealed that there was an inverse correlation
gp78 protein known as a pro-metastatic ERAD E3 ubiquitin between miR-224/-520c and TUSC3 expression in lung cancer
ligase12,13. Additionally, the HRD1 has been shown to limit the patient samples (Fig. 1g, h and Supplementary Fig. 1f, g). These
progression of breast cancers14. However, HRD1-dependent data suggest that miR-224/-520c are responsible for TUSC3
ERAD showed anti-apoptotic activity against ER-stress induced downmodulation and could be used as a biomarker and a
cell death15. Also, HRD1 protein could promote cancer pro- therapeutic target for NSCLC. Additionally, TUSC3 proteins were
gression by cytosolic p53 degradation through its E3 ubiquitin found to be undetectable in 18 of the 50 paired samples, which
ligase activity13,16,17 could be mediated by other mechanisms such as chromosomal
The loss of the chromosomal arm 8p where TUSC3 gene deletion or hypermethylation on TUSC3 gene, as well as our
locates is associated with cancer progression and TUSC3 defi- proposed mechanism by miR-224 and miR-520c in lung
ciency is frequently observed in advanced stage tumors18–21. cancer18,21,27.
The roles of TUSC3 in cancer progression showed context-
dependent manner in several solid tumors. TUSC3 was initially
identified as a homozygous deleted gene in metastasized prostate TUSC3 deficiency enhances the metastatic potential of lung
cancer patients and its deficiency upregulated cancer progres- cancer. To understand the roles of TUSC3 downmodulation in
sion and tumorigenesis in ovarian, prostate, glioblastoma and NSCLC metastasis, we generated TUSC3 knock-down (KD)
pancreatic cancers20–24. However, it has been also reported that H460 and A549 cells using shTUSC3s and found that TUSC3
TUSC3 was associated with genetic amplification or enhanced KD cells showed enhanced migration and invasion. Next, we
cancer progression in head and neck cancer and colorectal rescued TUSC3 expression by overexpressing TUSC3 mutants
cancer20,25,26. In particular, the controversial observations were (smTUSC3) harboring silent mutations for the shRNA target
reported in TUSC3-dependent oncogenesis in lung cancer27–30. sequences and showed a decrease in migration and invasion
To date, none of the plausible mechanisms investigated TUSC3- (Fig. 2a and Supplementary Fig. 2a, 4c). Moreover, the ectopic
dependent regulation of lung cancer progression and metas- expression of TUSC3 into HeLa and HCT116 cells, known as
tasis. miRNAs, the most abundant endogenous small non- TUSC3 null cells, consistently suppressed migration and inva-
coding RNA, are often dysregulated in most cancers, which sion potential (Supplementary Fig. 3a, b)36–38. To identify
consequently controls key regulators in tumorigenesis31. miR- downstream events in TUSC3 deficient NSCLC cells, we per-
520c was initially characterized as a pro-metastatic miRNA in formed PCR arrays with 84 known metastasis-related genes and
breast cancer whereas miR-520c/-373 functional cluster func- found that genes involved in cell-to-cell interaction and extra
tioned as a metastatic suppressor in estrogen receptor negative cellular matrix rearrangement were reasonably regulated in
breast cancer32,33. However, in lung cancer, the function of miR- both TUSC3 KD H460 cells (Fig. 2b and Supplementary Fig. 3c
520c remains unexplored and that of the miR-224 was also and Supplementary Dataset 1). These findings are consistent
controversial in NSCLC tumorigenesis34,35. Here we show that with the previous reports regarding the physiological role of
miR-224/-520c-mediated TUSC3 suppression enhances the TUSC3 being a member of oligosaccharyltransferase (OST)
metastatic potential of NSCLC through the altered ER-stress complex that mediates N-linked protein glycosylation, which
responses and HRD1-dependnet ERAD. frequently involves cell-to-cell and matrix interaction as a
2 NATURE COMMUNICATIONS | (2018)9:5110 | DOI: 10.1038/s41467-018-07561-8 | www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-07561-8 ARTICLE
a b c
IHC for TUSC3
(Normalized by wild type control)

(N = 50 pairs, p < 0.001) N = 223, P = 0.003
Relative luciferase activity

n.s n.s
100 1.2
Case 1 Case 2 Case 3
80 1.0
Percent survival
0.8
metastatic Primary
60 0.6
0.4 * *
40 TUSC3 low
0.2
TUSC3 high
Matched
20 0
0 WT Mutant WT Mutant
0 1000 2000 3000 4000
MiR-224 MiR-520c
Days
d e f
miR-520c
1.2 NC Primary Metastasized
miR-224
A-miRs
TUSC3 mRNA relative
1 miR224 M.W
Ctrl
Ctrl
miR-224
miR520c M.W (kDa)
0.8
expression
(kDa) 1.0 0.23 0.45
P < 0.001
1.0 1.88
0.6 * 37 37
* TUSC3
0.4 **
miR-520c
*
0.2 Vinculin
100 100
0
H460 A549
g h miR520c Cytokeratin Merged RGB

miR224 Cytokeratin Merged RGB
miR224 TUSC3 Merged RGB miR520c TUSC3 Merged RGB

Metastasized Primary
Metastasized Primary
P < 0.001
P < 0.001
Fig. 1 miR-224 and -520c are responsible for TUSC3 deficiency in metastasized lung cancer patient tissues. a Expression analyses of TUSC3 by
immunohistochemistry (IHC) showing suppressed expression in matched metastatic samples. 50 primary and their matched lymph node metastasized
lung tumor tissues were analyzed. The summarized expressional scores are shown in Supplementary Fig. 1b. P-value was obtained by Pearson’s chi-
squared test. b PrognoScan based Kaplan Meier plot showing decreased patient survivals associated with TUSC3 deficiency (GSE31210). c Luciferase
reporter assays with TUSC3 3’UTR in miR-224 or -520c-overexpressing HEK293 cells. The pGL3-TUSC3-3’UTR plasmid or the mutants harboring deleted
target sequences for miR-224 or miR-520c were co-transfected with miR-224 (left panels) or miR-520c (right panels). The relative values were obtained
by normalizing to Renilla luciferase values. Bars represent means ± SD (n = 4) and the p-values were determined by two-tailed student t-test (*p < 0.001).
d Decreased TUSC3 mRNA expression in miR-224 or miR-520c overexpressing NSCLC. Bars indicate means ±SD (n = 3) and p-values were addressed by
two-tailed student t-test (*p < 0.01, **p < 0.05). e Western blot analyses showing that miR-224 and miR-520c suppressed the endogenous TUSC3 protein
in A549 (left panel) and H460 (right panel) cells. Pre-miR-224/-520C (left) or anti-miR-224/-520c (right) were ectopically expressed in A549 (left) or
H460 (right) cells. f In situ hybridization showing the enhanced miR-224 and miR-520c expression in lymph node metastasized lung tissues compared to
their corresponding primary tumors. Total numbers of cases were summarized in Supplementary Fig. 1e. g, h Co-expression analyses of TUSC3 and miR-
224 (g) and miR-520c (h). The images in the miR-224 and miR-520c sections show a lung adenocarcinoma after co-expression of TUSC3 (fluorescence
red and RGB brown) with miR-224 (g, fluorescence blue and RGB blue) or miR-520c (h, fluorescence blue and RGB blue). The chi-squared test statistic
was generated and the null hypotheses that the expression of TUSC-3, miR-224, and miR-520 was equal in primary versus the metastatic tumors or that
the expression of TUSC3 and a given miRNA was equal in the primary and metastatic tumors tested using 2 degrees of freedom. The scale bars indicate
150 µm (a), 100 µm (b), and 200 µm (g, h), respectively
posttranslational modifier39–41. Furthermore, Gene Chip From orthotopic xenograft studies, we confirmed the role of
Human Transcriptome Arrays (HTA 2.0) were utilized in A549 TUSC3 downregulation in promoting cancer metastasis. 5 × 105
TUSC3 knock-out (KO) cells generated by CRISPR KO con- H460 TUSC3 KD cells (sh#2C1) were intravenously injected into
structs (Supplementary Fig. 2). After Gene Set Enrichment nude mice. After 4 weeks, the metastatic nodules from H460
Analysis (GSEA), metastasis-related gene signatures were TUSC3 KD cells were increased compared to controls (Supple-
shown to be further enriched in TUSC3 KO cells and in mentary Fig. 3d). Consistently, stronger bioluminescence was
Tunicamycin (TM)-treated TUSC3 KO cells (Fig. 2c and Sup- detected in the orthotopically xenografted mice injected with
plementary Dataset 2)42–44. These data suggest that TUSC3 A549 GFP+/luc+ TUSC3 KD (#sh3C1-Luc) cells (Fig. 2d). These
downregulation enhances metastasis-related gene signatures data strongly suggest that the downmodulation of TUSC3
and mediates ER-stress induced oncogenesis. enhances the metastatic potential of lung cancer in vivo.

a b sh#2C1/sh Ctrl
1.E–06
* **
Relative flurescence units

1.E–05
7 ** * shc
6 #2A1
1.E–04
P-Value
5 #2C1
4 1.E–03
3 1.E–02
2
1 1.E–01
0
1.E+00
Mock smTUSC3 Mock smTUSC3
–3 –1 1 3
Migration Invasion Log 2 (Fold)
c d
Ctrl vs TUSC3KO Ctrl + TM vs TUSC3KO+TM ×104
Photon flux rate (106)

5 weeks
p = 0.021 2.0
Metastasis-related genes 20
0.5 NES = 1.850
0.6 sh-con 1.5
NES = 2.035 15
Enrichment score
Enrichment score
0.4 0.5
P = 0.002 0.4 P < 0.001
0.3 sh-tusc3
0.2
FDRq = 0.009 0.3 FDRq = 0.001
10 1.0
0.1 0.2
0 0.1
–0.1 0 5 0.5
–0.2 –0.1
0
2 4 5 A549_ctrl
TUSC3 KO positively correlated TUSC3 KO positively correlated
Post injection
(Weeks)
Migration-related genes ×104 ×104
0.6 4 weeks 5 weeks 12.0
0.4 NES = 1.711 NES = 2.335 7.0
Enrichment score
0.5
Enrichment score
0.3 P < 0.001 P < 0.001

0.4
0.2 FDRq = 0.012 0.3 FDRq = 0.001 10.0
0.1 0.2
0 0.1 5.0 8.0
–0.1 0
–0.2 –0.1
–0.2
6.0
–0.3 3.0
4.0
1.0 2.0
KEGG_cell adhesion A549_#sh3C1

0.4 0.4
NES = 1.406
Enrichment score
NES = 1.476
Enrichment score
0.3 P = 0.023 P = 0.013

0.3
0.2 FDRq = 0.095
0.1
0.2
FDRq = 0.052
f
0
0.1 H460 H460-TUSC3KO
–0.1 0
–0.1 Ctrl A-miRs Ctrl A-miRs
Case 1
e A-miRs A-miRs
Case 2
siTUSC3
siTUSC3
M.W M.W
Ctrl
Ctrl
SC
SC
(kDa) (kDa)
37 37 6.0
TUSC3 TUSC3 **
*
100 Vinculin 100 Vinculin
Tumor area (Log 10)
5.5
Anchorage independent
* *
120 * *
5.0
cell growth (%)
90 *
60 4.5
30
4.0
0
A-NC + – – + – –
A-miRs – + + – + + 3.5
scrambled + + – + + – Ctrl A-miRs Ctrl A-miRs
situsc3 – – + – – +
A549 H460 H460 H460-TUSC3KO
Moreover, miR-224 and miR-520c enhanced cell migratory and corresponding controls, which rescued in H460 TUSC3 KO
invasive abilities (Supplementary Fig 3e, f) and epistatic analyses expressing anti-miR-224/-520c miRNAs (Fig. 2f and Supplemen-
using anti-miR-224/-520c and siTUSC3 siRNAs consistently tary Fig. 2).
showed that the suppression of both miRNAs decreased
metastatic potential including anchorage independent cell
growth, cell migration and invasion, which was rescued by ATF6α activity is enhanced in TUSC3 deficient cells. As an ER
siTUSC3 siRNAs treatment (Fig. 2e and Supplementary Fig. 3g– resident protein, TUSC3 is known as a binding protein with the
3i). Consistently, metastatic potential to lung tissues was Oligosaccharytransferase (OST) complex. The studies regarding
suppressed in mice xenografts subcutaneously injected with TUSC3 X-ray structure and biochemical analyses showed that
H460 expressing anti-miR-224/520c miRNAs compared to their TUSC3 enhanced N-linked glycosylation on the substrates

Fig. 2 TUSC3 deficiency enhances metastatic potential of NCLC in vitro and in vivo. a Increased migration and invasion of TUSC3 knockdown (KD) cells.
Two different TUSC3 KD cells (sh#2A1 and sh#2C1) were transfected with empty or mutant TUSC3 (smTUSC3) harboring three silence mutations
corresponding to shTUSC3 shRNA target sequences. After 48 h, the cells were introduced into migration (left panels) and invasion (right panels)
chambers. Bars represent means ± SD (n = 3) and p-value were calculated by student t-test (*p < 0.05 and **p < 0.01). b Differential expression of 84
known metastasis-related genes in H460 TUSC3KD cells (sh#2C1). The lists of genes showing differential expression is shown in Supplementary
Dataset 1. c Gene set enrichment analysis (GSEA) for metastasis, migration and cell adhesion-related gene sets in A549 control or TUSC3KO cells with
DMSO (left panels) or Tunicamycin (TM, 3ug/ml; right panels) for 16 h. The lists of gene sets are shown in Supplementary Dataset 2 (d) IVIS in vivo
images showing enhanced colonization ability of the A549 TUSC3 KD (sh#3C1). The cells (1 × 106) were intravenously injected into nude mice and the
bioluminescence was obtained in weekly basis from 3 weeks of injection. Bars indicate means ± SD (n = 3) and the p-values were addressed by two-tailed
student t-test. e Anchorage independent cell growth is modulated by miRNA-dependent TUSC3 downregulation. Anti-miR-224 and -miR-520c were co-
transfected with either scrambled or TUSC3 siRNAs for 48 h. After that, the cells were placed into an Agar Matrix Layer for 7 days. The numbers of
colonies were quantified with a standard MTS assay. The changed expression of endogenous TUSC3 was analyzed by Western blot analysis (upper panel)
or qRT-PCR analysis (Supplementary Fig. 3g). Bars indicate means ± SD (n = 4) and p-value were calculated by two-tailed student t-test (*p < 0.005 and
**p < 0.001). f Lung metastasis of subcutaneously xenografted mice showing enhanced lung metastasis. The 5 × 105 of the indicated cells were
subcutaneously injected into 4 nude mice, respectively. After 4 weeks, the lung tissues were harvested and dissected in 50-micron depth followed by H&E
stain. Tumor areas were measured by Image J software. p-value were calculated by two tailed student t-test (*p < 0.001, **p < 0.005)
through Cys-X-X-Cys (CXXC) motif on its ER-luminal KO cells (Fig. 3e). The reconstitution of TUSC3 or TUSC3 CCSS
domain39–41. The notion that an abnormal glycosylation in ER mutant in H460 TUSC3KO cells recovered PERK and IRE1α
is a typical ER stress inducible condition motivated us to analyze expression whereas the TUSC3-deleted mutant did not show
UPR activation in TUSC3 deficient cells. Further analysis using recovered expression (Fig. 3f). More importantly, the recon-
GSEA showed that the UPR related genes were highly enriched in stitution of the TUSC3 gene in TUSC3 KO cells followed by
TUSC3 KO cells without any stimulation and further enriched in subcellular fractionation assay showed decreased ATF6α protein
TM-treated TUSC3 KO compared to their corresponding con- activation, while the rescued activation of ATF6α was subtle in
trols (Fig. 3a and Supplementary Dataset 3). However, we could TUSC3 CCSS mutant-expressing cells (Fig. 3g). These data sug-
not find any significant difference in the expressional and spatial gest that C-X-X-C motif is responsible for TUSC3-mediated
changes of TUSC3, and the expression of miR-224/-520c in ATF6α activation, which is distinct from the TUSC3-dependent
response to ER stress induction (Supplementary Fig. 4). There- PERK and IRE1α regulation. Therefore, we believe that the
fore, we hypothesize that ER stress induction is not a regulator of altered protein N-glycosylation by TUSC3 downmodulation is
TUSC3 expression, but rather genetic and/or epigenetic responsible for the activation of the ATF6α. To analyze whether
mechanisms could be responsible for TUSC3 deficiency such as HRD1-dependent ATF6α activation could mediate the role of
the induction of the miR-224/-520c, promoter hypermethylation TUSC3 in cancer metastasis, we performed a series of rescue
and/or genetic deletion of chromosome 8p arm18,38,45. experiments. The ATF6α downmodulation enforced by siATF6α
To elucidate the mechanism underlying TUSC3-dependent siRNAs suppressed the clonogenicity and invasion whereas the
UPR regulation, a series of Western blot analyses were performed. other UPR responders, IRE1α and PERK suppression did not
Interestingly, the expression of IRE1α and PERK proteins, and affect those abilities in A549 or H460 TUSC3KO cells (Supple-
their downstream target genes was suppressed in response to ER mentary Fig. 6a–c). The suppression of the metastatic potential
stress induction in A549 TUSC3KO cells, while the suppression mediated by ATF6α in TUSC3 deficiency was further confirmed
was restored in A549 TUSC3/HRD1 double KO (DKO) cells by orthotopic xenograft mice models showing reduced the
(Fig. 3b). Conversely, the induction of ER stress enhanced nuclear numbers and areas of metastatic foci in H460 TUSC3KO cells
localized and chromatin-bound ATF6α proteins in TUSC3 KO stably expressing shATF6α (Fig. 3h). Furthermore, IHC for co-
cells, which failed to be rescued by TUSC3/HRD1 DKO. expression analysis using anti-TUSC3 and anti-ATF6α antibodies
Consequently, ATF6α promoter activity was enhanced in A549 in primary and metastasized lung cancer patient samples showed
TUSC3 KO cells (Fig. 3c and Supplementary Fig. 5a–c). The that enhanced and/or nuclear localized ATF6α protein was highly
expression of XBP1 known to be induced by active ATF6α inversely correlated with TUSC3 expression. Moreover, those
protein was consistently increased in A549 TUSC3KO cells, proteins were mutually exclusive in the four cases among five
which was not rescued in A549 TUSC3/HRD1 DKO cells. cases having both expression of ATF6α and TUSC3 proteins
Furthermore, the XBP1 cleavage induced by active IRE1α was (Fig. 3i). Additionally, the activation of ATF6α was more found to
suppressed in A549 TUSC3KO cells with ER stress induction, also be in metastasized lung cancer patient samples (Supplementary
rescued in A549 TUSC3/HRD1 DKO cells46 (Supplementary Fig. 6d, e). These data strongly suggest ATF6α has pro-metastatic
Fig. 5d). Additionally, ER-heat shock proteins, GRP78 and 94 property by, at least, mediating TUSC3-dependent metastatic
known as an ATF6α responder and promoting cancer progres- regulation. Next, we also analyzed the metastatic potential of
sion, were observed to be increased in A549 TUSC3 KO cells HRD1 gene using HRD1KO cells. We observed that the enhanced
(Fig. 3d and Supplementary Fig. 5e, f)10. These data suggest that metastatic potential by TUSC3 KO was rescued by HRD1/TUSC3
the activation of UPR in TUSC3 deficient cell could be mediated DKO in vitro and in vivo (Fig. 3j and Supplementary Fig. 6f, g),
by ATF6α rather than PERK and IRE1α proteins. Moreover, suggesting that TUSC3-dependent metastasis is likely to be linked
HRD1 protein is responsible for the downmodulation of the to HRD1-dependent signaling pathways.
IRE1α and PERK proteins, but the activation of ATF6α was
regulated in an HRD1-independent manner.
TUSC3 deficiency enhances HRD1-dependent metastatic
potential. To monitor the TUSC3-dependent ERAD efficiency in
C-X-X-C motif in TUSC3 is responsible for activating ATF6α. a live cell, we employed A1AT-NHK-ddVenus GFP mutant47.
To analyze the relationship between glycosylating ability and We generated stable cells in A549 TUSC3 KO cells by trans-
anti-metastatic role of TUSC3, we performed epistasis analyses by duction with the recombinant lenti-A1AT-NHK-ddVenus virus
employing wild type TUSC3 or TUSC3 CCSS mutant into TUSC3 and the accumulated mutant GFP proteins were monitored under

a REACTOME_UNFOLDED_PROTEIN_RESPONSE b
Ctrl vs Ctrl+TM TUSC3KO vs TUSC3KO+TM
Ctrl KO#4 DKO#6
Enrichment score
0.8 0.7 NES=2.367
Enrichment score
NES=2.865 0.6
0.7 p<0.001 M.W
p<0.001
0.6 0.5
FDRq<0.001 TG
0.5
0.4
FDRq<0.001 0.4
0.3
(kDa) 0 3 6 9 12 24 0 3 6 9 12 24 0 3 6 9 12 24
(1 uM,h)
0.3 0.2
0.2
0.1
0.1
100 IRE1α
0
0
50 XBP1S
Ctrl+TM positively correlated TUSC3KO+TM positively correlated
150 PERK
Ctrl vs TUSC3KO Ctrl+TM vs TUSC3KO+TM
0.6
Enrichment score
Enrichment score
0.5 NES=1.906 0.5 NES=1.913

0.4 p<0.002 0.4
p<0.001 p-EIF2α
0.3 FDRq<0.0038 0.3 FDRq<0.0044 37
0.2 0.2
0.1 0.1
EIF2α
0 0
37
B-Actin
37
c d siATF6α e
M.W DKO#6 Ctrl KO#4
(kDa) TG DMSO TM TM
0 6 12 0 6 12 0 6 12 (1 uM, h)
C #4 #5 C #4 #5 C #4 #5
Chromatin-bound
75 Signal peptide
75
GRP78
ATF6α Long
50 ER luminal oxidoreductase domain
PARP1 75 Transmembrane domain

100 Short
Calnexin 0 250 347, 348
75
GRP94
75 TUSC3 CSVC
100
ATF6α 75
GRP75 TUSC3 CCSS
ER/Golgi
75 SSVS
100 PARP1
Vinculin Del-TUSC3 CSVC
100
75 Calnexin
f g Empty TUSC3 CCSS

M.W TM
Empty TUSC3 CCSS Del-TUSC3 (kDa) 0 6 9 24 0 6 9 24 0 6 9 24
TG (5 ug/ml, h)
M.W 0 3 6 0 3 6 0 3 6 0 3 6
(kDa) (1 uM, h) 50 ATF6α
Nuclear
100 IRE1α
PARP1
100
50 XBP1-S Calnexin
75
150 PERK
100 ATF6α
75
p-EIF2α
ER/Golgi
37
100 PARP1
37 EIF2α
75 Calnexin
100 Vinculin 37 TUSC3
h i Primary Metastasized j
ATF6α
KO
shATF6α
DKO
TUSC3
100
100
Lung tumors for ATF6α and TUSC3
KO
*
Tumor area (%)
80
Tumor area (%)
shATF6α KO
80
ATF6α+ / ATF6α+ / ATF6α– / ATF6α– / 60
60
TUSC3+ TUSC3– TUSC3+ TUSC3– 40
40 *
5 12 32 29 20
DKO
20
0
(6.4%) (15.3%) (41.0%) (37.2%) 0
KO DKO
KO shATF6α
(Chi square = 59.9, p <0.001)
a fluorescence microscope and, quantified by FACS and Western NHK-ddVenus protein, which was dependent on HRD1 protein
blot analyses. As a result, we found that A1AT-NHK-ddVenus (Fig. 4b and Supplementary Fig. 7d). Additionally, stably
accumulated more in A549 TUSC3 KO cells compared to con- expressed A1AT-NHK-ddVenus proteins did not accumulated in
trols (Fig. 4a and Supplementary Fig. 7a). The accumulation of response to MG132 treatment in both A549 HRD1 KO and
A1AT-NHK-ddVenus in A549 TUSC3 KO cells were abrogated TUSC3/HRD1 DKO cells (Supplementary Fig. 7e, f), indicating
by overexpression of the TUSC3 gene (Supplementary Fig. 7b, c). that the regulation of A1AT-NHK-ddVenus protein by TUSC3
Also, anti-miR-224/-520c decreased the accumulation of A1AT- downregulation was completely dependent on HRD1 protein.

Fig. 3 TUSC3 deficiency selectively modulates Unfolded Protein Responses. a GSEA plots showing upregulation of the genes involved in Unfolded Protein
Reponses. b Western blot analyses showing weakened IRE1α-XBP1 and PERK-EIF2α pathways in A549 TUSC3 KO cells whereas enhanced in TUSC3/HRD1
DKO cells in response to ER stress induction. c Subcellular fractionation assay showing enhanced chromatin-bound ATF6α in A549 TUSC3 KO and HRD1/
TUSC3 DKO cells. The indicated cells were stimulated by 1.0 uM of TG for 6 or 9 h, respectively. Anti-Calnexin or anti-PARP1 antibody was used for ER/
Golgi or Nuclear fraction markers, respectively. d Increased ATF6α-dependent ER heat-shock proteins in TUSC3 deficient cells. The cells were transfected
by scrambled or siATF6α siRNAs for 48 h followed by exposing to DMSO or TM (3 ug/ml, 16 h). Western blot analysis shows elevated expression of
GRP78 and GRP94 in A549 TUSC3 KO cells. A mitochondrial heat-shock protein, GRP75 was used for negative control. e Schematic diagram showing
primary structure of TUSC3 protein and its mutants (TUSC3 CCSS). CSVC indicates the amino acids in C-X-X-C motif. f Rescued IRE1α and PERK
expression by reconstitution of TUSC3 or its CCSS mutant. g Restored nuclear localization of the ATF6α protein by the reconstitution of TUSC3 but not by
CCSS mutant in A549 TUSC3 KO cells. h Rescued the colonization ability of H460 TUSC3KO cells by suppressing ATF6α expression. 1 × 106 of the control
cells or ATF6α knock-downed TUSC3 KO ells was intravenously injected into four NOD scid gamma mice. p-value was calculated by unpaired student t-
test (*p = 0.002). The data for the tumor area from the control cells-injected mice are shared with Fig. 5g. i Co-expression analysis of ATF6α with TUSC3
protein showing inverse correlation between TUSC3 and ATF6α activation in lung cancer patient samples. p-value was obtained by Chi square analysis.
The scale bar is shown as 150 µm. j Rescued colonization ability of TUSC3/HRD1 DKO cells. The tumor area was calculated as the total area of lung
occupied by cancer is field of view using Image J software. The region for the cancer was expressed as percentage. p-value was calculated by unpaired
student t-test (*p < 0.001)
Taken together, these data suggest that TUSC3 deficiency recovered by overexpressing TUSC3 or CCSS mutant in H460
increased HRD1-dependent ERAD activity, mediated by miR-224 TUSC3KO cells treated with DMSO or tunicamycin (Fig. 4h and
and miR-520C. Supplementary Fig. 8i). In addition, p53-dependent anti-meta-
To examine the direct association of two ER resident proteins, static function was confirmed in orthotopically xenografted mice
we performed immunoprecipitation (IP) assays using pcDNA- with H460 TUSC3KO cells overexpressing TP53 gene (Fig. 4i).
TUSC3-V5-His or its mutants, and pCMV6-HRD1-Flag con- Furthermore, co-expression analyses of the TUSC3 protein with
structs. We found that the HRD1 protein interacted with wild IRE1α or p53 protein consistently showed that there was a
type TUSC3 and its CCSS mutant but not with the deletion significant correlation between TUSC3 and IRE1α protein and
mutant of TUSC3, further confirmed by inverse IP-Western blot similar results were observed in metastasized lung cancer patient
analyses (Figs. 3e, 4c and Supplementary Fig. 8a). In addition, a tissues with p53 proteins (Fig. 4j and Supplementary Fig. 8j, k).
series of pulse-chase experiments with A549 HRD1 KO or Also, GSEA plots indicated that genes suppressed by TP53 were
TUSC3/HRD1 DKO cells showed that TUSC3 protein levels were enriched in A549 TUSC3KO cells, which was exacerbated in
not significantly changed in the absence or presence of the HRD1 response to tunicamycin treatment (Fig. 4k and Supplementary
protein (Supplementary Fig. 8b, c). Consistently, in vivo Dataset 4)49. Taken together, TUSC3 downmodulation strength-
ubiquitination of TUSC3 protein did not show any significant ens HRD1-dependent metastatic potential although biochemical
change in response to HRD1 expression and/or ER stress studies should be addressed how TUSC3 regulates the interaction
induction (Supplementary Fig. 8d). Therefore, we believe that between HRD1 and its substrates.
the interaction of TUSC3 protein to HRD1 does not affect TUSC3
protein stability.
To characterize the exact role of HRD1-dependent ERAD TUSC3 deficiency negatively regulates p53-NM23H1 pathway.
regulation in TUSC3 deficient cells, we tested HRD1 affinity to To understand the mechanism by which TUSC3 deficiency
two known substrates, p53 and IRE1α16,17,48. The endogenous enhanced metastatic potentials through p53 regulation, we ana-
HRD1 protein was precipitated by anti-HRD1 antibody in H460 lyzed metastatic suppressor, NM23H1/250,51. First, we observed
and H460 TUSC3 KO cells followed by Western blot analyses for that the NM23H1/2 proteins were suppressed in TUSC3 deficient
HRD1 substrates. As a result, the interactions between HRD1 cells, which was further suppressed in response to ER stress
proteins and its substrates were found to be enhanced by TUSC3 induction (Fig. 5a and Supplementary Fig. 9a, b). Also, siRNA
downregulation, which was further confirmed by a rescue treatment of the TP53 decreased NM23H1/2 protein (Fig. 5b).
experiment that the overexpression of TUSC3 gene in H460 Consistently, p53 accumulation by treating with a MDM2 inhi-
TUSC3KO cells decreased the binding affinity of HRD1 protein bitor, Nutlin3a, recovered expression of NM23H1/2 protein in
to the substrates compared to the KO control. In addition to the A549 TUSC3KO cells (Fig. 5c), suggesting that p53 suppression
known substrates of HRD1, we identified PERK protein as a in TUSC3 deficient cells could be responsible for NM23H1/
binding protein with HRD1 and its interactions to HRD1 protein 2 suppression. Moreover, rescued NM23H1/2 expression was
was enhanced upon TUSC3 deficiency. However, another UPR observed in A549 TUSC3 KD cells treated with siHRD1 siRNAs
responder, ATF6α was not elucidated in the same immune (Fig. 5d). Next, we sought to determine the involvement of miR-
complex (Fig. 4d). 224/-520c in NM23H1/2 regulation and found that the sup-
To establish the competitive binding activity of TUSC3 to pression of miR-224/-520c increased NM23H1/2 expression,
HRD1 and p53, we performed an in vitro competition assay with which was diminished upon TUSC3 reduction (Fig. 5e). Addi-
p53 protein and found that its interaction to p53 protein was tionally, the ectopic expression of miR-224 or miR-520c
weakened by TUSC3 protein in HEK293 cells (Fig. 4e). Moreover, decreased NM23H1/2 protein expression levels (Supplementary
poly-ubiquitinated p53 protein was upregulated in A549 TUSC3 Fig. 9c, d). According to in silico tools, NM23H1/2 was not
KO cells, and rescued upon TUSC3 reconstitution (Fig. 4f). predicted as a target for miR-224 or miR-520c (data not shown),
Consistently, p53 ubiquitination was diminished by TUSC3 suggesting that the alteration of NM23H1/2 expression by
overexpression and similar results were found with IRE1α miR-224/-520c was indirect effects. Finally, reconstituted
(Supplementary Fig. 8e, f). Also, IRE1α and p53 protein was NM23H1 gene repressed the migratory and invasive abilities of
decreased in A549 TUSC3 KO cells, which was slightly rescued by TUSC3 KD cells, which was further confirmed by orthotopically
HRD1 deficiency in A549 cells (Fig. 4g and Supplementary xenografted mice with NM23H1-overexpressing H460TUSCKO
Fig. 8g, h). Moreover, the suppressed p53 expression was cells (Fig. 5f,g). These data suggest that miR-224/-520c-dependent

a c IP: e f
0 10 20 MG132
Ctrl
Input anti-His KO#4
M.W. (uM, 24 h)
KO
KO
KO
Del-TUSC3
Del-TUSC3
C
C
(kDa) Input IP:
IP-anti-HA
P53-Ub(n)
75 A1AT-NHK M.W. (10%) anti-flag
TUSC3
TUSC3
Empty
Empty
CCSS
CCSS
37 M.W. (kDa)
GAPDH p53-GFP
(kDa) 100
75
Anti-His
HRD1
A1AT-NHK intensify
15,000 37
(Log 10, mean)
Ctrl *
KO#4 37 Anti-His p53
Input
10,000 25
HRD1 – + + + – + + + TUSC3
HRD1
75 p53-GFP – + + + – + + + Ub-HA + + +
5000 Empty + +
50 p53-GFP + + +
TUSC3 – – TUSC3 – – +
0
MG132 TM + + +
0 20
(uM, 16 h)
b d Input IP- g
(5%) anti-HRD1
DKO#6 Ctrl KO#4
TUSC3
TUSC3
pcDNA
pcDNA
M.W. M.W. TM
(kDa) – + – + – + (5.0 ug/ml, 1 h)
IgG
Ctrl
Ctrl
(kDa)
M.W. Ctrl Anti-miRs MG132
150 50 p53
(kDa) 0 5 10 20 5 10 20 (uM, 24 h) PERK
100
75 75 GRP78
75 A1AT-NHK
37 GAPDH 150 Vinculin
IRE1α 100
100
75
8000
A1AT-NHK intensify
50
p53
(Log 10, mean)
37 h
6000 25
** 75 M.W. Empty TUSC3-his CCSS-his
TM
4000 HRD1
50 (kDa) 0 3 6 12 0 3 6 12 0 3 6 12 (5 ug/ml, h)
*
2000 37 37
* Anti-His
100
75 ATF6α
0 50
MG132 50 p53
0 20 20 20 *
(uM, 16 h)
50 100 Vinculin
k
*GAPDH
oc
iR
RD
37
m
M
siH
ti-
An
i j k Suppressed genes by TP53

p53 TUSC3 Merged
Ctrl vs TUSC3KO
0.5
NES= 1.854
Enrichment score
0.4
p < 0.001
0.3 FDRq<0.001
0.2
0.1
0
–0.1
** *
30
Number of foci
TUSC3KO positively correlated

20 IRE1α TUSC3 Merged
Ctrl+TM vs TUSC3KO + TM
Enrichment score
0.6 NES= 2.511

10 0.5 p <0.001
0.4 FDRq< 0.001
0.3
0.2
0 0.1
0
rl
53
pt
Ct
TP
Em
H460 TUSC3KO
TUSC3KO+TM positively correlated
Fig. 4 TUSC3 deficiency enhances HRD1-dependent pro-metastatic property by enhancing ERAD activity. a Enhanced A1AT-NHK-ddVenus GFP
accumulation in A549 TUSC3KO cells. The A1AT-NHK-ddVenus stable A549 control (Ctrl) or TUSC3 KO cells were generated by transduction with
pLL-Lenti-A1AT-NHK-ddVenus virus. The protein accumulation was monitored after incubating the cells with MG132 (20 uM) for 16 h. The accumulation
was quantified by FACS analyses (lower panel) or Western blot analyses using anti-GFP antibody (upper panel). b Decreased accumulation of A1AT-
NHK-ddVenus by miR-224 and -520c suppression. Bars indicate means ±SD (n = 3) and p-values were obtained by two-tailed student t-test (a and b,
*p < 0.001, **p < 0.02). c Co-immunoprecipitation assays with TUSC3 and HRD1 proteins. pcDNA-TUSC3-V5-His or its mutants (TUSC3-CCSS and Del-
TUSC3) was co-transfected with pCMV6-HRD1-Flag in HEK293 cells. d The changed affinity of HRD1 protein to its substrates by TUSC3. The endogenous
HRD1 protein was precipitated with anti-HRD1 antibody in H460 control, TUSC3 KO or TUSC3-reconstituting TUSC3 KO cells. e In vitro competition assay
with p53 and TUSC3 proteins in HEK293 cells. f In vivo ubiquitination assay of p53 protein showing enhanced p53 ubiquitination in response to
TUSC3 downmodulation. g Decreased p53 protein in H460 TUSC3 KO cell and rescued in HRD1/TUSC3 DKO cells. The indicated cells were exposed by
5.0 ug/ml of TM for 1 h and subject to Western blot analysis. h Rescued p53 protein in TUSC3 or TUSC3-CCSS mutant expressing H460 TUSC3 KO cells.
i Re-suppressed colonization effect of TUSC3KO cells upon TP53 reconstitution. Tail-vein injection was performed using the indicated cells, and
subsequently the lung tissues were harvested after 4 weeks. p-values were obtained by student t-test (*p = 0.00031, **p < 0.038). j Co-expression
analyses between TUSC3 and p53 (upper panels) or IRE1α (lower panels) in lung cancer patient samples. The p53 and IRE1α are shown in fluorescence
green whereas TUSC3 is shown in fluorescence red. Scale bar indicates 150 µm. k GSEA plots indicating gene sets suppressed by TP53 were enriched in
TUSC3KO cells either DMSO or TM treatment compared to the controls

a b c
siP53
M.W Ctrl KO#4 KO#5
SC
M.W
M.W DMSO TM (kDa) (kDa) – + – + – + Nut3a
(kDa) C #4 #5 C #4 #5 50 p53
20 50 p53
NM23H1/2 20 20
NM23H1/2 NM23H1/2
37 GAPDH 37 37 GAPDH
Vinculin
d e f * shc
6
Anti-miR224/520
5 #2A1
situsc3
Relative fold
* #2C1
A-NC
Ctrl sh#3A1 M.W 4
M.W
SC
– – + – – + siHRD1 (kDa) 3
(kDa) – + + – + + TM 20
25 NM23H1/2 2
NM23H1/2 1
37 GAPDH GAPDH
37 0
Empty NM23H1 Empty NM23H1
g Migration Invasion
KO
NM23H1
100
Tumor area (%)
KO
80
60
NM23H1
40
20
0
KO NM23H1
Fig. 5 Metastatic suppressor, NM23H1/2 is regulated by TUSC3 through p53 regulation. a The reduction of NM23H1/2 protein in TUSC3 KO cells in
response to ER stress induction. The cells were incubated by 5.0 ug of TM for 9 h and subsequently harvested for Western blot analysis. b Decreased
NM23H1 protein in p53 deficient cells. The H460 cells were transfected with siTP53 siRNAs for 48 h, and the expression of p53 and NM23H1/2 proteins
was measured by Western blot analysis. c Restored NM23H1/2 proteins by p53 protein accumulation in A549 TUSC3 KO cells. The cells were incubated
with Ntulin3a for 24 h, and Western blot analysis was performed to measure the expression of NM23H1/2. d HRD1-regulated NM23H1/2 protein in A549
TUSC3 KD cells. The A549 TUSC3 KD cells were transfected with scrambled or siHRD1 siRNAs. After 48 h, the cells were exposed to TM (5 ug/ml) for 6 h
followed by Western blot analysis using indicated antibodies. e Accumulated NM23H1/2 proteins upon miR-224/-520c downregulation. The miRNA KD
cells were transfected by scrambled or siTUSC3 siRNAs for 48 h, and the NM23H1/2 protein was measured by Western blot analysis. f The reduced
migration and invasion of A549 TUSC3 KD cells in response to NM23H1 reconstitution. H460 TUSC3 KD cells were transfected by pCMV6-NM23H1
plasmid for 24 h. Subsequently, the cells were harvested and subject to the chambers of migration or invasion assays. Error bars indicate means ± SD (n =
3) and the p-values were calculated by two-tailed student t-test (*p < 0.03). g Restored expression of NM23H1 decreased colonization ability of H460
TUSC3KO cells. The cells overexpressing NM23H1 generated by the transduction of lent-NM23H1 virus. 1 × 106 of NM23H1 overexpressing TUSC3KO cells
or control KO cells was intravenously injected into four NSG mice. The tumor area was obtained by Image J software by measuring the field of view in lung
occupied by cancer. The region for the cancer was expressed as percentage. p-value was calculated by unpaired student t-test (*p < 0.001)
TUSC3 suppression regulates NM23H1/2 expression, which Discussion

regulates the metastatic potential in NSCLC. An ER-membrane protein, TUSC3 functions in cancer patho-
Here we present a study characterizing the role of miRNAs- genesis by context dependent manner, but remains controversial
224/-520c-induced TUSC3 downregulation and its contribution in lung cancer. Like most other cancer types, it was initially
to the metastatic potential in NSCLC. In normal or early localized reported that TUSC3 gene frequently methylated in NSCLC
lung tumor, the expression of miR-224/-520c was low while compared to blood lymphocytes27. Moreover, the expression of
TUSC3 expression was maintained at normal level. As a result, TUSC3 gene was associated with lymph node metastasis and
TUSC3 regulates HRD1-dependent ERAD through direct inter- consistently suppressed in advanced lung cancers including small
action and is involved in N-linked protein glycosylation to some cell lung cancer and lung adenocarcinoma, a subtype of
proteins through C-X-X-C motif (Fig. 6, left panel). However, NSCLC30,52. Subsequent cell biological assays showed TUSC3
genetic deletion and/or epigenetic modification such as induction inhibited lung cancer cell proliferation and induced cell death52.
of miR-224/520c in advanced lung cancers is responsible for However, these findings have been challenged by opposite results
TUSC3 deficiency. As the consequence of aberrant N- that the methylation of the TUSC3 gene was associated with
glycosylation on some proteins, TUSC3 deficiency enhances the increased survival of patients with lung cancer28. Additionally,
UPR pathways and the E3 ubiquitin ligase activity of HRD1, TUSC3 enhanced lung cancer cell proliferation associated with
resulting in the suppression of IRE1α and PERK proteins, Hedgehog signaling pathway, which was confirmed in tumor
which can reduce excessive UPR. Moreover, another xenograft mice model and NSCLC patient samples29.
HRD1 substrate, p53 is decreased, which weaken the activation Regarding this, our current study claims four novel findings to
of the TP53-NM23H1/2 pathway, which consequently enhances support the pro-metastatic role of TUSC3 deficiency. First, we
the metastatic potential of NSCLC (Fig. 6, right panel). observed that TUSC3 expression was frequently suppressed in the

Normal/Localized cancer cells Metastatic cancer cells

Low High
miR-224/miR-520c miR-224/miR-520c
Genetic
alteration
cys cys
TUSC3 TUSC3 deficiency
Regulating HRD1- Protein Enhanced HRD-1- Abnormal

dependent ERAD N-Glycosylation dependent ERAD N-Glycosylation
Normal UPR / ERAD p53 IRE-1α PERK ATF6α
UPR activation
NM23H1/2
Cancer
metastasis
Fig. 6 The schematic diagram showing the working hypothesis by which miR-224/-520c-induced TUSC3 suppression enhanced metastatic potential of
NSCLC through the alteration of UPRs and HRD1-dependent ERAD
metastasized lung cancer patient samples compared to primary the overall impact on N-linked protein glycosylation by TUSC3
lung cancer, possibly mediated by the increased expression of downregulation remained subtle because of the functional
miR-224 and/or miR-520c (Fig. 1 and Supplementary Fig. 1). redundancy of MagT1 protein, another OST associated protein53.
Additionally, the role of miR-224 also remained controversial and However, TUSC3 x-ray structure and biochemical analyses
that of miR-520c has not been characterized in lung cancer revealed that TUSC3 mediated N-linked protein glycosylation
pathogenesis34,35. Our current study could also provide a through Cys-X-X-Cys (C-X-X-C) motif involving disulfide inter-
mechanistic insight to the unrecognized role of the miRNAs. actions that makes stable interaction of OST complex to their
Second, a series of in vitro and in vivo experiments using TUSC3 substrates39,41. Moreover, TUSC3 specifically enhanced the effi-
KD and/or KO cells consistently showed that TUSC3 deficiency ciency of N-linked glycosylation on integrin β1 and lectin protein
enhanced the metastatic potential of NSCLC (Fig. 2 and Sup- family, which enhanced cancer progression in ovarian and pros-
plementary Fig. 2, 3). Third, we also characterized the mechanism tate cancers22,54. Additionally, the alteration of posttranslational
underlying TUSC3 deficiency. Utilizing TUSC3KO and TUSC3/ modification is frequent ER-stress inducible condition. Indeed, the
HRD1 DKO cells, we found that ATF6α pathway, but not IRE1α ATF6α in TUSC3 deficient cells was still activated in HRD1/
and PERK pathways was selectively enhanced in TUSC3 deficient TUSC3 DKO cells and was only minimally rescued by TUSC3-
cells, possibly mediated the function of TUSC3 deficiency in lung CCSS mutant overexpression, suggesting that the activation of
cancer metastasis. Lastly, TUSC3 regulated HRD1-dependnet ATF6α by TUSC3 deficiency is not dependent on HRD1 protein
ERAD and its deficiency consequently enhanced the HRD1 and the function of the substrate stabilization through C-X-X-C
activity. As a result, the HRD1-dependent ERAD substrates, the motif in TUSC3 is an important for ATF6α activation (Fig. 3e, g
PERK, IRE1α and p53 protein were decreased. Moreover, p53- and Supplementary Figa. 5a–5c). Therefore, we believe that the
NM23H1/2 tumor suppressive pathway in TUSC3 deficient lung upregulation of UPRs in TUSC3 KO cells could be caused by
cancer cells was suppressed (Fig. 5 and Supplementary Fig. 9). defects in glycome profiles.
Interestingly, IHC results also showed that TUSC3 was over- We also characterized how three UPRs are independently
expressed in other stromal cells regardless of the expression of regulated in TUSC3 deficient cells. A series of rescue experiments
TUSC3 protein in cancer cells, suggesting that the TUSC3- using TUSC3 or TUSC3-CCSS mutant gene in TUSC3KO and/or
dependent metastatic regulation could be a cell autonomous effect HRD1/TUSC3 DKO cells showed that IRE1α and PERK proteins
(Fig. 1a). were downregulated whereas ATF6α was activated in TUSC3
To date, the significance of UPR activation is not well under- deficient cells with ER stress induction. Importantly, HRD1/
stood due to their dual roles in terms of cancer progression. One TUSC3 DKO cells exhibited restored expression of IRE1α and
of the accepted principles regarding the role of UPR in cancer is PERK proteins compared to the TUSC3 KO cells. Furthermore,
the ability of the UPR to activate tumor suppressive pathways TUSC3 and its null mutant (TUSC3-CCSS) for N-linked glyco-
when the cells were under acute and strong UPR activation sylation can bind to the HRD1 protein. When we reconstituted
conditions4,5. Our current observation that TUSC3 deficient cells the TUSC3 or TUSC3-CCSS in TUSC3 KO cells, the PERK and
were enriched for gene sets with metastatic signatures compared IRE1α proteins were found to be rescued in both TUSC3 and the
to control cells led us to investigate the relations between TUSC3 CCSS mutant expressing TUSC3 KO cells (Figs. 3, 4 and Sup-
and ER stress response. Although TUSC3 is a known subunit of plementary Fig. 4). Moreover, IRE1α was known to be a substrate
the OST complex responsible for N-linked protein glycosylation, for HRD1-dependent ERAD in synovial fibroblasts and this

regulation was conserved in lung cancer cells, and the PERK, but purchased from Origene. Anti-IRE1α (#ab96481, 1:50 dilution) antibody for IHC
not ATF6α protein was identified in HRD1 complex (Figs. 3, 4 analysis was purchased from ABcam. Uncropped scans of most important blots are
shown in the Supplementary Fig. 10.
and Supplementary Fig. 7f)48. Consequently, among three UPR
responders, PERK and IRE1α were suppressed by the enhanced
HRD1-dependent ERAD whereas ATF6α activation is not Cell culture and nucleic acid delivery. The parental TUSC3KO cells, H460 and
A549 were purchased from ATCC company. The H460 and A549 (Lung cancer),
dependent on HRD1 protein, but rather activated by TUSC3- HeLa (Cervical Carcinoma) and HCT116 (Colorectal Carcinoma) cells were cul-
dependent N-linked glycosylation. We believe that the differential tured in RPMI 1640 supplemented with 10% fetal bovine serum (FBS) and HEK293
regulation of the three pathways was to protect the cells from (Human Embryonic Fibroblast) cells were grown in DMEM supplemented with
excessive UPR activation suppressing tumor progression and 10% FBS (SIGMA). The transient expression of the plasmids and shRNAs was
obtained by using Lipofectamine 2000, Lipofectamine 3000, RNAiMAX, or Lipo-
metastasis, which makes TUSC3-dependent UPR oncogenic4,5. fectamine Plus reagents according to the manufacturer’s protocol (Invitrogen). The
The most prominent dysregulated gene, TP53 was often contamination of the cell lines used in this study were regularly monitored by a
deleted in lymph node metastasized larynx and pharynx carci- specific kit (Invivogen).
nomas where TUSC3 gene was also deleted although there was no
any evidence that one gene was deleted along with the other The generations of TUSC3 knock-down cells. TUSC3 shRNAs and recombinant
gene55. Co-expression analysis using anti-TUSC3 and anti-p53 lentivirus containing a pool of three target-specific shRNAs of TUSC3 were purchased
antibodies showed that there was significant correlation between from Santa Cruz Biotechnology, Inc. For the generation of TUSC3 knock down (KD)
cells, H460 or A549 cells were transfected with TUSC3 shRNA plasmids. The
those in the lymph node metastasized lung cancer patient samples transfectants were selected using Puromycin reagent for 2 weeks. After that, the cells
(Fig. 4j and Supplementary Fig. 8j). Moreover, p53 protein was were sorted into each single cell and grown for additional 2 weeks. KD candidates
also known to be one of the best characterized substrates in were examined for TUSC3 expression by Western blot, qRT-PCR and/or IF analyses.
HRD1-dependent ERAD16,17. Additionally, HRD1 activity to its The knock-down efficiencies were confirmed with two independent experiments. The
sequences for shRNAs used in this study are shown in Supplementary Table 1.
substrates was enhanced in response to TUSC3 deficiency (Fig. 4
and Supplementary Fig. 7). Also, the expression of the NM23H1/
The generations of TUSC3 knock-out cells. The CRISPR knock out (KO) con-
2 known as a metastatic suppressor was transcriptionally regu- structs for TUSC3 gene and HRD1 gene were purchased from Santa Cruz Bio-
lated by p53 protein and the suppressed NM23H1/2 expression technology, Inc. The cells were transfected with TUSC3 CRISPR/Cas9 KO plasmid.
was observed in our PCR array analyses using H460 TUSC3 KD After 48 h, GFP-positive cells were isolated by ARIA FACS sorter and each single
cells (Fig. 2b, Supplementary Fig. 3c and Supplementary Data- cell of TUSC3 KO candidates were plated into 96 well plates. After 10–20 days, the
cells were harvested and prepared for Western blot and RT-PCR analyses using
set 1)51. These observations motivated us to analyze TP53- anti-TUSC3 antibody, and Taqman probe or Cybergreen PCR primers. For the
NM23H1/2 pathway in TUSC3 deficient cells. Indeed, the p53 TUSC3 and HRD1 double KO cells, the TUSC3 KO cells were transfected with
protein was regulated by the altered HRD1-mediated ERAD CRISPR/Cas9 KO plasmid and HRD1 HDR plasmids for 48 h. After that the cells
activity in TUSC3KO cells and TP53-NM23H1/2 pathway func- were selected in Puromycin-containing media for 7 days. The KO candidates were
tionally rescued the metastatic potential in TUSC3 deficient cells plated into 96 well plates with a single cell. The KO candidates were validated by
qRT-PCR analyses of HRD1 Taqman probe and by Western blot analyses with
in vitro and in vivo (Figs. 4, 5 and Supplementary Fig. 8, 9). anti-HRD1 antibody. The expression and changed genomic DNA sequences are
Taken together, we propose that miR-224/-520c dependent shown in Supplementary Fig. 2. The information about the sgRNAs and the pri-
TUSC3 downregulation enhances the metastatic potentials of mers for genomic DNA PCRs to validate knock-out is shown in Supplementary
NSCLC by imbalance of the activation of the UPR and enhancing Table 1. The qRT-PCR, Western blot analyses or IF assays to confirm the KO were
performed with two or three independent experiments.
HRD1-dependent p53 protein suppression. Our current observa-
tion could provide important insight to understanding the con-
Quantitative RT-PCR. Total RNA was prepared using TRIZOL according to the
troversial roles of TUSC3 in lung cancer metastasis and the specific manufacturer’s protocol (Invitrogen). The 0.5–2 ug of total RNA was subject to first
roles of UPR and HRD1 regulation in TUSC3-deficient lung cancer, strand synthesis reaction following that thestandard TaqMan miRNA or gene
as an anti-cancer therapeutic strategy against NSCLC. expression assay protocol was performed (Applied Biosystems). The results were
produced and monitored in a GeneAmp PCR 9700 Thermocycler. The compara-
tive CT cycles for the quantification of the genes or miRNAs interested was
Methods quantified with ABI Prism 7900HT detection system (Applied Biosystems). Sub-
Plasmids and siRNAs. pCMV6-TUSC3, pCMV6-NM23H1, and pCMV6- sequently, the values were than normalized by RNU44 and/or RNU48 for the
Synoviolin (HRD1)-Flag plasmids were purchased from Origene. pcDNA-IRE1α- miRNAs and GAPDH, β-actin, and/or OAZ1 for the gene expression, respectively.
GFP, pcDNA-TP53-GFP, pcDNA-Ubiquitin-HA plasmids were purchased from The sequence information about the probes and/or PCR primers used in current
Addgene Company. For the TUSC3, the cDNAs were subcloned into pcDNA4.1- study are shown in Supplementary Table 2. The results of qRT-PCR were per-
V5-His vector with EcoR I and Xho I restriction sites or pcDNA3-HA vector with formed by three or four technical replicates of two or three independent samples
Hind III restriction site. The 3’UTR of TUSC3-1 and TUSC3-2 plasmids were with two independent experiments
purchased from Origene and the 3’UTRs were subcloned into pGL3 Vectors with
Xho I restriction site. On-Target Smart Pools containing 4 different siRNAs on
each target gene of TUSC3, TP53, HRD1, ATF6α or NM23H1 were purchased In situ hybridization and co-expression analysis. Lung cancer tissues were
from Dharmacon company. The primer sets for gene manipulation are shown in purchased from US Biomax, Inc and the LNA probes for miR-224 and miR-520c
Supplementary Tables 1. were purchased from EXIQON. The in situ hybridization reaction for the miRNAs
was done followed by the IHC detection of the TUSC3 on the same section. The
section was then analyzed with Nuance computer system which separates each
Antibodies. All dilution factors for the antibodies shown at next of lot number.
colorimetric base signal as a different fluorescent color, then mixes them to
Otherwise 1:100 dilution of the antibodies from Santacruz Biotech. Inc, or 1:1000
determine if there is co-expression. Optimal detection of TUSC3 by IHC was
dilution from the other companies were used for the analysis. Anti-FlagM2-resin
determined using the Leica Bond Max to be a dilution of 1:100 with a pretreatment
(#A2220, 1:5000 dilution), anti-GM130 (#G7295) and anti-TUSC3 (#SAB4503183,
in the manufacturer’s proteinase K. The optimal detection of miR-224 and miR-
1:500 dilution) antibody were purchased from Sigma-Aldrich for IHC and
520c was determined in proteinase K with a concentration of the digoxigenin-
Immunofluorescence assay. Anti-TUSC3 antibodies for Western blot analyses were
tagged LNA probe of 0.5 pmol/ul. For IHC of TUSC3 protein, each 5 ug/ml of anti-
purchased from (LifeSpan Biosciences, Inc. #LS-C384735; ProteinTech #16039-1,
TUSC3 antibody was used. The statistical analysis was performed with the Pear-
1:500 dilution). Anti-Calnexin (#9956), anti-CHOP (#9956), anti-phosphor-eIF2α
son’s chi-squared test by using the InStat Statistical Analysis Software.
(#3398), anti-GAPDH (#5174), anti-GRP78 (#3177), anti-HRD1 (#14773), anti-
IRE1α (#3294), anti-NM23H1/2 (#5353), anti-PARP1(#9532), anti-PERK (#5683),
anti-Vinculin (#13901) antibodies were purchased from Cell Signaling Technology. Co-expression analyses. After using a standard optimization protocol that
Anti-ATF6α (#sc-22799), anti-eIF2α (#sc-133132), anti-GRP94 (#sc-393402), anti- included positive controls known to have the target of interest, we tested the tissues
GRP75 (#sc-13967), anti-HA (#sc-805), anti-P53 (#sc-126), anti-P21 (#sc-397), for the following antigens: TUSC3, p53, IRE1α and ATF6α. Each core in the TMA
anti-Ubiquitin (#sc-8017), and anti-XBP1 (#sc-8015) antibodies were purchased was scored as positive (at least 20% of the cancer cells showing a strong signal) or
from Santacruz Biotech. Inc. Anti-V5 (#R960-25, 1:5000) antibody was purchased negative (less than 20% score). A given tissue was tested for two different antigens
from Invitrogen. Anti-His tag (#MCA139. 1:5000 dilution) antibody was purchased using fast red as the chromogen for one target and immunohistochemistry using
from AbD Serotech. Anti-Flag tag (#TA50011, 1:5000 dilution) antibody was DAB (brown) as the second chromogen with hematoxylin as the counterstain. The

results were then analyzed by the Nuance and InForm systems in which each FBS was added and the cells were grown in a standard cell incubator for 4–7 days.
chromogenic signal is separated and mixed to determine the percentage of the cells s. Subsequently the cell-matrix layer was dissolved by adding a solubilization buffer
expressing two targets after convered to a fluorescence-based signalt56,57. and the living cells was quantified by a standard protocol of MTS assay (Invit-
trogne). The anchorage independent cell growth in control and TUSC3KD or
miRNA KD cells was performed twice, independently.
Migration and invasion assays. The experimental procedures of migration and
invasion assays followed the manufacture’s protocols (Calbiochem and TREVIGEN
Inc). Briefly, the cells were plated and transfected with indicated plasmids and/or Statistical analysis. The results were analyzed using ANOVA and/or two tailed
miRNAs, respectively. After 24 h, the cells were incubated with corresponding cell student t-test. Data are presented as mean ± standard deviation (s.d.) or standard
culture media with reduced serum percentage (1%) for additional 12–24 h. Sub- error of the mean (S.E.M.) of two or more independent biological replicates. The
sequently the cells were washed with PBS and 5 × 105 or 1 × 106 cells were plated statistical method was not used to predetermine sample size. The experiments and
into the upper chambers of the Migration/Invasion assays in the cell culture media outcome assessment were performed in a blinded way. Only p-values <0.05 were
without FBS. The media supplemented with 10% FBS were added into lower considered significant.
chambers to use as a chemoattractant. After 16–24 h for migration assay and
36–48 h for invasion assay, the upper chambers were transferred into a new plate Bioinformatics analysis. Bioinformatics analysis was performed by using these
with detaching solutions contained Calcein AM for 0.5–1 h. The fluorescence was specific programs: Targetscan and Pictar.
analyzed at an excitation wavelength of 485 nm and an emission wavelength of
520nm. The experiments regarding migration and/or invasion abilities with control
and/or modified cells were successfully repeated two or three times. Data availability
The GEO accession number for the data of transcriptome analysis in A549 and A549
TUSC3 KO (KO#4) cells was reported in this paper is GSE76515. The PrognoScan
Xenograft experiments. Animal experiments were conducted after approval of based Kaplan Meier plot shown at Fig. 1b was generated by previous report
the Institutional Animal Care and Use Committee, the Ohio State University. The (GSE31210). All relevant data are available from the authors upon request. A
nude mice each were intravenously injected with 5x105 of H460 TUSC3 KD cells or reporting summary for this Article is available as a Supplementary Information file.
control cells (Jackson laboratory). Four weeks after injection with H460 TUSC3 KD
cells, the mice were euthanized, and their lungs were biopsied. The samples were
prepared for H&E staining and the number of foci was counted under light Received: 17 May 2018 Accepted: 9 November 2018
microscope (Nikon, Eclipse 50i). For the IVIS in vivo imaging, the stable A549
TUSC3 knockdown and GFP+/luc+ cells were generated, and nude mice were
injected (i.v.) with the 1 × 106 A549 TUSC3 KD and GFP+/luc+ cells or control
cells. Luciferase activity was monitored using the IVIS in vivo imaging system
weekly for 6 weeks by i.v. injecting in vivo luciferin reagents (Promega). For the
in vivo tail-vein injection with TUSC3 KO and HRD1/TUSC3 DKO cells or H460
TUSC3KO, H460 ATF6α KD/TUSC3KO, H460 NM23H1/TUSC3KO cells References
(0.1–1 × 106 cells) were injected intravenously into two different groups consisting 1. Valastyan, S. & Weinberg, R. A. Tumor metastasis: molecular insights and
of 4-6 NSG mice. After 4 weeks, the colonization ability was obtained by calculating evolving paradigms. Cell 147, 275–292 (2011).
the tumor area in the lung tissue with Image J software. The xenograft experiments 2. Molina, J. R., Yang, P., Cassivi, S. D., Schild, S. E. & Adjei, A. A. Non-small cell
were performed once. For the detection of lung cancer metastasis after sub- lung cancer: epidemiology, risk factors, treatment, and survivorship. Mayo
cutaneous injections, the H460, H460/anti-miR-224/-520c, H460 TUSC3KO, or Clin. Proc. 83, 584–594 (2008).
H460 TUSC3KO/anti-miR-224/-520c cells were subcutaneously injected into four 3. Herbst, R. S., Heymach, J. V. & Lippman, S. M. Lung cancer. N. Engl. J. Med.
nude mice, respectively. After 4 weeks, the mice were euthanized, and lung 359, 1367–1380 (2008).
metastases were evaluated under microscope after H&E stain. Specifically, the 4. Clarke, H. J., Chambers, J. E., Liniker, E. & Marciniak, S. J. Endoplasmic
biopsied lung tissues were dissected with 6 different sections with 50 microns in reticulum stress in malignancy. Cancer Cell. 25, 563–573 (2014).
depth per mice. Tumor areas were measured by Image J software. 6-12 weeks old 5. Wang, M. & Kaufman, R. J. The impact of the endoplasmic reticulum protein-
mice were used in this study and their ages were matched to each group within an folding environment on cancer development. Nat. Rev. Cancer 14, 581–597
experiment and the animals were evenly allocated to each group in an experiment (2014).
in the case that their genders were different. No statistical analysis was used to 6. Upton, J. P. et al. IRE1alpha cleaves select microRNAs during ER stress to
predetermine the sample size. derepress translation of proapoptotic Caspase-2. Science 338, 818–822 (2012).
7. Wang, M., Wey, S., Zhang, Y., Ye, R. & Lee, A. S. Role of the unfolded protein
response regulator GRP78/BiP in development, cancer, and neurological
Recombinant lentiviral production. A1AT-NHK-ddVenus cDNA was subcloned
disorders. Antioxid. Redox Signal. 11, 2307–2316 (2009).
into pLL-CMV-Puromycin lentiviral vector. The TUSC3 lentivirus was harvested
by co-transfection of pLL-TUSC3 and packing constructs in HEK293 cells. After 8. Baumeister, P. et al. Endoplasmic reticulum stress induction of the Grp78/BiP
72 h, the supernatant was collected and enriched for the recombinant viruses by promoter: activating mechanisms mediated by YY1 and its interactive
incubating with a virus precipitation solution, PEG-iTtm (System Biosciences) for chromatin modifiers. Mol. Cell. Biol. 25, 4529–4540 (2005).
16 h at 4 ˚C. The recombinant lentivirus was recollected by centrifuging the 9. Yoshida, H., Haze, K., Yanagi, H., Yura, T. & Mori, K. Identification of the cis-
samples at 1500×g for 45 min at 4 ˚C. acting endoplasmic reticulum stress response element responsible for
transcriptional induction of mammalian glucose-regulated proteins.
Involvement of basic leucine zipper transcription factors. J. Biol. Chem. 273,
Subcellular fractionation assay. The cells were incubated with DMSO or TM at 33741–33749 (1998).
indicated time and concentration. After that the ER-Golgi and Nuclear fractions 10. Lee, A. S. Glucose-regulated proteins in cancer: molecular mechanisms and
were extracted by a subcellular fractionation protocol (ThermoFisher Scientific.). therapeutic potential. Nat. Rev. Cancer 14, 263–276 (2014).
Anti-Calnexin, anti-GM130, and anti-PARP1 antibodies were used for ER, Golgi 11. Baldridge, R. D. & Rapoport, T. A. Autoubiquitination of the Hrd1 ligase
and Nuclear fraction markers, respectively. Three independent experiments were triggers protein retrotranslocation in ERAD. Cell 166, 394–407 (2016).
consistently repeated in TUSC3 KO cells. 12. Fang, S. et al. The tumor autocrine motility factor receptor, gp78, is a
ubiquitin protein ligase implicated in degradation from the endoplasmic
PCR arrays for metastasis regulating genes. The overall procedure was per- reticulum. Proc. Natl Acad. Sci. USA 98, 14422–14427 (2001).
formed by the manufacturer’s instruction (SABiosciences). Briefly, total RNA 13. Shmueli, A., Tsai, Y. C., Yang, M., Braun, M. A. & Weissman, A. M. Targeting
purified from TUSC3 KD cells or control cells was subject to PCR array plates of gp78 for ubiquitin-mediated proteasomal degradation by Hrd1: cross-talk
containing probes for 84 known metastasis regulating genes. The standard qRT- between E3s in the endoplasmic reticulum. Biochem. Biophys. Res. Commun.
PCR was subsequently performed. Six housekeeping genes were used for nor- 390, 758–762 (2009).
malizing the expression of the genes (GAPDH, β-actin, β-microglobulin, Riboso- 14. Xu, Y. M. et al. HRD1 suppresses the growth and metastasis of breast cancer
mal protein large, PO, and HPRT1). The data from the qRT-PCR was analyzed by cells by promoting IGF-1R degradation. Oncotarget. https://doi.org/10.18632/
using RT2 profiler PCR Array Data analysis software (SABiosciences). The control oncotarget.5733 (2015).
and TUSC3KD cells were triplicated from independent cultured sets and then the 15. Kaneko, M., Ishiguro, M., Niinuma, Y., Uesugi, M. & Nomura, Y. Human
PCR array was performed once. HRD1 protects against ER stress-induced apoptosis through ER-associated
degradation. FEBS Lett. 532, 147–152 (2002).
Soft agar colony formation assay. Overall procedure was followed according to 16. Qu, L. et al. Endoplasmic reticulum stress induces p53 cytoplasmic
manufacturer’s protocol (Cell Biolabs, Inc.). A base agar matrix layer was prepared localization and prevents p53-dependent apoptosis by a pathway involving
in 96 well plate by solidifying at 4 ˚C for 30 min After that, a top agar matrix layer glycogen synthase kinase-3beta. Genes Dev. 18, 261–277 (2004).
was mixed with the 0.5 × 106 cells/ml of the cells and placed into upper layer of the 17. Yamasaki, S. et al. Cytoplasmic destruction of p53 by the endoplasmic
base agar matrix at 4 ˚C for 30 min Lastly, RPMI media supplemented with 10% reticulum-resident ubiquitin ligase 'Synoviolin'. EMBO J. 26, 113–122 (2007).

18. Cai, Y. et al. Loss of chromosome 8p governs tumor progression and drug 47. Grotzke, J. E., Lu, Q. & Cresswell, P. Deglycosylation-dependent fluorescent
response by altering lipid metabolism. Cancer Cell. 29, 751–766 (2016). proteins provide unique tools for the study of ER-associated degradation.
19. Kandoth, C. et al. Mutational landscape and significance across 12 major Proc. Natl Acad. Sci. USA 110, 3393–3398 (2013).
cancer types. Nature 502, 333–339 (2013). 48. Gao, B. et al. Synoviolin promotes IRE1 ubiquitination and degradation in
20. Vasickova, K., Horak, P. & Vanhara, P. TUSC3: functional duality of a cancer synovial fibroblasts from mice with collagen-induced arthritis. EMBO Rep. 9,
gene. Cell. Mol. life Sci. 75, 849–857 (2018). 480–485 (2008).
21. Bova, G. S. et al. Homozygous deletion and frequent allelic loss of chromosome 49. Subramanian, A. et al. Gene set enrichment analysis: a knowledge-based
8p22 loci in human prostate cancer. Cancer Res. 53, 3869–3873 (1993). approach for interpreting genome-wide expression profiles. Proc. Natl Acad.
22. Horak, P. et al. TUSC3 loss alters the ER stress response and accelerates Sci. USA 102, 15545–15550 (2005).
prostate cancer growth in vivo. Sci. Rep. 4, 3739 (2014). 50. Smith, S. C. & Theodorescu, D. Learning therapeutic lessons from metastasis
23. Kratochvilova, K. et al. Tumor suppressor candidate 3 (TUSC3) prevents the suppressor proteins. Nat. Rev. Cancer 9, 253–264 (2009).
epithelial-to-mesenchymal transition and inhibits tumor growth by 51. Chen, S. L. et al. P53 is a regulator of the metastasis suppressor gene Nm23-
modulating the endoplasmic reticulum stress response in ovarian cancer cells. H1. Mol. Carcinog. 36, 204–214 (2003).
Int. J. Cancer 137, 1330–1340 (2015). 52. Peng, Y. et al. TUSC3 induces autophagy in human non-small cell lung cancer
24. Jiang, Z. et al. TUSC3 suppresses glioblastoma development by inhibiting Akt cells through Wnt/beta-catenin signaling. Oncotarget https://doi.org/10.18632/
signaling. Tumour Biol. https://doi.org/10.1007/s13277-016-5072-4 (2016). oncotarget.17674 (2017).
25. Gutierrez, V. F. et al. Genetic profile of second primary tumors and 53. Molinari, F. et al. Oligosaccharyltransferase-subunit mutations in
recurrences in head and neck squamous cell carcinomas. Head Neck 34, nonsyndromic mental retardation. Am. J. Hum. Genet. 82, 1150–1157 (2008).
830–839 (2012). 54. Vanhara, P. et al. Loss of the oligosaccharyl transferase subunit TUSC3
26. Gu, Y. et al. TUSC3 promotes colorectal cancer progression and epithelial- promotes proliferation and migration of ovarian cancer cells. Int. J. Oncol. 42,
mesenchymal transition (EMT) through WNT/beta-catenin and MAPK 1383–1389 (2013).
signalling. J. Pathol. 239, 60–71 (2016). 55. Guervos, M. A. et al. Deletions of N33, STK11 and TP53 are involved in the
27. Zemliakova, V. V. et al. [Profile of methylation of certain tumor growth development of lymph node metastasis in larynx and pharynx carcinomas.
suppressing genes in non-small cell lung cancer]. Mol. Biol. 37, 983–988 Cell. Oncol. 29, 327–334 (2007).
(2003). 56. Jeon, Y. J. et al. A set of NF-kappaB-regulated microRNAs induces acquired
28. Duppel, U., Woenckhaus, M., Schulz, C., Merk, J. & Dietmaier, W. TRAIL resistance in lung cancer. Proc. Natl Acad. Sci. USA 112, E3355–E3364
Quantitative detection of TUSC3 promoter methylation -a potential (2015).
biomarker for prognosis in lung cancer. Oncol. Lett. 12, 3004–3012 (2016). 57. Nuovo, G. False-positive results in diagnostic immunohistochemistry are
29. Gu, Y. et al. Oncogenic function of TUSC3 in non-small cell lung cancer is related to horseradish peroxidase conjugates in commercially available assays.
associated with Hedgehog signalling pathway. Biochim. Biophys. Acta 1863, Ann. Diagn. Pathol. 25, 54–59 (2016).
1749–1760 (2017).
30. Yu, X. et al. Tumor suppressor candidate 3 as a novel predictor for lymph
node metastasis in lung cancer patients. Oncol. Lett. 12, 5099–5105 (2016). Acknowledgements
31. Croce, C. M. Causes and consequences of microRNA dysregulation in cancer. This work was supported by grants from the National Cancer Institute (1R35CA197706-
Nat. Rev. Genet. 10, 704–714 (2009). 01) to C.M.C. and National Natural Science Foundation of China (81672305) to R.C.
32. Keklikoglou, I. et al. MicroRNA-520/373 family functions as a tumor
suppressor in estrogen receptor negative breast cancer by targeting NF- Author contributions
kappaB and TGF-beta signaling pathways. Oncogene 31, 4150–4163 (2012). C.M.C. guided the project. Y.J.J. conceived the idea and wrote manuscript. Y.J.J, T.K., and
33. Huang, Q. et al. The microRNAs miR-373 and miR-520c promote tumour R.C. designed the experiments. YJ.J. and R.C. designed and performed in vivo mice
invasion and metastasis. Nat. Cell Biol. 10, 202–210 (2008). studies. G.N. performed in situ hybridization and immunohistochemistry experiments.
34. Cui, R. et al. MicroRNA-224 promotes tumor progression in nonsmall cell Y.J.J., T.K., P.J., D.P., R.C., S.R., J.J., H.S., and Y.P. performed in vitro experiments. Y.J.J,
lung cancer. Proc. Natl Acad. Sci. USA 112, E4288–E4297 (2015). T.K., I.N., Y.J., and S.K. analyzed cancer patient samples. Y.J.J., R.C., N.J., S.R. Y.J., and S.
35. Zhu, D. et al. Decreased microRNA-224 and its clinical significance in non- S. performed mice xenograft experiments and analyses. Y.J.J., B.L., and J.K. conducted
small cell lung cancer patients. Diagn. Pathol. 9, 198 (2014). GSEA analyses. Y.J.J., J.E.G., and P.C. established the system for ERAD detection. Y.J.J.,T.
36. Cherepanova, N. A. & Gilmore, R. Mammalian cells lacking either the K., R.C., Y.K., and M.G. analyzed and interpreted data. G.L., P.N., and C.M.C. pre-
cotranslational or posttranslocational oligosaccharyltransferase complex revised manuscript.
display substrate-dependent defects in asparagine linked glycosylation. Sci.
Rep. 6, 20946 (2016).
37. MacGrogan, D., Levy, A., Bova, G. S., Isaacs, W. B. & Bookstein, R. Structure Additional information
and methylation-associated silencing of a gene within a homozygously deleted Supplementary Information accompanies this paper at https://doi.org/10.1038/s41467-
region of human chromosome band 8p22. Genomics 35, 55–65 (1996). 018-07561-8.
38. Ferreira, H. J. et al. Circular RNA CpG island hypermethylation-associated
silencing in human cancer. Oncotarget 9, 29208–29219 (2018). Competing interests: The authors declare no competing interests.
39. Mohorko, E. et al. Structural basis of substrate specificity of human
oligosaccharyl transferase subunit N33/Tusc3 and its role in regulating protein Reprints and permission information is available online at http://npg.nature.com/
N-glycosylation. Structure 22, 590–601 (2014). reprintsandpermissions/
40. Cherepanova, N. A., Shrimal, S. & Gilmore, R. Oxidoreductase activity is
necessary for N-glycosylation of cysteine-proximal acceptor sites in
glycoproteins. J. Cell. Biol. 206, 525–539 (2014).
Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in
41. Schulz, B. L. et al. Oxidoreductase activity of oligosaccharyltransferase
published maps and institutional affiliations.
subunits Ost3p and Ost6p defines site-specific glycosylation efficiency. Proc.
Natl Acad. Sci. USA 106, 11061–11066 (2009).
42. Azare, J. et al. Constitutively activated Stat3 induces tumorigenesis and
enhances cell motility of prostate epithelial cells through integrin beta 6. Mol.
Open Access This article is licensed under a Creative Commons
Cell. Biol. 27, 4444–4453 (2007).
Attribution 4.0 International License, which permits use, sharing,
43. Puiffe, M. L. et al. Characterization of ovarian cancer ascites on cell invasion,
proliferation, spheroid formation, and gene expression in an in vitro model of adaptation, distribution and reproduction in any medium or format, as long as you give
epithelial ovarian cancer. Neoplasia 9, 820–829 (2007). appropriate credit to the original author(s) and the source, provide a link to the Creative
44. Wu, Y., Siadaty, M. S., Berens, M. E., Hampton, G. M. & Theodorescu, D. Commons license, and indicate if changes were made. The images or other third party
Overlapping gene expression profiles of cell migration and tumor invasion in material in this article are included in the article’s Creative Commons license, unless
human bladder cancer identify metallothionein 1E and nicotinamide N- indicated otherwise in a credit line to the material. If material is not included in the
methyltransferase as novel regulators of cell migration. Oncogene 27, article’s Creative Commons license and your intended use is not permitted by statutory
6679–6689 (2008). regulation or exceeds the permitted use, you will need to obtain permission directly from
45. Ishwad, C. S. et al. Frequent allelic loss and homozygous deletion in the copyright holder. To view a copy of this license, visit http://creativecommons.org/
chromosome band 8p23 in oral cancer. Int. J. Cancer 80, 25–31 (1999). licenses/by/4.0/.
46. Yoshida, H., Matsui, T., Yamamoto, A., Okada, T. & Mori, K. XBP1 mRNA is
induced by ATF6 and spliced by IRE1 in response to ER stress to produce a
© The Author(s) 2018
highly active transcription factor. Cell 107, 881–891 (2001).

MiRNA-mediated TUSC3 Deficiency Enhances UPR

Uploaded by

Copyright:

Available Formats

MiRNA-mediated TUSC3 Deficiency Enhances UPR

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

MiRNA-mediated TUSC3 Deficiency Enhances UPR

Uploaded by

Copyright:

Available Formats

ARTICLE

DOI: 10.1038/s41467-018-07561-8 OPEN

miRNA-mediated TUSC3 deﬁciency enhances UPR

NATURE COMMUNICATIONS | (2018)9:5110 | DOI: 10.1038/s41467-018-07561-8 | www.nature.com/naturecommunications 1

2 NATURE COMMUNICATIONS | (2018)9:5110 | DOI: 10.1038/s41467-018-07561-8 | www.nature.com/naturecommunications

(Normalized by wild type control)

Relative luciferase activity

(kDa) 1.0 0.23 0.45

g h miR520c Cytokeratin Merged RGB

miR224 TUSC3 Merged RGB miR520c TUSC3 Merged RGB

NATURE COMMUNICATIONS | (2018)9:5110 | DOI: 10.1038/s41467-018-07561-8 | www.nature.com/naturecommunications 3

Relative flurescence units

Photon flux rate (106)

0.3 P < 0.001 P < 0.001

KEGG_cell adhesion A549_#sh3C1

0.3 P = 0.023 P = 0.013

TUSC3 KO positively correlated TUSC3 KO positively correlated

4 NATURE COMMUNICATIONS | (2018)9:5110 | DOI: 10.1038/s41467-018-07561-8 | www.nature.com/naturecommunications

NATURE COMMUNICATIONS | (2018)9:5110 | DOI: 10.1038/s41467-018-07561-8 | www.nature.com/naturecommunications 5

0.5 NES=1.906 0.5 NES=1.913

PARP1 75 Transmembrane domain

f g Empty TUSC3 CCSS

100 Vinculin 37 TUSC3

6 NATURE COMMUNICATIONS | (2018)9:5110 | DOI: 10.1038/s41467-018-07561-8 | www.nature.com/naturecommunications

NATURE COMMUNICATIONS | (2018)9:5110 | DOI: 10.1038/s41467-018-07561-8 | www.nature.com/naturecommunications 7

i j k Suppressed genes by TP53

TUSC3KO positively correlated

0.6 NES= 2.511

8 NATURE COMMUNICATIONS | (2018)9:5110 | DOI: 10.1038/s41467-018-07561-8 | www.nature.com/naturecommunications

TUSC3 suppression regulates NM23H1/2 expression, which Discussion

NATURE COMMUNICATIONS | (2018)9:5110 | DOI: 10.1038/s41467-018-07561-8 | www.nature.com/naturecommunications 9

Normal/Localized cancer cells Metastatic cancer cells

TUSC3 TUSC3 deficiency

Regulating HRD1- Protein Enhanced HRD-1- Abnormal

Normal UPR / ERAD p53 IRE-1α PERK ATF6α

10 NATURE COMMUNICATIONS | (2018)9:5110 | DOI: 10.1038/s41467-018-07561-8 | www.nature.com/naturecommunications

NATURE COMMUNICATIONS | (2018)9:5110 | DOI: 10.1038/s41467-018-07561-8 | www.nature.com/naturecommunications 11

12 NATURE COMMUNICATIONS | (2018)9:5110 | DOI: 10.1038/s41467-018-07561-8 | www.nature.com/naturecommunications

NATURE COMMUNICATIONS | (2018)9:5110 | DOI: 10.1038/s41467-018-07561-8 | www.nature.com/naturecommunications 13

You might also like