Altered polymerase theta expression promotes chromosomal instability in salivary adenoid cystic carcinoma

Abstract Genomic instability (GIN) plays a key role in cancer progression. The disorders of polymerase theta (POLQ) were reported to contribute to GIN and progression in many cancers. Here, we found that POLQ over‐expression was related to salivary adenoid cystic carcinoma (SACC) progression and poor prognosis. Then, we investigated the role and mechanism of POLQ in the GIN in SACC. GIN was assessed by chromosome staining with DAPI and Giemsa, as well as qRT‐PCR of the mitosis‐related gene expression. Meanwhile, PCR‐SSCP was used to evaluate microsatellite instability. Modulation of POLQ expression increased chromosomal instability and enhanced the sensitivity to etoposide without impacting microsatellite stability. Mechanistically, POLQ regulated genome stability by promoting the expression of the error‐prone alt‐NHEJ‐related protein PARP1, and down‐regulating c‐NHEJ‐ and HR‐related proteins KU70 and RAD51. In vitro CCK, Transwell assays and in vivo murine xenograft models indicated that the PARP inhibitor olaparib suppressed SACC growth in the case of etoposide‐induced DNA damage. Bioinformatic analysis identified CEBPB as a potential POLQ‐regulating transcription factor. In summary, our research provides new insights into the mechanisms of SACC chromosomal instability and identifies new potential targets for SACC treatment.

treatment strategies for controlling SACC progression and ultimate improving prognosis.
Tumours are characterized by genomic instability (GIN), which leads to a higher probability for genomic changes and the acquiring of deleterious mutations that may lead to tumour progression. 6 GIN includes chromosomal instability (CIN) and microsatellite instability (MSI). 7 In response to radiotherapy, most tumour cells exhibit an excessive GIN that results in cell apoptosis, but a small portion of tumour cells acquire non-fatal GIN, which enhances tumour heterogeneity, evolution and progression. 8 Recent studies identified the more genomic mutations in recurrent/metastatic adenoid cystic carcinoma (ACC) compared with primary ACC. 9 Furthermore, more than half of SACC cases characterized with GIN had worse outcomes compared with the SACC cases without GIN. 10 These studies indicate that GIN promotes SACC progression, thus, suggesting that the mechanism underlying GIN is key to inhibit SACC progression.
In normally proliferating cells, DNA damage or mitotic defects are repaired by multiple DNA repair pathways. In human cells, the double-strain break (DSB) repair includes the major DNA repair manner, including the classical non-homologous end joining (c-NHEJ) and homologous recombination (HR) pathways, as well as the errorprone repair manner, the alternative non-homologous end-joining (alt-NHEJ) pathway. In mammals, the c-NHEJ and HR pathways are the most precise DSB repair pathways that maintain chromosomal stability throughout the cell cycle. When c-NHEJ and/or HR function are deficient, the alt-NHEJ pathway will be activated. 11 However, the alt-NHEJ pathway is easy to cause GIN, such as DNA sequence loss, chromosome translocation and other events; so, it is also regarded as a main driver of cancer progression. 12,13 Polymerase theta (POLQ or POLθ) is an A-family DNA polymerases that is widely expressed in eukaryotes and is highly expressed in tumour cells and some normal human tissues. POLQ functions as a key factor in the alt-NHEJ pathway and participates in many cellular processes, including DNA repair and DNA replication. 14 Increased expression of POLQ has also been detected in many kinds of tumours, such as lung cancer, colorectal cancer and gastric cancer. 15 A higher expression of POLQ was significantly related to the poor survival in colorectal cancer and breast cancer . 16,17 Studies showed that POLQ-addicted cells and cancers had a similar GIN feature, showing an increased prevalence of alt-NHEJ-related microhomology-flanked deletions. 18 POLQ depletion also reduces aberrant replication products and chromosomal rearrangements. 19 However, there were few studies concerning the effect of POLQ on genomic stability, progression and prognosis in SACC. Here, we found that POLQ expression was related to SACC progression and poor prognosis. Through examining the function and regulatory mechanism of POLQ in SACC, we found that inhibiting POLQ or POLQ-related pathways increased the sensitivity of SACC cells to DNA damage. Furthermore, we identified a new regulator of POLQ, the bZIP transcription factor CEBPB, which could bind to POLQ promoter. This study provides the evidence for inhibiting SACC progression and optimizing current therapies for the treatment of SACC.

| Patients and follow-up
The data of the normal salivary glands and the SACC tissues that were surgically resected at Dalian Medical University (Dalian, China) between January 2001 and December 2013 were summarized and re-examined according to our previous study. 20 The pathological diagnosis in this study was conducted by two head and neck can-  Table 1.

| RNA extraction and quantitative real-time PCR (qRT-PCR)
RNA extraction and qRT-PCR were performed as previously described. 22  Real-Time PCR System (Thermo Fisher Scientific, 4484642) and the specific primers for human genes (Takara). The qRT-PCR results were obtained from three independent replicates. The relative changes in gene expression were summarized based on the 2 −ΔΔ CT method. All of the primer sequences are listed in Table S1.

| Western blot
Protein concentrations of cell lysates were determined using a BCA protein assay (KeyGen, KGP902). Equal amounts of protein were separated by 8%, 10%, or 15% SDS-PAGE and transferred onto nitrocellulose filter membranes (Solarbio, YA1710) or polyvinylidene difluoride membranes (GE, 10600023) by electroblotting. The membranes were blocked and then, incubated with primary antibodies at 4°C overnight. The primary antibodies are listed in Table S2.
Anti-GAPDH antibody and anti-VINCULIN antibody were used as loading controls. After washing, the membranes were incubated with secondary antibodies. The blots were washed, and signals were visualized with WesternBright™ ECL (Advansta, K-12045-D50).
The slides were then observed using an Olympus BX43 microscope (Olympus Corporation).

Factor
No. control group (saline with the same volume by the same injection way on that day).

Average
Tumour size was limited to less than 1000 mm 3 in compliance with animal welfare regulations. After 14 days of treatment, the mice were euthanized with CO 2 , and lack of respiration and heartbeat were used as an indicator of death. The xenografts were measured and retrieved. The specimens were divided into two sets: a half was kept frozen and stored at −80°C, and the remaining half was prepared as a paraffin block.

| Cell proliferation, cell migration and invasion assays
The

| Cytotoxicity assay
The MTT assay was performed to evaluate drug cytotoxicity in vitro.

| Dual-luciferase reporter assay
The wild-type POLQ promoter sequence was cloned into the GPL4-Basic vector and the CEBPB wild-type sequence was cloned into the   Table S3. 23 For SSCP, 5 μl of the PCR products mixed with 1 μl of 6× TriTrack DNA Loading Dye (Thermo Scientific™, SM1211) were loaded on a 1.0 mm 8% non-denaturing PAGE gel and electrophoresed at room temperature, 80 V, for ~2 h. The gel was then, analysed by silver staining (Solarbio, G7210) and imaging. The shift of forhead bands compared with the control group after electrophoresis was considered as the positive MSI.

| Statistical analysis
All data were statistically analysed with IBM SPSS statistics 20.0

| POLQ was highly expressed in relatively malignant SACC cell lines and tumours
To examine the expression of POLQ in SACC, we first compared the median expression of POLQ in tumour and normal samples from the GEPIA website (http://gepia.cance r-pku.cn/detail.php?gene=POLQ).
Since POLQ was widely expressed in various human organs and tissues ( Figure S1A), we extracted the median expression values from  Figure 1C). The clinical information of tumour cases and results of POLQ immunohistochemistry staining was summarized in Figure 1D and  Figure 1E).
Collectively, these observations implicated an enhancement of POLQ on SACC progression.

| Aberrant expression of POLQ promoted CIN in SACC
Then, we investigated whether POLQ played a role in genomic sta- POLQ-OE and shPOLQ SACC-83 cells exhibited microsatellite stability at the five loci, suggesting that they were microsatellite stable cells ( Figure 2D). These data suggested that POLQ influenced little microsatellite stability in SACC-83 cells.
Together, these results suggested that the dysregulated POLQ expression increased the level of CIN, but not MSI, in SACC cell lines.

| POLQ regulated SACC cell proliferation, migration and invasion in vitro and tumour size in vivo
To assess the role of POLQ in SACC malignancy, we examined To evaluate the effects of POLQ on SACC in vivo, we injected SACC-83 cells with over-expressed POLQ into nude mice. As shown in Figure 3D, the xenograft growth rate in the POLQ-OE group was decreased compared with the control SACC-83 xenografts. In addition, POLQ over-expression dramatically decreased the size of SACC tumours. These results suggested that POLQ over-expression inhibited SACC xenograft growth.

| Altered POLQ expression increased the sensitivity of SACC cell lines to etoposide
Because POLQ was related to DNA repair, we evaluated the effect POLQ over-expression was related to poor prognosis in many cancers. 16,17 We, therefore, examined the potential efficacy of POLQ inhibition under DNA damage in SACC tumours. We established xenografts in athymic mice using shPOLQ SACC-LM and control cell lines, and treated mice with etoposide to determine the impact of altered POLQ expression on SACC tumours under DNA damage. We found that POLQ suppression combined with etoposide treatment showed greater efficacy in reducing tumour growth compared with etoposide alone ( Figure 5C).

| POLQ negatively regulated RAD51 and KU70 expressions
Our results showed that both over-expressed and suppressed POLQ induced CIN, even independent of etoposide treatment. However, why the over-expression of POLQ was related to poor prognosis in cancer and the differential mechanisms between POLQ overexpressing and suppressing cells were not clear. CIN is primarily associated with defects in the DNA damage response. Therefore, we evaluated the effects of POLQ on DNA damage repair pathways.
In POLQ-OE and shPOLQ SACC cells, γH2AX levels were increased compared with the corresponding control groups ( Figure 6A). The levels of RAD51, a key protein of HR pathway, and

| The PARP1 inhibitor olaparib improved the sensitivity of SACC to etoposide
As our results suggested a correlation between POLQ and PARP1, we explored the effect of a treatment targeting PARP1 on SACC in the single treatment groups ( Figure 7F). Collectively, these data confirmed that SACC tumours with olaparib treatment were more sensitive to DNA damage in vitro and in vivo.

| The regulatory effects of CEBPB on POLQ in SACC cell lines
To examine whether POLQ expression was regulated by CEBPB, we

| DISCUSS ION
As a slow growing malignant tumour hard to be diagnosed, 31,32 SACC is considered as a biologically deceptive and frustrating head and neck tumour. SACC (especially the solid pattern type) displays heterogeneity with aggressive behaviour and treatment resistance, which lead to a poor long-term prognosis. 33 GIN is a characteristic of human cancers.
The presence of GIN in most SACC cases and the higher GIN levels in recurrent/metastatic ACC 1 suggest that GIN is an important driver in SACC progression. Therefore, we examined the GIN-related POLQ expression in SACC. Bioinformatics analysis suggested that POLQ promote the progression of most glandular epithelial cancers, however, there was no data on SACC reported ( Figure S1). POLQ expression in the highly metastatic SACC cell line SACC-LM and solid SACC was higher than thoes in SACC-83 cells ( Figure 1A-B) and other types of SACC, respectively ( Figure 1C-D). POLQ over-expression was related to the poor SACC prognosis ( Figure 1E). These findings suggest that POLQ promotes SACC progression.
Whether POLQ, the key factor in the alt-NHEJ error-prone repair pathway, promotes or inhibits GIN requires to be elucidated. Some studies showed that cells lacking POLQ exhibited the hypersensitivity to various DNA damaging agents, 34 implicating that POLQ inhibited GIN. However, other evidences suggested that POLQ over-expression increased DNA damage. 17 A recent review showed that the alt-NHEJ pathway mediated by POLQ had the characteristics of 'error guaranteed', and the aberrant POLQ expression could cause GIN: when POLQ was down-regulated, DNA repair capability was decreased and GIN increased; in contrast, when POLQ was over-expressed, the alt-NHEJ pathway was over-activated, and the error prone repair led to a large number of genomic scars and GIN increased. 24 Thus far, few study concerned the relationship between POLQ and GIN in SACC. In this study, we examined the relationship between POLQ and GIN in SACC. DNA repair defects can lead to chromosome breaks, which induce nuclear dysmorphology, including the precursor structures of CIN, such as crescent, and the apoptosisrelated morphological changes, such as micronucleus and abnormal size. 35,36 We found a large number of apoptosis-related morphological changes in SACC cells with the aberrant POLQ expression. We We investigated the effect of POLQ on SACC mitosis by detect-

ing CENP-A and HJURP expression. Centromere protein A (CENP-A)
is a centromere specific variant of canonical histone H3 and indispensable for cancer progression. 38 Over-expression of CENP-A cause CIN-related mitosis defects and centromere dysfunction, which are the hallmarks of cancer. 38 Holliday junction recognition protein (HJURP), the chaperone of CENP-A, deposits CENP-A at centromeres. 38 HJURP is over-expressed and related to the poor prognosis in some cancers. 38 The loss of CENP-A or HJURP leads to cell senescence, 39,40 and the depletion of CENP-A causes mouse embryonic lethality. 41 We found that POLQ positively regulated CENP-A and HJURP transcription, suggesting that POLQ influenced chromosomal stability and SACC progression by regulating mitosis.
These findings suggested that POLQ regulated chromosomal stability through impacting DNA damage. Furthermore, POLQ impact microsatellite stability little ( Figure 2D). Mismatch repair defect is an important reason of MSI, and MLH1 expression in SACC was also impacted little by POLQ ( Figure 6C). Previous studies have shown microsatellite stability in ACC. 9 These results further prove that POLQ works as a regulator of CIN, instead of MSI in SACC.
We found that POLQ over-expression inhibited SACC cell prolif-   45 We, therefore, examined the sensitivity of POLQ over-expressing or inhibited cells to etoposide. As a high dose of etoposide may induce too many irreparable DSBs, we used approximately half of the maximally inhibitory concentration (IC 50 ) to avoid inducing too many DNA damages to overwhelm POLQ function. We found that whether POLQ expression We also found both over-expression and inhibition of POLQ  50 We found that olaparib suppressed SACC-83 cell proliferation in a concentration-dependent manner.
The etoposide and olaparib combination suppressed SACC-83 proliferation, migration and invasion in vitro and inhibited SACC xenograft growth in vivo. These findings suggested that olaparib improved the inhibitory effect of etoposide on SACC. 42,43 Olaparib inhibits the chromosomal translocation produced by IR or topoisomerase II inhibitor etoposide, suggesting that chromosome translocation is decreased by the suppression on alt-NHEJ pathway. Therefore, the combined application of olaparib and etoposide has no accumulated effect.
To investigate the mechanism regulating POLQ, we examined the regulators upstream to POLQ. Most DNA polymerases are directly regulated by transcription factors 29,30 ; therefore, we speculated that POLQ was regulated by a transcription factor. CCAAT/ enhancer binding protein beta (CEBPB) is a bZIP transcription factor in many cellular processes, such as proliferation, differentiation and apoptosis. 51 Abnormal over-expression of CEBPB has been reported in many human cancers and is related to cancer progression and poor prognosis. CEPBP is regarded key to inhibit cancer progression and treatment resistance. 52 Bioinformatics data and the verification experiments suggested that CEBPB directly bond to POLQ promoter. However, we just detected the slight changes of POLQ in the CEBPB-OE and siCEBPB SACC-83 cells. We suspect that there are other mechanisms preventing POLQ expression from changing.
We plan to identify the relationship between CEBPB and POLQ in future studies.
In summary, our results indicate that POLQ inhibits the expressions of DNA repair factors, KU70 and RAD51, but promotes PARP1 expression to regulate DNA damage repair and CIN formation ( Figure 10). Therefore, POLQ inhibits the excessive CIN to reduce the risk of cell death and enhance cellular tolerance to DNA damage in SACC. POLQ over-expression is also a CIN promoting factor by accumulating mutations which promote SACC heterogeneity, evolution and progression. POLQ depletion or olaparib treatment inhibits SACC growth with DNA damaging agent by inducing excessive CIN.
Moreover, CEBPB is functionally related with POLQ, which should be pursued in further research.

CO N FLI C T O F I NTE R E S T
None of authors have conflicts of interest to declare.

DATA AVA I L A B I L I T Y S TAT E M E N T
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.