URI1 amplification in uterine carcinosarcoma associates with chemo-resistance and poor prognosis

Abstract

Uterine carcinosarcoma (UCS) is a rare type of cancer and accounts for 5% of uterine malignancies. However, UCS patients suffer a high prevalence of chemo-resistance and a very poor prognosis compared to uterine cancer patients. URI is a chaperone with functions in transcription. We analyzed the somatic URI1 copy number variation in 57 post-menopausal non-metastatic UCS patients in comparison to 363 uterine corpus endometrial carcinomas. URI1 amplification was detected in 40% (23/57) of primary UCS and 5.5% (20/363) of uterine carcinomas. UCS patients with URI1 amplification exhibited 13% (3/23) tumor-free survival compared to 41% (14/34) in the absence of URI amplification (P=0.023). URI1 amplification (OR=6.54, P=0.027), weight (OR=1.068, P=0.024), hypertension (OR=3.35, P=0.044), and tumor stage (OR=2.358, P=0.018) associated with poor survival. Patients treated with hormone replacement therapy (OR=15.87, P=0.011) displayed enhanced overall survival. Combined radiation and chemotherapy improved patient survival (median survival=2043 days) compared to single (median survival=597 days) or no treatment (median survival=317 days, P=0.0016). Importantly, patients with URI1 amplification had poor response to adjuvant treatment compared to control group (P=0.013). Tumors with URI1 amplification displayed decreased transcription of genes encoding tumor suppressor and apoptotic regulators and increased expression of genes regulating oncogenesis, survival and metastasis. Overexpression of URI1 in a cultured cell model induced ATM expression and resistance to cisplatin. Our findings suggest that high prevalence in UCS may associate with poor prognosis and worse response to adjuvant treatment.

Introduction

Uterine carcinosarcoma (UCS), also known as malignant mixed Mullerian tumor (MMMT), is an undifferentiated uterine malignancy with characteristics of both carcinoma and sarcoma. The endometrial carcinoma occurs within the inner layer of tissue lining of the uterus, whereas the sarcoma arises from the outer layer of muscle of the uterus. USC makes up five percent of all uterine cancers [1]. In the United States, approximately 2 per 100,000 women develop UCS annually [2]. Because of the aggressive nature of UCS, only 35% of patients survive five years after diagnosis. UCS and endometrial carcinomas have similar risk factors. Both malignancies are associated with obesity, diabetes, hypertension, smoking, nulliparity, and use of estrogen ortamoxifen [3-9]. By contrast, progestin-containing contraceptives or postmenopausal hormone therapy reduces the risk of both types of cancers [10-13]. The primary management of UCS is surgical, although adjuvant radiotherapy treatment (RT) and/or chemotherapy are often used. While one study reported that adjuvant RT after surgery decreased the risk of pelvic cancer recurrence compared to surgery alone [14], it did not provide any overall survival benefit [14]. Additional studies have provided evidence to support the benefit of aggressive adjuvant chemotherapy that combines RT with DNA damage-based chemotherapy [15-17]. Currently, it remains challenging to predict outcomes for UCS patients and the mechanism underlying patient relapse is unclear. Therefore, a reliable preoperative biomarker that predicts primary surgical response and identifies patients who would benefit from aggressive adjuvant therapy is urgently needed.

The unconventional prefoldin RPB5 interactor 1 (URI1) was originally identified as a scaffold protein that binds RNA polymerase II [18]. URI plays important roles in regulating gene expression. URI interacts with several transcription factors including transcription factor IIF (TFII), and the androgen receptor (AR) in prostate cancer cells [19-21]. It also interacts with a chromatin remodeling complex (PAF1) and translation initiation factors [22,23]. Current clinical findings have shown the correlation between URI1 and human cancers. For example, one study reported URI1 amplification in 10% of ovarian cancers and increased URI protein level correlates with tumor aggressiveness [24]. URI upregulation increases the expression of the “p53 and DNA damage-regulated gene 1” (PDRG1), suggesting a function for URI in DNA damage repair [19]. Decreased URI1 expression inhibits cell proliferation and induces apoptosis in ovarian and liver cancer cells [25,26]. In cervical cancer, increased URI expression is also associated with a high tumor grade [27].

Here, we hypothesize that URI1 amplification is associated with poor clinical outcome in UCS patients. We also reconstituted URI1 upregulation in a culture cell model and investigated URI overexpression in the DNA damage response.

Materials and methods

Patient cohort

We analyzed UCS and uterine corpus endometrial cacinoma samples with corresponding normal tissue from clinically annotated patients in The Cancer Genome Atlas (TCGA) Data Portal. All samples met freedom-to-publish criteria without restriction or limitations. The uterine corpus endometrial carcinomas cohort has been reported previously [28]. All UCS patients were postmenopausal (prior bilateral ovariectomy or >12 months since last menstrual period with no prior hysterectomy). All UCS specimens were surgically resected prior to systemic treatment and all patients received complete surgery as the primary treatment. UCS Patients did not have metastasis at the time of surgery. The clinical stage, age at diagnosis, tumor invasion percent, local lymphatic status, and surgical margin, were recorded. Patient race, weight, menopausal hormone therapy, hormonal contraceptives use, tamoxifen use, hypertension, diabetes, full-term pregnancies, and history of other malignancies were also recorded. Treatment response, time of disease relapse and date of death after initial diagnoses were recorded after the initial treatment. Lymph node positivity was determined by H & E staining and immunohistochemistry. We excluded patients who received neoadjuvant treatment. We also excluded patients with a history of tamoxifen use or with colorectal cancer, as some colon cancers can metastasize to the uterus. Using these criteria, we identified 57 UCS patients and determined URI1 copy number variation and mRNA expression.

Statistical analysis

Primary therapy outcome, tumor-free status, new tumor event, and surgical margins between URI1 amplification and control groups were compared using the chi-square test. The differences in age at diagnosis, and follow up time between two groups were evaluated with an unpaired t-test. Tumor stage between control and URI1 amplifed groups was evaluated with the Mann-Whitney test. Spearman’s rank correlation coefficient was used to analyze pairwise correlation between patient characteristics. Disease-associated variables were analyzed by univariate and multivariate Cox regression to test correlation with survival. Pairwise correlation efficiency between variables was calculated by Spearman’s rho. Somatic URI1 copy number abnormalities were identified by SNP microarray with putative copy number=2 from GISTIC 2.0 and genome-wide exome sequencing as described previously [28,29]. Overall survival was calculated using the Kaplan-Meier method, and the significance between groups was calculated by Wilcoxon test. Statistical analyses were performed using Prism version 6.0 (GraphPad Software, La Jolla, CA), cBioPortal [30], and SPSS version 13.0 (SPSS, Inc., Chicago, IL). All in vitro experiments were performed three times independently and the error bars represent the standard deviation with significance calculated by nonparametric t-test.

Cell survival assay and reagents

RL 95-2 cells (ATCC, Manasas, VA), a uterine/endometrial cancer line, were cultured in DMEM: F12 medium with 0.005 mg/ml insulin and 10% FBS. Cells were plated in 96-well plates at a concentration of 1,000 cells per well. After 48 hours of treatment, cell viability was measured using the CyQUANT assay (Life Technologies, Waltham, MA), as per manufacturer’s instruction. Cisplatin is from Sigma-Aldrich (St. Louis, MO). ATM inhibitor is from Selleckchem (Houston, TX) (Catalog #S7136). Anti-URI1 antibody is from Santa Cruz Biotechnology (Santa Cruz, CA) (Catalog #376011). Anti-phospho-ATR (Catalog #2853), phospho-ATM (Catalog #5883), ATM (Catalog #2873), ATR (Catalog #2790), and phospho-γH2AX (Catalog #9718) antibodies are from Cell Signaling (Berverly, MA). Anti-tubulin antibody (Catalog #489P) is from Biolegend (San Diego, CA).

Quantitative reverse transcription-polymerase chain reaction (qRT-PCR)

qRT-PCR to quantitate ATM mRNA expression used SYBR Green (Applied Biosystem, Foster City, CA). Expression of RPL-19 was used as the internal control. ATM primers are as following: F: AGACCGCGTGATACTGGATG. R: TCACTGTCACTGCACTCGGA.

Result

Patient cohort

This series of UCS (n=57) has a median age of diagnosis at 68 yrs, and median follow-up of 497 days (Table 1). In this cohort, stage I, II, III and IV cases are 39%, 9%, 35%, and 18% respectively. Pathology examination found 15 (26%) cases with positive pelvic lymph nodes involved and 9 (16%) cases with positive aortic lymph nodes involved (Table 1). Surgical margins were assessed in 12 (21%) cases (Table 1). Among these 57 patients, 29 (51%) experienced tumor recurrence (Table 1). After surgery, 7 patients received only RT and 17 patients received only chemotherapy (Table 1). RT in combination with chemotherapy was given to 19 patients and 14 patients did not receive any adjuvant treatment (Table 1). In this cohort, 56 (98%) have TP53 mutations, 22 (39%) have PI3KCA mutations, and 16 (28%) have PTEN mutations.

Table 1.

Clinicopathologic characteristics of the overall cohort

Characteristic	Value
Patient size	57
Age (y)
Median	68
Range	51-90
Race
White	44
Black	9
Asian	3
Unknown	1
Follow up (days)
Median	497
Range	8-3115
Stage
I	22
II	5
III	20
IV	10
Positive pelvic lymph nodes	15 (26%)
Positive aortic lymph nodes	9 (16%)
Tumor recurrence	29 (51%)
Surgical margin	12 (21%)
Depth of myometrial invasion %
Median	40
Range	0-100
Residual tumor
R0	34
R1	2
R2	10
Rx	11
Adjuvant treatment
No	14
Radiation treatment	7
Chemotherapy	17
Radiation+chemo	19

URI1 amplification in UCS associates with decreased tumor-free status

We first compared clinical and pathological characteristics in cases with normal or elevated URI1 copy number. URI1 amplification (URI1 amp) was detected in 23 (40%) patients (Table 2). Interestingly, we found URI1 amplification in only 20 (5.5%) uterine corpus endometrial carcinomas (n=363) (Table S1). In UCS patients, the control group and the URI1 amplified group have a similar age of diagnosis (69.2±9.9 vs. 70.0±8.4 yrs, P=0.57) and follow-up time (743±206 vs. 800±208 days, P=0.79) (Table 2). The URI1 amp group contains 65% of tumors in III/IV stage, compared to 44% of the control group (P=0.2) (Table 2). Patients with URI1 amplification trended toward deeper myometrial invasion (55%), higher chance of pelvic lymph node involvement (35%), and increased positive surgical margins (30%), compared to the control group (33%, 21%, and 15%, respectively) (Table 2). To determine the correlation of URI1 copy number and patient outcome, we then compared disease progression in the control and URI1 amp groups (Table 2). After primary therapy, 43% of patients in the URI1 amp cohort had partial or complete response compared to 61% in the control group (P=0.18) (Table 2). Importantly, 41% of patients in the control group had tumor-free survival while only 13% of patients with URI1 amplification remained tumor-free (P=0.023).

Table 2.

URI1 amplification associates with patient pathology and prognosis

URI1 amplification	No	Yes	P-value
Sample size	n=34	n=23
Age (yrs)	69.2±9.9	70.0±8.4	0.57
Stage			0.2
I	16 (47%)	6 (26%)
II	3 (9%)	2 (8%)
III	9 (26%)	11 (48%)
IV	6 (18%)	4 (17%)
Primary therapy outcome success1	21 (61%)	10 (43%)	0.18
Tumor free survival	14 (41%)	3 (13%)	0.023
Tumor recurrence	18 (53%)	11 (48%)	0.7
Days of follow-up	743±206	800±208	0.79
Median survival days	771±142	685±198	0.51
Pelvic lymph nodes	7 (21%)	8 (35%)	0.23
Aortic lymph nodes	5 (15%)	4 (17%)	0.31
Depth of myometrial invasion %2	33±10.2	55±15.2	0.19
Surgical Margin	5 (15%)	7 (30%)	0.15

¹Defined as complete or partial remission;

²Depth of myometrial invasion divided by depth of myometrial thickness.

URI1 amplification, weight, hypertension, and tumor stage associate with UCS patient overall survival

Next, we investigated what pre-operative characteristics and adjuvant treatments associate with patient survival. Among the 14 morphologic variables in our study, URI1 amplification (odds ratio (OR)=6.54, confidential interval (95% CI) 1.123-34.7, P=0.027), increasing weight (OR=1.068, 95% CI 1.009-1.13, P=0.024), hypertension (OR=3.35, 95% CI 1.032-11.898, P=0.044), and tumor clinical stage (OR=2.358, 95% CI 1.108-5.018, P=0.018) were significantly correlated with mortality (Table 3). A history of menopausal hormone therapy (OR=0.063, 95% CI 0.008-0.527, P=0.011) and chemotherapy (OR=0.074, 95% CI 0.014-0.381, P=0.002) were significantly associated with patient overall survival (Table 3). Correlation coefficient analysis revealed that URI1 amplification does not associate with any preoperative factors except hormonal contraceptives (r _s=-0.521, P=0.011) (Table S2). Patients who have a history of using hormonal contraceptives tend to have a normal URI1 copy number in USC malignancies. These data suggest that URI1 amplification could be used as an independent factor to predict patient prognosis.

Table 3.

Multivariate logistic regression model of characteristics associated with survival in UCS

Variables	Univariant			Multivariant
	P-value	OR	95% CI	P-value	OR	95% CI
URI1amp	0.551	1.240	0.612-2.513	0.027	6.543	1.234-34.7
Age at diagnosis	0.338	0.992	0.977-1.008	0.235	1.058	0.964-1.162
Weight (l b)	0.599	1.005	0.986-1.025	0.024	1.068	1.009-1.13
Hormonal contraceptives use	0.895	0.916	0.25-3.352	0.053	11.754	0.967-142.831
Menopausal hormone therapy	0.055	0.230	0.051-1.031	0.011	0.063	0.008-0.527
Hypertension	0.003	3.217	1.476-7.008	0.044	3.505	1.032-11.898
Diabetes	0.341	1.595	0.61-4.168	0.248	0.395	0.082-1.91
Pregnancies	0.171	0.769	0.528-1.12	0.24	0.708	0.398-1.259
Clinical stage	0.033	1.414	1.028-1.946	0.026	2.358	1.108-5.018
Pelvic lymph nodes positive	0.928	0.964	0.429-2.163	0.204	0.448	0.129-1.549
Aortic lymph nodes positive	0.090	2.021	0.897-4.555	0.697	0.774	0.214-2.807
Surgical margin	0.368	1.430	0.657-3.112	0.967	0.97	0.222-4.232
Radiation therapy	0.011	0.386	0.184-0.807	0.894	0.924	0.289-2.961
Chemotherapy	0.115	0.553	0.264-1.155	0.002	0.074	0.014-0.381

OR: Odds ratio, 95% CI: 95% confidence interval.

To analyze whether RT and chemotherapy provide benefit to patients with or without URI1 amplification, we analyzed the overall survival in the control and URI1 amp groups. Patients who received adjuvant RT or chemotherapy alone had significantly improved overall survival (median survival=597 days) compared to the untreated group (median survival=317 days, P=0.026) (Figure 1A). Patients with combined RT and chemotherapy had the best overall survival (median survival=2043 days) compared to single adjuvant treatment (P=0.029) or the non-treated group (P=0.0024) (Figure 1A). URI1 amplification in non-treated patients did not significantly affect overall survival (P=0.58) (Figure 1B). Among UCS patients who received single adjuvant treatment, the URI1 amp group had a much worse prognosis (median survival=442 days) compared to control patients (median survival=771 days, P=0.013) (Figure 1C). A combination of chemotherapy and RT dramatically improved overall survival in both URI1 amp (median survival=2043 days) and control group (median survival=3115 days) and the difference between these groups was not significant (P=0.053) (Figure 1D). Thus, our data indicates that patients with URI1 amplification are less responsive to single adjuvant treatments while RT-chemotherapy combination can significantly improve patients overall survival in both control and URI1 amp patients.

Figure 1.

Association between URI1 amplification and overall survival in UCS. A. Patients with adjuvant RT or chemotherapy (n=24) had significantly improved overall survival compared to patients without adjuvant treatment (n=14, P=0.026). Patients with combination of adjuvant RT and chemotherapy (n=19) have better overall survival compared to single treatment (P=0.029) and non-treated control (P=0.0024). B. Comparison of patient overall survival in control (n=8) and URI1 amplified group (n=6) without adjuvant treatment (P=0.58). C. After adjuvant RT or chemotherapy, URI1 amplified group (n=9) has the worse overall survival compared with control group (n=15, P=0.013). C. Comparison of URI1 amplified (n=8) and control group (n=11) overall survival after adjuvant combination of RT and chemotherapy (P=0.53).

URI1 amplification correlates with alteration in cancer-related gene expression and induces DNA damage-resistance through ATM upregulation

To elucidate the potential mechanism underlying the association between URI1 amplification and poor prognosis, we compared the expression of 220 genes that regulate cell cycle, proliferation, apoptosis, cancer metastasis, and differentiation. We specifically excluded genes on chromosome 19 where the URI gene resides to avoid genome linkage effects as a result of URI amplification. We identified 33 genes with significant mRNA expression alteration, among which 13 genes were downregulated and 20 were upregulated (Figure 2A and Table S3). Functional cluster analysis showed significant alteration in apoptosis and cell proliferation pathways. For example, URI1 amp tumors have decreased expression of the pro-apoptotic enzyme CASP8 (P=0.0035), the tumor suppressor PTEN (P=0.0082), the membrane-bound cytochrome with tumor suppressor and autophagy functions CYB5A (P=0.0022), and the transcriptional activator of proliferative genes E2F5 (P=0.031. Figure 2B). UCS tumors with amplified URI1 also express increased levels of the anti-apoptotic factor AKT2 (P=0.0024), the proto-oncogene NRAS (P=0.0044), the metastatic promoting matrix metalloproteinase MMP21 (P=0.0081), and the MAPK activator and mediator of cell growth MAP2K1 (P=0.0020, Figure 2C).

Figure 2.

Gene expression alteration in UCS with URI1 amplification. A. After analyzing 220 cancer related genes expression in control (n=34) and URI1 amplified tumors (n=23), heat map representation shows 34 genes with significant expression alternation. B. UCS tumors with URI1 amplification have decreased levels of PTEN (P=0.0082), CASP8 (P=0.0035), CYB5A (P=0.0022), and E2F5 (P=0.031). C. UCS tumor with URI1 amplification have increased expression of NRAS (P=0.0044), AKT2 (P=0.0024), MAP2K1 (P=0.0020), and MMP21 (P=0.0081).

Our data above showed that patients with URI1 amplification were less responsive to adjuvant therapies that induce DNA damage (Figure 1C). To elucidate the mechanism underlying such treatment-resistance, we investigated cell response to DNA damage upon URI1 upregulation. Overexpression of URI1 significantly increased cell survival after cisplatin treatment of the uterine/endometrial cancer cell line RL95-2 (Figure 3A). Cells with increased URI expressed a much higher level of phospho-ATM and phospho-ATR after cisplatin-induced DNA damage (Figure 3B). However, URI over-expressing cells also displayed dramatically lower γH2AX phosphorylation levels compared to controls cells after cisplatin treatment, suggesting less DNA-damage accumulation and therefore a URI1-mediated genome protective effect (Figure 3B). In a time course, we found that the phospho-γH2AX levels in URI-overexpressing cells reached peaks after 3 hours of cisplatin treatment and diminished quickly, while the DNA damage in control cells kept accumulating for over 24 hours (Figure S1). This suggests that although DNA damage pathways are activated, the resultant γH2AX foci are more rapidly resolved in URI1 amplified cells. Therefore, URI1 amplification promotes resistance to DNA damage. We also found that URI-induced ATM upregulation was at the level of transcription with mRNA expression increasing over 4-fold (Figure 3C). Importantly, treatment with cisplatin and ATM inhibitor together resulted in cell death in URI1 overexpressing cells compared to cisplatin alone, suggesting that URI-induced chemo-resistance is ATM-dependent and this combination could be useful for the treatment of chemo-insensitive patients with URI amplification (Figure 3D).

Figure 3.

URI1 induces ATM expression and promotes DNA damage resistance. A. RL 95-2 cells were transfected with control plasmid or URI1. After 48 hours of cisplatin treatment, cell survival was measured using the CyQUANT assay. Error bars represent standard deviation from three independent experiments. (*P<0.0001, **P<0.05). B. RL 95-2 cells were transfected with URI1 and treated with cisplatin for 48 hours. Cell lysates were subjected to western bot analysis. C. RL 95-2 cells were transfected with URI1. After 48 hours, cell mRNA was extracted and reversely transcribed to cDNA. ATM mRNA expression was quantified by q-RT PCR. (*P<0.0001). D. RL 95-2 cell were transfected with control plasmid or URI1. After 48 hours of cisplatin (5 μM) or/and ATM inhibitor (100 nM) treatment, cell survival was measured using the CyQUANT assay. (*P<0.001).

Discussion

URI was first identified as a chaperone protein and a component of the RNA polymerase complex [18]. In later studies, we and others have demonstrated its function in both cytoplasm and nucleus, regulating proliferation and apoptosis. For example, URI has proliferative and anti-apoptotic effects in hepatitis, breast and colorectal cancer cell lines, but not in immortalized but non-transformed renal and liver cells [18,26]. Studies in prostate cancer, however, revealed an anti-proliferative role of URI [31,32]. Further, a recent study reported URI1 amplification in 10% of ovarian cancer [24]. In this cohort of ovarian cancer patients, increased URI expression associated with larger tumor size, higher grade and chemo-resistance [24]. Ovarian cancer cell lines carrying URI1 copy number variations are rapamycin-resistant and more proliferative [24]. Two more recent publications showed increased mRNA expression of URI1, along with several other potential oncogenes such GAB2 and PAK4, in endometrial cancer [27,33]. The TP53 mutation rate (98%) in the cohort we analyze is much higher than a previously reported 32%, from a study in 25 UCS cases [34]. This may be due to the high percentage (53%) of stage III and IV carcinosarcoma in our cohort, compared to other reported cases (27%) [6].

Although UCS presents with undifferentiated features and a poor prognosis, risk factors from our analysis include body weight, history of hypertension, tumor stage, and local lymph node involvement, which all correlated with poor survival. Our analysis showed that URI1 amplification negatively associates with tumor-free survival after primary treatment (P=0.023) and increases patient risk of death by more than 6.5 fold. In contrast, a history of menopause hormone therapy appeared to have a dramatic protective effect in UCS overall survival by over 15-fold. Besides standard surgical treatment, UCS patients were often given adjuvant RT, or chemotherapy depending on the tumor stage. Despite the fact that UCS patients with URI1 amplification have a worse response to RT or chemotherapy compared to control patients, our analysis suggests that combined RT and chemotherapy might provide significant benefit to patient overall survival with URI1 amplification. It is interesting that progesterone or progestin use in hormone contraception is inversely associated with URI1 amplification (r _s=-0.52, P=0.011). A number of studies suggest that current use of oral contraceptives appears to increase the risk of breast cancer, and cervical cancer [35,36]. In contrast, women who use oral contraceptives have reduced risks of ovarian and endometrial cancer [37,38]. It requires further study to elucidate the potential mechanism underlying the association between progesterone use and URI1 amplification.

Our in vitro data indicated that increased URI promotes ATM expression in uterine cancer cells. As a result, cells became less responsive to cisplatin treatment. Importantly, the combination of cisplatin and ATM inhibitor reversed URI-induced DNA damage-resistance, suggesting the potential benefit of including ATM inhibitor with current chemo- and radiation-therapy in patients with URI1 amplification. Other mechanisms could also contribute to URI-induced chemo-resistance. For example, most UCS tumors carry TP53 mutation or deletion. Although DNA damage pathways are activated and upregulated, they appear to be ignored and γH2AX foci are more rapidly resolved in URI1 amplified cells. This could result in URI-dependent protection from DNA damage induced cell death.

One limitation in our study is the lack of reliable UCS models. As an alternative, we used the RL 95-2 cell line, which is derived from patients with moderately differentiated uterine carcinoma. An UCS patient derived cell line would be a better model to investigate this disease at molecular and cellular level. A further retrospective study utilizing a larger patient cohort may provide a better understanding of the impact of URI1 copy number variation in UCS. Overall, our study indicates that URI1 amplification in UCS strongly associates with a poor prognosis as a result of protection from DNA damage. The idea that URI1-amplification dependent chemo-resistance can be overcome by inhibiting ATM is a potential translational off shoot of our study.

Disclosure of conflict of interest

None.