Somatic Mutations of PPP2R1A in Ovarian and Uterine Carcinomas

Abstract

Exome sequencing of ovarian clear-cell carcinoma has identified somatic mutations in PPP2R1A, a subunit of protein phosphatase 2A. The present study was performed to determine the frequency of PPP2R1A mutations in exon 5, which harbors previously reported mutation hot spots, and adjacent exon 6, in 209 ovarian and 56 uterine tumors of various histologic subtypes. PPP2R1A mutations were demonstrated in 10 of 110 type I ovarian tumors (9.1%) including low-grade serous, low-grade endometrioid, clear-cell, and mucinous carcinomas. In contrast, none of 71 type II ovarian (high-grade serous) carcinomas exhibited PPP2R1A mutations. Moreover, PPP2R1A mutations were observed in 2 of 30 type I uterine (endometrioid) carcinomas (6.7%) and 5 of 26 type II uterine (serous) carcinomas (19.2%). Of the 18 mutations, 13 affected the R182 or 183, and there were 5 novel mutations including 3 involving S256, 1 involving W257, and 1 involving P179. All mutations were located in the α-helix repeats near the interface between the A subunit and the regulatory B subunit of the enzyme complex. These data provide new evidence that PPP2R1A somatic mutations occur in certain types of uterine and ovarian neoplastic lesions, especially uterine serous carcinomas, and suggest that mutation of PPP2R1A may participate in the pathogenesis of ovarian type I and uterine type II carcinomas.

A recent genome-wide sequencing analysis of all exons from ovarian clear cell carcinomas led to the discovery of somatic missense mutations in PPP2R1A in approximately 7% of these tumors.¹ PPP2R1A encodes a constant regulatory subunit of the protein phosphatase 2A holoenzyme, which is one of four major serine-threonine phosphatases.² Protein phosphatase 2A is composed primarily of a catalytic C subunit and a scaffolding subunit A (PPP2R1A and PPP2R1B) and a regulatory subunit B. Protein phosphatase 2A can be regulated by binding of at least 18 different regulatory B subunits to the core enzyme. The regulatory B subunit has been implicated in controlling the substrate specificity, cellular localization and enzymatic activity of protein phosphatase 2A, which participates in negative regulation of cellular proliferation among several other cellular functions.^2,3 Moreover, the regulatory subunit competes with virus-producing proteins including the small T antigens of the papovaviruses SV40 and polyoma, and the middle T antigen of polyoma,⁴ which suggests that protein phosphatase 2A has a role in the pathogenesis of virus-associated tumors.

Uterine carcinomas can be classified as type I or type II, and develop along unique molecular pathways. Type I uterine tumors are composed of endometrioid carcinomas, which frequently harbor mutations in CTNNB1, PTEN, ARID1A (BAF250A), and PIK3CA, whereas type II tumors are composed of serous carcinomas, which contain TP53 mutations in most cases.⁵ Patients with type I tumors are typically younger, and carcinoma develops at an earlier stage and has a more indolent clinical course than in patients with type II tumors. Type I uterine carcinomas arise from endometrial hyperplasia, whereas type II uterine carcinomas develop from endometrial intraepithelial carcinoma. Compared with uterine carcinoma, ovarian carcinoma is far more complex and includes more histologic subtypes than uterine tumors do.⁶ To provide a conceptual framework for the study of the pathogenesis of ovarian cancer, a dualistic model was proposed that organizes the clinical, pathologic, and molecular features of ovarian cancer by categorizing them into type I and type II groups, similar to the classification of uterine carcinoma.⁷ Type I tumors include low-grade serous, low-grade endometrioid, clear-cell, and mucinous carcinomas, which develop in a stepwise fashion from well-established precursor lesions such as borderline tumors and endometriosis. Type I tumors typically are manifested as a large mass confined to one ovary, and are associated with a relatively good prognosis. They typically exhibit mutations in KRAS, PIK3CA, PTEN, and CTNNB1, and rarely in TP53.^6,8 In contrast, type II tumors include high-grade serous carcinomas, malignant mixed mesodermal tumors (carcinosarcomas), and undifferentiated carcinomas. Type II tumors constitute most ovarian cancers and account for most deaths. They are detected at an advanced stage (stages II to IV) in greater than 75% of cases, grow rapidly, and are highly aggressive. Type II tumors, of which high-grade serous carcinoma is the prototypic type, are chromosomally highly unstable,^9,10 and harbor TP53 mutations in most cases but rarely demonstrate mutations in KRAS, PIK3CA, PTEN, and CTNNB1.^11,12

The present study sought to extend the previous observation of somatic mutations of PPP2R1A in ovarian clear-cell carcinomas¹ by examining the mutational profiles of a variety of histologic types of ovarian and uterine carcinomas that have been broadly classified as type I and type II tumors.^5,7 Sequence analysis was performed in 256 ovarian and uterine tumors, focusing on exon 5, which harbors the codons that encode amino acids 182 and 183, which, according to the previous study, represent mutational “hot spots” in PPP2R1A.¹ Exon 6 was also analyzed in all tumors.

Materials and Methods

Tissue Specimens

Tissue specimens from 265 ovarian and uterine tumors were analyzed for mutations at exons 5 and 6 including the codons 182 and 183 of PPP2R1A.¹ Type I ovarian tumors included 44 clear-cell carcinomas, 6 mucinous tumors, 20 low-grade serous carcinomas, and 40 low-grade endometrioid carcinomas. Type I uterine carcinoma included 30 low-grade (grade 1) endometrioid carcinomas. Type II ovarian tumors included 71 high-grade serous carcinomas. Type II uterine carcinomas included 26 serous carcinomas, 3 of which contained clear-cell carcinoma components. Those uterine carcinomas with mixed serous and clear-cell histologic features were classified as serous carcinoma. The diagnosis of uterine serous carcinomas was supported at immunohistochemistry, which demonstrated diffuse p53 staining and a high (>50%) Ki-67 labeling index. One hundred sixty-seven specimens were affinity-purified using BerEP4 antibody-conjugated magnetic beads (Invitrogen Corp., Carlsbad, CA) following a previously described protocol¹³. The remaining samples were obtained from FFPE tissues after microdissection to enrich tumor cells. The ovarian clear-cell carcinomas analyzed represented new cases that had not been studied in the previous report.¹ In addition, 14 serous, 13 mucinous, and 1 seromucinous ovarian borderline tumors were analyzed. Specimens were obtained from The Johns Hopkins Hospital or the National Taiwan University Hospital. H&E-stained sections were reviewed to confirm the diagnosis before performing the experiments. All human tissue was collected anonymously using protocols approved by the institutional review boards of both hospitals.

DNA Extraction and Mutation Analysis

Genomic DNA was isolated using a DNeasy Blood and Tissue Kit (Qiagen, Inc., Valencia, CA) for BerEP4 bead–purified tumor cells, and a QIAamp DNA Micro Kit (Qiagen, Inc.) for paraffin-embedded tissues. To enrich tumor cells from paraffin sections, pipette tips were used to individually scrape off the normal tissues, followed by tumor tissues, which were collected in separate microfuge tubes. For tumor tissues, only regions containing at least 60% of tumor cells were dissected. The selection of normal and tumor areas was made by two pathologists (I.M.S. and E.K.). PCR amplification was performed using the following primer pairs, which were designed to amplify the genomic DNA fragment of exon 5 flanking the arginine couplet at codons 182 and183: forward primer 5′-TACTTCCGGAACCTGTGCTC-3′ and reverse primer 5′-CCAGGAAGCAAAACTCACCT-3′. The PCR primers used to amplify exon 6 were: forward primer 5′-GTTCCTGCCCATGAAAGAGA-3′, reverse primer 5′-TTATTGCTCAAACGCCCAAT-3′, and sequencing primer 5′-AATGGTTCCATCGGCCTAAT-3′.

PCR conditions were as follows: 94°C for 2 minutes; three cycles at 94°C for 15 seconds, 64°C for 30 seconds, and 70°C for 30 seconds; three cycles at 94°C for 15 seconds, 61°C for 30 seconds, and 70°C for 30 seconds; three cycles at 94°C for 15 seconds, 58°C for 30 seconds, and 70°C for 30 seconds; and 30 cycles at 94°C for 15 seconds, 57°C for 30 seconds, and 70°C for 30 seconds, followed by 70°C for 5 minutes. Sanger DNA sequencing was performed by Beckman Coulter Genomics (Danvers, MA) using the sequencing primer 5′-CAAAACTCACCTGCTCGTCA-3′. Mutational analysis was performed using a software package (Mutation Surveyor; SoftGenetics LLC, State College, PA). To confirm somatic mutations, DNA from matched normal tissue was sequenced using the same method. All of the detected mutations were repeated at least twice to confirm the results and rule out potential PCR errors.

Results

Of 265 tumor tissues analyzed, PPP2R1A mutations at either exon 5 or 6 were detected in 18 tumors (Table 1). Their associated normal tissues contained wild-type sequences, confirming that the PPP2R1A mutations were somatic. The tumors with PPP2R1A mutations were of various subtypes of ovarian or uterine tumors except ovarian high-grade and low-grade serous carcinoma, in which no somatic mutations were found. Specifically, in ovarian carcinomas, PPP2R1A mutations were found in 4 of 44 ovarian clear-cell carcinomas (9.1%), 2 of 6 ovarian mucinous carcinomas (33.3%), and 4 of 40 ovarian low-grade endometrioid carcinomas (10%). In addition, PPP2R1A mutation was also recorded in 1 of 28 ovarian borderline tumors (3.6%). The mutation was observed in the seromucinous borderline tumor at codon 183 but not in the other borderline tumors. In uterine carcinomas, mutations were detected in 5 of 26 serous carcinomas (19.2%) (including those cases with mixed clear-cell and serous carcinoma), of which mutation frequency was greater than 6.7% in the uterine endometrioid carcinomas. Among the mutations, 1 uterine carcinoma demonstrated mixed clear-cell and serous components. Thus, 18 PPP2R1A mutations were identified. Of these, 13 mutations affected the R182 and R183 hot spots, which were identified in a previous report,¹ and there were 5 novel mutations including 3 involving S256, 1 involving W257, and 1 involving P179 (Table 2). All PPP2R1A mutations identified were missense mutations and were heterozygous. Representative chromatograms of somatic alterations from an ovarian low-grade endometrioid carcinoma (6026T) and a uterine serous carcinoma (319T) are shown in Figure 1.

Table 1.

Summary of PPP2R1A Mutation in Uterine and Ovarian Tumors

Tumor	Tumor type	No. of cases	Cases with mutation, no. (%)
Uterine cancer
Endometrioid	I	30	2 (6.7)
Serous⁎	II	26	5 (19.2)
Ovarian cancer
Low-grade serous	I	20	0 (0)
Clear cell	I	44	4 (9.1)
Low-grade endometriod	I	40	4 (10)
Mucinous	I	6	2 (33.3)
High-grade serous	II	71	0
Ovarian borderline tumor
Serous		14	0
Mucinous		13	0
Seromucinous		1	1 (100)

^⁎Three uterine carcinomas exhibited mixed features of clear-cell and serous carcinoma.

Table 2.

Mutation Profiles of PPP2R1A, KRAS, and PIK3CA

Sample	Histologic subtype	Gene
		PPP2R1A	K-RAS	PIK3CA
TW-7	Ovarian clear cell	R183W (547 C>T)	No mutation	No mutation
TW-11	Ovarian clear cell	R183W (547 C>T)	No mutation	No mutation
TW-24	Ovarian clear cell	R183Q (548 G>A)	No mutation	No mutation
TW-28	Ovarian clear cell	R183W (547 C>T)	No mutation	No mutation
OV81†	Ovarian clear cell	R183W (547 C>T)	No mutation	No mutation
109T†	Ovarian clear cell	R182W (544 C>T)	No mutation	E545K (1633 G>A)
192TCS†	Ovarian clear cell	R183G (547 C>G)	G12D (35 G>A)	No mutation
KK†	Clear-cell carcinoma cell line	R183Q (548 G>A)	No mutation	E545A (1634 A>C)
OVISE†	Clear-cell carcinoma cell line	R183W (547 C>T)	No mutation	No mutation
OVTOKO†	Clear-cell carcinoma cell line	R183G (547 C>G)	No mutation	No mutation
KT-41	Mucinous	R183Q (548 G>A)	G12D (35 G>A homo)‡	No mutation
KT-44	Mucinous	R183Q (548 G>A)	G12D (35 G>A)	No mutation
KT-36	Ovarian endometrioid	R183W (547 C>T)	No mutation	E545A (1634 A>C)
TS	Ovarian endometrioid	R183W (547 C>T)	No mutation	M1043T (3128T>C)
OS	Ovarian endometrioid	R182W (544 C>T)	No mutation	N1044K (3132T>A)
EM-17T	Uterine endometrioid	R183W (547 C>T)	No mutation	No mutation
EM-21T	Uterine endometrioid	R182W (544 C>T)	No mutation	H1047R (3140 A>G)
314TCS	Ovarian seromucinous borderline	R183W (547 C>T)	G12D (35 G>A)	H1047R (3140 A>G)
6026TCS	Ovarian endometrioid	R183Q (548 G>A)	No mutation	No mutation
702TCS	Uterine serous	S256Y (767 C>A)	No mutation	No mutation
319TCS	Uterine serous	S256F (767 C>T)	No mutation	H1047R (3140 A>G)
UPSC_5§	Uterine serous	P179R (536 C>G)	NA	NA
UCC_1T§	Uterine serous/clear cell	W257G (769 T>G)	NA	NA
UCC_3T§	Uterine serous	S256F (767 C>T)	NA	NA
PA10X¶	Pancreatic carcinoma	R183W (547 C>T)	G12R (34 G>C)	No mutation

^†PPP2R1A mutation identified in a previous report.¹

^‡Unless otherwise indicated, all mutations are heterozygous.

^§Genomic DNA isolated from paraffin-embedded tissues.

^¶PPP2R1A mutation identified in a previous report.¹⁴

Figure 1.

Examples of somatic mutations in PPP2R1A. Chromatogram of the sequences demonstrates a somatic mutation (548G>GA, 183R>RQ) in an ovarian endometrioid carcinoma, 6026T, and a novel mutation (767C>TC, 256S>SF) in a uterine serous carcinoma, 319T. The matched normal tissues (6026N and 319N) do not show the mutations. Both tumor tissues were affinity-purified. Arrows indicate the nucleotides with sequence mutation.

In addition, mutation profiles of PPP2R1A, KRAS, and PIK3CA were determined. This analysis included cases with PPP2R1A mutations that had been previously reported (three clear-cell carcinoma tissue samples, three clear-cell carcinoma cell lines, and one pancreatic cancer tissue),^1,14 to increase the number of cases. The results for mutation status of KRAS, PIK3CA, and PPP2R1A are given in Table 2. In contrast to the mutually exclusive pattern of KRAS and BRAF mutations previously reported,^15,16 a mutually exclusive mutation pattern between PPP2R1A and KRAS or between PPP2R1A and PIK3CA was not observed. Of 25 tumors with PPP2R1A mutations, 5 tumors harbored concurrent KRAS mutations and 8 tumors harbored concurrent PIK3CA mutations (Table 2). Sequencing PIK3CA and KRAS in 54 ovarian and uterine carcinomas with wild-type PPP2R1A demonstrated a mutation frequency in PIK3CA and KRAS similar to that in tumors that had mutations in PPP2R1A. Of those PPP2R1A wild-type cases, 19 (35%) demonstrated PIK3CA mutations and 5 (9%) demonstrated KRAS mutations. There was no significant correlation of PPP2R1A mutation status with either KRAS or PIK3CA mutations (P > 0.2, Fisher exact test).

Discussion

The present study, by using an independent set of tumor samples, not only verified that somatic PPP2R1A mutations occur in nearly 10% of ovarian clear-cell carcinomas¹ but also demonstrated that PPP2R1A mutations were detected in other types of ovarian and uterine carcinomas except ovarian serous carcinoma, the conventional type of ovarian cancer. Also identified were novel somatic mutations located outside the previously reported hot-spot region of residues of R182 and R183. The results of this study should have several implications in understanding the molecular pathogenesis of both ovarian and uterine carcinomas. It also provides a biological foundation for studying the roles of PPP2R1A mutations in the development of uterine serous carcinoma.

The presence of PPP2R1A mutations in type I but not type II ovarian carcinomas is of great interest and provides further molecular genetic evidence to support the dualistic model of ovarian carcinogenesis. Although the overall frequency of PPP2R1A mutations in ovarian clear-cell, endometrioid, and mucinous carcinomas was only 9.1%, this observation suggests that PPP2R1A mutations most likely contribute a functional effect in development of some type I ovarian tumors, whereas type II ovarian carcinomas such as high-grade serous carcinomas may use PPP2R1A-indepent pathways for their development. Furthermore, the data suggest that PPP2R1A mutations may be related to those tumors arising either from endometrium or endometriosis. This is because uterine endometrioid and serous carcinomas develop from endometrium and ovarian clear-cell and endometrioid carcinomas, and seromucinous borderline tumors are closely associated with endometriosis.^17–20 Thus, PPP2R1A mutations were observed in15 of 140 tumors (10.7%) related to endometrium or endometriosis, and the frequency was significantly greater than 2.4% in tumors not related to endometrium and endometriosis (P = 0.0072, Fisher exact test). The present study evaluated only gynecologic malignant lesions; however, other tumor types may also harbor PPP2R1A mutations. For example, a previous genome-wide sequencing analysis of pancreatic cancer identified a point mutation at the R183 residue of PPP2R1A in 1 of 24 tumors.¹⁴ Further studies analyzing PPP2R1A mutations in other types of cancer in humans are necessary to determine whether such mutations are generally enriched in tumors related to endometrium and endometriosis.

That PPP2R1A mutations occurred only in uterine serous carcinomas (uterine type II) but not in ovarian high-grade serous carcinomas (ovarian type II) suggests that although both tumor types share several clinicopathologic features, the pathogenesis is different in uterine serous carcinomas and ovarian high-grade serous carcinomas. The relatively high frequency of PPP2R1A mutations (19.2%) in uterine serous carcinoma warrants further investigation of its biological role in the development of this highly aggressive uterine carcinoma. In addition to TP53 mutations, PPP2R1A mutation is the most common sequence mutation in uterine serous carcinoma. All mutations in uterine serous carcinomas were located outside the R182 and R183, which raises the possibility that distinct mechanisms may be involved in generating the mutations at different locations.

Because the protein phosphatase 2A is implicated in regulation of signaling pathways including MAPK and AKT,^21,22 whether mutations of PPP2R1A coexisted with KRAS or PIK3CA mutations was analyzed because genes that regulate the MAPK and AKT pathways are frequently mutated in ovarian and uterine type I tumors.^5,6 From the perspective of cancer genetics, the lack of a mutually exclusive pattern between PPP2R1A and KRAS mutations or between PPP2R1A and PIK3CA mutations suggests that PPP2R1A mutations likely do not participate in signaling pathways involving KRAS and PI3K. However, future cell biology studies are needed to delineate whether there is cross-talk between protein phosphatase 2A pathway and the MAPK or AKT pathway.

A previous study¹ of ovarian clear cell carcinomas demonstrated that all of the PPP2R1A mutations were located in the pairs of arginine residues at R182 and R183.¹ However, analysis of the 18 PPP2R1A mutations in the present study identified new mutation positions at P179, S256, and W257. All mutations were located in the codons that are evolutionary conserved, and structural biology studies have demonstrated that the amino acids P179, R182, and R183 are located in the α-helix repeat 5, and amino acids S256 and W257 in the α-helix repeat 7. Both repeats are near the interface between the A subunit and the regulatory B subunit.²³ Mutations in several residues in the interface of A and B subunits such as P179, M180, and W257 have been reported to disrupt the interaction between these two subunits.²⁴ Because the mutations observed in this study were all at or close to those positions, it can be speculated that missense mutations at those residues could alter the binding affinity and/or specificity between A and B subunits and, thus, affect substrate recognition and/or phosphatase activity. All mutations identified in the previous and present studies were heterozygous missense mutations, which suggests a dominant attribute of PPP2R1A mutations in cancer pathogenesis. The mutated subunit A proteins encoded by mutant PPP2R1A may compete with their wild-type counterparts in binding to subunit B and forming a stable and functional enzyme complex. As a result, the enzymatic activity of protein phosphatase 2A, which may be important for tumor suppression, is reduced in the presence of mutant PPP2R1A proteins. This is the preferred interpretation, although there are other possibilities. For example, PPP2R1A may have a tumor-suppressor function, and the mutant proteins may function in a dominant negative manner. Additional studies should be undertaken to clarify the biological role of mutant PPP2R1A proteins in cancer development.

In summary, the present study provides new evidence of PPP2R1A mutations in several histologic subtypes of ovarian and uterine neoplasms in addition to ovarian clear-cell carcinoma. The relatively high frequency of PPP2R1A mutations in uterine (type II) serous carcinomas suggests that the alterations in the protein phosphatase 2A pathway may participate in its pathogenesis, which is different from ovarian (type II) high-grade serous carcinoma. The observation that all PPP2R1A mutations involve the α-helix repeats between the interface of subunits A and B strongly suggests that binding of both subunits is critical in cancer development. Considered together, the results indicate that PPP2R1A is a new cancer-associated gene that deserves further functional exploration to understand the roles of mutations in this gene in regulating protein phosphatase 2A functions and in tumor development.