Impaired PTPN13 Phosphatase Activity in Spontaneous Or HPV-Induced Squamous Cell Carcinomas Potentiates Oncogene Signaling Through the MAP Kinase Pathway

Oncogene (2009) 28, 3960–3970 & 2009 Macmillan Publishers Limited All rights reserved 0950-9232/09 $32.00 www.nature.com/onc ORIGINAL ARTICLE Impaired PTPN13 phosphatase activity in spontaneous or HPV-induced squamous cell carcinomas potentiates oncogene signaling through the MAP kinase pathway

AC Hoover1, GL Strand2, PN Nowicki3, ME Anderson1, PD Vermeer4, AJ Klingelhutz5, AD Bossler2, JV Pottala4, WJAJ Hendriks6 and JH Lee4,7

1Department of Otolaryngology—Head and Neck Surgery, Roy J. and Lucille A. Carver College of Medicine, University of Iowa, Iowa City, IA, USA; 2Department of Pathology, Roy J. and Lucille A. Carver College of Medicine, University of Iowa, Iowa City, IA, USA; 3Gynecologic Oncology, Roy J. and Lucille A. Carver College of Medicine, University of Iowa, Iowa City, IA, USA; 4Cancer Biology Research Center, Sanford Research/USD, Sioux Falls, SD, USA; 5Department of Microbiology, Roy J. and Lucille A. Carver College of Medicine, University of Iowa, Iowa City, IA, USA; 6Department of Cell Biology, Radboud University Nijmegen Medical Centre, Nijmegen, The Netherlands and 7Sanford ENT-Head and Neck Surgery, Sanford Health, Sioux Falls, SD, USA

Human papillomaviruses (HPVs) are a causative factor in frequently target key cellular pathways that are also over 90% of cervical and 25% of head and neck squamous altered in non-viral cancers. Because viral genes alter cell carcinomas (HNSCCs). The C terminus of the high- these pathways in a mechanistically consistent manner, risk HPV 16 E6 oncoprotein physically associates with studies of their function often serve as a starting point to and degrades a non-receptor protein tyrosine phosphatase understanding non-viral mechanisms of transformation. (PTPN13), and PTPN13 loss synergizes with H-RasV12 or In most viral cancers, synergistic cellular changes must ErbB2 for invasive growth in vivo. Oral keratinocytes that occur for malignant progression to occur. Therefore, it have lost PTPN13 and express H-RasV12 or ErbB2 show is important to study viral gene function in the context enhanced Ras/RAF/MEK/Erk signaling. In co-transfec- of these cellular changes. The following study examines tion studies, wild-type PTPN13 inhibited Ras/RAF/ a synergy between human papillomaviruses (HPV) viral MEK/Erk signaling in HEK 293 cells that overexpress oncogene function and cellular changes that lead to ErbB2, EGFR or H-RasV12, whereas an enzymatically invasion. High-risk HPVs promote cancerous growth inactive PTPN13 did not. Twenty percent of HPV- through overexpression of two multifunctional viral negative HNSCCs had PTPN13 phosphatase mutations oncoproteins, E6 and E7. Their known transforming that did not inhibit Ras/RAF/MEK/Erk signaling. functions include inactivation of pRB by E7 and Inhibition of Ras/RAF/MEK/Erk signaling using MEK degradation of p53 and activation of telomerase by E6 inhibitor U0126 blocked anchorage-independent growth in (Longworth and Laimins 2004). E6 oncoproteins from cells lacking PTPN13. These findings show that PTPN13 HPV subtypes that are high risk for malignant progres- phosphatase activity has a physiologically significant role sion also contain a C-terminal PDZ-binding motif in regulating MAP kinase signaling. (PDZBM), which has a poorly understood yet necessary Oncogene (2009) 28, 3960–3970; doi:10.1038/onc.2009.251; role in malignant transformation. PDZBMs are short published online 7 September 2009 C-terminal amino-acid sequences capable of binding PDZ domain containing proteins (Jelen et al., 2003). Keywords: human papilloma virus; HPV E6 oncogene; We have previously investigated the transforming PTPN13; MAP Kinase; ErbB2 effects of the E6 PDZBM of HPV type 16 in HPV- related head and neck squamous cell cancers (HNSCCs) (Spanos et al., 2008b) and cervical cancer (PN Nowicki Introduction et al., unpublished data) and have shown that it physically associates with and induces loss of PTPN13, Malignant transformation often occurs through ran- a non-receptor protein tyrosine phosphatase that con- dom, accumulated genetic changes resulting in char- tains five PDZ domains. In addition, HPV 16 E6 or acteristic features shared by nearly all cancers (Hanahan short hairpin RNA (shRNA) mediated PTPN13 loss V12 and Weinberg 2000). It is estimated that viral gene synergizes with H-Ras for invasive growth in vitro and expression has a role in 20% of cancers. Viral genes in vivo models of HNSCC (Spanos et al., 2008a, 2008b). Besides our data, PTPN13 has been reported as a putative tumor suppressor in a wide range of epithelial Correspondence: JH Lee, Department of Otolaryngology, Sanford cancers (including breast, colon and hepatocellular HPV Research Institute, 1310 22nd Street, Sioux Falls, SD 57105, (Wang et al., 2004; Yeh et al., 2006; Ying et al., 2006)). USA. E-mail: [email protected] Analysis of synergistic changes associated with PTPN13 Received 2 December 2008; revised 6 July 2009; accepted 11 July 2009; loss in colon cancers showed that a majority had published online 7 September 2009 mutations in the MAP kinase pathway (Wang et al., 2004). Oncogene signaling through the MAP kinase pathway AC Hoover et al 3961 Although some reports show significant association V660E expression construct. E6D146À151 is a deletion between Ras mutations and HPV in cervical cancers mutant of HPV16, which degrades p53 but lacks a (Lee et al., 1996; Landro et al., 2008), direct activating PDZBM and does not induce the loss of PTPN13 Ras mutations (like H-RasV12) are less common in (Spanos et al., 2008b). The shPTPN13 cells lack HNSCCs (Yarbrough et al., 1994; Hardisson 2003; Lu PTPN13 due to an shRNA mechanism (Spanos et al., et al., 2006). Ras pathway stimulation may alternatively 2008b). Through western blot, we confirmed that be achieved in HNSCCs by overexpression of mem- retroviral transduction with ErbB2 V660E increases brane-bound growth factor receptors, most notably the total and phoshorylated ErbB2 as compared with the ErbB family of receptor tyrosine kinases. The four parental cells (Figure 1a). Previously described PTPN13 members of this family (ErbB1–4) are commonly levels (Spanos et al., 2008b) were not altered by the overexpressed in HNSCCs and are associated with addition of ErbB2 V660E (not shown). activation of several major cancer-associated signaling To determine whether HPV 16 E6 PDZBM-mediated cascades, including signal transducers and activators degradation of PTPN13 is required for invasive growth of transcription (STAT’s), Ras/RAF/MEK/Erk (MAP in synergy with ErbB2, we injected 1 Â 106 cells from Kinase) and PI3 Kinase/AKT (Ford and Grandis each ErbB2 V660E-expressing cell line subcutaneously 2003). ErbB2 specifically is overexpressed in up to into the right hind legs of C57BL/6 mice (5 mice per 47% of HNSCCs (Cavalot et al., 2007), and when group). Weekly caliper measurements were used to combined with the expression of E6/E7 causes invasive calculate average mouse leg circumferences and estimate growth in primary oral keratinocytes, although the tumor growth rates over time (Figure 1b). All mice mechanism of HPV/ErbB2 synergy and the contribution receiving MTE HPV 16 E6/ErbB2 V660E or MTE of the E6 PDZBM were not explored (Al Moustafa shPTPN13/ErbB2 V660E cells formed tumors and met et al., 2004). criteria for killing within 50 and 30 days, respectively, Therefore, we have investigated whether the common but no mice receiving MTE HPV 16 E6D146À151/ErbB2 HNSCC oncogene, ErbB2, synergizes with HPV 16 E6- V660E cells showed signs of tumor formation for more induced PTPN13 loss to result in invasive growth than 80 days (Figure 1c). Representative mouse legs in vivo. To understand how PTPN13 loss alters cell from each group are shown at 2 weeks post-injection signaling promoting invasion, we investigated the (Figure 1d). MTE cells expressing both HPV 16 E6 and phosphorylation status of the relevant effector pathway E7, and ErbB2 V660E formed tumors at the similar signaling components in the presence or absence of rate as HPV 16 E6/ErbB2 V660E (data not shown). functional PTPN13. We describe a mechanism of Thus, oncogenic forms of ErbB2 (Figure 1) and H-Ras PTPN13’s phosphatase: the regulation MAP Kinase (Spanos et al., 2008b) synergize with HPV 16 E6 in a cascade signaling. Furthermore, we provide evidence similar PDZBM-dependent manner, suggesting that that PTPN13 loss of function may have a crucial role in growth factor receptor-mediated Ras activation may potentiating MAP Kinase cascade signaling down- be important to allow invasive growth in HPV-related stream of multiple different Ras-activating oncogenes HNSCCs. found in both HPV-positive and -negative epithelial cancers. PTPN13 loss correlates with increased Erk phosphorylation Our in vivo findings show that the expression of ErbB2 Results V660E or H-RasV12 is tumorigenic in oropharyngeal keratinocytes only if PTPN13 levels are decreased in ErbB2 synergizes with PTPN13 loss allowing invasive parallel, suggesting a potential role for PTPN13 in the growth in vivo regulation of cancer signaling pathways affected in Previously, we have shown that H-Rasv12 expression common by both ErbB2 and its downstream effector, synergizes with HPV 16 E6 or shRNA-mediated Ras. To assess what common signaling pathways are PTPN13 loss to allow invasive growth (Spanos et al., altered by the loss of PTPN13 and Ras/ErbB2 expres- 2008b). Here, we sought to determine if ErbB2, a sion, we examined phosphorylation levels of Erk and v12 common HNSCC oncogene that activates Ras, would AKT. MTE HPV 16 E6D146À151/H-Ras , MTE HPV also cause invasive growth when PTPN13 is absent. In 16 E6/H-Rasv12 and MTE shPTPN13/ H-Rasv12 cells human cancers, ErbB2 overexpression is oncogenic, but were grown to confluency, serum-starved for 24 h, and in rodent cancer models activating mutations are protein lysates were collected for immunoblot. Although required for efficient tumor formation (Moasser, we observed variation in phospho-AKT levels among 2007). We therefore cloned mouse ErbB2 cDNA into a the MTE cell lines, we found no correlation between retroviral vector and performed site-directed mutagen- phospho-AKT levels, in vivo invasive growth potential esis to introduce a transmembrane region point muta- and the presence/absence of PTPN13. However, cell tion (ErbB2 V660E) corresponding to that present in a lines with decreased PTPN13 showed increased levels constitutively active rat ErbB2 mutant (neuNT) (Moas- of phospho Erk1/2 (Figure 2a), a finding that cor- ser, 2007). Previously described mouse tonsil epithelial relates with tumor-forming ability in vivo. MTE (MTE) cell lines stably expressing HPV16 E6, HPV16 cells expressing ErbB2 V660E showed similar results E6D146À151 or shPTPN13 (Hoover et al., 2007; Spanos (Figure 2b). Immunohistochemistry of mouse tumor et al., 2008b) were then transduced with the ErbB2 samples showed increased levels of phospho-Erk

Oncogene Oncogene signaling through the MAP kinase pathway AC Hoover et al 3962

Δ E6 146-151 E6 shPTPN13 ErbB2 V660E - + - + - +

ErbB2

P-ErbB2

Actin

MTE E6Δ ErbB2 146-151 MTE E6 ErbB2 V660E V660E

MTE shPTPN13 ErbB2 V660E

Figure 1 ErbB2 synergizes with PTPN13 loss during invasive growth in vivo.(a) MTE cell lines stably expressing HPV 16 E6, E6D146À151 or PTPN13 shRNA (shPTPN13) were retrovirally transduced to express activated mouse ErbB2 (ErbB2 V660E). Functional protein expression, as compared with non-transduced cell lines, was conﬁrmed by immunoblot with antibodies to total and phosphorylated ErbB2. Actin immunostaining served as loading control. (b, c) One million cells from each line were injected subcutaneously into the right hind legs of C57BL/6 mice (n ¼ 5 per group). Mouse leg circumferences and tumor growth rates were calculated on the basis of weekly caliper measurements of leg dimensions, and mice were killed once tumors reached 20 mm in greatest dimension. The three treatment groups were compared with E6D using two-sided Dunnett adjustment for multiple comparisons developed by Hsu (Hsu 1992). The shPTPN13 and E6 groups were signiﬁcantly different than the control group (Po0.0001 and P ¼ 0.009, respectively). The median survival times in the same order of reporting as above were 18, 44 and 80 days, respectively. (d) Representative mouse legs from each group are shown at approximately 2 weeks post-injection.

compared with overlying normal epithelium. A repre- UMSCC90 cells have decreased PTPN13 levels com- sentative MTE shPTPN13 ErbB2 V660E tumor is pared with UMSCC84 cells (Spanos et al., 2008b). shown in Figure 2c. We also examined phospho-Erk Consistent with our ﬁndings in the mouse, human tonsil levels in primary human tonsil epithelial cells and epithelial cells expressing E6D146À151 and E7 showed those expressing combinations of HPV 16 E6, decreased phospho-Erk levels compared with cells v12 HPV 16 E6D146À151, E7 and H-Ras , as well as in one expressing wild-type E6 and E7. The addition of HPV 16-positive (UPCI-SCC90) and one HPV-negative H-RasV12 further increased phospho-Erk levels. More- (UMSCC 84) head and neck squamous cell carcinoma over, UPCI-SCC90 cells show increased phospho-Erk cell line. In previous work, we have shown that compared with human tonsil epithelial primary cells.

Oncogene Oncogene signaling through the MAP kinase pathway AC Hoover et al 3963 Δ E6 146-151 E6 E6E7 shPTPN13

H-RasV12 + +++ Δ Erk 1/2 - E6 146-151 E6 E6E7 shPTPN13 ErbB2 V660E - +++ +

P-Erk 1/2 ErbB2

AKT Erk 1/2

P-AKT P-Erk 1/2

GAPDH AKT

MTE shPTPN13 ErbB2 V660E P-AKT

GAPDH Erk 1/2

HTE HTE HTE E6Δ E6E7 SCC 90 SCC 84 146-151 E6E7 /E7 H-RasV12

P-Erk 1/2

P-Erk 1/2 GAPDH

Figure 2 PTPN13 loss correlates with increased Erk 1/2 phosphorylation. (a, b) Lysates of serum-starved MTE cell lines stably V12 expressing human papillomavirus (HPV) 16 E6, E6 and E7, E6D146À151 or PTPN13 shRNA (shPTPN13) as well as H-Ras or ErbB2 V660E were subjected to immunoblot analysis with indicated antibodies. (c) Immunohistochemistry for total and phosphorylated Erk 1/2 was performed on a representative tumor derived from a mouse implanted with MTE shPTPN13 ErbB2 V660E cells. Insets display control immunostainings of parallel sections to reveal antibody speciﬁcity. (d) HNSCC cell lines (SCC 90 and SCC 84) as well as V12 human tonsil epithelial (HTE) primary cells or those stably expressing HPV 16 E6, E7, E6D146À151 and H-Ras as indicated were subject to immunoblot analysis with antibody to phosphorylated Erk 1/2. Immunostaining for GAPDH was performed to monitor equal loading of protein samples.

Interestingly, the HPV-negative UMSCC84 cell line also human PTPN13, ErbB2 (wild type), EGFR (ErbB1), showed elevated phospho-Erk, although to a lesser H-RasV12 or an empty vector control. After 24 h of degree than UPCI-SCC90. These findings in human and serum starvation, protein lysates were collected and mouse cells suggest that induced loss of PTPN13 in vector expression confirmed by immunoblot. As ex- conjunction with Ras activation permits potentiation of pected, transfection with ErbB2, H-RasV12 or EGFR, as signaling through the MAP kinase pathway. compared with empty vector control conditions, increased levels of phospho-Erk, which was inhibited by PTPN13 attenuates MEK and Erk phosphorylation co-expression of PTPN13 (Figure 3). We observed a To further study the role PTPN13 has in regulating similar change in phospho-MEK, the immediate up- growth factor receptor-mediated signaling, we per- stream activator of Erk. Interestingly, Phospho-AKT formed co-transfection experiments in HEK 293 cells levels were not increased by transfection with Ras/ and examined MAP Kinase cascade signaling in ErbB2/EGFR or inhibited by PTPN13 (Figure 3). As a response to PTPN13 status, postulating that if loss previous report showed that PTPN13 may dephosphor- permits enhanced signaling then replacement of ylate ErbB2 directly (Zhu et al., 2008), we also examined PTPN13 should suppress signaling. Cells were split to levels of phospho-ErbB2 (tyr 1248), and we also equal, subconfluent densities and co-transfected as observed a small effect in the presence of wild-type indicated with mammalian expression vectors for PTPN13 (Figure 4). These findings show that under

Oncogene Oncogene signaling through the MAP kinase pathway AC Hoover et al 3964 ErbB2 - ++ PTPN13 --+ H-Ras - + + PTPN13 PTPN13 -+ -

ErbB2 PTPN13

P-ErbB2 ErbB2

AKT P-ErbB2

P-AKT AKT

Erk 1/2 P-AKT

P-Erk 1/2 Erk 1/2

MEK 1/2 P-Erk 1/2

P-MEK 1/2 GAPDH

GAPDH

EGFR - ++ PTPN13 - - +

PTPN13

EGFR

P-AKT

P-Erk 1/2

MEK 1/2

P-MEK 1/2

GAPDH

Figure 3 PTPN13 inhibits MAP Kinase cascade signaling. HEK 293 cells were transfected with mammalian expression vectors for (a) ErbB2, (b) H-RasV12 or (c) EGFR, as well as PTPN13 or an equivalent empty vector control. Two days post-transfection serum-starved cells were lysed and subjected to immunoblot analysis using indicated antibodies. GAPDH immunostaining served as loading control.

serum-starved conditions, PTPN13 consistently attenu- frequently co-occurred with mutations in the Ras/MAP ates MEK 1/2 and Erk 1/2 phosphorylation in response Kinase cascade (Wang et al., 2004). To determine to activation by various oncogenes common in cervical whether PTPN13 phosphatase activity regulates MAP and/or head and neck cancers. Kinase signaling, we performed site-directed mutagenesis to create a phosphatase dead PTPN13 construct for The phosphatase activity of PTPN13 is required to use in our co-transfection studies. An identical mutation attenuate MAP Kinase activation resulting in a phosphatase domain cysteine-to-serine Apart from its phosphatase, PTPN13 has many substitution (PTPN13 C2389S) has been shown to domains likely important for localization, binding abolish PTPN13 catalytic activity (Dromard et al., potential substrates or possibly regulatory roles 2007). We repeated the HEK 293 transfection assay (Erdmann, 2003). The functional signiﬁcance of the with expression vectors for EGFR/ErbB2 and either phosphatase domains in preventing transformation has wild-type or C2389S PTPN13 constructs. Immunoblot not been shown. In fact, PTPN13 phosphatase domain analysis shows a phosphatase domain function of null mice show only minor phenotypic defects (Wansink PTPN13 that is necessary to inhibit MAP Kinase et al., 2004; Nakahira et al., 2007). Contrary to this signaling in the presence of ErbB2 or EGFR, as transgenic mouse data, a recent screen of colorectal evidenced by increased levels of phospho-Erk and cancers suggests loss of phosphatase function that may phospho-MEK in PTPN13 C2389S conditions com- promote cancer because phosphatase domain mutations pared with wild-type PTPN13 (Figure 4).

Oncogene Oncogene signaling through the MAP kinase pathway AC Hoover et al 3965 EGFR ErbB2 Wild Wild PTPN13 C2389S PTPN13 C2389S Type Type PTPN13 PTPN13

ErbB2 EGFR P-EGFR Phospho ErbB2

Percent P-ErbB2 67.5 82.2 Percent P-EGFR 37.7 25.2

Phospho AKT Phospho AKT

Phospho Erk 1/2 Phospho Erk 1/2

MEK MEK

Phospho MEK Phospho MEK

Percent P-MEK 26.9 63.8 Percent P-MEK 24.5 74.7

GAPDH GAPDH

Figure 4 PTPN13s phosphatase activity is required to attenuate MAP Kinase signaling. HEK 293 cells were co-transfected as indicated with (a) ErbB2 or (b) EGFR and either wild-type or phosphatase dead (C2389S) PTPN13 expression constructs. Two days post-transfection serum-starved cells were lysed, subjected to immunoblot analysis using indicated antibodies and densitometry was calculated as described in Supplementary Methods. After transfection with either the wild type or C2398S, the percent phosphorylated MEK, ErbB2 and EGFR is indicated.

Mutations in HPV-negative HNSCCs do not attenuate an artificial substrate from HNSCC 132. Results MAP kinase signaling indicated that, as compared with wild-type PTPN13, The above findings support the hypothesis that loss of both mutants from HNSCC 132 behaved similar to the PTPN13 phosphatase activity is instrumental in onco- catalytically dead C2389S PTPN13 version (Supplemen- genic signaling. To examine whether HPV-negative tary Figure 1). We also repeated the transfection assay human tumors show alternative ways to abrogate in HEK 293 cells, and two of the three mutants, one PTPN13 enzymatic function, we sequenced PTPN13 from each HNSCC, did not reduce levels of phospho- exons 44–48 (encoding the PTP domain) from 10 HPV- Erk 1/2 (Figure 5b). The result of phospho-Erk 1/2 negative human HNSCC specimens. Two patients inhibition does not correspond with DifMUP phopha- showed a total of four mutations in the coding sequence tase activity because the P2408L mutant does not show (one patient had three). This percentage (20%) com- activity but is able to inhibit Erk phosphorylation. An pares well with that observed in colorectal cancers (9%) examination of phospho-Erk levels, by immunohisto- (Wang et al., 2004). The mutations were cross-matched chemistry, in the human cancers from which the with known single-nucleotide polymorphisms to rule out mutations were isolated showed focal areas of dense common polymorphisms and were aligned with the staining compared with little/none in normal tonsillar protein sequence to determine in which domains the epithelium (Figure 5c). Many of HNSCC samples mutations occurred. HNSCC 132 contained one muta- without phosphatase domain mutations also showed tion in exon 45, (6874C>T P2276S), and two mutations increased levels of phospho-Erk (data not shown), in exon 46 (7271C>T P2408L and 7316T>C L2423P). indicating that there may be multiple mechanisms HNSCC 134 contained one mutation (c.7205C>T through which MAP Kinase activity is increased in T2386I) in exon 46. The mutations occurring in exon HNSCCs. Our findings suggest that PTPN13 phospha- 46 were either within, or close to, important phosphatase domain mutations are one mechanism through tase domain functional groups (the phosphatase do- which increased MAP Kinase activity occurs in main-binding loop or WPD loop). Therefore, we chose HNSCCs. to determine if these mutations affect PTPN13 phosphatase activity and prevent MAP kinase potentiation MEK inhibition by U0126 abrogates Erk phosphorylation in conjunction with wild-type human ErbB2. We and anchorage-independent growth in tumorigenic cell performed site-directed mutagenesis to create corre- lines that have lost PTPN13 sponding mutations in our PTPN13 expression con- Impaired PTPN13 phosphatase activity in HNSCCs struct and assessed the PTP activity of immunopurified may alter signaling cascades other than the MAP kinase P2408L and L2423P mutant proteins using DifMUP as pathway. To determine the importance of MAP Kinase

Oncogene Oncogene signaling through the MAP kinase pathway AC Hoover et al 3966 HNSCC134

Wild PTPN13 - T2386I Normal Tonsil HNSCC 132 HNSCC 134 Type ErbB2 + ++

PTPN13

ErbB2 PTPN13

P-Erk 1/2

P-AKT

HNSCC 132 Erk 1/2 Wild PTPN13 - Type P2408L L2423P ErbB2 +++ +

PTPN13

ErbB2 P-Erk 1/2

P-Erk 1/2 Bar=100um P-AKT

Figure 5 A subset of human papillomavirus (HPV)-negative head and neck squamous cell carcinomas harbor PTPN13 phosphatase domain mutations. DNA sequencing of HPV negative, parafﬁn-embedded HNSCC samples revealed phosphatase domain mutations in two out of the ten patients. (a) HEK 293 cells were co-transfected with ErbB2 and PTPN13 constructs containing phosphatase domain point mutations corresponding to those found in the two HNSCC samples. Cells were serum-starved following transfection and immunoblot was performed using indicated antibodies. (b) Immunohistochemistry of these two HNSCC samples and of normal tonsil tissue was performed to assess PTPN13 expression and the levels of total and phosphorylated Erk 1/2. Insets display control immunostainings of parallel sections to reveal antibody speciﬁcity.

signaling for the tumorigenic process and assess the Discussion possible therapeutic potential of our findings, we examined the levels of phospho-Erk 1/2 in response to The conversion of a normal epithelial cell to a malignant the MEK inhibitor U0126 in two tumorigenic cell lines: one requires multiple cellular alterations (Hahn and a cell line derived from a mouse tumor generated by Weinberg, 2002). Our findings show that the loss of injection with MTE E6E7/ErbB2 V660E cells, and MTE enzymatic activity of a key phosphatase (PTPN13) shPTPN13/ErbB2 V660E cells characterized above. For synergizes with aberrant MAP Kinase signaling result- both cell lines, treatment with increasing doses of U0126 ing from hyperactivity of epithelial oncogenes, such as correlated with decreasing levels of phospho-Erk ErbB2 and H-Ras, to allow invasive growth in vivo. (Figure 6a). To determine the physiological significance Importantly, this synergy in potentiating malignant of decreased Erk phosphorylation, we examined growth is relevant both for HPV-positive as well as anchorage-independent growth (AIG) colony-forming HPV-negative tumors. efficiencies during U0126 treatment. AIG correlates Our data strongly suggest that HPV 16-mediated with invasive potential in vivo (Zhan et al., 2004; Reddig PTPN13 loss synergizes with ErbB2 activity during and Juliano, 2005), and we have previously shown a invasive growth in HPV-related head and neck cancers, correlation between PTPN13 loss and AIG in both a finding that expands on several previously published mouse and human cell lines (Spanos et al., 2008a, reports. ErbB2 and H-Ras have previously been known 2008b). In the cell lines examined, colony-forming to synergize with E6 and E7 to allow invasive growth efficiencies decreased in a dose-dependent response to (Al Moustafa et al., 2004; Schreiber et al., 2004). In U0126 that correlates with a decrease in phospho-Erk addition, ErbB2/EGFR overexpression occurs in a large (Figure 6b). These studies provide evidence that HPV proportion of cervical cancers (Perez-Regadera et al., 16-related cancers are dependent on enhanced MAP 2009), the majority of which are HPV positive. We show Kinase signaling during invasive growth, and pharma- that the HPV 16 E6 PDZBM degrades PTPN13 and this cological inhibition at the level of MEK may serve as a induced loss is needed to synergize with ErbB2. The useful therapy. site(s) where PTPN13 acts to exert this control on MAP

Oncogene Oncogene signaling through the MAP kinase pathway AC Hoover et al 3967 MTE shPTPN13 ErbB2 V660E MTEE6E7 ErbB2 V660E Tumor

DMSO 1µM 5 µM 20 µM DMSO 1µM 5 µM 20 µM

P-Erk 1/2

GAPDH

Anchorage Independent Growth Colony Forming Efficiency (%)

DMSO 1 µM U0126 5 µM U0126 20 µM U0126 MTE shPTPN13 5.4 6.7 3.5* 0** ErbB2 V660E

MTE E6E7 ErbB2 15.5 16.3 14.4 0.7** V660E Tumor

Figure 6 MEK inhibition by U0126 prevents anchorage-independent growth in tumorigenic cell lines. (a) A cell line derived from a mouse injected with MTE E6E7 ErbB2 V660E cells and MTE shPTPN13 ErbB2 V660E cells were treated for 4 h with 0, 1, 5 or 20 mM U0126 in serum-free medium. The levels of phosphorylated Erk 1/2 in the respective lysates were assessed by western blot using GAPDH immunostaining as a loading control. (b) The same cell lines were plated in triplicate, suspended in semisolid agar in 12 mm transwells, and fed growth media with indicated concentrations of U0126 for approximately 2 weeks. Average colony-forming efficiencies for each condition are indicated. Significant differences compared with dimethylsulfoxide (DMSO) controls were obtained as indicated (P-values of * ¼ 0.002 and **o0.0001). kinase signaling is still not clear. Using an independent evidence that did not correlate with phosphatase activity experimental system, Zhu et al. (2008) have shown that of a mutant and its ability to inhibit Erk phosphoryla- PTPN13 physically associates with and decreases tion. Mutation P2408L showed very little phosphatase phosphorylation of ErbB2 at phosphotyrosine 1248, activity by DifMUP assay, yet when transfected with and that PTPN13 loss enhances carcinogenic signaling ErbB2 is able to inhibit MAP kinase phosphorylation downstream of ErbB2. Our findings are largely in the same as wild type. This finding could be explained agreement; we show an effect on ErbB2-mediated by a difference in enzymatic specificity for the native signaling at the levels of MEK and Erk in response to substrate versus the artificial DifMUP assay. An PTPN13 and also show a change in ErbB2 phospho- alternative explanation is that other domains of the rylation at tyrosine 1248 that correlated with PTPN13 phosphatase are important in forming a possible status. The fact that both ErbB2 and H-RasV12 were complex to inhibit MAP Kinase signaling. The mechan- potentiated by PTPN13 loss and PTPN13 inhibited ism of inhibition and the binding partners involved will MAP kinase signaling downstream of multiple oncogenes be the focus of future studies. (ErbB2, EGFR, H-RasV12) suggests that the phospha- Although we have shown a specific synergy with tase target that inhibits MAP kinase signaling may not H-RasV12 and ErbB2 in vivo, our findings show that only be limited to ErbB2 tyrosine 1248. PTPN13 loss likewise alters signaling of EGFR, another Taken together, the above findings suggest a mechan- common epithelial oncogene. Other receptor tyrosine ism of how PTPN13 loss of function synergizes with kinases or pathways that activate MAP kinase signaling MAP Kinase activating oncogenes to promote invasive may also synergize with PTPN13 disruption during growth across both HPV-positive and -negative epithe- invasive growth. A number of MAP Kinase activating lial cancers. Our identification of non-functional oncogenes/mutations have been reported in cancer types PTPN13 mutants that allow aberrant MAP Kinase known to have decreased or mutated PTPN13. Examples signaling in HPV-negative HNSCC specimens is con- includeErbB3intheheadandneck(FordandGrandis, sistent with studies correlating PTPN13 phosphatase 2003), RAF in colorectal (Wang et al., 2004; Barault et al., domain mutations in colorectal cancers with alterations 2008) and insulin-like growth factor receptor and RAF in the Ras/MAP Kinase cascade (Wang et al., 2004). in hepatocellular carcinomas (Avila et al., 2006; Hopfner Also, mounting evidence points to PTPN13 as a et al., 2008). Although our findings provide some insight potential tumor suppressor in a broad range of other regarding the role of PTPN13 in MAP kinase signaling, epithelial cancers, including breast, cervical, gastric and many questions remain unanswered, such as what is the hepatocellular carcinomas (Wang et al., 2004; Yeh et al., target of the phosphatase and what are the roles of the 2006; Ying et al., 2006; Revillion et al., 2009). There non-phosphatase domains (Erdmann, 2003; Abaan and still some question as to which domain of PTPN13 is Toretsky, 2008). Future work will be required to better required for MAP kinase inhibition. We show some understand the complete mechanism of suppression.

Oncogene Oncogene signaling through the MAP kinase pathway AC Hoover et al 3968 From a clinical perspective these findings provide (Figure 1c). The one mortality in the E6D group was related rationale for the development of better-targeted thera- to a non-tumor-related anesthesia death. This is indicated in pies in cancers that have lost PTPN13 function. HPV- the survival graph. related cancers, in particular, offer a distinct opportu- nity for such therapies, as the viral oncogenes can be Cell lysis and immunoblot rapidly identified in clinical tumor samples and are Cell lysis and immunoblot analyses were carried out as described previously (Hoover et al., 2007). Blots were developed on Kodak associated with consistent changes in cell signaling Biomax MR film using SuperSignal West Pico and Femto pathways. It would also be possible to test which Chemiluminescent Substrates (Pierce Biotechnology, Rockford, tumors have synergizing oncogenes that potentiate IL, USA). The following antibodies were used: Anti Fap-1 MAPK signaling. We have shown that Erk phosphor- (Santa Cruz Biotechnology, Santa Cruz, CA, USA, sc-15356), ylation and AIG in cells lacking PTPN13 can be Anti GAPDH (Ambion, Austin, TX, USA, AM4300), Anti- inhibited at the level of MEK with U0126; however, phospho-AKT1/PKBa (ser 473) (Upstate, Billerica, CA, USA, U0126 has been shown to be ineffective in vivo, likely 05–669), Anti-phospho-p44/42 MAPKinase (thr202/204) (Cell due to poor solubility/bioavailability (Wang et al., Signaling, Danvers, MA, USA, #9101), Anti-phospho-erbB2 2007). Clinical trials have been undertaken for MEK (tyr 1248) (Upstate, 06–229), Anti-MapKinase Erk1/Erk2 inhibitors that show greatly improved pharmacological (Calbiochem, San Diego, CA, USA, 442704), Anti-AKT/PKB (Upstate, 05–591), Anti-phospho-MEK1/2 (ser217/221) (Cell properties in vivo compared with U0126 (Rinehart et al., Signaling, # 9154), and anti-human-c-erbB2 oncoprotein 2004; Lorusso et al., 2005; Adjei et al., 2008). These new (DakoCytomation, Carpinteria, CA, USA, A 0485). pharmacological MEK inhibitors may have utility in improving outcomes in HPV-related cancers. Transfection assay HEK 293 cells were routinely cultured in Dulbecco’s modified Eagle’s medium supplemented with 10% fetal bovine serum Materials and methods and 1% penicillin/streptomycin. Twenty-four hours before transfection, cells were split and plated at 6 Â 106 cells/10 cm Retroviral constructs and site-directed mutagenesis plate. Co-transfections were carried out as per manufacturers The following mammalian expression vectors were used in instructions using Polyfect (Qiagen, Valencia, CA, USA, 301105) transfection experiments: pQCXIX (empty vector) (Clontech, and 4 mg of each indicated plasmid. Twenty-four hours post- Mountain View, CA, USA), pcDNA3.1 (empty vector), transfection, plates were washed twice with phosphate-buffered pCMV5-PTPN13 (a gift from Ze’ev Ronai), pcDNA3.1- saline and serum-starved for 16–24 h in Dulbecco’s modified ErbB2, pBabe Puro-H-Rasv12, pBabe Puro-EGFR (Addgene Eagle’s medium before protein lysis. plasmid 11011). Methods for creating the activated mouse ErbB2 retroviral construct (pQCXIH-ErbB2 V660E) and the Anchorage-independent growth assay and MEK inhibitor PTPN13 mutant constructs are described in Supplementary (U0126) studies Methods. U0126 (Cell Signaling # 9903) was solubilized in dimethlysulf- oxide at 8 mg/ml before being added to Dulbecco’s modified Eagle’s medium at indicated concentrations. To create the Retroviral transduction and selection MTE E6E7 ErbB2 V660EA tumor cell line, tumor tissue was Mouse tonsil epithelial cell lines stably expressing HPV 16 E6/ dissected from the leg of a freshly killed male C57BL/6 mouse, E7, E6, E6D146À151, or shPTPN13 have been described previ- injected 3 weeks before with MTE E6E7 ErbB2 V660E cells. ously and were routinely cultured in E-Media (Hoover et al., Tissue was finely minced using a scalpel and subsequent 2007; Spanos et al., 2008b). Retroviral transduction with treatment with Dispase II. Cells were cultured using previously pQCXIH-ErbB2 V660E was performed using standard methods described methods (Hoover et al., 2007). Indicated amounts of (Hoover et al., 2007). Cell lines were selected in E-Media U0126 was applied to the cells in Eagle’s minimum essential containing 25–100 mg/ml Hygromycin until non-transduced medium for 48 h before lyses and subsequent immunoblots. controls had died (approximately 10 days). Following selection, Anchorage-independent growth assays were carried out by cells were maintained in E-Media with 5 mg/ml Hygromycin. applying a 100 ml base layer of 1% noble agar to 12 mm transwells with 0.4 mM polyester membrane inserts (Corning, In vivo invasive growth Lowell, MA, USA, # 3460) and allowed to solidify at 4 1C. Experiments were carried out as has been previously described Cells were suspended at 2 Â 103 cells/ml in a mixture of (Hoover et al., 2007; Spanos et al., 2008b). Examination of 0.33% noble agar and E-Media at 37 1C. A volume of 500 ml these tumors by a pathologist showed invasive growth into of this suspension was added in triplicate to transwells and muscle and capillary/lymphatic growth as was seen in our past allowed to harden 5 min at 4 1C. Transwells were added model (Spanos et al., 2008a, b). In brief, MTE cells were to sterile, 12-well tissue culture plates containing enough harvested from tissue culture plates using trypsin (0.25%), and growth media to cover the 1% agar layer (approximately resuspended at 103 cells/ml in E-Media. Using an 18-gauge 1.5 ml). Media was changed every other day and supple- needle, 100 ml of cell suspension was injected subcutaneously in mented with U0126 at indicated concentrations. Cells were the right hind leg of each C57BL/6 mouse. Leg circumference grown in suspension for 17 days. Colonies reaching diameters was estimated by weekly caliper measurements and using the of 100 mm or greater were counted. AIG percent colony- equation: circumference ¼ p(3(a þ b)À[(a þ 3b) Â (3a þ b)]1/2), forming efficiency was determined using the equation: (AIG where a is longest leg dimension, and b the shortest. colonies/cells seeded) Â 100. Analysis of variance statistical Mice were humanely killed once tumors reached 2 cm in analysis was used to calculate significance. greatest dimension, or the animal became emaciated or had functional leg impairment. An 80-day survival study of 20 mice Patient samples in four equally sized groups was analysed using the log-rank Formalin-fixed, paraffin-embedded oropharyngeal squamous test with Kaplan–Meier estimated survival functions cell carcinoma cases and tonsil tissues from tonsillectomies

Oncogene Oncogene signaling through the MAP kinase pathway AC Hoover et al 3969 were obtained from the Department of Pathology archives at Supplementary Methods. PTPN13 mutations were analysed University of Iowa Hospitals and Clinics (UIHC). This study for significance using several methods. Each mutation was was approved by the Institutional Review Board of the compared with a list of known PTPN13 single-nucleotide University of Iowa. To enrich the amount of DNA collected polymorphisms using the NCBI and Ensembl databases from tumor cells for PCR and sequencing, grossly visible (www.ncbi.nlm.nih.gov, www.ensembl.org). Sequence conser- tumor was isolated from surrounding normal tumor by vation among 10 amniotic vertebrates was analysed using macrodissection. Ensembl’s GeneSeqAlignView (www.ensembl.org). PTP amino-acid conservation was analysed using an alignment of 5 Immunohistochemistry PTPs (Villa et al., 2005). BLOSUM scores were calculated Formalin-fixed, paraffin-embedded tissues were deparaffinized using a BLOSUM matrix. Each mutation was analysed for in xylene, antigen unmasked in citrate buffer, and endogenous functional consequence using SIFT technology (blocks.fhcr- c.org). The location of each mutation within the phosphatase peroxidases quenched with 2% H2O2. Tissues were blocked with 10% bovine serum albumin. PTPN13 was detected using domain was identified using available crystal structure data sc-1138 (Santa Cruz Biotechnology), total ERK1/2 was (Villa et al., 2005). detected using 4695 and p-ERK1/2 was detected using 4370 (Cell Signaling). Incubation with primary antibody (1:125 for PTPN13, 1:1000 for total ERK1/2, 1:400 for p-ERK1/2) was Conflict of interest carried out overnight at 4 1C. Secondary antibody incubation and detection were carried out with the LSAB þ system (for The authors declare no conflict of interest. sc-1138) or EnVision þ system (for 4695, 4370) according to the manufacturer’s instructions (DakoCytomation). Slides were co-stained with hematoxylin. Blocking peptide (PTPN13, Acknowledgements ERK1/2) or secondary antibody (p-ERK) alone controls were completed to validate antibody specificity. JHL was supported by an NIDCR 1R01DE018386-01A1 and ACH was supported by an HHMI medical student fellowship. PTPN13 sequencing We acknowledge Jan Schepens’ technical expertise for Exons 44–48 of PTPN13 were amplified from genomic DNA completing the phosphatase assays and Catherine Christo- using PCR. The conditions for the PCRs are described in the pherson for manuscript preparation.

References

Abaan OD, Toretsky JA. (2008). PTPL1: a large phosphatase with a Hardisson D. (2003). Molecular pathogenesis of head and neck split personality. Cancer Metastasis Rev 27: 205–214. squamous cell carcinoma. Eur Arch Otorhinolaryngol 260: 502–508. Adjei AA, Cohen RB, Franklin W, Morris C, Wilson D, Molina JR Hoover AC, Spanos WC, Harris GF, Anderson ME, Klingelhutz AJ, et al. (2008). Phase I pharmacokinetic and pharmacodynamic study Lee JH. (2007). The role of human papillomavirus 16 E6 in of the oral, small-molecule mitogen-activated protein kinase kinase anchorage-independent and invasive growth of mouse tonsil 1/2 inhibitor AZD6244 (ARRY-142886) in patients with advanced epithelium. Arch Otolaryngol Head Neck Surg 133: 495–502. cancers. J Clin Oncol 26: 2139–2146. Hopfner M, Schuppan D, Scherubl H. (2008). Growth factor receptors Al Moustafa AE, Foulkes WD, Benlimame N, Wong A, Yen L, and related signalling pathways as targets for novel treatment Bergeron J et al. (2004). E6/E7 proteins of HPV type 16 and ErbB-2 strategies of hepatocellular cancer. World J Gastroenterol 14: 1–14. cooperate to induce neoplastic transformation of primary normal Hsu J. (1992). The factor analytic approach to simultaneous inference oral epithelial cells. Oncogene 23: 350–358. in the general linear model. J Computational and Graphical Statistics Avila MA, Berasain C, Sangro B, Prieto J. (2006). New therapies for 1: 151. hepatocellular carcinoma. Oncogene 25: 3866–3884. Jelen F, Oleksy A, Smietana K, Otlewski J. (2003). PDZ domains— Barault L, Veyrie N, Jooste V, Lecorre D, Chapusot C, Ferraz JM common players in the cell signaling. Acta Biochim Pol 50: 985–1017. et al. (2008). Mutations in the RAS-MAPK, PI(3)K (phosphatidy- Landro ME, Dalbert D, Picconi MA, Cuneo N, Gonzalez J, Vornetti S linositol-3-OH kinase) signaling network correlate with poor et al. (2008). Human papillomavirus and mutated H-ras oncogene in survival in a population-based series of colon cancers. Int J Cancer cervical carcinomas and pathological negative pelvic lymph nodes: a 122: 2255–2259. retrospective follow-up. J Med Virol 80: 694–701. Cavalot A, Martone T, Roggero N, Brondino G, Pagano M, Lee JH, Lee SK, Yang MH, Ahmed MM, Mohiuddin M, Lee EY. Cortesina G. (2007). Prognostic impact of HER-2/neu (1996). Expression and mutation of H-ras in uterine cervical cancer. expression on squamous head and neck carcinomas. Head Neck Gynecol Oncol 62: 49–54. 29: 655–664. Longworth MS, Laimins LA. (2004). Pathogenesis of human Dromard M, Bompard G, Glondu-Lassis M, Puech C, Chalbos D, papillomaviruses in differentiating epithelia. Microbiol Mol Biol Freiss G. (2007). The putative tumor suppressor gene PTPN13/ Rev 68: 362–372. PTPL1 induces apoptosis through insulin receptor substrate-1 Lorusso PM, Adjei AA, Varterasian M, Gadgeel S, Reid J, Mitchell dephosphorylation. Cancer Res 67: 6806–6813. DY et al. (2005). Phase I and pharmacodynamic study of the oral Erdmann KS. (2003). The protein tyrosine phosphatase PTP-Basophil/ MEK inhibitor CI-1040 in patients with advanced malignancies. Basophil-like. Interacting proteins and molecular functions. Eur J J Clin Oncol 23: 5281–5293. Biochem 270: 4789–4798. Lu SL, Herrington H, Reh D, Weber S, Bornstein S, Wang D et al. Ford AC, Grandis JR. (2003). Targeting epidermal growth factor (2006). Loss of transforming growth factor-beta type II receptor receptor in head and neck cancer. Head Neck 25: 67–73. promotes metastatic head-and-neck squamous cell carcinoma. Hahn WC, Weinberg RA. (2002). Rules for making human tumor Genes Dev 20: 1331–1342. cells. N Engl J Med 347: 1593–1603. Moasser MM. (2007). The oncogene HER2: its signaling and Hanahan D, Weinberg RA. (2000). The hallmarks of cancer. Cell 100: transforming functions and its role in human cancer pathogenesis. 57–70. Oncogene 26: 6469–6487.

Oncogene Oncogene signaling through the MAP kinase pathway AC Hoover et al 3970 Nakahira M, Tanaka T, Robson BE, Mizgerd JP, Grusby MJ. (2007). Villa F, Deak M, Bloomberg GB, Alessi DR, van Aalten DM. (2005). Regulation of signal transducer and activator of transcription Crystal structure of the PTPL1/FAP-1 human tyrosine phosphatase signaling by the tyrosine phosphatase PTP-BL. Immunity 26: mutated in colorectal cancer: evidence for a second phosphotyrosine 163–176. substrate recognition pocket. J Biol Chem 280: 8180–8187. Perez-Regadera J, Sanchez-Munoz A, De-la-Cruz J, Ballestin C, Lora Wang D, Boerner SA, Winkler JD, LoRusso PM. (2007). Clinical D, Garcia-Martin R et al. (2009). Negative prognostic impact of the experience of MEK inhibitors in cancer therapy. Biochim Biophys coexpression of epidermal growth factor receptor and c-erbB-2 in Acta 1773: 1248–1255. locally advanced cervical cancer. Oncology 76: 133–141. Wang Z, Shen D, Parsons DW, Bardelli A, Sager J, Szabo S et al. Reddig PJ, Juliano RL. (2005). Clinging to life: cell to matrix adhesion (2004). Mutational analysis of the tyrosine phosphatome in color- and cell survival. Cancer Metastasis Rev 24: 425–439. ectal cancers. Science 304: 1164–1166. Revillion F, Puech C, Rabenoelina F, Chalbos D, Peyrat JP, Freiss G. Wansink DG, Peters W, Schaafsma I, Sutmuller RP, Oerlemans F, (2009). Expression of the putative tumor suppressor gene PTPN13/ Adema GJ et al. (2004). Mild impairment of motor nerve repair PTPL1 is an independent prognostic marker for overall survival in in mice lacking PTP-BL tyrosine phosphatase activity. Physiol breast cancer. Int J Cancer 124: 638–643. Genomics 19: 50–60. Rinehart J, Adjei AA, Lorusso PM, Waterhouse D, Hecht JR, Natale Yarbrough WG, Shores C, Witsell DL, Weissler MC, Fidler ME, RB et al. (2004). Multicenter phase II study of the oral MEK Gilmer TM. (1994). ras mutations and expression in head and neck inhibitor, CI-1040, in patients with advanced non-small-cell lung, squamous cell carcinomas. Laryngoscope 104: 1337–1347. breast, colon, and pancreatic cancer. J Clin Oncol 22: 4456–4462. Yeh SH, Wu DC, Tsai CY, Kuo TJ, Yu WC, Chang YS et al. (2006). Schreiber K, Cannon RE, Karrison T, Beck-Engeser G, Huo D, Genetic characterization of fas-associated phosphatase-1 as a Tennant RW et al. (2004). Strong synergy between mutant ras and putative tumor suppressor gene on chromosome 4q21.3 in HPV16 E6/E7 in the development of primary tumors. Oncogene 23: hepatocellular carcinoma. Clin Cancer Res 12: 1097–1108. 3972–3979. Ying J, Li H, Cui Y, Wong AH, Langford C, Tao Q. (2006). Spanos WC, Geiger J, Anderson ME, Harris GF, Bossler AD, Smith Epigenetic disruption of two proapoptotic genes MAPK10/JNK3 RB et al. (2008a). Deletion of the PDZ motif of HPV16 E6 and PTPN13/FAP-1 in multiple lymphomas and carcinomas preventing immortalization and anchorage-independent growth in through hypermethylation of a common bidirectional promoter. human tonsil epithelial cells. Head Neck 30: 139–147. Leukemia 20: 1173–1175. Spanos WC, Hoover A, Harris GF, Wu S, Strand GL, Anderson ME Zhan M, Zhao H, Han ZC. (2004). Signalling mechanisms of anoikis. et al. (2008b). The PDZ binding motif of human papillomavirus Histol Histopathol 19: 973–983. type 16 E6 induces PTPN13 loss, which allows anchorage- Zhu JH, Chen R, Yi W, Cantin GT, Fearns C, Yang Y et al. (2008). independent growth and synergizes with ras for invasive growth. Protein tyrosine phosphatase PTPN13 negatively regulates Her2/ J Virol 82: 2493–2500. ErbB2 malignant signaling. Oncogene 27: 2525–2531.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc)

Oncogene