Human Pathology (2014) 45, 276–284
www.elsevier.com/locate/humpath
Original contribution
Activating transcription factor 2 in mesenchymal tumors☆,☆☆ Makoto Endo MD, PhD a,b,c,d , Le Su PhD e , Torsten O. Nielsen MD, PhD a,b,⁎ a
Genetic Pathology Evaluation Centre, University of British Columbia, Vancouver, British Columbia, Canada Department of Pathology and Laboratory Medicine, University of British Columbia, Vancouver, British Columbia, Canada c Department of Anatomic Pathology, Kyushu University, Fukuoka, Japan d Department of Orthopaedic Surgery, Kyushu University, Fukuoka, Japan e Biomedical Research Centre, University of British Columbia, Vancouver, British Columbia, Canada b
Received 10 May 2013; revised 27 August 2013; accepted 11 September 2013
Keywords: Activating transcription factor 2; Transcription factor; Mesenchymal tumor; Translocation-associated sarcoma; Immunohistochemistry
Summary Activating transcription factor 2 (ATF2) is a member of activator protein 1 superfamily, which can heterodimerize with other transcription factors regulating cell differentiation and survival. ATF2 assembles into a complex with the synovial sarcoma translocation, chromosome 18 (SS18)– synovial sarcoma, X breakpoint (SSX) fusion oncoprotein, and the transducin-like enhancer of split 1 (TLE1) corepressor, driving oncogenesis in synovial sarcoma. The fusion oncoproteins in many other translocation-associated sarcomas incorporate transcription factors from the ATF/cAMP response element binding or E26 families, which potentially form heterodimers with ATF2 to regulate transcription. ATF2 may therefore play an important role in the oncogenesis of many mesenchymal tumors, but as yet, little is known about its protein expression in patient specimens. Herein we perform immunohistochemical analyses using a validated specific antibody for ATF2 expression and intracellular localization on a cohort of 594 malignant and 207 benign mesenchymal tumors representing 47 diagnostic entities. Melanoma served as a positive control for nuclear and cytoplasmic staining. High nuclear ATF2 expression was mainly observed in translocation-associated and/or spindle cell sarcomas including synovial sarcoma, desmoplastic small round cell tumor, endometrial stromal sarcoma, gastrointestinal stromal tumor, malignant peripheral nerve sheath tumor, and solitary fibrous tumor. Cytoplasmic ATF2 expression was less frequently seen than nuclear expression in malignant mesenchymal tumors. Benign mesenchymal tumors mostly showed much lower nuclear and cytoplasmic ATF2 expression. © 2014 Elsevier Inc. All rights reserved.
☆
Disclosure/Conflict of interest: The authors declare no conflict of interest. This work was supported by grants from the Canadian Cancer Society Research Institute (grant no. 701582), an Extension Grant from the Canadian Cancer Society Research Institute (grant no. 2012-701558), and the Liddy Shriver Sarcoma Initiative. M.E. is a Japan Society for the Promotion of Science research fellow. ⁎ Corresponding author. Genetic Pathology Evaluation Centre, University of British Columbia, Room 509-2660 Oak St, Vancouver, British Columbia, Canada V6H 3Z6. E-mail address: [email protected] (T. O. Nielsen). ☆☆
0046-8177/$ – see front matter © 2014 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.humpath.2013.09.003
ATF2 in mesenchymal tumors
277
1. Introduction
2. Materials and methods
A wide variety of transcription factors have been demonstrated to have oncogenic and/or tumor suppressor activities in human malignancies [1]. Most genes located at the breakpoint of translocation-associated sarcomas (TASs) retain DNA-binding domains from transcription factors such as DNA-damage-inducible transcript 3, Friend leukemia integration 1 transcription factor, v-ets avian erythroblastosis virus E26 oncogene homolog, ETS translocation variant 1, and activating transcription factor 1 (ATF1), implicating these transcription factors in sarcoma tumorigenesis [2]. In addition, in synovial sarcoma, the synovial sarcoma translocation, chromosome 18 (SS18)–synovial sarcoma, X breakpoint (SSX) fusion oncoprotein (which has no DNA-binding domain) forms a complex with ATF2 and transducin-like enhancer of split 1 (TLE1), allowing the SS18-SSX fusion oncoprotein to bind DNA through ATF2 and regulate the downstream genes expression related to its tumorigenesis via TLE1's interactions with chromatin-modifying enzymes [3]. ATF2 is a member of the ATF/cAMP response element binding (CREB) family, comprising ATF1 (involved in Ewing sarcoma breakpoint region 1 [EWSR1]-ATF1 fusions in clear cell sarcoma and EWSR1-ATF1 or fused in sarcoma [FUS]ATF1 fusions in angiomatoid fibrous histiocytoma), cAMP response element binding protein 3-like 1 (CREB3L1) and CREB3L2 (involved in FUS-CREB3L1 or FUS-CREB3L2 fusions in low-grade fibromyxoid sarcoma), and other ATF or CREB proteins [2,4]. ATF2 contains 2 functional domains: an N-terminal transactivation domain and a C-terminal DNAbinding domain [5]. In the above fusions, the activating domain is lost, replaced by EWSR1 or FUS. The retained DNA-binding domain contains the basic leucine zipper motif, which enables ATF2 to dimerize and to bind to cAMP response element (CRE) sequences in the promoters of target genes [6,7]. Recently, ATF2 has been found to exhibit oncogenic and tumor suppressor activities, depending on tumor type and intracellular localization [1,5,7–9]. In the nucleus, ATF2 plays an important role in transcription, whereas in the cytoplasm, ATF2 impairs mitochondrial membrane potential, resulting in mitochondrial-based cell death [1,5,9]. The implication that ATF2 is oncogenic in the nucleus but tumor suppressive in the cytoplasm is supported by immunohistochemical findings that weaker cytoplasmic and stronger nuclear staining correlates with poor prognosis in melanomas [8]. ATF2 expression and localization appear to play an important role in resistance of melanoma cells to chemotherapeutic drugs and radiation [10–12]. Although these recent findings support the biological and clinical importance of ATF2 in human malignancies, neither the intracellular localization nor the expression level of ATF2 protein has yet been well characterized in mesenchymal tumors. Here, we performed a large-scale ATF2 immunohistochemical expression study using tissue microarrays (TMAs), including synovial sarcoma together with a wide variety of human mesenchymal tumor specimens.
2.1. Tumor samples and TMAs In this study, both previously described and more recently constructed TMAs were used. Each TMA contains 0.6-mm cores derived from representative viable diagnostic areas identified by a subspecialty pathologist (T.O.N.). Published TMAs used in this study are TMA 01-003 [13] (synovial sarcoma differential diagnosis array, 82 cases in duplicate), TMA 03-008 [14] (chondroid tumor array, 121 cases in duplicate), TMA 06-007 [15] (liposarcoma array, 69 cases in triplicate), TMA 06-001B [16] (gastrointestinal stromal tumor array, 150 cases in duplicate), TMA 08-019 [17] (endometrial stromal sarcoma array, 59 cases in duplicate), TMA 09-006 [17] (epithelioid sarcoma differential diagnosis array, 53 cases in duplicate), TMA 10-009 [17] (alveolar soft part sarcoma–alveolar rhabdomyosarcoma–desmoplastic small round cell tumor array, 12 cases in triplicate), and TMA malignant peripheral nerve sheath tumor [18] (malignant peripheral nerve sheath tumor differential diagnosis array, 176 cases in duplicate). Five additional TMAs not previously published were used in this study, assembled using a manual tissue microarrayer (Beecher Instruments, Inc, Silver Springs, MD) from formalinfixed, paraffin-embedded tissue as described previously [19]. TMA 09-008 (pigmented villonodular synovitis array) contains singular or triplicate cores of 55 cases of pigmented villonodular synovitis. TMA 12-004 (B-cell lymphoma 2, BCL2-positive tumor array) contains triplicate cores of 10 spindle cell lipomas, 8 cases of dermatofibrosarcoma protuberans, 10 solitary fibrous tumors, and 7 Ewing sarcomas. TMA 12-005 (pediatric mesenchymal tumor array) contains duplicate cores of 4 hepatoblastomas, 30 fibromas/fibromatoses (including 12 lipofibromatoses, 10 superficial or digital fibromatoses, 3 cases of fibromatosis colli, 2 calcifying aponeurotic fibromas, 2 Gardner fibromas, 1 irritation fibroma, and 2 fibromas not otherwise specified), 23 Wilms tumors, 9 Ewing sarcomas, 8 alveolar rhabdomyosarcomas, 7 synovial sarcomas, 1 nodular fasciitis, 6 myofibromas, 2 myositis ossificans, 9 cases of fibrous hamartoma of infancy, 8 desmoid-type fibromatoses, 8 fibrosarcomas, 12 embryonal rhabdomyosarcomas, 2 congenital mesoblastic nephromas, 1 gastrointestinal stromal tumor, and 2 fetal rhabdomyomas. TMA12-006 (TAS array) includes duplicate cores of 2 myxoid liposarcomas, 4 synovial sarcomas, 1 malignant peripheral nerve sheath tumor, 1 extraskeletal myxoid chondrosarcoma, 1 endometrial stromal sarcoma, and 1 clear cell sarcoma. TMA12-010 (pleomorphic sarcoma array) contains duplicate cores of 5 undifferentiated pleomorphic sarcomas and 5 dedifferentiated liposarcomas. TMA 06-003 (multitumor array, 76 cases) including a single core of carcinoids, a variety of carcinomas, malignant lymphomas, melanomas, mesotheliomas, and normal tissues was used for context with the melanomas considered as a
278 positive control for staining and interpretation. Normal tissues on this array included 9 examples of fat, 2 of colon, and 1 each of adrenal gland, cerebellum, fallopian tube, heart, liver, lung, and squamous mucosa.
2.2. Western blotting and RNA interference Antibody specificity was tested by Western blotting using human synovial sarcoma cell line SYO-1 [20] (kindly provided by Dr Akira Kawai, National Cancer Center Hospital, Tokyo) with or without RNA interference for ATF2. The small interfering RNA specific for ATF2 was purchased from Dharmacon (Lafayette, CO) and had a knockdown efficiency of 60.3% by quantitative reverse transcription polymerase chain reaction assessment. At 60% confluence, cells were transfected with the small interfering RNA using Lipofectamine RNAiMAX transfection reagent (Invitrogen, Carlsbad, CA) according to the manufacturer's instructions. Lysate was harvested 48 hours after transfection. Cells were washed twice with phosphate-buffered saline and incubated with radioimmunoprecipitation assay buffer (Santa Cruz Biotechnology, Santa Cruz, CA) for 35 minutes. Protein extracts were run in 10% to 12% sodium dodecyl sulfate– polyacrylamide gel electrophoresis and transferred to nitrocellulose membrane. Blots were incubated with anti-ATF2 polyclonal antibody (1:1000, C-19; Santa Cruz Biotechnology), and signals were visualized using the Odyssey Infrared System (LI-COR Biosciences, Lincoln, NE).
2.3. Immunohistochemistry Immunohistochemical staining methodology was essentially as described previously [19]. Antigen retrieval was performed using a steamer for 30 minutes in 1× ethylenediaminetetraacetic acid buffer (pH 8.0). Sections were incubated with anti-ATF2 antibody at 1:100 dilution for 60 minutes at room temperature, and then Dako EnVision antirabbit secondary antibody was applied. The NovaRed Substrate Kit (Vector Labs, Burlingame, CA) was used as the visualization chromogen. Slides were then counterstained with hematoxylin and mounted. Nuclear and cytoplasmic stainings of tumor cells were scored separately by light microscopy. Nuclear staining was evaluated based on the percentage of the positive cells staining above background, ranging 0% to 100%. Cytoplasmic staining was assessed as described by others [21], based on an immunoreactive score (IRS) calculated by multiplying an intensity score by a proportion score: no visible staining, intensity score 0; weak intensity, score 1; moderate intensity, score 2; strong intensity, score 3; 0% of positive cells, proportion score 0; N0 to 10% of positive cells, proportion score 1; N10% to 50% of positive cells, proportion score 2; N50% to 80% of positive cells, proportion score 3; and N80% of positive cells, proportion score 4, resulting in an IRS of 0 to 12. Cytoplasmic staining was also scored using the
M. Endo et al. following scale: 0, no staining; 1, weak staining; 2, moderate staining; and 3, intense staining, as described previously [8]. In that 4-tier scoring system, scores 0 and 1 were regarded as negative, and scores 2 and 3 were defined as positive [8]. Among replicate cores, percentage scores were averaged. When there was discrepancy in the cytoplasmic intensity of ATF2 staining among replicate cores, the higher intensity was used to calculate the IRS. Cores with less than 50 viable tumor cells were excluded. One trained subspecialty pathologist (M.E.) evaluated or reviewed the results of the immunohistochemical staining.
2.4. Statistical analysis Comparison of the distribution of the nuclear staining positive rates between groups was made using the Wilcoxon rank sum test. Statistical significance was defined as P b .05. Data analysis was performed with JMP software (version 9.0.2) (SAS Inc., Cary, NC). Diagnostic groupings were prespecified. The category of TAS comprises mesenchymal malignancies characterized by known fusion events: synovial sarcoma, alveolar soft part sarcoma, clear cell sarcoma, dermatofibrosarcoma protuberans, desmoplastic small round cell tumor, endometrial stromal sarcoma, Ewing sarcoma, extraskeletal myxoid chondrosarcoma, myxoid liposarcoma, low-grade fibromyxoid sarcoma, solitary fibrous tumor, and alveolar rhabdomyosarcoma. Non–translocation-associated sarcoma represented all sarcomas other than TAS and gastrointestinal stromal tumor. The spindle cell sarcoma category refers to nonpleomorphic tumors from the following categories: synovial sarcoma, dermatofibrosarcoma protuberans, fibrosarcoma, gastrointestinal stromal tumor, leiomyosarcoma, low-grade fibromyxoid sarcoma, malignant peripheral nerve sheath tumor, and solitary fibrous tumor.
2.5. Digital images Digital images of immunostained and hematoxylin and eosin–stained TMAs were acquired using a BLISS imager (Bacus Laboratories, Lombard, IL). A relational database was constructed that correlates scoring and identification information with images of each core. This information is publicly accessible at http://www.gpecimage.ubc.ca (username: atf2; password: atf2).
3. Results Western blotting with the anti-ATF2 antibody used for subsequent immunohistochemistry studies is presented in Fig. 1. Cell lysates without small interfering RNA treatment showed a single band at the expected molecular weight of ATF2 (70 kd). In contrast, the band was greatly diminished after small interfering RNA knockdown of ATF2, confirming specificity of the antibody.
ATF2 in mesenchymal tumors On formalin-fixed, paraffin-embedded whole tissue sections of synovial sarcoma, nuclear ATF2 staining intensity was uniform throughout the section (4 cases examined on whole sections). In the same way, on TMA specimens (n = 68 synovial sarcomas), nuclear ATF2 staining showed relatively uniform intensity in almost all positive cases. In contrast, cytoplasmic staining intensity varied widely in positive cases. Representative cases showing positive or negative ATF2 staining are shown in Fig. 2. Among assessed normal tissues, nuclear ATF2 expression was observed in colonic epithelium, and cytoplasmic expression was observed in hepatocytes.
279 Histograms showing the distribution of nuclear and cytoplasmic ATF2 expression scores in representative tumors are presented in Fig. 3, and results of all specimen types by diagnosis are detailed in the Table. In total, 485 (81.6%) of 594 malignant and 122 (58.9%) of 207 benign mesenchymal tumors showed greater than 1% nuclear positivity, whereas 178 (30.0%) of 594 malignant and 20 (9.7%) of 207 benign mesenchymal tumors showed at least weak cytoplasmic staining (IRS ≥1). Among malignant and borderline mesenchymal tumors, the highest proportions of nuclear ATF2 expression were observed in synovial sarcoma (mean score of percentage of positive cells, 67.3%), dermatofibrosarcoma protuberans (58.7%), endometrial stromal sarcoma (68.2%), gastrointestinal stromal tumor (56.3%), malignant peripheral nerve sheath tumor (57.0%), and solitary fibrous tumor (77.8%). In contrast, chondrosarcoma (5.0%) and well-differentiated liposarcoma (10.3%) showed low ATF2 nuclear expression. Relatively high cytoplasmic expression of ATF2 was detected in endometrial stromal sarcoma (2.88) and dermatofibrosarcoma protuberans (1.88), although none of the 2 exceeded the mean value in the examined melanoma cases. Among benign mesenchymal tumors, spindle cell lipoma showed high nuclear (62.0%) and cytoplasmic (2.8) ATF2 expression, and perineurioma showed high nuclear expression (65.9%). In contrast, enchondroma, fibroma/fibromatosis, and lipoma revealed low expression levels in both nucleus and cytoplasm. Among nonmesenchymal tumors, melanoma showed high nuclear (73.8%) and cytoplasmic (mean IRS, 3.12) expression. Relatively high ATF2 expression was observed in carcinomas (45.2%; 2.10), but the expression was distinctly lower in the examined normal tissues (5.7%; 0.33). For several specific diagnostic entities, relatively few (b10) cases were available precluding confident assessment of ATF2 expression frequency. Among these, the strong nuclear staining seen in both included cases of desmoplastic small round cell tumor may warrant further assessment. Comparisons of nuclear ATF2 expression between prespecified categories of benign versus malignant mesenchymal tumors, TASs versus non-TASs, and spindle cell sarcoma versus non–spindle cell sarcoma are presented in Fig. 4. Nuclear ATF2 expression levels were significantly higher in malignant mesenchymal tumors (P b .0001), TASs (P b .0001), and spindle cell sarcomas (P b .0001) than in cases not falling into these prespecified groups.
4. Discussion Fig. 1 Western blotting with anti-ATF2 antibody on human synovial sarcoma cell line SYO-1. Cell lysates are shown without (−) and with (+) treatment with a small interfering RNA specific for the ATF2 transcript (knockdown efficiency was 60.3% by quantitative polymerase chain reaction). The expected molecular weight of ATF2 is 70 kd.
To the best of our knowledge, this is the first article presenting an immunohistochemical expression profile of ATF2 in a large series of mesenchymal tumors. Our study reveals that the nuclear expression of ATF2 is significantly higher in both low-grade and high-grade malignancies
280
M. Endo et al.
Fig. 2 ATF2 immunostaining in representative tissue cores (core diameter, 0.6mm). A, Synovial sarcoma: nuclear expression 90%; cytoplasmic expression negative 0% (IRS 0). B, Gastrointestinal stromal tumor: nuclear expression 80%; cytoplasmic expression strong/80% (IRS 9). C, Endometrial stromal sarcoma: nuclear expression 90%; cytoplasmic expression weak/65% (IRS 3). D, Ewing sarcoma: nuclear expression 95%; cytoplasmic expression moderate/80% (IRS 6). E, Solitary fibrous tumor: nuclear expression 92.5%; cytoplasmic expression negative/0% (IRS 0). F, Chondrosarcoma: nuclear expression 0%; cytoplasmic expression negative/0% (IRS 0). G, Myxoid liposarcoma: nuclear expression 10%; cytoplasmic expression negative/0% (IRS 0). H, Melanoma: nuclear expression 98%; cytoplasmic expression moderate/80% (IRS 6).
ATF2 in mesenchymal tumors
281
Fig. 3 Histograms showing the distribution of nuclear and cytoplasmic expression of ATF2 in representative high–nuclear expression tumors. Histogram of melanoma is shown as a positive control and for comparison.
than in benign mesenchymal tumors, with synovial sarcoma, dermatofibrosarcoma protuberans, endometrial stromal sarcoma, gastrointestinal stromal tumor, malignant peripheral nerve sheath tumor, and solitary fibrous tumor showing particularly high nuclear ATF2 expression. In the literature, enhanced nuclear ATF2 expression
has been reported in metastatic melanomas [8] and skin carcinomas [22]. Biologically, it is plausible that higher nuclear expression of ATF2 is seen in malignant tumors over benign lesions and normal tissue because ATF2 works as an effector of the mitogen-activated protein kinase signaling pathway [9].
282 Table
M. Endo et al. ATF2 expression in mesenchymal tumors
Diagnosis
n
Malignant and borderline mesenchymal tumors n ≥ 10 Synovial sarcoma a,b 68 Chondrosarcoma, conventional 37 17 Dermatofibrosarcoma protuberans a,b Endometrial stromal sarcoma a 59 Ewing sarcoma a 21 Gastrointestinal stromal tumor b 145 Liposarcoma, myxoid a 33 Liposarcoma, well differentiated 10 Malignant peripheral nerve 74 sheath tumor b Rhabdomyosarcoma, embryonal 11 Solitary fibrous tumor a,b 15 n b10 Alveolar soft part sarcoma a 8 Angiosarcoma 4 Atypical fibroxanthoma 2 Chondrosarcoma, dedifferentiated 4 Chondrosarcoma, mesenchymal 3 Clear cell sarcoma a 8 Desmoid-type fibromatosis 7 Desmoplastic small round cell tumor a 2 Epithelioid hemangioendothelioma 2 Epithelioid sarcoma 9 Extraskeletal myxoid chondrosarcoma a 4 Fibrosarcoma b 8 6 Leiomyosarcoma b Liposarcoma, dedifferentiated 5 Liposarcoma, pleomorphic 6 Low-grade fibromyxoid sarcoma a,b 5 Osteosarcoma, conventional 4 Osteosarcoma, parosteal 2 Rhabdomyosarcoma, alveolar a 9 Undifferentiated pleomorphic sarcoma 6 Benign mesenchymal tumors n ≥10 Enchondroma 17 Fibroma/fibromatosis 26 Lipoma 10 Neurofibroma 43 Pigmented villonodular synovitis 52 Schwannoma 22 Spindle cell lipoma 10 n b 10 Chondroblastoma 4 Chondromyxoid fibroma 3 Congenital mesoblastic nephroma 2 Fetal rhabdomyoma 2 Myofibroma 6 Myositis ossificans 2 Nodular fasciitis 1 Osteochondroma 3 Perineurioma 4
Nuclear expression
Cytoplasmic expression
Mean score (%) SE
95% CI
Mean score (IRS) SE
95% CI
Positivity (%)
67.3 5.0 58.7 68.2 31.1 56.3 20.1 10.3 57.0
4.0 2.4 8.6 2.9 7.7 2.3 4.9 7.0 4.3
59.3-75.2 0.1-9.8 40.3-77.0 62.4-74.0 15.1-47.2 51.6-60.9 10.2-30.0 0-26.1 48.4-65.5
0.60 0.027 1.88 2.88 0.95 0.97 0.03 0.4 1.23
0.15 0.027 0.49 0.34 0.40 0.15 0.03 0.4 0.21
0.30-0.90 0-0.082 0.84-2.92 2.20-3.57 0.12-1.79 0.66-1.27 0-0.09 0-1.30 0.81-1.65
1.5 0 11.8 20.3 4.8 10.3 0 10 8.1
14.5 77.8
5.1 7.5
3.3-25.8 0.36 61.6-94.0 1.6
0.24 0.74
0-0.91 0 0.02-3.18 13.3
36.6 56.9 41.3 45.0 0 48.2 0.7 80.0 51.3 28.6 34.4 9.4 30.1 28.8 46.8 17.5 0.6 0 20.4 61
14.1 14.5 41.3 20.0 0 15.5 0.7 3.3 28.8 9.2 14.3 5.1 10.1 10.9 11.4 5.0 0.6 0 9.8 7.9
2.25 0.5 1.5 0 0 1.75 0 4 0 0 0 0.38 0.67 4.4 2.33 0.8 0 0 1.11 6.33
1.16 0.5 1.5 0 0 0.88 0 0 0 0 0 0.38 0.42 0.75 1.20 0.49 0 0 0.56 1.09
25.0 0 0 0 0 12.5 0 0 0 0 0 0 0 60.0 33.3 0 0 0 0 83.3
0 0.077 0 0.047 0.35 0.59 2.8
0 0.077 0 0.047 0.13 0.34 0.61
0 0 0 0 0 0 0 0 0
0 0 0 0 0 0
0.7 1.3 0.03 18.2 28.2 29.0 62.0 0 34.2 0 0.8 0 0 7.5 0 65.9
0.6 0.9 0.03 3.6 3.3 6.9 6.6 0 25.9 0 0.8 0 0 0 21.2
0-1.9 0-3.2 0-0.1 11.0-25.4 21.6-34.9 14.7-43.3 47.0-77.0
0 0
0-0 0 0-0.24 0 0-0 0 0-0.14 0 0.08-0.61 1.9 0-1.30 9.1 1.42-4.18 50.0 0 0 0 0 0 0 0 0 0
ATF2 in mesenchymal tumors
283
Table (continued) Diagnosis
n
Nuclear expression
Cytoplasmic expression
Mean score (%) SE Malignant nonmesenchymal tumors n ≥ 10 Carcinoma Melanoma Wilms tumor n b 10 Carcinoid Hepatoblastoma Malignant lymphoma Mesothelioma Normal tissue a b
40 17 23
45.2 73.8 15.1
2 4 3 4 18
1.5 22.5 48.3 30.5 5.7
95% CI
Mean score (IRS) SE
6.2 32.8-57.7 2.10 7.8 57.4-90.3 3.12 4.2 6.5-23.7 1.04 1.5 15.3 24.0 18.6 3.3 0-12.6
4 0 2 1.75 0.33
95% CI
Positivity (%)
0.38 0.63 0.32
1.33-2.87 25.0 1.78-4.45 23.5 0.39-1.70 0
0 0 2 0.63 0.33
100 0 33.3 0 5.6
0-1.04
Translocation-associated sarcoma. Spindle cell sarcoma.
Interestingly, the sarcomas with the highest nuclear ATF2 expression all fall into the categories of TAS and/or spindle cell sarcoma. The activator protein 1 (AP1) transcription factor superfamily including ATF and CREB is known to form homodimers or heterodimers with other AP1 factors to induce transcriptional activities [1]. High nuclear expression of ATF2 may therefore reflect high AP1 family transcription factor activity in these cancers, in which some of the fusion genes retain AP1 family dimerization domains [2]. In addition to TASs, spindle cell sarcomas including gastrointestinal stromal tumor and malignant peripheral nerve sheath tumor showed high ATF2 nuclear expression, suggesting that the AP1 superfamily may also have some role in these tumors. In synovial sarcoma, recent advances in understanding fusion oncoprotein biology have revealed that SS18-SSX forms a complex bridging ATF2 to TLE1, leading to abnormal silencing of ATF2 target genes [3]. ATF2 guides SS18-SSX to its target gene promoters, where TLE1 functions to recruit repressor complexes. In this study, we investigated ATF2 expression in human mesenchymal tumor clinical specimens including 68 synovial sarcomas and found particularly high levels of nuclear ATF2 expression in synovial sarcomas, supporting its fundamental role in the biology of this cancer. TLE1, the other direct partner of SS18-SSX, has now been shown by several groups to be highly expressed at the immunohistochemical level in synovial sarcomas [18,23–26]. TLE1 is particularly sensitive and specific for synovial sarcoma and has become a useful diagnostic marker for this disease. ATF2, although as important to the biology of synovial sarcoma, is unlikely to perform as well as TLE1 as a diagnostic marker because its specificity is lower, possibly reflecting a more widespread role for ATF2 in sarcoma biology. Interestingly, although ATF2 can promote cancer by inducing transcription activity when localized in the nucleus, it can also act as a tumor suppressor by enhancing apoptotic events at the level of the mitochondria when present in the cytoplasm [5]. These contradictory functions of ATF2 have
been supported by immunohistochemical data showing that strong cytoplasmic and weak nuclear staining of ATF2 is associated with better prognosis in melanoma [8]. In our series, almost all mesenchymal tumors showed lower cytoplasmic ATF2 immunoreactivity than in equivalently stained and scored melanomas, with the only exceptions being dedifferentiated liposarcoma and undifferentiated pleomorphic sarcoma. ATF2's biological role in the cytoplasm of mesenchymal tumors has yet to be clarified. Our study has 2 important limitations: the numbers of rare tumors available for assessment and a lack of clinical data allowing assessment of the prognostic value of ATF2. We examined only a few examples for several rare tumor entities because there are more than 100 categories of mesenchymal tumors, many extremely rare. For such rare tumors, further investigation might be necessary to draw firm conclusions about ATF2 expression. ATF2 expression is known to affect the survival of patients with melanoma [8], so ATF2 could be prognostic in mesenchymal tumors as well, the investigation of which would require linkage to patient outcomes. In summary, we have described that TASs and spindle cell sarcomas, including synovial sarcoma, dermatofibrosarcoma protuberans, endometrial stromal sarcoma, gastrointestinal stromal tumor, malignant peripheral nerve sheath tumor, and solitary fibrous tumor, show higher nuclear ATF2 expression than other mesenchymal tumors. This work supports molecular functional data implicating ATF2 in the biology of synovial sarcoma and suggests a potential wider involvement in the oncogenic processes driving molecularly and morphologically related cancers.
Acknowledgments We thank Christine Chow and Doris Gao for their technical assistance. Specimen access was provided through the BC Bone and Soft Tissue Tumour Bank (protocol H08-01717).
284
Fig. 4 Boxplots and histograms showing the difference in nuclear ATF2 expression between benign and malignant mesenchymal tumors (A), or TAS and non-TAS (B), or spindle cell sarcoma and non–spindle cell sarcoma (C). Nuclear ATF2 expression rate is significantly higher in malignant mesenchymal tumor (P b .0001), TAS (P b .0001), and spindle cell sarcoma (P b .0001).
References [1] Lau E, Ronai ZA. ATF2—at the crossroad of nuclear and cytosolic functions. J Cell Sci 2012;125:2815-24. [2] Riggi N, Cironi L, Suva ML, Stamenkovic I. Sarcomas: genetics, signalling, and cellular origins. Part 1: the fellowship of TET. J Pathol 2007;213:4-20. [3] Su L, Sampaio AV, Jones KB, et al. Deconstruction of the SS18-SSX fusion oncoprotein complex: insights into disease etiology and therapeutics. Cancer Cell 2012;21:333-47. [4] Manna PR, Dyson MT, Stocco DM. Role of basic leucine zipper proteins in transcriptional regulation of the steroidogenic acute regulatory protein gene. Mol Cell Endocrinol 2009;302:1-11.
M. Endo et al. [5] Bhoumik A, Ronai Z. ATF2: a transcription factor that elicits oncogenic or tumor suppressor activities. Cell Cycle 2008;7:2341-5. [6] Nagadoi A, Nakazawa K, Uda H, et al. Solution structure of the transactivation domain of ATF-2 comprising a zinc finger-like subdomain and a flexible subdomain. J Mol Biol 1999;287:593-607. [7] Lopez-Bergami P, Lau E, Ronai Z. Emerging roles of ATF2 and the dynamic AP1 network in cancer. Nat Rev Cancer 2010;10:65-76. [8] Berger AJ, Kluger HM, Li N, et al. Subcellular localization of activating transcription factor 2 in melanoma specimens predicts patient survival. Cancer Res 2003;63:8103-7. [9] Gozdecka M, Breitwieser W. The roles of ATF2 (activating transcription factor 2) in tumorigenesis. Biochem Soc Trans 2012;40:230-4. [10] Ronai Z, Yang YM, Fuchs SY, Adler V, Sardana M, Herlyn M. ATF2 confers radiation resistance to human melanoma cells. Oncogene 1998;16:523-31. [11] Ivanov VN, Ronai Z. p38 protects human melanoma cells from UVinduced apoptosis through down-regulation of NF-kappaB activity and Fas expression. Oncogene 2000;19:3003-12. [12] Yang YM, Dolan LR, Ronai Z. Expression of dominant negative CREB reduces resistance to radiation of human melanoma cells. Oncogene 1996;12:2223-33. [13] Nielsen TO, Hsu FD, O'Connell JX, et al. Tissue microarray validation of epidermal growth factor receptor and SALL2 in synovial sarcoma with comparison to tumors of similar histology. Am J Pathol 2003;163: 1449-56. [14] Ng TL, Gown AM, Barry TS, et al. Nuclear beta-catenin in mesenchymal tumors. Mod Pathol 2005;18:68-74. [15] Cheng H, Dodge J, Mehl E, et al. Validation of immature adipogenic status and identification of prognostic biomarkers in myxoid liposarcoma using tissue microarrays. HUM PATHOL 2009;40:1244-51. [16] Steigen SE, Straume B, Turbin D, et al. Clinicopathologic factors and nuclear morphometry as independent prognosticators in KIT-positive gastrointestinal stromal tumors. J Histochem Cytochem 2008;56: 139-45. [17] Pacheco M, Nielsen TO. Histone deacetylase 1 and 2 in mesenchymal tumors. Mod Pathol 2012;25:222-30. [18] Terry J, Saito T, Subramanian S, et al. TLE1 as a diagnostic immunohistochemical marker for synovial sarcoma emerging from gene expression profiling studies. Am J Surg Pathol 2007;31:240-6. [19] Sun Y, Turbin DA, Ling K, et al. Type I gamma phosphatidylinositol phosphate kinase modulates invasion and proliferation and its expression correlates with poor prognosis in breast cancer. Breast Cancer Res 2010;12:R6. [20] Kawai A, Naito N, Yoshida A, et al. Establishment and characterization of a biphasic synovial sarcoma cell line, SYO-1. Cancer Lett 2004;204:105-13. [21] Knippen S, Loning T, Muller V, Schroder C, Janicke F, MildeLangosch K. Expression and prognostic value of activating transcription factor 2 (ATF2) and its phosphorylated form in mammary carcinomas. Anticancer Res 2009;29:183-9. [22] Chen SY, Takeuchi S, Urabe K, et al. Overexpression of phosphorylated-ATF2 and STAT3 in cutaneous angiosarcoma and pyogenic granuloma. J Cutan Pathol 2008;35:722-30. [23] Kosemehmetoglu K, Vrana JA, Folpe AL. TLE1 expression is not specific for synovial sarcoma: a whole section study of 163 soft tissue and bone neoplasms. Mod Pathol 2009;22:872-8. [24] Jagdis A, Rubin BP, Tubbs RR, Pacheco M, Nielsen TO. Prospective evaluation of TLE1 as a diagnostic immunohistochemical marker in synovial sarcoma. Am J Surg Pathol 2009;33:1743-51. [25] Knosel T, Heretsch S, Altendorf-Hofmann A, et al. TLE1 is a robust diagnostic biomarker for synovial sarcomas and correlates with t(X;18): analysis of 319 cases. Eur J Cancer 2010;46:1170-6. [26] Foo WC, Cruise MW, Wick MR, Hornick JL. Immunohistochemical staining for TLE1 distinguishes synovial sarcoma from histologic mimics. Am J Clin Pathol 2011;135:839-44.
www.elsevier.com/locate/humpath
Original contribution
Activating transcription factor 2 in mesenchymal tumors☆,☆☆ Makoto Endo MD, PhD a,b,c,d , Le Su PhD e , Torsten O. Nielsen MD, PhD a,b,⁎ a
Genetic Pathology Evaluation Centre, University of British Columbia, Vancouver, British Columbia, Canada Department of Pathology and Laboratory Medicine, University of British Columbia, Vancouver, British Columbia, Canada c Department of Anatomic Pathology, Kyushu University, Fukuoka, Japan d Department of Orthopaedic Surgery, Kyushu University, Fukuoka, Japan e Biomedical Research Centre, University of British Columbia, Vancouver, British Columbia, Canada b
Received 10 May 2013; revised 27 August 2013; accepted 11 September 2013
Keywords: Activating transcription factor 2; Transcription factor; Mesenchymal tumor; Translocation-associated sarcoma; Immunohistochemistry
Summary Activating transcription factor 2 (ATF2) is a member of activator protein 1 superfamily, which can heterodimerize with other transcription factors regulating cell differentiation and survival. ATF2 assembles into a complex with the synovial sarcoma translocation, chromosome 18 (SS18)– synovial sarcoma, X breakpoint (SSX) fusion oncoprotein, and the transducin-like enhancer of split 1 (TLE1) corepressor, driving oncogenesis in synovial sarcoma. The fusion oncoproteins in many other translocation-associated sarcomas incorporate transcription factors from the ATF/cAMP response element binding or E26 families, which potentially form heterodimers with ATF2 to regulate transcription. ATF2 may therefore play an important role in the oncogenesis of many mesenchymal tumors, but as yet, little is known about its protein expression in patient specimens. Herein we perform immunohistochemical analyses using a validated specific antibody for ATF2 expression and intracellular localization on a cohort of 594 malignant and 207 benign mesenchymal tumors representing 47 diagnostic entities. Melanoma served as a positive control for nuclear and cytoplasmic staining. High nuclear ATF2 expression was mainly observed in translocation-associated and/or spindle cell sarcomas including synovial sarcoma, desmoplastic small round cell tumor, endometrial stromal sarcoma, gastrointestinal stromal tumor, malignant peripheral nerve sheath tumor, and solitary fibrous tumor. Cytoplasmic ATF2 expression was less frequently seen than nuclear expression in malignant mesenchymal tumors. Benign mesenchymal tumors mostly showed much lower nuclear and cytoplasmic ATF2 expression. © 2014 Elsevier Inc. All rights reserved.
☆
Disclosure/Conflict of interest: The authors declare no conflict of interest. This work was supported by grants from the Canadian Cancer Society Research Institute (grant no. 701582), an Extension Grant from the Canadian Cancer Society Research Institute (grant no. 2012-701558), and the Liddy Shriver Sarcoma Initiative. M.E. is a Japan Society for the Promotion of Science research fellow. ⁎ Corresponding author. Genetic Pathology Evaluation Centre, University of British Columbia, Room 509-2660 Oak St, Vancouver, British Columbia, Canada V6H 3Z6. E-mail address: [email protected] (T. O. Nielsen). ☆☆
0046-8177/$ – see front matter © 2014 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.humpath.2013.09.003
ATF2 in mesenchymal tumors
277
1. Introduction
2. Materials and methods
A wide variety of transcription factors have been demonstrated to have oncogenic and/or tumor suppressor activities in human malignancies [1]. Most genes located at the breakpoint of translocation-associated sarcomas (TASs) retain DNA-binding domains from transcription factors such as DNA-damage-inducible transcript 3, Friend leukemia integration 1 transcription factor, v-ets avian erythroblastosis virus E26 oncogene homolog, ETS translocation variant 1, and activating transcription factor 1 (ATF1), implicating these transcription factors in sarcoma tumorigenesis [2]. In addition, in synovial sarcoma, the synovial sarcoma translocation, chromosome 18 (SS18)–synovial sarcoma, X breakpoint (SSX) fusion oncoprotein (which has no DNA-binding domain) forms a complex with ATF2 and transducin-like enhancer of split 1 (TLE1), allowing the SS18-SSX fusion oncoprotein to bind DNA through ATF2 and regulate the downstream genes expression related to its tumorigenesis via TLE1's interactions with chromatin-modifying enzymes [3]. ATF2 is a member of the ATF/cAMP response element binding (CREB) family, comprising ATF1 (involved in Ewing sarcoma breakpoint region 1 [EWSR1]-ATF1 fusions in clear cell sarcoma and EWSR1-ATF1 or fused in sarcoma [FUS]ATF1 fusions in angiomatoid fibrous histiocytoma), cAMP response element binding protein 3-like 1 (CREB3L1) and CREB3L2 (involved in FUS-CREB3L1 or FUS-CREB3L2 fusions in low-grade fibromyxoid sarcoma), and other ATF or CREB proteins [2,4]. ATF2 contains 2 functional domains: an N-terminal transactivation domain and a C-terminal DNAbinding domain [5]. In the above fusions, the activating domain is lost, replaced by EWSR1 or FUS. The retained DNA-binding domain contains the basic leucine zipper motif, which enables ATF2 to dimerize and to bind to cAMP response element (CRE) sequences in the promoters of target genes [6,7]. Recently, ATF2 has been found to exhibit oncogenic and tumor suppressor activities, depending on tumor type and intracellular localization [1,5,7–9]. In the nucleus, ATF2 plays an important role in transcription, whereas in the cytoplasm, ATF2 impairs mitochondrial membrane potential, resulting in mitochondrial-based cell death [1,5,9]. The implication that ATF2 is oncogenic in the nucleus but tumor suppressive in the cytoplasm is supported by immunohistochemical findings that weaker cytoplasmic and stronger nuclear staining correlates with poor prognosis in melanomas [8]. ATF2 expression and localization appear to play an important role in resistance of melanoma cells to chemotherapeutic drugs and radiation [10–12]. Although these recent findings support the biological and clinical importance of ATF2 in human malignancies, neither the intracellular localization nor the expression level of ATF2 protein has yet been well characterized in mesenchymal tumors. Here, we performed a large-scale ATF2 immunohistochemical expression study using tissue microarrays (TMAs), including synovial sarcoma together with a wide variety of human mesenchymal tumor specimens.
2.1. Tumor samples and TMAs In this study, both previously described and more recently constructed TMAs were used. Each TMA contains 0.6-mm cores derived from representative viable diagnostic areas identified by a subspecialty pathologist (T.O.N.). Published TMAs used in this study are TMA 01-003 [13] (synovial sarcoma differential diagnosis array, 82 cases in duplicate), TMA 03-008 [14] (chondroid tumor array, 121 cases in duplicate), TMA 06-007 [15] (liposarcoma array, 69 cases in triplicate), TMA 06-001B [16] (gastrointestinal stromal tumor array, 150 cases in duplicate), TMA 08-019 [17] (endometrial stromal sarcoma array, 59 cases in duplicate), TMA 09-006 [17] (epithelioid sarcoma differential diagnosis array, 53 cases in duplicate), TMA 10-009 [17] (alveolar soft part sarcoma–alveolar rhabdomyosarcoma–desmoplastic small round cell tumor array, 12 cases in triplicate), and TMA malignant peripheral nerve sheath tumor [18] (malignant peripheral nerve sheath tumor differential diagnosis array, 176 cases in duplicate). Five additional TMAs not previously published were used in this study, assembled using a manual tissue microarrayer (Beecher Instruments, Inc, Silver Springs, MD) from formalinfixed, paraffin-embedded tissue as described previously [19]. TMA 09-008 (pigmented villonodular synovitis array) contains singular or triplicate cores of 55 cases of pigmented villonodular synovitis. TMA 12-004 (B-cell lymphoma 2, BCL2-positive tumor array) contains triplicate cores of 10 spindle cell lipomas, 8 cases of dermatofibrosarcoma protuberans, 10 solitary fibrous tumors, and 7 Ewing sarcomas. TMA 12-005 (pediatric mesenchymal tumor array) contains duplicate cores of 4 hepatoblastomas, 30 fibromas/fibromatoses (including 12 lipofibromatoses, 10 superficial or digital fibromatoses, 3 cases of fibromatosis colli, 2 calcifying aponeurotic fibromas, 2 Gardner fibromas, 1 irritation fibroma, and 2 fibromas not otherwise specified), 23 Wilms tumors, 9 Ewing sarcomas, 8 alveolar rhabdomyosarcomas, 7 synovial sarcomas, 1 nodular fasciitis, 6 myofibromas, 2 myositis ossificans, 9 cases of fibrous hamartoma of infancy, 8 desmoid-type fibromatoses, 8 fibrosarcomas, 12 embryonal rhabdomyosarcomas, 2 congenital mesoblastic nephromas, 1 gastrointestinal stromal tumor, and 2 fetal rhabdomyomas. TMA12-006 (TAS array) includes duplicate cores of 2 myxoid liposarcomas, 4 synovial sarcomas, 1 malignant peripheral nerve sheath tumor, 1 extraskeletal myxoid chondrosarcoma, 1 endometrial stromal sarcoma, and 1 clear cell sarcoma. TMA12-010 (pleomorphic sarcoma array) contains duplicate cores of 5 undifferentiated pleomorphic sarcomas and 5 dedifferentiated liposarcomas. TMA 06-003 (multitumor array, 76 cases) including a single core of carcinoids, a variety of carcinomas, malignant lymphomas, melanomas, mesotheliomas, and normal tissues was used for context with the melanomas considered as a
278 positive control for staining and interpretation. Normal tissues on this array included 9 examples of fat, 2 of colon, and 1 each of adrenal gland, cerebellum, fallopian tube, heart, liver, lung, and squamous mucosa.
2.2. Western blotting and RNA interference Antibody specificity was tested by Western blotting using human synovial sarcoma cell line SYO-1 [20] (kindly provided by Dr Akira Kawai, National Cancer Center Hospital, Tokyo) with or without RNA interference for ATF2. The small interfering RNA specific for ATF2 was purchased from Dharmacon (Lafayette, CO) and had a knockdown efficiency of 60.3% by quantitative reverse transcription polymerase chain reaction assessment. At 60% confluence, cells were transfected with the small interfering RNA using Lipofectamine RNAiMAX transfection reagent (Invitrogen, Carlsbad, CA) according to the manufacturer's instructions. Lysate was harvested 48 hours after transfection. Cells were washed twice with phosphate-buffered saline and incubated with radioimmunoprecipitation assay buffer (Santa Cruz Biotechnology, Santa Cruz, CA) for 35 minutes. Protein extracts were run in 10% to 12% sodium dodecyl sulfate– polyacrylamide gel electrophoresis and transferred to nitrocellulose membrane. Blots were incubated with anti-ATF2 polyclonal antibody (1:1000, C-19; Santa Cruz Biotechnology), and signals were visualized using the Odyssey Infrared System (LI-COR Biosciences, Lincoln, NE).
2.3. Immunohistochemistry Immunohistochemical staining methodology was essentially as described previously [19]. Antigen retrieval was performed using a steamer for 30 minutes in 1× ethylenediaminetetraacetic acid buffer (pH 8.0). Sections were incubated with anti-ATF2 antibody at 1:100 dilution for 60 minutes at room temperature, and then Dako EnVision antirabbit secondary antibody was applied. The NovaRed Substrate Kit (Vector Labs, Burlingame, CA) was used as the visualization chromogen. Slides were then counterstained with hematoxylin and mounted. Nuclear and cytoplasmic stainings of tumor cells were scored separately by light microscopy. Nuclear staining was evaluated based on the percentage of the positive cells staining above background, ranging 0% to 100%. Cytoplasmic staining was assessed as described by others [21], based on an immunoreactive score (IRS) calculated by multiplying an intensity score by a proportion score: no visible staining, intensity score 0; weak intensity, score 1; moderate intensity, score 2; strong intensity, score 3; 0% of positive cells, proportion score 0; N0 to 10% of positive cells, proportion score 1; N10% to 50% of positive cells, proportion score 2; N50% to 80% of positive cells, proportion score 3; and N80% of positive cells, proportion score 4, resulting in an IRS of 0 to 12. Cytoplasmic staining was also scored using the
M. Endo et al. following scale: 0, no staining; 1, weak staining; 2, moderate staining; and 3, intense staining, as described previously [8]. In that 4-tier scoring system, scores 0 and 1 were regarded as negative, and scores 2 and 3 were defined as positive [8]. Among replicate cores, percentage scores were averaged. When there was discrepancy in the cytoplasmic intensity of ATF2 staining among replicate cores, the higher intensity was used to calculate the IRS. Cores with less than 50 viable tumor cells were excluded. One trained subspecialty pathologist (M.E.) evaluated or reviewed the results of the immunohistochemical staining.
2.4. Statistical analysis Comparison of the distribution of the nuclear staining positive rates between groups was made using the Wilcoxon rank sum test. Statistical significance was defined as P b .05. Data analysis was performed with JMP software (version 9.0.2) (SAS Inc., Cary, NC). Diagnostic groupings were prespecified. The category of TAS comprises mesenchymal malignancies characterized by known fusion events: synovial sarcoma, alveolar soft part sarcoma, clear cell sarcoma, dermatofibrosarcoma protuberans, desmoplastic small round cell tumor, endometrial stromal sarcoma, Ewing sarcoma, extraskeletal myxoid chondrosarcoma, myxoid liposarcoma, low-grade fibromyxoid sarcoma, solitary fibrous tumor, and alveolar rhabdomyosarcoma. Non–translocation-associated sarcoma represented all sarcomas other than TAS and gastrointestinal stromal tumor. The spindle cell sarcoma category refers to nonpleomorphic tumors from the following categories: synovial sarcoma, dermatofibrosarcoma protuberans, fibrosarcoma, gastrointestinal stromal tumor, leiomyosarcoma, low-grade fibromyxoid sarcoma, malignant peripheral nerve sheath tumor, and solitary fibrous tumor.
2.5. Digital images Digital images of immunostained and hematoxylin and eosin–stained TMAs were acquired using a BLISS imager (Bacus Laboratories, Lombard, IL). A relational database was constructed that correlates scoring and identification information with images of each core. This information is publicly accessible at http://www.gpecimage.ubc.ca (username: atf2; password: atf2).
3. Results Western blotting with the anti-ATF2 antibody used for subsequent immunohistochemistry studies is presented in Fig. 1. Cell lysates without small interfering RNA treatment showed a single band at the expected molecular weight of ATF2 (70 kd). In contrast, the band was greatly diminished after small interfering RNA knockdown of ATF2, confirming specificity of the antibody.
ATF2 in mesenchymal tumors On formalin-fixed, paraffin-embedded whole tissue sections of synovial sarcoma, nuclear ATF2 staining intensity was uniform throughout the section (4 cases examined on whole sections). In the same way, on TMA specimens (n = 68 synovial sarcomas), nuclear ATF2 staining showed relatively uniform intensity in almost all positive cases. In contrast, cytoplasmic staining intensity varied widely in positive cases. Representative cases showing positive or negative ATF2 staining are shown in Fig. 2. Among assessed normal tissues, nuclear ATF2 expression was observed in colonic epithelium, and cytoplasmic expression was observed in hepatocytes.
279 Histograms showing the distribution of nuclear and cytoplasmic ATF2 expression scores in representative tumors are presented in Fig. 3, and results of all specimen types by diagnosis are detailed in the Table. In total, 485 (81.6%) of 594 malignant and 122 (58.9%) of 207 benign mesenchymal tumors showed greater than 1% nuclear positivity, whereas 178 (30.0%) of 594 malignant and 20 (9.7%) of 207 benign mesenchymal tumors showed at least weak cytoplasmic staining (IRS ≥1). Among malignant and borderline mesenchymal tumors, the highest proportions of nuclear ATF2 expression were observed in synovial sarcoma (mean score of percentage of positive cells, 67.3%), dermatofibrosarcoma protuberans (58.7%), endometrial stromal sarcoma (68.2%), gastrointestinal stromal tumor (56.3%), malignant peripheral nerve sheath tumor (57.0%), and solitary fibrous tumor (77.8%). In contrast, chondrosarcoma (5.0%) and well-differentiated liposarcoma (10.3%) showed low ATF2 nuclear expression. Relatively high cytoplasmic expression of ATF2 was detected in endometrial stromal sarcoma (2.88) and dermatofibrosarcoma protuberans (1.88), although none of the 2 exceeded the mean value in the examined melanoma cases. Among benign mesenchymal tumors, spindle cell lipoma showed high nuclear (62.0%) and cytoplasmic (2.8) ATF2 expression, and perineurioma showed high nuclear expression (65.9%). In contrast, enchondroma, fibroma/fibromatosis, and lipoma revealed low expression levels in both nucleus and cytoplasm. Among nonmesenchymal tumors, melanoma showed high nuclear (73.8%) and cytoplasmic (mean IRS, 3.12) expression. Relatively high ATF2 expression was observed in carcinomas (45.2%; 2.10), but the expression was distinctly lower in the examined normal tissues (5.7%; 0.33). For several specific diagnostic entities, relatively few (b10) cases were available precluding confident assessment of ATF2 expression frequency. Among these, the strong nuclear staining seen in both included cases of desmoplastic small round cell tumor may warrant further assessment. Comparisons of nuclear ATF2 expression between prespecified categories of benign versus malignant mesenchymal tumors, TASs versus non-TASs, and spindle cell sarcoma versus non–spindle cell sarcoma are presented in Fig. 4. Nuclear ATF2 expression levels were significantly higher in malignant mesenchymal tumors (P b .0001), TASs (P b .0001), and spindle cell sarcomas (P b .0001) than in cases not falling into these prespecified groups.
4. Discussion Fig. 1 Western blotting with anti-ATF2 antibody on human synovial sarcoma cell line SYO-1. Cell lysates are shown without (−) and with (+) treatment with a small interfering RNA specific for the ATF2 transcript (knockdown efficiency was 60.3% by quantitative polymerase chain reaction). The expected molecular weight of ATF2 is 70 kd.
To the best of our knowledge, this is the first article presenting an immunohistochemical expression profile of ATF2 in a large series of mesenchymal tumors. Our study reveals that the nuclear expression of ATF2 is significantly higher in both low-grade and high-grade malignancies
280
M. Endo et al.
Fig. 2 ATF2 immunostaining in representative tissue cores (core diameter, 0.6mm). A, Synovial sarcoma: nuclear expression 90%; cytoplasmic expression negative 0% (IRS 0). B, Gastrointestinal stromal tumor: nuclear expression 80%; cytoplasmic expression strong/80% (IRS 9). C, Endometrial stromal sarcoma: nuclear expression 90%; cytoplasmic expression weak/65% (IRS 3). D, Ewing sarcoma: nuclear expression 95%; cytoplasmic expression moderate/80% (IRS 6). E, Solitary fibrous tumor: nuclear expression 92.5%; cytoplasmic expression negative/0% (IRS 0). F, Chondrosarcoma: nuclear expression 0%; cytoplasmic expression negative/0% (IRS 0). G, Myxoid liposarcoma: nuclear expression 10%; cytoplasmic expression negative/0% (IRS 0). H, Melanoma: nuclear expression 98%; cytoplasmic expression moderate/80% (IRS 6).
ATF2 in mesenchymal tumors
281
Fig. 3 Histograms showing the distribution of nuclear and cytoplasmic expression of ATF2 in representative high–nuclear expression tumors. Histogram of melanoma is shown as a positive control and for comparison.
than in benign mesenchymal tumors, with synovial sarcoma, dermatofibrosarcoma protuberans, endometrial stromal sarcoma, gastrointestinal stromal tumor, malignant peripheral nerve sheath tumor, and solitary fibrous tumor showing particularly high nuclear ATF2 expression. In the literature, enhanced nuclear ATF2 expression
has been reported in metastatic melanomas [8] and skin carcinomas [22]. Biologically, it is plausible that higher nuclear expression of ATF2 is seen in malignant tumors over benign lesions and normal tissue because ATF2 works as an effector of the mitogen-activated protein kinase signaling pathway [9].
282 Table
M. Endo et al. ATF2 expression in mesenchymal tumors
Diagnosis
n
Malignant and borderline mesenchymal tumors n ≥ 10 Synovial sarcoma a,b 68 Chondrosarcoma, conventional 37 17 Dermatofibrosarcoma protuberans a,b Endometrial stromal sarcoma a 59 Ewing sarcoma a 21 Gastrointestinal stromal tumor b 145 Liposarcoma, myxoid a 33 Liposarcoma, well differentiated 10 Malignant peripheral nerve 74 sheath tumor b Rhabdomyosarcoma, embryonal 11 Solitary fibrous tumor a,b 15 n b10 Alveolar soft part sarcoma a 8 Angiosarcoma 4 Atypical fibroxanthoma 2 Chondrosarcoma, dedifferentiated 4 Chondrosarcoma, mesenchymal 3 Clear cell sarcoma a 8 Desmoid-type fibromatosis 7 Desmoplastic small round cell tumor a 2 Epithelioid hemangioendothelioma 2 Epithelioid sarcoma 9 Extraskeletal myxoid chondrosarcoma a 4 Fibrosarcoma b 8 6 Leiomyosarcoma b Liposarcoma, dedifferentiated 5 Liposarcoma, pleomorphic 6 Low-grade fibromyxoid sarcoma a,b 5 Osteosarcoma, conventional 4 Osteosarcoma, parosteal 2 Rhabdomyosarcoma, alveolar a 9 Undifferentiated pleomorphic sarcoma 6 Benign mesenchymal tumors n ≥10 Enchondroma 17 Fibroma/fibromatosis 26 Lipoma 10 Neurofibroma 43 Pigmented villonodular synovitis 52 Schwannoma 22 Spindle cell lipoma 10 n b 10 Chondroblastoma 4 Chondromyxoid fibroma 3 Congenital mesoblastic nephroma 2 Fetal rhabdomyoma 2 Myofibroma 6 Myositis ossificans 2 Nodular fasciitis 1 Osteochondroma 3 Perineurioma 4
Nuclear expression
Cytoplasmic expression
Mean score (%) SE
95% CI
Mean score (IRS) SE
95% CI
Positivity (%)
67.3 5.0 58.7 68.2 31.1 56.3 20.1 10.3 57.0
4.0 2.4 8.6 2.9 7.7 2.3 4.9 7.0 4.3
59.3-75.2 0.1-9.8 40.3-77.0 62.4-74.0 15.1-47.2 51.6-60.9 10.2-30.0 0-26.1 48.4-65.5
0.60 0.027 1.88 2.88 0.95 0.97 0.03 0.4 1.23
0.15 0.027 0.49 0.34 0.40 0.15 0.03 0.4 0.21
0.30-0.90 0-0.082 0.84-2.92 2.20-3.57 0.12-1.79 0.66-1.27 0-0.09 0-1.30 0.81-1.65
1.5 0 11.8 20.3 4.8 10.3 0 10 8.1
14.5 77.8
5.1 7.5
3.3-25.8 0.36 61.6-94.0 1.6
0.24 0.74
0-0.91 0 0.02-3.18 13.3
36.6 56.9 41.3 45.0 0 48.2 0.7 80.0 51.3 28.6 34.4 9.4 30.1 28.8 46.8 17.5 0.6 0 20.4 61
14.1 14.5 41.3 20.0 0 15.5 0.7 3.3 28.8 9.2 14.3 5.1 10.1 10.9 11.4 5.0 0.6 0 9.8 7.9
2.25 0.5 1.5 0 0 1.75 0 4 0 0 0 0.38 0.67 4.4 2.33 0.8 0 0 1.11 6.33
1.16 0.5 1.5 0 0 0.88 0 0 0 0 0 0.38 0.42 0.75 1.20 0.49 0 0 0.56 1.09
25.0 0 0 0 0 12.5 0 0 0 0 0 0 0 60.0 33.3 0 0 0 0 83.3
0 0.077 0 0.047 0.35 0.59 2.8
0 0.077 0 0.047 0.13 0.34 0.61
0 0 0 0 0 0 0 0 0
0 0 0 0 0 0
0.7 1.3 0.03 18.2 28.2 29.0 62.0 0 34.2 0 0.8 0 0 7.5 0 65.9
0.6 0.9 0.03 3.6 3.3 6.9 6.6 0 25.9 0 0.8 0 0 0 21.2
0-1.9 0-3.2 0-0.1 11.0-25.4 21.6-34.9 14.7-43.3 47.0-77.0
0 0
0-0 0 0-0.24 0 0-0 0 0-0.14 0 0.08-0.61 1.9 0-1.30 9.1 1.42-4.18 50.0 0 0 0 0 0 0 0 0 0
ATF2 in mesenchymal tumors
283
Table (continued) Diagnosis
n
Nuclear expression
Cytoplasmic expression
Mean score (%) SE Malignant nonmesenchymal tumors n ≥ 10 Carcinoma Melanoma Wilms tumor n b 10 Carcinoid Hepatoblastoma Malignant lymphoma Mesothelioma Normal tissue a b
40 17 23
45.2 73.8 15.1
2 4 3 4 18
1.5 22.5 48.3 30.5 5.7
95% CI
Mean score (IRS) SE
6.2 32.8-57.7 2.10 7.8 57.4-90.3 3.12 4.2 6.5-23.7 1.04 1.5 15.3 24.0 18.6 3.3 0-12.6
4 0 2 1.75 0.33
95% CI
Positivity (%)
0.38 0.63 0.32
1.33-2.87 25.0 1.78-4.45 23.5 0.39-1.70 0
0 0 2 0.63 0.33
100 0 33.3 0 5.6
0-1.04
Translocation-associated sarcoma. Spindle cell sarcoma.
Interestingly, the sarcomas with the highest nuclear ATF2 expression all fall into the categories of TAS and/or spindle cell sarcoma. The activator protein 1 (AP1) transcription factor superfamily including ATF and CREB is known to form homodimers or heterodimers with other AP1 factors to induce transcriptional activities [1]. High nuclear expression of ATF2 may therefore reflect high AP1 family transcription factor activity in these cancers, in which some of the fusion genes retain AP1 family dimerization domains [2]. In addition to TASs, spindle cell sarcomas including gastrointestinal stromal tumor and malignant peripheral nerve sheath tumor showed high ATF2 nuclear expression, suggesting that the AP1 superfamily may also have some role in these tumors. In synovial sarcoma, recent advances in understanding fusion oncoprotein biology have revealed that SS18-SSX forms a complex bridging ATF2 to TLE1, leading to abnormal silencing of ATF2 target genes [3]. ATF2 guides SS18-SSX to its target gene promoters, where TLE1 functions to recruit repressor complexes. In this study, we investigated ATF2 expression in human mesenchymal tumor clinical specimens including 68 synovial sarcomas and found particularly high levels of nuclear ATF2 expression in synovial sarcomas, supporting its fundamental role in the biology of this cancer. TLE1, the other direct partner of SS18-SSX, has now been shown by several groups to be highly expressed at the immunohistochemical level in synovial sarcomas [18,23–26]. TLE1 is particularly sensitive and specific for synovial sarcoma and has become a useful diagnostic marker for this disease. ATF2, although as important to the biology of synovial sarcoma, is unlikely to perform as well as TLE1 as a diagnostic marker because its specificity is lower, possibly reflecting a more widespread role for ATF2 in sarcoma biology. Interestingly, although ATF2 can promote cancer by inducing transcription activity when localized in the nucleus, it can also act as a tumor suppressor by enhancing apoptotic events at the level of the mitochondria when present in the cytoplasm [5]. These contradictory functions of ATF2 have
been supported by immunohistochemical data showing that strong cytoplasmic and weak nuclear staining of ATF2 is associated with better prognosis in melanoma [8]. In our series, almost all mesenchymal tumors showed lower cytoplasmic ATF2 immunoreactivity than in equivalently stained and scored melanomas, with the only exceptions being dedifferentiated liposarcoma and undifferentiated pleomorphic sarcoma. ATF2's biological role in the cytoplasm of mesenchymal tumors has yet to be clarified. Our study has 2 important limitations: the numbers of rare tumors available for assessment and a lack of clinical data allowing assessment of the prognostic value of ATF2. We examined only a few examples for several rare tumor entities because there are more than 100 categories of mesenchymal tumors, many extremely rare. For such rare tumors, further investigation might be necessary to draw firm conclusions about ATF2 expression. ATF2 expression is known to affect the survival of patients with melanoma [8], so ATF2 could be prognostic in mesenchymal tumors as well, the investigation of which would require linkage to patient outcomes. In summary, we have described that TASs and spindle cell sarcomas, including synovial sarcoma, dermatofibrosarcoma protuberans, endometrial stromal sarcoma, gastrointestinal stromal tumor, malignant peripheral nerve sheath tumor, and solitary fibrous tumor, show higher nuclear ATF2 expression than other mesenchymal tumors. This work supports molecular functional data implicating ATF2 in the biology of synovial sarcoma and suggests a potential wider involvement in the oncogenic processes driving molecularly and morphologically related cancers.
Acknowledgments We thank Christine Chow and Doris Gao for their technical assistance. Specimen access was provided through the BC Bone and Soft Tissue Tumour Bank (protocol H08-01717).
284
Fig. 4 Boxplots and histograms showing the difference in nuclear ATF2 expression between benign and malignant mesenchymal tumors (A), or TAS and non-TAS (B), or spindle cell sarcoma and non–spindle cell sarcoma (C). Nuclear ATF2 expression rate is significantly higher in malignant mesenchymal tumor (P b .0001), TAS (P b .0001), and spindle cell sarcoma (P b .0001).
References [1] Lau E, Ronai ZA. ATF2—at the crossroad of nuclear and cytosolic functions. J Cell Sci 2012;125:2815-24. [2] Riggi N, Cironi L, Suva ML, Stamenkovic I. Sarcomas: genetics, signalling, and cellular origins. Part 1: the fellowship of TET. J Pathol 2007;213:4-20. [3] Su L, Sampaio AV, Jones KB, et al. Deconstruction of the SS18-SSX fusion oncoprotein complex: insights into disease etiology and therapeutics. Cancer Cell 2012;21:333-47. [4] Manna PR, Dyson MT, Stocco DM. Role of basic leucine zipper proteins in transcriptional regulation of the steroidogenic acute regulatory protein gene. Mol Cell Endocrinol 2009;302:1-11.
M. Endo et al. [5] Bhoumik A, Ronai Z. ATF2: a transcription factor that elicits oncogenic or tumor suppressor activities. Cell Cycle 2008;7:2341-5. [6] Nagadoi A, Nakazawa K, Uda H, et al. Solution structure of the transactivation domain of ATF-2 comprising a zinc finger-like subdomain and a flexible subdomain. J Mol Biol 1999;287:593-607. [7] Lopez-Bergami P, Lau E, Ronai Z. Emerging roles of ATF2 and the dynamic AP1 network in cancer. Nat Rev Cancer 2010;10:65-76. [8] Berger AJ, Kluger HM, Li N, et al. Subcellular localization of activating transcription factor 2 in melanoma specimens predicts patient survival. Cancer Res 2003;63:8103-7. [9] Gozdecka M, Breitwieser W. The roles of ATF2 (activating transcription factor 2) in tumorigenesis. Biochem Soc Trans 2012;40:230-4. [10] Ronai Z, Yang YM, Fuchs SY, Adler V, Sardana M, Herlyn M. ATF2 confers radiation resistance to human melanoma cells. Oncogene 1998;16:523-31. [11] Ivanov VN, Ronai Z. p38 protects human melanoma cells from UVinduced apoptosis through down-regulation of NF-kappaB activity and Fas expression. Oncogene 2000;19:3003-12. [12] Yang YM, Dolan LR, Ronai Z. Expression of dominant negative CREB reduces resistance to radiation of human melanoma cells. Oncogene 1996;12:2223-33. [13] Nielsen TO, Hsu FD, O'Connell JX, et al. Tissue microarray validation of epidermal growth factor receptor and SALL2 in synovial sarcoma with comparison to tumors of similar histology. Am J Pathol 2003;163: 1449-56. [14] Ng TL, Gown AM, Barry TS, et al. Nuclear beta-catenin in mesenchymal tumors. Mod Pathol 2005;18:68-74. [15] Cheng H, Dodge J, Mehl E, et al. Validation of immature adipogenic status and identification of prognostic biomarkers in myxoid liposarcoma using tissue microarrays. HUM PATHOL 2009;40:1244-51. [16] Steigen SE, Straume B, Turbin D, et al. Clinicopathologic factors and nuclear morphometry as independent prognosticators in KIT-positive gastrointestinal stromal tumors. J Histochem Cytochem 2008;56: 139-45. [17] Pacheco M, Nielsen TO. Histone deacetylase 1 and 2 in mesenchymal tumors. Mod Pathol 2012;25:222-30. [18] Terry J, Saito T, Subramanian S, et al. TLE1 as a diagnostic immunohistochemical marker for synovial sarcoma emerging from gene expression profiling studies. Am J Surg Pathol 2007;31:240-6. [19] Sun Y, Turbin DA, Ling K, et al. Type I gamma phosphatidylinositol phosphate kinase modulates invasion and proliferation and its expression correlates with poor prognosis in breast cancer. Breast Cancer Res 2010;12:R6. [20] Kawai A, Naito N, Yoshida A, et al. Establishment and characterization of a biphasic synovial sarcoma cell line, SYO-1. Cancer Lett 2004;204:105-13. [21] Knippen S, Loning T, Muller V, Schroder C, Janicke F, MildeLangosch K. Expression and prognostic value of activating transcription factor 2 (ATF2) and its phosphorylated form in mammary carcinomas. Anticancer Res 2009;29:183-9. [22] Chen SY, Takeuchi S, Urabe K, et al. Overexpression of phosphorylated-ATF2 and STAT3 in cutaneous angiosarcoma and pyogenic granuloma. J Cutan Pathol 2008;35:722-30. [23] Kosemehmetoglu K, Vrana JA, Folpe AL. TLE1 expression is not specific for synovial sarcoma: a whole section study of 163 soft tissue and bone neoplasms. Mod Pathol 2009;22:872-8. [24] Jagdis A, Rubin BP, Tubbs RR, Pacheco M, Nielsen TO. Prospective evaluation of TLE1 as a diagnostic immunohistochemical marker in synovial sarcoma. Am J Surg Pathol 2009;33:1743-51. [25] Knosel T, Heretsch S, Altendorf-Hofmann A, et al. TLE1 is a robust diagnostic biomarker for synovial sarcomas and correlates with t(X;18): analysis of 319 cases. Eur J Cancer 2010;46:1170-6. [26] Foo WC, Cruise MW, Wick MR, Hornick JL. Immunohistochemical staining for TLE1 distinguishes synovial sarcoma from histologic mimics. Am J Clin Pathol 2011;135:839-44.
