The Ubiquitin-Proteasome Pathway Mediates Gelsolin Protein Downregulation in Pancreatic Cancer

Home , Annexin, Catenin, Gelsolin (cellular), Nestin (protein), Plakophilin

Xiao-Guang Ni,1 Lu Zhou,2 Gui-Qi Wang,1 Shang-Mei Liu,3 Xiao-Feng Bai,4 Fang Liu,5 Maikel P Peppelenbosch,2 and Ping Zhao4

1Department of Endoscopy, Cancer Institute and Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing, China; 2Department of Cell Biology, University Medical Center Groningen, University of Groningen, Groningen, The Netherlands; 3Department of Pathology, 4Department of Abdominal Surgery, and 5State Key Laboratory of Molecular Oncology, Cancer Institute and Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing, China

A well-known observation with respect to cancer biology is that transformed cells display a disturbed cytoskeleton. The under- lying mechanisms, however, remain only partly understood. In an effort to identify possible mechanisms, we compared the proteome of pancreatic cancer with matched normal pancreas and observed diminished protein levels of gelsolin—an actin filament severing and capping protein of crucial importance for maintaining cytoskeletal integrity—in pancreatic cancer. Additionally, pancreatic ductal adenocarcinomas displayed substantially decreased levels of gelsolin as judged by Western blot and immunohistochemical analyses of tissue micoarrays, when compared with cancerous and untransformed tissue from the same patients (P < 0.05). Importantly, no marked downregulation of gelsolin mRNA was observed (P > 0.05), suggesting that post- transcriptional mechanisms mediate low gelsolin protein levels. In apparent agreement, high activity ubiquitin-proteasome pathway in both patient samples and the BxPC-3 pancreatic cancer cell line was detected, and inhibition of the 26s proteasome system quickly restored gelsolin protein levels in the latter cell line. The status of ubiquitinated gelsolin is related to lymph node metastasis of pancreatic cancer. In conclusion, gelsolin levels are actively downregulated in pancreatic cancer and enhanced targeting of gelsolin to the ubiquitin-proteasome pathway is an important contributing factor for this effect. Online address: http://www.molmed.org doi: 10.2119/2008-00020.Ni

INTRODUCTION Among the mechanisms that mediate and even silencing of the gelsolin gene Pancreatic cancer is one of the most deviant cytoskeletal structure in cancer has been documented in many types of virulent malignances with an overall cells is an aberrant expression of gelsolin. human cancers, including bladder, five-year survival rate of only 3–5% and Gelsolin is a Ca2+- and polyphosphoinosi- breast, lung, prostate, gastric, and ovar- a median survival time after diagnosis tide 4, 5-bisphosphate (PIP2)-regulated ian cancers (5–10). Strikingly, forced ex- of < 6 months (1). Despite this immense actin filament severing and capping pression of gelsolin can cause cancer cell clinical problem, pancreatic cancer biol- protein (2). Gelsolin was first isolated morphological reversion and greatly re- ogy remains poorly understood in com- from rabbit lung macrophages as a mod- duce colony-forming ability in vitro, as parison with other cancers. For in- ulator of the cytoplasmic actin gel-sol well as tumorigenicity in vivo in various stance, although it has been recognized transformation (3), and is generally con- tumor types (11). However, downregula- for almost 40 years that transformed sidered one of the most important regu- tion of gelsolin protein levels in pancre- cells display a disturbed actin cyto- lators of the actin cytoskeleton (4). Thus, atic cancer has not been demonstrated skeletal structure, the mechanisms me- gelsolin is an obvious candidate for ex- convincingly, nor have the molecular diating disturbed actin filament organi- plaining altered cytoskeletal reorganiza- mechanisms that could mediate such zation in pancreatic cancer remain tion in pancreatic cancer. In support for downregulation been identified. largely obscure. such a notion, diminished expression In the present study, we provide con- vincing evidence that gelsolin protein levels become severely downregulated in Address correspondence and reprint requests to Ping Zhao, Department of Abdominal pancreatic cancer. This downregulation Surgery, Cancer Institute and Hospital, Chinese Academy of Medical Sciences and differs from most other tumor types be- Peking Union Medical College, NO.17 Panjiayuan, Chaoyang District, P.O. Box 2258, Beijing cause the downregulation is not mirrored 100021, P.R.China.Phone: 86-10-87711782; Fax: 86-10-87711782; E-mail: nixiaoguang@ at the mRNA level, apparently excluding yahoo.com.cn. pretranslational mechanisms. In apparent Submitted February 15, 2008; Accepted for publication June 6, 2008; Epub (www.molmed. agreement, we see that gelsolin is specifi- org) ahead of print June 11, 2008. cally routed into the ubiquitination-

582 | NI ET AL. | MOL MED 14(9-10)582-589, SEPTEMBER-OCTOBER 2008 RESEARCH ARTICLE

proteasome system and that inhibition of treated with 0.1% DMSO as a vehicle Primary data collection and analysis this system, at least in vitro, is sufficient control. were carried out using the GenePix Pro to restore gelsolin protein levels. Thus, 4.0 (Axon Instruments) software pack- downregulation of gelsolin is an intrinsic Antibody Microarray Detection age. The antibody spot was excluded part of the pancreatic cancerous process Procedure and Data Analysis from further analysis if it had a signal: which, unique to this type of cancer, Antibody microarrays (BD Clontech background ratio of < 1.5 in both chan- seems to be partly mediated by ubiquitin- 380, Cat. K1847-1, Lot. 2070683) were nels. For each spot, the median of ratios proteasome machinery. obtained from BD Biosciences Clontech was used in subsequent analysis. The (Palo Alto, CA, USA), and sample prepa- data were imported into the BD Clontech MATERIALS AND METHODS ration and processing procedures were Ab Microarray Analysis Workbook. The performed as described in the antibody relative fluorescence units of C-Cy5/N-Cy3 Patients and Specimens microarray user manual. Briefly, to re- from slide 1 (ratio 1) and N-Cy5/C-Cy3 Fresh tissue samples of 11 pancreatic duce the influence of biological heteroge- from slide 2 (ratio 2) were calculated ductal adenocarcinomas and their corre- neity among tumor samples derived and an internally normalized ratio (INR) sponding distant normal counterparts from different patients, 200 mg from each was obtained by computing the square were obtained at the time of resection of seven pancreatic ductal adenocarci- root of ratio 1/ratio 2. This value repre- with informed consent from Cancer In- noma samples were pooled together and sents the abundance of an antigen in stitute and Hospital (CIH), Chinese pulverized using liquid nitrogen. An pancreatic cancer relative to that of nor- Academy of Medical Sciences (CAMS), equivalent pool was made from matched mal pancreas. The value of INR > 1.5 or and Peking Union Medical College normal pancreas. Total protein was ex- < 0.7 was considered as a significantly (PUMC) between November 2001 and tracted from 200 mg of pooled cancer different protein expression level be- March 2003. The samples were snap- sample or pooled normal pancreas using tween the two samples. frozen in liquid nitrogen before storage extraction/labeling buffer, and then at –80° C. None of them received preop- mixed with Cy3 or Cy5 (Amersham Bio- Measurement of Gelsolin mRNA erative radiotherapy or chemotherapy. sciences, Piscataway, NJ, USA) sepa- Expression by Semiquantitative Formalin-fixed paraffin-embedded tis- rately. To prevent skewing of the array Reverse Transcriptase-PCR sue blocks containing pancreatic cancer results, cancer sample (C) and normal Total RNA was isolated from pancre- and normal pancreatic tissue were col- sample (N) were each split into two atic cancer specimens (n = 9) and their lected from the archives of the Depart- equal portions. Each portion was then la- corresponding distant normal counter- ments of Surgery at CIH, CAMS, and beled with either Cy3 or Cy5 to produce parts and cell line BxPC-3 using RNeasy PUMC between January 1991 and Au- four samples: C-Cy3, C-Cy5, N-Cy3, and MinElute cleanup kit (QIAGEN, Valen- gust 2002, and subjected to tissue mi- N-Cy5. There are two same slides in the cia, CA, USA) according to the manufac- croarray construction. The relevant kit. C-Cy5 (40 µg) and N-Cy3 (40 µg) turer’s instruction. RNA quality was as- ethical committee approved of the inves- were mixed and added to slide 1, and sessed on agarose gel electrophoresis tigations performed. N-Cy5 (40 µg) and C-Cy3 (40 µg) were and spectrophotometric analysis. Re- mixed and added to slide 2. Following verse transcription (RT) reactions were Cell Culture and Drug Treatment 30 min incubation at room temperature, performed on 5 µg of total RNA using Human pancreatic adenocarcinoma the slides were washed, dried, and SuperScript First-Strand synthesis for cell line BxPC-3 was obtained from scanned using the GenePix 4000B Mi- RT-PCR II kit (Invitrogen, Carlsbad, CA, American Type Culture Collection croaray Scanner (Axon Instruments, USA) at 42° C for 80 min, and 0.5–1 µg (Manassas, VA, USA). BxPC-3 cells Union City, CA, USA). There are 768 aliquots of the cDNA then were sub- were cultured in RPMI 1640 medium spots in this antibody microarray. Each jected to RT-PCR. The specific gelsolin supplemented with 10% heat-inacti- antibody is arrayed by duplicate spots primer sequences were used as follows: vated fetal bovine serum, 100 U mL–1 side-by-side on the slide. Besides the an- 5’-CCACTGCGTCGCGGGG-3’ (sense) penicillin, and 100 µg/mL streptomycin tibody capture-dependent spots, there and 5’-GGCAGCCAGCTCAGCCATG-3’ at 37° C in 5% CO2. Specific proteasome are four paired spots pre-labeled with (antisense). The PCR step was performed inhibitor lactacystin (Sigma-Aldrich, St. Cy3 and Cy5 bovine serum albumin using Taq DNA polymerase (Invitrogen). Louis, MO, USA) was dissolved in di- (BSA) located near the outermost corners As an internal control, GAPDH was am- methyl sulfoxide (DMSO). Cells were of the printed area that serve as orienta- plified to ensure cDNA quality and seeded on 100 mm plates. After 70% to tion markers. Two paired BSA spots, quantity for each RT-PCR reaction. Band 90% confluent, cells were treated with not labeled with fluorophore, serve as intensity was quantified using Quantity 10 µM lactacystin in serum-free me- negative controls. Therefore, this slide One 4.4.1 software (Bio-Rad Laboratories dium for the indicated times. Cells were contains 378 different antibodies. Inc, Hercules, CA, USA).

MOL MED 14(9-10)582-589, SEPTEMBER-OCTOBER 2008 | NI ET AL. | 583 UBIQUITINATION-MEDIATED DEGRADATION OF GELSOLIN IN PANCREATIC CANCER

Western Blot enhanced chemiluminescence kit (Santa heated at 100° C in 10 mM pH 6.0 citrate Total tissues (n = 11) and cell lysates Cruz Biotechnology). Parallel Western buffer for 10 min to retrieve antigens. were prepared in extraction buffer con- blot was probed with an anti-α-tubulin After incubation with mouse anti-gelsolin taining 50 mM Tris-HCl (pH 7.4), 150 mM monoclonal antibody (Santa Cruz monoclonal antibody, gelsolin immunos- NaCl, 1% Triton-100, 0.1% SDS, 1 mM Biotechnology) as a loading control. taining was detected by adding biotiny- EDTA, 1 mM AEBSF, 20 µg/mL apro- Band intensity was quantified using lated anti-mouse secondary antibody and tinin, and 20 µg/mL leupeptin. Equal Quantity One 4.4.1 software (Bio-Rad). streptavidin-horseradish peroxidase amounts of total protein (10 µg) were (Zymed Laboratories, South San Fran- separated by 10% SDS-PAGE and trans- Immunohistochemistry cisco, CA, USA). 3, 3’-Diaminobenzidine ferred to Hybond-P polyvinylidene diflu- Pancreatic cancer tissue microarray was was used as a chromogen, and hema- oride (PVDF) membrane (Amersham constructed as previously described (12), toxylin was used for counterstaining. Pharmacia Biotech, Piscataway, NJ, with a total of 256 tissue spots consisting USA). After blocking with 5% nonfat dry of 32 pancreatic ductal adenocarcinomas, Immunoprecipitation milk in PBS with 0.1% Tween-20, mem- 6 mucinous adenocarcinomas, 4 acinar Total tissue and cell extracts were branes were probed with mouse anti- cell carcinomas, 7 islet cell carcinomas, lysed with lysis buffer (25 mM Tris-HCl gelsolin monoclonal antibody (1:1000 8 carcinomas of the ampulla of Vater, [pH 7.5], 150 mM NaCl, 1% NP-40, dilution in PBS; BD Biosciences Pharmin- and 38 normal pancreatic tissues. The 1 mM EDTA, 1 mM sodium orthovana- gen, San Diego, CA, USA), followed by streptavidin-peroxidase method was date) containing protease inhibitors subsequent incubation with horseradish used for the immunostaining of gelsolin. 1 mM PMSF, 1 mM AEBSF, 10 µg/mL peroxidase-conjugated goat anti-mouse Briefly, after deparaffinization in xylene aprotinin, and 10 µg/mL leupeptin. secondary antibody (1:3000 dilution in and rehydration in grade ethanol, en- After centrifugation, the clarified super- PBS, Santa Cruz Biotechnology, Santa dogenous peroxidase activity was blocked natant was collected, and protein con- Cruz, CA, USA). Visualization of the by incubation in 3% hydrogen peroxide centration was determined by Bradford protein bands was performed by the for 10 min. Tissue sections then were method. Whole cell lysate containing

Table 1. Differentially expressed proteins in pancreatic ductal adenocarcinoma compared with normal pancreas detected by antibody microarray.

Upregulated in pancreatic cancer Downregulated in pancreatic cancer Name Function Name Function

UbcH6 (ubiquitin- E2 ubiquitin-conjugating DCC (deleted in colorectal Tumor suppressor gene conjugating enzyme H6) enzyme carcinoma) GABA b R2 (γ-aminobutyric G-protein-coupled HPV-16 L1 (human papillomavirus Human papillomavirus type 16 capsid acid type B receptor, receptor 51 type 16 capsid protein L1) protein L1 subunit 2) Plakophilin 2a Diverse roles in desmosomes RACK1 (receptor for activated Receptor of activated PKC and β-catenin signaling C kinase 1) Inhibitor 2 Inhibit the activity of Gelsolin Actin filament regulatory protein protein phosphatase 1 Nestin Intermediate filament Rabaptin-5 Effector of the small GTPase Rab 5 protein ShcC Shc family protein, DBP2 (DEAD box protein 2) RNA helicase involved in pre-mRNA regulation of Ras pathway splicing PRK2 (protein-kinase The Serine/threonine family IKKα/1 (inhibitor of nuclear factor Serine/threonine kinase, active NF-kα C-related kinase 2) of protein kinase, PKC kappa-B kinase α subunit) subfamily Neurogenin 3 Transcription factor, c-Cbl Proto-oncogene, E3 ubiquitin-protein mediate neuronal ligase differentiation STAT 3 (signal transducer Transcription activator FXR2 (fragile × mental retardation RNA binding protein and activator of syndrome related protein 2) transcription 3) NHE-3 (Na+/H+ exchanger 3) Regulation of Na+/H+ SRP54 (signal recognition particle Route proteins from the cytosol to the ER homeostasis 54 kDa protein)

584 | NI ET AL. | MOL MED 14(9-10)582-589, SEPTEMBER-OCTOBER 2008 RESEARCH ARTICLE

10 µg of protein from each sample was of pancreatic cancer tissue to normal tis- gated, prompting studies into whether prepared for Western blot analysis using sue of the same patients on antibody the observations made using the anti- anti-gelsolin antibody. For gelsolin im- capture-based proteomic chips. As this body capture chips can be confirmed munoprecipitation analysis, 1 ml of cell methodology is strongly biased toward using conventional technology. To de- lysate corresponding to 0.5 mg of total highly expressed proteins, it is particu- tect the alterations of gelsolin mRNA cellular protein first was precleared by larly useful to identify differential ex- and protein expression, the total RNA incubating with protein G-agarose pression of cytoskeletal components, and protein were extracted from the beads (Roche Diagnostics, Mannheim, which is the most abundant class of pro- pancreatic cancer cell line, pancreatic Germany) at 4° C for 1 h. The collected teins in the cellular proteome. Impor- ductal adenocarcinoma tissues, and supernatant was then incubated at 4° C tantly, we did observe significant differ- their corresponding distant normal pan- with 1.5 µg of mouse anti-gelsolin ences in the protein levels of some creatic tissues. Gelsolin mRNA was de- monoclonal antibody or nonimmune cytoskeletal components (Table 1), but termined by RT-PCR method with cyto- mouse IgG (Zhongshan Biotechnology, gelsolin especially was strongly down- plasm gelsolin gene full-length specific Beijing, China) for overnight with rota- regulated in pancreatic cancer. As gel- primers (from 96 bp to 2,360 bp). Ampli- tion. The immune complex was precipi- solin is an obvious candidate to regulate fication products were analyzed by 1% tated by incubation with 30 µLof the aberrant cytoskeletal structure, fur- agarose gel electrophoresis. Gelsolin protein G-agarose for 3 h at 4° C. The ther experiments were initiated to estab- mRNA can be detected in these extracts. agarose beads were pelleted by centrifu- lish whether reduced gelsolin expression No significant difference was observed gation and washed three times with is a bona-fide candidate mechanism for between nine matched pancreatic cancer lysis buffer. The beads were suspended explaining disturbed cytoskeletal organi- tissues and normal pancreas (P > 0.05). in 2×Laemmli sample buffer and boiled zation in pancreatic cancer. Representative electrophoretic analyses for 10 min. Protein G-agarose beads of the RT-PCR products of gelsolin and were removed from the complex by cen- The Expression of Gelsolin mRNA and control GAPDH are shown in Figure 1A trifugation at 10,000g for 5 min. The su- Protein in Pancreatic Cancer Tissues and 1B. pernatant was loaded onto 10% SDS- and Cell Line Importantly, however, gelsolin protein PAGE for Western blot analysis with Asupportive role for reduced gelsolin levels are diminished substantially in monoclonal antibody to gelsolin and levels in pancreatic cancer includes ob- ductal pancreatic cancer. Protein extracts ubiquitin (Santa Cruz Biotechnology). servations that downregulation of gel- from pancreatic tissues and cell cultures solin mRNA, via various genetic mecha- were prepared for Western blot analysis Statistical Analysis nisms, is an important characteristic of using monoclonal anti-gelsolin antibody. Statistical analysis was performed many tumor types. Whether gelsolin This antibody can detect specific bands using the SPSS 11.5 software package mRNA also is downregulated in ductal migrating at 90 kDa. Western blot re- (SPSS, Chicago, IL, USA). The Student pancreatic cancer had not been investi- vealed that 73% (8/11) of pancreatic t test was used to determine the statistical significance of Western blot or RT-PCR band intensity. Values were expressed as mean ± SD of at least three experiments. The chi-square test was used to assess the difference of gelsolin expression among different groups scored by immunohistochemistry. The correlation between gelsolin and each variable was assessed with the Spearman Rank correlation test. P value of less than 0.05 was considered statistically significant.

RESULTS Figure 1. The expression of gelsolin mRNA and protein in pancreatic cancer. (A,C) Repre- sentative RT-PCR gel and Western blot. (B,D) Band intensity was quantified using Quantity Proteomic Chip Analysis One software. The expression of GAPDH and α-tubulin was examined as an internal con- In an effort to obtain insight into the trol. Results are expressed as the mean ± SD of three separate experiments. *P < 0.05, molecular mechanisms that govern aber- compared with their corresponding cancer samples. C, pancreatic ductal adenocarci- rant cytoskeletal structure in pancreatic noma tissues; N, corresponding distant normal pancreatic tissues; BxPC-3, pancreatic cancer, we compared the holo-proteome cancer cell line.

MOL MED 14(9-10)582-589, SEPTEMBER-OCTOBER 2008 | NI ET AL. | 585 UBIQUITINATION-MEDIATED DEGRADATION OF GELSOLIN IN PANCREATIC CANCER

pancreas was 71% (27/38). In contrast, gelsolin expression in pancreatic adenocarcinoma was reduced significantly (χ2 = 14.41, P < 0.0001), and only 29% (12/42) pancreatic adenocarcinomas were scored as positive for gelsolin expression. There were no obvious differ- ences among ductal adenocarcinoma (31%, 10/32), mucinous adenocarcinoma (17%, 1/6), acinar cell carcinoma (25%, 1/4), and islet cell carcinoma (43%, 3/7) (P = 0.758). However, it was found that the positive expression rate of gelsolin in pancreatic ductal adenocarcinomas with lymph node metastasis (71.4%, Figure 2. Immunohistochemical analysis of 5/7) increased markedly (χ2 = 6.28, P = gelsolin expression on pancreatic cancer 0.02) compared with negative lymph Figure 3. Effect of proteasome inhibitor tissue microarray.(A,B) Overview of the H&E lactacystin enhances the gelsolin expres- node types (20%, 5/25). There were no staining and gelsolin immunohistochemical sion. BxPC-3 cells were treated with pro- statistically significant correlations be- staining on pancreatic cancer tissue micro- teasome inhibitor lactacystin, DMSO or array.(C) (200×) and (D) (400×) The gelsolin tween gelsolin expression and histologi- serum-free vehicle control for 12 h, 24 h, or expression in normal pancreas. (E) (200×) cal differentiation of pancreatic ductal 48 h. (A) After incubation, lysates were 2 and (F) (400×) The gelsolin expression in adenocarcinoma (χ = 0.283, P=0.868). prepared and an equal amount of pro- pancreatic ductal adenocarcinoma with- Further experiments were initiated to tein (10 µg) from each lysate was ana- out lymph node metastasis. (G) (200×) and investigate the molecular mechanisms lyzed by Western blot for gelsolin. Arrow at (H) (400×) The gelsolin expression in pancre- mediating downregulation of gelsolin 90 kDA indicates gelsolin. Arrow at 54 kDA atic ductal adenocarcinoma with lymph expression in the absence of reduced indicates α-tubulin. (B) Band intensity was node metastasis. mRNA levels. quantified by Quantity One software. α-Tubulin was examined as an internal control. Results are expressed as the ductal adenocarcinoma tissues showed Effect of Proteasome Inhibitor mean ± SD of three separate experiments. extremely low or undetectable expres- Lactacystin on the Gelsolin Expression *P < 0.05, compared with DMSO controls sions of gelsolin compared with in Pancreatic Cancer Cells of each group treated with lactacystin. matched normal pancreatic tissues (P < An obvious explanation for reduced **P < 0.01, compared with 0 h control of 0.05). There was a weak gelsolin expres- gelsolin levels in pancreatic cancer de- each group treated with lactacystin. sion in pancreatic cancer cell line BxPC-3 spite equal levels of gelsolin mRNA is (Figure 1C,1D). Thus downregulation of that gelsolin is targeted efficiently for gelsolin is a feature of pancreatic cancer, degradation to proteasome in pancreatic Gelsolin Protein Undergoes but this effect seems to involve a non- cancer. To test this possibility directly, we Ubiquitination in Pancreatic Cancer classical post-transcriptional effect. employed the BxPC-3 pancreatic cancer Degradation by the proteasome is cell line as model for in vivo pancreatic thought to be triggered by the polyubiq- The Correlation of Gelsolin Expression cancer that is accessible to pharmacologic uitination of a target protein. To deter- with Clinicopathologic Factors of manipulation by proteasome inhibitors. mine whether the proteasome-mediated Pancreatic Cancer Importantly, we observed that when the degradation of gelsolin is accompanied Downregulation of gelsolin protein BxPC-3 cell line was treated with spe- by increased levels of poly-ubiquitinated was confirmed by immunohistochem- cific 26s proteasome inhibitor lactacystin gelsolin, we investigated the ubiquitina- istry. We carried out immunohistochem- (10 µM), gelsolin expression quickly in- tion status of gelsolin in extracts of pan- ical studies for gelsolin on the paraffin- creased as compared with DMSO vehicle creatic cancer cell BxPC-3 treated in the embedded pancreatic cancer tissue control (P < 0.05), and markedly in- absence or presence of lactacystin (10 µM, microarray (Figure 2). In the normal creased gelsolin levels being evident 12 h treatment). Upon subsequent im- pancreas, there were diffuse strong cyto- from 12 h of treatment onwards (P < munoprecipitation with an anti-gelsolin plasmic immunostaining for gelsolin in 0.01) (Figure 3). Thus it seems that the antibody, followed by Western blot using duct cells and acinar cells, but occasion- proteasome is indeed an important factor monoclonal ubiquitin antibody and gel- ally nuclei appeared to stain. The posi- in pancreatic cancer cells regulating sta- solin antibody, polyubiquitinated gel- tive expression rate of gelsolin in normal bility of gelsolin proteins. solin is clearly visible in BxPC-3 cell ex-

586 | NI ET AL. | MOL MED 14(9-10)582-589, SEPTEMBER-OCTOBER 2008 RESEARCH ARTICLE

Figure 5. Correlation of gelsolin protein levels with pancreatic cancer lymph node metastasis. (A) Representative Western blot analysis of pancreatic cancer tissues with and without lymph node (LN) metastasis. Figure 4. Detection of gelsolin-ubiquitin conjugates. BxPC-3 cells were exposed to lacta- Arrow at 90 kDA indicates gelsolin. Arrow at cystin or DMSO vehicle for 12 h. Equal amounts of whole cell lysates were subjected to 54 kDA indicates -tubulin. (B) Band inten- immunoprecipitation with anti-gelsolin antibody or nonimmune mouse IgG, followed by α sity was quantified using Quantity One soft- Western blot (WB) with anti-gelsolin antibody to detect unmodified gelsolin (A) or with ware. -Tubulin serves as an internal con- anti-ubiquitin antibody to detect gelsolin-ubiquitin conjugates (B). α trol. (C) Gelsolin protein in the pancreatic cancer tissues with and without lymph node metastasis is immunoprecipitated (IP) tracts (Figure 4). Thus, the proteasome- gelsolin protein of the pancreatic cancer by anti-gelsolin antibody, and then Western dependent degradation of gelsolin in this tissues with and without lymph node blot (WB) analyzes the ubiquitin-tagged pancreatic cancer cell is accompanied by metastasis by anti-gelsolin antibody, and gelsolin with anti-ubiquitin antibody.Bracket a high level of ubiquitin ligase activity analyze its ubiquitin status, we find indicates gelsolin-Ub conjugates. toward gelsolin in an experimental sys- downregulation of ubiquitin ligase activ- tem. Subsequently, experiments were ity toward gelsolin in pancreatic cancers initiated to ascertain whether a similar with lymph node metastasis conditions. mechanism is involved. Gelsolin is one mechanism is operative in pancreatic (Figure 5C). This provides further sup- of the most important actin structure- cancer in vivo. port for the notion that gelsolin protein regulating proteins, and its expression in Immunohistochemical results show levels are controlled by the extent of almost all eukaryotic cells is testimony that gelsolin expression in pancreatic ubiquitin ligase activity toward gelsolin to its fundamental importance in main- cancer with lymph node metastasis is in pancreatic cancer. taining an organized actin cytoskeleton higher than that observed in negative (15,16). The physiological functions of lymph node metastasis (see above). In DISCUSSION the 90kDa gelsolin protein are regulated apparent agreement, Western blot analy- One of the most fundamental charac- by the intracellular Ca2+ concentration sis of fresh pancreatic cancer tissue with teristics of malignant and transformed and by its binding to diphosphoinosi- and without lymph node metastasis cells is the aberrant organization of the tides, both of which serve to activate its shows re-expression of gelsolin in the ad- cytoskeleton (13). The mechanisms actin-severing potential, capping the vanced lymph node positive phase of the which mediate transformation-associated barbed ends, and promoting nucleation disease. Levels of gelsolin protein, not cytoskeletal organization are only partly of polymerization (17). In many can- mRNA, in pancreatic cancer with lymph understood (14). Here we present evi- cers, gelsolin downregulation has been node metastasis (n = 3) are higher than dence based on proteomic comparison, identified as a factor mediating aberrant those in negative lymph node metastasis immunohistochemistry, and classical cytoskeletal organization, substantially (n = 8) (P < 0.05) (Figure 5A,5B), corrobo- biochemistry that diminished expres- reduced levels, for instance, in 77.8% rating the immunohistochemical results. sion of gelsolin through a transcription- (14/18) of bladder cancers (5), 70% Strikingly, when we immunoprecipitate independent ubiquitination-dependent (21/30) of breast cancers (6), 55%

MOL MED 14(9-10)582-589, SEPTEMBER-OCTOBER 2008 | NI ET AL. | 587 UBIQUITINATION-MEDIATED DEGRADATION OF GELSOLIN IN PANCREATIC CANCER

(48/68) of non-small cell lung cancers with lymph node negative pancreatic GSN 1 and GSN 2 siRNAs significantly (NSCLC) (7), 68.1% (15/22) of gastric cancers. This result is consistent with an reduced cell motility. In wound-healing cancers (9), and 85% (64/75) of ovarian earlier study of Thomson et al. (18), who assays, both GSN 1 and GSN 2 siRNAs cancers (10). In this study, pancreatic already reported increased gelsolin ex- significantly impaired the ability of cancer also is added to the growing list pression in lymph node positive cancers pancreatic cancer cells to migrate into of malignancies displaying reduced gel- and would suggest that the expression the wounded monolayer (18). These re- solin protein levels, gelsolin expression of the putative gelsolin-specific ubiquitin sults demonstrated that gelsolin might being diminished in 71% (30/42) of the ligase active in the earlier phases of the contribute to the motility of pancreatic cases of pancreatic cancer as compared transformation process is lost at later cancer cells. with matched control tissue. stages. In this respect, pancreatic cancer In conclusion, our results show signifi- Nevertheless, the mechanisms mediat- would resemble gastric cancer, where, cantly lower levels of gelsolin protein in ing diminished gelsolin expression in in earlier stages of the disease, Smad4- the early stages of human pancreatic different cancer cells are not very clear. specific ubiquitin ligases seem impli- adenocarcinoma. This reduction of gel- It has been reported that there were no cated in disease progression, but where solin protein seems dependent on the major mutations, gross rearrangements, their expression is lost at later stages of ubiquitin-proteasome-dependent degra- or deletions in the coding sequence of the disease, possibly as a consequence of dation of gelsolin in the pancreatic can- the gelsolin gene in human breast cancer the increased genetic instability at these cer cell, defining a novel pathway for and NSCLC cells (6,7). Recently, Noske et later stages (19). In apparent agreement, such downregulation in cancer. Study al. found inactivation of gelsolin might Thompson et al. reported that high levels of the correlation between proteasome- be mediated by epigenetic modification of nuclear gelsolin were associated with mediated gelsolin degradation and pan- (methylation and histone deacetylation) reduced patient survival times, which creatic cancer progression should have in human ovarian carcinoma (10). We was an independent prognostic parame- important clinical significance in the di- observed, however, gelsolin protein, but ter for pancreatic cancer (18). Thus, re- agnosis, treatment, and prognosis of not mRNA, was downregulated in pan- emergence of gelsolin expression would human pancreatic and other cancers. creatic cancer. Thus, an alternative non- be associated with the most advanced classical mechanism might mediate this aggressive stages of the disease. ACKNOWLEDGMENTS downregulation. The high activity of In earlier stages of pancreatic cancer, We thank Dr. Bing Guo for the sample ubiquitin-proteasome machinery toward loss of gelsolin expression may well be collection. This work was supported by gelsolin observed in the present study important. It was found that transfection the 10th Five-year National Key Technolo- seems to provide an appealing explana- of the human bladder cancer cell line gies R&D Program (Grant 2004BA703B11) tion for the gelsolin degradation. To test UMUC-2 using exogenous full-length and the National Natural Science Foun- this hypothesis, we use specific protea- gelsolin cDNA inhibited cell growth, dation of China (Grant 30500582). some inhibitor lactacystin to inhibit the reduced the colony-forming ability and ubiquitin-proteasome pathway in the the tumorigenicity in vivo. The cell cycle REFERENCES pancreatic cancer cell line. It was found of bladder cancer cells was arrested or 1. Jemal A, et al. (2007) Cancer statistics, 2007. CA that a proteasome inhibitor could in- delayed at the G2/M phase after these Cancer J. Clin. 57:43–66. 2. Sun HQ, Yamamoto M, Mejillano M, Yin HL. crease the levels of gelsolin protein. Im- cells were transduced with recombinant (1999) Gelsolin, a multifunctional actin regula- munoprecipitation results showed that adenovirus encoding wild-type gelsolin tory protein. J. Biol. Chem. 274:33179–82. gelsolin conjugated with ubiquitin in (Ad-GSN) in vitro. In the orthotopic blad- 3. Yin HL, Stossel TP. (1979) Control of cytoplasmic pancreatic cancer cells. The ubiquitin- der cancer model, the mass of the tumor actin gel-sol transformation by gelsolin, a cal- conjugated forms of gelsolin protein was approximately 90% less in Ad-GSN cium-dependent regulatory protein. Nature 281: 11 583–6. were increased after treatment with pro- treated animals than in controls. Thus, 4. Ono S. (2007) Mechanism of depolymerization teasome inhibitor lactacystin. Taken to- for increasing the size of the transformed and severing of actin filaments and its signifi- gether, these results indicate that gelsolin compartment, lower gelsolin expression cance in cytoskeletal dynamics. Int. Rev. Cytol. protein might be ubiquitinated in pan- may well be an important factor. At later 258:1–82. et al creatic cancer and degraded by the ubiq- stages, re-expression of gelsolin may 5. Tanaka M, . (1995) Gelsolin: a candidate for suppressor of human bladder cancer. Cancer Res. uitin-proteasome pathway. help the tumor with forming metastases. 55:3228–32. Importantly, later in the transforma- In human pancreatic adenocarcinoma 6. Asch HL, et al. (1996) Widespread loss of gelsolin tion sequence, gelsolin expression in- cell lines, RNA interference was used to in breast cancers of humans, mice, and rats. Can- creases again, as suggested by our obser- reduce gelsolin protein levels with two cer Res. 56:4841–5. 7. Dosaka-Akita H, et al. (1998) Frequent loss of gel- vation that lymph node positive gelsolin-targeting small interfering RNAs solin expression in non-small cell lung cancers of pancreatic cancers react stronger toward (siRNAs), GSN 1 and GSN 2. In transloca- heavy smokers. Cancer Res. 58:322–7. anti-gelsolin antibodies as compared tion assays using Boyden-style chambers, 8. Lee HK, Driscoll D, Asch H, Asch B, Zhang PJ.

588 | NI ET AL. | MOL MED 14(9-10)582-589, SEPTEMBER-OCTOBER 2008 RESEARCH ARTICLE

(1999) Downregulated gelsolin expression in hyperplastic and neoplastic lesions of the prostate. Prostate 40:14–9. 9. Kim JH, et al. (2004) Downregulation of gelsolin and retinoic acid receptor beta expression in gastric cancer tissues through histone deacety- lase 1. J. Gastroenterol. Hepatol. 19:218–24. 10. Noske A, et al. (2005) Loss of Gelsolin expression in human ovarian carcinomas. Eur. J. Cancer 41: 461–9. 11. Sazawa A, et al. (2002) Adenovirus mediated gelsolin gene therapy for orthotopic human bladder cancer in nude mice. J. Urol. 168:1182–7. 12. Bai XF, et al. (2004) Overexpression of annexin 1 in pancreatic cancer and its clinical significance. World J. Gastroenterol. 10:1466–70. 13. Yamazaki D, Kurisu S, Takenawa T. (2005) Regu- lation of cancer cell motility through actin reor- ganization. Cancer Sci. 96:379–86. 14. Kedrin D, van Rheenen J, Hernandez L, Condeelis J, Segall JE. (2007) Cell motility and cytoskeletal regulation in invasion and metastasis. J. Mammary Gland Biol. Neoplasia 12:143–52. 15. Silacci P, et al. (2004) Gelsolin superfamily proteins: key regulators of cellular functions. Cell. Mol. Life Sci. 61:2614–23. 16. dos Remedios CG, et al. (2003) Actin binding proteins: regulation of cytoskeletal microfilaments. Physiol. Rev. 83:433–73. 17. McGough AM, Staiger CJ, Min JK, Simonetti KD. (2003) The gelsolin family of actin regulatory proteins: modular structures, versatile functions. FEBS Lett. 552:75–81. 18. Thompson CC, et al. (2007) Pancreatic cancer cells overexpress gelsolin family-capping proteins, which contribute to their cell motility. Gut. 56:95–106. 19. Mishra L, Katuri V, Evans S. (2005) The role of PRAJA and ELF in TGF-beta signaling and gastric cancer. Cancer Biol. Ther. 4:694–9.

MOL MED 14(9-10)582-589, SEPTEMBER-OCTOBER 2008 | NI ET AL. | 589