Chromatin Remodeling Factors and BRM/BRG1 Expression as ... · BRM staining (38.1%; P 0.016). These results support the notion that BRM and BRG1 participate in two distinct chromosome

Featured Article

Chromatin Remodeling Factors and BRM/BRG1 Expression asPrognostic Indicators in Non-Small Cell Lung Cancer

Junya Fukuoka,1,6 Takeshi Fujii,1

Joanna H. Shih,2 Tatiana Dracheva,1

Daoud Meerzaman,1 Audrey Player,1

Kyeong Hong,1 Sharon Settnek,3 Ajay Gupta,3

Kenneth Buetow,1,3 Stephen Hewitt,4

William D. Travis,5 and Jin Jen1

1Laboratory of Population Genetics, 2Biometric Research Branch,3Center for Bioinformatics, 4Laboratory of Pathology, NationalCancer Institute, Bethesda, Maryland; 5Department of Pulmonary andMediastinal Pathology, Armed Forces Institute of Pathology,Washington, D.C.; and 6Department of Pathology, Shiga Universityof Medical Science, Shiga, Japan

ABSTRACTWe immunohistochemically examined 12 core proteins

involved in the chromatin remodeling machinery using atissue microarray composed of 150 lung adenocarcinoma(AD) and 150 squamous cell carcinoma (SCC) cases. Most ofthe proteins showed nuclear staining, whereas some alsoshowed cytoplasmic or membranous staining. When theexpression patterns of all tested antigens were considered,proteins with nuclear staining clustered into two majorgroups. Nuclear signals of BRM, Ini-1, retinoblastoma,mSin3A, HDAC1, and HAT1 clustered together, whereasnuclear signals of BRG1, BAF155, HDAC2, BAF170, andRbAP48 formed a second cluster. Additionally, two thirds ofthe cases on the lung tissue array had follow-up information,and survival analysis was performed for each of the testedproteins. Positive nuclear BRM (N-BRM) staining corre-lated with a favorable prognosis in SCC and AD patientswith a 5 year-survival of 53.5% compared with 32.3% forthose whose tumors were negative for N-BRM (P � 0.015).Furthermore, patients whose tumors stained positive forboth N-BRM and nuclear BRG1 had a 5 year-survival of72% compared with 33.6% (P � 0.013) for those whosetumors were positive for either or negative for both mark-ers. In contrast, membranous BRM (M-BRM) staining cor-related with a poorer prognosis in AD patients with a 5year-survival of 16.7% compared with those without M-

BRM staining (38.1%; P � 0.016). These results support thenotion that BRM and BRG1 participate in two distinctchromosome remodeling complexes that are functionallycomplementary and that the nuclear presence of BRM, itscoexpression with nuclear BRG1, and the altered cellularlocalization of BRM (M-BRM) are useful markers for non-small cell lung cancer prognosis.

INTRODUCTIONLung cancer has the highest cancer incidence and mortality

in the world (1). Even in localized stages, the prognosis ofnon-small cell lung cancer (NSCLC) has remained poor over thelast decades, despite curative surgery. At present, tumor stagingis the strongest prognostic indicator (2, 3), and the 5-yearsurvival rates range from 61% for stage IA disease and 24% forstage IIB disease to �10% for stage III disease (4, 5). Despiteintensive research, little has changed in the understanding andmanagement of NSCLC. Detailed histopathological features andknown clinical prognostic factors have been of limited help inpredicting lung cancer outcome.

Studies have shown that modification of chromatin struc-ture is an essential step in gene regulation primarily mediated bychromatin remodeling proteins. Among these proteins, histoneis known to establish a dynamic molecular interface and play anactive role in the regulation of transcription (6–8). Often, tran-scription is also regulated by other cofactors, and the balance ofchromatin remodeling activities may be crucial to ensure accu-rate responses to developmental or environmental cues and toprevent the transition of normal cells into cancer cells (9).

The SWI/SNF complex is a major complex of ATP-dependent chromatin remodeling factors and contributes to theregulation of gene expression by altering chromatin structure(10–14). SWI/SNF complex, originally identified in yeast, iscomposed of 11 characterized subunits, and SWI2/SNF2 is thecentral ATPase (15, 16). Human SWI/SNF complexes containone of the two core catalytic ATPase subunits, BRM or BRG1(14, 17–19). BRM and BRG1 further diverge in their subunitcompositions and are thought to participate in specialized cel-lular functions (19). Growing genetic and molecular evidenceindicates that subunits of the SWI/SNF complex can act astumor suppressors in human and mice (7, 20–22). Recently,concomitant loss of BRG1 and BRM expression was reported asa poor prognostic indicator in NSCLC (23). Results from bio-chemical and transfection studies also suggest that SWI/SNFmay interact and control the transcriptional activity of a varietyof genes involved in cellular growth and transformation (24–30).

The recent development of tissue microarray (TMA) tech-nique allows for high-throughput molecular analysis of a largeset of paraffin-embedded surgical specimens and makes it pos-sible to effectively evaluate many proteins in a specific biolog-ical pathway (31–35). Here, we focused on ATP-dependentchromatin remodeling and histone-modifying factors by exam-ining their expression status, relationship, and clinical correla-

Received 11/4/03; revised 2/13/04; accepted 4/6/04.The costs of publication of this article were defrayed in part by thepayment of page charges. This article must therefore be hereby markedadvertisement in accordance with 18 U.S.C. Section 1734 solely toindicate this fact.Note: J. Jen is also an associate professor in the Department of Otolar-yngology and the Oncology Center, Johns Hopkins University, Schoolof Medicine. T. Fujii is currently at the Department of LaboratoryMedicine, International Medical Center, Tokyo, Japan.Requests for reprints: Jin Jen, 41 Library Drive, Room D702, Be-thesda, Maryland 20892. Phone: (301) 435-8958; Fax: (301) 435-8963;E-mail: [email protected].

4314 Vol. 10, 4314–4324, July 1, 2004 Clinical Cancer Research

Research. on March 13, 2020. © 2004 American Association for Cancerclincancerres.aacrjournals.org Downloaded from

http://clincancerres.aacrjournals.org/

tion in 300 pulmonary adenocarcinoma (AD) and squamous cellcarcinoma (SCC) cases.

MATERIALS AND METHODSClinical Samples. Three hundred lung cancer cases (150

pulmonary AD and 150 SCC cases) were selected from theArmed Forces Institute of Pathology archive, based on thediagnosis and the quality and quantity of the available tissue onthe paraffin block. Demographic and clinical data were collectedat the time of block acquisition (Table 1). A total of 246 patientshad complete clinical information. Survival time and outcomedata were available for 211 patients. The tumors were stagedaccording to the International Union against Cancer’s tumor-node-metastasis (TNM) classification and histologically sub-typed and graded according to WHO guidelines (36).

Lung TMA. The most representative tumor areas werecarefully selected based on the matched H&E-stained slides andmarked directly on the donor block. The cylindrical tissuesample was cored (diameter, 0.6 mm) from the selected regionin the donor block and extruded directly into the recipient blockwith coordinated wells. One hundred adjacent nondiseased lungtissues from AD and SCC cases, along with 50 nonpulmonarynormal tissues, were included in the same array block (Fig. 1).Multiple 4-�m sections were cut with a microtome using anadhesive-coated tape (Instrumedics). Sections attached with ad-hesive tapes were floated on the water, transferred to chargedglass slides (Fisherbrand, Superfrost/Plus), and allowed to drywhile keeping the tape on the tissue slides. Before H&E orimmunohistochemical staining, adhesive tapes were peeled offthe glass slides immediately after immersing the slide in theTPC solvent (Instrumedics). Every 50th section from each tissuearray block was stained with H&E and examined for histopa-thology and tissue retention. The remaining sections were storedat room temperature until use.

Antibody Selection. The primary selection of the anti-bodies was based on protein participation in chromatin remod-eling machinery, commercial availability of the antibodies, andreliability of the immunohistochemical staining. For the SWI/SNF complex, both central core catalytic subunits (BRG1 andBRM) and other BRG1-associated factors, including BAF170and BAF155, and Ini-1 were selected for this study. HDAC1and HDAC2 are histone deacetylases and were selected because

of their known interaction with the SWI/SNF complex (8, 26).Retinoblastoma (RB) protein is also known to interact withHDAC and SWI/SNF in association with RbAP48 (37). mSin3Ais an evolutionarily conserved corepressor and a regulator of cellgrowth and was selected because it complexes with SWI/SNFcomplex and HDACs (26). Histone acetyltransferase (HAT) 1and Tip60 were also selected as an important protein of HATcomplex. HAT interacts and stabilizes SWI/SNF to bind pro-moter nucleosome (38, 39). AE1/AE3 and PGP9.5 were used tostandardize the staining quality of the tissues on the array. Thedetailed source information and optimal staining conditions forall antibodies are listed in Table 2.

Immunohistochemical Staining. Immediately after re-moving the adhesive tapes, the sections were deparaffinizedwith xylene and rehydrated through graded alcohols into water.Heat-induced epitope retrieval was performed by placing slidesin plastic coplin jars in citrate buffer (pH 6.0; BioGenex, San-down, NH) using a decloaking chamber (Biocare Medical, Wal-nut Creek, CA) for about 30 min. After heat-induced epitoperetrieval treatment, endogenous peroxidase was blocked with0.3% H2O2 in PBS for 20 min followed by washing twice inPBS. When a microwave was used for heat-induced epitoperetrieval instead of the decloaking chamber, specimens wereheated in a microwave oven for 4 min while keeping the buffertemperature around 95°C. Tissues were placed in the humidchamber and incubated with normal blocking serum for 30 min.The slides were then incubated with the primary antibodies at anoptimized titer as shown in Table 2 and diluted using Universalblocking reagent (BioGenex) for 60 min. After washing threetimes with PBS, each series of sections was incubated for 30min with biotinylated secondary antibodies (Vector Laborato-ries) diluted to 1:250 by Universal blocking reagent, washedthree times in PBS, and then incubated for 45 min with avidin-biotin complex method reagent (Vectastain Elite ABC kit; Vec-tor Laboratories). The reaction products were rinsed twice withPBS, placed in 0.05 M Tris-HCl buffer (pH 7.5) for 5 min, andthen developed in liquid 3,3�-diaminobenzidine (DAKO) for 3min. After the development, sections were washed twice withdistilled water, lightly counterstained with Mayer’s hematoxy-lin, dehydrated, cleared, and mounted with resinous mountingmedium. All procedures were carried out at room temperature.

Table 1 Clinical characteristics of all non-small cell lung cancer patients used in this study

Variable Category No. of patients (n � 230) Median survival (yrs)

Gender Male 166 2.8Female 64 4.6

Mean age � SD (yrs) 64.6 � 9.2 (36–86)Stage I 115 ] 4.2II 29

III 30 ] 1.3IV 7Vital status Alive 52 7.3a

Dead [ Tumor related 46 1.6Other cause 20 1.7Unknown 80 2.8

Median follow-up (yrs) 3.2 � 3.2a Median term of last follow-up.

4315Clinical Cancer Research



Western Analysis and Immunohistochemistry of LungCancer Cell Lines. To determine the specificity of the anti-bodies, a panel of nine lung cancer cell lines was obtained fromAmerican Type Culture Collection (Manassas, VA) and used forboth immunohistochemistry and Western analyses.1 Briefly,confluent cultures of each cell line were prepared for total celllysate in SDS-PAGE sample buffer or scraped to obtain the cell

pellet and fixed in 4% paraformaldehyde. The pellets were thentransferred into 15-ml tubes, dehydrated, and embedded in par-affin. The paraffin blocks were further processed and sectionedfor immunohistochemical staining using the protocols describedin Table 2. For Western analyses, cell lysates were sonicatedand centrifuged to remove the insoluble material, heated at 95°Cfor 5 min, and stored at �80°C. The protein quantity for eachlysate was adjusted based on the �-tubulin signal, and theproteins of interest were analyzed using standard protocols (40).The protein filter was incubated for 1 h at room temperaturewith anti-BRM (n19), BRG1, BAF155, and Ini-1 antibodies atdilutions of 1:20, 1:200, 1:200, 1:200, respectively. Immuno-

1 J. Fukuoka, A unified approach to standardize immunohistochemicalstaining using paraffin embedded cell arrays and Western blot, manu-script in preparation.

Fig. 1 Organization of lung tumortissue microarray (TMA). A, H&Estaining of the entire TMA section.B, scheme of TMA. Tissues samplesare represented by small rectanglesthat are arrayed in groups of 25 each.Tumor types are indicated as light gray(adenocarcinoma) or black (squamouscell carcinoma). The clear areas arecontrol tissues including normal lungfrom patients with adenocarcinoma(#1), or squamous cell carcinoma(#2), and normal nonpulmonary or-gans (#3).

4316 Chromatin Remodeling Factors in NSCLC



blots were washed with TBS [50 mM Tris (pH 7.6) and 0.15 M

NaCl] containing 0.1% Tween 20 for 15 min, incubated withhorseradish peroxidase-conjugated secondary antibody (diluted1:10,000), and visualized using enhanced chemiluminescenceWestern blotting detection reagents (Amersham Biosciences,Buckinghamshire, United Kingdom).

TMA Immunohistochemistry Scoring. The optimizedstaining condition for lung tumor microarray was determinedbased on the coexistence of both positive and negative cells inthe same tissue sample. Signals were considered positive whenreaction products were localized in the expected cellular com-ponent. The criteria for the staining were scored as follows:distribution score was scored as 0 (0%), 1 (1–50%), and 2(51–100%) to indicate the percentage of positive cells in alltumor cells present in one tissue. The intensity of the signal(intensity score) was scored as 0 (no signal), 1 (weak), 2(moderate), and 3 (marked). The total of distribution score andintensity score was then summed into a total score (TS) of TS0(sum � 0), TS1 (sum � 2), TS2 (sum � 3), and TS3 (sum �4–5). This scoring system was modified from previous conven-tions due to the smaller size of the tissue samples on TMA (41).Throughout this study, TS0 or TS1 was regarded as negative,whereas TS2 or TS3 was regarded as positive.

For each antibody staining, specimens with questionablesignals due to insufficient tumor cells, tumor cells that weredifficult to distinguish from other inflammatory or interstitialcells, or tissues with homogenously weak signals were excludedfrom further evaluation. Tissue sections with only necrotic tu-mor cells or tissues with high homogenous background werealso excluded. Twenty-seven cases with an intensity score of �3by all tested antibodies were excluded from all analysis. Fur-thermore, due to the small size of the tissues in TMA, somepositive samples could be scored as negative due to samplingerror. In our study, 16 samples showed no AE1/AE3 signal(TSO) in 262 scorable tumors, whereas 9 samples were TS1,despite the fact that all cancer tissues should be positive (42).Therefore, the percentage of positive cases underscored wasestimated to be (16 � 9)/(262 � 16 � 9) 100 � 10.5% of thepositive samples. This estimate is also consistent with the stain-ing status of PGP9.5, which was positive in 50% of TMA

samples (127 of 253 samples) and 54% in NSCLC cases (54 of98) when analyzed using conventional pathological sections(43).

Statistical Analysis. Comparisons of clinicopathologicalfactors (i.e., age at diagnosis, gender, histological type, or path-ological stage) between positive and negative groups were per-formed using the 2 test. Because a relatively large number ofpatients in our cohort died of unknown causes, only overallsurvival was considered for each antibody staining and prog-nostic analyses. The log-rank test was used for comparingsurvival distributions between positive and negative groups ofstaining, and Kaplan-Meier curves were plotted for the twogroups. Clinical factors were accounted for by fitting Cox pro-portional hazard models. When the number of observations in agroup is small (�30), the usual P value of the log-rank testrequiring a large sample size may not be valid. Consequently, apermutation test was used to evaluate the significance of thedifference between survival curves. Specifically, the associationbetween the staining status and survival outcome was randomlypermuted. A log-rank test statistic was generated with the per-muted data. This procedure was repeated 2000 times, and the Pvalue was the proportion of permutations with the same or largervalue of the log-rank test statistic as observed in the actualstudy. A similar permutation test for the Cox proportional haz-ards model was used, where the association between stainingstatus and survival outcome along with the clinical factors wasrandomly permuted. Association of binary staining status be-tween each pair of antibodies was examined using the 2 test,and the association was considered significant if the P value�0.01. Association between antibody staining and clinical fac-tors (gender, age at diagnosis, or histological type) was alsoanalyzed using the 2 test, where P � 0.01 was consideredsignificant.

For protein association analyses, hierarchical agglomera-tive clustering was applied to the total score of staining usingaverage linkage and 1 minus the Spearman rank correlation asthe distance metric. Clustering reproducibility was assessed bythe method proposed by McShane et al. (44)

Image Database of Lung TMAs. To facilitate data ex-change and sharing, all immunohistochemistry images along

Table 2 Antibodies and protocols used for immunohistochemical staining

Antibody Species Source Clone Type Protocol Titration Retrieval

HDAC1 Ra Santa Cruz H-51 IgG2a ABC 50 DCHDAC2 R Santa Cruz H-54 IgG ABC 200 MWBRG1 R Santa Cruz H-88 IgG ABC 200 DCBRM G Santa Cruz C-20 IgG ABC 100 DCBRM G Santa Cruz N-19 IgG ABC 100 DCBAF170 R Santa Cruz H-116 IgG ABC 100 MWBAF155 R Santa Cruz H-76 IgG ABC 100 MWRbAp48 G Santa Cruz K-15 IgG ABC 300 MWRB M Zymed RB-1 IgG1, � ABC 100 DCIni1 R Santa Cruz H-300 IgG ABC 50 DCmSin3A R Santa Cruz AK-11 IgG ABC 50 DCHAT1 G Santa Cruz C-20 IgG ABC 1000 DCTip60 G Santa Cruz K-17 IgG ABC 100 DCAE1/AE3 M DAKO AE1/AE3 IgG1 ABC 100 NonePGP9.5 R Biogenesis n/a n/a ABC 8000 DCa R, rabbit; DC, Decloaking Chamber; MW, microwave; G, goat; M, mouse; n/a, not available.




with the associated clinicopathological information of the tu-mors are available at the LPG Image Portal.2 LPG Image wasdeveloped leveraging the National Cancer Institute Center forBioinformatics cancer Images Portal (caIMAGE), which allowsresearchers to submit and retrieve histology images and anno-tations. The database is highly interactive and can be searched

based on patient’s gender, age, tumor histology, and experimen-tal findings for each of the antibodies described in this study.

RESULTSClinicopathological Variables and Their Associations

with Patient Survival. Overall, 273 cases had adequatelystained tissues for analysis, and clinical information was avail-able for 230 cases. Among them, age at diagnosis was availablefor 229 cases, with an average of 64.6 years (range, 36–86years). Gender was available for 230 cases and included 642 http://cancerimages.nci.nih.gov/caIMAGE/index.jsp.

Fig. 2 Chromatin remodelingfactors in lung cancer cell lines.A, immunoblotting of BRG1,BRM, BAF155, and Ini-1 amongnine lung cancer cell lines. TheEKVX cell line showed a bandreactive to anti-BRM but non-immunoreactive for anti-BRG1antibody. The apparent sizes ofBRG1 and BRM were a littlelarger than the known molecu-lar size (180 kDa) but are thesame as those shown in the datasheet from the supplier (http://www.scbt.com/). BAF155 andIni-1 showed an immunoreac-tive band in eight cell lines.H358 appeared to be negativefor Ini-1. B, anti-BRM immu-nohistochemical staining of thesame lung cancer cells. Hop92did not show positive stainingby immunohistochemistry butshowed a weak positive bandon Western blot. Consistentwith immunoblotting, immuno-histochemical signals of BRMwere lower than those of BRG1in lung cancer cell lines. H520,EKVX, and H2170 showedmembranous and cytoplasmicstaining, whereas H358 andH226 showed weak and focalnuclear staining.




female and 166 male patients. Pathological stage was availablefor 181 cases. Follow-up time or time to death was available in196 cases, with a median follow-up of 3.2 years. Fifty twopatients were alive at the last follow-up.

Female patients had a better prognosis in overall survivalwith a median survival of 4.6 years in contrast to a mediansurvival of 2.8 years for males (P � 0.028). As expected, stageI and II patients had a significantly better prognosis thanstage III and IV patients (P � 0.001). The median survival forstage I/II and stage III/IV patients was 4.2 and 1.3 years,respectively. There was no significant survival difference be-tween patients with AD and those with SCC. The essentialclinical information of the study cohort is summarized inTable 1.

Specificity of the Antibodies. We compared the proteinstaining of selected antibodies by Western and immunohisto-chemical analyses using a panel of nine lung cancer cell lines.BRG1 expression was observed in five of nine cell lines,whereas BRM expression was detectable in six of the nine celllines, although at much lower levels than BRG1 (Fig. 2A).Furthermore, five of these six cell lines (H520, H358, EKVX,H226, and H2170) had positive immunohistochemical staining,and three of these cell lines (H520, EKVX, and H2170) showedmembranous and cytoplasmic staining with no nuclear staining(Fig. 2B). BAF155 showed reactive band in all nine cell lines,whereas one cell line (H266) had a weaker band. Ini-1 expres-sion was generally weak but detectable in eight cells and ap-peared to be absent in H358 cells (Fig. 2A).

Expression of Proteins as Detected by Immunohisto-chemistry. A summary of the staining features for each anti-body is listed in Table 3. As expected, the staining of mostproteins including RB, HAT1, HDAC1, HDAC2, RbAP48,BAF170, and mSin3A was located exclusively in the tumor cellnucleus. However, BAF155, Ini-1 BRG1, and BRM showedboth nuclear and cytoplasmic or nuclear and membranous stain-ing (Table 3; examples in Fig. 3). Only Tip60 showed mostlycytoplasmic staining. Expression of HDAC2, nuclear BRM (N-BRM), RB, and nuclear BAF155 was significantly higher inSCC than in AD (P values are 0.001, �0.001, 0.005, and �.001,respectively). No protein was observed more frequently in ADthan SCC. No significant staining differences were observedbetween gender, age (�60 years versus �60 years), or stage(I/II versus III/IV). In the normal lung, all bronchiolar epitheliashowed positive nuclear signal for all tested antibodies, except

Ini-1, which showed both nuclear and cytoplasmic staining.Tip60 showed only cytoplasmic staining. For pneumocytes,BRG1 and Ini-1 were positive for most cells; BAF155, BAF170,and HDAC2 were weak to moderately positive in less than halfof cells; whereas the remaining antibodies were moderatelypositive in more than half of the pneumocytes.

Patterns of Association between Antibodies. We ex-amined the association of staining status between each pair ofthe tested proteins using 2 test. Among a total of 128 compar-isons(16 16/2), we observed 12 and 11 significant associa-tions in AD and SCC, respectively. Of these associations, sixwere common for both SCC and AD and included associationsbetween HDAC1 and RB, mSin3A, and HAT1; mSin3A andnuclear Ini-1 and HAT1; and HDAC2 and nuclear BAF155.Among the AD cases, staining of BAF170 was positively asso-ciated with HDAC2, nuclear BRG1 (N-BRG1), and RbAP48;N-BRG1 was also positively associated with RbAP48, whereasHDAC1 expression was correlated with nuclear Ini-1. For SCC,N-BRG1 staining was positively associated with HDAC2, nu-clear BAF155, and Tip60, whereas staining of the RB proteinwas correlated with mSin3A and HAT1 (Fig. 4A).

To gain insight into the potential interaction of the proteins,we organized the expression patterns of all tested antibodiesusing unsupervised clustering analysis (44). Two major groupsof nuclear protein staining were observed (Fig. 4, B and C):nuclear staining of BRM, Ini-1, RB, mSin3A, HDAC1, andHAT1 clustered as one group; whereas nuclear staining ofBAF155, HDAC2, BAF170, RbAP48, and BRG1 clustered as asecond group. Application of the reproducibility measures ofMcShane et al. (44) showed that these two groups of proteinassociations were highly reproducible (robustness index �90%). All of the nonnuclear signals branched outside these twoclusters further support the reliability of the nuclear proteinassociations (Fig. 4C).

Prognostic Significance. When the expression of eachof the 12 proteins was associated with clinical survival, onlyBRM staining demonstrated prognostic significance. PositiveN-BRM staining was observed in 51 cases among 193 caseswith survival outcomes (Fig. 5A). Kaplan-Meier analysis indi-cated that the overall 5-year survival for patients whose tumorswere N-BRM positive was 53.5% compared with 32.3% forthose whose tumors were negative for N-BRM (P � 0.015). Incontrast, M-BRM staining was observed in 30 of 268 cases, andonly four of these tumors also had nuclear staining. Of the 18

Table 3 Immunohistochemical staining of chromatin remodeling factors using TMAa

The actual number of cases is shown for each scoring, whereas percentages are given for � and � cells. N, nuclear, C, cytoplasmic, M,membranous. Total scores 2 and 3 were designated as positive, whereas scores 0 and 1 were recognized as negative.

Totalscore

BRG1 BRM BAF155 Ini-1 BAF170 HDAC1 HDAC2 RbAP48 RB mSin3A HAT1 Tip60

N C N M N C N C N N N N N N N C

0 74 221 156 222 122 162 135 148 43 73 110 55 71 72 27 841 44 4 45 16 51 21 17 19 25 20 28 37 45 9 38 362 77 7 38 15 44 69 73 53 60 75 46 78 78 60 54 643 50 13 29 15 30 21 25 28 122 70 63 71 68 100 127 54

� 48 92 74 89 70 67 58 47 27 36 54 35 44 29 37 37� 52 8 26 11 30 33 42 33 73 64 46 65 56 71 63 63

a TMA, tissue microarray.




cases with survival data, M-BRM correlated with a poorersurvival in AD with a 5-year survival of 16.7% for those whosetumors were membrane positive compared with 38% for thosewhose tumors were negative for M-BRM staining (P � 0.016).The number of the M-BRM-positive cases (six) with SCC didnot allow meaningful statistical evaluation.

Previous study has shown that concomitant loss of BRG1

and BRM was a poor prognostic indicator (23). Consistent withthis observation, our study also showed that patients whosetumors were positive for both nuclear BRG1 and BRM (n � 15)had a 5-year survival of 72% compared with 33.6% for the restof the cases (P � 0.013). However, after adjusting for genderand stage, the P value was 0.068 for N-BRM and 0.072 forN-BRM/N-BRG1. In contrast, the predictive value of M-BRM

Fig. 3 Examples of immunohistochemical staining by BRG1 and BRM antibodies in primary lung tissues. A, BRG1 in normal lung. Most of the cellsincluding pneumocytes and interstitial cells were positively stained in the nucleus. B, a lung tumor negative for BRG1 [TS0]. C, a tumor positive forBRG1 (TS3). D, BRM in normal lung. Most of the cells including pneumocytes and other interstitial cells are positive in the nucleus. E, tumor withnegative BRM staining (TS0); the surrounding inflammatory cells were positive. F, tumor positive for BRM (TS3). G and H, membranous BRMstaining in tumor cells. I, high magnifications (400) of images shown in G and H. Diameter of tissue core � 0.6 mm.




remained strong after adjustments, with a P value of 0.0065. Theassociated Kaplan-Meier curves are shown in Fig. 5B.

DISCUSSIONChromatin remodeling factors have been the subject of

strong interest in human cancers. However, little is known abouttheir role in lung cancer. Here, we investigated the expression of12 chromatin remodeling factors in NSCLC by immunohisto-chemical staining of a TMA containing 300 primary tumorsamples.

Association between Antibodies. Chromatin remodel-ing factors form complexes, and SWI/SNF is one of the mostwell-examined complexes (8, 10). In our study, we observedmany statistically significant associations between pairwisecomparisons of protein expression. Among these associations,about 50% were shared between AD and SCC. The remaining50% were unique to each tumor type.

The SWI/SNF complex is known to be composed of onecatalytic ATPase containing core protein (either BRG1 or BRM)and the other BRG1-binding proteins, i.e., BAF170, BAF155,Ini-1, BAF47, BAF60a, BAF60b, BAF60c, BAF250, and

BAF50 (10, 17, 19). It is also known to associate with someHDACs, such as HDAC1 and HDAC2, RbAP48, and RB (7, 10,45, 46). There are increasing numbers of associated proteinsknown to interact with this complex. Because the role of SWI/SNF complex in transcription is broad, variation of the associ-ated proteins in the complex is expected for different cellularfunctions. Among the associations observed in the current study,association between nuclear signals of BRG1 and HDAC2,BAF170, nuclear BAF155, and RbAP48, as well as HDAC1 andmSin3A, Ini-1, and RB have already been reported (10, 26,45–47). Using 2 test and clustering analysis, we showed thatthere were two distinct groups among these proteins (Fig. 4, Band C). Indeed, most of the known interactions were consistentwith our 2 test. Significantly, correlations between RB andHAT1 and mSin3A seen in this study have not been reportedpreviously. HAT complexes and SWI/SNF complexes interactwith each other for transcription. To stabilize the SWI/SNF onthe promoter region, HAT complexes are thought to increase theretention of SWI/SNF on the promoter (38). However, our studyonly revealed the coexistence of potentially interacting proteinsand cannot address whether these associations are due to direct

Fig. 4 Association of staining between each pair of antibodies. A, P values of 2 test between each antibody are shown for adenocarcinoma (grayarea) and squamous cell carcinoma (white area). P � 0.01 is considered significant and is indicated in the table. B, one-way clustering indicatingthe relatedness of expression patterns among all antibodies. Note that all nonnuclear signals clustered distantly from the nuclear signals of the sameantibody. C, identification of two highly correlated sets of chromatin remodeling proteins in NSCLC.




interaction within one complex or merely coexist in the samecell.

Prognostic Significance of BRM and BRG1. BRG1and BRM are two highly homologous ATPases that are func-tionally equivalent and form the core of the mammalian SWI/SNF chromatin remodeling complex (17). Recently, concomi-tant loss of BRG1 and BRM was reported as a poor prognosticindicator by Reisman et al. (23). In their study, 6 cases that werenegative for both BRG1 and BRM had a poorer survival ratecompared with the other 54 NSCLC cases (41 AD and 19 SCCcases). The antibody used could not distinguish the two proteins.

Here, we used antibodies that appeared to be specific for BRG1and BRM, respectively, based on Western analyses and immu-nohistochemistry (Figs. 3A and 5A). Expression of BRM in lungcancer cells was generally low but detectable as a single band atthe expected size and showed nonoverlapping expression withBRG1 by immunohistochemistry analysis. In lung cancer celllines, three of five immunopositive cell lines had membranousstaining, whereas two others showed only nuclear staining (Fig.3B). Taken together, these findings suggest the following: first,the anti-BRM antibody used in this study did not cross-reactwith BRG1; second, both membranous and nuclear staining

Fig. 5 Immunohistochemistry scoring and Kaplan-Meier survival analyses for BRG1 and BRM in non-small cell lung cancer. A, immunohisto-chemical scoring for nuclear BRG1, nuclear BRM, and membranous BRM signals. Grids corresponded to tumor arrangements on the TMA as shownin Fig. 1B (gray and black blocks). Total scores (TSs) are shown, and the colors corresponded to the TS values. Red, TS3; yellow, TS2; white,TS1/TS0; black, excluded or missing tissues. B, Kaplan-Meier survival plots. The overall patient survival status for antibodies and selected cellularlocations are shown. The number of patients in each arm and the associated P values are as indicated.




represented the same BRM antigen; and third, membranousstaining of BRM is present in primary lung cancers and may bemore common in lung cancer cell lines.

We also compared all combinations of BRG1 and BRMstaining with survival. Consistent with those reported by Reis-man et al. (23), coexpression of N-BRG1/N-BRM correlatedwith a trend of better prognosis than any other combinations ofthese two markers. Significantly, BRG1 and BRM belonged todifferent clusters based on their expression status among thetested tumors in our study. Kadam and Emerson (48) showedthat BRG binds to zinc finger proteins though its NH2-terminaldomain and is not present in BRM complex, whereas BRMinteracts with two ankylin repeat proteins that are a part of theNotch signal transduction pathway. Our results are consistentwith the knowledge that those two proteins do not exist in thesame complex (26, 49) and indicate that BRG and BRM mayparticipate in distinct but complementary cellular processes thatare required for cell regulation.

In our study, 30 of 268 NSCLC cases showed a distinctivemembranous BRM signal, whereas only 4 of those had coex-pression of nuclear signal as well. Staining with COOH termi-nus and NH2 terminus anti-BRM antibody (clone C-20 andN-19; Santa Cruz Biotechnology) showed a consistent membra-nous staining pattern (30 of 268 versus 26 of 263 countablesamples). The correlation efficiency P value between two anti-bodies is 10�14. In our study, N-BRM was associated withbetter prognosis in both SCC and AD tumors, whereas M-BRMwas associated with a poorer prognosis only in AD. Althoughthe exact mechanism is unclear, this difference in BRM cellularlocation may reflect the abnormal translocation of BRM in thecancer or indicate that BRM also participates in cellular pro-cesses other than chromatin remodeling. Because concomitantexpression of N-BRG1/N-BRM correlated with a better prog-nosis, we reasoned that M-BRM status may be negatively cor-related with N-BRG1 as well. However, 15 of 28 M-BRM-positive tumors were also N-BRG1 positive. Therefore, thischange of localization more likely reflects BRM alterations inlung cancer cells independent of BRG1.

Antibody Specificity and Staining Quality Control. Itis worth noting that tissue samples placed on TMA representonly a small portion of the whole tumor. Some samples willdemonstrate negative staining due to tissue sampling or heter-ogeneity. Furthermore, some of the antibodies used here havenot been extensively studied by immunohistochemical methods.To ensure specificity, we used Western analysis and immuno-histochemistry on lung cancer cell lines to evaluate and optimizethe tissue staining conditions. We also relied on well-establishedantigens, AE1/AE3 and PGP9.5, to evaluate the true positiverate of each antibody and used stringent criteria to excludesamples with uniformly weak stainings. In our pilot tests, weestimated that a number approximately equal to 10% of thepositive samples may be falsely negative for each staining.Therefore, antibodies with a high percentage of positivelystained samples will have a higher amount of false negativecases, making it less likely to identify significant clinical asso-ciation. On the other hand, although each chromatin remodelingfactor is expected to be present in the majority of cancer cells,all proteins investigated here have portions of negatively stainedcases that far exceed the estimated number of false negative

tumors (Table 3). Some of the negative staining reported heremay represent a down-regulation of these proteins in the tumorsor the lower sensitivity of the antibody. Our consideration ofTS1 staining as negative may also have contributed to the lowerestimation of antibody staining reported. For instance, our im-munoblotting data showed that BRG1 was detected in a similarproportion of lung cancer cells lines as observed in TMA,whereas BAF155 and Ini-1 were only positive in �50% of theprimary tumors compared with the data on lung cancer celllines. Furthermore, because the tumor blocks were collectedfrom pathologists worldwide, the quality and processing historyof the blocks could vary greatly, and the clinical pathologicaland survival information for each case are less than perfect.These factors may also have contributed to the marginal statis-tical significance observed for some proteins, particularly afteradjusting for other clinical factors.

Nonetheless, our study is one of the largest series of lungcancer analyses using an immunohistochemical method to ex-amine the expression status of chromatin remodeling factors(35). We showed that BRM appeared to be a central proteinwhose cellular location and staining status appeared to correlatewith overall survival in NSCLC patients. Furthermore, we pro-pose that BRM may participate in distinct cellular processes thatcontrol growth, depending on the type of lung cancer cells. Wealso showed that the expression patterns of the chromosomeremodeling factors branched into two distinct and reproducibleclusters and argue that these clusterings may correspond to thetrue biological interactions of these factors. Clearly, the unique-ness and biological significance of these associations will haveto rely on further analyses in other tissues and tumor types aswell as validations by biological and biochemical assays. Theprognostic trend of BRM reported here will also await confir-mation using cohorts with more complete clinical pathologicalinformation.

ACKNOWLEDGMENTSWe thank Drs. Michael Emmert-Buck and John Gillespie for

critical comments regarding this study and Anne Wenzel and CraigHicks for editing the manuscript. We also express our gratitude to thepathology follow-up division at the Armed Forces Institute of Pathologyfor making follow-up and clinical information available for this study.

REFERENCES1. Parkin DM, Pisani P, Ferlay J. Global cancer statistics. CA-CancerJ Clin 1999;49:33–64.2. Colby TV, Corrin B, Shimosato Y, Brambilla E, Travis WD, editors.Histological typing of lung and pleural tumors. Berlin: Springer; 1999.3. Junker K. Prognostic factors in stage I/II non-small cell lung cancer.Lung Cancer 2001;33(Suppl 1):S17–24.4. Travis WD, Travis LB, Devesa SS. Lung cancer. Cancer 1995;75:191–202.5. Mountain CF. The international system for staging lung cancer.Semin Surg Oncol 2000;18:106–15.6. Wade PA. Transcriptional control at regulatory checkpoints by his-tone deacetylases: molecular connections between cancer and chroma-tin. Hum Mol Genet 2001;10:693–8.7. Neely KE, Workman JL. The complexity of chromatin remodelingand its links to cancer. Biochim Biophys Acta 2002;1603:19–29.8. Klochendler-Yeivin A, Muchardt C, Yaniv M. SWI/SNF chromatinremodeling and cancer. Curr Opin Genet Dev 2002;12:73–9.




9. Davis PK, Brackmann RK. Chromatin remodeling and cancer. Can-cer Biol Ther 2003;2:22–9.10. Narlikar GJ, Fan HY, Kingston RE. Cooperation between com-plexes that regulate chromatin structure and transcription. Cell 2002;108:475–87.11. Langst G, Becker PB. Nucleosome mobilization and positioning byISWI-containing chromatin-remodeling factors. J Cell Sci 2001;114:2561–8.12. Kingston RE, Narlikar GJ. ATP-dependent remodeling and acety-lation as regulators of chromatin fluidity. Genes Dev 1999;13:2339–52.13. Burns LG, Peterson CL. The yeast SWI-SNF complex facilitatesbinding of a transcriptional activator to nucleosomal sites in vivo. MolCell Biol 1997;17:4811–9.14. Wu C. Chromatin remodeling and the control of gene expression.J Biol Chem 1997;272:28171–4.15. Breeden L, Nasmyth K. Cell cycle control of the yeast HO gene:cis- and trans-acting regulators. Cell 1987;48:389–97.16. Recht J, Osley MA. Mutations in both the structured domain andN-terminus of histone H2B bypass the requirement for Swi-Snf in yeast.EMBO J 1999;18:229–40.17. Phelan ML, Sif S, Narlikar GJ, Kingston RE. Reconstitution of acore chromatin remodeling complex from SWI/SNF subunits. Mol Cell1999;3:247–53.18. Tyler JK, Kadonaga JT. The “dark side” of chromatin remodeling:repressive effects on transcription. Cell 1999;99:443–6.19. Wang W, Cote J, Xue Y, et al. Purification and biochemicalheterogeneity of the mammalian SWI-SNF complex. EMBO J 1996;15:5370–82.20. Decristofaro MF, Betz BL, Rorie CJ, et al. Characterization ofSWI/SNF protein expression in human breast cancer cell lines and othermalignancies. J Cell Physiol 2001;186:136–45.21. Versteege I, Sevenet N, Lange J, et al. Truncating mutations ofhSNF5/INI1 in aggressive paediatric cancer. Nature 1998;394:203–6.22. Klochendler-Yeivin A, Fiette L, Barra J, et al. The murine SNF5/INI1 chromatin remodeling factor is essential for embryonic develop-ment and tumor suppression. EMBO Rep 2000;1:500–6.23. Reisman DN, Sciarrotta J, Wang W, Funkhouser WK, WeissmanBE. Loss of BRG1/BRM in human lung cancer cell lines and primarylung cancers: correlation with poor prognosis. Cancer Res 2003;63:560–6.24. Muchardt C, Bourachot B, Reyes JC, Yaniv M. ras transformationis associated with decreased expression of the brm/SNF2alpha ATPasefrom the mammalian SWI-SNF complex. EMBO J 1998;17:223–31.25. Reyes JC, Barra J, Muchardt C, et al. Altered control of cellularproliferation in the absence of mammalian brahma (SNF2alpha). EMBOJ 1998;17:6979–91.26. Sif S, Saurin AJ, Imbalzano AN, Kingston RE. Purification andcharacterization of mSin3A-containing Brg1 and hBrm chromatin re-modeling complexes. Genes Dev 2001;15:603–18.27. Strobeck MW, Knudsen KE, Fribourg AF, et al. BRG-1 is requiredfor RB-mediated cell cycle arrest. Proc Natl Acad Sci USA 2000;97:7748–53.28. Ito T, Yamauchi M, Nishina M, et al. Identification of SWI�SNFcomplex subunit BAF60a as a determinant of the transactivation poten-tial of Fos/Jun dimers. J Biol Chem 2001;276:2852–7.29. Roberts CW, Galusha SA, McMenamin ME, Fletcher CD, OrkinSH. Haploinsufficiency of Snf5 (integrase interactor 1) predisposes tomalignant rhabdoid tumors in mice. Proc Natl Acad Sci USA 2000;97:13796–800.

30. Guidi CJ, Sands AT, Zambrowicz BP, et al. Disruption of Ini1 leadsto peri-implantation lethality and tumorigenesis in mice. Mol Cell Biol2001;21:3598–603.

31. Kallioniemi OP, Wagner U, Kononen J, Sauter G. Tissue microar-ray technology for high-throughput molecular profiling of cancer. HumMol Genet 2001;10:657–62.

32. Nocito A, Kononen J, Kallioniemi OP, Sauter G. Tissue microar-rays (TMAs) for high-throughput molecular pathology research. Int JCancer 2001;94:1–5.

33. Kononen J, Bubendorf L, Kallioniemi A, et al. Tissue microarraysfor high-throughput molecular profiling of tumor specimens. Nat Med1998;4:844–7.

34. Horvath L, Henshall S. The application of tissue microarrays tocancer research. Pathology 2001;33:125–9.

35. Bremnes RM, Veve R, Gabrielson E, et al. High-throughput tissuemicroarray analysis used to evaluate biology and prognostic significanceof the E-cadherin pathway in non-small-cell lung cancer. J Clin Oncol2002;20:2417–28.

36. WHO. Histological typing of lung and pleural tumours, 3rd ed.Geneva, Switzerland: Springer-Verlag; 1999.

37. Brehm A, Miska EA, McCance DJ, et al. Retinoblastoma proteinrecruits histone deacetylase to repress transcription. Nature 1998;391:597–601.

38. Hassan AH, Neely KE, Workman JL. Histone acetyltransferasecomplexes stabilize swi/snf binding to promoter nucleosomes. Cell2001;104:817–27.

39. Kimura A, Horikoshi M. Tip60 acetylates six lysines of a specificclass in core histones in vitro. Genes Cells 1998;3:789–800.

40. Meerzaman D, Shapiro P, Kim K. Involvement of the MAP kinaseERK2 in MUC1 mucin signaling. Am J Physiol Lung Cell Mol Physiol2001;281:L86–91.

41. Guinee DG Jr, Travis WD, Trivers GE, et al. Gender comparisonsin human lung cancer: analysis of p53 mutations, anti-p53 serum anti-bodies and C-erbB-2 expression. Carcinogenesis 1995;16:993–1002.

42. Hammar SP. Lung and pleural neoplasms. In: Dabbs DJ, editor.Diagnostic immunohistochemistry. Philadelphia: Churchill Livingstone;2001. p. 267–312.

43. Hibi K, Westra WH, Borges M, et al. PGP9.5 as a candidate tumormarker for non-small-cell lung cancer. Am J Pathol 1999;155:711–5.

44. McShane LM, Radmacher MD, Freidlin B, et al. Methods forassessing reproducibility of clustering patterns observed in analyses ofmicroarray data. Bioinformatics 2002;18:1462–9.

45. Zhang HS, Gavin M, Dahiya A, et al. Exit from G1 and S phase ofthe cell cycle is regulated by repressor complexes containing HDAC-Rb-hSWI/SNF and Rb-hSWI/SNF. Cell 2000;101:79–89.

46. Nicolas E, Morales V, Magnaghi-Jaulin L, et al. RbAp48 belongs tothe histone deacetylase complex that associates with the retinoblastomaprotein. J Biol Chem 2000;275:9797–804.

47. Muchardt C, Yaniv M. When the SWI/SNF complex remodels. Thecell cycle. Oncogene 2001;20:3067–75.

48. Kadam S, Emerson BM. Transcriptional specificity of human SWI/SNF BRG1 and BRM chromatin remodeling complexes. Mol Cell2003;11:377–89.

49. Xue Y, Canman JC, Lee CS, et al. The human SWI/SNF-B chro-matin-remodeling complex is related to yeast rsc and localizes at kin-etochores of mitotic chromosomes. Proc Natl Acad Sci USA 2000;97:13015–20.




2004;10:4314-4324. Clin Cancer Res Junya Fukuoka, Takeshi Fujii, Joanna H. Shih, et al. as Prognostic Indicators in Non-Small Cell Lung CancerChromatin Remodeling Factors and BRM/BRG1 Expression

Updated version

http://clincancerres.aacrjournals.org/content/10/13/4314

Access the most recent version of this article at:

Cited articles

http://clincancerres.aacrjournals.org/content/10/13/4314.full#ref-list-1

This article cites 46 articles, 17 of which you can access for free at:

Citing articles

http://clincancerres.aacrjournals.org/content/10/13/4314.full#related-urls

This article has been cited by 26 HighWire-hosted articles. Access the articles at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

SubscriptionsReprints and

[email protected] at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

Permissions

Rightslink site. (CCC)Click on "Request Permissions" which will take you to the Copyright Clearance Center's

.http://clincancerres.aacrjournals.org/content/10/13/4314To request permission to re-use all or part of this article, use this link



http://clincancerres.aacrjournals.org/content/10/13/4314.full#ref-list-1

http://clincancerres.aacrjournals.org/content/10/13/4314.full#related-urls

http://clincancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]



Documents

Chromatin Remodeling Factors and BRM/BRG1 Expression as ... · BRM staining (38.1%; P 0.016). These results support the notion that BRM and BRG1 participate in two distinct chromosome