Abstract
Purpose: The goal of this study was to establish a common set of molecules that regulate cell invasion in head and neck cancer (HNC).
Experimental Design: Five invasive sublines derived from HNC cell lines were established using the Matrigel selection method. Proteomic technology, MetaCore algorithm, and reverse transcriptase-PCR methods were used to search for molecules that contribute to the invasion phenotype. Cellular functional analyses and clinical association studies were applied to examine the significance of the molecules.
Results: Fifty-two proteins were identified in more than two of the four independent proteomic experiments, including 10 (19%) molecular chaperones. Seven chaperones were confirmed to be differentially expressed in five sublines, Hsp90α, Hsp90β, Hsp90-B1/Gp96, Hsp70-A5/Grp78, and HYOU1, that upregulate, whereas Hsp60 and glucosidase-α neutral AB (GANAB) downregulate. Four molecules were further investigated. In all cell lines, knockdown of Hsp60 or GANAB and silencing of Gp96 or Grp78 considerably enhanced or reduced cell migration and invasion, respectively. Clinical association studies consistently revealed that low levels of Hsp60 or GANAB and high levels of Gp96 or Grp78 are significantly associated with advanced cancer (P < 0.001 to P = 0.047, respectively, for the four molecules) and poor survival (P < 0.001 to P = 0.025, respectively, for the four molecules).
Conclusion: Our study defined molecular chaperones as a common set of proteins that regulate the invasion phenotype of HNC. Loss of the tumor suppression function of Hsp60 or GANAB and acquisition of the oncogenic function of Gp96 or Grp78 contribute to aggressive cancers. These molecules may serve as prognostic markers and targets for cancer drug development. Clin Cancer Res; 17(14); 4629–41. ©2011 AACR.
In this study, we identified 52 proteins whose expression patterns are commonly altered in 5 invasive head and neck cancer sublines. The proteins included 10 (19%) molecular chaperones. Seven proteins had confirmed differential expression patterns, and 4 were further investigated. Cellular studies showed that Gp96 and Grp78 play positive roles in the regulation of cell migration and invasion, whereas Hsp60 and GANAB play negative roles. Clinical studies have consistently revealed that high levels of Gp96 and Grp78 and low levels of Hsp60 and GANAB are significantly associated with advanced cancer and poor survival. Thus, losing the tumor suppressive function of Hsp60 or GANAB and acquiring the oncogenic function of Gp96 or Grp78 contribute to aggressive cancer. These proteins may serve as prognostic markers in the prediction of patient survival and as targets for cancer drug development.
Introduction
Head and neck cancer (HNC) is the 6th most prevalent cancer in the worldwide, with an estimated over 500,000 new cases being diagnosed annually (1, 2). HNC is characterized by an aggressive growth phenotype and early metastasis that lead to difficult tumor control. Identification of molecular markers and effective treatment regimens for metastasis are urgently needed. Recently, several investigators have conducted global expression profiling experiments to identify genes linked to cellular invasion. However, there was little overlap in the genes identified by each study, probably because of differences in the study designs. For example, different experimental approaches were used to identify invasion-related genes, including the direct comparison of 2 sets of samples with different invasive capabilities (3–5) and the comparison of cancer cell lines with normal keratinocytes (6). However, a major disadvantage of these approaches lies in the heterogeneity of the samples. To reduce heterogeneity and obtain a common set of data on molecules involved in HNC invasion, in this study, we used an in vitro Matrigel invasion model to establish 5 highly invasive HNC sublines, following a global survey of the invasion-associated molecules by proteomic methods. A total of 52 proteins were identified with high frequency, including a significant fraction of molecular chaperones (19%). We therefore further investigated whether this group of proteins possessed roles in cellular invasion.
Molecular chaperones are required for the stability and activity of a wide range of client proteins that are involved in many biological processes, including signal transduction, cellular trafficking, chromatin remodeling, cell growth, differentiation, and reproduction (7, 8). However, their functions in regulation of cell mobility or invasion have not been much addressed. In this study, we confirmed 7 proteins differentially expressed in 5 HNC invasion sublines. We further validated the biological functions and clinical significance of 4 proteins, and discussed the implications in the carcinogenic mechanisms.
Materials and Methods
Establishment and characterization of highly invasive HNC cell sublines
Five HNC cell lines were used, including 2 nasopharyngeal cancers (BM1 and BM2; ref. 9), 1 oral cancer (OECM1; ref. 10) and 2 pharyngolaryngeal cancers (Fadu and Detroit; ref. 11). All the cell culture conditions were the same as previously reports (9–11). The Matrigel invasion method was used to establish HNC cell sublines with a high invasive capability, similar to what has been previously described (12). Four generations of HNC invasive sublines were established. Although different cells showed varying levels of invasiveness, the invasive ability of the HNC sublines increased with ascending generations. Compared to the parental cells, the invasive ability of the 5 different cell lines increased between 13- and 21-fold after 4 generations (Supplementary Fig. S1). These results indicate that highly invasive sublines of HNC cells were successfully established.
Subcellular protein extraction and comparative proteomic analysis
To better identify cell proteomes, subcellular proteins were extracted as previously described (13). Fractions containing 50 μg of protein were separated by electrophoresis on an 8% to 16% gradient gel. These protein bands were then scanned by a densitometer and analyzed by ImageMaster software. Protein bands with differential expressions were excised, extracted, and identified using MALDI-TOF mass spectrometry as previously described (14, 15). Peptide mass fingerprinting was done using the Mascot search engine (Matrix Science) and the National Center for Biotechnology Information protein–protein BLAST database.
Prediction and analysis of the network pathways associated with invasion
Network analyses of differentially expressed genes were done using the MetaCore Analytical suite (GeneGo Inc.) as previously described (12). MetaCore was used to calculate the statistical significance (P value) of the probability of assembly from a random set of nodes (genes) of the same size as the input list. To build the network of differentially expressed genes, we applied the shortest paths algorithm to establish direct paths between selected objects.
RNA extraction and reverse transcriptase-PCR analysis
Total RNA was extracted from cells using the TRIzol reagent (Gibco BRL) following the manufacturer's instructions. The cDNA synthesis and PCR reaction primarily followed methods that have been previously described (16). The primer sequences are listed in Supplementary Table S1. The PCR products were analyzed by 1.5% agarose gel electrophoresis. The density of each band was determined after normalization to an actin control band using the Gel Image System (Scion Corporation).
Protein extraction and immunoblot analysis
The protein extraction and immunoblot analysis were done as previously described (17). Briefly, cellular proteins were separated by SDS-PAGE and transferred to a nitrocellulose membrane. The membrane was hybridized with primary antibodies and then incubated with horseradish peroxidase-conjugated secondary antibodies. The primary antibodies used were anti-Hsp60 (clone: MAB3514, Chemicon), anti-glucosidase-α neutral AB (GANAB; clone: C-16, Santa Cruz Biotech.), anti-Gp96 (clone: 9G10.F8.2, NeoMarkers), and anti-Grp78 (clone N-20, Santa Cruz Biotech.). The membranes were developed and exposed to x-ray film. Using the Gel Image System (Scion Corporation), the density of each band was determined after normalization to an actin control band.
Cloning and transfection of short hairpin RNA plasmids
The pTOPO-U6 vector was used to construct Hsp60-, GANAB-, Gp96-, and Grp78-short hairpin RNA (shRNA) plasmids as previously described (9). Briefly, a 15 to 18 sense and antisense complementary hairpin oligonucleotide was generated against a specific mRNA sequence of each gene, following cloned into pTOPO-U6 vector. The sequences of the specific shRNA oligonucleotides are listed in Supplementary Table S2.
For cellular transfection, the Lipofectamine 2000 reagent and Opti-MEM medium (Invitrogen) were used, as previous described (10, 11).
Cell growth, migration, and invasion assays
Cell growth, migration, and invasion assays were done as previously described (10). Briefly, cell growth was monitored by counting cells on a daily basis. Cell migration was evaluated using the in vitro wound-healing assay. Cell invasion assays were done using a BioCoat Matrigel and Transwell invasion chamber. The invasive ability of these cells was determined by 2 days of incubation and by counting the cells that had passed through the Matrigel-coated membrane into the lower chamber.
Immunohistochemistry
Immunohistochemistry analysis was done using a streptavidin–biotin complex system (LSAB2 system; Dako), similarly as previous described (16). These primary antibodies were anti-Hsp60 (clone: MAB3514, Chemicon), anti-GANAB (clone: C-16, Santa Cruz Biotech.), anti-Gp96 (clone: 9G10.F8.2, NeoMarkers), and anti-Grp78 (clone N-20, Santa Cruz Biotech.). The color was developed with AEC substrate chromogen (LSAB2 system). The staining reactions were determined by microscopic examination. The immunoreactivity was evaluated by subjective assessment of the median staining intensity, as negative (0: no staining), weak (+1), moderate (+2) or strong (+3).
Patient characteristics
This study was approved by the Institutional Review Broad of the Human Investigation Committee in our institution. Written informed consent was obtained from all patients participating in this study. A total of 78 patients visiting the Chang Gung Memorial Hospital (Taoyuan, Taiwan) were recruited for this study. Biopsies of the tumor sample and grossly normal mucosa cells were obtained from each subject before chemotherapy or radiotherapy. The patients included 74 (95%) males and 4 (5%) females with an age range of 28 to 76 years (mean age: 49.8 years). All patients received radical surgery and postoperative adjuvant treatment. Thirty-six patients (46%) received postoperative radiotherapy and 11 had concurrent cisplatin based chemotherapy. The median follow-up time was 5.34 years (range: 1.2–8.5 years). At the end of study, 20 patients had died of disease. The 3- and 5-year disease-specific survival rates were 81% and 76%, respectively.
Clinical association study
Biopsies of cancer and grossly normal mucosa tissue were obtained from each subject before chemotherapy or radiotherapy. Proteins were extracted from tissues and subjected to immunoblot analysis as described earlier to determine the expression level of the chaperone proteins (Hsp60, GANAB, Gp96, and Grp78). To define the relative levels of protein expression in clinical samples, the band density of each tumor sample was normalized with an internal control (actin protein expression) and compared with that of normal tissue from the same patient. The cutoff points were determined after calculation of the receiver operating characteristic curve for best fit of sensitivity and specificity. For Hsp60 and GANAB, the protein expressions in tumor tissues lower than 1.2-fold of the normal counterparts were defined as low level. In Gp90 and Grp78, the protein expressions in tumor tissues greater than 1.8-fold of the normal counterparts were defined as high level.
The Pearson's chi-square test was used to examine the association of chaperone protein expressions and clinicopathologic features, including TNM stage. Survival curves were calculated by the Kaplan–Meier method with a log-rank test. All P values were 2-sided, and the significance level was set at a P < 0.05.
Results
Comparative proteomics and network pathway prediction
Comparative proteomic analysis was done in the parental and the sublines of the HNC cell lines. Four independent experiments were done, and a representative example is shown in Supplementary Fig. S2(A). A total of 420 protein bands were identified by mass spectrometry. These bands represented 184 proteins. Fifty-two proteins were identified more than twice among different cell types, indicating the significance of these proteins in regulation of the invasion phenotype, as summarized in Table 1. Of these 52 proteins, 18 (35%) function as cytoskeleton or adhesion molecules, 10 (19%) function as molecular chaperones, 9 (17%) are metabolic enzymes, 8 (15%) are involved in transcriptional or translational regulation, and 7 (13%) are involved in cellular signaling transduction (Supplementary Fig. S2B).
NCBI number . | Name . | Scorea . | Mass (Da)b . | pIc . | Sequence coveraged . | Up/down regulatione . |
---|---|---|---|---|---|---|
Molecular chaperone (N = 10) | ||||||
gi|83699649 | Hsp 90, α | 209 | 98,082 | 5.07 | 55% | Up |
gi|46249928 | Hsp 90, β | 234 | 83,212 | 4.97 | 50% | Up |
gi|61656607 | Hsp 90-B-1/Gp96 | 250 | 92,282 | 4.47 | 47% | Up |
gi|62089222 | Hsp 70-1A variant | 212 | 77,448 | 5.97 | 41% | Up |
gi|34783614 | Hsp 70-1B | 182 | 70,009 | 5.62 | 56% | Up |
gi|48257068 | Hsp 70-8, isoform 1 | 170 | 70,854 | 5.37 | 60% | Up |
gi|6900104 | Hsp 70-A5/Grp78 | 321 | 72,288 | 5.07 | 52% | Up |
gi|47938913 | HYOU1 protein | 126 | 75,297 | 5.55 | 41% | Up |
gi|31542947 | Hsp60/Chaperonin | 234 | 60,986 | 5.7 | 61% | Down |
gi|38197033 | GANAB | 108 | 55,401 | 6.26 | 40% | Down |
Cytoskeleton/adhesion (N = 18) | ||||||
gi|53791219 | Flamin A | 340 | 277,332 | 5.70 | 40% | Up |
gi|62089364 | Filamin B, β | 193 | 170,056 | 5.29 | 38% | Up |
gi|29436380 | MYH9 | 286 | 158,653 | 5.78 | 60% | Up |
gi|50513540 | Neurofibromin 2 | 110 | 64,570 | 9.03 | 61% | Up |
gi|4506787 | IQ motif-GTPase-1 | 398 | 189,134 | 6.08 | 45% | Up |
gi|39795300 | MAP6 | 43 | 77,024 | 8.88 | 69% | Up |
gi|2606094 | Cyr61 | 40 | 42,041 | 8.68 | 71% | Up |
gi|54696696 | Annexin A1 | 101 | 38,690 | 6.57 | 66% | Up |
gi|50845388 | Annexin A2 | 250 | 40,386 | 8.53 | 59% | Up |
gi|55665463 | MACF1 | 97 | 505,330 | 5.22 | 50% | Up |
gi|14719392 | Cofilin 2 | 76 | 18,725 | 7.66 | 72% | Up |
gi|18088719 | Tubulin, β | 163 | 49,640 | 4.79 | 61% | Up |
gi|46249758 | Villin 2 (ezrin) | 280 | 69,199 | 5.94 | 58% | Up |
gi|1477646 | Plectin | 291 | 518,173 | 5.57 | 26% | Up |
gi|62897681 | Calreticulin precursor | 105 | 46,890 | 4.3 | 53% | Up |
gi|62088870 | Type VII collagen-α1 | 96 | 96,749 | 5.66 | 29% | Up |
gi|299626 | EMS1 | 197 | 61,599 | 5.24 | 59% | Up |
gi|24657579 | Vinculin, isoform VCL | 188 | 116,663 | 5.83 | 46% | Down |
Metabolic enzyme (N = 9) | ||||||
gi|38013966 | TKT | 210 | 58,174 | 6.51 | 69% | Up |
gi|35505 | Pyruvate kinase | 230 | 57,841 | 7.58 | 60% | Up |
gi|13325287 | Enolase | 189 | 47,139 | 7.01 | 74% | Up |
gi|15012036 | Glutathione transferase | 130 | 23,327 | 5.43 | 66% | Up |
gi|4758304 | Disulfide isomerase-4 | 78 | 72,887 | 4.96 | 40% | Up |
gi|89573929 | G3PDH | 204 | 24,605 | 8.68 | 64% | Up |
gi|30582761 | Phosphoglucomutase 1 | 93 | 1,331 | 6.2 | 64% | Up |
gi|5803225 | Tyr/Tryp-monooxygenase | 157 | 29,155 | 4.63 | 71% | Up |
gi|12804929 | NAD 2 (mitochondrial) | 147 | 35,537 | 8.92 | 70% | Up |
Transcription/translation (N = 8) | ||||||
gi|10863945 | ATP-dep DNA helicase II | 102 | 82,652 | 5.55 | 38% | Up |
gi|29126861 | TIF- 4A, isoform 2 | 87 | 46,460 | 5.32 | 50% | Up |
gi|62896661 | TEF-1α, 1 variant | 110 | 50,110 | 8.98 | 45% | Up |
gi|4503483 | TEF-2 | 111 | 95,277 | 6.41 | 46% | Up |
gi|62088704 | HnRNP-K isoform variant | 135 | 48,774 | 5.48 | 59% | Up |
gi|75517570 | HnRNP-A1 | 139 | 29,368 | 9.19 | 67% | Up |
gi|14124942 | Ribophorin I | 118 | 64,542 | 6.1 | 60% | Up |
gi|189306 | Nucleolin | 76 | 76,298 | 4.59 | 29% | Up |
Signal transduction/others (N = 7) | ||||||
gi|5174447 | G protein, β2-like | 82 | 35,055 | 7.6 | 56% | Up |
gi|32455266 | Peroxiredoxin 1 | 127 | 22,096 | 8.67 | 52% | Up |
gi|13654237 | DNA-activated kinase | 335 | 468,788 | 6.75 | 28% | Up |
gi|40789059 | KIAA0051 | 461 | 19,1298 | 6.14 | 46% | Up |
gi|5031703 | SH3- binding protein | 100 | 52,132 | 5.38 | 53% | Up |
gi|482321 | TNF-α–induced protein | 58 | 39,625 | 8.41 | 43% | Up |
gi|62131678 | 14-3-3-ϵ | 143 | 26,487 | 4.76 | 65% | Down |
NCBI number . | Name . | Scorea . | Mass (Da)b . | pIc . | Sequence coveraged . | Up/down regulatione . |
---|---|---|---|---|---|---|
Molecular chaperone (N = 10) | ||||||
gi|83699649 | Hsp 90, α | 209 | 98,082 | 5.07 | 55% | Up |
gi|46249928 | Hsp 90, β | 234 | 83,212 | 4.97 | 50% | Up |
gi|61656607 | Hsp 90-B-1/Gp96 | 250 | 92,282 | 4.47 | 47% | Up |
gi|62089222 | Hsp 70-1A variant | 212 | 77,448 | 5.97 | 41% | Up |
gi|34783614 | Hsp 70-1B | 182 | 70,009 | 5.62 | 56% | Up |
gi|48257068 | Hsp 70-8, isoform 1 | 170 | 70,854 | 5.37 | 60% | Up |
gi|6900104 | Hsp 70-A5/Grp78 | 321 | 72,288 | 5.07 | 52% | Up |
gi|47938913 | HYOU1 protein | 126 | 75,297 | 5.55 | 41% | Up |
gi|31542947 | Hsp60/Chaperonin | 234 | 60,986 | 5.7 | 61% | Down |
gi|38197033 | GANAB | 108 | 55,401 | 6.26 | 40% | Down |
Cytoskeleton/adhesion (N = 18) | ||||||
gi|53791219 | Flamin A | 340 | 277,332 | 5.70 | 40% | Up |
gi|62089364 | Filamin B, β | 193 | 170,056 | 5.29 | 38% | Up |
gi|29436380 | MYH9 | 286 | 158,653 | 5.78 | 60% | Up |
gi|50513540 | Neurofibromin 2 | 110 | 64,570 | 9.03 | 61% | Up |
gi|4506787 | IQ motif-GTPase-1 | 398 | 189,134 | 6.08 | 45% | Up |
gi|39795300 | MAP6 | 43 | 77,024 | 8.88 | 69% | Up |
gi|2606094 | Cyr61 | 40 | 42,041 | 8.68 | 71% | Up |
gi|54696696 | Annexin A1 | 101 | 38,690 | 6.57 | 66% | Up |
gi|50845388 | Annexin A2 | 250 | 40,386 | 8.53 | 59% | Up |
gi|55665463 | MACF1 | 97 | 505,330 | 5.22 | 50% | Up |
gi|14719392 | Cofilin 2 | 76 | 18,725 | 7.66 | 72% | Up |
gi|18088719 | Tubulin, β | 163 | 49,640 | 4.79 | 61% | Up |
gi|46249758 | Villin 2 (ezrin) | 280 | 69,199 | 5.94 | 58% | Up |
gi|1477646 | Plectin | 291 | 518,173 | 5.57 | 26% | Up |
gi|62897681 | Calreticulin precursor | 105 | 46,890 | 4.3 | 53% | Up |
gi|62088870 | Type VII collagen-α1 | 96 | 96,749 | 5.66 | 29% | Up |
gi|299626 | EMS1 | 197 | 61,599 | 5.24 | 59% | Up |
gi|24657579 | Vinculin, isoform VCL | 188 | 116,663 | 5.83 | 46% | Down |
Metabolic enzyme (N = 9) | ||||||
gi|38013966 | TKT | 210 | 58,174 | 6.51 | 69% | Up |
gi|35505 | Pyruvate kinase | 230 | 57,841 | 7.58 | 60% | Up |
gi|13325287 | Enolase | 189 | 47,139 | 7.01 | 74% | Up |
gi|15012036 | Glutathione transferase | 130 | 23,327 | 5.43 | 66% | Up |
gi|4758304 | Disulfide isomerase-4 | 78 | 72,887 | 4.96 | 40% | Up |
gi|89573929 | G3PDH | 204 | 24,605 | 8.68 | 64% | Up |
gi|30582761 | Phosphoglucomutase 1 | 93 | 1,331 | 6.2 | 64% | Up |
gi|5803225 | Tyr/Tryp-monooxygenase | 157 | 29,155 | 4.63 | 71% | Up |
gi|12804929 | NAD 2 (mitochondrial) | 147 | 35,537 | 8.92 | 70% | Up |
Transcription/translation (N = 8) | ||||||
gi|10863945 | ATP-dep DNA helicase II | 102 | 82,652 | 5.55 | 38% | Up |
gi|29126861 | TIF- 4A, isoform 2 | 87 | 46,460 | 5.32 | 50% | Up |
gi|62896661 | TEF-1α, 1 variant | 110 | 50,110 | 8.98 | 45% | Up |
gi|4503483 | TEF-2 | 111 | 95,277 | 6.41 | 46% | Up |
gi|62088704 | HnRNP-K isoform variant | 135 | 48,774 | 5.48 | 59% | Up |
gi|75517570 | HnRNP-A1 | 139 | 29,368 | 9.19 | 67% | Up |
gi|14124942 | Ribophorin I | 118 | 64,542 | 6.1 | 60% | Up |
gi|189306 | Nucleolin | 76 | 76,298 | 4.59 | 29% | Up |
Signal transduction/others (N = 7) | ||||||
gi|5174447 | G protein, β2-like | 82 | 35,055 | 7.6 | 56% | Up |
gi|32455266 | Peroxiredoxin 1 | 127 | 22,096 | 8.67 | 52% | Up |
gi|13654237 | DNA-activated kinase | 335 | 468,788 | 6.75 | 28% | Up |
gi|40789059 | KIAA0051 | 461 | 19,1298 | 6.14 | 46% | Up |
gi|5031703 | SH3- binding protein | 100 | 52,132 | 5.38 | 53% | Up |
gi|482321 | TNF-α–induced protein | 58 | 39,625 | 8.41 | 43% | Up |
gi|62131678 | 14-3-3-ϵ | 143 | 26,487 | 4.76 | 65% | Down |
aScore, probability based scoring after Mascot peptide search with mass fingerprint (50). The scores greater than 67 are considered significant (P < 0.05).
bMass (Da), specific protein mass provided by Mascot peptide search with the mass spectrum data from MALDI-TOF mass spectrometric analysis.
cpI, isoelectric point provided by Mascot peptide search with the mass spectrum data from MALDI-TOF mass spectrometric analysis.
dSequence coverage, the percentage of sequence with identified peptides over predicted target protein.
eUp/downregulation, upregulation or downregulation of the specific protein in invasion subline cells.
The 52 proteins identified in several proteomes were imported into MetaCore, and pathways associated with invasiveness were analyzed. Seven regulatory pathways were found to be significantly associated with invasiveness (P < 10−23): regulation of apoptosis, actin cytoskeleton organization and biogenesis, mechanism of double-strand break repair, cellular response to stress, branching morphogenesis of a tube, pathway in mitochondrial ornithine transport, and cell cycle regulation (Supplementary Table S3).
To more specifically determine the most significant pathway in the regulation of cell invasion in HNC, the shortest path analytical model of MetaCore was applied. A total of 33 of the 52 identified genes were matched to network pathways (Fig. 1A). To validate these predicted networks, some of the network proteins were examined differentially expressed between parental and subline cells. As shown in Fig. 1B, several molecules were shown to be involved in the regulation of the invasion phenotype: the MMP2 extracellular protein; membrane protein Annexin-II; cytoplasmic c-Raf, ERK1/2, IKK-α, and 14-3-3-sigma signaling proteins; and nuclear STAT1 and c-Myc transcription factors. These results indicate that the invasion phenotype is exquisitely controlled by complicated mechanisms in human cells.
Molecular chaperone proteins have consistently altered expression in invasive cells
It is interesting to note that, in addition to cytoskeleton and adhesion molecules, molecular chaperones represented a significant fraction (10/52, 19%) of the proteins we identified (Table 1). We therefore explored whether this group of proteins contributes to cellular invasion in HNC. Seven of the proteins that appeared frequently in the proteomic screening were selected for confirmation studies. Reverse transcriptase (RT)-PCR analysis was done and the results are shown in Fig. 1C. As shown, Hsp90α/Hsp90-AA, Hsp90β/Hsp90-AB, Hsp90-B1/Gp96, Hsp70-A5/Grp78, and HYOU1 were consistently upregulated in all 5 invasive sublines compared with the corresponding parental cells (P < 0.001 in all the molecules). Hsp60 and GANAB were downregulated in all the invasive sublines (P < 0.001 in both molecules). The gene expression levels were also determined by RT-quantitative PCR method and the similar results were obtained (Supplementary Fig. S3). These results suggest that these proteins play common and important roles in regulation of invasive phenotype in HNC.
Cell growth was suppressed by Hsp60 and GANAB and promoted by Gp96 and Grp78
To shed more light on the biological functions of these chaperones, the 4 molecules (Hsp60, GANAB, Gp96, and Grp78) that exhibited the greatest change in expressions were selected for further study. Two specific shRNA constructs were designed against each protein, and the knockdown efficacy of each plasmid was determined by immunoblot analysis. As shown in Supplementary Fig. 4A, there were different levels of RNA expression in the presence of each of the 2 shRNA constructs, with 1 (Sh-2 of Hsp60, Sh-2 of GANAB, Sh-1 of Gp96, and Sh-2 of Grp78) showing more significant suppression (over 80% for all 4 genes). Therefore, we used the more potent shRNA constructs for further study. The specificity of these shRNA constructs was examined by Western blot analysis for the 4 proteins Gp96, Grp78, Hsp60, and GANAB, as shown in the Supplementary Fig. 4B. Results showed that the specificity for each shRNA construct was very high, with only minimal effects on the other chaperone proteins.
The effects of the shRNAs on cell growth were determined. In general, treatment with either Hsp60-sh or GANAB-sh enhanced cell growth, whereas treatment with both Gp96-sh and Grp78-sh inhibited cell growth. Using Fadu cells for example, cell growth rates increased with time when treated with Hsp60-sh or GANAB-sh, whereas the cell growth rates were significantly reduced upon Gp96-sh or Grp78-sh treatment (Supplementary Fig. S5). This phenomenon was consistently observed in other HNC cells, with a 1.3- to 1.6-fold increase in cells treated with either Hsp60-sh or GANAB-sh (P = 0.003–0.04 in all cell lines; Fig. 2A). Similarly, consistent with Fadu cells, substantial reductions were found in other HNC cells when treated with Gp96-sh (27%–54%, P < 0.03 in all cell lines) or Grp78-sh (18%–53%, P < 0.02 in all cell lines; Fig. 2A). These results suggest that Hsp60 and GANAB play negative roles, whereas Gp96 and Grp78 play positive roles in the regulation of cell growth.
Cell migration and invasion were inhibited by Hsp60 and GANAB and augmented by Gp96 and Grp78
An in vitro wound-healing assay was done to determine the effect that silencing these chaperones has on cell migration. As shown in Fig. 2B, both Hsp60-sh and GANAB-sh transfected cells migrated much faster than control cells. After 24 hours, the Hsp60-sh cells showed a 1.49-fold faster movement and GANAB-sh cells showed a 1.78-fold faster movement compared to the vector transfectants. After 48 hours, the wounded area was completely closed in the shRNA transfected cells, whereas the wounded area in vector transfected cells was not closed. The silencing of Gp96 and Grp78 had different effects. Both Gp96-sh and Grp78-sh transfected cells migrated much more slowly than control cells. After 48 hours, the wounded area remained significantly larger when compared to that in vector transfected cells, with a 44% and a 57% decrease for Gp96-sh and Grp78-sh transfectants, respectively. These results suggest that Hsp60 and GANAB negatively regulate cell migration, whereas Gp96 and Grp78 possess functions that positively regulate cell migration.
The invasive ability of the cells was determined using the Matrigel invasion assay. Cells treated with both Hsp60-sh and GANAB-sh showed a significant increase in invasive ability compared to control cells, as shown in Supplementary Fig. S6 and Fig. 2C. After 2 days, all 3 HNC cell lines that had Hsp60 silenced displayed a 1.6- to 3.9-fold (P < 0.001 to P = 0.018) increased ability to invade. The effect of GANAB silencing was even more dramatic. In all GANAB-sh cells, a 2.2- to 24.1-fold increase in invasiveness (P < 0.001 to P = 0.014) was observed. Treatment with either Gp96-sh or Grp78-sh resulted in a substantial decrease in cell invasion, as shown in the Supplementary Fig. S6 and 2C. After 2 days, all 3 Gp96-silenced cell lines displayed a 6% to 24% suppressed invasive ability (P < 0.001 in all 3 cell lines). Similarly, the 3 Grp78-silenced cell lines displayed a 4% to 23% suppressed invasive ability after 5 days (P < 0.001 in all 3 cell lines). These results suggest that Hsp60 and GANAB have negative roles, whereas Gp96 and Grp78 play crucially positive roles in the regulation of cell invasion.
Low-level expression of Hsp60 and GANAB and high-level expression of Gp96 and Grp78 are correlated with cancer aggressiveness and poor survival
To understand the clinical significance of these chaperone proteins in cancer, paired tumor and adjacent grossly normal tissues from 78 patients with HNC were obtained for study. For each tissue sample, total protein was extracted and subjected to immunoblot analysis. A representative example of the results is shown in Fig. 3A. As shown, Hsp60 and GANAB were significantly downregulated in many of the cancer tissues, whereas Gp96 and Grp78 were substantially upregulated. Results of immunohistochemistry analysis also support these findings. In general, the lower level of Hsp60 and GANAB and the higher level of Gp96 and Grp78 were found in advanced cancers in comparison with early staged diseases (Fig. 3B).
To determine the potential association between cancer status and protein expression, the Pearson's chi-square method was used for statistical analysis. Results are summarized in Table 2. For Hsp60 and GANAB proteins, low expression was correlated with more aggressive cancer, as in pathological T stage (P < 0.001 and P = 0.023 in Hsp60 and GANAB, respectively), N stage (P < 0.001 for both Hsp60 and GANAB), and overall stage (P < 0.001 for both Hsp60 and GANAB). Conversely, for Gp96 and Grp78, high expression was correlated with advanced cancer, as in pathological T stage (P = 0.047 and P = 0.024 for Gp96 and Grp78, respectively), N stage (P = 0.029 and P = 0.034 for Gp96 and Grp78, respectively), and overall stage (P = 0.016 and P = 0.007 for Gp96 and Grp78, respectively).
Clinical status . | Na . | HSP60 (%) . | GANAB (%) . | Gp96 (%) . | Grp78 (%) . | ||||
---|---|---|---|---|---|---|---|---|---|
. | . | Low . | High . | Low . | High . | Low . | High . | Low . | High . |
Pathological T stage | |||||||||
T1–T2 | 60 | 19 (32) | 41 (68) | 25 (42) | 35 (58) | 36 (60) | 24 (40) | 41 (68) | 19 (32) |
T3–T4 | 18 | 16 (89) | 2 (11) | 13 (72) | 5 (28) | 6 (33) | 12 (67) | 7 (39) | 11 (61) |
P value | <0.001 | 0.023 | 0.047 | 0.024 | |||||
Pathological N stage | |||||||||
N = 0 | 55 | 16 (29) | 39 (71) | 19 (34) | 36 (66) | 34 (62) | 21 (38) | 38 (69) | 17 (31) |
N > 0 | 23 | 19 (83) | 4 (17) | 19 (83) | 4 (17) | 8 (35) | 15 (65) | 10 (43) | 13 (57) |
P value | <0.001 | <0.001 | 0.029 | 0.034 | |||||
Overall stage | |||||||||
I-II | 46 | 9 (20) | 37 (80) | 14 (30) | 32 (70) | 30 (65) | 16 (35) | 34 (74) | 12 (26) |
III-IV | 32 | 26 (81) | 6 (19) | 24 (75) | 8 (25) | 12 (37) | 20 (63) | 14 (44) | 18 (56) |
P value | <0.001 | <0.001 | 0.016 | 0.007 | |||||
Differentiation | |||||||||
Well | 30 | 12 (40) | 18 (60) | 13 (43) | 17 (57) | 17 (57) | 13 (43) | 22 (73) | 8 (27) |
bM-P | 48 | 23 (48) | 25 (52) | 25 (52) | 23 (48) | 25 (52) | 23 (48) | 26 (54) | 22 (46) |
P value | 0.494 | 0.795 | 0.693 | 0.091 | |||||
Total | 78 | 35 (45) | 43 (55) | 38 (49) | 40 (51) | 42 (54) | 36 (46) | 48 (61) | 30 (39) |
Clinical status . | Na . | HSP60 (%) . | GANAB (%) . | Gp96 (%) . | Grp78 (%) . | ||||
---|---|---|---|---|---|---|---|---|---|
. | . | Low . | High . | Low . | High . | Low . | High . | Low . | High . |
Pathological T stage | |||||||||
T1–T2 | 60 | 19 (32) | 41 (68) | 25 (42) | 35 (58) | 36 (60) | 24 (40) | 41 (68) | 19 (32) |
T3–T4 | 18 | 16 (89) | 2 (11) | 13 (72) | 5 (28) | 6 (33) | 12 (67) | 7 (39) | 11 (61) |
P value | <0.001 | 0.023 | 0.047 | 0.024 | |||||
Pathological N stage | |||||||||
N = 0 | 55 | 16 (29) | 39 (71) | 19 (34) | 36 (66) | 34 (62) | 21 (38) | 38 (69) | 17 (31) |
N > 0 | 23 | 19 (83) | 4 (17) | 19 (83) | 4 (17) | 8 (35) | 15 (65) | 10 (43) | 13 (57) |
P value | <0.001 | <0.001 | 0.029 | 0.034 | |||||
Overall stage | |||||||||
I-II | 46 | 9 (20) | 37 (80) | 14 (30) | 32 (70) | 30 (65) | 16 (35) | 34 (74) | 12 (26) |
III-IV | 32 | 26 (81) | 6 (19) | 24 (75) | 8 (25) | 12 (37) | 20 (63) | 14 (44) | 18 (56) |
P value | <0.001 | <0.001 | 0.016 | 0.007 | |||||
Differentiation | |||||||||
Well | 30 | 12 (40) | 18 (60) | 13 (43) | 17 (57) | 17 (57) | 13 (43) | 22 (73) | 8 (27) |
bM-P | 48 | 23 (48) | 25 (52) | 25 (52) | 23 (48) | 25 (52) | 23 (48) | 26 (54) | 22 (46) |
P value | 0.494 | 0.795 | 0.693 | 0.091 | |||||
Total | 78 | 35 (45) | 43 (55) | 38 (49) | 40 (51) | 42 (54) | 36 (46) | 48 (61) | 30 (39) |
NOTE: Total of 78 head and neck cancer patients were recruited. In each cancer tissue, the expression levels of the 4 proteins (HSP60, GANAB, Gp96, and Grp78) were determined as low or high levels by Western blot analysis, as described in Materials and Methods. The Pearson's chi-square test was used to examine the association of these protein expressions with clinicopathologic features, including pathological T stage (T1–T2 vs. T3–T4), pathological N stage (N = 0 vs. N > 0), overall stage (stages I–II vs. stages III–IV), and the differentiation status (well differentiation vs. moderate to poor differentiation). P < 0.05 was considered statistical significance.
aN, number of patients.
bM–P, moderate to poor differentiation.
The association of patient survival and protein expression was examined using the Kaplan–Meier method with a log-rank test. As shown in Fig. 3C, lower levels of Hsp60 or GANAB were associated with adverse treatment outcomes (5-year disease-specific survival was 93% vs. 56%, P < 0.001 for Hsp60, and 90% vs. 62%, P = 0.006 for GANAB). Conversely, higher levels of Gp96 or Grp78 were associated with a poor treatment outcome (5-year disease-specific survival was 55% vs. 85%, P = 0.025 for Gp96, and 63% vs. 87%, P = 0.005 for Grp78).
Discussion
In HNC, tumor invasion and lymph node metastasis are common causes of treatment failure. To investigate the invasion phenotype of HNC, we established 5 sublines of highly invasive HNC cells. We used these cells to identify a common set of proteins that regulate the invasion phenotype. Proteomic analysis identified 52 proteins whose expression patters were frequently altered in the invasive sublines (Table 1). Network pathway prediction revealed that several cellular processes may respond to invasion, suggesting the involvement of complex circuits in the regulation of the invasion phenotype. Furthermore, specific proteins in the predicted pathways were identified (Fig. 1B), suggesting a high probability that the algorithmic analysis in this study is valid.
The biological functions of the identified proteins in the invasive sublines were examined. In addition to cytoskeleton adhesion molecules, molecular chaperone proteins represented a considerable fraction (19%) of the identified proteins. These molecular chaperone proteins included 3 proteins that belong to the Hsp90 family and 4 proteins that belong to the Hsp70 family (Table 1). We further confirmed that 7 proteins were differentially expressed in the 5 invasive HNC sublines. Of these, 5 (Hsp90α/Hsp90-AA, Hsp90β/Hsp90-AB, Hsp90-B1/Gp96, Hsp70-A5/Grp78, and HYOU1) were consistently upregulated, and 2 (Hsp60 and GANAB) were consistently downregulated (Fig. 1C). The 4 proteins with the greatest changes in the expression levels were selected for functional validation. Although Hsp60 and GANAB had only a marginal effect on cell growth, the knockdown of these 2 genes considerably enhanced cell migration and invasion. This suggests that these 2 proteins possess tumor suppressive functions, especially on the regulation of cell mobility (Fig. 2). In contrast, the knockdown of Gp96 or Grp78 drastically reduced cell growth, migration, and invasion, suggesting that these 2 proteins possess oncogenic functions (Fig. 2). Although the underlined mechanism between these chaperone proteins and invasion is still unclear, because there is no cross interactions between these 4 chaperones (Supplementary Fig. S4B), the effects of these molecules leading to cell invasions may be through mutual independent pathways.
A number of molecular chaperones, including Hsp60, are typically located in mitochondria and are presumed to function mainly within this organelle. However, there is now accumulating evidence that these chaperones are also located in a variety of cellular compartments where they do important functions (17–19). For example, Hsp60 in the cytosol can interact with procaspase-3 or p53 to orchestrate survival, whereas disruption of these complexes can exert an apoptotic effect (20, 21). All of these reports indicate that Hsp60 is involved in multiple cellular functions related to maintaining homeostasis. Recently, aberrant expression of Hsp60 was found in many clinical cancer tissues; however, this protein has been reported to be both positively and negatively correlated with cancer status. For example, elevated expression of Hsp60 was found in cervical, prostate, and breast cancers (22–24). On the other hand, decreased expression of this protein was reported in bronchial and bladder cancers (25, 26). Our results are in agreement with the latter findings that Hsp60 expression was lost during cancer progression (Fig. 3A and B). Furthermore, lower levels of Hsp60 expression were associated with clinicopathological stage and poor survival (Table 2, Fig. 3C), indicating that loss of the Hsp60 tumor suppression function may lead to advanced malignancy.
GANAB, also named the glucosidase II-α subunit, is a neutral glucosidase involved in the transaction and folding of newly synthesized glycoproteins in the endoplasmic reticulum (ER; refs. 27, 28). Disruption of GANAB led to the accumulation of misfolded glycoproteins and the induction of the unfolded protein response (28). GANAB is also a key regulator of glycosylation. Deletion of GANAB led to a novel N-glycosylation mechanism in the biosynthesis of variant cell surface glycoproteins (29). The loss of glycosidase II is associated with polycystic liver disease, in which hepatocystin fails to assemble with GANAB during carbohydrate processing, leading to altered cellular proliferation and differentiation (30, 31). Aside from polycystic liver disease, to the best of our knowledge, GANAB has not been reported to be associated with other diseases. In this study, we first showed that GANAB is associated with cancer development, through negative regulation of cell migration and invasion abilities (Fig. 2). This downregulation further supports a correlation between GANAB expression and more aggressive cancer and poor survival (Table 2, Fig. 3). Because N-glycosylation of several cellular proteins may change tumor invasion and metastatic ability (32, 33), GANAB may participate in carcinogenesis through the regulation of the N-glycosylation of specific client proteins.
There are 3 types of Hsp90s in mammalian cells: (a) cytosolic Hsp90, which includes Hsp90α/Hsp90-AA and Hsp90β/Hsp90-AB; (b) ER Hsp90, Hsp90-B1/Gp96/Grp94/endoplasmin; and (c) mitochondrial Hsp90, Trap-1 (7, 34). Because several oncoproteins are client proteins of cytosolic Hsp90, many Hsp90 inhibitors have been developed as anticancer therapeutic agents in preclinical and clinical evaluations (35, 36). The ER Hsp90, Gp96, was first identified to be associated with malignancy by screening a human teratocarcinoma cDNA library with mouse Gp96, which encodes the TRA1 protein (37). As knowledge increased, Gp96 was found to play an important role in the cell-mediated immunity of the antitumor response by forming stable complexes with tumor-derived antigenic peptides that in turn present these peptides to MHC class I complexes (38). Recently, Gp96 overexpression has been observed in several cancers, suggesting a direct link between this protein and malignant diseases (39–42). In this study, we found that the levels of both cytosolic (Hsp90α and Hsp90β) and endoplasmic reticulum (Gp96) Hsp90s are elevated in the invasive subline cells, and the level of Gp96 was higher than that of other Hsp90s (Fig. 1C). We further showed that high levels of Gp96 are associated with more aggressive phenotypes and poor survival of HNC patients (Table 2, Fig. 3). In addition, silencing Gp96 significantly suppressed cell growth and invasion (Fig. 2). Therefore, Gp96 may actively regulate multiple cancer behaviors and contribute to malignant transformation.
Grp78 (also named Hsp70-5, HspA5, or Bip) is a well-characterized ER chaperone that is part of the Hsp70 family (43). Previously, the study of Grp78 emphasized the cytoprotective and antiapoptotic function of Grp78 in response to ER stress, which leads to the modulation of chemosensitivity (44). Recently, it has been shown that Grp78 plays more roles than originally appreciated. Studies have shown that this protein is also expressed on the cell surface and may actively regulate multiple malignant phenotypes (45–48). Silencing of Grp78 expression may result in the loss of PTEN tumor suppression and oncogenic AKT activation (46). Grp78 may also form a complex with the cripto protein at the cell surface, and knockdown of this complex may eventually lead to suppression of cell proliferation and invasion (47, 48). In agreement with these reports, Grp78 has been found to be overexpressed in many cancers, including lung, colon, esophageal, gastric, and oral carcinomas (39–42, 49). In this study, we found high levels of Grp78 in HNC tissues, and these levels were associated with more aggressive status and poor prognosis (Table 2, Fig. 2). In addition, silencing of Grp78 significantly suppressed cell growth and invasion (Fig. 2). Thus, Grp78 has been shown to be an important oncogenic protein in many aspects.
Understanding the mechanisms underlying the regulation of invasion provides insight into the management of bulky or metastatic cancers. Our study shows the great potential of several molecular chaperones, especially Hsp60, GANAB, Gp96, and Grp78, in clinical applications involving HNC prognosis and treatment.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Grant Support
This study is supported by Grants from Chang Gung Memorial Hospital (CMRPG390421, J.T. Chang) and National Science Counsel of Taiwan (NSC 98-2314-B-182-039-MY3).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.