Skip to main content

Thank you for visiting You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

DSG3 is overexpressed in head neck cancer and is a potential molecular target for inhibition of oncogenesis


To identify genes that could potentially serve as molecular therapeutic markers for human head and neck cancer (HNC), we employed differential display analysis to compare the gene expression profiles between HNC and histopathologically normal epithelial tissues. Using reverse transcription–polymerase chain reaction and Western blot analysis, desmoglein 3 (DSG3) was identified as being differentially expressed at both the RNA and protein levels. Of 56 patients assayed, 34 (61%) had overexpression of DSG3, which correlated statistically with T stage (P=0.009), N stage (P=0.047), overall stage (P=0.011), tumor depth (P=0.009) and extracapsular spread in lymph nodes (P=0.044), suggesting that DSG3 participates in carcinogenesis of HNC. Consistent with the clinical findings, inhibition of DSG3 by RNA interference (RNAi) significantly reduced cell growth and colony formation to 57–21% in three HNC cell lines. Use of an in vitro wound healing and Matrigel invasion assays, we found that cell migration and invasive ability were also inhibited to 30–48% in three cell lines tested. An in vivo xenograft study showed that administration of DSG3-RNAi plasmid significantly inhibited tumor growth for 2 months in BALB/C nude mice. In conclusion, DSG3 is identified overexpressed in HNC, with the degree of overexpression associated with clinicopathologic features of the tumor. Inhibition of DSG3 significantly suppresses carcinogenic potential in cellular and in vivo animal studies. These findings suggest that DSG3 is a potential molecular target in the development of adjuvant therapy for HNC.


Head and neck cancer (HNC) is one of the most frequent cancers worldwide, with an estimated 500 000 or more new cases diagnosed annually (Vokes et al., 1993). This disease occurs much more frequently in males than in females (Vokes et al., 1993). Epidemiologic studies of HNC show a strong association with environmental carcinogens, especially the consumption of tobacco, alcohol or betel quid (Decker and Goldstein, 1982; Liao et al., 2004). The overall 5-year survival rate for patients with HNC is among the lowest of the major cancers and has not changed during the past two decades (Clayman et al., 1996). The standard treatment for patients with this malignancy is surgery or radiation or a combination of the two (Clayman et al., 1996). Although local control and survival rates are acceptable for early stage tumors, in advanced disease, bulky tumors or lymph node (LN) metastasis is a common cause of treatment failure. Reported 5-year disease-specific survival rates for tumor stages I, II, III and IV are 70–85, 45–60, 20–40 and 10–23%, respectively (Clayman et al., 1996). Even if there is a good treatment response, patients with advanced disease often suffer from substantial functional and cosmetic morbidity, which decreases the quality of life. Therefore, better therapies are needed for this malignant disease. A better understanding of the molecular mechanism of HNC may lead to improved therapeutic techniques.

Global genetic survey techniques that allow assessment of a wide range of genes at one time have been developed. However, the genetic alterations leading to HNC have not yet been well studied. The lack of this information hinders the investigation of molecular targets for the prevention, diagnosis and treatment of HNC. In this study, using a differential display technique, we identified desmoglein 3 (DSG3) overexpressed in HNC. The significance of DSG3 in clinical pathology, in cellular function and the potential effect of DSG3 knockdown in xenograft tumor mice were evaluated. Results of these studies may provide a foundation for further study of potential clinical applications in diagnosis, prognosis or therapeutics.


DSG3 is an HNC-associated gene

Using normal and tumor tissue from patients with HNC, we previously globally surveyed and identified several cancer-associated genes using differential display (Chang et al., 2005). A total of 30 differentially expressed bands were selected. Most of the genes were overexpressed in the tumor tissue, whereas a few were under-expressed. The abnormally expressed bands were extracted, re-amplified, cloned and subjected to sequencing analysis. The size of the cDNA ranged from 198 to 436 bp. After blasting their sequences through GenBank databases for identification search, we further interestingly found that two of the clones matched DSG3 (Unigene locus: Hs.1925). This gene was overexpressed in tumor tissues, suggesting it may play an important role in HNC carcinogenesis.

To verify DSG3 expression in HNC and to understand the tissue specificity of this gene in different malignant diseases, various cancer cell lines were examined by Western blot analysis, including oral fibroblast (ORF), keratinocyte, oral carcinoma cells (OECM1, KB, SCC25), nasopharyngeal carcinoma cells (BM1, 076), as well as other non-squamous malignancies: colon (CT-29), liver (SK, Mal), prostate (PC3, Dn145), bladder (U1) and neuron (SF268) cancer cells. As shown in the Figure 1a, ORFs and keratinocytes have non-detectable or basal levels of DSG3 expression, in contrast to high levels of this gene in other oral and nasopharyngeal cancer cells. Other non-HNC cells, however, has minimum levels of DSG3 expression (Figure 1b). These results indicate that DSG3 may be a tissue-specific molecule for the development of head and neck squamous cell carcinoma.

Figure 1

Differential expression of DSG3 in various HNC cells and other non-squamous malignant cells. Various cell lines were examined, including ORF, ORK, oral carcinoma cells (OECM1, KB, SCC25), nasopharyngeal carcinoma cells (BM1, 076) and other non-squamous malignancies: colon (CT-29), liver (SK, Mal), prostate (PC3, Dn145), bladder (U1) and neuron (SF268) cancer cells, as indicated in the top of each line. Cellular protein was extracted and subjected to Western blot analysis. Actin protein expression was determined in each cell line as internal control.

DSG3 is overexpressed in cancer tissues and correlated with clinical stages

To understand subcellular localization of DSG3 in normal and cancer cells, confocal microscopy examination was performed using immunofluorescent staining with anti-DSG3 antibody for normal ORFs, oral keratinocytes (ORK) and HNC cells (OECM1, BM1), whereas staining with 4,6-diamidino-2-phenylindole (DAPI) for nucleus was used as control. As shown in the Figure 2, DSG3 showed very low levels of expression in normal ORF and ORK cells, in contrast to high levels of expressions in HNC cells (OECM1 and BM1). Furthermore, in the highly expressed cancer cells, DSG3 was mainly located in the plasma membrane, which was inconsistent with previous report (Garrod et al., 2002; Yin and Green, 2004). To validate the clinical significance of DSG3 in HNC, normal and cancerous tissues from 56 patients with HNC were obtained for study. For each tissue sample, total RNA was extracted and subjected to reverse transcription–polymerase chain reaction (RT–PCR) analysis for DSG3 mRNA expression (Figure 3a). Using a cutoff value of a twofold greater expression in tumor versus normal tissues, 34 patients (61%) were found to have DSG3 protein overexpression in the tumor tissues (Figure 3b). In 40 pairs of normal and tumor tissues, the correlation coefficient between mRNA and protein was 85%, indicating that the protein levels were generally consistent with the mRNA expression levels. Pearson's χ2 method was used for statistical analysis to determine the potential association between DSG3 mRNA expression and disease status (Table 1). There was no statistical correlation of DSG3 level with cellular differentiation (P=0.153). However, significant correlations were found between DSG3 overexpression and T stage (P=0.009), N stage (P=0.047), overall stage (P=0.011), tumor depth (P=0.009) and extracapsular spread in LNs (P=0.044). These results suggest that DSG3 participate in cancer growth and invasion.

Figure 2

Immunofluorescent staining and confocal microscopy examination for DSG3 localization in normal (ORF and ORK) and cancer (OECM1, BM1) cells. DSG3 was detected using a DSG3-specific primary antibody with an FITC-conjugated secondary antibody and shown in green imaging. DAPI staining was performed for nuclear localization and shown in blue. Marge imaging was shown to indicate subcellular localization of DSG3 in a cell.

Figure 3

Overexpression of DSG3 in tumor (T) tissues compared to paired grossly normal (N) counterparts from patients with HNC. (a) DSG3 expression at the mRNA level as determined by RT–PCR. Expression of 18S RNA served as an internal control. (b) DSG3 expression at the protein level as determined by Western blot. Expression of actin protein served as an internal control.

Table 1 Association of DSG3 with clinicopathologic status

DSG3 knockdown inhibits cell growth and colony formation

To validate whether DSG3 functions in cell growth regulation, three different HNC cell lines (OECM1, KB, BM1) were used to examine changes in cellular phenotypes after DSG3 knockdown by RNA interference (RNAi). Two plasmids clones of DSG3-RNAi were designed; their sense and antisense hairpin structures are illustrated in Figure 4a. These two clones showed different levels of RNA expression, with DSG3-RNAi-1 (clone 2323) at minimum level and DSG3-RNAi (clone 2761) was much more significant of suppression, after transfection of these RNAi plasmids for 1 day (Figure 4b). As RNAi is a potent inhibitor, we use this plasmid clone for further study. After transfection with DSG3-RNAi for 4 days into OECM1, KB and BM1 cells, these cells were harvested for analysis. As shown in Figure 4c–e, DSG3 was significantly knocked down after 2 days of transfection compared to the vector control, suggesting the effectiveness of this RNAi design.

Figure 4

DSG3 expression was knockdown by RNAi. (a) Two plasmids clones of RNAi were designed: DSG3-RNAi-1 (clone 2323) and DSG3-RNAi (clone 2761), with their sense and antisense hairpin structures illustration. (b) Differential effectiveness of suppressing DSG3 expression in the two clones of DSG3-RNAi plasmids was shown in OECM1 cells for example. For each experiment, 5 × 105 cells were seeded in a 10-mm dish plate and transfected with two clones of RNAi or vector plasmids into OECM1 cells. Cells were further cultured for 1 day and the cellular proteins were extracted for the determination of DSG3 expression by Western blot analysis. (ce) Effects of DSG3-RNAi on DSG3 protein expression in OECM1, KB and BM1 cells, respectively. Cells were seeded with 5 × 105 cells, transfected with DSG-RNAi and further cultured for 4 days. Cellular proteins were extracted for the determination of DSG3 expression by Western blot analysis.

The cellular effects of DSG3-RNAi were next examined. The treatment of DSG3-RNAi resulted in a gradual and significant decrease in cell growth in all of three cell lines tested (Figure 5a–c). At day 4, cell growth was reduced to 21, 56 and 52%, respectively, for OECM1, KB and BM1 cells. Consistently, cellular colony formation was significant suppression by RNAi (Figure 6a). The numbers of cell colonies of DSG3 knockdown cells were reduced to 40, 57 and 31%, respectively, for OECM1, KB and BM1 cells (Figure 6b). Colony size and cell density were also lower than for the controls. Suppression of DSG3 expression apparently reduces cell growth and colony-forming ability in HNC cells.

Figure 5

DSG3 knockdown inhibits cell growth in three HNC cell lines: (a) OECM1, (b) KB and (c) BM1. For each experiment, either 5 × 105 cells (OECM1 and BM1) or 1 × 106 cells (KB) were seeded in a 10-mm dish plate, transfected with DSG3-RNAi or vector plasmids and further cultured for 4 days. Cell numbers were determined every 24 h. Experiments were performed in triplicate.

Figure 6

DSG3 knockdown inhibits colony formation in three HNC cell lines. (a) Effects of DSG3-RNAi on colony formation, with experiments performed in duplicate. After transfection with DSG3-RNAi or vector plasmids, various numbers of cells (300–1000 for OECM1 and BM1, 1000–3000 for KB) were seeded in a six-well plate and incubated for 7 days to allow colony formation. Cell colonies were visualized by 5% crystal violet staining. (b) Quantitative determination of colony numbers in DSG3-RNAi or vector treatment cells. Cell number over 50 in a cluster was defined as a signal colony. Cell samples with initial seeding of 300 in OECM1 or BM1, and 1000 in KB were determined in duplicate as indicated at the bottom of the figure.

DSG3 knockdown suppresses cell migration and invasion

Given that DSG3 is a desmosome protein involved in cell-to-cell adhesion, we examined whether DSG3 also participates in cell migration and invasion. Cell migration and invasion were determined using in vitro wound healing and Matrigel Transwell invasion assays. Monolayer cultures of HNC cells, either transfected with DSG3-RNAi or control vector, were wounded by a micropipette tip. Cell migration toward the wounded area was observed (Figure 7). Although the different cell lines migrated at different rates, fewer of the DSG3-RNAi transfectants in each cell line migrated toward the wounded area compared to controls. At 18 h, the wounded area was almost completely covered by OECM1 and BM1 control cells, whereas only a few transfected cells moved into the area. KB cell migration was much slower than that of the other two cell lines, although the control cells appeared less dense than the DSG3-RNAi-treated cells, indicating that the control cells were starting to migrate toward the wounded area. These results indicate that DSG3 has a positive regulatory role in cell migration. For the invasion assay, DSG3-RNAi-transfected cells were seeded in the upper chamber of Matrigel-coated Millicells. The number of cells invading the lower chamber was determined daily. As shown for OECM1, KB and BM1 cells, after 1 day of RNAi treatment, the number of invasive cells in each cell line was reduced (Figure 8a–c). By 2 days, there was more than a 50% inhibition of invasion in KB and BM1 cells. By day 5, the invading cells were reduced to 30, 43 and 46%, respectively, in OECM1, KB and BM1 cells. These results indicate that DSG3 also participates in cell invasion. Apparently, knockdown of DSG3 expression suppresses the ability of HNC cells to migrate and invade.

Figure 7

DSG3 knockdown reduces cell migration in three HNC cell lines. After transfection of either DSG3-RNAi or the vector plasmids, 2 × 106 transfectants were seeded per well in a six-well fibronectin-coated plate and incubated for 6 h to allow monolayer cell formation. Cells were wounded by a micropipette tip and then incubated in the presence of 1% FBS culture medium for 24 h. Cell migration toward the wounded area was observed and photographed.

Figure 8

DSG3 knockdown reduces cell migration in three HNC cell lines: (a) OECM1, (b) KB and (c) BM1 cells. A total of 2 × 105 transfectants were seeded per well in a 24-well plate coated with Matrigel in a Millicell invasion chamber and incubated at 37°C for 5 days. The numbers of cells that migrated through the Matrigel to the lower chamber were determined every 24 h. Each experiment was performed in triplicate.

DSG3 knockdown inhibits tumor growth in vivo

To investigate the effects of DSG3-RNAi treatment on tumor growth in vivo, we established a xenograft KB tumor in BALB/C nude mice. After injection of KB tumor cells, DSG3-RNAi or vector plasmids were administered intravenously into five animals per treatment group and continuously monitoring for a total of 60 days. Figure 9a shows the average tumor volume over the 56 days of study between these two groups. Tumors in mice given DSG3-RNAi had sustained, significant growth arrest compared to the controls. On average, DSG3-RNAi treatment decreased tumor growth by 74% at day 37 (P=0.003) and by 83% at day 56 (P<0.001).

Figure 9

Effect of DSG3-RNAi on tumor xenograft BALB-C/nu mice. (a) The mice were subcutaneously injected with 107 KB cancer cells. After 3 days, 50 μg of either DSG3-RNAi plasmid or the control vector plasmid in a total of 50 μl normal saline solution was injected intravenously, followed by a booster injection with 25 μg of plasmid twice a week for 6 weeks. The tumors were monitored for total of 60 days. The tumor volume was calculated as length × width × height, each dimension having been measured by calipers. Tumor volume was significantly greater at 56 days in the mice given control vector as compared with those given DSG3-RNAi (two-sided t-test, P<0.001). (b) Six weeks after grafting, tumors were removed and subjected to IHC analysis of DSG3 (radish color) and counterstained with HE (blue). Hair follicle tissues were also stained as control to ensure the successfulness of IHC staining.

In order to examine whether the suppression of tumor growth is associated with the effect of injected DSG3-RNAi, the xenograft tumors were dissected to examine DSG3 expression by immunohistochemical (IHC) method. As shown in the Figure 9b, the DSG3 protein was knocked down in the RNAi-injected xenograft tumor compared to the higher levels in the vector-injected tumor, and the expressions of DSG3 in hair follicles of the two mice were used as internal controls. These results indicated that the injected DSG3-RNAi leading to the inhibition of tumor growth was associated with DSG3 suppression. These results demonstrate the significant effect of DSG3 suppression on HNC growth and suggest that DSG3 may be a molecular target for treatments aimed at decreasing oncogenesis.


DSG3 is one of the components in the desmosome. Desmosomes are button-like points of intercellular contact, which couple cytoskeletal elements to the plasma membrane at cell-to-cell or cell-to-substrate adhesions. Whereas the related adherent junctions associate with microfilaments at cell-to-cell interfaces, desmosomes are tailored to anchor stress-bearing intermediate filaments at sites of strong intercellular adhesion. The resulting supracellular scaffolding plays a key role in providing mechanical integrity to tissues. These cellular rivets are dynamic structures subject to transcriptional and post-translational regulatory signals, and also participate in cell morphogenesis, differentiation and proliferation (Green and Gaudry, 2000; Garrod et al., 2002; Yin and Green, 2004). A disorder related to DSG3 is pemphigus vulgaris (Amagai et al., 1991), an autoimmune diseases. In pemphigus vulgaris, autoantibodies against DSG3 have been found in patients' serum accompanied by pathological loss of cell-to-cell adhesion. Similarly, DSG3 autoantibodies have also been found in paraneoplastic pemphigus, an autoimmune blistering disease that occurs in association with neoplasms such as non-Hodgkin's lymphoma and chronic lymphocytic leukemia (Amagai et al., 1998). It has been proposed that the autoantibodies play a pathogenic role in the loss of keratinocyte adhesion and blistering.

Reports with regard to clinical association of DSG expression with cancer are few and vary. Using Western blot analysis, Harada et al. (1996) found various abnormal DSG expression isoforms (including 150 and/or 130 kDa of DSG1 and/or DSG3) in different squamous carcinoma cell lines. By an IHC study, Krunic et al. (1996) found that DSG1 and DSG3 staining are, in general, distributed unevenly in subcellular compartment and reduced in skin squamous cell carcinoma in comparison to keratoacanthoma. Using similar method, Imai et al. (1995) found a reduction of DSG expression in advanced oral carcinoma. However, a recent study by Gordon et al. (2003) found that the expression of one or more myoepithelial markers, such as DSG3, was associated with a high invasive capacity compared with cells with a pure luminal phenotype. Recently, a large-scale microarray study by Chung et al. (2004) using 60 head and neck squamous cell carcinoma samples found that a subtype of tumors contained genes involved in desmosome function, such as DSG3, are overexpressed in poor outcome patients, which may associate with cancer-invasive and metastatic potential. In this study, we found that DSG3 is specifically overexpressed in head and neck squamous carcinoma cells (Figures 1, 2 and 3), but not in other non-squamous malignancy (Figures 1 and 2) or normal mucosa tissues (Figures 2 and 3), which is consistent with the finding by Chung et al. (2004). The conflicting data produced by various investigators may be owing to different assay methods as well as patient sampling. As short of specific antibody, earlier work in studying DSG expression cannot differentiate isoforms (Imai et al., 1995; Harada et al., 1996; Krunic et al., 1996), whereas the clone of antibody we used is able to specifically recognize DSG3 extracellular domain with high affinity, thus minimizing the crossreactivity with other DSG isoforms. Furthermore, most investigators examine protein expressions between different tumor tissues (Imai et al., 1995; Harada et al., 1996; Krunic et al., 1996; Gordon et al., 2003), but we analyse DSG3 mRNA and protein levels in the paired grossly normal mucosa and cancer tissues obtained form the same patients (Figure 3). These results provide clearer information regarding the change of DSG3 expression after cellular transformation.

In this study, we first demonstrated that DSG3 is overexpressed in HNC and that the overexpression is correlated with clinical characteristics of the tumor. The correlation of DSG3 with clinical T stage (Table 1) may be explained on a cellular level by the role of DSG3 in the regulation of growth and colony formation (Figures 5 and 6). DSG3's correlation with N stage (Table 1) corresponds on the cellular level to its participation of DSG3 in the regulation of cell migration and invasion (Figures 7 and 8). The fact that knockdown of DSG3 significantly inhibits xenograft tumor growth suggests that it may function in HNC carcinogenesis (Figure 9).

Desmosomes comprise proteins from at least three distinct gene families: cadherins, armadillo proteins and plakins. The desmosomal cadherins are further divided into DSG1–3 and desmocollins (DSC1–3). The armadillo proteins include plakoglobin, plakophilins (PKP1–3) and p0071. The plakin family proteins include desmoplakins, plectin and the cell envelope proteins envoplakin and periplakin. These desmosome proteins are kinetically coordinated to assemble desmosome. For example, DSG3 can compensate the loss of DSG3-mediated adhesion function in DSG3 knockout mice, suggesting that one DSG can compensate for loss function of another molecule (Hanakawa et al., 2002b). However, the levels of DSG must maintain certain dynamic levels to form desmosome structures and maintain intact cell–cell borders. It has been shown that DSG1 and DSG3 expressed at low level were able to incorporate into desmosome, but at a high-level expression of DSG1 in contrast, they disrupted desmosome structure (Hanakawa et al., 2002a). Therefore, the levels of DSG3 may not directly correlate with desmosome numbers or associated with cell–cell contact strength. The mechanism underlying DSG3's association with carcinogenesis is still unclear. Perhaps, DSG3 expression indicates the need of cancer cells in growth, whereas overexpression of this gene renders a ‘quenching’ effect, which binds a subset of associated desmosome proteins, thus disrupting the stability or kinetic of the desmosome structure, leading to facilitation of cell migration and invasion.

Understanding the mechanisms underlying oncogenesis has wide-ranging implications for targeting the treatment of cancer. In particular, treatment directed at molecules such as DSG3 that are overexpressed in tumor tissues may minimize cytotoxic effects on normal cells. Our study provides a foundation for further investigation into the manipulation of DSG3 in the treatment for HNC.

Materials and methods

Patients, tissues and cells

Fifty-six consecutive HNC patients seen at the Otorhinolaryngology or Head and Neck Surgery clinics at Chang Gung Memorial Hospital (Taoyuan, Taiwan) were enrolled. Written informed consent was obtained from all participating patients. The characteristics of these HNC patients were summarized in the Table 2. The patients included 52 (93%) males and four (7%) females, with an age range of 32–74 years (median 51 years). A total of 44 (79%) consumed alcohol, 51 (91%) smoked tobacco and 50 (89%) chewed betel nut. Biopsies of cancer and grossly normal mucosa tissue were obtained from each subject before chemo- or radiotherapy. Part of the diseased tissue was dissected and frozen immediately in liquid nitrogen until used for molecular assays. The remaining sample was fixed in formalin and processed for routine histopathologic examination. All patients had undergone a series of clinical evaluations, including assessment of tumor extension before treatment and the tumor response to therapy. All cancers were histologically graded: well differentiated, moderately differentiated, poorly differentiated or undifferentiated according to the World Health Organization classification. The diagnosis, clinical staging and identification of the anatomic site of the HNC were based on the American Joint Committee on Cancer Tomor Node Metasis (TNM) Classification of Malignant Tumors.

Table 2 Characteristics of head and neck cancer patients recruited in this study

HNC cell lines, including oral cancer-derived KB (American Type Culture Collection number: CCL17), OECM1 (Yang et al., 2001) and nasopharyngeal cancer-derived BM1 (Liao et al., 1998), were used. KB cells were grown in minimal essential medium (MEM), whereas OECM1 and BM1 were maintained in Roswell Park Memorial Institute 1640 medium supplemented with 10% fetal bovine serum (FBS) and antibiotics (100 U/ml penicillin, 100 U/ml streptomycin and 0.25 μg/ml amphotericin B) at 37°C, 5% CO2.

Differential display and gene identification

The RNA from seven pairs of grossly normal and tumor tissue were extracted for differential display analysis, as described previously (Chang et al., 2005). Briefly, approximately 20 mg of tissue was obtained from each sample for RNA extraction using TRIzol reagent (Giboc BRL, Rockville, MD, USA). Differential display was performed using reagents from RNA Image Kits (GenHunter, Brockline, MA, USA). cDNA was synthesized from 200 ng total RNA by reverse transcriptase using one of three anchor primers, H-T11A/G/C. The cDNA was PCR-amplified using one of the anchor primers and eight arbitrary primer sets, H-AP1 to H-AP8. The PCR fragments were separated by electrophoresis and the differentially expressed cDNA fragment was re-amplified by PCR reaction, following cloning into the thymine/adenine cloning vector plasmids (Invitrogen, Carlsbad, CA, USA). Each plasmid clone was purified and subjected to automatic sequencing by using the vector-specific primer. Alignment of insert sequences was performed and the BLAST program was used for homology searches in the NCBI GenBank database.

RNA extraction and RT–PCR

The mRNA expression of DSG3 was determined by RT–PCR. Total RNA was extracted by using TRIzol reagent (Gibco BRL, Rockville, MD, USA) following the manufacturer's instructions. The reverse transcription reaction was performed by incubation of a reaction mixture containing 30 ng RNA, 100 pmol of poly-T oligonucleotide, 4 U of reverse transcriptase (AMV, HT Biotech Ltd, UK), 10 U of RNase inhibitor (CalBiochem, San Diego, CA, USA) and 25 mM dNTP in a total of 30 μl reaction buffer at 37°C for 1 h. For each PCR run, a master mix was prepared on ice with 1 × TaqMan buffer, 5 mM MgCl2, 200 μ M dNTP, 300 nM of each primer, 1 U of AmpliTaq Gold DNA Polymerase and 3 μl of cDNA solution in a total of 30 μl. PCR reactions were carried out with DSG3 primers (forward primer 5′-IndexTermGTG TGA GAT GCC ACG CAG CT-3′ and reverse primer 5′-IndexTermGCC CCA TCA GCG TAG TCC TT-3′) for a total of 35 cycles at 94°C for 40 s, 56°C for 40 s and 72°C for 1 min, using an ABI Prism 5700 Sequence Detection System (Perkin-Elmer Applied Biosystems, Forster City, CA, USA). The PCR products were analysed by 2% agarose gel electrophoresis, stained with ethidium bromide, photographed by illuminating with 254 nm UV and each band density was determined. To define the relative expression of DSG3, the band density of each tumor sample was compared with that from the normal tissue from the same patient, after normalization with an internal control (18S RNA expression), similar to the previously described procedure (Chang et al., 2002). DSG3 expression in tumor tissue greater than twofold the normal counterpart was defined as overexpression.

Protein extraction and Western blot analysis

To determine the DSG3 protein level, subcellular proteins were extracted, and the enriched membrane fraction proteins were used. Tissue samples (50 mg) were homogenized in 300 μl of a lysis buffer (50 mM N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid, pH 7.5, 150 mM NaCl, 10% glycerol, 1.5 mM MgCl2) and incubated on ice for 30 min. After centrifuging at 14 000 g for 30 min at 4°C, the supernatant was collected and saved as cytosol proteins. The remaining cell pellet was lysed in the same buffer with additional 1% Triton X-100 and incubated on ice for 1 h. After centrifuging at 14 000 g for 30 min at 4°C, the supernatant was collected and saved as enriched membrane proteins. For Western blot analysis, 20 μg of the membrane proteins were separated by 8% sodium dodecyl sulfate–polyacrylamide gel electrophoresis following transferring to nitrocellulose membrane. The membrane was hybridized with anti-DSG3 antibody (clone 5H10, Santa Cruz Biotech, Santa Cruz, CA, USA), then incubated with secondary antibodies conjugated with horseradish peroxidase (Santa Cruz Biotech, Santa Cruz, CA, USA). The protein image was developed by using a Renaissance Western Blot Chemiluminescence Reagent kit (NEN Life Science Products, Boston, MA, USA) following autoradiography. To define the relative expression of DSG3 in clinical samples, the band density of each tumor sample was compared with that of the normal tissue from the same patient, after normalization with an internal control (actin protein expression). DSG3 expression in tumor tissue greater than twofold of the normal counterpart was defined as overexpression.

Immunofluorescent staining and Confocal microscopy

Before cell seeding, the glass coverslips were coated with 1% poly-L-lysine (Sigma) for 10 min in room temperature. After air drying of the coverslips, 3 × 105 of the cells were seeded into a coverslip, following incubation in 37°C overnight. After washing with PBS, cells were fixed in 3.7% formaldehyde in PBS buffer for 10 min. The fixed cells were then permeabilized with a permeation buffer (0.05% Triton X-100 in PBS) at 4°C for 10 min, and blocked with 1% FBS in PBS at 37°C for 30 min. The coverslips were incubated with anti-DSG3 antibody (clone 5H10, Santa Cruz Biotech, Santa Cruz, CA, USA), rinsed and stained with fluorescein isothiocyanate (FITC)-conjugated secondary antibody. Coverslips were mounted with mounting medium containing DAPI fluorescence (Vector laboratories, Burlingame, CA, USA). Fluorescence was visualized on a confocal laser microscope (Leica TCS Sp2 MP).

Statistical analysis

The Pearson χ2 test was used to examine the association of DSG3 expression and clinicopathologic features, including TNM stage and histopathologic characteristics. All P-values were two-sided, and the significance level was set at a P of <0.05.

Cloning of RNAi plasmid

The pTOPO-U6 vector was used to construct a DSG3-RNAi plasmid similarly as described previously (Chang et al., 2006), in which a 22-nt sense and antisense hairpin oligonucleotide was generated complementary to DSG3 mRNA. This hairpin oligonucleotide included two restriction enzyme cleavage sites corresponding to the blunt end and overhang end matching EcoRV- and BbsI-digested pTOPO-U6. Two RNAi sequences were designed, naming DSG3-RNAi and DSG3-RNAi-1, in which the first RNAi nucleotide complementary to DSG3 mRNA 2761 and 2323 bp, respectively. The sequence of the DSG3-RNAi oligonucleotide was 5′-IndexTermUUG UUA AGT GCC AGA CUU GAA GCU UGA AGU CUG GCU CUU AAC AAU UUU U-3′, and DSG3-RNAi-1 was 5′-IndexTermUCG GAG CAG CCA CUG GAG GAA GCU UGC UCC AGU GGC UGC UCC GAU UUU U-3′. Ligation between the hairpin oligonucleotides and pTOPO-U6 at these cloning sites produced the DSG3-RNAi plasmid.

Plasmid transfection and cell growth assay

For plasmid transfection, cells were seeded at a density of 5 × 105 in a 100-mm dish and cultured for 16 h. When 60% culture confluence was reached, cells were transfected with 6 μg of DSG3-RNAi or the vector plasmids using Lipofectamine 2000 reagent (Invitrogen, Carlsbad, CA, USA) in Opti-MEM medium (Invitrogen, Carlsbad, CA, USA) for 16 h, after which the medium was replaced with fresh complete medium. The transfection efficiency is approximately 70% in all three types of the HNC cell lines (OECM1, KB, BM1). Alteration of cellular phenotype was determined. To determine cell growth, cell numbers were counted daily with a hemocytometer.

Colony formation assay

A total of 300–3000 cells, transfected with either DSG3-RNAi or vector plasmids, were seeded onto six-well plates and allowed to grow without moving for 7 days in complete culture medium containing 30% FBS. The number of cell colonies was counted after staining with 5% crystal violet for 15 min.

Cell migration assay

Cell migration ability was evaluated by an in vitro wound-healing assay. After transfection with either DSG3-RNAi or vector plasmid, the cells were reseeded to a six-well fibronectin-coated plate (Coring Incorporated, Coring, NY, USA) and incubated for 6 h to allow monolayer cell formation. Cells were wounded by a micropipette tip and then incubated in the presence of 1% FBS culture medium for 24 h in a tissue culture incubator. Cell migration toward the wounded area was observed and photographed.

Cell invasion assay

A cell invasion assay was performed by using a BioCoat Matrigel (Becton Dickinson Biosciences, Bedford, MA, USA) and Millicell invasion chamber (Millipore Corporation, Bedford, MA, USA). The Matrigel, 100 μl diluted at 5 mg/ml, was coated onto the membrane of the Millicell upper chamber with a pore size of 8 μm in a 24-well plate for 4 h at 37°C, followed by washing with PBS. Cells, transfected with either DSG3-RNAi or vector plasmids in 1% FBS medium, were seeded 1 × 105 in the upper chamber. The lower chamber contained complete culture medium, which included 10% FBS in order to trap cell invasion. The invasion ability was determined for 4 days by counting cells in the lower chamber, which had previously passed through the Matrigel-coated membrane.

Mice and the xenograft tumor

A total of 10 female BALB/C null mice at 5 weeks of age were used for the experiment. The mice were subcutaneously injected in the upper portion of the hind limb with 107 KB cells. Three days after tumor cell xenografting, the mice were randomly divided into two groups of five mice each. The experimental group was injected intravenously with 50 μg of DSG3-RNAi plasmid in 50 μl PBS, followed by a booster of 25 μg of the plasmid in 25 μl PBS twice a week for a total of 6 weeks. The control group was injected with pTOPO-U6 vector plasmid following the same schedule as the experimental group. The mice were monitored for 56 days and the tumor dimensions were measured daily with calipers. The tumor volume was calculated as length × width × height.

Immunohistochemistry of xenograft tumor

Six weeks after grafting, tumors were removed and subjected to immunohistochemistry (IHC) analysis. Hair follicle tissues were also stained as internal control. Tissue samples were fixed in formaldehyde solution and embedded in paraffin. The sections were dewaxed twice in xylene for a total 10 min, and rinsed in alcohol and gradient alcohol/water mixtures for 5 min each solution. After rinsing in distilled water, antigen retrieval was carried out by boiling in 10 mM citrate buffer (pH 6.0) for 10 min. Endogenous peroxidase was blocked in hydrogen peroxide for 5 min and washed with PBS twice for 5 min. IHC analysis was performed using a streptavidin–biotin complex system (LSAB2 system; Dako, Carpinteria, CA, USA). Slides were then incubated with anti-DSG3 antibody (clone 32-6300, Zymed Laboratories Inc, South San Francisco, CA, USA) overnight at 4°C. Sections were washed and incubated with biotinylated secondary antibody for 1 h at room temperature, and developed the color with 3-amino-9-ethyl carbazole (AEC) substrate chromogen (LSAB2 system) for 10 min at room temperature. Each sample was counterstained with hematoxylin (HE) according to the manufacturer's suggested protocol (Zymed Laboratories Inc). The staining reactions were determined by microscopic examination.


  1. Amagai M, Klaus-Kovtun V, Stanley JR . (1991). Autoantibodies against a novel epithelial cadherin in pemphigus vulgaris, a disease of cell adhesion. Cell 67: 869–877.

    CAS  Article  Google Scholar 

  2. Amagai M, Nishikawa T, Nousari HC, Anhalt GJ, Hashimoto T . (1998). Antibodies against desmoglein 3 (pemphigus vulgaris antigen) are present in sera from patients with paraneoplastic pemphigus and cause acantholysis in vivo in neonatal mice. J Clin Invest 102: 775–782.

    CAS  Article  Google Scholar 

  3. Chang JT, Chen IH, Liao CT, Wang HM, Hsu YM, Hung KF et al. (2002). A reverse transcription comparative real-time PCR method for quantitative detection of angiogenic growth factors in head and neck cancer patients. Clin Biochem 35: 591–596.

    CAS  Article  Google Scholar 

  4. Chang JT, Lu YC, Chen YJ, Tseng CP, Chen YL, Fang CW et al. (2006). hTERT phosphorylation by PKC is essential for telomerase holoprotein integrity and enzyme activity in head neck cancer cells. Br J Cancer 94: 870–878.

    CAS  Article  Google Scholar 

  5. Chang JT, Wang HM, Chang KW, Chen WH, Wen MC, Hsu YM et al. (2005). Identification of differentially expressed genes in oral squamous cell carcinoma (OSCC): overexpression of NPM, CDK1 and NDRG1 and underexpression of CHES1. Int J Cancer 114: 942–949.

    CAS  Article  Google Scholar 

  6. Chung CH, Parker JS, Karaca G, Wu J, Funkhouser WK, Moore D et al. (2004). Molecular classification of head and neck squamous cell carcinomas using patterns of gene expression. Cancer Cell 5: 489–500.

    CAS  Article  Google Scholar 

  7. Clayman GL, Lippman SM, Laramore GE, Hong WK . (1996). Cancer Medicine In: Holland JF, Frei E, Bast RC, Kufe DW, Morton DL, Weichselbaum R (eds). Williams and Wilkins: Philadelphia, pp. 1645–1709.

    Google Scholar 

  8. Decker J, Goldstein JC . (1982). Risk factors in head and neck cancer. N Engl J Med 306: 1151–1155.

    CAS  Article  Google Scholar 

  9. Garrod D, Merritt AJ, Nie Z . (2002). Desmosomal cadherins. Curr Opin Cell Biol 14: 537–545.

    CAS  Article  Google Scholar 

  10. Gordon LA, Mulligan KT, Maxwell-Jones H, Adams M, Walker RA, Jones JL . (2003). Breast cell invasive potential relates to the myoepithelial phenotype. Int J Cancer 106: 8–16.

    CAS  Article  Google Scholar 

  11. Green KJ, Gaudry CA . (2000). Are desmosomes more than tethers for intermediate filaments? Nat Rev Mol Cell Biol 1: 208–216.

    CAS  Article  Google Scholar 

  12. Hanakawa Y, Amagai M, Shirakata Y, Yahata Y, Tokumaru S, Yamasaki K et al. (2002a). Differential effects of desmoglein 1 and desmoglein 3 on desmosome formation. J Invest Dermatol 119: 1231–1236.

    CAS  Article  Google Scholar 

  13. Hanakawa Y, Matsuyoshi N, Stanley JR . (2002b). Expression of desmoglein 1 compensates for genetic loss of desmoglein 3 in keratinocyte adhesion. J Invest Dermatol 119: 27–31.

    CAS  Article  Google Scholar 

  14. Harada H, Iwatsuki K, Ohtsuka M, Han GW, Kaneko F . (1996). Abnormal desmoglein expression by squamous cell carcinoma cells. Acta Dermatol Venereol 76: 417–420.

    CAS  Google Scholar 

  15. Imai K, Kumagai S, Nakagawa K, Yamamoto E, Nakanishi I, Okada Y . (1995). Immunolocalization of desmoglein and intermediate filaments in human oral squamous cell carcinomas. Head Neck 17: 204–212.

    CAS  Article  Google Scholar 

  16. Krunic AL, Garrod DR, Smith NP, Orchard GS, Cvijetic OB . (1996). Differential expression of desmosomal glycoproteins in keratoacanthoma and squamous cell carcinoma of the skin: an immunohistochemical aid to diagnosis. Acta Dermatol Venereol 76: 394–398.

    CAS  Google Scholar 

  17. Liao CT, Chang JT, Wang HM, Chen IH, Lin CY, Chen TM et al. (2004). Telomerase as an independent prognostic factor in head and neck squamous cell carcinoma. Head Neck 26: 504–512.

    Article  Google Scholar 

  18. Liao SK, Perng YP, Shen YC, Chung PJ, Chang YS, Wang CH . (1998). Chromosomal abnormalities of a new nasopharyngeal carcinoma cell line (NPC-BM1) derived from a bone marrow metastatic lesion. Cancer Genet Cytogenet 103: 52–58.

    CAS  Article  Google Scholar 

  19. Vokes EE, Weichselbaum RR, Lippman SM, Hong WK . (1993). Head and neck cancer. N Eng J Med 328: 184–194.

    CAS  Article  Google Scholar 

  20. Yang CC, Tu SF, Chang RC, Hsu SF, Wu CH . (2001). In vitro cellular response of retinoic acid treated human oral cancer cell lines. Zhonghua Yi Xue Za Zhi (Taipei) 64: 357–363.

    CAS  Google Scholar 

  21. Yin T, Green KJ . (2004). Regulation of desmosome assembly and adhesion. Semin Cell Dev Biol 15: 665–677.

    CAS  Article  Google Scholar 

Download references


This work was supported by Chang Gung Memorial Hospital Grant (CMRPD32017 and CMRPD140141) and National Science Research Grant of Taiwan (NSC 94-2745-B-182-004-URD). We thank Mary Jeanne Buttrey, MD, for correction of the manuscript.

Author information



Corresponding author

Correspondence to A-J Cheng.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Chen, YJ., Chang, J., Lee, L. et al. DSG3 is overexpressed in head neck cancer and is a potential molecular target for inhibition of oncogenesis. Oncogene 26, 467–476 (2007).

Download citation


  • DSG3
  • head neck cancer
  • clinical association
  • migration and invasion
  • therapeutic target

Further reading


Quick links