Introduction

Squamous cell carcinoma (SCC) of the larynx is a malignant neoplasm that shows different incidences throughout the world. The Mediterranean basin shows one of the highest incidences and, in Southern Europe, it is the sixth most common cancer among men [19]. The molecular mechanisms involved in the pathogenesis of these tumors are not well known. In recent years, the participation of cell-cycle regulatory elements in the development and progression of human tumors has gained interest [7]. Particularly, we and others have demonstrated the frequent alterations of different cell-cycle regulatory genes including CCND1, p16INK4a, p53, and p21Waf1 in laryngeal carcinomas [8–10, 18], indicating that deregulation of cell-cycle mechanisms may play an important role in the malignant transformation of laryngeal epithelium.

CDK4 was initially identified as the catalytic subunit present in the G1 CDK/Cyclin D complexes of mammalian cells [16] and, similarly to other G1 regulatory genes, may have oncogenic potential [5]. CDK4 has been mapped to 12q13, a chromosomal region that includes a number of genes such as MDM2, GADD153, SAS, and GLI, which are frequently amplified in different sarcomas of bone and soft tissues and gliomas [6, 11, 23]. CDK4 overexpression, independent of gene amplification, seems to be a more frequent phenomenon, and it has been observed in different human tumors [13, 15]. Gene mutations in the p16INK4 binding domain of CDK4 have been also identified as alternative oncogenic events in rare cases of familial and sporadic malignant melanoma with no alterations of p16INK4 gene [4, 24, 26, 27]. The role of CDK4 in the pathogenesis of head and neck carcinomas is not well known, although a putative prognostic value of CDK4 and CCND1 protein overexpression was described in laryngeal carcinomas [2].

To determine whether CDK4 mRNA overexpression was the result of gene amplification, we analyzed CDK4 mRNA expression and gene status in a series of primary SCC of the larynx. The results were compared with clinical and pathological features of the patients and CCND1 mRNA expression levels in the tumors. Our results indicate that CDK4 mRNA is frequently overexpressed in laryngeal carcinomas independently of obvious genetic alterations. CDK4 mRNA deregulation occurs early in the development of the tumors. However, the coordinate overexpression of CDK4 and CCND1 mRNAs could be implicated in the progression of these neoplasms.

Materials and methods

Patients and tissues

Fresh tissue was obtained from 60 patients who underwent surgery for SCC of the larynx at our institution. Tumor and normal samples were dissected from the surgical specimens, snap-frozen in liquid nitrogen, and stored at −80°C until used. The remaining of the specimen was routinely fixed and processed for histopathological analysis. Clinical information was retrieved from the clinical charts of the patients and is summarized in Table 1.

Table 1 Clinicopathological characteristics of the patients and CDK4 and CCND1 expression
Full size table

RNA extraction and Northern-blot analysis

Total RNA was isolated from frozen tissues by guanidine isothiocyanate extraction and cesium chloride gradient centrifugation as previously described. Tissue sections were evaluated histologically to ensure that neoplastic cells accounted for, at least, 75% of the total cell number [8]. Fifteen micrograms of total RNA was electrophoresed on a denaturing 1.2% agarose formaldehyde gel and transferred to Hybond-N membranes (Amersham, Buckinghamshire, UK). The membranes were prehybridized, hybridized at 45°C with the full cDNA CDK4 probe [26], and washed at 55°C. Probe was radiolabeled using a random primer DNA labeling kit (Promega, Madison, WI, USA) with α-32P dCTP. Quantification of the hybridization signals was obtained by optical densitometry with the UVP GelBase software package as described elsewhere [8]. Cases with a tumor/normal ratio higher than twofold were considered to overexpress CDK4. After stripping, the membranes were rehybridized with the CCND1 probe and analyzed in the same way.

Quantitative RT-PCR

cDNA was synthesized from 1 μg of DNAase-treated total RNA using random hexamers following the directions of the manufacturer (Taqman Reverse Transcription Reagents, Applied Biosystems). Quantitative PCR was performed to assess human CCND1 and CDK4 gene expression using ABI PRISM 7900 Sequence Detector System and Assay on Demand Technology (Applied Biosystems). Assay IDs were Hs00277039_m1 and Hs00364847_m1 for CCND1 and CDK4, respectively. PCR reactions were done using Taqman Universal Master Mix (Applied Biosystems). Amplification conditions were 2 min at 50°C for UNG activation and 10 min at 95°C for TaqGold activation and predenaturation followed by 40 cycles at 95°C for 15 s and 60°C for 1 min. β-Glucoronidase gene (GUS; Applied Biosystems) was used as endogenous control as recommended by the manufacturer.

DNA extraction, Southern-blot, and SSCP analyses

DNA was extracted by conventional methods using proteinase-K digestion and phenol-chloroform extraction. Fifteen micrograms of DNA from normal and tumoral tissue from 31 matched normal and tumor tissues was digested with EcoRI and analyzed by Southern blot. All those cases were previously analyzed by Northern blot. Membranes were hybridized with the same probe used for Northern-blot analysis, at 42°C, and exposed at −70°C for 2 weeks.

Exon 2 of CDK4 was analyzed by nonisotopic SSCP in 38 of the 60 cases analyzed by Northern blot that included the 31 analyzed by Southern blot. A total of 0.5 μg of DNA was used as template in a standard PCR reaction containing PCR buffer, dNTPs 10 mM each, primers 1 mM each, 1 U of Taq DNA polymerase, and 5% DMSO in a reaction volume of 50 μl. Primers—modified from Zuo et al. [27]—were 5′-GCTGCAGGTCATACCATCCT-3′ (sense) and 5′-ATCACACCCCACCTATAGG-3′ (antisense). A modification of the touchdown PCR strategy (step-down PCR) was used. Conditions were: a denaturation step of 5 min at 94°C, 4 cycles at 94°C for 45 s (denaturation), at 66°C for 1 min (annealing), and at 72°C for 1 min (extension), 4 cycles with annealing at 63°C, 4 cycles with annealing at 60°C, 4 cycles with annealing at 57°C, and 30 cycles with annealing at 55°C followed by a last step of 5 min at 72°C. Sixteen microliters was digested with RsaI, and 2 μl of the digested product was electrophoresed at room temperature in 15% polyacrylamide gels and silver-stained according to a previously described method [9].

Statistical methods

Categorical data were analyzed by means of contingency tables. Probability was calculated with the Fisher’s exact test. Data were analyzed using the SPSS statistical package (SPSS, Chicago, IL, USA). Significance was accepted with an α-risk of 0.05.

Results

CDK4 mRNA expression

CDK4 mRNA expression was analyzed in 60 matched pairs of normal and tumor samples by Northern blot. A single 1.9-kb transcript was detected in all normal tissue samples. Forty-two carcinomas (70%) showed CDK4 mRNA overexpression with a range of two- to 13-fold of their normal counterpart (Fig. 1). Northern-blot results were validated by Q-RT-PCR in a series of nine cases (six with and three without CDK4 overexpression) with a 100% of concordance between Northern blot and Q-RT-PCR (p = 0.012, Fisher’s exact test). A similar number of tumors overexpressing CDK4 were observed in early and advanced stages. Similarly, no significant differences were observed in CDK4 expression levels according to location or grade of differentiation of the tumors (Table 1). The overall disease-free interval after surgery was independent of CDK4 overexpression. Cases with CDK4 overexpression had a mean disease-free time of 65 vs 56 months for cases without it, for the 47 cases with available follow-up.

Fig. 1
figure 1

a Northern blot analysis of CDK4 mRNA expression in nine matched samples of normal mucosa (N) and squamous cell carcinoma of the larynx (T). The numbers of the cases are arbitrary. Tumors 1, 2, 3, 5, 6, 7, and 9 show CDK4 mRNA overexpression, whereas similar expression levels in the normal and tumor samples were detected in cases 4 and 8. Ethidium bromide staining of the membranes after transfer is shown as loading control. b Southern blot analysis of the CDK4 gene of the same cases shown in A. Digestion with EcoRI produces two bands of 4.6 and 3.8 kb, respectively. No amplifications, rearrangements, or other gross genetic abnormalities are detected. Differences in intensity are due to unbalanced loading, as shown by the β-actin probe used as loading control

Full size image

CDK4 gene analysis

To determine if the CDK4 overexpression observed in the tumors was due to genetic alterations, 31 paired normal and tumor samples, previously analyzed for CDK4 overexpression by Northern blot, and including 24 cases with CDK4 overexpression, were examined by Southern blot. The two expected bands of 3.8 and 4.6 kb were observed in all cases. No gene amplification, deletions, or rearrangements were detected in any of the tumors (Fig. 1). In addition, a mutational analysis of CDK4 exon 2, which included the p16INK4a binding domain, was performed by SSCP in 38 out of the 60 cases analyzed by Northern blot, including all cases analyzed by Southern blot. No altered migration patterns were identified in any of the tumors analyzed.

CCND1 mRNA expression

CCND1 mRNA expression was examined in the same cases (Fig. 2). Sixteen cases showed CCND1 overexpression, with levels ranging from two- to 18-fold of their normal counterparts. Northern-blot results were validated by Q-RT-PCR in a series of ten cases (four with and six without CCND1 overexpression) with a 100% of concordance between both methods (p = 0.005, Fisher’s exact test; Fig. 3). A significant association was found between CCND1 and advanced stage of disease because 75% of the cases with overexpression were in stage 4 (p = 0.003, Fisher’s exact test) and between CCND1 and CDK4 mRNA overexpression: 15 of the 16 (94%) tumors with CCND1 overexpression also showed high expression levels of CDK4 (Table 2; p = 0.023, Fisher’s exact test). Only one case, in stage 3, showed CCND1 overexpression without simultaneous CDK4 upregulation.

Fig. 2
figure 2

Comparative analysis of CDK4 and CCND1 expression. The same blots were hybridized with the CDK4 probe (top), washed, and rehybridized with the CCND1 probe (middle). Ethidium bromide staining of the membranes after transfer is shown as loading control (bottom). The two cases with CCND1 overexpression (1 and 5) show also CDK4 overexpression. The case numbers are arbitrary and are not coincidental with those depicted in Fig. 1

Full size image
Fig. 3
figure 3

Amplification plots of Q-RT-PCR of CCND1 (top) and GUS (bottom). CCND1 curves that appear shifted to the left indicate CCND1 overexpression. Q-RT-PCR was done to ensure that mRNA degradation was not the cause of NB unbalances. GUS curves are tightly packed indicating that mRNA samples contain similar amounts of the control gene in the aliquots used for the analysis

Full size image
Table 2 Association between CDK4 and CCND1 overexpression
Full size table

Discussion

In this study, we examined the CDK4 gene structure and mRNA expression in a series of laryngeal carcinomas. CDK4 overexpression was a frequent phenomenon in these tumors (70%), and it was independent of gross structural alterations of the gene. High expression levels were already observed in early-stage tumors, suggesting that CDK4 mRNA upregulation is an early phenomenon in the development of SCC of the larynx. However, the similar incidence of CDK4 mRNA overexpression in early and advanced carcinomas indicates that other factors must be involved in the progression of the tumors.

CDK4 protein overexpression in laryngeal carcinomas was reported by Dong et al. They used immunohistochemistry to detect abnormally high numbers of cells expressing CDK4 protein. They also reported a correlation between CDK4, CCND1, and PCNA (proliferative) indexes and suggested that CDK4 overexpression had prognostic value in overall survival [2]. Unfortunately, we could not investigate CDK4 protein expression in those cases because there was no additional tissue available in a significant fraction of them.

The mechanisms leading to protein overexpression differ from gene to gene. CCND1 overexpression can be explained in most of the cases as the result of gene amplification [8], but this is not a constant for CDK4. CDK4 amplification was described in sarcoma and glioma cell lines [6, 11], as well as in primary tumors [13, 23], and it is usually associated with CDK4 mRNA overexpression [11, 21]. However, CDK4 mRNA upregulation has been described in the absence of CDK4 amplification in some high-grade gliomas, peripheral neuroendocrine tumors [21], malignant melanomas [13], and ovarian carcinomas [15]. Similarly, no CDK4 gene amplifications were detected in our study, including tumors with mRNA overexpression, indicating that other mechanism may be involved in CDK4 deregulation in SCCs of the larynx. The correlation observed by Dong et al. between CDK4 and proliferative indexes could suggest that CDK4 expression was merely a reflection of higher cell proliferation in neoplastic tissues. This is very unlikely because experimental evidence shows that CDK4 levels remain unchanged when mammalian keratinocytes are induced to proliferate [22] or to terminally differentiate [14], in contrast with other cell types such as macrophages [17]. In the present study, we observed that virtually all cases overexpressing CCND1 also showed CDK4 upregulation and were in stage 4. These findings suggest that concomitant overexpression of both genes may be an important phenomenon that can confer a selective growth advantage in the progression of laryngeal carcinomas. This human model parallels in vitro observations indicating that synthesis of both CCND1 and CDK4 is necessary for G1 progression but the rate-limiting factor in CDK activity is often determined by the levels of a cyclin rather than of a kinase [3, 17]. Although differences in expression could be the result of mRNA degradation, we think that this is very unlikely. Q-RT-PCR obtained from the same RNA samples used in Northern-blot analysis shows control gene curves to be densely packed, indicating that there is no significant variation in RNA degradation between samples, and differences in target signals represent real differences in the target gene mRNA (Fig. 3).

CDK4 itself may have oncogenic activity. However, it has been related to the ability of CDK4 to bind to, and sequester, p16INK4a or other cyclin-dependent kinase inhibitors rather than to its enzymatic activity [5]. Therefore, CDK4 overexpression is proposed as an alternative mechanism to p16INK4a inactivation in the tumorigenesis of different human tumors including malignant melanomas [13] or gliomas [6] because alterations of p16INK4a and CDK4 occur alternatively but not simultaneously in the same tumor. p16INK4a gene is also inactivated in a number of laryngeal carcinomas by mutations or hypermethylation associated with concomitant allelic deletion [9], and these alterations occur in tumors with normal retinoblastoma expression. The CDK4 overexpression observed in our series may also participate in the inactivation of Rb/p16INK4a pathway in laryngeal carcinomas.

An alternative oncogenic mechanism to CDK4 overexpression is the mutation in certain residues of exon 2. Mutations in this region render the CDK4 protein resistant to the inhibitory effect of p16INK4a by preventing the binding of these proteins. Different mutations have been described in occasional human malignant melanomas. However, these alterations seem to be rare in other tumors [1, 24, 26]. Concordant with previous observations in nonmelanoma tumors, no evidence of mutations was found in this series of laryngeal carcinomas.

Several mitogenic agents have been shown to induce CDK4 expression in different tumor models [3, 12, 16, 25]. Particularly, IL6 induces Rb phosphorylation via CDK4 overexpression in multiple myeloma cells [25]. Similarly, androgen treatment increases CDK4 expression both at mRNA and protein levels in an androgen-dependent prostatic cancer cell line [12]. Recently, a switch in CDK4 promoter response to transcription factors between normal and transformed mammary epithelial cells was described [20]. This widens the putative mechanisms that can alter the cell-cycle regulation in cancer cells and provides an explanation for gene overexpression in the absence of structural alterations of the gene.

In conclusion, our findings indicate that CDK4 mRNA overexpression is a frequent phenomenon in SCCs of the larynx and support a role of this particular CDK in the pathogenesis of these neoplasms. This overexpression, previously detected by others at the protein level, already occurs at mRNA level, indicating that it responds to transcriptional rather than translational mechanisms. The absence of gross genetic alterations suggests the participation of other factors, yet uncharacterized, in the enhanced CDK4 transcription observed in the neoplastic tissues, and, as it has been stated before, this events could reflect changes in the elements that regulate CDK4 transcription from normal to neoplastic cells. The association between CDK4 and CCND1 overexpression suggests a cooperation between these two factors in laryngeal carcinoma progression. CCND1 overexpression occurring upon a background of CDK4 upregulation in the neoplastic cells could produce a selective growth advantage in laryngeal squamous carcinoma progression.