Virchows Archiv

, Volume 458, Issue 4, pp 413–419

PDCD4 nuclear loss inversely correlates with miR-21 levels in colon carcinogenesis


  • Matteo Fassan
    • Department of Medical Diagnostic Sciences & Special Therapies, Surgical Pathology & Cytopathology UnitUniversity of Padua-Italy
  • Marco Pizzi
    • Department of Medical Diagnostic Sciences & Special Therapies, Surgical Pathology & Cytopathology UnitUniversity of Padua-Italy
  • Luciano Giacomelli
    • Surgical Pathology & Cytopathology UnitAzienda Ospedaliera di Padova
  • Claudia Mescoli
    • Department of Medical Diagnostic Sciences & Special Therapies, Surgical Pathology & Cytopathology UnitUniversity of Padua-Italy
  • Kathrin Ludwig
    • Department of Medical Diagnostic Sciences & Special Therapies, Surgical Pathology & Cytopathology UnitUniversity of Padua-Italy
  • Salvatore Pucciarelli
    • Department of Oncological and Surgical Sciences, Clinica Chirurgica IIUniversity of Padua-Italy
    • Department of Medical Diagnostic Sciences & Special Therapies, Surgical Pathology & Cytopathology UnitUniversity of Padua-Italy
    • Surgical Pathology & Cytopathology UnitAzienda Ospedaliera di Padova
    • Surgical Pathology UnitIstituto Oncologico Veneto–IOV-IRCCS
Original Article

DOI: 10.1007/s00428-011-1046-5

Cite this article as:
Fassan, M., Pizzi, M., Giacomelli, L. et al. Virchows Arch (2011) 458: 413. doi:10.1007/s00428-011-1046-5


Programmed cell death 4 (PDCD4) has recently been demonstrated to be a new tumor suppressor gene involved in colon carcinogenesis. PDCD4 immunohistochemical expression was assessed in 300 polypoid lesions of the colon mucosa (50 hyperplastic polyps [HP], 50 serrated adenomas [SA], 50 tubular adenomas with low-grade-intraepithelial neoplasia [LG-IEN], 50 tubular adenomas with high-grade-IEN [HG-IEN]), and in 50 colon adenocarcinomas (CRC). As normal controls, we considered 50 biopsy samples obtained from patients with irritable bowel syndrome (N). We further investigated PDCD4 messenger RNA (mRNA) levels by quantitative real-time polymerase chain reaction (PCR) in a different series of N, LG-IEN, HG-IEN, and CRC biopsy samples. miR-21 expression (an important PDCD4-expression regulator) was also determined by quantitative real-time PCR and in situ hybridization. Normal colocytes and HP featured strong PDCD4 nuclear immunostaining whereas a significantly lower PDCD4 nuclear expression was observed in dysplasia (low- and high-grade adenomas and SA) and invasive CRC. PDCD4 immunostaining and mRNA levels decreased significantly as the phenotypic changes occurring during colon carcinogenesis progressively increased (p < 0.001). As expected, miR-21 expression was significantly upregulated in preneoplastic/neoplastic samples, consistent with PDCD4 downregulation. These results consistently support the use of nuclear PDCD4 immunohistochemical downregulation as a novel biomarker for the diagnosis of dysplastic/neoplastic lesions in colon biopsy samples.


PDCD4miR-21Tumor suppressor geneColon biopsyColon cancer


Colorectal cancer (CRC) is one of the most common malignancies of the industrialized world [13]. Colorectal oncogenesis is a multistep process that passes through various grades of glandular dedifferentiation (dysplastic stage) and eventually progresses to invasive adenocarcinoma [37]. In the adenoma–carcinoma sequence, the stepwise phenotypic shift is determined by a concurrent stepwise accumulation of genetic alterations [4].

Programmed cell death 4 (PDCD4) has recently been characterized as a new tumor suppressor gene involved in the apoptotic machinery, suppressing cell transformation and invasion by interacting with translation initiation factors eIF4A and eIF4G [818]. Molecules regulated by PDCD4 include p21, Cdk4, ornithine decarboxylase, carbonic anhydrase II, and JNK/c-Jun/AP-1 [18]. PDCD4 expression is consistently downregulated in several human cancers and cancer cell lines [17, 1926].

Different mechanisms have been implicated in PDCD4 dysregulation. Among others, the oncogenic microRNA miR-21 (hsa-miR-21) has been shown to specifically target the PDCD4 3′ untranslated region (3′ UTR), which negatively regulates PDCD4 expression [19, 23].

In their seminal work, Mudduluru and colleagues demonstrated a significant PDCD4 downregulation in both colonic adenomas and CRC [17], and we have recently added evidence that PDCD4 is more frequently downregulated in metastases than in their corresponding primary CRCs [23]. PDCD4 expression has also been shown to be an independent prognostic factor in CRC patients [17].

Few translational studies have explored the use of PDCD4 expression to support routine biopsy findings in colon precancerous/cancerous lesions [18]. To investigate the diagnostic usefulness of PDCD4 immunostaining in the colorectal cancer setting, the protein’s expression was assessed in a series of biopsy samples representing the whole phenotypic spectrum of colon oncogenesis. We also examined miR-21 expression levels in dysplastic/neoplastic biopsy samples by quantitative real-time PCR (qRT-PCR) and in situ hybridization (ISH).

Materials and methods

cDNA microarray analysis

The Oncomine database and gene microarray analysis tool, a repository for published cDNA microarray data ( [27, 28], was explored (on 03 January 2011) for PDCD4 messenger RNA (mRNA) expression in non-neoplastic colonic tissues, colon adenomatous lesions, and primary CRC. Oncomine algorithms were used to perform a statistical analysis of the differences in PDCD4 expression because it allows for multiple comparisons between different studies [2729]. Only studies with analytical results with a p < 0.005 were considered.


The cases considered in the present study were retrospectively collected from the files of the Surgical Pathology & Cytopathology Unit at the University of Padua. The institute’s ethical regulations on research conducted on human tissues were followed.

In all, 360 endoscopic biopsy samples were obtained from patients with different types of sporadic colonic polyps, i.e., 50 hyperplastic polyps (HP), 50 serrated adenomas (SA), 65 tubular adenomas with low-grade intraepithelial neoplasia (LG-IEN), and 65 tubular adenomas with high-grade IEN (HG-IEN), and from 65 patients with CRC, all of them well-differentiated (G1) tumors. Another 65 normal colonic mucosa biopsy samples (N) were obtained from patients who underwent colonoscopy for irritable bowel syndrome (Clinica Chirurgica II–University of Padua). No samples were obtained from cases of familial adenomatous polyposis syndrome.

Histology and immunohistochemistry

All biopsy specimens were immediately fixed in 10% buffered formalin and embedded in paraffin. Serial histological sections 4–6 μm thick were obtained from each paraffin block selected. The histological sections were stained with hematoxylin and eosin. The original diagnosis was confirmed by histologically re-evaluating all cases. Immunohistochemical staining was done automatically (Ventana Benchmark XT system; Touchstone, AZ) [19, 25, 30] for PDCD4 (catalog #HPA001032; Atlas Antibodies, Stockholm, Sweden; 1:100) according to the manufacturer’s instructions. Sections were lightly counterstained with hematoxylin. Appropriate positive and negative controls were run concurrently. In cancer samples, the presence of positive stromal and inflammatory cells served as an internal control. PDCD4 nuclear expression was jointly scored by two pathologists (MF and CM) unaware of the patients’ clinical history. As described previously [17, 19], nuclear PDCD4 staining was scored on a four-tiered scale (score 0, no nuclear staining; score 1, ≥1 ≤30% positive nuclear staining; score 2, >30% ≤ 70% positive nuclear staining; score 3, ≥71% positive nuclear staining. A total of 50 N, 50 HP, 50 SA, 50 LG-IEN, 50 HG-IEN, and 50 CRC biopsy samples were considered.

Quantitative real-time polymerase chain reaction

Formalin-fixed paraffin-embedded biopsy samples were deparaffinized with xylene at 50°C for 3 min. Total RNA extraction was done using the RecoverAll kit (Ambion Inc, Austin, TX) according to the manufacturer’s instructions. qRT-PCR analysis was performed using the GeneAmp PCR 9700 thermocycler (Applied Biosystems, Foster City, CA), and gene expression levels were quantified using the ABI Prism 7900HT Sequence Detection System (Applied Biosystems), as previously described [31]. Primers’ sequences for PDCD4 were: forward 5′-ggcctccaaggagtaagacc-3′; reverse 5′-aggggtctacatggcaactg-3′. GAPDH was used as the internal control gene (forward 5′-aagggaaggttgctggatagg-3′; reverse 5′-cacatccacctcctccacatc-3′). The NCodeTM miRNA qRT-PCR method (Invitrogen, Carlsbad, CA) was used to detect and quantify mature miR-21 (primer sequence: 5′-CGGTAGCTTATCAGACTGATGTTGA-3′) according to the manufacturer’s instructions. Normalization was done with the small nuclear RNA U6B (RNU6B; Invitrogen). PCR reactions were run in triplicate, including no-template controls. The data were analyzed using the comparative CT method. Biopsy samples of 10 N, 10 LG-IEN, 10 HG-IEN, and 10 CRC were considered for the qRT-PCR study.

In situ hybridization

In situ hybridization was performed using the GenPoint™ Catalyzed Signal Amplification System (DakoCytomation) according to the manufacturer’s protocol. Briefly, slides were incubated at 60°C for 30 min and deparaffinized as described in [24, 32]. Sections were treated with Proteinase K (DakoCytomation) for 30 min at room temperature, rinsed several times with dH2O, and immersed in 95% ethanol for 10 s before air-drying. The slides were pre-hybridized at 49–56°C for 1 h with mRNA ISH buffer (Ambion) before incubation overnight at 49–56°C in buffer containing the 5′-biotin-labeled miR-21 miRCURY™ LNA detection probe (Exiqon, Woburn, MA) or the scrambled negative control probe (U6, Exiqon) at a final concentration of 200 nM. The slides were washed in both Tris-buffered saline with Tween (TBST) and GenPoint™ stringent wash solution (54°C for 30 min), then exposed to H2O2 blocking solution (DakoCytomation) for 20 min, and then further blocked in a blocking buffer (DakoCytomation) for 30 min before they were exposed to primary streptavidin–horseradish peroxidase (HRP) antibody, biotinyl tyramide, secondary streptavidin–HRP antibody, and DAB chromogen solutions, following the manufacturer’s protocol. The slides were then briefly counterstained in hematoxylin and rinsed with TBST and water before mounting. Biopsy samples of five N, five LG-IEN, five HG-IEN, and five CRC were considered for the ISH study.

Statistical analysis

Differences and correlations between groups were tested by applying the modified Kruskal–Wallis nonparametric test for trend, Pearson’s correlation, and the t test. P values <0.05 were considered significant. All statistical assessments were performed with STATA software (Stata Corporation, College Station, Texas, USA).


The PDCD4 gene is downregulated in colon adenomas and CRC

PDCD4 gene expression was analyzed by checking different publicly available colon microarray studies using the Oncomine database and gene microarray data analysis tools [27, 28]. The analysis considered the mRNA expression levels for each of the studies involved. Two studies showed a significant downregulation of PDCD4 mRNA expression levels in adenomas by comparison with normal colonic mucosa (Fig. 1a; Sabates-Bellver et al. [33], t test p = 1.48E−16, fold change = −2.524, 32 normal and 25 adenomas; Gaspar et al. [34], t test p = 2.49E−4, fold change = −1.709, 22 normal and 56 adenomas). In two other independent data sets, a significant decrease in PDCD4 mRNA expression levels was apparent in CRC samples (Fig. 1b; Zou el al. [35], t test p = 1.91E−5; fold change = −2.230; eight normal and nine CRC; Kaiser et al. [36], t test p = 1.67E−6; fold change = −2.617; 5 normal and 89 CRC). To further support these data, we investigated PDCD4 mRNA levels in a series of dysplastic/neoplastic polypoid lesions of the colon mucosa by qRT-PCR analysis. A significant PDCD4 downregulation was found in samples of LG-IEN (with a 0.7-fold decrease; t test p = 0.043), HG-IEN (with a 1.5-fold decrease; t test p = 0.001), and CRC (with a 2.1-fold decrease; t test p < 0.001; Fig. 1c).
Fig. 1

PDCD4 downregulation during colon carcinogenesis. a–bPDCD4 expression (mRNA) array results of four microarray data sets [3336] are considered; statistical significance is calculated using the Oncomine website ( a Box plots show differences in mRNA expression in normal colonic mucosa (gray) versus adenoma (white). Class analysis: Sabates-Bellver 2007 (t test p = 1.48E−16, fold change = −2.524, 32 normal and 25 adenomas) [33]; Gaspar 2008 (t test p = 2.49E−4, fold change = −1.709, 22 normal and 56 adenomas) [34]. b Box plots show differences in mRNA expression in normal colonic mucosa (gray) versus adenocarcinoma (white). Class analysis: Zou 2002 (t test p = 1.91E−5; fold change = −2.230; eight normal and nine CRC) [35]; Kaiser 2007 (t test p = 1.67E−6; fold change = −2.617; 5 normal and 89 CRC) [36]. cPDCD4 mRNA levels in the present series of polypoid lesions of the colon mucosa by qRT-PCR analysis. PDCD4 is significantly downregulated in IEN and CRC (filled circles represent individual samples; horizontal lines indicate the mean)

Nuclear PDCD4 expression is downregulated during colon carcinogenesis and in serrated adenomas

Normal colonic mucosa biopsy samples consistently featured a strong PDCD4 expression in the nucleus coexisting with a moderate–strong cytoplasmic PDCD4 expression (Fig. 2a–c). In normal and cancer tissue samples, non-cancerous and non-epithelial cells (fibroblasts, lymphocytes, smooth muscle cells, and endothelia) always featured PDCD4 nuclear expression (considered as a positive internal control; not considered for immunohistochemistry (IHC) scoring) [17, 19, 25, 26, 37].
Fig. 2

Representative PDCD4 immunostaining of non-cancerous colonic tissues and preneoplastic/neoplastic lesions of carcinogenic cascade. a Immunohistochemical scores of nuclear PDCD4 in the different histological lesions (white score 0, light gray score 1, dark gray score 2; black score 3). Nuclear PDCD4 staining was significantly lower in preneoplastic/neoplastic lesions (i.e., SA, LG-IEN, HG-IEN, and K) than in normal tissues or hyperplastic polyps (p < 0.001). b–c In native colonic mucosa, colocytes featured strong nuclear and moderate–strong cytoplasmic staining. d HP maintain strong nuclear and weak cytoplasmic immunostaining. Note the presence in SA (e), LG-IEN (f–g), and HG-IEN (h) samples of residual normal colocytes with PDCD4-positive nuclei. i–j Little or no cytoplasmic immunostaining was associated to HG-IEN. k CRC showing scattered PDCD4-positive nuclei (original magnifications, ×20 and ×40)

Hyperplastic polyps retained a PDCD4 expression that was strong in the nuclei and moderate in the cytoplasm (Fig. 2d). Nuclear PDCD4 loss was consistently observed in dysplastic (LG-IEN and HG-IEN) and neoplastic samples (Kruskal–Wallis, p < 0.001; Fig. 2). In particular, samples of LG-IEN showed moderate–strong cytoplasmic staining coexisting with a significant nuclear PDCD4 downregulation; HG-IEN and CRC biopsy samples showed little or no cytoplasmic protein expression (Fig. 2f–k). In SA, the dysplastic component of the polypoid lesions featured much the same PDCD4 expression as in LG-IEN (Fig. 2e).

Overall, PDCD4 nuclear downregulation (i.e., scores 0 and 1) consistently discriminates dysplastic/neoplastic (scores 0–1, n = 188; scores 2–3, n = 12) versus non-dysplastic/neoplastic (scores 0–1, n = 8; scores 2–3, n = 92) samples (sensibility = 94%; specificity = 92%).

miR-21 is overexpressed in preneoplastic and neoplastic polypoid lesions of the colon mucosa

The oncomiR miR-21 directly regulates PDCD4 by binding to its 3′ UTR target sequence with a 100% match sequence at nt228–249 [19]. We investigated miR-21 expression in dysplastic and neoplastic polypoid lesions of the colon mucosa by qRT-PCR analysis. A significant miR-21 upregulation was found in samples of LG-IEN (with a 1.5-fold increase; t test p = 0.007), HG-IEN (with a 2.2-fold increase; t test p = 0.001), and CRC (with a 4.8-fold increase; t test p < 0.001; Fig. 3a). To further confirm these data, we investigated miR-21 expression by ISH, which confirmed a consistently significant overexpression in IEN and CRC samples (Fig. 3b). miR-21 expression was identifiable as a granular brown cytoplasmic staining consistently expressed by cancerous epithelia.
Fig. 3

miR-21 is overexpressed in dysplastic and neoplastic polypoid lesions of the colonic mucosa. a Different miR-21 expressions between different biopsy samples by qRT-PCR analysis. miR-21 was significantly upregulated in IEN lesions and CRC. Each filled circle represents an individual sample; horizontal lines indicate the mean. b Representative examples of miR-21 expression in normal colonic epithelium and polypoid lesions by ISH analysis. The presence of miR-21 is revealed by a granular brown stain. (original magnifications, ×20 and ×40). c Inverse correlation between the expression of miR-21 and PDCD4 among the 40 biopsy samples analyzed by qRT-PCR. (symbols represent individual samples)

As expected, an inverse correlation (r = −0.3611) between the expression of miR-21 and PDCD4 was observed among the 40 biopsy samples analyzed by qRT-PCR (Fig. 3c).


PDCD4 has been characterized as a new tumor suppressor gene that is downregulated in several human malignancies, including CRC [18]. In their seminal study, Göke and colleagues reported loss of nuclear PDCD4 staining in six of seven CRC specimens [26]. Following this preliminary observation, Mudduluru and colleagues investigated the usefulness of PDCD4 as a diagnostic marker in the adenoma/carcinoma sequence [17]. They did not consider the grade of the dysplastic lesions examined, but they found a significant downregulation in PDCD4 expression in adenoma and CRC samples, and they suggested that PDCD4 has promise as a marker of progression in the process of colon mucosa transformation. In a series of 71 CRC patients, they also found loss of total or nuclear PDCD4, an independent predictor of disease-specific and overall survival [17]. Taken together, the available information points to a role for PDCD4 as a key colonic mucosa regulatory gene.

We explored the immunohistochemical expression of PDCD4 in a series of sporadic colonic polypoid lesions and CRC biopsy samples. As expected, we found that PDCD4 nuclear expression decreased significantly as the lesions became dedifferentiated. In fact, normal colonic mucosa and HP biopsy samples consistently featured a strong nuclear and moderate–strong cytoplasmic PDCD4 expression (Fig. 2a–d) whereas LG-IEN and HG-IEN lesions showed a weak–moderate cytoplasmic immunostaining. A similar nuclear-to-cytoplasmic shift was documented by Mudduluru and colleagues in the adenoma–carcinoma sequence [17]. We found this shift in the Barrett’s esophagus setting too, with a significant and consistent decrease in nuclear PDCD4 in both dysplastic and neoplastic samples [19]. The reasons behind this particular subcellular reallocation are only partially understood. In vitro studies have shown that, under normal growth conditions, PDCD4 is located mainly in the nucleus, while it moves to the cytoplasm on serum withdrawal [8]. The nuclear-to-cytoplasmic shift documented in preneoplastic/neoplastic lesions could thus theoretically result from epigenetic (and/or genetic) mutations.

Serrated adenomas show characteristic molecular changes not commonly seen in traditional colorectal adenomas, and they probably progress to colorectal cancer via a different pathway (the so-called serrated neoplasia pathway) [38], so we decided to study the “serrated” dysplastic component in terms of PDCD4 expression. The similar nuclear loss observed in SA and IEN lesions would suggest that PDCD4 is associated with a major tumor suppressor function that is soon lost in different pathways of carcinogenesis.

Colonoscopy has acquired a central role in CRC detection and prevention, thanks to its impact on the adenoma–carcinoma sequences (the “traditional” and the “serrated”). Observational data support a role for colonoscopy and polypectomy in reducing the incidence of CRC and the related mortality [7, 39]. Dysplasia (particularly high-grade dysplasia) is still the fundamental element in cancer prediction for most preneoplastic GI conditions, and it is an independent marker of cancer risk [40]. The histological evidence of dysplasia may be very subtle and difficult to diagnose, however, even in the presence of well-established diagnostic criteria, because of the intrinsic characteristics of biopsy samples (i.e., shrinkage, fragmentation, and “burning”). Pathologists must therefore strive to reduce the uncertainties in the diagnosis of dysplastic lesions and to improve the management of CRC-prone asymptomatic patients. From the practical standpoint, PDCD4 immunohistochemical analysis could be valuable in discriminating between dysplastic and non-dysplastic lesions, and between SA and HP. PDCD4 (and its nuclear staining) may therefore be associated with more traditional IHC markers in the routine histological assessment of colonic biopsy samples.

miR-21 has been described as being upregulated in CRC and an important regulator of PDCD4 expression [16]. Unlike most mRNAs, moreover, miRNAs are long-lived in vivo and very stable in vitro, a feature that might be crucial in a clinical setting and enables the analysis of formalin-fixed, paraffin-embedded samples. Confirming these data, we found miR-21 significantly upregulated in IEN and CRC biopsy samples, pointing to a potential diagnostic role for this miR in discriminating neoplastic from non-neoplastic lesions.

In conclusion, our findings further support the primary role of PDCD4 as a tumor suppressor gene of the colonic mucosa and consistently point to PDCD4 being a promising additional diagnostic tool for discriminating normal colonic tissues from benign adenomas and CRC in biopsy samples. Further studies on larger series should investigate the prognostic role of PDCD4 nuclear staining in stratifying dysplastic lesions into appropriate categories of CRC evolution.


This work has been partly supported by an AIRC Regional grant 2009 (assigned to M.R.). The authors are grateful to Vincenza Guzzardo, Mariangela Balistreri, Vanni Lazzarin, and Cristiano Lanza for their technical assistance.

Conflict of interest

The authors declare that they have no conflict of interest.

Copyright information

© Springer-Verlag 2011