1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. References
  7. Supporting Information

Insulin-like growth factor II mRNA-binding protein 3 (IMP3) is an RNA-binding protein expressed in embryonic tissues and multiple cancers. To investigate the role of IMP3 in hepatocellular carcinoma (HCC), its protein expression in the surgically resected unifocal tumors of 377 HCC patients (296 men and 81 women) with ages ranging from 7 to 88 years (mean, 55.49 years) was analyzed by immunohistochemistry. IMP3 was expressed in 255 (67.6%) of 377 resected unifocal primary HCCs. IMP3 protein was predominantly expressed in tumor border and invasive front, and it was more abundant in the satellite nodules and tumor thrombi than in the main tumors. The expression correlated with high α-fetoprotein (>200 ng/mL, P < 1 × 10−7), larger tumor size (>5 cm, P = 0.006), high tumor grade (P < 1 × 10−7), and high tumor stage with vascular invasion and various degrees of intrahepatic metastasis (P < 1 × 10−7). IMP3 expression predicted early tumor recurrence (P < 1 × 10−7) and was a strong indicator of poor prognosis (P < 0.0001). Depletion of IMP3 with RNA interference in HCC cell line HA22T caused a decrease in cell motility, invasion, and transendothelial migration. Microarray analysis revealed that IMP3 depletion was associated with downregulation of multiple genes involved in tumor invasion. Conclusion: Our results indicate that IMP3 plays an important role in tumor invasion and metastasis and is a strong prognostic factor for patients with HCC. (HEPATOLOGY 2008.)

Hepatocellular carcinoma (HCC) is one of the most common fatal malignancies worldwide, especially in Southeast Asia and Africa.1 Its incidence is increasing in Western countries, mainly because of the prevalence of chronic hepatitis C infection.2 Despite the recent advancement in early detection and patient management, the prognosis remains unsatisfactory, mainly because of the advanced tumor stage caused by intrahepatic tumor spread and frequent tumor recurrence.3 Hence, there is intense interest in the identification of reliable molecular markers to predict early tumor recurrence (ETR) and poor prognosis to benefit patients by allowing better management planning and identification of potential therapeutic targets.

VICKZ (Vg1 RBP/Vera, IMP-1,2,3, CRD-BP, KOC, ZBP-1; acronym suggested by Yaniv and Yisraeli4) proteins are a group of highly conserved RNA-binding proteins that contain two RNA recognition motifs at the N terminus and four heterogeneous nuclear ribonucleoparticles K homology domains at the C terminus.4, 5 The diverse nomenclature of these proteins reflects the fact that they were cloned from different species by a number of laboratories working on different problems. All members of the family share 70% or greater identity throughout their 550 to 590 amino acid proteins, with the highest level of conservation in the canonical RNA-binding motifs.4 The VICKZ proteins express abundantly in embryonic tissues, including the placenta, but they are virtually undetectable in mature tissues.6 The functions of VICKZ proteins are beginning to be deciphered. These proteins were found to be involved in RNA trafficking (Vg1 RBP, Vera),7, 8 RNA stabilization (CRD-BP,9 and translation repression (IMP-1-3).5

The three human homologs of VICKZ proteins were initially identified during a search for trans-acting factors associated with the leader 3 sequence of insulin-like growth factor-II messenger RNA (mRNA); hence they were designated as insulin-like growth factor II mRNA-binding proteins 1, 2, and 3 (IMP1-3).5 IMP3, also known as KOC (KH domain containing protein overexpressed in cancer), is encoded by a 4158-nucleotide RNA transcript, resulting in a protein of 580 amino acids. The gene is located at chromosome 7p11.5, a locus frequently amplified in multiple cancers.10–12 IMP3 is expressed in a wide range of mouse and human fetal tissues in a time-dependent and cell-dependent manner,6, 13 but it is undetectable in human adult tissues.6 Therefore, IMP3 appears to play an important role in the differentiation and maturation process of early embryogenesis. Increased IMP3 expression was reported in carcinomas of pancreas, kidney, uterus, and many other malignant tumors.6, 14, 15 Hence, it is considered an oncofetal protein. The function of IMP3 is beginning to be unraveled. IMP3 binds to insulin-like growth factor (IGF) II transcripts and represses its translation in the late development of mouse embryo.5 The Xenopus homolog of IMP3, Vg1 RBP, mediates the location of Vg1 RNA to the vegetal cortex of Xenopus oocyte.16 Recently, IMPs have been shown to promote cell adhesion and invadopodia formation, suggesting that IMP3 plays a role in cell migration and invasion during development and cancer progression.17

IMP3 has been reported to be expressed in HCC,18 but the clinicopathological significance is largely unknown, and the functional role of IMP3 in cancer remains to be proved. In this study, we demonstrate that IMP3 promotes cancer cell motility, invasion, and migration through the endothelial layer, and its expression is strongly associated with high tumor stage, and it predicts poor prognosis in patients with HCC.

Patients and Methods

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. References
  7. Supporting Information

Liver Samples.

From 1982 to 1998, 377 surgically resected, unifocal, primary HCCs, which received detailed pathological assessment and regular follow-up at the National Taiwan University Hospital, were selected for this study. The study was conducted according to the regulations of the Ethics Committee, and the specimens were anonymous and analyzed in a blinded manner. The patients included 296 men and 81 women with a mean age of 55.49 years (range, 7-88 years). Serum hepatitis B surface antigen was detected in 252 cases and anti-hepatitis C virus antibody in 112, including 24 positive for both. Liver cirrhosis was found in 121 cases (32.1%). Three hundred forty-eight were Child-Pugh class A, 25 class B, and 4 class C. None had received transhepatic arterial embolization or chemotherapy before surgery. After surgery, all patients received laboratory examinations such as serum α-fetoprotein at 1-month to 6- month intervals, and ultrasonography of liver at 3-month to 12-month intervals. Tumors were detected in 134 patients within 1 year, which is defined as early tumor recurrence. Two hundred sixty-nine patients died within 10 years. Follow-up periods for survivors ranged from 27 to 222 months (median, 89 months).

Histology Study and Tumor Staging.

Surgically resected specimens were formalin-fixed and paraffin-embedded. Histological sections cut at 5-μm thickness were stained with hematoxylin-eosin and reviewed by one of the authors (H.C.H.) to determine tumor grade and stage. The tumor grade was based on the criteria proposed by Edmonson and Steiner.19 The tumors were staged according to the American Joint Committee on Cancer system.20 One hundred sixty-four were stage I, 113 stage II, and 97 stage III. The margins of surgical specimens were inked and checked under microscope. Only completely resected specimens were included in this study.

Immunohistochemical Analysis.

Tissue sections (5 μm) were dewaxed and rehydrated. Antigen retrieval was done by incubating the slides in 0.01 M citric acid buffer at 100°C for 10 minutes. After blocking with 3% H2O2 and 5% fetal bovine serum, the slides were allowed to react with a monoclonal antibody against IMP3 (1:100, L523S; Dako Cytomation, Carpinteria, CA) or high mobility group AT-hook 2 (HMGA2) (1:50, Abcam, Cambridge, UK) at 4°C overnight. The slides were then incubated with polymer–horseradish peroxidase reagent (BioGenex, San Ramon, CA). The peroxidase activity was visualized with diaminobenzidine tetrahydroxychloride solution (BioGenex). The sections were counterstained with hematoxylin. Dark brown cytoplasmic staining was defined as positive, and no staining was defined as negative. For negative control, we replaced the primary antibody with 5% fetal bovine serum. The proportion of tumor cells positive for IMP3 varied considerably, ranging from diffuse positive (3+) (>50%), focal, or heterogeneous (2+) (10%-50%) to positive in a small amount of tumor cells (1+) (<10%).

Analysis of p53 and β-Catenin Mutations.

Mutations of the p53 tumor suppressor gene were analyzed in 219 tumors by direct sequencing spanning exon 2 to exon 11 as described previously.21 Mutations of the β-catenin gene were analyzed in 251 cases by direct sequencing of exon 3 as described previously.22

Cell Culture.

Liver cancer cell line HA22T, breast cancer cell lines MDA-MB-231, and 293GP2 cells were grown in Dulbecco's modified Eagle medium with 10% fetal bovine serum. Lung adenocarcinoma cell line CL1-0 was maintained in Roswell Park Memorial Institute 1640 medium with 10% fetal bovine serum. Human umbilical vein endothelial cells were obtained as described by Jaffe et al.23 and maintained in M199 culture medium supplemented with 20% fetal bovine serum, 75 mg/L endothelial cell growth factor, and 0.1 g/L heparin. All of the cells were grown at 37°C in a humidified atmosphere composed of 95% air and 5% CO2. The cells were passaged with trypsin/ethylenediaminetetra-acetic acid at 80% to 90% confluence.

Plasmid, Transfection, and Retroviral Infection.

The open reading frame of IMP3 was amplified from HeLa cells by reverse transcription polymerase chain reaction (RT-PCR) and cloned to the expression vector pCMV-Tag 2B (Stratagene, La Jolla, CA) to generate pCMV-IMP3 vectors for transfection using the Lipofectamine 2000 reagent (Invitrogen, Carlsbad, CA). FLAG-tagged IMP3 was cut from the pCMV-IMP3 construct and subcloned to the NotI-BamHI site of pQCXIP (Clontech, Mountain View, CA) to generate pQCXIP-IMP3. pQCXIP-IMP3 was cotransfected with pVSV-G (Clontech) into the retroviral package cell 293-GP2 to produce retrovirus. After incubating with medium containing retroviral particles for 2 days, the target cells were treated with puromycin (1 μg/ml) (Clontech) for 2 weeks to select the cells with stable integration of retroviral vectors.

RNA Interference.

The RNA interference (RNAi) oligos synthesized by Ambion (Austin, TX), corresponding to nts 1667-1686 (si-IMP3-1) and 495-514 (si-IMP3-2) of the human IMP3 mRNA transcript (GenBank Accession no. NM_006547) were used. The negative control RNAi was also purchased from the Ambion Company. We used Lipofectamine 2000 (Invitrogen) to transfect the RNA oligos into cultured cells, following the manufacturer's protocol.

Western Blot.

Protein samples (60 μg each) were separated in 10% sodium dodecyl sulfate polyacrylamide gel electrophoresis and then electrotransferred onto nitrocellulose membrane (Amersham, Buckinghamshire, UK). The membranes were allowed to react with primary and secondary antibodies at optimum dilution, and the immunoreactive signals were detected using an enhanced chemiluminescence kit (Amersham Pharmacia Biotech).

Cell Proliferation Assay.

To measure cell survival and proliferation, the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-2H-tetrazoliumbromide (MTT) assay was performed. Cells (2000 each) were seeded into 96-well plates and incubated at 37°C in a humidified atmosphere with 5% CO2. At the appropriate time, MTT (2 mg/mL in phosphate-buffered saline was added and incubated for 4 hours. The resulted color reaction product, MTT formazan, was extracted with dimethyl sulfoxide and the absorbance at 570 nm measured.

Boyden Chamber Assay.

For invasion assays, we used modified Boyden chambers with filter inserts (pore size, 8 μm) coated with Matrigel (40 μg; Collaborative Biomedical, Becton Dickinson Labware) in 24-well dishes (Nucleopore, Pleasanton, CA). Cells (2 × 104) in 100 μL OPTI-MEM (Invitrogen) were placed in the upper chamber, and 1 mL of Dulbecco's modified Eagle's medium was placed in the lower chamber. After 24 hours in culture, cells were fixed in methanol for 15 minutes and then stained with 0.05% crystal violet and 4′-6-diamidino-2-phenylindole in phosphate-buffered saline for 15 minutes. Cells on the upper side of the filters were removed with cotton-tipped swabs, and the filters were washed with phosphate-buffered saline. Cells on the underside of the filters were viewed and counted under a fluorescent microscope. Each group was plated in triplicate in each experiment, and each experiment was repeated at least three times. The cell motility assay was done in the same way as the invasion assay except for the absence of Matrigel coating and the incubation time was 20 hours.

For transendothelial cell migration assay, human umbilical vein endothelial cells were grown in Boyden chamber until confluence. Cells (2 × 104) in 100 μL OPTI-MEM (Invitrogen) were labeled with CytoTracker Orange (Invitrogen) and placed in the upper chamber, and 1 mL Dulbecco's modified Eagle's medium was placed in the lower chamber. After 24 hours in culture, cells on the upper side of the filters were removed with cotton-tipped swabs, and the migrated cells on the underside were viewed and counted under a fluorescent microscope.

RNA Isolation and Microarray Analysis.

After 72 hours, RNAi-transfected HA22T cells were harvested and total RNA was isolated using the TRIzol (Life Technologies, Invitrogen), then extracted using RNeasy mini kit (Qiagen, Valencia, CA), according to the instructions of the manufacturer. The complementary DNA derived from paired RNA samples from the siIMP3-1 and negative control transfected HA22T cells were labeled with Cy3-dCTP or Cy5-dCTP, and then applied to microarray chip. The microarray experiment and data analysis were done by the Welgene Biotech (Taipei, Taiwan) using the Agilent Oligo Chip (Human 1A Oligo Chip V2, which include 22,000 unique 60-mer oligos representing 18,716 unique human genes sourced from RefSeq and Incyte's Genomics Foundation Database). Microarray data scanned by laser scanner were measured to generate values and ratios of gene expression between the hybridized samples.

Real-Time PCR.

Real-time PCR was performed with the ABI PRISM 7000 Sequence Detection System (Applied Biosystems, Foster City, CA) using the SYBR Green method.

Statistical Analysis.

The data analyses were carried out using Epi Info (version 3.3.2; Centers for Disease Control and Prevention) and Statistical Analysis System (version 9.1; SAS Institute Inc., Cary, NC) software. Correlation between IMP3 expression and clinicopathologic parameters was evaluated by using chi-squared test and Fisher's exact test. Survival rates were calculated by using the Kaplan-Meier method, and difference in survival curves was analyzed by using the log-rank test. Multivariant analyses of stage were conducted by fitting multiple logistic regression models,24 and then times to death were analyzed by fitting multiple Cox's proportional hazards models.25 Basic model-fitting techniques for (1) variable selection, (2) goodness-of-fit assessment, and (3) regression diagnostics (including residual analysis, influence analysis, and check of multicollinearity) were used in our regression analysis to assure the quality of analysis result.24, 25 Two-tailed P < 0.05 was considered statistically significant.


  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. References
  7. Supporting Information

Expression of IMP3 in HCC.

IMP3 protein was detected in the tumor cell cytoplasm in 255 of 377 HCCs (67.6%) on immunohistochemical stain. The IMP3 protein expression in HCC varied considerably and showed a heterogeneous distribution ranging from diffuse positive (84 cases), heterogeneous to focal positive (72 cases), to positive in a small amount of tumor cells (99 cases) (Fig. 1). In those with heterogeneous to focal or trace IMP3 expression, IMP3 was predominantly expressed at the peripheries of tumor nests (Fig. 1C), invasive fronts (Fig. 1D), and satellite nodules (Fig. 1D). Notably, the intravascular tumor thrombi in large and small portal vein branches often exhibited more intense and diffuse immunoreactivity than the main tumors (Fig. 1E). In large necrotic areas of a huge tumor, viable tumor cells around necrotic areas also showed stronger IMP3 expression (Fig. 1F). No staining was seen in the nontumorous liver tissue adjacent to the tumors.

thumbnail image

Figure 1. Expression of IMP3 in HCC. (A) Diffuse expression of IMP3 in a high-grade HCC. (B) No immunostaining of IMP3 in an encapsulated HCC. (C) Expression of IMP3 was more prominent at the periphery of a tumor mass. (D) Expression of IMP3 in invasive front and a satellite nodule. (E) A portal vein tumor embolus with strong IMP3 expression. (F) Strong IMP3 expression in the cells adjacent to necrotic area. T, tumor; C, capsule; N, nontumorous liver tissue; TC, tumor center; IF, invasive front; S, satellite nodule; PVT, portal vein tumor emboli.

Download figure to PowerPoint

To verify the immunostaining is specific, we analyzed the mRNA levels of 40 nontumorous liver tissues and 114 HCCs by real-time PCR. The RNA levels of IMP3 in tissues were highly correlated with the result of immunohistochemistry (Supplementary Fig. 1), indicating the immunostaining was specific for IMP3 and the overexpression of IMP3 in HCCs was mainly attributable to activation of transcription.

Correlation of Clinicopathological Factors and IMP3 Expression in HCC.

To elucidate the role of IMP3 in HCC progression, we correlated IMP3 protein expression with a variety of clinicopathological features. As shown in Table 1, IMP3 expression occurred more frequently in tumors of younger patients (55 years; P = 0.017), and in hepatitis B virus–associated HCC (P = 0.0055). IMP3 expression also correlated with high serum alpha-fetoprotein level (>200 ng/mL; P < 1 × 10−7). Histopathologically, IMP3 expression was associated with large (>5 cm) and high-grade (grade 3-4) tumors, P = 0.0057 and P < 1 × 10−7, respectively. Expression of IMP3 was infrequent in tumors with complete encapsulation (OR = 0.2, P = 0.0000005). Importantly, high-stage (stage II-III) HCCs, which had vascular invasion and various degree of intrahepatic metastasis, had much more frequent IMP3 expression as compared with low-stage HCC (stage I) (odds ratio [OR] = 5.8 for stage II versus stage I, OR = 9.75 for stage III versus stage I, both P < 1 × 10−7). Moreover, patients with IMP3-positive HCCs were more likely to have ETR (OR = 4.93, P < 1 × 10−7) and a lower 10-year survival rate than HCC without IMP3 expression, P < 0.0001 (Fig. 2A). Further analysis revealed that IMP3 expression affected the survival rate of patients with large HCCs (>5 cm), but not small HCCs (Fig. 2B).

Table 1. Univariate Analysis of IMP3 Protein Expression and Clinicopathological Risk Factors in Patients with Unifocal HCC
  IMP3 Protein Expression
TotalYes, n (%)OR (95% CI)P
Age (years)>55216132 (61)1.0 
 ≤55161123 (76)2.06 (1.28-3.33)0.017
SexMale296193 (65)1.0 
 Female8162 (76)1.74 (0.96-3.20)0.053
HBsAg(−)12170 (58)1.0 
 (+)252182 (72)1.89 (1.17-3.06)0.0055
AFP (ng/mL)≤200207108 (52)1.0 
 >200160138 (86)5.75 (3.30-10.09)1 × 10−7
Tumor size (cm)≤5188114 (60)1.0 
 >5185137 (74)1.85 (1.17-2.95)0.0057
Tumor grade1-2254147 (57)1.0 
 3-4120106 (88)5.51 (2.89-10.66)<1 × 10−7
Encapsulation(−)333239 (72)1.0 
 (+)4214 (33)0.2 (0.09-0.41)0.0000005
Tumor stageI16473 (44)1.0 
 II11393 (82)5.80 (3.16-10.73)<1 × 10−7
 III9786 (87)9.75 (4.64-20.92)<1 × 10−7
Early tumor recurrence(−)17996 (53)1.0 
 (+)134114 (85)4.93 (2.73-8.97)<1 × 10−7
p53 mutation(−)12473 (59)1.0 
 (+)9573 (77)2.32 (1.23-4.40)0.0052
β-Catenin mutation(−)218156 (71)1.0 
 (+)3315 (45)0.32 (0.74-0.96)0.0017
thumbnail image

Figure 2. Kaplan-Meier analysis of overall survival in 377 patients with HCC. (A) Patients with IMP3-positive HCC had significantly lower survival rate than those with IMP3-negative HCC. (B) Patients with IMP3-positive large (>5 cm) HCC had a much worse prognosis than the other three groups.

Download figure to PowerPoint

P53 and β-catenin are the two most frequently mutated genes in HCC.22 Mutations of p53 are associated with tumor progression and poor prognosis, whereas a mutation of β-catenin is more frequently seen in low-stage tumors and is associated with a favorable prognosis.22 IMP3 expression is more frequently seen in tumors with p53 mutation (P = 0.0052), and in those without β-catenin mutation (P = 0.0017).

Multivariate Analysis of Factors Related to Survival Rate and Tumor Stage.

Our previous studies have identified several molecular markers for prediction of prognosis of patients with surgically resected HCCs.22, 26–31 To identify independent factors for predicting patient survival, we put clinicopathological factors and these molecular markers, including IMP3, in a multivariate analysis using Cox's proportional hazards models. The factors analyzed are listed in Table 2. Tumor stage and ETR were the most significant independent histopathological and clinical factors in predicting patient survival (Table 3). Of the molecular markers analyzed, only tumor-associated trypsin inhibitor expression had a marginal significance as a prognostic factor independent of tumor stage and ETR.

Table 2. Clinicopathologic and Molecular Factors Analyzed in Multivariate Analysis
Clinicopathologic FactorsMolecular Factors
AgeGene mutation
Sex p53
Hepatitis B infection β-Catenin
HCV infectionGene expression
Cirrhosis Aurora-A
Child-Pugh score CD24
α-Fetoprotein Glypican-3
Size IMP3
Grade KIAA0101
Capsular invasion L2DTL
Stage Osteopontin
Early tumor recurrence PAP
Table 3. Multivariate Analysis of Risk Factors Associated with Survival and Tumor Stage of the Patients with Unifocal HCC
 EstimateSEWaldPOR/Hazard Ratio
 Stage II-III0.54450.21316.52760.01061.724
 Early recurrence1.48340.211249.3234<0.00014.408
 TATI expression0.41280.21033.85440.04961.511
Stage II-III     
 Tumor size0.17010.06546.76370.00931.185
 IMP3 expression2.27790.574115.7420<0.00019.756
 KZAA0101 expression1.23620.56884.72420.02973.443
 β-Catenin mutation-2.03740.68288.90370.00280.130
 L2DTL expression1.61230.52709.36150.00225.014

To determine which factors contribute to high tumor stage, we used multiple logistic regression models to analyze the factors listed in Table 2 and found tumor size, absence of β-catenin expression, and expression of IMP3, KIAA0101, and L2DTL were independent factors for high tumor stage. Notably, IMP3 expression was the strongest factor associated with high tumor stage (OR = 9.756, P < 0.0001).

Effect of IMP3 Expression on Tumor Cell Motility, Invasion, and Transendothelial Migration.

As described, IMP3 protein expression was a strong factor in predicting high tumor stage, which had vascular invasion and various extents of intrahepatic metastasis. We speculated that IMP3 might play a role in tumor invasion. Because all the liver cancer cell lines available to us have abundant IMP3 expression (Supplementary Fig. 2), we used RNAi to deplete IMP3 in cultured cells. To avoid erroneous off-target effect, all experiments were performed with two different small interfering RNA (siRNA) oligos (siIMP3-1 and siIMP3-2). Treatment with both siRNAs for 72 hours reduced the IMP3 protein level to approximately 15% of the control cells (Fig. 3A). To rule out the possibility that the effect of IMP3 on invasiveness is attributable to enhanced cell proliferation, we measured the in vitro proliferation rate of IMP3-depleted HA22T cells by MTT assay and found that depletion of IMP3 did not interfere with cell proliferation. In fact, IMP3-depleted cells proliferated slightly faster than control cells (Fig. 3B). In transwell assays, IMP3 depletion significantly reduced cell motility, invasion, and migration through the endothelial cell layer (Fig. 3C-E). Overexpression of IMP3 in HA22T by retroviral transduction did not further enhance the invasiveness of this highly invasive HCC cell line (data not shown). Overexpression of IMP3 in the noninvasive pulmonary adenocarcinoma cell line CL1-0 significantly enhanced tumor cell invasion in a dose-dependent manner (Fig. 3F). Overexpression of IMP3 in breast cancer cell line MDA-MB-231 also showed a similar result (Supplement Fig. 3). IMP1 and IMP3 were reported to be involved in the formation of invadopodia.18 However, in IMP3-depleted HA22T cells, the percentage of cells with invadopodia did not decrease significantly (Supplement Fig. 4). Hence, the effect of IMP3 on tumor invasiveness could not be accounted for by its effect on invadopodia formation.

thumbnail image

Figure 3. (A) Depletion of IMP3 in HCC cell line HA22T by siRNA. (B) Depletion of IMP3 in HA22T did not affect the growth of cells, as measured by MTT assay. (C-E) Depletion of IMP3 inhibited the motility (C), invasion through Matrigel (D), and transendothelial migration (E). (F) Overexpression of IMP3 in CL1-0 cell line enhanced invasion through Matrigel.

Download figure to PowerPoint

Expression Profile of IMP Depleted Cells.

To elucidate the function of IMP3 and the potential mechanism of IMP3 siRNA-induced suppression of tumor cell invasion, we did a microarray assay to compare the expression profiles of siIMP3-1 and control siRNA-treated HA22T cells. Of the 80 most significantly down-regulated genes in IMP3 depleted HA22T cells, 47 were known genes (Table 4). Of them, 13 were surface receptors and 15 were secretory proteins. Many of them belonged to the same family, such as KLRC, CXCL, epidermal growth factor, semaphorin, and synaptotagmin families. Notably, 10 of the 47 genes are known to be involved in cell motility and tumor invasion.32–35 E-cadherin is the gene most significantly up-regulated in IMP3-depleted cells. To verify the microarray findings, the expression of the selected genes was analyzed by real-time RT-PCR, and the results were in agreement with the microarray data in all genes tested.

Table 4. Genes Most Down-Regulated in IMP3-Depleted HA22T Cells*
GeneFold ChangeP Value
  • *

    Genes without known function are not listed.

  • Genes known to be involved in tumor invasion and cell migration.

  • Genes confirmed by real-time RT-PCR.

Secreted protein  
 Chemokine (C-X-C motif)  
  Interleukin 8 (IL-8, CXCL8)5.96<0.01
  Synaptotagmin IV3.840.01
  Synaptotagmin I2.94<0.01
 Epidermal growth factor family  
  Epithelial mitogen homolog (EPGN)3.24<0.01
  Epidermal growth factor3.01<0.01
  Semaphorin 3A3.65<0.01
  Semaphorin 3E2.99<0.01
 Other secretory proteins  
  Colony stimulating factor 33.87<0.01
  Collagen, type XI, alpha 13.72<0.01
  BMP binding endothelial regulator3.32<0.01
 Killer cell lectin-like receptor subfamily C  
 Bradykinin receptor  
  Bradykinin receptor B13.82<0.01
  Bradykinin receptor B22.990.00
 Other receptors  
  Coagulation factor II receptor-like 25.19<0.01
  Secretin receptor3.440.03
 Interleukin 13 receptor, alpha 23.24<0.01
 Toll-like receptor 63.23<0.01
 Glutamate receptor, metabotropic 83.18<0.01
 Interleukin 1 receptor accessory protein3.11<0.01
 Neuropeptide FF receptor 22.82<0.01
 Other proteins  
  POU domain, class 2, associated factor 14.230.01
  Ubiquitin-conjugating enzyme E2E 13.68<0.01
  Ankyrin repeat domain 223.43<0.01
  Zinc finger CCCH-type containing 12A3.19<0.01
  Serine palmitoyltransferase, long chain base subunit 33.18<0.01
  High mobility group AT-hook 2 (HMGA2)3.16<0.01
  Spectrin, alpha, erythrocytic 13.08<0.01
  Nucleosomal binding protein 12.91<0.01
  Glycerol kinase2.910.04
  Glutamyl aminopeptidase (ENPEP)2.850.04
  Leucine-rich repeats and IQ motif containing 22.82<0.01
  Heterogeneous nuclear ribonucleoprotein A02.79<0.01
  Stearoyl-CoA desaturase2.78<0.01
  Ataxin 72.77<0.01

Correlation of IMP3 and HMGA2 Expression.

HMGA2 is an oncofetal protein involved in cell proliferation, neoplastic transformation, and tumor invasion.36 We first verified the effect of depletion of IMP3 by real-time PCR analysis of HMGA2 expression on siIMP3-1 and siIMP3-2–treated HA22T cells (Fig. 4A). Expression of HMGA2 was analyzed by immunohistochemistry in 58 tumors, and 14 tumors (24%) showed nuclear staining of HMGA2. All 14 tumors were IMP3 positive (P = 0.0002). Most cases have a scattered staining pattern (Fig. 4B-E). To be noteworthy, similar to IMP3, HMGA2 also tended to express at tumor periphery and invasive front. HMGA2 was not expressed in nontumorous liver parenchyma in all cases tested.

thumbnail image

Figure 4. Expression of HMGA2 and IMP3. (A) Depletion of IMP3 in HA22T by siIMP3-1 and siIMP3-2 down-regulated the expression of HMGA2 as determined by real-time RT-PCR. (B-E) Colocalization of HMGA2 (B,D) and IMP3 (C,E) in HCCs. In (D) and (E), both HMGA2 and IMP3 localized in tumor periphery. T, tumor; C, capsule; L, liver.

Download figure to PowerPoint


  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. References
  7. Supporting Information

HCC is one of the most common fatal malignancies worldwide. Tumor stage is the most important histological prognostic factor. The 5-year survival rate of patients with high stage (stage II-III) surgically resectable HCCs is approximately 20%, less than one third of patients with low-stage (stage I) HCCs (approximately 60%).29 Hence, it is important to identify the factors associated with high tumor stage and elucidate the mechanisms of tumor progression. In this study, we found that IMP3 protein was expressed in a high fraction of HCC. The IMP3 protein expression correlated with serum hepatitis B surface antigen positivity, high alpha-fetoprotein, high tumor grade, and high tumor stage and was a strong marker of poor prognosis. Furthermore, the expression of IMP3 was the most significant factor among 10 clinicopathological and 13 molecular factors to be associated with high tumor stage, with an OR of 9.756. IMP3 protein tended to express at tumor borders, satellite nodules, and portal vein tumor thrombi. These findings led us reasonably to believe that IMP3 expression is an important factor related to the enhanced metastatic potential of HCC and hence poor prognosis.

To verify its role in tumor invasion potential, we then depleted IMP3 in human HCC cell line HA22T and confirmed that loss of IMP3 inhibited tumor cell motility, invasion, and transendothelial migration in the Boyden Chamber assays. Overexpression of IMP3 in pulmonary adenocarcinoma cell line CL1-0 and breast cancer cell line MDA-MB-231 enhanced the invasion capacity of these cell lines, indicating the invasion-promoting effect of IMP3 is not cell type specific. Our observations suggest that IMP3 play a crucial role in conferring tumor cells' invasive and metastatic capacity.

Despite recent reports of IMP3 expression in multiple cancers,14–16, 37–39 the function of IMP3 remains largely unknown. The only known RNA target of IMP3 is IGF-II.5 However, we found that the IGF-II protein levels in the culture medium, as determined by enzyme-linked immunosorbent assay in IMP3 depleted or overexpressed cells, did not change significantly in the cell lines we used (data not shown). Besides, IGF-II is an autocrine growth factor overexpressed in multiple cancers,32 but IMP3 binds to IGF-II transcripts and represses its translation.5 These observations make IGF-II unlikely to be responsible for the phenotype of IMP3 expression in cancers. To identify other targets of IMP3, we examined the expression profile of IMP3 depleted cells and found many of the top transcripts down-regulated by IMP3 depletion were known to be involved in cell movement and tumor invasion.33–35 Of them, we were particularly interested in the oncofetal protein HMGA2. HMGA2 is a nonhistone chromatin binding protein. Its expression in carcinomas is highly correlated with tumor invasion.36 In the HCCs analyzed, all HMGA2-positive tumors were also IMP3 positive, indicating that IMP3 may be a crucial factor for regulating HMGA2 expression. Interestingly, HMGA2 is known to be regulated by tumor suppressor microRNA let-7.40 IMP3 may regulate the stability of mRNA transcripts of HMGA2 by binding to the mRNA and protecting it from microRNA-mediated RNA degradation.

Cancer invasion and metastasis is a multistep and complex process involving cooperation of many proteins. Although our in vitro function assay demonstrated that depletion of IMP3 inhibited cell migration and invasion, the strong correlation of IMP3 with invasion and high tumor stage cannot be completely explained by the function of IMP3 itself. Highly invasive cancer cells frequently coexpressed many proteins involved in cell proliferation and tumor invasion. The expression of IMP3, besides its own function, may be a surrogate marker for the activation of cell proliferation and migration programs, which induced the highly malignant phenotype seen in clinical samples.

The restricted expression of IMP3 in a wide spectrum of cancers and the lack of expression in normal mature tissues make it a potential target of cancer therapy. A phase I clinical trial of immunotherapy targeting IMP3 in non–small cell lung cancer has been done.41 Our study also suggests that IMP3 may be a potential therapeutic target for HCC.


  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. References
  7. Supporting Information
  • 1
    El-Serag HB, Rudolph KL. Hepatocellular carcinoma: epidemiology and molecular carcinogenesis. Gastroenterology 2007; 132: 25572576.
  • 2
    Heathcote EJ. Prevention of hepatitis C virus-related hepatocellular carcinoma. Gastroenterology 2004; 127: S294S302.
  • 3
    Zhou XD. Recurrence and metastasis of hepatocellular carcinoma: progress and prospects. Hepatobiliary Pancreat Dis Int 2002; 1: 3541.
  • 4
    Yaniv K, Yisraeli JK. The involvement of a conserved family of RNA binding proteins in embryonic development and carcinogenesis. Gene 2002; 287: 4954.
  • 5
    Nielsen J, Christiansen J, Lykke-Andersen J, Johnsen AH, Wewer UM, Nielsen FC, et al. A family of insulin-like growth factor II mRNA-binding proteins represses translation in late development. Mol Cell Biol 1999; 19: 12621270.
  • 6
    Mueller-Pillasch F, Lacher U, Wallrapp C, Micha A, Zimmerhackl F, Hameister H, et al. Cloning of a gene highly overexpressed in cancer coding for a novel KH-domain containing protein. Oncogene 1997; 14: 27292733.
  • 7
    Havin L, Git A, Elisha Z, Oberman F, Yaniv K, Schwartz SP, et al. RNA-binding protein conserved in both microtubule- and microfilament-based RNA localization. Genes Dev 1998; 12: 15931598.
  • 8
    Deshler JO, Highett MI, Abramson T, Schnapp BJ. A highly conserved RNA-binding protein for cytoplasmic mRNA localization in vertebrates. Curr Biol 1998; 8: 489496.
  • 9
    Prokipcak RD, Herrick DJ, Ross J. Purification and properties of a protein that binds to the C-terminal coding region of human c-myc mRNA. J Biol Chem 1994; 269: 92619269.
  • 10
    Takada H, Imoto I, Tsuda H, Sonoda I, Ichikura T, Mochizuki H, et al. Screening of DNA copy-number aberrations in gastric cancer cell lines by array-based comparative genomic hybridization. Cancer Sci 2005; 96: 100110.
  • 11
    Suzuki T, Maruno M, Wada K, Kagawa N, Fujimoto Y, Hashimoto N, et al. Genetic analysis of human glioblastomas using a genomic microarray system. Brain Tumor Pathol 2004; 21: 2734.
  • 12
    Nessling M, Richter K, Schwaenen C, Roerig P, Wrobel G, Wessendorf S, et al. Candidate genes in breast cancer revealed by microarray-based comparative genomic hybridization of archived tissue. Cancer Res 2005; 65: 439447.
  • 13
    Mueller-Pillasch F, Pohl B, Wilda M, Lacher U, Beil M, Wallrapp C, et al. Expression of the highly conserved RNA binding protein KOC in embryogenesis. Mech Dev 1999; 88: 9599.
  • 14
    Yantiss RK, Woda BA, Fanger GR, Kalos M, Whalen GF, Tada H, et al. KOC (K homology domain containing protein overexpressed in cancer): a novel molecular marker that distinguishes between benign and malignant lesions of the pancreas. Am J Surg Pathol 2005; 29: 188195.
  • 15
    Jiang Z, Chu PG, Woda BA, Rock KL, Liu Q, Hsieh CC, et al. Analysis of RNA-binding protein IMP3 to predict metastasis and prognosis of renal-cell carcinoma: a retrospective study. Lancet Oncol 2006; 7: 556564.
  • 16
    Schwartz SP, Aisenthal L, Elisha Z, Oberman F, Yisraeli JK. A 69-kDa RNA-binding protein from Xenopus oocytes recognizes a common motif in two vegetally localized maternal mRNAs. Proc Natl Acad Sci U S A 1992; 89: 1189511899.
  • 17
    Vikesaa J, Hansen TV, Jonson L, Borup R, Wewer UM, Christiansen J, et al. RNA-binding IMPs promote cell adhesion and invadopodia formation. EMBO J 2006; 25: 14561468.
  • 18
    Zhang J, Chan EK. Autoantibodies to IGF-II mRNA binding protein p62 and overexpression of p62 in human hepatocellular carcinoma. Autoimmun Rev 2002; 1: 146153.
  • 19
    Edmonson HA, Steiner PE. Primary carcinoma of the liver: a study of 100 among 489,000 necropsies. Cancer (Phila) 1954; 7: 462503.
  • 20
    Fleming ID, Cooper JS, Henson DE, Hutter RVP, Kennedy BJ, Murphy GP, et al. AJCC Cancer Staging Manual. 5th ed. Philadelphia: Lippincott-Raven; 1997: 98126.
  • 21
    Hsu HC, Tseng HJ, Lai PL, Lee PH, Peng SY. Expression of p53 gene in 184 unifocal hepatocellular carcinomas: association with tumor growth and invasiveness. Cancer Res 1993; 53: 46914694.
  • 22
    Hsu HC, Jeng YM, Mao TL, Chu JS, Lai PL, Peng SY. β-Catenin mutations are associated with a subset of low-stage hepatocellular carcinoma negative for hepatitis B virus and with favorable prognosis. Am J Pathol 2000; 157: 763770.
  • 23
    Jaffe EA, Hoyer LW, Nachman RL. Synthesis of anti-bemophilic factor antigen by cultured human endothelial cells. J Clin Invest 1973; 52: 27572764.
  • 24
    Hosmer DW, Lemeshow S. Applied logistic regression. 2nd ed. New York: John Wiley & Sons, 2000.
  • 25
    Hosmer DW, Lemeshow S. Applied survival analysis: regression modeling of time to event data. New York: John Wiley & Sons; 1999.
  • 26
    Yuan RH, Jeng YM, Chen HL, Lai PL, Pan HW, Hsieh FJ, et al. Stathmin overexpression cooperates with p53 mutation and osteopontin overexpression, and is associated with tumour progression, early recurrence, and poor prognosis in hepatocellular carcinoma. J Pathol 2006; 209: 549558.
  • 27
    Yuan RH, Jeng YM, Chen HL, Hsieh FJ, Yang CY, Lee PH, et al. Opposite roles of human pancreatitis-associated protein and REG1A expression in hepatocellular carcinoma: association of pancreatitis-associated protein expression with low-stage hepatocellular carcinoma, β-catenin mutation, and favorable prognosis. Clin Cancer Res 2005; 11: 25682575.
  • 28
    Jeng YM, Peng SY, Lin CY, Hsu HC. Overexpression and amplification of Aurora-A in hepatocellular carcinoma. Clin Cancer Res 2004; 10: 20652071.
  • 29
    Peng SY, Ou YH, Chen WJ, Li HY, Liu SH, Pan HW, et al. Aberrant expressions of annexin A10 short isoform, osteopontin and α-fetoprotein at chromosome 4q cooperatively contribute to progression and poor prognosis of hepatocellular carcinoma. Int J Oncol 2005; 26: 10531061.
  • 30
    Lee YC, Pan HW, Peng SY, Lai PL, Kuo WS, Ou YH, et al. Overexpression of tumour-associated trypsin inhibitor (TATI) enhances tumour growth and is associated with portal vein invasion, early recurrence and a stage-independent prognostic factor of hepatocellular carcinoma. Eur J Cancer 2007; 43: 736744.
  • 31
    Yuan RH, Jeng YM, Pan HW, Hu FC, Lai PL, Lee PH, et al. Overexpression of KIAA0101 predicts high stage, early tumor recurrence, and poor prognosis of hepatocellular carcinoma. Clin Cancer Res 2007; 13: 53685376.
  • 32
    LeRoith D, Roberts CT Jr. The insulin-like growth factor system and cancer. Cancer Lett 2003; 195: 127137.
  • 33
    Miyazaki H, Patel V, Wang H, Edmunds RK, Gutkind JS, Yeudall WA. Down-regulation of CXCL5 inhibits squamous carcinogenesis. Cancer Res 2006; 66: 42794284.
  • 34
    Nutt CL, Matthews RT, Hockfield S. Glial tumor invasion: a role for the upregulation and cleavage of BEHAB/brevican. Neuroscientist 2001; 7: 113122.
  • 35
    Wells A, Kassis J, Solava J, Turner T, Lauffenburger DA. Growth factor-induced cell motility in tumor invasion. Acta Oncol 2002; 41: 124130.
  • 36
    Motoyama K, Inoue H, Nakamura Y, Uetake H, Sugihara K, Mori M. Clinical significance of high mobility group A2 in human gastric cancer and its relationship to let-7 microRNA family. Clin Cancer Res 2008; 14: 23342340.
  • 37
    Zheng W, Yi X, Fadare O, Liang SX, Martel M, Schwartz PE, et al. The oncofetal protein IMP3: A novel biomarker for endometrial serous carcinoma. Am J Surg Pathol. 2008; 32: 304315.
  • 38
    Li C, Rock KL, Woda BA, Jiang Z, Fraire AE, Dresser K. IMP3 is a novel biomarker for adenocarcinoma in situ of the uterine cervix: an immunohistochemical study in comparison with p16(INK4a) expression. Mod Pathol 2007; 20: 242247.
  • 39
    Simon R, Bourne PA, Yang Q, Spaulding BO, di Sant'Agnese PA, Wang HL, et al. Extrapulmonary small cell carcinomas express K homology domain containing protein overexpressed in cancer, but carcinoid tumors do not. Hum Pathol 2007; 38: 11781183.
  • 40
    Lee YS, Dutta A. The tumor suppressor microRNA let-7 represses the HMGA2 oncogene. Genes Dev 200; 21: 10251030.
  • 41
    Nemunaitis J, Meyers T, Senzer N, Cunningham C, West H, Vallieres E, et al. Phase I Trial of sequential administration of recombinant DNA and adenovirus expressing L523S protein in early stage non-small-cell lung cancer. Mol Ther 2006; 13: 11851191.

Supporting Information

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. References
  7. Supporting Information

Additional Supporting Information may be found in the online version of this article.

hep22459-SupplementaryFigure1.tif881KSupplementary Figure 1.Correlation of mRNA expression and immunohistochemical staining of IMP3 in 40 nontumorous liver tissues and 114 HCCs. Real-time RT-PCR was done to determine the amount of IMP3 in tissue samples. The RNA amount was highly correlated with the extent of immunohistochemical staining in a dose-dependent manner. All P value of the differences between different group were <0.05. L: nontumorous liver parenchyma, 0+: tumors without IMP3 expression, 1+, 2+, 3+: tumors with 1+, 2+, 3+ IMP3 expression by immunohistochemistry.
hep22459-SupplementaryFigure2.tif1619KSupplementary Figure 2.Expression of IMP3 in liver cancer cell lines.
hep22459-SupplementaryFigure3.tif11343KSupplementary Figure 3.Effect of IMP3 expression on breast cancer cell line MDA-MB-231. pQCXIP-IMP3 and control vector were transduced into MDA-MB-231 cells by retrovirus. Compared with those with empty vector (A), MDA-MB-231 cells with IMP3 overexpression invaded through Matrigel more efficiently (B). (C) Statistical analysis of three wells for each group in one experiment.
hep22459-SupplementaryFigure4.tif9165KSupplementary Figure 4.Invadopodia formation in IMP3 depleted HA22T cells. After treated with siRNA for 3 days, the cells were plated on slides coated with fibronectin. Invadopodia was shown by colocalization of phosphotyrosine (green) and actin (red). (A) Control cells. (B) IMP3 depleted cells. 200 cells in each slide were counted to determine the percentage of cells with invadopodia. As shown in (C), the percentage of cells with invadopodia was similar in IMP3-depleted cells and control cells.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.