CRD-BP/IMP1 expression characterizes cord blood CD34+ stem cells and affects c-myc and IGF-II expression in MCF-7 cancer cells.

The coding region determinant-binding protein/insulin-like growth factor II mRNA-binding protein (CRD-BP/IMP1) is an RNA-binding protein specifically recognizing c-myc, leader 3' IGF-II and tau mRNAs, and the H19 RNA. CRD-BP/IMP1 is predominantly expressed in embryonal tissues but is de novo activated and/or overexpressed in various human neoplasias. To address the question of whether CRD-BP/IMP1 expression characterizes certain cell types displaying distinct proliferation and/or differentiation properties (i.e. stem cells), we isolated cell subpopulations from human bone marrow, mobilized peripheral blood, and cord blood, all sources known to contain stem cells, and monitored for its expression. CRD-BP/IMP1 was detected only in cord blood-derived CD34(+) stem cells and not in any other cell type of either adult or cord blood origin. Adult BM CD34(+) cells cultured in the presence of 5'-azacytidine expressed de novo CRD-BP/IMP1, suggesting that epigenetic modifications may be responsible for its silencing in adult non-expressing cells. Furthermore, by applying the short interfering RNA methodology in MCF-7 cells, we observed, subsequent to knocking down CRD-BP/IMP1, decreased c-myc expression, increased IGF-II mRNA levels, and reduced cell proliferation rates. These data 1) suggest a normal role for CRD-BP/IMP1 in pluripotent stem cells with high renewal capacity, like the CB CD34(+) cells, 2) indicate that altered methylation may directly or indirectly affect its expression in adult cells, 3) imply that its de novo activation in cancer cells may affect the expression of c-Myc and insulin-like growth factor II, and 4) indicate that the inhibition of CRD-BP/IMP1 expression might affect cancer cell proliferation.

The coding region determinant-binding protein/insulin-like growth factor II mRNA-binding protein (CRD-BP/IMP1) 1 recognizes at least four RNAs. CRD-BP/IMP1 binds specifically to 1) one of the two independently cis-acting, c-myc mRNA instability elements (1), 2) the 5Ј-untranslated region of the leader 3Ј IGF-II mRNA, which represents the major embryonic form of this message (2), 3) the H19 RNA, a gene product exhibiting an oncofetal pattern of expression and whose role remains enigmatic (3), and 4) the neuron specific tau mRNA (4), which encodes for a microtubule-associated protein localized primarily in the cell body and axon of developing neurons.
Normally, CRD-BP/IMP1 is mostly expressed in embryonal tissues. Analysis of total RNA from mouse embryos indicated a peak of CRD-BP/IMP1 expression at embryonic day 12.5 followed by a decline toward birth and its disappearance in neonatal mice shortly after birth (3). In normal adult tissues, most studies failed to detect CRD-BP/IMP1 mRNA even with the highly sensitive technique of RT-PCR (3,(5)(6)(7)(8). However, exceptions have been reported; CRD-BP/IMP1 expression has been detected by immunohistochemistry in the spermatogenic cells in human testis (5), whereas by RT-PCR a very faint CRD-BP/IMP1-specific band has been detected in material derived from adult mouse intestines (9). The expression levels detected in adult mouse intestines are substantially lower than those observed in the corresponding embryonal tissue, indicating perhaps its selective expression in certain rare intestinal cells in adulthood. The de novo expression or overexpression of CRD-BP/IMP1 has been reported in a variety of human neoplasias and in some of these tumor types characterizes the vast majority of the specimens studied (7, 8, 10 -12). Furthermore, autoantibodies against CRD-BP/IMP1 have been recorded in the sera from 6 different cancer type patients, with frequencies ranging from 7.1 to 16.5% (13). Two recent animal studies clearly demonstrated a crucial role for CRD-BP/IMP1 in both development and cancer onset. CRD-BP/IMP1-deficient mice exhibit dwarfism, impaired gut development, and increased perinatal mortality (9), whereas transgenic mice exhibiting targeted expression of CRD-BP/IMP1 in adult mammary tissue develop tumors (14).
CRD-BP/IMP1 has been attributed various roles in determining the post-transcriptional fate of its RNA targets. CRD-BP/IMP1 acts as a nucleocytoplasmic shuttling protein (5) exhibiting a distinct pattern of localization in the cytoplasm (15). The protein is distributed along with microtubules and in motile cells transports toward the leading edge. Its nuclear export and cytoplasmic movement depend on RNA binding, implying * This work was supported by a grant from the Regional Operational Program Attika 20 (MIS code 59605GR (to M. P.) The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
§ Contributed equally to this work. ¶ To whom correspondence should be addressed: Cancer Immunology Immunotherapy Center, Saint Savas Hospital, 171 Alexandras Ave., 115 22 Athens, Greece. Tel.: 30-10-6409459; Fax: 30-10-6409516; E-mail: perez@ciic.gr. 1 The abbreviations used are: CRD-BP/IMP1, c-myc coding region determinant binding protein/insulin-like growth factor II mRNA-binding protein; siRNA, short interfering RNA; CB, cord blood; BM, bone marrow; MPB, mobilized peripheral blood;5azaC, 5Ј-azacytidine; RT, reverse transcription; IGF, insulin-like growth factor; Ct, cycle threshold. that CRD-BP/IMP1 recognizes its targets into the nucleus and thereby defines their cytoplasmic fate. It has been postulated that CRD-BP/IMP1 may play a role in polarizing genetic information via cytoplasmic RNA localization, a critical mechanism especially in developmental systems for the generation of subcellular asymmetries in protein abundance. The co-association to the tau mRNA of CRD-BP/IMP1 with HuD (4), a protein involved in trafficking of this mRNA from the cell body to the distal parts of axon in developing neurons (16), supports a role in mRNA localization for CRD-BP/IMP1. The original studies implicated CRD-BP/IMP1 in the control of the translation of IGF-II leader 3Ј mRNA (2) and of the stability of c-myc mRNA (1,17). However, these latter functions may be secondary to the former, as the mRNA localization process may implicate its protection from decay as well as its translational repression while the message is transported along the cytoskeleton to the site where it will be translated. The two animal studies, previously mentioned, have not shed light but have rather perplexed the issue of the CRD-BP/IMP1 mode of action. Although the overexpression of CRD-BP/IMP in the NIH-3T3 cell line suppresses leader 3Ј IGF-II mRNA translation (2) in the null mice, not only leader 3Ј but leader 4Ј IGF-II mRNA as well are redistributed from polysomes to RNPs, indicating their reduced translational efficiency (9). On the other hand, in mice expressing de novo CRD-BP/IMP1 in mammary tissue, a 100-fold increase in the levels of IGF-II mRNA has been recorded, albeit without any increase in the IGF-II intracellular protein levels (14). Regarding the role of CRD-BP/IMP1 upon c-myc mRNA fate, the available data from a cell-free mRNA decay system indicate that CRD-BP/IMP1 binding protects c-myc mRNA from endonucleolysis, leading to a 4 -8-fold increase in its half-life and steady state levels (1,17). Nevertheless, no alterations of c-myc mRNA levels have been detected in the early stages of mammary transformation in transgenic mice with targeted CRD-BP/IMP1 expression in the tissues of CRD-BP/ IMP1 null mice (9,14). The above indicates that the action of CRD-BP/IMP1 upon its RNA targets is more complex than initially thought and furthermore implys that in transformed adult cells CRD-BP/IMP1 may exert "illegitimate" roles due to its expression and/or overexpression in an inappropriate cellular context. Thus, the identification of the normal cell types that express CRD-BP/IMP1 could contribute toward the elucidation of its role in normal cellular processes as well as in tumorigenesis. In this study we show that among normal bone marrow (BM), mobilized peripheral blood (MPB), and cord blood (CB) cell subpopulations this gene is expressed exclusively within the CB CD34 ϩ stem cell compartment and not in their adult counterparts. Moreover, we show that CRD-BP/ IMP1 knock-down in the MCF-7 cancer cell line is followed by c-myc mRNA and protein down-regulation and by IGF-II mRNA up-regulation, providing evidence that CRD-BP/IMP1 may affect the expression of these two genes whose abnormal expression is associated with tumorigenesis. Finally, we observed a reduction in cell proliferation rates after CRD-BP/ IMP1 knock-down. Although modest and reversible due to the transient action of the exogenously added siRNAs, this reduction implied that the sustained inhibition of CRD-BP/IMP1 expression may affect cancer cell physiology, and if so, CRD-BP/IMP1 may provide a putative therapeutic intervention target.

CB, BM, and MPB Samples and Mononuclear Cell Purification and
Phenotyping-Human CB (n ϭ 29) samples from umbilical veins of normal preterm or full term (32-40 weeks) infants, granulocyte colonystimulating factor-MPB (n ϭ 10), and bone marrow aspirate samples (BM, n ϭ 5) were obtained after informed consent and processed within 4 h of collection. Mononuclear cells were isolated by Ficoll-Hypaque centrifugation using standard procedures. The remaining red blood cells in CB samples were eliminated by negative selection with antiglycophorin A microbeads (Miltenyi Biotec, Germany) according to the manufacturer's instructions. CD34 ϩ cells were isolated with a CD34 selection kit (Miltenyi Biotec). The CD34 Ϫ fraction was then serially positively selected for CD14 ϩ , CD56 ϩ , CD19 ϩ , and CD3 ϩ using microbeads conjugated with the respective monoclonal antibodies (all purchased by Miltenyi Biotec). For the estimation of cell purity, isolated populations were washed twice with ice-cold phosphate-buffered saline, 1% bovine serum albumin followed by incubation with saturating concentrations of the appropriate monoclonal antibodies (CD3-FITC clone UCHT1, Pharmingen; CD14-PE clone RM052, Immunotech; CD19-PE clone HIB19, BD Biosciences; CD56-PC5 clone N901/NKH1, Immunotech; CD34-PE clone 8G12, BD Biosciences) for 15 min at room temperature. Thereafter, cells were washed twice in ice-cold phosphatebuffered saline, 1% bovine serum albumin and fixed with 1% paraformaldehyde in phosphate-buffered saline. Samples were analyzed using FACSCalibur (BD Biosciences) and CellQuest analysis software. Isolated blood subpopulations had purity higher than 95%. The negative for all the above marker fractions (Lin Ϫ CD34 Ϫ ) was also obtained and tested in parallel with the other samples.
Ex Vivo Culture of BM CD34 ϩ Cells and 5Ј-Azacytidine (5azaC) Treatment-96-Well plates were seeded with 1-2ϫ10 5 cells/well BMderived CD34 ϩ cells in StemSpan medium (StemCell Technologies) with or without 50 ng/ml stem cell factor, 20 ng/ml FLT-3 ligand, 25 ng/ml thrombopoietin, 10 ng/ml interleukin-6 in the presence or not of 10 Ϫ6 M demethylating agent 5azaC (Sigma-Aldrich) and cultured for 15 days. Every 4 days half of the medium was discarded and replenished by fresh medium containing (or not) freshly added cytokines and 5azaC. At the end of the culture period separate cell samples were analyzed by fluorescence-activated cell sorter for the expression of CD34 and by RT-PCR for CRD-BP/IMP1 expression.
Cell Culture and siRNA Transfection-The human breast cancer cell line MCF-7 (ATCC, Manassas, VA) was cultured in minimum essential medium-␣ medium (Invitrogen) supplemented with 10% fetal bovine serum (Invitrogen), 2 mM L-glutamine (Invitrogen), 100 units/ml recombinant human insulin (Lilly), and 50 g/ml gentamicin at 37°C in a CO 2 incubator. Twenty-one nucleotide double-stranded RNAs directed against the CRD-BP/IMP1 mRNA were chemically synthesized (Xeragon, Inc.). The sense and antisense sequences were: siCRD-BP-1, r(CCUGGCUGCUGUAGGUCUU)dTT and r(AAGACCUACAGCAGC-CAGG)dTT; siCRD-BP-2, r(GCUGGAGACCCACAUAGGU)dTT and r(ACGUAUGUGGGUCUCCAGC)dTT. Control non-silencing, fluorescein label siRNA was purchased from Xeragon, Inc. (catalog number 80-1132). c-myc siRNA and negative control siRNAs (non-labeled) were from Ambion, Inc. (catolog number 4604). All siRNAs were diluted in 100 mM potassium acetate, 30 mM HEPES-KOH, pH 7.4, 2 mM magnesium acetate to a final concentration of 20 M and stored in aliquots at Ϫ70°C. MCF-7 cells were regularly passaged to maintain exponential growth. The day before transfection ϳ4.5 ϫ 10 5 cells were plated per well in 6-well plates. Transient transfection of siRNAs was carried out using the siPORT lipid transfection agent (Ambion, Inc.) following the manufacturer's instructions. The control non-silencing fluorescein-labeled siRNA was used to optimize siRNA and siPORT lipid concentrations. Cellular siRNA uptake was estimated by fluorescence-activated cell sorter analysis 24 h after transfection. After optimization, each experiment was carried out using a final volume of 5 l of siPORT lipid and a final concentration of 100 nM siRNA. Cells from duplicate wells in each experiment were pooled 48 h after transfection, and total RNA was isolated as described below. For the analysis of protein expression levels cells from three wells were pooled 48 h post-transfection, and total protein was isolated as described below. Each experiment was performed three times.
RNA Extraction and Reverse Transcription-Polymerase Chain Reaction-Total RNA was extracted from isolated cells using the NucleoSpin RNA II kit (Machrey-Nagel) according to the manufacturer's protocol. First strand cDNA synthesis was performed using ϳ1 g of total RNA, oligo(dT) primers, and the SuperScript III RNase H(-) reverse transcriptase (Invitrogen) according to preamplification system protocol in a total volume of 20 l. Two l of this cDNA material was used for PCR amplification with Taq Platinum (Invitrogen) in a nested PCR reaction of a total volume of 50 l. In the first round of amplification, the following oligos were used: forward 1, 5Ј-CAT GCC CTG AAG GTC TCC TA-3Ј, and reverse 1, 5Ј-AAG GTC TTG CAA CGA GGA GA-3Ј. Two l of this PCR product was used as a template in a second round of amplification using an internal pair of primers with the following sequences: forward 2, 5Ј-GAG CAG ATA GCA CAG GGA CC-3Ј and reverse 2, 5Ј-ATC TTA CAA GCG GAG GAG CA-3Ј. The cycling condi-tions for both rounds were 33 cycles of denaturation in 94°C for 1 min, annealing in 61°C for 1 min, and elongation in 72°C for 1 min 20 s.
To monitor the integrity of the mRNA of each sample, we performed PCR using 2 l of the same cDNA material and the following primers specific for B2-microglobulin: forward, 5Ј-GTC TGG GTT TCA TCC ATC CG-3Ј, and reverse, 5Ј-TCA TCC AAT CCA AAT GCG GC-3Ј. Cycling conditions were 35 rounds of 94°C for 1 min (denaturing), 60°C for 1 min (annealing), and 72°C for 2 min (extension). All the amplified products were subjected to1.2% agarose gel electrophoresis containing ethidium bromide dye solution (Invitrogen) and visualized by UV light.
Relative Quantitation of CRD-BP/IMP1, c-myc, and IGFII mRNA Levels-To evaluate the relative levels of CRD-BP/IMP1, c-myc, and IGF-II mRNAs we performed quantitative real time PCR analysis on a Rotor-Gene (CORBET) thermal cycler.
For CRD-BP/IMP1 and ␤2-microglobulin, we used LUX custom-synthesized primers labeled with a single fluorophore close to the 3Ј end in a hairpin structure (Invitrogen). The primers for CRD-BP/IMP1 were the following: CRDFLUX, 5Ј-CAT CGG ACG AGC ACT GGG CGA G-3Ј, and CRDR. 5Ј-TTT GGG CAC CGA ATG TTC AA-3Ј. For the housekeeping B2-microglobulin gene we used the fluorophore-labeled human ␤2M-certified LUX primer set (Invitrogen).
The reactions were performed in a total volume of 25 l using the Platinum qPCR SuperMix UDG (Invitrogen) according to the manufacturer's recommendations for Rotor-Gene and 2 l of the cDNA material prepared as described above. The cycling protocol consisted of 50°C for 2 min and 95°C for 2 min (uracil-DNA glycosylase reaction) and 50 cycles of denaturation at 95°C for 15 s, annealing at 60°C for a 30-s extension at 72°C for 30 s and plate read. To confirm amplification specificity we performed a melting curve analysis at the end of each cycling.
For IGF-II we used the following specific oligonucleotides: forward, 5Ј-ctg gag acg tac tgt gct acc ccc-3Ј, and reverse, 5Ј-gtg tca tat tgg aag aac ttg ccc-3Ј, and Platinum SYBR Green qPCR SuperMix UDG (Invitrogen). The reaction was set up according to the manufacturer's instructions using 2 l of the same cDNA material in a total volume of 25 l. The amplification protocol consisted of 50°C for 2 min and 95°C for 2 min (uracil-DNA glycosylase reaction) and 50 cycles of denaturation at 95°C for 15 s, annealing at 65°C for a 30-s extension at 72°C for 30 s, heat at 83°C for 2 s, and plate read.
For c-myc we used the Real Time PCR for Human c-Myc Gene Expression Primer and Probe Set (Maxim Biotech, Inc.) that contains a fluorescence resonance energy transfer probe bearing a fluorophore reporter and a quencher molecule. The reaction was set up using 2 l of the cDNA and the thermal protocol consisted of 3 cycles of 95°C for 30 s and 55°C for 60 s and 50 cycles of 95°C for 45 s and 55°C for 60 s and plate read.
In every amplification experiment serial dilutions of total mRNA isolated from MCF-7 cells were carried out and reverse-transcribed under the above-mentioned conditions to produce a standard curve (for both target and housekeeping gene). The Ct values for the transcripts were plotted on the respective standard curve to obtain arbitrary units for each transcript. CRD-BP/IMP1, c-myc, or IGF-II obtained units were divided by those of ␤2-microglobulin for normalization. Normalized values were analyzed by Student's t test, and p values of Ͻ0.05 were considered significant.
For the quantitation of IGF-II production after siRNA transfections, supernatants were collected at 48 and 72 h post-transfection from 6-well plate cultures and analyzed by using the DSL-10 Ϫ2600 ACTIVE ® Non-extraction IGF-II enzyme-linked immunosorbent assay kit (Diagnostic Systems Laboratories, Inc.) following the manufacturer's instructions.
[ 3 H]Thymidine Incorporation Assay-Assays were performed in 96well plates. The day before transfection, 1 ϫ 10 4 cells were plated per well. Transient transfection of siRNAs at a final concentration of 100 nM was carried out using 0.5 l of siPORT lipid transfection agent. Cells transfected with Cont-si (negative control), CRD-si1, CRD-si2, or c-myc siRNAs were incubated for 24, 48, or 72 h and pulsed for the last 6 h of culture with 2 Ci/well [ 3 H]thymidine ([ 3 H]TdR, 30 -40Ci/mmol, Amersham Biosciences). Subsequently, cells were harvested, and [ 3 H]TdR uptake was measured in a microbeta counter (Wallac, PerkinElmer Life Sciences). Results are expressed in counts/min (cpm). Six wells per siRNA were analyzed, and the experiments were performed three times. Obtained cpm were analyzed by the Student's t test, and p values of Ͻ0.05 were considered significant.

Study of CRD-BP/IMP1 Expression in Bone Marrow, Mobilized Peripheral Blood, and Cord Blood Cell Subpopula-
tions-In all organisms so far studied CRD-BP/IMP1 expression was found to be predominant in embryonal tissues. However, these studies have not addressed the question of whether CRD-BP/IMP1 expression is indiscriminate in all embryonal cell types or it characterizes certain cell populations with stem cell properties. BM, MPB, and CB are known to contain cells with stem cell characteristics. The best-defined hematopoietic stem cells (HSCs) are found within the CD34 ϩ and the Lin Ϫ CD34 Ϫ cell populations, although in the latter at very low frequencies (18,19). We monitored for CRD-BP/IMP1 expression in cell subpopulations isolated from adult BM, adult MPB, and CB. Practically homogenous populations (purity Ͼ95%) were isolated after selection using immunomagnetic beads as described under "Materials and Methods." We employed the nested RT-PCR approach to detect CRD-BP/IMP1 expression. This technique has been used previously to monitor for CRD-BP/IMP1 expression in tumor samples and normal tissues. Using this approach CRD-BP/IMP1 has been found to be specifically expressed in fetal and not in normal adult human tissues (7, 8, 10, 11).
The results of a typical experiment are shown in Fig. 1. Among the different cell subpopulations tested, the CB CD34ϩ subpopulation was the only one found to express CRD-BP/ IMP1. All 29 CB 34 ϩ samples studied were found positive, whereas none of the 5 BM CD34 ϩ and 10 MPB CD34 ϩ or any of the Lin ϩ (CD19 ϩ , CD56 ϩ , CD14 ϩ , and CD3 ϩ ) subpopulations isolated from 5 different CB, 3 MPB, and 3 BM expressed CRD-BP/IMP1. Lin Ϫ CD34 Ϫ cells were also found negative for CRD-BP/IMP1. This observation indicated that CRD-BP/IMP1 expression may not be an indiscriminate feature of embryonal cells but is rather a stem cell characteristic.
5azaC Treatment Activates CRD-BP/IMP1 Expression in Adult BM CD34 ϩ Cells-It is noteworthy that the highly sensitive Nested-RT-PCR approach could not detect trace amounts of CRD-BP/IMP1 expression even in the enriched adult CD34 ϩ cell subpopulations. The absence of leaky expression in the adult counterparts of CB CD34 ϩ cells indicated very tight post-natal regulation of this gene. A common regulatory mechanism of gene expression during development and differentiation also involved in the permanent silencing of embryonal genes in adult tissues is the methylation of CpG dinucleotides in their regulatory regions (20). It has been shown previously that the CRD-BP/IMP1 proximal promoter region has a high CG content and is enriched in CpG dinucleotides (8). To investigate the putative involvement of CpG methylation in the epigenetic silencing of CRD-BP/IMP1 gene, we cultured adult BM CD34 ϩ cells for a period of 15 days in the presence or absence of the demethylating agent 5azaC, quiescent or stimulated with a cytokine mixture that promotes the exit from G 0 /G 1 of primitive HSC populations (21). We detected the de novo activation of CRD-BP/IMP1 expression in actively proliferating BM CD34 ϩ cells after demethylating treatment (Fig.  2). On the contrary, no expression was detected in the nondividing, 5azaC-treated cells. These data indicated that CpG methylation directly or indirectly is involved in the tight regulation of CRD-BP/IMP1 in the adult BM CD34 ϩ cells and furthermore implys that altered methylation, an epigenetic modification frequently encountered in malignancies (22,23), may contribute to its de novo activation.
Effects of CRD-BP/IMP1 Knock-down upon c-myc and IGF-II Expression-We employed the siRNA methodology to knock down CRD-BP/IMP1 expression in the MCF-7 human breast cancer cell line to monitor for any putative effect upon c-myc and IGF-II expression. These cells, besides being easily transfectable, carry few extra c-myc gene copies (24), and thus, we anticipated that any effect of CRD-BP/IMP1 knock-down upon c-myc mRNA levels could be readily detectable. Small, synthetic, double-stranded RNAs can effectively mediate gene silencing through the RNA interference mechanism and have been successfully used to knock down the expression of a myr-iad of genes in cells from various organisms (25). After optimization of siRNA and siPORT lipid concentrations, ϳ85% of the cells were labeled positive with a control non-silencing fluorescein-labeled siRNA (Fig. 3A). Furthermore, a specific c-myc siRNA (MYC-si) from Ambion was used as a positive control and, the no-silencing (non-labeled) siRNA (Cont-si), provided in the same kit, was used as a negative control to monitor the efficiency of siRNA targeting. The usage of the MYC-si resulted in a reduction of 75% of c-myc mRNA levels compared with Cont-si at 48 h after transfection (p Ͻ 0.01, data not shown). Two chemically synthesized 21-nucleotide siRNAs targeting CRD-BP/IMP1 mRNA at positions 1114 -1134 (CRD-si1) and 1468 -1488 (CRD-si2) were used, and the effects upon CRD-BP/ IMP1 mRNA levels were analyzed. Total RNA was isolated 48 h after transfection, and CRD-BP/IMP1 mRNA levels were estimated using the real-time quantitative RT-PCR approach. CRD-BP/IMP1 levels in each sample were normalized against ␤2-microglobulin levels monitored with the same methodological approach in a separate reaction. The results of a typical experiment are depicted in Fig. 3B. The mean values (ϮS.D.) of normalized CRD-BP/IMP1 mRNA levels in cells treated with the siRNAs targeting CRD-BP/IMP1 are depicted in Fig. 3C as a percentage relative to the normalized CRD-BP/IMP1 mRNA levels in cells treated with the Cont-siRNA. The results indicated that CRD-BP/IMP1 mRNA levels at 48 h after transfection were reduced by 80 and 75% upon using CRD-si1 and CRD-si2, respectively, compared with Cont-si (p Ͻ 0.01 for both siRNAs).
To monitor the effects of CRD-BP/IMP1 knock-down upon c-myc and IGF-II mRNA levels, MCF-7 cells were transfected with either the CRD-si1 or CRD-si2 21-nucleotide siRNAs. Total RNA was isolated 48 h after transfection, and c-myc and IGF-II mRNA levels were estimated using the real-time quantitative RT-PCR approach and normalized against ␤2-microglobulin levels. The mean values (ϮS.D.) of normalized c-myc mRNA levels in cells treated with the siRNAs targeting CRD-BP/IMP1 are depicted in Fig. 4A as a percentage relative to the normalized c-myc mRNA levels in cells treated with the Contsi RNA and show a 3-fold decrease in c-myc mRNA levels after CRD-BP/IMP1 knock-down (p Ͻ 0.01 for both siRNAs). To assess the effects of CRD-BP/IMP1 knock-down upon c-Myc protein levels, Western blot analysis was performed using whole cell lysates prepared at 48 h after siRNA transfection. The results depicted in Fig. 4, B and C, show that the c-myc mRNA reduction was accompanied by a reduction of c-Myc protein levels. These data provided for the first time direct evidence that the intracellular down-regulation of CRD-BP/ IMP1 may affect the c-myc mRNA and protein levels, substantiating the in vitro findings of Bernstein et al. (1). Regarding the IGF-II, the estimated mean values (ϮS.D.) of normalized mRNA levels are depicted in Fig. 4D and show an approximately 15-fold increase in IGF-II mRNA levels after CRD-BP/ IMP1 knock-down (p Ͻ 0.01, for both siRNAs). To assess whether the increase in IGF-II mRNA levels after CRD-BP/ IMP1 knock-down was accompanied by an increase in the production of IGF-II, both Western blot analysis of whole cell lysates and enzyme-linked immunosorbent assay analysis of conditioned medium, collected at 48 h post-transfection, were performed. We were not able to detect the IGF-II protein by either approach due to the fact that the protein levels were below the detection limits of both methods. This is not surprising because, whereas in MCF-7 cells IGF-II mRNA is readily detected by conventional mRNA analysis methods, the levels of the corresponding protein are low in the conditioned medium, i.e. between 5 and 15 ng/ml per 10 7 cells (26,27). Nevertheless, we cannot exclude the possibility that IGF-II production is

CRD-BP/IMP1 Expression in HSC and Function in MCF-7 Cells
increased after CRD-BP/IMP1 knock-down, as has been reported for K562 cells (28).

CRD-BP/IMP1 Knock-down Reduces MCF-7 Proliferation
Rates-To assess any putative effect of CRD-BP/IMP1 knockdown and of the concomitant c-myc mRNA and protein downregulation upon the cell proliferation rates, we measured [ 3 H]TdR uptake at 24, 48, 72, and 96 h after transfection. As shown in Fig. 5A, a reduction of ϳ22% in [ 3 H]TdR incorporation rates was observed with both CRD-BP siRNAs at 24 and 48 h post-transfection, comparable with that obtained with the MYC-si. Furthermore, we detected the up-regulation of the cyclin kinase inhibitor p21 WAF/CIP1 in cells transfected with either the CRD-BP/IMP1 or the c-Myc siRNAs (Fig. 5B). This up-regulation was most likely due to c-myc down-regulation, since as has been shown previously, c-Myc represses the promoter of this growth arrest gene (29). The antiproliferative effects were lost at 72 h post-transfection, and at the same time, the enhanced DNA synthesis was detected in MYC-sitransfected cells, most likely reflecting a substantial cell entry into the S phase as the effect of the siRNA wears out. The same effect in cells treated with the CRD-BP/IMP1 siRNA was observed at 96 h post-transfection. This delay could be due to differences in the recovery time of the two mRNAs to pretreatment levels, or CRD-BP/IMP1 may affect, additionally to c-myc, targets whose recovery may be delayed. The above-mentioned observations indicated that any putative augmented IGF-II production after CRD-BP/IMP1 knock-down could not compensate for the antiproliferative action of c-myc down-regulation in MCF-7 cells. DISCUSSION CRD-BP/IMP1 is an oncofetal RNA-binding protein that specifically recognizes at least four RNAs, i.e. the c-myc and leader 3Ј IGF-II mRNAs, the H19 RNA, and the neuron-specific tau mRNA. The critical role for CRD-BP/IMP1 in development is revealed by the finding that CRD-BP/IMP1 null mice exhibited dwarfism, impaired gut development, and high perinatal lethality (9). Growth retardation in null mice increased progressively from day E17.5, mainly due to tissue hypoplasia, and the spatiotemporal pattern of CRD-BP/IMP1 expression revealed increased expression in different embryonal days in various expanding and developing organs up to day E17.5, where its expression generally decreased. To address the question of whether CRD-BP/IMP1 expression characterizes particular embryonic cell types, we analyzed isolated CB cell subpopulations and found its expression to be restricted exclusively to the CD34 ϩ stem cell compartment. These cells are formed mainly in the fetal liver and spleen early in development and in the last trimester of pregnancy begin to migrate to BM through the blood stream, reaching the umbilical cord, where they are found at the time of birth. Thus, the above finding implies a role for CRD-BP/IMP1 gene in fetal stem cell physiology. Independent of the stage of human development, the ability to reconstitute the hematopoietic system upon transplantation in vivo resides mostly within the CD34 ϩ population. However, various studies have shown that CB, BM, and PB CD34 ϩ cells exhibit distinct functional characteristics regarding their selfrenewal, differentiation, and proliferative properties and their repopulating capacity in NOD/SCID mice (19,30,31). These differences are also reflected in their gene expression profile (19, 30 -34), and our data show that CRD-BP/IMP1 is also differentially expressed in these cell populations.
It has been postulated that CRD-BP/IMP1 could affect the posttranscriptional fate of its RNA targets by determining their sub-cellular localization, by modulating their half-life, or by affecting their translatability. Herein we showed that targeting of CRD-BP/IMP1 expression in a human breast cancer cell line is followed by c-myc mRNA and protein down-regulation and by IGF-II mRNA up-regulation. Previous studies have shown elevated, albeit not statistically significant, expression of c-myc in CD34 ϩ fetal or CB cells compared with their adult BM counterparts (34). BM CD34 ϩ cells are in a deep dormant state,and a marked increase of c-myc mRNA levels is observed upon cytokine and growth factor stimulation, probably due to the activation of various regulatory mechanisms acting upon c-myc expression. Quiescent adult CD34 ϩ ells exposed to growth factors are rapidly committed and differentiate in vitro after a limited number of cell divisions. On the other hand, CD34 ϩ cells in earlier ontogenenetic stages possess distinct proliferation characteristics, and those isolated from fetal blood in the earlier stages of development are actively cycling. Until recently no data existed regarding a functional role of c-myc in HSC physiology. These cells, upon entrance to the cell cycle, are obliged to select self-renewal or differentiation, and it is well established that Notch signaling and HOXB4 expression promote the former. Satoh et al. (35) have provided evidence indicating that c-myc is a gene target and could be the downstream mediator of Notch and HOXB4 genes in HSC. The expression of a c-Myc-inducible form in murine HSC not only enhanced the expression of the same genes as Notch and HOXB4 but also induced their growth without disrupting their biologic properties. The ectopic activation of c-Myc increases cell survival, and proliferation under all cytokine combinations studied does not compromise HSC immature characteristics and augments their reconstituting activities upon transfusion to lethally irradiated mice. On the other hand, Wilson et al. (36) have shown recently that although c-myc might be dispensable for mice BM HSC self renewal, it plays a crucial role for promoting differentiation to committed progenitors. Their model proposes that fine tuning of c-myc expression in the two daughter cells after HSC division generates asymmetries mainly regarding cell adhesion molecules, retaining one of the cells to the HSC niche and promoting leaving and differentiation of the other (36). Based on these observations it cannot be excluded that CRD-BP/IMP1 expression in fetal CD34 ϩ cells could actually play a crucial role in their self-renewal, differentiation, and proliferation properties and their repopulating activity by affecting either c-myc expression or its mRNA intracellular distribution.
The abnormal expression of three of the thus far identified CRD-BP/IMP1 targets has been detected in human tumors and this together with the oncofetal pattern of CRD-BP/IMP1 expression led to the hypothesis that its de novo activation in tumors may be associated with the process of carcinogenesis. Conclusive evidence that CRD-BP/IMP1 acts as an oncogene comes from experiments with transgenic mice exhibiting targeted expression of the gene in their adult life in mammary glands (14). In this study quantitative analysis of CRD-BP/ IMP1 RNA target levels in mammary tissues, collected after the activation of CRD-BP/IMP1 expression but before the onset of preneoplastic or neoplastic lesions, has shown the elevated expression of IGF-II mRNA and H19 RNA but not the increase of c-myc mRNA or protein levels. This finding sticks because CRD-BP/IMP1 stabilizes c-myc mRNA levels in a cell-free mRNA decay system. To elucidate any putative effect of CRD-BP/IMP1 upon c-myc expression in intact cells, we employed the siRNA approach to knock-down CRD-BP/IMP1 expression, and we found that this was followed by a decrease in c-myc mRNA and protein levels. These data implied that c-myc mRNA could be an in vivo CRD-BP/IMP1 target and substantiated the notion that CRD-BP/IMP1 de novo expression in tumors might perturb c-myc normal regulation. Furthermore, the data indicated that targeting of CRD-BP/IMP1 could have pronounced effects upon c-myc expression. This is of particular interest since c-myc is one of the most frequently deregulated genes in malignances, and its transforming action has been linked to accelerated cell proliferation rates, cell growth, and metabolism and to the inhibition of cell differentiation (37)(38)(39)(40). On the other hand, we detected a 15fold increase in the IGF-II mRNA levels after CRD-BP/IMP1 knock-down. However, we were not able to detect whether this was followed by a concomitant increase in IGF-II produced or secreted levels. Controversial data regarding the CRD-BP/ IMP1 mode of action upon IGF-II have been provided previously by the studies of Nielsen et al. (2), Hansen et al. (9), and Tessier et al. (14). Various hypotheses could be proposed to explain these differences, for example, cell type-specific modes of action or the execution of illegitimate roles by CRD-BP/IMP1 in adult tissues and/or in tumor cells. IGF-II can stimulate various cellular responses acting as a cell survival factor or mitogenic factor and can also modify metabolism. IGF-II augments in vitro both proliferation and differentiation of BM CD34 ϩ cells, whereas it supports the maintenance of a great number of progenitor cells, thus suggesting a possible role in the expansion of committed progenitors (41). Furthermore, IGF-II functions as an autocrine factor in certain tumor types, including breast carcinomas (42)(43)(44). Herein we showed that CRD-BP/IMP1 knock-down and the concomitant c-myc downregulation resulted to a reduction in cell proliferation rates in MCF-7 cells. The above finding implied that whatever the effect of CRD-BP/IMP1 knock-down upon IGF-II production in these cells, the down-regulation of c-myc expression suffices to reduce proliferation rates. Despite the fact that the antiproliferative effects observed were modest and transient, most likely due to the limitations of the methodology, these data indicated that a sustained reduction or inhibition of CRD-BP/IMP1 expression could have more pronounced effects upon cancer cell physiology, and if so, CRD-BP/IMP1 may provide a therapeutic intervention target. Recently, Liao et al. (28) reported that CRD-BP/IMP1 knock-down has no effect upon c-myc mRNA and protein levels, whereas it promotes cell proliferation via the up-regulation of secreted IGF-II in the K562 leukemic cell line. Conflicting with our data, these observations may be due to differences in the CRD-BP/IMP1 expression levels between the two cell lines. The presence of 50 c-myc mRNA copies and 400,000 CRD-BP/IMP1 protein molecules per K562 cytoplasm has been estimated (28). Upon using the quantitative RT-PCR methodology, we calculated the CRD-BP/IMP1-glyceraldehyde-3-phosphate dehydrogenase mRNA ⌬Ct values for K562 and MCF-7 to be ϳ 6 and ϳ14.5, respectively, 2 indicating the substantially lower CRD-BP/IMP1 expression in MCF-7 cells.
Thus, whereas we cannot exclude the possibility that CRD-BP/ IMP1 may exert differential cell type-specific actions, a plausible explanation for the discrepancies observed in the two cell lines could be that CRD-BP/IMP1, being in vast excess, has not been depleted to the extent that can affect c-myc mRNA expression in K562 cells. Thus, with the c-myc levels unchanged, the increased IGF-II expression may result in the promotion of proliferation, as observed by Liao et al. (28).
The de novo CRD-BP/IMP1 expression has been detected in human tumors of different origins, and in some of these tumor types CRD-BP/IMP1 expression characterizes the vast majority of the samples studied (7, 8, 10 -12). However, little is known about the mechanisms responsible for the activation of this gene in cancer. The amplification of CRD-BP/IMP1 gene locus has been detected in breast carcinomas (45); nevertheless, the high percentage of CRD-BP/IMP1-expressing samples found in tumor samples from this organ (10) cannot be explained solely by the presence of this genomic abnormality. The data presented herein indicate that this gene is tightly regulated in adult normal cells, at least in the hematopoietic tissue. However, our findings could not exclude the possibility that certain rare cells might exist in other adult tissues expressing CRD-BP/IMP1, as implied by the identification of low expression levels in mouse adult intestines (9). Upon culturing adult BM CD34ϩ cells in the presence of 5AzaC, we identified the re-activation of CRD-BP/IMP1 expression, and this finding indicated that altered methylation may be directly or indirectly involved in the silencing of this gene in adult non-expressing cells. Our data implied that altered methylation, a common characteristic of cancer cells influencing the expression of oncogenes and oncosuppressor genes (22,23), may affect the expression of CRD-BP/IMP1 in cancer. An extended study of the methylation status of CRD-BP/IMP1 promoter region in tumor samples would prove whether altered methylation is directly responsible for the de novo activation of CRD-BP/IMP1 in neoplasias.
In conclusion, our data point toward a normal role for CRD-BP/IMP1 in the biology of pluripotent stem cells with high renewal capacity, as with the CB CD34 ϩ cells, and indicate that epigenetic mechanisms are involved in the regulation of its expression in adult non-expressing cells and imply that its de novo activation in cancer cells may affect the expression of c-myc and IGF-II, whose role in the control of cell growth is well established.