Two Isoforms of the RNA Binding Protein, Coding Region Determinant-binding Protein (CRD-BP/IGF2BP1), Are Expressed in Breast Epithelium and Support Clonogenic Growth of Breast Tumor Cells

Abstract

CRD-BP/IGF2BP1 has been characterized as an "oncofetal" RNA binding protein typically highly expressed in embryonic tissues, suppressed in normal adult tissues, but induced in many tumor types. In this study, we show that adult breast tissues express ubiquitous but low levels of CRD-BP protein and mRNA. Although CRD-BP mRNA expression is induced in breast tumor cells, levels remain ∼1000-fold lower than in embryonic tissues. Despite low expression levels, CRD-BP is required for clonogenic growth of breast cancer cells. We reveal that because the most common protein isoform in normal adult breast and breast tumors has an N-terminal deletion (lacking two RNA recognition motif (RRM) domains) and is therefore missing antibody epitopes, CRD-BP expression has been under-reported by previous studies. We show that a CRD-BP mutant mouse strain retains expression of the shorter transcript (ΔN-CRD-BP), which originates in intron 2, suggesting that the impact of complete ablation of this gene in mice is not yet known. Either the full-length CRD-BP or the N-terminally truncated version can rescue the clonogenicity of CRD-BP knockdown breast cancer cells, suggesting that clonogenic function is served by either CRD-BP isoform. In summary, although CRD-BP expression levels are low in breast cancer cells, this protein is necessary for clonogenic activity.

Keywords: IMP1; RNA binding protein; RNA metabolism; RNA-protein interaction; alternate transcription site; breast cancer; clonogenicity; mRNA; mammary gland; multifunctional protein.

FIGURE 1.

Analysis of CRD-BP transcripts shows that adult mouse breast epithelial cells express a CRD-BP transcript that lacks exons 1 and 2 (ΔN-CRD-BP). A, schematic diagrams show the gene structure for CRD-BP, with exons numbered as per NCBI annotation (accession NM_009951.4), together with the proteins predicted from the differential exon linkage products. A proposed novel start site (dotted circle) for transcription of a short form mRNA(s) (new exons identified by 5′-RACE depicted as red rectangles) is indicated along with the proposed start codon (ATG) in exon VI. Primer binding sites are indicated for canonical exons 1, 2, and 3 (F1, F2, and F3, respectively), and a reverse primer in exon 15 (R), along with forward primers that recognize intron 2 sequences (Fi2a and Fi2b). The annotated 7-kb 3′-UTR is shown, together with alternative polyadenylation sites (crosses) and let7 miRNA binding sites (white circles). The detailed mRNA sequences deduced from 5′-RACE analysis are also listed. The alternative transcription start site in intron 2 is predicted to produce different proteins, designated full-length (FL) and ΔN (N-terminally deleted). Functional domains of the CRD-BP protein are shown. RRM, RNA recognition motif. B, RT-PCR analysis was performed using the primers described in A to analyze RNA isolated from wild-type MEFs and adult mouse mammary cell lines (EP and EN cells). C, RNA samples from MEFs, primary adult mouse mammary epithelial cells (MECs), and adult mouse mammary cell lines (EP and EN cells) were also analyzed using qRT-PCR to determine total CRD-BP mRNA levels (compared with MEFs) and the relative distribution of the FL and ΔN CRD-BP isoforms.

FIGURE 2.

The ΔN-CRD-BP transcript is expressed in human breast cancer cells. A, schematic diagram depicting CRD-BP gene structure as in Fig. 1A (based on NCBI accession NM_006546.3). The novel exon located in intron 2 is labeled Exon i2, and the location of the additional forward primer (Fi2) is shown. B, RNA was isolated from the human cell lines indicated, and RT-PCR analysis (as described for Fig. 1B) was performed using primers to an intronic region (orthologous to the mouse sequence) in the human CRD-BP gene (see also Fig. 4A). C, isoform-specific expression patterns of CRD-BP mRNAs were assayed by qRT-PCR analysis as described for Fig. 1C, and relative expression was compared with MCF7 breast cancer cell line. D, to evaluate whether the CRD-BP mRNA species detected express the long let7 miRNA regulated 3′-UTR, RNA preparations from the indicated cells/tissues were analyzed by qRT-PCR analysis. Results are shown comparing signals from a primer set amplifying a region in the 7-kb-long 3′-UTR to signals from a primer set amplifying total CRD-BP (amplicon located in exon 15). E, RNASeq data from the Cancer Genome Atlas was analyzed to determine the expression of CRD-BP exons 1–15 in normal tissue (purple line) compared with primary breast tumor tissues, divided by subtype. Tumor numbers: basal, n = 142; Her2, n = 67; Luminal A (Lum-A), n = 434; Luminal B (Lum-B), n = 194. Exons are numbered along the x axis, whereas the y axis shows relative abundance. Potential exon linkage products (based on their approximately similar abundance) are indicated with horizontal black arrows.

FIGURE 3.

A CRD-BP mutant mouse strain retains expression of the ΔN-CRD-BP isoform. A, a schematic diagram showing the structure of the mutant CRD-BP allele, created by insertional mutagenesis with a cassette containing βgeo-pA-PGK-PURO-Stop (35), together with primer locations (indicated as for Fig. 1A). B, RNA isolated from wild-type and CRD-BP^hypo MEFs was analyzed by RT-PCR analysis using the primers indicated (left panel). qRT-PCR was used to assess total CRD-BP mRNA levels (compared with wild-type cells) and to measure the relative expression of the ΔN-CRD-BP isoform (right panel). C, wild-type and CRD-BP^hypo MEFs were fixed and stained with anti-CRD-BP antibodies as indicated (detailed under “Experimental Procedures”; nuclei were counterstained with DAPI). Epitope locations for antibodies used in this study are indicated in the CRD-BP schema depicted above the images. Scale bars, 50 μm. D, Western blots of protein lysates from wild-type and CRD-BP^hypo MEFs were probed with the CRD-BP antibodies indicated. E, Western blots of protein lysate from 293T cells probed with the C-terminally reactive antibody (kind gift from David Herrick).

FIGURE 4.

Homologous alternative transcription start sites in intron 2 for mouse and human CRD-BP genes. A, the human and mouse CRD-BP alleles are illustrated, together with summary data for detection of the active histone mark (H3K4Me3), summed transcription factor binding activity (from ENCODE ChIP-Seq data), and distribution of CpG islands often associated with active promoters. The bottom panel shows the sequence conservation across intron 2 for human and mouse CRD-BP, as well as filtered ChIP-Seq data exclusively showing RNA pol2 (POLR2A) binding. B, Kozak sequence analysis identified the predicted start codon for translation from the ΔN-CRD-BP isoform with the alternative transcription initiation site in intron 2. Amino acid sequences for full-length and ΔN-CRD-BP are shown.

FIGURE 5.

Expression of CRD-BP in human breast tumors. A, immunohistochemical analysis of a breast tumor microarray and control tissue sections (liver and colon) stained for CRD-BP using the VS mAb to CRD-BP (diaminobenzidine/brown stain), targeted against the C-terminal domain of CRD-BP, and counterstained with hematoxylin (pink). Representative stains of each subtype of breast tumor along with a normal breast tissue section are shown. Stains of normal human colon and liver are shown for comparison. Scale bars, 50 μm for low magnification images (left column) and 5 mm for high magnification images (right column). mag, magnification. B, the relative levels of CRD-BP mRNA were assessed from expression arrays associated with the Cancer Genome Atlas data set (670 human breast tumors; probe sets target 3′-UTR (53)) divided by tumor subtype, defined by their PAM50 signature (as indicated, including normal-like tumors). CRD-BP transcript levels are expressed as log₂ fold change over the median CRD-BP expression level across all tumor types. The boxes enclose the 25^th to 75^th percentile data points, with the horizontal bars showing the median values. The whiskers demark the 10^th and 90^th percentiles. *, p < 0.05. C, RNA isolated from primary human normal (N1–N2) and tumor (T1–T9) breast tissue was analyzed by qRT-PCR with the same primers used for Fig. 2C to assess the relative expression of CRD-BP isoforms. D, qRT-PCR analysis was performed as for Fig. 2D to assess the expression of the long 7-kb 3′-UTR of CRD-BP mRNA in primary human normal and tumor breast tissue.

FIGURE 6.

CRD-BP is required for clonogenic growth of breast tumor cell lines. A, human breast cell lines were transfected with an shRNA construct targeting the human CRD-BP gene (CRD-BP shRNA) or scrambled shRNA control (Scr shRNA). Knockdown efficiency of total CRD-BP mRNA was assessed 48 h post-transfection by qRT-PCR analysis. Representative knockdown efficiencies for two cell lines are shown (experiments were repeated at least three times). B, human breast cell lines were subcultured 24 h post-transfection with CRD-BP shRNA (CRD-BP shRNA) or a scrambled shRNA control (Scr shRNA) either to clonal density (10⁴ cells/10-cm² plate) for functional assay or by passaging (2-fold dilution) for assessment of knockdown efficiency (A). Colony formation was assessed by crystal violet staining ∼1 week postsubculture (as detailed under “Experimental Procedures”). Representative images of this assay are shown (left panel), and results were quantified for each cell type (right panel). C, the clonogenicity of stable CRD-BP knockdown mouse mammary cells was measured (EP and EN cells stably transduced with lentiviral shRNA constructs). Knockdown efficiencies are shown, determined by qRT-PCR analysis. Two CRD-BP shRNA constructs were used, and results were normalized to the control shRNA-transduced condition. D, crystal violet staining was performed as for B in mouse mammary EP and EN cells stably transduced with one of two CRD-BP shRNA constructs (CRD-BP shRNA-1 or shRNA-2) or a control shRNA construct (Control shRNA). Representative images from colony formation assays are shown, and quantified at right.

FIGURE 7.

Both full-length and ΔN-CRD-BP isoforms can rescue clonogenicity of CRD-BP knockdown breast cancer cells. A, MCF7 cells were transfected with human CRD-BP shRNA construct (or a scrambled shRNA construct as a control), fixed 48 h later, and stained with VS mAb to CRD-BP, illustrating the dynamic range of this visual assay. Representative images are shown. Scale bars, 10 μm. B, MCF7 cells were transfected with expression constructs encoding mouse full-length or ΔN-CRD-BP isoforms (or an empty vector as a control) and stained as for A. Higher magnification pictures show the subcellular distribution of CRD-BP. Scale bars, 50 μm for the top panel and 10 μm for the bottom panel. C, to test the rescue activity of the full-length and ΔN-CRD-BP species, combinations of specific or control shRNA to human CRD-BP, together with overexpressed mouse CRD-BP isoforms (or an empty vector control), were transfected into MCF7 cells. Overexpression of both isoforms was assessed by Western blot analysis using CRD-BP antibodies to the N-terminal domain (Cell Signaling) or the C-terminal domain (gift from Jeff Ross; left panel; see also epitope locations marked in Fig. 3C) and also by qRT-PCR analysis of total CRD-BP mRNA levels (right panel). D, MCF7 cells transfected with the indicated constructs were subcultured to clonal densities as for Fig. 5A (experiments were repeated at least three times). After ∼1 week, colonies were stained with Crystal Violet dye (optical density quantified below). ***, p < 0.05; **, p = 0.06; *, p = 0.08.