Genetic Alterations Disrupting the Nuclear Localization of the Retinoblastoma-related Gene RB2/p130 in Human Tumor Cell Lines and Primary Tumors

Cell Culture and Transfection.

The cell lines were obtained from the ATCC (Manassas, VA) and the European Collection of Animals Cell Cultures. The four human osteosarcoma cell lines (Saos-2, Hos, MG-63, and U2OS from ATCC) were grown at 37°C in DMEM supplemented with 15% fetal bovine serum, whereas CCRF-CEM (acute T lymphoblastic leukemia), Molt-4 (acute T lymphoblastic leukemia), and Daudi (B lymphoblast Burkitt’s lymphoma) from the ATCC and Jurkat (leukemia T-cell lymphoblast) from the European Collection of Animals Cell Cultures were grown in RPMI 1640 plus 10% fetal bovine serum.

Saos-2 cells were plated at a concentration of 1 × 10⁶ cells/plate in triplicate. After 24 h, the cells were transfected by the standard calcium phosphate precipitation method (25) .

Immunofluorescence and Confocal Microscopy Analysis.

All cell lines were fixed in 4% paraformaldehyde in 1× PBS and permeabilized with 0.8% Triton X-100 for 15 min. Slides were incubated to block nonspecific binding with 2% BSA and 3% normal goat serum in PBS (immunoreaction buffer) for 30 min at 37°C, reacted with the primary monoclonal anti-Rb2/p130 antibody (clone 10; Transduction Laboratories, Lexington, KY) diluted in immunoreaction buffer 1:50 for 3 h at 37°C, and then reacted with secondary FITC-conjugated rabbit antimouse IgG (Sigma, St. Louis, MO) for 1 h at 37°C. DNA was counterstained with 4′,6-diamidino-2-phenylindole (Sigma) to assess the nuclear domains. The samples were analyzed by a Zeiss LSM410 (Carl Zeiss) confocal laser scanning microscope equipped with a 100× oil immersion lens (numerical aperature = 1.4) and a 488/514 nm argon laser. Image acquisition, recording, and filtering were performed on the z series of confocal data (stacks) by an Indy 4600 graphic workstation (Silicon Graphics) as described previously (31) .

Western Blot Analysis.

Whole cell lysates were prepared by resuspending cell pellets in 200μ l of lysis buffer (50 mm Tris-HCl, 5 mm EDTA, 250 mm NaCl, 50 mm NaF, 0.1% Triton, 0.1 mm Na₃VO₄, and protease inhibitors 1 mm phenylmethylsulfonyl fluoride and 1 μg/ml aprotinin and leupeptin). The lysates were cleared by centrifugation for 15 min at 13,000 × g at 4°C, and total protein extracts were determined (25) . The protein (40 μg) was denatured by boiling in 2× Laemmli buffer[ 62.5 mm Tris-HCl (pH 6.8), 2% SDS, 10% glycerol, 5% β-mercaptoethanol, and bromphenol blue] and size-fractionated by electrophoresis in 6% SDS-polyacrylamide gel. The electrophoretic transfer of the protein to a polyvinylidene difluoride membrane (Millipore) was performed in 3-(cyclohexylamino)propanesulfonic acid buffer [10 mm 3-(cyclohexylamino)propanesulfonic acid and 2% methanol (pH 11)]. The membrane was blocked with 5% fat-free dried milk in TBS-T buffer [2 mm Tris, 13.7 mm NaCl, and 0.1% Tween-20 (pH 7.6)] and incubated with the primary monoclonal antibody (Transduction Laboratories) at a dilution of 1:500 in 3% milk. After several washes, the membrane was incubated with antimouse antibody coupled with horseradish peroxidase (Amersham) for 1 h and detected using the enhanced chemiluminescence system (Dupont NEN; Ref. 32 ).

PCR and SSCP Analysis.

The PCR reaction mixture (50 μl) contained genomic DNA at a final concentration of 4 ng/μl, 0.2 mm of each of the four deoxynucleotide triphosphates, 2 units of Klen TaqI (Ab Peptides), and the intron primer panels (exon 19, 5′-AGGTCCTATCACCAAGGGTGT-3′; exon 19 reverse primer, 5′-GCTTAGTTACTTCTTCAAGGC-3′; exon 20, 5′-GAGAAAGTTAATATCCTAGCTG-3′; exon 20 reverse primer, 5′-GTGAATGGTCCATATATAAATCA-3′; exon 21, 5′-TGGTTTAGCACACCTCTTCAC-3′; exon 21 reverse primer, 5′-GCTTAGCACAAACCCTGTTTC-3′; exon 22, 5′-CTGAGCTATGTGCATTTGCA-3′; exon 22 reverse primer, 5′-AAGGCTGCTGCTAAACAGAT-3′) at a final concentration of 0.4 μm each. Thirty-five cycles of denaturation (95°C, 1 min), annealing (55°C, 1 min), and extension (72°C, 1 min) linked to one cycle at 72°C for 7 min were carried out in a thermal cycler (Perkin-Elmer, Norwalk, CT). For all of the tumor cell lines, reverse transcription-PCR was performed using primers 5′-ATTTAGCAGCTGTCCGC-3′ (forward) and 5′-AAGGCTGCTGCTAAACAGAT-3′ (reverse) that amplified the region from nucleotide 2603–3540. The annealing temperature used was 56°C, and the cDNAs amplified were 937 bp.

MDE gel solution (FMC BioProducts, Rockland, ME; supplied as 2× liquid concentrate) was used for SSCP analysis. For a 100-ml total volume, 25 ml of 2× MDE gel solution, 6 ml of 10× TBE buffer, 69 ml of deionized water, 40 μl of N,N,N′,N′-tetramethylethylenediamine, and 400 μl of 10% ammonium persulfate were used as recommended by AT Biochem. Genomic PCR products (2 μl) were added to 9 μl of stop solution and heated to 94°C for 2 min. The denatured DNA was placed directly on ice for several minutes and then run through the MDE gel at 8 W constant power for 8 h at 15°C in 0.6× TBE running buffer. The bands were visualized with a silver staining method (Bio-Rad, Hercules, CA).

Sequence Analysis.

The PCR products of genomic DNAs and cDNAs were resolved on a 1.5% ethidium bromide-stained agarose gel. Bands were cut from the gels. DNA was purified using the QUIAquick gel extraction kit (Qiagen, Santa Clarita, CA) and used for automated DNA forward and reverse sequencing using dideoxy terminator reaction chemistry for sequence analysis on the Applied Biosystem Model 373A DNA sequencer. The forward and reverse sequences were repeated more times for each PCR product.

Site-directed in Vitro Mutagenesis.

Oligonucleotides encoding the putative bipartite NLS of pRb2/p130, the region encoding amino acids 1082–1102, were synthesized with the addition of an ATG transcriptional start site at the NH₂ terminus, 5′ HindIII and 3′ BamHI restriction sites, and two adenosine residues just before the BamHI site to keep the heterologous fusion protein between the bipartite NLS and the NH₂ terminus of the EGFP protein in frame. The oligonucleotides were annealed and ligated into the HindIII and BamHI restriction sites of the pEGFP-N1 expression construct to form the pEGFP-N1-NLS construct. pEGFP-N1-NLS-NQ1, pEGFP-N1-NLS-NQ2, and pEGFP-N1-NLS-NQ1&2 were constructed as described previously, except that the oligonucleotides synthesized encoded point mutations that altered amino acids Lys¹⁰⁸² to Asn and Arg¹⁰⁸³ to Gln in the first bipartite site and Lys¹¹⁰⁰ to Asn and Arg¹¹⁰² to Gln in the second site, and the Lys-to-Asn and Arg-to-Gln mutations were combined in both sites, respectively.

The point mutations in the bipartite NLS of full-length Rb2/p130 cDNA were formed by an extra long PCR technique developed in our laboratory that allows site-specific mutagenesis (33) . Wild-type Rb2/p130 cDNA with an exogenous HA epitope at the COOH terminus (pcDNA3-Rb2/p130-HA) as well as with the HA epitope and the c-myc major NLS (HAN) at the COOH terminus (pcDNA3-Rb2/p130-HAN) in the pcDNA3 vector were used as the PCR template. The pcDNA3-Rb2/p130- constructs contained the equivalent Lys-to-Asn and Arg-to-Gln mutations in the bipartite NLS as described above in the first site (NLS-NQ1), the second site (NLS-NQ2), or both (NLS-NQ1&2), with either the HA or HAN epitope, as indicated. The pcDNA3-Rb2-p130-WT and pCMV-CD20 constructs, which express the wild-type pRb2/p130 without any epitope tags and the interleukin 2 receptor, respectively, were described previously (25) .

Flow Cytometry Analysis and Colony Formation Assays.

Flow cytometry analysis (FACS) was carried out according to the procedure described previously (25) . Ten μg of DNA (pcDNA3, pcDNA3-Rb2/p130-WT, pcDNA3-Rb2/p130-HA, pcDNA3-Rb2/p130-HAN, pcDNA3-Rb2/p130-NLS-NQ1-HA, pcDNA3-Rb2/p130-NLS-NQ1-HAN, pcDNA3-Rb2/p130-NLS-NQ2-HA, pcDNA3-Rb2/p130-NLS-NQ2-HAN, pcDNA3-Rb2/p130-NLS-NQ1&2-HA, or pcDNA3-Rb2/p130-NLS-NQ1&2-HAN) were cotransfected with 2 μg of pCMV-CD20. Eighteen h after transfection, the cells were washed twice with 1× PBS and once with culture medium and incubated with fresh medium at 37°C. The cells were collected with 1× PBS and 0.1% EDTA at 48 h, processed for FACS analysis by incubation with the FITC-conjugated anti-CD20 monoclonal antibody (Becton Dickinson, Franklin Lakes, NJ), fixed in 70% ethanol, stained with propidium iodide, and treated with RNase A, as described previously (25) . FACS analysis was performed on a Coulter Elite apparatus, and data from 1 × 10⁴ CD20-positive cells were used to determine the cell cycle distribution of the selected cells.

The colony formation assay was performed by transfecting triplicate dishes of Saos-2 cells with 10 μg of the indicated DNAs according to the procedure described previously (25) . Saos-2 cells were selected with 800 μg/ml G418 for 3 weeks and then stained with 1% methylene blue in 50% ethanol.

Detection of EGFP and EGFP-Fusion Proteins.

Sterile glass coverslips were placed into 10-cm tissue culture dishes, and Saos-2 cells were plated 1 × 10⁶ cells/dish. Forty-eight h after transfection with the indicated pEGFP-N1 (Clontech, Palo Alto, CA) constructs and fusion proteins, cells were washed three times with 1× PBS and fixed directly on coverslips in freshly made 4% paraformaldehyde for 30 min at room temperature. Cells were then washed twice with 1× PBS, counterstained with 0.04 mg/ml propidium iodide in 1× PBS, mounted onto a glass microscope slide with a drop of PBS, and sealed with rubber cement. Slides were examined with a Leitz Wetzlar fluorescence-equipped microscope. Images for illustration purposes were obtained using a cooled charge-coupled device camera (Princeton Instruments, Trenton, NJ). The images of the EGFP and EGFP-fusion proteins were superimposed on those of the propidium iodide staining.

Expression of pRb2/p130 in Tumor Cell Lines.

To study the status of pRb2/p130, the expression of the protein was determined by immunocytochemistry in four osteosarcoma cell lines and four lymphoid tumor cell lines as well as in normal peripheral blood cells using a monoclonal antibody against the NH₂-terminal region of pRb2/p130.

As depicted in Fig. 1⇓ , all of the lymphoid tumor cell lines show a prevailing localization of the protein at the cytoplasmic level (Fig. 1, a–d⇓ ), whereas normal human lymphocytes (Fig. 1e⇓ ) used as a control and the osteosarcoma cell lines (Fig. 1, f–i⇓ ) all exhibit an exclusive nuclear localization of pRb2/p130.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

All cell lines were reacted with monoclonal anti-Rb2/p130 antibody. Confocal sections (1.5 μm apart) of FITC fluorescence were performed. In all of the lymphoid cell lines, a clear cytoplasmic localization of fluorescence was detectable (a, CCRF-CEM; b, Molt-4; c, Jurkat; d, Daudi cell lines). Normal human lymphocytes were used as a control. In the osteosarcoma cell lines, the fluorescence appears brightest in specific regions of the nucleus, which correspond with the inner nuclear domains (e, normal lymphocytes; f, Saos-2; g, Hos; h, MG63; i, U2OS). Bar, 2 μm.

Because mutations within the bipartite NLS in the COOH terminus of RB/p105 abrogate nuclear localization in certain tumors, we hypothesize that the cell lines that showed a cytoplasmic localization of the pRb2/p130 protein may harbor mutations disrupting the RB2/p130 NLS.

Mutational Screening.

In pRb2/p130, like pRb/p105 (12) , a putative NLS consisting of two clusters of basic residues separated by a stretch of amino acids is present in the COOH terminus of the protein. This sequence is highly homologous with the pRb/p105 NLS (Table 1)⇓ (in bold). In addition, the majority of the point mutations in RB/p105 are located in the B domain and COOH-terminal region (34) . To verify whether or not the intracellular distribution of the pRb2/p130 protein depends on mutations that affect NLS motifs differently, we studied the structure of exons 19–22 of the RB2/p130 gene that encode the B domain and the COOH terminus of the protein, where the putative NLS is located (30) . Genomic DNA sequences from coding exons 19–22 were amplified and screened for mutation by SSCP analysis. The direct PCR products were sequenced to identify the actual mutations. To reduce artifactual misincorporation generated by Taq polymerase, we used high-fidelity Taq, and forward and reverse sequences were repeated three times for each sample using PCR products derived from different amplifications. Human placental DNA and DNA from normal peripheral blood lymphocytes were used as controls for PCR reactions, SSCP analysis, and sequences. CCRF-CEM cells showed a homozygous insertion in exon 21, and Daudi and Molt-4 cell lines showed a heterozygous insertion that caused the loss of the bipartite NLS resulting from a shift in the coding frame with a consequent stop codon upstream of the NLS present in exon 22 (Table 2)⇓ . The Jurkat cell line showed two heterozygous mutations located at the end of exon 21 that caused the loss of the two serines upstream of the NLS. Moreover, two other heterozygous mutations inside the first and second regions of the bipartite NLS were present in this cell line (Table 2)⇓ . These compound mutations caused the loss of nuclear localization of the protein as shown by immunocytochemistry (Fig. 1⇓ ).

None of the osteosarcoma cell lines demonstrated any insertions or mutations that could alter the putative NLS, even if point mutations in exons 19, 20, and 21 were identified (Table 2)⇓ .

To rule out the hypothesis that mutations of the RB2/p130 gene are a product of cell growth in culture, we performed the same genetic analysis on primary Burkitt’s lymphomas positive for EBV. The mutations and insertions present in four primary tumors were the same as those found in the Daudi cell line, which is derived from a Burkitt’s lymphoma (Table 2)⇓ .

Because some of the mutations found in the lymphoid cell lines caused a premature stop codon, predicting a shorter trascription product with respect to wild type, we performed a Western blot analysis to detect the molecular weight of the pRb2/p130 protein in these cell lines.

Western Blot Analysis.

Western blot analysis of the cell lysates revealed the wild-type phosphorylated and hypophosphorylated forms (M_r 130,000/M_r 120,000) of the protein in the osteosarcoma cell lines (Saos and Hos) and in three of the lymphoid cell lines examined (Jurkat, Daudi, and Molt-4). One band at a different molecular weight with respect to the wild type is detectable in the same lymphoid cell lines (Fig. 2⇓ ). In the Jurkat cell line, only the unphosphorylated form of M_r 120,000 was present because two heterozygous mutations upstream of the NLS changed two serines into two arginines, altering two putative sites of phosphorylation (see Table 2⇓ ). Therefore, mutations resulting in a persistent hypophosphorylated form of the protein should result in an enhanced growth suppressive activity, but the concomitant presence in this cell line of two point mutations in the bipartite NLS prevents this activity, confining the protein to the cytoplasm. An abnormal band at M_r 116,000 was present in CCRF-CEM, Daudi, and Molt-4 cells, resulting from a guanosine or cytosine insertion, respectively, that caused a frameshift in the sequence and a premature stop codon with a protein of predicted molecular weight of 116,000. In addition, the presence of the M_r 130,000/M_r 116,000 and M_r 130,000/M_r 120,000/M_r 116,000 bands in Daudi and Molt-4 cells, respectively, was the result of different heterozygous mutations in these cell lines that disrupt the phosphorylation status in a different way (see Table 2⇓ ). In fact, these mutations also changed the amount of serines and threonines that could be phosphorylated.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Western blot analysis of whole cell lysates using a monoclonal anti-pRb2/p130 antibody that recognizes the NH₂-terminal region of pRb2/p130. The M_r 130,000 and M_r 120,000 wild-type bands of the hyperphosphorylated and hypophosphorylated forms are shown. The M_r 116,000 mutant form derived from the frameshift insertion induced by the insertion of guanosine (G) or cytosine (C) is also present.

Identification of the pRb2/p130 NLS.

To determine the functional consequences of disruptions in the putative NLS of pRb2/p130, we first determined whether or not this region can serve in and of itself as a NLS. The putative NLS of pRb2/p130 from amino acids 1082–1102 was fused to the NH₂ terminus of EGFP in the pEGFP-N1-NLS expression vector, which expresses a human codon-optimized, red-shifted green fluorescent protein that can be fused to heterologous proteins serving as a fluorescent tag (Clontech). EGFP is a low molecular weight protein that lacks any localization signal and is equally distributed in the nuclear and cytoplasmic compartments (Fig. 3B⇓ ). Point mutations were also constructed in the first region of the putative bipartite NLS, resulting in a change of amino acids Lys¹⁰⁸² to Asn and Arg¹⁰⁸³ to Gln (pEGFP-N1-NLS-NQ1), and in the second site, resulting in a change of amino acids Lys¹¹⁰⁰ to Asn and Arg¹¹⁰² to Gln (pEGFP-N1-NLS-NQ2), as well as the combination mutations in both bipartite sites (pEGFP-N1-NLS-NQ1&2). This was done because similar point mutations in the downstream sequences of the bipartite NLS of pRb/p105 (12) and nucleoplasmin disrupted nuclear localization (35) . The plasmids were transfected into Saos-2 cells, and the locations of the ectopically expressed proteins were determined by fluorescence microscopy. Saos-2 cells were chosen because pRb2/p130 is found to be exclusively nuclear in these cells, confirming that they do not harbor any mutations in other proteins that may affect nuclear shuttling directed by this region. The cells were counterstained with propidium iodide. As shown in Fig. 3⇓ , the level of background green fluorescence was determined to be nonsignificant in mock-transfected cells (Fig. 3A⇓ ). In pEGFP-N1-transfected cells, EGFP was expressed ubiquitously in the cell (Fig. 3B⇓ ). Fusion of the wild-type putative bipartite NLS to EGFP localized expression exclusively in the nucleus (Fig. 3C⇓ ). EGFP fused to point mutations in either the upstream region (Fig. 3D⇓ ) or downstream region of the bipartite NLS (Fig. 3E⇓ ) resulted in primarily nuclear expression. Mutations in both bipartite sites resulted in an expression pattern without any nuclear targeting, and the fusion protein equilibrated between the nuclear and cytoplasmic compartments (Fig. 3F⇓ ) because it was of such low molecular weight. This was the same expression pattern seen with EGFP alone (compare Fig. 3, B and F⇓ ). Therefore, this region of pRb2/p130 serves as a bipartite NLS where both the upstream and downstream signals can independently dictate nuclear expression. Only combined mutations in both the bipartite sites resulted in a complete loss of exclusive nuclear expression. This is consistent with the mutational data from the lymphoid cell lines, where both sites of the bipartite NLS were disrupted by insertions or by point mutations.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Localization in Saos-2 cells of ectopic EGFP expression and EGFP-fusion proteins to the wild-type and mutant forms of the putative bipartite NLS of pRb2/p130. Saos-2 cells were transfected with (A) mock (no background green fluorescence), (B) pEGFP-N1 (diffuse expression in the cytoplasm and nucleus), (C) pEGFP-N1-NLS (exclusively nuclear expression), (D) pEGFP-N1-NLS-NQ1 (primarily nuclear expression), (E) pEGFP-N1-NLS-NQ2 (primarily nuclear expression), and (F) pEGFP-N1-NLS-NQ1&2 (diffuse expression in the cytoplasm and the nucleus). Images of propidium iodide counterstaining and EGFP or EGFP-fusion protein expression patterns were captured on a fluorescence microscope and then overlaid.

Functional Consequences of NLS Disruption.

The effects of mutations on the growth-suppressive function of pRb2/p130 were analyzed by recreating the NQ point mutations by PCR-based site-directed mutagenesis in the full-length pRb2/p130 protein in the pcDNA3 mammalian expression vector that drives gene expression by the constitutive CMV promoter. Each of the constructs was tagged at the COOH terminus with a single HA epitope (HA tag from Hemophilus influenzae) so that exogenous expression could be distinguished from that of the endogenous protein. Each of the mutants and wild-type plasmids was expressed at approximately the same level, as determined by immunoprecipitation and Western blot analysis with the HA epitope (data not shown). Saos-2 cells were transfected with the mutant plasmids as well as with vector alone and with plasmids expressing the wild-type pRb2/p130 protein as controls, and their effects on cellular proliferation were measured by FACS analysis. The wild-type Rb2/p130 cDNA (pcDNA3-Rb2/p130) with or without an exogenous HA epitope at the COOH terminus (pcDNA3-Rb2/p130-HA) as well as with the HA epitope and the c-myc major NLS (HAN) at the COOH-terminus (pcDNA3-Rb2/p130-HAN) in the pcDNA3 vector were used. Additionally, the NLS mutants (NLS-NQ1-HA, NLS-NQ2-HA, or NLS-NQ1&2-HA) and the NLS mutants with the HA epitope and the c-myc major NLS (HAN) at the COOH terminus (NLS-NQ1-HAN, NLS-NQ2-HAN, or NLS-NQ1&2-HAN) were also used. Transient transfections using 10 μg of either plasmid were performed along with 2 μg of pCMV-CD20 encoding for the interleukin 2 receptor. Data from 1 × 10⁴ CD20-sorted cells were used to determine the cell cycle distribution of the selected cells. With respect to mock-transfected cells, in which 42% of cells were in G₀-G₁, 32% in S phase, and 26% in G₂-M phase, the amount of G₀-G₁-blocked cells was increased to about 60% in either pRb2/p130 WT-, pRb2/p130 HA-, and pRb2/p130 HAN-transfected cells. Mutations within either the upstream or downstream region of the bipartite NLS had no significant effect (G₀-G₁ about 60%) on the growth-suppressive activity of pRb2/p130 (pRb2/p130 NQ1-HA, pRb2/p130 NQ1-HAN, pRb2/p130 NQ2-HA, and pRb2/p130 NQ2-HAN). Their expression still led to a G₀-G₁ phase growth arrest that was consistent with the localization data, demonstrating that each of the bipartite signals can independently direct nuclear transport (Fig. 3, D and E⇓ ). On the contrary, the combined mutations in both bipartite sites (pRb2/p130 NQ1&2-HA) almost completely abolished the growth-suppressive activity of pRb2/p130 because the amount of cells in G₀-G₁ was identical to that of mock-transfected cells (G₀-G₁ about 42%). Addition of the c-myc major NLS to the COOH terminus of this mutant (pRb2/p130 NQ1&2-HAN) by targeting its expression to the nucleus restored its G₀-G₁ growth arresting activity to that of the wild-type protein (G₀-G₁ almost 60%). This demonstrated that the reduced biological activity of pRb2/p130 conferred by these mutations was indeed conferred by their disruption of the nuclear localization of pRb2/p130 and not by the induction of gross conformational changes and/or the abrogation of binding to critical protein targets. In addition, the growth-suppressive effects of pRb2/p130 overexpression were not merely due to the toxicity of the protein because point mutations in the protein effectively abrogated this activity. Furthermore, pRb2/p130 must be in the nucleus to regulate G₁ progression. To rule out the hypothesis that the growth-suppressive effects of pRb2/p130 overexpression could be because of mere toxicity of the protein, we performed colony assays to study the long-term effects of mutant pRb2/p130 overexpression. As shown in Table 3⇓ , the mutant NLS-NQ1&2-HA that carries mutations within either the upstream or downstream region of the bipartite NLS showed no significant growth-suppressive activity with respect to the wild-type pRb2/p130. Insertion of the c-myc major NLS into the COOH terminus of mutant NLS-NQ1&2-HA to make the construct NLS-NQ1&2-HAN by targeting its expression to the nucleus restored the growth arresting activity to that of the wild-type protein in the colony assay, confirming the results of the FACS analysis.