Inhibitor of DNA Binding 4 (ID4) is highly expressed in human melanoma tissues and may function to restrict normal differentiation of melanoma cells

Abstract

Melanoma tissues and cell lines are heterogeneous, and include cells with invasive, proliferative, stem cell-like, and differentiated properties. Such heterogeneity likely contributes to the aggressiveness of the disease and resistance to therapy. One model suggests that heterogeneity arises from rare cancer stem cells (CSCs) that produce distinct cancer cell lineages. Another model suggests that heterogeneity arises through reversible cellular plasticity, or phenotype-switching. Recent work indicates that phenotype-switching may include the ability of cancer cells to dedifferentiate to a stem cell-like state. We set out to investigate the phenotype-switching capabilities of melanoma cells, and used unbiased methods to identify genes that may control such switching. We developed a system to reversibly synchronize melanoma cells between 2D-monolayer and 3D-stem cell-like growth states. Melanoma cells maintained in the stem cell-like state showed a striking upregulation of a gene set related to development and neural stem cell biology, which included SRY-box 2 (SOX2) and Inhibitor of DNA Binding 4 (ID4). A gene set related to cancer cell motility and invasiveness was concomitantly downregulated. Intense and pervasive ID4 protein expression was detected in human melanoma tissue samples, suggesting disease relevance for this protein. SiRNA knockdown of ID4 inhibited switching from monolayer to 3D-stem cell-like growth, and instead promoted switching to a highly differentiated, neuronal-like morphology. We suggest that ID4 is upregulated in melanoma as part of a stem cell-like program that facilitates further adaptive plasticity. ID4 may contribute to disease by preventing stem cell-like melanoma cells from progressing to a normal differentiated state. This interpretation is guided by the known role of ID4 as a differentiation inhibitor during normal development. The melanoma stem cell-like state may be protected by factors such as ID4, thereby potentially identifying a new therapeutic vulnerability to drive differentiation to the normal cell phenotype.

Figure 1. Formation of 3D-MBs under hESC conditions.

(A) Survey of melanoma cell lines for the ability to form attached 3D-spheroids in hESC medium. The 1205Lu cell line formed attached 3D-spheroids, the 451Lu cell line formed floating spheroids, while the other tested cell lines displayed only various degrees of aggregation. (B) Formation of attached 1205Lu 3D-MBs after transfer to hESC medium over time. (C) To confirm that 1205Lu MBs formed by aggregation, 1205Lu GFP-labeled cells and 1205Lu nuclear dsRed-labeled were mixed and cultured as described in Panel A. (D) Mature 1205Lu 3D-MBs were removed by pipetting and were re-plated under standard monolayer media growth conditions, under which they reseeded a new monolayer. Shown is monolayer outgrowth after 96 hours.

Figure 2. Expression microarray analysis of monolayer versus 3D-MB cultures.

(A) Diagram of experimental design for comparison of transcriptomes of monolayer and 3D-MB cells. OM, original monolayer; MB, melanoma bodies; NM, new monolayer. (B) Raw microarray data for six housekeeping genes in OM, MB and NM RNA samples: ACTB, Beta-Actin; ACTG1, Gamma 1 Actin; GAPDH, Glyceraldehyde-3-Phosphate Dehydrogenase; LDH1, Lactate Dehydrogenase; UBC, Ubiquitin C; VIM, Vimentin. An average value was calculated from 10 spots, for each gene. (C) Sampling of genes upregulated 5-fold or greater in the MB sample compared to the OM and NM samples (MB>OM, MB>NM). Genes are ordered according to the fold-upregulation in the MB sample versus OM sample (MB>OM). Samples were normalized versus an average of three housekeeping genes, GAPDH, ACTB and HSP90 (see Panel B), and error bars represent the standard deviation. Grey filled circles indicate genes categorized with the GO term “developmental process.” (D) Sampling of genes downregulated 5-fold or greater in the MB sample compared to the OM and NM samples (MB<OM, MB<NM). Genes are ordered according to fold-downregulation in the MB sample versus OM sample (MB<OM). Grey filled circles indicate genes categorized with the GO terms “inflammatory response” or “cell motility.” (E) Venn diagram showing the number of upregulated and downregulated (unshared and shared) genes when OM, MB and NM are compared. (F) Heat maps of genes upregulated and downregulated in MB samples compared to OM and NM samples.

Figure 3. Gene Ontology analysis of genes upregulated and downregulated.

GO analysis was performed to determine biological processes associated with MB-upregulated and MB-downregulated genes. Bars represent the number of regulated genes. MB-upregulated genes clustered mainly to developmental processes (top panel), while MB-downregulated genes clustered mainly to signaling, motility and invasion-related processes (bottom panel).

Figure 4. Neural stem cell features of 3D-MBs.

(A) Data extracted from 1205Lu microarray analysis shows upregulation of neural stem cell genes in MB versus OM samples, normalized to average expression of three housekeeping genes, GAPDH, ACTB and HSP90. Error bars indicate standard deviation. (B) Analysis of neural stem cell gene upregulation in 1205Lu MBs using quantitative reverse transcription real time PCR, as normalized to GAPDH expression. Errors bars represent the standard deviation from triplicate samples. (C) Western blot analysis after switching 1205Lu monolayer cells to hESC conditions. (D) Immuofluorescence detection of neural stem cell markers in 1205Lu MBs and monolayer cultures. (E) Dual staining of MBs with TUJ1 and SOX2 antibodies. (F) Heterotypic MBs formed with 1205Lu and GFP-labeled HeLa cells were transferred to standard media, resulting in HeLa cells with neural features migrating from MBs.

Figure 5. Staining of human melanoma tissue microarray samples with anti-ID4.

Figure 6. Knockdown of ID4 or SOX2 inhibit MB formation.

(A) Diagram of experimental design. (B) Effectiveness of siRNAs in knockdown of ID4, SOX2 and NOTCH1 in hESC media. Western blots were carried out using LIN28 siRNA as a negative control. (C) The 1205Lu cells were treated with the indicated siRNAs, followed by challenge with hESC medium. MB formation was monitored. (D) The1205Lu cells were treated with individual ID4 siRNA, followed by challenge with hESC medium. Cell morphology was monitored after 7 days of culture in hESC media. (E) Western blots were carried out to analyze the effectiveness of individual ID4 siRNA. Lanes: mix, a mix of three siRNAs targeting ID4; con, untreated; 1,2,3, independent siRNAs targeting ID4.

Figure 7. Analysis of defects in MB formation.

(A) Diagram of experimental design. Nuclear DsRed-labeled 1205Lu cells were treated with siRNAs, mixed with untreated GFP-labeled 1205Lu cells, and plated under hESC conditions. (B) Imaging of MB formation. Merged image shows readout of red-green mixed (yellow dashed circle) versus green-only (green dashed circle) MBs. In the phase contrast image, the MBs corresponding to the merged images are indicated (white dashed circle). Right, FACS analysis was used to independently monitor cell growth or survival of green and red cells in hESC medium.

Figure 8. Knockdown of ID4 promotes differentiation of 1205Lu cells.

(A) Diagram of experimental design. (B) Representative images showing differentiation phenotype of 1205Lu cells after ID4 siRNA knockdown, hESC media challenge (five days), and switching to 254 medium (96 hours). (C) Western blot testing of ID4 knockdown using shRNAs delivered by lentivectors. (D) Representative images showing differentiation phenotype of 1205Lu cells after ID4 shRNA knockdown and hESC media challenge. (E) Representative images showing differentiation phenotype after ID4 shRNA delivery to 1205Lu cultures without exposure to hESC medium. Phenotype was enhanced after 96 hours in 254 medium.

Figure 9. Representative images showing differentiation phenotype of 1205Lu cells after ID4 shRNA delivery, followed by growth in 254 medium for 96 hours.

Figure 10. Model for role of ID4 as a differentiation guardian.

Model proposes that melanoma cells can undergo phenotype-switching to acquire stem-like cell features. Growth in hESC media can synchronize this behavior. The proposed role of ID4 as part of the stem cell-like transcriptional signature is to act as a “differentiation guardian” to prevent progression of cancer cells to a more normal differentiated state.