|Home | About | Journals | Submit | Contact Us | Français|
This article has been retractedRetraction in:
Identifying the functions of proteins, which associate with specific subnuclear structures, is critical to understanding eukaryotic nuclear dynamics. Sp100 is a prototypical protein of ND10/PML nuclear bodies, which colocalizes with Daxx and the proto-oncogenic PML. Sp100 isoforms contain SAND, PHD, Bromo and HMG domains and are highly sumoylated, all characteristics suggestive of a role in chromatin-mediated gene regulation. A role for Sp100 in oncogenesis has not previously been defined. Using selective Sp100 isoform-knockdown approaches, we show that normal human diploid fibroblasts with reduced Sp100 levels rapidly senesce. Subsequently, small rapidly dividing Sp100 minus cells emerge from the senescing fibroblasts and are found to be highly tumorigenic in nude mice. The derivation of these tumorigenic cells from the parental fibroblasts is confirmed by micro-satellite analysis. The small rapidly dividing Sp100 minus cells now also lack ND10/PML bodies, exhibit genomic instability and p53 cytoplasmic sequestration. They also have activated MYC, RAS and TERT pathways and express mesenchymal to epithelial (MET) trans-differentiation markers. Reintroduction of expression of only the Sp100A isoform is sufficient to maintain senescence and to inhibit emergence of the highly tumorigenic cells. Global transcriptome studies, quantitative PCR and protein studies as well as immunolocalization studies during the course of the transformation reveal that a transient expression of stem cell markers precedes the malignant transformation. These results identify a role for Sp100 as a tumor suppressor in addition to its role in maintaining ND10/PML bodies and in the epigenetic regulation of gene expression.
The development of solid tumors is a multi-step process in which cells acquire increasingly malignant characteristics and typically involves the activation of oncogenes and the inactivation of tumor suppressor and pro-apoptotic genes, often by epigenetic silencing effectors (1). These progressive steps are regulated in part by an abnormal tissue microenvironment (2), but, several molecular checkpoints must also be altered to generate a tumor from a normal cell (3). Normal cells typically divide for a highly restricted number of generations, then enter a state of replicative senescence that can be characterized by acidic β-galactosidase staining (reviewed in (4). Normal human fibroblasts immortalize only at very low frequency and basically seem to be resistant to spontaneous transformation. This said, senescent human fibroblasts have been demonstrated to have the capacity to both stimulate the proliferation of premalignant and malignant epithelial cells in culture, and to increase the tumorigenicity of premalignant epithelial cells in mouse xenografts (5, 6). Thus, senescent human fibroblasts can create a tissue microenvironment that promotes multiple stages of tumor evolution through the senescence-associated secretory phenotype or SASP. This phenotype develops slowly and only after DNA damage, when senescent fibroblasts begin secreting IL-6, IL-8 and other cytokines that promote transformation of epithelial cells (7).
The constitutive proteins in ND10 or PML bodies (8, 9) are PML, Sp100 and Daxx (10), which play critical roles in oncogenic transformation (11–13), interferon-mediated viral resistance (14, 15) and response to PML-directed cancer therapeutics(16). They have also been shown to play a role in the onset of cellular senescence (17–19). PML triggers senescence when overexpressed in human diploid fibroblasts. Both PML and Sp100 are upregulated by interferon (IFN) and are part of an intrinsic anti-viral defense mechanism (14, 15, 20, 21).
Sp100 is a single-copy gene located on human chromosome 2q37 (22), and produces four alternatively spliced Sp100 isoforms. All isoforms share the 476 N-terminal amino acids with the most abundant isoform - Sp100A aberrantly migrating at 100 kDa (23). The three minor isoforms contain a SAND domain, so named after Sp100, AIRE, NucP41/75 and DEAF1, transcriptional regulators that bind to DNA. A highly conserved tryptophan occurs at the DNA binding interface of the SAND domains of each of these proteins. An Sp100B variant, constructed with a mutation of this tryptophan (W655Q), loses its ability to repress transcription of HSV-1 virus (14) as well as IFN-dependent suppression of viral gene expression (15).
Here we report that suppression of expression of all 4 Sp100 isoforms leads to immortalization of normal human neonatal fibroblasts. This immortalization is is preceded by the transient expression of embryonic stem cell (ESC) markers, is accompanied by a mesenchymal to epithelial trans-differentiation event (MET) and results in an acquired capacity to form tumors in nude mice.
Human embryonic kidney 293 (HEK293), 293T and human diploid foreskin fibroblast (BJ cells obtained from ATTC at population doubling of 23) were maintained in Dulbecco’s modified Eagle medium (DMEM) supplemented with 10% FCS and 1% antibiotics. All cells were grown at 37°C in a humidified 5% CO2 atmosphere. Certain cells cultures were treated with 1000 U/ml of IFNβ for 18 h as indicated.
For lentiviral pLKO.1 vector-based knock-down, A11 (clones TRCN000019224), B1 (clones TRCN000019226) and A7 (clone TRCN000019227) were purchased from Open BioSystems (Huntsville, AL). To create S2 knock-down of Sp100 we used a previously characterized construct based on the pLKO.1 vector (24). The protocol followed was described previously (15). Passage number after transduction with lentivirus is used since the cell numbers varied considerably per confluent culture dish specifically during the onset of senescence. Population doubling time is therefore substantially higher for the BJsm cells than BJV or BJ-S cells.
The protocol followed was described previously (15), with minor modifications. Cells were analyzed using a Leica confocal laser-scanning microscope. ND10 bodies were visualized using the following antibodies: monoclonal antibody (mAb) PG-M3 recognizes PML (Santa Cruz Biotechnology, Santa Cruz, CA) (1:500 dilution), rabbit antibody AB1380 against Sp100 (Chemicon International, Temecula, CA) (1:1000 dilution).
BJ cells or 293T cells were cultured in DMEM plus 10% FCS and were transfected with plasmids using the Lipofectamine 2000 (Invitrogen) reagent according to the manufacture’s protocol. In vitro luciferase assays were performed as described previously (15). The GFP-tagged Sp100 constructs were previously described (14).
Total cellular protein extract was prepared as described previously (15). The following antibodies were applied: rabbit anti-Sp100 antibody AB1380 (1:30,000 dilution), rabbit anti-Daxx antibody 2133 (1:300 dilution) (25), mAb anti-GFP B-2 (1:2,000 dilution) (Santa Cruz Biotechnology, Santa Cruz, CA), and mAb anti-tubulin (1:10,000 dilution) (Sigma, St Louis, MO).
Total cellular RNA was prepared using Trizol reagent (Life Technologies, Rockville, MD) as described previously (15) PCR products were run on 2% agarose gels. For quantitative real-time-PCR (qRT-PCR) analysis, 2.5 μg of total RNA were used to synthesize cDNA using the Superscript First-Strand Synthesis System for RT-PCR (Invitrogene) according to the manufacturer’s protocol. 40-cycle qRT-PCR products from 50 ng cDNA and 250 nM each of gene-specific primers (see Supplementary Information, Table S1) were detected using SYBR green and standard thermal cycler conditions in 7900HT Fast Real Time PCR System (ABI). Each sample was run in triplicate on 384-well plate. Endogenous GAPDH expression was used for internal normalization.
Genomic DNA was prepared from BJ, BJsm and 293T cells using DNeasy Blood and Tissue kit (QIAGEN). Genotyping was performed on genomic DNA using the Identity Mapping Kit (Cornell) according to the manufacture’s protocol. Briefly, 60 ng of genomic DNA was used per PCR reaction with a set of five highly polymorphic tetra-nucleotide microsatellites, FES/FPS, vWA31, D22S417, D10S526, and D5S592 for genotype determination. Amplified products were analyzed for fragment size on an Applied Biosystems 3130xl Genetic Analyzer.
Reverse transcription was conducted using TaqManR® microRNA reverse transcription kit (ABI) with 100 ng total RNA and Human Multiplex RT primer pool v1.0 in 15 ml reaction volume. qRT-PCR reactions with 3-fold diluted cDNAs were assembled in 384-well plates in triplicate and assayed in the ABI 7900HT Fast Real-Time PCR System. The individual assay IDs were 002307 for hsa-let-7a and 002405 for hsa-let-7c, respectively. miRNA expression was normalized by small nucleolar RNAs (RNU44 and RNU48) measured in each individual sample using individual control TaqManR® assays, and fold changes were calculated using the delta-delta Ct method (26).
We performed microarray analysis of mRNAs expression in BJV, collected at p15; BJ-S, collected at p11; and BJsm cells, collected at p57, when no cells with BJ-S morphology could be found in the population of BJsm cells. In addition samples were treated for 24 h with 1000 μ/ml IFNβ. IFNβ was included as it allowed the isolation of Sp100 dependent gene expression. A constant amount (400 ng) of total RNA was amplified, as recommended by Illumina. Samples were hybridized to the Illumina WG-6v2 human whole genome bead arrays. Illumina BeadStudio v.3.0 software was used to export expression levels and to determine p-values for each probe for each sample. Arrays were normalized between each other and filtered to remove non-informative probes. A probe was called non-informative if it had detection p-value>0.05 in all samples or if maximum ratio between expression values was not at least 1.2 between at least two samples.
Data for 18 samples (6 groups, 3 replicates) was tested for differentially expressed genes using one-way ANOVA analysis with significance set to p-value<0.05. False Discovery Rate was calculated according to Storey JD (27). Significantly differentially expressed genes as determined by ANOVA analysis were clustered with the k-means algorithm using from 2 to 25 clusters as input parameters, taking the median value of 3 replicates as an expression for a specific gene in any condition. Visual inspection was used to identify the maximum number of clusters that represent variability of the data. The heat maps were generated using 2-way hierarchical clustering using Euclidean distance to cluster samples/conditions and Spearman correlation distance to cluster genes. Pathway analysis was carried out using Ingenuity Pathways Analysis software, using Ingenuity Core Analysis (IPA 6.0, IngenuityR® Systems) with Benjamini-Hochberg multiple testing corrected p-value<0.01 as a significance threshold. Fisher exact test was used to test significance of an enrichment of genes with specific annotation (transcription factor target, biological process, function, etc.) in a list with significance threshold set to p-value<0.05.
Our microarray profiling data can be found at the Gene Expression Omnibus (GEO) database under accession GSE20613.
ShRNA-directed Sp100 suppression was used to explore the role of Sp100 and its isoforms in human diploid fibroblast growth. To target the various isoforms, human diploid fibroblasts (BJ cells) were transduced with two pools of lentivirus expressing shRNA. Viruses A7 and S2 targeted all isoforms; viruses A11 and B1 targeted only the SAND-domain-containing isoforms (28) (Fig. 1A). Control BJ cells were transduced with empty vector and named BJV cells. Transduced cells were selected with 0.5 μg/ml puromycin for 3 days followed by serial passageing. Suppression of Sp100 isoform expression was confirmed by quantitative RT-PCR (Fig. 1B). BJ cells transduced to suppress only SAND-containing Sp100 isoforms are designated BJ-SAND cells. As expected expression of the Sp100A isoform, which lacks a SAND domain, is unaffected in the BJ-SAND cells and retains its sumoylated modifications and its interferon inducability (Fig. 1C, lane 7 and 8). All isoforms are suppressed in BJ-S cells (Fig. 1C, lane 3 and 4), a suppression that cannot be overcome by interferon activation.
The vector-transduced BJV cells continued to actively proliferate and did not senesce within the time fame investigated, in agreement with previous observations (29). By contrast, both BJ-S and BJ-SAND cells proliferated slowly after shRNA transduction and showed enhanced senescence associated (SA)-β-galactosidase staining between passages 10–13 post-transduction. Cultures at this point were comprised of large flat cells. The BJ-SAND cells became multinucleated, showed intense staining and stopped proliferation at passage 16 post-transduction (Fig. 2C and F).
During the onset of senescence, which occurs between passages 11 to 15 post-transduction, smaller rapidly growing cells emerge (designated BJ small or BJsm) (Fig. 1D). These small cells are initially tightly associated with dendritic extensions of the large flattened fibroblasts, (Fig. 1D, BJ-S/sm p13, see also Supplemental Fig. S1). This emergence was followed by the development of clones with a cobble stone-like morphology and strong self-adherence (Fig. 1D, BJsm p15 and Fig. 2B). As determined by z-axis confocal microscopy BJsm cells have approximately five times the height as the flat BJV and BJ-S cells (Supplemental Fig. S2). These BJsm cells arise multi-focally in each culture and rapidly outgrow the BJ-S cells within 3–4 passages. Figure 1E compares the growth rate of BJV, BJsm cells with HEK293 cells, a fast growing cell line maintained in the laboratory, which also lacks Sp100. These HEK293 cells do not adhere to each other. BJsm cells grow substantially faster than parental BJ fibroblasts and faster than HEK293 cells and they do not express senescence-associated SA-β-galactosidase (Fig. 2B). These apparently immortalized cells arose in multiple, independently derived cultures of BJ-S cells, and have gone through more than 100 passages. BJsm cells lack all mRNA Sp100 isoforms, as determined by qRT-PCR (Fig. 1B), and exhibit a significant reduction in PML and Daxx proteins (Fig. 1C, lane 5 and 6). Concomitantly, ND10/PML bodies are reduced from >20 per nucleus in BJ-S cells to approximately one per nucleus in BJsm cells (Fig. 2E). These results strongly suggest that loss of Sp100 leads to rapid immortalization and morphologic transformation of primary human fibroblasts.
Previous observations had shown that elevated expression of the minor PML IV isoform induces senescence (30). Using qPCR analysis we found that expression of PML IV mRNA increased substantially in BJ-SAND as one might have expected but surprisingly it is also increased in the fast growing BJsm cells. However, no apparent increase of PML protein was detected in BJ-SAND cells using Western blots (Fig 1B). We therefore measured the relative expression level of the major PML I isoform by qRT-PCR analysis in the BJ, BJ-S, BJsm and BJ-SAND cell lines. We found that PML I transcript levels are equivalent to levels in all cell lines except the transformed BJsm where it is substantially increased (4 fold) (Supplemental information Figure S6). The apparently normal PML I expression in BJ, BJ-S and BJ-SAND is in contrast to high expression of PML IV mRNA in senescing BJ-SAND (Fig 1B), demonstrating that a change in the PML splicing pattern has occurred. The increased expression of PML IV and PML I transcripts in BJsm cells should have no consequence as all PML protein isoforms are not detected (Fig. 1B). The cause and consequence of the reduction of PML and Daxx proteins in BJsm cells can presently not be evaluated.
To exclude the possibility that immortalization and transformation are the result of lentivirus integration, we repeated the infection and serial passaging experiment several times. In 5/8 independent repeats we observed the appearance of BJsm cells but never observed immortalization of BJ cells transduced with vector only lentiviruses or with lentiviruses suppressing only SAND-domain containing Sp100 isoforms. Thus immortalization as a result of off-target or integration events is highly unlikely.
We also addressed the concern that the transformed BJsm cells may have been derived from some low level of contaminating cells in the original fibroblast cultures. We initially excluded the 293T cells, used to produce the lentiviruses, as a possible contaminant by testing BJsm cells for the presence of large T-antigen expressed in 293T cells. No T-antigen expression was detected in either the fibroblasts or the emerging transformed cell (Supplemental Fig. S3A). The identity of the emerging cells as being derived from the parental BJ fibroblasts and not from some contaminant was further confirmed by micro satellite comparative analysis of the parental BJ cells, the emergent transformed BJsm cells and the 293T cells used to prepare the shRNA vectors. We found a perfect identity of the micro-satellite markers detected in the fibroblast and transformed cells while the 293 cells exhibited an altered panel of markers. This further eliminates 293 cells as a source of the emerging transformed cells but more importantly confirms the derivation of the BJsm transformed cells from the parental BJ fibroblasts (Supplemental Fig. S3B).
The immortalized BJsm cells were also demonstrated to form spheres in low or serum free medium (Supplemental Figure S1). When such spheres were transferred back to medium with 10% FBS, all cells reformed monolayers, indicating that these cells have properties of tumor-spheres. We then determined whether the BJsm cells had the potential to form tumors in nude mice. Subcutaneous injection of 1×106 BJsm cells resulted in large tumors after 6 weeks, whereas, mice injected with BJ and BJV cells were tumor free even after 3 months. HeLa cells were used as a positive control (Fig. 1F). Histological examination of BJsm derived tumors showed tightly packed cells with a nucleus/cytoplasm ratio significantly higher than those derived from HeLa cells (Fig. 1F, lower panel). We found no evidences of differentiation into different cell types. Thus, while suppression of just the Sp100 isoforms with a SAND domain, leads to accelerated senescence of normal human fibroblasts, complete suppression of all Sp100 isoforms (including Sp100A) leads to the emergence of immortalized cells which can form tumors in nude mice.
We further examined the molecular basis of the immortalization and transformation in BJsm cells. A hallmark of dividing cells that bypass replicative senescence is genomic instability and DNA damage (31). DNA break recognition is known to activate the ATM kinase followed by γH2AX phosphorylation (32). Our analysis of γH2AX foci showed that most BJV cells had no γH2AX foci, a few had 2–5 foci and only a very few had over a hundred foci (a selected image, Fig. 2G, shows all three phenotypes). BJ-S cells also exhibited no more than five foci per cell (Fig. 2H). However, almost all BJsm cells were positive for a large number of γH2AX foci (Fig. 2I). Thus a high level of DNA breaks occur in BJsm cells suggesting pronounced genomic instability in cells lacking Sp100.
In previous studies, BJ cells have been transformed by activation of the MYC and RAS pathways and by suppression of p53 and RB (33) so we examined their mRNA expression levels in the BJsm cells. While MYC and RAS were found to be upregulated, surprisingly we found p53 expression to also be highly upregulated in BJsm cells. Exhaustive sequencing of p53 transcripts from BJsm cells did not find any inactivating mutations.
However, we also found that the p53 target gene p21CIP1 was not expressed (Fig. 3A), suggesting that p53 was likely functionally inactivate, perhaps due to cytoplasmic sequestration. To investigate this possibility, we immuno-stained BJV, BJ-S and BJsm cells with anti-p53 and anti-HSP70 antibodies. We found p53 aggregated with the highly induced HSP70 only in the cytoplasm of the BJsm cells. This suggests that although p53 transcription is high, a large part of wt p53 is inactivate due to its sequestration in the cytoplasm with HSP70 (Supplemental Fig. S4).
To further address the hypothesis that Sp100A is acting as a tumor suppressor in our system, we transfected BJsm cells with a vector expressing an RFP-tagged Sp100A and followed the growth of the transfected cells using fluorescence microscopy. We found that the RFP-expressing BJsm cells did not divide and rapidly disappeared from the culture. To determine, whether this phenomenon might be due to reactivation of the wild type p53, we measured the response of a p53 activated p21CIP1 luciferase reporter plasmid to the introduction of the RFP tagged Sp100A expression vector. We found that ectopic Sp100A expression enhanced activity of the p21CIP1 promoter while the isoforms with SAND domains suppressed p21CIP1 expression. To determine whether this suppression might be due to the DNA binding capacity of the SAND domain, we made a single amino acid mutation (W-Q) in the DNA binding site of the suppressive Sp100B isoform yielding the mutant Sp100BQ. Indeed, transfection of BJsm with the Sp100BQ mutant reversed the suppression. Moreover, the effect of SP100BQ was specific to the p21CIP1 promoter, as we did not see any effect on the PIG3 promoter driven reporter although both promoters were responsive to transfection of p53 (Fig 3C). Based on these results, we propose that the tumor-suppressive p53 functions are inactivated by its sequestration in the cytoplasm in the absence of the Sp100A isoform.
The different Sp100 isoforms can interact with each other and by shifting the isoform balance in favor of Sp100A by reducing the SAND domain containing isoforms leads to early senescence. We then examined the effects of changing the balance of the Sp100 isoforms by increasing the level of Sp100A in wild type BJ cells. Contrary to the BJsm cells BJ cells survive the introduced RFP-Sp100A expression. However, the induction of p21CIP1 is low to non-existant (Fig. 3D). Since normal fibroblasts have very little p53 (none detected by western blot; see Fig. 3A) we tested whether the lack or presence of very low amounts of p53 can account for these negative results. When we transfected BJ cells with p53 alone, we doubled the p21CIP1 reporter signal likely due to the residual amount of Sp100A present in the BJ fibroblasts. Co-transfection of p53 and Sp100A, however, substantially increased p21CIP1 promoter activity to 5 times the control level (Fig. 3D). This shows that Sp100A alone does not directly activate p21CIP1 but synergizes with p53 in activating the p21CIP1 promoter.
Another hallmark of cells that bypass replicative senescence is the re-expression of telomerase which counteracts the progressive shortening of telomeres (34). We find a robust induction of TERT mRNA in BJsm cells, consistent with the by-pass of senescence, but not in the BJV or BJ-S cells (Fig. 3B). Since the expression of TERT is regulated by MYC, we also tested MYC expression in all BJ cell lines. We found that the MYC gene was highly expressed in BJsm cells at both the mRNA and protein levels (Fig. 3A and B) suggesting that MYC expression is also deregulated in BJsm cells. Together these data show Sp100 knock-down results in the deregulation of p53 and MYC, which in turn contribute to the transformation of BJsm cells. To gain further insight into the underlying mechanism of the BJ-S to BJsm transformation, we compared gene expression patterns in BJV, BJ-S, and BJsm cells using Illumina microarrays. We also evaluated gene expression as a function of the interferon response of each of the 3 cell types as this function is often altered in transformed cells (35). We first focused on identifying transcription factors that were differentially expressed between the transformed BJsm cells and the BJV, BJ-S cells. We identified 261 transcription factors with significantly changed expression levels in BJsm cells in this comparison (Supplemental Fig. S5). Two of the factors LEF1 and E2F2, upregulated in BJsm cells support accelerated proliferation of cells and in addition, LEF1 targets MYC which is also up-regulated in these cells. Because of the important role suggested for MYC in this transformation event we also compared the expression of known MYC target genes in the 3 cell lines. We found that 53% of 333 core MYC regulated genes (36) were highly induced in BJsm cells (Supplement Fig. S5). An additional comparison with the more extensive MYC target gene database described by Chandriani et al. (37) revealed that 546 of 1934 MYC-regulated genes were alsoinduced in BJsm cells (Fig. 3D). The deregulation of many of these target genes compared to BJV cells, including expression of LIN28B a known suppressor of the let7-RAS pathway, was confirmed by RT-PCR (partially shown in Fig. 3B and F)(38). Since LIN28B ( or LIN28) has been shown to determine the level of mature let-7 in cancer cells (37) we used qRT-PCR to confirm the suppression of let-7a and let-7d levels in BJsm cells (Fig. 3F). One of the known targets of let-7 is KRAS (39). In agreement with suppression of let-7, we found the induction of KRAS at the protein (Fig. 3H) but not at the mRNA level in BJsm cells (Fig. 3G). An observation that is consistent with the removal of a translational block. This increase in KRAS protein was further supported by our microarray data which showed expression of 503 genes associated with an activated KRAS signature (40) also changed significantly in BJsm cells (Supplemental Fig. S5). KRAS is frequently activated through mutations that correlate with MGMT suppression (41). We found a reduction of MGMT mRNA levels in BJsm cells (Fig. 3G). Together these results suggest that in addition to p53 and MYC, the RAS oncogenic pathway is deregulated in BJsm cells.
Further analysis of the microarray data also revealed evidence for expression of embryonic stem cell markers in BJsm cells. We found that 325 genes reported to belong to the embryonic stem cell (ESC) expression program (42) were significantly upregulated in BJsm compared to BJ-S cells in our microarray studies. However, the canonical POU5F1, NANOG, SOX2, and KLF4 reprogramming factors (RFs) (43–45) were not found to be activated in these cells. To determine whether these RFs were only expressed transiently during the transformation process, we used semi-quantitative RT-PCR to examine expression of these genes in BJ-S cells at different passages spanning the process of their transformation to BJsm cells. Expression of POU5F1 and SOX2 mRNA was found to transiently increase at passages 13–15, with expression being suppressed again by passage 22. NANOG mRNA was transiently increased, only at passage 14–15, while KLF4 mRNA increased at the earliest tested, passage 11 (Fig. 4A). Expression of all RFs was undetectable by RT-PCR after passage 22 concordant with the microarray data. Significantly, the appearance of SOX2, POU5F1 and NANOG mRNA coincides with complete suppression of Sp100 mRNA.
We also confirmed the expression of individual RFs in BJ-S cells by immunofluorescence microscopy. Nuclear SOX2 protein staining appeared at passage 13 (Fig. 4C-H), in complete agreement with RT-PCR and Western blotting analysis (Fig. 4A and B), and increased with passage number. Importantly, SOX2 is expressed early in the large, flat BJ-S cells before overt transformation is evident (Fig. 4E). SOX2 expression was never observed in BJV cells expressing SP100. We observed similar expression patterns for POU5F1 and NANOG (data not shown). These results suggested that BJ cells transiently express markers consistent with an ESC-like state, before progressing to KRAS-mediated transformation.
We have indicated above that BJsm cells form a cobblestone pattern in culture, that they are tightly attached to each other and that they form aggressive tumors with epithelial morphology in vivo. The morphological changes observed during the BJ-S to BJsm transition suggest a mesenchymal-epithelial trans-differentiation event has occurred. Analysis of the microarray data focused on this possibility showed there was a suppression of transcription factors required for the mesenchymal phenotype: TWIST1, SNAI1, SNAI2, JUB (Ajuba) already evident in BJ-S cells, and that BJsm cells exhibited an additional suppression of the mesenychmal phenotype markers: CDH2 (N-cadherin), S100A4 (FSP1), VIM (vimentin), FN1 (fibronectin) and SERPINE1 (PAI-1). Induction of the expression of epithelial markers ACTA2 (aortic smooth muscle actin) and CDH1 (E-cadherin) (46) was confirmed in BJsm cells by RT-PCR. E-cadherin induction and the reduced expression of vimentin and fibronectin proteins was also confirmed by Western blot (Fig. 5A and B). Moreover, we found that BMP7 which functions as an inhibitor of epithelial-mesenchymal transition (EMT) was induced in BJsm cells (47). Two recent publications have suggested that MET is a critical initiating event during the derivation of induced pluripotent stem cells from mouse fibroblasts (48, 49).
In this study, we present evidence that Sp100 functions as a key regulator of cellular maintenance in that its expression in a normal human diploid fibroblast cell line promotes a program of senescence, thus preventing expression of programs for immortalization and malignant transformation. We also show that expression of the Sp100A isoform synergizes with p53 to activate the p21CIP1 promoter and that a shift in the balance between the SAND domain containing isoforms and Sp100A is sufficient to push BJ cells into a senescent state. It is also noteworthy that the Sp100A isoform, which is only known to contain the heterochromatin protein (HP1) binding domain and the Sp100 dimerization domain, is necessary and sufficient to maintain the normal program leading to senescence and thus suppresses transformation in BJ fibroblasts. This implies that chromatin association of Sp100A is required to maintain the normal senescence program by preventing entry into programs that support continued cell division and transformation. Failure to senesce as a result of the stochastic imbalance of the Sp100A isoform induced by shRNA suppression of all Sp100 isoforms may allow a new program to be activated. This new program apparently is facilitated by the transient expression of some stem cell markers and is reversible to some extent, as adding back Sp100A to transformed BJsm cells stops cell division resulting in the death of the transformed cells, likely through the Sp100A mobilization of the extensive amount of sequestered wild-type p53. Thus, it appears that Sp100 can function as a tumor suppressor in this system. Since both the SAND and the HMG domains of Sp100 have the potential for sequence specific DNA recognition (50, 51) and may bind to non-methylated CpG sequences in both viral and cellular DNA, the possibility is raised that direct DNA binding of the SAND domain containing Sp100 isoforms is involved in gene regulation modified by their potential heterodimerization with Sp100A. Regardless of the mode of recognition, it is likely that Sp100 proteins regulate endogenous gene expression as they do viral promoters (15), and thus may influence the development of the transformed phenotype. The short time frame required for the emergence of the malignant transformed cells following complete knockdown of Sp100, the high frequency of emergence within a few doubling times and the reversibility of the high replicative activity by reintroduction of Sp100A make it unlikely that transformation results from the accumulation of multiple heritable genetic mutations in the cell cultures. Whether loss of Sp100 alone can induce tumors in vivo is not known. However, homozygous deletions of Sp100 have been shown for mantel cell lymphoma (52); although whether this event is causative is not known.
In summary our data indicate that Sp100 null cells rapidly exhibit a transformed state characterized by activated KRAS, MYC and TERT pathways, cytoplasmic sequestered p53 and pronounced genomic instability. This transformation is preceded by the transient expression of an ESC-like transcription pattern and acquisition of an epithelial phenotype. We show that the loss of a expression of a single gene, Sp100, can reprogram the normal differentiation program of human fibroblasts and mimics the reported reprogramming of human fibroblasts by several externally introduced reprogramming factors (53).
We thank Dmitri Gurevitch for H&E staining of tumor samples, James Hayden for advice with microscopes, Calen Nichols for technical assistance with micro-array hybridization and Dr. R. Everett for reagents.
*Financial support: This study was supported by funds from NIH AI 41136, NIH GM 57599, NIH HD 34612 and the Mathers Foundation and the PA DOH Commonwealth Universal Research Enhancement Program, NIHCore grant CA 010815 is acknowledged for the support of the microscopy, genomics and sequencing facilities. This project is funded in part under a grant with the Pennsylvania Department of Health.
Disclosure of Potential Conflicts of Interest: No potential conflicts of interest were disclosed.