Growth factor independence 1 expression in myeloma cells enhances their growth, survival, and osteoclastogenesis

Bone marrow (BM) aspirates were collected in heparin from 9 healthy subjects and 36 patients with plasma cell disorders (Additional file 1 Tables S1 and S2). Non-adherent marrow mononuclear cells were collected, and highly purified MM cells (> 90%) were isolated by magnetic cell fractionation with anti-CD138 MicroBeads (Miltenyi Biotec Inc., San Diego, CA) [18] as previously described [19, 20]. The remaining adherent cells were cultured for 21 days with media changes every 4 days to obtain BMSCs that were used at passages 2 and 3.

Myeloma cell lines were purchased from ATCC (Manassas, VA) (MM.1S; H929) and from Leibniz-Institut DSMZ (Braunschweig, Germany) (MOLP-8) or generously provided by Drs. Louis Stancato (U266), Kenneth Anderson (KMS-11), and Nicola Giuliani (JJN3). HEK293-T cell line was purchased from ATCC (Manassas, VA), and SAKA-T cell line was generated by our group [21].

Fox Chase Beige SCID female mice (4–6 weeks of age) (Charles Rivers, Indianapolis, IN) were inoculated intratibially (IT) with 10⁵ MM.1S cells stably transduced with empty vector (EV) or overexpressing Gfi1 (Gfi1 o/e) in 20 μl of PBS. Mice were maintained and handled in accordance with the Guide for the Care and Use of Laboratory Animals on a protocol approved by the Indiana University IACUC. The animals were followed for 8 weeks before euthanasia due to large tumor development. X-ray images of dissected tibias were acquired on a viva CT 40 scanner (Scanco Medical) at a resolution of 21 m isotropic, reconstructed, and segmented for 3D display using the instruments analysis algorithm software (Sanco Medical Evaluation Program V6).

Human plasma Igλ concentrations were measured after 5 weeks post-IT injection to define successful myeloma engraftment and at the end of the study to evaluate tumor burden using commercially available ELISA kits, according to manufacturer’s instructions (Bethyl Laboratories, Inc., Montgomery, TX, USA).

The hind limbs were fixed with formalin and decalcified in 10% EDTA for 2 weeks. The tissues were processed as previously described [22], and the sections stained with hematoxylin and eosin, and tartrate-resistant alkaline phosphatase (TRAP) (Sigma-Aldrich, St. Louis, MO, USA). The sections were scored on a Leica DM LB compound microscope outfitted with a Q-Imaging Micropublisher Cooled CCD color digital camera (W. Nuhsbaum Inc., McHenry, IL, USA).

Mouse bone marrow cells were flushed from the long bones of 3–5-month-old mice, and non-adherent cells were collected and incubated in αMEM supplemented with M-CSF (10 ng/ml) for 48–72 h to generate bone marrow monocytes as previously described [23]. For myeloma-osteoclast co-cultures, M-CSF-generated bone marrow monocytes were plated into 96-well culture plates (1 × 10⁵ cells/well) in αMEM supplemented with RANKL (50 ng/ml) and MM.1S EV, and MM.1S Gfi1 o/e myeloma cells (5000 cells/well) were added 6 h later. The cultures were continued for 4 days and then fixed in 4% paraformaldehyde, washed with PBS, and stained for TRAP using a leucocyte acid phosphatase staining kit (Sigma-Aldrich, St. Louis, MO, USA). TRAP-positive cells with three or more nuclei were scored as osteoclasts by counting with an Olympus CKX41 inverted microscope using a × 10 objective.

Statistical analyses were performed using Prism software (Irvine, CA). The differences between groups were compared using a two-tailed unpaired Student t test or ANOVA. Statistically significant difference was set at p < 0.05, and results are expressed as mean ± SEM. Representative data from at least three biologic replicates are shown.

The specific details for the other experimental methods and procedures employed (chemicals and antibodies; plasmid and lentiviral constructs; immunoprecipitation and Western blotting; real-time RT-PCR (qPCR); ChIP assay and immunofluorescence) are listed in Additional file 2 (available on the journal website).

Gfi1 mRNA levels were significantly increased in human CD138+ cells from MM patients compared with normal donors (Additional file 1: Table S1) (Fig. 1a). We then compared Gfi1 mRNA levels (Additional file 1: Table S2) in different stages of the disease and found that they were increased in relapsed MM patients compared with MGUS patients and newly diagnosed MM patients (Fig. 1b) suggesting that Gfi1 levels correlate with disease progression. Gfi1 protein levels were also significantly higher in CD138+ cells from MM patients and MM cell lines compared with normal donors (Fig. 1c, d).

We next examined the effects of modulating Gfi1 levels in human MM cell lines that expressed wild-type (wt), mutant, or haploinsufficient p53. Knockdown (KD) of Gfi1, using two different shRNA (#1 and #2) (Fig. 2a), significantly increased the expression of the Bcl-2 family pro-apoptotic genes BAX, PUMA, and NOXA (Fig. 2b, c) in H929 cells (p53-wt) and significantly decreased cell growth and viability after 24 and 72 h, compared to scrambled shRNA-transduced control cells (Fig. 2d). The decreased viability of Gfi1-KD in MM cells (shRNA#1) resulted from enhanced apoptosis, as shown by increased caspase 3 activation and DNA fragmentation (Fig. 2e, f) and increased levels of sub-G0 cell cycle fraction when compared to scrambled control shRNA (Additional file 3: Figure S1A). Importantly, Gfi1-KD (shRNA#1) also induced cell death in MM cells with altered p53. Gfi1-KD increased cleaved Mcl-1 and caspase 3 levels in JJN3 (p53-haploinsufficient) and RPMI-8266 (p53-mutant) MM cells (Additional file 3: Figure S1B).

We then determined the effects of overexpression (o/e) of Gfi1 in MM cells (Fig. 3a). MM.1S Gfi1 o/e enhanced proliferation of MM cells compared with MM.1S cells transduced with empty vector (EV), as shown by a significantly decreased CellTrace staining after 72 h (Fig. 3b). Consistent with these results, MM.1S Gfi1 o/e cells displayed enhanced mitosis with a significantly increased percentage of cells in “G2+M” phase compared with MM.1S EV cells (Fig. 3c), while “S” phase levels were only slightly higher in these cells. The enhanced viability of MM cells overexpressing Gfi1 was not restricted to p53-wt cells. MTT assays showed that Gfi1 overexpression in JJN3 cells increased their metabolic activity (Additional file 4: Figure S2A). Importantly, Gfi1 o/e partially protected MM cells from bortezomib (Btz)-induced apoptosis. Btz treatment dose-dependently decreased MM.1S EV cell viability after 24 and 48 h (Fig. 3d), while Gfi1 o/e significantly enhanced MM.1S viability (Fig. 3d) and significantly decreased caspase 3 activation (Fig. 3e). This effect was also observed in Gfi1 o/e JJN3 cells, which had significantly enhanced viability when exposed to Btz compared with EV controls (Additional file 4: Figure S2B). However, Gfi1 o/e did not protect MM cells from treatment with dexamethasone (data not shown). These results suggest that Gfi1 plays a key role in MM cell survival and contributes to proteasome inhibitor resistance.

Since Gfi1 binds p53 [12] and KD of Gfi1 in p53-wt MM cells increased expression of Bcl-2 family proteins, we examined the contribution of p53-Gfi1 binding to the survival of p53-replete MM cells. Gfi1 acts as an epigenetic regulator of gene transcription by recruiting histone deacetylases (HDACs) to promoters of target genes [24]. Since both Gfi1 and p53 undergo post-translational modification, with p53 acetylation activating p53-mediated transcription [25, 26], we determined the role of Gfi1 and p53 acetylation in MM cell survival. Treatment of HEK293-T cells, transfected with plasmids carrying the mouse Gfi1 cDNA, with TSA (Zn²⁺-dependent HDAC inhibitor (HDACi)) and NAM (NAD+-dependent HDACi) revealed acetylation of the lysine residues of Gfi1 (Fig. 4a, top). TSA plus NAM treatment also resulted in increased acetylated Gfi1 in MM.1S myeloma cells (Fig. 4a, bottom). Moreover, this treatment induced high levels of p53 acetylation in H929 cells (Fig. 4b), as did nanomolar concentrations of actinomycin D (Fig. 4c), a known inducer of p53 acetylation at low doses [27]. p53 acetylation also increased the levels of p53 target proteins (BAX, PUMA) (Fig. 4b, c). Because p53 acetylation increased p53 activity, we assessed the ability of p53 to bind to the NOXA and BAX promoters. Treatment with HDACi significantly increased the relative enrichment of p53 at these promoters (Fig. 4d).

Since both Gfi1 and p53 can be acetylated in MM cells and their acetylation induced activation of pro-apoptotic genes and decreased viability (Fig. 4d, e), we next assessed if acetylated Gfi1 binds p53. HDACi treatment markedly decreased p53 binding to Gfi1 in MM.1S (Fig. 4a, bottom) and H929 cells (Fig. 4f). Further, immunofluorescence and cell fractionation studies showed that HDACi treatment shifts their cellular distribution. Gfi1 and p53 were co-localized in the cytosol of H929 cells (Fig. 4g), with increased amounts of these proteins in the cytosolic versus nuclear fractions (Fig. 4h). HDACi treatment increased acetylated p53 levels in the nucleus but did not increase nuclear Gfi1 (Fig. 4g, h).

To determine if acetylation of Gfi1, p53, or both was responsible for the enhanced apoptosis seen with HDACi treatment, we analyzed a set of Myc-tagged mGfi truncation constructs for the presence of acetylated lysines. We found several contained acetylated lysines (Fig. 4i), with the most intense bands in an overlapping region from amino acid 291 to 300 that contained K292 in the second zinc finger. We then mutated lysine 292 to arginine (K292R) of Gfi1 and transfected wt Gfi1 and the K292R mutant into HEK293-T cells. Wt Gfi1 co-immunoprecipitated with endogenous p53, but this binding was lost when the cells were exposed to acetylating conditions that strongly acetylated the lysine residues (Fig. 4j). Transfection of HEK293-T cells with the Gfi1 K292R mutant (which will block acetylation of the critical K292 residue in Gfi1, but should not affect p53 acetylation) resulted in decreased p53 binding in untreated cells, but no further loss of p53 interaction was observed in the presence of acetylating conditions (Fig. 4j). This result may suggest that Gfi1-K292 is part of the interaction region for p53, which would also explain why acetylation of K292 could make the complex dissociate. These results demonstrate that Gfi1 acetylation decreases Gfi1 interaction with p53 in human MM cells and that Gfi1-p53 binding prevents p53 binding to the promoters of its pro-apoptotic target genes. Therefore, the acetylation status of Gfi1 appears to be a determining factor contributing to Gfi1’s effects on MM cell survival.

Gfi1 levels are elevated in early B cell development and decrease in mature B cells [8], but how Gfi1 expression is regulated in MM cells is unknown. Since multiple factors increase MM cell survival, growth, and chemoresistance [28], including adhesive interactions between MM cells and BMSC, IL-6, TNFα, and sphingosine-1-phosphate (S1P), we tested their effects on Gfi1 expression in MM cells. Adhesive interactions between MM cells and BMSC increased Gfi1 expression 1.5-fold in MM cells, at both the transcriptional and protein level (Fig. 5a) as did IL-6 (Fig. 5b). S1P and TNFα had variable effects on Gfi1 levels in MM cell lines (Fig. 5b). Both IL-6 treatment and MM-BMSC adhesive interactions significantly enhanced Gfi1 protein levels, which were associated with increased levels of the pro-survival Mcl-1 protein levels in MM cells (Fig. 5a, right panel; c; d). Since Mcl-1 is a direct and functional target gene of Gfi1 in p210BCR/ABL-transformed cells [29] and plays an important role in cell proliferation and survival, we tested if Gfi1 and Mcl-1 protein expression levels were correlated in MM cells. Mcl-1 mRNA levels were significantly increased in MM.1S Gfi1 o/e cells compared with control MM.1S EV cells (Fig. 6a). Further, adhesive interactions with BMSC (SAKA-T normal human BMSC cell line) enhanced Gfi1 mRNA expression in H929 MM cells as early as 4 h, and this induction positively correlated with the increased Mcl-1 expression (Fig. 6b). Importantly, Gfi1 protein levels in MM cell lines and primary CD138+ MM cells significantly and highly correlated with Mcl-1 protein expression (Fig. 6c). Thus, BM microenvironmental factors known to sustain MM cell growth and survival also regulate Gfi1.

We next determined the effects of Gfi1 overexpression in MM cells in vivo. Consistent with prior studies, 80% of evaluable animals injected intratibially with either MM.1S EV cells (6/7) or MM.1S Gfi1 o/e cells (7/9) successful engrafted MM cells, with detectable human Igλ levels in plasma as an estimate of tumor burden. Plasma Igλ concentrations at sacrifice were higher in MM.1S Gfi1 o/e-injected mice as compared with MM.1S EV-injected mice, although this enhanced MM growth did not reach statistical significance (Fig. 7a). Histologic analysis of tumor burden in MM bearing tibiae was difficult to assess due to extensive cortical bone destruction in bones injected with the MM.1S Gfi1 o/e cells as compared with MM.1S EV-injected mice (data not shown). The trend towards more aggressive tumors in the bone marrow of mice bearing Gfi1 o/e MM.1S as compared with EV controls might be due to the modulation of the c-Myc oncogene by the Gfi1 levels (Additional file 5: Figure S3A), Interestingly, examination of tibial X-rays of animals displaying similar Igλ showed that MM.1S Gfi1 o/e cells caused greater and more extensive cortical and trabecular bone destruction than MM.1S EV cells (Fig. 7b). μCT analysis confirmed the significantly greater bone loss in the injected tibiae of MM.1S Gfi1 o/e-bearing mice (Fig. 7c) with a significant decrease in bone volume relative to the volume of calcified tissue (BV/TV) and trabecular number (Tb.N.) (Fig. 7d). Importantly, histologic analysis of tartrate-resistant acid phosphatase (TRAP)-stained sections showed increased osteoclast (OCL) numbers in the Gfi1 o/e MM.1S-bearing tibias compared with EV MM.1S-bearing tibias (Fig. 7e, histology). Analysis of OCL surface area (Oc.S/BS) demonstrated a marked increase in the Gfi1 o/e MM.1S-injected tibias that did not reach statistical significance (Fig. 7e, bar graph), again reflecting the extreme bone loss seen in these animals. Interestingly, the histologic analysis with TRAP staining demonstrated that the Gfi1 o/e MM.1S-injected tibias had larger OCL that contained more nuclei/OCL (Fig. 7e, histology). We then determined if increased Gfi1 expression in the MM cells increased OCL formation by culturing purified mouse OCL precursors with Gfi1 o/e and EV MM.1S cells for 72 h. TRAP staining of the cultures showed significantly increased OCL numbers (p < 0.05) in co-cultures containing Gfi1-overexpressing MM cells (Fig. 7f). Further, OCL formed in co-cultures of Gfi1 o/e MM.1S were larger and contained more nuclei/OCL than OCL formed in EV MM.1S control cell co-cultures. Consistent with this findings, in preliminary studies, we found that MM.1S Gfi1 o/e cells produce higher protein levels of IL-6 and integrin α4 as well as mRNA levels of RANKL and IL-6 and secrete higher levels of MIP-1α than MM.1S EV controls (Additional file 5: Figure S3 B, C, and D). These results support that Gfi1 o/e in MM cells enhances OCL precursor fusion and OCL formation.