Effect of combining glucocorticoids with Compound A on glucocorticoid receptor responsiveness in lymphoid malignancies

Glucocorticoids (GCs) are a cornerstone in the treatment of lymphoid malignancies such as multiple myeloma (MM) and acute lymphoblastic leukemia (ALL). Yet, prolonged GC use is hampered by deleterious GC-related side effects and the emergence of GC resistance. To tackle and overcome these GC-related problems, the applicability of selective glucocorticoid receptor agonists and modulators was studied, in search of fewer side-effects and at least equal therapeutic efficacy as classic GCs. Compound A (CpdA) is a prototypical example of such a selective glucocorticoid receptor modulator and does not support GR-mediated transactivation. Here, we examined whether the combination of CpdA with the classic GC dexamethasone (Dex) may improve GC responsiveness of MM and ALL cell lines. We find that the combination of Dex and CpdA does not substantially enhance GC-mediated cell killing. In line, several apoptosis hallmarks, such as caspase 3/7 activity, PARP cleavage and the levels of cleaved-caspase 3 remain unchanged upon combining Dex with CpdA. Moreover, we monitor no additional inhibition of cell proliferation and the homologous downregulation of GR is not counteracted by the combination of Dex and CpdA. In addition, CpdA is unable to modulate Dex-liganded GR transactivation and transrepression, yet, Dex-mediated transrepression is also aberrant in these lymphoid cell lines. Together, transrepression-favoring compounds, alone or combined with GCs, do not seem a valid strategy in the treatment of lymphoid malignancies.

Dexamethasone (Dex) was purchased from Sigma Aldrich, dissolved in ethanol (EtOH) and stored at -20˚C. Compound A (CpdA) was synthesized as described by Louw et al. [26], dissolved in EtOH, flushed with an inert gas (N 2 -vapours), protected from light and was stored at -80˚C. Recombinant murine TNFα, obtained from the VIB protein service facility, was dissolved in cell culture medium and used at a final concentration of 2000IU/ml. The total solvent concentration in all experiments was kept equal in each condition.

RT-qPCR
Total RNA was isolated using an RNeasy mini kit (Qiagen), according to the manufacturer's instructions. The resulting RNA concentration was measured using a Biodrop (Isogen). Reverse transcription (RT) was performed using an iScript cDNA synthesis kit (Bio-Rad). The resulting cDNA was used as a template for quantitative PCR (qPCR) reactions using the Lightcycler 480 SYBR Green I Master mix (Roche diagnostics), following the manufacturer's protocol. The qPCR reaction protocol includes: a) activation enzyme and initial denaturation, 5' at 95˚C; b) 40 cycles of denaturation 15" at 95˚C, hybridization and elongation 45" at 60˚C, and was performed on a Lightcycler 480 system (384-well plate format, Roche diagnostics). The primer sequences are available in S1 Table. Each condition was performed in triplicate and the resulting Cq values were analyzed using qBasePlus (Biogazelle) and normalized to the reference genes: SDHA, RPL13A and YWHAZ [40,41]. Statistical analyses were performed on log transformed data.

Nucleus-cytoplasm fractionation
After washing the cells with ice-cold PBS, the cells were lysed in hypotonic buffer (20mM HEPES pH = 7.0, 20% glycerol, 10mM NaCl, 1.5mM MgCl 2 , 0.2mM EDTA, 0.1% TritonX100, supplemented with Halt protease and phosphatase inhibitor cocktail, EDTA-free (Thermo scientific)) and centrifuged for 10min at 4˚C and 80rcf, separating the cytoplasmic from the nuclear fraction. Next, the pelleted nuclei were resuspended in hypertonic buffer (hypotonic buffer supplemented with 500mM NaCl) to disrupt the nuclear membrane and rotated for 30min at 4˚C, followed by centrifugation for 5min at 4˚C and 21130rcf. From each fraction, the protein concentration was determined and maximally 10μg of total protein was further processed for WB analysis.

CellTiter-Glo and Caspase-Glo 3/7 assays
After 72h of treatment (see figure legends), cells were subjected to a CellTiter-Glo cell viability or a Caspase-Glo 3/7 assay (Promega). The CellTiter-Glo or Caspase-Glo 3/7 reagent was reconstituted by adding the CellTiter-Glo or Caspase_Glo 3/7 substrate to the CellTiter-Glo or Caspase_Glo 3/7 buffer, respectively, and was equilibrated at room temperature. A volume of CellTiter-Glo or Caspase_Glo 3/7 reagent equal to the volume in the well was added. The contents were mixed on an orbital shaker to induce cell lysis and the plate was incubated for 10' (CellTiter-Glo) or 1h (Caspase_Glo 3/7) to stabilize the luminescent signal. Luminescence was recorded using a Spectramax Paradigm spectrophotometer (Beckman Coulter) with SoftMax-Pro 6.1 software.

Cell proliferation assay
At the indicated time points, cells were stained with trypan blue (1:1 ratio with cells) and counted using the Countess automated cell counter (Invitrogen).

Statistical analyses
Results are presented as scatter, dot or bar plots, in which the mean +/-standard error of the mean (SEM) are depicted. When scatter dot plots are used, the open circles (o) represent the mean of the individual biological replicates. Statistical analyses were performed using Graph-Pad Prism 7 software. When the means of 2 groups were compared, a two-tailed unpaired ttest was used. When the means of 2 variables (e.g. induction and concentration) of more than 2 groups were compared, a two-way ANOVA with Tukey's or Sidak's multiple comparisons post-test was used. Results were designated significant when the P-value (P) < 0.05: Ã = P < 0.05, ÃÃ = P < 0.01, ÃÃÃ = P < 0.001, ÃÃÃÃ = P < 0.0001, ns = non-significant.

CpdA does not support nuclear accumulation of GR yet induces HSP70 gene expression
CpdA was shown to induce nuclear translocation of GR in different cell models, albeit to varying degrees [27,28,42]. Therefore, we first assayed nuclear accumulation of GR upon CpdA treatment in GC-sensitive MM1.S cells. We monitored the degree of cellular fractionation by determining PARP and tubulin levels, as controls for nuclear and cytoplasmic fractions, respectively ( Fig 1A). As a positive control, Dex is used as it strongly induces nuclear translocation of GR [43]. In contrast, Fig 1A shows that CpdA does not support a marked nuclear accumulation of GR.
Recently, Beck et al. reported that CpdA induces Hsp70 gene expression in A549 cells in a GR-dependent manner [44]. Hence, as a control for the activity of CpdA, we monitored the expression levels of HSPA1A, one of the genes coding for Hsp70 [45], in MM1.S after 6h treatment with CpdA. As shown in Fig 1B, CpdA strongly induces HSPA1A expression compared to solvent control.

Adding CpdA to Dex treatment does not substantially enhance GCmediated cell killing
Next, we determined GC responsiveness of the GC-sensitive MM1.S (MM) and C7-14 (ALL) cells and the GC-resistant MM1.R (MM) and C1-15 cells (ALL) [46,47]. We treated these cells for 72h with a concentration range (10 -4 M-10 -10 M) of Dex, CpdA or the combination hereof and measured cell viability using MTT assays. Fig 2A and 2B show that the cell viability of GCsensitive MM1.S and C7-14 cells decreases with Dex treatments in a concentration-responsive manner. As expected, GC-resistant MM1.R and C1-15 do not respond to Dex treatment ( Fig  2C and 2D). In addition, CpdA alone does not decrease the cell viability of GC-sensitive MM1. S and C7-14 cells (Fig 2A and 2B), except slightly (10-20%) at 10 -6 M in MM1.S and MM1.R cells and in all cell lines at the higher (10 -4 M and 10 -5 M) concentrations. This mild effect of In addition, this experiment was repeated with a fixed (lower) concentration of CpdA (10 -6 M) and varying concentrations (10 -5 M-10 -9 M) of Dex. Fig 2E shows that combined Dex and CpdA treatments leads to a minor reduction (on average 10%) of MM1.S cell viability at each concentration, yet, overall no statistical interaction can be shown between Dex and CpdA.

Dex and CpdA combination neither augments GC-induced apoptosis nor increases cell proliferation inhibition
We further zoomed in on GC-mediated apoptosis and evaluated whether 72h treatment with a limited concentration range (10 -6 M-10 -8 M) of Dex, CpdA or the combination of both could increase caspase 3/7 activity of GC-sensitive MM1.S and C7-14 cells. Fig 3A and 3B shows that Dex treatment elevates the caspase 3/7 activity in MM1.S and C7-14 cells with increasing concentrations. Yet, maximal caspase 3/7 activity in C7-14 cells is reached at lower concentration Dex (10 -7 M) compared to MM1.S cells (10 -6 M). CpdA treatment does not augment caspase 3/ 7 activity at any concentration in C7-14 cells, while in MM1.S cells this activity is slightly enhanced at 10 -6 M CpdA (Fig 3A and 3B). Equimolar Dex and CpdA combination also do not augment caspase 3/7 activity compared to Dex treatment in MM1.S or C7-14 cells. In addition, we repeated this experiment with a fixed concentration of CpdA (10 -5  Protein levels of other apoptosis hallmarks, such as PARP cleavage and cleaved-caspase 3, were also assayed following 72h treatment with Dex (10 -6 M or 5.10 -8 M), CpdA (5.10 -6 M) or a combination thereof. CpdA treatment alone does not change the cleaved-caspase 3 levels, and PARP cleavage is not induced in both MM1.S and C7-14 cells (Fig 3C and 3D). Cleaved-caspase 3 levels are the highest and PARP cleavage is the strongest with 10 -6 M Dex treatment and added CpdA does not change these levels (Fig 3C and 3D), possibly because a plateau is reached. In line with this hypothesis, combining 5.10 -8 M Dex with CpdA increases the levels of cleaved-caspase 3 compared to 5.10 -8 M Dex alone in both cell lines, yet, this is not the case for PARP cleavage (Fig 3C and 3D).
In addition, cell proliferation was assayed by treating MM1.S and C7-14 cells for 72h with Dex (10 -6 M), CpdA (5.10 -6 M) or both combined. As expected, the number of living cells increases in the control condition in function of time, indicative of proliferating cells (Fig 3E  and 3F). In contrast, Dex treatment blocks proliferation of the cells, which is most pronounced at 72h in C7-14 cells. CpdA does not affect cell proliferation as the number of living cells/ml is comparable to the control condition. Consistently, Fig 3E and 3F show that the combination of Dex and CpdA largely reflects the number of living cells/ml of Dex treatment alone.

GR protein levels are not sustained by combining Dex and CpdA
CpdA was previously shown to protect GR from homologous downregulation in various cells [16,39]. Therefore, we wondered whether the addition of CpdA to Dex treatment might sustain GR protein levels in MM1.S, C7-14 and C1-15 cells. To this end, cells were treated for 72h with a concentration range (10 -5 M-10 -9 M) of Dex, CpdA or a combination hereof (Fig 4A), and GR protein levels were determined. MM1.R cells are not taken along as they are GR-negative (S3A Fig). GR levels in GC-resistant C1-15 cells remain largely unaltered, regardless of any treatment (Fig 4A and 4B). In MM1.S and C7-14 cells, GR protein levels decrease by Dex treatment in a concentration-responsive manner, except in C7-14 cells at 10 -8 M Dex, where GR levels even increase compared to control. Although this 10 -8 M Dex-effect is already present after 24h, the increase in GR levels is most pronounced after 72h (S3B Fig). CpdA treatment alone does not affect GR protein levels, but again kills cells at 10 -5 M in MM1.S cells, as evident from the lacking GAPDH levels (Fig 4A). Equimolar concentrations of CpdA and Dex is unable to rescue GR from degradation in MM1.S and C7-14 cells, and GR levels largely resemble those of Dex treatment alone.
To exclude that the latter is due to the use of equimolar concentrations, we treated the cells with a fixed CpdA concentration on top of varying Dex concentrations (10 -6 M-10 -9 M). Fig 4B shows that using this treatment scheme, GR protein levels are also not protected from degradation and are strongly reduced in a concentration-dependent manner compared to control.

CpdA does not alter Dex-mediated transactivation of GRE-driven genes
Our group recently reported that CpdA can modulate Dex-instigated gene expression [29]. To test whether this also occurs in lymphoid cell lines, cells were treated for 6h with Dex, CpdA or the combination hereof, followed by QPCR analyses of GR and of its target genes GILZ and FKBP5. Both GILZ and FKBP5 are strongly induced upon Dex treatment in MM1.S and C7-14 cells (Fig 5) but also in C1-15 cells (S4A Fig), the latter regardless of the cells' resistance towards GC-mediated apoptosis. GR mRNA levels are unaffected by 6h Dex treatment in MM1.S cells, but are upregulated in C7-14 and C1-15 cells (Fig 5 and S4A Fig). At large, CpdA does not support transactivation of GRE-driven genes, except for a mild increase in GILZ expression in MM1.S cells (Fig 5A). Moreover, combining Dex and CpdA does not alter the expression levels of these genes compared to Dex treatment (Fig 5), except for the GR mRNA levels in C1-15 cells, which are elevated compared to Dex (S4A Fig).

CpdA does not support transrepression of pro-inflammatory genes, in presence or absence of Dex
We subsequently wondered whether CpdA supports transrepression in the lymphoid cell lines under study and whether CpdA modulates transrepression instigated by Dex-activated GR. Hence, we prestimulated the cells 1h with Dex, CpdA or a combination thereof, followed or not by 5h of a pro-inflammatory stimulus (TNFα) and assayed the mRNA levels of A20, NFKBIA (IκBα), ICAM and RANTES.
A20 and NFKBIA, both key inhibitors of NF-κB, are induced upon TNFα stimulation and both have a GR-and NF-κB-binding site in their corresponding promoters [48,49] (Fig 6). Dex augments NFKBIA expression levels in MM1.S and C7-14 cells, and only in the latter even beyond the TNFα-induced NFKBIA expression (Fig 6, S4B Fig). TNFα-induced A20 mRNA expression is not further induced by Dex. Moreover, CpdA does not significantly decrease TNFα-stimulated A20 or NFKBIA mRNA levels in any cell line. Also, the combination of Dex and CpdA does not differentially affect the TNFα-induced expression of A20 and NFKBIA compared to Dex treatment (Fig 6, S4B Fig).
The pro-inflammatory genes ICAM and RANTES are upregulated by TNFα treatment, while Dex only mildly reduces this stimulation for RANTES in MM1.S cells (Fig 6A). Remarkably, in MM1.S cells CpdA stimulates ICAM expression in the absence of TNF, albeit mildly. Nevertheless, in all cell lines, CpdA is unable to significantly reduce TNFα-stimulated expression of ICAM and RANTES and also the combination of Dex and CpdA does not result in further inhibition of TNFα-induced pro-inflammatory gene expression compared to Dex (Fig 6,  S4B Fig).

Discussion
The potential of SEGRAMs to reduce the number and intensity of GC-coupled side effects with improved therapeutic efficacy and prolonged responsiveness [3,15], is an attractive route to explore in lymphoid malignancies. Therefore, we investigated whether the combination of classic GCs with transrepression-favoring SEGRAMs could promote GC-induced apoptosis and/or postpone GC resistance by protecting GR from degradation.
CpdA did not substantially induce GRE-driven gene expression in MM1.S, C7-14 or C1-15 cells, in accordance with previous reports on other cell types [27,28,30,50]. The lack of CpdA to induce GR dimerization was proposed to underlie its dissociated behavior [30,42] and was also linked to decreased nuclear import and increased nuclear export [51]. In line herewith, CpdA indeed did not lead to a marked GR nuclear enrichment in MM1.S cells. Varying degrees of CpdA-instigated GR nuclear translocation were also reported in LNCaP-GR prostate cancer cells [28], CT5.3hTERT cancer-associated fibroblasts [52] and A549 cells [29], and thus seems a cell type-dependent phenomenon. In addition, CpdA's dissociated behavior could arise from a different conformational change in GR, leading to altered coregulator recruitment and slightly altered target gene preferences [27,29,53,54]. Consistent herewith, CpdA increased GILZ and ICAM expression in MM1.S and C7-14 cells, respectively. Alternatively, CpdA is able to GR-independently upregulate DUSP1 in airway smooth muscle cells as part of a mechanism by which CpdA blocks production of GC-resistant chemokines [55]. Similarly, Desmet et al. observed a CpdA-mediated upregulation of DUSP1 in intestinal epithelial cells and to a lesser extent in A549 cells [29]. In addition, HSPA1A mRNA was upregulated by CpdA in MM1.S cells, which was also reported for A549 cells, albeit as a GR-dependent phenomenon [44]. The fact that DNA itself can act as a sequence-specific allosteric regulator of GR [56], might be an additional contributor as to why certain GR ligands discriminate between GR transactivation of particular genes. Principally, CpdA could not modulate Dexinduced transactivation, except for an increased GR expression in C1-15 cells, as compared to Dex, again suggesting cell type-and gene-specific modulation.
TNF-induced pro-inflammatory gene expression was only modestly reduced by Dex in MM1.S cells and additional CpdA did not alter these Dex effects. Also CpdA alone failed to inhibit TNF-induced pro-inflammatory gene expression. This aberrant transrepression profile on top of a lack in transactivation may underpin why adding CpdA to Dex treatment largely failed to enhance GC-mediated apoptosis and to additionally inhibit cell proliferation of these lymphoid cell lines. The actual mechanisms underlying GC-induced apoptosis in lymphoid cells are not completely elucidated [4,57]. For instance, there is no consensus whether either transactivation of pro-apoptotic genes (e.g. BIM, GILZ), or the transrepression of proinflammatory genes (e.g. IL-6), anti-apoptotic genes (e.g. Bcl-XL) and cell cycle promoting genes (e.g. cyclin D1), is the most crucial mechanism governing GC-induced apoptosis [8,18,19,21,58,59]. It is most likely that both mechanisms contribute to GC-mediated cell death and that also non-genomic mechanisms are important [12]. Anyhow, we found that GC-induced killing of lymphoid cells was only slightly strengthened by adding (10 -6 M) CpdA and given its dissociated profile, this suggests that transrepression alone is most likely not sufficient to provoke GC-mediated apoptosis. In addition, CpdA alone has two different effects depending on the concentration that is used: 10 -4 M-10 -5 M CpdA (high concentration) is cytotoxic, while 10 -6 M CpdA and lower concentrations are not. These results differ from the studies of Lesovaya and coworkers, showing CpdA induced PARP cleavage and upregulated Bim and p53 expression in leukemia cells [38], with nanomolar amounts of CpdA able to reduce the cell growth of MM1.S cells [39]. However, in the same study CpdA also reduced the cell growth of GR-negative MM1.R cells at 10 -7 M [39], indicating that the observed results might have a GR-independent component.
In contrast, we only observed massive cell death with CpdA at very high concentrations (10 -4 M-10 -5 M), which was independent of the cells' GR status and responsiveness to GCs. This agrees with Wüst and colleagues, who showed that various cell types (lymphocytes, fibroblasts, neuronal cells) can undergo massive GR-independent cell death with high dose CpdA [31]. This is ascribed to CpdA's instability and thus its cyclization into an aziridine intermediate, which is known to have alkylating properties. This process especially occurs in buffers (e.g. PBS at higher pH) and after longer incubation periods [31]. However, this does not mean that in another study CpdA's effects are also per se GR-independent, as this most definitely also depends on the cell context (e.g. inflammation vs. cancer) and cell type. Prolonged treatment with CpdA was shown to protect GR from homologous downregulation in fibroblast-like synoviocytes, isolated from rheumatoid arthritis patients. Treatment for 24h with CpdA sustained GR protein levels, while Dex already downregulated GR levels after 6h [16]. CpdA treatment of CEM and NCEB cells for 24h also preserved GR levels [39]. In contrast, here, addition of CpdA to Dex treatment was unable to prevent homologous downregulation of GR in lymphoid cell lines. The proposed combination strategy is thus unable to prolong GC responsiveness in lymphoid cell lines via sustained and protected GR protein levels.
Besides in lymphoid malignancies, CpdA was also described to have anti-cancer properties in solid cancers [4]. In prostate cancer, CpdA was shown to act as a combined AR antagonist and GR agonist, resulting in the inhibition of prostate tumor growth and the induction of apoptosis in vitro [28]. The latter effect was even more pronounced when CpdA was combined with Bortezomib, as this proteasome inhibitor results in GR accumulation [36]. Also in bladder cancer, CpdA was reported to inhibit cell proliferation and induce cell cycle arrest and apoptosis in GR+/AR+ cells, and to reduce tumor growth more strongly than Dex in a xenograft model [37]. In addition, Chen and coworkers demonstrated that in triple negative breast cancer (TNBC), where GCs are given as adjuvant, CpdA regulates only a small number of genes that are not involved in carcinogenesis. This is in sharp contrast to Dex, which regulates a large set of genes that are associated with TNBC progression and drug resistance [60]. The latter exemplifies the potential for SEGRAMs in solid tumors.
Taken together, we favor CpdA's classification as a selective GR modulator (not ligand), which refers to its dissociated behavior, meaning that CpdA supports GR transrepression but not GR transactivation. Yet, this terminology does not exclude alternative action modes of CpdA such as non-genomic mechanisms, or targeting of other nuclear receptors or transcription factors. Finally, our study demonstrates that dissociated, transrepression-favoring compounds, such as CpdA, do not seem a valid therapeutic strategy in the treatment of lymphoid malignancies, but can hold promise for the treatment of solid cancers and inflammation.
Supporting information S1 Table. RT-qPCR primer sequences. were normalized accordingly. The scatter plots represent the mean +/-SEM of 3 independent experiments. Statistical analysis was performed using GraphPad Prism 7, using a two-way ANOVA with Sidak's multiple comparison post-test, comparing Dex vs. Dex/CpdA per concentration. Only significant differences are displayed: ÃÃ = P < 0.01. (C) MM1.S cells were treated for 24h or 48h with solvent, 10 -6 M Dex or a limited CpdA concentration range (10 -5 M-10 -7 M). Protein lysates were subjected to WB analysis, determining the protein levels of PARP (89 and 113kDa) and cleaved-caspase 3