A Novel Approach to Safer Glucocorticoid Receptor–Targeted Anti-lymphoma Therapy via REDD1 (Regulated in Development and DNA Damage 1) Inhibition

Though the treatment of hematologic malignancies has evolved over the past decades, glucocorticoids (Gcs) remain the essential component of combination therapies, and the glucocorticoid receptor (GR) is a well-recognized target for blood cancers (1, 2). Unfortunately patients chronically treated with Gcs develop numerous metabolic and atrophic adverse effects including osteoporosis, muscle waste, and skin atrophy that strongly affect their life quality (2–4). Thus, there is a great need for the development of safer GR-targeted therapies.

GR is a transcription factor that regulates gene expression via transactivation (TA) that requires GR binding to the responsive elements in gene promoters, and transrepression (TR) that is mediated, among other mechanisms, by binding and blocking other TFs including proinflammatory NF-kB and AP-1 (2, 5, 6). GR TR is an important mechanism underlying therapeutic anti-inflammatory and anticancer effects of Gcs, and it usually does not require GR DNA binding (7). GR TA regulates GR signaling linked to gluconeogenesis and catabolism, and often mediates steroid atrophic effects in different tissues (2, 8). Thus, for years the major efforts in the field have been directed toward the development of selective GR agonists/modulators (SEGRAM) that shift GR activity toward TR (9). The alternative approach could be the combination of Gcs with compounds protecting tissues against steroid adverse effects.

Recently, mTOR/Akt inhibitor REDD1 (Regulated in development and DNA damage 1) was identified as the key mediator of Gcs-induced atrophy in different tissues. Indeed, in REDD1 KO animals, muscle and skin were protected against Gcs atrophy (10, 11). At the same time in REDD1 KO mice, the anti-inflammatory potential of Gcs was completely preserved (10), suggesting that inhibitors of REDD1 expression could dissociate therapeutic and adverse atrophogenic effects of Gcs.

As REDD1 pharmacologic inhibitors are not known, we applied drug repurposing approach. Using bioinformatics screening of Library of Integrated Network-Based Cellular Signatures (LINCS) database of approximately 20,000 transcriptional signatures induced by FDA-approved and experimental drugs (http://lincsproject.org/LINCS/), we identified a significant number of putative inhibitors of REDD1 expression among PI3K/Akt/mTOR inhibitors (12, 13) including classical experimental inhibitors such as LY294002 and Wortmannin (WM), and drugs in clinical trials such as Rapamycin (Rapa) and AZD8055. This was completely unexpected as PI3K/Akt/mTOR inhibitors block major proproliferative signaling in cells. We demonstrated that compounds from this pharmacologic class indeed inhibited REDD1 in keratinocytes and attenuated skin atrophy induced by topical Gcs but did not interfere with their anti-inflammatory effects (12, 13).

The goal of this work was to evaluate whether the same approach (Gcs+PI3K/Akt/mTOR inhibitor) is applicable to increase safety of systemic Gcs used for blood cancer. We analyzed the effects of PI3K/Akt/mTOR inhibitors AZD8055, WM, and LY294002 on REDD1 expression and GR function in leukemia and lymphoma cells. We assessed atrophogenic effects of systemic Dexamethasone (Dex)±LY294002 or Rapa, previously validated as REDD1 inhibitor (13), using osteoporosis and skin atrophy mouse models. We also monitored antilymphoma effects of this combination therapy in human lymphoma xenograft model to assure that the regimens chosen for tissues protection remain optimal for anticancer activity. We found that LY294002 and Rapa prevented development of Dex-induced osteoporosis and skin atrophy, and enhanced Dex anti-lymphoma effects.

LY294002, WM, and AZD8055 were from LC Laboratories, Dex and other chemicals were from Sigma-Aldrich.

Putative REDD1 inhibitors were identified by computational screening of LINCS library as described in ref. (12). The molecular signature of each compound in each experiment is presented at LINCS as a list of differentially expressed genes, ordered by descending expression fold change. The top putative REDD1 inhibitors were selected according to the number of LINCS experiments in which REDD1 was within 100 most down-regulated genes in treated cells. For statistical computing, we used the R project version 3.2.5 (https://www.r-project.org/). Selected compounds are presented in Supplementary Table S1.

Acute lymphoblastic leukemia cell line CEM and mantle cell lymphoma cell line Granta were obtained from the ATCC and cultured in RPMI-1640 (Thermo Fisher) with 10% FBS (HyClone), sodium pyruvate (10 mmol/L), HEPES (10 mmol/L), and antibiotics (referred thereafter as complete medium; Thermo Fisher) as described (14). Peripheral blood samples were collected from healthy volunteers. The ethical Committee of the Blokhin Cancer Research Center approved this study, and written-informed consents were provided. Mononuclear cell fraction was isolated by 1.077 g/mL Ficoll-Isopaque density-gradient centrifugation and cultured in the complete medium. Cells were pretreated with Rapa (0.1–5 μmol/L), LY294002 (10–100 μmol/L), WM (1–10 μmol/L), and AZD8055 (0.1–2 μmol/L) or vehicle for 6 hours, and then were treated for Dex (1–10 μmol/L) or vehicle for 24 to 144 hours as indicated.

Western blot analysis was performed as described in ref. (10). The following antibodies were used: GR (H-300; Santa Cruz Biotechnology), phospho-GR (Ser211), phospho-4E-BP1 (Thr37/46), phospho-Akt (Ser473), Akt, 4E-BP1, cleaved and full-length PARP, GAPDH, HDAC1 (Cell Signaling Technology), and REDD1 (Proteintech Group).

Cell viability was assessed using the MTT assay as follows: cells were incubated at 37°C for 3 hours with 20 μL of MTT solution (5 mg/mL). Supernatant was discarded, and 150 μL of 100% DMSO was added to each well. Optical density was measured at 495 nm using a microplate reader MultiScan MCC 340 (Labsystems). The IC₅₀ values were calculated using Quest Graph IC₅₀ Calculator free software. Isobolograms were constructed as described previously (15). In brief, IC₅₀ values of PI3K inhibitors or Dex as single agents were plotted as 1 on the x and y axes, respectively. Various concentrations of PI3K inhibitors and for Dex were combined. The IC₅₀ values for each compound in the combination were calculated and plotted as a percentage of each single drug. The disposition of data points below, above, or on the line of additivity indicates synergy, antagonism, or additivity, correspondingly.

Apoptosis induction was evaluated using Western blot analysis of PARP cleavage and propidium iodine (PI) staining as described in (14).

Total RNA isolation, reverse transcription, and Q-PCR were performed as described in ref. (10). Primers were designed with NCBI Primer-BLAST (Supplementary Table S2). Results were normalized to the expression of the housekeeping RPL27 gene (10).

GRE.Luc, NF-κB.Luc, and control mCMV.Luc reporter cells were generated as described in ref. (13) using viral stocks obtained from Northwestern University SDRC DNA/RNA delivery Core. Luciferase activity was measured as described in ref. (10).

Ten million Granta cells in 200 μL BD Matrigel Matrix were subcutaneously injected into the right flank of 7-week-old female nu/nu mice (Taconic). Animals were randomized and treated every 48 hours with LY294002 (20 mg/kg), Rapamycin (5 mg/kg) of vehicle (30% PEG3350, 4% DMSO, 5% Tween 20 in PBS) followed by Dex (1 mg/kg) of vehicle (5% DMSO in PBS) 6 hours later. Body weight was recorded twice a week. Tumor size was measured twice a week by digital calipers, and drug interaction mode was calculated based on combination index method described in refs. (16, 17). Combination index (CI) of Dex + Rapa/LY294002 was calculated as ratio (observed fraction value of combination group)/(expected fraction value of combination group). The observed fractions were calculated as ratio (mean value of experiment)/(mean value of control), and the expected fraction values of combination effect between combined groups and single drug groups were calculated as expected fraction = (mean observed fraction of Dex) × (mean observed fraction of Rapa/LY294002). CI < 1, CI = 1, and CI > 1 indicate a synergy, additive effect, and antagonism, respectively.

Glucocorticoid-induced osteoporosis was induced as described in ref. (18) by i.p. injection of Dex (10 mg/kg) in 12-week-old BALB/c female mice (“Stolbovaya”, Moscow region, Russia) every 48 hours for 5 weeks. Six hours prior to Dex, animals were treated i.p. with LY294002 (20 mg/kg) and Rapamycin (5 mg/kg) of vehicle (30% PEG3350, 4% DMSO, 5% Tween-20 in PBS). Body weight was recorded twice a week.

All animal experiments were in compliance with Northwestern University Animal Care and Use Committee (ACUC) and Blokhin NMRCO–approved protocols.

Xenograft and skin samples were fixed in formaldehyde and embedded in paraffin, and sections were stained with hematoxylin and eosin. The bone samples after formaldehyde fixation were decalcinated in decalcination buffer (0.5 mol/L EDTA, 0.7 mmol/L NaOH, pH = 7.0) for 14 days, then washed with PBS twice, and further processed according to standard histology protocol. Ki67 and Caspase 3 were detected by immunostaining with antibodies to Ki-67 and Caspase-3 (Cell Signaling Technology).

Quantification of the epidermal width, dermal adipose thickness, bone thickness, and cellularity of xenograft tissue was performed in sections stained with hematoxylin and eosin. At least 5 individual fields per slide of each individual sample/treatment (at least 50 images/treatment group) were analyzed using Axioplan2 microscope software (Carl Zeiss). The epidermal, dermal adipose, or bone thickness and cell density in treated animals is presented as % to control animals.

Mean and SD values were calculated using Microsoft Excel software. The treatment effects in each experiment were compared by one-way ANOVA or t test. Differences between groups were considered significant at P < 0.05. All in vitro experiments were repeated 3 times. In animal experiments, we used 10 animals/experimental group.

REDD1 pharmacologic inhibitors are not known. Thus, we searched for the inhibitors of REDD1 expression using bioinformatics screening of a large LINCS library of transcriptomic signatures of approximately 20,000 small-molecule drugs tested at large range of concentrations in approximately 50 human cell lines (http://lincsproject.org/LINCS/). We scored compounds according to the number of LINCS experiments in which REDD1 expression was significantly inhibited (REDD1 was within top 100 downregulated genes) and found that inhibitors of the PI3K/mTOR/Akt pathway represented the major pharmacologic class among the top inhibitors of REDD1 expression. Even though, it was a rather unexpected outcome, we have recently validated REDD1-suppressive effects of several PI3K/Akt/mTOR inhibitors in human keratinocytes (12, 13).

For this study, we selected three compounds: classical experimental PI3K inhibitors WM and LY294002; and dual mTOR inhibitor AZD8055 that recently entered clinical trials as anticancer drug (ref. 19; Supplementary Table S1). We also used mTOR inhibitor, Rapamycin (Rapa), as a reference REDD1 inhibitor characterized previously (13). It is known that PI3K activation leads to the phosphorylation and activation of Akt, which in turn activates mTOR in both mTORC1 and mTORC2 complexes (20). Thus, we expected that the selected inhibitors could affect to some degree all branches of PI3K/Akt/mTOR signaling.

The anti-REDD1 effects of these compounds were tested in mantle cell lymphoma Granta and acute lymphoblastic leukemia CEM cells, which are widely used for the screening of novel chemotherapeutics for blood cancer treatment (1, 2, 14).

The optimal regimen of cell treatment: pretreatment with REDD1 inhibitors for 6 hours followed by Dex treatment for 24 hours was chosen as described (12, 13). Dex alone significantly induced REDD1 expression at mRNA/protein levels in both Granta and CEM cells. All tested compounds, LY294002 (50 μmol/L), WM (10 μmol/L), and AZD8055 (1 μmol/L) prevented REDD1 induction (Fig. 1; Supplementary Fig. S1).

The selected regimen and concentrations of REDD1 inhibitors strongly downregulated PI3K/Akt/mTOR signaling monitored by phosphorylation of mTOR substrate 4E-BP1 (eukaryotic initiation factor 4E binding protein 1), and Akt at Ser473, the major site regulating Akt activity (Fig. 1B and D). PI3K/Akt/mTOR inhibitors also significantly reduced basal expression of REDD1 mRNA (in CEM cells, Fig. 1A) and protein (more pronounced in Granta cells, Fig. 1D).

REDD1 is a GR target gene (10, 21). To assess whether REDD1 inhibitors could modify GR function, we evaluated LY294002, WM, and AZD8055 effect on GR TA and TR using CEM and Granta cells transduced with GRE.Luc and NF-kB.Luc reporters. All three tested compounds inhibited Dex-induced GRE.Luciferase activity in both cell lines (Fig. 2A and C). Moreover, they also negatively affected basal GR activity when applied individually, with LY294002 being the most effective. Attenuation of GR TA was further confirmed by Q-PCR analysis of known GR-target genes including GR chaperone FKBP51 (12, 14), mediator of Gcs anti-inflammatory effects GILZ (22, 23), transcriptional activator KLF9 (24), and inhibitor of cytokine production MKP-1 (ref. 25; Fig. 2B and D; Supplementary Fig. S2A and S2B).

NF-kB activity was strongly reduced according to kB.Luciferase test by Dex as well as by PI3K/Akt/mTOR inhibitors during individual treatment, and in case of Dex combination with AZD8055 or WM, we observed additive inhibitory effect in both cell lines (Fig. 2E and G). This finding was further corroborated by Q-PCR analysis of negative regulation of several endogenous genes such as cell-cycle regulators cyclins, CCND1/D2, and IL7R involved in lymphoid cell survival (26), and to the lesser extent, a marker of malignant lymphoid cells CD86 (Fig. 2F and H).

To identify the molecular mechanisms underlying GR functional changes induced by PI3K/Akt/mTOR inhibitors, we assessed how these compounds affected major steps in GR activation: GR phosphorylation at Ser211 and nuclear translocation that are critical for GR TA (27, 28).

All studied compounds prevented GR phosphorylation induced by Dex in both cell types to a different degree (Fig. 2I). Two out of three tested PI3K inhibitors, LY294002 and AZD8055, in addition to the effect on GR phosphorylation significantly prevented GR nuclear import (Supplementary Fig. S2C).

PI3K/Akt/mTOR inhibitors were developed as anticancer drugs. To assess whether and how PI3K/Akt/mTOR inhibitors interacted with Dex anti-lymphoma effects, we first evaluated the efficacy of the combined treatment in vitro using such readouts as cell viability (MTT test) and apoptosis (PARP cleavage and sub–G₁-phase cell population determined by FACS). We tested cytotoxic effect of Dex and PI3K inhibitors in wide concentration range (LY294002 1–500 μmol/L, WM 0.1–25 μmol/L, AZD8055 0.05–5 μmol/L; Supplementary Fig. S3) and determined IC₅₀ values after 24-hour treatment (Supplementary Table S3). PI3K inhibitors at optimal, close to IC₅₀ concentrations (50 μmol/L for LY294002, 1 μmol/L for AZD8055, and 10 μmol/L for WM) decreased CEM and Granta survival by 70% to 90%, and Dex (1 μmol/L) by 50% to 55% when used alone for 6 days (Fig. 3A and E). At the same time, normal lymphocytes from healthy volunteers were significantly more resistant to cytotoxic effects of combinational treatments (Supplementary Fig. S4).

To assess the type of interactions between Dex and PI3K/Akt/mTOR inhibitors in antilymphoma tests, we used determined IC₅₀ values for inhibitors and Dex after 24-hour treatment (Supplementary Table S3), generated isobolograms, and used statistical models to determine synergy, additivity, or antagonism between Dex and PI3K inhibitors (Fig. 3B–D, F–H). Isobologram analysis revealed synergy in antiproliferative effects of Dex and LY294002 in both cell types (Fig. 3B and F). Synergy was also determined for Dex and Rapa, used as reference REDD1 inhibitor (Supplementary Fig. S5). The effects of Dex and AZD8055 or WM on lymphoma cell survival were mostly additive (Fig. 3C, D, G, and H).

Next, we examined effect of Dex combined with LY294002, AZD8055, WM, or reference inhibitor Rapa on apoptosis. When used alone, Dex and all three PI3K inhibitors as well as Rapa stimulated apoptosis in CEM and Granta cells as assessed by increase in sub–G₁-phase cell population (Fig. 3J; Supplementary Fig. S5). LY294002 cooperated with Dex in apoptosis induction; other inhibitors increased apoptosis development compared with Dex only in both cell types. These findings were confirmed by Western blot analysis of PARP cleavage (Fig. 3K), that revealed the especially pronounced cooperative proapoptotic effects between LY294002 and Dex in CEM cells.

To assess whether REDD1 inhibitors from PI3K/Akt/mTOR pharmacologic class enhance Dex antilymphoma activity in vivo, we used previously established s.c. Granta xenograft model in athymic nu/nu female mice (2). We selected LY294002 that displayed synergistic antilymphoma effects with Dex in vitro. In addition, we used as a positive control mTOR inhibitor Rapamycin whose synergistic anticancer effects with Dex in vitro were reported (29). We previously characterized Rapa as REDD1 inhibitor and tissue protector against atrophic effects of topical Gcs in skin (13).

Animals were randomized to different treatment groups after tumor volume reached 50 mm³. All compounds were delivered via i.p. injections of LY29400 (20 mg/kg), Rapa (5 mg/kg), or vehicle in the morning followed by Dex treatment (1 mg/kg) or vehicle 6 hours later following the optimal regimens for REDD1 inhibition developed in vitro (2, 13). All treatment regiments were only minimally toxic according to the steady increase in body weight monitored twice weekly (Supplementary Fig. S6). Tumor volume was measured also twice a week (Supplementary Fig. S7).

At the end-point of experiment (27th day of treatment), the average tumor volume in control was 1,489 ± 161.3 mm³. In groups treated with Dex and LY294002 individually, the tumor volume decreased by approximately 40% and approximately 70% correspondingly (Fig. 4A). We observed the remarkable 85% decrease in tumor volume in LY294002+Dex group. Similar cooperation in anticancer effects was observed for Rapa and Dex (∼75% decrease in tumor volume in Rapa-treated group and ∼85% in Rapa + Dex group), which was in line with published data (29–32). To determine the mode of combined action of Dex and LY294002/Rapa, we used the CI approach based on the difference between the expected and observed mean tumor volumes in Dex+LY294002 or Dex + Rapa groups compared with control. Calculated CI values of 0.87 for Dex + LY294002 and 0.85 for Dex+Rapa indicated synergistic mode of interaction.

The reduction in xenograft tumor volume in Dex+LY294002 and Dex+Rapa groups reflected drastic decrease of tumor cellularity, by 50% to 70% in groups with combined treatments with superior effect of Dex + LY294002 (Fig. 4B and C). It is known that the major effects of Gcs in transformed lymphoid cells are growth inhibition and apoptosis (33, 34). Thus, we evaluated proliferative activity of the cells in xenografts as well as apoptosis by immunostaining with proliferation marker Ki67 and apoptosis marker caspase 3. We observed the lowest number of Ki67-positive cells and the highest percentage of apoptotic cells in the xenografts from LY294002+Dex-treated mice (Fig. 4D–G) that further confirmed that LY294002 cotreatment enhanced anti-lymphoma effects of Dex. Effects of Rapa on antiproliferative and proapoptotic Dex activity were less pronounced, but still statistically significant for increased apoptosis induction.

We selected LY294002 that most significantly modulated GR function and displayed synergistic antilymphoma effects with Dex both in vitro and in vivo, to test its tissue-protective effect against atrophy induced by systemic exposure to Dex.

The mechanisms of Gcs atrophic effects are complex and depend on the tissue. In skin and muscle, the GR molecular signature appeared to be similar (10), and in both tissues, Gcs significantly shift metabolism to catabolic processes that affect lipids, sphingolipids, and proteins (10, 35). The steroid-induced skin atrophy involves all skin compartments and results in overall skin thinning, changes in collagen synthesis, atrophy of epidermis, and dermal adipose (8, 36, 37).

In bone, Gcs induce the shift of bone cell populations toward bone-resorbing osteoclasts via increased osteoclast formation and inhibition of proliferation of bone-forming osteoblasts leading to an increased risk of fragility and bone fractures (38). One of the major signaling pathways that regulates osteoclast formation from their precursors as well as their activation and survival during normal bone remodeling and in a variety of pathologic conditions is mediated by RANK (receptor activator of nuclear factor κB)/RANKL (RANK ligand) signaling (39). Another important regulator of osteoclasts is cytokine Osteoprotegerin (OPG) that acts as a decoy receptor for RANKL. Thus, OPG protects the skeleton from excessive bone resorption, and RANKL/OPG ratio is frequently used as an important determinant of bone mass and skeletal integrity (39). Collagen family proteins (especially Col1a1 whose polymorphisms are associated with age-related osteoporosis) also play a key role in bone formation, structure, and mineralization (40, 41).

To test whether REDD1 inhibitors protect against systemic atrophic effects of Gcs, we used previously described model of Gcs-induced osteoporosis in Balb/c female mice (18) and assessed atrophy in both bone and skin induced by high-dose i.p. Dex injections (10 mg/kg) every 48 hours for 5 weeks. LY294002 (20 mg/kg), Rapa (5 mg/kg), or vehicle were injected 6 hours prior animal treatment with Dex. Over the course of the experiment, general animal health, as determined by cage-side observations and animal body weight, was similar between control and experimental groups; moreover, we saw a normal steady increase in animal body weight in all groups during the experiment (Supplementary Fig. S8).

We revealed the tendency in decrease of femoral bone thickness between control and Dex-treated animals (Fig. 5A). Unfortunately, this tendency was not statistically significant, possibly due to the duration of treatment. However, changes in widely accepted osteoporosis markers, RANKL (associated with bone resorption), OPG (protector from bone resorption), and RANKL/OPG ratio, clearly indicated that in Dex-treated group, femur bones started to be affected by osteoporosis: RANKL/OPG ratio was significantly increased after 5-week treatment (Fig. 5B). Importantly, both LY294002 and Rapa induced downregulation of RANKL expression, upregulation of OPG expression, and, correspondingly, the significant decrease of RANKL/OPG ratio toward control level. Moreover, expression of Col1a1 and especially Col2a1 decreased after Dex chronic treatment and was restored in bone samples from animals treated with LY294002 or Rapa (Supplementary Fig. S9).

To test whether LY294002 and Rapamycin could prevent development of skin atrophy induced by systemic Dex delivery, we assessed morphologic changes in skin of Balb/c female mice. In agreement with previously published results (8), chronic treatment with systemic Dex led to a very significant cutaneous atrophy: epidermal thickness was reduced by approximately 50%, and dermal adipose tissue was reduced by approximately 70% (Fig. 5C–E). Remarkably, LY294002 and Rapamycin partially protected skin from these hypoplastic effects of Dex: LY294002+Dex-treated skin displayed only 35% and 45% reduction in epidermal and dermal adipose thickness accordingly. The protective effects of Rapamycin were similar.

Significant success in cancer treatment is turning many cancers including blood cancer into chronic diseases. Thus, there has been a growing interest to the treatment-related adverse effects affecting health and quality of life during/following cancer therapy. The focus of studies presented here is on the use of tissue protectors to mitigate adverse effects of systemic Gcs without impact on their anti-lymphoma potential.

Gcs are very effective against a number of hematologic malignancies. However, their significant atrophic and metabolic adverse effects motivated development of novel steroidal and non-steroidal GR activators and modulators (SEGRAM) with reduced side effects. Numerous SEGRAM have been synthesized and tested by industry and academia; however, the success was rather limited (42). There also have been efforts to limit systemic side effects of Gcs using short-living Gcs-based prodrugs that act locally and then undergo deactivation (42). Unfortunately, only few SEGRAM reached clinical trials but have never been approved for wide-scale clinical use.

We propose here an alternative approach to safer GR-targeted therapies via combination of Gcs with tissue protectors which we selected by the bioinformatics screening of small-molecule drugs based on their negative effect on atrophogene REDD1. As far as we know, this is the first successful attempt to use drug repurposing approach to seek for compounds preventing atrophy.

We tested here antiatrophogenic compounds from the pharmacologic class of PI3K/Akt/mTOR inhibitors. The unexpected finding that PI3K/Akt/mTOR inhibitors can potently inhibit the expression of Akt/mTOR genetic regulator REDD1 confirms the existence of a little known feedback loop in Akt/mTOR signaling. As shown here and in our recent work (13), this loop tightly regulates REDD1 expression at RNA/protein levels. However, the molecular mechanisms underlying transcriptional regulation of REDD1 at basal level by mTOR have yet to be identified. Interestingly, mTOR also regulates REDD1 proteasomal degradation, and thus, mTOR inhibitors can reduce REDD1 protein stability (43).

One of our most intriguing findings was the capacity of numerous PI3K/Akt/mTOR inhibitors to modify GR function shifting overall GR activity toward therapeutically important TR both in lymphoid cell types (Fig. 2), and also in epithelial cells as assessed by Luciferase test and by global changes in Gcs transcriptome (12, 13). These effects could underlie the feedback loop in PI3K/Akt/mTOR signaling after the exposure of Gcs.

The inhibition of GR TA followed inhibition of GR phosphorylation and decreased GR nuclear import. It is known that GR is phosphorylated on multiple residues by MAPKs, cyclin-dependent kinases, casein kinase II, and glycogen synthase kinase-3β (GSK-3β; refs. 27, 44). Interestingly, some of PI3K inhibitors such as LY294002 do have off-target effects mediated by binding to BET proteins (Bromodomain and Extra-Terminal motif proteins) regulating the histone acetylation. As GR is also involved into epigenetic regulation of gene expression, possible cross-talk between Gcs/GR and BET signaling could affect the expression of GR target genes, especially cytokines and chemokines implicated into anti-inflammatory Gcs response (45–47).

Interestingly, Akt itself can be directly involved in GR phosphorylation and GR function tuning (48). These literature data suggest that PI3K/Akt/mTOR inhibitors affected GR phosphorylation status via GR kinases as mTOR and Akt cross-talk with signaling pathways involving p38, GSK-3beta is known (27, 44). At the same time, the exaggeration of GR TR branch (associated with anti-inflammatory and anticancer effects of Gcs) in part reflects profound negative effect of PI3K inhibitors on NF-kB activation (12, 13).

The directional changes in GR function (downregulation of GR TA) and blockage of atrophogene REDD1 suggested that PI3K/Akt/mTOR indeed could protect tissues against steroid atrophy. Using two different models of steroid atrophy–skin atrophy and osteoporosis, we validated protective effects of PI3K inhibitors when they and Gcs were delivered systemically. Even though the causative role of REDD1 in skin and muscle atrophy was proven experimentally (10, 11), the mechanisms underlying REDD1 effects in bone will be an interesting direction for future investigation.

The PI3K/AKT/mTOR pathway plays a central role in regulating growth and survival of lymphoid cells, and constitutive PI3K activation is frequently observed in different types of lymphoid malignancies (49, 50). Importantly, in our work, combination of Gcs with PI3K inhibitors enhanced antilymphoma effects of Dex in vitro and in vivo, in human lymphoma xenograft model, and the therapeutic effects of PI3K inhibitor + Dex combinations ranged from cooperative to synergistic (especially in case of Rapamycin and LY294002). The mechanism underlying cooperative cytotoxicity of Gcs and PI3K inhibitors could involve the attenuation of Gcs-induced prosurvival autophagy and shift to Gcs-dependent apoptosis in cells lacking REDD1 (51). Interestingly, pharmacologic inhibition of PI3K, mTOR, and Akt could overcome resistance/increase sensitivity of leukemia cells to Gcs (52).

Overall, our findings provide a proof of principle for using compounds from PI3K/Akt/mTOR inhibitors class to prevent atrophic effects induced by chronic systemic treatment with Gcs. This combination also appeared to be exceptionally promising due to enhanced antilymphoma activity of Gcs. In addition to atrophy induced in different tissues, chronic treatment with Gcs results in metabolic, gastrointestinal, and cardiovascular side effects (53). As some of those adverse effects including glucose metabolism also strongly depend on TA branch of GR signaling (54), it is conceivable that PI3K/Akt/mTOR inhibitors could potentially ameliorate some metabolic adverse effects of Gcs, suggesting clinical applications for wide range of different diseases and disorders treated with Gcs.

No potential conflicts of interest were disclosed.

Conception and design: E.A. Lesovaya, M.G. Yakubovskaya, I. Budunova

Development of methodology: E.A. Lesovaya, A.V. Savinkova, O.V. Morozova, A. Klopot, J.T. Dudley

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A.V. Savinkova, O.V. Morozova, E.S. Lylova, E.P. Kulikov, K.I. Kirsanov, A. Klopot, G. Baida

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): E.A. Lesovaya, A.V. Savinkova, M.G. Yakubovskaya, L.I. Gordon, B. Readhead, J.T. Dudley

Writing, review, and/or revision of the manuscript: E.A. Lesovaya, A.V. Savinkova, E.M. Zhidkova, M.G. Yakubovskaya, L.I. Gordon, I. Budunova

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): E.M. Zhidkova

This work is supported by R01GM112945 and R01AI125366 (to I. Budunova and J.T. Dudley), HESI-THRIVE grant (to I. Budunova), and Russian Science Foundation grant 17-75-20124 (to E.A. Lesovaya). The authors would like to thank University of Chicago Genomics Facility, and Northwestern University SDRC (5 P30 AR057216) morphology and phenotyping, skin tissue engineering, and DNA/RNA delivery Cores for technical support.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.