Loss of Serum and Glucocorticoid-Regulated Kinase 3 (SGK3) Does Not Affect Proliferation and Survival of Multiple Myeloma Cell Lines

Multiple myeloma (MM) is a generally fatal plasma cell cancer that often shows activation of the phosphoinositide 3-kinase/Akt (PI3K/Akt) pathway. Targeted pharmacologic therapies, however, have not yet progressed beyond the clinical trial stage, and given the complexity of the PI3K/Akt signalling system (e.g. multiple protein isoforms, diverse feedback regulation mechanisms, strong variability between patients) it is mandatory to characterise its ramifications in order to better guide informed decisions about the best therapeutic approaches. Here we explore whether serum and glucocorticoid-regulated kinase 3 (SGK3), a potential downstream effector of PI3K, plays a role in oncogenic signalling in MM cells—either in concert with or independent of Akt. SGK3 was expressed in all MM cell lines and in all primary MM samples tested. Four MM cell lines representing a broad range of intrinsic Akt activation (very strong: MM.1s, moderate: L 363 and JJN-3, absent: AMO-1) were chosen to test the effects of transient SGK3 knockdown alone and in combination with pharmacological inhibition of Akt, PI3K-p110α, or in the context of serum starvation. Although the electroporation protocol led to strong SGK3 depletion for at least 5 days its absence had no substantial effect on the activation status of potential downstream substrates, or on the survival, viability or proliferation of MM cells in all experimental contexts tested. We conclude that it is unlikely that SGK3 plays a significant role for oncogenic signalling in multiple myeloma.


Introduction
Multiple myeloma (MM) is a haematologic cancer caused by mature, antibody-producing Bcells (plasma cells) [1]. It accounts for 10% of all haematological malignancies and has an incidence rate in Europe of 4.5-6/100,000/year, affecting primarily the elderly population [2]. Due to ageing societies the incidence is thus continuously rising. Most patients have benefited from the recent introduction of novel therapeutics such as proteasome inhibitors and IMiDs, working banks, which were stored in liquid nitrogen. Subsequently, MM cell cultures were freshly grown from these frozen stocks at 3-4 month intervals. The cell culture parameters were 5% CO 2 at 37°C in RPMI-1640 medium supplemented with 10% FBS, 1 mM sodium pyruvate, 2 mM glutamine, 100 U/ml penicillin, and 100 μg/ml streptomycin. INA-6 cells were supplied with 2 ng/ml recombinant human interleukin-6. All MM cell line cultures were regularly checked for mycoplasma negativity [26]. The purification of primary MM samples through microbead selection of CD138-positive cells is described in detail in [27]. Patient characteristics and treatment regimens pertaining to the primary MM samples are listed in Table 1.

Ethics statement
Primary MM cells were isolated from bone marrow aspirates of MM patients that were acquired on occasions of routine diagnostic interventions after informed consent of patients (Ethics Committee of the Medical Faculty of Würzburg University, reference numbers: 73/05, 76/13).

Electroporation of MM cells and drug treatment of electroporated cells
The protocol for MM cell electroporation with siRNA oligonucleotides and subsequent purification of viable cells is described in detail in [29]. The SGK3 stealth siRNA was used at a concentration of 3 μM in the electroporation mixtures. A stealth siRNA targeting enhanced green fluorescent protein (EGFP) was used in the control electroporations. For drug treatments of electroporated MM cells, the cells were seeded on day 2 post-electroporation into 96 well plates in 100 μl of full medium and at densities of between 15,000 and 50,000 cells per well, dependent on the specific cell line and the intended assay (higher cell numbers for alamarBlue assays, lower numbers for annexin V FACS measurements). Drugs were dissolved in DMSO and kept as frozen stock solutions, from which they were freshly diluted in full medium to either 2x (single drug additions) or 4x (two-drug combinations) working solutions. Of these either 100 μl or 50 μl were added per well of MM cells to yield final volumes of 200 μl. Cells were then cultured for an additional 3-5 days prior to cell death/viability analyses.

Mitochondrial activity/viability assay
The alamarBlue colorimetric assay was used to determine knockdown and drug effects on the viability (i.e. the collective and interconnected effects on metabolism, proliferation and cell death) of electroporated MM cells. Between 25,000 (AMO-1, JJN-3, L-363) and 50,000 (MM.1s) cells were seeded per well (96-well plates) and each test condition was prepared in triplicate. Measurements and data analyses were performed as described in the manufacturer's manual (MorphoSys, Oxford, UK). The size of effects was calculated relative to DMSO-treated controls (= 100% values).

Data analysis
Dose-effect curves were calculated from at least three independent experiments by non-linear regression analysis (sigmoidal shape, variable slope setting) using GraphPad Prism 3.0 (Graph-Pad Software, La Jolla, CA, U.S.A.). Regression analysis was not performed for data sets when it was obvious that a dose-effect relationship did not exist (MK-2206 treatment of AMO-1 cells).

Expression of SGK3 in multiple myeloma cells
In order to characterise SGK3 protein expression in MM cells in relation to expression or activation of other components of the PI3K/Akt system we initially performed Western analyses with extracts from MM cell lines and primary MM samples. The level of SGK3 varied considerably between different MM cell lines, but the protein was always detectable (Fig 1). Its expression pattern was similar to that of PI3K-p110α, a major determinant of PI3K to Akt signaling in MM [15,31], and to that of pan-Akt (Fig 1). A major drawback for functional SGK3 analyses is the apparent lack of antibodies for reliable Western assessment of SGK3 phosphorylation under regular culture conditions (i.e. steady-state levels). Although SGK3 phosphorylation across a range of solid cancer cell lines has been reported [22], with the exception of a single cell line [32] such signals have otherwise only been shown in the context of genetically engineered strong protein overexpression [32,33]. We used pairs of SGK3 depleted/control transfected MM samples, in conditions with or without IL-6/IGF-1-mediated stimulation of the PI3K system, to test a number of anti-phospho-SGK3 antibodies in Western blotting. However, we never detected signals that could specifically be attributed to phospho-SGK3 (data not shown, the antibodies tested are listed in the Methods section). Staining for phosphorylated substrates downstream of PI3K (phospho-Akt) and downstream of Akt and, potentially, SGK3 (phospho-FOXO1/3A, phospho-PRAS40, phospho-GSK-3β) confirmed the pattern for phospho-Akt previously reported for MM cell lines [10], with either strong (MM.1s, OPM-2), intermediate (JJN-3, KMS-11, L-363, MOLP-8) or low/undetectable levels (AMO-1, INA-6, U266). Whereas the phospho-PRAS40 pattern closely matched that of phospho-Akt, the patterns of phospho-GSK-3β and phospho-FOXO1/3A showed no such correlation, leaving an Akt independent contribution via SGK3 a possibility (Fig 1). Activity of the Ras/MAPK pathway (represented by phospho-ERK1/2) and the JAK/STAT pathway (represented by phospho-STAT3) was not correlated to activation of any of the PI3K/Akt system components.
SGK3 was also found to be present in all primary MM samples tested (note: the apparently very strong signal in sample pMM-6 is in part the result of rearward smearing off of a strong band that runs slightly lower than full-length SGK3) (Fig 2). Although the phosphorylation levels of PI3K system components showed strong differences between the individual primary samples, the presence of phospho-Akt was more stringently correlated with phosphorylation of downstream substrates than in MM cell lines (Fig 2). Taken together, these analyses showed that SGK3 expression in MM cells appears to be ubiquitous, but that its presence is not obviously correlated to a particular activity pattern of potential downstream substrates.

Analysis of SGK3 knockdown in MM cells
Because specific pharmacologic inhibitors for SGK3 are not available, we decided to use siRNA-mediated knockdown to investigate the consequences of SGK3 depletion in MM cells. Transient transfection of MM cell lines with RNAi oligonucleotides can efficiently be achieved through electroporation and results in substantial target depletion for at least 5 days [29]. The MM cell lines chosen in our experiments represent an intrinsically phospho-Akt negative line (AMO-1), two that are moderately phospho-Akt positive (JJN-3, L-363) and one that is strongly phospho-Akt positive (MM.1s), and whose phospho-Akt signal is dependent on PI3K-p110α activity [15]. L-363 cells also represent the rare case of an MM cell line with an activating mutation in PIK3CA. Electroporation of the stealth siRNA against SGK3 (stSGK3) into MM cells very effectively decreased SGK3 protein levels, and the knockdown remained strong even until day 7 post-electroporation (Fig 3A and 3B; a stealth siRNA targeting EGFP (stEGFP) was used in the respective control transfections). However, Western analyses showed that SGK3 depletion was without effect on the intrinsic activity levels of PI3K/Akt pathway components in all MM cell lines tested (Fig 3B). Furthermore, measurements of cell death (annexin V/PI staining (Fig 3C)) and of mitochondrial activity (alamarBlue assay (Fig 3D)) at day 4 post-electroporation did not reveal differences between SGK3 knockdown cells and their respective controls, except perhaps a slight impairment in the viability of SGK3-depleted JJN-3 cells. Cell cycle/proliferation analyses based on BrdU incorporation into newly synthesized DNA led to identical results for control and SGK3 knockdown cells (assay performed on day 4 post-electroporation, i.e. when SGK3 levels had reached their low point) (Fig 3E). Lastly, the evaluation of FACS-based mean forward scatter readings, again taken for cell samples at day 4 post-electroporation, showed that absence of SGK3 did not affect cell volume. These results are in contrast to, for example, the at least modest effects reported for SGK3 depletion in certain cell lines derived from solid cancers (breast [34], hepatic [35], prostate [36]). On the other hand, siRNA-mediated knockdown of Akt [10] or of PI3K-p110α [15] can substantially affect cell survival/viability in PI3K-p110α/Akt dependent multiple myeloma cell lines like MM.1s, L-363 and JJN-3. The consistent lack of effects in this context therefore argues against a major role of SGK3 on its own for maintenance of MM cell proliferation and survival.

Analysis of SGK3 knockdown in combination with inhibition of Akt
We next tested if SGK3 knockdown might be more harmful within the context of simultaneous inhibition of Akt, because both enzymes occupy equivalent positions in signalling pathways downstream of PI3K [19]. SGK3 could thus represent a second pillar of PI3K-mediated growth and survival signalling and its function in MM cells might only be indispensible in conditions of Akt inhibition, or its inhibition might enhance the effects of Akt blockade. We tested two allosteric inhibitors of Akt, MK-2206 [37,38] and the structurally unrelated compound Akti 1,2 Fig 2. SGK3 expression in relation to (activated) signalling components of the PI3K/Akt system in primary MM cells. Shown are Western blots prepared from frozen pellets of primary MM cells purified by CD138 microbead selection. The material was used to load two gels, with the representative β-actin control belonging to the same blot on which PI3K-p110α, P-FOXO1/3A, P-Akt (Thr308) and pan-Akt were also stained. Of note, the phospho-Akt (Thr308) antibody also stained a slightly larger band in most primary MM samples (marked by "?") that was not visible in any MM cell line. This band runs slightly higher, though, than the SGK3 band as stained on the parallel blot. Staining for P-PRAS40 (CST; no. 2997) was performed after staining for P-GSK-3β. Since both antibodies were raised in rabbit the latter signal reappeared in the P-PRAS40 blot. (also known as Akt inhibitor VIII) [39], both of which bind in a pocket between the pleckstrinhomology domain and the kinase domain and thus prevent Akt phosphorylation/activation [40]. Both compounds abrogated the constitutive phosphorylations of Akt and of the Akt downstream substrates FOXO1/3A and PRAS40 in MM cells, although MK-2206 was effective at lower concentrations than Akti1,2 ( Fig 4A; downregulation of phospho-Akt signals with Akti1,2 only shown for the strongly phospho-Akt positive MM.1s cells). These inhibitors did not affect phosphorylation of the upstream kinase PDK-1 at position Ser241, which is presumed to be essential for PDK-1 activity [41]. Both Akt inhibitors were titrated to MM cells electroporated with either stealth siRNA against EGFP (electroporation control cells, green dots in Fig 4B and 4C) or against SGK3 (red dots in Fig 4B and 4C). The drugs were added 2 days after electroporation, i.e. at a time when cell integrity was expected to be fully re-established and when SGK3 knockdown was mostly achieved. Cells were then cultured for another 3 days prior to cell death (annexin V/PI) or viability (alamarBlue) measurements. Additionally, for the MK-2206 treatments, another control titration with cells taken from the regular cell culture (blue dots in Fig 4B) was simultaneously performed to account for potential effects caused by the electroporation procedure. Cell line MM.1s, which is strongly dependent on Akt signalling for survival [10], showed complete killing curves for both inhibitors, whereas the phospho-Akt negative MM cell line AMO-1 was unaffected (lower and not dose-dependent viability values in the AMO-1 alamarBlue assay for treatment with MK-2206 were mostly a consequence of the electroporation treatment per se) (Fig 4B). L-363 cells displayed little to moderate Akt inhibitor-induced acute cell death but were more profoundly affected in their viability, a result in keeping with the role that Akt signalling has for maintenance of metabolic activity [42]. In general, treatment of SGK3 knockdown cells with Akt inhibitors induced marginally higher rates of apoptosis and/or decreases in proliferation/viability compared to respective control cells, but the differences were at best slight and, for MM.1s cells treated with MK-2206, indistinguishable from the effects on untransfected control cells. Taken together, these experiments therefore do not support the notion that SGK3 might act as a redundant safeguard mechanism to maintain growth and survival of MM cells under conditions of Akt blockade.

SGK3 knockdown in combination with MEK blockade
We have previously shown that the combined blockade of PI3K and MEK1,2 in the phospho-Akt negative MM cell line AMO-1 leads to strong synergistic induction of cell death. This effect was observed for the combination of pan PI3K/mTOR inhibitor PI103 with MEK1,2 inhibitor PD184352 [11], as well as for the combination of PI3K-p110α inhibitor BYL-719 with MEK1,2 inhibitor PD0325901 [15]. However, no such effect was found for the combination of Akt inhibitor Akti1,2 with MEK1,2 inhibitor PD98059 [10]. We therefore tested if the PI3K-dependent survival signal in this combination blockade with MEK inhibitors is transmitted by SGK3 (Fig 5). AMO-1 cells electroporated with either stealth siRNA against SGK3 (gray bars) or against EGFP (white bars) were kept in cell culture for two days to recover from the effects of electroporation and to downregulate SGK3. Cells were then subjected to single or combination drug treatments for 3 additional days, and apoptosis was measured by annexin V/PI staining and FACS analysis. As expected, the combination of PI3K-p110α and MEK1,2 inhibition (BYL-719 + PD0325901) strongly affected AMO-1 cell survival, whereas each drug alone had little impact, and this effect was not mirrored by combinations of Akt and MEK inhibitor (MK-2206 + PD0325901 or Akti1,2 + PD0325901) (Fig 5A). However, apoptosis induction in   [15]. Of note, the strong synergistic effect observed in AMO-1 cells for the combination of PI3K-p110α inhibitor BYL-719 and MEK1,2 inhibitor PD0325901 is not observed when the PI3K-p110α inhibitor is substituted with an Akt inhibitor (MK-2206 or Akti1,2), but it is also not mirrored by the combination of MEK1,2 inhibition and SGK3 depletion.
doi:10.1371/journal.pone.0122689.g005 PD0325901-treated AMO-1 cells depleted for SGK3 was no more effective than in control cells (Fig 5A, third columns from the left), arguing against SGK3 as the crucial mediator of PI3K-p110α dependence in this cell line. Although SGK3 knockdown cells exhibited slightly increased rates of cell death in some settings (i.e. for treatment with BYL-719 or with PD0325901 + MK-2206; Fig 5A) the effects were always small and within the limits of experimental variation. Taken together, these experiments provided no evidence for a functional role of SGK3 downstream of PI3K-p110α in AMO-1 cells. L-363 cells, which harbour an activating PIK3CA mutation, also showed no substantial increases in cell death when pharmacological blockade of MEK1,2, Akt or PI3K-p110α was performed in cells devoid of SGK3 (Fig 5B).

Serum dependence of MM cells in the context of SGK3 depletion
We finally tested if SGK3 knockdown had any influence on the survival of MM cells in conditions of serum deprivation, because extrinsic serum-dependent growth and survival signals should at least in parts be transmitted by PI3K. MM cells were electroporated with stSGK3 or stEGFP siRNAs and kept in cell culture for 2 days in order to recover and to downregulate SGK3. Cells were then washed 3 times with PBS and resuspended in full medium without FBS. The concentration of FBS was then adjusted for the different test conditions and cell death determined after 3 additional days in culture (Fig 6). In contrast to effects described for the hepatocellular carcinoma cell line QGY-7701 [35] SGK3 knockdown had no influence on the serum dependence of any of the MM cell lines tested (Fig 6).

Relevance of SGK3 for the growth and survival of MM cells
The PI3K/Akt system is generally recognised as an important signalling and survival axis in tumor biology, including in MM [10,15,43,44,45], but clinically approved targeted molecular therapies are currently still limited to inhibitors of the mTOR components [17]. However, a large number of candidate compounds including isoform-specific PI3K inhibitors as well as pan-PI3K or Akt inhibitors, and drugs exhibiting dual target specificities towards PI3K/mTOR are in preclinical and clinical evaluation [46]. A phase III trial in MM with perifosine, an alkylphospholipid that leads to downregulation of Akt [47], has failed to show additional therapeutic benefit when the drug was combined with bortezomib and dexamethasone. However, a recent phase I trial with the pan-Akt inhibitor GSK2110183 (afuresertib) revealed clinically relevant responses in some MM patients [18], thus suggesting that Akt blockade can be beneficial if the right patients and therapeutic settings can be identified. Because our preclinical analyses of the Akt and PI3K isoforms in MM have revealed strong variability in the intrinsic activity of this pathway and in the response of MM cells to its blockade [10,15], we tested if SGK3 might provide a significant escape route from PI3K or Akt blockade-mediated apoptosis. SGK3 has recently been described as an oncogenic player in some solid cancers. Most prominently, it was identified as a crucial component in PI3K downstream signalling in cancer cells that harbour activating PIK3CA mutations, a condition that is often associated with relatively low intrinsic Akt activity [22]. SGK3 has also been described as androgen or estrogen inducible survival kinase in prostate [36] and breast cancer [34], respectively, and the SGK3 locus constitutes a genetically amplified lesion in hepatic cancer [35]. However, these examples also show that for SGK3 no consistent oncogenic mechanism is implied, and an immunohistochemical study in ovarian cancer did not find SGK3 phosphorylation (activation) to be correlated to either the phospho-Akt status or the PIK3CA mutation status [48].
In our experiments we detected SGK3 protein expression in all MM cell lines and primary MM samples tested, but our functional analyses found no evidence for a substantial role of this kinase in the maintenance of growth and survival of MM cells. A clear limitation of our study is the lack of functional analyses with primary MM cells, which have been impossible due to the absence of suitable pharmacologic inhibitors, the non-transfectability of primary MM cells and a general dearth in unambiguous functional readouts for intrinsic or extrinsic SGK3 activation. However, the four transfectable MM cell lines chosen for our experiments do represent a broad range of intrinsic PI3K/Akt activation levels, and they represent MM cell lines that strongly (MM.1s), moderately (L-363, JJN-3) or not at all respond with apoptosis/viability decreases to blockade of Akt or PI3K/PI3K-p110α [10,11,15]. L-363 also represents the rare case of an MM cell line with an activating PIK3CA mutation. Furthermore, stealth siRNA-mediated transient SGK3 knockdown led to reliable and strong SGK3 protein depletion for more than 5 days, and thus permitted the addition of small molecule pathway inhibitors to MM samples virtually devoid of SGK3. Based on previous experience with Akt and PI3K-p110α blockade the time frame for these experiments should have been sufficient to establish functional effects [10,11,15]. The fact that SGK3 depletion alone or in combination with Akt inhibition failed to show an SGK3-dependent contribution to the survival or viability of MM cells therefore argues against a prominent role for SGK3 in the oncogenic deregulation of the PI3K/Akt system in multiple myeloma, and against its utility as a potential therapeutic target in this disease. Based on the limited scope of our study it can certainly not be ruled out that either in a clinical setting or perhaps for smaller MM subgroups SGK3 blockade might still prove beneficial. However, at the current stage of SGK inhibitor development no options exist to rigorously assess this.