The Rationale for the Dual-Targeting Therapy for RSK2 and AKT in Multiple Myeloma

Multiple myeloma (MM) is characterized by remarkable cytogenetic/molecular heterogeneity among patients and intraclonal diversity even in a single patient. We previously demonstrated that PDPK1, the master kinase of series of AGC kinases, is universally active in MM, and plays pivotal roles in cell proliferation and cell survival of myeloma cells regardless of the profiles of cytogenetic and genetic abnormalities. This study investigated the therapeutic efficacy and mechanism of action of dual blockade of two major PDPK1 substrates, RSK2 and AKT, in MM. The combinatory treatment of BI-D1870, an inhibitor for N-terminal kinase domain (NTKD) of RSK2, and ipatasertib, an inhibitor for AKT, showed the additive to synergistic anti-tumor effect on human MM-derived cell lines (HMCLs) with active RSK2-NTKD and AKT, by enhancing apoptotic induction with BIM and BID activation. Moreover, the dual blockade of RSK2 and AKT exerted robust molecular effects on critical gene sets associated with myeloma pathophysiologies, such as those with MYC, mTOR, STK33, ribosomal biogenesis, or cell-extrinsic stimuli of soluble factors, in HMCLs. These results provide the biological and molecular rationales for the dual-targeting strategy for RSK2 and AKT, which may overcome the therapeutic difficulty due to cytogenetic/molecular heterogeneity in MM.


Effects of BI-D1870 and Ipatasertib in 6 Human Myeloma Cell Lines (HMCLs)
We first looked at the baseline activation status of PDPK1, RSK2, AKT, and ERK1/2 in six HMCLs. PDPK1 and RSK2 were found to be activated through phosphorylation in all six HMCLs examined, while the activation statuses of ERK and AKT were variable among HMCLs. AKT was found to be active in NCI-H929, OPM-2, KMS-12-BM, and KMS-28-PE cells, while only faint expression of p-AKT Ser473 was detected in AMO-1 and RPMI8226 cells ( Figure 1A). Following that, we examined the growth inhibitory effects of BI-D1870, a specific inhibitor for RSK2-NTKD, and ipatasertib, an inhibitor for AKT, in six HMCLs. The treatment with BI-D1870 showed the dose-dependent growth inhibitory effect on all HMCLs examined ( Figure 1B). In contrast, we observed cellular context-dependent growth inhibition by ipatasertib in HMCLs studied; while ipatasertib exerted the dosedependent growth inhibitory effect on four HMCLs of NCI-H929, OPM-2, KMS-12-BM, and KMS-28-PE cells with active p-AKT Ser473 , it has no or only modest growth inhibitory effect on AMO-1 and RPMI8226 cells without active AKT ( Figure 1A,C). In addition, with these growth inhibitory effects, treatment with BI-D1870 repressed the phosphorylation of RSK2 Ser227 , while treatment with ipatasertib for 48 h exerted the optimal inhibition of AKT activity as shown by the dephosphorylation of PRAS40 Thr246 , which is the substrate of AKT. Unlike other cancers [33][34][35][36], both BI-D1870 and ipatasertib showed no cross-reaction to their respective target molecules. In addition, the combination of the two agents did not result in further inactivation of RSK2 and AKT compared with the effect induced by a single agent in the two HMCLs ( Figure 1D). Int. J. Mol. Sci. 2022, 23, x FOR PEER REVIEW 3 with these growth inhibitory effects, treatment with BI-D1870 repressed the phosph lation of RSK2 Ser227 , while treatment with ipatasertib for 48 h exerted the optimal inhib of AKT activity as shown by the dephosphorylation of PRAS40 Thr246 , which is the subs of AKT. Unlike other cancers [33][34][35][36], both BI-D1870 and ipatasertib showed no c reaction to their respective target molecules. In addition, the combination of the agents did not result in further inactivation of RSK2 and AKT compared with the e induced by a single agent in the two HMCLs ( Figure 1D).  The IC 50 values of BI-D1870 for  NCI-H929, OPM-2, KMS12-BM, KMS28-PE, AMO-1, and RPMI8226 cells  The combination of BI-D1870 and ipatasertib was then tested in four RSK2-NTKDand AKT-activated HMCLs. The normalized isobologram analysis revealed that the two agents had additive to synergistic growth inhibitory effects (Figure 2A), and also showed favorable dose-reduction indices (DRIs) of BI-D1870 and ipatasertib at most effect levels in their combination setting (Table S1). In addition, the combination of BI-D1870 and ipatasertib showed significantly more powerful growth suppression compared to the effect by either ipatasertib or BI-D1870 alone, not only in HMCLs, but also in primary myeloma cells from a patient who acquired resistance to PI, IMiDs, and anti-CD38 MoAb ( Figure 2B). Cell cycle analysis revealed that BI-1870 treatment increased the proportion of cells in the sub G1 fraction, implying the induction of apoptosis, whereas ipatasertib treatment blocked the G1/S transition without increasing cells in the sub G1 fraction. However, when ipatasertib was added to BI-D1870, there was significantly increased cells in sub G1 fraction compared to that induced by BI-D1870 alone ( Figure 2C). This was also confirmed by the significant increase in Annexin-V-positive cells, when ipatasertib was added to BI-D1870 compared to that induced by BI-D1870 alone ( Figure 2D,E). These findings suggest that dual targeting of RSK2-NTKD and AKT has a favorable combinatory anti-tumor effect in HMCLs when it is accompanied by apoptosis induction.

Mechanisms of Apoptosis Induced by the Combination of BI-D1870 and Ipatasertib in HMCLs
We next investigated the mechanisms underlying apoptosis induction by the combination of BI-D1870 and ipatasertib in HMCLs. Results of Caspase assays showed that, while ipatasertib did not cause the activation of caspases, the addition of ipatasertib onto BI-D1870 resulted in significant enhancement of the activities of Caspase-3/7, Caspase-8, and Caspase-9 compared to those induced by BI-D1870 alone ( Figure 3A). Western blot (WB) analyses also revealed the more prominent processing of Caspase-3, Caspase-8, and Caspase-9, and the resultant increase in cleaved PARP by the combination of the two agents compared to that induced by BI-D1870 alone ( Figures 3B and S1). These also implicated the involvement of both cell-intrinsic and extrinsic apoptotic pathways in the anti-tumor effect by the dual targeting of RSK2-NTKD and AKT.
With these results, we then investigated the expression changes in apoptosis regulators. As a result, the combination of BI-D1870 and ipatasertib resulted in the changes in two major pro-apoptotic BH3-only proteins, BIM and BID, as shown by the increased expression of three BIM isoforms and the emergence of truncated BID (t-BID) compared with those induced by either BI-D1870 or ipatasertib alone in two HMCLs examined ( Figures 3C and S1). BIM induction was at least partially mediated by transcriptional activation ( Figure 3D). However, in NCI-H929 cells, knocking down either BIM or BID did not completely prevent the induction of apoptosis by the combination of BI-D1870 and ipatasertib ( Figure S2). In addition, we found no prominent/meaningful expression change in other BH3-only proteins, such as BAD or PUMA, BAX, anti-apoptotic proteins, such as BCL-2, BCL-X L , and MCL-1, and series of molecules involved in death receptor-mediated apoptosis pathways, such as DR4, DR5, FAS, TNFR1, FLIP, and TRAIL ( Figures S1 and S3). These findings suggested the involvement of an unknown mechanism in apoptosis induced by the concomitant blockade of RSK2-NTKD and AKT.
blocked the G1/S transition without increasing cells in the sub G1 fraction. However, when ipatasertib was added to BI-D1870, there was significantly increased cells in sub G1 fraction compared to that induced by BI-D1870 alone ( Figure 2C). This was also confirmed by the significant increase in Annexin-V-positive cells, when ipatasertib was added to BI-D1870 compared to that induced by BI-D1870 alone ( Figure 2D,E). These findings suggest that dual targeting of RSK2-NTKD and AKT has a favorable combinatory anti-tumor effect in HMCLs when it is accompanied by apoptosis induction.

Mechanisms of Apoptosis Induced by the Combination of BI-D1870 and Ipatasertib in HMCLs
We next investigated the mechanisms underlying apoptosis induction by the combination of BI-D1870 and ipatasertib in HMCLs. Results of Caspase assays showed that, while ipatasertib did not cause the activation of caspases, the addition of ipatasertib onto  Expression levels relative to control (untreated cells) are shown below each band measured by densitometric analysis using Image-J software. ACTB was used as an internal control. (D) Transcriptional levels of BIM in treated cells relative to untreated cells. Mean value with S.D. of triplicate data. * p < 0.05, ** p < 0.01 *** p < 0.001, **** p < 0.0001.

Molecular Effects Induced by the Dual Blockade of RSK2-NTKD and AKT
To understand the molecular mechanism underlying the anti-MM effect by the dual blockade of RSK2-NTKD and AKT, the study performed gene expression profile (GEP)

Molecular Effects Induced by the Dual Blockade of RSK2-NTKD and AKT
To understand the molecular mechanism underlying the anti-MM effect by the dual blockade of RSK2-NTKD and AKT, the study performed gene expression profile (GEP) analysis to comprehensively analyze commonly modulated genes by the combinatory treatment with BI-D1870 and ipatasertib in comparison with those modulated either by BI-D8170 or ipatasertib alone in NCI-H929 and OPM-2 cells (Table S2). Overall, the correlation coefficients for gene expression changes between NCI-H929 and OPM-2 cells were 0.23, 0.16, and 0.30 after the treatment by BI-D1870, ipatasertib, and the combination of two agents, respectively, indicating the existence of weak correlation only with the combination of treatment, but not with single agents between the two HMCLs ( Figure 4A). However, as shown in Figure 4B, 144, and 78, genes were found to be commonly upregulated by BI-D1870 and ipatasertib respectively, while 62 and 72 genes were commonly downregulated by BI-D1870, and ipatasertib, in the two HMCLs. There were only 38 upregulated and 19 downregulated genes that overlap between genes modulated by the sole treatment of BI-D1870 or ipatasertib. It also noted that, 176 genes were commonly upregulated, while 194 genes were commonly downregulated by the combination of BI-D1870 and ipatasertib in the two HMCLs. Importantly, 94 upregulated genes and 135 downregulated genes were significantly modulated only when HMCLs were exposed to combinatory treatment, indicating the occurrence of more molecular changes only when HMCLs were exposed to the combinatory treatment of BI-D1870 and ipatasertib ( Figure 4B). While the overall molecular response to the inhibition of RSK2-NTKD, AKT, and dual blockade of both kinases is variable and cellular context-dependent at the transcriptional level, the findings suggest the induction of more common molecular impact by the dual blockade of RSK2 and AKT compared to that induced by the blockade of either RSK2 or AKT alone. We also validated the GEP results by adding quantitative reverse transcription-polymerase chain reaction (qRT-PCR) for the expression of several randomly selected genes ( Figure S4 and Table S2).
Then, we used gene set enrichment analysis (GSEA) to compare gene expression changes caused by RSK2-NTKD and AKT blockade to changes caused by RSK2 or AKT blockade alone. In brief, we first performed gene clustering in NCI-H929 and OPM-2 cells using BI-D1870, ipatasertib, and their combination treatments ( Figure 4C and Table 1).
As judged by the false discovery rate q-value, various gene sets were found to be more significantly modulated by the combination of BI-D1870 and ipatasertib, compared with the treatment by either BI-D1870 or ipatasertib alone. Significant gene sets downregulated by the combination of two agents (Cluster C1 in Figure 4C; Table 1) included those for various oncogenic signatures, such as gene sets driven by upregulation of MYC and mTOR, and re-exposure to serum, interleukin (IL)-2 or IL-15 after starvation, and gene sets inactivated by gene knockdown of RPS14, the blockade of vascular endothelial growth factor (VEGF), and cellular differentiation, among others ( Figure 5A,B; Table 2). Significant gene sets upregulated by the combination of two agents (Cluster C5 in Figure 4C; Table 1) Table 2). Indeed, the co-treatment of BI-D1870 and ipatasertib caused the moderate inhibition of phospho-mTOR and the reduction in c-MYC expression ( Figure 6). ase chain reaction (qRT-PCR) for the expression of several randomly selected genes (Figure S4 and Table S2).
Then, we used gene set enrichment analysis (GSEA) to compare gene expression changes caused by RSK2-NTKD and AKT blockade to changes caused by RSK2 or AKT blockade alone. In brief, we first performed gene clustering in NCI-H929 and OPM-2 cells using BI-D1870, ipatasertib, and their combination treatments ( Figure 4C and    Figure 4C. Figure 4C Gene    Figure 4C. Figure 4C Gene sets inactivated by gene knockdown of RPS14, the blockade of vascular endothelial growth factor (VEGF), and cellular differentiation, among others ( Figure 5A,B; Table 2). Significant gene sets upregulated by the combination of two agents (Cluster C5 in Figure  4C; Table 1) included those downregulated by mTOR and MYC and upregulated by the knockdown of STK33, a pro-tumor serine-threonine kinase (Figure 5C,D; Table 2). Indeed, the co-treatment of BI-D1870 and ipatasertib caused the moderate inhibition of phospho-mTOR and the reduction in c-MYC expression ( Figure 6).   cantly enriched by the combination treatment of BI-D1870 and ipatasertib compared to the treatment by either BI-D1870 or ipatasertib alone. Significance was shown by FDR(q-value) scores. -log10(FDR-q-value) score bigger than 1.3 is significant. Figure 6. WB for MYC, mTOR, and p-mTOR: Cells were treated for 48 h at the indicated concentrations of agents. Expression levels relative to control (untreated cells) are shown below each band measured by densitometric analysis using Image-J software. ACTB was used as an internal control.  Figure 6. WB for MYC, mTOR, and p-mTOR: Cells were treated for 48 h at the indicated concentrations of agents. Expression levels relative to control (untreated cells) are shown below each band measured by densitometric analysis using Image-J software. ACTB was used as an internal control.

Discussion
RAS/RAF/ERK and PI3K/AKT pathways are frequently activated by both various cell-intrinsic oncogenic events and extrinsic stimuli, and they play critical roles in myeloma pathophysiology. However, the clinical development of pathway-directed therapeutics targeting either RAS, RAF, PI3K, or AKT as the single target has failed in yielding results until today [37,38]. We are of the opinion that, although RSK2 is the bottleneck output hub of RAS/RAF/ERK pathway [39,40], the therapeutic effect of single targeting of RAS or RAF might be hampered by the PDPK1-mediated constitutive active status of RSK2, which is independent of RAS/RAF/ERK activity in myeloma cells, while the blockade of AKT alone has only cytostatic effects but no major cell killing effect, as was found in this study in myeloma cells.
In this study, we discovered that the combination of BI-D1870 and ipatasertib had a synergistic antiproliferative effect in HMCLs with active RSK2-NTKD and AKT. Although ipatasertib alone is not effective in inducing apoptosis in HMCLs, the addition of ipatasertib onto BI-D1870 significantly augmented the cell death-inducing effect with apoptosis, which was accompanied by the enhanced activation of both cell-intrinsic and -extrinsic apoptosis pathways. Indeed, dual blockade of RSK2 and AKT enhanced the transcriptional upregulation of BIM in HMCLs examined, those were consistent with other cancerous diseases [41]. The molecular effect by the combination of BI-D1870 and ipatasertib presumably promoted Casease-9 activation and led to the resultant activation of the cell-intrinsic apoptosis pathway. In addition, BI-D1870 treatment caused the processing of Caspase-8 to its cleaved form in HMCLs, which was consistent with a previous report showing that the blockade of RSK2 abrogates Caspase-8 stability [42]. However, it is unclear how the addition of ipatasertib onto BI-D1870 increased Caspase-8 activity remains to be verified. Indeed, Caspase-8 activation was not observed by ipatasertib treatment alone. We hypothesize that Caspase-3 may be involved in Caspase-8 feedback activity to promote proteolytic processing [43].
At the molecular level, no significant correlation in gene expression changes after RSK2 or AKT blockade was observed between the two HMCLs studied, which may reflect molecular heterogeneity between the two HMCLs studied. Despite this, a positive correlation of gene expression changes emerged following the combinatory blockade of the two molecules in the same two HMCLs. Moreover, the combinatory blockade of RSK2 and AKT caused gene expression changes in a larger number of genes, compared to those induced by the blockade of either RSK2 or AKT alone. Thus, the combining blockade of RSK2 and AKT has more robust molecular effects on HMCLs, raising the expectation for the universal efficacy on myeloma cells of diverse molecular features from different patients.
With this setting, GSEA disclosed that, in comparison with the sole blockade of either RSK2 or AKT, the combination of blockade in the two kinases proved more powerful effects on several critical gene sets involved in myeloma pathophysiologies, such as those associated with MYC, mTOR, STK33, ribosomal biogenesis, or cell-extrinsic stimuli of soluble factors. MYC is a multifunctional oncogenic transcription factor that affects various cell biological processes, including cell proliferation, apoptosis, and DNA damage in various cancerous diseases, including myeloma [44,45]. Furthermore, MYC is frequently overexpressed through various overlapping mechanisms, such as cytogenetic, epigenetic, and IRF4-mediated transcriptional mechanisms, in myeloma cells, and plays critical roles in cell survival and proliferation of myeloma cells, thereby constituting the fundamental process in both the development and progression of MM [45][46][47][48]. The therapeutic efficacy of IMiDs, such as lenalidomide and pomalidomide, has been known to be partly mediated by the repression of MYC through the blockade of IKZF family proteins and ARID2 [49,50]. Given a serious prognostic consequence of acquiring of the resistance to IMiDs-containing therapy in MM treatment, it is expected that the dual blockade of RSK2 and AKT constitutes a new strategy that potently overcomes IMiDs resistance in MM. mTOR activation by cell-extrinsic stimuli, such as by IL-6, VEGF, and insulin-like growth factor-1, also contribute to the disease progression of MM by promoting cell survival, cell metabolism, the cross-talk with the ubiquitin-proteasome system, and angiogenesis [51][52][53][54]. With this regard, it was reasonable to note that the dual blockade of RSK2 and AKT can also cause the downregulation of gene sets associated with cell-extrinsic stimuli (serum, IL-2, IL-15, and VEGF). Given that the inhibition of mTORC2 has the synergistic anti-MM effects with a proteasome inhibitor and IMiD [51,52], the dual blockade of RSK2 and AKT may be expected to enhance the effects of proteasome inhibitors and IMiDs on myeloma cells. It was also appealing that the simultaneous inhibition of RSK2 and AKT causes the downregulation of genes associated with RPS14, which is an essential component of ribosomal biogenesis. Although the pathologic role of abnormal expression of RSP14 has not been known in MM, the blockade of RSP14 has been shown to exhibit a tumor-suppressive role regardless of TP53 status in solid and hematologic cancers [55]. STK33 has been known to participate in the transcriptional activation of c-MYC [56], and also contributes to the angiogenetic process as the mediator of hypoxia-inducible factor-1α/VEGF signaling [57]. Because the functional involvement of STK33 in MM has been unknown, more research is needed to clarify its role.
Despite the remarkable improvement of prognosis of patients with MM by the advent of combinatory therapies of PIs, IMiDs, and monoclonal antibodies, MM still remains to be mostly incurable. Therefore, the therapeutic strategy that can overcome the resistance to the combination of the three class agents is an unmet medical need in MM. Reportedly, the incidence of mutations within the RAS/ERK pathways is significantly increased in MM refractory to PIs and IMiDs [58]. Mechanistically, oncogenic activation of RAS/RAF/ERK signaling is one of the mechanisms that enhance proteasome capacity, while treatment with PI causes the phosphorylation of AKT [59,60], and both mechanisms contribute to the resistance to proteasome inhibition in myeloma cells. AKT-mediated sustained eIF4E expression and C/EBP translation have been proposed as the resistant mechanisms for IMiDs [61]. Moreover, the resistant mechanisms for current standard therapeutics for MM include aberrant MYC overexpression [50], and the increased rate of protein translation [62]. Considering these, one of the clinical positions of the dual-targeting therapy for RSK2 and AKT may be theoretically the salvage setting after treatment failure of PIs, and/or IMiDs with or without monoclonal antibodies for CD38 or SLAMF-7. Unfortunately, we were not able to examine the in vivo anti-myeloma effect of the combination of BI-D1870 and ipatasertib, due to the uncertain in vivo performance of BI-D1870, and also due to the lack of a promising in vivo bioavailable inhibitor specific for RSK2-NTKD. This was the limitation of the current study, which urges the generation of in vivo bioavailable inhibitor for RSK2-NTKD. To implement this strategy in a clinical environment, the clinical development of bioavailable dual inhibitors for RSK2 and AKT is expected for MM [63].
In conclusion, our study revealed that the dual blockade of RSK2 and AKT exhibits anti-tumor effects in myeloma cells via the regulation of multidirectional cell-intrinsic and extrinsic molecular mechanisms for various critical molecules and biologic processes in myeloma development and progression, and, thus, is an attractive candidate of new therapeutic strategy for MM with diverse molecular backgrounds.

Cell Proliferation Assay
Cells were seeded in a flat-bottomed 96-well plate at 2.0 × 10 5 cells/mL in 200 µL medium per well. After 48 h of treatment with various concentrations of BI-D1870, ipatasertib, or their combinations for the absolute number of viable cells was analyzed with Cell Counting Kit-8 (CCK-8) assay (Dojindo Molecular Technologies, Kamimashiki, Kumamoto, Japan) using a SpectraMax iD3 multi-mode microplate reader (Molecular Devices, San Jose, CA, USA). Combination index values and DRIs were analyzed using CalcuSyn software (Biosoft, Cambridge, Cambridgeshire, UK). Synergism is defined as more than the expected additive effect with CI < 1. DRI = 1 indicates no dose reduction, whereas DRI > 1 and <1 indicate favorable and unfavorable dose-reduction, respectively [66].

Analysis of Cell Cycle and Detection of Apoptosis
After 48 h of treatment with an agent(s), cells were washed with phosphate buffer saline (PBS), treated with PBS containing 0.1% Triton X-100, and stained with propidium iodide (PI) (Sigma Aldrich, St. Louis, MN, USA). The DNA content was measured using FACSCalibur (Becton Dickinson, Franklin Lakes, NJ, USA). Cells were counterstained with Annexin-V-fluorescein isothiocyanate and PI using the Annexin-V-FLUOS Staining Kit (Roche, Basel, Basel-Stadt, Switzerland) for apoptosis analysis, and flow cytometric analysis was performed. The data were analyzed using the FLOWJO software (Tomy Digital Biology, Tokyo, Japan).

qRT-PCR
The RNeasy Mini Kit was used to isolate total RNA from NCI-H929 and OPM-2 cells (QIAGEN, Hilden, North Rhine-Westphalia, Germany). One microgram of total RNA was reverse transcribed to cDNA using a High-Capacity Reverse Transcription Kit (Applied Biosystems, Foster City, CA, USA). qRT-PCR was carried out using a QuantStudio 3 system (Applied Biosystems) to quantify the expression level of BIM. Real-time RT-PCR primer probes (BIM; Hs00708019_s1, ACTB; Hs01060665_g1) were purchased from Applied Biosystems. qRT-PCR for YPEL3, JUND, APAF1 and CDKN1B was performed using Fast SYBR Green Master Mix with a StepOne Plus instrument (Applied Biosystems). Primers were purchased from Hokkaido Systems Science Co., Ltd. (Sapporo, Japan) ( Table S3). The transcriptional levels of target genes were adjusted following the level of ACTB.

Microarray Analysis and GSEA
After treating NCI-H929 and OPM-2 cells with ipatasertib and/or BI-D1870 at their IC50s for 24 h, total RNA was isolated using mirVana TM miRNA Isolation Kit (Invitrogen, Waltham, MA, USA). GEP was analyzed with a Clariom S array (Affymetrix, Santa Clara, CA, USA), a GeneChip WT Plus Reagent Kit (Thermo Fisher Scientific, Waltham, MA, USA), and a GeneChip Scanner 7G (Affymetrix). The robust multichip array method was used to normalize the expression data. Differentially expressed genes (DEGs) were identified using one-way analysis of variance and post hoc Tukey's honestly significant difference test after the fact. Adjusted p values for multiple comparisons were considered significant if they were less than under 0.05. Hierarchical clustering among DEGs was performed using Euclidean distance and Ward's method. GSEA was performed using the software GSEA and the oncogenic signature gene sets from Molecular Signature Database (MSigDB) with default parameters (https://www.gsea-msigdb.org, accessed on 1 December 2021) [32,67].

Statistical Analysis
Statistical analyses were carried out using EZR (Saitama Medical Center, Jichi Medical University, Omiya, Saitama, Japan), a graphical user interface for R (The R Foundation for Statistical Computing, Vienna, Austria). The student's t-test was used to compare continuous variables between groups, and a p-value less than 0.05 was considered significant.  Data Availability Statement: Any data or material that support the findings in this study can be made available by the corresponding author upon request.