Withaferin A was purchased from Enzo Life Sciences (Lausen, Switzerland). Bortezomib, MTT, XTT, calcein, EH, VRP, and R123 were purchased from Sigma–Aldrich (Buchs, Switzerland).

Dried and powdered roots of W. somnifera (1.6 kg) were extracted three times with CH₂Cl₂/MeOH (1:1, 4 L) (Xu et al., 2009). Extracts were combined and evaporated under reduced pressure to afford the crude extract (108.5 g) which was partitioned between 80% aq. MeOH and hexanes. The 80% aq. MeOH fraction which contained withanolides was diluted to 50% aq. MeOH with water and extracted with CHCl₃. Evaporation of CHCl₃ under reduced pressure gave an extract (22.1 g) enriched with withanolides. A portion (6.0 g) of this extract was subjected to size-exclusion chromatography over a column of Sephadex LH-20 (160.0 g) made up in hexanes/CH₂Cl₂ (1:4) and eluted with hexanes/CH₂Cl₂ (1:4, 1.5 L), CH₂Cl₂/acetone (3:2, 1.0 L), CH₂Cl₂/MeOH (1:1, 1.0 L), and finally with MeOH (0.5 L). Thirty fractions (125 mL each) were collected and combined based on their TLC profiles to yield 18 combined fractions [A (1133.8 mg), B (859.3 mg), C (334.6 mg), D (212.5 mg), E (134.4 mg), F (140.9 mg), G (182.4 mg), H (428.4 mg), I (280.9 mg), J (154.7 mg), K (43.1 mg), L (16.7 mg), M (65.8 mg), N (120.5 mg), O (350.8 mg), P (246.2 mg), Q (512.4 mg), and R (620.3 mg)]. Fraction G which contained the majority of withanolides was subjected to column chromatography over silica gel (3.5 g) made up in Et₂O and eluted with Et₂O followed by increasing amounts of MeOH in Et₂O to give seven sub-fractions [G1 (3.4 mg), G2 (3.0 mg), G3 (1.5 mg), G4 (26.1 mg), G5 (82.5 mg), G6 (48.5 mg), and G7 (10.1 mg)]. The sub-fraction G4 on trituration with Et₂O afforded WND as a white solid (18.2 mg); [α]D25 +78 (c 0.2, MeOH) [liter +80] (Lavie et al., 1968). The ¹H NMR, ¹³C NMR, and mass spectroscopic data were consistent with those reported (Roumy et al., 2010).

Multiple myeloma cancer stem cells were purchased from Celprogen (Torrance, CA, United States). They were isolated from a bone marrow lesion of a MM patient, and are CD44⁺, CD166⁺ and CD138^- (Medema, 2013; Issa et al., 2016; Issa and Cuendet, 2017). The tumorigenicity of MM-CSCs, as determined by Celprogen, is <1000 cells (Medema, 2013; Issa et al., 2016). MM-CSCs were maintained in the hematopoietic stem cell medium StemPRO-34 (Life Technologies, Zug, Switzerland), supplemented with 10% FBS (Nuaillé, France), 100 IU/mL penicillin and 250 μg/mL streptomycin (Life Technologies), and 2 mM L-glutamine. MM-CSCs were utilized at passages between 4 and 12 to preserve a stem cell state. The tumoral PCs RPMI 8226 (RPMIs), the DEX-sensitive MM1.S and the DEX-resistant MM1.R were obtained from LGC standards (Middlesex, United Kingdom), and were maintained in RPMI-1640 medium supplemented with 10% FBS, 100 IU/mL penicillin, and 250 μg/mL streptomycin. RPMIs are mainly composed of CD138⁺ cells, but have a fraction of 2–5% CD138^- cells known to be tumorigenic (Matsui et al., 2004). Normal hematopoietic CD34⁺ stem cells (NSCs) were purchased from Life Technologies, and maintained in StemPRO-34 medium, supplemented with Cytokine Mix E (PromoCell, Heidelberg, Germany) and 2 mM L-glutamine. NSCs were utilized at passages between 3 and 6 to preserve a stem cell state. MM-CSCs, NSCs, RPMIs, MM1.S, and MM1.R cells were passaged every 2–3 days, and were maintained in a humidified atmosphere supplemented with 5% CO₂ at 37°C (standard conditions).

Multiple myeloma cancer stem cells were plated in 96-well plates at a density of 5,000 cells per well and allowed to adhere for 24 h, whereas RPMIs, MM1.S, and MM1.R cells were plated at a density of 15,000 cells per well and treated immediately with WND or positive controls. After 72 h incubation, cells were incubated for 2 h with 20 μL of MTT solution (5 mg/mL) for MM-CSCs, or for 4 h with 50 μL XTT solution (1 mg/mL) for RPMIs, MM1.S, and MM1.R cells. Cell growth was computed as the absorbance of each test well divided by that of the vehicle control wells and then multiplied by 100. The concentration that resulted in 50% reduction in cell growth (IC₅₀) was computed from at least five different concentrations using GraphPad Prism 5.0 software.

Multiple myeloma cancer stem cells were plated in 6-well plates at a density of 100,000 cells per well and allowed to adhere for 24 h before WND treatment at which about 300,000 cells were present in the well. RPMIs and NSCs were seeded at 300,000 cells per well in 6-well plates and were treated immediately with increasing concentrations of WND for 24 h. Flow cytometric analyses were performed on MM-CSCs, RPMIs, and NSCs to examine the effects of WND on cell viability/mortality using calcein staining for viability and EH staining for mortality. For apoptosis, MM-CSCs and RPMIs were assessed after 24 h WND treatment using the AV/PI staining (Life Technologies). Following treatment with WND, floating and adherent cells were collected, washed with PBS, adjusted to 10⁶ cells/mL, and then incubated with the corresponding stain at room temperature for 15 min. Cells were then analyzed using the Attune NxT Acoustic Focusing Flow Cytometer (Applied Biosystems, Life Technologies, Zug, Switzerland).

To determine whether or not WND was an inhibitor of P-gp efflux transport, R123 accumulation assays were performed. MM-CSCs were plated at a density of 15,000 cells per well in 96-well plates, and were allowed to adhere overnight. MM-CSCs were washed with PBS and incubated in 250 μL of 10 μM R123 solution with or without WND or controls for 30 min at 37°C. MM-CSCs were then washed once with PBS and incubated with 50 μL of 1 μL/mg DAPI stain. VRP was used as positive control inhibitor of P-gp efflux transport. MM-CSCs were then washed once with PBS and intracellular R123 fluorescence was immediately quantified using the Cytation 3 Cell Imaging Reader (Biotek, Winooski, VT, United States). The R123 fluorescence was normalized to DAPI nuclear stain. The R123 levels in drug-treated MM-CSCs were compared to that in vehicle control. The average normalized R123 accumulation was multiplied by 100 and divided by the average R123 accumulation vehicle control-treated MM-CSCs.

Multiple myeloma cancer stem cells, NSCs, and RPMIs were lysed, and total RNA was collected from cell lysates with or without WND treatment using the Aurum total RNA Mini Kit (Bio-Rad, Cressier, Switzerland). Reverse transcription was conducted using 0.5 μg RNA, the “high capacity total RNA to cDNA” Reverse Transcriptase (Life Technologies) and a PCR standard thermal cycler (Applied Biosystems). One microliter cDNA was amplified by quantitative PCR using a SYBR Green PCR Kit (Applied Biosystems) and a Step One Plus Real-Time PCR Thermal Cycler (Applied Biosystems). Custom primers for GAPDH, 18S, and P-gp were designed on http://bioinfo.ut.ee/primer3-0.4.0/ and obtained from Life Technologies (Koressaar and Remm, 2007; Untergasser et al., 2012). Relative gene expression was calculated by normalizing to the average of two housekeeping genes 18S and GAPDH using the ΔΔC_T method.

Each experiment was conducted at least three times, and all data were expressed as mean ± SEM. An unpaired t-test was employed for statistical comparisons that involved two groups. A one-way ANOVA was employed for multiple comparisons in experiments with one independent variable, and a Bonferroni test was used for post hoc analysis. A significant difference in mean values between groups was considered when p < 0.05.

Using MTT and XTT assays, the effect of WND on cell growth in MM-CSCs, RPMIs, MM1.S, and MM1.R cells was investigated. Results indicated that a 72 h WND treatment inhibited MM-CSCs growth with an IC₅₀ value of 88.4 ± 13.9 nM, and that of RPMIs with an IC₅₀ of 76.1 ± 8.0 nM (Table 1). The IC₅₀ values of WND in MM1.S and MM1.R cells were 107.2 ± 14.2 nM and 106.3 ± 11.0 nM, respectively (Table 1). Statistical analysis indicated that the IC₅₀ values of WND were similar between MM-CSCs and RPMIs, and between MM1.S and MM1.R cells. Results also showed that the IC₅₀ values of WFA were significantly higher in the resistant cells, MM-CSCs and MM1.R, than in the sensitive ones RPMIs and MM1.S (Table 1) (Issa and Cuendet, 2017). The behavior of BTZ, a control drug, was similar to that of WFA. This shows the resistant phenotype in MM-CSCs and MM1.R cells (Table 1).

The cytotoxic effect of a 24 h treatment with WND was measured by flow cytometric analyses of calcein/EH staining in MM-CSCs and RPMIs (Figures 2A,B). The percentage of viable (calcein⁺/EH^-) MM-CSCs was significantly reduced from 97.7% with vehicle control treatment to 67.9 and 49.4% with 600 and 800 nM WND, respectively. The percentage of dying (calcein⁺/EH⁺) MM-CSCs significantly increased starting at 400 nM WND, and that of dead (calcein^-/EH⁺) MM-CSCs significantly increased starting at 600 nM WND. Similarly, RPMIs viability was reduced from 80.9% with vehicle control to 67.2, 41.2, 35.5, and 29.4% with 200, 400, 600, and 800 nM WND, respectively. The percentage of dying RPMIs did not appear to significantly change with increasing doses of WND; however, the percentage of dead RPMIs significantly increased starting at 200 nM.

The effect of WND on CD34⁺ hematopoietic NSCs viability was assessed to determine if WND exhibited any degree of selectivity toward MM-CSCs. The IC₅₀ value of WND with respect to viability (calcein⁺ cells) was calculated in MM-CSCs and NSCs after a 72 h treatment with WND. The IC₅₀ value was 0.28 ± 0.06 μM in MM-CSCs and greater than 0.8 μM in NSCs, suggesting a minimum of 2.75-fold selectivity. The cytotoxic effect on MM-CSCs and RPMIs prompted us to investigate whether the mechanism of cell death in response to WND was due to apoptosis. Flow cytometric analyses of AV/PI staining showed significant apoptotic effects in both MM-CSCs and RPMIs in response to a 24 h WND treatment (Figures 2C,D). The percentage of viable cells (AV⁺/PI^-) was significantly reduced with 800 nM WND in MM-CSCs, and starting with 200 nM WND in RPMIs. The percentage of apoptotic cells (AV⁺/PI⁺ and AV^-/PI⁺) was significantly increased with 800 nM WND in MM-CSCs, and starting with 400 nM WND in RPMIs.

Next, the mRNA expression of P-gp in MM-CSCs and in RPMIs was compared (Figure 3A). Results showed that P-gp expression was 12.5-fold lower in RPMIs than in MM-CSCs. In addition, a 24 h WND treatment in MM-CSCs did not affect P-gp mRNA expression at the tested doses (Figure 3B). The observation that MM-CSCs are resistant to the cytostatic effects of BTZ and WFA, but not to WND, combined with the observation that MM-CSCs highly expresses P-gp in comparison to RPMIs, led to investigate whether the antiproliferative activity of BTZ, WFA, and WND was dependent on the function of P-gp efflux transporter. The ability of VRP, a known P-gp inhibitor, to sensitize MM-CSCs to BTZ, WFA, and WND was then measured. A 5 μM VRP treatment sensitized MM-CSCs to both BTZ and WFA, but not to WND (Table 2). This was evidenced by the significant reduction in IC₅₀ values with both WFA and BTZ after VRP treatment. In contrast, the IC₅₀ value of WND was similar with or without VRP (Table 2). In RPMIs, which express substantially lower P-gp mRNA levels, the use of VRP did not significantly lower the IC₅₀ values (Table 2).

The intracellular accumulation of R123, a substrate of P-gp transporter, was measured in the P-gp-highly expressing MM-CSCs and was found to significantly increase in a dose-dependent manner in MM-CSCs treated with VRP, starting at a dose of 1.25 μM (Figure 4A). Similarly, WFA caused a significant dose-dependent increase in R123 accumulation starting at 0.62 μM (Figure 4B). WND did not appear to affect the R123 intracellular accumulation up until 400 nM. However, a 600 nM WND treatment significantly increased R123 intracellular accumulation (Figure 4C). BTZ, a molecule previously shown to be affected by P-gp-mediated resistance (O’Connor et al., 2013), did not affect R123 accumulation at the tested doses (Figure 4D).