Parthenolide Treatment Results in a Dose-Dependent Decrease in Lymphoma Cell Viability
To determine the effect of parthenolide on mantle cell lymphoma (MCL) and diffuse large B-cell lymphoma (DLCL) viability, MTT assays were performed after 24 h exposure of MCL and DLCL cells to increasing doses of parthenolide. As shown in , the viability of Granta, Rec-1 and HF4B cells (MCL cell lines) decreased in a dose-dependent fashion after exposure to parthenolide, with an EC50
of 7.5±0.2, 5.3±0.2 and 5.2±0.3 µM, respectively. Parthenolide similarly decreased the viability of the DLCL cell lines SUD-HL6 and OCI-Ly19, with an EC50
of 9.3±0.4 and 7.3±0.3 µM, respectively, (). Parthenolide also had a potent dose-dependent effect on the viability of DLCL cells cultured directly from a patient biopsy specimen (DLCL 27B) having an EC50
3.0±0.11 µM (). In our subsequent studies, we used the Granta MCL line to elucidate the mechanism of parthenolide action.
Parthenolide treatment results in a dose-dependent decrease in lymphoma cell viability.
Parthenolide Attenuates the Levels of Exofacial Free Thiols as Assessed by Flow Cytometry
To directly assess the effect of parthenolide on Granta cell exofacial thiols we used an Alexa Fluor 633 coupled maleimide compound (ALM). Maleimide is a nucleophile that covalently binds to free thiol groups, and when coupled with the charged Alexa Fluor 633 dye becomes cell impermeable, allowing for the examination of cell surface thiol levels. As shown in , exposure of Granta cells to 30 µM parthenolide for 3 h resulted in a decrease in exofacial free thiols by approximately 36% relative to that of vehicle treated control cells. In contrast, treatment with cell-impermeable glutathione (GSH) increased the levels of exofacial free thiols by approximately 49% (). When cells were pretreated with GSH for 2 h, washed and exposed to 30 µM parthenolide for 3 h there was no parthenolide induced decrease in exofacial free thiol levels (). Pretreatment with the cell permeable monoethyl ester (GSHee) or the GSH precursor N-acetyl-L-cysteine (NAC) similarly protected exofacial thiols from modification by parthenolide (data not shown). The histogram shown in is representative of 3 independent experiments. The MFI (mean fluorescent intensity) of the control cells from all 5 experiments varied only by +/− 8%, and the direction of each experimental result in relationship to that of the control cells was the same in each experiment. Taken together, the data suggest that parthenolide modulates exofacial free thiol groups; a process that is inhibited by pretreatment with GSH.
Parthenolide attenuates the levels of exofacial free thiols as assessed by flow cytometry.
To control for the possibility that GSH pretreatment alters intracellular redox, which would result in an increase in surface free thiols, we evaluated the effect of GSH on parthenolide induced ROS generation (Supplementary Fig. S1
). Whereas pretreatment for 2 h with 5 mM of cell permeable GSHee or NAC partially inhibited parthenolide induced ROS generation, a 2 h pretreatment with 5 mM GSH had no effect on intracellular ROS generation. Similar results were seen using parthenolide at doses of 5 and 10 µM for 15, 30 and 60 min of exposure (data not shown). If GSH was altering intracellular redox it should have attenuated parthenolide induced ROS generation similar to that seen with GSHee and NAC. Taken together, these data suggest that the effect of GSH on exofacial free thiols is a direct effect and not due to an effect on intracellular redox.
To control for the possibility that exofacial thiol modification is a result of intracellular H2
, generated by parthenolide, which would be anticipated to diffuse across the cell membrane and directly modify exofacial thiols, we used the cell impermeable H2
scavenger, catalase. Exposure of Granta cells to 5 mM H2
resulted in exofacial protein oxidation, (as demonstrated by Alexa Fluor staining; Supplemental Fig. S2A
), which is blocked by 500 µM catalase. In contrast, exposure of Granta cells to 30 µM parthenolide results in exofacial protein oxidation, which is not blocked by catalase (Supplemental Fig. S2B
). Taken together, parthenolide is not modifying exofacial thiols through an H2
Parthenolide Attenuates the Levels of Exofacial Free Thiols as Assessed by Western Blot Analysis
To further confirm that parthenolide modifies exofacial free thiols, we used a biotinylated thiol reactive reagent, N-(biotinoyl)-N-(iodoacetyl) ethylendiamine (BIAM), which binds to free thiol groups and is detected by streptavidin-peroxidase staining of western blots. As shown in , lane 1, plasma membrane enriched cell lysates from untreated Granta cells contain significant numbers of proteins having free thiol groups (with each band corresponding to a protein having a free thiol group able to interact with BIAM). After exposure of Granta cells to 10 (, lane 2) or 30 µM (, lane 3) parthenolide for 3 h, however, there is a dose dependent decrease in the intensity of the bands corresponding to ~22kD and ~12kD suggesting that parthenolide is modifying the free thiols of these proteins, making them unreactive to BIAM.
Parthenolide attenuates the levels of exofacial free thiols as assessed by western blot analysis.
We next determined whether the cell impermeable GSH attenuated the effect of parthenolide on the free thiols of the ~22 and~12kD proteins. Treatment of Granta cells with 5 mM GSH alone (, lane 4) had no significant effect on the intensity of the bands corresponding to the ~22 and ~12kD proteins as compared to that seen in untreated control cells (, lane 1). In contrast, a 2 h pre-treatment with 5 mM GSH followed by exposure to 30 µM parthenolide for 3 h significantly attenuates the decrease in the intensity (and thus modification) of the bands corresponding to the ~22 and ~12kD proteins seen with a 3 h exposure of 30 µM parthenolide alone (compare lane 3, parthenolide alone, with lane 5, GSH plus parthenolide).
As a control for the oxidation of protein free thiols, H2
was used. Exposure of Granta cells to H2
resulted in decreased signals in multiple bands that included the ~22kD and ~12kD bands (, lane 6
) modified by parthenolide. Equal protein loading in each lane was demonstrated with actin staining. Membrane purity is shown by blotting with an anti-transferrin antibody (Supplementary Fig. S3
). Taken together with the flow cytometric data presented in , parthenolide modulates exofacial free thiols, including those of proteins having molecular weights of ~22kD and ~12kD, and such modifications are blocked by pretreatment with GSH.
Identification of Surface Thioredoxin-1 as One of the Targets of Parthenolide
We hypothesized that one of the candidate proteins modified by parthenolide was thioredoxin-1, a 12kD protein containing thiol groups amenable to parthenolide modification. We first evaluated whether thioredoxin-1 can be detected on the Granta cell surface. Granta cells were stained with an anti-thioredoxin antibody and analyzed by flow cytometry (). The mean fluorescent intensity (MFI) of staining with anti-thioredoxin antibody is approximately 85% greater than that of the secondary antibody alone, confirming that thioredoxin is on the Granta cell surface.
Identification of surface thioredoxin-1 as one of the targets for parthenolide.
We next determined whether parthenolide could directly interact with free thiol groups within purified human thioredoxin-1. Thioredoxin-1 was co-incubated with maleimide conjugated to a polyethylene glycol spacer having terminal methyl groups (MM(PEG)24). Maleimide binds to free thiols on thioredoxin and upon SDS-PAGE separation and blotting with an anti-thioredoxin antibody gives a characteristic ladder pattern which is a mixture of thioredoxin with different molecular weights due to MM(PEG)24 conjugation to ≥1 of the 4 free thiols of thioredoxin (, lane 2). However, when thioredoxin was pre-incubated with different concentrations of parthenolide (ranging from 5 to 50 µM), the sites available for MM(PEG)24 disappeared in a dose dependent manner (as evidenced by a decrease in laddering; , lanes 3−5), and were completely abolished in the presence of 50 µM parthenolide (, lane 6). Taken together, parthenolide directly binds to free thiol groups on thioredoxin-1. As expected, co-incubation of thioredoxin-1 with the thiol alkylator, N-ethylmaleimide (NEM), also reduced the sites available for MM(PEG)24 binding thus decreasing thioredoxin laddering (, lane 7).
Finally, we determined whether parthenolide was directly modifying membrane associated thioredoxin-1 in Granta cells by BIAM staining and neutravidin pull-down. As shown in , lane 1, one of the biotinylated proteins pulled down by neutravidin is thioredoxin. However, when Granta cells were treated with 30 µM parthenolide for 3 h, and then treated with BIAM, none of the proteins pulled down by neutravidin reacted with the anti-thioredoxin antibody (, lane 2), suggesting that parthenolide has modified thioredoxin free thiols, making them unavailable to BIAM binding and thus they are not pulled-down by neutravidin. In contrast, when cells were first treated with 5 mM GSH for 2 h, and then exposed to 30 µM parthenolide for 3 h, the pulled down proteins were recognized by the anti-thioredoxin antibody (, lane 4), suggesting that the free thiol groups were protected from parthenolide (via an as yet unknown mechanism), and thus available for interaction with BIAM. Lastly, GSH treatment alone had no effect on the thioredoxin pull-down (, lane 3).
To exclude the possibility that the effect of parthenolide on thioredoxin free thiols was an indirect effect, due to a change in the pH of the buffer elicited by parthenolide (and the thiol modifications thus due to the change in pH), we evaluated the pH in the buffer, in the presence and absence of parthenolide, with both color pH indicator strips and a pH meter. Parthenolide had no effect on the pH of the buffer (data not shown).
Parthenolide Cytotoxicity Is Blocked by Thiol Antioxidants
To gain insight into whether the parthenolide induced reduction in free exofacial thiols is a mechanism through which parthenolide mediates its cytotoxic effect, we next determined whether the effect of parthenolide on Granta cell death can be blocked by pretreatment with GSH, GSHee and NAC using the same conditions that blocked the parthenolide induced decrease in exofacial free thiols. As parthenolide can form covalent adducts directly with GSH and NAC (data not shown), the antioxidants were extensively washed out prior to the addition of parthenolide. As shown in , a 3 h exposure of Granta cells to 30 µM parthenolide was sufficient to result in a significant decrease in cell viability after 14 h (25% of control as measured by MTT assay), although there was no significant cell death seen within the 3 h duration of treatment itself (data not shown). When cells were pretreated with thiol antioxidants for 2 h, the antioxidants washed out, and the cells exposed to 30 µM parthenolide for 3 h, cell viability increased from 25% without antioxidant pretreatment to approximately 85% with 5 mM NAC, 5 mM GSEee or 5 mM GSH pretreatment, respectively (). Similar results were seen when the cells were exposed to 10 µM parthenolide (data not shown). In contrast, when Granta cells were pretreated with parthenolide (10 or 30 µM) for 3 h followed by antioxidant exposure, the cytotoxic effects of parthenolide were not inhibited (data not shown).
Parthenolide cytotoxicity is blocked by pre-treatment with thiol antioxidants.
NFκB Antagonism and JNK Activation by Parthenolide Is Blocked by GSH
As parthenolide has been shown to induce malignant cell death by inhibiting NFκB activation and/or activating JNK 
, we next determined whether GSH, which blocked the effect of parthenolide on exofacial thiols, could alter the effect of parthenolide on NFκB and JNK.
Using a highly sensitive and specific ELISA-based assay that detects only activated NFκB subunits 
, exposure of Granta cells to 30 µM parthenolide for 3 hours resulted in an approximately 5-fold decrease in p65 DNA binding activity (). These inhibitory effects of parthenolide on p65 DNA binding activity, however, were reversed by preincubating the cells with 5 mM GSH ().
Parthenolide inhibition of NFκB and activation of JNK is inhibited by GSH.
Since the post-translational modification of p65 often determines the transcriptional activity of this molecule, we next evaluated the effect of parthenolide on the key phosphorylation site, serine 536, within the p65 molecule. Exposure of Granta cells to 30 µM parthenolide for 3 hours resulted in a significant decrease in the phosphorylation state of p65 (, lane 2) which was abrogated by pretreatment with 5 mM GSH for 2 hours (, lane 4). That the total p65 levels under these experimental conditions remained unaltered suggests that the findings shown in are truly due to the modification of p65 by such experimental conditions.
As the activation of NFκB is intimately coupled with the signal induced phosphorylation and subsequent proteosomal degradation of the NFκB inhibitor, IκB-α, we next evaluated the effect of parthenolide on phospho- and total IκB-α levels. As shown in , exposure of Granta cells to 30 µM parthenolide for 3 hours resulted in a decrease in phospho-IκB-α, which is highly consistent with the observed effects of parthenolide on p65 DNA binding activity and modification of p65 (, respectively). It is of interest that the levels of total endogenous IκB-α were decreased in the cells exposed to GSH (, lanes 3
). This may be a result of the reported effects of GSH on IκB-α degradation 
Finally the effect of GSH on parthenolide-mediated activation of JNK was examined by conducting immunoblot analyses. Analogous to the observations outlined in the above experiments, parthenolide-dependent phosphorylation of JNK isoforms (p46 and p54) was completely blocked in the cells pre-treated with GSH ().