The HIV Protease Inhibitor Nelfinavir Downregulates Akt Phosphorylation by Inhibiting Proteasomal Activity and Inducing the Unfolded Protein Response

Abstract

HIV protease inhibitors(HPIs), which have been used to treat HIV patients since the mid 1990s, have been shown to downregulate the phosphatidylinositol 3-kinase (PI3K)-Akt pathway. Because this pathway is frequently activated in human malignancies and associated with resistance to ionizing radiation, we investigated and confirmed that HPIs could radiosensitize cells. However, the mechanism underlying this downregulation was unclear, prompting the investigations in this report. In this paper we show that nelfinavir inhibits proteasome activity. Inhibition of the proteasome leads to endoplasmic reticulum-based stress with accumulation of misfolded proteins, which triggers the unfolded protein response (UPR). As part of the UPR, the alpha subunit of eukaryotic translation initiation factor 2 (eIF2α) is phosphorylated, resulting in a decrease in global protein synthesis and induction of the feedback regulator growth arrest and DNA damage-inducible protein (GADD34), which acts as a phosphatase in complex with protein phosphatase 1. This complex dephosphorylates eIF2α; however, our data also suggest that this phosphatase activity can dephosphorylate Akt. Furthermore, our data indicate that nelfinavir decreases Akt phosphorylation by triggering this response. These findings may have important implications in understanding how nelfinavir may increase radiation sensitivity and also result in downregulation of the PI3K/Akt pathway.

Introduction

The phosphatidylinositol 3-kinase (PI3K)/Akt pathway is thought to play an important role in the development of cancers possibly through effects on cell proliferation, adhesion, migration, invasion, apoptosis, and angiogenesis [1,2]. Activation of this pathway is very common in human cancers. This can occur through direct mechanisms such as amplification/mutation of the PI3K subunits [3,4] or Akt overexpression [5]. The PI3K/Akt pathway can also be indirectly activated in cancers through loss of phosphatase and tensin homolog deleted on chrorose 10 protein [2,6], mutation of Ras [7,8], and mutation/amplification of the epidermal growth factor receptor (EGFR) [9,10]. We and others have shown that activation of the PI3K/Akt pathway in human cancers results in resistance to ionizing radiation both in cells in tissue culture and in tumor xenografts in vivo [11–14]. For this reason, we have been interested in targeting this pathway as a means of increasing radiation sensitivity. Because the PI3K/Akt pathway is so commonly activated in tumors but not in normal tissues, inhibition of this pathway should offer some selectivity in the treatment of many cancers. There is currently a great deal of ongoing research to develop drugs targeting the PI3K/Akt pathway that are safe to use in people. In a previous article, we reported that HIV protease inhibitors (HPIs), including nelfinavir, could decrease Akt phosphorylation and increase the sensitivity of cells to radiation [15]. We tested five first-generation HPIs and found that three of them (nelfinavir, amprenavir, saquinavir) inhibited Akt signaling [15]. Of the three, we felt nelfinavir was the most efficacious. However, the mechanism by which nelfinavir decreases Akt phosphorylation remains unclear.

The HPIs are peptidomimetics that inhibit the HIV aspartyl protease, a retroviral enzyme that cleaves the viral gag-pol polyprotein and is necessary for the production of infectious viral particles [16]. These drugs have been used for over a decade to treat patients with HIV infection and are fairly safe. However, they are associated with lipid and metabolic disturbances including hyperlipidemia, insulin resistance, peripheral lipoatrophy, central fat accumulation, and hepatic steatosis [17]. Akt, especially the Akt2 isoform [18], plays a key role in the coordinated regulation of growth and metabolism by the insulin/insulinlike growth factor signaling pathway [19]. Therefore, it is possible that the insulin resistance caused by the HPIs could be related to the decrease in Akt phosphorylation that we have noted.

Data are emerging that the first-generation HPIs (including nelfinavir) inhibit proteasome function [20,21]. Parker et al. [20] have found that nelfinavir inhibits the chymotryptic activity of the 20S proteasome by 50% at 4 µmol/l. The proteasome performs a surveillance function by controlling proteolysis of regulatory proteins, such as those involved in cell cycle progression and apoptosis. Inhibition of the proteasome leads to excessive accumulation of misfolded proteins in the endoplasmic reticulum (ER). This leads to the unfolded protein response (UPR) [20], which serves to alleviate ER stress [22,23]. Under nonstress conditions, immunoglobulin heavy chain binding protein (BiP) (also known as GRP78) is bound to the ER-luminal domains of a number of transmembrane kinases including RNA-dependent protein kinase-like ER kinase (PERK) preventing its activation [24]. After excessive accumulation of proteins in the ER, BiP preferentially binds to unfolded proteins and dissociates from PERK, thereby rendering the latter active [24]. PERK then dimerizes and phosphorylates eukaryotic translation initiation factor 2α (eIF2α) on serine 51 [24]. Phosphorylated eIF2α (P-eIF2α) globally decreases protein synthesis, thereby providing the stressed cells time to clear misfolded proteins from the ER and facilitate recovery [25,26]. P-eIF2α also increases translation of a few UPR-related transcripts such as those encoding activating transcription factor 4 and growth arrest and DNA damage-inducible protein (GADD34) [24]. GADD34 complexes with PP1 to form a phosphatase that functions in a negative feedback loop to reverse eIF2α phosphorylation and limit the UPR [27].

In this study we explore the effect of nelfinavir on ER stress and on the expression of various downstream proteins including P-eIF2α, PP1, GADD34, and BiP. We relate this to the dephosphorylation of Akt and construct a model in which nelfinavir's effect on Akt is related to its induction of the UPR. Understanding how nelfinavir decreases Akt phosphorylation may have important implications in understanding how the drug leads to insulin resistance, or in designing drugs that are better radiosensitizers.

Methods

Cells

SQ20B cells were obtained from American Type Culture Collection (Rockville, MD). They were cultured in DMEM (Fisher Scientific, Pittsburgh, PA) supplemented with 10% FBS (Atlanta Biologicals, Norcross, GA), penicillin (100 U/ml), and streptomycin (100 mg/ml) (Gibco/BRL, Gaithersburg, MD) at 37°C in humidified 5% carbon dioxide-95% air.

SQ20B cells were transfected with the plasmids myc-tagged PP1 (T320A) (kindly provided by Dr. Xu Wanping from Dr. Necker's Lab, National Cancer Institute, Bethesda, MD) [28] and Flag-tagged GADD34 (generously released by Dr. David Ron, New York University) [29] using lipofectamine (Invitrogen, Carlsbad, CA) according to the manufacturer's instructions.

Drugs

Nelfinavir was bought for research use from the inpatient pharmacy at the Hospital of the University of Pennsylvania. Nelfinavir came as solid caplets. It was ground into a fine powder and subsequently dissolved in 100% ethanol. LY294002, tunicamycin, thapsigargin, okadaic acid, calyculin A, MG132, and cyclohexamide were purchased from Sigma-Aldrich (St. Louis, MO).

Western Blotting

Cells were lysed without trypsinization by rinsing culture dishes once with PBS followed by lysis with reducing Laemlli sample buffer. Samples were boiled, sheared, clarified by centrifugation, and stored at -20°C. Samples containing equal amounts of protein were separated on a 12% SDS-polyacrylamide gel and blotted onto nitrocellulose membranes. Membranes were blocked in PBS containing 0.1% Tween 20 and 5% powdered milk before addition of primary antibody. Polyclonal antibodies against Akt, phospho-Akt (Ser473), eIF2α, phospho-eIF2α, and phospho-PERK were purchased from Cell Signaling (Danvers, MA). Monoclonal antibodies against β-actin and FLAG (M2) were obtained from Sigma-Aldrich. Monoclonal antibody against c-myc (4E10), polyclonal antibodies against GADD34, CHOP (GADD153), and BiP (GRP78) were from Santa Cruz Biotechnology (Santa Cruz, CA). Polyclonal antibody against PP1 was from Calbiochem (San Diego, CA). All antibodies were diluted 1:1000 in 5% milk in PBS-T overnight at 4°C. Antibody binding was detected using the ECL chemiluminescence kit (Amersham, Arlington Heights, IL). Images were digitized using an Arcus II scanner, and figures were assembled with Adobe Photoshop 3.0 and Microsoft Power Point.

Radiation Survival Determination

Cells in exponential growth phase were counted and plated in 60-mm dishes containing 4 ml of medium. The cells were allowed to attach and drugs were added to cultures at least 1 hour before radiation. Cells were irradiated with a Mark I cesium irradiator (J.L. Shepherd, San Fernando, CA) at a dose rate of 1.6 Gy/min. Colonies were stained and counted 10 to 14 days after irradiation. A colony by definition had >50 cells. The surviving fraction was calculated by dividing the number of colonies formed by the number of cells plated times plating efficiency. Each point on the survival curve represents the mean surviving fraction from at least three replicates.

Immunoprecipitation

SQ20B cells were treated with vehicle (ethanol) and 10 µmol/l nelfinavir for 24 hours, washed in PBS, and lysed in RIPA buffer (20 mmol/l Tris, pH 7.5, 150 mmol/l NaCl, 1 mmol/l EDTA, 1 mmol/l EGTA, 1% Triton X-100, 2.5 mmol/l sodium pyrophosphate, 1 mmol/l β-glycerol phosphate, 1 mmol/l Na₃VO₄, 1 µg/ml leupeptin, and 2 mmol/l PMSF). Sample lysates were collected and cleared. Cell lysates (250 µg) were immunoprecipitated with 2.5 µg Akt or PP1 antibody overnight at 4°C. This was then rocked with 20 µl slurry of Protein A beads (Invitrogen) for 2 hours at 4°C. Beads were washed four times with RIPA buffer, boiled in 6x SDS-PAGE sample buffer and run on 10% SDS-PAGE gel. Protein was transferred to nitrocellulose then blotted for Akt and PP1.

[³⁵S]Methionine Incorporation Assays

Cells were treated for 24 hours with nelfinavir, then the procedure outlined by Koumenis et al. [30] was followed. In brief, cells were harvested and equal numbers of cells (5 x 10⁵) were plated into 60-mm dishes in media that contained no methionine to ensure optimal incorporation of [³⁵S]methionine. After 10 hours in methionine-free media, cells were labeled with [³⁵S]methionine (50 µCi/ml) for 30 minutes. Cells were washed three times with PBS, scraped off the dishes in PBS, and centrifuged at 5000g. The cell pellet was resuspended in PBS containing 2 mmol/l EDTA and lysed by three rounds of freeze-thawing. TCA precipitation of macromolecules and scintillation counting were performed as follows: Briefly, 10 µl of sample was added to a mixture of 250 µl of water, 50 µl of BSA (1 mg/ml), and 100 µl of TCA (50%, wt/vol). The samples were kept on ice for <20 minutes and then filtered through glass filters (GC filters, VWR Scientific, West Chester, PA) under vacuum. The filter was washed three times with ice-cold TCA (5%, wt/vol) and with ethanol (95%, vol/vol) and then immersed into 5 ml of scintillation liquid (Ecolume, ICN, Costa Mesa, CA). Disintegrations per minute were calculated as the percentage of [³⁵S]methionine incorporation relative to that incorporated by untreated cells. Results were normalized as the percent increase or decrease of [³⁵S]methionine incorporation compared with that of untreated control cells.

Proteasome Activity

SQ20B cells were plated on 100-mm dishes and treated for 24 hours with nelfinavir at various concentrations (0, 5, 10, 15, or 20 µmol/l). Cell lysates were collected and assayed for 20S proteasome activity as detailed in Calbiochem's (San Diego, CA) proteasome isolation kit (cat. no. 539176). Briefly, this kit uses proteasome affinity beads that bind the proteasome in equal amount of protein. The beads are washed and incubated in a reaction buffer containing a fluorogenic substrate such as Suc-Leu-Leu-Val-Tyr-AMC (Calbiochem, cat. no. 539142) to measure the proteolytic activity of the proteasome on the affinity beads. This is compared to an AMC standard curve, which was done as described in a Chemicon International (Temecula, CA) 20S proteasome activity assay kit (cat. no. APT280). As a control, treatment with MG132 was used.

Results

Nelfinavir Inhibits Akt, Results in Radiation Sensitization, and Inhibits the Proteasome

Western blot analysis shows that nelfinavir (10–20 µmol/l) was able to downregulate Akt phosphorylation as could the PI3K inhibitor LY294002 (Figure 1A). We have previously published that 5 µmol/l could also downregulate Akt but with longer treatment (3 days) [15]. We performed clonogenic kill curves with nelfinavir and LY294002 individually and in combination. Both drugs sensitized SQ20B cells to radiation (Figure 1B), but the combination gave no additive effect, suggesting that they work through the same pathway. We measured 20S proteasome activity in SQ20B cells after treatment with nelfinavir for 24 hours (Figure 1C). Nelfinavir concentrations of 5 µmol/l resulted in 30% less proteasome activity with a leveling of dose response in the 5- to 15-µmol/l range.

Figure 1.

Nelfinavir inhibits Akt, results in radiation sensitization and inhibits the proteasome. (A) SQ20B cells were treated with LY294002 (LY) or nelfinavir (Nfv) at the indicated concentrations or control carrier (C). Twenty-four hours later, cells were harvested and Western blotting was performed for P-Ser Akt or β-actin (loading control). (B) SQ20B cells were plated and allowed to attach for 4 hours. Cells were then treated with nelfinavir (5 µmol/l) and/or LY294002 (10 µmol/l). One hour later, they were irradiated. Ten to 14 days later, dishes were stained and colonies were counted. (C) 20S proteasome activity in SQ20B cells was measured with a Calbiochem proteasome isolation kit (cat. no. 5339176) after treatment with 0 to 20 µmol/l nelfinavir for 24 hours.

SQ20B Cells Exhibit ER Stress and UPR with Nelfinavir

To show that there was a relationship between downregulation of Akt phosphorylation and proteasomal inhibition, SQ20B cells were treated for 24 hours with 0 to 40 µmol/l of nelfinavir (Figure 2A). Dephosphorylation of the Ser473 site of Akt starts at 5 µmol/l and is complete by 40 µmol/l. The total Akt level also decreased with high concentrations of nelfinavir, perhaps due to the exaggeration of the attenuation of global protein synthesis. There was no decrease in total Akt at the 5- to 10-µmol/l range even after treatment for 3 days (data not shown). Phosphorylation of eIF2α, hallmark of the UPR, started at 5 µmol/l, peaked at 10 µmol/l, but then subsequently decreased in association with increased expression of GADD34, which in complex with PP1 forms a phosphatase that negatively regulates P-eIF2α. The levels of BiP and PP1 did not change. We then treated SQ20B cells with 10 µmol/l nelfinavir and harvested protein at different time points (0–24 hours) after treatment (Figure 2B). There was a decrease in P-Akt with increased duration of time of exposure to nelfinavir but without change in total Akt levels. By 24 hours, we saw increased phosphorylation of PERK and eIF2α. In addition, at 24 hours, we started seeing expression of GADD34. After 72 hours of exposure to 10 µmol/l nelfinavir, there was complete downregulation of P-Akt with accompanying decrease in P-eIF2α and an increase in GADD34 (data not shown); however, the levels of BiP and PP1 did not change. Therefore, these data demonstrate that, in a dose- and time-dependent manner, the downregulation of P-Akt correlates with ER stress as manifested by P-eIF2α and subsequent translation of GADD34.

Figure 2.

Nelfinavir induces proteins associated with ER stress and the UPR. In (A) to (D), SQ20B cells were seeded, then exposed to drug 4 hours later. (A) Cells were treated with nelfinavir (Nfv) at the indicated concentrations. Twenty-four hours later, protein samples were harvested. (B) Cells were treated with 10 µmol/l nelfinavir. At the times indicated, protein samples were harvested from replicate dishes for protein. (C) Cells were treated with 1 µmol/l of the proteasomal inhibitor MG132. At the times indicated, protein samples were harvested from replicate dishes. (D) Cells were treated with either tunicamycin or thapsigargin at the indicated concentrations. Twenty-four hours later, protein samples were harvested. In (A) through (D), Western blotting was performed for using antibodies as indicated. (E) [³⁵S]Methionine incorporation assays were performed on SQ20B cells treated with the indicated concentration of nelfinavir for 24 hours. Sixty minutes before the end of treatment, cells were labeled with [³⁵S]methionine (50 µCi/ml). Results were normalized as the percent increase or decrease of [³⁵S]methionine incorporation compared with untreated control cells.

If the effects of nelfinavir on Akt phosphorylation are indeed mediated by proteasomal inhibition, then a proteasomal inhibitor should have the same effect. MG132 is a known inhibitor of the proteasome [31]. We treated SQ20B cells with 1 µmol/l MG132 and harvested protein at 0 to 24 hours (Figure 2C). Akt phosphorylation was downregulated by 24 hours and corresponded to a increase in P-eIF2α and induction of GADD34. PP1 levels remain constant although the level of BiP does increase.

To demonstrate that the coupling of P-Akt downregulation to UPR induction was not isolated to nelfinavir and MG132, we used the agents tunicamycin and thapsigargin, which are widely used for induction of the UPR in cell culture. Tunicamycin depletes intracellular calcium stores to trigger the UPR, whereas thapsigargin inhibits protein glycosylation [32,33]. Figure 2D shows that both tunicamycin and thapsigargin could decrease P-Akt to undetectable levels with increasing concentrations of the drugs. The decrease in Akt phosphorylation was accompanied by induction of the UPR as evidenced by the phosphorylation of eIF2α and translation of GADD34 and BiP (Figure 2D). These results further support the hypothesis that one cell pathway couples P-Akt downregulation with UPR induction in SQ20B cells.

Another hallmark of UPR induction is a global reduction in protein synthesis. To verify that nelfinavir induced ER stress, we measured new protein synthesis with [³⁵S]methionine labeling in SQ20B cells treated with varying concentrations of the drug (Figure 2E). Figure 2E shows that there was a dose-dependent decrease in protein synthesis. At doses of 5 and 10 µmol/l, new protein synthesis was decreased to 66% and 40%, respectively, of that seen in control cells. However, even this level of decrease in protein synthesis was insufficient to cause SQ20B cells to undergo apoptosis or retard the growth of SQ20B cells (data not shown).

PP1 but Not PP2A Mediates Akt Phosphorylation

Because there was a temporal correlation between dephosphorylation of Akt and the induction of GADD34, which in complex with PP1 has phosphatase activity toward eIF2α, we wanted to see whether Akt might be a target of the GADD34/PP1 phosphatase. Cells were pretreated with calyculin A, which inhibits PP1, then exposed to nelfinavir (Figure 3A). Nelfinavir downregulated Akt phosphorylation when cells were pretreated with 0 or 1 nmol/l calyculin, neither of which affected the level of PP1. However, with higher doses of calyculin (>10 nmol/l), the level of PP1 decreased and the phosphorylation of eIF2α increased, as expected. There was a concomitant increase in the phosphorylation of Akt, and its phosphorylation was no longer responsive to nelfinavir. In the first four lanes of this blot we do not see eIF2α phosphorylation with nelfinavir treatment alone because the signal with calyculin was so high; the film had to be removed quickly. However, with longer exposure, we do see eIF2α phosphorylation with nelfinavir alone (data not shown). Calyculin inhibits PP2A as well as PP1; therefore, we also treated cells with okadaic acid, which is a more specific inhibitor of PP2A. We saw no difference in P-eIF2α or P-Akt with the addition of okadaic acid to nelfinavir compared with nelfinavir alone (data not shown).

Figure 3.

Inhibition of PP1 reverses the effect of nelfinavir on Akt phosphorylation. SQ20B cells were seeded, then exposed to drug 4 hours later. (A) Cells were pretreated with calyculin A (CA) at the indicated concentration for 1 hour before treatment with nelfinavir (10 µmol/l). Twenty-four hours later, protein was harvested. (B) SQ20B cells were pretreated with calyculin A or okadaic acid (OA) at the indicated concentration for 1 hour before treatment with tunicamycin (1.5 µg/ml). Twenty-four hours later, protein was harvested. In (A) and (B), gel electrophoresis was performed followed by Western blotting using antibodies as indicated.

To confirm that inhibition of PP1 by calyculin also reverses the P-Akt downregulation that occurs in response to a known inducer of the UPR, SQ20B cells were pretreated with calyculin or okadaic acid followed by treatment with tunicamycin. In Figure 3B, calyculin completely abolished P-Akt downregulation, dramatically increased eIF2α phosphorylation, and decreased PP1 levels. In contrast, treatment with okadaic acid had little or no effect on the levels of P-Akt and P-eIF2α. Collectively, these data support the idea that PP1 can function as part of a phosphatase that can dephosphorylate both eIF2α and Akt.

GADD34 Induction Promotes Downregulation of Akt in SQ20B Cells

Although our results suggest that the activation of PP1 phosphatase is responsible for P-Akt downregulation in SQ20B cells, PP1 protein levels remained unchanged with treatment with nelfinavir, MG132, tunicamycin, and thapsigargin, indicating that eIF2α phosphatase activity was not regulated by the UPR by altering translation of PP1. We also analyzed PP1 phosphorylation, which is another means of regulating PP1 activity, and this was unchanged during UPR activation (data not shown). PP1 can also bind to its substrates such as Akt, which would be another mechanism by which Akt activity could be modulated [28,34]. We first confirmed that the treatment with the drug for 24 hours did not change the level of either PP1 or total Akt (Figure 4A). To study whether nelfinavir perturbed the association between Akt and PP1, we performed coimmunoprecipitations using cell lysates from control or nelfinavir-treated SQ20B cells (Figure 4B). Compared with 10% input, which represents equal loading of Akt and PP1, no obvious changes in the association between Akt and PP1 were detected with nelfinavir treatment. Both antibodies could coimmunoprecipitate equal amounts of Akt and PP1 from control and nelfinavir-treated cells. These results indicate that nelfinavir does not lead to Akt dephosphorylation by changing the interaction between PP1 and Akt.

Figure 4.

Inhibition of PP1 reverses the effect of nelfinavir on Akt phosphorylation. (A) SQ20B cells were seeded. Four hours later, they were treated with 10 µmol/l nelfinavir (Nfv). Twenty-four hours later, protein was harvested and subjected to gel electrophoresis. Western blotting was performed for Akt and PPI1. (B) Same as in (A) except after 24 hours of nelfinavir treatment, cells were lysed in RIPA buffer and immunoprecipitation was performed using antibody against either Akt (left) or PP1 (right). Immunoprecipitated proteins were run on gel electrophoresis and Western blotting was performed for both Akt and PP1.

In complex with PP1, GADD34 dephosphorylates P-eIF2α [27]. Others have found that GADD34 expression is induced late in the UPR [24], presumably as a negative feedback mechanism to limit eIF2α activity by promoting its dephosphorylation. We have also observed that GADD34 expression was induced with activation of UPR signaling elicited by the different reagents including nelfinavir, MG132, thapsigargin, and tunicamycin (Figure 2). This induction occurred with a direct temporal correlation with dephosphorylation of Akt, suggesting that there might be a cause and effect relationship. To confirm the role of GADD34 in the downregulation of P-Akt, we transfected SQ20B cells with GADD34 cDNA or a constitutively active mutant of PP1 (T320 in which the phosphorylation site has been mutated from T to A). The Western blots in Figure 5 demonstrate that P-Akt levels in either of these two transfectants were significantly lower than in the vector control. Therefore, Akt is dephosphorylated in response to increased PP1 phosphatase activity. Treatment with nelfinavir further downregulated P-Akt. As expected, the phosphorylation of eIF2α was decreased in cells transfected with GADD34; surprisingly, this did not occur in cells transfected with PP1(T320). A possible explanation for this is that the point mutation may prevent binding of the GADD34-PP1 (T320A) complex to eIF2α but not to Akt.

Figure 5.

Expression of GADD34 or constitutively active PP1 decreases Akt phosphorylation. SQ20B cells were transfected with empty vector or vector expressing either PP1(T320) (constitutively active mutant) or GADD34. Twenty-four hours later, cells were exposed to nelfinavir (10 µmol/l). Twenty-four hours later, cells were harvested for protein and gel electrophoresis was performed followed by Western blotting using antibodies as indicated.

Discussion

We have been interested in understanding the mechanism by which the protease inhibitor nelfinavir decreases Akt phosphorylation. The primary impetus for this has been our finding that tumor cells can be radiosensitized by the drug both in vitro and in vivo. Because nelfinavir has been used for over a decade in the treatment of HIV patients with an acceptable toxicity profile, it has the potential of being moved quickly into clinical trials to be tested as a radiosensitizer. Our previous results suggest that the ability of nelfinavir to radiosensitize is dependent on its downregulation of P-Akt [15]. Therefore, we were very interested in determining how this downregulation occurred mechanistically. Our model as to how nelfinavir leads to Akt dephosphorylation is shown in Figure 6. Nelfinavir inhibits the proteasome (Figure 1C) at similar concentrations (micromolar) achievable in the serum of patients on the drug [35] and at the same doses at which Akt phosphorylation is inhibited. Although this does not prove that proteasomal inhibition/ER stress is required for Akt dephosphorylation by nelfinavir, this idea is supported by the fact that structurally unrelated agents that inhibit the proteasome (MG132) or induce ER stress (thapsigargin, tunicamycin) also decrease Akt phosphorylation (Figure 2, C and D). Nelfinavir induces ER stress, which leads to phosphorylation of eIF2α and global inhibition of protein synthesis. However, the synthesis of selected proteins such as GADD34 is actually increased under ER stress. GADD34, in complex with PP1, dephosphorylates eIF2α as part of a negative feedback loop. We observed a temporal correlation between GADD34 induction and P-Akt downregulation, suggesting that there might be a cause and effect relationship between the two. To show this, we performed transfections with GADD34 and constitutively active PP1 and found that both of these also downregulated P-Akt. These results do not specifically place the GADD34/PP1 phosphatase activity downstream of nelfinavir. However, treatment of cells with calyculin A, which inhibits both PP1 and PP2A, increased Akt phosphorylation and made the cells refractory to nelfinavir. This result indicates that either PP1 or PP2A likely lies between nelfinavir and Akt dephosphorylation. We repeated the experiment using okadaic acid, which has a unique activity profile against phosphatases. Okadaic acid inhibits PP2A at low concentrations (IC₅₀ <0.1 nmol/l), but it inhibits both PP2A and PP1 at higher concentrations (IC₅₀ = 150 nmol/l) [28]. At a concentration of up to 100 nmol/l, okadaic acid has no effect on Akt phosphorylation (Figure 3B), ruling out PP2A as the responsible phosphatase. Taken together with the transfection data with PP1, we conclude that PP1 is the phosphatase that dephosphorylates Akt in response to nelfinavir.

Figure 6.

Model of Akt dephosphorylation by nelfinavir. Nelfinavir induces ER stress and the unfolded response (UPR). As part of this response, protein translation is globally inhibited as a result of phosphorylation of eIF2α. However, expression of selected proteins such as activating transcription factor 4 and GADD34 is actually increased. GADD34 complexes with PP1 to form a phosphatase that dephosphorylates eIF2α and, according to our hypothesis, Akt as well.

Several investigators have implicated PP2A in regulating Akt phosphorylation [36–38]; however, these studies were all performed on cells of nonepithelial origin. In contrast, our okadaic results suggest that PP1, not PP2A, is the relevant phosphatase. Consistent with our results, Xu et al. [28] showed in SKBR3 breast carcinoma cells that endogenous Akt and PP1 associated with each other. They also showed that transient overexpression of constitutively active PP1 promoted Akt dephosphorylation, and, using an in vitro phosphatase assay, they confirmed that Akt is a substrate of PP1.

It is interesting that other agents currently in clinical trial such as bortezomib (PS341) also exert antitumor activity by inhibiting the proteasome, leading to the UPR [39]. Furthermore, in one study, the level of phosphorylated Akt was decreased in a dose-dependent manner with bortezomib [40]. Bortezomib has been approved for use in relapsed or refractory multiple myeloma as single-agent therapy. The grade 3/4 toxicity associated with it include thrombocytopenia (29%), diarrhea (11%), anemia (11%), and neutropenia (10%) [41]. The toxicity seen with nelfinavir in HIV patients within 1 to 2 months of use are mostly gastrointestinal and consists of grade 1/2 diarrhea (57%) and, occasionally, grade 3/4 (9%), which readily responds to antidiarrheal medication and rarely requires discontinuation of nelfinavir [35]. We also did not see single-agent activity, changes in the growth rate, or apoptosis in tumor cells with nelfinavir at clinically relevant doses. At higher doses (20–40 µmol/l) we clearly saw apoptosis with nelfinavir alone. We propose a dual-hit hypothesis of radiation sensitization. In solid tumor, we can exert a low-grade stress response with nelfinavir, and the addition of radiation results in increased cell kill. It is possible bortezomib therapy at lower doses would also result in decreased toxicity and radiosensitization.

In summary, we have shown that the HIV protease inhibitor nelfinavir decreases Akt phosphorylation indirectly by inhibiting the proteasome and triggering the UPR response. This results in radiosensitization at clinically relevant doses of nelfinavir. The toxicity and safety profile of nelfinavir has been well studied since the mid 1990s. It is clinically poised for a trial with radiation. In the meantime, knowledge of the mechanism by which nelfinavir results in radiation sensitization may be useful in designing newer agents that might have even less toxicity and might be more effective radiosensitizers.