Proteasome inhibitor PS-341 (VELCADE) induces stabilization of the TRAIL receptor DR5 mRNA through the 3′-untranslated region Free

Tumor necrosis factor–related apoptosis-inducing ligand (TRAIL) is a proapoptotic protein thought to be a potential anticancer agent due to its ability to target tumor but not normal cells and its in vivo activity against a wide range of tumor types including melanoma, breast, and prostate cancer (for reviews, see refs. 1–3). Cytotoxic agonist antibodies directed against TRAIL receptors have been developed and are useful against multiple tumor types (4–6) and these antibodies have entered clinical trials. The proteasome inhibitor PS-341 (VELCADE, bortezomib; ref. 7) has been shown to greatly enhance the killing activity of TRAIL (8–11) by increasing TRAIL receptors, DR4 (TRAIL R1) and DR5 (TRAIL R2; ref. 12). This increase in TRAIL receptors occurs via the inhibition of ubiquitinated TRAIL receptor degradation and by increasing TRAIL receptor mRNA (12). In both yeast and multiple myeloma cells in culture (13, 14), the addition of PS-341 both increases and decreases specific mRNAs, suggesting that regulation of gene transcription by this agent is a general phenomenon not limited to the TRAIL receptor or prostate cancer cells.

The mechanism by which PS-341 controls the transcription of DR5 is complex, but in multiple myeloma and head and neck cancer cells, the unfolded protein response is involved (15–17). PS-341 induces PERK, an ER stress-specific eIF-2α kinase, ATF4, an ER stress-induced transcription factor, and its proapoptotic target CHOP/GADD153 (15–17). Because the DR5 upstream region contains a CHOP-binding sequence, induction of these proteins can stimulate the transcription of this gene (18, 19). In addition to the CHOP binding site, the first intron of the DR5 gene contains binding sites for both p53 and nuclear factor-κB: DR5 was first cloned as a p53-responsive gene (20, 21), and regulation of nuclear factor-κB activity can modulate the transcription of this gene (22). The transcriptional regulation of DR5 is complex, and mapping the promoter region reveals multiple potential transcription factor binding sites.

An alternative to transcriptional control mechanisms for modulating DR5 levels is mRNA stability regulation. Many inflammatory mediators, growth factors, and cytokines, such as tumor necrosis factor-α, interleukin-1β, interleukin-8, granulocyte macrophage colony-stimulating factor, macrophage inflammatory protein 1α, and vascular endothelial growth factor, are all regulated by mRNA stability (23–28). These mRNAs contain class II AU-rich elements found in the 3′-untranslated region (UTR) that bind protein factors, which regulate mRNA half-life. In addition, there are AU-rich elements (class I) that are mostly found in the 3′-UTR of proto-oncogenes, e.g., c-Fos (29). Proteins that can directly bind to AU-rich elements to promote mRNA decay include tristetraprolin (30, 31), BRF1 (32), and TIA-1 (33), whereas mRNA-stabilizing proteins include the AUF-1 isoforms and the HuR protein (34–36). The levels and activity of these mRNA-binding proteins can be regulated both transcriptionally and post-translationally.

To further understand how PS-341 increases DR5, we investigated whether PS-341 treatment increases the stability of DR5 mRNA. This potential mechanism is suggested by the observation that the DR5 3′-UTR contains several AU-rich regions capable of binding proteins that post-transcriptionally regulate this mRNA (37). We find that PS-341 treatment increases the stability of DR5 mRNA half-life from 4 to 10 h, enhances the binding of cytoplasmic proteins to the DR5 3′-UTR mRNA, and increases cytoplasmic levels of mRNA-binding proteins.

LNCaP, PC-3, and DU145 prostate cancer cell lines were purchased from American Type Culture Collection. LNCaP cells were cultured in RPMI 1640, and PC-3 and DU145 cells were cultured in DMEM (Hyclone) supplemented with 10% fetal bovine serum (Hyclone) and 100 μg/mL penicillin-streptomycin (Invitrogen). Cells were grown in 5% CO₂ at 37°C. Antibodies used for Western blot analysis at a dilution of 1:1,000 include a HuR monoclonal antibody (3A2), TIAR, TIA-1, tristetraprolin, hnRNP C1/C2, glyceraldehyde-3-phosphate dehydrogenase (GAPDH), and Lamin B1 all purchased from Santa Cruz Biotechnology. A TIAR monoclonal antibody used for immunoprecipitation was purchased from BD Biosciences. The AUF-1 antibody (1:4,000 dilution) was obtained from Upstate Biotechnology. The anti-HA antibody was obtained from Covance. Anti-Flag M2 antibody, anti-Flag beads, and anti-HA beads were from Sigma. Protein A agarose beads were supplied by Invitrogen. All secondary antibodies conjugated to horseradish peroxidase were from Amersham Biosciences and diluted 1:5,000. The riboprobe in vitro transcription kit was purchased from Promega. All other chemicals were purchased from Sigma, unless otherwise indicated.

A full-length DR5 clone was obtained from Origene. Using this clone as a template, DR5 3′-UTR fragments were generated by PCR amplification (see Supplementary Table S1 for oligomers)¹

For mRNA stability experiments, 24 h after transfection, LNCaP cells were treated with DMSO or PS-341 (0.5 μmol/L, 12 h) followed by actinomycin D (2 μg/mL), and cells were harvested for mRNA quantification. For analysis of small interfering RNA (siRNA) duplex effects and EGFP reporter expression 24 h after electroporation, cells were treated with PS-341 (0.1 μmol/L) for 12 h followed by the addition of actinomycin D (2 μg/mL). The cells were then harvested at regular time intervals for mRNA measurements. For protein analysis in cytoplasmic extracts, siRNA duplex transfected cells were harvested after 36 h of electroporation. All of the siRNA duplexes (see Supplementary Table S1)¹ were obtained from Dharmacon.

For transfection of cDNA into LNCaP cells, 1.0 × 10⁷ cells/mL were electroporated at 280 mV and 950 μF using a Bio-Rad Gene Pulser II System. Alternatively, LipofectAMINE (Invitrogen) was also used for transfection. For the assessment of cell viability, scrambled or HuR siRNA-transfected cells were plated onto a 96-well plate. Twenty-four hours after transfection, the cells were treated with PS-341, TRAIL, or the combination for 18 h and cell viability measured by the MTS assay (38).

Cytoplasmic extracts were prepared from the pelleted cells by hypotonic lysis buffer following the protocol as reported by Ford et al. (39). Nuclei were pelleted, and the protein was extracted with nuclear extraction buffer [10 mmol/L HEPES (pH 7.9), 0.41 mol/L KCl, 1.5 mmol/L MgCl₂, 0.5 mmol/L DTT, 25% glycerol, 0.2 mmol/L EDTA, 0.1 mmol/L phenylmethylsulfonyl fluoride (PMSF)], and the extract was centrifuged at 14,000 rpm for 10 min. For other Western blotting experiments, the pelleted cells were lysed in a buffer containing 20 mmol/L Tris-HCl (pH 7.4), 150 mmol/L NaCl, 2 mmol/L EDTA, 10% glycerol, 1% Triton X-100, 1 mmol/L PMSF, and protease inhibitor cocktail. The lysates were cleared by centrifugation at 14,000 rpm for 15 min and the supernatant was stored at -80°C. Total protein (30 μg) was resolved by 12% SDS-PAGE and transferred to polyvinylidene difluoride membrane followed by blocking in 3% bovine serum albumin.

RNase protection assays were done according to the Riboquant protocol described by BD PharMingen. For RNase protection assays, the DR5 probe was PCR amplified from the full-length clone, whereas the GAPDH probe was amplified from the LNCaP cDNA prepared for real-time PCR. Both probes were cloned into a pSP70 vector (Promega).

For UV cross-linking experiments, an aliquot of radiolabeled transcript (1 × 10⁶ counts/min) and yeast RNA (10 μg) with or without competing unlabeled RNA transcripts used at the indicated molar excess were added to the mixture containing LNCaP cytoplasmic extract (50 μg), 15 mmol/L HEPES (pH 7.6), 40 mmol/L KCl, 5 mmol/L MgCl₂, 1% glycerol, yeast RNA (10 μg), heparin (10 μg), and 65 units RNasin in a total volume of 50 μL. Mixtures were incubated for 10 min at 22°C and exposed to short-wave (254 nm) UV irradiation at a distance of 7 cm for 30 min. Transcripts that were not cross-linked to protein were digested with RNase A (0.5 mg/mL) and RNase T1 (10 units/mL) at 37°C for 30 min.. For immunoprecipitation of proteins bound to radiolabeled transcripts, after treatment with RNases, the mixtures were incubated with protein A-agarose beads (Invitrogen) precoated with 1 μg HuR (Santa Cruz Biotechnology), TIAR (BD PharMingen), or nonspecific IgG antibodies overnight at 4°C. The beads were washed with NT2 buffer (40) six times and resolved in 10% SDS-PAGE, dried, and subjected to autoradiography and phosphorimaging for quantitation.

Immunoprecipitation of endogenous RNA-protein complexes were done according to the method of Tenenbaum et al. (40). LNCaP cytoplasmic extracts (200 μg) were incubated (overnight at 4°C) with protein A-agarose beads (Invitrogen), which were precoated with 2 μg HuR, TIA-1, TIAR, and hnRNP C1/C2 antibodies (Santa Cruz Biotechnology) and AUF-1 antibodies (Upstate Biotechnology) in a mixture containing 50 mmol/L Tris-HCl (pH 7.4), 150 mmol/L NaCl, 1 mmol/L MgCl₂, 0.05% NP-40, 15 mmol/L EDTA, 1 mmol/L DTT, 10 mg/mL yeast RNA, 1 mg/mL heparin, and 65 units RNasin. After incubation, the beads were washed six times with NT2 buffer [50 mmol/L Tris-HCl (pH 7.4), 150 mmol/L NaCl, 1 mmol/L MgCl₂, 0.05% NP-40] and further incubated with 100 μL NT2 buffer containing 0.1% SDS and 0.5 mg/mL proteinase K (30 min, 50°C). The RNA was extracted from these immunoprecipitates using TRIzol-containing glycoblue (Ambion) as a coprecipitant and was quantitated by PCR.

For real-time reverse transcription-PCR, RNA isolated from the immunoprecipitation samples was reverse transcribed using random hexamers and SuperScript II RT (Invitrogen) and the resulting cDNA was used for PCR amplification using gene-specific primer pairs. Quantitative PCR conditions for these reactions were as follows: 95°C for 10 s, 58°C for 30 s, and 72°C 10 s for 40 cycles.

PCR was carried out for Fig. 6A in the immunoprecipitated samples under the following conditions: 94°C for 30 s, 50°C for 45 s, and 68°C for 45 s for 25 cycles and the resulting product was separated with 1% agarose gel electrophoresis.

In LNCaP prostate cancer cells, proteasomal inhibition by PS-341 for 18 h induced a 3-fold increase in DR5 protein levels (Fig. 1A). To examine whether the protein increase is secondary to DR5 mRNA accumulation in LNCaP cells, we measured DR5 mRNA after PS-341 treatment using a RNase protection assay (Fig. 1B). To measure this increase, quantitative real-time PCR was done (Fig. 1C). Results indicate that mRNA levels increased after 5 h of PS-341 treatment, and by 24 h, a 6-fold increase was evident. Results observed in these experiments with PS-341 were identical to those obtained with the proteasome inhibitor MG-132 (data not shown). Because the DR5 gene has a long 3′-UTR, we hypothesized that the mRNA may be post-transcriptionally regulated. To examine this possibility, we measured mRNA half-life after actinomycin D treatment. PS-341 and actinomycin D concentrations were carefully selected to avoid synergistic toxicity induced by the use of these agents. LNCaP cells were treated with PS-341 (500 nmol/L) for 12 h followed by actinomycin D (2 μg/mL). These experiments reveal that DR5 mRNA half-life is ∼4 h in LNCaP cells, whereas the half-life of DR5 mRNA after PS-341 treatment is dramatically increased to at least 8 h by RNase protection assay and quantitative PCR (Fig. 1D). We next evaluated DU145 and PC-3 cells and found that PS-341 treatment increased DR5 mRNA in the human prostate cancer cell line, DU145, but did not stabilize mRNA (data not shown; Supplementary Fig. S1A).¹ Variability in the 4- and 8-h time points can be accounted for by the difference in loading that is seen with the GAPDH control (Supplementary Fig. S1A).¹ Treatment of PC-3 cells with this compound neither stabilized DR5 mRNA nor increased its half-life (Supplementary Fig. S1A).¹ Although the cell lines were human prostate cancer cells, results suggest that the ability of this compound to regulate mRNA stability is cell type specific.

To examine whether the 3′-UTR of DR5 has a general effect on the stability of other genes, we cloned the EGFP coding region into the pcDNA3.1 vector followed by the 3′-UTR of the DR5 mRNA (Fig. 2). LNCaP cells were transiently transfected with EGFP plasmid alone or with an EGFP-3′-UTR plasmid. Figure 2 shows that PS-341 treatment stabilized EGFP-3′-UTR mRNA with a messenger half-life of >8 h, whereas in DMSO treated cells the half-life was between 3 and 4 h. In comparison with the half-life of the EGFP-3′-UTR mRNA, the EGFP message that did not contain the UTR was not different in control or PS-341-treated LNCaP cells. Thus, the 3′-UTR of DR5 mRNA likely contains regulatory sequences that mediate the stability of this transcript after PS-341 treatment.

To examine PS-341-induced changes in protein binding to the 3′-UTR of DR5, ∼450-bp fragments of the UTR were cloned into a pSP70 vector. These plasmids were used to generate a radiolabeled transcript for identifying proteins that could regulate mRNA stability (Fig. 3A).

The generated radiolabeled fragments were UV cross-linked after incubation with the cytoplasmic extract prepared from LNCaP cells treated with and without PS-341. After UV cross-linking, the unbound transcripts were digested with RNases, and the cross-linked material was fractionated via 10% SDS-PAGE. The binding of proteins to these 3′-UTR fragments are increased by PS-341 treatment of LNCaP cells, with the most prominent bands being ∼36 and 45 kDa (Fig. 3B). Increased binding of these proteins to the F-1 fragment occurs in a time-dependent manner with maximal increase at ∼17 h after PS-341 treatment (Fig. 3C). To examine the importance the F-1 and F-4 fragments to the regulation of mRNA stability, these fragments were cloned into the EGFP reporter and transfected into the LNCaP cells. By quantitative PCR, we find that treatment of cells with PS-341 stabilized the mRNA containing each of these 3′-UTR fragments. This suggests that each of these fragments plays a role in regulating DR5 mRNA stability by this compound (Fig. 3D). It is possible that their additive effects would be similar to the entire 3′-UTR.

Shorter deletion constructs of the F-1 and F-4 fragments were generated to select critical elements (see line diagrams in Fig. 4A and B) that bind mRNA-interacting proteins. The F-12 fragment had a protein binding pattern comparable with F-1 (Fig. 4A). To ascertain the critical stabilizing elements, we further divided F-12 into two equal shorter transcripts of 75 bp: F-121 and F-122 (Fig. 4A). Autoradiographic images of the UV cross-linking experiments for F-121 and F-122 clearly indicate that the F-122, and not F-121 fragment, binds to cytoplasmic proteins; again, protein binding is increased by PS-341 treatment (Fig. 4A). Interestingly, F-122 fragment sequence analysis reveals two overlapping AU-rich nonamers (UUUAUUUAUUUAUUU), which we have denoted F-122 AU-rich element. The F-4 region also strongly bound increased amounts of protein after PS-341 treatment. Analysis of smaller fragments (Fig. 4B) of this UTR fragment shows that the 71-bp section (denoted F-421) can bind these proteins. This fragment contains a significant U-rich stretch of nucleotides from 3,556 to 3,587 with the potential to interact with mRNA-binding proteins.

To identify whether the proteins binding fragments of the 3′-UTR were identical for both F-1 and F-4 regions, we did UV cross-linking experiments using either the 78-bp F-122 or F-421 as cold competitors (Fig. 4C). As shown in Fig. 4C, the F-421 fragment competed for protein binding to the F-1 fragment. Likewise, the F-122 fragment competed for protein binding to the F-4 fragment (Fig. 4D).

To measure AU-rich binding proteins after PS-341 treatment, we prepared cytoplasmic and nuclear extracts from LNCaP cells treated with or without PS-341 for 18 h. These extracts were fractionated via 12% SDS-PAGE for subsequent Western blot analysis. Figure 5A shows an increase in cytoplasmic but not nuclear mRNA-binding protein including the AUF-1 isoforms, HuR, and hnRNP C1/C2. These proteins are increased 1.4-, 1.7-, and 4.8-fold respectively. In additional experiments (Fig. 6A for example), HuR increased between 2.2- and 2.4-fold. To evaluate whether specific proteins showed increased binding to the DR5 3′-UTR mRNA, we examined the proteins bound to this mRNA before and after PS-341 treatment. First, mRNA-binding proteins in the cytosol were immunoprecipitated using antibodies specific to HuR and TIAR (Fig. 5B); then, the extent of binding to the DR5 mRNA was evaluated by real-time PCR. Figure 5B shows a 5-fold increase in DR5 mRNA in the HuR samples immunoprecipitated from PS-341-treated LNCaP cells, whereas immunoprecipitation with TIAR did not bring down any DR5 transcript. To quantitate these changes more accurately, we did quantitative real-time PCR using DR5 primers (Fig. 5C). With this method, we report an 8-fold increase in HuR bound to DR5 mRNA after PS-341 treatment, but we observed no significant alteration in the binding of other AU-rich proteins, AUF-1, TIA-1, and TIAR, to the DR5 mRNA (Fig. 5C).

To assess whether the HuR protein interacts specifically with the F-1 and F-12 3′-UTR fragments, we examined the ability of antibodies specific to mRNA-binding proteins to immunoprecipitate radiolabeled F-1 and F-12 fragments. Autoradiographic images of the immunoprecipitated F-1 and F-12 transcripts (Fig. 5D) show that after PS-341 treatment HuR binding, but not TIAR binding, increased 3-fold to both F-1 and F-12 3′-UTR fragments. Together, these results confirm that the HuR binding protein is involved in DR5 mRNA stabilization after PS-341 treatment.

As shown in Fig. 6A, siRNA targeted at HuR markedly decreased HuR protein in control and PS-341-treated LNCaP cells but had no effect on GAPDH protein. This siRNA decreased the levels of HuR mRNA (data not shown). Figure 6B shows that knockdown of HuR using siRNA markedly decreases the binding of the 36-kDa protein induced by PS-341 to both F-122 and F-421 fragments of the 3′-UTR (Fig. 6B, lanes 3 and 4 and lanes 9 and 10). In comparison, knockdown of the AUF-1 protein (see Supplementary Fig. S1)¹ did not affect either the major 36- or 45-kDa bands seen on UV cross-linking (Fig. 6B, lanes 5 and 6 and lanes 11 and 12). HuR knockdown decreases the half-life of DR5 mRNA (Fig. 6C) from >8 to <4 h even after treatment with PS-341. Thus, HuR protein accounts for one of the major proteins bound to the 3′-UTR and knockdown of this protein negatively regulates the ability of PS-341 to stabilize DR5 mRNA.

To evaluate the possibility that PS-341 increased HuR protein by preventing its degradation by the proteasome, a HuR cDNA containing a Flag tag (a gift from J.A. Steitz) was transfected with or without a HA-ubiquitin plasmid. Western blot analysis of the extract with Flag antibody showed that the transfected HuR plasmid is expressed in LNCaP cells (Fig. 6D, 36-kDa band) and that treatment with PS-341 not only increased the amount of this protein but also increased HuR fragments (compare lanes 1 and 2, blunt arrows in Fig. 6D). When HA-ubiquitin is overexpressed along with HuR followed by the treatment with PS-341, HuR ubiquitination is markedly increased, and the 36-kDa band of HuR is shifted to a higher molecular weight band (Fig. 6D, lane 4). To further examine the possibility that HuR was ubiquitinated, this protein was immunoprecipitated with Flag beads and then Western blotted with a Flag antibody. Only in cells transfected with Flag HuR and HA-ubiquitin did PS-341 treatment induce higher molecular weight bands of HuR protein (Fig. 6D, lane 8, bracket). If this Western blot is then stripped and probed with a HA antibody, the high molecular weight bands recognized by this antibody (Fig. 6D, lane 10) are shown to contain ubiquitin. Similarly, if the transfected proteins were immunoprecipitated with HA antibody and then Western blotted with Flag antibody, a high molecular weight band was seen only when doubly transfected cells were treated with PS-341 (Fig. 6D, lane 14). Thus, it is possible that increases in HuR induced by PS-341 treatment could be secondary to the inhibition of the degradation of this protein.

Here, we provide data to support a novel mechanism underlying the elevation of DR5 by PS-341. PS-341 treatment increases DR5 mRNA by stabilizing the mRNA half-life from 4 to ∼10 h. Our results suggest that this increase in stability is secondary to binding of specific proteins to the DR5 3′-UTR. First, we showed that transferring the 3′-UTR of DR5 to a heterologous mRNA, EGFP, allows this transcript stabilization via PS-341 treatment of transfected LNCaP cells. Second, we show that PS-341 increases protein binding to fragments F-1 and F-4 of the DR5 3′-UTR that contain A/U-rich sequences. Third, we provide evidence that PS-341 treatment increases cytoplasmic multiple mRNA-binding proteins, including HuR, which is known to stabilize mRNAs. Fourth, we show that PS-341 treatment increased HuR protein bound to the DR5 3′-UTR. Finally, we provide data to suggest that protein knockdown with siRNA decreases the ability of PS-341 to prolong the half-life of this mRNA.

Our experiments show that PS-341 treatment has the potential to block the degradation of HuR and increases its ubiquitination leading to elevated HuR. We have not ruled out the possibility that increases in HuR protein in the cytoplasm arise from inhibition of various signal transduction pathways, that is, AMP-activated protein kinase (41, 42) that regulate transport of this protein from the nucleus to the cytosol. We have focused our studies on LNCaP cells, because in this cell type PS-341 markedly enhances TRAIL-induced tumor cell killing by elevating DR5 (12). However, we show that the ability of PS-341 to prolong mRNA half-life is clearly cell type specific; this does not occur with either PC-3 or DU145 human prostate cancer cell lines. It is possible that differences among these cells are secondary to the availability of proteasome subunits. Recent experiments show that the proteasome expression levels and subunit composition in B-cell lines influence response to proteasome inhibition (43).

It is possible in the absence of PS-341 treatment that negative regulators of mRNA stability bind and regulate the DR5 3′-UTR and that changes in HuR protein levels lead to the replacement of these proteins on the mRNA with stabilization. We have observed that overexpression of tristetraprolin in HeLa cells suppresses the expression of DR5 mRNA.²

The observation that the HuR protein and AUF-1 are both increased by PS-341 treatment complements the observation that AUF-1 is also ubiquitinated and the levels of this protein are controlled by the proteasome (45–48). However, the AUF-1 family of four proteins is complicated, with p37 and p40 isoforms being ubiquitinated and having mRNA destabilizing activity (45–48). When evaluating the stability of mRNAs that contain binding sites for both HuR and AUF-1 proteins (44), the stability or degradation of a specific mRNA was determined by the amount of each protein present in the cytoplasm. Using siRNA targeted at all four AUF-1 isoforms, we did not find major changes in the proteins bound to the DR5 3′-UTR, suggesting that this protein alone does not play a role in regulating the DR5 mRNA. To evaluate the importance of HuR levels alone on TRAIL plus PS-341 killing, we knocked down HuR with siRNA and then treated LNCaP cells with each agent alone or in combination. Although knockdown of HuR did significantly increase cell viability, the change in the number of viable cells was not large (Supplementary Fig. S2).¹ Possible reasons to explain this could be that both the 36-kDa protein HuR and the unknown 45-kDa protein may be coordinately regulating the posttranscriptional process. Both proteins may need to be simultaneously knocked down to see a major effect. Alternatively, PS-341 has multiple other effects, including elevating levels of BH3 proteins Noxa, Bik, and Bim (10, 49), which could be as important to apoptosis as increases in DR5 levels.

In summary, our results suggest a novel mechanism, mRNA stabilization, by which the proteasome inhibitor PS-341 regulates DR5 mRNA and its protein. This work further clarifies the mechanism by which PS-341 enhances sensitivity of cancer cells to the apoptotic agent TRAIL. Modulating the amount or activity of mRNA-binding proteins has great promise for enhancing cancer chemotherapy.

No potential conflicts of interest were disclosed.

We thank Dr. Joan A. Steitz (Howard Hughes Medical Institute, Yale University) for the gift of Flag-HuR plasmid, Dr. Ann-Bin Shyu (University of Texas Health Science Center) for the gift of the rat-GAPDH plasmid that was used to normalize all quantitative PCR, Dr. Baby G. Tholanikunnel (Medical University of South Carolina) for the Sh-RNA plasmid for HuR in the initial part of the study, the Millennium Corp. for providing PS-341, and the Statistical Core of the Hollings Cancer Center, headed by Dr. Elizabeth Garrett-Mayer, which carried out the analysis of these data.