|Home | About | Journals | Submit | Contact Us | Français|
The polysomal ribonuclease 1 (PMR1) mRNA endonuclease forms a selective complex with its translating substrate mRNAs where it is activated to initiate mRNA decay. Previous work showed tyrosine phosphorylation is required for PMR1 targeting to this polysome-bound complex, and it identified c-Src as the responsible kinase. c-Src phosphorylation occurs in a distinct complex, and the current study shows that 90-kDa heat shock protein (Hsp90) is also recovered with PMR1 and c-Src. Hsp90 binding to PMR1 is inhibited by geldanamycin, and geldanamycin stabilizes substrate mRNA to PMR1-mediated decay. PMR1 is inherently unstable and geldanamycin causes PMR1 to rapidly disappear in a process that is catalyzed by the 26S proteasome. We present a model where Hsp90 interacts transiently to stabilize PMR1 in a manner similar to its interaction with c-Src, thus facilitating the tyrosine phosphorylation and targeting of PMR1 to polysomes.
Posttranscriptional regulation of mRNA turnover is a major mechanism regulating gene expression (reviewed in Garneau et al., 2007 ), and work done in the past several years has elucidated the fundamental mechanisms of decay and has begun to clarify regulation of the decay process and its relationship to translation. The decay of most mRNAs begins with poly(A) shortening, followed by exonuclease-mediated decay involving removal of the cap, 5′-3′ decay catalyzed by Xrn1, and 3′-5′ decay catalyzed by the exosome. The identification in 2003 (Sheth and Parker, 2003 ) of processing bodies (P bodies) as focal concentrations of the decapping proteins Dcp1 and Dcp2, Xrn1, and other proteins involved in 5′-3′ mRNA decay led to the early conclusion that these are sites of mRNA decay. More recent work shows P bodies play a larger role in mRNA storage and silencing, and using quite different approaches we (Murray and Schoenberg, 2007 ) and others (Stoecklin et al., 2005 ) showed that unstable mRNAs can decay simultaneously from both ends and that 5′-3′ and 3′-5′ decay are functionally linked. A notable feature of the exonuclease-mediated decay process is that it acts on mRNAs that are no longer bound by translating ribosomes.
For reasons yet to be determined, some mRNAs decay by endonucleolytic cleavage while they are still engaged by ribosomes and undergoing active translation. Because endonucleolytic decay acts on specific mRNAs their targeting to this pathway must be dictated by sequence elements within the mRNAs that are bound by one or more proteins that recruit the endonuclease to the translating messenger ribonucleoprotein (mRNP). Polysomal ribonuclease 1 (PMR1) was the first mRNA endonuclease to be identified (Dompenciel et al., 1995 ), but others include G3BP (Gallouzi et al., 1998 ), IRE1 (Hollien and Weissman, 2006 ), and a yet to be identified erythroid-specific endonuclease (Wang and Kiledjian, 2000 ). PMR1 was initially identified as a polysome-associated ribonuclease activity whose appearance on Xenopus liver polysomes coincides with the disappearance of serum protein mRNAs during estrogen induction of yolk protein gene transcription (Pastori et al., 1991a ,b ). Subsequent work showed that mammalian PMR1 catalyzes the initial steps in the decay of nonsense-containing β-globin mRNA in erythroid cells (Stevens et al., 2002 ; Bremer et al., 2003 ). PMR1 is a member of the peroxidase gene family that is made as an 80-kDa precursor (PMR80) that is processed to the active 60-kDa form (PMR60) (Chernokalskaya et al., 1998 ). Estrogen has no effect on the amount of this protein in Xenopus hepatocytes; rather, it causes a 21-fold increase in unit activity of the polysome-bound enzyme (Cunningham et al., 2001 ).
Our recent work has focused on identifying how selectivity in PMR60 targeting to substrate mRNA is determined. This process can be replicated in transfected mammalian cells, and using this we showed that endonuclease-mediated decay involves sequences on PMR60 required for targeting the ribonuclease to polysomes and ongoing translation of its substrate mRNA (Yang and Schoenberg, 2004 ). This is a selective process that involves formation of an mRNP complex of PMR60 with its translating substrate mRNA. This work was facilitated by the use of a catalytically inactive form of PMR60 (PMR60°), and domain mapping experiments showed that sequences in the C-terminus are required both for polysome targeting and mRNA decay. The C-terminal polysome-targeting domain contains a consensus Src homology 2 site, and tyrosine phosphorylation at position 650 is required for PMR60 targeting to polysomes and mRNA decay (Yang et al., 2004 ). This was the first demonstration of a role for tyrosine kinase signaling in mRNA decay, and our recent identification of c-Src as the responsible kinase (Peng and Schoenberg, 2007 ) raises the possibility that the transforming activity of this protooncogene may in part result from increased decay of mRNAs encoding proteins that regulate cell growth.
Initial evidence linking c-Src to PMR60 came from experiments in which immunoprecipitated myc-tagged PMR60° was incubated in vitro with [γ-32P]ATP (Peng and Schoenberg, 2007 ). Tyrosine kinases commonly form a complex with their substrates and undergo autophosphorylation, and this experiment resulted in 32P labeling of three proteins: PMR60°, c-Src, and an unidentified 90-kDa protein. Here, we identify this protein as 90-kDa heat shock protein (Hsp90), we show that this interaction is required for endonuclease-mediated mRNA decay, and we show that PMR60 is an inherently unstable protein that upon inhibition of Hsp90 is rapidly degraded by the proteasome.
Monoclonal antibodies to the c-Myc epitope tag (9E10), Hsp70 (W27), myc antibody-coupled agarose beads, rabbit polyclonal antibodies to Hsp90 (H-114) and actin (C-11) were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Mouse monoclonal antibody (mAb) (AC88) to Hsp90 was purchased from Abcam (Cambridge, MA), and horseradish peroxidase-coupled rabbit anti-mouse IgG and goat anti-rabbit IgG were purchased from Santa Cruz Biotechnology.
Cos-1 cells were cultured in DMEM plus 10% fetal bovine serum and 2 mM l-glutamine. U2OS cells were maintained in McCoy's 5A medium plus 10% fetal bovine serum. Cos-1 and U2OS cells in log phase growth were transfected using FuGENE 6 (Roche Diagnostics, Indianapolis, IN) or Lipofectamine (Invitrogen, Carlsbad, CA) following the manufacturer's protocols. Geldanamycin (GA) was purchased from InvivoGen (San Diego, CA) and dissolved in dimethyl sulfoxide (DMSO). Proteasome inhibitor pyrazylcarbonyl-Phe-Leuboronate (PS-341; Millennium Pharmaceuticals, Cambridge, MA) was dissolved in DMSO. To generate a line of cells stably expressing PMR60°-tandem affinity (TAP) tag, U2OS cells were transfected with plasmid pcDNA3-myc-PMR60°-TAP (Yang and Schoenberg, 2004 ), and the monolayer was trypsinized 24 h later and transferred into a 100-mm-diameter culture dish. A mixed culture of stably transfected cells was then selected by 2-wk growth in medium containing 100 μg/ml Geneticin (G418 sulfate; Invitrogen).
Cytoplasmic extracts were prepared as described previously (Yang and Schoenberg, 2004 ). Briefly, cells were washed twice with ice-cold phosphate-buffered saline, and then they were suspended in cell lysis buffer [10 mM HEPES-KOH, pH 7.5, 10 mM KCl, 5 mM MgCl2, 50 mM NaF, 0.5% Nonidet P-40 (vol/vol), 2 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, and 25 μl/ml protease inhibitor mixture (Sigma-Aldrich, St. Louis, MO) and 10 μl/ml phosphatase inhibitor cocktail (Sigma-Aldrich)]. After incubation for 15 min on ice, the cells were lysed with 30 strokes of a Dounce homogenizer (A pestle), and nuclei were removed by centrifugation for 15 min at 2000 × g. Where indicated, RNase A was added at the final concentration of 50 ng/ml followed by 30 min on ice before immunoprecipitation (Yang et al., 2006 ).
Cos-1 cells (9 × 106) were harvested 40 h after transfection with plasmids pcDNA3-GFP-TAP or pcDNA3-Myc-PMR-TAP. Cytoplasmic extracts in 4 ml of lysis buffer were incubated with rocking for 2 h at 4°C with 200 μl of Fast-Flow IgG Sepharose 6 (GE Healthcare, Little Chalfont, Buckinghamshire, United Kingdom), followed by four washes with 1 ml of IPP150 buffer (10 mM Tris-HCl, pH 7.5, 150 mM NaCl, and 0.1% NP-40) and two washes with 1 ml of Tev cleavage buffer (10 mM Tris-HCl, pH 7.5, 150 mM NaCl, 0.1% NP-40, 0.5 mM EDTA, and 1 mM dithiothreitol). Bound protein was released by incubating with 50 U of Tev protease (Invitrogen) for 2 h at 25°C in 200 μl of TEV cleavage buffer, followed by centrifugation at 1000 × g to pellet the beads. These were washed twice with 400 μl of Tev cleavage buffer, and the washes and initial eluate were combined. Thirty microliters was removed for analysis by SDS-polyacrylamide gel electrophoresis (PAGE) and silver staining (Figure 1A), and the remaining sample was trichloroacetic acid precipitated, dissolved in 50 μl of SDS sample buffer, and applied to a 10% SDS-PAGE gel. The 90-kDa band identified by Coomassie Blue staining was excised and placed into in 5% acetic acid in water to prevent bacterial contamination. This was digested with trypsin, and 19 tryptic fragments were identified as Hsp90 by liquid chromatography-tandem mass spectrometry (LC-MS/MS) in The Ohio State University Mass Spectrometry and Proteomics Facility.
Cells (2 × 106) in a 10-cm dish were collected by scraping and lysed as described above in 0.5 μl of lysis buffer. For immunoprecipitation with myc antibody, 450 μl of cell lysate was incubated with 10 μl of a 50% suspension of mAb-coupled agarose beads on a rocking platform for 3 h at 4°C. For immunoprecipitation with rabbit anti-Hsp90 antibody, 450 μl of cell lysate was incubated with 30 μl of rabbit polyclonal antibody for 3 h at 4°C, followed by overnight incubation at 4°C on a rocking platform with 30 μl of protein A-agarose (Santa Cruz Biotechnology). The beads were washed four times with IPP150 buffer, and then they were suspended in SDS sample buffer. For Western blot analysis, the immunoprecipitates were separated on a 10% SDS-PAGE gel and electroblotted onto Immobilon-P membrane (Millipore, Billerica, MA). The membrane was blocked for 1 h at 25°C in 5% nonfat dry milk in Tris-buffered saline/Tween 20 buffer (20 mM Tris-HCl, pH 7.5, 150 mM NaCl, and 0.1% Tween 20), and then it was incubated with the primary antibody for 4 h at 25°C, washed, and incubated with horseradish peroxidase-conjugated secondary antibody for 1 h. Blots were developed with SuperSignal West Pico Chemiluminescent Substrate (Pierce Chemical, Rockford, IL).
U2OS cells (8 × 105) were transiently transfected with plasmids expressing albumin and luciferase mRNA plus empty vector (pcDNA3), or plasmid expressing catalytically active myc-PMR60 (Peng and Schoenberg, 2007 ). DMSO or 1 μM geldanamycin was added the next day, and total RNA was isolated 24 h later using TRIzol reagent (Invitrogen). The antisense albumin riboprobe was prepared with the MAXIscript In Vitro Transcription kit (Ambion, Austin, TX) by using T7 promoter from a pcRII-Topo plasmid containing exons 14 and 15 of Xenopus albumin cDNA. The antisense firefly luciferase riboprobe was synthesized with T3 promoter from a pBluescript(SK) plasmid containing the first 153 nucleotides of firefly luciferase cDNA. Ribonuclease protection assay was done as described previously (Yang and Schoenberg, 2004 ) with 5 μg of total RNA hybridized to 600 pg of each riboprobe by using the Ribonuclease Protection Assay III kit (Ambion). Protected probe was separated on a denaturing 6% polyacrylamide-urea gel and quantified by PhosphorImager (GE Healthcare) analysis.
A plasmid expressing PMR60° with a C-terminal tandem affinity tag (PMR60°-TAP; Yang and Schoenberg, 2004 ) was used to increase the recovery efficiency of proteins associated with PMR60°. This was transfected into Cos-1 cells and cytoplasmic complexes bound to IgG-Sepharose through the TAP tag were eluted by Tev protease cleavage. The recovered proteins were separated on an SDS-PAGE gel, and the 90-kDa band identified by Coomassie Blue staining was excised, digested with trypsin, and analyzed by LC-MS/MS. Nineteen peptides were sequenced, each of which identified the 90-kDa band as Hsp90 (Figure 1A). To confirm that Hsp90 is a PMR60-binding protein, complexes recovered with immobilized myc mAb were washed with increasing concentrations of NaCl before elution with SDS sample buffer and analysis by Western blot (Figure 1B). In agreement with the preceding results, Hsp90 was recovered with PMR60° but not with a green fluorescent protein (GFP) control. Eighty-six percent of the Hsp90 remained after washing with 300 mM NaCl and 55% remained after washing with 450 mM NaCl. Even after 600 mM NaCl, 32% of Hsp90 seen in the control sample remained bound to PMR60°, evidence of a strong interaction between these proteins.
RNase digestion is commonly used to determine whether two proteins interact directly or are recovered through shared binding to RNA (Yang et al., 2006 ). Results in Figure 1C show that prior treatment with RNase A has no impact on the recovery of Hsp90 with immunoprecipitated myc-PMR60°, supporting a direct interaction between these proteins. Final confirmation that Hsp90 is a PMR60°-binding protein is shown by the reciprocal immunoprecipitation of myc-PMR60° but not myc-GFP with antibody to Hsp90 (Figure 1D).
The different functional domains of PMR60 are shown at the top of Figure 2A. The central portion of the protein makes up the catalytic core, and PMR60 targeting to polysomes depends on c-Src phosphorylation of Y650 in the C-terminal domain. c-Src binds to two repeating PXXP motifs (Peng and Schoenberg, 2007 ), and the N-terminal 50 amino acids has a unique stress-responsive domain that recruits PMR60° to stress granules by stress-induced binding to TIA-1 (Yang et al., 2006 ). In addition, PMR60° binds to the cytoskeleton-regulatory proteins mammalian enabled and vasodilator activated phosphoprotein through a domain that lies between the stress-responsive domain and the c-Src binding sites (Peng, et al., manuscript in preparation). To identify domains that may be involved in binding Hsp90, Cos-1 cells were transfected with the battery of myc-tagged N- and C-terminal deletions diagrammed in Figure 2A together with myc-GFP, and protein was recovered with immobilized myc antibody was analyzed by Western blot with antibodies to Hsp90 and the myc tag (Figure 2B). Hsp90 was recovered with full-length PMR60° (lane 2) but not GFP (lane 1), and its recovery was unaffected by any of the N- or C-terminal deletions (lanes 3–8). However, Hsp90 is not recovered with protein lacking the central domain of PMR60° (ΔM, lane 9). Because the corresponding central portion of PMR60° deleted from the ΔM construct is unstable, we cannot state with certainty whether this is the principal site of Hsp90 binding, or whether binding was lost due to changes in protein folding.
The N-terminal portion of Hsp90 is an ATP binding domain whose function is essential for its interaction with client proteins. The binding here of ansamycin antibiotics (e.g., GA) blocks the binding of ATP and inhibits these interactions (Prodromou et al., 1997 ; Stebbins et al., 1997 ). To determine whether PMR60° is an Hsp90 client protein, U2OS cells transiently expressing PMR60° were treated for 24 h with GA or DMSO, and cytoplasmic protein recovered by immunoprecipitation with immobilized myc antibody was analyzed by Western blot for PMR60° and Hsp90 (Figure 3A). GA treatment reduced the amount of input PMR60° by 20%, but more strikingly it reduced the binding of Hsp90 by 70%, indicating that PMR60° is an Hsp90 client protein.
To determine whether the interaction of Hsp90 with PMR60 influences endonuclease-mediated mRNA decay, U2OS cells transiently expressing catalytically active PMR60, albumin, and luciferase mRNA were treated with GA, DMSO, or left untreated, and then RNA was recovered 24 h later and analyzed by RNase protection assay (Figure 3B). The amount of albumin mRNA in each sample was normalized to luciferase mRNA by PhosphorImager analysis (Yang and Schoenberg, 2004 ), and the quantified results are shown beneath the autoradiogram. Albumin mRNA is reduced to 40% of control by coexpression of PMR60 (lane 4), and this was unaffected by DMSO; however, 1 μM GA inhibited PMR60-mediated mRNA decay, returning the level of albumin mRNA to 75% of control. Similar results were seen with 2 and 3 μM GA (data not shown). These data indicate that disrupting the interaction of PMR60 with Hsp90 reduces the efficiency of endonuclease-mediated mRNA decay, most likely by interfering with the proper folding of PMR60.
Alterations in protein folding in GA-treated cells commonly result in reduced levels of Hsp90 client proteins (e.g., Doong et al., 2003 ). This is best seen with endogenous proteins or stably transfected cells because protein overexpression that commonly occurs with transient transfection can dampen the impact of GA on the degradation of Hsp90 client proteins (Kim et al., 2006 ). To examine more carefully the interaction of Hsp90 with PMR60° and the impact of GA on this, we generated a line of U2OS cells stably expressing catalytically inactive PMR60° with a TAP tag and we used Tev protease cleavage to recover complexes from IgG-Sepharose. Results in Figure 4A confirm that Hsp90 is selectively recovered with PMR60°. Note that Tev protease cleaves within the TAP tag to generate the smaller form of PMR60° seen in lane 4.
These cells were used to test whether results using transient transfection masked a more significant role for Hsp90 in controlling the folding and stability of PMR60°. In the experiment shown in Figure 4B, PMR60°-TAP–expressing cells were treated for 24 h with increasing amounts of GA, and total protein was analyzed by Western blot. PMR60°-TAP disappeared in a concentration-dependent manner, with 50% lost with 0.1 μM GA and 85% with 1 μM GA. GA had no impact on actin in these samples, and it increased the amount of Hsp90 at concentrations ≥1 μM. These data indicate that PMR60° is indeed an Hsp90 client protein, and they point to a disruption in PMR60 folding as the likely cause of the GA-induced stabilization of albumin mRNA in Figure 3B.
Cycloheximide inhibition of translation is commonly used to study protein turnover, but because it sequesters PMR60° in the polysome-bound complex with its substrate mRNA it cannot be used to study PMR60° protein decay. Instead, we looked at the rate at which GA destabilizes PMR60° in the stable U2OS cell line (Figure 4C). In this experiment, cells were treated with (lanes 2–6) or without 1 μM GA (lanes 7–11) for 2–24 h, and changes in PMR60°-TAP, Hsp90, and β-actin were monitored by Western blot. As in the preceding experiment, there was a small increase in Hsp90 by 12 h in GA and no change in β-actin. However, 83% of PMR60°-TAP disappeared within 2 h of GA addition, and it remained at this repressed level throughout the next 24 h. Based on these results, we conclude that PMR60° binding to Hsp90 is required for its proper folding, and inhibiting this process with GA results in the rapid disappearance of PMR60°.
For many Hsp90 client proteins, the misfolding caused by GA inhibition of ATP binding results in their degradation by the 26S proteasome (e.g., Akt [Doong et al., 2003 ], Raf1 [Stancato et al., 1997 ], and cystic fibrosis transmembrane conductance regulator [Loo et al., 1998 ]). To determine whether this accounts for the lower amount of PMR60° seen after GA in Figure 4, we examined the impact of the highly selective proteasome inhibitor PS-341 (Kisselev and Goldberg, 2001 ; Kisselev et al., 2006 ) (Figure 5A). In this experiment, PMR60°- TAP–expressing U2OS cells were incubated for 1 h in the medium containing increasing concentrations of PS-341, followed by 16 h in 1 μM GA. PS-341 inhibited the GA-induced destabilization PMR60° in a dose-dependent manner, indicating that the 26S proteasome catalyzes the degradation of PMR60°. Inhibition of lysosomal proteases with chloroquine and NH4Cl had no impact on GA-induced decay (data not shown).
In addition to its impact on Hsp90-associated protein folding GA increases the amount of Hsp70 (Lawson et al., 1998 ), and in GA-treated cells the binding of Hsp70 is associated with proteasome-mediated degradation of several Hsp90 client proteins (Bonvini et al., 2004 ; Kim et al., 2006 ). To determine whether this applies to PMR60°, cells stably expressing PMR60°-TAP were treated with 1 μM GA for 1–6 h, and the complexes recovered by Tev protease cleavage after binding to IgG-Sepharose were analyzed by Western blotting with antibodies to the myc tag on PMR60° and Hsp70 (Figure 5B). The GA-induced increase in Hsp70 over time in lanes 1–5 mirrors the loss of PMR60°, and little Hsp70 was recovered with PMR60° from control (DMSO-treated) cells (lane 6). Unexpectedly, the kinetics of Hsp70 binding did not match the loss of PMR60°, and increased recovery of Hsp70 was only seen after the majority of PMR60° was lost. The reason for this is not known, although it may reflect differences in turnover of PMR60 present in the different complexes in which it resides in the cell.
Previous reports linking mRNA decay to proteasome-mediated protein turnover looked at the stabilization of an ARE-containing mRNA resulting from the degradation of the 37-kDa isoform of Auf1 (hnRNP D; Laroia et al., 1999 , 2002 ). The results presented here show that the PMR1 mRNA endonuclease is an Hsp90 client protein. Treating cells with a drug that inhibits Hsp90 client protein folding disrupts the interaction between Hsp90 and PMR1, stabilizes PMR1 target mRNA, and activates the rapid proteasome-mediated degradation of PMR1. The model in Figure 6 incorporates these results with our current view of PMR1-mediated mRNA decay.
The properties of the PMR1 mRNA endonuclease are consistent with its role as a functional integrator of signal transduction. The 60-kDa active form of PMR1 (PMR60) is found primarily in two complexes. Complex I is an mRNP that contains PMR60, its substrate mRNA, and the protein(s) (shown in red) that bring these together. This is bound by translating ribosomes, and it is here that endonuclease-mediated decay takes place (Yang and Schoenberg, 2004 ). Complex I is released from polysomes by treating cells with puromycin before lysis or by adding EDTA to cytoplasmic extracts, and the dissociated complex sediments at ~680 kDa on glycerol gradients. PMR60 must be ‘licensed’ to join complex I and participate in mRNA decay by phosphorylation of a tyrosine (Y650) located in a defined polysome-targeting domain (Yang et al., 2004 ). This occurs in an ~140-kDa complex (complex II), the principal components of which are PMR60 and its activating tyrosine kinase c-Src (Peng and Schoenberg, 2007 ). c-Src phosphorylation of PMR60 can be activated by epidermal growth factor (EGF); however, this is likely just one of several signaling events that participate in this process.
We initially identified c-Src as a 60-kDa protein that was phosphorylated in vitro along with PMR60 and an unknown 90-kDa protein when immunoprecipitated PMR60 was incubated with [γ-32P]ATP (Peng and Schoenberg, 2007 ). The data in Figure 1 identify the 90-kDa protein as Hsp90, and subsequent experiments using GA, a compound that interferes with Hsp90 client protein folding by occupying the ATP binding, indicate that transient association of Hsp90 with PMR60 is required for its proper folding and accumulation within the cell. GA caused a modest reduction in the amount of PMR60° in transiently transfected cells (Figure 3A), but in cells that stably express PMR60 83% is lost within 2 h (Figure 4C). These results are consistent with protein overexpression in transiently transfected cells blunting the degradation of misfolded PMR60.
Hsp90 is a key integrator of several signal transduction pathways, and the effects of GA on mRNA decay in Figure 3B and on PMR60 in Figure 4C suggest that signaling through Hsp90 may also regulate endonuclease-mediated decay. The stabilization of PMR60° by the selective proteasome inhibitor PS-341 (Figure 5A) indicates that, like many Hsp90 client proteins, PMR60 is degraded by the proteasome. For some of these proteins, Hsp90 is first replaced by Hsp70 (Kim et al., 2006 ); however, this does not seem to be the case for PMR60°, because increased binding of Hsp70 occurs after most of it has already been degraded (Figure 5B).
Finally, we noted previously that c-Src is elevated in many cancers, and we proposed this may enhance tumor cell growth by promoting the decay of PMR60 substrate mRNAs, some of which may encode growth-regulatory proteins (Peng and Schoenberg, 2007 ). Hsp90 stabilizes c-Src in a manner similar to that of PMR60 (Xu et al., 1999 ), and its involvement with both of these proteins is consistent with this hypothesis. Furthermore, these findings raise the possibility that Hsp90, c-Src inhibitors, or a combination may inhibit tumor cell growth in part by inhibiting endonuclease-mediated mRNA decay.
We thank members of the Schoenberg lab for helpful comments on this work. This work was supported by National Institutes of Health grant GM-38277 (to D.R.S.). Support for core facilities was provided by center grant P30 CA-16058 to The Ohio State University Comprehensive Cancer Center.
This article was published online ahead of print in MBC in Press (http://www.molbiolcell.org/cgi/doi/10.1091/mbc.E07-08-0774) on November 28, 2007.