Search Technical Information:

Minireviews

The ERK Signal Transduction Pathway

Date 1/1/2004

Its Role in Growth Factor Signaling and Cancer

First printed in R&D Systems' 2004 Catalog.

Contents

Growth factors, through receptor tyrosine kinases, recruit a large network of signaling proteins to execute their cellular programs.

The first of these networks to be discovered was the Ras-Raf-ERK signal transduction cascade, defined by extracellular signal-regulated kinase-1 (ERK1) and ERK2.1 One of four mitogen-activated protein kinase (MAPK) signaling pathways, the ERK phosphorylation cascade's importance in intracellular signaling has been compared to the role of the Krebs cycle in energy metabolism.2 The ERK cascade functions in cellular proliferation, differentiation, and survival, and its inappropriate activation is a common occurrence in human cancers.

Figure 1. The activated membrane-spanning epidermal growth factor receptor (EGF R) becomes a platform for the assembly of a signaling complex that includes the cytoplasmic growth factor receptor bound protein 2 (Grb2) and son of sevenless (SOS), which activates the membrane-bound GTPase, Ras.

Upstream from ERK

During growth factor stimulation, the ERK phosphorylation cascade is linked to cell surface receptor tyrosine kinases (RTKs) and other upstream signaling proteins with known oncogenic potential (Figure 1).3

RTKs

Membrane spanning cell surface receptors of the RTK family are endowed with intrinsic tyrosine kinase activity, catalyzing the transfer of the a-phosphate of ATP to the hydroxyl groups of tyrosines on target proteins. All RTKs contain a frequently glycosylated extracellular ligand binding domain, connected through a single transmembrane helix to the cytoplasmic domain. The cytoplasmic domain contains a conserved protein tyrosine kinase (PTK) core and regulatory regions that are subjected to autophosphorylation and phosphorylation by other kinases.4 Nearly all RTKs are monomers at the cell membrane, with ligand binding or ectopic overexpression resulting in receptor dimerization and tyrosine autophosphorylation in trans.

As the first RTK to be discovered,5 the epidermal growth factor receptor (EGF R, also known as ErbB1 from the v-erb-B transforming protein of an avian retrovirus) has helped establish many of the principles of RTK function.6 EGF R forms an RTK subfamily with three other closely related receptors: the orphan receptor ErbB2 (also known as HER2 and neu); ErbB3 (also known as HER3 and characterized by an impaired kinase domain); and ErbB4 (also known as HER4).7 Activating mutations and transforming overexpression, mimicking receptor oligomerization of EGF R and its fellow family members, have been implicated in numerous cancers, including mammary carcinomas, squamous carcinomas and glioblastomas,3 and will be our model RTK here for further analysis.

Adaptors

The EGF receptor contains at least nine tyrosine residues capable of phosphorylation in its cytoplasmic domain, and seven of these are autophosphorylation sites.8 Tyr autophosphorylation sites on EGF R and other RTKs provide a mechanism for the recognition and assembly of signaling complexes, functioning as binding sites for Src homology 2 (SH2) and phosphotyrosine binding (PTB) domains of a variety of signaling proteins.4 Therefore, activated RTKs become a platform for the recognition and recruitment of a specific complement of signaling proteins. One such signaling protein is growth factor receptor bound protein 2 (Grb2). Grb2, a cytosolic adaptor, contains a central SH2 domain flanked by two Src homology 3 (SH3) domains that allow it to constitutively associate with the proline-rich regions of the nucleotide exchange factor son of sevenless (SOS).9 Phosphotyrosine 1068 of the activated EGF R is a binding site for the SH2 domain of Grb2,10 either directly or through the assistance of another SH2 adaptor, Shc.11 The recruitment of Grb2 from the cytoplasm to the plasma membrane brings SOS near the membrane-bound product of proto-oncogene c-ras. Through guanine exchange, SOS enhances GDP release and GTP binding to Ras, converting this GTPase into its active conformation.

Ras

Ras is a notable member of the large family of GTPases, proteins that bind and hydrolyze GTP. First discovered as transforming oncogenes of murine sarcoma viruses, three highly related 21 kDa mammalian proteins, Harvey-Ras (H-Ras), Kirsten-Ras (K-Ras), and Neuroblastoma-Ras (N-Ras) have been identified.12 Activating mutations of these Ras isoforms, which impair GTPase activity and stabilize the GTP bound state, are found in nearly one-third of all human cancers, making these oncoproteins among the most potent transforming polypeptides known.1

Ras family members are anchored to the cytoplasmic face of the plasma membrane by carboxyl-terminal farnesylation. This localization to the inner leaflet brings Ras into close proximity with SOS, stimulating the exchange of GDP bound to Ras with GTP from the cytosol. This exchange activates Ras conformationally, allowing it to interact with a number of downstream effectors.13 Within the ERK signaling cascade, active Ras functions as an adaptor that binds to effector Raf kinases with high affinity, causing their translocation to the cell membrane, where Raf activation takes place.14

Figure 2. The phosphorylation cascade regulates the core module, including MAP3K Raf, MAPK kinase MEK, and MAPK ERK.

The Core Module

Activation of all MAPKs is regulated by a central three-tiered core signaling module, comprised of an apical MAPK kinase kinase (MAP3K), a MAPK kinase (MEK or MKK), and a downstream MAPK (Figure 2).

Raf

Like Ras, the MAP3K Raf was first discovered in the form of a mutant retroviral transforming agent, v-raf.15 Raf is a Ser/Thr protein kinase, catalyzing the phosphorylation of hydroxyl groups on specific Ser and Thr residues.16 Mammals possess 3 Raf proteins, ranging from 70 to 100 kDa in size: the ubiquitously expressed and best studied Raf-1; the more obscure A-Raf, with high levels in muscle and urogenital tissues;17 and B-Raf, present in multiple isoforms and strongly expressed in fetal brain and adult cerebrum.18 In addition to their different cell-specific expression, the phenotypes of raf knock-out mice also vary. A-raf deficient mice survive birth with intestinal and neurological abnormalities,19 while mice with a targeted disruption of the B-raf gene die of vascular defects during mid-gestation.20 Raf-1-/- mice also die in utero with a number of developmental defects.21 These results indicate that Raf-1 may serve a general role in tissue formation, whereas A-Raf and B-Raf fulfill more specialized duties.

Recruitment to the plasma membrane by GTP-bound Ras is the initiating event in Raf activation. The effector domain of Ras binds Raf at two locations in the MAP3K's N-terminus, the Ras-binding domain (RBD) and the cysteine-rich domain (CRD), with binding at both sites necessary for activation.22 Different Ras isoforms appear to activate Raf with varying ability, despite binding in vitro with comparable affinity. For example, K-Ras both recruits Raf-1 to the plasma membrane more efficiently, and activates the recruited Raf-1 more potently than H-Ras.23 It has also been suggested that B-Raf is the primary target of oncogenic Ras isoforms.24 Recently, activating mutations of B-raf were reported in 66% of malignant melanomas, all within the kinase domain and most converting Val599 to Glu.25

Raf-1 has four activating phosphorylation sites that are Ras-inducible and confer full Raf activation: Ser338, Tyr341, Thr491 and Ser494. Mutation of these sites to phosphomimetic acidic residues results in constitutive activity independent of Ras.16 Recently, p21-activated protein kinase (PAK) family members have been shown to phosphorylate Raf at Ser338 and induce activity,26 though these findings have been disputed.27 Other kinases that phosphorylate these activating residues have not been identified, and are the subject of considerable speculation. Phosphorylation also inhibits Raf. Raf-1 phosphorylated at Ser259 by Akt isoforms appears to create an auto-inhibitory conformation state maintained by 14-3-3 dimers,28 and mutation of this Ser to Ala restores the basal activity of Raf-1. The Ser/Thr phosphatases PP1 and PP2A target Ser259, thus promoting Raf-1 activation.29 Inhibition of these phosphatases increases inactive 14-3-3/Raf-1 complexes.

MEK

Phosphorylated Raf activates MAPK/ERK kinase 1 (MEK1) and MEK2, also known as MKK1 and MKK2.30 Surprisingly, no naturally occurring oncoproteins derived from MEKs have been found, although expression of constitutively active forms transforms fibroblasts that produce tumors in nude mice.31 Disruption of mouse mek1 is lethal in utero, with mutant embryos dying from defective placental vascularization, suggesting a role for MEK in angiogenesis.32 MEK1 and MEK2, about 45 kDa each, share 80% sequence identity. It is unclear why two MEKs exist, although conservation of both forms throughout eukaryotic species suggests non-redundant functions, as does MEK1 gene disruption. Both MEKs are expressed ubiquitously in mammalian cells at micromolar levels, although some tissue-specific variation has been noted.33

Raf family activation of MEK1 and MEK2 occurs through phosphorylation of two Ser residues at positions 217 and 221 found in the activation loop.34 MEKs can be partially activated by phosphorylation at either site, and substitution of these sites with acidic amino acids enhances basal activity.31 While Raf isoforms are enzymes of relatively low abundance, the high concentration of MEKs allows for amplification of signaling.35 Different Raf isoforms activate MEK1 and MEK2 differentially: A-Raf is a weak activator; B-Raf activates MEK1 preferentially; and Raf-1 efficiently activates both MEKs.36 Raf-1 contains two separate MEK binding sites, with substrate interaction greatly enhanced following phosphorylation of Raf-1 at Ser338.37 Two regulatory phosphorylation sites on MEK outside the activation loop either positively or negatively regulate the MAPKK. The first, at Ser298 and phosphorylated by PAK1, may help prime MEK1 for activation by Raf-1.38 Conversely, in vivo phosphorylation by an unknown kinase at Ser212, a site conserved in all MAPKKs, sharply decreases MEK1 activity.39

ERK

Also known as MAPK3 and MAPK1, the MAP kinases ERK1 and ERK2 are 44- and 42-kDa Ser/Thr kinases, respectively, with 90% sequence identity in mammals. Initially isolated and cloned as kinases activated in response to insulin and NGF,40,41 ERK1 and ERK2 are both expressed in most, if not all, mammalian tissues, with ERK2 levels generally higher than ERK1. Knock-out studies in mice demonstrate that either ERK may at least partially compensate for the other's loss, although ERK1 has been found to specifically regulate thymocyte maturation.2 Dual Thr and Tyr phosphorylations activate both ERKs, at Thr202/Tyr204 for human ERK1 and Thr185/Tyr187 for human ERK2. Unlike MEK, significant ERK activation requires phosphorylation at both sites, with Tyr phosphorylation preceding that of Thr.42 As with MEK, no in vivo mutations have been identified that activate ERK.

Because ERKs and other MAPKs require both Thr and Tyr phosphorylation for full activity, dual specificity phosphatases (DSPs) that dephosphorylate both sites are uniquely positioned to regulate MAPK signal transduction cascades. At least 9 DSPs, also termed MAPK phosphatases (MKPs), have been identified in mammalian cells.43 DSPs frequently associated with ERK inactivation include MKP3, MKP4, and phosphatase of activated cells 1 (PAC1). MKP3, also termed PYST1, is probably the best-studied DSP, present in many tissues and most specific for ERKs versus other MAPKs. MKP4, expressed in kidney, placenta, and embryonic liver, strongly dephosphorylates ERKs but shows some reactivity toward JNK and p38 as well. The hematopoietically expressed PAC1 also shows limited reactivity with JNK and p38 in addition to ERKs, and is upregulated transcriptionally by p53.44 In addition to DSPs, the phosphatases PP2A and HePTP have been implicated in ERK2 dephosphorylation at Thr185 and Tyr187, respectively.45 The finding that multiple phosphatases inactivate ERKs suggests that the duration and extent of ERK activation is controlled by the balanced activities of MEKs and these phosphatases.

Recent studies have revealed scaffolding as a mechanism that helps the ERK cascade transduce signals with both high efficiency and specificity. First discovered in lower eukaryotes, experiments in mammals have focused on two scaffolding proteins in particular, kinase suppressor of Ras (KSR) and MEK partner 1 (MP1). KSR appears to tether Rafs, ERKs and particularly MEKs, redistributing MEK from a soluble fraction into a high molecular weight complex.46 Experiments with KSR-deficient mice indicate that KSR is not absolutely required, but enhances signaling from Ras.47 The scaffolding protein MP1 tethers MEK with ERK, and has enhanced specificity for MEK1 and ERK1 over MEK2 and ERK2, favoring the activation of ERK1.48 The physiological significance of this differential activation is not understood, but reduction of MP1 using RNA interference results in defective ERK signal transduction.49

Downstream from ERK

The ERKs are proline-directed protein kinases, phosphorylating Pro-neighboring Ser or Thr residues. Docking sites present on physiological substrates confer additional specificity.50 These docking interactions, through non-catalytic regions on ERK, team with scaffolding proteins to ensure signaling fidelity and enzymatic efficiency both to and from the MAPK. Downstream, activated ERK regulates growth factor-responsive targets in the cytosol and also translocates to the nucleus where it phosphorylates a number of transcription factors regulating gene expression (Figure 3).

Figure 3. ERK1 and ERK2 feed back on the ERK signal transduction pathway; initiate transcription directly or indirectly via phosphorylation of ribosomal protein S6 kinases (RSKs), mitogen- and stress-activated protein kinases (MSKs), and ternary complex factors (TCFs); and regulate translation indirectly by inducing tRNA and rRNA synthesis.

Cytoplasmic Targets

Cytosolic substrates for ERK include several pathway components involved in ERK negative feedback regulation. Multiple residues on SOS are phosphorylated by ERK following growth factor stimulation.51 SOS phosphorylation destabilizes the SOS-Grb2 complex, eliminating SOS recruitment to the plasma membrane and interfering with Ras activation of the ERK pathway. Negative feedback by ERK has also been proposed to occur through direct phosphorylation of the EGF receptor at Thr669.52 The physiological importance of this phosphorylation is currently unclear, however. Finally, ERKs have also been demonstrated to negatively regulate themselves by phosphorylating MKPs, which reduces the degradation of these phosphatases through the ubiquitin-directed proteasome complex.53

MAPK-interacting kinase 1 (MNK1) and MNK2 are cytosolic Ser/Thr protein kinases initially discovered using two-hybrid screens for ERK-interacting proteins.54 Both ERK and p38, but not JNK, activate MNK by phosphorylation at Thr197 and Thr202. Active MNK1, and possibly MNK2, upregulates eukaryotic initiation factor-4E (eIF-4E) in vitro, through phosphorylation at Ser209. This upregulation under physiological conditions has been questioned, however.55 Phosphorylation of eIF-4E at Ser209 is believed to strengthen this initiation factor's affinity for 7-methylguanosine cap structures, directing ribosomes to the 5' ends of mRNAs and enhancing translation efficiency.

ERK1 and ERK2 regulate transcription indirectly by phosphorylating the 90 kDa ribosomal protein S6 kinases (RSKs), a family of broadly expressed Ser/Thr kinases activated in response to mitogenic stimuli, including growth factors and tumor-promoting phorbol esters.56 A highly conserved feature common to all RSK family members is the presence of tandem non-identical catalytic domains, involved in both exogenous phosphorylation and auto-activation.57 These domains are activated in a sequential manner by a series of phosphorylations following the binding of active ERK1 or ERK2 to an ERK docking site located at the extreme carboxyl terminus of cytoplasmic RSK.58 On RSK1, these sequential phosphorylation sites include Thr573, Ser380 and Ser221. Active RSKs appear to play a major role in transcriptional regulation, translocating to the nucleus and phosphorylating such factors as the product of proto-oncogene c-fos at Ser362, serum response factor (SRF) at Ser103, and cyclic AMP response element-binding protein (CREB) at Ser133.59,60 Although RSK1 was initially purified and named based on its ability to phosphorylate the ribosomal protein S6 in vitro, this translational component is apparently the physiological substrate for the p70 S6 kinases, and not the RSKs.61

Nuclear Targets

Upon phosphorylation, nuclear translocation of ERK1 and ERK2 is critical for both gene expression and DNA replication induced by growth factors.62 In the nucleus, ERK phosphorylates an array of targets, including transcription factors and a family of RSK-related kinases, the mitogen- and stress-activated protein kinases (MSKs).63 MSK1 and MSK2, activated by both ERK and p38, share the same tandem kinase structure as the RSKs, and also appear to be activated by sequential phosphorylation following MAPK docking. On MSK1, these sequential sites include Thr581, Ser376 and Ser212. MSKs phosphorylate and activate the AP-1 component ATF1 at Ser63,64 and may be more important in vivo than RSKs in CREB phosphorylation at the activating Ser133.65 Using knock-out mice, MSKs were also found to phosphorylate Histone H3 at Ser10 and Ser28, and the high-mobility-group protein HMG-14 at Ser6, facilitating the rapid induction of immediate early genes following mitogenic stimulation.66

Probably the best-characterized transcription factor substrates of ERKs are ternary complex factors (TCFs), including Elk-1, which is directly phosphorylated by ERK1 and ERK2 at multiple sites, including the activating Ser383.67 Upon complex formation with SRF, phosphorylated TCFs transcriptionally activate the numerous mitogen-inducible genes regulated by serum response elements (SREs).68 The SREs regulating these genes are sufficient to confer inducibility to EGF, as well as to serum and phorbol esters. TCFs Sap1 and Sap2 are also phosphorylated by ERK, as are other Ets family members.69 Another direct target of ERK, at least in vitro, is the product of proto-oncogene c-myc, a short-lived transcription factor involved in multiple aspects of growth control. Following phosphorylation at Thr58 and Ser62 within its transactivation domain, Myc activates transcription as a heterodimeric partner with Max.70

Finally, cellular growth and proliferation require protein synthesis, and the ERK pathway has been recently demonstrated to directly link growth factor signaling to ribosome biogenesis. Following serum induction, ERK phosphorylates the BRF1 subunit of RNA polymerase (pol) III-specific transcription factor TFIIIB, both in vitro and in vivo, at an unknown site.71 As with MNK activation, phosphorylation of this pol III subunit enhances translational efficiency, inducing tRNA and 5S rRNA synthesis. Previous experiments demonstrated that ERK also upregulates synthesis of ribosomal RNA by pol I through phosphorylation of upstream binding factor (UBF) at Thr117 and Thr201 following EGF treatment.72

Conclusions

With aberrations in the ERK cascade implicated in a high proportion of human cancers, many emerging therapies target proteins in the pathway.73 Upstream, more than a dozen drugs, including humanized monoclonal antibodies and small molecule inhibitors, are in clinical trials for EGF R family inhibition.74 One antibody, Herceptin, is approved for use in breast cancer patients overexpressing HER2. Several Ras inhibitors have reached Phase I to III clinical testing, including antisense oligonucleotides to limit expression and farnesyl transferase inhibitors to prevent membrane anchoring. Although no downstream inhibitors have reached trials, transcription factors have recently gained interest as the most direct and mechanistically relevant targets of the ERK pathway.75 Within the core module, a number of small kinase inhibitors against Raf and MEK have entered clinical development.76 This has occurred despite several concerns, including the dilution of ATP site-directed inhibitors by the high intracellular concentration of ATP, limited inhibitor specificity due to a common catalytic mechanism and other structural similarities among all kinases, and the importance of these kinases in so many processes unrelated to cellular proliferation.73 Ultimately, as the pathway is further revealed, novel components will provide the researcher with new therapeutic strategies.

As these candidate therapies against ERK signaling components undergo development and enter trials, reagents that monitor their targets' inhibition are critical for future success.73,76 Unfortunately, it is frequently unclear in oncology research if a poor response is due to an inconsequential target or if the target was inhibited insufficiently. Providing a "proof of principle" tool for both the laboratory and the clinic, biomarker assays that correlate the extent of target suppression with the efficacy of treatment are needed.

References

  1. Seger, R. & E. Krebs (1995) FASEB J. 9:726.
  2. Pages, G. et al. (1999) Science 286:1374.
  3. Blume-Jensen, P. & T. Hunter (2001) Nature 411:355.
  4. Schlessinger, J. (2000) Cell 103:211.
  5. Downward, J. et al. (1984) Nature 307:521.
  6. Schlessinger, J. (2002) Cell 110:669.
  7. Zwick, E. et al. (1999) Trends Prot. Sci. 20:408.
  8. Levkowitz, G. et al. (1999) Mol. Cell 4:1029.
  9. Li, N. et al. (1993) Nature 363:85.
  10. Rojas, M. et al. (1996) J. Biol. Chem. 271:27456.
  11. Pelicci, G. et al. (1992) Cell 70:93.
  12. Shimizu, K. et al. (1983) Proc. Natl. Acad. Sci. USA 80:2112.
  13. Avruch, J. et al. (1994) Trends Biochem. Sci. 19:279.
  14. Jelinek, T. et al. (1996) Mol. Cell. Biol. 16:1027.
  15. Rapp, U. et al. (1983) Proc. Natl. Acad. Sci. USA 80:4218.
  16. Chong, H. et al. (2003) Cell. Signal. 15:463.
  17. Lee, J. et al. (1996) Oncogene 12:1669.
  18. Barnier, J. et al. (1995) J. Biol. Chem. 270:23381.
  19. Pritchard, C. et al. (1996) Curr. Biol. 6:614.
  20. Wojnowski, L. et al. (1997) Nat. Genet. 16:293.
  21. Mikula, M. et al. (2001) EMBO J. 20:1952.
  22. Fridman, M. et al. (2000) J. Biol. Chem. 275:30363.
  23. Yan, J. et al. (1998) J. Biol. Chem. 273:24052.
  24. Marais, R. et al. (1997) J. Biol. Chem. 272:4378.
  25. Davies, H. et al. (2002) Nature 417:949.
  26. Zang, M. et al. (2002) J. Biol. Chem. 277:4395.
  27. Chiloeches, A. et al. (2001) Mol. Cell. Biol.
    28. 21:2423.
  28. Zimmermann, S. et al. (1999) Science 286:1741.
  29. Jaumot, M. & J. Hancock (2001) Oncogene 20:3949.
  30. Zheng, C. & K. Guan (1993) J. Biol. Chem. 268:11435.
  31. Mansour, S. et al. (1994) Science 265:966.
  32. Giroux, S. et al. (1999) Curr. Biol. 9:369.
  33. Brott, B. et al. (1993) Cell Growth Differ. 4:921.
  34. Alessi, D. et al. (1994) EMBO J. 13:1610.
  35. Huang, C. & J. Ferrell (1996) Proc. Natl. Acad. Sci.
    36. USA 93:10078.
  36. Kolch, W. (2000) Biochem. J. 351:289.
  37. Xiang, X. et al. (2002) J. Biol. Chem. 277:44996.
  38. Coles, L. & P. Shaw (2002) Oncogene 21:2236.
  39. Gopalbhai, K. et al. (2003) J. Biol. Chem. 278:8118.
  40. Boulton, T. et al. (1990) Science 249:64.
  41. Boulton, T. et al. (1991) Cell 65:663.
  42. Ferrell, J. & R. Bhatt (1997) J. Biol. Chem. 272:19008.
  43. Camps, M. et al. (2000) FASEB J. 14:6.
  44. Yin, Y. et al. (2003) Nature 422:527.
  45. Zhou, B. et al. (2002) J. Biol. Chem. 277:31818.
  46. Stewart, S. et al. (1999) Mol. Cell. Biol. 19:5523.
  47. Nguyen, A. et al. (2002) Mol. Cell. Biol. 22:3035.
  48. Schaeffer, H. et al. (1998) Science 281:1668.
  49. Teis, D. et al. (2002) Dev. Cell 3:803.
  50. Tanoue, T. et al. (2000) Nature Cell Biol. 2:110.
  51. Chen, D. et al. (1996) J. Biol. Chem. 271:6328.
  52. Northwood, I. et al. (1991) J. Biol. Chem. 266:15266.
  53. Brondello, J.-M. et al. (1999) Science 286:2514.
  54. Waskiewicz, A. et al. (1997) EMBO J. 16:1909.
  55. Knauf, U. et al. (2001) Mol. Cell. Biol. 21:5500.
  56. Chen, R.-H. et al. (1991) Mol. Cell. Biol. 11:1861.
  57. Dalby, K. et al. (1998) J. Biol. Chem. 273:1496.
  58. Gavin, A.-C. & A. Nebreda (1999) Curr. Biol. 9:281.
  59. Chen, R.-H. et al. (1993) Proc. Natl. Acad. Sci. USA 90:10952.
  60. Xing, J. et al. (1996) Science 273:959.
  61. Chung, J. et al. (1992) Cell 69:1227.
  62. Brunet, A. et al. (1999) EMBO J. 18:664.
  63. Deak, M. et al. (1998) EMBO J. 17:4426.
  64. Gupta, P. & R. Prywes (2002) J. Biol. Chem. 277:50550.
  65. Wiggin, G. et al. (2002) Mol. Cell. Biol. 22:2871.
  66. Soloaga, A. et al. (2003) EMBO J. 22:2788.
  67. Marais, R. et al. (1993) Cell 73:381.
  68. Gille, H. et al. (1995) EMBO J. 14:951.
  69. Price, M. et al. (1995) EMBO J. 14:2589.
  70. Gupta, S. et al. (1993) Proc. Natl. Acad. Sci. USA 90:3216.
  71. Felton-Edkins, Z. et al. (2003) EMBO J. 22:2422.
  72. Stefanovsky, V. et al. (2001) Mol. Cell 8:1063.
  73. Dancey, J. & E. Sausville (2003) Nat. Rev. Drug Dis. 2:296.
  74. Grunwald, V. & M. Hidalgo (2003) J. Natl. Cancer Inst. 95:851.
  75. Darnell, J. (2002) Nat. Rev. Cancer 2:740.
  76. Sebolt-Leopold, J. (2000) Oncogene 19:6594.