(Received for publication, February 4, 1997, and in revised form, May 14, 1997)
From the Division of Genetics, Department of Molecular and Cell Biology, University of California, Berkeley, California 94720
This work describes the phosphorylation of Saccharomyces cerevisiae Ras proteins and explores the physiological role of the phosphorylation of Ras2 protein. Proteins expressed from activated alleles of RAS were less stable and less phosphorylated than proteins from cells expressing wild-type alleles of RAS. This difference in phosphorylation level did not result from increased signaling through the Ras-cAMP pathway or reflect the primarily GTP-bound nature of activated forms of Ras protein per se. In addition, phosphorylation of Ras protein was not dependent on proper localization of the Ras2 protein to the plasma membrane nor on the interaction of Ras2p with its exchange factor, Cdc25p. The preferred phosphorylation site on Ras2 protein was identified as serine 214. This site, when mutated to alanine, led to promiscuous phosphorylation of Ras2 protein on nearby serine residues. A decrease in phosphorylation may lead to a decrease in signaling through the Ras-cAMP pathway.
The RAS genes were discovered as the transforming factors of acutely oncogenic retroviruses and were later recognized as virally transduced forms of cellular proto-oncogenes. Cellular RAS genes are activated by mutation in a wide variety of human cancers, providing the first example of a common molecular oncogenic mechanism for cancer (1-5).
Ras genes encode small guanine nucleotide-binding proteins and are highly conserved among eukaryotes. Mammalian cells have at least three distinct Ras proteins, Ha-Ras, N-Ras, and Ki-Ras (6), all of which can act as oncogenes when activated by certain mutations. The Ras superfamily of proteins function as molecular switches that regulate cell growth and differentiation. Ras protein is in its active form when bound to GTP and in its inactive form when bound to GDP. In this way, the ratio of GTP to GDP-bound forms of Ras protein determines the amount of Ras-mediated signaling.
The proper function of Ras protein requires cycling between the GTP and GDP-bound states in response to a nutrient or differentiation signal. All Ras proteins have intrinsic GTPase activity, and the regulation of the ratio of GTP to GDP-bound forms of the protein is modulated partially through regulation of this activity by proteins that stimulate the GTPase activity (7). In fact, mutant forms of Ras protein that fail to hydrolyze GTP produce unregulated cell growth or inappropriate differentiation.
Signal transduction pathways involving Ras proteins are responsible for the developmental fates of several different tissues in metazoans (8-11). Studies of Ras in these signal transduction cascades have led to a unified model for metazoan Ras signaling through a mitogen-activated kinase cascade. Saccharomyces cerevisiae contains two RAS genes, RAS1 and RAS2 (12), that both modulate growth. In contrast to all metazoans studied to date, the effector for RAS in S. cerevisiae is adenylyl cyclase, rather than a member of the mitogen-activated kinase pathway (13). Thus, although the function of RAS is conserved from S. cerevisiae to metazoans, the effectors for these Ras proteins have diverged.
In all species examined, Ras proteins are subject to several post-translational modifications at their carboxyl termini. These carboxyl-terminal modifications are conserved among Ras proteins from metazoans to yeast and are dependent on the presence of a carboxyl-terminal prenylation motif. The modifications include farnesylation of the most carboxyl-terminal cysteine (14), followed by endoproteolytic removal of the three carboxyl-terminal residues (15), and subsequent methylation of the carboxyl group on the newly generated carboxyl-terminal prenyl-cysteine (16). There may also be a reversible palmitoylation of the cysteine residue adjacent to the prenyl-cysteine (17). These modifications play a significant role in the proper localization of Ras protein to the plasma membrane.
The activity of some alleles of Ha-Ras is modulated by phosphorylation (18). Some naturally occurring activated mutant Ha-RasVal12 proteins have a second mutation that changes an alanine to a threonine at position 59 (A59T). Phosphorylation occurs on these Ha-RasVal12-Thr59 mutant proteins on threonine 59. This A59T mutation has been detected only in viral Ras oncoproteins that also contain the G12V mutation as well, and may arise as a suppressor of G12V. Indeed, Ha-RasVal12-Thr59 protein transforms fibroblasts less efficiently than does Ha-RasVal12 protein (18).
The two yeast Ras proteins, Ras1p and Ras2p, are also phosphorylated (19). However these phosphorylations have not been characterized with respect to their position in the protein or their biological function. Because phosphorylation frequently plays a regulatory role in biology, and because the regulation of Ras proteins in yeast is still poorly understood, we characterized the phosphorylation of yeast Ras protein to determine what role(s) it might play.
Yeast strains were grown in casamino acid medium lacking uracil and/or tryptophan (CAA), yeast minimal medium (YM), or rich medium (YPD) (Difco) containing 2% glucose (Sigma). 5-FOA was used at a concentration of 1 g/liter to select against URA3 function. Diploids were sporulated on minimal sporulation medium (20). Solid media contained 2% agar (Difco). Yeast cells were transformed by the lithium acetate method (21). Doubling times of yeast strains were determined by measuring the A600 of logarithmically growing cultures at 30 °C. For analysis of growth of phosphorylation mutants, all drugs or compounds were added to the YPD media after autoclaving.
All cloning and other molecular methods were performed according to standard methods (22) unless otherwise specified. DNA sequencing was performed on double-stranded templates by the di-deoxy chain termination method, either using the Sequenase 2.0 kit (U. S. Biochemical Corp.) on double-stranded denatured templates with 35S-dATP or by using cycle sequencing with fluorescent nucleotides on double-stranded templates and an ABI automated sequencer.
PCR1 ReactionsAll PCR products for cloning were made with Vent DNA polymerase (New England Biolabs) at 1 unit per 100-µl reaction. The reactions were in buffer containing 10 mM KCl, 20 mM Tris, pH 8.8, 10 mM (NH4)2SO4, 2 mM MgSO4, 0.1% Triton X-100, 250 µM each dNTP, 1 µM each primer, and 1-10 ng of template DNA. Denaturing temperature was 94 °C. Annealing temperature was 2-5 °C below the Tm of the primer with the lowest Tm. Extension was at 72 °C for 1 min for each kilobase of DNA amplified. Amplification was for 35 cycles.
To sew PCR products together, 2 µl of each template product were added to the standard buffer above with no oligo primers and an addition of 0.1 mM tetramethylammonium chloride. Five cycles were then run in this buffer in the absence of primers to generate full-length product template; the flanking primers were then added at 1 µM concentration, and 35 additional cycles were done to amplify the full-length product.
Oligo-directed MutagenesisOligonucleotide-directed mutagenesis was performed as described previously (23) using single-stranded DNA generated from RAS2 plasmid pJR1039 and RAS2Val19 plasmid pJR1040.
Heat Shock Sensitivity TestsAnalysis of heat shock sensitivity was performed by growing strains for 2 days at 30 °C into stationary phase (A600 4-5) in liquid casamino acid medium lacking tryptophan. Cells were diluted 10 × in distilled H2O and subjected to a 90-s heat shock at 55 °C in a thermocycler (MJ Research Inc.). Cells were then further diluted in distilled H2O and plated onto solid casamino acid medium lacking tryptophan. Percentage survival from heat shock was calculated by comparing the number of colonies arising from heat-shocked cells to the number obtained from cells from the same initial dilution that had not been heat shocked.
Strain ConstructionCells deleted for CDC25 and
carrying a CEN TRP1-marked CDC25 plasmid
(pCDC25-TRP1) were transformed with a high copy
URA3-marked plasmid carrying TPK1 (pJR1857). The
resulting strain RAS1 RAS2 cdc25 pCDC25-TRP1
pTPK1-URA3 (JRY5416) was grown in medium containing tryptophan to allow loss of the CDC25-TRP1 plasmid. These
cells were then replica-plated to medium lacking tryptophan to identify colonies that had lost the CDC25-TRP1 plasmid, one of which
became strain RAS1 RAS2 cdc25
pTPK1-URA3
(JRY5417), which is genotypically cdc25
but otherwise
isogenic to JRY5416.
The ras1 ras2
pTPK1 strain (JRY5404) was
constructed in several steps, JRY3001 MAT
ade2 his3 leu2 trp1
ura3 can1 (W303 background) was transformed with a PCR product
designed to disrupt the RAS1 open reading frame. The PCR
product was synthesized using Vent DNA polymerase using oligo primers
JW040 and JW041. Primer JW040
5
-CAGGTAAACAAAATTTTCCCTTTTTAGAACCATATGGTACTGAGAGTGCACC-3
contained 30 bp complementary to the non-coding strand of
RAS1 just upstream of the ATG start codon and generated a
NdeI site (underlined) utilizing the ATG start codon. The
remaining 16 bp (in bold type) were complementary to the region of the
pRS vectors (24) flanking the insertion of the selectable marker genes
for these vectors. Primer JW041
5
-CAAAACCATGTCATATCAAGAGAGCAGGATCGCGGTATTTCACACCGC-3
contained 31 bp complementary to the coding strand of
RAS1 just downstream of the stop codon and 16 bp (in bold
type) complementary to the vector flanking the insertion of the
selectable marker genes for these vectors. This primer pair was used to
amplify the HIS3 gene from vector pRS303. The resulting PCR
product contained the entire HIS3 gene flanked by 30 and 31 bp of RAS1 sequence 5
and 3
of the HIS3 gene,
respectively, to target recombination of HIS3 into the
RAS1 locus. The PCR product was used to transform JRY3001,
and His+ transformants were selected. In approximately half
the His+ transformants, the RAS1 gene was
replaced by HIS3, as confirmed by DNA blot hybridization
(data not shown). The resulting strain, MAT
ade2 his3 leu2
trp1 ura3 can1 ras1
::HIS3, was JRY5412. This strain
was the parent of the bcy1
strain described below
(JRY5410). The ras1
::HIS3 strain was
transformed with the TPK1-URA3 high copy plasmid pJR1857
(JRY5427) and then deleted for RAS2 using a
ras2
::LEU2 disruption plasmid (pJR1858). In
this ras2::LEU2 the LEU2 gene replaced
the RAS2 coding region from the ClaI site 560 bp
upstream of the ATG start codon to the MscI site 636 bp into
the open reading frame. Disruption was confirmed by the inability of
resulting LEU2 transformants to grow on solid medium
containing 5-FOA, which selects against the pTPK1-URA3
plasmid, which is required for viability in the ras1
ras2
strain. Approximately half the resulting LEU2
transformants were unable to grow on 5-FOA. One of these was designated
JRY5404 and had the genotype MAT
ade2 his3 leu2 trp1 ura3 can1
ras1
::HIS3 ras2
::LEU2
pTPK1-URA3.
The bcy1 strain (JRY5410) was constructed using the
ras1
::HIS3 strain described above (JRY5412)
using a linear fragment from plasmid pJR1071 that contained the
BCY1 gene disrupted by the LEU2 gene. The
disruption was confirmed by testing heat shock sensitivity of the
LEU2 transformants (data not shown).
The ras1::HIS3 strains (JRY5407 and JRY5408)
were constructed in two steps to create an unmarked deletion (25). The
RAS1 disruption plasmid (pJR1877) lacked the 270-bp
HpaI-MscI fragment internal to RAS1. A
partial digest of this plasmid with HindIII, which cut
within the remaining RAS1 sequence, created a linear fragment that directed insertion to the RAS1 locus
(JRY5428). The resulting URA3 transformants were streaked on
5-FOA medium to select for a recombination event that removed the
vector with the URA3 gene, leaving behind an unmarked
partial deletion allele of RAS1 (JRY5407). The disruption
allele was confirmed by DNA blot hybridization. The strain with the
ras1 disruption allele was transformed with a URA3
RAS2Val19 plasmid (pJR1861) (JRY5408).
The phenotypes of various RAS2 alleles were evaluated after integrating these alleles in a strain lacking the wild-type RAS1 and RAS2 alleles (JRY5404). To construct TRP1-integrating versions of the desired RAS2 allele the XhoI-BamHI fragment containing the RAS2 alleles from various plasmids was subcloned into TRP1-marked integrating vector YIplac204, which had been digested with both SalI and BamHI. The resulting plasmids were then digested with NdeI and re-ligated to remove the small polylinker-containing region, making EcoRV a unique site within the TRP1 gene. Linearization of these plasmids with EcoRV directs integration into the TRP1 locus.
Plasmid ConstructionsThe high copy TRP1-marked RAS2 plasmid (pJR1244) was constructed by cloning the 1.9-kb EcoRI-SalI fragment containing the entire RAS2 gene from pJR827 into pRS424. The TRP1-marked CEN RAS2 plasmid (pJR1051) was constructed by cloning the 1.9-kb EcoRI-SalI fragment into pRS314. The TRP1-marked CEN RAS2Val19 plasmid (pJR1052) was constructed by cloning the 1.9-kb EcoRI-SalI fragment containing the entire RAS2Val19 gene from pJR828 into pRS314. Plasmid pJR827 was constructed by cloning the 1.9-kb ClaI-HindIII RAS2 gene into pRS306. Plasmid pJR828 was constructed by cloning the 1.9-kb ClaI-HindIII fragment containing the entire RAS2Val19 gene into pRS306. Plasmid pJR1039 was constructed by cloning the 1.9-kb EcoRI-SalI RAS2 fragment from pJR827 into pRS316. Plasmid pJR1040 was constructed by cloning the 1.9-kb EcoRI-SalI RAS2Val19 fragment from pJR828 into pRS316. The CEN TRP1-marked RAS2 plasmid (pJR1878) was generated by cloning a XhoI-BamHI fragment containing the entire RAS2 gene from pJR1039 into pRS414.
The high copy URA3-marked TPK1 plasmid (pJR1857)
was constructed by cloning a PCR fragment containing the entire
TPK1 gene into pRS426. The TPK1 PCR product was
amplified using Vent DNA polymerase using oligo primers JW094 and
JW095. Primer JW094, 5-GGAAGCTTGCAAAGCCAGTGTCCCAACG-3
was
complementary to the non-coding strand at the 5
end of the
TPK1 gene and generated a HindIII site. Primer
JW095, 5
-GGAAGCTTGAACGCCGTTCCATCTCCG-3
was complementary to the coding strand at the 3
end of TPK1 and also
generated a HindIII site. The resulting PCR product was
purified with a Millipore ultrafree-MC microcentrifuge unit, digested
with HindIII, gel-purified, and cloned into pRS426,
resulting in pJW167, in which the open reading frame of TPK1
was in the same orientation as URA3.
The high copy RAS1 plasmid pJR1859 was constructed by
cloning a HindIII fragment containing the entire
RAS1 gene into the high copy TRP1 vector
YEplac112. In the resulting plasmid (pJR1859) the RAS1 was
in the same orientation as TRP1. The high copy
RAS1Val19 plasmid (pJR1860) was constructed by
cloning a PCR fragment containing the entire
RAS1Val19 gene into the high copy TRP1
vector YEplac112. The RAS1Val19 PCR fragment was
generated by the PCR sewing method. First, two PCR products were made
both of which incorporated the RAS1Val19 mutation
into the product by use of a mutagenic oligo primer. One product
amplified the RAS1 region 5 of amino acid 19, utilizing oligonucleotides JW134 and T7 as primers. T7 annealed to the polylinker region of RAS1 plasmid pJW99 that was used as the template
for the PCR reaction. Primer JW134
5
-GCAGATTTACCAACGCCAACTCCACCGACAACTAC-3
was
complementary to the coding strand of RAS1 and introduced the Gly-19
Val mutation (bold type, underlined). The other product amplified the RAS1 region 3
of amino acid 19, utilizing
oligonucleotide JW133 and reverse as primers. Reverse primer annealed
to the polylinker region of RAS1 plasmid pJW99 which was
used as the template for the PCR reaction. Primer JW133
5
-GTAGTTGTCGGTGGAGTTGGCGGTGGTAAATCTGC-3
was complementary to the non-coding strand of RAS1 and
introduced the Gly-19
Val mutation (bold type, underlined). Both
products were purified, and 2 µl of each of the previous products
were used as template for a PCR sewing reaction (see below) utilizing primers T7 and reverse. The resulting full-length
RAS1Val19 PCR product was purified, digested with
HindIII, and cloned into the TRP1-marked high
copy vector YCplac112 to create pJR1860, in which
RAS1Val19 was in the same orientation as the
TRP1 gene. The presence of the RAS1Val19
mutation was confirmed by sequencing. The RAS1 plasmid used
as template for the PCR reactions described above (pJR1879) was
constructed by cloning the HindIII fragment containing
RAS1 into the polylinker of pRS425.
A plasmid encoding a fusion protein in which the green fluorescence
protein (GFP-S65Y) was joined to the amino terminus of the
RAS2 coding region was used to evaluate the localization of Ras2p in living cells. This plasmid was derived from a high copy TRP1-marked vector (pJR1136, a gift from Cindy Trueblood)
that contained a HindIII-XbaI fragment with the
strong constitutive GPD promoter and the PGK1 terminator. The
RAS2 coding region was inserted between the promoter and the
terminator by inserting an EcoRI-SalI fragment
(from pJR1880) into the EcoRI and SalI sites of
pJR1136, forming pJR1881. As a result, RAS2 was under the
control of the GPD promoter. In the next step, a very bright variant of
the GFP gene (S65Y) (26) was amplified from a plasmid pRH431 (a gift
from Randy Hampton) with the oligonucleotide primers JW121 and JW122.
5-CCGAGCTCATGAGTAAAGGAGAAGAAC-3
(JW121) was
complementary to the non-coding strand of vbGFP and
introduced a SacI site (underlined) just 5
of the ATG start
codon of vbGFP (in bold type), and
5
-CCGAATTCTTTGTATAGTTCATCCATG-3
(JW122) was complementary
to the coding strand of vbGFP and introduced an
EcoRI site (underlined) just 5
of the stop codon for
vbGFP. This EcoRI site allowed a
SacI-EcoRI fragment of GFP to be cloned into the
GPD-RAS2 plasmid (pJR1881) creating an amino-terminal fusion
of GFP to the RAS2 gene (pJR1862). The
GFP-ras2-SSIIS plasmid (pJR1863) was constructed by
substituting the wild-type RAS2 gene on an
EcoRI-SalI fragment from pJR1862 with an
EcoRI-SalI PCR fragment containing the
ras2-SSIIS allele. The ras2-SSIIS PCR fragment
was generated as above from plasmid pEG(KG)RAS2-SSIIS utilizing oligonucleotide primers JW067 and JW127.
5
-CCGAATTCATGCCTTTGAACAAG-3
(JW067) was
complementary to the non-coding strand at the 5
end of
RAS2, and 5
-CCGTCGACGCGTTTCTACAACTATTTCC-3
(JW127) was complementary to the coding strand of the RAS2
gene prior to the cysteine to serine mutations.
Plasmids marked with the TRP1 gene and containing the serine to alanine mutations at RAS2 positions 198, 202, 207, and 214 (pJR1864, pJR1865, pJR1866, and pJR1867) were created by subcloning the 1.0-kb PstI fragments from pJR1868, pJR1869, pJR1870, and pJR1871, respectively, into the vector fragment of PstI-digested RAS2 plasmid pJR1244 and selecting the correct orientation. Plasmid pJR1868, pJR1869, pJR1870, and pJR1871 were generated by oligo-directed mutagenesis using single-stranded DNA from RAS2Val19 plasmid pJR1040 as template. Oligonucleotide primers JW115, JW116, JW117, and JW118 which were complementary to the coding strand were used to generate pJR1868, pJR1869, pJR1870, and pJR1871, respectively.
Creation of Other Serine to Alanine Mutations by PCR SewingPCR products were generated under standard conditions, and
sewing products were made as described above using flanking primers JW137 and JW124. These sewing products were transformed into a ras1 ras2
yeast strain (JRY5404) that carries a
URA3-marked RAS2 plasmid, simultaneously with
linearized TRP1-marked RAS2-gap plasmid, pJR1882,
that was deleted for the HpaI fragment of RAS2, and TRP1 transformants were selected. Strains were then
grown on medium containing 5-FOA to select for successful gap repair. Plasmids from these strains were recovered (27) and transformed into
Escherichia coli strain DH5
. DNA sequencing was used as a
final check on the sequence of each mutation created.
The RAS2-gap plasmid, pJR1882, was constructed by digesting RAS2 plasmid pJR1244 with HpaI and re-ligating to remove the 1.5-kb RAS2-containing insert.
Immunoblot AnalysisSamples were harvested at various
culture densities, as described, and frozen at 80 °C until
processing. Immunoblot analysis was performed essentially as described
previously (28), only the rat monoclonal antibody, Y13-259 (Oncogene
Sciences), was used at a 1:250 dilution at 4 °C.
Cells were grown to an
A600 of approximately 1.0 in medium with 5 µM KH2PO4 and 10 µM
(NH4)2SO4 to limit cells for
phosphate and sulfate (LPSM). 2 A600 units of
cells were pelleted and resuspended in 2 ml of fresh LPSM. 1 mCi of
inorganic 32P (Amersham Corp.) or 1 mCi of
35S-trans-label (ICN) was added to cells that were
metabolically labeled at 30 °C for 3 h, unless otherwise
specified. Labeled cells were then pelleted and washed with 1 ml of YM
media with 0.1% NaN3, and the pellets were frozen at
80 °C.
Labeled cells were thawed and resuspended on ice in 50 mM Tris, pH 7.5, 20 mM MgCl, 100 mM NaCl, 0.1% Nonidet P-40, 1 mM dithiothreitol, 1% aprotinin, 1 mM phenylmethylsulfonyl fluoride (NLB) at 10 OD/ml. Cells were disrupted with glass beads (29). Cell lysates were removed to a clean chilled tube, and glass beads were washed with one additional volume of NLB. Lysate and wash were pooled and centrifuged at full speed in a microcentrifuge at 4 °C for 10 min to pellet cell debris. The supernatants were added to 1.5 × volume of bovine serum albumin-pretreated activated charcoal, vortexed, and pelleted repeatedly until no charcoal residue remained.
Protein G-Sepharose beads conjugated to the anti-Ras antibody Y13-259 (Oncogene Sciences) were prepared using standard procedures (30).
ImmunoprecipitationsEquivalent protein for each lysate was
added to a 10-µl bed volume (100-µl slurry) of protein
G-Sepharose/Y13-259 beads and incubated end over end at 4 °C for
2 h. Lysate was removed, and beads were washed 3 × 100 cv
with ice-cold NLB, and then with 3 × 100 cv with ice-cold NLB
lacking Nonidet P-40. Immunoprecipitated proteins were eluted from
beads in 20 mM Tris, pH 8.0, 20 mM EDTA, 2%
SDS, 2 mM GTP, 2 mM GDP at 70 °C for 5 min.
Eluate was removed from beads, spun at 4 °C for 5 min to remove any
insoluble material, and stored at 20 °C.
Proteins from
immunoprecipitations were electrophoretically separated on 12%
SDS-polyacrylamide gels, unless otherwise stated. Gels were fixed in
25% isopropyl alcohol, 10% HOAc for 30 min, dried at 60 °C for
1 h, and placed on film at 80 °C and/or exposed to a
PhosphorImager cassette (Molecular Dynamics).
A yeast strain
deleted for RAS1 and RAS2 and carrying the
p(EG)KG-RAS2 plasmid (31), which expresses a GST-Ras2 fusion
protein from the galactose-inducible Gal1 promoter (JRY5429), was grown in 5 liters of YM with 2% raffinose as a carbon source to an
A600 of 0.2 at 30 °C. Galactose was added to
4% and the cells continued to grow overnight at 30 °C for 4 more
doublings. Cells were harvested and washed with STMN (0.3 M
sorbitol, 0.1 M NaCl, 5 mM MgCl, 10 mM Tris, pH 7.4) and frozen at 80 °C. Protein was
purified as published (31) only no detergent was added. Protein was
eluted by digesting the bead-bound protein with thrombin 2 × for
30 min at room temperature. Purification of Ras2p with 12 amino acids from GST was to near-homogeneity from the single affinity
chromatographic step. A total of 5 mg of Ras2 protein was obtained from
the 5 liters of induced cells. The identity of this protein as Ras was confirmed by immunoblot (data not shown). The resulting Ras2 protein was then further purified by reverse phase high performance liquid chromatography on a C18 column. The protein invariably eluted as two
peaks that could not be resolved under the conditions used. A portion
of the purified protein was treated with alkaline phosphatase, and the
treated protein was eluted as a single peak on high performance liquid
chromatography analysis, indicating that the second peak represented
phosphorylated Ras2p.
Ras2 peptides were generated by digestion with LysC, AspN, or ArgC (Boehringer Mannheim) according to manufacturer's instructions. Peptides were then individually purified by high performance liquid chromatography over a C18 reverse phase column and subjected to analysis by mass spectrometry.
In most cases in which phosphorylation plays a
regulatory role on a protein, the half-life of phosphorylation is
shorter than the half-life of the protein. Therefore we compared the
half-life of the entire Ras2 protein pool and the half-life of
phosphorylated Ras2 protein. A strain deleted for ras1 and
ras2 and carrying a high copy RAS2 plasmid
(JRY5430) was pulse-labeled with either 32Pi or
35S-trans-label and chased into cold medium. The Ras2p was
then immunoprecipitated and analyzed by SDS-PAGE. The half-life of the
Ras2 protein pool under these conditions was greater than 5 h,
whereas the half-life of the phosphorylated pool was less than 1 h
(Fig. 1). Based on these data, phosphorylation of Ras2p was dynamic and was therefore potentially a regulatory
modification.
Activated RAS Alleles Produced Hypophosphorylated Ras Protein
If phosphorylation of Ras2p regulated its activity, then
mutations that affect the activity of Ras2p might affect its level of
phosphorylation. We therefore tested whether constitutively activated
alleles of Ras had an altered level of phosphorylation. Cells with null
alleles of chromosomal RAS1 and RAS2 (JRY5404) were transformed with a high copy plasmid carrying either
RAS2 (pJR1244) or RAS2Val19(pJR1912),
which encodes a Ras2 protein with impaired GTPase activity due to the
G19V mutation. This mutation is analogous to the G12V mutation of
oncogenic forms of human Ras protein. The resulting yeast strains,
ras1 ras2
pRAS2 (JRY5430) and ras1
ras2
pRAS2Val19 (JRY5470), were radiolabeled
with 32P or grown in the same medium without radiolabel.
The Ras proteins from the labeled cells were immunoprecipitated and
analyzed for phosphorylation, and the level of Ras protein in the
unlabeled cells was analyzed by immunoblot. Cells carrying only
RAS2Val19 had 10-fold less steady-state level of Ras
protein than cells carrying only RAS2 (data not shown).
Thus, to evaluate the effect of the RAS2Val19 allele
on phosphorylation of Ras protein per se, the difference in
steady-state protein levels was compensated for when loading immunoprecipitates onto the gel. The Ras2 protein was significantly more phosphorylated than Ras2Val19 protein (Fig.
2A). At face value, these data indicated that
activated Ras2p was hypophosphorylated relative to wild-type Ras2p.
The effect of activating mutations on Ras2 protein steady-state level
and phosphorylation state was unprecedented and unexpected. Therefore,
the impact of an activating mutation on Ras1 protein phosphorylation
was evaluated to determine whether the effect of an activating mutation
in RAS2 reflected a general trend for activated alleles of
RAS or was an idiosyncrasy of the Ras2 protein. Cells
lacking chromosomal RAS1 and RAS2 (JRY5404) were
transformed with a high copy plasmid carrying either RAS1
(pJR1859) or RAS1Val19 (pJR1860). The resulting
yeast strains, ras1ras2
pRAS1 (JRY5405) and
ras1
ras2
pRAS1Val19 (JRY5406), were
radiolabeled with 32P or grown in the same medium without
radiolabel. As before, the Ras proteins from the labeled cells were
immunoprecipitated and analyzed for the extent of phosphorylation,
whereas the level of Ras proteins in the unlabeled cells was analyzed
by immunoblot. As with Ras2Val19p, cells carrying only
RAS1Val19 expressed approximately 10-fold less
steady-state level of Ras protein than cells carrying only
RAS1 (data not shown). Also, as before, the difference in
steady-state protein levels was therefore compensated for when loading
the gel to compare the extent of phosphorylation of activated
versus non-activated alleles of RAS1. Ras1p was
more phosphorylated than Ras1Val19p (Fig. 2B).
Thus, RAS1 and RAS2 shared the following two
features: activated alleles of either protein resulted in a substantial decrease in steady-state Ras protein level; and even when taking changes in protein level into account, the activated alleles resulted in a substantial decrease in the phosphorylation of both Ras1p and
Ras2p.
To determine whether the observed decreased phosphorylation of Ras2Val19 protein was a result of more rapid turnover of the Ras2Val19 protein, we analyzed the turnover rate of the phosphorylated forms of the Ras2Val19 protein compared with the turnover rate of the phosphorylated forms of wild-type Ras2p. A strain deleted for ras1 and ras2 and carrying a high copy RAS2 plasmid (JRY5430) or a high copy RAS2Val19 plasmid (JRY5470) was pulse-labeled with 32Pi and chased into cold medium. The Ras2p was then immunoprecipitated and analyzed by SDS-PAGE. Although the Ras2Val19 protein attained a phosphorylated state more slowly that wild-type Ras2p, the phosphorylated pool of the Ras2Val19 protein was significantly more stable that the phosphorylated Ras2p pool (Fig. 2C), suggesting that rapid turnover of the Ras2Val19 protein was not responsible for its lower steady-state level of phosphorylation. Rather, Ras2Val19 protein was apparently a poorer substrate for the putative Ras2p kinase and/or the Ras2p phosphatase.
Ras Protein Phosphorylation Was Not a Means of Feedback RegulationThe reduction in phosphorylation of
Ras2Val19 protein suggested that the phosphorylation of Ras
proteins might be tied in some way to the activity of the
cAMP-dependent kinase pathway, which is over-stimulated by
Ras2Val19p. Because the physiological effects of
RAS2Val19 are dominant, it was possible to test
whether feedback regulation from the cAMP-dependent protein
kinases was responsible for the differential phosphorylation of Ras2
and Ras2Val19 protein. To test whether expression of
Ras2Val19 protein could prevent the phosphorylation of
wild-type Ras2p in the same cell, two strains with the genotypes
ras1 RAS2 pRAS2Val19 (JRY5408) and
ras1
RAS2 (JRY5407) were radiolabeled with
32P and evaluated for Ras protein phosphorylation by
immunoprecipitation. The presence of RAS2Val19 did
not prevent the phosphorylation of wild-type Ras2 protein in cells with
both RAS2 and RAS2Val19 alleles (Fig.
3A, compare lanes 2 and
3). Therefore, reduced phosphorylation of Ras2p was not due
to enhanced activity of the cAMP signaling pathway.
As an independent evaluation of the role of elevated cAMP levels on the
phosphorylation state of Ras proteins, other modifications that cause
activated phenotypes in yeast cells were tested for whether they
reduced the phosphorylation of Ras2p. Overexpression of
TPK1, which encodes one of several
cAMP-dependent protein kinases, yields activated phenotypes
similar to those associated with RAS2Val19 (32).
Strains disrupted for BCY1 the regulatory subunit of the
cAMP-dependent protein kinases TPK1,
TPK2, and TPK3 also have activated phenotypes
(32). Strains containing these activating mutations were examined for
extent of phosphorylation of the wild-type Ras2 protein. Although the
strains ras1 ras2
pTPK1
pRAS2-2µ (JRY5409) and ras1
bcy1
pRAS2-2µ (JRY5410) were significantly more heat
shock-sensitive than their isogenic parent strain (JRY5413) (data not
shown), the levels of Ras2 protein phosphorylation in these three
strains were virtually indistinguishable (Fig. 3B). In
addition, the steady-state level of Ras2p in these mutant strains was
also unaffected (data not shown), demonstrating that neither decreased
steady-state level of Ras2Val19p nor decreased
Ras2Val19 protein phosphorylation was due to
hyper-stimulation of the cAMP pathway per se.
Ras protein is localized at the plasma membrane and must be membrane-localized to promote growth. Mutations in RAS that prevent proper localization prevent stimulation of proliferation (33) although sufficiently high overexpression of localization-defective Ras protein can provide enough Ras function for viability. Therefore we tested whether the localization of Ras protein to the plasma membrane was necessary for its efficient phosphorylation.
Previous localization studies of Ras protein in yeast were of limited resolution due, at least in part, to the weak localization signal with existing antibodies in immunolocalization experiments. To better assay Ras protein localization, a fusion gene was constructed that joined the structural gene for green fluorescence protein (GFP) to the amino terminus of the coding region of RAS2. The GFP-RAS2 fusion gene was expressed from the strong constitutive GPD promoter and terminated with the PGK1 terminator (pJR1862). A fusion gene was also constructed in which the carboxyl-terminal cysteine codons of RAS2 (CCIIS-OH) were mutated to serine codons (SSIIS-OH) (pJR1863). The protein encoded by this mutated form of RAS2, ras2-SSIIS, cannot be processed at its carboxyl terminus and should not be membrane-localized.
Both GFP-Ras2 fusion proteins, GFP-Ras2 and GFP-Ras2-SSIIS,
complemented the growth defect of the RAS null strain (data
not shown). The strains expressing each of the GFP-Ras2 fusion proteins as their only functional Ras protein, ras1 ras2
pGFP-RAS2 (JRY5414) and ras1
ras2
pGFP-ras2-SSIIS (JRY5415), were visualized by fluorescence
microscopy to determine the localization of the GFP-Ras2 fusion
proteins. As expected, the GFP-Ras2 protein was primarily localized to
the plasma membrane, whereas the GFP-ras2-SSIIS protein was localized
throughout the cytoplasm (data not shown). There was no difference in
Ras protein expression levels in these two strains as judged by
immunoblot analysis (Fig. 4).
To test whether phosphorylation of Ras2 protein was affected by its localization, the two strains were radiolabeled with 32P, and the GFP-Ras2 proteins in these cells were immunoprecipitated and analyzed for the extent of phosphorylation. There was no detectable difference in the levels of phosphorylation between the Ras protein that was localized and that which was not (Fig. 4). Thus membrane localization was not required for efficient phosphorylation of Ras protein in yeast.
Deletion of CDC25 Did Not Affect the Phosphorylation State of Ras ProteinRas2Val19 protein, which is primarily GTP-bound, has a weaker interaction with Cdc25 guanine nucleotide exchange protein than does wild-type Ras2 protein, which is primarily GDP-bound (34). Because the GDP-bound form of Ras2 protein associates more tightly with Cdc25p, and because the GDP-bound form of Ras2p is more phosphorylated than predominantly GTP-bound Ras2Val19 protein, the Cdc25 dependence of Ras2 protein phosphorylation was evaluated.
The effect of a CDC25 deletion on the phosphorylation state
of Ras protein was tested as above using the isogenic pair of strains
RAS1 RAS2 cdc25 pCDC25-TRP1 pTPK1
(JRY5416) and RAS1 RAS2 cdc25
pTPK1 (JRY5417).
There was no detectable difference in the level of phosphorylation of
Ras proteins in cells with or without functional Cdc25 protein (Fig.
5). Therefore, association with the guanine nucleotide
exchange factor was not necessary for the efficient phosphorylation or
de-phosphorylation of Ras protein.
Identification of a Phosphorylation Site on Ras2 Protein
The experiments described above established that Ras2p phosphorylation was dynamic and ruled out several possible explanations for why and how Ras2p phosphorylation was modulated. We then turned to a different approach for elucidating the role of Ras phosphorylation by evaluating phenotypes associated with mutant forms of Ras2p that lacked the ability to be efficiently phosphorylated. To construct such an allele, it was necessary to identify the phosphorylated residue(s) in Ras2p. As mentioned previously, Ras2 protein is phosphorylated only on one or more serine residues (19). However, there are 31 serine residues in mature Ras2p, and previous studies did not determine the number or stoichiometry of phosphorylation (19). The site(s) of Ras2p phosphorylation were determined through a combination of biochemical and genetic experiments and then the serines at these potential phosphorylation site(s) were changed to alanines. These studies were followed by in vivo analysis of phosphorylation-defective Ras2 proteins.
The predicted mass for Ras2 protein carrying 12 amino acids from the GST fusion, and accounting for post-translational farnesylation, proteolysis, and methyl esterification is 35,797 daltons (Da). The actual mass of the purified Ras2 protein was determined using electrospray mass spectrometry. Two protein species were detected in this sample. One protein, which represented 60-70% of the total protein, had a mass of 35,891.5 ± 2 Da. The other protein, which represented 30-40% of the total protein, had a mass of 35,971.4 ± 3 Da. These two masses differed by 80 Da, which is the mass of a phosphate. Thus, 60-70% of the Ras2 protein was phosphorylated. Surprisingly, the mass of the lower molecular weight species in this sample was 95 Da larger than the predicted mass of Ras2p. These 95 Da could have represented an additional phosphate (80 Da) plus an additional methylation (15 Da), or some other modification or combination of modifications that totaled 95 Da. However, as described below, it is unlikely that these 95 Da result in part from a second phosphorylation.
Identification of a Phosphorylated Ras2 PeptideThe
endoproteolytic enzymes LysC, AspN, and ArgC were used to digest
purified Ras2 protein, and the resulting peptides from these digestions
were purified and subjected to analysis by mass spectrometry to
identify phosphorylated peptides. Any phosphorylated peptide was
expected to have a mass 80 Da larger than the mass predicted from the
amino acid sequence of that peptide. Utilization of three different
endoproteases allowed identification of numerous overlapping
non-phosphorylated peptides which accounted for nearly the entire
amino-terminal half of the Ras2 protein (Fig. 6) and suggested that the conserved amino-terminal half of the Ras2 protein was not phosphorylated in vivo.
Few peptides were isolated from the carboxyl-terminal domain of the Ras2 protein. However, one peptide fraction was a mixture of two peptides whose masses differed by 80 Da (Fig. 6, peptide 19). Approximately 60% of the protein in this peptide fraction had a mass of 6117 Da; the remaining 40% had a mass of 6197 Da. This ratio was similar to the ratio of the larger and smaller masses observed for the full-length Ras2 protein.
The mass of this phosphopeptide indicated that it represented amino acids 194-259 of Ras2p. This prediction was confirmed by amino-terminal sequence analysis of the peptide. There are 8 serines within this phosphopeptide. However, the measured mass of a peptide corresponding to residues 220-239 (Fig. 6, peptide 20), excluded the 4 serine residues in this peptide as sites of phosphorylation. The remaining 4 serine residues in the phosphopeptide (194-259), Ser-198, Ser-202, Ser-207, and Ser-214, were potential phosphorylation sites.
Mutagenesis of the Serines in Phosphopeptide 194-259To
determine whether the phosphorylation on peptide 194-259 was, in fact,
the only phosphorylation on Ras2p, each of the 4 serine codons in this
portion of the RAS2 gene was individually mutagenized to
alanine codons. The resulting alleles, ras2Ala198
(pJR1864), ras2Ala202 (pJR1865),
ras2Ala207 (pJR1866), and
ras2Ala214 (pJR1867) were transformed on high copy
TRP1-based plasmids into a RAS null strain
(JRY5404), and all four mutants provided Ras function (data not shown).
The resulting strains ras1 ras2
pras2Ala198 (JRY5418), ras1
ras2
pras2Ala202 (JRY5419), ras1
ras2
pras2Ala207 (JRY5420), and ras1
ras2
pras2Ala214 (JRY5421) were evaluated
for Ras2 protein phosphorylation as described above and compared with
the strain expressing wild-type RAS2 in high copy (JRY5430).
Only the ras2Ala214 allele produced Ras2 protein
that was less phosphorylated than wild-type protein (Fig.
7, compare lanes 6 and 7)
suggesting that serine 214 was a phosphorylation site in Ras2 protein.
The remaining three mutant proteins were phosphorylated to a level
indistinguishable from that on wild-type protein (Fig. 7, compare
lanes 3, 4, and 5 to lane 2).
Steady-state level of Ras2 protein in all these strains was equivalent
(Fig. 7, immunoblot).
The extent of phosphorylation on Ras2Ala214p was 56% that of the wild-type protein. Therefore, substitution of serine 214 with a non-phosphorylatable alanine residue decreased the level of phosphorylation on full-length Ras2 protein by 44%. These data suggested that serine 214 was indeed a phosphorylation site for Ras2 protein. However, the residual phosphorylation of this mutant protein indicated that a second site was phosphorylated.
Search for the Second Phosphorylation Site in Ras2 ProteinOnly 5 serine residues remained unaccounted for by
peptide mass spectrometry data and were therefore still candidates for a second phosphorylation site: serine 6, serine 262, serine 273, serine
285, and serine 291 (Fig. 6). To determine whether one of these
remaining serines was a second phosphorylation site, a series of
site-directed mutants were constructed that eliminated various
combinations of these and other serine residues (Fig. 8A). All of the mutant alleles were
transformed into a RAS null strain (JRY5404) on high copy
TRP1-marked plasmids, and all were able to complement the
growth defect of a null strain (data not shown).
Evaluation of protein phosphorylation of these mutant Ras2 proteins revealed several points of interest (Fig. 8A). Mutation of serine residue 6, 262, 273, 285, or 291 to alanine had no effect on the level of phosphorylation of these Ras2 mutant proteins (Fig. 8A, lanes 7, 9, 12, and 14). Therefore, none of these serine residues was likely to serve as a phosphorylation site in vivo. In all cases, a decrease in the level of Ras2 phosphorylation occurred only in combinations containing the S214A mutation (Fig. 8A, lanes 8 and 13). In addition, all Ras2 proteins with the S214A mutation were less phosphorylated (Fig. 8A, lanes 8 and 13), demonstrating that serine 214 was a preferred site of phosphorylation in vivo.
Interestingly, when additional serine or threonine residues near serine 214 were mutated to alanine in cis to the S214A mutation, the level of Ras2 protein phosphorylation could be reduced below the level observed when the S214A mutation was present alone (Fig. 8B). The greater the number of serine (or threonine) to alanine changes in cis to S214A, the lower the level of phosphorylation on Ras2p (Fig. 8B, compare lanes 3, 4 and 5). Thus, although serine 214 was the preferred phosphorylation site for Ras2p, mutation of serine 214 to alanine resulted in the phosphorylation of other serine or threonine residues near serine 214.
Characterization of Cells Expressing the ras2Ala214 AlleleThe identification of a preferred phosphorylation site for
Ras2p and subsequent mutagenesis of this serine residue to alanine allowed testing of what phenotype(s) was associated with
reduction of Ras2p phosphorylation levels. TRP1-marked
versions of five different alleles of RAS2
(ras2Ala214, RAS2Val19-Ala214,
ras2Ala198-Ala202-Ala207-Ala214-Ala235-Ala238,
RAS2, RAS2Val19) were integrated at the
TRP1 locus in a ras1 ras2
null strain (JRY5404) to produce an otherwise isogenic set of strains (JRY5422, JRY5423, JRY5424, JRY5425, and JRY5426, respectively). Analysis of
growth rate of these five integrated strains demonstrated that presence
of a phosphorylation mutation in RAS2 or in cis
to the activating mutation RAS2Val19 had no effect
on the doubling time of these strains (Table II). Thus
neither phosphorylation of Ras2p nor the serine residues at these
positions contributed substantially to the essential function of Ras2p
under these conditions. However, this assay was relatively insensitive
to decreases in Ras2 function, because strains with less than 5%
wild-type Ras2p levels grow at normal rates.2
|
|
A widely used measure of RAS2Val19 function is the ability of this mutation to render strains highly sensitive to heat shock. Interestingly, combining the serine 214 to alanine mutation with the activating RAS2Val19 mutation resulted in a 20-fold increase in heat shock resistance (Table II). This result suggested that a decrease in Ras2 protein phosphorylation led to a detectable decrease in signaling through the Ras-cAMP pathway.
Growth on Other MediaThe five integrated strains were tested
for growth on multiple different media and at different temperatures.
This analysis included media containing drugs known to affect cAMP
production in other organisms (caffeine) and that block synthesis of
isoprenoids (lovastatin), which are necessary for the
post-translational processing of Ras proteins. However, these
components had no differential effect on growth of strains expressing
the five different alleles (data not shown). However, a differential
effect on growth was observed among these five strains when they were
grown on YPD medium containing formamide, which is an increasingly
popular reagent for identifying conditional alleles in yeast (35).
Cells containing Ras2Val19 protein grew more poorly on
medium containing formamide than cells with wild-type Ras2 protein
(Fig. 9). Strains expressing phosphorylation-defective
alleles of Ras2 protein grew better on medium containing formamide than
the wild-type strain (Fig. 9). In addition, an isogenic strain
expressing only the ras2-23 temperature-sensitive allele of
RAS2 also grew better on formamide than the wild-type
strain. Together, these data suggested that formamide sensitivity
provided a sensitive phenotypic read-out of Ras signaling as follows:
increases in signaling through the Ras-cAMP pathway (expression of
Ras2Val19) led to an increased sensitivity to formamide,
and decreases in signaling through the Ras-cAMP pathway (expression of
ras2-23 or a phosphorylation-defective mutant Ras protein)
led to decreased formamide sensitivity. These results were consistent
with the suppression of heat shock sensitivity observed when the S214A mutation was present in cis to the
RAS2Val19 mutation.
Because phosphorylation has been shown to alter the localization of some proteins, we tested whether mutations in the phosphorylation site(s) affected the localization of Ras2p. Proteins from the five strains above were fractionated into membrane and cytosolic pools. The localization of the Ras2 proteins in these cells was then analyzed by immunoblot. Reduction of phosphorylation had no effect on the localization of Ras2 protein (data not shown).
This study provided a systematic evaluation of phosphorylation of Ras1p and Ras2p of S. cerevisiae with a focus on Ras2p. Both Ras1p and Ras2p were found to be phosphorylated, and at least in the case of Ras2p, the phosphorylation was dynamic and occurred on roughly 60% of the Ras proteins. Studies of the half-life of Ras2p and of the half-life of its phosphorylation established that Ras2p phosphorylation was dynamic and potentially served a regulatory role. Interestingly, activated forms of both Ras1p and Ras2p were hypophosphorylated, and these proteins were present at only 10-20% the level of wild-type Ras protein.
We explored potential links between Ras protein phosphorylation and the role of Ras protein in nutrient sensing and signaling through the Ras-cAMP pathway. Multiple lines of evidence suggest the activity of the Ras-cAMP pathway is under tight feedback control. Intracellular levels of cAMP are unusually high when the activities of the cAMP-dependent protein kinases are lowered by mutation. In addition a lower level of cAMP is found in cells with an elevated level of a cAMP-dependent protein kinase (22). Both of these effects are dependent on the presence of RAS and CDC25. Because the modulation of cAMP levels in response to mutations of the cAMP-dependent protein kinase is dependent on RAS, we tested whether modulation of cAMP levels involved phosphorylation of Ras2p in a feedback loop by the cAMP-dependent protein kinases. However, in this study, several lines of evidence indicated that the difference between the phosphorylation of wild-type Ras2p and the hypophosphorylated Ras2Val19 protein did not reflect the activity level of the cAMP response pathway.
Because Cdc25p is partially responsible for regulating the ratio of GTP- and GDP-bound Ras proteins in yeast cells, and because GTP-bound Ras protein associates more poorly with Cdc25p than does GDP-bound Ras protein, we tested whether phosphorylation of Ras2 protein was modulated through an interaction with Cdc25p. The level of phosphorylation of Ras protein was not affected by the presence or absence of CDC25. Thus, Ras2p did not need association with Cdc25 protein to become phosphorylated.
The proper localization of Ras2 protein to the plasma membrane was not necessary for its phosphorylation, as localization-defective alleles of Ras2 protein were phosphorylated to a level indistinguishable from wild-type. However, because GFP-Ras2-SSIIS protein was sufficient to promote growth when present at elevated levels, presumably some of the mutant protein was, in fact, reaching the plasma membrane. Although it was possible that some of the Ras2-SSIIS protein was phosphorylated during transient associations with the membrane, if this were the mechanism by which the Ras2-SSIIS protein became phosphorylated, it is unlikely that the level of phosphorylation on the mutant and wild-type proteins would have been the same.
We anticipated that an allele of RAS2 that encoded a protein unable to be phosphorylated would provide insight into any regulatory role of the phosphorylation. The site of Ras2p phosphorylation was determined by a combination of peptide mapping by mass spectrometry and site-directed mutagenesis. These studies indicated only a single site of Ras2p phosphorylation and that approximately 60% of the Ras molecules were phosphorylated. The serine at position 214 was the preferred site of phosphorylation; however, the protein kinase(s) responsible for Ras2p phosphorylation was able to phosphorylate a variety of serine residues in the vicinity of position 214 when this site was altered. This region of the Ras2 protein of yeast is missing from mammalian Ras, for which three-dimensional structural information is available, precluding insight into the structural basis of this phosphorylation flexibility.
The phosphorylation of Ras2p was unaffected by changes in the activity of the Ras-cAMP response pathway as described above. However, changes in the phosphorylation state of Ras2p appeared to affect signaling through the Ras-cAMP pathway: phosphorylated forms of Ras2p appeared to be more active than non-phosphorylated forms based upon the phenotypes of mutant proteins that were significantly reduced for phosphorylation. Presumably Ras2Val19 protein is hyperactive despite its decreased phosphorylation state, because it is primarily bound to GTP and therefore relieved from modulation by phosphorylation. We found two phenotypes associated with decreases in Ras2 protein phosphorylation. The first was the 20-fold increase in survival upon heat shock of cells expressing the RAS2Val19-Ala214 allele compared with cells expressing RAS2Val19. Thus a single point mutation that led to a decrease in phosphorylation led to a detectable decrease in signaling through the Ras-cAMP pathway. The second phenotype was a decrease in sensitivity to formamide of strains harboring mutant Ras2 protein with significantly reduced phosphorylation. Formamide is thought to destabilize protein-protein interactions by disrupting hydrogen bonding (35). It is therefore possible that the phosphorylation on Ras2 protein is affecting the interaction of Ras2p with another protein(s) in the Ras signaling cascade, perhaps with the cyclase-associated protein CAP (see below). Because activated alleles of RAS were more sensitive to formamide, and because cells expressing a hypomorphic allele of RAS, ras2-23, were more resistant to formamide, formamide sensitivity in this context was a phenotypic readout for Ras signaling. If so, a decrease in phosphorylation led to a decrease in signaling through the Ras-cAMP pathway. Although we cannot rule out the possibility that any protein with six point mutations might exhibit a similar phenotype, the observation that a single point mutation at position 214, which reduces the phosphorylation of Ras2p by only 40%, can suppress the heat shock sensitivity of RAS2Val19 by 20-fold, suggested that it was the decrease in phosphorylation and not merely six substitutions that led to an altered phenotype.
The subtlety of these phenotypes suggested that phosphorylation of Ras2p may modulate Ras function but was not necessary for growth. Moreover, the phosphorylation of Ras2p occurred on a residue in the hypervariable domain of Ras2p which is dispensable for viability. However, we could not exclude the possibility that phosphorylation of Ras protein was essential, but only a small quantity of phosphorylated protein was sufficient for viability, because even in strains with multiple serine to alanine mutations there remained a detectable pool of phosphorylated protein.
A full cellular response to Ras2Val19 protein is dependent upon the adenylyl cyclase-associated protein (CAP), whereas CAP is not necessary for wild-type Ras function (see below) (36). CAP is a multifunctional protein; the amino-terminal third binds to adenylyl cyclase and has been implicated in activation of the cAMP-dependent protein kinase pathway. The carboxyl-terminal domain of CAP binds monomeric actin and serves a cytoskeletal regulatory function. In addition, CAP contains a short proline-rich domain between the Ras-related and cytoskeletal-related portions of the protein, which interacts with other proteins including yeast actin-associated protein Abp1p, and with cortical actin-containing structures (37). These observations suggest CAP may coordinate Ras-cAMP signaling and cytoskeletal organization.
Deletion of SRV2, which encodes CAP, is lethal in some strain backgrounds. Alleles of SRV2 that are deleted for the amino-terminal Ras-related portion of the protein are viable but suppress some phenotypes associated with activated RAS. In fact, alleles of SRV2 that encode proteins deleted for the RAS-responsive portion of CAP significantly decrease the ability of Ras2Val19 protein to stimulate the production of cAMP (38). However, elimination of the same portion of SRV2 has no effect on the ability of wild-type Ras2 protein to stimulate the production of cAMP, suggesting that activated and non-activated alleles of Ras protein interact differently with CAP (36). Perhaps the phosphorylation of Ras protein is related to its interaction with CAP. RAS also plays a role in the completion of mitosis in addition to its essential role in the G1 phase of the cell cycle (39). It is possible that the phosphorylation of Ras protein plays a role in the mitotic function of Ras protein. There has been no analysis of altered mitotic function(s) of activated alleles of RAS.
Finally, we address the mass discrepancy identified in this study. Our measurements of the mass of Ras2p exceeded the calculated molecular mass of the protein by 175 daltons. As described above, 80 daltons were accounted for by the phosphorylation of Ras2p on serine 214. At present, we do not know the basis of the remaining 95-dalton discrepancy. One possibility is that the reported sequence of Ras2 contains errors that add up to 95 daltons. However, we have sequenced the cloned RAS2 used in these studies and confirmed the published sequence. A second phosphorylation accounting for 80 of the remaining 95 daltons is inconsistent with our data. At present we favor the view that the 95 daltons represent one or more previously uncharacterized modifications of Ras2p which, based upon the confirmed mass of several overlapping Ras2 peptides from the conserved amino-terminal domain, presumably occurs in the hypervariable domain of the protein.
We thank David King for his tireless efforts at the mass spectrometer during the peptide analysis stage of these studies and without whom the phosphorylation site would not have been mapped. We also thank Sharlene Zhou and Michael Moore who provided peptide sequence. We thank Cindy Trueblood for her contributions to strain and plasmid constructions described in this work. We thank Randy Hampton and Stig Hansen for stimulating discussions and their excellent suggestions. We thank the members of the lab for their support.