Biochemical characterization of a novel KRAS insertion mutation from a human leukemia.

A novel alteration in exon 1 of KRAS was detected by single strand conformational polymorphism analysis of DNA amplified from the bone marrow of a 4-year-old child with myeloid leukemia. Sequencing of this mutant allele revealed an insertion of three nucleotides between codons 10 and 11 resulting in an in-frame insertion of glycine. Expression of the mutant protein in NIH 3T3 cells caused cellular transformation, and expression in COS cells activated the Ras-mitogen-activated protein kinase signaling pathway. Surprisingly, Ras·GTP levels measured in COS cells established that this novel mutant accumulates to 90% in the GTP state, considerably higher than a residue 12 mutant. Biochemical analysis confirmed that the higher Ras·GTP levels correspond to a dramatic decrease in intrinsic GTP hydrolysis as well as resistance to GTPase-activating proteins. This mutation is the first dominant Ras mutation found in human cancer that does not involve residues 12, 13, or 61, and its biochemical properties should help elucidate the mechanism of oncogenic activation.

A novel alteration in exon 1 of KRAS was detected by single strand conformational polymorphism analysis of DNA amplified from the bone marrow of a 4-year-old child with myeloid leukemia. Sequencing of this mutant allele revealed an insertion of three nucleotides between codons 10 and 11 resulting in an in-frame insertion of glycine. Expression of the mutant protein in NIH 3T3 cells caused cellular transformation, and expression in COS cells activated the Ras-mitogen-activated protein kinase signaling pathway. Surprisingly, Ras⅐GTP levels measured in COS cells established that this novel mutant accumulates to 90% in the GTP state, considerably higher than a residue 12 mutant. Biochemical analysis confirmed that the higher Ras⅐GTP levels correspond to a dramatic decrease in intrinsic GTP hydrolysis as well as resistance to GTPase-activating proteins. This mutation is the first dominant Ras mutation found in human cancer that does not involve residues 12, 13, or 61, and its biochemical properties should help elucidate the mechanism of oncogenic activation.
Mutations of RAS proto-oncogenes are among the most common alterations detected in human cancer cells (1,2). Activated RAS oncogenes function as dominant alleles and encode proteins with aberrant biochemical activity. Wild-type Ras proteins regulate cellular growth and differentiation by cycling between inactive GDP-bound and active GTP-bound forms (reviewed in Refs. 3 and 4). Oncogenic mutant proteins show defective intrinsic GTP hydrolysis and therefore accumulate elevated levels of Ras⅐GTP. This, in turn, leads to increased signaling through effector proteins. The defect in intrinsic GTP hydrolysis is augmented by an insensitivity to the Ras GTPaseactivating proteins p120-GAP and neurofibromin, which nor-mally bind to Ras⅐GTP and accelerate the rate of hydrolysis to Ras⅐GDP (reviewed in Ref. 5).
Ras activation leads to increased flux through a number of effector pathways, including the Raf/mitogen-activated protein (MAP) 1 kinase cascade (6). In normal cells, this is primarily achieved by activation of nucleotide exchange on Ras. In contrast, tumor cells with activated RAS oncogenes accumulate high levels of Ras⅐GTP because of a defect in the rate of GTP hydrolysis. A wide variety of mutations have been described which mimic oncogenic RAS by transforming the phenotype of cultured cell lines, yet only residues glycine 12, glycine 13, and glutamine 61 have been found mutated in actual human tumors. This observation suggests that such point mutations are uniquely poised to alter the biochemical properties of Ras such that a constitutive signal is induced in the cell.
In analyzing the DNA of a patient with myeloid leukemia, we found a novel activating mutation in the KRAS gene that involves the insertion of three nucleotides between codons 10 and 11, such that an anomalous glycine residue is introduced. We denote this mutant 10 Gly 11 to signify this genetic alteration. This mutant Ras protein shares with the codon 12, 13, and 61 mutations the ability to activate the Ras pathway and morphologically transform cell lines. Analysis of the nucleotide dissociation and hydrolysis activities reveals similarities, but also differences, with other activating mutations.

EXPERIMENTAL PROCEDURES
Materials-The 10 Gly 11 mutation was introduced by polymerase chain reaction (PCR) into a plasmid encoding the Glu-epitope tagged K-Ras(4B) protein for baculovirus expression (7). An NcoI-EcoRI fragment from this plasmid was introduced into the NcoI-EcoRI sites of pTrc99A (8) in order to generate pTrc 10 Gly 11 for bacterial expression. An NcoI-XbaI fragment from the baculovirus plasmids for K-Ras(wildtype), K-Ras( 10 Gly 11 ), and K-Ras(Asp 12 ) was ligated into the SacI-XbaI sites of pcDB (9) using a SacI-NcoI linker in order to generate the corresponding pcDB-Ras plasmids. Ras and GAP proteins were purified as described previously (7,10). The pEXV3-ERK2-tag plasmid has also been previously described (11).
Mutation Detection Techniques-Genomic DNA was extracted from bone marrow cells, amplified with RAS-specific oligonucleotide primers by PCR, and screened for mutations by single strand conformational polymorphism (SSCP) analysis exactly as described elsewhere (12). Mutations were confirmed by DNA sequence analysis of cloned PCR products using Sequenase, version 2.0.
MAP Kinase Assays-COS cell plates were transfected with 2 g of pEXV3-ERK2-tag plasmid and 5 g of pcDB-Ras plasmid by electroporation (13). In order to measure MAP kinase activity, the ERK2-tag protein was selectively immunopurified using antibodies directed to the epitope tag. Cells were lysed in 20 mM Tris, pH 8, 137 mM NaCl, 10% glycerol, 1% Triton X-100, 50 mM NaF, 1.5 mM MgCl 2, 1 mM EGTA, 1 mM vanadate, 1 mM Pefabloc, 10 g/ml leupeptin, 10 g/ml aprotinin, and cell debris was removed by centrifugation. 25 g of ERK2-tag antibodies and 10 l of protein A-Sepharose FF (Pharmacia Biotech Inc.) were added and incubated for 1 h at 4°C with constant rotation. The Sepharose beads were pelleted and washed three times with lysis buffer and then once with kinase buffer (30 mM Tris, pH 8, 20 mM MgCl 2 , 2 mM MnCl 2 ). Kinase activity was then measured by adding 30 l of kinase buffer containing 1 M ATP (300 Ci/mmol) and 7 g of myelin basic protein (Upstate Biotechnology, Inc.) and incubating for 30 min at 30°C with constant agitation. Assays were stopped by addition of SDS sample buffer and analyzed on 14% polyacrylamide gels. Incorporation of 32 P into the myelin basic protein was judged by autoradiography and quan- tified on an Ambis radioanalytic scanner.
Transformation Assays-Focus formation assays in NIH 3T3 cells were performed as described previously (14).
GTP Loading in COS Cells-COS cell plates were transfected with 5 g of pcDB-Ras plasmids by electroporation and grown for 48 h in DMEM supplemented with 10% fetal calf serum. The cells were then washed three times in phosphate-free DMEM and labeled with [ 32 P]orthophosphate (500 Ci per sample) overnight in phosphate-free DMEM containing 10 mM HEPES, pH 7.4, and 1 mg/ml bovine serum albumin at 37°C in a 5% CO 2 incubator. Analysis of Ras-bound nucleotides was performed exactly as described previously (15). GTP Loading in Vitro-To measure the ratio of GTP/GDP at equilibrium, 0.1 M Ras protein was incubated with 2.5 M [␣-32 P]GTP (20 Ci/mmol) for 5 h at 37°C in Buffer A (20 mM HEPES, pH 7.3, 2 mM MgCl 2 , 2 mM dithiothreitol, 0.1 mg/ml bovine serum albumin). Reactions were terminated by adding to 20 l of Y13-259 protein G-Sepharose beads and allowing them to bind for 40 min at 4°C. These beads were washed four times with 1 ml of PBS containing 2 mM MgCl 2 and eluted by adding 20 l of elution buffer followed by heating at 68°C for 20 min. These samples were analyzed as described previously (15).
Intrinsic and GAP-stimulated GTP Hydrolysis-Assays were performed in Buffer A using the protein concentrations indicated in the figures and the legends. GTP hydrolysis was monitored by assaying [ 32 P]phosphate release from [␥-32 P]GTP as described previously (15).

RESULTS
SSCP analysis of KRAS exon 1 fragments amplified from the bone marrow DNA of a 4-year-old boy with acute myelogenous leukemia revealed two abnormal fragments that migrated more slowly than the normal bands (Fig. 1A). The intensities of the normal and abnormal fragments were equivalent. Surprisingly, allele-specific oligonucleotide analysis showed no mutations at codons 12 or 13, and sequencing of multiple clones showed no alterations from codon 12 through the end of exon 1 (12). After the abnormality persisted on multiple independent PCR-SSCP experiments, sequence analysis of cloned PCR products exon 1 demonstrated an in-frame 3-nucleotide insertion (GGA) between codons 10 and 11 in multiple clones (Fig. 1B). We designated this mutant 10 G 11 to indicate the insertion of a glycine residue between amino acids 10 and 11.
In order to confirm that this is truly an activating mutation, the K-Ras( 10 Gly 11 ) cDNA was subcloned into the mammalian expression vector pcDB, expressed in NIH 3T3 cells and scored for the ability to form morphologically transformed foci. Table  I depicts the results of three independent experiments comparing K-Ras( 10 Gly 11 ) with K-Ras(wild-type) and K-Ras(Asp 12 ), the most common Ras mutation found in human cancers. These K-Ras data confirm that 10 Gly 11 is also able to cause focus formation at a similar frequency to that of K-Ras(Asp 12 ). In these assays, K-Ras(wild-type) had no effect.
As a further test for activation of Ras signaling, the ability to activate the Raf-MEK-ERK pathway was examined. As shown in Fig. 2, K-Ras( 10 Gly 11 ) induced an activation of recombinant ERK2 activity when cotransfected into COS cells. This activation was considerably higher than that caused by K-Ras(wildtype) and similar to that caused by K-Ras(Asp 12 ).
Since the ability of oncogenic Ras to activate downstream effectors derives from an increased percentage of bound GTP, the nucleotides bound to the various K-Ras proteins were determined in transfected COS cells. The percentage of Ras⅐GTP was determined after labeling the unstimulated, transfected cells with inorganic [ 32 P]phosphate (Fig. 3A). Surprisingly, the Ras⅐GTP levels of K-Ras( 10 Gly 11 ) (93.5%) were considerably higher than those of K-Ras(Asp 12 ) (45.7%). This suggested that the K-Ras( 10 Gly 11 ) intrinsic GDP dissociation rate may be considerably higher, the GTP hydrolysis rate may be considerably lower, or that an endogenous cellular factor may selectively down-regulate K-Ras(Asp 12 ).
In order to test this latter possibility, purified K-Ras proteins were incubated with excess GTP and the nucleotide dissociation and hydrolysis activities allowed to equilibrate for 5 h. Bound nucleotides were then analyzed using methods similar to those for the COS cell labelings (Fig. 3B). The Ras⅐GTP levels on the mutant proteins were comparable in COS cells and in solution, while the Ras⅐GTP levels on the wild-type protein was considerably lower in the COS cells (7.0%) than in the cell-free reaction (31.1%). This observation suggests that neither K-Ras( 10 Gly 11 ) nor K-Ras(Asp 12 ) is down-regulated in the COS cells. In contrast K-Ras(wild-type) is markedly down-regulated in the COS cells, and this is consistent with the hypothesis that the cellular GAPs are active on K-Ras(wild-type) but not on either mutant protein.
Since the difference in Ras⅐GTP levels between K-Ras( 10 Gly 11 ) and K-Ras(Asp 12 ) is apparently not due to cellular factors, it appeared that intrinsic properties should be accountable. Measurement of intrinsic GTPase activities confirmed this hypothesis (Fig. 4A). While K-Ras(Asp 12 ) has only a 2-fold lower GTPase rate than K-Ras(wild-type), that of K-Ras( 10 Gly 11 ) is over 10-fold lower. In addition, both mutant proteins are insensitive to both p120-GAP (Fig. 4B) and neurofibromin (Fig. 4C), consistent with the observed insensitivity of these Ras proteins to endogenous COS cell down-regulators.
Ras⅐GTP levels are determined by a counterbalance between   20 5 Ϯ 1 the rates of GTP hydrolysis and nucleotide dissociation. Therefore, it was of interest to compare the nucleotide dissociation rates of the mutant proteins. As shown in Table II, while the GTP dissociation rates are similar, the GDP dissociation rate off K-Ras( 10 Gly 11 ) is faster than that off K-Ras(Asp 12 ). This property further accentuates the biochemical differences between the mutant proteins and contributes to the dramatic differences seen in Fig. 3. As discussed below, these findings support the idea that intrinsic biochemical properties are important determinants of the transforming potency of mutant Ras proteins. The altered nucleotide dissociation rates collaborate with reduced GTPase activity and insensitivity to the Ras GAPs to raise the Ras⅐GTP levels.

DISCUSSION
The discovery of a novel activating RAS mutation in the bone marrow of a child with myeloid leukemia has implications for the role of Ras proteins in human cancer. Deregulated signaling through Ras appears to play a central role in myeloid leukemogenesis and may occur by at least three different genetic mechanisms (16). First, 20 -40% of leukemic bone marrows show activating RAS mutations (1,2,12). Second, the chimeric Bcr-Abl protein that is produced in chronic myelogenous leukemia stimulates nucleotide exchange on Ras by binding to the upstream adaptor protein Grb-2 (17). Third, the neurofibromatosis type 1 (NF1) tumor suppressor gene encodes neurofibromin, a GAP for Ras. Children with NF1 are predisposed to myeloid leukemia, and we have shown that these leukemias commonly delete the normal parental NF1 allele (18) and show elevated levels of Ras⅐GTP (15). The fact that we detected this novel KRAS mutation in the bone marrow of a child with leukemia led us to consider the possibility that the 10 Gly 11 insertion might specifically desensitize the protein to neurofibromin while leaving the sensitivity to p120-GAP intact. This is not the case (Fig. 4), and this novel mutation confers resistance to both GAP proteins and markedly reduces the intrinsic rate of GTP hydrolysis.
Although 10 Gly 11 is the first insertional RAS mutation de-tected in a human cancer, previous data suggest that insertion mutations can activate Ras proteins. Specifically, recombinant HRAS insertions in the region surrounding codon 12 were found to induce focus formation and soft agar growth (19). Also, activation of the KRAS gene in chemically induced rodent tumors was shown to involve duplication of residues in the codon 12 region (20). More recently, a mutation in the Rasrelated TC21 gene in a human leiomyosarcoma cell line was found to be due to an insertion in the region of TC21 corresponding to codon 12 of Ras (21). While little biochemical data have been reported for any of these mutants, all appear to activate the transforming potential of the proteins. However, it is of particular interest to note that the wild-type TC21 allele was lost in the leiomyosarcoma cell line (21). This is in contrast to the K-Ras( 10 Gly 11 ) mutant reported here, which is clearly a dominant mutation that arose in a leukemia that retains one normal KRAS allele (Fig. 1).
In the oncogenic activation of RAS genes during tumor initiation and progression, the biochemical properties of the Ras proteins are subjected to rigorous selection criteria. Previously, mutations of only three residues, namely Gly 12 , Gly 13 , and  seventh and eighth lanes). B, purified Ras proteins were incubated with excess [␣-32 P]GTP for 5 h at 37°C before immunoprecipitation and separation of bound nucleotides by thin-layer chromatography. Again, the order is vehicle control (Vehicle) (first and second lanes), K-Ras(wild-type) (Wildtype) (third and fourth lanes), K-Ras( 10 Gly 11 ) (10G11) (fifth and sixth lanes), or K-Ras(Asp 12 ) (Asp12) (seventh and eighth lanes). In both panels, the percentages of Ras⅐GTP were determined on an Ambis radioanalytic scanner and are shown at the bottom.
Gln 61 , had been found in human tumors. The identification of the K-Ras( 10 Gly 11 ) mutation in a primary human tumor solidifies our understanding of the biochemical criteria for mutation selection. This mutation activates the GDP dissociation rate and dramatically inhibits the GTP hydrolysis rate intrinsic to the Ras protein. At the same time, the ability to interact with effectors is preserved, as judged by the ability to activate the MAP kinase pathway and induce cellular transformation.
The insertion between Gly 10 and Ala 11 falls within the phosphate binding loop of Ras (22). From the three-dimensional structure of Ras, it appears that the side chain of Lys 16 makes contacts with the carbonyls of Gly 10 and Ala 11 , and it is possible that the glycine insertion in K-Ras( 10 Gly 11 ) disrupts this interaction. This disruption could account for the altered biochemical properties of K-Ras( 10 Gly 11 ), since Lys 16 also makes critical contacts with the ␥-phosphate of GTP. The structural and mechanistic consequences of phosphate binding loop mutations are still not understood (23). It will be of interest to determine the structure of K-Ras( 10 Gly 11 ) in more detail, since this may shed light on the mechanism of GTP hydrolysis.
The biochemical properties of K-Ras( 10 Gly 11 ) underscore the importance of GAP proteins in limiting cellular growth by down-regulating Ras. In this context, it is striking that a difference in Ras⅐GTP levels between K-Ras(wild-type) and K-Ras(Asp 12 ) proteins is only evident in transfected COS cells (compare Fig. 3, A and B). This clearly implicates a cellular factor that down-regulates Ras as an important determinant of resistance to transformation. Like other mutant Ras proteins, K-Ras( 10 Gly 11 ) is insensitive to GAPs (Fig. 4). Genetic experiments in yeast and mammalian cells also implicate GAPs as playing a central role in controlling growth. Yeast strains that have inactivated the NF1 homologs ira1 and ira2 display increased sensitivity to heat shock, presumably because these mutants are unable to down-regulate Ras (24). Murine embryos that are homozygous for targeted disruptions of either the p120-GAP or Nf1 genes are nonviable (25)(26)(27). Finally, humans with NF1 are predisposed to myeloid leukemia and other types of cancer, and these tumor cells frequently show loss of the normal NF1 allele (5,18). While these biochemical and genetic data emphasize that GAP function is essential to properly regulate Ras signaling in vivo, the intrinsic GTPase activities of Ras proteins strongly influence their transforming potential. Thus, a Ras mutant with Gly 12 replaced by Pro or Ala is not transforming, despite insensitivity to GAPs, since intrinsic GTPase and GDP dissociation rates are not sufficiently altered (28). Analysis of K-Ras( 10 Gly 11 ) provides additional insights into the specific requirements for determining the transforming potency of Ras mutants. A better understanding of the biochemical defects associated with oncogenic Ras may lead to improved treatments for a number of human cancers.  4. Rates of intrinsic and GAP-catalyzed GTP hydrolysis on K-Ras(wild-type) (Wildtype) (closed circles), K-Ras( 10 Gly 11 ) (10G11) (closed squares), and K-Ras(Asp 12 ) ( Asp12 ) (closed triangles). A, time course of intrinsic GTP hydrolysis on purified Ras proteins. The background without Ras protein is shown with open circles. B, p120-GAP-catalyzed GTP hydrolysis on the purified Ras proteins. C, neurofibromincatalyzed GTP hydrolysis on the purified Ras proteins. The GAP-related domain of neurofibromin (NF1-GRD) was used in this experiment.