The efficacy of monoclonal antibodies targeting the epidermal growth factor receptor (EGFR) pathway for treatment of colorectal cancer (CRC) depends on predictive biomarkers that can identify likely responders or non-responders to therapy [1
]. Downstream of EGFR, the RAS oncogene family members, including the KRAS and NRAS proto-oncogene GTPase isoforms, are critical molecular gatekeepers of cell signaling pathways that control cell growth, proliferation, differentiation, and survival [2
]. Gain-of-function mutations in KRAS are reported in 35–40% of CRC cases [3
], while activating NRAS mutations are detected at a much lower frequency ranging from 3–5% [5
]. Mutations in either gene result in hyperactive RAS signaling and constitutive activation of growth signals to downstream effectors and are thus predictive of unfavorable prognosis and poor response to anti-EGFR therapy [6
Despite the apparent homology and similar structural organization between KRAS and NRAS, numerous studies point to isoform-specific functions and differences in tissues and cancer hallmarks affected. KRAS mutations are more common in pancreatic, lung, and colon cancer whereas NRAS is more commonly mutated in melanomas and acute myeloid leukemia [7
]. Further, mutations in analogous hotspot codons (codons 12, 13, and 61) display different frequencies as well as distinct structural and functional defects [8
]. KRAS and NRAS mutations are generally considered mutually exclusive but there are recent reports of coexisting mutations in a caecal adenocarcinoma and contiguous tubulovillous adenoma [12
]. As a preemptive action, and based on limited clinical evidence, the American Society of Clinical Oncology and the European Society of Medical Oncology have issued revised guidelines for the management of metastatic colorectal cancer patients. RAS mutation profiling now includes codons 12, 13, 59, 61, and 117 for both KRAS and NRAS, and additionally codon 146 for KRAS [13
]. The non-redundancy of KRAS and NRAS functions, and the high level of tumor heterogeneity observed in CRC, however, suggest wide-ranging clinical and therapeutic implications. At present, functional characterization of these analogous mutations and identification of additional and rare mutations are highly relevant in light of emerging mechanisms of chemoresistance and therapeutic failure.
A retrospective targeted next-generation sequencing study of histopathologically verified malignant CRC patient tumor samples from the University of the Philippines Manila National Institutes of Health [15
] identified both canonical and novel or reported but uncharacterized somatic mutations in KRAS and NRAS. One of the mutations occurs within the hotspot KRAS codon 12 (NM_004985.3: c.34G>A), which switches the non-polar glycine residue at position 12 with a polar, uncharged serine residue (p.12G>S; G12S). This mutation has previously been identified as an independent prognostic factor of worse overall survival in CRC patients with colorectal liver metastases [16
]. Another missense mutation detected in the colorectal tissue samples occurs within the hotspot exon 2 of KRAS (NM_004985.3: c.175G>A), in codon 59 (p.59A>T; A59T). This mutation lies within the guanosine triphosphate (GTP) binding region of KRAS which implies potential impact on wild-type KRAS protein function. Although included in the list of KRAS mutations to account for, its utility as a negative predictor of response to therapy is not firmly established. A case report of a metastatic CRC patient with this mutation shows an apparent response to the anti-EGFR panitumumab, based on limited radiographic and carcinoembryonic antigen biomarker responses [17
]. The functional consequences of these two KRAS mutations are yet to be evaluated, although the Catalogue of Somatic Mutations in Cancer database (COSMIC db) gives them a FATHMM [18
] prediction score of 0.98, thus classifying them as both deleterious and pathogenic. Another putatively activating missense mutation was also detected within KRAS exon 3 (NM_004985.3: c. 410A>G) in codon 137 (p.137Y>C; Y137C) which was not found to be reported in COSMIC db and is thus deemed novel. Meanwhile, analysis of NRAS sequences revealed a mutation in codon 11 (p.11A>VA11V). This mutation (NM_002524.5: c. 32C>T) has not been previously reported in the literature and is hence considered novel, with its phenotypic consequences and functional impact on NRAS protein function yet to be determined.
This study reports the functional characterization of the non-canonical KRAS G12S and KRAS A59T mutations, and the novel KRAS Y137C and NRAS A11V mutations identified in Filipino colorectal cancer patient tumor samples. Their effects on cell proliferation, apoptosis, migration, cytoskeletal organization, and gross morphology, as well as their ability to promote epithelial-mesenchymal transition (EMT) and activate the mitogen-activated protein kinase (MAPK) response pathway were assessed. Divergent and overlapping oncogenic properties of the KRAS and NRAS variants were observed and can have therapeutic implications in patients with heterogenous tumors harboring mutations in one or both isoforms. Translatability of these into negative predictive markers for response to anti-EGFR therapy, however, awaits validation from clinical studies involving patients with the requisite mutational profile.
2. Materials and Methods
2.1. Site-Directed Mutagenesis and Generation of KRAS and NRAS Mutant Constructs
The open reading frame (ORF) fragments of wild-type KRAS isoform b (NM_004985.3) and KRAS G12D canonical mutant variants [19
], and the wild-type NRAS isoform 1 (NM_002524.5) and NRAS G12D canonical mutant variants (Yu, R.T and Garcia, R., unpublished) used in this study are available in the laboratory and were previously cloned into the pTargeTTM
mammalian expression vector. Site-directed mutagenesis was performed with the pTargeT-wild type (WT) KRAS construct as a template to generate the novel KRAS mutants, using primers listed in Table 1
. The pTarget-WT NRAS construct was used as a template for site-directed mutagenesis to generate the NRAS variants used in this study. The primers used are listed in Table 2
The mutant KRAS and NRAS variant fragments were amplified in a polymerase chain reaction (PCR) mixture containing a final concentration of 1X PCR buffer (Titanium® Taq PCR buffer, Clontech Laboratories, Inc., Mountainview, CA, USA), 0.125 μM of each deoxynucleoside triphosphate (Promega, Madison, WI, USA), 2 μM each of the forward and reverse primers, 1U Taq polymerase (Titanium® Taq polymerase, Clontech Laboratories, Inc.), and 50 ng of the pTargeT-WT KRAS or pTargeT-WT NRAS template. Amplification was performed with an initial denaturation temperature of 94 °C for 5 min, followed by 25–30 cycles of denaturation at 94 °C for 30 s, annealing at 55 °C for 30 s, and extension at 72 °C for 30 s, with a final extension step at 72 °C for 10 min. The amplified 5‘- and 3‘-fragments of each KRAS and NRAS mutant were then used as templates (25 ng of each fragment) for overlap-extension PCR to generate the full-length KRAS and NRAS mutant coding sequences, using the wild-type KRAS or NRAS outer F and R primers and the same PCR conditions stated above. The amplified products were cloned directly into the pTargeTTM vector via TA-cloning. Cloning orientation and sequence identity were confirmed by Sanger sequencing.
2.2. Cell Culture and Transfection of NIH3T3 and HCT116 Cells
The murine embryonic fibroblast cell line NIH3T3 and the human colorectal carcinoma cell line HCT116 were sourced from the American Type Culture Collection (Manassas, VA, USA; Cat. No. CRL-1658 and CCL-247). NIH3T3 cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM; Gibco; Thermo Fisher Scientific, Inc., Waltham, MA, USA) supplemented with 10% newborn calf serum (NBCS; Gibco; Thermo Fisher Scientific, Inc.), 100 U/mL penicillin/streptomycin, and 3.7 g/L sodium bicarbonate. HCT116 cells were cultured in Roswell Park Memorial Institute 1640 medium (RPMI-1640; Gibco; Thermo Fisher Scientific, Inc.) supplemented with 10% fetal bovine serum (FBS; Gibco; Thermo Fisher Scientific, Inc.), 50 U/mL penicillin/streptomycin, and 2.0 g/L sodium bicarbonate. All cells were maintained in a humidified incubator at 37 °C with 5% CO2
. NIH3T3 cells have routinely been used for characterizing RAS oncogenes and their mutant variants, given the cell line’s distinct advantage of not requiring complementary cooperative mutations for oncogenic transformation to manifest [20
]. For other cancer hallmark assays, especially those requiring an epithelial phenotype, HCT116 was used. Although a cancer cell line with a KRAS G13D mutant background, its oncogenic phenotype can still be modulated with transfected oncogenes [21
NIH3T3 cells were seeded for assays in 12-well plates at 40,000 cells/well, while HCT116 cells were seeded for assays in 12-well plates at 300,000 cells/well to achieve 80–90% confluence upon transfection. NIH3T3 cells were transfected with 1 μg pTargeTTM constructs using FuGENE® HD transfection reagent (Promega, Madison, WI, USA). HCT116 cells were transfected with 2 μg pTargeTTM constructs using Lipofectamine® 2000 (Invitrogen; Thermo Fisher Scientific, Inc., Carlsbad, CA, USA). Control setups were transfected with an equivalent amount of a mammalian reporter vector expressing the green fluorescent protein ZsGreen1 (pmR-ZsGreen1, Clontech Laboratories, Inc., Mountain View, CA, USA) to assess transfection efficiency by fluorescence microscopy. A transfection efficiency of 70% and 80% was routinely observed for NIH3T3 and HCT116 cells, respectively. For the morphological characterization experiments, the pTargeTTM constructs were co-transfected with pmR-ZsGreen1 at a vector ratio of 1:3 (pmR-ZsGreen:pTargeT) to allow the observation of cells that had high probabilities of being transfected and are overexpressing the KRAS and NRAS variants. The amounts and ratios of plasmid constructs and of FuGENE® HD or Lipofectamine 2000 were optimized to achieve at least 70% transfection efficiency between 24 and 72 h post-transfection for all morphological, functional, and molecular characterization experiments.
2.3. Cell Proliferation Assay
NIH3T3 cells were seeded in 96-well plates at 2500 cells/well and were transfected 24 h after seeding with 200 ng of each pTargeTTM construct in triplicates. Transfected cells were maintained in DMEM media supplemented with 2.5% NBCS. At 48 and 72 h post-transfection, the number of metabolically active cells per setup was measured after the addition of 10 µL CellTiter 96® Aqueous One-Cell Proliferation Assay Reagent (Promega) per well, and incubation at 37 °C until color development. Absorbance values of each setup were measured at 460 nm with a colorimetric plate reader (FLUOstar Omega Microplate Reader, BMG LABTECH, Cary, NC, USA). Cell counts were calculated from a standard curve generated using serial dilutions of an untransfected cell suspension and by plotting the number of cells vs. A460. Mean cell counts were calculated per setup for each time point.
2.4. Caspase 3/7 Assay
HCT116 cells were seeded in 96-well plates at 10,000 cells/well and transfected in triplicates, as described above. Transfected cells were incubated in RPMI media supplemented with 4% FBS alone, or with 6 mM sodium butyrate for induction of apoptosis. Ten microliters of Caspase-Glo® 3/7 Assay reagent (Promega) was added to each well at 24 h post-induction. The plates were incubated at ambient temperature for 2 h with gentle shaking. Mean luminescence readings per setup was measured with the FLUOstar Omega Microplate Reader.
2.5. Wound Healing Assay
NIH3T3 cells were seeded in 12-well plates and transfected as described above. A thin artificial wound was created on the transfected cell monolayers at >90% confluence using sterile toothpicks. Cells were afterwards maintained in DMEM media supplemented with 2.5% NBCS. Wound closure was monitored via time-lapse microscopy for a period of 20 h, capturing the same field of view per well per setup at 1 h intervals at 20x magnification using an Olympus IX83 inverted fluorescence microscope (Olympus Corporation, Tokyo, Japan). Percent open wound area was analyzed using the TScratch software [22
]. A linear graph of % open wound area vs. time was generated per setup, and the slope of each line equation was obtained to determine the rate of wound closure. The average rate of three field views per setup was plotted and statistically analyzed.
2.6. Western Blotting
The rabbit monoclonal anti-E-cadherin (24E10) (1:5000; Cat. No. 3195S), mouse monoclonal anti-N-cadherin (13A9) (1:5000; Cat. No. 13A9), rabbit monoclonal anti-vimentin (DZ1H3) XP® (1:1000; Cat. No. 5741S), rabbit anti-p44/42 MAPK (Erk1/2) (137F5) (1:1000; Cat. No. 4695S), rabbit anti-phospho-p44/42 MAPK (Erk1/2) (Thr202/Tyr204) (1:1000; Cat. No. 9101S), and mouse monoclonal anti-GAPDH (14C10) (1:10,000; Cat. No. 2118S) antibodies were from Cell Signaling Technology (Danvers, MA, USA). The goat anti-mouse IgG (H+L) (1:5000; Cat. No. 31430) and goat anti-rabbit IgG (1:10,000; Cat. No. 31460) secondary antibodies conjugated with horseradish peroxidase were from Invitrogen (Thermo Fisher Scientific, Inc.).
NIH3T3 cells were harvested 48 h post-transfection. Total protein was extracted using radioimmunoprecipitation lysis buffer (150 mM NaCl, 1.0% IGEPAL®
CA-630, 0.5% sodium deoxycholate, 0.1% SDS, 50 mM Tris, pH 8.0) (Sigma-Aldrich Corp.; Merck KGaA, St. Louis, MO, USA) supplemented with protease inhibitors (104 mM AEBSF, 80 μM aprotinin, 4 mM Bestatin, 1.4 mM E-64, 2 mM leupeptin, and 1.5 mM pepstatin A) (protease inhibitor cocktail; Sigma-Aldrich Corp.) and quantified through the bicinchoninic acid assay (BCA) method. For polyacrylamide gel electrophoresis, 30 µg of total protein per setup was loaded into Any kD™ Mini-PROTEAN®
TGX Stain-Free™ Protein Gels (Bio-Rad Laboratories, Inc., Hercules, CA, USA) and blotted onto polyvinylidene difluoride (PVDF) membranes. The membranes were blocked for 1 h at room temperature with 5% w
bovine serum albumin, heat shock fraction (Sigma-Aldrich Corp.) in 1 X Tris-buffered saline (TBST; 20 mM Tris, 150 mM NaCl, 0.1% Tween 20), and then probed overnight at 4 °C with the primary antibodies described above. After washing thrice with 1 X TBST, the membranes were incubated with the appropriate secondary antibodies for 1 h at room temperature. Signals were developed with enhanced chemiluminescence substrate and imaged using the ChemiDoc Touch Imaging System (Bio-Rad Laboratories, Inc.) using optimal exposure settings. Gene expression levels were obtained by densitometric analysis of digitized band intensities normalized against Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) or total protein loaded in stain-free gels, using GelQuant.NET software (v1.8.2. Biochemlabsolutions, University of California, San Francisco, CA, USA) provided by biochemlabsolutions.com. Total protein loaded in stain-free gels has been reported to provide superior accuracy and reliability in protein semi-quantification compared to commonly used housekeeping genes and was thus also used for protein expression normalization in this study to support our data [23
2.7. Actin Cytoskeleton Staining
NIH3T3 cells were seeded at 8000 cells/well in Millicell® EZ 8-well chamber slides (Merck KGaA, Darmstadt, Germany) and transfected with 600 ng of each pTargeTTM construct 24 h after seeding. Transfected cells were fixed with 4% paraformaldehyde at 48 h post-transfection for 20 min on ice, then permeabilized with 0.1% Triton X-100 in 1X PBS for 15 min at room temperature. After washing with 1X PBS, cells were blocked with 1% BSA in PBS for 20 min at room temperature, and then incubated in a 1:100 dilution of tetramethylrhodamine-conjugated phalloidin (Invitrogen; Thermo Fisher Scientific, Inc.) in 1X PBS for 1 h at room temperature with gentle shaking. The cells were again washed with 1X PBS before counterstaining the nuclei with Hoechst 33258 (1 µg/µL) for 5 min at room temperature. After the final washing step in 1X PBS, the cells were mounted in SlowFadeTM Diamond antifade mountant (Invitrogen; Thermo Fisher Scientific, Inc.) and were visualized under an inverted fluorescence microscope (IX83, Olympus Corporation), using a red fluorescent filter (λex/λem: 490/525 nm) to visualize filamentous actin structures, and a blue fluorescent filter (λex/λem: 355/465 nm) to visualize the nuclei.
2.8. Observation of Gross Morphology
NIH3T3 cells were seeded at 10,000 cells/well in 24-well plates and co-transfected with 500 ng of each pTargeTTM
construct together with 100 ng of empty pmR-ZsGreen1 vector 24 h after seeding. Morphological appearance (i.e., size, refringency, presence of filopodia, presence of lamellipodia, and depolarization) of transfected fibroblasts were examined under an inverted brightfield microscope (Olympus IX51, Olympus Corporation) 72 h post transfection. To quantitatively compare the transforming effect on cellular morphology by the different variants of KRAS and NRAS, the percentage of cells exhibiting transformed characteristics was determined for each transfection setup. Each transfected well was viewed in three different fields under 40x magnification. Using the Fiji image processing software (v1.52i, University of Wisconsin-Madison, Madison, WI, USA) [25
], fibroblasts with aberrant morphology were counted for each documented field. A total cell count per view was also performed. The mean percentage of morphologically transformed cells was then computed for all three fields of view and statistically compared among all setups.
2.9. ELK-TAD Luciferase Reporter Assay
To measure the ability of KRAS and NRAS variants to activate the mitogen-activated protein kinase (MAPK) response pathway, ETS domain transcription factor ELK-1-responsive luciferase reporter HEK293 cells (Signosis Inc. Silicon Valley, San Francisco, CA, USA; Cat. No. SL-0040-FP) were seeded at a density of 10,000 cells/well in 96-well plates. This cell line derived from human embryonic kidney stably expresses the firefly luciferase reporter gene under the control of the ELK1 response element, Gal4-upstream activating sequence (UAS), along with a fusion protein linking the transactivation domain (TAD) of ELK1 to Gal4. HEK293 cells were maintained in DMEM supplemented with 10% FBS and 100 U/mL penicillin/streptomycin and kept in a humidified incubator at 37 °C with 5% CO2.
HEK293 cells were transfected with 200 ng of each pTargeTTM construct 24 h after seeding. Media was changed to DMEM supplemented with 4% FBS 24 h post-transfection to deprive the cells of serum that can mask measurable effects which can be attributed to the KRAS or NRAS mutations. At 48 h post-transfection, cells were lysed through the addition of 20 µL of Passive Lysis Buffer (Promega Corporation; Cat. No. E1941) per well and incubation for 20 min at room temperature. Cell lysates were transferred into opaque 96-well plates prior to the addition of 100 µL of the Luciferase Assay Reagent (Promega Corporation; Cat. No. E1483). Luminescence signals per well were measured using a plate reader (FLUOstar Omega Microplate Reader, BMG LABTECH) under the pre-set Firefly luciferase protocol settings.
Parallel wells were also transfected with each pTargeTTM construct to obtain the relative cell count through fluorescent staining of DNA for normalization of luciferase expression. Media was removed at 48 h post-transfection and replaced with 100 µL of CyQuant NF reagent from the CyQUANT NF Cell Proliferation Assay kit (Thermo Fisher Scientific Inc.; cat. no. C35007) per well. Plates were then incubated at 37 °C for 1 h prior to measuring fluorescence readings per well using a plate reader (FLUOstar Omega Microplate Reader, BMG LABTECH) with an excitation wavelength of 485 nm, and emission detection at 530 nm. Luminescence readings from the luciferase assay were then normalized against fluorescence readings indicating relative cell count per setup and statistically analyzed.
2.10. Bioinformatics-Based Analysis of Potential Functional Impact of KRAS and NRAS Mutations
The predicted effects of each non-canonical/novel RAS mutation were first analyzed using the Polymorphism Phenotyping (POLYPHEN-2; version 2, Harvard Genetics, Cambridge, MA, USA) [26
], Sorting Intolerant from Tolerant (SIFT; version 5.2.2, Bioinformatics Institute, Singapore, Republic of Singapore) [27
], and Align GVGD (Huntsman Cancer Institute, University of Utah, Salt Lake City, UT, USA) [28
] prediction tools.
The impact of each non-canonical and novel mutation on KRAS and NRAS protein structure was assessed by building homology models recapitulating each mutation, using the solved 3D structure of human KRAS Q61H (PDB:3GFT) and wild-type HRAS (PDB:5DB30) as reference for the KRAS variants, and human NRAS (PDB:5UHV) as reference for the NRAS variants. Each KRAS mutant model was then built using the Accelrys Discovery Studio Client 2.5 (Dassault Systèmes BIOVIA, San Diego, CA, USA) Homology Modeler (HM) protocol, and the global main chain root-mean-square distance (RMSD) and Overlay Similarity (OS) of each mutant model was computed based on structural superimposition with KRAS WT. On the other hand, the NRAS mutant models were built using the SWISS-MODEL web-based client (https://swissmodel.expasy.org/
). The global main chain RMSD and OS of each mutant model was computed based on structural superimposition with NRAS WT using the BIOVIA Discovery Studio Visualizer (Dassault Systèmes BIOVIA).
Additionally, the effect of the KRAS mutations on GTP binding was assessed using the LigandFit docking module. The LigandFit docking algorithm involves the need to define binding sites of the KRAS protein based on known interacting amino acids. Monte Carlo-based methods to search for optimal ligand conformations were then done upon the initiation of the docking protocol, after which ligand fitting with GTP was carried out. Best-fit docked models of each KRAS mutation were selected based on the model with the highest Dock Score, and the binding energies of each model were compared among mutants.
2.11. Statistical Analysis
All data were statistically analyzed using unpaired two-tailed t-test to measure differences between two setups. Analysis of variance with post-hoc Tukey’s honest significant difference (HSD) was used to test significant differences between multiple setups. Data from all quantitative experiments were presented as mean ± standard error. In all tests, significance value was defined as * P < 0.05; ** P < 0.01; and *** P < 0.001.
In this study, the oncogenic impact of four non-canonical/novel KRAS and NRAS mutations identified from malignant CRC tumor samples [15
] were investigated. The two non-canonical KRAS mutations, G12S in exon 2 (NM_004985.3: c.34G>A; p.12G>S; COSMIC ID: COSM517) and A59T in exon 3 (NM_004985.3: c.175G>A; p.59A>T; COSMIC ID: COSM546) were previously reported in the COSMIC database, but have yet to be investigated for their functional effects upon extensive literature review. The KRAS G12S mutation has been reported to occur in 4.9%–5.7% of colorectal cancer samples harboring mutations in the KRAS gene [44
]. The KRAS A59T mutation has been detected in 24 tissue samples from the large intestine [44
], stomach [45
], salivary gland [46
], lung [44
], and cervix [47
], of which 13 cases (54%) were identified in CRC [44
]. The novel mutants KRAS Y137C (NM_004985.3: c. 410A>G; p.137Y>C) and NRAS A11V (NM_002524.5: c. 32C>T; p.11A>V) have not been previously reported to occur in CRC and as such, their functional impact and phenotypic consequences also remain unknown.
Mutations in the RAS gene have been described in numerous studies, and mutant RAS has since been validated as a driver of tumor initiation and maintenance [8
]. Mutations in KRAS hotspots impair the intrinsic activity of RAS by locking the mutant versions of the protein in a GTP-bound active conformation or prohibiting the binding of GTPase-activating proteins (GAPs) that facilitate GTP hydrolysis. This leads to the constitutive activation of downstream signaling cascades, including the MAPK pathway and PI3K/AKT pathway that is independent of ligand-induced activation from EGFR [48
]. This, in turn, results in uncontrolled cellular proliferation, eventually leading to tumor formation. The mutational status of KRAS is also highly concordant between primary tumor formation and metastasis, suggesting a role in the early process of carcinogenesis [48
]. Although less studied functionally and mechanistically, NRAS mutations have been associated with worse prognosis and overall survival when compared to KRAS mutant- or wild type KRAS-driven metastatic CRC [49
]. Further, patients who are carriers of NRAS mutations (7.7%; 1/13) and are wild type for KRAS were shown to respond less to cetuximab treatment compared to NRAS wild type carriers (38.1%; 110/289). This demonstrates the utility and importance of including NRAS, as well as other downstream effectors of EGFR such as BRAF and PIK3CA, in mutational profiling especially among patients who are wild type for KRAS [5
In silico analyses show that similar to the canonical KRAS G12D and NRAS G12D mutations, the non-canonical KRAS G12S and A59T mutations as well as the novel NRAS A11V mutation have high potential functional impact, based on three amino acid substitution analysis tools: PolyPhen-2, SIFT, and Align GVGD. Further, the novel KRAS Y137C mutation has a higher Grantham Difference (GD) score compared to KRAS G12D and was predicted to be in the highest class of mutations potentially altering protein function, according to Align GVGD. These results suggest that the rare mutations cause significant changes in RAS protein structure due to their critical positions within the protein. Protein structure modeling using the Homology Modeler protocol of Discovery Studio further reveals KRAS A59T as the mutant with the highest deviation from wild type by RMSD. The canonical mutant KRAS G12D, as expected, and the non-canonical mutant KRAS G12S, on the other hand, have the lowest overlay similarity to wild type, suggesting subtly nuanced effects on protein structure conferred by these mutations. Molecular docking simulations and analysis of ligand-receptor interactions between GTP and the KRAS mutants revealed marked differences in amino acid interactions. G12S did not mimic the charge-based shift in atomic interactions facilitated by G12D, most likely due to serine being uncharged. A59T and Y137C had marked and minimal effects on predicted binding interactions, respectively. In silico structural modeling of the novel NRAS variant A11V vis-à-vis the canonical mutant NRAS G12D revealed insignificant changes in the overall protein structure of both variants. OS calculations further revealed closer similarity in the structures of wild type NRAS and A11V compared with G12D, pointing to more significant effects on function by the subtle structural changes in the latter.
Functional and morphological characterization confirmed the transformative effects of the non-canonical and novel KRAS and NRAS mutations. Similar to NIH3T3 cells transfected with the canonical KRAS mutant G12D, cells transfected with KRAS G12S, A59T, and Y137C all displayed increased proliferation and migration rates. Increased vimentin expression in the epithelial cell line HCT116 was also observed upon transfection with the canonical KRAS G12D, the non-canonical KRAS G12S and the novel KRAS mutant Y137C, compared to KRAS wild type-expressing cells and vector only control. There was no apparent effect on E-cadherin expression which may be indicative of partial EMT. Cells expressing the canonical mutant NRAS G12D and the novel mutant NRAS G11V both showed no apparent effects on proliferation, migration and EMT. Instead, both NRAS mutants conferred increased resistance to apoptosis in HCT116, compared to wild type NRAS and vector only control. On the other hand, the canonical KRAS G12D, the non-canonical KRAS G12S and KRAS A59T, and the novel KRAS Y137C, did not cause resistance to apoptosis.
In contrast to the divergent effects of KRAS and NRAS mutants on the aforementioned hallmarks, all KRAS and NRAS mutants studied caused both altered F-actin organization and morphological changes when transfected into NIH3T3 cells, compared to wild type and vector only controls. Cytoplasmic shrinkage, disorganized actin filaments, and the presence of cellular protrusions were apparent in all mutant setups. Gross morphological changes such as cell rounding, refringency, and smaller cell size were also observed. Overall, these findings are consistent with the observed migratory phenotypes of the rare KRAS mutants and reveal an apparent effect of the rare NRAS A11V mutation on cell motility and adhesion which was not observed in the scratch wound assays. Cytoskeletal reorganization and pseudopodial formation are highly suggestive of a migratory phenotype, as these are among the first few steps taken by the cell to initiate movement [43
]. The results therefore suggest that the expression of NRAS G12D and A11V may induce cells to initiate migration via their effects on the cytoskeleton but may not be involved in cellular motility, per se.
Identification and elucidation of the signaling pathways involved in these phenotypic readouts can shed light on this apparent discrepancy. RAS-mediated cancer progression is known to engage multiple effectors. PI3Kα and PLC ε, for instance, can mediate cytoskeletal reorganization whereas RALGDS can regulate migration [50
Differences in ability of the various mutants to signal via the downstream effector ERK and to activate ELK-1, a downstream nuclear target and well-characterized substrate of ERK, hint at distinct signaling pathways engaged by mutations in KRAS and NRAS. While cells transfected with KRAS G12S and KRAS A59T showed elevated p-ERK levels and significant ELK-1 activation similar to KRAS G12D, KRAS Y137C did not show any difference compared to KRAS wild type and vector only controls. NRAS wild type, NRAS G12D, and NRAS A11V did not have any effects on p-ERK levels but NRAS G12D appears to activate ELK-1 significantly.
Taken together, the results of this study point to overlapping and distinct phenotypic consequences of KRAS and NRAS mutations. Although preliminary, the p-ERK and ELK-1 activation assays provided initial data suggesting that different RAS mutants may engage different signaling pathways. Deviations from wild type structure, however minor, can affect amplitude and duration of binding to downstream effectors. While in silico modeling and ligand docking simulations were able to capture differences in predicted binding interactions, phenotypic characterization of the mutants only confirmed their oncogenicity and net effect on distinct cancer hallmarks. More mechanistic studies are warranted to identify the signaling pathways involved in mediating the different phenotypic readouts, to determine the differential oncogenic potencies of the mutants, and to determine if the structural constraints imposed by the different mutations may have altered association with downstream effectors. Mouse models, on the other hand, may help determine whether these mutations have a role in colorectal cancer initiation and progression, as well as resolve their functional consequences in vivo.