Structural characterization of NRas Q61L and Q61R mutants and their potential affect on intrinsic hydrolysis. by Shores Salter

Transcription

1 Structural characterization of NRas Q61L and Q61R mutants and their potential affect on intrinsic hydrolysis by Shores Salter B.S. in Chemistry, Northeastern University A thesis submitted to The Faculty of the College of Science of Northeastern University in partial fulfillment of the requirements for the degree of Master of Science April 22, 2016 Thesis directed by Carla Mattos Professor of Chemistry and Chemical Biology

2 ACKNOWLEDGEMENTS There have been many individuals along the way that have contributed to my successful journey through six years of higher education at Northeastern University. First and foremost I would like to thank my parents, Bob and Lorraine, and two older brothers, Ian and Cam, for their assistance and support along the way. I would also like to thank Roseann Sdoia, Mike Materia, and Shana Cottone with whom I have formed a life changing, family-like relationship with over the past three years that has been a continued source of personal inspiration. Everything becomes easier when you surround yourself with a great support system, and having family and friends behind me throughout my six years at Northeastern has meant the world to me. To my fellow lab members, thank you for all the assistance and insight you have all shared with me. Everyday I felt as though I learned something new either through literature or through discussions with fellow lab members who enlightened me on topics with which I was less familiar. Specifically, thank you to Dr. Christian Johnson who provided great mentorship in my first year in the Mattos Lab. Christian guided my incorporation into the lab and while teaching me tremendous amounts about the Ras protein and its role in cancer research, he also helped me learn new laboratory techniques with which I had limited experience. Additionally, Jillian Parker was also a great reference when questions arose during my research. She had direct influence on the completion of my thesis as she helped loop NRasQ61R-GppNHp crystals that eventually led to crystal structure. I also would like to thank Northeastern University as a wholly institution for granting me the opportunity to first complete my undergraduate studies and then for the opportunity to complete my graduate degree here as well. Specifically, I would like to thank Graham Jones whose persuasiveness, determination, and vision were largely responsible for my decision to join ii

3 the Northeastern Chemistry Department. Additionally, I would like to thank my undergraduate and graduate academic counselors, Katie Dioli and Cara Shockley. They each have been very approachable and helpful throughout my Northeastern experience. Thank you to my graduate thesis committee members, Penny Beuning and Zhaohui Sunny Zhou, for taking the time to scrutinize my thesis. I have worked closely with you both whether it be through classroom experiences or group meetings and I cherish your opinions and feedback. Finally, I would like to thank my advisor through the entire graduate process, Carla Mattos. Thank you Carla for granting me the opportunity to work in your lab as an undergraduate student, and also for allowing me to continue on in completion of my graduate degree. It has been a long enlightening experience in becoming an accredited scientist, and your mentorship along the way has contributed immensely to my professional maturation and growth as a scientist and an individual. iii

4 ABSRACT OF THESIS Ras GTPases are a subfamily of a larger family of hydrolase enzymes, which function as key regulators of signal transduction pathways. Site-specific mutations in nucleotide and effector protein binding regions have the capacity to render Ras in a constitutively active state leading to unregulated cellular proliferation and tumor growth. Mutants of the NRas isoform have been found to be a leading cause of melanomas, however drug targeting of such mutants has remained largely unsuccessful. Lack of clinical success in the targeting of Ras mutants is due in part to the dynamic active and allosteric sites of the protein, which are connected through interactions critical to the successful hydrolysis of GTP in the catalytic conformation of Ras. Our investigation is focused on the characterization of the structures of two highly transforming NRas Q61 mutants, Q61L and Q61R. Comparison of the mutant structures with previously investigated structures of wild-type NRas as well as with other Ras structures in the PDB, reveals how communication networks connecting distant regions of the protein are impaired by the mutations, leading to constitutively active Ras. We propose the NRas mutants at residue 61 found commonly in melanomas preferentially affect the Ras/Raf/MEK/ERK downstream signaling pathway critical in the regulation of cellular proliferation. iv

5 TABLE OF CONTENTS Acknowledgements Abstract of Thesis Table of Contents Lists of Figures List of Tables ii iv v vi vi Introduction 1 Methods and Materials 10 Results 15 Discussion 27 Conclusions 31 Appendices 32 References 36 v

6 LIST OF FIGURES AND TABLES Page Number Figure 1. Generalized Ras GTPase cycle 2 Figure 2. Ras state 1 and state 2 Conformations 4 Figure 3. Bridging and Nucleophilic water molecules in catalytic Ras 7 Figure 4. Allosteric switch conformations, R state vs. T state 8 Figure 5. NRasQ61L-GppNHp active site 18 Figure 6. NRasQ61R-GppNHp active site 18 Figure 7. NRasQ61L-GppNHp, NRasQ61R-GppNHp, 1CTQ active sites 19 Figure 8. R97 Orientation 22 Figure 9. Allosteric bonding network conserved between NRasQ61 mutants 23 Figure 10. α3 switch II H-bonding network, HRas-GppNHp R state 24 Figure 11. α3 switch II H-bonding network, NRasQ61L-GppNHp 25 Figure 12. α3 switch II H-bonding network, NRasQ61R-GppNHp 25 Table 1. NRasQ61L-GppNHp data collection and structure refinement data 32 Table 2. NRasQ61R-GppNHp data collection and structure refinement data 33 Table 3. α3/l7 shift measurements 34 Table 4. Allosteric bonding network NRasQ61L-GppNHp, measurements 34 Table 5. Allosteric bonding network NRasQ61R-GppNHp, measurements 34 Table 6. α3 switch II H-bonding network NRasQ61L-GppNHp, measurements 35 Table 7. α3 switch II H-bonding network NRasQ61R-GppNHp, measurements 35 vi

7 Introduction Over 20 years have been dedicated to the investigation and characterization of the structural biology of the signaling protein Ras. Since its ability to transform mammalian cells into a neoplastic state was first identified in the 1980 s, Ras has been a popular oncogene targeted in the field of cancer research [1, 2]. Ras is a hub signaling protein involved in many signaling pathways with downstream functions in the regulation of cell proliferation, differentiation, and survival [1]. Active Ras has been shown to interact with multiple effector proteins such as Raf [3], RalGDS [4], PI3K [5], and NORE1A [6] with each interaction influencing a different cellular response. The most well studied signaling pathway involving Ras is the Ras/Raf/MEK/ERK pathway, which is a mediator of cellular proliferation [7]. The membrane bound Ras protein functions via a binary switch mechanism depicted below in Figure 1, which balances the intracellular levels of active and inactive states of the GTPase in the regulation of many cellular processes [8]. In its inactive or off state Ras is bound in complex with guanine diphosphate (GDP), while in its active or on state it is rather found in complex with guanine triphosphate (GTP) [9]. Guanine Nucleotide Exchange Factors (GEFs) are regulatory proteins that mediate the release of GDP and subsequent loading of GTP into the active site of Ras [10], thus stimulating the signaling protein into its activated state. Meanwhile a secondary regulatory protein, GTPase Activating Proteins (GAPs), accelerate the normally slow rate of intrinsic hydrolysis of GTP to GDP facilitating downstream signaling inactivation [11, 12]. Evidence has shown that site-specific mutations occurring primarily at residues glycine 12 (G12), glycine 13 (G13), and glutamine 61 (Q61) convert Ras into an active proto-oncogene. Mutations at these nucleotide-binding residues favor GTP binding, trapping Ras in a constitutively active state, which leads to unregulated cellular proliferation [13]. 1

8 P$ P$ P$ Ras$$ GTPase$ GDP$ G D P GAP$ GEF$ P$ P$ P$ Ras$ GTPase$ GTP$ G T P Fig. 1. A generalized representation of the binary switch mechanism that regulates activation and inactivation of Ras GTPase signaling. Ras GTPase exists as three isoforms that are encoded by proto-oncogenes (NRAS, KRAS, and HRAS), with each isoform composed of a catalytic G domain and a C-terminal hypervariable region (HVR) [14]. The HVR is a small portion at the C-terminal end of Ras, stretching from residues This terminal region undergoes post-translational modifications resulting in highly divergent HVR sequences between N, K, and HRas isoforms [15]. HVR modifications for each isoform act as a communication network responsible for Ras-membrane localization, having a direct impact on protein function. HVR importance is highlighted by its role in the placement of the catalytic G domain at the cytoplasmic membrane surface, leading to an inherent interaction between the active site of Ras and membrane proteins [14]. G domain placement is guided by Ras-membrane interactions that are modified by bound-nucleotide state, posttranslational modifications at the HVR [16], and isoform specific residues along α-helix 4 (α4) and α-helix 5 (α5) of the allosteric lobe [17]. Structurally the three isoforms have a similar catalytic G domain that is composed of two separate lobes: the effector and allosteric lobes. The effector lobe is completely conserved between all isoforms and is critical to signaling, as key 2

9 residues in this region are directly involved in effector protein recognition and nucleotide binding [17]. Isoform sequence differences outside the HVR are primarily observed in the allosteric lobe that is 90% conserved between isoforms, and contribute to the communication network between the active site of Ras and the cellular membrane [17]. The G domain (residues 1-166) is the catalytic portion of Ras, which contains key catalytic residues involved in the hydrolysis of GTP to GDP [10]. Structurally, it comprises six β-sheets and five α-helices in the following order, β1-α1-β2-β3-α2-β4-α3-β5-α4-β6-α5 [14]. Active site residues that participate in the catalytic hydrolysis of GTP are found in the N- terminal half of the protein, within residues 1-86, which is referred to as the effector lobe [10, 18]. Within this region are secondary motifs β1-α1-β2-β3-α2-β4. The active site comprises the P loop (residues 10-14), switch I (residues 30-40), and switch II (residues 60-76), and mutations within these regions often lead to oncogenesis [19]. In active Ras, residues in these regions converge over the β and γ-phosphates of GTP and form H-bonds to coordinate nearby water molecules that are critical to hydrolysis and signal inactivation [15]. Meanwhile the C-terminal half of the Ras G domain is referred to as the allosteric lobe and includes residues [10]. Secondary structure motifs α3-β5-α4-β6-α5 are found within this region. Residues within the allosteric lobe are important in the formation of salt-bridges and bonding networks that are essential to the connection between the N and C-terminals of the Ras G domain [20]. As will be discussed further in this investigation, the orientation of α3 of the allosteric lobe relative to switch II of the active site is critical in determining whether Ras-GTP is in a favorable or unfavorable conformation for intrinsic hydrolysis [21]. 3

10 Loop'7' α3' α5' α2' Switch'II' α4' P'Loop' GppNHp' Switch'I' Fig. 2. The globular structure of Ras. Secondary motifs that contribute to the catalytic conformation of the protein are labeled. The structural difference between state 1 and state 2 Ras is evident through comparison of the switch I region in each structure. Ras in the state 1 conformation is colored in light blue (PDB code: 4EFL). Ras in the state 2 conformation is colored in green (PDB code: 3K8Y). Switch I in the state 1 conformation is flipped away from the active site of Ras, while the switch I region in state 2 Ras adopts a closed conformation as the Y32 side chain stacks over the γ-phosphate of GppNHp. Oncogenic mutations of the NRAS isoform have been identified in tumor types such as melanoma, acute myeloid leukemia, and thyroid cancer [22]. Currently, approved therapies for melanoma treatment are primarily focused on the inhibition of BRAF proto-oncogene downstream signaling as part of the MEK/ERK pathway, which is found in approximately half of all melanomas [23, 24]. Recent studies have shown a combination of immunotherapeutic drugs such as the BRAF inhibitor vemurafenib and other oncogene inhibitors such as ipilimumab (CTLA-4 inhibitor) elicit significant patient response in down-regulating metastatic melanoma 4

11 tumor cell growth [25, 26]. However, high rates of resistance to BRAF inhibition have been observed in patients within a year of treatment [24, 25]. This deficiency of BRAF targeted inhibition therapies leads to an opportunity for the development of novel therapeutics designed for the inhibition of NRAS mutant proto-oncogenes. NRAS mutations at residues Q61, G12, and G13 occur at observed rates of 60%, 35%, and 5%, respectively, in all NRAS-linked tumor types [9]. Q61 mutants have been shown to predominate in cutaneous melanomas as 80% of melanoma-linked NRAS mutants occur at residue Q61 [27]. This finding agrees with the fact NRAS wild-type codon 61 (CAA) contains a pyrimidine doublet, which has been shown to be a target for UV radiation induced damage [28]. Behind BRAF, NRAS is the second leading mutant oncogene found in metastatic melanomas, at an approximate rate of 15-30% [29, 30]. To date, Ras has been considered an undruggable proto-oncogene as attempts to directly inhibit the protein have not translated into clinical success [31]. However recent preclinical studies have offered potential therapies designed to inhibit the unregulated signaling of Ras mutants through indirect pathways [31]. Structure elucidation of highly transforming Ras mutants is the first step in understanding inhibition pathways. The elusiveness of Ras for drug targeting is due primarily to its dynamic and flexible active site, particularly switch I and switch II regions [13, 19]. The active site of Ras undergoes different conformational states that can be linked to regulatory and effector protein interactions, which guide the hydrolysis of GTP. Different conformational states, referred to as state 1 and state 2 in Figure 2 above, are identified by the positioning of switch I in relation to the bound nucleotide [32]. State 1 describes an opened conformation in which Y32 and switch I are directed away from the nucleotide, and hydrolysis of GTP is required to be a GAP-catalyzed reaction [13, 33]. A more tightly packed protein is observed in the state 2 conformation of Ras, 5

12 in which the hydroxyl group of Y32 is closely nestled to the γ-phosphate of GTP and switch I is closed over the nucleotide [13]. Ras in its state 2 conformation is observed when it participates in the Raf/MEK/ERK pathway as Raf kinase-ras Binding Domain (Raf-RBD) has been shown to have binding specificity to the switch I region of Ras influencing a closed conformation [13, 33]. Open and closed conformations will be used in this investigation to describe state 1 and 2, respectively. Sufficient hydrolysis of Ras-GTP has long been thought to be a biological process entirely dependent upon GAP catalysis through a Ras/RasGAP interaction at the conserved active site [11]. GAPs have been proposed to enhance Ras hydrolysis rates through different mechanisms, such as stabilizing the catalytic conformation of the protein as well as the stabilization of a charged transition state formed during γ-phosphate cleavage [34, 35]. The arginine finger of GAPs has been observed to participate in the active site of the opened conformation of Ras stabilizing a negative charge that develops on the nucleotide during its loose transition state of hydrolysis, ultimately facilitating γ-phosphate cleavage [36]. More recently however, studies have suggested Ras in the Ras/Raf-RBD complex adopts a state 2 closed conformation similar to the Raps/Raf-RBD complex where intrinsic hydrolysis is favored [37, 38]. The mechanism of intrinsic hydrolysis is dependent upon the complete ordering of catalytic components of the Ras active site as seen in Figure 3. The complete ordering of the Ras active site includes catalytic residues such as the Q61 side chain, as well as the positioning of two water molecules referred to as the bridging and nucleophilic water molecules [39, 40]. Previous studies have shown the transition state of the intrinsic hydrolysis mechanism also adopts a loose conformation, and the positioning of the bridging water molecule is critical in replacing the role the GAP arginine finger plays in GTP hydrolysis in the opened Ras conformation [40-42]. 6

13 Q61$ G60$ W175$ W189$ O1G$ T35$ Y32$ Fig. 3. The active site of HRas in the catalytic, closed conformation (PDB code: 3K8Y). The positions of the bridging (W189) and catalytic (W175) water molecules are critical to the catalytic conformation of Ras. The interaction between W189 and W175 with O1G of the γ- phosphate have lengths of 2.6Å and 2.8Å respectively. The bridging water molecule, placed between Y32 and Q61, functions to neutralize a negative charge that forms on the γ-phosphate during hydrolysis. Once switch I has been stabilized in the closed conformation the α-helix 3 and loop 7 (α3/l7) regions undergo a second conformational change that affects switch II residues, referred to as the allosteric switch [14]. Closed Ras transitions rapidly between allosteric switch conformations, referred to as the R and T state, which favor and disfavor intrinsic hydrolysis respectively. R and T state HRas can be seen in Figure 4 below. While in the R state conformation α3/l7 of Ras interacts with α4, shifting the helical region approximately 3Å towards the allosteric lobe and away from switch II of the active site. In contrast, α3/l7 in the T state conformation is observed to remain in closer proximity to switch II residues, resulting in 7

14 Loop$7$ α3$ α4$ Switch$II$ GppNHp$ Fig. 4. The allosteric switch conformations of Ras. Ras in the R state catalytic conformation is colored in green (PBD code: 3K8Y). Ras in the T state anticatalytic conformation is colored in yellow (PDB code:2rge). The conformational effects caused the allosteric switch mechanism of state 2 Ras is observed by the shift of α3 toward α4 in catalytic Ras creating room for the ordering of switch II residues. The α3 helix in T state Ras maintains a closer conformation to the effector lobe resulting in the disorder of switch II residues as seen in the above crystal structure. overall sequence disorder of N-terminal switch II residues, seen in Figure 4, and sluggish rates of intrinsic hydrolysis [21]. This investigation will be focused on the structure elucidation of two highly transforming NRas Q61 mutants, leucine (Q61L) and arginine (Q61R), and the structural comparison with the previously studied structure of wild-type NRas and other Ras proteins in the PDB [43]. Previous studies have shown Q61 mutants Q61L and Q61R to be highly transforming in NRas-linked metastatic tumors with Q61R mutants having a particular high expression in melanoma tumors [26, 27, 30]. Structure elucidation and qualitative analysis of intrinsic 8

15 hydrolysis potential of each mutant relative to the wild-type will be useful as novel therapeutics are developed to target highly transforming NRas mutants. This is because intrinsic hydrolysis activated by ligand binding at a remote allosteric site is now thought to be the mechanism through which signaling is inactivated in the presence of Raf [37]. Thus, while GAPs may still play a role in the depletion of available free Ras-GTP, impaired intrinsic hydrolysis is likely to play a key role in NRas Q61 mutants associated with melanomas. Additionally, structural insight into the active conformation of NRas proteins is valuable as the NRas isoform has been the least studied of the Ras isoforms, thus there is very limited published structural data. 9

16 Methods and Materials Site-directed mutagenesis, cell growth, and expression of wild-type and mutant NRAS: All mutagenesis procedures were done using the truncated version of the NRAS gene as our DNA template, yielding a truncated Ras protein (18 kda). The truncated version of NRas encompasses the effector and allosteric lobes but does not include the HVR, as it contains residues The current investigation is focused on the comparison of wild-type NRas- GppNHp with Q61 mutants leucine (NRasQ61L-GppNHp) and arginine (NRasQ61R-GppNHp). The same mutagenesis, cell growth, expression, and purification procedures were carried out for all NRas proteins. The QuikChange II Site-Directed Mutagenesis Kit protocol from Stratagene was followed to amplify site-directed mutants of NRAS using a pair of complementary oligonucleotide primers containing our desired mutations. The DNA coding for NRAS mutant genes were encoded in the pet21 vector. TOP10 E. coli cells were then transformed with pet21 encoding NRas and allowed to grow on Luria Broth (LB) agar plates containing ampicillin [44]. Sequence accuracy of transfected Ras was confirmed by isolating the DNA plasmid of the transfected E. coli cells following the E.Z.N.A Plasmid DNA Mini Kit (manufacturer), which was then outsourced for sequencing (Eurofins Operon). Once the accuracy of the DNA plasmid sequence was confirmed BL21 competent E. coli cells were transformed with the pet21 vector for each respective NRAS mutant. Competent, transfected BL21 cells were then used to create glycerol stocks for each mutant, which were stored at -80 o C. Glycerol stocks were prepared for the following NRAS mutants: Q61L, G12V, G12D, Q61K, Q61R, Q61H, and G12S. To obtain sufficient protein expression for purification of NRas wild-type and mutant proteins, transfected BL21 cells for each respective mutant were grown overnight in 200 ml of LB media (25 g/l) containing ampicillin (50 mg/l) at 37 o C [45]. After the overnight growth, 25 ml of the cell 10

17 solution was added to six 1-liter LB + amp solutions, and grown for approximately minutes at 37 o C. Once each LB solution grew within a density of OD, as measured by a spectrometer at OD600, 1 ml of IPTG (120 mg/ml to 120 µg/ml final concentration) was added to each flask to induce protein expression. Protein expression was induced at 32 o C for six hours. After six hours of protein expression BL21 cells were pelleted by centrifugation (13,000 rpm for 20 minutes) and stored at -80 o C. Purification of NRas: The purification of each NRas protein was followed as previously described by Dr. Greg Buhrman for the purification of HRas [21]. Each cell pellet was resuspended and solubilized in 100 ml Buffer A (20 mm Tris, 5 mm MgCl 2, 50 mm NaCl, 5% Glycerol, 1 mm DTT, 20 µm GDP, ph 8.0) on ice with added protease inhibitors (Benzamidine, Leupeptin, and E-64). Cells were lysed through five cycles of sonication. Cellular debris was removed by centrifugation and by the addition of 35 ul of 10% w/v poly(ethyleneimine) (PEI). After PEI precipitation the protein solution was centrifuged, then syringe filtered using a 5.0-µm filter followed by a µm filter. The filtered protein solution was injected into a chilled superloop, which was then connected to a HiPrep QFF 16/10 anion exchange column (GE Healthcare). The protein solution was added to the column using Buffer A and then eluted off the column into fraction tubes using a 40% gradient of Buffer B (20 mm Tris, 5 mm MgCl 2, 1 M NaCl, 5% Glycerol, 1 mm DTT, 20 µm GDP, ph 8.0). Fractions containing our desired NRas protein were determined by SDS- PAGE analysis. NRas-containing fractions were concentrated to <5 ml using a Millipore 10,000 MW cutoff concentrator. Next, concentrated NRas solutions were added to a HiPrep Sephacryl S-200 HR gel filtration column (GE Healthcare). Using Buffer A, the protein solution was loaded and eluted at 11

18 a rate of 1 ml/min into fraction tubes. Again, NRas containing fractions were identified by SDS- PAGE analysis and concentrated to <5 ml. After concentration the protein solution was injected onto a HiTrap Q HP, 5 x 1 ml anion exchange column. The protein was loaded onto the column using Buffer A and then eluted into fraction tubes using a 20% gradient of Buffer B. SDS-PAGE analysis was used to identify NRas containing fractions, and then the identified fractions were concentrated to <2 ml. Finally, a Bradford assay was used to determine the concentration of each protein solution at 595 nm [46]. At this point, our concentrated NRas solutions were in the inactive GDP bound conformation. For our investigation we were interested in observing NRas in its active, or GTP bound state. Instead of GTP we exchanged GDP for a non-hydrolyzable GTP analogue, guanosine 5ʹ-[β,γ-imido]triphosphate (GppNHp) [47]. For this exchange we used an exchange buffer (32 mm Tris, 200 mm Ammonium Sulfate, 10 mm DTT, 0.15% n-octylglucopyranoside, ph 8.0) and a stabilization buffer (20 mm HEPES, 50 mm NaCl, 20 mm MgCl 2, 1 mm DTT, ph 7.5) [21]. Lab member Derion Reid was responsible for the GppNHp exchange of the NRas wild-type protein while I exchanged the NRasQ61L and NRasQ61R mutants following the same protocol. Concentrated mutant NRas was added in 2 ml increments over a chilled NAP-25 column that had been pre-equilibrated with 15 ml of exchange buffer. Protein fractions were collected in 1.5 ml microcentrifuge tubes and fractions containing protein were identified using Bradford reagent. Protein containing fractions were pooled into a 15 ml conical tube and 1 mg of GppNHp was added for every 10 mg of NRas protein. This protein sample was allowed to incubate in the presence of p-aminobenzamidine cross-linked beads (SIGMA) for 45 minutes at 37 o C. After 45 minutes 1 M MgCl 2 (20 mm final concentration) was added to the protein solution to quench the exchange reaction. After centrifugation the exchanged protein solution 12

19 was then added in 2 ml increments over a chilled Nap-25 column pre-equilibrated with stabilization buffer. Again, protein-containing fractions were identified using the Bradford reagent, pooled, then concentrated to <500 µl. The concentration of the exchanged NRas solution was determined by Bradford assay, and 50 µl aliquots were flash-frozen and stored at - 80 o C. Protein crystallization conditions, X-ray crystallography, and structure refinement: NRasQ61L-GppNHp protein stocks (5.9 mg/ml) were thawed on ice and screened for crystal growth using the hanging drop crystallization method [48]. Crystal screen conditions from Crystal Screen 1 and 2 as well as PEG/Ion Screen 1 and 2 (Hampton Research) were added to the individual wells (200 µl/well) of four 24-well plates. Crystals were grown by mixing 2 µl of crystal screen solutions with 2 µl of thawed protein solution on a glass slide, and then the slide was inverted over the corresponding 24-well plate creating a sealed environment. The crystals were allowed to grow undisturbed at 18 o C and were observed for crystal formation every week. After two weeks, crystal growth was observed in multiple wells. The best crystals grew in the well containing Crystal Screen 18 (0.2 M Magnesium acetate tetrahydrate, 0.1 M Sodium cacodylate trihydrate ph 6.5, 20 % w/v Polyethylene glycol 8,000). Crystals from this well were cryoprotected in 30% glycerol, looped, and flash frozen in liquid Nitrogen. NRas-GppNHp (3.5 mg/ml) was purified exactly as outline above, however the nucleotide exchange, crystallization, and structure refinement was carried out by lab member Derion Reid. NRas-GppNHp crystals grew in the same conditions as NRasQ61L-GppNHp (0.2 M Magnesium acetate tetrahydrate, 0.1 M Sodium cacodylate trihydrate ph 6.5, 20 % w/v Polyethylene glycol 8,000) [43]. Crystallization conditions for NRasQ61R-GppNHp (8.2 mg/ml) were screened differently than the Q61L mutant. The sitting drop method [48] was used 13

20 for crystal screening, as 1 µl of protein solution was mixed with 1 µl of crystal screen and allowed to crystalize in wells of a 96-well plate. The best crystal formation occurred in the well containing Crystal Screen #22. Condition optimization was tested in a 24-well plate by changing salt and PEG concentrations, and crystalizing a 4-µL drop as described above. Protein crystals formed and were collected from the following condition: 0.2 M Sodium acetate trihydrate, 0.1 M Tris hydrochloride ph 8.5, 30% w/v Polyethylene glycol 4,000. Protein crystals were analyzed on a MicroMax-007 Rigaku X-ray generator. Diffractions patterns were collected over 360 frames, integrated, and scaled using HKL-3000R [49]. Structure determination was carried out using PHENIX and COOT programs, used for solving macromolecular protein structures [50, 51]. The initial phasing model was based on the published structure for HRas-GppNHp, 3K9L [21]. Modifications to each crystal structure were carried out using COOT while PHENIX was used for the refinement of the overall structure. 14

21 Results Crystallization of NRas protein was different for each variant. Crystal growth was optimized based on conditions such as high salt and protein concentrations as well as nonphysiological temperature and ph. Multiple conditions were conducive to crystal growth, however all crystals are not grown to the same quality as identified by X-ray diffraction patterns [52]. Crystals of NRasQ61L-GppNHp mutant were crystallized with symmetry of the P space group and diffracted to a resolution of 2.00 Å in conditions optimized from Crystal Screen #18 (Hampton Research): 0.2 M Magnesium acetate tetrahydrate, 0.1 M Sodium cacodylate trihydrate ph 6.5, 20% w/v Polyethylene glycol 8,000. These were the same crystal conditions in which previously investigated NRas-GppNHp was crystallized [43]. Interestingly, NRas-GppNHp was also modeled into the P space group at a resolution of 1.7 Å, allowing for a direct comparison of mutation driving structural differences between the two proteins [43]. Meanwhile, NRasQ61R-GppNHp crystal growth occurred in conditions optimized from Crystal Screen 22 (Hampton Research): 0.2 M Sodium acetate trihydrate, 0.1 M Tris hydrochloride ph 8.5, 30% w/v Polyethylene glycol 4,000. Additionally, NRasQ61R-GppNHp was crystallized with symmetry of the P space group to a resolution of 1.8 Å. Space groups P and P are two of the most commonly found space groups of protein structures published on the PDB [53]. Data collection and refinement statistics are presented in Tables 1 and 2 for NRasQ61L-GppNHp and NRasQ61R-GppNHp structures respectively. It must be noted that different space groups have a major influence on the structure of proteins modeled within their constraints [54, 55]. HRas crystals with symmetry of space group R32, have so far been the least constrained by crystal contacts, with a switch I stabilized in the closed conformation as observed in the Ras/Raf complex and switch II free of crystal contacts 15

22 [21]. Structural modeling effects due to the different space groups are taken into account during this investigation. Active site crystal structure of highly transforming NRas mutants Q61L and Q61R: Lab member Dr. Christian Johnson provided a thorough structural analysis of NRas- GppNHp, the structure of which will be referenced throughout this investigation in comparison to my investigated Q61 mutant structures [43]. In both mutant structures switch I is stabilized by crystal contacts in the state 2 conformation associated with the more active state of Ras-GTP. A catalytically competent state of Ras is reliant upon the order of active site residues and the positioning of the bridging and nucleophilic water molecules, as seen in Figure 3, that participate in the hydrolysis reaction [40]. As mentioned, the NRasQ61L-GppNHp crystal was modeled with symmetry of space group P Crystal packing of the P space group most closely mimics the conformation of Ras/RasGAP complex [11]. As a result of the similarity to the Ras/RasGap conformation, the Y32 side chain of the Q61L mutant adopts an opened conformation while other switch I residues are closed over the GppNHp nucleotide as seen in Figure 5. The hydroxyl group of the Y32 side chain does not participate in the active site. Rather the hydroxyl group forms a direct H-bond with the GppNHp γ-phosphate of a neighboring symmetry molecule. An identical conformation is observed in the switch I region of the wildtype structure and other Ras proteins modeled in the P space group [39]. In this crystal form there is minor disordering of side chains along the N-terminal end of switch II in both the wildtype NRas-GppNHp and NRasQ61L-GppNHp structures. Meanwhile the switch II backbone is well ordered, which is not the case in the structure of HRas mutants modeled in the P space group [13]. In the HRas structures of several Q61 mutants, the mutant side chain is part of a hydrophobic cluster involving residues from both switch I and switch II. In spite of different 16

23 crystal packing, the aliphatic leucine side chain of NRasQ61L-GppNHp is placed in position to contribute to a similar hydrophobic cluster formed over the nucleotide, which includes residues Y32, P34, I36, L61, and Y64 depicted in Figure 5. The hydrophobic cluster is similar to one observed in the Ran-GppNHp/importin-β complex that buries the pre-catalytic water molecule and GppNHp γ-phosphate, leading to impairment of the GTP hydrolysis reaction until importin-β is released [56]. The extension of the leucine side chain into the hydrophobic cluster near the active site is facilitated by the unwinding of α2 at the N-terminal end of the switch II region. The shortened length of α2 of the Q61L mutant, compared to wild-type HRas-GppNHp modeled in the P space group (PDB code: 1CTQ) [39], allows for switch II residues in the elongated catalytic loop 4 region to extend towards switch I. This conformational difference provides the leucine side chain the opportunity to interact more intimately with the hydrophobic cluster that buries the γ-phosphate end of GppNHp. The α2 motif in the NRasQ61L-GppNHp mutant includes residues S65 to T74, while α2 in the HRas-GppNHp extends to the N-terminal end of switch II including residues E62, E63, and Y64 [39]. The length of α2 is conserved between our mutant structures as the switch II motif adopts a similar conformation in NRasQ61R-GppNHp. 17

24 Y64$ L61$ O1G$ G60$ W20$ I36$ Y32$ P34$ T35$ GppNHp$ Fig. 5. NRasQ61L-GppNHp active site. The aliphatic Leucine side chain is part of a hydrophobic cluster that buries the pre-catalytic water molecule and GppNHp γ-phosphate. This cluster includes residues Y32, P34, I36, L61, and Y64. In this structure the Y64 side chain is disordered. The Y32 side chain is flipped away from the γ-phosphate due to crystal contacts of the P space group. Although buried beneath the hydrophobic cluster, the nucleophilic water molecule, W20, is in position for catalysis as it is coordinated by T35 (2.7Å), G60 (3.1Å), and γ- phosphate (2.7Å). There is no bridging water molecule observed. G60$ R61$ Y64$ O1G$ W20$ I36$ Y32$ T35$ GppNHp$ P34$ Fig. 6. NRasQ61R-GppNHp active site. The aliphatic portion of ther61 side chain contributes to the hydrophobic pocket as it closes over the γ-phosphate of GppNHp. The charged guanidinium group is solvent exposed. In this structure Y64 is well ordered and clearly isolates the GppNHp nucleotide and pre-catalytic water molecule from the surrounding bulk solvent. The pre-catalytic water molecule, W20, is in position for nucleophilic attack as it is coordinated by T35 (2.9Å), G60 (3.3Å), and γ-phosphate (2.7Å). However, there is no bridging water molecule due to the direct interaction between Y32 and the γ-phosphate (2.5Å). 18

25 Y64$ 61$ G60$ I36$ Y32$ P34$ T35$ Fig. 7. Active site of NRasQ61L-GppNHp (cyan), NRasQ61R-GppNHp (magenta), and HRas- GppNHp (PDB code: 1CTQ) (pink). There is no bridging water and the nucleophilic water molecule, which is conserved between NRasQ61L-GppNHp and NRasQ61R-GppNHp, is coordinated by residues T35, G60, and the γ-phosphate. The nucleophilic water and GppNHp γ- phosphate are isolated from the bulk solvent due to the hydrophobic cluster formed by residues Y32, P34, I36, L61, and Y64 closing over the active site. The NRasQ61R-GppNHp mutant was crystallized with symmetry of the P space group, and a clear difference between P and P space groups can be observed in Figure 7 when comparing the orientation of the Y32 side chain in each respective structure. Switch I of the Q61R structure is also in the state 2 conformation. However, as seen in Figure 6, it remains entirely closed as the Y32 tyrosyl group stacks over the GppNHp γ-phosphate forming a direct H-bond interaction with O1G (2.5 Å). The well-ordered arginine side chain also directly participates in the active site, as its aliphatic portion contributes to the hydrophobic cluster seen in the Q61L mutant and Ran-GppNHp/importin-β complex [56]. Due to the long side chain, the charged guanidinium group of arginine behaves similarly to the amine group of lysine in the previously characterized HRasQ61K-GppNHp mutant (PDB code: 2RGD) as it is exposed to the bulk solvent [21]. GppNHp$ 19

26 Also important to the active conformation of Ras is the placement of the bridging and nucleophilic water molecules [40]. Highly-transforming Ras Q61 mutants previously published in the PDB often lack a bridging water molecule, which is typically positioned between Y32 and Q61 to neutralize the negatively charged transition state that forms during intrinsic hydrolysis [13, 14, 40]. The shift of Y32 in mutant structures towards the nucleotide, as observed in our NRasQ61R-GppNHp structure, creates a direct interaction between the hydroxyl group of Y32 and the GppNHp γ-phosphate (2.5Å) as seen in Figure 6. This direct interaction leaves no room for positioning of the bridging water molecule. A bridging water molecule is also missing in the Q61L structure, as the Y32 side chain is in an opened conformation making a direct interaction with the γ-phosphate of a symmetry-related molecule. On the other hand, the nucleophilic water molecule is present and conserved between mutant structures Q61L and Q61R as W20 and W8 form a direct H-bond (2.7 Å) with O1G of the GppNHp γ-phosphate in the respective structures. The amide of the residue G60 backbone, O1G of the γ-phosphate, and the carbonyl group of T35 coordinate the pre-catalytic water, as observed in the HRas-GppNHp structure. The positioning of α3 without calcium acetate in the allosteric site: The transition between different conformational states of Ras has been shown by NMR and molecular dynamics simulations to be a rapid transition that occurs at a nanosecond timescale [37, 57, 58]. HRas-GTP transitions between a primarily T state conformation while in solution to an increased presence of R state upon crystallization, as indicated by the majority of PDB HRas-GTP structures expressing the R state conformation. It has been proposed that this transition can be guided by the crystallization of Ras proteins in reservoir solutions containing calcium acetate [14, 21]. R and T state conformation comparison of two R32 modeled HRas- GppNHp proteins [14] shows the R state is stabilized as a calcium ion and acetate molecule bind 20

27 at the allosteric site of Ras. H-bonds formed between the guanidinium group of R97 and acetate ions shown in Figure 8 work to anchor α3 closer to α4, initiating the allosteric-active site communication network observed in the catalytic R state conformation of Ras. Our investigated proteins were not crystallized in the presence of calcium acetate, thus there is no substrate binding at the allosteric sites. Upon calcium acetate ligand binding, α3 of R state HRas-GppNHp (3K8Y) experiences an approximate 3-Å shift towards α4 relative to α3 of T state HRas- GppNHp (2RGE), as seen in Figure 4 [14]. This shift can be confidently determined to be a cause of an allosteric binding effect due to the minimal crystal contacts along α3 and switch II regions of the R32 space group. When observing the orientation in α3 of wild-type NRas, Q61L, and Q61R mutant structures we observed an approximate shift of 1.4 Å, 2.0 Å, and 1.4 Å toward α4 relative to the 2RGE structure, respectively. Shift measurements of C-terminal α3 and Loop 7 residues for all three structures can be found in Table 3. The orientation of α3 in all three investigated structures is not indicative of whether the proteins exhibit R or T state behavior. A definitive conclusion cannot be made whether the orientation of α3 is a result of intramolecular interactions or rather a result of the particular crystal contact effects observed in proteins modeled into P and P space groups. 21

28 Loop#7# Acetate# α3# R97# α4# Fig. 8. The orientation of R97 at the allosteric site is conserved between NRasQ61L-GppNHp (cyan) and NRasQ61R-GppNHp (magenta). The R97 side chain in each mutant points into a hydrophobic pocket formed between α3, α4, and Loop 7 residues. In the R state conformation (3K8Y, green) the R97 side chain interacts with the acetate ion (3.1 Å) placed in the allosteric site, thus the guanidinium group of the R97 side chain points outward towards the protein surface and is exposed to the bulk solvent. While α3 in NRasQ61L-GppNHp and NRasQ61R-GppNHp structures are not fully shifted towards α4 of the allosteric lobe, as seen in HRas-GppNHp R state conformation (3K8Y), the two helices are still tethered through a conserved water-mediated H-bonding network linking α3, L7, α4, and α5. The interaction between R97 and the acetate ion is essential for a large shift of α3 as seen in the 3K8Y structure. Due to the lack of ligand binding the R97 side chain of each mutant structure takes on a different conformation than seen in the 3K8Y structure. As depicted in Figure 8, the charged R97 guanidinium group in mutant structures points into the hydrophobic core between α3, α4, and Loop 7 facilitating a communication network linking α3 residues with residues of the C-terminal end of the protein. The H-bonding network illustrated in Figure 9 involves residues R97, E98, K101, D108, Y137, and Y166 as well as critically placed water molecules that are conserved between Q61L and Q61R mutants. Detailed bonding measurements are presented in Tables 4 and 5 for the Q61L and Q61R structures, respectively. This extensive 22

29 Y166$ D108$ Loop$7$ K101$ α5$ R97$ E98$ N94$ Fig. 9. Conserved H-bonding network at the allosteric site of NRasQ61L-GppNHp (green) and NRasQ61R-GppNHp (cyan) mutant structures. communication network heavily influences the positioning of α3 while not in the presence of calcium acetate. The minor shift of α3 in each structure is further stabilized upon an interaction between the N94 hydroxyl group of α3 and Y137 amine group of α4 (2.8 Å for each protein). A similar interaction involving H94 and Y137 is conserved in R state HRas-GppNHp [21]. In each mutant structure the shift of α3 away from the effector lobe through direct interactions formed with α4 of the allosteric lobe creates additional space between α3 and the switch II region compared to the T state conformation of Ras. Communication network between α3 and switch II residues: The α3/l7 shift of the allosteric switch mechanism is critical to create space for the formation of a H-bonding network stretching between α3 and switch II residues in the catalytic conformation of Ras. This previously investigated bonding network acts as the second network that ultimately links the allosteric and active sites of the protein, and coordinates the pre-catalytic Q61 side chain in position for hydrolysis [21]. The R state HRas-GppNHp α3 and switch II bonding network is presented in Figure 10 below. The 2-Å shift of α3 in NRasQ61L-GppNHp Y137$ α3$ 23

30 D69$ Q99$ R68$ W391$ Y71$ Y96$ W384$ W373$ W372$ S65$ Y64$ E63$ E62$ W176$ Q61$ GppNHp$ Fig. 10. Water-mediated network linking α3 and switch II residues investigated in the R state HRas-GppNHp structure (PDB code: 3K8Y) (21). The guanidinium group of R68 is the central facilitator of many H-bond interactions, and the multiple water molecules lead to an intimate bonding network linking the effector and allosteric lobes. creates room for a communication network that stretches the length of switch II, albeit a different, looser network than observed in Figure 10. The water-mediated H-bonding network in the Q61L mutant includes residues G60, L61, R68, D69, Y71, Y96, and Q99 that form a looser linkage between the C-terminal of α3 and the active site. Specific bond length information is listed in Table 6 that correspond to interactions detailed in Figure 11. Interestingly, another loose bonding network is observed between α3 and switch II in the NRasQ61R-GppNHp structure. The communication network observed in the Q61R mutant structure involves interactions between residues T58, G60, R68 D69, Y96, and Q99. Specific bond lengths can be found in Table 7 corresponding to interactions depicted in Figure 12 for the Q61R mutant. 24

31 D69$ Y71$ Q99$ R68$ Y64$ Y96$ W113$ W112$ E63$ W9$ L61$ W24$ G60$ GppNHp$ Fig. 11. Water-mediated H-bonding network linking α3 and switch II residues in NRasQ61LGppNHp mutant structure. Structures are colored as above. Y71$ D69$ Q99$ R68$ Y64$ Y96$ T58$ W118$ E63$ W9$ G60$ W113$ R61$ GppNHp$ Fig. 12. Water-mediated H-bonding network linking α3 and switch II residues in NRasQ61RGppNHp mutant structure. Y71 side chain is flipped away from the hydrophobic core likely leading to a looser bonding network between the R68 side chain and N-terminal switch II residues. 25

32 While a similar H-bonding network is observed, a contrasting aspect of the Q61L and Q61R mutant structures is the participation of Y71 in the hydrophobic core of the protein. Positioning of Y71 in the hydrophobic core has a stabilizing influence on the critical placement of R68 between α3 and switch II regions. In the catalytic conformation of Ras the guanidinium group of the R68 side chain is the facilitator of many H-bonding interactions as it leads to a compact interaction between Q99 of α3 and N-terminal switch II residues shown in Figure 10 (15, 21). In the Q61L structure Y71 stabilizes the positioning of R68 through van der Waals forces, likely influencing the orientation of the lengthy side chain to adopt a similar conformation observed in Figure 10 as R68 extends closer to N-terminal switch II residues and further from C-terminal α3 residues. This extension is evident by the stretched interaction between the OH group of Q99 and NH 2 of R68 (3.5 Å) in the Q61L structure. Oppositely, residue Y71 of the Q61R structure is flipped away from the hydrophobic core and towards the protein s surface as observed in HRas-GppNHp T state [14]. The orientation of Y71 potentially affects R68 and Q99 side chains in the Q61R structure as the closer functional groups are able to form a stronger interaction (3.1 Å). The closer interaction between R68 with α3 residues results in poor interaction between R68 and N-terminal switch II residues. 26

33 Discussion Previous studies have investigated the hydrolysis rates of uncomplexed HRas-GTP and mutant HRasQ61L-GTP and showed that the mutant was able to hydrolyze GTP through an intrinsic mechanism albeit at slower rates than its wild-type counterpart [13, 14]. Further experimental data has shown that binding of Raf to the HRasQ61L mutant promoted the ordering of switch II in the anticatalytic conformation associated with the T state [13]. MD simulations of uncomplexed mutant HRasQ61L-GTP and HRasQ61L-GTP/Raf-RBD were used to investigate the flexibility affects experienced by HRas when in complex [37]. Upon comparison of MD simulation results it was clear Raf-RBD binding led to a rigidified switch II region favoring the anticatalytic conformation in mutant HRas, whereas the catalytic conformation was favored in complexed wild-type HRas. Increased flexibility along switch II residues and increased calcium specificity at the allosteric site favor the catalytic R state conformation in the wild-type protein when in complex with Raf. Oncogenic mutants NRasQ61L-GppNHp and NRasQ61R-GppNHp were crystallized in their uncomplexed forms. Since hydrolysis data of HRasQ61L-GppNHp/Raf complex shows stabilization of the anticatalytic conformation and given that the structure of the mutants presented here have features of this same T state conformation, it is likely that the anticatalytic conformation is also stabilized for NRas mutants in the presence of Raf [13]. Wild-type NRas has a significantly decreased hydrolysis rate in the presence of Raf-RBD, while the affect of Raf- RBD on HRas hydrolysis is minimal [43]. This is consistent with a more prominent effect of NRasQ61 mutants on the Ras/Raf/MEK/Erk pathway, which is observed to be in an activated state in 80% of cutaneous melanomas [24, 59]. 27

34 The direct interaction between Y32 and the γ-phosphate as well as the hydrophobic active site in each mutant structure are the leading structural features promoting T state conformation. Direct interaction of the Y32 hydroxyl group and O1G of the γ-phosphate leaves no space for the bridging water molecule that is critical to the intrinsic mechanism. Without the bridging water molecule in position to donate a proton to the charged transition state Ras is unable to complete the hydrolysis reaction intrinsically. Additionally, the hydrophobic cluster that buries the precatalytic water molecule in both Q61L and Q61R mutants dramatically impairs the ability of Ras to cleave the γ-phosphate of GTP. It has been proposed in the catalytic conformation of Ras a water channel is observed linking the pre-catalytic water molecule to the bulk solvent, which functions to facilitate positioning of the pre-catalytic water as well as product release after hydrolysis [60]. This water-mediated pathway is not present in the NRas mutants due to the residue 61 side chains that extend directly over the nucleophilic water molecule forming the hydrophobic cluster shown in Figure 7. Certain interactions in the mutant structures, such as the interaction between N94 and Y137 or the water-mediated H-bonding network between α3 and switch II, are interesting as they suggested a potential R state conformation. However, these interactions are likely a result of crystal packing effects rather than intramolecular forces. The water-mediated H-bonding network between α3 and switch II is not conserved between mutants. Additionally, neither water network is conserved relative to the network in the R state HRas-GppNHp that places the Q61 side chain in position for catalysis [21]. The looser connections observed in NRasQ61L and NRasQ61R is characteristic of the anticatalytic conformation leading to the poor coordination of the residue 61 side chain, as seen in Figures 11 and 12. In the catalytic conformation of HRas-GppNHp the N- terminal extension of the α2 motif is a result of the H-bonding network between α3 and switch II. 28