Comprehensive Multidimensional NMR Studies on the Interaction Between Ras-Like Gtpases and Bisphenols

Comprehensive multidimensional NMR studies on the interaction between Ras-like GTPases and Bisphenols

Dissertation to obtain the degree Doctor Rerum Naturalium (Dr. rer. nat.) at the Faculty of Chemistry and Biochemistry of the Ruhr University Bochum

Miriam Schöpel, M. Sc. Biochemistry

July 2016

This thesis was completed from May 2013 to July 2016 at the group of biomolecular NMR at the Ruhr University Bochum under the supervision of Prof. Dr. Raphael Stoll.

Chair: 1st Examiner: Prof. Dr. Raphael Stoll 2nd Examiner: Prof. Dr. Christian Herrmann Date of Final Examination:

Ever tried. Ever failed. No matter. Try Again. Fail again. Fail better.

-Samuel Beckett-

Content

Abstract ...... 1 Scope of this thesis ...... 3 Part I: Introduction ...... 4

1. Introduction ...... 5

1.1. Guanine Nucleotide-binding Proteins (GNBPs) ...... 5 1.2. The small GTPase Ras ...... 9 1.3. The Ras protein in cancers, drugging the undruggable? ...... 12 1.4. Rap-Proteins ...... 15 1.5. Adenylation of small GTPases by Legionella Pneumophila effectors ...... 16 1.6. Bisphenols ...... 17 1.7. NMR to characterise protein-ligand interaction ...... 21

Part II: Published Papers ...... 24

2. Published Paper ...... 25 2.1. Paper I: ...... 27 Bisphenol A Binds to Ras Proteins and Competes with Guanine Nucleotide Exchange: Implications for GTPase-Selective Antagonists ...... 27 2.2. Paper II: ...... 57 The Bisphenol A Analogue Bisphenol S binds to K-Ras4B - implications for ‘BPA-free’ plastics ...... 57 2.3. Paper III ...... 65 Different Bisphenols bind to Ras isoforms and induce a pseudo-active conformation ...... 65

Part III: Summary ...... 83

3. Summary ...... 84 4. Conclusion & Outlook ...... 88

Bibliography ...... 89 Curriculum Vitae ...... 97 Acknowledgements ...... 100

Abstract

Ras proteins are key players in a series of cellular signalling pathways, such as proliferation and cell growth. The human Ras family consists of three different members, H-Ras, K-Ras (two isoforms: 4A and 4B), and N-Ras. The Ras gene is tragically famous for being mutated in about 30 % of human cancers. Ras proteins fulfil their cellular role by acting as binary molecular switches: The “on”- or “off” state is characterised by the bound nucleotide (GTP = “on”, GDP = “off”). From a NMR (nuclear magnetic resonance) spectroscopist´s point of view, a protein is a collection of resonances with different frequencies, according to their chemical environment. The popular representation of this point of view is the 2D 1H-15N TROSY- HSQC spectrum, in which every amide proton of an amino acid (besides proline) is represented by a single cross peak. In a simplified manner, the 2D 1H-15N TROSY- HSQC spectrum represents the fingerprint of the protein. Both of Ras´ activation states can be distinguished in these spectra, because of the dynamical and structural properties of each nucleotide binding state. Bisphenols (BPs) are the topic of a heated, as well scientific but also public, discussion. Chemically they share a common attribute in form of two ringed aromatic systems, which are connected by a sp3-hybridised central carbon atom. This bridging moiety can be optionally substituted, leading to a variety of different bisphenols. BPs are used in industrials processes as plasticisers and can be found in different consumer products, like plastic bottles, cashier receipts, and canned foods. This ubiquitous exposure is reflected by the fact, that these compounds are also detectable in human blood, urine, and sweat. In this thesis, the effects of small organic compounds from the class of Bisphenols on the structural and functional properties of Ras and its interaction partners were investigated. The identification of the binding pocket was carried out by the analysis of chemical shift perturbations (CSP-NMR) within a series of 2D 1H-15N TROSY-HSQC spectra. The pocket is located between the α2-helix of Switch II and the central β-sheet of the Ras protein. By plotting the amino acids CSPs versus the ligand concentration, the KD value of every residue can be derived. The extracted KD values are in milli- to micromolar range, an affinity in which also some of Ras´ protein protein interactions are found.

Moreover, several chemical shift perturbations for amino acids were observed, which are not part of the binding pocket: G13, D57, G60 (part of the nucleotide binding domain), and Y157. The chemical shift of tyrosine 157 is a valuable tracer for Ras´ toggling between “on”- and “off”-state. The CSPs of these particular showed a ligand- induced shift into the same direction as observed for the active form of the protein, which leads to the conclusion that the different BPs induce a pseudo-activation of the Ras protein. Judged from the NMR spectra, this form has to be different from the original active form, since the specific line broadening of the above mentioned amino acids (G13, D57, and G60) is not observed. In addition to this finding, a fluorescence assay was performed, that probes the SOS-mediated (son of sevenless) nucleotide exchange. The observed rates showed a significant decrease, which can be explained by two models: Firstly, a sterically interference of the ligand upon binding in the Ras- SOS interface and, secondly, a lower affinity of SOS towards Ras, because it is likely that SOS does not bind to the pseudo-activated form of Ras. In a next step, Rap1A and Ras isoforms were tested in order to investigate the selectivity of BPs for small GTPases. The isoforms H-Ras and N-Ras, showed the same binding mode as found for K-Ras4B, except for N-Ras, that exhibits a slightly lower KD. Remarkably, Rap1A showed no binding of bisphenolic compounds. A possible explanation could be that the Switch II region of this proteins is more rigid or at least is characterised by different Switch II dynamics compared to the Ras isoforms, with the result that this protein does not offer a binding pocket for BPs. With this thesis, a direct link between BPs and cancer related Ras isoforms is identified. Based on the observation that BPs modulate the structural and functional properties of these proteins further research is needed to place the results in a larger, more medical, context.

Scope of this thesis

Protein ligand interactions are essential to almost all processes occurring in living cells, as correct signal transmission is essential for the adaption to the environment. Additionally, they harbour the opportunity to characterise interaction partners on an atomic level and the possibility of therapeutically intervention. In this work, the interaction of various bisphenols, in total 14 (see table 1 in the introduction part), and different Ras proteins has been investigated. The Ras proteins tested, included the three Ras isoforms (H-, N-, K4A, and K4B), and Rap1A (Ras related protein 1). The protein Ras is directly related to various cancer types (like lung, colorectal, and pancreatic cancer), because it controls fundamental processes during cell growth. Rap1A is sequentially highly related to the Ras isoforms but also shows significant differences, including amino acid replacements. The Bisphenols investigated varied in the substitution pattern of the central carbon atom, in order to characterise their structure activity relationship (SAR) properties. Moreover, the NMR-based screening of the different bisphenols was standardised in order to obtain comparable affinity constants (KD-values). The ultimate goal was to evaluate the different bisphenols and proteins based on their interaction, in order to obtain a deeper insight in the molecular details of Ras interaction between the protein and bisphenols.

Part I: Introduction

1. Introduction

1.1. Guanine Nucleotide-binding Proteins (GNBPs)

Cells are constantly sending messages, checking nutrient levels and growth rates within the cell and with other cells. These messages need to be explicit, so that they can be heard over the busy bustle inside the crowded cytoplasm. One way to strengthen signals is to link them to a process that is chemically irreversible, like the cleavage of ATP or GTP. Adenosine triphosphate (ATP) and guanosine triphosphate (GTP) are used in living cells as cofactors for various biochemical transformation reactions. Both nucleotides share their structural motifs: the nucleobase itself (adenosine or guanosine), a ribose sugar, and three phosphates. The nucleobase is attached at the first position of the ribose, whereas the triphosphate group is attached to the fifth position.

Figure 1: Adenosine triphosphate (ATP, left) and guanosine triphosphate (GTP, right). Both nucleotides share their structural motifs: the nucleobase itself (adenosine or guanosine), a ribose sugar, and three phosphates. The nucleobase is attached at the first position of the ribose, whereas the triphosphate group is attached to the fifth position.

ATP is regarded as "molecular unit of currency" of intracellular energy generation and transfer reactions1. In contrast to this, the hydrolysis of GTP to GDP mainly plays a regulatory role in biochemical processes, including cell growth, cell differentiation, as well as vesicular and nuclear transport. Proteins capable of binding guanosine nucleotides are called guanosine binding proteins (GNBPs = guanine nucleotide-binding proteins). An eukaryotic cell contains around 100-150 GNBPs, which compete for their common nucleotide substrate2. The superfamily of monomeric small GTPases includes proteins with a size of 20-25 kDa and a common fold, consisting of a six-stranded mixed β-sheet and five α-helices on

6 both sides. The other big group within the GNBPs is formed by the heterotrimeric G- Protein, which consist of α, β, and ɣ subunits. These are activated by the G-protein coupled receptors, which are the medicinal target of approximately 40% of all drugs3. The regulatory role of the small GTPases is fulfilled through their capability to act as binary switches by toggling between an “on” and “off” state. These states are chemically characterised by the bound nucleotides: GTP (=on) and GDP (=off). Since the intrinsic GTP hydrolysis, performed by the small GTPases, is very slow (10 – 9 1/s), it is accelerated by GTPase activating proteins (GAPs) by several magnitudes4. The hydrolysis product GDP is not automatically released, because the protein has the same affinity for GTP and GDP. Therefore, the exchange process is also catalysed by a group of different proteins, which are surprisingly structurally totally unrelated: the group of guanine exchange factors (GEFs).

Figure 2: The GTPase cycle involves the exchange of GDP by Guanine Nucleotide Exchange Factors (GEFs), which control the nucleotide exchange, by increasing the dissociation rate. Once activated, binding of GTP enables the proteins to interact with various effectors. The mechanism of GEF action involves the direct insertion of certain amino acids into the nucleotide binding domain (NBD). By this action, the affinity of the GTPase towards the nucleotide is reduce. The bound GTP is the common timing mechanism to return them to the GDP-bound 'OFF'-state, thereby completing the GTPase cycle.

On an atomic level, the GEF protein firstly binds the binary GNBP-nucleotide complex and thereby forms a trimeric complex. In a second step the GEF inserts a catalytic residue in the near of/inside the P-Loop that results in a structural disorder. Thus the GNBP protein is not able to bind the nucleotide any longer and this leads to a nucleotide free, binary GNBP-GEF complex. In the final step, the protein is replenished with a nucleotide, which is normally GTP, because of its tenfold higher intracellular concentration5. GAPs also interact directly with GNBPs, as they offer a catalytic residue and/or stabilise the transformation6. Extensions at the C-terminus and the N- terminus play an important role in the interaction of GNBP with other proteins. GNBPs share four to five conserved sequence motifs (G1-G4/G5), which enable the protein to bind and release the nucleotide. These motifs are therefore in proximity to the nucleotide binding site (NBD).

Figure 3: GNBPs share 4-5 conserved sequence motifs (G1-G4/G5): The G1 motif consists of GxxxxGKS/T and is also known as the P loop (P = phosphate, because of the phosphate binding capability). The G2 motif is represented by the Switch I and is only conserved in one single amino acid, a threonine. The next motif, G3, is represented by in total four residues, with two conserved amino acids: DxxG. G4 (N/TKxD) and G5 (not totally conserved, but often the motif SAK).

The G1 motif consists of GxxxxGKS/T (where x is any amino acid) and is also known as the P loop (P = phosphate, because of the phosphate binding capability). On an atomic level, the amino group of the lysine side chain contacts the oxygens of the β- and ɣ-phosphate. In accordance with ATP-binding proteins, the P loop is also termed

8 the Walker A motif7. The G2 motif is represented by the Switch I and is only conserved in one single amino acid, a threonine. The next motif, G3, is represented by four residues, with two conserved amino acids: DxxG, which are involved in the stabilisation of the Mg2+ ion. G4 (N/TKxD) and G5 (not totally conserved, but often SAK) are both involved in binding the guanine base. The G5´s alanine directly faces the oxygen of the nucleobase carbonyl group. The communication between the G-Domain and its effector proteins is based on two regions: The Switch I and Switch II. Both regions undergo dramatic structural changes, depending on which nucleotide is bound. The underlying mechanism is known as the “loaded-spring mechanism” as introduced by Vetter et al. in 20012,8.

Figure 4: Schematic presentation of the loaded spring model. The canonical switch mechanism involves the interaction of the γ-phosphate with Thr and Gly of Switch I and II, which are released upon GTP hydrolysis. Modified from Wittinghofer et al8.

This universal mechanism involves two amino acids, a threonine (the conserved one of the Switch I) and a glycine, which are when GTP is bound in a loaded state characterised by distinct conformations of both amino acids. In the process of GTP- hydrolysis, the ɣ-phosphate is released and the residues are allowed to relax into different conformations. Interestingly, this time the amino acid side chains, are not involved, but rather the amide groups itself, which are bound to the oxygens of the ɣ- phosphate. Also, the side chain of threonine is involved in binding to Mg2+.

1.2. The small GTPase Ras

The first systematic approach for classification of the superfamily of small Ras (Rat sarcoma) related GTPases was carried out by Valencia et al. in 19919. Since then the availability of sequences and the methods to analyse phylogenetic relationship have significantly improved. Therefore, Rojas et al. presented in 2012 a updated revision of the evolution of the superfamily10. Hitherto (2016), 167 proteins have been identified, which are part of the human Ras superfamily. The biggest family is that of the Rab proteins (Ras-related in brain) with 65 members. These proteins are important for the trafficking of proteins between different cell organelles11. Followed by 39 Ras proteins, which are discussed in more detail in further sections. 30 proteins represent the family of Arf (ADP ribosylation factors) proteins, which is the most diverse and divergent family. They regulate membrane trafficking and phospholipid metabolism12.

Figure 5: The Ras superfamily of small GTPases, which consists in total of 167 members. The biggest family is those of the Rab family with 65 Rabs, these proteins are important for the trafficking of proteins between different cell organelles. Followed by 39 Ras proteins. 30 proteins represent the family of Arf proteins, which regulate membrane trafficking and phospholipid metabolism. The next family is the Rho family, which consists of 22 members. Those are involved in signalling networks related to actin regulation, cell-cycle progression, and gene expression. The family of Ran proteins is represented by just one member. The corresponding structures are shown in the same orientation, the Switch I and Switch II are indicated in blue and orange, respectively. The nucleotide is shown in cyan and the Mg2+ as a green sphere.

The next family is the Rho (Ras homologue) family, which consists of 22 members. These proteins are involved in signalling networks related to actin regulation, cell-cycle progression, and gene expression13. The family of Ran (Ras related nuclear protein) proteins is represented by just one member, although there are several copies of this gene in plants. The Ran-protein is the most highly expressed protein in human14. The Ras protein is probably the most famous small GTPase, because of its importance for cell growth, differentiation and survival. In the human proteome, four isoforms of Ras are found: N-, H-Ras, K-Ras4A and the alternatively spliced K-Ras4B. These isoforms share a high sequence similarity (>95%) and mainly differ in a region called the “hyper variable region” (HVR), which is build up by the carboxy-terminal 25 amino acids (only 8 % amino acid identity) and is the basis of the different cellular localisation and action. The final four residues of the HVR region are known as CAAX-box, which is the target of posttranslational modification. The Ras proteins go through four steps of modification: isoprenylation, proteolysis, methylation and palmitoylation.

Figure 6: Schematic presentation of the 4 Ras isoforms: H-Ras, N-Ras, K-Ras4A and K-Ras4B are highly homologous throughout the conserved G domain (amino acids 1–166). The C-terminal hypervariable domain (amino acids 166–188/189) specifies membrane localisation through posttranslational modifications that include the farnesylation of each isoform on the C-terminal CAAX motif and palmitoylation of cysteines on H-Ras, N-Ras, and K-Ras4A (highlighted in yellow). Membrane localisation of K-Ras4B is facilitated by a stretch of lysines.

Isoprenylation involves the enzyme farnesyltransferase (FTase), which transfers a farnesyl group from farnesyl pyrophosphate (FPP) to the pre-Ras protein. This step is crucial for the proper biological function15. Then, the farnesylated-proteins converting enzyme 1 (FACE-1) cleaves right after the farnesyl cysteine to release –AAX. This is followed by endoproteolytic shortening and carboxyl methylation. Both steps significantly increase the hydrophobicity of the C-terminus16. Some GTPases are also palmitoylated once or even twice, H-Ras, N-Ras and K-Ras4A at C181 and C184, respectively. K-Ras4B does not get an additional prenyl anchor, but possess a polybasic stretch (of six lysines). In order to regulate a dynamic trafficking of KRas-4B between the plasma- and endomembranes, a serine (S181) is phosphorylated by protein kinase C17.The Ras proteins controls the intracellular processes mostly through the ERK-MAPK (extracellular signal–regulated kinases, mitogen-activated protein kinase) pathway that regulates cell growth and differentiation18.

Figure 7: Receptor tyrosine kinase (RTK) mediated activation of Ras: The pathway is activated by binding of an extracellular ligand (e.g. EGF, epidermal growth factor), leading to an autophosphorylation of RTKs´ internal residues. The SH2 domain of GRB2 binds to these residues and promotes SOS membrane localisation. Thus the Ras proteins is activated and triggers a cascade of phosphorylation steps, including Raf-, MEK-, and ERK-phosphorylation.

The MAPKs are a group of protein serine/threonine kinases, which are activated by dual phosphorylation at the tripeptide motif Thr-Xaa-Tyr. The pathway is initiated by the binding of extracellular factors to heterotrimeric GPCR (G-Protein coupled receptors) or RTKs (receptor-linked tyrosine kinases), particularly the EGFR (epidermal growth factor receptor). This receptor stimulates Ras indirectly by recruiting the guanine exchange factor SOS to the cell membrane. SOS exists in a complex with the adapter protein GRB2 (growth factor receptor-bound protein-2). Upon receptor activation, the GRB2/SOS complex is translocated to the membrane by binding to autophosphorylated residues in the RTKs. Active Ras recruits and activates the protein Raf (rat fibrosarcoma), that initiates a cascade of protein phosphorylations: Phosphorylated MEK (MAPK/ERK kinase) phosphorylates ERK (extracellular signal– regulated kinases). Phosphorylated ERK proceeds from the cytoplasm into the nucleus, where it subsequently phosphorylates nuclear targets like transcription factors. The phosphorylated, and thereby activated, transcription factors turn on gene expression of target genes.

1.3. The Ras protein in cancers, drugging the undruggable?

Ras has been identified as an oncogene and is found to carry mutations in more 20%- 30 % of human cancers, mostly in pancreatic (90 %), lung (40 %), and colorectal types (50 %)19. As already mentioned, the proper Ras dependent signalling is based on the interaction with GAP and GEF proteins20. When Ras is mutated, this interaction is disturbed in such way, that the intrinsic and GAP-mediated GTPase activity is altered. This leads to a GTP-bound state (constitutively on) and an aberrant signal transduction. In contrast to resting cells, which are predominantly bound to GDP (>95%), this active cells become mainly GTP loaded (80%)21. In other words: In cancers driven by mutant Ras, the protein is locked in the active, constitutively GTP- bound state, through a defect in the switch-off mechanism. The mutations mainly occur at residue G12 (G12V and G12D) in the P-loop and the catalytic residue Q61 (Q61K)19,22. In the mid-2000s, when the farnesyltransferase-inhibitors failed to show an effect on cancer cells, the interest in Ras-based anti-cancer drugs suffered a setback23,24. At least since 2013, when the American National Cancer Institute initiated a newly designed Ras Project, this topic attracted new attention, in a way that now anti-Ras

13 drugs tackle the Ras dependent signalling network at different points 25. Even though, the FTIs failed, membrane association remain as a potential target. Today, the focus lays on two other CAAX modification enzymes: ICMT (Protein-S-isoprenylcysteine O- methyltransferase)26 and Rce127,28 (CAAX prenyl protease 2). In addition, the inhibition by ligands of the chaperone protein prenyl-binding protein phosphodiesterase, a protein modulator of Ras’ membrane trafficking has been described29,30. Furthermore the downstream signalling performed by Ras is addressed, which is particularly challenging, since Ras is involved in a variety of signal networks31. A promising idea is to block the Ras binding domain (RBD) of the Raf-kinase by a synthetic molecule, which mimics the Ras protein. In a very recent paper by Athuluri- Divakar et al., a styryl-benzyl sulfone (called ribosertib) is presented, that acts as a Ras-mimetic and interacts with the RBDs of Raf kinases, Ral-GDS, and PI3Ks. This binding leads to an inability to interact with Ras, thus disrupting the following signal pathways32. The search for direct Ras binding ligands was also successful, although, the Ras protein was believed to be undruggable since it lacks hydrophobic cavities33. In 2012 Maurer et al. and Sun et al. described two ligands, DCAI (4,6-dichloro-2-methyl-3- aminoethyl-indole)34 and indolic compound (Compound 13, (S)-N-(2-((1H-indol-3- yl)methyl)-1H-benzo[d]imidazol-5-yl)pyrrolidine-2-carboxamide)35 that bind to mutants (G12V and G12D; respectively) version of the Ras protein (see figure 8).

Figure 8: The Ras ligands: a.) Maurer et al. presented the ligand DCAI (4,6-dichloro-2-methyl-3- aminoethyl-indole). b.) Sun et al. showed that also a Bisindole (here compound 13) is capable of binding to the G12D-K-Ras4B protein.

The DCAI compound has a KD value of 1.1±0.5 mM, derived from NMR experiments and blocked the nucleotide exchange and release reactions with an IC50 of 342 ± 22 μM and 155 ± 36 μM, respectively. Compound 13 binds to K-Ras(G12D) with

an affinity of KD = 190 µm and an inhibition of the SOS-meditated nucleotide exchange of IC50 = 58±8 %. Both ligands bind at the interface of Ras and SOS and thus prevent a proper Ras SOS complex formation by a sterical clash. Interestingly, both authors mention a tyrosine (Y71, part of the Switch II), that plays an important role in that way that this residue occupies a hydrophobic pocket in the apo crystal structure. Upon ligand binding this residue has to be replaced by the ligand. In the end, this replacement process leads to the exposure of a hydrophobic cavity of approx. 7 × 7 Å at the opening and a depth of 5 Å. This molecule binding site is located between helix α2 and the core β-sheet, β1–β3. The interacting amino acids include K5, L6, V7, I55, L56, and T74. In 2013, Ostrem et al. published another ligand that bind irreversibly to a mutant version of the Ras protein (G12C)36. The synthesised compounds rely on the mutant cysteine for binding (acrylamide and sulphonamide coupling groups) and therefore do not affect the wild-type protein. The binding pocket of these compounds is different to those of Maurer and Sun et al. and is located between the central β-sheet of Ras, and the α2-(Switch II) and α3-helices. Like the other binding pocket this binding pocket does not exist in the crystal structures of GDP-bound wildtype Ras and is probably highly dynamic. In 2014, Burns et al. published the identification and characterisation of a functionally important small molecule binding site on the Ras:SOS:Ras complex37. Analogous to other interfacial inhibitors (like the Brefeldin A, which targets the Arf1:Sec7 domain complex38), this approach could be used to render Ras incapable of interacting with effector proteins. Also in 2014, Sun et al. presented another publication, in which they introduced a second ligand binding site. They attached the previously identified ligands of 2012 covalently to the first binding site (to an artificially introduced cysteine at position 39) and performed a second fragment screen. However, the second-site hits bind to both modified (S39C) and native K-Ras4B protein with affinities ranging from 0.3 to 3 mM.

In order to achieve a better KD-value, this two independently binding hits then have to be linked to each other as introduced by the SAR by NMR approach (see chapter 1.5).

1.4. Rap-Proteins

When searching for Ras-related genes in a human cDNA library, three proteins Rap1A, Rap1B, and Rap2 were identified by Pizon et al. in 198839,40. In the following year, the proteins were mapped to the chromosomes 1p12-p13, 12q14, and 13q34, respectively41. Rap1 shares around 50 % amino acid identity, 80 % amino acid homology, a nearly identical structure, and same effector binding surfaces (root mean square deviation of 0.7 Å for homologous residues) with the previously discussed Ras protein (see figure 9)42. Rap1 was identified as a Ras antagonist, because it is capable to bind several Ras effectors43–46. Later a direct function of Rap1 was described, which is the regulation of integrin-mediated cell adhesion and the inhibition of Rho-mediated cell contraction47– 50. In 1995, Herrmann and co-workers demonstrated the importance of the amino acids positioned at 30 and 31 of Ras and Rap in discriminating downstream effector partners51. In the Ras protein D30 and E31 are found, whereas in the Rap protein E30 and K31 are positioned at the protein−protein binding interface. Following studies involved the reversion mutations of Rap to Ras and emphasised, that these residues

Figure 9: Sequence Alignment of four Ras isoforms (K-Ras4B, H-Ras, and N-Ras) and Rap1A. The green box indicates the P-loop (nucleotide binding), in red and blue Switch I and Switch II (GEF-, GAP- and effector interaction) are shown. The yellow box indicates the HVR region (hypervariable region, important for membrane localisation) including the poly-lysine stretch for Rap1A and K- Ras4B as well as the CAAX-motif for every protein.

16 are involved in the interactions with downstream binding partners. The charge reversion mutation Rap K31E and the double mutation Rap E30D/K31E resulted in lowering the KD between the Rap mutant and the Raf to values similar to those of interactions between wild-type Ras and Raf52. Subsequently it was shown that these two residues play also a fundamental role in the Rap1A RalGDS-interaction (Ral guanine dissociation stimulator)42,53. In 1990, when a human platelet cDNA library was screened a second Rap-2 isoform was found and therefore called Rap-2B54. The protein is 90% identical to Rap-2A with the highest variability at the carboxyl terminus of the protein. Interestingly, both Rap2 isoforms contain a phenylalanine at position 39 rather than a serine like in Rap-1 and Ras proteins, which is supposed to guarantee specific effector interactions55,56. In 2007, a new member of the Rap superfamily was discovered and named Rap-2C, so up to now in total there are 5 family members57. Interestingly, the protein´s posttranslational lipid modification is similarly heterogeneous as in the Ras isoforms: Both Rap1 isoforms and Rap-2B are geranylgeranylated whereas Rap-2A and Rap- 2C are farnesylated. Additionally, both Rap-1 forms possess a polybasic region within their HVR, whereas the three Rap-2 isoforms are palmitoylated at two cysteine residues (located also in the HVR) and lack polybasic residues. This could explain, why the Rap-2 isoforms fail to antagonise cellular transformation caused by the Ras protein.

1.5. Adenylation of small GTPases by Legionella Pneumophila effectors

Rab proteins, that play central roles in intracellular vesicular trafficking, are AMPylated by Legionella Pneumophila. The infection with this pathogen results in Legionnaires´ disease, leading to severe, often fatal, pneumonia in humans58. Legionella produces a variety of proteins, including an effector of Rab (termed DrrA or SidM), which binds the protein, translocates it to the plasma membrane, and adenylates it59. The adenylation takes place at Tyr77 (Tyr71 in Ras proteins) that is part of the Switch II. Recently, it has been shown that this irreversible modification stabilises the active Rab1b state by locking the switch in the active signalling conformation independent of the originally nucleotide loading state60.

1.6. Bisphenols

The synthesis of Bisphenol A was published for the first time in 1891 by the Russian chemist A.P. Dianin, who described the condensation of acetone (therefore the A in Bisphenol A) with two phenols61.

Figure 10: Synthesis of Bisphenol A from phenol and acetone, as firstly discovered by A.P. Dianin in 1891.

Along the same reaction pathway, various of bisphenols were synthesised. Chemically, this family of compounds is characterised by an optionally substituted central carbon/sulphur atom, which is flanked by two hydroxyphenyl functionalities.

Table 1: Overview of the different Bisphenols tested in this thesis. The corresponding CAS number, the reactants, and the systematic name (IUPAC) are shown.

Bis- Systematic name Structural formula CAS Reactants phenol (IUPAC)

2,2-Bis(4- Phenol A 80-05-7 hydroxyphenyl) Acetone propane

Phenol 2,2-Bis(4- AF 1478-61-1 Hexafluor- hydroxyphenyl) acetone hexafluoropropane

1,1-Bis(4- Phenol AP 1571-75-1 hydroxyphenyl)-1- Acetophenone phenyl-ethane

Phenol 2,2-Bis(4- B 77-40-7 Butanone hydroxyphenyl)butane

Phenol Bis-(4-hydroxyphenyl) BP 1844-01-5 Benzophenone diphenylmethane

Phenol Bis(4-hydroxyphenyl)- C 2 14868-03-2 Dichlormethane 2,2-dichlorethylene

Phenol 1,1-Bis(4- E 2081-08-5 Acetaldehyde hydroxyphenyl)ethane

Bis(4- Phenol F 87139-40-0 hydroxyphenyl)methan Formaldehyde e

9,9-Bis(4- Phenol hydroxyphenyl)fluoren FL 3236-71-3 Fluorenon e

1,3-Bis(2-(4- M 13595-25-0 Phenol hydroxyphenyl)-2- propyl)benzene

1,4-Bis(2-(4- P 2167-51-3 Phenol hydroxyphenyl)-2- propyl)benzene

Phenol Bis(4- S 80-09-1 Sulfur trioxide hydroxyphenyl)sulfone

1,1-Bis(4- Phenol Z 843-55-0 hydroxyphenyl)- Cyclohexanone cyclohexane

In the 1930s, it was discovered that Bisphenols (BPs) can be used as synthetic estrogen by mimicking its structure62. The bisphenols were later quickly replaced by different estradiols, which showed higher impact upon hormone balance. Nowadays BPs are known endocrine disrupting chemicals (EDC), which interfere with the

19 physiological function of estrogen63. In the 1950s, it was found that BPA can be used in the synthesis of different plastics64. Since then, BPs are used in variety of plastic products as plasticising agents, like epoxy resins and polycarbonate plastics.

Figure 11: Synthesis of polycarbonate from Bisphenol A and phosgene, the reaction is catalysed by a strong acid, such as hydrochloric acid (HCl).

Bisphenol A (BPA) is one of the mostly produced chemicals worldwide with approximately 10 billion tons in 201165. BPA, which central carbon is accompanied by two methyl and two phenolic groups, is used in food can linings, thermal papers, and other daily life plastic products. The main source for human exposal to bisphenols is based on the uptake through food and drinks66. Therefore BPs are found in human serum67, urine68, adipose69, and placental tissues70. Outside the human body, significant amounts of BPs can be found in drinking and wastewater, air, as well as dust66. The second main source is thermal paper, like cashier receipts, in which the BPA is not polymerised. This makes it more available for exposure than polymerised BPA in resins or plastics71. Bisphenols in plastics can be identified by the plastic identification code, that divides the plastics into seven broad classes.

Figure 12: The plastics code used for a proper recycling: the number 1 signifies that the product is made out of polyethylene terephthalate (PETE), 2 means high-density polyethylene (HDPE), 3 is polyvinyl chloride (PVC), 4 stands for low-density polyethylene (LDPE), 5 implies polypropylene (PP),6 encodes polystyrene (PS), 7 represents other plastics, such as acrylic, nylon, polycarbonate, epoxy resins, and polylactic acid.

These codes were introduced for recycling purposes. In general, plastics that are labelled 1, 2, 3, 4, 5, and 6 are unlikely to contain BPA. Plastics of type 7 are simply all others types of plastics, also the mentioned epoxy resins and polycarbonate

20 plastics, that contain BPA. A special case is flexible PVC (type 3), which is softened by the usage of BPA72. BPA is related to a range of effects in humans and laboratory animals, like embryonic development73, asthma74, obesity, and other metabolic syndromes75. Presumably, this effects are caused by the endocrine disrupting action of BPs and the resulting misbalance in hormone status76. In fact, it has been shown in 2012, that different bisphenols bind and activate estrogen receptors (ER) α and β. Interestingly, the binding mode differs from that of 17β-estradiol77. With the help of crystallographic analysis, they were able to show that the tested BPA, Bisphenol C (BPC, and Bisphenol AF (BPAF) bind in two distinct modes. The authors presume that this may be the basis of the BPs´ different activities on the ERs. Due to its hormone-like properties and connection to several illnesses, the usage of BPA in consumer products is nowadays a topic of heated public discussion. Starting in 2008, this discussion led to investigations by several governments on BPAs safety. The European Food Safety Authority (EFSA) released in 2003, 2008, 2009, 2010, 2011, and 2015 reviews regarding this compound, each time based on new scientific input. On every occasion EFSA stated, that their experts were not able to identify any evidence, that the usage of BPA is not safe, based on the known human exposure levels to BPA78. In several studies it was observed that the BPA blood and urine concentration of infants were higher than these of grownups78,79. The BPA detoxification process involves several mechanisms in the liver, which are not totally present in infants and children, so that the BPA remains longer in the body. This is why several countries decided to ban the usage of BPA in baby bottles (for an infant the main BPA source). For example, the USA, Canada, Germany, Switzerland, the Netherlands, Denmark. In February 2016, the French government declared, that it intends to propose BPA as a REACH (Registration, Evaluation, Authorisation and Restriction of Chemicals) regulation candidate. When this proposal is successful, BPA will be registered as a substances of very high concern (SVHC) and its production and import will be limited80.

1.7. NMR to characterise protein-ligand interaction

The molecular recognition process between macromolecular receptors and molecular ligands is fundamental to various cellular processes, like the transmission of cellular signals, cell-cell communication, DNA-transcription, and DNA-translation. The term ligand refers, in the biochemical field, to a small organic compounds, which binds to a macromolecule (proteins, DNA, and RNA), which are then called receptors.

The affinity of a protein for a ligand is described by the dissociation constant KD. The binding process is kinetically expressed as an equilibrium condition, balanced by the association and dissociation. Therefore, the KD can be characterised by the dissociation (off rate) divided by association (on rate), where the off rate has to be the slower step, otherwise there would be no binding. In the context of the atomic mass law, this value is given as the product of the molar protein and ligand concentration divided by the molar concentration of the protein- ligand complex. Thermodynamically, when the binding of receptor and ligand molecules are in solution, the molar Gibbs free energy ΔG or the binding affinity is related to the dissociation constant KD by the formula shown in figure 13 c).

Figure 13: The dissociation constant (KD) expressed by the mass law, a kinetic approach, and thermodynamically: a) In the context of the atomic mass law, this value is given as the product of the molar protein and ligand concentration divided by the molar concentration of the protein ligand complex. b) By a kinetic approach the KD can be characterised by the dissociation (off rate) divided by association (on rate). c) Thermodynamically, when the binding of receptor and ligand molecules are in solution, the molar Gibbs free energy ΔG or the binding affinity is related to the dissociation constant KD.

As the binding between proteins and ligand is non-covalent, it is dominated by hydrogen bonding, electrostatic interactions, hydrophobic, and van der Waals forces. The affinities of typical biochemical interaction are in the µM range for the weakest interactions and in the pM range for the strongest. On a standard free energy level these values relate to approximately –50 kJ mole-1 for the tightest interactions to approximately –17 kJ mole-1 for the weaker binding events81. NMR (nuclear magnetic resonance) is a powerful tool, when it comes to characterise these dynamic, often weak, and predominantly transient interaction between a protein and a ligand. In brief, this technique utilises the magnetic properties of particular atomic nuclei (in the biomolecular field typical nuclei are 1H, 13C, 15N, and 19F). These nuclei, when placed in a strong magnetic field, absorb electromagnetic radiation at a nuclei- specific frequency. Based on this the chemical features, the structure and dynamic of macromolecules are accessible. The 2D 1H-15N TROSY-HSQC spectrum (heteronuclear single quantum coherence/correlation) was firstly described by Bodenhausen et al. in 198082. In biomolecular NMR the first axis (acquisition domain) is mostly the proton (1H) and the other (indirect domain) for a heteronucleus, which is usually 13C-carbon or 15N- nitrogen. The spectrum contains a cross peak for each unique proton covalently attached to the heteronucleus (besides of proline, which has no amide proton). In other words, the 2D-1H-15N-HSQC is the fingerprint of the protein, that reports the J-1H 15N coupling. In 1997, Pervushin et al. developed a specialised NMR spectrum with an improvement in line-width, which was introduced as the TROSY-HSQC (Transverse relaxation- optimised spectroscopy). This technique allows the suppression of transverse relaxation in multidimensional NMR experiments (mainly 1H-15N), which is based on constructive use of interference between DD (dipole-dipole) coupling and CSA (chemical shift anisotropy)83. This technique extended the molecular mass range that can be studied by NMR, thus also large proteins and protein complexes, like the 900 kD GroEL GroES complex and the 670 kDa 20S proteasome are now observable by NMR-spectroscopy84,85. The analysis of SAR (structure-activity relationships) by NMR-spectroscopy focuses on drug design starting from protein structure determination to characterisation and optimisation of lead compounds86. When a small ligand binds to a protein, the local magnetic environment changes for the amino acid residues involved in binding.

This can be detected via NMR as resonance changes in the 2D 1H-15N spectra. The NMR chemical shift perturbations correspond to amino acid residues of the protein involved in the interaction process, thus enabling the binding site to be mapped onto the protein surface. By screening a library of compounds, ligands can be chosen based on the changes they cause in the 2D 1H-15N NMR spectra of the protein. Usually, these ligands exhibit a low to moderate affinity with values in the millimolar to micromolar range. Two ligands, that bind independently to nearby sites can be chemically linked in order to create a new ligand, that exhibits a noticeable better affinity. This approach was applied successfully for a new BCL-XL ligand87, an XIAP-ligand.

Figure 14: The SAR by NMR (structure–activity relationship) approach86: In the first step a library of ligands gets titrated upon an 15N-enriched protein, via NMR the binding of a ligand is confirmed (ligand A). Additionally, a second screen can be used to find a second ligand which binds to another site (ligand B). These two ligands are then linked in order to obtain a better dissociation constant 103 (KD). Modified from Klages et al. .

Part II: Published Papers

2. Published Paper

The accepted manuscripts are:

1. Bisphenol A Binds to Ras Proteins and Competes with Guanine Nucleotide Exchange: Implications for GTPase-Selective Antagonists

Schoepel, M.; Jockers, K. F. G.; Dueppe, P. M.; Autzen, J.; Potheraveedu, V. N.; Ince, S.; Yip, K. T.; Heumann, R.; Herrmann, C.; Scherkenbeck, J.; Stoll, R. Bisphenol A Binds to Ras Proteins and Competes with Guanine Nucleotide Exchange: Implications for GTPase-Selective Antagonists. J. Med. Chem. 2013, 56, 9664–9672.

2. The Bisphenol A Analogue Bisphenol S binds to K-Ras4B - implications for ‘BPA-free’ plastics

Schoepel, M.; Herrmann, C.; Scherkenbeck, J.; Stoll, R. The Bisphenol A Analogue Bisphenol S Binds to K-Ras4B - Implications for “BPA-Free” Plastics. FEBS Lett. 2016, 590, 369–375.

The submitted manuscripts are:

1. Different Bisphenols bind to Ras isoforms and induce a pseudo-active conformation

Miriam Schöpel, Klaus Kock, Xueyin Zhong, Bastian Kohl, Christian Herrmann, Stefanie Löffek, Iris Helfrich, Hagen S. Bachmann, Jürgen Scherkenbeck, and Raphael Stoll

Submitted at 19th July, 2016 at Angewandte Chemie Communication (Impact factor: 11.709 in 2015)

Other publications, that are not considered in this thesis:

1. Sequence-selective molecular recognition of the C-terminal CaaX-boxes of Rheb and related Ras-proteins by synthetic receptors

Düppe, P. M.; Tran Thi Phuong, T.; Autzen, J.; Schoepel, M.; Yip, K. T.; Stoll, R.; Scherkenbeck, J. Sequence-Selective Molecular Recognition of the C-Terminal CaaX- Boxes of Rheb and Related Ras-Proteins by Synthetic Receptors. ACS Chem. Biol. 2014, 9, 1755–1763.

DOI: 10.1021/cb5002075

2.1. Paper I:

Bisphenol A Binds to Ras Proteins and Competes with Guanine Nucleotide Exchange: Implications for GTPase- Selective Antagonists

Article

pubs.acs.org/jmc

Bisphenol A Binds to Ras Proteins and Competes with Guanine Nucleotide Exchange: Implications for GTPase-Selective Antagonists Miriam Schöpel,†,§ Katharina F. G. Jockers,†,§ Peter M. Düppe,‡ Jasmin Autzen,‡ Veena N. Potheraveedu,† Semra Ince,† King Tuo Yip,† Rolf Heumann,† Christian Herrmann,† Jürgen Scherkenbeck,*,‡ and Raphael Stoll*,† †Faculty of Chemistry and Biochemistry, Ruhr University of Bochum, Universitatsstraßë 150, D-44780 Bochum, Germany ‡Faculty of Chemistry, University of Wuppertal, Gaußstraße 20, D-42119 Wuppertal, Germany

*S Supporting Information

ABSTRACT: We show for the first time that bisphenol A (10) has the capacity to interact directly with K-Ras and that Rheb weakly binds to bisphenol A (10) and 4,4′- biphenol derivatives. We have characterized these interactions at atomic resolution suggesting that these compounds sterically interfere with the Sos-mediated nucleotide exchange in H- and K-Ras. We show that 4,4′-biphenol (5) selectively inhibits Rheb signaling and induces cell death suggesting that this compound might be a novel candidate for treatment of tuberous sclerosis-mediated tumor growth. Our results propose a new mode of action for bisphenol A (10) that advocates a reduced exposure to this compound in our environment. Our data may lay the foundation for the future design of GTPase- selective antagonists with higher affinity to benefit of the treatment of cancer because K- Ras inhibition is regarded to be a promising strategy with a potential therapeutic window for targeting Sos in Ras-driven tumors.

■ INTRODUCTION proliferation and inhibition of apoptosis, including that of tumor cells. Thus, mTOR has gained much interest as a K-Ras4B and Ras homologue enriched in the brain (Rheb) are 11,15 small GTPases that belong to the Ras superfamily of guanine therapeutic target in cancer. Insulin as well as other growth nucleotide-binding proteins and are related to Rap and Ral.1 To factors stimulate the GTP loading of Rheb via the inhibition of achieve their full physiological potential, GTPases toggle the tuberous sclerosis complex (TSC) 1 and 2, a tumor suppressor protein complex that acts as a Rheb GTPase between an inactive, GDP-bound state and an active, GTP- 15,16 bound state.2 Therefore, small GTPases function as molecular activating protein (GAP). Ras-like G-proteins are not only negatively regulated by GAPs but also positively by guanine switches in living cells and are key players in intracellular 2 3 nucleotide exchange factors (GEFs). GEFs interact directly signaling. ffi Ras has been identified as an oncogene and is found to carry with G-proteins, thereby lowering the a nity of the G-protein mutations in more than 20% of human cancers.2,4−6 Several for its bound nucleotide. As a consequence, this nucleotide is released and replaced by excess bulk GTP under physiological isoforms of Ras exist, such as K-, N-, and H-Ras. Mutations of 2 K-Ras, for example, are frequently found in pancreatic, colon, conditions. The son of sevenless (Sos) protein serves as a GEF and lung carcinomas.7 Ras and in particular K-Ras have as it catalyzes the rate-limiting step of replenishing the level of attracted widespread attention in cancer drug development activated, GTP-bound K-Ras4B in the cell. Note that a bona fi fi 17 initiatives.8 de GEF for Rheb has not been identi ed yet. − The function of Rheb has been studied in a variety of The inhibition of such protein protein interactions by organisms, especially in Drosophila and mammalian cells.9 means of small molecules is regarded as a crucial strategy in These reports underscore the role of Rheb as a molecular cancer therapy; however, despite some progress in recent years, − switch in many cellular processes such as cell volume growth, the inhibition of protein protein interactions by small cell cycle progression, neuronal axon regeneration, autophagy, molecules still represents a formidable challenge in many nutritional deprivation, oxygen stress, and cellular energy cases. The reason behind this is the fact that these interactions status.10−12 The effects of Rheb are mediated via the are usually mediated by large surface regions and not by small mammalian target of rapamycin (mTOR), which exists in two areas; consequently, each residue contributes only minimally to different multiprotein complexes: the rapamycin-sensitive the overall binding free energy. Nevertheless, during the past mTORC1, which is responsible for the modulation of protein years several small molecules and zinc-chelating compounds translation, and TORC2, which mediates the spatial control of cell growth by regulating the actin cytoskeleton.13,14 In Received: August 20, 2013 mammals, mTOR activity plays an important role in Published: November 22, 2013

Chart 1. Selected Structures from Table S1, Supporting Information, Discussed in Detail

have been identified and characterized as Sos antagonists of K- ■ RESULTS AND DISCUSSION 8,18−21 Ras We have applied multidimensional heteronuclear NMR spec- Inspired by the results of Fesik and co-workers who troscopy and chemical shift perturbation analysis in order to performed an NMR screening of a large 11.000 member characterize the interactions between low molecular weight library, we designed a small fragment library based mainly on compounds with K-Ras and Rheb, respectively (Chart 1, Table the privileged structure concept and the “rule of five” [Chart 1 S1, Supporting Information, and Figures 1a,b and 3a,b). The and Table S1, Supporting Information, which lists all identified binding sites were then used as experimental compounds tested in this study and also provides information restraints in a molecular docking procedure (Figures 2, 4, and on the structure−activity relationship (SAR)].19,22−24 Further- S7, Supporting Information).31,32 Structures obtained from more, SARs of known Ras inhibitors were also taken into these docking calculations with the HADDOCK software account.8 This focused library, containing not more than 100 package were refined by OPLS_2005 force field minimizations compounds, was screened by multidimensional NMR spec- with water as solvent (Figure S5, Supporting Information). fi troscopy to identify structures that bind to Rheb, for which In the re ned docking model, the complex between Rheb small molecule inhibitors still remained unknown. The best and 4,4′-biphenol is stabilized by three hydrogen bridges Rheb-fragments were additionally examined for their binding between Ile69 and Ile78 to one 4,4′-biphenol hydroxyl group and another hydrogen bond between the second 4,4′-biphenol and selectivity for K-Ras (Table S1, Supporting Information). fi The NMR screening of Rheb revealed that almost all hydroxyl group and Lys109 in the re ned docking model, which explain the observed chemical shift perturbation. The hydrophilic compounds exhibited no significant chemical shift hydrogen bonds fix the 4,4′-biphenol (5) horizontally at the top perturbations, while in the group of the more lipophilic, polar fi of a deep lipophilic binding pocket with the two aromatic rings structures several compounds could be identi ed as ligands. conformationally twisted by approximately 30°. On the basis of Remarkably, most binding molecules contained structural the observed significant NMR chemical shift perturbations, the elements that had either a linear form, resembling a biphenyl lipophilic binding site includes the flexible switch II amino acids 2 3 scaffold or a bend, induced by an sp -orsp-hybridized bridge Tyr67, Ile69, Phe70-Ile78, and Tyr81 as well as the residues between the phenyl rings (Chart 1). In the case of Rheb, the Leu103, Met106, Val107, and Lys109 located in α-helix 3. The most significant chemical shifts were found for bisphenol A bottom of this pocket is mainly formed by residues Tyr67 and (10), 4,4′-biphenol (5), and 4,4′-dihydroxybenzophenone (9). Tyr81. As the switch II region of Rheb has been shown to Here, we also show for the first time that bisphenol A (10) can exhibit an increased flexibility on the pico- to nanosecond time directly bind to both K-Ras (Figures 1a,b, 2, S1, and S2, scale, conformational selection might play a role in ligand 36 Supporting Information) and Rheb (Figures 3a,b, 4, S3, S4, and binding to Rheb. However, the KD value extracted from S6−S8, Supporting Information), suggesting an entirely new multidimensional NMR spectroscopy of 4,4′-biphenol and mode of action for this ubiquitous compound. According to the Rheb is only approximately 1500 ± 200 μM (Figure S6, Breast Cancer Fund, bisphenol A (10) is one of the most Supporting Information). Separating the two phenol moieties of 4,4′-biphenol (5)by frequent chemicals humans are exposed to as it is a building 2 block of polycarbonate plastics and hence is present in many introducing an sp carbon, like in benzophenone, or by a quaternary sp3 carbon with limited conformational freedom, as household products, such as plastic food containers and eating found in bisphenol A (10), introduces a kink in the structure utensils. It has been suggested that bisphenol A (10) might directing the aromatic rings deeply into the binding pocket. (partly) cause cardiovascular diseases, breast and prostate 25−28 The rather low KD value extracted from multidimensional cancers, and neuronal disorders, to name but a few. NMR spectra of bisphenol A (10) and Rheb is approximately Bisphenol A (10) is regarded as an endocrine disrupting 1800 ± 500 μM (Figure S3, Supporting Information). In the chemical as it is capable of disturbing the normal activity of refined docking model, bisphenol A (10)isfixed in the 29,30 (estrogen) nuclear hormone receptors. However, neither lipophilic pocket by three hydrogen bridges, similar to 4,4′- the molecular basis of this process has been addressed nor the biphenol (5). In Rheb, Ser68 and Gln72 form hydrogen bonds complex impact of bisphenol A (10) on living cells is fully with one hydroxyl group of bisphenol A (10) and Tyr67, and understood to date. one of the residues forming the bottom of the pocket is

9665 dx.doi.org/10.1021/jm401291q | J. Med. Chem. 2013, 56, 9664−9672 Journal of Medicinal Chemistry Article

Figure 1. (a) NMR chemical shift perturbation for K-Ras bound to GDP upon titration with bisphenol A (10). The left panel shows the region of the 1H−15N HSQC spectra of K-Ras/GDP titrated with bisphenol A (10) for Leu56 and Asp57 (molar ratio of K-Ras/bisphenol A (10) ranging from 1:0 (shown in black) to 1:6 (shown in magenta)). The right panel shows a corresponding region of the 1H−15N HSQC spectra for Gly60, Thr74, and Gly75 of K-Ras/GDP titrated with bisphenol A (10) (molar ratio of K-Ras/bisphenol A ranging from 1:0 (shown in black) to 1:6 (shown in magenta)). (b) Weighted chemical shift differences of Ras/GDP titrated with bisphenol A (10). Weighted chemical shift differences plotted versus the amino acid sequence are shown at the top. On the basis of the previously published assignments and this NMR titration 39 experiment, the KD of bisphenol A (10)/K-Ras is 600 ± 200 μM as shown at the bottom. The KD values in this study have been determined by using a binding isotherm as described previously.20,35,36,46 involved in another hydrogen bond with the second hydroxyl located deep inside the pocket. While the binding situation is group. Additional contributions to the overall binding energy similar for 4,4′-dihydroxybenzophenone (9), the conformation- arise from a π-cation interaction between the side-chain amino ally highly flexible 4,4′-methylenediphenol (11) induces only group of Lys109 and the phenol group of bisphenol A (10) minor chemical shift changes in Rheb. Significantly reduced

9666 dx.doi.org/10.1021/jm401291q | J. Med. Chem. 2013, 56, 9664−9672 Journal of Medicinal Chemistry Article

Figure 2. Weighted chemical shift differences of K-Ras/GDP titrated with bisphenol A (10). On the left, a HADDOCK model of K-Ras/GDP in complex with bisphenol A (10) is shown. On the right, the observed weighted chemical shift differences are projected onto the molecular surface of K-Ras. No significant weighted chemical shift differences could be observed on the back side of K-Ras (data not shown). For the color code, please refer to Figure 1b. chemical shift perturbations were also found for the loaded K-Ras, thereby rendering this low molecular weight diphenylether (12). compound a potential inhibitor of K-Ras activation (Figure 5). Bisphenol A (10) also binds to K-Ras, however, at a different It has been shown previously that K-Ras and Rheb are site, which includes N-terminal residues of β-strand 1, as well as involved in the mechanisms of apoptosis or cellular − switch I and II residues. On the basis of the previously growth.11,36 38 Therefore, we determined dose−response published assignments and these NMR titration experiments, curves for bisphenol A (10) and 4,4′-biphenol (5) on cellular the KD of bisphenol A (10)/K-Ras is 600 ± 200 μM and is degeneration and signaling. By means of direct microscopic therefore significantly higher in comparison to that of Rheb counting, we found an increase in cellular degeneration at (Figure 1a,b).39 The binding site in K-Ras identified in this concentrations of bisphenol A (10) exceeding 100 μM (Figure study for bisphenol A (10) is identical to the one described S9, Supporting Information) indicating a toxic effect beyond 19,20 ff recently for structurally different compounds. Much to our this concentration. We then tested the e ect of various surprise, 4,4′-biphenol (5) did not show any binding to K-Ras concentrations of 4,4′-biphenol (5) on cellular degeneration and thus turns out to be a selective ligand for Rheb. On the after 4 h of treatment using the MTT assay. Degenerating cells basis of our structural analyses, it is evident that the bisphenol A were observed already at 25 μM reaching a maximum at 100 (10) binding site on K-Ras is considerably smaller than the μM (Figure 6) and decreasing thereafter until 200 μM was corresponding binding site on Rheb. Stretched, linear reached. Beyond 200 μM, cellular degeneration increased again. ′ Because of the weak but selective binding of 4,4′-biphenol molecules such as 4,4 -biphenol (5) cannot be accommodated ff in this pocket. As expected, a simulation using the Autodock (5) to Rheb but not to K-Ras, we tested the e ect on ′ software also confirmed only a very weak binding of 4,4′- intracellular signaling. Interestingly, at concentrations of 4,4 - biphenol (5) inducing maximal cell death (100 μM), a biphenol (5). pronounced decrease in S6 ribosomal protein activating In accordance with recent studies of K-Ras ligands, our phosphorylation (phospho-S6) was found (Figure 6). Thus, refined docking models show that two hydroxyl- or amino- 4,4′-biphenol (5)-induced cell death was correlated with a substituted (hetero-) aromatic rings separated by a sp2- 3 blockade of phospho-S6, which is involved in the regulation of hybridized or a sp -carbon with limited conformational freedom ff 8,18−20 protein synthesis. However, there was no detectable e ect of represent characteristic motives for K-Ras binding. bisphenol A (10) on the canonical Rheb signaling pathway Furthermore, for optimal binding the aromatic rings must be (data not shown). Although bisphenol A (10) binds to both K- able to adopt almost a right angle (Figure S5, Supporting Ras and Rheb, the data suggest that the cell death-inducing − Information). Interestingly, very similar structure activity effects by bisphenol A (10) are not due to the regulation of the relationships have been reported for the potent inhibitory canonical signaling activity of Rheb. ff 2+ 33 e ect of bisphenol A (10) on voltage-activated Ca channels. Taken together, we have designed a small fragment library in fi Bisphenol A (10) signi cantly reduces the Sos-mediated this study based mainly on the privileged structure concept and nucleotide exchange reaction of both H- and K-Ras by a factor Lipinski’s “rule of five”.23 Furthermore, structure−activity of 1.6 (Figure 5). For H-Ras, even a 2.5-fold reduction of Sos- relationships of known Ras inhibitors were also taken into mediated exchange was found (data not shown). This account. This focused library, containing not more than 100 corroborates our in silico molecular models that predict a steric compounds, was screened by multidimensional NMR spec- clash between bisphenol A (10) and Sos, thereby preventing troscopy to identify structures that bind to Rheb, for which complex formation (Figure 5). This is in accordance with small molecule inhibitors still remained unknown. The best recent studies in which a comparable stoichiometric excess of Rheb ligands were additionally examined for their binding and − small molecular compounds were used.19 21,34,35 In addition, selectivity for K-Ras. The NMR screening of Rheb revealed that the affinity of bisphenol A (10) is in the micromolar range and almost all hydrophilic compounds exhibited no significant therefore comparable to the affinity between Ras bound to NMR chemical shift perturbations, while in the group of the GDP and its GEF, such as Sos.21,34,35 Obviously, bisphenol A more lipophilic, polar structures several compounds could be (10) might well compete with Sos for the binding of GDP- identified as ligands. Remarkably, most binding molecules

9667 dx.doi.org/10.1021/jm401291q | J. Med. Chem. 2013, 56, 9664−9672 Journal of Medicinal Chemistry Article

Figure 3. (a) NMR chemical shift perturbation of Rheb/GDP titrated with bisphenol A (10). The left panel shows the region of the 1H−15N HSQC spectra for Ile78 of Rheb/GDP titrated with bisphenol A (10) (molar ratio of Rheb/bisphenol A (10) ranging from 1:0 (shown in black) to 1:15 (shown in magenta)). The right panel shows the region of the 1H−15N HSQC spectra for Phe70 of Rheb/GDP titrated with bisphenol A (10) (molar ratio of Rheb/BPA ranging from 1:0 (shown in black) to 1:15 (shown in magenta)). (b) Weighted chemical shift differences of Rheb/GDP titrated with bisphenol A (10). The weighted chemical shift differences of Rheb/GDP titrated with bisphenol A (10) plotted versus the ligand concentration is shown in the top panel. On the basis of this NMR titration experiment, the KD of bisphenol A (10)/Rheb is approximately 1830 ± 470 μM as shown in the bottom panel. The assignments of Rheb/GDP have been previously published.46 contained structural elements that had either a linear form, multidimensional NMR data, we were able to map the binding resembling a biphenyl scaffold or a bend, induced by an sp2-or site of these ligands and to generate structural models of both sp3-hybridized bridge between the phenyl rings. In the case of Rheb and K-Ras in complex with these ligands showing for the Rheb, the most significant chemical shifts were found for first time that bisphenol A (10) and 4,4′-biphenol (5) bisphenol A (10) and 4,4′-biphenol (5). On the basis of the derivatives have indeed the capacity to interact directly with

9668 dx.doi.org/10.1021/jm401291q | J. Med. Chem. 2013, 56, 9664−9672 Journal of Medicinal Chemistry Article

Figure 4. Weighted chemical shift perturbation of Rheb bound to GDP upon titration with bisphenol A (10). On the left, a HADDOCK model of Rheb/GDP bound to bisphenol A (10) is shown. On the right, the observed weighted chemical shift diﬀerences are projected onto the molecular surface of Rheb. For the color code, refer to Figure 3b.

Figure 6. Measurement of cell death in response to various concentrations of 4,4′-biphenol (5). HeLa cells were treated with increasing concentrations of 4,4′-biphenol (5) for 4 h as indicated. Results are reported as the mean value ± SEM (standard error mean) from triplicates. Statistically signiﬁcant diﬀerences are indicated as * p < 0.1 and ***p ≤ 0.05 using Student’s t test. The insert shows that 4,4′-biphenol (5) causes a decrease of phospho-S6 protein levels at 100 μM.

remarkably different behavior of bisphenol A (10) and 4,4′- biphenol (5) might help to design future experiments to decipher the signaling cascade of Rheb. As mentioned before, a GEF has not been characterized for Rheb on a molecular level.17 However, it could be shown previously that Rheb can enhance apoptosis via ASK-1 indicating that the signaling cascade of Rheb could differ from other small GTPases.36 Interestingly, we show here that 4,4′-biphenol (5) selectively inhibits Rheb signaling and induces cell death suggesting that this compound might be a novel candidate for treatment of tuberous sclerosis-mediated tumor growth. ■ CONCLUSIONS Here, we show for the first time that bisphenol A (10) can ff Figure 5. Sos-mediated nucleotide exchange assay. E ect of bisphenol directly bind to both K-Ras and Rheb with an affinity A(10) on Sos-mediated K-Ras activation (top panel). (Bottom panel) comparable to that of known low molecular compounds that Model of K-Ras[green]/Sos[blue] (PDB code 1BKD) superimposed with bisphenol A (10)[red]. interact with K-Ras. We could also establish that these compounds sterically interfere with the Sos-mediated nucleotide exchange in H- and K-Ras. This suggests an entirely new K-Ras and/or Rheb, respectively. In addition, we have mode of action of bisphenol A (10), a ubiquitous compound in characterized these interactions at atomic resolution and our civilization. According to the Breast Cancer Fund, established that these compounds sterically interfere with the bisphenol A (10) is one of the most frequent chemicals Sos-mediated nucleotide exchange in H- and K-Ras. The humans are exposed to as it is a building block of polycarbonate

9669 dx.doi.org/10.1021/jm401291q | J. Med. Chem. 2013, 56, 9664−9672 Journal of Medicinal Chemistry Article plastics and hence is present in many household products, such NMR Spectral Analysis and Molecular Docking Studies. 42 as plastic food containers and eating utensils. It has been Assignment and data handling were performed using NMRView and CcpNmr Analysis.43 Docking was performed using HADDOCK 2.131 suggested that bisphenol A (10) might cause cardiovascular 44 fi diseases, breast and prostate cancers, and neuronal disorders, to and CNS 1.3. The topology and parameter les for the low molecular weight compounds were generated using the PRODRG name but a few. In contrast to most of the small GTPase family server (www.http://davapc1.bioch.dundee.ac.uk/prodrg/). Docking members, a GEF has not been characterized for Rheb on a interfaces were defined by ambiguous interaction restraints (AIR) molecular level. However, mutations in the Rheb GTPase using the coordinate sets 1XTQ, 2L0X (ensemble of structures), and activating protein tuberous sclerosis protein 1 (TSC1, 4DSO. For K-Ras, Leu6, Val9, Leu56, Thr74, and Gly75 were selected harmatin) result in enhanced Rheb activity generating ectopic as active residues. In the case of Rheb, residues Ile69-Gly80 and neural progenitor cells. Using a MTT-based and an S6-Kinase Met106-Val107 were set to be active in the docking protocol. A total phosphorylation assay, we describe here a selective binding of of 1000 structures were generated using a rigid body docking 4,4′-biphenol (5) to Rheb thus delivering a novel target which procedure in HADDOCK. The 200 best scoring structures thereof were subjected to semiflexible simulated annealing, followed by a directly inhibits the TSC1 downstream signaling without fi ff re nement in explicit water. Finally, structures with lowest a ecting Ras signaling. On the one hand, our results do HADDOCK scores were selected for further analysis. Structure suggest a new mode of action for bisphenol A (10), which, in visualization and superposition based on RMSD values for Cα, C, and turn, advocates a reduced exposure to this compound in our N atoms were performed by PyMol.47 environment. On the other hand, our data might also lay the Sos-Mediated Nucleotide Exchange Assay. The Sos-mediated nucleotide exchange assay was performed as previously published.35 foundation for future design of GTPase-selective antagonists fl fl with higher affinity to benefit the treatment of cancer because Brie y, Ras bound to uorescent mant-GDP was incubated with a 100- fold molar excess of GDP in the presence and in the absence of Sos. K-Ras inhibition is regarded to be a promising strategy with a Inhibitory compounds were added at various concentrations. The potential therapeutic window for targeting Sos in Ras-driven fl 19,20 uorescence was excited at 360 nm and detected at 450 nm, and its tumors. Finally, we believe that our work represents an time course was recorded with a Perkin-Elmer LS50B instrument. important contribution not only to the biological activity of Determination of Cell Numbers. Total number of cells were bisphenol A (10) and phenol derivatives but also opens an counted by DAPI (4′,6-diamidino-2-phenylindole) (Sigma) staining, avenue to the design of GTPase-selective antagonists for new and the fraction of degenerating cells was determined by counting cells 6 therapies in the treatment of cancer and brain disease. with nuclear fragmentation. 1.5 × 10 cells per well were plated in 6 well plates and incubated at 37 °C/10% CO2 for 24 h. Cells were treated with various dilutions of bisphenol A (10) dissolved in DMSO ■ EXPERIMENTAL SECTION ° and incubated for 24 h at 37 C/10% CO2. As a control, cells were also Recombinant Proteins and Their Purification. Isotopically treated with a maximal concentration of DMSO (0.1%), yielding 9.7 ± enriched full-length proteins were expressed and purified as previously 2% of cell death considered here as background (Figure S9, 36,39 Supporting Information). Results are reported as the mean value ± published. Western blot analysis was also performed essentially as fi described previously.40 UV/visible spectra were recorded with an SEM (standard error of the mean) from triplicates. The signi cance of difference was assessed by one-way Student’s t test with p ≤ 0.05 Analytik Jena SPECORD 200 spectrometer. fi Low Molecular Weight Compounds and Solvents. Com- considered as statistically signi cant. pounds tested in this study were purchased with ≥95% purity, as MTT Assay. The MTT assay was used to determine the viability of ′ × 4 judged by GC and/or LC, from Alfa Aesar, ABCR, ACROS Organics, HeLa cells in the presence of 4,4 -biphenol (5). Cells (6 10 ) were ° AppliChem, Dr. Ehrensdorfer GmbH, Merck, Fluka, MAYBRIDGE, plated in each 96 wells and incubated for 20 h at 37 C/10% CO2. ′ Sigma-Aldrich, and TCI Europe. Deuterated solvents for NMR Cells were then treated with various dilutions of 4,4 -biphenol (5) dissolved in DMSO. Cells were also treated with the respective measurements were obtained from Deutero GmbH. concentrations of DMSO without 4,4′-biphenol (5), and background Multidimensional NMR Spectroscopy. All spectra were effects of DMSO on cell viability were subtracted. Plates were recorded at 298 K on a Bruker DRX600 spectrometer equipped incubated with 4,4′-biphenol (5) for 4 h at 37 °C/10% CO . A 5 mg/ with pulsed field gradients and a triple resonance probe head. The 2 mL stock of MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazo- NMR spectra with Watergate solvent suppression were recorded at lium bromide, Sigma) was diluted 1:10 into prewarmed medium. 600.13 MHz proton frequency and at 298 K. The 1D 1H NMR spectra Culture media were then replaced with diluted MTT solution. Plates were recorded with a time domain of 32 k data points and a spectral were then incubated for another 3 h at 37 °C/10% CO2. After width of 12019.23 Hz. The sweep width of the two-dimensional incubation, supernatants were removed, and 100 μL of 100% DMSO homonuclear spectra was 12019.23 Hz in the direct 1H and 2736.70 15 was added. Plates were then placed on an orbital shaker for 10 min, Hz in the indirect N dimension. The free-induction decay was and the absorbance was recorded at 570 and 620 nm. The percentage acquired for 340.9 ms, and the dwell time was set to 41.6 μs. The of degenerating cells in presence of 4,4′-biphenol (5) was expressed as binding studies were performed in accordance with literature 36 41 absorbance relative to that of the control. Results are reported as the procedures. All spectra were processed with NMRPipe and ± fi ff 42 43 mean value SEM from triplicates. The signi cance of di erence was analyzed with NMRView or CcpNmr Analysis. assessed by Student’s t test with p ≤ 0.05 considered as statistically Protein Binding Studies. NMR titrations using putative ligands significant. of Rheb and K-Ras4B consisted of monitoring changes in chemical S6-Kinase Phosphorylation Assay. Cells (1.5 × 106) were shifts and line widths of the backbone amide resonances of uniformly seeded per well in 6-well plates and incubated for 24 h at 37 °C/10% 15N enriched Rheb or K-Ras4B samples as a function of ligand CO2. Cells were serum starved for 24 h and then restimulated with concentration. This resulted in a series of 1H−15N HSQC NMR serum alone or serum supplemented with bisphenol A (10) or 4,4′- 22 spectra, following the procedure of SAR by NMR. Quantitative biphenol (5) (Figure 6). Cells were then incubated for 4 h at 37 °C/ analysis of ligand-induced chemical shift perturbation was performed 10% CO2. Untreated cells were taken as a control. Cells were lysed, by applying Pythagoras’ equation to the weighted chemical shifts as and the protein content was measured using the Bradford assay45 to previously published.36,39 Weighted chemical shift perturbations ensure equal loading. Western blot analysis was also performed regarded to be significant are shown in color in Figures 1a and b, 3a essentially as described previously.40 Antibodies used in this study and b, as well as S6, Supporting Information. The assignments of K- comprised antiphospho S6-protein (Cell Signaling Technology) and 39 Ras amide resonances were obtained from (BMRB 18529). KD anti-tubulin (Sigma). Proteins were visualized by using enhanced values were derived as previously described.19,22,35,36,46 chemiluminescence reagents (Thermo Scientific).

9670 dx.doi.org/10.1021/jm401291q | J. Med. Chem. 2013, 56, 9664−9672 Journal of Medicinal Chemistry Article ■ ASSOCIATED CONTENT (9) Urano, J.; Tabancay, A. P.; Yang, W.; Tamanoi, F. The Saccharomyces cerevisiae Rheb G-protein is involved in regulating *S Supporting Information 1 −15 canavanine resistance and arginine uptake. J. Biol. Chem. 2000, 275, List of compounds used in this study; 600 MHz H N 11198−11206. HSQC spectra of K-Ras/GDP titrated with bisphenol A; (10) Manning, B. D.; Cantley, L. C. Rheb fills a GAP between TSC electrostatic surface potential K-Ras/GDP in complex with and TOR. Trends Biochem. Sci. 2003, 28, 573−576. 1 15 bisphenol A (10); 600 MHz H− N HSQC spectra of Rheb/ (11) Aspuria, P.-J.; Tamanoi, F. The Rheb family of GTP-binding GDP titrated with bisphenol A (10); electrostatic surface proteins. Cell. Signalling 2004, 16, 1105−1112. potential Rheb/GDP in complex with bisphenol A (10); (12) Park, K. K.; Liu, K.; Hu, Y.; Smith, P. D.; Wang, C.; Cai, B.; Xu, superposition of K-Ras/GDP in complex with bisphenol A B.; Connolly, L.; Kramvis, I.; Sahin, M.; He, Z. Promoting axon (10); 600 MHz 1H−15N HSQC spectra of Rheb/GDP titrated regeneration in the adult CNS by modulation of the PTEN/mTOR with 4,4′-biphenol (5); weighted chemical shift perturbation of pathway. Science 2008, 322, 963−966. Rheb bound to GDP upon titration with 4,4′-biphenol (5); (13) Garami, A.; Zwartkruis, F. J. T.; Nobukuni, T.; Joaquin, M.; electrostatic surface potential Rheb/GDP in complex with 4,4′- Roccio, M.; Stocker, H.; Kozma, S. C.; Hafen, E.; Bos, J. L.; Thomas, biphenol (5); and measurement of degenerating cells in G. Insulin activation of Rheb, a mediator of mTOR/S6K/4E-BP response to various concentrations of bisphenol A (10). This signaling, is inhibited by TSC1 and 2. Mol. Cell 2003, 11, 1457−1466. material is available free of charge via the Internet at http:// (14) Hall, M. N. mTOR-what does it do? Transplant. Proc. 2008, 40, − pubs.acs.org. S5 S8. (15) Inoki, K.; Li, Y.; Xu, T.; Guan, K.-L. Rheb GTPase is a direct target of TSC2 GAP activity and regulates mTOR signaling. Genes Dev. ■ AUTHOR INFORMATION 2003, 17, 1829−1834. Corresponding Authors (16) Li, Y. TSC2: filling the GAP in the mTOR signaling pathway. *(J.S.) Phone: +49 202 439 2654. Fax: +49 202 439 3464. E- Trends Biochem. Sci. 2004, 29,32−38. mail: [email protected]. (17) Rehmann, H.; Brüning, M.; Berghaus, C.; Schwarten, M.; *(R.S.) Phone: +49 234 32 25466. Fax: +49 234 32 05466. E- Köhler, K.; Stocker, H.; Stoll, R.; Zwartkruis, F. J.; Wittinghofer, A. mail: [email protected]. Biochemical characterisation of TCTP questions its function as a guanine nucleotide exchange factor for Rheb. FEBS Lett. 2008, 582, Author Contributions § 3005−3010. M.S. and K.F.G.J. equally contributed to this work. (18) Rosnizeck, I. C.; Spoerner, M.; Harsch, T.; Kreitner, S.; Notes Filchtinski, D.; Herrmann, C.; Engel, D.; König, B.; Kalbitzer, H. R. The authors declare no competing ﬁnancial interest. Metal-bis(2-picolyl)amine complexes as state 1(T) inhibitors of activated Ras protein. Angew. Chem., Int. Ed. 2012, 1,1−6. ■ ACKNOWLEDGMENTS (19) Sun, Q.; Burke, J. P.; Phan, J.; Burns, M. C.; Olejniczak, E. T.; Waterson, A. G.; Lee, T.; Rossanese, O. W.; Fesik, S. W. Discovery of We are grateful to the Deutsche Krebshilfe (109776 and ﬁ small molecules that bind to K-Ras and inhibit Sos-mediated 109777) and the DFG (SFB 642) for generous nancial activation. Angew. Chem., Int. Ed. 2012, 51, 6140−6143. support. (20) Maurer, T.; Garrenton, L. S.; Oh, A.; Pitts, K.; Anderson, D. J.; Skelton, N. J.; Fauber, B. P.; Pan, B.; Malek, S.; Stokoe, D.; Ludlam, M. ■ ABBREVIATIONS USED J. C.; Bowman, K. K.; Wu, J.; Giannetti, A. M.; Starovasnik, M. A; GAP, GTPase activating protein; GEF, guanine nucleotide Mellman, I.; Jackson, P. K.; Rudolph, J.; Wang, W.; Fang, G. Small- exchange factor; HSQC, heteronuclear single quantum molecule ligands bind to a distinct pocket in Ras and inhibit SOS- coherence; NMR, nuclear magnetic resonance; SAR, struc- mediated nucleotide exchange activity. Proc. Natl. Acad. Sci. U.S.A. ture−activity-relationship; Sos, son of sevenless 2012, 109, 5299−5304. (21) Palmioli, A.; Sacco, E.; Abraham, S.; Thomas, C. J.; Di Domizio, ■ REFERENCES A.; De Gioia, L.; Gaponenko, V.; Vanoni, M.; Peri, F. First experimental identification of Ras-inhibitor binding interface using a (1) Reuther, G. W.; Der, C. J. The Ras branch of small GTPases: Ras water-soluble Ras ligand. Bioorg. Med. Chem. Lett. 2009, 19, 4217− ’ family members don t fall far from the tree. Curr. Opin. Cell Biol. 2000, 4222. 12, 157−165. (22) Shuker, S. B.; Hajduk, P. J.; Meadows, R. P.; Fesik, S. W. (2) Vetter, I. R.; Wittinghofer, A. The guanine nucleotide-binding Discovering high-affinity ligands for proteins: SAR by NMR. Science − switch in three dimensions. Science 2001, 294, 1299 1304. 1996, 274, 1531−1534. (3) Scheffzek, K.; Ahmadian, M. GTPase activating proteins: (23) Lipinski, C. A.; Lombardo, F.; Dominy, B. W.; Feeney, P. J. structural and functional insights 18 years after discovery. Cell. Mol. Experimental and computational approaches to estimate solubility and Sci. 2005, 62, 3014−3038. permeability in drug discovery and development settings. Adv. Drug (4) Schubbert, S.; Shannon, K.; Bollag, G. Hyperactive Ras in − developmental disorders and cancer. Nat. Rev. Cancer 2007, 7, 295− Delivery Rev. 2001, 46,3 26. 308. (24) DeSimone, R. W.; Currie, K. S.; Mitchell, S. A.; Darrow, J. W.; (5) Hanahan, D.; Weinberg, R. A. Hallmarks of cancer: the next Pippin, D. A. Privileged structures: applications in drug discovery. − generation. Cell 2011, 144, 646−674. Comb. Chem. High Throughput Screening 2004, 7, 473 494. (6) Pylayeva-Gupta, Y.; Grabocka, E.; Bar-Sagi, D. RAS oncogenes: (25) Braun, J. M.; Yolton, K.; Dietrich, K. N.; Hornung, R.; Ye, X.; weaving a tumorigenic web. Nat. Rev. Cancer 2011, 11, 761−774. Calafat, A. M.; Lanphear, B. P. Prenatal bisphenol A exposure and early (7) Slebos, R.; Hoppin, J.; Tolbert, P. K-ras and p53 in pancreatic childhood behavior. Environ. Health Perspect. 2009, 117, 1945−1952. ’ cancer: association with medical history, histopathology, and environ- (26) Braun, J. M.; Hauser, R. Bisphenol A and children s health. Curr. mental exposures in a population-based study. Cancer Epidemiol., Opin. Pediatr. 2011, 23, 233−239. Biomarkers Prev. 2000, 9, 1223−1232. (27) Lang, I. a; Galloway, T. S.; Scarlett, A.; Henley, W. E.; Depledge, (8) Wang, W.; Fang, G.; Rudolph, J. Ras inhibition via direct Ras M.; Wallace, R. B.; Melzer, D. Association of urinary bisphenol A binding-is there a path forward? Bioorg. Med. Chem. Lett. 2012, 22, concentration with medical disorders and laboratory abnormalities in 5766−5776. adults. J. Am. Med. Assoc. 2008, 300, 1303−1310.

9671 dx.doi.org/10.1021/jm401291q | J. Med. Chem. 2013, 56, 9664−9672 Journal of Medicinal Chemistry Article

(28) Sugiura-Ogasawara, M.; Ozaki, Y.; Sonta, S.; Makino, T.; Suzumori, K. Exposure to bisphenol A is associated with recurrent miscarriage. Hum. Reprod. 2005, 20, 2325−2329. (29) Wetherill, Y. B.; Akingbemi, B. T.; Kanno, J.; McLachlan, J. a.; Nadal, A.; Sonnenschein, C.; Watson, C. S.; Zoeller, R. T.; Belcher, S. M. In vitro molecular mechanisms of bisphenol A action. Reprod. Toxicol. 2007, 24, 178−198. (30) Rogers, J. A.; Metz, L.; Yong, V. W. Review: Endocrine disrupting chemicals and immune responses: a focus on bisphenol-A and its potential mechanisms. Mol. Immunol. 2013, 53, 421−430. (31) Dominguez, C.; Boelens, R.; Bonvin, A. M. J. J. HADDOCK: a protein-protein docking approach based on biochemical or biophysical information. J. Am. Chem. Soc. 2003, 125, 1731−1737. (32) De Vries, S. J.; Van Dijk, A. D. J.; Krzeminski, M.; Van Dijk, M.; Thureau, A.; Hsu, V.; Wassenaar, T.; Bonvin, A. M. J. J. HADDOCK versus HADDOCK: new features and performance of HADDOCK2.0 on the CAPRI targets. Proteins 2007, 69, 726−733. (33) Deutschmann, A.; Hans, M.; Meyer, R.; Haberlein,̈ H.; Swandulla, D. Bisphenol A inhibits voltage-activated Ca(2+) channels in vitro: mechanisms and structural requirements. Mol. Pharmacol. 2013, 83, 501−511. (34) Zhang, B.; Zhang, Y.; Shacter, E.; Zheng, Y. Mechanism of the guanine nucleotide exchange reaction of Ras GTPase - evidence for a GTP/GDP displacement model. Biochemistry 2005, 44, 2566−2576. (35) Lenzen, C.; Cool, R. H.; Prinz, H.; Kuhlmann, J.; Wittinghofer, A. Kinetic analysis by fluorescence of the interaction between Ras and the catalytic domain of the guanine nucleotide exchange factor Cdc25Mm. Biochemistry 1998, 37, 7420−7430. (36) Karassek, S.; Berghaus, C.; Schwarten, M.; Goemans, C. G.; Ohse, N.; Kock, G.; Jockers, K.; Neumann, S.; Gottfried, S.; Herrmann, C.; Heumann, R.; Stoll, R. Ras homolog enriched in brain (Rheb) enhances apoptotic signaling. J. Biol. Chem. 2010, 285, 33979−33991. (37) Rodenhuis, S.; SLEBOS, J. The ras oncogenes in human lung cancer 1−3. Am. Rev. Respir. Dis. 1990, 142, S27−S30. (38) Ehrkamp, A.; Herrmann, C.; Stoll, R.; Heumann, R. Ras and Rheb signaling in survival and cell death. Cancers 2013, 5, 639−661. (39) Vo, U.; Embrey, K. J.; Breeze, A. L.; Golovanov, A. P. 1H, 13C and 15N resonance assignment for the human K-Ras at physiological pH. Biomol. NMR Assignments 2013, 7, 215−219. (40) Chakrabarty, K.; Serchov, T.; Mann, S. A.; Dietzel, I. D.; Heumann, R. Enhancement of dopaminergic properties and protection mediated by neuronal activation of Ras in mouse ventral mesencephalic neurones. Eur. J. Neurosci. 2007, 25, 1971−1981. (41) Delaglio, F.; Grzesiek, S.; Vuister, G. W.; Zhu, G.; Pfeifer, J.; Bax, A. D. A multidimensional spectral processing system based on pipes. J. Biomol. NMR 1995, 6, 277−293. (42) Johnson, B. A.; Blevins, R. A. NMRView, a computer program for the visualization and analysis of NMR data. J. Biomol. NMR 1994, 4, 603−614. (43) Vranken, W. F.; Boucher, W.; Stevens, T. J.; Fogh, R. H.; Pajon, A.; Llinas, M.; Ulrich, E. L.; Markley, J. L.; Ionides, J.; Laue, E. D. The CCPN data model for NMR spectroscopy: development of a software pipeline. Proteins 2005, 59, 687−696. (44) Simonson, T.; Warren, G. L. Overview of CNS QHTML graphical interface (CNS program). Acta Crystallogr., Sect. D 1998, 54, 905−921. (45) Bradford, M. A rapid and sensitive method for the quantitation of microgram quantities of protein using the principle of protein dye binding. Anal. Biochem. 1976, 72, 248−254. (46) Berghaus, C.; Schwarten, M. Sequence-specific 1H, 13C, and 15N backbone assignment of the GTPase rRheb in its GDP-bound form. Biomol. NMR Assignments 2007, 1,45−47. (47) Delano, W. L., The PyMol Molecular Graphics System; Delano Scientiﬁc: Palo Alto, CA, 2002.

9672 dx.doi.org/10.1021/jm401291q | J. Med. Chem. 2013, 56, 9664−9672 Supporting Information

Bisphenol A binds to Ras proteins and competes with guanine nucleotide exchange - implications for GTPase-selective antagonists

Miriam Schöpel,$,‡, Katharina F.G. Jockers,$,‡, Peter M. Düppe,§, Jasmin Autzen,§, Veena N. Potheraveedu,$, Semra Ince,$, King Tuo Yip,$, Rolf Heumann,$, Christian Herrmann,$, Jürgen Scherkenbeck,§,*, and Raphael Stoll,$,*

[$] Faculty of Chemistry and Biochemistry, Ruhr University of Bochum, Universitätsstr. 150, D- 44780 Bochum, Germany, Fax: (+) 49 234 32 05466, email: [email protected]

[§] Faculty of Chemistry, University of Wuppertal, Gaußstr. 20, D-42119 Wuppertal, Germany, Fax: (+) 49 202 439 3464, email: [email protected]

[‡] Equally contributing authors

[*] Corresponding authors

Table of contents

Table S1 List of compounds used in this study Pages S2-S10

Figures S1-S9 Additional Figures Pages S11-S20

S1 TABLE

Table S1. List of low molecular weight compounds comprising pool A to H tested in this study. The relative affinity for Rheb is given arbitrarily as (-) for no detectable significant binding affinity and as (+/++/+++) for significant binding in order to provide SAR properties. This qualitative scale describes a gradually increased chemical shift perturbation from minute (+) to strong (+++) as observed by NMR spectroscopy. Examples for the latter are shown in Figures 1a and 3a.

pool A

structure name CAS affinity

2-Naphthoxyacetic acid 120-23-0 -

L-Pipecolic acid 3105-95-1 -

5-Benzimidazolecarboxylic acid 15788-16-6 -

3-Pyridinepropionic acid 3724-19-4 -

Indole-2-carboxylic acid 1477-50-5 -

4-Phenoxybenzoic acid 2215-77-2 ++

Nicotinic acid 59-67-6 -

S2 3-(4-Hydroxyphenyl)propionic acid 501-97-3 +

Indole-6-carboxylic acid 1670-82-2 -

(4-Benzyloxy)phenylacetic acid 6547-53-1 +

pool B

structure name CAS Affinity

2-Acetylpyridine 1122-62-9 -

2-Phenylbenzimidazole 716-79-0 -

2-(2-Pyridyl)benzimidazole 1137-68-4 +

4-Phenylpyridine 939-23-1 ++

Benzohydroxamic acid 495-18-1 -

Coumarin 91-64-5 -

S3 Uracil 66-22-8 -

Purine 120-73-0 -

Thieno[3,2-b]pyridin-7-ol 107818-20-2 -

N,N′-Diphenylthiourea 102-08-9 -

pool C

structure name CAS affinity

4-Amino-1-benzylpiperidine 50541-93-0 -

4-Pyrrolidinopyridine 2456-81-7 -

1-(2-Pyrimidyl)piperazine 20980-22-7 -

Indazole 271-44-3 -

S4 Melamine 108-78-1 -

1-(3,4-Dichlorophenyl)piperazine 57260-67-0 ++

5-Aminouracil 932-52-5 -

Benzamidine 618-39-3 -

Adenine 73-24-5 -

2-Aminobenzimidazole 934-32-7 -

pool D

structure name CAS affinity

3-Benzoylpropionic acid 2051-95-8 -

5-Phenyl-2-furoic acid 52938-97-3 +

S5 2-Naphthoic acid 93-09-4 +

2,3-Diphenylpropionic acid 3333-15-1 -

Diphenylacetic acid 117-34-0 -

Benzofuran-2-carboxylic acid 496-41-3 -

Quinaldic acid 93-10-7 -

Imidazole-2-carboxylic acid 16042-25-4 -

Biphenyl-4-carboxylic acid 92-92-2 +

pool E

structure name CAS affinity

trans-Cinnamic acid 140-10-3 -

Coumarin-3-carboxylic acid 531-81-7 -

S6 3-(1-Pyrrolyl)benzoic acid 61471-45-2 -

4-Biphenylacetic acid 5728-52-9 ++

pool F

structure name CAS affinity 2-Benzofuranyl methyl ketone 1646-26-0 -

2-Hydroxybenzimidazole 615-16-7 -

4,4 -Dihydroxybiphenyl 92-88-6 +++ ′

Benzamide 55-21-0 -

1,3-Diphenylurea 102-07-8 -

4-Benzyloxybenzyl alcohol 836-43-1 +

4-Benzoylpyridine 14548-46-0 +

6-Hydroxy-1-tetralone 3470-50-6 -

S7 Carbazole 86-74-8 -

Benzophenone 119-61-9 ++

pool G

structure name CAS affinity

1-(2-Pyridyl)piperazine 34803-66-2 -

N-Phenyl-o-phenylenediamine 534-85-0 ++

3-Acetylindole 703-80-0 +

4-Phenylbenzylamine 712-76-5 ++

2-Amino-3-nitropyridine 4214-75-9 -

2-Aminopyrimidine 109-12-6 -

3,4-Diaminobenzophenone 39070-63-8 ++

5-Amino-2-methylindole 7570-49-2 -

S8 4-Phenylmorpholine 92-53-5 -

pool H

structure name CAS affinity

4-(Benzyloxy)phenol 103-16-2 +

4,4'-Dihydroxydiphenyl ether 1965-09-9 ++

Benzidine 92-87-5 -

4-Phenoxyphenol 831-82-3 ++

4-(4-Hydroxy-Phenoxy)-Benzoic 500-76-5 + acid

4,4′-Diacetylbiphenyl 787-69-9 +

4 -Hydroxy-4-biphenylcarboxylic ′ 58574-03-1 - acid

S9 Hydroquinone 123-31-9 +

Bis(4-hydroxyphenyl)methane 620-92-8 ++

4,4′-Dihydroxybenzophenone 611-99-4 ++

Bisphenol A 80-05-7 +++

S10 FIGURES

Figure S1. 600 MHz 1H-15N HSQC spectra of K-Ras/GDP titrated with Bisphenol A (10) at pH 7.8 and 298 K. The spectrum of K-Ras/GDP in the absence of Bisphenol A (10) is shown in black. The spectrum shown in red has been recorded at a molar ratio between K-Ras/GDP and Bisphenol A (10) of 1:6.

S11

Figure S2. Electrostatic surface potential K-Ras/GDP in complex with Bisphenol A (10).

S12

Figure S3. 600 MHz 1H-15N HSQC spectra of Rheb/GDP titrated with Bisphenol A (10) at pH 7.8 and 298 K. The spectrum of Rheb/GDP in the absence of Bisphenol A (10) is shown in black. The spectrum shown in red has been recorded at a molar ratio between Rheb/GPB and Bisphenol A (10) of 1:15.

S13

Figure S4. Electrostatic surface potential Rheb/GDP in complex with Bisphenol A (10).

S14

Figure S5. Superposition of K-Ras/GDP, represented as a ribbon, in complex with Bisphenol A (10) shown in cyan, DCAI shown in magenta, and (((S)-N-(2-((1-methyl-1H-indol-3-yl)-methyl)- 1H-benzo[d]imidazol-5-yl)pyrrolidine-2-carboxamide)) shown in brown.19,20

S15

S16

Figure S6. 600 MHz 1H-15N HSQC spectra of Rheb/GDP titrated with 4,4’-Biphenol (5) at pH 7.8 and 298 K. The upper panel (page S16) shows the spectrum of Rheb/GDP in the absence of any ligand is shown in black. The spectrum shown in red has been recorded at a molar ratio between Rheb/GDP and 4,4’-Biphenol of 1:15. In the middle panel (page S16), the weighted chemical shift differences of Rheb/GDP titrated with 4,4’-Biphenol (5) plotted versus the amino acid sequence are shown. The extracted KD value of 4,4’-Biphenol and Rheb is approximately 1540±230 μM (lower panel, page S17).

S17

Figure S7. Weighted chemical shift perturbation of Rheb bound to GDP upon titration with 4,4’- Biphenol (5). On the left, a HADDOCK model of Rheb/GDP bound to 4,4’-Biphenol (5) is shown. On the right, the observed weighted chemical shift differences are projected onto the molecular surface of Rheb. For the color code, please refer to Figure S6.

S18

Figure S8. Electrostatic surface potential Rheb/GDP in complex with 4,4’-Biphenol (5).

S19

Figure S9. Measurement of degenerating cells in response to various concentrations of Bisphenol A (10). HeLa cells were treated with increasing concentrations of Bisphenol A (10) for 24 hours as indicated (for details see methods). Results are reported as mean value ± SEM (Standard Error Mean) from triplicates. Statistically significant differences are indicated as *** p<0.005, **p≤0.05; using Student’s T-test. Phospho-S6 level was not changed (data not shown).

S20 57

2.2. Paper II:

The Bisphenol A Analogue Bisphenol S binds to K-Ras4B - implications for ‘BPA-free’ plastics

The Bisphenol A analogue Bisphenol S binds to K-Ras4B – implications for ‘BPA-free’ plastics Miriam Schopel€ 1, Christian Herrmann1,Jurgen€ Scherkenbeck2 and Raphael Stoll1

1 Faculty of Chemistry and Biochemistry, Ruhr University of Bochum, Germany 2 Faculty of Chemistry, University of Wuppertal, Germany

Correspondence K-Ras4B is a small GTPase that belongs to the Ras superfamily of guanine R. Stoll, Faculty of Chemistry and nucleotide-binding proteins. GTPases function as molecular switches in cells Biochemistry, Ruhr University of Bochum, and are key players in intracellular signalling. Ras has been identiﬁed as an Universitatsstr.€ 150, D-44780 Bochum, oncogene and is mutated in more than 20% of human cancers. Here, we Germany Fax: +49 234 32 05466 report that Bisphenol S binds into a binding pocket of K-Ras4B previously Tel: +49 234 32 25466 identiﬁed for various low molecular weight compounds. Our results advocate E-mail: [email protected] for more comprehensive safety studies on the toxicity of Bisphenol S, as it is frequently used for Bisphenol A-free food containers. (Received 20 October 2015, revised 21 December 2015, accepted 1 January 2016, Keywords: Bisphenol A; Bisphenol S; K-Ras4B; NMR spectroscopy; available online 3 February 2016) nucleotide exchange; plastic food containers doi:10.1002/1873-3468.12056

Edited by Christian Griesinger

Recently, we could show for the first time that Bisphe- public discussion as it is a well-known endocrine dis- nol A can directly bind to H-Ras, K-Ras4B and Rheb, rupting chemical (EDC). EDCs interfere with the suggesting an entirely new mode of action for this physiological function of oestrogen by mimicking its compound [1]. Bisphenols (BPs) are abundant in mod- structure [8]. Additionally, BPA has been connected to ern life, as they are used in a variety of plastic prod- a number of chronic diseases, including obesity, ucts as plasticising agents, such as food container, asthma and cancer [9–12]. Because of public pressure baby bottles and thermal papers, like cashier receipts and new governmental restrictions producers replaced [2–4]. Humans are mainly exposed throughout canned BPA by its chemical analogue BPS (BPS, 4,40-Sulpho- foods to BPs so that BPs are found in human blood nyldiphenol, CAS 80-09-1) [13]. BPS contains a central serum, urine and sweat [5–7]. Chemically, this family sulphonyl group and two flanking phenol groups, simi- of compounds is characterised by an optionally substi- lar to BPA. In ‘BPA-free’-labelled thermal paper like tuted central carbon/sulphur atom that is flanked by in cashier receipts boarding passes, for example, BPS two hydroxyphenyl functionalities (Fig. 1). The most is used as a developer [4]. common member of this family Bisphenol A (BPA, Although BPA has been studied more thoroughly to 4,40-(propane-2,2-diyl)diphenol, CAS 80-05-7) whose date, there is mounting evidence that BPS shows a sim- central carbon is substituted with two methyl and two ilar or at least comparable physiological effect by func- phenolic groups, is a topic in heated scientific and tioning as an EDC [13–15]. A very recent publication

Abbreviations 1D, one-dimensional; 2D, two-dimensional; BPA, Bisphenol A; BPS, Bisphenol S; BPs, bisphenols; DCAI, 4,6-dichloro-2-methyl-3-aminoethyl- indole; EDC, endocrine-disrupting chemical; GAP, GTPase-activating protein; GDI, guanine nucleotide dissociation inhibition; GDP, guanosine diphosphate; GEF, guanine nucleotide exchange factor; GTP, guanosine triphosphate; HSQC, heteronuclear single quantum coherence; KD, dissociation constant; NMR, nuclear magnetic resonance; RMSD, root mean square deviation; SD, standard deviation; Sos, son of sevenless.

FEBS Letters 590 (2016) 369–375 ª 2016 Federation of European Biochemical Societies 369 Bisphenol S binds to K-Ras4B M. Schopel€ et al.

its guanine exchange factor SOS (Son of Sevenless) with the help of ﬂuorescence spectroscopy [21].

Experimental section

Recombinant proteins and their puriﬁcation K-Ras4B was expressed and puriﬁed as previously published [1,22].

Low molecular weight compounds and solvents BPS (BPS, 4,40-Sulphonyldiphenol, CAS 80-09-1) was purchased from Sigma-Aldrich (St. Louis, MO, USA). Deuterated solvents for NMR measurements were obtained from Deutero GmbH (Kastellaun, Germany). Fig. 1. Chemical structures of Bisphenol A (upper panel) and Bisphenol S (lower panel). Multidimensional NMR spectroscopy showed that low doses of BPA and BPS induce preco- The 1D- and 2D-NMR spectra were recorded as previ- cious hypothalamic neurogenesis in embryonic zebra- ously published and so were the NMR-based binding fish, which can lead to an improper fine-tuning of the studies [1]. brain in later development [16]. K-Ras4B belongs to the family of small GTPases, Proteins binding studies a group of enzymes that hydrolyse guanosine triphosphate (GTP) to guanosine diphosphate (GDP). The assignments of K-Ras4B backbone amide reso- Through toggling between an active (GTP-loaded) nances were obtained from the BMRB database entry and an inactive (GDP-loaded) state, K-Ras4B acts as 18529 [22]. Binding of BPS to 0.4 mM K-Ras4B was a molecular switch within the cell [17]. Ras-like detected by comparing 2D sensitivity-enhanced 1H-15N G-proteins are not only negatively regulated by GAPs HSQC NMR spectra of the protein in the absence and (GTPase-activating proteins) but also positively by presence of BPS. In order to determine the binding guanine nucleotide exchange factors (GEFs) [17]. The pocket the chemical shift perturbation (∆p.p.m.) was son of sevenless (Sos) protein serves as a GEF as it plotted versus the amino acid sequence. Chemical shift catalyses the rate-limiting step of replenishing the perturbations greater than two standard deviations level of activated, GTP-bound K-Ras4B in the cell were regarded to be significant. KD values were [18]. Furthermore, K-Ras4B is a known protein onco- derived by using the standard fitting equation from the gene, as it is found to carry mutations in 20 % of ORIGIN software (‘Hyperbl’ Origin OriginLab, human cancers [19]. Both proteins’ activation states Northampton, MA, USA). and the binding of small molecules to K-Ras4B can be investigated through NMR (nuclear magnetic reso- NMR spectral analysis and molecular docking nance) spectroscopy, which exploits the magnetic studies properties of certain atomic nuclei. In addition, 2D1H-15N-NMR provides the opportunity to deter- Assignment and data handling were performed using mine the KD value (dissociation constant), a valuable CCPNMR [23]. Docking was performed using HADDOCK parameter to evaluate the impact of a ligand on a 2.1 (High Ambiguity Driven protein–protein DOCK- protein [20]. Based on our previous studies, which ing) [24] and CNS 1.2 [25] as previously published [1]. had shown that BPA binds with a KD value of Docking interfaces were defined by ambiguous interac- 600 lM to the small GTPases K-Ras4B, our objective tion restraints (AIR) using the atomic coordinate set in this study was to examine the binding of BPS, 4DSO. Val8, Ile36, Leu56, Thr74 and Gly75 were because of its comparable chemical structure, to the selected as active residues. The surrounding solvent protein oncogene K-Ras4B [1]. exposed amino acids were chosen as passive residues. Finally, we carried out nucleotide exchange experi- Structure visualisation and superimposition based on ments, which observe the interaction between Ras and RMSD values for Ca, C, and N atoms were performed

370 FEBS Letters 590 (2016) 369–375 ª 2016 Federation of European Biochemical Societies M. Schopel€ et al. Bisphenol S binds to K-Ras4B

using PYMOL (Delano, W. L., The PyMol Molecular Signiﬁcant shifts, with a perturbation bigger than Graphics System (2002) Delano Scientiﬁc, Palo Alto, once the SD, can be detected for K5, L6, V7, V8, V9, CA, USA). A11, T20, V29, Y32, D33, I36, E37, D38, K42, D54, I55, D57, T58, G60, M67, M72, L79 and Q99 (Fig. 3A). Weighted chemical shift differences twice Sos-mediated nucleotide exchange assay the SD correspond to G10, S39, L56, T74, G75 and The son of sevenless (Sos)-mediated nucleotide K101 (Fig. 3A). As small GTPases act as important exchange assay was performed as previously published molecular switches within the cell, they show a number [1,26]. of structural features, like the P-loop, the switch I and switch II region. The P-loop is responsible for proper Results binding of the guanosine nucleotide. In our study, we observed for A11 that is part of the P-loop a shift The 2D NMR protein-binding studies show that the higher than once the SD. Even more for G10, a shift BPA-replacement BPS binds to K-Ras4B. of twice the SD was detected. Additionally to the P- By comparing 2D 1H-15N-HSQC in the presence Loop, small GTPases harbour two switch regions, and the absence of BPS, we have investigated the termed switch I and switch II. For the switch I region, chemical shift perturbation induced by BPS. Overall, several shifts were observed including amino acids numerous shifts were detected (Fig. 2). The values of V29, Y32, D33 and I36-D38 with a shift of once the onefold and twofold of the standard deviation (SD) SD and S39 with a shift of twice the SD. At switch II, were subtracted from the measured chemical shift per- D57, T48, G60, M67 and M72 exhibit a shift of once turbation. higher the SD and twice the SD, respectively. Overall,

Fig. 2. Overall view of the NMR chemical shift perturbation at 600 MHz and at 298 K observed for K-Ras4B bound to GDP upon titration with Bisphenol S, ranging from the black reference to a ratio of 1 : 25, shown in magenta. In the upper left corner, the panel shows the region corresponding to the resonance of S39 with a signiﬁcant shift and I84, C118 and G151 without any shift as an internal control. The upper right corner illustrates residues G60, T74 and G75. The last two exhibit the largest shift of all amino acids, in concordance with being in closest proximity to the ligand. I36 is shown in the lower right corner, with a high ﬁeld shift upon ligand binding. Residues A130, E152 and L159, show no shift at all. The last panel in the lower left represents the shifts of V8, L56 and D57. Residues with an asterisks are used for docking.

FEBS Letters 590 (2016) 369–375 ª 2016 Federation of European Biochemical Societies 371 Bisphenol S binds to K-Ras4B M. Schopel€ et al.

Fig. 3. (A) Weighted chemical shift differences plotted versus the amino acid sequence. The largest chemical shift differences correspond to residues T74 and G75. Additionally, hydrophobic residues corresponding to amino acids L6-G10, I36 and S39 of the switch I region, residues D54-T58, Q99 and K101 show signiﬁcant chemical shift differences. (B) The observed weighted chemical shift differences are projected onto the molecular surface of K-Ras4B (PDB-ID 4DSO). Residues with a shift difference once or twice the SD are highlighted in orange and in red, respectively. In the lower right corner, the backside of K-Ras4B is shown, which does not experience any chemical shift perturbation at all in the presence of BPS.

T74 and G75 exhibit the largest shift, corresponding pocket was used to model the biomolecular complex to the closest proximity to BPS (Fig. 3A). of K-Ras4B and BPS. For this purpose, the docking The plotted chemical shift perturbation on the sur- software HADDOCK was used (Fig. 4). face of K-Ras4B (PDB-ID 4DSO) reveals a small It is apparent from the docking results that one molecule binding site close to helix a2 and near to the ring of the symmetrical BPS molecule penetrates into core b-sheet (b1–b3) that is the same as described for the binding pocket and remains in a twisted orienta- BPA (Fig. 3B) [1]. Hydrophobic amino acids that con- tion with respect to the second ring. The sulphonyl stitute to the bottom core of the pocket are L6 and group itself contacts at the upper part of the binding V5. This pocket is surrounded by L56, T74 and G75. pocket K5 and T74, in concordance with the titration In contrast to BPA, BPS induces additional chemical data. Q70 on the left-hand side, however, cannot be perturbation at amino acids I36, E37, D38 and S39 of detected in the 2D-1H-15N HSQC NMR spectra. D54 the switch I region. exhibits a chemical shift difference of one SD and is located on the right-hand side. The second ring is exposed to S39 on the right, to L56 on the bottom, NMR-based K -determination and molecular D and to E37 on the left side of the binding pocket. All docking studies of these residues showed a signiﬁcant chemical shift By plotting the weighted NMR chemical shift perturbation.

(p.p.m.) versus the ligand concentration the KD values can be extracted from protein ligand interaction stud- Sos-mediated nucleotide exchange assay ies. The KD value of the interaction between K- Ras4B and BPA was determined to be Bisphenol A signiﬁcantly reduces the Sos-mediated ~ 600 Æ 200 lM [1]. For BPS, we observed a much nucleotide exchange reaction by a factor of 1.6 [1]. weaker interaction with a KD of ~ 5.8 Æ 0.7 mM Much to our surprise, we were not able to detect any (Fig. 4). The value was calculated using Origin0s change in the exchange rate for Bisphenol S (BPS), implemented hyperbolic function. On the basis of although it binds as BPA to K-Ras4B, as judged 2D-1H-15N HSQC NMR spectra, we calculated the from the titration data. Also, varying the BPS con-

KD value for each amino acid and we computed centration from 1 to 3.3 mM showed no signiﬁcant mean value of six representative residue (V8, I36, effect on the Sos-mediated nucleotide exchange reac- S39, L56, T74 and G75). The identiﬁed binding tion (Fig. 5).

372 FEBS Letters 590 (2016) 369–375 ª 2016 Federation of European Biochemical Societies M. Schopel€ et al. Bisphenol S binds to K-Ras4B

Fig. 4. On the left side, the chemical shift differences (in p.p.m.) are plotted against the ligand concentration (in mM). For the binding pocket, six representative residues (V8, I36, S39, L56, T74 and G75) were chosen in order to calculate the KD value that yielded 5.78 Æ 0.69 mM. In the middle panel, a HADDOCK-based model is shown. In the very right panel, an enlarged version of the binding pocket is shown. The colour code corresponds to Fig. 3B.

nucleotide exchange in H-Ras and K-Ras4B. The with- drawal of BPA from baby bottles in the EU in 2011 and a year later in the US is the conclusion of scien- tiﬁc research, public concerns and a thereby triggered change in consumer behaviour. When searching for an alternative to BPA, many manufactures turned to its chemical analogue BPS, whose two phenol groups are

connected through a sulphone group (SO2) instead of a branched sp3-hybridised carbon atom. This work shows for the first time that BPS also binds to the small GTPase K-Ras4B. As described previously, K- Ras4B displays a small ligand binding site between switch I and switch II, which is close to helix a2 and to the core b-sheet (b1–b3) [1,27,28]. In comparison to BPA, BPS also induced chemical Fig. 5. Sos-mediated nucleotide exchange assay. There is no shift perturbation of amino acids I36 to S39 which detectable effect of BPS on the Sos-mediated K-Ras4B activation. indicates that the BPS molecule rises above this ligand- binding pocket and reaches out to amino acids I36 and Discussion S39, located further away. By comparing the size and orientation of BPS0s sulphone group with BPA0ssp3- This paper describes our effort to characterise the hybridised carbon atom, it is obvious that in BPS the interaction of BPS with the small GTPase K-Ras4B. phenolic moieties are restricted to a less flexible confor- Bisphenols, like BPA, are found in many plastics. mation with respect to each other (Fig. 6). They are mainly used as plasticisers in the inner poly- BPS0s first ring adopts the same orientation and carbonate plastic coating of food cans, because they the two oxygen atoms occupy the same space as the are strong, flexible, high-temperature tolerant and – methyl groups of BPA. BPA0s second ring orientates most importantly – cheap [3]. We have previously itself towards Q70, which leads to a bent orientation shown that BPA has the capacity directly to interact of both rings with respects to each other. In the case with both K- and H-Ras. We characterised the interac- of BPS, the first ring is twisted but the rings do not tion at atomic resolution using multidimensional adopt a tilted orientation to each other. This differ- NMR spectroscopy and established that this com- ence can be clearly observed in the docking structures pound sterically interferes with the Sos-mediated of both K-Ras4B/BPA and K-Ras4B/BPS. Taken

FEBS Letters 590 (2016) 369–375 ª 2016 Federation of European Biochemical Societies 373 Bisphenol S binds to K-Ras4B M. Schopel€ et al.

Fig. 6. Comparison of binding site of BPA and BPS based on HADDOCK structures. BPA [1] (on the right side) exhibits a higher degree of flexibility due to its sp3 hybridisation. BPS (on the left side) lacks this flexibility and remains in more rigid orientation. Residues marked with an asterisk are not detectable in the 2D 1H-15N-HSQC NMR spectra. together, the docking results suggest that the sul- replacement, we believe our results will therefore be phone group of BPS is sterically too demanding to beneficial for future toxicity studies of BPs. bind effectively to the very narrow binding pocket of K-Ras4B that is estimated to be around 7 9 7 Aat Acknowledgements the opening and 5 A deep (Fig. 6)[27]. This implies that not only the phenolic moiety but also the bridg- We are grateful to the Deutsche Krebshilfe (109776 ing atoms, such as a sp3 carbon atom or a sulphone and 109777), the DFG (SFB 642), and the RUB Plus group, located next to it, contribute to binding affin- Research School for generous financial support. ity and functional activity. On the one hand, the millimolar affinity of BPS is similar to the binding Author contributions strength of low molecular weight compounds previously identified, such as the indole derivate [28,29], M.S., C.H. and R.S. conducted experiments and anal- that did also not show an effect on the nucleotide ysed data. M.S., J.S. and R.S. designed research and exchange rate. On the other hand, compounds such wrote the manuscript. All authors reviewed the manu- as DCAI (4,6-dichloro-2-methyl-3-aminoethyl-indole), script and provided approval for submission. The BPA and indole derivates modified to a larger extent authors declare that they have no conflicts of interest, did influence the nucleotide exchange rate [1,27,28]. financial or otherwise, regarding the publication of this In addition, DCAI inhibits the activation of K-Ras4B manuscript. in HEK-T293 cells [27]. Obviously, these results suggest that the minimal KD value of those low molecu- References lar weight compounds required to interfere with the 1 Schoepel M, Jockers KFG, Dueppe PM, Autzen J, Sos-mediated exchange is ~ 700 lM [1,27,28]. Appar- Potheraveedu VN, Ince S, Yip KT, Heumann R, ently, low molecular weight compounds with a K D Herrmann C, Scherkenbeck J et al. (2013) Bisphenol A value in the micromolar range can impede nucleotide binds to Ras proteins and competes with guanine exchange because the affinity between K-Ras4B nucleotide exchange: implications for GTPase-selective (GDP) and its GEF, such as Sos, is in a similar antagonists. J Med Chem 56, 9664–9672. affinity regime [26,30,31]. BPS is a ligand of K- 2 Pivnenko K, Pedersen GA, Eriksson E and Astrup TF Ras4B, whose binding affinity is much lower than (2015) Bisphenol A and its structural analogues in that of BPA and thereby strips BPS off a significant household waste paper. Waste Manag 44,39–47. functional role, at least with regard to the Sos-medi- 3 Glausiusz BYJ (2014) The plastics puzzle. Nature 508, tated nucleotide exchange of K-Ras4B. In the light of 7–9. our previous observation that BPA binds to Ras 4 Liao C, Liu F and Kannan K (2012) Bisphenol s, a new GTPases, our new results on BPS reported here sug- bisphenol analogue, in paper products and currency gest a universal binding mode of BPs. As BPA-free bills and its association with bisphenol a residues. plastic products often contain BPS as a chemical Environ Sci Technol 46, 6515–6522.

374 FEBS Letters 590 (2016) 369–375 ª 2016 Federation of European Biochemical Societies M. Schopel€ et al. Bisphenol S binds to K-Ras4B

5 Vandenberg LN, Hauser R, Marcus M, Olea N and 19 Schubbert S, Shannon K and Bollag G (2007) Welshons WV (2007) Human exposure to bisphenol A Hyperactive Ras in developmental disorders and cancer. (BPA). Reprod Toxicol 24, 139–177. Nat Rev Cancer 7, 295–308. 6 Liao C, Liu F and Kannan K (2012) Bisphenol S, a 20 Williamson MP (2013) Using chemical shift new bisphenol analogue, in paper products and perturbation to characterise ligand binding. Prog Nucl currency bills and its association with bisphenol a Magn Reson Spectrosc 73,1–16. residues. Environ Sci Technol 46, 6515–6522. 21 Herrmann C, Martin GA and Wittinghofer A (1995) 7 Vandenberg LN, Chahoud I, Padmanabhan V, Quantitative analysis of the complex between p21ras Paumgartten FJR and Schoenfelder G (2010) and the Ras-binding domain of the human Raf-1 Biomonitoring studies should be used by regulatory protein kinase. J Biol Chem 270, 2901–2905. agencies to assess human exposure levels and safety of 22 Vo U, Embrey KJ, Breeze AL and Golovanov AP bisphenol A. Environ Health Perspect 118, 1051–1054. (2013) 1H, 13C and 15N resonance assignment for the 8 Kitamura S (2005) Comparative study of the endocrine- human K-Ras at physiological pH. Biomol NMR disrupting activity of bisphenol A and 19 related Assign 7, 215–219. compounds. Toxicol Sci 84, 249–259. 23 Vranken WF, Boucher W, Stevens TJ, Fogh RH, Pajon 9 Alonso-Magdalena P, Vieira E, Soriano S, Menes L, A, Llinas M, Ulrich EL, Markley JL, Ionides J and Burks D, Quesada I and Nadal A (2010) Bisphenol a Laue ED (2005) The CCPN data model for NMR exposure during pregnancy disrupts glucose homeostasis spectroscopy: development of a software pipeline. in mothers and adult male offspring. Environ Health Proteins 59, 687–696. Perspect 118, 1243–1250. 24 Dominguez C, Boelens R and Bonvin AMJJ (2003) 10 Midoro-Horiuti T, Tiwari R, Watson CS and HADDOCK: a protein-protein docking approach based Goldblum RM (2009) Maternal bisphenol A exposure on biochemical or biophysical information. J Am Chem promotes the development of experimental asthma in Soc 125, 1731–1737. mouse pups. Environ Health Perspect 118, 273–277. 25 Simonson T and Warren GL (1998) Overview of CNS 11 Keri RA, Ho S-M, Hunt PA, Knudsen KE, Soto AM QHTML graphical interfac @ – (CNS program). Acta and Prins GS (2007) An evaluation of evidence for the Crystallogr Sect D Biol Crystallogr 54, 905–921. carcinogenic activity of bisphenol A. Reprod Toxicol 24, 26 Lenzen C, Cool RH, Prinz H, Kuhlmann J and 240–252. Wittinghofer A (1998) Kinetic analysis by fluorescence 12 Vom Saal FS, Nagel SC, Coe BL, Angle BM and of the interaction between Ras and the catalytic domain Taylor JA (2012) The estrogenic endocrine disrupting of the guanine nucleotide exchange factor Cdc25Mm. chemical bisphenol A (BPA) and obesity. Mol Cell Biochemistry 37, 7420–7430. Endocrinol 354,74–84. 27 Maurer T, Garrenton LS, Oh A, Pitts K, Anderson DJ, 13 Glausiusz J (2014) Toxicology: the plastics puzzle. Skelton NJ, Fauber BP, Pan B, Malek S, Stokoe D (2012) Nature 508, 306–308. Small-molecule ligands bind to a distinct pocket in Ras 14 Grignard E, Lapenna S and Bremer S (2012) Weak and inhibit SOS-mediated nucleotide exchange activity. estrogenic transcriptional activities of bisphenol A and Proc Natl Acad Sci USA 109,5299–5304. bisphenol S. Toxicol In Vitro 26, 727–731. 28 Sun Q, Burke JP, Phan J, Burns MC, Olejniczak ET, 15 Hashimoto Y, Moriguchi Y, Oshima H, Kawaguchi M, Waterson AG, Lee T, Rossanese OW and Fesik SW Miyazaki K and Nakamura M (2001) Measurement of (2012) Discovery of small molecules that bind to K-Ras estrogenic activity of chemicals for the development of and inhibit Sos-mediated activation. Angew Chem Int new dental polymers. Toxicol In Vitro 15, 421–425. Ed Engl 51, 6140–6143. 16 Kinch CD, Ibhazehiebo K, Jeong J-H, Habibi HR and 29 Sun Q, Phan J, Friberg AR, Camper DV, Olejniczak Kurrasch DM (2015) Low-dose exposure to bisphenol ET and Fesik SW (2014) A method for the second-site A and replacement bisphenol S induces precocious screening of K-Ras in the presence of a covalently hypothalamic neurogenesis in embryonic zebrafish. Proc attached first-site ligand. J Biomol NMR 60,11–14. Natl Acad Sci USA 112, 201417731. 30 Palmioli A, Sacco E, Abraham S, Thomas CJ, Di 17 Vetter IR and Wittinghofer A (2001) The guanine Domizio A, De Gioia L, Gaponenko V, Vanoni M and nucleotide-binding switch in three dimensions. Science Peri F (2009) First experimental identification of Ras- 294, 1299–1304. inhibitor binding interface using a water-soluble Ras 18 Simon MA, Bowtell DDL, Dodson GS, Laverty TR ligand. Bioorg Med Chem Lett 19, 4217–4222. and Rubin GM (1991) Ras1 and a putative guanine 31 Zhang B, Zhang Y, Shacter E and Zheng Y (2005) nucleotide exchange factor perform crucial steps in Mechanism of the guanine nucleotide exchange reaction signaling by the sevenless protein tyrosine kinase. Cell of Ras GTPase–evidence for a GTP/GDP displacement 67, 701–716. model. Biochemistry 44, 2566–2576.

FEBS Letters 590 (2016) 369–375 ª 2016 Federation of European Biochemical Societies 375 65

2.3. Paper III

Different Bisphenols bind to Ras isoforms and induce a pseudo-active conformation

COMMUNICATION

Different Bisphenols bind to Ras isoforms and induce a pseudo- active conformation

Miriam Schöpel[a], Klaus Kock[a], Xueyin Zhong[a], Bastian Kohl[a], Christian Herrmann[a], Stefanie Löffek[b], Iris Helfrich[b], Hagen S. Bachmann[b], Jürgen Scherkenbeck[c], and Raphael Stoll[a]*

Abstract: The protein family of small GTPases controls cellular orders of magnitude. The G-domain is common in all GTPases processes by acting as a binary switch. This powerful function is and two regions, termed switch I and switch II, are the structural performed by toggling between an “on”-(GTP-loaded) and an “off”- basis of the GTP/GDP-transition. Probably, Ras is the best (GDP-bound) state. These states are regulated by guanine- known member of the family of small GTPases, because it is an nucleotide-exchange factors (GEFs) and GTPase-activating proteins oncogene and is mutated in 30 % of human cancers[2,3].In cells, (GAPs). The most prominent family members are the Ras-isoforms three isoforms of Ras exist: K-Ras (spliced as K-Ras4A and K- H-Ras, N-Ras, and K-Ras, which are highly related and frequently Ras4B), H-Ras, and N-Ras.∼ The first 168 amino acids are mutated in cancer. The Ras related protein (Rap-1A) shares about almost identical, whereas the hypervariable region (HVR), which 50% sequence homology with Ras and competes for Ras effectors. is responsible for locating GTPases at different cellular locations Bisphenols are widespread in modern life, because of their industrial important for effector regulation[4–6], shares only 15 % sequence application as plasticisers. BPA is definitely the most known member, similarity. Rap-1A (Ras related protein) is a homologue of Ras and gained scientific as well as public attention as an endocrine (>50%) and shares the nearly the same effector proteins and is disrupting chemical (EDC) that eventually led to a replacement of thereby able to bind also to effectors like the Raf1[7,8] Recently, BPA. However, compounds used to replace BPA still contain the new efforts were undertaken to find small antagonistic ligands molecular scaffold of bisphenols. Here we present the first that bind to the GDP-loaded K-Ras4B isoform [9–13]. It could comprehensive study on the interaction of small GTPases (Ras- previously be shown, that not only BPA but also BPS bind to isoforms, Rap-1A) with various bisphenols: BPA, BPAF, BPB, BPE, GDP-bound K-Ras4B[14,15]. Furthermore, BPA, but not BPS, BPF, and an amine-substituted BPAF-derivate. We could show that interferes with the protein-protein interaction between Ras and all of these bisphenols select all Ras-Isoforms but not Rap-1A and its GEF Son of Sevenless (Sos). In general, bisphenols are bind to a common site on these proteins, which induces their known as endocrine disrupting chemicals (EDC) that bind to the pseudo-active conformation. endocrine receptor. [16–18] Herein we report, to our knowledge, the first comprehensive study on the interaction of in total 14 Bisphenols and Ras-related proteins, such as K-Ras4B, H-Ras, Small GTPases are characterised by their ability to bind and N-Ras, and Rap-1A. hydrolase guanosine triphosphate (GTP) to guanosine In order to identify the binding pocket, all bisphenols were diphosphate (GDP). Within the cell, they act as a binary switch titrated stepwise, with each titration step monitored by NMR since they are active when GTP is bound and inactive when spectra. In a “SAR by NMR”-like approach we tested 14 different bound to GDP[1]. On the one hand, the hydrolysis rate is accelerated by proteins called GAPs (GTPase activating proteins. Guanine exchange factors (GEFs), on the other hand, are proteins that lower the affinity of Ras for GDP by several

[a] M. Schöpel, B.Sc, M.Sc, Dr. Klaus Kock, Prof. Dr. C. Herrmann, Prof. Dr. R. Stoll Faculty of Chemistry and Biochemistry Ruhr University of Bochum Universitätsstr. 150, D-44780 Bochum, Germany Fax: (+) 49 234 32 05466 E-mail: [email protected] Homepage: www.rub.de/bionmr [b] Dr. Stefanie Löffek, PD Dr. Iris Helfrich, PD Dr. Hagen S. Bachmann, M.Sc. Universitätsklinikum Essen Hufelandstr. 55 D-45147 Essen, Germany [c] Prof. Dr. J. Scherkenbeck Faculty of Mathematics and Natural Sciences University of Wuppertal Gaußstr. 20, D-42119 Wuppertal, Germany [*] Corresponding author Scheme 1. Bisphenols tested in this study, with varying bridging moieties at [] We are grateful to the DFG (SFB 642), Deutsche Krebshilfe (109776 3 and 109777) as well as the RUB Research SchoolPlus for generous the central sp -hybridised carbon atom and. AFX was used to characterise the financial support. binding of one phenolic ring to K-Ras4B.

Supporting information for this article is available on the WWW under http://www.angewandte.org or from the author.

COMMUNICATION

bisphenols, which all vary in their bridging moiety flanked by the shift of the amino acids D57 and G60 are probably triggered by two phenolic groups (Scheme 1, Suppl. Inf.). Two of the the interaction of the amide group of L56, which faces directly bisphenols were also substituted at the phenolic ring itself (BPC the ring surface of the bound bisphenol and thus forwards this and BPNH2). Due to the high protein and ligand concentration information from D57 to G60. Bisphenol E, which contains one required some of the bisphenols were too insoluble in order to proton and one methyl group, exhibited a better KD value of be tested by NMR spectroscopy, as monitored by 1H NMR. This 6.5 ± 0.7 mM and showed the same pattern as observed for BPF, affected three different bisphenols: BPAP, BPBP, and BPFL that with higher magnitude of chemical shift perturbation. As already have all at least one additional ring. shown, BPA that bears two CH3 groups has a KD value of The titration of K-Ras4B with BPC, BPM, BPP, BPZ, and the 600 ±200 µM[14], and shows mainly the same pattern as found related compounds, hexachlorophen, and bromochlorophen, for BPE. In addition, the observed chemical shift perturbation for lead to a denaturation of the protein, as judged by visible G13 suggest the allosteric effect upon ligand binding as this precipitation of samples and NMR spectroscopy. In total, 7 out of amino acid is close to the nucleotides’ ɣ-phosphate. This is close 14 Bisphenols could not be analysed by NMR-based titration to a site that is occupied with a glycerol molecule in a previously experiments due to their extremely low solubility. reported Ras structure complexed with other low molecular weight compounds in the same pocket that harbours BPs[9]. This site was also (only) found to be occupied by a ligand covalently bound to a K-Ras(G12C) mutant[12]. Nonetheless, the NMR chemical shift perturbation analysis advocate an allosteric effect of BPs binding to Ras on the region close to the nucleotides’ ɣ- phosphate rather than a second binding site for BPs[14]. It has been shown, that the introduction of methyl groups has a thermodynamically stabilising (steric) effect on the bisphenol [20] geometry . This could be the explanation for the improved KD value, since the ligand geometry is restrained and better fits to the surface of the protein. Taken together, the improvement of

the KD value arises from the fixation and stabilisation of the bisphenol in a more rigid orientation, which seems to be st Figure 1 Denotation of the different binding clusters, the 1 (ligand) ring (L6, beneficial for the interaction with the Ras protein. In case of BPB I55, L56, T74, G75), the not surface exposed hydrophobic cluster (L79, C80, that contains one methyl and one ethyl substitution we observed I93), the NBD (nucleotide binding cluster, D57 and G60), and their location a KD deterioration (KD value of 3.6 ± 0.7 mM). Supposedly, the within the GTPase fold. ethyl group is sterically too demanding and prevents the exact ring orientation required for tighter binding. It is well known that a fluorine substitution can lead to a change in the preferred Bisphenol F that carries two protons at the central carbon atom molecular conformation and can increase the binding affinity by binds to K-Ras4B with a KD value of 14 ± 1.6 mM. Furthermore, exploiting specific fluorine protein interactions. Consistently, we Bisphenol F caused additional significant chemical shift observed for BPAF, which carries two fluoromethyl groups, a perturbations for amino acids V9, D57, G60, L79, C80, and I93, slightly better KD of 347 ± 5 µM (Fig. S1). This interaction was which therefore have to be induced by the second ring and/or by further corroborated by 1D 19F NMR active spectroscopy that the bridging moiety (Fig. 1, 2a, S1). Much to our surprise these revealed a downfield shift of 2.15 ppm of BPAF upon protein amino acids were, in part, neither surface-exposed nor at the binding (Fig. S4). In order to investigate the competitive action of binding pocket itself. BPAF towards SOS, we carried out a titration of 14N-SOS to 15N- Based on their location in the common GTPase fold, these Ras in order to create a complex of around 100 kDa, which residues can be grouped into two clusters (Fig. 1). The first leads to a broadening of resonances of beyond detection due cluster is constituted by the amino acids L6, V9, L79, C80, and rotational correlation effects. This can be observed at a ratio of I93. L6 is a “first-ring”-residue and is therefore directly affected 1:1 and is more pronounced when more SOS is added (Fig. S3). by the ligand (Fig. 1). The amide faces the side of the ring, as Due to the low solubility of BPs, co-crystals could not be judged by the vector of the chemical shift perturbation induced obtained despite extensive screening and soaking trials. [19] by the aromatic ring current . We suppose, that the interaction However, judged from molecular dockings using the HADDOCK of the phenolic moiety and L6 is strengthened by the second ring, software suite, BPAF binds at the interface of SOS and RAS and so that L6 is able to transmit the binding information to V8 and antagonises Ras/SOS protein complex formation. V9 of β-sheet one. The next residues that establish this chain of contacts are L79/C80 (β-sheet 4) and finally I93 (α-helix 3). The

COMMUNICATION

Figure 2: a.) Overall view of the NMR chemical shift perturbation at 600 MHz and at 298 K observed for K-Ras4B bound to GDP upon titration with Bisphenol AF (upper panel) and BPNH2 (lower panel), ranging from the black reference to a ratio of 1: 25, shown in magenta. The enlarged panels illustrate residues G60, T74 and G75. b.) Comparison of certain amino acids (Y157 and T74) upon nucleotide loading (purple=GPPNHP), and the addition of ligand (cyan=GPPNHP+BPAF) The addition of different ligands leads to a pseudo-activation of the GTPase, based on the resonance of Y157 upon titration with different bisphenolic ligands (orange=AFX, red=BPF, teal=BPE, pink=BPA, violet=BPB, blue=BPAF, green=BPNH2, light blue=BPS). c.) Level of BPAF-induced pseudo-activation reported in % d.) Orientation of BPNH2 based on HADDOCK models, the first ring is located in the binding pocket formed by K5, L6, I55, L56, T74 and G75. The second ring faces the solvent.

characteristic, which impairs a precise determination of its KD When BPAF was then titrated to this complex, previously NMR spectroscopy. Therefore, we performed GDI assays that disappeared resonances emerged again. exploit the protein-bound fluorescence of mant-GDP. This

The aromatic rings of the BPAF analogue BPNH2 are amino- fluorescent nucleotide analogue is exchanged in the presence of

substituted at the meta-position and offer the opportunity to form the Ras-GEF SOS. The KD value obtained is 199±99 µM, which

hydrogen bonds. When BPNH2 was titrated to Ras we observed is better than the one of the unsubstituted BPAF (Fig. 3). This intermediate-exchange behaviour for the backbone amide result also underlines the potential physiological impact of the

resonances of amino acids L56, D57, T74, and G75 which protein-protein interaction, since BPNH2 is able to compete with indicates a change in binding kinetics at 600 MHz Lamor SOS (Fig. 3). A comparison of all compound-induced chemical frequency (Fig. 2a, 2d). For this binding mode, disappearance shift perturbations showed that some chemical shift vectors and reappearance of peaks at different resonances is evolve always into the same direction whereas NMR resonances

COMMUNICATION

of other amino acids display high directional inhomogeneity in finding implies that the interaction between K-Ras4B and its 2D 1H-15N HSQC (Fig. S5). This must be caused by the bridging downstream effector proteins might indeed be influenced by BPs. region of the BPs, since this part is different for each Bisphenol. L6 and L56, which are part of the pocket base, showed always the same shift pattern, and are therefore not affected by the bridging region. An opposed behaviour was observed for M67 and T74, which are part of the binding pocket rim (Fig. S5). Since the isoforms of K-Ras4B are highly related in terms of their amino acid sequence, we also tested H-Ras and N-Ras for their potential to interact with BPAF. For H-Ras and BPAF, the titration of the isoforms yields a KD value of 395 ± 4 µM that compares well to K-Ras4B. Not surprisingly, also the shift pattern is exactly the same (Fig. S2). For N-Ras (1-170), we observed a similar pattern of chemical shift perturbations, albeit the 2D 1H-15N-HSQC NMR spectrum is not assigned (Fig. S2).

In addition, the KD value was in the same range (770 ±7 µM). These results show that both of the Ras-isoforms interact equally to K-Ras4B with BPAF and that therefore the HVR- Region does not affect the ligand binding. Rap-1A, which shares an overall sequence similarity of 50 % with the Ras isoforms was also tested. Surprisingly, we were not able to detect any significant chemical shift perturbations induced by BPAF, although the ligand was clearly detectable in the control 1D 1H-NMR spectra and is thus sufficiently soluble (Fig. S2). The 1st ring contact residue T74 is changed to an asparagine residue in Rap-1A. In addition, the whole hydrophobic cluster is varied from V9 to L9, L79 to A79, C80 to L80, and I93 to L93, whereas L56 and D57 are not altered. Because of these numerous changes it is not possible to determine exactly what the reason for Rap-1A incapability to bind BPAF is. Nonetheless, it can be speculated that all in all subtle but concerted changes of the binding pocket of the GPTase(s) modulate their affinity towards different bisphenols. Figure 3: SOS-mediated nucleotide exchange assay (upper panel). A thorough analysis of all chemical shift perturbations reveals This Ras-GDI assays reveals KD value of 199±99 µM for BPNH2. MTT proliferation assay (lower four panels). The IC50 values in Hela cells that the inactive, GDP-bound, K-Ras4B-GTPase adopts an exposed for 24, 48 and 72 hrs to BPs were determined as follows: active conformation (Fig. 2b). The chemical shift difference for 53.79 µmol (BPA), 17.78 µmol (BPAF), 33.01 µmol (BPNH2), 119.30 µmol Y157 clear shows that the compounds gradually shift the (BPS). conformational equilibrium from the inactive to the active form[21]

(Fig. 2b,c). In quantitative terms, the amide resonance of Y157 is shifted by binding BPNH2 approx. 64 % to the resonance frequency of Y157 in active, GppNp-bound state (Fig. 5c, S5, Summary Cellular processes are controlled by the protein family of small S6). For BPAF, BPA, and BPS this effect is less pronounced GTPases that act as a binary on/off switch. Bisphenols are (58 %, 49%, 20 %, respectively). We note that these different widespread in modern life, because of their industrial application levels of inducing the active state of Ras correlate with KD values as plasticisers. The comprehensive study presented here is the determined by NMR and IC50 values can be predicted from the the first to scrutinise the interaction between small Ras-related vectorial analysis of the chemical shift perturbation and are GTPases (Ras-isoforms, Rap-1A) and various bisphenols: BPA, independent of the nucleotide loading state (Fig. 3, S5, S6). We BPAF, BPB, BPE, BPF, BPNH . We could show that bisphenols therefore have coined the term ‘pseudo-activation of Ras’. 2 Interestingly, this effect has also been observed in other small bind selectivity to a subset of small GTPases and induce their [22] pseudo-active conformation. Not only could we determine their GTPases: Rab1b, a main regulator of membrane trafficking . IC values but we also show that all of these bisphenols interact This protein is AMPylated by the Legionella effector protein DrrA 50 at tyrosine 77 (Y71 in Ras proteins) and is stabilised by this with all Ras isoforms but not with Rap-1A through utilising a modification in an active state, as shown by molecular dynamics common site on these proteins that induces their pseudo-active [23] conformation. In conclusion, we show here that the Ras isoforms simulation .It is interesting to note that the characteristic line are off-target proteins that should be considered in the future BP broadening for switch I and II of GppNHp-loaded GTPase does not occur for GDP-bound K-Ras4B when complexed with BPs. toxicity evaluation. Together with our previous studies on BPS, Apparently, the micro- to millisecond dynamics of the switch the results of our current study underlines the demand for further toxicity studies of BPs. regions that is characteristic for activated GTPases is absent in the GDP/K-Ras4B/BP complex (Figure 2b.). This important

COMMUNICATION

Acknowledgements 730110. We are grateful to the Deutsche Krebshilfe (109776 and 109777), the DFG (SFB 642), and the RUB Research [12] J. M. Ostrem, U. Peters, M. L. Sos, J. A. Wells, K. M. SchoolPlus for generous financial support. Shokat, Nature 2013, 503, 548–51.

[13] Q. Sun, J. P. Burke, J. Phan, M. C. Burns, E. T. Olejniczak, Keywords: Bisphenols • K-Ras4B • Rap-1A • NMR A. G. Waterson, T. Lee, O. W. Rossanese, S. W. Fesik, spectroscopy • pseudo-active conformation Angew. Chemie Int. Ed. 2012, 51, 6140–6143.

[1] I. R. Vetter, A. Wittinghofer, Science 2001, 294, 1299–304. [14] M. Schoepel, K. F. G. Jockers, P. M. Düppe, J. Autzen, V. N. Potheraveedu, S. Ince, K. T. Yip, R. Heumann, C. [2] I. Prior, P. Lewis, C. Mattos, Cancer Res. 2012, 72, 2457– Herrmann, J. Scherkenbeck, et al., J. Med. Chem. 2013, 56, 2467. 9664–72.

[3] S. Schubbert, K. Shannon, G. Bollag, Nat. Rev. Cancer [15] M. Schoepel, C. Herrmann, J. Scherkenbeck, R. Stoll, 2007, 7, 295–308. FEBS Lett. 2016, 590, 369–375.

[4] K. Nomura, H. Kanemura, T. Satoh, T. Kataoka, J. Biol. [16] F. S. Vom Saal, S. C. Nagel, B. L. Coe, B. M. Angle, J. A. Chem. 2004, 279, 22664–73. Taylor, Mol. Cell. Endocrinol. 2012, 354, 74–84.

[5] M. A. Daniels, E. Teixeiro, J. Gill, B. Hausmann, D. [17] L. N. Vandenberg, T. Colborn, T. B. Hayes, J. J. Heindel, D. Roubaty, K. Holmberg, G. Werlen, G. A. Holländer, N. R. J. R. Jacobs, D.-H. Lee, T. Shioda, A. M. Soto, F. S. vom Saal, Gascoigne, E. Palmer, Nature 2006, 444, 724–9. W. V Welshons, et al., Endocr. Rev. 2012, 33, 378–455.

[6] M. Schmick, N. Vartak, B. Papke, M. Kovacevic, D. C. [18] K. Pivnenko, G. A. Pedersen, E. Eriksson, T. F. Astrup, Truxius, L. Rossmannek, P. I. H. Bastiaens, Cell 2014, 157, Waste Manag. 2015, 44, 39–47. 459–71. [19] C. S. Wannere, P. V. R. Schleyer, Org. Lett. 2003, 5, 605– [7] C. Herrmann, G. Horn, M. Spaargaren, A. Wittinghofer, C. 8. Herrmann, M. Spaargaren, A. Wittinghofer, J. Biol. Chem. 1996, 271, 6794–6800. [20] J. Z. Dávalos, R. Herrero, J. C. S. J. C. S. J. Costa, L. M. N. B. F. L. M. N. B. F. L. M. N. B. F. Santos, J. F. Liebman, J. [8] N. Nassar, G. Horn, C. Herrmann, A. Scherer, F. Z. D valos, R. Herrero, J. C. S. J. C. S. J. Costa, L. M. N. B. McCormick, A. Wittinghofer, Nature 1995, 375, 554–60. F. L. M. N. B. F. L. M. N. B. F. Santos, J. F. Liebman, et al., J. Phys. Chem. A 2014, 118, 3705–3709. [9] T. Maurer, L. S. Garrenton, A. Oh, K. Pitts, D. J. Anderson, N. J. Skelton, B. P. Fauber, B. Pan, S. Malek, D. Stokoe, et [21] M. J. Smith, B. G. Neel, M. Ikura, Proc. Natl. Acad. Sci. al., Proc. Natl. Acad. Sci. U. S. A. 2012, 109, 5299–304. 2013, 110, 4574–9.

[10] M. C. M. Burns, Q. Sun, R. N. Daniels, D. Camper, J. P. [22] M. P. Müller, H. Peters, J. Blümer, W. Blankenfeldt, R. S. Kennedy, J. Phan, E. T. Olejniczak, T. Lee, A. G. Waterson, Goody, A. Itzen, Science 2010, 329, 946–9. O. W. Rossanese, et al., Proc. … 2014, 111, 3401–6. [23] M. P. Luitz, R. Bomblies, E. Ramcke, A. Itzen, M. [11] F. Shima, Y. Yoshikawa, M. Ye, Proc. … 2013, DOI Zacharias, Sci. Rep. 2016, 6, 19896. 10.1073/pnas.1217730110/- /DCSupplemental.www.pnas.org/cgi/doi/10.1073/pnas.1217

Supporting Information

Experimental Section

This section describes the expression and purification of proteins, multidimensional NMR-based titration experiments, molecular docking studies, the SOS-mediated nucleotide exchange assay, and the MTT assay.

Protein expression and purification

15N-enriched K-Ras4B was purified as previously published[1].

15N-enriched H-Ras4A (p21) and a truncated version of 15N-enriched Rap1A (1-170), both in the pGEX4T1-vector, were expressed as GST-tagged fusion proteins in BL21-T1R cells. The expression and purification was the same for both proteins. A 50 ml pre-culture of LB (LB, 10 g/L tryptone, 5 g/L yeast extract, 5 g/L NaCl) was inoculated with a single colony and incubated over night at 37 °C. This culture was then diluted to 1 l LB, cells were incubated until an OD600=1 and harvested. The cell pellet was resuspended and transferred in 250 ml of supplemented M9 media (50 mM Na2HPO4, 20 mM 15 KH2PO4, 9 mM NaCl, 4 g/L glucose, 0.5 g/L N-NH4Cl, 0.1 mM CaCl2, 2 mM MgSO4, 10 mg/L thiamine, 10 mg/L biotin, and 150 mg/L ampicillin, pH 7.4). After incubation for 1 h at 37 °C, the temperature was changed to 30 °C and the protein expression was induced by the adding 1 mM Isopropl β-D-1- thiogalactopyranoside (IPTG). After 4 h, the cells were centrifuged and were stored at -20 °C until needed. Cells were resuspended in Tris buffer (50 mM Tris 7.4, 150 mM NaCl, 5 mM MgCl2, 2 mM DTT) that contained a iture of protease ihiitors Oplete™ EDTA-free, Roche). Cells were lysed using a Microfluidizer and cell debris was removed by centrifugation. The batch purification process involved the immobilisation of the protein on glutathione beads (Protino Glutathione Agarose 4B, Machery Nagel), followed by two washing steps and overnight on-column-proteolysis with the thrombin (Sigma Aldrich) at 8 °C. An additional elution step was applied to the supernatant. The purity of the two fractions was analysed by SDS-PAGE and the protein was concentrated to 0.4 mM, frozen in liquid nitrogen, and stored at -80 °C.

15N-enriched N-Ras was purified using the pTac-vector system. The vector was transformed in BL21T1R-cells and the cells were grown as described for H-Ras and Rap1a, with the exception that the

1 period of protein production expression was changed from 4 h to 18 h at 37 °C. The protein purification was carried out as published[2].

For the expression and purification of 14N SOS-RBD protein, the pProEX-Htb-SOS-RBD-vector was transformed in BL21(DE3) cells and a pre-culture of 50 ml LB media was inoculated with a single colony and incubated over night at 37 °C. The cells were then diluted with 1 L of LB media and grown until an OD600 of 0.8 was reached. Isopropl β-D-1-thiogalactopyranoside (0.1 mM final concentration) was added to induce protein expression at 18 °C overnight. After harvesting, cells were stored at -20 °C. The protein was purified by using Ni Sepharose 6 Fast Flow (GE Healthcare Life Sciences) and a GE ÄKTA Purifier FPLC. Finally, the protein was applied to a size exclusion column (GE Superdex 75 column) and eluted with 20 mM HEPES pH 7,4, 150 mM NaCl; 1mM DTT. The Protein was concentrated and frozen in liquid nitrogen.

Low molecular weight compounds and solvents

AFX (4-(Trifluoromethyl)phenol, CAS 402-45-9), Bisphenol A (BPA, 2,2-Bis(4-hydroxyphenyl)propane, CAS 80-05-7), Bisphenol AF (BPAF, 2,2-Bis(4-hydroxyphenyl)hexafluoropropane, CAS 1478-61-1), Bisphenol AP (BPAP, ,′-(1-Phenylethylidene)bisphenol, CAS 571-75-1), Bisphenol BP (BPBP, Bis(4- hydroxyphenyl)diphenylmethane, CAS 1844-01-5), Bisphenol C (BPC, 2,2-Bis(4-hydroxy-3- methylphenyl)propane, CAS 79-97-0), Bisphenol FL (BPFL, ,′-(9- Fluorenylidene)diphenol, CAS 3236-71-3), Bisphenol M (BPM, ,′-(1,3- Phenylenediisopropylidene)bisphenol, CAS 13595-25-0), Bisphenol NH2 (BPNH2, 2,2-Bis(3-amino-4- hydroxyphenyl)hexafluoropropane, CAS 83558-87-6), Bisphenol P (BPP, ,′-(1,4- Phenylenediisopropylidene)bisphenol, CAS 2167-51-3), Bisphenol S (BP“, ,’- Sulfonyldiphenol, CAS 80-09-1), Bisphenol Z (BPZ, 4,4′-Cyclohexylidenebisphenol, CAS 843-55-0) were purchased from Sigma-Aldrich (St. Louis, MO). Deuterated solvents for NMR measurements were obtained from Deutero GmbH (Kastellaun, Germany). Bisphenol B, (BPB, 2,2-Bis(4-hydroxyphenyl)butane, CAS 77-40-7), Bisphenol E (BPE, 1,1-Bis(4- hydroxyphenyl)ethane, CAS 2081 08 5), and Bisphenol F (BPF, ,′-Methylenediphenol, CAS 620-92-8) were purchased from TCI Germany (Eschborn, Germany).

Since it has previously been shown that bisphenols were able to bind to K-Ras4B we questioned whether a single aromatic (phenolic) ligand would be also capable of binding. To address this, we also tested a smaller fragment [4-(trifluoromethyl)phenol, AFX]. Its affinity is rather weak and prevented a determination of its KD value as the saturation could not be reached before the protein started to denature. Nonetheless, AFX partly showed the same binding pattern as observed for the bisphenols BPA and BPS. This included a significant chemical shift perturbation for the amino acids L6, I55, L56, T74, and G75 that cluster to form a defined ligand binding site. We denote these residues as first-rig-residues (Fig. 1).

Bis- Systematic name (IUPAC) Structural formula CAS Comment phenol

2,2-Bis(4-hydroxyphenyl) A 80-05-7 K =0.6 ± 0.2 mM propane D

1478- 2,2-Bis(4-hydroxyphenyl) AF K =0.4 ± 0.1 mM 61-1 hexafluoropropane D

1571- 1,1-Bis(4-hydroxyphenyl)-1- AP insoluble 75-1 phenyl-ethane

2,2-Bis(4- B 77-40-7 K =3.6 ± 0.7 mM hydroxyphenyl)butane D

1844- Bis-(4-hydroxyphenyl) BP insoluble 01-5 diphenylmethane

14868- Bis(4-hydroxyphenyl)-2,2- denaturation of C 2 03-2 dichlorethylene protein

2081- 1,1-Bis(4- E K =7 ± 0.7 mM 08-5 hydroxyphenyl)ethane D

87139- Bis(4- F K =14 ± 2 mM 40-0 hydroxyphenyl)methane D

9,9-Bis(4- 3236- FL hydroxyphenyl)fluorene insoluble 71-3

13595- 1,3-Bis(2-(4-hydroxyphenyl)- denaturation of M 25-0 2-propyl)benzene protein

2167- 1,4-Bis(2-(4-hydroxyphenyl)- denaturation of P 51-3 2-propyl)benzene protein

S 80-09-1 Bis(4-hydroxyphenyl)sulfone KD=6 ± 0.7 mM

843-55- 1,1-Bis(4-hydroxyphenyl)- denaturation of Z 0 cyclohexane protein

83558- 1,1-Bis(4-hydroxyphenyl)- NH K =0.4 ± 0.1 mM 2 87-6 cyclohexane D

Multidimensional NMR-based titration experiments

The titrations were performed as previously published[1]. In order to decrease solubility artefacts, the different ligands were titrated in the same % v/v steps, using different stock concentration.

19F NMR Spectroscopy

The 19F NMR spectra were recorded on a Bruker DPX 250 MHz spectrometer. The reference experiment was recorded on sample of 0.2 mM BPAF in PBS 7.4. Upon addition the K-Ras4B protein in a molar ratio of 1:1, a downfield shift of 2.15 ppm could be observed.

HADDOCK-based modelling

Assignment and data handling were performed using ccpNMR. Docking was performed using HADDOCK 2 and CNS 1.2 as previously published[1]. Docking interfaces were defined for each compound by active and passive ambiguous interaction restraints (AIRs) and using the atomic coordinate set 4DSO deposited in the RCSB-database[3]. The active and passive AIRs were chosen judged from the highest chemical shift and the location, as indicated on the HADDOCK homepage[4]. For AFX, BPF, BPE, and BPA, the overall binding pattern did not change, so that the residues L6, I55, L56, D57, T74, G75 were selected as AIRs. BPB´s replacement of one methyl group to an ethyl group did have an influence, so that M72 was chosen over I55 as an AIR. The only bisphenol analogue with a sulphone moiety, let to a significant changed pattern, so that E37, S39, L56, M67, T74, G75 were selected as active residues. The BPAF structure is based on one central carbon atom which is flanked by two phenolic moieties and two trifluoromethyl groups. The AIRs are more alike to those chosen for the carbon-based bisphenols, so that L6, I24, L56, D57, T74, G75 were selected. In the case of

BPNH2 the binding motif changed for many amino acids from a fast exchange to intermediate exchange behaviour. Thus, these were then selected as AIRs for molecular docking calculations. In addition, amino acids corresponding to the nucleotide binding site showed line broadening effects. However, this issue was not considered in the docking process, since it was regarded as an allosteric effect. In this case, the residues L6, I24, L56, M72, T74, and G75 were chosen as AIRs. The surrounding water accessible amino acids were defined as passive residues (T3, K5, I36, E37, D38, S39, Y40, R41, L52, D54, D69, Q70, R73, E76, K104) and remained the same for all BPs, unless they had already been chosen as active residues. Structure visualisation and superimposition based on

RMSD values for Cα, C, and N atoms were performed using PyMol (Delano, W. L., The PyMol Molecular Graphics System (2002) Delano Scientific, Palo Alto, CA, USA).

Competitive Titration of 14N SOS-RBD and BPAF on 15N-enriched GDP-bound K-Ras4B

When 14N SOS-RBD is titrated to 15N-enriched GDP-bound K-Ras4B (0.4 mM), a protein-protein complex of approx. 100 kDa is formed. This increase in molecular weight leads to a broadening of NMR resonances, because of rotational correlation effects. This effect can already be observed in 2D 1H-15N HSQC spectra at a molar ratio of 1:1 (14N SOS-RBD:15N-enriched GDP-bound K-Ras4B) and is more pronounced when more SOS is added to adjust a molar ratio of 1:2. Addition of an increasing amount of BPAF to the 1:2 15N-enriched K-Ras4B/ 14N SOS complex (1:2:0.5, 1:2:1, 1:2:4) recovers the

6 backbone amide proton NMR resonances in 2D 1H-15N HSQC spectra. The acquired NMR spectrum is similar to the 15N-enriched GDP-bound K-Ras4B:BPAF 1H-15N HSQC spectra spectrum.

SOS-mediated nucleotide exchange assay

The assay was performed as prescribed[1]. A solution of 1µM Ras*mantGDP, 200µM GDP was mixed with different concentrations of BPNH2 and incubated at 20°C for 5-10 min. To prevent insolubility of the BPNH2, a final DMSO concentration of 5 v/v % was used. After that, 0.5µM of SOS was added and the change in fluorescence (excitation wavelength, 366 nm; emission wavelength, 442 nm) was recorded with a Perkin Elmer LS50B instrument.

MTT proliferation assay

The IC50 values were assessed by MTT assay. Hela cells were seeded (1100 cells/well) in a 96-well microtiter plate followed by overnight incubation. The day after, cells were treated with individual bisphenols at indicated concentrations or corresponding DMSO (as control) in 100 µl medium up to 72 hours. Thereafter, 10 μl MTT solutio g/l was added to eah well for h. The iture was removed carefully via pipetting, and the remaining formazan crystals formed were dissolved by 100 μl DM“O/ % SDS/ 0.01 M acetic acid. After 15 min, the absorbance of each well was read at 570 nm (reference 620 nm) with an absorbance reader. Blank values were subtracted and IC50 values were calculated by using Prism 6 (GraphPad Software).

Abbreviations

1D, one-dimensional; 2D, two-dimensional; BPs, Bisphenols; BPA, Bisphenol A; BPS, Bisphenol S;; EDC, endocrine disrupting chemical; GAP, GTPase activating protein; GDI, guanine nucleotide dissociation inhibition; GDP, guanosine diphosphate; GTP, guanosine triphosphate; GEF, guanine nucleotide exchange factor; HSQC, heteronuclear single quantum coherence; KD dissociation constant; NMR, nuclear magnetic resonance; RMSD, root mean square deviation; SD, standard deviation; Sos, son of sevenless, BP, bisphenols; BPA, BPAF, BPB, BPE, BPF, BPNH2.

Supporting Figures

Supporting Figure 1. On the left hand side, the chemical shift differences (in ppm.) are plotted against the ligand concentration (in mM). For the binding pocket, six representative residues (L6, V9, Y40, L56, T74 and G75) were chosen in order to calculate the K D value that yielded 347 ± 5 µM. In the right hand panel, a HADDOCK- based model is shown, in which the amino acids with a weighted chemical shift above once the standard deviation (SD) are annotated in orange and those with a shift with a value twice the SD in red. The BPAF molecule is represented in green.

Supporting Figure 1. Overall view of the NMR chemical shift perturbation at 600 MHz and at 298 K observed in 2D 1H-15N HSQC spectra of H-Ras, K-Ras4B, N-Ras, and Rap-1A bound to GDP upon titration with Bisphenol AF, ranging from the black reference to a ratio of 1 : 25, shown in magenta. Rap-1A did not show significant binding at all. 8

Supporting Figure 3. Competitive titration of 15N-enriched GDP-bound K-Ras4B with 14N SOS-RBD and BPAF at 600 MHz and at 298 K. Different 2D 1H-15N HSQC spectra of this titration are shown, starting with the K-Ras4B GDP protein only (reference in black). In dark and lighter blue spectra with the GEF-protein SOS added (molar ratios of 1:1 and 1:2) are depicted. The resulting line broadening of resonances is obvious. In green, the ligand BPAF is added in a molar ratio of 1:2:0.5. In orange and pink, the 1:2:2 and 1:2:4 molar titration steps are shown. The recovery of all backbone amide proton NMR is clearly visible. It is important to note, that the recovered resonances of residues from the binding pocket exhibit chemical shift perturbations compared to ligand-free 2D 1H-15N HSQC spectra of 15N-enriched GDP-bound K-Ras4B. 9

Supporting Figure 4. 1D 19F-Spectra at 250 MHz and at 298 K. In black, the reference BPAF in PBS at pH 7.4 is shown. In blue, the chemical shift corresponding to the protein-bound BPAF (in a molar ratio of 1:1) is presented.

Supporting Figure 5. In the left hand panel, a vectorial shift analysis of L6, L56 (in pink) as representative residues of the binding pocket is shown based on the addition 1 15 of BPNH2 derived from the chemical shift perturbation observed in 2D H- N HSQC. The different shifts follow the same pattern, besides of L56 in BPS. In cyan, two representatives from the rim of the binding pocket are shown, the vectorial shift pattern of M67 and T74 are characterised by a high dispersion in the 1H as well as in the 15N dimension. In the right hand panel, the L6, L56, M67, and T74 are highlighted 11 in pink and cyan on the protein surface superimposed with structure of a BPNH 2 docked onto GDP-bound K-Ras4B.

Supporting Figure 6. Column chart of the pseudo-activation rates of different bisphenols and AFX (Trifluormethylphenol) based on the chemical shift perturbation of Y157, normalised against the shift of the GPPNHP-activated Ras (100%). AFX does not lead to a pseudo-activation of Ras. In contrast, BPS and BPF cause a rate of around 20 %, whereas BPE and BPB induce 37.3 % and 32.8 %, respectively. Remarkably, BPA

evokes a rate of nearly 50 %, only surpassed by BPAF (57.5 %) and BPNH 2 (64.2 %).

Acknowledgements

We are grateful to the Deutsche Krebshilfe (109776 and 109777), the DFG (SFB 642), and the RUB Research SchoolPlus for generous financial support.

Part III: Summary

3. Summary

The Ras oncogene family has been extensively studied over the last 3 decades, with up to now more than 40,000 scientific articles published88. Due to its role as an oncogene, Ras has attracted a huge interest from the scientific community. Oncogenes are genes that are closely linked to cancer, since mutations change the function of the encoded protein. This creates the malignant properties that are needed for cancer to grow and spread. The Ras protein was among the first to be discovered as an oncogene in 1982 by several groups89–92. Subsequently, tremendous effort was carried out to understand the function of Ras in cellular proliferation and to characterise its structural features. Highlights of this efforts were the crystallisation of the Ras protein alone93, the Ras-RasGAP complex94, and the Ras-RasGEF complex95. Inhibiting Ras´ cancer related actions were then the next step, but all failed in reaching clinical application23,24, which led scientist to think of Ras as an undruggable target in combatting cancer. However, recent data show that there might be indeed a chance to tackle the protein in different ways to finally achieve a Ras-targeted therapy. Despite the absence of hydrophobic cavities, several groups published papers, that describe new Ras ligands, which bind directly34–37. Interestingly, the depicted binding pocket cannot be ascertained in the published apo-crystal structures of Ras, and is only visible upon ligand binding. In other words, the search for ligand based on the crystal structure alone does not make sense in the case of the Ras proteins, without taking into account the dynamic features of the Ras proteins, especially when Ras is loaded with GDP. Maurer et al. described in 2012 Ras ligands, that bind in a pocket between helix α2 and the core β-sheet, β1–β334. The same pocket was also reported by Sun et al. in the same year35. Both authors described indolic ligands, with one or two indolic moieties respectively. In 2014, Burns et al. characterised another ligand binding pocket, which was only found, when the first binding pocket was blocked by an covalently bound ligand37. This thesis is mainly based on two peer-reviewed publications and a submitted manuscript. The first one was published in November 2013 in the Journal of medicinal chemistry (impact factor in 2013: 5.6). This paper has been cited up to now (July 2016) twelve times. In this paper, multidimensional heteronuclear NMR spectroscopy and

85 chemical shift perturbation analysis were applied, in order to characterise the interactions between low molecular weight compounds and K-Ras4B. The binding pocket includes N-terminal residues of β-strand 1, as well as Switch I and Switch II and is identical to the pocket already mentioned above. With the help of a docking model, it was shown that for optimal binding aromatic rings of BPA must be able to adopt almost a right angle. In addition, it has been shown that Bisphenol A significantly reduces the SOS-mediated nucleotide exchange reaction of both H- and K-Ras4B by a factor of 2.5 and 1.6, respectively. A possible explanation for this observation could be a steric clash between Bisphenol A and SOS, that prevents complex formation. In the next publication the question of a potential binding of BPS (Bisphenol S) was answered, with the result that BPS is also a K-Ras4B ligand. BPS is used as a substitute for BPA in everyday plastic materials, as shown in several publications96,97. There is mounting evidence, that BPS shows a similar or at least comparable physiological effect by functioning as an EDC 98,99. The obtained binding affinity of 5.8±0.7 mM is much lower than that of BPA. A potential explanation of this finding could be the different 3D structure of BPA and BPS: In BPS the two times methyl substituted carbon atom is replaced by a sulphonyl group, which leads to a different geometry of the molecule, since the free electron pairs of the oxygens force the atoms to occupy a bent orientation. Both ligands´ first rings exhibit the same binding mode, as judged from the chemical shift perturbations of the corresponding amino acids (e.g. L6, I55, and L56). In contrast to the first ring, the second ring of BPA and BPS is orientated differently. BPS induces chemical shift perturbations of amino acids I36 and S39, which are located further away from the originally described binding pocket. BPS probably rises above the pocket and reaches out for these residues. In contrast, the second ring of BPA point towards Q70. These structural features can be observed in the corresponding docking structures. Additionally, the SOS-mediated nucleotide exchange was not affected by this ligand.

In accordance with similar results by other groups, we correlated the KD value with the nucleotide exchange reaction. The fact that BPS binds with lower affinity can now be understood in a structural content of the BPS-Ras interaction. These results were published in the FEBS Letters journal (impact factor 2015: 3.169) in February 2016.

In order to characterise the interactions of bisphenolic compounds and small Ras-like GTPases to a greater extend, a group of, in total, 14 Bisphenols, which mainly differ in their bridging region, together with other compounds were chosen to be tested on K- Ras4B, Ras isoforms, and Ras-like GTPases. In a structure activity relation (SAR) by NMR approach the bisphenols were characterised according to their KD value, their docking mode, and their ability to interfere with the SOS mediated nucleotide exchange. The newly tested BPs include: BPF (two protons), BPE (one proton & one methyl group), BPB (one methyl & one ethyl group), BPAF (two trifluoromethyl groups), and BPNH2 (two trifluoromethyl groups, and two meta substitutions with an amine group at each ring). The results of these BPs together with the already published BPA and BPS show, that the best KD values were obtained for BPA, BPAF, and BPNH2, with values in the high micromolar range (600 µM, 347 µM, and 198 µM, respectively). The 3D geometry of bisphenols is sterically stabilised by the introduction of methyl groups100. It is well possible, that the improvement of the KD value is a result of this stabilisation that generates a more rigid orientation, which seems to be beneficial in a pre-fit like manner for the interaction with the Ras protein. Additionally, an introduction of trifluoromethyl (-CF3) groups instead of methyl (-CH3) increases the binding affinity, presumably through exploiting specific fluorine protein interactions101,102. The BPAF molecule was also tested for its ability to bind to the Ras-Isoforms, H-Ras and N-Ras, with the result that both proteins bind the molecule with KD values in the same range as observed for K-Ras4B. Since the main difference between these isoforms is the so-called HVR (hypervariable region), this particular region is bot involved in the molecular interactions. The next protein tested was then Rap1A, which exhibits a sequence identity over all residues of around 50 %. Remarkably, there was no significant shift observable upon titrating of BPAF, although the ligand was clearly detectable in the control 1D 1H-NMR spectrum. To explain this observation, we focused on the sequence of Rap1A. The first ring contact residue T74 is replaced by an asparagine residue (N74) in Rap1A. In addition, the entire hydrophobic cluster is changed from V9 to L9, L79 to A79, C80 to L80, and I93 to L93. In contrast, L56 and D57 are not altered. Because of these numerous changes, it is not possible to clearly state why Rap1A is not capable of binding BPAF. Through analysing all compound-induced chemical shift perturbations for every Bisphenol, a compound “pseudo on”-conformation of the Ras protein is revealed. As

87 introduced by Smith et al., the amino acid Y157 is a valuable tracer for the nucleotide loading state or Ras. It can be observed, that the compounds gradually shift the conformational equilibrium from the inactive to the active form although the nucleotide itself is not changed. In quantitative terms, the cross peak of the BPNH2 is shifted around 64 %, BPAF 58 %, BPA 49 % and BPS 20 % to the ppm values of the active GTP-Ras conformation. Moreover, the active Ras protein (GPPNHP-loaded) showed no binding to any Bisphenol, which could be explained by the different dynamical features of the active and inactive GTPase and the presence or absence of the binding pocket. Interestingly, this 'pseudo-activation' has also been observed in other small GTPases: Rab-1B, a main regulator of membrane trafficking. This protein is AMPylated by the Legionella effector protein DrrA at tyrosine 77 (equivalent to Y71 in Ras proteins) and is stabilised by this modification in an active state, as shown by molecular dynamics simulation. Interestingly, the typical line broadening of the GTP form is not observed, so there is an obvious dynamical difference between the pseudo-activation and the wildtype GTP-form of Ras. Taken together, it was shown for the first time that various bisphenolic compounds bind and pseudo-activate Ras isoforms. In sharp contrast, Rap1A does not bind to Bisphenols.

4. Conclusion & Outlook

Herein a comprehensive study of the interaction of Ras related proteins and a variety of Bisphenols (BPs) is presented. In a structure activity related approach it was shown, that the introduction of two trifluoromethyl groups at the bridging moiety of the bisphenolic compounds is the best choice, regarding the KD. In a second step, the phenolic ring itself was substituted with an amine group. This lead to a slight improvement of the KD and resulted in a different dynamical status of the small GTPase Ras. By the detailed observation of a loading state tracer cross peak (Y157), a “pseudo-activation” of the protein was observed. The term “pseudo” was chosen, since the spectrum (ligand bound, GDP) is not similar to those corresponding to the inactive (ligand free, GDP) or active (ligand free, GPPNHP). Moreover, the active Ras protein (GPPNHP-loaded) showed no binding to any Bisphenol. Additionally, the two isoforms N-Ras and H-Ras were tested upon their ability to bind to BPAF. The observed KD was in the same range as the KD of K-Ras4B. Interestingly, the sequentially related protein Rap1A, whose 2D 1H-15N-TROSY spectrum was firstly introduced in this thesis, did not show any binding to the Bisphenols. Indeed, this has to be further investigated in future experiments, in which, for example, the sequential differences between Ras and Rap could be erased by a “Quikchange” mutagenesis. Despite intensive trials, up to now it was not possible to obtain co-crystals of Bisphenols and K-Ras4B, due to the fact that BPs are only poor soluble in water. The synthesis of Bisphenols derivates with a higher water solubility is fundamental to further crystallisation trials. These co-crystals could then help to design ligands, that show a higher affinity to K-Ras4B.

Bibliography

1. Knowles, J. R. Enzyme-catalyzed phosphoryl transfer reactions. Annu. Rev. Biochem. 49, 877–919 (1980). 2. Vetter, I. R. & Wittinghofer, A. The guanine nucleotide-binding switch in three dimensions. Science 294, 1299–304 (2001). 3. Overington, J. P., Al-Lazikani, B. & Hopkins, A. L. How many drug targets are there? Nat. Rev. Drug Discov. 5, 993–6 (2006). 4. Gibbs, J. B., Sigal, I. S., Poe, M. & Scolnick, E. M. Intrinsic GTPase activity distinguishes normal and oncogenic ras p21 molecules. Proc. Natl. Acad. Sci. U. S. A. 81, 5704–8 (1984). 5. Lodish, H. et al. Molecular Cell Biology. (2000). 6. Gideon, P. et al. Mutational and kinetic analyses of the GTPase-activating protein (GAP)-p21 interaction: the C-terminal domain of GAP is not sufficient for full activity. Mol. Cell. Biol. 12, 2050–2056 (1992). 7. Saraste, M., Sibbald, P. R. & Wittinghofer, A. The P-loop--a common motif in ATP- and GTP-binding proteins. Trends Biochem. Sci. 15, 430–4 (1990). 8. Wittinghofer, A. & Vetter, I. R. Structure-function relationships of the G domain, a canonical switch motif. Annu. Rev. Biochem. 80, 943–971 (2011). 9. Valencia, A., Chardin, P., Wittinghofer, A. & Sander, C. The ras protein family: evolutionary tree and role of conserved amino acids. Biochemistry 30, 4637–48 (1991). 10. Rojas, A. M., Fuentes, G., Rausell, A. & Valencia, A. The Ras protein superfamily: Evolutionary tree and role of conserved amino acids. J. Cell Biol. 196, 189–201 (2012). 11. Zerial, M. & McBride, H. Rab proteins as membrane organizers. Nat. Rev. Mol. Cell Biol. 2, 107–117 (2001). 12. Munro, S. The Arf-like GTPase Arl1 and its role in membrane traffic. Biochem. Soc. Trans. 33, 601–5 (2005). 13. Heasman, S. J. & Ridley, A. J. Mammalian Rho GTPases: new insights into their functions from in vivo studies. Nat. Rev. Mol. Cell Biol. 9, 690–701 (2008). 14. Rojas, A. M. & Valencia, A. in Ras Superfamily Small G Proteins Biol. Mech. 1 3–23 (Springer Vienna, 2014). doi:10.1007/978-3-7091-1806-1_1

15. Aronheim, A. et al. Membrane targeting of the nucleotide exchange factor Sos is sufficient for activating the Ras signaling pathway. Cell 78, 949–61 (1994). 16. Gao, J., Liao, J. & Yang, G.-Y. CAAX-box protein, prenylation process and carcinogenesis. Am. J. Transl. Res. 1, 312–25 (2009). 17. Bivona, T. G. et al. PKC regulates a farnesyl-electrostatic switch on K-Ras that promotes its association with Bcl-XL on mitochondria and induces apoptosis. Mol. Cell 21, 481–93 (2006). 18. Gysin, S., Salt, M., Young, A. & McCormick, F. Therapeutic strategies for targeting ras proteins. Genes Cancer 2, 359–72 (2011). 19. Schubbert, S., Shannon, K. & Bollag, G. Hyperactive Ras in developmental disorders and cancer. Nat. Rev. Cancer 7, 295–308 (2007). 20. Cherfils, J. & Chardin, P. GEFs: structural basis for their activation of small GTP- binding proteins. Trends Biochem. Sci. 24, 306–311 (1999). 21. Hayes, T. K., Stratford, J. K., Wang-Gillam, A. & Der, C. J. in Ras Superfamily Small G Proteins Biol. Mech. 1 157–171 (Springer Vienna, 2014). doi:10.1007/978-3-7091-1806-1_7 22. Schubbert, S., Bollag, G. & Shannon, K. Deregulated Ras signaling in developmental disorders: new tricks for an old dog. Curr. Opin. Genet. Dev. 17, 15–22 (2007). 23. Whyte, D. B. et al. K- and N-Ras are geranylgeranylated in cells treated with farnesyl protein transferase inhibitors. J. Biol. Chem. 272, 14459–64 (1997). 24. Fiordalisi, J. J. et al. High affinity for farnesyltransferase and alternative prenylation contribute individually to K-Ras4B resistance to farnesyltransferase inhibitors. J. Biol. Chem. 278, 41718–27 (2003). 25. Thompson, H. US National Cancer Institute’s new Ras project targets an old foe. Nat. Med. 19, 949–950 (2013). 26. Rotblat, B., Ehrlich, M., Haklai, R. & Kloog, Y. The Ras inhibitor farnesylthiosalicylic acid (Salirasib) disrupts the spatiotemporal localization of active Ras: a potential treatment for cancer. Methods Enzymol. 439, 467–89 (2008). 27. Mohammed, I. et al. 8-Hydroxyquinoline-based inhibitors of the Rce1 protease disrupt Ras membrane localization in human cells. Bioorg. Med. Chem. 24, 160– 178 (2016). 28. Manolaridis, I. et al. Mechanism of farnesylated CAAX protein processing by the

intramembrane protease Rce1. Nature 504, 301–305 (2013). 29. Chandra, A. et al. The GDI-like solubilizing factor PDEδ sustains the spatial organization and signalling of Ras family proteins. Nat. Cell Biol. 14, 148–58 (2012). 30. Zimmermann, G. et al. Small molecule inhibition of the KRAS-PDEδ interaction impairs oncogenic KRAS signalling. Nature 497, 638–42 (2013). 31. Mitin, N., Rossman, K. L. & Der, C. J. Signaling interplay in Ras superfamily function. Curr. Biol. 15, R563–74 (2005). 32. Athuluri-Divakar, S. K. et al. A Small Molecule RAS-Mimetic Disrupts RAS Association with Effector Proteins to Block Signaling. Cell 165, 643–55 (2016). 33. Cox, A. D. & Der, C. J. Ras history: The saga continues. Small GTPases 1, 2– 27 (2010). 34. Maurer, T. et al. Small-molecule ligands bind to a distinct pocket in Ras and inhibit SOS-mediated nucleotide exchange activity. Proc. Natl. Acad. Sci. U. S. A. 109, 5299–304 (2012). 35. Sun, Q. et al. Discovery of small molecules that bind to K-Ras and inhibit Sos- mediated activation. Angew. Chemie - Int. Ed. 51, 6140–6143 (2012). 36. Ostrem, J. M., Peters, U., Sos, M. L., Wells, J. A. & Shokat, K. M. K-Ras(G12C) inhibitors allosterically control GTP affinity and effector interactions. Nature 503, 548–51 (2013). 37. Burns, M. C. M. et al. Approach for targeting Ras with small molecules that activate SOS-mediated nucleotide exchange. Proc. … 111, 3401–6 (2014). 38. Mossessova, E., Corpina, R. A. & Goldberg, J. Crystal Structure of ARF1 and Sec7 Complexed with Brefeldin A and Its Implications for the Guanine Nucleotide Exchange Mechanism. Mol. Cell 12, 1403–1411 (2003). 39. Pizon, V., Chardin, P., Lerosey, I., Olofsson, B. & Tavitian, A. Human cDNAs rap1 and rap2 homologous to the Drosophila gene Dras3 encode proteins closely related to ras in the ‘effector’ region. Oncogene 3, 201–4 (1988). 40. Pizon, V., Lerosey, I., Chardin, P. & Tavitian, A. Nucleotide sequence of a human cDNA encoding a ras-related protein (rap1B). Nucleic Acids Res. 16, 7719 (1988). 41. Rousseau-Merck, M. F., Pizon, V., Tavitian, A. & Berger, R. Chromosome mapping of the human RAS-related RAP1A, RAP1B, and RAP2 genes to chromosomes 1p12----p13, 12q14, and 13q34, respectively. Cytogenet. Cell

Genet. 53, 2–4 (1990). 42. Ragain, C. M., Newberry, R. W., Ritchie, A. W. & Webb, L. J. Role of electrostatics in differential binding of RalGDS to rap mutations E30D and K31E investigated by vibrational spectroscopy of thiocyanate probes. J. Phys. Chem. B 116, 9326–9336 (2012). 43. Kitayama, H., Sugimoto, Y., Matsuzaki, T., Ikawa, Y. & Noda, M. A ras-related gene with transformation suppressor activity. Cell 56, 77–84 (1989). 44. Leach, S. D., Berger, D. H., Davidson, B. S., Curley, S. A. & Tainsky, M. A. Enhanced Krev-1 expression inhibits the growth of pancreatic adenocarcinoma cells. Pancreas 16, 491–8 (1998). 45. Damak, S., Harnboonsong, Y., George, P. M. & Bullock, D. W. Expression of human Krev-1 gene in lungs of transgenic mice and subsequent reduction in multiplicity of ethyl carbamate-induced lung adenomas. Mol. Carcinog. 17, 84– 91 (1996). 46. Kondoh, N. et al. The S29 ribosomal protein increases tumor suppressor activity of K rev-1 gene on v-K ras-transformed NIH3T3 cells. Cancer 89, 2–7 (1996). 47. Bos, J. L. et al. Semaphorin signaling meets rap. Sci. Signal. 5, pe6 (2012). 48. Sebzda, E., Bracke, M., Tugal, T., Hogg, N. & Cantrell, D. A. Rap1A positively regulates T cells via integrin activation rather than inhibiting lymphocyte signaling. Nat. Immunol. 3, 251–258 (2002). 49. McLeod, S. J., Shum, A. J., Lee, R. L., Takei, F. & Gold, M. R. The Rap GTPases regulate integrin-mediated adhesion, cell spreading, actin polymerization, and Pyk2 tyrosine phosphorylation in B lymphocytes. J. Biol. Chem. 279, 12009–19 (2004). 50. Birukova, A. A., Tian, X., Tian, Y., Higginbotham, K. & Birukov, K. G. Rap-afadin axis in control of Rho signaling and endothelial barrier recovery. Mol. Biol. Cell 24, 2678–88 (2013). 51. Herrmann, C. et al. Differential Interaction of the Ras Family GTP-binding Proteins H-Ras, Rap1A, and R-Ras with the Putative Effector Molecules Raf Kinase and Ral-Guanine Nucleotide Exchange Factor. J. Biol. Chem. 271, 6794– 6800 (1996). 52. Nassar, N. et al. Ras/Rap effector specificity determined by charge reversal. Nat. Struct. Biol. 3, 723–729 (1996). 53. Kiel, C., Selzer, T., Shaul, Y., Schreiber, G. & Herrmann, C. Electrostatically

optimized Ras-binding Ral guanine dissociation stimulator mutants increase the rate of association by stabilizing the encounter complex. Proc. Natl. Acad. Sci. U. S. A. 101, 9223–8 (2004). 54. Ohmstede, C. A., Farrell, F. X., Reep, B. R., Clemetson, K. J. & Lapetina, E. G. RAP2B: a RAS-related GTP-binding protein from platelets. Proc. Natl. Acad. Sci. U. S. A. 87, 6527–31 (1990). 55. Miertzschke, M. et al. Characterization of interactions of adapter protein RAPL/Nore1B with RAP GTPases and their role in T cell migration. J. Biol. Chem. 282, 30629–42 (2007). 56. Nonaka, H. et al. MINK is a Rap2 effector for phosphorylation of the postsynaptic scaffold protein TANC1. Biochem. Biophys. Res. Commun. 377, 573–8 (2008). 57. Guo, Z. et al. Cloning and characterization of the human gene RAP2C, a novel member of Ras family, which activates transcriptional activities of SRE. Mol. Biol. Rep. 34, 137–44 (2007). 58. Diederen, B. M. W. Legionella spp. and Legionnaires’ disease. J. Infect. 56, 1– 12 (2008). 59. Müller, M. P. et al. The Legionella effector protein DrrA AMPylates the membrane traffic regulator Rab1b. Science 329, 946–9 (2010). 60. Luitz, M. P., Bomblies, R., Ramcke, E., Itzen, A. & Zacharias, M. Adenylylation of Tyr77 stabilizes Rab1b GTPase in an active state: A molecular dynamics simulation analysis. Sci. Rep. 6, 19896 (2016). 61. Dianin, A. P. Condensation of ketones with phenols. J. Russ. Phys. Chem. Soc 488–517 (1891). 62. Dodds, E. C. & Lawson, W. Synthetic Estrogenic Agents without the Phenanthrene Nucleus. Nature 137, 996–996 (1936). 63. Rogers, J. A., Metz, L. & Yong, V. W. Review: Endocrine disrupting chemicals and immune responses: a focus on bisphenol-A and its potential mechanisms. Mol. Immunol. 53, 421–30 (2013). 64. Cotter, R. J. Engineering Plastics: A Handbook of Polyarylethers. (Taylor & Francis, 1995). 65. Vom Saal, F. S., Nagel, S. C., Coe, B. L., Angle, B. M. & Taylor, J. A. The estrogenic endocrine disrupting chemical bisphenol A (BPA) and obesity. Mol. Cell. Endocrinol. 354, 74–84 (2012). 66. Vandenberg, L. N., Hauser, R., Marcus, M., Olea, N. & Welshons, W. V. Human

exposure to bisphenol A (BPA). Reprod. Toxicol. Elmsford NY 24, 139–177 (2007). 67. Ribeiro-Varandas, E., Viegas, W., Sofia Pereira, H. & Delgado, M. Bisphenol A at concentrations found in human serum induces aneugenic effects in endothelial cells. Mutat. Res. 751, 27–33 (2013). 68. Matsumoto, A. et al. Bisphenol A levels in human urine. Environ. Health Perspect. 111, 101–4 (2003). 69. Ben-Jonathan, N., Hugo, E. R. & Brandebourg, T. D. Effects of bisphenol A on adipokine release from human adipose tissue: Implications for the metabolic syndrome. Mol. Cell. Endocrinol. 304, 49–54 (2009). 70. Arbuckle, T. E. et al. Exposure to free and conjugated forms of bisphenol A and triclosan among pregnant women in the MIREC cohort. Environ. Health Perspect. 123, 277–84 (2015). 71. Pivnenko, K., Pedersen, G. A., Eriksson, E. & Astrup, T. F. Bisphenol A and its structural analogues in household waste paper. Waste Manag. 44, 39–47 (2015). 72. Fiege, H. et al. in Ullmann’s Encycl. Ind. Chem. (Wiley-VCH Verlag GmbH & Co. KGaA, 2000). doi:10.1002/14356007.a19_313 73. Kinch, C. D., Ibhazehiebo, K., Jeong, J.-H., Habibi, H. R. & Kurrasch, D. M. Low- dose exposure to bisphenol A and replacement bisphenol S induces precocious hypothalamic neurogenesis in embryonic zebrafish. Proc. Natl. Acad. Sci. 112, 201417731 (2015). 74. Midoro-Horiuti, T., Tiwari, R., Watson, C. S. & Goldblum, R. M. Maternal Bisphenol A Exposure Promotes the Development of Experimental Asthma in Mouse Pups. Environ. Health Perspect. 118, 273–277 (2009). 75. Rubin, B. S. & Soto, A. M. Bisphenol A: Perinatal exposure and body weight. Mol. Cell. Endocrinol. 304, 55–62 (2009). 76. Vandenberg, L. N. et al. Hormones and endocrine-disrupting chemicals: low- dose effects and nonmonotonic dose responses. Endocr. Rev. 33, 378–455 (2012). 77. Delfosse, V. et al. Structural and mechanistic insights into bisphenols action provide guidelines for risk assessment and discovery of bisphenol A substitutes. Proc. Natl. Acad. Sci. U. S. A. 109, 14930–5 (2012). 78. EFSA. Scientific Opinion on the risks to public health related to the presence of

bisphenol A (BPA) in foodstuffs. EFSA J. 13, (2015). 79. Edginton, A. N. & Ritter, L. Predicting plasma concentrations of bisphenol A in children younger than 2 years of age after typical feeding schedules, using a physiologically based toxicokinetic model. Environ. Health Perspect. 117, 645– 652 (2009). 80. European Chemicals Agency. Current SVHC intentions - ECHA. (2016). 81. Kuriyan, J., Konforti, B. & Wemmer, D. The molecules of life : physical and chemical principles. 82. Bodenhausen, G. & Ruben, D. J. Natural abundance nitrogen-15 NMR by enhanced heteronuclear spectroscopy. Chem. Phys. Lett. 69, 185–189 (1980). 83. Pervushin, K., Riek, R., Wider, G. & Wüthrich, K. Attenuated T2 relaxation by mutual cancellation of dipole-dipole coupling and chemical shift anisotropy indicates an avenue to NMR structures of very large biological macromolecules in solution. Proc. Natl. Acad. Sci. U. S. A. 94, 12366–12371 (1997). 84. Fiaux, J., Bertelsen, E. B., Horwich, A. L. & Wüthrich, K. NMR analysis of a 900K GroEL GroES complex. Nature 418, 207–11 (2002). 85. Sprangers, R. & Kay, L. E. Quantitative dynamics and binding studies of the 20S proteasome by NMR. Nature 445, 618–622 (2007). 86. Shuker, S. B. et al. Discovering high-affinity ligands for proteins: SAR by NMR. Science (80-. ). 274, 1531–1534 (1996). 87. Petros, A. M. et al. Discovery of a potent and selective Bcl-2 inhibitor using SAR by NMR. Bioorg. Med. Chem. Lett. 20, (2010). 88. Fernandez-Medarde, A. & Santos, E. Ras in Cancer and Developmental Diseases. Genes Cancer 2, 344–358 (2011). 89. Cooper, G. M. Cellular transforming genes. Science 217, 801–6 (1982). 90. Santos, E., Tronick, S. R., Aaronson, S. A., Pulciani, S. & Barbacid, M. T24 human bladder carcinoma oncogene is an activated form of the normal human homologue of BALB- and Harvey-MSV transforming genes. Nature 298, 343– 347 (1982). 91. Parada, L. F., Tabin, C. J., Shih, C. & Weinberg, R. A. Human EJ bladder carcinoma oncogene is homologue of Harvey sarcoma virus ras gene. Nature 297, 474–478 (1982). 92. Taparowsky, E. et al. Activation of the T24 bladder carcinoma transforming gene is linked to a single amino acid change. Nature 300, 762–765 (1982).

93. Pai, E. F. et al. Refined crystal structure of the triphosphate conformation of H- ras p21 at 1.35 A resolution: implications for the mechanism of GTP hydrolysis. EMBO J. 9, 2351–9 (1990). 94. Scheffzek, K. et al. The Ras-RasGAP Complex: Structural Basis for GTPase Activation and Its Loss in Oncogenic Ras Mutants. Science (80-. ). 277, 333– 338 (1997). 95. Boriack-Sjodin, P. . A., Margarit, S. . M., Bar-Sagi, D. & Kuriyan, J. The structural basis of the activation of Ras by Sos. Nature 394, 337–343 (1998). 96. Liao, C., Liu, F. & Kannan, K. Bisphenol s, a new bisphenol analogue, in paper products and currency bills and its association with bisphenol a residues. Environ. Sci. Technol. 46, 6515–22 (2012). 97. Glausiusz, J. Toxicology: The plastics puzzle. Nature 508, 306–8 (2014). 98. Hashimoto, Y. et al. Measurement of estrogenic activity of chemicals for the development of new dental polymers. Toxicol. Vitr. 15, 421–425 (2001). 99. Grignard, E., Lapenna, S. & Bremer, S. Weak estrogenic transcriptional activities of Bisphenol A and Bisphenol S. Toxicol. Vitr. an Int. J. Publ. Assoc. with BIBRA 26, 727–31 (2012). 100. Davalos, J. Z., Herrero, R., Costa, J., Santos, L. M. N. B. F. & Liebman, J. F. Energetic and structural study of bisphenols. J. Phys. Chem. A 118, 3705–3709 (2014). 101. Böhm, H.-J. et al. Fluorine in medicinal chemistry. Chembiochem 5, 637–43 (2004). 102. Müller, K., Faeh, C. & Diederich, F. Fluorine in pharmaceuticals: looking beyond intuition. Science 317, 1881–6 (2007). 103. Klages, J. et al. NMR-based screening: a powerful tool in fragment-based drug discovery. Mol. Biosyst. 2, 318 (2006).

Curriculum Vitae

Personal Data

Miriam Schöpel Born on 24 March, 1989 in Bochum [email protected]

Doctorate

5/2013 - 10/2016, Ruhr University Bochum

Education

10/2011 – 4/2013 Master of Biochemistry, Ruhr University Bochum GBM-Price for the best Master degree in Biochemistry 10/2008 – 10/2011 Bachelor of Biochemistry, Ruhr University Bochum 1999 - 2008 “Abitur”, CF-Gauß-Gymnasium Gelsenkirchen

Academic Achievements

Poster

 11-13th March, 2014, Retreat with AG Bayer, Erndtebrück: Bisphenol A binds to Ras proteins and competes with guanine nucleotide exchange ̶ implications for GTPase selective antagonists.

 24-29th Aug, 2014, XXVI ICMRBS in Dallas, TX, USA: Bisphenol A binds to Ras proteins and competes with guanine nucleotide exchange ̶ implications for GTPase selective antagonists. Poster #119, Travel funding by the Research School Bochum

 6th Nov, 2014, Research Day, Research School Bochum, Ruhr University Bochum: Bisphenol A binds to the cancer related protein K-Ras Best poster presentation in life science  11-13th May, 2015, Summer school of the SFB 642, Velen: Bisphenol A analogues bind to Ras proteins and compete with guanine nucleotide exchange

 11th June, 2015, Science Day, Research School Bochum, Ruhr University Bochum: Bisphenol A binds to Ras proteins and competes with guanine nucleotide exchange ̶ implications for GTPase selective antagonists. Poster

Talks

 11-13th May, 2015, Summer school of the SFB 642, Velen: Bisphenol A analogues bind to Ras proteins and compete with guanine nucleotide exchange

Peer-reviewed publications

 Accepted manuscripts

1. Bisphenol A Binds to Ras Proteins and Competes with Guanine Nucleotide Exchange: Implications for GTPase-Selective Antagonists

Schoepel, M.; Jockers, K. F. G.; Dueppe, P. M.; Autzen, J.; Potheraveedu, V. N.; Ince, S.; Yip, K. T.; Heumann, R.; Herrmann, C.; Scherkenbeck, J.; Stoll, R. J. Med. Chem. 2013, 56, 9664–9672. Declaration: Reprinted (adapted) with permission from © Copyright 2013 American Chemical Society, DOI: 10.1021/jm401291q

2. The Bisphenol A Analogue Bisphenol S binds to K-Ras4B - implications for ‘BPA-free’ plastics

Schoepel, M.; Herrmann, C.; Scherkenbeck, J.; Stoll, R. The Bisphenol A Analogue Bisphenol S Binds to K-Ras4B - Implications for “BPA-Free” Plastics. FEBS Lett. 2016, 590, 369–375. Declaration: Reprinted (adapted) with the permission from © Copyright 2016 Wiley, DOI: 10.1002/1873-3468.12056

 Other publications, that are not considered in this thesis:

Sequence-selective molecular recognition of the C-terminal CaaX-boxes of Rheb and related Ras-proteins by synthetic receptors

Düppe, P. M.; Tran Thi Phuong, T.; Autzen, J.; Schoepel, M.; Yip, K. T.; Stoll, R.; Scherkenbeck, J. Sequence-Selective Molecular Recognition of the C-Terminal CaaX-Boxes of Rheb and Related Ras-Proteins by Synthetic Receptors. ACS Chem. Biol. 2014, 9, 1755–1763.

 Submitted manuscripts

Different Bisphenols bind to Ras isoforms and induce a pseudo-active conformation

Miriam Schöpel, Klaus Kock, Xueyin Zhong, Bastian Kohl, Christian Herrmann, Stefanie Löffek, Iris Helfrich, Hagen S. Bachmann, Jürgen Scherkenbeck, and Raphael Stoll

100

Acknowledgements

Thanks to…

Prof. Dr. Stoll for giving me the opportunity to work in his department, being a great Doktorvater, and the introduction to the power of NMR.

Prof. Dr. Herrmann for being my 2nd supervisor and for inducting me into the complex field of small GTPases.

My colleagues of the biomolecular spectroscopy department:

 Bastian Kohl for being a true friend and my biggest help in the lab & spectrometer. Your creativity has always been inspirational to me.  Xueyin Zhong for the crystallisation trials and his friendship.  Stefanie Pütz for helping me in the lab.  Iris Langstein for the support in the final phase of my lab work and her friendship.  Hans Jochen Hauswald for his relentless dedication to the NMR spectrometers.  Thanks to my undergraduate students (in chronical order): Fabian Wendt, Madeline Puschmann, Oliver Arnolds, and Sascha Shkura.

The Members of Herrmanns lab: Miriam Kutsch (Miri II), Semra Ince, Fabian Klümpers, Oktavian Krenczyk, Sergii Sshydlovskyi, and Klaus Kock.

My friends for giving support & love, when I needed it the most. My brothers Clemens (+Katrin) and Martin. My mother for believing in me. My beloved sister Betti (+Julia +Sally) for always standing by my side.