Structure and Kinetics of the PTEN Tumor Suppressor: Investigation of Solution and Membrane-Associated States
Cancer is an umbrella term for a class of diseases all characterized by uncontrolled cell growth. In 2008, the average person had a 20% probability of being diagnosed with cancer before the age of 75 and a 11% chance of dying from it . There are two categories of genes whose alteration can result in cancer: 1) oncogenes that promote cell growth and 2) tumor-suppressor genes that inhibit cell division and survival. PTEN is the second-most frequently mutated gene in human cancer  after p53, which in turn is mutated in half of all tumors . While PTEN has a protein phosphatase capacity, it fulfills its cancer protective role through its ability to dephosphorylate the lipid PI(3,4,5)P3, which in turn shuts down the PI3K/Akt signaling pathway thereby downregulating cell growth. Despite its critical role in preventing aberrant cell proliferation, there is no structural information available on membrane-bound PTEN. The crystal structure of a highly truncated membrane-free PTEN mutant was determined , but the deleted N-terminal and C-terminal tails have postulated membrane-association and regulatory roles, respectively.
Since PTEN’s regulatory function has been postulated to be membranemediated, it is crucial to identify the binding mechanism and the contribution of various lipid species to the overall kinetics. In this thesis, we first describe a novel biomimetic construct called a tethered bilayer lipid membrane (tBLM) which allows for the simultaneous characterization of the bilayer by multiple techniques while allowing for an exquisite control of lipid composition. To validate the biological relevance of tBLMs, we visualized their optical homogeneity using fluorescence microscopy (FM), quantified the defect-density using electrochemical impedance spectroscopy (EIS) and the lipid diffusivity using two-photon fluorescence correlation spectroscopy (2P-FCS). We formulated a protocol that allowed for the preparation of defect-free planar bilayers which were able to reproduce the fluidity of free-standing membranes (such as lipid vesicles) without compromising on long-term stability (the issue with black lipid membranes) while decoupling the proximal (to the substrate) leaflet from the distal leaflet (the issue with solid-supported lipid bilayers).
We then proceeded to quantify the binding affinities of wt PTEN, an autismrelated mutant H93R PTEN, a Cowden syndrome-related mutant C124S PTEN and the truncated crystal structure PTEN mutant to the anionic lipids phosphatidylserine (PS), phosphatidylinositol (4,5)-bisphosphate [PI(4,5)P2] and phosphatidylinositol (3,4,5)-trisphosphate [PI(3,4,5)P3] individually, as well as to membranes composed of biologically-relevant mixtures of these lipids. We show that the membrane-association of PTEN is very sensitive to lipid composition. It had earlier been determined that PTEN has a very weak affinity for the zwitterionic phosphatidylcholine (PC) lipids [Kd > 500 μM] . wt PTEN is initially attracted to the membrane through interactions with PS lipids [Kd=22.0±0.5 μM; Bmax=155±3 ng/cm2]. This is then followed by specific association with PI(4,5)P2 lipids [Kd=0.4±0.1 μM; Bmax=23±1 ng/cm2] which allows for the efficient dephosphorylation of the PI(3,4,5)P3 catalytic substrate [Kd=2.4±0.2 μM]. We also observed that wt PTEN shows an order of magnitude stronger affinity to membranes containing both PS and PI(4,5)P2 [Kd=0.04±0.01 μM] indicating a cooperative binding effect. The H93R PTEN mutation is spatially separated from the PI(4,5)P2-binding module (PBM), the CBR3 PS-binding motif in the C2 domain as well as the active site. Yet, it exhibits a 4-fold strengthened affinity to PS-containing membranes [Kd(H93R)=3.1±0.3 μM], a 3-fold weakened affinity to PI(4,5)P2-containing membranes [Kd(H93R)=1.3±0.2 μM] and a 75% reduced phosphatase activity  with respect to the wt. The catalytically inactive C124S mutation allows us to show that PTEN’s association with PI(4,5)P2 and PI(3,4,5)P3 is independent and non-competitive [Kd(PI(4,5)P2) = 0.32±0.03 μM; Kd(PI(3,4,5)P3)=0.12±0.03 μM; [Kd(PI(4,5)P2:PI(3,4,5)P3=1:1) = 0.13±0.01 μM], indicating distinct binding sites. The truncated PTEN mutant has a stronger association to PS lipids [2.5≤Kd(truncated)≤4.9 μM] compared to wt PTEN due to the increase in net charge of the protein by +14 as a result of the deletions and the loss of any unfavorable interactions between the tail and the body of the protein. We also observe a two-fold decreased affinity to PI(4,5)P2-bearing membranes compared to wt PTEN [Kd(truncated)=0.77±0.07 μM], likely due to the absence of six residues from the PI(4,5)P2 binding module (PBM).
The SPR binding measurements serve a dual purpose of quantifying PTEN’s membrane association as well as identifying suitable conditions for performing neutron reflectivity (NR) measurements to determine the structure of membrane-bound PTEN. We studied four systems: H93R PTEN bound to a PS-bearing membrane, wt bound to a PS-bearing membrane, wt PTEN bound to a PS+PI(4,5)P2-bearing membrane and the truncated PTEN mutant bound to a PS+PI(4,5)P2-bearing membrane. The NR data was fit using the conventional slab/box model  as well as the new continuous distribution model that was recently developed by our group . All four protein neutron scattering length density (nSLD) profiles are distinct, implying unique membrane-bound states. Both wt PTEN bound states show a 60 ˚A extension of the protein along the bilayer normal while the H93R bound state is more compact at an extension of just 45 ˚A. We suggest that this is primarily due to the C-terminal tail being located distal to the membrane for the wt PTEN, unlike for H93R PTEN, although the H93R point mutation could also result in a conformational change in the core domains of the protein. In all cases, there is minimal penetration of the protein into the lipid headgroups indicating an interfacial association of PTEN with the lipid bilayer
We estimated the orientation of membrane-bound PTEN using the SASSIE  conformational generator in combination with Euler angle rotational analysis. Assuming that the core domains of the protein are unchanged from the crystal structure, wt PTEN binds to PS at an angle given by (θ, ϕ) = (30◦, 30◦) while wt PTEN binds to membranes containing both PS and PI(4,5)P2 at an angle given by (θ, ϕ) = (10◦, 300◦) where (θ, ϕ) = (0◦, 0◦) corresponds to the proposed membrane binding orientation of the protein, as predicted by the crystal structure . This analysis fails when applied to the H93R PTEN NR data, likely indicating a deviation in the secondary structure of the mutant from the crystal structure.
Finally, we performed complementary all-atom molecular dynamic (MD) simulations which allowed us to study the molecular-level details of PTEN’s equilibrium conformation(s), both in solution as well as in a membrane-bound state, while using the experimental results as a source of validation. The association of PTEN with a PS-bearing membrane results in a conformational change of the protein which provides the active site with easier access to the PI(3,4,5)P3 catalytic substrate. The C-terminal tail of membrane-bound PTEN is in a relatively compact conformation and is located distal to the membrane, making minimal contacts with the body of the protein, as suggested by the NR data. However, the tail is extended in solution, allowing it to associate with the CBR3 PS-binding motif of the C2 domain, thereby obstructing membrane association. While this is only one of the conformations that the tail can adopt, the PEST phosphorylation sites on the C-terminal tail are spatially adjacent to Lysines on the C2 domain. Consequently, multiple phosphorylations of the C-terminal tail could lock the protein in a ‘closed’ state where the tail interacts with the PS-binding sites, thereby excluding the possibility of membrane-association . This implies the phosphorylation of the C-terminal tail is a plausible mechanism for PTEN regulation.
In combination, these results from a broad spectrum of investigations provide an entirely new perspective on the activation and regulation of the PTEN tumor suppressor. The detailed molecular picture that arises is urgently needed to help define future research investigation into PTEN’s tumor suppressor role and aid in the search for pharmaceutical targets to counteract the adverse impact of PTEN mutations. They also provide a reference structure for a lipid phosphatase in its active state on a thermally disordered, in-plane fluid membrane. We showcase the ability of seemingly innocuous point mutations to disrupt the membrane-association process through a combination of altered interactions and conformational changes. Our data supports a regulatory role for the disordered C-terminal tail, based on its ability to defy the structure-function paradigm by interfering with PTEN’s ability to bind to the lipid membrane, thereby reducing its catalytic activity.