Michael Grabe

Professor

UCSF Cardiovascular Research Institute
Country flagUnited States

Research Interests

ApplyKite TwinLearns as you apply

Make your next move.

Prepare your application

Review drafts before you send them.

Explore related searches

Articles (18)

TMEM16A is a Ca 2+ -activated anion channel that provides direct electrical feedback to the plasma membrane in response to intracellular Ca 2+ . Its conductive state remains unresolved, leaving questions about gating, Cl - permeation, and modulation by Ca 2+ , depolarization, and lipids. To investigate the open state, we performed molecular dynamics simulations of TMEM16A bound to the putative open-state blocker 1PBC. After inhibitor removal, the putative, pore-lining helix TM4 developed kinks at two sites: an upper site that opens the pore for Cl - permeation, and a deeper site causing constriction. A conserved hydrophobic network between TM3 and TM4 persisted in most open structures but separated during extreme dilation, allowing lipids to transiently block the pore. Patch-clamp recordings indicated that the intact network promotes activation. Further simulations yielded >60 Cl - permeation events and a single-channel conductance matching experiments. Additional electrostatic and kinetic modeling indicated that TMEM16A’s transition from outward-rectification to Ohmic conductance with increasing Ca 2+ results from a weak voltage dependence of Ca 2+ binding, which act cooperatively to open the pore. HIGHLIGHTS Removal of the inhibitor 1PBC from TMEM16A induces spontaneous transitions to ion conductive and non-conductive states via bending in TM4 at two different locations. Simulations suggest that the open state is stabilized by a small hydrophobic network between TM3 and 4 and disrupting this network biases channel closure in electrophysiological recordings. A kinetic model of conduction based on energetics from the all-atom MD simulations coupled to continuum calculations gives a linear current-voltage curve consistent with the fully open conformation and removal of 1 Ca 2+ switches to an outwardly rectifying state via electrostatic influence on the Cl - energy profile. However, the rectification is too weak to match experiment, but a cooperative model of Ca 2+ binding with weak voltage-dependence does match experiment.

Year:

2026

Deep learning has advanced the design of static protein structures, but the controlled conformational changes that are hallmarks of natural signaling proteins have remained inaccessible to de novo design. Here, we describe a general deep learning–guided approach for de novo design of dynamic changes between intradomain geometries of proteins, similar to switch mechanisms prevalent in nature, with atomic-level precision. We solve four structures that validate the designed conformations, demonstrate modulation of the conformational landscape by orthosteric ligands and allosteric mutations, and show that physics-based simulations are in agreement with deep-learning predictions and experimental data. Our approach demonstrates that new modes of motion can now be realized through de novo design and provides a framework for constructing biology-inspired, tunable, and controllable protein signaling behavior de novo.

Year:

2025

The inner mitochondrial membrane proteins ATP/ADP carrier protein 1 (AAC1) and Uncoupling protein 1 (UCP1) belong to the SLC25 mitochondrial carrier family. AAC1 is responsible for ATP/ADP exchange, while UCP1-dependent proton transport, which also requires small molecules known as activators, is the basis of brown fat thermogenesis. Arachidonic acid (AA) is an endogenous activator capable of inducing proton transport in both proteins. As such, both AAC1- and UCP1-dependent proton transport are potential targets of weight loss drugs. While AAC1 structures have long been available, only recently have structures of UCP1 been determined. Unfortunately, no AA-bound structure of either protein is available. To explore their interactions with AA, we performed molecular dynamics (MD) simulations of both proteins. Six parallel simulations of each protein were run with an average length of just over 6 μs, for a total of 75 μs of aggregate simulation across both proteins. AA bound deeply between transmembrane helix (TM) helices or in the central cavity of AAC1 in 14 events and between TM helices of UCP1 in 6 events. All AA involved in these deep binding events came from the intermembrane space-facing (C) leaflet. In AAC1, AA most often bound between TM1/TM2 and TM5/TM6. In four cases the fatty acid bound at the bottom of the central cavity rather than in an interhelical groove. In UCP1, all but one deeply bound AA sat between TM5 and TM6. No AA fully entered the cavity as observed in AAC1. In addition to entering the proteins, AAs were enriched around them in the surrounding membrane adjacent to the TM helices. While both protein structures exhibit hydrophobic stretches separating the intermembrane space (IMS) from the matrix, water wires formed through both AAC1 and UCP1, connecting the bulk water in both regions. Grotthuss shuttling along water wires has been proposed as a possible mechanism of AAC1/UCP1-dependent proton transport, but water wires are not present in experimental structures and have not previously been reported in MD simulations. Calculations of electric potentials along these water wires find a large 0.75–1 V electrostatic barrier along water wires through AAC1 and a substantially smaller such barrier of ~0.5 V through UCP1.

Year:

2025

Cryo-electron microscopy (cryoEM) is a powerful tool for atomic- and molecular-resolution structure determination, while molecular dynamics (MD) simulations are similarly powerful tools for predicting molecular trajectories. Given the challenges in estimating biomolecule dynamics with cryoEM alone, MD simulations are employed to forecast molecular motions and to interpret cryoEM reconstructions. Few methods, however, can evaluate MD predictions directly. Here, we use multislice wave propagation to project sampled snapshots of MD trajectories, either coarse-grained (CG) or all-atom (AA), into simulated cryoEM 3D reconstructions. We compared simulated and experimental images of low- and high-curvature membranes to show that MD simulations qualitatively reflect the fluidity and thus the contrast of biological membranes observed by cryoEM. MD simulations also correctly predicted bilayer dimensions for single component flat bilayers observed in cryoEM images. However, Martini3 CG-MD simulations failed to predict changes in membrane thickness induced by high curvature and with heterogeneous lipid compositions. We pinpointed the misbehavior of polyunsaturated lipid tails and cholesterol in Martini3 simulations as the main error sources contributing to inaccurate bilayer thicknesses. Our comparisons also explain membrane structure discrepancies between cryoEM and small angle X-ray scattering (SAXS). Further testing of MD predictions by direct comparisons between simulated and experimental cryoEM images should lead to the development of more accurate MD force fields. Statement of Significance Molecular dynamics (MD) simulations are frequently employed to predict the dynamics of biological macromolecules and assemblies, but these predictions remain difficult to validate experimentally. Cryo-electron microscopy (cryoEM) can be used to directly image the conformational ensemble of macromolecules, but images of Coulombic potential cannot be easily compared to snapshots of atoms from MD simulations. Here, we show that a physics-based multislice image projection algorithm accurately converts MD trajectories to simulated cryoEM 2D images and 3D reconstructions. Using this approach, we identify consistencies and discrepancies between MD simulations and cryoEM experiments. Notably, coarse-grained MD performs poorly compared to all-atom MD when simulating membranes composed of mixtures of lipids that include cholesterol and polyunsaturated lipids, providing observables for MD force field improvement.

Year:

2025

Lysosomes process cellular waste and coordinate responses to metabolic challenge. Central to lysosomal homeostasis are phosphoinositide lipids, key signaling molecules which establish organelle identity, regulate membrane dynamics and are tightly linked to the pathophysiology and therapy of lysosomal storage disorders, neurodegeneration, and cancer. Phosphatidylinositol 3,5- bisphosphate (PI(3,5)P2) interacts with multiple lysosomal membrane proteins and plays a critical role in regulating lysosomal pH by directly inhibiting the chloride/proton antiporter ClC-7, though the molecular mechanism of this inhibition remains unclear. Here, using a combination of functional, structural, and computational analysis, we demonstrate that PI(3,5)P2 binding dramatically remodels the structure of ClC-7 by inducing close association between cytosolic and transmembrane domains. Disease-causing mutations show increased transport activity through loss of PI(3,5)P2 binding and subsequent inhibition. Conversely, ClC-7 activation is correlated with dissociation and increased disorder of the cytoplasmic domain along with novel transmembrane domain conformations, revealing a mechanistic link between specific lysosomal lipids, transporter regulation, and the enigmatic basis of the ClC-7 slow gate.

Year:

2025

Open Reading Frame 9b (Orf9b), an accessory protein of SARS-CoV and –2, is involved in innate immune suppression through its binding to the mitochondrial receptor Translocase of Outer Membrane 70 (Tom70). Previous structural studies of Orf9b in isolation revealed a β-sheet-rich homodimer; however, structures of Orf9b in complex with Tom70 revealed a monomeric helical fold. Here, we developed a biophysical model that quantifies how Orf9b switches between these conformations and binds to Tom70, a requirement for suppressing the type 1 interferon response. We used this model to characterize the effect of lipid binding and mutations in variants of concern to the Orf9b:Tom70 equilibrium. We found that the binding of a lipid to the Orf9b homodimer biases the Orf9b monomer:dimer equilibrium towards the dimer by reducing the dimer dissociation rate ~100 fold. We also found that mutations in variants of concern can alter different microscopic rate constants without significantly affecting binding to Tom70. Together, our results highlight how perturbations to different steps in these coupled equilibria can affect the apparent affinity of Orf9b to Tom70, with potential downstream implications for interferon signaling in coronavirus infection.

Year:

2025

Plant survival requires cellular sensing systems that detect and respond to nutrient and ion fluctuations. The ability to monitor and adjust solute concentrations is essential for managing environmental stresses, particularly salt stress, yet the molecular mechanisms underlying cellular ion sensing remain largely unknown. Here, we combine luminal pH measurements with mathematical modelling of ion transport to investigate the role of the Trans-Golgi Network/Early Endosome (TGN/EE) for cellular ion homeostasis. Our results demonstrate that cytosolic concentrations of Cl - , Na + and K + directly impact ion transport activity at the TGN/EE, thereby influencing luminal pH dynamics. Specifically, we find that the TGN/EE lumen becomes more alkaline when plants are exposed to elevated NaCl concentrations, indicating a previously unrecognised role for endomembrane pH regulation in salt stress responses. Furthermore, we demonstrate that TGN/EE ion transport mutants that fail to exhibit NaCl-induced alkalisation display hypersensitivity to Na + . These findings indicate that the translation of elevated cytosolic ion concentrations into TGN/EE luminal alkalisation, represents an important mechanism for conferring Na + tolerance in plant cells.

Year:

2025

Open Reading Frame 9b (Orf9b), an accessory protein of SARS-CoV and -2, is involved in innate immune suppression through its binding to the mitochondrial receptor Translocase of Outer Membrane 70 (Tom70). Previous structural studies of Orf9b in isolation revealed a β-sheet-rich homodimer, however, structures of Orf9b in complex with Tom70 revealed a monomeric helical fold. Here, we developed a biophysical model that quantifies how Orf9b switches between these conformations and binds to Tom70, a requirement for suppressing the type 1 interferon response. We used this model to characterize the effect of lipid binding and mutations in variants of concern to the Orf9b:Tom70 equilibrium. We found that the binding of a lipid to the Orf9b homodimer biases the Orf9b monomer:dimer equilibrium towards the dimer by reducing the dimer dissociation rate ∼100-fold. We also found that mutations in variants of concern can alter different microscopic rate constants without significantly affecting binding to Tom70. Together our results highlight how perturbations to different steps in these coupled equilibria can affect the apparent affinity of Orf9b to Tom70, with potential downstream implications for interferon signaling in coronavirus infection.

Year:

2025

Biological membranes are complex and dynamic structures with different populations of lipids in their inner and outer leaflets. The Ca 2+ -activated TMEM16 family of membrane proteins plays an important role in collapsing this asymmetric lipid distribution by spontaneously and bidirectionally scrambling phospholipids between the two leaflets, which can initiate signaling and alter the physical properties of the membrane. While evidence shows that lipid scrambling can occur via an open hydrophilic pathway (groove) that spans the membrane, it remains unclear if all family members facilitate lipid movement in this manner. Here, we present a comprehensive computational study of lipid scrambling by all TMEM16 members with experimentally solved structures. We performed coarse-grained molecular dynamics (MD) simulations of 27 structures from five different family members solved under activating and non-activating conditions, and we captured over 700 scrambling events in aggregate. This enabled us to directly compare scrambling rates, mechanisms, and protein–lipid interactions for fungal and mammalian TMEM16s, in both open (Ca 2+ -bound) and closed (Ca 2+ -free) conformations with statistical rigor. We show that all TMEM16 structures thin the membrane and that the majority of scrambling (>90%) occurs at the groove only when TM4 and TM6 have sufficiently separated. Surprisingly, we also observed 60 scrambling events that occurred outside the canonical groove, over 90% of which took place at the dimer–dimer interface in mammalian TMEM16s. This new site suggests an alternative mechanism for lipid scrambling in the absence of an open groove.

Year:

2025

The asymmetric resting distribution of the three major phospholipid classes on the mammalian plasma membrane, with phosphatidylserine and phosphatidylethanolamine mostly on the inner leaflet, and phosphatidylcholine mostly on the outer leaflet, is maintained by ATP-dependent flippases and floppases that exhibit headgroup selectivity. Upon signaling cues, this asymmetry can be dissipated by various phospholipid scramblases, allowing cells to respond to stimuli and adapt to different physiological contexts. The prevailing view in the field is that phospholipid scramblases on the plasma membrane act without headgroup preference. Here we report contrary experimental evidence based on a phospholipid scrambling assay that quantifies the fluorescence polarization of nitrobenzoxadiazole-labeled phospholipids for kinetic monitoring of phospholipid scrambling on the plasma membrane of living cells. Our experiments reveal that the plasma membrane-residing calcium-activated phospholipid scramblase TMEM16F preferentially acts on phosphatidylserine and phosphatidylcholine over phosphatidylethanolamine.

Year:

2025

Year:

2025

Year:

2025

Year:

2025

Year:

2025

Year:

2025

Year:

2025

Year:

2024

Year:

2024

Collaborators (17)

Susan M. Lea

Member and Endowed Chair

St. Jude Children's Research Hospital

UNITED STATES
View details

Brian K. Shoichet

University of California, San Francisco

UNITED STATES
View details

Nevan Krogan

Professor

University of California, San Francisco

UNITED STATES
View details

Simon Newstead

David Phillips Professor of Molecular Biophysics

University of Oxford

UNITED KINGDOM
View details

Colin Ophus

Associate Professor

Stanford University

UNITED STATES
View details

Karin Schumacher

Professor

Ruprecht Karls Universität Heidelberg

GERMANY
View details

Monita Muralidharan

University of California, San Francisco

UNITED STATES
View details

William F. DeGrado

Professor

University of California

UNITED STATES
View details

Danielle Swaney

University of California, San Francisco

UNITED STATES
View details

David Crottès

Associate professor

Université de Tours

FRANCE
View details

Stefanie Wege

Project Leader

University of Bonn

GERMANY
View details

Joanne Parker

University of Oxford

UNITED KINGDOM
View details

Tanja Kortemme

Professor

University of California, San Francisco

UNITED STATES
View details

Lily Jan

professor

University of California, San Francisco

UNITED STATES
View details

Colin Smith

Assistant Professor

Wesleyan University

UNITED STATES
View details

James Fraser

Professor

University of California, San Francisco

UNITED STATES
View details

Yuh Nung Jan

Jack and DeLoris Lange Endowed Chair

University of California, San Francisco

UNITED STATES
View details