Identifying the minimal sets of distance restraints for FRET-assisted protein structural modeling.

Proteins naturally occur in crowded cellular environments and interact with other proteins, nucleic acids, and organelles. Since most previous experimental protein structure determination techniques require that proteins occur in idealized, non-physiological environments, the effects of realistic cellular environments on protein structure are largely unexplored. Recently, Förster resonance energy transfer (FRET) has been shown to be an effective experimental method for investigating protein structure in vivo. Inter-residue distances measured in vivo can be incorporated as restraints in molecular dynamics (MD) simulations to model protein structural dynamics in vivo. Since most FRET studies only obtain inter-residue separations for a small number of amino acid pairs, it is important to determine the minimum number of restraints in the MD simulations that are required to achieve a given root-mean-square deviation (RMSD) from the experimental structural ensemble. Further, what is the optimal method for selecting these inter-residue restraints? Here, we implement several methods for selecting the most important FRET pairs and determine the number of pairs Nr that are needed to induce conformational changes in proteins between two experimentally determined structures. We find that enforcing only a small fraction of restraints, Nr/N≲0.08, where N is the number of amino acids, can induce the conformational changes. These results establish the efficacy of FRET-assisted MD simulations for atomic scale structural modeling of proteins in vivo. Significance: Determining protein structure in vivo is essential for understanding protein function. Most protein structures have been studied in non-physiological conditions using x-ray crystallography, NMR spectroscopy, and cryo-electron microscopy. Thus, we do not know whether the cellular environment significantly affects protein structure. We emphasize the benefits of FRET-assisted molecular dynamics simulations in characterizing protein structure in vivo at the atomic scale. We identify the minimum number of FRET pairs that can induce conformational changes in several proteins, including one that has been characterized using in-cell NMR.

Oleic acid differentially affects lipid droplet storage of de novo synthesized lipids in hepatocytes and adipocytes.

Lipogenesis is a vital but often dysregulated metabolic pathway. Here we use optical photothermal infrared imaging to quantify lipogenesis rates of isotopically labelled oleic acid and glucose concomitantly in live cells. In hepatocytes, but not adipocytes, we find that oleic acid feeding at 60 µM increases the number and size of lipid droplets (LDs) while simultaneously inhibiting storage of de novo synthesized lipids in LDs. Our results demonstrate alternate regulation of lipogenesis between cell types.

Oleic acid differentially affects de novo lipogenesis in adipocytes and hepatocytes.

Lipogenesis is a vital but often dysregulated metabolic pathway. We report super-resolution multiplexed vibrational imaging of lipogenesis rates and pathways using isotopically labelled oleic acid and glucose as probes in live adipocytes and hepatocytes. These findings suggest oleic acid inhibits de novo lipogenesis (DNL), but not total lipogenesis, in hepatocytes. No significant effect is seen in adipocytes. These differential effects may be due to alternate regulation of DNL between cell types and could help explain the complicated role oleic acid plays in metabolism.

Site-Specific Crosslinking Reveals Phosphofructokinase-L Inhibition Drives Self-Assembly and Attenuation of Protein Interactions.

Phosphofructokinase is the central enzyme in glycolysis and constitutes a highly regulated step. The liver isoform (PFKL) compartmentalizes during activation and inhibition in vitro and in vivo respectively. Compartmentalized PFKL is hypothesized to modulate metabolic flux consistent with its central role as the rate limiting step in glycolysis. PFKL tetramers self-assemble at two interfaces in the monomer (interface 1 and 2), yet how these interfaces contribute to PFKL compartmentalization and drive protein interactions remains unclear. Here, we used site-specific incorporation of noncanonical photocrosslinking amino acids to identify PFKL interactors at interface 1, 2, and the active site. Tandem mass tag-based quantitative interactomics reveals interface 2 as a hotspot for PFKL interactions, particularly with cytoskeletal, glycolytic, and carbohydrate derivative metabolic proteins. Furthermore, PFKL compartmentalization into puncta was observed in human cells using citrate inhibition. Puncta formation attenuated crosslinked protein-protein interactions with the cytoskeleton at interface 2. This result suggests that PFKL compartmentalization sequesters interface 2, but not interface 1, and may modulate associated protein assemblies with the cytoskeleton.

Site-specific crosslinking reveals Phosphofructokinase-L inhibition drives self-assembly and attenuation of protein interactions.

Phosphofructokinase is the central enzyme in glycolysis and constitutes a highly regulated step. The liver isoform (PFKL) compartmentalizes during activation and inhibition in vitro and in vivo, respectively. Compartmentalized PFKL is hypothesized to modulate metabolic flux consistent with its central role as the rate limiting step in glycolysis. PFKL tetramers self-assemble at two interfaces in the monomer (interface 1 and 2), yet how these interfaces contribute to PFKL compartmentalization and drive protein interactions remains unclear. Here, we used site-specific incorporation of noncanonical photocrosslinking amino acids to identify PFKL interactors at interface 1, 2, and the active site. Tandem mass tag-based quantitative interactomics reveals interface 2 as a hotspot for PFKL interactions, particularly with cytoskeletal, glycolytic, and carbohydrate derivative metabolic proteins. Furthermore, PFKL compartmentalization into puncta was observed in human cells using citrate inhibition. Puncta formation attenuated crosslinked protein-protein interactions with the cytoskeleton at interface 2. This result suggests that PFKL compartmentalization sequesters interface 2, but not interface 1, and may modulate associated protein assemblies with the cytoskeleton.

Chemical interactions modulate λ6-85 stability in cells.

Because steric crowding is most effective when the crowding agent is similar in size to the molecule that it acts upon and the average macromolecule inside cells is much larger than a small protein or peptide, steric crowding is not predicted to affect their folding inside cells. On the other hand, chemical interactions should perturb in-cell structure and stability because they arise from interactions between the surface of the small protein or peptide and its environment. Indeed, previous in vitro measurements of the λ-repressor fragment, λ6-85 , in crowding matrices comprised of Ficoll or protein crowders support these predictions. Here, we directly quantify the in-cell stability of λ6-85 and distinguish the contribution of steric crowding and chemical interactions to its stability. Using a FRET-labeled λ6-85 construct, we find that the fragment is stabilized by 5°C in-cells compared to in vitro. We demonstrate that this stabilization cannot be explained by steric crowding because, as anticipated, Ficoll has no effect on λ6-85 stability. We find that the in-cell stabilization arises from chemical interactions, mimicked in vitro by mammalian protein extraction reagent (M-PER™). Comparison between FRET values in-cell and in Ficoll confirms that U-2 OS cytosolic crowding is reproduced at macromolecule concentrations of 15% w/v. Our measurements validate the cytomimetic of 15% Ficoll and 20% M-PER™ that we previously developed for protein and RNA folding studies. However, because the in-cell stability of λ6-85 is reproduced by 20% v/v M-PER™ alone, we predict that this simplified mixture could be a useful tool to predict the in-cell behaviors of other small proteins and peptides.

Spatiotemporal Heterogeneity of De Novo Lipogenesis in Fixed and Living Single Cells.

De novo lipogenesis (DNL) is a critical metabolic process that provides the majority of lipids for adipocyte and liver tissue. In cancer, obesity, type II diabetes, and nonalcoholic fatty liver disease DNL becomes dysregulated. A deeper understanding of the rates and of subcellular organization of DNL is necessary for identifying how this dysregulation occurs and varies across individuals and diseases. However, DNL is difficult to study inside the cell because labeling lipids and their precursors is not trivial. Existing techniques either can only measure parts of DNL, like glucose uptake, or do not provide spatiotemporal resolution. Here, we track DNL in space and time as isotopically labeled glucose is converted to lipids in adipocytes using optical photothermal infrared microscopy (OPTIR). OPTIR provides submicron resolution infrared imaging of the glucose metabolism in both living and fixed cells while also reporting on the identity of lipids and other biomolecules. We show significant incorporation of the labeled carbons into triglycerides in lipid droplets over the course of 72 h. Live cells had better preservation of lipid droplet morphology, but both showed similar DNL rates. Rates of DNL, as measured by the ratio of 13C-labeled lipid to 12C-labeled lipid, were heterogeneous, with differences within and between lipid droplets and from cell to cell. The high rates of DNL measured in adipocyte cells match upregulated rates of DNL previously reported in PANC1 pancreatic cancer cells. Taken together, our findings support a model where DNL is locally regulated to meet energy needs within cells.

An in Vitro Cytomimetic of In-Cell RNA Folding.

To discover the cytomimetic that accounts for cytoplasmic crowding and sticking on RNA stability, we conducted a two-dimensional scan of mixtures of artificial crowding and sticking agents, PEG10k and M-PERTM . As our model RNA, we investigate the fourU RNA thermometer motif of Salmonella, a hairpin-structured RNA that regulates translation by unfolding and exposing its ribosome binding site (RBS) in response to temperature perturbations. We found that the addition of artificial crowding and sticking agents leads to a stabilization and destabilization of RNA folding, respectively, through the excluded volume effect and surface interactions. FRET-labels were added to the fourU RNA and Fast Relaxation Imaging (FReI), fluorescence microscopy coupled to temperature-jump spectroscopy, probed differences between folding stability of RNA inside single living cells and in vitro. Our results suggest that the cytoplasmic environment affecting RNA folding is comparable to a combination of 20 % v/v M-PERTM and 150 g/L PEG10k.

Spliceosomal SL1 RNA binding to U1-70K: the role of the extended RRM.

The RNA recognition motif (RRM) occurs widely in RNA-binding proteins, but does not always by itself support full binding. For example, it is known that binding of SL1 RNA to the protein U1-70K in the U1 spliceosomal particle is reduced when a region flanking the RRM is truncated. How the RRM flanking regions that together with the RRM make up an 'extended RRM' (eRRM) contribute to complex stability and structural organization is unknown. We study the U1-70K eRRM bound to SL1 RNA by thermal dissociation and laser temperature jump kinetics; long-time molecular dynamics simulations interpret the experiments with atomistic resolution. Truncation of the helix flanking the RRM on its N-terminal side, 'N-helix,' strongly reduces overall binding, which is further weakened under higher salt and temperature conditions. Truncating the disordered region flanking the RRM on the C-terminal side, 'C-IDR', affects the local binding site. Surprisingly, all-atom simulations show that protein truncation enhances base stacking interactions in the binding site and leaves the overall number of hydrogen bonds intact. Instead, the flanking regions of the eRRM act in a distributed fashion via collective interactions with the RNA when external stresses such as temperature or high salt mimicking osmotic imbalance are applied.

Cellular Sticking Can Strongly Reduce Complex Binding by Speeding Dissociation.

While extensive studies have been carried out to determine protein-RNA binding affinities, mechanisms, and dynamics in vitro, such studies do not take into consideration the effect of the many weak nonspecific interactions in a cell filled with potential binding partners. Here we experimentally tested the role of the cellular environment on affinity and binding dynamics between a protein and RNA in living U-2 OS cells. Our model system is the spliceosomal protein U1A and its binding partner SL2 of the U1 snRNA. The binding equilibrium was perturbed by a laser-induced temperature jump and monitored by Förster resonance energy transfer. The apparent binding affinity in live cells was reduced by up to 2 orders of magnitude compared to in vitro. The measured in-cell dissociation rate coefficients were up to 2 orders of magnitude larger, whereas no change in the measured association rate coefficient was observed. The latter is not what would be anticipated due to macromolecular crowding or nonspecific sticking of the uncomplexed U1A and SL2 in the cell. A quantitative model fits our experimental results, with the major cellular effect being that U1A and SL2 sticking to cellular components are capable of binding, just not as strongly as the free complex. This observation suggests that high binding affinities measured or designed in vitro are necessary for proper binding in vivo, where competition with many nonspecific interactions exists, especially for strongly interacting species with high charge or large hydrophobic surface areas.

Cytoskeletal Drugs Modulate Off-Target Protein Folding Landscapes Inside Cells.

The dynamic cytoskeletal network of microtubules and actin filaments can be disassembled by drugs. Cytoskeletal drugs work by perturbing the monomer-polymer equilibrium, thus changing the size and number of macromolecular crowders inside cells. Changes in both crowding and nonspecific surface interactions ("sticking") following cytoskeleton disassembly can affect the protein stability, structure, and function directly or indirectly by changing the fluidity of the cytoplasm and altering the crowding and sticking of other macromolecules in the cytoplasm. The effect of cytoskeleton disassembly on protein energy landscapes inside cells has yet to be observed. Here we have measured the effect of several cytoskeletal drugs on the folding energy landscape of two FRET-labeled proteins with different in vitro sensitivities to macromolecular crowding. Phosphoglycerate kinase (PGK) was previously shown to be more sensitive to crowding, whereas variable major protein-like sequence expressed (VlsE) was previously shown to be more sensitive to sticking. The in-cell effects of drugs that depolymerize either actin filaments (cytochalasin D and latrunculin B) or microtubules (nocodazole and vinblastine) were compared. The crowding sensor protein CrH2-FRET verified that cytoskeletal drugs decrease the extent of crowding inside cells despite also reducing the overall cell volume. The decreased compactness and folding stability of PGK could be explained by the decreased extent of crowding induced by these drugs. VlsE's opposite response to the drugs shows that depolymerization of the cytoskeleton also changes sticking in the cellular milieu. Our results demonstrate that perturbation of the monomer-polymer cytoskeletal equilibrium, for example, during natural cell migration or stresses from drug treatment, has off-target effects on the energy landscapes of proteins in the cell.

An in vitro mimic of in-cell solvation for protein folding studies.

Ficoll, an inert macromolecule, is a common in vitro crowder, but by itself it does not reproduce in-cell stability or kinetic trends for protein folding. Lysis buffer, which contains ions, glycerol as a simple kosmotrope, and mimics small crowders with hydrophilic/hydrophobic patches, can reproduce sticking trends observed in cells but not the crowding. We previously suggested that the proper combination of Ficoll and lysis buffer could reproduce the opposite in-cell folding stability trend of two proteins: variable major protein-like sequence expressed (VlsE) is destabilized in eukaryotic cells and phosphoglycerate kinase (PGK) is stabilized. Here, to discover a well-characterized solvation environment that mimics in-cell stabilities for these two very differently behaved proteins, we conduct a two-dimensional scan of Ficoll (0-250 mg/ml) and lysis buffer (0-75%) mixtures. Contrary to our previous expectation, we show that mixtures of Ficoll and lysis buffer have a significant nonadditive effect on the folding stability. Lysis buffer enhances the stabilizing effect of Ficoll on PGK and inhibits the stabilizing effect of Ficoll on VlsE. We demonstrate that a combination of 150 mg/ml Ficoll and 60% lysis buffer can be used as an in vitro mimic to account for both crowding and non-steric effects on PGK and VlsE stability and folding kinetics in the cell. Our results also suggest that this mixture is close to the point where phase separation will occur. The simple mixture proposed here, based on commercially available reagents, could be a useful tool to study a variety of cytoplasmic protein interactions, such as folding, binding and assembly, and enzymatic reactions. SIGNIFICANCE STATEMENT: The complexity of the in-cell environment is difficult to reproduce in the test tube. Here we validate a mimic of cellular crowding and sticking interactions in a test tube using two proteins that are differently impacted by the cell: one is stabilized and the other is destabilized. This mimic is a starting point to reproduce cellular effects on a variety of protein and biomolecular interactions, such as folding and binding.

Quantifying protein dynamics and stability in a living organism.

As an integral part of modern cell biology, fluorescence microscopy enables quantification of the stability and dynamics of fluorescence-labeled biomolecules inside cultured cells. However, obtaining time-resolved data from individual cells within a live vertebrate organism remains challenging. Here we demonstrate a customized pipeline that integrates meganuclease-mediated mosaic transformation with fluorescence-detected temperature-jump microscopy to probe dynamics and stability of endogenously expressed proteins in different tissues of living multicellular organisms.

Cell Volume Controls Protein Stability and Compactness of the Unfolded State.

Macromolecular crowding is widely accepted as one of the factors that can alter protein stability, structure, and function inside cells. Less often considered is that crowding can be dynamic: as cell volume changes, either as a result of external duress or in the course of the cell cycle, water moves in or out through membrane channels, and crowding changes in tune. Both theory and in vitro experiments predict that protein stability will be altered as a result of crowding changes. However, it is unclear how much the structural ensemble is altered as crowding changes in the cell. To test this, we look at the response of a FRET-labeled kinase to osmotically induced volume changes in live cells. We examine both the folded and unfolded states of the kinase by changing the temperature of the media surrounding the cell. Our data reveals that crowding compacts the structure of its unfolded ensemble but stabilizes the folded protein. We propose that the structure of proteins lacking a rigid, well-defined tertiary structure could be highly sensitive to both increases and decreases in cell volume. Our findings present a possible mechanism for disordered proteins to act as sensors and actuators of cell cycle or external stress events that coincide with a change in macromolecular crowding.

Soluble Zwitterionic Poly(sulfobetaine) Destabilizes Proteins.

The widespread interest in neutral, water-soluble polymers such as poly(ethylene glycol) (PEG) and poly(zwitterions) such as poly(sulfobetaine) (pSB) for biomedical applications is due to their widely assumed low protein binding. Here we demonstrate that pSB chains in solution can interact with proteins directly. Moreover, pSB can reduce the thermal stability and increase the protein folding cooperativity relative to proteins in buffer or in PEG solutions. Polymer-dependent changes in the tryptophan fluorescence spectra of three structurally-distinct proteins reveal that soluble, 100 kDa pSB interacts directly with all three proteins and changes both the local polarity near tryptophan residues and the protein conformation. Thermal denaturation studies show that the protein melting temperatures decrease by as much as ∼1.9 °C per weight percent of polymer and that protein folding cooperativity increases by as much as ∼130 J mol-1 K-1 per weight percent of polymer. The exact extent of the changes is protein-dependent, as some proteins exhibit increased stability, whereas others experience decreased stability at high soluble pSB concentrations. These results suggest that pSB is not universally protein-repellent and that its efficacy in biotechnological applications will depend on the specific proteins used.

Non-Steric Interactions Predict the Trend and Steric Interactions the Offset of Protein Stability in Cells.

Although biomolecules evolved to function in the cell, most biochemical assays are carried out in vitro. In-cell studies highlight how steric and non-steric interactions modulate protein folding and interactions. VlsE and PGK present two extremes of chemical behavior in the cell: the extracellular protein VlsE is destabilized in eukaryotic cells, whereas the cytoplasmic protein PGK is stabilized. VlsE and PGK are benchmarks in a systematic series of solvation environments to distinguish contributions from non-steric and steric interactions to protein stability, compactness, and folding rate by comparing cell lysate, a crowding agent, ionic buffer and lysate buffer with in-cell results. As anticipated, crowding stabilizes proteins, causes compaction, and can speed folding. Protein flexibility determines its sensitivity to steric interactions or crowding. Non-steric interactions alone predict in-cell stability trends, while crowding provides an offset towards greater stabilization. We suggest that a simple combination of lysis buffer and Ficoll is an effective new in vitro mimic of the intracellular environment on protein folding and stability.

Labeling for Quantitative Comparison of Imaging Measurements in Vitro and in Cells.

Qualitative imaging of biomolecular localization and distribution inside cells has revolutionized cell biology. Most of these powerful techniques require modifications to the target biomolecule. Over the past 10 years, these techniques have been extended to quantitative measurements, from in-cell protein folding rates to complex dissociation constants to RNA lifetimes. Such measurements can be affected even when a target molecule is just mildly perturbed by its labels. Here, the impact of labeling on protein (and RNA) structure, stability, and function in cells is discussed via practical examples from the recent literature. General guidelines for selecting and validating modification sites are provided to bring the best from cell biology and imaging to quantitative biophysical experiments inside cells.

Binding, folding and insertion of a ß-hairpin peptide at a lipid bilayer surface: Influence of electrostatics and lipid tail packing.

Antimicrobial peptides (AMPs) act as host defenses against microbial pathogens. Here we investigate the interactions of SVS-1 (KVKVKVKVdPlPTKVKVKVK), an engineered AMP and anti-cancer ß-hairpin peptide, with lipid bilayers using spectroscopic studies and atomistic molecular dynamics simulations. In agreement with literature reports, simulation and experiment show preferential binding of SVS-1 peptides to anionic over neutral bilayers. Fluorescence and circular dichroism studies of a Trp-substituted SVS-1 analogue indicate, however, that it will bind to a zwitterionic DPPC bilayer under high-curvature conditions and folds into a hairpin. In bilayers formed from a 1:1 mixture of DPPC and anionic DPPG lipids, curvature and lipid fluidity are also observed to promote deeper insertion of the fluorescent peptide. Simulations using the CHARMM C36m force field offer complementary insight into timescales and mechanisms of folding and insertion. SVS-1 simulated at an anionic mixed POPC/POPG bilayer folded into a hairpin over a microsecond, the final stage in folding coinciding with the establishment of contact between the peptide's valine sidechains and the lipid tails through a "flip and dip" mechanism. Partial, transient folding and superficial bilayer contact are seen in simulation of the peptide at a zwitterionic POPC bilayer. Only when external surface tension is applied does the peptide establish lasting contact with the POPC bilayer. Our findings reveal the influence of disruption to lipid headgroup packing (via curvature or surface tension) on the pathway of binding and insertion, highlighting the collaborative effort of electrostatic and hydrophobic interactions on interaction of SVS-1 with lipid bilayers.

How does solvation in the cell affect protein folding and binding?

The cellular environment is highly diverse and capable of rapid changes in solute composition and concentrations. Decades of protein studies have highlighted their sensitivity to solute environment, yet these studies were rarely performed in situ. Recently, new techniques capable of monitoring proteins in their natural context within a live cell have emerged. A recurring theme of these investigations is the importance of the often-neglected cellular solvation environment to protein function. An emerging consensus is that protein processes in the cell are affected by a combination of steric and non-steric interactions with this solution. Here we explain how protein surface area and volume changes control these two interaction types, and give recent examples that highlight how even mild environmental changes can alter cellular processes.

A quantitative connection of experimental and simulated folding landscapes by vibrational spectroscopy.

For small molecule reaction kinetics, computed reaction coordinates often mimic experimentally measured observables quite accurately. Although nowadays simulated and measured biomolecule kinetics can be compared on the same time scale, a gap between computed and experimental observables remains. Here we directly compared temperature-jump experiments and molecular dynamics simulations of protein folding dynamics using the same observable: the time-dependent infrared spectrum. We first measured the stability and folding kinetics of the fastest-folding ß-protein, the GTT35 WW domain, using its structurally specific infrared spectrum. The relaxation dynamics of the peptide backbone, ß-sheets, turn, and random coil were measured independently by probing the amide I' region at different frequencies. Next, the amide I' spectra along folding/unfolding molecular dynamics trajectories were simulated by accurate mixed quantum/classical calculations. The simulated time dependence and spectral amplitudes at the exact experimental probe frequencies provided relaxation and folding rates in agreement with experimental observations. The calculations validated by experiment yield direct structural evidence for a rate-limiting reaction step where an intermediate state with either the first or second hairpin is formed. We show how folding switches from a more homogeneous (apparent two-state) process at high temperature to a more heterogeneous process at low temperature, where different parts of the WW domain fold at different rates.