Approximating Projections of Conformational Boltzmann Distributions with AlphaFold2 Predictions: Opportunities and Limitations.

Protein thermodynamics is intimately tied to biological function and can enable processes such as signal transduction, enzyme catalysis, and molecular recognition. The relative free energies of conformations that contribute to these functional equilibria evolved for the physiology of the organism. Despite the importance of these equilibria for understanding biological function and developing treatments for disease, computational and experimental methods capable of quantifying the energetic determinants of these equilibria are limited to systems of modest size. Recently, it has been demonstrated that the artificial intelligence system AlphaFold2 can be manipulated to produce structurally valid protein conformational ensembles. Here, we extend these studies and explore the extent to which AlphaFold2 contact distance distributions can approximate projections of the conformational Boltzmann distributions. For this purpose, we examine the joint probability distributions of inter-residue contact distances along functionally relevant collective variables of several protein systems. Our studies suggest that AlphaFold2 normalized contact distance distributions can correlate with conformation probabilities obtained with other methods but that they suffer from peak broadening. We also find that the AlphaFold2 contact distance distributions can be sensitive to point mutations. Overall, we anticipate that our findings will be valuable as the community seeks to model the thermodynamics of conformational changes in large biomolecular systems.

Rosetta Energy Analysis of AlphaFold2 models: Point Mutations and Conformational Ensembles.

There has been an explosive growth in the applications of AlphaFold2, and other structure prediction platforms, to accurately predict protein structures from a multiple sequence alignment (MSA) for downstream structural analysis. However, two outstanding questions persist in the field regarding the robustness of AlphaFold2 predictions of the consequences of point mutations and the completeness of its prediction of protein conformational ensembles. We combined our previously developed method SPEACH_AF with model relaxation and energetic analysis with Rosetta to address these questions. SPEACH_AF introduces residue substitutions across the MSA and not just within the input sequence. With respect to conformational ensembles, we combined SPEACH_AF and a new MSA subsampling method, AF_cluster, and for a benchmarked set of proteins, we found that the energetics of the conformational ensembles generated by AlphaFold2 correspond to those of experimental structures and explored by standard molecular dynamic methods. With respect to point mutations, we compared the structural and energetic consequences of having the mutation(s) in the input sequence versus in the whole MSA (SPEACH_AF). Both methods yielded models different from the wild-type sequence, with more robust changes when the mutation(s) were in the whole MSA. While our findings demonstrate the robustness of AlphaFold2 in analyzing point mutations and exploring conformational ensembles, they highlight the need for multi parameter structural and energetic analyses of these models to generate experimentally testable hypotheses.

Cucurbit[7]uril Enhances Distance Measurements of Spin-Labeled Proteins.

We report complex formation between the chloroacetamide 2,6-diazaadamantane nitroxide radical (ClA-DZD) and cucurbit[7]uril (CB-7), for which the association constant in water, Ka = 1.9 × 106 M-1, is at least 1 order of magnitude higher than the previously studied organic radicals. The radical is highly immobilized by CB-7, as indicated by the increase in the rotational correlation time, τrot, by a factor of 36, relative to that in the buffer solution. The X-ray structure of ClA-DZD@CB-7 shows the encapsulated DZD guest inside the undistorted CB-7 host, with the pendant group protruding outside. Upon addition of CB-7 to T4 Lysozyme (T4L) doubly spin-labeled with the iodoacetamide derivative of DZD, we observe the increase in τrot and electron spin coherence time, Tm, along with the narrowing of interspin distance distributions. Sensitivity of the DEER measurements at 83 K increases by a factor 4-9, compared to the common spin label such as MTSL, which is not affected by CB-7. Interspin distances of 3 nm could be reliably measured in water/glycerol up to temperatures near the glass transition/melting temperature of the matrix at 200 K, thus bringing us closer to the goal of supramolecular recognition-enabled long-distance DEER measurements at near physiological temperatures. The X-ray structure of DZD-T4L 65 at 1.12 Å resolution allows for unambiguous modeling of the DZD label (0.88 occupancy), indicating an undisturbed structure and conformation of the protein.

Asymmetric conformations and lipid interactions shape the ATP-coupled cycle of a heterodimeric ABC transporter.

Here we used cryo-electron microscopy (cryo-EM), double electron-electron resonance spectroscopy (DEER), and molecular dynamics (MD) simulations, to capture and characterize ATP- and substrate-bound inward-facing (IF) and occluded (OC) conformational states of the heterodimeric ATP binding cassette (ABC) multidrug exporter BmrCD in lipid nanodiscs. Supported by DEER analysis, the structures reveal that ATP-powered isomerization entails changes in the relative symmetry of the BmrC and BmrD subunits that propagates from the transmembrane domain to the nucleotide binding domain. The structures uncover asymmetric substrate and Mg2+ binding which we hypothesize are required for triggering ATP hydrolysis preferentially in one of the nucleotide-binding sites. MD simulations demonstrate that multiple lipid molecules differentially bind the IF versus the OC conformation thus establishing that lipid interactions modulate BmrCD energy landscape. Our findings are framed in a model that highlights the role of asymmetric conformations in the ATP-coupled transport with general implications to the mechanism of ABC transporters.

Generation and characterization of a zebrafish knockout model of abcb4, a homolog of the human multidrug efflux transporter P-glycoprotein.

The ATP-binding cassette subfamily B member 1 (ABCB1), encoding a multidrug transporter referred to as P-glycoprotein (Pgp), plays a critical role in the efflux of xenobiotics in humans and is implicated in cancer resistance to chemotherapy. Therefore, developing high-throughput animal models to screen for Pgp function and bioavailability of substrates and inhibitors is paramount. Here, we generated and validated a zebrafish knockout line of abcb4, a human Pgp transporter homolog. CRISPR/Cas9 genome editing technology was deployed to generate a frameshift mutation in exon 4 of zebrafish abcb4. The zebrafish abcb4 homozygous mutant exhibited elevated accumulation of fluorescent rhodamine 123, a substrate of human Pgp, in the intestine and brain area of embryos. Moreover, abcb4 knockout embryos were sensitized toward toxic compounds such as doxorubicin and vinblastine compared to the WT zebrafish. Immunostaining for zebrafish Abcb4 colocalized in the endothelial brain cells of adult zebrafish. Transcriptome profiling using Gene Set Enrichment Analysis uncovered that the 'cell cycle process,' 'mitotic cell cycles,' and 'microtubule-based process' were significantly downregulated in the abcb4 knockout brain with age. This study establishes and validates the abcb4 knockout zebrafish as an animal model to study Pgp function in vivo. Unexpectedly it reveals a potentially novel role for zebrafish abcb4 in age-related changes in the brain. The zebrafish lines generated here will provide a platform to aid in the discovery of modulators of Pgp function as well as the characterization of human mutants thereof.

Cucurbit[7]uril Enhances Distance Measurements of Spin-Labeled Proteins.

We report complex formation between the chloroacetamide 2,6-diazaadamantane nitroxide radical (ClA-DZD) and cucurbit[7]uril (CB-7), for which the association constant in water, Ka = 1.9 × 106 M-1, is at least one order of magnitude higher than the previously studied organic radicals. The radical is highly immobilized by CB-7, as indicated by the increase of the rotational correlation time, τrot, by a factor of 36, relative to that in the buffer solution. The X-ray structure of ClA-DZD@CB-7 shows the encapsulated DZD guest inside the undistorted CB-7 host, with the pendant group protruding outside. Upon addition of CB-7 to T4 Lysozyme (T4L) doubly spin-labeled with the iodoacetamide derivative of DZD, we observe the increase in τrot and electron spin coherence time, Tm, along with the narrowing of inter-spin distance distributions. Sensitivity of the DEER measurements at 83 K increases by a factor 4 - 9, compared to the common spin label such as MTSL, which is not affected by CB-7. Inter-spin distances of 3-nm could be reliably measured in water/glycerol up to temperatures near the glass transition/melting temperature of the matrix at 200 K, thus bringing us closer to the goal of supramolecular recognition-enabled long-distance DEER measurements at near physiological temperatures. The X-ray structure of DZD-T4L 65 at 1.12 Å resolution allows for unambiguous modeling of the DZD label (0.88 occupancy), indicating undisturbed structure and conformation of the protein.

Determining how MAFA and MAFB transcription factors activity is influenced by structural differences predicted by AlphaFold2.

MAFA and MAFB are related basic-leucine-zipper domain containing transcription factors which have important regulatory roles in a variety of cellular contexts, including pancreatic islet hormone producing α and ß cells. These proteins have similar as well as distinct functional properties, and here we first used AlphaFold2, an artificial intelligence-based structural prediction program, to obtain insight into the three-dimensional organization of their non-DNA binding/dimerization sequences. This analysis was conducted on the wildtype (WT) proteins as well the pathogenic MAFA Ser64Phe (MAFA S64F ) and MAFB Ser70Ala (MAFB S70A ) mutants, with structural differences revealed between MAFA WT and MAFB WT in addition to MAFA S64F and MAFA WT , but not MAFB S70A and MAFB WT . Functional analysis disclosed that the inability to properly phosphorylate at S70 in MAFB S70A , like S65 in MAFA S64F , greatly increased protein stability and enabled MAFB S70A to accelerate cellular senescence in cultured cells. Significant differences were also observed in the ability of MAFA, MAFA S64F , MAFB, and MAFB S70A to cooperatively stimulate Insulin enhancer-driven activity in the presence of other islet-enriched transcription factors. Experiments performed on protein chimeras disclosed that these properties were greatly influenced by structural differences found between the WT and mutant proteins. In general, these results revealed that AlphaFold2 predicts features essential to protein activity.

Approximating conformational Boltzmann distributions with AlphaFold2 predictions.

Protein dynamics are intimately tied to biological function and can enable processes such as signal transduction, enzyme catalysis, and molecular recognition. The relative free energies of conformations that contribute to these functional equilibria are evolved for the physiology of the organism. Despite the importance of these equilibria for understanding biological function and developing treatments for disease, the computational and experimental methods capable of quantifying them are limited to systems of modest size. Here, we demonstrate that AlphaFold2 contact distance distributions can approximate conformational Boltzmann distributions, which we evaluate through examination of the joint probability distributions of inter-residue contact distances along functionally relevant collective variables of several protein systems. Further, we show that contact distance probability distributions generated by AlphaFold2 are sensitive to points mutations thus AF2 can predict the structural effects of mutations in some systems. We anticipate that our approach will be a valuable tool to model the thermodynamics of conformational changes in large biomolecular systems.

Generation and characterization of a zebrafish knockout model of abcb4, a homolog of the human multidrug efflux transporter P-glycoprotein.

The ATP-binding cassette subfamily B member 1 (ABCB1), encoding a multidrug transporter referred to as P-glycoprotein (Pgp), plays a critical role in the efflux of xenobiotics in humans and is implicated in cancer resistance to chemotherapy. Therefore, developing high throughput animal models to screen for Pgp function and bioavailability of substrates and inhibitors is paramount. Here, we generated and validated a zebrafish knockout line of abcb4 , a human Pgp transporter homolog. CRISPR/Cas9 genome editing technology was deployed to generate a frameshift mutation in exon 4 of zebrafish abcb4 . The zebrafish abcb4 homozygous mutant exhibited elevated accumulation of fluorescent rhodamine 123, a substrate of human Pgp, in the intestine and brain area of embryos. Moreover, abcb4 knockout embryos were sensitized toward toxic compounds such as doxorubicin and vinblastine compared to the WT zebrafish. Immuno-staining for zebrafish Abcb4 colocalized in the endothelial brain cells of adult zebrafish. Transcriptome profiling using Gene Set Enrichment Analysis (GSEA) uncovered that the 'cell cycle process,' 'mitotic cell cycles,' and 'microtubule-based process' were significantly downregulated in the abcb4 knockout brain with age. This study establishes and validates the a bcb4 knockout zebrafish as an animal model to study Pgp function in vivo. Unexpectedly it reveals a potentially novel role for zebrafish abcb4 in age-related changes in the brain. The zebrafish lines generated here will provide a platform to aid in the discovery of modulators of Pgp function as well as the characterization of human mutants thereof.

Interplay between Nrf2 and αB-crystallin in the lens and heart of zebrafish under proteostatic stress.

A coordinated oxidative stress response, partly triggered by the transcription factor Nrf2, protects cells from the continual production of reactive oxygen species. Left unbuffered, reactive oxygen species can lead to protein aggregation that has been implicated in a spectrum of diseases such as cataract of the ocular lens and myopathy of the heart. While proteostasis is maintained by diverse families of heat shock proteins, the interplay between the oxidative and proteostatic stress responses in the lens and heart has not been investigated. Capitalizing on multiple zebrafish lines that have compromised function of Nrf2 and/or the two zebrafish small heat shock proteins αBa- and αBb-crystallin, we uncovered a transcriptional relationship that leads to a substantial increase in αBb-crystallin transcripts in the heart in response to compromised function of Nrf2. In the lens, the concomitant loss of function of Nrf2 and αBa-crystallin leads to upregulation of the cholesterol biosynthesis pathway, thus mitigating the phenotypic consequences of the αBa-crystallin knockout. By contrast, abrogation of Nrf2 function accentuates the penetrance of a heart edema phenotype characteristic of embryos of αB-crystallin knockout lines. Multiple molecular pathways, such as genes involved in extracellular interactions and implicated in cardiomyopathy, are revealed from transcriptome profiling, thus identifying novel targets for further investigation. Together, our transcriptome/phenotypic analysis establishes an intersection between oxidative stress and chaperone responses in the lens and heart.

Asymmetric conformations and lipid interactions shape the ATP-coupled cycle of a heterodimeric ABC transporter.

To illuminate the structural origin of catalytic asymmetry of heterodimeric ABC transporters and how it shapes the energetics of their conformational cycles, we used cryo-electron microscopy (cryo-EM), double electron-electron resonance spectroscopy (DEER), and molecular dynamics (MD) simulations, to capture and characterize conformational states of the heterodimeric ABC multidrug exporter BmrCD in lipid nanodiscs. In addition to multiple ATP- and substrate-bound inward-facing (IF) conformations, we obtained the structure of an occluded (OC) conformation wherein the unique extracellular domain (ECD) twists to partially open the extracellular gate. In conjunction with DEER analysis of the populations of these conformations, the structures reveal that ATP-powered isomerization entails changes in the relative symmetry of the BmrC and BmrD subunits that propagates from the transmembrane domain (TMD) to the nucleotide binding domain (NBD). The structures uncover asymmetric substrate and Mg 2+ binding which we hypothesize are required for triggering ATP hydrolysis preferentially in one of the nucleotide-binding sites. MD simulations demonstrated that multiple lipid molecules, identified from the cryo-EM density maps, differentially bind the IF versus the OC conformation thus modulating their relative stability. In addition to establishing how lipid interactions with BmrCD modulate the energy landscape, our findings are framed in a distinct transport model that highlights the role of asymmetric conformations in the ATP-coupled cycle with implications to the mechanism of ABC transporters in general.

Structural analysis of the P132L disease mutation in caveolin-1 reveals its role in the assembly of oligomeric complexes.

Caveolin-1 (CAV1) is a membrane-sculpting protein that oligomerizes to generate flask-shaped invaginations of the plasma membrane known as caveolae. Mutations in CAV1 have been linked to multiple diseases in humans. Such mutations often interfere with oligomerization and the intracellular trafficking processes required for successful caveolae assembly, but the molecular mechanisms underlying these defects have not been structurally explained. Here, we investigate how a disease-associated mutation in one of the most highly conserved residues in CAV1, P132L, affects CAV1 structure and oligomerization. We show that P132 is positioned at a major site of protomer-protomer interactions within the CAV1 complex, providing a structural explanation for why the mutant protein fails to homo-oligomerize correctly. Using a combination of computational, structural, biochemical, and cell biological approaches, we find that despite its homo-oligomerization defects P132L is capable of forming mixed hetero-oligomeric complexes with WT CAV1 and that these complexes can be incorporated into caveolae. These findings provide insights into the fundamental mechanisms that control the formation of homo- and hetero-oligomers of caveolins that are essential for caveolae biogenesis, as well as how these processes are disrupted in human disease.

Template-free prediction of a new monotopic membrane protein fold and assembly by AlphaFold2.

AlphaFold2 (AF2) has revolutionized the field of protein structural prediction. Here, we test its ability to predict the tertiary and quaternary structure of a previously undescribed scaffold with new folds and unusual architecture, the monotopic membrane protein caveolin-1 (CAV1). CAV1 assembles into a disc-shaped oligomer composed of 11 symmetrically arranged protomers, each assuming an identical new fold, and contains the largest parallel ß-barrel known to exist in nature. Remarkably, AF2 predicts both the fold of the protomers and the interfaces between them. It also assembles between seven and 15 copies of CAV1 into disc-shaped complexes. However, the predicted multimers are energetically strained, especially the parallel ß-barrel. These findings highlight the ability of AF2 to correctly predict new protein folds and oligomeric assemblies at a granular level while missing some elements of higher-order complexes, thus positing a new direction for the continued development of deep-learning protein structure prediction approaches.

Adamts10 controls transforming growth factor ß family signaling that contributes to retinal ganglion cell development.

Although mutations in ADAMTS10 have long been known to cause autosomal recessive Weill-Marchesani Syndrome which is characterized by short stature and ocular abnormalities, more recent work has shown that certain mutations in ADAMTS10 cause glaucoma in dogs. In humans, glaucoma is the leading cause of irreversible vision loss that affects tens of millions of people world-wide. Vision loss in glaucoma is a result of neurodegeneration of retinal ganglion cells that form the inner-most layer of the retina and whose axons form the optic nerve which relays visual information to the brain. ADAMTS10 contributes to the formation of microfibrils which sequester latent transforming growth factor ß (TGFß). Among its many biological functions, TGFß promotes the development of retinal ganglion cells and is also known to play other roles in glaucoma pathogenesis. The aim of this study was to test the hypothesis that ADAMTS10 plays a role in retinal ganglion cell development through regulation of TGFß signaling. To this end, Adamts10 expression was targeted for reduction in zebrafish embryos carrying either a fluorescent reporter that labels retinal ganglion cells, or a fluorescent reporter of pSmad3-mediated TGFß family signaling. Loss of adamts10 function in zebrafish embryos reduced retinal ganglion cell reporter fluorescence and prevented formation of an ordered retinal ganglion cell layer. Targeting adamts10 expression also drastically reduced constitutive TGFß signaling in the eye. Direct inhibition of the TGFß receptor reduced retinal ganglion cell reporter fluorescence similar to the effect of targeting adamts10 expression. These findings unveil a previously unknown role for Adamts10 in retinal ganglion cell development and suggest that the developmental role of Adamts10 is mediated by active TGFß family signaling. In addition, our results show for the first time that Adamts10 is necessary for pSmad3-mediated constitutive TGFß family signaling.

Integrated AlphaFold2 and DEER investigation of the conformational dynamics of a pH-dependent APC antiporter.

The Amino Acid-Polyamine-Organocation (APC) transporter GadC contributes to the survival of pathogenic bacteria under extreme acid stress by exchanging extracellular glutamate for intracellular γ-aminobutyric acid (GABA). Its structure, determined in an inward-facing conformation at alkaline pH, consists of the canonical LeuT-fold with a conserved five-helix inverted repeat, thereby resembling functionally divergent transporters such as the serotonin transporter SERT and the glucose-sodium symporter SGLT1. However, despite this structural similarity, it is unclear if the conformational dynamics of antiporters such as GadC follow the blueprint of these or other LeuT-fold transporters. Here, we used double electron-electron resonance (DEER) spectroscopy to monitor the conformational dynamics of GadC in lipid bilayers in response to acidification and substrate binding. To guide experimental design and facilitate the interpretation of the DEER data, we generated an ensemble of structural models in multiple conformations using a recently introduced modification of AlphaFold2 . Our experimental results reveal acid-induced conformational changes that dislodge the Cterminus from the permeation pathway coupled with rearrangement of helices that enables isomerization between inward- and outward-facing states. The substrate glutamate, but not GABA, modulates the dynamics of an extracellular thin gate without shifting the equilibrium between inward- and outward-facing conformations. In addition to introducing an integrated methodology for probing transporter conformational dynamics, the congruence of the DEER data with patterns of structural rearrangements deduced from ensembles of AlphaFold2 models illuminates the conformational cycle of GadC underpinning transport and exposes yet another example of the divergence between the dynamics of different families in the LeuT-fold.

SPEACH_AF: Sampling protein ensembles and conformational heterogeneity with Alphafold2.

The unprecedented performance of Deepmind's Alphafold2 in predicting protein structure in CASP XIV and the creation of a database of structures for multiple proteomes and protein sequence repositories is reshaping structural biology. However, because this database returns a single structure, it brought into question Alphafold's ability to capture the intrinsic conformational flexibility of proteins. Here we present a general approach to drive Alphafold2 to model alternate protein conformations through simple manipulation of the multiple sequence alignment via in silico mutagenesis. The approach is grounded in the hypothesis that the multiple sequence alignment must also encode for protein structural heterogeneity, thus its rational manipulation will enable Alphafold2 to sample alternate conformations. A systematic modeling pipeline is benchmarked against canonical examples of protein conformational flexibility and applied to interrogate the conformational landscape of membrane proteins. This work broadens the applicability of Alphafold2 by generating multiple protein conformations to be tested biologically, biochemically, biophysically, and for use in structure-based drug design.

Principles of Alternating Access in LeuT-fold Transporters: Commonalities and Divergences.

Found in all domains of life, transporters belonging to the LeuT-fold class mediate the import and exchange of hydrophilic and charged compounds such as amino acids, metals, and sugar molecules. Nearly two decades of investigations on the eponymous bacterial transporter LeuT have yielded a library of high-resolution snapshots of its conformational cycle linked by solution-state experimental data obtained from multiple techniques. In parallel, its topology has been observed in symporters and antiporters characterized by a spectrum of substrate specificities and coupled to gradients of distinct ions. Here we review and compare mechanistic models of transport for LeuT, its well-studied homologs, as well as functionally distant members of the fold, emphasizing the commonalities and divergences in alternating access and the corresponding energy landscapes. Our integrated summary illustrates how fold conservation, a hallmark of the LeuT fold, coincides with divergent choreographies of alternating access that nevertheless capitalize on recurrent structural motifs. In addition, it highlights the knowledge gap that hinders the leveraging of the current body of research into detailed mechanisms of transport for this important class of membrane proteins.

Molecular architecture of the human caveolin-1 complex.

Membrane-sculpting proteins shape the morphology of cell membranes and facilitate remodeling in response to physiological and environmental cues. Complexes of the monotopic membrane protein caveolin function as essential curvature-generating components of caveolae, flask-shaped invaginations that sense and respond to plasma membrane tension. However, the structural basis for caveolin's membrane remodeling activity is currently unknown. Here, we show that, using cryo-electron microscopy, the human caveolin-1 complex is composed of 11 protomers organized into a tightly packed disc with a flat membrane-embedded surface. The structural insights suggest a previously unrecognized mechanism for how membrane-sculpting proteins interact with membranes and reveal how key regions of caveolin-1, including its scaffolding, oligomerization, and intramembrane domains, contribute to its function.

Modeling of protein conformational changes with Rosetta guided by limited experimental data.

Conformational changes are an essential component of functional cycles of many proteins, but their characterization often requires an integrative structural biology approach. Here, we introduce and benchmark ConfChangeMover (CCM), a new method built into the widely used macromolecular modeling suite Rosetta that is tailored to model conformational changes in proteins using sparse experimental data. CCM can rotate and translate secondary structural elements and modify their backbone dihedral angles in regions of interest. We benchmarked CCM on soluble and membrane proteins with simulated Cα-Cα distance restraints and sparse experimental double electron-electron resonance (DEER) restraints, respectively. In both benchmarks, CCM outperformed state-of-the-art Rosetta methods, showing that it can model a diverse array of conformational changes. In addition, the Rosetta framework allows a wide variety of experimental data to be integrated with CCM, thus extending its capability beyond DEER restraints. This method will contribute to the biophysical characterization of protein dynamics.

Sampling alternative conformational states of transporters and receptors with AlphaFold2.

Equilibrium fluctuations and triggered conformational changes often underlie the functional cycles of membrane proteins. For example, transporters mediate the passage of molecules across cell membranes by alternating between inward- and outward-facing states, while receptors undergo intracellular structural rearrangements that initiate signaling cascades. Although the conformational plasticity of these proteins has historically posed a challenge for traditional de novo protein structure prediction pipelines, the recent success of AlphaFold2 (AF2) in CASP14 culminated in the modeling of a transporter in multiple conformations to high accuracy. Given that AF2 was designed to predict static structures of proteins, it remains unclear if this result represents an underexplored capability to accurately predict multiple conformations and/or structural heterogeneity. Here, we present an approach to drive AF2 to sample alternative conformations of topologically diverse transporters and G-protein-coupled receptors that are absent from the AF2 training set. Whereas models of most proteins generated using the default AF2 pipeline are conformationally homogeneous and nearly identical to one another, reducing the depth of the input multiple sequence alignments by stochastic subsampling led to the generation of accurate models in multiple conformations. In our benchmark, these conformations spanned the range between two experimental structures of interest, with models at the extremes of these conformational distributions observed to be among the most accurate (average template modeling score of 0.94). These results suggest a straightforward approach to identifying native-like alternative states, while also highlighting the need for the next generation of deep learning algorithms to be designed to predict ensembles of biophysically relevant states.