AlphaFold-assisted structure determination of a bacterial protein of unknown function using X-ray and electron crystallography.

Macromolecular crystallography generally requires the recovery of missing phase information from diffraction data to reconstruct an electron-density map of the crystallized molecule. Most recent structures have been solved using molecular replacement as a phasing method, requiring an a priori structure that is closely related to the target protein to serve as a search model; when no such search model exists, molecular replacement is not possible. New advances in computational machine-learning methods, however, have resulted in major advances in protein structure predictions from sequence information. Methods that generate predicted structural models of sufficient accuracy provide a powerful approach to molecular replacement. Taking advantage of these advances, AlphaFold predictions were applied to enable structure determination of a bacterial protein of unknown function (UniProtKB Q63NT7, NCBI locus BPSS0212) based on diffraction data that had evaded phasing attempts using MIR and anomalous scattering methods. Using both X-ray and micro-electron (microED) diffraction data, it was possible to solve the structure of the main fragment of the protein using a predicted model of that domain as a starting point. The use of predicted structural models importantly expands the promise of electron diffraction, where structure determination relies critically on molecular replacement.

A Highly Ordered Nitroxide Side Chain for Distance Mapping and Monitoring Slow Structural Fluctuations in Proteins.

Site-directed spin labeling electron paramagnetic resonance (SDSL-EPR) is an established tool for exploring protein structure and dynamics. Although nitroxide side chains attached to a single cysteine via a disulfide linkage are commonly employed in SDSL-EPR, their internal flexibility complicates applications to monitor slow internal motions in proteins and to structure determination by distance mapping. Moreover, the labile disulfide linkage prohibits the use of reducing agents often needed for protein stability. To enable the application of SDSL-EPR to the measurement of slow internal dynamics, new spin labels with hindered internal motion are desired. Here, we introduce a highly ordered nitroxide side chain, designated R9, attached at a single cysteine residue via a non-reducible thioether linkage. The reaction to introduce R9 is highly selective for solvent-exposed cysteine residues. Structures of R9 at two helical sites in T4 Lysozyme were determined by X-ray crystallography and the mobility in helical sequences was characterized by EPR spectral lineshape analysis, Saturation Transfer EPR, and Saturation Recovery EPR. In addition, interspin distance measurements between pairs of R9 residues are reported. Collectively, all data indicate that R9 will be useful for monitoring slow internal structural fluctuations, and applications to distance mapping via dipolar spectroscopy and relaxation enhancement methods are anticipated. Supplementary Information: The online version contains supplementary material available at 10.1007/s00723-023-01618-8.

Cryo-EM structures of the D290V mutant of the hnRNPA2 low-complexity domain suggests how D290V affects phase separation and aggregation.

Heterogeneous nuclear ribonucleoprotein A2 (hnRNPA2) is a human ribonucleoprotein that transports RNA to designated locations for translation via its ability to phase separate. Its mutated form, D290V, is implicated in multisystem proteinopathy known to afflict two families, mainly with myopathy and Paget's disease of bone. Here, we investigate this mutant form of hnRNPA2 by determining cryo-EM structures of the recombinant D290V low complexity domain. We find that the mutant form of hnRNPA2 differs from the WT fibrils in four ways. In contrast to the WT fibrils, the PY-nuclear localization signals in the fibril cores of all three mutant polymorphs are less accessible to chaperones. Also, the mutant fibrils are more stable than WT fibrils as judged by phase separation, thermal stability, and energetic calculations. Similar to other pathogenic amyloids, the mutant fibrils are polymorphic. Thus, these structures offer evidence to explain how a D-to-V missense mutation diverts the assembly of reversible, functional amyloid-like fibrils into the assembly of pathogenic amyloid, and may shed light on analogous conversions occurring in other ribonucleoproteins that lead to neurological diseases such as amyotrophic lateral sclerosis and frontotemporal dementia.

Structure-based design of nanobodies that inhibit seeding of Alzheimer's patient-extracted tau fibrils.

Despite much effort, antibody therapies for Alzheimer's disease (AD) have shown limited efficacy. Challenges to the rational design of effective antibodies include the difficulty of achieving specific affinity to critical targets, poor expression, and antibody aggregation caused by buried charges and unstructured loops. To overcome these challenges, we grafted previously determined sequences of fibril-capping amyloid inhibitors onto a camel heavy chain antibody scaffold. These sequences were designed to cap fibrils of tau, known to form the neurofibrillary tangles of AD, thereby preventing fibril elongation. The nanobodies grafted with capping inhibitors blocked tau aggregation in biosensor cells seeded with postmortem brain extracts from AD and progressive supranuclear palsy (PSP) patients. The tau capping nanobody inhibitors also blocked seeding by recombinant tau oligomers. Another challenge to the design of effective antibodies is their poor blood-brain barrier (BBB) penetration. In this study, we also designed a bispecific nanobody composed of a nanobody that targets a receptor on the BBB and a tau capping nanobody inhibitor, conjoined by a flexible linker. We provide evidence that the bispecific nanobody improved BBB penetration over the tau capping inhibitor alone after intravenous administration in mice. Our results suggest that the design of synthetic antibodies that target sequences that drive protein aggregation may be a promising approach to inhibit the prion-like seeding of tau and other proteins involved in AD and related proteinopathies.

Cryo-EM structure determination of small therapeutic protein targets at 3 Å-resolution using a rigid imaging scaffold.

Cryoelectron microscopy (Cryo-EM) has enabled structural determination of proteins larger than about 50 kDa, including many intractable by any other method, but it has largely failed for smaller proteins. Here, we obtain structures of small proteins by binding them to a rigid molecular scaffold based on a designed protein cage, revealing atomic details at resolutions reaching 2.9 Å. We apply this system to the key cancer signaling protein KRAS (19 kDa in size), obtaining four structures of oncogenic mutational variants by cryo-EM. Importantly, a structure for the key G12C mutant bound to an inhibitor drug (AMG510) reveals significant conformational differences compared to prior data in the crystalline state. The findings highlight the promise of cryo-EM scaffolds for advancing the design of drug molecules against small therapeutic protein targets in cancer and other human diseases.

Development and atomic structure of a new fluorescence-based sensor to probe heme transfer in bacterial pathogens.

Heme is the most abundant source of iron in the human body and is actively scavenged by bacterial pathogens during infections. Corynebacterium diphtheriae and other species of actinobacteria scavenge heme using cell wall associated and secreted proteins that contain Conserved Region (CR) domains. Here we report the development of a fluorescent sensor to measure heme transfer from the C-terminal CR domain within the HtaA protein (CR2) to other hemoproteins within the heme-uptake system. The sensor contains the CR2 domain inserted into the ß2 to ß3 turn of the Enhanced Green Fluorescent Protein (EGFP). A 2.45 Å crystal structure reveals the basis of heme binding to the CR2 domain via iron-tyrosyl coordination and shares conserved structural features with CR domains present in Corynebacterium glutamicum. The structure and small angle X-ray scattering experiments are consistent with the sensor adopting a V-shaped structure that exhibits only small fluctuations in inter-domain positioning. We demonstrate heme transfer from the sensor to the CR domains located within the HtaA or HtaB proteins in the heme-uptake system as measured by a âˆ¼ 60% increase in sensor fluorescence and native mass spectrometry.

Low complexity domains of the nucleocapsid protein of SARS-CoV-2 form amyloid fibrils.

The self-assembly of the Nucleocapsid protein (NCAP) of SARS-CoV-2 is crucial for its function. Computational analysis of the amino acid sequence of NCAP reveals low-complexity domains (LCDs) akin to LCDs in other proteins known to self-assemble as phase separation droplets and amyloid fibrils. Previous reports have described NCAP's propensity to phase-separate. Here we show that the central LCD of NCAP is capable of both, phase separation and amyloid formation. Within this central LCD we identified three adhesive segments and determined the atomic structure of the fibrils formed by each. Those structures guided the design of G12, a peptide that interferes with the self-assembly of NCAP and demonstrates antiviral activity in SARS-CoV-2 infected cells. Our work, therefore, demonstrates the amyloid form of the central LCD of NCAP and suggests that amyloidogenic segments of NCAP could be targeted for drug development.

The Shr receptor from Streptococcus pyogenes uses a cap and release mechanism to acquire heme-iron from human hemoglobin.

Streptococcus pyogenes (group A Streptococcus) is a clinically important microbial pathogen that requires iron in order to proliferate. During infections, S. pyogenes uses the surface displayed Shr receptor to capture human hemoglobin (Hb) and acquires its iron-laden heme molecules. Through a poorly understood mechanism, Shr engages Hb via two structurally unique N-terminal Hb-interacting domains (HID1 and HID2) which facilitate heme transfer to proximal NEAr Transporter (NEAT) domains. Based on the results of X-ray crystallography, small angle X-ray scattering, NMR spectroscopy, native mass spectrometry, and heme transfer experiments, we propose that Shr utilizes a "cap and release" mechanism to gather heme from Hb. In the mechanism, Shr uses the HID1 and HID2 modules to preferentially recognize only heme-loaded forms of Hb by contacting the edges of its protoporphyrin rings. Heme transfer is enabled by significant receptor dynamics within the Shr-Hb complex which function to transiently uncap HID1 from the heme bound to Hb's ß subunit, enabling the gated release of its relatively weakly bound heme molecule and subsequent capture by Shr's NEAT domains. These dynamics may maximize the efficiency of heme scavenging by S. pyogenes, enabling it to preferentially recognize and remove heme from only heme-loaded forms of Hb that contain iron.

Micro-electron diffraction structure of the aggregation-driving N terminus of Drosophila neuronal protein Orb2A reveals amyloid-like ß-sheets.

Amyloid protein aggregation is commonly associated with progressive neurodegenerative diseases, however not all amyloid fibrils are pathogenic. The neuronal cytoplasmic polyadenylation element binding protein is a regulator of synaptic mRNA translation and has been shown to form functional amyloid aggregates that stabilize long-term memory. In adult Drosophila neurons, the cytoplasmic polyadenylation element binding homolog Orb2 is expressed as 2 isoforms, of which the Orb2B isoform is far more abundant, but the rarer Orb2A isoform is required to initiate Orb2 aggregation. The N terminus is a distinctive feature of the Orb2A isoform and is critical for its aggregation. Intriguingly, replacement of phenylalanine in the fifth position of Orb2A with tyrosine (F5Y) in Drosophila impairs stabilization of long-term memory. The structure of endogenous Orb2B fibers was recently determined by cryo-EM, but the structure adopted by fibrillar Orb2A is less certain. Here we use micro-electron diffraction to determine the structure of the first 9 N-terminal residues of Orb2A, at a resolution of 1.05 Å. We find that this segment (which we term M9I) forms an amyloid-like array of parallel in-register ß-sheets, which interact through side chain interdigitation of aromatic and hydrophobic residues. Our structure provides an explanation for the decreased aggregation observed for the F5Y mutant and offers a hypothesis for how the addition of a single atom (the tyrosyl oxygen) affects long-term memory. We also propose a structural model of Orb2A that integrates our structure of the M9I segment with the published Orb2B cryo-EM structure.

Crystal structure of mevalonate 3,5-bisphosphate decarboxylase reveals insight into the evolution of decarboxylases in the mevalonate metabolic pathways.

Mevalonate 3,5-bisphosphate decarboxylase is involved in the recently discovered Thermoplasma-type mevalonate pathway. The enzyme catalyzes the elimination of the 3-phosphate group from mevalonate 3,5-bisphosphate as well as concomitant decarboxylation of the substrate. This entire reaction of the enzyme resembles the latter half-reactions of its homologs, diphosphomevalonate decarboxylase and phosphomevalonate decarboxylase, which also catalyze ATP-dependent phosphorylation of the 3-hydroxyl group of their substrates. However, the crystal structure of mevalonate 3,5-bisphosphate decarboxylase and the structural reasons of the difference between reactions catalyzed by the enzyme and its homologs are unknown. In this study, we determined the X-ray crystal structure of mevalonate 3,5-bisphosphate decarboxylase from Picrophilus torridus, a thermoacidophilic archaeon of the order Thermoplasmatales. Structural and mutational analysis demonstrated the importance of a conserved aspartate residue for enzyme activity. In addition, although crystallization was performed in the absence of substrate or ligands, residual electron density having the shape of a fatty acid was observed at a position overlapping the ATP-binding site of the homologous enzyme, diphosphomevalonate decarboxylase. This finding is in agreement with the expected evolutionary route from phosphomevalonate decarboxylase (ATP-dependent) to mevalonate 3,5-bisphosphate decarboxylase (ATP-independent) through the loss of kinase activity. We found that the binding of geranylgeranyl diphosphate, an intermediate of the archeal isoprenoid biosynthesis pathway, evoked significant activation of mevalonate 3,5-bisphosphate decarboxylase, and several mutations at the putative geranylgeranyl diphosphate-binding site impaired this activation, suggesting the physiological importance of ligand binding as well as a possible novel regulatory system employed by the Thermoplasma-type mevalonate pathway.

Cryo-EM structure of RNA-induced tau fibrils reveals a small C-terminal core that may nucleate fibril formation.

In neurodegenerative diseases including Alzheimer's and amyotrophic lateral sclerosis, proteins that bind RNA are found in aggregated forms in autopsied brains. Evidence suggests that RNA aids nucleation of these pathological aggregates; however, the mechanism has not been investigated at the level of atomic structure. Here, we present the 3.4-Å resolution structure of fibrils of full-length recombinant tau protein in the presence of RNA, determined by electron cryomicroscopy (cryo-EM). The structure reveals the familiar in-register cross-ß amyloid scaffold but with a small fibril core spanning residues Glu391 to Ala426, a region disordered in the fuzzy coat in all previously studied tau polymorphs. RNA is bound on the fibril surface to the positively charged residues Arg406 and His407 and runs parallel to the fibril axis. The fibrils dissolve when RNase is added, showing that RNA is necessary for fibril integrity. While this structure cannot exist simultaneously with the tau fibril structures extracted from patients' brains, it could conceivably account for the nucleating effects of RNA cofactors followed by remodeling as fibrils mature.

The structure of the Clostridium thermocellum RsgI9 ectodomain provides insight into the mechanism of biomass sensing.

Clostridium thermocellum is actively being developed as a microbial platform to produce biofuels and chemicals from renewable plant biomass. An attractive feature of this bacterium is its ability to efficiently degrade lignocellulose using surface-displayed cellulosomes, large multi-protein complexes that house different types of cellulase enzymes. Clostridium thermocellum tailors the enzyme composition of its cellulosome using nine membrane-embedded anti-σ factors (RsgI1-9), which are thought to sense different types of extracellular carbohydrates and then elicit distinct gene expression programs via cytoplasmic σ factors. Here we show that the RsgI9 anti-σ factor interacts with cellulose via a C-terminal bi-domain unit. A 2.0 Å crystal structure reveals that the unit is constructed from S1C peptidase and NTF2-like protein domains that contain a potential binding site for cellulose. Small-angle X-ray scattering experiments of the intact ectodomain indicate that it adopts a bi-lobed, elongated conformation. In the structure, a conserved RsgI extracellular (CRE) domain is connected to the bi-domain via a proline-rich linker, which is expected to project the carbohydrate-binding unit ~160 Å from the cell surface. The CRE and proline-rich elements are conserved in several other C. thermocellum anti-σ factors, suggesting that they will also form extended structures that sense carbohydrates.

Insight into the molecular basis of substrate recognition by the wall teichoic acid glycosyltransferase TagA.

Wall teichoic acid (WTA) polymers are covalently affixed to the Gram-positive bacterial cell wall and have important functions in cell elongation, cell morphology, biofilm formation, and ß-lactam antibiotic resistance. The first committed step in WTA biosynthesis is catalyzed by the TagA glycosyltransferase (also called TarA), a peripheral membrane protein that produces the conserved linkage unit, which joins WTA to the cell wall peptidoglycan. TagA contains a conserved GT26 core domain followed by a C-terminal polypeptide tail that is important for catalysis and membrane binding. Here, we report the crystal structure of the Thermoanaerobacter italicus TagA enzyme bound to UDP-N-acetyl-d-mannosamine, revealing the molecular basis of substrate binding. Native MS experiments support the model that only monomeric TagA is enzymatically active and that it is stabilized by membrane binding. Molecular dynamics simulations and enzyme activity measurements indicate that the C-terminal polypeptide tail facilitates catalysis by encapsulating the UDP-N-acetyl-d-mannosamine substrate, presenting three highly conserved arginine residues to the active site that are important for catalysis (R214, R221, and R224). From these data, we present a mechanistic model of catalysis that ascribes functions for these residues. This work could facilitate the development of new antimicrobial compounds that disrupt WTA biosynthesis in pathogenic bacteria.

Prospecting for natural products by genome mining and microcrystal electron diffraction.

More than 60% of pharmaceuticals are related to natural products (NPs), chemicals produced by living organisms. Despite this, the rate of NP discovery has slowed over the past few decades. In many cases the rate-limiting step in NP discovery is structural characterization. Here we report the use of microcrystal electron diffraction (MicroED), an emerging cryogenic electron microscopy (CryoEM) method, in combination with genome mining to accelerate NP discovery and structural elucidation. As proof of principle we rapidly determine the structure of a new 2-pyridone NP, Py-469, and revise the structure of fischerin, an NP isolated more than 25 years ago, with potent cytotoxicity but hitherto ambiguous structural assignment. This study serves as a powerful demonstration of the synergy of MicroED and synthetic biology in NP discovery, technologies that when taken together will ultimately accelerate the rate at which new drugs are discovered.

Structure of human factor VIIa-soluble tissue factor with calcium, magnesium and rubidium.

Coagulation factor VIIa (FVIIa) consists of a γ-carboxyglutamic acid (GLA) domain, two epidermal growth factor-like (EGF) domains and a protease domain. FVIIa binds three Mg2+ ions and four Ca2+ ions in the GLA domain, one Ca2+ ion in the EGF1 domain and one Ca2+ ion in the protease domain. Further, FVIIa contains an Na+ site in the protease domain. Since Na+ and water share the same number of electrons, Na+ sites in proteins are difficult to distinguish from waters in X-ray structures. Here, to verify the Na+ site in FVIIa, the structure of the FVIIa-soluble tissue factor (TF) complex was solved at 1.8 Šresolution containing Mg2+, Ca2+ and Rb+ ions. In this structure, Rb+ replaced two Ca2+ sites in the GLA domain and occupied three non-metal sites in the protease domain. However, Rb+ was not detected at the expected Na+ site. In kinetic experiments, Na+ increased the amidolytic activity of FVIIa towards the synthetic substrate S-2288 (H-D-Ile-Pro-Arg-p-nitroanilide) by ∼20-fold; however, in the presence of Ca2+, Na+ had a negligible effect. Ca2+ increased the hydrolytic activity of FVIIa towards S-2288 by ∼60-fold in the absence of Na+ and by ∼82-fold in the presence of Na+. In molecular-dynamics simulations, Na+ stabilized the two Na+-binding loops (the 184-loop and 220-loop) and the TF-binding region spanning residues 163-180. Ca2+ stabilized the Ca2+-binding loop (the 70-loop) and Na+-binding loops but not the TF-binding region. Na+ and Ca2+ together stabilized both the Na+-binding and Ca2+-binding loops and the TF-binding region. Previously, Rb+ has been used to define the Na+ site in thrombin; however, it was unsuccessful in detecting the Na+ site in FVIIa. A conceivable explanation for this observation is provided.

Inhibition of amyloid formation of the Nucleoprotein of SARS-CoV-2.

The SARS-CoV-2 Nucleoprotein (NCAP) functions in RNA packaging during viral replication and assembly. Computational analysis of its amino acid sequence reveals a central low-complexity domain (LCD) having sequence features akin to LCDs in other proteins known to function in liquid-liquid phase separation. Here we show that in the presence of viral RNA, NCAP, and also its LCD segment alone, form amyloid-like fibrils when undergoing liquid-liquid phase separation. Within the LCD we identified three 6-residue segments that drive amyloid fibril formation. We determined atomic structures for fibrils formed by each of the three identified segments. These structures informed our design of peptide inhibitors of NCAP fibril formation and liquid-liquid phase separation, suggesting a therapeutic route for Covid-19. ONE SENTENCE SUMMARY: Atomic structures of amyloid-driving peptide segments from SARS-CoV-2 Nucleoprotein inform the development of Covid-19 therapeutics.

Isolation and X-ray Crystal Structure of an Electrogenerated TEMPO-N3 Charge-Transfer Complex.

Advances in radical-based catalytic reactions have created a demand for understanding their mechanistic underpinnings. Here, we present the isolation, structural elucidation, and theoretical analysis of a catalytically relevant charge-transfer species formed between the azidyl radical and (2,2,6,6-tetramethylpiperidin-1-yl)oxyl (TEMPO). The unusual bond angles and pancake bonding between these two fragments highlight the weak bonding interactions present in this complex. This X-ray structure validates computational predictions as well as mechanistic proposals of TEMPO-mediated radical azidation reactions.

CryoEM structure of the low-complexity domain of hnRNPA2 and its conversion to pathogenic amyloid.

hnRNPA2 is a human ribonucleoprotein (RNP) involved in RNA metabolism. It forms fibrils both under cellular stress and in mutated form in neurodegenerative conditions. Previous work established that the C-terminal low-complexity domain (LCD) of hnRNPA2 fibrillizes under stress, and missense mutations in this domain are found in the disease multisystem proteinopathy (MSP). However, little is known at the atomic level about the hnRNPA2 LCD structure that is involved in those processes and how disease mutations cause structural change. Here we present the cryo-electron microscopy (cryoEM) structure of the hnRNPA2 LCD fibril core and demonstrate its capability to form a reversible hydrogel in vitro containing amyloid-like fibrils. Whereas these fibrils, like pathogenic amyloid, are formed from protein chains stacked into ß-sheets by backbone hydrogen bonds, they display distinct structural differences: the chains are kinked, enabling non-covalent cross-linking of fibrils and disfavoring formation of pathogenic steric zippers. Both reversibility and energetic calculations suggest these fibrils are less stable than pathogenic amyloid. Moreover, the crystal structure of the disease-mutation-containing segment (D290V) of hnRNPA2 suggests that the replacement fundamentally alters the fibril structure to a more stable energetic state. These findings illuminate how molecular interactions promote protein fibril networks and how mutation can transform fibril structure from functional to a pathogenic form.

Crystal structure of a conformational antibody that binds tau oligomers and inhibits pathological seeding by extracts from donors with Alzheimer's disease.

Soluble oligomers of aggregated tau accompany the accumulation of insoluble amyloid fibrils, a histological hallmark of Alzheimer disease (AD) and two dozen related neurodegenerative diseases. Both oligomers and fibrils seed the spread of Tau pathology, and by virtue of their low molecular weight and relative solubility, oligomers may be particularly pernicious seeds. Here, we report the formation of in vitro tau oligomers formed by an ionic liquid (IL15). Using IL15-induced recombinant tau oligomers and a dot blot assay, we discovered a mAb (M204) that binds oligomeric tau, but not tau monomers or fibrils. M204 and an engineered single-chain variable fragment (scFv) inhibited seeding by IL15-induced tau oligomers and pathological extracts from donors with AD and chronic traumatic encephalopathy. This finding suggests that M204-scFv targets pathological structures that are formed by tau in neurodegenerative diseases. We found that M204-scFv itself partitions into oligomeric forms that inhibit seeding differently, and crystal structures of the M204-scFv monomer, dimer, and trimer revealed conformational differences that explain differences among these forms in binding and inhibition. The efficiency of M204-scFv antibodies to inhibit the seeding by brain tissue extracts from different donors with tauopathies varied among individuals, indicating the possible existence of distinct amyloid polymorphs. We propose that by binding to oligomers, which are hypothesized to be the earliest seeding-competent species, M204-scFv may have potential as an early-stage diagnostic for AD and tauopathies, and also could guide the development of promising therapeutic antibodies.

Serial femtosecond crystallography on in vivo-grown crystals drives elucidation of mosquitocidal Cyt1Aa bioactivation cascade.

Cyt1Aa is the one of four crystalline protoxins produced by mosquitocidal bacterium Bacillus thuringiensis israelensis (Bti) that has been shown to delay the evolution of insect resistance in the field. Limiting our understanding of Bti efficacy and the path to improved toxicity and spectrum has been ignorance of how Cyt1Aa crystallizes in vivo and of its mechanism of toxicity. Here, we use serial femtosecond crystallography to determine the Cyt1Aa protoxin structure from sub-micron-sized crystals produced in Bti. Structures determined under various pH/redox conditions illuminate the role played by previously uncharacterized disulfide-bridge and domain-swapped interfaces from crystal formation in Bti to dissolution in the larval mosquito midgut. Biochemical, toxicological and biophysical methods enable the deconvolution of key steps in the Cyt1Aa bioactivation cascade. We additionally show that the size, shape, production yield, pH sensitivity and toxicity of Cyt1Aa crystals grown in Bti can be controlled by single atom substitution.