A 5+1 assemble-to-activate mechanism of the Lon proteolytic machine.

Many AAA+ (ATPases associated with diverse cellular activities) proteins function as protein or DNA remodelers by threading the substrate through the central pore of their hexameric assemblies. In this ATP-dependent translocating state, the substrate is gripped by the pore loops of the ATPase domains arranged in a universal right-handed spiral staircase organization. However, the process by which a AAA+ protein is activated to adopt this substrate-pore-loop arrangement remains unknown. We show here, using cryo-electron microscopy (cryo-EM), that the activation process of the Lon AAA+ protease may involve a pentameric assembly and a substrate-dependent incorporation of the sixth protomer to form the substrate-pore-loop contacts seen in the translocating state. Based on the structural results, we design truncated monomeric mutants that inhibit Lon activity by binding to the native pentamer and demonstrated that expressing these monomeric mutants in Escherichia coli cells containing functional Lon elicits specific phenotypes associated with lon deficiency, including the inhibition of persister cell formation. These findings uncover a substrate-dependent assembly process for the activation of a AAA+ protein and demonstrate a targeted approach to selectively inhibit its function within cells.

Structural basis for the hydrolytic activity of the transpeptidase-like protein DpaA to detach Braun's lipoprotein from peptidoglycan.

IMPORTANCE: Cross-linking reaction of Braun's lipoprotein (Lpp) to peptidoglycan (PG) is catalyzed by some members of the YkuD family of transpeptidases. However, the exact opposite reaction of cleaving the Lpp-PG cross-link is performed by DpaA, which is also a YkuD-like protein. In this work, we determined the crystal structure of DpaA to provide the molecular rationale for the ability of the transpeptidase-like protein to cleave, rather than form, the Lpp-PG linkage. Our findings also revealed the structural features that distinguish the different functional types of the YkuD family enzymes from one another.

Robust Design of Effective Allosteric Activators for Rsp5 E3 Ligase Using the Machine Learning Tool ProteinMPNN.

We used the deep learning tool ProteinMPNN to redesign ubiquitin (Ub) as a specific and functionally stimulating/enhancing binder of the Rsp5 E3 ligase. We generated 20 extensively mutated─up to 37 of 76 residues─recombinant Ub variants (UbVs), named R1 to R20, displaying well-folded structures and high thermal stabilities. These UbVs can also form stable complexes with Rsp5, as predicted using AlphaFold2. Three of the UbVs bound to Rsp5 with low micromolar affinity, with R4 and R12 effectively enhancing the Rsp5 activity six folds. AlphaFold2 predicts that R4 and R12 bind to Rsp5's exosite in an identical manner to the Rsp5-Ub template, thereby allosterically activating Rsp5-Ub thioester formation. Thus, we present a virtual solution for rapidly and cost-effectively designing UbVs as functional modulators of Ub-related enzymes.

Cryo-EM reveals the structure and dynamics of a 723-residue malate synthase G.

Determination of sub-100 kDa (kDa) structures by cryo-electron microscopy (EM) is a longstanding but not straightforward goal. Here, we present a 2.9-Å cryo-EM structure of a 723-amino acid apo-form malate synthase G (MSG) from Escherichia coli. The cryo-EM structure of the 82-kDa MSG exhibits the same global folding as structures resolved by crystallography and nuclear magnetic resonance (NMR) spectroscopy, and the crystal and cryo-EM structures are indistinguishable. Analyses of MSG dynamics reveal consistent conformational flexibilities among the three experimental approaches, most notably that the α/ß domain exhibits structural heterogeneity. We observed that sidechains of F453, L454, M629, and E630 residues involved in hosting the cofactor acetyl-CoA and substrate rotate differently between the cryo-EM apo-form and complex crystal structures. Our work demonstrates that the cryo-EM technique can be used to determine structures and conformational heterogeneity of sub-100 kDa biomolecules to a quality as high as that obtained from X-ray crystallography and NMR spectroscopy.

Structural basis for the helical filament formation of Escherichia coli glutamine synthetase.

Escherichia coli glutamine synthetase (EcGS) spontaneously forms a dodecamer that catalytically converts glutamate to glutamine. EcGS stacks with other dodecamers to create a filament-like polymer visible under transmission electron microscopy. Filamentous EcGS is induced by environmental metal ions. We used cryo-electron microscopy (cryo-EM) to decipher the structure of metal ion (nickel)-induced EcGS helical filament at a sub-3Å resolution. EcGS filament formation involves stacking of native dodecamers by chelating nickel ions to residues His5 and His13 in the first N-terminal helix (H1). His5 and His13 from paired parallel H1 helices provide salt bridges and hydrogen bonds to tightly stack two dodecamers. One subunit of the EcGS filament hosts two nickel ions, whereas the dodecameric interface and the ATP/Mg-binding site both host a nickel ion each. We reveal that upon adding glutamate or ATP for catalytic reactions, nickel-induced EcGS filament reverts to individual dodecamers. Such tunable filament formation is often associated with stress responses. Our results provide detailed structural information on the mechanism underlying reversible and tunable EcGS filament formation.

Structural and biological insights into Klebsiella pneumoniae surface polysaccharide degradation by a bacteriophage K1 lyase: implications for clinical use.

BACKGROUND: K1 capsular polysaccharide (CPS)-associated Klebsiella pneumoniae is the primary cause of pyogenic liver abscesses (PLA) in Asia. Patients with PLA often have serious complications, ultimately leading to a mortality of ~ 5%. This K1 CPS has been reported as a promising target for development of glycoconjugate vaccines against K. pneumoniae infection. The pyruvylation and O-acetylation modifications on the K1 CPS are essential to the immune response induced by the CPS. To date, however, obtaining the fragments of K1 CPS that contain the pyruvylation and O-acetylation for generating glycoconjugate vaccines still remains a challenge. METHODS: We analyzed the digested CPS products with NMR spectroscopy and mass spectrometry to reveal a bacteriophage-derived polysaccharide depolymerase specific to K1 CPS. The biochemical and biophysical properties of the enzyme were characterized and its crystal structures containing bound CPS products were determined. We also performed site-directed mutagenesis, enzyme kinetic analysis, phage absorption and infectivity studies, and treatment of the K. pneumoniae-infected mice with the wild-type and mutant enzymes. RESULTS: We found a bacteriophage-derived polysaccharide lyase that depolymerizes the K1 CPS into fragments of 1-3 repeating trisaccharide units with the retention of the pyruvylation and O-acetylation, and thus the important antigenic determinants of intact K1 CPS. We also determined the 1.46-Å-resolution, product-bound crystal structure of the enzyme, revealing two distinct carbohydrate-binding sites in a trimeric ß-helix architecture, which provide the first direct evidence for a second, non-catalytic, carbohydrate-binding site in bacteriophage-derived polysaccharide depolymerases. We demonstrate the tight interaction between the pyruvate moiety of K1 CPS and the enzyme in this second carbohydrate-binding site to be crucial to CPS depolymerization of the enzyme as well as phage absorption and infectivity. We also demonstrate that the enzyme is capable of protecting mice from K1 K. pneumoniae infection, even against a high challenge dose. CONCLUSIONS: Our results provide insights into how the enzyme recognizes and depolymerizes the K1 CPS, and demonstrate the potential use of the protein not only as a therapeutic agent against K. pneumoniae, but also as a tool to prepare structurally-defined oligosaccharides for the generation of glycoconjugate vaccines against infections caused by this organism.

Condition-dependent structural collapse in the intrinsically disordered N-terminal domain of prion protein.

Prion protein is composed of a structure-unsolved N-terminal domain and a globular C-terminal domain. Under limited trypsin digestion, mouse recombinant prion protein can be cleaved into two parts at residue Lys105. Here, we termed these two fragments as the N-domain (sequence 23-105) and the C-domain (sequence 106-230). In this study, the structural properties of the N-domain, the C-domain, and the full-length protein were explored using small-angle X-ray scattering, analytical ultracentrifugation, circular dichroism spectroscopy, and the 8-anilino-1-naphthalenesulfonic acid binding assay. The conformation and size of the prion protein were found to change sensitively under the solvent conditions. The positive residues in the sequence 23-99 of the N-domain were found to be responsible for the enhanced flexibility with the salt concentration reduced below 5 mM. The C-domain containing a hydrophobic patch tends to unfold and aggregate during a salt-induced structural collapse. The N-domain collapsed together with the C-domain at pH 5.2, whereas it collapsed independently at pH 4.2. The positively charged cluster (sequence 100-105) in the N-domain contributed to protecting the exposed hydrophobic surface of the C-domain.

Effect of SARS-CoV-2 B.1.1.7 mutations on spike protein structure and function.

The B.1.1.7 variant of SARS-CoV-2 first detected in the UK harbors amino-acid substitutions and deletions in the spike protein that potentially enhance host angiotensin conversion enzyme 2 (ACE2) receptor binding and viral immune evasion. Here we report cryo-EM structures of the spike protein of B.1.1.7 in the apo and ACE2-bound forms. The apo form showed one or two receptor-binding domains (RBDs) in the open conformation, without populating the fully closed state. All three RBDs were engaged in ACE2 binding. The B.1.1.7-specific A570D mutation introduces a molecular switch that could modulate the opening and closing of the RBD. The N501Y mutation introduces a π-π interaction that enhances RBD binding to ACE2 and abolishes binding of a potent neutralizing antibody (nAb). Cryo-EM also revealed how a cocktail of two nAbs simultaneously bind to all three RBDs, and demonstrated the potency of the nAb cocktail to neutralize different SARS-CoV-2 pseudovirus strains, including B.1.1.7.

Direct Visualization of a 26 kDa Protein by Cryo-Electron Microscopy Aided by a Small Scaffold Protein.

Cryo-electron microscopy (cryo-EM)-based structure determination of small proteins is hindered by the technical challenges associated with low signal-to-noise ratios of their particle images in intrinsically noisy micrographs. One solution is to attach the target protein to a large protein scaffold to increase its apparent size and, therefore, image contrast. Here we report a novel scaffold design based on a trimeric helical protein, E. coli ornithine transcarbamylase (OTC), fused to human ubiquitin. As a proof of principle, we demonstrated the ability to resolve a cryo-EM map of a 26 kDa human ubiquitin C-terminal hydrolase (UCHL1) attached to the C-terminus of ubiquitin as part of the trimeric assembly. The results revealed conformational changes in UCHL1 upon binding to ubiquitin, namely, a significant displacement of α-helix 2, which was also observed by X-ray crystallography. Our findings demonstrated the potential of the trimeric OTC scaffold design for studying a large number of ubiquitin interacting proteins by cryo-EM.

Structural polymorphism and substrate promiscuity of a ribosome-associated molecular chaperone.

Trigger factor (TF) is a highly conserved multi-domain molecular chaperone that exerts its chaperone activity at the ribosomal tunnel exit from which newly synthesized nascent chains emerge. TF also displays promiscuous substrate binding for a large number of cytosolic proteins independent of ribosome binding. We asked how TF recognizes a variety of substrates while existing in a monomer-dimer equilibrium. Paramagnetic nuclear magnetic resonance (NMR) and electron spin resonance (ESR) spectroscopy were used to show that dimeric TF displays a high degree of structural polymorphism in solution. A series of peptides has been generated to quantify their TF binding affinities in relation with their sequence compositions. The results confirmed a previous predication that TF preferentially binds to peptide fragments that are rich in aromatic and positively charged amino acids. NMR paramagnetic relaxation enhancement analysis showed that TF utilizes multiple binding sites, located in the chaperone domain and part of the prolyl trans-cis isomerization domain, to interact with these peptides. Dimerization of TF effectively sequesters most of the substrate binding sites, which are expected to become accessible upon binding to the ribosome as a monomer. As TF lacks ATPase activity, which is commonly used to trigger conformational changes within molecular chaperones in action, the ribosome-binding-associated disassembly and conformational rearrangements may be the underlying regulatory mechanism of its chaperone activity.

Multiresolution Imaging Using Bioluminescence Resonance Energy Transfer Identifies Distinct Biodistribution Profiles of Extracellular Vesicles and Exomeres with Redirected Tropism.

Extracellular particles (EPs) including extracellular vesicles (EVs) and exomeres play significant roles in diseases and therapeutic applications. However, their spatiotemporal dynamics in vivo have remained largely unresolved in detail due to the lack of a suitable method. Therefore, a bioluminescence resonance energy transfer (BRET)-based reporter, PalmGRET, is created to enable pan-EP labeling ranging from exomeres (<50 nm) to small (<200 nm) and medium and large (>200 nm) EVs. PalmGRET emits robust, sustained signals and allows the visualization, tracking, and quantification of the EPs from whole animal to nanoscopic resolutions under different imaging modalities, including bioluminescence, BRET, and fluorescence. Using PalmGRET, it is shown that EPs released by lung metastatic hepatocellular carcinoma (HCC) exhibit lung tropism with varying distributions to other major organs in immunocompetent mice. It is further demonstrated that gene knockdown of lung-tropic membrane proteins, solute carrier organic anion transporter family member 2A1, alanine aminopeptidase/Cd13, and chloride intracellular channel 1 decreases HCC-EP distribution to the lungs and yields distinct biodistribution profiles. It is anticipated that EP-specific imaging, quantitative assays, and detailed in vivo characterization are a starting point for more accurate and comprehensive in vivo models of EP biology and therapeutic design.

Structural Basis for the Differential Regulatory Roles of the PDZ Domain in C-Terminal Processing Proteases.

Carboxyl (C)-terminal processing proteases (CTPs) participate in protective and regulatory proteolysis in bacteria. The PDZ domain is central to the activity of CTPs but plays inherently different regulatory roles. For example, the PDZ domain inhibits the activity of the signaling protease CtpB by blocking the active site but is required for the activation of Prc (or Tsp), a tail-specific protease that degrades SsrA-tagged proteins. Here, by structural and functional analyses, we show that in the unliganded resting state of Prc, the PDZ domain is docked inside the bowl-shaped scaffold without contacting the active site, which is kept in a default misaligned conformation. In Prc, a hydrophobic substrate sensor distinct from CtpB engages substrate binding to the PDZ domain and triggers a structural remodeling to align the active-site residues. Therefore, this work reveals the structural basis for understanding the contrasting roles of the PDZ domain in the regulation of CTPs.IMPORTANCE Prc, also known previously as Tsp, is the founding member of the carboxyl-terminal processing protease (CTP) family of PDZ domain-containing proteases that include CtpA and CtpB. The substrate-binding PDZ domain is responsible for regulating the protease activity of CTP proteases; however, the regulatory role of PDZ domain is stimulatory in Prc but inhibitory in CtpA/B. By determining a series of crystal structures of Prc in the unliganded resting state, this study presents the structural basis for PDZ-dependent activation of Prc, the results of which explain the contrasting roles of the PDZ domain in the regulation of the protease activity of CTPs.

Automation of the kidney function prediction and classification through ultrasound-based kidney imaging using deep learning.

Prediction of kidney function and chronic kidney disease (CKD) through kidney ultrasound imaging has long been considered desirable in clinical practice because of its safety, convenience, and affordability. However, this highly desirable approach is beyond the capability of human vision. We developed a deep learning approach for automatically determining the estimated glomerular filtration rate (eGFR) and CKD status. We exploited the transfer learning technique, integrating the powerful ResNet model pretrained on an ImageNet dataset in our neural network architecture, to predict kidney function based on 4,505 kidney ultrasound images labeled using eGFRs derived from serum creatinine concentrations. To further extract the information from ultrasound images, we leveraged kidney length annotations to remove the peripheral region of the kidneys and applied various data augmentation schemes to produce additional data with variations. Bootstrap aggregation was also applied to avoid overfitting and improve the model's generalization. Moreover, the kidney function features obtained by our deep neural network were used to identify the CKD status defined by an eGFR of <60 ml/min/1.73 m2. A Pearson correlation coefficient of 0.741 indicated the strong relationship between artificial intelligence (AI)- and creatinine-based GFR estimations. Overall CKD status classification accuracy of our model was 85.6% -higher than that of experienced nephrologists (60.3%-80.1%). Our model is the first fundamental step toward realizing the potential of transforming kidney ultrasound imaging into an effective, real-time, distant screening tool. AI-GFR estimation offers the possibility of noninvasive assessment of kidney function, a key goal of AI-powered functional automation in clinical practice.

The C-terminal D/E-rich domain of MBD3 is a putative Z-DNA mimic that competes for Zα DNA-binding activity.

The Z-DNA binding domain (Zα), derived from the human RNA editing enzyme ADAR1, can induce and stabilize the Z-DNA conformation. However, the biological function of Zα/Z-DNA remains elusive. Herein, we sought to identify proteins associated with Zα to gain insight into the functional network of Zα/Z-DNA. By pull-down, biophysical and biochemical analyses, we identified a novel Zα-interacting protein, MBD3, and revealed that Zα interacted with its C-terminal acidic region, an aspartate (D)/glutamate (E)-rich domain, with high affinity. The D/E-rich domain of MBD3 may act as a DNA mimic to compete with Z-DNA for binding to Zα. Dimerization of MBD3 via intermolecular interaction of the D/E-rich domain and its N-terminal DNA binding domain, a methyl-CpG-binding domain (MBD), attenuated the high affinity interaction of Zα and the D/E-rich domain. By monitoring the conformation transition of DNA, we found that Zα could compete with the MBD domain for binding to the Z-DNA forming sequence, but not vice versa. Furthermore, co-immunoprecipitation experiments confirmed the interaction of MBD3 and ADAR1 in vivo. Our findings suggest that the interplay of Zα and MBD3 may regulate the transition of the DNA conformation between B- and Z-DNA and thereby modulate chromatin accessibility, resulting in alterations in gene expression.

Covariate-adjusted heatmaps for visualizing biological data via correlation decomposition.

Motivation: Heatmap is a popular visualization technique in biology and related fields. In this study, we extend heatmaps within the framework of matrix visualization (MV) by incorporating a covariate adjustment process through the estimation of conditional correlations. MV can explore the embedded information structure of high-dimensional large-scale datasets effectively without dimension reduction. The benefit of the proposed covariate-adjusted heatmap is in the exploration of conditional association structures among the subjects or variables that cannot be done with conventional MV. Results: For adjustment of a discrete covariate, the conditional correlation is estimated by the within and between analysis. This procedure decomposes a correlation matrix into the within- and between-component matrices. The contribution of the covariate effects can then be assessed through the relative structure of the between-component to the original correlation matrix while the within-component acts as a residual. When a covariate is of continuous nature, the conditional correlation is equivalent to the partial correlation under the assumption of a joint normal distribution. A test is then employed to identify the variable pairs which possess the most significant differences at varying levels of correlation before and after a covariate adjustment. In addition, a z-score significance map is constructed to visualize these results. A simulation and three biological datasets are employed to illustrate the power and versatility of our proposed method. Availability and implementation: GAP is available to readers and is free to non-commercial applications. The installation instructions, the user's manual, and the detailed tutorials can be found at http://gap.stat.sinica.edu.tw/Software/GAP. Supplementary information: Supplementary Data are available at Bioinformatics online.

RIM C2B Domains Target Presynaptic Active Zone Functions to PIP2-Containing Membranes.

Rapid and efficient synaptic vesicle fusion requires a pool of primed vesicles, the nearby tethering of Ca2+ channels, and the presence of the phospholipid PIP2 in the target membrane. Although the presynaptic active zone mediates the first two requirements, it is unclear how fusion is targeted to membranes with high PIP2 content. Here we find that the C2B domain of the active zone scaffold RIM is critical for action potential-triggered fusion. Remarkably, the known RIM functions in vesicle priming and Ca2+ influx do not require RIM C2B domains. Instead, biophysical experiments reveal that RIM C2 domains, which lack Ca2+ binding, specifically bind to PIP2. Mutational analyses establish that PIP2 binding to RIM C2B and its tethering to the other RIM domains are crucial for efficient exocytosis. We propose that RIM C2B domains are constitutive PIP2-binding modules that couple mechanisms for vesicle priming and Ca2+ channel tethering to PIP2-containing target membranes.

Acetylome of Acinetobacter baumannii SK17 Reveals a Highly-Conserved Modification of Histone-Like Protein HU.

Lysine acetylation is a prevalent post-translational modification in both eukaryotes and prokaryotes. Whereas this modification is known to play pivotal roles in eukaryotes, the function and extent of this modification in prokaryotic cells remain largely unexplored. Here we report the acetylome of a pair of antibiotic-sensitive and -resistant nosocomial pathogen Acinetobacter baumannii SK17-S and SK17-R. A total of 145 lysine acetylation sites on 125 proteins was identified, and there are 23 acetylated proteins found in both strains, including histone-like protein HU which was found to be acetylated at Lys13. HU is a dimeric DNA-binding protein critical for maintaining chromosomal architecture and other DNA-dependent functions. To analyze the effects of site-specific acetylation, homogenously Lys13-acetylated HU protein, HU(K13ac) was prepared by genetic code expansion. Whilst not exerting an obvious effect on the oligomeric state, Lys13 acetylation alters both the thermal stability and DNA binding kinetics of HU. Accordingly, this modification likely destabilizes the chromosome structure and regulates bacterial gene transcription. This work indicates that acetyllysine plays an important role in bacterial epigenetics.

Structural basis of adaptor-mediated protein degradation by the tail-specific PDZ-protease Prc.

Peptidoglycan (PG) is a highly cross-linked, protective mesh-like sacculus that surrounds the bacterial cytoplasmic membrane. Expansion of PG is tightly coupled to growth of a bacterial cell and requires hydrolases to cleave the cross-links for insertion of nascent PG material. In Escherichia coli, a proteolytic system comprising the periplasmic PDZ-protease Prc and the lipoprotein adaptor NlpI contributes to PG enlargement by regulating cellular levels of MepS, a cross-link-specific hydrolase. Here, we demonstrate how NlpI binds Prc to facilitate the degradation of its substrate MepS by structural and mutational analyses. An NlpI homodimer binds two molecules of Prc and forms three-sided MepS-docking cradles using its tetratricopeptide repeats. Prc forms a monomeric bowl-shaped structure with a lid-like PDZ domain connected by a substrate-sensing hinge that recognizes the bound C terminus of the substrate. In summary, our study reveals mechanistic details of protein degradation by the PDZ-protease Prc bound to its cognate adaptor protein.

Lactose Binding Induces Opposing Dynamics Changes in Human Galectins Revealed by NMR-Based Hydrogen-Deuterium Exchange.

Galectins are ß-galactoside-binding proteins implicated in a myriad of biological functions. Despite their highly conserved carbohydrate binding motifs with essentially identical structures, their affinities for lactose, a common galectin inhibitor, vary significantly. Here, we aimed to examine the molecular basis of differential lactose affinities amongst galectins using solution-based techniques. Consistent dissociation constants of lactose binding were derived from nuclear magnetic resonance (NMR) spectroscopy, intrinsic tryptophan fluorescence, isothermal titration calorimetry and bio-layer interferometry for human galectin-1 (hGal1), galectin-7 (hGal7), and the N-terminal and C-terminal domains of galectin-8 (hGal8NTD and hGal8CTD, respectively). Furthermore, the dissociation rates of lactose binding were extracted from NMR lineshape analyses. Structural mapping of chemical shift perturbations revealed long-range perturbations upon lactose binding for hGal1 and hGal8NTD. We further demonstrated using the NMR-based hydrogen-deuterium exchange (HDX) that lactose binding increases the exchange rates of residues located on the opposite side of the ligand-binding pocket for hGal1 and hGal8NTD, indicative of allostery. Additionally, lactose binding induces significant stabilisation of hGal8CTD across the entire domain. Our results suggested that lactose binding reduced the internal dynamics of hGal8CTD on a very slow timescale (minutes and slower) at the expense of reduced binding affinity due to the unfavourable loss of conformational entropy.

Entropic stabilization of a deubiquitinase provides conformational plasticity and slow unfolding kinetics beneficial for functioning on the proteasome.

Human ubiquitin C-terminal hydrolyase UCH-L5 is a topologically knotted deubiquitinase that is activated upon binding to the proteasome subunit Rpn13. The length of its intrinsically disordered cross-over loop is essential for substrate recognition. Here, we showed that the catalytic domain of UCH-L5 exhibits higher equilibrium folding stability with an unfolding rate on the scale of 10-8 s-1, over four orders of magnitudes slower than its paralogs, namely UCH-L1 and -L3, which have shorter cross-over loops. NMR relaxation dynamics analysis confirmed the intrinsic disorder of the cross-over loop. Hydrogen deuterium exchange analysis further revealed a positive correlation between the length of the cross-over loop and the degree of local fluctuations, despite UCH-L5 being thermodynamically and kinetically more stable than the shorter UCHs. Considering the role of UCH-L5 in removing K48-linked ubiquitin to prevent proteasomal degradation of ubiquitinated substrates, our findings offered mechanistic insights into the evolution of UCH-L5. Compared to its paralogs, it is entropically stabilized to withstand mechanical unfolding by the proteasome while maintaining structural plasticity. It can therefore accommodate a broad range of substrate geometries at the cost of unfavourable entropic loss.