Abstract

Open in a separate window

Misfolding and aggregation of the protein remain some of the most common phenomena observed in neurodegeneration. While there exist multiple neurodegenerative disorders characterized by accumulation of distinct proteins, what remains particularly interesting is the ability of these proteins to undergo a conformational change to form aggregates. TDP-43 is one such nucleic acid binding protein whose misfolding is associated with many neurogenerative diseases including amyotrophic lateral sclerosis (ALS) and fronto-temporal lobar degeneration (FTLD). TDP-43 protein assumes several different conformations and oligomeric states under the diseased condition. In this review, we explore the intrinsic relationship between the conformational variability of TDP-43 protein, with a particular focus on the RRM domains, and its propensity to undergo aggregation. We further emphasize the probable mechanism behind the formation of these conformations and suggest a potential diagnostic and therapeutic strategy in the context of these conformational states of the protein.

Neurodegeneration and Protein Aggregation

All cells, including neurons, experience death. In particular, neurodegeneration involves a gradual loss of specific neurons, which coincides with the clinical syndromes witnessed typically in a rapidly aging population. Neurodegeneration is an umbrella term used to define a variety of diseases, some of which include Alzheimer’s disease (AD),^1,2 Parkinson’s disease (PD),³ Huntington’s disease (HD),⁴ Guam-Parkinsonism,⁵ amyotrophic lateral sclerosis (ALS),⁶ and fronto-temporal lobar degeneration (FTLD).⁷ Each of these distinct diseases involves the deposition or aggregation of distinct proteins in different anatomical regions of the brain, often reflecting the symptoms of the diseases. For example, in AD, we observe major deposition of Aβ and tau protein in the medial temporal lobe and neocortical structures, and nonfunctionality of this region results in cognitive impairment, including memory loss, difficulty in reading and writing, loss of impulse control, etc.^8,9 In contrast, for ALS, the deposition of the TAR DNA binding protein is primarily seen in the upper and lower motor neurons. The degeneration of the upper motor neuron gives rise to hyper-excitability and spasticity, while the degeneration of the lower motor neuron results in muscle weakness, fasciculation, and muscular atrophy, etc.¹⁰ Among the causative factors that have been consistently observed in neurodegenerative diseases are chronic environmental stress, deposition of protein aggregates, and inefficient clearance of the protein aggregates from the cell due to dysfunctional ubiquitin-proteasome system, nonfunctional autophagosomal/lysosomal system, excitotoxic insult, synaptic failure, mitochondrial dysfunction, and neuroinflammation.¹¹ Compelling evidence suggests that oligomerization of the proteins leading to the deposition of amyloids is the fundamental cause of neuronal loss and degeneration.^12,13 Different diseases show deposition of different proteins such as Aβ and tau in Alzheimer’s, α-synuclein in Parkinson’s, polyglutamine protein in Huntington’s disease, and TDP-43 in ALS and FTLD. Despite the differences in their primary sequence and overall structural fold of the soluble precursor protein, the aggregates are characterized by highly ordered cross-β sheet rich structure.^14,15 Therefore, there appears to be a common underlying mechanism that might populate generic conformational states along the energy landscape of aggregation-prone proteins. Characterization of these conformational states and understanding the kinetic barriers between them could shed light on new avenues to detect and target pathological protein.

In this review, we discuss how alterations in local environmental conditions influence the conformational plasticity of proteins in general, mainly examining this phenomenon through the lens of energy landscape. Our discussion centers on the TDP-43 protein with a particular focus on understanding the role of RRM domains in aggregation and lists out the potential mechanisms involved behind the formation of different conformational states for TDP-43 protein.

TDP-43 is a Major Protein Aggregate in ALS and FTLD

The gene coding for TAR DNA binding protein is well conserved among Drosophila, human, mouse, and Caenorhabditis elegans.¹⁶ The 43 kDa protein, TDP-43, was first discovered in 1996 for its role in binding to the TAR DNA of HIV and repressing the transcription process.¹⁷ TDP-43 was later studied for its role as an exon-skipping promoter during the splicing of apolipoprotein A-II (apoA-II) and cystic fibrosis transmembrane regulator (CFTR) transcripts. Despite its various vital functions, TDP-43 protein was found to be a major protein deposited in the spinal cord and brain of patients suffering from ALS and FTLD.^18,19 Since then, several neurodegenerative diseases such as Alzheimer’s, Parkinson’s, Huntington’s, Guam-Parkinsonism dementia, and Perry’s syndrome have shown intraneuronal deposition of TDP-43 aggregates.²⁰⁻²³ More recently, limbic-predominant age-related TDP-43 encephalopathy-neuropathological change (LATE-NC) was recognized as a brain disorder affecting older adults showing accumulation of TDP-43.²⁴ Together, these diseases showing deposition of TDP-43 protein were termed TDP-43 proteinopathies.²⁵

TDP-43 Is a Nucleic Acid Binding Protein with an Intrinsically Disordered Region

Structurally, TDP-43 protein has four major domains, an N-terminal domain: NTD (aa, 1–76), two RNA recognition motif domains: RRM1 (amino acids 104–176) and RRM2 (amino acids 192–262) linked via a linker region, and a C-terminal domain: CTD (aa 274–414).²⁶⁻²⁹ The protein has two signal sequences: a nuclear localization signal (NLS; aa 82–98) and a nuclear export signal (NES; aa 239–250), through which it shuttles between the nucleus and cytoplasm to perform its function.³⁰ TDP-43 is primarily a nuclear protein but in the case of ALS, the protein is redistributed, showing an increased deposition in the form of inclusions in the cytoplasm.³¹ Due to high aggregation propensity and poor solubility, the complete structure of TDP-43 remains unresolved until now, but the role of individual domains in aggregation is being addressed.

Experimental evidence suggests that TDP-43 protein remains in a monomer–dimer equilibrium inside the cell under normal conditions.^32,33 The process of dimerization is shown to be facilitated by the N-terminal domain of TDP-43.³⁴⁻³⁶ It has been reasoned that the dimeric NTD enhances the solubility of the protein through its pre-mRNA splicing activity and protects against inclusion formation.³⁷ In contrast to this idea, studies have also suggested that dimerization may be involved in the aggregation of the protein.³²⁻³⁴ These studies imply that the N-terminal region may increase the local concentration of the protein which can serve as a prerequisite for aggregation reaction.^38,39 The C-terminal domain of TDP-43 consists of a glycine-rich region and a segment containing polar uncharged amino acids, such as glutamine and asparagine (Q/N). This domain architecture is highly comparable to the prionlike domains of several proteins such as Sup35, Fused-In-Sarcoma (FUS), TATA-box binding protein associated factor 15 (TAF-15), and Ewing sarcoma breakpoint region 1 (EWSR1) and heterogeneous nuclear ribonucleoproteins (hnRNPs) family of proteins.⁴⁰ A vast number of studies in the literature focus on the contribution of the CTD toward the aggregation of the protein. A chief reason for using the C-terminal region as a model domain is because it harbors the majority of the mutations and phosphorylation sites.⁴¹⁻⁴⁴ A detailed review of the CTD of TDP-43 and its involvement in the diseased condition has been reviewed previously.^45,46 Additionally, it has also been seen that C-terminal fragments of sizes ~25–35 kDa are observed in the inclusion bodies.^47,48 Cleavage of these fragments occurs at the caspase site located on the RRM2 domain of TDP-43 protein,⁴⁷ prompting us to hypothesize that partial unfolding of the RRM domain might also be pivotal in the aggregation process.

Role of RRM Domains of TDP-43 in Function and Aggregation

Conformational Changes within the RRM Domains of TDP-43 during Stress Conditions: Formation of Oligomer with Molten Globulelike Structure

An NMR study, coupled with X-ray fiber diffraction data, highlights a strong similarity in the structure of amyloid fibrils formed by full-length TDP-43 and TDP-35 (90–414 residue).⁶¹ The C-terminus showed a polymorphic structure when forming fibrils alone or in combination with other domains. This suggests that both the N-terminal and RRM domains are crucial in forming the core structure of TDP-43.⁶¹ The RRM domains of TDP-43 have been studied individually⁶² and in tandem^63,64 and shown to undergo aggregation. Two regions within the RRM domains have been identified to play a vital role in the aggregation process: residues 166–173 in RRM1 and residues 246–255 in RRM2.^56,65

Studies from our lab showed that protonation of residue H166 causes destabilization of the RRM domains causing it to partially unfold and result in the formation of nativelike oligomers with an intact secondary structure and a broken tertiary structure (molten globule).^63,64,66 Thus, the protonation–deprotonation of His166 residue regulates the assembly disassembly of protein molecules (Figure ​Figure11).^63,66 Upon prolonged stress, these oligomers can undergo a conformational change to form ThT-positive aggregates, thus acting as a bridge to link the native and the aggregation energy landscape (Figure ​Figure11).⁶⁷ Our time-dependent analysis of the aggregation process uncovered a multistep structural transformation (Figure ​Figure22), which we have modeled to reveal four major steps, each characterized by varying energy barriers. The sequence of these steps depends on the protein concentration (Figures ​Figures22 and ​and33).⁶⁷ Our data suggest that the initial partial unfolding, triggered by side-chain disruption from protonation, is sufficient to overcome the major energy barrier and drive the protein through a series of conformational changes following a linear (isodesmic) polymerization mechanism (Figure ​Figure33).⁶⁷ Consequently, the population and stability of the different states of protein (monomer, oligomer, or aggregates) were shown to be highly dependent on the intramolecular and intermolecular interactions, which are modulated by changing the electrostatic conditions through pH and salt in the solution (Figure ​Figure11).^68,69 Several in vitro studies have demonstrated that a change in the pH destabilizes the protein molecules and induces a conformational change to form aggregates under prolonged stress conditions.^64,70−73 In some proteins, protonation has been shown to act as a molecular switch controlling the equilibrium between native and the partially unfolded molecules that can prime misfolded structures.^{64,66,74−76} Consistent with this idea, it has been shown that assembly disassembly is one of the most energy-efficient ways to mitigate stress, unlike post-translational modification or degradation.⁷⁷ Additionally, the time scales of formation of these higher-order assemblies are also less, allowing the cells to quickly respond to stress-like conditions.⁷⁷ Formation of such molten globule states were also observed for other neurodegenerative proteins such as β2-microglobulin (β2m),^78,79 prion,⁸⁰ and p53.^81,82

Open in a separate window

Figure 1

Formation of metastable, molten-globulelike oligomers that bridges the folding and aggregation energy landscapes. Reproduced from ref (63). Copyright 2019 American Chemical Society. The cartoon on the right represents the conformational changes in the RRM domains of TDP-43 driven by altered electrostatics. Reproduced from ref (68). Copyright 2023 American Chemical Society.

Open in a separate window

Figure 2

RRM domains of TDP-43 undergo a multistep structural transformation upon protonation of a buried ionizable residue, leading to the formation of a fibrillar structure. The kinetic data presented here monitors the time scales of different structural conformations during the aggregation process. (A) Changes in the protein’s secondary structure, monitored through far-UV circular dichroism. (B) Formation of ThT-binding pockets. (C and D) Increase in protein assembly size as tracked by dynamic light scattering. (E) Changes in the local tertiary structure, observed via tryptophan fluorescence for the protein before (blue line) and after (yellow line) transfer into aggregation buffer. Data from these panels were analyzed using single or double exponential equations, and (F) the apparent rate constants derived from these fits. The black-filled circles in panels A, B, and D indicate signal of the protein before transferring into the aggregation condition. Adapted with permission from ref (67). Copyright 2023 John Wiley and Sons.

Open in a separate window

Figure 3

Multistep model for the aggregation of the RRM domains of TDP-43. This kinetic model is built using the apparent rate constants obtained from Figure ​Figure22. The model illustrates how the protein concentration influences the rate-limiting step in the aggregation process. It depicts four major stages in the aggregation process, each corresponding to a distinct change in the conformation and size of the protein assembly over time. Adapted with permission from ref (67). Copyright 2023 John Wiley and Sons.

Electrostatics in the cell are often altered either by post-translational modification of side chains or by the presence of mutations in the protein. For example, two crucial electrostatic mutations are present within RRM domains, P112H and D169G.^83,84 The mutation D169G in the RRM1 domain does not affect the aggregation propensity but induces a local conformational change due to the loss of hydrogen bond with T115 residue.⁸⁵ We also demonstrated that these mutants do not alter the global structure of the tandem RRM domains. However, under pH and salt stress, the aggregation kinetics of P112H increases significantly as compared to the wild-type sequence, while D169G remains largely comparable to the wild-type sequence. Under oxidation stress or upon binding to zinc ions, these domains undergo oligomerization or formation of ThT-positive aggregates.⁸⁶⁻⁸⁸ Structural studies have also revealed that the RRM2 domain of TDP-43 populates an intermediate state that has been proposed to act as a link between the folding and the misfolding energy landscape.⁸⁹ All these studies reinforce the notion that destabilization of RRM domains through changes in protonation, oxidation, or zinc-binding can induce partial unfolding and expose hydrophobic patches in the protein, leading to aggregate formation. Conversely, stabilization of protein through RNA/DNA binding has been shown to mitigate the process of aggregation.^58,59,66

RRM domains, in general, contribute to the architecture of many aggregation-prone proteins such as FUS, hnRNPs, TAF-15, and EWSR1, which are associated with various neurodegenerative diseases.^40,90 Remarkably, around ~240 proteins within the human proteome contain prionlike low complexity domain, many of which also harbor a canonical RNA Recognition Motif.^10,91 Numerous studies have focused on the role of low complexity domain in misfolding and aggregation, but investigations with emphasis on the role of RRM domain in aggregation is only now evolving.⁹²

Conformational Variability in the Aggregates from Brain Patients

Polymorphism in the Amyloid Fold of the Aggregates

Under diseased conditions, both full-length TDP-43 and truncated C-terminal fragments (CTFs) form granulo-filamentous assemblies with diameters of 10–15 nm.^118−121 Cryo-EM structures from two individuals suffering from ALS with FTLD degeneration (type B) revealed an identical double-spiral-shaped fold, distinct from those observed using recombinant TDP-43 invitro.^122−124 This ordered core was formed by the fragment (G282–Q360) with flanking regions forming a fuzzy coat. In contrast, TDP-43 assemblies from type A FTLD-TDP individuals exhibited a distinct extended structural fold resembling a chevron.¹²⁵ Although the same region contributed to the formation of the ordered core, the authors hypothesized that the local structural variations within individual filaments might give rise to alternative conformations. These structural differences may explain the distinct pathologies observed in different types of FTLD-TDP, which are characterized by the brain distribution of aggregated TDP-43.¹²⁵ The different local environmental conditions might likely influence the conformation, giving rise to distinct amyloid-fold.

Under in vitro conditions, different environmental conditions such as pH, temperature, mechanical agitation, or ionic strength give rise to different assemblies.^126,127 However, once a scaffold is formed, the growth continues, keeping the atomistic order the same, even when the solution condition is changed. A specific polymorph can seed the same type of architecture even under different solution conditions, similar to those observed in crystalline systems.¹²⁸ This has been studied in the case of Aβ₄₀, where the formation of striated ribbons and twisted fibrils formed under agitation and quiescent conditions, respectively, when used to act as a seed, propagated similar structures under different conditions.¹²⁹ This might be due to the intrinsic stability of the extremely ordered architecture and the high energy barriers between the different polymorphs.

In general, polymorphism seen in the amyloid can be attributed to the side-chain conformation, backbone conformation, or supramolecular assembly. Several different proteins such as Aβ40,^130−133 α-synuclein,^134−136 islet amyloid polypeptide (IAPP),¹³⁷ tau,^138,139 and prion¹⁴⁰ show polymorphic amyloid fibrils. A range of different conformations adopted by the protein can often contribute distinctly to its toxicity and transmission.^141,142 Unlike a native fold which is evolutionarily selected, the formation of an amyloid fold is not optimized by evolutionary pressure, and hence, there is a multitude of different arrangements that becomes accessible to the polypeptide sequence.¹²⁸

Modulation of the Energy Landscape: Targeting the Native State, Intermediate, and the Aggregated States of the Protein

The energy landscape concept illustrates the conformational space available for a polypeptide sequence to form a structure, resembling a funnel where the ruggedness depends on various factors (Figure ​Figure44).^143,144 In a complex microenvironment, the energy landscape will typically be rough for a multifunctional protein, which can often be correlated to frustration in the folding process.¹⁴⁵ Experimental evidence suggests that for a well-folded globular protein to undergo the formation of amyloid or amorphous aggregates, it requires partial unfolding and destabilization of its native structure (Figure ​Figure44).^{12,13,146−150} Many aggregation-prone proteins show the formation of partially unfolded structures under destabilizing conditions such as β2 microglobulin,⁷⁶ prion,^151,152 superoxide dismutase 1 (SOD1),¹⁵³ and transthyretin (TTR).¹⁵⁴ Transient formations of such partially unfolded species allow specific electrostatic, hydrophobic, and intermolecular interactions that enable the protein to undertake key structural rearrangements to undergo fibril formation.¹⁴⁷

Open in a separate window

Figure 4

Folding energy landscape indicates the search for a polypeptide to achieve its native state. The aggregation energy landscape illustrated by the dark brown color represents the different kinds of intermediates and assembly formation under stresslike conditions. The various energy minima on the landscape represents the possibility of formation of different intermediates and their respective stability.

Given that the structural conformations formed early during the aggregation process are often reversible, targeting these conformations by stabilizing them and driving the equilibrium away from higher-order aggregates offers an opportunity to develop therapeutic modulators (Figure ​Figure55). For example, tafamidis is a small molecule kinetic stabilizer that stabilizes the native tetrameric state of TTR and is FDA-approved for TTR amyloidosis.¹⁵⁵ Similar kinetic stabilizers are being developed to stabilize the native dimer of the immunoglobulin light chain, which can form aggregation-prone conformations in the immunoglobulin light chain amyloidosis.¹⁵⁶ Native state stabilization as a therapeutic strategy has more recently been adapted for treating TDP-43 proteinopathies. Leveraging Cryo-EM structures of TDP-43 fibrils, the study proposed to develop a high-throughput screening assay to identify and develop a native state stabilizer.¹⁵⁷ One such native state stabilizer for TDP-43 can be RNA which has been consistently reported to stabilize the protein and prevent it from undergoing aggregation.¹⁵⁷

Open in a separate window

Figure 5

Complex multistep aggregation process and the heterogeneity observed during the aggregation reaction. The pathway illustrates the formation of different intermediates, both on-pathway and off-pathway, as well as generation of different assemblies, including oligomers. Potential therapeutic strategies include stabilizing the native state, diverting the reaction to off-pathway intermediates, or developing small molecules that bind to the aggregates.

Another approach would be to target the on-pathway oligomers and intermediates during the aggregation process (Figure ​Figure55). For example, cysteine-reactive small molecules, designed from conformational motifs in dimeric and hexameric intermediates of human β2m, promote off-pathway tetramer production that inhibits aggregation.¹⁵⁸ Yet another approach is to directly target the aggregates formed. Along these lines, a recent study successfully utilized a molecule “baicalein”, a structure-correcting agent that can convert the existing TDP-43 aggregates into a functional oligomeric state in the disease model.¹⁵⁹ While baicalein was shown to effectively mitigate the proteinopathies in the mouse model, it still remains to be seen if it can effectively refold the TDP-43 protein into a nonpathological conformation.¹⁶⁰ An interesting tactic developed by Eisenberg et al. involved designing mini proteins that could specifically target and cap the growing ends of α-synuclein, Aβ, and tau fibrils.¹⁶¹ Using this approach, they were successfully able to slow down the seeded aggregation of these proteins in vitro, in human cells, and in C. elegans models of AD and PD.¹⁶¹ Multiple studies have adapted these strategies to specifically target the fibrillar structure and slow down or inhibit the aggregation of proteins.^162−165 More detailed kinetic design principles used for guidance and development of inhibitors for fibril growth can be found in a recently published study.¹⁶⁶

Conclusion

Given the significant variation in TDP-43 proteinopathies, characterized by diverse conformations of the protein, understanding and characterizing these different structures are crucial for developing therapeutic approaches. Potential therapeutic strategies include using natural ligands like RNA to stabilize the protein, developing small molecules to target the TDP-43 fibril core, or creating phosphomimetic substitutions. This review provides a broad perspective on the various conformational states observed for the TDP-43 protein. By utilizing the framework of the energy landscape, we explore how targeting these key structures could open new avenues for treating these diseases.

Acknowledgments

Special Issue

Published as part of ACS Omegaspecial issue “Celebrating the 25th Anniversary of the Chemical Research Society of India”.

Author Present Address

^§ Johns Hopkins University, Baltimore, Maryland 21205, United States

Notes

The authors declare no competing financial interest.

References