*3.3. Modulation of IDPs by Chaperones and Co-Chaperones*

Aggregates produced in neurodegenerative diseases have been shown to respond to changes in levels of molecular chaperones, suggesting the possibility of therapeutic intervention and a role for chaperones in disease pathogenesis [217]. The heat shock protein Hsp90 promotes neurodegenerative disorders indirectly [218]. Tau protein accumulation is regulated by a (Hsp90) chaperone system. This chaperone is able to bind Tau, causing a conformational change that allows tau's phosphorylation by glycogen synthase kinase (GSK3β), leading to tau aggregation [219]. The inhibition of this chaperone results in the reduction of tau phosphorylation levels, due to reduction of GSK3β levels [220]. Another approach was performed, the use of a co-chaperone of Hsp90, ATPase homolog 1 (Aha1), this protein is an activator of Hsp90. This approach promoted the increase of the production of aggregated tau in vitro and in mouse model of neurodegenerative disease. Moreover, inhibition of Aha1 reduced tau accumulation in cultured cells. Thus, Aha1 is an interesting target to the treatment of Alzheimer's disease [221].

The Hsp70 is a protein stabilizer, has a cellular protection against neurodegeneration of the central nervous system [218]. Members of the Hsp70 family, such as Hsp70 and Hsc70, bind to misfolded proteins and somehow send them to the lysosome–autophagy pathway or ubiquitin–proteasome system for degradation [222,223]. Folding and degradation of proteins are linked through co-chaperones, such as C-terminus of HSP70-interacting protein (CHIP) and HSJ1 (DNAJB2) [224] which regulate the decisions determining whether misfolded proteins are refolded or degraded. The CHIP is associated with α-synuclein inclusions and act as a co-chaperone, altering its aggregation and enhancing the degradation of the misfolded α-synuclein [225].

Another important chaperone is cyclophilin 40 (CyP40) that is a cis/trans peptidyl-prolyl isomerase (PPIase) and is involved in regulation and orientation of proline residues [226,227]. Tau protein is rich in proline residues and its residues, usually found in β-turns, are involved in tau aggregation propensity [228]. Based on this information, Baker and coworkers [229] demonstrated that CyP40 possess the ability to dissolve amyloids fibrils in vitro. Nuclear magnetic resonance experiments showed that CyP40 acts specifically on proline rich residues performing the disaggregation of tau fibrils and oligomers. This cyclophilin could also interact with others aggregated proteins containing proline, like α-synuclein [229].

#### **4. Known Drugs Acting on IDPs**

Despite IDPs abundance in eukaryotes, currently there are no FDA-approved drugs specifically targeting these proteins, only experimental and speculative ones (i.e. drugs that have been evaluated by the United States Food and Drug Administration agency and had their marketing sanctioned). Some experimental drug examples are prevalent in the literature, such as those targeting p53-MDM2, c-Myc-Max, and EWS-Fli1 complexes, while some others are less discussed [230–233]. In this section we provide an overview of the pharmacological modulation of IDPs, neurodegenerative and otherwise.

Intrinsically disordered proteins are normally considered aggregation-avoidant, due to their high proportion of charged residues (as opposed to patches of folding-inducing, hydrophobic residues). Such "non-folding" plasticity is proposed to be advantageous for proteins with multiple partners [234,235]. However, some of the IDPs, as described earlier, are found in conformational diseases and amyloid formation (e.g., in Alzheimer's disease, Parkinson's disease). The suppression of fibril formation, thus, is of therapeutic interest. Drug candidates in this front include molecular tweezers (Table 1), which are ligands designed to bind lysine and arginine specifically, perturbing aggregation [236–238]. These positively-charged residues are prone to interact with negatively-charged regions in the fibril-forming monomers [237]. The SEN1576 compound, a 5-aryloxypyrimidine inhibitor of synaptotoxic Aβ aggregation (Table 1) was shown to be safe and orally bioavailable with good brain penetration [239].

Fragments of amyloid fibrils also served as templates for non-natural amino acid inhibitors of amyloid fibril formation (D-TLKIVW) [240], while the ELN484228 (Table 1) compound was shown to be protective in cell models for vesicular dysfunction via α-Synuclein [235]. Alterations of the neuroleptic agent chlorpromazine allowed for enhanced 20S proteasome activation, inducing degradation of IDPs, such as tau and α-synuclein, but not of structured proteins [241]. These chlorpromazine-derived molecules, despite showing noteworthy potential (even as tools to study the proteasome 20S gate regulation), may interfere with other, physiological, non-pathologic disordered proteins to a still unstudied extent. A naphthoquinone-tryptophan hybrid (NQTrp) (Table 1) was shown to be effective in model systems for tau aggregation [242].

Regarding tumor-associated IDPs, the most commonly mutated gene in human cancers, the tumor suppressor protein p53, is key to cell cycle signaling. It is regulated by binding to various partners, including MDM2 and Taz2 [243–245]. Despite being highly disordered, p53 is not itself the target for the currently screened drug candidates aiming the p53-MDM2 complex. Instead, these ligands aim to occupy the p53-binding site in MDM2. The inhibitors nutlins (Table 1) (*cis*-imidazoline analogs currently in phase-I clinical trials), were shown to be potent against multiple cancerous cell lines, including breast cancer, colorectal cancer, lung cancer, osteosarcoma, prostate cancer, and renal cancer [246–249].

The c-Myc-Max complex involves the IDP transcription factor c-Myc that is activated by binding to Max, being expressed constitutively in various cancer cells [250]. Inhibitor candidates target the disordered c-Myc in this case, including peptidomimetic inhibitors [249,250], the small molecules 10058-F4, 10074-G5 (Table 1), and some others [250–257]. The oncogenic fusion protein EWS-Fli1 is an IDP exclusively present in Ewing's sarcoma [257]. As with c-Myc-Max inhibitors, the small molecule inhibitor YK-4–279 (Table 1) targets the disordered EWS-Fli1 protein directly [257,258].

The AF9-AF4 dimer is found in acute leukemias and is composed by two disordered fusion proteins. The AF9 protein is responsible for turning hematopoietic cells oncogenic [259,260]. The AF4-derived peptide of amino acid sequence PFWT was shown to inhibit AF9 when used in combination with established chemotherapeutic agents [261,262], while some non-peptidic inhibitor candidates have been identified by high-throughput screening [263]. The protein-tyrosine phosphatase 1B (PTP1B), a reticular non-transmembrane enzyme, has been validated as therapeutic target for diabetes, obesity, and breast cancer, due to its role as negative regulator of insulin and leptin signaling [264]. Protein-tyrosine phosphatase 1B has an elongated disordered carboxy-terminus, to which trodusquemine (Table 1) (MSI-1436, a natural product) binds [265]. This aminosterol

acts allosterically, stabilizing an inactive form of the enzyme by binding to a non-catalytic disordered site [265].


**Table 1.** Known drugs acting on IDPs (selected examples).

<sup>\*</sup> Unless otherwise specified, these names are directly taken from their lead-compound coding and have no meaning on their own. EWS: Ewing Sarcoma; MDM2: oncoprotein (murine double minute 2); Myc: Myelocytomatosistranscription factor homolog; NQTrp: naphthoquinone-tryptophan hybrid; PHF6: Tau-derived peptide (amino acid sequence VQIVYK); PTP1B: protein-tyrosine phosphatase 1B.

Inhibitors of α-synuclein aggregation are considered a promising approach. From in vitro studies, a few lead molecules were identified, such as EGCG (epigallocatechin gallate) [266]. Iron is known to induce the aggregation of α-synuclein. Deferiprone is an iron chelator used in thalassemic patients. Two clinical trials have shown a decrease in iron content in the substantia nigra of some PD patients, while a trend for improved motor scores was seen for all enrolled patients [267]. Although copper can induce aggregation of α-synuclein in vitro, levels of copper in the substantia nigra of PD patients are up to 50% lower than that of age-matched controls. The Cu2+ complex of diacetylbis-(4-methylthiosemicarbazone), called Cu2+(atsm), showed neuroprotective action in different animal models of PD [268], prompting for a phase I trial. From in vivo studies, promising results were obtained for KYP-2047, an inhibitor of prolyl oligopeptidase, an enzyme shown to interact with α-synuclein [269–271]; the diphenyl-pyrazole compound anle138b, was shown to cross the blood-brain barrier of mice and to reduce aggregation of α-synuclein [269]. The NPT200-11 compound prevented the formation of oligomers of α-synuclein and improved neuropathological symptoms in transgenic mice [272] and it has already been subjected to a phase I clinical trial. Another promising compound, NPT088, is a fusion protein of a general amyloid interaction motif derived from a bacteriophage [273] and a fragment of human immunoglobin, developed by Proclara Biociences, entered a phase I clinical trial for AD [274]. Phase II clinical trial of NPT088 is expected to include PD patients. Intrabodies, a single chain variable fragment of immunoglobulin expressed intracellularly, have been developed to target oligomeric and fibrillary α-synuclein, conferred neuroprotection, apparently by shifting the dynamics of the aggregation process [275–277]. Addition of a proteasome-addressing sequence to intrabodies targeted pathological forms of α-synuclein to degradation, NbSyn87PEST, directed towards the C-terminal region, and VH14PEST, directed against the NAC hydrophobic interaction domain, effectively degraded α-synuclein in cultured cells [278]. In rats overexpressing wild-type α-synuclein, these proteasome-targeted intrabodies (or nanobodies) decreased the levels of pathological aggregates, increased striatal dopamine levels and improved motor function [279]. Research in this promising field moves to find ways to deliver these compounds in adequate levels in specific areas of the brain, probably by using viral vectors.

Immunotherapies against α-synuclein, based on the evidence of an extracellular pathological protein during spreading of PD to different brain structures, show promising results in animal models [280]. Besides opsonization of the pathological protein for clearance, it is likely that antibodies could block further oligomerization of α-synuclein. Both passive (humanized monoclonal antibodies) and active (vaccine) immunization are being pursued. Pharmaceutical companies have joined the efforts and early clinical trials have been concluded or are under way. A brief description of the more advanced planned immunotherapies follows. A phase I trial was conducted by Roche for PRX002, a monoclonal antibody against the C-terminus of α-synuclein. It was well tolerated and reduced by 96% the levels of serum α-synuclein [281,282]. Affitope PD01A, a synthetic α-synuclein-mimicking peptide developed by Affiris for active immunization, had the first pilot study in 21 PD patients concluded in May 2018. It elicited a specific antibody response and showed good safety and tolerability profiles in a long-term (4 years) outpatient setting. Results of Affitope PD03A phase I clinical trial indicated no severe off-target effects, and a dose-dependent production of antibodies that cross-reacted with the intended α-synuclein epitope. Results from animal studies demonstrated that the antibodies raised against these antigens crossed the blood-brain barrier, decreasing the levels of aggregated α-synuclein, thereby improving motor function [283,284].

Regarding tauopathies, various therapeutic approaches have been tested, aiming to inhibit aggregation of tau, either directly or by preventing its interaction with some partners, and removal of toxic conformers and fibrillated tau [80]. However, the enormous effort put on finding ways to revert or delay the neurodegeneration symptoms associated to fibrillar tau, or to prevent the onset of tauopathies, has been so far unsuccessful, partly due to the intrinsically disordered nature of tau, which hampers drug design based on structural approaches. Small molecules that inhibit tau aggregation in vitro are considered promising leads to anti-tauopathy drugs [285] and the number of new tau inhibitory molecules grows steadily [286]. Nevertheless, there are unanswered questions regarding their effectiveness in vivo and the potential non-specific effects on normal tau physiology that could impact heavily on the CNS. The most studied small inhibitory molecules belong to distinct chemical groups, such as phenotiazines, cyanines, rhodanines, and arylmethines [287,288]. Peptides derived from neuroprotective proteins like NAP (amino acid sequence NAPVSIPQ) and D-SAL (all D-amino acid sequence SALLRSIPA) [289,290], enantiomeric peptides [291], and RNA/DNA aptamers [292] are also attractive components of future anti-tauopathy therapies. Some natural molecules present in cellular medium, such vitamin B12 [293] and 8-nitro-CGMP [294], are known to inhibit tau aggregation through oxidation of its cysteine residues. Drugs that bind tau, inducing formation of intramolecular disulfide bonds, such as methylene blue [295] or cinnamon derivatives [296], are potential frames for developing specific tau aggregation inhibitors. Taking into account that accumulation of hyperphosphorylated tau is a hallmark of AD and other neurodegenerative disorders, inhibitors of kinases, particularly of glycogen sintase kinase 3β and of Fyn, a member the Src-family of non-receptor tyrosine kinases, have drawn much attention for their anti-tauopathy potential [80]. Another strategy focuses on dual inhibitors that would interfere on tau aggregability and simultaneously block its interaction with protein partners, particularly kinases [297,298]. Other attempts to develop an anti-tauopathy drug have focused on inhibiting tau interaction with proteases like beta-secretases [299], caspases [300], and calpain [301], and chaperones such as Hsp90 [302], among others.

Tau-targeted immunotherapy began in 2013 [303], and since then a dozen of different types of immunological strategies were subject of clinical trials, including two active immunizations (vaccines) and humanized monoclonal antibodies directed towards distinct tau epitopes aiming passive immunization (reviewed in References [304,305]). These clinical trials are still at early phases and only limited data on the outcomes have been disclosed so far [306,307]. To achieve a successful immunotherapy to treat tauopathies, antibodies should be capable of neutralizing at least one of the many diseased isoforms of tau, either intracellularly or in the extracellular space, and interrupt the processes that lead to tau fibrillation and the neuron-to-neuron spreading. Ideally, the antibodies do not bind to normal tau and are able to cross the blood–brain barrier. In the case of a tau vaccine, the senescence of the immune system of the elderly has to be considered [304]. AADvac1 was conceived as the first vaccine against AD, using as immunogen a tau peptide previously identified to be essential for its pathological aggregation. Active immunization with this peptide elicits antibodies against a stretch of tau's primary sequence (amino acid residues 294–305) and to conformational epitopes as well, targeting mainly extracellular tau, reducing its oligomerization. Tested in different animal models of AD, AADvac1 raised a protective humoral immune response with antibodies that discriminated between normal and pathological tau, reduced the level of neurofibrillary pathology in rat brains and lowered the content of disease-specific hyperphosphorylated tau [308]. Phase I clinical trial of AADvac1, conducted in 2013–2015 in patients aged 50–85 years with mild-to-moderate AD immunized weekly for 12 weeks, revealed a favorable safety profile and 29 out of 30 patients given AADvac1 developed an IgG response [309]. After 72 weeks, and booster doses of AADvac1, patients who had developed higher IgG titers showed lower hippocampal atrophy and cognitive decline rates and only mild adverse side effects [310]. A second active immunotherapy against tau has the compound ACI-35 as the immunogen, a peptide containing tau's phospho-epitope pS396/pS404, in a liposome-based formulation able to elicit antibodies against abnormal hyperphosphorylated tau in P301L tau-mice [311].

Attempts of passive immunotherapy utilize humanized monoclonal antibodies, mostly of IgG1 or IgG4 isotypes, which are directed towards stretches of tau's primary sequence known to be involved in the oligomerization of the protein, or to extracellular seeding-capable forms of truncated tau [304]. Phase I clinical trial of ABBV-8E12, one of such humanized monoclonal antibodies [312], revealed a satisfactory safety profile in 30 patients with the progressive supranuclear palsy tauopathy, receiving single doses (2.5 to 50 mg/kg) of the antibody, with no signs of immunogenicity against it [313]. Phase I trials of other two anti-tau antibodies, C2N-8E12 and BMS-986168, were also conducted [307].

Something that is noteworthy, and that demonstrates the close interplay between amyloid β peptide and tau in causing neurodegenerative diseases, therapeutic interventions aimed at one pathology can ameliorate symptoms of the other. This is the case of immunization of triple transgenic AD-like mice with a full-length DNA of amyloid β1–42 peptide, which showed a 40% reduction in the brain content of the amyloid β1–42 concomitant with a 25–50% decrease of total tau and different phosphorylated tau isoforms [314]. Conversely, passive immunization with antibodies against tau's fragments 6–18 and 184–195 protected triple transgenic AD-like mice by reducing amyloid precursor protein in the CA1 region of hypothalamus and in amyloid plaques [315,316].

#### **5. Status and Challenges in Drug Development for IDPs**

The limited number of drugs targeting IDPs currently available (see previous section) may look disappointing, considering the physiological relevance of these proteins. One should be careful, though, to not exaggerate the current lack of IDP-specific drugs as being a reflection of disorder as a limitation for drug development.

The rational drug design strategy has been used successfully since the 1980s [317–319]. It depends on knowledge of the three-dimensional structure of the target protein, based on which ligands (usually inhibitors) are planned with aid of computational tools [320]. By definition, IDPs do not have a single, major conformation, occurring in dynamic conformational ensembles [10], and there is difficulty in using such traditional techniques to design IDP ligands [321]. Hence, most cases of IDP drug development were carried out by experimental screening and not by rational design [322]. Still, detection of IDP "hits" (potential initial drug candidates) through high throughput screening of compounds has been challenging [321] and computational methods achieved some success in predicting good candidates [323].

For IDPs with recognizable/determined metastable structures in their conformational ensembles, such structures could be used for rational drug design. However, IDPs are expected to be promiscuous, acting as hubs for multiple cellular processes [324]. This scenario, described as "protein clouds" [325] has its complexity further increased, with IDP ligands being described as "ligand clouds around protein clouds" [326]. Such roles in protein–protein interactions (PPIs) make IDPs especially interesting as drug targets, but the development of molecules targeting PPIs has been in itself, challenging [327–330].

A large difference is expected between the entropic loss and the enthalpic gain upon binding of a small ligand to an IDP, but some of them were shown to be capable of forming adaptable, specific interfaces for small molecule binding [231,235,321]. Intrinsically disordered proteins are difficult targets, since their interactions with small molecules are weaker and more transient, and the entropic loss is greater, in comparison to structured proteins [331]. The fragment-based drug design approach allows fragments to sample large amounts of chemical space, reducing the number of compounds for screening, with different fragments that bind at different regions of IDPs being able to be linked together via an appropriate linker [331]. Such fragments usually require hydrogen bonds to achieve detectable binding, generating an enthalpic gain that compensates for the entropic loss upon binding of the small molecules, lowering the free energy of the protein upon binding [331].

Still, the over-representation of IDPs in disorders, as summarized by the D<sup>2</sup> concept (for "disorder in disorders") [11] and the D3 concept (for "disorder in degenerative disorders") [16], points to these proteins as promising therapeutic targets. Attempts to detect potentially druggable cavities in IDPs have identified at least 14 targets that could be subjected to rational drug design [233]. A more general estimate is that 9% of detected cavities may be druggable in IDPs, in comparison to 5% in ordered proteins [233]. These observations are especially interesting, considering that current drugs target around 500 proteins, less than 10% of the estimated potential target list, with very strict classes (such as enzymes and G-protein coupled receptors) accounting for more than 70% of them [230,331,332].

There have been major advances in the detection and prediction of IDP features in protein sequences [293], something that will surely help in the identification of these special drug targets. Major breakthroughs are also being achieved by the combination of experimental methods (especially

NMR and fluorescence techniques) with computational modeling and molecular dynamics simulation of IDPs [294,322]. The latter is one of the few methods that allow for the description of IDPs in their conformational ensemble, instead of a single (or just a few) conformations [132,322,324,333–337].
