Final Report Summary - STAMYEV (Structural and biochemical basis of protein amyloid evolution)
The publishable summary has to include 5 distinct parts described below:
• An executive summary (not exceeding 1 page).
Amyloid fibrils are protein aggregates formed by certain proteins and they underlie an increasing number of fatal diseases such as Alzheimer´s disease, Familial Amyloidotic Polyneuropathy (FAP) and atherosclerosis that are leading causes of death in the World. Amyloid formation is an intriguing phenomenon poorly understood but found in different species through evolution from bacteria to humans. In the STAMYEV project we investigated how the molecular mechanisms of amyloid formation emerged during the evolution of protein structures as a way to understand such mechanisms and learn strategies to prevent them.
We used two model proteins: apolipoprotein A-I (apoA-I), a cholesterol transporter involved in atherosclerosis and familial amyloidosis, and transthyretin (TTR), a thyroid hormone transporter that forms amyloid in FAP. We studied in detail their amyloid-formation mechanisms using mutagenesis and several biophysical methods. We characterized the amyloid mechanisms of apoA-I variants with single amino-acid mutations and gained insight into the misfolding process at a molecular level. Importantly, we described a new rare mutant apoAIE34K identified in amyloidosis patients and used it to exemplify how a single mutation can alter local conformation dynamics and trigger unfolding and aggregation. We have established, using TTR from fish as an evolutionary model, that amyloid mechanisms are not fully conserved in evolution: specific changes in length and properties of the TTR N-terminus alter the stability of the protein and decrease or abolish amyloid formation. Our findings suggest that during evolution, while TTR’s N-terminus shortened, the protein became more stable but also more prone to form amyloid in vitro highlighting that stability may not correlate inversely with amyloid propensity as previously believed. We could relate the N-terminal of TTR with its amyloid propensity and provide the basis to better understand the amyloid formation process and how to target it. Additionally, we have identified a novel amyloid-forming protein: the extracellular matrix cartilage acid protein (CRTAC) that has an unresolved structure and function but which is associated with a number of pathologies, including multiple sclerosis. We have shown that human and fish CRTAC form amyloid fibrils and this provided the basis for a new ongoing project to produce the crystal structure of human, fish and bacteria CRTAC and clarify CRTAC’s function and role in disease. Overall, this work contributed to advance the general understanding of the amyloid phenomenon and its evolution and revealed important molecular mechanisms of amyloid formation highly relevant and fundamental for the identification of strategies or molecules to block this pathologic process.
• A summary description of project context and objectives (not exceeding 4 pages).
The project STAMYEV addressed the molecular and evolutionary basis for misfolding of functional proteins into aggregates that form amyloid fibrils and are associated with amyloid diseases. Amyloid formation is one of the most intriguing issues in modern biology and over 40 diseases are directly associated with protein amyloid formation, including Alzheimer's disease, atherosclerosis and familial amyloid polyneuropathy (FAP). Amyloid aggregates can derive from proteins sharing no sequence similarity but they all share a common highly organized β-sheet rich structure. Such characteristic structures seem to be evolutionary conserved as they have been identified in bacteria, fungi, insects, fish and humans. In some cases, amyloid aggregates seem to have functional roles and curiously prions (an amyloid structure derived from the prion protein) can propagate the disease just through protein interactions without genetic material. Because not all proteins unfold into amyloid aggregates in vivo, we hypothesized that the mechanisms of misfolding into amyloid can be dictated/influenced by the evolution of protein structures. We believe, that by investigating such mechanisms between evolutionary distant homologues we can learn important information to elucidate and help prevent aggregation and disease.
To address our hypothesis, we proposed to use two model proteins with diverse structural and functional properties: apolipoprotein A-I (apoA-I), an α-helical lipid-binding protein that transports cholesterol and is involved in atherosclerosis and familial amyloidosis, and transthyretin (TTR), a β-sheet-rich tetramer that functions as a thyroid hormone and retinol transporter and forms amyloid in FAP. TTR is particularly well described in different evolutionary species and was chosen as an evolutionary model of amyloid mechanisms using human and fish homologues and mutant variants. Additionally, we introduced Cartilage Acidic Protein as a new evolutionary model as it was recently identified as an amyloid-forming protein and its comparative analysis between human, fish and cyanobacteria was proposed.
The objectives proposed (considered as work packages-WP) and the respective tasks, including deviations from the original Annex I of grant agreement (DOW) (see Table1) were the following:
WP1- Apolipoprotein AI misfolding
Objective 1) Characterize the molecular mechanism of human apoA-I misfolding into amyloid
Within this goal we proposed to analyse the stability, lipid-binding affinity and misfolding of human apoA-I mutants that cause hereditary amyloidosis. We hypothesize that these mutants have reduced stability and are aggregation-prone due to reduced lipid binding affinity
Task 1- Design, express and purify apoA-I wild type (wt) and mutants. We proposed to analyze the amyloid formation mechanisms of a new apoA-I amyloidogenic mutation, hApoA-I E34K, recently identified in 2 patients. For that we planned to produce a recombinant full-length wt and mutant protein, human (h) hApoA-I E34K and also a more stable variant containing the N-terminal section of the apoA-I protein: E34K delta(185-243) and the N-terminal of two other described mutant F71Ydelta(185-243), L159Rdelta(185-243) for comparative controls. This was a deviation from the original plan as a result of new data that became available at the start of the project and a new collaboration with the laboratory that identified the new amyloidogenic mutation mutant E34K in patients.
Task 2- ApoA-I:lipid-complexes. This step included, i) complex formation of apoA-Iwt and mutants with phosphatidyl cholines (model lipid) and ii) comparison of the structure and stability of the proteins as lipid-bound and lipid-free proteins
Task 3- Monitor conformational changes and misfolding of full length and N-terminal ApoA-Iwt and mutants This task involved monitoring the conformational changes by biophysical methods (Circular Dichroism and Fluorescence).
WP2- Transthyretin (TTR) misfolding
Objective 2) Characterize TTR unfolding and aggregation and assess the role of the N-terminus in this process using a comparative approach
We proposed to test the hypothesis that the N-terminal segment of TTR plays an important structural role in its stability and amyloid formation.
Task 4- Production of human and fish TTR variants: involved the design, expression and purification of several TTR protein forms (wild type and mutants) from human (h) and sea bream (sb): hTTRWT, hTTRG6S (naturally occurring novel mutant), sbTTRwt, sbTTRΔ1-6, sbTTRΔ1-12 and lamprey TTR (lampTTR).
Task 5- Stability and amyloid formation of human and fish TTR variants: in this task, we proposed to test fibril formation of hTTRwt, sea bream sbTTRwt and N-terminal mutants by fluorescence, assess fibril morphology by EM, and analyze stability and fibril formation mechanism using biophysical methods. We also included additional studies of amyloid presence/formation inside cells transfected with human, sb and lamp TTR constructs that will continue to after the project.
Task 6- Crystallization of lamprey TTR: This task was not originally planned but was proposed as a contingency plan, in order to assess if the N-terminal section of the protein modified protein and tetramer structure and to understand how this might contribute to evolution of protein stability (sb and h TTR crystal structure is available). Our goal was to provide a new TTR structure not previously described and potentially reveal structural evolutionary features of TTR, namely the position and impact on protein structure and the tetramer of the N-terminus, which is longer in Lamprey than in human (9 amino acids longer) or sea bream (6 amino acids longer).
WP3- Cartilage acidic protein (CRTAC) aggregation and crystallization
Objective 3) Test CRTAC amyloid formation and crystallize CRTAC β-propeller domain
This goal was added as an important asset to this project considering that CRTAC proteins have been conserved during evolution, are strongly aggregation-prone and play a putative role in disease. We proposed to test the propensity of CRTAC to form amyloid and provide novel information on their structure and evolution by comparing human (hCTRAC1) and two homologues of sea bass CRTAC (dlCRTAC1a and dlCRTAC1b). We also started a collaboration with the group of Professor Liebermann at the Georgia Institute of Technology aimed at solving the crystal structure of the beta-propeller region of 3 evolutionary representative CRTAC homologues: human (hCRTAC), sea bream (sparus aurata) (saCRTAC) and cyanobacter (Synechococcus sp) (syCRTAC). The goal was to have the proteins ready to start the trials and proceed with this task through our collaboration, even after completion of the MSCA project.
Task 7- Characterize CRTAC proteins, its structure, stability and aggregates and crystallize hCRTAC, saCRTAC and cyCRTAC. We proposed to test whether fish and human CRTAC homologues form amyloid fibrils. This involved the design of constructs of human, sea bream and cyanobacteria CRTAC1 β-propeller domain, optimize expression of soluble forms of the proteins and obtain large amounts of pure proteins for crystallization.
• A description of the main S&T results/foregrounds (not exceeding 25 pages)
Main results within each WP:
WP1- Apolipoprotein AI misfolding
Molecular mechanism of human apoA-I misfolding into amyloid
We proposed to identify molecular mechanisms of apoAI amyloid formation, in particular the rare mutant E34K that is involved in a severe form of systemic amyloidosis. The apoA-IE34K is a very interesting and novel mutation recently identified in apoA-I amyloidosis patients. This mutation is located at the top of the unstable helix bundle formed by the lipid-free apoA-I monomer, and is postulated to be critical for amyloid formation. We believed this mutation would provide invaluable information on the stability and aggregation mechanism of apoA-I. However, due to the low stability of the full-length protein and the reported high stability of the apoA-Iwt N-terminal region (residues 1-184) that was previous expressed and crystallized in the apoA-Iwt we decided to focus on this region that contains all known sites of amylolidogenic apoAI (AApoAI) mutations, ii) three out of four amyloidogenic segments that likely initiate protein misfolding, and iii) the 9-11 kDa N-terminal fragments found in AApoAI deposits. Initially, we produced full-length apoA-I wt and E34K mutant which was designed and cloned into the vector pMalcx4 (the wt cloned into pDEST-His6-MBP vector) for expression in E. coli BL21 as MBP-fusion proteins and purified by affinity chromatography. Because mutant E34K expresses in low amounts and is very unstable a new construct was designed and produced with codon usage optimization. We generated the N-terminal region (1-184) of ApoA-IE34K and explored its conformation, stability and dynamics in vitro and in silico using molecular dynamics (MD) simulations. Other naturally occurring variants and their C-terminal truncated regions were used as controls including apoA-IWT and two disease-causing mutants: F71Y (the most conservative AApoAI mutation), and L159R, a non-amyloidogenic mutation that causes aberrant lipid metabolism and increases the risk of atherosclerosis. This allowed validation of the C-terminal truncated version of the protein as a model to study protein dynamics in MD simulations.
We designed constructs and expressed all the truncated mutants E34Kdelta(185-243), apoA-IWTdelta(185-243) F71Ydelta(185-243), L159Rdelta(185-243) using the vector pMalcx4 and fusion to MBP to improve solubility. The proteins were successfully expressed in E. coli BL21 and purified by affinity and size-exclusion chromatography. Mutational effects on the protein structure, stability, and dynamics in specific regions were determined using circular dichroism (CD) spectroscopy (Fig.1) and molecular dynamics (MD) simulations and are shown in Fig. 2.
All C-terminally truncated proteins were monodisperse in solution and formed dimers observed by size-exclusion chromatography (Fig. 2B), similar to what was described for the crystallized construct. Far-UV CD spectra indicated a high α-helical conformation for all proteins at 25 °C (Fig. 2D), with the helix content ranging from 56 ± 5% for delta(185-243)E34K to 62 ± 5% for delta(185-243)WT, similar to 60% alpha-helix (or ~110 residues) reported previously for delta(185-243)WT 1.
Comparison of apoA mutants with full-length WT, which showed 50 ± 5% alpha-helix, including ~110 residues in the globular domain2, suggested that this domain had very similar secondary structure in the full-length and in the truncated WT. The helical content in the globular domain did not significantly change upon C-terminal truncation and was perturbed slightly by point mutations including E34K.
Near-UV CD spectra were used to probe the aromatic packing in the helix-bundle domain. This domain contains all four tryptophans and five out of seven tyrosines in apoA-I, which dominate its near-UV CD. Importantly, each mutant showed a distinct spectrum indicating altered aromatic packing (Fig. 1E). WT, F71Y, and L159R showed very similar near-UV CD spectra in their C-terminally truncated and full-length forms. These results show that delta(185-243) truncation did not significantly change the aromatic packing in the globular domain of all proteins explored. However, this packing was significantly altered by point mutations in this domain, such as Glu35Lys, F71Y, and L159R.
Protein stability was probed by thermal denaturation wherein changes in the alpha-helical content were monitored by CD at 222 nm. All proteins showed cooperative alpha-helical unfolding (Fig. 1F) that was independent of the heating rate, indicating thermodynamic reversibility. The transition midpoint, which was measured with an STD of 1.5 °C, was Tm=57 °C in truncated WT and F71Y. Truncated E34K showed Tm=52 °C (Fig. 1F). Truncated L159R showed the lowest Tm=37 °C, 20 °C
The rank order of stability emerging from the current study of globular domains was WT=F71Y>E34K>L159R (Fig. 2F), which is in agreement with that observed for the full-length proteins, WT≥F71Y>L159R 3. This agreement suggests that, similar to their globular domains, full-length E34K is less stable than WT and F71Y, but more stable than L159R apoA-I. This result is consistent with the clinical finding that E34K mutation carriers have normal plasma levels of apoA-I and HDL, unlike L159R carriers whose plasma levels are reduced probably due to reduced protein stability.
In summary, our results revealed that the E34K mutation has little effect on the overall secondary structure in the globular domain but decreased its stability, as evidenced by a ~5 C reduction in Tm. Secondly, this and other point mutations in the helix-bundle domain have distinct near-UV CD signatures and hence, distinct aromatic side chain packing. Thirdly, despite its destabilizing effects, the C-terminal truncation has no significant effects on the secondary structure and aromatic packing in the helix bundle of WT, F71Y, L159R and, probably, E34K. These results validate the use of C-terminally truncated variants as structural models for understanding their full-length counterparts. As such, the C-terminally truncated variants were employed in all MD simulations, with initial coordinates drawn from the crystal structure of apoA-Idelta(185-243)WT 1.
The MD simulations were performed by our collaborators John Straub and Afra Panahi and helped unravel the early steps in the misfolding of variant proteins. The monomeric protein instead of the crystallographic dimer was chosen for the simulations, as the free apoA-I monomer is thought to form the precursor of amyloid and facilitates more extensive simulations.
The helix bundle in all proteins remained stable after high-temperature simulations but showed partial loss of secondary structure, mainly in residues 34 - 81 (Fig. 2A) decreasing to 60% in the final model. The latter agreed with the 61% alpha-helix content determined by CD spectroscopy in delta(185-243) WT in solution (Fig. 2). Similarly, mutant proteins showed a decrease in their helical content by ~20% due to partial unfolding (Fig. 2), in agreement with 56 - 62% alpha-helix observed in solution by far-UV CD (Fig. 2). All proteins showed full or partial unfolding in the region 34 - 81 (Fig. 2) denoting a flexible secondary structure consistent with large-scale motions proposed to mediate the helix-bundle opening during apoA-I binding to HDL 1,4. Mutation-specific differences were observed in the local secondary structure. This included the helical conformation at Tyr18 and downstream. Tyr18, which is located near a helical kink in the helix bundle, is in the middle of the major amyloidogenic segment, Leu14-Leu22 (Fig. 2B). In WT, this hydrophobic segment is fully helical and thus protected from misfolding, yet in all mutants this region shows an apparent helical loss (Fig. 2A, B). We propose that such a mutation-induced loss of structural protection in this adhesive segment helps initiate protein misfolding in amyloid. Partial loss of helical structure in the major amyloidogenic segment of mutant proteins, suggested by our MD simulations (Fig. 2) suggests a possible molecular basis for amyloid formation by these mutants.
MD simulation studies further showed that the mutations influence global molecular motions (not shown). The mutational effects on the molecular motion were not localized to the mutation site but distributed across the protein, progressively increased from F71Y to E34K to L159R. A similar rank order was observed for the experimentally determined thermal stability, WT>F71Y>E34K>L159R. Our observations consistently show that E34K, which is much less conservative than F71Y substitution, is also more disruptive. Notably, the most disruptive mutation, L159R, is non-amyloidogenic. These results support the idea that structural destabilization alone is not sufficient to make the protein amyloidogenic, and suggest that local protein structure and dynamics in sensitive regions are important.
MD simulations also revealed significant mutational effects on the conformation of Trp72 and suggested that alternative conformations of Trp72 and perhaps Trp8 at the bottom of the helix bundle were mainly responsible for the differences in the Trp packing among the variant proteins observed by near-UV CD (Fig. 1E). We propose that “open” conformations of Trp72, which were particularly highly populated in the two amyloidogenic mutants, F71Y and E34K, help initiating protein misfolding by perturbing the structure in the bottom of the helix bundle and allowing transient water entry into its hydrophobic core. Such perturbations are expected to increase solvent exposure of the nearby Tyr18 and other residues from the major amyloidogenic segment Leu14-Leu22, thus favoring protein misfolding.
WP2- Transthyretin misfolding
TTR unfolding and aggregation and the role of the N-terminus in this process using a comparative approach
All TTR variants were successfully produced and purified as showed by SDS-PAGE in Fig. 2A. In all cases high yields of protein were obtained. Establishing the expression and purification methods was essential to perform the proposed studies and also set the basis for new relevant projects that will continue after the reporting period.
We then performed a thorough comparative analysis of the stability and amyloid formation of three TTRs evolutionarily distant and with significant changes in the N-terminal length: human (h) TTR, sea bream (sb) TTR and lamprey (lamp) TTR. Gel filtration reported on the apparent stability of the tetramers showing that, in common with hTTR, sbTTR and lampTTRs form homogenous tetramers and retain a tetrameric conformation in solution (Fig. 4A). Because the TTR tetramer is generally very stable and does not loose structure with heating up to 100C, we used Urea 3M to destabilize the tetramer before performing a melting scan. Circular dichroism melting data at an ellipticity minimum 220 nm of the TTRs in 3M Urea showed (Fig. 4B) a strong scan rate dependence, indicating kinetic effects. Heating from 50 to 95oC at 30 C/h showed that the apparent melting temperatures Tm,app (which report on stability) increased from the TTR from the more ancient (lamprey) to the more evolved human protein. These results suggest that during evolution from agnathan to humans, TTR not only gained affinity for THs but also apparently became more amyloid-prone and, paradoxically, more thermostable. Increased stability may have evolved as a protective mechanism against amyloidosis. To test fibril formation, we measured Thioflavin-T fluorescence after incubation with TTS at 37oC for 48h (Fig. 4B) and results revealed that lampTTR is less amyloidogenic than sb or hTTR. These results were confirmed by electron microscopy (Fig. 4D) that revealed dense fibrillar agglomerates for hTTR, earlier-stage aggregates (short fibrils/protofibrils) for sbTTR and non-fibrillar aggregates for lampTTR. These results suggest that, surprisingly, fibrillation propensity of TTR correlates directly with protein thermostability.
The structure of lampTTR has not been previously described and its markedly longer N-terminus may suggest an alternative or more stable configuration. To obtain information on lampTTR structure and possibly on the localization of the N-terminal within the structure we tried to crystalize the protein. Crystal screens were based on previously published conditions for h and sb TTR. Crystals shaped as needles were obtained in 2.4 m ammonium sulfate, 7% glycerol, and 0.2 m sodium acetate, pH 5-6.5. Crystals were analyzed by SDS-PAGE and the presence of a single band suggests protein crystals and not salt. The obtained crystals will be tested by x-ray and data will be collected to produce a lampTTR 3D structure.
WP3- Cartilage acidic protein (CRTAC) aggregation and crystallization
CRTAC amyloid formation and crystallization of CRTAC β-propeller domain
We proposed to test the stability and propensity of CRTACs to aggregate and form amyloid structures. Additionally, we aimed to provide novel information on their structure and evolution by solving the crystal structure of the beta-propeller region of 3 evolutionary representative CRTAC homologues: human (hCRTAC), sea bream (sparus aurata) (saCRTAC) and cyanobacter (Synechococcus sp) (cyCRTAC).
The studies to compare TTR stability and amyloid formation were performed as a comparative analysis between the human CRTAC (hCRTAC) and two sea bass CRTAC1 variants (dlCRTAC1a and dlCRTAC1b). The corresponding sequences were cloned into a pET11a vector, expressed in E. coli and purified by continuous elution electrophoresis (Model 491 Prep Cell).
The secondary structure of human and teleost CRTAC1’s was evaluated using far-UV CD spectra measurements. All CRTACs had high β-sheet content (Fig. 5A) and hCRTAC1 and dlCrtac1a with an ellipticity minimum at 211nm shared the greatest structural similarity.
CRTAC1s stability was assessed by measuring CD ellipticity minimum (210 nm) changes during heating from 20 ºC to 95 ºC, which denotes thermal unfolding. Complete loss of secondary structure, was not achieved even at 95 ºC, suggesting high thermostability of the protein (Fig. 5B). The loss of secondary structure started around 50 ºC and was much more pronounced for hCRTAC1 compared to dlCrtac1a and dlCrtac1b. The high thermostability of CRTAC1’s was further confirmed using intrinsic fluorescence spectroscopy (Fig. 5C) when no appreciable emission maximum shift, suggestive of tertiary structure unfolding, was detected after heating, but only in the presence of Guanidine Hydrochloride (6M) (Fig. 5D) indicating the proteins are susceptible to chemical but not thermal unfolding. Since CRTAC1 forms high molecular weight aggregates spontaneously in solution, the observed thermostability may also be a consequence of a high thermoresistance of the formed aggregates.
The presence of high molecular aggregates was confirmed by size-exclusion chromatography (SEC) (Fig. 6A) and PAGE (Fig. 6B) but the monomeric form of the proteins was also found is solution as suggested by the smaller peaks with higher retention volumes of approximately 14-15 ml (molecular weight approximately 40-70kDa). However, for dlCrtac1b, SDS-PAGE revealed that the protein in the higher retention volume fraction contained a mixture of both aggregates and monomers suggesting that monomeric dlCrtac1b was highly unstable and aggregated immediately after isolation.
In order to evaluate whether CRTAC1s form amyloid structures in vitro we incubated samples of CRTAC1s for 2 weeks at 37ºC and analysed them over time to follow the progression of aggregation and the morphology of the aggregates by TEM (Fig. 7). After 1 week at 37ºC both human and piscine CRTAC1 solutions contained elongated oligomeric structures that formed dense clusters of amyloid aggregates. After 2 weeks at 37ºC dlCrtac1a and hCRTAC1 samples had fewer aggregation clusters and more disperse smaller aggregation units while dlCrtac1b contained very well defined curvilinear protofibrils. In general aggregates of hCRTAC1 and dlCrtac1a formed faster and had less defined morphology than those of dlCrtac1b. No significant change in the aggregation patterns of the CRTAC1 proteins analysed was identified during 3 weeks incubation (not shown). In general aggregates, hCRTAC1 and dlCrtac1a formed faster and had less defined morphology than those of dlCrtac1b.
In order to clarify the structure of CRTACs and its evolution we planned to solve the crystal structure of three evolutionary distant CRATC homologues: human (hCRTAC1), sea bream (sparus aurata) (saCRTAC1b) and cyanobacter (Synechococcus sp) (cyCRTAC1). Because CRTACs are aggregation prone and hard to crystallize, we focused specifically on the most conserved and potentially more soluble beta-propeller region (this region has been crystallized in other beta-propeller proteins). Knowing the expression and purification of the proposed proteins might require thorough optimization we set the goal to produce soluble and sufficient amounts of proteins within this project and continue with the crystallography trials after the project. For this task we established a new collaboration that will continue after this project with the group of Prof. Liebermann at the Georgia Institute of Technology. Prof. Liebermann has extensive experience on the crystallization of difficult beta-propeller proteins.
The coding sequence of the β-propeller domain of the hCRTAC1, saCRTAC1b and cyCRTAC1 was each inserted in the vector pMalc4X for expression in E. coli BL21 as MBP-fusion proteins. After extensive optimization, we could obtain a significant amount of protein in the soluble phase (Fig. 8) that is now being purified and will be further used for crystallization trials.
Summary
Compliance with foreseen goals/milestones
Year 1
• Scientific aims achieved (within modified plan) and results analysed
• 1 manuscript published; 1 manuscript in preparation; 1 oral presentation
• 1 conference attended; 1 workshop attended; > 10 seminars attended
• Training aims achieved
Year 2
• Scientific aims achieved (within modified plan) and results analysed
• 1 manuscript published; 1 manuscript submitted; 1 manuscript in preparation; 3 oral presentation
• 1 conference attended; 5 workshops attended; > 15 seminars attended
• Training aims achieved
Year 3
• Scientific aims achieved (within modified plan) and results analysed
• 2 manuscripts published; 1 manuscript submitted; 2 manuscripts in preparation; 4 oral presentation
• 3 conferences attended; 9 workshops attended; > 20 seminars attended
• Training aims achieved
• New research position offered at Harvard Medical School
• Funding application for 2 new projects submitted shortly after reporting period 2
Training received:
Biophysical methods: Circular dichroism (CD) spectroscopy, Differential scanning calorimetry, pressure perturbation calorimetry (DSC and PPC), Transmission electron microscopy (TEM), Protein Crystallography, Fluorescence methods. Cell culture, fluorescence microscopy, immunohistochemistry (returning host)
Science communication: oral presentations, workshops, publication of a layman summary, review/book chapter writing
Management and interpersonal skills: management of the MSCA project (budget and science) independently, managed 7 collaborations, including MSCA project partners
Mentorship and leadership: helped training graduate students in the laboratory: teaching assistant at Boston University graduate program, courses and seminars.
• The potential impact (including the socio-economic impact and the wider societal implications of the project so far) and the main dissemination activities and exploitation of results (not exceeding 10 pages).
WP1- Apolipoprotein AI misfolding
Amyloid diseases, including neurodegenerative and systemic amyloidosis, affect millions of patients worldwide and are a major public health challenge. Apolipoproteins, a family of lipid surface-binding proteins are overrepresented in amyloidosis. Hereditary apolipoprotein A-I (apoA-I) amyloidosis is a life-threatening type of systemic amyloidosis due to specific amino-acid mutations in the apoA-I protein. Mutant apoA-I proteins aggregate and deposit, often as N-terminal fragments in various organs, causing organ damage. So far, the molecular events triggering protein aggregation in this and other amyloid diseases are not completely understood.
In these studies, we collaborated with Sophie Valleix from Université Paris-Descartes and John E. Straub at Boston University to develop in-depth investigation of a new apoA-I mutation identified in patients with severe amyloidosis: apoA-I-E34K. We contributed to elucidate the molecular basis for the amyloidogenic properties of the apoA-IE34K variant and compared its conformation and dynamics with those of other disease-causing variants, Phe71Tyr (amyloidogenic) and Leu159Arg (non-amyloidogenic). We found that the mutations reduced protein stability and altered aromatic side chain packing and caused local helical unfolding. Our findings exemplify how diverse mutations alter local conformational dynamics in similar sensitive regions of a protein molecule. This phenomenon is probably not limited to apoA-I, but may extend to other globular proteins and their disease-causing variants
This contribution to elucidation of early steps on protein unfolding, along with improved diagnostics, can advance molecular understanding and therapeutic targeting of amyloid disorders. The study of amyloidosis as a basis to understand the process so effective treatments can be identified is of considerable socio-economic interest. This is particularly so in Europe where the average age of the population is shifting upwards and makes diseases of the aging population relevant from a series of socio-economic perspectives. The costs of amyloidogenic diseases for society are considerable in terms of health care needs, loss of work, injury but also because of their impact on the structure and functioning of societies.
WP2- Transthyretin misfolding
Transthyretin amyloidosis is another type of amyloid disease with high prevalence in countries like Portugal and Sweden. Human Transthyretin transports thyroid hormones in the blood in a tetrameric conformation. When the tetramer is destabilized the protein unfolds and aggregates into amyloid fibrils that deposit in organs or the nervous system causing severe and lethal polyneuropathies and cardiomyopathies. We had previously characterized TTR evolution using fish TTR models and hypothesized that N-terminal region, which was remodeled during evolution, could be involved in amyloid formation. In the STAMYEV project we provided evidences that the N-terminus is involved in the protein stability and modulates amyloid formation during evolution. We found that TTR from the ancient agnathan, lamprey does not form fibril despite the lower stability of the protein. Lamprey TTR has a longer N-terminus that could be responsible for blocking TTR aggregation. With this work we set up the grounds to further investigate in detail the molecular interactions in lamprey TTR that prevent the protein to enter the amyloid-formation pathway. We hope in the future to translate this information into targeting human TTR amyloid formation in disease and provide insights into development of new therapies.
The output of this work can continue to be shared between the outgoing host, Boston University, USA and the returning host, Canter of Marine Sciences, Portugal. The host in Boston works in direct collaboration with the Boston University Amyloidosis centre where basic science is combined with clinical interventions. Boston is a large hub of pharmaceutical companies, several of them directly involved in finding therapies for TTR amyloidosis, and working directly with the Amyloidosis centre in clinical trials. Any future outcomes from this project have the potential to impact the patients.
I will now transfer the skills, knowledge and scientific output of the STAMYEV project into my new position at Harvard Medical School (the laboratory will be transferred in the near future to the Cardiovascular Institute at Stanford School of Medicine). I will be working at the laboratory of Professor Liao, to help develop a project focused on Transthyretin amyloid cardiomyopathies. In this new project we will develop animal models of transthyretin amyloidosis and we will further test the N-terminal hypothesis in vivo, including different mutants. We will work in close collaboration with medical/clinical researches and pharmaceutical industry increasing to potential for the development of treatment strategies.
WP3- Cartilage acidic protein (CRTAC) aggregation and crystallization
Cartilage acidic protein1 (CRTAC1) is an extracellular matrix protein of chondrogenic tissue and its function remains unclear. CRTAC is putatively involved with diseases of the human cardiovascular, haematological, neurological, respiratory and urinary systems and potentially in multiple sclerosis. CRTAC also belongs to a large family of beta-propeller proteins that in mammals have been associated with diseases, including amyloid diseases such as Alzheimer’s. With our studies we found that CRTAC1 forms high molecular weight thermo-stable aggregates and revealed for the first time that such aggregates can consist of amyloid-like structures that might be related to its disease-association. We placed CRTAC protein within the landscape of the amylome and further contributed to the general understating of CRTAC1’s and beta-propeller family evolution and function. Overall, we provided new insights into the potential biological function of CRTAC1 in vertebrates and its possible role in disease. We have opened up the pathway for targeted studies in human to establish the link between the aggregating potential of this protein and amyloidogenic pathologies. Within my future projects at Stanford School of Medicine I will aim to further develop such studies. We will have access to TTR amyloidosis patient’s samples where I can try to identify the presence of CRTAC proteins in the aggregated form and investigate its putative interactions with different proteins in their amyloid conformation (e.g. TTR).
Overall, the project has had impact in diverse areas such as fundament biology, medicine, health, society and has carried out the groundwork for new targeted interventions in amyloid diseases. Amyloid related disorders represent a tremendous medical and socio-economic problem, specifically in ageing societies, such as in Europe or North America which are also two major hubs of amyloid research. Alzheimer’s disease, for example, is estimated to affect more than and 5 million people in Europe alone and where annual societal cost are estimated to exceed 55 billion €. The understanding of amyloid formation is a clear priority in the context of degenerative diseases and our studies that are aimed towards understanding the evolution of the protein structures and dynamics of amyloidogenesis directly contribute to that goal. Specifically, we helped to: 1) establish a novel molecular mechanism of apoA-I misfolding in Apolipoprotein amyloidosis and atherosclerosis, 2) support the novel hypothesis that TTR N-terminus plays a key role in TTR misfolding in amyloid, and 3) establish a link between evolutionary adaptation and amyloid formation by fish and human TTR.
Impact on scientific and personal development
The project had also a tremendous impact on my personal scientific development. I had the opportunity to deepen my knowledge and broaden my skills in structural biology, biophysics of protein folding and amyloid through expert training from Dr. Gursky, her laboratory team and collaborators. In addition to specific scientific training, the project challenged me to further develop/acquire complementary transferable competencies specially directed towards scientific maturity and independency (detailed below). I attended several training courses and seminars toward development of skills in teaching, mentorship, science communication, time management, leadership, scientific writing, job application and interviewing. I participated and presented my work in conferences and in seminars both nationally and internationally and I joined science outreached activities. I was also able to manage the project independently, both scientifically and administratively. I established and managed new collaborations. At the end of the project I was offered a position of Associate Scientist at Harvard Medical School and shortly after I will transition to Stanford University School of Medicine as faculty Instructor towards establishing a future independent position. I will continue to foster collaborations stemming from the MSCA project and disseminate its outcomes.
• The address of the project public website, if applicable as well as relevant contact details.
Not applicable
• An executive summary (not exceeding 1 page).
Amyloid fibrils are protein aggregates formed by certain proteins and they underlie an increasing number of fatal diseases such as Alzheimer´s disease, Familial Amyloidotic Polyneuropathy (FAP) and atherosclerosis that are leading causes of death in the World. Amyloid formation is an intriguing phenomenon poorly understood but found in different species through evolution from bacteria to humans. In the STAMYEV project we investigated how the molecular mechanisms of amyloid formation emerged during the evolution of protein structures as a way to understand such mechanisms and learn strategies to prevent them.
We used two model proteins: apolipoprotein A-I (apoA-I), a cholesterol transporter involved in atherosclerosis and familial amyloidosis, and transthyretin (TTR), a thyroid hormone transporter that forms amyloid in FAP. We studied in detail their amyloid-formation mechanisms using mutagenesis and several biophysical methods. We characterized the amyloid mechanisms of apoA-I variants with single amino-acid mutations and gained insight into the misfolding process at a molecular level. Importantly, we described a new rare mutant apoAIE34K identified in amyloidosis patients and used it to exemplify how a single mutation can alter local conformation dynamics and trigger unfolding and aggregation. We have established, using TTR from fish as an evolutionary model, that amyloid mechanisms are not fully conserved in evolution: specific changes in length and properties of the TTR N-terminus alter the stability of the protein and decrease or abolish amyloid formation. Our findings suggest that during evolution, while TTR’s N-terminus shortened, the protein became more stable but also more prone to form amyloid in vitro highlighting that stability may not correlate inversely with amyloid propensity as previously believed. We could relate the N-terminal of TTR with its amyloid propensity and provide the basis to better understand the amyloid formation process and how to target it. Additionally, we have identified a novel amyloid-forming protein: the extracellular matrix cartilage acid protein (CRTAC) that has an unresolved structure and function but which is associated with a number of pathologies, including multiple sclerosis. We have shown that human and fish CRTAC form amyloid fibrils and this provided the basis for a new ongoing project to produce the crystal structure of human, fish and bacteria CRTAC and clarify CRTAC’s function and role in disease. Overall, this work contributed to advance the general understanding of the amyloid phenomenon and its evolution and revealed important molecular mechanisms of amyloid formation highly relevant and fundamental for the identification of strategies or molecules to block this pathologic process.
• A summary description of project context and objectives (not exceeding 4 pages).
The project STAMYEV addressed the molecular and evolutionary basis for misfolding of functional proteins into aggregates that form amyloid fibrils and are associated with amyloid diseases. Amyloid formation is one of the most intriguing issues in modern biology and over 40 diseases are directly associated with protein amyloid formation, including Alzheimer's disease, atherosclerosis and familial amyloid polyneuropathy (FAP). Amyloid aggregates can derive from proteins sharing no sequence similarity but they all share a common highly organized β-sheet rich structure. Such characteristic structures seem to be evolutionary conserved as they have been identified in bacteria, fungi, insects, fish and humans. In some cases, amyloid aggregates seem to have functional roles and curiously prions (an amyloid structure derived from the prion protein) can propagate the disease just through protein interactions without genetic material. Because not all proteins unfold into amyloid aggregates in vivo, we hypothesized that the mechanisms of misfolding into amyloid can be dictated/influenced by the evolution of protein structures. We believe, that by investigating such mechanisms between evolutionary distant homologues we can learn important information to elucidate and help prevent aggregation and disease.
To address our hypothesis, we proposed to use two model proteins with diverse structural and functional properties: apolipoprotein A-I (apoA-I), an α-helical lipid-binding protein that transports cholesterol and is involved in atherosclerosis and familial amyloidosis, and transthyretin (TTR), a β-sheet-rich tetramer that functions as a thyroid hormone and retinol transporter and forms amyloid in FAP. TTR is particularly well described in different evolutionary species and was chosen as an evolutionary model of amyloid mechanisms using human and fish homologues and mutant variants. Additionally, we introduced Cartilage Acidic Protein as a new evolutionary model as it was recently identified as an amyloid-forming protein and its comparative analysis between human, fish and cyanobacteria was proposed.
The objectives proposed (considered as work packages-WP) and the respective tasks, including deviations from the original Annex I of grant agreement (DOW) (see Table1) were the following:
WP1- Apolipoprotein AI misfolding
Objective 1) Characterize the molecular mechanism of human apoA-I misfolding into amyloid
Within this goal we proposed to analyse the stability, lipid-binding affinity and misfolding of human apoA-I mutants that cause hereditary amyloidosis. We hypothesize that these mutants have reduced stability and are aggregation-prone due to reduced lipid binding affinity
Task 1- Design, express and purify apoA-I wild type (wt) and mutants. We proposed to analyze the amyloid formation mechanisms of a new apoA-I amyloidogenic mutation, hApoA-I E34K, recently identified in 2 patients. For that we planned to produce a recombinant full-length wt and mutant protein, human (h) hApoA-I E34K and also a more stable variant containing the N-terminal section of the apoA-I protein: E34K delta(185-243) and the N-terminal of two other described mutant F71Ydelta(185-243), L159Rdelta(185-243) for comparative controls. This was a deviation from the original plan as a result of new data that became available at the start of the project and a new collaboration with the laboratory that identified the new amyloidogenic mutation mutant E34K in patients.
Task 2- ApoA-I:lipid-complexes. This step included, i) complex formation of apoA-Iwt and mutants with phosphatidyl cholines (model lipid) and ii) comparison of the structure and stability of the proteins as lipid-bound and lipid-free proteins
Task 3- Monitor conformational changes and misfolding of full length and N-terminal ApoA-Iwt and mutants This task involved monitoring the conformational changes by biophysical methods (Circular Dichroism and Fluorescence).
WP2- Transthyretin (TTR) misfolding
Objective 2) Characterize TTR unfolding and aggregation and assess the role of the N-terminus in this process using a comparative approach
We proposed to test the hypothesis that the N-terminal segment of TTR plays an important structural role in its stability and amyloid formation.
Task 4- Production of human and fish TTR variants: involved the design, expression and purification of several TTR protein forms (wild type and mutants) from human (h) and sea bream (sb): hTTRWT, hTTRG6S (naturally occurring novel mutant), sbTTRwt, sbTTRΔ1-6, sbTTRΔ1-12 and lamprey TTR (lampTTR).
Task 5- Stability and amyloid formation of human and fish TTR variants: in this task, we proposed to test fibril formation of hTTRwt, sea bream sbTTRwt and N-terminal mutants by fluorescence, assess fibril morphology by EM, and analyze stability and fibril formation mechanism using biophysical methods. We also included additional studies of amyloid presence/formation inside cells transfected with human, sb and lamp TTR constructs that will continue to after the project.
Task 6- Crystallization of lamprey TTR: This task was not originally planned but was proposed as a contingency plan, in order to assess if the N-terminal section of the protein modified protein and tetramer structure and to understand how this might contribute to evolution of protein stability (sb and h TTR crystal structure is available). Our goal was to provide a new TTR structure not previously described and potentially reveal structural evolutionary features of TTR, namely the position and impact on protein structure and the tetramer of the N-terminus, which is longer in Lamprey than in human (9 amino acids longer) or sea bream (6 amino acids longer).
WP3- Cartilage acidic protein (CRTAC) aggregation and crystallization
Objective 3) Test CRTAC amyloid formation and crystallize CRTAC β-propeller domain
This goal was added as an important asset to this project considering that CRTAC proteins have been conserved during evolution, are strongly aggregation-prone and play a putative role in disease. We proposed to test the propensity of CRTAC to form amyloid and provide novel information on their structure and evolution by comparing human (hCTRAC1) and two homologues of sea bass CRTAC (dlCRTAC1a and dlCRTAC1b). We also started a collaboration with the group of Professor Liebermann at the Georgia Institute of Technology aimed at solving the crystal structure of the beta-propeller region of 3 evolutionary representative CRTAC homologues: human (hCRTAC), sea bream (sparus aurata) (saCRTAC) and cyanobacter (Synechococcus sp) (syCRTAC). The goal was to have the proteins ready to start the trials and proceed with this task through our collaboration, even after completion of the MSCA project.
Task 7- Characterize CRTAC proteins, its structure, stability and aggregates and crystallize hCRTAC, saCRTAC and cyCRTAC. We proposed to test whether fish and human CRTAC homologues form amyloid fibrils. This involved the design of constructs of human, sea bream and cyanobacteria CRTAC1 β-propeller domain, optimize expression of soluble forms of the proteins and obtain large amounts of pure proteins for crystallization.
• A description of the main S&T results/foregrounds (not exceeding 25 pages)
Main results within each WP:
WP1- Apolipoprotein AI misfolding
Molecular mechanism of human apoA-I misfolding into amyloid
We proposed to identify molecular mechanisms of apoAI amyloid formation, in particular the rare mutant E34K that is involved in a severe form of systemic amyloidosis. The apoA-IE34K is a very interesting and novel mutation recently identified in apoA-I amyloidosis patients. This mutation is located at the top of the unstable helix bundle formed by the lipid-free apoA-I monomer, and is postulated to be critical for amyloid formation. We believed this mutation would provide invaluable information on the stability and aggregation mechanism of apoA-I. However, due to the low stability of the full-length protein and the reported high stability of the apoA-Iwt N-terminal region (residues 1-184) that was previous expressed and crystallized in the apoA-Iwt we decided to focus on this region that contains all known sites of amylolidogenic apoAI (AApoAI) mutations, ii) three out of four amyloidogenic segments that likely initiate protein misfolding, and iii) the 9-11 kDa N-terminal fragments found in AApoAI deposits. Initially, we produced full-length apoA-I wt and E34K mutant which was designed and cloned into the vector pMalcx4 (the wt cloned into pDEST-His6-MBP vector) for expression in E. coli BL21 as MBP-fusion proteins and purified by affinity chromatography. Because mutant E34K expresses in low amounts and is very unstable a new construct was designed and produced with codon usage optimization. We generated the N-terminal region (1-184) of ApoA-IE34K and explored its conformation, stability and dynamics in vitro and in silico using molecular dynamics (MD) simulations. Other naturally occurring variants and their C-terminal truncated regions were used as controls including apoA-IWT and two disease-causing mutants: F71Y (the most conservative AApoAI mutation), and L159R, a non-amyloidogenic mutation that causes aberrant lipid metabolism and increases the risk of atherosclerosis. This allowed validation of the C-terminal truncated version of the protein as a model to study protein dynamics in MD simulations.
We designed constructs and expressed all the truncated mutants E34Kdelta(185-243), apoA-IWTdelta(185-243) F71Ydelta(185-243), L159Rdelta(185-243) using the vector pMalcx4 and fusion to MBP to improve solubility. The proteins were successfully expressed in E. coli BL21 and purified by affinity and size-exclusion chromatography. Mutational effects on the protein structure, stability, and dynamics in specific regions were determined using circular dichroism (CD) spectroscopy (Fig.1) and molecular dynamics (MD) simulations and are shown in Fig. 2.
All C-terminally truncated proteins were monodisperse in solution and formed dimers observed by size-exclusion chromatography (Fig. 2B), similar to what was described for the crystallized construct. Far-UV CD spectra indicated a high α-helical conformation for all proteins at 25 °C (Fig. 2D), with the helix content ranging from 56 ± 5% for delta(185-243)E34K to 62 ± 5% for delta(185-243)WT, similar to 60% alpha-helix (or ~110 residues) reported previously for delta(185-243)WT 1.
Comparison of apoA mutants with full-length WT, which showed 50 ± 5% alpha-helix, including ~110 residues in the globular domain2, suggested that this domain had very similar secondary structure in the full-length and in the truncated WT. The helical content in the globular domain did not significantly change upon C-terminal truncation and was perturbed slightly by point mutations including E34K.
Near-UV CD spectra were used to probe the aromatic packing in the helix-bundle domain. This domain contains all four tryptophans and five out of seven tyrosines in apoA-I, which dominate its near-UV CD. Importantly, each mutant showed a distinct spectrum indicating altered aromatic packing (Fig. 1E). WT, F71Y, and L159R showed very similar near-UV CD spectra in their C-terminally truncated and full-length forms. These results show that delta(185-243) truncation did not significantly change the aromatic packing in the globular domain of all proteins explored. However, this packing was significantly altered by point mutations in this domain, such as Glu35Lys, F71Y, and L159R.
Protein stability was probed by thermal denaturation wherein changes in the alpha-helical content were monitored by CD at 222 nm. All proteins showed cooperative alpha-helical unfolding (Fig. 1F) that was independent of the heating rate, indicating thermodynamic reversibility. The transition midpoint, which was measured with an STD of 1.5 °C, was Tm=57 °C in truncated WT and F71Y. Truncated E34K showed Tm=52 °C (Fig. 1F). Truncated L159R showed the lowest Tm=37 °C, 20 °C
The rank order of stability emerging from the current study of globular domains was WT=F71Y>E34K>L159R (Fig. 2F), which is in agreement with that observed for the full-length proteins, WT≥F71Y>L159R 3. This agreement suggests that, similar to their globular domains, full-length E34K is less stable than WT and F71Y, but more stable than L159R apoA-I. This result is consistent with the clinical finding that E34K mutation carriers have normal plasma levels of apoA-I and HDL, unlike L159R carriers whose plasma levels are reduced probably due to reduced protein stability.
In summary, our results revealed that the E34K mutation has little effect on the overall secondary structure in the globular domain but decreased its stability, as evidenced by a ~5 C reduction in Tm. Secondly, this and other point mutations in the helix-bundle domain have distinct near-UV CD signatures and hence, distinct aromatic side chain packing. Thirdly, despite its destabilizing effects, the C-terminal truncation has no significant effects on the secondary structure and aromatic packing in the helix bundle of WT, F71Y, L159R and, probably, E34K. These results validate the use of C-terminally truncated variants as structural models for understanding their full-length counterparts. As such, the C-terminally truncated variants were employed in all MD simulations, with initial coordinates drawn from the crystal structure of apoA-Idelta(185-243)WT 1.
The MD simulations were performed by our collaborators John Straub and Afra Panahi and helped unravel the early steps in the misfolding of variant proteins. The monomeric protein instead of the crystallographic dimer was chosen for the simulations, as the free apoA-I monomer is thought to form the precursor of amyloid and facilitates more extensive simulations.
The helix bundle in all proteins remained stable after high-temperature simulations but showed partial loss of secondary structure, mainly in residues 34 - 81 (Fig. 2A) decreasing to 60% in the final model. The latter agreed with the 61% alpha-helix content determined by CD spectroscopy in delta(185-243) WT in solution (Fig. 2). Similarly, mutant proteins showed a decrease in their helical content by ~20% due to partial unfolding (Fig. 2), in agreement with 56 - 62% alpha-helix observed in solution by far-UV CD (Fig. 2). All proteins showed full or partial unfolding in the region 34 - 81 (Fig. 2) denoting a flexible secondary structure consistent with large-scale motions proposed to mediate the helix-bundle opening during apoA-I binding to HDL 1,4. Mutation-specific differences were observed in the local secondary structure. This included the helical conformation at Tyr18 and downstream. Tyr18, which is located near a helical kink in the helix bundle, is in the middle of the major amyloidogenic segment, Leu14-Leu22 (Fig. 2B). In WT, this hydrophobic segment is fully helical and thus protected from misfolding, yet in all mutants this region shows an apparent helical loss (Fig. 2A, B). We propose that such a mutation-induced loss of structural protection in this adhesive segment helps initiate protein misfolding in amyloid. Partial loss of helical structure in the major amyloidogenic segment of mutant proteins, suggested by our MD simulations (Fig. 2) suggests a possible molecular basis for amyloid formation by these mutants.
MD simulation studies further showed that the mutations influence global molecular motions (not shown). The mutational effects on the molecular motion were not localized to the mutation site but distributed across the protein, progressively increased from F71Y to E34K to L159R. A similar rank order was observed for the experimentally determined thermal stability, WT>F71Y>E34K>L159R. Our observations consistently show that E34K, which is much less conservative than F71Y substitution, is also more disruptive. Notably, the most disruptive mutation, L159R, is non-amyloidogenic. These results support the idea that structural destabilization alone is not sufficient to make the protein amyloidogenic, and suggest that local protein structure and dynamics in sensitive regions are important.
MD simulations also revealed significant mutational effects on the conformation of Trp72 and suggested that alternative conformations of Trp72 and perhaps Trp8 at the bottom of the helix bundle were mainly responsible for the differences in the Trp packing among the variant proteins observed by near-UV CD (Fig. 1E). We propose that “open” conformations of Trp72, which were particularly highly populated in the two amyloidogenic mutants, F71Y and E34K, help initiating protein misfolding by perturbing the structure in the bottom of the helix bundle and allowing transient water entry into its hydrophobic core. Such perturbations are expected to increase solvent exposure of the nearby Tyr18 and other residues from the major amyloidogenic segment Leu14-Leu22, thus favoring protein misfolding.
WP2- Transthyretin misfolding
TTR unfolding and aggregation and the role of the N-terminus in this process using a comparative approach
All TTR variants were successfully produced and purified as showed by SDS-PAGE in Fig. 2A. In all cases high yields of protein were obtained. Establishing the expression and purification methods was essential to perform the proposed studies and also set the basis for new relevant projects that will continue after the reporting period.
We then performed a thorough comparative analysis of the stability and amyloid formation of three TTRs evolutionarily distant and with significant changes in the N-terminal length: human (h) TTR, sea bream (sb) TTR and lamprey (lamp) TTR. Gel filtration reported on the apparent stability of the tetramers showing that, in common with hTTR, sbTTR and lampTTRs form homogenous tetramers and retain a tetrameric conformation in solution (Fig. 4A). Because the TTR tetramer is generally very stable and does not loose structure with heating up to 100C, we used Urea 3M to destabilize the tetramer before performing a melting scan. Circular dichroism melting data at an ellipticity minimum 220 nm of the TTRs in 3M Urea showed (Fig. 4B) a strong scan rate dependence, indicating kinetic effects. Heating from 50 to 95oC at 30 C/h showed that the apparent melting temperatures Tm,app (which report on stability) increased from the TTR from the more ancient (lamprey) to the more evolved human protein. These results suggest that during evolution from agnathan to humans, TTR not only gained affinity for THs but also apparently became more amyloid-prone and, paradoxically, more thermostable. Increased stability may have evolved as a protective mechanism against amyloidosis. To test fibril formation, we measured Thioflavin-T fluorescence after incubation with TTS at 37oC for 48h (Fig. 4B) and results revealed that lampTTR is less amyloidogenic than sb or hTTR. These results were confirmed by electron microscopy (Fig. 4D) that revealed dense fibrillar agglomerates for hTTR, earlier-stage aggregates (short fibrils/protofibrils) for sbTTR and non-fibrillar aggregates for lampTTR. These results suggest that, surprisingly, fibrillation propensity of TTR correlates directly with protein thermostability.
The structure of lampTTR has not been previously described and its markedly longer N-terminus may suggest an alternative or more stable configuration. To obtain information on lampTTR structure and possibly on the localization of the N-terminal within the structure we tried to crystalize the protein. Crystal screens were based on previously published conditions for h and sb TTR. Crystals shaped as needles were obtained in 2.4 m ammonium sulfate, 7% glycerol, and 0.2 m sodium acetate, pH 5-6.5. Crystals were analyzed by SDS-PAGE and the presence of a single band suggests protein crystals and not salt. The obtained crystals will be tested by x-ray and data will be collected to produce a lampTTR 3D structure.
WP3- Cartilage acidic protein (CRTAC) aggregation and crystallization
CRTAC amyloid formation and crystallization of CRTAC β-propeller domain
We proposed to test the stability and propensity of CRTACs to aggregate and form amyloid structures. Additionally, we aimed to provide novel information on their structure and evolution by solving the crystal structure of the beta-propeller region of 3 evolutionary representative CRTAC homologues: human (hCRTAC), sea bream (sparus aurata) (saCRTAC) and cyanobacter (Synechococcus sp) (cyCRTAC).
The studies to compare TTR stability and amyloid formation were performed as a comparative analysis between the human CRTAC (hCRTAC) and two sea bass CRTAC1 variants (dlCRTAC1a and dlCRTAC1b). The corresponding sequences were cloned into a pET11a vector, expressed in E. coli and purified by continuous elution electrophoresis (Model 491 Prep Cell).
The secondary structure of human and teleost CRTAC1’s was evaluated using far-UV CD spectra measurements. All CRTACs had high β-sheet content (Fig. 5A) and hCRTAC1 and dlCrtac1a with an ellipticity minimum at 211nm shared the greatest structural similarity.
CRTAC1s stability was assessed by measuring CD ellipticity minimum (210 nm) changes during heating from 20 ºC to 95 ºC, which denotes thermal unfolding. Complete loss of secondary structure, was not achieved even at 95 ºC, suggesting high thermostability of the protein (Fig. 5B). The loss of secondary structure started around 50 ºC and was much more pronounced for hCRTAC1 compared to dlCrtac1a and dlCrtac1b. The high thermostability of CRTAC1’s was further confirmed using intrinsic fluorescence spectroscopy (Fig. 5C) when no appreciable emission maximum shift, suggestive of tertiary structure unfolding, was detected after heating, but only in the presence of Guanidine Hydrochloride (6M) (Fig. 5D) indicating the proteins are susceptible to chemical but not thermal unfolding. Since CRTAC1 forms high molecular weight aggregates spontaneously in solution, the observed thermostability may also be a consequence of a high thermoresistance of the formed aggregates.
The presence of high molecular aggregates was confirmed by size-exclusion chromatography (SEC) (Fig. 6A) and PAGE (Fig. 6B) but the monomeric form of the proteins was also found is solution as suggested by the smaller peaks with higher retention volumes of approximately 14-15 ml (molecular weight approximately 40-70kDa). However, for dlCrtac1b, SDS-PAGE revealed that the protein in the higher retention volume fraction contained a mixture of both aggregates and monomers suggesting that monomeric dlCrtac1b was highly unstable and aggregated immediately after isolation.
In order to evaluate whether CRTAC1s form amyloid structures in vitro we incubated samples of CRTAC1s for 2 weeks at 37ºC and analysed them over time to follow the progression of aggregation and the morphology of the aggregates by TEM (Fig. 7). After 1 week at 37ºC both human and piscine CRTAC1 solutions contained elongated oligomeric structures that formed dense clusters of amyloid aggregates. After 2 weeks at 37ºC dlCrtac1a and hCRTAC1 samples had fewer aggregation clusters and more disperse smaller aggregation units while dlCrtac1b contained very well defined curvilinear protofibrils. In general aggregates of hCRTAC1 and dlCrtac1a formed faster and had less defined morphology than those of dlCrtac1b. No significant change in the aggregation patterns of the CRTAC1 proteins analysed was identified during 3 weeks incubation (not shown). In general aggregates, hCRTAC1 and dlCrtac1a formed faster and had less defined morphology than those of dlCrtac1b.
In order to clarify the structure of CRTACs and its evolution we planned to solve the crystal structure of three evolutionary distant CRATC homologues: human (hCRTAC1), sea bream (sparus aurata) (saCRTAC1b) and cyanobacter (Synechococcus sp) (cyCRTAC1). Because CRTACs are aggregation prone and hard to crystallize, we focused specifically on the most conserved and potentially more soluble beta-propeller region (this region has been crystallized in other beta-propeller proteins). Knowing the expression and purification of the proposed proteins might require thorough optimization we set the goal to produce soluble and sufficient amounts of proteins within this project and continue with the crystallography trials after the project. For this task we established a new collaboration that will continue after this project with the group of Prof. Liebermann at the Georgia Institute of Technology. Prof. Liebermann has extensive experience on the crystallization of difficult beta-propeller proteins.
The coding sequence of the β-propeller domain of the hCRTAC1, saCRTAC1b and cyCRTAC1 was each inserted in the vector pMalc4X for expression in E. coli BL21 as MBP-fusion proteins. After extensive optimization, we could obtain a significant amount of protein in the soluble phase (Fig. 8) that is now being purified and will be further used for crystallization trials.
Summary
Compliance with foreseen goals/milestones
Year 1
• Scientific aims achieved (within modified plan) and results analysed
• 1 manuscript published; 1 manuscript in preparation; 1 oral presentation
• 1 conference attended; 1 workshop attended; > 10 seminars attended
• Training aims achieved
Year 2
• Scientific aims achieved (within modified plan) and results analysed
• 1 manuscript published; 1 manuscript submitted; 1 manuscript in preparation; 3 oral presentation
• 1 conference attended; 5 workshops attended; > 15 seminars attended
• Training aims achieved
Year 3
• Scientific aims achieved (within modified plan) and results analysed
• 2 manuscripts published; 1 manuscript submitted; 2 manuscripts in preparation; 4 oral presentation
• 3 conferences attended; 9 workshops attended; > 20 seminars attended
• Training aims achieved
• New research position offered at Harvard Medical School
• Funding application for 2 new projects submitted shortly after reporting period 2
Training received:
Biophysical methods: Circular dichroism (CD) spectroscopy, Differential scanning calorimetry, pressure perturbation calorimetry (DSC and PPC), Transmission electron microscopy (TEM), Protein Crystallography, Fluorescence methods. Cell culture, fluorescence microscopy, immunohistochemistry (returning host)
Science communication: oral presentations, workshops, publication of a layman summary, review/book chapter writing
Management and interpersonal skills: management of the MSCA project (budget and science) independently, managed 7 collaborations, including MSCA project partners
Mentorship and leadership: helped training graduate students in the laboratory: teaching assistant at Boston University graduate program, courses and seminars.
• The potential impact (including the socio-economic impact and the wider societal implications of the project so far) and the main dissemination activities and exploitation of results (not exceeding 10 pages).
WP1- Apolipoprotein AI misfolding
Amyloid diseases, including neurodegenerative and systemic amyloidosis, affect millions of patients worldwide and are a major public health challenge. Apolipoproteins, a family of lipid surface-binding proteins are overrepresented in amyloidosis. Hereditary apolipoprotein A-I (apoA-I) amyloidosis is a life-threatening type of systemic amyloidosis due to specific amino-acid mutations in the apoA-I protein. Mutant apoA-I proteins aggregate and deposit, often as N-terminal fragments in various organs, causing organ damage. So far, the molecular events triggering protein aggregation in this and other amyloid diseases are not completely understood.
In these studies, we collaborated with Sophie Valleix from Université Paris-Descartes and John E. Straub at Boston University to develop in-depth investigation of a new apoA-I mutation identified in patients with severe amyloidosis: apoA-I-E34K. We contributed to elucidate the molecular basis for the amyloidogenic properties of the apoA-IE34K variant and compared its conformation and dynamics with those of other disease-causing variants, Phe71Tyr (amyloidogenic) and Leu159Arg (non-amyloidogenic). We found that the mutations reduced protein stability and altered aromatic side chain packing and caused local helical unfolding. Our findings exemplify how diverse mutations alter local conformational dynamics in similar sensitive regions of a protein molecule. This phenomenon is probably not limited to apoA-I, but may extend to other globular proteins and their disease-causing variants
This contribution to elucidation of early steps on protein unfolding, along with improved diagnostics, can advance molecular understanding and therapeutic targeting of amyloid disorders. The study of amyloidosis as a basis to understand the process so effective treatments can be identified is of considerable socio-economic interest. This is particularly so in Europe where the average age of the population is shifting upwards and makes diseases of the aging population relevant from a series of socio-economic perspectives. The costs of amyloidogenic diseases for society are considerable in terms of health care needs, loss of work, injury but also because of their impact on the structure and functioning of societies.
WP2- Transthyretin misfolding
Transthyretin amyloidosis is another type of amyloid disease with high prevalence in countries like Portugal and Sweden. Human Transthyretin transports thyroid hormones in the blood in a tetrameric conformation. When the tetramer is destabilized the protein unfolds and aggregates into amyloid fibrils that deposit in organs or the nervous system causing severe and lethal polyneuropathies and cardiomyopathies. We had previously characterized TTR evolution using fish TTR models and hypothesized that N-terminal region, which was remodeled during evolution, could be involved in amyloid formation. In the STAMYEV project we provided evidences that the N-terminus is involved in the protein stability and modulates amyloid formation during evolution. We found that TTR from the ancient agnathan, lamprey does not form fibril despite the lower stability of the protein. Lamprey TTR has a longer N-terminus that could be responsible for blocking TTR aggregation. With this work we set up the grounds to further investigate in detail the molecular interactions in lamprey TTR that prevent the protein to enter the amyloid-formation pathway. We hope in the future to translate this information into targeting human TTR amyloid formation in disease and provide insights into development of new therapies.
The output of this work can continue to be shared between the outgoing host, Boston University, USA and the returning host, Canter of Marine Sciences, Portugal. The host in Boston works in direct collaboration with the Boston University Amyloidosis centre where basic science is combined with clinical interventions. Boston is a large hub of pharmaceutical companies, several of them directly involved in finding therapies for TTR amyloidosis, and working directly with the Amyloidosis centre in clinical trials. Any future outcomes from this project have the potential to impact the patients.
I will now transfer the skills, knowledge and scientific output of the STAMYEV project into my new position at Harvard Medical School (the laboratory will be transferred in the near future to the Cardiovascular Institute at Stanford School of Medicine). I will be working at the laboratory of Professor Liao, to help develop a project focused on Transthyretin amyloid cardiomyopathies. In this new project we will develop animal models of transthyretin amyloidosis and we will further test the N-terminal hypothesis in vivo, including different mutants. We will work in close collaboration with medical/clinical researches and pharmaceutical industry increasing to potential for the development of treatment strategies.
WP3- Cartilage acidic protein (CRTAC) aggregation and crystallization
Cartilage acidic protein1 (CRTAC1) is an extracellular matrix protein of chondrogenic tissue and its function remains unclear. CRTAC is putatively involved with diseases of the human cardiovascular, haematological, neurological, respiratory and urinary systems and potentially in multiple sclerosis. CRTAC also belongs to a large family of beta-propeller proteins that in mammals have been associated with diseases, including amyloid diseases such as Alzheimer’s. With our studies we found that CRTAC1 forms high molecular weight thermo-stable aggregates and revealed for the first time that such aggregates can consist of amyloid-like structures that might be related to its disease-association. We placed CRTAC protein within the landscape of the amylome and further contributed to the general understating of CRTAC1’s and beta-propeller family evolution and function. Overall, we provided new insights into the potential biological function of CRTAC1 in vertebrates and its possible role in disease. We have opened up the pathway for targeted studies in human to establish the link between the aggregating potential of this protein and amyloidogenic pathologies. Within my future projects at Stanford School of Medicine I will aim to further develop such studies. We will have access to TTR amyloidosis patient’s samples where I can try to identify the presence of CRTAC proteins in the aggregated form and investigate its putative interactions with different proteins in their amyloid conformation (e.g. TTR).
Overall, the project has had impact in diverse areas such as fundament biology, medicine, health, society and has carried out the groundwork for new targeted interventions in amyloid diseases. Amyloid related disorders represent a tremendous medical and socio-economic problem, specifically in ageing societies, such as in Europe or North America which are also two major hubs of amyloid research. Alzheimer’s disease, for example, is estimated to affect more than and 5 million people in Europe alone and where annual societal cost are estimated to exceed 55 billion €. The understanding of amyloid formation is a clear priority in the context of degenerative diseases and our studies that are aimed towards understanding the evolution of the protein structures and dynamics of amyloidogenesis directly contribute to that goal. Specifically, we helped to: 1) establish a novel molecular mechanism of apoA-I misfolding in Apolipoprotein amyloidosis and atherosclerosis, 2) support the novel hypothesis that TTR N-terminus plays a key role in TTR misfolding in amyloid, and 3) establish a link between evolutionary adaptation and amyloid formation by fish and human TTR.
Impact on scientific and personal development
The project had also a tremendous impact on my personal scientific development. I had the opportunity to deepen my knowledge and broaden my skills in structural biology, biophysics of protein folding and amyloid through expert training from Dr. Gursky, her laboratory team and collaborators. In addition to specific scientific training, the project challenged me to further develop/acquire complementary transferable competencies specially directed towards scientific maturity and independency (detailed below). I attended several training courses and seminars toward development of skills in teaching, mentorship, science communication, time management, leadership, scientific writing, job application and interviewing. I participated and presented my work in conferences and in seminars both nationally and internationally and I joined science outreached activities. I was also able to manage the project independently, both scientifically and administratively. I established and managed new collaborations. At the end of the project I was offered a position of Associate Scientist at Harvard Medical School and shortly after I will transition to Stanford University School of Medicine as faculty Instructor towards establishing a future independent position. I will continue to foster collaborations stemming from the MSCA project and disseminate its outcomes.
• The address of the project public website, if applicable as well as relevant contact details.
Not applicable
