Transthyretin amyloidosis (ATTR) belongs to a class of disorders caused by protein misfolding and aggregation. ATTR is a disabling disorder of autosomal dominant trait, where transthyretin (TTR) forms amyloid deposits in different organs, causing dysfunction of the peripheral nervous system. We previously discovered that amyloid fibrils from ATTR patients are glycated by methylglyoxal. Even though no consensus has been reached about the actual role of methylglyoxal-derived advanced glycation end-products in amyloid diseases, evidence collected so far points to a role for protein glycation in conformational abnormalities, being ubiquitously found in amyloid deposits in Alzheimer's disease, dialysis-related amyloidosis and Parkinson's diseases. Human fibrinogen, an extracellular chaperone, was reported to specifically interact with a wide spectrum of stressed proteins and suppress their aggregation, being an interacting protein with TTR. Fibrinogen is differentially glycated in ATTR, leading to its chaperone activity loss. Here we show the existence of a proteostasis imbalance in ATTR linked to fibrinogen glycation by methylglyoxal.
Transthyretin amyloidosis (ATTR) belongs to a class of disorders that occur as a consequence of protein misfolding and aggregation, with transthyretin (TTR) forming insoluble cross β-fibre amyloid deposits [1–4]. ATTR is a disabling disorder of autosomal dominant trait, where protein is deposited at different organs, causing dysfunction of the peripheral nervous system (PNS) [1,4]. Partial suppression of organ function is eventually reached, resulting ultimately in full function loss, affecting the patient's life quality and causing death 10–20 years after disease onset . ATTR is presently recognized as a worldwide disease, with major incidence foci in Portugal, Sweden and Japan [3,6,7].
The most widely accepted molecular model of ATTR relates variations of TTR tetramer stability with TTR point mutations . According to this model, point mutations promote the dissociation of the protein tetramer, followed by misfolding of the monomers into an aggregation-prone conformation. Several TTR single point mutations are associated with ATTR. The most common one is the replacement of the valine residue at position 30 by methionine (V30M) .
Although TTR tetramer stability may be a significant factor in the initiation of amyloid fibril formation, there are a number of observations related to ATTR pathogenesis that must be considered in order to have a solid understanding of this disease. First, mutant TTR, which is present from birth, does not evolve to amyloid fibril prior to late adult life [10–12]. It is unreasonable to consider that the late onset of the disease is only due to tetramer destabilization, since the mutant protein is present from birth [10–12]. Secondly, wild-type (WT) TTR presents intrinsic amyloidogenicity, since its aggregation in older individuals causes systemic senile amyloidosis (SSA) . Finally, protein misfolding and aggregation are common features to many neurodegenerative diseases referred to as ‘conformational disorders’ [14–17]. These observations suggest that changes in the normal protein homoeostasis and post-translational modifications might contribute to pathogenesis of these conditions, since the decreased ability of the proteostasis network to cope with inherited misfolding-prone proteins, aging and/or metabolic/environmental stress appears to trigger or exacerbate proteostasis diseases [15–17].
Protein glycation is a ubiquitous hallmark of neurodegenerative diseases of amyloid type , and the accumulation of advanced glycation end-products (AGEs) has also been linked to protein aggregation . The most powerful glycation agent in vivo is methylglyoxal (MG), formed in all living cells as an unavoidable and non-enzymatic by-product of glycolysis . MG irreversibly modifies arginine and lysine side chains, resulting in a chemically heterogeneous group of AGEs, termed MAGEs (methylglyoxal-derived advanced glycation end-products) [21,22]. Argpyrimidine and hydroimidazolones are specific markers of protein glycation by MG in arginine residues, whereas Nε-(carboxyethyl)lysine (CEL) is derived from the specific reaction of MG with lysine residues [22,23].
MAGEs are found in several conformational pathologies such as Alzheimer's disease [24–27], dialysis-related amyloidosis  and Parkinson's disease . ATTR is no exception to this pattern since we discovered that amyloid deposits from ATTR patients are glycated by MG . Evidence collected so far points to a more intriguing contribution, as it seems that MAGEs can be implicated in the progression and outcome of the senescence phenomena and related pathologies. This strengthens the hypothesis that MG-derived protein glycation is involved in conformational disease and that glycation can be the missing link between metabolic factors proteostasis imbalance and disease.
Fibrinogen is a 340 kDa glycoprotein synthesized by the liver and assembled as a dimer of three different polypeptide chains bound to each other, termed α, β and γ [30,31]. Recent evidence from our group showed that, in the plasma of ATTR patients, fibrinogen is overexpressed and interacts with TTR . Interestingly, fibrinogen has also been described as an important player in other amyloid-like pathologies (Alzheimer's and Parkinson's diseases) [33,34]. In addition, human fibrinogen was reported to specifically interact with a wide spectrum of stressed proteins and suppress their aggregation [30,35–37]. Evidence suggests a potential role of fibrinogen in misfolding diseases as a molecular chaperone. Thus it is likely that, in patients with different pathologies, but sharing common molecular mechanisms, the increased levels of fibrinogen occur in response to a common increased need of extracellular chaperone activity [36,37].
In the present study, aggregation and fibre formation assays were performed in the presence of fibrinogen isolated from the plasma of healthy control subjects and ATTR patients. Significant differences were observed between the chaperone activity of fibrinogen-enriched fractions from healthy controls and ATTR patients. Since these individuals present different glycation profiles, mapping fibrinogen glycation by ultra-high-resolution mass spectrometry [Fourier-transform ion cyclotron resonance (FTICR)–MS] was performed. A structurally differential glycation pattern was found in fibrinogen from ATTR patients regarding healthy individuals. Moreover, glycated fibrinogen (GF) loses its capability as a chaperone and of interacting with TTR.
Our results support the existence of an imbalance in the proteostasis machinery in ATTR pathogenesis, as fibrinogen chaperone activity seems to be impaired in ATTR patients. Furthermore, this extracellular chaperone is differentially glycated regarding healthy individuals, suggesting that glycation by MG causes its chaperone activity loss.
Human samples and plasma collection
Blood samples from control healthy individuals and ATTR Portuguese type TTR V30M patients (three subjects each, all males, age range 23–38 years in the two cohorts) were collected in citrate-containing tubes. Samples were centrifuged at 1800 g for 5 min at 4°C. None of the individuals in the two cohorts presented any diabetic indication. The supernatant plasma was kept frozen at −80°C until further analysis. All subjects were characterized by gene typing to be heterozygous carriers for the V30M mutation except for the controls that do not bear any known TTR mutation. Heterozygosity of ATTR patients was later verified in our laboratory using an FTICR–MS-based assay [38,39]. At the time of transplantation, ATTR patients showed peripheral polyneuropathy or autonomic polyneuropathy without signs of amyloid deposition . All individuals gave informed written consent and the protocol was approved according to EEC ethical rules at Curry Cabral Hospital, Lisbon.
Preparation of fibrinogen-enriched fraction (FEC)
FECs were prepared by a modified ice-cold ethanol fractionation procedure . Human plasma from ATTR and healthy control subjects was diluted (1:10) in ultrapure water and precipitated by the slow addition of 0.22 volume of ice-cold 50% ethanol, lowering the temperature to −3°C. After centrifugation (10 min at 10000 g), the precipitate was washed with 0.5 original volume (O.V.) of 7% ethanol at −3°C and the solution was again centrifuged using the previous conditions. The precipitate was re-collected and dissolved in 0.25 O.V. 55 μM trisodium citrate buffer, pH 6.5, at 30°C for 30 min. The solution was cooled to 0°C followed by the addition of 20% ice-cold ethanol to a final concentration of 2%. After a spin-down, a mucus-like precipitate was removed and 20% ice-cold ethanol was added again to a final concentration of 8%, followed by centrifugation (10 min at 10000 g).
In vitro fibrinogen glycation by methylglyoxal
High purity MG was prepared by fractional distillation under reduced pressure under a nitrogen atmosphere as previously described . Once prepared, MG solutions were standardized by enzymatic assay with glyoxalase I and II . Purity was verified by HPLC analysis. For in vitro glycation, human fibrinogen prepared from plasma (1 mg/ml, Calbiochem) in 0.1 M sodium phosphate buffer (pH 7.4) was incubated with 0.2, 0.5, 2.5 and 10 mM of MG for 4 days at 37°C with stirring at 800 rpm, in the presence of NaN3 (Merck) to prevent microbial growth. Only one batch of human CF was used for this work.
Protein aggregation and amyloid fibre formation assay
Human insulin (SAFC Biosciences) was prepared in 0.1 M sodium acetate buffer, pH 4.6 (with 137 mM NaCl and 2.7 mM KCl) to a final concentration ranging from 5 to 100 μM.
Each TTR variant, 0.8 mg/ml solution (10 mM phosphate, 100 mM KCl and 1 mM EDTA, pH 7.0) of the different TTR variants was prepared in acetate buffer (200 mM sodium acetate, 100 mM KCl and 1 mM EDTA) at pH 4.6; 0.002% of sterile filtered NaN3 (Merck) was included in the aggregation mixtures.
Protein aggregation was followed by absorbance at 330 nm increase due increased light scattering in time. To monitor fibril formation, aggregation mixtures were added to a solution of 50 mM glycine/NaOH, pH 8.5, buffer with 0.5 μM thioflavin T. Fibril formation was revealed by the appearance of new emission maxima of the thioflavin T fluorophore, at 450 nm and 482 nm respectively, corresponding to the described maxima in its fluorescence spectra after binding to amyloid fibrils of different nature. Fluorescence measurements were made on a Fluorolog-3 instrument (Horiba Jobin Yvon) in a quartz cuvette with 1 cm optical path in Instrument Control Center v2.2.13 software. Both excitation and emission bandwidths were 2.5 nm and fluorescence measurements correspond to an average of ten readings. Fluorescence intensity time courses were recorded at 482 nm (excitation at 450 nm).
To monitor fibrinogen chaperone activity, the same reaction mixtures described previously were incubated with 0.5 μM and 1 μM fibrinogen [enriched fraction and Calbiochem® fibrinogen (CF) respectively], for 12–72 h at 37°C with constant stirring (900 rpm). Aggregates and fibre formation were detected as described before.
Polyacrylamide gel electrophoresis
Fibrinogen chains were analysed by SDS/PAGE in mini-gel format (7 cm×7 cm Tetra system from Bio-Rad Laboratories, 10% gel). For Western blotting, 2 μg was used per lane. For peptide mass fingerprint and glycation site mapping, 8 μg was used per lane. Prior to electrophoresis, samples were added of reduction buffer [6.25 mM Tris/HCl, pH 6.8, 20% (v/v) glycerol, 2% (w/v) SDS and 5% (v/v) s-mercaptoethanol] and heated at 100°C for 2 min. Protein bands were stained with Coomassie Brilliant Blue G-250 (Bio-Rad Laboratories).
Proteins were transferred from the polyacrylamide gel to PVDF membranes (Millipore). Membranes were blocked for 1 h at room temperature with TBS-T [10 mM Tris/HCl and 150 mM NaCl, pH 7.5, with 0.1% (v/v) Tween 20] containing 5% (w/v) non-fat dried skimmed milk powder. Membranes were then incubated overnight at 4°C in TBS-T containing 1% (w/v) non-fat dried skimmed milk powder with the primary antibody anti-argpyrimidine (JaICA monoclonal antibody, 1:10000 dilution), anti-fibrinogen (Calbiochem polyclonal antibody, 1:10000 dilution) and anti-TTR (Dako, 1:2000 dilution). Membranes were washed three times, 15 min each, with TBS-T and incubated for 3 h at room temperature with the secondary antibodies: anti-mouse IgG (Sigma–Aldrich, 1:4000 dilution) and anti-rabbit IgG (Sigma–Aldrich, 1:4000 dilution). Immunoreactivity was detected by chemiluminescence following the manufacturer's instructions (Pierce ECL Western Blotting Substrate).
In-gel protein digestion
In-gel protein digestion was performed as previously described [42-44]. Briefly, protein bands were excised, washed in ultrapure water, destained in 50% acetonitrile (ACN) and subsequently dehydrated with 100% ACN. Cysteine residues were reduced with 10 mM DTT and alkylated with 50 mM iodoacetamide. Gel pieces were dehydrated using 100% ACN and rehydrated at 4°C in digestion buffer containing either 50 mM NH4HCO3 with 6.7 ng/μl trypsin (modified porcine trypsin, proteomics grade, Promega) or 25 mM sodium phosphate buffer, pH 7.8, with 10 ng/nl endoproteinase Glu-C (sequencing grade, Roche). After 45 min, excess supernatant was removed and discarded, and 50 μl of 50 mM NH4HCO3 (for trypsin digestion) or 25 mM sodium phosphate buffer, pH 7.8 (for Glu-C digestion) were added. Digestions were allowed to proceed at 37°C overnight (∼16 h). After digestion, the remaining supernatant was removed and stored at −20°C until further analysis.
Samples were desalted and concentrated in home-made micro columns containing reverse-phase medium POROS R2 or OLIGO R3 (Applied Biosystems) and eluted sequentially to the MALDI target AnchorChip (BrukerDaltonics) with the appropriated matrix. For α-cyano-4-hydroxycinnamic acid (CHCA, Fluka), 10 μg/μl matrix solutions were prepared in 0.1% trifluoroacetic acid (TFA) with 20%, 50% or 80% ACN. The 10 μg/μl 2,5-dihydroxybenzoic acid matrix (DHB, Fluka) was prepared in 0.1% TFA with 10%, 50% or 80% ACN.
Peptide mixtures were analysed by MALDI–FTICR–MS in a Bruker Apex UltraQe, Apollo II ESI–MALDI combi-source (Bruker Daltonics), with a 7 T magnet (Magnex Corporation). Monoisotopic peptide masses were determined using the SNAP 2 algorithm in Data Analysis software version 3.4 (Bruker Daltonics). External calibration was performed by using BSA tryptic digest spectrum, processed and analysed with Biotools 3.1 (Bruker Daltonics). For protein identification purposes, monoisotopic peptide masses from non-GF peptides were used for database search using Mascot (Matrix Science; http://www.matrixscience.com). Data were submitted and analysed with BioTools 3.1 (Bruker Daltonics). Database searches were performed against Swiss-Prot, a non-identical protein sequence database (http://www.uniprot.org/uniprot/). The following criteria were used to perform the search: (1) mass accuracy better than 5 ppm; (2) one missed cleavage in peptide masses; and (3) carbamidomethylation of cysteine and oxidation of methionine as fixed and variable amino acid modifications respectively. Criteria used for protein identification in the Mascot software were (1) significant homology scores achieved in Mascot; (2) at least 20% sequence coverage and four peptide matches; and (3) similarity between the protein molecular mass calculated from the gel and from the identified chain.
For the analysis of glycated peptides, BioTools was used to compare the obtained monoisotopic peptide masses with predicted monoisotopic mass values considering optional modifications by MG. In the case of tryptic digests, hydrolysis was considered not to occur at modified sites and two missed cleavages were allowed.
For MS/MS analysis, a MALDI–TOF/TOF 4800 Plus mass spectrometer (Applied Biosystems) was used. Glycated peptides identified with BioTools were used to create an inclusion list for tandem experiments. MS/MS analyses were performed using CID (collision-induced dissociation) with 1 kV collision energy at a pressure of 1×106 Torr. Two thousand laser shots were collected for each MS/MS spectrum using a fixed laser intensity of 4000 V. Raw data were generated by the 4000 Series Explorer Software v3.0 RC1 (Applied Biosystems). MS/MS data were analysed in Data Explorer 4.5 (Applied Biosystems).
Plasma TTR co-purification
Fibrinogen and GF (1 mg) were coupled on two Aminolink columns as described in the Aminolink Plus immobilization kit manufacturer's protocol. Columns were washed with PBS and were incubated with human plasma samples (1:10 dilution in PBS) for 1 h at room temperature, with continuous rocking. After washing the columns four times with PBS, proteins were eluted by gravity-flow with 2 ml of 0.1 M glycine/HCl, pH 3.0, following neutralization with 0.1 M NaOH. Western blotting was performed after resolving the eluted protein samples by SDS/PAGE blotted on to a PVDF membrane. Membranes were blocked in 5% non-fat dried skimmed milk powder, and the proteins were probed against 1:2000 diluted polyclonal rabbit anti-TTR antibody (Dako) overnight, and then incubated with 1:4000 diluted horseradish peroxidase-labelled anti-rabbit IgG.
TTR expression and purification
TTR human genes (for WT and V30M variant) were amplified by PCR. PCR was carried out using primers with recombination sequences (‘Gateway att’ sites) added to the 5′ and 3′ end of the gene (forward primer 5′-GGGGACAAGTTTGTACAAAAAAGCAGGCTTCGAAGGAGATAGAACCATGGGCCCTACGGGCACCG-3′ and reverse primer 5′-GGGGACCACTTTGTACAAGAAAGCTGGGTATTCCTTGGGATTGGTGACG-3′). PCR amplification products were inserted into the pDONR™221 plasmid from the Gateway System (Invitrogen) and sequenced. The gene cassette in the Gateway Entry clone was then transferred to a Gateway Destination vector Pet23-b with His tag at the C-terminus using the proprietary enzyme mixture, ‘LR Clonase’ (Invitrogen).
Recombinant human TTR variants were expressed in Escherichia coli strain BL21(DE3). A preculture (10 ml) was grown overnight at 37°C in LB broth with ampicillin (100 μg/ml) and inoculated into 2.5 litres of fresh LB medium. Protein expression was induced with 1 mM IPTG when the attenuance at 600 nm reached 0.5. After growth for 4 h at 37°C with shaking, cells were harvested by centrifugation at 5000 g for 10 min at 4°C.
Cell pellets were suspended in 50 ml of lysis buffer (50 mM Tris/HCl, 400 mM NaCl, 10 mM imidazole and 0.5 mM PMSF, pH 7.5) and lysed by sonication at 4°C. After centrifugation at 5000 g for 30 min at 4°C, supernatant was added to 1.5 ml of Ni2+-nitrilotriacetic acid (Ni-NTA) beads (Qiagen). The lysate/Ni-NTA mixture was mixed gently by shaking at room temperature for 2 h. Beads were washed three times for 5 min with 10 ml of washing buffer (50 mM Tris/HCl, 400 mM NaCl, 10 mM imidazole, 0.5 mM PMSF and 10% glycerol, pH 7.5). Elution of the fusion proteins was carried with 5 ml of elution buffer (50 mM Tris/HCl, 400 mM NaCl, 250 mM imidazole and 0.5 mM PMSF, pH 7.5). Solutions of recombinant TTR variants were dialysed against 20 mM sodium acetate, 6 mM HEPES and 0.8 mM PMSF, pH 7.5 and then lyophilized.
Purified TTR protein variants were subjected to gel filtration on a Sephacryl S-200 HR column (GE Healthcare) pre-equilibrated with 30 mM HEPES 10 mM NaOAc, pH 7.4. Elution was performed at a flow rate of 0.5 ml/min in the same buffer. Apparent molecular masses were assessed based on elution volumes of suitable markers (BSA, 66.5 kDa; SmpB, 18 kDa).
Atomic force microscopy
Tapping mode AFM was performed in air, on a multimode AFM instrument with Nanoscope IIIa controller from Digital Instruments, Bruker. Etched Si probes with a spring constant of 42 N·m−1 and a resonance frequency of about 300 kHz (TESP, Bruker), and a scan rate of approximately 1.6 Hz were employed for imaging. Approximately 50 μl of sample was placed on to a freshly cleaved atomically flat mica surface for 20 min and dried under nitrogen.
Fibrinogen chaperone activity is decreased in ATTR patients
The description of fibrinogen as a molecular chaperone  highlights its potential role in misfolding diseases, in particular in ATTR, since it was found to be an interacting partner of TTR . Hence, we investigated whether fibrinogen from ATTR patients and control subjects displayed a differential chaperone activity.
Fibrinogen from ATTR patients and healthy individuals was purified using a modified fibrinogen-enrichment protocol. By Western blot analysis we observed that fibrinogen is abundant in the enriched fraction, whereas TTR and albumin were not detected (Figures 1B and 1C). Indeed, the most abundant protein in all plasma electrophoretic profiles corresponds to human serum albumin (HSA) that is absent from the enriched fraction (Figure 1A).
Fibrinogen chaperone activity is altered in ATTR patients
SDS/PAGE analysis of control and ATTR fibrinogen-enriched samples shows homogeneous protein profiles between the two sets of individuals and within subjects in the same group (Figure 1D). In the SDS/PAGE of the enriched fraction the most abundant proteins present are the α, β and γ chains of fibrinogen and whose identities were confirmed by peptide mass fingerprint (Figures 1D and 1E). Furthermore, using gel-densitometry analysis, we observed that fibrinogen represents 50% of the whole protein content in the enriched fraction (Supplementary Figure S1).
To evaluate the chaperone activity of fibrinogen, a well-established amyloid model protein, insulin, was used. Aggregation and fibre formation were induced under acidic conditions and detected by light scattering and thioflavin T fluorescence respectively [42,43] (Supplementary Figure S2). Human CF was able to suppress both aggregation and fibre formation of insulin in vitro in a dose-dependent manner (Figures 1F and 1G). Aggregation assays were repeated in the presence of fibrinogen-enriched fractions from ATTR and from healthy individuals. Our results show that the enriched fraction from healthy individuals decreased insulin aggregation by 50% (Figure 1H) (P=0.0280), and decreases fibre formation by 40% (Figure 1I) (P=0.0018). Fibrinogen from ATTR patients did not show statistical significant effects on the suppression of insulin aggregation neither on fibre accumulation. To rule out any consequence of a molecular crowding effect on the observed chaperone activity caused by other proteins present in the fibrinogen-enriched fractions , we conducted the same assay using BSA at the same respective molar protein concentration of the enriched fraction. We observed that BSA does not influence the observed chaperone effects (Figures 1H and 1I). To confirm that fibrinogen was indeed responsible for the suppressive effect on insulin aggregation, polyclonal anti-fibrinogen antibody was added to the reaction mixture containing the fibrinogen-enriched fraction from healthy individuals. This antibody binds selectively to fibrinogen, preventing its interaction with natural target proteins, thus inhibiting fibrinogen chaperone activity . Our results show that the fibrinogen-enriched samples from healthy subjects incubated with this antibody exhibit an aggregation and fibre formation pattern similar to that of insulin alone (Supplementary Figure S3). This confirms that, despite the heterogeneous composition of the enriched fraction, fibrinogen is the major component involved in the observed reduction in aggregation and fibre formation. These results strongly support the hypothesis that fibrinogen chaperone-like activity is decreased in ATTR patients compared with healthy individuals.
Fibrinogen from ATTR patients displays an increased glycation pattern
We have previously found that TTR fibres from ATTR individuals are glycated by MG  and that ATTR patients show increased glycation levels in some plasma proteins . Considering the fact that protein glycation by MG is known to exert effects on the activity of chaperone proteins , we speculated that this could also be the case of fibrinogen in the context of ATTR pathogenesis.
To evaluate the effects of glycation on chaperone activity, CF was glycated in vitro by incubation with increasing concentrations of MG from 0 to 10 mM. Glycation by MG was probed by Western blotting , confirming an increase in the MAGEs; the signal being proportional to the molecular mass increment of the fibrinogen chains. Fibrinogen incubated with different concentrations of MG was later incubated with insulin to determine the impact on its chaperone activity (Figures 2A and 2B).
Effect of GF on insulin aggregation
Since fibrinogen glycation by MG in vitro decreases its chaperone activity, we investigated whether the differential fibrinogen glycation in ATTR patients could explain the observed absence of reduction in insulin aggregation. Therefore, both fibrinogen-enriched fractions collected from healthy subjects and from ATTR patients were analysed by SDS/PAGE and argpyrimidine-modified proteins detected by Western blotting using a monoclonal antibody specific for this MAGE (Figures 2C and 2D) . Higher levels of argpyrimidine-modified proteins were found to be present in ATTR subjects by comparison with healthy individuals (Figure 2C). This was particularly noticeable in a protein band with a molecular mass slightly under 75 kDa that shows a stronger signal for argpyrimidine only in ATTR individuals (Figure 2C). This protein was identified as being the α-chain of fibrinogen.
Mapping fibrinogen glycation in ATTR
To investigate the molecular effects of fibrinogen glycation on its chaperone activity, we identified the location of the glycated residues in fibrinogen from ATTR patients and from healthy subjects. Since fibrinogen is a large multidomain glycoprotein composed of two pairs of three non-identical polypeptide chains (Aα, Bβ and γγ) [30,31], it is quite difficult to map its post-translational modification either by top-down or conventional bottom-up strategies. An added problem is the low abundance of glycation . Hence, the first step was to optimize MS detection and characterization of MAGE-modified peptides using in vitro GF (Supplementary Figure S4). To increase protein sequence coverage and the likelihood of detecting MG-modified residues, we combined three methods: (1) different micro-chromatography conditions, by using different resins and sequential peptide elution; (2) parallel use of more than one protease; and (3) different MALDI matrices. Combining peptide matches obtained with all of these methods allowed us to increase the average sequence coverage from 31% (classic approach based on trypsin digestion, R2 sample clean up and single matrix elution) to an outstanding sequence coverage of 82%, more adequate for post-translational modification search. The ultra-high resolving power of FTICR–MS was critical to resolve closely adjacent isotopic series, thus enhancing peptide ion detection and contributing to increased protein sequence coverage.
In vitro GF polypeptide chains were separated by SDS/PAGE, followed by in-gel digestion, sample cleanup and spectra acquisition. Given that MG-HI, argpyrimidine and CEL are the most frequent MAGE , we considered the respective mass increments for arginine and lysine residues  (Supplementary Figure S5).
We found a total of 56 different peptide ions exclusively for the GF mass spectra whose mass values corresponded to fibrinogen peptides containing MG adducts: 24, 21 and 11 for fibrinogen's α, β and γ chains respectively. In Figure 3(A) we show two spectra for the fibrinogen β chain of non-glycated (top) and GF (bottom) peptides. We confirmed the sequence of the non-glycated peptides by tandem MS (Figures 3B and 3F) and assigned the observed mass shifts (Figure 3C) corresponding to one and two MG-HI adducts (Figure 3D) and one argpyrimidine (Figure 3G). We detected a total of 36 different ions corresponding to peptides with ten MG-H1, 16 argpyrimidine and 16 CEL, for all fibrinogen chains. Interestingly, peptides containing more than one MG adduct were also found, for example two peptides, not present in the non-GF β chain, shown in Figures 3(A) and 3(B), corresponding to different MAGEs of the same peptide (TVNSNIPTNLRVLRSILE, m/z 2039.16). The peak with m/z 2147.19 matches this peptide with two MG-HI adducts in both arginine residues 166 and 169, determination further confirmed by MS/MS (Figure 3D), whereas the peak with m/z 2093.17 corresponds to the same peptide, but with only one mass adduct. We found using MS/MS that this last peptide carrying a MG-HI adduct was in fact a mixture of two peptides with one MG-HI adduct in each arginine residue. The parent ion is a contribution of peptides glycated exclusively in Arg166 or in Arg169 (Figure 3C). Another peptide (ALLQQERPIRNSVDE, m/z 1767.94) was detected containing either MG-HI or argpyrimidine adducts (Figures 3E and 3F). These data confirm the inherent complexity and heterogeneity of glycation .
In vitro glycation sites
To compare the glycation landscape of fibrinogen in ATTR patients with that from healthy subjects, we applied the MS method used to characterize the in vitro glycation pattern of fibrinogen.
The data acquired show greater spectral complexity, less intense signals and lower signal to noise ratio relative to the data obtained for in vitro studies, as expected when comparing post-translational modifications from in vitro and in vivo samples. Considering only non-glycated peptides and combining the data from all individuals in the same group, we obtained similar sequence coverage values for fibrinogen from both ATTR and control individuals, 85% and 83.6% respectively and a lower inter-individual variation.
Figure 3(A) shows a close up of one spectrum from a healthy subject's fibrinogen α chain. The peak in red, absent from the ATTR fibrinogen α chain spectra (Figure 3B), corresponds to one MG-HI adduct of the peptide WKALTDMPQMRME (m/z 1636.75). As shown in Figure 4, the peak with m/z 1779.87, matching the peptide MKGLIDEVNQDFTNR with one CEL adduct in Lys52 (Figure 4D) is present in the ATTR spectra, not being detected in the control's fibrinogen α chain.
In vivo glycation sites
We detected a total of 46 MAGEs for ATTR subjects (an average of 15 per individual), corresponding to 15 MG-HI, 15 argpyrimidine and 16 CEL adducts, for all three chains. In healthy control individuals, the total number of MAGEs found was 18 (making an average of six per individual) and from those eight were MG-HI, five were argpyrimidine and five were CEL adducts (Supplementary Figure S6, Supplementary Table S1).
Analysis of the distribution of fibrinogen glycation sites from ATTR patients show that glycated residues are mainly found on its globular domains (Figure 4C). In contrast, the fibrinogen glycated residues from control individuals show a wider and seemingly random distribution (Figure 4E). Moreover, glycation by MG is more extensive in fibrinogen from ATTR patients, and the in vivo results reproduce the inherent complexity and heterogeneity of glycation reactions observed for in vitro glycation analysis, particularly for such a large protein. Moreover, glycation seems to have a specific distribution in fibrinogen's globular domains, suggesting that this metabolic alteration can be related to fibrinogen's reduced chaperone capacity in these individuals.
Fibrinogen chaperone activity prevents TTR fibril formation
Glycation causes protein structural changes thus affecting protein–protein interactions. TTR interacts in a very stable way with fibrinogen (Figure 5A). However, the eluted TTR amount is reduced upon binding to glycated compared with non-GF. This indicates that glycation decreases with the interaction between these two proteins.
Fibrinogen glycation promotes TTR fibril formation
The ability of fibrinogen to inhibit TTR aggregation was investigated using recombinant TTR. As human TTR is mainly tetrameric, it was important to confirm whether the recombinant TTR variant could form a tetramer under physiological conditions. As presented in Figure 5(B), the elution profile obtained by size-exclusion chromatography, for both TTR WT and TTR V30M, showed in each case a single peak, with an estimated molecular mass of 60 kDa, the approximate mass of the tetramer.
TTR partially unfolds under acidic conditions, forming amyloid fibril, that is the hallmark of pathological conditions such as ATTR and SSA.
To evaluate the effects of glycation on fibrinogen's chaperone activity over TTR, we compared the ability of GF and non-GF to prevent TTR amyloidogenesis at pH 4.6 using light scattering and thioflavin T binding (Figure 5C). TTR WT amyloid fibril formation is lower than TTR V30M, as expected. AFM was used to confirm the presence of TTR amyloid fibres. After 12 h of incubation, WT TTR predominantly aggregates into fibrils with a length of 2−3 nm, although longer fibres were also present (Figures 5F and 5H).
As shown in Figures 5(C) and 5(D), TTR WT and TTR V30M aggregation and amyloid fibre formation were significantly suppressed by fibrinogen in a dose-dependent manner (P<0.05). This observation is in agreement with the effect of fibrinogen on insulin aggregation. Using AFM in the presence of fibrinogen WT, TTR fibrils were not detected after 12 h (Figure 5G).
Upon TTR fibril formation, in the absence of fibrinogen, no detectable TTR was found in the supernatant (Figure 5E, lane S1), most TTR being found in the insoluble fraction (Figure 5E, lane P1). By contrast, in the presence of fibrinogen, almost all TTR was recovered in the supernatant (Figure 5E, lane S2) and virtually no detectable TTR was found in the insoluble fraction (Figure 5E, lane P2). These results show the ability of fibrinogen to maintain TTR soluble and properly folded.
The extracellular space imposes an additional challenge to protein stability, because of its higher oxidizing environment and mechanical stress, due to the continuous blood pumping, enhancing protein unfolding and aggregation. Molecular chaperones are essential to intra- and extra-cellular proteostasis. Extracellular protein misfolding and aggregation underlie many of the most serious amyloid diseases, including ATTR, Alzheimer's disease, Huntington's disease and Parkinson's disease .
Concerning ATTR, extracellular chaperones such as clusterin have been implicated in the pathology of ATTR and SSA [52,53]. Another extracellular protein, fibrinogen, was found by us to be an interacting partner of TTR, being noteworthy that higher levels of fibrinogen were found in V30M ATTR individuals than in healthy control subjects [32,54]. Increased levels of fibrinogen were also found in both Alzheimer's disease and vascular dementia . Moreover, fibrinogen specifically interacts and suppresses aggregation of a wide spectrum of stressed proteins . In the present study, we demonstrated that fibrinogen from ATTR patients displays an impaired chaperone capacity, probably due to differential glycation by MG, when compared with fibrinogen from control subjects.
ATTR is an amyloid disease that is highly heterogeneous at the phenotypic level, regarding penetrance, incidence and also symptomatology. Considering TTR tetramer stability as a significant factor in the initiation of amyloid fibril formation, and that variant forms of TTR accelerate tetramer dissociation, one should expect that fibres would be essentially composed of TTR mutant forms, which is not the case. Furthermore, the amyloidogenic TTR variant, present from birth, does not cause amyloid fibril formation prior to adult life. This evidence strongly supports the hypothesis that there are non-genetic factors involved in the pathogenesis of amyloid fibril formation  that enhances disease penetrance and incidence. Several markers have been related to organism and cell senescence. In particular, some of these markers are related to stress phenomena and to the action of by-products resulting from metabolic processes. The formation and accumulation of AGEs has been linked to the progression of age-related diseases such as diabetes, Parkinson's disease and SSA [57–59].
Research concerning the role of protein glycation in disease progression is a field of study in great expansion . As with any post-translational modification, glycation has been described as having a significant effect on protein folding, conformation, stability, turnover and function  which may be associated with the cell and tissue damage observed in aging and several related pathological conditions , including ATTR . Our results show that fibrinogen from ATTR patients displays an increased glycation profile compared with healthy control subjects as well as a differential structural pattern. In ATTR, fibrinogen presents MAGEs mainly on the globular domains. Indeed FTICR–MS analysis allowed us to map the extension of this modification in both ATTR and control individuals. It was interesting to observe that, although the majority of the sequence pattern for glycation distribution was not conserved within individuals, three-dimensional motifs were preserved for AGEs in ATTR patients: MG-derived modifications in ATTR individuals have a specific spatial distribution preferentially localizing at the fibrinogen's globular domains. These observations reveal the inherent complexity and heterogeneity of glycation and suggest that this post-translational modification reduces the fibrinogen chaperone activity in ATTR individuals. Fibrinogen's globular domains are rich in basic amino acid residues and do not present a hydrophobic core, suggesting that this region is probably not directly involved in the chaperone activity. In fact, structural data on human fibrinogen indicate that this molecule must be flexible, mainly in the coiled-coil regions. We propose that glycation of target residues is sufficient to destabilize fibrinogen's quaternary structure and thus affect its ability to suppress protein aggregation, since it is already established that glycation is capable of affecting enzymatic activity [20,62]. In agreement with our observations, we speculate that the differential glycation profile observed in ATTR patients can explain the loss of chaperone activity observed in these individuals.
ATTR patients have an abnormal metabolism of glucose , as they show a high hypoglycaemic and hyperinsulinaemic profile after being administered with glucose. In fact, a modest increase in cellular glucose metabolism results in a substantial increase in AGEs accumulation . This suggests that an elevated cellular uptake of this molecule may contribute to the formation of glycated proteins.
Gonçalo da Costa and Carlos Cordeiro conceived and designed the experiments. Daniel Fonseca, Samuel Gilberto, Cristina Ribeiro-Silva, Raquel Ribeiro, Inês Batista Guinote, Susana Saraiva, Ricardo Gomes and Ana Viana performed the experiments. Gonçalo da Costa, Patrick Freire, Ana Viana, Ana Ponces Freire and Carlos Cordeiro analysed the data. Elia Mateus and Eduardo Barroso contributed reagents/materials/analysis tools. Daniel Fonseca, Samuel Gilberto, Gonçalo da Costa and Carlos Cordeiro wrote the paper.
We acknowledge Nurse Margarida, from Hospital de Curry Cabral, Lisboa, Portugal, for her outstanding co-operation in this work regarding sample collection.
This work was supported by the Fundação para a Ciência e a Tecnologia [grant numbers SFRH/BPD/74711/2010 (to I.B.G.), SFRH/BPD/41037/2007 (to R.A.G.), IF/00808/2013 (to A.V.), IF/00359/2014 (to G.C.), PEst-OE/QUI/UI0612/2011, UID/MULTI/00612/2013, PTDC/QUI/123060/2010, RECI/BBB-BEP/0104 and REDE/1501/REM/2005]; and the Amyloidosis Foundation 2011 Junior research grant.
advanced glycation end-product
Fourier-transform ion cyclotron resonance
human serum albumin
systemic senile amyloidosis
These authors contributed equally to this work.