Proposing a minimal set of metrics and methods to predict probabilities of amyloidosis disease and onset age in individuals

Transthyretin (TTR) protein amyloid

Human transthyretin (TTR, Uniprot # P02766) is a tetrameric, beta-sheet-rich, blood protein important in the transport of thyroxine and retinol. Blood-borne TTR is produced and secreted from the liver predominantly, although the choroid plexus of the brain does express high amounts [6]. In plasma, the TTR tetramer is in dynamic equilibrium with a monomeric form (Figure 1) with the degree of dissociation dependent on individual genetics and exogenous factors like binding of small hormones. TTR is less stable in monomeric form. TTR monomers may misfold and be incorporated into supramolecular arrangements of beta-sheets known as cross-beta structures [7–10]. These cross-β assemblies are stabilized by an extensive network of hydrogen bonds that cause them to form insoluble amyloids and resist proteolysis. Initially, misfolding is a spontaneous process [11, 12]. Once present, TTR oligomers may act in prion-like fashion [13, 14] to catalyze further misfolding of TTR into cross-β assemblies that accumulate into the fibrils and larger TTR amyloids as observed with multiple other amyloidoses [7, 15].

Biological factors promoting amyloidosis

Three general factors have been linked with certainty to the development of TTR amyloidosis: age, gender, and stability of the TTR tetramer. Development of TTR amyloidosis increases with age, occurs more frequently and earlier in men than women, and is enhanced by mutations that decrease structural stability of TTR tetramers [1–5]. It is unknown how these factors impact the half-life in plasma. The removal pathways have not been fully characterized, but in rats occurs predominantly in the liver [16]. TTR can be removed from plasma by proteolysis of dissociated monomers, and and by aggregation into fibrils, β–amyloids, and amyloid deposits in tissues. The rates and mechanisms for removal of fibrils and amyloid deposits are poorly understood, if indeed such pathways are significant. The relative amounts of TTR removed via these pathways depend on the relative rates of two processes, proteolysis and amyloid formation. Both are preceded by tetramer dissociation to monomers (Figure 1). If proteolysis is rapid enough to maintain TTR monomer concentration sufficiently low, transformations leading to amyloid deposits are not significant. With rapid protease activity, TTR monomer concentration in plasma is determined by tetramer stability. But when protease activity decreases relative to synthesis, monomer accumulates and can aggregate. Aggregation has an n^th order dependence on monomer concentration and even a small increase of monomer concentration greatly increases the probability of oligomer formation [14].

Amyloid accumulation can be slowed by small molecules which bind and stabilize the tetramer (tafamidis, diflunisal, and AG10), and limit monomer formation [17, 18] and thus slow but do not cure the disease. Another treatment approach arises from the observation that the TTR plasma concentration in an individual is maintained at a near constant value both before and after an amyloidosis develops [19, 20]. The rate of removal due to protein degradation generally decreases with age [21–23]. It is expected that if proteolysis slows to where it cannot keep up with the rate of formation of monomers, TTR is removed by formation of fibrils and β-amyloid deposits. We therefore hypothesize that the age-related imbalance in proteolysis initiates TTR amyloid diseases. Identifying the proteolysis pathways would provide the next step toward developing a preventive cure for TTR amyloidosis.

Individual differences in development of amyloidosis

Differences in lifestyle factors like diet, internal physiology, and mutant variants contribute to individual differences in the timeline of amyloidosis development. All three factors quantitatively affect the dissociation of the tetramer into monomer [9, 24] and because removal presumably proceeds via monomer, the relative rates of proteolysis and aggregation differ among individuals. Any TTR mutation, physiological condition, or exogenous factor that stabilizes the TTR tetramer also decreases monomer concentration. This favors a smaller ratio of aggregation rate to proteolysis rate, which delays or prevents amyloidosis. Based on clinical reports we expect physiological differences such as differences in plasma proteins and small binding molecules affect tetramer stability and thus development of amyloidosis [1, 3, 25–29]. These physiological differences may explain differences in the age of onset in men and women, but the in vivo studies required to quantitate these effects have not yet been done. Exogenous factors related to life-style and diet which modify metabolites and hormone signaling may also be important modulators of amyloidosis, but the significance of these factors is also unknown.

Mutations in the gene coding for TTR were evolutionarily selected to enhance its role as a transporter of thyroxine and retinol, but play an important role in amyloidosis. Since TTR mutations affecting amyloidosis become evident after the age of reproduction, no selective Darwinian pressure hindered acceptance and propagation of mutations that affect amyloidosis as long as the mutations had no substantial effect on transport functions. As a consequence, the TTR gene exhibits a large number (150 and counting) of antagonistic variants and a few protagonistic variants such as T119M, V30M, and L55P. [30, 31]. The most common variant is designated as ‘wildtype’.

The many mutations of TTR proteins appear to have minor effect on the ability of the monomers to be incorporated into β-fibrils and amyloids [9, 14]. All form remarkably similar wrapped, lamellar cross β-sheet structures [14, 31]. Even though mutant changes in single amino acids at the various locations on the TTR protein do not appreciably interfere with formation of β-sheet aggregates, these same changes do cause different tissue tropisms and corresponding differences in phenotypic expressions of the disease according to genotype [32]. For example, the amyloids formed from wildtype TTR show a predominantly cardiac tissue tropism where the β-amyloids accumulate in heart muscle. But, the symptoms often differ among individuals having the same TTR mutation and may range from carpal tunnel syndrome to neuropathy, spinal stenosis, gastrointestinal problems, Hashimoto’s disease, etc. and various combinations of these.

Some of the diversity in symptoms may arise because mutant subjects are frequently heterozygous with wildtype and mutant TTR co-expressed in equal amounts. Tetramer formation from mutants produces mixed hetero/homo-tetramers (i.e. wt₄; wt₃mt₁; wt₂mt₂; wt₁mt₃; mt₄). When the tetramers dissociate and monomers aggregate into fibrils, both wildtype and mutant subunits can be incorporated into the β-sheet structured oligomers. The fibrils thus contain mutant-specific variations in the protein structures at their surfaces that may create reactive surfaces which bind to cellular membranes with complementary binding groups. Mutant-specific differences in oligomer/fibril binding to cellular membranes could explain differing phenotypic patterns of TTR amyloidosis.

An integrated model for transthyretin (TTR) amyloidosis.

Studies of dissociation of tetrameric TTR have implicated both kinetics and thermodynamics in the amyloidogenic behavior of TTR. Two observations in the literature suggest that focusing on kinetics may be important: 1) the concentration of monomer in equilibrium with the tetramer is orders of magnitude lower than the concentration of tetramer [9, 26, 33–35], and 2) the exchange rate of monomers between wild-type tetramers is much faster at 4° C than at 25° C and even slower at 37° C [36]. Taken together, these two observations indicate the tetramer undergoes a fast dissociation followed by a temperature-dependent conformational change of the monomer. As a consequence of the conformation change in the monomer, at 4° C the association is much slower than the dissociation reaction and allows promiscuous mixing of monomers. This model accounts for both the slow exchange rate and inverted temperature dependence.

Figure 1 graphically summarizes the relation of amyloidosis to TTR concentrations, stabilities of TTR tetramers ((TTR)₄), and fates of monomers (TTR). This graphic makes it clear that quantifying the relative rates for monomer transformations is key to understanding development or lack of development of amyloidosis. The first step in removal of TTR is dissociation of the tetramer to monomer. Monomers that are not removed by proteolysis irreversibly aggregate into fibrils and β-amyloid deposits. The path from monomer to β-amyloid deposits occurs extracellularly and may include misfolded monomers, protofilaments, filaments, and fibrils, but it is generally agreed that monomers are folded into β-structures before, or upon their incorporation into, fibrils and maintain this structure in the fibrils and further into the construction of amyloids [37].

Formalizing a minimal kinetics mechanism (Figure 1) allows us to build a quantifiable model. Assuming the rate of dissociation and association between monomer and tetramer is fast relative to the rates of proteolysis and aggregation, the rates of formation and removal of TTR from plasma are described by the following equations.

$\text{Amino}\text{acids}\to {(\text{TTR})}_4$

$$\text{d}[{(\text{TTR})}_4]\text{/dt}=+{\text{k}}_{\text{synthesis}}$$(1)

$\text{TTR}\to \text{amino}\text{acids}$

$$\text{d}[\text{TTR}]\text{/dt}=-{\text{k}}_{\text{Deg1}}[\text{TTR}]$$(2)

$\text{TTR}\to \text{growth of TTR aggregates}$

$$\text{d}[\text{TTR}]\text{/dt}=-{\text{k}}_{\text{aggregate}}[\text{TTR}]$$(3)

Assuming the rate of production of TTR tetramer (equation 1) is constant, proteostasis is maintained by balancing the rates of removal (equations 2 and 3) with the rate of production, and therefore the sum of the rates of reactions 2 and 3 must equal 4*k_synthesis (equation 4).

$${\text{k}}_{\text{Deg1}}[\text{TTR}]+{\text{k}}_{\text{aggregate}}[\text{TTR}]=4({\text{k}}_{\text{synthesis}})$$(4)

Solving for the concentration of monomer, [TTR] gives

$$[\text{TTR}]=\text{4}({\text{k}}_{\text{synthesis}})\text{/}({\text{k}}_{\text{Deg1}}+{\text{k}}_{\text{aggregate}})$$(5)

Equation 5 suggests the concentration of TTR monomer depends on the relative values of the rate constants for the production and removal processes. The removal rate constants change with conditions that affect the rates of removal of the monomer. In the absence of aggregate formation, equation 5 predicts that the plasma monomer concentration is a constant that depends solely on the stability of the tetramer. Also, the ratio of k_synth to k_Deg1 may differ between individuals, so the concentration of monomer may differ as well. Therefore, an individual’s TTR amyloidosis risk could be assessed by measuring the individual’s serum TTR concentration and the stability of their TTR tetramer with respect to the monomer. The greater the concentration of tetramer and the lower the stability, the greater the chance that amyloid deposits will cause disease earlier in life. The ratio of a quantitative measure of tetramer stability to TTR serum concentration over time in a single individual (longitudinal) or across populations (cross-sectionally) at specific ages is thus predicted to be a useful index of the likelihood of an individual developing a TTR amyloid disease later in life. Individuals with the most destabilizing mutations or cellular conditions in which amyloidosis is initiated at earlier ages would need to be compared as a unique population cross section or longitudinally.

These relative stabilities in combination with the frequently observed fact that unfolding of protein tertiary and secondary structures are endothermic [38–40] allows us to construct a Gibbs energy diagram for amyloidosis (Figure 2).

${(\text{TTR})}_{\text{4}}\leftrightarrow \text{4TTR}\leftrightarrow {\text{4TTR}}^{\text{mis}}$(6)

Here TTR^mis is the misfolded monomer, either unfolded or a β-pleated sheet structure. Figure 2 shows that any change that decreases the Gibbs energy of the tetramer increases the activation energy for dissociation and decreases the concentration of monomer whether the monomer concentration is determined by thermodynamics or by kinetics. Drugs that bind to and stabilize the tetramer therefore decrease the monomer concentration. The figure also illustrates that misfolded monomers in vivo are tagged and hydrolyzed or aggregated because refolding and reentering the tetramer structure has a very low probability, i.e. amyloids are very stable relative to the unfolded intermediate. The likelihood of developing an amyloid disease is thus predictable from the Gibbs energy for dissociation of the tetramer and the serum concentration of an individual. Relative Gibbs energies of mutant variants and the effects of drug binding can be measured in near in vivo conditions as shown below.

Review of methods for determining TTR tetramer concentration and stability in serum samples from individuals

Measuring an individual’s TTR concentration and stability within the context of their blood serum is important clinically because it accounts for individual variations in environmental variables like hormone concentration or TTR binding partners. Any variable which may change TTR conformational distributions and stabilities would change the predicted risk of amyloidosis.

Although multiple methods exist to measure concentration and stability, we discuss methods which seem to us most relevant and easiest to employ. Current mass spectrometry based methods represent a viable approach for monitoring both concentration and structural stability within the complex mixture of the blood. These methods are simple to multiplex and currently employed for a wide variety of clinical diagnostics [41–43].

Concentration measurements

Multiple high quality reviews have been published on the topic of mass spectrometry for measuring protein concentration [44–47]. An optimized clinical assay for TTR might include a panel of internal standards composed of synthetic isotopically labeled peptides versions of natural TTR peptides (Figure 3A). This type of test could allow for simultaneous measurement of TTR concentration and test for known pathogenic mutations. Panels of internal standards for concentration measurement are commonly used for small molecules like drugs of abuse as well as protein specific measurements [48] and optimized assays are highly specific, require very little sample, and can be run on thousands of samples around the clock. Little or no change beyond selection of the peptides to be measured are required to monitor TTR versus any other protein.

Stability measurements

Protein structural stability has historically been assessed as the resistance to denaturation [49, 50]. As shown in Figure 3 the monomer sequence Figure 3A folds into both beta sheet and alpha helical secondary structures Figure 3B. The relative resistance to denaturation can be estimated based on non-covalent bonding within the secondary structure, and would be different in an alpha helix versus a beta sheet. If a reporter is located in a beta sheet rich region versus the alpha helical region, different stabilities would be expected based on the location of the reporter and the folding state (unfolded, monomeric, tetrameric, or amyloid) of the protein. For example, residues 95 to 101 are within the alpha helix and thought to be distant from any protein-protein interfaces in the tetramer (Figure 3C based on reference [33]). Based on the structure in Figure 3, the alpha helical section of the protein may have a similar structural stability regardless of protein oligomeric structure (tetramer versus monomer). The edges of the beta sheets on the other hand gain new binding interactions in the tetramer and would be significantly modified relative to the monomer. In amyloid, there is an overall loss of alpha helical structure and increase in the beta sheet for each monomer [5]. Thus, measuring the fold stabilities along the entire sequence of the protein could be valuable for differentiating between monomer, tetramer, and fibril/amyloid forms of TTR. Further, measuring the TTR structural stability in the context of blood serum (in situ) is important as ligands within the blood serum bind and stabilize the tetramer beta sandwich which could be read out as a shift in the structural stability of these sections of the protein.

Several mass spectrometric techniques have been published recently which allow in situ measurements of structural stability along the entire sequence of the protein [43, 51–53]. These methods can be generalized to chemical versus temperature denaturation of the protein (Figure 4). Both methods can measure sequence specific stabilities across TTR within a complex mixture of proteins like blood serum. Both methods also measure individual unfolding curves within complex mixtures by treating small aliquots to gradually increasing denaturation conditions. Each method has individual advantages and disadvantages. Thermal is fast, requires minimal sample handling, and uses the entire protein sequence as a reporter. However, the thermal mechanism functions through a precipitation step that is not fully characterized and may change with unknown variables. The chemical method uses covalent modification of specific amino acids to report unfolding. This is an advantage in that all the protein stays in solution and both labeled and unlabeled forms of the peptide can be measured and signal compared for substrate and product. The disadvantage of the chemical method is that the distribution of modifiable amino acids is nonuniform across any protein. For example, methionine labeling has been used to monitor chemical denaturation and has the sensitivity to identify stability changes due to drug binding [54] or secondary structure differences [55]. Unfortunately, only a single methionine is found in the N-terminal portion of the TTR protein (Figure 3A). Fortunately, a variety of different labeling methods exist in the literature that may be appropriate for monitoring protein structure [56].

Most of the published studies using mass spectrometry unfolding assays have used yeast or cell culture. To test whether these mass spectrometry methods might be appropriate for blood serum we performed preliminary tests. We found the thermal denaturation technique was not effective for blood serum. We tested temperatures up to 100° C, but there was no evidence of precipitation. This may be due to the extreme thermal stability of Albumin [57], stabilizing the rest of the proteins in the serum. In comparison, proteins from tissue homogenate precipitated efficiently beginning at 40C, and precipitation increased as predicted up to 70C, at which point 90% of the protein had precipitated from solution, as reported in the literature [52]. To test the chemical denaturation method, we used guanidine denaturation and histidine modification by iodine to assay blood serum. We observed differential behavior for several peptides and measured significant differences in the denaturation midpoint of two different sections of TTR (Figure 5). Although both of these sequences (A: 42-53, B:101-123) are expected to contain beta strands, the peptide A has a lower overall stability (denatures at lower concentrations) in these conditions than peptide B.

After optimization of modification chemistry for residues in sections of interest, this type of method has the potential to provide the stability measurements needed to assess risk of amyloidosis. This will require significant effort to validate in a cross-sectional study, and we expect that longitudinal measurements could be most fruitful if made throughout a subject’s lifetime.

Summary

1. This biophysical model and proposed measurement methods can be used directly with current human subjects or with banked blood samples. We propose that it will be of great benefit in current clinical trials to understand treatment efficacy.

2. The rate of synthesis of TTR is constant. For proteostasis, the rate of removal of TTR must equal the rate of synthesis. TTR in plasma is largely in the tetrameric form, (TTR)₄, but dissociates to give very low, but significant concentrations of dimers and monomers.

3. Removal of TTR from plasma proceeds via monomers.

4. Monomers undergo two processes that remove them from solution, proteolysis or aggregation. The combined rates of these two pathways equals the total rate of monomer removal, which is also equal to the rate of production of monomer via dissociation of tetramer. Depending on the relative rates, either of the two reaction pathways could account for anywhere from 100% to 0% of the rate of monomer removal.

5. The critical monomer concentration for aggregation is unknown, however the cause of aggregation develops slowly over time. Once amyloidosis begins, the rate of development of amyloidosis is determined by the rate of monomer incorporation into various aggregates that lead to fibrils and amyloids.

6. Destabilizing tetramer by pleiotropic mutations leads to greater dissociation of monomer and a higher, variant-dependent concentration of TTR monomer in plasma. Mutations are not required for TTR amyloidosis formation; point mutations only modify the equilibrium concentrations in Figure 6. Amyloidosis caused by wild-type TTR follows the same mechanism as amyloidosis caused by variants of TTR and thus should be considered as variants of the same disease for purposes of clinical studies.

7. Amyloidosis begins when the rate of TTR proteolysis decreases relative to the rate of amyloid formation and monomer concentration increases sufficiently to allow significant oligomerization into fibrils and amyloids. The cause of a decrease in the rate of proteolysis of TTR remains to be identified.

When the tetramer is stabilized by drugs or stabilizing mutations, the concentration of tetramer will increase in plasma to a steady-state level determined by the rate of proteolysis.

Chemical denaturation

Blood serum from a healthy human control was diluted with PBS to 10 mg/ml and divided into 9 fractions containing 0.02 milliliters. We added 0.027 milliliter guanidine of different stock solutions to each fraction to bring final guanidine concentration to 0, 0.4, 0.9, 1.3, 1.7, 2.2, 2.6, 3.0, 3.5M and partially unfold the proteins in the individual fractions as shown in Figure 4. This mixture was allowed to incubate at 37° C for 30 minutes at which point iodine/KI solution (0.3 molar) was added to achieve final concentration of 2.9 nmole in 0.05 milliliter. The iodine-labeling reaction proceeded for 10 minutes until 100mM imidazole was added to quench the reaction. At this point high concentration guanidine (6M) was added to fully denature all protein. Tris(2-carboxyethyl)phosphine (TCEP, final conc 10mM) was added to reduce disulfide linkages and chloroacetamide (40mM final concentration) was added to cap the reduced cysteines. The mixture was placed on top of a 30kD spin concentrator (VWR) and washed 2 times with ammonium bicarbonate (25 mM, ph=8.5). Trypsin (mass spectrometry grade) was added to the mixture on top of the filter and allowed to digest the protein overnight at 37° C. After digestion, peptides were spun through the filter and concentrated under vacuum for analysis by mass spectrometry.

Thermal denaturation

Blood serum was diluted with PBS to 2 mg/ml and 30uL volume was added to 10 PCR tubes. A PCR thermocycler (Biorad T100) was used to create an equally spaced gradient of temperature with 10 steps from 37 to 63 or 37 to 100 Celsius. Samples were incubated at these temperatures for 3 minutes to partially unfold the proteins in the individual fractions, then cool down to 4° C for 3min to quench the reaction as shown in Figure 4. The samples were then removed from the thermocycler and spin at 14000xG at 4° C for 20mins to precipitate the aggregated protein. The soluble fraction was collected for protein digestion. At this point high concentration guanidine (6M) was added to fully denature protein. Tris(2-carboxyethyl)phosphine (TCEP, final conc 10mM) was added to reduce disulfide linkages and chloroacetamide (40mM final concentration) was added to cap the reduced cysteines. The mixture was placed on top of a 30kD spin concentrator (VWR) and washed 2 times with ammonium bicarbonate (25 mM, ph=8.5). Trypsin (mass spectrometry grade) was added to the mixture on top of the filter and allowed to digest the protein overnight at 37° C. After digestion peptides were spun through the filter and concentrated under vacuum for analysis by mass spectrometry.