Protein Expression Analysis by 2D-DIGE
To quantitatively investigate proteome changes of yeast cells in response to TTR expression we analysed and compared the proteomes of yeast BY4741 cells carrying the plasmid without the insert (control), TTR-wt (BTTR-wt) and TTR-L55P variant (BTTR-L55P). TTR expression was confirmed by MS analysis (data not shown) and western blot, where similar expression levels of TTR-wt and TTR-L55P were detected (). To analyse the presence of TTR aggregates, a protein aggregate filtration assay was performed (). This microfiltration method is based on the finding that high-molecular mass amyloid-like aggregates are SDS-insoluble, being therefore retained in a 0.2 µm blocked membrane. In fact, if the nitrocellulose membrane is blocked previously to the filtration procedure, TTR-containing insoluble protein aggregates are retained while soluble TTR do not bind to the blocked membrane, in contrast to a regular dot-blot assay using non-blocked nitrocellulose membrane (). Thus this method is indeed a protein aggregation filter trap assay. Using blocked nitrocellulose, a positive signal was only obtained in the crude extract indicating the presence of TTR in the insoluble fraction. As shown in , TTR aggregates insoluble in 2% SDS were clearly observed in yeast cells expressing the amyloidogenic L55P variant (BTTR-L55P). SDS-insoluble aggregates were not detected in BTTR-wt and, as expected, in the control (). In addition, a substantially higher TTR amount was found in the insoluble protein fraction of BTTR-L55P in comparison to BTTR-wt (). Thus, even though no cell toxicity and growth defects were observed (), TTR-L55P variant, when expressed in yeast, forms high-molecular mass amyloid-like aggregates.
Characterization of TTR expression in yeast.
The 2D electrophoretic maps contained around 1800 spots (). From these, a total of 78 protein spots were detected with a statistically significant change in abundance (ANOVA p<0.05) with an absolute variation ≥1.3-fold from at least one experimental group. Examples of three spot patterns are shown in . Spot 1461 increase its abundance only in BTTR-L55P while spot 1643 decrease its abundance in both BTTR-wt and BTTR-L55P. The spot 1698 is differentially expressed in both experimental groups but shows a significantly higher abundance in BTTR-L55P compared to BTTR-wt. A principal component analysis (PCA) analysis shows that 2D gel images cluster into three well separated groups (), indicating a clear differentiation between the expression of the non-amyloidogenic TTR-wt and the amyloidogenic TTR-L55P form with significant changes in protein abundances. In fact, by directly comparing the 2D-DIGE maps of BTTR-wt and BTTR-L55P with the control, a total of 24 and 75 spots, respectively, were differentially expressed, highlighting a much higher induced change upon expression of the amyloidogenic TTR variant. Interestingly, of the 24 protein spots with changes in abundance between BTTR-wt/control, 21 spots were also differentially expressed in BTTR-L55P/control with a similar fold variation. These changes are likely to be due to the heterologous TTR expression and not because of TTR misfolding. However, 54 protein spots exhibited significant changes in abundance exclusively upon TTR-L55P expression, suggesting that relevant proteins involved in the cell response to TTR misfolding and aggregation are revealed with this approach.
2D-DIGE differential protein expression analysis.
Protein Identification and Gene Ontology Analysis
All spots highlighted in were picked and trypsin digested using the Ettan Spot Handling Workstation and the proteins were identified by MALDI-TOF-TOF MS. With this approach, we were able to identify the corresponding proteins in 73 spots, resulting in the identification of 70 unique proteins (). For the majority of the identified proteins, the molecular mass and isoelectric points determined on the 2D gel are consistent. In some cases, the same protein is identified in different spots across the 2D gel with different molecular mass and isoelectric point suggesting the presence of post-translational modifications and/or protein isoforms. All spots representing the same protein have a very similar regulation (see for example spot 991 and 994 both identified as alcohol dehydrogenase (ADH1) where a similar trend and fold variation was observed). In seven spots, more than one protein was identified (see ). In some cases, MSMS data allowed the identification of a particular protein isoform (example, spot 613 identified as HSP75). In other cases, this was not possible and thus both protein isoforms are shown in (for example, spot 1542 identified as enolase 1 and/or enolase 2).
Differentially expressed proteins identified by MALDI-TOF-TOF MSMS.
A significant change in protein abundance was clearly detected for spot number 1492, absent from the control (). This spot was unequivocally identified as human TTR. This is a noteworthy observation for two main reasons: first, it shows that our experimental system leads to a high TTR expression level; second, the detection of this expected difference validates the approach we chose to detect quantitative differences in protein abundances.
The identified proteins were categorised into functional groups and cellular location using Gene Ontology annotations. The 70 identified proteins fell into 10 functional categories (). About 45% are proteins related to cell metabolism, including carbohydrate (16 unique proteins, 23%), amino acid (6 proteins, 9%), energy (4 proteins, 6%), nucleotide (4 proteins, 6%) and lipid metabolism (1 protein, 1%). A significant number of the identified proteins (17 proteins; 24%) are involved in translation, including ribosomal proteins and translational factors. Noteworthy, a high number of the identified differentially expressed proteins are related to protein folding and degradation pathways (13 proteins, 19%). Several of these proteins have been described as stress response ones, involved in the response to an increase protein misfolding (as Hsp70 protein family). Proteins involved in transport (2 proteins, 3%), cell redox-homeostasis (3 proteins, 4%) and proteins with unknown or poorly characterized function (4 proteins, 6%) were also identified.
Gene ontology characterization of the identified differentially expressed proteins.
Concerning cellular location, a high number of identified proteins were from mitochondria (27%), while 22% were from cytoplasm, 12.5% from the plasma membrane and nucleus ().
Proteome Changes Induced by the Amyloidogenic TTR Variant
As described above, a clear differentiation was evident between the proteome of cells expressing the non-amyloidogenic TTR-wt and the highly amyloidogenic TTR-L55P variant. In BTTR-wt, 22 proteins were differentially expressed (), with 15 proteins up-regulated and 7 down-regulated (, grey). By contrast, in cells expressing the amyloidogenic TTR-L55P, significant proteome changes were induced with 67 unique proteins differentially expressed, 49 being up-regulated and 18 down-regulated (, detailed in , black). To further explore the involvement of the uncovered pathways involved in TTR misfolding, we performed a functional enrichment analysis using DAVID (Database for Annotation, Visualization and Integrated Discovery). This analysis revealed 8 functional clusters with a significantly enrichment score (). Significant changes in cell metabolism (namely glucose and amino acid metabolism) and also in the regulation of translation and protein synthesis was noticeable. Biological themes related to plasma membrane and mitochondria proteins and a functional enrichment in the molecular chaperones network and in protein refolding was observed. In contrast, the DAVID analysis using the proteins differentially expressed in BTTR-wt revealed three significant clusters only: mitochondrial matrix (FDR of 6.10E-07), tricarboxylic acid cycle (FDR of 3.60E-06) and plasma membrane enriched fraction (FDR of 5.40E-03).
Expression profile in BTTR-wt and BTTR-L55P versus the control.
Figure 5 Detailed expression profiles for all the identified differentially expressed proteins, according to its functional categories: (A) cell metabolism; (B) unknown function; (C) Cell redox homeostasis; (D) protein folding and degradation; (E) translation. (more ...)
Functional annotation enrichment analysis of the identified proteins using the Database for Annotation, Visualization and Integrated Discovery (DAVID ) v6.7.
The high number of the differentially expressed proteins involved in metabolic processes hints that the cell response to protein misfolding stress is accompanied by active metabolic changes. The major metabolic pathways altered are illustrated in . Following TTR-L55P expression, we detected an up-regulation of several glycolytic enzymes ( and ), pointing to an increase in glucose catabolism. In addition, the two enzymes that catalyse ethanol formation from pyruvate (PDC1, pyruvate decarboxylase 1 and ADH1, alcohol dehydrogenase) are down-regulated ( and ). This could reflect a shift in the pyruvate fate, from alcoholic fermentation to the TCA cycle and oxidative phosphorylation. Indeed, the TCA cycle enzymes citrate synthase (CISY1) aconitate hydratase (ACON) and isocitrate dehydrogenase (IDHP), which are responsible for the synthesis of α-ketoglutarate from acetyl-CoA, are up-regulated in BTTR-L55P in compassion to control cells. Consistently, an increased abundance of ATP synthase (ATPB) and a down-regulation of the pentose phosphate pathway enzyme transketolase (TKT1) and (DL)-glycerol-3-phosphatase 1 (GPP1), an enzyme involved in glycerol synthesis, was detected. Altogether, these results point to the channelling of glucose catabolism through the TCA cycle, leading to an increase ATP production via cell respiration (). Notably, this metabolic change is not apparent in BTTR-wt ( in grey). In these cells, D-glucose consumption and ethanol production are similar ().
Major metabolic pathways altered in BTTR-L55P.
The increased ATP demand is surely related with the higher energy needed to actively refold or degrade misfolded proteins. In fact, a similar metabolic change was apparent in the response to heat shock that is also characterized by an increased protein unfolding and aggregation 
. An undesirable by-product of an increased ATP formation via cell respiration is the formation of reactive oxygen species such as superoxide anion which may activate the oxidative stress response. This explains the up-regulation found exclusively in BTTR-L55P of superoxide dismutase [Cu-Zn] (SODC). These data may suggest a link between protein misfolding and oxidative cellular stress derived from a higher ATP demand. It has been show that oxidative modifications may facilitate aggregation of amyloidogenic proteins 
. Upon aging, cellular defences towards oxidative stress are compromised and oxidative protein modifications are also likely to accumulate, which may be synergistically linked to the disease onset 
Another major set of differentially expressed proteins are involved in translation, with a complex pattern of expression (6 proteins down-regulated and 10 up-regulated in BTTR-L55P) (). These changes may reflect an adaptive response to cellular stress. It was reported that elongation factors are up-regulated in response to stress conditions, such as oxidative stress 
. In addition to its canonical role in translation, unique cellular activities, such as nuclear export, cytoskeleton organization and apoptosis, have been attributed to elongation factor protein family in eukaryotes 
. Interestingly, a potential role in protein quality control and co-translational degradation has been suggested for these proteins 
. Elongation factors interact with the 26S proteasome and this association increases when translation is inhibited 
. Thus, the identified proteins may be important in the cell response to protein misfolding in a more complex way than a simple activation or inhibition of protein synthesis. Although further studies are needed to clarify this issue, our findings provide a good starting point by revealing potential protein targets.
Several stress-response proteins involved in protein folding and/or degradation were also identified, with 13 proteins differentially expressed in BTTR-L55P (10 up-regulated and only 3 down-regulated, ). Some heat shock proteins were also found up-regulated in BTTR-wt by a similar fold variation (HSP77, HSP60, HSP72 and HSP10; , in grey). Interestingly, HSP77, HSP60 and its co-chaperone HSP10 
are mitochondrial-resident chaperones, involved in folding of newly imported proteins to the mitochondria. The other identified proteins involved in protein folding and degradation changed their abundance only in BTTR-L55P. The potential role of these proteins in TTR misfolding and aggregation is illustrated in .
TTR misfolding and protein quality control mechanisms.
Two interesting protein revealed by our study are the cyclophilin FK506-binding protein 1 (FKBP) and cyclophilin A (CYPH). These proteins, that have a peptidyl-prolyl-cis-trans isomerase (PPIase) activity, were up-regulated in BTTR-L55P (). FKBP changed its abundance also in BTTR-wt but it increases significantly in BTTR-L55P (1.4 fold vs 2.2, ). The involvement of PPIases in the cell response to protein misfolding and aggregation is still unclear. However, recent experiments implied this protein family in conformational neurodegenerative disorders. It was showed that FKBP52 overexpression reduced the Aβ peptide toxicity and increases the lifespan of flies expressing Aβ peptide, whereas loss of function of FKBP52 exacerbated these Aβ phenotypes 
. Our previous results revealed that PPIase cyclophilin H is one of the major TTR interactuant in human plasma of ATTR patients 
. In yeast cells, FKBP was shown to interact with the heat shock factor 1, a major regulator of the cell response to stress conditions such as heat shock and protein misfolding 
. In addition to FKBP, CYPH was up-regulated by 1.7 fold exclusively in BTTR-L55P. The human homologue of yeast CYPH is a major Aβ-peptide interactuant found in the brain and elevated levels of this protein were reported in human Alzheimeŕs disease brains 
. Moreover, it was showed that the Aβ oligomeric form has a greater affinity for CYPH, hinting for a relevant role of this protein in protein aggregation 
. Interestingly, yeast CYPH interacts with several proteasome regulatory subunits and also with ubiquitin and SMT3 (yeast homologue of mammalian SUMO1) 
suggesting a role in protein degradation associated processes. Altogether, these observations suggest an important role of PPIases in disease development and cellular responses to protein aggregation in the context of conformational disorders and are good protein targets for further studies.
DIGE Screen identified three additional molecular chaperones exclusively in BTTR-L55P: an increased abundance of HSP71 (SSA1 gene) and HSP72 (SSA2 gene) and a reduction in HSP75 (SSB1 gene). Using an animal model of ATTR, it was recently observed that TTR deposition leads to an increase in HSP70 expression 
, in agreement with our 2D-DIGE analysis. However, contrary to our study, the interplay between the different isoforms of the Hsp70 was not revealed, which is highly relevant considering that a functional difference between members of SSA and SSB is apparent 
. Both classes of chaperones affect de novo
prion formation in yeast, although with opposite effects 
. After heat shock, a similar trend was observed with the SSB isoform being down-regulated while the SSA isoforms is up-regulated 
Besides HSP70, the endoplasmic reticulum lumen resident protein disulfide isomerase (PDI) was up-regulated exclusively in BTTR-L55P, which appears to be in concordance with an up-regulation of the HSP70 resident endoplasmatic reticulum chaperone BiP in tissues affected with TTR aggregation 
. Thus, even though a simple model organism was used in this study and TTR was used as a model amyloidogenic protein, it is likely that relevant protein targets, even for familial amyloidosis, was revealed by this screen.
The action of molecular chaperones is an essential first step to avoid protein aggregation. At this point, misfolded proteins are either refolded or degraded, avoiding their accumulation. It is now recognized that some molecular chaperones are also involved, either directly or indirectly, in protein disposal. HSP71, found up-regulated exclusively in BTTR-L55P, and its co-chaperone HSP40 specifically recognize misfolded protein domains and escort them for proteasome degradation 
. Yeast HSP71 is in fact known to interact with proteasome regulatory subunits (like the 26S RPN 2) 
. Noteworthy, the ubiquitin-like protein SMT3 was found up-regulated exclusively in BTTR-L55P (). This protein displays 50% sequence identity with mammalian SUMO1 protein and is essential for yeast viability. Protein sumoylation is known to be involved in neurodegenerative disorders such as Alzheimer’s 
and Huntington’s diseases 
although its exact role is presently controversial. For the Huntington protein, sumoylation renders the protein more soluble and apparently more toxic by inhibiting its aggregation into inclusion bodies 
. In Alzheimer’s disease, SUMO1 overexpression in tissue cultured cells co-transfected with the APP gene suppresses Aβ fragment accumulation 
. Studies in human samples of ATTR patients showed that TTR aggregates lead to a significant increase in ubiquitin conjugates and an impairment of the ubiquitin-proteasome system was observed 
Two proteins involved in protein folding and degradation were down-regulated exclusively in BTTR-L55P: UBX domain-containing protein 1 (UBX1) and protein BMH2 (Brain Modulosignalin Homologue, member of the ubiquitous 14-3-3 gene family) (, in black). UBX1 is known to interact with proteasome regulatory subunits and with ubiquitylated proteins in vivo
, being required for the degradation of an ubiquitylated model substrate 
. Interestingly, UBX1 also interacts with FKBP 
, with several isoforms of the HSP70 family (HSP71 
and HSP77 
) and also with PDI 
, proteins differentially expressed in response to TTR-L55P expression. Concerning BMH2, it is likely to be related to carbohydrate metabolism and stress response 
. It was observed that the expression of BMH proteins is altered after exposure to several stress conditions such as heat shock or dithiotreitol 
. To our knowledge, no association between 14-3-3 proteins and TTR aggregation has been described until now.
Three proteins with unknown or poorly characterized functions were also identified (). In addition to a down-regulation of AIM29 (already detected in BTTR-wt), the UPF0001 protein YBL036C was also down-regulated, while the uncharacterized protein YJL217W and cytochrome c oxidase assembly protein COX14 increase its abundance. The biological function of these proteins is not yet known. Some data seem to relate the YJL217W protein in the regulation of enolase1 
. This could be related with the detected increase in glucose catabolism and, considering that yeast enolase1 also function as a heat shock protein 
, with the misfolding of TTR-L55P. This is an interesting hypothesis that requires further investigation, feasible in yeast considering that the ENO1 yeast null mutant is viable. COX14 Is an integral mitochondria membrane protein and a yeast null mutant for this protein displays a respiratory growth deficiency 
. Thus, the increased expression of this protein may results in an improved mitochondria function. No significant human homology was found for this protein. Finally, UPF0001 protein YBL036C also increased significantly its abundance in BTTR-L55P. A BLAST homology search revealed that this yeast protein shares 43% identity with proline synthase co-transcribed bacterial homolog protein, whose function is nevertheless not yet known.