Life expectancy has exponentially increased in the past decades, bringing huge challenges to healthcare for the treatment of age-related diseases [1] [2] [3] . In this way, it is important to promote healthy aging and so, aging biomarkers are needed to timely assess the health status of elderly people [1] [3] [4] [5] .

The decline in proteostasis, i.e. a decrease in the activity and/or expression of the components of the unfolded protein response, is linked to the aging process [6] . This leads to the accumulation of aberrant proteins that can organize into aggregates that are potentially toxic to the cell [7] . Protein aggregates are extensively studied in the context of neurodegenerative diseases [8] [9] [10] , however only few studies have so far focused on protein aggregation during natural aging [11] [12] . Therefore, specific protein aggregates not associated to a specific disease can be potential biomarkers of physiological aging. To identify these aggregates, we need reliable methodologies that can detect changes in protein conformation in biological samples in a simple, accurate and inexpensive manner. Metabolomic approaches, such as Fourier Transform Infrared Spectroscopy, seem to be reliable and promising to study protein aggregation, since age-related metabolic modifications can result in structural changes that modify IR spectra [13] [14] [15] . In order to verify the applicability of FTIR in the study of protein aggregates, one needs to use model organisms whose proteins are known to aggregate. Saccharomyces cerevisiae is a well-studied model in the context of aging and several studies have been carried out to evaluate the mechanisms involved in cellular senescence, including protein aggregation [16] [17] [18] .

In this study, we have verified the applicability of FTIR to evaluate protein aggregation in S. cerevisiae. We have used specific mutants of S. cerevisiae that have the tendency to aggregate proteins at high level (unpublished). Our data clearly indicate that FTIR can be used to detect such protein aggregates.

In this study Saccharomyces cerevisiae, an extensively used unicellular model [17] [20] [21] [22] , was used to study protein alterations, with particular interest in PA. Wild-type and stress yeasts with induced PA were used. Figure 1 shows brightfield and fluorescence microscopy images of cells of S. cerevisiae, with and without protein aggregation. In WT strains cells show GFP fluorescence homogenously distributed in the whole cytoplasm, while in PA cells, fluorescence is concentrated in foci (Figure 1). Table 1 presents the percentage of cells with PA in both WT and mutant strains. WT clones have a reduced number of cells with protein aggregates. As expected, mutant strains present extensive PA.

To evaluate the potential of FTIR to detect protein aggregation, FTIR spectra of both WT and PA strains were acquired in the spectral range between 4000 and 600 cm⁻¹. An average spectrum of both WT and PA strains is present in

Figure 2, with emphasis in the region between 1700 and 1500 cm⁻¹ (Figure 2(b)), which is mainly responsible for the protein absorption (both Amide I and Amide II). Spectra were then subject to direct analysis with visualization of the spectroscopic signals related to biochemical pattern of all yeast samples. Once each spectroscopic signal is linked to a defined chemical group and/or chemical family of compounds it is possible to identify and compare the main differences between the two yeast groups. Since the spectra were obtained by directly inserting the colonies on the crystal, the dominant spectral signals were due to the water absorption and the remaining signals were underneath, so it was almost impossible to see differences between both spectra, only with direct spectra analysis. With the objective of increasing the differences between groups,

2^nd derivative processed by the Savitzky-Golay algorithm was applied to the spectra (Figure 2(c)).

For a better comparison and characterization among the different sample groups, mainly in the protein region, principal component analysis (PCA) was applied to the 2^nd derivative spectra (1700 - 1500 cm⁻¹) of yeast samples. PCA provides score and loading diagrams, allowing identifying and characterizing wild-type and mutant strains. 1700 - 1500 cm⁻¹ spectral region allowed to observe protein alterations in the samples since it is correlated with protein secondary structure (amide I and amide II). PCA results are presented in Figure 3. We did not use PC1 and PC2 for our statistical analysis sincethese PCs represent mostly water contribution (strong signal at ~1650 cm⁻¹) and do not represent differences in protein conformation between samples. In score scatter plot

(Figure 3(a)) it is possible observe that the wild-type (WT) samples are located mainly in negative PC3 and the stress (PA) samples are mainly in positive PC3 region. PCA loadings allow the comprehension of the distributional pattern of the samples (Figure 3(b)) and the identification of molecular alterations that might be present. PA samples are characterized by spectroscopic signals located at 1683, 1645, 1628 and 1622 cm⁻¹ and WT samples are characterized by the peak at 1654 cm⁻¹. To understand this discrimination, it is essential to assign these peaks to the corresponding biological contributor. In accordance to literature, the peak at 1654 cm⁻¹, here related to WT strains, is associated to α-helix structures in proteins [23] [24] [25] [26] , which are usually present in more soluble structures and so less prone to aggregate. Conversely, peaks at 1683 cm⁻¹ and 1628 cm⁻¹, associated with mutant strains, are related to β-sheet structures in fungal proteins [26] [27] . It is important to notice that β-sheet structures are usually related with less soluble and more prone to aggregate proteins, so these peaks could be directly related with the presence of PA in mutant strains, corroborating microscopy results. Also, the peak at 1622 cm⁻¹ is assigned to a shift in amide I bands, commonly related with the presence of protein aggregates [13] [26] [28] . The peak at 1645 cm⁻¹ is related to β-turns structures in amide I band of proteins [23] [26] [29] [30] [31] .

A PLS analysis was also applied to the 2^nd derivative spectra (1700 - 1500 cm⁻¹ spectral region) to explore the correlation between the spectra profile and percentage of protein aggregation evaluated by fluorescence microscopy in the yeasts samples. A calibration model with two latent variables was built (y = 0.5385154x + 11.932844) with a R² of 0.5385148, a RMSEC of 16.96% and of RMSEP of 20.6 % (Figure 4(a)). The PLS B-coefficient built with factor 2 showed that 1685, 1645, 1628 and 1622 cm⁻¹ are positively correlated with the samples with higher amounts of protein aggregation (Figure 4(b)). The relation between the PA samples and these spectroscopic signals, related with the presence of β-sheet structures and protein aggregates, corroborate the PCA results confirming that it is possible to evaluate the presence of protein aggregation by FTIR spectroscopy.

One of the drawbacks of this study is the fact that we did not determine the exact number of cells placed in ATR crystal. We used a 10 µL swab for all samples, however it is impossible to be completely certain that the number of cells was the same in each measurement.

These data confirm that FTIR has the potential to identify changes in protein conformation, possibly related with the presence of protein aggregates in a eukaryotic model. The fact that FTIR is a fast, non-destructive, simple and inexpensive metabolomic analysis makes it of particular interest to the study of biological samples. In fact, in the future, we intend to apply FTIR to study mice tissue samples and to determine the pattern of protein aggregation during aging in mammals and ultimately identify new aging biomarkers.

The Authors declare that there is no conflict of interest.

This work was supported by iBiMED - Institute of Biomedicine, University of Aveiro and the Fundação para a Ciência e Tecnologia of the Ministry of Education and Science of Portugal [grants number UID/BIM/04501/2014; BI/UI98/ 7395/2016 to S.M. and SFRH/BD/86866/2012 to J.T.].

Magalhães, S., Graça, A., Tavares, J., Santos, M.A.S., Delgadillo, I. and Nunes, A. (2018) Saccharomyces cerevisiae as a Model to Confirm the Ability of FTIR to Evaluate the Presence of Protein Aggregates. Spectral Analysis Reviews, 6, 1-11. https://doi.org/10.4236/sar.2018.61001