Sample preparation for FTIR transflection spectroscopy of dried samples

The collection of organs for this study was harvested from an unrelated project utilising rats and the entire study was specifically approved by the Monash University, ‘School of Biomedical Sciences Animal Ethics Committee A’. The animal care and experimental procedures were carried out in accordance with the Australian ‘Code of Practice for the Care and Use of Animals for Scientific Purposes’. These animals were sacrificed by 100mg/kg pentobarbitone intraperitoneal injection and organs collected immediately afterwards.

At postmortem, liver and heart were immediately excised from adult rats, trimmed of connective tissue and sliced into pieces of approximately 1 mm in thickness. The tissue pieces were immediately transferred into isotonic saline solution. For samples undergoing desiccation, alcohol and formalin fixation, fresh-frozen sectioning was required. To do this, randomly selected pieces were mounted with OCT (optimal cutting temperature) medium onto a cryotome (Leica model CM 3050S, sectioning temperature −25°C) and sectioned. In order to obtain good transflection FTIR spectra of tissue samples, the tissue was sectioned to 5 μm, a level of thickness that allows the infrared light to penetrate. To ensure a high level of homogeneity, sections taken from both liver and heart tissue were taken consecutively and immediately mounted onto Ag/SnO₂ coated infrared reflective “Mirr-IR” slides (Kevley Technologies). The sample preparation varied according to one of the following three methods:

A) Desiccator dehydrated tissue. The fresh tissue pieces were taken from the saline solution, immediately sectioned as described above, mountedon“Mir-IR”slides and placed in a desiccator loaded with silica gel for 48 hours.

B) Alcohol dehydrated tissue. The fresh tissue pieces were immersed in absolute alcohol for10 minutes then sectioned as above and allowed to air dry on the “Mirr-IR” slides.

C) Formalin dehydrated tissue. The tissue pieces were stored in formalin for 30 days prior to being sectioned and mounted on “Mirr-IR” slides as described above and allowed to air dry.

Sample preparation for ATR-FTIR spectroscopy of hydrated tissues

D) Fresh / hydrated tissue. ATR-FTIR spectra were obtained directly from freshly excised fully hydrated tissue by positioning each tissue piece flush against the sampling ATR crystal.

FTIR micro-spectroscopy of dried tissue sections

Spectra were collected with an Agilent “Stingray” FTIR micro-spectroscopy system which combines a 7000 series rapid-scan spectrometer, a 600 UMA microscope and a liquid nitrogen cooled mercury-cadmium-telluride 64 x 64 pixel focal plane array (FPA) detector. For each tissue section three spectral maps of 32 x 32 pixels were obtained from different locations within the section. (4 pixel aggregation and 64 co-added scans). Background spectra were collected from a region free of tissue of the FTIR-reflective slide.

ATR-FTIR spectroscopy of fresh hydrated tissue sections

ATR-FTIR spectra of the hydrated tissue pieces were recorded on a Bruker IFS Equinox FTIR system paired with a Golden Gate single bounce diamond micro-ATR accessory. All FTIR spectra were recorded over the range 4000–800 cm⁻¹ at 6 cm⁻¹ spectral resolution and with a zero fill factor of 2.

Protein bands

The spectral region 1800–1500 cm⁻¹ (Fig. 2(B) and Fig. 4(B)) corresponds primarily to absorptions attributable to proteins. The principle protein marker band is the amide I absorption centered around 1654 cm⁻¹. The amide I band position and shape is sensitive to the secondary structures of proteins [28] but is also particularly susceptible to distortions from light scattering effects and hence, it is essential that such distortions are removed to avoid misinterpretation of the protein secondary structure information [26]. In the second derivative spectra this band was resolved into a set of 3 amide I peaks at ∼1680, 1654 and 1634 cm⁻¹. In the hydrated tissue sample’s ATR-FTIR spectrum the1634 cm⁻¹band appeared strong and was shifted to ∼1631 cm⁻¹: this is due to the presence of the underlying OH bending mode of water at ∼1633 cm⁻¹. In comparing the amide I band in the second derivative spectra of the heart and liver tissues, marked differences are apparent in the relative strengths of the 3 underlying band features. This is due to the different overall protein composition of the two tissue types. For both tissue types and for all sample preparation methods the central and strongest peak (∼1654 cm⁻¹) is associated with both α-helical secondary structure and random disordered structure [29,30]. The position and intensity of this peak was largely unchanged except for in the formalin fixed liver tissue where the 1654 cm⁻¹ absorption appeared weaker in intensity. An amide I absorption associated with β-sheet protein secondary structure was apparent in the ethanol fixed and desiccated heart tissues at ∼1634 cm⁻¹but was blue shifted to 1638 cm⁻¹ in the formalin fixed heart tissue samples. Conversely, this same β-sheet band was observed at ∼1638 cm⁻¹ in the ethanol fixed and desiccated liver tissues but was red shifted to 1631 cm⁻¹ in the formalin fixed liver tissues. The highest wavenumber value component peak of the amide I group was the absorption at 1681 cm⁻¹ associated with β-turn secondary structure. For both tissue types this band changed only slightly between samples dried by desiccation or fixed with formalin but for the samples prepared using water substitution by ethanol there was a significant increase in the intensity of this peak for both tissue types and for the heart tissue a shift to a lower frequency (1671 cm⁻¹) was also observed.

Considering these observed changes in the underlying features of the amide I peak together, and considering the varied mixture of protein content present in these tissue samples, it is not possible from this work alone to unambiguously determine what changes in protein secondary structure or protein folding or unfolding events may have occurred as a result of the various applied sample preparation treatments. However, changes in the profile and intensity of the amide I band were clearly observed for both the formalin fixed and ethanol substitution dried samples compared with the desiccator dried samples and the ATR-FTIR spectra obtained from hydrated samples. The changes in the amide I band profile reflect significant differences in hydrogen-bonding strengths and differences in transition dipole coupling as these are the factors that give the amide I band its sensitivity to protein secondary conformational changes. Both the formalin fixation and the ethanol dehydration methods altered the amide I band profile and thus affected its usefulness for reliable determination of percentages of secondary structures present in the tissues. These methods are indicating that they do not reflect the true living native hydrated state of the proteins. A significant reduction in the intensity of the amide I peak was observed with formalin fixation. This has previously been reported by Faolain and colleagues and attributed to a loss of secondary amide upon formation of tertiary amides through the cross linking action of formalin [27]. In the same work, a shift of the amide I peak following tissue fixation with formalin was also reported as was observed in this work.

Mie scatter induced red shifts of these bands may have been a difficulty that was encountered in earlier studies. At that time the spectrum distorting influence of scattering phenomena was not understood fully and algorithmic corrections were not available. In the current work, scattering effects have been removed prior to the analysis and interpretation of the spectra. A shift of the amide I and amide II bands to lower wavenumber values after ethanol fixation for preservation of bone that was previously reported by Pleshko et al [31] and cited in later works [32] but it was not observed in our work. Pleshko’s assignment to ethanol induced hydrogen-bonding changes and/or protein secondary conformational changes is therefore not of general applicability to all protein mixtures and may be a mistaken attribution of Mie scatter induced spectral changes. In addition, they reported that formalin fixation did not alter the amide I band profile, contrary to our observations.

Lipid bands

The spectral region 3100–2800 cm⁻¹ (Fig. 2(A) and Fig. 4(A)) corresponds to lipids.The 3012 cm⁻¹ band assigned to olefins or unsaturated fatty acids [33,34] were observed in the FTIR-ATR spectra obtained from both types of hydrated tissues. In both tissue types this band was absent following ethanol dehydration. In the heart tissue this band was detected following sample preparation by both formalin fixation and by desiccator drying. But in the liver tissue this band was absent following formalin fixation. Similarly, the bands at 2924 and 2853 cm⁻¹ assigned to ν_asCH₂ and ν_sCH₂ of lipids, respectively, were greatly decreased in intensity or absent for both tissue types when ethanol was used in the sample preparation. Presumably, this indicates complete extraction of lipids as a result of the ethanol water substitution treatment. It was also observed that these bands were slightly red shifted (∼2 cm⁻¹) when formalin fixation had been used. The two bands at 2960 and 2874 cm⁻¹ were relatively unchanged in both tissue types and for all sample preparation methods. These ν_asCH₃ and ν_sCH₃ absorptions are often assigned to lipids and also methyl groups contained in nucleic acids and proteins. Given that the CH₂ bands were absent in the ethanol treated samples due to lipid extraction, these bands must be predominantly due to methyl groups contained in either proteins or nucleic acids or both and are therefore relatively insensitive to lipid content, at least in these samples.

Mixed contribution region—carbohydrates, proteins, nucleic acids

The spectral region 1475–950 cm⁻¹ (Fig. 2(C) and Fig. 4(C)) corresponds to the bands from several macromolecules: carbohydrates, proteins and nucleic acids. The second derivative spectra in the region 1390–1475 cm⁻¹ exhibited a complex mixture of band features for both tissue types. These absorptions are predominately due to bending modes of CH₂ and CH₃ groups. For both tissue types the desiccated and the ethanol treated samples demonstrate absorptions that were remarkably similar suggesting that this region is also not particularly sensitive to lipid content. Comparing the hydrated ATR spectra and the formalin fixed tissue spectra with the other two dried tissue preparations, significant wave number shifts and splitting of the CH₂ and CH₃ band features were observed (Table 1 and Table 2). In the case of the formalin fixed tissues, this can be attributed to the cross-linking action of the fixative on proteins. The spectra of the formalin fixed tissues also exhibited band shifts for many of the peaks in the spectral region from approximately 1330–1150 cm⁻¹ which can be associated with amide III modes and in particular with the spectrum of collagen.

In the heart tissue spectra two bands characteristic of collagen at 1205 and 1285 cm⁻¹ were observable in the wet tissue spectra but only weakly detected in the formalin fixed and ethanol dehydrated tissue spectra. These collagen marker bands were apparently absent from the desiccated sample spectra. Bands in the region 1160–950 cm⁻¹ are largely due to stretching vibrations of CO groups in carbohydrates although CO in the ribose backbone of nucleic acids and the ν_sPO₂^- (symmetric phosphate stretching) modes also contribute absorbance at around 1080 cm⁻¹. In the desiccated and ethanol treated liver tissue samples the triplet of bands due to glycogen (1154, 1080, 1035, 1022 cm⁻¹) appeared strong in the spectra but were weak in the wet spectra and altered to a broad band feature in the formalin fixed tissues. Establishment of glycogen levels is important for the diagnosis of many diseases [35,36]. Clearly the spectra indicate that formalin fixation has some effect on glycogen chemistry in tissues. In this regard, there are conflicting reports in the literature in relation to the effects of formalin on tissue glycogen. Formalin has been used as a method for assay extraction of glycogen from tissues [37], whereas others have reported that formalin is effective for localizing and preserving glycogen content in tissues [38]. Our findings support the latter; there was no evidence that formalin removes glycogen from either the heart or liver tissue.