Food Research International 66 (2014) 123–131

Contents lists available at ScienceDirect

Food Research International

j ourna l homepage: www.e lsev ie r .com/ locate / foodres

Raman spectroscopic study of effect of the cooking temperature and timeon meat proteins

Daniel T. Berhe a, Søren B. Engelsen a, Marchen S. Hviid b, René Lametsch a,⁎a Department of Food Science, Faculty of Science, University of Copenhagen, Rolighedsvej 30, DK-1958 Frederiksberg C, Denmarkb Danish Meat Research Institute, Gregersensvej 9, DK-2630 Taastrup, Denmark

⁎ Corresponding author. Tel.: +45 35333483; fax: +45E-mail address: rla@life.ku.dk (R. Lametsch).

http://dx.doi.org/10.1016/j.foodres.2014.09.0100963-9969/© 2014 Elsevier Ltd. All rights reserved.

a b s t r a c t

a r t i c l e i n f o

Article history:Received 18 July 2014Accepted 8 September 2014Available online 16 September 2014

Chemical compounds studied in this article:Tyrosine (PubChem CID: 6057)Tryptophan (PubChem CID: 6305)

Keywords:Protein denaturationRaman spectraThermal treatmentChemometrics

Changes in protein secondary structure during thermal treatment ofwholemeatwere studied using Raman spec-troscopy. Pork longissimus thoracis was heat treated at 50 to 70 °C for 2–10 h and 110 average Raman spectrawere collected from all the combinations. The effect of cooking temperature onmeat protein structurewas highlypronounced compared to cooking time. Detailed information about the changes in protein structure affected bycooking temperature and timewas revealed and Raman spectra from different cooking temperatureswere read-ily discriminated by principal component analysis. Temperature had a significant effect on the intensity ratio oftyrosine (Tyr) and tryptophan (Trp) and on cooking loss. Good correlationswere found between theRaman spec-tra and cooking temperature (R2 = 0.96), cooking loss (R2 = 0.82) and cooking time (R2 = 0.78). Raman spec-troscopy proved to be a useful technique to follow the effect of cooking temperature and time onmeat proteins inintact muscle.

© 2014 Elsevier Ltd. All rights reserved.

1. Introduction

Heat treatment ofmeat andmeat products leads to structural chang-es in meat proteins. The structural changes inmyofibrillar proteins, sar-coplasmic proteins and connective tissues will therefore have an effecton tenderness and/or texture and other quality parameters of a product(Liu, Zhao, Xiong, Xie, & Qin, 2008; Martens, Stabursvik, & Martens,1982; Shao, Zou, Xu, Wu, & Zhou, 2011; Xu, Han, Fei, & Zhou, 2011).Some of the changes are transversal and longitudinal shrinkage of themuscle fibers, aggregation and gel formation of the sarcoplasmic pro-teins and shrinkage and solubilization of the connective tissues(Christensen, Bertram, Aaslyng, & Christensen, 2011; Christensen,Ertbjerg, Aaslyng, & Christensen, 2011; Palka, 1999; Tornberg, 2005).The secondary structure of proteins which is stabilized, for example,by hydrogen bonding and hydrophobic interactions, is commonly af-fected by heat treatment (Liu et al., 2008; Tornberg, 2005) and also bypH (Liu et al., 2008). Therefore, the effect of heat treatment on the sec-ondary structure and consequently on themicroenvironment is respon-sible for changes in functional properties of meat proteins duringprocessing (Herrero, Carmona, & Careche, 2004; Lan et al., 1995; Palka& Daun, 1999; Xu et al., 2011). Martens et al. (1982) reported that firm-ness, the force necessary to compress themeat strip about 30% with themolar teeth, was dramatically increased in temperature ranges where

35283341.

denaturation of myofibrillar proteins took place. The authors (Martenset al., 1982) suggested that this relationship could be due to increasedaggregation of proteins after thermal denaturation of myofibrillar pro-teins. During heat-induced gelation of extracted pork myofibrillar pro-teins, the unfolding of the secondary structure and the exposure ofburied hydrophobic amino acids were observed (Xu et al., 2011). Suchchanges showed a positive relationship with the increased hardness ofthe gel which might be due to increased conformational changes andhydrophobic interactions as a function of heating temperature (Xuet al., 2011). Myofibrillar proteins, accounting for 50 to 55% of thetotal protein in muscle (Tornberg, 2005), are therefore governing con-stituents for the quality of heat treated products.

Cooking temperature and time are the main determinant factorswhich need attention when dealing with the heat treatment of meat.The effect of these two factors on the structure of meat proteins(Christensen, Bertram, et al., 2011; Christensen, Ertbjerg, et al., 2011)and in turn on product quality (Christensen, Bertram, et al., 2011;Christensen, Ertbjerg, et al., 2011; Martens et al., 1982; Mortensen,Frøst, Skibsted, & Risbo, 2012) has been studied intensively. Accordingto studies using differential scanning calorimetry (DSC), there arethree maximum transition temperatures: 56.5, 65.8 and 79.4 °C whichcould be ascribed to myosin, sarcoplasmic proteins and connective tis-sues and actin in pork muscle, respectively (Xiong, Brekke, & Leung,1987). As mentioned above, conformational changes in myofibrillarproteins, sarcoplasmic proteins and connective tissues determine prop-erties of the final product. It is, therefore, needed to look for fast and

124 D.T. Berhe et al. / Food Research International 66 (2014) 123–131

non-destructivemethodswhich can be used tomonitor thefinal qualityof a product.

Raman spectroscopy is a useful technique for monitoring changes inprotein structure and provides information at the molecular level. Itmeasures molecular vibrations but instead of measuring absorptionsuch as in infrared (IR) spectroscopy, Raman spectroscopy measuresscattering. The principle of Raman spectroscopy is that a sample is ir-radiated with a strong monochromatic laser light which interacts witha molecule in the sample. Most of the photons are scattered from thesample with the same energy level as the incident light (Rayleigh scat-tering) whereas very few photons are scattered with a lower frequencycompared with the frequency of the incident light (Stokes scattering).The difference in frequency of the incident light and inelasticallyscattered light is called the Raman shift (Δν in cm−1) or simply justthe Raman spectrum. This spectrum contains information about thefundamental molecular vibrations as is also found in the IR spectrum.Near infrared (NIR) is another spectroscopic technique. It measuresovertones and the combination tones of the molecular vibrationswhereas IR and Raman spectroscopy measures the fundamentals. Themain advantage of Raman spectroscopy over NIR is that Raman spec-troscopy has a higher selectivity and that the spectra are not dominatedby molecular vibrations of water which dominate the infrared and NIRspectra of many food samples (Ellekjaer & Isaksson, 1992). In contrast,Raman spectroscopy has the disadvantage of being sensitive to samplefluorescence.

In the past few years, Raman spectroscopy has been used to predictthe quality of meat and meat products (Herrero, Carmona, Cofrades, &Jimenez-Colmenero, 2008; Pedersen, Morel, Andersen, & Engelsen,2003; Schmidt, Scheier, & Hopkins, 2013; Wang, Lonergan, & Yu,2012). Beattie, Bell, Farmer, Moss, and Patterson (2004) found a goodcorrelation between Raman spectra and consumer-perceived qualityparameters of roasted beef. Another promising result has also been re-ported in sheep meat with a correlation (R2) of 0.86 and 0.83 betweenRaman spectra recorded using a handheld instrument and shearforce and cooking loss, respectively (Schmidt et al., 2013). Wang et al.(2012) used Raman spectra to predict the sensory quality of pork loin(cooked at 70 °C) and found that prediction accuracy for tendernessand juiciness was significantly higher than that for chewiness. Theauthors (Wang et al., 2012) suggested that this apparent discrepancyin prediction accuracy could be due to the fact that Raman spectra ofmeat may be more closely related to protein composition and/orstructures of myofibrillar components than that of connectivetissues.

The main objective of the present study was to investigate the effectof cooking temperature and time on the meat proteins' structure in in-tact muscle using Raman spectroscopy. The second objective of thisstudy was to predict cooking temperature and time using Raman spec-tra collected from cooked then cooled pork meat.

2. Material and methods

2.1. Meat sample

Four female pigs with a slaughter weight of 74.3 to 77 kg werestunned using CO2 and further dressing was completed within30 min post-mortem (PM) in a slaughter house (UCR, Roskilde,Denmark). The carcasses were chill-stored at 4 °C for 24 h PM.Longissimus thoracis was excised from both right and left sides ofthe pigs at the 24 h PM and was transported to the University of Co-penhagen for laboratory analysis. The pH of the muscle was 5.52–5.58. Skin, back-fat and visible connective tissues were removedfrom the muscle. Sampling was made between the 9th and 13thribs from both sides of the pigs and the sampled area was cut intochops (4–5 cm length). Each chop was vacuum packed separatelyand stored at −20 °C until use.

2.2. Heat treatment and sample preparation

The frozen chopswere thawed at 4 °C for 15h in their vacuumpackedplastic bags. The thawed chops were removed from the plastic bags andcut into small pieces of samples (2 × 2 × 3 cm3, height × width × length,respectively). The length of the samples was cut along the direction ofmuscle fibers. Weight of each sample was recorded then each samplewas re-vacuum packed and heat treated in a preheated water bath (ICC“Roner”, Frinox Aps, Hillerød, Denmark) at different cooking tempera-tures with an interval of 2 °C (50, 52, 54 … 68 and 70 °C) for differentcooking times with an interval of 2 h (2, 4, 6, 8 and 10 h) in each cookingtemperature. This design gave 55 treatments from all the combinations(11-temperature point × 5-time point). Each treatment was in duplicate(55 × 2 = 110 samples). Therefore, ten pieces of meat samples wereplaced at a time in a water bath in each temperature point. Then twosamples (the duplicate) were removed at each time point and werecooled in running cold tap water for 10min tominimize further cooking.Heat treatedmeat samples were removed from the vacuum packed plas-tic bags and dried using paper tissue. The final weight was recorded tocalculate cooking loss. Finally, each cooked sample was cut into slices(5 mm-thickness) across the fiber direction and the slices were readyfor Raman spectroscopy measurement.

Cooking losswas calculated byweighing the sample before and afterheat treatment, as mentioned above, and expressed as a percentage ofthe original weight.

2.3. Differential scanning calorimetric (DSC) analysis

A raw pork meat sample (32.4 mg) was placed in a sample cell of adifferential scanning calorimeter (DSC 1) (STARe System, Mettler Tole-do, Schwerzenbach, Switzerland) to follow themajor endothermic tran-sitions in the pork muscle. An empty sealed aluminium crucible wasplaced in the reference cell. The sample was scanned from 30 °C to90 °C with a heating rate of 2 °C/min.

2.4. Raman spectroscopic measurement

Ramanmeasurementsweremade on thenew cut surface of themeatslices using the RamanRxn1 instrument (Kaiser Optical Systems, Inc.,Michigan, 132 USA). The slices were placed on ametal plate under ami-croscope with a 10× objective to collect the Raman scattering from thesamples. The instrument was equipped with a 785 nm laser (Invictus,Kaiser Optical Systems Inc., Michigan, USA), a single holographic gratingand a thermoelectric cooled charge-coupled device detector, operated at−40 °C. The laser was focused on each sample parallel to the musclefiber direction, and the laser power on the sample was 120 mW. Thespectra were acquired using an average spectrum of 2 scans each withan exposure time of 10 s and were stored as Raman shifts in the rangeof 1800–200 cm−1. Four spectra were collected by making grids with adistance of 4 mm on the surface of each slice and the average of the 4spectra was used for data analysis (110 samples × the average of fourspectra per sample = 110 average spectra). Occasional Raman spectraof high fat content were removed during the measurement in order tofocus on the structural changes of proteins.

2.5. Chemometrics

Raman spectra, cooking loss data, cooking temperature and cookingtimewere imported intoMatLab software (MATLAB2014a) (Mathworks,Massachusetts, USA) for data processing. The MatLab and Partial LeastSquares (PLS) Toolbox (Version 7.5 Eigenvector Research, Inc.;Wenatch-ee, WA, USA) were used to perform principal component analysis (PCA)(Wold, Esbensen, & Geladi, 1987) and partial least squares (PLS) regres-sion (Wold, Martens, & Wold, 1983) on the dataset. The Raman spectrawere preprocessed using extended multiplicative signal correction(EMSC) (Martens, Nielsen, & Engelsen, 2003) and then mean centered.

125D.T. Berhe et al. / Food Research International 66 (2014) 123–131

PCA was performed in order to retrieve information about systematicspectral variations in relation to the treatments. The PLS regression wasperformed to evaluate the relationship between Raman spectra andcooking loss, cooking temperature and cooking time. The developedPLS regression models were cross-validated using a contiguous block inorder to evaluate their performance. Analysis of variance (ANOVA) onthe relative intensity ratio of tyrosine (Tyr) (I856 cm

−1/I827 cm−1) and

tryptophan (Trp) (I758 cm−1/I1450 cm

−1) from the preprocessed Ramanspectra as well as the analysis of the cooking loss data were performedusing the SAS software (SAS version 9.4) following the Split-plotANOVA procedure. Datawas analyzedwith thewhole plot being cookingtemperature and the sub-plot treatment being cooking time. The randomeffect included replicates × cooking temperature.When F-testswere sig-nificant, means were compared using least significant differences (LSD).

3. Results

3.1. Differential scanning calorimetry

Fig. 1 shows a representative DSC thermogram of raw whole porkmeat scanned from 30 °C to 90 °C with a heating rate of 2 °C/min.Three endothermic transitions namely at 53.4 °C, 61.6 °C and 76.1 °Cwere observed (Fig. 1). These three endothermic transition regions arein agreement with the literature (Xiong et al., 1987). These transitionregions have beenwell known and are ascribed tomyosin, sarcoplasmicproteins and connective tissues and actin, respectively (Tornberg, 2005;Xiong et al., 1987).

3.2. Raw Raman spectra

In this study, a relatively small sample dimension was used in orderto achieve quick temperature equilibrium and obtain the informationabout the denaturation of meat proteins. The raw Raman spectra ofsamples cooked at different cooking temperatures (50–70 °C) for 4 hare presented in Fig. 2. The spectra are colored according to the cookingtemperature in order to uncover any systematic difference in the rawspectra due to the cooking temperature. The spectra from samplescooked at 50 and 52 °C have high background fluorescence whicheven grows higher at 54, 56 and 58 °C then decreased as the cooking

Fig. 1. Differential scanning calorimetry (DSC) therm

temperature was increased (Fig. 2). The same trend was observed forall the other cooking times.

3.3. Effect of cooking temperature and time-preprocessed Raman spectra ofpork muscle

Preprocessed Raman spectra from the samples cooked at differentcooking temperatures (50–70 °C) for 4 h are presented in Fig. 3A.Table 1 shows the assignments of Raman bands, indicated with num-bers in the figure (Fig. 3A). The spectra show a gradual decrease inintensity in the regions of amide-I (1655–1650 cm−1), CH-bending(1340 cm−1) and C\C stretch (940–934 cm−1) as the cooking temper-ature increases from 50 °C to 70 °C (Fig. 3B–D). On the contrary, a grad-ual increase in intensity at 1668–1663 cm−1 (amide-I = N β-sheet andrandom coil) and 1243–1237 cm−1 (amide-III = N random coil and β-sheet) are observed as the cooking temperature increases (Fig. 3B–C).Fig. 3B also shows that samples cooked below 60 °C had wider Ramanpeaks with lower intensity in the amide-I region whereas samplescooked especially at the maximum temperatures (68 and 70 °C) had anarrow peak in the amide-I region.

Furthermore, there was a decrease in intensity in the S\S stretchingband as the cooking temperature increased from 50 to 70 °C (Fig. 3E).

3.4. Changes in local microenvironment

Both cooking temperature and cooking time proved to have a signif-icant effect on the intensity ratio of the Tyr doublet (I856 cm

−1/I827 cm

−1) (Fig. 4 left and right, respectively). The intensity ratio wassignificantly higher in samples cooked at 54 °C than at all the othercooking temperatures (Fig. 4 left). The intensity ratio increased signifi-cantly as the cooking time increased (Fig. 4 right). The higher standarddeviation in Fig. 4 (right) was due to the high effect of cooking temper-ature in each cooking time. Therewas no significant interaction effect ofcooking temperature and cooking time on the intensity ratio of the Tyrdoublet (Fig. not shown). The intensity ratio of Trp was calculated bydividing its intensity to the intensity of the CH2 bending band at1450 cm−1 as the latter band has been expected not to change duringconformational changes (Herrero et al., 2004). There was a significanteffect of cooking temperature on the intensity ratio of Trp (I758 cm

−1/

ogram of longissimus thoracis of pork muscle.

Fig. 2. Raw Raman spectra (1800–400 cm−1) from longissimus thoracis of pork cooked at different cooking temperatures (50–70 °C) for 4 h. Colored according to cooking temperature(n = 11). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Fig. 3. Raman spectra preprocessed using extended multiplicative signal correction (EMSC) from longissimus thoracis of pork cooked at different cooking temperatures (50 °C–70 °C) for4 h. Colored according to cooking temperature. Selected regions: 1800–400 cm−1 (A), 1680–1630 cm−1 (B), 1350–1220 cm−1 (C), 960–920 cm−1(D) and 580–500 cm−1 (E). See Table 1for the assignment of the labeled bands (n = 11). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

126 D.T. Berhe et al. / Food Research International 66 (2014) 123–131

Table 1Assignment of themost common bands in Raman spectra ofmeat. The band numbers cor-respond to bands in Fig. 2A.a

Band nr. Band position (cm−1) Assignment

1 1685–1645 Amide-I (β-sheet, random coil and α-helix)2 1606 Tyrosine, phenylalanine, tryptophan3 1554 Tryptophan4 1450 CH3, CH2, CH− all bending5 1341 CH− bending6 1309–1229 Amide-III (α-helix, random coil and β-sheet)7 1003 Phenylalanine8 940 C\C stretching (α-helix)9 850 Tyrosine10 830 Tyrosine11 760 Tryptophan12 560–510 S\S

a Adopted from: Chen & Lord, 1974; Herrero, 2008; Wei et al., 2008.

127D.T. Berhe et al. / Food Research International 66 (2014) 123–131

I1450 cm−1) (Fig. 5 left). The intensity ratio of Trp was significantly

higher at 56 °C and 54 °C (Fig. 5). Cooking time also had a significant ef-fect on the intensity ratio of Trp (Fig. 5 right). There was no interactioneffect of cooking temperature and time on the intensity ratio of Trp.

3.5. Discrimination of Raman spectra frommeat samples— PCA calculation

3.5.1. Effect of cooking temperatureFig. 6A shows the PCA-score plot for all the Raman spectra colored

according to the cooking temperature. Two outliers were removedbased on their high values in Hotelling T2. The first principal component(PC-1) was able to show the general stepwise structural changes inthe meat samples cooked at 50 to 70 °C (Fig. 6A). The loadings plot forPC-1 (Fig. 6B) reveals that the samples cooked above 60 °C were posi-tively correlated to the high intensity of bands at the amide-I(1668 cm−1 = N β-sheet) and amide-III (1238 cm−1 = N β-sheet) re-gions. In contrast, a negative correlation was found for the bands fromCH bending (1340 cm−1), C\C stretch (937 cm−1 = N α-helix), Tyr(856 and 827 cm−1), Trp (around 757 cm−1) and S\S stretching(530 cm−1). PC-2 scores discriminate the samples cooked at 50–52 °Cfrom all the other samples in general and from samples cooked at 54–58 °C in particular (Fig. 6A). The loadings plot for PC-2 showsthat there was a negative contribution from bands at 1642 cm−1,1340 cm−1, 934 cm−1 and 521 cm−1, and a positive contributionfrom bands at 1668, 1236, 1010 cm−1 (Fig. 6C).

3.5.2. Effect of cooking timePCAs were calculated separately for each cooking temperature in

order to investigate the effect of cooking time (2–10 h) at each cookingtemperature. In general, these analyses showed no clear discrimination

Fig. 4.Mean relative intensity ratio of tyrosine (I856 cm−1/I827 cm−1) of longissimus thoracis of po

(right) (n = 110). Different letters indicate significant differences (P b 0.05). Bars represent st

between the different cooking times (Fig. not shown). However, sam-ples cooked for the different cooking times either at 62 or 64 °C showeda tendency to be discriminated in their respective PCA-score plot. ThePC-1 vs PC-2 score plot (Fig. 7 left) shows that there is a tendency fora discrimination of samples cooked below and above 6 h at the cookingtemperature of 62 °C. The corresponding loadings plot for PC-1 showsthat samples cooked for more than 6 h at 62 °C were negatively corre-lated to a high intensity of bands at 1645 and 1332 and a region assignedto S\S stretching (537 and 514 cm−1) (Fig. 7 right). Although the peakwas not well defined, there was also a contribution of the C\C stretch(934 cm−1 = N α-helix) (Fig. 7 right).

3.6. Prediction of cooking loss, cooking temperature and time

Both cooking temperature and time had a significant effect oncooking loss which is shown in Fig. 8. There was no significant interac-tion effect of cooking temperature and cooking time on cooking loss(Fig. not shown). Cooking loss increased significantly as cooking tem-perature and/or time increase (Fig. 8 left and right, respectively). PLSregression revealed that cooking loss can be predicted from theRaman spectra using 2 latent variables with a very good fit: the squaredcorrelation coefficient (R2) of 0.82 (Table 2). The rootmean square errorof cross validation (RMSECV) was 2.82% (Table 2). By inspection of theregression coefficients, a positive contribution mainly from β-sheet(1668 and 1238 cm−1) and a negative contribution from α-helix(1643 and 936 cm−1) were revealed (Fig. not shown). Also, a negativecontribution from the Tyr (around 856 and 827 cm−1), Trp (around756 cm−1) and S\S stretch (530 cm−1) was revealed (Fig. not shown).

Prediction of cooking temperature using the Raman spectra showedvery high correlation (R2 = 0.96) with a prediction error (RMSECV) of1.26 °C (Table 2). Four latent variables were required to predict thecooking temperature. In contrast, the PLS prediction of cooking time re-sulted in a poorermodel (R2= 0.78)which required six latent variables(Table 2). Similar bands (around 1668, 1644, 1340, 1237, 937, 856, 827,756 and 530 cm−1) in LV-1 which contributed to the prediction ofcooking loss were also the main contributors for the prediction ofcooking temperature and time (Fig. not shown). Apart from this, therewas a clear negative contribution from: 939, 855, 827, 756, and536 cm−1 in LV-2 for the prediction of cooking time which was notthe case in the models for the prediction of cooking loss and cookingtemperature.

4. Discussion

In this study, Raman spectroscopy was used to elucidate conforma-tional changes in meat proteins during cooking. The higher backgroundintensity in samples cooked at 54, 56 and 58 °C might be due to the

rk cooked at different cooking temperatures (left) (n= 110) and different cooking timesandard deviation of means.

Fig. 5.Mean relative intensity ratio of tryptophan (I758cm−1/I1450 cm−1) of longissimus thoracis of pork cooked at different cooking temperatures (left) (n = 110) and different cooking

times (right) (n = 110). Different letters indicate significant differences (P b 0.05). Bars represent standard deviation of means.

128 D.T. Berhe et al. / Food Research International 66 (2014) 123–131

effect of the dynamic process of denaturation of myosin in this range oftemperature (Fig. 2). The unfolding ofmyosinmolecules in turn leads toexposure of aromatic hydrophobic residues primarily Trp, phenylala-nine (Phe) and Tyr (Boyer, Joandel, Ouali, & Culioli, 1996), which areknown to cause fluorescence (Chan, Gill, & Paulson, 1992). This is

Fig. 6. Principal component analysis (PCA) of Raman spectra (1800–400 cm−1) from longissimcooking times (2–10 h). PC-score plot (PC-1 vs PC-2) colored according to cooking temperatuthe variation in the Raman spectra (n = 108). (For interpretation of the references to color in

confirmed in the present study where the intensity ratio of Tyr andTrp was significantly higher at 54 and 56 °C, respectively (Figs. 4 and5) (see below for clarification). The reduced background at the highertemperatures (above 60 °C) (Fig. 2) might be due to the involvementof the hydrophobic residues in the formation of protein aggregates. In

us thoracis of pork cooked at different cooking temperatures (50 °C–70 °C) for differentre (A) and PC-loadings plot for PC-1 (B) and PC-2 (C). PC-1 and PC-2 explained 87.25% ofthis figure legend, the reader is referred to the web version of this article.)

Fig. 7. Principal component analysis (PCA) of Raman spectra (1800–400 cm−1) from longissimus thoracis of pork cooked at 62 °C for different cooking times (2–10 h). PC-score plot (PC-1vs -2) colored according to cooking time (left) and PC-loadings plot for PC-1 (right). (n = 10).

129D.T. Berhe et al. / Food Research International 66 (2014) 123–131

a study on myosin solutions from fish, Chan et al. (1992) found that thefluorescence decreased as temperature increased during heat treatmentfrom 22 to 60 °C. Furthermore, this study showed that the remainingbackground fluorescence (Fig. 2) was best compensated for byemploying the EMSC preprocessing (Fig. 3A).

The preprocessed Raman spectra (Fig. 3A–E) reveal changes in fre-quency and intensity in the Raman bands which are useful for under-standing the denaturation process of meat proteins. The changes areprimarily indicators for changes in the secondary structure of the poly-peptide back bone, disulfide bonds and in turn the microenvironment ofthe amino acid side chains. The Raman bands were assigned based onthe literature (Chen & Lord, 1974; Herrero, 2008; Wei, Zhang, Halas, &Hartgerink, 2008). Model compounds/polymers have been used for theassignment of Raman bands (Chen & Lord, 1974; Wei et al., 2008) and ithas been observed that it is difficult to find the exact position of the con-formations especially in the amide-III regions (Chen & Lord, 1974). Thiscould be due to different factors such as the strength of the hydrogenbond. Chen and Lord (1974) made assignments of Raman bands in theamide-I region as follows: α-helix (1655–1650 cm−1), random-coil(around 1665 cm−1) and β-pleated sheet (around 1666 cm−1) andamide-III region as: α-helix (1300–1265 cm−1), random coil (1253–1243 cm−1) and β-sheet (1235–1229 cm−1) using known polymers.

Fig. 8.Mean cooking loss (%) of longissimus thoracis of pork cooked at different cooking temperdicate significant differences (P b 0.05). Bars represent standard deviation of means.

The same regions showed changes in position and/or intensity in thepresent study (Fig. 3B–C). Thewider Raman peakswith a lower intensi-ty in the amide-I region for samples cooked below 60 °C compared to anarrow peak for samples cooked at themaximum temperatures (68 and70 °C) can also indicate the complexity of the system. The changes inposition and/or intensity of the Raman band in this study are in goodagreement with other studies which found an increase in β-pleatedsheet and a decrease in α-helix in intact muscle (Beattie, Bell,Borggaard, & Moss, 2008), meat batters (Herrero, Carmona, Cofrades,et al., 2008; Herrero, Carmona, Lopez-Lopez, & Jimenez-Colmenero,2008; Shao et al., 2011) and extracted myofibrillar proteins (Xu et al.,2011) during heat treatment. As pointed out above, the primary causefor the changes in intensity and frequency of bands (below 60 °C)might be the denaturation of myosin where changes in its structurelead to shifting position of bands in the amide-I (1650 cm−1) andamide-III regions (Carew, Stanley, Seidel, & Gergely, 1983). Denatur-ation of connective tissues and sarcoplasmic proteins (Fig. 1) may alsohave contributed to the changes in frequency and intensity of theRaman bands as they denature mainly above 60 °C (Voutila, Mullen,Ruusunen, Troy, & Puolanne, 2007; Xiong et al., 1987). In relation tomeat quality, a higher amount of β-sheet has been positively correlatedto increased sensory toughness in beef (Beattie et al., 2004), and to

atures (left) (n = 110) and different cooking times (right) (n= 110). Different letters in-

Table 2Partial least squares (PLS) regressionmodels for predictions of cooking loss, cooking tem-perature and time based on Raman spectra from cooked pork longissimus thoracis(n = 110).

R2 RMSECV #LVs

Cooking loss (%) 0.82 2.82 2Cooking temperature (°C) 0.96 1.26 4Cooking time (h) 0.78 1.36 6

R2 = squared correlation coefficient; RMSECV = root mean squared error of cross-validation; #LVs = number of latent variables.

130 D.T. Berhe et al. / Food Research International 66 (2014) 123–131

higher penetration force values in cooked meat batters (Herrero,Carmona, Lopez-Lopez, et al., 2008).

Disulfide bonds are known to stabilize the tertiary structure of pro-teins. Raman bands at the S\S stretching vibrations of disulfide bondsregion: 508, 528 and 540 cm−1 are assigned to gauche–gauche–gauche,gauche–gauche–trans and trans–gauche–trans, respectively (Sugeta, Go,& Miyazawa, 1973; Van Wart & Scheraga, 1986). Xu et al. (2011)found that the intensity of the S\S stretching around 545 cm−1was de-creased as heating temperature was increased from 30 to 70 °C duringgel formation of extracted pork myofibrillar proteins. They ascribedthis to the conversion of the trans–gauche–trans conformations to theother conformations as the temperature increased. Therefore, the de-crease in intensity at 530 cm−1 (gauche–gauche–trans) and broadeningof the band could indicate the loss of the native form of the disulfidebonds in the present study.

The intensity ratio for Tyr doublet (I856 cm−1/I827 cm

−1) and for Trp(I758 cm

−1/I1450 cm−1) was calculated in order to investigate the change

in the local microenvironment during the thermal treatment. In thisstudy, the intensity ratio for Tyr was above 1 (Fig. 4) indicating thatthe Tyr residues were involved in hydrogen bonding by acting as bothhydrogen acceptor and donor in a polar environment (Wei et al.,2008; Xu et al., 2011). It was interesting to see the significantly higherintensity ratio of Tyr at 54 °C which was the temperature where an en-dothermic peak assigned to myosin was observed (Fig. 1). The decreasein the intensity ratio of Tyr and Trp as the temperature was increasedmight be due to the involvement of these residues in hydrophobic inter-actions after they have been exposed to the aqueous environment in thehigher temperature. Our results are in agreement with previous work(Xu et al., 2011) who found similar trends during heat-induced gel for-mation. The increase in intensity ratio of the Tyr doublet as a function oftime might be due to a positive effect of cooking time on the polarmicroenvironment.

PCA-score plot (Fig. 6A) revealed the systematic conformationalchanges in the meat proteins discussed above. In general, inspectingthe score and loadings plots (Fig. 6), PC-1 reflects the increase in ran-dom coil and/or β-sheet structure and decrease in disulfide bond asthe temperature increases from 50 to 70 °C whereas PC-2 reveals thatthere was a decrease in α-helix structure and a start for the increasein random coil and β-sheet structures below 60 °C. PC-2 also explainsa loss in the gauche–gauche–trans conformation as the temperature in-creased. With regard to the effect of cooking time (Fig. 7), it could besuggested that such clustering of the Raman spectra at the specificcooking temperatures (62 and 64 °C) might be due to structural chang-es in connective tissues during the extended cooking time which favorscollagenase (Tornberg, 2005). Similarly, an endothermic peak assignedto sarcoplasmic proteins and connective tissues was observed in thesame temperature range in the DSC thermogram (Fig. 1). The higher in-tensity ratio of both Tyr and Trp at 64 °C compared to the other temper-atures higher than 60 °C (Figs. 4 and 5) can support our suggestion. Ingeneral, the absence of a clear effect of cooking time in most of thecooking temperatures seems to be in accordance with other objectivemeasurement results (Christensen, Ertbjerg, et al., 2011) where in-creased cooking time from 3 to 8 h at cooking temperatures of 48, 53,58 and 63 °C did not affect the shear force value of the longissimus

dorsi of pigs. However, it is worth mentioning that prolonged cookingtime (above 10 h) could affect shear force value of longissimus dorsifrom pork (Christensen, Ertbjerg, et al., 2011). The novelty of the pres-ent study is, therefore, that Raman spectroscopy can provide suchbasic information about the stepwise changes in the secondary struc-ture and in turn the hydrophobic environment in intact meat rapidlywithout following tedious procedures (homogenization, incubation, ex-traction etc.) andwithout using chemicals. The structural changes in theconnective tissues were not pronounced as compared to the changes inmyosin. This could be due to the small amount of connective tissue inthe muscle (Christensen, Ertbjerg, et al., 2011) and/or other reasons.Therefore, further investigation is needed to study the structural chang-es in connective tissues during heat treatment using muscles with ahigh amount of connective tissues at higher temperatures.

The increase in cooking loss as a function of cooking temperatureand/or cooking time is in good agreement with the literature(Christensen, Bertram, et al., 2011; Christensen, Ertbjerg, et al., 2011;Martens et al., 1982; Palka& Daun, 1999). This parameter has a negativecorrelationwith the juiciness of a cooked product (Martens et al., 1982).Interestingly, previous work by Palka and Daun (1999) reported thehighest increase in cooking loss in the range of 50 to 60 °C which inthis study was particularly pronounced when the temperature in-creased from 56 to 58 °C (Fig. 8). The very good fit (R2 = 0.82)(Table 2) of themodel to predict the cooking loss shows the robustnessof the model and is in good agreement with previous works (Beattieet al., 2008; Schmidt et al., 2013). The positive contribution of β-sheetcan indicate that the structural changes in the meat proteins both inthe myofibrils and connective tissue were the causes for the loss ofwater holding capacity of the myofibrillar proteins (Palka, 1999). Itwas interesting to see the contribution of the same regions (1668,1642, 1340, 1235, 1009, 937, 758 and 520 cm−1) of the Raman spectrafor the prediction of cooking loss, temperature and time though therewere differences in higher latent variables and the models used differ-ent numbers of latent variables. This can indicate the general inter-relationship among these parameters with denaturation of meatproteins.

5. Conclusion

This study demonstrated that Raman spectroscopy can providevaluable information about stepwise conformational changes in thesecondary structure of meat proteins, disulfide bond and in turn thehydrophobic environment at the molecular level in intact muscle. Theeffect of cooking temperature on the structure of meat proteins waspronounced compared to the cooking time. Raman spectra from sam-ples cooked below and above 60 °C were discriminated. Bands atamide-I, amide-III and disulfide bond regions were found to be themost important spectral regions for the discrimination. Significantchanges in the intensity ratio of bands from Tyr and Trp residues werefound indicating changes in the hydrophobicity of the microenviron-ment during the thermal treatment. Cooking time did not show aclear effect in most of the cooking temperatures. It was possible to pre-dict cooking temperature and cooking loss with a very good correlationand also cooking time using Raman spectra and PLS regression. The po-tential of Raman spectroscopy to predict cooking temperature could beof potential interest for themeat industry for quality control as it is pos-sible to predict the core temperature of a given cooked meat productafter it has been cooked and chilled.

Acknowledgments

This study was funded by the University of Copenhagen and DanishMeat Research Institute (DMRI).We thank Frans van der Berg, Carl EmilAae Eskildsen and Claus Borggaard for their helpful discussions duringthe data analysis. Linda de Sparra Terkelsen is thanked for her help dur-ing sample collection.

131D.T. Berhe et al. / Food Research International 66 (2014) 123–131

References

Beattie, J. R., Bell, S. E. J., Borggaard, C., & Moss, B.W. (2008). Preliminary investigations onthe effects of ageing and cooking on the Raman spectra of porcine longissimus dorsi.Meat Science, 80, 1205–1211.

Beattie, R. J., Bell, S. J., Farmer, L. J., Moss, B., & Patterson, D. (2004). Preliminary investiga-tion of the application of Raman spectroscopy to the prediction of the sensory qualityof beef silverside. Meat Science, 66, 903–913.

Boyer, C., Joandel, S., Ouali, A., & Culioli, J. (1996). Determination of surface hydrophobic-ity of fast and slowmyosins from rabbit skeletal muscles: implication in heat-inducedgelation. Journal of the Science of Food and Agriculture, 72, 367–375.

Carew, E. B., Stanley, H. E., Seidel, J. C., & Gergely, J. (1983). Studies of myosin and its pro-teolytic fragments by laser Raman spectroscopy. Biophysical Journal, 44, 219–224.

Chan, J. K., Gill, T. A., & Paulson, A. T. (1992). The dynamics of thermal denaturation of fishmyosins. Food Research International, 25, 117–123.

Chen, M. C., & Lord, R. C. (1974). Laser-excited Raman spectroscopy of biomolecules. VI.Polypeptides as conformational models. Journal of the American Chemical Society, 96,4750–4752.

Christensen, L., Bertram, H. C., Aaslyng, M.D., & Christensen, M. (2011). Protein denatur-ation and water–protein interactions as affected by low temperature long time treat-ment of porcine longissimus dorsi. Meat Science, 88, 718–722.

Christensen, L., Ertbjerg, P., Aaslyng, M.D., & Christensen, M. (2011). Effect of prolongedheat treatment from 48 °C to 63 °C on toughness, cooking loss and color of pork.Meat Science, 88, 280–285.

Ellekjaer, M. R., & Isaksson, T. (1992). Assessment of maximum cooking temperatures inpreviously heat treated beef. Part 1: near infrared spectroscopy. Journal of the Scienceof Food and Agriculture, 59, 335–343.

Herrero, A.M. (2008). Raman spectroscopy for monitoring protein structure in musclefood systems. Critical Reviews in Food Science and Nutrition, 48, 512–523.

Herrero, A.M., Carmona, P., & Careche, M. (2004). Raman spectroscopic study of structuralchanges in hake (Merluccius merluccius L.) muscle proteins during frozen storage.Journal of Agricultural and Food Chemistry, 52, 2147–2153.

Herrero, A.M., Carmona, P., Cofrades, S., & Jimenez-Colmenero, F. (2008). Raman spectro-scopic determination of structural changes in meat batters upon soy protein additionand heat treatment. Food Research International, 41, 765–772.

Herrero, A.M., Carmona, P., Lopez-Lopez, I., & Jimenez-Colmenero, F. (2008). Raman spec-troscopic evaluation of meat batter structural changes induced by thermal treatmentand salt addition. Journal of Agricultural and Food Chemistry, 56, 7119–7124.

Lan, Y. H., Novakofski, J., McCusker, R. H., Brewer, M. S., Carr, T. R., & McKeith, F. K. (1995).Thermal gelation properties of protein fractions from pork and chicken breast mus-cles. Journal of Food Science, 60, 742–747.

Liu, R., Zhao, S., Xiong, S., Xie, B., & Qin, L. (2008). Role of secondary structures in the ge-lation of porcine myosin at different pH values. Meat Science, 80, 632–639.

Martens, H., Nielsen, J. P., & Engelsen, S. B. (2003). Light scattering and light absorbanceseparated by extended multiplicative signal correction. Application to near-infraredtransmission analysis of powder mixtures. Analytical Chemistry, 75, 394–404.

Martens, H., Stabursvik, E., & Martens, M. (1982). Texture and colour changes in meatduring cooking related to thermal denaturation of muscle proteins 1. Journal ofTexture Studies, 13, 291–309.

Mortensen, L. M., Frøst, M. B., Skibsted, L. H., & Risbo, J. (2012). Effect of time and temper-ature on sensory properties in low-temperature long-time sous-vide cooking of beef.Journal of Culinary Science & Technology, 10, 75–90.

Palka, K. (1999). Changes in intramuscular connective tissue and collagen solubility of bo-vine M. semitendinosus during retorting. Meat Science, 53, 189–194.

Palka, K., & Daun, H. (1999). Changes in texture, cooking losses, andmyofibrillar structureof bovine M. semitendinosus during heating. Meat Science, 51, 237–243.

Pedersen, D. K., Morel, S., Andersen, H. J., & Engelsen, S. B. (2003). Early prediction ofwater-holding capacity in meat by multivariate vibrational spectroscopy. MeatScience, 65, 581–592.

Schmidt, H., Scheier, R., & Hopkins, D. L. (2013). Preliminary investigation on the relation-ship of Raman spectra of sheep meat with shear force and cooking loss.Meat Science,93, 138–143.

Shao, J. H., Zou, Y. F., Xu, X. L., Wu, J. Q., & Zhou, G. H. (2011). Evaluation of structuralchanges in raw and heated meat batters prepared with different lipids usingRaman spectroscopy. Food Research International, 44, 2955–2961.

Sugeta, H., Go, A., & Miyazawa, T. (1973). Vibrational spectra and molecular conforma-tions of dialkyl disulfides. Bulletin of the Chemical Society of Japan, 46, 3407–3411.

Tornberg, E. (2005). Effects of heat on meat proteins — implications on structure andquality of meat products. Meat Science, 70, 493–508.

Van Wart, H. E., & Scheraga, H. A. (1986). Agreement with the disulfide stretching fre-quency–conformation correlation of Sugeta, Go, and Miyazawa. Proceedings of theNational Academy of Sciences, 83, 3064–3067.

Voutila, L., Mullen, A.M., Ruusunen, M., Troy, D., & Puolanne, E. (2007). Thermal stabilityof connective tissue from porcine muscles. Meat Science, 76, 474–480.

Wang, Q., Lonergan, S. M., & Yu, C. (2012). Rapid determination of pork sensory qualityusing Raman spectroscopy. Meat Science, 91, 232–239.

Wei, F., Zhang, D., Halas, N. J., & Hartgerink, J.D. (2008). Aromatic amino acids providingcharacteristic motifs in the raman and SERS spectroscopy of peptides. The Journal ofPhysical Chemistry B, 112, 9158–9164.

Wold, S., Esbensen, K., & Geladi, P. (1987). Principal component analysis. Chemometricsand Intelligent Laboratory Systems, 2, 37–52.

Wold, S., Martens, H., & Wold, H. (1983). The multivariate calibration problem in chemis-try solved by the PLS method. Lecture Notes in Mathematics(973), 286–293.

Xiong, Y. L., Brekke, C. J., & Leung, H. K. (1987). Thermal denaturation of muscle proteinsfrom different species and muscle types as studied by differential scanning calorime-try. Canadian Institute of Food Science and Technology Journal, 20, 357–362.

Xu, X. L., Han, M. Y., Fei, Y., & Zhou, G. H. (2011). Raman spectroscopic study of heat-induced gelation of pork myofibrillar proteins and its relationship with textural char-acteristic. Meat Science, 87, 159–164.