- Email: [email protected]
Predicting tenderness of fresh ovine semimembranosus using Raman spectroscopy
Meat Science 97 (2014) 597–601
Contents lists available at ScienceDirect
Meat Science journal homepage: www.elsevier.com/locate/meatsci
Predicting tenderness of fresh ovine semimembranosus using Raman spectroscopy Stephanie M. Fowler a,b,⁎, Heinar Schmidt c, Remy van de Ven d, Peter Wynn a,b, David L. Hopkins b,e a
School of Animal and Veterinary Sciences, Charles Sturt University, Wagga Wagga, Australia Graham Centre for Agricultural Innovation, NSW Department of Primary Industries and Charles Sturt University, Wagga Wagga, Australia Research Centre of Food Quality, University of Bayreuth, Kulmbach, Germany d Orange Institute of Agriculture, NSW Department of Primary Industries, Orange, Australia e Centre for Sheep and Red Meat Development, NSW Department of Primary Industries, Cowra, Australia b c
a r t i c l e
i n f o
Article history: Received 14 October 2013 Received in revised form 21 January 2014 Accepted 20 February 2014 Available online 12 March 2014 Keywords: Shear force Meat quality assessment Lamb Raman
a b s t r a c t A hand held Raman probe was used to predict shear force (SF) of fresh lamb m. semimembranosus (topside). Eighty muscles were measured at 1 day PM and after a further 4 days ageing (5 days PM). At 1 day PM sarcomere length (SL) and particle size (PS) were measured and at 5 days PM, SF, PS, cooking loss (CL) and pH were also measured. SF values were regressed against Raman spectra using partial least squares regression and against traditional predictors (e.g. SL) using linear regression. The best prediction of SF used spectra at 1 day PM which gave a root mean square error of prediction (RMSEP) of 11.5 N (Null = 13.2) and the squared correlation between observed and cross validated predicted values (R2cv) was 0.27. Prediction of SF based on the traditional predictors had smaller R2 values than using Raman spectra justifying further study on Raman spectroscopy. © 2014 Elsevier Ltd. All rights reserved.
1. Introduction Tenderness, juiciness and flavour are all factors which influence the eating quality of meat. Of these, tenderness is critical as a tough steak is unacceptable (Wood et al., 2008), although tenderness of lamb may have a lesser importance than other sensory attributes (Thompson et al., 2005). It has been established that the tenderness of meat is determined by the interaction between the connective tissue that creates ‘background toughness’, the myofibrillar structure (Damez & Clerjon, 2008) and the changes to these structures post mortem (Hopkins & Geesink, 2009). Consequently, considerable research has focused on the ability of technologies to objectively measure tenderness. A review of such technologies has highlighted Raman spectroscopy as having potential to be used for online measurement of meat quality traits (Damez & Clerjon, 2008). Based on the in-elastic scattering of light which can provide information about molecular composition and structure, Raman spectroscopy has potential for use in muscle food systems as it is rapid, non-destructive, non-invasive, not sensitive to varying water content and is not based on the absorption of light (Li-Chan, 1996). Previous research has not overlooked these advantages and several studies have demonstrated that Raman is a useful tool in predicting sensory traits in cooked ⁎ Corresponding author at: School of Animal and Veterinary Science, Locked Bag 588, Boorooma St, Wagga Wagga, NSW 2678, Australia. E-mail address: [email protected] (S.M. Fowler).
http://dx.doi.org/10.1016/j.meatsci.2014.02.018 0309-1740/© 2014 Elsevier Ltd. All rights reserved.
beef (Beattie, Bell, Farmer, Moss, & Patterson, 2004) and in explaining a large variation (R2 = 0.77) in shear force of pork (Beattie, Bell, Borggaard, & Moss, 2008). However, these studies used a bench top instrument that is not suitable for industrial application. Alternatively, Schmidt, Scheier, and Hopkins (2013) used a hand held Raman device suitable for online application to predict shear force of lamb meat with good predictability (R2 = 0.79 and 0.86 for two sample groups). Despite using the hand held Raman device, the industrial application of this study is limited as samples in the study were frozen and then thawed prior to measurement. Further to this, none of these studies cited have reported on the measurement of Raman spectroscopy, shear force and other predictors of tenderness on the same piece of meat. This study reports for the first time, the potential of a Raman hand held device to predict the shear force of fresh intact lamb m. semimembranosus (topside) and a comparison of Raman spectroscopy with the traditional predictors of shear force. 2. Materials and methods 2.1. Samples Topside (product identification number HAM 5077; Anonymous, 2005) samples were removed from 80 lamb carcases over 4 consecutive days (20 samples per day) from the same abattoir. Topsides were randomly selected from different consignments and thus were of unknown
598
S.M. Fowler et al. / Meat Science 97 (2014) 597–601
backgrounds, age and gender, to represent the various animals typically processed by the abattoir in order to achieve a spread in shear force levels. The cap muscle (m. gracilis) and m. adductor were removed to leave the m. semimembranosus (SM) which was the muscle of measurement.
2.2. Raman spectroscopy Raman spectroscopic measurements were conducted at ambient room temperature on day 1 post mortem on a fresh cut surface of the intact SM with the epimysium removed (Fig. 1). Twelve positions were scanned using a Raman hand held device (Schmidt, Sowoidnich, & Kronfeldt, 2010) perpendicular to the muscle fibre, over the face where the m. adductor had been removed. Spectra were recorded using 70 mW of laser power and an integration time of 3.75 s with no repetitions. After scanning and removal of sections for measurement of traditional indicators, SM samples were vacuum packed and held at 1 °C for 4 days. At 5 days post mortem, SM samples were removed from the vacuum packs and allowed to ‘bloom’ for 2 h before a freshly cut surface was rescanned.
2.3. Traditional predictors of tenderness At 1 day post mortem, sections were removed for sarcomere length analysis using the laser diffraction method (Bouton, Harris, Shorthose, & Baxter, 1973) and particle size analysis (PSA) (Karumendu, van de Ven, Kerr, Lanza, & Hopkins, 2009). At 5 days post mortem, shear force tests were conducted on blocks (mean weight 65 g) cut from the middle of the muscle after Raman scanning. Blocks were cooked at 71 °C for 35 min and analysed using a Lloyd texture analyser with a vee-blade as described by Hopkins, Toohey, Kerr, and van de Ven (2011). The average of 6 repetitions was reported except when the co-efficient of variation exceeded 24% in which case the median of the 6 repetitions was used (Hopkins, Kerr, Kerr, & van de Ven, 2012). Shear force blocks were weighed before and after cooking to determine cooking loss. The pHu was determined using 2.5 g of muscle homogenate in 10 ml of 5 mM iodoacetate/ 150 mM KCl (pH adjusted to 7.0), as described by Dransfield, Etherington, and Taylor (1992). Another section was taken for particle size analysis (Karumendu et al., 2009) at 5 days post mortem.
2.4. Data analysis Background interference was determined using a ‘dark’ scan, also taken at ambient temperature, which was subtracted from the spectra before they were saved as raw data. Prior to analysis non-meat spectra were identified using Principal Component Analysis (PCA). Scores were calculated with MATLAB 7.9.0 (R2009b) software (The MathWorks Inc., Natick, MA, USA). Any scores outside an established threshold value were checked and removed if intensities corresponding to lamb meat were not present in the spectra. Following this, 12 spectra for each sample were averaged, the wavelength range was restricted to a range of 500–1800 cm−1 and then the spectra were pre-processed by dividing each by its l2-norm (square root of sum of squared intensities). Models for predicting shear force which included Raman spectra were fitted using partial least squares (PLS) regression analysis, using the package pls (Martin, Hopkins, Gardner, & Thompson, 2006) under R (R Core Team, 2013). The number of latent variables (LV) was determined using 20 replications of 8-k fold cross validation and selecting the model with the minimum average root mean square error of prediction (RMSEP). Shear force predictions for each observed shear force value, for each SM sample, were then obtained, based on the selected number of LV, using the Leave-One-Out (LOO) cross validation method. An approach combining Raman spectra and the traditional predictors of shear force was also fitted using PLS regression. Models for the prediction of shear force based on traditional predictors of shear force, sarcomere length, pHu, cooking loss and/or PSA, omitting Raman spectra, were fitted using linear regression with R computer software (R Core Team, 2013). 3. Results In Table 1, summary results for the traditional predictors of shear force are given. This data demonstrates that shear force measurements had a large range but none were below 27 N, which indicates none were very tender (Hopkins, Hegarty, Walker, & Pethick, 2006). Comparison of PSA at 1 and 5 days post mortem demonstrates the effects of proteolysis as particle size decreased with ageing. Prediction errors (as quantified by the root mean square error of prediction (RMSEP)) for models using different combinations of the traditional predictors sarcomere length (SL), cooking loss (CL), pHu and/or PSA, with and without Raman spectra at 1 and 5 days are presented in Table 2. None of the traditional measures were significant predictors of shear force alone or jointly (P N 0.05). Based on the RMSEP criterion, the best model for predicting shear force of SM at 5 days post mortem was based on Raman spectra collected on day 1 (Table 2), using 3 latent variables. The squared correlation between the Leave-One-Out cross validated predicted and observed shear force values was R2cv = 0.27 (Fig. 2.). Changes to Raman spectra due to ageing are complex and are outside the scope of this paper, however the results reported here indicate that the prediction of shear force using Raman spectra taken on day 1 is better (R2cv = 0.27; RMSEP = 11.48) than the prediction using Raman spectra collected 5 days post mortem (R2cv = 0.17; RMSEP = 12.20). Overall, this is a 1.4 N reduction in RMSEP when compared to the Null prediction model Table 1 Mean, standard deviation, and range of shear force (N), cooking loss (%), sarcomere length (μm), pHu and PSA (μm) of lamb m. semimembranosus (n = 80).
Fig. 1. Hand held Raman spectroscopy sensor head measuring a fresh intact lamb SM, with the epimysium removed.
Trait
Ageing (days)
Mean
SD
Range (min, max)
Shear force (N) Cooking loss (%) Sarcomere length (μm) pHu PSA (μm) PSA (μm)
5 5 1 1 1 5
51.4 19.2 1.70 5.61 229 166
13.1 3.7 0.11 0.11 46.8 40.8
29.2–78.4 0.24–28.8 1.46–1.99 5.52–6.23 159–455 95–322
S.M. Fowler et al. / Meat Science 97 (2014) 597–601 Table 2 The RMSEP for models using traditional predictors and/or Raman spectra to predict shear force values (N) of lamb SM at 5 days post mortem. Model covariates
RMSEP (N)
Relative reduction RMSEP (%)
Null Cooking loss Sarcomere length pHu PSA CL, SL, pHu and PSA Raman spectra (1 day) Raman spectra (5 day) Raman spectra (1 day) + CL, SL, pHu and PSA Raman spectra (5 day) + CL, SL, pHu and PSA
13.2 13.3 13.2 13.8 13.5 13.4 11.5 12.2 11.8 12.5
– −0.8 0 −4.5 −2.3 −1.5 12.9 7.6 10.6 5.3
(root mean square error of predicted shear force values based on average shear force leaving one observation out at a time). As illustrated in Fig. 3, the largest changes between the averaged spectra of the 5 toughest and 5 most tender SM occurred between the 800 cm−1 and 1125 cm−1 wavelengths. 4. Discussion 4.1. Prediction of shear force using Raman spectra Our results gave a relatively poor accuracy for predicting shear force in comparison to previous Raman research on lamb, as Schmidt et al. (2013) reported higher coefficients of determination (R2) of 0.72 and 0.86 when predicting shear force values from carcases collected from two different sites (RMSEC = 3.2% and RMSECV = 8.7% and RMSEC = 2.6% and RMSECV = 7.6%, respectively). While both studies have been conducted on lamb, there is a limited ability to compare between reported results as disparities exist in experimental parameters and design. Sample handling varies greatly between these studies as Schmidt et al. (2013) conducted Raman measurements on 3 cm thick sections of loin which had been aged for 5 days post mortem, frozen at −20 °C and thawed prior to measurement. During freezing, changes to the biochemical characteristics of muscle occur as water moves out of cells and cellular membranes (Sikorski, 1978). This movement of water damages surrounding protein side chains by rupturing the hydrophilic/ hydrophobic adherences and causing mechanical damage to the muscle structures and increasing salt levels, which further reduces the number of hydrophilic protein groups (Sikorski, 1978). As Raman scattering in muscle is characterised by the vibrations of chemical bonds of the peptide backbone and amino acid side chains, it is sensitive to the damage that is induced by freezing and thawing (Herrero, 2008a; Li-Chan,
Predicted Shear Force (N)
80 70 60 50 40 30 30
40
50
60
70
80
Observed Shear Force (N) Fig. 2. Cross validated prediction of shear force values (N) at 5 days post mortem with Raman spectra collected on day 1 post mortem and analysed with 3 latent vectors (R2 = 0.27, RMSEP = 11.5).
599
Nakai, & Hirotsuka, 1994). Consequently, it is problematic when comparing results between Raman studies that have used fresh versus frozen then thawed samples. Li-Chan (1996) and Herrero (2008b) have addressed how some of the biochemical changes are reflected in the Raman spectra due to hydrophobic interactions after storage (Herrero, 2008b) and the denaturation of myosin (Li-Chan, 1996). However, changes to Raman spectra as a result of water movement and decreased water holding capacity, which occur when lamb is frozen (Muela, Sañudo, Campo, Medel, & Beltrán, 2010) have not yet been characterised. It is hypothesised that movement of water within the myofibril during freezing removes the overlap effect of solvent water within Raman bands that reflect the peptide backbone conformation. Removal of this overlap effect reveals the changes in secondary structure of proteins in the amide I band as well as removal of the miscellaneous side chain vibrations that contribute to the amide III band (Li-Chan, 1996). Consequently, the Raman spectra of the secondary protein structures become more distinct and this enables a better prediction of shear force values. Previous studies using Raman spectroscopy to measure porcine m. longissimus and beef silverside also produced relatively high R2 values for predicting shear force of 0.77 and 0.75 respectively (Beattie et al., 2004, 2008). However, as with previous studies conducted on lamb, a direct comparison between results reported in literature and models created in the research reported here cannot be made as there is little accordance between experimental design and parameters for Raman spectroscopic measurement. Although all studies aim to predict shear force values measured using a Warner-Bratzler texture analyser with Raman spectra, as with Schmidt et al. (2013) both of these studies (Beattie et al., 2004, 2008) conducted Raman measurements on meat samples that had been subjected to freezing and thawing prior to measurement. Further complicating any direct comparison between prediction models, the number of independent samples measured in these studies was small, chemometric analysis of spectral data varies, spectra have been collected using a bench top Raman device with varying wavelengths and the samples have been measured on a rotating stage using various accumulation and integration times. 4.2. Prediction of shear force using traditional measures Previous studies have used coefficients of determination (R2) to describe the relationship between traditional measures such as sarcomere length, PSA or pHu and shear force (Bouton, Carroll, Fisher, Harris, & Shorthose, 1973; Hood & Tarrant, 1981; Hopkins et al., 2007; Smulders, Marsh, Swartz, Russell, & Hoenecke, 1990). However, it needs to be recognised that as linear correlations are dependent on the range of the data set, they are not a good measure of the merit of the calibration between data sets which have different ranges (Davies & Fearn, 2006). Therefore it is problematic to compare between studies where different experimental designs and treatments have been applied to artificially influence results and highlight interrelationships between the traditional predictors and shear force. For example, Bouton, Harris, and Shorthose (1971) induced a greater range of pHu values (5.6–7.0), while others have studied different muscles (Geesink, Sujang, & Koohmaraie, 2011; Karumendu et al., 2009), species (Smulders et al., 1990) and processing practices including ageing periods and hanging methods (Bouton, Harris, et al., 1973). While informative about the impact of extremes in traditional measures on shear force values, these studies may not be indicative of the range or distribution of values for traditional measures of lamb SM which are typically processed. Consequently, the R2 values previously reported may not be good estimates of the differences which would be found in standard carcases commonly processed (Davies & Fearn, 2006). Relationships between traditional indicators, such as pH and shear force values aren't always linear and may be confounded by interrelationships with other factors (Hopkins et al., 2006). Given this it is not surprising that while the variation in shear force explained by
600
S.M. Fowler et al. / Meat Science 97 (2014) 597–601
Fig. 3. Changes in intensities of background corrected Raman spectra from tender (29–32 N; black) and tough (74–78 N; grey) fresh intact lamb SM.
traditional predictors in this study agrees with some other studies (Hopkins, Toohey, Lamb, Kerr, van de Ven, & Refshauge, 2011), it does not agree with others (Bouton, Carroll, et al., 1973; Hopkins et al., 2006; Karumendu et al., 2009). A comparison between the root mean square error of prediction (RMSEP) to give a more robust explanation of the variation of predictions using different co-variates (Davies & Fearn, 2006) suggested that no significant improvement in prediction models was gained by combining traditional predictors with Raman spectra (Table 2). When this is considered with the accuracy of the prediction models, Raman spectra at 1 day post mortem has the greatest potential as an accurate and precise predictor of tenderness. As with previous Raman studies on meat (Beattie et al., 2004, 2008; Schmidt et al., 2013), it is not possible to assign changes in the Raman spectra of tough and tender lamb with certainty, although a tentative interpretation of spectra is plausible based on the relationship between Raman spectra and the composition and structure of proteins. Spectra collected in this data set (Fig. 3.) exhibit the typical Raman signal for muscle tissue with peaks in intensity representing key chemical bond vibrations of amino acids including tryptophan, tyrosine and phenylalanine, as well as those of protein backbone conformations and secondary protein structures (Pézolet, Pigeon, Ménard, & Caillé, 1988). The largest changes in intensities between spectra from tough and tender SM relate to the tyrosine doublet at wavelengths 826 cm− 1 and 853 cm−1 and the α-helix peak at 930 cm−1 (Fig. 3). These are similar to those found by Schmidt et al. (2013) who reported that the tyrosine doublet weakened in tough lamb and therefore the presence of this doublet signal could be used to classify tender lamb. The results of Beattie et al. (2008) support this, suggesting that shifts in tyrosine bands are a direct reflection of proteolysis given the amount of free tyrosine increases with ageing in normal quality pork. However, Fig. 3 also indicates that there is a difference in the intensity ratio between the two peaks that combine to create the tyrosine doublet. In the tender SM samples, the tyrosine doublet peak intensities at 826 cm− 1 to 853 cm− 1 have a ratio of 0.56:0.61 counts per second (c/s) while in tough SM samples the intensities have a ratio of 0.49:0.48 c/s. This equates to a 12% difference in the peak height at 853 cm−1 between tough and tender samples, as there was a 9% increase for tender samples and a 3% decrease for tough SM. While tyrosine is an amino acid and therefore present in many proteins, it is known that the relative loss of intensity of tyrosine at approximately 830 cm−1 is indicative of the environment of hydrogen bonds in the aromatic side chain of tyrosine (Bonnier et al., 2011). As the phenolic hydroxyl group binds tightly
when oxygen is the donor, but relatively loosely when hydrogen serves this function (Nelson & Cox, 2008), a change in what functions as the donor in the phenyl group of tyrosine may be indicative of the hydrophobicity of the proteins as well as their ability to resist oxidation and reduction (Nelson & Cox, 2008). Consequently, changes in peak intensities may indicate that the tough SM contain a greater number of tyrosine moieties with hydroxyl groups bound to oxygen, therefore reducing the ability of proteolysis to cleave these amino acids in myofibrillar proteins. Larger disparities between intensities exist at the wavelength range between 870 and 1122 cm−1, which represents a combination of bands including tryptophan (Trp), the C\N stretch and C\C stretch vibrations (Krimm & Bandekar, 1986). Previous Raman protein studies have identified the importance of the C\C stretch vibrations at approximately 930 cm−1 as they represent the α-helical secondary structures of proteins (Herrero, 2008b). Therefore, Fig. 3 suggests a decrease in the number of α-helical protein structures in tough lamb SM. When secondary protein structures convert from α-helices into β-sheets or random coils due to denaturation the intensity of the Raman signal corresponding to α-helices at approximately 930 cm− 1 becomes broader and weaker (Tu, 1986), which could contribute to the loss of intensity peak and symmetry in spectra of tough SM. This agrees with Beattie et al. (2004) who suggested that tough meat is characterised by a larger number of β-sheets in comparison to tender meat. However, the presence of the α-helical protein backbone structures is dependent on sample orientation. Therefore inducing polarised changes parallel to the bonds results in a different excitation pattern in contrast to changes induced perpendicular to the bonds (Pézolet et al., 1988). Consequently, by placing samples on a rotating stage, Beattie et al. (2008) and Beattie et al. (2004) would have artificially reduced the contribution of α-helices to the spectra by continually changing the angle at which the laser enters the sample relative to the fibre orientation and reducing the time during which α-helical segments would be polarised. As muscle is a closed system, the attributes of muscle which determine the effects of proteolysis and the subsequent amount of myofibrillar degradation during ageing are predetermined by the biophysical and biochemical properties of the sample at processing. It is hypothesised that this change in myofibrillar structure weakens the Raman signal when spectra are measured at 5 days. If changes to ionic strength during ageing are affecting the ability of Raman to predict shear force it is hypothesised that collecting spectra as the muscle enters rigor would improve the prediction, as ionic strength of sarcoplasmic fluid rapidly rises as the muscle enters rigor (Ouali, 1992).
S.M. Fowler et al. / Meat Science 97 (2014) 597–601
5. Conclusion Overall it is difficult to determine the ability of Raman spectroscopy to predict shear force values of intact lamb, as there is currently no opportunity to compare these results with other studies using the same sample handling and experimental design. Therefore, the accuracy and precision of these predictions need to be validated. However, this study does indicate that Raman spectra may be a better indicator of variation in shear force compared to the traditional measures of sarcomere length, cooking loss, pHu and PSA. Direct comparison of the Raman spectra demonstrated that the discrimination between tough and tender fresh intact SM can be made using the intensity of spectral peaks that correspond to the tyrosine doublet at 826 and 853 cm−1 and αhelix at 930 cm−1. The impact of early post mortem events and collagen on the prediction of shear force using Raman spectroscopy is yet to be determined and the next phase of work will need to address these issues. Acknowledgements This work has been financially supported by the Australian Meat Processor Corporation (AMPC) and Meat and Livestock Australia (MLA), as is the senior author by scholarship. The authors also acknowledge the contribution of Matt Kerr, Tracy Lamb and Kristy Bailes (NSW DPI) who assisted during the measurement of the samples. References Anonymous (2005). Handbook of Australian meat (7th ed.). Brisbane, Australia: AUSMEAT Limited. Beattie, J. R., Bell, S. E. J., Borggaard, C., & Moss, B. W. (2008). Preliminary investigations on the effects of ageing and cooking on the Raman spectra of porcine longissimus dorsi. Meat Science, 80, 1205–1211. Beattie, J. R., Bell, S. J., Farmer, L. J., Moss, B. W., & Patterson, D. (2004). Preliminary investigation of the application of Raman spectroscopy to the prediction of the sensory quality of beef silverside. Meat Science, 66, 903–913. Bonnier, F., Mehmood, A., Knief, P., Meade, A.D., Hornebeck, W., Lambkin, H., Flynn, H., McDonagh, V., Healy, C., Lee, T. C., Lyng, F. M., & Byrne, H. J. (2011). In vitro analysis of immersed human tissues by Raman microspectroscopy. Journal of Raman Spectroscopy, 42, 888–896. Bouton, P. E., Carroll, F. D., Fisher, A. L., Harris, V., & Shorthose, W. R. (1973). Effect of altering ultimate pH on bovine muscle tenderness. Journal of Food Science, 38, 816–822. Bouton, P. E., Harris, P. V., & Shorthose, W. R. (1971). Effect of ultimate pH upon the water holding capacity and tenderness of mutton. Journal of Food Science, 36, 435–439. Bouton, P. E., Harris, P. V., Shorthose, W. R., & Baxter, R. I. (1973). Comparison of the effects of aging, conditioning and skeletal restraint on the tenderness of mutton. Journal of Food Science, 38. Damez, J. -L., & Clerjon, S. (2008). Meat quality assessment using biophysical methods related to meat structure. Meat Science, 80, 132–149. Davies, A.M., & Fearn, T. (2006). Back to basics: Calibration statistics. Spectroscopy Europe, 18, 31–32. Dransfield, E., Etherington, D. J., & Taylor, M.A. J. (1992). Modelling post-mortem tenderisation—II: Enzyme changes during storage of electrically stimulated and non-stimulated beef. Meat Science, 31, 75–84. Geesink, G. H., Sujang, S., & Koohmaraie, M. (2011). Tenderness of pre- and post rigor lamb longissimus muscle. Meat Science, 88, 723–726. Herrero, A.M. (2008). Raman spectroscopy a promising technique for quality assessment of meat and fish: A review. Food Chemistry, 107, 1642–1651. Herrero, A.M. (2008). Raman spectroscopy for monitoring protein structure in muscle food systems. Critical Reviews in Food Science and Nutrition, 48, 512–523.
601
Hood, D. E., & Tarrant, P. V. (1981). The problem of dark-cutting in beef. The Netherlands: Martinus Nijhoff Publishers. Hopkins, D. L., & Geesink, G. H. (2009). Protein degradation post mortem and tenderisation. In M. Du, & R. McCormick (Eds.), Applied muscle biology and meat science (pp. 149–173). USA: CRC Press, Taylor & Francis Group. Hopkins, D. L., Hegarty, R. S., Walker, P. J., & Pethick, D. W. (2006). Relationship between animal age, intramuscular fat, cooking loss, pH, shear force and eating quality of aged meat from young sheep. Australian Journal of Experimental Agriculture, 46, 879–884. Hopkins, D. L., Kerr, M. J., Kerr, M. G., & van de Ven, R. (2012). Within sample variance for shear force testing of lamb meat. Proceedings of the 29th Biennial Conference of the Australian Society of, Animal Production (pp. 31). Hopkins, D. L., Stanley, D. F., Toohey, E. S., Gardner, G. E., Pethick, D. W., & van de Ven, R. (2007). Sire and growth path effects on sheep meat production 2. Meat and eating quality. Australian Journal of Experimental Agriculture, 47, 1219–1228. Hopkins, D. L., Toohey, E. S., Kerr, M. J., & van de Ven, R. (2011). Comparison of two instruments (G2 Tenderometer and a Lloyd Texture analyser) for measuring the shear force of cooked meat. Animal Production Science, 51, 71–76. Hopkins, D. L., Toohey, E. S., Lamb, T. A., Kerr, M. J., van de Ven, R., & Refshauge, G. (2011). Explaining the variation in the shear force of lamb meat using sarcomere length, the rate of rigor onset and pH. Meat Science, 88, 794–796. Karumendu, L. U., van de Ven, R., Kerr, M. J., Lanza, M., & Hopkins, D. L. (2009). Particle size analysis of lamb meat: Effect of homogenization speed, comparison with myofibrillar fragmentation index and its relationship with shear force. Meat Science, 82, 425–431. Krimm, S., & Bandekar, J. (1986). Vibrational spectroscopy and conformation of peptides, polypeptides, and proteins. In C. B. Anfinsen, J. T. Edsall, & M. F. Richards (Eds.), Advances in protein chemistry (pp. 181–364). : Academic Press. Li-Chan, E. C. Y. (1996). The applications of Raman spectroscopy in food science. Trends in Food Science & Technology, 7, 361–370. Li-Chan, E. C. Y., Nakai, S., & Hirotsuka, M. (1994). Raman spectroscopy as a probe of protein structure in food systems. In R. Y. Yada, R. L. Jackman, & J. L. Smith (Eds.), Protein structure–function relationships in food (pp. 163–198). NZ: Blackie Academic and Professional Glasgow. Martin, K. M., Hopkins, D. L., Gardner, G. E., & Thompson, J. M. (2006). Effects of stimulation on tenderness of lamb with a focus on protein degradation. Australian Journal of Experimental Agriculture, 46, 891–896. Muela, E., Sañudo, C., Campo, M. M., Medel, I., & Beltrán, J. A. (2010). Effect of freezing method and frozen storage duration on instrumental quality of lamb throughout display. Meat Science, 84, 662–669. Nelson, D. L., & Cox, M. M. (2008). Principles of biochemistry (5th ed.). New York, USA: Freeman and Company. Ouali, A. (1992). Proteolytic and physicochemical mechanisms involved in meat texture development. Biochimie, 74, 251–265. Pézolet, M., Pigeon, M., Ménard, D., & Caillé, J. P. (1988). Raman spectroscopy of cytoplasmic muscle fiber proteins. Orientational order. Biophysical Journal, 53, 319–325. R Core Team (2013). R: A language and environment for statistical computing. (Vienna, Austria): R Foundation for Statistical Computing. Schmidt, H., Scheier, R., & Hopkins, D. L. (2013). Preliminary investigation on the relationship of Raman spectra of sheep meat with shear force and cooking loss. Meat Science, 93, 138–143. Schmidt, H., Sowoidnich, K., & Kronfeldt, H. D. (2010). A prototype hand-held raman sensor for the in situ characterization of meat quality. Applied Spectroscopy, 64, 888–894. Sikorski, Z. E. (1978). Protein changes in muscle foods due to freezing and frozen storage. International Journal of Refrigeration, 1, 173–180. Smulders, F. J. M., Marsh, B. B., Swartz, D. R., Russell, R. L., & Hoenecke, M. E. (1990). Beef tenderness and sarcomere length. Meat Science, 28, 349–363. Thompson, J. M., Hopkins, D. L., D'Souza, D. N., Walker, P. J., Baud, S. R., & Pethick, D. W. (2005). The impact of processing on sensory and objective measurements of sheep meat eating quality. Australian Journal of Experimental Agriculture, 45, 561–573. Tu, A. T. (1986). Peptide backbone conformation and microenvironment of protein side chain. In R. J. H. Clark, & R. E. Hester (Eds.), Spectroscopy of biological systems. : John Wiley and Sons Ltd. Wood, J.D., Enser, M., Fisher, A. V., Nute, G. R., Sheard, P. R., Richardson, R. I., Hughes, S. I., & Whittington, F. M. (2008). Fat deposition, fatty acid composition and meat quality: A review. Meat Science, 78, 343–358.
Contents lists available at ScienceDirect
Meat Science journal homepage: www.elsevier.com/locate/meatsci
Predicting tenderness of fresh ovine semimembranosus using Raman spectroscopy Stephanie M. Fowler a,b,⁎, Heinar Schmidt c, Remy van de Ven d, Peter Wynn a,b, David L. Hopkins b,e a
School of Animal and Veterinary Sciences, Charles Sturt University, Wagga Wagga, Australia Graham Centre for Agricultural Innovation, NSW Department of Primary Industries and Charles Sturt University, Wagga Wagga, Australia Research Centre of Food Quality, University of Bayreuth, Kulmbach, Germany d Orange Institute of Agriculture, NSW Department of Primary Industries, Orange, Australia e Centre for Sheep and Red Meat Development, NSW Department of Primary Industries, Cowra, Australia b c
a r t i c l e
i n f o
Article history: Received 14 October 2013 Received in revised form 21 January 2014 Accepted 20 February 2014 Available online 12 March 2014 Keywords: Shear force Meat quality assessment Lamb Raman
a b s t r a c t A hand held Raman probe was used to predict shear force (SF) of fresh lamb m. semimembranosus (topside). Eighty muscles were measured at 1 day PM and after a further 4 days ageing (5 days PM). At 1 day PM sarcomere length (SL) and particle size (PS) were measured and at 5 days PM, SF, PS, cooking loss (CL) and pH were also measured. SF values were regressed against Raman spectra using partial least squares regression and against traditional predictors (e.g. SL) using linear regression. The best prediction of SF used spectra at 1 day PM which gave a root mean square error of prediction (RMSEP) of 11.5 N (Null = 13.2) and the squared correlation between observed and cross validated predicted values (R2cv) was 0.27. Prediction of SF based on the traditional predictors had smaller R2 values than using Raman spectra justifying further study on Raman spectroscopy. © 2014 Elsevier Ltd. All rights reserved.
1. Introduction Tenderness, juiciness and flavour are all factors which influence the eating quality of meat. Of these, tenderness is critical as a tough steak is unacceptable (Wood et al., 2008), although tenderness of lamb may have a lesser importance than other sensory attributes (Thompson et al., 2005). It has been established that the tenderness of meat is determined by the interaction between the connective tissue that creates ‘background toughness’, the myofibrillar structure (Damez & Clerjon, 2008) and the changes to these structures post mortem (Hopkins & Geesink, 2009). Consequently, considerable research has focused on the ability of technologies to objectively measure tenderness. A review of such technologies has highlighted Raman spectroscopy as having potential to be used for online measurement of meat quality traits (Damez & Clerjon, 2008). Based on the in-elastic scattering of light which can provide information about molecular composition and structure, Raman spectroscopy has potential for use in muscle food systems as it is rapid, non-destructive, non-invasive, not sensitive to varying water content and is not based on the absorption of light (Li-Chan, 1996). Previous research has not overlooked these advantages and several studies have demonstrated that Raman is a useful tool in predicting sensory traits in cooked ⁎ Corresponding author at: School of Animal and Veterinary Science, Locked Bag 588, Boorooma St, Wagga Wagga, NSW 2678, Australia. E-mail address: [email protected] (S.M. Fowler).
http://dx.doi.org/10.1016/j.meatsci.2014.02.018 0309-1740/© 2014 Elsevier Ltd. All rights reserved.
beef (Beattie, Bell, Farmer, Moss, & Patterson, 2004) and in explaining a large variation (R2 = 0.77) in shear force of pork (Beattie, Bell, Borggaard, & Moss, 2008). However, these studies used a bench top instrument that is not suitable for industrial application. Alternatively, Schmidt, Scheier, and Hopkins (2013) used a hand held Raman device suitable for online application to predict shear force of lamb meat with good predictability (R2 = 0.79 and 0.86 for two sample groups). Despite using the hand held Raman device, the industrial application of this study is limited as samples in the study were frozen and then thawed prior to measurement. Further to this, none of these studies cited have reported on the measurement of Raman spectroscopy, shear force and other predictors of tenderness on the same piece of meat. This study reports for the first time, the potential of a Raman hand held device to predict the shear force of fresh intact lamb m. semimembranosus (topside) and a comparison of Raman spectroscopy with the traditional predictors of shear force. 2. Materials and methods 2.1. Samples Topside (product identification number HAM 5077; Anonymous, 2005) samples were removed from 80 lamb carcases over 4 consecutive days (20 samples per day) from the same abattoir. Topsides were randomly selected from different consignments and thus were of unknown
598
S.M. Fowler et al. / Meat Science 97 (2014) 597–601
backgrounds, age and gender, to represent the various animals typically processed by the abattoir in order to achieve a spread in shear force levels. The cap muscle (m. gracilis) and m. adductor were removed to leave the m. semimembranosus (SM) which was the muscle of measurement.
2.2. Raman spectroscopy Raman spectroscopic measurements were conducted at ambient room temperature on day 1 post mortem on a fresh cut surface of the intact SM with the epimysium removed (Fig. 1). Twelve positions were scanned using a Raman hand held device (Schmidt, Sowoidnich, & Kronfeldt, 2010) perpendicular to the muscle fibre, over the face where the m. adductor had been removed. Spectra were recorded using 70 mW of laser power and an integration time of 3.75 s with no repetitions. After scanning and removal of sections for measurement of traditional indicators, SM samples were vacuum packed and held at 1 °C for 4 days. At 5 days post mortem, SM samples were removed from the vacuum packs and allowed to ‘bloom’ for 2 h before a freshly cut surface was rescanned.
2.3. Traditional predictors of tenderness At 1 day post mortem, sections were removed for sarcomere length analysis using the laser diffraction method (Bouton, Harris, Shorthose, & Baxter, 1973) and particle size analysis (PSA) (Karumendu, van de Ven, Kerr, Lanza, & Hopkins, 2009). At 5 days post mortem, shear force tests were conducted on blocks (mean weight 65 g) cut from the middle of the muscle after Raman scanning. Blocks were cooked at 71 °C for 35 min and analysed using a Lloyd texture analyser with a vee-blade as described by Hopkins, Toohey, Kerr, and van de Ven (2011). The average of 6 repetitions was reported except when the co-efficient of variation exceeded 24% in which case the median of the 6 repetitions was used (Hopkins, Kerr, Kerr, & van de Ven, 2012). Shear force blocks were weighed before and after cooking to determine cooking loss. The pHu was determined using 2.5 g of muscle homogenate in 10 ml of 5 mM iodoacetate/ 150 mM KCl (pH adjusted to 7.0), as described by Dransfield, Etherington, and Taylor (1992). Another section was taken for particle size analysis (Karumendu et al., 2009) at 5 days post mortem.
2.4. Data analysis Background interference was determined using a ‘dark’ scan, also taken at ambient temperature, which was subtracted from the spectra before they were saved as raw data. Prior to analysis non-meat spectra were identified using Principal Component Analysis (PCA). Scores were calculated with MATLAB 7.9.0 (R2009b) software (The MathWorks Inc., Natick, MA, USA). Any scores outside an established threshold value were checked and removed if intensities corresponding to lamb meat were not present in the spectra. Following this, 12 spectra for each sample were averaged, the wavelength range was restricted to a range of 500–1800 cm−1 and then the spectra were pre-processed by dividing each by its l2-norm (square root of sum of squared intensities). Models for predicting shear force which included Raman spectra were fitted using partial least squares (PLS) regression analysis, using the package pls (Martin, Hopkins, Gardner, & Thompson, 2006) under R (R Core Team, 2013). The number of latent variables (LV) was determined using 20 replications of 8-k fold cross validation and selecting the model with the minimum average root mean square error of prediction (RMSEP). Shear force predictions for each observed shear force value, for each SM sample, were then obtained, based on the selected number of LV, using the Leave-One-Out (LOO) cross validation method. An approach combining Raman spectra and the traditional predictors of shear force was also fitted using PLS regression. Models for the prediction of shear force based on traditional predictors of shear force, sarcomere length, pHu, cooking loss and/or PSA, omitting Raman spectra, were fitted using linear regression with R computer software (R Core Team, 2013). 3. Results In Table 1, summary results for the traditional predictors of shear force are given. This data demonstrates that shear force measurements had a large range but none were below 27 N, which indicates none were very tender (Hopkins, Hegarty, Walker, & Pethick, 2006). Comparison of PSA at 1 and 5 days post mortem demonstrates the effects of proteolysis as particle size decreased with ageing. Prediction errors (as quantified by the root mean square error of prediction (RMSEP)) for models using different combinations of the traditional predictors sarcomere length (SL), cooking loss (CL), pHu and/or PSA, with and without Raman spectra at 1 and 5 days are presented in Table 2. None of the traditional measures were significant predictors of shear force alone or jointly (P N 0.05). Based on the RMSEP criterion, the best model for predicting shear force of SM at 5 days post mortem was based on Raman spectra collected on day 1 (Table 2), using 3 latent variables. The squared correlation between the Leave-One-Out cross validated predicted and observed shear force values was R2cv = 0.27 (Fig. 2.). Changes to Raman spectra due to ageing are complex and are outside the scope of this paper, however the results reported here indicate that the prediction of shear force using Raman spectra taken on day 1 is better (R2cv = 0.27; RMSEP = 11.48) than the prediction using Raman spectra collected 5 days post mortem (R2cv = 0.17; RMSEP = 12.20). Overall, this is a 1.4 N reduction in RMSEP when compared to the Null prediction model Table 1 Mean, standard deviation, and range of shear force (N), cooking loss (%), sarcomere length (μm), pHu and PSA (μm) of lamb m. semimembranosus (n = 80).
Fig. 1. Hand held Raman spectroscopy sensor head measuring a fresh intact lamb SM, with the epimysium removed.
Trait
Ageing (days)
Mean
SD
Range (min, max)
Shear force (N) Cooking loss (%) Sarcomere length (μm) pHu PSA (μm) PSA (μm)
5 5 1 1 1 5
51.4 19.2 1.70 5.61 229 166
13.1 3.7 0.11 0.11 46.8 40.8
29.2–78.4 0.24–28.8 1.46–1.99 5.52–6.23 159–455 95–322
S.M. Fowler et al. / Meat Science 97 (2014) 597–601 Table 2 The RMSEP for models using traditional predictors and/or Raman spectra to predict shear force values (N) of lamb SM at 5 days post mortem. Model covariates
RMSEP (N)
Relative reduction RMSEP (%)
Null Cooking loss Sarcomere length pHu PSA CL, SL, pHu and PSA Raman spectra (1 day) Raman spectra (5 day) Raman spectra (1 day) + CL, SL, pHu and PSA Raman spectra (5 day) + CL, SL, pHu and PSA
13.2 13.3 13.2 13.8 13.5 13.4 11.5 12.2 11.8 12.5
– −0.8 0 −4.5 −2.3 −1.5 12.9 7.6 10.6 5.3
(root mean square error of predicted shear force values based on average shear force leaving one observation out at a time). As illustrated in Fig. 3, the largest changes between the averaged spectra of the 5 toughest and 5 most tender SM occurred between the 800 cm−1 and 1125 cm−1 wavelengths. 4. Discussion 4.1. Prediction of shear force using Raman spectra Our results gave a relatively poor accuracy for predicting shear force in comparison to previous Raman research on lamb, as Schmidt et al. (2013) reported higher coefficients of determination (R2) of 0.72 and 0.86 when predicting shear force values from carcases collected from two different sites (RMSEC = 3.2% and RMSECV = 8.7% and RMSEC = 2.6% and RMSECV = 7.6%, respectively). While both studies have been conducted on lamb, there is a limited ability to compare between reported results as disparities exist in experimental parameters and design. Sample handling varies greatly between these studies as Schmidt et al. (2013) conducted Raman measurements on 3 cm thick sections of loin which had been aged for 5 days post mortem, frozen at −20 °C and thawed prior to measurement. During freezing, changes to the biochemical characteristics of muscle occur as water moves out of cells and cellular membranes (Sikorski, 1978). This movement of water damages surrounding protein side chains by rupturing the hydrophilic/ hydrophobic adherences and causing mechanical damage to the muscle structures and increasing salt levels, which further reduces the number of hydrophilic protein groups (Sikorski, 1978). As Raman scattering in muscle is characterised by the vibrations of chemical bonds of the peptide backbone and amino acid side chains, it is sensitive to the damage that is induced by freezing and thawing (Herrero, 2008a; Li-Chan,
Predicted Shear Force (N)
80 70 60 50 40 30 30
40
50
60
70
80
Observed Shear Force (N) Fig. 2. Cross validated prediction of shear force values (N) at 5 days post mortem with Raman spectra collected on day 1 post mortem and analysed with 3 latent vectors (R2 = 0.27, RMSEP = 11.5).
599
Nakai, & Hirotsuka, 1994). Consequently, it is problematic when comparing results between Raman studies that have used fresh versus frozen then thawed samples. Li-Chan (1996) and Herrero (2008b) have addressed how some of the biochemical changes are reflected in the Raman spectra due to hydrophobic interactions after storage (Herrero, 2008b) and the denaturation of myosin (Li-Chan, 1996). However, changes to Raman spectra as a result of water movement and decreased water holding capacity, which occur when lamb is frozen (Muela, Sañudo, Campo, Medel, & Beltrán, 2010) have not yet been characterised. It is hypothesised that movement of water within the myofibril during freezing removes the overlap effect of solvent water within Raman bands that reflect the peptide backbone conformation. Removal of this overlap effect reveals the changes in secondary structure of proteins in the amide I band as well as removal of the miscellaneous side chain vibrations that contribute to the amide III band (Li-Chan, 1996). Consequently, the Raman spectra of the secondary protein structures become more distinct and this enables a better prediction of shear force values. Previous studies using Raman spectroscopy to measure porcine m. longissimus and beef silverside also produced relatively high R2 values for predicting shear force of 0.77 and 0.75 respectively (Beattie et al., 2004, 2008). However, as with previous studies conducted on lamb, a direct comparison between results reported in literature and models created in the research reported here cannot be made as there is little accordance between experimental design and parameters for Raman spectroscopic measurement. Although all studies aim to predict shear force values measured using a Warner-Bratzler texture analyser with Raman spectra, as with Schmidt et al. (2013) both of these studies (Beattie et al., 2004, 2008) conducted Raman measurements on meat samples that had been subjected to freezing and thawing prior to measurement. Further complicating any direct comparison between prediction models, the number of independent samples measured in these studies was small, chemometric analysis of spectral data varies, spectra have been collected using a bench top Raman device with varying wavelengths and the samples have been measured on a rotating stage using various accumulation and integration times. 4.2. Prediction of shear force using traditional measures Previous studies have used coefficients of determination (R2) to describe the relationship between traditional measures such as sarcomere length, PSA or pHu and shear force (Bouton, Carroll, Fisher, Harris, & Shorthose, 1973; Hood & Tarrant, 1981; Hopkins et al., 2007; Smulders, Marsh, Swartz, Russell, & Hoenecke, 1990). However, it needs to be recognised that as linear correlations are dependent on the range of the data set, they are not a good measure of the merit of the calibration between data sets which have different ranges (Davies & Fearn, 2006). Therefore it is problematic to compare between studies where different experimental designs and treatments have been applied to artificially influence results and highlight interrelationships between the traditional predictors and shear force. For example, Bouton, Harris, and Shorthose (1971) induced a greater range of pHu values (5.6–7.0), while others have studied different muscles (Geesink, Sujang, & Koohmaraie, 2011; Karumendu et al., 2009), species (Smulders et al., 1990) and processing practices including ageing periods and hanging methods (Bouton, Harris, et al., 1973). While informative about the impact of extremes in traditional measures on shear force values, these studies may not be indicative of the range or distribution of values for traditional measures of lamb SM which are typically processed. Consequently, the R2 values previously reported may not be good estimates of the differences which would be found in standard carcases commonly processed (Davies & Fearn, 2006). Relationships between traditional indicators, such as pH and shear force values aren't always linear and may be confounded by interrelationships with other factors (Hopkins et al., 2006). Given this it is not surprising that while the variation in shear force explained by
600
S.M. Fowler et al. / Meat Science 97 (2014) 597–601
Fig. 3. Changes in intensities of background corrected Raman spectra from tender (29–32 N; black) and tough (74–78 N; grey) fresh intact lamb SM.
traditional predictors in this study agrees with some other studies (Hopkins, Toohey, Lamb, Kerr, van de Ven, & Refshauge, 2011), it does not agree with others (Bouton, Carroll, et al., 1973; Hopkins et al., 2006; Karumendu et al., 2009). A comparison between the root mean square error of prediction (RMSEP) to give a more robust explanation of the variation of predictions using different co-variates (Davies & Fearn, 2006) suggested that no significant improvement in prediction models was gained by combining traditional predictors with Raman spectra (Table 2). When this is considered with the accuracy of the prediction models, Raman spectra at 1 day post mortem has the greatest potential as an accurate and precise predictor of tenderness. As with previous Raman studies on meat (Beattie et al., 2004, 2008; Schmidt et al., 2013), it is not possible to assign changes in the Raman spectra of tough and tender lamb with certainty, although a tentative interpretation of spectra is plausible based on the relationship between Raman spectra and the composition and structure of proteins. Spectra collected in this data set (Fig. 3.) exhibit the typical Raman signal for muscle tissue with peaks in intensity representing key chemical bond vibrations of amino acids including tryptophan, tyrosine and phenylalanine, as well as those of protein backbone conformations and secondary protein structures (Pézolet, Pigeon, Ménard, & Caillé, 1988). The largest changes in intensities between spectra from tough and tender SM relate to the tyrosine doublet at wavelengths 826 cm− 1 and 853 cm−1 and the α-helix peak at 930 cm−1 (Fig. 3). These are similar to those found by Schmidt et al. (2013) who reported that the tyrosine doublet weakened in tough lamb and therefore the presence of this doublet signal could be used to classify tender lamb. The results of Beattie et al. (2008) support this, suggesting that shifts in tyrosine bands are a direct reflection of proteolysis given the amount of free tyrosine increases with ageing in normal quality pork. However, Fig. 3 also indicates that there is a difference in the intensity ratio between the two peaks that combine to create the tyrosine doublet. In the tender SM samples, the tyrosine doublet peak intensities at 826 cm− 1 to 853 cm− 1 have a ratio of 0.56:0.61 counts per second (c/s) while in tough SM samples the intensities have a ratio of 0.49:0.48 c/s. This equates to a 12% difference in the peak height at 853 cm−1 between tough and tender samples, as there was a 9% increase for tender samples and a 3% decrease for tough SM. While tyrosine is an amino acid and therefore present in many proteins, it is known that the relative loss of intensity of tyrosine at approximately 830 cm−1 is indicative of the environment of hydrogen bonds in the aromatic side chain of tyrosine (Bonnier et al., 2011). As the phenolic hydroxyl group binds tightly
when oxygen is the donor, but relatively loosely when hydrogen serves this function (Nelson & Cox, 2008), a change in what functions as the donor in the phenyl group of tyrosine may be indicative of the hydrophobicity of the proteins as well as their ability to resist oxidation and reduction (Nelson & Cox, 2008). Consequently, changes in peak intensities may indicate that the tough SM contain a greater number of tyrosine moieties with hydroxyl groups bound to oxygen, therefore reducing the ability of proteolysis to cleave these amino acids in myofibrillar proteins. Larger disparities between intensities exist at the wavelength range between 870 and 1122 cm−1, which represents a combination of bands including tryptophan (Trp), the C\N stretch and C\C stretch vibrations (Krimm & Bandekar, 1986). Previous Raman protein studies have identified the importance of the C\C stretch vibrations at approximately 930 cm−1 as they represent the α-helical secondary structures of proteins (Herrero, 2008b). Therefore, Fig. 3 suggests a decrease in the number of α-helical protein structures in tough lamb SM. When secondary protein structures convert from α-helices into β-sheets or random coils due to denaturation the intensity of the Raman signal corresponding to α-helices at approximately 930 cm− 1 becomes broader and weaker (Tu, 1986), which could contribute to the loss of intensity peak and symmetry in spectra of tough SM. This agrees with Beattie et al. (2004) who suggested that tough meat is characterised by a larger number of β-sheets in comparison to tender meat. However, the presence of the α-helical protein backbone structures is dependent on sample orientation. Therefore inducing polarised changes parallel to the bonds results in a different excitation pattern in contrast to changes induced perpendicular to the bonds (Pézolet et al., 1988). Consequently, by placing samples on a rotating stage, Beattie et al. (2008) and Beattie et al. (2004) would have artificially reduced the contribution of α-helices to the spectra by continually changing the angle at which the laser enters the sample relative to the fibre orientation and reducing the time during which α-helical segments would be polarised. As muscle is a closed system, the attributes of muscle which determine the effects of proteolysis and the subsequent amount of myofibrillar degradation during ageing are predetermined by the biophysical and biochemical properties of the sample at processing. It is hypothesised that this change in myofibrillar structure weakens the Raman signal when spectra are measured at 5 days. If changes to ionic strength during ageing are affecting the ability of Raman to predict shear force it is hypothesised that collecting spectra as the muscle enters rigor would improve the prediction, as ionic strength of sarcoplasmic fluid rapidly rises as the muscle enters rigor (Ouali, 1992).
S.M. Fowler et al. / Meat Science 97 (2014) 597–601
5. Conclusion Overall it is difficult to determine the ability of Raman spectroscopy to predict shear force values of intact lamb, as there is currently no opportunity to compare these results with other studies using the same sample handling and experimental design. Therefore, the accuracy and precision of these predictions need to be validated. However, this study does indicate that Raman spectra may be a better indicator of variation in shear force compared to the traditional measures of sarcomere length, cooking loss, pHu and PSA. Direct comparison of the Raman spectra demonstrated that the discrimination between tough and tender fresh intact SM can be made using the intensity of spectral peaks that correspond to the tyrosine doublet at 826 and 853 cm−1 and αhelix at 930 cm−1. The impact of early post mortem events and collagen on the prediction of shear force using Raman spectroscopy is yet to be determined and the next phase of work will need to address these issues. Acknowledgements This work has been financially supported by the Australian Meat Processor Corporation (AMPC) and Meat and Livestock Australia (MLA), as is the senior author by scholarship. The authors also acknowledge the contribution of Matt Kerr, Tracy Lamb and Kristy Bailes (NSW DPI) who assisted during the measurement of the samples. References Anonymous (2005). Handbook of Australian meat (7th ed.). Brisbane, Australia: AUSMEAT Limited. Beattie, J. R., Bell, S. E. J., Borggaard, C., & Moss, B. W. (2008). Preliminary investigations on the effects of ageing and cooking on the Raman spectra of porcine longissimus dorsi. Meat Science, 80, 1205–1211. Beattie, J. R., Bell, S. J., Farmer, L. J., Moss, B. W., & Patterson, D. (2004). Preliminary investigation of the application of Raman spectroscopy to the prediction of the sensory quality of beef silverside. Meat Science, 66, 903–913. Bonnier, F., Mehmood, A., Knief, P., Meade, A.D., Hornebeck, W., Lambkin, H., Flynn, H., McDonagh, V., Healy, C., Lee, T. C., Lyng, F. M., & Byrne, H. J. (2011). In vitro analysis of immersed human tissues by Raman microspectroscopy. Journal of Raman Spectroscopy, 42, 888–896. Bouton, P. E., Carroll, F. D., Fisher, A. L., Harris, V., & Shorthose, W. R. (1973). Effect of altering ultimate pH on bovine muscle tenderness. Journal of Food Science, 38, 816–822. Bouton, P. E., Harris, P. V., & Shorthose, W. R. (1971). Effect of ultimate pH upon the water holding capacity and tenderness of mutton. Journal of Food Science, 36, 435–439. Bouton, P. E., Harris, P. V., Shorthose, W. R., & Baxter, R. I. (1973). Comparison of the effects of aging, conditioning and skeletal restraint on the tenderness of mutton. Journal of Food Science, 38. Damez, J. -L., & Clerjon, S. (2008). Meat quality assessment using biophysical methods related to meat structure. Meat Science, 80, 132–149. Davies, A.M., & Fearn, T. (2006). Back to basics: Calibration statistics. Spectroscopy Europe, 18, 31–32. Dransfield, E., Etherington, D. J., & Taylor, M.A. J. (1992). Modelling post-mortem tenderisation—II: Enzyme changes during storage of electrically stimulated and non-stimulated beef. Meat Science, 31, 75–84. Geesink, G. H., Sujang, S., & Koohmaraie, M. (2011). Tenderness of pre- and post rigor lamb longissimus muscle. Meat Science, 88, 723–726. Herrero, A.M. (2008). Raman spectroscopy a promising technique for quality assessment of meat and fish: A review. Food Chemistry, 107, 1642–1651. Herrero, A.M. (2008). Raman spectroscopy for monitoring protein structure in muscle food systems. Critical Reviews in Food Science and Nutrition, 48, 512–523.
601
Hood, D. E., & Tarrant, P. V. (1981). The problem of dark-cutting in beef. The Netherlands: Martinus Nijhoff Publishers. Hopkins, D. L., & Geesink, G. H. (2009). Protein degradation post mortem and tenderisation. In M. Du, & R. McCormick (Eds.), Applied muscle biology and meat science (pp. 149–173). USA: CRC Press, Taylor & Francis Group. Hopkins, D. L., Hegarty, R. S., Walker, P. J., & Pethick, D. W. (2006). Relationship between animal age, intramuscular fat, cooking loss, pH, shear force and eating quality of aged meat from young sheep. Australian Journal of Experimental Agriculture, 46, 879–884. Hopkins, D. L., Kerr, M. J., Kerr, M. G., & van de Ven, R. (2012). Within sample variance for shear force testing of lamb meat. Proceedings of the 29th Biennial Conference of the Australian Society of, Animal Production (pp. 31). Hopkins, D. L., Stanley, D. F., Toohey, E. S., Gardner, G. E., Pethick, D. W., & van de Ven, R. (2007). Sire and growth path effects on sheep meat production 2. Meat and eating quality. Australian Journal of Experimental Agriculture, 47, 1219–1228. Hopkins, D. L., Toohey, E. S., Kerr, M. J., & van de Ven, R. (2011). Comparison of two instruments (G2 Tenderometer and a Lloyd Texture analyser) for measuring the shear force of cooked meat. Animal Production Science, 51, 71–76. Hopkins, D. L., Toohey, E. S., Lamb, T. A., Kerr, M. J., van de Ven, R., & Refshauge, G. (2011). Explaining the variation in the shear force of lamb meat using sarcomere length, the rate of rigor onset and pH. Meat Science, 88, 794–796. Karumendu, L. U., van de Ven, R., Kerr, M. J., Lanza, M., & Hopkins, D. L. (2009). Particle size analysis of lamb meat: Effect of homogenization speed, comparison with myofibrillar fragmentation index and its relationship with shear force. Meat Science, 82, 425–431. Krimm, S., & Bandekar, J. (1986). Vibrational spectroscopy and conformation of peptides, polypeptides, and proteins. In C. B. Anfinsen, J. T. Edsall, & M. F. Richards (Eds.), Advances in protein chemistry (pp. 181–364). : Academic Press. Li-Chan, E. C. Y. (1996). The applications of Raman spectroscopy in food science. Trends in Food Science & Technology, 7, 361–370. Li-Chan, E. C. Y., Nakai, S., & Hirotsuka, M. (1994). Raman spectroscopy as a probe of protein structure in food systems. In R. Y. Yada, R. L. Jackman, & J. L. Smith (Eds.), Protein structure–function relationships in food (pp. 163–198). NZ: Blackie Academic and Professional Glasgow. Martin, K. M., Hopkins, D. L., Gardner, G. E., & Thompson, J. M. (2006). Effects of stimulation on tenderness of lamb with a focus on protein degradation. Australian Journal of Experimental Agriculture, 46, 891–896. Muela, E., Sañudo, C., Campo, M. M., Medel, I., & Beltrán, J. A. (2010). Effect of freezing method and frozen storage duration on instrumental quality of lamb throughout display. Meat Science, 84, 662–669. Nelson, D. L., & Cox, M. M. (2008). Principles of biochemistry (5th ed.). New York, USA: Freeman and Company. Ouali, A. (1992). Proteolytic and physicochemical mechanisms involved in meat texture development. Biochimie, 74, 251–265. Pézolet, M., Pigeon, M., Ménard, D., & Caillé, J. P. (1988). Raman spectroscopy of cytoplasmic muscle fiber proteins. Orientational order. Biophysical Journal, 53, 319–325. R Core Team (2013). R: A language and environment for statistical computing. (Vienna, Austria): R Foundation for Statistical Computing. Schmidt, H., Scheier, R., & Hopkins, D. L. (2013). Preliminary investigation on the relationship of Raman spectra of sheep meat with shear force and cooking loss. Meat Science, 93, 138–143. Schmidt, H., Sowoidnich, K., & Kronfeldt, H. D. (2010). A prototype hand-held raman sensor for the in situ characterization of meat quality. Applied Spectroscopy, 64, 888–894. Sikorski, Z. E. (1978). Protein changes in muscle foods due to freezing and frozen storage. International Journal of Refrigeration, 1, 173–180. Smulders, F. J. M., Marsh, B. B., Swartz, D. R., Russell, R. L., & Hoenecke, M. E. (1990). Beef tenderness and sarcomere length. Meat Science, 28, 349–363. Thompson, J. M., Hopkins, D. L., D'Souza, D. N., Walker, P. J., Baud, S. R., & Pethick, D. W. (2005). The impact of processing on sensory and objective measurements of sheep meat eating quality. Australian Journal of Experimental Agriculture, 45, 561–573. Tu, A. T. (1986). Peptide backbone conformation and microenvironment of protein side chain. In R. J. H. Clark, & R. E. Hester (Eds.), Spectroscopy of biological systems. : John Wiley and Sons Ltd. Wood, J.D., Enser, M., Fisher, A. V., Nute, G. R., Sheard, P. R., Richardson, R. I., Hughes, S. I., & Whittington, F. M. (2008). Fat deposition, fatty acid composition and meat quality: A review. Meat Science, 78, 343–358.