Application of advanced light microscopic techniques to gain deeper insights into cheese matrix physico-chemistry
- 669 Downloads
Despite considerable advances in confocal microscopy and related technologies, advanced light microscopy techniques have been rarely used to study the microstructure of cheese matrices. Here, we demonstrate the potential of advanced microscopic techniques, namely fluorescence lifetime imaging (FLIM), confocal Raman spectroscopy and non-linear microscopy (two-photon excitation and second harmonic generation microscopy), to study localized physicochemical properties of cheese matrices. These methods allow for precise localization of individual chemical components and help to determine their spatial organization in three dimensions. We discuss both fluorescent labelling and label-free methods that provide valuable information about the localized environment. Overall, these new technologies will enable a greater understanding of the influence of manufacturing processes on cheese quality and consistency.
KeywordsConfocal microscopy Cheese matrix Fluorescence lifetime Second harmonic generation Two-photon excitation Confocal Raman microscopy
Cheese is a complex soft matter composed of a protein matrix entrapping fat, water, minerals and dispersed bacterial colonies and their metabolites. The characteristic structure, texture and flavour properties of cheese result from a sequence of dynamic physicochemical, microbial and biochemical processes starting in the cheese vat during manufacture and continuing throughout ripening and cold storage. The textural and rheological properties of a cheese are dependent on interactions between its structural units, including those between fat globules and the protein matrix, fat globules coated with casein micelles which may interact with cheese matrix, moisture content, pools of free fat, levels of proteolysis and casein matrix bond strength (Everett and Auty 2008; Hickey et al. 2015).
Fat may be entrapped in the cheese matrix as inert filler or as a copolymer within the casein matrix. It can exist in globular form, consisting of droplets ranging in size from 0.2 to 15 μm with an average of 4 μm (Lopez et al. 2007) surrounded by milk fat globular membrane material (MFGM) or it may exist as non-globular free fat. The MFGM is a layer of surface active material which envelops the fat globule and is composed of proteins, enzymes, glycoprotein, triacylglycerols, phospholipids, glycolipids, cholesterol and other minor components (Lopez and Briard-Bion 2007; Hickey et al. 2015). Certain cheese manufacture processes such as high cook temperatures result in coalescence of fat globules along with fracturing of the MFGM during the ensuing pressing of the curds results in the formation of free fat and the release of MFGM material (Lopez and Briard-Bion 2007; Hickey et al. 2015). Amphipolar properties of the phospholipids in the MFGM also enable protein-fat interactions which are essential for matrix structure and enable fat globules to act as emulsifiers within the matrix stabilizing both oil-in-water and water-in-oil emulsions. This is probably responsible for bacterial colonies within the cheese matrix locating at the protein-fat interface in contact with the MFGM due to the hydrophobic nature of LAB bacteria (Banks 1991; Lopez et al. 2006; Pereira et al. 2009; Gallier et al. 2010).
Calcium and phosphorus are the key mineral components within cheese matrices and participate in the protein network that forms through milk coagulation and have a major effect on the structure and texture of cheese (Upreti and Metzger 2006; Lucey and Fox 2015). Increased levels of acidification during cheese manufacture results in conversion of colloidal calcium phosphate to the soluble form which is lost in the whey (Sheehan and Guinee 2004). Excess calcium in the cheese post manufacture can lead to the development of calcium lactate crystals (Upreti and Metzger 2006).
Starter bacteria are integral to cheese manufacture and ripening. Their primary role is to acidify milk during the initial fermentation and then during ripening to lyse thereby releasing protease and lipase enzymes into the cheese matrix resulting in the breakdown of peptides, amino acids and fatty acids leading to the development of volatile flavour compounds (Hickey et al. 2015). The work of Jeanson et al. (2011) showed bacterial colonies to be randomly distributed and static within the cheese matrix, while other studies (Laloy et al. 1996; Lopez et al. 2006) showed bacteria to be located in direct contact with the MFGM or located at the casein-fat interface. During ripening, residual sugars such as lactose and galactose provide substrate for the growth of non-starter lactic acid bacteria (NSLAB), while metabolites of starter fermentation such as lactate enable growth of secondary/adjunct cultures such as propionic acid bacteria.
Given the complexity of cheese as a physical and biological system and considering its manufacture from a raw material which can vary markedly in composition, there is a continued need to understand and control the factors that influence cheese quality and consistency. Traditionally, research in the area of cheese microstructure has relied on scanning or transmission electron microscopy and basic confocal microscopy of fluorescently labelled cheese samples. Scanning electron microscopy (SEM) provides detailed topographic images of the cheese surface (or fracture surfaces in the case of freeze-fracture cryo-SEM), differentiating fats, proteins salt crystals and even individual bacterial cells and colonies. However, standard electron microscopy cannot provide complete 3D information about spatial arrangement of individual cheese components. On the other hand, confocal laser scanning microscopy (CLSM), in conjunction with proper fluorescence staining, can provide three-dimensional visualization of fat and protein components or separation of dead and live bacteria inside the cheese matrix (Auty et al. 2001; El-Bakry and Sheehan 2014).
When detailed information about local chemical properties or spatial organization of individual cheese components is required, e.g. individual lipid constituents (triglycerides, phospholipids), distribution of calcium lactate crystals, local pH, or distribution of trace components, advanced microscopic techniques prove themselves to be valuable tools in food research. Thus, with recent development in materials sciences including advanced microscopic techniques, it is now timely to apply these techniques to gain a deeper understanding of cheese matrices and their role in cheese quality and consistency. The objective of this research was to evaluate the application of a range of advanced light microscopic techniques in research into the physicochemical properties of cheese matrices.
2 Materials and methods
2.1 Cheese preparation
2.1.1 Starter strains
Thermophilic starter cultures typically used in Swiss-type cheese manufacture (Scott 1981), i.e. S. thermophilus TH3 (ST) and L. helveticus LHB02 (LH), were purchased from Chr. Hansen Ltd. (Little Island, Co. Cork, Ireland) as individual frozen concentrates and stored at -80 °C until cheese manufacture.
2.1.2 Cheese manufacture
Raw milk was obtained from a local dairy company, standardized to a protein to fat ratio of 1.01:1, held overnight at <10 °C, pasteurized at 72 °C for 15 s, and pumped at 32 °C into cylindrical jacketed stainless steel vat (500 L) with automated variable speed cutting and stirring equipment (APV, Schweiz AG, Worb, Switzerland). Cheesemilk was heated to 34 °C and inoculated with 10 mg.L−1 of frozen direct vat starter (DVS) ST and 5 mg.L−1 of frozen DVS LH (Chr. Hansens Ltd.). After a 60-min ripening period, chymosin (Chymax plus, Chr. Hansens Ltd.) diluted to 1:6 with de-ionized water, was added at a level of 18 mL per 100 kg of milk. A coagulation time of 35 min was allowed prior to cutting of the coagulum. After a 10-min healing period, the curd/whey mixture was cooked by steam injection into the jacket of the vat with constant stirring. Maximum scald employed was 50 °C. Curds were cooked at a rate of 0.5 °C.min from 34 to 45 °C and at 1 °C.min from 45 °C to maximum scald. Curds were pitched at pH 6.3.
The curds were drained, cheddared, milled at pH 5.3, salted at a rate of 1.45 % (w/w) and pressed overnight on a horizontal press at 264.6 kPa in 20 kg blocks. The cheeses were then vacuum packed and ripened at 8 °C for up to 230 days.
2.1.3 Cheese preparation for light microscopy
Thin slices of cheese, measuring approximately 5 × 5 × 3 mm, were freshly cut from a cheese sample using a scalpel.
Neutral lipids were stained with a lipid specific Nile blue fluorescent dye (Sigma-Aldrich, St. Louis, MO), prepared in acetone with concentration of 42 μg.mL−1.
Phospholipids were labelled with N-(lyssamine Rhodamine-B sulfonyl)-1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (Rh-DOPE; Avanti Polar Lipids Inc., Birmingham, England), dispersed in chloroform. To stain phospholipids, about 200 μL of the staining of Rh-DOPE was added to the coverslip. After chloroform evaporation, the coverslip was pressed against the slice of cheese to permit diffusion of the stain for 30 min in the dark at 4 °C.
2.2 Advanced techniques in optical microscopy: basic principles and instrumentation
2.2.1 Confocal laser scanning fluorescence microscopy
Confocal laser scanning fluorescence microscopy (CLSM) has long been a standard tool in cell biology and became popular in other areas of science, including food research. It is based on fluorescence, a phenomenon where a photon of suitable wavelength is absorbed by a specific dye molecule, followed by an emission of a photon with somewhat longer wavelength (lower energy). With a set of fluorescence filters, the distribution of the fluorescent can be observed with high contrast and specificity. The main disadvantage of simple widefield fluorescence observation is the blur and haze caused by out-of-focus fluorescence, particularly in thick samples, a common situation in food research. The elaborate optical arrangement of a CLSM effectively blocks the out-of-focus light, employing laser excitation of a single spot in the sample, fast raster scanning in three dimensions and efficient multichannel detectors (Pawley 2006). With a CLSM, one obtains a full 3D representation of the spatial distribution of fluorescent molecules in the sample with high contrast and resolution.
The CLSM is a basic platform that can be easily extended to perform advanced microscopic imaging, such as two-photon excitation, second harmonic generation, fluorescence lifetime or Raman microscopy.
2.2.2 Fluorescence lifetime imaging microscopy
Fluorescence lifetime imaging microscopy (FLIM) measures the characteristic time a fluorescent molecule spends in the excited state (fluorescence lifetime). This lifetime is typically a few nanoseconds in length and it is specific for given fluorophore. Moreover, the fluorescence lifetime of many fluorescent dyes is a sensitive indicator of local chemical properties, such as ion concentration, pH, oxygen concentration or presence of other fluorescent molecules (Lakowicz 2006). Consequently, FLIM can not only be used to distinguish different fluorophores on the basis of their characteristic lifetimes (rather than their spectral properties) but also to distinguish among different environments based on changes in lifetime of the same fluorophore, for example local pH within the cheese matrix (Burdikova et al. 2015). Fluorescence lifetime microscopy can also provide additional valuable information in fat composition studies when combined with various fluorescent dyes (e.g. Nile blue, Rhodamine-DOPE). Better understanding of the composition of lipid droplets in cheese matrix may improve the functional and nutritional properties of food products (Lopez and Briard-Bion 2007).
Fluorescence lifetime measurements of Nile blue- and Rhodamine-DOPE-stained cheese samples were performed with Zeiss LSM780 NLO confocal microscope based on Axio Observer. Z1 inverted microscope (Carl Zeiss MicroImaging, GmbH) equipped with Chameleon Vision (Coherent, Inc.) pulsed IR laser (0.1 ps pulses, 80 MHz repetition rate) tuned to 860 nm for two-photon fluorescence excitation and Becker & Hickl hybrid detector (HPM-100-40) and TCSPC hardware (SPC-150, Becker & Hickl, GmbH). Fluorescence was collected through an IR-blocking filter and 580–640 nm emission filter with a Plan Apo 63×/1.4 oil IR objective lens. The measured fluorescence decay curves were analyzed with Becker & Hickl SPCImage fitting software to yield the mean fluorescence lifetime in 256 × 256 pixel lifetime images.
2.2.3 Two-photon excitation and second harmonic generation imaging microscopy
Second harmonic generation (SHG) imaging is non-linear optical process with two requirements—high excitation intensity and non-centrosymmetric molecules with long range order present in the sample (Bianchini and Diaspro 2008). No fluorescence takes place in this process and the wavelength of the emitted light (as the name suggests) is exactly one half of the excitation wavelength. Generally, the same pulsed laser source used for two-photon excitation (TPE) is also suitable for SHG, and these two signals can be easily separated spectrally. As no labelling is needed for SHG imaging (Diaspro et al. 2002), it is a vital method for visualizing certain specific components of the sample, e.g. collagen in tissues (Filová et al. 2010; Šepitka et al. 2012). Some low symmetry crystals also emit strong SHG signal (Haupert et al. 2012) so this technique can be applied to detect calcium lactate crystals within cheese matrices.
TPE is a non-linear process where a fluorescent molecule simultaneously absorbs two photons of low energy (long wavelength, typically in the near infrared part of the spectrum) in order to reach an exited state. The dye molecule then emits a photon, whose energy is somewhat lower than the sum of the excitation photon energies (the emission wavelength is longer than one half of the excitation wavelength). However, for this phenomenon to appear, the excitation intensity must be exceptionally high (megawatts per square micrometre) and this is only achievable with dedicated pulsed lasers (pulses of the order of picoseconds or shorter) in a laser point scanning setup (Diaspro et al. 2006). The main advantages of two-photon excitation microscopy are (i) lower absorption and scattering of the IR excitation light in many kinds of samples (e.g. biological tissues), (ii) due to non-linear effects, no fluorescence originates from outside the focus plane (Burdíková et al. 2010), so there is no need to block the out-of-focus light, and this greatly enhances the fluorescence collection efficiency in highly scattering samples. These two effects together facilitate fluorescence microscopy deep within a scattering sample.
SHG signal of calcium lactate crystals, together with TPE fluorescence, were collected with TCS SP8 AOBS MP confocal microscope (Leica Microsystems, GmbH) equipped with InSight DeepSee (Spectra-Physics, Inc.) pulsed IR laser (0.1 ps pulses, 80 MHz repetition rate) tuned to 860 nm. Internal spectral detectors were used for fluorescence acquisition and the frequency doubled signal was observed through 420–440 nm bandpass filter in transmission geometry.
2.2.4 Raman confocal imaging
Raman scattering is a process of inelastic scattering of photons, where part of the photon energy is transferred to molecular vibrations. It is a very weak process (typically less than 1 ppm of incoming photons are inelastically scattered), however, the spectrum of the scattered photons (Raman spectrum) shows a very high level of detail. Every molecule in the sample has its characteristic spectrum, reflecting the energies of vibrational modes of that molecule. With known spectra of individual constituents (often found in specialized databases), one can decompose the measured Raman spectrum (using, e.g. principal component analysis (PCA) (Shinzawa et al. 2009)) into the contributions of the individual chemical species. When combined with a confocal microscopic setup, the method (Hollricher and Ibach 2011) is a powerful tool to determine the chemical composition of the sample with micrometre resolution, in 3D and without any labelling. Here, we show the potential and limitations of this technique in determining the local composition of cheese samples.
A slice of cheese was cut using a sharp knife and the local Raman spectrum was taken using inVia confocal Raman microscope (Renishaw, PLC) with 785 nm excitation (laser power of about 60 mW in the sample plane) and 20× objective. Laser spot is visible in the middle of Fig. 5. Vertical size of the laser spot is about 20 μm.
In order to extract quantitative information from the acquired Raman spectra, the statistical classification of the spectra was carried out adopting a rolling-circle procedure for spectral processing. In this way, the slowly varying spectral background is suppressed. Also automated subtraction of cosmic ray peaks, and the Savitzky–Golay procedure for subsequent noise filtering was employed. Finally, all the data obtained from the studied samples was analyzed using standard principal component analysis (PCA) algorithms in order to distinguish possible difference between them.
2.2.5 Cryo-electron microscopy
Cheese samples were imaged in a scanning electron microscope using an Alto2500 cryo-stage system (Gatan UK Ltd., Abingdon, UK.). Sections of cheese 1 × 1 × 4 mm were cut at room temperature with a razor blade, then plunge frozen in liquid nitrogen slush at −207 °C, followed by freeze fracturing under vacuum in the cryo-chamber using a precooled scalpel blade. The fractured surfaces were then etched at −85 °C for 15 min, cooled to −125 °C, and coated with sputtered platinum (10 mA for 60 s). Samples were transferred under vacuum to the cold stage in a Zeiss Supra 40VP field emission scanning electron microscope (Carl Zeiss AG, Darmstadt, Germany), where they were maintained at −125 °C and imaged using an accelerating voltage of 1 kV.
3 Results and discussion
3.1 Calcium lactate crystals visualized with SHG microscopy
Calcium lactate crystals form when Ca2+ and lactate ions exceed their solubility, supersaturate the serum phase of the cheese and then crystallize at nucleation sites within cheese matrices. Most starter bacteria ferment lactose to l-lactate, however, considerable amounts of d-lactate are formed by the activity of NSLAB during cheese ripening. The lower solubility of Ca-d-lactate then causes crystal formation (Lawrence et al. 2004). Crystallization of calcium lactate on the surface of Cheddar cheese is a common and troublesome defect (Pearce et al. 1973). The deposits are visible to the naked eye; products containing crystal deposits are not aesthetically pleasing and can be sometimes rejected by the retailer/consumer who may confuse them with mould contamination.
Calcium lactate crystals can be visualized using X-ray crystallography but we will miss the chemical information about the placement fat and protein around the crystals. The crystalline deposits are also frequently observed by standard light microscopy SEM or TEM in processed cheese products (Kalab 1995). Using various techniques such as electron microscopy and X-Ray diffraction analysis, infrared spectroscopy and energy dispersive X-Ray spectrometry analysis (EDSA), several crystal types have been identified, e.g. monoclinic calcium pyrophosphate dehydrate, disodium phosphate dodecahydrate, unreacted melting salts, tyrosine, calcium citrate, fatty acids, protein and lactose (Kondo et al. 1990). The composition of crystalline structures in natural cheeses and PCPs was studied by with SEM EDSA and X-Ray diffraction (Washam et al. 1985). Calcium specific fluorescent stain Alizarin Red was used to identify calcium lactate crystals by optical microscopy (Yiu 1985).
In comparison, SHG microscopy does not allow us to chemically characterize the composition of crystals, but this label-free microscopic technique allows us to visualize the calcium lactate crystals inside cheese matrix, and with combination with standard CLSM, we can get also the information about the relationship between crystals and the surrounding environment like location of fat, protein, phospholipids, live/dead bacteria in three dimensions (Online Resource 1). This localization after further investigation can further illustrate the relationship between calcium lactate crystals and bacterial, fat and mineral placement inside the cheese matrix.
3.2 Lipids visualized with FLIM
Currently it is possible to detect phospholipids using spectroscopy, nuclear magnetic resonance spectroscopy, particularly with P-NMR. With HPLC-ELSD, it is possible to determine the composition and type of phospholipids present, but it does not provide information about the placement and location of the phospholipids in within the cheese matrix. Utilization of FLIM methodology provides more accurate information about the placement of phospholipids in the cheese matrix. To achieve separation of different types of fats inside the cheese matrix, we used the same methods as used previously (Lopez et al. 2008) for CLSM.
3.3 Characterization of chemical composition of cheese matrix by confocal Raman microscopy
To demonstrate the capability of Raman spectroscopy for the compositional analysis and the spatial visualization of cheese samples, we have recently performed experiments in which spectra were acquired point-by-point, at a few selected positions on a cut sample surface (Fig. 5). Consequently, some relevant molecular complexes (Fig. 6) could be identified in this way, with some spatial resolution so that e.g. inhomogeneous distribution of beta-carotenoids is visible. As an example, this analytical approach could help to determine the underlying causes of the development of a pink color defect, persistent in a wide range of ripened cheese varieties (Daly et al. 2012).
However, this type of localized investigation for species and/or compound identification is time consuming and often does not collect sufficient or complete (multi-spectra) information about the whole sample, particularly where there is likely to by inhomogeneity within the sample. A possible solution to the problem of localized measurements, to build up a compositional map, has come about with the commercial introduction of line-scan techniques. This approach is around for some time, and has recently been applied to the study of biological samples (Samek et al. 2010). In essence, rather than measuring the sample point-by-point the technique relies on line scanning in which multiple locations are sampled simultaneously. With the many, rapidly acquired spectra—up to tens of thousands of spectra—‘data mining’ for analysis and identification becomes possible, using standard chemometric techniques. This can be also combined with confocal imaging to quickly build up 3D maps of the surface.
Given the desire within the cheese industry to achieve a greater understanding of the influence of manufacture and ripening parameters on cheese quality and consistency, there is a continued interest in exploiting new analytical techniques to gain new insights into the physicochemical characteristics of cheese matrices. In this study, a range of advanced light microscopic techniques were evaluated to determine their potential to provide a better understanding of cheese matrices at the micro-scale and in particular to aid research into determining the level of localized inhomogeneity within the cheese matrix. This study concluded that (i) second harmonic generation microscopy technique facilitates visualization of calcium lactate crystals with reference to the fat and protein components within the cheese matrix; (ii) two-photon excitation fluorescent lifetime microscopic techniques enables a more precise determination and location of fats and phospholipids within the cheese matrix, and (iii) the Raman confocal microscopy offers a new and very powerful methodology to analyze and fingerprint different chemical components of the cheese matrix.
This work was funded by the Dairy Levy Trust, Ireland, with additional support from the Ministry of Education, Youth and Sports of the Czech Republic (LO1212) together with the European Commission (ALISI No. CZ.1.05/2.1.00/01.0017).We also thank Dr. Jana Pala (Leica Microsystem, Mannhein, Germany) for image acquisition on Leica TCS SP8 microscope and the W. M. Keck Center for Cellular Imaging (KCCI) at the University of Virginia for expert advice regarding fluorescence lifetime imaging.
Conflict of interest
Zuzana Burdikova, Zdenek Svindrych, Cian Hickey, Martin G. Wilkinson, Mark A. E. Auty, Ota Samek, Silvia Bernatova, Vladislav Krzyzanek, Ammasi Periasamy, and Jeremiah J. Sheehan declare that they have no conflict of interest.
Statement of human and animal rights
This article does not contain any studies with human or animal subjects performed by any of the authors.
- Banks W (1991) Milk lipids. Bull Int Dairy Fed 260:3–6Google Scholar
- Diaspro A, Fronte P, Raimondo M, et al. (2002) Functional imaging of living Paramecium by means of confocal and two-photon excitation fluorescence microscopy. pp 24–31Google Scholar
- Hollricher O, Ibach W (2011) Confocal Raman microscopy. Springer, BerlinGoogle Scholar
- Lawrence RC, Gilles J, Creamer LK et al (2004) Cheddar cheese and related dry-salted cheese varieties, 3rd edn. Elsevier Academic Press, LondonGoogle Scholar
- Pearce KN, Creamer LK, Gilles J (1973) Calcium lactate deposits on rindless cheddar cheese. N Z J Dairy Sci Technol 8:3–7Google Scholar
- Scott R (1981) Cheesemaking practice. Elsevier Applied Science Publishers, LondonGoogle Scholar
- Washam CJ, Kerr TJ, Hurst VJ, Rigsby WE (1985) A scanning electron microscopy study of crystalline structures on commercial cheese. Dev Ind Microbiol 26:749–761Google Scholar
- Yiu SH (1985) A fluorescence microscopic study of cheese. Food Microstruct 4:99–106Google Scholar