Naturally occurring monomeric κ-casein and αs2-casein in bovine milk were purified by ion-exchange chromatography in order to localize potential intrachain disulphide bridges. Enzymic cleavage of the proteins followed by mass spectrometry and amino acid sequence analysis of cystine-containing peptides revealed the presence of an intrachain disulphide bond in both proteins.

The non-protein nitrogen (NPN) of samples of Parmigiano-Reggiano cheese ripened for 6 and 15 months was fractionated by ion-exchange chromatography on a Cu2+-Chelex column to separate oligopeptides from free amino acids. Peptide components were isolated by reversed-phase HPLC and identified by fast atom bombardment–mass spectrometry (FAB–MS). Only the NPN fraction of 6 month old cheese samples contained enough peptides to be further characterized. On the basis of FAB–MS spectral results, 39 oligopeptides were identified, the main components being phosphopeptides. Two sets of both intact and partly dephosphorylated peptides, accounting for a total of 19 phosphopeptides, were formed by the hydrolysis of β–casein and belonged to regions 1–20 and 6–28 of β–casein. The formation and potential role of these peptides in cheese is discussed.

Samples of lactic coagulum were obtained from various depths in a lactic casein silo. The casein content at the base of the silo was in the range 38–62 g/kg compared with 25–30 g/kg in the bulk of the silo. This layer of higher casein concentration, called heavy coagulum, was up to l m deep and appeared long before the bulk milk had fully coagulated. It had a lower pH than the bulk, e.g. pH 4·35 when the bulk pH was 5·2. The heavy coagulum layer was removed from the silo last during pump-out. A laboratory coagulation tube was designed to permit more intensive study of heavy coagulum formation. Starter bacteria levels in the heavy coagulum were 10–160 times those in the bulk. Heavy coagulum had a visibly different curd structure (a much more granular gel), produced differently shaped extrusion curves, and had a higher gel strength than coagulum from the bulk. The heavy coagulum phenomenon had many similarities to minor sludge formation in Cottage cheese manufacture.

Heat-induced gelation, an important functional property of β-lactoglobulin, was studied by measuring the rheological properties of both the A and B variants of the protein during and after heat treatment within a range of pH, temperature and concentration. Gel electrophoresis was used to determine the extent of denaturation and disulphide bond crosslinking of some samples. Both variants formed gel networks on heating at temperatures > 75 °C, and under most conditions the storage modulus (G′) of βlactoglobulin A gels was higher than the G′ of β-lactoglobulin B gels, in particular after cooling to 25 °C. A minimum protein concentration of 50 g/1 was required for gel formation at pH 7·0 in 0·1 M-NaCl by both variants at 80 °C. Increasing the protein concentration above 50 g/1 increased G′, the extent of increase being much greater for the A variant than the B variant. G′ of variant A gels was not much influenced by pH whereas G′ of variant B gels decreased slightly from pH 3 to pH 6 and increased between pH 6 and pH 9. When mixtures of the two variants were gelled G′ increased at the temperature of heating (80 °C) and after cooling (25 °C) as the relative quantity of variant A was increased. Comparisons of the loss of discrete protein bands from electrophoretic gels (native-PAGE, SDS-PAGE and SDS-PAGE of reduced samples) showed that heating β-lactoglobulin solutions of 100 g/1 at pH 7 in 0·1 M-NaCl and at 75, 80 and 85 °C caused a faster loss of both native and SDS-soluble β-lactoglobulin A than of β-lactoglobulin B. It was concluded that the loss of native β-lactoglobulin structure from these solutions during heating was faster than the formation of disulphidelinked aggregates, which was faster than gel formation for both β-lactoglobulin A and β-lactoglobulin B, and that each of these reactions was faster for β-lactoglobulin A than for β-lactoglobulin B. This contrasts with conclusions drawn from some previous studies and may arise from the differences in protein concentration between the present study (∼ 100 g/1) and the previous ones (< ∼ 10 g/1).

The percolation model of Steventon et al. (1991) successfully detected changes in the mechanism of whey protein concentrate gelation with pH and temperature by comparing simulated with experimental gelation times. The results demonstrated that at 71 °C and pH 5·2 the formation of an aggregate from two denatured protein molecules (initiation) was the rate-determining step in the gelation process, while at pH 7·0 the addition of denatured protein molecules to an aggregate (propagation) was rate-determining. At pH 5·9, the gelation process was also initiation limited, with the rate being slower than for pH 5·2 solutions, probably owing to electrostatic effects. Analysis of the temperature dependence of the gelation time, and percolation analysis both showed that there was a change in the rate-controlling reaction at 73 °C for gelation at pH 5·2 and 7·0. In the case of pH 7·0 gelation, this change in rate-controlling reaction was not due to a change from denaturation- to aggregation-controlled gelation, but was probably due to a change in the relative rates of interactions between protein molecules. Gelation at pH 5·2 was aggregation-controlled at temperatures below 73 °C, and denaturation-controlled at higher temperatures; there appeared to be another change in rate-limiting reaction at 80–85 °C without a change in mechanism (i.e. it remained denaturation-limited). The activation energies of the rate-limiting reactions determined from analysis of the temperature dependence of gelation time and from percolation analysis were in agreement. This is evidence that the changes in the rate-controlling reaction of whey protein concentrate gelation with temperature and pH were real.

Two dimensional NMR spectroscopy was used to study the structure of a peptide composed of the N-terminal 42 amino acid residues of β-casein. The peptide was obtained by enzymic cleavage using endoproteinase Asp-N. Complete sequence-specific 1H NMR assignment was performed for the peptide at three Ca2+ concentrations (0, 22 and 37 mM). The NMR results show that the peptide was highly flexible and adopted multiple conformations. No stable secondary structures were present; however, the peptide had some regions with non-random structure. The region between residues Leu16 and Asn27 adopted conformations with an increased contribution of α-helical structure, a so-called nascent helix. Two regions, Glu11-SerP15 and Lys29-Phe33 showed an increased population of conformations with extended structures. Addition of Ca2+ induced chemical shift changes for the backbone amide protons, especially around the phosphoserine region and around the suggested α-helical structure, indicating that the addition of Ca2+ stabilized the structure already present in the apo form of the peptide.

A simple and rapid method for determination of retinol in whole milk is described. Milk (400 mg) was diluted with a water–methanol–ethanol mixture (55:9:36 by vol.) and saponified with potassium hydroxide in a culture tube. Retinol was extracted into 10ml heptane-di-isopropyl ether (3:1) of which 100 μl was injected into the HPLC. Fluorescence detection ensured the specificity and sensitivity. The critical points in the analysis were the extraction of retinol from fatty samples and the instability of retinol towards photo-oxidation and air oxidation in alkaline solvents. There was an average recovery of 98·5 ± 2·4% and a CV of 3·6%.

The incorporation of serum proteins into the synthetic fat globule membrane in milks that had been heated and subsequently homogenized has been quantitatively determined using electrophoresis and densitometry. The results indicated that the amounts of serum proteins present in the fat globule membrane were smaller in milk that was heated before homogenization than in milk that had been homogenized and then heated. The thickness of the adsorbed layer of protein on the surfaces of the fat particles, measured using photon correlation spectroscopy in dissociating and non-dissociating buffer solutions, was also lower in the former case (∼ 20 nm) compared with the latter (∼ 34 nm). The kinetics of adsorption of both β–lactoglobulin and α-lactalbumin in heated homogenized milk on the vacant ‘sites’ on the synthetic fat globule membrane followed similar trends at 75 and 90 °C.

The odour impact compounds of raw, pasteurized and UHT bovine milk were investigated using vacuum extraction and extract dilution sniffing analysis using CharmAnalysis™. Fifteen odour peaks with Charm values between 10 and 3443 were detected. Of these peaks, twelve were identified as hexanal, ethyl butanoate, 2-heptanone, heptanal, dimethyl sulphone, l-octen-3-ol, ethyl hexanoate, 2-nonanone, nonanal, benzothiazole, 2-undecanone, indole and one as a mixture of 2-tridecanone and δ-decalactone. Dimethyl sulphone, indole and one unknown compound (retention index 1154) were the only ones detected as odour impact compounds in all three types of milk. Ethyl butanoate and ethyl hexanoate were found as powerful odorants only in raw milk. A further predominant odorant for this milk was dimethyl sulphone, which was the most important odorant in pasteurized milk. 2-Hexanone and 2-nonanone were identified as the most intense volatile flavour compounds of UHT milk. The powerful odorants described can be used as indicators of the aroma quality of heated milk.

Eight strains of Geotrichum candidum isolated from cheese were compared for aroma production. Twenty-four components, including alcohols, methyl ketones, fatty acids and esters, were identified by gas–liquid chromatography and gas chromatography–mass spectrometry of the volatile compounds produced by G. candidum cultures. Strains could be differentiated by their ability to produce dimethyldisulphide, phenol, phenylethanol and methyl ketones.

The arsenic content of cows' milk consumed in Spain was determined. A procedure using dry ashing hydride generation atomic absorption spectrometry was developed for the purpose. Experimental conditions for the destruction of organic matter were established and an interference study was carried out. The method developed has a detection limit of 0·1 ng/g, a relative SD of 5%, and a recovery (mean ± CSI) of 100 ± 4%. The accuracy of the method was checked by analysis of a certified sample of IAEA milk powder (mean ± CSI: certified, 4·85 ± 0·32 ng/g; found, 5·04 ± 0·81 ng/g). The proposed procedure was used to analyse cows' milk samples. The extremely low arsenic levels found do not present toxicological problems.

The role of interfacial protein in determining the heat stability of recombined milk was investigated by removing serum protein prior to homogenization and reincorporating it after homogenization. In addition, the surface protein composition of recombined fat globules was probed by analyses of protein load and by quantification of the individual surface protein components using FPLC. In the absence of serum protein, substantially more casein was bound to the fat surface during homogenization. Despite this, the detrimental effect of homogenization on heat stability did not occur when serum protein had been removed from the system. Reincorporation of serum protein after homogenization caused the heat coagulation time–pH profile to revert to a form very similar to that observed without removing serum protein from the system. Thus, adsorption of serum protein did not affect heat stability. It is more likely that heat-induced interactions of serum protein with surface-adsorbed casein promoted heat coagulation. Fat surface area rather than casein load affected these interfacial protein-protein interactions during heating.

An inhibition ELISA (enzyme-linked immunosorbent assay) was developed for the determination of caseinomacropeptide (CMP) in order to estimate the proteolysis of κ-casein due to the enzymes of psychrotrophic bacteria in bulk raw milk. The CMP present in milk was quantified specifically by an antibody. The limit of detection was ∽ 0·1 μg/ml and the CV was < 10%. This method was used to study the proteolytic activity of three strains of psychrotrophic Pseudomonas fluorescens in raw milk and to analyse different raw milk samples supplied by four dairy plants. The proteolytic activity for different strains of psychrotrophs and for different milk samples varied considerably, but no correlation was established between the level of microbial flora and κ-casein proteolysis. It is thus not possible to determine the extent of proteolysis from the bacterial count alone. However, by CMP determination in bulk raw milk samples after 6 d storage at 4°C, the mean κ-casein proteolysis was ∽ 4%. Among the milk samples analysed that contained < 107 cfu psychrotrophs/ml, 30% exhibited a proteolysis of κ-casein < 0·5%, i.e. < 5μg CMP/ml.

Trans fatty acids (TFA) formed during biohydrogenation by ruminant animals were quantified in cows' milk fat by means of Fourier transform i.r. (FTIR) spectroscopy. When samples were analysed as the derived methyl esters, a spectral subtraction technique resulted in TFA values that were not biased by the unspecific absorption of intact triacylglycerols or influenced by the shape of the trans band. Austrian milk fat samples contained 20–50 g TFA/kg. Results obtained by analysis of TFA as triacylglycerols compared favourably with those obtained by using methyl esters. The findings of the FTIR method were further substantiated by applying argentation chromatography in combination with capillary gas chromatography to the same set of samples. Besides trans monoenoic acids, non-methylene interrupted cis, trans octadecadienoic acids contributed significantly to the total trans content. Two novel TFA, trans-11, cis-15 octadecadienoic and cis-9, trans-13 octadecadienoic acids, were identified in milk fat by means of their retention behaviour on argentation chromatography and gas chromatography as well as by an oxidative degradation procedure.

Autolysis of and proteolysis by various Lactococcus lactis subsp. cremoris strains were monitored in cheese ‘juice’ extracted by hydraulic pressure up to 63 d ripening. Viability was lowest for strain AM2 (non-bitter), intermediate for strain HP (bitter) and highest for the defined mixed strains G11/C25 (non-bitter). Autolysis monitored by the levels of the intracellular marker enzymes lactate dehydrogenase (EC 1.1.1.27), glucose-6-phosphate dehydrogenase (EC 1.1.1.49) and post-proline dipeptidyl aminopeptidase proceeded in the order AM2 > G11/C25 > HP. Differences in autolysis between strains did not appear to be due to differences in stabilities of the marker enzymes, populations of non-starter lactic acid bacteria or levels of the marker enzymes in the strains. Proteolysis, as measured by gel permeation FPLC and free amino acid analysis of the cheese juice was highest for AM2, intermediate for G11/C25 and lowest for HP. The results of this study provided some evidence that different Lactococcus strains used for cheesemaking had different autolytic patterns during ripening, the effects of which on ripening and flavour development have not yet been clearly demonstrated.

A sandwich ELISA (enzyme-linked immunosorbent assay) of the two-site type has been successfully developed for the detection of cows' milk in goats' or ewes' milk. The assay uses two monoclonal antibodies (MAb) raised in mice against cows' β-lactoglobulin (β-lg). These MAb recognize different epitopes of the β-lg, which are sufficiently distinct to allow simultaneous binding of the corresponding antibodies. One of the MAb recognizes a species-specific epitope of the bovine β-lg and was adsorbed to a plastic microtitration plate (capture antibody). The second MAb was labelled with peroxidase and used to detect the captured cows' β-lg. Factors affecting assay performance were investigated. The optimized assay is highly specific, reproducible (intra- and inter-assay CV were 8 and 13% respectively) and sensitive: as little as 5 ng β-lg/ml or 1 part cows' milk per 100000 parts goats' or ewes' milk can be detected. The technique is robust, cheap, rapid, reliable and suitable for high sample throughput, semi-automation and screening surveys. The MAb used guarantee the high specificity of the assay and indefinite reagent supply of constant quality once approved by collaborative national or international trials.

The ability of Mozzarella cheese to stretch and form long fibres was evaluated by two simple instrumental tests. The first test, which was basically a tensile test, resembled the way the consumer assesses the stretchability of cheese as a pizza topping. The extensibility measurements of the test correlated well with the sensory stringiness scores. The second, fundamental test was a compressive elongation of the melted cheese. The elongational viscosity of the melted cheese was calculated; this could be used as an index of the ability of the cheese to form strings when it is stretched.

Untreated whole milk, skim milk, sweetened milks, sweetened and unsweetened creams and ice-cream mixes were preincubated with a culture of Enterococcus faecium, then subjected to heat treatment in a pilot-scale plate heat exchanger using a hold of 15 s at 76°C. The liquids were examined for γ-glutamyltranspeptidase (GGTP) activity and total streptococcal count before and after heat treatment and the results plotted against water activity (aw). There was a good correlation between reduction in GGTP activity, destruction of streptococci and aw, demonstrating the potential of the enzymic assay for assessing the severity of HTST heat treatments above the minimum for pasteurization.

A simple and rapid competitive enzyme-linked immunosorbent assay method has been developed to detect the residual activity of chymosin in various types of cheese without extraction. In a cheese of the Saint-Paulin type, concentrations of chymosin detected at the end of cheesemaking in curd and whey represented 22·5 and 73·5% respectively of the total amount added to milk, so most of the chymosin was detected using this method. In a cooked pressed cheese of the Comté type, only 14% of the chymosin added to milk was detected in curd and all the chymosin in whey was inactivated. It was thus possible to demonstrate the presence of active residual chymosin in cooked pressed cheese. When measurements were carried out in 12 different cheeses, chymosin concentrations ranged between 384 ng/g in a Camembert cheese and 86 ng/g for an 8-month old Comté cheese. Chymosin appeared to be highly stable during cheese ripening.

A total of 529 bacterial strains have been isolated from milk and cream sampled at different sites in a dairy production plant under conditions selective for aerobic sporeforming bacteria. Identification with classical methods based on morphological, physiological and biochemical criteria showed Bacillus licheniformis to be the most frequently occurring Bacillus sp. The investigation also revealed 62 unidentified strains. Classical identification methods were time consuming (3–7d), lacked specificity and —because of their dependence on phenotypic gene expression—sometimes produced ambiguous results. Consequently, a colony hybridization method developed for the identification of B. licheniformis strains and using non-radioactive labelled 23S rRNA targeted oligonucleotide probes was also used. Identification of B. licheniformis with borth classical and hybridization methods revealed diverging identification results for 70 strains.