Effects of enzymatic lactose hydrolysis on thermal markers in lactose-free UHT milk

Abstract

Lactose-free products are more susceptible to chemical and physical modifications during heating and storage, due to the release of glucose and galactose during enzymatic processing, both more reactive than lactose. The present study demonstrates the effect of enzymatic lactose hydrolysis on 5-hydroxymethylfurfural (HMF), whey protein nitrogen index (WPNI) and lactulose used as thermal markers for UHT milk process monitoring. Six milk leading brands which provided regular and lactose-free UHT milk were selected, giving a total of 12 UHT milk samples analyzed in authentic duplicates. All lactose-free samples showed high levels of HMF index (42.15 µmol L⁻¹, against 13.11 µmol L⁻¹ for regular samples) and low lactulose contents (13.03 mg 100 mL⁻¹, against 35.59 mg 100 mL⁻¹ of regular ones). High variations in HMF (55–85%) and lactulose (42–91%) intra-brand analysis indicated that both markers are influenced by the lactose hydrolysis process. The paired t test indicated there was no difference among WPNI indexes of regular and lactose-free milks suggesting that this thermal marker is suitable to infer about heat damage in lactose-free dairy matrices.

Electronic supplementary material

The online version of this article (10.1007/s13197-020-04561-9) contains supplementary material, which is available to authorized users.

Introduction

Different compounds modified or produced during thermal processing, known as thermal markers, are used to identify the heating damages (Feinberg et al. 2006). 5-Hydroxymethylfurfural (HMF), for instance, is an intermediate compound of Maillard reaction and is used as a thermal marker for UHT and sterilization processes (Morales et al. 2000). HMF quantities are typically measured by spectrophotometry, based on a colorimetric reaction with thiobarbituric acid (TBA) (Keeney and Bassette 1958). However, this method has low stability and specificity, as TBA can also react with other aldehydes compounds present in milk (Ritota et al. 2017), making it necessary to employ other analytical methods, mainly high performance liquid chromatography (HPLC) (Ferrer et al. 2000; Morales et al. 2000). Nevertheless, the colorimetric method is still widely used due to its simple and cost-effective implementation.

One of the most important parameters used to monitor heat damage is the whey protein nitrogen index (WPNI), as the whey proteins denaturation and aggregation also depend on the temperature and time of heating. The WPNI index represents the amount of heat-undenatured whey protein nitrogen (soluble in saturated sodium chloride solution) expressed in milligrams of WPN per milliliter of milk, and it is commonly determined by tubidimetric method (Ritota et al. 2017). Different studies report the use of HPLC (Feinberg et al. 2006; Lan et al. 2010) and capillary electrophoresis coupled to mass spectrometry (Jones et al. 1998) to evaluate the content and structure of protein fractions, as β-lactoglobulin in heated dairy products. However, these methods have high analytical cost, require skilled labor and show low analytical frequency, which prevents them from being chosen as routine tools in the dairy industry.

Lactulose is a disaccharide formed exclusively during milk heating and international regulatory agencies has suggested it as thermal marker. Lactulose is used to differentiate thermal processes (e.g.: pasteurization, UHT, sterilization) and also to classify the UHT processing (direct or indirect) (Feinberg et al. 2006; Lan et al. 2010). The International Dairy Federation (IDF) establishes a range of lactulose (10–60 mg 100 mL⁻¹) for products submitted to the UHT process (Elliott et al. 2005). Official standard methods include HPLC (ISO 11868/IDF 147 2007) and enzymatic methodologies (ISO 11285/IDF 175 2004), but they are time and reagent consuming what makes them difficult to be implemented as routine assays.

Currently the production of lactose-hydrolyzed milks has increased (Dekker et al. 2019), and they are classified according to their residual lactose content as “lactose-free” or “low-lactose” (EFSA 2010; BRASIL 2017). The industrial process of the lactose hydrolysis is mainly performed through enzymatic degradation by using microbial β-galactosidase. The glucose and galactose residues released during this process might intensify chemical and physical modifications in milk due to their higher reactivity (Messia et al. 2007). Some thermal markers can also be influenced by this intrinsic change in the carbohydrate composition caused by the lactose hydrolysis. In this study, we investigated the influence of lactose hydrolysis on the occurrence of thermal markers used for UHT process monitoring of lactose-free milks by using total HMF and WPNI indexes, and the lactulose content.

Experimental

Reagents and samples

Sodium chloride (NaCl) and hydrochloric acid (HCl) were purchased from Vetec Química Fina (Rio de Janeiro, RJ, Brazil). Milk standards for WPNI modeling were obtained from American Dairy Products Institute (ADPI) (Elmhurst, IL, USA). Trichloroacetic acid (TCA), thiobarbituric acid (TBA), and oxalic acid were obtained from Merck (Darmstadt, HE, Germany). Standards for HMF, lactulose, lactose, 2,6-pyridinedicarboxylic acid (PDC), and cetyltrimethylammonium bromide (CTAB) reagents were purchased from Sigma Aldrich (St. Louis, MO, USA). All the solutions were prepared by using deionized water by reverse osmosis system (Millipore, Bedford, MA, USA).

Six commercial Brazilian brands of UHT milk, with the largest market share in both regular and lactose-free milk were selected. One sample of semi skimmed regular UHT milk and one sample of semi skimmed lactose-free UHT milk of each brand were collected randomly. A total of 12 UHT milk samples were analyzed in authentic duplicates.

Analytical methods

The HMF and WPNI indexes were determined based on spectrophotometric methods originally proposed by Keeney and Bassette (1958) and Kuramoto et al. (1958), respectively. The quantification modeling for HMF and WPNI consisted in 6 concentration levels (3, 6, 9, 12, 15, and 20 µmol L⁻¹) and (7.86, 6.41, 4.97, 3.52, 2.08, and 0.63 mg WPN mL⁻¹), respectively. For total HMF index quantification, all milk samples were submitted to boil during 1 h with 0.3 mol L⁻¹ oxalic acid, followed by protein precipitation by TCA 40% w/v addition and supernatant collection after centrifugation (10,786×g for 10 min). Considering the lactose-free samples are more reactive due to the presence of glucose and galactose in their composition, higher amounts of total HMF were expected, comparing to regular milks. So, the supernatant of all lactose-free samples was diluted with deionized water at 1:5 ratio. For WPNI quantification, the dilution step was not necessary since the same protein content is expected for both regular and lactose-free samples.

Lactose and lactulose levels were quantified by capillary zone electrophoresis with indirect detection in ultraviolet region (CZE-UV) according to our previous work (Neves and de Oliveira 2020). Agilent ChemStation software (Rev. B.04.03) was used for electrophoresis data acquisition and processing. Samples preparation was performed by TCA 40% w/v solution at 1:1 ratio (sample-acid solution) followed by centrifugation (10,786×g) and filtration with 0.45 µm membrane (RC45/13) Macherey–Nagel^® (Bethlehem, PA, USA) before injection. Lactose quantification: samples extract of regular UHT milk were diluted 1:5 (extract volume-final volume), while lactose-free milks extracts had a 1:2 dilution. The dilution step was done to obtain a good electrophoretic signal, since high amounts of lactose is present in sample’s composition and compromise the analyte quantification. The dilution ratio was defined according to the expected average of the lactose content of each matrix (regular and lactose-free). Lactulose quantification: samples extracts were injected without dilution due to their low amounts. The single point standard addition (SPSA) quantification method was used since it is more suitable for high complexity matrices. A representative electrophoretic profile of lactulose quantification in both regular and lactose-free milk samples is given in Online Resource 1, showing the lactose-lactulose isomers separation in regular milk matrices and the lactulose-glucose-galactose separation in lactose-free milk matrices.

Statistical analysis

The statistical assumptions of residues normality, homoscedasticity, lack of fit and regression significance of HMF and WPNI modeling were evaluated using Microsoft Excel^® 14.0 ver. (Microsoft Office, 2010) as well as the statistical analysis of all data.

Results and discussion

Evaluation of statistical modeling for HMF and WPNI indexes

Data independence was achieved by analyzing all authentic replicates at all levels randomly, for both proposed models. According to Table 1, a normal and homoscedastic profile with 95% confidence (p value > 0.05 and C_calculated < C_critical, respectively) was obtained. There was no lack of fit (F_calculated < F_critical) and a high regression significance (F_calculated >F_critical) was achieved, indicating that both models are able to predict HMF and WPNI indexes in milk samples.

Table 1.

Model parameters and statistical assumptions for WPNI and HMF

Parameter	WPNIa (mg WPN mL−1)	HMFb (µmol L−1)
Slope	− 4.41 (± 0.23)	0.00996 (± 0.0001)
Intercept	93.92 (± 1.17)	0.00091 (± 0.0018)
LOFc	0.30 (3.26)*	0.77 (3.26)*
SIGd	343 (4.49)*	4138 (4.49)*
R2 adjusted	0.95	0.99

Residual normality (Shapiro–Wilk test): ^ap = 0.063; ^bp = 0.915. Homoscedasticity (Cochran test): ^aC_calc = 0.11 (C_crit = 0.62); ^bC_calc = 0.21 (C_crit = 0.62)

*Critical values in parentheses, where: ^cLOF (lack of fit): F_crit (ν₁ = 4, ν₂ = 12); ^dSIG (regression significance): F_crit (ν₁ = 1, ν₂ = 16). Where ν₁: numerator degree of freedom (LOF: m − p; SIG: p − 1) and ν₂: denominator degree of freedom (LOF: n − m; SIG: n − p), where p: number of parameters, m: number of levels, n: number of observations. 95% confidence interval

Thermal markers quantification

Table 2 provides the quantitative data, nutritional composition according to product’s labeling and paired t test using the mean values of thermal markers of both regular and lactose-free samples of all brands, at 95% of confidence. Additionally, scattering plots were built to verify the relationship between the thermal markers and the quantified lactose content (Fig. 1).

Table 2.

Quantitative data of lactose content and thermal markers evaluated, and label composition of all UHT milk samples collected

Brand	Matrixa	Lactoseb (g 100 mL−1) Mean (± sd)	Thermal markers						Nutritional compositiona (g 100 mL−1)
			Total HMFc (µmol L−1)		WPNIc (mg WPN mL−1)		Lactuloseb (mg 100 mL−1)		Protein	Fat	Lactose	Glucose	Galactose
			Mean (± c.i.)	% CVintra	Mean (± c.i.)	% CVintra	Mean (± sd)	% CVintra
1	REG	4.78 (± 0.02)	11.85 (± 0.49)	82.1	4.28 (± 0.75)	9.8	34.12 (± 0.63)	n.q	3.3	1.1	4.7	n.i	n.i
	LF	0.1177 (± 0.0003)	44.68 (± 1.22)		4.92 (± 0.75)		n.q		3.3	1.1	n.i	2.3	2.3
2	REG	4.80 (± 0.00)	17.98 (± 0.55)	78.1	2.59 (± 0.75)	73.4	20.13 (± 0.01)	49.6	3.1	1.0	4.5	n.i	n.i
	LF	0.0967 (± 0.0003)	62.34 (± 1.77)		0.82 (± 0.75)		9.68 (± 0.04)		3.2	1.0	n.i	2.5	2.5
3	REG	4.63 (± 0.01)	13.56 (± 0.50)	57.7	1.75 (± 0.77)	9.0	55.72 (± 3.10)	90.1	3.2	0.8	5.0	n.i	n.i
	LF	0.1174 (± 0.0002)	32.23 (± 0.86)		1.99 (± 0.84)		12.36 (± 0.80)		3.2	0.8	n.i	2.3	2.4
4	REG	4.70 (± 0.02)	12.61 (± 0.50)	85.0	3.13 (± 0.76)	2.8	36.44 (± 0.16)	42.3	3.0	1.0	5.0	n.i	n.i
	LF	0.1065 (± 0.0004)	50.60 (± 1.40)		3.26 (± 0.75)		19.65 (± 0.22)		3.0	1.0	n.i	2.5	2.5
5	REG	4.71 (± 0.00)	9.24 (± 0.49)	78.6	3.55 (± 0.79)	1.1	19.42 (± 0.13)	n.q	3.3	1.1	4.7	n.i	n.i
	LF	0.0964 (± 0.0002)	32.38 (± 0.86)		3.61 (± 0.79)		n.q		3.3	1.1	n.i	2.4	2.3
6	REG	4.80 (± 0.01)	13.41 (± 0.50)	55.4	2.57 (± 0.77)	7.6	47.73 (± 0.97)	90.7	3.2	1.3	4.6	n.i	n.i
	LF	0.1071 (± 0.0001)	30.67 (± 0.82)		2.31 (± 0.78)		10.42 (± 0.21)		3.2	1.4	n.i	n.i	n.i
Global valuesd	REG	4.74 (± 0.07)	13.11 (± 2.86)	74.3	2.98 (± 0.88)	4.0	35.59 (± 14.53)	65.6	3.18 (± 0.12)	1.05 (± 0.16)	4.75 (± 0.21)	–	–
	LF	0.1070 (± 0.0094)	42.15 (± 12.74)		2.82 (± 1.43)		13.03 (± 4.56)		3.20 (± 0.11)	1.07 (± 0.20)	–	2.38 (± 0.11)	2.42 (± 0.08)
Matrix effecte		n.a	6.44*		0.47*		3.41**		n.a

% CV_intra coefficient of variation calculated intra-brand, considering both REG and LF samples from the same brand, n.q. not quantified, n.a. not applicable, n.i. not informed

*t_critical = 2.57 (n = 6, α = 0.05)

**t_critical = 3.18 (n = 4, α = 0.05)

^aDairy matrix classification according to the lactose content, where REG: regular milk, and LF: lactose-free milk, and nutritional composition as reported on the product’s label

^bStandard deviation in parentheses for authentic duplicates (n = 2)

^cConfidence interval calculated based on model parameters

^dGlobal means considering all replicates, standard deviation in parentheses (n = 6)

^ePaired t test considering REG and LF thermal indicators

Fig. 1.

Scatter plots showing the concentration of different thermal markers based on lactose content. a Total HMF index, b WPNI index, and c lactulose versus the content of lactose found in regular and lactose-free milk samples

HMF index

Lactose-free samples showed higher total HMF values when compared to regular ones. Ferrer et al. (2000), using reverse phase HPLC with UV detection, showed total HMF indexes of 4.71 and 25.20 µmol L⁻¹ for regular and low-lactose UHT milks, respectively, corresponding to a variability of ca. 90% comparing these matrices. Previously, we detected variations higher than 80%, comparing conventional and lactose-free UHT milk (Neves et al. 2016). Here, the intra-brand CV showed variations from 55 to 85%, suggesting the influence of the lactose hydrolysis on HMF formation. The paired t test indicated that HMF index of regular milks differ from the lactose-free ones, at 95% of confidence (t_calculated > t_critical), supporting the intra-brand analysis. Some studies revealed an additional influence of storage conditions in the Maillard reaction occurrence, showing that lactose-hydrolyzed milks are more susceptible to physical and chemical modifications during storage, exhibiting higher levels of chemical markers compared to conventional milks (Jansson et al. 2014).

Total HMF values plotted versus lactose content (Fig. 1) show the formation of two isolated groups. The first group, characterized by samples labeled as “lactose-free”, showed HMF indexes above 30 µmol L⁻¹ and low lactose concentrations. The second group exhibited HMF values below 20 µmol L⁻¹ and regular lactose content (4.5–5.0 g 100 mL⁻¹). The graphical representation suggests that HMF is influenced by the lactose hydrolysis, supporting the results found by the paired t test.

Our results are in agreement with previous findings from the literature, which indicate values of total HMF in regular UHT milk below 30 µmol L⁻¹ (range 3.46–28.4 µmol L⁻¹) detected by different methodologies (Morales et al. 1996, 2000; Morales and Jiménez-Pérez 1999; Ferrer et al. 2000; Neves et al. 2016), allow us to suggest that this marker can cautiously be used as dairy matrix classifier. However, its use as thermal marker is not recommended as it can be influenced by storage conditions. Alternatively, this compound can act as a control tool to predict the intensity of flavor and color modifications in both regular and lactose-free milks.

WPNI index

All regular and the majority of lactose-free samples were classified as “medium heat treatment” (1.51–5.99 mg WPN mL⁻¹ product), while lactose-free sample of Brand 2 received “high heat treatment” (< 1.50 mg WPN mL⁻¹) according to ADPI classifications (ADPI 2009). A recent study indicated that low rates of whey proteins denaturation on the casein micelles surface allow the formation of casein clusters, since whey proteins tend to stabilize the milk systems avoiding casein micelle collapse over the storage period (Gaur et al. 2018). As a result, it is assumed that “medium heat treatment” ADPI classification is desired in long shelf life products, like UHT milk.

Low variability in intra-brand analysis was detected (global CV_intra-brand = 4%) despite Brand 2 showing variations of ca. 70% in its WPNI index. Low WPNI values indicate high degree of soluble proteins destabilization caused by product overexposure to heating (Silva 2004; ADPI 2009). The higher HMF index (62.34 µmol L⁻¹) exhibited by Brand 2 lactose-free samples associated with the lower WPNI value may partially support the hypothesis of overheating. Brand 2 was re-tested for all thermal markers and the results were quite similar to those depicted in Table 2 (re-tested data not shown).

Both lactose-free and regular milk samples dispersed at a similar WPNI range (Fig. 1), while the paired t test indicated that there were not significant differences between lactose-free and regular WPNI values (t_calculated < t_critical). These findings suggest that the WPNI index is not influenced by the lactose hydrolysis process and can act as a valuable tool to infer about the heat treatment damage of different dairy matrices.

Lactulose

Regular UHT milks exhibited a wide lactulose content range (19.4–55.7 mg 100 mL⁻¹) which can be associated with different combinations of temperature and time exposure to heat, and/or with different UHT processes (direct and indirect). Nevertheless, all regular milks were within the range recommended by the IDF for UHT products, suggesting they were not overexposed to heating. Additionally, our findings are in accordance to several studies which found lactulose concentrations ranging 12.0–40.0 mg 100 mL⁻¹ using IDF official methods (Morales et al. 2000; Feinberg et al. 2006). Regarding lactose-free milks, our results are in accordance with some literature data that also revealed low levels of lactulose formation in reduced-lactose milks compared to regular ones, when both are submitted to the same thermal processing (Messia et al. 2007; Ruiz-Matute et al. 2012). Therefore, the classical lactulose threshold for regular UHT milks is not appropriate for lactose-free ones.

The intra-brand variability range (42–91%) and the paired t test indicating significant differences among the lactulose levels of lactose-free and regular UHT milks (t_calculated > t_critical) suggest the influence of lactose hydrolysis on lactulose formation. In addition, considering that lactose hydrolysis can occur before or after thermal treatment, and the ability of microbial β-galactosidase to hydrolyze both disaccharides (lactose and lactulose), the use of lactulose as a thermal marker for lactose-hydrolyzed milk becomes inconclusive. Fructose and tagatose determinations may be used as an alternative to infer about the heat damage in lactose-hydrolyzed products (Martínez-Villaluenga et al. 2008; Ruiz-Matute et al. 2012).

Our data also showed that lactose-free samples formed a cluster at lactulose levels below 20 mg 100 mL⁻¹_, while the regular ones had higher contents of this indicator (Fig. 1). However, data from the literature reported that both lactose-free and regular milks may exhibit lactulose levels (ca. 10.0 mg 100 mL⁻¹) if they are submitted to mild UHT processing (Elliott et al. 2005). So, lactulose is also not recommended as a matrix classifier.

Nutritional composition

The complete compositional analyses were beyond the scope of our work. However, the nutritional composition (fat, protein and carbohydrates) indicated by the producers on the product’s label (Table 2) was considered to infer about the influence of lactose hydrolysis on such macronutrients. Both regular and lactose-free milks showed similar fat and protein contents, indicating that they are not influenced by the lactose hydrolysis process. Regarding the lactose content of regular milks, values from label and CZE-UV were quite similar (4.75 and 4.74 g 100 mL⁻¹, respectively). Very low lactose amounts exhibited by lactose-free samples indicate that almost all lactose content was converted into glucose and galactose, as indicated in the product’s label. Some studies showed slight differences between glucose and galactose contents, with galactose showing lower contents with respect to glucose. This usually occurs due to the formation of galactooligosaccharides during transgalactosylating action of β-galactosidase (Messia et al. 2007; Ruiz-Matute et al. 2012). Although the centesimal analyses were not performed, our results were able to identify the relation between the thermal markers and the lactose hydrolysis and can help both the industry and the regulatory agencies to use them according to their characteristics.

Conclusion

Our results were able to identify the relationship between the thermal markers and the lactose hydrolysis and can help both the industry and the regulatory agencies to use them according to their characteristics. Lactose-free milks showed higher HMF indexes and lower lactulose contents compared to regular ones, while the lactose hydrolysis process did not significantly influence the WPNI index. The WPNI index can be considered the most appropriate thermal marker to infer about heat damage of lactose-free milks.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Funding

The authors thank to the Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq), Instituto Nacional de Ciência e Tecnologia de Bioanalítica (INCTBio–Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP) Grant No. 2014/50867-3 and CNPq Grant No. 465389/2014-7), Fundação de Amparo à Pesquisa do Estado de Minas Gerais (FAPEMIG; Rede Mineira de Química (RQ-MG) CEX—RED-00010-14) and Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) of Brazil for financial support and fellowships.