Evaluation of the volatilomic potentials of the Lactobacillus casei 431 and Lactobacillus acidophilus La-5 in fermented milk

ABSTRACT

This study was carried out to investigate the contribution of two strains of Lactobacilli; Lactobacillus casei 431 and Lactobacillus acidophilus La-5 to the volatile organic compounds (VOCs) in fermented milk. The samples were subjected to the headspace-solid phase micro-extraction (HS-SPME) and gas chromatography mass spectrometry (GC-MS). Prior to the HS-SPME, four different silica fibres were screened for their ability to retain VOCs in the fermented milk. The fibre that retained the highest amount of VOCs was selected and used for the subsequent analyses. A total of 105 compounds made up of a wide range of VOCs such as acids, alcohols, aldehydes, aromatic compounds, esters and ketones were identified. The ketones were the most abundant group of compounds produced by the two strains of lactic acid bacteria (LAB) investigated. In addition, L. casei 431 was able to produce higher concentrations of typical flavour compounds like 2,3-butanedione, 2-heptanone, acetoin and 2-nonanone compared to L acidophilus La-5. As such, the ability of these strains to produce these distinct typical flavour compounds suggests their potential as starter cultures for the production of fermented dairy products with enhanced flavour/or aroma.

RESUMEN

El presente estudio se propuso investigar la contribución que hacen dos cepas de Lactobacilli, Lactobacillus casei 431 y Lactobacillus acidophilus La-5, a los compuestos orgánicos volátiles (VOC) en la leche fermentada. Las muestras fueron sometidas a microextracción en fase sólida de espacio de cabeza (HS-SPME) y a espectrometría de masas por cromatografía de gases (GC-MS). Antes de realizar la HS-SPME se examinaron cuatro diferentes fibras de sílice con la finalidad de determinar su capacidad de retener los VOC en la leche fermentada. Así, se seleccionó la fibra que retuvo mayor cantidad de VOC para utilizarla en análisis posteriores. Se identificaron 105 compuestos integrados por una amplia gama de VOC, a saber: ácidos, alcoholes, aldehídos, compuestos aromáticos, ésteres y cetonas. El grupo más abundante de compuestos producidos por las dos cepas de bacterias del ácido láctico (LAB) investigadas fue el de cetonas. El segundo grupo más copioso fue el de ácidos. Los compuestos derivados de aldehídos, alcoholes y ésteres tuvieron escasa presencia. Además, en comparación con L. acidophilus La-5, L. casei 431 produjo concentraciones más altas de compuestos saborizantes típicos como 2,3-butanodiona, 2-heptanona, acetoína y 2-nonanona. En conclusión, la capacidad de las cepas investigadas para producir distintos compuestos de sabores típicos sugiere su potencial para ser empleadas como cultivos iniciadores destinados a la elaboración de productos lácteos fermentados con sabor/aroma mejorado.

KEYWORDS:

• Lactobacillus acidophilus La-5
• Lactobacillus casei 431
• milk fermentation
• volatile organic compounds

PALABRAS CLAVE:

• Lactobacillus acidophilus La-5
• Lactobacillus casei 431
• fermentación de leche
• compuestos orgánicos volátiles

1. Introduction

Health-conscious consumers are constantly demanding for dairy products with high nutritional quality and sensory satisfaction. Whilst consumers are more interested in products that are low in fat and sugar, overall flavour quality remains the core determinant for their food choices (Potts & Peterson, 2019). Lactic acid bacteria (LAB) play a major role in the production of characteristic flavours during fermentation and storage that affects the quality and acceptability of fermented dairy products. The overall flavour of fermented milk is a combination of non-volatile and volatile organic compounds (Engels et al., 1997). The nature and type of aroma compounds produced by Lactobacillus sp often depends on the species and strains used (Cui et al., 2019). The compounds responsible for the aroma are quite diverse and they include; alcohols, aldehydes, esters, fatty acids, ketone and sulphur compounds among others (Beshkova et al., 2003; Bozoudi et al., 2015; Dan, Wang, Wu et al., 2017; Marilley & Casey, 2004; G. Smit et al., 2005; Valduga et al., 2010). In the same way, the precursors are as diverse as the chemical classes of the aroma compounds.

The precursors include the three main components of food matter namely carbohydrates, lipids and proteins. In addition, aroma formation is a complex process involving the generation of precursors and their subsequent conversion into aroma compounds (Smid & Kleerebezem, 2014). These aroma compounds which often results from amino acid catabolism by LAB (Yvon & Rijnen, 2001) generally impart favourable flavours to the fermented milk (Lin et al., 2016). The amino acid converting ability of LAB varies greatly from strain to strain (G. Smit et al., 2005). Therefore, an adequate selection of LAB based on metabolic activity is required to produce dairy products with good flavour, adequate acidity and viscosity. The flavour of a dairy product is extremely complex due to the great number of compounds present which have different polarities, volatilities and occurs in a wide range of concentrations (Lasekan et al., 2013). Therefore, sample preparation such as extraction and concentration of compounds remains a critical area in volatile analysis. Continuous liquid-liquid extraction has been used in sample preparation for the determination of dairy volatiles (Xanthopoulos et al., 1994). Other extraction methods include; ultrasound (Abesinghe et al., 2019), supercritical fluid extraction-gas chromatography (Tuomala & Kallio, 1996) and headspace solid-phase micro-extraction (HS-SPME) (Dan, Wang, Jin et al., 2017).

HS-SPME has become a popular technique widely used to analyse volatile compounds in several matrices, among them, fermented milks (Lasekan et al., 2016). SPME coupled with gas chromatography mass spectrometry (SPME-GC-MS) are commonly used for the qualitative and quantitative analysis of volatile organic compounds (VOCs) in various dairy products (Condurso et al., 2008; Tunick et al., 2013). The volatilomic potentials of Lactobacillus sp. has been reported in many studies in which different species and strain were used (Irlinger et al., 2017). For example, L. casei has been shown to produce 2,3-butanedione (diacetyl) and acetoin, that gives the buttery flavor in fermented dairy products (Cui et al., 2019). L. acidophilus has the ability to produce acetaldehyde and diacetyl. The amounts of acetaldehyde produced by L. acidophilus have an impact on the flavor of dairy products (Østlie et al., 2003). The individual LAB strains’ ability to produce an array of different volatile compounds is a powerful tool for selecting cultures with special flavor attributes (Cuffia et al., 2019). VOCs of fermented milk have been widely studied in recent years. However to the best of our knowledge, there are no reported studies on the volatilomic potential of these two strains of LAB (i.e. Lactobacillus acidophilus La-5 and Lactobacillus casei 431).

The purpose of the present study was to characterize the volatile compound profiles of two strains of LAB (Lactobacillus acidophilus La-5 and Lactobacillus casei 431) during milk fermentation, using SPME-GC-MS technique.

2. Materials and methods

2.1. Bacterial isolates and chemicals

Lactobacillus sp. (L. acidophilus La-5 and L. casei 431) was obtained from Chr. Hansen A/S as frozen direct-to-vat starters (DVS) (Gayrettepe, Kuala Lumper, Malaysia). De Man Rogosa Sharpe, (MRS) broth (Merck, Germany) and C₈-C₄₀ n-alkanes were obtained from Merck chemicals (Darmstadt, Germany). Each strain of the frozen LAB was propagated in 10 mL MRS broth in three successive transfers at 37°C for 24 h. To separate bacteria, the cultures were centrifuged at 5000 g /15 min. The biomass was washed twice (2×) with distilled water. The LAB were subsequently inoculated into 100 mL of reconstituted skim milk (12% w/v), then incubated at 37°C for 24 h as preculture to obtain approximately ~10⁸ colony forming units (CFU)/mL. Three inoculation treatments were made. Each of the pre-cultures was inoculated (2% v/v) separately into the pasteurized 1 L whole milk and incubated at 37°C for 24 h. The obtained fermented milk products (~ pH 4.6) were stored for volatile analysis at 4°C.

2.2. SPME fibre screening

Different silica fibres [StableFlex polydimethylsiloxane (PDMS), 100 μm; StableFlex polydimethylsiloxane/ divinylbenzene (PDMS/DVB), 65 μm; StableFlex polydimethylsiloxane/ carboxen (PDMS/CAR), 75 μm; StableFlex divinylbenzene/ carboxen/polydimethylsiloxane (DVB/CAR/PDMS), 50/30 μm] obtained from Supelco (Bellefonte, PA, USA) were chosen for the optimization study. Previously, all fibres were conditioned according to the manufacturer’s instructions. Each fibre was exposed to the headspace of the milk fermented by L. casei 431 using similar temperature and time (50°C/20 min). Desorption was carried out for 15 min at 250°C. All analyses were performed in triplicate. The fibre that retained the highest amount of VOCs was selected and used for the subsequent analyses. For that, the chromatographic profiles of the volatile compounds were integrated and the peak area values were averaged and expressed as arbitrary units

2.3. HS-SPME optimization

A factorial central composite rotatable design (CCRD) was used to optimize HS-SPME conditions (Chatterjee et al., 2012). For CCRD design, N is the number of experiments which consists of 2 ⁿ factorial points and 2 ⁿ axial points (α = 1.4142) and n_c central points. n is the number of independent variables. The temperature extraction (°C) and extraction time (min) were chosen as variables for the optimization and the levels of each variable are shown in Table 1. The responses for the total peak area of VOCs analysed were correlated using a second order polynomial model Equation (1)(1) $y=\beta o+\sum _{i=1}^k\beta ixi\sum _{i=1}^k\beta ii{x}_1^2+\sum _{i=1}^{k-1}\sum _{j=1}^k\beta ijxixj+\epsilon$(1)

(1) $y=\beta o+\sum _{i=1}^k\beta ixi\sum _{i=1}^k\beta ii{x}_1^2+\sum _{i=1}^{k-1}\sum _{j=1}^k\beta ijxixj+\epsilon$(1)

Table 1. Factors, levels and experimental domain of the conditions applied to optimize the extraction by HS-SPME.

Tabla 1. Factores, niveles y dominio experimental de las condiciones aplicadas para optimizar la extracción mediante HS-SPME

Variables	Coded variables
	-α	−1	0	1	α
Extraction temperature (T, ◦C)	34	40	55	70	76
Extraction time (t, min)	12	20	40	60	68

Where y is the forecasted response; βo: a constant; βi: the linear coefficient; βii: the quadratic term coefficient; βij: the interaction coefficient and ε: the error associated with the model. The obtained data are presented in Tables 1 and 2 respectively.

Table 2. Experimental conditions and values of response (total area) obtained for the CCRD for the HS-SPME optimization.

Tabla 2. Condiciones experimentales y valores de respuesta (área total) obtenidos para el CCRD de la optimización mediante HS-SPME

	Factors**
Experiment	T (°C)	Extraction temperature (°C)	t (min)	Extraction time (min)	Response*** (Total peak area of the compounds)
1	−1.41	34.00	0	40.00	3.90 E + 07
2	−1	40.00	−1	20.00	3.60 E + 07
3	−1	40.00	1	60.00	3.97 E + 07
4	0	55.00	−1.41	12.00	3.24 E + 07
5*	0	55.00	0	40.00	4.54 E + 07
6*	0	55.00	0	40.00	4.34 E + 07
7*	0	55.00	0	40.00	4.33 E + 07
8*	0	55.00	0	40.00	4.42 E + 07
9*	0	55.00	0	40.00	4.43 E + 07
10	0	55.00	1.41	68.00	5.33 E + 07
11	1	70.00	−1	20.00	1.98 E + 07
12	1	70.00	1	60.00	2.46 E + 07
13	1.41	76.00	0	40.00	1.91 E + 07

*: central point repetition; **: with α = 1.4142; ***: expressed in arbitrary units.

*: punto central de repetición; **: con α = 1.4142; ***: expresado en unidades arbitrarias.

2.4. Headspace solid-phase micro-extraction of fermented milk

About 5.0 g of the fermented milk was weighed into a 20 mL vial sealed with a screw-top cap and a PTFE/Silicon septum (Fisher Scientific Inc., Ireland Ltd). The HS-SPME analysis was carried out according to the design established in Table 2. A PDMS/CAR fibre selected from the optimization study (Figure 1) was exposed to the headspace of samples. Then the fibre was inserted into the GC injector port provided with 0.75 mm i.d. liner, and the analytes were thermally desorbed at 250°C for 5 min (split-less mode).

Figure 1. The total peak area of compounds obtained by different fibre coatings. Each fibre was exposed to the headspace of the fermented milk by L. casei 431 for 20 min and 50°C.

Figura 1. Área de pico total de compuestos obtenidos mediante diferentes recubrimientos de fibra. Cada fibra fue expuesta al espacio de cabeza de la leche fermentada por L. casei 431 durante 20 minutos a 50°C

2.5. GC-MS analysis

The method described by Aunsbjerg et al. (2015) was employed for the GC-MS analyses. The tentative identification of VOCs extracted by HS-SPME method from the two types of fermented milk products was carried out using a GC-MS System, Agilent 5890 (Agilent, Stockport, UK) instrument equipped with a GC-17A ver.3 GC with a flame ionization detector (FID) (Shimadzu, Japan), operating in a split-less mode. The fibre was kept in the injector at 250°C for 5 min, with helium gas (purity 99.99%) as carrier gas at a constant flow rate of 1.0 mL/min during desorption. The volatile compounds were separated using a ZEBRON ZB-FFAP column (Agilent, UK), 50 m $\times$ 0.2 mm internal diameter $\times$ 0.33 µm film thickness (Crawford Scientific, UK). The oven temperature program was as follow: initial temperature was 40 °C, held for 1 min, increased at 5°C /min to 70°C, and finally 5°C/min to 100°C. The final temperature was then ramped at 10°C/min to 240°C to allow a run time of 30 min. The mass spectrometry was set to record 33–450 atomic mass units, at a sampling rate of 1.11 scans s⁻¹.

2.6. Identification of volatile compounds

Volatiles were tentatively identified by comparing their mass spectra with the mass spectral database (NIST Standard Reference Database 1A V17, Agilent Technologies Inc.) and Kovats retention index (RI). Calculation of the retention indices was carried out by Kovats method (Equation 2(2) $RI\left(x\right)=100\left(z\right)+100\times \frac{Rt\left(x\right)-Rt\left(z\right)}{Rt\left(y\right)-Rt\left(z\right)}$(2) ), using a normal paraffin mixture (C₈-C₄₀) as external references (Lasekan & Hussein, 2018).

(2) $RI\left(x\right)=100\left(z\right)+100\times \frac{Rt\left(x\right)-Rt\left(z\right)}{Rt\left(y\right)-Rt\left(z\right)}$(2)

Where RI (x) is the retention index of the unknown compound (x); x: unknown compound that is being analysed; z: the number of carbon atoms of the n-alkane eluted before (x); y: the number of carbon atoms of the retention time n-alkane eluted after (x); Rt (x): the retention time of the unknown compound (x); Rt (z): the retention time of the n-alkane eluted before (x); and Rt (y): the retention time of the n-alkane eluted after (x). Bianchi et al. (2007) and Manral et al. (2008).

2.7. Gas chromatography-olfactometry (GC–O)

The olfactometric analysis of the fermented milk volatiles was carried out with a Trace Ultra 1300 GC (Thermo Fischer, Scientific, Waltham, MA, USA) with a ZB-FFAP column (Agilent, UK), 50 m x 0.2 mm internal diameter x 0.33 μm film thickness (Crawford Scientific, UK) and an olfactory detector port (ODP 3), (Gerstel, Mulheim, Germany). The transfer line to the GC-O sniffing port was held at 250°C. The volume of the extracted milk samples analysed and the oven temperature program was similar to those employed for the GC-MS above. Sniffing was carried out by three trained panellists with over two years of training. These panellists exhibited normalized responses which produced strong agreement.

2.8. Statistical analysis

Analysis of variance (ANOVA) was carried out to determine the most fitted response surface models. Statistical analysis was carried out using the Minitab, Version 18, (Minitab Inc., USA). In addition, Pareto charts were established to analyse the outcomes of the volatiles profiles. The visualization of the differences between the overall VOCs in the fermented milk samples was carried out using the Gene Cluster 3.0/Java Treeview (Windows, Mac OSX, Linux, UNIX) software.

3. Results and discussion

3.1. Selecting for the best SPME fibre

Selecting the best coating fibre from the different fibres is an important step in the optimization process which usually relies on the chemical nature of the targeted analyte. Figure 1 shows the mean of the total peak areas obtained from the four different SPME fibres tested with respect to their capacity to extract the VOCs of the fermented milk. Results revealed significant differences (p < .05) in the fibres’ ability to extract volatiles. In the current study, PDMS/CAR was able to extract the highest amount of volatile compounds (i.e.163 compounds) as reflected in its higher mean peak area (19.80 × 10⁷) (Figure 1). This was followed by the PDMS/DVB (6.0 × 10⁷) with 60 compounds and PDMS/DVB/CAR as well as PDMS extracted the least amount of volatile compounds (< 3.0 × 10⁷). Therefore, we selected the PDMS/CAR which gave the highest mean peak area value for the assessment of the characteristic volatile profile of the fermented milk products.

3.2. Optimization of the HS-SPME

Studies have shown that the headspace analysis of VOCs by HS-SPME is greatly affected by the vapour pressure of volatile compounds in the vial (Lasekan et al., 2016; Pellati et al., 2005). In addition, the extraction temperature, equilibrium time and extraction time are the major factors influencing the vapour pressure and the equilibrium of the VOCs in the headspace (Pellati et al., 2005). Based on the results of previous studies, the extraction time and extraction temperature were therefore selected and optimized (Table 2). Thirteen experiments of the design were carried out at random and the responses are presented in Table 2. The results and the predicted regression coefficients of the fitted mathematical models with their corresponding R² values, and lack of fit test are provided in Table 3. The significance of each term was determined using the F-ratio and the P value. In this study, the amount of VOCs extracted from the headspace of the fermented milk was significantly (p < .05) influenced by both the main and interactive effects of the extraction temperature and extraction time. The quadratic effect of the independent variables had no significant influence (p > .05) on the amount of VOCs extracted. This observation was further corroborated by the Pareto chart (Figure 2(a)) which showed that increase in either factors (i.e. extraction temperature and extraction time) resulted in significant increases (p < .05) in the amount of VOCs extracted. Similar observations were reported for the extraction of VOCs from Chili (Junior et al., 2011) and Brazilian sugar cane spirits (Zacaroni et al., 2017). To provide a visual relationship between the independent factors and the amount of VOCs extracted a 3D response surface plot of the peak area of the VOCs against extraction time and extraction temperature is presented in Figure 2(b). Results showed that increasing the extraction temperature resulted in an increase in the amount of VOCs. However, when the temperature was increased beyond 50°C, there was significant reduction (p < .05) in the amount of VOCs extracted. This observation is due to the fact the SPME is an exothermic reaction (Pellati et al., 2005). The extraction of VOCs decreases with increase in extraction temperature. In the light of these results the best response within the range studied was defined as an extraction time of 60 min and an extraction temperature of 50°C.

Table 3. Coefficients and tests of the significance of the variables obtained for the extraction of the volatile organic compounds from fermented milk.

Tabla 3. Coeficientes y pruebas de significancia de las variables obtenidas para la extracción de compuestos orgánicos volátiles de la leche fermentada

Regression coefficient	Total peak area	P level
b0	44189275	0.000*
x1	−7443838	0.002*
x2	4747395	0.017*
x1x2	−9018029	0.001*
$x_1^2$	−2115295	0.235
$x_2^2$	263309	0.906
R^2	0.9025
Regression (P value)	0.001*
Lack of fit (F value)	61.14
Lack of fit (P value)	0.001*

X₁, Extraction temperature; X₂, extraction time; b_o = constant; P values * significant (p < 0.05)

X₁, Temperatura de extracción; X₂, tiempo de extracción; b_o = constante; * Valores P significativos (p < 0.05)

Figure 2. (a): Pareto chart for responses (total peak area versus different conditions tested (b): response surface plot for total peak are versus extraction time, and extraction temperature.

Figura 2. (a): Gráfico de Pareto de las respuestas (área de pico total versus diferentes condiciones probadas); (b): gráfico de superficie de respuesta para el pico total versus tiempo de extracción y temperatura de extracción

3.3. Volatile organic compounds analysis in the fermented milk

The formation of VOCs in milk and dairy products is complex and it includes glycolysis, lipolysis as well as proteolysis of milk components which is brought about by the enzymatic mechanisms of microflora within the dairy matrix (G. Smit et al., 2005). Suffice to say that the production of flavour compounds in fermented dairy products is strain dependent. In this study, both L. casei 431 and L. acidophilus La-5 have the ability to produce VOCs, such as ketones, aldehydes, alcohols, and acids. More than hundred compounds were tentatively identified in the volatile fractions of the fermented milk samples. Of this, eighty-three compounds were identified in the L. casei milk while eighty-four compounds were identified in the La-5 milk. The compounds included seven aldehydes, twenty-one ketones, fifteen acids, twelve esters, nineteen alcohols, thirteen aromatic hydrocarbons, 12 hydrocarbons and six furans (Table 4). As shown in Figure 3, ketones and acids demonstrated to be major VOCs, while the other compounds were minor in the fermented milk samples. Although, the qualitative profile of both L. casei milk and La-5 milk were similar, the relative abundance of the compounds of different chemical families was different. The differences between the L. casei milk and La-5 milk were mainly due to the composition and abundance of their ketones, acids, and alcohols. For instance, in this study, 21 ketones were identified in the fermented milk samples (Table 4). Some of these important flavour VOCs were present at relatively high levels. The most prominent of these ketones were the creamy/buttery 2,3-butanedione (12.2–25.5%), fruity 2 heptanone (15.2–17.89%) and the herbal 2-nonanone (1.67–2.67%). These ketones are metabolised from citrate (Pan et al., 2014). Ketones have variously been reported as important flavour compounds in fermented milk (Chaves et al., 2011; Dan, Wang, Wu et al., 2017; Milesi et al., 2010). 2,3-Butanedione is a diketone which was the most abundant ketone in the L. casei milk with percentage area twice that of the La-5 milk. In addition, this compound can readily be converted into acetoin by the enzyme diacetyl reductase (Rattray et al., 2003). Surprisingly, low levels of acetoin were obtained for both La-5 and L. casei milks. In contrast, 2-heptanone and 2-nonanone contributed significantly to the total amount of ketones in both milk samples. These compounds have been reported as the major contributors to the creamy fresh flavour in fermented milk (Pionnier & Hugelshofer, 2006).

Table 4. Volatile compounds identified by HS-SPME-GC–MS/GC-O analysis in fermented milk.

Tabla 4. Compuestos volátiles identificados mediante el análisis HS-SPME-GC–MS/GC-O en leche fermentada

				Area (%)
				Fermented milk by:
No.	Volatile Compound	Chemical Formula	RI*	L. casei 431	L. acidophilus La-5	Aroma description
Aldehydes
1	3-methyl-Butanal	C5 H10O	643	0.527	0.602	Malty
2	Hexanal	C6H12O	806	1.455	9.266	Fatty, green
3	(z)-2-Heptenal	C7H12O	913	0.032	-
4	(E)-2-Octenal	C8H14O	1013	0.036	-	Fresh cucumber
5	3,7-Dimethyl-6-octenal	C10H18O	1125	-	0.028	Herbal
6	Benzaldehyde	C7H6O	982	0.173	0.027	Almond-like
7	5-Hydroxymethylfurfural	C6H6O3	1163	0.028	0.019	Herbal
	Total			2.251	9.942
Ketones
8	Acetone	C4H8O2	455	2.004	1.936
9	2-Butanone	C4H8O	555	1.247	-
10	2-Pentanone	C5H10O	654	1.833	-
11	2,3-Butanedione (Diacetyl)	C4H6O2	691	25.529	20.216	Creamy,buttery
12	1-Hydroxy-2-propanone	C3 H6O2	698	0.217	0.454
13	3-Hydroxy-2-butanone (Acetoin)	C4H8O2	717	0.490	0.133	Buttery,Creamy
14	2-Hexanone	C6H12O	754	-	0.243
15	Acetyl valeryl	C7H12O2	989	0.253	-	Yogurt-like
16	2-Heptanone	C7H14O	853	17.986	15.213	Fruity
17	2-Octanone	C8H16O	952	0.165	0.110	Cheese-like
18	2-Nonanone	C9H18O	1052	2.670	1.671	Herbal
19	2-Decanone	C10H20O	1151	-	0.015	Sour
20	2-Undecanone	C11H22O	1251	0.410	0.249
21	4-Nonanone	C9H18O	1052	0.004	-	Green
22	6,7-Dodecanedione	C12H22O2	1486	0.016	-
23	2(5H)-Furanone	C4H4O2	807	0.120	0.081
24	2-Tridecanone	C13 H26O	1449	0.055	0.050
25	2-Pentadecanone	C15H30O	1648	0.018	0.013
26	5-Hydroxymethyldihydrofuran-2-one	C5H8O3	1129	0.017	-
27	5-Heptyldihydro-2(3H)-furanone	C11H20O2	1483	-	0.016	Creamy
28	Dihydro-4-hydroxy-2(3H)-furanone	C4H6O3	1013	0.012	-	Caramel-like
	Total			53.046	40.400
Acids
29	Acetic acid	C2H4O2	576	13.065	19.525	Sweaty
30	Propanoic acid	C3H6O2	676	0.251	0.148
31	Butanoic acid	C4H8O2	775	2.347	1.732	Yoghurt, rancid
32	Pentanoic acid	C5H10O2	875	0.050	0.061
33	Oxalacetic acid	C4H4O5	1268	0.014	-
34	Hexanoic acid	C6H12O2	974	3.290	3.042
35	Heptanoic acid	C7H14O2	1073	0.045	0.067
36	Octanoic acid	C8H18O2	1173	0.709	1.262
37	Nonanoic acid	C9H18O2	1272	-	0.071
38	n-Decanoic acid	C10H20O2	1372	0.091	0.015
39	9-Decenoic acid	C10H18O2	1362	-	0.011
40	15-Hydroxypentadecanoic acid	C15H30O3	2111	-	0.028
41	Dodecanoic acid	C12H24O2	1570	-	0.047
42	Tetradecanoic acid	C14H28O2	1769	-	0.047
43	n-Hexadecanoic acid	C16H32O2	1968	-	0.074
	Total			19.862	26.13
Esters
44	2- Ethylhexy acetate	C4H6O2	576	0.006	-
45	Ethyl Acetate	C4H8O2	586	0.738	4.873	Sweet
46	2-Methyl propanoate	C4H8O2	711	0.079	0.036	Butter
47	3-Methyl butanoate	C5H10O2	811	0.269	-	Sweat
48	2-Methyl butanoate	C5H10O2		-	0.118	Acidic
49	Pentyl acetate	C7H14O2	884	0.017	-
50	Isobutyl-2-methyl-2-propenoate	C7H12O2	852	-	0.064
51	2-Ethyl-1-hexyl acetate	C10H20O2	1118	0.003	-
52	Sec-butyl crotonate	C8H14O2	1181	0.078	-
53	Methyl-2-oxopropanoate	C11H20O3	1418	-	0.021
54	Cyclohexyl decanoate	C16H30O2	1842	-	0.050
55	Decyl decanoate	C20H40O2	2177	0.090	-
	Total			1.280	5.162
Alcohols
56	Cyclobutanol	C4H8O	668	0.054	-
57	Isobutanol	C4H10O	646	-	0.960	Bitter
58	Eucalyptol	C10H18O	1059	-	1.199	Camphoraceous
59	3-Methyl-1-butanol	C5H12O	697	-	0.552	Caramel
60	1-Pentanol	C5H12O	761	0.915	1.115
61	2-Heptanol	C7H16O	879	0.298	0.077	Apricot
62	3-Methyl-2-buten-1-ol	C5H10O	746	0.046	0.222	Citrus
63	1-Hexanol	C6H14O	860	0.421	0.248
64	6-Undecanol	C11H24O	1277	0.011	-
65	2-Octanol	C8H18O	979	0.039	-
66	1-Octen-3-ol	C8H16O	969	0.180	0.039	Mushroom
67	1-Heptanol	C7H16O	960	0.122	-	Green
68	2-Nonanol	C9H20O	1078	0.184	0.038
69	1-Octanol	C8H18O	1059	0.097	0.050
70	Tetrahydro-2-furanol	C4H8O2	777	0.049	-
71	1-Nonanol	C9H20O	1159	0.064	0.025
72	2-Furanmethanol	C5H2O6	885	0.852	0.522
73	2-Undecanol	C11H24O	1277	0.056	-
74	3-Methyl phenol	C7H8O	1014	0.005	0.007	Smoky
	Total			3.393	5.054
Aromatic Hydrocarbons
75	Toluene	C7H8	794	0.035	0.343
76	Ethylbenzene	C8H10	893	0.014	0.015
77	o-Xylene	C8H10	907	-	0.009
78	ρ-Xylene	C8H10	907	0.009	-
79	D-Limonene	C10H16	1018	-	0.235
80	Cyclohexane	C6H12	719	-	0.056
81	Styrene	C8H8	883	0.034	-
82	Benzene, 1-methyl-4-(1-methylethyl)-	C10H14	1042	-	0.254
83	Benzene, 4-ethyl-1,2-dimethyl-	C10H14	1119	-	0.016
84	Benzene, propyl-	C9H12	992	-	0.019
85	Allyl n-octyl ether	C11H22O	1181	0.046	-
86	Pentylpyridine	C10H15 N	1192	-	0.960	Fat
87	2-Methylpyridine	C6H7 N	886	0.007	-	Sweat
	Total			0.145	1.907
Alkanes
88	Hexane	C6H14	618	1.811	-
89	Hexane,3,3-dimethyl-	C8H18	732	0.568	-
90	Heptane,2,2,4,6,6-pentamethyl-	C12H26	981	0.181	0.259
91	Undecane	C11H4	1115	-	0.073
92	Dodecane	C12H26	1214	0.034	-
93	Butane, 2,3-dimethyl-	C6H14	489	-	0.065
94	Heptane, 3-ethyl-4-methyl-	C10H22	887	0.035	-
95	Heptan,3-[1,1-dimethylethoxy)methyl]-	C12H26O	1141	0.026	0.038
	Total			2.655	0.435
Alkenes
96	1-Undecene	C11H22	1105	-	0.044
97	5-Undecene, (E)-	C10H16	958	-	0.013
98	1,3-Hexadiene,3-ethyl-2-methyl	C9H16	868	0.115	-
99	Benzene, 1-ethenyl-4ethyl-	C10H12	1096	-	0.003
	Total			0.115	0.06
Furans
100	Furan	47H4O	553	0.311	-
101	Furan,3-methyl-	C5H6O	642	0.367	0.303
102	Furan,2-ethyl-	C6H8O	742	0.095	0.137
103	2-n-Butyl furan	C8H12O	941	0.006	-
104	Furan,2-pentyl-	C9H14O	1040	0.390	0.319
105	Furfural	C5H4O2	831	0.080	0.009
	Total			1.249	0.768

*: The retention indexes (RI) of unknown compounds on ZB-FFAP 30 m X 0.2 column (Agilent Technologies Inc., USA) calculated against the GC-MS retention time of n-alkanes (C3-C40); -: not detected.

*: Los índices de retención (RI) de compuestos desconocidos en la columna ZB-FFAP 30 m X 0.2 (Agilent Technologies Inc., EE. UU.), calculados contra el tiempo de retención GC-MS de n-alcanos (C3-C40); -: no detectado.

Figure 3. The relative contents of the different classes of volatile compounds (VOCs) detected in the fermented milk samples.

Figura 3. Contenidos relativos de las diferentes clases de compuestos volátiles (VOC) detectados en las muestras de leche fermentada

Similar to the ketones, the alcohols are important flavour compounds in dairy products (Pan et al., 2014). The fermented milk samples produced an array of alcohols such as; 3-methyl-1-butanol, 3-methyl-2-butenol, 1-hexanol, 2-furanmethanol, 1-octen-3-ol and 2-nonanol all of which occurred in appreciable amount. It is worthy to note that the La-5 milk had a higher total relative alcohol abundance compared to the L. casei milk. Most of these alcohols have been reported in dairy products and some of them are known to impart good flavour to dairy products. For example, 1-octen-3-ol which has a low threshold value (1.5 μg/L) is known to contribute good flavour to dairy products even at low concentrations (Friedrich & Acree, 1998).

On the other hand, fifteen acids were detected in the different fermented milk samples. Among these, the majority were acetic acid, butanoic acid, hexanoic acid and octanoic acid. In this study, the acetic acid percentage abundance ranged from 13.1 to 19.5%, with La-5 milk recording higher value of 19.5%. High levels of acetic acid have been reported to contribute to the ‘tart’ flavour of yogurt and ‘vinegary’ taste in fermented milk (Panagiotidis et al., 2001). Apart from acetic acid, hexanoic acid and octanoic acid which were in lower concentrations than acetic acid have been reported as contributing fatty/cheesy flavour to fermented milk (Pan et al., 2014).

Ethyl acetate was the most significant ester detected in the fermented milk samples. Other esters such as 3/2-methylbutanoate, 2-methylpropanoate, pentyl acetate, 2-ethylhexyl acetate occurred in low concentrations in the fermented milk samples. The contribution of esters to the flavour of dairy products is concentration dependent. At low concentration, esters contribute positively to the overall flavour balance, while at high concentrations, they produces fruity flavour defect (Liu et al., 2004).

Fewer numbers of aldehydes were detected in the fermented milk samples. The aldehydes were the malty 3-methylbutanal, hexanal, (z)-2-heptenal, (E)-2-octenal, 3,7-dimethyl-6-octenal, benzaldehyde, and 5-hydroxymethyl furfural. 3-Methylbutanal and hexanal were the predominant aldehydes detected. In numerous food products, aldehydes are key-flavour compounds. Sensorially, they are discerned as malty, almond-like. 3-Methylbutanal is among the most important volatiles in fermented dairy products and it is produced by synthesis action of micro-organisms via both the Ehrlich pathway and via amino-acid biosynthetic pathways of branched-chain amino acids valine, isoleucine and leucine (B. A. Smit et al., 2009).

In this study, there were many hydrocarbons and aromatic hydrocarbons in the fermented cow milk samples. These compounds were possibly related to the milk, feed given to the herd, the milk transport containers, materials used during production, and packaging materials (Yüksel & Bakırcı, 2015). For example, limonene can be transferred directly from the cows’ feeds (Viallon et al., 1999).

Finally, to better visualise the differences between the overall volatile compound profiles generated by the two strains of LAB, a heat map (Figure 4) of their volatilomic potentials was generated. The dark green blocks indicated compounds exhibiting higher peak areas while the red blocks were for compounds showing much lower peak areas. The heat map shows that L. casei 431 exhibited higher abundances for the following compounds; 2-heptanone, 2,3-butanedione, 2-pentanone, 2-butanone, 2-nonanone, butanoic acid, acetoin, 2-furanmethanol, 2-methylbutanoate etc. (Dark green color). On the other hand L. acidophilus La-5 exhibited higher abundances for acetic acid, hexanol, ethyl acetate, isobutanol, oactanoic acid, 3-methyl-1-butanol, 2-hexanone etc. (Dark green color). Overall, L. casei 431 exhibited higher abundances in critical chemical groups (i.e. ketones, esters, alcohols and aromatic hydrocarbons) as compare to L. acidophilus La-5.

Figure 4. Heatmap of the volatile organic compounds (VOCs) identified in fermented milk samples. The most abundant chemical groups are shown in green color while the least abundant group is shown in red color.

Figura 4. Mapa de calor de los compuestos orgánicos volátiles (VOC) identificados en muestras de leche fermentada. Los grupos químicos más abundantes aparecen en color verde, mientras que el grupo menos abundante aparece en color rojo

4. Conclusion

According to the results of this study, PDMS/CAR is the best fibre for the extraction of VOCs in fermented milk. In addition, analysis of the VOCs produced during the lactic acid fermentation of Lactobacillus casei 431 and Lactobacillus acidophilus La-5, led to the identification of a wide range of VOCs made up of 7 aldehydes, 21 ketones, 15 acids, 12 esters, 19 alcohols, 12 hydrocarbons, 13 aromatic compounds and 6 furans. Furthermore, the ketones represented a consistently higher percentage in the milk produced by the two strains of the LAB investigated. The ketones were followed by the acids. The aldehydes, alcohols and esters produced relatively lower percentage abundances in the milk samples. The ability of these LAB strains to produce distinct flavour compounds suggests their potential as starter cultures to enhance the development of dairy products. In this study, L. casei 431 was able to produce more typical flavour compounds such as 2,3-butanedione, 2-heptanone, acetoin and 2-nonanone compared to L acidophilus La-5. Overall, L. casei 431 exhibited higher abundances in the production of critical chemical groups (i.e. ketones, esters, alcohols and aromatic hydrocarbons) as compared to L. acidophilus La-5.

Acknowledgments

The authors are grateful for the extensive financial support received from University Putra Malaysia research scheme Grant (9478500). Also, the authors are grateful for LAB strains supplied by Chr. Hansen A/S Malaysia.

Disclosure statement

The authors declares no competing interest

Additional information

Funding

This work was supported by the University Putra Malaysia [research scheme Grant (9478500)].