Introduction

Proteases have been investigated for practical purposes, as cleaning and hair removing agents, in detergents, leather processing, silver recovery, medical applications, food processing, chemical industry, and waste treatment [1, 2]. Moreover, their wide diversity and action specificity range have attracted the attention of biotechnologists worldwide [3]. Although protease production is an inherent property of all organisms, only those microorganisms producing a substantial amount of extracellular protease can be commercially exploited [4].

It is well known that extracellular protease production by microorganisms is greatly influenced by media components, especially carbon and nitrogen sources, as well as physical factors such as pH, temperature, inoculum size, dissolved oxygen level, and incubation time [5]. However, there is not an optimal medium for the production of protease by different microbial species because each of them requires specific cultivation conditions [6].

Soybean flour (SF) is an inexpensive component of media for protease production by bacteria, being used as nitrogen, carbon, and vitamins source; therefore, its use as alternative component for proteases production by yeasts would be of a certain industrial concern [7–12].

The use of microorganisms as biotechnological sources of industrially relevant enzymes has stimulated a renewed interest in the identification of news producers such as yeasts [13, 14]. Several researchers [13, 15–17] have demonstrated that yeasts are able to produce enzymes with biotechnological potential, but only a few yeast proteases have been studied for alternative potential applications [18–20].

Terrestrial and marine yeasts have been described as protease producers such as Candida lipolytica [3], Yarrowia lipolytica [3], and Aureobasidium pullulans [2]. Other important sources of yeasts are many cheese varieties and raw milks, where they live in association with lactic bacteria [15, 16]. Dairy products offer a special niche for the development of yeasts with specific activity [21]. Different caseinolytic yeasts have been identified, including some species of Kluyveromyces and Candida such as Candida punicea, Candida lipolytica, Candida aquatica and Candida curiosa [14].

The aims of this work were (a) to select yeasts species from raw milk able to release extracellular proteases, (b) to identify the variables most influencing protease production by Candida buinensis by means of a 23 factorial design, (c) to determine the best conditions for its activity and stability, and (d) to estimate the thermodynamic parameters of its thermal inactivation.

Materials and Methods

Milk Samples and Yeast Isolation

The milk samples were collected in a dairy products cooperative at the Cachoeirinha district of Pernambuco, Brazil. The milk samples were filtered through quantitative filter and utilized for microorganism isolation.

Samples were serially diluted tenfold in sterile distilled water (102, 104, 106, and 108); 0.001 mL of the dilutions were inoculated in Man Rogosa Sharpe agar medium (Difco, Detroit, MI) and incubated for 3 days at 30 °C. The yeast colonies were randomly selected, through the macroscopic and microscopic characteristics of the colonies, from each plate and purified by subsequent streaking on Sabourand agar (Difco, Detroit, MI). The slants thus obtained were maintained at 4 °C.

Yeast Identification

The identification was realized by morphological characterization [18], reproduction type [22], physiological characterization based on urea hydrolysis [23], sugar and salt assimilation tests, and fermentations tests [22].

Screening of Proteolytic Yeasts

Strains isolated and identified were inoculated in the Manachini solution [16] containing gelatin (10% w/v) for 72 h at 30 °C and 150 rpm in orbital shaker. The cell-free supernatants were clarified by centrifugation (5,500×g) for 15 min, and 120 µL were placed into cup plates containing milk–gelatin–agar medium. Each plate was incubated for 3 days at 30 °C. The ability to digest casein was demonstrated by the formation of transparent halos on the plates and expressed in millimeter. Positive strains were submitted to the quantitative test described by Moreira et al. [8]. The proteolytic activity was determined in the cell-free supernatants obtained by centrifugation (5,500×g) for 5 min at 25 °C. A 1.0% (w/v) Azocasein (Sigma Chemical, St. Louis, MO) in 0.1 M Tris–HCl buffer solution (pH 7.6) containing 1 mM CaCl2 was used as substrate [24]. One activity unit was defined as the amount of enzyme that produced a 1.0 increase in the optical density within 1 h at 440 nm. The residual protein concentration after the proteolytic activity assay was determined according to Bradford [25], using bovine serum albumin as a standard.

Fermentation Conditions

The fermentations were carried out in Erlenmeyer flasks (250 mL) containing 50 mL of the culture medium described by Porto et al. [7], consisting in 0.1% (w/v) NH4Cl, 0.06% (w/v) MgSO4⋅7H2O, 0.453% (w/v) K2HPO4, and 0.1 mL mineral solution (1 mg/mL FeSO4⋅7H2O, 1 mg/mL MnCl2⋅4H2O, 1 mg/mL ZnSO4⋅H2O and 1 mg/mL CaCl2) and SF at concentration according to the selected full factorial design described later. After inoculating cells up to a concentration of 106 cells/mL in Erlenmeyer flasks containing the production medium, fermentations were performed in an orbital shaker at 200 rpm for 72 h. Samples were collected every 24 h and assayed for pH, total protein, proteolytic activity, and biomass concentration by dry weight.

Experimental Design and Modeling of Enzymatic Activity

Runs were carried out according to a 23 full factorial design in order to identify and quantify the effects and interactions of SF concentration, initial pH of the medium, and temperature (independent variables) on the enzyme activity (response), taken as an indirect measure of protease production. To this purpose, a set of ten experiments with two replicates at the central points was performed. The range and levels of the components (factors or independent variables) under study are given in Table 1.

Table 1 Factor levels employed in the 23 design used for the study on protease production by Candida buinensis.
Full size table

The following linear regression model was employed to predict the response:

$$ Y = {b_0} + \sum {{b_i}{x_i} + \sum {{b_i}{x_j} + \sum {{b_{ij}}{x_i}{x_j}} } } $$
(1)

where b 0 is the interception coefficient, b i are the linear coefficients, b ij are the interaction coefficients and x i and x j are the coded values of the independent variables.

The goodness of fit of the model was evaluated by the determination coefficient (R 2) and the analysis of variance (ANOVA) at a confidence level of p ≤ 0.05. All statistical and graphical analyses were carried out using the “Statistic 8.0” software (StatSoft, Tulsa, OK).

Effects of the pH and Temperature on the Enzymatic Activity and Stability

For the determination of the optimum pH of the enzyme, the pH of the reaction mixture containing buffer and 1% (w/v) azocasein was varied over the range 5.2–9.0. The buffers used were 0.1 M citrate–phosphate (pH 5.0–6.2), 0.1 M sodium phosphate (pH 5.8–8.0), and 0.1 M Tris–HCl (pH 7.2–9.0). The pH stability was determined by pre-incubating the cell-free supernatants into the selected buffer in the same pH range. For the determination of optimum temperature, the reaction mixture containing 1% (w/v) azocasein in 0.1 M Tris–HCl buffer (pH 7.6) and the cell-free supernatants were incubated over a temperature range of 25–80 °C. Tests of activity and stability at variable pH and temperature were performed, varying the incubation time from 30 to 120 min. After incubation, the samples were submitted to determination of residual proteolytic activity.

Kinetic and Thermodynamic Parameters of Thermal Inactivation

The progressive inactivation of proteases in cell-free supernatants was kinetically described as a first-order irreversible reaction, whose temperature-dependent rate constant, k (min−1), was estimated at each temperature from the semi-log plot of the residual activity with respect to the initial one (ψ) versus time (t):

$$ \ln \,\psi = - k\,t $$
(2)

On the other hand, all the thermodynamic parameters of thermal inactivation of proteases were estimated by a combination of the well-known Arrhenius and Eyring equations.

In particular, the activation energy, E*(kJ/mol), has been estimated by regression from the slope of the straight line obtained plotting lnk versus 1/T, according to Arrhenius:

$$ \ln {k} = \ln A - \frac{{E*}}{{RT}} $$
(3)

where A is the frequency factor or Arrhenius constant (min−1), R the ideal gas constant (8.3145 J/mol K), and T the absolute temperature (K).

The related enthalpy (ΔH*) and entropy (ΔS*) of activation were estimated by regression analysis of ln(k/T) versus 1/T [26], according to the equation of Eyring [27] derived from the transition state theory:

$$ \ln \frac{{{k}}}{T} = \ln \frac{{{k_B}}}{h} + \frac{{\Delta S*}}{R} - \frac{{\Delta H*}}{{RT}} $$
(4)

where k B is the Boltzmann constant (1.381∙10−23 J/K) and h the Plank’s constant (1.104∙10−35 J min).

Finally, the free energy of activation (ΔG*) was estimated from the corresponding values of enthalpy and entropy by the equation:

$$ \Delta G* = \Delta H* - T\Delta S* $$
(5)

Results and Discussion

Yeast Isolation and Identification

A total of 51 microorganisms were isolated in the milk samples, and 17.6% of these were identified summarily as yeasts. The low yeast number in raw milk may be due to competitive utilization for the substrates especially by the quickly growing psychotropic bacteria or to inhibition by metabolites excreted by bacteria. The species isolated and identified were deposited in the URM collection of Federal University of Pernambuco, Brazil (Table 2).

Table 2 Identification and proteolytic activity of yeast species isolated from raw cow milk produced in Cachoeirinha-PE, Brazil.
Full size table

In this study, the Candida genus was the most frequently encountered one in the cow milk samples. These results are in accordance with those reported for fermented milks or handicraft cheese [15], milks of Apulia region, Italy [16], and African fermented milks [28, 29].

Selection of the Proteolytic Yeast

As shown by the transparent halos for casein degradation ability (Table 2), all the species isolated and identified were positive for qualitative proteolytic activity, even if the subsequent quantitative proteolytic tests were carried out only for the species showing positive halos with diameter larger than 10 mm. The Candida species tested showed satisfactory proteolytic activity (between 4.8 and 23.5 U/mg), a result that agrees with the ones obtained by Corbo et al. [16], who observed the ability of several Candida strains present in dairy products to hydrolyze the milk protein. Since the species that showed the highest proteolytic activity was C. buinensis (23.5 U/mg), it was selected for subsequent protease production in SF medium.

Protease Production

The results of protease production, in terms of protease activity in the cell-free supernatants obtained by C. buinensis fermentations performed according to the 23 full factorial design, are summarized in Table 3. The numbers from one to eight refer to the experimental conditions corresponding to combinations of the variables tested, while nine and ten to the central point.

Table 3 Results of protease production, in terms of protease activity in the cell-free supernatants, by Candida buinensis fermentations performed according to the 23 full factorial design.
Full size table

The highest protease production by C. buinensis (573 U/mL) took place after 48 h in the fermentations performed at 35 °C, pH 6.5, and 0.1–0.5% (w/v) SF; therefore, such a fermentation time was always adopted for the statistical analysis. This activity was one order of magnitude higher than that reported for chymotrypsin-like protease by a mutant of Tricoderma harzianum (80 U/mL) after 2–3 days [30].

Some significant findings of the statistical analyses are evident in Fig. 1, where the Pareto chart shows the estimated effects of the independent variables (pH, temperature, and SF concentration) and of their interactions on the proteolytic activity, according to a decreasing order of magnitude. Whereas the length of each bar is proportional to the standardized effect, bars extending beyond the vertical line correspond to the effects statistically significant at a confidence level of 95%. It should be noted that only the pH main effect (3) and its interaction with temperature (1*3) were significant.

Fig. 1
figure 1

Pareto chart showing the individual effects of variables (1 temperature, 2 soybean flour concentration, 3 pH) and their interactions (1*2, 1*3, 2*3, 1*2*3) on the extracellular proteolytic activity of C. buinensis cell-free supernatants

Full size image

In particular, the negative sign of the pH effect means that protease production was favored by a pH decrease and achieved maximum values (547–573 U/mL) at pH 6.5. Moreover, the negative effect also of the interaction between pH and temperature means that the simultaneous decrease in the levels of both independent variables led to an improvement in protease production.

This behavior is consistent with the well-known strong dependence of cell growth and enzyme production on the extracellular pH [20], although optimum pH is well known to differ from one organism to the other. In particular, Chi et al. [2] observed that a decrease in the initial pH of culture medium improved the release of extracellular alkaline protease by A. pullulans, with the highest yield occurring at pH 6.0.

Contrarily to the generally accepted opinion that extracellular protease secretion by microorganisms is substantially influenced by medium components, including carbon source, nitrogen source, and trace elements [31], the level of SF, utilized in the present study as nitrogen, vitamin, and carbon source, did not significantly influence protease production according to the full factorial design. The protease activity by C. buinensis was in fact the same (573 U/mL) at both tested SF levels (0.1% and 0.5%) at pH 6.5 and 35 °C. This result, which suggests that the highest protease production could be ensured by a minimum amount of this component, is of a certain industrial concern, taking into account that around 30–40% of the production cost of industrial enzymes can be accounted for the cost of the growth medium [32].

The value of the determination coefficient (R 2 = 0.80) indicates that the model explained 80% of the variability data. The statistical significance of the model (Eq. 1) was also confirmed by the low significant lack of fit (p = 0.06) and the very low pure error (4.5) (Table 4). Moreover, it was found to adequately fit the data at a confidence level of 95% (p ≤ 0.05). These results demonstrate that both temperature and pH were important independent variables governing the protease production by C. buinensis.

Table 4 Analysis of variance for protease production.
Full size table

Effect of pH on Activity and Stability

Proteases produced by C. buinensis were then submitted to tests of relative activity versus time performed in the pH range 5.2–9.0 in the presence of three different buffers.

As shown in Fig. 2, the proteolytic activity progressively increased with the pH, reached a maximum value (531 U/mL) at pH 7.2 in a Tris–HCl buffer, and then drastically fell under alkaline conditions to a minimum value at pH 9.0. Similar results were found by Bolumar et al. [33] for protease produced by Debaryomyces hansenii that showed maximum activity at pH 7.5, although this enzyme was active in a broad range (pH 5.0–9.5). Contrarily to these results, optimum proteolytic activities under alkaline conditions were observed for A. pullulans (pH 9) [2], Candida albicans URM3622 (pH 8.2) [20], and Bacillus sp. strain MO-1 (pH 8.0) [34].

Fig. 2
figure 2

Effect of pH on the proteolytic activity of C. buinensis cell-free supernatants. Proteolytic activity is expressed as percentage with respect to its maximum value at pH 7.2 (531 U/mL)

Full size image

In the present work, we also observed retention of more than 98% of the maximum activity at pH 6.0 after 30 min and 42% at pH 7.5 after 90 min, and negligible activity at pH 5.8 and 9.0 (Fig. 3). These optima are quite different from the acidic and alkaline proteases reported by Nakayama et al. [35] for Bacillus sp. strain NTAP-1 and by Lima et al. [20] for C. albicans, respectively, but in reasonable agreement with those of common bacterial collagenases, which are known to have optimum pH in the range 7–8 [36].

Fig. 3
figure 3

pH stability profile of C. buinensis protease in cell-free supernatants after 30, 60, and 90 min of incubation. Values are expressed as percentage residual activity with respect to the initial one

Full size image

Effect of Temperature on Activity and Stability

Figure 4 shows that the cell-free supernatants activity progressively decreased with increasing temperature within the selected range; the maximum proteolytic activity (534 U/mL) did in fact occur at the lowest temperature (25 °C). This temperature is quite different from the optimum values reported for different proteases from Candida caseinolytica (37 °C) [3], D. hansenii (37 °C) [33], A. pullulans (45 °C) [2], and C. albicans (45 °C) [20].

Fig. 4
figure 4

Effect of temperature on the proteolytic activity of C. buinensis cell-free supernatants. Proteolytic activity is expressed as percentage with respect to its maximum value at 25 °C (534 U/mL)

Full size image

Figure 5 shows the semi-log plots of extracellular protease thermal stability expressed as residual activities at different temperatures with respect to the initial ones. Up to 30 °C, the residual activity after 90 min kept higher than 68%, whereas it fell to less than 6% for T ≥ 60 °C. Beyond 60 °C, a drastic reduction was observed, with the residual activity falling to zero at 80 °C, thereby highlighting complete thermal inactivation. Ramírez-Zavala et al. [14] reported that the activity of a serine carboxypeptidase from Kluyveromyces marxianus decreased significantly over 60 °C, with almost complete inactivation (>99%) after 30 min at 55 °C. Ma et al. [37] observed that an alkaline protease from A. pullulans was stable up to 20 °C, whereas inactivated rapidly over 45 °C and totally at 50 °C for 30 min.

Fig. 5
figure 5

Semi-log plot of the residual activity of C. buinensis proteases versus time in cell-free supernatants at different temperatures

Full size image

From the slopes of these straight lines, we estimated with fair correlation (0.86 ≤ R 2 ≤ 0.96) the values of the first-order kinetic constant (k), which varied from 0.039 min−1 at 25 °C to 0.0498 min−1 at 80 °C, corresponding to half-lives of 178 and 13.9 min, respectively. These values confirm the satisfactory thermostability of proteases in the cell-free supernatants at temperatures not exceeding 30 °C.

As explained in “Materials and Methods”, the estimation of the thermodynamic parameters of the system required the elaboration of the Arrhenius and Eyring type plots illustrated in Fig. 6. In particular, according to Eq. 3, the Arrhenius plot of lnk versus 1/T allowed estimating the activation energy of protease thermal inactivation in the cell-free supernatants.

Fig. 6
figure 6

Arrhenius (white circle) and Erying (filled square) plots of thermal stability of C. buinensis proteases in cell-free supernatants

Full size image

Table 5 lists the activation parameters referred to the above-supposed irreversible first-order inactivation. The activation enthalpy exhibited a positive value (ΔH* = 37.3 kJ/mol), consistently with the fact that the enzyme inactivation is an endothermic process. Moreover, the activation entropy was negative (ΔS* = −197.5 J/mol K) as expected by the formation of a transition state with more rigid structure compared to the ternary one of the stable proteases. As previously suggested, to explain similar thermodynamic profile of clavulanic acid degradation (ΔH* = 36.5 kJ/mol; ΔS* = −219.7 J/mol K; ΔG* = 103.5 kJ/mol), protease inactivation could have been influenced by some interaction with other substances either initially present in the medium or produced by the microorganism [38]. Furthermore, the activation energy (40 kJ/mol) well compares with that reported for clavulanic acid degradation (E* = 39 kJ/mol), which proves the general validity of the proposed inactivation model.

Table 5 Thermodynamic parameters of thermal inactivation of protease from Candida buinensis in cell-free supernatants.
Full size table

Conclusions

The isolation and identification of proteolytic yeast strains from raw cow milk allowed selecting a strain of C. buinensis as the best protease producer. The influence of three independent variables (temperature, pH and SF concentration) and interactions among them on the extracellular protease production by this yeast was investigated through a 23 full factorial design. The thermodynamic parameters of the enzyme thermoinactivation were also estimated by a combination of the Arrhenius and Eyring models. The protease exhibited important properties such as broad stability ranges of pH and temperature. Due to these properties, they could be profitably exploited in industrial applications as an environmentally friendly hydrolysis agent alternative to the chemical ones. Then, the use of an inexpensive medium as cheap nitrogen source, as the SF proposed in this study, could be of great significance to scale up this process to the industrial scale.