**Surfactant Imprinting Hyperactivated Immobilized Lipase as E**ffi**cient Biocatalyst for Biodiesel Production from Waste Cooking Oil**

#### **Huixia Yang and Weiwei Zhang \***

State Key Laboratory of High-efficiency Utilization of Coal and Green Chemical Engineering, School of Chemistry and Chemical Engineering, Ningxia University, Yinchuan 750021, China; yhx6668297@sina.com **\*** Correspondence: zhangww@nxu.edu.cn; Tel.: +86-0951-2062004

Received: 12 October 2019; Accepted: 29 October 2019; Published: 1 November 2019

**Abstract:** Enzymatic production of biodiesel from waste cooking oil (WCO) could contribute to resolving the problems of energy demand and environment pollutions.In the present work, *Burkholderia cepacia* lipase (BCL) was activated by surfactant imprinting, and subsequently immobilized in magnetic cross-linked enzyme aggregates (mCLEAs) with hydroxyapatite coated magnetic nanoparticles (HAP-coated MNPs). The maximum hyperactivation of BCL mCLEAs was observed in the pretreatment of BCL with 0.1 mM Triton X-100. The optimized Triton-activated BCL mCLEAs was used as a highly active and robust biocatalyst for biodiesel production from WCO, exhibiting significant increase in biodiesel yield and tolerance to methanol. The results indicated that surfactant imprinting integrating mCLEAs could fix BCL in their active (open) form, experiencing a boost in activity and allowing biodiesel production performed in solvent without further addition of water. A maximal biodiesel yield of 98% was achieved under optimized conditions with molar ratio of methanol-to-WCO 7:1 in one-time addition in hexane at 40 ◦C. Therefore, the present study displays a versatile method for lipase immobilization and shows great practical latency in renewable biodiesel production.

**Keywords:** biodiesel; waste cooking oil; lipase immobilization; interfacial activation; functionalized magnetic nanoparticles

#### **1. Introduction**

Over the past decades, biodiesel has attracted great interest as a sustainable alternative for fossil fuels in virtue of the depletion of fossilized fuel resources and their environmental impacts [1]. Biodiesel is a renewable and clean energy, and possess favorable advantages in combustion emission like low emissions of CO, sulfur free, low hydrocarbon aroma, high cetanenumber, and high flash point [2].

The conventional chemical technologies for biodiesel production involve the use of acid or basic catalysts (i.e., NaOH, KOH, and H2SO4), thus numerous disadvantages are inescapable, for example, high corrosive procedure, high energy consumption, high quantities of waste pollution, and costly in efficient product separation processes [3]. Furthermore, in order to prevent the hydrolysis reaction and saponification, high quality oils are required, with low contents of water and free fatty acids [4].

Feedstocks used for biodiesel can be allocated five categories, including edible vegetable oils, non-edible plant oils, animal fats, microalgae oils, and waste oils [5]. The global application of first-generation biodiesel produced by using edible oils, was restricted due to food scarcity and high cost of the edible oils [6]. Biodiesel production from waste cooking oils (WCO) could be a promising and cost effective candidate in handling issues associated with energy crisis, environmental concerns, and total cost reduction of biodiesel production [7]. Moreover, 15 million tons of WCO are produced annually throughout the world [8], bringing great challenge in reasonable management

of such oils on account of environment concerns [9]. However, using WCO as raw material is quite challenging as it contains a high amount of free fatty acids (FFAs) and water which could hinder the homogeneous alkaline-catalyzed transesterification in conventional biodiesel production processes [10]. Complete conversion of these low-quality feedstocks like WCO could be accomplished in enzymatic biodiesel production without saponification. Therefore, enzyme-catalyzed transesterification has become a laudable potential alternative for biodiesel synthesis.

Particularly, lipases are foremost and efficient enzymes implemented in biodiesel production. Lipase-catalyzed process exhibits key advantages such as no soap formation, high-purity products, easy product removal, adaptable to different biodiesel feedstock, environmentally friendly, and mild operating conditions [5]. However, the commercialization of enzymatic biodiesel production remains complicated, because of high price and low stability of lipases as well as low reaction rate of biocatalysis. Heterogeneous enzyme-catalyzed transesterification using immobilized lipases is a possible solution to these problems [11].

Immobilization of enzymes has been investigated for many years, and lipase can generally be immobilized by various techniques such as cross-linking, adsorption, entrapment and encapsulation [12,13]. Thereinto, cross-linked enzyme aggregates (CLEAs) is a cheap and efficient strategy for enzyme immobilization, which has broad applicability over numerous enzyme classes. Owing to its outstanding resistance to organic solvents, extreme pH, and high temperatures, CLEAs has attracted growing attention in cost effective biocatalysis [14]. Nevertheless, small particle size and low mechanical stability of CLEAs could directly affect mass transfer and stability under operational conditions, thus accordingly cause problems in practical use [15]. An alternative approach for circumventing compressed construction of CLEAs is to use "smart" magnetic CLEAs (mCLEAs). Magnetic nanoparticles (MNPs) could provide enhanced stability over repeated uses, especially for enzymes having low amount of lysine residues on their surface. Besides, mCLEAs could perform easily separation using a permanent magnet, affording novel combinations of bioconversions and down-streaming processes, thus provide the necessary reduction in enzymecosts to enable commercial viability.

Among various types of nanomaterials, MNPs have attracted substantial attention in enzyme immobilizations. However, bare MNPs tend to aggregate due to their high surface energy and are easily oxidized in the air leading to loss of magnetism and dispersibility, thus limiting their exploitation in practical applications [16]. The surface modification with an organic or an inorganic shell is an appropriate strategy to address these issues. Due to their excellent biocompatibility, slow biodegradation, high surface area-to-volume-ratio, and unique mechanical stability, Hydroxyapatite (HAP) could be a proper inorganic surface coating for MNPs [17]. Moreover, HAP-coated MNPs can be easily functionalized with organosilanes, and consequently has great application potential in enzyme immobilization.

*Burkholderia cepacia* lipase (BCL)is one of the most widely used lipases in biocatalysis [18]. On account of its versatility to accommodate a wide variety of substrates, high heat resistance, and good tolerance to polar organic solvents, BCL has been extensively used in various biotechnological processes, especially for biodiesel production. The active site of BCL is shielded by a mobile element, called the lid or flap [19]. The displacement of lid or flap to closed or open position, which directly impacts the accessibility of active site, determines the enzyme in an in active or active conformation. In general, substrate access to the underlying active site is prohibited in its closed configuration. As the stabilization of the open conformation of all lipases could remarkably increase their catalytic activity, a favorable method to obtain highly active biocatalysts should try to immobilize lipases in their most active form (open conformation).

Generally, the preparation of immobilized enzyme with enhanced activity and stability is a persistent goal of the biotechnology industry to seek maximum profit. Therefore, developing a simple and efficient approach for lipase interfacial activation in immobilization is highly desirable. Bioimprinting is a commonly used method for achieving hyperactivation of lipases in organic media. The principle of bioimprinting is to "anchor" the enzyme in its active form, which could be achieved by binding with imprint molecules (such as surfactants, natural substrates, substrate analogs etc.). From an applied point of view, the dramatic hyperactivation of lipases by low concentrations of surfactants is an expeditious and facile method for lipase interfacial activation [20].

To develop an efficient and environmentally benign process for the biodiesel production from waste cooking oils, in the present study surfactant imprinting strategy on BCL was implemented in combination with mCLEAs immobilization using HAP-coated MNPs. Subsequent cross-linking could "lock" BCL in its favorable conformation, while HAP-coated MNPs could facilitate the recovery of immobilized BCL and simplify the biodiesel purification. To the best of our knowledge, this is the first report on BCL immobilization integrating surfactant imprinting and mCLEAs. The optimal conditions for mCLEAs preparation, along with the effect of different surfactants (anionic, cationic, and non-ionic) on the catalytic activity of BCL mCLEAs in transesterification were studied. The optimized surfactant-activated BCL mCLEAs was further used in transesterification of waste cooking oils to biodiesel. In addition, a detailed analysis of solvents, methanol-to-oil molar ratio, and temperatures on the yield of biodiesel production was presented. The results obtained in the research are expected to provide a reliable basis for further exploration of lipase immobilization and efficient biodiesel production in industry.

#### **2. Results and Discussion**

#### *2.1. Preparation and Characterization of Immobilized Lipase*

In this study, the prepared MNPs encapsulated by hydroxyapatite (HAP) were used as immobilization supports. The amino functionalization of HAP-coated MNPs was carried out using 3-aminopropyltrimethoxysilane (APTES) for efficient enzyme attachment. Typically, the preparation procedure of immobilization supports and surfactant-activated BCL mCLEAs were performed according to the scheme shown in Scheme 1. The prepared magnetic supports and immobilized BCL were characterized by fourier transform infrared spectroscopy (FT-IR), transmission electron microscope (SEM) and vibrating sample magnetometer (VSM).

FTIR characterization was performed to investigate the chemical composition of functionalized MNPs and immobilized BCL. Spectra were recorded on over the region from 4000 to 400 cm−1. As shown in Figure 1, the strong peak at 588 and 639 cm−<sup>1</sup> corresponds to the stretching vibration of Fe-O bond. The characteristic absorption bands related to the HAP appease at 565 and 1044 cm<sup>−</sup>1, which are assigned to phosphate groups [21]. In the IR spectrum of modified MNPs and BCL mCLEAs, the characteristic absorption bands related to the functional groups of HAP emerged clearly, which demonstrated the successful incorporation of MNPs with HAP. For all immobilized lipases, including BCL CLEAs, BCL mCLEAs and surfactant-activated BCL mCLEAs, the typical IR bands responsible for the lipase that were chemically covalent-bonded to the functionalized MNPs were observed at 1642 cm−<sup>1</sup> for amide I (C=O stretching vibration) and at 1539 cm−<sup>1</sup> for amide II (N-H bending vibration), respectively. Besides, compared with the results shown in Figure 1, aliphatic C-H stretch band at 2859 and 2927 cm<sup>−</sup>1, corresponding to C-H stretching vibrations, are clearly observed in all immobilized lipases, which also indicated the successful loading of lipase.

**Figure 1.** Spectra of (**A**) Fe3O4 MNPs, (**B**) hydroxyapatite coated magnetic nanoparticles (HAP-coated MNPs), (**C**) 3-aminopropyltrimethoxysilane (APTES)-HAP-coated MNPs, (**D**) BCL CLEAs, (**E**) BCL mCLEAs, (**F**) Triton-activated BCL mCLEAs.

In order to assess morphology, size and composition of functionalized MNPs and immobilized BCL, SEM images were collected and illustrated in Figure 2. As seen in Figure 2, bare Fe3O4 MNPs formed significantly dense agglomeration, because of their high surface energy and strong dipole-dipole interactions. It is obvious that the structure of Fe3O4 MNPs becomes looser and more evenly distributed after being functionalized with HAP (Figure 2B) and APTES (Figure 2C), suggesting that surface modification is favorable for preventing aggregation of Fe3O4 MNPs. At the same time, the rough surface of Fe3O4 MNPs also increased the surface area for attachment of enzyme.

The crucial structure factors in aggregated-based enzyme immobilization, including morphological topographies, structural arrangement and size, play an important role in affecting substrate affinity and stability of biocatalyst [22]. Besides, the particle size of enzymes is an important property of any heterogeneous catalysis since it can directly affect the diffusion of substrates and catalytic efficiency, especially in the internal enzymes of highly compact aggregates [23]. SEM images (Figure 2D) of standard BCL CLEAs revealed no defined morphologies and large size particles. Moreover, standard BCL CLEAs presented a uniform and compact surface with the presence of few tiny pores. On the contrary, after the incorporation of functionalized MNPs, BCL mCLEAs formed spherical structures and small particle sizes, which could reduce inner steric hindrance in closely packed CLEAs. It is noteworthy that the presence of functionalized MNPs displayed large active surface available for lipase immobilization, therefore were important for development of a stabilized enzyme-matrix. Furthermore, a loose and homodispersed structure of Triton-activated BCL mCLEAs was found in Figure 2F, suggesting that the formation of large aggregates were forbidden by the imprinting of surfactants. From the SEM outcomes, it can be discerned that, thanks to the coating of surfactants, lipase could be uniformly dispersed on functionalized MNPs, which could contribute to a wider

surface area with more catalytic sites and decrease the diffusion limit. Consequently, compared with standard BCL CLEAs, Triton-activated BCL mCLEAs could perform superior catalytic efficiency.

**Figure 2.** Images of(**A**) Fe3O4 MNPs, (**B**) HAP-coated MNPs, (**C**) APTES-HAP-coated MNPs, (**D**) BCL CLEAs, (**E**) BCL mCLEAs, (**F**) Triton-activated BCL mCLEAs.

The magnetic property of functionalized MNPs and immobilized BCL were measured using VSM. The hysteresis curves of the Fe3O4 MNPs, HAP-coated MNPs, APTES-HAP-coated MNPs, BCL mCLEAs and Triton-activated BCL mCLEAs shown in Figure 3, exhibited a perfect sigmoidal behavior, corresponding to a typical superparamagnetism.

**Figure 3.** Hysteresis loops of Fe3O4 MNPs, HAP-coated MNPs, APTES-HAP-coated MNPs, BCL mCLEAs and Triton-activated BCL mCLEAs. The inner shows the easy magnetic separation of Triton-activated BCL mCLEAs in reaction mixture.

With further functionalization of MNPs, the saturation magnetization value decreased and correlated with the increase of the core-shell layer. Interestingly, it is obviously observed that the saturation magnetization value of Triton-activated BCL mCLEAs increased visibly compared to BCL mCLEAs. It might be due to the uniform dispersion of lipase on MNPs and availability of large surface area which decreased the shielding-effect of the out layer substances. As seen in Figure 3 (inner), Triton-activated BCL mCLEAs showed fast response (6s) to the external magnetic field andcould be easily recovered from the reaction mixture. After removing the external magnetic field, the magnetic

immobilized BCL redispersed rapidly by a slight shake, indicating good dispersion and efficient recyclability in industrial application.

#### *2.2. Optimization of the Immobilization Conditions*

In this study, the enzymes were precipitated by adding water-miscible organic solvents (acetone, ethanol and 2-propanol), PEG 800 and ammonium sulfate. The optimum precipitant was selected by measuring the transesterification activity of the corresponding BCL mCLEAs. Compared with free BCL, all BCL mCLEAs prepared using different precipitants performed higher transesterification activity in organic solvent. Among the protein precipitants evaluated, ammonium sulfate showed maximum recovery of activity (Figure 4a), therefore was further used in BCL immobilization.

**Figure 4.** (**a**) Precipitant type and (**b**) glutaraldehyde concentration on the activities of BCL mCLEAs.

Traditionally, glutaraldehyde has been extensively used as the cross-linking agent to prepare CLEAs of various enzymes and exhibited a strong effect on activity and particle size of enzyme aggregates. The activity recovery of CLEAs greatly depends on the type of enzyme and the concentration of glutaraldehyde [24]. Lower glutaraldehyde concentration affects the cross-linking efficiency, which might result in enzyme leakage in immobilization, while excessive glutaraldehyde can induce the flexibility of enzymes and the active site availability, consequently, decreasing the activity recovery of CLEAs [25,26]. In this study, the influence of glutaraldehyde concentration on activity of BCL mCLEAs was investigated by using various concentrations of glutaraldehyde in cross-linking. As shown in Figure 4b, the optimum glutaraldehyde concentration of BCL mCLEAs was 2.0% (v/v).

#### *2.3. Hyperactivation of BCL mCLEAs with Surfactants*

A pivotal challenge in lipase immobilization is to open the lid of lipases and fix their open form for the exposure of active site. Surfactant imprinting is an efficient approach to activate lipases by facilitating lid-opening. Like other lipases, BCL also consists of a mobile element at the surface, which composed of two helical elements (a5- and a9-helices) and covers the active site [18]. To improve the catalytic performance of BCL mCLEAs, BCL was imprinted in the presence of surfactants prior to immobilization. Thus, four different surfactants with different properties (cationic, anionic and non-ionic) were investigated for modulating the activity of BCL mCLEAs in biodiesel production. As seen in Figure 5, Triton X-100 exhibited maximum effect on the enhancement of lipase activity in low surfactant concentration, while the addition of sodium bis-2-(ethylhexyl) sulfosuccinate (AOT) showed the least influence. The optimal surfactant in proper concentration acting as a bipolar agent, could simulate the amphiphilic environment to benefit the exposure of hydrophobic regions in the active site. Meanwhile, surfactants may also promote the dissociation of large aggregates formed, thus slightly increase the enzymatic activity of lipase (Figure 2).

**Figure 5.** Four different surfactants activation on activity of surfactant-activated BCL mCLEAs in biodiesel production.

However, the increase of surfactant concentration led to gradual decrease of biodiesel yield in all cases, indicating that surfactants showed positive and negative effect on the activity of lipase. Additional detergent molecules may bind to the active site region of lipase, blocking the substrate access, inducing inhibition [27]. Compared with ionic surfactants (AOT and cetrimonium bromide (CTAB)), nonionic surfactants (Triton X-100 and Tween 80) were preferred aiming at regulating the activity of BCL (Figure 5). As the main interaction between the enzyme and nonionic surfactants is hydrophobic interaction while anionic or cationic surfactants perform electrostatic interactions [28], mild hydrophobic interaction between BCL and the surfactant might be important to trigger the interfacial activation mechanism. Therefore, nonionic Triton X-100 and Tween 80 were further studied to confirm the optimal amphiphile and surfactant concentration. As performed in Figure 6, the maximum hyperactivation of BCL mCLEAs was observed in the pretreatment of BCL with 0.1 mM Triton X-100, and the optimal Triton-activated BCL mCLEAs were used for further experiments.

**Figure 6.** Surfactants (Triton X-100 and Tween 80) concentration in surfactant-activated BCL mCLEAs preparation.

#### *2.4. Biodiesel Production*

For the economic feasibility of biodiesel production, solvents, methanol-to-oil molar ratio, and reaction temperature are important variables to optimize for transesterification step. As a result of oxidative reactions occurring during cooking and long-term storage in air, WCO generally exhibits a dramatic increase in viscosity and saponification value [7]. Compared to the fresh oil, high viscose WCO is not favored in biodiesel production. Using solvents could reduce the viscosity of the reaction medium and decrease the diffusion limitations, while it might also directly affect the enzyme structure and activity. In general, hydrophobic solvents could promote the interface and stabilize lipases on their open assembly, causing the hyperactivation of these enzymes. To select the most suitable medium, five hydrophobic solvents commonly used in transesterification were tested in biodiesel production (Figure 7a). It can be clearly seen that biodiesel yield is remarkably dependent on the type of solvent. Overall, Triton-activated BCL mCLEAs exhibited higher activity than BCL mCLEAs and free BCL in all the solvents tested, and the changing trend of their activity in various solvents was accord with

BCL mCLEAs and free BCL. In case of Triton-activated BCL mCLEAs, the best results were achieved using n-hexane with a yield of up to 94% biodiesel, which was 3.3-fold higher than that in free BCL catalyzed reaction. Interestingly, surfactant hyperactivation in combination with immobilization could fasten lipase in their active conformation, allowing biodiesel production performed in solvent without further addition of water, which was in accordance with earlier reports [29,30].

**Figure 7.** Reaction parameters on biodiesel production catalyzed by free BCL, BCL mCLEAs and Trion-activated BCL mCLEAs and reusability of immobilized BCL. (**a**) Solvents, (**b**) molar ratio of methanol to oil, (**c**) temperature, (**d**) reusability.

The methanol:oil molar ratio can have a significant effect on the reaction yield because excess methanol increases the reaction rate and drives high yield of biodiesel, while a high concentration of methanol leads to inactivation of lipases. In this study, experiments were performed at different molar ratios of methanol to WCO ranging from 3:1 (stoichiometric ratio) to 11:1 both in hexane and cyclohexane with methanol added only once. As shown in Figure 7b, Triton-activated BCL mCLEAs exhibited higher biodiesel yields in one-time addition of methanol under all the experimental conditions, especially when hexane was used as solvent. Meanwhile, owing to the significant inhibitory effect of excess methanol in one time addition, free BCL showed low activity in biodiesel production. It is worth noting that yields of Triton-activated BCL mCLEAs catalyzed reactions in hexane exceeded 90% in a wide range of methanol-to-WCO ratio, while the reaction yield of BCL mCLEAs catalyzed reaction decreased with methanol-to-WCO ratio exceeding 6:1. Consequently, it can be confirmed that surfactants pretreatment provided not only hyperactivation but also protection to lipases from denaturation in excess methanol. In addition, the maximum biodiesel yield was observed at a methanol-to-WCO ratio of 7:1 for Triton-activated BCL mCLEAs. Consequently, the minimal stoichiometric methanol-to-WCO ratio of 7:1 was chosen in further experiments.

Most of the enzymatic transesterification depends on temperature, which could enhance reaction rate and improve the dispersion of immobilized particles in reaction medium with better mass transfer between the reactants [31]. However, thermal denaturation of the enzyme might occur with elevation of temperature, typically according to the property of enzyme and immobilized methods. The effect of temperature on the yield of biodiesel during the transesterification of WCO has been investigated over a temperature range from 35 to 55 ◦C. According to Figure 7c, the optimum operational temperature for Triton-activated BCL mCLEAs and BCL mCLEAs was 40 ◦C, with biodiesel yields of 98% and 76% respectively after 24 h, and further increase of temperature will result in decrease of biodiesel yields simultaneously. Besides, Triton-activated BCL mCLEAs showed better activity below 40 ◦C, while BCL mCLEAs performed higher biodiesel yield over 40 ◦C. The suitable covalent cross-linking

with functionalized MNPs provided extra structure stabilization in mCLEAs, requiring much more energy to the disruption of this stable structure than free enzyme [32]. Nevertheless, the accessible active site of lipases achieved by surfactants pretreatment might be more sensitive to high temperature denaturation [33].

In summary, the optimal reaction conditions for Triton-activated BCL mCLEAs catalyzed transesterification of WCO are as follows: hexane used as solvent, molar ratio of methanol-to-WCO 7:1 in one-time addition, reaction temperature 40 ◦C. To verify the feasibility of the whole process at a larger scale, transesterification of WCO were performed under optimal conditions adding proper amount of Triton-activated BCL mCLEAs (the initial content of BCL was 240 mg in immobilization) to a mixture of 1 g WCO in 20 mL hexane. The biodiesel yield reached 94% after shaking at 40 ◦C for 48 h. Triton-activated BCL mCLEAs showed good activity and stability under higher oil content, indicating the possibility of its scale-up application in bioreactor systems.

#### *2.5. Reusability*

Reusability of immobilized enzyme is a chief criterion for its cost-effective use for potential industry applications. The utilization of functionalized MNPs facilitates the consequent reuse of immobilized enzyme. To investigate the reusability of BCL mCLEAs and Triton-activated BCL mCLEAs, the immobilized lipases were recovered by magnetic separation, and applied in the consecutive batches of biodiesel reactions under optimized conditions. Assessments of the operational stability were analyzed for 6 cycles and presented in Figure 7d. As observed, Triton-activated BCL mCLEAs showed no significant loss in the catalytic activity after subsequent consecutive reuse for 4 cycles, and kept 82% relative activity after continuous running 5 cycles. Meanwhile, the relative activity of BCL mCLEAs was 55% after 5 cycles, implying that BCL could possess good long-term stability with surfactant pretreatment. The protein denaturation in one time addition of methanol and byproduct inhibition might be account for the decrease in biodiesel yield in long-term reuses [34].

#### **3. Materials and Methods**

#### *3.1. Materials*

*Burkholderia cepacia* lipase (powder, Amano Lipase PS, ≥3000 U/g) and fatty acid methyl ester standards were purchased from Sigma-Aldrich (St. Louis, MO, USA). Also, 3-aminopropyl triethoxysilane (APTES), glutaraldehyde (25%, v/v) and 2-phenyl ethanol (>98%, CP) were supplied by Aladdin (Shanghai, China). Sodium bis-2-(ethylhexyl) sulfosuccinate (AOT) were procured from Acros (USA). Waste cooking oil (WCO) was obtained from local restaurant around Ningxia University campus (Yinchuan, China) with the following fatty acid compositions: 10.48% palmitic acid, 15.04% stearicacid, 38.44% oleic acid, 23.76% linoleic acid, and 1.72% linolenic acid. The WCO sample was filtered to separate impurities and solids in the oil. The physical properties of WCO are saponification value of 197.3 mg KOH/g, acid value of 4.37 mg KOH/g, and average molecular weight of 870.9 g/mol. All other chemicals were of analytical or chromatographical grade and used as purchased.

#### *3.2. Preparation of Magnetic Support*

Preparation of HAP-coated MNPs was carried out according to the previously reported method [35]. Initially, MNPs cores were prepared by the conventional co-precipitation method. Typically, FeCl2·4H2O (1.1 g) and (3.0 g) of FeCl3·6H2O were dissolved in 90 mL deionized water under the protection of argon, with subsequent addition of 25% ammonia solution (30 mL) under vigorous stirring at room temperature. After stirring for 30 min, a 60 mL aqueous solution composed of Ca(NO3)2·4H2O (7.1 g) and (NH4)2HPO4 (2.3 g) adjusted to pH=11 was added drop wise to the above suspension under continuous stirring. Subsequently, the resultant mixture was heated to 90 ◦C and stirred for 2 h. After cooling to room temperature and aging in the mother solution overnight, the obtained

precipitates were washed several times with deionized water until neutral and lyophilized for 12 h. The HAP-coated MNPs were obtained by calcining the materials in air at 300 ◦C for 3 h.

To obtain 3-aminopropyl trimetoxysilane functionalized HAP-coated MNPs (APTES-HAP-coated MNPs), HAP-coated MNPs (1.0 g) were suspended in a solution composed of 30 mL anhydrous toluene and 0.44 g of APTES. The mixture was refluxed under Ar atmosphere for 12 h, and then washed several times with ethanol, magnetically separated, and subsequently lyophilized prior to use.

#### *3.3. Lipase Immobilization*

BCL mCLEAs were produced according to the procedure described in Scheme 1. Firstly, 10 mg of APTES-HAP-MNPs were dispersed in 1 mL of BCL solution (10 mg/mL, 0.1 M phosphate buffer, pH 7.0) and shaken for 15 min at 30 ◦C. Then 5 mL of precipitant was added with stirring at 4 ◦C for 30 min. After precipitation, glutaraldyhyde was added drop wise into the suspension and stirred for 3 h at 30 ◦C. Afterwards, BCL mCLEAs were collected by centrifugation and washed thrice with phosphate buffer and deionized water, lyophilized and finally stored at 4 ◦C.

During optimization of the immobilization conditions, the effects of precipitants (acetone, ethanol, isopropanol, PEG 800 (1 g/mL), and saturated ammonium sulfate solution) and concentration of glutaraldehyde on the activity recovery of BCL mCLEAs were investigated.

The surfactant-activated BCL mCLEAs was prepared using cationic (CTAB), anionic (AOT) and nonionic (Tween 80 and Triton Triton X-100) surfactants at various concentrations. Then, 1 mL of BCL solution and appropriate amount of surfactant were mixed and stirred at 4 ◦C for 30 min. After incubated for 24 h at 4 ◦C, the suspended solution was sequentially used for BCL mCLEAs preparation under optimal conditions.

#### *3.4. Characterization*

The prepared support matrix and immobilized lipase described above were characterized using FTIR, SEM and VSM. The Fourier transform infrared (FTIR) spectra were acquired using a Perkin Elmer Frontier spectrometer (Spectrum Two, Waltham, MA, USA) equipped with an Attenuate Total Reflection (ATR) accessory. Samples were analyzed as KBr pellets in the range of 400 to 4000 cm−<sup>1</sup> at a resolution of 0.5 cm−1. The morphology of the particle surface was observed using a scanning electron microscope (SEM, Sigma HD, ZEISS, Germany), with deposition of a thin coating of gold onto the samples prior to analyses. The magnetic properties were detected by a vibrating sample magnetometer (VSM, MicroSense EZ9, Lowell, MA, USA) at room temperature.

#### *3.5. Activity Assay*

In studying the optimal conditions for BCL mCLEAs preparation, the enzymatic transesterification activities of free lipase and immobilized BCLs were assayed via transesterification reaction of 2-phenyl ethanol with vinyl acetate according to the method introduced previously [36]. The reaction mixture contained 10 mg of 2-phenylethanol, 1 mL of vinyl acetate and 10 mg of lipase (the initial content of BCL was 10 mg in preparing BCL CLEAs and BCL mCLEAs), and the reactions were carried out at 30 ◦C with continuous shaking at 220 rpm. After 24 h of reaction, samples were withdrawn and analyzed by high-performance liquid chromatography (HPLC). All experiments were repeated at least three times. The relative activity of BCL mCLEAs was calculated with the following equation:

$$\text{Relative activity } (\%) = \frac{\text{Transversity (i)} \text{ of immobtized BCL}}{\text{Transsterification yield of free BCL}} \times 100$$

#### *3.6. Enzymatic Transesterification for Biodiesel Production*

The transesterification of WCO were carried out at 40 ◦C in a 10 mL screw-capped vessel for 24 h with continuous shaking at 220 rpm. Unless otherwise stated, a typical reaction mixture consisted of 50 mg WCO, 2.0 mL hexane, 10 mg of lipase (the initial content of BCL was 10 mg in preparing BCL CLEAs and BCL mCLEAs) and methanol using methanol: oil molar ratio of 4:1. Single factor optimization was conducted to determine optimal reaction parameters for transesterification of WCO to biodiesel. Various conditions including kinds and concentration of surfactants, solvents, molar ratio of methanol to oil and temperature (◦C) were investigated. The transesterification reaction of large scale with 1 g WCO were carried out as described in Section 2.4. All biodiesel reactions were performed in dried solvents without any water added. The yield of biodiesel (20 μL) was analyzed in different time intervals using gas chromatography.

#### *3.7. Analytical Methods*

HPLC was conducted with Shimadzu LC-2010A HT apparatus using C18 column (UltimateXB-C18, 5 μm, 4.6 × 150 mm, Welch). The samples were analyzed with a mixture of MeOH/water = 80:20 (v/v) as eluent at 0.8 mL/min for 9 min at 254 nm.

Fatty acid methyl esters (FAMEs) were analyzed by a Fuli9790 plus gas chromatography (Fuli, Zhejiang, China) fitted with a flame ionization detectorcity (FID, Zhejiang, China), and a KB-FFAP column (30 m × 0.32 mm × 0.25 μm). Nitrogen gas was a carrier at continues flow of 1.0 mL/min. The oven (Zhejiang, China) temperature was set and at 160 ◦C maintained for 2 min, then a heating ramp was applied up to 240 ◦C at a rate of 10 ◦C /min, and the temperature of the oven was maintained at 240 ◦C for 15 min. The temperatures of the injector (Zhejiang, China) and the detector (Zhejiang, China) were set at 270 and 280 ◦C, respectively. Methyl tridecanoate was used as internal standard, and the biodiesel yield (%) was calculated by peaks area of standard FAME peaks.

#### *3.8. Reusability*

The reusability of Triton-activated BCL mCLEAs and BCL mCLEAs for the transesterification of WCO were also investigated under optimal conditions. After each batch reaction, immobilized BCL was recovered by magnetic separation and washed with n-hexane. The washed biocatalyst was reused consecutively in repetitive cycles. The biodiesel yield of the first reaction was set as 100% and the FAMEs yield in the subsequent reactions was calculated accordingly.

#### **4. Conclusions**

A facile and effectual surfactant imprinting method to expose the lipase active site integrating amino functionalized HAP-coated MNPs was established to immobilize CLEAs of BCL attaining enhanced activity and stability. The as-prepared Triton-activated BCL mCLEAs was subsequent processed in enzymatic transesterification of waste cooking oil for biodiesel production, and showed 98% biodiesel yield under optimal conditions, which was 5.3-fold higher than the free lipase. This study proved that hyperactivation with surfactant could significantly improve the resistance of lipase to methanol in one-time addition, when compared to BCL mCLEAs and free BCL. In addition, surfactant imprinting in combination with immobilization could fasten lipase in their active conformation, allowing biodiesel production performed in solvent without further addition of water, and thus displayed priority in downstream purification of biodiesel over ordinary immobilization methods. Besides, the green immobilization with functionalized MNPs facilitates fast and easy recovery of lipase, and the corresponding immobilized BCL was reused for 4 cycles without significant loss in the catalytic activity. Furthermore, this work provides a promising approach for immobilization of other lipases, which can be used with success in green and clean production processes.

**Author Contributions:** Conceptualization, W.Z.; Investigation, H.Y.; supervision, funding acquisition, writing—original draft preparation and review and editing, W.Z.

**Funding:** This research was funded by the National Natural Science Foundation of China (No. 21865023), the Natural Science Foundation of Ningxia (No. 2019AAC03022), Scientific Research Foundation of the Higher Education Institutions of Ningxia (No. NGY2017045), Major Innovation Projects for Building First-class Universities in China's Western Region (No. ZKZD2017003), and the Scientific Research Start Funds of Ningxia University Talent Introduction (No. BQD2015012).

**Acknowledgments:** The help from Fu Zheng (School of Physics & Electronic-Electrical Engineering, Ningxia University) for magnetization measurements is gratefully recognized.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Response Surface Methodology Approach for Optimized Biodiesel Production from Waste Chicken Fat Oil**

#### **Fatima Shafiq 1, Muhammad Waseem Mumtaz 1,\*, Hamid Mukhtar 2, Tooba Touqeer 1, Syed Ali Raza 3, Umer Rashid 4,\*, Imededdine Arbi Nehdi 5,6 and Thomas Shean Yaw Choong <sup>7</sup>**


Received: 1 February 2020; Accepted: 26 February 2020; Published: 6 June 2020

**Abstract:** Biodiesel is gaining acceptance as an alternative fuel in a scenario where fossil fuel reserves are being depleted rapidly. Therefore, it is considered as the fuel of the future due to its sustainability, renewable nature and environment friendly attributes. The optimal yield of biodiesel from cheap feed stock oils is a challenge to add cost effectiveness without compromising the fuel quality. In the current experiment, waste chicken fat oil was taken as the feedstock oil to produce biodiesel through the chemical and enzymatic route of transesterification. The process of chemical transesterification was performed using KOH and sodium methoxide, while enzymatic transesterification was done by using free *Aspergillus terreus* lipase and *Aspergillus terreus* lipase immobilized on functionalized Fe3O4 nanoparticles (Fe3O4\_PDA\_Lipase) as biocatalysts. The physico-chemical properties of the understudy feedstock oil were analyzed to check the feasibility as a feedstock for the biodiesel synthesis. The feedstock oil was found suitable for biodiesel production based upon quality assessment. Optimization of various reaction parameters (the temperature and time of reaction, catalyst concentration and methanol-to-oil mole ratio) was performed based on the response surface methodology (RSM). The maximum yield of biodiesel (90.6%) was obtained from waste chicken fat oil by using Fe3O4\_PDA\_Lipase as an immobilized nano-biocatalyst. Moreover, the above said optimum yield was obtained when transesterification was done using 6% Fe3O4\_PDA\_Lipase with a methanol-to-oil ratio of 6:1 at 42 ◦C for 36 h. Biodiesel production was monitored by FTIR spectroscopic analysis, whereas compositional profiling was done by GC–MS. The measured fuel properties—cloud point, pour point, flash point, fire point and kinematic viscosity—met the biodiesel specifications by American Society for Testing and Materials (ASTM).

**Keywords:** biodiesel; transesterification; immobilized lipase; RSM; fuel properties

#### **1. Introduction**

The rapid industrial growth and population explosion have built an immense pressure on natural resources, including fossil fuels. The whole world is determined to find suitable solutions in context with the forthcoming energy crisis. The world is in search of alternate sources of fuel to reduce its dependency on conventional fuels. Biodiesel has emerged as a promising alternative fuel in recent years due to its renewable nature and environment friendly attributes. Biodiesel may be characterized as alkyl esters of fatty acids and may be utilized easily in diesel engines without major alterations [1]. The emissions of CO and NOx from diesel burning are issues of keen interest as both are greenhouse gases and responsible for tropospheric ozone formation. It is an established fact based on the work of many researchers that, comparative to conventional diesel, combustion of biodiesel produces less CO and unburnt hydrocarbons but higher NOx emissions, probably due to a higher oxygen content in biodiesel [2,3].

Initially, synthesis of biodiesel was extensively carried out using vegetable oils and seed oils of non-edible origin. Usually, the production of biodiesel from edible oils is not cost-effective and these vegetable oils are used in food, hence are valuable. To avoid the problems associated with cost, edibility and food shortage, biodiesel production from non-edible fractions of food and related wastes is gaining sound gravity. Similarly, the biodiesel preparation from non-edible seeds like *Jatropha* is not completely feasible as the cultivation of non-edible seed oil plants may create competition with edible crops on shrinking fertile agriculture land [4]. Recently many studies were carried out for biodiesel synthesis from low cost vegetal and animal-based feed stocks like waste cooking oils and animal fats by reducing their viscosity through the transesterification process [5]. The chemical and enzymatic transesterification processes are adopted to convert fatty acids of feedstock into their alkyl esters. Both transesterification modes have their own modalities and advantages but may be optimized for high quality biodiesel [6].

Chicken fat is a poultry waste that can be used to produce biodiesel. The fat content in chicken is about 10% by weight, which is very high, and its cost is low. Commercial broiler chicken meat was reported to have relatively high contents of polyunsaturated lipids as compared with organic chicken [7]. Researchers have reported that chicken fat constitute about 25% to 35% saturated and 40% to 75% unsaturated fatty acids. Palmitic acid, along with stearic acid, linoleic acid and oleic acid, are major fatty acids in chicken fat [8,9]. The fats can be converted into alkyl esters by the process of transesterification. In an alkali-catalyzed transesterification reaction, both the glyceride and alcohol should be extensively free of water contents as the water compels the reaction to partly change into a saponification reaction, resulting in soap formation [6]. Sodium hydroxide and potassium hydroxide are commonly used as alkali catalysts, but they result in water formation during transesterification, that is why sodium and potassium methoxides are preferred for biodiesel production. Alkali catalysts are good especially for those feed stocks that contain minimal acid value. However, if the acid value of the feed stock is high, then it is recommended to perform pre-treatment acid esterification to reduce the free fatty acid contents before performing base-catalyzed transesterification of the feedstock [10].

On the other hand, enzyme-catalyzed transesterification is gaining acceptance and is considered technically comparable to alkali transesterification. This method normally employs lipase as a catalyst. Lipase-catalyzed transesterification of feedstock oils with a relatively higher free fatty acid content can be carried out without performing any pre-treatment acid esterification step that is normally required in case of alkaline transesterification [11]. However enzymatic transesterification is a high-cost process, because enzymes can be denatured easily in the presence of short-chain alcohols and it is difficult to recover [12]. To cope with these problems, enzymes are immobilized on various supports to enhance their durability. Immobilized enzymes are adoptable to harsh conditions as compared to the free enzymes and are easy to recover. Immobilization of enzymes on the matrix and beads may reduce the enzyme activity by blocking its active site and lowering the mass transfer. However, due to very small size and Brownian movement of nanoparticles, these are a potent choice for enzyme immobilization [13]. There are few reports on enzymatic transesterification of chicken fat oil [14,15].

In the present work the transesterification process was optimized to synthesize biodiesel from a cheaper source in the form of waste chicken fat. The relative effects of various catalysts and their concentration were studied and optimized for improved yields of biodiesel by involving the methanol-to-oil ratio along with reaction time and temperature. The synthesized biodiesel was also analyzed for fuel properties to check its feasibility for use in a compression ignition (CI) engine.

#### **2. Results and Discussion**

#### *2.1. Physico-Chemical Characterization of Waste Chicken Fat Oil (WCFO)*

The pre-analysis tests of WCFO revealed that the acid value of the oil was 6.56 ± 0.05 mg KOH/g, saponification value 200 <sup>±</sup> 7.50 mg KOH/g, refractive index 1.46 <sup>±</sup> 0.01, density 0.85 <sup>±</sup> 0.07 g/cm3, iodine number 75 ± 10.70 g iodine/100 g and the peroxide value was computed as (5.5 ± 0.50 meqO2/kg). These values were depicted comparable with that reported by previous studies [16]. Chicken fat oil has a high acid value, which is why the acid esterification of the feedstock was done prior to the alkaline transesterification to reduce the free fatty acid content and avoid saponification.

#### *2.2. Optimization of Biodiesel Production Process*

The experimental results obtained after performing reactions as per CCRD were statistically analyzed to select the most appropriate model from the linear, 2F1, cubical and quadratic models. The model that was best suited was chosen by considering the *p*-values, R<sup>2</sup> values, lack-of-fit tests and adjusted R<sup>2</sup> values. It was observed that the quadratic model was most suited for both the chemical and enzymatic routes of biodiesel production (Table 1).


**Table 1.** Summary of selected quadratic models.

The summary statistics clearly determined the fitness of quadratic models for chemical as well as enzymatic biodiesel production process for WCFO.

#### *2.3. Graphs of Predicted vs. Actual Values*

The predicted vs. actual value graphs for biodiesel yield depicts the fitness of the selected quadratic model. The graphs of predicted vs. actual values are shown in Figure 1, where Figure 1a–d describes the predicted vs. actual graphs based on experimental data about yield of biodiesel obtained through the transesterification of waste chicken fat oil by Fe3O4\_PDA\_Lipase (Figure 1a), *Aspergillus terreus* lipase (Figure 1b), sodium methoxide (Figure 1c) and KOH (Figure 1d). The distribution of the data along the straight line and the small difference between the predicted and actual value reveals the fitness of the quadratic model for all four experimental designs.

**Figure 1.** Graphs of predicted vs. actual yield for waste chicken fat oil-based biodiesel by Fe3O4\_PDA\_Lipase (**a**), *Aspergillus terreus* lipase (**b**), Sodium methoxide (**c**) and KOH (**d**).

#### *2.4. Optimization of Reaction Parameters for Manufacturing of Biodiesel Using Chicken Fat Oils*

The enzymatic transesterification of waste chicken oil using Fe3O4\_PDA\_Lipase as a bio-catalyst resulted in optimal biodiesel yield when transesterification reactions were performed by employing 6% Fe3O4\_PDA\_Lipaseconcentration with a 6:1 molar ratio of methanol to oil, at 42 ◦C for 36 h. While in case of enzymatic transesterification by *Aspergillus terreus* lipase, a 1% enzyme concentration, methanol-to-oil ratio of 6:1 and reaction temperature of 35 ◦C for 36 h were the optimal process conditions. However, when the sodium methoxide-catalyzed transesterification of WCFO was conducted, the optimum conditions for the reaction were a 1% catalyst level and a 6:1 methanol:oil mole ratio at 60 ◦C for a 1.25 h reaction time (Table 2). The optimum biodiesel yield in case of a potassium hydroxide (KOH)-catalyzed reaction was obtained at a 1% catalyst concentration, 1 h of reaction time, a methanol-to-oil ratio of 6:1 and 60 ◦C. Highest biodiesel yield was obtained for the nano-biocatalyst (Fe3O4\_PDA\_Lipase), which might be due to the high stability and activity of the immobilized enzyme at an elevated temperature and adoptability towards harsh conditions [17]. Moreover, the lipase can also convert the free fatty acids present in the feedstock to FAMEs. Lower yield obtained by the free lipase can be explained by the reduction of enzyme activity due to denaturation of the free enzymes at a higher temperature, which is required for biodiesel production from chicken fat oil, and the presence of short-chain alcohol [18]. For chemical transesterification, sodium methoxide was proven to be better than KOH, because sodium methoxide did not produce water, which might be responsible for saponification, and the separation of glycerol from biodiesel could be difficult, thus reducing process

efficiency. Comparable results for enzymatic and chemical transesterification of waste chicken fat oil have been reported in the published literature. Coppini et al. has reported a 90.61% biodiesel yield from chicken fat by using a 0.3 wt % NS-40116 enzyme, 1.5 of methanol:oil and 1.5 wt % water at 45 ◦C for 24 h [11]. Da Silva et al. has reported a 77% esterification yield by using 0.3 wt % lipase, 1:4.5 methanol:oil and 2 wt % water at 30 ◦C in 24 h [15]. Alptekin et al. has reported an 87.4% biodiesel yield from waste chicken fat using a 1% concentration of a KOH catalyst and a 6:1 methanol-to-oil ratio at 60 ◦C [19]. Mata et al. has reported a 76.8% biodiesel yield by transesterification of chicken fat using a 0.8% KOH catalyst, 6:1 methanol:oil at 60 ◦C for 2 h [20]. The few variations in the results are probably due to the different fatty acid profiles of chicken fats and different enzyme sources.

**Table 2.** Optimized factors for biodiesel synthesis via enzymatic and chemical modes of transesterification of chicken fat oil.


#### *2.5. ANOVA for Transesterification Data of WCFO*

The influence of various reaction parameters such as linear factors, 1st order interactions and quadratic expressions on percentage biodiesel yield are described in the ANOVA table (Table 3). The terms (a)–(d) represents the quadratic models based on findings of Fe3O4\_PDA\_Lipase, *Aspergillus terreus* lipase, sodium methoxide and KOH-catalyzed transesterification of WCFO, respectively. The statistical analysis depicted that the linear term, A—reaction time, had a significant impact for models a, b and c on biodiesel yield (*p* < 0.0001, 0.0003 and 0.0003, respectively), which were <0.05, while for model d it was not significant. The linear term B–reaction temperature, showed *p* values of 0.1743, <0.0001, 0.0004 and <0.0001 for models a, b, c and d, respectively. The Fe3O4\_PDA\_Lipase catalyzed transesterification was not affected significantly by temperature change in the selected range. Reaction temperature significantly affected the biodiesel yield for *Aspergillus terreus* lipase, which was temperature sensitive. The *p* values for the linear term C—CH3OH:Oil, was <0.05 for model (a) and (c) but it was 0.1524 for model (b) and 0.7970 for model (d), which is >0.05. D—catalysts/biocatalysts concentration, was proven to have a significant effect on biodiesel yield for all the four models. A previous report on *Jatropha curcas* seed oil transesterification showed the significant impact of catalyst concentration, methanol-to-oil molar ratio, reaction temperature and reaction time on biodiesel yields [21] In case of Model (a), the 1st order interaction terms AC, AD and CD were found to be significant having *p*-values of 0.0007, 0.0001 and 0.0018, respectively, which were less than 0.05; however, for Model (b) only BD and CD were found significant. In case of Model (c), the 1st order interaction variables, i.e., AD and CD, were significant with *p*-values of 0.0065 and 0.0017 being less than 0.05; for Model (d), only AC 1st order interactions were imparting a significant impact on biodiesel yield with p-values lower than 0.05. Where the quadratic terms C<sup>2</sup> and D<sup>2</sup> were significant for Models (a) having a *p* < 0.05, for Model (b) the statistical significance was noted among the quadratic terms B2, C<sup>2</sup> and D2. In Model (c), B2 and C2 were significant, while in the case of Model (d), A<sup>2</sup> and D2 were significantly affecting the biodiesel yield with *p* < 0.05.


**Table 3.** RSM-based ANOVA for transesterification of waste chicken fat oil (WCFO).

Note: Fe3O4\_PDA\_Lipase (a), *Aspergillus terreus* lipase (b), sodium methoxide (c) and KOH (d). SS stands for sum of squares and MS is mean square.

The 3D surface plots of the significant 1st order interaction terms are presented in Figure 2. Figure 2a–c presents the significant 1st order interaction terms of Model (a). Figure 2a shows the 3D surface plot between the methanol-to-oil ratio and reaction time; it reveals that the yield increases with an increase in reaction time and methanol:oil, but further increases in the methanol-to-oil ratio resulted in a decreased biodiesel yield. The joint impact of time and concentration to increase the biodiesel yield is given in Figure 2b. Figure 2c presents the 3D plot between bio-catalyst/enzyme concentration and methanol:oil for Model (a). The plot shows that enzyme concentration directly increases the biodiesel yield but the increase of metahonl:oil after a specific limit decreases the biodiesel yield.

Figure 2d,e presents the 3D response surface plots for Model (b). Figure 2d reveals the relation between catalyst concentration and reaction temperature; an increase in temperature decreases the biodiesel yield probably due to the denaturation of free enzyme. Figure 2e presents possible impact of methanol:oil and bio-catalyst/enzyme concentration on biodiesel yield.

Figure 2f,g are the 3D plots of the significant 1st order interaction terms of Model (c). Figure 2f presents the relation between catalyst concentration and reaction time. It is observed that increase in both parameters increases the biodiesel yield. Figure 2g shows the relation between the methanol-to-oil ratio and catalyst concentration. The catalyst concentration increased the biodiesel yield but by further increasing the methanol-to-oil ratio up to certain level, however beyond optimal level, a decrease in biodiesel yield was noted.

Figure 2h presents the response of surface plot on the only significant interaction term of Model (d), which is between the reaction time and methanol-to-oil ratio. It showed that biodiesel yield increased with time but after a specific period any further increase in time was not effective.

**Figure 2.** Response surface graphs for the significant 1st order interaction terms of Model (**a**) (A × C), (**b**) (A × D), (**c**) (C × D); Model (b), (**d**) (B × D)**,** (**e**) (C × D); Model (c), (**f**) (A × D)**,** (**g**) (C × D); Model (d), (**h**) (A × C).

#### *2.6. FTIR Spectroscopic Analysis of Feedstock Oil, Biodiesel and Composition of Fatty Acid Methyl Esters*

Asymmetric bending of the CH3 group was observed in the region between 1425 and 1447 cm−<sup>1</sup> and in the region ranging from 1188 to 1200 cm−<sup>1</sup> which were basic characteristic peaks of biodiesel. While the C=O stretch vibrations observed in the region between 1700 and 1800 cm−<sup>1</sup> and CH2 asymmetric and symmetric stretching vibrations appeared at 2900–3100 cm−<sup>1</sup> were present in FTIR spectra of both feedstock oil and synthesized biodiesel samples. However, signals in the 1390–1400 cm−<sup>1</sup> region confirmed the O–CH2 group and in the 1095–1101 cm−<sup>1</sup> region defined the asymmetric axial stretching of O–CH2–C for WCFO in FTIR spectra; however, these bands were absent in their respective biodiesel spectra. The above spectroscopic observations were according with the findings of a previous study [22]. The palmitic acid methyl ester (C16:0) 17.96%, stearic acid methyl ester (C18:0) 20.85%, oleic acid methyl ester (C18:1) 42.92% and linoleic acid methyl ester (C18:2) 16.54%, respectively, were the major FAMEs (Table 4). The current findings were found comparable with those reported by a previous study [16].

**Table 4.** The fatty acid methyl ester (ME) composition of the synthesized biodiesel.


#### *2.7. Fuel Characteristics of WCFO Biodiesel*

Fuel analysis plays a vital role in the evaluation of the manufactured biodiesel for its technical compatibility with conventional diesel. Fuel analysis of the understudy biodiesel samples was carried out in accordance with ASTM standard methods and the findings are mentioned below.

Kinematic viscosity is considered as one of the most significant fuel properties, as it is related to the resistance of flow that mainly occurs due to internal friction. If a biofuel contains greater values of kinematic viscosity, it will result in poor fuel atomization or incomplete combustion. Flash point and fuel volatility are inversely related to each other. Similarly, high values of cloud point generally result in problems such as fuel-line clogging. The kinematic viscosity (mm2/s), flash point (◦C), fire point (◦C), pour point (◦C) and cloud point (◦C) values for WCFO biodiesel are given in Table 5. The mentioned fuel-quality parameters were found comparable with those from previous studies [19,20].

**Table 5.** Fuel characteristics of WCFO biodiesel.


#### **3. Materials and Methods**

Chemicals and reagents of analytical grade were utilized during study and were procured from Sigma-Aldrich (Saint Louis, MO, USA) andMerck (Darmstadt, Germany). Lipase from *Aspergillus terreus* was produced through fermentation at the Institute of Industrial Biotechnology, GC University, Lahore, Pakistan. The nano-biocatalyst (Fe3O4\_PDA\_Lipase) prepared and characterized in our previous work has been used as the immobilized lipase for biodiesel production from chicken fat oil [17]. The chicken fat was collected from the local market of Gujrat City, Pakistan.

#### *3.1. Pre-Treatment of Feedstock*

The collected waste chicken fat was heated at 100 ◦C to convert it into liquid. The liquid was then filtered to remove the solid waste. Since the chicken fat oil contains a high free fatty acid (FFA) content, alkaline catalysts are not suitable for un-treated chicken fat oil. In this case, acid esterification was used to reduce the acid value before alkaline transesterification. For this purpose, the chicken fat oil was taken in three neck flasks equipped with a thermometer and a glass condenser. The third neck was used to withdraw the sample. Oil was homogenized by heating and stirring at 600 rpm. Briefly, 50 mg concentrated sulfuric acid and 2.25 g methanol for each gram of FFA present in the oil was mixed in a beaker. The acid value of the sample was checked after specific time intervals by taking small aliquots. The process was carried out till the acid number reduced to the desired value. After completion of the process, the mixture was put in a separating funnel. Three layers were formed after some time. The top layer consisted of unreacted methanol and the lower layer was water while the middle layer was fatty acid methyl esters (FAMEs) and oil. The middle layer was collected for chemical transesterification [23]. The enzymatic transesterification, however, was done without pre-acid esterification. The collected oil was subjected to analysis for some basic parameters, including saponification value, acid value (AV) and peroxide value. The degree of unsaturation of feedstock oil was determined by iodine number. Similarly, the specific gravity was also determined along with density and the refractive index.

#### *3.2. Experimental Design*

The Central Composite Response Surface Methodology (CCRD–RSM) was used to evaluate the impact of the different conditions, namely A) reaction time, B) reaction temperature, C) CH3OH: oil and D) the catalyst/biocatalyst's concentration on percentage biodiesel yield, for different transesterification routs using catalysts (KOH, CH3ONa, free *Aspergillus terreus*lipase and Fe3O4\_PDA\_Lipase). The ranges of the selected parameters for the four models are presented in Table 6.


**Table 6.** The ranges of reaction parameters: reaction time, temperature, CH3OH-to-oil ratio and enzyme concentration used for optimization studies of enzymatic and chemical transesterification.

In each case, thirty experiments were carried out as per CCRD factorial design. Chemically, this process was performed in three neck flasks equipped with a temperature regulator. A stirrer and reflux condenser were also attached with the flask. The reactions were accomplished at 500 RPM.

#### 3.2.1. Chemical Transesterification

Briefly, 50 g of pre-treated waste chicken fat oil (WCFO) in a flat-bottom three-neck flask was subjected to pre-heating for moisture removal. The molecular weight of the understudy chicken fat oil (873.72 g/mol) was calculated as per a previously reported method [24]. The next step involved the mixing of known amounts of methanol and catalysts, respectively. The resultant mixture was transferred gently to the flask. The whole mixture present in the flask was kept for heating with stirring. According to the CCRD design, the reaction conditions were maintained. The reaction was allowed to proceed, and on completion the mixture was separated into two layers. The upper layer was taken and further processed to obtain the refined biodiesel.

#### 3.2.2. Enzymatic Transesterification

For enzymatic biodiesel production, firstly waste chicken oil was mixed with methanol. A specific amount of immobilized lipase (Fe3O4\_PDA\_Lipase) for design (a) and free *Aspergillus terreus* lipase for design (b) was introduced to the oil/methanol solution and the reaction mixture was subjected to orbital shaking at 200 rpm with a 0.5% water content (with respect to oil), for a specific time period [25]. The CCRD was followed to set the alcohol to oil molar ratio, enzyme concentration, reaction temperature and time. After completion of the reaction, the glycerol was removed to obtain crude biodiesel, which was purified to get refined biodiesel. Magnetic nano-biocatalyst was separated from biodiesel and glycerol by magnetic decantation.

For the optimization studies, suitable statistical models based on experimental data were employed. Linear, 2FI, cubical and quadratic models were tested. The lack-of-fit test values, model significance (*p*-value), the R<sup>2</sup> and adjusted R2 values provided the base to select most appropriate statistical model. Finally, the response surface graphs were utilized to check the influence of the studied reaction conditions on the yield of biodiesel.

#### *3.3. Quantification and Characterization of Synthesized Biodiesel*

For the FTIR spectroscopic study, a Carry660 FTIR spectrophotometer (Agilent Technologies, Santa Clara, CA, United States) was used and FTIR spectra were drawn over 400–4000 cm−<sup>1</sup> scanning range.

The biodiesel from waste chicken oil was subjected to GC–MS analysis in order to evaluate the fatty acid methyl esters content (FAMEs). For this purpose, the GC–MS (QP 2010) instrument with a dB 5 column (Shimadzu, Japan) having diameter of 0.15 mm was used. The sample size (1.0 μL) was taken with a split ratio of 1:100, while a source of carrier gas, helium, was used having a 1.20 ml/min flow rate. The oven temperature was kept at 160.0 to 260.0 ◦C with a ramp rate of 4 ◦C per minute. The scanning of mass was done from 40.0 to 560.0 *m*/*z*. The detection of FAMEs was ascertained by comparing the relative retention time of each discrete FAMEs with reliable standards of FAMEs and by comparison with the NIST mass spectral library.

Fuel characteristics of the biodiesel were evaluated by some test experiments utilizing the ASTM standard procedure, i.e., cloud point (ASTM D 2500), viscosity (ASTM D 455), pour point (ASTM D 97) and flash point (ASTM D 93) [19]. The measurements were made in triplicate and the results were analyzed with the help of statistical tools.

#### **4. Conclusions**

The waste chicken fat oil was transformed into biodiesel by alkaline and enzymatic transesterifications. The reaction time, temperature, methanol:oil ratio and catalyst concentration were selected for the process optimization. Among all the catalysts and enzymes used, Fe3O4\_PDA\_Lipase-catalyzed transesterification of the studied feedstock oil was proved to be the most efficient to give maximum biodiesel yield. On the other hand, in case of chemical catalysis, CH3ONa was proved to be better than KOH when chicken fat oil was used as the feedstock. FTIR spectroscopy and GC/MS characterization further confirmed biodiesel formation. The compositional profiles and fuel characteristics of the synthesized biodiesel showed a promising compatibility of WCFO as a potential candidate for biodiesel production for future fuel regimes.

**Author Contributions:** The idea and concept for the current research was floated by, F.S., T.T., M.W.M., H.M. and U.R.; the methodology was also designed by F.S., M.W.M. and U.R.; the enzyme production was carried out by H.M.; the writing of the original draft was prepared by F.S. with the help and supervision of M.W.M., H.M., S.A.R., T.S.Y.C.; and I.A.N. reviewed the manuscript for improvement of the final version. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** One of the authors acknowledges his gratitude to King Saud University (Riyadh, Saudi Arabia) for the support/technical assistance of this research through Researchers Supporting Project number (RSP-2019/80).

**Conflicts of Interest:** The author declares no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Review* **Challenges and Opportunities in Identifying and Characterising Keratinases for Value-Added Peptide Production**

**Juan Pinheiro De Oliveira Martinez 1, Guiqin Cai 1, Matthias Nachtschatt 1, Laura Navone 1, Zhanying Zhang 1, Karen Robins 1,2 and Robert Speight 1,\***


Received: 8 January 2020; Accepted: 24 January 2020; Published: 3 February 2020

**Abstract:** Keratins are important structural proteins produced by mammals, birds and reptiles. Keratins usually act as a protective barrier or a mechanical support. Millions of tonnes of keratin wastes and low value co-products are generated every year in the poultry, meat processing, leather and wool industries. Keratinases are proteases able to breakdown keratin providing a unique opportunity of hydrolysing keratin materials like mammalian hair, wool and feathers under mild conditions. These mild conditions ameliorate the problem of unwanted amino acid modification that usually occurs with thermochemical alternatives. Keratinase hydrolysis addresses the waste problem by producing valuable peptide mixes. Identifying keratinases is an inherent problem associated with the search for new enzymes due to the challenge of predicting protease substrate specificity. Here, we present a comprehensive review of twenty sequenced peptidases with keratinolytic activity from the serine protease and metalloprotease families. The review compares their biochemical activities and highlights the difficulties associated with the interpretation of these data. Potential applications of keratinases and keratin hydrolysates generated with these enzymes are also discussed. The review concludes with a critical discussion of the need for standardized assays and increased number of sequenced keratinases, which would allow a meaningful comparison of the biochemical traits, phylogeny and keratinase sequences. This deeper understanding would facilitate the search of the vast peptidase family sequence space for novel keratinases with industrial potential.

**Keywords:** keratinase; serine protease; metalloprotease; peptidase; keratin hydrolysis; keratin waste; valorisation; bioactive peptides

#### **1. Introduction**

Millions of tonnes of waste keratin are produced every year in the poultry, meat processing, leather and wool textile industries. The global poultry meat processing industry alone produces 40 <sup>×</sup> 106 tonnes of waste feathers annually [1]. With the transition away from the fossil fuel-centric economy to a sustainable circular economy, the valorisation of keratin materials addresses the waste problem and facilitates the integration of waste keratin into new value chains to enable a circular economy.

Traditionally, keratin waste has been sent to landfill or rendering, or used as fertilizer, feather meal or incinerated [2,3]. There is, however, an opportunity for livestock industries to produce higher value products from waste keratin. There are multiple thermochemical methods available to prepare hydrolysed keratin for various value-adding opportunities [4]. However, the use of peptidases with keratinolytic activity for keratin hydrolysis protects the integrity of the keratin amino acids in most cases and allows control over the peptide size in the hydrolysate that is not readily achievable with other methods [5]. This degree of control allows the production of bespoke medical biomaterials, smart biocomposites, protein feed supplements with enhanced nutritional and bioactive properties as well as personal care products with enhanced functional and bioactive properties.

Identifying peptidases with keratinolytic activity is an inherent problem associated with the search for new enzymes. Keratinase activity however appears to be dependent on the accessibility of the keratin substrate to the enzyme [6,7]. Thermochemical or biochemical treatment of the keratin, with emphasis on the reduction of the disulphide bond and disruption of other important bonds involved in the structural stability of keratin like isopeptide, hydrogen and glycolytic bonds [6,8–10], appears to be the prerequisite for enzymatic hydrolysis. Sulphitolysis, which involves reduction of the disulphide bond in keratin, often acts synergistically with keratinases in nature [6,7]. Although destabilization of the keratin structure is a prerequisite for keratin hydrolysis, not all peptidases can hydrolyse keratin. Peptidases like trypsin, papain and pepsin cannot hydrolyse keratin as efficiently as peptidases with keratinolytic activity, even if the reduction of disulphide bond has already occurred [11]. The elucidation of the unique characteristics of peptidases with keratinolytic activity that differentiate them from the other peptidases, would be an important breakthrough in the search for new and robust keratinases for the valorisation of keratin waste.

This paper reviews twenty sequenced peptidases with keratinolytic activity from the serine protease and metalloprotease families by comparing their biochemical characteristics and will highlight the difficulties associated with the interpretation of these data.

#### **2. Keratin: A Complex and Strong Structure**

Keratins are important structural proteins produced by vertebrate epithelia that have various physiological function. Keratins can act as a protective barrier to water, against infection or cushion tissue from mechanical impact. The two main types of keratins proteins are α-keratin and β-keratins. These two types are further divided into acidic or basic, soft or hard, and have different molecular weights [4,9,12,13]. The following section describes the complexity of the keratin structure, which provides insight into the resistance of keratin to hydrolysis. This review will concentrate on hard α-keratin and β-keratin, γ-keratins and the keratin-associated proteins, which are common to mammalian hair, bristles, wool, hooves, horns and feathers.

α-Keratin has an α-helix structure, which is stabilized by hydrogen bonding and the presence of multiple cysteines forming disulphide bridges. α-Keratin is characterized by a lower sulphur content compared to other keratins and a molecular mass of 60–80 kDa [4]. Hard α-keratin is the major protein of mammalian fibres, nails, hooves and horns. In contrast, hard β-keratins are characteristic of the hard, cornified epidermis of reptiles and birds, e.g., feathers, claws and scales, and have a twisted β-sheet-like structure. They also form the major component of the fibre cuticle. The β-keratin pleated sheets consist of β-strands, which are laterally packed and can have a parallel or antiparallel orientation. The β-sheets are held together by hydrogen bonds and the planar nature of the peptide bond, which results in the stable pleated β-sheet [13]. β-Keratins have a molecular mass of 10–22 kDa. A third type of keratin, γ-keratin, is a globular protein with a high sulphur content and a molecular weight of about 15 kDa. This keratin, along with keratin-associated proteins, form the matrix between the microfibrils and microfibrils of the fibre cortex of mammalian fibres and stabilize the structure of the cortex via extensive disulphide bridge formation.

The complex structural organization of all mammalian fibres is very similar [8]. The hair fibre consists of an outermost cuticle layer, which is composed of overlapping flattened scale-like cells that form a protective sheath around the cortex [8]. The major protein of the fibre cuticle is β-keratin [4]. The cortex is composed of hard α-keratin intermediate filaments embedded in a sulphur-rich matrix. These filaments surround the medulla when present, as is the case for coarser fibres. The cell membrane complex binds the cuticle and cortical cells.

The cuticle layer is laminated and consists of the following layers—the cuticle filament-associated surface membrane, the cystine-rich exocuticular *a*-layer, the lower exocuticle and the endocuticle, which contains only low levels of sulphur-containing amino acids and constitutes the inner lining of the cuticle [8]. The outermost layer of the cuticle provides a hydrophobic barrier, which protects the fibre surface from water and chemical compounds. This cuticle filament-associated surface membrane is 2–7 nm thick and composed of highly cross-linked proteins and lipids. The major fatty acid of the cuticle surface lipids found in human and animal hair is 18-methyleicosanoic acid [14]. It is covalently linked to the protein matrix below by a thioester linkage and the protein matrix is cross-linked by isopeptide bonds [15]. An isopeptide bond results from the transglutaminase-catalysed formation of an amide bond between the amino acid side chains of the amino acid residues in the keratin protein, for example, lysine and glutamine [9].

The cortical cells are assembled as keratin intermediate filaments and have a diameter of 7–8 nm in all mammalian fibres [8]. These intermediate filaments form ordered aggregates or microfibrils and macrofibrils depending on species and function (Figure 1). The hard α-keratin intermediate filaments are assembled from tetramers, a pair of laterally aligned and antiparallel dimeric molecules. On average, keratin intermediate filaments contain eight tetramers. In the case of wool, the cortex region is composed of an orthocortex and paracortex with different intermediate filament/matrix packing. The proportion of ortho- and paracortex in the wool fibre determines the degree of crimping [13].

**Figure 1.** Structure of keratin. Adapted from work in [12] under the Creative Commons Attribution 4.0 International license (https://creativecommons.org/licenses/by/4.0/deed.en).

Keratin peptide heterodimers are formed when a type I (acidic) polypeptide chain and a type II (basic) polypeptide chain align in parallel. Each polypeptide chain is composed of a central α-helical region (about 46 nm in length) with non-helical head and tail domains [13]. The head and tail domains are rich in cysteine, glycine and tyrosine amino acids. Disulphide and isopeptide bonds are formed with other keratin intermediate filaments, cysteine-rich matrix proteins and keratin associated proteins, which stabilize the fibre [8–10]. The disulphide bonds along with the N-acetyl glucosamine-glycosylated serine and threonine in the head and tail domains also stabilize the heterodimers [6].

#### **3. Thermochemical Methods of Keratin Degradation**

There are multiple thermochemical methods available to prepare hydrolysed keratin for various value-adding opportunities, with specific processes chosen depending on the end-use [4]. Thermochemical methods include solubilization of keratin in organic solvents, ionic liquids or by hydrothermal methods; oxidation or reduction of the disulphide bridges; disruption of the hydrogen bonds with compounds like urea; and acid or base hydrolysis.

The composition of the final hydrolysate will depend on the method used to hydrolyse the keratin. Some of the thermochemical processes result in a hydrolysate containing a highly diverse mix of keratin-derived peptides and free amino acids and others are more specific. However, in most cases, the amino acid composition is modified. The processes and hydrolysate products will be described in more detail in the following section.

After solubilization of keratin with solvents like with *N,N*-dimethylformamide or dimethyl sulfoxide, precipitation is required with acetone and drying to produce a powder of keratin [16]. The major drawback of this method is the use of large quantities of solvents, which need to be recycled or incinerated. Solubilization can also be achieved with ionic liquids. Xie et al. used the ionic liquid, 1-butyl-3-methylimidazolium chloride for the solubilization of wool keratin, which disrupted

the hydrogen bonds in the keratin macromolecules [17]. The keratin peptides were precipitated from the resulting hydrolysate with methanol. Ionic liquids are more expensive than traditional solvents and extraction of the keratins from the ionic liquid can be difficult.

Hydrothermal treatment is usually carried out at temperatures of 80–140 ◦C and steam pressures of 10–15 psi. Acid or base can be added to speed up the process of solubilization [18]. Under conditions of high temperature and pressure, the thermally unstable amino acids, glutamine and asparagine are degraded [19]. If base is added to this process then lysine, methionine and tryptophan are also destroyed [20,21]. Modified amino acids, lysinoalanine and lanthionine are also formed from lysine and cystine, respectively. Heating of proteins leads to a degree of racemization of the free and bound L-amino acids [22–24].

Reduction with reducing agents like thioglycolate [4], dithiothreitol [25], 2-mercaptoethanol [26], sodium sulphite [27], bisulphites [28] or cysteine [29] combined with high concentrations of compounds like urea, thiourea or surfactants, which disrupt the hydrogen bonds stabilising the keratin structure, results in the production of kerateine [30]. Kerateine contains cysteine thiol and cysteine sulfonate in place of the disulphide bonds. Kerateine is less soluble in water and can be re-cross-linked if exposed to an oxidant [4].

The microstructure of wool keratin after treatment for 4 h at 65 ◦C with 2-mercaptoethanol, EDTA, high concentrations of urea and pH 9 was investigated by Cardamone [31]. Analysis of the hydrolysate revealed a defined mixture of microfibrillar and intermediate filaments. This mixture of subunits was suitable for producing self-assembling biomaterials.

Oxidation of keratin by oxidants like peracetic acid [32] or peroxycarboximidic acid [33] leads to the formation of keratose. Keratose contains sulfonic acid groups and cysteic acid instead of the disulphide bonds [4]. These keratoses are hydroscopic, water soluble and the disulphide bridges cannot spontaneously re-form under oxidative conditions. Keratoses are not as stable as kerateines.

Oxidative sulphitolysis has been patented and commercialized to produce three functional keratin protein and peptide products. These products are based on S-sulphonated keratin intermediate filaments, S-sulphonated keratin high-sulphur proteins and keratin peptides [34]. The process aims at maintaining the structural integrity of the keratin proteins. The cystine groups in the wool keratin are converted to S-sulfocysteine using sodium sulphite or sodium metabisulfite and then oxidized with cupraammonium hydroxide. The intermediate filaments and peptides can undergo crosslinking by reductive desulfonation of the cysteines in the filaments and peptides and subsequent reformation of the intermolecular disulphide bonds.

One of the disadvantages of alkaline hydrolysis of keratins is the modification or degradation of amino acids (Table 1). Alkaline hydrolysis of keratins at higher temperatures results in the degradation of the thermally unstable amino acids, asparagine, glutamine, arginine, serine, threonine and cysteine [5]. Lysinoalanine and 8-aminoalanine are formed under alkaline conditions [35,36]. Another modification that occurs is the racemization of free or bound L-amino acids to the D-enantiomers [23,37,38]. Free amino acids racemize ten times slower than bound amino acids [24]. Following, for example, prolonged treatment of wool keratin at 70 ◦C and pH 9–11, lanthionyl residues [31] and dehydroalanine [39] are formed from cystine. Cystine and hydroxy amino acids were destroyed if the alkaline treatment was performed in the presence of reducing agents [40].



Note: \* Dehydroalanine is probably formed from the cleavage of the C-S bond in cystine.

Acid hydrolysis of keratins leads to the loss of some amino acids like serine, threonine, tyrosine and cystine and the conversion of asparagine, glutamine, methionine and tryptophan into other compounds ([5,19,46] Table 2). Polypeptides, resulting from the acid hydrolysis of keratin, have a more amorphous structure than alkaline hydrolysates, because most of the hydrogen bonds are broken during this process [47]. A typical acid hydrolysis of keratin uses hydrochloric acid [48,49] or sulphuric acid [50] at high temperatures.

Zhang et al. showed that acid hydrolysis was not as effective as other treatments mentioned above [49]. Wool keratin was hydrolysed with 4M hydrochloric acid at 95 ◦C for 24 h, resulting in 33% solubilization of the wool keratin. Increasing the treatment time had no effect on the yield, suggesting that there is a recalcitrant portion of the keratin resistant to acid hydrolysis.

Thermochemical methods offer cheap and versatile processes for hydrolysing keratin for a variety of applications. However, the use of harsh chemicals and conditions, the lack of ability to control the processes in most cases and the often unfavourable modification of the amino acids or peptides present environmental problems and peptide mixes that would be unsuitable for some applications. Using enzymes working under mild conditions to catalyse the hydrolysis offers a favourable alternative.


**Table 2.** Amino acid modification during acid treatment.

#### **4. Microbial Degradation of Keratin**

The first peptidases with keratinolytic activity were found in *Bacillus* sp. and *Streptomyces* sp. and belong to the serine peptidase family [51]. The ability to degrade keratin is widespread and has been identified in bacteria and fungi [4,52]. Keratin-degrading microorganisms have been isolated from many sources like skin, feathers, hair, nails, soil, geothermal hot stream and wastewater, which is reflected in the optimum pH and temperature of the keratinase activity of these microorganisms. The pH optimums of keratinases range from pH 5.5 for the fungal keratinase from *Trichophyton mentagrophytes* [53] to pH 12.5 for the keratinase from *Brevibacillus* sp. AS-S10-11 [54]. Although, temperature optimums vary from 30 ◦C for the keratinase from *Brevibacterium luteolum* [55] to 100 ◦C for the keratinase from *Fervidobacterium islandicum* AW-1 [56].

Publications from 2018 and 2019 report the isolation of diverse species of bacteria like *Streptomyces* sp. [57], *Aeromonas hydrophila* FB3 [58], *Pseudomonas putida* KT2440 [59] and *Serratia marcescens* EGD-HP20 [60,61] with keratinolytic activity. However, the number of *Bacillus* strains with keratinolytic activity prevailed over any other genus of bacteria [62–86]. Valorisation of waste feathers [5,65,66,87] and the replacement of the traditional, highly polluting hide dehairing step used in the leather industry with a more environmentally friendly enzymatic step using keratinases [2,55,79,88] were the dominant themes of these papers.

Despite the interest in the enzymatic hydrolysis of keratin, mechanisms of keratin degradation in microorganisms are not fully understood. There is evidence that microbial degradation of keratin proceeds via a consortium of enzymes (Figure 2 [6,7,89]).

Disruption of the keratin structure is an essential step in the breakdown of keratin by keratinases. Various mechanisms have been suggested for fungal systems. Disulphide bond reductases and the intracellular cysteine dioxygenase can break the structure-stabilizing disulphide bridges in keratin [6,7,90]. Cysteine dioxygenase in conjunction with aspartate aminotransferase produces the reducing agent, sulphite, from cysteine, which is secreted into the surroundings and contributes to the chemical reduction of the disulphide bond. The reduction of the disulphide bonds aids access of the endoproteases (serine protease family), exoproteases (metalloprotease family) and

oligopeptidase (metalloprotease family) to the keratin fibres or feathers. It has also been found that the membrane-bound redox system of the cell can cleave the disulphide bonds in keratin. The mechanical pressure exerted by fungal mycelia penetrating the keratin structure can also contribute to the disruption of this structure, facilitating access of the keratinase to the substrate. In nature, these mechanisms act synergistically with keratinases and speed up the degradation of keratin. Auxiliary proteins, like lytic polysaccharide monooxygenases (LPMOs), have been found associated with keratin degradation [6]. It is thought that they contribute to α- and β-keratin degradation. Until now LPMOs were thought to be associated with cellulose, chitin, hemicellulose and starch degradation only. It is possible that these enzymes hydrolyse the glycolytic bond between N-acetylglucosamine and serine and threonine in the head and tail region of the intermediate filaments, which contributes to the destabilization of the keratin structure.

**Figure 2.** Possible mechanisms for microbial degradation of keratin (LPMO = lytic polysaccharide monooxygenase).

However, examples of peptidases with keratinolytic activity that do not need the assistance of disulphide reducing enzymes or agents have also been reported. Pillai et al. isolated a serine protease from *Bacillus subtilis* P13 with reductase and keratinase activities [91]. The isolated enzyme was able to decompose feathers and dehair hides.

He et al. analysed the enzyme consortium involved in the hydrolysis of feathers by a specific strain of *Bacillus subtilis* and identified four of the enzymes involved in keratin hydrolysis [74]: a serine protease with keratinase and disulphide bond-reducing activity; a peptidase T; a γ-glutamyltransferase, which generates a free cysteinyl group from glutathionine; and a cystathionine γ-synthase, which catalyses the production of L-cystathionine from homoserine ester and cysteine. The L-cystathionine is further converted to methionine and ammonia is released.

#### **5. Characterisation and Comparison of Keratinases from S1, S8 and M4 Peptidase Families**

Many articles characterising organisms capable of degrading keratin and their possible industrial applications have been published. Yet, there are few articles that report enzyme sequences and investigate the molecular and biochemical characteristics of the enzymes produced by these organisms [92,93]. The first paper that explored the molecular aspects of a keratinase produced by *Bacillus licheniformis* was published by Lin et al., 1995. Since then, more than 40 keratinases have been sequenced. To date, peptidases with keratinolytic activity from six different peptidase families have been identified: S1, S8, M4, M5, M14 and M28. Most of the characterized keratinases are produced by *Bacilli* and are members of the S8 serine peptidase family. There are currently over 127,000 peptidase sequences from the S1 (70919), S8 (38270), M4 (6403), M5 (145), M14 (11202) and M28 (904) families deposited on the MEROPS peptidases database. These 127,000 peptidase sequences represent an enormous unmined potential for the discovery of new peptidases with keratinolytic activity if the requisite properties of a peptidase with keratinolytic activity can be identified.

The S1 family sequences, when pairwise aligned, show a minimum value of 27.27% and a maximum of 97.22% identity, with an average of 61.48% for the four available sequences. The S8 family has a minimum of 13.69% and a maximum of 99.72% identity, with an average of 63.29% for the 13 available sequences, and the M4 family has 25.56% identity between the two available sequences. Although many of the characterized enzymes have been produced by the native unmodified organism [94–98], several

examples involve heterologous expression. Different organisms have been used for recombinant production, including yeast such as *Komagataella Pastoris* (*Pichia Pastoris*) [99] and bacteria such as *Escherichia coli* [100–110] and *Streptomyces lividans* [111]. Including the pre-pro-domains with the catalytic domain in heterologous systems have been shown to maintain enzyme activity and secretion [99,102,107,110] and inclusion of C-domains, when present, is important for substrate binding and recognition [105]. Replacing the native signal peptide for the *E. coli* signal peptide when expressing in *E. coli* has also led to higher levels of expression [101].

In this section the biochemical data of twenty sequenced peptidases with keratinolytic activity from the S1 and S8 peptidase families (serine proteases) and the M4 peptidase family (metalloprotease) are compared (Table 3). Difficulties associated with the interpretation of these data are also highlighted. The selection is based on the availability of sequence and biochemical data. The M5, M14 and M28 peptidase families were excluded because each family had only one biochemically characterized example with full sequence data available.

**Table 3.** Keratinolytic microorganisms and their keratinases from the S1, S8 and M4 keratinases selected for this study.


Note: <sup>1</sup> NCBI GenBank nucleotide accession number.

#### *5.1. S1, S8 and M4 Peptidase Families*

The S1 family is the largest family of serine proteases. The active site of S1 peptidases contains the catalytic triad, His, Asp and Ser. All enzymes characterized in this family are endopeptidases. The four peptidases in Table 3 belong to the S1A family represented by chymotrypsin as the type-example. The hydrophobic amino acid at the P1 site determines the specificity of these peptidases [113,114].

The S8 family is currently the second largest serine protease family and the most widely characterized to date [114,115]. Most of the keratinases are found in the subfamily S8A including the 14 keratinases in Table 3. They are represented by subtilisin as the type-example. Their active site contains the catalytic triad of Asp, His and Ser. In general, these enzymes are endopeptidases [116], active between neutral and moderately alkaline pH and many are thermostable [117]. Most enzymes in this family are not specific, usually cleaving after a hydrophobic residue in the peptide substrate [114,117]. S1 and S8 families are examples of convergent evolution as they catalyse the same reaction but have no sequence homology. Two calcium-binding sites contribute to thermal stability in many members of these families [114,117].

Two keratinases in Table 3 belong to the M4 family. They are characterized by a catalytic zinc ion tetrahedrally coordinated in the active site by a histidine and glutamate present in a HEXXH motif, another glutamate residue and water [118]. Most members of this family are endopeptidases and active at neutral pH. The preferred cleavage site occurs at a hydrophobic residue followed by leucine, phenylalanine, isoleucine or valine. These peptidases are stabilized by Ca2<sup>+</sup> [119].

Independent of their families, keratinases usually cleave aromatic and hydrophobic amino acid residues at the P1 position. Keratins are composed of 50 to 60% aromatic and hydrophobic residues, which could partially explain the keratinase specificity for keratin [120–122]. Most of these peptidases are stabilized by divalent cations like Ca2<sup>+</sup> and are extracellular [119,123,124].

#### *5.2. Problems Associated with Keratinase Assays*

There are several issues with the current methods used to characterize keratinases. The assays are not standardized in the literature in terms of reaction conditions and substrates. The most common method used to measure keratinase activity is a colorimetric assay that uses the commercially available derivative of wool, keratin azure [125] or azokeratin (sulfanilic acid-azokeratin [126]). However, batch variability and the fact that the chromogenic agents are only bound to the outer portion of the substrate compromises reproducibility. Quantification of the soluble peptides generated by hydrolysis of keratin has also been used to determine the effectiveness of keratinases on keratin substrates. Common quantification methods used are Bradford [95,127], Lowry [128,129] or measurement of absorption at 280 nm [106,111] (see Table 4). Each of these methods have several limitations. The Coomassie Blue dye used in the Bradford method preferentially reacts with arginine and lysine in the peptide mix and alkaline pH and detergents interfere with the reaction. The Folin–Ciocalteu dye used in the Lowry method oxidizes the aromatic amino acids residues in the protein and is affected by reducing agents. Only tyrosine, tryptophan and cysteine absorb at 280 nm and other compounds like DNA in the solution can interfere with the measurement [130]. The simplest and probably most accurate method for quantifying keratinase activity is the measurement of weight loss when the insoluble keratin substrates like mammalian hair fibres, feathers or wool are solubilised through hydrolysis [96,127].

Table 4 describes a variety of substrates that have been used to assay keratinase activity in the literature. The substrates that were used include keratin azure (wool), keratin powder, soluble keratin, keratin (undefined), feathers and feather meal powder. It was not possible to ascertain the source and integrity of most of these keratin substrates from the papers. The pretreatment of these substrates is also an important aspect in determining the keratinase activity. Keratin powder and solubilized keratin were generally obtained from commercial sources; however, the sources and preparation were not described. Pretreatments like autoclaving and milling [103,107], or treating with solvents at high temperatures [106], are known methods for keratin powder preparation from the literature. In the case of the rK27 keratinase, the feather powder used in the assay was autoclaved and dried at 60 ◦C [103]. These preparation methods, as already described in Section 3, would compromise the keratin structure. The keratinases, KerRP [96], Ker-A1 [97] and SAPB [102] were assayed on keratins of unknown source. In the WF146 protease assay, the feather substrate was washed with ethanol prior to use in the assay, which would likely remove the protective lipid layer [108].

Co-treatment can also affect the integrity of the keratin substrates during enzymatic hydrolysis [125,131,132]. Except for KerQ7 [104], all assays in Table 4 were carried under alkaline conditions between pH 8 and pH 12.5 and temperatures ranging from 50 to 80 ◦C. These conditions would most likely contribute to the weakening the keratin structure. Keratinase assays with SAPDZ [100], KerQ7 [104], KERDZ [94], and KERAK-29 [95] were supplemented with the divalent cations Ca2<sup>+</sup> or Mn2+. Divalent cations are known to stabilize serine proteases [114,117]. Other keratinase studies added reducing agents, like β-mercaptoethanol (protease C2 [106], WF146 protease [108]) or dithiothreitol (SFP2 [99]) to the assay mixture. Reducing agents are known to break the disulphide bond leading to disruption of the keratin.


**Table 4.** Keratinase pH and temperature optimums of the selected S1, S8 and M4 keratinases with associated assay conditions. See text for further details of the assays.

Note: Source organism, accession numbers and references can be found in Table 3; \* in some cases pH and temperature optimums were determined on both casein and keratin substrates. The casein optimums are in brackets; Temp = temperature; PT = pretreatment; CT = co-treatment; β-ME = β-mercaptoethanol; DTT = dithiothreitol. Quantification methods, where available, are in brackets after the assay condition description.

The challenges with the keratinase assays discussed above highlight the need for standardized assays and substrates used to test keratinases and the challenges faced in attempting to compare and analyse data from the literature when the assays are not comparable.

#### *5.3. The E*ff*ect of Additives on Selected S1, S8 and M4 Keratinases*

Various additives were tested on the selected S1, S8 and M4 keratinases-cationic, anionic and neutral detergents, oxidizing agents, reducing agents, mono- and divalent metals. Table 5 contains a summary of additives that had a positive effect on keratinase activity. A positive effect was defined as ≥ 110% activity compared to the control without additive. Some of the papers used keratin as the assay substrate, some used casein and in some cases, both were tested.


**Table 5.** Influence of additives on the activity of selected S1, S8 and M4 keratinases. Numbers in brackets correspond to the % activity compared to 100% in the absence of the additive.

Note: Source organism, accession numbers and references can be found in Table 3; Bold = tested on a keratinous substrate; Not bold = tested on casein; \* = Only tested for binding; \*\* included for comparison with KerSMD; underlined = denaturing detergents; DTT = dithiothreitol; β-ME = β-mercaptoethanol; LAS = linear alkylbenzene sulfonate; SDS = sodium dodecyl sulfate; TAED = tetraacetylethylenediamine; TTAB = tetradecyltrimethylammonium bromide; CHAPS = 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate; CTAB = cetrimonium bromide.

Despite there being large differences in concentrations of metals additives, incubation time and temperature, in general, supplementation with Ca2<sup>+</sup> showed the largest increase in activity except for KerSMF [101] and kerA1 [97]. In the case of KerSMF, Ca2<sup>+</sup> had no effect on activity and in the case of kerA1, Mg2<sup>+</sup> addition increased activity by 199% compared to 123% for Ca2+. In general, Ca2<sup>+</sup> > Mg2<sup>+</sup> > Mn2<sup>+</sup> had a positive effect on all the S1 and S8 keratinases (Table 5). The effect of these divalent metals on M4 metalloproteases is discrete compared to serine proteases. Only the addition of magnesium and manganese ions resulted in keratinase activity slightly above the control without additives. These divalent ions have been described to stabilize the active structure of the peptidases by binding to the metal-binding sites [100]. Other explanations for the higher activity are possible stabilization of enzyme/substrate complex [101] or formation of salt or ion bridges that maintain the enzyme conformation [97,122,128]. Furthermore, these metal ions reduce thermal denaturation [134]. Lin et al. observed that aqualysin, a thermostable peptidase from the S8 family, was only stable at high temperatures in the presence of 1 mM Ca2<sup>+</sup> [135].

Several studies were carried out on the keratinase activity in the presence of metal ions (Zn2<sup>+</sup>, Cu2<sup>+</sup>, Co2<sup>+</sup>, Ba2<sup>+</sup>, Sn2<sup>+</sup> and Ni2<sup>+</sup>) were carried out (Table 5). The addition of the metal ions improved activity between 10% and 29% except for SAPDZ [100], where Zn2<sup>+</sup> addition increased activity by 80% and Cu2<sup>+</sup>addition increased activity of SFP2 [99] by 49%. Li et al. characterized SFP1, a non-keratinolytic peptidase similar to SFP2 and produced by the same organism [99]. It showed increased activity with copper ions, possibly due to the stabilization of the enzyme. Copper ions acting as a stabilizer has rarely been described in previous serine protease studies, and it is not known whether there is a copper-binding site stabilizing the enzyme [136]. In another example, peptidases were more stable in the presence of copper ions, which resulted in a reduction in both autolysis and thermal inactivation rates [137].

Detergents, in general, enable the disruption or formation of hydrophobic and hydrophilic bonds and assist in the extraction of proteins into aqueous media [138]. Detergents can act as denaturing agents on enzymes. Denaturing detergents are anionic (SDS, LAS) or cationic (CTAB, TTAB). They denature proteins by breaking protein–protein interactions. Non-denaturing detergents are non-ionic (Triton X-100, Tweens, cholate, saponin) or zwitterionic (CHAPS, sulfobetaine, zwittergent), and their action is milder and enzyme function is usually maintained. In most cases the addition of denaturing and non-denaturing detergents resulted in an increase in activity (110–150%). However, the addition of the non-ionic detergents to the assay mixture with keratin as substrate of rK27 had a dramatic effect on activity compared to the control without detergent [103]. Activities of 677% (Triton X-100), 242% (Tween 80), 461% (saponin) and 276% (cholate) were achieved. The addition of the anionic denaturing detergent, SDS to the assay increased the keratinase activity to 186%. The addition of the non-ionic detergents, Tween 40, Tween 60, Tween 80 and Triton X-305 to the assay mixture with keratin as substrate for the M4 keratinase, RecGEOker [109], showed increased activity to 180%, 133%, 122% and 153%, respectively. This example showed a definite trend of increasing activity with decreasing Tween 80 (monounsaturated C18 derivative) < Tween 60 (saturated C18 derivative) < Tween 40 (saturated C16 derivative). The partial solubilizing action of detergents on the insoluble keratin substrate might explain why both denaturing and non-denaturing detergents have a positive effect on keratin hydrolysis. There are insufficient examples to confirm this Tween effect on keratinases in general.

The reduction of disulphide bonds, destabilizes keratins and acts synergistically with keratin hydrolysis in nature [6,8–10]. Sodium sulphite, dithiothreitol (DTT) and β-mercaptoethanol were tested on some of the keratinases in Table 5. The reducing agents had a positive effect on all S8 keratinases tested with keratin as the substrate. The increase in activity ranged from 115% for Na2SO3 (KerSMF [101]) to 623% for β-mercaptoethanol (YT06 protease [98]) except in the case of KBALT [63], where β-mercaptoethanol had no effect on the activity. β-Mercaptoethanol doubled the activity of SAPB [102] when tested with casein as substrate. DTT also increased the activity of the M4 keratinase, RecGEOker (139% [109]), when tested with keratin as substrate. None of the S1 enzymes were tested

with keratin and reducing agents. However, the two assays with casein and reducing agent showed on one hand, no effect from β-mercaptoethanol on KERAK-29, [95] and on the other hand, a considerable effect on SFP2 (DTT, 278%; β-mercaptoethanol, 235% [99]). It should be noted that where disulphide bonds present in the enzyme are essential for function the inclusion of reducing agents may negatively affect activity.

Chaotropic agents are comparable to detergents, breaking non-covalent interactions and allowing protein denaturation [139–141]. Urea and isopropanol are chaotropic agents (Table 5). The activity of SAPB [102] was increased to 165% in the presence of urea compared to the control and the activity of NAPase [111] was increased to 130% in the presence of isopropanol [111].

The effect of the oxidizing agents, H2O2 and sodium hypochlorite, was also studied on three S8 and S1 keratinases, SAPB [102], rK27 [103] and KERAK-29 [95]. Activity was significantly increased in all cases (Table 5).

In most cases the effect of additives like divalent cations, detergents, reducing agents, chaotropic agents and oxidizing agents have a positive effect on keratinase activity. Nearly all compounds capable of disrupting the integrity of the keratin structure without inactivating the keratinase appear to have a positive effect on keratinase activity. The effect of compounds disrupting the keratin structure was, in some cases like rK27 [103], significant.

#### *5.4. Substrate Specificity*

Table 6 summarizes the substrate specificity data of the selected keratinases from the literature. In general, a variety of keratins and other proteins like gelatin, casein and albumin were tested. To compare the selected keratinases, the values in Table 6 have been normalized using the activity of designated keratin substrates (keratin azure, keratin, feather or wool) as 100% activity.

KerQ7 [104] was the only keratinase in Table 6 tested on multiple types of keratins. KerQ7 showed a preference for the β-keratin-rich feather meal and feathers. The activity on feather meal was only 16% higher than feathers. The activity on rabbit hair, goat hair and bovine hair was 88%, 74% and 50% of the activity on feathers, respectively, whereas activity on wool was only 12%. These substrates are rich in α-keratins [4,9,12,13]. Substrate fibre thickness and fibre surface area may also contribute to the variations in enzyme activity. Nonetheless, the keratinase activity toward various substrates is likely to be multifactorial. KerSMD and KerSMF, from *Stenotrophomonas maltophilia*, showed less activity towards feather powder and wool than keratin azure [101]. KerSMD and KerSMF had similar activity on feather powder (54% and 71%, respectively) and wool (59% and 78%, respectively). However, KerSMD showed an activity of 1589% towards soluble keratin compared to an activity of 126% for KerSMF on the same substrate.

No trends were observable on non-keratin substrates. For example, SAPDZ [100] showed 81% activity on gelatin compared to keratin, whereas the activity of kerA1 [97] and SAPB [102] on gelatin was 22% and 146% compared to keratin, respectively. The same inconsistencies can be seen with casein. The activities of SAPB [102] and KerSMD [101] on casein are 153% and 2800%, respectively, compared to keratin azure, whereas KerSMF [101] has only slightly lower activity on casein (91%) compared to keratin azure.

Keratinases are known for their activity on "hard-to-degrade" proteinogenic substrates. Most of the characterized keratinases in the literature are also capable of degrading collagen, which is an example of another complex and hard-to-degrade substrate [142]. A study in 2008 characterized the first keratinase without collagenase activity [143]. Only three enzymes from the S8 family were tested on collagen or azocoll (azocollagen). Vpr [105] presented collagenase activity (129%) and while SAPDZ [100] did not. Protease C2 [106] showed a surprisingly high activity (24000%) on azocoll compared to keratin azure (100%). KERDZ [94], from the S1 peptidase family, had no activity on collagen, whereas RecGEOker [109], belonging to the M4 metallopeptidase family, was able to hydrolyse collagen. The differences in activity between substrates may be attributed to the specific peptide sequences in the substrates and the sequence specificity of the enzymes.

**Table 6.** Substrate specificity of the selected S1, S8 and M4 keratinases. Numbers in brackets correspond to the % activity relative to other substrates.


Note: Source organism, accession numbers and references can be found in Table 3; Activity normalized to the following substrates—<sup>1</sup> keratin azure, <sup>2</sup> keratin; <sup>3</sup> feather, <sup>4</sup> wool; BAEE = N-α-benzoyl-L-arginine ethyl ester; TAME = N-α-p-tosyl-L-arginine methyl; BCEE = benzoyl-citrulline ethyl ester; BTEE = N-benzoyl-L-tyrosine ethyl ester; ATEE = N-acetyl-L-tyrosine ethyl ester; DTNB = 5,5 -dithiobis-(2-nitrobenzoic acid); Azocoll = commercially available azocollagen.

Some enzymes also showed esterase activity, which may be of importance for facilitating enzyme access to the substrate. Fatty acids of the cuticle surface are linked via a thioester to the protein matrix below in keratin fibres and feathers [14]. Only three enzymes in Table 6 have been characterized on ester substrates. The two S8 family peptidases—SAPDZ [100] and KerRP [96]—appear to have similar ester substrate affinity with both showing activity against N-benzoyl-L-tyrosine ethyl ester (BTEE) and N-acetyl-L-tyrosine ethyl ester (ATEE). In contrast, KERDZ (S1 family) had no activity towards these substrates but was active towards N-α-benzoyl-L-arginine ethyl ester (BAEE), N-α-p-tosyl-L-arginine methyl (TAME) and benzoyl-citrulline ethyl ester (BCEE) [94].

In general, the peptidases from S1, S8 and M4 families (Table 6) present varied substrate specificities. There are limited examples in the S1 and M4 families to detect trends but even within the S8 peptidases examples there were no obvious substrate preferences.

#### **6. Potential Applications of Keratinases**

New keratinases with improved properties for commercialization and the keratin hydrolysates they produce represent an opportunity for adding value to keratin waste.

Commercial keratinases are sold for a variety of applications (Table 7) such as the degradation of infectious prions, as supplements for animal feed to improve its nutritional value, removal of corns and calluses from skin, treatment of acne and nail fungi and, they are also incorporated into cosmetic skin peeling and depilatory creams [6,52,144]. Other applications include the use in cleaning products for unblocking drainpipes and septic tanks.



Source: <sup>1</sup> [6], <sup>2</sup> [127], <sup>3</sup> [52].

There are also a number of promising applications of keratinases that have not been commercialized to date: dag or manure balls removal from cattle hides and tails [145]; extraction of glucocorticoids from chicken feathers to monitor the stress level in poultry breeding and production programmes [146] extraction of chicken feather cholesterol as a precursor to bile salts that can be used to produce bio-emulsifiers and biosurfactants in the cosmetic industry [18]; selective hydrolysis of wool from wool/polyester or mixed textiles to facilitate textile recycling [147]; and dehairing of hides in the leather industry [64,65,82].

The use of keratinases for the processing of keratinous waste might be advantageous for high value products. The use of enzymes instead of thermochemical methods for keratin hydrolysis reduces chemical modification arising from harsh chemical hydrolysis and might allow a degree of control of the peptide composition that is produced. Keratin hydrolysates are widely used in protein feed supplements [18]. Feather waste, for example, is hydrolysed with saturated steam under high pressure (sometimes with the addition of lime) to produce feather meal, which is used as a feed supplement [148]. These conditions lead to the loss or modification of some of the amino acids, which impacts the nutritional value and digestibility of the feather meal. Hydrolysis with keratinases might offer an alternative, which reduces the energy requirements of the process and enhances the nutritional value of the supplement.

Keratin-derived bioactive peptides have been reported in the literature. These peptides have a range of activities like antimicrobial [149], antihypertensive [150], anti-inflammatory [151–154], antioxidant [149,150,155], inhibition of early stage amyloid aggregation [156], antidiabetic [157] or anti-aging [158–160] depending on the keratin source and the method of preparation. Producing protein feed supplements with antioxidant or anti-inflammatory properties as well as skin and hair

products with antioxidant, anti-inflammatory, antimicrobial or anti-aging properties would most likely increase the value of these products.

Keratin peptides and subunits can spontaneously self-assemble [161]. This property can be exploited to form biomaterials like hydrogels, films, sponges, scaffolds and nanofibres for tissue engineering, wound healing, fibroblast cultivation and treatment of burns [161–164]. The production of smart biocomposites is also of interest. An example is the production of transparent plastic film containing citric acid from wool hydrolysate [165]. The plastic has excellent biocidal activity and could be used as a functional packaging for food.

The examples described above demonstrate the commercial potential of keratinases and the large number of opportunities they offer for adding value to keratin waste by producing bioactive protein feed supplements, personal care products and biomaterials from keratin hydrolysates.

#### **7. Discovery and Future Research**

The standout problem with the characterization of keratinases, demonstrated by the analysis of the assay conditions in this review, is lack of standardization of the keratinase assay combined with the small number of sequenced peptidases with keratinolytic activity that have been biochemically characterized. Both of these issues hamper the identification and comparison of true keratinases. Current experimental conditions vary in temperature, pH, buffer types and concentration, additives, substrates and their pretreatment biasing possible conclusions. It is unclear whether some proteases are keratinases or whether pretreatment or co-treatment influences their keratinolytic activity to some degree.

The uncertainty in defining keratinases and highly variable characterization of keratinases in the literature increases the challenge of finding new keratinases based on literature data or from sequence databases. However, the discovery of new keratinases is critical for expanding the opportunities for waste keratin valorisation. It would be desirable to identify new keratinases with high activities and specificities enabling control over cleavage sites, peptide molecular weights and amino acid side chain modifications.

Standardized experiments combined with phylogenetic studies and sequence analyses are needed. Standardized experiments, which avoid pre- or co-treatments, would determine the true protease activity on keratin substrates and reduce possible experimental biases. An in-depth phylogenetic analysis would help to clarify the position of keratinases within the phylogenetic trees of the peptidase families in which they are found and may help focus the search for new peptidases with keratinolytic activity. A comprehensive sequence analyses, aimed at the identification of conserved sites between peptidases with keratinolytic activity, as well as the presence of specific domains that possibly contribute to their ability to hydrolyse keratin, may assist in the development of algorithms to search the vast sequence space of the peptidase families.

**Author Contributions:** Conceptualization, R.S., K.R., J.P.D.O.M., L.N., M.N., G.C. and Z.Z.; investigation, J.P.D.O.M. and K.R.; writing—original draft preparation, J.P.D.O.M. and K.R.; writing—review and editing, R.S., L.N., M.N. and G.C.; supervision, R.S.; project administration, R.S.; funding acquisition, R.S. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the Australian Government Department of Agriculture, grant number RnD4Profit-16-03-002.

**Acknowledgments:** This project is supported by Meat and Livestock Australia through funding from the Australian Government Department of Agriculture as part of its Rural R&D for Profit program and the partners.

**Conflicts of Interest:** The authors declare no conflicts of interest. This manuscript was approved for publishing by Meat and Livestock Australia and the Australian Government Department of Agriculture.

#### **References**

1. Tesfaye, T.; Sithole, B.; Ramjugernath, D. Valorisation of chicken feathers: A review on recycling and recovery route—current status and future prospects. *Clean Technol. Environ. Policy* **2017**, *19*, 2363–2368. [CrossRef]


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*
