**3. Discussion**

Keratinase is a protease that is very robust with broad application in industry. In this work, isolation of a new potential keratinase producer from the *Bacillus* genera due to its high keratinolytic activity [28]. Five colonies were successfully isolated from poultry waste and soil in Selangor using heat treatment method for the selection of spore-forming *Bacillus* species. As one of the exclusivities of *Bacillus* species is characterized by endospore formation, heat treatment is the most common and simplest method use to infer the presence of spore-forming *Bacillus* [29,30]. Under extreme environment, spore-forming *Bacillus* spp. develop endospore; a metabolically inactive dormant cell to

protect themselves against the harsh environment [31]. Once the environment returns to favorable conditions for growth, the cell will proceed with its vegetative life cycle and continue to germinate. The purpose of sporulation in this experiment is to kill other vegetative cells, leaving only dormant endospore cells to survive. Of all five cultures tested, only isolate UPM-AAG1 shows positive growth response in FMA, spore positive, and highest keratinase activity, as well as the highest bacterial growth count; hence, this isolate was selected for further studies. This method of isolation of keratinase producing organism has been reported before [32], where feather as a substrate was used as sole carbon and nitrogen source. A few keratinolytic *Bacillus* spp. that utilize feather solely as carbon and nitrogen sources include *Bacillus pumilus* GRK [26] and *Bacillus cereus* LAU 08 [27].

Physiological and biochemical identification of isolate AAG1 revealed a rod-shaped structure indicating their Gram-positive characteristic. Gram-positive keratinase producer is not exclusive to *Bacillus* spp. as other Gram-positive bacteria have been reported, including *BreviBacillus parabrevis* [33], *Micrococcus luteus*, and *Actinobacter* sp. [34]. On the basic of 16srDNA sequencing and phylogenetic analysis, the keratinolytic isolate UPM-AAG1 was tentatively identified as *Bacillus* sp. strain UPM-AAG1. Many of the major commercial keratinolytic bacteria come from the genus *Bacillus* spp. [28,35], chiefly due to its generally regards as safe (GRAS) property [36]. The keratinolytic bacteria in this study were isolated from a poultry farm environment, and numerous keratinolytic bacteria have been isolated from poultry farm soils, such as *Bacillus subtilis* DP [25] and five keratinolytic strains of *Bacillus* spp. [37], to name a few, making them predictable as the feather-contaminated soils offer rich sources of keratin for selective enrichment [25,38].

Despite being the most studied and widely documented, the major drawback in keratinase study is to optimize keratinase production using feather keratin as the sole carbon and nitrogen sources. Addition of supplements will incur a high cost when production is scaled up (Table 7).


**Table 7.** Summary of keratinase production by *Bacillus* spp.


**Table 7.** *Cont*.

In keratinase research, the main objective is to maximize keratinase production through manipulating external and internal parameters [46]. Most of the optimization studied involving keratinase involved conventional optimization through one factor-at-a-time (OFAT) [22,41,47] or both conventional and statistical approach [17,48] but with non-keratin carbon and nitrogen sources (Table 7). By carrying out optimization relying on no additional supplements, as well as optimizing process at ambient temperature suiting Malaysia (from 24 to 32 ◦C), will increase the chances of a successful keratinase production by local small and medium enterprise (SME) companies. To date, only very few keratinase-producing *Bacillus* spp. bacteria have been optimized using feather as the sole carbon and nitrogen sources (Table 7). Additional C and N sources supplementation may not work during actual feather degradation as the augmented bacterium may choose to utilize the easily assimilable C and N source rather than the feather itself. In addition, competition with the easily assimilable C and N sources by indigenous bacteria may outcompete the augmented bacterium resulting in a lower production of keratinase and poor degradation of feather waste [49]. Compared to many keratin-degrading *Bacillus* spp. (Table 7), *Bacillus* sp. UPM-AAG1 produce a relatively good keratinase activity (60.1 U/mL) in a shorter time (24 h) at 30 ◦C, features that make this bacterium suitable for the requirement of the local SME company where keratinase production should be optimum or acceptable activity at ambient temperature. However, the applicability of this strain in real world conditions need to be tested, and this remains the limit of this work.

RSM CCD's result showed that all factors—temperature, inoculum size (v/v), pH, and feather concentration (w/v)—exert positive effects to the model with feather concentration forming a major contributor. This result is similar to Yusuf et al. [48], where feather concentration was found to be the most significant substrates for keratinase production when Plackett-Burman (PB) was used in the screening process. Apart from that, Govarthanan et al. [24] also reported the same result where a significant increase in keratinase production was observed when feather was used as substrate. The inoculum size was reported to give significant effect towards keratinase production in *Bacillus licheniformis* ER-15 [43]. This is because inoculum size significantly affects the growth profile of aerobic microorganism. Further, a neutral to alkaline pH were reported to promote keratinase production in various microorganism [50,51] with the exception of a few including *Bacillus subtilis* [52] where the highest activity occurs at acid to neutral pH range (pH 5–7). Apart from that, temperature also plays an important role in the production of keratinase enzyme. Generally, most of the reported

keratinases work optimally in between 28 to 50 ◦C [24,39]. The ANOVA analysis result of keratinase activity obtained through RSM indicated that the model is adequate with a correlation coefficient, R<sup>2</sup> of the model was 0.9569 and an adjusted R<sup>2</sup> value of 0.9167 showing a high correlation between the experimental data design (Table 5). The nearer the R<sup>2</sup> value to 1, the better the accuracy of the model. The "Pred R-Squared" of 0.8165 was in consistent agreemen<sup>t</sup> to the "Adj R-Squared" of 0.9167 indicating an acceptable degree of correlation between the observed value and predicted values [53], although "Pred R-Squared" value of >0.9 is more desirable in many cases [54]. A ratio > 4 for the Adeq Precision value is sought and the result from this study with a value of 15.586 indicates a good signal to noise ratio [55]. The large lack of fit F value is normally sough and, with a value of 0.94, suggests an insignificant lack of fit relative to the pure error [56,57]. The *p*-value for the lack of fit value was 0.5666, and this demonstrated the model appropriateness for the optimal region. The Model F-value of 23.79 implies the model is significant. There is only a 0.01% chance that "Model F-value" this large could occur due to noise. Values of "Prob > F" less than 0.0500 indicate model terms are significant [58]. Significant model terms in this case were A, B, C, D, A2, C2, D2, BC, BD. Further 3D analysis of the model showed an escalated pattern in keratinase production when the temperature was increased, and pH maintained in the targeted range (Figure 3a). The same escalating pattern was also observed in Figure 3b and c where keratinase production increased only when the temperature was increased and could not increase further with increasing in inoculum size and feather concentrations, respectively. Verification of the model with a predicted value close to the actual value showed the reliability of the experiment to predicted precise condition, thus supporting the accuracy of the model over 95%.

The amino acid profile of hydrolysate of *Bacillus* sp. UPM-AAG1 revealed the presence of both essential amino acid and non-essential amino acids. The result is in accordance with Reference [19], that demonstrated the presence 17 different amino acid acquired from fermentation of *Bacillus cereus* utilizing chicken feather as sole their carbon and nitrogen sources. The fermented hydrolysate was rich with nutritionally essential amino acid particularly lysine, threonine, and methionine. Similarly, Ghosh et al. [59] reported on purified keratinolytic protease from feather waste hydrolysate by *Bacillus cereus* DCUW that comprises of 17 different amino acids.

#### **4. Materials and Methods**

#### *4.1. Azokeratin and Keratinase Assay*

Azo keratin substrate was prepared [60] with the modification where, instead of ball milling, the feather was cut into small pieces with a scissor. One gram of a finely cut white chicken feather, 20 mL deionized water, and 10% of NaHCO3 were mixed in a 100 mL round bottom flask. Separately, 0.174 g of sulfanilic acid was dissolved in 5 mL of 0.2 M NaOH. Next 0.069 g of NaNO2 was added to the suspension. The solution then was acidified with 0.4 mL of 5 M HCl for 2 min and neutralized by 0.4 mL of 5 M NaOH. The prepared solution was then added to finely cut feather keratin and mixed properly for 10 min. The reaction mixture then was filtered. Insoluble azo keratin was retrieved and rinsed with deionized water. The azo keratin was then suspended in water and shaken for 2 h at 50 ◦C. The pH of the filtrate and absorbance readings were taken periodically until the pH of the filtrate reached 6.0–7.0 and the absorbance value was less than 0.01 [60]. The resulting azokeratin (Supplementary Figure S1) is utilized for keratinase assay. All experiments were carried out three times unless otherwise stated.

The keratinase activity was determined using azo keratin as a substrate. 5 mg azo keratin substrate was added to 1.5 mL mini centrifuges tube together with 800 μL of 0.1 M phosphate buffer pH 8.0. Then, 200 μL of enzyme supernatant was added to the mixture. The mixture was vortexed thoroughly and incubated at 30 ◦C for 30 min in a water bath. The enzymatic reaction was stopped by 200 μL of 10% (w/v) trichloroacetic acid added to the mixture, and the absorbance was read at 450 nm (DTX 800-Multimode detector, Beckman Coulter, Brea, CA, USA). Control was prepared by adding trichloroacetic acid (TCA) to the mixture before the enzyme. One unit of keratinase activity was

defined as 0.010 unit increase in the absorbance at 450 nm compared to control [48]. All experiments were carried out in triplicate, unless stated otherwise.

#### *4.2. Isolation and Screening of Bacillus sp. with Keratinolytic Activity*

Soil samples and poultry waste specimens were collected from a waste collection area of a poultry research farm in Universiti Putra Malaysia. One percent (w/v) of each soil samples and poultry waste were dissolved in 10 mL of sterilized phosphate bu ffer saline (PBS) and incubated in 80 ◦C water bath for 10 min to further biased the selection towards spore-forming *Bacillus* species. The PBS medium used was adopted from Dulbecco and Vogt [61]. The suspension (100 μL) was spread on nutrient agar (NA) supplemented with keratin substrate. The plates were incubated at room temperature (25 ◦C) for 24–48 h. Surviving bacteria that showed di fferent morphology and high hydrolysis zone on NA were further re-streaked on NA until pure cultures were obtained [62].

All potentially isolated keratinase-producing bacteria were screened according to the ability to develop endospore under stress environment in a sporing medium (pH 7.0) composed of g/L: 1.6 NH4 Cl, 0.9 K2HPO4, 0.6 KH2 PO4, 0.2 MgSO4·7H2O, and 0.07 CaCl·2H2O, 0.01 FeSO4·7H2O and 0.01 EDTA for two days at 25 ◦C under shaking condition at 150 rpm. All strains were spore stained with malachite green and safranin according to Reference [63]'s method and observed under a light microscope (Olympus BX.40F4, Japan) with 100× magnification [64]. Positive endospore-forming isolate were further screened based on bacterial growth (CFU/mL) on feather meal agar (FMA) composed of (g/L); 1.0 feather, 0.5 NaCl, 0.7 K2HPO4, 0.001 MgSO4·6H2O and 15.0 bacteriological agar and keratinase assay with 1% feather as sole carbon and nitrogen sources [48].

#### *4.3. Morphological, Biochemical and Molecular Identification of Keratinolytic Microorganism*

The identity of the selected bacterium was further identified morphologically using Gram staining method and a series of biochemical test (oxidase test, catalase, Voges-Proskauer, nitrate, citrate, lipase, and gelatinase) [64]. Meanwhile, molecular identification confirmation was performed by 16S rDNA sequence analysis using a 24 h culture of the bacterial cell using InnuPREP Bacteria DNA Kit (Analytik Jena, Jena, Germany) according to the manufacturer's protocol. Amplification of the partial 16S rRNA gene was carried out using universal primers. The PCR mixtures comprise of mixtures of 1μL of 5 mM 27F (5-AGA GTT TGATCC TGG CTC AG-3) and 1429R (5-TAC GGT TACCTT GTT ACG ACTT-3) of forward and reverse primer, 1 μL of DNA sample, 12.5 μL of Master mix 2 × Taq (Vivantis Technologies Sdn. Bhd., Selangor, Malaysia) and 9.5 μL sterile deionized water for a final volume of 25 μL. The polymerase chain reaction was accomplished using a gradient thermocycler (Hercuvan, Milton, UK) under the following conditions: 3 min initial denaturation at 94 ◦C, 29 cycles denaturation for 1 min at 94 ◦C, 1 min of annealing at 58 ◦C, 2 min of extension for 10 min, and final extension at 72 ◦C for 10 min with incubation at 4 ◦C. Successfully amplified DNA fragments were analyzed on 1% (w/v) agarose gel [65].

#### *4.4. Sequence Analysis and Phylogenetic Analysis*

The selected sequence was analyzed using BLASTn [66]. Twenty sequence alignment with more than 95% similarity was selected for further analysis using neighbor-joining method, as in Ref. [67], fitting to the distances of Jukes-Cantor [68]. Phylogenetic analysis was done using PHYLIP software v3.696 (http://evolution.genetics.washington.edu/phylip.htmL). *E. coli* strain U5/41 was used as the outgroups in the cladogram for identification analysis. The confidence level of each branch was calculated by 1000 bootstraps replicates. The constructed tree was viewed using Tree View version 1.6.6.

#### *4.5. Optimization of Keratinase Activity Using Response Surface Methodology*

The e ffect of four factors namely temperature, inoculum size (v/v), pH, and feather concentration (w/v) on keratinase production was screened statistically using Plackett-Burman factorial design (PBFD) to verify the significance of the named factors in the production of keratinase. The experimental design and statistical analysis were performed using statistical software Design-Expert® 6.0.8 (Stat-Ease, Minneapolis, MN, USA). Each independent factor was evaluated at two different levels; minimum and maximum levels (+1, −1) as shown in Table 8. Keratinase activity was analyzed as the response. The independent factors that show significance by PBFD were optimized further for their interaction effects by composite design (CCD) of response surface methodology (RSM). Each independent factor was studied at five different level: <sup>−</sup>α, −1, 0, +1, +α. Keratinase activity was evaluated as a response based on 30 experimental design. All experiments were conducted in triplicate, and keratinase activity was examined as the response using a second-order polynomial equation as below:

$$y = \sum\_{i=1}^{k} \beta\_i X\_i + \sum\_{i}^{k} \beta\_{ii} X\_{i2} + \sum\_{1 \le i \le j}^{k} \beta\_{ij} X\_i X\_j$$

where *Y* is the predicted response, *X* is the independent factor that is affected by *Y*, *k* is the number of factors, β0 is the constant term, β*i* is the linear coefficient, βii is the *i* the quadratic coefficient, and β*ij* is the *ij* the interaction coefficient, whereas *i* and *j* = 1,2,3 and *i* - *j* are coefficient in the model. The significance of each coefficient in the equation was determined by Fisher's F test and analysis of variances (*p* < 0.05). The experimental design and statistical analysis were performed using statistical software Design-Expert® 6.0.8 (Stat-Ease, Minneapolis, MN, USA). All experiments were conducted in triplicate.

**Table 8.** Experimental factors and level of minimum and maximum range for statistical screening using Plackett-Burman factorial design (PBFD).


#### *4.6. Amino Acid Profile of Hydrolysate*

Amino acid profile of hydrolysate was performed according to a previous method [69], with slight modifications. The amino acid profile of sample hydrolysate was determined using an HPLC system (Agilent 1200, Agilent Technologies, Santa Clara, CA, USA). The sample was subjected to automated pre-column derivatization using orthopthalaldehyde (OPA) run through the injector program. The injector program protocols were as follows where 2.5 μL were drawn from a borate buffer vial (0.4 min, pH 10.2). Next, 0.5 μL of the sample was drawn from a sample vial, followed by mixing with 3 μL in a wash port five times and waiting for 0.2 min. Next, 0.5 μL of orthopthalaldehyde (OPA) was drawn, followed by mixing of 3.5 μL in wash port 6 times. Next, 32 μL of injection diluent (1 mL of mobile phase A + 15 μL of concentrated H3PO3) was mixed with 20 μL in seat 8 times. The sample was injected, then wait for 0.10 min and valve bypass. The mobile phase A consisted of 10 mM of Na2HPO4, 10 mM Na2B4O7, pH 8.2, and mobile phase B (acetonitrile-methanol-water; 45:45:10, v/v). A programmed gradient elution was performed from 2% B to 57% B for 7 min, followed by 57% B to 100% B for 8.4 min, with a flow rate of 1.5 mL/min at 40 ◦C. Amino acid detection was detected with a 250 nm Diode Array Detector (DAD) detector using an amino acid standard solution (Sigma-Aldrich, St. Louis, MO, USA).
