Industrial Crops and Products 95 (2017) 202–206

Contents lists available at ScienceDirect

Industrial Crops and Products

journal homepage: www.elsevier.com/locate/indcrop

The use of factorial design for ferulic acid production by co-culture

Kamaliah Abdul Samad ∗ , Norazwina Zainol
Faculty of Chemical and Natural Resources Engineering, Universiti Malaysia Pahang, Lebuhraya Tun Razak, 26300, Gambang, Kuantan, Pahang, Malaysia

a r t i c l e i n f o a b s t r a c t

Article history: Several factors can influence the use of co-culture for ferulic acid production from banana stem waste.
Received 9 June 2016 The aim of this study was to analyse factors of temperature, pH, agitation, water-to-substrate ratio,
Received in revised form 19 August 2016 volume of inoculum, fermentation time, and type of co-culture by employing a 24 fractional factorial
Accepted 20 October 2016
design (FFD). The results showed the order of contribution effects towards ferulic acid production:
pH > type of co-culture > volume of inoculum > agitation > fermentation time > temperature > water-to-
Keywords:
substrate ratio. Four factors including pH, agitation, type of co-culture, and volume of inoculum were
Ferulic acid
shown to have significant effects on ferulic acid production. Among the determining factors, the main
Banana stem waste
Co-culture
factor of pH and the interaction of temperature and fermentation time had the strongest effect on fer-
Fractional factorial design ulic acid production. The result indicated that FFD was useful to improve the ferulic acid production by
considering all the interactions of variables involved.
© 2016 Elsevier B.V. All rights reserved.

1. Introduction Simoes, 2014). Donaghy et al. (1998) successfully screened some

Ferulic acid (FA) was first isolated from Ferula foetida and it
presents in plant cell wall components as covalent side chains.
Component of lignocelluloses is found with dihydroferulic acid col-
lectively. It gives rigidity to the cell wall by making the crosslink
between polysaccharides and lignin (Kumar and Pruthi, 2014). FA
was discovered in most monocot plant such as wheat bran, corn
kernel, sugar beet pulp, and rice of bran oil, with varying quantity
from 5 to 28 g/kg (Fazary and Ju, 2007). Extensive studies have been
performed on conversion of lignocellulosic material to FA. Agricul-
tural residues are sources of lignocellulosic material. Banana stem
waste (BSW) is a lignocelluloses waste, which consists of 15.42%
lignin, 53.45% cellulose, and 28.56% hemicellulose (Silveira et al.,
2008).
Microorganisms are used to degrade lignin and hemicellulose in
waste material by producing lignin-degrading enzyme to produce
ferulic acid (Hasyierah et al., 2008). A few microorganisms have
been identified to be able to hydrolyse FA from the materials of cell
wall (Faulds and Williamson, 1995). Several enzymes such as fer-
ulic acid esterase (FAE) and xylanase produced by microorganisms
act synergistically in maximising the degradation of plant cell wall
to release FA (Huang et al., 2011). Bacillus spp. was found as the
most potent extracellular enzyme producer (Balasubramanian and
∗ Corresponding author.
E-mail addresses: kamaliahabdulsamad@ymail.com (K.A. Samad),
azwina@ump.edu.my (N. Zainol).
Bacillus spp. as FAE producers, while Heck et al. (2002) found that
Bacillus subtilis contributed the highest xylanase activity in their
research. The utilisation of multi-enzyme by co-culture provides
a promising method to improve the product yield. However, best
condition of a biological process should be determined to ensure a
maximum yield of a product.
Design expert software (DX) is a tool that can be used efficiently
for design of experiment (DOE). The software provides a function
known as full factorial design which is useful to study and deter-
mine the influences of several variables necessary in a process
(Golshani et al., 2013). A few research works have been accom-
plished by employing factorial design for ferulic acid production
effectively. Ismail and Zainol (2014) proved that, by employing fac-
torial design, the best variable of temperature was determined in
producing the highest content of FA from BSW using sugarcane
press machine. In addition, Mussatto et al. (2007) also success-
fully applied the factorial design approach in maximising the FA
extraction from brewer spent grain through alkaline hydrolysis
method. They found that temperature had the greatest effect, fol-
lowed by the reaction time and NaOH concentration. Currently,
the application of factorial design on the production of FA by co-
culture through fermentation has never been reported. Therefore,
this research is a value added to expand the knowledge on the pro-
duction of FA by microorganisms. In this study, FFD was applied to
determine the significant factors involved in fermentation of BSW
by co-culture to produce FA. The effects of seven factors on FA pro-
duction were analysed using 24 fractional factorial design. FFD is a

http://dx.doi.org/10.1016/j.indcrop.2016.10.028
0926-6690/© 2016 Elsevier B.V. All rights reserved.
 K.A. Samad, N. Zainol / Industrial Crops and Products 95 (2017) 202–206
good alternative to full factorial design, due to many factors were
involved in this research.
2. Materials and methods
2.1. Microorganism
In this study, three strains of Bacillus spp. were isolated from
soil, which were Bacillus thuringiensis Bt407, Bacillus cereus strain
CCM 2010 and Bacillus pumilus SAFR-032. The bacterial strains were
acclimatised for a month in BSW before the isolation. The strains
were identified by 16S rRNA sequence analysis method. All the
strains were grown and maintained on nutrient agar plate.
2.2. Substrate
BSW was obtained from banana plantation in Kuantan, Pahang.
The stem was cleaned to remove any dirt attached before being
used as substrate. The stem was cut into a cube (1 cm). The sub-
strates were prepared at different ratios by mixing them with water
and blended. Initial pH was adjusted according to Table 1. The sub-
strate was prepared in 250-ml Erlenmeyer flask for each set of
experiment and sterilised at 121 ◦ C for 15 min.
2.3. Preparation of co-culture
Before starting the experiment, bacterial strains of Bacillus
pumilus SAFR-032, Bacillus cereus strain CCM 2010 and Bacillus
thuringiensis Bt407 were resuscitated by plating onto nutrient agar
plate and incubated at 37 ◦ C for 24 h. A single colony of each strain
was scraped and inoculated into a universal bottle containing 10 ml
of nutrient broth. The cultures were grown for 22 h at 37 ◦ C in an
incubator. A total of 10% of the enriched cultures were added to
50 ml sterile nutrient broth in 100-mL Erlenmeyer flask and incu-
bated further for 22 h at 37 ◦ C. Co-culture of A (B. thuringiensis, B.
cereus, and B. pumilus) and B (B. thuringiensis and B. cereus) were
prepared. B. thuringiensis, B. cereus, and B. pumilus were grown as
co-culture by adding a total of 50 ml stationary phase inoculum
of single culture at equal volume into 1000-mL Erlenmeyer flask
containing 500 ml of nutrient broth. Inoculum was incubated in an
incubator shaker with mild shaking for another 22 h at 37 ◦ C.
2.4. Experimental set-up
Fermentation of BSW involving co-culture was performed under
aseptic technique. Inoculum of co-culture at stationary phase was
inoculated into 250-ml Erlenmeyer flask containing substrate.
Incubation was conducted in an incubator shaker at different tem-
peratures and agitation speed according to Table 1. Sample was
collected at various times (Table 1) and centrifuged. The FA content
in the supernatant was quantified using high performance liquid
chromatography (HPLC). The outputs of the experimental design
were analysed with Design Expert software to evaluate the effects
of each factor involved.
2.5. Factorial design methodology
In this study, Design Expert software (Version 8.0.6, State-Ease)
was used. The screening was performed through fractional fac-
torial design (FFD) to determine the experimental variables and
interactions that have significant influence on the production of
203
ferulic acid. The experimental data were analysed by FFD to fit the
following first order polynomial equation (Eq. (1)):

n
Y = ˇ0 + ˇi Xi (1)
i=1
where Y represents the value of the response, ˇ0 is the constant
coefficient, n is the number of variables, ˇi represents the coeffi-
cient of the linear parameters, and Xi represents the coefficient of
the interaction parameters.
A 24 fractional factorial design consisting of 16 runs were per-
formed for all the parameters, as shown in Table 2.
2.6. Analytical method
Samples were analysed according to Chamkha et al. (2001) with
modification using HPLC (Agilent 1100 system). It was equipped
with Agilent Zorbaq SB-AQ C18 analytical column with an isocratic
mobile phase consisted of acetonitrile, distilled water, and acetic
acid (30:69.5:0.5, v/v). The flow rate was set at 0.6 ml/min. Ferulic
acid was injected via 25-ml injection loop and measured using a
diod array detector (DAD) at 280 nm. Prior to the analysis, sam-
ples were prepared by centrifugation at 5800 rpm for 15 min. The
supernatant were filtered into vial using 0.45 ␮m nylon syringe
filter.
3. Results and discussion
3.1. Statistical analysis for ferulic acid production from BSW
Table 2 shows the amount of ferulic acid obtained within
the range of 40.22 to 488.41 mg/kg. The highest content of FA
(488.41 mg/kg) was obtained when the fermentation was per-
formed for 24 h at 26 ◦ C, initial pH of 9.5, agitation at 150 rpm,
water-to-substrate concentration one of 2:1, with 2% inoculum of
co-culture B (B. cereus and B. thuringiensis) as shown in standard
order 7.
The analysis of variance (ANOVA) analysis for FA production was
done to determine the significance of the model (Table 3). The sta-
tistical significance of regression equation can be checked using
F-values, while the p-values were used to examine the significance
of each coefficient. From the model, the F-value is 171.31 and the
p-value is very low (p < 0.0001). There was only 0.01% chance that
the model F-value this large could occur due to noise. Generally, a
good model is implied by the greater of the calculated F-value sev-
eral times compared to the tabulated value (Hamzaoui et al., 2008).
The smaller p-values show the more the significance of the corre-
sponding variable (Masoumi et al., 2011). The model term effect of
B, C, E, G, AC, AE, AF, and BD were statistically significant in affect-
ing the ferulic acid production. While, the model term of A, D, and
F were not significant, as their p-values are larger than 0.05.
The R2 from the ANOVA is used to indicate how close the data to
the regression line. A good fitting model shows R2 of more than 80%
(Karazhiyan et al., 2011). The satisfactory R2 value was obtained
from this analysis (0.9979), which indicates that the model fits the
experimental and predicted values well. The final equations in term
of actual factors were determined as follows:
FAA = 120.64 − 5.24A + 77.73 B + 1.22C + 29.87D − 16.09E
− 8.69F − 0.03AC + 0.37AE + 0.27AF − 4.02BD (2)
FAB = 164.58 − 15.24A + 77.73 B + 1.22C + 29.87D − 16.09E
− 8.69F − 0.03AC + 0.37AE + 0.27AF − 4.02BD (3)
204 K.A. Samad, N. Zainol / Industrial Crops and Products 95 (2017) 202–206

Table 1
Variables and their coded and actual levels used in the method of 24 fractional factorial experiments.

No. Variables Coded Type of factor Actual values of coded levels Units

−1 +1
◦
1 Temperature A Numerical 26 40 C
2 pH B Numerical 5.5 9.5 pH units
3 Agitation C Numerical 0 150 rpm
4 Water-to-substrate ratio D Numerical 2:1 10:1 w/w
5 Volume of inoculum E Numerical 2 10 %, v/v
6 Fermentation time F Numerical 24 72 hours
7 Type of co-culture G Categorical Aa Ba
a
Co-culture A (B. cereus, B. pumilus and B. thuringiensis) and B (B. cereus and B. thuringiensis).

Table 2
The design of the 24 fractional factorial experiments.

Standard Order Coded values of variables FA (mg/kg)

A B C D E F G

1 −1 −1 −1 −1 −1 −1 −1 87.2027
2 +1 −1 −1 −1 +1 −1 +1 40.2194
3 −1 +1 −1 −1 +1 +1 −1 237.364
4 +1 +1 −1 −1 −1 +1 +1 423.303
5 −1 −1 +1 −1 +1 +1 +1 56.2657
6 +1 −1 +1 −1 −1 +1 −1 114.081
7 −1 +1 +1 −1 −1 −1 +1 488.409
8 +1 +1 +1 −1 +1 −1 −1 263.808
9 −1 −1 −1 +1 −1 +1 +1 133.519
10 +1 −1 −1 +1 +1 +1 −1 153.829
11 −1 +1 −1 +1 +1 −1 +1 289.284
12 +1 +1 −1 +1 −1 −1 −1 218.093
13 −1 −1 +1 +1 +1 −1 −1 162.206
14 +1 −1 +1 +1 −1 −1 +1 97.1652
15 −1 +1 +1 +1 −1 +1 −1 290.215
16 +1 +1 +1 +1 +1 +1 +1 350.176

Table 3
Test of significance for regression coefficient.

Source CoefficientEstimate Sum ofSquares F Value p-valueProb > F

Model 212.82 259171.29 171.31 <0.0001 significant

A-Temperature −5.24 438.81 3.19 *0.1486
B-pH 107.26 184075.84 1338.41 <0.0001
C-Agitation 14.97 3585.38 26.07 0.0070
D-Water-to-substrate ratio −1.01 16.33 0.12 *0.7477
E-Volume of inoculum −18.68 5581.41 40.58 0.0031
F-Fermentation time 7.02 789.13 5.74 *0.0747
G-Type of co-culture 21.97 7723.90 56.16 0.0017
AC −16.25 4223.05 30.71 0.0052
AE 13.10 2746.19 19.97 0.0111
AF 45.74 33474.66 243.39 <0.0001
BD −32.13 16516.58 120.09 0.0004
Residual 550.13
Cor Total 259721.42

R2 = 0.9979, *Values of p-values greater than 0.05 indicating the model terms are not significant.

where FAA and FAB are referred as the response of ferulic acid yield
using co-culture A and co-culture B, respectively. A is temperature,
B is pH, C is agitation, D is water-to-substrate ratio, E is volume of
inoculum, and F is fermentation time. Factors of A, B, C, D, E, and
F are referred as the main effects, while AC, AE, AF, and BD are the
interaction effects.
3.2. Main and interaction effects of the factors
Fig. 1 shows the Pareto chart representing the main and inter-
action effects of factors involved in the process. The chart is very
useful to analyse the most significant factors. The height of the bars
presents the impact of the factors. The t-values of the bars are the
values of the square root of the F-values from ANOVA. Two limit
lines, namely Bonferroni limit line and t-value limit line, present
the t-value of the effects. The value of both lines are 6.254 and
2.776, respectively. The Pareto chart clearly shows that the factors
of B, C, E, G, AC, AE, AF, and BD exceed the t-value limit of 2.776 and
have significant effects for ferulic acid production.
Among the factors, the main factor of pH remarkably exceeded
the Bonferroni limit, which present a much stricter level of 0.025
by cutting p-value of 0.05 into the half. Salleh et al. (2011) also
reported that pH was strongly affecting the extraction of ferulic acid
from paddy straw. Meanwhile, the interaction between tempera-
ture and fermentation time was the most significant interaction
that exceeded the Bonferroni limit. In fermentation process, the
suitable condition of temperature is necessary for FA production.
The FA content from the malted barley was decreased at high tem-
perature as a result of denatured enzyme that supposedly used to
release ferulic acid from the cell wall (Humberstone and Briggs,
 K.A. Samad, N. Zainol / Industrial Crops and Products 95 (2017) 202–206
Fig. 1. Pareto chart showing the relative effect of various factors on ferulic acid production.
2000). According to Shah and Pathak (2010), a significant coeffi-
cient is represented by the coefficients with t-value of effect above
the Bonferroni line, while a coefficient is probable to be significant
when the t-value of effect is between Bonferroni line and t-value
limit line.
3.3. Production of ferulic acid through fermentation
Co-culture (B. cereus and B. thuringiensis) fermentation of BSW
released 488.41 mg/kg of ferulic acid. The result was 53% higher
compared to FA found by Ismail and Zainol (2014) although the
same type of raw material was used. In their experiment, about
227.4 mg/kg of FA was extracted through mechanical extraction
process using sugarcane press machine. Besides, different agricul-
tural wastes to obtain ferulic acid using microbial also were also
studied. Sarangi and Sahoo (2010) presented that about 275 mg/kg
of FA was produced by the incubation of wheat bran with Staphy-
lococcus aureus for 6 days, By comparison, the amount of FA from
BSW was 43% higher.
Using other method, Salleh et al. (2011) reported that about
5181 mg/kg of FA was obtained from paddy straw via alkaline
hydrolysis using 2 M of NaOH at 100 ◦ C. It is obvious that FA was
extracted at higher amount (91%) using the chemical extraction
method compared to biological method done in this research. How-
ever, the use of biological method is considered natural for FA
production compared to alkaline hydrolysis method (Kumar and
Pruthi, 2014). Besides, it could cost less due to low energy consump-
tion (Verardi et al., 2012). In this study, a low temperature (26 ◦ C)
and a short fermentation time (24 h) were enough to efficiently
release ferulic acid from BSW.
4. Conclusions
The production of FA from BSW by co-culture was studied
using 24 fractional factorial design. From the analysis, pH, agita-
tion, volume of inoculum, and type of co-culture were identified
as significant effects on FA production. In addition, the interac-
tion of temperature and agitation, temperature and volume of
205
inoculum, temperature and fermentation time, and pH and water-
to-substrate ratio were also significant. Among the factors, the main
factor of pH and the interaction of temperature and fermentation
time were determined to be the most effective in influencing the
FA production. The findings revealed that FFD has the ability in
examining the effects of a large number of factors contributing to
a process effectively with a minimum number of experiments.
Acknowledgments
The authors would like to gratefully acknowledge the Faculty of
Chemical and Natural Resources Engineering, University Malaysia
Pahang for providing the facilities to accomplish the research.
References
Balasubramanian, N., Simoes, N., 2014. Bacillus pumilus S124A carboxymethyl
cellulase; a thermo stable enzyme with a wide substrate spectrum utility. Int. J.
Biol. Macromol. 67, 132–139.
Chamkha, M., Patel, B.K., Garcia, J.L., Labat, M., 2001. Isolation of Clostridium
bifermentans from oil mill wastewaters converting cinnamic acid to
3-phenylpropionic acid and emendation of the species. Anaerobe 7, 189–197.
Donaghy, J., Kelly, P.F., McKay, A.M., 1998. Detection of ferulic acid esterase
production by Bacillus spp. and lactobacilli. Appl. Microbiol. Biotechnol. 50,
257–260.
Faulds, C.B., Williamson, G., 1995. Release of ferulic acid from wheat bran by a
ferulic acid esterase (FAE III) from Aspergillus niger. Appl. Microbiol.
Biotechnol. 43, 1082–1087.
Fazary, A.E., Ju, Y.H., 2007. Feruloyl esterases as biotechnological tools: current and
future perspectives. Acta Biochim. Biophys. Sin. 39, 811–828.
Golshani, T., Jorjani, E., Chelgani, S.C., Shafaei, S.Z., Nafechi, Y.H., 2013. Modeling
and process optimization for microbial desulfurization of coal by using a
two-level full factorial design. Int. J. Min. Sci. Technol. 23, 261–265.
Hamzaoui, A.H., Jamoussi, B., M’nif, A., 2008. Lithium recovery from highly
concentrated solutions: response surface methodology (RSM) process
parameters optimization. Hydrometallurgy 90, 1–7.
Hasyierah, Noor, Mohamed Zulkali, M.S., Ku Syahidah, M.D., Ku, I., 2008. Ferulic
acid from lignocellulosic biomass: review. Proc. MUCE, T2008.
Heck, J.X., Hertz, P.F., Ayub, M.A., 2002. Cellulase and xylanase productions by
isolated Amazon Bacillus strains using soybean industrial residue based
solid-state cultivation. Braz. J. Microbiol. 33, 213–218.
Huang, Y.C., Chen, Y.F., Chen, C.Y., Chen, W.L., Ciou, Y.P., Liu, W.H., Yang, C.H., 2011.
Production of ferulic acid from lignocellulolytic agricultural biomass by
Thermobifida fusca thermostable esterase produced in Yarrowia lipolytica
transformant. Bioresour. Technol. 102, 8117–8122.
206 K.A. Samad, N. Zainol / Industrial Crops and Products 95 (2017) 202–206
Humberstone, F.J., Briggs, D.E., 2000. Extraction and assay of ferulic acid esterase
from malted barley. J. Inst. Brew. 106, 21–30.
Ismail, S.N., Zainol, N., 2014. Factorial analysis of ferulic acid extraction from
banana stem waste. Int. J. Appl. Eng. Res. 9, 6823–6833.
Karazhiyan, H., Razavi, S.M., Phillips, G.O., 2011. Extraction optimization of a
hydrocolloid extract from cress seed (Lepidium sativum) using response surface
methodology. Food Hydrocolloids 25, 915–920.
Kumar, N., Pruthi, V., 2014. Potential applications of ferulic acid from natural
sources. Biotechnol. Rep. 4, 86–93.
Masoumi, H.R.F., Kassim, A., Basri, M., Abdullah, D.K., 2011. Determining optimum
conditions for lipase-catalyzed synthesis of triethanolamine (TEA)-based
esterquat cationic surfactant by a Taguchi robust design method. Molecules 16,
4672–4680.
Mussatto, S.I., Dragone, G., Roberto, I.C., 2007. Ferulic and p-coumaric acids
extraction by alkaline hydrolysis of brewer’s spent grain. Ind. Crops Prod. 25,
231–237.
Salleh, N.H.M., Daud, M.Z.M., Arbain, D., Ahmad, M.S., Ismail, K.S.K., 2011.
Optimization of alkaline hydrolysis of paddy straw for ferulic acid extraction.
Ind. Crops Prod. 34, 1635–1640.
Sarangi, P.K., Sahoo, H.P., 2010. Ferulic acid production from wheat bran using
Staphylococcus aureus. New York Sci. J. 3, 79–81.
Shah, M., Pathak, K., 2010. Development and statistical optimization of solid lipid
nanoparticles of simvastatin by using 23 full-factorial design. AAPS
PharmSciTech 11, 489–496.
Silveira, M.L.L., Furlan, S.A., Ninow, J.L., 2008. Development of an alternative
technology for the oyster mushroom production using liquid inoculum. Food
Sci. Technol. (Campinas) 28, 858–862.
Verardi, A., Ricca, E., De Bari, I., Calabrò, V., 2012. Hydrolysis of Lignocellulosic
Biomass: Current Status of Processes and Technologies and Future
Perspectives. INTECH Open Access Publisher.