JOURNAL OF BIOSCIENCE AND BIOENGINEERING © 2008, The Society for Biotechnology, Japan

Vol. 105, No. 3, 232–237. 2008

DOI: 10.1263/jbb.105.232

Optimization of Critical Medium Components Using Response Surface

Methodology for Phenazine-1-Carboxylic Acid Production
by Pseudomonas sp. M-18Q
Li-Li Yuan,1 Ya-Qian Li,1 Yi Wang,1 Xue-Hong Zhang,1 and Yu-Quan Xu1*
Key Laboratory of Microbial Metabolism, Shanghai Jiao Tong University,
800 Dongchuan Road, Shanghai 200240, P.R. China1

Received 6 September 2007/Accepted 15 December 2007

The optimal flask-shaking batch fermentation medium for phenazine-1-carboxylic acid (PCA)
production by Pseudomonas sp. M-18Q, a qscR chromosomal inactivated mutant of the strain
M18 was studied using statistical experimental design and analysis. The Plackett–Burman design
(PBD) was used to evaluate the effects of eight medium components on the production of PCA,
which showed that glucose and soytone were the most significant ingredients (P <0.05). The steep-
est ascent experiment was adopted to determine the optimal region of the medium composition.
The optimum composition of the fermentation medium for maximum PCA yield, as determined
on the basis of a five-level two-factor central composite design (CCD), was obtained by response
surface methodology (RSM). The high correlation between the predicted and observed values in-
dicated the validity of the model. A maximum PCA yield of 1240 mg/l was obtained at 17.81 g/l
glucose and 11.47 g/l soytone, and the production was increased by 65.3% compared with that using
the original medium, which was at 750 mg/l.

[Key words: Pseudomonas sp. M-18Q, phenazine-1-carboxylic acid, Plackett–Burman design (PBD), response
surface methodology (RSM), medium optimization]

Through the direct suppression of fungal phytopathogens,
phenazine-1-carboxylic acid (PCA), a broad-spectrum anti-
biotic, is effective for disease control (1). It is confirmed
that PCA has the capacity to inhibit the mycelial growth of
several agricultural important fungal pathogens, including
Fusarium spp. and Pythium spp. (2). It was reported that
PCA produced by Pseudomona fluorescens 2–79 is an es-
sential component for the effective control of wheat take-all
disease caused by Gaeumannomyces graminis var. tritici
(3). Our studies showed that PCA, which is produced by
Pseudomonas sp. M18, one of the plant-growth-promoting
rhizobacteria (PGPR), has a strong suppressive inhibitory
activity against Fusarium spp. and several other fungal phyto-
pathogens of agricultural economic significance as well (4).
In China, PCA has been named Shenqinmycin and gained a
Pesticide Registration Certification by the Ministry of Agri-
culture because of its high efficiency against various phyto-
pathogens, low toxicity, and good environmental compati-
bility.
At present, the contradiction between low production and
high demands is becoming acute, giving rise to the urgency
of addressing the problem of increasing the production of
PCA. The qscR gene, encoding the quorum-sensing regulator
QscR, a luxR-type transcriptional regulator, was first found
to be a quorum-sensing control repressor in Pseudomonas
* Corresponding author. e-mail: xuyq@sjtu.edu.cn
phone: +86-21-34204347 fax: +86-21-34204854
aeruginosa (5). It controlled the production of many exo-
proteins and secondary metabolites via a hierarchial quorum-
sensing (QS) regulatory cascade (5, 6). Pseudomonas sp.
M-18Q, constructed by inactivating the chromosomal qscR,
which negatively regulates PCA production in Pseudomo-
nas sp. M18, was observed to have a higher potential for
producing more PCA than the wild type of strain M18 (7).
Aside from genetic alteration, culture components also affect
the yield of PCA. However, conventional methods of opti-
mization involve changing variable-to-variable while main-
taining all others at a fixed level; these often do not yield re-
liable results because interactions between different compo-
nents are neglected. Besides, they are laborious, time-con-
suming, and impractical. Recently, many statistical experi-
mental design methods have been developed to eliminate
these drawbacks. For example, the Plackett–Burman design
(PBD), which ignores interactions among factors, has been
applied to minimize the number of fermentation runs (8–10).
Only the most effective factors with high significance levels
will be selected for further optimization, while others with
lower significance levels or small effects on response value
will be omitted in further experiments (11). In addition, PBD
has been widely applied to medium optimization and formu-
lation of multicomponents among others (12, 13). The steep-
est ascent experiment, a preliminary experiment to determine
the optimum level, moves sequentially along the path of the
steepest ascent, that is, in the direction of the maximum in-
crease in response value (14). Then, the central composite de-

232
VOL. 105, 2008 RSM FOR PCA PRODUCTION BY PSEUDOMONAS SP. M-18Q 233

sign (CCD) and response surface method (RSM) (15, 16) TABLE 1. Eight medium factors and their levels used in PBD
are further adopted to determine the relationships between Low level High level
variables and responses. Moreover, the optimum of each vari- Variable
Variables −1 +1
code
able will be obtained by differential approximation. This (g/l) (g/l)
technique has been widely applied to optimizing parameters X1 Glucose 8 12
and factors for fermentation media for various microorgan- X2 Fructose 4 6
isms (17–19). The correction of results was assured and a X3 Dummy I −1 1
maximum PCA yield of 1240 mg/l was gained after medium X4 Glycerol 4 6
X5 Dummy II −1 1
optimization using Minitab 15.0 software. X6 Dummy III −1 1
X7 Mannitol 4 6
MATERIALS AND METHODS X8 Peptone 11 16.5
X9 Soybean Powder 27.5 41.25
Microorganism, medium, inoculation, and shake-flask cul- X10 Soytone 16.3 24.45
ture conditions Strain of Pseudomonas sp. M18 was isolated X11 KNO3 5 7.5
and the mutant M-18Q was constructed in our laboratory (7). They
were stored in tubes containing 60% glycerol (volumetric basis) at
−70°C. They were activated by streak plate method in King’s were glucose, fructose, glycerol, mannitol, peptone, soybean pow-
Medium B (KMB) (20) with 15 g/l agar and incubated at 28°C for der, soytone, and KNO3. These eight medium components were
24 h. Antibiotics for screening pure culture of strain M-18Q and chosen for further optimization.
wild-type strain M18 were used at the following concentrations Plackett-Burman design PBD (11) is an efficient and effec-
(µg/ml): for mutant strain M-18Q, gentamicin (Gm) 30, ampicillin tive approach to the systematic investigation and evaluation of the
(Ap) 100; for wild-type strain M18, Ap 100. All the media were effects of medium components. Each independent variable was
sterilized by autoclaving at 121°C for 20 min. The initial pH was tested at two levels, a high (+1) level and a low (−1) level. Table 1
maintained within the 7.0–7.5 range. shows the factors and their levels used in the experimental design,
Fermentation was divided into two stages: seed growth and PCA whereas Table 2 shows the detail of the design.
production. One colony from the strain M-18Q was inoculated into Path of the steepest ascent (descent) experiment PBD served
50 ml/250 ml KMB broth and incubated in a C25KC rotatory shaker as a local approximation in a small region close to the initial oper-
(New Brunswick Scientific) at 220 rpm for 14 h, while 3% (v/v) of ating conditions but far from where the process exhibited curvature.
the culture (OD ≈ 3.5) was inoculated into 100 ml/500 ml liquid me- The path of steepest ascent (descent) was adopted to determine a
dium in conical flasks incubated in a HYG-B shaker for PCA fer- suitable direction by increasing or decreasing the concentrations of
mentation at 170 rpm for 72 h. The components of the medium to variables according to the sign of the main effects to improve pro-
be optimized were as follows (g/l): glucose 8, fructose 4, mannitol 4, duction (14). For the purpose of maximizing PCA production, we
peptone 11, soybean powder 27.5, soytone 16.3, KNO3 5, and 4 g/l used the direction of the steepest ascent.
glycerol. Central composite designs and response surface method
Assay of PCA production Mixed samples of 180 µl of cul- CCD was used for investigating the region of the response surface
ture acidified to pH 4.0 with 20 µl of 6 N HCl and 540 µl of chloro- in the neighborhood of the optimum. For two factors, a fractional
form were agitated well and centrifuged at 12,000 rpm for 5 min, factorial design with five replications of center points and four star
and the lower layer was collected and subjected to high-perfor- points, which allows curvature estimation, is typically recommended
mance liquid chromatography (HPLC; Shimadzu LC-8A; Shimadzu, to have a total number of 13 runs (22). The star points for this de-
Kyoto) using C18 reverse-phase column (250 mm*4.6 mm, 5 µm) sign had a value of 1.41421, which can maintain rotatability. The
eluted with methanol-water (70 :30/V: V) at a flow rate of 1 ml/min. actual values of the variables and the design matrix are shown in
PCA was detected by UV spectroscopy at 248 nm, and the reten- Tables 3 and 4.
tion time was 1.80 min (21). Using software Minitab15.0, we obtained the response surface
Experimental design In preliminary experiments, culture model, which is a second-order model that was confirmed by sta-
conditions, various carbon and nitrogen sources, and inorganic salt tistical analysis.
were evaluated for their suitability to sustain good PCA production
by strain M-18Q. The data obtained indicated that the major vari-
ables affecting the performance of the culture in terms of PCA yield

TABLE 2. Matrix of PBD and results of evaluation of factors affecting PCA production by strain M-18Q
Run X1 X2 X3 X4 X5 X6 X7 X8 X9 X10 X11 PCA (mg/l)
1 1 −1 1 −1 −1 −1 1 1 1 −1 1 932.06
2 1 1 −1 1 −1 −1 −1 1 1 1 −1 644.90
3 −1 1 1 −1 1 −1 −1 −1 1 1 1 637.68
4 1 −1 1 1 −1 1 −1 −1 −1 1 1 828.63
5 1 1 −1 1 1 −1 1 −1 −1 −1 1 975.41
6 1 1 1 −1 1 1 −1 1 −1 −1 −1 895.81
7 −1 1 1 1 −1 1 1 −1 1 −1 −1 684.26
8 −1 −1 1 1 1 −1 1 1 −1 1 −1 536.61
9 −1 −1 −1 1 1 1 −1 1 1 −1 1 966.07
10 1 −1 −1 −1 1 1 1 −1 1 1 −1 737.20
11 −1 1 −1 −1 −1 1 1 1 −1 1 1 485.90
12 −1 −1 −1 −1 −1 −1 −1 −1 −1 −1 −1 753.24
X3, X5 and X6 were dummies; other codes were for eight factors of the medium to be optimized.
234 YUAN ET AL. J. BIOSCI. BIOENG.,

TABLE 3. Selected nutrient ranges for PCA optimization

Coded level and actual value
Variable code Variable (g/l)
−1.414 −1 0 +1 +1.414
X1 Glucose 7.5 10 16 22 24.5
X2 Soytone 5.9 8 13 18 20.07

TABLE 4. CCD matrix employed for glucose and higher production of PCA. The path of steepest ascent started
soytone independent variables from the center of the PBD and moved along the path in which
Coded level the concentration of glucose increased to 1.5 g/l, whereas
Run that of soybean powder decreased to 2.0 g/l (23). Table 6
X1 X2 PCA (mg/l)
1 0.00000 1.41421 863.95
shows the result of the path of steepest ascent experiment.
2 1.00000 1.00000 993.66 The maximum production of PCA was obtained when the
3 0.00000 0.00000 1210.00 parameters included 16 g/l glucose and 13 g/l soytone. This
4 1.00000 −1.00000 1116.40 result suggested that the point might be near the region of
5 0.00000 0.00000 1237.13 the maximum response.
6 0.00000 0.00000 1193.82 Central composite design and response surface method
7 −1.41421 0.00000 986.65
The data shown in Table 4 were explained by the following
8 0.00000 −1.41421 1083.92
9 0.00000 0.00000 1247.61 second-order polynomial equation (24, 25):
10 0.00000 0.00000 1236.19 Y = β0 + Σβi*Xi + Σβii*Xi2 + Σβij*Xi*Xj (1)
11 1.41421 0.00000 1198.88
12 −1.00000 −1.00000 1136.02 where Y stands for the predicted response; β0, βi, βii and βij
13 −1.00000 1.00000 771.82 are the offset term, linear effect, squared effect, and interac-
tion effect, respectively; Xi and Xj are the coded indepen-
dent variables or factors.
Response results shown in Table 4 were analyzed using
RESULTS AND DISCUSSION
Minitab 15.0 software. The t-test and P values were used to
PBD for screening important medium factors for PCA identify the effect of each factor on PCA production (Table 7).
production The importance of the eight medium compo- Glucose, soytone, and the interaction of the two selected
nents, namely, glucose, fructose, glycerol, mannitol, peptone, variables had a significant effect on PCA yield (p <0.05), as
soybean powder, soytone, and KNO3 for PCA production was well as the quadratic terms of glucose and soytone. The fit
investigated using PBD. Table 5 shows the effects of these of the model was checked by the coefficient of determina-
components on PCA production. The effects of glucose and tion R2, which was calculated to be 0.9688, indicating that
soytone were (+) 158.4 and (−) 222.7, respectively, and both 96.88% of the variability in the response can be explained
had confidence levels above 95%. Hence, they were con- by the model. The model can be shown as follows:
sidered as the most significant factors that affect PCA pro-
Y (mg/l) = 1224.95 + 62.79X1 − 99.75X2 − 73.31X1*X1
duction. Others had small effects and low confidence levels
− 132.73X2*X2 + 60.36X1*X2 (2)
(P > 0.05) and were considered insignificant. In the Pareto
chart (Fig. 1), the largest effects (most important factors) are where Y is the response, that is PCA concentration, and X1
presented in the upper portion and then progress down to and X2 are the coded values of glucose and soytone, respec-
the smallest effects (least important factors). In addition, it tively.
directly shows that the most important factors determining Table 8 presents Fisher’s F-test of analysis of variance
PCA production were glucose and soytone. (ANOVA), which also proves that this regression was statis-
Path of steepest ascent experiment PBD results in- tically significant (P < 0.0001) at 95% of confidence level.
dicated that the glucose effect was positive, whereas that of In order to gain a better understanding of the effects of the
soytone was negative. Thus, increasing glucose concentra- variables on the production of PCA, the predicted model
tion and decreasing soytone concentration could result in a was presented as 3D response surface graphs (Fig. 2).

TABLE 5. Analysis of PB design on optimization of culture medium in shake flask culture

Term Effect Coef SE Coef T-value Pr > |t|
Constant 756.5 21.10 35.84 0.000
X1 158.4 79.2 21.10 3.75 0.033
X2 −71.6 −35.8 21.10 −1.70 0.188
X3 32.3 16.2 21.10 0.77 0.499
X4 −62.5 −31.2 21.10 −1.48 0.235
X5 −25.8 −12.9 21.10 −0.61 0.584
X6 21.1 10.5 21.10 0.50 0.652
X7 −222.7 −111.3 21.10 −5.28 0.013
X8 95.6 47.8 21.10 2.27 0.108
R-Sq = 94.67%; R-Sq (adjust) = 80.47%.
VOL. 105, 2008 RSM FOR PCA PRODUCTION BY PSEUDOMONAS SP. M-18Q 235

TABLE 6. Design and results of path of steepest ascent experiment

Factor PCA
Run
X1 (g/l) X2 (g/l) (mg/l)
1 10 21 677.51
2 11.5 19 734.57
3 13 17 794.34
4 14.5 15 848.03
5 16 13 1235.72
6 17.5 11 1215.55
7 19 9 963.91
8 20.5 7 959.00
9 22 5 858.61
10 23.5 3 459.32

FIG. 1. Pareto chart of eight-factor standard effects on PCA pro-
duction. The important terms were soytone and glucose.
On the basis of medium optimization, the quadratic model
predicted that the largest amount of PCA was 1250 mg/l,
when the X1 code level was 0.302 and that of X2 was −0.307,
which were 17.81 g/l glucose and 11.47 g/l soytone, respec-
tively.
An experiment of 15 single colonies that incorporated the
optimized variables and other components of medium that
were kept the same as their original levels, was carried out
to verify the results. Figure 3 shows that 9 out of 15, that is,
60% of strains, exceeded 1250 mg/l in PCA production, and
the average PCA yield of single colonies reached 1240± 71
mg/l, which increased by 65.3% compared with those that
yielded 750 ± 46 mg/l in the original medium (Fig. 3). When
the optimized medium obtained from the above experiment
was applied to PCA fermentation using the wild-type strain
M18, a PCA yield of 62 mg/l was obtained, which was not
much different from 75 mg/l obtained in pigment production
media (PPM) (26). Studies suggested that qscR inactivation
changes the utilization of nutrition and energy sources in the
strain M18, and that the optimized medium might be unique
to the mutant strain M-18Q for PCA production.
The search for new and improved strains is one of the
major and common activities for increasing antibiotics pro-
duction. The mature technique of homologous recombination
was adopted to deactivate chromosomal qscR, a quorum-
sensing controlled repressor gene in Pseudomonas sp. M-18,
in order to obtain the mutant strain M-18Q (7). The prelimi-
nary results indicated that the PCA yield advanced four- to
five-fold compared with the wild-type strain M18 in PPM.
It is proved that genetic engineering is indeed an effective
means that could contribute to the increase in PCA produc-
tion.
Our results demonstrated that in addition to developing a
high-yielding strain, nutrient components also play an im-
portant role in PCA production and accumulation. When
provided a mixture of carbon energy sources, bacteria pref-
erentially utilize those they can metabolize most rapidly.
For the majority of bacteria, the preferred carbon energy
source is glucose. Studies revealed that glucose is absolute-
ly necessary for the maximum production of PCA by wild-
type strain M18 and its mutants as a carbon energy source

TABLE 7. Regression coefficients and their significance for response surface quadratic model
Term Coef StDev Coef T P
Constant 1224.95 15.78 77.640 0.000
X1 62.79 12.47 5.034 0.002
X2 −99.75 12.47 −7.997 0.000
X1*X1 −73.31 13.38 −5.481 0.001
X2*X2 −132.73 13.38 −9.923 0.000
X1*X2 60.36 17.64 3.422 0.011
S = 35.2793 PRESS = 50964.6
R-Sq = 96.88%; R-Sq (adjust) = 94.65%.

TABLE 8. ANOVA of regression model

Source DF Seq SS Adj SS Adj MS F P
Regression 5 270467 270467 54093.4 43.46 0.000
Linear 2 111151 111151 55575.5 44.65 0.000
Quadratic 2 144740 144740 72370.2 58.15 0.000
Interactions 1 14576 14576 14575.7 11.71 0.011
Residual error 7 8712 8712 1244.6
Lack of fit 3 6732 6732 2243.9 4.53 0.089
Pure error 4 1981 1981 495.2
Total 12 279179
DF, Degree of freedom; SS, sum of squares; MS, mean square.
236 YUAN ET AL.
FIG. 2. 3D surface graph and its corresponding contour plot of soy-
tone vs. glucose for PCA (mg/l) production.
FIG. 3. PCA production by Pseudomonas sp. M-18Q single colo-
nies in optimized medium and in original medium.
(27, 28). From experience, we know that the rapid metabo-
lization of glucose accelerates cell growth in a short time,
and that a certain threshold level of the strain M18 culture
population in the exponential phase is necessary for synthe-
sizing and secreting PCA as cells enter the stationary phase.
Thus, it is not surprising that glucose is a significant factor
(p < 0.05), as shown by PBD results.
Just as bacteria have preferred carbon energy sources,
they also have preferred sources of nitrogen. It was reported
that organic nitrogen is superior to inorganic nitrogen in the
fermentation process of PCA (27, 28). In a previous experi-
ment, it was found that when soytone is utilized as a nitro-
gen source, then the obtained PCA yield is almost twice that
obtained from peptone with at same amount of total nitro-
gen in PPM. It can be deduced that soytone other than pep-
tone may contain inorganic salts or growth factors that are
beneficial for obtaining a higher yield of PCA. Every coin
has two sides; it is also true for soytone in PCA fermenta-
tion. As shown by PBD and RSM results, soytone as an im-
portant factor has a negative effect, and redundant soytone
may block PCA synthesis by strain M-18Q. On the basis of
data analysis, we appropriately decreased the amount of soy-
tone and finally obtained a satisfactory result.
In addition, the interaction between the above two fer-
mentation variables also cannot be neglected in PCA pro-
duction. The results of this study indicated that the suitable
ratio and proper balance between carbon and nitrogen sources
is important for obtaining a higher yield.
J. BIOSCI. BIOENG.,
PCA, the key component of microorganism-based anti-
microbial Shenqinmycin, can effectively prevent and cure
pimiento epidemic disease, the wilt of watermelon sprout,
and melon vine withering (29). It will definitely bring many
economic and environmental benefits.
This study proved that statistical experimental designs
provide an efficient and economical measure for optimizing
information on the effects of each factor. We also found that
PBD enable the screening for the most important factors for
PCA yield, and that the path of steepest ascent experiment
provided a local approximation for easily carrying out the
subsequent experiments. Furthermore, we also found that
CCD as well as RSM provided the optimum levels of two
variables. The results were confirmed using Minitab 15.0
software. A maximum PCA production of 1240 mg/l was
achieved with the following optimized factors in this fer-
mentation process: 17.81 g/l glucose and 11.47 g/l soytone.
ACKNOWLEDGMENTS
This study was supported by the 863 Programs of the People’s
Republic of China (no. 2006AA10A209) and the Shanghai Leading
Academic Discipline Project (no. B203).
REFERENCES
1. Thomashow, L. S. and Weller, D. M.: Current concepts in
the use of introduced bacteria for biological disease control:
mechanisms and antifungal metabolites. Plant–Microbe Interact.,
1, 187–236 (1996).
2. Brisbane, P. G., Harris, J. R., and Moen, R.: Inhibition of
fungi from wheat roots by Pseudomonas fluorescens 2–79
and fungicides. Soil Biol. Biochem., 21, 1019–1025 (1989).
3. Thomashow, L. S. and Weller, D. M.: Role of a phenazine
antibiotic from Pseudomonas fluorescens in biological con-
trol of Gaeumannomyces graminis var. tritici. J. Bacteriol.,
170, 3499–3508 (1988).
4. Xu, Y. Q., Tang, W. N., and Zheng, Y. L.: Screening of
Pseudomonas M18 and its biological control of cucumber wilt
disease under commercial greenhouse. J. Shanghai Jiaotong
Univ., 33, 210–213 (1999).
5. Chugani, S. A., Whitely, M., Lee, K. M., D’Argenio, D.,
Manoil, C., and Greenberg, E. P.: QscR, a modulator of
quorum-sensing signal synthesis and virulence in Pseudomonas
aeruginosa. Proc. Natl. Acad. Sci. USA, 98, 2752–2757
(2001).
6. Ledgham, F., Ventre, I., Soscia, C., Foglino, M., Sturgis,
J. N., and Lazdunski, A.: Interactions of the quorum sensing
regulator QscR: interaction with itself and the other regulators
of Pseudomonas aeruginosa LasR and RhlR. Mol. Microbiol.,
48, 199–210 (2003).
7. Wang, Y., Yan, A., Huang, X. Q., Zhang, X. H., and Xu,
Y. Q.: Construction of Pseudomonas sp. M18 qscR–Mutant
and its regulation on biosynthesis of PCA and Plt. Acta
Microbiol. Sin., 47, 254–259 (2007).
8. Castro, P. M. L., Hayter, P. M., Ison, A. P., and Bull, A. T.:
Application of statistical design to the optimization of culture
medium for recombinant interferon-gamma production by
Chinese hamster ovary cells. Appl. Microbiol. Biotechnol.,
38, 84–90 (1992).
9. Escamilla, E. M., Dendooven, L., Magaña, I. P., Parra, S.,
and De la Torre, M.: Optimization of gibberellic acid pro-
duction by immobilized Gibberellafujikuroi mycelium in flu-
idized bioreactors. J. Biotechnol., 76, 147–155 (2000).
10. Xu, C. P., Kim, S. W., Hwang, H. J., Choi, J. W., and Yun,
VOL. 105, 2008
J. W.: Optimization of submerged culture conditions for my-
celial growth and exo-biopolymer production by Paecilomyces
tenuipes C240. Process Biochem., 38, 1025–1030 (2003).
11. Plackett, R. L. and Burman, J. P.: The design of optimum
multifactor experiments. Biometrika, 33, 305–325 (1946).
12. Naveena, B. J., Altaf, Md., and Bhadriah, K.: Selection of
medium components by Plackett-Burman design for produc-
tion of L(+) lactic acid by Lactobacillus amylophilus GV6 in
SSF using wheat bran. Bioresour. Technol., 96, 485–490
(2005).
13. Loukas, Y. L.: A Plackett-Burnam screening design directs
the efficient formulation of multicomponent DRV liposomes.
J. Pharm. Biomed. Anal., 26, 255–263 (2001).
14. Montgomeryd, D. C.: Design and analysis of experiments,
3rd ed. John Wiley and Sons, New York (1991).
15. Mead, R. and Pike, D. J.: A review of response surface
methodology from a biometrics viewpoint. J. Biometrics, 31,
803–851 (1975).
16. Mathews, R. J., Scott, R. G., and Morgan, S. L.: Character-
ization of an enzymatic determination of arsenic (IV) based on
response surface methodology. Anal. Chim. Acta, 133, 169–
182 (1981).
17. Carla, J. and Roberto, I. C.: Optimization of xylitol produc-
tion by Candida guilliermondii FTI 20037 using response sur-
face methodology. Process Biochem., 36, 1119–1124 (2001).
18. Burkert, J. F. M., Maugeri, F., and Rodrigues, M. I.: Opti-
mization of extracellular lipase production by Geotrichum sp.
using factorial design. Bioresour. Technol., 91, 74–89 (2004).
19. Liu, C. H. and Wu, P. S.: Optimization of adenoviral produc-
tion in human embryonic kidney cells using response surface
methodology. J. Biosci. Bioeng., 103, 406–411 (2007).
20. King, E. O., Ward, M. K., and Raney, D. E.: Two simple
RSM FOR PCA PRODUCTION BY PSEUDOMONAS SP. M-18Q 237
media for the demonstration of pyocyanin and fluorescein. J.
Lab. Clin. Med., 44, 301–307 (1954).
21. Ge, Y. H., Huang, X. Q., Wang, S. L., Zhang, X. H., and
Xu, Y. Q.: Phenazine-1-carboxylic acid is negatively regulated
and pyoluteorin positively regulated by gacA in Pseudomonas
sp. M18. FEMS Microbiol. Lett., 237, 41–47 (2004).
22. Kennedy, M. and Krouse, D.: Strategies for improving fer-
mentation medium performance. J. Ind. Microbiol. Biotechnol.,
23, 456–475 (1999).
23. Chang, Y. N., Huang, J. C., and Lee, C. C., Shin, I. L., and
Tzeng, Y. M.: Use of response surface methodology to opti-
mize culture medium for production of lovastatin by Monascus.
ruber. Enzyme Microb. Technol., 30, 889–894 (2002).
24. Vohra, A. and Satyanarayana, T.: Statistical optimization
of the medium components by response surface methodology
to enhance phytase production by Pichia anomala. Process
Biochem., 37, 999–1004 (2002).
25. Myers, R. H. and Montgomery, D. C.: Response surface
methodology, 2nd ed. Wiley, New York (2002).
26. Levitch, M. E. and Stadtman, E. R.: A study of the biosyn-
thesis of phenazine-1-carboxylic acid. Arch. Biochem. Biophys.,
106, 194–199 (1964).
27. Geng, H. F., Cheng, F., and Xu, Y. Q.: Mutation of Pseu-
domonas fluorescen spp. M18 producing phenazine carboxy-
lic acid and its fermentation conditions. Ind. Microbiol., 33,
1–5 (2003).
28. Luo, P., Xu, Y. Q., Zhang, X., and Chen, F.: Study on fer-
mentation condition for PCA production by Pseudomonas
fluorescens. M18. Acta Agric. Shanghai, 18, 66–71 (2002).
29. Xu, Y. Q.: Green microorganism based antimicrobial: shen-
qinmycin (M18). Fine Specialty Chem., 20, 8–9 (2004).