Bioresource Technology 101 (2010) 1904–1912

Contents lists available at ScienceDirect

Bioresource Technology

journal homepage: www.elsevier .com/locate /bior tech

Use of response surface methodology to evaluate the effect of metal ions(Ca2+, Ni2+, Mn2+, Cu2+) on production of antifungal compoundsby Paenibacillus polymyxa

Waseem Raza, Wu Hongsheng, Shen Qirong *

Jiangsu Provincial Key Lab for Organic Slid Waste Utilization, Nanjing Agricultural University, No. 1, Weigang Road, Nanjing, 210095 Jiangsu Province, China

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

Article history:Received 9 June 2009Received in revised form 5 October 2009Accepted 12 October 2009Available online 6 November 2009

Keywords:Antifungal compoundsMetal ionsPaenibacillus polymyxaQuadratic model

0960-8524/$ - see front matter � 2009 Elsevier Ltd. Adoi:10.1016/j.biortech.2009.10.029

* Corresponding author. Tel.: +86 13901586468; faE-mail address: shenqirong@njau.edu.cn (S. Qiron

The effects of four metal ions (Ca2+, Ni2+, Mn2+ and Cu2+) were evaluated on growth and production ofantifungal compounds by Paenibacillus polymyxa SQR-21 and a quadratic predictive model was developedusing response surface methodology (RSM). The results revealed, Mn2+ and Ni2+ showed most positivesynergistic interactive affect on production of antifungal compounds followed by the positive interactivesynergistic affect of Cu2+ and Ni2+ and then Mn2+ and Cu2+. While the interactive effect of Ca2+ with allother three metals inhibited the production of antifungal compounds. The Mn2+ (P = 0.0384), Ni2+

(P = 0.0004) and Cu2+ (P = 0.0117) significantly affected the production of antifungal compounds whilethe effect of Ca2+ (P = 0.1851) was less significant. The maximum growth (OD600 = 1.55) was obtainedat 500 (0), 125 (0), 100 (�2) and 37.5 (0) lM levels and the maximum size of inhibition zone (31 mm)was measured at 400 (�1), 150 (1), 400 (1) and 25 lM (�1) levels of Ca2+, Mn2+, Ni2+ and Cu2+, respec-tively. The RSM model provided an easy and effective way to determine the interactive effect of metalions on production of antifungal compounds by P. polymyxa SQR-21 so that optimum media recipescan be developed to produce maximum amounts of antifungal compounds under laboratory and com-mercial fermentation conditions.

� 2009 Elsevier Ltd. All rights reserved.

1. Introduction

The culture medium for antibiotic-producing microorganismsusually contains varying amounts of trace elements necessary forgrowth and antibiotic production. However, amounts and avail-ability of trace elements in natural and complex substrate materi-als vary with the particular source. The addition of heavy metals toa liquid culture can significantly affect growth and antibiotic pro-duction by microorganisms as reported by Pavl and Banerjee(1983) who observed a promotive effect of Cu2+, Mn2+, Zn2+ andFe2+ on growth and antibiotic yield of Streptomyces galbus. This reg-ulatory effect of trace metals or medium mineral content on micro-bial secondary metabolism has been recorded for a variety ofspecies (Weinberg, 1977), but little work has been reported forPaenibacillus species. Among Paenibacillus spp., Paenibacillus poly-myxa (P. polymyxa) strains are capable of suppressing plant dis-eases and promoting plant growth (Ryu et al., 2006). P. polymyxastrains are known to produce two types of peptide antibiotics.One group comprises the antibiotics active against bacteria, includ-ing the polymyxins (Shoji et al., 1977). The other group is made upof the peptide antibiotics active against fungi and gram-positive

ll rights reserved.

x: +86 2584432420.g).

bacteria, includes fusaricidins A, B, C, and D (Beatty and Jensen,2002). In addition, there are many reports where the nature ofthe inhibitory agent is undefined (Dijksterhuis et al., 1999). In thisexperiment, we used P. polymyxa strain SQR-21 that was tested forall possible antifungal compounds and it was found that it pro-duced only fusaricidin type antifungal compounds, composed ofa group of cyclic depsipeptides that have molecular masses of883, 897, 948, and 961 Da, with an unusual 15-guanidino-3-hydroxypentadecanoic acid moiety bound to a free amino group(Raza et al., 2009). The fusaricidin biosynthetic gene (fusA) clusterwas cloned and sequenced. It spans 32.4 kb, including an openreading frame, encoding a six-module nonribosomal peptide syn-thetase (Li and Jensen, 2008). In this study, we used four metal ions(Ca2+, Ni2+, Mn2+, Cu2+) to evaluate their effect on the production offusaricidin type antifungal compounds. The effect of these metalions has been widely reported. For example, The addition of0.03% copper sulfate, increased the production of the antibioticshydroheptin and chartreusin by Streptomyces chartreusis IMRU3962 (Tunac and McDaniel, 1985). Nickel inhibited growth, pig-ment and antibiotic production by Streptomyces galbus (Raytapadaret al., 1995). Mn2+ had a marked effect on growth and bulbiforminproduction by Bacillus subtilis (Mahmood, 1970). Calcium has beenidentified as an inhibitory substance for the biosynthesis of efroto-mycin in resting cells of Nocradia lactamdurans (Cover et al., 1991).

W. Raza et al. / Bioresource Technology 101 (2010) 1904–1912 1905

Useful information can be obtained by studying the effect ofindividual metal ions on metabolic products of microorganisms,but in natural environments, metals are present together in variousconcentrations. Therefore, there is a need to assess their effects atdifferent concentrations to find out their interactive behavior andoptimum recipe concentrations to obtain maximum metabolicproducts. In addition, fusaricidin seems to have great potentialfor industrial use according to the recent reports on its germicidalactivity against pathogenic gram-positive bacteria and plant path-ogenic fungi and thereby fusaricidin is in increasing demand (Razaet al., 2009; Beatty and Jensen, 2002). For this purpose, it is essen-tial to evaluate the interactive effects of four metal ions (Ca2+, Ni2+,Mn2+, Cu2+) on the production of fusaricidin type antifungal com-pounds. The traditional ‘‘four-factor; three-level” technique usedfor optimizing a multi-variable system with all potential combina-tions, not only require many experiments but also may result inwrong conclusions. Under these circumstances, RSM is an attrac-tive alternative that can be used to study the effects of several fac-tors influencing the dependent responses by varying the factorssimultaneously and carrying out a limited number of experiments(Arteaga et al., 1994). RSM is a collection of statistical techniquesfor designing experiments, building models and searching opti-mum conditions of factors for desirable responses and this canbe achieved easily by fitting experimental data to an easily avail-able statistical software like Design-Expert (Version 7.1.6; Stat-Ease, Inc., Minneapolis, MN, USA) (Montgomery, 1991). RSM is cur-rently the most popular optimization technique because of itscomprehensive theory, reasonably high efficiency and simplicityand has been successfully applied in many areas of biotechnologysuch as optimization of a cultural medium (Wang and Liu, 2008)and aqueous two-phase extraction of bacteriocin (Li et al., 2001).However, it has not previously been used to study and evaluatethe effect of metal ions on the production of fusaricidin type anti-fungal compounds. Therefore, an experiment was planned toinvestigate the effect of four metal ions on growth and productionof antifungal compounds by P. polymyxa SQR-21 and to develop aquadratic predictive model. In addition, the effects of metal ionson overall metabolic activity of P. polymyxa were also determinedby measuring intra and extracellular proteins and carbohydratesand intracellular lipids and total RNA contents.

2. Methods

2.1. Bacterial and fungal cultures

A chitinase-deficient and fusaricidin-producing strain, P. poly-myxa SQR-21, and a pathogenic strain, Fusarium oxysporum f. sp.cucumerinum (F. oxysporum), were provided by the Soil–MicrobeInteraction Laboratory, Nanjing Agriculture University, Nanjing,China. The P. polymyxa SQR-21 culture was maintained on LaurieBertani (LB) medium plates and was stored at �80 �C in trypticsoya broth (TSB) containing 20% glycerol for further use. The fungalpathogen, F. oxysporum, was maintained by cultivation on PDAplates for 3 days at 28 �C and then the plates were sealed withparafilm and stored at 4 �C. The pathogen was subcultured onto afresh PDA plate after one month.

Table 1Factors and levels of four variables chosen for the trail.

Factors Symbols Levels

Coded �2 �1 0 +1 +2

Ca2+ (lM) x1 300 400 500 600 700Mn2+ (lM) x2 75 100 125 150 225Ni2+ (lM) x3 100 200 300 400 500Cu2+ (lM) x4 12.5 25 37.5 50 62.5

2.2. Isolation, purification and molecular weight determination ofantifungal compounds

For the isolation of antifungal compounds, tryptone medium(tryptone, 10; NaCl, 5 and sucrose, 10 g L�1; pH 7.5) was inoculatedwith 1 ml of overnight culture of SQR-21 in tryptone broth andincubated in an incubator shaker (170 rpm, 30 �C). After 4 days, li-quid culture was centrifuged at 12000g for 10 min and superna-

tants were pooled. Active compounds were extracted two timeswith an equal volume of n-butanol. The extract was concentratedby using a rotary evaporator and the residues were dissolved inmethanol. For the purification of antifungal compounds, activecompounds were eluted with chloroform and methanol (50:50)from a silica gel column (2 cm � 24 cm). Twenty six fractions(10 ml each) were collected, dried in a rotary evaporator and dis-solved in 1 ml methanol. Active fractions, identified by agar diffu-sion assay against F. oxysporum on PDA, were pooled. These activefractions were further purified on a Sephadex LH-20 column(2 cm � 24 cm) which was equilibrated and eluted with methanol.Active fractions located by bioassay were pooled. Antifungal sub-stances from Sephadex LH-20 were purified further with reversephase high performance liquid chromatography (EttanTM LC, GEHealthcare life sciences) by using a Grace Vydac C18 238TP52(5 lm) column and eluted with 5–95% of gradient B (0.05% formicacid in acetonitrile) in 80 min, with a flow rate of 0.2 ml min�1.While gradient A was 0.05% formic acid. The column fractions weremonitored for absorbance at 214 nm. After agar diffusion assayagainst F. oxysporum on PDA, two active peaks were located andantifungal material was subjected to liquid chromatography–massspectroscopic analyses (LC–MS) (Agilent 1100 HPLC-1946A MSP;Agilent Technologies, Santa Clara, USA) for molecular weight deter-mination. The MS analysis was done by electrospray ionization inpositive ion mode.

2.3. Preparation of the metal ion media and assay for the antifungalactivity

Liquid-culture experiments were performed in 100 ml of tryp-tone broth in 500 ml Erlenmeyer flasks. Initial Ca2+, Ni2+, Mn2+

and Cu2+ contents in tryptone broth determined by Spectra AA,220 FS atomic absorbance spectrometer, were 110, 0.7, 3.4 and0.6 lM, respectively. For the estimation of optimal concentrationsof each metal ion for maximum production of antifungal com-pounds, one factor (one metal ion) with three concentration levelexperiments were conducted. The cultures were supplementedwith CaSO4, MnSO4, NiCl2 and CuSO4, for Ca2+, Ni2+, Mn2+ andCu2+, respectively. The levels considered for Ca2+ were 200, 400and 600, for Mn2+ 50, 100 and 150 lM, for Ni2+ these levels were100, 200 and 400 lM and for Cu2+ these levels were 25, 50 and100 lM. Each experiment had three replicates including controlcultures without supplemented metal ions. The isolation of anti-fungal compounds was done as described in Section 2.2. The con-centrated n-butanol extracts were dissolved in 500 ll methanoland used to determine antifungal activity by agar diffusion assayusing F. oxysporum as an indicator pathogen. After three days, theinhibition zone was measured. The metal concentrations thatshowed maximum production of antifungal compounds were usedas the highest level (+1 level) for interactive study.

For the interactive study of metal ions, four-factor (Ca2+, Ni2+,Mn2+, Cu2+) with five concentration level experiments were con-ducted and, after sterilization, the cultures were supplementedwith CaSO4, MnSO4, NiCl2 and CuSO4 for Ca2+, Ni2+, Mn2+ andCu2+, respectively, according to the proportion of metal ions men-tioned in Table 1. Each experiment had three replicates.

1906 W. Raza et al. / Bioresource Technology 101 (2010) 1904–1912

2.4. Cellular protein, carbohydrate and lipid contents

The P. polymyxa SQR-21 liquid culture samples (2 ml) were cen-trifuged (12,000g) for 10 min. The pellets were suspended in 2 mlof deionized water for washing and centrifuged three times. Thesepellets were used for the determination of total intracellular pro-tein (IP), carbohydrate (IC) and lipid contents (IL). For measuringtotal protein contents, the washed cells were resuspended indeionized water and incubated in 1 N NaOH at 90 �C for 10 minto solubilize cellular protein. Proteins were measured by the meth-od of Bradford (1976) with bovine serum albumin standards rang-ing from 10–100 lg ml�1. Total carbohydrate was estimated inrinsed-cell samples by the phenol–sulfuric acid method (Duboiset al., 1956). The lipid contents of bacterial cells were measuredusing the phosphoric acid–vanillin reagent method of Izard andLimberger (2003) with Triolein standards ranging from 10 to100 lg. The cell free liquid culture was used for the estimation ofextracellular protein (EP) and polysaccharide (EPS) contents bythe above-described methods. Before assaying protein, the result-ing EPS solution was dialyzed using a membrane of 1000 Damolecular weigh cut-off against ultra pure water for 2 days at4 �C to remove the small molecules and entrained media residues.The lyophilized cell pellets were used to measure cellular dryweight.

2.5. RNA extraction and primers design

Total RNA was isolated by using the Trizol reagent (InvitrogenTM,Shanghai) according to manufacturer’s instructions. To removeDNA contamination, 10 U DNaseI (Takara, Dalian) along with 20U RNase inhibitor (Takara, Dalian) (37 �C, 40 min) were used inthe reaction mixture of 50 ll containing 20–50 lg RNA. RNA wasestimated by determining the absorbance at 260 nm. Specificprimers for fusaricidin synthetase gene (fusA) (111 bp) and 16SRNA gene (16s) (210 bp) were designed by using primer premier5 software (PREMIER biosoft international). The designed primerswere as follows: fusA1: 50 GCAGAGGATGATAGTGTTGGTC 3’, fasA2:5’ CAGCACATCATGCGTTCC 3’, 16s1: 5’ CATTCATCGTTTACGGCGT 3’and 16s2: 5’ TGTTAATCCCGAGGCTCACT 3’.

2.6. Reverse transcription and real time PCR assay

For the synthesis of first stand cDNA, 3 lg of RNA, 200 U ofRevertAidTM M-MuLV reverse transcriptase (Fermantas), 20 U RNaseinhibitor (TaKaRa, Dalian), 0.2 lg of Random hexamer primer and1 mM dNTP in the total volume of 20 ll were used. RT was per-formed using the following parameters: denaturation for 5 minat 65 �C, incubation for 60 min at 42 �C and inactivation for 5 minat 95 �C. Target genes from cDNA were amplified separately using3 ll aliquots of RT product as template and 30 pmol of each primerpair (fusA1 and fusA2, 16s1 and 16s2). Reaction mixtures for PCRcontained 2.5 U Taq polymerase, 20 nmol of dNTP and100 nmol Mg2+. The PCR protocol included a 5 min incubation at95 �C, followed by 30 cycles, each including 30 s at 94 �C, 30s at58 �C, 1 min at 72 �C and then 2 min at 72 �C. Amplified productswere separated, stained and viewed to check band intensity andcDNA quality. Singleplex relative Real Time PCR was performedusing an iCycler MyiQTM single color Real Time PCR detection sys-tem (BioRad). Reactions mixtures (20 ll) contained 1 mM primers,3 ll cDNA and 10 ll of SYBR�Premix Ex TaqTM (Perfect Real Time)(TaKaRa, Dalian) including TaKaRa Ex TaqTM HS and SYBR� GreenI, dNTP and buffer. The PCR protocol included a 10 min denatur-ation at 95 �C followed by 40 cycles with 95 �C for 30 s, 58 �C for30 s and 72 �C for 1 min. Detection of the fluorescent productwas carried out at the end of the 72 �C extension period (2 min).After PCR, these samples were heated from 58 to 95 �C and fluores-

cence was monitored to determine the Tm. Two microliter of cDNAfrom each treatment were mixed to prepare relative standards. Theexperiment was repeated twice.

2.7. Methodology and design of experiments

The RSM, an empirical modeling technique used to estimate therelationship between a set of controllable experimental factors andobserved results, is currently one of the most popular optimizationtechniques in the field of biotechnology and the most commonexperimental design used in RSM is the Central Composite Design(CCD) (Wang and Liu, 2008). The CCD is optimized design for fittingquadratic model that has equal predictability in all directions fromthe center. The numbers of experimental points in the CCD are suf-ficient to test statistical validity of the fitted model and lack-of-fitof the model (Arteaga et al., 1994). In addition, the central point inCCD is replicated several times to estimate the error due to exper-imental or random variability (Liu and Tzeng, 1998). In this exper-iment, a CCD with four variables (Ca2+, Ni2+, Mn2+, Cu2+) at fiveconcentration levels was performed that allowed minimal numberof experimental runs to obtain the quadratic predictive model(Montgomery, 1991). The CCD was applied using Design-Expertsoftware. Thirty experiments were augmented with six replica-tions at the center points to evaluate the pure error. The fourparameters, Ca2+, Ni2+, Mn2+ and Cu2+ concentration, were chosenas main variables and designated as x1, x2, x3, and x4, respectively.The low, middle and high levels, of each variable were designatedas �2, �1, 0, +1 and + 2, respectively (Table 1).

The behavior of the system was explained by the following sec-ond-degree polynomial equation:

Y ¼ B0 þXn

i¼1

Bixi þXn

i<j

Bijxixj þXn

j¼1

Bjjx2j ð1Þ

where Y is the response; b0 is a constant, bi is the linear, bij is thesecond-order interaction, bjj is the quadratic coefficients and xi isthe non-coded in-dependent variables. It can be observed that inthe present study, four variables are involved and hence n takesthe value 4. Thus, by substituting the value 4 for n, Eq. (1) becomes:

Y ¼ B0 þ B1x1 þ B2x2 þ B3x3 þ B4x4 þ B12x1x2 þ B13x1x3

þ B14x1x4 þ B23x2x3 þ B24x2x4 þ B34x3x4 þ B11x21 þ B22x22

þ B33x23 þ B44x24 ð2Þ

where Y is the predicted response, x1, x2, x3 and x4 are the input vari-ables; B0 is a constant; B1, B2, B3 and B4 are the linear coefficients;B12, B13, B14, B23, B24 and B34 are the cross product coefficients;B11, B22, B33 and B44 are the quadratic coefficients.

3. Results and discussion

After the purification of antifungal material by RP-HPLC, two ac-tive peaks were located by agar diffusion assay against F. oxyspo-rum. LC-MS analysis showed that each peak was comprised oftwo inseparable compounds; the first had molecular weights of947.5 and 961.5 Da (M + H)+ while the second had molecularweights of 883.5 and 897.5 Da (M + H)+ (Fig. 1). Different chro-matographic conditions were used but further separation of anti-fungal compounds was not achieved. Considering the literatureand our results, the first peak compounds having molecularweights of 947.5 and 961.5 Da (M + H)+ were designated as fusaric-idin C and fusaricidin D, respectively while second peak com-pounds having molecular weights of 883.5 and 897.5 Da (M + H)+

were designated as fusaricidin A and fusaricidin B, respectively(Beatty and Jensen, 2002; Raza et al., 2009). These four fusaricidinsare not available commercially so we estimated the production of

Fig. 1. RP-HPLC peak profile of antifungal material from Paenibacillus polymyxaSQR-21. Antifungal activity was found in peaks eluted at 38.4 and 39.9 min, asindicated by the arrows (A) and mass spectroscopic analysis of antifungalcompounds, showing molecular weights (947.5 and 961.5 Da) and (883.5 and897.5 Da) of two peaks that showed antifungal activity (B).

Fig. 2. RT-PCR products originating from cDNA. Total RNA was extracted fromP. polymyxa SQR-21 grown in submerged culture and treated with four concentra-tions of Ca2+ (a = 0, b = 200, c = 400, d = 600 lM Ca2+), Mn2+ (a = 0, b = 50, c = 100 andd = 150 lM), Ni2+ (a = 0, b = 100, c = 200 and d = 400 lM) and Cu2+ (a = 0, b = 25,c = 50 and d = 100 lM). M = DNA molecular marker; A–D = positive controls for 16sRNA gene;+a to + d = fusA (+RT); �a to �d = negative controls for fusA (�RT).

W. Raza et al. / Bioresource Technology 101 (2010) 1904–1912 1907

antifungal compounds (fusaricidins) by measuring the diametersof inhibition zones.

For the estimation of optimal concentration of each metal ionfor maximum production of antifungal compounds, the influenceof one factor (one metal ion) was determined with three concen-tration level experiments. The results indicated that an increasein Ca2+, Ni2+ and Mn2+ concentrations in the liquid medium signif-icantly increased the production of antifungal compounds and therelative expression of fusA. In contrast, in case of Cu2+, maximumantifungal activity and relative expression of fusA occurred at50 lM Cu2+ while at the highest level of Cu2+ (100 lM), decreasein the production of antifungal compounds and in the relativeexpression of fusA was determined. The increase in intensity ofthe bands corresponding to fusA mRNA suggests an increase ingene expression with increasing concentrations of metal ions(Fig. 2). Growth, as measured by optical density (OD600), increasedas the concentration of Ca2+ decreased, increased as the Ni2+ con-centration increased, but was not affected significantly by Mn2+

and Cu2+ (Table 2).Analysis of intra and extracellular protein and carbohydrate and

intracellular lipid contents indicated that an increase in Ca2+ con-centration in the liquid medium decreased the IP and EP content,but increased the IC, IL, EPS and RNA content. The increase inNi2+ concentration decreased the IL content and increased the ICand EPS content, while IP, EP and RNA levels were increased up

to 100 lM Ni2+ concentration. At higher levels of Ni2+ (200 and400 lM), decrease in IP, EP and RNA levels was determined. An in-crease in the Mn2+ concentration in the liquid medium increasedthe IL, EP, EPS and RNA levels while the effect on IP and IC contentwas less significant. An increase in Cu2+ concentration increased IPand EPS levels, decreased the IC, EP and RNA levels, but had no sig-nificant effect on IL levels.

The values of responses (size of inhibition zone and OD600) ob-tained under different experimental conditions of combined appli-cation of metal ions are summarized in Table 3. Based onparameter estimates, RSM provides an empirical relationship be-tween the response variables (size of inhibition zone and OD600)and the test variables under consideration. The values of theregression coefficients were calculated by applying multipleregression analysis to the inhibition zone data, and an equationof the second order polynomial could be written as:

YInhibition zone ¼ 21:54� 0:42x1 þ 0:68x2 þ 1:35x3 þ 0:86x4

� 0:94x1x2 � 0:64x1x3 � 0:87x1x4 þ 0:54x2x3

� 1:10x2x4 � 0:90x3x4 þ 0:31x21 þ 0:48x2

2

þ 0:35x23 þ 1:43x2

4 ð3Þ

The values of the regression coefficients were calculated byapplying multiple regression analysis on the experimental data ofOD600 and an equation of the second order polynomial could bewritten as:

YGrowth ¼ 1:48þ 0:00361x1 � 0:0038x2 � 0:033x3 � 0:00392x4

� 0:020x1x2 þ 0:0038x1x3 � 0:047x1x4 � 0:0032x2x3

� 0:013x2x4 � 0:0036x3x4 � 0:013x21 � 0:018x2

2

� 0:0035x23 � 0:013x2

4 ð4Þ

The adjusted determination coefficient R2Adj is the correlation

measure for testing the goodness of fit of the regression equation.The value of R2

Adj (0.74) for Eq. (3) and R2Adj (0.71) for Eq. (4) indi-

cates a high degree of correlation between observed and predicted

Table 2Effect of the different Ca2+, Ni2+, Mn2+ and Cu2+ concentrations on OD600. Inhibition zone (IZ), intracellular protein (IP), carbohydrate (IC) and lipid contents (IL), extracellularprotein (EP) and polysaccharide contents (EPS), total RNA and relative expression (RE) of fusA of P. polymyxa SQR-21 against Fusarium oxysporum after four days incubation. Valueshaving same letters are less significant with each other at P = 0.05.

Ion (lM) OD600 IZ (mm) IP (lg/mg) IC (lg/mg) IL (lg/mg) EP (lg/ml) EPS (lg/ml) RNA (lg/mg) RE

Ca2+

0 0.77 a 13.5 d 261 a 91.9 b 69.5 b 467 a 709 c 15.7 c 0.32 c200 0.67 b 16.5 c 192 b 114.7 a 98.1 a 312 b 858 b 26.1 b 0.59 b400 0.65 b 17.8 b 178 bc 112.5 a 96.9 a 317 b 878 a 28.9 a 0.63 b600 0.60 c 19.2 a 149 c 111.6 a 105.8 a 321 b 853 b 27.2 ab 0.78 aLSD 0.04 0.85 33.23 16.21 9.53 21.4 10.49 2.21 0.06

Ni2+

0 0.74 c 14.7 c 404 b 64.1 c 80.5 a 353 b 709 b 21.0 b 0.44 c100 0.95 b 18.5 b 519 a 78.2 b 69.6 ab 562 a 751 b 28.7 a 0.84 b200 0.98 ab 22.6 a 421 b 86.2 a 57.1 bc 367 b 863 ab 20.5 b 1.19 a400 1.08 a 23.9 a 317 c 75.0 b 54.0 c 314b 1110 a 17. 7 c 0.98 abLSD 0.10 2.47 16.73 2.74 10.68 69.12 172.7 2.36 0.26

Mn2+

0 1.087 18.1 d 371.6 92.8 63.5 b 436 c 1248 c 21.5 b 0.47 c50 1.080 20.1 c 382.1 85.6 67.7 ab 521 a 1530 a 24.8 b 0.49 c100 1.034 22.2 b 391.6 96.3 71.1 a 517 a 1473 a 30.6 a 0.73 b150 1.025 23.7 a 399.7 98.1 72.2 a 467 b 1369 b 33.5 a 0.91 aLSD NS 0.90 NS NS 4.70 11.79 84.24 2.97 0.18

Cu2+

0 1.156 14.4 b 321 c 90.2 a 111.2 324 a 896 b 16.3 a 0.57 c25 1.165 15.8 b 337 bc 74.7 b 109.6 269 b 1562 b 14.8 a 0.91 b50 1.170 18.0 a 370 a 70.5 b 112.2 298 ab 1909 a 12.8 b 1.36 a100 1.176 15.7 b 358 ab 60.7 c 112.3 305 ab 1385 c 12.5 b 0.71 cLSD NS 1.27 7.33 4.76 NS 33.99 90.76 1.51 0.13

Table 3Central composite design arrangements and responses.

Trial no. Variables Y for inhibition zone Y for OD600

x1 x2 x3 x4 Observed Predicted Observed Predicted

1 �1 �1 �1 �1 16.67 17.47 1.40 1.392 1 �1 �1 �1 21.50 21.97 1.50 1.523 �1 1 �1 �1 20.67 22.25 1.43 1.454 1 1 �1 �1 23.00 22.50 1.48 1.505 �1 �1 1 �1 21.00 21.25 1.29 1.336 1 �1 1 �1 24.34 23.67 1.54 1.497 �1 1 1 �1 31.00 28.70 1.37 1.378 1 1 1 �1 24.50 26.86 1.44 1.469 �1 �1 �1 1 26.67 25.36 1.48 1.5110 1 �1 �1 1 24.00 25.86 1.48 1.4511 �1 1 �1 1 25.00 25.23 1.49 1.5112 1 1 �1 1 22.67 21.47 1.37 1.3713 �1 �1 1 1 26.00 26.06 1.46 1.4214 1 �1 1 1 25.00 24.47 1.37 1.3915 �1 1 1 1 28.00 28.58 1.40 1.4116 1 1 1 1 24.00 22.75 1.33 1.3117 �2 0 0 0 24.00 23.36 1.44 1.4118 2 0 0 0 22.00 22.03 1.40 1.4419 0 �2 0 0 23.00 21.84 1.41 1.4220 0 2 0 0 24.34 24.89 1.42 1.3921 0 0 �2 0 21.00 20.34 1.55 1.5222 0 0 2 0 25.34 25.39 1.38 1.3923 0 0 0 �2 27.00 25.31 1.47 1.4524 0 0 0 2 28.00 29.08 1.40 1.4125 0 0 0 0 21.67 21.34 1.47 1.4726 0 0 0 0 20.67 21.34 1.50 1.4727 0 0 0 0 23.25 21.34 1.47 1.4728 0 0 0 0 21.50 21.34 1.50 1.4729 0 0 0 0 21.40 21.34 1.45 1.4730 0 0 0 0 20.77 21.34 1.49 1.47

1908 W. Raza et al. / Bioresource Technology 101 (2010) 1904–1912

values of the size of inhibition zone and growth (OD600), respec-tively. Wang and Liu (2008) also showed a high degree of correla-tion between observed and predicted values. Analysis of variance(ANOVA) is a statistical testing model, which is required to testthe significance and adequacy of the model. Here the ANOVA of

the regression model demonstrates that the model is highly signif-icant, as is evident from the calculated F-value (6.80 for the size ofinhibition zone and 6.05 for OD600) and very low probability value(P = 0.0003 for inhibition zone and 0.0007 for OD600). The modelalso shows that lack of fit is not significant statistically for size of

Table 4Analysis of variance for predictive equation for the size of inhibition zone (antifungal compound production by P. polymyxa SQR-21).

Source Sum of squares df Mean square F-value Prob > F

Model 205.54 14 14.68 6.80 0.0003 SignificantResidual 32.40 15 2.16Lack of fit 28.09 10 2.81 3.26 0.1023 Not significantPure error 4.31 5 0.86Cor. total 237.93 29

SD = 1.47; R2 = 0.86; R2Adj = 0.74.

W. Raza et al. / Bioresource Technology 101 (2010) 1904–1912 1909

inhibition zone as well as for growth (OD600). As this is evidentfrom the higher calculated F-values than the tabular F-value F0.05

(14, 10) = 4.60 at 0.05 level. Therefore, the model was found to beadequate for the prediction of optimal conditions within the rangeof variables (Ca2+, Ni2+, Mn2+, Cu2+) employed (Table 4).

The coefficient values of Eq. (3) (Table 5) were calculated andtested for their significance using the Design-Expert software.The P-values are used as a tool to check the significance of eachof the coefficients, which in turn can indicate the pattern of theinteractions between the variables. The smaller the value of P,the more significant is the corresponding coefficient. It can be seenfrom Table 5 that the linear coefficients (x2, x3 and x4), quadraticterm coefficients (x2

4) and cross product coefficients (x1x2, x1x4,x2x4 and x3x4) are significant. The other term coefficients (x1, x2

1,x2

2, x23, x1x3, and x2x3) are not significant.

The entire relationships between reaction factors and responsescan be better understood by examining the planned series of re-sponse surface plots (Fig. 3) generated from the predicted model(Eq. (3)) by using the Design-Expert software. Fig. 3A shows the3D response surface plot at varying Ca2+ and Mn2+ concentrationsand at fixed Ni2+ (300 lM) and Cu2+ (37.5 lM) concentrations.From Fig. 3A, it can be seen that the size of inhibition zones in-creased with the increase in concentration of Mn2+ at low levelsof Ca2+ and Ca2+ increased the size of inhibition zone at low levelof Mn2+ while at higher levels of Mn2+ and Ca2+, the interactive ef-fect of both metal ions was inhibitory but not the least. The 3D re-sponse surface plot of Fig. 3B also shows similar results. The Fig. 3Bgives the size of inhibition zones as a function of Ca2+ and Ni2+ con-centrations at fixed Cu2+ (37.5 lM) and Mn2+ (125 lM) concentra-tions. The size of inhibition zones as a function of Ca2+ and Cu2+

concentrations at fixed Mn2+ (125 lM) and Ni2+ (300 lM) concen-trations also showed similar results as above (data not shown).One noticeable observation in all above-mentioned figures (3A

Table 5Test of significance for regression coefficients.

Model term Coefficient estimate df S.E.

Intercept 21.54 1 0.60x1 �0.42 1 0.30x2 0.68 1 0.30x3 1.35 1 0.30x4 0.86 1 0.30x1x2 �0.94 1 0.37x1x3 �0.64 1 0.37x1x4 �0.87 1 0.37x2x3 0.54 1 0.37x3x4 �1.10 1 0.37x2x4 �0.90 1 0.37x2

10.31 1 0.28

x22

0.48 1 0.28

x23

0.35 1 0.28

x24

1.43 1 0.28

CI = Confidence limit.* <0.5.** <0.1.

and 3B) is that, according to the results of interactive effect ofNi2+, Cu2+ and Mn2+ with Ca2+; Ni2+, Cu2+ and Mn2+ dominants inincreasing the size of inhibition zone at lower levels of Ca2+ as com-pared with increase in size of inhibition zone by Ca2+ at lower lev-els of Ni2+, Cu2+ and Mn2+. Analyzing the Mn2+–Ni2+ plot (Fig. 3C), asignificant positively synergistic interactive effect on the produc-tion of antifungal compounds was observed at fixed Ca2+

(500 lM) and Cu2+ (37.5 lM) levels. The Mn2+ and Ni2+ positive ef-fect on size of inhibition zone was maximum at highest levels.According to the Fig. 3D, Cu2+ shows maximum increase in the sizeof inhibition zone at lower levels of Mn2+ while Mn2+ shows max-imum increase in the size of inhibition zone at lower levels of Cu2+.However, at higher concentrations a slight decrease in the size ofinhibition zone was observed. The response surface plots of the ef-fect of Ni2+ and Cu2+ and their mutual interactions depicted (datanot shown) significant positive synergistic effect on the productionof antifungal compound by P. polymyxa SQR-21 that increases withthe increase of Ni2+ and Cu2+ levels.

The data obtained in this experiment clearly indicated that theinteractive affect of metals ions on the production of metabolitesby P. polymyxa SQR-21 differed significantly from that when onlyone metal ion was applied. The optimal metal ion concentrationsfor the maximum production of antifungal compounds observedin one-factor with three concentration level experiments were600, 150, 400 and 50 lM for Ca2+, Mn2+, Ni2+ and Cu2+, respectively.However, when the interactive effects of metal ions were observed,all metal ions except Ca2+, increased the production of antifungalcompounds and the maximum size of inhibition zone (31 mm)was measured at 400 (�1), 150 (1), 400 (1) and 25 lM (�1) levelsof Ca2+, Mn2+, Ni2+ and Cu2+, respectively. In our experiment, Mn2+

and Ni2+ showed the strongest positive synergistic interactive ef-fect on the production of antifungal compounds followed by inter-active effect of Cu2+ and Ni2+ and then Mn2+ and Cu2+. The

95% Cl low 95% Cl high F-value Prob > F

20.26 22.82 – –�1.06 0.22 1.93 0.1851

0.041 1.32 5.15 0.0384*

0.71 1.99 20.18 0.0004**

0.22 1.50 8.23 0.0117*

�1.72 �0.15 6.51 0.0221*

�1.43 0.14 3.08 0.0996�1.66 �0.092 5.67 0.0309*

�0.24 1.32 2.17 0.1614�1.89 �0.32 9.02 0.0089**

�1.68 �0.11 5.95 0.0276*

�0.29 0.91 1.21 0.2889

�0.12 1.07 2.88 0.1105

�0.25 0.95 1.56 0.2301

0.84 2.03 26.10 0.0001**

Fig. 3. Response surface plots of the effect of Mn2+ and Ca2+ (A), Ni2+ and Ca2+ (B), Mn2+ and Ni2+ (C) and Cu2+ and Mn2+ (D) and their mutual interactions on the production ofantifungal compound by Paenibacillus polymyxa SQR-21.

1910 W. Raza et al. / Bioresource Technology 101 (2010) 1904–1912

interactive effect of Ca2+ with the other three metals resulted in theinhibition of the production of antifungal compounds. Similar ef-fects of these metal ions on the production of metabolites by differ-ent microorganisms have been reported. For example, calciuminhibited the biosynthesis of efrotomycin in resting cells of Nocra-dia lactamdurans (Cover et al., 1991) and cephamycin C in Strepto-myces clavuligerus (Chimel et al., 1986). Ni2+ increased theproduction of cephabacins in Lysobector Iactamgenus (Roh et al.,1992) and fortimicin A production in Micromonospora olivoasteros-pora (Yamamoto et al., 1977). Mn2+ stimulated polyene productionin Streptomyces (Liu et al., 1975). Pavl and Banerjee (1983) reportedthe promotive effect of Cu2+, Mn2+, Zn2+ and Fe2+ on growth andantibiotic yield of Streptomyces galbus. The addition of 0.03% cop-per sulfate, increased the production of antibiotics; hydroheptin

and chartreusin, by Streptomyces chartreusis IMRU 3962 (Tunacand McDaniel, 1985).

The results regarding the interactive effect of metal ions on bac-terial growth indicate that all metal ions except Ca2+ decreased thegrowth (OD600) of P. polymyxa in liquid culture as depicted from Eq.(4) although Ca2+, surprisingly, decreased the growth (OD600) of P.polymyxa in one factor with three concentration level experiments.The inhibitory effect of Ni2+ was significant while the effect of allother metal ions (Ca2+, Mn2+ and Cu2+) was less significant. Themaximum growth, as determined by measurement of OD600 (valueof 1.55) was obtained for the combinations of Ca2+, Mn2+, Ni2+ andCu2+ at concentrations of 500 (0), 125 (0), 100 (�2) and 37.5 (0)lM, respectively. Similar effects of these metal ions on the growthof other microorganisms have been reported. Growth of Escherichia

Fig. 4. The effect of optimal metal ions concentrations for maximum production of antifungal compounds (Ca2+ = 400 (�1), Mn2+ 150 (1), Ni2+ = 400 (1), and Cu2+ = 25 lM(�1)) on the growth (OD600), intracellular protein, carbohydrate and lipid contents and on extracellular protein and polysaccharide contents of P. polymyxa SQR-21.Ck = control; MI = metal ions.

W. Raza et al. / Bioresource Technology 101 (2010) 1904–1912 1911

coli was increased with the addition of calcium (Onoda et al., 2000).Nickel ion inhibited the biomass of Streptomyces coelicolor (Abbasand Edwards, 1990). Murata et al. (2005) reported the inhibitoryeffect of Cu2+ on bacterial growth and Mn2+ inhibited the growthof Leptothrix discophora (Adams and Ghiorse, 1985).

The metal ions concentrations that gave maximum productionof antifungal compounds were used to evaluate their effects onoverall metabolic activity of P. polymyxa SQR-21. The resultsshowed that OD600, IL, IP and EP contents were decreased whileEPS and IC contents were increased (Fig. 4). These results clearlyindicated that metal ions (Ca2+, Mn2+, Ni2+ and Cu2+) have a regula-tory effect, either, directly or indirectly, on bacterial secondarymetabolism mainly in the production of fusaricidins. The decreasein IP contents by 20% that was far less than decrease in EP content(76%). This observation suggests that these metal ions might possi-bly trigger the expression of proteins involved in the production offusaricidins and a transfer of resources towards the production ofantibiotic rather than other cellular processes such as extracellularproteins. This hypothesis is supported by the observation thatexpression of fusA increase with an increase in metal ions concen-trations in the liquid medium. The decrease in the concentration ofextracellular proteins also indicated toxic effects of these metalions on the production of extracellular enzyme. In one-factor withthree concentration level experiments, Ca2+ decreased the extra-cellular protein contents and in four-factor with five concentrationlevel experiments, Ca2+ seems to be inhibiting production ofmetabolite as it decreased the production of antibiotic by P. poly-myxa. So Ca2+ ion might be exerting some toxic effects on the pro-duction of extracellular enzyme, which is in agreement withprevious findings of Macedo et al. (1999) and Sami et al. (2008)who reported the decrease in the production of extracellular en-zymes by calcium.

Table 6Verified results of the model equation.

Trial no. Variables

x1 x2 x3 x4

1 400 100 280 252 410 110 285 273 420 120 290 294 430 130 300 315 440 140 305 336 450 150 310 357 460 160 315 378 470 170 320 399 480 180 325 4110 490 190 330 43

In addition to the decrease in IP contents, the IL contents werealso decreased and residual energy was being used for IC and EPSsynthesis. The cells of gram-positive bacteria like P. polymyxa arerigid and relatively insensitive to shear forces because of their thickcell wall. The exopolysaccharides that they form might play a cru-cial role in metal biosorption and precipitation as P. polymyxastrains have been used in the biosorption of copper (Acostalet al., 2005). The biological and chemical characteristics of theseuptake processes are important as an aid in the understanding ofthe role of metallic ions in basic cellular functions. The presenceof different metal ions such as Ca2+, Pb2+ and Mg2+ as well as oxideminerals of iron, aluminium and calcium influence the types andquantity of polysaccharides, proteins and enzymes secreted bythe bacteria (Reed, 1987). The mechanisms of how metal ions in-crease or decrease the growth and antifungal activity are compli-cated and have not yet been elucidated. In our case, metal ionsmight be interfering with secondary metabolism more general toenzymes or cellular processes. They may have stimulated or sup-pressed the synthesis of the prepeptide or the activation of theappropriate prepeptide maturation enzymes and the transportout of the cell. The Ca2+ binding sites were predicted to be presentin NisP peptidase, which cleaves the leader peptide from the pre-cursor nisin (Siezen et al., 1995). Since the precursor is devoid ofantibacterial activity (Meer van der et al., 1993), metal ions arethought to be involved in the activation or inactivation of the lea-der peptidase. The increase in concentration of antibiotic in themedium can be a consequence of the increase in cell wall perme-ability of SQR-21 promoted by metal ions. This is in agreementwith Petit-Glatron et al. (1993) who studied the capacity of the cellwall to concentrate metal ion and proposed that the increased con-centration of metal ions in the microenvironment of the cell wallcould play an important role in the last step of the secretion. An-

Y for Inhibition zone Y for OD600

Observed Predicted Observed Predicted

19.0 18.7 0.516 0.51420.5 19.9 0.551 0.52920.8 21.1 0.567 0.54422.5 22.2 0.573 0.56022.9 23.3 0.579 0.57623.0 24.5 0.583 0.59225.8 25.6 0.592 0.60726.3 26.7 0.611 0.62428.9 27.9 0.622 0.63829.3 29.1 0.686 0.654

1912 W. Raza et al. / Bioresource Technology 101 (2010) 1904–1912

other possibility is that the metal ions activated enzymes whoseactivity results in a change in the regulatory functions of the cellin favor of different secondary metabolites especially fusaricidins.This information will aid in developing fermentation technologyfor the production of maximum antibiotic. More research is neededat the cellular level, however, to evaluate the behavior of metalions alone and in different combinations on production of anti-fungal compounds by P. polymyxa.

4. Verification of predictive model

The adequacy of second order polynomial model equations(Eqs. (3) and (4)) was validated by conducting 10 verificationexperiments under different random combinations of metal ions(with in the tested area) and the results are shown in Table 6.The validation data were analyzed by using the SPSS software (Ver-sion 16.0, SPSS Inc.). The correlation coefficient (R) between theexperimental and predicted values of size of inhibition zone(which is an indication of production of antifungal compoundsby P. polymyxa SQR-21) is 0.9624 while correlation coefficient be-tween the experimental and predicted values of growth (OD600) ofP. polymyxa SQR-21 is 0.8805. The results indicate that the exper-imental values of size of inhibition zone and growth are in goodagreement with the predicted values and support the conclusionthat the models of Eqs. (3) and (4) provide satisfactory and accu-rate results for the prediction of optimal conditions within therange of variables (Ca2+, Ni2+, Mn2+, Cu2+) employed.

5. Conclusion

The results showed the marked advantage of using RSM overtraditional techniques to investigate the effect of metal ions ongrowth and production of antifungal compounds. It allowed us toobtain a regression equation predicting the growth and productionof the antifungal compounds and identification of the most impor-tant metal ions in this process. The Mn2+ (P = 0.0384), Ni2+

(P = 0.0004) and Cu2+ (P = 0.0117) significantly affected the produc-tion of antifungal compounds but the effect of Ca2+ (P = 0.1851)was not significant. This regression equation model is the firstapplication of RSM to determine the interactive effect of metal ionson the production of antifungal compounds by P. polymyxa SQR-21so that optimum media recipes can be developed to produce max-imum amounts of antifungal compounds under laboratory andcommercial fermentation conditions.

Acknowledgements

This research work was financially supported by the NationalNature Science Foundation of China (40871126), China Scienceand Technology Ministry, 973 Programme (2007CB109304) and863 Programme (2006AAD10Z416), and the China AgricultureMinistry (2006-G62). We thank Professor Warren Dick at OHIOState University, USA for improving the English of the manuscript.

References

Abbas, A.S., Edwards, C., 1990. Effects of metals on Streptomyces coelicolor growthand actinorhodin production. Appl. Environ. Microbiol. 56, 675–680.

Acostal, M.P., Valdman, E., Leite, S.G.F., Battaglini, F., Ruzal, S.M., 2005. Biosorptionof copper by Paenibacillus polymyxa cells and their exopolysaccharide. World J.Microbiol. Biotechnol. 21, 1157–1163.

Adams, L.F., Ghiorse, W.C., 1985. Influence of manganese on growth of a sheath lessstrain of Leptothrix discophora. Appl. Environ. Microbiol. 49, 556–562.

Arteaga, G.E., Li-Chan, E., Vasquez-Arteaga, M.C., Nakai, S., 1994. Systematicexperimental designs for product formula optimization. Trend. Food. Sci.Technol. 5, 243–254.

Beatty, P.H., Jensen, S.E., 2002. Paenibacillus polymyxa produces fusaricidin-typeantifungal antibiotics active against Leptosphaeria maculans, the causative agentof blackleg disease of canola. Can. J. Microbiol. 48, 159–169.

Bradford, M.M., 1976. A rapid and sensitive method for the quantitation ofmicrogram quantities of protein utilizing the principle of protein-dye binding.Analy. Biochem. 72, 248–254.

Chimel, A., Brzeszczynska, A., Kabza, B., 1986. Biosynthesis of cephamycin by restingcells of Streptomyces lactamdurans. Acta Microbiol. Polon. 35, 251–257.

Cover, W.H., Kirpekar, A.C., George, H., 1991. A calcium inhibition of efrotomycinproduction by Nocardia lactamdurans. J. Indust. Microbiol. Biotech. 7, 41–44.

Dijksterhuis, J., Sanders, M., Gorris, L.G.M., Smid, E.J., 1999. Antibiosis plays a role inthe context of direct interaction during antagonism of Paenibacillus polymyxatowards Fusarium oxysporum. J. Appl. Microbiol. 86, 13–21.

Dubois, M., Gilles, K.A., Hamilton, J.K., Rebers, P.A., Smith, F., 1956. Colorimetricmethod for determination of sugars and related substances. Anal. Chem. 28,350–356.

Izard, J., Limberger, R.J., 2003. Rapid screening method for quantitation of bacterialcell lipids from whole cells. J. Microbiol. Meth. 55, 411–418.

Li, J., Jensen, S., 2008. Nonribosomal biosynthesis of fusaricidins by Paenibacilluspolymyxa PKB1 involves direct activation of a d-amino acid. Chem. Biol. 15,118–127.

Li, C., Bai, J., Li, W., Cai, Z., Ouyang, F., 2001. Optimization of conditions forbacteriocin extraction in PEG/salt aqueous two-phase systems using statisticalexperimental designs. Biotechnol. Prog. 17, 366–368.

Liu, B.L., Tzeng, Y.M., 1998. Optimization of growth medium for the production ofspores from Bacillus thuringiensis using response surface methodology.Bioprocess Eng. 18, 413–418.

Liu, C.M., McDaniel, L.E., Schaffner, C.P., 1975. Factors affecting the production ofCandicidin. Antimicrob. Agents Chemother. 7, 196–202.

Macedo, J.M.B., Gottschalk, L.M.F., Bon, E.P.S., 1999. Lignin peroxidase and proteaseproduction by Streptomyces viridosporus T7A in the presence of calciumcarbonate. Appl. Biochem. Biotechnol. 79, 735–744.

Mahmood, M., 1970. Trace elements for growth and bulbiformin production byBacillus subtilis. J. Appl. Microbiol. 35, 1–5.

Meer van der, J.R., Polman, J., Beerthuyzen, M.M., Siezen, R.J., Kuipers, O.P., de Vos,W.M., 1993. Characterization of the Lactococcus lactis nisin A operon genes nisP,encoding a subtilisin-like serine protease involved in precursor processing, andnisR, encoding a regulatory protein involved in nisin biosynthesis. J. Bacteriol.175, 2578–2588.

Montgomery, D.C., 1991. Design and Analysis of Experiments, third ed. Wiley andSons Inc., New York. pp. 39–76.

Murata, T., Kanao-Koshikawa, M., Takamatsu, T., 2005. Effects of Pb, Cu, Sb, In andAg contamination on the proliferation of soil bacterial colonies, soildehydrogenase activity and phospholipid fatty acid profiles of soil microbialcommunities. Water Air Soil Pollut. 164, 103–118.

Onoda, T., Enokizono, J., Kaya, H., Oshima, A., Freestone, P., Norris, V., 2000. Effects ofcalcium and calcium chelators on growth and morphology of Escherichia coli L-Form NC-7. J. Bacteriol. 182, 1419–1422.

Pavl, A.K., Banerjee, A.K., 1983. Determination of optimum conditions for antibioticproduction by Streptomyces galbus. Folia Microbiol. 28, 397–405.

Petit-Glatron, M.F., Grajcar, L., Munz, A., Chambert, R., 1993. The contribution of thecell wall to a transmembrane calcium gradient could play a role in Bacillussubtilis protein secretion. Mol. Microbiol. 9, 1097–1106.

Raytapadar, S., Datta, R., Paul, A.K., 1995. Effects of some heavy metals on growth,pigment and antibiotic production by Streptomyces galbus. Acta Microbiol.Immunol. Hung. 42, 171–177.

Raza, W., Yang, X., Wu, H., Wang, Y., Xu, Y., Shen, Q., 2009. Isolation andcharacterization of fusaricidin type compounds producing strain ofPaenibacillus polymyxa SQR-21 active against Fusarium oxysporum F.sp.nevium. Eur. J. Plant Pathol. doi:10.1007/s10658-009-9496-1.

Reed, G., 1987. Industrial Microbiology, fourth ed. CBS Publishers, New Delhi.Roh, J.W., Bang, J.H., Nam, D.H., 1992. Nutritional requirements of Lysobacter

lactamgenus for the production of cephabacins. Biotechnol. Lett. 14, 455–460.Ryu, C.M., Kim, J., Choi, O., Kim, S.H., Park, C.S., 2006. Improvement of biological

control capacity of Paenibacillus polymyxa E681 by seed pelleting on sesame.Biol. Cont. 39, 282–289.

Sami, A.J., Awais, M., Shakoori, A.R., 2008. Preliminary studies on the production ofendo-1, 4-b-D-glucanases activity produced by Enterobacter cloacae. African J.Biotechnol. 7, 1318–1322.

Shoji, J., Kato, T., Hinoo, H., 1977. The structure of polymyxin S. Studies onantibiotics from the genus Bacillus, XXI.. J. Antibiot. 30, 1035–1041.

Siezen, R.J., Rollema, H.S., Kuipers, O.P., Vos de, W.M., 1995. Homology modelling ofthe Lactococcus lactis leader peptidase NisP and its interaction with theprecursor of the antibiotic nisin. Protein Eng. 8, 117–125.

Tunac, J.B., McDaniel, L.E., 1985. Effect of phosphate and copper on the fermentationof hydroheptin. Appl. Environ. Microbiol. 50, 1192–1195.

Wang, Z.W., Liu, X.L., 2008. Medium optimization for antifungal active substancesproduction from a newly isolated Paenibacillus sp. Using response surfacemethodology. Biores. Technol. 99, 8245–8251.

Weinberg, E.D., 1977. Mineral element control of microbial secondary metabolism.In: Weinberg, E.D. (Ed.), Microorganisms and Minerals. Marcel Dekker Inc., NewYork, pp. 289–316.

Yamamoto, M., Okachi, R., Kawamoto, I., Nara, T., 1977. Fortimicin A production byMicromonospora olivoasterospora in a chemically defined medium. J. Antibiot.30, 1064–1072.