Factorial analysis of tricarboxylic acid cycle intermediates for optimization of zeaxanthin production from Flavobacterium multivorum


Paul S. Bernstein, Department of Ophthalmology and Visual Sciences, 50 North Medical Drive, Moran Eye Center, University of Utah School of Medicine, Salt Lake City, UT 84132, USA (e-mail: paul.bernstein@hsc.utah.edu) .


Aims:  To study the effect of intermediates of the tricarboxylic acid (TCA) cycle on the production of zeaxanthin from Flavobacterium multivorum in order to optimize production of this xanthophyll carotenoid.

Methods and Results:  The concentration of selected TCA cycle intermediates (malic acid, isocitric acid and α-ketoglutarate) was optimized in shake flask culture, using a statistical two-level, three-variable factorial approach. The carotenoid production profile was also studied in the optimized medium at various growth phases. Optimized medium resulted in a sixfold increase in volumetric production of zeaxanthin (10·65 ± 0·63 μg ml−1) using malic acid (6·02 mm), isocitric acid (6·20 mm) and α-ketoglutarate (0·02 mm). The majority of zeaxanthin was produced in the late logarithmic growth phase whereas a substantial amount of β-cryptoxanthin and β-carotene were observed in the early logarithmic phase.

Significance and Impact of the Study:  This study demonstrates improvement of zeaxanthin production from F. multivorum which might aid in the commercialization of zeaxanthin production from this microbe.


Zeaxanthin (3,3′-dihydroxy-β-carotene) is an oxygenated carotenoid which may play a critical role in the prevention of cancer (Nishino et al. 2002) and age-related macular degeneration, the leading cause of blindness in the developed world (Snodderly 1995; Moeller et al. 2000). Zeaxanthin has also been used as a colourant in the cosmetic and food industries (Hadden et al. 1999).

Commercial demand for zeaxanthin is mainly fulfilled by chemical synthesis because there are few microbes that synthesize zeaxanthin as their main carotenoid (Johnson and Schroeder 1996). Despite the availability of a variety of synthetic carotenoids, there is currently renewed interest in microbial sources of carotenoids partially driven by public bias against synthetic chemical additives. Moreover, chemical synthesis of carotenoids involves a number of complex chemical conversion reactions, so efficient microbial production may be more cost-effective. The production of β-carotene and astaxanthin by Dunaliella sp. and Haematococcus pluvialis, respectively, are now well developed technologies. These micro-algae accumulate carotenoids as high as 4–10% (w/w) of their cell dry weight, which makes them a commercially viable source (Nelis and DeLeenheer 1991; Ausich 1997).

Recent publications indicate interest in the microbial production of zeaxanthin (Ruther et al. 1997; Jin et al. 2003). Among them, Flavobacterium multivorum, a nonfastidious and nonpathogenic bacterium which rapidly accumulates zeaxanthin is considered to be an important potential microbial source of zeaxanthin (Pasamontes et al. 1997; Alcantara and Sanchez 1999; Masetto et al. 2001); however, low carotenoid yield and a lack of information about optimization of media for carotenoid production restrict the possibilities of its use (Nelis and DeLeenheer 1991).

Increase in cell mass and carotenoid productivity will lead to an increase in total carotenoid production. The utility of culture broth supplementation with tricarboxylic acid (TCA) cycle intermediates as stimulants for carotenoid production has been previously reported for several microbes (Bjork and Neujahr 1969; Noparatnaraporn et al. 1986; Alcantara and Sanchez 1999), but their quantitative optimization and interactive effect were never reported. In these reports a variety of TCA intermediates affect carotenoid production at different levels. As these intermediates are sequentially made in the metabolic cycle, they are expected to have interactive effects on carotenoid production.

The conventional method of ‘one variable at a time’ used for optimization of medium is usually time consuming and often fails to assess the additive response of the variables. A factorial approach for optimization is convenient and can yield several fold improvements in the process as demonstrated in previously published studies (Florencio et al. 1998; Vazquez and Martin 1998; Bhosale and Gadre 2001; Ramirez et al. 2001). In this report, the effect of TCA intermediates on zeaxanthin production optimization from F. multivorum was studied.

Materials and methods


A 5% (v/v) inoculum of F. multivorum in the logarithmic phase (12 h), grown in the basal medium mentioned below was used throughout the studies (O.D.500 nm 1·3).

Reagents and chemicals

The media ingredients, sodium salts of TCA intermediates and β-carotene, were purchased from Sigma Chemical. HPLC grade methylene chloride, methanol, and hexane were obtained from Fisher Scientific. Synthetic zeaxanthin and β-cryptoxanthin were from Hoffmann-La Roche (Basel, Switzerland), and lutein prepared from marigold flowers was a gift from Kemin Foods (Des Moines, Iowa).

Bacterial strain and growth medium

Flavobacterium multivorum ATCC 55238 was obtained from American Type Culture Collection (ATCC, Manassas, VA, USA). The culture was maintained on Yeast-Malt extract (YM) agar containing (g l−1) glucose 35, malt extract 30, yeast extract 20, peptone 10, MgSO4·7H2O 0·2 and agar 25, at pH 6·0. The basal liquid growth medium for shake flask studies contained (g l−1) glucose 25, yeast extract 10, peptone 10, MgSO4·7H2O 0·2, NaCl 1·5, K2HPO4 1, MnCl2 0·1, FeSO4·7H2O 0·01 and CoCl2·6H2O 0·01 at pH 7·0. pH was adjusted to 7·0 at the beginning of each experiment.

Pigment extraction from the bacterial cells

One millilitre of culture broth was centrifuged at 1800 g for 10 min at 4°C. The supernatant was discarded, and media components in the cell pellet were washed away three times by suspension in sterile saline (pH 7·0) and centrifugation. The cell mass was subjected to sonication using a sonic dismembrator (Fisher Scientific, Model number F60) in the presence of 1 ml of cold, oxygen-free methanol containing 0·01% butylated hydroxy toluene (BHT) (w/v) for 30 s (output power 5). The sonicated sample was centrifuged to remove the white cell pellet. The supernatant contains extracted carotenoids.

High-performance liquid chromatography

The pigments in 1 ml methanol were dried by vacuum evaporation in a Speedvac Plus (SC110 Savant Instruments, Inc., Holbrook, NY, USA) and re-dissolved in 1 ml of HPLC mobile phase [hexane: dichloromethane: methanol: N,N′-di-isopropylethylamine (80 : 19·2 : 0·7 : 0·1)]. HPLC separation was carried out at a flow rate of 1·0 ml min−1 on a cyano column (Microsorb 25 cm length × 4·6 mm id; Rainin Instrument Co. Woburn, MA, USA). The column was maintained at room temperature, and the HPLC detector was operated at 450 nm. Peak identities were confirmed by photodiode-array spectra and by coelution with authentic standards as necessary.

Cell growth measurement

Cell growth was monitored by measurement of turbidity at 500 nm with a u.v.-visible spectrophotometer (Smart SpecTM 3000; BioRad Laboratories, Inc., Hercules, CA, USA). Samples were diluted suitably (so as to have absorbance between 0·2 and 0·8) with double-distilled water, and absorbance was measured immediately at 500 nm. For cell dry weight (CDW) estimation, the 10 ml sample was centrifuged at 2850 g for 10 min at 4°C and washed twice with double-distilled water by suspension and centrifugation. The supernatant was discarded, and the cell pellet was then taken to constant weight in an oven at 80°C. The relationship between absorbance measured at 500 nm for cell growth and estimated CDW was observed to be CDW = 0·432 × O.D. at 500 nm at all stages of growth.

Effect of TCA cycle intermediates

To study the effect of TCA cycle intermediates, F. multivorum was studied for growth and carotenoid production in the liquid basal medium supplemented with sodium salts of TCA intermediates (10 mm) listed in Table 1. The experiments were performed in quadruplicate in shake flask culture at 30°C on a rotary shaker at 250 rev min−1.

Table 1.  Effect of TCA cycle intermediates (10 mm) on growth and zeaxanthin production from Flavobacterium multivorum grown for 44 h in shake flask culture
TCA cycle intermediateGrowth (O.D.500 nm)Zeaxanthin
(μg g−1)(μg ml−1)
Control3·38 ± 0·031·04 ± 0·011·77 ± 0·09
Acetic acid4·50 ± 0·140·68 ± 0·011·54 ± 0·13
Citric acid4·01 ± 0·011·01 ± 0·022·04 ± 0·17
Fumaric acid4·80 ± 0·560·73 ± 0·011·75 ± 0·01
Isocitric acid7·85 ± 0·070·95 ± 0·013·61 ± 0·06
α-Ketoglutarate5·52 ± 0·161·15 ± 0·013·37 ± 0·40
Malic acid5·16 ± 0·221·61 ± 0·034·19 ± 0·18
Oxaloacetic acid4·31 ± 0·430·73 ± 0·011·48 ± 0·11
Succinic acid3·70 ± 0·140·74 ± 0·011·34 ± 0·02

Factorial design

Two-level, three-variable factorial experiments were designed using the approach described earlier (Box et al. 1978; Davies 1993; Bhosale and Gadre 2001). After initial experiments, the three best TCA intermediates were selected for further studies based on higher quantitative levels of zeaxanthin. The effects of selected TCA intermediates [malic acid (MA), isocitric acid (ICA) and α-ketoglutarate (αKG)] were studied using the two-level factorial design given in Table 2. Response was assessed by determination of volumetric production of zeaxanthin (μg ml−1). The change in the response from the ‘minus’ to ‘plus’ level was calculated as effect (E) of the variable under study.

Table 2.  Results of factorial experiment using Flavobacterium multivorum grown for 44 h
Flask no.Design matrix (mm)Growth (O.D.500 nm)β-Cryptoxanthin (μg ml−1)Zeaxanthin (μg ml−1)Effect (E)*
  1. *E0 represents average effect of eight experiments; EMAEICAEKG represents main effects. EMA·ICA, EMA·ICA, EICA·KG represent two factor interactive effects and, EMA·ICA·KG represents three factor interactive effect of variables (MA, malic acid; ICA, isocitric acid; KG, α-ketoglutarate).

(a) First factorial
Centroid1010106·53 ± 0·180·06 ± 0·045·16 ± 0·21
15557·05 ± 0·210·25 ± 0·012·57 ± 0·09E0: 2·04
215557·95 ± 0·070·27 ± 0·043·28 ± 0·10EMA: 0·47
351559·85 ± 0·210·36 ± 0·013·36 ± 0·23EICA: 0·56
41515510·35 ± 0·20·45 ± 0·013·58 ± 0·30EKG: −2·29
555156·60 ± 0·420·11 ± 0·010·32 ± 0·16EMA·ICA: −0·12
6155155·75 ± 0·770·15 ± 0·040·90 ± 0·01EMA·KG: 0·05
7515154·10 ± 0·140·13 ± 0·010·94 ± 0·08EICA·KG: 0·01
81515153·75 ± 0·210·17 ± 0·021·39 ± 0·01EMA·ICA·KG: 0·70
(b) Second factorial
Centroid11·0211·25·0211·11 ± 0·180·29 ± 07·84 ± 0·0
16·026·20·0215·56 ± 0·110·63 ± 0·0410·65 ± 0·63E0: 5·78
216·026·20·0211·52 ± 0·280·62 ± 0·025·50 ± 0·31EMA: 0·12
36·0216·20·0213·40 ± 0·280·66 ± 0·016·45 ± 0·63EICA: −0·62
416·0216·20·0215·46 ± 0·190·92 ± 0·227·55 ± 0·49EKG: −3·50
56·026·210·028·24 ± 0·120·22 ± 0·011·59 ± 0·15EMA·ICA: −1·60
616·026·210·0213·12 ± 0·280·27 ± 0·016·62 ± 0·25EMA·KG: 2·30
76·0216·210·028·84 ± 0·330·13 ± 0·044·19 ± 0·28EICA·KG: 0·22
816·0216·210·028·36 ± 0·620·12 ± 0·023·68 ± 0·30EMA·ICA·KG: −2·67

Data obtained from the factorial experiments were fitted into a Yates algorithm for evaluating the relative importance of each variable, according to Box et al. (1978). E0 represents the average effect of eight experiments carried out in the 23 factorial design matrix, which was calculated by means of an algorithm developed by Yates. The Yates algorithm is applied to the observations after they have been rearranged in standard order as detailed by Box et al. 1978. The main effects of the quantitative variables (EMA, EICA, EαKG), the interaction effects of two variables (EMA·ICA, EICA·KG, and EMA·KG) and the three-factor interaction (EMA·ICA·KG) was calculated as per Box et al. 1978. This method for the calculation of main and interaction effects was followed for both first and second factorial experiments. New treatment combinations of different variables were determined in a search for the optimum combination using the method of steepest ascent, as recommended by Davies (1993).

The method of steepest ascent was applied to optimize response when the process moves from basal level to a point in the factor space where gain in zeaxanthin level was maximized. Step size from the centroid of the factorial experiment was calculated from the main effects only. The positive value of main effect indicates that increases in the quantity of the variable will increase the yield. The six steepest ascent steps in the most important factor during each factorial experiment should equal the difference between its level in the factorial experiment. However, the step size for the other two factors was calculated by considering the effect of the most important factor along with their individual effect (Davies 1993). The second factorial (two-level) was performed above the best point in the steepest ascent experiment. This was repeated until no further increase was noted. All the experiments were carried out in duplicate.

Growth and carotenoid production in optimized medium

Growth and carotenoid production by F. multivorum were studied in quadruplicate using optimized medium in shake flasks at 30°C on a rotary shaker at 250 rev min−1. Samples were removed periodically and analyzed for cell growth as well as carotenoid production. Carotenoid values are expressed in volumetric units (μg ml−1 of culture broth) and/or cellular accumulation (μg g−1 of cell dry weight). The percentage (%) of total carotenoids is expressed as total amount of individual carotenoid present when compared with total carotenoid content added together.


Effect of TCA cycle intermediates

The effect of single additions of TCA cycle intermediates on growth and zeaxanthin production is shown in Table 1. Most of the TCA cycle intermediates had a positive effect on growth compared with control. In the control medium, zeaxanthin, β-cryptoxanthin, and β-carotene were 1·77 ± 0·09, 0·24 ± 0·03 and 0·076 ± 0·01 μg ml−1, respectively. Supplementation led to higher cell mass at the end of fermentation (44 h). Isocitric acid, malic acid and α-ketoglutarate had maximum stimulatory effect on volumetric production of zeaxanthin (μg ml−1; Table 1). During growth, supplementation of citric acid and oxaloacetate displayed a decrease in pH (3–4) of the growth medium irrespective of the presence of buffering salts (pH 7·0), which might be the reason for relatively poor growth and volumetric carotenoid content. Addition of acetic and succinic acid also displayed lower volumetric carotenoid content when compared with control.

Cellular accumulation (μg g−1) of carotenoids was not affected much by supplementation with TCA intermediates. Maximum accumulation was observed with malic acid supplementation. In most cases, zeaxanthin was found to be 90 ± 2% (w/w) of the total carotenoid content, with the remainder as β-cryptoxanthin (8 ± 2%), β-carotene (1·5 ±0·25%), and lutein (0·5 ± 0·1%).

Optimization of TCA cycle intermediates

Based on high volumetric zeaxanthin levels in the above experiment, isocitric acid, malic acid, and α-ketoglutarate were selected for further studies on optimization and interactive effects. Preliminary experiments using 10 mm supplementations of each of the three intermediates resulted in threefold improvement in zeaxanthin production (5·16 ± 0·21 μg ml−1) with a concomitant fourfold decrease in the β-cryptoxanthin level (0·06 μg ml−1).

When the effect (E) of these intermediates on zeaxanthin production was studied using a factorial design, the first factorial experiment indicated that the individual effect of α-ketoglutarate was much stronger when compared with isocitric acid and malic acid (Table 2a). However, the effect was negative, indicating that a decrease in the α-ketoglutarate level will lead to an increase in zeaxanthin production. However, the effect of isocitric acid and malic acid was in the positive direction. The interactive effect of malic acid and isocitric acid was slightly more significant than the other two interactive effects. However, its negative value suggested that in order to reach the optimum value, a simultaneous step down in quantitative values is required. The step size for decrease or increment was determined using the method of steepest ascent. The steps were calculated using main effects only.

Experiments along the direction of steepest ascent yielded the best volumetric zeaxanthin production (7·84 ±0·01 μg ml−1) at malic acid, isocitric acid and α-ketoglutarate concentrations of 11·02, 11·20 and 5·02 mm, respectively (Table 3a). The yield was 4·3-fold higher than the basal level. β-Cryptoxanthin was 1·9 ± 0·2% (0·15 ±0·05 μg ml−1) of the total carotenoid at this stage.

Table 3.  Results of steepest ascent experiments using Flavobacterium multivorum grown for 44 h
Flask no.MAICA (mm)α-KGGrowth (O.D.500 nm)β-Cryptoxanthin (μg ml−1)Zeaxanthin (μg ml−1)
(a) Steepest ascent experiment no. 1
0 level1010106·53 ± 0·180·06 ± 0·045·16 ± 0·21
110·3410·48·348·26 ± 0·340·08 ± 0·014·91 ± 0·07
210·6810·86·688·68 ± 0·390·08 ± 0·024·77 ± 0·12
311·0211·25·0211·38 ± 0·250·15 ± 0·057·63 ± 0·47
411·3611·43·3613·43 ± 0·310·11 ± 0·015·89 ± 0·15
511·7011·61·78·26 ± 0·370·16 ± 0·027·15 ± 0·07
612·0411·80·0410·33 ± 0·980·11 ± 0·035·30 ± 0·42
(b) Steepest ascent experiment no. 2
0 level11·0211·205·0211·11 ± 0·180·15 ± 0·217·84 ± 0·07
111·0710·913·410·78 ± 0·310·13 ± 0·015·54 ± 0·65
211·1310·621·815·09 ± 0·751·21 ± 0·065·21 ± 0·03
311·1810·330·215·66 ± 0·480·21 ± 0·019·79 ± 0·16
411·2410·04012·91 ± 0·440·32 ± 0·018·35 ± 0·43
511·309·75012·80 ± 1·310·30 ± 0·108·05 ± 0·21
611·359·46012·83 ± 0·240·24 ± 0·017·80 ± 0·28

When the second factorial was performed with the above point at the center of the cube, the individual effect of α-ketoglutarate was observed to be much stronger towards the negative direction (Table 2b). The highest level of zeaxanthin (10·65 ± 0·63 μg ml−1) was observed at malic acid, isocitric acid and α-ketoglutarate levels of 6·02, 6·20 and 0·02 mm respectively, displaying sixfold improvement in the zeaxanthin production when compared with the control. β-Cryptoxanthin's level went up to 0·63 ±0·04 μg ml−1 and thus represented 5·5 ± 0·2% of the total carotenoid level. The low level of α-ketoglutarate in the optimized medium was observed to be influential as a slight decrease in the zeaxanthin level (9·85 ± 0·02 μg ml−1) was observed in the absence of α-ketoglutarate.

Experiments performed along the direction of steepest ascent resulting from the second factorial design showed no further improvement in volumetric zeaxanthin production (Table 3b).

Growth and carotenoid production in optimized medium

Growth and carotenoid production profiles of F. multivorum ATCC 55238 were studied in optimized medium and were compared with the basal medium (Fig. 1). The maximum specific growth rate (μmax) and maximum biomass achieved in optimum medium were 0·37 h−1 and 6·5 g l−1, respectively, when compared with 0·21 h−1 and 2·0 g l−1 obtained in the basal medium.

Figure 1.

Growth pattern (O.D. at 500 nm) and production profiles (μg ml−1) of zeaxanthin, β-cryptoxanthin, and β-carotene in Flavobacterium multivorum ATCC 55238 grown in basal (□) and optimized medium (bsl00001)

In the optimized medium, the production profile indicated that zeaxanthin, β-cryptoxanthin and β-carotene were the major carotenoids produced during the initial logarithmic phase up to 20 h. Maximum production rate for zeaxanthin, β-cryptoxanthin, and β-carotene during this phase was 0·068, 0·113 and 0·108 μg ml−1 h−1, respectively. At 20 h, the total carotenoid content was found to be 8·18 ± 0·2 μg ml−1, and zeaxanthin, β-cryptoxanthin, and β-carotene were in the proportion of 66:13:21.

During the late logarithmic phase (20–32 h), a drastic increase in zeaxanthin production rate was observed with a concomitant decrease in β-cryptoxanthin and β-carotene content. Maximum production rate for zeaxanthin was observed to be 0·43 ± 0·3 μg ml−1 h−1. At this stage in the optimized medium, the specific zeaxanthin accumulation rate was maximum (198·9 μg g−1 h−1) when compared with basal medium (17·8 μg g−1 h−1). At 44 h, total carotenoid content was observed to be 10·74 ± 0·6 μg ml−1, and zeaxanthin, β-cryptoxanthin, and β-carotene were in the proportion of 93:5:2. A trace of lutein production was also observed in the late hours of growth. In the basal medium, maximum zeaxanthin production rate was observed between 12 and 16 h (0·16 μg ml−1 h−1). After 44 h, the total carotenoid level was 2·04 μg ml−1, representing zeaxanthin, β-cryptoxanthin, and β-carotene in the proportion of 89:6:5.


The production of carotenoids from biological sources has been an area of extensive investigation in the last decade. Microbial production of β-carotene and astaxanthin are well known technologies; however, there is a need to develop microbial technology for other carotenoids as well. Among several microorganisms reported, nonphotosynthetic Flavobacterium sp. are the only ones that have potential for commercialization for zeaxanthin production (Nelis and DeLeenheer 1991). Recently, cultural studies indicate the possibilities of hyperproduction from Flavobacterium sp. (Alcantara and Sanchez 1999). However, F. multivorum used in this study and several other reported strains need a ‘strategy to improve fermentation medium’ in order to increase zeaxanthin concentration, yield and volumetric productivity.

In this study, peptone was chosen as a nitrogen source despite of the fact that it is rich in glutamate and other amino acids. Our initial experiments with the medium revealed that an increase in peptone concentration did not lead to an increase in carotenoid production from Flavobacterium. It can be assumed that the amount of glutamate was just sufficient to support the growth of the bacteria without influencing carotenoid production. Moreover, peptone is a complex medium, and it is difficult to assess the interactive response of its individual components. Therefore experiments were performed with supplementation of TCA intermediates keeping suitable media control.

Intermediates of the TCA cycle are known to form carbon skeletons for carotenoid and lipid biosynthesis in microbes. There are two possible explanations available in the literature for the enhancement of carotenoid production from microbes by supplementation of TCA cycle intermediates (Bjork and Neujahr 1969; An 2001).

It was proposed that high respiratory and TCA cycle activity is associated with production of the large quantities of reactive oxygen species (free radicals and singlet oxygen), which are known to enhance carotenoid production. They play an important role in the stimulation of carotenogenesis as a part of inherently complex microbial regulatory defense mechanisms that function at both the gene and the protein levels (Schroeder and Johnson 1995; An 2001). The increase in the specific carotenoid accumulation between 20–32 h of growth phase was indicative of hypersynthesis of carotenoid as a part of a defense mechanism against reactive oxygen species.

TCA cycle intermediates supplied to Xanthophyllomyces dendrorhous displayed twofold improvement in the volumetric production of astaxanthin. Alcantara and Sanchez (1999) proposed that the positive effect of TCA cycle intermediates could be because of an increase in the pool of oxaloacetate, which is decarboxylated to pyruvate, which in turn increases the acetyl CoA pool, the starting substance for isoprenoid synthesis. Flavobacterium sp. displayed increased growth and zeaxanthin production upon supplementation of 10 mm concentrations of intermediates of the TCA cycle.

Intermediates of the TCA cycle displayed improvement in zeaxanthin production upon supplementation to basal growth medium. However, several fold further improvement could be obtained by a systematic factorial approach of quantitative optimization of selected intermediates. A factorial approach has advantages over a one-factor-at-a-time method, which lacks the ability to detect interaction among factors. The factorial approach in this study detected both effects of TCA cycle intermediates and interactions among TCA cycle intermediates because every observation in the factorial design gave information about all factors.

In both factorial experiments, α-ketoglutarate was observed to be more influential than the other two selected components. It is difficult to explain the direct effect of α-ketoglutarate in the carotenoid biosynthetic pathway especially when in the absence of malic acid and isocitric acid, α-ketoglutarate (10 mm) led to threefold improvement in the zeaxanthin level. Bjork and Neujahr (1969) suggested that specific actions of the TCA intermediates on some key enzymes involved in isoprenoid biosynthesis pathway like thiolase and acetyl-CoA-carboxylase can influence the carotenoid level. It can be hypothesized that in the presence of malic acid and isocitric acid, some kind of concentration-dependent inhibition of carotenoid biosynthetic enzymes by α-ketoglutarate might lead to a decrease in carotenoid yield.

The optimum values for TCA intermediates was observed to be 6·02, 6·20 and 0·02 mm for malic acid, isocitric acid and α-ketoglutarate, respectively, addition of which showed sixfold improvement in the zeaxanthin production (10·65 ± 0·63 μg ml−1). Carotenoid production profiles indicated the presence of significant amounts of β-carotene and β-cryptoxanthin in the early stages of growth. This was in accordance with McDermott et al. (1973) who proposed that β-carotene and β-cryptoxanthin are intermediates in zeaxanthin biosynthesis, wherein β-carotene and β-cryptoxanthin are ultimately hydroxylated in the later stages of growth to form zeaxanthin. The quantitative levels of zeaxanthin after fermentation reported in this study were much higher than recently reported values from Flavobacterium sp. (Alcantara and Sanchez 1999) and comparable with Flavobacterium sp. grown in a four liter fermenter (Masetto et al. 2001).

Thus, supplementation of TCA cycle intermediates in the culture medium leads to improvements in zeaxanthin production. Scale-up studies in high volume bioreactors using the optimized medium proposed in this study might lead to further improvement of zeaxanthin production from F. multivorum. A systematic approach of scale up using batch, fed-batch, continuous culture fermentation, and high cell density culture will determine the competitive future of F. multivorum as a zeaxanthin source, which in turn may facilitate commercial nutritional supplementation against diseases such as age-related macular degeneration and cancer.


This work was supported by National Institute of Health Grant EY-11600 and by Research to Prevent Blindness, Inc. (New York, NY, USA). PSB is a Sybil B. Harrington Research to Prevent Blindness Scholar in macular degeneration research.