Biosynthesis of a novel C30 carotenoid in Bacillus firmus isolates



Gerhard Sandmann, Molecular Biosciences, Goethe University Frankfurt, Max-von-Laue Str. 9, D-60438 Frankfurt, Germany. E-mail:



Pigmented Bacillus spp. with probiotic properties have been isolated. In the yellow-/orange-coloured strains, the carotenoid pigments present have been characterized. In contrast, the carotenoids present in the Bacillus isolates coloured red await identification. The present article reports progress on the elucidation of the pigment biosynthetic pathway in these red-pigmented Bacillus firmus strains.

Methods and Results

A combination of UV/Vis, chromatographic and mass spectrometry (MS) has revealed the properties of the predominant pigment and the end-point carotenoid of the pathway to be methyl 4,4′-diapolycopene-dioate after transmethylation. The diglycosyl ester of 4,4′-diapolycopene-dioate persists in vivo prior to chemical treatment. Different mutants and inhibitor treatment were employed to establish the C30 biosynthesis pathway with all precursors and intermediates to 4,4′-diapolycopene-dioate detected, which include 4,4′-diapophytene and all desaturation intermediates to 4,4′-diapolycopene and 4,4′-diapolycopene-dialdehyde. To cultures synthesizing the 4,4′-diapolycopene-dioate derivative and those in which its formation was inhibited, oxidative stress was induced by peroxide treatment. Conditions that decreased the growth rate of the pigmented cells by only 30% caused a complete growth inhibition of the culture devoid of the 4,4′-diapolycopene-dioate derivative.


This finding demonstrates the diversity of C30 carotenoid biosynthesis in Bacillus species and the antioxidative function of the 4,4′-diapolycopene-dioate derivative in B. firmus cells.

Significance and Impact of the Study

It could be shown that the C30 4,4′-diapolycopene-dioate derivatives protect pigmented B. firmus from peroxidative reactions. Under oxidative conditions, this can be an ecological advantage over nonpigmented (=noncarotenogenic) strains that are equally abundant.


Carotenoids are commonly found in pigmented bacteria. These terpenoids posses a chain of 30, 40 or 50 carbons. C40 carotenoids with an acyclic or bicyclic structures with modifying groups at the ionone rings are present in most bacterial groups (Goodwin 1980) including especially photoautotrophic species (Takaichi 1999). C50 carotenoids are restricted to several groups of Gram-positive bacteria, including species from the genera Micrococcus, Halobacterium, Corynebacterium and Flavobacterium (Britton 1988). A prominent example is decaprenoxanthin with two substituted ε-end groups. Its biosynthesis and the function of the genes involved were elucidated for Corynebacterium glutamicum (Krubasik et al. 2001). C30 carotenoids are another minor group. They are found in unrelated genera and species like Methylobacterium rhodium (formerly Pseudomonas rhodos) (Kleinig et al. 1979), Methylomonas (Tao et al. 2005), Staphylococcus aureus (Marshall and Wilmoth 1981), heliobacteria (Takaichi et al. 2003), Rubritella squalenifaciens (Shindo et al. ,2007, 2008a), Planococcus maritimus (Shindo et al. 2008b) and Halobacillus halophilus (Köcher et al. 2009). The C30 carotenoid products of bacterial pathways include staphyloxanthin (Pelz et al. 2005) derived from diaponeurosporene, a methyl glucosyl apo-lycopenoate derivative (Osawa et al. 2010), acylglycosyl monoesters (Shindo et al. 2008a,b) or diacylglycosyl diesters both derived from 4,4′-diapolycopene (Shindo et al. 2007). The best studied carotenogenic gene cluster for the C30 pathway is the one from Staph. aureus, of which all genes involved in staphyloxanthin synthesis were functionally assigned (Pelz et al. 2005).

Most Bacillus species are nonpigmented. However, yellow Bacilli have been reported, some time ago (Duc et al. 2005). In a survey of new Bacillus isolates, yellow-, orange- and red-pigmented strains were found (Khaneja et al. 2010). Preliminary studies with the desaturase inhibitor diphenylamine (DPA) resulted in the accumulation of diapophytoene, indicating the existence of a C30 carotenoid biosynthesis pathway. In the present investigation, we analysed the carotenoids from two red-pigmented Bacillus isolates both related to B. firmus (Khaneja et al. 2010). Using a combination of different chromatographic and spectroscopic procedures, the structures of the carotenoid intermediate products of the pathway were elucidated.

Materials and methods

Strains, cultivation and treatment

Bacillus firmus strains SF241 and GB1 (Khaneja et al. 2010) were grown with Luria Bertani medium usually for 3 days at 28°C. The cells were scraped off the plates and freeze-dried. Colourless mutant SF241-W2 was obtained by mutagenesis with ethyl methanesulfonate (EMS) following the protocol of Dubnau et al. (1973). To liquid cultures of a density of 109 cells per ml, EMS was added to a final concentration of 5%. After incubation for 20 min, cells were washed, were grown overnight in a fresh medium and were plated out for the screening of colonies with whitish pigmentation. Large-scale growth for the isolation of C492 from GB1 was for 2 days in a Biostat UD 100 fermenter (Braun, Melsungen, Germany) filled with 50 l LB medium, gassed with 8 l h−1 air and stirred with 150 rev min−1 at the optimum fermenter temperature of 32°C.

Peroxidation experiments were carried out with GB1 cultures inoculated to an OD600 of 0·03 by immediate application of cumene hydroperoxide (250 μmol l−1 final concentration) and by growth determination from the exponential phase of cultures grown for 20 h. Parallel cultures were treated additionally with 50 μmol l−1 DPA which was applied directly after inoculation for the complete inhibition of formation of coloured carotenoids. This concentration was the lowest to obtain full inhibition of coloured carotenoid formation and to minimize the growth inhibition.

Carotenoid extraction and HPLC separation

Carotenoids were extracted from freeze-dried liquid cultures harvested prior to sporulation by heating at 60°C for 20 min in dichloromethane/methanol (1 : 1) and partitioned into diethyl ether prior to HPLC analysis. Carotenoid acids were obtained by saponification with 10% KOH in methanol for 15 min at 60°C or in case of transmethylation with 6% KOH. Carotenoid acids partitioned into ether after the pH was brought to 2–3 by the addition of HCl. Carotenoids were analysed by HPLC on a non-end-capped polymeric 3-μm C30 column (YMC, Wilmington, NC, USA) (Sander et al. 1994) at a temperature of 26°C and eluted with a linear gradient from methanol/water (96 : 4) to methanol/ter-butyl methyl ether/water (26 : 70 : 4) over 30 min (Breitenbach et al. 2001). Spectra were recorded online using a Kontron DAD 440 diode array detector (Kontron Instruments, Neufahrn, Germany). Reference compounds were generated in Escherichia coli transformed with plasmids to synthesize diapophytoene, diaponeurosporene and diapolycopene (Raisig and Sandmann 2001) or 4,4′-diapolycopene-4-al (Tao et al. 2005). Diacylglycosyl-4,4′-diapolycopene-4,4′-dioate was isolated from Methylobacterium rhodium (Kleinig et al. 1979). Polarity of carotenoids was compared by TLC on activated silica plates developed with toluene/ethyl acetate/methanol (50 : 30 : 20) (Taylor and Davies 1983).

HPLC and mass spectroscopy

Mass spectrometry (MS) was performed after separation by HPLC prior to on-line MS as described above with an altered mobile phase to facilitate ionization. Mobile phase was comprised of (A) methanol containing 0·1% formic acid (by volume) and (B) tert-butyl methyl ether containing 0·1% formic acid (by volume). These solvents were used in a gradient mode starting at 100% (A) for 5 min, then stepped to 95% (A) for 4 min, followed by a linear gradient over 30 min to 25% (A). After this, gradient (A) was stepped down to 10% over 10 min. Initial conditions (100% A) were restored for 10 min after the gradient to re-equilibrate the system. The flow rate used was 0·2 ml min−1. The ionization mode employed was atmospheric pressure chemical ionization (APCI) operating in a positive mode (Thermo Scientific, San Jose, CA, USA). Capillary and APCI vaporization temperatures were set at 225 and 450°C, respectively, and the gas flow (nitrogen) at 80 units. APCI source settings were as follows: source voltage at 4·5 kV, source current of 5 μA and a capillary voltage of 3 V. A full MS scan was performed from 300 to 1500 m/z, and MS/MS spectra were recorded at normalized collision energy of 35% and isolation width of 1 m/z.


Properties of C492

Recently, a series of coloured Bacilli have been isolated (Khaneja et al. 2010). Among them were B. firmus isolates SF241 and GB1 with a novel red carotenoid assigned as C492 because of its main absorbance at 492 nm. Carotenoid extracts from both strains were separated by HPLC. The carotenoid patterns of SF241 (Fig. 1a) and GB1 (Fig. 1b) are very similar. Within the characteristic chromatograms, peak 1 of the HPLC trace occurred at 18·3 min and was identified as the all-trans isomer of C492. Its typical spectrum is shown in Fig. 2(a). One of the advantages of the C30 column used here for carotenoid separation is the discrimination of cis/trans isomers (Breitenbach and Sandmann 2005). Compounds 1′ and 1′′ represent different cis isomers as indicated by an additional cis peak at 380 nm and a shift of the main maxima from 492 to 490 and 485 nm, respectively. The spectrum of C492 indicates the presence of 13 conjugated double bonds as chromophore (Takaichi and Shimada 1992). Among the known carotenoids, C492 closely resembles diacylglycosyl-4,4′-diapolycopene 4,4′-diacid, the main carotenoid isolated from M. rhodium (Kleinig et al. 1979), which has the same spectrum as C492. However, diacylglycosyl-4,4′-diapolycopene-4,4′-dioate in the chromatogram derived from M. rhodium extracts (Fig. 1c) exhibited a longer retention time of 25·9 min for the all-trans isomer (Fig. 1c, peak 2) than the carotenoid C492 found in B. firmus SF241 and GB1 (Fig. 1a,b).

Figure 1.

Carotenoid accumulation in Bacillus strains SF241 (a) and GB1 (b). Trace (c) shows diacylglycosyl-4,4′-diapolycopene-4,4′-dioate extracted from Methylobacterium rhodium; (d) and (e) are KOH hydrolysates from the major carotenoid of SF241 and diacylglycosyl-4,4′-diapolycopene-4,4′-dioate, respectively. The carotenoid in trace (f) is the reference compound 4,4′-diapolycopene-4,4′-dioate produced in Escherichia coli.

Figure 2.

On-line absorbance spectra of C492 eluted in peak 1, peak 2 and peak 3 (a), 4,4′-diapophytoene (b), 4,4′-diapo-ζ-carotene (c), 4,4′-diapophytofluene (d), 4,4′-diaponeurosporene (e), 4,4′-diapolycopene (f) and 4,4′-diapolycopene-4,4′-dial (g). Absorption spectrum of peak 3 in Fig. 1(h). Spectra of peak 2 (diacylglycosyl-4,4′-diapolycopene-4,4′-dioate from Methylobacterium rhodium) correspond with those of peak 1 (C492 in SF241 and GB1) and peak 3 (hydrolysis product of C492) in Fig. 1 with 4,4′-diapolycopene-4,4′-dioate standard.

Polarity of C492 and diacylglycosyl-4,4′-diapolycopene 4,4′-diacid and their hydrolysis products was compared by TLC. In a system using toluene/ethyl acetate/methanol (50 : 30 : 20), C492 has an Rf value of 0·3 compared to 0·75 for diacylglycosyl-4,4′-diapolycopene 4,4′-diacid. This indicates that C492 is more polar than diacylglycosyl-4,4′-diapolycopene 4,4′-diacid. Upon hydrolysis of C492, the resulting product is ether soluble only after acidification to pH 3, indicating that it is a free carotenoid acid. Its polarity (Rf value 0·45) is between C492 and diacylglycosyl-4,4′-diapolycopene 4,4′-diacid. On HPLC, the hydrolysis product 3 of C492 labelled as peak 3 in Fig. 1(d) and also of diacylglycosyl-4,4′-diapolycopene 4,4′-diacid (Fig. 1e, peak 3) showed the same chromatographic behaviour. This acid could be identified with a reference compound as 4,4′-diapolycopene 4,4′-diacid (Fig. 1f) synthesized in E. coli and a commercial standard (data not shown). From these TLC/HPLC results, it can be concluded that C492 is an ester with an alcohol that increases the polarity of the acid, unlike the acylglycoside groups in diacylglycosyl-4,4′-diapolycopene 4,4′-diacid which decreases the polarity because of the attached fatty acids. Quantification of the carotenoid contents of SF241 and GB1 gave 176·6 ± 11·8 and 146·2 ± 39·1 μg g−1 dry weight of C492, respectively.

HPLC mass spectroscopy of pathway product

It was attempted to acquire the MS spectra of the unhydrolysed parent carotenoid (C492). Good HPLC separation and photo diode array (PDA) detection were obtained. However, it was found that this component was not amenable for MS analysis, either directly on-line following HPLC separation or off-line after purification by HPLC/TLC. In addition, both direct infusion mass spectrometry (DiMS) and atmospheric pressure chemical ionization (APCI) or electrospray ionization (ESI) ionization in positive and negative modes were also tested with no success. Despite the different attempts to pre-purify the C492 using solid phase extraction cartridges (normal and weak anion-exchange phases) or by RP-HPLC, the identification of this carotenoid by MS remains unresolved. We presume that either the chemical nature of this compound makes it unable for APCI and ESI ionization or unstable under those conditions, or that co-eluting matrix components induce signal suppression preventing the detection of the C492 in the MS. Nonetheless, the transmethylation products of C492 were easily detected and identified by HPLC-MS (Fig. 3). Upon treatment of C492 with 6% KOH in methanol, the resulting products were the mono- and the dimethyl esters of 4,4′-diapolycopene-4,4′-diacid. This is indicated by the molecular masses m/z of 489 and 475, respectively. In addition, individual fragments corresponding to the loss of water and methanol could be found for the monomethyl ester with m/z of 457 and 443, respectively, together with a combination of both losses with m/z of 425. Owing to its two ester groups, the dimethyl esters of 4,4′-diapolycopene-4,4′-diacid yield a fragment by the loss of one (m/z of 457) and of two methanol molecules (m/z of 425). The reference compound 4,4′-diapolycopene-4,4′-diacid was also subjected as a standard to LC/MS analysis. Besides its molecular mass m/z of 461, it yielded the same fragment (m/z of 457) by the loss of water as did the monomethyl 4,4′-diapolycopene-4,4′-diacid.

Figure 3.

Mass spectra of the products of C492 from Bacillus firmus GB1 obtained after transmethylation with 6% KOH in methanol. Corresponding structures and fragmentation patterns are provided according to LC/MS analysis.

Diphenylamine treatment to elucidate the early steps of biosynthesis

Diphenylamine is an inhibitor of bacterial phytoene desaturases (Sandmann and Fraser 1993). Its application in low doses facilitates the accumulation of different intermediates of the desaturation pathway (Fig. 4). Application of 2·5 μmol l−1 DPA to SF241 resulted in the detection of 4,4′-diaponeurosporene (retention time 25·3 min) and 4,4′-diapolycopene (retention time 28·8 min) in trace 4A. Increases in the DPA concentration to 5 μmol l−1 yielded 4,4′-diapophytofluene (retention time 19·0 min) and 4,4′-diapo-ζ-carotene (retention time 24·8 min) in trace 4B and 4,4′-diapophytoene (retention time 14·6 min) in trace 4C. Moreover, HPLC analysis of the colourless mutant SF241-W2 showed that it exclusively contained 4,4′-diapophytoene as only carotenoid (Fig. 4d). These carotenes were identified by their corresponding absorption spectra shown in Fig. 2 and by the same retention times of the appropriate standards synthesized in E. coli (Fig. 4e).

Figure 4.

Precursor carotenoids in Bacillus strain SF241 treated with diphenylamine 2·5 μmol l−1 (a) and 5 μmol l−1 (b and c) and of mutant SF241-W2 (d). Trace (e) separates a mixture of 4,4′-diapophytoene (DPhy), 4,4′-diapophytofluene (DPf), 4,4′-diapo-ζ-carotenes (DZC), 4,4′-diapolycopene (DLy) produced in Escherichia coli and trace (f) of 4,4′-diapolycopene-4,4′-dial (DLyD) as reference compounds. DNeu is 4,4′-diaponeurosporene.

Mutant SF241-W2 is colourless. HPLC analysis showed that it exclusively contained 4,4′-diapophytoene as only carotenoid (Fig. 4d). An unusual carotenoid was also found in trace A of Fig. 5 with a retention time of 32·7 min. and an absorbance spectrum of 472, 505 and 535 nm (Fig. 2g). The same retention time and an identical spectrum were obtained for the reference compound 4,4′-diapolycopene-4,4′-dial (Fig. 4f).

Figure 5.

Intermediates of the carotenogenic pathway of Bacillus firmus strains GB1 and SF241 leading to a polar 4,4′-diapolycopene-4,4′-dioic acid ester as end product.

Antioxidative property of C492-containing Bacillus firmus

The peroxidation experiments were carried out with GB1 and GB1 treated with 50 μmol l−1 of DPA which completely inhibited the formation of coloured carotenoids. This treatment was the only choice because the white mutant SF241-W2 exhibited poor growth with a rate <50% of the wild type. Growth rates were determined as increase in OD600 per h from the exponential phase during cultivation for 20 h (Table 1). The growth rate for GB1 treated with DPA was 77% compared to the culture without DPA. The applied concentrations are slightly toxic but lower DPA concentrations were not sufficient for the complete inhibition of coloured carotenoid synthesis. To both cultures, cumene peroxide that causes oxidative stress was added. Like any peroxide which, for example, is generated by photosensitized reactions, it can start radical chain reactions. This leads to the degradation of cell components and impairment of growth (Krinski and Yeumb 2003). Cumene peroxide affected the growth of the pigmented culture to a similar extent as DPA application. However, when carotenoid formation was inhibited by DPA, the growth rate slowed dramatically (0·1% of untreated control). This finding demonstrates that the 4,4′-diapolycopene-4,4′-diacid derivative in B. firmus GB1 and SF241 can protect the cells from peroxide-mediated oxidation reactions.

Table 1. Growth of Bacillus firmus strain GB1 in the presence of 250 μmol l−1 of cumene hydroperoxide (CHP) and of 50 μmol l−1 diphenylamine (DPA)
 Growth rate (ΔOD600 per h 10−3)
  1. OD600 was determined from the exponential phase of a 20-h LB culture; treatment of DPA inhibited the formation of coloured carotenoids completely.

GB182·4 ± 10·4
GB1 + DPA63·4 ± 8·8
GB1 + CHP56·0 ± 6·1
GB1 + DPA + CHP1·7 ± 0·3


The pigmentation of Bacillus species ranges from yellow via orange to red. Although their spectra resembled those of carotenoids, several of their structures have not been identified to date (Khaneja et al. 2010). Among them is a red carotenoid found in B. firmus isolates. Its nature and the carotenoid pathway of B. firmus have been established (Fig. 5) by HPLC-PDA-MS analysis of the end product as well as by the accumulation and identification of a series of 4,4′-diapocarotenes and 4,4′-diapolycopene-4,4′-dial intermediates after inhibition by DPA (Fig. 4). The presence of 4,4′-diapophytoene in mutant SF241-W2 indicates complete inactivation of 4,4′-diapophytoene desaturase. Its synthesis in the wild type from farnesyl pyrophosphate by the crtM gene product is the starting reaction of a 4,4′-diapo carotenoid pathway. This initial reaction determines the C30 pathway found in a few other bacteria (Pelz et al. 2005; Tao et al. 2005; Perez-Fons et al. 2011). In the following desaturation steps, 4,4′-diapophytoene is converted via 4,4′-diapophytofluene, 4,4′-diapo-ζ-carotene and 4,4′-diaponeurosporene to 4,4′-diapolycopene by CrtN, which is closely related to CrtI C40 phytoene desaturase (Raisig and Sandmann 2001). At this stage, the pathway diverts from the pathway in Bacillus indicus HU36, which branches towards glycosyl-4′-methyl-apolycopenoate fatty acid esters as end products (Perez-Fons et al. 2011). In B. firmus, oxidation of C4 and C4′ proceeds via 4,4′-diapolycopene-4,4′-dial to 4,4′-diapolycopene-4,4′-diacid (Fig. 5), which is a hydrolysis product of the final carotenoid of the pathway. DPA-dependent accumulation of 4,4′-diapolycopene-4,4′-dial indicates that oxidation of the aldehyde to the acid is catalysed by a CrtI/CrtN-related enzyme (Raisig and Sandmann 2001). All genes for the conversion of diapophytoene to 4,4′-diapolycopene-4,4′-diacid have been cloned from Methylomonas strain 16a (Tao et al. 2005). The end product of the pathway in B. firmus is a 4,4′-diapolycopene-4,4′-diacid derivative as shown by HPLC-MS, TLC and chemical characterization. The carotenoid synthesized by B. firmus via the 4,4′-diapocarotenoid pathway is related to the carotenoids found in M. rhodium and R. squalenifaciens, which are mono- and diglucosyl (Kleinig et al. 1979) or monoxylosyl esters (Shindo et al. 2007) with a fatty acid of different chain length esterified to C-2, C-4 or C-6 of the sugar moiety (Shindo et al. 2008b). However, carotenoid in the B. firmus resembles slightly to the acetyl-4,4′-diapolycopene-4,4′-dioate identified recently in the Bacillus-related Sporosarcina aquimarina (Steiger et al. 2012). In vitro experiments already demonstrated the antioxidative potential of other diapolycopenedioic acid derivatives (Shindo et al. 2008a,b). We were able to show this antioxidative protection for the GB1 carotenoid in the cells of B. firmus (Table 1). In addition, mutation of 4,4′-diapophytoene desaturase as well as DPA treatment resulted in decreased growth (Table 1), which may be a consequence of inhibition of carotenogenesis. It has already been shown that glucose repression of carotenogenesis in Bacilli including B. firmus decreased the cell growth only in pigmented strains (Manzo et al. 2011). This is another indication that carotenoids may be involved in the stress response of coloured Bacilli. A series of 27 B. firmus isolated have been obtained from seawater (Rüger and Koploy 1980). Half of them were nonpigmented and the others exhibited orange or yellow pigmentations. The B. firmus strains that can synthesize the 4,4′-diapolycopene-4,4′-diacid derivative have the advantage of being protected from peroxidative reactions.


The authors are grateful for funding through the EU FP7 Colorspore project no. 207948.