Vol. 55, No. 12

Profiling and Quantitation of Bacterial Carotenoids by LiquidChromatography and Photodiode Array Detection

H. J. NELIS AND A. P. DE LEENHEER*

Laboratoria voor Medische Biochemie en voor Klinische Analyse, Harelbekestraat 72, B-9000 Ghent, Belgium

Received 11 May 1989/Accepted 14 September 1989

An analytical method for the profiling and quantitative determination of carotenoids in bacteria is described.Exhaustive extraction of the pigments from four selected bacterial strains required treatment of the cells withpotassium hydroxide or liquefied phenol or both before the addition of the extracting solvent (methanol or

diethyl ether). The carotenoids in the extracts were separated by nonaqueous reversed-phase liquid chroma-tography in conjunction with photodiode array absorption detection. The identity of a peak was considereddefinitive only when both its retention time and absorption spectrum, before and after chemical reactions,matched those of a reference component. In the absence of the latter, most peaks could be tentatively identified.Two examples illustrate how in the analysis of pigmented bacteria errors may result from using nonchromato-graphic procedures or liquid chromatographic methods lacking sufficient criteria for peak identification.Carotenoids of interest were determined quantitatively when the authentic reference substance was availableor, alternatively, were determined semiquantitatively.

Carotenoid-containing bacteria and fungi are of consider-able academic (11) and industrial (20) interest. Much litera-ture has appeared on the isolation, purification, and identi-fication of microbial carotenoids (11, 16). In contrast, fewreports have dealt with the routine small-scale determinationof these pigments in fungi and bacteria. Some recent studieson the biotechnological potential of carotenoid-producingmicroorganisms continue to use spectrophotometry as a toolto monitor improvements in pigment yield in relation to,e.g., environmental factors (7, 26, 27). However, nonchro-matographic procedures fail to detect qualitative changes inthe pigment pattern and are also subject to interferences,e.g., from colored medium components adsorbed on thecells (26).

Despite the popularity of high-performance liquid chroma-tography in the carotenoid area (29), the application of thistechnique to the analysis of pigmented bacteria and fungi hasbeen limited so far. Of the few existing methods (10, 12, 15,19, 28), only three involve a quantitative determination, one

of fungal (19) and two of bacterial (15, 28) carotenoids.However, none addresses the issue of exhaustive pigmentisolation. Most high-performance liquid chromatographysystems reported for the separation of bacterial and fungalcarotenoids employ normal-phase chromatography (10, 12,15), which is more subject to variability than reversed-phasechromatography. Peak identification is sometimes onlybased on one criterion, i.e., cochromatography of unknownswith authentic reference substances (19). More often, ab-sorption spectra of collected peaks are also determined toconfirm their identity, either off-line in a spectrophotometer(15, 28) or on-line by a manual wavelength-ratioing proce-dure (10). A more modern approach would consist of record-ing the absorption spectra with the aid of a photodiode arraydetector. The potential of this powerful technique for theidentification of carotenoids in general has been outlined byRuedi (29). Applications include the pigment profiling ofArtemia (24), algae (17), orange juice (8), vegetables (13, 14),and palm oil (25) but so far not of bacteria and fungi.

This report describes the application of nonaqueous re-

* Corresponding author.

versed-phase chromatography in conjunction with photo-diode array detection and improved exhaustive extractionprocedures to the separation and identification of carot-enoids in bacteria. Pigments of interest were quantitatedwith authentic reference components or, in the absence ofthe latter, were determined semiquantitatively.

(This work was presented in part at the 8th InternationalSymposium on Carotenoids [Boston, Mass., 27 to 31 July1987], abstr. 9).

MATERIALS AND METHODS

Chemicals and reagents. Canthaxanthin, echinenone, iso-zeaxanthin, and isocryptoxanthin were gifts from Hoffmann-La Roche (Basel, Switzerland). p-Carotene was obtainedfrom Fluka (Buchs, Switzerland); lycopene was from SigmaChemical Co. (St. Louis, Mo.). Acetonitrile (Janssen Chim-ica, Beerse, Belgium) and methanol and dichloromethane(both from Hoechst AG, Frankfurt, Federal Republic ofGermany) were chemically pure. The latter two were redis-tilled in a spinning band apparatus (B/R Instruments, Pasa-dena, Md.). All other chemicals were analytical grade andpurchased from E. Merck AG (Darmstadt, Federal Republicof Germany) or Janssen Chimica. Bond Elut Alumina N(neutral) minicolumns (500 mg, 2.8 ml) were obtained fromAnalytichem (Harbor City, Calif.). Liquefied phenol was

prepared by adding water (8 parts) to molten phenol (92parts).

Bacterial strains and growth conditions. Brevibacterium sp.strain KY-4313 was a gift from T. Oka (Kyowa HakkoKogyo Co., Tokyo, Japan). It was maintained on brain heartinfusion agar (Oxoid Ltd., Basingstoke, United Kingdom).Mass cultures were carried out at 30°C in 500-ml shake flaskscontaining 100 ml of hydrocarbon medium as describedpreviously (36) or, alternatively, in brain heart infusion broth(21).Rhodobacter capsulatus (ATCC 23782) and Rhodomicro-

bium vannielii (ATCC 17100) were obtained from the Amer-ican Type Culture Collection (Rockville, Md.). Growthconditions for these strains have been described previously(6, 34, 35). When Rhodobacter capsulatus was grown non-

axenically in a continuous-flow photobioreactor, sodium

3065

APPLIED AND ENVIRONMENTAL MICROBIOLOGY, Dec. 1989, p. 3065-30710099-2240/89/123065-06$02.00/0Copyright © 1989, American Society for Microbiology

on April 7, 2020 by guest

http://aem.asm

.org/D

ownloaded from

3066 NELIS AND DE LEENHEER

molybdate and chloroxuron were added to reduce contami-nation (6).Micrococcus roseus (ATCC 516), donated by J. J. Cooney

(Boston, Mass.), or its equivalent CCM 839, originating fromthe Czechoslovak Collection of Microorganisms, Brno,Czechoslovakia), was grown according to the procedure ofCooney and Berry (2).

Extraction of carotenoids from bacterial cells. (i) Bacteriagrown in nonhydrocarbon medium. Slightly different extrac-tion procedures were used for Brevibacterium sp. strainKY-4313 on the one hand and Rhodobacter capsulatus,Rhodomicrobium vannielii, and M. roseus on the otherhand. The difference pertains to the treatment or nontreat-ment of the cells with KOH before treatment with liquefiedphenol.To about 10 mg of freeze-dried cells of Brevibacterium sp.

strain KY-4313 was added, under vigorous vortex mixing,0.3 ml of potassium hydroxide (60%, wt/vol), followed by 3ml of methanol. After centrifugation and isolation of thesupernatant, the same step was repeated once more. Theresidue was treated with 0.4 ml of liquefied phenol (vortexmixing for no longer than 30 s) and extracted with 4 ml ofdiethyl ether. The ethereal supernatant was combined withthe methanolic extracts. The internal standard (P-apo-8'-carotenal) was added, and the carotenoids were extractedafter the addition of 5 ml of sodium chloride (5%, wt/vol) and4 ml of diethyl ether. The ether layer was isolated, succes-

sively washed with potassium hydroxide (2%, wt/vol) andwater, dried over anhydrous sodium sulfate, and evaporatedto dryness under nitrogen. The residue was dissolved in 0.2to 2 ml of the chromatographic solvent (depending on theconcentration of the carotenoids), and a 100-,u sample was

injected on the liquid chromatographic column.For the isolation of the carotenoids from the three other

species, cells (about 10 mg) were directly treated with 0.4 mlof liquefied phenol and extracted with 3 ml of methanol. Ingeneral, two to three steps were required to ensure completebleaching of the residue. Potassium hydroxide (3 ml; 60%,wt/vol) was added to the combined extracts, and the mixturewas left at room temperature for 2 h, except for the extractof M. roseus, which was worked up immediately. After theaddition of 6 ml of sodium chloride (5%, wt/vol), the pig-ments were reextracted in 10 ml of diethyl ether. Theethereal extract was further treated as described above forBrevibacterium sp. strain KY-4313. Canthaxanthin and ,-carotene were used as internal standards for the semiquan-titative analysis of Rhodobacter capsulatus and Rhodomi-crobium vannielii, respectively.

(ii) Bacteria grown in hydrocarbon medium (Brevibacteriumsp. strain KY-4313). After growth of Brevibacterium sp.

strain KY-4313 in hydrocarbon medium, the fermentationbroth readily separated into a lower aqueous layer and a

hydrocarbon (paraffin) top layer containing the cells. Waterwas removed as completely as possible by vacuum aspira-tion and centrifugation. The resulting paste, correspondingto approximately 300 to 400 mg of dry cells, was extractedtwice with 15 ml of a mixture of diethyl ether-methanol (1:1,vol/vol). The cells were subsequently treated with 1.5 ml ofliquefied phenol and extracted with 15 ml of diethyl etheruntil complete bleaching of the residue. All extracts were

combined in a 100-ml volumetric flask brought to volumewith diethyl ether. A 2-ml sample was supplemented with theinternal standard (,-apo-8'-carotenal), 1 ml of methanol was

added, and the solution was concentrated to about 1 mlunder nitrogen to remove all the diethyl ether. Hexane (5 ml)and 4 ml of potassium hydroxide (2%, wt/vol) were added,

and the mixture was shaken to extract the carotenoids. Thehexane layer was dried over sodium sulfate and applied ontop of a Bond Elut Alumina N minicolumn preequilibratedwith hexane. After being washed with hexane to remove allparaffin, the retained carotenoids were eluted with diethylether. The eluate was evaporated to dryness, and the residuewas reconstituted with 2 ml of the chromatographic solvent.A 100-.lI sample was injected on the liquid chromatographiccolumn.

(iii) Recovery of pigments. The recovery o,f pigments wasconsidered to be complete when, after repeated treatment ofthe cells with the extracting agent(s), the cellular residue andthe supernatant had become visually colorless and when atthe same time the use of alternative extraction procedures nolonger resulted in a detectable chromatographic peak. Tocompare the efficiency of different extraction procedures, wedetermined the pigments in the extracts either quantitatively(see below) or semiquantitatively (i.e., by spectropho-tometry). The highest value obtained was arbitrarily set at100%, and all other results were referred to this maximumvalue.

Liquid chromatography. (i) Apparatus and chromato-graphic conditions. The high-performance liquid chromatog-raphy system consisted of a pump (5020; Varian, Palo Alto,Calif.), an injector (N 60; Valco, Houston, Tex.) fitted witha 100-,ul loop, a photodiode array detector, and an integrator(SP 4100; Spectra-Physics, San Jose, Calif.). A commercial5-,um Zorbax ODS column (Du Pont Co., Wilmington, Del.)was eluted with mixtures of acetonitrile-methanol-dichloro-methane, either 40:50:10 (vol/vol) (eluent A) or 70:15:15(voUvol) (eluent B), at a flow rate of 1 ml/min. The detectorsignal was monitored at 470 nm.For peak identification, the retention times of unknown

peaks were compared with those of reference substances.The absorption spectrum of each peak was recorded andmemorized with the aid of an HP 1040 A photodiode arraydetector, connected to an HP 9121 dual disk drive and an HP7470 A plotter (all from Hewlett-Packard Co., Palo Alto,Calif.).

(ii) Quantitation of carotenoids. When authentic referencesubstances were at hand (e.g., canthaxanthin, echinenone,n-carotene), quantitation was based on the measurement ofpeak height ratios (compound of interest versus internalstandard). Calibration curves were constructed by plottingpeak height ratios versus known amounts of the compoundto be determined.

In the absence of a reference substance, a semiquantita-tive determination was done with a structural analog of the(tentatively identified) compound of interest. In this case,peak area ratios (compound of interest versus internal stan-dard) were used which were corrected for differences in themolar absorption coefficient (literature values [4]) of thecompound of interest and the analog, respectively.

RESULTS

Extraction of carotenoids from bacteria. (i) Bacteria grownin nonhydrocarbon medium. Vigorous mixing of cells of thefour selected bacterial species with various organic solventsled to incomplete recovery of carotenoids (Table 1). Theresidue was thoroughly bleached by treating the cells withliquefied phenol before the addition of the extracting solvent(Table 1). The efficiency of the phenol approach for Brevi-bacterium sp. strain KY-4313 is evidenced from a compari-son with another presumably exhaustive extraction methodwith boiling methanol, which was also found to result in a

APPL. ENVIRON. MICROBIOL.

on April 7, 2020 by guest

http://aem.asm

.org/D

ownloaded from

DETERMINATION OF CAROTENOIDS IN BACTERIA 3067

TABLE 1. Relative effectiveness of different solvents for extraction of carotenoids from bacteriaa

% Pigment recovery from:Pretreatment Extracting

of cells solvent Brevibacterium sp. Rhodobacter Rhodomicrobium Micrococcusstrain KY-4313 capsulatus vannielii roseus

None CH30H 77.7 46.5 9.2None Acetone 59.4 7.9 0 22.4KOH (60%, wt/vol) CH30H 100.0 73.4 27.5 23.8Liquefied phenol CH30H 91.4 100.0 100.0 100.0

a Vortex-type mixing of freeze-dried cells with an excess of solvent.

colorless residue. A repeated quantitative determination ofcanthaxanthin in a sample of freeze-dried cells of Brevibac-terium sp. strain KY-4313 yielded 89.5 + 2.3 ,ug/g (n = 7)with boiling methanol versus 99.0 + 1.1 ,ug/g (n = 8) withphenol. The difference between the two values was statisti-cally highly significant (P < 0.001). However, potassiumhydroxide was still slightly superior to phenol in promotingthe extraction of canthaxanthin from Brevibacterium sp.

strain KY-4313 (Table 1) but, surprisingly, failed to liberatemost echinenone and 1-carotene, two biosynthetic precur-

sors of canthaxanthin, from the cells. After two extractionswith KOH-methanol, all canthaxanthin had been recoveredbut as much as 50% of the echinenone and 90% of the1-carotene remained in the residue. Both pigments could beefficiently extracted with phenol-methanol or phenol-diethylether. KOH was also less effective than phenol in promotingthe extraction of carotenoids from the three other speciesstudied (Table 1).

Carotenoids could be readily reextracted in diethyl etherfrom alkalinized phenol-methanol-water mixtures. M. roseus

was an exception in that the bulk of the pigments remainedin the aqueous layer. No phenol partitioned into the etherphase.The stability of carotenoids in the presence of phenol was

variable. A 20-min exposure at room temperature of syn-

thetic 1-carotene to pure liquefied phenol in the absence ofmethanol or diethyl ether resulted in 90% degradation (13 to30% within 1 min), as opposed to only a 3% loss forcanthaxanthin (no loss within 1 min). However, the break-down of 1-carotene was arrested upon the addition of the

1

0 8 MIN.FIG. 1. Liquid chromatographic profile of an extract of Brevi-

bacterium sp. strain KY-4313. Eluent A. Peak identities: 1, all-trans-canthaxanthin; 2, cis-canthaxanthins; 3, P-apo-8'-carotenal(internal standard); 4, echinenone; 5, ,8-carotene.

extracting solvent, after which the compound remainedstable for at least 30 min at 0°C.

(ii) Bacteria (Brevibacterium sp. strain KY-4313) grown inhydrocarbon medium. Treatment of hydrocarbon-grownBrevibacterium sp. strain KY-4313 cells with organic sol-vents to isolate the carotenoids resulted in coextraction ofresidual hydrocarbon adsorbed onto the cells. Canthaxan-thin, echinenone, and the internal standard were separatedfrom the coextracted paraffin on Bond Elut Alumina Nminicolumns. No oily residue was obtained after evapora-tion of the eluted fraction, indicating that all paraffin hadbeen removed during the washing step with hexane. Thereproducibility of this cleanup step was excellent. A repeti-tive application of 2-ml samples of an extract of Brevibacte-rium sp. strain KY-4313 in hexane (mean concentration, 582pug/liter) resulted upon high-performance liquid chromato-graphic determination in a relative standard deviation of1.8% (n = 9) for canthaxanthin.

Liquid chromatography. (i) Pigment profiling in bacterialextracts. The efficiency of nonaqueous reversed-phase liquidchromatography on Zorbax ODS (22) for the separation ofbacterial carotenoids can be ascertained from Fig. 1 through4.

All major peaks in Fig. 1 through 3 could be tentatively ordefinitively identified on the basis of their absorption spectraand comparison with reference components. The principalpigments of Brevibacterium sp. strain KY-4313 were foundto be canthaxanthin (peak 1), echinenone (peak 4), and13-carotene (peak 5) (Fig. 1). The natural compounds coe-luted with their respective synthetic equivalents, and their

A47O A8 5 C

1

67

3

0 48 0 48 0 4 8MKFIG. 2. Liquid chromatographic profiles of extracts of Rhodo-

bacter capsulatus grown under different conditions ([A] iron excess,grown nonaxenically; [B] iron excess, grown axenically; [C] irondeficiency). Eluent B. Peak identities (tentative): 1, spheroidenone;2, cis isomers of spheroidenone; 3, spheroidene; 4, 5, 6, and 7,unidentified.

tL13

4

.ji 5

VOL. 55, 1989

on April 7, 2020 by guest

http://aem.asm

.org/D

ownloaded from

3068 NELIS AND DE LEENHEER

230 330 430 530 nmFIG. 5. Absorption spectra of spheroidenone ( ), spheroi-

dene (-- -), and peak 4 from Fig. 2. (....).

FIG. 3. Liquid chromatographic profile of an extract of Rhodo-microbium vannielii. Eluent B. See Table 2 for tentative peakidentification.

absorption spectra closely matched (all-trans-canthaxanthin,Xmax of 476 nm; echinenone, Xmax of 465 nm; ,-carotene,Xmax of [430], 453, and 481 nm). Peak 2 was assigned to amixture of cis-canthaxanthins, as indicated by its absorptionspectrum (Xmax of 363 and 468 nm) and its coelution withsynthetic cis-canthaxanthin (24). After reduction on a micro-scale of the ketocarotenoids all-trans-canthaxanthin andechinenone with lithium aluminum hydride (H. J. Nelis andA. P. De Leenheer, manuscript in preparation), two morepolar peaks were obtained which had the same retentiontime and absorption spectra ('max of 453 and 478 nm in bothcases) as the predicted reaction products, isozeaxanthin andisocryptoxanthin, respectively.The reported occurrence of spheroidene and spheroide-

none in Rhodobacter capsulatus (31) was tentatively con-firmed on the basis of the elution position and the absorptionspectra (Fig. 5) of the two peaks in Fig. 2A. Unfortunately,no synthetic reference components could be acquired forthese carotenoids. Because of the possible biotechnologicalinterest of spheroidenone, attempts were made to increasethe cellular level of this valuable red pigment by changing thegrowth conditions of the organism. Some of these attemptswere successful in that spheroidenone (plus some cis iso-mers) indeed came to supersede spheroidene (Fig. 2B), a

A470

0 4 8 0 4 MIN.

FIG. 4. (A) Liquid chromatographic profile of an extract of M.roseus (CCM 839). (B) Chromatogram of synthetic all-trans-can-thaxanthin. Eluent B. All peaks in panel A are unidentified. Peak 2is all-trans-canthaxanthin.

change which was reflected in a deep red appearance of thecells. Iron deficiency in the growth medium likewise led tointensely red cells but yielded a totally different carotenoidprofile (Fig. 2C). All peaks now displayed spectra quitedifferent from the typical ketocarotenoid spectrum of sphe-roidenone. The absorption spectrum of peak 4 is shown inFig. 5. Absorption maxima of the other peaks were 448, 475,and 506 nm (peaks 5 and 6) and 418, 443, and 473 nm (peak7). These unknown peaks were not further characterized.

Figure 3 and Table 2 show how peaks in a more complexcarotenoid profile originating from Rhodomicrobium van-

nielii could still be tentatively identified in the absence ofreference compounds. Only one peak (peak 8) was defini-tively characterized (as lycopene) according to the criteriaoutlined above. Identification of the other peaks was tenta-tive based on their absorption spectra, retention behavior,and established pigment patterns reported in the literature(1, 31) (Table 2). The absorption spectra of the major peaksclosely matched the literature data for the various carot-enoids reportedly occurring in Rhodomicrobium vannielii(Table 2). Moreover, the resulting elution sequence wasconsistent with the relative polarity of the peaks with refer-ence to lycopene, as expressed by their respective partitioncoefficients (hexane-95% methanol) (Table 2) (9).

In contrast, the experimentally obtained pigment patternof M. roseus did not agree at all with the one repeatedlyreported by one particular research group (2, 3, 32, 33, 38).The claim that this species contains canthaxanthin as thepredominating carotenoid could not be substantiated. Only asmall fraction of the orange color of the cells proved extract-able with methanol, the solvent recommended by Cooneyand co-workers (2, 3, 38). Exhaustive pigment extractionwas achieved with phenol-methanol, but upon alkalinizationand addition of diethyl ether, the bulk of the pigmentaccumulated at the interface of the two phases. Attempts tochromatograph these acidic carotenoids have been unsuc-cessful so far. The occurrence of astacene was ruled out byusing a liquid chromatographic system specially designed forthis compound (23). The chromatographic profile of thecarotenoids that did partition into diethyl ether is shown inFig. 4. It was similar to the one obtained after (incomplete)extraction with pure methanol, indicating that the phenolhad left the pigments essentially untouched. Although peak 1(Fig. 4A) coeluted with all-trans-canthaxanthin (Fig. 4B), itsabsorption spectrum was totally inconsistent with this struc-tural assignment (Fig. 6). None of the other peaks displayeda symmetrical absorption spectrum without fine structure,typical of ketocarotenoids.

(ii) Quantitative determination. Canthaxanthin (all-transplus cis) was quantitatively determined in Brevibacterium

A B1 2

L!f I1~- -II I I . .

APPL. ENVIRON. MICROBIOL.

on April 7, 2020 by guest

http://aem.asm

.org/D

ownloaded from

DETERMINATION OF CAROTENOIDS IN BACTERIA 3069

230 330 430 530 nm

FIG. 6. Absorption spectra of peak 1 from Fig. 4A ( ) and ofauthentic all-trans-canthaxanthin (peak 2, Fig. 4B) (... ).

sp. strain KY-4313 by using the authentic substances forcalibration and P-apo-8'-carotenal as an internal standard.Repetitive analysis of freeze-dried cells grown in nonhydro-carbon medium yielded a relative standard deviation of 4.0%(x = 191 ,ug/g, n = 10). For the determination of canthaxan-thin in hydrocarbon-grown Brevibacterium sp. strain KY-4313, a relative standard deviation of 3.2% (x = 355 ,ug/g, n= 8) was obtained. The minimum detectable quantities(absolute amount injected) of carotenoids in the presentsystem are in the low nanogram range. For example, given avalue of 1 ng of canthaxanthin, this would correspond forBrevibacterium sp. strain KY-4313 to a sensitivity, in termsof concentration, of 0.2 ,ug/g (in case the residue is dissolvedin 0.2 ml).The main pigments of Rhodobacter capsulatus and Rho-

domicrobium vannielii were determined semiquantitatively.Lycopene was used as a standard for calibration because ofits structural analogy to the (tentatively identified) carot-enoids of interest. Owing to their totally unknown nature,the pigments of M. roseus could not be quantitated.

DISCUSSION

The majority of methods for the extraction of carotenoidsfrom bacteria rely on the use of acetone or methanol (1, 3, 7,12, 15, 28). However, these polar organic solvents often failto liberate all pigment (5, 18, 37), which was confirmed byour own observations. Several investigators included suchadditives as trichloroacetic acid (5) or potassium hydroxide(37) to ensure exhaustive pigment extraction. One reportrecommends a phenol-glycerol (3:1) mixture as an extractingagent (18). In our hands, pure liquefied phenol was found torelease the intracellular carotenoids nearly instantaneously,

thus making them amenable to a subsequent transfer intomethanol or diethyl ether. In a second step, diethyl etheralso served to reextract the pigments, leaving the phenol inthe alkalinized water-methanol phase. At the same time, thealkali destroyed any chlorophyll that might have beenpresent in extracts from photosynthetic bacteria (Rhodobac-ter capsulatus and Rhodomicrobium vannielii). Althoughliquefied phenol appears to act as a universal extraction-promoting agent, the instability of certain carotenoids inphenol may limit its general use. However, degradation can

be largely overcome by minimizing the exposure of thepigment to phenol (c30 s), i.e., by adding the extractingsolvent as soon as possible.Hydrocarbon-grown bacteria pose additional extraction

problems because these cells are coated with a paraffin layertightly adhering to their hydrophobic surface. A preliminaryremoval of the hydrocarbon with organic solvents provednot feasible without afflicting a concomitant substantial lossof pigment. The final procedure devised for Brevibacteriumsp. strain KY-4313 combined hydrocarbon removal withexhaustive carotenoid extraction. The hydrocarbon was

subsequently eliminated from the extract by chromatogra-phy on alumina. This step was necessary to avoid a final oilyresidue that could not be solubilized in the chromatographicsolvent before injection and to protect the reversed-phasecolumn.Nonaqueous reversed-phase chromatography (22) has

proved to be a reliable technique for the separation ofbacterial carotenoids. The inadequacy of a nonchromato-graphic analytical method was clearly evidenced from theresults obtained with Rhodobacter capsulatus: a spectro-photometric determination of the red pigment in iron-defi-cient cells would have readily led to the wrong conclusion ofan increased cellular spheroidenone level. However, theexample of M. roseus demonstrated that even with an

efficient chromatographic separation, pitfalls may arise fromusing only one criterion (retention time) for the identificationof peaks in bacterial extracts. The indispensable secondcriterion is provided by the absorption spectrum of the peak,recorded with the aid of the photodiode array detector. If theabsorption spectrum of an unknown peak agreed with that ofa reference component displaying the same retention, iden-tification was considered definitive, particularly when thisagreement persisted after chemical reactions affecting spe-cific functional groups (e.g., reduction of ketocarotenoids tohydroxy derivatives in extracts from Brevibacterium sp.strain KY-4313). The experimentally determined pigment

TABLE 2. Tentative identification of major carotenoid peaks in extracts of Rhodomicrobium vannielii (see Fig. 3)

Tentative structure elucidationePneoak Identity No. of double Functional

bonds group(s)

2 460, 487, 519 460, 488, 522 66:34 Rhodovibrin 12 OH, OCH33 448, 475, 508 448, 474, 506 76:24 Rhodopin 11 OH4 470, 497, 532 468, 499, 533 84:16 Spirilloxanthin 13 OCH3, OCH37 461, 489, 521 460, 485, 520 94:6 Anhydrorhodovibrin 12 OCH38 448, 475, 508 448, 474, 505 100:0 Lycopenef 119 418, 443, 473 416, 440, 470 100:0 Neurosporene 9

a Peaks 1, 5, and 6 were unidentified; peak 10 is a-carotene (internal standard).b Experimentally determined in the chromatographic solvent CH2CI2-CH30H-CH3CN (15:15:70, vol/vol).c Literature data (4): absorption maxima in acetone, except for neurosporene (ethanol).d Partition coefficient hexane-95% methanol (9).e With the reported pigment pattern (1, 11, 31) as a guideline.f Definitive identification versus reference substance.

VOL. 55, 1989

on April 7, 2020 by guest

http://aem.asm

.org/D

ownloaded from

3070 NELIS AND DE LEENHEER

profile of Brevibacterium sp. strain KY-4313 agreed with theone described previously (30). In the absence of referencecomponents, a tentative structural assignment is usuallypossible if the pigment pattern has been reported in theliterature (Rhodobacter capsulatus and Rhodomicrobiumvannielii). The absorption spectra alone of the pigments ofM. roseus provided insufficient information to permit atentative structure determination without the aid of moresophisticated spectrometric techniques (mass spectrometry,nuclear magnetic resonance).As far as quantitation of bacterial carotenoids is con-

cerned, the analytical approach described meets the require-ments of a routine micromethod. Only milligram amounts ofsample are required, and the simplicity of the extractionprocedure ensures a high sample throughput. Therefore, themethod is particularly suitable to monitor efforts to increasethe yield of bacterial carotenoids as part of biotechnologicalprojects, which usually require the analysis of large numbersof samples. Such quantitative analyses are now routinelydone in connection with studies of the potential valorizationof Brevibacterium sp. strain KY-4313 (21) and Rhodobactercapsulatus (6) as natural sources of ketocarotenoids.

ACKNOWLEDGMENTS

This work was supported by FGWO contracts 3.0048.86 and3.0046.87 (Belgian National Foundation for Scientific Re.earch).H.J.N. acknowledges his permanent position of research associatefrom the Belgian National Foundation for Scientific Research(NFWD).We are indebted to W. Verstraete (Faculty of Agricultural Sci-

ences, State University of Ghent) for donating samples of photosyn-thetic bacteria. The excellent technical assistance of K. Van denHaesevelde is gratefully acknowledged.

LITERATURE CITED1. Britton, G., R. K. Singh, T. W. Goodwin, and A. Ben-Aziz. 1975.

The carotenoids of Rhodomicrobium vannielii (Rhodospiril-laceae) and the effect of diphenylamine on the carotenoidcomposition. Phytochemistry 14:2427-2433.

2. Cooney, J. J., and R. A. Berry. 1981. Inhibition of carotenoidsynthesis in Micrococcus roseus. Can. J. Microbiol. 27:421-425.

3. Cooney, J. J., H. W. Marks, and A. M. Smith. 1966. Isolationand identification of canthaxanthin from Micrococcus roseus. J.Bacteriol. 92:342-345.

4. De Ritter, E., and A. E. Purcell. 1981. Carotenoid analyticalmethods, p. 815-923. In J. C. Bauernfeind (ed.), Carotenoids ascolorants and vitamin A precursors. Academic Press, Inc.,Orlando, Fla.

5. Downs, J., and D. E. F. Harrison. 1974. Studies on the produc-tion of pink pigment in Pseudomonas extorquens NCIB 9399growing in continuous culture. J. Appl. Bacteriol. 37:65-74.

6. Driessens, K., J. Liessens, S. Masduki, W. Verstraete, H. Nelis,and A. De Leenheer. 1987. Production of Rhodobacter capsula-tus ATCC 23782 with short residence time in a continuous flowphotobioreactor. Process Biochem. 22:160-164.

7. Dumenil, G., M. Laget, A. Cremieux, M. Millon, and M.Brousse. 1983. Etude des conditions de pigmentation d'unesouche bacterienne methylotrophe facultative: Corynebacte-rium sp. XG. Ann. Pharm. Fr. 41:427-435.

8. Fisher, J. F., and R. L. Rouseff. 1986. Solid-phase extractionand HPLC determination of 13-cryptoxanthin and a- andcarotene in orange juice. J. Agric. Food Chem. 34:985-989.

9. Foppen, F. H. 1971. Tables for the identification of carotenoidpigments. Chromatogr. Rev. 14:133-298.

10. Gillan, F. T., and R. B. Johns. 1983. Normal-phase HPLCanalysis of microbial carotenoids and neutral lipids. J. Chro-matogr. Sci. 21:34-38.

11. Goodwin, T. W. 1980. The biochemistry of the carotenoids, vol.1, plants, 2nd ed., p. 257-345. Chapman & Hall, Ltd., London.

12. Kester, A. S., and R. E. Thompson. 1984. Computer-optimizednormal-phase high-performance liquid chromatographic separa-tion of Corynebacterium poinsettiae carotenoids. J. Chro-matogr. 310:372-378.

13. Khachik, F., and G. R. Beecher. 1987. Application of a C-4513-carotene as an internal standard for the quantification ofcarotenoids in yellow/orange vegetables by liquid chromatogra-phy. J. Agric. Food Chem. 35:732-738.

14. Khachik, F., G. R. Beecher, and N. F. Whittaker. 1986. Sepa-ration, identification and quantification of the major carotenoidand chlorophyll constituents in extracts of several green vege-tables by liquid chromatography. J. Agric. Food Chem. 34:603-616.

15. Korthals, H. J., and C. L. M. Steenbergen. 1985. Separation andquantification of pigments from natural phototrophic microbialpopulations. FEMS Microbiol. Ecol. 31:177-185.

16. Liaaen-Jensen, S. 1978. Chemistry of carotenoid pigments, p.233-247. In R. K. Clayton and W. R. Sistrom (ed.), The photo-synthetic bacteria. Plenum Publishing Corp., New York.

17. Mantoura, R. F. C., and C. A. Llewellyn. 1984. Trace enrich-ment of marine algal pigments for use with HPLC-diode arrayspectroscopy. J. High Resol. Chromatogr. Chromatogr. Com-mun. 7:632-635.

18. Merritt, K., and N. J. Jacobs. 1978. Characterization andincidence of pigment production by human clinical group Bstreptococci. J. Clin. Microbiol. 8:105-107.

19. Nam, H. S., S. Y. Cho, and J. S. Rhee. 1988. High-performanceliquid chromatographic analysis of major carotenoids fromRhodotorula glutinis. J. Chromatogr. 448:445-447.

20. Nelis, H. J., and A. P. De Leenheer. 1989. Microbial productionof carotenoids other than 1-carotene, p. 43-80. In E. J. Van-damme (ed.), Biotechnology of vitamins, pigments and growthfactors. Elsevier Applied Science, London.

21. Nelis, H. J., and A. P. De Leenheer. 1989. Reinvestigation ofBrevibacterium sp. strain KY-4313 as a source of canthaxan-thin. Appl. Environ. Microbiol. 55:2505-2510.

22. Nelis, H. J. C. F., and A. P. De Leenheer. 1983. Isocraticnonaqueous reversed-phase liquid chromatography of carot-enoids. Anal. Chem. 55:270-275.

23. Nelis, H. J. C. F., and A. P. De Leenheer. 1988. Reversed-phaseliquid chromatography of astacene. J. Chromatogr. 452:535-542.

24. Nelis, H. J. C. F., P. Lavens, L. Moens, P. Sorgeloos, J. A.Jonckheere, G. R. Criel, and A. P. De Leenheer. 1984. Cis-canthaxanthins: unusual carotenoids in the eggs and the repro-ductive system of female brine shrimp Artemia. J. Biol. Chem.259:6063-6066.

25. Ng, J. H., and B. Tan. 1988. Analysis of palm oil carotenoids byHPLC with diode-array detection. J. Chromatogr. Sci. 26:463-469.

26. Okagbue, R. N., and M. J. Lewis. 1984. Use of alfalfa residualjuice as a substrate for propagation of the red yeast Phaffiarhodozyma. Appl. Microbiol. Biotechnol. 20:33-39.

27. Okagbue, R. N., and M. J. Lewis. 1985. Influence of mixedculture conditions on yeast-wall hydrolytic activity of Bacilluscirculans WL-12 and on extractability of astaxanthin from theyeast Phaffia rhodozyma. J. Appl. Bacteriol. 59:243-255.

28. Palmisano, A. C., S. E. Cronin, and D. J. Des Marais. 1988.Analysis of lipophilic pigments from a phototrophic microbialmat community by high performance liquid chromatography. J.Microbiol. Methods 8:209-217.

29. Ruedi, P. 1985. HPLC-a powerful tool in carotenoid research.Pure Appl. Chem. 57:793-800.

30. Sakurai, H., K. Kato, T. Sakai, Y. Masuda, and T. Kuriyama.1971. Studies on microbial utilization of petroleum. I. Separa-tion and characterization of carotenoids produced by a speciesof Brevibacterium in hydrocarbon media. Bull. Chem. Soc. Jpn.44:481-484.

31. Schmidt, K. 1978. Biosynthesis of carotenoids, p. 729-750. InR. K. Clayton and W. R. Sistrom (ed.), The photosyntheticbacteria. Plenum Publishing Corp., New York.

32. Schwartzel, E. H., and J. J. Cooney. 1974. Isolation and char-acterization of pigmentation mutants of Micrococcus roseus.Can. J. Microbiol. 20:1007-1013.

APPL. ENVIRON. MICROBIOL.

on April 7, 2020 by guest

http://aem.asm

.org/D

ownloaded from

DETERMINATION OF CAROTENOIDS IN BACTERIA 3071

33. Schwartzel, E. H., and J. J. Cooney. 1974. Action of light onMicrococcus roseus. Can. J. Microbiol. 20:1015-1021.

34. Segers, L., and W. Verstraete. 1985. Chemical control of eucary-otic and blue-green algae in anaerobic photoreactors culturingRhodospirillaceae. Experientia 41:99-101.

35. Suhaimi, M., J. Liessens, and W. Verstraete. 1987. NH4+-Nassimilation by Rhodobacter capsulatus ATCC 23782 grownaxenically and non-axenically in N and C rich media. J. Appl.Bacteriol. 62:53-64.

36. Tanaka, A., K. Kato, and S. Fukui. 1971. Studies on theformation of vitamins and their functions in hydrocarbon fer-mentation. IX. Production of carotenoids from hydrocarbons byBrevibacterium. J. Ferment. Technol. 49:778-791.

37. Taylor, R. F. 1980. Chromatography of carotenoids and retin-oids. Adv. Chromatogr. 22:157-213.

38. Ungers, G. E., and J. J. Cooney. 1968. Isolation and character-ization of carotenoid pigments of Micrococcus roseus. J. Bac-teriol. 96:234-241.

VOL. 55, 1989

on April 7, 2020 by guest

http://aem.asm

.org/D

ownloaded from