Proc. NatI. Acad. Sci. USAVol. 89, pp. 9321-9325, October 1992Biochemistry

Functional expression of zeaxanthin glucosyltransferase fromErwinia herbicola and a proposed uridine diphosphate binding site

(carotenoid/znaxthin diglucoside/UDP-glucose)

BHUPINDER S. HUNDLE*, DAVID A. O'BRIEN*, MARIE ALBERTI*, PETER BEYERt, AND JOHN E. HEARST*t*Department of Chemistry, University of California, and Division of Chemical Biodynamics, Lawrence Berkeley Laboratory, Berkeley, CA 94720; andtInstitute of Biology, Freiburg Germany Institut fur Biologie 2, Albert-Ludwigs-Universitat, Schanzlestrasse 1, D-7800 Freiburg, Federal Republic of Germany

Communicated by Melvin Calvin, June 15, 1992

ABSTRACT Erwinia herbicola, a nonphotosynthetic bac-terium, is yellow colored due to the accumulation of unusuallypolar carotenoids, primarily mono- and diglucosides of zea-xanthin. We have cloned and expressed the gene for the enzymethat catalyzes the glucosylation of znthin. The enzyme hasan apparent molecular mass of 45 kDa on an SDS/polyacryl-amide gel, which is consistent with its calculated molecularmass. In vitro enzymatic activity was demonstrated usingUDP-["4C]glucose and eaxanthin as substrates. The productzeaanthin diglucoside and its intermediate monoglucosidewere identified by thin layer chromatography. The optimumpH and temperature ranges of the enzyme are 7.0-7.5 and32-3TC, respectively. A hydropathy plot indicates no appar-ent membrane-spanning regions, and biochemical experimentssuggest that the enzyme is weakly membrane-associated. Theamino acid sequence derived from the zeaxanthin glucosyl-transferase gene shows a small region of high similarity withother glucuronosyl- and glucosyltransferases that use eitherUDP-activated glucuronic acid or a sugar as one of theirsubstrates. Based on these similarities, we propose that thisconserved sequence is part of the UDP binding site.

Numerous hydrophobic compounds are converted to morewater-soluble products via condensation with activated glu-curonic acid or an activated sugar. The enzymes catalyzingsuch condensation reactions are known as glucuronosyl- orglycosyltransferases. The lipophilic substrate must have anappropriate substituent moiety, typically a hydroxyl or car-boxyl group, that can be covalently modified via glycosyla-tion.

In mammals, UDP-glucuronosyltransferase is primarily aliver microsomal enzyme that catalyzes the transfer of glu-curonic acid from UDP-glucuronic acid to many endogenoussubstances such as steroids, bilirubin, thyroid hormones, oramines. This condensation reaction is of critical importancein the removal of many exogenous compounds, such asdrugs, food additives, pesticides, and other ingested com-pounds, which may be toxic or carcinogenic (1).

In plants, glycosylation offlavanones is common. It is mostwidely studied in citrus fruit. The bitter taste of grapefruit isdue to the presence offlavanone neohesperidosides, whereasthe isomeric flavanone rutinosides predominant in orangesand lemon are tasteless (2). One of the bronz loci (Bz-McCallele) in maize codes for UDP-glucose flavonoid glucosyl-transferase, which catalyzes one of the last steps in antho-cyanin biosynthesis. This enzyme has been cloned and se-quenced (3).

Ecdysteroids are insect molting hormones that are essen-tial for normal development in lepidopteran species. A bac-ulovirus, Autographa californica, blocks insect molting byinterfering with ecdysteroid biosynthesis. The virus achieves

this by inserting a gene that codes for ecdysteroid glucosyl-transferase into the host genome (4). The product of thisenzyme is a glucosylecdysteroid that is not recognized by thedeveloping larvae, and as a result, the molting process isarrested.

In bacteria, an example of glycosylation of a lipophilicsubstrate occurs in carotenoid biosynthesis. Carotenoids aremethyl-branched C40 compounds composed of polyisopreneunits, which have been further desaturated to produce achromophore of conjugated double bonds. These compoundstypically serve to protect plants and bacteria from photoox-idative damage. Many common carotenoids, such as a- and,B-carotenes and their xanthophyll derivatives, have alsoundergone cyclization of their termini into six-memberedrings (5). Two examples of ring-glycosylated cyclic caro-tenoids have been identified: rhamnosylated zeaxanthin inCorynebacteria (6) and glucosylated zeaxanthin in Erwiniaherbicola (7) and Erwinia uredovora (8). The diglucosylatedzeaxanthin in Erwinia is among the most polar carotenoidsfound in nature [e.g., zeaxanthin has a solubility of 12.6 ppm(this work), whereas zeaxanthin mono- and diglucosides havesolubilities of 100 and 800 ppm in water, respectively (9)]. Thegenes coding for zeaxanthin diglucoside production fromfarnesyl pyrophosphate are clustered in the two Erwiniaspecies. This gene cluster from Er. herbicola has been clonedand expressed in Escherichia coli, resulting in yellow-coloredEs. coli (10). The various enzymatic steps have been assignedto specific loci within the Erwinia carotenoid gene clustersvia mutagenesis, and the cognate genes have been sequencedin both Er. herbicola (B.S.H., M.A., P.B., and J.E.H.,unpublished data) and Er. uredovora (8).The first portion of the carotenoid biosynthetic pathway is

common to plant and bacterial systems and consists of threecondensations of isopentenyl pyrophosphate units to formthe C20 geranylgeranyl pyrophosphate, which is then dimer-ized to the first C40 carotenoid, phytoene. All these reactionsutilize water-soluble phosphorylated substrates, and the en-zyme activities for several of these reaction steps have beenisolated in vitro from a variety of organisms (11). However,subsequent enzymatic activities in the carotenoid biosynthe-sis pathways of these organisms utilize lipophilic substratesand are difficult to isolate in vitro. Most of the enzymes forcarotenoid biosynthesis after phytoene are thought to bemembrane bound and may form a multienzyme complex. Nohomogeneous protein fractions catalyzing individual post-phytoene reactions have been isolated thus far.With the availability of sequenced DNA from Er. herbi-

cola,§ we have been able to express the enzyme zeaxanthinglucosyltransferase in Es. coli under the control of an induc-ible T7 RNA polymerase promoter. The experiments in thisreport demonstrate the enzyme's in vitro activity. A se-

tTo whom reprint requests should be addressed.§The sequence reported in this paper has been deposited in theGenBank data base (accession no. M87280).

9321

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Dow

nloa

ded

by g

uest

on

May

24,

202

1

Proc. Natl. Acad. Sci. USA 89 (1992)

quence comparison with other UDP-glycosyltransferasessuggests a putative UDP binding domain.

MATERIALS AND METHODS

Organisms and Growth Conditions. The Es. coli strainscarrying the Er. herbicola genes contained on plasmidspAPU211 (B.S.H. et al., unpublished data) and pAPUX (Fig.1) were grown at 370C in LC medium [10g oftryptone (Difco),5 g of yeast extract (Difco), and 5 g of NaCl per liter] byselection for ampicillin resistance using ampicillin (Sigma) at100 jg/ml.PCR Amplification of the crtX Gene. The Er. herbicola

zeaxanthin glucosyltransferase gene crtX was amplified via aPCR using AmpliTaq DNA polymerase (Perkin-Elmer/Cetus) in 100 pl of the standard buffer (12, 13). The N-ter-minal primer (5'-GGGATACCATATGAGCCATTTTGC-CATTG, 1 AM) contained an Nde I restriction site to permitthe ligation of the amplified gene directly into the translationstart site of the pET-3b expression vector (14). The C-ter-minal primer (5'-AAATCAGATCTCTCACGATACGCTCT-CACT, 1 ,M) was designed to hybridize to a region imme-diately downstream of the gene and to contain a Bgl II site topermit insertion into the BamHI site of pET-3b. LinearizedpAPU211 plasmid containing the Er. herbicola carotenoidgene cluster was used as the template (0.42 Ag or 500 pM).The amplification cycle was as follows: denaturation at 940Cfor 1 min, annealing at 450C for 1.5 min, a gradual temperatureincrease to 72TC over a 2-min period, and a further primerextension at 720C for 1 min. Amplification was performed for30 cycles.

Plasmids and Cloning Techniques. Plasmid pAPU211 wasderived from pPL376 (10) and contains the essential Er.herbicola genes for carotenoid production (B.S.H. et al.,unpublished data). Plasmid pAPUX (Fig. 1) was constructedby ligating PCR-amplified crtX into the Nde I and BamHIrestriction sites of the pET-3b vector (14). The plasmid was

cloned and maintained in Es. coli DH5a cells. Clones con-taining pAPUX were screened by the mini-prep method ofRiggs and McLachlan (15). All nucleic acid and enzymaticmanipulations were performed according to standard pub-lished procedures (16) or manufacturer's protocols. For theoverexpression of glucosyltransferase, plasmid pAPUX wastransformed into Es. coli BL21(DE3).SDS/Gel Electrophoresis. Intact Es. coli cells were centri-

fuged and resuspended in 400 pl of 1 x sample buffer per ml

EcoRl 5.55 HindIII 0.02

Shine Dai:garno

promoter

FIG. 1. Construction of plasmid pAPUX. Ap represents theampicillin-resistance gene. crtX is the Er. herbicola zeaxanthinglucosyltransferase gene. Upstream of the crtX gene at an appro-priate distance is a T7 promoter and a Shine-Dalgarno sequence.Downstream of crtX is a strong 17 terminator.

of culture and boiled for 4 min. Cell lysate fractions were alsotreated similarly, and proteins were analyzed by SDS/PAGEon a 10%1 gel with the discontinuous buffer system ofLaemmli(17). Proteins were stained with Coomassie blue R250 anddestained in methanol/acetic acid/water, 4:1:5 (vol/vol).To obtain proteins for N-terminal sequencing, an unstained

gel was electroblotted onto an Immobilon membrane, ac-cording to Immobilon tech protocol TP006 (Millipore). Themembrane was stained with Coomassie blue, and bands ofinterest were cut from the membrane and sequenced.

Preparation of Enzyme Extract. Es. coli BL21(DE3) cellscontaining the plasmid pAPUX were grown to an OD6N, valueof0.4 and induced with 0.4mM isopropyl P-D-thiogalactosidefor 45 min. All further steps were carried out at 40C unlessotherwise stated. Cells were harvested by centrifugation at4200 x g for 10 min. The resulting pellet from a 100-ml cellsuspension was resuspended in 1.0-4.0 ml of0.05M Tris HCl(pH 7.5), containing 1 mM 2-mercaptoethanol and 0.1 mMphenylmethylsulfonyl fluoride. Crude cell lysate was ob-tained by one passage through a French pressure cell at13,000 psi (1 psi = 6.9 kPa). Large cell debris and unbrokencells were removed by centrifugation at 3100 x g for 5 min.

Fractionation of the crude cell lysate was performed bycentrifuging the sample at 13,000 x g for 15 min. The pelletobtained in this manner is indicative of segregation of excessoverexpressed protein in inclusion bodies. The low-speedsupernatant was ultracentrifuged at 100,000 x g for 90 min.The pelleted membrane fraction was resuspended in 240 Al ofthe Tris buffer described above. A 120-jud fraction was savedfor analysis, and the remaining 120 lja was washed with 4 mlofTris buffer and again ultracentrifuged for 90 min. This finalpellet was resuspended in 120 iml ofTris buffer and is referredto as the washed-membrane fraction. Protein concentrationswere determined by the Bradford assay (18).Enzyme Assay. Glucosyltransferase activity was assayed

by measuring the incorporation of [14C]glucose into zeaxan-thin. The assay mixture was prepared as follows: 20-80 ul ofenzyme extract, 13.5 nmol of zeaxanthin in 5 ,ul of acetone,and 100 nmol of [14C]UDP-glucose at a specific activity of2.95 mCi/mmol (1 Ci = 37 GBq), in 0.05 M Tris'HCl (pH 7.5)containing 1 mM 2-mercaptoethanol to a final volume of 100,ul. UDP-glucose, radiolabeled at all six carbon atoms, waspurchased from DuPont/NEN at a specific activity of 295.1mCi/mmol and diluted with unlabeled UDP-glucose (Sigma).Zeaxanthin was a gift from Hoffmann-La Roche. Zeaxanthinwas quantified based on an Ellv of 2340 at 452 nm in acetone(19). Incubation was performed at 37C for 5 h in the dark, andthe reaction was stopped by the addition of 200 pA ofchloroform/methanol, 2:1 (vol/vol). Reaction products wereidentified by TLC. To separate products, 40 ,l of thechloroform extract from each assay mixture was applied ona silica gel TLC plate (Whatman). Two solvent systems wereused. The TLC was first developed in solvent system A[petroleum ether/ether/acetone, 1:1:1 (vol/vol)] to separateunreacted zeaxanthin from the products. After air drying theTLC plate for 30 min, solvent system B [petroleum ether/ether/methanol, 1:1:1 (vol/vol)] was used to separate zea-xanthin monoglucoside from zeaxanthin diglucoside. TheTLC plates were scanned for radioactivity using a Phosphor-Imager 400 series (Molecular Dynamics, Sunnyvale, CA),and spots were quantified using IMAGEQUANT Version 3.15software supplied by Molecular Dynamics.

Protein Sequence Comparisons. The deduced amino acidsequence of CrtX from Er. herbicola was compared with theprotein sequence data bases. Data base searches were madeusing the FASTDB program (20) of the Intelligenetics Suite ofsequencing software. Data banks searched were the ProteinIdentification Resource Version 28 (March 31, 1991) andSwiss-Prot Version 18 (May 199i). To identify commonmotifs or boxes of similarity, short overlapping sequences of

9322 Biochemistry: Hundle et al.

Dow

nloa

ded

by g

uest

on

May

24,

202

1

Proc. Natl. Acad. Sci. USA 89 (1992) 9323

48 amino acid residues were used to probe the data bases.Sequences were aligned using the Intelligenetics GENEALIGNMENT program.

RESULTSThe plasmid construct pAPUX is outlined in Fig. 1. PCR-amplified Er. herbicola DNA containing crtX was insertedimmediately beyond the T7 promoter and Shine-Dalgarnosequence of the pET-3b expression vector. A 410 T7 termi-nation signal is located immediately downstream of the crtXinsert.

This pAPUX construct was maintained in Es. coli DH5acells, and the crtX gene was expressed in Es. coli BL21(DE3)by induction with isopropyl j3-D-thiogalactoside. The accu-mulation of a newly formed 45-kDa protein was observed ona SDS/10% polyacrylamide gel (Fig. 2). The calculatedmolecular mass based on the derived amino acid sequence ofthe protein is also 45 kDa. The identity ofthis band as the crtXgene product was confirmed by N-terminal amino acid se-quencing. An additional band of slightly lower molecularmass was observed only as a minor component in the crudeextract but was a major component in the supernatantfractions.

Extracts from plasmid-containing Es. coli strains wereassayed for in vitro glucosyltransferase activity by incubationwith zeaxanthin and UDP-[14C]glucose. When an extractfrom induced Es. coli BL21(DE3) cells containing pAPUXwas used, significant amounts of the yellow-colored radio-labeled products, zeaxanthin mono- and diglucosides, wereobserved. As a negative control, an extract from Es. coliBL21(DE3) cells containing the pET-3b vector without crtXwas incubated with zeaxanthin and UDP-[14C]glucose in thein vitro assay, and no product was observed (data not shown).Thin layer chromatographic analysis of the products of thisassay is shown in Fig. 3, lane 1. Solvent system A was usedto separate unreacted zeaxanthin (Rf = 0.8) from the reactionproducts. Subsequent use of solvent system B resulted in theresolution of the polar zeaxanthin mono- and diglucosides(Fig. 3, lane 1). Authentic samples of zeaxanthin and thezeaxanthin mono- and diglucosides [Er. herbicola carotenoidextract (7)] were used as standards for comparison (Fig. 3,lanes 3 and 4). The second substrate UDP-[14C]glucose didnot move from the origin and, being colorless, was onlydetected as a radiolabeled spot (lane 2). Phosphorimaging ofthe TLC plate showed the two colored spots at Rf values of

FIG. 2. SDS/PAGE in a 10o gel showing expression of zeaxan-thin glucosyltransferase. Lanes: 1, protein standard markers of 66,45, 24, and 18.4 kDa; 2 and 3, total protein from uninduced andinduced whole Es. coli cells containing the plasmid pAPUX, respec-tively (note: far less total protein was loaded in lane 2); 4, 12.4 Ag ofcrude extract obtained by using a French press; 5, 4.2 ,ug of inclusionbody fraction; 6, 22.7 jsg of low-speed (13,000 x g) supernatantfraction; 7, 30.2 ,ug of high-speed (100,000 x g) supernatant fraction;8, 12.8 ug of unwashed membrane fraction; 9, 3.4 ,ug of washedmembrane fraction.

A

Solvent A+

Solvent B+

B1 2 3 4 1 2

't

=, a g | i | .:.'...._ gi ill | _ W ,nwt_>y=. e i:_ | | | X <._ n -* S * EbbLi$:'F4|'

SxwSJ l l i ,".,:.1^. | | | ${>2=->2$2

*, I I l Siae. _:

, ' | | | | | l gl'''-'',:$ * - * }t7:.g * - * '.tX'5 * * | 2:;z$1: _.i1 | | l _ _s | | |N l | l r.IIE E * | _;<il:,| | | l _IEII l - l l * l l 311S:S | | lwfli l - _ l l * l _19_| | | l121l| * -l - __,lS | - _ | __'| I | | I _. ;g | | * | =';i$$^^R$, S * - l | i'

| | | i F$ |:R | rM} wvv.. -si#n M - - -g , S -1so _s,.,_ | l$ i | '-y:''"'='' ^

X1 W I ,1 l, ,! X w .-..:' w '&_

FIG. 3. Zeaxanthin glucosyltransferase assay results showingformation of zeaxanthin mono- and diglucosides. (A) TLC platestained with iodine vapor. Lanes: 1, incubation of induced Es. coliBL21(DE3) cell lysate containing pAPUX with zeaxanthin andUDP-[14C]glucose showing formation ofzeaxanthin mono- and diglu-cosides; 2, UDP-[W4C]glucose; 3, zeaxanthin; 4, Es. coli (pPL376)carotenoid extract, containing mono- and diglucosides as majorcarotenoids. Solvent fronts for solvents A and B are indicated. (B)Phosphorimage of the TLC plate. Lanes: 1, incorporation of 14C inboth products, mono- and diglucosides; 2, standard UDP-[14C]glucose remaining at the origin.

1.0 and 0.3 in solvent system B to be radioactive, thusindicating that [14C]glucose was incorporated into both prod-ucts.The in vitro activity of the enzyme was assayed at 37°C

from pH 6.5 to pH 9.0, and maximum activity was found atpH 7.0-7.5. Similarly, activity at pH 7.4 was assayed from 23to 42°C and found to be optimal at 32-37°C (data not shown).The addition of larger amounts of enzyme extract did notincrease the yield of glucosylated products, indicating thatenzyme was not the limiting component. Also, concentra-tions of UDP-glucose >1 mM did not increase the productyield (data not shown). However, the addition of increasingamounts of zeaxanthin from 3 to 50 nmol/0. 1 ml to the assaydid result in a roughly proportional increase in products (Fig.4A). To determine an optimal time at which cells should beharvested after isopropyl B-D-thiogalactoside induction,crude enzyme extracts were assayed for activity at varioustimes after induction. Fifty percent ofthe maximum observedenzyme activity was present in cells harvested after 0.5 h, and=80%o of maximum activity was reached in 1 h (Fig. 4B).To determine the intracellular localization of active gluco-

syltransferase, various cell fractions were assayed for activ-ity. The glucosyltransferase activity was present in both thewashed-membrane fraction and the high-speed supernatantfraction. In assays ofthe whole-cell lysate and all supernatantfractions, the zeaxanthin diglucoside was produced in signif-icantly higher amounts than the monoglucoside (Fig. 4C).The specific enzymatic activity was higher in the crudeextract than in the later fractions, indicating the loss of someactivity during fractionation.A hydropathy plot of the derived amino acid sequence of

Er. herbicola CrtX was prepared by the method of Kyte andDoolittle (21), in which a hydropathy index of + 1.6 or greater

Biochemistry: Hundle et al.

Dow

nloa

ded

by g

uest

on

May

24,

202

1

Proc. Natl. Acad. Sci. USA 89 (1992)

A

6

B 2._

0

0

ut-

04

0

bo

C7

Induction time, h

C

3

Fraction

FIG. 4. Plots of enzyme activity versus increasing amount ofzeaxanthin (A) or isopropyl 1B-D-thiogalactoside induction time (B),after which cells were harvested and fractionated. Curves: 1, mono-and diglucoside; 2, diglucoside; 3, monoglucoside. (C) Cell lysatefractions: 1, crude cell extract; 2, low-speed supernatant; 3, high-speed supernatant; 4, unwashed membranes; 5, washed membranes.Bars: open, monoglucoside; solid, diglucoside; hatched, mono- anddiglucoside. Enzyme activity was determined in 100 ul containing20-80 ,ul of enzyme extract, 13.5 nmol of zeaxanthin in 5 ,ul ofacetone (or as indicated), 100 nmol of [14C]UDP-glucose at a specificactivity of 2.95 mCi/mmol in 0.05 M Tris'HCl (pH 7.5) containing 1mM 2-mercaptoethanol. Yields of both products were determined bythe phosphorimaging of TLC plates as described in Fig. 3 andcompared with a known 14C standard.

for a segment containing at least 19 amino acid residues isindicative of a membrane-spanning region (plot not shown).No unambiguous membrane-spanning regions were identi-fied using these criteria; however, two regions were foundthat contain primarily hydrophobic residues. Both of these

regions, from residues 115 to 131 and from residues 324 to334, contain multiple stretches of hydrophobic amino acidsthat are of insufficient length to constitute transmembranehelices.The CrtX amino acid sequence has been aligned for com-

parison with the following sequences: insect ecdysteroidUDP-glucosyltransferase (4); maize UDP-glucose flavonoidglucosyltransferase (3); and human (22), rat (23), and mouse(24) UDP-glucuronosyltransferases (Fig. 5). The three mam-malian glucuronosyltransferases show good identity amongthemselves but not with the other proteins. Between the twospecies of Erwinia, there is 56% identity. The main region ofhomology present in all six sequences is the area marked bythe shaded bar in Fig. 5.

DISCUSSIONThe observation that [14C]glucose is incorporated into twoproducts is consistent with the carotenoid biosynthesis path-way for Er. herbicola, in which a two-step glucosylation ofzeaxanthin was proposed based on the accumulation of thetwo pigments, zeaxanthin mono- and diglucosides (7).

UDP-glucose UDP UDP-glucose UDP

zeaxanthin a zeaxanthin monoglucoside a zeaxanthin diglucoside

In a standard reaction mixture, 100 nmol of UDP-glucoseand 13.5 nmol of zeaxanthin were incubated with 0.12 mg ofprotein from a crude enzyme extract in 0.1 ml. Increasing theamount of enzyme or UDP-glucose did not result in asignificant increase in the amount of products. This indicatesthat in this particular assay, neither the enzyme nor theUDP-glucose was the limiting factor. However, product yieldwas greatly increased by increasing the amount of zeaxan-thin. Despite the apparent solubility limitations, levels ofzeaxanthin substrate as high as 50 nmol/0.1 ml continued toproduce an increase in the glucosylated products. Also, thezeaxanthin diglucoside was found in significantly higheramounts than the monoglucoside, despite the fact that thediglucoside is the product of the second condensation reac-tion. The monoglucoside, as an intermediate, never reachesvery high levels compared with zeaxanthin, the originalsubstrate. These results are consistent with a model in whichthe binding of zeaxanthin to the enzyme is the rate-limitingstep for the reaction sequence. It is not known whether themonoglucoside intermediate must completely dissociatefrom the enzyme complex to undergo further reaction.

- 347 qnWfnQrAVLRH.Kkmr-aAF:. qgq-5 sdEa aea a ;PMvQ-r--.-

Mz 349 VpWapQvAVLRHpsvgAFv-UHauwaSVLEgLss vPMa'r',> Sf!>DQ As3--

CrtX 297 VsfvdQpryvaean 1vT-H4GG-NtVLdALaaat:v'a'TLsf7QPa-aAFR'A%:

H 354 YKWiPQNDLLGHPKT-AFITHGGA:NGiYkAIspr _ MvGVPLFaDQPDNIAHM,,MzA-.

R 355 YKWLP2NDLLGHPKTCAFv:HNGGANG1YEAIYHG:?MIGIPLFG;,DQPDNIAHKw.I:P.

355 YKWLPQNDLLGHPKTKAFVGHGGANGvYEAIYHIC-_PMI7 FLFGeQhDhNIAHMiVArR

FIG. 5. Sequence similarities of the UDP-glucosyl moiety bind-ing proteins. Only the region around the conserved domain isshown. Protein sequences, sources, and Swiss-Prot accession num-bers are as follows: I, insect ecdysteroid UDP-glucosyltransferase,UDPE$NPVAC; Mz, maize UDP-glucose flavonoid glucosyltrans-ferase, UFO1$MAIZE; CrtX, Er. herbicola zeaxanthin UDP-glucose transferase, GenBank accession no. M87280; H, R, and M,human (UDP1$HUMAN), rat (UDP4$RAT), and mouse(UDP1$MOUSE), UDP-glucuronosyltransferases, respectively.Uppercase type and vertical lines indicate identical amino acidresidues in adjacent sequences, asterisks identify residues conservedamong at least five of the six sequences, lowercase type identifiesnonconserved residues, and the overbar represents the main regionof homology.

9324 Biochemistry: Hundle et al.

Dow

nloa

ded

by g

uest

on

May

24,

202

1

Proc. Natl. Acad. Sci. USA 89 (1992) 9325

A significant increase in monoglucoside yield as a fractionof total products could be seen at high zeaxanthin levels (Fig.4B), as expected from a zeaxanthin-limited reaction. Com-pared with the supernatants, membrane preparations pro-duced relatively higher levels ofthe monoglucoside (Fig. 4C).This increase in monoglucoside yield is most likely the resultof the greater availability of zeaxanthin substrate, since thislipophilic compound is expected to partition into the mem-brane (nonpolar phase).The glucosyltransferase activity was present in both the

washed membrane fraction and the supernatant fraction.After fractionation, washed membranes exhibited >5-foldhigher activity per mg of total protein when compared withthe high-speed supernatant, indicating that the protein ismembrane-associated. A hydropathy plot of the enzyme'sderived amino acid sequence does not show any clear mem-brane-spanning regions. However, it is possible that one ofthe two hydrophobic domains (see Results), instead of beingan internal region of the protein, could serve as an "anchor"to associate the protein with the membrane. Phospholipiddependence of another UDP-glucose-requiring enzyme in-volved in maize sterol glucosylation has been demonstratedby Ury et al. (25). Our conclusion is that CrtX is looselyassociated with the membrane and is partially dissociatedfrom the membrane during fractionation.

Several factors could account for the enzyme activityremaining in the high-speed supernatant fraction. It is pos-sible that the enzyme is still associated with small membranefragments that are too small to be sedimented. Mowat andArias (26), using electron microscopy, have shown that thetransfer of glucuronosyltransferase activity from liver mi-crosomes to the supernatant obtained by ultracentrifugationat 100,000 x g results merely from reduction in the size ofmembrane vesicles and not from free enzyme. Alternatively,it is possible that the enzyme is associated with the membranein vivo but that the membrane is not absolutely required forsome basal level of activity. For maximum activity of theenzyme, membranes might be essential simply because thezeaxanthin substrate is a lipophilic compound located in themembrane.We observe the formation of a second band on the SDS/

PAGE gel upon centrifugation of the cell lysate to isolatemembranes (Fig. 2). It appears that this protein, which is onlyslightly smaller in molecular mass than the original overex-pressed product, is the result of proteolytic cleavage of theoriginal enzyme. This second protein is segregated in thesupernatant and may be formed after some of the originalenzyme is freed from the membrane, making it more acces-sible to proteases. The appearance of this second band is alsoconcomitant with a net loss ofenzyme activity observed uponfractionation. The combined enzyme activities of both the100,000 x g supernatant plus membrane fractions are lessthan that found in the original lysate (Fig. 4C) on an activityper mg of protein basis. Because this lower band is not foundin a significant level in cells lysed by boiling, it is possible thatthe release and proteolysis of the enzyme results from thefractionation itself and does not reflect a normal in vivoprocess.The deduced amino acid sequence of Er. herbicola CrtX

shows a negligible amount of overall homology with the othersequenced UDP-glucosyl- and glucuronosyltransferases.However, there is one distinct region of homology that ispresent in each of the six sequences (Fig. 5). Since the onlycommon feature among these enzymes is that they requireUDP-activated substrates, we conclude that the followingsequence is most likely a portion of the UDP binding site:

QX13TX2GX7LX4PX4PX3DQX4A, where X represents avariable residue.

In summary, we have overproduced the enzyme zeaxan-thin glucosyltransferase from Er. herbicola and demon-strated its in vitro activity. The enzyme apparently is not anintegral membrane protein but is loosely attached to themembrane. We have proposed a UDP-binding sequence thatis shared by the known transferases that require a UDP-activated second substrate.

We thank R. Tuveson for providing plasmid pPL376, D. Koh foroligonucleotide synthesis, M. Moore of the University of CaliforniaBerkeley Microchemical Facility for N-terminal amino acid sequenc-ing, and D. Cook and D. Burke for critical reading of the manuscript.This work was supported by National Institute of Environment andHealth Postdoctoral Training Grant 2 T32 ES07075-11 (B.S.H.) andin part by the office of Basic Energy Sciences, Biological EnergyDivision, Department of Energy under Contract DE-AC030-76SF00098 (J.E.H.) and the Deutsche Forschungsgemeinschaft(P.B.).

1. Dutton, G. J. (1980) Glucuronidation of Drugs and OtherCompounds (CRC, Cleveland), pp. 1-78.

2. Bar-Peled, M., Lewinsohn, E., Fluhr, R. & Gressel, J. (1991)J. Biol. Chem. 266, 20953-20959.

3. Ralston, E. J., English, J. J. & Dooner, H. K. (1988) Genetics119, 185-197.

4. O'Reilly, D. R. & Miller, L. K. (1989) Science 245, 1110-1112.5. Straub, 0. (1987) in Key to Carotenoids, eds. Pfander, H.,

Gerspacher, M., Rychener, M. & Schwabe, R. (Birkhaeuser,Basel), pp. 11-171.

6. Nybraaten, G. & Liaaen-Jensen, S. (1974) Acta Chem. Scand.Ser. B 28, 1219-1224.

7. Hundle, B. S., Beyer, P., Kleinig, H., Englert, G. & Hearst,J. E. (1991) Photochem. Photobiol. 54, 89-93.

8. Misawa, N., Nakagawa, M., Kobayashi, K., Yamano, S.,Izawa, Y., Nakamura, K. & Harashima, K. (1990) J. Bacteriol.172, 6704-6712.

9. Pfander, V. H. & Hodler, M. (1974) Helv. Chim. Acta 57,1641-1651.

10. Perry, K. L., Simonitch, T. A., Harrison-Lavoie, K. J. & Liu,S. T. (1986) J. Bacteriol. 168, 607-612.

11. Britton, G. (1983) in The Biochemistry of Natural Pigments(Cambridge Univ. Press, Cambridge, U.K.).

12. Saiki, R. K., Gelfand, D. H., Stoffel, S., Scharf, S. J., Higuchi,R., Horn, G. T., Mullis, K. B. & Erlich, H. A. (1988) Science239, 487-491.

13. Loh, E. Y., Elliot, J. F., Cwirla, S., Lanier, L. L. & Davis, M.(1989) Science 243, 217-220.

14. Studier, W. F., Rosenberg, A. H., Dunn, J. J. & Dubendorff,J. W. (1990) Methods Enzymol. 185, 60-89.

15. Riggs, M. G. & McLachlan, A. (1986) BioTechniques 4, 312-313.

16. Maniatis, T., Fritsch, E. F. & Sambrook, J. (1989) MolecularCloning:A Laboratory Manual (Cold Spring Harbor Lab., ColdSpring Harbor, NY).

17. Laemmli, U. K. (1970) Nature (London) 227, 680-685.18. Bradford, M. M. (1976) Anal. Biochem. 72, 248-254.19. Davies, B. H. (1976) Chem. Biochem. Plant Pigm. 2, 38-165.20. Brutlag, D. L., Deutricourt, J. P., Maulik, S. & Relph, J. (1990)

Comput. Appl. Biosci. 6, 237-245.21. Kyte, J. & Doolittle, R. F. (1982) J. Mol. Biol. 157, 105-132.22. Jackson, M. R., McCarthy, L. R., Harding, D., Wilson, S.,

Coughtrie, M. W. H. & Burchell, B. (1987) Biochem. J. 242,581-588.

23. Mackenzie, P. I. (1987) J. Biol. Chem. 262, 9744-9749.24. Kimura, T. & Owens, I. S. (1987) J. Biochem (Tokyo) 168,

515-521.25. Ury, A., Benveniste, P. & Bouvier-Nave, P. (1989) Plant

Physiol. 91, 567-573.26. Mowat, A. P. & Arias, I. M. (1970) Biochim. Biophys. Acta

212, 65-78.

Biochemistry: Hundle et al.

Dow

nloa

ded

by g

uest

on

May

24,

202

1