ManipulationandCompensationof SteroidalGl.coalkaloidBiosynthesis in Potatoes
K.F. McCue', P.V. Allen, L.V.T. Shepherd 2 . A. Blake 2 , D.R. Rockhold', R.G. Novy,D. Stewart, H.V. Davies and W.R. Belknap'
'Agricultural Research Service, 800 Buchanan Street, Albany, CA 94710 USA2 Quality, Health and Nutrition Programme, Scottish Crop Research Institute, lnvergowrie,
Dundee, DD2 5DA UK3 Aberdeen Research and Extension Center, 1693 5 2700 W. Aberdeen, ID 83210 USA
Keywords: Solanurn tuberosurn, steroidal glycoalkaloids, glue osyltransferase,galactosyltransferase. rhamnosyltransferase
AbstractSteroidal glycoalkaloids (SCAs) are undesirable secondary metabolites
produced in solanaceous plants including potato,tomato and eggplant. Two tn-glycosylated alkaloids, ct-chaconine and a-solanine, occur naturally in potato tubersand can accumulate to excessive levels due to mechanical, environmental, andgenetic perturbations. We have identified members of the steroidal-alkaloid glycosyltransferasese (Sgt) gene family coding the LJDP-galactose: solanidinegalactosyltransferase (Sgtl), the U DP-glucose:solaflidifle glucosyltransferase (Sgt2),and the UDP_rhamnose:3sOlaflifle/13-chacon1ne rhamnosyltransferase (Sgt3).Reverse genetic manipulation of the Sgt gene family members using transgenic linesexpressing active antisense gene constructs results in a shift of SGA pathway flux, inturn resulting in compensation by other pathway intermediates and products. NewSgt gene family members are being studied to verify the substrate specificity andbiosynthetic sequence. Additional studies are underway to use the members of thisgene family to completely block accumulation of SCAs in tubers.
INTRODUCTIONPotatoes (So/an urn tuberosurn) produce bitter compounds called steroidal
glycoalkaloids (SGA5) (Gregory, 1984). SGAs are also found in relatives of the potato,tomato (Lvcopersicon esculenturn) (Zimowski, 1996), and eggplant (Solanurn melongena)(Paczkowski et al., 1998). High levels of SGAs are undesirable. When potatoes areexposed to light they turn green and start to accumulate SGAs. High levels can also becaused by bruising, adverse environmental conditions during growth, and during breedingof potatoes for new varieties (Kuc, 1984, Sinden et al., 1984).
We have isolated and characterized three steroidal-alkaloid glycosyltransferase(Sgt) genes involved in the biosynthesis of SGAs (Fig. 1). The first gene isolated was SgtI(Moehs et al., 1997). Originally believed to encode the UDP-glucose:solanidifleglucosyltransferase, the in vitro function of this gene was later determined to be that ofthe UDP-galactose:solanidine galactosyltransferase (McCue et al., 2005). Transgenicplants regenerated with antisense transgenes or empty vector controls routinely showstable variations in glycoalkaloid content +1-30% in healthy true to type lines (McCue etal., 2003) that is not attributable to transgene activity.
Using protein sequence homology a second Sgt gene family member wasidentified and isolated, and the in vivo and in vitro function for Sgt2 identified this geneas that encoding the true UDP-glucose:solanidlne glucosyltransferase (McCue et al.,2006). A third member of the Sgt gene family has now been characterized and the in vivofunction of the Sgt3 gene is that of the rhamnose:13-steroidal glycoalkaloid (-solanineII3-chaconine) rhamnosyl transferase (McCue et al., 2007).
As part of a program to improve potatoes with biotechnology we are isolating andcharacterizing the natural potato genes that encode the enzymes responsible forglycosylation in the SGA biosynthetic pathway. By re-introducing the genes into potatoes
VI h International Solanaceae ConferenceEds.: D.M. Spooner et al. 343Acta Hon. 745, ISHS 2007
in an antisense orientation, we can block the sugar addition and alter the accumulation ofS(IAs. We have now assigned genes to four of the six steps in theglycosylation of SGA5
inI hese genes are being compared and used to identify genes encoding the additional steps
the pathway. The effect on glycoalkaloid accumulation in transgenic plant linescontaining active antisense constructs is shown. The effects of wounding on geneexpression should serve as a tool for aiding the identification of the remaining genefamily members.
MATERIALS AND METHODS
40Plant MaterialsRound white potato (Solanuin tuberosurn) cultivars 'Lenape' (Akeley et al., 1968)
and Desirée' were used in transgenic experiments. Plants were grown in the glass housesin Albany, California, or Invergowrie, Dundee, Scotland.
Sgi Sequence Identification and ComparisonNovel Sgt sequences were obtained by searching the translated potato EST
database with the predicted SGTI protein sequence. Homologous sequences wereamplified from a wounded tuber cDNA library as previously described (McCue et al.,2005). Sequences are compared using the Blast server and by ClustaIW alignment, andBoxshade was used to format aligned sequences.
Steroidal Glycoalkalojd DeterminationsLevels of SGAs were quantified from glasshouse-grown 'Lenape.' Freeze-dried
material was produced from 'Lenape' and 'Desirée' tubers as described by McCue et al.,2005 and Defernez et at., 2004. Glycoalkaloid levels were determined by an LC-MSmethod on a Thermo LCQ-DECA (Hemel Hempstead, UK) LC-MS as previouslydescribed (McCue et at., 2006). Analysis was performed on two extractions from each ofthree independent minitubers for each line.
Protein Purification and AnalysisProtein was expressed in yeast using pYES (Invitrogen) constructs as previously
described (McCue et al., 2005, McCue et al., 2006). Protein purification was carried outusing the His protein isolation system (Sigma) according to manufacturer's instructions.Purification was assessed via SDS PAGE and staining was done with Coomassie blue.Elution of recombinant proteins was monitored by Western blot analysis using the anti-VS epitope antibody (Invitrogen). Solanidine glycosyl trans ferase assays were carried outas described (McCue et at., 2005). Products were separated using anion exchange resin(Stapleton et at., 1991). No activity is reported for reactions with less than twofold theaverage dpm counts of the blank reactions. Values are the mean of duplicate assays.
RNA BlotsPotato tubers were wounded by cutting and incubated in the dark. Leaves were
wounded by crushing 50% of the leaf surface with a hemostat. Wounded and controltissues were frozen in liquid nitrogen, ground, and extracted for RNA as previouslydescribed (Verwoerd et al., 1989). RNA was isolated and fractionated by agarose gelelectrophoresis, and transferred to a charged nylon membrane (Roche) (Rickey andBelknap, 1991). RNA blots were hybridized with random-primed (GE Healthcare),double-stranded probes of the: SgtJ cDNA (Moehs et al., 1997) (GenBank Accession no.U82367), Sgt2 cDNA (McCue et at., 2006) (GenBank Accession no. DQ2 18276), or theSgt3 eDNA (McCue, GenBank Accession no. DQ266437) (McCue et al., 2007).
344
10 RESULTS AND DISCUSSION
Sgt Sequence Identity and Protein HomologySgt2 and Sg13 were identified as the most similar translated sequence matches to
the SGT1 protein at 61% and 43% identity respectively. SGT2 and SGT3 are 44%identical. A ClustaIW alignment of the coding regions of SGT1, SGT2 and SGT3 isshown in Fig. 2, revealing additional homologies across the predicted peptides. Thenucleotide sequences for the different Sgt genes show a greater percent identity than theprotein sequences, with Sgt2 and Sgt3 being 69% and 58% identical to SgtJ, respectively,
and Sgt2 and Sgt3 sharing 58% identity.
Substrate Specificity of SGTI and SGT2Protein activity was confirmed by in vitro studies on the activity of recombinant
SGTI and SGT2 protein with solanaceous SGAs and either UDP-[ 3 H]glucose or IJDP-[ 3 H]galactose as the sugar donor (Table I). While the recombinant SGTI is capable ofutilizing UDP-glucose as a substrate in vitro, this activity is drastically less than theactivity observed with UDP-galactose. Surprisingly the activity with the tomato SGAs ishigher than with the native substrate solanidine regardless of the sugar donor. Therecombinant SGT2 was readily capable of utilizing UDP-glucose as a substrate in vitrowith all three solanaceous aglycones (solanidine, solasodine, and tomatidine) withmaximum activity obtained with the potato aglycone solanidine. No activity was observedwith UDP-galactose as a substrate for any of the aglycones tested. The threshold foractivity was set at twofold the reaction blank.
Steroidal Glycoalkaloid AccumulationThe SGA levels were measured in glass house-grown minitubers from 2005.
Fig. 3 shows the levels of component alkaloids a-solanine and a-chaconine in thetransgenic Lenape' lines with active antisense Sgt genes, the empty vectorpBINPLUS/ARS (pBPA), or Lenape' control lines. The plant lines are sorted bytransgene.
Expression of an effective antisense SgtI gene in potato tubers resulted in severelyreduced a-solanine accumulation and elevated levels of a-chaconine, resulting inessentially unchanged total SGA levels. This is consistent with the enzymatic data whereSGTI is the only enzyme with detectable gal actosyltransferase activity. Flow of carboninto the common precursor of both SGT1 and SGT2 appears to be controlled not by aspecific SGA end product but by the total level of cc-SGAs. This is supported by thefinding that SGT I activity was partially inhibited in the presence of either u-SGAs(McCue et al., 2005).
The antisense Sgt2 gene partially inhibits the accumulation of a-chaconine, theleast desirable of the two major SGA species with a significant concomitant increase in a-solanine accumulation. The incomplete inhibition of a-chaconine accumulation may beexplained by the enzymatic data. Although SGT2 has a strict substrate specificity forUDP-glucose as the sugar donor SGT1 does exhibit glucosyltransferase activity.Although there is no significant glucosyltransferase activity with solanidine shown inTable I this activity was observed for SGTI in an earlier study (Moehs et al., 1997) andthus may account for a-chaconine accumulation in the presence of effective antisenseSgr2. The accumulation of more a-solanine than a-chaconine in line 1615 is the oppositeof what is normally observed, suggesting a weakly effective antisense transgene causing asignificant increase in a-solanine accumulation. However, the level of a-chaconine is notsignificantly different than controls and this effect was not observed previously orsupported by suppression of Sgt2 mRNA (McCue et al., 2006).
Introduction of the antisense Sg13 gene resulted in a significant decrease in theaccumulation of both SGAs that are usually found. Because a single antisense transgeneis capable of blocking accumulation of both end products, this suggests that either SGT3is the only J3-SGA rhamnosyltransferase or that if there are distinct -solanine and 3-
345
Mon
chaconine rhamnosyltransferases, that they share sufficient nucleic acid identity to bedown-regulated by a single antisense transcript. These results confirmed earlier studiesthat showed reductions in both ci—solanine and a—chaconine with a resultingaccumulation of both -SGA intermediates (McCue et al., 2007).
Analysis of variance for a-solanine and ci-chaconine indicates that the changes inci-solanine content for all three antisense lines are significantly different than the controlsas are the changes in ci-chaconine content in the Sgtl and Sgt3 lines. There is nosignificant difference between the empty vector transformed lines and the 'Lenape'controls.
Wound-Induced Expression of the Sgt Gene FamilyIt is well known that accumulation of glycoalkaloids is subject to a variety of
environmental factors including wounding (Sinden et al., 1984). To examine the effect ofwounding on Sgt mRNA abundance, total RNA was extracted from wounded leaves andtubers and probed with SgtI, Sgt2 or Sg13 (Fig. 4). In tubers the expression pattern isvirtually identical for all three Sgt gene family members. Transcript levels in total mRNAfrom tubers decreases slightly after two hours, increases by 16 hours, subsequentlydecreasing by 48 hours. In leaves all three genes exhibit different patterns. SgtI does notshow an increase in mRNA abundance after wounding. Conversely, Sgt2 and Sgt3divergent at two hours converge with strong induction at 16 hours.
These genes and information from them will be used to isolate and identify theremaining members of the Sgt gene family to be used in future efforts to modify theaccumulation of SGAs in potatoes.
ACKNOWLEDGEMENTSThe authors acknowledge funding by the Agricultural Research Service National
Programs (CRIS Project Number 5325-21420-001-OOD) (K.F.M., P.V.A., D.R.R..W.R.B.). the European Commission (Grant Number QLRT-1999-00765) (H.V.D.,L.V.T.S.), and the Scottish Executive Environment and Rural Affairs Department (A.B,D.S., H.V.D., L.V.T.S.). The mention of a trademark or proprietary product does notconstitute a guarantee or warranty of the product by the United States Department ofAgriculture and does not imply its approval to the exclusion of other products that may besuitable.
Literature CitedAkeley, R.V., Mills, W.R.. Cunningham, C.E. and Watts, J. 1968. Lenape: a new potato
variety high in solids and chipping quality. Amer. Potato J. 45:142-15 1.Defernez, M., Gunning, Y.M., Parr, A.J., Shepherd, L.V., Davies, H.V. and Colquhoun,
I.J. 2004. NMR and HPLC-UV profiling of potatoes with genetic modifications tometabolic pathways. J. Agri. Food Chem. 52:6075-6085.
Gregory, P. 1984. Glycoalkaloid composition of potatoes: diversity and biologicalimplications. Amer. Potato J. 61:115-122.
Kuc, J. 1984. Steroid glycoalkaloids and related compounds as potato quality factors.Amer. Potato J. 61:123-135.
McCue, K.F., Corsini, D.L., Allen, P.V., Rockhold, D.R., Maccree, M.M., Shepherd,L.V.T.. Moehs, C.P., Joyce, P., Davies, H. and Belknap, W.R. 2003. Reduction oftotal steroidal glycoalkaloids in potato tubers using antisense constructs of a geneencoding a solanidine glucosyl transferase. Acta Hort. 619:77-86.
McCue, K.F., Shepherd, L.V.T., Allen, P.V., Maccree, M.M., Rockhold, D.R., Corsini, D.,Davies, H. and Belknap, W.R. 2005. Metabolic compensation of steroidalglycoalkaloid biosynthesis in transgenic potato tubers: using reverse genetics toconfirm the in vivo enzyme function of a steroidal alkaloid galactosyltransferase.Plant Sci. (Shannon) 168:267-273.
McCue, K.F., Allen, P.V., Shepherd, L.V.T., Blake, A., Whitworth, J., Maccree, M.M.,Rockhold, DR., Stewart, D., Davies, H. and Belknap, W.R. 2006. The primary in vivo
346
steroidal alkaloid glucosyltransferase from potato. Phytochemistry 67:1590-1597.McCue, K.F., Allen, P.V., Shepherd, L.V.T., Blake, A., Maccree, M.M., Rockhold, DR.,
Novy, R., Stewart, D., Davies, H. and Belknap. W.R. 2007. Potato glycosterolrhamnosyltransferase, the terminal step in triose side chain biosynthesis.Phytochemistry 68:327-334.
Moehs, C.P., Allen, P.V., Friedman, M. and Belknap, W.R. 1997. Cloning and expressionof solanidine UDP-glucose glucosyltransferase from potato. Plant J. 11 :227-236.
Paczkowski, C., Kalinowska, M. and Wojciechowski, Z.A. 1998. The 3-0-glucosylationof steroidal sapogenins and alkaloids in eggplant (So/an urn melon gena); evidence fortwo separate glucosyltransferases. Phytochemistry 48:1151-1159.
Rickey, T.M. and Belknap, W.R. 1991. Comparison of the expression of several stress-responsive genes in potato tubers. Plant Mol. Biol. 16:1009-1018.
Sinden, S.L., Sanford. L.L. and Webb, R.E. 1984. Genetic and environmental control ofpotato glycoalkaloids. Amer. Potato J. 61:141-155.
Stapleton, A., Allen, P.V., Friedman, M. and Belknap, W.R. 1991. Purification andcharacterization of solanidine glucosyltransferase from the potato (Solanurntuberosum). J. Agri. Food Chem. 39:1187-1193.
Verwoerd, T.C., Dekker, B.M. and Hoekema, A. 1989. A small-scale procedure for therapid isolation of plant RNAs. Nucleic Acids Res. 17:2362.
Zimowski, J. 1996. Enzymatic glycosylation of tomatidine in tomato plants. Adv. Exp.Med. Biol. 404:71-80.
Tables
Table 1. SGT I and SGT2 substrate preferences. Glycosyltransferase activity of therecombinant SGT:his fusion proteins using UDP-H]-glucose or UDP-[3H]-galactoseas sugar donors, and solanidine, solasodine and tomatidine as steroid glycoalkaloidreceptors. Data from McCue et al., 2005; McCue et al., 2006.
Receptor SGT SGT2(nmol .min'•mg)(nmol.min' mg')
UDP-[3H]-GlucoseSolanidine NA 18.1Solasodine 2.7 14.9Tomatidine 3.0 15.8
UDP-[3H]-GalactoseSolanidine 6.4 NA -Solasodine 16.8 NATomatidine 20.9 NA
Values represent the mean of duplicate assays corrected for the control reaction using DMSO in theabsence of aglycone. NA (no activity) is reported for protein samples with <0.5 nmol .min' .mg' (twicethe background).
1911 IVA
,qMFigures Solanidine
UDP-gal , UDP-gluSGTI""hSGT2
y-solanine,/-chaconine
tJDP-!u 4.—UDP-rha
3-solanine3-chaconine
UDP-rha-_4 SGT3—UDP-rha
cx-solaninecx-chaconine
Fig. 1. SGA biosynthesis. Glycosylation steps from the aglycone solanidine for the twobranches of SGAs. UDP-galactose:solanidine gal actosyl trans ferase (SGT I), UDP-glucose:solanidine glucosyltransferase (SGT2) and UDP-rhamnose: 3-steroidalglycoalkaloid rhamnosyltransferase (SGT3).
NLHVLFLPH IPLVNAARLFASR VK ILTT NALLFRS IDHVLFLPYGHIIPLVNAARLFASR VKVTILTT NAILFRSSIDNV F PYHI PLVRLFAVTINALLF SS D
SGTII"IVI'IY PS EVGLPEGIEF 'YIIKPM ((SIPiS[V(IL' PEVGLP GIEN(511p 1'IIQYPNID['vIPr
C'ITIP'V ,LP NI NPI GSV VC'I LQFI L
PDCIFSDMYFPWTVDIA L IPR L N151 NL Y PEE
I PDCIFSDMYFPWTVDIA L IPR LFNYllI NLR YEPHI P I SDM FPWTVDE IPR FR KP
SGTISIT2S FPGLPD IKFKLSQLTDDL
Dr1iPULPD
F PGLP 1KSI L DS I H T ELK
IT I. Iç)V P NO CSGT2 231 IT C' N 'N P I
I '1 NT
SGT1KPKSVLYVSFGSM FPE L IA L SNVPFI£K KSVLYVSFGS IRFPE L IA L S PFIWV
P MEN'S LLP IWV R
IGT-LED FE L IKGW PQL I NSA CFMTHrGVLEAI DVPMIYL I5(1 LIIKGWAPQL IL lISA GGF'1'IT9flGWI', N1EIIiVrVV E,
N FK II WAPQ IL M, I SlIP TW, rLF"1 'T - S N
SGTI VIIG DVWN I A1R' C'MYIGA IN
SGT 3 IGADVWN PSIRLM
SKMA AT GGSS NLTALI lENSKMAKNA EGGSS NLTALI lENSLKIMD
TEEGGSSL LI 1K
I
Fig. 2. ClustaiW alignment of SGT I. SGT2. and SGT3 deduced amino acid sequencesshowing regions of identity (dark shading) and similarity (light shading).
348
175
150
125
100
75
50
25
0NN
0)U)
CouO -)U,- ,N NCC 0 00C-CDCDNN-CDCDC)C)
C,a<<CCCC
Caa
U)0)0)0)0)COCOC)C)
U)U)(1)0)aa
Plant Line
Fig. 3. SGA content of tubers from potatoes expressing Sgt antisense transgenes. Levelsof u-solanine and a-chaconine suppression and compensation in the Sgt antisenselines, empty vector (pBPA), and two independent Lenape' control lines. Valuesrepresent data for three replicate glass house-grown minitubers.
Wounded TubersWounded LeavesTime (h) Time (h)
0 2 6 16 24 480 25 16
Sgt Sgt
Sgt2 Sgt2
Sgt3 WSgt
RNA L19Fii 1th RNA
Fig.
4. Wound induction of Sgt], Sgt2 and g1$ RNA in control and wounded potatotubers and leaves. 'Lenape' tubers and leaves were wounded for the timesindicated and total RNA (20 pg/lane tubers, 10 pg/lane leaves) was probed withSgtl, Sgr2 and Sgt3. The ethidium bromide-stained gel is shown as a loadingreference.
349
