Insertional polymorphism in introns 4 and 10 ofthe maize β-glucosidase gene glu1

Asim Esen and Hema Bandaranayake

Abstract: The majorβ-glucosidase isozyme Glu1 is encoded by a highly polymorphic gene (glu1) in maize. Theglu1gene comprises 12 exons and 11 introns. Two of these introns, introns 4 and 10, show insertional polymorphism: thosein allele glu1-1 (represented by inbred line OH7B) are longer than those in other inbred genotypes and in two teosintes(Zea mexicanaand Zea parviglumis) surveyed. Sequence data revealed that an increase in the length of intron 4 from150 to 477 bp in OH7B is due to a short (11 bp) tandem duplication and a large insertion sequence of 313 bp plus a4-bp (5′ ATAG 3′) direct repeat. The 313-bp insertion sequence (referred to as mzsTn-1) has all the features of atransposon, having a 25-bp well-conserved (3/25 mismatches) inverted repeat sequence at its termini flanked by a 4-bpdirect repeat. The increase in length from 1041 to 1302 bp in intron 10 of OH7B is due to a 259-bp insertion sequence(referred to as mzsTn-2) plus a 2-bp (5′ TA 3′) direct repeat. The mzsTn-2 element also possesses all the hallmarks ofa transposon: a 34-bp well-conserved (3/34 mismatches) inverted repeat sequence at its termini flanked by a 2-bpdirect repeat. The mzsTn-1 element belongs to a new family of inverted repeat elements, while mzsTn-2 belongs to theStowawayfamily of inverted repeat elements. Analysis of PCR products from amplifications off genomic-DNAtemplates, using primers derived from the inverted repeat termini, and Southern blotting data suggest that both smalltransposons are members of a multigene family. The occurrence of two different small transposons in introns of thesameglu1 allele in inbred OH7B and their absence in other genotypes suggest that they have moved into thisglu1allele recently through mediation of their autonomous counterparts that are active in OH7B or in its ancestry.

Key words: β-glucosidase, maize, intron, small transposon, polymorphism.

Résumé: L’isoenzyme majeure de laβ-glucosidase (Glu1) est codée par un gène très polymorphe (glu1) chez lemaïs. Le gèneglu1 comprend 12 exons et 11 introns. Deux de ces introns, les introns 4 et 10, montrent unpolymorphisme insertionnel : ils sont de plus grande taille chez l’allèleglu1-1 (présent chez la lignée fixée OH7B) quechez les allèles présents chez les autres génotypes et chez les deux téosintes (Zea mexicanaet Zea parviglumis)examinés. L’étude de la séquence nucléotidique a révélé que l’accroissement de la taille de l’intron 4 (de 150 à 477 pbchez OH7B) est dû à une courte séquence (11 pb) répétée en tandem de même qu’à une grande insertion de 313 pbqui s’accompagne de la duplication d’un segment de 4 pb (5′ ATAG 3′). La séquence d’insertion de 313 pb (appeléemzsTn-1) montre toutes les caractéristiques d’un transposon : des séquences terminales répétées inversées mesurant 25pb et qui sont bien conservées (3 mésappariements sur 25) de même que la présence de séquences répétées de 4 pb depart et d’autre de l’insertion. L’accroissement de la taille de l’intron 10 (de 1041 à 1302 pb chez OH7B) est attribuableà une séquence d’insertion de 259 pb (appelée mzsTn-2) qui est bordée par une répétition de 2 pb (5′ TA 3′).L’élément mzsTn-2 possède également toutes les caractéristiques d’un transposon : des séquences terminales répétéesinversées mesurant 34 pb et qui sont bien conservées (3 mésappariements sur 34) de même que la présence deséquences répétées de 2 pb de part et d’autre de l’insertion. L’élément mzsTn-1 appartient à une nouvelle familled’éléments à séquences inversées répétées alors que l’élément mzsTn-2 appartient à la famille des élémentsStowaway.Une analyse des produits d’amplification PCR réalisées sur de l’ADN génomique avec des amorces dérivées desséquences terminales et des analyses Southern suggèrent que ces deux petits transposons sont membres de famillesmultigéniques. La découverte de deux petits transposons différents dans des introns du même allèleglu1 chez OH7B etleur absence chez d’autres génotypes suggèrent qu’ils se sont insérés dans cet allèle récemment par suite de l’activitédes éléments autonomes correspondants, lesquels seraient actifs chez OH7B ou chez ses ancêtres.

Mots clés: β-glucosidase, maïs, intron, petit transposon, polymorphisme.

[Traduit par la Rédaction] Esen and Bandaranayake 604

Genome41: 597–604 (1998) © 1998 NRC Canada

597

Corresponding Editor: R.J. Kemble.

Received September 23, 1997. Accepted March 31, 1998.

A. Esen1 and H. Bandaranayake.Department of Biology, Virginia Polytechnic Institute and State University, Blacksburg,VA 24061–0406, U.S.A.1 Author to whom all correspondence should be addressed (e-mail: esen@vt).

Gen

ome

Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

Dep

osito

ry S

ervi

ces

Prog

ram

on

11/1

7/14

For

pers

onal

use

onl

y.

β-Glucosidase (β-D-glucoside glucohydrolase, EC 3.2.1.21)catalyzes the hydrolysis of aryl and alkylβ-D-glucosides, aswell as short chainβ-linked oligosaccharides (e.g., cellobioseand laminoribose, etc.). The enzyme occurs widely inprokaryotes and eukaryotes. In plants,β-glucosidases are im-plicated in a variety of functions. These include: (i) defenseagainst herbivores and other pests through the release ofcoumarins, saponins, thiocyanates, HCN, hydroxamic acids,and terpenes (Bell 1981; Conn 1981; Niemeyer 1988; Nisius1988; Poulton 1990; Sahi et al. 1990); (ii ) hydrolysis ofglucoconjugates of gibberellins, auxins, abscisic acid, andcytokinins (Smith and van Staden 1978; Schliemann 1984;Matsuzaki and Koiwai 1986; Wiese and Grambow 1986;Brzobohaty et al. 1993); (iii ) lignification (Dharmawardhanaet al. 1995); and (iv) hydrolysis ofβ-linked oligosaccharidesresulting from cellulolysis during seed germination (Leahet al. 1995).

In maize,β-glucosidase is encoded by two nuclear genes(glu1 and glu2) and localized in plastids. Theglu1 gene isthe most polymorphic (>30 alleles) gene on record in maizeor any other organism (Goodman and Stuber 1983). Incontrast, theglu2 gene is not polymorphic (M. Shahid andA. Esen, unpublished). The major function ofβ-glucosidaseappears to be in defense against pests by releasing toxichydroxamic acids (mostly 2,4-dihydroxy-7-methoxy-1,4-benzoxazin-3-one (DIMBOA)) from their glucosides.DIMBOA–glc constitutes up to 1% of the dry weight inyoung maize parts, making it the most abundant hydroxamicacid glucoside (80–90% of total) and also the most abundantphysiological substrate.β-Glucosidase and the substrate(DIMBOA–glc) are physically separated in different com-partments within the cell. Physical injury and cell lysis dis-rupt the compartmentalization and bring the enzyme and thesubstrate together, leading to the release of the toxicaglycone DIMBOA.

Transposable elements have been found to be an integralpart of any genome that has been carefully surveyed so far.Active and inactive transposable elements are believed toconstitute about 50% of the maize genome (Freeling 1984).Examples of virtually all types of transposable elements areencountered in plants (Wessler et al. 1995). Of particular in-terest among them are the miniature inverted repeattransposable elements (MITEs). These elements have well-defined terminal inverted repeats; short central regions,which vary in length from <80 to 343 bp and are unlikely tocode for any protein; internal complementarity to form ahairpinlike secondary structure; and target sequences that are2–3 nucleotides long. The best documented cases of MITEsin plants are theTourist (113–343 bp) andStowaway(80–323 bp) families (Bureau and Wessler 1994a, 1994b). Bothfamilies are widespread in both monocots and dicots, andthey are typically found in introns and flanking regions ofgenes and contribute to the regulation of plant genes, in ad-dition to other effects.

To understand the molecular basis of the extensive multi-ple allelism at theglu1 gene leading to >31 allozymes, wehave sequenced theglu1 cDNAs and genomic DNAs fromselected inbred lines that are homozygous for particularGlu1 allozymes. During the course of such research, we

have also determined the genomic sequence and organiza-tion of theβ-glucosidase geneglu1 in the maize inbred lineK55 (H. Bandaranayake and A. Esen, GenBank accessionNo. U44773, direct submission). The transcribed portion ofthe glu1 gene comprises 12 exons interrupted by 11 intronsand shows high similarity to the genomic organization oftwo myrosinase (β-thioglucosidase) genes fromArabidopsisthaliana (Xue et al. 1995) and to that ofβ-glucosidase genesfrom barley (Leah et al. 1995) and cassava (S. Liddle, J.Hughes, and M.A. Hughes, GenBank accession No. X94986,direct submission). When we examined the organization ofthe glu1 gene in other maize genotypes (inbred lines) and inZea mexicanaand Zea parviglumisby amplifying overlap-ping fragments from genomic DNA templates, we found thatintrons 4 and 10 were approximately 300 and 250 bp, re-spectively, longer in a maize inbred line (OH7B) than inother genotypes. The purpose of this study was to elucidatethe molecular basis of length variation in introns 4 and 10 ofthe glu1 gene, as well as that of the electrophoretic mobilitydifferences between allozymes encoded by theglu1 alleles 1(inbred line OH7B) and 7 (inbred line K55). We report herethat the length increase in both introns is due to insertions ofsmall transposonlike elements, and that two amino acid sub-stitutions explain the mobility difference between the Glu1allozymes 7 and 1.

DNA isolation, amplification, and sequencingGenomic DNA of the maize inbred lines K55, B73, and OH7B

and maize relativesZ. parviglumis(presumptive maize ancestor)and Z. mexicanawas isolated according to a miniprep procedure(Dellaporta 1994). The entire span of theglu1 gene was amplifiedby PCR in three overlapping fragments from genomic DNA tem-plate of inbred line K55. Amplification primers were derived fromselected regions of the K55glu1 cDNA (allele 7) and also fromparts of theglu1 genomic DNA that had been sequenced earlier inour laboratory (A. Esen and M. Shahid, unpublished; H.Bandaranayake and A. Esen, unpublished). The three PCR prod-ucts were sequenced in both directions by the cycle-sequencingmethod (Epicentre Technologies, Madison, Wis.). Similarly, thegenomic segment of the OH7Bglu1 gene (allele 1) was amplifiedin three overlapping fragments using the same primers as thoseused for K55. The regions corresponding to 12 exons, and to intronregions flanking the exons, were sequenced. In addition, the twoinsertion sequences (mzsTn-1 and mzsTn-2) responsible forinsertional polymorphism in OH7Bglu1 introns 4 and 10 were se-quenced. Sequence data were checked for similarity against thosein the data base, using the BLAST software. In these data-basesearches, the entire insertion wase used as the query sequence. Thesequences that showed high similarity to mzsTn-1 and mzsTn-2were retrieved and aligned using the J. Hein alignment method inthe software MEGALIGN (DNASTAR).

Amplifications of introns 4 and 10 and their flankingregions

The regions of theglu1 gene that include introns 4 and 10 wereamplified from genomic DNA templates of K55 (allele 7), B73 (al-lele 7), OH7B (allele 1),Z. parviglumis, and Z. mexicana. Thegenomic fragment including intron 4 was amplified using theprimer pair β-glu94 (CTGCGAATATATATTATTGCTAC) andβ-glu81 (GTGATCGAGTCTTTGTGATC). Similarly, the genomicfragment including intron 10 was amplified from DNA of the same

© 1998 NRC Canada

598 Genome Vol. 41, 1998

Gen

ome

Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

Dep

osito

ry S

ervi

ces

Prog

ram

on

11/1

7/14

For

pers

onal

use

onl

y.

genotypes using the primer pairβ-glu84 (GCTAGATGAGTCA-CTATGAC) and β-glu85 (TGTTTGAGTCAATATCTCGCA).PCR was performed in a MJ Research PT-100 thermocycler. Reac-tion conditions were initial denaturation at 94°C for 2 min, fol-lowed by 36 cycles of denaturation at 94°C for 20 s, annealing at58°C (forβ-glu84 andβ-glu85) and 60°C (forβ-glu94 andβ-glu81)for 20 s, and extension at 72°C for 3 min, and then a final exten-sion at 72°C for 7 min. The amplification products wereelectrophoresed through 1% agarose gels, visualized by ethidiumbromide staining, and analyzed. In addition, a primer (β-glu105,GGGTSTGTTTGGTTKGGCT) was designed and used to amplifythe intron-4 insertion, while another primer (β-glu106, CTCCCT-CCGRTTTCCTATTAG) was designed and used to amplify theintron-10 insertion. These primers were derived from the invertedrepeat region of each transposon, so that a single primer could beused to specifically amplify the mzsTn-1 and mzsTn-2 related se-quences from the genomic DNA templates of K55, B73, OH7B,Z. parviglumis, and Z. mexicana. Reaction conditions were thesame as stated above, except that annealing was at 46oC. The an-nealing temperature used was 12–16°C below the calculatedTm ofeach primer, to allow amplification of divergent members of eachfamily in the genome. The amplicons were electrophoresedthrough 1% agarose gels and visualized by ethidium bromide stain-ing.

Southern blot analysisGenomic DNAs of OH7B and K55 were digested with the re-

striction enzymesBamHI, EcoRI, HindIII, KpnI, and PstI. About20 µg of each digest was electrophoresed through a 0.8% agarosegel, blotted onto Nytran filters, and probed with32P-labeled (ran-dom primer labeling) OH7B intron 4 and intron 10 insertionsequence DNAs, which were amplified by PCR using the primersβ-glu105 and 106, respectively. Hybridization was carried out at65°C and the filters were washed in 1× SSC (1× SSC: 0.15 MNaCl plus 0.015 M sodium citrate) – 0.1% SDS, followed by 0.5×SSC – 0.1% SDS.

Organization of the glu1 gene and differences betweenalleles 1 and 7

The glu1 gene of maize spans a length of about 5 kb(4939 bp) comprising 12 exons and 11 introns. The 12 exonsspan a length of about 2 kb. Alleles 7 (K55) and 1 (OH7B)differ from each other by 12 nucleotide substitutions in theircoding regions. Three of these substitutions result in aminoacid substitutions: two in the mature protein and one in thetransit peptide. Two amino acid substitutions (K107 to N107and A423 to D423 in the mature protein) explain the electro-phoretic mobility difference between allozymes Glu1-1(OH7B) and Glu1-7 (K55), where the former has a higherelectrophoretic mobility than the latter in alkaline nativePAGE (polyacrylamide gel electrophoresis) (data notshown), owing to the two extra negative charges.

Insertional polymorphism in introns 4 and 10That insertional polymorphism was present in introns 4

and 10 of theglu1 gene was a surprising result. When wecompared the sizes of the PCR products containing introns 4and 10, they were 742 and 2242 bp, respectively, in OH7B(Fig. 1, lanes 2 and 8) versus 415 and 1981 bp, respectively,in the other genotypes (Fig. 1, lanes 1, 3–5, 7, and 9–11). In-terestingly, Z. parviglumisgenomic DNA template yielded alonger PCR product (Fig. 1, lanes 5 and 11) in addition to

the one that is the same size as that found in genotypeswithout insertion in introns 4 and 10. These data suggest thatadditional length polymorphism exists in introns 4 and 10 orin other introns of theglu1 gene, and thatZ. parviglumis,not being an inbred genotype, is heterozygous for differentlength variant alleles.

Intron 4 insertion and its relation to other smalltransposons

After sequencing the PCR products that contain introns 4and 10, it was found that intron 4 was 478 bp long in theOH7B glu1 gene compared with a length of 150 bp in K55and apparently also in four other genotypes included in theanalysis. The increase in the length of intron 4 in OH7B wasdue to a short tandem duplication of the sequenceAAAAATGATCT (11 bp) and to a large insertion sequenceof 313 bp plus a 4-bp (5′ ATAG 3′) direct repeat after nucle-otide 16 near the 5′ end of intron 4 (Fig. 2). The 313-bp in-sertion sequence (referred to as mzsTn-1 hereinafter) has allthe features of a transposable element, having a 25-bp well-conserved (3/25 mismatches, Fig. 3) inverted repeat se-quence at its termini flanked by a 4-bp direct repeat. Asearch of the data base using the BLAST software revealedthat mzsTn-1 had homology to sequences found in the 5′-flanking regions of three other maize genes,waxy (wx), U3small nuclear (sn) RNA (mu3.7), and B-I. It had 89% se-quence identity to a transposonlike 318-bp insertion se-

© 1998 NRC Canada

Esen and Bandaranayake 599

Fig. 1. Length polymorphism in introns 4 and 10 of theglu1gene among genotypes of maize and related taxa. Lane M, DNAsize markers; lanes 1–5: amplification products obtained usinggenomic DNA templates and a primer pair (β-glu81/β-glu94)derived from intron 4 sequence bracketing the insertion site ofmzsTn-1: lane 1, K55; 2, OH7B; 3, B73; 4,Z. mexicana; and 5,Z. parviglumis; lanes 7–11: amplification products obtained usinggenomic DNA templates and a primer pair (β-glu84/β-glu85)bracketing the insertion site of mzsTn-2 in intron 10: lane 7,K55; 8, OH7B; 9, B73; 10,Z. mexicana; 11, Z. parviglumis; and6 and 12, no DNA template (negative) controls.

Gen

ome

Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

Dep

osito

ry S

ervi

ces

Prog

ram

on

11/1

7/14

For

pers

onal

use

onl

y.

quence (GenBank accession number X06934) found 1.6 kbupstream of thewx gene in the maize inbred line W23 (Spellet al. 1988). The difference between mzsTn-1 and itshomologue near thewx gene is that the latter has a 15-bp in-verted repeat that is less perfect (3/15 mismatches, Fig. 3).One can recognize a 25-bp terminal inverted repeat in thewxversion of the mzsTn-1 as well, but the 5′-end repeat unit ap-pears to have nucleotide substitutions, making it consider-ably different (8/25 mismatches) from theglu1 (3/25mismatches) andmu3.7(4/25 mismatches) versions (Fig. 3).In fact, the 5′-end units of the terminal inverted repeats areidentical in theglu1 andmu3.7mzsTn-1s, but they each dif-fer from thewx andB-I mzsTn-1 5′-end units (Fig. 3). At the3′ end, theglu1 and mu3.7mzsTn-1 repeat units differ bytwo nucleotides, but again they each differ from thewx andB-I mzsTn-1s by up to 9 nucleotides, indicating much higherdivergence at the 3′ ends of these homologous elements. It isconceivable that the 25 bp long terminal inverted repeat wepostulate is actually 15 bp, as proposed by Spell et al.(1988),and that the remaining 10-bp segment is involved in the for-mation of intra-strand DNA secondary structures (e.g., hair-pin and stem–loop), as suggested by Bureau and Wessler(1992) for members of theTourist family of small transposons.

The OH7B mzsTn-1 also shows 83% sequence identitywith a maize U3 snRNA pseudogene (mu3.7, GenBank ac-cession No. Z29642) antisense strand (thus the insertion is inreverse orientation), spanning nucleotides 882 through 1193(Leader et al. 1994). As mentioned above, the 25-bp in-verted repeat sequence shows perfect identity between theOH7B glu1 (intron 4) mzsTn-1 and themu3.7mzsTn-1 atthe 5′ end (25/25) and near perfect identity (23/25) at the 3′end. Themu3.7mzsTn-1 is flanked by a 3-bp (5′ TAA 3′) di-rect repeat sequence, the same as the one that occurs in thewx version of mzsTn-1 but different from that of theglu1version (5′ ATAG 3′). Although inverted repeat termini ofthe glu1 mzsTn-1 have higher sequence identity with thoseof themu3.7mzsTn-1 than with those of thewx mzsTn-1, itsinternal (core) segment sequence shares higher sequenceidentity with thewx mzsTn-1 than with themu3.7mzsTn-1.

The mu3.7mzsTn-1 was inserted 34-bp upstream of the up-stream sequence element (USE) and 69-bp upstream of theTATA box, based on the location of these promoter elementsin the wild-type allele of the U3 snRNA gene (mu3.8). Con-ceivably, the insertion produced a transcriptionally inactiveU3 snRNA allele (i.e., pseudogene) either instantly or overtime.

Another data-base entry with which the OH7Bglu1mzsTn-1 has sequence similarity (45%) is the 5′-flanking re-gion of the maize regulatory geneB-I (GenBank accessionNo. X70790), which codes for a transcription factor in-volved in the regulation of anthocyanin biosynthesis(Radicella et al. 1992). In this case, the insertion is 252 bplong and is about 300-bp upstream of the TATA box, with noapparent effects on the expression of theB-I gene. Althoughthe sequence similarity between the OH7Bglu1 mzsTn-1and theB-I gene mzsTn-1 is somewhat low (45%), owing to5 deletions ranging in length from 7 to 23 bp, it is unlikelyto be fortuitous. The 5′-end unit of the terminal inverted re-peat inB-1 mzsTn-1 appears to have a 7-bp (CTTTTTT) in-sertion (Fig. 3, underlined). The postulated repeat unit showshigher similarity (21/25) to the repeat units of theglu1 andmu3.7mzsTn-1s when this 7-bp region is not included in thesequence comparison. Thus, the comparisons of the fourmembers of the mzsTn-1 family suggest that it is an oldtransposon family and that its variants of different ages arescattered throughout the maize genome. TheB-I genemzsTn-1 may represent a much earlier insertion event, whilethe glu1 gene mzsTn-1 may represent a rather recent inser-tion, because of its nearly perfect terminal inverted repeatsand its absence from the alleles of theglu1 gene in othermaize inbred lines and teosintes. Moreover, the cladogramderived from the alignment data (Fig. 3) indicates that thewx andglu1 mzsTn-1s are derived most recently and from acommon ancestor, which links them to themu3.7mzsTn-1,and thus these three mzsTn-1s represent a lineage. The samecladogram suggests that the lineage leading to theB-ImzsTn-1 separated from the lineage leading to the thewx,glu1, andmu3.7mzsTn-1s a long time ago.

© 1998 NRC Canada

600 Genome Vol. 41, 1998

Fig. 2. Diagrams showing length polymorphism in introns 4 and 10 of theglu1 gene between two maize genotypes (OH7B, top; K55,bottom). The locations and lengths of mzsTn-1 and an 11-bp duplication (Duplic.) in intron 4, and of mzsTn-2 in intron 10, of OH7Bare shown as solid blocks, while the original introns 4 and 10 before insertion (as in K55) are shown as clear blocks. Exons flankingeach intron are shown as stippled blocks. All diagrams are drawn to scale.

Gen

ome

Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

Dep

osito

ry S

ervi

ces

Prog

ram

on

11/1

7/14

For

pers

onal

use

onl

y.

Intron 10 insertion and its relation to other smalltransposons

As for the 259-bp insertion sequence (referred to asmzsTn-2) in intron 10 of the OH7Bglu1 gene, its insertionwas responsible for increasing the length of this longestglu1intron from 1041 to 1302 bp, making it one of the longest

three introns found in plant genes (Fig. 1, lane 8; Fig. 2; L.J.Destefano, personal communication). It is apparent thatmzsTn-2 is a member of theStowawayfamily of inverted re-peat elements that has recently been catalogued and de-scribed (Bureau and Wessler 1994b). The Stowawayfamilyelements occur widely in both monocots and dicots and vary

© 1998 NRC Canada

Esen and Bandaranayake 601

Fig. 3. Alignment of mzsTn-1 (313 bp) homologues found near thewx gene (Spell et al. 1988) and in theglu1 (this study),mu3.7(Leader et al. 1994), andB-I (Radicella et al. 1992) genes, all of maize. The white letters with black background in the alignmentshow identities among the sequences of the four different elements, while the black letters with white background show thedifferences. Postulated 5′ and 3′ inverted repeat termini are boxed. The cladogram below the alignment shows the phylogeneticrelationships between mzsTn-1 (glu1) and its homologues.

Gen

ome

Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

Dep

osito

ry S

ervi

ces

Prog

ram

on

11/1

7/14

For

pers

onal

use

onl

y.

in length from 80 to 323 bp. The mzsTn-2 element also pos-sesses all the hallmarks of a transposable element: a well-conserved (3/34 mismatches) 34-bp inverted repeat sequenceat its termini flanked by a 2-bp (TA) direct repeat. However,mzsTn-2 differs from otherStowawayelements by having a34-bp terminal inverted repeat. The 5′-end of the 34-bp re-gion contains an 11-bp consensus terminal inverted repeatthat is found in other members of the family (Bureau andWessler 1994b). It is conceivable that mzsTn-2 represents anew subfamily of theStowawayfamily, different from the

previously described members of that family. Alternatively,the 23-bp downstream region that follows the 11-bp consen-sus region in the 34-bp terminal inverted repeat is involvedin the formation of such intrastrand DNA secondary struc-tures as hairpins and stem–loops. A BLAST search of thedata base for sequences showing similarity to mzsTn-2 pro-duced three hits (Fig. 4): the sorghum PEP carboxylase gene(Lepiniec et al. 1993; GenBank accession number X65137),a 30-kb rice genomic fragment containing the ADP-glucosepyrophosphorylase subunitSh2 and NADPH-dependent

© 1998 NRC Canada

602 Genome Vol. 41, 1998

Fig. 4. Alignment of mzsTn-2 homologues found in theglu1 gene of maize (this study), in the PEP carboxylase gene of sorghum(Lepiniec et al. 1993), in a 30-kb rice genomic fragment containing the ADP-glucose pyrophosphorylase subunitSh2and the NADPH-dependent reductaseA1 genes (Chen and Bennetzen 1997), and in the rice cyclophilin 2 (cyp2) gene (Bucholz et al. 1994). Theglu1mzsTn-2 antisense strand is aligned with the sense strands of its sorghum and rice homologues. The white letters with blackbackground in the alignment show identities among the sequences of the four different elements, while the black letters with whitebackground show the differences. Postulated 5′ and 3′ inverted repeat termini are boxed. The cladogram below the alignment shows thephylogenetic relationships between mzsTn-2 (glu1) and its homologues.

Gen

ome

Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

Dep

osito

ry S

ervi

ces

Prog

ram

on

11/1

7/14

For

pers

onal

use

onl

y.

reductaseA1 genes (Chen and Bennetzen 1996; GenBankaccession No. U70541) and the rice cyclophilin 2 (cyp2)gene (Bucholz et al. 1994; GenBank accession No. Z29469).The antisense strand of mzsTn-2 in intron 10 had the highestsequence similarity to the sense strands of its homologues insorghum and rice, indicating that its orientation is invertedwith respect to its homologues. The insertion site in the PEPcarboxylase gene (SVPEPCG) is 839 bp upstream of thetranscription start site, apparently with no effects on the ex-pression of the gene. mzsTn-2 and its sorghum homologuehave 67% sequence identity, but the latter is probably nolonger capable of transposing, because nearly half the lengthof its 3′-end inverted repeat sequence is modified by shortinsertions and nucleotide substitutions. Surprisingly, mzsTn-2 shares 64 and 63% sequence identity, respectively, with itsrice homologue found in thecyp2gene and in the fragmentcontaining theSh2and A1 genes. The mzsTn-2 homologuein the cyp2 gene encompasses the region from nucleotide121 to nucleotide 382, which is about 850 nucleotides up-stream of the coding region, while the one near thea1 geneis 173 nucleotides downstream of the stop codon. In bothcases the 3′-end inverted repeat unit could not be identifiedwith certainty (Fig. 4). Interestingly, the cladogram derivedfrom the alignment data predicts a closer relationship (amore recent divergence from a common ancestor) betweenSVPEPCG (sorghum) andcyp2 (rice) mzsTn-2 homologuesthan between SVPEPCG andglu1. This close relationship isobviously an artifact, because maize and sorghum belong tothe same grass tribe (Andropogoneae) within the subfamilyPanicoideae and diverged from a common ancestor 20 mil-lion years ago, while sorghum and rice not only belong todifferent subfamilies of the Poaceae but also diverged fromeach other 50 million years ago (Bennetzen and Freeling1993). The extensive mismatch between the inverted repeattermini of PEP carboxylase insertion and the difficulty ofidentifying the 3′-end inverted repeat element of the ricecyp2anda1 gene insertions suggest that these elements areprobably no longer capable of transposing. Our BLASTsearch using the entire mzsTn-1 element as the query se-quence failed to reveal any of theStowawayelements thatoccur in nearly 50 different plant genes (the data-base searchof Bureau and Wessler 1994b), except the three mentionedabove. This was probably due to the fact that our mzsTn-2showed very low homology to other members of theStow-away family.

Multiplicity of maize small transposonsSouthern blot data (not shown) indicated that mzsTn-1

and mzsTn-2 occur in multiple copies in the genome, be-cause digests of all five restriction enzymes yielded morethan 30–40 bands hybridizing with each probe. However,fragment-size multiplicity was much greater in blots probedwith mzsTn-1 than in those probed with mzsTn-2. Bureauand Wessler (1992) found the copy number of selectedTour-ist family members to be 103 to ~5 × 104 in the genome. Asimilar degree of multiplicity is likely to be the case with theStowaway family, as well as with the family to whichmzsTn-1 belongs, as is evident from the many bands hybrid-izing with a specific probe on Southern blots (Spell et al.1988; this study). Clearly, mzsTn-2 is aStowawayfamilymember, based on the similarity of its inverted repeat and di-

rect repeat sequences to those of theStowawayfamily, and itis expected to occur in high-copy number.

The analysis of PCR amplicons resulting from the use ofsingle primers derived from the inverted repeat termini re-vealed numerous fragments ranging in size from 313 to ca.2500 bp, in the case of mzsTn-1 related sequences, and from259 to ca. 2000 bp, in the case of mzsTn-2 related sequences(data not shown). The amplicon profiles of all 5 genotypeshad weakly staining bands, including the expected specific313- and 259-bp amplicons. These data also suggest that allthe maize genotypes and the two teosintes tested had multi-ple copies of mzsTn-1 and mzsTn-2 related sequences intheir genomes. Although we have not attempted to deter-mine the copy numbers of mzsTn-1 and mzsTn-2, it is likelythat a given band detected by Southern blot analysis and af-ter PCR amplification includes fragments corresponding totwo or more subfamilies.

Significance and implicationsOur data showing the involvement of two different small

transposons (mzsTn-1 and mzsTn-2) in producing lengthpolymorphism in two introns of theglu1 gene have the fol-lowing broader implications. (i) They support the hypothesisthat there are many high copy number short transposable el-ement families, which were named MITEs by Wessler et al.(1995), in the genomes of maize and other flowering plants.(2) Short transposons must have played an important role ingene mutation and genome evolution, as is evident from datashowing their preponderance in both monocots and dicots.Most notably, the data-base searches and analysis conductedby the laboratory of S.R. Wessler (Bureau and Wessler1994a, 1994b) revealed the existence of numerous smalltransposon families, whose members were in the flanking re-gions and introns of many plant genes. Movements of thesesmall transposons into exons would produce either crypticsplice sites or nonfunctional mRNAs and (or) protein prod-ucts, often having deleterious and lethal consequences. Inthe OH7Bglu1 gene, the mzsTn-1 insertion in intron 4 andthe mzsTn-2 insertion in intron 10 are apparently at posi-tions not affecting the processing (e.g., recognition and exci-sion of introns) of the precursor mRNA by the spliceosomemachinery. In plant introns, the basis for recognition ofintrons by the spliceosome machinery prior to excision ap-pears to be in AU-rich sequences upstream of the 3′-end ac-ceptor site (AG) (Luehrsen and Walbot 1994a, 1994b);mzsTn-1 is 58% AU-rich, while mzsTn-2 is 71% AU-rich. Apractical use of insertional polymorphism in introns wouldbe for cultivar or genotype identification. If a catalogue ofallele-specific small transposons that are present in intronsof genes is developed, one could amplify such regions byPCR using primers flanking the insertion sites and discrimi-nate cultivars or genotypes from each other on the basis ofthe presence versus absence of such small transposons.

The question of whether or not mzsTn-1 and mzsTn-2 arebona fide transposable elements will remain unanswered un-til their movement is demonstrated. Such demonstrationwould have been straightforward had an easily scorable mu-tant phenotype been produced by their movement. The factthat these two unrelated small transposons are found in dif-ferent introns of the sameβ-glucosidase allele but are absentin other genotypes suggests that they have recently moved

© 1998 NRC Canada

Esen and Bandaranayake 603

Gen

ome

Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

Dep

osito

ry S

ervi

ces

Prog

ram

on

11/1

7/14

For

pers

onal

use

onl

y.

into their current locations in the OH7Bglu1 allele. Thusthey are capable of moving given the right genetic back-ground. The data also suggest that these two smalltransposons may use a common mechanism of transposition,or are mobilized by the same activator.

mzsTn-1 and mzsTn-2, being 313 and 259 bp long, re-spectively, are too small to code for transposase activity.Therefore, they must be dependent on specific transposasesencoded by other genes that mediate the transposition ofsmall insertion sequence families such as mzsTn-1 andmzsTn-2. Alternatively, both elements are deletion variantsof longer and complete transposons that possess transposasecoding capacity. The OH7B genetic background is thenlikely to have the activator required for transposition ofthese nonautonomous elements. It is postulated thattransposons might constitute up to one-half of the maize ge-nome (Freeling 1984). This would make the maize genome(and other genomes with an active transposon load) a dy-namic “playground” for numerous small- and large-transposon families, so that genomes are locked into a pathof continuous random change.

The sequence data pertaining to the small transposons,mzsTn-1 and mzsTn-2, reported in this paper appear in theGenBank data base under the accession numbers U60560and U60561, respectively.

This research was supported by National Science Founda-tion grants (IBN-9118770 and IBN-9318134) to A.E. Theauthors also thank Professor John E. Doebley for seeds ofthe teosinte accessions used in this study.

Bell, A. 1981. Biochemical mechanisms of disease resistance.Annu. Rev. Plant Physiol.32: 21–81.

Bennetzen, J.L., and Freeling, M. 1993. Grasses as a single geneticsystem: genome composition collinearity and compatibility.Trends Genet.9: 259–261.

Brzobohaty, B., Moore, I., Christofferson, P., Bako, L., Campos,N., Schell, J., and Palme, K. 1993. Release of active cytokininby a β-glucosidase localized to the maize root meristems. Sci-ence (Washington, D.C.),262: 1051–1054.

Bucholz, W.G., Harris-Haller, L., DeRose, R.T., and Hall, T.C.1994. Cyclophilins are encoded by a small gene family in rice.Plant Mol. Biol. 25: 837–843.

Bureau, T.E., and Wessler, S.R. 1992.Tourist: a large family ofsmall inverted repeat elements frequently associated with maizegenes. Plant Cell,4: 1283–1294.

Bureau, T.E., and Wessler, S.R. 1994a. Mobile inverted-repeat ele-ments of theTourist family are associated with the genes of manycereal grasses. Proc. Natl. Acad. Sci. U.S.A.91: 1411–1415.

Bureau, T.E., and Wessler, S.R. 1994b. Stowaway: a new family ofinverted repeat elements associated with the genes of both mono-cotyledonous and dicotyledonous plants. Plant Cell,6: 907–916.

Chen, M., and Bennetzen, J.L. 1996. Sequence composition andorganization in theSh2/A1-homologous region of rice. PlantMol. Biol. 32: 999–1001.

Conn, E.E. 1981. Cyanogenic glycosides.In Biochemistry ofplants. Vol 7.Edited byP.K. Stumpf and E.E. Conn. AcademicPress, N.Y. pp. 479–500.

Dellaporta, S. 1994. Plant miniprep and microprep; versions 2.1–2.3. In The maize book.Edited byM.Freeling and V. Walbot.Springer–Verlag, N.Y. pp. 522–525.

Dharmawardhana, D.P., Ellis, B.E., and Carlson, J.E. 1995. Aβ-glucosidase from lodgpole pine xylem specific for the ligninprecursor coniferin. Plant Physiol.107: 331–339.

Freeling, M. 1984. Plant transposable elements and insertion se-quences. Annu. Rev. Plant Physiol.35: 277–298.

Goodman, M.M., and Stuber, C.W. 1993. Maize.In Isozymes inplant genetics and breeding. Part B.Edited byS.D. Tanksley andT.J. Oton. Elsevier, New York.

Leader, D.J., Connelly, S., Filipowicz, W., and Brown, J.W.S.1994. Characterization and expression of a maize U3 snRNAgene. Biochim. Biophys. Acta,1219: 145–147.

Leah, R., Kigel, J., Svendsen, I., and Mundy, J. 1995. Biochemicaland molecular characterization of a barley seedβ-glucosidase. J.Biol. Chem.270: 15 789 – 15 797.

Lepiniec, L., Keryer, E., Philippe, H., Gadal, P., and Cretin, C. 1993.Sorghum phosphoenolpyruvate carboxylase gene family: structure,function and molecular evolution. Plant Mol. Biol.21: 487–502.

Luehrsen, K.R., and Walbot, V. 1994a. Addition of A- and U-richsequence increases the splicing efficiency of a deleted form of amaize intron. Plant Mol. Biol.24: 449–463.

Luehrsen, K.R., and Walbot, V. 1994b. Intron creation andpolyadenylation in maize are directed by AU-rich RNA. GenesDev. 8: 1117–1130.

Matsuzaki, T., and Koiwai, A. 1986. Germination inhibition instigma extracts of tobacco and identification of MeABA, ABA,and ABA-β-D-glucopyranoside. Agric. Biol. Chem.50: 2193–2199.

Niemeyer, H.M. 1988. Hydroxamic acids (4-hydroxy-1,4-benzoxazin-3-ones), defence chemicals in the Gramineae.Phytochemistry,27: 3349–3358.

Nisius, A. 1988. The stromacentre inAvenaplastids: an aggrega-tion of β-glucosidase responsible for the activation of oat-leafsaponins. Planta,173: 474–481.

Poulton, J.E. 1990. Cyanogenesis in plants. Plant Physiol.94: 401–405.Radicella, J.P., Brown, D., Tolar, L.A., and Chandler, V.L. 1992.

Allelic diversity of the maize B regulatory gene: different leaderand promoter sequences of two B alleles determine distinct tis-sue specificities of anthocyanin production. Genes Dev.6:2152–2164.

Sahi, S.V., Chilton, M.-D., and Chilton, W.S. 1990. Corn metabo-lites affect growth and virulence ofAgrobacterium tumefaciens.Proc. Natl. Acad. Sci. U.S.A.87: 3879–3883.

Schliemann, W. 1984. Hydrolysis of conjugated gibberellins byβ-glucosidases from dwarf rice (Oryza sativaL. cv. Tan-ginbozu).J. Plant Physiol.116: 123–132.

Smith, A.R., and van Staden, J. 1978. Changes in endogenouscytokinin levels in kernels ofZea maysL. during imbibition andgermination. J. Exp. Bot.29: 1067–1075.

Spell, M.L., Baran, G., and Wessler, S.R. 1988. An RFLP adjacentto the maizewx gene has the structure of a transposable ele-ment. Mol. Gen. Genet.211: 364–366.

Wessler, S.R., Bureau, T.E., and White, S.E. 1995. LTR-retrotransposons and MITEs: important players in the evolutionof plant genomes. Curr. Opin. Genet. Dev.5: 814–821.

Wiese, G., and Grambow, H. 1986. Indole-3-methanol-β-D-glc andindole-3-carboxylic acid-β-D-glc are products of indole-3-aceticacid degradation in wheat leaf segments. Phytochemistry,25:2451–2455.

Xue, J., Jorgensen, M., Pihlgren, U., and Rask, L. 1995. Themyrosinase gene family inArabidopsis thaliana: gene organiza-tion, expression and evolution. Plant Mol. Biol.27: 911–922.

© 1998 NRC Canada

604 Genome Vol. 41, 1998

Gen

ome

Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

Dep

osito

ry S

ervi

ces

Prog

ram

on

11/1

7/14

For

pers

onal

use

onl

y.