bare retrotransposons produce multiple groups of rarely polyadenylated transcripts from two...

BARE retrotransposons produce multiple groups of rarelypolyadenylated transcripts from two differentiallyregulated promoters

Wei Chang1 and Alan H. Schulman1,2,*

1MTT/BI Plant Genomics Laboratory, Institute of Biotechnology, Viikki Biocenter, University of Helsinki, P.O. Box 56,

Helsinki, Finland, and2zPlant Genomics, Biotechnology and Food Research, MTT Agrifood Research Finland, Myllytie 10, FIN-31600 Jokioinen,

Finland

Received 11 March 2008; revised 29 April 2008; accepted 19 May 2008; published online 15 July 2008.*For correspondence (fax +358 9 191 58952; e-mail [email protected]).

Summary

The BARE retrotransposon family comprises more than 104 copies in the barley (Hordeum vulgare) genome.

The element is bounded by long terminal repeats (LTRs, 1829 bp) containing promoters and RNA-processing

motifs required for retrotransposon replication. Members of the BARE1 subfamily are transcribed, translated,

and form virus-like particles. Very similar retrotransposons are expressed as RNA and protein in other cereals

and grasses. The BARE2 subfamily is, however, non-autonomous because it cannot produce the GAG capsid

protein. The pattern of plant development implies that inheritance of integrated copies should critically

depend, in the first instance, on cell-specific and tissue-specific expression patterns. We examined

transcription of BARE within different barley tissues and analyzed the promoter function of the BARE LTR.

The two promoters of the LTR vary independently in activity by tissue. In embryos TATA1 was almost inactive,

whereas transcription in callus appears to be less tightly regulated than in other tissues. Deletion analyses of

the LTR uncovered strong positive and negative regulatory elements. The promoters produce multiple groups

of transcripts that are distinct by their start and stop points, by their sequences, and by whether they are

polyadenylated. Some of these groups do not share the common end structures needed for template switching

during replication. Only about 15% of BARE transcripts are polyadenylated. The data suggest that distinct

subfamilies of transcripts may play independent roles in providing the proteins and replication templates

for the BARE retrotransposon life cycle.

Keywords: retrotransposon life cycle, promoter, transcriptional regulation, polyadenylation, mRNA, barley.

Introduction

The long terminal repeat (LTR) retrotransposons are ubi-

quitous in eukaryotes and comprise large proportions of

many plant genomes (Feschotte et al., 2002; Kumar and

Bennetzen, 1999; Schulman and Kalendar, 2005). The life

cycle of active LTR retrotransposons comprises the same

basic steps, facilitated by the same major proteins, as that of

retroviruses except that the assembled particle is not

enveloped for passage out of the cell (Vicient et al., 2001b;

Wicker et al., 2007; Wright and Voytas, 2002). These steps

comprise transcription, translation of the proteins encoded

by the retrotransposon, packaging into a virus-like particle,

reverse transcription, nuclear localization, and finally inte-

gration of the cDNA copy back into the genome.

Transcription of a retrotransposon therefore represents

the first regulatory opportunity because it controls produc-

tion of the RNA that serves as the template for both reverse

transcription and translation. Long terminal repeat retro-

transposon transcriptional promoters and their regulatory

regions are contained within the LTRs. Although the process

of reverse transcription renders the two LTRs identical, the 5¢LTR drives transcription whereas the 3¢ LTR serves as the

transcription terminator.

The final replication step, integration, must occur in cells

clonally ancestral to gametes to be heritable in plants that

reproduce only sexually. Hence, tissue-specific control

mechanisms can affect the evolutionary success and geno-

40 ª 2008 The AuthorsJournal compilation ª 2008 Blackwell Publishing Ltd

The Plant Journal (2008) 56, 40–50 doi: 10.1111/j.1365-313X.2008.03572.x

mic prevalence of retrotransposon families. Some, generally

low-copy, retrotransposons are normally transcriptionally

silent but are stress-inducible (Grandbastien et al., 2005; Liu

et al., 2004; Takeda et al., 1998), whereas abundant elements

such as BARE appear more generally active (Jaaskelainen

et al., 1999). Transcriptional analysis of the latter is needed

to clarify control of their propagation and production of

heritable copies.

BARE, a well-characterized and active Copia LTR retro-

transposon family, is bounded by 1.8-kb LTRs that surround

a central coding domain. The BARE elements are highly

abundant in Hordeum genomes, which they have helped to

shape (Kalendar et al., 2000; Shirasu et al., 2000; Suoniemi

et al., 1996a; Vicient et al., 1999). Furthermore, closely

related copies are found in other cereals, which are both

transcribed (Vicient and Schulman, 2002) and translated

(Vicient et al., 2001a). The BARE family is divided between

two main subfamilies, BARE1 and BARE2 (Tanskanen et al.,

2007). The recently identified BARE2 subfamily is highly

similar to BARE1 but lacks a functional coding region for

the capsid protein GAG, due to a specific, conserved

deletion of the start codon (Tanskanen et al., 2007; Vicient

and Schulman, 2005). However, it contains the signals and

structures needed for packaging into BARE1 virus-like

particles (VLPs) and replication by the BARE1 enzymes

(Sabot and Schulman, 2006). In barley, BARE2 is actually

more abundant than the BARE1 host. This suggests that

BARE2 is a non-autonomous subfamily of retrotransposons

that is parasitic on BARE1 (Sabot and Schulman, 2006;

Tanskanen et al., 2007). Such parasitism appears to require

co-expression of the BARE1 and BARE2 transcripts and the

BARE1 proteins, but this has not been investigated until

now.

More generally, retrotransposon transcripts are either

translated, reverse-transcribed, or both. It has remained an

open question for both retrotransposons and retroviruses as

to whether the self-same transcripts serve both functions.

Here, in order to shed light on the dynamics of the BARE life

cycle and the interplay of BARE1 and BARE2 subfamilies,

we have examined the transcripts of this retrotransposon

family. Promoter choice in different intact tissues, control of

transcription, transcription initiation, and termination and

polyadenylation were investigated.

Results

A comparison of BARE promoter choice in different

barley tissues

The BARE retrotransposon is unusual in containing two

promoters, TATA1 and TATA2 (TATA-box like motifs;

Figure 1). The question of how both promoters may be

under selection for maintenance prompted us to investi-

gate the relative activity of TATA1 and TATA2 in leaves,

(a)

(b)

(c)

(d)

Figure 1. Diagram of the BARE region under experimental consideration.

(a) Key features. The long terminal repeat (LTR) is shown as a thick box, the untranslated leader (UTL) region between the 5¢ LTR and the N-terminal end of GAG

(arrow) as a thin box. The positions of TATA1 (T1) and TATA2 (T2) are marked with bent arrows. The UTL is stippled, and is represented as for transcripts originating

from TATA1. Restriction sites relevant to constructs described in the text are shown with their position according to accession Z17327.

(b) Probes used in RNase protection assays (upper black boxes), primers used for the first-strand cDNA synthesis (middle gray boxes), and primer used in PCR

amplification (white box, bottom). Arrow bars indicate the direction of priming.

(c) Positions of nested PCR primers for mapping the 3¢ ends of transcripts.

(d) Identified termination sites and predictions for the R domain of BARE transcripts. Stop sites are indicated by arrows, those above the LTR for polyadenylated

RNAs and those below for un-polyadenylated RNA. The untranslated region (UTR) is indicated by stipples, the region of the 3¢ LTR beyond the end of transcription by

hatches. Start sites are indicated by dashed lines. The resulting predicted R domain is shown by brackets above and below the LTR, respectively, for polyadenlyated

and un-polyadenylated transcripts.

Retrotransposon BARE transcription 41

ª 2008 The AuthorsJournal compilation ª 2008 Blackwell Publishing Ltd, The Plant Journal, (2008), 56, 40–50

roots, shoots, embryos, and callus by the RNase protec-

tion assay (Melton et al., 1984) to evaluate expression

(Figure 2).

Probes P1 and P2 were designed to match the consensus

sequence of the available BARE LTRs. The products of the

probes protected by cellular RNA and by control sense-

strand RNA, together with the other reaction controls

(Figure 2), show that the probes are specific. Each probe

generates protected fragments corresponding to the sizes

expected from hybridization to the main transcript classes

(described below and in Figure 5). However, because TATA2

is downstream of TATA1, the P2 signal is derived from

transcription of both TATA1 and TATA2, whereas P1 reports

TATA1 activity only. Relative expression levels from TATA1

and TATA2 were calculated so as to take this into account.

We compared the strength of TATA1 and TATA2 by using

the 18S RNA probe P18 as the universal internal control. The

curves of Figure 2(b) are normalized to the highest peak and

compressed along the Y-axis; hence, the P1 and P2 peaks

appear similar in height. Embryos showed the highest total

signal as measured by the P2 probe (4.57 · the signal from

18S), followed by roots (3.75), callus (3.25), shoots (3.0), and

leaves (1.9). The TATA1 promoter, as measured by the P1

probe, showed fairly similar activities between the different

tissues of about 1 · 18S. Therefore, differences in total

transcript levels between the tissues can be attributed to

differences in TATA2 activity. When the P1 and P2 signals

were normalized to a constant 18S signal, TATA2 accounted

for an activity of 41.1 normalized units in embryos, 33.4 in

roots, 24.5 in callus and shoots, and 12.5 in leaves.

The probes could also bind to read-through transcripts

not driven by the LTR. To check this, we searched the dbEST

database, which contains about 479 000 barley expressed

sequence tags (ESTs), for matches to BARE by BLASTN.

Only one possible read-through transcript from barley and

two from wheat were seen (data not shown). Hence, the

RNase protection results are unlikely to have been affected

by read-through transcripts.

Control of BARE transcription in barley tissues

Previous assays of the promoter activity of the BARE1 LTR

were carried out only in leaf protoplasts (Suoniemi et al.,

1996b), which may not accurately reflect expression in

organized tissues. Transient expression assays (Figure 3)

were carried out by particle bombardment using full-length

and deleted LTR constructs (Table 1 and Figure 4) in order to

reveal regulatory elements. We investigated leaves,

embryos, and callus cultures, but concentrated on leaves

(Figure 4a) as they are better bombardment targets.

The removal of 654 bp from the 5¢ end of the LTR to form

construct C reduces promoter activity by 30% in leaves

(Figure 4a), indicating the presence of positive elements

upstream of this position. However, construct GH, contain-

ing only the TATA2 box, yields as much activity as the full-

length LTR, suggesting that negative regulatory elements

are present between nucleotide position (nt) 963 and nt

P1

Leaf

Root

Callus

Shoot

Leaf

Root

Shoot Callus

P2

A

B

C

D

Sample (a)

(b)

Pro

be P

1 P

rob

e P2

P1

Leaf

Embryo

Yeast

Sample

Pro

be P

1

P2

Leaf

Embryo

Yeast

Pro

be P

1

Figure 2. BARE transcription analysis by ribonuclease protection assays.

(a) Comparison of TATA1 and TATA2 use in total RNA from four different

tissues. The main peaks labeled in the chromatographs correspond to: P2,

probe P2 fragment protected by BARE TATA1 and TATA2 transcripts; P1, probe

P1 fragments protected by BARE TATA1 transcripts; P18, probe P18S fragment

protected by 18S RNA; P18S, probe P18S itself. Control assays among the

samples: P1, 54 pg of P1 was hybridized with corresponding synthetic, sense-

strand transcripts as a positive control for transcripts using TATA1; P2, 99 pg

of P2 was hybridized with corresponding synthetic, sense-strand transcripts as

a positive control for TATA1 + TATA2 transcripts; A, P1 and P18S without

RNase digestion; B, P2 and P18S without RNase digestion; C, 54 pg of P1 was

hybridized with 15 lg of yeast total RNA as a negative control for transcripts

using TATA1; D, 99 pg of P2 was hybridized with 15 lg of yeast total RNA as a

negative control for total transcripts. For the leaf, root, shoot, and callus, 15 lg

of total RNA was probed with 54 pg of P1 or P2 and 80 pg of P18S. The X-axis at

the bottom shows elution time during capillary electrophoresis. Small peaks

eluted early (left side of graphs) are digested, unprotected RNA.

(b) Comparison of transcription levels between leaves and embryos. Probes

and RNA amounts are as for (a). For tissue samples, the small peak for probe P1

between P1 and P2 may result from transcripts initiating at nt 1363 and 1383,

and that just to the left of P2 (with probe P2) may result from transcripts

initiating at nt 1685, 1686, and 1689 (see Figure 5).

42 Wei Chang and Alan H. Schulman


1444. Construct BP contains only the final 568 bp of the LTR

and displays 52% of the activity seen with the full-length

LTR, pointing to a strong positive regulatory element for

TATA2 in the 166-bp piece between nt 1444 and nt 1611 that

differentiates constructs BP and GH.

We investigated the role of regions downstream of TATA2

on reporter enzyme expression in leaves. When 426 bp at

the 3¢ end of construct C is deleted to form construct CD2

(Figure 4a), LUC activity returns to that of a full-length LTR.

Because enzyme activity rather than mRNA per se is

measured in the assay, this increase in activity may be due

to deletion of the many previously reported (Suoniemi et al.,

1996b) ATG start codons upstream of the reporter gene,

rather than to an increase in promoter strength. Construct T2

contains most (except for the 5¢ 38 nt) of the positive

regulatory regions of construct GH and of the full-length LTR

between nt 309 and 962, but it also retains one-quarter of the

negative regulatory region defined by the comparison

between constructs C and GH. It displays only 40% of full

LTR activity, thereby narrowing the extent of the negative

and positive regulatory regions to, respectively, 127 bp

between nt 1444 and 1611 and 38 bp between nt 1444 and

1481.

The inactivity of construct T0, lacking TATA1, TATA2, and

all LTR sequences downstream of TATA2, establishes that

the first 910 bp of the LTR do not contain a cryptic promoter.

Likewise, the inactivity of construct H demonstrates that the

last 338 bp of the LTR lack an independent promoter. The

results reported here with constructs GH and T2 correspond

(a)

(d)

(b) (c)

(e)

Figure 3. Histochemical analysis of the BARE

long terminal repeat (LTR) promoter by GUS

activity staining in different tissues following

bombardment.

(a) Seven-day-old leaf.

(b) Twenty-four-hour germinated embryo.

(c) Forty-eight-hour-old shoot.

(d) Seven-day callus culture.

(e) Seven-day root.

Table 1 Constructs used in particle bombardment for investigationof BARE transcription regulation

Construct LTR fragmenta Nucleotide rangeb Comments

LTR XhoI 309–2179C PvuI–XhoI 963–2179GH DdeI–XhoI 1444–2179BP StyI–XhoI 1611–2179CD2 PvuI–SacI 963–1753H SacI–XhoI 1757–2179 No TATA boxT0 BstEII 309–1217 No TATA boxD4 StuI 309–1420 Only TATA1T1 StyI + SacI–XhoI 309–1614 + 1757–2179 Only TATA1SA StuI–SacI 1421–1753 Only TATA2SN StuI–SnaBI 1421–2043 Only TATA2T2 BstEII + AviII–XhoI 309–1217 + 1481–2179

aThe full-length LTR ranges over nt 309–2137, with the untranslatedleader extending to the putative translational start at nt 3392.An intact 5¢ end begins at nt 309.bNumbering as in accession number Z17327.

LUC activityTATA(a)

(b)

% LTR activity309 2179 0 50 100

BP

luc NOS

luc NOS

luc NOS

luc NOS

luc NOS

luc NOS

luc NOS

luc NOS

LTR

C 2

GH

C

BP

T2

H

T0

21

TATA

% LTR activitybp

luc NOS

luc NOS

luc NOS

LTR

SA

21

309 2179 0 50 100

BPBP luc NOS

4

T1 luc NOS

bp

Construct

Figure 4. Contribution of BARE long terminal repeat (LTR) regions to

promoter activity in bombarded tissues.

(a) Comparison of various constructs in leaves. A diagram of the constructs

(full details in Table 1) is shown on the left, luciferase activity relative to the

full-length LTR on right. Error bars are shown.

(b) Comparison of responses in embryo (black) and callus (brown), relative to

the full-length LTR in each tissue.



well to those from the earlier protoplast assays (Suoniemi

et al., 1996b). However, construct BP has more than 60% of

the LTR activity in leaves and less than 50% in protoplasts,

and construct C gives 90% of the LTR activity in leaves but

125% in protoplasts. These differences may relate to varia-

tion in the presence of regulatory factors in protoplasts

compared to intact tissues, to the stressed state of protop-

lasts, or to both factors.

Callus tissue is a useful experimental system for studying

the BARE life cycle due to the abundance of BARE translation

products in callus (as there is also in embryos) compared

with their dearth in leaves (Jaaskelainen et al., 1999; Vicient

et al., 2001a). For transcriptional comparisons between

embryos and callus, we chose constructs to test the relative

strengths of TATA1 and TATA2: the full-length LTR, BP and

SA, having only TATA2, and T1 and D4 (Figure 4b). Construct

T1 contains TATA1 and the rest of the LTR, but lacks TATA2

and its downstream region. Construct D4 contains TATA1

and the initial part of LTR but lacks TATA2 and both its

upstream and downstream regions. The results were nor-

malized to the function of the full-length LTR, and showed

that both TATA1 and TATA2 were active in callus. TATA2

had more absolute activity in embryos than in callus.

Construct SA showed that the upstream regulatory region

of TATA2 is sufficient to give strong expression in the

embryo, with expression in callus approaching the full-

length LTR. Compared with the full-length LTR, construct BP

gave about 40% activity in embryos and about 75% in callus,

differing in turn from leaves and protoplasts as described

above.

Determination of the start site for BARE transcripts

Given the tissue-specific differences in promoter choice, we

have examined the 5¢ ends of BARE transcripts from various

tissues in more detail by 5¢ rapid amplification of cDNA

ends (RACE). For TATA1, we were also thereby able to

separate the results for BARE1 and BARE2 (Figure 5).

Shoots yielded clones for transcripts from both BARE sub-

families and TATA boxes. Callus cells contained both

TATA1-driven BARE2 transcripts and TATA2-driven BARE

transcripts. Multiple, closely spaced 5¢ ends were obtained,

with four different 5¢ ends for TATA1 products in 13 se-

quenced RACE clones and five different 5¢ ends for TATA2

products among 24 RACE clones (Figures 5 and 6). The

nucleotide polymorphisms visible within the sequences

indicate that the transcripts are derived from different

groups of BARE elements, rather than representing

incomplete extensions. Only one sequence was found that

was prematurely truncated at the 5¢ end.

Sequence alignments (Figure 6) demonstrate that TATA1

transcripts starting at position 1348 (numbered according to

accession Z17327) belong to the BARE1 subfamily (Fig-

ure 6a, group A) and are highly similar to the BARE1 type

element (Z17327). This position is 42-bp downstream of

TATA1. Transcripts starting at positions 1351, 1363, and

1383 (Figures 5 and 6b) belong to the BARE2 subfamily. The

TATA1 promoter region and transcriptional starts of BARE2

LTRs contain many small indels and point mutations com-

pared with BARE1, hence the nucleotide numbering accord-

ing to BARE1 is imprecise. The TATA2 transcripts initiate

at positions 1675, 1678, and 1685 (Figures 5 and 6). The

majority of TATA2 shoot transcripts begin 19-bp down-

stream of the TATA box, less than the predicted 33 bp. The

TATA2 box, visible in the transcripts originating from

TATA1, is completely conserved at position 322 in the

alignment (Figure 6b). Due to the similarity of the sequence

downstream of TATA2 between BARE1 and BARE2

(Figure 6b), it is difficult to assign TATA2 transcripts to a

particular subfamily.

Callus transcripts differ from shoot transcripts not only by

start site but also by sequence (Figures 5 and 6b, groups I1,

I2, J). Given the active BARE translation in callus, it is striking

that of 37 sequenced transcripts none of the TATA1 products

from callus were derived from BARE1, which is translation-

ally competent. This indicates that GAG translation products

seen in callus are likely to be from BARE1 TATA2 transcripts.

12931656

T1 T2

1814

Tissue

Copies

Variants

Element

StartGro

up

1348

1351

BARE-1

BARE-2

1

1

2

5

shoot

callus

A.

(a)

(b)

B.

1363

1383

BARE-2

BARE-2

1

1

3

1

callus

shoot

C.

D

.

1675

1678

1685

1686

1689

?

?

?

?

?

2

3

1

2

1

5

9

1

7

2

shoot

shoot

shoot

callus

callus

F.

G.

H.

I.

J.

37 14

Stop

1343BARE-212callusE.

Figure 5. Transcriptional start sites for BARE.

(a) Diagram of the promoter region of the long terminal repeat (LTR). White

box, untranscribed region; gray box, untranslated leader according to the

longest transcript.

(b) Diagram, abundance, and types of 5¢ rapid amplification of cDNA ends

(RACE) products cloned from shoot transcripts. Diagonally hatched boxes

indicate transcripts of TATA1 and stippled boxes transcripts of TATA2.

‘Group’ refers to sequences sharing the same start site. ‘Copies’ is the number

cloned and sequenced of each group out of a total of 35. ‘Variants’ are distinct

sequences within each group. ‘Element’ is the retrotransposon subfamily

from which the transcripts are derived, with ? meaning that subfamily cannot

be resolved. ‘Start’ refers to the transcriptional start position for each group.



The most common TATA2 transcripts from callus begin

10 bp further downstream than those found in shoots (nt 352

in the alignment; position 1686, Z17327). The callus

transcripts, furthermore, begin only 1–4 bp downstream of

the putative RNA cap site (Manninen and Schulman, 1993a).

These data, taken together, suggest that a different set of

BARE elements is transcribed in callus compared with those

in differentiated plant tissues, and that BARE expression in

callus may be less tightly regulated as well.

BARE transcript polyadenylation

Regulation of the polyadenylation of the retrotransposon

transcript is virtually unexplored in plants. Retroviruses may

produce both polyadenylated and non-polyadenylated

transcripts (Ratnasabapathy et al., 1990); many non-poly-

adenylated virus transcripts are efficiently translated (Gallie

and Walbot, 1990; Zeenko and Gallie, 2005). We investigated

the frequency of polyadenylation BARE transcript by cloning

BARE RNA independent of the presence of poly(A) tails,

using RNA ligase to ligate linkers to total RNA prior to PCR

(RLM-PCR). Following 3¢ linker ligation, RT-PCR was carried

out with primer RLM-2 matching BARE in combination either

with E1820, which amplifies only poly(A) RNA, or with

primer E2146, which matches the linker and amplifies

independent of the presence of a poly(A) tail (Figure 7).To

investigate the proportion of polyadenylated transcripts

among all BARE transcripts, more than 100 of these products

were cloned, then screened by PCR with the same primers.

(a)

(b)

Figure 6. Alignment of BARE rapid amplification of cDNA ends (RACE)

products with the BARE1 and BARE2 DNA sequences. Letters refer to

transcript groups, numbers to the variant type within each group (for details

see Figure 5).

(a) Alignment of group A transcripts to BARE1. Transcripts of group A are

aligned separately from the others due to the large number of indels in the

start region between BARE1 and BARE2.

(b) Alignment of group B–J transcripts (for details see Figure 5) to BARE1 and

BARE2 genomic sequences.

Poly(A) tail All tails

RT (–)

RT (+)

RT (–)

RT (+)

500600700800

400

300

200

100

Figure 7. Amplification of BARE 3¢ ends by RNA ligase-mediated (RLM) rapid

amplification of cDNA ends (RACE).

Left, amplification only of transcripts containing a poly(A) tail using primer

E1820. Right, amplification of all BARE 3¢ ends using primer E2146. RT())

control reactions lacked reverse transcriptase. A 100-bp ladder is indicated.

The poly(A) products, although not visible in the ‘all tails’ lane, are faintly

present and can be isolated as clones from the reaction.



A total of 15.5% contained poly(A) tails. The tails varied from

32 to 36 nt in length.

BARE transcript termination

A repeat, the R domain, is present at both ends of a retro-

transposon transcript. It is formed during transcription by

the LTR segment lying between the start of transcription and

its termination. It allows pairing between the RNA transcript

and the first cDNA strand, essential for replication. Given the

variations in transcript polyadenylation and initiation posi-

tion, we examined whether transcript termination sites also

vary, and how this would affect the R domain. The termi-

nation sites were established by sequencing both poly-

adenylated and non-polyadenylated clones derived from the

T4 RNA ligase-mediated (RLM) cloning. Eight cloned

products were checked for insertion length, and five were

sequenced. Of these, four terminate at positions 11539,

11559, 11561, and 11561 (by alignment with Z17327). This

places the termination points 21 and 23 bp immediately after

TATA1. One terminated 189 bp downstream from the TATA1

initiation point. The more upstream termination sites do not

provide any R domain to transcripts initiated from TATA2

(Figure 1d). However, the non-polyadenylated clones dis-

played termination sites 60, 105, 500, and 697nt (at positions

11598, 11642, 12037, and 12147 by alignment with Z17327)

beyond the TATA1 initiation site. The data indicate that

multiple transcript classes are produced, displaying distinct

R-domain structures.

Discussion

Success for retrotransposons may be defined as the ability

to propagate in the genome to high copy number without

seriously compromising the success of the host itself. In the

absence of strong negative selection, a positive feedback

loop leads an increasing proportion of the retrotransposon

population to represent ones that are expressed under

conditions where daughter copies can be effectively inte-

grated and inherited. In this regard, the BARE family can be

regarded as highly successful, particularly the BARE2 sub-

family parasitic on BARE1. BARE is abundant (Vicient et al.,

2001a), transcribed (Suoniemi et al., 1996b), translated

(Jaaskelainen et al., 1999), and associated with insertional

polymorphisms even at single sites (Kalendar et al., 2000).

The first step in retrotransposon replication is transcrip-

tion of the integrated copies. Transcription and transcript

processing offer avenues of cellular control over retrotrans-

poson activity. Here, we have examined the transcription

and transcripts of BARE retrotransposons in barley tissues.

The BARE LTR initiates transcription and is unusual in

possessing two promoters. We found that both promoters

are active in callus. TATA2 is much more active in embryos

than in leaves, with intermediate activity in roots and callus,

whereas TATA1 is virtually inactive in embryos. The results

from the RNase protection assays, bombardment assays,

and 5¢ RACE are all consistent regarding the activity of

TATA1 versus TATA2, although the assays report somewhat

different relative values. The results suggest that, because

new daughter copies in embryos but not in leaves are

inheritable, control of TATA2 plays a much more important

role than control of TATA1 in the evolution of the BARE

family. The analyses of LTR promoter constructs indicate

that positive regulatory regions are located between nt 309

and 963 and nt 1444 and 1611, while a negative regulatory

region can be found between nt 963 and nt 1444. The nt 309–

963 region, when searched against a plant cis-acting

element database (http://www.dna.affrc.go.jp/PLACE/),

reveals many potential regulatory motifs, the roles of which

remain to be explored.

We found four start sites for TATA1 and five for TATA2.

For TATA1, cDNA sequence and start site correlations show

that different BARE groups favor particular start sites. In

shoots, both BARE1 and BARE2 use TATA1, but only BARE2

does so in callus. Of the BARE2 transcripts, the one cloned

from shoots has a different start site, further downstream,

from any of the 11 isolated from callus, which belong to

several distinct sequence groups. Different start sites were

also seen for TATA2-driven transcripts in both shoots and

callus, though for these it is not possible to distinguish those

originating from BARE1 from those of BARE2.

Our earlier results (Suoniemi et al., 1996b) indicated a

start for TATA1 at nt 1323 and for TATA2 at nt 1670, both for

callus. Our current results (Figure 5) differ from these by

25–60 nt for TATA1 and from 5 to 19 nt for TATA2, which is

unsurprising. The earlier positions were estimated from the

relative mobilities of RNA fragments in sequencing ladders,

whereas the current positions have been determined directly

by sequencing. Furthermore, the earlier estimates were

made without reference to BARE2, which was unknown at

the time.

Whole-genome transcriptional analyses indicate that

putative alternative transcriptional start sites are not uncom-

mon in Arabidopsis (Alexandrov et al., 2006). However,

BARE represents a multi-gene family on a scale not seen for

cellular genes in Arabidopsis or elsewhere. Furthermore, the

presence not only of multiple start sites but also of multiple

stop sites and non-polyadenylated transcripts shows that

retrotransposon transcription differs radically from general

cellular transcription in plants. Variation in the start and

termination positions correlates with sequence variation

within BARE, implying that BARE subgroups may differ in

their replication rather than each individual BARE being

transcribed from multiple sites.

We found, furthermore, that the termination site depends

on whether a BARE transcript is polyadenylated. Only about

16% of BARE transcripts were polyadenylated. Although

de-adenylation is a key step in mRNA degradation (Beelman



and Parker, 1995; Feldbrugge et al., 2001), two features of the

cloned cDNAs suggest that those lacking poly(A) tails are not

de-adenylated transcripts. First, the length of the poly(A)

tails was homogeneous even when the cDNAs were cloned

by T4 RNA ligase and without an oligo-dT primer, showing

no evidence of the poly(A) shortening typical of de-adeny-

lation (Feldbrugge et al., 2001). Furthermore, the non-

adenylated transcripts terminate at distinct points (60, 105,

500 nt) further downstream from TATA1 in the 3¢ LTR than

the termination sites of the polyadenylated transcripts

(most, 20 nt downstream from TATA1). Hence, the non-

adenylated transcripts are not shortened versions of those

with poly(A) tails.

Retrotransposons are transcribed in the nucleus, like

cellular genes, but are replicated and translated in the

cytoplasm. Distinct classes of RNA utilize different export

pathways (Cullen, 2003). Polyadenylation, lacking in most

BARE1 transcripts, is an important identity element in

defining and controlling export of mRNA (Fuke and Ohno,

2008). Our results therefore beg the question of nuclear

export of non-adenylated BARE transcripts. In preliminary

experiments (W.C. and A.S., unpublished results), isolated

BARE VLPs appear to contain exclusively or mostly non-

polyadenlyated BARE RNA, providing evidence not only for

the export but also for the packaging of these transcripts.

Cellular histone and some retroviral transcripts, which like

BARE are both unspliced and not polyadenylated, rely on a 3¢stem–loop structure and recognition by the mRNA exporter

receptor TAP (Erkmann et al., 2005; LeBlanc et al., 2007). It

remains to be seen if a similar mechanism acts on BARE

transcripts.

Retrotransposon transcripts have two distinct roles, as

mRNA for translation and as templates for cDNA synthesis,

and whether the same RNA serves both purposes is unknown

(Sabot and Schulman, 2006). As a prelude to replication,

retrotransposon transcripts are generally packaged two to a

VLP. Examination of the BARE packaging signals indicates

that BARE2 can be packaged together with BARE1 in trans-

complementation of its lack of its own GAG (Sabot and

Schulman, 2006; Tanskanen et al., 2007). However, it is not

the BARE1–BARE2 distinction but that of transcript structure

more generally that is critical. The data here indicate that

multiple transcript classes are produced from BARE, display-

ing distinct R-domain structures, with the polyadenylated

RNAs containing short ones (Figure 1d). End structures have

been shown to be critical in replication of synthetic plant

retrotransposon constructs (Bohmdorfer et al., 2005).

Here, we have demonstrated that major classes of native

BARE transcripts would not be able to engage in strand

transfer during cDNA synthesis if co-packaged, due to

the lack of a common R domain stemming from variation

in the initiation and termination sites. Only a minority of the

transcripts with long predicted R domains were found, for

example, among shoot RNA. This provides an indication

that the trans-preference concept for retroviral RNAs, where

two pools of mRNAs co-exist – one to direct synthesis of the

proteins required for the life cycle and the other to serve as

the template for the replication of the virus genome

(Nikolaitchik et al., 2006) – applies to the LTR retrotranspo-

sons as well.

The structure of the RNA transcript probably also plays a

critical role, not only in LTR retrotransposon replication but

also in translation. Translation of cellular mRNA involves the

formation of a loop, where the 5¢ cap and 3¢ poly(A) tail are

brought together through the interaction of a variety of

initiation factors and poly(A)-binding proteins (Kawaguchi

and Bailey-Serres, 2002). The long R domains, present in the

BARE transcripts that are not polyadenylated, may provide

for a strategy similar to that of many plant RNA viruses,

where stem–loop-mediated translation is used (Dreher and

Miller, 2006). BARE2 transcripts must be packaged into

BARE1 GAG, yet which population of BARE transcripts is

translated remains to be established. At least in callus, only

TATA2 produces translationally competent BARE tran-

scripts. The translational competence of non-polyadeny-

lated BARE transcripts needs to be established.

In conclusion, several features of our data suggest that a

dynamic system regulates the transcription and replication

of distinct BARE subfamilies. First, the existence of two

TATA boxes allows independent selective forces to act on

their corresponding regulatory regions. Although TATA1 is

virtually inactive in embryos, which can give rise to clones of

cells ultimately leading to gametes, it is nonetheless main-

tained. At present we do not know, due to technical

limitations, whether TATA1 is active in floral tissues directly

giving rise to gametes. Second, subfamilies of BARE1 and

BARE2 give rise to mRNAs with distinct start sites, and these

may replicate as separate populations due to the presence

of different R domains. Lastly, the presence of both poly-

adenylated and non-polyadenylated transcripts of varying

structures suggests the existence of separately replicating

pools. The compartmentalization of BARE copies within the

genome by expression, RNA processing, and replication

creates avenues for segmental evolution of the family and

interacts with an opposing force, that of recombination

between distinct groups of these retrotransposons (Sabot

and Schulman, 2007; Tanskanen et al., 2007; Vicient and

Schulman, 2005).

Experimental procedures

Plant materials

Barley (Hordeum vulgare L.) cv. Himalaya was germinated asepti-cally at room temperature for 24 h before harvesting embryos. Forroots and shoots, Himalaya seeds were sown in 15-cm pots filledwith vermiculite and grown in a controlled environment roomilluminated with 18 h light at 5 · 104 lux 103 lEm)2 sec)1 photo-synthetically active radiation (PAR)]. Fresh shoots and roots were



excised after 3.5 days and leaves were collected after 7 days. Calluslines derived from barley cv. Kymppi were harvested 10–14 daysafter subculture. Tissues of the same age as prepared forbombardments were frozen in liquid nitrogen and stored at )70�Cfor total RNA isolation.

In vitro synthesis and gel purification of probes

Probes for ribonuclease protection assays were synthesized byin vitro transcription using T7 RNA polymerase on templates ofsynthetic DNA containing the T7 promoter. Several partially single-stranded templates were prepared by annealing a synthetic oligo-nucleotide, complementary only to the promoter region of thelonger bottom strand of the template. The resulting oligonucleotideprobes are shown relative to the LTR in Figure 1. The templates forprobe P1 and its complementary strand were generated by in vitrotranscription from annealed oligos A and B, and oligos A and C,respectively. The template for probe P2 and its complementarystrand were generated from oligos A and D, and oligos A and E,respectively. The oligos consisted of: oligo A: 5¢-GGATCCTAAT-ACGACTCACTATAGG-3¢; oligo B, 5¢-CCGGGAGTGACGAATGGG-TTCCGGATGTTCACCGGGGAGGGGGGGCAGCCCTATAGTGAGTC-GTATTAGGATCC-3¢; oligo C, 5¢-GGAAAAGT(CT)GTCTCAAATCAGCCCTAAAACCTCC(GA)TATATACCTATAGTGAGTCGTATTAGGATC-3¢;oligo D, 5¢-TTTAGGGCTGATTTGAGAC(GA)ACTTTTCCTATAGTGA-GTCGTATTAGGATCC-3¢; oligo E, 5¢-GGAAAAGT(CT)GTCTCAAAT-CAGCCCTAAACCTATAGTGAGTCGTATTAGGATCC-3¢. The sametemplates were also used for making probes to detect antisenseRNA.

In vitro transcription reactions were made with the T7 MEGA-shortscript� kit (Ambion, http://www.ambion.com/) according to themanufacturer’s protocol. Fluorescent probes were labeled withfluorescein-12-UTP at a 1:1 ratio with labeled UTP. The resultingprobes (Figure 1) were: P1, 5¢-CCCUCCCCGGUGAACAUCCG-GAACCCAUUCGUCACUCCCGG-3¢, a 41-mer and P2, 5¢-GGAAAAG-U(CU)GUCUCAAAUCAGCCCUAAA-3¢, a 28-mer. The internalcontrol RNA probe P18S was generated by in vitro transcriptionwith T3 polymerase from the pTRI RNA 18S (Ambion) antisensecontrol template, which contains an 80-bp fragment from a highlyconserved region of the 18S ribosomal gene. The probe wasdenatured, purified on a 15% 2-amino-2-(hydroxymethyl)-1,3-pro-panediol (TRIS)-borate-EDTA (TBE)–urea ready-made gel, cut out,ethanol precipitated, and quantified by spectrophotometer.

Ribonuclease protection assays

Ribonuclease protection assays (RPAs) were made with the RPA IIIkit (Ambion) according to the manufacturer’s instructions, exceptthat total RNA was precipitated before adding the probe. Becausethe probe–target pairs are A–U rich, we used RNase T1 rather thanRNase A to remove unprotected single-stranded RNA. Fifteenmicrograms total RNA from each tissue was employed in the assay.The precipitated, protected fragment generated by the assay wasdissolved in 10 ll loading buffer containing formamide (95%,deionized), EDTA (0.3%), dextran blue (0.01 M) and NaOH (20 mM),then heat denatured (92�C for 3 min) before loading on a 5% acryl-amide sequencing gel (Pharmacia LKB ALF DNA sequencer, http://www4.gelifesciences.com).

For comparison of the expression levels in the different tissues,an 18S RNA probe was used to normalize for RNA loading. Thefluorescent probe P18S produced by in vitro transcription is 99 bp inlength (Figure 2a, second peak of graphs A and B). The 18 Sribosomal RNA, when hybridized to probe P18S, protected an 80-bpfragment that could be detected as a peak in the presence of both P1

and P2 (Figure 2, Peak 18). The data from each assay were firstnormalized by the protected 18S peak and then the peak heightscompared for the P1 and P2 probes. The background peaks at theleft end of each chromatograph represent cleaved probe followingthe protection. Negative controls were carried out by hybridizing P1and P2 to yeast total RNA; no protected peaks at the P1 and P2positions were found (Figure 2a, graphs C and D; Figure 2b, yeastsamples). In each experiment, positive controls were carried out byhybridizing P1 and P2 to a four-fold excess of their correspondingsense-strand transcripts, producing single peaks (Figure 2, chro-matographs P1 and P2) due to perfectly matching probes andtargets. Given the length of the BARE transcript (�7.1 kb), amaximum of 0.5% of BARE transcripts with the message pool (ourunpublished results), and the size of the P1 and P2 probes, theprobes were present at a 12-fold (P1) to 28-fold (P2) molar excessover the target transcripts.

Transient transformation

Transient transformation was carried out by bombardment (PDS1000/He particle device, Bio-Rad, http://www.bio-rad.com/), using avacuum of 28 in Hg, a He pressure of 900 lb in)2 (p.s.i.), and a targetdistance of 6 cm. Three pieces of leaf, each 1.5 cm in length, or 25embryos were placed in the center of a bombardment plate con-taining 0.8% solidified agar. Concentrated callus culture was placedon the center of a plate with a 1.5-cm diameter flat surface.

The different LTR–LUC constructs were mixed with a GUS controlplasmid (pBI221) at a ratio of 1:1, and precipitated onto goldparticles (�1 lm diameter) prior to bombardment. The precipitationmixture contained 1.25 mg gold particles, 6.25 lg plasmid DNA,25 ll CaCl2 (2.5 M), and 10 ll spermidine-free base (Sigma S0266,http://www.sigmaaldrich.com/) at 0.1 M. The mixture was centri-fuged briefly and the supernatant removed. After washing with100% ethanol, the particle–DNA pellet was resuspended in 30 llethanol. For each bombardment, 7.5 ll of particle–DNA suspensionwas spread onto the surface of the macrocarrier. After bombard-ment, tissues were incubated on the sealed agar plate for 24 h atroom temperature in the dark. Preliminary experiments showed thatbombardment at room temperature rather than at 4�C, using afull-length LTR promoter, gave four-fold higher expression.Room temperature was used in all subsequent assays. Signalstrength was reduced if the particles were coated with CaCl2 but nospermidine-free base. Microprojectile aggregates were dispersedby bombardment.

LUC and GUS assays

Expressed LUC and GUS proteins were extracted for enzymaticmeasurement by submerging the bombarded tissues in extractionbuffer containing 50 mM Na phosphate pH 7.4, 1% polyvinylpyrr-olidone (PVP) 360 000, 2 mM EDTA, and 20 mM DTT and thenhomogenizing. The extracts were pelleted in a microcentrifuge for10 min at 4�C. Cleared lysates of 10 ll were assayed both withthe Promega LUC kit (Promega E1501, http://www.promega.com/)and Tropix GUS-Light kit (Applied Biosystems, http://www.appliedbiosystems.com/) according to the manufacturers’ direc-tions. In each experiment, two sets of bombardments were carriedout, one containing an LTR–luc deletion construct mixed with thefull-length LTR-uidA control and the other containing the full-lengthLTR–luc fusion mixed with the full-length LTR–uidA control. Eachpair was measured in four replicates. The final results were calcu-lated as the ratio EXO–LUC/LTR–GUS:LTR–LUC/LTR–GUS, whereEXO–LUC is the LTR deletion fused to luc, LTR–LUC is the full-lengthLTR fused to luc, and LTR–GUS is the full-length LTR fused to uidA.



This approach minimizes variation in promoter activity due todifferences between experiments in the efficiencies of samplebombardment and extraction.

Histochemical staining

Visualization of GUS activity by staining was performed bysubmerging tissues, which had been bombarded with LTR–GUSconstructs, for 48 h at room temperature in a stain buffer contain-ing 2 mM 5-bromo-4-chloro-3-indolyl-beta-D-galactopyranoside(X-Gal), 0.1% Triton X-100, 0.1 M NaHPO4. The blue stain was thenvisualized by light microscopy.

5¢ RACE-PCR

5¢ RACE-PCR was used to determine 5¢ RNA ends. Total RNA wasfirst treated with RNase-free DNase and then purified on a column(Qiagen 28204, http://www.qiagen.com/). The DNase-treated RNAwas tested for DNA contamination by PCR, using two primer pairsfor integrase (Suoniemi et al., 1998) that amplify well from thehighly repeated BARE integrase domain present in barley genomicDNA. Untreated total RNA and genomic DNA served as controls forthe DNA contamination test. The reverse transcription reaction forproduction of cDNA was carried out using the SMART RACE cDNAAmplification Kit (Clontech K1811-1, http://www.clontech.com/) on1 lg callus or short RNA according to the manufacturer’s instruc-tions. The positions of the specific primers, P-PBS (5¢-CTACGCAT-GAACCTAGCTCATGATGCC-3¢) and ATGA (5¢-CTGGTTGGCCCACG(CT)GAGCCATT(GA)ATCTACAACA(CT)A-3¢), are shown in Figure 1.They respectively matched the BARE PBS, which is found adjacentto the 5¢ LTR, and the beginning of GAG coding region, which isinternal to the PBS. For the second step, an additional gene-specificprimer, PW1 (5¢-AAGCCTGACGACGCTCCGGAGGTTGGT-3¢; Fig-ure 1), was used for the PCR amplification. The RACE experimentswere carried out with a touchdown PCR (Don et al., 1991) protocol toincrease the specificity of the products. The reaction consisted of:five cycles of 94�C for 5 sec, 72�C for 2 min; five cycles of 94�C for 5sec, 70�C for 10 sec, 72�C for 2 min; 35 cycles of 94�C for 5 sec, 68�Cfor 10 sec, 72�C for 2 min; a final extension at 72�C for 5 min; storageat 4�C. The PCR product was purified (Minielute PCR purification kit,Qiagen), then cloned into the pGEM-T vector (Promega). Coloniescontaining inserts were screened by PCR, using the gene-specificprimer PW1 together with the Universal primer mix from thekit. Control reactions lacking reverse transcriptase produced noproduct.

Analysis of 3¢ transcript ends

To determine the sequence and presence of polyadenylation at the3¢ ends of BARE transcripts, RNA ligase-mediated (RLM) RT-PCRwas carried out. Total RNA was isolated (RNAasy kit, Qiagen)then ligated to the phosphorylated linker E2147 (5¢-ACT-CTGCGTTGTTACCACTGCTT-3¢) in a 70-ll ligation reaction con-taining 12 lg RNA, 7 ll 10 · T4 RNA ligase buffer, 5 ll T4 RNAligase, and 5 ll of linker E2147 (0.3 lg ll)1). The RNA was firstdenatured at 95�C for 2 min then cooled on ice before adding to thereaction. The ligation reaction was incubated at 37�C for 3.5 h andthen purified on a MiniElute RNA clean-up column (Qiagen). ThecDNA was synthesized by using a first-strand synthesis kit (Clon-tech) with oligo E2146, which is complementary to the E2147 linker,as the primer. The cDNA was amplified using the conditionsdescribed above for 5¢ RACE PCR.

For polyadenylated transcripts, RT-PCR was carried out asfollows. Total RNA isolated as above was treated three times witha DNA-free kit (Ambion AM-1906). To test for DNA contamination,PCR was first carried out directly on 5 lg RNA using a pair of well-conserved primers to the LTR. To make cDNA, 1 lg RNA was primedwith oligo E1820 (AAGCAGTGGTAACAACGCAGAGTACT30NA).Nested PCR was then carried out in two rounds, the first roundwith primers RLM-1 (CGCAGGGTCTGCACACTTAAGGTTC) and E2146 (AAGCAGTGGTAACAACGCAGAGT, which matches part ofE1820) and the second round with primers RLM-2 (GTTGGTTACC-GAATGTTGTTCGGAGTC) and E2146, respectively. The RT-minuscontrols produced no amplification products. The RT-PCR productswere cloned and sequenced.

Acknowledgements

We thank Paivi Laamanen for her excellent technical assistance withsequencing gels for the NPA work, Annu Suoniemi and Anne-MariNarvanto for their contributions to the LTR constructs, and LarsPaulin for helpful technical suggestions. We also thank CarlosVicient and Francois Sabot for discussion and comments. This workwas supported by a grant to WC from the Academy of Finland(project 106949).

References

Alexandrov, N.N., Troukhan, M.E., Brover, V.V., Tatarinova, T.,

Flavell, R.B. and Feldmann, K.A. (2006) Features of Arabidopsisgenes and genome discovered using full-length cDNAs. PlantMol. Biol. 60, 69–85.

Beelman, C.A. and Parker, R. (1995) Degradation of mRNA ineukaryotes. Cell, 81, 179–183.

Bohmdorfer, G., Hofacker, I.L., Garber, K., Jelenic, S., Nizhynska, V.,

Hirochika, H., Stadler, P.F. and Bachmair, A. (2005) UnorthodoxmRNA start site to extend the highly structured leader of retro-transposon Tto1 mRNA increases transposition rate. RNA, 11,1181–1191.

Cullen, B.R. (2003) Nuclear RNA export. J. Cell Sci. 116, 587–597.Don, R.H., Cox, P.T., Wainwright, B.J., Baker, K. and Mattick, J.S.

(1991) ‘Touchdown’ PCR to circumvent spurious priming duringgene amplification. Nucl. Acids Res. 19, 4008.

Dreher, T.W. and Miller, W.A. (2006) Translational control in positivestrand RNA plant viruses. Virology, 344, 185–197.

Erkmann, J.A., Sanchez, R., Treichel, N., Marzluff, W.F. and Kutay,

U. (2005) Nuclear export of metazoan replication-dependenthistone mRNAs is dependent on RNA length and is mediatedby TAP. RNA, 11, 45–58.

Feldbrugge, M., Arizti, P., Sullivan, M.L., Zamore, P.D., Belasco, J.G.

and Green, P.J. (2001) Comparative analysis of the plant mRNA-destabilizing element, DST, in mammalian and tobacco cells.Plant Mol. Biol. 49, 215–223.

Feschotte, C., Jiang, N. and Wessler, S. (2002) Plant transposableelements: where genetics meets genomics. Nat. Rev. Genet. 3,329–341.

Fuke, H. and Ohno, M. (2008) Role of poly (A) tail as an identityelement for mRNA nuclear export. Nucl. Acids Res. 36, 1037–1049.

Gallie, D.R. and Walbot, V. (1990) RNA pseudoknot domain oftobacco mosaic virus can functionally substitute for a poly(A) tailin plant and animal cells. Genes Dev. 4, 1149–1157.

Grandbastien, M.A., Audeon, C., Bonnivard, E., Casacuberta, J.M.,

Chalhoub, B., Costa, A.P., Le, Q.H., Melayah, D., Petit, M., Poncet,

C., Tam, S.M., Van Sluys, M.A. and Mhiri, C. (2005) Stress



activation and genomic impact of Tnt1 retrotransposons inSolanaceae. Cytogenet. Genome Res. 110, 229–241.

Jaaskelainen, M., Mykkanen, A.-H., Arna, T., Vicient, C., Suoniemi,

A., Kalendar, R., Savilahti, H. and Schulman, A.H. (1999) Retro-transposon BARE-1: expression of encoded proteins and forma-tion of virus-like particles in barley cells. Plant J. 20, 413–422.

Kalendar, R., Tanskanen, J., Immonen, S., Nevo, E. and Schulman,

A.H. (2000) Genome evolution of wild barley (Hordeum sponta-neum) by BARE-1 retrotransposon dynamics in response to sharpmicroclimatic divergence. Proc. Natl. Acad. Sci. USA 97, 6603–6607.

Kawaguchi, R. and Bailey-Serres, J. (2002) Regulation of transla-tional initiation in plants. Curr. Opin. Plant Biol. 5, 460–465.

Kumar, A. and Bennetzen, J. (1999) Plant retrotransposons. Annu.Rev. Genet. 33, 479–532.

LeBlanc, J.J., Uddowla, S., Abraham, B., Clatterbuck, S. and

Beemon, K.L. (2007) Tap and Dbp5, but not Gag, are involved inDR-mediated nuclear export of unspliced Rous sarcoma virusRNA. Virology, 363, 376–386.

Liu, Z.L., Han, F.P., Tan, M., Shan, X.H., Dong, Y.Z., Wang, X.Z.,

Fedak, G., Hao, S. and Liu, B. (2004) Activation of a rice endoge-nous retrotransposon Tos17 in tissue culture is accompanied bycytosine demethylation and causes heritable alteration in meth-ylation pattern of flanking genomic regions. Theor. Appl. Genet.109, 200–2009.

Manninen, I. and Schulman, A.H. (1993a) BARE-1, a copia-like ret-roelement in barley (Hordeum vulgare L.). Plant Mol. Biol. 22,829–846.

Melton, D.A., Krieg, P.A., Rebagliati, M.R., Maniatis, T., Zinn, K. and

Green, M.R. (1984) Efficient in vitro synthesis of biologically activeRNA and RNA hybridization probes from plasmids containing abacteriophage SP6 promoter. Nucl. Acids Res. 12, 7035–7056.

Nikolaitchik, O., Rhodes, T.D., Ott, D. and Hu, W.S. (2006) Effects ofmutations in the human immunodeficiency virus type 1 Gag geneon RNA packaging and recombination. J. Virol. 80, 4691–4697.

Ratnasabapathy, R., Sheldon, M., Johal, L. and Hernandez, N. (1990)The HIV-1 long terminal repeat contains an unusual element thatinduces the synthesis of short RNAs from various mRNA andsnRNA promoters. Genes Dev. 4, 2061–2074.

Sabot, F. and Schulman, A. (2007) Template switching can createcomplex LTR retrotransposon insertions in Triticeae genomes.BMC Genomics, 8, 247.

Sabot, F. and Schulman, A.H. (2006) Parasitism and the retrotrans-poson life cycle in plants: a hitchhiker’s guide to the genome.Heredity, 97, 381–388.

Schulman, A.H. and Kalendar, R. (2005) A movable feast: diverseretrotransposons and their contribution to barley genomedynamics. Cytogenet. Genome Res. 110, 598–605.

Shirasu, K., Schulman, A.H., Lahaye, T. and Schulze-Lefert, P. (2000)A contiguous 66 kb barley DNA sequence provides evidence forreversible genome expansion. Genome Res. 10, 908–915.

Suoniemi, A., Anamthawat-Jonsson, K., Arna, T. and Schulman,

A.H. (1996a) Retrotransposon BARE-1 is a major, dispersedcomponent of the barley (Hordeum vulgare L.) genome. PlantMol. Biol. 30, 1321–1329.

Suoniemi, A., Narvanto, A. and Schulman, A.H. (1996b) The BARE-1retrotransposon is transcribed in barley from an LTR promoteractive in transient assays. Plant Mol. Biol. 31, 295–306.

Suoniemi, A., Tanskanen, J., Pentikainen, O., Johnson, M.S. and

Schulman, A.H. (1998) The core domain of retrotransposonintegrase in Hordeum: predicted structure and evolution. Mol.Biol. Evol. 15, 1135–1144.

Takeda, S., Sugimoto, K., Otsuki, H. and Hirochika, H. (1998)Transcriptional activation of the tobacco retrotransposonTto1 by wounding and methyl jasmonate. Plant Mol. Biol. 36,365–376.

Tanskanen, J.A., Sabot, F., Vicient, C. and Schulman, A.H. (2007)Life without GAG: the BARE-2 retrotransposon as a parasite¢sparasite. Gene, 390, 166–174.

Vicient, C.M., Jaaskelainen, M., Kalendar, R. and Schulman, A.H.

(2001a) Active retrotransposons are a common feature of grassgenomes. Plant Physiol. 125, 1283–1292.

Vicient, C.M., Kalendar, R. and Schulman, A.H. (2001b) Envelope-containing retrovirus-like elements are widespread, transcribedand spliced, and insertionally polymorphic in plants. GenomeRes. 11, 2041–2049.

Vicient, C.M. and Schulman, A.H. (2002) Copia-like retrotranspo-sons in the rice genome: few and assorted. Genome Lett. 1, 35–47.

Vicient, C.M. and Schulman, A.H. (2005) Variability, recombination,and mosaic evolution of the barley BARE-1 retrotransposon. Mol.Biol. Evol. 61, 275–291.

Vicient, C.M., Suoniemi, A., Anamthawat-Jonsson, K., Tanskanen,

J., Beharav, A., Nevo, E. and Schulman, A.H. (1999) Retrotrans-poson BARE-1 and its role in genome evolution in the genusHordeum. Plant Cell, 11, 1769–1784.

Wicker, T., Sabot, F., Hua-Van, A., Bennetzen, J., Capy, P.,

Chalhoub, B., Flavell, A.J., Leroy, P., Morgante, M., Panaud, O.,

Paux, E., SanMiguel, P. and Schulman, A.H. (2007) A unifiedclassification system for eukaryotic transposable elements. Nat.Rev. Genet. 8, 973–982.

Wright, D.A. and Voytas, D.F. (2002) Athila4 of Arabidopsis andCalypso of soybean define a lineage of endogenous plant retro-viruses. Genome Res. 12, 122–131.

Zeenko, V. and Gallie, D.R. (2005) Cap-independent translation oftobacco etch virus is conferred by an RNA pseudoknot in the5¢-leader. J. Biol. Chem. 280, 26813–26824.



bare retrotransposons produce multiple groups of rarely polyadenylated transcripts from two...

Documents