Plant Biotechnology Journal

(2005)

3

, pp. 571–582 doi: 10.1111/j.1467-7652.2005.00147.x

© 2005 Blackwell Publishing Ltd

571

Blackwell Publishing, Ltd.Oxford, UKPBIPlant Biotechnology Journal1467-7644© 2005 Blackwell Publishing Ltd? 20052?571Original Article

Root-specific repressor-operator gene complexTehryung Kim et al.

Engineering a root-specific, repressor-operator gene complex

Tehryung Kim

1,

†, Rebecca S. Balish

2,

†, Andrew C. P. Heaton

1

, Elizabeth C. McKinney

1

, Om Parkash Dhankher

3

and Richard B. Meagher

1,

*

1

Department of Genetics, University of Georgia, Athens, GA 30602

2

Department of Microbiology, Miami University, Oxford, OH 45056

3

Department of Plant, Soil, and Insect Science, University of Massachusetts Amherst, Amherst, MA 10062

Summary

Strong, tissue-specific and genetically regulated expression systems are essential tools in

plant biotechnology. An expression system tool called a ‘repressor-operator gene complex’

(ROC) has diverse applications in plant biotechnology fields including phytoremediation,

disease resistance, plant nutrition, food safety, and hybrid seed production. To test this

concept, we assembled a root-specific ROC using a strategy that could be used to construct

almost any gene expression pattern. When a modified

E. coli lac

repressor with a nuclear

localization signal was expressed from a rubisco small subunit expression vector,

S1pt::lacIn

,

LacIn protein was localized to the nuclei of leaf and stem cells, but not to root cells. A LacIn

repressible

Arabidopsis

actin expression vector

A2pot

was assembled containing upstream

bacterial

lacO

operator sequences, and it was tested for organ and tissue specificity using

β

-glucuronidase (

GUS

) and mercuric ion reductase (

merA

) gene reporters. Strong GUS

enzyme expression was restricted to root tissues of

A2pot::GUS/S1pt::lacIn

ROC plants,

while GUS activity was high in all vegetative tissues of plants lacking the repressor.

Repression of shoot GUS expression exceeded 99.9% with no evidence of root repression,

among a large percentage of doubly transformed plants. Similarly, MerA was strongly

expressed in the roots, but not the shoots of

A2pot::merA/S1pt::lacIn

plants, while MerA

levels remained high in both shoots and roots of plants lacking repressor. Plants with MerA

expression restricted to roots were approximately as tolerant to ionic mercury as plants

constitutively expressing MerA in roots and shoots. The superiority of this ROC over the

previously described root-specific tobacco RB7 promoter is demonstrated.

Received 11 November 2004

revised 10 May 2005;

accepted 18 May 2005.

*

Correspondence

(fax 706-542-1387;

e-mail meagher@uga.edu)

†These two authors contributed equally in

a collaborative effort.

Keywords:

phytoremediation, plant

pathogens, nutrition, resistance, hybrid seed,

gene silencing.

Introduction

Strong, tissue-specific expression systems are essential plant-

biotechnology tools. We devised a strategy for building repressor-

operator gene-complexes (ROCs) that could be used to target

almost any expression pattern and tested it by constructing

a simple root-specific system with wide applications to plant

biotechnology. The general approach to creating a ROC is to

use the restricted expression of one promoter to limit the

expression pattern of a second, more widely expressed

promoter or to repress the expression of a promoter under

particular genetic circumstances. In this manuscript, we

utilized a shoot-specific promoter to limit the expression of a

vegetative constitutive promoter to root expression.

Our root-specific ROC combined the well-characterized

genetic regulatory elements from the

E. coli lac

operon and

two strongly expressed plant genes,

SRS1

and

ACT2

, as out-

lined in Figure 1. The strategy employs a modified

Lac

I repres-

sor protein for the permanent repression in shoots of target

genes expressed from the

ACT2

promoter. First, we assem-

bled a

Lac

I repressible vector,

A2pot

, in which the

ACT2

promoter region was modified to contain wild-type

E. coli lacO

sequences, which are the DNA binding sites for the repressor

.

The wild-type

Arabidopsis ACT2

actin promoter directs strong

572

Tehryung Kim

et al

.

© Blackwell Publishing Ltd,

Plant Biotechnology Journal

(2005),

3

, 571–582

constitutive gene expression in all vegetative shoot, root,

and inflorescence tissues and cell types in young and mature

plants (An

et al

., 1996; Kandasamy

et al

., 2002). Second, we

constructed the

S1pt::lacIn

gene in which a modified

E. coli

Lac

I repressor protein was expressed using

SRS1

regulatory

elements in the vector

S1pt

. Various configurations of the

Glycine max SRS1 ribulose bisphosphate carboxylase (RUBISCO)

small subunit

promoter direct strong light-induced expression

in shoots, but no expression has ever been detected from this

promoter in root tissues, even when assayed with sensitive

reporters (Shirley

et al

., 1987; Dhankher

et al

., 2002). Third,

β

-glucuronidase (

GUS

) and mercuric ion reductase (

merA

)

sequences were cloned into

A2pot

to test the repressible

expression directed by this ROC. Fourth, we demonstrate that

both reporter constructs were strongly expressed in shoots

and roots, except in the presence of

S1pt::lacIn

, which shut

down the

lacO

-modified promoter in shoots, leaving strong

activity in roots.

Results

Constructing a root-specific ROC

The

A2pt

expression vector contains promoter, 5

′

, and 3

′

UTR sequences, and polyadenylation signals from the

ACT2

gene (Kandasamy

et al

., 2002). The

A2pt

sequence

was modified into a

Lac

I repressible expression vector,

A2pot

,

by adding two 25-bp

lacO

operator sequences from the

E. coli lac

operon, as diagrammed in Figure 1. These sequences

were inserted 75 bp apart, one immediately after the

ACT2

TATA box sequence and the other immediately after the

multiple start sites of transcription, and replaced two 25-bp

stretches of native sequence. The two 25-bp regions of

ACT2

native promoter that were replaced are poorly conserved

with the closely related

Arabidopsis ACT8

constitutive actin

promoter sequence and, thus, were predicted to be non-

essential for the strong constitutive expression of

ACT2

(An

et al

., 1996).

The

S1pt

expression vector has promoter, 5

′

, and 3

′

UTR

sequences and polyadenylation signals from the light-induced

shoot-specific

RUBISCO small subunit

gene

SRS1

(Materials

and methods).

S1pt

was used to drive expression of a modi-

fied

E. coli lac

repressor,

LacIn

, in shoots, and hence limit

the expression of

A2pot

to roots, as diagrammed in Figure 1.

The

lacIn

DNA coding sequence was modified from the wild-

type

E. coli

sequence so that it encoded a protein with

the C-terminal eukaryotic nuclear localization signal (NLS),

SSVVHPKKKRKV. The

lacIn

sequence was cloned as an

in-frame translational fusion following the ATG initiation

codon in

S1pt

to make

S1pt::lacIn

(Figure 1).

Immunolocalization of LacIn protein

T

1

generation transgenic plants expressing

S1pt::lacIn

were

obtained by selecting for the linked hygromycin resistance

gene (

hptII

) marker. The tissue specificity of

LacIn

repressor

expression was assayed immuno-histochemically using a

Lac

I-specific antibody in T

2

generation transgenic

Arabidopsis

seedlings and plants

.

LacIn protein was localized to shoot

Figure 1 Construction of a root-specific repressor-operator gene complex based on shoot-repression of a constitutive promoter. (a) The S1pt::lacIn gene encoded a modified bacterial lac repressor that contained a nuclear localization signal (NLS). The lacIn sequence was transcribed from the strong, light-induced, shoot-specific regulatory elements from the SRS1 gene. The A2pot vector used a modified constitutive actin ACT2 promoter regulatory elements that contained two bacterial lacO operator sequences flanking the start site for transcription (ts). A2pot was used to express two target genes of interest, merA and GUS. When S1pt::lacIn was expressed in shoots, the lacIn protein (green) entered the nucleus and repressed A2pot::GUS or A2pot::merA transcription. In roots, where no repressor protein was expressed, and the A2pot::GUS and A2pot::merA constructs were still active. (b) When plants expressing LacIn (green) from the S1pt::lacIn gene in shoots were crossed with those constitutively expressing MerA (blue) from the A2pot::merA gene, the hybrids expressed MerA primarily in roots.

Root-specific repressor-operator gene complex

573

© Blackwell Publishing Ltd,

Plant Biotechnology Journal

(2005),

3

, 571–582

tissues only of

S1pt::lacIn

plants, shown in Figure 2. Compar-

ison of LacIn localization with DNA-specific DAPI staining

revealed that essentially all shoot nuclei contained repressor

protein. LacIn staining was concentrated in the nuclei with

little staining detected in the cytoplasm, suggesting the

composite NLS was functional in directing efficient nuclear

transport. LacIn was found in the nuclei of essentially all leaf

cells from seedlings and older plants. No LacIn protein was

detected in the nuclei or cytoplasm of root cells from

S1pt::lacIn

plants (Figure 2c), demonstrating the specificity of

the expression in shoot tissues. Furthermore, no immuno-

staining was observed in the shoots or roots of wild-type

plants (Figure 2b,d), demonstrating the specificity of the

immunolocalization assay.

Expression pattern of a GUS reporter with root-specific

ROC

The

β

-glucuronidase (GUS) reporter gene sequence was

cloned under control of A2pot sequences to test the tissue

specificity of the ROC system outlined in Figure 1. T1 genera-

tion transgenic plants were obtained by selecting for the

linked kanamycin resistance gene (nptII ). A2pot::GUS expres-

sion was examined in T2 generation transgenic seedlings and

plants. Histochemical staining for GUS activity showed that

A2pot::GUS was highly expressed in all vegetative tissues, as

shown at the top of Figure 3. A closer examination of these

vegetative tissues revealed strong expression in all cell types,

as reported previously for wild-type ACT2 promoter-GUS

fusions stained for 24 h (An et al., 1996). Longer staining of

the A2pot::GUS plants made the vegetative organs and

tissues opaque blue with significant blue pigment leaching

into the medium, further demonstrating the strength of GUS

enzyme expression. After 2 or 24 h of staining with substrate,

no significant blue GUS precipitate was observed in pollen,

ovules, or developing seeds (not shown), consistent with the

lack of expression of wild-type ACT2 promoter-GUS fusions

in reproductive tissues (An et al., 1996). Thus, it does

not appear that insertion of the lacO sequences in A2pot

significantly altered the strong, vegetative expression pattern

from that of the wild-type ACT2 promoter. Fifty of 50 T1

generation plants transformed with this construct showed

the same expression pattern, with a several-fold variation

in the levels of expression among the various lines.

Plants expressing both the A2pot::GUS and S1pt::lacIn

genes were obtained by cotransformation of the two con-

structs and coselection for the growth of seedlings expressing

both nptII and hptII markers, respectively. The majority of the

transgenic plant lines demonstrating resistance to both drugs

showed little or no GUS expression in shoot tissues, while

root expression was already quite strong in these plants after

2 h of histochemical staining. A typical doubly transformed

plant is shown in Figure 3c,d. Cross sections of histochemically

stained root confirmed that all tissue and cell types in roots

exhibited strong GUS activity, including root hairs, epidermis,

cortex, vascular cylinder (Figure 3g,h), and root tip (not shown),

consistent with the known root expression pattern of the

wild-type actin ACT2 promoter (An et al., 1996). Root stain-

ing was indistinguishable from that in A2pot::GUS plants

lacking the repressor. Among this majority of plants showing

Figure 2 Expression of the LacIn protein in nuclei. Subcellular localization of LacIn in shoot (a, b) and root (c, d) cells of transgenic plants expressing S1pt::lacIn. The LacIn protein was localized in dissociated and fixed cells using the commercial anti-LacI rabbit polyclonal antibody (#217449 Stratatene) and FITC-goat-anti-rabbit antisera (Sigma) (left panel, a–d). LacIn protein was detected in the nucleus of the leaf (a, left), but not root (c, left) cells of S1pt::lacIn plants and was not detected in wild-type plants (b, d, left). Cells were costained with DAPI to localize DNA and nuclei (right panel a-d). Young leaves and roots of T2 generation plants were examined in these experiments.

574 Tehryung Kim et al.

© Blackwell Publishing Ltd, Plant Biotechnology Journal (2005), 3, 571–582

root-specific expression, occasional weak staining was observed

in cotyledons and at the distal margins of the shoots of newly

emerging leaves (not shown). Nontransgenic control plants

showed no GUS staining of shoots or roots (Figure 3e,f).

In order to quantify both expression and repression levels

we examined GUS activity in the roots and shoots of a

population of independently transformed plant lines using the

fluorescent substrate MUG that can be quantified over a wide

dynamic range. We compared the data for 60 T1 generation

plant lines transformed with both A2pot::GUS + S1pt::lacIn

to data from 15 T1 plants transformed with A2pot::GUS alone.

The 15 A2pot::GUS plants lacking repressor had shoot fluo-

rescence levels that averaged 150% of that in roots when data

were normalized for fresh shoot and root weights (Figure 4).

Figure 3 The root-specific expression of β-glucuronidase from the A2pot::GUS sequence in transgenic Arabidopsis expressing the LacIn repressor in leaves. Seedlings transformed with the A2pot::GUS gene alone (a and b); seedlings cotransformed with both A2pot::GUS and S1pt::lacIn (c and d); and wild-type seedlings were examined for GUS enzyme expression. GUS expression in shoots (a) was knocked down by the expression of S1pt::lacIn in shoots (c), but not in roots (d). Cross section of root tissue confirmed that all types of cells in root tissue exhibited GUS activity (d).

Root-specific repressor-operator gene complex 575

© Blackwell Publishing Ltd, Plant Biotechnology Journal (2005), 3, 571–582

The doubly transformed plants showed a wide range of

fluorescence phenotypes from strong to weak repression of

shoot activity by LacIn. Therefore, we binned these plants

into three groups: those showing strong, intermediate, or

weak shoot repression based on the ratios of shoot to root

fluorescence (Figure 4a). Among the 60 cotransformants, the

17 plants with strong shoot repression showed shoot expres-

sion levels that averaged much less than 0.2% of root values

(Figure 4b). In other words, the average root levels were more

than 500 times higher than shoot levels. Some of these plants

showed root GUS expression with 1000- to 5000-fold lower

shoot expression levels as shown in Figure 4b for the two

plants in this group with the highest and lowest levels of total

root GUS activity. The 18 intermediate plants showed shoot

activity that was 10% to 15% of the root expression levels. The

25 plants in the weakly repressed group showed shoot levels

of GUS expression that were equal to the root levels and,

thus, only slightly repressed relative to control plants lacking

repressor. While the observed fluorescence values for plants

from each group varied widely (Figure 4b), the absolute levels

of root GUS activity did not appear to correlate with the pres-

ence or absence of the repressor construct S1pt::lacIn. This

suggested LacIn repressor was never expressed in sufficient

quantities in roots to repress root GUS expression. Stably

transformed T2 plant lines maintained similar levels of shoot

repression and strong root expression as shown for a plant

with strong repression in Figure 3c,d.

Root-specific expression of a merA reporter

We assembled a second reporter construct, A2pot::merA,

designed to express a modified E. coli mercuric ion reductase

merA gene that could be repressed by LacIn (see Materials

and methods). Plants were independently transformed with

S1pt::lacIn and A2pot::merA and lines were identified that

contained single T-DNA insertions based on segregation

analysis of T2 generation plants. Ten independent plant lines

with high levels of LacIn or MerA protein expression in shoots

were identified by Western blot analysis for protein expres-

sion. Representative plants with single A2pot::merA trans-

gene insertions are shown in Figure 5 (see lanes labelled

S1pt::lacIn-5, A2pot::merA-1, A2pot::merA-5, A2pot::merA-

10). These single insertion lines and their genetic hybrids with

S1pt::lacIn-5 were examined for organ-specific MerA protein

expression. The shoots and roots of the three lines expressing

A2pot::merA alone showed high levels of MerA protein. In

contrast, MerA protein expression was significantly repressed

in shoots of A2pot::merA × S1pt::lacIn hybrid plants. The low

levels of MerA expression in shoots corresponded with LacIn

Figure 4 Quantification of relative β-glucuronidase shoot and root activities. Fluorometric assays were used to quantify β-glucuronidase (MUG) activity in shoots and roots of T1 generation plants cotransformed with both A2pot::GUS and S1pt::LacIn and control plants containing the A2pot::GUS construct alone. A. Shoot fluorescence is expressed as a percent of root fluorescence of four-week-old plants. The 60 doubly transformed plants were grouped into those showing strong (99–100%), intermediate (70–90%), or weak repression (0–65%). Average root levels were defined as 100%. Standard errors are shown for shoot data. B. Shoot and root MUG fluorescence values are shown for the two individual plants with the weakest and strongest root MUG activity from each group in panel A along with the average values for each group. These data demonstrate the wide range of fluorescent activity obtained in populations of transformed plant lines. A & B. All GUS data were normalized to fresh shoot or root weights.

576 Tehryung Kim et al.

© Blackwell Publishing Ltd, Plant Biotechnology Journal (2005), 3, 571–582

protein expression from the S1pt::lacIn-5 parent repressor

line.

These Western results were quantified as summarized in

Figure 5f. The levels of shoot-specific MerA expression were

repressed more than 20-fold in some lines, when crossed

to the S1pt::lacIn-5 repressor line. For example, in the

S1pt::lacIn-5 × A2pot::merA-1 hybrid the levels of MerA were

not detectable over background as compared to the strong

MerA expression observed in the A2pot::merA-1 parent line.

Other lines like S1pt::lacIn-5 × A2pot::merA-10 were repressed

in shoots sevenfold below the A2pot::merA-10 parent line.

The average repression of MerA in shoots, among the three

shoot samples shown in frame a, was approximately nine-

fold. The limitation in quantifying repression in these experi-

ments is that shoot MerA levels below 5% of that shown for

the strong A2pot::merA lines in Figure 5 cannot be detected

on Western blot. Thus, the dynamic range of these measure-

ments, approximately 20-fold, underestimated the levels of

MerA repression in shoots. These levels of repression were

reproducible for each line in repetitions of this experiment.

Hybrid plants were examined through the T3 generation and

still showed the same strong, root-specific expression pat-

tern, suggesting this root-specific ROC is genetically stable.

The hybrid plants expressing MerA protein predominantly

in roots (A2pot::merA × S1pt::lacIn) were plated on media

with 80 µM Hg(II) (Figure 6). These plants were as tolerant to

Hg(II) as were plants expressing MerA from the A2pot::merA

construct alone without LacIn repression. As expected from

previous studies, wild-type seeds and seedlings die on these

same Hg(II) concentrations. Thus, it appears that detoxifica-

tion of Hg(II) to Hg(0) in roots is sufficient to confer resistance

to above-ground portions of the plant.

ROC controlled MerA expression was compared with that

from the tobacco RB7 root-specific gene expression system

(Yamamoto et al., 1991). Twenty independent transgenic

Arabidopsis seedling expressing the RB7::merA fusion were

made and the most Hg(II) resistant plant line was selected

for further study. MerA expression levels in roots from ROC

system were approximately fivefold higher compared to those

with RB7::merA construct (Figure 6a). Root growth on the

Hg(II) containing medium was much higher for S1pt::lacIn

× A2pot::merA plants with than for the RB7::merA plants,

which barely grew on the mercury-containing medium

(Figure 6b).

Discussion

ROCs could be built using repressor proteins and operator

DNA binding sites from several known systems of gene

regulation. The lac operon is one of the best-characterized reg-

ulatory systems and was among the first discovered in E. coli

(Pardee et al., 1958; Jacob and Monod, 1959). The tetrameric

LacI repressor protein binds naturally to a lacO operator DNA

sequence in the E. coli lac promoter region, repressing tran-

scription of downstream genes for lactose utilization. Addi-

tion of a small molecule inducer releases the repressor and

activates transcription. The E. coli lacI repressor and lacO

Figure 5 Assays of root-specific MerA protein expression: Western blot assays of MerA protein in roots and shoots of plants expressing A2pot::merA and S1pt::lacIn were compared to plants expressing either gene alone and wild-type plants. The complementary expression of LacIn in shoots is shown for comparison. Mouse monoclonal antibodies were used to probe Western blots for MerA (a & b) and LacIn (e). Immunoreactions were enhanced and images quantified as described in Materials and methods. Coomassie-stained samples of total shoot and root protein are shown to illustrate the approximately equal loading among samples (c & d). All lanes of the polyacrylamide gels were loaded with 20 µg of total protein. The levels of MerA among transgenic lines lacking repressor, A2pot::merA-1, A2pot::merA-5, and A2pot::merA-10 varied by about 30%, being 58.6, 77.6, and 80.6 density units, respectively. f. Relative levels of MerA protein expression with and without the S1pt::LacIn gene. The levels of MerA protein in each line were normalized to the levels of MerA in the roots of the parent lines lacking LacIn. The data for the three sets of plant lines examined above (a & b) were combined to give standard errors. Image J was used to quantify the intensity of bands from Western blots (http://rsb.info.nih.gov/ig/).

Root-specific repressor-operator gene complex 577

© Blackwell Publishing Ltd, Plant Biotechnology Journal (2005), 3, 571–582

operator sequences have been combined with eukaryotic

sequences to construct inducible target gene expression

systems in mammalian and plant cells (Fuerst et al., 1989;

Ulmasov et al., 1997). These constructs are the basis for a few

commercial mammalian and plant expression vectors designed

for inducible gene expression such as LacSwitch from Strata-

gene (LaJolla, CA). However, problems with low expression of

the target genes and inefficient chemical induction have

prevented the widespread application of this technology in

animal and plant cells (Fuerst et al., 1989; Hu and Davidson,

1990; Wilde et al., 1992). Furthermore, none of the published

and commercially available lac operon-based regulatory sys-

tems are designed for constitutive organ- or tissue-specific

expression of target genes.

The results presented herein show that expression of LacIn

repressor from S1pt::lacIn in conjunction with reporters

expressed from the A2pot cassette resulted in strong, consti-

tutive root-specific expression of the reporter proteins. The

LacIn protein was localized to the nucleus of leaf, but not root

cells, and clearly worked as an effective repressor of the A2pot

expression constructs in shoots. Furthermore, the insertion in

A2pot of two 25-bp lacO sequences, one immediately after

the TATA box and one after the start of transcription, did not

appear to hinder strong expression of either GUS or MerA

from the A2pot promoter (Figures 4 and 5, respectively).

S1pt::LacIn-driven repression of shoot GUS activity from

A2pot::GUS was very strong in a subset of the 60 doubly

transformed plants, averaging greater than 500-fold in 28%

of the plants and exceeding 2000-fold repression in many

plants. Repression levels are likely a function of both the level

of LacIn repressor and genome position affecting the avail-

ability of the two operator sequences in the A2pot transgenes

to bind repressor. Furthermore, plants with repressor activity

generally showed high levels of GUS or MerA expression in

roots, levels equivalent to that observed without repressor.

Thus, it appears the S1p::lacIn construct expressed insuffi-

cient LacIn repressor in roots to cause measurable repression

in below-ground tissues.

We used the Lac repressor protein to permanently silence

expression of two specific reporter genes in shoots as part of

a root-specific ROC. The root-specific gene expression strat-

egy described does not require the lac system to have induc-

ible properties. For repression to occur in shoot cells, the Lac

repressor protein needed to be stable, to be concentrated in

the nucleus, and to maintain its strong affinity for operator

sequence. The affinity of the wild-type bacterial repressor for

the operator is moderate in strength, about ∼Kd = 10−12 mole/L,

depending somewhat upon the conditions used to assay

affinity (Betz et al., 1986; Whitson et al., 1986). Weak GUS

expression was occasionally detected from the A2pot::GUS

construct in cells at distal leaf margins in the presence of

the repressor transgene in the intermediately and strongly

repressed lines. This leaky expression could have been due to

inaccessibility of the operator DNA sequences. Alternatively,

Figure 6 MerA protein levels and mercury tolerance with different root-specific expression systems. (a) Western blot assays of MerA protein in roots of plants expressing A2pot::merA and S1pt::lacIn, A2pt::merA, and RB7::merA were compared to wild-type (WT) plants. Mouse monoclonal antibody mAbMerA was used to probe membrane imprints for MerA. Coomassie-stained duplicate gel samples of total shoot and root protein are shown to illustrate the approximately equal loading among samples. (b) The tolerance of hybrid plants coexpressing the A2pot::merA and S1::lacIn genes was compared to plants expressing A2pt::merA or RB7::merA, and wild-type. Transgenic seeds were germinated on half-strength MS media containing 0 or 80 µM Hg(II) and grown vertically for 4 weeks.

578 Tehryung Kim et al.

© Blackwell Publishing Ltd, Plant Biotechnology Journal (2005), 3, 571–582

it may have been due to reduced transcription of S1pt::lacIn

gene or more rapid turnover of the repressor RNA or protein,

or some combination of these effects occurring in some small

number of cells. Looking toward improving this system fur-

ther, it is likely that tighter binding to lacO sequences of a fur-

ther modified LacIn protein, should lower the concentration

of repressor required for complete repression and, thus, lead

to more efficient repression in shoots. Mutant LacI repressor

proteins with 10 000-fold higher affinities for operator

sequences and mutant operator DNA sequences with 30-

fold higher affinities for repressor have been reported in

bacteria (Pfahl, 1979; Betz et al., 1986; Falcon and Matthews,

2000). Thus, there are many options for further improving ROC

repression stringency using elements from the lac operon.

The root-specific ROC developed has a wide potential

application, because strong, constitutive, root-specific pro-

moters are essential to many areas of plant biotechnology,

including enhanced plant nutrition, control of plant root dis-

eases, food safety, and phytoremediation of contaminated

soils. For example, the enhanced uptake of essential nutrients

such as iron, zinc, potassium, and/or nitrate into plant roots

should enhance plant yield and the nutritional quality of

foods (Guerinot and Eide, 1999; Guerinot, 2001). Root-specific

protection from pathogens would improve crop and food safety

by both restricting expression of antifungal and antinematode

agents to roots and preventing the synthesis of native toxins

by these and associated pathogens. Numerous fungal path-

ogens target roots, such as those causing sudden death syn-

drome in soybean, take-all disease in wheat, and wilt diseases

in vegetable crops (Iqbal et al., 2002; Anjaiah et al., 2003;

Gutteridge et al., 2003; Lievens et al., 2003). Nematodes nearly

all target roots and several diseases such as soybean cyst

nematode (Lewers et al., 2002) and nematode root-knot

(Giblin-Davis et al., 2003) cause significant annual loss of

crops to US agriculture. Targeting the synthesis of fungal-

specific and nematode-specific toxins with the root-specific

ROC described herein should improve the yield and quality of

many food, forage, and animal feed crops (Gao et al., 2000).

Another application with profound health and environmen-

tal implications is the phytoremediation of toxic pollutants in

the soil. Among the genes that need to be enhanced for root

expression are those encoding catabolic enzymes that degrade

organic pollutants, plasma membrane transporters that uptake

pollutants, transporters involved in the short-distance move-

ment of pollutants from root-epidermal cells to the root vas-

cular system, and enzymes involved in the transformation of

elemental pollutants like mercury and arsenic to less toxic

chemical species (Meagher, 2000). A key to many engineered

phytoremediation strategies is plant tolerance to toxins once

they enter plant cells (Salt et al., 1995). The root-specific

expression of MerA provided high levels of mercury tolerance

to transgenic plants. Based on data presented herein, we

have sufficient root expression of MerA in various root tissues

to protect these tissues from ionic mercury’s toxic effects by

converting reactive Hg(II) to relatively nontoxic, nonreactive

metallic mercury, Hg(0).

Considering the need for root-specific promoters, several

previously examined, native root-specific genes might have

been chosen as a source of promoter and terminator sequences,

but none of these could be depended upon to drive strong

target gene expression in all root tissues and cells. For exam-

ple, large numbers of P450 encoding genes may be root-

specific, but no single P450 of the more than 270 CYP genes

in Arabidopsis has been identified that is both strongly

expressed and expressed in all root tissues (Xu et al., 2001).

Other root-specific genes, including the Zea maize zmGRP3,

ZRP2, and ZRP4 genes (Held et al., 1993, 1997; Goddemeier

et al., 1998) and tomato LeRse-1 gene (Lauter, 1996) all

showed relatively weak levels of root-specific expression.

Localization of ZRP2 and ZRP4 mRNAs suggests their

expression is concentrated in cortical parenchyma cells of

young roots and the root endodermis, respectively. In

transgenic plants, a few plant promoters driving root-specific

expression of reporters have been described, such as Arabi-

dopsis SCARECROW (Sabatini et al., 2003) and tobacco

TobRD2 (also NtQPT1) (Conkling et al., 1990) and TobRB7

(Yamamoto et al., 1991). Again, reporter expression patterns

from these promoters are restricted to subsets of tissues: root

meristem, root cortex, and root vascular cylindar, respectively.

Finally, promoter constructs derived from the Arabidopsis

Pyk10 gene drive moderate levels of expression in transgenic

plants that is relatively root-specific reporter (Nitz et al., 2001).

We tested the TobRB7 system and found MerA root activity

undetectable in most plant lines (data not shown). While the

level of total MerA protein in the most active RB7::merA plant

line was reasonably high, these plants still were not appreci-

ably mercury resistant (Figure 6). Perhaps this is due to the

lack of RB7 promoter activity in root epidermal and vascular

tissues (Yamamoto et al., 1991) leading to the death of these

tissues from Hg(II) toxicity. We also tested merA reporter

gene expression in transgenic plants driven from the root

promoter, NPQpT1 (TobRD2) (Conkling et al., 1990) and did

not obtain significant mercury resistance in a number of

transgenic plants (not shown). The negative results from these

experiments with two tobacco promoters, including low levels

of MerA protein in roots, the lack of any significant mercury

resistance, and substantial leaky expression in shoots of nearly

all plants motivated our development of a root-specific ROC.

Root-specific repressor-operator gene complex 579

© Blackwell Publishing Ltd, Plant Biotechnology Journal (2005), 3, 571–582

Engineered ROCs are needed to create a variety of strong

organ- or tissue-specific gene expression patterns and to

create genetic regulatory systems not found among natural

plant promoters. For example, it should be a simple extension

of this work to create a genetically regulated pollen-specific

ROC that would have wide application in the hybrid seed

industry. In the well-accepted BarStar/Barnase system (Beals

and Goldberg, 1997) male-sterile elite line plants are gener-

ated by the anther-specific expression of the toxic ribonucle-

ase, barnase, leading to male sterility. When this male-sterile

line is crossed to a second fully fertile elite plant line express-

ing barstar ribonuclease inhibitor the hybrid seeds are fully

fertile. The BarStar/Barnase system is very specific for these

two proteins. However, a much more widely applicable ROC

could be set up for hybrid seed production. The first male-

sterile elite parent A would be engineered by expressing a

toxic gene, such as that encoding a ribonuclese or RNA inter-

ference construct, from an anther-specific promoter contain-

ing lac operator sequences. A second elite parent, B, would

be engineered for the anther-specific expression of LacIn

repressor, but would remain fully fertile. Crosses between A

and B, with B as the male parent, would produce fully fertile

hybrid seeds and plants due to the repression of the toxic

gene in anthers. Variations on this mechanism could be used

to create fully sterile genetically modified plants that could be

released into the environment with safe containment of their

germplasm (Daniell, 2002).

The root-specific ROC described resulted in strong root

expression levels of GUS and MerA reporter proteins and

GUS reporter levels were repressed by three orders of mag-

nitude in the shoots of some plant lines. The wide variation

observed in both absolute expression levels and in the levels

of shoot repression suggests genome position effects play a

significant role in the performance of this ROC. In the future

we anticipate making improvements to our lac-based ROCs,

testing the root-specific system for application in useful field-

adapted plant species, and developing male-sterility systems

for hybrid seed production and complete-sterility for the

release of genetically modified plants. More immediately, we

will test the root-specific expression of transporters and

detoxifying enzymes that will assist with the development of

plants for the phytoremediation of mercury and arsenic.

Materials and methods

Plasmid construction

A 1072-bp lacI gene coding sequence (GenBank accession

#AY042185) was PCR amplified from plasmid pSK9118

(pMS421) (Grana et al., 1988) supplied by the Sidney Kush-

ner laboratory. The sense primer LacIs-NcoXho (TAGTAAG-

GAG GAACCACCTC GAGGCCATGG GTAAACCAGT

AACGTTATAC GAT) introduced sequences for XhoI and NcoI

sites at the AUG initiation codon on the 5′ end, and the anti-

sense primer LacIa-NLSBamHind (ÁTGTAAGCTT GGATCCTCAA

ACCTTTCTCT TCTTCTTAGG ATGAACAACA GAAGACTGCC

CGCTTTCCAG TCGGGAAA) introduced codons for the NLS

(SSVVHPKKKRKV-stop) and BamHI and HindIII sites at the 3′end of the lacI sequence. This artificial NLS was comprised of

sequences from the SV40 T-antigen (PKKKRKV) (Goldfarb

et al., 1986) and ARP7 (SVVHRK) (Kandasamy et al., 2003).

The amplified LacIn fragment was cleaved by XhoI and HindIII

and ligated into the compatible replacement region of pBlue-

script-SKII (Stratagene) to make the plasmid placIn. The lacIn

coding sequence was subcloned into the NcoI-HindIII replace-

ment region of the pS1pt plasmid to make S1pt::lacIn (Shirley

et al., 1987; Dhankher et al., 2002). The entire cassette

containing the lacIn encoding sequence and SRSI flanking

sequences was subcloned into the KpnI and SacI replace-

ment region of pCambia (Hajdukiewicz et al., 1994).

To construct A2pot, two copies of the 25-bp wild-type

bacterial lac operator sequence 5′GTGGAATTGT GAGCG-

GATAA CAATT were substituted for two sequences in the

ACT2 gene promoter: one immediately following the TATA

box and the other following the transcriptional start site as

shown in Figure 1. A two-fragment overlap extension PCR

mutagenesis strategy (Ho et al., 1989) was used to assemble

the modified ACT2 promoter sequence and replace the cor-

responding sequences of A2pt. The first round PCR paired

sense primer ACT2p-KpnS (GGTACCTGAT CTCAAATACA

TTGATA) with antisense primer Act2p-LacO1A (GCCGGA-

GATT CAAAACGGCT GATGAAAGTG AGGAGGACAA

CGAGACAATT CAATTGTTAT CCGCTCACAA TTCCACTTAT

ATAGGCGGGT TTATCTCTT), and sense primer Act2p-LacO2S

(AGCCGTTTTG AATCTCCGGC GACTTGACAG AGAAGAACAA

GGATGTGGAA TTGTGAGCGG ATAACAATTT AATCCAG-

GAGATTCATTCTC CGTTTTGAA) with antisense primer ACT2p-

1100 A (TAGCTATAAT CGAGCTAACT GAT). The resulting

814- and 310-bp fragments were assembled into a single

fragment in a second round of PCR using the same flanking

primers, ACT2p-KpnS and ACT2p-1100A. This fragment was

digested with KpnI and EcoRI and cloned into the corre-

sponding promoter replacement region of ACT2pt cassette

in bluescript, creating pA2pot.

Both the GUS and merA sequences were cloned into the

A2pot cassette to make the A2pot::GUS and A2pot::merA

genes. The 1851-bp GUS sequence was PCR amplified from

pBI221 (Tanaka et al., 1990) using the sense primer GUS-S

580 Tehryung Kim et al.

© Blackwell Publishing Ltd, Plant Biotechnology Journal (2005), 3, 571–582

(TAGAGTTCTA GAATAAAGGA GGAAAAACCG GTACCCCATG

GGATTACGTCCTG TAGAAACCCC AA) to add a XbaI site and

a NcoI site at the ATG codon and using the antisense primer

GUS-A (TTCGATCTCG AGGAGCTCGG ATCCTCATTG TTT-

GCCTCCC TGCTGCGGTT) to add XhoI, SacI and BamHI sites.

The merA77 sequence was cloned as a NcoI-BamHI fragment

into the corresponding replacement region of the A2pot vec-

tor to make A2pot::merA (Figure 1). The KpnI-SacI fragments

containing the A2pot::merA and A2pot::GUS sequences were

moved into the replacement region of pBIN19.

The TobRB7-5A promoter (Yamamoto et al., 1991) was cloned

as a XbaI-BamHI fragment into the compatible replacement

region upstream of the merA77 gene in a pBluescript-merA

construct. The TobRB7-5A promoter-merA gene fusion was

subcloned as a XbaI and KpnI fragment into the compatible

replacement region pBin19 vector with a nos terminator

(Bevan, 1984).

Plant and bacterial growth conditions

Plants were grown on agar media with MS salts or in soil with

16-h light and 8-h darkness. Agar media was supplemented

with various concentrations of mercury chloride (HgCl2) as

indicated.

The A2pot::GUS, A2pot::merA, and S1pt::lacIn constructs

were introduced alone and together into Arabidopsis thal-

iana (ecotype Columbia) by Agrobacterium mediated trans-

formation using the vacuum infiltration procedure (Ye et al.,

1999). T1 generation plants were selected for kanamycin

resistance (35 µg/mL) encoded on the pBIN19 (Bevan, 1984)

derived reporter clones and/or for hygromycin resistance

(50 µg/mL) encoded by the pCambia (http://www.cambia.org)

derived clone S1pt::lacIn. Selection was carried out in 1/2MS

media on agar plates for 7 days after plating seeds. Seedlings

were then moved to nonselective media for 2 weeks and

then transplanted to soil. T2 and T3 generation plants were

examined for the phenotypes described herein.

Histochemical and fluorometric β-glucuronidase assay

GUS activity was histochemically assayed using 4-week-old

seedlings. Seedlings were selected on the media with antibiotics

for 1 week and then transferred to the same media without

antibiotics. The incubation solution was modified from Stomp

(1992) and contained 10 mM EDTA, 0.1 M Na2HPO4, 0.1 M

NaH2PO4, 0.5 mM K ferricyanide, 0.5 mM K ferrocyanide,

and 1.0 mg/mL X-gluc. Tissue was added and incubated for

2 h at 37 °C with gentle shaking and thus modified from the

original staining protocol used for strong actin promoters

(An et al., 1996). Tissues were then treated with two

changes of 70% ethanol over 24 h to remove chlorophyll

and photographed under bright field lighting on a dissecting

microscope.

Root tissues were examined under a compound microscope

to evaluate the staining of individual cells. After staining

(as above) and before ethanol treatment, the tissues were

fixed with 2% glutaraldehyde in 0.2 M sodium phosphate

buffer pH 7.2, dehydrated by the sequence of ethyl glycerol

monoacetate–ethanol–propanol–butanol, infiltrated and

embedded in glycolmethacrylate (HistoResin, Co.), sectioned

at 8 micron thickness, then mounted on glass slides (Kim

et al., 1999).

GUS activity was quantified fluorometrically using 4-week-

old seedlings. GUS was extracted by grinding shoot and root

sections frozen in liquid nitrogen and suspending in an

extraction buffer that contained 50 mM NaHPO4 (pH 7.0),

10 mM β-mercaptoethanol, 0.1% sodium lauryl sarcosine,

10 mM Na2EDTA, and 0.1% Triton X-100. The extract was

reacted with 2 mM 4-methylumbelliferyl β-D-glucuronide

(MUG, Sigma, St Louis, MO) for 1 h at 37 °C, stopped with

0.2 M Na2CO3 stop buffer, and fluorescence was measured

using 360 nm excitation and 480 nm emission and a Synergy

HT Fluorometer (BioTec, Inc., Winooski, VT).

MerA and LacIn protein detection and localization

LacIn and MerA proteins were assayed on Western blots

using polyclonal rabbit antibody (Stratagene) and mono-

clonal antibody (Rugh et al., 1998), respectively. LacI and GUS

protein standards were obtained from (Stratagene) and

(G7896, Sigma), respectively. Total protein levels in plant

extracts were determined as described previously (Kandasamy

et al., 1999). LacIn protein was localized to shoot cell nuclei

using a previously described immunolocalization protocol in

which the fixed tissues were partially digested with sufficient

cell wall-degrading enzymes to release pieces of plant tissue,

but not individual cells (Kandasamy and Meagher, 1999;

Kandasamy et al., 1999).

Acknowledgements

We wish to thank Gay Gragson for editorial comments, Dilip

Shah for thoughts on root pathogens, and Sidney Kushner

for helpful insights into the E. coli lac operon. This work was

supported by USDA grant # 2001-35100-10652 to R.S.B.,

EPA Star Fellowship Program (U-915647-01-0) to A.C.P.H.,

and DOE EMSP grant #DE-FG07-96ER20257 and NIH grant

(GM 36397-18) to R.B.M.

Root-specific repressor-operator gene complex 581

© Blackwell Publishing Ltd, Plant Biotechnology Journal (2005), 3, 571–582

References

An, Y.Q., McDowell, J.M., Huang, S., McKinney, E.C., Chambliss, S.and Meagher, R.B. (1996) Strong, constitutive expression of theArabidopsis ACT2/ACT8 actin subclass in vegetative tissues. PlantJ. 10, 107–121.

Anjaiah, V., Cornelis, P. and Koedam, N. (2003) Effect of genotypeand root colonization in biological control of fusarium wilts inpigeonpea and chickpea by Pseudomonas aeruginosa PNA1. Can.J. Microbiol. 49, 85–91.

Beals, T.P. and Goldberg, R.B. (1997) A novel cell ablation strategyblocks tobacco anther dehiscence. Plant Cell, 9, 1527–1545.

Betz, J.L., Sasmor, H.M., Buck, F., Insley, M.Y. and Caruthers, M.H.(1986) Base substitution mutants of the lac operator: in vivo andin vitro affinities for lac repressor. Gene, 50, 123–132.

Bevan, M.W. (1984) Binary Agrobacterium vectors for plant trans-formation. Nuc. Acids. Res. 12, 8711–8721.

Conkling, M.A., Cheng, C.-I., Yamamoto, Y.T. and Goodman, H.M.(1990) Isolation of transcriptionally regulated root-specific genesfrom tobacco. Plant Physiol. 93, 1203–1211.

Daniell, H. (2002) Molecular strategies for gene containment intransgenic crops. Nat. Biotechnol. 20, 581–586.

Dhankher, O.P., Li, Y., Rosen, B.P., Shi, J., Salt, D., Senecoff, J.F.,Sashti, N.A. and Meagher, R.B. (2002) Engineering tolerance andhyperaccumulation of arsenic in plants by combining arsenatereductase and gamma-glutamylcysteine synthetase expression.Nat. Biotechnol. 20, 1140–1145.

Falcon, C.M. and Matthews, K.S. (2000) Operator DNA sequencevariation enhances high affinity binding by hinge helix mutants oflactose repressor protein. Biochemistry, 39, 11074–11083.

Fuerst, T.R., Fernandez, M.P. and Moss, B. (1989) Transfer of theinducible lac repressor/operator system from Escherichia coli to avaccinia virus expression vector. Proc. Natl. Acad. Sci. USA, 86,2549–2553.

Gao, A.G., Hakimi, S.M., Mittanck, C.A., Wu, Y., Woerner, B.M.,Stark, D.M., Shah, D.M., Liang, J. and Rommens, C.M. (2000)Fungal pathogen protection in potato by expression of a plantdefensin peptide. Nat. Biotechnol. 18, 1307–1310.

Giblin-Davis, R.M., Williams, D.S., Bekal, S., Dickson, D.W., Brito, J.A.,Becker, J.O. and Preston, J.F. (2003) ‘Candidatus pasteuria usgae’sp. nov., an obligate endoparasite of the phytoparasitic nematodeBelonolaimus longicaudatus. Int. J. Syst. Evol. Microbiol. 53, 197–200.

Goddemeier, M.L., Wulff, D. and Feix, G. (1998) Root-specific expres-sion of a Zea mays gene encoding a novel glycine-rich protein,zmGRP3. Plant Mol. Biol. 36, 799–802.

Goldfarb, D.S., Gariepy, J., Schoolnik, G. and Kornberg, R.D. (1986)Synthetic peptides as nuclear localization signals. Nature, 322,641–644.

Grana, D., Gardella, T. and Susskind, M.M. (1988) The effects ofmutations in the ant promoter of phage P22 depend on context.Genetics, 120, 319–327.

Guerinot, M.L. (2001) Improving rice yields – ironing out the details.Nat. Biotechnol. 19, 417–418.

Guerinot, M.L. and Eide, D. (1999) Zeroing in on zinc uptake in yeastand plants. Curr. Opin. Plant Biol. 2, 244–249.

Gutteridge, R.J., Bateman, G.L. and Todd, A.D. (2003) Variationin the effects of take-all disease on grain yield and quality ofwinter cereals in field experiments. Pest Manag. Sci. 59, 215–224.

Hajdukiewicz, P., Svab, Z. and Maliga, P. (1994) The small, versatilepPZP family of Agrobacterium binary vectors for plant transforma-tion. Plant Mol Biol. 25, 989–994.

Held, B.M., John, I., Wang, H., Moragoda, L., Tirimanne, T.S., Wurtele, E.S.and Colbert, J.T. (1997) Zrp2: a novel maize gene whose mRNAaccumulates in the root cortex and mature stems. Plant Mol. Biol.35, 367–375.

Held, B.M., Wang, H., John, I., Wurtele, E.S. and Colbert, J.T. (1993)An mRNA putatively coding for an O-methyltransferase accumu-lates preferentially in maize roots and is located predominantly inthe region of the endodermis. Plant Physiol. 102, 1001–1008.

Ho, S.N., Hund, H.D., Horton, R.M., Pullen, J.K. and Pease, L.R. (1989)Site-directed mutagenesis by overlap extension using thepolymerase chain reaction. Gene, 77, 51–59.

Hu, M.C. and Davidson, N. (1990) A combination of derepression ofthe lac operator-repressor system with positive induction by glu-cocorticoid and metal ions provides a high-level-inducible geneexpression system based on the human metallothionein-IIA pro-moter. Mol. Cell Biol. 10, 6141–6151.

Iqbal, J., Afzal, J., Yaegashi, S., Ruben, E., Triwitayakorn, K., Njiti, N.,Ahsan, R., Wood, J. and Lightfoot, A. (2002) A pyramid of locifor partial resistance to Fusarium solani f. sp. glycines maintainsMyoinositol-1-phosphate synthase expression in soybean roots.Theor. Appl. Genet. 105, 1115–1123.

Jacob, F. and Monod, J. (1959) Genes of structure and genes of reg-ulation in the biosynthesis of proteins. C R. Hebd. Seances Acad.Sci. 249, 1282–1284.

Kandasamy, M.K. and Meagher, R.B. (1999) Actin–organelle inter-action: association with chloroplast in Arabidopsis leaf mesophyllcells. Cell Motil. Cytoskeleton, 44, 110–118.

Kandasamy, M.K., McKinney, E.C. and Meagher, R.B. (1999) The latepollen-specific actins in angiosperms. Plant J. 18, 681–691.

Kandasamy, M.K., McKinney, E.C. and Meagher, R.B. (2002)Functional non-equivalency of actin isovariants in Arabidopsis.Mol. Biol. Cell, 13, 251–261.

Kandasamy, M.K., McKinney, E.C. and Meagher, R.B. (2003)Cell cycle–dependent association of Arabidopsis actin-relatedproteins AtARP4 and AtARP7 with the nucleus. Plant J. 33, 939–948.

Kim, T., Choudhury, M.K.U. and Wetzstein, H.Y. (1999) A quantitativeand histological comparison of GUS expression with different pro-moter constructs used in microprojectile bombardment of peanutleaf tissue. In Vitro Cell. Dev. Biol. Plant, 35, 51–56.

Lauter, F.-R. (1996) Root-specific expression of the LeRse-1 gene intomato is induced by exposure of the shoot to light. Mol. Gen.Genet. 252, 751–754.

Lewers, K., Heinz, R., Beard, H., Marek, L. and Matthews, B. (2002)A physical map of a gene-dense region in soybean linkage groupA2 near the black seed coat and Rhg (4) loci. Theor. Appl. Genet.104, 254–260.

Lievens, B., Brouwer, M., Vanachter, A.C., Levesque, C.A.,Cammue, B.P. and Thomma, B.P. (2003) Design and developmentof a DNA array for rapid detection and identification of multipletomato vascular wilt pathogens. FEMS Microbiol. Lett. 223, 113–122.

Meagher, R.B. (2000) Phytoremediation of toxic elemental andorganic pollutants. Curr. Opin. Plant Biol. 3, 153–162.

Nitz, I., Berkefeld, H., Puzio, P.S. and Grundler, F.M. (2001) Pyk10,a seedling and root specific gene and promoter from Arabidopsisthaliana. Plant Sci. 161, 337–346.

582 Tehryung Kim et al.

© Blackwell Publishing Ltd, Plant Biotechnology Journal (2005), 3, 571–582

Pardee, A.B., Jacob, F. and Monod, J. (1958) The role of the induciblealleles and the constrtutive alleles in the synthesis of beta-galactosidase in zygotes of Escherichia coli. C R Hebd. SeancesAcad. Sci. 246, 3125–3128.

Pfahl, M. (1979) Tight-binding repressors of the lac operon: selec-tion system and in vitro analysis. J. Bacteriol. 137, 137–145.

Rugh, C.L., Senecoff, J.F., Meagher, R.B. and Merkle, S.A. (1998)Development of transgenic yellow poplar for mercury phytoreme-diation. Nat. Biotechnol. 16, 925–928.

Sabatini, S., Heidstra, R., Wildwater, M. and Scheres, B. (2003)SCARECROW is involved in positioning the stem cell niche in theArabidopsis root meristem. Genes Dev. 17, 354–358.

Salt, D.E., Blaylock, M., Kumar, N.P.B.A., Dushenkov, V., Ensley, B.D.,Chet, I. and Raskin, I. (1995) Phytoremediation: a novel strategyfor the removal of toxic metals from the environment using plants.Biotechnology, 13, 468–474.

Shirley, B.W., Berry-Lowe, S.L., Rogers, S.G., Flick, J.S., Horsch, R.,Fraley, R.T. and Meagher, R.B. (1987) 5′ proximal sequences of asoybean ribulose-1,5-bisphosphate carboxylase small subunit genedirect light and phytochrome controlled transcription. Nucl. AcidsRes. 15, 6501–6514.

Stomp, A.M. (1992) GUS Protocols: Using the GUS Gene as aReporter of Gene Expression (Gallagher, S.R., ed.), pp. 103–112.Palo Alto, CA: Academic Press.

Ye G.-N., Stone, D., Pang, S.-Z., Creely, W., Gonzalez, K. and Hinchee, M.

(1999) Arabidopsis ovule is the target for Agrobacterium in plantavacuum infiltration transformation. Plant J. 19, 249–257.

Tanaka, A., Mita, S., Ohta, S., Kyozuka, J., Shimamoto, K. andNakamura, K. (1990) Enhancement of foreign gene expression bya dicot intron in rice but not in tobacco is correlated with anincreased level of mRNA and an efficient splicing of the intron.Nucl. Acids Res. 18, 6767–6770.

Ulmasov, B., Capone, J. and Folk, W. (1997) Regulated expressionof plant tRNA genes by the prokaryotic tet and lac repressors.Plant Mol. Biol. 35, 417–424.

Whitson, P.A., Olson, J.S. and Matthews, K.S. (1986) Thermody-namic analysis of the lactose repressor-operator DNA interaction.Biochemistry, 25, 3852–3858.

Wilde, R.J., Shufflebottom, D., Cooke, S., Jasinska, I., Merryweather, A.,Beri, R., Brammar, W.J., Bevan, M. and Schuch, W. (1992) Controlof gene expression in tobacco cells using a bacterial operator-repressor system. EMBO J. 11, 1251–1259.

Xu, W., Bak, S., Decker, A., Paquette, S.M., Feyereisen, R. andGalbraith, D.W. (2001) Microarray-based analysis of gene expres-sion in very large gene families: the cytochrome P450 gene super-family of Arabidopsis thaliana. Gene, 272, 61–74.

Yamamoto, Y.T., Taylor, C.G., Acedo, G.N., Cheng, C.L. andConkling, M.A. (1991) Characterization of cis-acting sequencesregulating root-specific gene expression in tobacco. Plant Cell, 3,371–382.