Modifying the pollen coat protein composition in Brassica

Modifying the pollen coat protein composition in Brassica

Elizabeth Foster, Danielle Schneiderman, Michel Cloutier, Stephen Gleddie and Laurian S. Robert*

Eastern Cereal and Oilseed Research Centre, Agriculture and Agri-Food Canada, 2091 K.W. Neatby Bldg.,

960 Carling Ave., Ottawa, ON K1A 0C6, Canada

Received 12 March 2002; accepted 29 April 2002.*For correspondence (fax +1613 759 1701; e-mail [email protected])

Summary

The interactions between pollen and stigma are essential for plant reproduction and are made possible

by compounds, such as proteins and lipids, located on their surfaces. The pollen coat is formed in part

by compounds synthesized in, and released from, the tapetum, which become transferred to the pollen

coat late in pollen development. In the Brassicaceae the predominant proteins of the mature pollen coat

are the tapetal oleosin-like proteins, which are highly expressed in, and ultimately transferred from, the

tapetum. Here we report the modi®cation of the protein composition of the pollen coat by the addition

of an active enzyme which was synthesized in the tapetum. The marker enzyme b-glucuronidase (GUS)

was successfully targeted to the pollen coat in transgenic Brassica carinata plants expressing GUS

translationally fused to a B. napus tapetal oleosin-like protein (BnOlnB;4). To our knowledge this is the

®rst demonstration of the targeting of an enzyme to the pollen coat.

Keywords: tapetum, pollen, pollen coat, tapetal oleosin-like protein, GUS

Introduction

The pollen coat (also known as the tryphine or pollenkitt)

®lls the pits of the exine pollen wall layer and consists

mostly of lipids and proteins (Heslop-Harrison et al., 1974).

The pollen coat serves a variety of functions such as

protecting pollen from the environment, attracting insect

pollinators, adhering pollen grains to insects and the

stigma surface, aiding hydration and germination, and

participating in self-compatibility and incompatibility (Bih

et al., 1999; Dickinson, 1994; Elleman and Dickinson, 1986,

1990; HuÈ lskamp et al., 1995). In Brassica, after a pollen

grain adheres to the stigma, the pollen coat ¯ows out from

the pits of the exine to form a contact zone between the

pollen grain and stigma surfaces. It is within this contact

zone that pollen hydration will occur in a compatible

reaction (Wolters-Arts et al., 1998).

The importance of the pollen coat for reproduction is

revealed by several studies. ECERIFERUM (CER) mutants

of Arabidopsis thaliana, which are de®cient in epicuticular

wax biosynthesis, lack pollen coat lipids and are condi-

tionally male sterile (Fiebig et al., 2000; HuÈ lskamp et al.,

1995; Preuss et al., 1993). Pollen coat lipids and fertility are

restored in a mutant-suppressor plant line (Fiebig et al.,

2000). The grp17-1 mutant of Arabidopsis, which lacks one

of the tapetal oleosin-like proteins, exhibits delayed pollen

hydration (May®eld and Preuss, 2000). Proteins involved in

self-incompatibility, including the male determinant S-

locus cysteine rich protein (SCR), are also located within

the pollen coat of Brassica oleracea (Cabrillac et al., 2001;

Shiba et al. 2001).

The predominant protein component of the pollen coat

in mature pollen of B. napus and related species are the

tapetal oleosin-like family of proteins (Murphy and Ross,

1998; Ross and Murphy, 1996). Tapetal oleosin-like mRNAs

are highly expressed and accumulate speci®cally in the

tapetum (Robert et al., 1994; Ross and Murphy, 1996),

which is the sporophytic cell layer surrounding and

nourishing the developing pollen grains. The tapetal

oleosin-like proteins are also expressed in the tapetum

where they are associated with unique cytoplasmic lipid

bodies called tapetosomes (Murphy and Ross, 1998).

However, unlike tapetal oleosin-like mRNAs, the tapetal

oleosin-like proteins persist after tapetal degeneration and

are released into the locule as part of the tapetosomes

prior to becoming localized to the mature pollen coat

(Murphy and Ross, 1998).

Structurally, tapetal oleosin-like proteins possess three

distinct domains, a hydrophobic central domain similar to

that of seed oleosins, ¯anked by two relatively hydrophilic

The Plant Journal (2002) 31(4), 477±486

ã 2002 Blackwell Science Ltd 477

N- and C-terminal domains (Robert et al., 1994; Ross and

Murphy, 1996). Although full-length tapetal oleosin-like

proteins initially accumulate in the tapetum, they become

cleaved into their mature forms after tapetal degeneration

at or near the junction between the hydrophobic central

domain and the C-terminal domain (Murphy and Ross,

1998). The Brassica pollen coat has been shown to contain

the mature C-terminal portions of the tapetal oleosin-like

proteins.

Here we report the modi®cation of the protein compos-

ition of the pollen coat in Brassica. The marker enzyme b-

glucuronidase (GUS) was targeted from the tapetum to the

pollen coat via a translational fusion with the B. napus

tapetal oleosin-like protein BnOlnB;4. This is the ®rst

demonstration of the targeting of an active enzyme to

the pollen coat. This strategy of pollen coat modi®cation

may be useful in the study of pollen/stigma interactions.

Results

Transformation of Brassica carinata with the TOG

translational fusion

The tapetal oleosin-like BnOlnB;4 gene was previously

isolated from B. napus based on its abundant expression

in anthers (Robert et al., 1994). We constructed a transla-

tional fusion between the BnOlnB;4 gene, including the

promoter and the entire coding region with the exception

of the TGA stop codon, and the uidA gene encoding b-

glucuronidase (GUS; Figure 1a). The junction between the

upstream BnOlnB;4 and downstream uidA sequences was

veri®ed to con®rm the integrity of the open reading frame

(Figure 1b).

The resulting tapetal oleosin-like::GUS (TOG) transla-

tional fusion construct was introduced into Ethiopian

mustard (B. carinata), a readily transformable relative of

B. napus, by Agrobacterium-mediated transformation.

Approximately 20 lines were retained after selection on

kanamycin.

Prior to screening for expression of the TOG construct in

the putative transgenic plants, anther development in

B. carinata ¯owers was determined histologically (data not

shown). The tapetum is present within anthers of ¯ower

buds from 2- to 6-mm in length but is absent in 7-mm

buds. Flowers open and release mature pollen once the

buds have reached approximately 8 mm in length. Anther

development in B. carinata parallels that of B. napus, in

which ¯ower bud length can be correlated to develop-

mental events within the anther (Scott et al., 1991).

However, B. carinata buds tend to be somewhat larger

than those of B. napus at similar developmental stages. In

B. napus, tapetal degeneration occurs at about 5 mm bud

length and ¯oral opening occurs by approximately 6±

7 mm bud lengths.

The putative TOG transgenic B. carinata plants were

initially screened histochemically for GUS enzymatic

activity of the fusion protein in anthers isolated from 5-

Figure 1. Diagram of the tapetal oleosin-like::GUS (TOG) translationalfusion construct.(a) Schematic map of the B. napus OlnB;4 promoter and coding regionsfused to the uidA sequence encoding b-glucuronidase (GUS) and thenopaline synthase terminator (TER). The BnOlnB;4 coding region iscomplete except for the stop codon. The ATG translational start site ofBnOlnB;4 and some restriction sites are indicated.(b) The reading frame between the BnOlnB;4 and GUS coding regions ispreserved in the synthetic nucleotide sequences used in the construction.Codons and corresponding amino acids are indicated.

Figure 2. Expression of the TOG translational fusion in transgenicB. carinata line 22 during anther development in 2-mm, 3-mm, 4-mm, 5-mm, 6-mm, 7-mm and 8-mm ¯ower buds.(a) Northern blot analysis using total anther RNA, a GUS probe and highstringency conditions. Molecular weight of the TOG mRNA is indicatedon the left.(b) Western blot analysis using the anti-GUS antibody and approximatelyequal quantities of proteins from isolated anthers. The molecular weightsof TOG proteins are indicated on the left.(c) Fluorogenic analysis of GUS enzymatic activity in isolated anthers.GUS activity is expressed in 104 pmol MU min±1 mg protein±1 6 standarderror with n = 3. For controls, see Figure 3.

478 Elizabeth Foster et al.

ã Blackwell Science Ltd, The Plant Journal, (2002), 31, 477±486

mm ¯ower buds. Approximately 9 TOG lines exhibited

GUS activity at various levels while the remaining 11 lines

failed to exhibit detectable GUS activity (data not shown).

GUS expression levels in anthers isolated from 5-mm buds

were also measured in all 9 GUS-positive TOG lines by

¯uorogenic analysis (data not shown). Southern blot

analyses of genomic DNA extracted from the GUS-positive

TOG transgenic plants revealed 1±5 intact copies of the

gene fusion (data not shown). Based on these data, TOG

lines were selected as described below for detailed

analyses.

The TOG fusion is expressed during anther development

Two transgenic TOG lines showing high levels of GUS

activity were selected for further analysis during anther

development. Results obtained with TOG line 5 were

comparable to that of TOG line 22 and are thus presented

only for the latter in subsequent analyses. A Northern blot

hybridized with a GUS probe revealed a transcript of

approximately 3 kb, consistent with the predicted size of

the TOG (transcript) (Figure 2a). The TOG mRNA was

detected ®rst in anthers isolated from approximately 3-mm

¯ower buds and then accumulated rapidly, reaching its

maximum level in anthers from 5-mm buds. Thereafter

steady-state levels of the TOG mRNA decreased somewhat

in anthers from buds 6±7 mm in length, correlating with

the onset of tapetal degeneration. By the late pollen

maturation stage (8-mm buds), the TOG mRNA was no

longer detectable in anthers. Thus, the pattern of TOG

mRNA accumulation in anthers is similar to that previously

described for the BnOlnB;4 gene (Robert et al., 1994). As a

con®rmation, similar Northern blot analysis using a

BnOlnB;4 probe also detected the approximately 3 kb

TOG mRNA, as well as higher levels of the native

BnOlnB;4 mRNA at approximately 1.6 kb (data not shown).

The pattern of TOG protein accumulation differed from

that of the TOG transcript. Western blot analysis with an

anti-GUS primary antibody revealed a protein at approxi-

mately 125 kDa, consistent with the molecular weight

predicted for a full-length fusion protein between

BnOlnB;4 and GUS (Figure 2b). The full-length TOG fusion

protein was ®rst detected in anthers from 4-mm ¯ower

buds prior to tapetal degeneration, peaked in 6-mm buds

and was no longer observed in 8-mm buds. Another

protein was detected at approximately 115 kDa in anthers

from 6- to 8-mm buds. The molecular weight of this

polypeptide is consistent with that predicted for the

mature TOG fusion protein following processing at or

near the beginning of the BnOlnB;4 C-terminal domain

(Murphy and Ross, 1998; Figure 2b). A second protein of

approximately 97 kDa was also detected in the later stages

of anther development (7±8 mm buds). Proteins cross-

reacting with the anti-GUS antibody were not detected in

untransformed B. carinata anthers (see below, Figure 3b).

As a con®rmation, Western blot analysis with an anti-

BnOlnB;4 primary antibody revealed the same proteins as

described for the anti-GUS antibody (data not shown). In

addition to the TOG proteins, lower molecular weight

native tapetal oleosin-like proteins were detected with the

anti-BnOlnB;4 antibody in both transgenic TOG and

untransformed B. carinata plants. Two proteins were

detected of about 54 and 62 kDa in 4±7 mm ¯ower buds

and two proteins of about 38 and 46 kDa persisted during

late anther development (data not shown). The multiple

forms of the native tapetal oleosin-like protein detected in

Figure 3. Localization of RNA, protein expression and GUS activity inanthers, pollen grains and pollen coats from plants transformed with theTOG translational fusion or the transcriptional fusions BnOlnB;4-GUS andSta 44-GUS.(a) Northern blot analysis using the GUS probe at high stringency.Molecular weights of the TOG or GUS mRNAs are indicated on the left.(b) Western blot analysis using the anti-GUS antibody. The molecularweights of the TOG proteins are indicated on the left. Pollen coatsamples (p. coats) were puri®ed from anthers harvested prior to anthesisfrom recently opened ¯owers.(c) Fluorogenic analysis of GUS enzymatic activity is expressed in 104

pmol MU min±1 mg protein±1 6 standard error with n = 3. Samples wereobtained from anthers of 5-mm B. carinata ¯ower buds (TOG transgenicand untransformed control) or 4- to 5-mm B. napus buds (BnOlnB;4-GUSand Sta 44-GUS) and from pollen of 8-mm B. carinata buds (TOGtransgenic and untransformed control) or 6- to 7-mm B. napus buds(BnOlnB;4-GUS and Sta 44-GUS).

Modifying the pollen coat protein composition 479




B. carinata with the anti-BnOlnB;4 antibody were reminis-

cent of the multiple forms which have been identi®ed in

B. napus with an anti-BnOlnB;3 antibody (Murphy and

Ross, 1998). The detection of multiple forms in B. napus

was thought to occur due to the similarity of amino acid

sequence between BnOlnB;3 and BnOlnB;4, as well as the

alternative splicing of tapetal oleosin-like genes (Murphy

and Ross, 1998). We probably detected multiple forms of

the native tapetal oleosin-like proteins in B. carinata for

similar reasons. In B. napus, our anti-BnOlnB;4 antibody

detected the same tapetal oleosin-like proteins (data not

shown) as previously described with the anti-BnOlnB;3

antibody (Ross and Murphy, 1996).

GUS enzymatic activity was also examined throughout

anther development (Figure 2c). GUS activity was initially

detected in anthers from 3-mm ¯ower buds and persisted

throughout anther development. Thus, GUS activity was

detected before, during and after tapetal degeneration

indicating that both the full-length and the processed

forms of the TOG fusion protein are enzymatically active.

The TOG fusion protein is targeted to the pollen

Expression of the TOG gene fusion was then examined in

isolated pollen grains at the late pollen maturation stage

(8-mm ¯ower buds). In Northern analyses using a GUS

probe, all GUS-positive TOG transgenic lines exhibited the

approximately 3 kb TOG mRNA in anthers isolated from 5-

mm buds, but not in pollen isolated from 8-mm buds (TOG

line 22, Figure 3a; remaining 8 GUS-positive TOG lines,

data not shown). However, in Western blot analyses of a

high-expressing TOG line (Figure 3b), the full-length TOG

fusion protein (approximately 125 kDa) was detected with

the anti-GUS antibody in anthers from 5-mm buds prior to

tapetum degeneration, whereas the mature 115 kDa

protein and the 97 kDa protein were detected in pollen

isolated from 8-mm buds. As a con®rmation, the same

TOG proteins were detected using the anti-BnOlnB;4

antibody as were the native tapetal oleosin-like proteins

(data not shown). Furthermore, GUS enzymatic activity

was also localized to pollen grains isolated from 8-mm

buds, in addition to being detected in anthers isolated

from 5-mm buds (Figure 3c). Expression of the TOG

construct was not detected by ¯uorogenic GUS analysis

in the high-expressing TOG lines in leaf, stem, 8 mm

emasculated ¯ower bud and pistil tissues (data not

shown).

To determine whether a translational fusion is necessary

to target proteins from the tapetum to pollen grains,

expression of a GUS transcriptional fusion to the tapetum-

speci®c BnOlnB;4 promoter (BnOlnB;4-GUS) was analysed

in transgenic B. napus. GUS expression directed by the

BnOlnB;4 promoter was previously shown to be speci®c to

the tapetum in anthers isolated from 3- to 5-mm ¯ower

buds of B. napus (Hong et al., 1997b). The approximately

2 kb BnOlnB;4 promoter is the same as that used to direct

expression of the TOG translational fusion.

Like the TOG mRNA, the GUS mRNA (approximately

2 kb) transcriptionally fused to the BnOlnB;4 tapetal

oleosin-like promoter was detected in transgenic

B. napus prior to tapetal degradation in 4±5 mm ¯ower

buds as previously reported (Hong et al., 1997b), and was

undetectable in pollen isolated from 6- to 7-mm buds after

the tapetum had degenerated just prior to ¯oral opening

(Figure 3a). As predicted, Western blot analyses of these

plants with an anti-GUS antibody detected the GUS

protein in anthers from 4- to 5-mm buds (Figure 3b).

However, unlike TOG, the anti-GUS antibody did not

detect cross-reacting proteins in pollen grains isolated

from 6- to 7-mm buds. Accordingly, ¯uorogenic analysis

revealed GUS activity in anthers isolated from 4- to 5-mm

buds but not in pollen isolated from 6- to 7-mm buds

(Figure 3c). These data indicate that unless fused transla-

tionally to the tapetal oleosin-like protein BnOlnB;4, the

GUS enzyme produced in the tapetum does not associate

with pollen.

For further comparison, a GUS transcriptional fusion to

a B. napus polygalacturonase promoter (Sta 44-GUS)

which directs high levels of expression in pollen late in

development (Robert et al., 1993), was also included. As

previously reported (Hong et al., 1997a), we con®rmed that

the GUS mRNA was present in 4±5 mm buds of B. napus

transformed with Sta 44-GUS and was also detected at

high levels in pollen isolated from 6- to 7-mm buds

(Figure 3a). Western blot analysis with the anti-GUS

antibody revealed the GUS protein (Figure 3b) and ¯uoro-

genic analysis showed GUS activity (Figure 3c) in anthers

isolated from 4- to 5-mm buds as well as in isolated mature

pollen from 6- to 7-mm buds. The persistence of GUS

expression after tapetal degeneration in these transgenic

Figure 4. Immunogold localization of tapetal oleosin-like, TOG and GUS proteins during anther development.Anthers from untransformed B. carinata 5-mm (a) and 8-mm (b) ¯ower buds reacted with the anti-BnOlnB;4 antibody in tapetosome lipid bodies andpollen coats, respectively. Anthers from 5-mm (c) and 8-mm (d) buds of transgenic B. carinata line 22 containing the TOG translational fusion reacted withthe anti-GUS antibody in tapetosomes and pollen coats, respectively. Examples of elaioplast (e) and tapetosome (t) lipid bodies and the exine (ex), intine(in) and pollen coat (pc) layers are indicated on panels (a) to (d). The anti-GUS antibody reacted with the tapetum in anthers from the 4-mm bud stage(e) but not with the 7-mm bud stage (f) of transgenic B. napus containing the BnOlnB;4-GUS transcriptional fusion, and within pollen grains at the 5-mm(g) and 7-mm (h) bud stage of transgenic B. napus containing the Sta 44-GUS transcriptional fusion. Scale bars equal 1.6 (a,b) and 1.1 (c-h) mm.



B. napus plants relates to the Sta 44-GUS construct driving

expression within the pollen grain itself.

The TOG fusion protein is localized to the pollen coat

To determine whether the TOG translational fusion protein

was targeted to the pollen coat, pollen coats were puri®ed

by cyclohexane solubilization and Western blot analyses

were performed using the anti-GUS antibody. As pre-

dicted, the mature 115 kDa TOG protein was found in

pollen coats puri®ed from pollen of open ¯owers imme-

diately prior to anther dehiscence (Figure 3b). Two add-

itional proteins of about 85 and 80 kDa also appeared in

mature open ¯ower pollen, however, the 97 kDa band,

noted previously in anthers from about 8-mm buds, was

no longer detectable. Western blot analyses of pollen coats

with the anti-BnOlnB;4 antibody detected, in addition to

native tapetal oleosin-like proteins, the mature 115 kDa

TOG protein, but not the 85 and 80 kDa proteins (data not

shown) suggesting that the 85 and 80 kDa proteins, unlike

the 115 kDa protein, no longer contain the 20 residues

recognized by the anti-BnOlnB;4 antibody. Alexander

staining and microscopic visualization veri®ed that the

pollen remains viable and intact following removal of the

pollen coats suggesting that the cross-reacting proteins do

not originate from within pollen (data not shown).

In contrast, pollen coats puri®ed from open ¯owers of

transgenic B. napus containing the transcriptional fusion

constructs BnOlnB;4-GUS or Sta 44-GUS did not exhibit

detectable proteins cross-reacting with the anti-GUS anti-

body (Figure 3b). These data indicate that the GUS protein

does not relocate to the pollen coat upon tapetal degen-

eration if it is not fused translationally to BnOlnB;4 and that

high levels of GUS expression within the pollen grain itself

do not necessarily result in the appearance of the GUS

protein in puri®ed pollen coats.

Immunogold localization was ®rst used to determine the

subcellular localization of native tapetal oleosin-like

proteins in anthers during development. Native tapetal

oleosin-like proteins were assessed in untransformed

B. carinata with the anti-BnOlnB;4 antibody, which cross-

reacted with the tapetosome lipid bodies of anthers from

5-mm ¯ower buds (Figure 4a). Gold particles were not

detected binding to another type of tapetal lipid body

(elaioplasts), or elsewhere in the anther. In anthers isolated

from 8-mm buds of untransformed B. carinata, where the

tapetum has disappeared and the tapetosomes have

disintegrated, gold particles are localized to the pollen

coat which ®lls the interstices of the exine indicating that

oleosin-like proteins are localized to the pollen coat

(Figure 4b). Similar results were observed with compar-

able developmental stages of B. napus anthers (data not

shown) as previously reported (Murphy and Ross, 1998).

The subcellular localization of the TOG protein during

anther development was also assessed by immunogold

localization in high-expressing TOG transgenic lines. The

anti-GUS antibody cross-reacted with the tapetosomes

within the tapetum of anthers isolated from 5-mm ¯ower

buds, but not elsewhere within the anther (Figure 4c). In

anthers isolated from 8-mm buds, the anti-GUS antibody

speci®cally cross-reacted with the pollen coat (Figure 4d).

These data indicate that the TOG translational fusion is

initially present within the tapetum associated with the

tapetosomes and ultimately becomes localized to the

pollen coat. Pre-immune serum exhibited only a back-

ground signal with anthers or pollen isolated from TOG

transgenic lines (data not shown). In another control, the

anti-GUS antibody did not cross-react with proteins from

untransformed anthers or pollen (data not shown).

For comparison, immunogold localizations were also

performed with transgenic B. napus containing the tape-

tal-expressed BnOlnB;4-GUS or the pollen-expressed Sta

44-GUS transcriptional fusions. In anthers isolated from 4-

mm ¯ower buds of B. napus transformed with BnOlnB;4-

GUS (which correspond to the same developmental stage

as 5-mm ¯ower buds of B. carinata), the anti-GUS anti-

body detected GUS protein dispersed throughout the

tapetum (Figure 4e). The lack of tapetosome localization

and the lower GUS expression level in the BnOlnB;4-GUS

plants as compared to the TOG plants (Figure 3b,c) likely

accounted for the lower number of gold particles observed

in the BnOlnB;4-GUS sections. In 7-mm bud anthers

(which correspond to the same developmental stage as

8-mm buds of B. carinata), the GUS protein was not

detected in the anther, locule or pollen (Figure 4f) consist-

ent with its disappearance after tapetal degradation. In

contrast, in B. napus transformed with Sta 44-GUS,

immunogold localization revealed the GUS protein to be

dispersed within the pollen cytoplasm in both 5-mm

(Figure 4g) and 7-mm bud anthers (Figure 4h). In Sta 44-

GUS transgenic plants, GUS activity had been shown to

increase in pollen during anther development (Hong et al.,

1997a). Accordingly, the number of gold particles associ-

ated with the GUS protein was found to be higher in the

later stage of pollen development. These data con®rm that

the GUS protein requires a translational fusion to

BnOlnB;4 for localization to the tapetosomes and ultim-

ately to the pollen coat.

In agreement with the immunolocalization of the TOG

protein to pollen, GUS histochemical staining was negli-

gible with pollen from 5-mm bud anthers (Figure 5a), but

pronounced with pollen from 8-mm bud anthers

(Figure 5b) in TOG plants. This indicates that GUS

enzymatic activity is localized to the pollen only after the

disappearance of the tapetum late in anther development.

GUS activity also persists following the release of pollen

from the anther, as pollen grains continued to exhibit GUS



histochemical staining for more than 2 months after

collection and storage under ambient conditions (data

not shown).

To indicate whether the GUS activity localized to pollen

of TOG transgenic plants was indeed the result of

sporophytic expression rather than gametophytic expres-

sion, GUS histochemical analysis was performed on TOG

lines containing single copy insertions. In 9 GUS-positive

progeny of each of two self-pollinated T0 plants, GUS

histochemical staining of pollen from 8-mm ¯ower buds

typically revealed about 98AÊ 0.2% GUS positive pollen

grains. In comparison, a mix of stained and unstained

pollen grains could be observed by GUS histochemical

staining of T1 progeny of a self-pollinated B. napus trans-

genic line containing a single copy of the pollen-expressed

Sta 44-GUS construct (Figure 5c). The frequency of GUS

staining of pollen from the TOG plants thus re¯ects the

enzymatic activity transferred to the pollen from the

sporophytic tapetum, rather than from gametophytic

expression. GUS histochemical staining did not occur

with pollen from untransformed plants (data not shown).

Discussion

Here we report the ®rst demonstration that the pollen coat

protein composition can be altered by targeting of an

active enzyme synthesized in the tapetum. Targeting to the

pollen coat was achieved with a translational fusion

between a B. napus tapetal oleosin-like protein gene,

BnOlnB;4, and the uidA gene encoding GUS (TOG) intro-

duced into B. carinata plants. Like the expression pattern

of the native gene, BnOlnB;4, TOG mRNA accumulated in

anthers in the tapetum and then disappeared at the time of

tapetal degeneration. However, the TOG proteins and GUS

enzymatic activity not only accumulated prior to tapetal

degeneration, but persisted after tapetal degradation in the

TOG lines.

TOG proteins are initially detected at full length

(approximately 125 kDa) and then become processed into

the predicted mature form (approximately 115 kDa) by the

time they are localized to the pollen. This cleavage is

consistent with that observed at or near the beginning of

the C-terminal domain of the native BnOlnB;4 and related

BnOlnB;3 proteins in B. napus (Murphy and Ross, 1998) as

well as similar proteins in B. carinata. In addition, another

version of the TOG protein of about 97 kDa appears

transiently late in pollen development. The 97 kDa protein

was likely formed by cleavage within the GUS portion of

the protein as it cross-reacts with the anti-BnOlnB;4

antibody. The 97 kDa protein is probably further cleaved

into one or both of the 85 and 80 kDa proteins associated

with mature pollen, neither of which cross-reacted with the

anti-BnOlnB;4 antibody.

These additional processing events appear to be occur-

ring at precise locations along the TOG peptide. Previous

analyses of the termini of the mature tapetal oleosin-like

protein in pollen coats of B. napus (Murphy and Ross,

1998) does not rule out additional cleavage products that

accumulate at lower levels. Thus, the additional cleavage

of the TOG protein to form the 85 and 80 kDa proteins may

be spurious, or re¯ect the actual processing pattern of

tapetal oleosin-like proteins. Additional processing may

re¯ect increased accessibility of the TOG proteins to

proteolysis resulting from the translational fusion. The

fates of the N-terminal and hydrophobic domains of the

full-length TOG, BnOlnB;4 and other tapetal oleosin-like

proteins are unknown.

The tapetal oleosin-like proteins, which lack a signal

peptide, use a unique targeting pathway to move from the

tapetum into the locule and ultimately to the pollen. The

TOG proteins are localized to tapetosomes within the

tapetum, remain associated with the tapetosomes follow-

ing tapetal degradation and then become localized to the

pollen coat of mature pollen. Presently, the mechanism(s)

by which proteins are targeted from the tapetum to the

pollen after tapetal degeneration are not well understood.

Thus, it is interesting to speculate whether the association

of proteins with lipid bodies, which occurs with tapetal

oleosin-like proteins, protects proteins during tapetal

degeneration and/or is ultimately required for pollen coat

targeting. Interestingly, in transgenic B. napus plants

containing the transcriptional fusion between the

BnOlnB;4 tapetal promoter and GUS, the GUS protein

disappears from anthers after tapetal degradation (results

not shown). This suggests that the persistence of GUS

after tapetal degeneration in the TOG plants may indeed

re¯ect protection of the protein due to its association with

the tapetosomes. In our study, targeting is not affected by

the addition of a lengthy translational fusion at the C-

terminus. However, it is not known what portion of the

tapetal oleosin-like protein is essential for targeting and

Figure 5. GUS histochemical staining of pollen from TOG and Sta 44-GUS transgenic plants.GUS histochemical staining of pollen from 5-mm (a) and 8-mm (b) budsof transgenic B. carinata line 22 containing the TOG translational fusionand from 7-mm (c) buds of T1 progeny of a self-pollinated B. napustransgenic line containing a single copy of the Sta 44-GUS transcriptionalfusion construct. Scale bars equal 20 mm.



whether the location of the translational fusion can be

varied.

Little is presently known about the function of tapetal

oleosin-like proteins. Tapetal oleosin-like proteins may act

as structural components of the tapetosomes (Murphy and

Ross, 1998), to sequester lipids within the tapetum (Wang

et al., 1997), to coalesce the pollen coat during pollen

maturation (Piffanelli and Murphy, 1998) or at many steps

during pollination (Murphy and Ross, 1998; Piffanelli and

Murphy, 1998; Ross and Murphy, 1996; Ruiter et al., 1997).

The only evidence to date of the function of the tapetal

oleosin-like proteins comes from an A. thaliana mutant,

grp17-1, which lacks the most abundant pollen coat

protein, the tapetal oleosin-like protein GRP17 (May®eld

and Preuss, 2000). The grp17-1 mutant exhibits a delayed

onset of pollen hydration because pollen fails to interact

normally with the stigma, delaying subsequent steps in

pollination and resulting in the reduced ability of grp17-1

mutant pollen to compete with wild-type pollen during

pollination.

Translational fusion of the tapetal oleosin-like gene to

GUS produces no obvious deleterious effect on tapetal or

pollen development, pollen maturation, pollination or

fertility. Variable expression levels of the TOG transgene

were observed in different transgenic lines. However, in all

lines the TOG expression level was signi®cantly below that

of native BnOlnB;4 transcripts and proteins. It will be

interesting to evaluate whether increased levels of

proteins targeted to the pollen coat will affect pollen

development or function.

The tapetal oleosin-like proteins have not been

described outside the Brassicaceae despite their predom-

inance in the tapetum and pollen coat of Brassica and

related species. We do not know whether genes encoding

divergent proteins, which possess similar functions to

tapetal oleosin-like genes, exist. We are presently evaluat-

ing whether our approach for modifying the pollen coat

protein composition could also be applied in species other

than Brassica.

Here we have shown that the protein composition of the

pollen coat can be modi®ed by the targeting of a

translational fusion protein from the tapetum. This study

indicates that the tapetal oleosin-like protein BnOlnB;4

provides an effective translational fusion partner to shuttle

proteins from the tapetum to the pollen coat. Moreover,

we have demonstrated that an enzyme targeted to the

pollen coat can remain active. Signi®cantly, the activity of

the GUS enzyme used in this demonstration persisted for

weeks after dehiscence. Thus, our strategy for the modi-

®cation of the pollen coat composition could provide a

novel opportunity to study the function of the pollen coat

in the interactions of pollen grains with stigmas, pollina-

tors and the environment. Furthermore, given the import-

ant role of the pollen coat in the pollen/stigma recognition

process, it may be possible to alter the interaction between

pollen and pistil, which may in turn have an impact on the

development of applications for the control of transgene

¯ow, the production of hybrid seed and the preservation of

germplasm.

Experimental procedures

Plant material and transformation

B. carinata A. Braun (Ethiopian mustard) breeding line C90-1163,obtained from Dr K. Falk, Saskatoon Research Centre, AAFC, wasgrown in growth cabinets or the greenhouse typically at 15°C day/10°C night or 20°C day/15°C night under natural and/or arti®ciallight. Agrobacterium-mediated transformation (strain EHA 105) ofB. carinata was performed essentially as described by Babic et al.,1998).

Plasmid construction

The tapetal oleosin-like protein/GUS translational gene fusion(TOG) was constructed by ligating an adapter of two annealedoligos containing KpnI sites, GATCCTCTAGAGGTACCG andGATCCGGTACCTCTAGAG, into the BamHI site located upstreamof the GUS coding region of pOB4G (Hong et al., 1997b). TheBnOlnB;4 promoter and coding region were then ligated as a unitupstream of the introduced KpnI site in frame with the GUScoding region of pOB4G to create the TOG plasmid as follows.The BnOlnB;4 promoter and coding region fragment was createdby inserting an adapter of two annealed oligos containing KpnIsites, TAGGTACCGAGCTCGGGGGATCC and TAGGATCCCCCGA-GCTCGGTACC, into the NdeI site located immediately 5¢- to theBnOlnB;4 TGA stop codon. A second NdeI site in the BnOlnB;4promoter had been previously removed by restriction digestionand ®lling in with Klenow. The TOG plasmid was sequenced tocon®rm the reading frame was preserved between the BnOlnB;4and GUS coding regions. Construction of the BnOlnB;4-GUS(pOB4G) and the B. napus Sta 44-GUS transcriptional fusions andB. napus transformations were described elsewhere (Hong et al.,1997a; Hong et al., 1997b).

Anther, pollen and pollen coat isolation

Anthers were carefully dissected from buds at different develop-mental stages corresponding to the length of the bud (measuredin mm) from the base to the tip of the closed sepals. Antherdevelopment was determined by staining resin-embedded sec-tions of buds at different lengths with Toluidine Blue. Pollen wasisolated according to a method modi®ed from that described byMurphy and Ross (1998). Brie¯y, anthers were gently squeezed inan Eppendorf tube with a disposable blue pestle (Eppendorf) tosuspend the pollen in extraction buffer (100 mM HEPES pH 7.5,10 mM KCl, 1 mM MgCl2, 1 mM EDTA, 1 mM DTT, 0.4 M sucrose,0.01% Triton X-100). The pollen grains were centrifuged at 1000 gfor 3 min and the pollen pellet was washed in extraction buffer.Pollen coats were puri®ed according to a method modi®ed fromthat described by Murphy and Ross (1998). Brie¯y, suspendedpollen was dried by centrifugation in a glass ®bre-plugged ®lterbasket at 20 000 g for 20 sec and the pollen coats extracted bysimilarly centrifuging cyclohexane through the dried pollen onthe ®lter into a new tube. Cyclohexane was evaporated under a



stream of nitrogen gas leaving the pollen coats as a residue.Pollen viability was determined using a 1 : 1 mixture of twoviability staining solutions containing malachite green, acidfuchsin and orange G (Alexander, 1969, 1980).

Northern analysis

Total RNA was isolated from anthers and pollen grains usingTrizol (Gibco BRL, Burlington, ON, Canada) according to themanufacturer's instructions. Five to 10 mg of total RNA wereelectrophoresed on 1±1.5% (w/v) agarose/formaldehyde gels andtransferred to Hybond-N nylon membranes (AmershamPharmacia Biotech, Baie d'Urfe, QC, Canada). Membranes werehybridized in a modi®ed Church aqueous phosphate buffer(Amersham Pharmacia Biotech) at 65°C with random-primed32P-labelled GUS (2 kb BamHI/SacI fragment of pBI121 (Clontech,Palo Alto, CA, USA)) or BnOlnB;4 (0.2 and 1.1 kb EcoRI fragmentsof the BnOlnB;4 cDNA clone Sta 41-9; Robert et al., 1994) probes.Blots were washed in 23 SSC, 0.1% SDS at 65°C and exposed toX-ray ®lm. Equal loading was assessed by A260 of the sample andby ethidium bromide staining of rRNA bands.

GUS enzymatic assays

GUS ¯uorogenic assays of tissue samples from stem, leaf, pistil,anther and pollen were performed essentially as described byJefferson (1987). Extracts were centrifuged to remove debris andthe supernatant was assayed for GUS activity and proteinconcentration using a modi®ed Bradford assay (Bio-Rad, Laval,QC, Canada). Fluorescence at timed intervals was measured withexcitation at 320±390 nm and emission at 415±650 nm using aHitachi F-2000 Fluorescent Spectrophotometer and the slope wasdetermined. The speci®c activity of the GUS enzyme was calcu-lated as pmol 4-methyl umbelliferone (MU) min±1 mg±1 totalprotein. GUS activity was estimated from the average of threereplicate assays. GUS histochemical staining of pollen wasperformed using a method modi®ed from that of Jefferson(1987) in a solution of 50 mM NaPO4 pH 7.0, 10 mM EDTA,0.5 mM K3[Fe(CN)6], 0.5 mM K4[Fe(CN)6], 0.1% sarcosyl, 0.1% b-mercaptoethanol, 0.1% Triton X-100, 1 mg ml±1 X-gluc (5-bromo-4-chloro-3-indolyl-b-D-glucuronic acid) at 37°C overnight andimaged using a Zeiss Axioplan 2 microscope. For the analysis ofsegregating progeny of single-copy TOG lines, 400±800 pollengrains were counted per GUS-positive plant.

SDS-PAGE and Western blotting

Protein samples were extracted directly in 23 loading buffer andseparated by SDS-PAGE according to the methods of Laemmli(1970). Protein concentrations were determined by a modi®edLowry RCDC Protein assay (Bio-Rad). Proteins were transferred toPVDF membranes (Bio-Rad) and blocked in 3% bovine serumalbumin, 5% skim milk powder in TBS (10 mM Tris pH 8.0,150 mM NaCl). A polyclonal anti-GUS rabbit IgG (MolecularProbes, Engere, OR, USA) was used at a 1 : 4000±5000 dilutionin 0.5% blocking solution (Roche, Laval, QC, Canada). Apolyclonal anti-BnOlnB;4 rabbit IgG was generated using asynthesized 20-mer peptide (LGIPESIKPSNIIPESIKPS; SymGen,San Carlos, CA, USA), corresponding to the ®rst 20 residues of theC-terminal domain of BnOlnB;4, conjugated to Keyhole limpethaemocyanin (KLH). The anti-BnOlnB;4 IgG was used at a 1 : 3000dilution. Proteins were detected using a 1 : 15000 dilution of goat

antirabbit IgG conjugated to horseradish peroxidase (Sigma,Oakville, ON, Canada) using BM chemiluminescence blottingsubstrate (Roche).

Immunogold localization

Anthers were ®xed in 0.8% glutaraldehyde, 4% paraformalde-hyde, 0.1 M NaPO4 buffer pH 7.2. After washing in 0.1 M NaPO4

pH 7.2, tissues were dehydrated in an ethanol series andin®ltrated with LR White acrylic resin (London Resin Co.,London, UK) over several days at 25°C. Following polymerizationof the resin at 50°C overnight, ultra-thin sections (approximately100 nm) were cut on a Reichert ultra-microtome and collected onnickel grids. Sections were incubated in 1% glycine in PBS (0.01 M

NaPO4 pH 7.4, 0.85% NaCl) for 30 min to inactivate residualaldehydes and blocked in 1% ovalbumin in PBS for 10 min.Antibody incubations were carried out with anti-GUS (1 : 1000dilution) or anti-BnOlnB;4 (1 : 100) primary antibodies in 0.01%ovalbumin in PBS followed by re-blocking in 1% ovalbumin inPBS and then with 10±15 nm-diameter gold-conjugated goatantirabbit secondary antibody (EY Laboratories, CA, USA) in0.01% ovalbumin in PBS for 1 h at 25°C. Three 5 minute washes inPBS were performed between each incubation or blocking step.After the procedure, residual salts were removed by washing inwater. As a control, samples were incubated with pre-immunerabbit serum. Samples were observed in a Zeiss EM902A trans-mission electron microscope.

Acknowledgements

We would like to acknowledge Dr Joanne Ross for advice on thepollen coat extraction method, Karri Hume for assistance devel-oping the pollen coat extraction method, Ann-Fook Yang andAmparo Jardine for assistance with the immunogold localiza-tions, Dr Kevin Falk for providing the B. carinata breeding line andDr Leonid Savitch and Dr Ravinder Sardana for reviewing themanuscript. ECORC contribution number 02±25.

References

Alexander, M.P. (1969) Differential staining of aborted andnonaborted pollen. Stain Technol. 44, 117±122.

Alexander, M.P. (1980) A versatile stain for pollen, fungi, yeastand bacteria. Stain Technol. 55, 13±18.

Babic, V., Datla, R.S., Scoles, G.J. and Keller, W.A. (1998)Development of an ef®cient Agrobacterium-mediatedtransformation system for Brassica carinata. Plant CellReports 17, 183±188.

Bih, F.Y., Wu, S.S.H., Ratnayake, C., Walling, L.L., Nothnagel, E.A.and Huang, A.H.C. (1999) The predominant protein on thesurface of maize pollen is an endoxylanase synthesized by atapetum mRNA with a long 5¢ leader. J. Biol. Chem. 274, 22884±22894.

Cabrillac, D., Cock, J.M., Dumas, C. and Gaude, T. (2001) The S-locus receptor kinase is inhibited by thioredoxins and activatedby pollen coat proteins. Nature 410, 220±223.

Dickinson, H. (1994) Simply a social disease? Nature 367, 517±518.Elleman, C.J. and Dickinson, H.G. (1986) Pollen±stigma

interactions in Brassica. IV. Structural re-organization in thepollen grains during hydration. J. Cell. Sci. 80, 141±157.

Elleman, C.J. and Dickinson, H.G. (1990) The role of the exine



coating in pollen±stigma interactions in Brassica oleracea L.New Phytol. 114, 511±518.

Fiebig, A., May®eld, J.A., Miley, N.L., Chau, S., Fischer, R.L. andPreuss, D. (2000) Alterations in CER6, a gene identical to CUT1,differentially affects long-chain lipid content on the surface ofpollen and stems. Plant Cell 12, 2001±2008.

Heslop-Harrison, J., Knox, R.B. and Heslop-Harrison, Y. (1974)Pollen-wall proteins: exine-held fractions associated with theincompatibility response in cruciferae. Theoret. Appl. Genet. 44,133±137.

Hong, H.P., Gerster, J.L., Datla, R.S.S., Albani, D., Scoles, G.,Keller, W. and Robert, L.S. (1997a) The promoter of a Brassicanapus polygalacturonase gene directs pollen expression ofb-glucuronidase in transgenic Brassica plants. Plant CellReports 16, 373±378.

Hong, H.P., Ross, J.H.E., Gerster, J.L., Rigas, S., Datla, R.S.S.,Hatzopoulos, P., Scoles, G., Keller, W., Murphy, D.J. andRobert, L.S. (1997b) Promoter sequences from two differentBrassica napus tapetal oleosin-like genes direct tapetalexpression of b-glucuronidase in transgenic Brassica plants.Plant Mol. Biol. 34, 549±555.

HuÈ lskamp, M., Kopczak, S.D., Horejsi, T.F., Kihl, B.K. and Pruitt,R.E. (1995) Identi®cation of genes required for pollen-stigmarecognition in Arabidopsis thaliana. Plant J. 8, 703±714.

Jefferson, R.A. (1987) Assaying chimeric genes in plants: the GUSgene fusion system. Plant Mol. Biol. Reporter 5, 387±405.

Laemmli, U.K. (1970) Cleavage of structural proteins during theassembly of the head of bacteriophage T4. Nature 227, 680±685.

May®eld, J.A. and Preuss, D. (2000) Rapid initiation ofArabidopsis pollination requires the oleosin-domain proteinGRP17. Nature Cell Biol. 2, 128±130.

Murphy, D.J. and Ross, J.H.E. (1998) Biosynthesis, targeting andprocessing of oleosin-like proteins, which are major pollen coatcomponents in Brassica napus. Plant J. 13, 1±16.

Piffanelli, P. and Murphy, D.J. (1998) Novel organelles and

targeting mechanisms in the anther tapetum. Trends PlantSci. 3, 250±253.

Preuss, D., Lemieux, B., Yen, G. and Davis, R.W. (1993) Aconditional sterile mutation eliminates surface componentsfrom Arabidopsis pollen and disrupts cell signaling duringfertilization. Genes Devel. 7, 974±985.

Robert, L.S., Allard, S., Gerster, J.L., Cass, L. and Simmonds, J.(1993) Isolation and characterization of a polygalacturonasegene highly expressed in Brassica napus pollen. Plant Mol. Biol.23, 1273±1278.

Robert, L.S., Gerster, J., Allard, S., Cass, L. and Simmonds, J.(1994) Molecular characterization of two Brassica napus genesrelated to oleosins which are highly expressed in the tapetum.Plant J. 6, 927±933.

Ross, J.H.E. and Murphy, D.J. (1996) Characterization of anther-expressed genes encoding a major class of extracellularoleosin-like proteins in the pollen coat of Brassicaceae. PlantJ. 9, 625±637.

Ruiter, R.K., Van Eldik, G.J., Van Herpen, R.M.A., Wullems, G.J.and Schrauwen, J.A.M. (1997) Characterization of oleosins inthe pollen coat of Brassica oleracea. Plant Cell 9, 1621±1631.

Scott, R., Dagless, E., Hodge, R., Paul, W., Sou¯eri, I. andDraper, J. (1991) Patterns of gene expression in developinganthers of Brassica napus. Plant Mol. Biol. 17, 195±207.

Shiba, H., Takayama, S., Iwano, M., Shimosato, H., Funato, M.,Nakagawa, T., Che, F.-S., Suzuki, G., Watanabe, M., Hinata, K.and Isogai, A. (2001) A pollen coat protein, SP11/scr,determines the pollen S-speci®city in the self-incompatibilityof Brassica species. Plant Physiol. 125, 2095±2103.

Wang, T.W., Balsamo, R.A., Ratnayake, C., Platt, K.A., Ting, J.T.L.and Huang, A.H.C. (1997) Identi®cation, subcellular localization,and developmental studies of oleosins in the anther of Brassicanapus. Plant J. 11, 475±487.

Wolters-Arts, M., Lush, W.M. and Mariani, C. (1998) Lipids arerequired for directional pollen-tube growth. Nature 392, 818±821.



Documents

Modifying the pollen coat protein composition in Brassica