Department of Botany, The University of Tennessee, Knoxville,
Tennessee 37996-1100
The Arabidopsis COP1 protein functions as a developmental
regulator, in part by repressing photomorphogenesis in darkness. Using
complementation of a cop1 loss-of-function allele with
transgenes expressing fusions of cop1 mutant proteins
and 
-glucuronidase, it was confirmed that COP1 consists of two
modules, an amino terminal module conferring a basal function during
development and a carboxyl terminal module conferring repression of
photomorphogenesis. The amino-terminal zinc-binding domain of COP1 was
indispensable for COP1 function. In contrast, the debilitating effects
of site-directed mutations in the single nuclear localization signal of
COP1 were partially compensated by high-level transgene expression. The carboxyl-terminal module of COP1, though unable to substantially ameliorate a cop1 loss-of-function allele on its own,
was sufficient for conferring a light-quality-dependent hyperetiolation
phenotype in the presence of wild-type COP1. Moreover, partial COP1
activity could be reconstituted in vivo from two non-covalently linked, complementary polypeptides that represent the two functional modules of
COP1. Evidence is presented for efficient association of the two
sub-fragments of the split COP1 protein in Arabidopsis and in a yeast
two-hybrid assay.
 |
INTRODUCTION |
The regulation of plant development
by light is mediated by positive and negative signaling pathways that
transduce environmental light signals from phytochromes, cryptochromes,
and UV light sensors to the level of gene expression (Khurana et al.,
1998
). The Arabidopsis constitutive photomorphogenesis 1 (COP1) protein
functions as a repressor of photomorphogenesis during seedling
germination in darkness by suppressing the expression of light
inducible transcripts in the nucleus. COP1 operates in collaboration
with other developmental regulators of the COP/DET/FUS group of
proteins, where DET and FUS stand for de-etiolated and fusca,
respectively (Deng et al., 1991; Fankhauser and Chory, 1997).
Regulation of COP1 activity, which is under negative control by light,
appears to involve a light-dependent and tissue-specific relocalization
of the COP1 protein across the nuclear envelope, with hypocotyl cell
nuclei containing high levels of COP1 in darkness and reduced levels in
constant light (von Arnim and Deng, 1994; von Arnim et al., 1997;
Osterlund and Deng, 1998). The -glucuronidase- (GUS) COP1 fusion
transgene used to demonstrate the light modulated subcellular
localization of COP1 is fully functional in complementing a
cop1 loss-of-function allele, cop1-5, suggesting
that the modulation of subcellular localization is functionally
significant (von Arnim et al., 1997).
The COP1 protein possesses four recognizable structural domains;
beginning at the amino terminus, these are a zinc-binding Ring-finger
motif, a potential coiled-coil domain (Helix), a central core domain,
and a domain of WD-40 repeats (Deng et al., 1992). Nuclear import of
COP1 is mediated by a single bipartite nuclear localization signal
(NLS) located in the core domain. Nuclear exclusion of COP1 requires a
cytoplasmic localization signal (CLS), which overlaps the Helix domain
(Stacey et al., 1999).
The COP1 protein may consist of two functional modules. Mild
cop1 mutant alleles are defective in the repression of
photomorphogenesis only, but remain viable and fertile. In contrast,
severe cop1 alleles are seedling lethal, accumulate large
amounts of anthocyanins during late embryogenesis, and phenotypically
resemble loss-of-function alleles of other genes in the
COP/DET/FUS group. Molecular analysis of an allelic series
of cop1 mutants revealed that a mild allele of
COP1, cop1-4, contains a stop codon at amino acid
position 283. The mutation results in a truncated product (COP1-4 or
COP1[1-282]) and, unless readthrough occurs, it will prevent
synthesis of the COP1 carboxyl terminal domains. In a converse manner,
overexpression of COP1-4 results in a mild dominant-negative phenotype
(McNellis et al., 1996). Therefore, COP1-4 appears to represent a
partly autonomous, amino terminal module (COP1N), which confers a basic function in plant growth and development (McNellis et al., 1994a). Hence the carboxyl terminal module of COP1 consisting of the core domain and the WD-40 repeats (COP1C) may extend the function of the
COP1N module by enabling the repression of
photomorphogenesis. This notion has recently received experimental
support because overexpression of COP1(293-675) fused to GUS caused
hypocotyl elongation in a wild-type background (Stacey et al., 1999).
In addition, truncation of the WD-40 domain abolished the dominant positive phenotype caused by overexpression of full-length COP1 (McNellis et al., 1994b; Torii et al., 1998).
To further delineate the relative functional contributions of the
individual COP1 domains, including its NLS, we have studied the
complementation and light-dependent overexpression phenotypes of a
series of recombinant COP1 alleles with known subcellular targeting
properties. Among other data we established that
co-expression of the COP1N and COP1C modules from separate polypeptides
in Arabidopsis reconstituted almost the entire wild-type COP1 activity.
COP1N and COP1C interacted during Arabidopsis immunoprecipitations and in the yeast two-hybrid system, suggesting stable binding between the
two non-covalently linked COP1 fragments.
| |
RESULTS |
Functional Analysis of COP1 Domains
Previous experiments had shown that full-length COP1 and specific
deletion mutants were able to confer a long hypocotyl phenotype during
germination of transgenic seedlings under white light conditions (Stacey et al., 1999). To examine whether any of the deletion mutants
retained COP1 activity, we tested the capacity of the 35S:GUS-COP1
mutant overexpression constructs to modify the phenotype of the lethal
allele cop1-5. Based on the severity of the cop1-5 phenotype
and the absence of detectable COP1 gene product, cop1-5 is
thought to represent the cop1 null phenotype (McNellis et
al., 1994a). Tables I and
II summarize the phenotypes of some of
the mutant alleles and transgenes used in this work. Figure
1 summarizes the data and Figure
2 shows representative examples of weak
and partial complementation phenotypes after germination in light or
darkness. Non-transgenic, wild-type Arabidopsis, as well as the
cop1-5 mutant are shown for comparison. The
cop1-5 mutants expressing intact GUS-COP1 look identical to
wild type (Fig. 3; von Arnim et al.,
1997).
View larger version (22K):
[in this window]
[in a new window]
|
Figure 1.
Structure of COP1 mutants. The
capacity of the respective mutants to complement the cop1-5
mutation is indicated on the right. All COP1 mutants were expressed as
GUS-fusions. Yes, Full complementation, i.e. a wild-type phenotype;
Partial, plants that retained a mild cop1 phenotype, but produced seeds
(e.g. Fig. 2, COP1[1-282]); Weak, a slight amelioration of the
cop1-5 seedling phenotype (Fig. 2, COP1[293-675]), which
was insufficient to overcome lethality of cop1-5. The
structural domains of COP1 are indicated by patterned boxes. CLS and
NLS denote the cytoplasmic and nuclear localizaton signal of COP1,
respectively.
|
|
View larger version (79K):
[in this window]
[in a new window]
|
Figure 2.
Complementation phenotypes of cop1-5
homozygous plants expressing COP1 mutant proteins. A cop1-5
mutant seedling and COP1 wild-type plants (WT) are shown for
comparison. The cop1-5 mutant looks identical in light and
darkness. All COP1 mutant proteins were expressed as GUS fusions.
Seedlings were germinated on agar medium for 5 d in darkness (top
row) or for 12 d in the light (bottom row). Bars = 1 mm.
|
|
View larger version (71K):
[in this window]
[in a new window]
|
Figure 3.
Complementation phenotypes of cop1-5
homozygous plants expressing the indicated NLS mutants of COP1 or
wild-type COP1 (wt). All proteins were expressed as GUS fusions. The
relative transgene expression levels are indicated by the GUS
activities (picomoles of methylumbelliferone per minute per microgram
protein) given below the panels. Seedlings were germinated on agar
medium for 5 d in darkness (top row) or for 12 d in the light
(bottom row). Bars = 1 mm. Note that the unmutated GUS-COP1
transgene complements the cop1-5 allele to the level of
wild-type Arabidopsis.
|
|
None of the deletion constructs tested fully complemented the
cop1-5 allele, indicating that all the deleted domains
tested here are indispensable for full COP1 activity. To address the extent of overexpression driven by the 35S promoter in our system, we
compared the signals for wild-type COP1 and a dilution series of
35S:GUS-COP1 on a western blot that was probed with a polyclonal COP1
antiserum (Fig. 4A). The levels of
GUS-COP1 and GUS-COP1(293-675) were approximately 10-fold higher than
that of endogenous COP1. Similar expression levels were found for other
GUS-COP1 deletion mutants, including GUS-COP1(105-675) (not shown).
Therefore, overexpression of the COP1 deletion fragments tested was
insufficient to overcome the mutational defect.
View larger version (79K):
[in this window]
[in a new window]
|
Figure 4.
Western-blot data and co-immunoprecipitation
between COP1-4 and GUS-COP1(293-675) (GUS-COP1C). A, Equal amounts of
total protein from wild-type non-transgenic seedlings ( ) or from
seedlings expressing GUS-COP1C or GUS-COP1 were separated by SDS-PAGE,
blotted, and probed with a polyclonal COP1 antiserum (pc), a COP1
monoclonal antibody (mc), or a GUS antibody (GUS). The monoclonal
antibody recognizes an N-terminal epitope absent from GUS-COP1C.
Migration positions of the COP1 proteins are indicated. Less than
full-length bands are due to protein degradation. For the GUS-COP1
sample, a 1:10 dilution was also loaded to compare the expression
levels of GUS-COP1 and wild-type COP1. The GUS-COP1 extracts were from
cop1-5 mutant seedlings and therefore lack the wild-type
COP1 signal. The affinity of the monoclonal antibody is insufficient
for western-blot detection of wild-type COP1. B, Immunoprecipitates
from cop1-4 mutant or COP1 wild-type seedlings
transgenic for GUS-COP1C were prepared using the polyclonal serum (pc)
or the monoclonal antibody (mc), which recognizes COP1-4, but not
COP1C. Antibody was omitted from controls (). The immunoprecipitates
were separated, blotted, and probed with an anti-GUS antiserum. Note
that GUS-COP1C was coprecipitated with each anti-COP1 antibody from
cop1-4 mutant extracts. One control immunoprecipitation was
conducted with an anti-GUS antiserum (GUS).
|
|
Among the deletion mutants, GUS-COP1-4 resulted in the most striking
complementation, a phenotype reminiscent of yet more severe than that
of the cop1-4 allele regarding hypocotyl elongation, cotyledon expansion, and anthocyanin accumulation. The original cop1-4 allele carries a stop codon at position 283 of
COP1. The GUS-COP1-4 transgene allowed survival and seed set
of soil grown cop1-5 mutants. This result demonstrates
conclusively that the COP1(1-282) fragment can function independently
of the WD-40 domain to confer a basal level of COP1 activity, even when
fused to GUS.
In contrast, deletion of the Ring-finger domain from COP1 resulted in a
more severe defect than deletion of the WD-40 and core domains. In
detail, GUS-COP1(105-675) retained only weak COP1 activity compared
with the cop1-5 allele, yielding a slightly more elongated
hypocotyl in darkness. In addition, partial expansion of the cotyledons
took place under light conditions (Fig. 2). A weakly ameliorated
phenotype was also seen with GUS-COP1(293-675) (Fig. 2) as well
as with GUS-COP1(1-392) and GUS-COP1-9 (Fig. 1). In contrast,
GUS-COP1(393-675) expressing a cytoplasmic WD-40 domain did not modify
the cop1-5 phenotype at all (Fig. 1). The lack of function
of GUS-COP1(1-392) is of interest in comparison with the substantial
activity conferred by GUS-COP1(1-282) and suggests that the core domain
of COP1 can interfere with the basal activity of the N-terminal domain.
Functional Analysis of an NLS
We then asked to what extent defects in the COP1 NLS would
compromise GUS-COP1 activity at the phenotypic level. For each of two
site-directed mutants in the NLS, which had shown reduced nuclear
accumulation of GUS activity (Stacey et al., 1999), two lines were
analyzed for the degree of complementation of the cop1-5 allele, and the level of transgene expression was determined by GUS
activity assay (Figs. 1 and 3). The mutants COP1 mut1 and COP1 mut2 and
even the double mutant COP1 mut1 mut2 all suppressed the seedling
lethality of cop1-5 when expressed at levels above 4.0 pmol
methylumbelliferone min
1
µg1 protein, and fertile plants homozygous
for the cop1-5 allele were recovered for all three
constructs. This expression level corresponds to the level shown in the
immunoblot of Figure 4A. However, the three mutants differed in the
extent of complementation. In darkness, COP1 mut1 showed the most
complete complementation at the seedling stage, followed by COP1 mut2
and COP1 mut1 mut2. As adults the COP1 mut2-complemented plants were
essentially wild type, whereas COP1 mut1 was slightly dwarfed and COP1
mut1 mut2 was severely dwarfed (average heights at maturity were 20 cm
for COP1 mut2 and wild type, 12 cm for COP1 mut1, and 4.5 cm for COP1 mut1 mut2).
Lines with reduced expression levels of the NLS mutants COP1 mut2 and
COP1 mut1 mut2 showed more severe cop1-like phenotypes after
germination in darkness. When grown in soil, no viable plants were
recovered for a COP1 mut1 mut2 line characterized by a GUS expression
level of 0.9 pmol methylumbelliferone min1
µg1 protein (Fig. 3). For COP1 mut1, all
lines analyzed had high expression levels. Taken together, these
results confirm that COP1 functions in the nucleus to repress
photomorphogenesis, and they demonstrate that the NLS is essential.
However, a defect in NLS activity can be partially compensated by a
high gene expression level.
It seemed plausible that Ser phosphorylation at sites adjacent to the
COP1 NLS, located between residues 293 and 314, might play a role in
NLS activity and consequently, in the activity of the overexpressed
COP1 (Moll et al., 1991; Jans et al., 1995; Jensen et al., 1998). Among
those Ser residues that are conserved in COP1 from Arabidopsis (GenBank
accession no. P43254), tomato (accession no. L24437), and pea
(accession no. P93471), S280 and S288 are closest to the NLS. To
eliminate the possibility of phosphorylation at these sites, the Ser
residues were changed to Ala. The site-directed mutants S280A and S288A
were fully capable of complementing the cop1-5 allele (Fig.
1) and caused hypocotyl elongation similar to intact GUS-COP1 in a
wild-type background (not shown). Therefore, we do not favor a model in
which COP1 NLS activity is controlled through phosphorylation at
adjacent sites. It obviously remains possible that phosphorylation at
non-conserved residues or at more remote residues modulates COP1 NLS activity.
Fragment Complementation between COP1N and COP1C Modules of
COP1
COP1-4 alone provides a basal function of COP1, which
allows the plant to reach the flowering stage, but renders it
constitutively photomorphogenic. COP1(293-675), on the other hand,
promotes hypocotyl elongation in a wild-type background and thus acts
as a repressor of photomorphogenesis (Stacey et al., 1999). We asked
whether co-expression of COP1-4, encoding COP1(1-282), and
GUS-COP1(293-675) as separate polypeptides would reconstitute full COP1
activity in transgenic Arabidopsis. When germinating in
darkness, cop1-4 mutant seedlings expressing the
GUS-COP1(293-675) protein showed wild-type hypocotyl length (Fig.
5, top), and approximately one-half of
the seedlings had closed cotyledons, whereas the remainder lacked an
apical hook and had slightly open cotyledons. A non-transgenic cop1-4 seedling is shown for comparison. Furthermore,
GUS-COP1(293-675) transgenic cop1-4 plants formed a larger
rosette, were taller, and produced more seeds than cop1-4
controls (Fig. 5, D-F).
View larger version (80K):
[in this window]
[in a new window]
|
Figure 5.
Complementation of the cop1-4 allele by
GUS-COP1(293-675). Transgenic seedlings or non-transgenic controls were
germinated in darkness for 5 d (top) or grown under light
conditions for 5 weeks (bottom). The two transgenic seedlings shown in
B and C represent the range of variation seen among complemented
seedlings. A and D, cop 1-4; B, C, and E, cop
1-4, GUS-COP1(293-675); F, wild type.
|
|
However, despite the striking complementation seen at the seedling
stage, not all cop1-4 mutant plants carrying a
GUS-COP1(293-675) transgene were complemented at the rosette stage.
When rosette leaves were stained for GUS activity, we noted that a
complemented phenotype correlated with a high transgene expression
level, whereas a cop1-4 like phenotype, i.e. small, rounded
rosette leaves, correlated with silenced GUS expression (Fig.
6, top row).
View larger version (79K):
[in this window]
[in a new window]
|
Figure 6.
Complementation of cop1-4, but
not of cop1-6, by GUS-COP1(293-675). Rosette leaves were
stained for GUS activity (top). Complementation of the
cop1-4 mutant was abolished by silencing of the transgene.
Bottom, Unstained plants are shown to demonstrate the lack of
complementation of the cop1-6 phenotype.
|
|
A second mild cop1 allele, cop1-6, which contains
a five-residue insertion within the NLS (McNellis et al., 1994a), was
not complemented by GUS-COP1(293-675), even when the transgene
expression level was high (Fig. 6). Therefore, the interaction between
the COP1 amino terminus and GUS-COP1(293-675) was not only
transgene expression dependent, but also allele specific. In summary
our results clearly demonstrate that a substantial fraction of COP1 activity can be recovered by expressing the two functional modules of
COP1 as two separate polypeptides. Henceforth the two modules are
referred to as COP1N and COP1C.
The hyperetiolation phenotype conferred by overexpression of
GUS-COP1(293-675) in the wild-type COP1 background, which can be seen
under white light and far-red light (Stacey et al., 1999), was partly
suppressed in the cop1-4 mutant (Fig.
7). Under blue light, red light, or
darkness, the GUS-COP1(293-675) transgene did not cause substantial
hypocotyl elongation in wild-type COP1 plants. Therefore,
the requirement of wild-type COP1 under these conditions could not be
tested. However, even under red light and blue light the amelioration
of the cop1-4 phenotype by GUS-COP1(293-675) can be seen clearly (Fig.
7). The dependence of the hypocotyl elongation under white light and
far-red on the presence of wild-type COP1 suggests that the hypocotyl
elongation caused by COP1C overexpression may be mediated by the
wild-type COP1 protein.
View larger version (22K):
[in this window]
[in a new window]
|
Figure 7.
Hypocotyl lengths of
cop1-4 mutant or wild-type COP1 seedlings with or
without the GUS-COP1(293-675) transgene after germination for 5 d
under the indicated constant light conditions. Error bars represent
standard deviations.
|
|
The COP1N and COP1C Modules Interact
The allele-specific complementation between COP1N and COP1C
suggested that the two modules of COP1 interact physically in the cell.
In this case one might find evidence for a stable complex between the
two proteins. In an alternate manner, the two modules might carry out
their respective cellular functions independently, but in this case one
would not expect to detect a physical association between the two
proteins. We tested COP1N and COP1C for a physical interaction in three assays, by co-immunoprecipitation, by
colocalization, and in the yeast two-hybrid assay.
Immunoprecipitates were prepared from light-grown cop1-4
mutant seedlings transgenic for GUS-COP1(293-675). The
antibodies used were a polyclonal antiserum raised against the
COP1(1-287) protein or a monoclonal antibody that recognizes an
N-terminal epitope within COP1. As expected, on western blots, neither
antiserum recognized the GUS-COP1(293-675) protein directly (Fig. 4A).
The anti-COP1N immunoprecipitates were gel
fractionated, immunoblotted, and probed with an anti-GUS antiserum to
reveal any coprecipitated GUS-COP1(293-675) protein (Fig. 4B).
GUS-COP1(293-675) was co-immunoprecipitated from light-grown
cop1-4 seedling extracts in an antibody-dependent fashion,
suggesting that COP1N and COP1C associate with
each other in Arabidopsis. Co-immunoprecipitation was also evident in
extracts from dark-grown seedlings (not shown). No signal was detected in anti-COP1 immunoprecipitates from transgenic seedlings harboring the
wild-type COP1 gene, suggesting that any association between wild-type
COP1 and COP1(293-675) may be unstable under our immunoprecipitation conditions. The absence of a stable interaction in the wild type further confirmed that the anti-COP1 antibodies did not recognize the
COP1 carboxyl terminus of GUS-COP1(293-675) directly.
The GUS-COP1(293-675) module (COP1C) carries a strong NLS, whereas
COP1-4 (COP1N) does not (Stacey et al., 1999). We asked whether an
association between the two COP1 modules might be indicated by
increased nuclear uptake of COP1N when co-expressed with COP1C. When a
green fluorescent protein (GFP) fusion of COP1-4 was co-expressed with
GUS-COP1(293-675) in onion epidermal cells, approximately one-half of
all cotransformed cells clearly showed nuclear GFP fluorescence,
specifically in nuclear foci (Fig. 8, top
panels). The nuclear foci are characteristic for nuclear COP1 proteins harboring the subnuclear localization signal located within COP1-4 (Stacey and von Arnim, 1999). When GFP-COP1-4 was expressed alone, a
small number of nuclear foci were detected in only one out of 20 transformed cells, but all other cells lacked nuclear foci (Fig.
8, C and D). This result is consistent with a physical association between GFP-COP1-4 and GUS-COP1(293-675), an association that is
sufficiently stable to promote cotransport into the nucleus of
GFP-COP1-4 as directed by the NLS of GUS-COP1(293-675).
View larger version (121K):
[in this window]
[in a new window]
|
Figure 8.
Cotransformation of onion epidermal cells with
GFP-COP1-4 and GUS-COP1(293-675). The representative nuclei shown
demonstrate that GFP-COP1-4 localizes to nuclear foci when co-expressed
with GUS-COP1(293-675), but not when expressed alone. Left, GFP
fluorescence; right, brightfield images of the same cells. Broken lines
demarcate the nuclei. The cytoplasmic GFP-COP1-4 protein is
concentrated in cytoplasmic inclusion bodies, which are outside the
field of view shown here. A and B, GFP-COP1-4 and GUS-COP1(293-675); C
and D, GFP-COP1-4 alone.
|
|
We also addressed the hypothesis that the interaction between
COP1N and COP1C was mediated by other plant cellular
proteins by testing whether COP1(1-287) and COP1(293-675) associate
with each other in the yeast two-hybrid assay as a heterologous system (Table III). Co-expression of a LexA
fusion of COP1(1-287) and an activation domain fusion of COP1(293-675)
did result in strong induction of reporter gene expression. The
interaction was as strong as the known interaction between two
COP1(1-287) fragments (McNellis et al., 1996), which served as a
positive control. Neither COP1(1-287) nor COP1(293-675) was able to
activate transcription on its own or in combination with control
partners, indicating that the interaction is most likely direct. Taken
together, these data suggest that fragment complementation of COP1
activity is mediated by direct binding of the fragments to each
other.
| |
DISCUSSION |
In this study we have clarified the modular structure-to-function
relationships of the Arabidopsis COP1 protein. Using functional reconstitution in transgenic Arabidopsis we discovered four noteworthy features. First, site-directed mutations of the COP1 nuclear
localization signal, although compromising COP1 activity, were less
detrimental than anticipated based on prior cell biological assays if
expressed at high levels from the cauliflower mosaic virus 35S
promoter. Second, deletion of the COP1 amino terminus including the
Ring-finger domain abolished COP1 function when assayed in a
cop1-5 mutant background, confirming that the Ring-finger
motif is necessary for COP1 activity (Stoop-Meyer et al., 1999). Third,
we demonstrated that COP1 function could be reconstituted in vivo from
a split protein, i.e. from two separately expressed coding regions that represent the N- and C-terminal modules of COP1. Fourth and finally, the hyperetiolation phenotype displayed by light-grown seedlings that
express a COP1 carboxyl-terminal fragment may be mediated by the
wild-type COP1 protein.
The Hyperetiolation Phenotype May Be Mediated by Wild-Type
COP1
Overexpression of the WD-40 domain with its associated NLS in the
core domain confers hyperetiolation in a wild-type background (Stacey
et al., 1999). Given that all deletion mutants, as well as the point
mutant COP1-9, proved to be non-functional when assayed in a
cop1-5 mutant background, it is clear that the
hyperetiolation phenotype does not depend on wild-type activity of the
particular COP1 fragment. In a converse manner, the COP1 NLS mutants
COP1 mut1 and mut2, which retained substantial COP1 activity, failed to
cause hyperetiolation (Stacey et al., 1999). These data are easily
explained if we postulate that the COP1 mutants cause hyperetiolation not by autonomously activating the etiolation pathway, but by interfering with the inactivation of endogenous wild-type COP1. This
hypothesis is consistent with the observation that the overexpression phenotype conferred by GUS-COP1(293-675) under white light and far-red
light was reduced in the cop1-4 mutant (Fig. 7).
The Biological Role of the COP1 NLS
NLSs have been delineated in a substantial number of plant
proteins, often using transient expression assays. Moreover, the effect
of mutations within an NLS on the biological function of the protein
has been tested in the case of viral and bacterial proteins,
specifically for Agrobacterium proteins involved in T-DNA
transfer and tumor formation (Gelvin 1998; Relic et al., 1998).
However, there are few if any instances in which the biological requirement for a NLS has been evaluated critically for a plant-encoded protein at the organism level (Liu et al., 1996). One classical bipartite NLS consisting of two clusters of basic residues had been
defined for COP1 using cell biological assays (Stacey et al., 1999).
Here we showed that at moderate expression levels, site-directed
mutants of the COP1 NLS, which eliminated either one or both basic
clusters, failed to complement the cop1-5 allele to various
degrees (Fig. 3). However, at high expression levels, mutation of
either basic cluster alone still allowed almost complete complementation and even mutation of both basic clusters yielded only
an intermediate cop1-like phenotype in dark-grown seedlings or in soil-grown plants (Fig. 3). From transient expression assays we
estimated that each NLS mutation reduced the nuclear accumulation at
least 10-fold, compared with the wild type (Stacey et al., 1999). It is
clear that the functional complementation assay used here appears to be
more sensitive than previous visual assays. Furthermore, throughout
this work the functional assay revealed a strict correlation between
integrity of the NLS and COP1 activity, indicating that activity of
COP1 depends on its nuclear localization (Fig. 3). Our results provide
experimental support for the notion that even large proteins lacking a
cytologically recognizable NLS and containing a CLS may gain
access to the nucleus under only mild overexpression conditions,
possibly by cotransport with other proteins.
Reconstitution of COP1 Activity from Protein Sub-Fragments
The mild phenotype of the cop1-4 allele, which carries
a stop codon at position 283, suggested that the encoded protein
(COP1N) was capable of performing a basal function of COP1 in
regulating plant growth independently of the COP1C module (McNellis et
al., 1994a). However, it could not be excluded that residual, but
quantitatively insufficient expression of the COP1C module occurred in
cop1-4 mutants, by readthrough of the cop1-4 stop
codon or by spurious translation initiation at Met M292 of COP1. We
showed here that expression of a recombinant GUS-COP1-4 protein rescued
the phenotype of the cop1 null allele cop1-5 to a
level slightly below that of the cop1-4 mutant. This result
argues against readthrough and suggests that
COP1N most likely functions independently of the rest of the protein in providing a basal element of COP1 activity.
The ability of GUS-COP1-4 to complement the cop1-5 allele
was surprising because GUS-COP1-4 protein was thought to be
predominantly cytoplasmic (Stacey et al., 1999). However, a low
efficiency of nuclear uptake may have been compensated by the
approximately 10-fold overexpression obtained for a 35S-driven COP1
transgene product. Consistent with this notion, when the expression
level of GUS-COP1-4 was reduced by gene silencing (Fig. 6), the
transgene failed to complement cop1-5 .
A genetic interaction between the COP1N module and the COP1C module had
been suggested because mutations within the C-terminal portion of the
WD-40 domain abolished the basal activity of the COP1N module,
resulting in a severe, seedling-lethal phenotype (McNellis et al.,
1994a). Moreover, such WD-40 mutations disrupted the targeting of the
COP1 protein to discrete nuclear foci, which is a function of the COP1N
module (Stacey and von Arnim, 1999). Considering the recessive nature
of the WD-40 mutations, they must act in cis, maybe through a
protein-to-protein contact between the COP1N and COP1C modules. These
results prompted us to seek experimental evidence for or against
a direct interaction between the COP1N and COP1C modules. Co-expression
of COP1N and a fragment representing the COP1C module
(GUS-COP1[293-675]) from unlinked genes reconstituted almost full
COP1 activity during seedling development (Fig. 5). This result alone
does not prove that COP1N and COP1C interact physically in vivo because
it remained possible that the two modules carry out their functions
independently in the cell. However, additional data consistent with a
physical interaction between COP1N and COP1C came from experiments
demonstrating co-immunoprecipitation and colocalization, and from the
yeast two-hybrid assay. Therefore, we favor the hypothesis that COP1N and COP1C form a stable complex, possibly a heterodimer, in Arabidopsis.
Even though COP1C interacted with the COP1N module encoded by the
cop1-4 allele, COP1C did not interact efficiently with
wild-type COP1 in immunoprecipitations and COP1C also did not
complement the cop1-6 allele. This result is not surprising
if one considers that COP1N is already covalently linked to a COP1C
module in the wild-type COP1 and cop1-6 alleles.
The native COP1C fragment may compete effectively with the free
GUS-COP1(293-675) protein for the same interface in COP1N. The
biological significance of the interaction between the COP1N and COP1C
modules is not yet clear. We are exploring the hypothesis that the
interaction may provide a means to regulate the activities of the three
known subcellular localization signals in COP1, a nuclear, a
cytoplasmic, and a subnuclear localization signal (Stacey and von
Arnim, 1999; Stacey et al., 1999).
Numerous cellular functions are performed by protein complexes
whose subunits are expressed from separate transcripts. In contrast,
reconstitution of activity of a single polypeptide from nonoverlapping
sub-fragments ("split protein" and "fragment complementation") has been demonstrated only occasionally. In the majority of cases, functional reconstitution was examined biochemically, often with the
goal of delineating protein-folding pathways (Marti, 1998; Goldberg and
Baldwin, 1999), or after transient expression (Schmidt-Rose and
Jentsch, 1997). Functional reconstitution of split protein activity in
vivo has also been accomplished in prokaryotes, at times with high
efficiency (Rubin and Levy, 1990; Shiba and Schimmel, 1992). In
contrast, to our knowledge efficient functional reconstitution by a
split protein in a eukaryotic cell has not been described before. It
was possible that such reconstitutions would generally be inefficient
and would depend on heterologous, covalently fused, dimerizing partner
proteins (Johnsson and Varshavsky, 1994; Remy and Michnik, 1999). In
the case of COP1 no heterologous partners may be involved, because the
COP1N module was untagged. In addition, even though the
GUS-tag, or other cellular proteins, may have stabilized the
dimerization between COP1C and COP1N, it remains likely that the
interaction between COP1N and COP1C was via a common interface.
| |
MATERIALS AND METHODS |
COP1 Mutants
Except for the GUS-COP1S280A and GUS-COP1S288A mutants, the
construction of the COP1 mutants has been described previously (von
Arnim and Deng, 1994; Stacey et al., 1999). All proteins were expressed
as GUS fusions under the control of a double 35S promoter. Mutants are
referred to by their amino acid coordinates; for example, COP1(105-675)
designates a deletion of residues 1 to 104 in the 675-amino acid COP1
protein. Some mutants are referred to by their given allele numbers,
e.g. COP1-4 (McNellis et al., 1994a). The S280A and S288A mutants are
Ser-to-Ala substitutions, which were generated with the pAlter system
(Promega, Madison, WI). COP1 mut1, COP1 mut2, and COP1 mut1 mut2 have
site-directed substitutions within the first, the second, or both basic
clusters of the bipartite NLS of COP1, respectively (Stacey et al.,
1999). The terms COP1N and COP1C refer to the amino-terminal and
carboxyl-terminal modules of COP1, which are represented in this work
by COP1-4 or COP1(1-287) and GUS-COP1(293-675), respectively.
Plant Growth and Transgenic Lines
For complementation, cop1-5 heterozygous plants
(Deng et al., 1992) were transformed via Agrobacterium
tumefaciens-mediated DNA transfer following the vacuum
infiltration procedure (Clough and Bent, 1998). In some cases
transgenic lines created earlier (Stacey et al., 1999) were crossed to
plants that were heterozygous for the loss-of-function allele
cop1-5 or homozygous for the mild allele
cop1-4. Complementation was scored in F2
populations segregating for the respective cop1 allele
and a single transgene locus. Full complementation was confirmed by
recovering viable seeds from a plant homozygous for the
cop1-5 mutation and testing for homozygosity of a
kanamycin resistance marker, which is linked to
cop1-5.
Seedlings were germinated on agar medium containing one-half-strength
Murashige and Skoog salts (Sigma, St. Louis) and 1% (w/v) Suc. For specific light treatments, seedlings were
illuminated with Snap-Lite light emitting diodes (Quantum Devices,
Barneveld, WI) at the following fluence rates: blue light, 10 µmol
m2 s1; far-red light, 15 µmol
m2 s1; and red light, 10 µmol
m2 s1. Digital images were taken on a
stereomicroscope (Olympus SZX12, Olympus, Tokyo) with a camera (Olympus
DP10, Olympus) and assembled into composites with Adobe Photoshop
software (Adobe Systems, Mountain View, CA).
Immunoprecipitation
Protein extracts were prepared from 6-d-old seedlings by
grinding in homogenization buffer {400 mM Suc, 10%
[w/v] glycerol, 10 mM KCl, 50 mM Tris
[tris(hydroxymethyl)-aminomethane], pH 8.0, and 10 mM
EDTA} with 2 µM leupeptin, 10 µg/mL aprotinin, and
0.5 mM phenylmethylsulfonyl fluoride as protease inhibitors
and spinning for 2 min. The supernatant was supplemented with one
volume of NP-40 buffer (150 mM NaCl, 50 mM
Tris, pH 8.0, and 1% [w/v] Igepal [Sigma]). Polyclonal or
monoclonal antibody was subsequently added and incubated at 4°C for
2 h. Antibodies used were 1 µg of an affinity-purified anti-COP1
polyclonal antibody raised against COP1(1-287) (McNellis et al., 1994a)
or 1 µg of IgG from a monoclonal anti-COP1 cell culture supernatant,
which recognizes an N-terminal epitope in COP1 (McNellis et al.,
1994b). Immunocomplexes collected on proteinA-agarose beads (Sigma)
were washed three times with NP-40 buffer and proteins were eluted by
incubation at 95°C with 2× Laemmli SDS-PAGE sample buffer. Proteins
were separated by SDS-PAGE on an 8% (w/v) denaturing polyacrylamide
gel, electroblotted to polyvinylidene difluoride membrane, and probed
with an anti-GUS polyclonal antiserum (Molecular Probes,
Eugene, OR). Blots were developed using an alkaline
phosphatase-linked secondary antibody and chemiluminescence detection
with disodium
3-(4-methoxyspiro[1,2-dioxetane-3,2'-{5'-chloro}tricyclo {3.3.1.13,7}decan]4-yl) (Roche Biochemicals, Basel) as substrate.
Yeast Two-Hybrid Assay and Reporter Assays
The EcoRI-SalI polylinker sequence
of pEG202 (Golemis et al., 1997) was replaced with an alternative
linker formed by annealing the oligonucleotides
5'-AATTCGGATCCATGGCCTAGGTAATTAAG-3' and
5'-TCGACTTAATTACCTAGGCCATGGATCCG-3', thereby forming plasmid
pAVA458 with stop codons in all three frames downstream of the linker
sequence. pAVA459 was created by replacing the polylinker of pJG4-5
(Golemis et al., 1997) with the alternative linker formed by the
oligonucleotides 5'-AATTCAGATCTACCATGGCCTAGGTAATTAAC-3' and
5'-TCGAGTTAATTACCTAGGCCATGGTAGATCTG-3'. COP1 fragments were sub-cloned
to pAVA458 for LexA DNA-binding domain fusions and to pAVA459 for B42
activation domain fusions. Plasmids were cotransformed (Gietz and
Schiestl, 1995) into yeast strain EGY148 (Invitrogen, Carlsbad, CA)
harboring the LexA:lacZ fusion plasmid pSH18-34. pRFHM1, encoding a
LexA-bicoid fusion, and pSH17-4, encoding a LexA-GAL4 fusion served as
negative and positive controls, respectively. -galactosidase assays
were carried out according to published methods (Ausubel et al., 1997).
At least four independent cotransformants were tested at least three
times for each plasmid combination.
GFP-COP1-4 and GUS-COP1(293-675) were co-expressed in onion
epidermal cells using particle bombardment (Stacey et al., 1999). GUS
activity was assayed fluorimetrically according to the method of
Jefferson (1987).
Thanks to Massimo Pigliucci for use of a spectroradiometer
and to Qing Gu for helpful comments on the manuscript.
Received March 14, 2000; accepted June 8, 2000.
TOP
ABSTRACT
INTRODUCTION
