Functional Genomic Analysis of the AUXIN RESPONSE
FACTOR Gene Family Members in Arabidopsis thaliana:
Unique and Overlapping Functions of ARF7 and ARF19
Yoko Okushima,a,1,2Paul J. Overvoorde,a,1,3Kazunari Arima,a,4Jose M. Alonso,b,5April Chan,aCharlie Chang,a
Joseph R. Ecker,bBeth Hughes,aAmy Lui,aDiana Nguyen,aCourtney Onodera,aHong Quach,aAlison Smith,a
Guixia Yu,aand Athanasios Theologisa,6
aPlant Gene Expression Center, Albany, California 94710
bSalk Institute for Biological Studies, La Jolla, California 92037
The AUXIN RESPONSE FACTOR (ARF) gene family products, together with the AUXIN/INDOLE-3-ACETIC ACID proteins,
regulate auxin-mediated transcriptional activation/repression. The biological function(s) of most ARFs is poorly understood.
Here, we report the identification and characterization of T-DNA insertion lines for 18 of the 23 ARF gene family members in
Arabidopsis thaliana. Most of the lines fail to show an obvious growth phenotype except of the previously identified arf2/
hss, arf3/ett, arf5/mp, and arf7/nph4 mutants, suggesting that there are functional redundancies among the ARF proteins.
Subsequently, we generated double mutants. arf7 arf19 has a strong auxin-related phenotype not observed in the arf7 and
arf19 single mutants, including severely impaired lateral root formation and abnormal gravitropism in both hypocotyl and
root. Global gene expression analysis revealed that auxin-induced gene expression is severely impaired in the arf7 single
and arf7 arf19 double mutants. For example, the expression of several genes, such as those encoding members of LATERAL
ORGAN BOUNDARIES domain proteins and AUXIN-REGULATED GENE INVOLVED IN ORGAN SIZE, are disrupted in the
double mutant. The data suggest that the ARF7 and ARF19 proteins play essential roles in auxin-mediated plant
development by regulating both unique and partially overlapping sets of target genes. These observations provide
molecular insight into the unique and overlapping functions of ARF gene family members in Arabidopsis.
The plant hormone auxin, typified by indole-3-acetic acid (IAA),
regulates a variety of physiological processes, including apical
dominance, tropic responses, lateral root formation, vascular
differentiation, embryo patterning, and shoot elongation (Davies,
(Abel and Theologis, 1996). Several classes of early auxin-
responsive genes have been identified including the Aux/IAA,
GH3, and SAUR-like genes (Abel and Theologis, 1996; Guilfoyle
et al., 1998). The GH3-like genes encode acyl adenylate–forming
isozymes (Staswick et al., 2002). Several GH3-like proteins
covalently modify IAA, jasmonic acid, or salicylic acid, indicating
that they play a global role in various hormone signaling path-
ways. The function of the SAUR-like genes is still unknown, but it
has been suggested that they may encode short-lived nuclear
proteins involved in auxin signaling by interacting with calmod-
ulin (Yang and Poovaiah, 2000; Knauss et al., 2003).
The Aux/IAAs have been among the first auxin-regulated
genes to be isolated and are the most characterized among
early auxin-responsive genes. They are encoded by a large gene
family in Arabidopsis thaliana with 29 members (Abel et al., 1995;
Reed, 2001; Liscum and Reed, 2002; Remington et al., 2004).
They encode short-lived nuclear proteins, and most of them
contain four highly conserved domains (I to IV) (Abel et al., 1994;
Reed, 2001). Each domain contributes to the functional proper-
ties of the protein. Domain II confers instability of the protein
(Worley et al., 2000; Ouellet etal., 2001). Domains III and IV serve
for homodimerization and heterodimerization with other Aux/IAA
gene family members as well as for heterodimerization with the
Auxin Response Factors (ARFs) (Kim et al., 1997; Ulmasov et al.,
1997, 1999a, 1999b). Domain I is responsible for the transcrip-
tional repressing activity of the proteins (Tiwari et al., 2004).
The ARF proteins are also encoded by a large gene family in
Arabidopsis (23 members). A typical ARF protein contains a B3-
ARF binds to auxin-responsive cis-acting elements (AuxREs)
found in the promoter region of auxin-responsive genes through
1These authors contributed equally to this work.
2Current address: Nara Institute of Science and Technology, Takayama
8916-5, Ikoma, Nara 630-0101, Japan.
3Current address: Macalester College, St. Paul, MN 55105.
4Current address: Department of Chemistry and BioScience, Faculty of
Science, Kagoshima University, Kagoshima 890-0065, Japan.
5Current address: Department of Genetics, North Carolina State Uni-
versity, Raleigh, NC 27695.
6To whom correspondence should be addressed. E-mail theo@nature.
berkeley.edu; fax 510-559-5678.
The author responsible for distribution of materials integral to the
findings presented in this article in accordance with the policy described
in the Instructions for Authors (www.plantcell.org) is: Athanasios
WOnline version contains Web-only data.
Article, publication date, and citation information can be found at
The Plant Cell, Vol. 17, 444–463, February 2005, www.plantcell.org ª 2005 American Society of Plant Biologists
its DNA binding domain (Abel et al., 1996; Ulmasov et al., 1997,
1999a). The amino acid composition of the middle region
between the DNA binding domain and domains III/IV determines
whether an ARF protein functions as an activator or repressor
(Ulmasov et al., 1999b; Tiwari et al., 2003). The Aux/IAA proteins
regulate auxin-gene expression through interaction with the ARF
proteins. The Aux/IAAs are targets for degradation by the
SCFTIR1complex, and most importantly, auxin mediates their
interaction with the proteolytic machinery (Gray et al., 1999,
2001; Ward and Estelle, 2001; Dharmasiri and Estelle, 2004).
Aux/IAA protein stability is a central regulator in auxin signaling.
Several gain-of-function Aux/IAA mutants, including shy2/iaa3
(Tian and Reed, 1999), axr2/iaa7 (Nagpal et al., 2000), bdl/iaa12
(Hamann et al., 2002), slr/iaa14 (Fukaki et al., 2002), arx3/iaa17
(Rouse et al., 1998), msg2/iaa19 (Tatematsu et al., 2004), and
iaa28-1 (Rogg et al., 2001), have been isolated by forward
genetics. These mutants have amino acid substitutions in highly
conserved residues of domain II, resulting in enhanced protein
in growth and development. Loss-of-function mutations of AUX/
IAAs do not show an obvious visible growth phenotype (Rouse
et al., 1998; Tian and Reed, 1999; Nagpal et al., 2000; P.J.
Overvoorde and Y. Okushima, unpublished data). Loss-of-
function mutants in five ARF genes have been previously
(Sessions et al., 1997; Nemhauser et al., 2000). Loss-of-function
mutations of ARF7/NPH4/MSG1/TIR5 result in impaired hypo-
cotyl response to blue light and other differential growth re-
sponses associated with changes in auxin sensitivity (Watahiki
and Yamamoto, 1997; Stowe-Evans et al., 1998; Harper et al.,
2000). Mutations in ARF5/MP interfere with the formation of
vascular strands and the initiation of the body axis in the early
embryo (Hardtke and Berleth, 1998). Mutations in ARF2/HSS
have been identified as suppressors of the hookless phenotype
(Li et al., 2004). ARF2 acts as a communication link between the
ethylene and the auxin signaling pathways for regulating hypo-
cotyl bending. Lastly, ARF8 functions in hypocotyl elongation,
and it is involved in auxin homeostasis (Tian et al., 2004). The
biological functions, however, of the remaining ARF gene family
members are unknown.
Here, we have employed a functional genomic strategy that
involves the identification of T-DNA insertion in the ARF gene
family members to elucidate some of the biological functions of
the ARF transcription factors. Most of the single arf T-DNA in-
sertion mutants fail to show an obvious growth phenotype. How-
ever, double mutants, such as arf7 arf19, show a strong auxin
phenotype that results in the absence of lateral root formation
than neither the arf7 nor arf19 single mutant expresses. The
results suggest that there are unique and overlapping functions
among related ARF gene family members in Arabidopsis.
The Arabidopsis ARF Gene Family
The Arabidopsis genome contains 23 ARF genes scattered
among the five chromosomes (Arabidopsis Genome Initiative,
2000; annotation version V5.0, Figure 1A). The locations of the
four previously described loss-of-function mutations, arf3/ett
(Sessions and Zambryski, 1995), arf5/mp (Hardtke and Berleth,
1998), arf7/nph4 (Harper et al., 2000), and arf2/hss (Li et al.,
2004), are highlighted in Figure 1A. A cluster of ARF genes,
ARF12, 13, 14, 15, 20, 21, and 22, is present in the upper arm of
chromosome I (Figure 1A). These genes share a high degree of
similarity among theiramino acid andnucleotide sequences (see
Supplemental Figure 1 and Table 1 online). ARF23 is a pseudo-
gene (see Supplemental Figure 1 online; Guilfoyle and Hagen,
2001). Phylogenetic analysis reveals that the genes fall into three
branches (marked with different colors in Figure 1B). Class I has
the most members (15) that can be subdivided into three
subclasses, Ia (five members, shaded brown), Ib (eight mem-
bers, shaded blue), and Ic (two members, shaded green). Their
2001; see Supplemental Figure 1 online), and some of them
Class II (shaded pink) has five members, and some of them
function as activators. Their middle region is rich in Glu (Ulmasov
et al., 1999a; Guilfoyle and Hagen, 2001). Class III (shaded
yellow) also contains three members that are the most divergent
compared with those encoded by the other two classes. ARF3
III and IV (Guilfoyle and Hagen, 2001), may potentially contain
highly divergent domains III and IV (see Supplemental Figure 1
online). Furthermore, ARF13 does not have domains III and IV in
this new alignment (see Supplemental Figure 1 online). The ARF
polypeptides vary in size ranging from;57 (ARF13) to;129 kD
(ARF7) (see Supplemental Table 2 online). This size variation is
primarily attributable to the different amino acid content in the
middle region (see Supplemental Figure 1 online).
RNA hybridization analysis reveals that ARF1-ARF9, ARF11,
ARF16, ARF17, ARF18, and ARF19 are expressed in light-grown
seedlings and various plant tissues, including roots, leaves,
flowers, and stems (Ulmasov et al., 1999a; data not shown). We
were unable to detect expression of the clustered ARF genes in
these various RNAs, and there are no ESTs or cDNAs for these
genes in public databases. Exploratory RT-PCR analysis using
cDNA from various tissues (see Methods) revealed that the
clustered genes are expressed during embryogenesis (see
Supplemental Figure 2B online). Transgenic plants expressing
the b-glucuronidase (GUS) reporter gene from the ARF12 and
ARF22 promoters show that the ProARF12:GUS is expressed only
in the developing seeds, and its expression is detected in the
entire seed, including embryos and the integument surrounding
the embryo (see Supplemental Figure 2F online). ProARF22:GUS
transgenic plants display an identical GUS expression pattern as
the ProARF12:GUS plants (data not shown).
Isolation of ARF T-DNA Insertion Mutants
We initiated this project using a PCR-based screening approach
to identify T-DNA insertion mutants for a large number of ARF
genes. A total of 80,000 T-DNA insertion line populations in the
Columbia ecotype were initially screened, and eight lines were
identified (Alonso et al., 2003). Subsequently, the laboratory
former Torrey Mesa Research Institute, and 10 additional lines
ARFs and Development445
were isolated (Sessions et al., 2002). More recently, we obtained
Taken together during the last 6 years, we identified 27 T-DNA
2 and Supplemental Table 3 online provide a summary of all the
All the lines have been backcrossed at least once and partially
Arabidopsis Biological Resource Center (http://www.biosci.
Phenotypes of Insertion Mutants
We were able to identify T-DNA insertion lines for arf3/ettin, arf5/
mp, arf7/nph4/msg1, and arf2/hss, and their reported pheno-
types were confirmed. Two independent arf3 alleles, arf3-1 and
arf3-2, have unusual gynoecium and floral patterning defects,
including an increased number of sepals and carpals (see Sup-
plemental Figures 3A to 3C online; Sessions et al., 1997). The
arf5-1 mutant fails to form root meristem and normal cotyledons
(see Supplemental Figure 3D online; Hardtke and Berleth, 1998),
and the arf7-1 mutant displays an impaired phototropic re-
sponse toward blue light (Figure 4F; Harper et al., 2000). The
arf2-6, arf2-7, and arf2-8 mutants have a pleiotropic phenotype,
including a long, thick, and wavy inflorescence stem, large
leaves, abnormal flower morphology, and late flowering under
long-day conditions (seeSupplemental Figure 3E online;Li etal.,
2004; Y. Okushima and A. Theologis, unpublished data). It has
been recently reported that arf8 seedlings have long hypocotyls
in various light conditions (Tian et al., 2004). We did not examine
the light-associated phenotype of arf8, but we saw longer in-
florescence stems in the mutant than those in the wild type
growth phenotype (Figure 3).
Because most of the arf T-DNA insertion mutants fail to show
an abnormal growth phenotype (Figure 3), we are generating
double and higher-order mutants among the various insertion
lines. So far, we have generated double mutants among closely
Supplemental Figure 4 online). The phenotype of arf1 arf2 is
similar but much stronger than that of arf2 (see Supplemental
and exhibits severe defects in flower development (see Supple-
mental Figure 4C online). The phenotypic and molecular char-
acterization of arf7 arf19 is presented below.
Isolation and Characterization of arf7 arf19 Double Mutants
and Reed, 2002; Remington et al., 2004). Given the close
Figure 1. The ARF Gene Family of Arabidopsis.
(A) Chromosomal location of ARF genes. The locations of 23 putative ARF genes on the Arabidopsis chromosomes (I to V) are shown according to
version 5.0 of the Arabidopsis Genome annotation submitted to GenBank. Mutants that have been isolated in the ARF gene are shown on the left side of
the chromosomes. The ARF genes clustered on chromosome I are boxed.
(B) Phylogenetic analysis. An unrooted dendogram was generated using ClustalW (Thompson et al., 1994). TreeView was used to generate the
graphical output (Page, 1996). The numbers at the branching points indicate the percentage of times that each branch topology was found during
bootstrap analysis (n ¼ 1000). The gene names, accession numbers, protein identifier, and the accession numbers of the full-length open reading
frames (ORFs) used for this analysis are also shown. Predicted ORFs from the genomic annotation were used for ARF14, ARF15, ARF21, and ARF23
(pseudogene) genes. The full-length ORFs of ARF2, ARF6, ARF7, ARF8, ARF11, ARF12, ARF13, ARF19, ARF20, and ARF22 were constructed during
this study. A differential spliced form of ARF13 has been cloned recently (accession number AY680406).
446 The Plant Cell
Figure 2. Location of T-DNA Insertions in the ARF Gene Family Members.
Boxes represent exons. T-DNA insertions with gray triangles denote lines whose characterization has been completed. T-DNA insertions with white
triangles denote lines not yet characterized.
ARFs and Development447
Figure 3. Phenotype of Mature Mutant Plants.
Three wild-type (left) and three mutant plants (right) are shown. The plants were grown at the same time. White dots indicate the boundariesbetweenthe
wild-type and the mutant plants.
448The Plant Cell
relationship of ARF7 and ARF19, we tested whether the arf19
mutant had an altered phototropic response similar to that
reported for nph4/arf7 (Liscum and Briggs, 1995). We found
type–like manner (Figure 4F). Mature arf7 mutant plants (nph4-1,
arf7-1, and msg1-2/nph4-102) do not show any gross develop-
mental defects, except that they have epinastic rosette leaves
and the length of the inflorescence stems is slightly shorter than
that of the wild-type plants (Figure 3; data not shown; Watahiki
and Yamamoto, 1997). These characteristics are more pro-
nounced in the arf7 arf19 double mutant. The appearance of
mature arf19 plants is identical to that of the wild type (Figures 3
and 4A). The results suggest that the expression of ARF7
functionally compensates for the loss of ARF19 expression
responsible for differential hypocotyl growth, but not vice versa.
We initially used the nph4-1 mutant (Liscum and Briggs, 1995)
as the arf7 allele for crossing into arf19-1 to generate the arf7
arf19 double mutant. Among the F2 population, approximately
one out of 16 plants had short and thin inflorescence stem and
small leaves. PCR analysis confirmed that these small plants
were double homozygous for both mutations. Because the orig-
inal nph4-1 line was screened from fast neutron-mutagenized
seeds carrying the homozygous recessive glabrous1 (gl1) muta-
tion (Liscum and Briggs, 1995), we backcrossed the nph4-1 and
nph4-1 arf19-1 to Columbia (Col) wild-type plants. The nph4-1
and nph4-1 arf19-1 mutant lines without the gl1 mutation were
used for further analysis.
The nph4-1 arf19-1 double mutant exhibits much stronger
auxin-related phenotypes than those of nph4-1 and arf19-1
short inflorescence stems, and their rosette leaves are small and
epinastic (Figures 4A to 4C; see Supplemental Figure 4 online;
data not shown). In addition, nph4-1 arf19-1 has reduced num-
bers of inflorescence stems, suggesting enhanced apical dom-
inance. By contrast, the flowers of nph4-1 arf19-1 appear to be
normal, and they fertilize normally (data not shown). The pheno-
type of nph4-1 arf19-1 is the most obvious at its seedling stage,
with its most prominent phenotype being severely impaired
lateral root formation (Figure 4B, Table 1). The primary roots of
arf19-1 produce as many lateral roots as the wild type, whereas
the arf7 mutant produces fewer lateral roots compared with the
wild type (Figure 4B, Table 1). The primary roots of the nph4-1
arf19-1 seedlings fail to produce lateral roots in 2-week-old
seedlings. However, nph4-1 arf19-1 seedlings start to generate
several lateral roots after ;2 weeks of growth, and their mor-
phological appearance is normal (Figure 4C; data not shown).
The nph4-1 arf19-1 mutant also displays agravitropic responses
in both hypocotyls and roots (Figure 4D). When seedlings are
grown vertically under dark conditions, the hypocotyl growth
orientation of arf7 is significantly skewed compared with the wild
type, whereas the arf19-1 mutant has a normal gravitropic
response (Figure 4D; Harper et al., 2000). Interestingly, in the
nph4-1 arf19-1 seedlings, regulation of growth orientation is dis-
rupted in both hypocotyls and roots, with the hypocotyls occa-
sionally growing downward and the roots upward (Figure 4D).
Also, the roots and hypocotyls of nph4-1 arf19-1 show reduced
gravitropic curvatures compared with the wild type when verti-
cally dark-grown seedlings are reoriented by 908 (data not
shown). The phototropic response toward blue light in hypocot-
yls of nph4-1 arf19-1 seedlings is disrupted as in the arf7 single
mutants (Figure 4F). We generated additional combinations of
arf7 arf19 double mutants using other alleles of arf7 and arf19 to
confirm the phenotypes of nph4-1 arf19-1. We used msg1-2/
nph4-102 (Watahiki and Yamamoto, 1997) and arf7-1 as the arf7
arf19 double mutant alleles, msg1-2 arf19-1, arf7-1 arf19-1,
nph4-1 arf19-2, msg1-2 arf19-2, and arf7-1 arf19-2 (Figures
4A, 4B, and 4D, Table 1; data not shown), display the same
root formation, and agravitropic response. These results confirm
that the phenotypes of nph4-1 arf19-1 are caused by the loss of
ARF7 and ARF19 function.
The phenotypes of the arf7 arf19 mutant are similar to those
reported for the solitary root (slr)/iaa14 mutant (Fukaki et al.,
2002). The slr mutant also shows strong auxin-related pheno-
and hypocotyls, small plant size, and few root hairs (Figures 4A,
seedlings exhibit severely impaired lateral formation, their pri-
mary roots start to produce lateral roots ;2 weeks from
germination (Figure 4C). By contrast, slr-1 seedlings do not
produce any lateral roots even after 4 weeks from germination
(Figure 4C; data not shown). We also examined the effect of
exogenous auxin on lateral root formation in the nph4-1 arf19-1
seedlings. Four-day-old light-grown seedlings of the wild type,
nph4-1 arf19-1, and slr-1 were transferred to medium containing
1 mM IAA. After an additional 3 d of incubation, wild-type
seedlings started to produce many lateral roots, but nph4-1
arf19-1 and slr-1 fail to produce any lateral roots. However, after
5 d of incubation on IAA, several lateral roots are induced in
nph4-1 arf19-1 but not in slr-1 (data not shown). Lower concen-
trations of IAA (1 to 100 nM) fail to induce lateral root formation in
nph4-1 arf19-1 even after 5 d of incubation (data not shown).
These results suggest that the auxin- induced lateral root
formation is inhibited in nph4-1 arf19-1, but is more severely
impaired in slr-1. Also, both slr-1 and arf7 arf19 mutants have
mutants, butslr-1 has smaller rosette leaves and shorter petioles
than arf7 arf19 (Figure 4C). The most striking phenotypic differ-
ence between the arf7 arf19 and slr-1 mutants is the root hair
formation. The slr-1 mutant has very few root hairs (Fukaki et al.,
mutants show normal root hair formation (Figure 4E).
Auxin Sensitivity of arf7 arf19
The arf7 single mutants display reduced auxin sensitivity in
hypocotyl growth, whereas they show normal auxin response in
the roots (Figures 5A and 5B; Watahiki and Yamamoto, 1997;
Stowe-Evans et al., 1998). By contrast, arf19-1 shows normal
auxin sensitivity in the hypocotyls and a mild but significant
resistance to exogenous auxin in the roots (Figures 5A and 5B).
arf19-2 (data not shown), suggesting that the auxin response is
slightly impaired in the roots of the arf19 single mutants.Interest-
ingly, the arf7 arf19 double mutants display severely reduced
ARFs and Development449
Figure 4. Phenotypes of the arf7 arf19 Double Mutant.
(A) Four-week-old soil-grown plants of the wild type, arf19-1, nph4-1, msg1-2, and arf7-1 (top) and the wild type, nph4-1 arf19-1, msg1-2 arf19-1, arf7-1
arf19-1, and slr-1 (bottom).
450 The Plant Cell
auxin sensitivity in both roots and hypocotyls (Figures 5A and
degree as in slr-1. The data suggest that the hypocotyl auxin
sensitivity is impaired in the arf7 single mutants, the root auxin
sensitivity is impaired in the arf19 single mutants, and both are
severely impaired in the arf7 arf19 double mutant. Surprisingly,
the slr-1 hypocotyls fail to elongate after transfer to dark
conditions, and exogenous auxin application does not affect
their hypocotyl growth (Figures 5B and 5C).
Expression Patterns of ARF7 and ARF19
ProARF19:GUS to gain a better understanding of the tissue-
specific expression of ARF7 and ARF19. The expression pat-
terns of ProARF7:GUS and ProARF19:GUS are distinct, with partial
expression is observed in the hypocotyls and petioles of
ProARF7:GUS seedlings (Figure 6A), whereas ProARF19:GUS ex-
pression is restricted to the vascular tissue in the aerial parts
(Figure 6B). In root tissue, unlike the aerial part, ProARF19:GUS is
strongly expressed throughout, including vascular tissue, the
meristematic region, root cap, root hair, and the sites of newly
forming lateral roots (Figures 6B, 6D, and 6J to 6L). By contrast,
ProARF7:GUS expression in the primary root is restricted to the
vascular tissues and is not detected in the meristematic region,
root cap, and root hairs (Figures 6A, 6C, and 6E to 6I). ProARF7:
GUS is expressed in the early stages of lateral root primordia
(Figure 5E). However, after the root primordia emerge from the
parental primary roots, the expression of ProARF7:GUS dissi-
pates from the meristematic region (Figures 6G to 6I). ProARF7:
generated transgenicplantswithProARF7:GUS and
root is elongated (data not shown). The results suggest that both
ARF7 and ARF19 are expressed in sites where lateral roots are
initiated, consistent with the observation of impaired lateral root
formation in the arf7 arf19 double mutants.
Although the loss of ARF19 function does not alter plant de-
velopment, overexpression of ARF19 has a dramatic effect on
plant morphology (Figures 7A to 7D). Overexpression of ARF19
results in alternation of root architecture (Figure 7D). The leaves
of Pro35S:ARF19 plants are narrower, elongated, and misshapen
(Figures 7B and 7C). The Pro35S:ARF19 plants exhibit strong
reduction in apical dominance and have a dwarf phenotype
(Figure 7A). They produce a small number of siliques and have
lower seed production (data not shown). The phenotype of
Pro35S:ARF19 plants is associated with higher levels of the
ARF19 transcript (Figure 7E).
Transcriptional Profiling of the arf7, arf19, and
arf7 arf19 Mutants
The auxin-related phenotypes of arf7, arf19, and arf7 arf19
mutants prompted us to perform detailed microarray analysis
with these mutants using the Affymetrix whole-genome ATH1
GeneChip. We used the nph4-1, arf19-1, and nph4-1 arf19-1
mutants as representatives for each mutant allele during this
experiment. Light-grown seedlings of the wild type, nph4-1,
arf19-1, and nph4-1 arf19-1 were treated for 2 h with the carrier
solvent ethanol (control sample) or 5 mM IAA (auxin-treated
sample). Each experiment was performed in triplicate, and total
RNA was independently isolated to generate biotin-labeled
cRNA for hybridization (see Methods).
Figure 8 shows the scatter plots representing the auxin-
regulated transcriptional profiles of wild-type, arf19-1, nph4-1,
and nph4-1 arf19-1 mutants. A cursory examination of these
scatter plots demonstrates that the loss of ARF7 and ARF19
causes gross changes in auxin-induced gene expression. The
wild-type scatter plot shows that the gene expression profile is
globally altered by exogenous auxin treatment. The scatter plot
of arf19-1 shows a similar degree of distribution as with the wild
type, suggesting that almost normal auxin-regulated gene ex-
pression is maintained in the arf19 single mutant (Figure 8).
However, the scatter plots of nph4-1 and nph4-1 arf19-1 display
a smaller degree of distribution than that of the wild type,
indicating that the auxin-mediated transcriptional regulation is
Table 1. Lateral Root Formation in arf7, arf19, and arf7
MutantNumber of Lateral Roots
Wild type (Col)
7.6 6 3.2
1.3 6 1.1
0.6 6 0.7
1.7 6 1.4
6.8 6 1.6
0.0 6 0.0
0.0 6 0.0
0.0 6 0.0
The number of lateral roots in 10-d-old seedlings was determined. The
numbers represent the average of more than 18 seedlings 6 SD.
Figure 4. (continued).
(B) Seventeen-day-old seedlings of wild type, arf19-1, nph4-1, msg1-2, arf7-1, nph4-1 arf19-1, msg1-2 arf19-1, arf7-1 arf19-1, and slr-1.
(C) Twenty-two-day-old seedlings of the wild type, nph4-1 arf19-1, and slr-1 grown on agar plates vertically.
(D) Gravitropic response of 3-d-old dark-grown seedlings of the wild type, arf19-1, nph4-1, msg1-2, arf7-1, nph4-1 arf19-1, msg1-2 arf19-1, arf7-1
arf19-1, and slr-1.
(E) Root hair formations of the wild type, arf19-1, nph4-1, and nph4-1 arf19-1.
(F) Phototropism of 3-d-old dark-grown seedlings of the wild type, arf19-1, nph4-1, msg1-2, arf7-1, nph4-1 arf19-1, and arf7-1 arf19-1. Seedlings were
exposed to unilateral blue light from the right for 8 h.
ARFs and Development 451
globally repressed in these mutants (Figure 8). We extracted the
auxin-regulated genes using the log2expression values from the
robust multichip analysis (RMA) output file (Irizarry et al., 2003)
and established rigorous statistical criteria based on a variance
measurement to generate auxin-regulated gene lists (see Meth-
ods). Among the 22,800 genes,only 203 met the criteria for more
than twofold auxin induction (I, induced genes), and 68 genes
met the criteria for more than twofold repression (R, repressed
genes). A complete list of all the auxin-regulated genes and how
they are affected by the mutants can be found in the Supple-
mental Tables 4 and 5 online. These gene lists include various
classes of known auxin-regulated genes, such as Aux/IAA, GH3,
SAUR, and ACS, consistent with similar studies reported pre-
viously (Tian et al., 2002; Ullah et al., 2003; Redman et al., 2004).
The genes identified as auxin-regulated (induced or repressed)
were functionally categorized to examine the auxin-regulated
cellular and metabolic processes affected by either or both loss-
of-function mutations of ARF7 and ARF19. Supplemental Figure
of the auxin-regulated genes is currently annotated as encoding
proteins of known or putative function.
only slightly responsive to auxin in the mutants (see Methods).
Among the 203 auxin-induced genes, 105 (51.7%), 14 (6.9%),
and 173 (85.2%) were identified as differentially regulated genes
by nph4-1, arf19-1, and nph4-1 arf19-1, respectively. Likewise,
22 (32.4%), 3 (4.4%), and 44 (64.7%) among the 68 auxin-
repressed genes were identified as differentially regulated genes
by nph4-1, arf19-1, and nph4-1 arf19-1, respectively. This com-
mutants revealed overlapping genes among these gene sets
(Figure 9). For example, among the 203 auxin-induced genes, 96
were similarly affected by the nph4-1 single and nph4-1 arf19-1
double mutants (Figure 9A, class I-D). The class I-D genes are
considered to be preferentially regulated by ARF7. Likewise,
eight auxin-induced genes are similarly affected in the arf19-1
single and nph4-1 arf19-1 double mutants (Figure 9A, class I-F).
These genes are considered to be preferentially regulated by
ARF19. By contrast, 64 auxin-induced genes are differentially
regulated only by the nph4-1 arf19-1 double mutant (Figure 9A,
redundantly regulated by both ARF7 and ARF19. Similar distri-
bution of differentially regulated genes is found among auxin-
repressed genes (Figure 9B, class R-D to R-G). Supplemental
that belong to each class (class I-A to I-H and class R-A to R-H).
Figure 10 shows the expression behavior of some representative
auxin-regulated genes of various functional categories in these
Figure 5. Auxin Sensitivity ofthe Wild Type, arf7, arf19, arf7 arf19, and slr
(A) Inhibition of root growth by exogenous auxin. Each value represents
the average of more than 10 seedlings. Bars represent SE of the average.
(B) and (C) Inhibition of hypocotyl elongation by exogenous auxin. Data
represent the mean of hypocotyl length as a percent of controls (B) or of
actual measurements (C). Bars represent SE of the average. See
Methods for experimental details.
452The Plant Cell
various classes. In addition to classical auxin-regulated genes,
such as IAA5, IAA14, and IAA19, various classes of genes in-
volved in ethylene biosynthesis and perception, phytohormone-
related, and cell wall biosynthesis and development show
defective auxin-regulated gene expression in the mutants,
Figure 7. Developmental Defects by ARF19 Overexpression.
(A) to (C) Growth inhibition in 5-week-old plants (A), first true leaves (B),
and 12-d-old light-grown seedlings (C).
(D) Alteration of root architecture in 10-d-old seedlings.
(E) Expression of ARF19 in overexpressing lines from 7-d-old light-grown
seedlings. ARF gene expression was assessed by RT-PCR as described
ACT8 transcript was used as an internal control. White and black arrow-
heads indicate the size of genomic and cDNA fragments, respectively.
Figure 6. Expression of GUS in ProARF7:GUS and ProARF19:GUS Trans-
(A) GUS expression in a 6-d-old light-grown ProARF7:GUS seedling.
(B) GUS expression in a 6-d-old light-grown ProARF19:GUS seedling.
(C) Root apex of a ProARF7:GUS seedling primary root.
(D) Root apex of a ProARF19:GUS seedling primary root.
(E) to (I) ProARF7:GUS expression in the vascular tissue of mature primary
root, lateral root primordia ([E] and [F], arrowhead), and developing
lateral roots ([G] to [I]).
(J) to (L) ProARF19:GUS expression in entire tissue of primary root and
developing lateral roots.
ARFs and Development 453
especially in nph4-1 arf19-1. A wide range of auxin-regulated
cellular and metabolic processes is affected by the loss of ARF7
and ARF19 gene function.
The transcriptional profile of the untreated control seedlings is
also altered in the nph4-1 arf19-1 double mutant. Comparison of
the transcriptional profiles between the nph4-1 arf19-1 mutant
and the wild type in the absence of auxin treatment reveals that
55 and 45 genes are induced and/or repressed twofold or higher
in nph4-1 arf19-1, respectively (Figure 11; see Supplemental
Tables 6 and 7 online). Interestingly, 20 of the 55 induced genes
in nph4-1 arf19-1 are involved in metabolism (see Supplemental
Table 6 online). Fewer genes have altered gene expression in
untreated nph4-1 and arf19-1 seedlings (Figures 11A and 11B).
Only two genes are repressed in arf19-1, and one of them is
gene expression in untreated seedlings. Figures 11C and 11D
show some representatives of induced or repressed genes in
nph4-1 arf19-1 or both nph4-1 and nph4-1 arf19-1.
The ARF gene family encodes transcriptional regulators that are
involved in auxin signaling. Despite their essential role in auxin-
mediated gene regulation, little is known regarding their biolog-
ical functions, except for very few of them studied by classical
molecular genetic analysis. Questions arise, such as why does
Arabidopsis have so many ARFs? What is the biological function
of each ARF? Which genes do they regulate? To answer these
questions, we have attempted to isolate loss-of-function T-DNA
insertion mutants for all the ARF gene family members using
provide a systematic strategy for analyzing gene function
(Borevitz and Ecker, 2004). We have identified T-DNA insertion
alleles for19outof23ARFgenes,andinitial characterization has
lines isolated. Among the 18 arf single mutants, obvious growth
phenotypes were observed only in the previously identified
Figure 8. Global Gene Expression Profiling.
MA plots (Dudiot et al., 2002) showing changes of auxin-regulated gene expression levels in the wild type, arf19-1, nph4-1, and nph4-1 arf19-1. Each
plot represents the log ratio of the average of the auxin-treated samples (I) to the control samples (C) [M ¼ log2(I/C)] versus overall average intensity
[A ¼ log2p(I*C)]. The genes induced by auxin treatment (M > 1) are highlighted in red, and the genes repressed by auxin treatment (M < ?1) are
highlighted in green. The data were further analyzed for variance to extract statistically valid auxin-regulated genes (see Methods).
454 The Plant Cell
mutants using forward genetics (i.e., arf2/hss, arf3/ett, arf5/mp,
and arf7/nph4). The rest of the arf single mutants fail to show an
obvious growth phenotype. However, in-depth analysis of these
inhibition of root elongation may detect biological phenotypes
associated with these lines. These ARFs may act redundantly in
auxin-mediated gene regulation and provide compensatory
functions during plant development. The expression of at least
two clustered ARF genes in a specific stage of embryogenesis
reinforces the concept of functional redundancy among the ARF
proteins. To query the concept of gene redundancy, we gener-
ated several double mutants among closely related ARF mem-
bers. Their phenotypic analysis indicates that related pairs of
ARFs, namely, ARF1/ARF2, ARF6/ARF8, and ARF7/ARF19, act
redundantly in a distinct developmental manner. During this
study, we focused on the redundant functions of ARF7 and
ARF19 using biological and molecular approaches. A similar
et al., 2004). The in planta interaction between ARF5 and ARF7
suggested by the experiments of Hardtke et al. (2004) raises the
possibility that different combinations of ARF heterodimers may
have various selective functions in regulating targeted gene
expression. Potential heterodimerization between ARF7 and
ARF19 is also suggested by the inhibition of auxin-induced
expession of genes such as At2g23060 (Hookless1-like) and
At4g22620 (AtSAUR-34) by either the arf7 or the arf19 mutant
(see Supplemental Figure 5 online; class I-E). Consequently, the
Figure 9. Comparative Analysis of Genes Differentially Regulated by Auxin in nph4-1, arf19-1, and nph4-1 arf19-1.
Differentially regulated genes in mutants among auxin-induced (A) and repressed (B) genes are shown. Each circle within the Venn diagram indicates
numbers and percentages (in parentheses) of genes with repressed induction or repression levels. Only those genes with greater than twofold fold
change ratio (FCR) in nph4-1, arf19-1, and nph4-1 arf19-1 were analyzed (see Methods). We defined each area of the Venn diagram from A to H, and
each class was further divided into two subgroups based on their auxin-induced expression profiles in the wild type. The genes classified into class D
are considered to be preferentially regulated by ARF7, and those classified into class F are considered to be preferentially regulated by ARF19. The
genes classified into classes E and G are considered to be redundantly regulated by ARF7 and ARF19. The class A genes have similar expression
profiles to class D genes. Likewise, class C genes have similar expression profiles to class F genes. The expression profiles of the representative genes
from each class are shown in Supplemental Figure 5 online.
ARFs and Development 455
Figure 10. The Expression Profiles of Representative Auxin-Regulated Genes in the Wild Type, nph4-1, arf19-1, and nph4-1 arf19-1.
The data represent the average relative intensity expression level of control (open bar) or auxin-treated (blue bar) samples from triplicate experiments.
Bars represent SD of the average. Boxes next to gene names indicate classification color codes according to Figure 9.
456The Plant Cell
T-DNA insertion mutants will be beneficial to understand auxin-
regulated processes mediated by ARF–ARF and ARF–Aux/IAA
interactions. Similar studies using reverse genetics have also re-
vealed unique and overlapping functions among the R2R3-MYB
et al., 1999; Parenicova et al., 2003; Pinyopich et al., 2003).
Unique and Overlapping Developmental Functions of
ARF7 and ARF19
Considering the phenotypes of arf7 and arf19 single mutants,
ARF7 appears to regulate auxin-dependent differential growth in
the hypocotyls, and ARF19 partially mediates auxin signaling in
Figure 11. Effect of the nph4-1, arf19-1, and nph4-1 arf19-1 Mutations on Global Gene Expression in Untreated Control Samples.
(A) Induced genes in the mutants under control conditions.
(B) Repressed genes in the mutants under control conditions. Each circle within the Venn diagram indicates the number of genes with greater than
twofold induction or repression.
(C) Expression profiles of induced classes of genes.
(D) Expression profiles of repressed classes of genes. Data represent the average relative intensity expression levels of control (open bar) or auxin-treated (blue
ARFs and Development457
the roots. The severity of their phenotypes is greatly enhanced in
the double mutant compared with the single mutations, demon-
severely impaired lateral root formation, agravitropic hypocotyls
and roots, and small organs and enhanced apical dominance in
aerial portions. These phenotypes are observed only in the arf7
these developmental events are redundantly regulated by ARF7
and ARF19. Expression of one ARF allows for functional com-
pensation for the loss of the other in arf7 and arf19 single
mutants. This may be because of the high similarity of these two
proteins. The analysis of promoter-GUS transgenic plants dem-
onstrated that there is a significant agreement between the
expression patterns and the developmental defects in the single
and double mutants. ProARF7:GUS is strongly expressed in the
hypocotyls, whereas ProARF19:GUS is strongly expressed in the
roots. Furthermore, expression of ProARF7:GUS is detected
throughout the hypocotyl, whereas the expression of ProARF19:
6A and 6B). However, despite the global ProARF19:GUS expres-
sion and an altered auxin sensitivity in arf19 root, only the arf7
mutants have slightly reduced numbers of lateral roots (Table 1),
suggesting that the ARF7 has a regulatory function in lateral root
initiation. The microarray experiments show that the auxin-
dependent induction of ARF19 is impaired in the nph4-1 mutant
(Figure 10A). Interestingly, the promoter region of ARF19 con-
tains two AuxREs (data not shown), suggesting that ARF7 may
directly modulate the expression of ARF19. This may provide an
alternative explanation for the apparent phenotype of the arf7
mutants. The inadequate auxin-mediated induction of ARF19
expression may have an additive effect on the loss of ARF7
function, yielding an obvious phenotype. We have not tested yet
whether the ARF7 and ARF19 proteins can complement the loss
of each other. Promoter-swapping experiments using trans-
genic arf7 and arf19 single or double mutants harboring ARF7
promoter:ARF19 and ARF19 promoter:ARF7 gene constructs
have the potential to clarify this issue.
ARF7 and ARF19 Regulate Both Unique and Partially
Overlapping Sets of Target Genes
The microarray data provide clear evidence for the unique and
redundant functions of ARF7 and ARF19 on auxin-mediated
gene expression. The almost complete lack of auxin-mediated
transcriptional regulation in the arf7 arf19 mutant is puzzling
(Figure 9). It implies that ARF7 and ARF19 are the only ARF
factors that are necessary and sufficient for auxin signaling in
7-d-old light-grown seedlings. Are the rest of the ARFs dispens-
able? The possibility exists that the majority of auxin-regulated
the ARF7/ARF19 pair. It should be noted that the adult arf7 arf19
plants, although smaller in size, have a normal appearance with
may not be critical for auxin-mediated transcriptional regulation
during the development of aerial organs. Such a proposition is
supported bythephenotypes oftwo otherARFmutants, arf5/mp
and arf3/ett; they control auxin-mediated gene regulation re-
sponsible for axial cell and gynoecium patterning during organo-
genesis, respectively, indicating that ARF5 and ARF3 may also
act in a particular developmental window. In addition, several
single and double arf mutants, including arf2, arf1 arf2, arf3, and
arf6 arf8, have flowers with abnormal morphology and/or poor
fertility, suggesting that these ARFs may act redundantly in
auxin-mediated gene regulation responsible for flower develop-
ment. Comparative microarray analysis with different double
mutants at different developmental stages has the potential to
clarify this view. Alternatively, the remaining ARFs may regulate
genes that are not auxin regulated at that particular develop-
mental stage. The current prevailing view that all ARFs regulate
auxin-mediated gene expression has not been tested experi-
mentally with vigor. Finally, the remaining ARFs may regulate
genes in a cell-specific manner (distinct cell types) that the
microarray analysis fails to detect. This last possibility points to
the necessity of conducting global expression studies in specific
cell types (Birnbaum et al., 2003).
Comparative analysis of the gene sets in which auxin-
mediated regulation was suppressed in nph4-1, arf19-1, and
nph4-1 arf19-1 mutants allowed us to classify the auxin-regu-
lated genes into gene sets preferentially regulated by ARF7 and
ARF19 alone or redundantly regulated by both ARF7 and ARF19
(Figure 9). The data suggest that the ARF7 and ARF19 regulate
both distinct and partially overlapping sets of target genes
(Figure 9). ARF7 appears to regulate many more auxin-induced
genes (47%) than ARF19 (4%), and ;30% of the auxin-induced
genes are redundantly regulated by ARF7 and ARF19. It is of a
contain at least one AuxRE (TGTCnC or GnGACA) in their;2-kb
promoter region (data not shown), suggesting that they are
directly regulated by these ARFs. This suggests that the ARF7
and ARF19 proteins have the capacity to act as transcriptional
activators or repressors of various auxin-regulated genes. The
current assignment of ARF7 and ARF19 solely as transcriptional
activators is not warranted. Although microarray analysis pro-
vides useful and a vast amount of information regarding the
genes regulated by the ARF7/ARF19 pair, more direct global
technologies, such as chromatin immunoprecipitation and DNA
CHIP (ChIP:CHIP), have the potential to identify target genes that
are regulated by this and other ARF pairs (Ren et al., 2000; Iyer
et al., 2001).
The lists of auxin-regulated genes in which expression is
inhibited in the mutants contain putative downstream targets of
ARF7 and ARF19. LATERAL ROOT PRIMORDIUM1 (LRP1) is
one such candidate gene. The expression level of LRP1 is
induced by auxin treatment in the wild type (Figure 10F; Ullah
et al., 2003), and its auxin-mediated induction is inhibited in
nph4-1 arf19-1 (Figure 10F). LRP1 is expressed during the early
stage of lateral root primordia (Smith and Fedoroff, 1995), and its
inhibition is consistent with impaired lateral root formation in the
nph4-1 arf19-1 mutant. Another potential candidate is the
AUXIN-REGULATED GENE INVOLVED IN ORGAN SIZE (AR-
GOS) gene, which is inhibited in auxin-treated and -untreated
nph4-1 and nph4-1 arf19-1 mutants (Figure 10F). Loss-of-
and larger plant sizes, respectively (Hu et al., 2003). The small
plant size of arf7 arf19 may be related to the low expression level
458 The Plant Cell