which subsequently regulates cytoskeletal organization and clathrin-mediated endocytosis of
PIN proteins (5, 17, 18, 22, 23). However, it is not
known how ABP1 transmits the auxin signal to
regulate these cytoplasmic responses. Here, we
demonstrate that the transmembrane kinase (TMK)
members of the receptor-like kinase family interact
with ABP1 on the cell surface in an auxin-dependent
manner and are required for the auxin-mediated
activation of ROP GTPase signaling.
To transduce extracellular auxin to the cytoplasmic responses, secreted ABP1 is expected to
communicate with the cytoplasm through a transmembrane docking protein (17, 18, 24). We hypothesized that TMKs serve as ABP1 docking proteins.
TMKs belong to a clade of receptor-like kinases
with four functionally overlapping members whose
founding member is TMK1. They contain an intracellular kinase domain, a single transmembrane
pass, and an extracellular domain with two regions of leucine-rich repeats (LRRs) separated
by a non-LRR region. TMKs affect multiple auxin-mediated processes (25). We found that tmk mutants are affected in the same responses regulated
by ABP1. The tmk1–/–;tmk2–/–;tmk3–/–;tmk4–/–
mutant (tmk1–/–;tmk234) displayed embryo lethality, though with a lower penetrance of lethality as
conferred by abp1 null mutations (fig. S1). Both
tmk1–/–;tmk234 and tmk1–/+;tmk234 seedlings
displayed single cotyledons and fused leaf-cups,
typically found in pin1-1 mutants (Fig. 1, A to
C; fig. S2A; and table S1). Some PINs (including
PIN1) modulate auxin efflux, are polarly distributed to the plasma membrane, and are regulated
by ABP1- and ROP GTPase–dependent auxin
signaling (17, 18, 26–28). PIN1 localization is also
affected in tmk1–/+tmk234 mutant (fig. S1, E to
G), as in abp1 or rop mutants (18, 28). A weak
abp1 allele (abp1-5) greatly enhances cotyledon
defects in tmk1–/+;tmk234, suggesting a functional
interaction between ABP1 and TMKs (table S1).
Auxin promotes the development of interdigitated pavement cells in the Arabidopsis leaf
epidermis through ABP1 and ROP GTPases
(18). The pavement cells of tmk1–/–;tmk234 showed
interdigitation defects similar to but stronger
than those observed in the abp1-5 mutant (Fig.
1, D and F). Just as in abp1-5, the pavement cell
defects in tmk1–/–;tmk234 mutants were not
rescued by auxin (Fig. 1, D to G) (18). The
abp1-5;tmk1–/–tmk234 quintuple mutant displayed
a phenotype similar to that of the tmk1–/–;tmk234
mutant (fig. S2C). These results suggest that
TMKs are required for auxin promotion of pavement cell interdigitation and support an overlapping function with ABP1 in this process.
Auxin activates both the ROP2- and ROP6-
dependent pathways in an ABP1-dependent man-
ner in leaf pavement cells (18). We determined
whether TMKs were also required for the rapid
activation of ROP2 and ROP6 GTPases by auxin,
similar to ABP1 (18). Green fluorescent protein
(GFP)–ROP2 and GFP-ROP6 transgenic plants
were crossed with the tmk1–/+;tmk234 mutant, and
tmk1–/–tmk234 plants containing GFP-ROP2 or
-ROP6 were isolated for ROP activity assays
(Fig. 2, A to D) (see supplementary materials and
methods). In GFP-ROP2 and GFP-ROP6 trans-
genic lines, the amount of active GFP-ROP2 and
GFP-ROP6 proteins increased nearly fourfold
upon treatment with 100 nM naphthaleneacetic
acid (NAA), as previously shown (Fig. 2, A to
D) (18). However, in tmk1–/–tmk234;GFP-ROP2
or tmk1–/–tmk234;GFP-ROP6 mutants, auxin-
mediated activation of GFP-ROP2 and GFP-
ROP6 was largely abolished (Fig. 2, A to D), as
in the abp1-5 mutant (18).
We next assessed the effect of tmk mutations
on ROP2 and ROP6 signaling targets in leaf
pavement cells. The ROP2 effector RIC4, lo-
calized to the plasma membrane especially at
the tip of the lobes, promotes the accumulation
of cortical actin microfilaments (F-actin) (Fig.
2E) (18, 22, 23). In the tmk1–/–;tmk234 mutant,
the lobe tip and plasma membrane distribution
of GFP-RIC4 was abolished, as in abp1-5 and
rop2RNAi;rop4-1 mutants (Fig. 2F) (18, 22).
Furthermore, cortical F-actin, which normally
accumulates at lobe sites (fig. S3A), was absent
from the cortical regions of tmk1–/–;tmk234 mu-
tant pavement cells, just as in abp1-5 mutants
(18, 22) (fig. S3B). The ROP6 effector RIC1
associates with cortical microtubules and pro-
motes their organization upon activation by
ROP6. The association of yellow fluorescent
protein (YFP)–RIC1 with cortical microtubules
was abolished in the tmk1–/–;tmk234 quadruple
mutants, causing the disorganization of corti-
cal microtubules, as observed in the abp1-5 and
rop6-1 mutants (18, 23) (Fig. 2, G to H; and fig.
S3, C and D). These results indicate that TMKs
participate in auxin perception or signaling that
activates both the ROP2 and ROP6 pathways in
leaf pavement cells, similar to ABP1.
Given the occurrence of both ROP2 and ROP6
activation at the plasma membrane (18, 22, 23),
their upstream signaling components are expected to localize to the cell surface as well. In
both the leaf pavement cells and mesophyll cells
of a p TMK1::TMK1-GFP transgenic line, TMK1-
GFP was localized to the plasma membrane (fig.
S4, A and B; and fig. S2, H and I). Although
ABP1 is mostly found in the endoplasmic reticulum (ER), a fraction of ABP1 was observed on
the cell surface in maize (8, 10, 11, 24, 29, 30).
To determine ABP1 distribution, we performed
immunogold histochemistry in conjunction with
transmission electron microscopy (TEM) (fig.
S5, B to H) and epifluoresence microscopy of
GFP-tagged ABP1 (fig. S5I). The TEM analysis
indicates that the majority of ABP1 localized to
the ER, whereas ~22% ABP1 is detected on the
plasma membrane in Arabidopsis root cells,
Fig. 1. Transmembrane kinase genes are required for auxin-mediated pavement cell interdigitation. (A to C) The cotyledon phenotype in the wild type (A), tmk1–/+;tmk234 mutant (B), and
pin1-1 mutant (C). Scale bars, 500 mm. (D to H) Pavement cell phenotype of the wild type with (D) or
without (E) auxin (20 nM NAA) treatment, and tmk1–/–;tmk234 quadruple mutant with (F) or without (G)
auxin (20 nM NAA) treatment. Scale bars, 10 mm. The degree of pavement cell interdigitation was
quantified by determining the average number of lobes per square micrometer of pavement cells (Ave
Lobe No./mm2) (H). Error bars indicate SD.