Formation of organelle-like N2-fixing symbiosomes
in legume root nodules is controlled by DMI2
Erik Limpens*, Rossana Mirabella*, Elena Fedorova, Carolien Franken, Henk Franssen, Ton Bisseling†, and Rene ´ Geurts
Laboratory of Molecular Biology, Wageningen University, Dreijenlaan 3, 6703 HA Wageningen, The Netherlands
Communicated by Marc C. E. Van Montagu, Ghent University, Ghent, Belgium, May 24, 2005 (received for review March 31, 2005)
In most legume nodules, the N2-fixing rhizobia are present as
organelle-like structures inside their host cells. These structures,
named symbiosomes, contain one or a few rhizobia surrounded by
a plant membrane. Symbiosome formation requires the release of
bacteria from cell-wall-bound infection threads. In primitive le-
gumes, rhizobia are hosted in intracellular infection threads that,
of symbiosomes is presumed to represent a major step in the
evolution of legume–nodule symbiosis, because symbiosomes fa-
cilitate the exchange of metabolites between the two symbionts.
Here, we show that the genes, which are essential for initiating
truncatula nodules in the region where symbiosome formation
occurs. At least one of these genes, encoding the receptor kinase
DOES NOT MAKE INFECTIONS 2 (DMI2) is essential for symbiosome
formation. The protein locates to the host cell plasma membrane
and to the membrane surrounding the infection threads. A partial
reduction of DMI2 expression causes a phenotype that resembles
the infection structures found in primitive legume nodules, be-
cause infected cells are occupied by large intracellular infection
threads instead of by organelle-like symbiosomes.
infection ? Medicago ? Rhizobium ? Nod factor
the various tissues of the root nodule, and, as a consequence, the
tissues are of graded age, with the youngest cells adjacent to
the meristem. Cell-wall-bound infection threads in these cells
grow toward and penetrate cells that are newly added to the
central tissue by the meristem. Here, unwalled infection droplets
extrude from the infection threads, after which the bacteria are
endocytosed into the cytoplasm (1). The rhizobia thus become
surrounded by a plant membrane and form organelle-like sym-
biosomes. Subsequent division of the symbiosomes ultimately
results in infected cells that become fully packed with N2-fixing
symbiosomes, requiring a major reorganization of the cytoskel-
etal and endomembrane system of the host cells, with symbio-
some membrane biogenesis and demand in infected cells being
?30 times greater than that required for plasma membrane
The formation of symbiosomes is presumed to represent a
major step in the evolution of legume nodule symbiosis, because
symbiosome formation does not occur in nodules formed on
legume species that form a symbiosis that is considered to be
more primitive (e.g., Andira spp., many species belonging to the
Fabaceae subfamily Caesalpinoideae) (3–5), and Parasponia spp.,
the only nonlegume species that can establish a symbiosis with
rhizobia (6). In these species, Rhizobium bacteria are not re-
leased into the nodule host cells but remain in infection threads,
called fixation threads, which are enclosed by a cell-wall-like
structure and in which the rhizobia can fix atmospheric nitrogen.
In the more advanced legumes, e.g., M. truncatula, the ability of
the rhizobia to fix nitrogen requires the accommodation of the
bacteria in the intracellular symbiosomal compartments, where
the bacteria find the right conditions for fixing atmospheric
edicago truncatula nodules have meristems at their apices.
By division, these meristems continuously add new cells to
The symbiotic interaction is set in motion by a molecular
dialogue between the two partners. The formation of a legume
nodule starts with cortical cell divisions, by which a primordium
is formed. In addition, the bacteria must enter the plant root. In
M. truncatula, entrance into the plant root is established by the
formation of infection threads in root hairs that have curled
specifically around a few bacteria. These infection threads guide
the bacteria to the primordia. There, the bacteria are released
from the infection threads into the host cells, upon which the
nodule primordium differentiates into a nodule (7). These early
steps of the nodulation process, e.g., infection thread and nodule
primordium formation, are induced by rhizobial signaling mol-
ecules, the so-called Nod factors. These Nod factors are, likely,
recognized by a specific LysM domain containing receptor
kinases [e.g., LYK3, LYK4 (8), and NFP (ref. 9; C. Gough and
R.G., unpublished data)] in M. truncatula that are analogous to
NFR1 and NFR5 in Lotus japonicus (10, 11). These LysM
receptor kinases trigger a signal-transduction cascade that is
essential to induce all early symbiotic events. Besides the LysM
receptor kinases, several other components of this Nod-factor-
induced signaling cascade have been identified, and, in M. trun-
catula, these components include the putative cation channel
DMI1 (12) [analogous to CASTOR and POLLUX in L. japoni-
cus (13)], the leucine-rich-repeat-containing receptor kinase
DMI2 (14) [orthologous to SYMRK in L. japonicus (15)], and
the calcium?calmodulin-dependent kinase DMI3 (16, 17).
cause a block at a very early step of nodule formation. Therefore,
such null mutants cannot reveal whether the genes also have a
function at later stages of the symbiosis. Here, we show that the
Nod factor signaling genes are expressed in the apex of the
nodule, in a few cell layers directly adjacent to the meristem,
which coincides with the place where rhizobia are released from
the infection threads into the host cells. Of these genes, DMI2
showed the highest expression level, and the protein is located in
the plasma membrane and infection thread membrane in these
cells. By creating mutants in which DMI2 expression is misregu-
lated by both (inducible) RNA interference (RNAi) and expres-
sion of DMI2 in the mutant background, by using a promoter
with a low activity in the nodule apex, we show that the receptor
kinase DMI2, besides being essential for nodule initiation, is a
key regulator of symbiosome formation. In this process, DMI2
is shown to be required in the distal part of the infection zone
to restrict infection thread growth and to switch to endocytosis
of the bacteria.
Materials and Methods
In Situ Hybridizations. In situ hybridizations were conducted on
14-d-old nodules according to procedures described by Van der
Wiel et al. in ref. 18. The nodules were fixed in 4% paraformal-
dehyde, supplemented with 0.25% glutaraldehyde in 10 mM
Abbreviation: qPCR, qualitative PCR.
*E.L. and R.M. contributed equally to this work.
†To whom correspondence should be addressed. E-mail: firstname.lastname@example.org.
© 2005 by The National Academy of Sciences of the USA
July 19, 2005 ?
vol. 102 ?
no. 29 ?
sodium phosphate buffer (pH 7.4) for 3 h. Fixed nodules were
dehydrated by passing through a series of graded ethanol baths
and embedded into paraffin. Sections (7 ?m) were dried over-
night on polylysine-coated slides at 37°C, deparaffinized with
xylene, and rehydrated by a graded ethanol series. Antisense
RNA probes were generated from ?200-bp subclones of the
coding regions of the target genes (DMI1: base pairs 1–203,
205–384, 477–691, 848-1049, 1100–1260, and 1572–1760; DMI2:
base pairs 373–617, 784-1039, 1200–1453, 1610–1864, and 1964–
2219; DMI3: base pairs 12–229, 221–467, 472–712, and 1198–
1388; and LYK3: base pairs 113–342, 483–694, 866-1079, 1258–
1486, and 1705–1947). For hybridization, a mixture containing
2 ? 106cpm of each probe was used. After washing, the slides
were coated with microautoradiography emulsion LM-1 (Am-
ersham Pharmacia) and exposed for 3 weeks at 4°C. The slides
were developed for 5 min in Kodak D-19 developer and fixed in
Kodak fixative. Sections were counterstained with toluidine blue
and mounted. For imaging, a Nikon Optiphot-2 bright-field
microscope was used.
Quantitative RT-PCR (qPCR). qPCR was conducted on RNA iso-
lated from nitrogen-starved, uninoculated roots and nodule
apices at 10 d postinoculation with Sinorhizobium meliloti strain
2011 (Sm2011). cDNA was synthesized from 1 ?g of total RNA
by using the TaqMan gold RT-PCR kit (PerkinElmer Applied
Biosystems) in a total volume of 50 ?l with the supplied hexamer
primers. qPCR reactions were performed in triplicate on 6.5 ?l
of cDNA by using the SYBR green RT-PCR Master kit
(PerkinElmer Applied Biosystems; 40 cycles at 95°C for 10 s and
at 60°C for 1 min), and real-time detection was performed on the
ABI PRISM 7700 sequence detector (Applied Biosystems) and
analyzed by using the program GENEAMP 5700 SDS (PerkinElmer
Applied Biosystems). The specificity of the PCR amplification
procedures was checked with a heat-dissociation step (from 60°C
to 95°C) at the end of the run and by agarose gel electrophoresis.
Results were standardized to the MtACTIN2 expression levels.
The primers used were MtACTIN2, 5?-TGGCATCACTCAG-
TACCTTTCAACAG-3? and 5?-ACCCAAAGCATCAAATA-
ATAAGTCAACC-3?; DMI1, 5?-GTTGCTGCAGATGGAGG-
GAAGAT-3? and 5?-GCGCCAGCCACAAAACAGTAT-3?;
DMI2, 5?-TGGACCCCTTTTGAATGCCTATG-3? and 5?-
TCCACTCCAACTCTCCAATGCTTC-3?; DMI3, 5?-TCATT-
GATCCCTTTTGCTTCTCGT-3? and 5?-GATGCTACTTC-
CTCTTTGCTGATGC-3?; and LYK3, 5?-TGCTAAGGG-
TTCAGCTGTTGGTA-3? and 5?-AAATGCCCTAGAAGTT-
Plasmids and Vectors. The full-length DMI2 coding sequence was
PCR-amplified from M. truncatula root cDNA by using primers
containing NheI and SacI restriction sites 5?-CTAGCTAGCAT-
GATGGAGTTACAAGTTATTAAG-3? and 5?-TCCGAGCTC-
The 3,103-bp fragment was subsequently cloned into pGEM-t
(Promega). The full-length DMI2 cDNA was cloned by using
NheI–SacI into a modified pBluescriptII SK? vector, containing
the CaMV 35S promoter from pMON999 (Monsanto) and a
NOS-terminator sequence with an introduced PacI restriction site,
and, subsequently, cloned by using HindIII–PacI into the binary
promoter, a 2,200-bp region upstream of the ATG start codon was
PCR-amplified on Medicago A17 genomic DNA by using primers
containing HindIII and BamHI sites: 5?-AAGCTTCAAATTTG-
GACCGAACTG-3? and 5?-GGATCCAACTTGAATCCAT-
GCTAACTAACT-3?. This fragment was used to replace HindIII–
DMI2-GFP fusion was constructed by using the Gateway vector
pK7FWG2 (20), modified to contain the red fluorescent marker
DsRED1 under the control of the constitutive Arabidopsis Ubiq-
uitin10 promoter (AatI–XbaI). DMI2, in its genomic context,
including promoter region and introns, was PCR amplified, cloned
into pENTR-D-TOPO (Invitrogen) by using primers 5?-
CACCTCTCCTTTTATCTTTTGCTTGTGG-3? and 5?-TCTC-
GGTTGAGGGTGTGACAAGG-3?, and recombined into
pK7FWG2-Q10::DsRED1 by using LR-Clonase (Invitrogen).
Plant Material and Rhizobial Strain. Medicago accession Jemalong
A17 and dmi2 mutant TR25 containing MtENOD11::GUS (21)
were used for transformations. A S. meliloti strain Sm2011
expressing GFP (8) was used to inoculate plants.
Agrobacterium rhizogenes-Mediated Transformation and RNAi. A.
rhizogenes-mediated RNAi and root transformation of 5-d-old
M. truncatula seedlings was performed according to procedures
described by Limpens et al. in ref. 19. Homogenously cotrans-
formed roots were selected, based on the expression of the red
fluorescent marker DsRED1. A 555-bp fragment of the DMI2
mRNA sequence (base pairs 1351–1905) was PCR-amplified
from nodule cDNA by using primers containing SpeI–AscI and
BamHI–SwaI restriction sites (underlined), 5?-ATACTAGTG-
GCGCGCCACCGTCCTCCTTGCTGATA-3? and 5?-ATG-
subsequently, cloned as an inverted repeat into pRedRoot (as
described by Limpens et al. in ref. 19). A Rhizobium-inducible
RNAi construct was made by using the Gateway vector
pK7GWIWG2(II) (20), modified to contain the red fluorescent
marker DsRED1 under the control of the constitutive Arabi-
dopsis Ubiquitin10 promoter (AatI–XbaI). This vector was fur-
MtENOD12 promoter (22). The MtENOD12 promoter (832 bp)
was PCR-amplified from the M. truncatula genomic DNA by
using primers containing SacI and SpeI restriction sites (under-
AGTCAAC-3? and 5?-ACTAGTTTAAGTAGTAATTTTAA-
TGTTAGTGC-3?. The 555-bp DMI2 fragment was cloned into
pENTR-D-TOPO (Invitrogen) and recombined into the modi-
fied Gateway pK7GWIWG2(II)-Q10::DsRED binary vector.
Histochemical Analysis and Microscopy. Histochemical ?-glucuron-
idase staining was performed according to the procedure de-
scribed by Jefferson et al. in ref. 23, with few modifications. Plant
material was incubated in 0.05% (wt?vol) X-Gluc (Duchefa,
Haarlem, The Netherlands) in 0.1 M sodium phosphate buffer
potassium ferricyanide. The roots were infiltrated for 30 min,
under vacuum, and further incubated at 37°C.
For electron microscopy, nodules were fixed for 3.5 h in a
mixture of 4% paraformaldehyde and 3% glutaraldehyde in 50
mM potassium phosphate buffer (pH 7.4). The nodules were
postfixed for 3 h with 1% osmium tetroxide, dehydrated through
an ethanol series, and embedded in London resin white. Ultra-
thin sections (60 nm) were obtained with a Reichert Ultracuts
ultratome (Leica), stained with 2% uranyl acetate and Reynolds’
lead citrate solution (24), and observed by using an EM208
electron microscope (Philips, Eindhoven, The Netherlands). For
light microscopy, semithin sections (0.6–1 ?m) were stained with
a 1:1 mixture of 1% toluidine blue and 1% methylene blue
solution and embedded in Paraplast (Electron Microscopy
Sciences, Hatfield, PA). Sections were viewed with a Nikon
Optiphot-2 microscope. Images where processed electronically
by using the program PHOTOSHOP 6.0 (Adobe Systems, San Jose,
CA). Imaging of DsRED1 or GFP fluorescence was done by
using the Leica MZIII fluorescence stereomicroscope and a
Nikon Optiphot-2 coupled to a mercury lamp.
Confocal imaging of DMI2-GFP was done by using a Zeiss
LSM 510 confocal laser scanning microscope [excitation 488
(GFP), 534 nm (propidium iodide)]. GFP emission was detected
www.pnas.org?cgi?doi?10.1073?pnas.0504284102 Limpens et al.
by using a 505- to 530-nm band-pass filter; propidium iodide
emission was detected in another channel with a 560- to 615-nm
band pass. Nodules were hand-sectioned, incubated for 10 min
in 0.2 ?g?ml propidium iodide, and washed with water.
Expression of the Nod Factor Signaling Genes in the Nodule. To
determine the expression pattern of the Nod factor signaling
genes in the nodule, in situ hybridizations were performed by
using LYK3, DMI1, DMI2, and DMI3 as probes. These in situ
hybridizations show that LYK3, DMI1, DMI2, and DMI3 are
expressed in the apices of nitrogen-fixing Medicago nodules (Fig.
1), consistent with the possibility that the Nod factor signaling
genes, besides being essential for initiating nodule formation,
also have a function at later stages of nodule development. In all
cases, the highest level of expression occurs in only a few cell
layers directly adjacent to the meristem, except for DMI1, which
is expressed in a slightly broader region. DMI2, especially, was
very abundant in those cell layers directly adjacent to the
meristem, which form the most distal part of the so-called
infection zone and correspond to the region where bacteria are
released from infection threads into the host cells and symbio-
somes subsequently divide (1). Real-time qPCR analysis on
RNA isolated from the nodule apex confirmed the relatively
strong expression of DMI2 in this region (Fig. 2). Real-time
qPCR analysis further showed that all three DMI genes are
expressed at markedly higher levels in the nodule apex, as
compared with roots, whereas expression of LYK3 is reduced.
DMI2 Is Essential for Symbiosome Formation. Because DMI2 tran-
scripts especially are highly abundant in the distal part of the
of DMI2 expression could specifically affect its function in the
nodule, while still allowing the first steps of Nod factor signaling
that lead to the formation of a nodule. Therefore, we made
mutants with reduced DMI2 expression by using RNAi, and we
expressed DMI2 in the mutant background, under the control of
a promoter with a lower activity in the nodule apex than the
RNAi generally results in a range of knockdown levels of a
targeted gene and can, thus, be used as a tool to identify weak
phenotypes. We exploited this characteristic to knock down
DMI2 expression by using Agrobacterium rhizogenes-mediated
root transformations and looked for phenotypes in the nodule.
A. rhizogenes-mediated RNAi, with the (constitutive) CaMV 35S
3), and, on only ?15% of the transformed RNAi roots (9 of 57),
some nodules formed upon inoculation with Sm2011. Histolog-
ical analysis of 12 of these nodules revealed, in 6 of them, the
presence of numerous wide infection threads in the central tissue
of the nodule but no symbiosomes (Fig. 4A). The other 6 nodules
showed infected cells filled with symbiosomes, similar to control
nodules (Fig. 4 C and F). This finding suggests that partial
knockdown of DMI2 expression causes extensive infection
thread growth in the nodule and blocks the release of rhizobia
from these infection threads.
Because the vast majority of the DMI2 knockdown lines did
not develop nodules, we tried to circumvent this early block of
tudinal sections of 14-d-old Medicago nodules. (A, C, E, and G) Bright-field
images of nodule sections hybridized with 35S-UTP labeled antisense DMI2
(black). (B, D, F, and H) Epipolarization images of A, C, E, and G, respectively.
M, meristem; ZII, infection zone; ZIII, fixation zone. (Scale bars: 200 ?m.)
In situ localization of DMI2, DMI1, DMI3, and LYK3 mRNA in longi-
LYK3, DMI1, DMI2, and DMI3 in roots (white) and the nodule apex (gray).
Relative transcript levels were determined by real-time qPCR and normalized
by using MtACTIN2 as reference.
Quantification of mRNA levels of the Nod factor signaling genes
(RNAi) roots determined by real-time qPCR using MtACTIN2 as reference. The
average of three independent control roots transformed with an empty
binary vector is shown.
Quantification of mRNA levels in five independent DMI2 knockdown
Limpens et al.
July 19, 2005 ?
vol. 102 ?
no. 29 ?
nodulation by using a Rhizobium-inducible RNAi system specif-
ically to knock down DMI2 expression in the nodule. To this end,
we used the Medicago ENOD12 promoter to drive the DMI2
RNAi hairpin construct (ENOD12::DMI2i). The ENOD12 gene
is specifically and strongly up-regulated upon rhizobial infection
and highly expressed in the distal part of the infection zone (18).
All transgenic ENOD12::DMI2i roots formed nodules (7 nodules
per root, n ? 14) upon inoculation with Sm2011. Sectioning
these nodules revealed, in ?30% of the transgenic nodules (7 of
23), an infection phenotype similar to that observed in the
CaMV 35S partial knockdown lines, namely, extensive infection
thread growth in the central tissue of the nodule but no
In addition to partial knockdown of DMI2 expression by
RNAi, we expressed DMI2 in the mutant background under the
control of a constitutive promoter with low activity in the nodule
apex. Based on ?-glucuronidase-expression data, we knew that
a CaMV 35S-derived promoter has a markedly lower expression
level than does the DMI2 promoter in the apex of the nodule
(data not shown). Introduction of 35S::DMI2 into the dmi2
knockout mutant TR25, by using A. rhizogenes transformation,
partially restored nodulation ability, and, on average, two nod-
ules formed on a 35S::DMI2-transformed TR25 root (n ? 43).
Sectioning these nodules (n ? 30) revealed, in all cases, a
infection thread growth in the central tissue of the nodule, and
no symbiosomes (Fig. 4 B and E). Because the used 35S
promoter is expressed in almost the entire nodule, we tested
(8 nodules per root, n ? 8) and the cytology of these nodules was
indistinguishable from that of wild-type nodules, showing that
ectopic expression of DMI2 does not affect infection thread
growth or symbiosome formation. Furthermore, the introduc-
tion of DMI2, under the control of its endogenous promoter,
fully complemented the TR25 mutant (8 nodules per root, n ?
12), with nodules showing a wild-type histology (data not
shown). These data indicate that the phenotype observed in the
35S::DMI2 TR25 nodules is due to misregulation of DMI2
expression in the nodule apex.
In a section of wild-type nodules, infection threads are
observed in only 10–20% of the infected cells (Fig. 4 C and F).
In contrast, the vast majority of the cells of the central tissue
in 1-?m-thick sections of the 35S::DMI2 TR25 nodules contain
numerous large infection threads, indicating that these infec-
tion threads fill a major part of these cells (Fig. 4 B and E).
Transmission electron microscopy showed that the mutant
infection threads contained numerous bacteria embedded in a
matrix with a low electron density and are bound by a cell wall
and a membrane (Fig. 4D). In comparison with wild-type
infection threads, these threads showed an increase in diam-
eter and extensive branching. Furthermore, these analyses
confirmed the absence of symbiosomes. Therefore, DMI2 in
the distal part of the infection zone of the nodule is required
to restrict infection thread growth and to switch to endocytosis,
by means of which, bacteria are released and N2-fixing sym-
biosomes are formed.
Both the DMI2 knockdown and the 35S::DMI2 TR25 mutant
nodules are white and so, most likely, lack leghemoglobin, a
protein required to facilitate oxygen transport to the symbio-
somes, essential for nitrogen fixation. In wild-type M. trunca-
tula nodules, the rhizobial nif (nitrogen fixation) genes are first
induced in the fixation zone, when symbiosomes have fully
filled the infected cells (R. Mirabella, personal communica-
tion). The induction of the rhizobial nif genes at such a
relatively late stage in the development of wild-type nodules
seems consistent with a lack of nitrogen fixation in the DMI2
mutant nodules. Furthermore, the 35S::DMI2 TR25 mutants
were unable to grow under nitrogen-limiting conditions. These
TR25 nodule, showing wide infection threads occupying most cells of the central tissue. (C) Longitudinal section (1 ?m) of a control nodule (transformed with
an empty binary vector). A typical zonation of indeterminate nodules can be seen. M, meristem; zII, infection zone; zIII, fixation zone. The arrow indicates
an infection thread inside a cell without releasing bacteria. The infection thread is surrounded by a fibrillar wall (arrowhead) and a membrane. (E) Closeup of
boxed area in B, showing infection threads in the distal part of the 35S::DMI2-transformed TR25 nodule occupying major parts of the cells but with no release
dark blue. it, infection thread. (Scale bars: A–C, 100 ?m; D, 1 ?m; and E and F, 20 ?m.)
Histology of nodules with reduced DMI2 expression levels and wild-type nodules. (A) Longitudinal sections (5 ?m) of 2-week-old nodules formed on
www.pnas.org?cgi?doi?10.1073?pnas.0504284102 Limpens et al.
data indicate that the mutant nodules were unable to fix
DMI2 Is Located in the Plasma Membrane and Infection Thread
Membrane. Because DMI2 is required to restrict infection thread
growth and to switch to the release of rhizobia in the distal part
of the infection zone of the nodule, we investigated whether
region. Therefore, a C-terminal GFP fusion was constructed by
using DMI2 in its genomic context; including introns and the
1.7-kb promoter region. The corresponding binary construct was
introduced into the dmi2–1 mutant TR25 (21) via A. rhizogenes-
mediated root transformation, and transgenic roots were se-
lected by using the red-fluorescent-selectable marker DsRED1
(9). Upon inoculation with Sm2011, N2-fixing nodules were
the dmi2 mutant TR25. To determine the localization of the
DMI2 protein, transgenic nodules were sectioned and analyzed
by confocal laser scanning microscopy. Strong fluorescent la-
beling could be observed in the plasma membrane as well as
infection thread membranes in cells in the distal part of the
infection zone (Fig. 5). In contrast, no GFP fluorescence was
observed in symbiosome membranes surrounding bacteria re-
leased into the host cytosol.
The receptor kinase DMI2 is an essential component of the
Nod factor signaling cascade, inducing the early steps of
nodulation, such as infection thread and nodule primordium
formation (14, 21). Here, we show that DMI2 is highly ex-
pressed in the apex of the nodule directly adjacent to the
meristem, where the receptor kinase is present in the host cell
plasma membrane as well as in the membrane surrounding
infection threads. There, DMI2 is a key regulator of organelle-
like symbiosome formation.
In situ hybridizations revealed high DMI2 expression levels in
two to three cell layers of the infection zone adjacent to the
meristem. Other components of the Nod factor signaling path-
way, e.g., DMI1, DMI3, and LYK3, were also expressed in these
cell layers, which correspond to the nodule cells where release of
bacteria from the infection thread takes place and symbiosomes
subsequently divide. It has been shown that the rhizobial nod
genes that are involved in Nod factor synthesis are still active in
these cell layers (25, 26), making it likely that Nod factor
signaling occurs in these cells. However, it cannot be excluded
that DMI2 controls symbiosome formation in a Nod-factor-
independent manner (27).
Because loss-of-function mutations in the Nod factor signaling
genes cause a block at a very early step of nodule formation, we
decided to reduce the expression of DMI2 to affect DMI2
functions in the nodule, while still allowing a nodule to be
formed. Knockdown of DMI2 expression via (inducible) RNAi,
as well as expression of DMI2 in the mutant background by
means of a 35S-derived promoter (with relatively low activity in
the nodule apex when compared with the DMI2 promoter),
resulted in extensive growth of infection threads and blocked the
subsequent release of bacteria from these infection threads. In
wild-type nodules, a switch from infection thread growth to
release of bacteria in the distal part of the infection zone is
required to facilitate symbiosome formation. Because this switch
is not occurring in the mutant nodules, it could be that the
sustained infection thread growth and suppressed release of
bacteria are interlinked processes. These data suggest that a
threshold level of DMI2 expression in the distal part of the
infection zone needs to be reached to induce the switch from
infection thread growth to release of bacteria. Because the
expression level of DMI2 is only partially reduced, it is possible
that, in addition to symbiosome formation, several other pro-
cesses in the nodule are controlled by DMI2, because these
processes might not be affected by this partial reduction. The
subcellular localization of DMI2 to the plasma membrane, as
well as to the membrane surrounding the infection threads,
suggests a direct function for DMI2 in the internalization of the
A similar block of symbiosome formation was observed in
strong DMI2 knockdown lines in Sesbania rostrata, as described
epidermal responses required for nodulation can be circum-
vented through intercellular colonization of the cortex at lateral
root bases, resulting in the formation of infection pockets from
which infection threads penetrate the cortex and nodules are
formed. It was shown that even 90% DMI2 knockdown still
allowed the formation of such infection pockets at lateral root
bases and subsequent nodule organ formation but blocked the
release of the bacteria from infection threads in the nodule (28).
In M. truncatula, the epidermal steps required for nodulation
cannot be circumvented, and similar knockdown levels cause an
epidermal block of nodulation (8).
The infection phenotype in mutant nodules with reduced
DMI2 expression shows some striking similarities to that of the
recently identified M. truncatula mutant numerous infections
polyphenolics (nip) (29). This mutant, similarly, contains nu-
merous bulbous infection threads in the central tissue of the
nodule, without the release of bacteria. However, in contrast to
dmi2 mutants, nip shows additional defects in lateral root
development, indicating that NIP has a more general function
in lateral root development, suggesting that DMI2 does not play
a role in lateral root development. Given the fact that DMI2
encodes a receptor kinase and already functions early in the Nod
factor signal-transduction cascade, either DMI2 regulates NIP
during nodule formation to facilitate symbiosome formation or
DMI2 and NIP control symbiosome formation via independent
Although the rhizobia present in the dmi2 mutant nodules do
not fix nitrogen, the infection phenotype in these nodules is
reminiscent of that of nodules formed on primitive legumes and
Parasponia (1, 3–6). In those nodules, bacteria are not released,
but, instead, N2-fixing rhizobia are hosted within intracellular
thread membrane (arrow). A confocal cross-section through an infection
thread is shown. No GFP labeling is observed around symbiosomes (S). Rhizo-
bia are stained with propidium iodide. (Scale bar: 20 ?m.)
Subcellular localization of DMI2 in the distal part of the infection
Limpens et al.
July 19, 2005 ?
vol. 102 ?
no. 29 ?
infection threads. The reason for this evolutionary step from, in Download full-text
principle, ‘‘extracellular’’ accommodation of the rhizobia to the
more intimate intracellular accommodation in organelle-like
symbiosomes could be to improve the exchange of metabolites
between the two partners, because of the absence of a cell wall
and a maximized interface. It is intriguing that this evolutionary
step of forming organelle-like symbiosomes has come under the
control of at least one of the genes essential for Nod factor
signaling in the epidermis.
We thank D. G. Barker for providing the TR25-ENOD11::GUS line and
S. de Nooijer, who participated in this work as an undergraduate student.
This work was supported by Horizon Breakthrough Grant 050-71-010
from the Dutch Genomics Initiative.
1. Brewin, N. J. (2004) Crit. Rev. Plant Sci. 23, 293–316.
2. Verma, D. P. S. & Hong, Z. (1996) Trends Microbiol. 4, 364–368.
3. De Faria, S. M., Sutherland, J. M. & Sprent, J. I. (1986) Plant Sci. 45, 143–147.
4. De Faria, S. M., McInroy, S. G. & Sprent, J. I. (1987) Can. J. Bot. 65, 553–558.
5. Sprent, J. I. (2001) Nodulation in Legumes (R. Bot. Gard., Kew, U.K.), 146 p.
6. Trinick, M. J. (1979) Can. J. Microbiol. 25, 565–578.
7. Oldroyd, G. D. & Downie, J. A. (2004) Nat. Rev. Mol. Cell Biol. 5, 566–576.
8. Limpens, E., Franken, C., Smit, P., Willemse, J., Bisseling, T. & Geurts, R.
(2003) Science 302, 630–633.
9. Amor, B. B., Shaw, S. L., Oldroyd, G. E. D., Maillet, F., Penmetsa, R. V., Cook,
D., Long, S. R., De ´narie ´, J. & Gough, C. (2003) Plant J. 34, 495–506.
10. Madsen, E. B., Madsen, L. H., Radutoiu, S., Olbryt, M., Rakwalska, M.,
Szczyglowski, K., Sato, S., Kaneko, T., Tabata, S., Sandal, N. & Stougaard, J.
(2003) Nature 425, 637–640.
11. Radutoiu, S., Madsen, L. H., Madsen, E. B., Felle, H. H., Umehara, Y.,
Gronlund, M., Sato, S., Nakamura, Y., Tabata, S., Sandal, N. & Stougaard, J.
(2003) Nature 425, 585–592.
12. Ane ´, J. M., Kiss, G. B., Riely, B. K., Penmetsa, R. V., Oldroyd, G. E., Ayax, C.,
Levy, J., Debelle ´, F., Baek, J. M., Kalo ´, P., et al. (2004) Science 303, 1364–1367.
13. Imaizumi-Anraku, H., Takeda, N., Charpentier, M., Perry, J., Miwa, H.,
Umehara, Y., Kouchi, H., Murakami, Y., Mulder, L., Vickers, K., et al. (2005)
Nature 433, 527–531.
14. Endre, G., Kereszt, A., Kevei, Z., Mihacea, S., Kalo ´, P. & Kiss, G. B. (2002)
Nature 417, 962–966.
15. Stracke, S., Kistner, C., Yoshida, S., Mulder, L., Sato, S., Kaneko, T., Tabata,
16. Levy, J., Bres, C., Geurts, R., Chalhoub, B., Kulikova, O., Duc, G., Journet,
E. P., Ane ´, J. M., Lauber, E., Bisseling, T., et al. (2004) Science 303, 1361–1364.
17. Mitra, R. M., Gleason, C. A., Edwards, A., Hadfield, J., Downie, J. A., Oldroyd,
G. E. & Long, S. R. (2004) Proc. Natl. Acad. Sci. USA 101, 4701–4705.
18. Van De Wiel, C., Norris, J. H., Bochenek, B., Dickstein, R. & Bisseling, T.
(1990) Plant Cell 2, 1009–1017.
19. Limpens, E., Ramos, J., Franken, C., Raz, V., Compaan, B., Franssen, H.,
Bisseling, T. & Geurts, R. (2004) J. Exp. Bot. 55, 983–992.
20. Karimi, M., Inze, D. & Depicker, A. (2002) Trends Plant Sci. 7, 193–195.
21. Catoira, R., Galera, C., De Billy, F., Penmetsa, R. V., Journet, E. P., Maillet,
F., Rosenberg, C., Cook, D., Gough, C. & De ´narie ´, J. (2000) Plant Cell 12,
22. Pichon, M., Journet, E. P., Dedieu, A., de Billy, F., Truchet, G. & Barker, D. G.
(1992) Plant Cell 4, 1199–1211.
23. Jefferson, A. R., Kavanagh, T. A. & Bevan, M. W. (1987) EMBO J. 6,
24. Reynolds, E. S. (1963) J. Cell Biol. 17, 208–212.
25. Sharma, S. B. & Signer, E. R. (1990) Genes Dev. 4, 344–356.
(1991) J. Bacteriol. 173, 4277–4287.
M. (1998) Mol. Plant–Microbe Interact. 11, 999–1008.
28. Capoen, W., Goormachtig, S., De Rycke, R., Schroeyers, K. & Holsters, M.
(2005) Proc. Natl. Acad. Sci. USA 102, 10369–10374.
29. Veereshlingam, H., Haynes, J. G., Penmetsa, R. V., Cook, D. R., Sherrier, D. J.
& Dickstein, R. (2004) Plant Physiol. 136, 3692–3702.
www.pnas.org?cgi?doi?10.1073?pnas.0504284102 Limpens et al.