Differential expression of nm23 genes in adult mouse dorsal root ganglia.

Abstract

Nm23 has been identified as a gene family encoding different isoforms of nucleoside diphosphate kinase (NDPK). This protein is a key enzyme in nucleotide metabolism and has been shown to play important roles in various cellular functions. In the present study, we have investigated the expression of three isotypes in mouse dorsal root ganglia. In situ hybridization and reverse transcriptase-polymerase chain reaction analysis demonstrated high levels of nm23-M1, -M2, and -M3 mRNA expression in peripheral nervous tissue. Moreover, in situ hybridization also displayed a specific nuclear localization for nm23-M2 mRNA. Immunohistochemistry with light and electron microscopy on isoform-specific antibodies revealed a differential subcellular distribution of NDPK isoforms. Isoform A was mainly cytosolic, showing only partial association with organelles. In contrast, isoform B was also found in the nucleus, which is in agreement with its proposed role as a transcription factor. The results also indicate a preferential association of isoform C with endoplasmic reticulum and plasma membranes in neuronal cells. Furthermore, isoform C appeared to combine with other NDPK isoforms as demonstrated by double-labeling evidence by electron microscopy and might be responsible for binding NDPK oligomers to membranes. Thus, isoform C may be considered as a protein of importance for maintaining intracellular pools of GTP in the vicinity of membranes and, hence, for transmembrane signaling. The results indicate a high expression of NDPK isoforms, not only in the central but also in the peripheral nervous system. Their different subcellular compartmentalization suggests that they have isoform-specific roles in neuronal cell physiology.

Full text

Page 1

Differential Expression of nm23 Genes
in Adult Mouse Dorsal Root Ganglia
PERRINE BARRAUD, LILIAN AMREIN, ERIC DOBREMEZ, SANDRINE DABERNAT,
KARINE MASSE, MONIQUE LAROU, JEAN-YVES DANIEL, AND MARC LANDRY*
EA DRED 483, Laboratoire de Biologie de la Diffe ´renciation et du De ´veloppement,
Universite ´ Victor Se ´galen, 33 076 Bordeaux Cedex, France
ABSTRACT
Nm23 has been identified as a gene family encoding different isoforms of nucleoside
diphosphate kinase (NDPK). This protein is a key enzyme in nucleotide metabolism and
has been shown to play important roles in various cellular functions. In the present study,
we have investigated the expression of three isotypes in mouse dorsal root ganglia. In situ
hybridization and reverse transcriptase-polymerase chain reaction analysis demon-
strated high levels of nm23-M1, -M2, and -M3 mRNA expression in peripheral nervous
tissue. Moreover, in situ hybridization also displayed a specific nuclear localization for
nm23-M2 mRNA. Immunohistochemistry with light and electron microscopy on isoform-
specific antibodies revealed a differential subcellular distribution of NDPK isoforms.
Isoform A was mainly cytosolic, showing only partial association with organelles. In
contrast, isoform B was also found in the nucleus, which is in agreement with its proposed
role as a transcription factor. The results also indicate a preferential association of
isoform C with endoplasmic reticulum and plasma membranes in neuronal cells. Fur-
thermore, isoform C appeared to combine with other NDPK isoforms as demonstrated by
double-labeling evidence by electron microscopy and might be responsible for binding
NDPK oligomers to membranes. Thus, isoform C may be considered as a protein of
importance for maintaining intracellular pools of GTP in the vicinity of membranes and,
hence, for transmembrane signaling. The results indicate a high expression of NDPK
isoforms, not only in the central but also in the peripheral nervous system. Their different
subcellular compartmentalization suggests that they have isoform-specific roles in neu-
ronal cell physiology. J. Comp. Neurol. 444:306–323, 2002.
© 2002 Wiley-Liss, Inc.
Indexing terms: electron microscope; nucleoside diphosphate kinase; dorsal root ganglia;
subcellular localization
A mouse nm23 gene (nm23-M1) was originally identi-
fied in several tumors as a putative metastasis suppressor
(Steeg et al., 1988a,b; Rosengard et al., 1989; Leone et al.,
1991, 1993). Subsequently, other highly homologous genes
were characterized and shown to code for nucleoside
diphosphate kinase (NDPK) isotypes. These enzymes have
been evidenced in a wide variety of organisms, including
the prokaryote Myxococcus xanthus (Munoz-Dorado et al.,
1990a), as well as Dictyostelium discoidium (Lacombe et
al., 1990), Drosophila (Biggs et al., 1990), Xenopus (Oua-
tas et al., 1997), rat (Kimura et al., 1990; Shimada et al.,
1993), and mouse (Urano et al., 1992). In human, up to six
distinct but highly related genes have been isolated:
nm23-H1 (Steeg et al., 1988a), nm23-H2 (Stahl et al.,
1991), DR-nm23 (Venturelli et al., 1995), nm23-H4 (Milon
et al., 1997), nm23-H5 (Munier et al., 1998), and nm23-H6
(Tsuiki et al., 2000).
NDPKs are ubiquitous enzymes combining in an active
heterohexamer that catalyzes
?-phosphate from ATP to nucleoside diphosphates by
forming a high-energy phosphorylated enzyme intermedi-
ate (Parks and Agarwal, 1973). Thus, NDPKs are consid-
ered to play key roles in the nucleotide metabolism of the
cell by synthesizing intracellular nucleoside triphosphates
the transferof the
Grant sponsor: the local committee of the “Ligue Nationale contre le
Cancer”; Grant sponsor: the “Association pour la Recherche Me ´dicale en
Aquitaine.”
*Correspondence to: Marc Landry, INSERM EPI 9914, Institut Francois
Magendie, 1, rue Camille Saint-Sae ¨ns, 33 077 Bordeaux Cedex, France.
E-mail: marc.landry@u-bordeaux2.fr
Received 21 February 2001; Revised 30 July 2001; Accepted 26 Novem-
ber 2001
Published online the week of February 11, 2002
THE JOURNAL OF COMPARATIVE NEUROLOGY 444:306–323 (2002)
© 2002 WILEY-LISS, INC.
DOI 10.1002/cne.10150

Page 2

(Kimura and Shimada, 1990; Lacombe et al., 1990;
Munoz-Dorado et al., 1990a,b; Nomura et al., 1992; Fuku-
chi et al., 1993). However, in addition to their role as
housekeeping enzymes, NDPK isotypes also display mul-
tifunctions possibly unrelated to their catalytic activities.
The diverse array of NDPK biological functions includes
roles in cell proliferation (Keim et al., 1992; Ohneda et al.,
1994), differentiation (Lakso et al., 1992; Amendola et al.,
1997), development (Dearolf et al., 1988a,b; Rosengard et
al., 1989; Wallet et al., 1990; Timmons et al., 1993), apo-
ptosis (Venturelli et al., 1995), and protein synthesis and
signal transduction by activating GTP-binding protein
(Chakrabarty, 1998; Klinker and Seifert, 1999; Zhang et
al., 1999).
Several lines of evidence suggest that nm23 gene prod-
ucts may also be important for neuronal functions. Muta-
tions in Drosophila NDPK caused abnormal development
of larval neural tissue (Timmons et al., 1993). NDPKs are
also detected in Xenopus (Ouatas et al., 1998) and mouse
(Lakso et al., 1992) neuronal tissue undergoing differen-
tiation and in the adult central nervous system (Dabernat
et al., 1999b). Nm23 gene products may be of importance
in maintaining the appropriate differentiation and signal-
ing pathways in neuronal tissues, as suggested by NDPK-
stimulated differentiation of PC12 cells in culture (Ger-
vasi et al., 1996; Ishijima et al., 1999).
For a better assessment of NDPKs neuronal functions, a
broad interest in possible specific tissular and subcellular
localizations of different isoforms has emerged from bio-
chemical and cellular studies (Kimura et al., 1990; Ouatas
et al., 1998; Dabernat et al., 1999a,b; Milon et al., 2000;
Tsuiki et al., 2000). Previous immunohistochemical stud-
ies aiming to provide a detailed analysis of NDPKs expres-
sion patterns have often been hindered by the use of
antibodies unable to discriminate between either isoforms
(Lakso et al., 1992) and resulting in somehow contradic-
tory results (Biggs et al., 1990; Lacombe et al., 1991;
Sastre-Garau et al., 1992; Sawan et al., 1994; Nakamura
et al., 1995; Kraeft et al., 1996; Pinon et al., 1999). More-
over, whereas the presence of NDPKs in the central ner-
vous system is well established, no data about NDPKs
expression in peripheral nervous system have been avail-
able to date.
The present study demonstrates high levels of nm23
genes expression in mouse dorsal root ganglia (DRG), a
peripheral neuronal tissue. In addition, differential sub-
cellular localizations of NDPK isoforms are shown and
their possible combinations are characterized by using
immunohistochemistry with specific antibodies under
electron microscopy.
MATERIAL AND METHODS
Animals and tissue preparation
Six C57BL/6 adult mice were used for in situ hybridiza-
tion. L4 and L5 DRGs were rapidly dissected out and
immediately frozen on dry ice.
Fifteen C57BL/6 adult mice were anesthetized and
perfusion-fixed for immunohistochemistry through the as-
cending aorta with 10 ml of saline solution (37°C) followed
by 50 ml of an ice-cold fixative (4% paraformaldehyde,
0.2% picric acid in 0.1 M phosphate buffer, pH 7.4). For
immunofluorescence, the L4 and L5 DRGs from 10 ani-
mals were rapidly dissected out, immersed in the same
fixative for 90 minutes, and rinsed for at least 24 hours in
0.1 M phosphate buffer (pH 7.4) containing 12% sucrose
and 0.01% sodium azide (Sigma, St. Louis, MO). Before
sectioning, DRGs were frozen in isopentane cooled to
?70°C on dry ice. For electron microscopy studies, the L5
DRGs from five animals were rapidly dissected out, im-
mersed in the same fixative for 90 minutes, and rinsed in
0.1 M phosphate buffer (pH 7.4). They were then post-
fixed in OsO4 (1% in 0.1 M phosphate buffer). All experi-
ments complied with the internationally accepted NIH
guidelines.
Reverse transcriptase-polymerase
chain reaction
Oligo primers.
Oligonucleotide sets of primers used in
polymerase chain reaction (PCR) amplification nm23
cDNA were selected in the 5? and 3? untranslated regions
according to published cDNA nucleotide sequences.
nm23-M1 (GenBank accession no. U85511), forward
primer location 63-81 of 5?-GCGGTAAAGCCTTGTCAT-
3?, reverse primer location 650-633 of 5?-GGTTCCTTC-
CAGGTCAAA-3?; nm23-M2
NM008705) forwardprimer
CTCGAGCGTACCTTCATT-3?, reverse primer location
591-574 of 5?-CAGTTCCAAAGTCTTTAT-3?; nm23-M3
(GenBank accession no. AF288689) forward primer loca-
tion 191-208 of 5?-CGAAGAGCTACTGCGGGA-3?, reverse
primerlocation 670-653of
AGGG-3?. The housekeeping gene used as a PCR standard
was glyceraldehyde-3-phosphate dehydrogenase (GAPDH)
(GenBank accession no. M32599) forward primer location
311-328of5?-GGTGCTGAGTATGTCGTG-3?,
primer location 598-581 of 5?-CTTCTGGGTGGCAG-
TGAT-3?.
Reverse transcriptase-PCR assay.
extracted from three mouse tissues by using Trizol (Gibco
BRL Life Technologies, Rockville, MD) according to the
protocol provided by the manufacturer. RNA was treated
with Rnase-free DNaseI (Gibco BRL) at 37°C for 20 min-
utes to remove potential genomic DNA contamination.
Reverse transcriptase (RT) products were generated from
1 ?g of total RNA by using Superscript II RT kit (Gibco
BRL).
Aliquots of the reverse transcribed cDNA preparation
were used for each PCR amplification (25-?l final volume)
containing 40 ?mol of each primer, 0.4 mM of each de-
oxynucleotide, 1? PCR buffer, 2.5 mM MgCl2, and 1 unit
of Taq DNA polymerase (Promega, Madison, WI).
PCR procedures were done in a Perkin-Elmer 9600
Thermocycler (Perkin Elmer, Norwalk, CT). Each PCR
cycle consisted of a denaturation step at 94°C for 30 sec-
onds, 30 seconds of annealing at 50°C, and 1 minute of
polymerization during 40 cycles. The final polymerization
step was extended to an additional 10 minutes at 72°C.
Data are expressed as mean ? SEM from triplicate exper-
iments.
Quantification.
Serial dilutions (2?, 5?, 10?, 20?,
50?, 100?, 200?, for nm23-M1 and -M2, and GAPDH
PCR amplifications and 2?, 3?, 4?, 5?, for nm23-M3
PCR amplifications) were prepared from RT products to
obtain a standard curve. Dilutions used in the PCR assay
were in the linear range of this curve. Samples were run
on 1.5% agarose gel containing 2% ethidium bromide, and
individual bands were quantified on a personal computer
with the NIH Image software. Values of nm23 intensity
(GenBank
location
accession
29-46
no.
5?- of
5?-TATGAAAGGCAACA-
reverse
Total RNA was
307Nm23 GENE EXPRESSION IN SPINAL GANGLIA

Page 3

were plotted on the standard curves, and the hybridiza-
tion signal was normalized with the corresponding hybrid-
ization signal obtained for GAPDH mRNA in the same
conditions.
In situ hybridization methodology
Probes.
Oligonucleotide probes were synthesized by
Eurogentec (Seraing, Belgium). Altogether, six oligonucle-
otides were used in this study for in situ detection of
nm23-M1 (n ? 2), nm23-M2 (n ? 2), and nm23-M3 (n ? 2)
mRNAs. Sequences were designed as follows from pub-
lished cDNA sequences. (A) Oligonucleotides complemen-
tary to nm23-M1 mRNA (Dabernat et al., 1999b): (1)
5?-CACTTTAATCTAACAGTGGCACGTAACACTGCACA-
GGGAGTCACG; (2) 5?-CTGTGAGAACAAGAGTAAGC-
AGGTAGAAAACCGGCACCGTCCTA. (B) Oligonucleo-
tides complementary to nm23-M2 mRNA (Dabernat et al.,
1999b): (1)5?-TAGAAAAGATGATCCATCCTGTCAGT-
GGGATGAAGAGCTCTGTCC;
GTGCTGAAAAGGATTCTGGTTTCTTCATGTCTAC. (C)
Oligonucleotides complementary to nm23-M3 mRNA
(GenBank, accession no. AF 288689): (1) 5?-GCTGACTG-
CAATAACATCTACTCATATAGCCAATGTCCGGCG; (2)
5?-GTCCCTGGACCGACTAGGTTGGGTTGACCTGCAC-
TCACACAGCAC.
All oligonucleotides were chosen in regions presenting
few homologies with sequences of related mRNAs, and
they were then checked against the Genbank database.
The oligonucleotides were radioactively labeled as previ-
ously described (Dagerlind et al., 1992) at the 3? end by
using terminal deoxynucleotidyl transferase (TdT; Amer-
sham) in a cobalt-containing buffer with [35S]dATP (Am-
ersham) to a specific activity of 1–4 ? 109 cpm/?g. They
were then purified by ethanol precipitation.
In situ hybridization procedure.
cut at 14 ?m thickness in a cryostat (Leica Instruments,
Nussloch, Germany) and thaw-mounted onto Super Frost
D Plus slides (CML, Nemours, France).
For the detection of either nm23-M1, nm23-M2, or
nm23-M3 mRNA, the sections were incubated as de-
scribed previously (Dagerlind et al., 1992), without any
pretreatment, with 0.5 ng per slide of one or both [35S]-
labeled specific oligonucleotides in a hybridization solu-
tion containing 50% formamide (Sigma), 4? SSC, 1? Den-
hardt’s solution, 1% sarcosyl (N-lauryl sarcosine, Sigma),
0.02 M phosphate buffer (pH 7.0), 10% dextran sulfate
(Sigma), 250 ?g/ml yeast tRNA (Sigma), and 50 ?g/ml
sheared and heat-denatured salmon sperm DNA (Sigma)
in the hybridization solution. After hybridization, the
slides were rinsed in 1? SSC four times for 15 minutes at
55°C followed by 30 minutes at room temperature.
Sections were then air-dried and dipped into Ilford K5
nuclear emulsion (Ilford, Mobberly, Cheshire, UK) diluted
1:1 with distilled water, exposed for 7 to 10 days, devel-
oped in Kodak D19 (Kodak, Rochester, NY) for 3 minutes
and fixed in Kodak 3000 for 8 minutes. After mounting in
glycerol, the sections were analyzed in a Zeiss Axiophot 2
microscope equipped with a darkfield condenser. Some
sections were counterstained with cresyl violet or bisben-
zimide to assess the labeled areas.
Immunohistochemistry
Antibodies.
Recombinant mouse NDPK proteins were
obtained with previously published protocols (Lascu et al.,
1997). The coding parts of nm23 genes were inserted into
(2)5?-ACCCATCAGTA-
Frozen DRGs were
the pJC20 expression vector to transform BL21 bacteria.
Production of NDPK proteins was induced by adding 1
mM IPTG. Proteins were then purified and NDPK activity
was measured according to published protocols (Kezdi et
al., 1976; Lascu et al., 1983). NDPK A and B isoforms were
lyophilized and anti-guinea pig antibodies were raised in
guinea pig by Eurogentec. Anti-NDPK C antibody was
raised in rabbit and was a gift from Dr M. Konrad (Max
Planck Institu ¨te, Go ¨ttingen, Germany). All antibodies
were subsequently purified on HiTrap NHS-activated col-
umns (Amersham) as described elsewhere (Pinon et al.,
1999).
Western blotting.
Western blotting was performed as
described by Martinez et al. (1995). Recombinant enzymes
(10 to 100 ng) were electrophoretically separated by so-
dium dodecyl sulfate resolving gels containing 10% poly-
acrylamide and transferred to an Immobilon-P membrane
(0.1 ?m, Millipore, Saint Quentin en Yvelines, France).
Immunodetection was carried out by using the above poly-
clonal anti-NDP kinase antibodies (1/5,000, A, 2.24 ?g/ml;
B, 3.42 ?g/ml; and C, 1.9 ?g/ml). Binding of the primary
antibody was visualized by using a secondary antibody
conjugated with peroxidase (Sigma). Specific signals were
detected by COVALIGHT™chemiluminescence reaction
kit (Valbiotech).
ELISA tests.
ELISA tests were performed according
to Ternynck and Avrameas (1991) by using 96-well plates
coated with 100 ?l of 1 ?g/ml solutions of the purified A, B,
and C recombinant NDP kinases. Secondary antibodies
were used at a 1/1000 dilution and detected by incubation
with the chromogenic substrate o-phenylenediamine dihy-
drochloride (Sigma).
Immunofluorescence.
The frozen L4 and L5 DRGs
were sectioned in a cryostat (Leica) at 14 ?m thickness
and thaw-mounted onto gelatin-coated slides. The sec-
tions were then processed for indirect immunofluores-
cence. Sections were incubated overnight at 4°C with an-
tisera diluted at 1:200 (anti-NDPK A and B) or 1:1,000
(anti-NDPK C) in phosphate-buffered saline (PBS; pH 7.4)
containing 0.5% bovine serum albumin (BSA, Sigma)
(PBS-BSA). They were then rinsed in PBS at room tem-
perature. Anti NDPK A and B guinea pig antibodies were
detected by using the avidin-biotin method: sections were
incubated with biotinylated anti-guinea pig antibody (1:
200, Vector, Burlingame, CA) in PBS-BSA for 2 hours at
room temperature, rinsed in PBS, and immersed in a
solutionof PBS-BSAcontaining
streptavidin (Vector) for 30 minutes at room temperature.
Anti-NDPK C antibody was incubated with fluorescein- or
rhodamine-RedX-conjugated secondary anti-rabbit anti-
body (1:100, Jackson Immunoresearch, West Grove, PA)
diluted in PBS-BSA for 2 hours at room temperature.
After final rinsing, all sections were mounted in a mixture
ofglycerol andPBS (9:1)
phenylenediamine (Sigma) (Platt and Michael, 1983). The
sections were examined in a Zeiss Axiophot 2 microscope
equipped with appropriate filter combinations and were
photographed and digitalized at 300 dpi input resolution.
Postembedding immunogold method.
dehydrated in increasing concentrations (70–100%) of
ethanol. After impregnation with propylene oxide for 10
minutes, tissues were embedded in Epon 812. Ultrathin
sections were cut from Epon-embedded L5 DRGs and col-
lected on uncoated nickel grids (200 mesh). The sections
on grids were then reacted with a saturated aqueous so-
1:200 fluorescein-
containing0.1%
para-
L5 DRGs were
308P. BARRAUD ET AL.

Page 4

lution of sodium metaperiodate (Sigma) for 45 minutes,
rinsed in distilled water and PBS-BSA, and then incu-
bated in 1% chicken egg albumin (Sigma) for 20 minutes.
The grids were transferred onto drops of primary antibod-
ies diluted at 1:200 in PBS-BSA and incubated for 48
hours. After extensive washing in PBS, the grids were
incubated in 1% chicken egg albumin and then for 2 hours
atroom temperaturewith
conjugated secondary antibody (goat anti-rabbit [GAR 5
nm] or goat anti-guinea pig [GAG 10 nm] IgG) diluted at
1:40 in PBS-BSA.
For double immunostaining, the grids were incubated
on one face together with both primary antibodies, guinea-
pig anti-NDPK A or B and rabbit anti-NDPK C, followed
by a mixture of the appropriate colloidal gold conjugates
with gold particles of different sizes (GAR 5 nm and GAG
10 nm). The grids were then rinsed and placed on drops of
1% glutaraldehyde for 15 minutes before final rinsing in
distilled water. Ultrathin sections were then counter-
stained with uranyl acetate and examined in a Philips
CM10 electron microscope.
Quantification.
The distribution of NDPK proteins in
different compartments of L5 DRG neurons was analyzed
from immunogold-treated sections under electron micros-
copy. The analysis was performed on photomicrographs at
a final magnification of ?39,000. The measurements were
performed on 3 normal adult mice and 12 neurons per
animal were analyzed for single and double immunohis-
tochemistry. The immunogold particles were identified
and counted in association with four subcellular compart-
ments, i.e., the plasma membrane, the endoplasmic retic-
ulum and Golgi apparatus, the nucleus, and the cytosol. In
the latter, immunoparticles were associated with no de-
tectable organelles. In single-labeling experiments, the
number of gold particles detectable in the four compart-
ments after incubation with preabsorbed antibodies was
also quantified.
For double-labeling experiments, we have distinguished
the particles alone from the particles colocalized with par-
ticles of another size. In the four subcellular compart-
ments investigated, we compared the labeling for NDPK A
alone with NDPK A colocalized with NDPK C (Fig. 13a),
labeling for NDPK B alone with NDPK B colocalized with
NDPK C (Fig. 13b), labeling for NDPK C alone with
NDPK C colocalized with NDPK A (Fig. 13c), and labeling
for NDPK C alone with NDPK C colocalized with NDPK B
(Fig. 13d). The results are expressed as percentages of the
total number of particles (alone and colocalized) corre-
sponding to a given NDPK, i.e., NDPK A (Fig. 13a), NDPK
B (Fig. 13b), and NDPK C (Fig. 13c,d). The statistical
analysis was carried out by using analysis of variance
(ANOVA) followed by the post hoc Dunnett test. A confi-
dence level of p ? 0.05 was considered significant.
Controls
As internal negative controls in PCR amplifications, we
omitted the template in the reaction mixture and included
RNA instead of cDNA. A positive control was obtained by
using 10 ng of plasmid containing the corresponding
cDNA isoform as template.
For in situ hybridization control experiments, an excess
of nonlabeled probes (100-fold) was added to this mixture.
In addition to ELISA tests and Western Blotting, im-
munohistochemical controls included preabsorbtion of an-
tisera with the corresponding recombinant mouse (NDPK
theappropriate gold-
A and B) or human (NDPK C) protein at a concentration of
10?6M. In some experiments, omission of primary anti-
bodies was followed by the avidin-streptavidin technique
or incubation with colloidal gold conjugates.
Computer analysis of Nm23-M3 amino acid
sequence
Possible conformations of the first 21 amino acid resi-
dues of the published sequence of mouse NDPK C (Gen-
Bank accession no. AF 288689) were analyzed by using
the TMpred procedure as proposed on the following web
site: http://www.ch.embnet.org/software/TMPRED_form.
html
RESULTS
Tissue distribution of nm23-M1, -M2, and -M3 mRNAs
encoding NDPK A, B, and C proteins in the mouse nervous
system was determined and quantified by RT-PCR (Figs.
1, 2). For in situ hybridization, the radioactively labeled
oligonucleotide probes were detected by autoradiography.
The signal resulted in silver grains overlying the neuronal
cell bodies, indicating nm23-M1, -M2, or -M3 mRNA (Figs.
3, 4). The specificity of the primary antibodies raised in
guinea pig (anti-NDPK A and B) or rabbit (anti-NDPK C)
was assessed by Western blotting (Fig. 5) and ELISA tests
(Fig. 6). These antibodies were detected by immunofluo-
rescence under light microscopy (Fig. 7) or by a postem-
bedding method with electron microscopy (Figs. 8–11) and
were subsequently quantified (Figs. 12, 13).
nm23 isoform expression in adult mouse
sensory ganglia
RT-PCR analysis.
Total RNA extracts from various
central nervous system areas were analyzed, and the re-
sults were compared with transcript levels found in the
peripheral nervous system, particularly the DRG, and in
Fig. 1.
analysis of nm23-M1, -M2, and -M3 mRNA distribution. RT products
from isolated mouse tissue were used in PCR reaction with nm23 or
glyceraldehyde-3-phosphate dehydrogenase primers. B, brain; Cx,
cortex; C, cerebellum; OB, olfactory bulb; P, pituitary; DRG, dorsal
root ganglia; A, adrenal; H, heart; AT, adipose tissue; T, testicle; K,
kidney.
Reverse transcription-polymerase chain reaction (RT-PCR)
309Nm23 GENE EXPRESSION IN SPINAL GANGLIA

Page 5

other tissues, i.e., heart, kidney, adrenal, testicular, and
adipose tissue (Fig. 1). nm23-M1 transcripts were prefer-
entially expressed in nervous tissue with high levels in
total brain extracts and DRG compared with non-neural
tissues (Fig. 2a). In contrast, high levels of nm23-M2
mRNA were found in all tissues investigated (Fig. 2b).
nm23-M3 mRNA was abundant in the nervous system but
also in testicle and kidney (Fig. 2c). For each of the three
mRNA species, DRG displayed one of the highest tran-
script amounts.
In situ hybridization.
Labeling was found in neuron
profiles, whereas no significant signals for any of the nm23
mRNAs were detected in fiber tracts (Fig. 3a,c,e). Practi-
cally all neuron profiles were labeled for nm23-M1, -M2,
and -M3. Silver grains indicating nm23-M1 and -M3
mRNAs were seen over neuronal cytoplasm as observed in
counterstained sections (Fig. 4a,b). No variations in grain
density were observed between populations of small, me-
dium, and large neuron profiles. However, nm23-M2
mRNA displayed a specific subcellular distribution within
the perikarya, because they were frequently detected over
the nucleus (Fig. 4c–f). This feature was most obvious in
large neurons.
Western blotting, ELISA tests, and controls
ern blot analysis demonstrated that no anti-NDPK anti-
body cross-reacted with another recombinant NDPK iso-
type (Fig. 5). As expected, the signal detected corresponds
to a molecular weight of 17 kDa for NDPK A and B,
similar to that of a NDPK monomer (Gilles et al., 1991).
The signal at 18 kDa for NDPK C can be explained by the
higher molecular weight of this isoform due to the
N-terminal presequence. ELISA tests confirmed that the
polyclonal anti-NDPK antibodies are specific of one iso-
form at the dilutions used for immunohistochemistry (Fig.
6).
No immunostaining was observed under light micros-
copy when antibodies were preadsorbed with the corre-
sponding recombinant peptide (data not shown) showing
that polyclonal anti-NDPK antibodies did not recognize
another antigen in the DRG tissue. Furthermore, only
weak, homogeneous staining was detected after incuba-
tion with preabsorbed antibodies under electron micros-
copy (see also below). Similarly, in situ hybridization with
an excess of unlabeled probes abolished radioactive sig-
nals in DRG (Fig. 3b,d,f).
Light microscope immunohistochemistry.
reactivity for NDPK A, B, and C isoforms was visualized
in mouse L4 and L5 DRG (Fig. 7). Labeling was restricted
to neuron profiles not affecting fiber tracts. No associa-
tions could be established between the intensity of label-
ing and the size of the neuron profiles. Differences were
noticed in the subcellular localization of NDPK protein
stainings. Thus, NDPK A-like immunoreactivity (-LI) was
diffuse, although restricted to the cytoplasm (Fig. 7a,b). In
contrast, NDPK B-LI was detected in the whole neuron
profiles, including the nucleus (Fig. 7c,d). Staining for
NDPK C was detected in the cytoplasm (Fig. 7e). At a high
magnification, strong NDPK C immunoreactivity ap-
peared punctate in DRG neurons not only in the cyto-
plasm (Fig. 7f) but also on the plasma membrane (Fig. 7g).
No labelings for Nm23 proteins were observed in the dor-
sal horn of the spinal cord, where fibers arising from DRG
neurons ended (data not shown).
Electron microscope immunohistochemistry.
tron microscopy was carried out to provide a more detailed
West-
Immuno-
Elec-
Fig. 2.
reaction from nm23-M1 (a), -M2 (b), and -M3 (c) mRNA. Values are
expressed as mean ? SEM from triplicate experiments. DRG: dorsal
root ganglia.
Quantification of reverse transcription-polymerase chain
310P. BARRAUD ET AL.

Page 6

Fig. 3.
mouse dorsal root ganglia after hybridization with35S-labeled probes
complementary to nm23-M1 (a), -M2 (c), or -M3 (e) mRNA and with
an excess of cold probes (b,d,f). Strong labeling was seen in virtually
Emulsion-dipped autoradiograms of sections of adult
all neuronal cells (a,c,e). No positive signals were detected in the
dorsal root ganglia with an excess of cold probes (b,d,f). All photomi-
crographs have the same magnification. Scale bar ? 50 ?m in f
(applies to a–f).
311Nm23 GENE EXPRESSION IN SPINAL GANGLIA

Page 7

Fig. 4.
sections of mouse dorsal root ganglia hybridized with
probes complementary to nm23-M1 (a), -M2 (c–f), or -M3 (b) mRNA.
Sections were observed in brightfield (a,b), epi-illumination (c,e),
darkfield with epi-illumination (d), or darkfield (f). nm23-M1 and -M3
Cresyl violet- (a,b) or bisbenzimide- (c–f) counterstained
35S-labeled
mRNA were seen in the cytoplasm, and the nucleus remained devoid
of labeling (open arrows in a and b). In some neuron profiles,
nm23-M2 mRNA were strongly expressed in the nucleus as revealed
by bisbenzimide counterstaining (arrows in c–f). Scale bars ? 50 ?m
in a,b,d (applies to d), f (applies to e).
312P. BARRAUD ET AL.

Page 8

analysis of the subcellular localization of NDPK proteins
and some morphologic informations about their possible
association in various cell compartments. Neurons of var-
ious size were rich in rough endoplasmic reticulum (RER),
ribosomes, and mitochondria and had a well developed
Golgi complex. Whatever the antibody used, gold particles
were found on the cell bodies with only a very few scat-
tered particles in myelinated fibers or in the extracellular
space. The mitochondria, lysosomes, and multivesicular
bodies remained unlabeled within the cell bodies. Gold
particles accumulated differentially in other subcellular
domains, depending on the antibody used.
Single immunohistochemistry for NDPK A protein re-
sulted in clusters of gold particles on the cytoplasm. These
were especially abundant in areas enriched in ribosomes
or RER (Fig. 8a). Most gold particles did not display any
association with the organelles (51.4% ? 7.8; p ? 0.01 vs.
any of the other compartments; Fig. 12a). However, ap-
proximately one quarter was associated with the RER
membrane, and almost one fifth was seen in the close
vicinity of the plasma membrane (Fig. 12a). Labeling in
the nucleus remained very low and not significant.
Single labeling for NDPK B was prominent in both the
cytoplasm (Figs. 9a–c, 12b) and the nucleus (35.1% ? 9.3;
Figs. 9c,d, 12b). One quarter of the gold particles was
detected in the cytosol (Fig. 9b,c, 12b). A similar propor-
tion was associated with the RER membrane (Fig. 9a, 12b)
and Golgi complex. Around one fifth of the gold particles
was also seen close to the plasma membrane (Fig. 12b). In
the cytosol, clusters of gold particles were frequently ob-
served in the vicinity of the nuclear membrane (Fig. 9b).
In some cases, the particles were observed close to special-
izations of the nuclear membrane, and possibly corre-
sponded to nuclear pores (Fig. 9b).
NDPK C labeling was mostly associated with various
types of membrane. The highest proportion of particles
was found in the vicinity of the RER membrane, mainly on
its outer face, in the Golgi complex and in vesicles budding
off from the trans-Golgi network (53.3% ? 8.4; p ? 0.01 vs.
any of the other compartments; Figs. 8b,c, 12c). High
levels of labeling were also detected on plasma membrane
(27.8% ? 7.2; Figs. 8d, 12c). In contrast, no labeling was
observed in the nucleus, except a very few scattered par-
ticles.
In all cases, preabsorbtion experiments displayed weak
staining, homogeneous in all compartments (Fig. 12). This
background staining represents 8.4%, 8.5%, and 8.6% of
the total number of gold particles counted in experiments
without preabsorbtion for the detection of NDPK A, B, and
C, respectively. Moreover, after preincubation with preab-
sorbed antibodies, the background staining consisted in
isolated particles that were never observed as clusters.
Clusters displaying 5-nm or 10-nm gold particles were
observed, and those gold particles were termed “not colo-
calized.” In clusters showing an association between par-
Fig. 5.
anti-nucleoside diphosphate kinase (NDPK) A (anti-A), anti-NDPK B
(anti-B), and anti-NDPK C (anti-C) antibodies. Recombinant enzymes
A and B electrophoretically migrated to the 17-kDa position and C to
the 18-kDa position in a 10% sodium dodecyl sulfate-polyacrylamide
gel electrophoresis. For anti-A and anti-B, 10 ng of recombinant
enzymes were loaded, whereas 100 ng of recombinant enzymes were
loaded for anti-C.
Western blot analysis of the specificity of the polyclonal
Fig. 6.
B, and anti-NDPK C polyclonal antibodies were tested by the enzyme-
linked immunosorbent assay method at the concentration indicated.
Plates were coated with 1 ?g/ml recombinant enzymes.
Anti-nucleoside diphosphate kinase (NDPK) A, anti-NDPK
313Nm23 GENE EXPRESSION IN SPINAL GANGLIA

Page 9

Fig. 7.
side diphosphate kinase (NDPK) A (M1, a,b), NDPK B (M2, c,d), or
NDPK C (M3, e–g) -immunoreactive neuron profiles in adult mouse
L5 dorsal root ganglia sections. a,b: The cytoplasm of neuron profiles
was homogeneously stained for NDPK A, but the nucleus remained
negative (arrows). c,d: The labeling for NDPK B was distributed
throughout the whole neuron profile, including the nucleus (arrows).
Immunofluorescence photomicrographs showing nucleo-
NDPK B was even restricted to the nucleus in some neuron profiles
(double arrowhead). e–g: At low magnification (e), the staining pat-
tern for NDPK C resembled that of NDPK A with labeling in the
cytoplasm (arrows in e,f). At higher magnification (f,g), a punctate
staining can be observed on the plasma membrane and within the
cytoplasm (arrowheads). Scale bar ? 50 ?m.
314P. BARRAUD ET AL.

Page 10

ticles of both sizes, gold particles were referred as “colo-
calized.” The quantitative data are summarized in Figure
13.
When combining immunostaining for NDPK A (10-nm
gold particles) and C (5-nm gold particles), double- and
single-labeled structures were visualized in all positive
subcellular compartments (Fig. 10). Large gold particles
indicating NDPK A were mainly found to be associated
with small particles, especially in the cytosol (31.4% ? 5.8;
P ? 0.05 vs. NDPK A alone; Fig. 13a). Conversely, only
Fig. 8.
beling for nucleoside diphosphate kinase (NDPK) A (a) or NDPK C
(b–d) in mouse L5 dorsal root ganglia. a: Gold particles (10 nm in
diameter, arrows) were observed in cytosolic areas, without associa-
tion with any organelles. b–d: Gold particles (5 nm in diameter) were
Electron photomicrographs showing single-immunogold la-
often associated with rough endoplasmic reticulum membranes (ar-
rows in c), vesicles budding off from the trans-Golgi network (arrow in
b), or plasma membrane (arrow in d). Some of the gold particles
corresponding to NDPK C can also be seen in the cytosol (arrowhead
in d). Scale bars ? 100 nm in a–d.
315 Nm23 GENE EXPRESSION IN SPINAL GANGLIA

Page 11

half of the small particles corresponding to NDPK C colo-
calized with large ones, whatever the subcellular compart-
ment analyzed (Fig. 13c). The heaviest labeling for NDPK
C was detected on the RER membrane (Fig. 10d) or Golgi
complex (Fig. 10a). The plasma membrane also displayed
high amounts of small gold particles (Fig. 13c).
Fig. 9.
beling for nucleoside diphosphate kinase (NDPK) B in mouse L5
dorsal root ganglia. Gold particles (5 nm in diameter) were detected on
the rough endoplasmic reticulum membrane (arrows in a). Gold par-
ticles were seen in the cytosol, close to the nuclear membrane (black
Electron photomicrographs showing single-immunogold la-
arrows in b and c), in the vicinity of nuclear pores (open arrow in b),
and on the nuclear envelope (arrowhead in c). Strong labeling was
also observed in the nucleus (arrow in d). N, nucleus. Scale bar ? 100
nm in d (applies to a–d).
316P. BARRAUD ET AL.

Page 12

When performing double labeling for NDPK B (10-nm
gold particles) and C (5-nm gold particles), a high
amount of large particles was seen in the nucleus with-
out colocalization with small particles (32.4% ? 5.8; p ?
0.05 vs. NDPK B ? NDPK C; Fig. 13b). In contrast,
large particles indicating NDPK B were mostly colocal-
ized with small particles in RER (Fig 11a,b) and Golgi
complex, cytosol (Fig. 11c–e), and on the plasma mem-
brane (Fig. 11d,e). Around half of the labeling for NDPK
C was associated with the RER membrane or Golgi
complex (Fig. 13d). Most of the small particles corre-
sponding to NDPK C seen in the Golgi complex or RER
membrane (Fig. 11b) were not colocalized with NDPK B
(36.4% ? 7.4; p ? 0.05 vs. NDPK C ? NDPK B; Fig.
13d). Intermediate levels of labeling were detected on
the plasma membrane (Fig. 11d,e), where similar num-
bers of small gold particles were colocalized or not co-
localized with NDPK B (Fig. 13d). Finally, the lowest
amounts of small gold particles were found in the cy-
tosol, and no labeling was observed in the nucleus for
NDPK C (Fig. 13d).
Computer analysis of nm23-M3 amino acid
sequence
TMpred prediction output for the analysis of the first 21
N-terminal amino-acid residues of NDPK C showed that
residues 1 to 17 were likely to form a transmembrane
helix with either possible orientation, i.e., N-terminus in-
side (score ? 1,789) or outside (score ? 1,565).
DISCUSSION
Expression in DRG neurons
General distribution.
exhibit specific tissue distribution. Our RT-PCR analysis
confirmed these earlier findings as well as the preferential
expression of nm23-M1 mRNA in neural tissues already
reported in rat (Kimura et al., 1990), mouse (Dabernat et
al., 1999b), and Xenopus (Ouatas et al., 1998). By compar-
ison, a large expression of mouse nm23-M2 mRNA has
been evidenced in a large variety of organs, including the
nervous system (Dabernat et al., 1999a,b). In the present
nm23 isotypes are known to
Fig. 10.
labeling for nucleoside diphosphate kinase (NDPK) A (10-nm gold
particles) and C (5-nm gold particles) in mouse L5 dorsal root ganglia.
Association of particles of both sizes (double arrows) can be observed
Electron photomicrographs showing double-immunogold
in the Golgi complex (a) and cytosol (b,c). Clusters of either small
(single arrow in c and d) or large (arrowhead in d) particles may
indicate NDPK hexamers devoid of one or the other isoform. Scale
bars ? 100 nm in a–d.
317 Nm23 GENE EXPRESSION IN SPINAL GANGLIA

Page 13

Fig. 11.
labeling for nucleoside diphosphate kinase (NDPK) B (10-nm gold
particles) and C (5-nm gold particles) in mouse L5 dorsal root ganglia.
Large particles were mostly associated with small ones in cytoplasmic
areas, including rough endoplasmic reticulum membranes (double
Electron photomicrographs showing double-immunogold
arrows in a and b), cytosol (double arrow in c, double arrowheads in
d and e), and plasma membrane (double arrows in d and e). Clusters
containing only small particles were also frequent on the rough en-
doplasmic reticulum membrane (arrow in b). M, mitochondria. Scale
bars ? 100 nm in a–e.
318P. BARRAUD ET AL.

Page 14

study, high levels of nm23-M3 mRNA were also demon-
strated in the nervous system. In contrast, previous stud-
ies have shown that additional isoforms, i.e., nm23-H4
and nm23-H5, are expressed in a tissue-dependent man-
ner and are found at low amounts in neural tissues (Milon
et al., 1997; Munier et al., 1998). In the present study, we
focused on mouse isoforms highly expressed in neurons,
i.e., NDPK A, B, and C, encoded by nm23-M1, -M2, and
-M3 genes, respectively.
The present data provide the first evidence for nm23
gene expression in DRG, a peripheral nervous tissue.
Moreover, in our experience, DRG exhibited the highest
nm23 mRNA rates compared with other neuronal areas. A
distinct, strong signal was also detected by in situ hybrid-
ization in DRG. Thus, sensory ganglia appear to be the
site of a strong synthesis of NDPKs.
The present work describes the in vivo expression of
nm23-M3 mRNA in nervous tissue, whereas so far DR-
nm23, the human gene homologous to nm23-M3, has only
been studied in vitro in various cell line models (Venturelli
et al., 1995; Amendola et al., 1997; Martinez et al., 1997).
NDPK expression has been detected in virtually all
sensory neurons but not in glial cells. Thus, nm23 gene
expression appears restricted to neuronal cells in DRG, as
in the central nervous system. At the cellular level, vari-
ations in grain densities remained apparently low and it is
unlikely that nm23 mRNAs expression levels are related
to a specific neuronal population. NDPKs may participate
in general neuronal functions in DRG rather than being
involved in specific transmission pathways (e.g., nocicep-
tive or proprioceptive).
Differential subcellular localization in adult DRG
neurons
mRNA localization.
As expected, transcripts encoding
for NDPK A and C were found in the cytosol. Surprisingly,
nm23-M2 mRNA was found not only in the cytosol but
also, if not preferentially, in the nucleus. Such a pattern
has been previously reported for NDPK X2 mRNA coding
for a NDPK isoform from Xenopus laevis (Ouatas et al.,
1998), which is similar to the corresponding human and
mouse NDPK B.
A nuclear localization has also been described in neuro-
nal cell types for various mRNA species, including vimen-
tin (Lawrence and Singer, 1986), alpha-2 adrenoreceptor
(Nicholas et al., 1993), alpha-actin (Kislauskis et al.,
1993), and SNAP-25a (Jacobsson et al., 1996). One possi-
bility is that our oligoprobe detected pre-mRNA, i.e., un-
spliced mRNA, in addition to mature cytoplasmic mRNA.
Another hypothesis is that nm23-M2 mRNA is translo-
cated from the nucleus to the cytoplasm through specific
mechanisms, possibly related to specific ribonucleopro-
teins (Jacobsson et al., 1996). Retention of nm23-M2
mRNA in the nucleus would result in an impaired trans-
location of this particular mRNA.
Protein localization.
Antibody-purification eliminated
any possible cross-reactivity between these antibodies, de-
spite the similarities between the antigens, as demon-
strated by Western blot analysis. Unspecific staining is
also very unlikely, because no signals have been observed
after preadsorption with the corresponding recombinant
proteins. Thus, control experiments carried out in this
study ensured the specificity of all antibodies used.
Different subcellular NDPK expression patterns have
already been described, including a granular and filamen-
Fig. 12.
side diphosphate kinase (NDPK) A (a), NDPK B (b), or NDPK C (c) in
four different subcellular compartments of three adult mice dorsal
root ganglia (black bars). The results are expressed as a percentage of
the total number of counted gold particles. NDPK A is prominent in
the cytosol, without being associated with any organelle. NDPK B was
found in all investigated subcellular areas, including nucleus. NDPK
C is especially abundant on the rough endoplasmic reticulum (RER)
and plasma membranes. **P ? 0.01. Quantification of the number of
gold particles detectable in the four compartments after incubation
with preabsorbed antibodies (striped bars). The results are expressed
as a percentage of the total number of gold particles counted in
experiment without preabsorbtion.
Quantification of single-immunogold labeling for nucleo-
319 Nm23 GENE EXPRESSION IN SPINAL GANGLIA

Page 15

tous cytoplasmic staining in interphasic cells (Pinon et al.,
1999) and a diffuse (Lakso et al., 1992) or punctate (Mar-
tinez et al., 1997) labeling. The present results provide
new insights about NDPK localizations in distinct subcel-
lular domains, i.e., cytosol, nucleus, RER, and plasma
membrane of adult mouse DRG.
Fig. 13.
side diphosphate kinase (NDPK) A and C (a,c), or NDPK B and C
(b,d) in four different subcellular compartments of three adult mice
dorsal root ganglia. For each of these subcellular areas, graphs indi-
Quantification of double-immunogold labeling for nucleo-
cate the percentage of gold particles corresponding to NDPK A (a),
NDPK B (b), or NDPK C (c and d) seen alone (black bars) or associated
with another type of gold particle (striped bars). RER, rough endo-
plasmic reticulum. *P ? 0.05.
320P. BARRAUD ET AL.