=-'17-=-6_ _ _ _ _ _ _ _ _ _ _ _ - - - - - - - LETIERSTONATURE---------'-'NATURE
337 12 JANUA_R_Y_l~2
Significant differences in the amount of perinuclear a -mRN A
were evident even in a single ARIA-treated myotube. In Fig.
4a, one cluster of nuclei is heavily labelled whereas another
group, approximately 50 f.lm away in the same cell, is not label-
led above background. Previous studies have shown that high
density clusters of surface AChRs occur near myotube nuclei 5- 7 .
We have noted a correlation between the local concentration of
a-mRNA and the overall density of surface AChRs in the same
region, but the precise spatial relationship between these
molecules remains to be determined.
Marked differences among nuclei in ARIA-treated cultures
were also evident after hybridization with a probe that detects
a partially spliced nuclear RNA encoding intron 6 of the a-
subunit gene 2 (Fig. 4b ). The striking heterogeneity in the distri-
bution of the exon probe is therefore probably due to different
rates of transcription of the a-subunit gene, and cannot be
attributed entirely to extra-nuclear factors such as the distribu-
tion of rough endoplasmic reticulum. We have not yet reliably
detected a signal from the intron probe in control cultures,
probably because the level of intron 6-containing nuclear RNA
is only 16% of that of mature a-mRNA 2 •
Our results indicate that ARIA does not activate all myotube
nuclei to the same degree. The distribution of grain densities in
ARIA-treated cultures was not obviously bimodal, but we
averaged grain counts over more than one nucleus in many
measurements, minimizing the distinction between responding
and non-responding nuclei. Likewise, the resolution of our
method does not allow us to determine if the distribution of
responding nuclei is graded or multimodal. In addition, we do
not know if the nuclei labelled at background levels are a distinct
population, or part of the same distribution as nuclei labelled
above background. These issues might be resolved if the signal-
to-noise ratio produced by the intron probe can be improved.
We cannot at present account for the observed differences
among nuclei. No characteristic morphological features of
nuclei were noted that correlate with their level of expression
of a-mRNA. The ability to respond to inducing factors may be
an intrinsic property of nuclei derived from a particular
myogenic lineage. Alternatively, the ability to respond may be
determined by the cytoplasmic environment of the nuclei. It is
unlikely that the heterogeneity is related to the distribution of
receptors for ARIA, because we have observed similar differen-
ces among nuclei in a-mRNA expression following treatment
with tetrodotoxin (unpublished data). It may also be relevant
in this regard that only a few nuclei in denervated chick muscle
(15-days after hatching) seem to be heavily labelled with an
a-subunit mRNA probe 8 .
Northern blot and in situ hybridization experiments have
shown that AChR subunit mRNAs are concentrated at synaptic
sites in vivo (refs 8, 9, and unpublished results). Although it is
clear that motor neurons induce the accumulation of receptors
in the postsynaptic membrane, our results indicate that myotube
nuclei are not equally susceptible to this trophic influence. A
subset of nuclei might be predisposed to participate in synapse
formation even before the nerve arrives.
We thank Marc Ballivet for the AChR clones and Bruce
Paterson for the cardiac actin clone. We also acknowledge Cindy
Nettrour and Rebecca Dill-Devor for technical assistance. This
work was supported by a Clinical Investigator Development
Award (D.A.H.) and a Physician Scientist Award (D.L.F.), both
from the NIH, and by grants from the NIH, the McKnight
Foundation, and the James S. McDonnell Foundation (G.D.F.).
Received 24 October; accepted 24 November 1988.
1. Usdin, T. B. & Fischbach, G. D. J. Cell Bioi. 103,493-507 (1986).
2. Harris, D. A., Falls, D. L. , Dill-Devor, R. M. & Fischbach, G. D. Proc. natn. Acad. Sci.
USA 85, 1983-1987 (1988).
3. Merlie, J.P. & Smith, M. M. J. Membrane Bioi. 91, 1-10 (1986).
4. Dubinsky, J. M., Morgan, M. & Fischbach, G. D. Devl Bioi. 130, 209-219 (1988).
5. Fischbach, G. D. & Cohen, S. A. Devl Bioi. 31, 147-162 (1973).
6. Bruner, J. M. & Bursztajn, S. Devl Bioi. 115, 35-43 (l986).
7. Englander, L L & Rubin. I L J. Cell Bioi. 104,87-95 (1987).
VOL.
8. Fontaine, 8., Sassoon, D., Buckingham, M. & Changeux, J.-P. EMBO 1. 7, 603-609 ( l988).
9. Merlie, J. P. & Sanes, J. R. Nature 317, 66-68 (1985).
10. Buc-Caron, M.-H., Nystrom, P. & Fischbach, G. D. Devl Bioi. 95,378-386 (1983).
11. Hayward, L J. & Schwartz, R. J. J. Cell. Bioi. 102, 1485-1493 (1986).
12. Eldridge, J., Zehner, Z. & Paterson, B. M. Gene 36,55-63 (1985).
13. Nef. P.. Oncyser, C., Alliod. C.. Couturier. S. & Ballivet. M. EMBO J. 7, 595-601 ( 1988).
Conversion of mdx myofibres from
dystrophin-negative to -positive by
injection of normal myoblasts
T. A. Partridge*, J. E. Morgan*, G. R. Coulton*,
E. P. Hoffmant & L. M. Kunkelt
*Department of Histopathology, Charing Cross and Westminster
Medical School, Fulham Palace Road, London W6 8RF, UK
t Division of Genetics, Childrens Hospital, Pediatrics, Harvard
Medical School and Howard Hughes Medical Institute, Boston,
Massachusetts 02115, USA
An important corollary to the recent advances in our understanding
of the primary cause of Duchenne muscular dystrophy 1 ~, is the
validation of genuine genetic homologues as animal models of the
disease 7 ' 8 in which potential therapies can be tested. The persistent
skeletal muscle necrosis that characterizes human Duchenne mus-
cular dystrophy9 is also seen in the mdx mouse 1{}-13 and is, in both,
a consequence of a deficiency of dystrophin6 ' 7 , probably within the
muscle fibres themselves 14- 16 • As injected muscle precursor cells
of one genotype can fuse with host muscle fibres of a different
genotype and express the donor genes 17•18, we decided to test grafts
of normal muscle precursor cells to see if they could induce
synthesis of dystrophin in innately dystrophin-deficient mdx muscle
fibres. We show that injected normal muscle precursor cells can
fuse with pre-existing or regenerating mdx muscle fibres to render
many of these fibres dystrophin-positive and so to partially or
wholly rescue them from their biochemical defect.
Mononucleated cells were prepared by enzymatic dissociation
of normal neonatal mouse muscle 18 . Between 5 x 10 5 and 4 x 106
cells were injected into the right extensor digitorum longus
(EDL) muscles of 5-27-day-old mdx hosts. Mice were killed
after 20-99 days and the EDL and adjacent tibialis anterior and
peroneus muscles removed from the injected and contralateral
non-injected legs. Successful fusion of donor muscle precursor
cells (mpc) into host (mdx) myofibres was verified by isoelectric
focusing of allotypic isoenzymes of glucose 6-phosphate
isomerase (GPI EC 53-1-9) 17 (Fig. 1). Donor mpc, homozygous
for GPI-lsb (producing the BB isoenzyme) were injected into
homozygous GPI-lsa hosts (producing the AA isoenzyme). The
AB GPI heterodimer can be formed only in mosaic host/donor
myofibres in which both types of myonucleus express their
individual GPI-1 alleles. 'Host' mdx strains (designated
mdx/ nude and M9) were bred to be homozygous for the GPI-1 sa
allele and also to avoid immune rejection of the injected cells.
Athymic 'mdx/nude' double mutants accept a wide range of
allografts, whereas the M9 strain is MHC-compatible with the
donor strains and is readily made tolerant to these strains at
birth 19 .
Of 70 mice analysed, 39 had the heterodimer AB GPI in one
or more muscles, indicating fusion of the injected cells into host
(mdx) myofibres (Table 1; Fig. 1). In 18 muscles the proportion
of AB heterodimer exceeded that of the donor (BB) isoenzyme
(see for example Fig. 1b, c and e), indicating that the majority
of surviving donor nuclei were resident in mosaic muscle fibres.
Grafts of normal mpc succeeded less frequently in young (5-7
day) mouse muscles (3 of 17), in which myofibres should be
acquiring new myonuclei during rapid growth 20 , than in regen-
erating muscle of 19-27 -day-old mice (36 of 53). ln the older
NATURE VOL. 337 12 JANUARY 1989
LETIERS TO NATURE
BB
increasing
AB
pH
AA
Fig. I GPI isoenzyme analysis of mdx muscles injected with mpc
derived from normal (C57Bl/ 10+ I+) donors (lanes a-e) or mpc
derived from mdx (C57Bl/ 10 mdx/ mdx) donors (lane f). a, EDL
muscle of mdx/ nude host, injected with 1.5 x 106 normal mpc at
7 days and examined 42 days later. This muscle contained only
the host AA GPI isoenzyme; presumably the injection had failed .
b, EDL muscle of M9 host, injected with 1.5 x 106 normal mpc
at 21 days and examined 50 days later. Host AA, donor BB and
the AB heterodimer isoenzymes were present. c, TA muscle of
mdx/ nude host, adjacent to an EDL which had been injected with
5 x 105 normal mpc at 23 days and examined 40 days later. Host
AA, donor BB and the AB heterodimer isoenzymes were present.
6
d, EDL muscle of mdx/nude host, injected with 1.7 x 10 normal
mpc at 25 days of age and examined 50 days later. Host AA, donor
BB and the AB heterodimer isoenzymes were present. e, EDL
muscle of mdx/ nude host, injected with 1 x 106 normal mpc at 27
days and examined 20 days later. Host AA, donor BB and the AB
heterodimer isoenzymes were present. f, ED L muscle of mdx/ nude
host, injected with 5 x 10 5 mdx mpc at 24 days and examined 21
days later. Host AA, donor BB and the AB heterodimer isoenzymes
were present. This indicates that the donor mdx cells had become
incorporated into the host mdx muscle fibres as myonuclei. No
dystrophin was detected in such muscles. g, Standard: skeletal
muscle from an F 1 (C57Bl/6J x 129/ ReJ) mouse, to show the loca-
tion of the AA, BB and heterodimer AB GPI isoenzymes. h,
Control: a mixture of C57Bl/6J and 129/ReJ skeletal muscles, to
show that the AA and BB GPI isoenzymes do not reassociate to
form the heterodimer AB GPI isoenzyme during isoelectric
focusing.
Methods. Glucose-6-phosphate isomerase isoenzymes were separ-
ated by isoelectric focusing on 0.8 mm agarose gels of extracts
from individual 5 fi.m cryostat sections. The enzyme was visualized
with an agarose overlay containing fructose-6-phosphate, glucose-
6-phosphate dehydrogenase, Nitroblue Tetrazolium and co-
factors 23
mice, fusion of donor mpc into host muscles was significantly
more frequent in the nude/mdx (27 of 34) hosts than in M9
hosts (9 of 19) (P = 0.038; 2 x 2, Yates-corrected x 2 test), suggest-
ing that there was some immunological impediment to mpc
grafting, even in hosts that are MHC-compatible and tolerant.
Heterodimeric GPI was sometimes found in a muscle adjacent
to the injected EDL (see for example Fig. 1c) where it may
indicate inaccurate injection or migration of mpc between con-
tiguous muscles 2 1•22 • Heterodimeric GPI was never observed in
the non-injected muscles of the contralateral leg.
To see if the normal mpc that had been incorporated into
mdx myofibres could produce dystrophin, selected muscles con-
taining a high proportion of heterodimeric GPI were tested for
their dystrophin content by immunoblotting and by immuno-
fluorescent staining. The immunoblot in Fig. 2 of injected mdx
EDLs from two mice indicates that dystrophin of normal size
is present at up to 30-40% of normal levels, as judged by
densitometry. No dystrophin was detected in the non-injected
contralateral EDL. Immunofluorescent staining of injected
muscles showed that dystrophin was in its normal sarcolemmal
location 23 - 25 (Fig. 3) in some 10-40% of muscle fibre profiles .
In some muscles, dystrophin-positive fibres were distributed
diffusely (Fig. 3a); in others they formed foci. In addition,
individual fibre profiles often showed both dystrophin-positive
and dystrophin-negative regions; this patchy pattern is strikingly
similar to that observed in young (10-day-old) mdx/ + hetero-
zygotes26, suggesting a similar type of local expression of
dystrophin from competent, admixed with non-competent,
177
Table I Summary of implantation experiments and results of GPJ
analyses
Incidence of heterodimer*
GPI in host
Host age Adjacent
Host Donor (days) EDL muscle
mdx/ nude Normalt 5-7 1/9 1/ 9
mdx/ nude Normalt 19-27 23/34 14/34
M9 Normalt 5 0/ 8 2/ 8
M9 Normal t 21-22 6/ 19 5/ 19
mdx/ nude mdx* 24 8/ 12 2/ 12
Two mouse colonies were generated to act as hosts. The mdx mutation
was bred from its normal GPI-Jsb strain (C57Bl/10) on to a background
homozygous for the GPI-1 sa allele as it is easier to detect small amount s
of donor cells of GPI-ls" allotype against this background than vice
versa. At the same time, to avoid allograft rejection, we arranged for
the host mouse to be either of the same H-2 type as the donor of the
mpc, or a nu/nu mouse in which the defective immune system permits
allograft survival. To achieve the former, mdx female mice were initially
crossed with male 129/ReJ (GPl-ls•) mice (both strains H-2b). The
resultant Fl progeny were self-bred and breeding stock was selected
from the F2 generation. F2 mice that were GPI-la" allotype and , as
assessed by muscle biopsy, mdx , were crossed to obtain a colony
(designated M9) of pure-bred mdx mice homozygous for GPI-ls" and
H-2 b. Tolerance was induced in propsective M9 hosts by injection at
birth of 5 x 106 donor strain spleen cells into the orbital branch of the
anterior facial vein. A similar breeding strategy was used to obtain a
colony of nu / nu mdx mice from female mdx mice crossed with male
nu / nu mice. Here in the F2 generation nu/ nu mdx male mice were
mated with non-nude female mdx mice (nu/ nu mice are poor mothers ).
Of the latter, only those identified as nu/ + (whose litters included
nu/nu progeny), were retained as a basis for the breeding colony. Here
too we selected mice homozygous for GPJ-ls 0 . Into the muscles of these
mdx/ nude mice or tolerant M9 mice , we injected normal mpc, prepared
from newborn C57BL/10 mice as previously described 17 • Mpc were
implanted by percutaneous injection into growing muscles of 5-7 day
old mice as previously described 18 Mice 19-27 days old were anaesthet-
ized, a slit was made in the skin overlying the EDL and the mpc were
injected along the length of the EDL muscle using a 5 ,..., Hamilton
syringe 16 . The skin overlying the injection site was then closed by
suturing. Between 20 and 99 days after implantation, the target EDL
muscle and the neighbouring tibialis anterior and peroneus muscles
were removed for analysis. GPI isoenzymes from single cryostat sections
were analysed by isoelectric focusing (see Fig. I, ref. 22). The presence
of the AB heterodimeric GPI isoenzyme, lying intermediate between
the AA or BB homodimers, indicates that the implanted mpc had been
incorporated into host muscle fibres 17 •
* Number containing heterodimer/ number examined.
t C57Bl/IOScSn wild type.
:j: C57 Bl/ 1OScSn mdx/ mdx.
myonuclei in mosaic muscle fibres . We did not see this
dystrophin immunostaining in non-injected contralateral control
muscles, or in muscles that had been injected with normal mpc
but contained no donor or heterodimer GPI.
We injected C57BI/ 10-mdx-derived mpc into mdx/ nude hosts
to control for the possibility that the injection procedure itself
might have induced dystrophin expression in mdx myofibres.
In muscles where the presence of heterodimer GPI confirmed
that the injected mdx mpc had become incorporated into the
mdx host myofibres (Fig. 1/; Table 1), dystrophin was not
detectable either by immunofluorescence or by immunoblotting.
The availability of anti-dystrophin antibodies permits assess-
ment of a possible replacement therapy for an inherited
myopathy in terms of the efficacy of introduction of the poten-
tially curative gene product. We have shown that a single injec-
tion of normal mpc into a growing or regenerating mdx muscle
can lead to synthesis of normally sized dystrophin in substantial
amounts and correctly located in a large proportion of muscle
fibres. This almost certainly reflects de novo synthesis of
dystrophin, rather than passive carriage of this protein by injec-
 178
~------------------------------------LETTERSTQNATUR£----------------~N~A~Tu~R~E~v~o~L~·~33~7~1~2~JA~N~u~A~R~Y~1~989
mdxl mdx2
-e .... -...
'1:1 '1:1
....v -....s I -s...
·=- u= ·=-
<lj (II 0 (II <lj
z
0
v
(II
0
(II
=0
u z
0
Mr (K)
427 .., Dystrophin
180
116
..,ATPase
84
58
48.5
Fig. 2 Immunoblot detection of dystrophin in injected mdx
muscles. Total muscle protein derived from the injected and non-
injected (control) EDL muscles of two mdx mice was analysed for
dystrophin content. The injected muscles have apparently normal
dystrophin present at up to 30-40% of levels found in normal
mouse muscle. The immunoblot, after use for dystrophin detection,
was reprocessed for detection of a fast-twitch isoform of the
Ca 2 +Mg2+ ATPase 41 as a control for the muscle protein content
of each lane. Relative molecular mass (M,) calibration is shown
to the left.
Methods. The portion of the ED L muscles remaining from the G PI
isoenzyme and dystrophin immunofluorescence analyses were
chipped out of the histological embedding medium at -20 oc and
processed for electrophoresis on 3.5%-12.5% gradient SDS-poly-
acrylamide gels and immunoblotted as previously described 6 ,
except that gels of 1.5 mm thickness were used and the bisacryl-
amide concentration in the 3.5% gel was 0.13%.
ted cells, because these cells were entirely mononucleate 17 ·22 •27
and dystrophin is not expressed before the formation of multinu-
cleate myotubes 28 ·29 . These findings contrast markedly with ear-
lier work on the inherited phosphorylase kinase deficiency in
the ICR/IAn mouse 18 , where incorporation of grafted normal
mpc into host myofibres was poor and, even when successful,
resulted in only a marginal increase of phosphorylase kinase
above basal levels. This difference may reflect the lack ofmyone-
crosis and regeneration in the ICR/IAn as compared with the
mdx myopathy. Law has studied the murine 'dy' dystrophi 0·3\
an autosomal recessive condition involving a major neuropathic
element 32 : the aetiology and pathogenesis of this disease are
uncertain 33 ·34 , so interpretation of the results is not straightfor-
ward. Nonetheless, the considerable functional improvement
achieved by injecting dy / dy muscles with normal mpc is
encouraf!ing.
Before cell implantation can be considered as a means of
therapy for DMD in man, a number of important problems
must be addressed. First, it is important to establish whether
the introduction of dystrophin into mdx myofibres by this means
will prevent their subsequent necrosis. Second, it is difficult to
see how the requirement for large numbers of human myogenic
cells can be met, except by large-scale tissue culture and cell-
sorting35. Third, the methods used to avoid immune rejection
£7 -~
.. ••
--~!.
"
\
t
~·
.,
'
- ./
).
'----
Fig. 3 Immunofluorescent visualization of dystrophin in mdx
muscles injected with mpc derived from normal (C57Bl/IO)
donors. a, Muscle of M9 host mouse injected at 21 days with
1.5 x 106 mpc derived from a normal (C75Bl/IO) donor and
examined 21 days later. Muscle fibre profiles exhibiting the charac-
teristic pattern of sarcolemmal dystrophin staining are found
throughout the field. They are interspersed with typical mdx fibre
profiles lacking this intense surface staining, and visible only by
virtue of background sarcoplasmic staining. In a few fibres (marked
with an asterisk), some regions of the sarcolemmal profile are
weakly stained or unstained and others are brightly stained. This
patchy pattern presumably reflects the restricted area of distribu-
tion of dystrophin around myonuclei derived from injected normal
mpc. Stain is streptavidin-Texas Red and the final magnification
is x 128. b, Muscle ofmdx/nude mouse injected at a time of muscle
growth (7 days) with 4 x I 06 mpc derived from a normal ( C57Bl/ I 0)
donor and examined 31 days later. This shows a mixture of com-
pletely stained, partially stained and unstained muscle fibres in
what appears to be an area of regeneration. It is uncertain whether
these donor cells entered the muscle fibres during growth, during
regeneration following the damage caused by the injection or
during the regeneration that follows the onset of mdx lesions at
around 21 days. The presence of some very small dystrophin-
positive fibres suggests that the latter mechanism is at least partially
responsible. Stain is streptavidin-Texas Red and the final mag-
nification is x200.
Methods. Cryostat cross-sections (5-8 J.Lm) of the muscles were
thawed on to gelatin-coated slides. PBS containing 5% horse serum
was used for blocking of the sections and for all reagent dilutions
and washes. The primary antibody was a mixture of crude sheep-
anti-mouse-dystrophin antisera directed against the dystrophin
antigens of molecular weight 30,000 and 60,000 and diluted I : 500.
The second antibody was a biotinylated donkey anti-sheep (Amer-
sham), diluted I: 250, followed by a streptavidin-Texas Red (BRL)
or streptavidin-fluorescein (Amersham) diluted I: 40 or I: 250
respectively.
in the present study would not be applicable to man, although
the absence of class I MHC antigens on mature muscle fibres
in man 36 and in mouse 37 suggests that hosts and donors may
require less stringent matching for muscle grafts than for other
tissues. It is also encouraging that long-term survival of muscle
allografts in the mouse has been achieved with cyclosporin
A30·38 , but the demonstration of both class I (ref. 36) and class
II (ref. 39) MHC antigens on muscle fibres in some circum-
stances encourages caution. Fourth, there is the formidable
problem of treating all the major muscle groups; as the migratory
powers of mpc have not been shown to exceed a few mil-
 - --LETIERSTONATURE
NATURE VOL 337 12 JANUA R:...:
Y-'-'19:...:8_.:_
9_ _ _ _ 179
limetres 22 •40 , injections would need to be spaced by not more
than 1 em apart. Finally, the pronounced cardiac abnormalities
in DMD may not easily be addressed by the techniques described
in this report.
We thank Dick Sullivan for help with histology and Ron
Barnett for photography. This work was supported by grants
from the Muscular Dystrophy Group of Great Britain and
Northern Ireland (T.A.P., J.E.M . and G.R.C.) and the Muscular
Dystrophy Association of the United States (L.M.K.), the Wil-
liam Randolph Hearst Fund (E.H.), and the National Institutes
of Health (L.M.K.). L.M.K. is an associate investigator of the
Howard Hughes Medical Institute. This work owes much to the
initial inspiration of Professor J. C. Sloper.
Received 5 Oct ober; accepted 2 December J 988.
l. Monaco, A. P. et al. Nature 323, 646- 650 ( 1986).
2. Burghes, A. H. M. eta/. Nature 328, 434-437 (!987).
3. Hoffman , E. P., Mo naco, A. P., Feener, C. C. & Kunkel, L. M. Science 238, 347-350 ( 1987).
4. Koenig, M. eta/. Cell 50, 509-517 (1987).
5. Koenig, M .. Monaco, A. P. & Kunkel, L. M. Ce/153, 219- 228 ( 1988).
6. Hoffman, E. H. eta/. New Eng/ 1. M ed. 318, 1363-1368 (1988).
7. Hoffman, E. H ., Brown, R. H. & Kunk el, L. M. Cell 51, 919-928 (1987).
8. Cooper, B. J. era/. Na ture 334, 154- 156 ( 1988).
9. Dubowitz, V. , Sewry. C. A. & Fitzsimm o ns, R. B. in Muscle Biopsy, a Practical Approach
2nd ed n (Bai llihe Tindall , London, Philadelphia and Toronto, 1985).
10. Bulfield, G., Si ll er, W. G. , Wight, P. A. l. & Moore , K. J. Prnc. natn. Acad. Sci. USA. 81,
11 89-1 192 ( 1984) .
11. Tanabe, Y., Esaki, K. & Nomura, T. A cta neuropath. 69,Yl-9 5 ( \986).
The mucosal vascular addressin is a
tissue-specific endothelial cell
adhesion molecule for
circulating lymphocytes
Maurice Nakache, Ellen Lakey Berg,
Philip R. Streeter & Eugene C. Butcher
Laboratory of Experimental Oncology, Department of Pathology,
Stanford University, Stanford, California 94305, USA and the
Palo Alto Veterans Administration Medical Center, Palo Alto,
California 94304, USA
Tissue position-dependent or address-dependent expression of cell
adhesion molecules has been proposed to play a part in cellular
positioning in a variety of systems, for example during neural
development, the metastasis of neoplasms, and the tissue-specific
homing of lymphocytes. The extravasation of blood-borne lympho-
cytes is regulated by interactions with the endothelium of
specialised venules, such as the high endothelial venules (HEY)
in organized lymph nodes and mucosal lymphoid tissues 1-4. At
least three separate lymphocyte-HEY recognition systems have
been described, one mediating tissue-selective lymphocyte interac-
tions with HEY in peripheral lymph nodes, another in mucosal
lymphoid organs, and a third in inflamed synovium4-8. We have
previously identified a tissue-specific 'vascular addressin' in the
mouse which is selectively expressed by venules mediating lym-
phocyte-homing into mucosal tissues 9 • To determine whether this
addressin is a specific adhesion molecule for lymphocytes, we have
isolated it by monoclonal antibody-affinity chromatography and
inserted it into supported phospholipid planar membranes 10•11 •
Lymphocytes bind to membranes containing the addressin, but not
to phospholipid bilayers or to control glycophorin-reconstituted
membranes. Only those lymphocytes and lymphoma cell lines
capable of binding to mucosal HEY adhere well to the isolated
addressin; peripheral lymph node HEY-specific or HEY-non-
binding cell lines do not bind. Binding is blocked by anti-addressin
antibody MECA-367. We conclude that the mucosal vascular
addressin is a tissue-specific endothelial cell-adhesion molecule
for lymphocytes, and suggest that it could regulate lymphocyte
12. Camwat h, J. & Shotton, D. 1. Neural. Sci. 80, 39- 54 I 1987).
13. Coulton, G. R., Morgan, J. E., Pa nridge, T. A. & Sloper, J. C. Neuropath . appl. Ne!lrobiol.
14, 53-70 ( 1988).
14. Thomso n, W. H. S., Sweetin , J. C. & Elton, R. A. Nature 249, 151-152 ( 1974).
15. Hoffm a n, E. P., Hudeck i, M.S .• Rosenbe rg. P. A.. Pollin a. C. M. & Kunkel , L. M . Neuro n.
1. 411-420 11988).
16. Morgan , 1. E., Co ulton , G. R. & Partridge, T. A. Muscle Nerve {i n th e press).
17. Watt , D. J ., Morgan, J. E. & Pa rtridge, T. A. Mu scle Nerve 7, 741-750 I 1984).
18. Morgan, J . E., Watt, D. J ., Sloper, J. C. & Partridge, T. A. J. Neurol. Sci. 86, 137- 147
{1988) .
19. Billingham, R. E., Brent, L. & Medawar, P. B. Phil. Trans. R. Soc. 8239, 1035- 1036 ( 1956).
20. Cardasis. C. A. & Cooper, G. W. 1. exp. Zoot. 191 ,347- 358 ( 1975).
21. Watt, D. 1., Morgan , J. E .. Cli fford, M . A. & Partridge, T. A. Anal. Embryo/. 175, 517-536
( 1987).
22. Morgan , J. E., Cou lton. G. R. & Partridge, T. A. J. Mu scle Res. Cell A1otil. 8, 386-396 ( 1987).
23. Sugita, H . et at. Proc. Japan Acad. 64, 37-39 (1988).
24. Zub rzycka-G aarn, E. E. eta/. Nature 333, 466-469 I 1988).
25. Bonill a, E. et al Cell 54, 477 - 452 ( 1988) .
26. Watkin s, S.C., Hoffman, E. P., Slayter, H. S. & Kunkel, L. M. Muscle & Nerve (sub mitted).
27. Morgan, J. E. J. Neurol. Sci. 85, 197-207 (1988).
28. Nude) , U. , Robzyk, K. & Yaffe , D. Nature 331 ,635- 638 {1988).
29. Lev, A., Feener, C., Kunkel , L. M . & Brown, R. H. I bini. Chern. 262, 158 17- 158 19 ( 1987).
30. Law, P. K., Goodwin , T. G. & Li . H-J. Trans. Proc. 20, 1114- 1119 ( 1988).
3 1. Law, P. K., Goodwin , T. G. & Wang, M.G. Muscle Nerve II , 525- 533 (1988).
32. Bradley, W. G. & Jenkinson , M . 1. Neuml. Sci. 18,227 - 247 11973).
33. Pet erson, A. C. Nature 248, 561 - 564 11974).
34. Ontell, M. Hum. Path. 17, 673 - 682 (1986)
35. Webst er, C. et al. Expl Cell R es. 174, 252-265 (19 88).
36. Appl eyard, S. T. , Dunn, M . J., Dubowitz, V. & Ro se, M. L. Lancet i. 361-363 ( 1985 ).
37. Pond er, B. A. J ., Wilkinson , M. M ., Wood, M. & Westwood , J. H. J. Hisrochem. Cytochem.
31 , 911 - 9 19 (1983 ).
38. WaH . D. J .• Morgan, J. E. & Partridge, T. A. Clin. exp. lmmun. 55, 419-426 ( 1984).
39. Retch er, J. A. 1. di11. Path 41 , 410- 414 ( 1988).
40. Lipton, B. H. & Schultz, E. Science 205, 1292-1294 119791.
41. Kapri elian, Z. & Fambrou gh, 0. M. Oevl Bioi. 124,490 - 503 11987) .
traffic into mucosal tissues by mediating attachment of blood-borne
cells to endothelium.
The mucosal vascular addressin (MAd) , a glycoprotein of
relative molecular mass ( M,) 58,000-66,000 (58 K-66 K) which
is highly expressed by HEY in mucosal lymphoid tissues,
was purified to near-homogeneity from detergent lysates of
mouse mesenteric lymph-node stroma by antibody-affinity
chromatography (see Fig. 1). Artificial planar lipid membranes
containing the isolated MAd were prepared on quartz slides,
and we assessed the ability of these membranes to su pport
binding of normal mouse lymphocytes and of TKl, a mouse
lymphoma that binds 3-5 times better than normal mouse lym-
phocytes to mucosal HEY in Peyer's patch frozen sections 6 .
Both normal lymphocytes and TKI cells bound better (7-10-
fold) to the addressin-containing than to the control gly-
cophorin-reconstituted membranes, or to membranes without
inserted proteins (see Fig. 2 for TK1 cells). Also, TKI cells
bound roughly 5 times more efficiently than lymph node cells
to the MAd-containing membranes . Under the conditions used,
up to 50% of applied TK1 lymphoma cells bound. Further
experiments showed that there was binding to addressin, like
binding to HEY in the Stamper-Woodruff frozen-section assai,
both at 7 oc and at 25 oc (not shown).
We compared binding of lymphomas that express functional
homing receptors for mucosal HEY with the binding of lym-
phomas which recognize only peripheral lymph-node HEY, or
which have no HEY-binding ability at all ( Fig. 3). TKJ and
TK43 , T-celllymphomas which bind to mucosal HEY in frozen
sections 6 • 12 , also bind well to MAd-reconstituted membranes.
In contrast, the peripheral lymph-node HEY-specific B-cellline
38C13 (ref. 13), and homing receptor-negative T cell (BW5147),
and B-lineage (Ll-2) lymphomas' 2 ·' 3 did not bind above the
background levels defined by adherence to glycophorin-recon-
stituted membranes. Thus the MAd selectively supports the
binding oflymphoid cells which are able to recognize and adhere
to mucosal HEY. The specificity of binding was further
confirmed in monoclonal antibody blocking experiments (Fig.
4) . Pretreatment of the membranes with anti-addressin antibody
MECA-367 reduced lymphocyte binding to background levels.
Control a ntibodies, including another anti-addressin antibody
MECA-89 which does not block lymphocyte binding to HEY
(ref. 9), had no significant effect.
The results indicate that the mucosal vascular addressin is a