Abstract |
The aim of this thesis is the analysis of the molecular interactions between neurons
and glia in the myelinated fibers of the nervous system. The glial cells, that form the
myelin sheath, and the neural axons interact and mutually affect each others fate and
development. These interactions are necessary for the formation of distinct areas on
the myelinated neural axons, which mediate the rapid propagation of the action
potentials. Myelinated fibers in both the PNS (peripheral nervous system) and CNS
(central nervous system) are well organized in the following distinct areas: the node
of Ranvier, the paranodes, the juxtaparanodes and the internode. The normal structure
of the node of Ranvier, and the perinodal areas, is disrupted during neurological
disorders such as multiple sclerosis (MS), Guillain-Barre syndrome etc. In the adult,
voltage gated ion channels are concentrated in these areas (Νa+ channels in the node
of Ranvier and Κ+ channels in the juxtaparanode) which are responsible for the
propagation of the action potential and in cases of defects in the myelin sheath these
channels are diffusely distributed. Lately, members of the cell adhesion molecules of
the Ig-superfamily were characterized as components of the macromolecular complex
that mediates the axo-glial interactions in the above areas. This research is
concentrated on the IgSF-CAM member, TAG-1, which is found highly enriched in
the juxtaparanodal regions of myelinated fibers, in both PNS and CNS. TAG-1 is, so
far, the only IgSF molecule identified in the juxtaparanodal region of myelinated
fibers. In this area, it is present in both axonal and glial membranes which are in close
apposition, and it is required for the clustering of the Shaker-type potassium channels
(K+) and Caspr2. It was recently shown that TAG-1 is able to associate in cis with
Caspr2 and in trans with the complex Caspr2-(TAG-1). This complex is necessary for
the accumulation of the Kv channels at the juxtaparanodes, since in Tag-/- mice the Kv
channels and Caspr2 are not clustered in the juxtaparanode.
We investigated the domains of TAG-1 responsible for its interactions with other
juxtaparanodal proteins. By using co-transfection experiments we found that the
IgC2-GFP and FNIII-GFP constructs we generated produce functional proteins which
are externalized to the cell membrane. We also established that the Caspr2 protein
does not interact directly with Κv channels but it interacts with the immunoglobulin
domains of hTAG-1 and not with its fibronectin domains. We then directed our
attention to the association of TAG-1 with Kv in brain lysates where they are able to
6
co-immunoprecipitate. This association is not present in the TAG-1 mutant brain. In
addition, K+ channels directly interact with the immunoglobulin (IgC2) domains of
hTAG-1 but not its fibronectin-like repeats. We also looked into the putative
association of TAG-1 with the gap junction protein connexin 29 which also localizes
in the juxtaparanodes. We have examined whether this juxtaparanodal localization
was perturbed in Tag1-/- mice as but they did not display any change in the
distribution of Cx29. We conclude the juxtaparanodal complex that holds the Kv
channels in this region does not include Cx29 but is specific to TAG-1 and Caspr2.
We provide evidence that TAG-1, Caspr2 and Kv form a specific juxtaparanodal
complex where the three proteins directly interact via the Ig domains of TAG-1. This
is the first evidence so far that addresses structural issues involved in the axo-glial
interactions known to be essential for the proper organization of myelinated fibers.
These results, taken together, refine our model of interactions of juxtaparanodal
proteins in that it predicts that the homophilic interaction between glial and axonal
TAG-1 occurs via the FN domains thus allowing the Ig domains to bind Caspr2 and
Kv.
It is known that TAG-1 is expressed in both neural and glial cell membranes and that
it has the ability to bind homophilically to itself via its fibronectin-like domains.
Although the model predicts a homophilic interaction between two molecules of
TAG-1, a glial one and a neural one, mediating the axoglial interaction of the
juxtaparanode, this has not been formally shown. To assess the role of TAG-1,
specifically in the CNS, and also in an attempt to rescue the phenotype of TAG-1
deficient animals, we produced DNA constructs that will allow the generation of
transgenic animals specifically expressing TAG-1 in myelin-making glia. In addition,
we generated another construct that would drive expression of TAG-1 in adult
neurons. By crossing these animals to the TAG-1 deficient background we will test
the function of TAG-1 specifically in the CNS glia as well as neurons.
In order to obtain a further insight into the putative function of glial or neuronal TAG-
1 in the formation of axoglial contacts and in the dynamics of their association in
living cells, we used explant and dissociated cultures of dorsal root ganglion (DRG)
neurons and Schwann cells from Tag1-/- and wild type (WT) mice, as well as cocultures
of Tag1-/- DRG neurons and WT Schwann cells (and vice versa), in
conditions allowing in vitro myelination. Using immunocytochemistry, we examined
the localization and enrichment of TAG-1, K+ channels and Caspr2 and it was
7
observed that, in addition to the fact that myelin was produced normally, the altered
phenotype observed in Tag1-/- mutant mice (no clustering of Caspr2 and K+ channels
at juxtaparanodes) can be retained. The most striking discovery of our work though,
was the fact that the internodal lengths on the processes of cultured Tag1-/- DRG’s
appeared variable and sometimes much shorter, compared to the wild type cultures.
This led us to measuring these internodal lengths. The statistical analysis of the
comparison between Tag1-/- and wild type derived segments showed that the Tag1-/-
internodes are significantly (P value: P <0.0001) shorter than the controls. Next,
immunohistochemistry on teased fibers from sciatic nerves of Tag1-/- and wild type
mice, allowed us to measure the length of the internodal segments in a similar manner
and confirm that in vivo the mutant internodal segments were shorter than the controls
(P value: P <0.0001) as well.
|