A Glial-Neuronal Signaling Pathway Revealed by Mutations in a Neurexin-Related Protein

See allHide authors and affiliations

Science  26 Feb 1999:
Vol. 283, Issue 5406, pp. 1343-1345
DOI: 10.1126/science.283.5406.1343


In the nervous system, glial cells greatly outnumber neurons but the full extent of their role in determining neural activity remains unknown. Here the axotactin (axo) gene ofDrosophila was shown to encode a member of the neurexin protein superfamily secreted by glia and subsequently localized to axonal tracts. Null mutations of axo caused temperature-sensitive paralysis and a corresponding blockade of axonal conduction. Thus, the AXO protein appears to be a component of a glial-neuronal signaling mechanism that helps to determine the membrane electrical properties of target axons.

Glial cells influence neural activity indirectly by insulating neurons and regulating their microenvironment, but they also have direct effects on neuronal signaling and plasticity (1). The molecules responsible for mediating communication between neurons and glia are largely unknown.

Neurexins are a family of neuronal cell surface proteins with proposed roles in cell adhesion and intercellular signaling (2). Three distinct genes encoding neurexins have been identified in vertebrates. Neurexins, together with the closely related NCP protein family [defined by Drosophilaand human neurexin IV, along with rat contactin-associated protein (Caspr, also known as paranodin)], constitute a protein superfamily (2). Studies indicate that members of the NCP family are involved in neuronal-glial interactions in both vertebrates and invertebrates. Drosophila neurexin IV mutants (nrx IV) disrupt septate junctions and impair the blood-brain barrier (3). Vertebrate Caspr may also mediate glial-dependent insulation of axons (4). The protein binding domains in the cytoplasmic tails of neurexins and NCPs enable members of the neurexin superfamily to provide a link between the extracellular environment and intracellular signaling pathways (2).

Identification of additional members of the neurexin superfamily and analysis of their functions in vivo should help elucidate their biological significance. Here we report the genetic and molecular characterization of the Drosophila axo gene, which encodes a member of the neurexin superfamily expressed in glia but is also apparently required for normal axonal membrane excitability.axo was identified by isolation of a recessive P element–induced mutation, axoP12 , in a screen for temperature-sensitive paralytic mutations (5). A single P element at 64C1-3 on the polytene chromosomes was detected inaxoP12 ; its precise excision in rare revertants confirmed that it caused the mutant phenotype.

Genomic DNA flanking the element was cloned and used to identify the corresponding axo transcript (6) (Fig. 1A). Northern (RNA) blot analysis revealed a 7.5-kb transcript that is missing inaxoP12 mutants but is restored in revertants (7) (Fig. 1B). The complete loss of a gene product inaxoP12 , rather than the production of a thermolabile protein, is consistent with other temperature-sensitive paralytic mutants in Drosophila in which the elimination or reduced expression of various ion channels under all conditions renders the organism hypersensitive to the physiological effects of elevated temperatures (8, 9).

Figure 1

Molecular analysis of the axolocus. (A) Genomic structure of the axolocus. The exon-intron boundaries of axo were determined from genomic sequence analysis. Three genes (axo, gene b, and gene c) are localized in the vicinity of the P-element insertion site as determined by Northern blot and cDNA analysis. The axo gene contains at least 13 exons that span more than 45 kb of genomic DNA. The P element (vertical arrow) is inserted 23 bases upstream of the first exon of axo. The entire transcript of gene b is nested within the fifth intron of axo. The transcript of gene c is located upstream of axo on the other side of the P element. The direction of each transcript is indicated by horizontal arrows. Theaxo28 allele is associated with a deletion of at least 10 kb at the 3′ end of the axo locus. The location of the deletion is indicated by the horizontal line. One deletion breakpoint is localized between exons 11 and 12, as determined by genomic PCR. The location of the other breakpoint is not determined (indicated by a dashed line). The genomic structure of genec is not completely mapped. (B) Northern blot analysis of axo and gene b. axo cDNA probes recognize a 7.5-kb band in the lane representing mRNA prepared from wild-type adult flies. This 7.5-kb band is undetectable inaxoP12 mutants and is fully restored in twoaxoP12 revertants,axoP12-R1 (R1) andaxoP12-R2 (R2). Complementary DNA probes from gene b recognize a 1.7-kb band. Neither this transcript nor the 4.2-kb transcript recognized by cDNA probes from gene c(not shown) is altered in the axoP12 mutant. (C) Diagram of the structural organization of AXO as compared with that of members of the neurexin superfamily. The other proteins represented are Drosophila neurexin IV (NRX IV), rat Caspr/paranodin, and rat neurexin Iα (2–4). Gaps (dotted lines) are introduced into the sequences of the other proteins to maximize their alignment with the AXO sequence. The percent amino acid identity between the different domains in AXO and the corresponding domains in NRX IV or neurexin Iα is shown. The asterisk indicates that the two domains are less than 20% identical. The cysteine-rich Kunitz-like domain is present only in AXO. In addition, all the other proteins except AXO contain a transmembrane domain. DISC, discoidin domain.

A set of overlapping cDNAs encompassing the entire axotranscript was isolated and sequenced (10). AXO contains several motifs, including an NH2-terminal hydrophobic signal sequence followed by a consensus cleavage site, three cysteine-rich epidermal growth factor (EGF) repeats, five laminin G (Lam G) repeats, and a fibrinogen β/γ-like (FIB) segment. The arrangement of these domains, together with computer-based sequence alignments, identified AXO as a member of the neurexin superfamily (Fig. 1C). Several features distinguished AXO from other members of the superfamily: a Kunitz-like domain (11); the presence of both a Lam G domain (characteristic of neurexins) and a FIB domain (characteristic of NCPs) adjacent to the second EGF repeat; and the absence of a transmembrane domain, which suggests that AXO is secreted rather than membrane-associated (12).

Beginning at embryonic stage 13, axo transcripts were detected in the differentiating nervous system in wild-type but not inaxoP12 embryos in a segmentally repeated pattern of discrete spots directly overlying the longitudinal axonal tracts (Fig. 2, A and C). This pattern was not coincident with neuronal cell bodies, which lie lateral and medial to the nerve bundles (13), but corresponded with the location of a subset of glial cells, including longitudinal glia and segmental boundary cells (14). The glial identity of these cells was confirmed by elimination of the axo signal in gcm(glial cells missing) embryos (15) (Fig. 2, C and D). This pattern of expression was maintained throughout the third larval instar. The distribution of the AXO protein (16) differed from that of the axotranscript: AXO coincided with axonal tracts in the ventral nerve cord, brain, and parts of the peripheral nervous system (Fig. 2E). Thus, AXO appeared to be secreted from glia and subsequently localized to axonal tracts.

Figure 2

Expression patterns of the axotranscript and encoded AXO protein. (A) In situ hybridization with an axo cDNA probe in whole-mount wild-type embryos (stage 13) oriented with anterior to the left. For methods, see (13). The top embryo shows a lateral view (ventral side down) and the bottom embryo shows a dorsal view. Arrow indicates the ventral ganglion; double-headed arrow indicates the nerve cord. The segmentally repeated pattern seen as a series of discrete spots in the ventral nerve cord is maintained in the third instar thoracic ganglion (B). Anterior is to the left. (C and D) Dissected embryonic ventral nerve cords, double-labeled by in situ hybridization with an axocDNA probe (blue) and immunostained with the BP102 monoclonal antibody (orange), which stains axonal tracts (15). Anterior is at the top. In (C), the location of the axo-expressing cells relative to the axonal tracts suggests their identity as a subset of longitudinal glia. This control embryo, derived from the intercross ofgcm/CyO males and females, is gcm/CyOin genotype. The nerve cord shown in (D) is from a gcm/gcmembryo from the same cross. gcm homozygous embryos are recognizable by the aberrant structure of their ventral nerve cords, revealed with the BP102 antibody. No axo-expressing cells are observed when glia are eliminated by the gcmmutation. (E) Immunostaining of a dissected wild-type nerve cord with a polyclonal antiserum to AXO. Anterior is to the left. In contrast to the distribution of the axo transcript seen in (A) and (C), the AXO protein is localized to longitudinal axon tracts (solid arrow) as well as to an axonal scaffold in the brain (open arrow). (F) Immunostaining of an axoP12 embryo with polyclonal antiserum to AXO. No staining is detectable. Scale bar, 50 μm in (A), (B), (E), and (F); 20 μm in (C) and (D).

These results suggested that impairment of some aspect of neuronal development or function caused by loss of AXO resulted in a paralytic phenotype. Immunohistochemical studies of axo mutants indicated that development of the embryonic nervous system was normal, as were the number and distribution of synaptic boutons at the larval neuromuscular junction. Thus, the paralytic phenotype apparently involved a functional rather than a structural defect.

Excitatory junctional potentials (EJPs) were recorded at the larval neuromuscular junction (17). At 20oC, the resting membrane potential and the amplitude and duration of evoked EJPs were normal in axoP12 larvae. However, EJPs were lost within 10 min at 38oC (in 15 out of 20 larvae) (Fig. 3A) (18). During this interval, the threshold required to evoke EJPs increased until transmission failed completely. At 38oC, electrotonic stimulation still elicited the graded release of neurotransmitter inaxoP12 larvae, demonstrating that both neurotransmitter release and postsynaptic response were apparently normal under these conditions (Fig. 3C). However, direct nerve recordings revealed a defect in axonal conduction. At 37oC, compound action potentials in mutant nerves were either lost or were reduced by more than 90% (n = 10), in contrast with wild-type nerves, in which no failure was seen (n = 7) (Fig. 3, D and E) (17).

Figure 3

Temperature-dependent failure of conduction in axo. (A) A representative set of recordings from one of the axoP12 larvae that showed a complete loss of the EJP at elevated temperature. At 20oC (top), the EJP appeared normal in amplitude and duration. However, after 10 min at 38oC (middle), the response could no longer be evoked. When the temperature returned to 20oC (bottom), the EJP fully recovered. (B) A representative set of recordings from a wild-type larva under the same conditions as in (A). In the wild-type larva, the EJP persisted at the elevated temperature for at least 20 min. (C) Graded electrotonic EJPs were still elicited in axoP12 larvae after the active response failed at 38oC by the positioning of the stimulating electrode closer to the nerve terminal and the use of stimuli of increasing intensity. After 10 min at 38oC, compound action potentials recorded with a suction electrode were severely diminished in axoP12 (D) but not in normal (E) larvae. Recovery was nearly complete upon return to room temperature.

Defects in septate junctions in Drosophila nrx IV mutants result in loss of the blood-brain barrier and the failure of action potential in high [K+] bathing solutions. Temperature-dependent failure of nerve conduction inaxoP12 larvae occurred even when larvae were bathed in low [K+] solutions, and failure was not exacerbated in high [K+] solutions. Thus, the blood-brain barrier was intact in axo mutants.

Taken together, the phenotypic and molecular properties of AXO suggest that it is a component of a glial-neuronal signaling pathway that is important in establishing the electrical properties of axonal membranes. We have previously shown that a temperature-sensitive defect in axonal conduction is conferred by mutations in structural or regulatory genes that result in reduced numbers of sodium channels (8). Moreover, kinesin mutations in Drosophilaalso impair action potential propagation as an apparent consequence of reduced sodium and potassium channel density in axonal membranes (19). Consequently, we hypothesize that anaxo-dependent signal is required for the normal expression, localization, or clustering of some set of ion channels. The aberrant expression or distribution of these channels in axo mutants would thus underlie the observed defect in nerve conduction at elevated temperatures. Recent work with vertebrate cell culture systems demonstrates that glia actively determine neuronal function by secreting factors that affect neuronal membrane excitability (1, 20). However, the identity of these factors is still unknown. AXO may be one such factor in Drosophila.

  • * Present address: Division of Neuroscience, Baylor College of Medicine, Houston, TX 77030, USA.

  • To whom correspondence should be addressed. E-mail: ganetzky{at}


View Abstract

Navigate This Article