Identification of Agrin in Electric Organ Extracts and Localization of Agrin-Like Molecules in Muscle and Central Nervous System

The basal lamina sheaths of muscle fibres in skeletal muscles survive trauma that causes the muscle fibres and the axons that innervate them to degenerate. New myofibres regenerate within the sheaths, axons grow to the original synaptic sites on them, and neuromuscular function is re-established. At the regenerated neuromuscular junctions, as at the original ones, the myofibre surface is characterized by high concentrations of acetylcholine receptors (AChRs) and acetylcholinesterase (AChE), both of which are crucial for synaptic transmission. Our studies on regenerating neuromuscular junctions in the frog have revealed that the synaptic portion of the myofibre basal lamina sheath contains molecules that direct the formation of AChR and AChE aggregates on regenerating myofibres (Burden, Sargent & McMahan, 1979; McMahan & Slater, 1984; Anglister & McMahan, 1985). Among the questions raised by this observation are the following. How are the synaptic organizing molecules in the basal lamina regulated? What is their mechan-Jm of action? Do these molecules play a role in the long-term maintenance of the neuromuscular junction as well as in restoring neuromuscular function after trauma? Are these the same molecules that mediate the formation of AChR and AChE aggregates at developing neuromuscular junctions in the embryo?

As a step towards answering these questions we have undertaken a series of studies aimed at identifying the synaptic organizing molecules at the neuromuscular junction and making specific markers for them. We selected as a source of such molecules the electric organ of Torpedo californica, which has a concentration of cholinergic synapses far greater than that of skeletal muscle. In previous reports (Godfrey et al. 1984; Wallace et al. 1985; Wallace, 1986; Magill et al. 1986) we documented that basal lamina-containing fractions of the electric organ are enriched for molecules, called agrin, that cause the formation of patches on cultured myotubes at which AChRs and AChE are aggregated. We demonstrated further that similar molecules can be extracted from muscle (Godfrey et al. 1984), but in much smaller amounts, and that monoclonal antibodies against agrin recognize molecules concentrated in the synaptic cleft at the neuromuscular junction (Fallon et al. 1985). These findings led to the agrin hypothesis: agrin is similar to the AChR- and AChE-aggregating molecules in the synaptic basal lamina (Fallon et al. 1985; Wallace et al. 1985). Here we present a brief account of new studies that have led to the identification and purification of agrin and have revealed that agrin-like molecules in the synaptic cleft at the neuromuscular junction are components of the basal lamina, which is consistent with the agrin hypothesis. We also present evidence that the cell bodies of motor neurones contain agrin-like molecules, which raises the possibility that the AChR- and AChE-aggregating molecules in the synaptic basal lamina are produced by the presynaptic cellular component of the neuromuscular junction.

In a previous report (Nitkin et al. 1983) we showed that agrin activity can be purified several thousand-fold by routine gel filtration and ion exchange chromatography. To identify and further purify agrin we have used immunoaffinity techniques.

We made monoclonal antibodies by immunizing mice with a partially purified preparation of agrin and screened the resulting hybridomas for those secreting antibodies that immunoprecipitated agrin activity (Fallon et al. 1985). At least seven of our anti-agrin antibodies recognize different epitopes (e.g. Magill et al. 1987). We found that at least four of the seven antibodies immunoprecipitated polypeptides of relative molecular masses (M_r) 70, 95, 135 and 150K (M_r 1000 = K) from our electric organ extracts (Fig. 1). Each of the monoclonal antibodies that immunoprecipitated native agrin was tested by immunoblot analysis for its ability to bind to denatured agrin. Specific labelling was detected for one of the monoclonal antibodies; it again bound to polypeptides at 70, 95, 135 and 150K (data not shown; but see Fig. 2).

To learn which of the polypeptides recognized by the antibodies had agrin activity we combined gel filtration column chromatography with immunoblot analysis. As shown in Fig. 2, the 150K and 95K polypeptides comigrate with the AChR/AChE-aggregating activity. Thus our antibodies recognize in electric organ extracts two forms of agrin plus two polypeptides which do not have AChR/AChE-aggregating activity.

We have now purified microgram amounts of agrin and the other related polypeptides by employing immunoaffinity techniques. Isolation of each of the polypeptides by preparative polyacrylamide gel electrophoresis has permitted analysis of their N-terminal amino acid sequence. Our aim is to develop probes which will enable us to use molecular genetic techniques to examine such questions as how the agrin polypeptides are related to each other, where and when they are synthesized, and if there are related peptides in other tissues or in muscles of other animal species.

Three polypeptides that cause AChR aggregation on cultured myotubes have been identified in neural tissues by other workers. Any or all of these molecules may play a role in the formation of AChR aggregates at the neuromuscular junction. They are a 42K polypeptide extracted from chick brain (Usdin & Fischbach, 1986), an 84K polypeptide, called sciatin, which is extracted from sciatic nerves and is virtually identical to transferrin (e.g. Oh & Markelonis, 1982), and calcitonin gene-related peptide (CGRP) which is 23K and is present in motor neurones (e.g. New & Mudge, 1986). The relative molecular mass of each clearly differs from that of the two forms of agrin. Moreover, agrin has little, if any, effect on the rate of receptor insertion into the myotube plasma membrane while the others have a profound effect in this regard. Thus, agrin is distinct from any of these molecules.

To determine whether agrin-like molecules in the synaptic cleft at the neuromuscular junction are components of the basal lamina we first damaged the frog’s cutaneous pectoris muscle by crushing it (Anglister & McMahan, 1985; McMahan & Slater, 1984). This resulted in degeneration and phagocytosis of all cellular components of the neuromuscular junctions - myofibres, axon terminals and Schwann cells - while leaving intact much of the myofibres’ basal lamina sheaths. Regeneration of myofibres and axons was prevented (McMahan & Slater, 1984). Three weeks after damage, the ‘muscle’ was removed from the frog and the empty basal lamina sheaths were incubated with anti-agrin antibodies. As shown in Fig. 3, the antibodies stained the synaptic portion of the myofibre basal lamina intensely.

Thus, molecules antigenically similar to agrin are concentrated in the synaptic basal lamina and these molecules remain adherent to the basal lamina for several days after muscle degeneration, as do the basal lamina’s AChR- and AChE-aggregating molecules (McMahan & Slater, 1984).

The synaptic basal lamina is not the only structure in muscle stained by anti-agrin antibodies. Fig. 3 illustrates that the antibodies also stain Schwann cell basal lamina. In addition, although the extrajunctional surface of twitch muscle fibres did not stain, the antibodies stained the extrajunctional surface of slow skeletal muscle fibres, but at a much lower level than at the neuromuscular junction. In certain species the antibodies stained the surface of smooth muscle fibres in the walls of blood vessels. The surface of smooth muscle fibres, like that of skeletal muscle fibres, is coated with basal lamina. Thus, agrin may be antigenically related to one or more basal lamina molecules that have a broad distribution in muscle.

One likely source of the AChR- and AChE-aggregating molecules in the synaptic basal lamina is the axon terminal. If the axon terminal were the source and if agrin were related to the AChR- and AChE-aggregating molecules in the synaptic basal lamina, one might expect to detect agrin in extracts of CNS regions rich in the cell bodies of motor neurones. As illustrated in Fig. 4, we indeed found that extracts of the electric lobe of the Torpedo brain, the region of the brain occupied by the cell bodies of the motor neurones that innervate the electric organ, contained molecules that caused the formation of AChR and AChE aggregates on cultured myotubes. The active molecules were immunoprecipitated with anti-agrin antibodies, indicating that they are similar, if not identical, to agrin. We have now detected functionally and antigenically similar molecules in extracts of spinal cord, another region rich in the cell bodies of motor neurones, from Torpedo, frog and chick.

We stained the electric lobe and spinal cords with the anti-agrin antibodies to determine the distribution of agrin-like molecules. In all cases the cell bodies of motor neurones stained (Fig. 4). The stain was distributed in patches in the cytoplasm, indicating that it is associated with organelles. No other neurones in the electric lobe and spinal cord (lumbosacral region) stained; thus, among neurone cell bodies in these regions, agrin-like molecules are selectively localized to those of motor neurones. Stain was also associated with capillaries and pia mater, which have a basal lamina. We are currently performing experiments to determine whether the agrin-like molecules in the motor neurones have agrin activity and whether these molecules are released at axon terminals to become incorporated into the synaptic basal lamina.