From the outset I hoped that what we learned about normal and injured spinal motor neurons might be useful in understanding diseased motor neurons. This hope was based in part upon the expectation that the more complete our inventory of the structural and molecular components of normal motor neurons, the more likely we can completely define the qualitative and quantitative changes in diseased motor neurons. That information would greatly aid our understanding of how diseased motor neurons die and could help us understand the causes of motor neuron diseases. This document is but one indication of how far away we still are from a comprehensive understanding of normal spinal motor neurons. On the other hand, new and highly sensitive methods of compositional analysis, including proteomic and genetic methods that can be applied to isolated motor neurons, show promise for greatly advancing our understanding of the molecular composition of these and other classes of neurons.

Another way in which our data may relate to motor neuron disease involves our characterization of the properties of the cell body cytoskeleton and its role in the axon reaction. Adult motor neurons can die if axonal transport is interrupted near the cell body and some spinal motor neurons in ALS patients and in animal models of motor neuron disease show signs of such interruption. Those signs include axonal swellings near the motor neuron cell body in ALS (Delisle and Carpenter, 1984), morphological changes in cell bodies typical of axonal interruption (Nakano and Hirano, 1987), and decreased axonal transport in animal models of motor neuron disease (Williamson and Cleveland, 1999).

Our studies on size-dependent differences in motor neurons could also eventually help to explain some aspects of ALS, since, as discussed below , large neurons may be most vulnerable to motor neuron disease (McIlwain, 1991). Moreover, we have applied many of the experimental techniques we developed for normal frog and rodent motor neurons to human spinal tissue and have gained some insight into the composition and structure of normal and diseased human motor neurons (e.g., ). Nonetheless, it is disappointing that, at the close of our work, we still do not know whether any of the information we have gained from normal and injured cells will significantly increase our understanding of motor neuron disease.

In addition to the research cited above, our laboratory has used post mortem human spinal tissue, generously donated to us by ALS patients and their families, to ask direct questions about diseased motor neurons. We have made limited progress in answering some of those questions.

A. Motor neurons in ALS

1. Are sensory nerves affected by ALS?

Despite the predominance of motor signs in the clinical presentation of ALS, there is ample evidence for subtle sensory abnormalities in this disease. A number of electrophysiological studies have reported evidence for sensory dysfunction (e.g., Radtke et al., 1986, Shefner et al., 1991, Theys et al., 1999). In addition, morphological and biochemical abnormalities have been found in sensory nerves in ALS. In a biochemical study of the spinal roots from patients with sporadic ALS, we found evidence for abnormal proteins within the dorsal roots, as well as the ventral roots in one-half of the ALS specimens examined (Brock and McIlwain, 1984). The abnormality involved the presence of glial fibrillary acidic protein (GFAP) and its breakdown products, which are derived from astrocytes and are not found in normal spinal roots. Since other investigators had previously reported morphological evidence for the presence of astrocytic intrusions called “glial bundles” within the dorsal and ventral roots of some patients with sporadic ALS (Ohama et al., 1981; Ghatak and Nochlin, 1982), we concluded that glial bundles were the source of the GFAP and its breakdown products that we observed on electrophoretic gels. Axonal atrophy and loss are likely associated with astrocytic extension into the spinal roots (Brock and McIlwain, 1984), since morphological evidence already existed for the preferential loss of large sensory axons in the dorsal root (Kawamura et al., 1981) and in peripheral nerve (Dyck et al., 1975) in sporadic ALS.  

The loss of other large non-motor neuronal cells, including large neurons in Clarke’s nucleus (Averback and Crocker, 1982) and spinal border cells (Williams et al., 1990), has also been implicated in ALS. It is possible that large neurons of any type – i.e., neurons with large cell body and axonal volumes – are more vulnerable to ALS than smaller neurons.

2. Are large motor neurons more vulnerable to ALS than small ones?

Pathological and clinical studies on patients with ALS, as well as on animal models of ALS (Feeney et al., 2001; Chiu et al., 1995; Zang and Cheema, 2002), indicated that motor neurons are progressively lost during the course of the disease. The rate of loss of motor neurons of different sizes may not be uniform. There is suggestive evidence from a variety of studies on motor neuron disease in human beings (Tsukagoshi et al.,1979; Oyanagi, 1980; Kawamura et al., 1981; Sobue et al.,1987) and in some animal models (Cork et al.,1989) that the largest spinal motor neurons are the first to be lost as the disease progresses. Our own laboratory showed that lumbar motor neurons with cell body areas exceeding 3000 mm2 represented only 19% of the total lumbar motor neurons isolated from patients with sporadic ALS, a marked decrease from normal lumbar spinal cord, where 69% of all isolated motor neurons exceeded that size (McIlwain, 1991).

At least three possible explanations can be given for a disproportionate, early decrease in the number of large motor neurons in motor neuron disease: 1) large motor neurons die faster than small ones; 2) large motor neurons atrophy earlier than small ones; 3) motor neurons never reached their maximal size before the disease began. There is no reason to favor the third possibility, particularly since studies on animal models of familial ALS have confirmed that prior to disease onset, spinal motor neurons attain their normal adult sizes ( Chiu et al., 1995; Feeney et al., 2001). It also seems clear that many intermediate- and small-sized motor neurons are lost at some stage as ALS progresses (e.g., Kawamura et al., 1981). If atrophy of large neurons explains the loss of large motor neurons in ALS, one would expect to find occasional examples of patients in the mid-course of the disease who have decreased numbers of large motor neurons, but increased numbers of smaller motor neurons. Such examples were found in studies of ALS patients by Kawamura et al. (1981) and Bergeron et al. (1994).

Our laboratory suspects that large neurons are more susceptible to ALS, based upon the evidence cited above that large size in both motor neurons and non-motor neurons appears to correlate with increased vulnerability to motor neuron disease.

3. Can one identify “sick” motor neurons isolated from ALS spinal cord?

The succinct answer to this question is that we were unable to do so, but continue to believe it is possible. Based upon many histopathological studies, some motor neurons in ALS are clearly more affected than others, but it is unknown how many, if any, might be normal.

Once we had developed our procedure for isolating individual motor neuron cell bodies, we searched for obviously impaired motor neurons, so that we could collect and compare them with normal-appearing cells from the same individual or from control tissue. We sought ways to identify affected motor neurons by light microscopy, exposing them to as few additional reagents and manipulations as possible. However, we were unable to detect consistently any morphological abnormalities in cell bodies isolated from ALS tissue. Occasionally, a sclerotic, hyaline-filled cell body was identified, but the vast majority of the isolated motor neurons were indistinguishable from normal cells. Simple stains, such as basophilic dyes for RNA, did not reveal reproducible differences among the isolated cell bodies. Most of the available immunohistochemical methods were impractical, because they required multiple manipulations of the cell suspensions.

Nonetheless, we believe it may ultimately be possible to achieve this goal, using the expanding number of cytological markers for ALS-affected motor neurons and newer, highly sensitive analytical methods, such as mass spectrometry.  

4. Are there significant differences in the protein composition of motor neuron cell bodies isolated from normal and ALS spinal cord?

This was the question about ALS we most wanted to answer. Lacking a method for narrowing our analyses of ALS motor neurons to clearly affected cells, we resorted to comparisons between all cell bodies isolated from ALS and normal spinal tissue. The analytical method we used was two-dimensional polyacrylamide gel electrophoresis (2-D PAGE). In one experiment performed in 1985, we isolated 452 cell bodies from the lumbar spinal cord of a 75 year-old female succumbing to sporadic ALS and 611 cell bodies from the lumbar spinal cord of a 76 year-old female who died of non-neurological causes. Spinal tissue from both individuals was obtained 2-3 hours after death. Isolated cell bodies were transferred to first dimension gels in O’Farrell’s lysis buffer + 1% SDS and their proteins resolved by 2-D PAGE. No qualitative differences were found among approximately 85 silver-stained protein spots on the ALS and control gels. In a second experiment, again no qualitative differences were found in the cell body protein patterns from a 54 year-old male ALS patient and a control subject. At the time of these experiments, we were not equipped to search for quantitative differences between ALS and control gels. In each experiment, we also looked for protein differences in gels of ALS and control cell bodies that localized at two different densities on the discontinuous sucrose gradient (the 0.9/1.2M interface and the 1.2/2.5M interface). No qualitative differences were detected in the protein patterns of cell bodies found at the two different densities.

These negative results helped us to see more clearly the difficulties we faced with this approach. Since we were unsure what fraction of cell bodies isolated from ALS tissue were abnormal, the negative results could mean that we were comparing mostly normal, surviving motor neurons from ALS tissue with normal cells from the control tissue. On the other hand, if abnormalities did exist on the gels, then our chances for obtaining useful results rested upon detecting quantitative differences between ALS and control gels and identifying the proteins whose content was altered. However, the identities of only a few major spots were known to us, because these experiments were done well before mass spectrometric analyses of spots on 2-D gels became available generally. The odds of finding a meaningful, reproducible quantitative difference and then identifying the relevant protein were not favorable. An additional factor that led us to suspend these experiments was an indication from all of our electrophoretic studies on isolated cell bodies from both frogs and human beings that a substantial fraction of the total cell body protein did not appear on the gels. This possibility was inferred from experiments in which 2-D gels loaded with known amounts of protein from spinal cord homogenates were compared with gels of isolated cell bodies containing the same amount of protein. The cell body gels always were more lightly stained and had fewer spots on them. Most of this difference may be a consequence of the high fraction of insoluble protein in the isolated cell bodies, especially when exposed to methylene blue , and may also involve the absence on the gels of proteins with isoelectric points or molecular weights outside the range of our electrophoretic conditions. If cytoskeletal abnormalities are essential features of some motor neuron diseases, then the development of methods to analyze insoluble proteins in motor neurons could become critical to our understanding of these disorders.

Current technologies now offer much brighter prospects for detecting qualitative and quantitative abnormalities in proteins of ALS motor neuron cell bodies. Genomic methods now permit one to quantify and compare the expression of particular genes in normal and abnormal cells. New proteomic techniques for quantifying protein staining on electrophoretic gels and identifying extremely small amounts of unknown proteins can focus comparisons directly upon proteins in normal and abnormal cells. By applying these new methodologies to motor neuron cell bodies isolated from control and ALS spinal cord, one may now be able to accomplish the goals we were unable to reach with the methods available to us.

5. Does the protein composition of spinal ventral gray matter from ALS and control patients differ significantly?

Glial fibrillary acidic protein (GFAP) and its breakdown products dominate the Coomassie blue-stained protein patterns on 2-D gels of water-soluble proteins extracted from the human ventral gray matter of lumbar or cervical spinal cord . This was found to be true of samples from both sporadic ALS and control tissue. The breakdown products were less obvious on silver-stained gels than on gels stained with Coomassie blue. On gels where 100 or more proteins were resolved, minor spots could be identified on ALS gels that were not present on control gels and vice versa, but these differences were not reproducible from gel to gel. We suspect that these differences may have been a consequence of slight variations in the total amount of protein on different gels, which particularly affects the detection of the weakly stained spots.

Similar, copious quantities of GFAP and its breakdown products were seen by us in the ventral gray matter of an infant who died of Werdnig-Hoffman disease, a motor neuron disease of infants. We do not know whether this is also true of control tissue from human infants. GFAP and its breakdown products were also found in much lower amounts in the ventral gray matter of the cow (Brock and McIlwain, 1985). The breakdown of GFAP probably occurs mainly after death, since the breakdown products were more abundant in bovine ventral gray matter that had remained at room temperature for 24 hours than in tissue from the same animal that was chilled immediately. If so, these products might be useful as a measure of tissue change due to post mortem delay. However, we have shown that many of the proteins in ventral gray matter and in isolated spinal motor neurons are relatively stable for at least 24 hours after death .

B. Skin changes in ALS

In 1991 my laboratory began to work with Dr. Seiitsu Ono of the Teikyo University School of Medicine and Dr. Mitsuo Yamauchi of the UNC School of Dentistry on two issues related to their research on skin changes in sporadic ALS. The first issue involved the development of a biochemical test for the diagnosis of sporadic ALS early in the disease. The second issue was how the skin changes in ALS are related to the changes occurring within the nervous system. Over the course of the next three years, we explored each of these two issues.

1. Can collagen breakdown products be used for early diagnosis of ALS?

Ono and Yamauchi (1992) had found very early, progressive changes in type I collagen cross-linking in the skin of ALS patients that we believed could form the basis of a laboratory diagnostic test for ALS. Even now, there exists no biochemical test for sporadic ALS that might aid in the rapid diagnosis of the disease. Such a test would not only help to relieve the uncertainty and expense often associated with making a definitive diagnosis of the disease, but eventually may be essential for diagnosing the disease early enough to minimize the loss of motor neurons, once an effective treatment of sporadic ALS is discovered.

We performed trial experiments on the urine of ALS patients to look for alterations in the content of a fragment of type I collagen that is progressively lost from the skin in ALS. The fragment, called histidinohydroxylysinonorleucine (HHL), contains a crosslink found in mature type I collagen. Its accelerated loss from the skin in ALS could result from either an increase in the degradation or a decrease in the synthesis of mature, HHL-containing type I collagen. Because very low amounts of HHL had to be detected, we chose to examine urine, which normally contains little or no protein, rather than blood, which is easier to collect, but contains large amounts of non-HHL protein.

Seven ALS patients in a local ALS support group donated urine samples, with seven family members serving as controls. Rather than collecting 24-hour urine samples, we normalized the HHL concentration to the creatinine concentration of each sample. Creatinine has a disadvantage that we recognized: i.e., it is made by muscle and is excreted in larger amounts as skeletal muscles atrophy. However, if the changes in HHL excretion in ALS were much larger than those in creatinine, HHL could become a useful indicator of ALS. If the loss of HHL from the skin were caused by accelerated breakdown of mature type I collagen, we predicted that the ratio of HHL to creatinine excretion within the first 1-2 years of the disease would increase if the increase in HHL excretion exceeded that of creatinine excretion. On the other hand, if inhibition of HHL synthesis caused the decrease in its content in ALS, we expected the ratio of HHL to creatinine in the urine to decline early in the course of the disease, as HHL was depleted from the skin and replaced at a slower rate.

In the first of two separate analyses of the urine samples, which were stored frozen until use, we found that the two ALS patients with the shortest disease duration (approx. 1.5 years) had higher ratios of HHL to creatinine than patients with disease durations of 2-14 years . This suggested that mature type I collagen in ALS skin was being broken down faster than normal early in the disease. In a repeat analysis of the same urine samples, one of those two patients again showed a high HHL to creatinine ratio, while the second patient had a normal ratio. This variability was also encountered in the analysis of the other ALS samples, as well as with the control samples. When the ratios of individual ALS patients were compared to their family member’s values or to an overall average of the control group ratio, no significant difference was observed between the ALS patients and the controls.

The upshot of these pilot experiments was that changes in HHL excretion in the urine were not likely to be large enough and reproducible enough to warrant a large-scale project. It is still possible that at earlier stages of ALS than we studied, HHL excretion is high enough to be useful as a biochemical indicator of the disease. Since these experiments were done, Ono, Yamauchi and their colleagues have shown that it is possible to detect changes of the 7S fragment of type IV collagen in the serum of ALS patients (Ono et al. 1998). This finding encourages the hope that an early diagnostic test for ALS may yet be developed that exploits our knowledge of the skin changes that occur in ALS.  

2. Are there protein cross-link abnormalities in motor neurons in ALS?

If we knew the relationship between the changes in the nervous system in ALS and the changes in the skin, we might understand more clearly the cause of motor neuron death in ALS. For example, collagen is present in ventral roots and spinal nerves (e.g., Shellswell et al., 1979). Since collagen likely protects the mechanical integrity of spinal roots and nerves that are often subjected to extraordinarily wide range of movements over many years, collagen abnormalities in spinal nerve collagen might cause significant damage over time in lower, if not upper motor neurons from mechanical stress to motor axons.

Another possible clue to an association between skin collagen and motor neurons relates to similarities in agents that can disrupt the organization of collagen in the skin and neurofilaments in motor neurons. Two lathyrogenic compounds, beta-aminoproprionitrile (BAPN) and beta, beta'-iminiodiproprionitrile (IDPN), which can interfere with the formation of mature collagen, can also produce motor axonal spheroids similar to those often seen in ALS. We formulated an hypothesis that similar kinds of cross-links may exist in neurofilaments in motor neurons and in collagen, both of which are very stable, long-lived proteins. This hypothesis was submitted to The Lancet, but was declined by that journal and was never published. It is included here in the hope that it may contain something of use to those who continue to explore the possible involvement of lathyrogens, age-dependent changes in the mechanical properties of spinal roots and nerves, disruption of axonal transport, and copper-requiring enzymes in the pathogenesis of motor neuron diseases .