It is common practice to demonstrate primary cilia using immunohistochemical staining for ACIII [22, 30], and on this basis we conclude that over-expression of a disease-causing SOD1 mutation significantly reduces the expression of primary cilia on motor neurons late in primary culture and in the lumbar spinal cord of adult animals in situ. However, it might be more conservative to conclude that there is a reduction in ACIII cilial staining, as it is theoretically possible that the ACIII staining has disappeared from an otherwise intact primary cilium. There could be important consequences in either event, and moreover, since our study focused on immunohistochemical alterations only, the potential importance of these findings would be broadened if there were additional deficits in cilial function in motor neurons still retaining ACIII staining.
Some technical issues bear discussion. Because of restrictions imposed by antibody suitability, species, and the availability of secondary stains, we chose to identify motor neurons indirectly. Usual stains for motor neurons, eg SMI32 or ChAT, result in a stain that is diffused and poorly suited to counterstaining for cilia. In consequence, we used an indirect method for identifying ciliated motor neurons. We first verified that nearly all large neurons (greater than 20 μm) in the anterior horn co-express ChAT and the neuronal stain NeuN, and then used NeuN, which is better suited to the identification of cilia, and size greater than 20 μm, as a surrogate for direct identification of motor neurons. The indirect identification of motor neurons could introduce error if some motor neurons in the hmSOD1 mice were pyknotic (and thereby not identified as motor neurons by size). However, there were very few NeuN positive cells less than 20 μm in lamina IX in either WT or G93A SOD1 mice, and thus pyknotic motor neurons would have to have lost their NeuN staining (which has been reported in neurons in an experimental stroke model ).
We used 40 μm sections and only counted cells where the nucleus could be identified in totality by imaging at multiple planes. It is possible that a primary cilium was present on some cells but out of section, and the cell thus misidentified as lacking a primary cilium. This error would not be large, and indeed, if there were somatic shrinkage with disease, it would bias against finding a difference in G93A SOD1 mice.
Even in diseased animals, some neurons appeared healthy and had normal appearing primary cilia, and thus there was considerable variability in cilial staining. This variability raises the possibility that the disease process is patchy with respect to cilial pathology, that there are differences in susceptibility between types of motor neurons, or that there is a differential expression of hmSOD1 between individual motor neurons in this model. (There is some evidence that the transgene copy number may not be fully stable in G93A SOD1 mice  but nothing is known of cellular mosaicism.)
In spite of these caveats, our cell culture results and our animal studies are mutually consistent, and consistent with the hypothesis that there is a reduction in the proportion of ciliated motor neurons in G93A SOD1 mice. Neurotoxic SOD1 mutations could affect primary cilial structure and function in several ways. There might be intra-ciliary protein accretion and/or altered intraflagellar transport, and in this light, mSOD1-induced changes in the normal anatomical barriers to the exchange of proteins between cytosolic and ciliary compartments might be important (ie the ciliary necklace and transitional fibers ).
A major unanswered question that will await further experimentation is whether the reduction in ciliated motor neurons simply reflects the underlying diseased state of these cells, or whether the loss may be more directly and causally related to the ALS disease process. There are several reasons to believe that a loss of primary cilia could be detrimental. Most directly related to this work, the loss of ACIII would be expected to disrupt cAMP second messenger signaling, in this case from an unknown G protein coupled receptor and unknown ligand .
Moreover, much, indeed perhaps all, Shh signaling occurs through the primary cilium, and disruption of the primary cilium alters the balance between canonical and non-canonical Wnt pathways . Both Shh and Wnt have been shown to exert neuroprotective function [17–20, 34, 35]. Similarly, other signaling pathways involve the primary cilium and may be neuroprotective, including MCH1, SSTR3, and PDGF [36–38]. Loss of cilial signaling (or reduced cilial signaling due to reduced cilial stability and increased cilial turnover) could render motor neurons more susceptible to neurotoxic challenge.
Last, the primary cilium derives from the maternal kinetophore, is expressed only on growth-arrested cells  (possibly for this reason), and perhaps not surprisingly many cilial signaling pathways have been implicated in cell cycle control. Specifically, Shh, Wnt, PDGF, and SSTR3 have all been implicated in cell cycle control [39–43]. Disruption of primary cilial function might alter the response of post-mitotic motor neurons to disease-induced cell cycle re-entry signals [44, 45], with possible adverse consequences .