Teleost fish RICH proteins are neuronal GAPs that are induced in retinal ganglion cells during optic nerve regeneration [44]. They are transported down the growing axon and they accumulate at varicosities and membrane blebs often associated with formation of branches [48]. They are homologous with mammalian CNPases [49, 50], sharing significant sequence homology on the C-terminal two-thirds region [44], and they display several functional properties in common, such as phosphodiesterase activity (2’,3’-cyclic nucleotide 3’-phosphodiesterase) [42, 44], membrane association (through prenylated C-terminus) [45], interaction with tubulin, and the ability to promote the branching of membranous extensions [46]. Interestingly, the genes encoding these proteins show significant differences in regulation of expression in the CNS. Mammalian CNPases are abundantly expressed in the mammalian brain and spinal cord, mainly due to very high levels of expression in glial cells involved in myelination (oligodendrocytes). Conversely, RICH proteins demonstrated moderate levels of expression in the teleost brain and retina; however, in the retina, the protein was strongly induced in the neurons that are in the process of axon regeneration [44]. The process of nerve regeneration is very complex, its success depending on a variety of both extrinsic factors and intrinsic neuronal GAPs that modulate the regeneration capabilities of the tissue [12, 13, 17–19]. Due to the difference in the regulation of expression between teleost RICH proteins and mammalian CNPases, RICH proteins could be an important contributing factor for the higher intrinsic nerve regeneration capacity of neurons in the CNS of cold-blooded vertebrates. These proteins would not be available to the mammalian neuronal counterparts after neurite damage, giving interest to studies aiming to understand the roles of RICH proteins during neurite regrowth.
The PC12 cell line has been utilized extensively to study the process of neuritogenesis at both the cellular and molecular levels. Several neuronal GAPs have been demonstrated to increase the structural plasticity of PC12 cells, such as GAP43, CAP23, c-Jun, ATF3, etc., with expression of the proteins enhancing neuritogenesis in response to NGF or by combination with other factors that trigger differentiation [51–53]. The expression of zRICH in stable transfectant PC12 cells also increased structural plasticity in response to NGF, although the effect occurred at later stages, by specifically promoting neurite branching [46]. Interestingly, a mutant version of the protein devoid of phosphodiesterase activity, termed zRICH(H334A), with a single amino acid substitution of a histidine located in the catalytic site, resulted in increased potency for enhancing neuritogenesis, suggesting a possible mechanism of self regulation and that this version of the protein could be used as a tool to enhance the intrinsic capacity of neurons for axon regeneration. These studies were limited by the need of fixing the cells and detecting expression through immunocytochemistry. To study the effects of this protein on neuritogenesis and nerve regeneration in further detail, new experimental procedures were designed that avoid the need for immunodetection of expression and facilitate the morphometric analysis of the differentiated cells. To be able to detect the expression levels of the zRICH(H334A) protein in living cells during the differentiation procedure, a eukaryotic plasmid was generated that encodes a fusion protein consisting of a monomeric version of RFP (DsRed-monomer) fused to the N-terminus of zRICH(H334A) (Fig. 1A). This design avoids interference with the C-terminal membrane localization motif. Stable transfectant PC12 cells were generated and demonstrated constitutive expression of the fusion protein by WB (Fig. 1B). Importantly, the protein allowed dynamic observation of expression under fluorescence microscopy and the capture of matching phase contrast and fluorescence microscopy images to facilitate the analysis of neuritogenesis (Fig. 2).
To study the effects of the expression of the fusion protein on neuritogenesis, a collection of images of differentiated neurons expressing zRICH were analyzed in detail by computer-assisted neurite tracing with the NeuronJ plugin of ImageJ image analysis software (Figs. 3A, 3B). The main effect of expression of RFP-zRICH(H334A) in PC12 cells during NGF induced neuritogenesis was an increase in neurite branching (approximately 2.4 fold increase versus control cells for the experiment shown in Fig. 3C). This effect matches well with previous results obtained with cells expressing unfused zRICH(H334A) [46], suggesting that the fusion of the fluorescent protein to the N-terminus did not block its function on promoting structural plasticity (although the magnitude of the effect detected was partially reduced). While the NeuronJ computer-assisted procedure facilitated the detailed analysis of effects on neuritogenesis, it still involves time-consuming tracing of neuritis [54]. A second morphometric procedure was developed based on previous studies demonstrating successful neuronal differentiation analysis by applying stereological methods to counting frames [55, 56]. In contrast to the NeuronJ method, this second procedure bypassed the need for neurite tracing, and was applied to random field images collected following a normalized and systematic procedure to avoid overlap and to represent the global neuronal population in the culture dish. Importantly, this method avoids time-consuming scanning for isolated neurons (Fig. 4A). ImageJ software was then used to count certain parameters on each image (Fig. 4B). Interestingly, the random field procedure was also able to detect the effect of RFP-zRICH(H334A) on neurite branching (detected as an increase in the Branching Points per Neuronal Cell Count parameter ratio, Fig. 4C). For a comparison of several characteristics of these methods, please see Supplementary Fig. 1 [Additional File 1].
Both the NeuronJ and random field procedures were able to detect changes in neurite morphology. While the random field procedure is more time-efficient, the differences in the two procedures could prove advantageous for specific experimental purposes, and the two methods can provide complementary information. For example, by comparing the results obtained with both methods, it is possible to discuss the effects of the protein at various levels of expression. The NeuronJ-based procedure can be applied to cells with specific levels of expression. For the experiments presented, cells with relatively high expression were analyzed (further subdivision was however not possible statistically as the number of cells matching requirements was limited). By utilizing pools of stable transfectants, the experiments avoided specific differences in the individual PC12 cells unrelated to the transfected gene. However, the pooled stable transfectants show characteristically wide variation of levels of expression for the introduced gene, which can span at least three orders of magnitude by flow cytometer analysis [57, 58]. Based on WB analyses, the average levels of protein expression in the PC12 stable transfectant pool exceed 0.5 ng of zRICH protein per µg of total cell protein (Fig. 1B), over 0.05% weight ratio, levels that would make it as abundant as some cytoskeleton-associated proteins [59, 60]. It can be speculated, based on estimates from the images obtained by fluorescence microscopy, that the NeuronJ analysis was performed with the neurons in the top 10% of levels of expression, possibly with levels higher than 10 fold above the average levels of the entire population. This would correspond to over 0.5% of the cellular protein weight ratio, becoming one of the most abundant proteins in the cell, probably close to the levels of tubulin itself. It is interesting to speculate, based on observations in previous WB analyses, that the levels of RICH expression in the stable transfectants could be similar to those in retinal ganglion cells in the process of nerve regeneration in zebrafish or goldfish (approximately 0.5-1 ng of RICH protein per µg of retina protein, where RGC are estimated to represent 5–10% of the retinal cells) [44]. The observed high levels of RICH expression during optic nerve regeneration, as well as previous functional analyses with RICH proteins and mammalian CNPases, suggested a role for these proteins in the regulation of the tubulin cytoskeleton and its interaction with the plasma membrane [45, 61]. The effects on neurite branching observed in the studies with PC12 cells expressing high levels of the RFP-zRICH(H334A) protein (Fig. 3C) are consistent with this hypothesis. On the other hand, the analysis of random fields allows an objective and relatively rapid analysis of a cell population by quantitating differentiation parameters on a per image basis. The results suggest that the effects on neurite branching observed with average levels of expression (Fig. 4C) match well with those observed with NeuronJ on cells with the highest levels of expression. Statistical comparisons of both variances and coefficients of variation for the effects detected with the two methods did not detect significant differences (F-tests, p > 0.05). Although not statistically significant, the classical analysis of selected cells with NeuronJ showed lower variation in the detection of the fold-effects on the parameters estimated (primary neurites, branching, arbor length). The procedure of analyzing random field images was used on a more diverse population of cells, probably explaining lower sensitivity to more moderate effects on parameters (such as for the Horizontal Grid Crosses, a parameter used to estimate neurite arbor length). Comparing effects of proteins at different levels on nerve regeneration can be important, to learn whether the effects do require very high levels of expression, or even if they differ at diverse levels. In several cases, for signaling molecules that are found at low levels in cells, artificial effects have been observed in experiments achieving high overexpression levels [62–64]. However, in this particular case, RICH and mammalian CNPases are expressed at fairly high levels in physiological systems, and the effects observed on the transfected PC12 neurons with the highest levels support a possible structural role on the tubulin cytoskeleton function during neuritogenesis. The experiments presented cannot tell whether the effects of the protein occur early during the generation of the branches from growth cones, or later by stabilization of the growing branches. Future studies analyzing video from continuous live microscopy could provide clues to answer this question.
Neurons obtained by differentiation of PC12 cells have been used previously to study neurite regrowth and regeneration [47, 65, 66]. When PC12 cells are differentiated with NGF, they become primed for rapid neurite regeneration through translational regulation and independent of new transcription [47]. Since the transfected cells can be analyzed for expression dynamically, without the need for immunodetection, they were used for neurite regeneration assays, by extending differentiation experiments to add a neurite injury protocol followed by a relatively short recovery period (Fig. 5A). The protocol was therefore designed to focus on regeneration rather than de novo differentiation. The expression of zRICH(H334A) promoted structural plasticity during neurite regeneration in PC12 cells by promoting a significant increase in branching (Fig. 5C).