10.1 Introduction

10.1.1 Features of Axon Development

The arrangement of retinal neurons in the brain reflects that of the light-sensitive cells in the retina and, ultimately, the visual world. During embryonic development, independent of the birthplace in the retina, retinal ganglion cell (RGC) axons extend in the direction of the optic nerve head, where they collect to exit the eye and form the optic nerve. In vertebrates, once past the midline optic chiasm in the ventral diencephalon, retinal ganglion cell axons grow to the optic tectum, their most prominent synaptic target in the midbrain, and arborize in a topographic array that, in essence, copies the spatial map in the retina onto the brain. Likewise, several other neuronal projections are concurrently established in the embryonic brain; so, how do axons succeed in finding their way?

Observations in vivo of developing axonal projections have discovered that their growth is highly directed, with axons navigating along a prescribed trajectory en route to their respective synaptic targets and making very few errors of navigation in the process (Crossland et al. 1974; Lance-Jones and Landmesser 1981; Holt and Harris 1983; Raper et al. 1983). This remarkable pathfinding fidelity depends on successive spatial signals—guidance cues—presented in the embryonic landscape and integrated by the growth cone, a sensory, and motile structure at the tip of developing axons. Axon trajectories are thus seemingly divided into shorter segments in such a way that the effort of navigating towards a distant target is reduced to the simpler task of reaching consecutive intermediate points.

Four evolutionary conserved families of signaling molecules that function as instructive—chemotactic—guidance cues are classically described for their widespread roles in axon guidance: netrins, semaphorins, slits, and ephrins. In addition to these chemical signals, growth cones are also instructed by cell–cell and cell–matrix physical adhesions that provide not only an effective roadmap for navigation but also an essential platform for the protrusive behavior of the growth cone (Lowery and Van Vactor 2009; O’Donnell et al. 2009). These contacts can be mediated by members of the integrin, cadherin, and, most prominently, immunoglobulin (IgG) superfamilies. Importantly, the actions of these various signals are not mutually exclusive, but rather coordinately act to ensure that axon navigation ensues unerringly. Indeed, the involvement of various instructive signals even along a short trajectory considerably diminishes the likelihood of guidance errors and promotes the necessary fidelity in the establishment of neuronal connections.

10.1.2 Axonal mRNA Localization: One in Thousands

The synapse, the structure formed by two communicating neurons, underlies one of the most striking features of nerve cells: the extreme cellular and molecular polarization of axons and dendrites (Barnes and Polleux 2009). Necessarily asymmetric in function, these two compartments receive an independent assortment of organelles, membrane components, and molecules from the cell body (Hirokawa and Takemura 2005). Subcellular RNA localization has emerged as a particularly prevalent and cost-efficient mechanism of outsourcing genomic information in these highly polarized cells where the site of transcription can be far removed from the final destination of the protein (Jung et al. 2012, 2014). Mechanistically, specific transcripts can be precisely localized to subcellular compartments using the “address” information harbored in their untranslated regions (UTRs), which function as cis-acting platforms for regulatory RNA-binding proteins (RBPs) and small noncoding RNAs (Andreassi and Riccio 2009). Subsequently, local, ‘on-site’ synthesis confers both spatial and temporal precision, as the new protein is present only where and when a biological demand for it exists (Fig. 10.1). It is this mRNA-based mechanism that, for example, allows the growth cone to enjoy a certain degree of functional autonomy in its guidance process (Jung et al. 2012).

Fig. 10.1
figure 1

Axonal mRNA localization and local protein synthesis. (A) Subcellular targeting of specific mRNAs depends on the recognition of localization barcodes by nuclear and cytoplasmic trans-acting factors (TAFs), which collectively associate as part of higher-order messenger ribonucleoprotein (mRNP) complexes. Most of the axon-targeting elements that have been identified are situated in the 3′UTR of the mRNAs and are decoded by various TAFs operating synchronously. Some functionally related transcripts share similar axon-targeting motifs and are regulated by common sets of TAFs, a property that allows these messages to be translated simultaneously with temporal and spatial precision. (B1) Upon recruiting additional adapter proteins (not depicted), mRNPs are shipped along cytoskeletal tracts by motor-driven active transport mechanisms towards their subcellular destination. Notably, mRNAs are maintained in a translationally dormant state during the assembly and transport phases. (B2) By modulating the activation of mTORC1 signaling and, in parallel, eliciting changes in the binding affinity of specific TAFs, various local stimuli, including guidance cues, can bring about concerted alterations in gene expression programs

Full size image

Still, it was not until the recent appreciation of the complexity of the axonal transcriptome—several independent genome-wide screens have identified thousands of mRNAs localizing in the axonal compartment of embryonic and adult neuronal cells (Taylor et al. 2009; Andreassi et al. 2010; Zivraj et al. 2010; Gumy et al. 2011)—that the functional significance of this cellular mechanism was fully grasped. Indeed, the prevailing view at the turn of the century was that axons did not synthesize proteins but that instead the entire proteome in this compartment was maintained by a constant provision of proteins synthesized in the cell body and transported along the axon (Jung et al. 2012; Lasek et al. 1973). Significant evidence linking local mRNA translation to many aspects of axonal biology has since overwhelmingly dismissed this notion. It is now known that axonal mRNA translation regulates not only growth cone guidance decisions (Campbell and Holt 2001; Brittis et al. 2002; Wu et al. 2005; Leung et al. 2006, 2013), where its involvement was originally studied, but also axon elongation (van Kesteren et al. 2006; Hengst et al. 2009), axon maintenance and degeneration (Andreassi et al. 2010; Cox et al. 2008; Yoon et al. 2012; Baleriola et al. 2014), as well as nerve injury and axon regeneration responses (Verma et al. 2005; Ben-Yaakov et al. 2012; Perry et al. 2012), among other processes (Jung et al. 2012).

In turn, the impact of local protein synthesis in such diverse cellular mechanisms underlines a crucial aspect of axonal RNA localization: the dynamic nature of the local transcriptome. Indeed, even within the same population of neurons, comparative profiling of two different developmental stages has revealed that axons contain ‘age’-specific mRNA pools. For example, mRNAs encoding synaptic vesicle proteins, which intuition suggests being irrelevant during the pathfinding stages, are only found in target-arrived axons (Zivraj et al. 2010). It is noteworthy, however, that all of the axonal populations analyzed to date appear to have a common core of transcripts, such as those encoding mitochondrial and ribosomal proteins (Jung et al. 2012), suggesting that these molecules are implicated in ‘everyday’ axon upkeep. This also seems to be the case with Tctp: its transcript is ranked among the most enriched in the axonal compartment of diverse embryonic and adult neuronal populations (Table 10.1), which indicates that Tctp has a constitutive axonal function.

Table 10.1 tctp is a highly enriched axonal mRNA. Although initial evidence put forward the idea that tctp was a modestly expressed gene in the brain (Thiele et al. 2000), recent genome-wide RNA profiling efforts have identified its mRNA among the most abundant axonal transcripts in both embryonic and adult neuronal populations.
Full size table

10.1.3 Axon Guidance and Cancer: Shared Features

The parallels between the processes of axon guidance and cancer cell invasion hinted that Tctp, a protein associated with malignancy, could play a particularly important role during the wiring of neuronal circuits (Box 10.1). Indeed, from the continuous changes in motility and adhesion, or the crosstalk with the surrounding environment, the challenges faced by a metastatic cell echo those overcome by a pathfinding growth cone as it navigates through the developing brain.

Box 10.1 Neuronal Connectivity and Cancer Metastasis: Historical Parallels?

Historically, the neuroscience field debated two explanatory hypotheses regarding the wiring of the nervous system. The “resonance theory” explained the developmental patterning of the central nerve tracts on a purely mechanical basis, by schemes of initially nonselective growth that, based on the validity of the connection formed, were later maintained or eliminated (Meyer 1998; Weiss 1936). A second framework proposed that selective chemical or electrical forces guided neuronal connections and found initial support in the experiments of John Langley in the late ninetieth century (Langley 1895). The extensive studies of Roger Sperry on how regenerating frog retinal ganglion cell axons are arranged when reinnervating their target categorically proved the latter hypothesis (Sperry 1943, 1944, 1963). In his most dramatic experiment, Sperry rotated the eye 180° on its dorsoventral axis after severing the optic nerve and noted that it lead to the animal having inverted vision; that is, the axons were originating from reversed positions in the eye yet managing to find their appropriate synaptic connections in the brain. He concluded that “the cells and fibers of the brain and cord must carry some kind of individual identification tags, presumably cytochemical in nature, by which they are distinguished one from another almost, in many regions, to the level of the single neuron” (Sperry 1963), a molecular view of the structuring of the nervous system which remains largely unchallenged to date (Zipursky and Sanes 2010).

This idea resonates with the seminal work of Stephen Paget, an English surgeon who published in 1889 what has come to be known as the “seed and soil” hypothesis, for it embodies an idea quite akin to that implied in Sperry’s chemoaffinity postulate. Paget noted, in the process of analysis of more than 900 autopsy records, that tumor metastasis contains an organ-specific, nonrandom character: “The evidence seems to me irresistible that in cancer of the breast the bones suffer in a special way (…) Some bones suffer more than others; the disease has its seats of election” (Paget 1889). From these observations, he inferred that metastases depend on certain cancer cells—the “seeds”—having a specific affinity for the environment of certain organs—the “soil”—correctly concluding, with sound resemblance to modern day theories of neural circuitry assembly, that only when both “seed” and “soil” were compatible would metastasis form (Fidler 2003; Valastyan and Weinberg 2011).

Curiously, a short incursion into the history of the classical axon guidance molecules reveals an association with cancer pathology dating back to their discovery, suggesting that common signaling pathways operate in both contexts. The Epha1 gene, for example, was cloned from a carcinoma cell line in 1987 in a screen for novel tyrosine kinase receptors with oncogenic potential, and the first Ephrin ligand was also described by a group working in the context of cancer (Pasquale 2010). Likewise, the Dcc gene, the prototypical Netrin-1 receptor, was originally identified as a tumor suppressor in advanced stages of colorectal carcinoma (hence its designation, deleted in colorectal cancer) (Mehlen et al. 2011). Only in the mid-1990s did their association with axon guidance mechanisms begin to be established (Tessier-Lavigne and Goodman 1996; Serafini et al. 1996).

However, the links between these processes are perhaps best illustrated by the recent characterization of frequent mutations and copy number variations in classical axon guidance genes in tumors derived from pancreatic ductal adenocarcinoma and liver fluke-associated cholangiocarcinoma patients (Biankin et al. 2012; Ong et al. 2012) or the ongoing cancer clinical trials targeting axon guidance molecules. Finally, it is also relevant to note that tctp is not the only cancer-associated transcript localizing in developing axons; in fact, the “cancer” gene ontology (GO) term is among the most significantly enriched in these axonal transcriptomes (Zivraj et al. 2010; Gumy et al. 2011), underlying the cellular and molecular commonalities that can be drawn between both contexts.

10.1.4 Axonal Mitochondria

Since, as a general rule, neuronal cells cannot be replaced throughout the individual’s lifetime (Dekkers et al. 2013), the preservation of functional neural circuits must necessarily rely on effective protective mechanisms of its components. Classical conjectures supported the view that the process of axonal degeneration ensued from deficient sustenance from the cell body (e.g., as a result of cell body death) (Pease and Segal 2014). However, it is now well established that the axonal degenerative cascade can be actively promoted by in situ death pathways and is counteracted by locally acting and, to some extent, axon-specific pro-survival mechanisms (Pease and Segal 2014). Moreover, adequate metabolic provision—and hence mitochondria—is pivotal to axonal function, as the demand for energy, metabolites, and calcium buffering is particularly elevated at axons terminals (e.g., to support synaptic transmission) (Friedman and Nunnari 2014). Indeed, many mitochondrial dysfunctions trigger neurodegenerative disorders with prominent axonal phenotypes (Nunnari and Suomalainen 2012; Delettre et al. 2000; Alexander et al. 2000), suggesting that axons are particularly vulnerable to compromised mitochondrial output. Similarly, a growing axon is dependent on adequate mitochondrial operation, as it requires the continuous provision of energy for its extension in the embryonic brain. It follows that neurons must preserve a damage-prone mitochondrial network to maintain functionality and integrity.

10.2 TCTP in Neuronal Circuitry Assembly

Given that the identification of tctp as a potential candidate of study stemmed from genome-wide profiling screens, we initially sought to validate that its transcripts localize to retinal ganglion cell axons and growth cones at a time when the Xenopus laevis retinotectal projection is developing.Footnote 1 In situ hybridization showed robust tctp signal in the optic fiber layer and in the optic nerve head axon-only structures through which retinal ganglion cell axons navigate to exit the eye. Additionally, in eye explants, tctp mRNA signal could be detected in the growth cone of retinal ganglion cell axons. In concordance with tctp mRNA axonal localization, Tctp protein was similarly detected in these retinal ganglion cell structures. Ample mRNA and protein signals were also found in the inner and outer plexiform layers, suggestive of localization in the neurites of other retinal neurons, as well as in the photoreceptor layer, populated by light-sensitive neurons, and the ciliary marginal zone, a neurogenic niche in the retina. Significantly, our initial investigations also showed that tctp expression is nearly tenfold higher than actb in retinal ganglion cell axons as measured by quantitative PCR,Footnote 2 confirming tctp as a highly enriched axonal transcript.

Further analyses revealed that Tctp is implicated in the development of the retinotectal projection (Fig. 10.2). Specifically, Tctp depletion using antisense morpholino oligonucleotides results in splayed projections that fail to innervate the optic tectum at the normal developmental time window (Fig. 10.2a). These effects are not a consequence of extracellularly acting Tctp, as normal retinal ganglion cell axons develop unerringly through a Tctp-deficient optic tract pathway (Fig. 10.2b). Moreover, in vivo time-lapse imaging of developing Tctp-depleted retinal axons revealed that their rate of extension was about half of that observed in controls, excluding the possibility that the axonal phenotypes observed are a result of an underlying delay in eye development.

Fig. 10.2
figure 2

Tctp is required for axon development in the embryonic visual system. (a) tctp knockdown in vivo was achieved using an antisense oligonucleotide morpholino (MO) delivered into both dorsal blastomeres of four-cell stage Xenopus laevis, which give rise to the entire central nervous system. The retinotectal projection was labeled by intra-ocular delivery of a fluorescent lipophilic dye (DiI) at stage 40, when pioneer axons have completed their stereotyped growth through the optic tract (OT) and reached their target area. Whereas control embryos consistently developed compact axon profiles and had innervated the optic tectum, Tctp deficiency resulted in stunted and splayed projections that lagged in their development. The retinotectal projection is depicted in orange. Dashed contour represents the contralateral, dye-filled eye. (b) Tctp displays IgE-dependent histamine-releasing activity and other cytokine-like extracellular roles. Consequently, it could regulate axon development through its effects in the embryonic brain environment. To test this possibility, we devised an approach that generates embryos deficient in Tctp only in one-half of the nervous system. Because the retinotectal projection projects contralaterally (i.e., axons from the left eye extend towards the right side of the brain), this methodology allowed us to probe the effects of a Tctp-deficient optic tract pathway. Overall, normal axons developed unaffected through the Tctp morphant environment, suggesting that the observed axon phenotypes are independent of Tctp acting extracellularly

Full size image

We began our characterization of Tctp axonal mode of action by focusing on mitochondria. This line of investigation unexpectedly arose while examining the histology of Tctp-depleted retinas for signs of delayed development. Curiously, although the gross stratification of the retina was unaffected, we noted obvious signs of degeneration in the photoreceptor layer of Tctp morphants. The subsequent finding that Tctp expression in these cells is confined to the mitochondria-rich inner segments, together with reports documenting Tctp as part of the mitochondrial proteome (Fountoulakis et al. 2002; Rezaul et al. 2005), suggested a potential link between Tctp and mitochondrial function.Footnote 3 These indeed proved to be insightful observations, as Tctp morphant retinas show reduced total ATP levels. Following on this result, we measured a ~20% decline in the membrane potential of mitochondria from Tctp-depleted axons,Footnote 4 as well as a significant decrease in the number of axonal mitochondria. Importantly, this decrease in axonal mitochondrial density was not accompanied by changes in overall mitochondrial biogenesis or mass, arguing for a phenotype with predominantly axonal repercussions. Indeed, examination of mitochondrial transport dynamics in axons showed that a higher proportion of these organelles move towards the cell body in axons deficient in Tctp than in controls, in line with previous reports showing that dysfunctional mitochondria are selectively “shipped” to the cell body for repair and/or degradation (Miller and Sheetz 2004; Sheng and Cai 2012).

How does mitochondrial dysfunction develop from Tctp deficiency? An interesting possibility stemmed from reports linking Tctp to the B-cell lymphoma 2 (Bcl2) family of proteins, which play key mediator roles of mitochondrial integrity and apoptosis (Czabotar et al. 2014). Significantly, embryonic sensory neurons lacking Bcl2, the prototypic member of this family, show reduced axon growth rates (Hilton et al. 1997), a phenotype we encountered in Tctp morphants. Particularly well defined is the association between Tctp and myeloid cell leukemia 1 (Mcl1) (Liu et al. 2005; Yang et al. 2005; Zhang et al. 2002), a neuroprotective Bcl2-related pro-survival factor (Mori et al. 2004), which prompted us to explore whether these two proteins shared a functional relationship in axons. We first showed that axonal Tctp interacts with Mcl1 using a proximity ligation assay, complementing previous biochemical data with an approach that allows the examination of protein–protein interactions with subcellular precision. Second, we looked for signs of unbalanced pro-survival signaling in Tctp-depleted axonsFootnote 5 (Liu et al. 2005; Amson et al. 2012; Rho et al. 2011; Vaseva and Moll 2009; Leu et al. 2004). Both cleaved Caspase-3 and P53 levels were found to be elevated in axons in the absence of Tctp. Third, consistent with the idea that Tctp works via Mcl1 and the survival machinery to regulate axon development, Mcl1 morphants show similar, albeit milder, axon misprojection phenotypes.Footnote 6 Finally, since the N-terminal region of Tctp is required for its pro-survival properties (Yang et al. 2005; Zhang et al. 2002), we were able to test whether Tctp pro-survival interactions are a requirement for normal axon development. To this end, we designed a transgene encoding a truncated Tctp protein that was devoid of pro-survival activity (Tctp40-172aa). Tctp40-172aa retains Tctp’s signature motifs, as well as the interactions domains required for the association with several known Tctp-interacting proteins, but lacks those necessary for the association with Mcl1 (Yang et al. 2005). Unlike full-length tctp, co-delivery of tctp 40-172aa with a tctp-targeting morpholino failed to prevent the abnormal development of the retinotectal projection resulting from Tctp deficiency. Collectively, our various findings suggest that Tctp regulates axon development through its association with the survival machinery of the axon (Fig. 10.3), and establish a novel and fundamental role for Tctp in vertebrate neural circuitry assembly (Roque et al. 2016).

Fig. 10.3
figure 3

Mechanistic insights into the role of Tctp in neuronal circuitry assembly. The normal physiologic scenario is illustrated in (a), whereas the consequences of Tctp deficiency on axon development programs uncovered by our study are shown in (b)

Full size image

10.3 Summary and Future Directions

Neurons are highly compartmentalized cells with great energy demands. Given their elongated morphology and unique metabolic requirements, mitochondrial operation needs to be appropriately regulated in these cells to sustain normal neuronal functioning. This assumes particular relevance at distal axon terminals, which require the localized presence of mitochondria to support growth, maintenance, and synaptic transmission (Sheng and Cai 2012). Significantly, our study identified Tctp as a key checkpoint for normal axon development by impacting on axonal mitochondrial homeostasis. Given the importance of maintaining an operational mitochondrial network during axon development and overall neuronal function, it is perhaps not surprising that all axonal populations analyzed to date at the transcriptome level contain a large proportion of mitochondria-related mRNAs (Jung et al. 2012). In fact, it has been demonstrated that up to 25% of all proteins synthesized in presynaptic nerve terminals become associated with mitochondria (Gioio et al. 2001). Hence, our efforts to characterize the function of Tctp in the context of axon development typify the significant biological investment put into supporting these organelles subcellularly.

Whereas we focused exclusively on examining the role of Tctp in axon development, future work should aim at elucidating its implications in the adult nervous system. Indeed, the decreased Tctp protein expression levels observed in Down syndrome and Alzheimer’s disease (Kim et al. 2001), pathologies associated with mitochondrial dysfunction (Friedman and Nunnari 2014; Pagano and Castello 2012), together with the finding that tctp is also among the most abundant transcripts in adult axons, prompt speculation that Tctp holds an important lifelong axonal function. However, given that Tctp is required for the assembly of neural circuitry, temporal control over its expression will be a key aspect of any future examination. This could be achieved with minimal effort by crossing the existing tctp-floxed heterozygous mice with an inducible, neuron-specific Cre recombinase strain (Chen et al. 2007; Susini et al. 2008). Considering that proper mitochondrial operation is an imperative of synaptic homeostasis (Sheng and Cai 2012), such strategy would, for example, allow one to study Tctp in the context of synaptic function independently of preceding defects in neural circuitry formation. Very much in line with these considerations, recent evidence collected from a neuron-like cell line suggests that Tctp can function as a promoter of dopamine release, a neurotransmitter involved in motor control and reward-motivated behaviors (Seo et al 2016).