In Silico analysis of newt A1EV cargo identifies genes involved in neural growth and development
EV cargos are key mediators of functional regulation in recipient cells. In our previous study, EV RNA cargo, isolated from newt A1 cells (Fig. 1a) was profiled by RNA-Seq and mapped using sequence similarity to human genome databases 18. As shown in Figure 1b, we had identified many unique mRNAs in newt A1EVs not found within the EV cargo of two mammalian cell lines, a cardiac-derived stem cell and normal human dermal fibroblasts. Closer scrutiny of mRNA unique to A1EVs revealed many of these mRNAs are involved in early embryonic development and neuronal differentiation. Gene Ontology Analysis (GOA) identified mRNA involved in several biological processes related to neuron function, including neuron projection growth and development; p-value = 2.93E-23 and 1.54E-19, respectively (Figure 1c, Supplemental Figure 1 and Supplemental Table 1). Additionally, much of the mRNA cargo generate gene expression products found within the cellular components of the neuron projections and synapses; p-values = 1.67E-21 and 4.75E-23, respectively (Supp. Fig. 2 and Supp. Table 2).
A1EVs drive neurite-like extensions in cell lines that model nerve growth
To determine if A1EVs can affect neuron differentiation or drive neurite growth, 1x108 A1EVs were added to cultured PC-12 cells, a rat pheochromocytoma line of the adrenal medulla that expresses the nerve growth factor receptor (NGF) and responds to NGF by generating plasma membrane projections that share similarities to neurites. After 48 hours of incubation with either A1EVs, A1 cell culture media in which EV generation had been suppressed (20µM GW4869), or media-only control, the PC-12 cells displayed many more neurite-producing cells with A1EV incubation than with the media control or with A1 media lacking EVs. (Supp. Fig. 3).
A1EVs promote neurite growth and branching in mammalian sympathetic neurons, in a dose-dependent manner
As PC-12 cells are not true neurons 19,20, the potential nerve growth properties of A1EVs were tested on sympathetic nerves derived from rat superior cervical ganglion (SCG). Sympathetic neurons were chosen as a study model amongst other nerve types, as sympathetic neuron growth and neurite development require minimal growth factors that may mask or confound the effects of A1EVs. Neonatal rat SCG neurons were plated sparsely to avoid inter-cell contact and multiple cultures were exposed to increasing concentrations of A1EVs and incubated for 12 hours. As described in the methods section, the A1EVs were thoroughly washed in PBS prior to neuron culture addition to remove any secreted factors that are not part of the A1EV cargo, such as Nerve Growth Factor (NGF), that could confound the effects of the A1EVs. Alpha acetylated tubulin staining revealed increases in neurite length after 12 hours corresponding to increases in A1EV concentration and 108 and 109 A1EVs/mL yielded the maximum neurite length (Fig. 2a,c). Interestingly, at 1x1010 EVs/mL, very little neurite growth occurred, far less than in the media-only control group. Sholl analysis 21, a measure of neurite complexity, of A1EV-treated neurons indicated that the average neurite growth and the number of neurite intersections also increases with increasing concentration of A1EV, until 1x1010 EVs/mL which showed little neurite growth and complexity (Fig. 2b,c). Similar effects on neurite growth were also demonstrated on sensory DRG neurons (Supp. Fig. 4). Since both 1x108 and 1x109 EVs/mL showed strong effects on the induction of neurite growth, 1x108 EVs/mL was chosen for all the ensuing experiments.
The rate of neurite growth increases following A1EV exposure, but growth cone number and length are similar to untreated control
Next, to determine the optimal time for A1EV-mediated induction of neurite growth, SCG sympathetic neurons were administered 108 A1EVs/mL or media alone, and neurite growth was assessed 1-, 2-, 4-, 6-, and 12-hours post-exposure. Enhanced neurite growth occurred within one hour of A1EV treatment, indicating that A1EVs may activate endogenous growth pathways via receptor signaling, as opposed to translation of delivered mRNA cargo (Fig. 3a, b). Although there appears to be a sudden shift in neurite growth rate in A1EV-treated neurons at 12 hours, this was likely due to the neurites coming into contact with one another. Growth cone length and number of projections was assessed by measuring the length and number of neurite tips that stained positive for filamentous actin (phalloidin), but negative for alpha acetylated tubulin. No significant difference was observed between A1EV-exposed and unexposed neuron cultures (Fig. 3c, d, e).
A1EV suppression diminishes mammalian neurite growth
To confirm that A1EVs were responsible for the enhanced neurite growth, and not some other secreted, non-vesicular factors, we added EV generation inhibitors (e.g., GW4869, an inhibitor of ceramide synthesis which suppresses A1EV secretion 18) to A1 cells in culture. To establish effective inhibitor concentrations, A1 cells were exposed to increasing concentrations of GW4869 and two other EV inhibitors (cambinol, and ketotifen fumarate) concurrently with plating (Fig. 4a). The lowest effective concentrations for A1EV suppression for GW4869, cambinol or ketotifen fumarate were 20uM, 10uM and 10uM, respectively. Next, SCG neurons were plated in complete media and, one hour afterwards, exposed to MEM only (control), 1x108 A1EVs/mL in MEM, or equivalent volumes of the MEM collected from inhibitor-exposed A1 cultures. As described in the methods section, the enriched A1EV MEM media was thoroughly washed in PBS during the ultrafiltration step to remove any remaining EV synthesis inhibitors. Neurons were then grown for 12 hours and analyzed by immunohistochemistry and confocal microscopy. While A1EV-exposed neurons demonstrated enhanced neurite growth, those neurons that received few or no A1EVs did not grow significantly faster than the MEM-only control neurons (Fig. 4b).
Fluorescently labelled A1EVs generation and neuron uptake
To track A1EVs in solution and assess uptake by cultured neurons, we generated an A1EV cell line that expressed a CD63-GFP fusion protein; this tetraspanin cell protein is a common surface marker used to identify and isolate EVs. Phase-contrast and fluorescent microscopy images of transfected A1 cells, show diffuse, cytosolic GFP expression as well as concentrated pockets of CD63-GFP within the Golgi (Fig. 5a). Two hours after exposure of cultured sympathetic neurons to the fluorescent A1EVs (5x108 A1EV/mL), co-localization of GFP-labelled A1EVs with Rab7, a late endosomal marker that co-localizes with internalized EVs 22 was observed along the length of the neurites (Fig. 5b).
A substantial body of evidence demonstrated that EV-resident mRNAs can be translated into functional proteins in recipient cells 23. Several mRNAs involved in early embryonic development and neuronal differentiation, such as RUNX3, are enriched in A1EVs (Fig. 1b). RUNX3 is considered as a critical determining factor in early outgrowth of neurons 24,25. While RUNX3 mRNA was enriched within the A1EVs, no RUNX3 protein was found within the EVs 18 The next logical step then was to examine whether RUNX3 mRNAs residing in A1EVs can be translated and contribute to the protein expression in recipient neurons. Analysis of protein expression, via Western Blot, in SCG neurons found significantly higher levels of RUNX3 following A1-EV exposure compared to media-only controls (Fig. 5c).
Enhanced neurite growth is supported by increased mitochondria and increased cellular respiration in A1EV-exposed neurons
The metabolic costs of facilitating higher neurite growth rates must be met with increased cellular respiration. We measured oxygen consumption rates (OCR) of cultured SCG neurons exposed to A1EVs (1x108/mL), equivalent volumes of media (control), or A1 culture media lacking A1EVs (20µM and 40µM GW4869) (Fig. 6). Spare respiratory capacity and ATP production at baseline were significantly increased in A1EV-exposed neurons compared to media-only treated cells (Fig. 6a). Additionally, as shown in Figure 6b and 6c, a concentration-dependent response was observed between increasing numbers of A1EVs (decreasing GW4869 concentration) and basal respiration, spare respiratory capacity, proton leak, and ATP production. Probing for the mitochondrial marker, TOM70, by Western blot in A1EV-exposed neurons (1x108/mL) revealed increased mitochondrial staining compared to media-only control (Fig. 6b), but no changes in mitochondrial localization within A1EV-treated neurons (Fig. 6c).
Canonical Pathway Analysis and validation of the IPA-predicted activation of NGF/ERK5 signaling pathway in SCG neurons treated with A1EV
Having demonstrated that A1EVs could increase neurite growth, we sought to investigate the underlying molecular mechanisms. When comparing RNA from neurons exposed to 1x108 A1EVs/mL or media only, sequencing data and Ingenuity Pathway Analysis (IPA) identified 36 significantly-altered canonical pathways (Fig. 7a). The top 2 regulated pathways were NGF signaling (-log(p-value) = 3.87) and ERK5 signaling (-log(p-value) = 2.43). For NGF signaling, 66% (75/114) of the pathway genes were upregulated, and 27% (31/114) downregulated. ERK5 signaling was also activated, with 65% (47/72) of the pathway genes upregulated and 26% (31/114) downregulated. Proteomic analysis of isolated A1EVs did not identify NGF within the EV cargo and RNA sequencing found negligible levels of NGF mRNA transcript (53 reads in 10 million) (18, supplemental tables).
To validate IPA-predicted activation of NGF/ERK5 signaling pathway in SCG neurons exposed to A1EVs, we examined key proteins within the NGF signaling pathway (TrkA, ERK5 and CREB) (Fig. 7b) by Western Blot. The expression levels of TrkA, ERK5 and CREB were significantly upregulated in the A1EV-treated neurons compared to media-only controls (Fig. 7c). Overexpression of ERK5 has been demonstrated to strengthen ERK5 signaling 26,27. The next logical step was to examine whether A1EV-mediated upregulation of ERK5 promoted downstream signaling activation by checking phosphorylation of ERK5 and CREB. Intriguingly, A1EV exposure resulted in increased phosphorylation of ERK5 and CREB. These findings validated that A1EV uptake activates NGF/ERK5 pathway in SCG neurons.
Apoptosis is reduced following A1EV exposure in NGF-deprived SCG neurons
In addition to governing neurite outgrowth, NGF signaling pathway activation also regulates neuronal survival through downstream activation of PI3K/AKT survival pathways 28. Depravation of NGF in SCG neurons activates the mitochondrial pathway of apoptosis through BCL-2-associated X protein (BAX) and the release of cytochrome c. We had previously described the pro-survival effects of A1EVs in mammalian cardiomyocytes undergoing oxidative stress through the activation of PI3K/AKT signaling, but we did not determine if the benefits were due to cell receptor activation or incorporation of A1EV cargo components that stimulate signaling pathways downstream of the receptor.
To examine whether A1EV-mediated NGF signaling could promote survival, SCG neurons were cultured for 48 hours in complete neural media + 10ng/mL of NGF, then rinsed and switched to NGF-free media. Neurons were exposed to 108 A1EVs/mL, or media only, at the time of the media change, cultured for an additional 12 hours without NGF, and then fixed. Neuron death was assessed by TUNEL staining and the number of neuron cell bodies (soma) present, an indicator of neuron apoptosis (Fig. 8a-c). The number of TUNEL-positive neurons was significantly decreased in A1EV-exposed cultures as compared with control. Additionally, tubulin staining revealed higher numbers of somas in NGF-depleted neuron cultures following A1EV exposure, compared to media-only control. Thus, A1EVs reduced primary neuron apoptosis induced by NGF withdrawal.
Blocking the TRKA receptor does not inhibit A1-EV-mediated neuron survival in NGF-depleted cultures
Although Fig 5b shows that internalized A1EVs co-localize with the Rab7 endocytic marker, it was unclear if A1EVs could also drive NGF/ERK5 signaling through activation of the Tropomyosin Receptor Kinase A (TRKA) receptor, the primary receptor of the NGF signaling pathway. To probe this question, primary SCG neurons were exposed to the TrkA receptor-specific inhibitor AG879 at two concentrations; 100nM, which partially inhibits NGF-promoted survial and 5uM, which fully blocks NGF-promoted survival 29,30 (Fig. 9a, b). SCG neurons cultured in complete neural media + NGF maintained a basal apoptosis rate of approximately 11%. 12 hours in NGF-depleted media resulted in increased TUNEL staining (71%), as expected. The addition of 100nM AG879 (AGlow) and 5uM AG879 (AGhigh) in the presence of NGF resulted in an increase of TUNEL-positive neurons (49%and 100%, respectively). The presence of A1EVs in the media without NGF rescued the neurons from apoptosis (19% vs. 71%), while neurons in NGF-depleted media plus A1EVs and AG879 exhibited fewer TUNEL+ neurons than controls (21%). High levels of AG879 (5uM) + A1EVs in NGF-depleted cultures resulted in a partial rescue, with a 29% decrease in TUNEL-positive staining compared to SCG cultures treated with 5uM AG879, that lack NGF. This indicates that, while A1EVs can protect neurons from NGF-depleted culture conditions, the mechanism can occur independent of the TrkA receptor. Thus, A1EV-mediated survival of neurons is accomplished through the NGF signaling pathway but can occur independently of the TRKA receptor.