TWE-PRIL reverse signalling suppresses sympathetic axon growth and tissue innervation

ABSTRACT TWE-PRIL is a naturally occurring fusion protein of components of two TNF superfamily members: the extracellular domain of APRIL; and the intracellular and transmembrane domains of TWEAK with no known function. Here, we show that April−/− mice (which lack APRIL and TWE-PRIL) exhibited overgrowth of sympathetic fibres in vivo, and sympathetic neurons cultured from these mice had significantly longer axons than neurons cultured from wild-type littermates. Enhanced axon growth from sympathetic neurons cultured from April−/− mice was prevented by expressing full-length TWE-PRIL in these neurons but not by treating them with soluble APRIL. Soluble APRIL, however, enhanced axon growth from the sympathetic neurons of wild-type mice. siRNA knockdown of TWE-PRIL but not siRNA knockdown of APRIL alone also enhanced axon growth from wild-type sympathetic neurons. Our work reveals the first and physiologically relevant role for TWE-PRIL and suggests that it mediates reverse signalling.


INTRODUCTION
Key to understanding how functional neural circuits are established in development is the identification and characterization of the signals that regulate the growth, branching and disposition of neural processes. A great deal has been learned about the nature of these signals and how they operate by studying neurons of the vertebrate peripheral nervous system (PNS), not least because of the ease with which PNS neurons can be cultured and the ease with which the innervation of various tissues can be studied in vivo. Pre-eminent among the signals that regulate the growth of neural processes is the nerve growth factor (NGF) family of neurotrophins (Huang and Reichardt, 2001). These secreted proteins are generally synthesized in the periphery by innervated tissues and have two main functions in PNS neuron development: they promote neuron survival and stimulate the growth and branching of axons within innervated tissues. As such, they play a key role in regulating neuron number and tissue innervation (Davies, 2003;Glebova and Ginty, 2004).
In addition to neurotrophins and other neurotrophic factors, members of the tumour necrosis factor superfamily (TNFSF) have recently been recognized to be physiologically relevant regulators of axon growth during development. The 19 members of this superfamily are best understood for their many crucial functions in the development and function of the immune system, but are increasingly recognized as having diverse roles in other tissues and organs (Hehlgans and Pfeffer, 2005). Several TNFSF members have been shown to modulate axon growth, generally without affecting neuronal survival. For example, in certain populations of neurons, GITRL (TNFSF18) and APRIL (a proliferating-inducing ligand, TNFSF13) enhance axon growth (McWilliams et al., 2017;O'Keeffe et al., 2008;Osório et al., 2014), whereas RANKL (TNFSF11) and LIGHT (TNFSF14) suppress the axon growthpromoting actions of neurotrophins (Gavalda et al., 2009;Gutierrez et al., 2013). Members of the TNFSF are type II transmembrane glycoproteins that are active both as membrane-integrated ligands and as soluble ligands after cleavage from the cell membrane. They exert their effects by binding to members of the tumour necrosis factor receptor superfamily (TNFRSF) (Hehlgans and Pfeffer, 2005). In addition to functioning as ligands, several membraneintegrated TNFSF members can act as reverse signalling receptors for their respective TNFRSF partners, which function as activating ligands (Sun and Fink, 2007). Recently, TNFSF reverse signalling has been shown to be physiologically relevant for the regulation of axon growth and tissue innervation in the developing PNS. For example, TNFR1-activated TNF reverse signalling (Kisiswa et al., , 2017Wheeler et al., 2014) and CD40-activated CD40L reverse signalling (McWilliams et al., 2015) enhance axon growth and promote the innervation of certain tissues in vivo. CD40activated CD40L reverse signalling also enhances the growth and branching of excitatory neuron dendrites, and suppresses the growth and branching of inhibitory neuron dendrites in the developing CNS (Carriba and Davies, 2017).
APRIL is best characterized for its role in regulating lymphocyte survival and activation (Schneider, 2005). In the developing nervous system, soluble APRIL enhances axon growth from cultured hippocampal pyramidal neurons and midbrain dopaminergic neurons, and in vivo studies have revealed that the initial nigrostriatal projection is significantly compromised in April −/− embryos (McWilliams et al., 2017;Osório et al., 2014). The April gene (Tnfsf13 -Mouse Genome Informatics) is adjacent to the gene that encodes another member of the TNFSF, TWEAK (TNF-like weak inducer of apoptosis, TNFSF12). A differentially spliced hybrid transcript is translated into a protein, TWE-PRIL, that comprises the extracellular domain of APRIL, and the cytoplasmic and transmembrane domains of TWEAK (Kolfschoten et al., 2003;Pradet-Balade et al., 2002). Although T-cells and monocytes express TWE-PRIL and the protein has been shown to stimulate cycling lymphoma cell lines (Pradet-Balade et al., 2002), no biological function has yet been ascribed to this protein in vivo.
The sympathetic neurons of the rodent superior cervical ganglion (SCG) are one of the most extensively studied populations of neurons in the developing PNS. These neurons innervate cranial tissues and become dependent on NGF for survival shortly after their axons reach and ramify within their target tissues (Davies, 2009;Glebova and Ginty, 2005). Our current in vivo and in vitro studies of SCG neurons of April −/− mice (which lack APRIL and TWE-PRIL), together with siRNA transfection in wild-type neurons to selectively knock down expression of particular proteins have revealed the first physiologically relevant role for TWE-PRIL. This functions as a suppressor of NGF-promoted sympathetic axon growth and accomplishes this role via a novel signalling mechanism; TWE-PRIL-mediated reverse signalling activated by one of the canonical APRIL receptors, BCMA (TNFRSF17), functioning as the ligand.

Expression of APRIL, TWEAK, TWE-PRIL and BCMA in SCG neurons
To begin to examine whether TWE-PRIL and/or APRIL play a role in developing sympathetic neurons, we used RT-qPCR to assess the expression of Twe-pril (Tnfsfm13 -Mouse Genome Informatics) and April mRNAs in SCG dissected from mice at stages throughout embryonic and postnatal development. We also measured the levels of transcripts encoding BCMA (TNFRSF17) and TACI (TNFRSF13B), the TNFRSF members that bind APRIL/TWE-PRIL (Bossen and Schneider, 2006). For completeness, we also measured the levels of Tweak mRNA in the developing SCG. To assess expression of Twe-pril mRNA, we used primers that bind either side of the junction of the regions encoding the APRIL and TWEAK components of this fusion protein. The relative levels of expression of these transcripts were quantified relative to their levels in adult spleen.
Twe-pril, April, Tweak and Bcma mRNAs were reliably and consistently detected in the developing SCG. April and Bcma mRNAs were first detected at E13, when the SCG becomes anatomically discernible and the earliest axons are growing to their targets. Twe-pril and Tweak mRNAs were first detected at E15, which is a day after the earliest axons reach their targets and have started to grow and branch within their targets (Fig. 1A). Taci mRNA was barely detectable and not accurately quantifiable at any of the embryonic and postnatal ages studied.
To confirm expression of the encoded proteins, we visualized these proteins by immunocytochemistry in dissociated SCG cultures established from P0 mice. TWE-PRIL and APRIL were localized using an antibody that recognizes an epitope in the extracellular domain of both TWE-PRIL and APRIL. This antibody strongly labelled virtually all SCG neurons (Fig. 1B). Labelling was not observed in cultures established from P0 April −/− mice (Fig. 1), confirming the specificity of this antibody. Virtually all SCG neurons were also labelled by anti-BCMA and by anti-TWEAK. Labelling with anti-TACI antibodies was barely above background. In double labelling experiments, the neurons were co-labelled by anti-APRIL/TWE-PRIL and anti-BCMA. These studies suggest that SCG neurons co-express TWE-PRIL/APRIL and BCMA.
Increased in vivo growth of sympathetic fibres in April −/− mice Given the evidence that APRIL promotes the early growth of axons from at least two kinds of CNS neurons (McWilliams et al., 2017;Osório et al., 2014) and our current demonstration that developing SCG neurons express TWE-PRIL/APRIL, we investigated whether absence of TWE-PRIL/APRIL affects the projection and early ramification of sympathetic axons in their targets by comparing the sympathetic innervation of April −/− embryos (which lack APRIL and TWE-PRIL) and April +/+ littermates. Immunolabelling-enabled three-dimensional imaging of solvent-cleared organs (iDISCO) (Renier et al., 2014) is particularly appropriate for examining this, as it facilitates imaging the sympathetic innervation of SCG targets in cleared heads at a stage of development when sympathetic fibres are growing and branching within their targets. Sympathetic fibres were labelled with an antibody against tyrosine hydroxylase (TH), a ratelimiting enzyme of noradrenaline synthesis. These studies were carried out at E16, which is 2 days after the earliest sympathetic fibres have reached their targets and have started to ramify in these targets, but is before the sympathetic innervation becomes too dense to reliably quantify with this technique.
The iDISCO preparations were visualized by light sheet fluorescent microscopy at optical sections of 5 µm (between 600 and 800 optical sections per animal, which included all rostrally directed fibres from both SCG). Individual optical sections were assembled into three-dimensional reconstructions of sympathetic fibres emanating from the SCG using Bitplane Imaris software (Bitplane Scientific Solutions, Zurich, Switzerland), which allows topographic manipulation of the reconstructions for accurate fibre tracing and measurement. Examination of collapsed z-stack light sheet images ( Fig. 2A) and Bitplane Imaris reconstructions ( Fig. 2B) showed that the gross anatomical disposition of individual fibres in whole heads of April +/+ and April −/− mice were similar at E16. However, measurements of the lengths of the longest SCG fibres (which terminate in nasal tissue at this stage) in Bitplane Imaris reconstructions revealed significantly longer fibres in April −/− mice compared with April +/+ littermates (Fig. 2C). Bitplane Imaris reconstructions facilitated semi-manual tracing of individual SCG fibres (Fig. 2D). The small number of caudally directed and short local SCG fibres were excluded from these tracings because the full extent of caudally directed fibres could not be followed because several of these had been severed when heads were collected for iDISCO preparations and because of the complexity of local SCG fibres. Measurements taken from these manual tracings revealed a highly significant, twofold increase in the combined lengths of fibres in E16 April −/− mice compared with April +/+ littermates (Fig. 2E). All analysis was carried out blind. Importantly, when quantification was carried out using a surface rendering programme that required no observer input following manual threshold optimization (Fig. 2F), April −/− heads still exhibited a highly significant increase in total sympathetic innervation compared with April +/+ littermates (Fig. 2G).
To exclude the possibility that these results may reflect differences in the level of expression of TH between the SCG neurons of April +/+ and April −/− mice, we used western blotting to estimate the relative levels of TH in the SCG of these mice. There was no significant difference in TH levels between the SCG of April +/+ and April −/− mice at E16 (Fig. S1). In summary, our studies suggest that SCG sympathetic fibres in April −/− embryos grow longer than in April +/+ littermates. To ascertain whether the long sympathetic fibre phenotype observed in April −/− embryos in vivo is replicated by sympathetic neurons cultured from these embryos, we compared the size of the axon arbors of SCG neurons cultured from April +/+ and April −/− mice. We established very low density dissociated cultures from E16, P0, P3 and P5 April +/+ and April −/− littermates in culture medium supplemented with 0.1 ng/ml NGF to stimulate axon growth and 25 μM of pancaspase inhibitor Boc _ D _ FMK to prevent apoptosis. After 16 h incubation, the neurons were labelled with the fluorescent vital dye calcein-AM, and images of individual neurons were acquired. In these short-term cultures, all neurites are structurally axons (Lein et al., 1995). Neurons cultured from April −/− mice had larger axon arbors than those grown from April +/+ mice, as shown by representative neurons in P0 cultures (Fig. 3A). These differences and how they change in development are clearly illustrated in Sholl plots, which provide a graphic illustration of axon length and branching with distance from the cell body (Sholl, 1953). The differences in axon arbor size between April −/− and April +/+ neurons were greatest in cultures established at E16 and P0. Differences were negligible in P3 cultures and were not discernable in P5 cultures (Fig. 3B).
NGF dose responses carried out at E16, P0 and P3 in the presence of the pancaspase inhibitor Boc _ D _ FMK revealed that the difference in total axon length in the neuron arbors between April −/− and April +/+ neurons increased with NGF concentration. As shown in P0 cultures, there were no differences in total axon length in the absence of NGF, and the differences between April −/− and April +/+ neurons increased with increasing NGF concentration up to 10 ng/ml and (the highest concentration of NGF used) (Fig. 3C). Very similar differences were observed in E16 cultures and a similar trend was observed in P3 cultures, although the differences were not as great (not shown).
Although there were significant differences in NGF-promoted axon growth between the SCG neurons of April +/+ and April −/− mice, there were no significant differences in NGF-promoted survival. In the absence of Boc _ D _ FMK, overlapping NGF survival dose responses were obtained from SCG neurons of April +/+ and April −/− mice, as illustrated for P0 cultures (Fig. 3D). Overlapping NGF survival dose responses were also observed in cultures of SCG neurons of April +/+ and April −/− littermates set up at E16 and P3 (not shown). In summary, these findings show that NGF-promoted axon growth, but not survival, is significantly greater from April −/− SCG neurons, compared with April +/+ SCG neurons. The enhanced response to NGF occurs over a window of perinatal development when axons are ramifying in their targets under the influence of target-derived NGF. Because SCG neurons cultured from April −/− mice mirror the in vivo phenotype, we were able to undertake additional in vitro studies to clarify the mechanism underlying this phenotype.
siRNA knockdown of TWE-PRIL/APRIL, but not APRIL alone, increases sympathetic axon growth To complement and extend our in vitro studies of NGF-promoted axon growth from SCG neurons cultured from April +/+ and April −/− mice, we used siRNA to knock down expression of TWE-PRIL, APRIL, TWEAK, BCMA and TACI in SCG neurons cultured from wild-type mice. We co-transfected P0 SCG neurons with specific siRNA oligonucleotides, together with a GFP expression vector to identify and label the transfected neurons. The neurons were (E) Scatter bar chart of the combined lengths of rostral SCG fibres measured in Bitplane Imaris reconstructions (mean and individual data points, n=6 separate embryos of each genotype, ***P<0.001, t-test). (F) Representative images of SCG fibre reconstruction using an automatic surface rendering programme based on Bitplane Imaris reconstructions of April +/+ and April −/− heads viewed in the transverse plane. (G) Scatter bar chart representing the total surface volume of rostral SCG fibres ( proportional to total fibre length) in reconstructions generated by the Imaris Surface Creation module (mean and individual data points, n=6 separate embryos of each genotype, ***P<0.001, t-test). Scale bars: 1 mm.
transfected by electroporation, after which they were plated in NGFcontaining medium and were imaged 16 h later. We used combinations of three different specific 21-mer siRNA oligonucleotides that have been individually validated by the manufacturer to knock down expression of the target protein in cell lines. Although it was not possible to knock down TWE-PRIL without knocking down either APRIL or TWEAK, by using siRNAs that target the parts of April and Tweak mRNAs that do not encode part of TWE-PRIL it was possible to knock down either APRIL or TWEAK without affecting TWE-PRIL expression (Fig. 4A). To control for non-specific effects of siRNA transfection, we transfected SCG neurons with a 21-mer doublestranded RNA oligonucleotide that has a scrambled sequence.
We used immunocytochemistry to confirm selective knockdown of these proteins in transfected neurons. BCMA siRNAs greatly reduced BCMA immunoreactivity, TWEAK siRNAs eliminated TWEAK immunoreactivity, and either APRIL siRNAs or APRIL/TWE-PRIL siRNAs eliminated APRIL/TWE-PRIL immunoreactivity (Fig. S2). TWEAK siRNAs did not affect APRIL/TWE-PRIL immunoreactivity and APRIL siRNAs did not affect TWEAK immunoreactivity (Fig. S3).
Neurons transfected with siRNAs that reduce expression of TWE-PRIL and APRIL (siRNAs that targeted sequences in exons 4 and 6 of the April gene, equivalent to exons 11 and 13 of the Twe-pril primary transcript, Fig. 4A) had significantly longer axons than those of neurons transfected with control RNA oligonucleotides (Fig. 4B,C). In contrast, neurons transfected with siRNAs that selectively knock down APRIL expression without affecting TWE-PRIL expression (siRNAs that targeted sequences in exon 1 of the April gene, Fig. 4A) had axons that were no longer than those of neurons transfected with control RNA oligonucleotides (Fig. 4B,C). Likewise, siRNA that specifically knocked down expression of TWEAK without affecting TWE-PRIL expression (siRNAs that targeted sequences in exon 7 of the Tweak gene) had no significant effect on axon growth (Fig. 4C). Knocking down expression of both APRIL and TWEAK without affecting TWE-PRIL expression (co-transfecting neurons with the siRNAs that selectively target APRIL plus siRNAs that target TWEAK) also had no significant effect on axon growth. However, co-transfecting neurons with the siRNAs that reduce expression of TWE-PRIL and APRIL plus the siRNAs that target TWEAK had had significantly longer axons than those of neurons transfected with control RNA oligonucleotides (Fig. 4C). Taken together, these findings suggest that TWE-PRIL, but neither APRIL nor TWEAK, participates in the suppression of NGFpromoted axon growth from cultured SCG neurons. This in turn suggests that absence of TWE-PRIL, rather than absence of APRIL, underlies the in vivo sympathetic phenotype of April −/− embryos.
In accordance with differences in the expression of BCMA and TACI (Fig. 1A), siRNAs that specifically targeted BCMA, but not siRNAs that targeted TACI, significantly increased axon growth from transfected SCG neurons (Fig. 4B,C). This suggests that BCMA, but not TACI, participates in the suppression of NGFpromoted axon growth from developing SCG neurons.
Rescue of the in vitro phenotype of April −/− SCG neurons by expressing TWE-PRIL The above findings suggest that the interaction between TWE-PRIL and BCMA suppresses NGF-promoted axon growth. However, these findings do not indicate the direction of signalling: either TWE-PRIL-activated BCMA-mediated forward signalling or BCMA-activated TWE-PRIL-mediated reverse signalling. To distinguish between these alternatives, we asked whether or not soluble APRIL was sufficient to reverse the enhanced NGFpromoted axon growth from April −/− SCG neurons (which would be consistent with forward signalling) or whether expression of membrane integrated TWE-PRIL is required.
The axon arbors of NGF-supplemented P0 April −/− SCG neurons were very similar with or without soluble recombinant APRIL in the culture medium (Fig. 5A). This suggests that soluble APRIL is unable to rescue the enhanced NGF-promoted axon growth phenotype of cultured April −/− SCG neurons. In contrast, the axon arbors of NGF-supplemented P0 April −/− SCG neurons transfected with a full-length TWE-PRIL expression plasmid were significantly shorter than those of control-transfected April −/− SCG neurons (neurons transfected with an empty expression plasmid) and were not significantly different from controltransfected April +/+ SCG neurons (Fig. 5B,C). This suggests that expression of full-length TWE-PRIL completely rescued the enhanced NGF-promoted axon growth phenotype of April −/− SCG neurons. These findings are consistent with the idea that TWE-PRIL-mediated reverse signalling suppresses NGFpromoted axon growth.
A wealth of evidence suggests that the intracellular domain of TNFSF members that are able to mediate reverse signalling is essential for this mode of signalling (Sun and Fink, 2007). To test the importance of the intracellular domain of TWE-PRIL for reverse signalling, we transfected neurons with a plasmid that expresses a truncated form of TWE-PRIL that lacks the entire intracellular domain derived from TWEAK apart from the initiation codon of this type II glycoprotein but retains the transmembrane domain derived from TWEAK and the extracellular domain derived from APRIL (ΔTWE-PRIL). The size of the axon arbors of NGFsupplemented P0 April −/− SCG neurons transfected with this expression plasmid were not significantly different from those of control-transfected April −/− SCG neurons (Fig. 5C). Expression of neither full-length TWEAK nor full-length BCMA were able to substitute for full-length TWE-PRIL in suppressing enhanced NGFpromoted axon growth from April −/− SCG neurons (Fig. 5C).
Interestingly, transfecting April −/− SCG neurons with a plasmid that expresses full-length APRIL was able to prevent enhanced NGF-promoted axon growth from April −/− SCG neurons. This suggests that overexpressed APRIL is able to mediate reverse signalling and is able to substitute for TWE-PRIL in suppressing NGF-promoted axon growth from April −/− SCG neurons.
Soluble APRIL replicates the in vitro phenotype of April −/− SCG neurons Although the above findings are consistent with the operation of BCMA-activated TWE-PRIL-mediated reverse signalling, in order to provide an additional test that signalling occurs in this direction, we studied the effects of soluble recombinant APRIL on NGFpromoted axon growth from wild-type SCG neurons. We reasoned that soluble APRIL should compete with endogenous membrane integrated TWE-PRIL for binding to BCMA. This should interfere with endogenous BCMA-activated TWE-PRIL-mediated reverse signalling, but should not inhibit BCMA-mediated forward signalling. Soluble APRIL caused a dose-dependent increase in NGF-promoted axon length, reaching a plateau at 100 ng/ml APRIL (Fig. 6A,B). The period of development over which soluble APRIL treatment enhanced NGF-promoted axon growth from wild-type SCG neurons was similar to the period of development over which SCG neurons cultured from April −/− mice displayed increased responsiveness to NGF. Sholl plots of NGF-supplemented E16, P0, P3 and P5 wild-type SCG neurons grown with and without soluble APRIL showed that the enhancement of NGF-promoted axon growth by APRIL was greatest in E16 and P0 cultures (Fig. 6C). Moreover, the magnitude by which APRIL enhanced NGFpromoted axon growth from wild-type SCG neurons was very similar to the magnitude of NGF-promoted axon growth from SCG neurons cultured from April −/− mice, compared with the magnitude of NGF-promoted axon growth from SCG neurons cultured from April +/+ mice (compare Figs 3B and 6C). The influence of soluble APRIL on NGF-promoted axon growth from wild-type SCG neurons was also related to NGF concentration. Although APRIL had no effect on axon growth from SCG neurons grown with low levels of NGF, it progressively enhanced NGF-promoted axon growth with increasing NGF concentration (Fig. 6D), as is the case with SCG neurons of April −/− mice compared with SCG neurons of April +/+ mice. As is the case with absence of APRIL, treatment with soluble APRIL had no effect on NGF-promoted survival. In the absence of Boc _ D _ FMK, overlapping NGF survival dose responses were observed from P0 SCG neurons grown with and without APRIL (Fig. 6E). Taken together, the above results indicate that soluble APRIL enhances NGF-promoted axon growth from wild- type SCG neurons, but not NGF-promoted survival, over the same period of development and to the same extent as a null mutation in the April gene.

TWE-PRIL reverse signalling suppresses axonal growth by supressing ERK1/ERK2 activation in response to NGF
Because activation of the MEK/ERK pathway contributes to the axon growth response to NGF (Gómez and Cohen, 1991;Goold and Gordon-Weeks, 2005;O'Keeffe et al., 2008) and is an essential step in the axon growth response of sympathetic neurons to TNFR1activated TNFα-mediated reverse signalling (Kisiswa et al., 2017), we investigated whether TWE-PRIL/APRIL modulates ERK activation by NGF and whether this plays a role in axon growth suppression by TWE-PRIL reverse signalling. We began by investigating whether the expression of TWE-PRIL/APRIL influences ERK activation by NGF. For these studies, we used western blotting to compare ERK activation in response to NGF in SCG neurons cultured at high density from April +/+ and April −/− mice. These studies revealed that the level of phospho-ERK1/ERK2 relative to total ERK1/ERK2 was significantly greater in NGF stimulated P0 SCG neurons cultured from April −/− mice compared with those cultured from April +/+ littermates (Fig. 7A,B).
To investigate whether suppression of NGF-stimulated ERK1/ ERK2 activation by TWE-PRIL reverse signalling contributes to axon growth suppression, we studied the effect of U0126, a selective MEK1/MEK2 inhibitor that interferes with MEK1/MEK2 dependent activation of ERK1/ERK2, on axon growth from SCG in which TWE-PRIL reverse signalling was either permitted or prevented. In these experiments, P0 SCG neurons from April −/− and April +/+ mice were plated at low density in medium containing either U0126 or U0124, and were treated with NGF after 1 h. Half of the cultures were additionally treated with soluble APRIL at this time to inhibit TWE-PRIL reverse signalling. Neurons in all cultures were imaged 16 h after plating. The control compound U0124 did not affect the marked difference in axon length between neurons cultured from April −/− and April +/+ mice. However, axon length in cultures established from April −/− mice was significantly reduced by U0126 to the level observed from April +/+ neurons cultured with U0124 (Fig. 7C). U0126 also significantly reduced the enhanced axon growth of wild-type neurons treated with APRIL to the level observed in wild-type neurons that had not been treated with APRIL (Fig. 7C). Taken together, these findings suggest that suppression of NGF-stimulated ERK activation contributes to the suppression of axon growth by TWE-PRIL reverse signalling.
We also explored the possibility that differences in the expression of NGF receptors might account for the differences in the activation of ERK1/ERK2 in April −/− and April +/+ SCG neurons by NGF. Western blotting revealed there were no differences in the intensity of the bands corresponding to TrkA (the NGF receptor tyrosine kinase) in SCG neurons from April −/− and April +/+ mice grown with different levels of NGF (Fig. 7A). In addition, qPCR revealed that there were no significant differences in the levels of TrkA mRNA and p75 mRNA (which encodes the common neurotrophin receptor) between April −/− and April +/+ SCG (Fig. S4). DISCUSSION We have shown that April-null mice, which lack both APRIL and TWE-PRIL, have significantly longer sympathetic fibres in vivo at the stage when sympathetic axons are starting to ramify in their targets under the influence of target-derived NGF. This implies that APRIL and/or TWE-PRIL are physiologically relevant suppressors of sympathetic fibre elongation in vivo. The axon arbors of sympathetic neurons cultured from April −/− mice were significantly larger than those of wild-type littermates grown in the presence of NGF, but were not significantly different in the absence of NGF. This suggests that APRIL and/or TWE-PRIL suppress NGFpromoted axon growth from developing sympathetic neurons. The difference in NGF-promoted axon growth between SCG neurons of April +/+ and April −/− mice was observed over a window of development from late foetal to early postnatal stages. Moreover, whereas the axon growth response of SCG neurons to NGF was affected by the April-null mutation, the survival response of these neurons to NGF was completely unaffected. This suggests that APRIL and/or TWE-PRIL suppress NGF-promoted axon growth, Fig. 7. NGF-stimulated ERK activation is enhanced in April −/− neurons and inhibition of ERK activation eliminates the differences in axon growth between April −/− and April +/+ neurons. (A) Representative western blots probed for phospho-ERK1/ERK2, total ERK1/ERK2, TrkA and β-III tubulin of lysates of P0 SCG neurons of April −/− and April +/+ mice that had been incubated with 25 µM Boc-D-FMK for 12 h before either 30 min treatment with NGF (0.1 or 10 ng/ml) or no treatment. (B) Densitometry of three separate experiments (mean±s.e.m. of phospho-ERK1/ERK2 relative to total ERK1/ERK2). **P<0.01, statistical comparison indicated. (C) Bar charts of total length of axons in the arbors of P0 SCG neurons of April −/− and April +/+ mice that were plated in medium containing either 10 µM U0126 or 10 µM U0124 and were treated with either 0.1 ng/ml NGF or 0.1 ng/ml NGF plus 100 ng/ml APRIL 1 h after plating. Axon arbors were imaged 16 h later. Mean±s.e.m. of data from over 50 neurons per condition (***P<0.001, statistical comparison indicated, ANOVA).
but not NGF-promoted neuron survival. These features echo emerging themes in the influence of the TNFSF on axon growth in the developing PNS. Whereas several TNFSF and TNFRSF members promote or inhibit axon growth, they have no or negligible effect on neuron survival (Gavalda et al., 2009;Gutierrez et al., 2013;Kisiswa et al., 2013;McWilliams et al., 2015;O'Keeffe et al., 2008). The effect of these factors on axon growth is transient. Typically, the neurons begin responding to these factors several days after they have started to respond to neurotrophins, and the developmental window over which they exert their effects is short compared with the extended period over which neurotrophins act. The transient influence of the TNFSF/TNFRSF member on axon growth in the developing PNS generally coincides with the perinatal period when axons are ramifying extensively in their targets. It is therefore likely that the TNFSF plays an important role refining innervation during this crucial period of development.
The enhanced NGF-promoted axon growth phenotype of SCG neurons cultured from April −/− mice was replicated by transfecting wild-type SCG neurons with siRNA that knock down expression of both TWE-PRIL and APRIL, but was not observed in SCG neurons transfected with siRNA that selectively knock down either APRIL or TWEAK. This suggests that TWE-PRIL, but neither APRIL nor TWEAK, is responsible for suppressing NGF-promoted axon growth. This in turn suggests that the increased growth of SCG nerve fibre branches in April −/− embryos in vivo is due to the absence of TWE-PRIL rather than the absence of APRIL.
TWE-PRIL comprises the extracellular ligand-binding domain of APRIL fused to the cytoplasmic and transmembrane domains of TWEAK (Kolfschoten et al., 2003;Pradet-Balade et al., 2002), and binds the same two members of the TNFRSF as APRIL, BCMA and TACI. Several findings suggest that BCMA, rather than TACI, is required for the suppression of NGF-promoted axon growth by TWE-PRIL. First, whereas Bcma mRNA is present in developing SCG from E13 onwards, Taci mRNA was barely detectable at any age. Second, BCMA immunoreactivity was clearly present in cultured neonatal SCG neurons, whereas TACI immunoreactivity was barely above background. Third, transfecting SCG with siRNA that knock down BCMA enhanced NGF-promoted axon growth as effectively as siRNA knockdown of TWE-PRIL, whereas siRNA directed against TACI had no effect on NGF-promoted axon growth.
Two observations suggest that BCMA and TWE-PRIL modulate axon growth by an autocrine mechanism. First, almost all neonatal SCG neurons were labelled by antibodies to BCMA and TWE-PRIL/APRIL, suggesting that these neurons co-express BCMA and TWE-PRIL/APRIL. Second, all in vitro experiments were carried out on SCG neurons cultured at exceptionally low density. As such, the neurons were non-contiguous and were cultured in a relatively large volume of medium. Co-labelling and very low-density culture studies of sympathetic and sensory neurons have likewise concluded that several members of the TNFSF and TNFRSF families are likely to exert their effects on axon growth by an autocrine mechanism. These include enhanced NGF-promoted sympathetic axon growth by GITRL/GITR (O'Keeffe et al., 2008) and CD40L/CD40 (McWilliams et al., 2015), and suppressed BDNF-promoted sensory axon growth by LIGHT/HVEM (Gavalda et al., 2009). In addition to autocrine signalling, the TNFSF/ TNFRSF can also regulate axon growth and tissue innervation by a paracrine mechanism, e.g. target-derived TNFR1 enhances sympathetic axon growth by activating TNF reverse signalling in axons (Kisiswa et al., , 2017.
It is likely that autocrine interaction of TWE-PRIL and BCMA suppresses NGF-promoted axon growth by a BCMA-activated TWE-PRIL-mediated reverse signalling mechanism. First, the enhanced NGF-promoted axon growth phenotype of SCG neurons cultured from April −/− mice is completely rescued by transfecting the neurons with a TWE-PRIL expression plasmid but not by treating the neurons with soluble APRIL. Because the expressed TWE-PRIL is able to integrate into the membrane it has the potential to mediate reverse signalling. Soluble APRIL, on the other hand, is not membrane integrated and only has the potential to activate BCMA-mediated forward signalling. Second, soluble APRIL enhances NGF-promoted axon growth of wild-type SCG neurons as effectively as deletion of April/Twe-pril or siRNA knockdown of TWE-PRIL. The most parsimonious explanation for this result is that soluble APRIL competes with endogenous membrane-integrated TWE-PRIL for binding to endogenous BCMA, and thereby inhibits endogenous BCMA from activating TWE-PRIL-mediated reverse signalling. Furthermore, these experiments also control for the efficacy of the soluble APRIL used in the phenotype rescue experiments.
We showed that enhanced NGF-promoted axon growth from April −/− SCG neurons can also be eliminated by transfecting the neurons with an APRIL expression plasmid. This suggests that membrane-integrated APRIL is also able to mediate reverse signalling and can substitute for TWE-PRIL in rescuing the phenotype of April −/− SCG neurons in vitro. However, siRNA that specifically knocks down APRIL in wild-type SCG neurons and leaves TWE-PRIL unaffected does not affect NGF-promoted axon growth. This suggests that endogenous APRIL reverse signalling does not normally contribute to the regulation of NGF-promoted axon growth. This apparent discrepancy may be because endogenous APRIL may be expressed at a much lower level than TWE-PRIL in developing SCG neurons. Moreover, it has been shown in HEK293 cells that APRIL is cleaved in the Golgi apparatus by a furin convertase to generate a biologically active secreted protein (Lopez-Fraga et al., 2001). This implies that APRIL is not expressed at the cell surface and therefore could not mediate reverse signalling. Because APRIL expression in our transfection experiments is under the control of a strong promoter, APRIL may have been expressed at levels that exceed the capacity of the Golgi apparatus to process it. Although our data do not support a role of endogenous APRIL-mediated reverse signalling in modulating NGF-promoted axon growth in SCG neurons, we cannot exclude the possibility that APRIL reverse signalling contributes to the regulation of sympathetic axon growth in vivo.
Contrary to the view that APRIL is produced solely as a secreted ligand (Lopez-Fraga et al., 2001), more recent studies have reported APRIL expression at the cell membrane of monocyte and macrophage cell lines, and primary macrophages (Lee et al., 2010a, b). Furthermore, it has been reported that TACI-Fc and an anti-APRIL antibody promote expression of pro-inflammatory mediators and inhibit phagocytosis and chemotaxis in the THP-1 monocyte cell line, suggesting that these effects are mediated by APRIL reverse signalling (Lee et al., 2010a). APRIL reverse signalling could not, however, be demonstrated in primary macrophages (Nys et al., 2013).
Several recent studies have reported that reverse signalling by members of the TNFSF is physiologically relevant for regulating axon and dendrite growth. TNFR1-activated TNF reverse signalling enhances the growth of sympathetic and sensory axons in vitro and plays a role in establishing target field innervation in vivo (Kisiswa et al., , 2017Wheeler et al., 2014). CD40-activated CD40L reverse signalling enhances sympathetic axon growth in vitro and promotes the sympathetic innervation of a subset of tissues in vivo (McWilliams et al., 2015). CD40L reverse signalling has also been recently shown to exert opposite effects in vivo on the growth and complexity of excitatory hippocampal pyramidal neuron dendrites and inhibitory striatal medium spiny neuron dendrites (Carriba and Davies, 2017).
We have shown that modulation of NGF-stimulated ERK activation plays a key role in mediating the effects of TWE-PRIL reverse signalling on axon growth. Likewise, modulation of ERK activation is a key step in effecting the influence of TNFR1activated TNF reverse signalling on axon growth . However, whereas TNF reverse signalling enhances NGFstimulated ERK activation, TWE-PRIL reverse signalling appears to have the opposite effect. Activation of PKC is a key step in ERK activation by TNF reverse signalling, and PKC activation has been shown to mediate the effects of CD40L reverse signalling on dendrite growth in the CNS. However, different PKC isoforms have been implicated in mediating the promotion and inhibition of dendrite growth by CD40L reverse signalling in different CNS neurons (Carriba and Davies, 2017). Whether PKC and whether distinct PKC isoforms play a role in mediating the axon growth inhibitory effects of TWE-PRIL reverse signalling are interesting questions for future research as is the potential modulation of other signalling pathways that influence axon growth.
In summary, our data suggest that TWE-PRIL reverse signalling is a physiologically relevant regulator of NGF-promoted sympathetic axon growth (Fig. 8). As TWE-PRIL reverse signalling has yet to be reported in any system, our study not only shows that TWE-PRIL can mediate reverse signalling, but also provides evidence for the first physiological role for TWE-PRIL.

Animals
This study was conducted on tissues obtained from CD1 wild-type mice (Mus musculus) and mice with a null mutation in the April gene (a gift from Raif Geha, Boston Children's Hospital, Harvard Medical School, Cambridge, MA, USA) that were backcrossed into a CD1 background. The April-null mice lack half of exon 3 and exons 4 to 6 of the April gene, which encode 126 amino acids of the extracellular domain (Castigli et al., 2004). As such, these mice lack both APRIL and TWE-PRIL. Breeding, housing and genotyping was approved by the Cardiff University Ethical Review Board and was performed within the guidelines of the Home Office Animals (Scientific Procedures) Act, 1986.
Neuron culture SCG were dissected from embryonic and postnatal CD1 mice. The ganglia were trypsinized and plated at very low density (∼500 neurons per dish) in poly-ornithine/laminin-coated 35 mm or four-well tissue culture dishes (Greiner) in serum-free Hams F14 medium (Davies et al., 1993) supplemented with 0.25% Albumax I (Life Technologies).
Neuronal survival was estimated by counting the number of attached neurons within a 12×12 mm grid in the centre of each dish 2 h after plating and again after 24 h, and expressing the 24 h count as a percentage of the 2 h count. Analysis of the size and complexity of axon arbors was carried out by labelling the neurons at the end of the experiment with the fluorescent vital dye calcein-AM (Life Technologies). For every condition in each experiment, images of at least 50 neurons were digitally acquired by fluorescence microscopy and analysed to obtain total axon length, number of branch points and Sholl profiles (Gutierrez and Davies, 2007). Statistical analyses were performed using one way ANOVA with Bonferroni-Holm post hoc test. Pair-wise comparisons were made using Student's t-test.
For siRNA experiments, dissociated suspensions of SCG neurons were electroporated with Silencer Select siRNAs (Life Technologies) directed against TWE-PRIL/APRIL, APRIL alone, TWEAK alone, BCMA or TACI, or with a scrambled control siRNA. Co-transfection of a GFP expression plasmid allowed identification of the transfected neurons and imaging of their processes.
Recombinant NGF and APRIL were obtained from R&D Systems. Where indicated, the culture medium was supplemented with the caspase inhibitor Boc-D-FMK (Calbiochem) to prevent neuronal apoptosis in the absence of or in low concentrations of NGF. Expression plasmids were obtained from OriGene and the truncated TWE-PRIL construct was synthesized by Eurofins.

Analysis of SCG fibres in vivo
Analysis of sympathetic fibre growth from the SCG was carried out by immunolabelling-enabled three-dimensional imaging of solvent-cleared organs (iDISCO) (Renier et al., 2014) of E16 April +/+ and April −/− heads. Briefly, the heads were fixed in 4% paraformaldehyde for 24 h and serially dehydrated in a series of increasing methanol concentrations in water. The samples were then incubated overnight in 66% dichloromethane (DCM)/ 33% methanol, bleached in chilled 5% H 2 O 2 to reduce tissue autofluorescence and then serially rehydrated in a series of decreasing methanol concentrations in water. After rehydration, the heads were washed in PBS containing 0.2% Triton X-100 (PTx.2) and were incubated in permeabilization solution (PTx.2, 20% DMSO and 0.3 M glycine) for 36 h at 37°C. After permeabilization, the samples were washed twice in PTx.2 and then incubated in blocking solution (6% donkey serum, 10% DMSO in PTx.2) for 36 h at 37°C. The heads were then incubated with rabbit polyclonal anti-tyrosine hydroxylase antibody (1:300, AB152, Millipore) in PTwH (0.2% Tween-20 and 10 mg/ml heparin in PBS) containing 5% DMSO and 3% donkey serum for 72 h at 37°C. Following extensive washing in PTwH, the samples were incubated with donkey antirabbit 647 Alexa Fluor secondary antibody (1:300, A-31573, Life Technologies) in PTwH plus 3% donkey serum for 72 h at 37°C. After further washing in PTwH, samples were cleared by dehydration in a series of methanol concentrations in water followed by 3 h incubation in 66% DCM/ 33% methanol. After washing in 100% DCM, the samples were placed in dibenzyl ether (DBE) until clear and imaged while submerged in DBE.
The specimens were imaged using a LaVision Light Sheet Ultramicroscope. Optical sections (5 µm) were acquired using a Neo high sensitivity CMOS camera. Captured z-axis images were analysed with Imaris 9 software (Bitplane Scientific Solutions), which allows for reconstruction of TH-positive fibres in three dimensions. The Filament Tracer module was used to detect fibres within the selected region of interest with semi-manual segmentation and the Surface Creation Module allowed for automized three-dimensional computerized reconstructions of THpositive fibres after manual threshold optimization (Schindelin et al., 2012). The April −/− data are expressed as a percentage of the mean of April +/+ data. All imaging and quantification was performed blind with genotypes determined after TH quantification.

RT-qPCR
The levels of April, Twe-pril, Tweak, Bcma, Taci, TrkA and p75mRNAs were quantified by RT-qPCR in total RNA extracted from dissected SCG relative to a geometric mean of mRNAs encoding the housekeeping enzymes glyceraldehyde phosphate dehydrogenase (GAPDH) and succinate dehydrogenase (SDHA). Total RNA was extracted from dissected tissues with the RNeasy Mini Lipid extraction kit (Qiagen) and 5 µl was reverse transcribed for 1 h at 45°C using the AffinityScript kit (Agilent) in a 25 µl reaction according to the manufacturer's instructions. cDNA (2 µl) was amplified in a 20 µl reaction volume using Brilliant III