Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Sci Signal. Author manuscript; available in PMC 2013 May 6.
Published in final edited form as:
PMCID: PMC3645873

Signaling to Transcription Networks in the Neuronal Retrograde Injury Response


Retrograde signaling from axon to soma activates intrinsic regeneration mechanisms in lesioned peripheral sensory neurons; however, the links between axonal injury signaling and the cell body response are not well understood. Here, we used phosphoproteomics and microarrays to implicate ~900 phosphoproteins in retrograde injury signaling in rat sciatic nerve axons in vivo and ~4500 transcripts in the in vivo response to injury in the dorsal root ganglia. Computational analyses of these data sets identified ~400 redundant axonal signaling networks connected to 39 transcription factors implicated in the sensory neuron response to axonal injury. Experimental perturbation of individual overrepresented signaling hub proteins, including Abl, AKT, p38, and protein kinase C, affected neurite outgrowth in sensory neurons. Paradoxically, however, combined perturbation of Abl together with other hub proteins had a reduced effect relative to perturbation of individual proteins. Our data indicate that nerve injury responses are controlled by multiple regulatory components, and suggest that network redundancies provide robustness to the injury response.


There are marked differences in the ability of different classes of neurons in the adult mammalian nervous system to regenerate after injury. Functional regeneration depends on the ability of injured neurons to integrate stimulatory and inhibitory signals from environmental cues (1, 2) and to mobilize intrinsic neurite growth programs (3). Initiation of intrinsic repair mechanisms requires that the neuronal cell body changes growth patterns in response to an axonal injury. A well-described example of such a switch is the “conditioning lesion” phenomenon observed in peripheral sensory neurons, which switch from arborizing to elongating process growth if previously subjected to a conditioning injury in vivo (4, 5). The conditioning lesion paradigm is widely used as a model to understand early events in the initiation of neuronal regeneration. Studies using this and other models have shown that information about the injury must be conveyed from the axonal lesion site to the cell body through retrograde transport mechanisms based on dynein-mediated trafficking of molecular signaling complexes (6, 7). Such complexes can assemble on linkers consisting of importins (8, 9), cytoskeletal components (10), or kinase signaling scaffolds (11, 12), which allow trafficking of various phosphoprotein signals (13). For example, a number of mitogen-activated protein kinases (MAPKs) are involved in regeneration and repair mechanisms in affected neurons from various species (1417), and MAPK family members are retrogradely transported after peripheral nerve injury (10, 1820).

Retrograde signals induce transcriptional reprogramming, thereby eliciting the regeneration response. Changes in the cell body that occur in response to retrograde injury signaling may include increased production of various molecules, including transcription factors (TFs), cytoskeletal proteins, cell adhesion and axon guidance molecules, and trophic factors and their receptors (3). A principal goal in the field has been to identify genes and proteins that regulate the outgrowth and regeneration of axons. Recent work has shown that the cell body response to injury can be reactivated by additional lesions weeks to months after the original injury (21, 22), giving hope that detailed understanding of this process might lead to clinical approaches applicable for both newly injured and long-term patients. Thus, identification of the main molecular determinants of the injury response is of both basic and potential clinical interest. Straightforward comparisons of injured versus uninjured neurons in differential gene screens are complicated by the fact that it is difficult to identify the genes responsible for regeneration in the context of the many transcriptional events induced by the stress of injury or by the changed connectivity and signaling of affected cells. Furthermore, in vivo screens are necessarily carried out at the tissue level, hence on heterogeneous mixtures of responsive and nonresponsive cells (23). Consequently, screens of differential gene expression after nerve injury have mostly resulted in long lists of genes encoding proteins associated with regeneration, without a clear identification of genes that encode proteins that might induce regeneration (3).

The difficulty of teasing out the effects of individual genes and proteins from the massive changes that occur after neuronal injury has led to the hope that it may be possible to identify “master regulators” that coordinate the regenerative program in neurons. The hypothesis underlying these efforts is that a relatively small complement of TFs or proteins that function as critical nodes in signaling or transcription networks (or both) might control much of the functional response to nerve lesions. Although a number of candidate regulators have been suggested (2431), thus far there has been no truly comprehensive attempt to delineate the combined signaling and transcriptional networks governing the neuronal response to injury. Here, we investigated the response to lesion in peripheral sensory neurons—a well-established model for studies of neuronal development, growth, and injury (6, 32)–to define networks linking axonal signaling to transcription in the cell body. Neuronal cell bodies in the L4 and L5 lumbar dorsal root ganglia (DRG) extend sensory axons to the periphery through the sciatic nerve. We combined analysis of the sciatic nerve axonal phosphoproteome with that of the L4 and L5 DRG transcriptome to identify transcriptional signatures enriched in injury-regulated genes and construct signaling networks connected to TFs identified from the transcriptome data. These combined signaling to transcription networks enabled us to identify candidate proteins involved in regulating the initiation of neurite regeneration.


Phosphoproteomics of axonal retrograde injury signaling in the sciatic nerve

We used an established and validated lesion-ligation model (33) to analyze the phosphoprotein composition of axoplasm accumulated at ligatures 24 hours after crush injury of rat sciatic nerve. The nerve was crushed 4 to 5 cm from the L4 and L5 DRG, and the ligation was placed ~2 cm proximal to the crush site. Comparison of ligature-enriched protein ensembles from uninjured and injured nerve allowed discrimination of four experimental groups: proximal and distal noninjured (NIP and NID, respectively) and proximal and distal injured (INP and IND, respectively). Each of these four groups is characterized by a distinct combination of injured versus uninjured anterograde and retrograde signals (Fig. 1A) [see (33)]. Phosphoproteins from these four groups were enriched and fractionated and then analyzed by tandem mass spectrometry (Fig. 1B and fig. S1). The analyses revealed ~5500 phosphopeptides altogether, constituting ~60% of all identified peptides and corresponding to a total of 1447 proteins (Fig. 1C), some of which are shared between groups. About 80% of the proteins were represented by one or two phosphopeptides, with the remainder showing up to 10 phosphopeptides per protein (Fig. 1D). Serine was by far the most prominent phosphorylation site represented in the data set, followed by threonine, with a small number of tyrosine phosphorylations (Fig. 1D and fig. S1).

Fig. 1
Phosphopeptide enrichment scheme and overview of retrograde injury phosphoproteome. (A) Schematic of experimental lesion-ligation model using rat sciatic nerve. NR, noninjured axon, retrograde signal; IR, injured axon, retrograde signal; NA, non-injured ...

This experimental model enables the identification of changes in protein ensembles undergoing retrograde transport after nerve injury through a comparison of distal ligation samples (33). Comparison of phosphopeptides in IND versus NID samples revealed 825 unique phosphorylations on 685 proteins, 417 of which were in NID and 268 in IND. Of these differentially phosphorylated proteins, 164 contained two or more phosphorylation sites, and ~40% of the proteins in this group were phosphorylated on different sites in the injured and noninjured samples. Overall, ~40% of the assigned phosphorylation sites have been previously reported or could be inferred by similarity to records in databases, whereas the remaining ~60% had not previously been reported. A comprehensive list of the phosphoproteins and phosphorylation sites used for further analyses is shown in table S1. Gene Ontology (GO) analyses of phosphoproteins with mutually exclusive phosphorylation sites in the injured and noninjured groups showed the highest enrichment for proteins in the signal transduction, guanosine triphosphatase activity, microtubule-related transport, and metabolism GO subcategories (table S1).

Transcriptomics of cell body responses in the DRG

We performed microarray analyses on adult rat L4 and L5 DRG after sciatic nerve lesion. In contrast to most previous differential gene expression studies of nerve injury, we examined a time frame of hours after injury to focus on transcriptional events that are a direct consequence of retrograde signals elicited by the lesion. We used Affymetrix Rat 230.2 arrays containing 31,100 probe sets to screen changes in gene expression in L4 and L5 DRG over a time course of 1 to 28 hours after a crush lesion of the sciatic nerve ~4 cm from the DRG. Array data were preprocessed and normalized, and differential expression was determined by analysis of variance (ANOVA) and false discovery rate (FDR), followed by K-means clustering. The 1-hour time point did not reveal significant differences in DRG gene expression relative to sham-operated controls and was subsequently used as a reference for comparison of gene expression over the course of the analysis. Analyses using P values up to 0.01 or FDR cutoffs below 15%, or both, showed that a widespread transcriptional response commenced 8 to 12 hours after injury, with the most robust responses apparent at 18 to 28 hours. This time lag might indicate a requirement for the arrival of dynein-driven retrograde signals to initiate the main transcriptional response. Heat maps of changes in gene expression that passed ANOVA and FDR tests at q = 0.001 are shown in Fig. 2. We subdivided the 1490 genes showing increased expression in response to injury into four clusters based on the kinetics of their up-regulation (Fig. 2A) and grouped the 1210 genes showing decreased expression into three clusters based on the kinetics of expression over the time course we evaluated (Fig. 2B). The corresponding gene lists are shown in table S2, and the annotated portions of these lists (~60% of the data sets) were used as the transcriptome input for subsequent computational analyses. Because of the high sensitivity of the ANOVA test, we chose a strict cutoff in q to avoid identification of a large number of false-positive genes. Thus, although the identified clusters contain the genes showing the most significant changes in expression, the total number of responding genes may actually be larger.

Fig. 2
Profiling of gene expression in DRG after sciatic nerve injury. (A) Heat map representation of K-means clustering of genes up-regulated in L4 and L5 DRG after sciatic nerve injury that passed ANOVA and FDR criteria as described in Materials and Methods. ...

TF-binding site analyses

We began our identification of regulatory components in the system by analyzing the promoter regions of co-regulated genes in the identified clusters to investigate the possible involvement of specific TFs in response to injury. Querying TF-binding site (TFBS) databases with the promoter sequences of co-regulated genes allows the identification and prioritization of TFs as candidate control nodes in the ensemble of transcriptional interactions occurring in a given system, known as the transcriptional networks (34, 35). The most widely used approach to reveal TFBS in the promoter regions of coexpressed genes is based on positional weight matrices (PWMs) constructed from collections of known binding sites for a given TF or TF family. TRANSFAC is the largest available collection of TFBSs and corresponding PWMs and has been used for analyses of regulatory regions in diverse gene classes (35, 36). In addition to the ANOVA analysis described above, we also generated sets of the genes showing the greatest difference in expression (induced or repressed) between injured and uninjured nerve for each individual time point by using two-sided t tests to compare replicates from each time point to the reference time point replicates, and FDR to control for multiple hypotheses. We scanned the gene sets thus obtained for overrepresentation of all available TFBS matrices curated in TRANSFAC 9 by comparison to equivalently sized control sets of nonresponding genes. This analysis revealed ~100 TFBS matrices that were significantly overrepresented in gene sets differentially regulated by injury. Because many of the matrices were highly redundant, these 100 TFBSs correspond to 453 TF candidates from 118 different families (fig. S2).

Intersecting signaling and transcriptional networks

We used the TRANSPATH database (37) as implemented in ExPlain to construct signaling networks consistent with the axoplasm phosphoproteomic data. TRANSPATH-generated networks such as that shown in Fig. 3A could be found for ~200 of the retrogradely transported phosphoproteins in our data sets. The redundancy and extensive cross talk characteristic of signal transduction networks is reflected in the fact that >1800 networks passed statistical significance filters for ~200 phosphoproteins (Fig. 3B). At this stage of the analysis, we juxtaposed the axonal signaling networks and ganglia gene regulation data sets. This intersection reduced the number of candidate networks to ~500, covering a wide range of signaling pathways (Fig. 3B, fig. S3, and table S3). The analyses revealed that multiple networks are implicated in the ganglionic transcriptional response at any given time point (Fig. 3C and table S4). The multiplicity of gene regulation events associated with each network at the different time points does not permit assignment of specific networks solely to increases or decreases in overall gene expression in this system.

Fig. 3
Linking axonal phosphoproteomes to cell body transcriptomes. Overlap analyses linking cell signaling networks based on phosphoproteins correlated with retrograde injury with activation or repression pathways for TFs implicated in the ganglia response. ...

The intersecting signaling and transcription data sets reduced the number of statistically significant candidate TFs from 453 (fig. S2) to 39 (Fig. 4), corresponding to 26 TF families with 44 TFBS matrices. Time-dependent clustering of these signaling network–associated TFs showed that nearly all of them are downstream of multiple signaling networks (Fig. 4). Two TF clusters showed a high correlation with changes in gene expression at later time points (18 to 28 hours after injury); these clusters included TFs in the STAT (signal transducer and activator of transcription), HNF (hepatocyte nuclear factor), USF (upstream stimulatory factor), Jun, Smad, SRF (serum response factor), and ER-α (estrogen receptor–α) families. Each network cascade involved multiple TFs, ranging from a handful per network to >30. Overall, the intersecting signaling to transcription networks covered ~28% of the retrogradely transported phosphoprotein data and ~41% of the gene regulation data. Thus, cross-referencing the two data sets enabled us to focus on the proteins, genes, and pathways most likely to be relevant to the neuronal injury response.

Fig. 4
TF involvement in signaling to transcription networks. TFs associated with key node–based networks at P < 0.05 at different time points after injury are depicted together with the net effect on gene expression on the left (red, up; green, ...

Candidate hub proteins in signaling networks activated by injury

The high interconnectivity (Fig. 3) and redundancy in TF targeting (Fig. 4) by the signaling networks suggest that the system should be resistant to perturbations of single components. However, a closer examination of phosphoprotein representation in the different networks showed multiple appearances of various proteins in many of the networks (Fig. 5A). A number of these overrepresented proteins have already been implicated in nerve regeneration (3), although most have not been mechanistically linked to retrograde injury signaling (6, 7). Proteins such as c-Abl, PDK1 (phosphoinositide-dependent protein kinase 1), different PKC (protein kinase C) isoforms, p38 (a MAPK), and others appear in >80% of the networks implicated in up-regulation (Fig. 5B) or down-regulation (Fig. 5C) of gene expression in DRG after sciatic nerve injury. Removal of these proteins in silico caused substantial reductions in the size and connectivity of the reconstructed networks, raising the possibility that such multiply connected proteins, hereby designated “hub proteins,” might be critical for the system and that their perturbation might have far-reaching effects.

Fig. 5
Hub proteins involved in multiple networks. (A) Occurrence of identified phosphoproteins in signaling networks implicated in up-regulation (red) or down-regulation (blue) of gene expression. (B and C) Note the relative overrepresentation of ~50 proteins, ...

We therefore selected a few of the most highly represented hub proteins for targeting in a functional screen, focusing on candidates positioned to affect pathways upstream of multiple TFs (Fig. 6A). Many kinase inhibitors are nonspecific, and even those thought to be highly specific can have widespread secondary and tertiary effects (38). We therefore limited this screen to those kinases for which we could obtain well-characterized small-molecule inhibitors or activators. We applied these small molecules to freshly dissociated and triturated adult mouse DRG neurons for 3 hours, after which we removed the compounds by washing and monitored neuronal morphology and outgrowth parameters on a substrate permissive for outgrowth over a period of 72 hours (Fig. 6B). Such transient applications of perturbing agents were intended to induce a specific perturbation of the retrograde injury signal elicited by the triturating lesion without interfering with later signaling events or with neurite outgrowth mechanisms per se (9). Figure S4 shows a comparison of the neuronal outgrowth and branching response of neurons cultured with this protocol and neurons from animals that underwent a prior conditional crush lesion of the sciatic nerve in vivo. To ascertain the effectiveness of this transient drug application approach, we used the pan–serine-threonine kinase inhibitor staurosporine as a positive control. Indeed, staurosporine caused significant dose-dependent inhibition of neurite length (Fig. 6C) with a concomitant increase in branching frequency (fig. S4C) without affecting cell viability. Because long-term application of staurosporine is highly toxic, the specific effects observed with transient application are consistent with a temporally restricted perturbation of signaling events.

Fig. 6
Effects of perturbation of selected hub proteins. (A) Network analysis prediction of TF regulation for signaling pathways incorporating the four indicated hub proteins. (B) Flow chart of the experimental procedure for transient perturbation of injury ...

Robustness in retrograde injury signaling

Having demonstrated that it is possible to reduce neurite outgrowth by transient blockade of injury-induced phosphorylation, we examined whether transient perturbations might also enhance outgrowth. The network analyses indicated that several isoforms of PKC are likely to be involved in injury-regulated gene expression, influencing at least 28 TFs (Fig. 6A), leading us to test the effects of the PKCα and PKCβ inhibitor bis(4-fluoroanilino)-phthalimide. This inhibitor increased overall neurite outgrowth, which reached four times that of controls after 48 hours in culture (Fig. 6D). In contrast, the general PKC activator PMA (phorbol 12-myristate 13-acetate) caused a factor of ~2 decrease in total neurite outgrowth at 48 hours after plating. Thus, transient pharmacological perturbations concomitant with injury can induce both decreases and increases in the neurite outgrowth capacity of DRG sensory neurons. However, attempts to modulate specific PKC isoforms indirectly through inhibition of diacylglycerol kinase pathways caused relatively minor effects.

We then tested the effects of inhibitors of Abl (Abelson proto-oncogene), AKT, and p38, all hub proteins highly overrepresented in signaling networks (Fig. 5) upstream of multiple TFs (Fig. 6A). Abl, a tyrosine kinase found in nearly all the different signaling networks, also acts as a key node in its own network. The net effect of the different phosphoproteins upstream of Abl should lead to Abl activation after injury (table S3), and the network analyses suggest that Abl activation should affect 25 TFs: activating 12, inhibiting another 12, and disinhibiting one (p53) (Fig. 6A). Application of imatinib, which specifically inhibits Abl through an interaction with its tyrosine kinase domain (39), within the 3-hour window after trituration reduced neurite outgrowth significantly at the 24-hour time point, with a less pronounced effect at later time points (Fig. 6E). Branching frequency increased concomitantly with inhibition of neurite outgrowth (fig. S4C).

AKT, another kinase represented in nearly all the networks (Fig. 5), acts as a prominent downstream factor in the insulin receptor substrate 2 (IRS2) network. The phosphoprotein network analyses predicted that AKT should also be activated by injury and in turn should affect the activity of 33 TFs: inhibiting 15, activating 5, disinhibiting 10, and having mixed effects on 3 (Fig. 6A). LY-294002, which was used to specifically inhibit AKT, caused a marked decrease in total neurite outgrowth by 72 hours after plating (Fig. 6E) without a parallel effect on branching frequency (fig. S4C).

Analysis of p38 key node–related networks showed that activation of p38 should activate 10 TFs, inhibit 14, disinhibit 5, and have mixed effects on different isoforms of CREB (cyclic adenosine 5′-monophosphate response element–binding protein) (Fig. 6A). Inhibition of p38 caused a decrease in total neurite outgrowth (Fig. 6E), along with an increase in neurite branching frequency (fig. S4C), at 24 hours after plating. However, the effects of p38 inhibition were transient and did not persist at later times (Fig. 6E).

The effects of the individual compounds on neurite length lead us to examine the consequences of combining them, thereby simultaneously inhibiting two different hub kinases in the 3 hours immediately after injury. Combinations of Abl and AKT inhibitors at various concentrations failed to significantly affect total neurite outgrowth and, in fact, were much less efficacious than either inhibitor alone (Fig. 6E), although there was an effect of the combination on branching frequency (fig. S4C). We observed similar results on neurite outgrowth for Abl inhibition combined with PKC activation (Fig. 6E). Combinations of p38 inhibition with PKC activation or AKT inhibition did not show any synergism (Fig. 6E and fig. S4C). Only the combination of p38 and Abl inhibition revealed significant synergism in reducing neurite outgrowth at 48 hours after plating (Fig. 6E). Together, these data show that it is difficult to obtain efficient combinatorial perturbation of retrograde injury signaling, indicating that the system is highly robust.


Here, we examined the link between axonal injury signaling and the ensuing cell body response by using phosphoproteomics to characterize axonal signaling ensembles in rat sciatic nerve and microarrays to characterize the transcriptional response in DRG. The primary data sets were dauntingly large, comprising 879 proteins and 2465 phosphorylation sites on one hand and ~4500 transcripts on the other. This may be a consequence of the major changes caused by injury, whereby disruption of subcellular compartmentalization may cause mixing of kinases to generate a massive phosphorylation response (40, 41). The extent and complexity of these data could also reflect that they were obtained from multicellular tissues challenged in vivo. Computational analyses and juxtaposition of the data sets provided clarification and insight, indicating that ~400 highly redundant signaling networks connect to 39 TFs to control a substantial part of the response to nerve injury in sensory neurons. Experimental perturbation of individual overrepresented signaling hub proteins affected neurite outgrowth in sensory neurons, but, paradoxically, in some cases, simultaneous application of inhibitors for two targets had a lesser effect than did each inhibitor alone. These data indicate that nerve injury responses are controlled by multiple regulatory components, and suggest that network redundancy provides robustness to the injury response system.

Despite the large size of the data sets and the number of candidate hub proteins and transcriptional regulators implicated by our analysis, additional transcriptional regulators and hub proteins may yet be identified because of lacunae in data collection and the limitations of the currently available computational approaches. A single phosphoproteomic screen as done here cannot provide saturating and quantitative coverage of all the proteins undergoing phosphorylation at different time points after sciatic nerve injury (42), and current coverage of gene annotation and validated TFBS predictions are far short of being genome-wide (43). Nonetheless, the intersecting networks account for ~28% of the retrogradely transported phosphoprotein data and ~41% of the gene regulation data. Given that both data sets must include confounding data from nonneuronal components in the source tissues and from other physiological influences, the approach of juxtaposing retrograde signaling data from nerve with ganglia transcriptomes allows rapid and efficient focus on those components of the data most relevant for retrograde injury signaling.

Our findings show that the major cell body transcriptional response to axonal injury occurs after a distinct temporal delay. Although rapid electrical signals may play a priming role in the cell body response to injury (44, 45), relatively few genes show substantial changes in expression 3 hours after injury, which suggests that slower motor-driven signals are required to elicit the main transcriptional response. These may be de novo signals activated by the injury or by the loss of trophic or other signals present before injury (7). Complexes linked to molecular motors were indeed shown to be involved in retrograde injury signaling (912), and the temporal pattern observed in our data is consistent with the time frame needed for arrival of motor-dependent signals from the injury site. Indeed, proteomic profiling of retrogradely transported components in ligature-concentrated nerve axoplasm revealed numerous phosphoproteins specific to injured nerve. The number of these candidate signals is orders of magnitude larger than those previously reported (6, 7). This was unexpected because most axonal retrograde signaling mechanisms reported to date have been focused on the transport of a limited number of molecules (46, 47). These mechanisms include trapping of a ligand-receptor complex in an endosome to ensure continuous activation of the receptor by the trapped ligand en route (4850) or binding of an activated kinase to a chaperone linker that protects it from dephosphorylation en route (10, 20). In contrast, our data suggest that numerous components of a signaling network might travel together in the axon, perhaps linking to the retrograde transport machinery through a hub component that binds directly to molecular motors or importin subunits. Such mass movement of signaling molecules could maintain the network in an activated state through reciprocal activation cycles or wavelike propagation mechanisms (51). Indeed, it seems that retrogradely transported signaling ensembles might rival the complexity of synaptic phosphoproteomes, also thought to function by kinase cascades through a multiplicity of molecules (52, 53).

The regulatory junctions identified here comprise ~50 hub phosphoproteins and 39 TFs, underscoring the complexity of the injury response process. The nerve injury response seems to be channeled through multiple and parallel pathways, integrating diverse inputs and controlling a complex transcriptional output. Such a plethora of signals may assist the neuronal cell body in assessing the extent of damage and distance of the injury site, ensuring an appropriate response (54). This differs from the more familiar kinase signaling networks identified in cancer, which appear to depend on a limited number of critical hubs (55). Thus, the multifaceted regulation we observe for intrinsic injury responses in lesioned peripheral neurons seems to have more in common with complex immunological responses or multigenic disorders than to such systems as oncogene-driven proliferating cells or diseases arising from monogenetic lesions. The multiplicity of networks affecting the activity of each TF suggests that the mechanism underlying the injury response is highly robust, and indeed, we did not observe complete block of neurite outgrowth response after pharmacological perturbation of the individual hub proteins we tested, perhaps also reflecting redundancy of critical signaling proteins (56). Moreover, combinations of pharmacological agents did not necessarily produce additive or synergic effects. Specifically, in the case of combinations with c-Abl inhibitor, the effects were weaker than for each compound alone; this finding indicates the existence of unpredicted modes of cross talk and a high degree of interconnectivity in the signaling networks, leading to robustness in the retrograde response.

Given the recent interest in combinatorial approaches to the development of therapies for nerve injury (1, 22, 57), our findings have potentially broad implications. Although many of the candidate hub proteins we identified are considered druggable targets by the pharmaceutical industry (58), the hope that modulation of a handful of master regulators might suffice to elicit regeneration may be overly naïve. Similarly, early optimism upon identification of initial candidates for extrinsic influences on regeneration (59) has largely been replaced with a more nuanced view as the true number of parallel pathways became known (60, 61). Nonetheless, although our data suggest that it will not be trivial to identify useful cocktails that elicit robust intrinsic regenerative responses, we hope that the approach outlined in this study will enable future efforts to focus on a relatively tractable number of regulators. It is likely that such efforts will have to include dynamic in vivo measurements of signal flux through the system to characterize candidate modulators and drug prototypes (62). We hope that the present effort to outline a comprehensive characterization of the retrograde injury response will provide a useful mechanistic basis for future efforts to understand and modulate this process.


For complete details of experimental procedures, please see the Supplementary Materials.

Axoplasm and mass spectrometry

Eight- to 12-week-old Wistar rats were subjected to crush lesion, ligation, and axoplasm preparation as described (33, 63). Axoplasm samples (in quadruplicate for each treatment group—injured and uninjured proximal and distal, totaling 16 phosphoproteome analyses) were denatured, reduced and alkylated, and digested with trypsin before phosphopeptide enrichment on 5-mm TiO2 beads (GL Sciences), followed by strong cation exchange chromatography on a PolyLC column. Eluted fractions were desalted and analyzed as described (64). Initial automated analysis of spectra was with ProteinProspector 4.25, using the UniProt.2006.04.21 databases. Sequences and phosphorylation site assignments reported in table S1 were verified by manual inspection of the relevant spectra.

Microarrays and TFBS analyses

L4/L5 DRG were dissected from crush-lesioned or control sham-operated animals at the indicated time points. DRG total RNA was extracted with Qiagen RNeasy, 8 to 10 µg of RNA per sample were reverse-transcribed, and biotinylated complementary RNA was then hybridized to rat genome 230.2 arrays (Affymetrix). Each set was performed in triplicate over seven time points for a total of 39 arrays including controls. Array data were preprocessed with the robust multichip analysis algorithm (65), and data quality was assessed by principal components analysis. Log2-transformed data were zero-transformed to the first time point. For ANOVA, gene profiles were standardized such that all genes would have the same SD. TFBS enrichment was evaluated in ExPlain 2.3 (Biobase) with gene sets passing an FDR threshold of 0.1 and fold change of 2 or more for each time point and their corresponding background sets. Promoter sequences were scanned from 600 base pairs (bp) upstream to 100 bp downstream of the predicted transcriptional start site for each such gene, and TFBSs were identified on the basis of TRANSFAC version 9. TFBS enrichment in test versus background sets was assessed by t test with a P value threshold of 0.05.

Network analyses

Phosphoproteomic data were used to generate protein signaling networks with TRANSPATH (37) in ExPlain 2.3. Key node–based networks as defined by ExPlain were built with the following search parameters: distance threshold, 4; connectivity penalty, 8; persistence reward, 0.5, with inclusion of superfamilies and transregulation reactions. Network score threshold was set to 10, P value to 0.001, and FDR to 0.05. Networks derived from microarray and TFBS analyses were cross-referenced with those derived from the phosphoproteins to generate combined networks for each time point.

High-content imaging screen

Eight compounds were tested for effects on neurite outgrowth on yellow fluorescent protein–expressing mouse DRG neurons (66) cultured in 96-well plates. Culture media and outgrowth-permissive substrates were as described (9). A dose-response profile was established for a 3-hour application of each agent before selection of the final test concentration. Neuronal cultures were imaged on an ImageXpress Micro System (Molecular Devices), and morphometric parameters were quantified with MetaXpress software (Molecular Devices) after subtracting results with vehicle. Treatment/control ratios were log-transformed and data were evaluated by one-way ANOVA (P < 0.01).

Supplementary Material

Supplemental figs

Supplemental figure 3

Click here to view.(284 bytes, webloc)

Supplemental table 1

Click here to view.(313 bytes, webloc)

Supplemental table 2

Click here to view.(313 bytes, webloc)

Supplemental table 3

Click here to view.(313 bytes, webloc)

Supplemental table 4

Click here to view.(313 bytes, webloc)


We thank J. L. Twiss, M. H. Tuszynski, and A. Blesch for helpful discussions. Funding: Supported by grants to M.F. from the Israel Science Foundation, the Christopher and Dana Reeve Foundation, and the International Institute for Research in Paraplegia; grants from the Dr. Miriam and Sheldon Adelson Medical Research Foundation to M.F., D.H.G., and A.L.B.; NIH grants to A.L.B. (NCRR P41RR001614 and NCRR RR012961); a European Molecular Biology Organization short-term fellowship to I.M.; and the Chaya Professorial Chair in Molecular Neuroscience to M.F.



Materials and Methods

Fig. S1. Enrichment of injury-related phosphopeptides.

Fig. S2. Initial analyses of TFBSs.

Fig. S3. Network structure of the 45 top-ranked networks. Color-coding as detailed in the legend to Fig. 3.

Fig. S4. Drug effects on morphometric parameters of the injury response in sensory neurons.

Table S1. Phosphoproteins identified in this study.

Table S2. Differentially expressed genes from the microarray data.

Table S3. Phosphoprotein representation in networks linking axonal signaling to the cell body transcriptional response.

Table S4. Linkage of signaling networks to gene regulation profiles at designated time points.


Author contributions: I.M., Y.S.-R., M.R., K.B.-Y., and I.R. performed the experiments. I.M., K.F.M., and A.L.B. were responsible for phosphoproteomic analyses. Y.S.-R., S.H.-S., G.C., and D.H.G. were responsible for microarrays. I.M., S.Y.D., O.S., and Y.P. were responsible for computational analyses. I.M. and M.F. designed the experiments, analyzed the data, and wrote the paper.

Competing interests: The authors declare no competing interests.


1. Benowitz LI, Yin Y. Combinatorial treatments for promoting axon regeneration in the CNS: Strategies for overcoming inhibitory signals and activating neurons’ intrinsic growth state. Dev. Neurobiol. 2007;67:1148–1165. [PubMed]
2. Zhou FQ, Snider WD. Intracellular control of developmental and regenerative axon growth. Philos. Trans. R. Soc. London Ser. B. 2006;361:1575–1592. [PMC free article] [PubMed]
3. Rossi F, Gianola S, Corvetti L. Regulation of intrinsic neuronal properties for axon growth and regeneration. Prog. Neurobiol. 2007;81:1–28. [PubMed]
4. Smith DS, Skene JH. A transcription-dependent switch controls competence of adult neurons for distinct modes of axon growth. J. Neurosci. 1997;17:646–658. [PubMed]
5. Neumann S, Woolf CJ. Regeneration of dorsal column fibers into and beyond the lesion site following adult spinal cord injury. Neuron. 1999;23:83–91. [PubMed]
6. Abe N, Cavalli V. Nerve injury signaling. Curr. Opin. Neurobiol. 2008;18:276–283. [PMC free article] [PubMed]
7. Rishal I, Fainzilber M. Retrograde signaling in axonal regeneration. Exp. Neurol. 2010;223:5–10. [PubMed]
8. Yudin D, Hanz S, Yoo S, Iavnilovitch E, Willis D, Gradus T, Vuppalanchi D, Segal-Ruder Y, Ben-Yaakov K, Hieda M, Yoneda Y, Twiss JL, Fainzilber M. Localized regulation of axonal RanGTPase controls retrograde injury signaling in peripheral nerve. Neuron. 2008;59:241–252. [PMC free article] [PubMed]
9. Hanz S, Perlson E, Willis D, Zheng JQ, Massarwa R, Huerta JJ, Koltzenburg M, Kohler M, van-Minnen J, Twiss JL, Fainzilber M. Axoplasmic importins enable retrograde injury signaling in lesioned nerve. Neuron. 2003;40:1095–1104. [PubMed]
10. Perlson E, Hanz S, Ben-Yaakov K, Segal-Ruder Y, Seger R, Fainzilber M. Vimentin-dependent spatial translocation of an activated MAP kinase in injured nerve. Neuron. 2005;45:715–726. [PubMed]
11. Cavalli V, Kujala P, Klumperman J, Goldstein LS. Sunday Driver links axonal transport to damage signaling. J. Cell Biol. 2005;168:775–787. [PMC free article] [PubMed]
12. Abe N, Almenar-Queralt A, Lillo C, Shen Z, Lozach J, Briggs SP, Williams DS, Goldstein LS, Cavalli V. Sunday Driver interacts with two distinct classes of axonal organelles. J. Biol. Chem. 2009;284:34628–34639. [PMC free article] [PubMed]
13. Hanz S, Fainzilber M. Retrograde signaling in injured nerve—the axon reaction revisited. J. Neurochem. 2006;99:13–19. [PubMed]
14. Haeusgen W, Boehm R, Zhao Y, Herdegen T, Waetzig V. Specific activities of individual c-Jun N-terminal kinases in the brain. Neuroscience. 2009;161:951–959. [PubMed]
15. Hollis ER, II, Jamshidi P, Low K, Blesch A, Tuszynski MH. Induction of corticospinal regeneration by lentiviral trkB-induced Erk activation. Proc. Natl. Acad. Sci. U.S.A. 2009;106:7215–7220. [PubMed]
16. Yan D, Wu Z, Chisholm AD, Jin Y. The DLK-1 kinase promotes mRNA stability and local translation in C. elegans synapses and axon regeneration. Cell. 2009;138:1005–1018. [PMC free article] [PubMed]
17. Hammarlund M, Nix P, Hauth L, Jorgensen EM, Bastiani M. Axon regeneration requires a conserved MAP kinase pathway. Science. 2009;323:802–806. [PMC free article] [PubMed]
18. Kenney AM, Kocsis JD. Peripheral axotomy induces long-term c-Jun amino-terminal kinase-1 activation and activator protein-1 binding activity by c-Jun and junD in adult rat dorsal root ganglia in vivo. J. Neurosci. 1998;18:1318–1328. [PMC free article] [PubMed]
19. Lindwall C, Kanje M. Retrograde axonal transport of JNK signaling molecules influence injury induced nuclear changes in p-c-Jun and ATF3 in adult rat sensory neurons. Mol. Cell. Neurosci. 2005;29:269–282. [PubMed]
20. Perlson E, Michaelevski I, Kowalsman N, Ben-Yaakov K, Shaked M, Seger R, Eisenstein M, Fainzilber M. Vimentin binding to phosphorylated Erk sterically hinders enzymatic dephosphorylation of the kinase. J. Mol. Biol. 2006;364:938–944. [PubMed]
21. Ylera B, Ertürk A, Hellal F, Nadrigny F, Hurtado A, Tahirovic S, Oudega M, Kirchhoff F, Bradke F. Chronically CNS-injured adult sensory neurons gain regenerative competence upon a lesion of their peripheral axon. Curr. Biol. 2009;19:930–936. [PubMed]
22. Kadoya K, Tsukada S, Lu P, Coppola G, Geschwind D, Filbin MT, Blesch A, Tuszynski MH. Combined intrinsic and extrinsic neuronal mechanisms facilitate bridging axonal regeneration one year after spinal cord injury. Neuron. 2009;64:165–172. [PMC free article] [PubMed]
23. Costigan M, Befort K, Karchewski L, Griffin RS, D’Urso D, Allchorne A, Sitarski J, Mannion JW, Pratt RE, Woolf CJ. Replicate high-density rat genome oligonucleotide microarrays reveal hundreds of regulated genes in the dorsal root ganglion after peripheral nerve injury. BMC Neurosci. 2002;3:16. [PMC free article] [PubMed]
24. Seijffers R, Mills CD, Woolf CJ. ATF3 increases the intrinsic growth state of DRG neurons to enhance peripheral nerve regeneration. J. Neurosci. 2007;27:7911–7920. [PubMed]
25. Hannila SS, Filbin MT. The role of cyclic AMP signaling in promoting axonal regeneration after spinal cord injury. Exp. Neurol. 2008;209:321–332. [PMC free article] [PubMed]
26. Qiu J, Cafferty WB, McMahon SB, Thompson SW. Conditioning injury-induced spinal axon regeneration requires signal transducer and activator of transcription 3 activation. J. Neurosci. 2005;25:1645–1653. [PubMed]
27. Raivich G, Behrens A. Role of the AP-1 transcription factor c-Jun in developing, adult and injured brain. Prog. Neurobiol. 2006;78:347–363. [PubMed]
28. Stam FJ, MacGillavry HD, Armstrong NJ, de Gunst MC, Zhang Y, van Kesteren RE, Smit AB, Verhaagen J. Identification of candidate transcriptional modulators involved in successful regeneration after nerve injury. Eur. J. Neurosci. 2007;25:3629–3637. [PubMed]
29. Park KK, Liu K, Hu Y, Smith PD, Wang C, Cai B, Xu B, Connolly L, Kramvis I, Sahin M, He Z. Promoting axon regeneration in the adult CNS by modulation of the PTEN/mTOR pathway. Science. 2008;322:963–966. [PMC free article] [PubMed]
30. Zou H, Ho C, Wong K, Tessier-Lavigne M. Axotomy-induced Smad1 activation promotes axonal growth in adult sensory neurons. J. Neurosci. 2009;29:7116–7123. [PMC free article] [PubMed]
31. Moore DL, Blackmore MG, Hu Y, Kaestner KH, Bixby JL, Lemmon VP, Goldberg JL. KLF family members regulate intrinsic axon regeneration ability. Science. 2009;326:298–301. [PMC free article] [PubMed]
32. Marmigère F, Ernfors P. Specification and connectivity of neuronal subtypes in the sensory lineage. Nat. Rev. Neurosci. 2007;8:114–127. [PubMed]
33. Michaelevski I, Medzihradszky KF, Lynn A, Burlingame AL, Fainzilber M. Axonal transport proteomics reveals mobilization of translation machinery to the lesion site in injured sciatic nerve. Mol. Cell. Proteomics. 2010;9:976–987. [PMC free article] [PubMed]
34. Garten Y, Kaplan S, Pilpel Y. Extraction of transcription regulatory signals from genome-wide DNA–protein interaction data. Nucleic Acids Res. 2005;33:605–615. [PMC free article] [PubMed]
35. Kel A, Voss N, Jauregui R, Kel-Margoulis O, Wingender E. Beyond microarrays: Finding key transcription factors controlling signal transduction pathways. BMC Bioinformatics. 2006;7(Suppl. 2):S13. [PMC free article] [PubMed]
36. Kel A, Voss N, Valeev T, Stegmaier P, Kel-Margoulis O, Wingender E. ExPlain: Finding upstream drug targets in disease gene regulatory networks. SAR QSAR Environ. Res. 2008;19:481–494. [PubMed]
37. Krull M, Pistor S, Voss N, Kel A, Reuter I, Kronenberg D, Michael H, Schwarzer K, Potapov A, Choi C, Kel-Margoulis O, Wingender E. TRANSPATH: An information resource for storing and visualizing signaling pathways and their pathological aberrations. Nucleic Acids Res. 2006;34:D546–D551. [PMC free article] [PubMed]
38. Pan C, Olsen JV, Daub H, Mann M. Global effects of kinase inhibitors on signaling networks revealed by quantitative phosphoproteomics. Mol. Cell. Proteomics. 2009;8:2796–2808. [PMC free article] [PubMed]
39. Houghton PJ, Germain GS, Harwood FC, Schuetz JD, Stewart CF, Buchdunger E, Traxler P. Imatinib mesylate is a potent inhibitor of the ABCG2 (BCRP) transporter and reverses resistance to topotecan and SN-38 in vitro. Cancer Res. 2004;64:2333–2337. [PubMed]
40. Lasiecka ZM, Yap CC, Vakulenko M, Winckler B. Compartmentalizing the neuronal plasma membrane from axon initial segments to synapses. Int. Rev. Cell Mol. Biol. 2009;272:303–389. [PubMed]
41. Chu CT, Levinthal DJ, Kulich SM, Chalovich EM, DeFranco DB. Oxidative neuronal injury. The dark side of ERK1/2. Eur. J. Biochem. 2004;271:2060–2066. [PMC free article] [PubMed]
42. Piggee C. Phosphoproteomics: Miles to go before it’s routine. Anal. Chem. 2009;81:2418–2420. [PubMed]
43. Badis G, Berger MF, Philippakis AA, Talukder S, Gehrke AR, Jaeger SA, Chan ET, Metzler G, Vedenko A, Chen X, Kuznetsov H, Wang CF, Coburn D, Newburger DE, Morris Q, Hughes TR, Bulyk ML. Diversity and complexity in DNA recognition by transcription factors. Science. 2009;324:1720–1723. [PMC free article] [PubMed]
44. Perlson E, Hanz S, Medzihradszky KF, Burlingame AL, Fainzilber M. From snails to sciatic nerve: Retrograde injury signaling from axon to soma in lesioned neurons. J. Neurobiol. 2004;58:287–294. [PubMed]
45. Mandolesi G, Madeddu F, Bozzi Y, Maffei L, Ratto GM. Acute physiological response of mammalian central neurons to axotomy: Ionic regulation and electrical activity. FASEB J. 2004;18:1934–1936. [PubMed]
46. Ibáñez CF. Message in a bottle: Long-range retrograde signaling in the nervous system. Trends Cell Biol. 2007;17:519–528. [PubMed]
47. Cosker KE, Courchesne SL, Segal RA. Action in the axon: Generation and transport of signaling endosomes. Curr. Opin. Neurobiol. 2008;18:270–275. [PMC free article] [PubMed]
48. Howe CL, Valletta JS, Rusnak AS, Mobley WC. NGF signaling from clathrin-coated vesicles: Evidence that signaling endosomes serve as a platform for the Ras-MAPK pathway. Neuron. 2001;32:801–814. [PubMed]
49. Watson FL, Heerssen HM, Bhattacharyya A, Klesse L, Lin MZ, Segal RA. Neurotrophins use the Erk5 pathway to mediate a retrograde survival response. Nat. Neurosci. 2001;4:981–988. [PubMed]
50. Riccio A, Pierchala BA, Ciarallo CL, Ginty DD. An NGF-TrkA–mediated retrograde signal to transcription factor CREB in sympathetic neurons. Science. 1997;277:1097–1100. [PubMed]
51. Kholodenko BN. Cell-signalling dynamics in time and space. Nat. Rev. Mol. Cell Biol. 2006;7:165–176. [PMC free article] [PubMed]
52. Emes RD, Pocklington AJ, Anderson CN, Bayes A, Collins MO, Vickers CA, Croning MD, Malik BR, Choudhary JS, Armstrong JD, Grant SG. Evolutionary expansion and anatomical specialization of synapse proteome complexity. Nat. Neurosci. 2008;11:799–806. [PMC free article] [PubMed]
53. Trinidad JC, Thalhammer A, Specht CG, Lynn AJ, Baker PR, Schoepfer R, Burlingame AL. Quantitative analysis of synaptic phosphorylation and protein expression. Mol. Cell. Proteomics. 2008;7:684–696. [PubMed]
54. Kam N, Pilpel Y, Fainzilber M. Can molecular motors drive distance measurements in injured neurons? PLoS Comput. Biol. 2009;5:e1000477. [PMC free article] [PubMed]
55. Amit I, Wides R, Yarden Y. Evolvable signaling networks of receptor tyrosine kinases: Relevance of robustness to malignancy and to cancer therapy. Mol. Syst. Biol. 2007;3:151. [PMC free article] [PubMed]
56. Kafri R, Springer M, Pilpel Y. Genetic redundancy: New tricks for old genes. Cell. 2009;136:389–392. [PubMed]
57. Lu P, Tuszynski MH. Growth factors and combinatorial therapies for CNS regeneration. Exp. Neurol. 2008;209:313–320. [PMC free article] [PubMed]
58. Sioud M, Leirdal M. Druggable signaling proteins. Methods Mol. Biol. 2007;361:1–24. [PubMed]
59. Schwab ME. Nogo and axon regeneration. Curr. Opin. Neurobiol. 2004;14:118–124. [PubMed]
60. Woolf CJ. No Nogo: Now where to go? Neuron. 2003;38:153–156. [PubMed]
61. Giger RJ, Venkatesh K, Chivatakarn O, Raiker SJ, Robak L, Hofer T, Lee H, Rader C. Mechanisms of CNS myelin inhibition: Evidence for distinct and neuronal cell type specific receptor systems. Restor. Neurol. Neurosci. 2008;26:97–115. [PubMed]
62. Hellerstein MK. Exploiting complexity and the robustness of network architecture for drug discovery. J. Pharmacol. Exp. Ther. 2008;325:1–9. [PubMed]
63. Rishal I, Michaelevski I, Rozenbaum M, Shinder V, Medzihradszky KF, Burlingame AL, Fainzilber M. Axoplasm isolation from peripheral nerve. Dev. Neurobiol. 2010;70:126–133. [PMC free article] [PubMed]
64. Trinidad JC, Specht CG, Thalhammer A, Schoepfer R, Burlingame AL. Comprehensive identification of phosphorylation sites in postsynaptic density preparations. Mol. Cell. Proteomics. 2006;5:914–922. [PubMed]
65. Bolstad BM, Collin F, Simpson KM, Irizarry RA, Speed TP. Experimental design and low-level analysis of microarray data. Int. Rev. Neurobiol. 2004;60:25–58. [PubMed]
66. Feng G, Mellor RH, Bernstein M, Keller-Peck C, Nguyen QT, Wallace M, Nerbonne JM, Lichtman JW, Sanes JR. Imaging neuronal subsets in transgenic mice expressing multiple spectral variants of GFP. Neuron. 2000;28:41–51. [PubMed]