Retrograde TrkAIII transport from ERGIC to ER: a re-localisation mechanism for oncogenic activity.

In human SH-SY5Y neuroblastoma (NB) cells, nascent immature N-glycosylated 110kDa TrkA moves rapidly from the endoplasmic reticulum (ER) to the Golgi Network (GN), where it matures into the 140kDa receptor prior to being transported to the cell surface, creating GN and cell surface pools of inactive receptor maintained below the spontaneous activation threshold by a full compliment of inhibitory domains and endogenous PTPases. In contrast, the oncogenic alternative TrkAIII splice variant is not expressed at the cell surface but re-localises to intracellular membranes, within which it exhibits spontaneous ERGIC/COPI-associated activation and oncogenic Akt signalling. In this study, we characterise the mechanism responsible for TrkAIII re-localisation. Spontaneous TrkAIII activation, facilitated by D4 IG-like domain and N-glycosylation site omission, increases spontaneous activation potential by altering intracellular trafficking, inhibiting cell surface expression and eliminating an important inhibitory domain. TrkAIII, spontaneously activated within the permissive ERGIC/COPI compartment, rather than moving in an anterograde direction to the GN exhibits retrograde transport back to the ER, where it is inactivated. This sets-up self-perpetuating TrkAIII re-cycling between the ERGIC and ER, that ensures continual accumulation above the spontaneous activation threshold of the ERGIC/COPI compartment. This is reversed by TrkA tyrosine kinase inhibitors, which promote anterograde transport of inactivated TrkAIII to the GN, resulting in GN-associated TrkAIII maturation to a 120kDa species that is degraded at the proteasome.


IntroductIon
The developmental and stress-regulated alternative TrkAIII splice variant of the NGF receptor tropomyosinrelated kinase, TrkA, is expressed by advanced stage human neuroblastomas (NBs), associates with poor prognosis and post therapeutic relapse in unfavourable high TrkA expressing tumours and exhibits oncogenic activity in NB models [1][2][3][4][5][6]. TrkAIII oncogenic activity is caused by exon 6/7 skipping, which results in the omission of the extracellular D4 IG-like domain and several N-glycosylation sites, important for optimising intracellular trafficking, correct cell surface localisation, ligand binding and preventing spontaneous ligandindependent receptor activation [1,[7][8][9]. As a consequence and in contrast to TrkA, TrkAIII is not expressed at the cell surface but re-localises to intracellular membranes, within which it exhibits spontaneous ligand-independent activation, leading to oncogenic signalling through Akt but not Ras from this alternative location. Signalling from TrkAIII differs to classical signalling from cell surface ligand-activated TrkA through Ras/MAPK, suggesting that the re-localisation of TrkAIII is fundamental for it's oncogenic activity [1]. This hypothesis is supported by recent reports that the oncogenic activity of several receptor tyrosine kinase (RTK) oncogenes depends www.impactjournals.com/oncotarget upon their re-localisation [10]. Mutation-activated FLT3 and c-KIT oncogenes, important in haematological malignancies, are retained within the ER, exhibit ERassociated spontaneous activation and oncogenic signalling through Akt/STAT5/PIM [11][12]. The Trk-T3 oncogene, involved in thyroid cancer, re-localises to ER exit sites (ERES), exhibits spontaneous activation within ERES-associated COPII vesicles and induces oncogenic signalling from this alternative location through Akt but not Ras, similar to that reported for the TrkAIII oncoprotein [1,13].
The re-localisation of FLT3-ITD and c-Kit-ITD to the ER results from duplication of the internal tandem juxta-membrane domains, which impairs ER exit and facilitates spontaneous activation by compromising an important receptor inhibitory domain [11,12]. The relocalisation of Trk-T3 to the ERES/COPII compartment results from TrkA extracellular domain substitution with a chimeric Trk-fused gene (TFG) sequences that regulate COPII vesicle formation at ERES, with the elimination of important TrkA extracellular domain sequences further facilitating spontaneous activation [13]. Whatever mechanisms result in RTK oncogene re-localisation, all induce oncogenic signalling from an alternative substrate context [1,[10][11][12][13].
Interestingly, the mere overexpression of TrkA in some cell types results in spontaneous intracellular activation and altered signalling. In human SV40immortalised 293 kidney cells, TrkA overexpression results in spontaneous activation within the ERGIC, which causes GN disruption and inhibits further GNassociated TrkA maturation. This augments intracellular accumulation, facilitates spontaneous activation and results in altered signalling through Akt but not Ras from this alternative site, also similar to that reported for the TrkAIII oncoprotein [1,14]. This cell-specific effect most likely reflects slower rates of intracellular TrkA trafficking, maturation and cell surface translocation, leading to aberrant accumulation and spontaneous activation within the ERGIC, perpetuated by GN disruption [14].
In pursuit of novel therapeutic strategies to inhibit RTK oncogenic activity, therefore, it is not only important to characterise the mechanisms responsible for oncogenic RTK activation and the oncogenic signalling pathways activated but also to identify the specific sites to which RTK oncogenes re-localise, and the mechanisms that cause re-localization and facilitate spontaneous activation. In the present study, we report a novel mechanism to explain the re-localisation of the TrkAIII oncoprotein in human NB cells. This mechanism is characterised by activationdependent retrograde TrkAIII recycling from the ERGIC back to the ER, which reduces anterograde TrkAIII transport to the GN, impedes further GN-associated TrkAIII maturation and sets-up self-perpetuating TrkAIII accumulation above the threshold for spontaneous activation within the ERGIC/COPI compartment, resulting in oncogenic Akt signalling from this alternative location. This mechanism is prevented by TrkA inhibitors, which promote anterograde trafficking of inactivated TrkAIII from the ERGIC to the GN, resulting in GN-associated maturation of TrkAIII to a novel 120kDa species that fails to reach the cell surface and is degraded at the proteasome. results trkA localises to two distinct pools of inactive receptors in sH-sY5Y cells We have previously reported [3] and confirm here ( Figure 1A) that TrkA trafficking from ER to GN and then to the cell surface in SH-SY5Y NB cells occurs without spontaneous activation and results in two distinct steadystate pools of inactive TrkA receptors at the cell surface and within the GN, the latter overlapping closely with the GN marker GM130 [3]. Treatment of TrkA SH-SY5Y cells with the PTPase inhibitor sodium orthovanadate (Vanadate, 0.1mM for 3 hours) induced tyrosine phosphorylation of cell surface but not GN-associated 140kDa TrkA, implicating endogenous PTPases in preventing spontaneous activation of cell surface 140kDa TrkA but not GN-associated TrkA ( Figure 1A

Activated trkAIII recycles back to the er
Metabolically labelling was used to further interrogate TrkA and TrkAIII trafficking over 60 minutes in density gradient ultracentrifugation separated membranes.

spontaneous trkAIII activation occurs within the erGIc/coPI vesicle compartment
In TrkAIII SH-SY5Y cells, confocal IF confirmed a close relationship between Y490 phosphorylated TrkAIII and the ERGIC marker ERGIC-53, when compared to the cis-GN marker GM130, the trans-GN marker TGN46 and the ER marker calnexin ( Figure 5A), confirming our previous report using non-confocal IF [3].
Brefeldin A (BFA; 5µg/ml for 6 hours) abrogated TrkAIII tyrosine phosphorylation, assessed by both indirect IF ( Figure 5B) and Western blotting of whole cell extracts ( Figure 5C). BFA caused re-distribution of GM130 in TrkAIII SH-SY5Y cells, consistent with disruption of a constitutively intact GN ( Figure 5D).

trkA inhibitors promote 120kda trkAIII maturation
Different TrkA inhibitors were compared in their capacity to promote 120kDa TrkAIII maturation. The TrkA inhibitors CEP-701, K252a, Go6976 and GW441756 [17][18][19][20][21] all abrogated TrkAIII tyrosine phosphorylation and promoted 120kDa TrkAIII maturation ( Figure 6A). In contrast, the hyper-activation of TrkAIII induced by sodium orthovanadate did not result in 120kDa TrkAIII maturation, supporting an inverse relationship between TrkAIII tyrosine phosphorylation and 120kDa maturation ( Figure 6B). In IF studies, all TrkA inhibitors decreased TrkAIII staining intensity and promoted more diffuse staining without evidence of cell surface expression, whereas sodium orthovanadate did not alter TrkAIII staining intensity or distribution ( Figure 6C). Combination of TrkA inhibitors with the Hsp90 inhibitor Geldanamycin (GA, 1µM) further reduced TrkAIII staining intensity but in the presence of the proteasome inhibitor MG-132 (1µM) TrkAIII staining intensity was restored in association with re-distribution to intercellular junctions and cellular projections, consistent with potential cell surface expression ( Figure 6C). These data suggest that TrkA inhibitors and GA promote TrkAIII degradation at the proteasome.

nocodozol and GA inhibit 120kda trkAIII Maturation
MG-132 (1µM for 3 and 6 hours) induced low-level 120kDa TrkAIII expression, suggesting that in TrkAIII SH-SY5Y cells a proportion of TrkAIII must arrive at the GN but upon 120kDa maturation is degraded at the proteasome and, therefore, not readily detected ( Figure  7A).
GA (1µM for 12 hours) inhibited TrkAIII tyrosine phosphorylation, did not reduce 100kDa TrkAIII levels nor induce 120kDa TrkAIII expression ( Figure 7C) but did inhibit 120kDa TrkAIII maturation induced by CEP-701 ( Figure 7D). MG-132 not only reversed this GA-inhibitory effect but also augmented 120kDa TrkAIII expression above that induced by CEP-701 alone ( Figure 7D), indicating that GA enhances 120kDa TrkAIII degradation at the proteasome. We also confirm our previous report [2], that MG-132 reverses GA abrogation of 140kDa TrkA expression in TrkA SH-SY5Y cells ( Figure 7D). GA did not reduce 100kDa TrkAIII or 110kDa TrkA levels, indicating that these immature receptor forms are relatively insensitive to degradation at the proteasome and are not stabilized by GA-sensitive interaction with Hsp90.
The thiol reducing agent dithiothreitol (DTT) inhibits protein folding, promotes protein retention within the ER and prevents protein trafficking to the GN [24]. DTT abrogated both CEP-701-induced 120kDa TrkAIII maturation and constitutive 140kDa TrkA maturation, without reducing immature receptor levels ( Figure 7E, left panel). This effect was not inhibited by MG-132, confirming it to be proteasome-independent, consistent with the necessity for transport to the GN for both 120kDa TrkAIII and 140kDa TrkA maturation ( Figure 7E)

dIscussIon
In this study, we characterise the re-localisation mechanism responsible for spontaneous activation micrographs demonstrating different degrees of association (yellow) between Y490-phosphorylated TrkAIII (pTrkAIII) (red) and GM130 (green) (top panels); pTrkAIII (red) and TGN46 (green) (2 nd panels); pTrkAIII (red) and ERGIC-53 (green) (3 rd panels); and pTrkAIII (red) and Calnexin (green) (bottom panels) in individual representative TrkAIII SH-SY5Y cells (magnification x100; bar = 10µm). b. Indirect non-confocal IF, demonstrating the degree of overlap (yellow) between TrkAIII (green) and phosphorylated TrkAIII (red) in untreated TrkAIII SH-SY5Y cells (Con) (top 3 panels) and BFA-treated (0.1µM for 3 hr) TrkAIII SH-SY5Y cells (bottom 3 panels; bar = 10µm). c. Western blots of whole cell extracts (30µg), demonstrating the effect of BFA (0.1µM for 0, 1, 3 and 6 hours) on total and tyrosine phosphorylated TrkAIII. d. Indirect non-confocal IF, demonstrating overlap (yellow) and general distribution of TrkAIII (green) and GM130 (red) in untreated (Con) and BFA-treated (0.1µM for 3 hr) TrkAIII SH-SY5Y cells. Western blots demonstrating: eI. TrkAIII co-immunoprecipitation of COP-I in density gradient ultracentrifugation-purified TrkAIII SH-SY5Y membranes; eII. The presence of tyrosine phosphorylated TrkAIII in TrkAIII immunoprecipitates from the same TrkAIII SH-SY5Y membrane preparation and eIII. The absence of TrkA/COPI co-immunoprecipitation and TrkA tyrosine phosphorylated in density gradient ultracentrifugation-purified intracellular membranes from TrkA SH-SY5Y cells. www.impactjournals.com/oncotarget  and subsequent oncogenic activity of the TrkAIII oncoprotein in SH-SY5Y NB cells. Exons 6-7 skipping that characterises TrkAIII, eliminates the D4 IG-like domain and several N-glycosylation sites involved in optimising intracellular trafficking and reducing spontaneous activation potential, increasing intracellular accumulation and facilitating spontaneous activation within the ERGIC/COPI vesicle compartment. This, in turn, promotes retrograde TrkAIII transport back to the ER and reduces anterograde transport to the GN, setting up a self-perpetuating mechanism for intracellular TrkAIII entrapment, accumulation, spontaneous activation and oncogenic Akt signalling from this alternative localisation. TrkA tyrosine kinase inhibitors abrogate this mechanism and promote TrkAIII transport to the GN, resulting in GNassociated TrkAIII maturation to a 120kDa species that is more readily degraded at the proteasome.
The marked difference in intracellular trafficking, cell surface expression and spontaneous activation exhibited by TrkA and TrkAIII in SH-SY5Y cells confirms the critical role played exons 6 and 7 encoded sequence in optimising intracellular trafficking from the ER to the GN, reducing intracellular spontaneous activation potential and promoting receptor maturation, an essential step in cell surface expression (this study and [2]). In the case of TrkA, this results in two predominant inactive steady state pools, one within the GN and the other at the cell surface (this study and [1][2][3]) both maintained inactive by a full complement of spontaneous activation inhibitory D4 and D5 IG-like domains, a full N-glycosylated status that reduces aberrant accumulation within the ERGIC, N-glycan maturation that further reduces aberrant intracellular accumulation by promoting translocation to the cell surface [1-4, 7, 8-10] and, in the case of cell surface TrkA, endogenous PTPase activity [2]. In contrast to TrkA, immature N-glycosylated 100kDa TrkAIII exits the ER but fails to adequately reach the GN. As a consequence, TrkAIII does not undergo significant GNassociated N-glycan maturation, fails to reach the cell surface and becomes entrapped between the ER and ERGIC intracellular membrane compartments, within which it accumulates and undergoes spontaneous ERGIC/ COPI-associated activation.
Metabolic labelling comparisons indicated that whilst TrkA and TrkAIII both exit the ER, TrkAIII recycles back to the ER from the ERGIC, resulting in entrapment and accumulation, whereas TrkA moves from the ERGIC to the GN. TrkAIII recycling from the ERGIC to the ER was prevented by TrkA inhibitors, which promoted TrkAIII trafficking to the GN, resulting in GNassociated TrkAIII N-glycan maturation. This indicates that retrograde recycling of TrkAIII back to the ER and reduced transport to the GN is TrkAIII activity-dependent. Subsequent TrkAIII entrapment and accumulation within the pre-GN compartment, combined with omission of the D4 IGC1 spontaneous activation prevention domain, further augment the potential for spontaneous intracellular activation. This novel activation-regulated re-localisation mechanism differs to that utilised by mutated FLT and Kit RTK oncogenes, which fail to satisfy ER quality control and are retained and activated within the ER [10][11][12], and the Trk-T3 oncogene that is retained and activated within the ERES/COPII compartment as a result of chimeric TFG sequences involved in COPII vesicle formation [13].
TrkAIII activation within an ERGIC/COPI vesicle context was confirmed using BFA, an inhibitor of ARF-1dependent COPI vesicle formation [21], which abrogated ERGIC-associated TrkAIII activation, was supported by the close relationship exhibited between COP-I positive membranes and tyrosine phosphorylated TrkAIII and by co-immunoprecipitation of tyrosine phosphorylated TrkAIII and COP-I from the same membrane fraction. The fact that BFA inhibited TrkAIII activity but did not promote TrkAIII 120kDa maturation is explained by its capacity to disrupt the GN (this study and [21]), which would prevent the transport of inactivated TrkAIII to the GN. BFA does not disrupt the ERGIC [25], further supporting the hypothesis of ERGIC-associated TrkAIII activation within a COP-I vesicle context but does prevent COP-II vesicle formation [21]. We do not exclude, therefore, a potential role for COP-II vesicles in regulating the spontaneous activation of TrkAIII. BFA inhibition of spontaneous TrkAIII activation in SH-SY5Y cells differs to a report that ERGIC-associated spontaneous TrkA activation in human kidney 293 cells is insensitivity to BFA and, therefore, COPI-independent [14].
The mechanism through which activated TrkAIII recycles back to the ER remains to be elucidated. TrkAIII does not contain KDEL-like sequences, suggesting a KDEL-R-independent mechanism but could complex with KDEL-cargos in active-form to facilitate retrograde transport in KDEL-R/COPI vesicles [26,27]. KDEL-R is a G-protein coupled receptor (GPCR) and induces Arf1dependent COPI vesicle formation upon cargo binding [27]. GPRCs activate immature intracellular TrkA [28][29][30], suggesting an intriguing possibility that cargo-activated KDEL-Rs may activate TrkAIII, resulting in retrograde transport of KDEL-R/COPI vesicles containing KDELRactivated TrkAIII. Although immature TrkA receptors were not activated under similar conditions, this may explain the selective activation of TrkAIII within the ERGIC/COPI compartment. We are investigating this possibility.
Alternatively, the recycling of activated TrkAIII from the ERGIC to ER may result from motor proteindriven, MT minus-end, retrograde receptor transport, analogous to that described for the retrograde transport of ligand-activated cell surface TrkA receptors [31][32][33]. Such a mechanism would not only explain recycling back to the ER but also the accumulation of active TrkAIII at the centrosome [3]. Whatever the mechanism, the net result is self-perpetuating TrkAIII re-localisation to the ER/ www.impactjournals.com/oncotarget ERGIC compartment, with the subsequent combination of entrapment and accumulation of this compromised receptor augmenting the potential for spontaneous activation.
TrkAIII activation within ERGIC/COPI compartment differs to activation of the TrkT3 oncogene, which occurs within the ERES/COPII compartment [13]. This can be explained by different re-localisation mechanisms. TrkT3 is trapped within the ERES/COPII compartment by chimeric TFG sequences, not present in TrkAIII, which regulate COPII vesicle formation [13]. Both mechanism, however, result in spontaneous activation and oncogenic Akt signalling from their respective locations, confirming that TrkA oncoproteins, compromised either by mutation or alternative splicing, may re-localise to different intracellular membranes but become activated when they accumulate above the local threshold for spontaneous activation.
In contrast to TrkAIII, TrkA did not exhibit spontaneous activation in SH-SY5Y cells, clearly implicating the extracellular D4 IG-like domain and associated N-glycosylation sites omitted from TrkAIII, in preventing spontaneous intracellular TrkA activation. This most likely depends upon the optimisation of intracellular TrkA trafficking to the GN, GN-associated N-glycan maturation and subsequent expression at the cell surface [7][8][9]. The optimisation of intracellular TrkA trafficking to the GN appears to be key in this respect, as illustrated by the recent report of spontaneous ERGICassociated activation of immature TrkA in human kidney 293 cells [14]. This cell specific effect, characterised by aberrant TrkA accumulation within the ERGIC above the spontaneous activation threshold, results in GN disruption, which promotes the entrapment of immature TrkA within the ERGIC, illustrating the absolute requirement for optimised intracellular trafficking to the GN in preventing spontaneous intracellular activation [14].
TrkAIII did not exhibit spontaneous activation throughout the ER but could be activated throughout the ER by the PTPase inhibitor sodium orthovanadate, implicating ER PTPases in maintaining ER-associated TrkAIII inactive [2]. This indicates that active TrkAIII recycling back to the ER would be inactivated within the ER prior to being re-transport back to the ERGIC in COPII vesicles, the formation of which would be enhanced by TrkAIII-induced Akt signalling [1,26]. Spontaneous TrkAIII activation within the GN was also minimal. This may reflect either the low level of TrkAIII in COPnegative GN membranes and/or the RTK inhibitory nature of the GN, which is exemplified by the steady state GNassociated pool of inactive TrkA, consistent with the GN function as natural reservoir for receptor tyrosine kinases during their maturation [34].
TrkA inhibitors promoted TrkAIII transport from the ER to GN, resulting in GN-associated TrkAIII maturation to 120kDa, analogous to GN-associated TrkA maturation from 110kDa to 140kDa (this study and [1,3]). In contrast, hyper-phosphorylation of TrkAIII induced by sodium orthovanadate did not result in 120kDa TrkAIII maturation, supporting an inverse relationship between TrkAIII activation and maturation. This bears similarity to ERGIC-associated activation of immature 110kDa TrkA receptors in human kidney 293 cells, which impedes 140kDa TrkA maturation by disrupting the GN [14]. Here, we found no evidence for GN disruption in TrkAIII SH-SY5Y cells, suggesting that impaired maturation of TrkAIII involves the reduced transport of activated TrkAIII to the GN, rather than GN disruption.
The proteasome inhibitor MG132 promoted lowlevel expression of 120kDa TrkAIII without influencing 100kDa TrkAIII levels in TrkAIII SH-SY5Y cells. This indicates that a small proportion of TrkAIII must constitutively arrive at the GN and mature to 120kDa but is not detected in the absence of MG132 due to degradation at the proteasome. Furthermore, MG132 augmented 120kDa TrkAIII levels induced by CEP-701, restored 120kDa TrkAIII expression to CEP-701-treated TrkAIII SH-SY5Y cells in which 120kDa TrkAIII expression had been abrogated by the Hsp90 inhibitor GA and, in the presence of TrkA inhibitors, caused the re-distribution of TrkAIII to cellular projections and intracellular junctions, suggesting cell surface expression. Together, these data indicate that: TrkA inhibitors promote GN-associated TrkAIII 120kDa maturation; 120kDa TrkAIII is more sensitive than 100kda TrkAIII to degradation at the proteasome; 120kda TrkAIII is stabilised by GA sensitive interaction the Hsp90 adding to our report that TrkAIII exhibits GA-sensitive interaction with Hsp90 [2], and that cell surface expression of 120kDa TrkAIII is prevented by its degradation at the proteasome.
In conclusion, we propose that the novel relocalisation mechanism characterised by spontaneous ERGIC/COPI-associated activation and subsequent retrograde recycling back to the ER is responsible for the entrapment, accumulation, spontaneous activation and subsequent oncogenic activity of TrkAIII and is initiated from the alternative location of the ERGIC/ COPI compartment. This highlights the importance of intracellular TrkAIII re-localisation for oncogenic activity, identifies the ERGIC/COPI compartment as a preferential site for spontaneous TrkAIII activation and provides an explanation for how TrkAIII, compromised by the omission of the D4 IGC1 spontaneous activationprevention domain, is able to overcome the threshold for spontaneous activation. The subsequent and altered oncogenic signalling through PI3K/Akt but not Ras/ MAPK results in increased survival, a pattern of malignant gene expression, increased genetic instability and promotion of stem cell-like behaviour rather than differentiation [1,3,4,6,35], which together not only promote tumour progression but may also be involved in tumour initiation. TrkA tyrosine kinase inhibitors abrogate this mechanism by promoting the anterograde transport of inactivated TrkAIII to the GN, resulting in GN-associated TrkAIII maturation to a 120kDa form that is degraded at the proteasome, reducing both intracellular accumulation and oncogenic activity of the TrkAIII oncoprotein.

Immunoprecipitation and western blots
Cells were extracted in lysis buffer (PBS containing 0.5% sodium deoxycholate, 1% NP40, 0.1% SDS, 1mM sodium orthovanadate, 1mM PMSF, 1µg/ml of pepstatin A and Aprotinin), protein concentrations were calculated by Bradford Assay (Sigma-Aldrich) and extracts pre-cleared with 1µg pre-immune IgG (1 hour at 4°C) and 20µl protein A Sepharose (Fast Flow, Sigma) for 20 minutes at 4°C. For immunoprecipitation, pre-cleared extracts (200-500µg for stable transfectants) were incubated with primary antibody at a concentration of 0.1-1µg/500µg total protein for 2-16 hours at 4°C. Following incubation, 20µl Protein A Sepharose (Fast flow, Sigma-Aldrich) in lysis buffer, was added and reactions incubated for 60 minutes at 4°C. Protein sepharose/IgG conjugates were collected by centrifugation (10,000xg for 5 minutes), washed 3 times in lysis buffer, re-suspended in SDS-PAGE sample buffer and subjected to reducing SDS-PAGE/Western blotting. Proteins were trans-blotted by electrophoresis onto Hybond C+ nitrocellulose membranes (Amersham Int. UK) and membranes air-dried. Non-specific protein binding site were blocked by 2 hours incubation in 5% non-fat milk in TBS, prior to incubation for 2-16 hours with primary antibodies diluted in blocking solution at 4°C. Following incubation, membranes were washed in TBS, incubated for 20 minutes with secondary HRPconjugated antibody diluted in blocking solution and immunoreactive species detected by chemiluminescence reaction, as directed (Amersham Int., Bedford, UK).

Metabolic labelling of trkAI and trkAIII
Briefly, 3-4 million cells per time point, preincubated for 8 hours in the presence or absence of 100nM CEP-701, were incubated for 2 hour in methionine and cysteine-free medium prior to exposure for 30 minutes to 100µCi/ml S 35 -labelled methionine/cysteine (Perkinwww.impactjournals.com/oncotarget Elmer). Cells were then washed with pre-warmed complete medium and chased in excess (1mM) cold methionine/cysteine (Sigma-Aldrich), for different times. Cell aliquots were taken at each time point and subjected to Histodenz TM ultracentrifugation fractionation as described above, membrane fractions separated by regular reducing SDS-PAGE and dried gels were subjected to autoradiography. Densitometric analysis was performed using Image J software.

Regular and confocal immunofluorescence (IF)
Cells grown on Nunc glass chamber slides (Sigma-Aldrich) were washed in PBS, fixed in 37% formaldehyde, permeabilized in cold methanol and processed for indirect immunofluorescence (IF). Slides were incubated for 1 h in blocking solution (1% bovine serum albumin in PBS-0.03% TX100) and then for 2 to 16 h with primary antibody in blocking solution at room temperature. Slides were washed three times in PBS-0.03% TX100, then incubated with secondary fluorochrome-conjugated antibody diluted in blocking solution for 1 h at room temperature, and then washed in PBS-0.03% TX100 and mounted using Vector Mount. For confocal analysis, slides were examined under a Nikon Eclipse 80i confocal microscope and images elaborated using EZ-C1 Gold software (version 3.80). For non-confocal analysis, slides were examined under a Zeiss Axioplan 2 fluorescence microscope, with digital camera and Leica M500 Image Manager software.

statistical analysis
Data were analysed statistically by Student's t-test. Statistical significance was associated with probabilities of < 0.05.