The neural cell adhesion molecule (NCAM) is involved in neural development and in plasticity in the adult brain. NCAM140 and NCAM180 isoforms are transmembrane proteins with cytoplasmic domains that differ only in an alternatively spliced exon in the NCAM180 isoform. Both isoforms can interact with several extracellular and cytoplasmic molecules mediating NCAM-dependent functions. Most identified intracellular interaction partners bind to both isoforms, NCAM140 and NCAM180. To identify further intracellular interaction partners specifically binding to NCAM180 the cytosolic domain of human NCAM180 was recombinantly expressed and applied onto a protein macroarray containing the protein library from human fetal brain. We identified the ubiquitin C-terminal hydrolase (UCHL1) which has been described as a de-ubiquitinating enzyme as a potential interaction partner of NCAM180. Since NCAM180 and NCAM140 are ubiquitinated, NCAM140 was included in the subsequent experiments. A partial colocalization of both NCAM isoforms and UCHL1 was observed in primary neurons and the B35 neuroblastoma cell line. Overexpression of UCHL1 significantly decreased constitutive ubiquitination of NCAM180 and NCAM140 whereas inhibition of endogenous UCHL1 increased NCAM's ubiquitination. Furthermore, lysosomal localization of NCAM180 and NCAM140 was significantly reduced after overexpression of UCHL1 consistent with a partial colocalization of internalized NCAM with UCHL1. These data indicate that UCHL1 is a novel interaction partner of both NCAM isoforms that regulates their ubiquitination and intracellular trafficking.
Structured digital abstract
cytoplasmic domain of human NCAM180
neural cell adhesion molecule
turned on after division
ubiquitin C-terminal hydrolase
Cell adhesion molecules are important players in neural development and plasticity. The neural cell adhesion molecule (NCAM) is a member of the immunoglobulin superfamily of Ca2+-independent cell adhesion molecules. It is widely expressed in the central and peripheral nervous system but also in other tissues. Several functions have been ascribed to NCAM in the nervous system, such as involvement in regulation of cell migration, synaptic plasticity, axonal growth and fasciculation [1-7]. NCAM is expressed as three major isoforms termed NCAM120, NCAM140 and NCAM180. NCAM140 and NCAM180 are transmembrane isoforms whereas NCAM120 is anchored to the membrane via glycosylphosphatidylinositol (GPI). The NCAM140 and NCAM180 isoforms differ only in an alternatively spliced exon resulting in an additional 267 cytosolic amino acids in NCAM180. Functions of NCAM are mediated by several interactions with other molecules through its extracellular domain and its cytoplasmic tail . Intracellular binding partners of NCAM include cytoskeletally associated proteins and signaling molecules. Among the cytoskeletal binding proteins that NCAM interacts with are spectrin, which builds a molecular bridge between NCAM and protein kinase C (PKC) β2, growth-associated protein 43, α- and β-tubulin, α-actinin, microtubule-associated protein 1A, β-actin, tropomyosin and RhoA-binding kinase-α. Further cytosolic interaction partners of NCAM include the proteins leucine-rich acidic nuclear protein, syndapin, TOAD-64, protein phosphatase 1 (PP1) and PP2A [9-15]. After initiation by homophilic or heterophilic binding, NCAM can activate several signaling molecules, e.g. PKA, phospholipase C (PLC) γ, PKC β2, receptor protein tyrosine phosphatase α and the non-receptor tyrosine kinase p59fyn. NCAM signaling is known to be required for NCAM-dependent neurite outgrowth and cell migration [16-23].
All three major isoforms of NCAM are extensively post-translationally modified including intracellular phosphorylation and ubiquitination of the NCAM140 and NCAM180 isoforms [24-28].
During the last decade it became obvious that ubiquitin, besides its role as a proteasomal degradation signal , can also regulate many other processes including endocytosis, cellular sorting and signaling. Ubiquitin is usually covalently attached to lysine residues of a target protein. Depending on how many ubiquitin molecules are attached to the substrate protein one can distinguish between mono-, multiple mono- or poly-ubiquitination. Adding another level of complexity, all different kinds of ubiquitin modification have been implicated in the regulation of endocytosis .
We found that ubiquitin regulates at least partially NCAM's endocytosis in neurons and that the major part of internalized NCAM is recycled to the cell surface whereas only a small amount becomes lysosomally degraded . The ubiquitination of a target protein is regulated by different enzyme systems. The specificity of ubiquitin attachment is usually controlled by ubiquitin ligases that bind to the target protein to transfer ubiquitin . In contrast, the specific removal of ubiquitin from target proteins is controlled by de-ubiquitinating enzymes (DUBs). Since ubiquitinated molecules can be sorted to either proteasomes or lysosomes for degradation, depending on the kind of ubiquitin modification , one important function for DUBs is to prevent cell surface molecules from degradation by removing ubiquitin. They will instead be recycled to the cell surface [31-34]. How ubiquitination of NCAM is regulated is not yet known.
Although many intracellular interaction partners of NCAM have been identified in the last few decades, many aspects of NCAM function and signal transduction remain unclear. Therefore, we aimed at identifying novel intracellular interaction partners of NCAM. We found the ubiquitin C-terminal hydrolase (UCHL1) to be a potential novel interaction partner that is functionally involved in ubiquitination and intracellular trafficking of NCAM140 and NCAM180.
Identification of potential interaction partners of the cytosolic domain of hNCAM180
To identify cytoplasmic proteins that interact with the cytoplasmic domain of human NCAM180 (hNCAM180cyt) recombinant hNCAM180cyt was expressed and purified by ligand affinity chromatography (Fig. 1). Purified hNCAM180cyt was applied on a protein macroarray comprising 25 300 expression clones of human fetal brain to screen for hNCAM180cyt binding partners. Detailed analysis of expression clones which had bound hNCAM180cyt revealed already identified binding partners of NCAM, i.e. PLC-γ , spectrin [9, 10], tubulin  and PP1 . Additionally, we detected further proteins that have not yet been identified as interaction partners of NCAM's cytoplasmic domain, interestingly including UCHL1 as a potential interaction partner of hNCAM180cyt.
Colocalization of NCAM and UCHL1
To determine if NCAM colocalizes with UCHL1 in neurons double immunofluorescence staining was performed. Endogenous NCAM and UCHL1 colocalized partially in cell somata and neurites of primary cortical neurons (Fig. 2A). Colocalization of UCHL1 and NCAM was also investigated in B35 cells, which is a well established cell culture model of central nervous system neurons [28, 35]. NCAM140 was included in the subsequent experiments since it is also ubiquitinated. NCAM180 or NCAM140, respectively, was expressed in B35 cells and analyzed for colocalization with endogenous UCHL1. For both NCAM isoforms a partial colocalization could also be observed in B35 cells (Fig. 2B). These data support a potential interaction between NCAM and UCHL1 in neurons.
Overexpression of UCHL1 decreases ubiquitination of NCAM
Earlier results showed that NCAM is constitutively ubiquitinated. When endocytosis of NCAM was induced by NCAM-specific antibodies, the ubiquitination level was drastically increased . We therefore investigated next whether UCHL1 influences the constitutive or endocytosis-induced ubiquitination level of NCAM. We found that the constitutive ubiquitination of NCAM180 was significantly reduced in the presence of UCHL1 (37.6 ± 13.8%) compared to the control (100%). In contrast, ubiquitination of NCAM180 was not altered in the presence or absence of UCHL1 when endocytosis of NCAM was induced. Inhibition of UCHL1 activity by the specific inhibitor LDN-91946 significantly increased constitutive NCAM180 ubiquitination (151.3 ± 4.6%) further confirming a functional role of UCHL1 in NCAM ubiquitination (Fig. 3). As for NCAM180, the constitutive ubiquitination of NCAM140 was significantly decreased by UCHL1 expression (24.0 ± 7.4%, Fig. 3) compared to 100% in the control. No significant effect of UCHL1 was detected when NCAM140 endocytosis was induced. To exclude an effect of UCHL1 on the ubiquitination machinery in general and to verify the specificity of UCHL1 overexpression on NCAM, we also analyzed whether the ubiquitination level of the related cell adhesion molecule L1 was influenced by its overexpression in B35 cells. No difference in ubiquitination of L1 could be observed in the presence or absence of UCHL1 (not shown).
UCHL1 decreases lysosomal degradation of NCAM
NCAM ubiquitination regulates at least partially NCAM's endocytosis . To determine whether UCHL1 influences NCAM endocytosis and/or intracellular trafficking by altering its ubiquitination these aspects were investigated next. For these experiments, NCAM endocytosis was induced by NCAM-specific antibodies. Interestingly, NCAM endocytosis was not influenced by overexpression of UCHL1 (not shown). However, significantly less NCAM180 and NCAM140 was detected in lysosomes after overexpression of UCHL1 (19.3 ± 1.4% for NCAM180 and 20.9 ± 1.4% for NCAM140) compared to the control (27.4 ± 2.5% for NCAM180 and 27.7 ± 1.8% for NCAM140; Fig. 4) indicating that de-ubiquitination of NCAM by UCHL1 might increase its recycling concomitantly with decreasing lysosomal degradation of NCAM.
UCHL1 colocalizes partially with internalized NCAM
If UCHL1 plays a role in recycling of NCAM to the cell surface it has to be associated not only with cell surface NCAM but particularly with internalized NCAM. Therefore we analyzed whether internalized NCAM colocalizes with UCHL1. In B35 cells, expressing either NCAM180 or NCAM140, a small portion of internalized NCAM colocalized with UCHL1 (Fig. 5A,B,D,E). The colocalization is highlighted in higher magnification (Fig. 5B,E, left side) and by the intensity profiles of the single channels of the confocal microscope (Fig. 5B,E, right side). After treatment with cyclic AMP B35 cells differentiate and develop growth cones similar to the growth cones of primary cortical neurons . Focusing on the growth cones of B35 cells an obvious colocalization of internalized NCAM180 or NCAM140, respectively, with endogenous UCHL1 could be observed (Fig. 5C,F).
In this work we identified UCHL1 as a possible novel interaction partner of hNCAM180cyt using a protein macroarray. An interaction between UCHL1 and NCAM appears highly likely since NCAM is ubiquitinated and UCHL1 is one of the most abundant proteins in the brain accounting for 1–2% of total soluble brain proteins [28, 36, 37]. It is exclusively localized in neurons which is comparable with the main expression of NCAM140 and NCAM180 isoforms in neurons . Consistently, we found a partial colocalization of UCHL1 and NCAM in primary neurons and B35 neuroblastoma cells further supporting a potential interaction between the two molecules. Unfortunately, we were not able to show an interaction directly by coprecipitation of NCAM180 and UCHL1 due to the proximity of the UCHL1 signals to the antibody signals from the immunoprecipitation and unspecific signals at the expected molecular weight. Therefore, we tested directly whether UCHL1 exerts functional effects on NCAM. NCAM140 was included in our studies since it is also ubiquitinated. We found that UCHL1 decreases the constitutive ubiquitination of both NCAM isoforms. This result indicated that UCHL1, which has been described not only as a DUB but, in dimeric form, also as a ubiquitin ligase or as a stabilizer of mono-ubiquitin independent of its enzymatic activity [37, 39-41], may act in the case of NCAM as a DUB.
A connection of NCAM with UCHL1 has already been demonstrated in a rat model. Reduced expression of NCAM180 after passive avoidance learning was accompanied by increased expression of UCHL1. Both decreased NCAM expression and increased UCHL1 expression were inhibited by administration of C3d peptide which results in NCAM-dependent temporary amnesia indicating the specificity of alterations of UCHL1 expression in NCAM-dependent learning processes . It should be noted that we did not observe a difference in UCHL1 expression in NCAM−/− mice compared with wild-type mice in brain lysates (not shown) indicating that NCAM absence itself does not alter UCHL1 expression. In contrast to our data the results from Foley et al. indicate a ubiquitin ligase function of UCHL1 leading to ubiquitination of NCAM and subsequent degradation . However, in that study ubiquitination of NCAM was not investigated after C3d treatment. The discrepancy with our results may be attributable to a possible dimeric form of UCHL1, to its mono-ubiquitin stabilizing function or to different NCAM-dependent mechanisms involved in the specific training system applied by Foley et al.  compared with our system. The latter hypothesis is not unlikely since NCAM can interact with several different molecules and activate different signal transduction pathways .
We reported in a previous study that triggering of NCAM endocytosis with NCAM-specific antibodies increased ubiquitination of NCAM and that ubiquitination may represent an internalization signal for NCAM . Interestingly, UCHL1 overexpression had no effect on NCAM's ubiquitination after induction of its endocytosis; exclusively constitutive ubiquitination of NCAM was affected. Likewise, we did not detect any effect of UCHL1 overexpression on endocytosis of NCAM. This observation fits to our previously described model that increased ubiquitination is necessary for effective NCAM endocytosis . Since the ubiquitination level after induction of NCAM endocytosis was not altered by UCHL1 no difference in NCAM endocytosis would be expected. We observed instead a reduced lysosomal degradation of internalized NCAM by UCHL1 overexpression. Triggering of NCAM endocytosis by NCAM-specific antibodies leads to clustering of NCAM at the cell surface . This may result in a different conformation of NCAM at the plasma membrane and could thereby mask the binding site for UCHL1 in the cytoplasmic domain of NCAM. In this model it is conceivable that UCHL1 may interact with the portion of NCAM at the cell surface (without triggering NCAM endocytosis) that is constitutively ubiquitinated, thus reducing its constitutive ubiquitination level. After triggering of NCAM endocytosis UCHL1 may only bind to NCAM after its internalization which would explain its inefficiency for altering the endocytosis of NCAM. In this scenario it is also comprehensible that no difference of NCAM ubiquitination could be observed since even after induction of NCAM endocytosis only a small amount of NCAM is internalized. Changes in ubiquitination of exclusively internalized NCAM would probably not be detectable with our methods. However, a small although not significant decrease of ubiquitination was observed for NCAM140 in the presence of UCHL1 after triggering NCAM endocytosis. In agreement with this hypothesis is our result that also internalized NCAM colocalizes with UCHL1 to a small extent in cell somata but with higher probability in growth cones. A logical conclusion would be that UCHL1 might increase recycling of NCAM by preventing lysosomal degradation (see Fig. 6 for the proposed model).
It has been demonstrated earlier that the balance between ubiquitination and de-ubiquitination is critical for a variety of cellular functions . For plasma membrane proteins it is known that mono-ubiquitination can act as a signal for endocytosis but also as a signal to sort and target proteins in the endosome into the multivesicular body/lysosomal pathway [45-47]. In contrast, DUBs can prevent membrane proteins from lysosomal degradation by de-ubiquitination and promote their recycling to the cell surface . Examples of this mechanism are the epithelial Na+ channel in kidney cells, the cystic fibrosis transmembrane conductance regulator CFTR and the β2-adrenergic receptor [32-34].
In neurons the balance between ubiquitination and de-ubiquitination affects synapse function. This has been demonstrated, for example, in genetic studies for the Drosophila de-ubiquitinating enzyme Fat facets and the ubiquitin ligase Highwire . Recently, Kowalski et al. showed that the DUB ubiquitin-specific protease-46 regulates de-ubiquitination of the glutamate receptor GLR-1 and thus controls the abundance of GLR-1 in the ventral nerve cord of Caenorhabditis elegans . UCHL1 is also known to play a role in synaptic function. Its disruption leads to defects in long-term facilitation and in synaptic function and structure [51-54] and it is therefore important for neural development and plasticity in adult brain. Endocytosis and recycling of membrane proteins are also required for correct synaptic function. How NCAM intracellular trafficking is necessary for neurite outgrowth and synaptic function is not yet known but one can speculate that NCAM endocytosis and recycling might influence the presence of other synaptic components at the synapse, especially since NCAM colocalization with UCHL1 seemed to be more pronounced in growth cones. NCAM can target for example the N-methyl-d-aspartate subtype of glutamate receptors to post-synaptic densities [55, 56].
In conclusion this study shows that UCHL1 regulates de-ubiquitination of NCAM and most probably recycling of NCAM and might therefore be important for NCAM-dependent functions in the nervous system.
Materials and methods
Hybridoma cells producing monoclonal antibody 123C3 against human NCAM were provided by R. Michalides (Amsterdam, The Netherlands), monoclonal NCAM antibody (5B8) producing hybridoma cells were kindly provided by R. Horstkorte (Halle, Germany) and antibody against human L1 (L1-11A, subclone of UJ 127.11) by P. Altevogt (Heidelberg, Germany) ; antibody against ubiquitin (P4D1) was purchased from Cell Signaling Technology (Beverley, MA, USA), tetra His-antibody was from Qiagen (Hilden, Germany) and monoclonal rabbit UCHL1 antibody from Abcam (Cambridge, UK). Secondary antibodies conjugated to Cy3 or horseradish peroxidase were from Dianova (Hamburg, Germany); Alexa-Fluor-633- or Alexa-Fluor-488-conjugated secondary antibodies and LysoTracker® were from Molecular Probes (Eugene, OR, USA).
Expression plasmids for human NCAM140 or NCAM180 in B35 cells have already been described [25, 58]. pcDNA3-L1 cDNA was provided by P. Maness (Chapel Hill, NC, USA). Human UCHL1 expression plasmid was kindly provided by L. Stefanis (Athens, Greece) and the prokaryotic expression plasmid pBJG1 containing a His-tag epitope by Gross et al. .
Expression and purification of recombinant proteins
The cytoplasmic region of hNCAM180cyt was generated using the following primers: forward 5′-GGA TTC GGA TTC ATG GAC ATC ACC TGC TAC TTC CTG AAC AAG-3′ and reverse 5′-CTC GAG CTC GAG TGC TTT GCT CTC GTT CTC CTT TG-3′, restricted with BamHI and XhoI and ligated into the prokaryotic expression plasmid pBJG1. The expression of hNCAM180cyt in BL21 (DE3) bacteria was induced with 1 mm isopropyl thio-β-d-galactoside for 3.5 h at 37 °C. Cells were harvested in NaCl/Pi containing 1% Triton-X100 and lysed by 10 freezing/thawing cycles with liquid nitrogen. hNCAM180cyt was purified by Ni-nitrilotriacetic acid affinity chromatography (Qiagen) according to the manufacturer's instructions. The purified protein was concentrated to a minimum of 1 mg·mL−1 in NaCl/Pi and fluorescently labeled with Dyomics DY-633 Fluorophore (Dyomics, Jena, Germany) according to the manufacturer's instructions.
The macroarray membrane (Source Bioscience imaGenes, Berlin, Germany) containing 25 300 different His-tagged expression clones of human fetal brain was rinsed in 96% ethanol to lyse bacterial colonies followed by ultrapure water. Excess colony material was removed using paper tissue. After washing with NaCl/Pi the membrane was blocked with Odyssey blocking buffer (Licor Biosciences, Bad Homburg, Germany) for 2 h at room temperature. Fluorescently labeled hNCAM180cyt (15 nm) was incubated with the prepared membrane in Odyssey blocking buffer NaCl/Pi/0.05%Tween-20 (ratio 1 : 1) for 16 h. After washing four times for 1 h with NaCl/Pi/0.05% Tween-20 and a final washing step with NaCl/Pi, signals were detected using the Licor Odyssey scanner. For removal of bound proteins the membrane was incubated with heated stripping buffer (2% SDS, 65.5 mm Tris/HCl pH 6.8, 100 mm β-mercaptoethanol) for 30 min and washed with NaCl/Pi. To determine expression levels of spotted proteins the membrane was blocked again with Odyssey blocking buffer and incubated with Alexa-Fluor-800-conjugated anti tetra-His-antibody (Qiagen) as described above. The resulting images were analyzed with Aida Image Analyzer version 4.24 (Raytest, Straubenhardt, Germany). The expression clones on the macroarray membrane are spotted twice in a characteristic pattern around a guiding point. Fluorescent signals were only counted as positive if both spots of a clone were significantly brighter than the background and the signal derived from binding of hNCAM180cyt to the corresponding expression clones significantly more intense than the signal derived from the expression control (tetra-His antibody). Correlation of positive spots to the corresponding expression clones was carried out using the Source Bioscience imaGenes Scoring Template allowing a mapping of the x, y positions of positive signals to the expression clones spotted on the membrane via the annotation table also available from Source Bioscience imaGenes.
Cell culture and transfections
B35 neuroblastoma cells have been described earlier . Cells were transiently co-transfected with full-length human NCAM140 or NCAM180 constructs (described in [25, 58]) and UCHL1 or empty expression plasmid, respectively, using Turbofect™ reagent (Fermentas, St Leon-Rot, Germany). Primary cortical neurons were prepared as described earlier . Immunohistochemical experiments were carried out after 6 days in vitro.
Endocytosis and colocalization studies
For induction of NCAM endocytosis cells were treated as described in . Briefly, endocytosis was induced for 1 h with 123C3 antibody at 37 °C. Cells were fixed with 8% paraformaldehyde in NaCl/Pi and were incubated for 30 min with Alexa-Fluor-633-conjugated anti-mouse antibodies (1 : 500 in blocking solution, 5% horse serum in NaCl/Pi) for the detection of cell surface associated NCAM. In some of the experiments all binding sites of the first antibody were subsequently saturated by incubation with rabbit anti-mouse immunoglobulins (0.25 mg·mL−1 in blocking solution). Then cells were postfixed for 5 min with 8% paraformaldehyde at 4 °C and permeabilized with 0.5% Triton X-100 (20 min). For colocalization of internalized NCAM with endogenous UCHL1 a 30-min incubation step with UCHL1-specific antibodies (1 : 100 in blocking solution) was performed. Internalized NCAM was visualized using Alexa-Fluor-488-conjugated anti-mouse antibodies (1 : 150) and UCHL1 with Cy3-conjugated anti-rabbit antibodies. Finally cells were embedded in Permafluor (Beckman-Coulter, Marseille, France). Images were taken using a Zeiss LSM510 MetaUV confocal microscope (Oberkochen, Germany). For colocalization of endocytosed NCAM with LysoTracker® endocytosis was induced for 1 h as described above. LysoTracker® Red (200 nm) was added additionally to medium during endocytosis induction. Cells were then treated as described for analysis of endocytosis with the only difference that cell surface NCAM was detected using anti-mouse Alexa-Fluor-633-conjugated secondary antibodies.
Immunoprecipitation and immunoblotting
For immunoprecipitation B35 cells were plated on plastic dishes 10 cm in diameter. Cells were transfected with the corresponding constructs and differentiated as described in . For some experiments cells were pretreated with UCHL1 inhibitor (LDN 91946; Sigma-Aldrich, Taufkirchen, Germany) for 6 h at a concentration of 100 μm or control treated and on some dishes endocytosis was induced for 1 h prior to cell lysis in RIPA buffer (c.c.pro, Oberdorla, Germany). Protein concentration was determined using the Biorad Dc protein assay (München, Germany). The same protein amounts of lysed cells were subjected to immunoprecipitation with 5B8 antibody and protein G sepharose (GE Healthcare, Freiburg, Germany) overnight at 4 °C. Immunoprecipitated NCAM was eluted by heating in sample buffer (5 min, 95 °C) and the supernatants were applied onto an SDS gel and subjected to immunoblot analysis as described in . To control the amount of precipitated NCAM, antibodies were removed from the membrane and the same membrane was again incubated with 5B8 antibodies (250 ng·mL−1) and corresponding peroxidase-conjugated secondary antibodies.
Quantification and statistical analysis
Endocytosis was quantified by counting NCAM-positive vesicles inside the cell. Quantification of internalized NCAM in lysosomes was performed by counting NCAM-positive vesicles and vesicles containing NCAM and LysoTracker®. These two values were set in relation to each other and the percentage of colocalization was calculated. Data were calculated from at least three different experiments with approximately 15 cells analyzed in each experiment. Statistical analysis was carried out using the unpaired t test and data are presented as the mean ± SEM. Densitometric analysis of the immunoblot experiments was carried out using imagej software.
We thank S. Schmidt for preparation of cortical neurons and H. Faraidun for subcloning hNCAM180cyt.