Kinesin-1 promotes post-Golgi trafficking of NCAM140 and NCAM180 to the cell surface

The neural cell adhesion molecule (NCAM, also known as NCAM1) belongs to the immunoglobulin superfamily of Ca²⁺-independent cell adhesion molecules. Alternative splicing of a single gene gives rise to three major isoforms named after their apparent molecular masses, NCAM120, NCAM140 and NCAM180. NCAM120 is anchored to the surface membrane through a glycosylphosphatidylinositol (GPI) anchor, whereas NCAM140 and NCAM180 are transmembrane glycoproteins with intracellular domains of different lengths (Walsh and Doherty, 1991).

NCAM is highly expressed in the developing central and peripheral nervous systems, where it plays important functional roles in cell adhesion, cell migration, axonal outgrowth and fasciculation, as well as synapse formation. In the adult NCAM is involved in regulating synaptic plasticity (Chazal et al., 2000; Chipman et al., 2014; Cremer et al., 1994, 1997; Doherty et al., 1990; Hinsby et al., 2004; Kiryushko et al., 2004; Maness and Schachner, 2007; Muller et al., 1996; Panicker et al., 2003; Sytnyk et al., 2004). Being expressed at the cell surface of neurons and glial cells, NCAM induces intracellular signal transduction pathways and cytoskeleton rearrangements in response to interactions with its extracellular domain (Büttner et al., 2003; Ditlevsen et al., 2008; Leshchyns'ka et al., 2003; Li et al., 2013; Nielsen et al., 2010; Pollerberg et al., 1987; Walmod et al., 2004).

Transport of NCAM from the cell body and through the translational machinery in cellular processes is a prerequisite for cell surface delivery. In axons, NCAM is distributed by fast axonal transport (Garner et al., 1986; Nybroe et al., 1986). In dendrites and glial cell processes, NCAM also needs to be distributed to the cell surface after its transport to the sites of insertion. The molecular mechanisms of the transport and cell surface delivery of NCAM are incompletely understood. It therefore deemed pertinent to search for the molecules that bind to the intracellular domain(s) of the two transmembrane NCAM isoforms and could be involved in NCAM transport and cell surface delivery. By protein macroarray screening using the intracellular domains of NCAM as baits, we identified kinesin light chain 1 (KLC1) as a new binding partner of the intracellular domains of NCAM140 and NCAM180.

KLC1 is part of kinesin-1, a motor protein that transports cargoes towards the plus end of microtubules in axons and dendrites (Brady, 1985; Hirokawa et al., 1991; Vale et al., 1985). Kinesin-1 is composed of two kinesin heavy chains (KHCs; KIF5A, KIF5B and KIF5C isoforms), which use ATP to energize the movement along microtubules, and two light chains (KLCs), which play a role in cargo attachment to kinesin-1 heavy chains (Cabeza-Arvelaiz et al., 1993; Hirokawa, 1998; Hirokawa et al., 1989; Niclas et al., 1994; Rahman et al., 1998; Verhey et al., 1998; Xia et al., 1998; Yang et al., 1990).

In the present study, we show that KLC1 binds directly to the eight-amino-acid sequence comprising amino acids 747–754 in the intracellular domains of human NCAM140 and NCAM180, and promotes cell surface delivery of NCAM.

To identify new intracellular interaction partners of the intracellular domains of human NCAM180 and NCAM140 (hNCAM180-ID and hNCAM140-ID), recombinantly expressed highly purified hNCAM180-ID and hNCAM140-ID were individually applied to a protein macroarray comprising 24,000 expression clones from human fetal brain. In addition to the already identified binding partners of NCAM, such as spectrin (Leshchyns'ka et al., 2003; Pollerberg et al., 1986), PLCγ and PP2A (Büttner et al., 2005), we observed binding of hNCAM180-ID and hNCAM140-ID to KLC1 (Q07866; note UniProt accession numbers are given in brackets) (Fig. 1A). hNCAM180-ID and hNCAM140-ID did not bind to kinesin light chain 2 (KLC2) (Q9H0B6), kinesin light chain 4 (KLC4) (Q9NSK0), kinesin-like protein KIF1A (Q12756), kinesin-like protein KIF2C (Q99661), kinesin-like protein KIF3C (O14782), kinesin-like protein KIF22 (Q14807), kinesin-like protein KIF15 (Q9NS87), kinesin-like protein KIF21B (O75037), kinesin-like protein KIFC3 (Q9BVG8), kinesin heavy chain isoform KIF5C (O60282), kinesin heavy chain isoform KIF5A (Q12840), kinesin-like protein KIF21A (Q7Z4S6), or chromosome-associated kinesin KIF4A (O95239).

To confirm the association between KLC1 and NCAM, they were immunoprecipitated from detergent lysate of 1-day-old mouse brains, and immunoprecipitates were subjected to western blot analysis with antibodies against NCAM and KLC1. NCAM immunoreactivity appeared as a broad protein band with an apparent molecular mass above 180 kDa (Fig. 1B,C). The broad range of this NCAM band composed of different molecular masses derives from the very heterogeneous lengths of polysialic acid chains attached to the NCAM protein backbone and has been observed to be a prominent phenomenon in immature brains (Hoffman et al., 1982; Rothbard et al., 1982). NCAM co-immunoprecipitated with KLC1 (Fig. 1B), and KLC1 co-immunoprecipitated with NCAM (Fig. 1C). Hence, KLC1 and NCAM are associated with each other.

Next, we aimed to identify the binding site for KLC1 in NCAM. Given that both hNCAM140-ID and hNCAM180-ID bound to KLC1 in the macroarray assay, we analyzed binding of KLC1 to amino acid sequences of hNCAM140-ID and hNCAM180-ID that were close to the plasma membrane. This amino acid sequence and the position of this stretch relative to the plasma membrane are identical in both transmembrane NCAM isoforms (Fig. 1D). To analyze the binding between KLC1 and this amino acid sequence, KLC1 was immunopurified from brain lysates and immobilized as bait on a plastic surface for ELISA. Chemically synthesized peptides corresponding to amino acid sequences 747–762, 763–776 and 755–769 of hNCAM140-ID and hNCAM180-ID were then probed for their ability to bind to KLC1. Dose-dependent binding of NCAM-ID-747–762 to KLC1 was detectable, whereas NCAM-ID-763–776 and NCAM-ID-755–769 did not bind (Fig. 1E). Given that NCAM-ID-747–762 but not NCAM-ID-755–769 binds to KLC1, we conclude that the amino acid sequence which is unique in NCAM-ID-747–762 (encompassing amino acids 747–754, CGKAGPGA) is required for binding to KLC1. To confirm this, CHO cells were co-transfected with KLC1 and KHC (KIF5C), together with non-mutated hNCAM180 or hNCAM180 containing a deletion of the amino acid sequence responsible for binding to KLC1 (hNCAM180-Δ747–754). The interaction between KLC1 and hNCAM180 was then analyzed by co-immunoprecipitation and western blot analysis. Non-mutated hNCAM180 but not hNCAM180-Δ747–754 co-immunoprecipitated with KLC1 (Fig. 1F).

To identify the site of putative interactions between NCAM and KLC1 in cells, cultured mouse cortical neurons were colabeled with antibodies against KLC1 and antibodies recognizing the intracellular domains of NCAM140 and NCAM180. Immunofluorescence analysis showed that KLC1 and NCAM were distributed along neurites and in neuronal somata (Fig. 2A). Noticeably, confocal microscopy showed small intracellular accumulations of NCAM colocalizing with accumulations of KLC1, particularly in somata (Fig. 2A). Replacement of the primary antibodies with control non-immune immunoglobulins from the same species that the antibodies were derived from eliminated the signal (Fig. 2B).

To verify that NCAM and KLC1 form complexes, a proximity ligation assay (PLA) was performed (Söderberg et al., 2006). In this analysis, the primary antibodies are detected with secondary antibodies conjugated to oligonucleotides, which provide amplifiable DNA strands when they are in close proximity to each other (<40 nm). Amplified products can then be detected with fluorescence signals. Results showed that the intracellular domains of NCAM140 and NCAM180 are in close proximity to KLC1 (Fig. 2C). In agreement with the immunofluorescence data, proximity ligation signals were observed in somata, along neurites and in growth cones (Fig. 2C), where both proteins are enriched (Morfini et al., 2001; Sytnyk et al., 2002). Omission of one of the antibodies erased the proximity ligation signal (Fig. 2D). No proximity ligation signals were detectable in neurons colabeled with antibodies against KLC2 and antibodies recognizing the intracellular domains of NCAM140 and NCAM180 (Fig. 2E).

To confirm that NCAM and kinesin-1 colocalize intracellularly, colocalization of NCAM with kinesin-1 was analyzed in neurons overexpressing human NCAM180 (hNCAM180) with GFP-tagged KLC1 and KHC (KIF5C) (hereafter GFP–KHC/KLC1). Neurons were labeled with antibodies specific for the extracellular domain of human, but not mouse, NCAM, which were applied to the cells before membrane permeabilization with detergent to detect cell surface hNCAM180 only. After detergent permeabilization, the total pool of hNCAM180 was immunolabeled with a different fluorochrome. Confocal microscopy showed that the intracellular accumulations of hNCAM180 in somata colocalized with GFP-tagged KHC and KLC1 (Fig. 2F).

To identify the subcellular compartment, in which NCAM colocalizes with kinesin-1, neurons co-transfected with hNCAM180 and GFP–KHC/KLC1 were colabeled with antibodies against hNCAM and the Golgi marker protein TGN38 (also known as TGOLN2). Analysis of the confocal sections of somata showed that intracellular accumulations of NCAM partially colocalized with GFP–KHC/KLC1 and TGN-38, suggesting that they associate in the Golgi and trans-Golgi compartments (Fig. 3A). In agreement, western blot analysis showed that both NCAM and KLC1 were present in Golgi and trans-Golgi organelles isolated from early postnatal mouse brains using sucrose gradient centrifugation (Fig. 3B). Although NCAM was detectable at similar levels in the Golgi and trans-Golgi fractions, KLC1 and KIF5A were found at higher levels in the trans-Golgi fraction, which was enriched in TGN38 when compared to the Golgi, as described previously (Hossain et al., 2014). These data suggest that KLC1 contributes to the intracellular transport of NCAM predominantly in trans-Golgi organelles.

Given that NCAM has crucial functions at the cell surface, we investigated whether kinesin-1 is involved in cell surface delivery of NCAM. CHO cells were co-transfected with hNCAM180 and either GFP–KLC1, GFP–KHC/KLC1 or, for control, GFP alone. In agreement with our observations on neurons, confocal fluorescence analysis of CHO cells co-transfected with hNCAM180 and GFP–KHC/KLC1 and analyzed for cell surface and total pools of hNCAM180 showed that intracellular accumulations of NCAM colocalized with GFP–KHC/KLC1 (Fig. 4A). Cell surface localization of hNCAM180 was then assayed using antibodies against the extracellular domain of hNCAM applied to live cells on ice to inhibit endocytosis of the antibodies (Fig. 4B). Images of the labeled cells were acquired by fluorescence microscopy to quantify hNCAM180 immunofluorescence intensities associated with the surface membrane of transfected cells. This analysis showed that levels of hNCAM at the cell surface of GFP–KLC1 or GFP–KHC/KLC1 co-transfected cells were ∼twofold higher when compared to cell surface hNCAM levels of GFP co-transfected cells (Fig. 4C). Cell surface hNCAM levels were not different between CHO cells co-transfected with hNCAM and GFP–KLC1 or GFP–KHC/KLC1. Given that KIF5 (KIF5A, KIF5B and KIF5C) is expressed endogenously by CHO cells (Hammond et al., 2012) and can form a functional motor with transfected GFP–KLC1 (Araki et al., 2007), our results suggest that it is sufficient for transport of NCAM preceding cell surface delivery.

In individual cells, levels of cell surface hNCAM180 correlated with levels of expression of GFP–KLC1 and GFP–KHC/KLC1 as measured by GFP fluorescence, further indicating that an increase in the expression of kinesin-1 results in increased cell surface delivery of NCAM (Fig. 4B). In contrast, cell surface levels of NCAM180 poorly correlated with GFP levels in control GFP co-transfected cells (Fig. 4B).

To analyze whether the intracellular domain of NCAM is required for kinesin-1-dependent cell surface delivery of NCAM, CHO cells were transfected with a construct of hNCAM deleted in its intracellular domain (hNCAMΔCT) and co-transfected with GFP–KHC/KLC1 or GFP. Surprisingly, deletion of the intracellular domain of NCAM did not abolish cell surface delivery of NCAM. Levels of cell surface hNCAMΔCT were similar in CHO cells co-transfected with GFP–KHC/KLC1 or GFP (Fig. 4D) and similar to the levels of full-length hNCAM180 in CHO cells co-transfected with GFP–KHC/KLC1 (Fig. 4D). Levels of cell surface hNCAMΔCT correlated poorly with the expression levels of GFP–KHC/KLC1 (Fig. 4E). Thus, the intracellular domain of NCAM is required for kinesin-1-regulated cell surface delivery of hNCAM180, whereas deletion of this domain results in kinesin-1-independent delivery.

The intracellular domain of NCAM interacts with adaptor protein 2 (AP-2), which recruits cell surface cargo proteins to clathrin-coated endocytotic vesicles (Shetty et al., 2013). Hence, we compared endocytosis rates of hNCAM180 and hNCAMΔCT by analyzing the internalization of NCAM antibodies in live CHO cells, an assay used by us previously to analyze endocytosis of CHL1, another cell adhesion molecule of the immunoglobulin superfamily (Tian et al., 2012). Transfected CHO cells were incubated with antibodies recognizing the extracellular domain of hNCAM. Cell surface and total pools of NCAM–antibody complexes were then labeled with Cy3- and Cy5-coupled secondary antibodies applied before and after detergent permeabilization of cells, respectively. The efficiency of NCAM endocytosis was estimated using confocal microscopy by analyzing the ratio of intensities of Cy5 fluorescence in Cy3-negative areas, representing endocytosed NCAM, and Cy5 fluorescence in Cy3-positive areas, representing surface NCAM. This analysis showed that endocytosis of hNCAMΔCT is reduced when compared to endocytosis of hNCAM180 (Fig. 4F). Accumulation of hNCAMΔCT at the cell surface could therefore be due to inhibited endocytosis.

Next, we analyzed whether cell surface delivery of hNCAM180 is affected by inhibition of the interaction between its intracellular domain and kinesin-1 using fragments of the intracellular domain of NCAM. CHO cells were co-transfected with hNCAM180 and GFP–KHC/KLC1 together with cDNA constructs encoding the entire intracellular domain of rat NCAM140 (NCAM-ID), its fragments NCAM-ID-729–750, NCAM-ID-748–777 and NCAM-ID-777–810 derived from the membrane proximal region, or the NCAM180-specific amino acid sequence encoded by exon 18 (NCAM180-ID-808–1078) (Fig. 5A). Immunofluorescence analysis of the cell surface levels of hNCAM180 showed that levels of hNCAM180 at the cell surface of cells co-transfected with GFP-KHC/KLC1 together with NCAM-ID are reduced to the levels observed in GFP co-transfected cells, indicating that NCAM-ID interferes with kinesin-1-dependent cell surface delivery of hNCAM180, probably by competing with hNCAM180 for binding to KLC1 (Fig. 5B). A similar effect was observed for CHO cells co-transfected with NCAM-ID-748–777 (Fig. 5B), which also contains the binding site for KLC1 (Fig. 1E). This effect was not seen with cells co-transfected with NCAM180-ID-808–1078 (Fig. 5C). Interestingly, transfection with NCAM-ID-729–750 and NCAM-ID-777–810 also reduced cell surface hNCAM180 levels (Fig. 5B). Because these amino acid stretches contain binding sequences for components of the exocyst complex (Chernyshova et al., 2011), we analyzed whether overexpression of the exocyst complex Sec8 (also known as EXOC4) subunit mutated in its phosphorylation sites (Sec8-Y401F,Y616F) interferes with the cell surface delivery of NCAM. We had previously shown that this mutant interferes with exocyst complex assembly at surface membranes (Chernyshova et al., 2011). Kinesin-1-dependent cell surface delivery of hNCAM180 was inhibited in cells transfected with Sec8-Y401F,Y616F (Fig. 5C). Kinesin-1-dependent cell surface delivery of NCAM was also abolished in cells co-transfected with hNCAM180-Δ747–754 (Fig. 5C). We conclude that amino acids 747–754 and also the submembrane regions of NCAM involved in the interaction with the exocyst complex are required for kinesin-1-dependent cell surface delivery of NCAM.

Given that kinesin-1 is required for intracellular trafficking, it is tempting to speculate that the interaction between NCAM and KLC1 is disrupted when NCAM is delivered to the cell surface. Interestingly, the amino acid sequence in NCAM-ID that is required for its binding to KLC1 overlaps with a recently identified binding site for p21-activated protein kinase (PAK1), which interacts with NCAM at the cell surface (Li et al., 2013). This observation prompted us to analyze whether PAK1 and KLC1 compete with each other for binding to NCAM-ID. To test this idea, we performed a pulldown assay with KLC1 immunopurified from the mouse brain and peptides from NCAM-ID. Beads with immobilized KLC1 were incubated with chemically synthesized NCAM-ID-747–762, NCAM-ID-763–776, and NCAM-ID-755–769. Slot blot analysis of the pulldowns showed that only NCAM-ID-747–762 bound to KLC1 (Fig. 6A), confirming the ELISA results (Fig. 1E). The binding of NCAM-ID-747–762 to KLC1 in the presence of immunopurified PAK1 was reduced to 7.0±0.3% (mean±s.e.m.) of the signal observed in the absence of PAK1, verifying that KLC1 and PAK1 compete for the binding to NCAM-ID-747–762.

To analyze whether PAK1 plays a role in the cell surface delivery of NCAM, hNCAM180 and GFP–KHC/KLC1 were co-transfected into CHO cells together with either full-length PAK1 where T423 in the catalytic domain of PAK1 was replaced with glutamic acid resulting in a constitutively active enzyme (CA-PAK1) (Sells et al., 1997) or with a truncated form of PAK1 consisting of amino acid residues 67–150, representing the autoinhibitory region of mouse PAK1 (DN-PAK1) (Hayashi et al., 2002). Immunofluorescence analysis of the transfected cells showed that cell surface levels of hNCAM were reduced in cells co-transfected with either of the two PAK1 forms (Fig. 6B). We conclude that overexpressed PAK1, and particularly its autoinhibitory domain, interferes with cell surface delivery of NCAM possibly by competing with KLC1 for binding to NCAM.

To investigate whether KLC1 promotes cell surface delivery of NCAM in neurons, cortical neurons were transfected with hNCAM180 together with GFP, GFP–KLC1 or GFP–KHC/KLC1. Surface hNCAM180 was detected with antibodies recognizing the extracellular domain of human but not mouse NCAM applied to live neurons and visualized with secondary antibodies coupled to the Cy3 fluorochrome. Neurons were then colabeled with antibodies specific for hNCAM180, which were applied to neurons after permeabilization of membranes with detergent, and visualized with the Cy5 fluorochrome to detect cell surface and internal pools of hNCAM180. The efficiency of hNCAM180 delivery to the cell surface of neuronal somata was then estimated using confocal microscopy and analyzing the ratio of surface NCAM levels (NCAM_surf, identified as Cy5 fluorescence intensity in Cy3-positive areas) and internal NCAM levels (NCAM_intern, identified as Cy5 fluorescence intensity in Cy3-negative areas). This analysis showed that the NCAM_surf:NCAM_intern ratio is higher in neurons co-transfected with GFP–KHC/KLC1 when compared to neurons co-transfected with GFP only (Fig. 7A,B), indicating increased transport of hNCAM180 from the internal pool to the cell surface. Overexpression of GFP–KLC1 alone resulted in only a slight increase in this ratio (Fig. 7A,B), probably due to the high expression of endogenous KLC1 in neurons.

To confirm that the interaction between NCAM and KLC1 is required for kinesin-1-dependent cell surface delivery of NCAM, the NCAM_surf:NCAM_intern ratio was analyzed in neurons transfected with hNCAM180-Δ747–754. Cell surface delivery of hNCAM180-Δ747–754 in GFP-transfected neurons was reduced when compared to that of non-mutated hNCAM180 (Fig. 7A,B) and remained unchanged, when GFP–KLC1 or GFP–KHC/KLC1 were overexpressed (Fig. 7A,B). To corroborate these data, we also analyzed whether knockdown of KLC1 expression affected hNCAM180 targeting from the Golgi complex to the cell surface. In neurons transfected with KLC1 small interfering RNA (siRNA), cell surface delivery of hNCAM180 was reduced to the level observed for hNCAM180-Δ747–754 in neurons co-transfected with control siRNA (Fig. 7C,D). Cell surface targeting of hNCAM180 was not affected by KLC2 siRNA (Fig. 7C,D). Our observations thus indicate that disruption of the interaction of NCAM180 with endogenous KLC1 reduces cell surface delivery of hNCAM180.

In accordance with our observations in CHO cells (Fig. 4D), complete deletion of the intracellular domain resulted in increased levels of hNCAMΔCT at the cell surface as reflected by the increased NCAM_surf:NCAM_intern ratio (Fig. 7A,B). Cell surface delivery of hNCAMΔCT was not affected by co-overexpression of GFP–KLC1 or GFP–KHC/KLC1 (Fig. 7A,B), indicating that cell surface insertion of hNCAMΔCT is independent of kinesin-1. To analyze whether endocytosis of hNCAMΔCT is affected, endocytosis rates were estimated by measuring the internalization of NCAM antibodies. Neurons transfected with hNCAM constructs and maintained for 2 days in culture were incubated with antibodies recognizing the extracellular domain of human but not mouse NCAM, and cell surface as well as total pools of NCAM–antibody complexes were visualized using secondary antibodies conjugated to different fluorochromes, applied before and after membrane permeabilization, respectively. The ratio of endocytosed and cell surface pools of NCAM in somata showed that internalization of hNCAMΔCT was slightly reduced when compared to endocytosis of hNCAM180 and hNCAM180-Δ747–754 (Fig. 8). Interestingly, internalization of hNCAM180 was also slightly reduced by co-transfection with GFP–KHC/KLC1 (Fig. 8).

Kinesin-1 is a motor protein enabling fast transport of a variety of cargoes along microtubules of neurites (Brady, 1985; Hirokawa et al., 1991; Kamal et al., 2000; Konecna et al., 2006; Lazarov et al., 2005; Vale et al., 1985). Kinesin-1 is known to carry a number of neuronal transmembrane proteins, including apolipoprotein E receptor 2 (ApoER2, also known as LRP8) (Verhey et al., 2001), amyloid precursor protein (Kamal et al., 2000; Lazarov et al., 2005), and calsyntenin-1 (also known as alcadein) (Araki et al., 2007; Konecna et al., 2006), which can associate directly or indirectly with KLC1 (Hirokawa et al., 2009). In this study, we identified KLC1 as a new binding partner of the intracellular domain of NCAM140 and NCAM180. We also show that kinesin-1 increases cell surface delivery of NCAM180 both in CHO cells and neurons. Our observation that intracellular accumulations of NCAM colocalize with kinesin-1 in the Golgi complex suggests that kinesin-1 promotes trafficking of NCAM-containing organelles from the Golgi compartment to the cell surface.

KLC1 is the main binding site for cargoes in kinesin-1 (Hirokawa, 1998; Hirokawa et al., 1989). Using macroarray screening, we demonstrated that KLC1 binds directly to hNCAM180-ID and hNCAM140-ID. Confirming the physiological role of this association, NCAM could be co-immunoprecipitated with KLC1 from mouse brain tissue lysates. The KLC1-binding motif within the intracellular domain of NCAM140 and NCAM180 comprises the amino acid sequence CGKAGPGA and does not contain tyrosine or tryptophan residues such as have been shown to be crucial for the interaction between KLC1 and other cargoes (Aoyama et al., 2009; Araki et al., 2007; Bracale et al., 2007; Dodding et al., 2011; Hayakawa et al., 2007; Konecna et al., 2006; Rosa-Ferreira and Munro, 2011; Schmidt et al., 2009; Verhey et al., 2001). Therefore, the identified sequence in NCAM represents a new binding motif for KLC1.

Interestingly, the complete deletion of the intracellular domain of NCAM does not abolish cell surface delivery, although it results in elimination of the binding sites for KLC1 and other intracellular binding partners, such as PAK1. This observation is not surprising, given that NCAMΔCT could be considered to be a functional and structural equivalent of the NCAM120 isoform of NCAM, which is anchored to the membrane through GPI and lacks an intracellular domain and, hence, is transported by mechanisms distinct from those of NCAM180 and NCAM140 (Garner et al., 1986; Nybroe et al., 1986). Our observation that cell surface levels of NCAMΔCT are similar to or even higher than those for NCAM180 indicates that KLC1 is not required for cell surface delivery of NCAM, but rather suggests that this interaction plays a role in regulation of NCAM transport. In agreement, cell surface delivery of NCAMΔCT is independent of kinesin-1 in our experiments, indicating that NCAMΔCT is delivered to the cell surface through another, kinesin-1-independent, pathway. Reduced endocytosis also contributes to the accumulation of NCAMΔCT at the cell surface, probably because the intracellular domains of NCAM140 and NCAM180 are required for their interaction with AP-2 (Shetty et al., 2013), which plays a role in the recruitment of the proteins to the clathrin-coated endocytic vesicles.

Somewhat unexpectedly, overexpression of kinesin-1 also reduced endocytosis of NCAM180. NCAM180 associates with the membrane–cytoskeleton linker protein spectrin (Pollerberg et al., 1986, 1987; Leshchyns'ka et al., 2003) and promotes polymerization of the spectrin cytoskeleton beneath the cell surface membrane (Leshchyns'ka et al., 2003; Sytnyk et al., 2006). The spectrin cytoskeleton inhibits endocytosis of transmembrane proteins by interfering with clathrin-coated pit formation (Puchkov et al., 2011; Kamal et al., 1998). It is therefore conceivable that kinesin-1-dependent cell surface delivery of NCAM180 increases the assembly of the spectrin cytoskeleton, which could interfere with the endocytosis of NCAM180. We also cannot exclude that other factors, such as endosomal transport, could be affected by kinesin-1.

Confirming the role of KLC1 in cell surface delivery of NCAM180, the dominant-negative fragment of the intracellular domain of NCAM140 and NCAM180, comprising the KLC1-binding site, reduced cell surface levels of NCAM180. Interestingly, two other fragments derived from the intracellular domain of NCAM also inhibited its delivery. These fragments, comprising amino acids 729–750 and 777–810, interact with the exocyst subunits Exo70 (also known as EXOC7) and Sec8, respectively (Chernyshova et al., 2011), which are essential for exocytosis of membranous organelles (Hsu et al., 2004). On the basis of these results it is conceivable that NCAM-ID-729–750 and NCAM-ID-777–810 inhibit cell surface delivery of NCAM180 by inhibiting the binding of the exocyst subunits to NCAM and exocytosis of NCAM-containing vesicles.

The role of PAK1 is interesting in the context of cell surface delivery of NCAM isoforms. PAK1 interacts with NCAM at the cell surface of growing neurites and growth cones (Li et al., 2013). It binds to the amino acid sequence in the intracellular domain of NCAM, which we now show also contains the binding site for KLC1 (Li et al., 2013). Indeed, PAK1 and KLC1 compete for the binding to NCAM. PAK1 contains an autoinhibitory domain that binds to the kinase domain of PAK1 to inactivate the enzyme. Although overexpression of constitutively active PAK1 reduces cell surface delivery of NCAM, suggesting that activated PAK1 inhibits this delivery, we consider this possibility unlikely because delivery is also reduced in cells overexpressing the autoinhibitory domain of PAK1, which inhibits endogenous PAK1 activity. These observations rather suggest that the autoinhibitory domain of PAK1, which is exposed in its active form enabling its interaction with other proteins (Say et al., 2010), regulates delivery of NCAM. We thus propose that this domain competes with KLC1 for binding to NCAM and contributes to the detachment of NCAM from KLC1 upon delivery of NCAM to the cell surface. In agreement with this idea, activation of NCAM functions induces activation of PAK1 (Li et al., 2013) and exocytosis in growth cones (Chernyshova et al., 2011). Interestingly, the proposed model also suggests that activation of NCAM by homophilic and heterophilic interactions at sites of contacts between cells can promote exocytosis of NCAM-containing vesicles to stabilize these contacts. A recent study shows that presynaptically expressed NCAM is required for the maintenance of the neuromuscular junction innervation field, and it is tempting to speculate that cell surface delivery of NCAM by kinesin-1 is important for stabilizing the neuromuscular junction (Chipman et al., 2014).

The functions of NCAM are further regulated by post-translational modifications, such as polysialylation (Gascon et al., 2007, 2010; Kleene and Schachner, 2004; Rønn et al., 2000; Seki and Rutishauser, 1998), and phosphorylation and ubiquitylation (Cassens et al., 2010; Diestel et al., 2004, 2007; Pollscheit et al., 2012; Sorkin et al., 1984; Wobst et al., 2012). Whether posttranslational modifications affect transport of NCAM by kinesin-1 is an intriguing question, which will have to be followed up by future studies.

In summary, we reveal a new role for kinesin-1 together with PAK1 in regulating the cell surface expression of the two transmembrane isoforms of NCAM. This new role sheds light on a different level of regulation of cell surface expression of NCAM at the strategic site to allow it influence cell-to-cell interactions in the developing and adult brain, particularly in view of its implications in nervous system disorders.

Rabbit polyclonal antibodies (pAb) against mouse NCAM (Westphal et al., 2010) (for western blot analysis and immunoprecipitation) and mouse monoclonal antibody (mAb) ERIC recognizing human, but not mouse, NCAM (for western blot and immunocytochemical experiments; Santa Cruz Biotechnology) react with the extracellular domain of all NCAM isoforms. Mouse mAb 5B8 against the intracellular domain of mouse NCAM140 and NCAM180 (for immunocytochemistry and PLA) was from the Developmental Studies Hybridoma Bank (University of Iowa, Iowa). Rabbit pAb against KLC1 (H-75, for immunoprecipitation, western blotting, immunocytochemistry and PLA), goat pAb against KLC2 (N-14, for PLA), rabbit pAb against TGN38 (M-290, for immunocytochemistry and western blotting), rabbit pAb against KIF5A (for western blotting) and non-immune immunoglobulins were from Santa Cruz Biotechnology. Mouse mAb against EEA1 (for western blotting) and mouse mAb against PDI (for western blotting) were from BD Biosciences. Secondary antibodies coupled to horseradish peroxidase (HRP), Cy2, Cy3 or Cy5 were from Jackson ImmunoResearch Laboratories.

C57BL/6J mice (1 to 3 days old) were used for preparation of neuronal cultures and biochemical experiments. All experiments were approved by the Animal Care and Ethics Committee of the University of New South Wales (permit 12/135B).

The cDNAs encoding the intracellular domain of human NCAM180 (hNCAM180-ID) and NCAM140 (hNCAM140-ID) used to produce recombinant proteins for the protein macroarray assay have been described (Homrich et al., 2014; Wobst et al., 2012).

For transfection of cells, cDNAs coding for full length non-mutated hNCAM180 (Diestel et al., 2004); hNCAMΔCT (Boutin et al., 2009); rat NCAM140-ID (Leshchyns'ka et al., 2003); rat NCAM-ID-729–750, rat NCAM-ID-777–810, and Sec8Y401F,Y616F (Chernyshova et al., 2011); rat NCAM-ID-748–777 (Li et al., 2013); GFP–KLC1 and GFP–KHC1 (generous gifts of Toshiharu Suzuki, Hokkaido University, Sapporo, Japan) (Araki et al., 2007; Kawano et al., 2012), and GFP (Clontech) were used. To construct cDNA coding for hNCAM180-Δ747–754, site-specific mutagenesis was carried out using the overlap extension PCR method (Ho et al., 1989) with hNCAM180 as a template. The following sense primers 5′-CGCGGATCCATGCTGCAAACTAAGGATCTC-3′ (pI) and 5′-GCGGTCAACCTG- - - - -AAGGGCAAAGAC-3′ (pIII) and antisense primers 5′-GTCTTTGCCCTT- - - - -CAGGTTGACCGC-3′ (pII) and 5′-CGGGATCCCGCATGCTCGAGTCATGCTTTGCTCTC-3′ (pIV) were used (dashed lines indicate regions of deleted amino acids). The mutant construct was checked for correctness by DNA sequencing (MWG Biotech, Ebersbach, Germany). NCAM180-specific exon18-coded sequence was amplified from rat NCAM180 cDNA (Leshchyns'ka et al., 2003) using the following primers: 5′-CTGCCTGCCGACACCACAGCCA-3′, and 5′-TTACTCGGTCTTTGCTGGTGCGGGC-3′, and was cloned into the pcDNA3 vector.

Control siRNA and siRNA for knockdown of KLC1 (sc-43881) and KLC2 (sc-146492) expression was from Santa Cruz Biotechnology.

hNCAM180-ID and hNCAM140-ID were expressed in BL21 (DE3) bacteria and purified by Nickel-nitrilotriacetic acid affinity chromatography (hNCAM180-ID) or by glutathione affinity chromatography (hNCAM140-ID) according to manufacturer's instructions (Qiagen) as described previously (Homrich et al., 2014; Wobst et al., 2012). Purified proteins were concentrated to a minimum of 1 mg/ml in phosphate-buffered saline (PBS) and fluorescently labeled with Dyomics DY-633 Fluorophore (Dyomics, Jena, Germany).

The macroarray membrane (Source Bioscience imaGenes, Berlin, Germany) containing 24,000 different His-tagged expression clones of human fetal brain was prepared as described previously (Homrich et al., 2014; Wobst et al., 2012). Purified and fluorescently labeled hNCAM180-ID (15 nM) and NCAM140-ID (13.5 nM), respectively, were incubated in Odyssey blocking buffer (LI-COR Biosciences) with the prepared membrane for 16 h. Signals were detected using the Licor Odyssey scanner (LI-COR Biosciences). Images were analyzed with Aida Image Analyzer version 4.24 (Raytest, Straubenhardt, Germany). Positive spots were identified using the Scoring Template and the annotation table of Source Bioscience imaGenes revealing the spotting pattern of the expression clones on the membrane (Homrich et al., 2014; Wobst et al., 2012).

Brains of 1-day-old mice were homogenized using a Potter homogenizer in buffer containing 5 mM Tris-HCl pH 7.4, 1 mM MgCl_2, 1 mM CaCl₂, 0.1 mM PMSF, EDTA-free protease inhibitor cocktail (Roche) and 0.32 M sucrose. Homogenates containing 1 mg of total protein were lysed for 30 min at 4°C with lysis buffer containing 50 mM Tris-HCl pH 7.5, 150 mM NaCl, 1% Triton X-100, 1 mM Na₄P₂O₇, 1 mM NaF, 2 mM Na₃VO₄, 0.1 mM PMSF and EDTA-free protease inhibitor cocktail. Lysates were centrifuged for 15 min at 20,000 g at 4°C. Supernatants were cleared with protein-A/G–agarose beads (Santa Cruz Biotechnology) for 3 h at 4°C and incubated with antibodies against KLC1, NCAM or non-immune IgG overnight at 4°C. Complexes were collected with protein-A/G-agarose beads for 3 h at 4°C. Beads were washed twice with lysis buffer and once with Tris-buffered saline (TBS), pH 7.4. Complexes were eluted from beads with non-reducing LDS sample buffer (Life Technologies) to detect KLC1 by western blot analysis. For detection of NCAM, 150 mM DTT was added to the samples. Samples were heated for 10 min at 70°C followed by the SDS-PAGE electrophoresis and western blot analysis.

Isolation of the cytoplasmic fraction, Golgi and trans-Golgi network (TGN) organelles was performed as described previously (Fath and Burgess, 1993; Sytnyk et al., 2002). Brains from 1- to 3-day-old mice were homogenized using a T10 homogenizer (IKA) in PKM buffer containing 100 mM KH₂PO₄ pH 6.5, 5 mM MgCl₂, 3 mM KCl, 0.1 mM PMSF, EDTA-free protease inhibitor cocktail and 0.5 M sucrose. Brain homogenates were centrifuged at 600 g for 10 min at 4°C. The supernatant was collected and centrifuged on a discontinuous density gradient of 0.7 and 1.3 M sucrose in PKM buffer (100 mM KH₂PO₄ pH 6.5, 5 mM MgCl₂, 3 mM KCl, 0.1 mM PMSF, EDTA-free protease inhibitor cocktail) at 105,000 g for 60 min at 4°C. The upper layer of the gradient was collected and centrifuged at 200,000 g for 45 min at 4°C. The supernatant containing the cytoplasmic protein fraction was used as source of endogenous KLC1 for ELISA and pulldown assays. The pellet was positive for EEA1 indicating that it contains endosomes. Membranous organelles were collected at the 0.7–1.3 M sucrose-PKM interface, adjusted to 1.25 M sucrose in PKM, overlaid with 1.1 M and 0.5 M sucrose in PKM and centrifuged at 90,000 g for 90 min at 4°C. Membranes collected at 1.25–1.1 M sucrose interface were enriched in PDI indicating that they contain endoplasmic reticulum (ER). Golgi and TGN organelles were collected at the 0.5–1.1 M sucrose interface, adjusted to 0.7 M sucrose in PKM and centrifuged for 15 min at 10,000 g at 4°C. The pellet containing Golgi stacks was collected. The supernatant containing TGN organelles was centrifuged at 259,000 g for 30 min at 4°C, and the pellet containing TGN organelles was collected. Organelles were resuspended in PEMS buffer containing 10 mM Pipes pH 7.0, 1 mM EGTA, 2 mM MgCl₂, 0.25 M sucrose, 0.1 mM PMSF and EDTA-free protease inhibitor cocktail.

Proteins were separated by 10% SDS-PAGE (Life Technologies) and electroblotted onto PVDF membranes (Millipore). Membranes were blocked with 5% skim milk powder in phosphate-buffered saline (PBS), pH 7.4, and incubated with appropriate primary antibodies followed by incubation with HRP-labeled secondary antibodies, which were visualized using Western Lightning Ultra reagent (PerkinElmer) or Luminata Crescendo Western HRP substrate (Millipore). Chemiluminescence images were obtained using the MicroChemi 4.2 imaging system (DNR Bio-Imaging Systems).

Rabbit pAb against KLC1 (1 µg/ml in Na₂CO₃ buffer, pH 9.6) were absorbed to the surface of 96-well MICROLON 600 plates (Greiner) overnight at 4°C. Wells were blocked with 1% BSA in TBS for 1 h at 37°C and incubated with the cytoplasmic fraction from mouse brain diluted in TBS with 0.1% Tween 20 (TBST) to purify and immobilize KLC1. Wells were washed with TBST and incubated for 1 h at 37°C with increasing concentrations of chemically synthesized biotinylated peptides corresponding to amino acid sequences 747–762, 763–776 and 755–769 of human NCAM140 and NCAM180 (Peptide 2.0) diluted in TBST. Wells were washed with TBST, incubated for 1 h at 37°C with HRP-coupled neutrAvidin (Thermo Fisher Scientific) in TBST containing 1% BSA, and washed with TBST. The color reaction was developed with 1 mg/ml o-phenylenediamine (Sigma). Absorbance was measured at 450 nm.

PAK1 was immunopurified from brain lysates (as described in the co-immunoprecipitation section) using agarose beads conjugated to rabbit pAb against PAK1 (Santa Cruz Biotechnology) as described previously (Li et al., 2013). PAK1 was eluted from the beads with 0.2 M glycine pH 4.0. The eluate containing PAK1 was immediately neutralized with 1 M Tris-HCl pH 8.0.

For immunopurification and immobilization of KLC1, the cytoplasmic protein fraction from mouse brain was cleared with protein-A–agarose beads (Santa Cruz Biotechnology) for 5 h at 4°C and incubated with rabbit pAb against KLC1 applied overnight at 4°C. KLC1 was precipitated with protein-A–agarose beads applied for 5 h at 4°C. Beads were washed five times with TBST and incubated with 2.86 nmol of chemically synthesized biotinylated peptides corresponding to amino acids sequences 747–762, 763–776 or 755–769 of human NCAM140 (Peptide 2.0) in TBST. Purified PAK1 (12.8 nmol) was added when indicated. Samples were incubated overnight at 4°C. Supernatants containing non-bound peptides were collected, and beads were washed four times with TBST. Peptides bound to the beads were eluted with 0.2 M glycine, pH 2.0, and immediately treated with 1 M Tris for adjustment to pH 7.4. Samples containing input, bound and non-bound peptides were immobilized on nitrocellulose membranes using a slot blot apparatus. Peptides were detected with HRP-coupled neutrAvidin.

Cultures of cortical neurons were prepared as described previously (Andreyeva et al., 2010). Neurons were obtained from 2-day-old mice and maintained in Neurobasal A medium supplemented with 2% B-27, glutaMAX and bFGF (2 ng/ml) (all from Life Technologies) on glass coverslips coated with poly-D-lysine (100 μg/ml). Neurons were transfected before plating using a Neon transfection system (Life Technologies). Neurons were analyzed 48 h after plating.

CHO cells were cultured in Dulbecco's modified Eagle's medium with Ham's F-12 (PAA Laboratories) containing 10% fetal bovine serum (Sigma) and antibiotics (Gentamicin/Amphotericin B, Life Technologies). At 1 day before transfection, cells were plated into 2-cm² wells on glass coverslips in culture medium without antibiotics. CHO cells were transfected with Lipofectamine 2000 (Life Technologies).

When indicated, cell surface hNCAM180 in transfected neurons was labeled by incubating live neurons with a mouse monoclonal antibody against the extracellular domain of human NCAM (ERIC) applied for 1 h at 37°C in a CO₂ incubator. Unless indicated otherwise, other steps were performed at room temperature. Neurons were fixed with 4% formaldehyde in PBS for 15 min, washed with PBS, blocked with 5% donkey serum in PBS for 15 min and labeled with anti-mouse-IgG Cy3-conjugated secondary antibodies applied in 1% donkey serum in PBS for 30 min. To visualize total hNCAM180 and other proteins, neurons were post-fixed with 2% formaldehyde in PBS for 5 min, washed with PBS, permeabilized with 0.25% Triton X-100 in PBS for 5 min, blocked with 5% donkey serum in PBS for 20 min, incubated with ERIC antibodies and antibodies against other antigens applied in 1% donkey serum in PBS overnight at 4°C, and labeled with Cy5-conjugated (for total hNCAM180) and Cy2-conjugated (for other antigens) secondary antibodies applied in 1% donkey serum in PBS for 45 min. Targeting of hNCAM180 to the cell surface was analyzed using confocal slice images of the somata of neurons. To calculate the NCAM_surf:NCAM_int ratio, cell surface and internal pools of hNCAM180 were estimated by measuring immunofluorescence intensities of Cy5 labeling in manually outlined Cy3-negative intracellular compartments (NCAM_int) and at the Cy3-positive cell surface membrane regions (NCAM_surf) using ImageJ (National Institute of Health, Bethesda, MD).

All antibodies were applied in 0.1% BSA in PBS. To label cell surface NCAM, CHO cells were placed on ice and incubated with ERIC antibodies applied for 20 min, followed by corresponding secondary antibodies applied for 15 min. CHO cells were then fixed with 4% formaldehyde in PBS for 15 min at room temperature and washed with PBS. When indicated, cells were also colabeled for the total hNCAM180. Cells were permeabilized with 0.25% Triton X-100 in PBS for 5 min, blocked with 1% BSA in PBS for 20 min, incubated with ERIC antibodies applied overnight at 4°C, and labeled with secondary antibodies for 45 min at room temperature. Levels of GFP fluorescence and cell surface hNCAM immunofluorescence were measured using ImageJ by manually outlining cells.

Endocytosis of hNCAM180 was analyzed by estimating NCAM antibody internalization from the cell culture medium as described previously (Puchkov et al., 2011; Tian et al., 2012). Transfected cells were incubated with ERIC antibody applied for 1 h at 37°C in a CO₂ incubator. Unless indicated otherwise, all further steps were performed at room temperature. Cells were fixed with 4% formaldehyde in PBS for 15 min, washed with PBS, blocked with 5% donkey serum in PBS for 15 min and labeled with Cy3-conjugated secondary antibodies applied in 1% donkey serum in PBS for 30 min to visualize cell surface NCAM. Cells were then post-fixed with 2% formaldehyde in PBS for 5 min, washed with PBS, permeabilized with 0.25% Triton X-100 in PBS for 5 min, blocked with 5% donkey serum in PBS for 20 min and incubated with Cy5-conjugated secondary antibodies applied in 1% donkey serum in PBS for 45 min to visualize cell surface and internalized pools of NCAM antibodies. NCAM antibody internalization was analyzed using confocal slice images of CHO cells and somata of neurons. Endocytosed NCAM and NCAM–antibody complexes were identified as Cy3-negative and Cy5-positive accumulations. To calculate the NCAM_end:NCAM_surf ratio, immunofluorescence intensities of Cy5 labeling were measured in manually outlined Cy3-negative intracellular compartments (NCAM_end) and at the Cy3-positive cell surface membrane regions (NCAM_surf) using ImageJ.

The PLA was performed as described previously (Li et al., 2013; Tian et al., 2012). Cultured neurons were fixed in 4% formaldehyde in PBS, washed with PBS, permeabilized with 0.25% Triton X-100 in PBS for 5 min, blocked with 1% BSA in PBS for 20 min, and incubated with antibodies against the intracellular domain of NCAM and KLC1 or KLC2 applied in 0.1% BSA in PBS overnight at 4°C. Further steps were performed using secondary antibodies conjugated with oligonucleotides (PLA probes, Olink Bioscience, Uppsala, Sweden) and Duolink II fluorescence kit (Olink Bioscience) in accordance with the manufacturer's instructions.

Coverslips were embedded in FluorPreserve reagent (Calbiochem). Immunofluorescence images were acquired using a confocal laser scanning microscope C1si, NIS Elements software, and oil Plan Apo VC 60× objective (numerical aperture 1.4), all from Nikon Corporation (Tokyo, Japan). Immunofluorescence intensity measurements were performed in ImageJ software. Immunofluorescence intensities were measured in arbitrary units defined as pixel values of 16-bit grayscale images.

All experiments were independently performed three to five times. Excel and GraphPad Prism 6 were used for statistical analyses. Differences between groups were analyzed using one-way ANOVA with Dunnett's multiple comparisons test.

We are grateful to Prof. Toshiharu Suzuki (Hokkaido University, Japan) for GFP–KLC1 and GFP–KHC1 constructs, to Dr Ohshima (Laboratory for Developmental Neurobiology, Brain Science Institute, Japan) for providing Pak1 constructs, and to Prof. Jörg Höhfeld and Dr Michael Dreiseidler for help with the protein macroarray.