Densin-180: revised membrane topology, domain structure and phosphorylation status

Abstract

Densin-180 is a core component of post-synaptic densities, the highly complex molecular assemblies that mediate signaling between neuronal cells. It is a multi-domain scaffold protein characterized by multiple leucine-rich repeat domains plus a single Psd95/Discs large/Zona occludens-1 domain. In its original topology model a single transmembrane segment was proposed with an extracellular N-terminus and an intracellular C-terminus. However, recently discovered in vivo phosphorylation sites are incompatible with this topology. Here, we discuss an all-intracellular and membrane-associated localization of Densin-180 that is consistent with and supported by all the latest experimental data. This revised topology which now includes also a phosphorylation-rich area will have deciding influence on future research involving Densin-180 and its signaling.

Densin-180: a key synaptic component

Densin-180, also known as Leucine-rich repeat-containing protein 7 (Lrrc7) or Protein Lap1, was first purified from rat post-synaptic densities (PSD) preparations, where it is highly enriched (Apperson et al. 1996) (Note: a unified nomenclature for proteins across species has been used here, with only the first letter in upper case). It was characterized as a 188 kDa, 1495 residue long, brain-specific protein containing 16 leucine-rich repeats (LRRs) within the 500 N-terminal residues, and one Psd95/Discs large/Zona occludens-1 (PDZ) domain within the 200 C-terminal residues (Fig. 1a).

Fig. 1.

(a) Originally proposed structure of Densin-180. In homology to the domain structure of GpIbα, Densin-180 was proposed as a transmembrane domain (TMD) protein with a large extracellular area containing the leucine-rich repeats (LRRs) domain and a Mucin-homolog domain, while the intracellular part features a PDZ (Densin-180) and an Abp-binding domain (Abp; GpIbα), respectively (Apperson et al. 1996). (b) In vivo phosphorylation sites on Densin-180. Clusters of Ser/Thr phosphorylation (black) and two sites of Tyr phosphorylation (grey) (http://www.phosphosite.org; Trinidad et al. 2006; Munton et al. 2007; Rikova et al. 2007; Trinidad et al. 2008) were found N-terminal of the originally proposed single transmembrane domain. These phosphorylation sites are incompatible with the the original topology shown in (a). Numbering is for murine sequence uniprot Q80TE7-1. (c) Summary of domains on Densin-180. The hallmark LRRs, PDZ and LAP-specific domains domains at the N and C-terminal segments place Densin-180 in the LAP protein family (Santoni et al. 2002), which otherwise consists only out of all-intracellular members. The N-terminal palmitoylation (Izawa et al. 2008) positioned in analogy to Erbin (zig-zag line; anchors the N-terminus of Densin-180 to the membrane). The region containing clusters of phosphorylation shown in (b) is highlighted in blue. The cytoplasmic binding areas of rat Densin-180 for CamkIIα and α-Actinin are indicated, as are the areas of human Densin-180 binding to β-Catenin, δ-Catenin, Maguin- 1, and Shank (Strack et al. 2000; Walikonis et al. 2001; Izawa et al. 2002; Ohtakara et al. 2002; Quitsch et al. 2005; Robison et al. 2005; Heikkila et al. 2007). (d) Revised topology for Densin-180. All-intracellular and membrane-anchored topology of Densin-180.

It has been shown that the C-terminal segment of Densin-180 interacts with key players of synaptic signaling namely α-Actinin, CamkIIα (Strack et al. 2000; Walikonis et al. 2001; Robison et al. 2005), Shank (Quitsch et al. 2005), Maguin-1 (Ohtakara et al. 2002), β-Catenin (Heikkila et al. 2007), and δ-Catenin (Izawa et al. 2002).

In particular, Densin-180 has the ability to position CamkIIα at the post-synaptic membrane (Strack et al. 2000) in close proximity of the calcium influx that results from synaptic stimulation. This interaction is regulated by the phosphorylation state of the two interacting proteins. It has been observed that autophosphorylation of CamkIIα (at Thr286) increased, and phosphorylation of Densin-180 slightly decreased, if at all, the binding affinity between these proteins (Strack et al. 2000; Walikonis et al. 2001). Densin-180 binding to CamkIIα does not compete with the well-characterized interaction of CamkIIα with the NMDA receptor subunit Nr2B (Strack et al. 2000). The latter is dependent on CamkIIα activity (Bayer et al. 2001).

Densin-180 also promotes dendritic branching in culture systems, a function that is negatively regulated by binding of its C-terminus to Shank (Quitsch et al. 2005). The reported interaction with β- and δ-Catenin, which in turn bind to Cadherins, highlights a potential role for Densin-180 in cell adhesion (Izawa et al. 2002; Heikkila et al. 2007). Moreover, Densin-180 is linked to Psd95 via Maguin-1 (Ohtakara et al. 2002).

In summary, all reported interactions of the C-terminal segment of Densin-180 occur with well-established cytoplasmic proteins thus supporting a cytoplasmic location of this region of the protein (Fig. 1c).

Transmembrane topology for Densin-180?

The N-terminal segments of Densin-180 have received far less attention. Apperson et al. identified an LRRs-domain in this moiety of the protein (Apperson et al. 1996) that matched well with the consensus of a variety of LRR proteins known at that time (Kobe and Deisenhofer 1995). Most of those LRR proteins contain either a single transmembrane domain (TMD) or are secreted proteins, consistent with an extracellular localization of the LRR-domains. In 1996 a transmembrane topology that contains also an extracellular Mucin-like domain was proposed for Densin-180 by analogy to the domain-line up of GpIbα (Apperson et al. 1996), (Fig. 1a). This proposed transmembrane topology has been widely adopted in numerous reviews, albeit with caution in some instances, e.g. (Colbran 2004; Kim and Sheng 2004).

In vivo phosphorylation data disagree with the prevailing transmembrane topology model of Densin-180

Recent mass-spectrometric analysis of the in vivo phosphorylation state of Densin-180 from tryptic digests of murine PSD preparations revealed a surprising number of phosphorylated serine and threonine residues (Trinidad et al. 2006, 2008; Munton et al. 2007) (Fig. 1b, Table S1). Except for Ser1392, all of them are located N-terminal of the proposed TMD. Further, analysis of human lung cancer tissue identified two sites of tyrosine phosphorylation within the 826–1213 cluster (http://www.phosphosite.org; Rikova et al. 2007). These findings are at odds with the widely accepted topology as shown in Fig. 1(a), as these phosphorylated residues would reside on the extracellular side of Densin-180.

The experimental evidence supporting extensive phosphorylation in the region of residues 826–1213 is extremely strong. The identification of these phosphorylation sites was accomplished by tandem mass spectrometry using stringent statistical interpretation criteria, combined with manual inspection of the individual spectra. In many cases multiple identifications of the same phosphorylation site was obtained from overlapping tryptic peptides, which were a result of incomplete tryptic digestion. The phosphorylation sites were identified across multiple experiments, and several sites independently identified by multiple laboratories. While a small percentage of phosphorylation sites identified by large-scale mass spectrometry experiments (or any analytical approach for that matter) will turn out to be incorrect, depending on the stringency applied to the interpretation of the tandem mass spectra the arguments listed above effectively address any concerns regarding whether the region from 826 to 1213 is extensively phosphorylated.

This phosphorylation-rich area identified in mouse is highly conserved in rat (99% sequence identity) where all potential sites of phosphorylation are present, and in human (90% sequence identity) with all potential sites present except for S1132.

Potential phosphorylation sites of the C-terminal segment have been investigated in previous studies on recombinant fusion proteins. These experiments identified phospho-Ser1288 and phospho-Ser1392 (all sites renumbered according to murine Q80TE7-1) with a potential role in Densin-180/CamkIIα interaction (Strack et al. 2000; Walikonis et al. 2001).

Mass-spectrometric analysis of native Densin-180 confirmed phospho-Ser1392, but phospho-Ser1288 was not detected (Walikonis et al. 2001; Trinidad et al. 2006, 2008; Munton et al. 2007). A failure to identify phospho-Ser1288 to date is possibly due to intrinsic limitations of mass-spectrometric analysis of very complex mixtures of tryptic digests (Kuster et al. 2005), or due to a low stoichiometry for that site in the particular samples analyzed.

Taken together, the mass-spectrometric results clearly identified multiple phosphorylation sites within the proposed extracellular domain of Densin-180. While there is sparse evidence in the literature for extracellular kinases and a few extracellular phosphorylation sites (e.g., Redegeld et al. 1999; Canton and Litchfield 2006), such a clustering of ecto-phosphorylation sites has never been reported and seems highly unlikely. In acknowledgement of these new data we therefore took into consideration the possibility of a different topology, namely an all-intracellular and membrane-associated localization of Densin-180 (Fig. 1d), as further outlined below.

Lack of evidence for a signal peptide and transmembrane region

The new phosphorylation data prompted us to re-evaluate the evidence for the presence of a signal peptide required for an extracellular location of the N-terminus, and to search for putative transmembrane segments, using a range of current bioinformatics tools.

The originally proposed TMD for Densin-180 at residues 1218–1241 is atypical because of the numerous charged and polar amino acids (4 and 5 residues, respectively) within the assigned region, which also contained two prolines (Apperson et al. 1996).

We found no evidence of a signal peptide in the murine Densin-180 protein sequence (Uniprot Q80TE7-1) using seven current structural domain prediction programs (Table S2). Similarly, none of these software tools predicted the TMD suggested in Fig. 1(a), and positioned in Fig. 1(b). Only one of the programs [TopPred (Mobyle project, France, http://mobyle.pasteur.fr; Table S2) using the GES-scale] yielded a putative TMD (with a borderline score of 0.90) from AA538–558, namely RSMCAPLPVAAQSTTLPSLSG. This sequence, apart from being located much further upstream than the original proposed area, seems unlikely to form a TMD for there are three prolines within the proposed stretch and a cluster of polar AA (QSTT– –S–S). While a TMD at that region would place all phosphorylation sites identified by mass-spectrometric approaches in an intracellular location, a single TMD in combination with a cytoplasmic C-terminal segment would require a signal-peptide, for which there is neither bioinformatics (see above) nor experimental (see below) evidence.

Thus, the analysis with novel bioinformatics tools suggests a cytoplasmic location of the full Densin-180 protein (Fig. 1d).

Glycosylation of Densin-180?

Only limited experimental data on the glycosylation status of Densin-180 is available. Initial sequence analysis suggested a Mucin-like domain at amino acid residues 820–910 (Fig. 1a), in analogy to GpIbα (Apperson et al. 1996). Mucin-like domains are characterized by high abundance of prolines, serines, and threonines, and are O-linked glycosylated (Desseyn et al. 2008), suggesting that also Densin-180 might be glycosylated.

Enzymatic digests with a range of glycosidases found no evidence for most types of glycosylation, including N-linked glycans of the high Mannose, hybrid and complex oligosaccharide type, and O-linked di-Galactose moieties (Apperson et al. 1996). Only prolonged incubation with Neuraminidase suggested the presence of sialic-acid modified residues, and digestion with O-sialoglycoprotein endoprotease suggested possible positioning of O-sialoglycosylation within the Mucin domain. The experimental evidence supporting this prediction is quite weak.

The cluster of in vivo phosphorylation sites identified by recent mass-spectrometric analysis is located within the postulated Mucin-like domain. If we accept that these phosphorylation sites are intracellular, then glycosylation in this area is intracellular as well. Intracellular O-glycosylation is not an uncommon phenomenon (Hart 1997) and may occasionally compete with phosphorylation on the same residue (Vosseller et al. 2006). In the absence of high-resolution experimental data on Densin-180, its in vivo glycosylation status remains, however, uncertain.

In the absence of clear evidence of glycosylation within the 826–1213 segment, we therefore prefer to label this area ‘phosphorylation-rich’ segment (Fig. 1c and d).

Palmitoylation of the N-terminal segment and membrane association

While most studies have focused on the C-terminal part of Densin-180, only few groups have addressed the function of the N-terminus. In functional terms, the LRRs have been found to be necessary for the association of Densin-180 with the plasma membrane (Quitsch et al. 2005). Very recently, palmitoylation of recombinant human Densin-180 and Erbin (see below) has been reported (Izawa et al. 2008). In analogy to the detailed mapping studies on Erbin, the palmitoylation of Densin-180 is predicted to occur at conserved residues Cys14/16.

Palmitoylation, an intracellular modification (Walsh 2006), explains membrane-associated localization of these molecules without the need for a TMD topology. In this regard it is interesting to notice that expression of a splice variant of Densin-180 lacking the putative TMD region, but still including the previously suggested signal peptide, led to a firmly membrane-associated protein, not to secretion of the protein (Jiao et al. 2008). This experimental evidence supports the bioinformatics prediction that Densin-180 lacks a signal peptide.

Further, the large central area following the LRRs inclusive the originally proposed TMD has an influence on subcellular distribution and filopodia outgrowth (Jiao et al. 2008). Deletion of residues 478–1372 led to a diffuse localization of recombinant Densin-180 near the plasma membrane instead of a punctate one, which could be explained by the lack of possible protein-interaction sites of the truncate Densin-180.

This membrane-associated, cytoplasmic localization readily explains the inaccessibility of full-length Densin-180 to surface biotinylation (Izawa et al. 2002). All the above-mentioned studies on the N-terminal part of Densin-180 have questioned the TMD topology of Densin-180 as their results are more in favor of an all-intracellular model. In summary, these data suggest a membrane-associated/anchored location for Densin-180 (Fig. 1d).

Densin-180 in non-neuronal tissue

Densin-180 has been described only in a small number of tissues outside of the nervous system. In kidney, Densin co-precipitates with nephrin from glomerular podocyte lysates. In this experiment the molecular weight was reported as 210 kDa protein in western blot analysis, following mass-spectrometric identification (Ahola et al. 2003). It is located at the slit diaphragm of the kidney, adherens junctions of podocytes, interacts with β–, and probably α–, catenin, and co-localizes with F-actin, (Heikkila et al. 2007). Upon disruption of junctions by calcium deprivation or puromycin aminonucleoside treatment, Densin is dispersed from podocyte plasma membranes.

Consistent with its location at the slit diaphragm in renal podocytes, Densin is also found at adherens junctions of testicular Sertoli cells (Lassila et al. 2007) and was further identified in pancreatic islet cells (Rinta-Valkama et al. 2007).

Surprisingly little is known about the role of Densin-180 in pathological conditions and human diseases. Increased Densin mRNA and protein levels have been reported in kidneys of CNF (Congenital Nephrotic Syndrome of the Finnish type) patients, where a mutation in the gene encoding for nephrin leads to the absence of the protein, hence resulting in strong proteinuria (Ahola et al. 2003).

Densin-180 is a member of the LAP protein family

The identification of several new proteins homologous to Densin-180, such as Scribble (Bilder and Perrimon 2000), Erbin (Borg et al. 2000), and Lano (Saito et al. 2001), has led to the definition of a new protein family, namely the LRR and PDZ domain-containing (LAP) proteins family (Bryant and Huwe 2000; Bilder et al. 2001; Wilson et al. 2001; Santoni et al. 2002). These proteins are characterized by 16 LRRs, 2 LAP-specific domains and 0–4 PDZ domains and are grouped into subfamilies according to the number of PDZ domains.

All of the more recently identified members of this family are thought to be cytosolic membrane-associated proteins. This includes Erbin, which shares the highest similarity to Densin-180 (Borg et al. 2000; Huang et al. 2001; Izawa et al. 2008). These two proteins are 68% identical, have both a single PDZ domain and together define the LAP1 protein subfamily (Santoni et al. 2002). Their LRRs, LAP-specific domains a and b, and even the region between the LAP-specific domains are conserved at a similar level (70%, 71%, 79%, and 54% identity, respectively). The phosphorylation-rich domain is much less conserved with 27% identity. While Erbin has been found phosphorylated in several studies (http://www.phosphosite.org) and has also been identified in PSDs (Trinidad et al. 2008), no sites of phosphorylation have so far been identified in synaptic preparations. Taken together this suggests that the two proteins might be differently regulated by phosphorylation events.

Possible additional function for Densin-180 arising from the all intracellular model

Although Densin-180 has been identified more than 10 years ago, surprisingly little is known about its function. This might be primarily because of the long prevailing view that the main part of the protein was extracellular and heavily glycosylated and hence much less accessible for interaction studies in heterologous expression systems than the relatively short C-terminus. The LRRs of Let-413 (Legouis et al. 2003) and Erbin (Legouis et al. 2003; Izawa et al. 2008), Scribble (Zeitler et al. 2004; Navarro et al. 2005) and Densin-180 (Quitsch et al. 2005; Jiao et al. 2008) have been found to be required for localization of the protein at the plasma membrane. While the possibility that this domain interacts directly with the plasma membrane cannot be ruled out, the specific localization of Erbin constructs (Legouis et al. 2003) and Densin-180 (Quitsch et al. 2005) at the basolateral versus the apical membrane suggests that the LRRs domains most likely interact with proteins that in turn have specific subcellular localization. Sur8, a protein with a LRRs domain closely related to the one of the LAP protein Let-413, binds to Ras, a small GTPase with membrane anchoring properties, and thereby influences tyrosine kinase receptor/Mapk signaling (Sieburth et al. 1998). While no small GTPases have yet been found to interact with the LRRs of LAP proteins, they are likely candidates of interaction and might therefore provide a link of LAP proteins with intracellular signaling cascades, and hence could provide a role for Densin-180 in integral parts of the post-synaptic signaling machinery that involves small GTPases. Further, as the PDZ domain binds different proteins than the LRRs, and binding of Shank seems to change the LRRs domain-provided membrane association (Quitsch et al. 2005), there might be also a role of Densin-180 in differentially regulating several signaling pathways downstream of synaptic transmission. The fact that there seem to be three different areas of interaction with proteins, namely the LRRs, the phosphorylation-rich and the PDZ domain could place Densin-180 as a hub in a critical cross-linking position between different signaling cascades.

Summary/revised topology

In summary, the latest experimental data and the results from structural prediction engines strongly suggest an all-intracellular topology for Densin-180 (Fig. 1d), in parallel with all other members of the LAP family. The palmitoylation and the LRR domain likely play an important part in anchoring the molecule to the plasma membrane (Izawa et al. 2008). The LRR domain of Densin-180 is most likely involved in protein–protein interactions with cytosolic proteins, in addition to the well-studied interaction of the PDZ domain.

The large central area of Densin-180, between the LRRs and the PDZ domain, is modified with multiple phosphorylations, very likely by several different kinases (Trinidad et al. 2006, 2008). Our revised topology model provides a new framework for the analysis of the phosphorylation state of Densin-180, it interaction with other proteins, its subcellular targeting, and function at postsynaptic densities.

Abbreviations used

LAP

LRR and PDZ domain-containing

LRR

leucine-rich repeat

PDZ

Psd95/Discs large/Zona occludens-1

PSD

post-synaptic density

TMD

transmembrane domain