Lipolysis-stimulated lipoprotein receptor: a novel membrane protein of tricellular tight junctions

Tricellular tight junctions (tTJs) are specialized structural variants of tight junctions that restrict the free diffusion of solutes at the extracellular space of tricellular contacts. Their presence at cell corners, situated in the angles between three adjacent epithelial cells, was identified early by electron microscopy, but despite their potential importance, tTJs have been generally ignored in epithelial cell biology. Tricellulin was the first molecular component of tTJs shown to be involved in their formation and in epithelial barrier function. However, the precise molecular organization and function of tTJs are still largely unknown. Recently, we identified the lipolysis‐stimulated lipoprotein receptor (LSR) as a tTJ‐associated membrane protein. LSR recruits tricellulin to tTJs, suggesting that the LSR‐tricellulin system plays a key role in tTJ formation. In this paper, we summarize the identification and characterization of LSR as a molecular component of tTJs.

Keywords: tricellular contact; tight junction; LSR; tricellulin

A belt of tight junctions (TJs) circumscribes the apical‐most regions of the lateral plasma membranes of epithelial cells.[[1]] Consequently, all routes through the intercellular space are sealed by TJs, and the cellular sheet is able to provide a barrier to the paracellular diffusion of solutes. However, this simple description ignores the frequent occurrence of tricellular contacts, where the vertices of three epithelial cells meet within the cellular sheet. Claudin‐based TJs are formed between two plasma membranes, but it is not immediately obvious how the extracellular space is effectively excluded at tricellular contacts, where three plasma membranes assemble. Observations of these regions in vertebrate epithelial cells by freeze‐fracture electron microscopy in the early 1970s identified a specialized type of tight junction, namely tricellular TJs (tTJs).[[3]] Images of freeze‐fracture replicas suggest that the belt of TJs is not continuous at tricellular contacts. In these regions, the most apical elements of the TJ strands on each side of the junction join and turn to extend in the basal direction (Fig. 1). These strands, which are considered to be the central sealing elements, are connected by short TJ strands extending from the TJs on both sides to form networks.[1] This results in three sets of central sealing elements assembling to exclude the extracellular space at the tricellular contacts. This arrangement is thought to impede the paracellular diffusion of solutes. These observations suggest that specialized proteins may be involved in the formation of tTJs. However, until recently, such molecules were not identified. Consequently, for a long period tTJs attracted little attention in the field of epithelial cell biology.

Graph: 1 Freeze‐fracture replica of tTJs in MDCK cells. TJ strands on both sides join to form the central sealing element (arrow) and extend to the basal direction. Scale bar: 200 nm.

An integral membrane protein, tricellulin was the first molecular constituent of tTJs to be identified in 2005.[7] Tricellulin has four transmembrane domains and shows sequence similarity to occludin, a TJ‐associated protein. Tricellulin is expressed in a variety of epithelial cells and is localized at the central sealing elements of tTJs. In RNAi‐mediated suppression and overexpression studies on cultured epithelial cells, it was observed that tricellulin was required for full development of transepithelial electrical resistance (TER) and for the generation of the paracellular barrier to macromolecules.[[7]] In addition, exogenously expressed tricellulin increased the cross‐linking of claudin‐based TJ strands in claudin‐expressing fibroblasts,[9] suggesting the possibility that tricellulin is also involved in the complex network of claudin‐based TJ strands in tTJs. However, there are still many unsolved questions regarding tTJs. The detailed mechanism underlying the function of tricellulin has not been clarified at the molecular level. It is likely that the peculiar localization of tricellulin in tTJs at cell vertices is important for its function, but the basis of this localization is unknown. It is also likely that there are other molecular constituents specific to tTJs. It is necessary to identify and characterize the tTJ‐associated molecules to fully understand their organization and function at the molecular level.

TJs cannot be highly purified for biochemical analyses by subcellular fractionation. Therefore, to identify novel TJ‐associated proteins, localization‐based screening of monoclonal antibodies generated against a TJ‐containing membrane fraction has often been used. Many of the key proteins of TJs, including ZO‐1,[10] cingulin,[11] occludin,[12] and JAM,[13] were discovered in this way. However, a limitation of this approach is that it depends on the antigenicity of putative junctional proteins, which cannot be controlled. To overcome this problem, we applied a localization‐based expression cloning method, namely FL‐REX (fluorescence localization‐based retrovirus‐mediated expression cloning), developed by Misawa and colleagues,[14] for the identification of novel molecular constituents of TJs, including tTJs. In this visual screening method, a green fluorescent protein (GFP)‐fused cDNA library constructed in a retrovirus‐based expression vector is first introduced into cells. The cells in which the exogenous GFP‐fusion proteins are observed to be localized at certain cellular structures are then selected under a fluorescence microscope. Their cDNAs integrated into the genome are finally cloned by genomic PCR. Using an FL‐REX screen using a GFP‐fusion library constructed from T84 human colon carcinoma cell‐derived poly(A)+ RNA, we identified lipolysis‐stimulated lipoprotein receptor (LSR) as a novel molecular constituent of tricellular contacts.[15] LSR is a single transmembrane protein of about 65 kD molecular mass. It contains an extracellular Ig‐like domain, a transmembrane domain, and a long cytoplasmic domain. Immunofluorescence analyses of cultured epithelial cells and of frozen sections of various mouse tissues revealed that LSR was localized at tricellular contacts of most epithelial tissues (Fig. 2). Furthermore, immunoreplica labeling revealed that LSR was concentrated in the central sealing elements of tTJs observed by electron microscopy, that is, a similar location to tricellulin. These results indicate that LSR is a novel molecular component of tTJs.[15]

Graph: 2 Immunofluorescence localization of LSR in epithelial cells in the mouse epidydimis. Occludin is also labeled to delineate TJs. LSR is highly concentrated in tricellular contacts. Scale bar: 10 μm.

To investigate the roles of LSR in tTJ formation and in epithelial barrier function, EpH4 mouse mammary epithelial cells with stable suppression of LSR expression by LSR‐specific shRNA were established. Immunofluorescence staining of LSR‐knockdown EpH4 cells with the antibody to occludin (an excellent marker of TJs) indicated that under subconfluent conditions tTJ formation was altered; occludin staining at tricellular contacts appeared to be discontinuous, and abnormal accumulation of occludin was often observed in these regions. Furthermore, LSR‐knockdown in EpH4 cells grown on permeable filters under confluent conditions showed reduced TER compared with parent EpH4 cells. Nevertheless, TER of LSR‐knockdown EpH4 cellular sheets (∼1000 ohm cm2) was still much higher than that of confluent Madin‐Darby canine kidney (MDCK) II cellular sheets (50–70 ohm cm2).[16] The effects on tTJ formation and TER in LSR‐knockdown EpH4 cells were canceled by re‐expression of HA‐tagged LSR.[15] These observations indicate that LSR is required for normal tTJ formation and to produce the high transepithelial electric resistance of the cellular sheet.

The most significant feature of LSR is its interaction with tricellulin. In tricellulin‐knockdown EpH4 cells, immunofluorescence staining showed that LSR was still located at tricellular contacts. In contrast, in LSR‐knockdown EpH4 cells, tricellulin lost its tricellular localization and was distributed throughout the basolateral membrane. This phenotype was canceled by reexpression of RNAi‐resistant LSR in LSR‐knockdown cells (Fig. 3).[15] These observations suggest that LSR recruits tricellulin to tricellular contacts. When a deletion mutant of LSR lacking three‐quarters of the C‐terminal cytoplasmic domain was introduced into LSR‐knockdown EpH4 cells, this construct was localized to tricellular contacts but tricellulin was absent, suggesting that the cytoplasmic region of LSR is necessary to recruit tricellulin.[15] LSR‐mediated tricellulin recruitment was also reproduced in fibroblasts.[15] When LSR was stably expressed in L cells, exogenous LSR was assembled into short lines of dots, part of which were cell–cell contacts. In contrast, tricellulin that was exogenously expressed in L cells was diffusely distributed throughout the plasma membrane. However, in LSR‐expressing L cells, exogenous tricellulin was colocalized with LSR as dots, again suggesting that LSR recruits tricellulin. When N‐terminal cytoplasmic domain‐deleted or C‐terminal cytoplasmic domain‐deleted tricellulin was expressed in LSR‐expressing L cells, the former, but not the latter, was colocalized with LSR. Furthermore, a non‐TJ–associated membrane protein CD9 fused with the C‐terminal cytoplasmic region of tricellulin was also colocalized with LSR, but CD9 was not.[15] Taken together, these results indicate that LSR recruits tricellulin to tTJs, and that the interaction between the cytoplasmic domain of LSR and the C‐terminal cytoplasmic domain of tricellulin is required for this recruitment.

Graph: 3 LSR recruits tricellulin to tricellular contacts. Images show double immunofluorescence staining of mouse EpH4 epithelial cells, tricellulin knockdown EpH4 cells (Tricellulin‐KD), LSR‐knockdown cells (LSR‐KD), and LSR‐knockdown cells expressing RNAi‐resistant HA‐tagged LSR (LSR‐KD/LSR) with anti‐LSR and anti‐tricellulin antibodies. Scale bar: 10 μm.

Identification of LSR presents a possible model for tTJ formation. In this model, LSR assembles at the corners of epithelial cells to generate a landmark for tTJ formation, and tricellulin is recruited to tricellular contacts via its interaction with LSR. Because tricellulin seems to have an affinity for claudins within the plasma membrane,[9] it might recruit claudin‐based TJ strands to tricellular contacts to form tTJs. Furthermore, the ability of tricellulin to increase cross‐linking of claudin‐based TJ strands in mouse L cells[9] suggests that it might help to connect the many short TJ strands at tTJs. Perhaps the most intriguing question now is how LSR recognizes tricellular contacts for its own localization. The LSR domain responsible and the factors that recruit LSR to tricellular contacts should be investigated in future studies.

LSR was originally identified as a triglyceride‐rich lipoprotein receptor.[17] LSR gene‐deficient mice had previously been reported to be embryonic lethal,[18] whereas heterozygous mice show increased fatty acid levels after food intake.[19] However, to date the relationship between these two functions of LSR—in tTJ formation and in lipoprotein uptake—remains elusive. To clarify the role of LSR in tTJ formation in vivo, tissue‐specific conditional knockout mice for the LSR gene should be generated and analyzed.

Recently, we found that LSR could not be detected in some epithelial tissues by immunofluorescence staining, despite tricellulin localization at tricellular contacts (unpublished observation). Given our conclusion described earlier that LSR recruits tricellulin to tricellular contacts, it seems that other proteins must assume the role of LSR. Two candidates are the LSR‐related proteins, ILDR1[20] and ILDR2/c1orf32/Lisch‐like.[[21]] Both are membrane proteins containing an Ig‐like domain and their overall structure shares a primary amino acid sequence with that of LSR. It would be of great interest to examine whether ILDR1 and ILDR2/c1orf32/Lisch‐like also share common functional features with LSR in terms of their possible localization at tTJs and recruitment of tricellulin. In particular, ILDR1 is intriguing because it has recently been reported to be a causal gene for autosomal‐recessive hearing impairment, in which tricellulin mutations were also identified.[[23]]

We thank Dr. Hiroyuki Sasaki (Jikei University School of Medicine) for providing an electron micrograph for Figure 1. M.F. is supported by a NEXT Program from the Japan Society for the Promotion of Science.

The authors declare no conflicts of interest.