A cellular model of albumin endocytosis uncovers a link between membrane and nuclear proteins

Albumin, the most abundant circulating plasma protein, binds a wide variety of exogenous and endogenous molecules and has multiple physiological functions, including maintenance of colloid osmotic pressure, antioxidant activity and immune modulation (Quinlan et al., 2005; Spinella et al., 2016). Albumin is mainly synthesized in liver and its level is balanced by renal or gastrointestinal excretion and catabolism (Fanali et al., 2012; Naldi et al., 2017). Disruption of albumin homeostasis is associated with a variety of human diseases. In healthy kidneys, small amounts of serum albumin cross the glomerular filtration barrier and are efficiently reabsorbed by renal tubular cells under physiological conditions (Birn and Christensen, 2006; Gekle, 2005). Albuminuria occurs when the permeability of the glomerular capillary wall is increased or when reabsorption of albumin by tubular cells is impeded (Comper et al., 2008; Pollock and Poronnik, 2007). Clinically, the presence of albuminuria contributes to the progression of chronic kidney disease, and moderate albuminuria over time is also an important risk factor for cardiovascular disease and early cardiovascular mortality (Lambers Heerspink and Gansevoort, 2015; Parving et al., 2015).

Cubilin (CUBN), encoded by the CUBN gene, is one of the essential albumin receptors. CUBN is a 460 kDa multiligand receptor expressed in renal and intestinal epithelial cells (Amsellem et al., 2010; Birn et al., 2000; Christensen et al., 2013). Other ligands of CUBN include transferrin, intrinsic factor (IF)-vitamin B₁₂, hemoglobin, immunoglobulin light chain, apolipoprotein A-I and high-density lipoprotein (Batuman et al., 1998; Birn et al., 2000, 1997; Gburek et al., 2002; Gianesello et al., 2017; Hammad et al., 2000; Kozyraki et al., 1999; Kozyraki et al., 2001; Nagai et al., 2011; Nielsen et al., 2016). CUBN contains an N-terminal stretch, eight epidermal growth factor (EGF)-like repeats and 27 CUB [complement C1r/C1 s, Uegf (epidermal growth factor-related sea urchin protein) and BMP1 (bone morphogenic protein 1)] domains (Kozyraki et al., 1998; Moestrup et al., 1998). CUBN lacks a transmembrane domain and an intracellular domain (Moestrup et al., 1998) and thus requires another molecule, amnionless (AMN), to function as a receptor complex, called CUBAM (Fyfe et al., 2004; Kalantry et al., 2001; Tanner et al., 2003). AMN is a 38-50 kDa type I transmembrane protein that is expressed in intestine and kidney, where CUBN is also abundantly expressed (Strope et al., 2004). AMN has a cysteine-rich stretch in its extracellular domain and two [FY]XNPX[FY] endocytosis signal motifs in its intracellular domain (Fyfe et al., 2004; Kalantry et al., 2001), to which two adaptor proteins, autosomal recessive hypercholesterolemia protein (ARH, also known as LDLRAP1) and disabled-2 (DAB2), bind (Pedersen et al., 2010). In the absence of AMN, CUBN is retained in the endoplasmic reticulum (ER) and is unable to be targeted to the plasma membrane (Ahuja et al., 2008; Coudroy et al., 2005; Fyfe et al., 2004; He et al., 2005; Strope et al., 2004).

The crucial role of CUBN in ligand endocytosis in renal and intestinal tissues has been established in animal models and human disease. Cubilin-deficient mice exhibited markedly decreased uptake of albumin by proximal tubule cells and resultant albuminuria (Amsellem et al., 2010). In humans, several variants in CUBN are associated with albuminuria, suggesting the crucial function of CUBN in renal clearance of albumin (Boger et al., 2011; Teumer et al., 2016). Furthermore, loss-of-function mutations either in CUBN or AMN are reported to cause the congenital disease Imerslund–Grasbeck syndrome (IGS), also known as juvenile megaloblastic anemia 1 (MGA1) (OMIM #261100) (Aminoff et al., 1999; Tanner et al., 2003, 2012). Patients with IGS present with low molecular weight proteinuria and megaloblastic anemia as a result of impaired IF-vitamin B₁₂ complex absorption in ileum (Gräsbeck, 2006; Tanner et al., 2012). One pathogenic mechanism of IGS is the disruption of correct intracellular trafficking of CUBAM to the cell membrane (Amsellem et al., 2010; Coudroy et al., 2005; Fyfe et al., 2004, 1991; He et al., 2005; Storm et al., 2011; Strope et al., 2004). We previously reported that AMN-dependent glycosylation of CUBN is crucial for membrane targeting (Udagawa et al., 2018) and several mutations of CUBN and AMN result in abrogation of AMN-dependent plasma membrane localization of CUBN (Udagawa et al., 2018). Another mechanism of IGS is inhibition of ligand binding to CUBN by mutations within its vitamin B₁₂/IF-binding region (CUB5-8) (Andersen et al., 2010; Tanner et al., 2012).

So far, the mechanism underlying the endocytosis of albumin by CUBN has remained controversial. Previous biochemical studies have identified two distinct domains in CUBN that directly bind to albumin: the 113 amino acid (aa) N-terminal stretch and the CUB 7 and CUB 8 domains (Yammani et al., 2001). However, accumulating evidence in humans suggests that mutations or variants in the C-terminal part (CUB9-27 domains) of CUBN rather than the CUB5-8 domains are associated with albuminuria (Ahluwalia et al., 2019; Bedin et al., 2020; Boger et al., 2011; Haas et al., 2018; Teumer et al., 2016; Zanetti et al., 2019). To date, the role the C-terminal CUB domains of CUBN in albumin endocytosis has not been experimentally demonstrated.

Albumin endocytosis in renal tubular cells is also regulated by another CUBN-binding protein, megalin (also known as LRP-2), a large transmembrane protein that belongs to the LDL receptor family (Eshbach and Weisz, 2017). Megalin forms a trimetric complex with CUBN and AMN (Ahuja et al., 2008) and also bears endocytosis signal motifs on its intracellular domain (Saito et al., 1994). By using genetic mouse models, megalin was shown to promote endocytosis of the CUBN-albumin complex (Amsellem et al., 2010). Although AMN is essential for surface expression of CUBN, whether AMN is involved in CUBN-mediated albumin endocytosis is not clear. Furthermore, the regulatory mechanism of CUBN-mediated endocytosis is poorly understood.

In the present study, we establish a quantitative assay to evaluate cellular albumin uptake in cultured cells mediated by CUBAM. We demonstrate that the C-terminal part of CUBN and the signal motifs of AMN are crucial for albumin endocytosis. We also identify nuclear valosin-containing protein-like 2 (NVL2) as a novel interacting protein of AMN and show that NVL2 is involved in CUBN-mediated endocytosis of albumin. Our findings reveal a novel function of a nucleolar protein in regulating ligand endocytosis through a cell surface receptor.

To analyze intracellular traffic or endocytosis by CUBAM in cultured cells, several short forms of CUBN have been used, including N-terminal CUBN fragments composed of the 113 aa N-terminal stretch, EGF repeats and the CUB1-4 domain [mini-CUBN (1-4)] (Udagawa et al., 2018) (Fig. 1A) or CUB1-8 domain [mini-CUBN (1-8)] (Coudroy et al., 2005; Fyfe et al., 2004) (Fig. S1A). We generated a mammalian expression vector encoding full-length human CUBN with a FLAG-tag on its C terminus (Fig. 1A). The full-length CUBN expressed in HEK293 T cells interacted with AMN (Fig. 1B) and was targeted to the cell membrane only when AMN was coexpressed (Fig. 1C), similar to mini-CUBNs (Udagawa et al., 2018).

The albumin endocytosis capacity of full-length CUBN was compared with that of mini-CUBN (1-4) or mini-CUBN (1-8). Cells expressing mini-CUBN (1-4), mini-CUBN (1-8) or full-length CUBN (Fig. S1A,B) were incubated in medium containing FITC-bovine serum albumin (BSA) for 1 h. The amounts of albumin endocytosed in the cells and CUBN expressed on cell surface were quantified by immunostaining without permeabilization. Cells expressing full-length CUBN, but not those expressing mini-CUBN (1-4) at the cell surface, demonstrated significantly increased internalization of albumin compared with CUBN-negative cells (Fig. 1D). Internalization of albumin by full-length CUBN was also confirmed by flow cytometry analysis (Fig. 1E,F). Although a previous study demonstrated that the CUB 7-8 domain of CUBN binds to albumin (Yammani et al., 2001), expression of mini-CUBN (1-8) failed to increase the cellular uptake of albumin (Fig. S1C,D). In cells expressing full-length CUBN, the signals of internalized albumin were colocalized with internalized CUBN and partially colocalized with EEA1, an early endosome marker (Mu et al., 1995) (Fig. 1G). These results suggest that albumin is internalized by full-length CUBN, but not by the N-terminal mini-CUBN (1-8), and colocalizes with CUBN in early endosomes.

AMN has two signal motifs in its intracellular domain that are essential for internalization of the IF-vitamin B₁₂ complex mediated by CUBAM (Pedersen et al., 2010). The AMN mutant that lacks both signal motifs (Sdel-AMN) (Fig. 2A) bound to CUBN and promoted CUBN membrane targeting in the same manner as wild-type AMN (Fig. 2B; Fig. S2A,B). However, cells expressing Sdel-AMN together with full-length CUBN demonstrated significantly impaired albumin uptake (Fig. 2C,D), suggesting that the endocytosis signal motifs of AMN are crucial for CUBAM-mediated albumin uptake.

To investigate whether the internalization of CUBAM is dependent on megalin, the endocytosis assay was performed in human cultured colorectal HCT116 cells, which do not express endogenous megalin mRNA (Udagawa et al., 2018). CUBN endocytosis dependency on AMN signal motifs was also observed in HCT116 cells, suggesting that CUBAM endocytosis occurs independently of megalin expression (Fig. S2C,D).

To investigate the regulatory mechanism of albumin endocytosis mediated by CUBAM, an interactome analysis of AMN using stable isotope labeling by amino acids in cell culture (SILAC) technology was performed (Fig. 3A). AMN was expressed in HEK293T cells that were cultured in medium containing heavy isotope-labeled amino acids, and empty vector was transfected into cells cultured in medium containing light isotope-labeled amino acids. The lysates from both cell samples were mixed, immunoprecipitated and subjected to nanoscale liquid chromatography coupled with tandem mass spectrometry (nano-LC-MS/MS). Nuclear valosin-containing protein-like (NVL) was among the candidate molecules interacting with AMN that contained higher ratios of heavy/light amino acids (Fig. 3B,C).

NVL is an AAA-ATPase (ATPase associated with various cellular activities). The human genome encodes two isoforms, NVL1 and NVL2, which are identical over the 750 C-terminal amino acids (Fig. 3D) (Germain-Lee et al., 1997). Although the physiological role of NVL1 has not been clarified, NVL2 has been reported to localize in the cell nucleolus and function in ribosome synthesis and telomerase activity (Her and Chung, 2012; Nagahama et al., 2004). Immunoprecipitation assays revealed that NVL2 but not NVL1 interacts with AMN (Fig. 3E; Fig. S3A,B). An interaction between endogenous NVL2 and AMN was also confirmed in CACO2 cells (Fig. 3F).

In HEK293T cells, which do not express endogenous AMN (Fig. S4A), endogenous NVL is located in the nuclear compartment. However, when AMN is overexpressed in these cells, endogenous NVL translocates to the extranuclear compartment (Fig. 4A). Proximity ligation assay (PLA), which detects the approximation of two molecules within 40 nm, showed signals indicating the proximity of AMN and NVL2 in the extranuclear compartment (Fig. 4B). Co-immunostaining with calnexin (CNX), an ER marker (Wada et al., 1991), revealed that NVL colocalized with AMN in the ER in AMN-expressing cells (Fig. 4C). CUBN and AMN are interdependently targeted to the plasma membrane (Fyfe et al., 2004; Udagawa et al., 2018). We also analyzed the localization of NVL2 and its interaction with AMN in cells expressing both CUBN and AMN. In cells coexpressing NVL2, AMN and CUBN, NVL2 partially colocalizes with β-catenin along the cell membrane (Ozawa et al., 1989) (Fig. 4D). PLA assay with co-staining of CNX or β-catenin showed the proximity of AMN and NVL2 both in the ER and along the plasma membrane in CUBAM-expressing cells (Fig. 4E,F). Furthermore, the interaction between NVL2 and CUBN along the cell surface was analyzed using cells expressing wild-type CUBN or the G653R mutant, which is defective in trafficking to the cell surface (Udagawa et al., 2018). HEK293T cells transiently expressing FLAG-tagged wild-type or mutant CUBN along with exogenous NVL2 and AMN were incubated in medium containing anti-FLAG antibody so that CUBN on the cell surface was labeled (Fig. S3C). Cell lysates were incubated with Protein G Sepharose beads, and bound proteins were analyzed by western blotting. In cells expressing the CUBN G653R mutant, levels of CUBN on the cell surface as well as AMN and NVL2 bound to CUBN on the cell surface were markedly reduced compared with cells expressing wild-type CUBN. These results demonstrate that NVL2 is expressed in extranuclear compartments, such as the ER and plasma membrane, in cells expressing AMN and CUBN.

In human renal tubules and intestine, NVL was located mainly in the nucleolus. However, distinct signals for NVL were also detected in the extranuclear compartment, colocalizing with AMN signals (Fig. 4G-I; Fig. S5).

To explore further the physical association of NVL2 and AMN, we analyzed the domain of AMN responsible for its interaction with NVL2 (Fig. 5A). Immunoprecipitation assay demonstrated that elimination of the intracellular domain of AMN (AMN-ΔICD) remarkably impaired its interaction with NVL2 (Fig. 5B). Pull-down analysis demonstrated that the intracellular domain of AMN (AMN-ICD) was sufficient to interact with NVL2 (Fig. 5C). Elimination of signal motifs in AMN-ICD (Fig. 2A) completely abrogated its interaction with NVL2 (Fig. 5D), and Sdel-AMN did not cause the translocation of NVL into cytoplasm that was observed with wild-type AMN (Fig. 5E). Therefore, both interaction of AMN and NVL2 and cytoplasmic translocation of NVL are dependent on the intracellular endocytosis signal motifs of AMN.

Because the signal motifs of AMN mediate endocytosis of CUBAM through interactions with the adaptor molecules ARH and Dab2 (Pedersen et al., 2010), we examined the interaction between NVL2 and the adaptor proteins by immunoprecipitation. Although we could not detect an interaction between NVL2 and Dab2, NVL2 bound to ARH only when AMN was coexpressed (Fig. S6A). PLA assay also confirmed the proximity of NVL2 and ARH in cells expressing AMN (Fig. S6C). The signals indicating the AMN-dependent proximity of NVL2 and ARH in CUBAM-expressing cells were partly colocalized with β-catenin along the plasma membrane (Fig. S6D). These results suggest that the intracellular signal motifs of AMN mediate formation of the NVL2-AMN-ARH trimeric complex along the plasma membrane.

We also investigated the domains of NVL2 that are responsible for its interaction with AMN (Fig. 5F). Immunoprecipitation assays demonstrated that elimination of the N terminus of NVL2 (NVL2-D1D2) remarkably impaired its interaction with AMN (Fig. 5G), suggesting that the N terminus of NVL2 is the domain essential for its interaction with AMN. Pull-down analysis demonstrated that the recombinant N terminus of NVL2 (NVL2-NT) was sufficient to interact with AMN (Fig. 5H).

NVL2 is an AAA-ATPase with two functional ATPase domains (AAA domains) in the central region (D1) and the C terminus (D2). We investigated whether the capacity of NVL2 for ATP hydrolysis or ATP binding influences the interaction between NVL2 and AMN. We used NVL2 with E365Q or E682Q mutation, which impairs the ATP-hydrolysis capacity of the AAA domain in D1 or D2, respectively, and NVL2 with K311M or K628M mutation, which impairs the ATP-binding capacity of the AAA domain in D1 or D2, respectively (Fig. 5F) (Lo et al., 2019). Immunoprecipitation assays showed that although the NVL2 mutants deficient in ATP hydrolysis (E365Q, E682Q or E365Q/E682Q) still interacted with AMN, the K311M or K311M/K628M mutation in NVL2 clearly disturbed its interaction with AMN (Fig. 5I). NVL2-AMN-ARH trimeric complex formation was also impaired by K311M mutation (Fig. S6B). These data indicated that although the N terminus plays an essential role in the interaction between NVL2 and AMN, the ATP-binding state of NVL2 influences the interaction with AMN.

To investigate the involvement of NVL2 in the endocytosis of CUBAM, we depleted NVL in HEK293T cells transiently expressing CUBN and AMN (Fig. 6A). Knockdown of endogenous NVL did not inhibit membrane targeting of CUBN (Fig. 6B).

We then examined the role of NVL on the endocytosis of CUBAM by endocytosis assay (Pedersen et al., 2010). FLAG-tagged mini-CUBN (1-4) on the cell surface was labeled with anti-FLAG antibody and incubated for 1 h at 37°C. Endocytosed CUBN molecules were detected by secondary antibody in the form of signals internalized into the cells. In control cells transfected with control siRNA, almost all the signals for CUBN were internalized after 1 h (Fig. 6C). On the other hand, internalized CUBN in two lines of NVL knockdown cells was decreased compared with control cells (Fig. 6C). We quantitatively evaluated the proportion of CUBN that was left on the cell surface after induction of endocytosis by calculating the proportion of CUBN that colocalized with β-catenin along the cell membrane (Ozawa et al., 1989). The proportion of CUBN left on the cell surface was significantly increased in NVL-depleted cells compared with control cells (Fig. 6D,E). The reduced CUBN internalization caused by NVL depletion was rescued by expression of siRNA-resistant NVL2 (Fig. 6F,G), confirming that NVL2 facilitates endocytosis of CUBN.

We also evaluated the functional influence of NVL on albumin uptake mediated by CUBAM. HEK293T cells expressing full-length CUBN and AMN with or without NVL knockdown were incubated in medium containing FITC-labeled BSA at 37°C for 1 h. Surface CUBN and β-catenin were visualized by immunostaining. In cells transfected with control siRNA, most of the albumin was internalized into the cells. In contrast, significant amounts of albumin were retained on the surface of NVL-depleted cells (Fig. 7A). Flow cytometry analysis demonstrated that uptake of FITC-BSA was significantly decreased in NVL knockdown cells compared with control cells (Fig. 7B). We also quantified the proportion of FITC-BSA retained on the cell surface by calculating the signals for CUBN on the cell surface colocalized with β-catenin. Significantly higher proportions of FITC-BSA signals were retained on the cell surface in NVL-depleted cells compared with control cells (Fig. 7C). The surface retention of albumin in NVL-depleted cells was rescued by expression of siRNA-resistant NVL2 (Fig. 7D). The role of NVL in albumin uptake was also confirmed in CACO2 cells, which express endogenous CUBN and AMN (Fig. S4B, Fig. S7A,B).

Involvement of the ATPase activity of NVL2 on CUBAM-mediated albumin uptake was also evaluated using cells expressing doxycycline (Dox)-inducible NVL2 (wild-type or E365Q/E682Q) (Yoshikatsu et al., 2015). Albumin uptake by cells expressing E365Q/E682Q NVL2 was significantly impaired (Fig. 7E,F) compared with cells expressing wild-type NVL2, suggesting that the ATP-hydrolysis capacity of NVL2 plays a crucial role in its regulation of CUBAM-mediated albumin endocytosis.

Because NVL2 is reported to play a role in ribosome biosynthesis (Nagahama et al., 2004, 2006), we examined whether impaired albumin uptake by NVL depletion was mediated by nucleolar stress (Boulon et al., 2010). Administration of actinomycin D or CX-5461, both of which inhibit Pol I-driven rRNA transcription (Carmo-Fonseca et al., 1992; Drygin et al., 2011), induced the release of nucleolin in the nucleoplasm (Fig. S8A), which indicated nucleolar disruption (Boulon et al., 2010). Notably, actinomycin D or CX-5461 did not affect CUBAM-mediated albumin uptake (Fig. S8B). These results suggest that reduced CUBAM-mediated albumin uptake caused by NVL depletion is not mediated by nucleolar stress.

Here, we have established a cellular model in which full-length CUBN is expressed on the cell surface to enable quantitative evaluation of albumin uptake. Our findings reveal a novel regulatory mechanism of endocytosis that is controlled by CUBN, AMN and NVL2 (Fig. 8).

Previous in vitro analyses using segmented fragments identified two distinct domains in CUBN that directly bind to albumin: the 113 aa N-terminal stretch and CUB 7 and CUB 8 domains (Yammani et al., 2001). However, the CUBN fragment comprising the N-terminal stretch, EGF repeats and CUB 1-8 domain was not sufficient for endocytosis of albumin in cultured cells (Fig. S1C,D). A previous study reported a patient with single-base homozygous deletion (c.8355delA) in exon 53 of CUBN, causing premature truncation of the encoded protein (p.S2785fsX19) in the CUB 20 domain; this patient presented proteinuria, but not anemia, suggesting that deletion of C-terminal CUB domains affects albumin uptake without interfering with IF-vitamin B₁₂ endocytosis (Ovunc et al., 2011). Other biallelic variants in the C terminus of CUBN have also been found in patients with isolated proteinuria (Bedin et al., 2020). Furthermore, recent GWAS studies identified polymorphisms affecting the C-terminal CUB domains (p.A1690V, p.N2157D, p.I2984V, p.A2914V) that are associated with albuminuria (Ahluwalia et al., 2019; Boger et al., 2011; Haas et al., 2018; Teumer et al., 2016; Zanetti et al., 2019). Our results have experimentally demonstrated the crucial role of the C-terminal CUB domains of CUBN in albumin endocytosis and provide theoretical basis for these genotype–phenotype associations.

Regarding the role of AMN in endocytosis, the intracellular signal motifs of AMN have been shown to be essential for the uptake of IF-vitamin B₁₂ complex, which binds to CUB 5-8 domains of CUBN (Pedersen et al., 2010). Because IF does not bind to megalin, AMN is thought to mediate internalization of the IF-vitamin B₁₂ complex in a megalin-independent manner (Fyfe et al., 2004; Pedersen et al., 2010). In vivo, intestinal uptake of dietary vitamin B₁₂ by CUBN occurs independently of megalin (Jensen et al., 2014). However, megalin-dependent internalization of several CUBN ligands has also been reported. For example, transferrin, another CUBN ligand, is internalized by CUBN in a megalin-dependent manner (Kozyraki et al., 2001). Amsellem and colleagues demonstrated that selective inactivation of either megalin or cubilin in mice impedes formation of albumin endocytosed vesicles, and inactivation of both megalin and cubilin did not increase albuminuria, suggesting that the main role of megalin in albumin reabsorption is to drive the internalization of CUBN-albumin complexes (Amsellem et al., 2010). In our cell culture system, we found that CUBN and AMN were sufficient for albumin uptake (Fig. 2; Fig. S2). These seemingly contradictory results could be attributed to a secondary effect of megalin knockout. One of the possibilities is the decreased CUBN expression in the megalin-deficient mouse (Amsellem et al., 2010), which might inhibit megalin-independent CUBAM-mediated endocytosis of albumin. To date, the regulatory mechanisms that control the expression, localization and function of CUBN by megalin have not been well elucidated. Previously, Ahuja and coworkers demonstrated that megalin affects the intracellular turnover of CUBN (Ahuja et al., 2008). The precise function of megalin in the CUBAM-mediated endocytosis of albumin should be examined in future studies.

Although the nucleolus has been classically recognized as a site of RNA polymerase I (RNA Pol I) transcription and ribosome biogenesis, it is now recognized to have multiple significant functions in cellular homeostasis (Nunez Villacis et al., 2018). Most nucleolar proteins are highly dynamic and shuttle between the nucleolus and nucleoplasm, and their translocation regulates DNA repair, RNA Pol II transcription, telomere maintenance, the stress response and apoptosis (Iarovaia et al., 2019). Furthermore, some nucleolar proteins can be trafficked from the nuclei to the extranuclear space. One example is the cyclin-dependent kinase inhibitor p21cip, which transits through the nucleolus on its way from the nucleus to the cytoplasm (Abella et al., 2010; Iarovaia et al., 2019). P21cip has a dual function in oncogenesis, depending mainly on its intracellular localization: p21cip functions as a tumor suppressor in the nucleus and as an oncoprotein in the cytoplasm (Abella et al., 2010). In the present study, NVL translocated from the nucleolus to the extranuclear region in a CUBAM-dependent manner, and extranuclear localization of NVL was confirmed in tissues such as intestine and renal tubules where CUBN and AMN are highly expressed. The finding that NVL2 is involved in albumin uptake reveals a novel nucleolar function in regulating ligand endocytosis mediated by cell surface receptors.

Several studies have reported on NVL in relation to pathological states. Expression of NVL has been found to associate with the invasion and prognosis of prostate cancer (Zhao et al., 2016). The associations of single nucleotide polymorphisms in NVL with psychiatric disorders or diabetic retinopathy have also been reported (Pollack et al., 2019; Wang et al., 2015). However, the functional mechanisms of NVL in the pathogenesis of these diseases are unknown. Genetic animal models of NVL will be required to unravel the differential, and potentially interrelated, roles of NVL in ribosome synthesis, regulation of telomerase activity and endocytosis in vivo.

Recent integrative analyses revealed that endolysosomal function in proximal tubular cells is involved in development and progression of chronic kidney disease (Qiu et al., 2018). The receptor-mediated endocytosis of glomerular-filtered toxic substances (Kuwahara et al., 2016) or excessive albumin (Lambers Heerspink and Gansevoort, 2015; Liu et al., 2015) in renal tubular cells exerts cytotoxic effects through a wide array of intracellular signaling pathways. The cellular model described here will help identify the molecular mechanism of albumin endocytosis in physiological and diverse pathologic states.

HEK293 T (female) and HCT116 (male) cells (ATCC, Manassas, VA, USA) were grown in Dulbecco's modified Eagle's medium (DMEM) with 10% fetal bovine serum (FBS) and sodium pyruvate. Flp-In T-REx-293 cells exhibiting doxycycline (Dox)-induced expression of Strep-tagII-tagged NVL2 (wild-type or E365Q/E682Q mutant) were used as described previously (Yoshikatsu et al., 2015). Protein expression was induced by adding 1 g/ml Dox (Sigma-Aldrich, St Louis, MO, USA) 48 h before the experiments. CACO2 (male) cells (ATCC) were grown in DMEM with 10% FBS, sodium pyruvate and nonessential amino acids (Nacalai Tesque, Japan) on collagen-coated plates. Cells were grown in a humidified atmosphere with 5% CO₂ and 95% air at 37°C. We checked for contamination frequently.

Plasmid transfection in HEK293T and Flp-In T-REx-293 cells was performed using PEI-MAX (Polysciences, Warrington, PA, USA) and in HCT116 cells using Lipofectamine 2000 reagent (Invitrogen, Carlsbad, CA, USA). The sequences of siRNAs used were as follows: non-silencing control (Merck, Cat# SIC-001, Kenilworth, NJ, USA), NVL siRNA#1 (5′-CCAGGGAAAGAAUACUUCAAACAUU-3′), NVL siRNA#2 (5′-CAGACCAGUUCAAAGCUCUUGGAUU-3′) and clathrin siRNA (5′-CCGGAAAUUUGAUGUCAAUACUUCA-3′). The siRNA duplexes were transfected into HEK293T and CACO2 cells using Lipofectamine RNAiMAX (Invitrogen) according to the manufacturer's instructions. In some experiments, actinomycin D (final concentration, 5 nM) or CX-5461 (1 µM) was added to the culture medium 3 h or 1 h before the experiment, respectively.

Full-length cDNA of human NVL1, NVL2, ARH, CUBN and segments comprising the 113 aa N-terminal stretch, EGF repeats and CUB 1-8 domain of CUBN were amplified by PCR and cloned into the pCMV-tag 4A vector (Stratagene, La Jolla, CA, USA). The full-length cDNA of human ARH was amplified by PCR and cloned into pCMV-HA, an HA-tagged vector based on pCMV-tag 4A. To generate pCMV-GFP, eGFP was amplified from a pEGFP-N1 vector using primers with an XhoI site in the forward primer and an ApaI site and putative splice acceptor site in the reverse primer. pCMV-tag 4A and the amplified eGFP were digested with XhoI and ApaI. Fragments were ligated to create pCMV-GFP. The pCMV-GFP and FLAG-NVL2 were digested with NotI and SalI, and fragments were ligated to create GFP-NVL2. Amplified and digested eGFR fragments were ligated with fragments of FLAG-mini-CUBN (1-4) digested with XhoI and ApaI to generate GFP-mini-CUBN (1-4). Mutant NVL2 constructs (E365Q, E682Q, K311M and K628M) were generated by PCR-based mutagenesis. siRNA-resistant GFP-NVL2 was generated by introducing silent mutations (5′-ACGTGAGCGGATCTTGCAGACGC-3′) using PCR-based mutagenesis. Myc-DDK-AMN, FLAG-rat-mini-CUBN (1-4) and myc-GFP-AMN were as described (Udagawa et al., 2018). Myc-tagged AMN was generated by inserting a stop codon (TAA) immediately after the myc-tag sequence in the myc-DDK-AMN vector using PCR-based mutagenesis. To generate myc-tagged AMN-ΔICD, the extracellular and transmembrane domain of AMN (1-378 aa) was amplified by PCR and cloned into the pCMV6-Entry vector (OriGene, Rockville, MD, USA) with a stop codon (TAA) inserted immediately after the myc sequence using PCR-based mutagenesis. To generate an Escherichia coli expression vector of GST-tagged AMN-ICD and NVL2-NT, the intracellular domain of AMN (379-453 aa) and the N terminus of NVL2 (1-264 aa) were amplified by PCR and cloned into the pGEX-4T-1 vector (GE Healthcare, Buckinghamshire, UK). The myc-tagged, myc-DDK-tagged and myc-GFP-tagged mutant Sdel-AMNs were generated by PCR-based mutagenesis using wild-type myc-AMN, myc-DDK-AMN or myc-GFP-AMN plasmid, respectively. pCMV-tag 4A was used as control empty vector.

Rabbit polyclonal anti-NVL antibody was affinity-purified from the serum of rabbits that had been injected with recombinant NVL protein (Nagahama et al., 2004). The following antibodies and reagents were obtained from commercial sources: mouse monoclonal anti-DYKDDDDK antibody (Cat# 014-22383, Fujifilm Wako, Japan) used as anti-FLAG antibody for immunoblotting; mouse monoclonal anti-FLAG antibody (Cat# F1804, Merck, Kenilworth, NJ, USA) used as anti-FLAG antibody for assays other than immunoblotting; rat monoclonal anti-HA antibody (Cat# 11867423001, Roche, Basel, Switzerland) used for immunoblotting and immunofluorescence; rabbit polyclonal anti-HA antibody (Cat# 561, Medical & Biological Laboratories, Japan) used for PLA assays; rabbit polyclonal anti-AMN antibody (Cat# HPA000817, Atlas Antibodies, Sweden); mouse monoclonal anti-calnexin antibody (Cat# sc-46669, Santa Cruz, Dallas, TX, USA); mouse monoclonal anti-nucleolin antibody (Cat# sc-17826, Santa Cruz); rabbit polyclonal anti-β-actin antibody (Cat# 4967, Cell Signaling, Danvers, MA, USA); mouse monoclonal anti-β-catenin antibody (Cat# 610154, BD Biosciences, Japan); mouse monoclonal anti-EEA1 antibody (Cat# M176-3, Medical & Biological Laboratories, Japan); anti-FLAG antibody bead (Cat# A2220-5ML, Merck); anti-myc antibody beads (Cat# 016-26503, Fujifilm Wako); anti-HA antibody beads (Cat# 014-23081, Fujifilm Wako); protein G Sepharose (Cat# 17-0618-02, GE Healthcare); glutathione Sepharose (Cat# 17-5132-01, GE Healthcare); FITC-BSA (Cat# A9771-60MG, Merck); actinomycin D (Cat# 00851-44, Nacalai Tesque, Japan); and CX-5461 (Cat# cs-0568, ChemScene, Monmouth Junction, NJ, USA).

Co-immunoprecipitation and western blotting were performed as previously described (Udagawa et al., 2018). Cells were lysed for 10 min on ice in a lysis buffer (20 mM Tris-HCl pH 7.5, 150 mM NaCl, 1 mM EDTA and 1% NP-40) containing a protease inhibitor cocktail (Nacalai Tesque, Japan). Lysates were clarified by centrifugation and incubated with agarose beads conjugated with anti-FLAG antibody, anti-myc antibody or anti-HA antibody for 1 h at 4°C. For immunoprecipitation with anti-AMN antibody, lysates were incubated with protein G Sepharose for 1 h at 4°C. After centrifugation, the supernatants were collected and incubated with anti-AMN antibody for 1 h at 4°C and then incubated with protein G Sepharose beads for another 1 h at 4°C. For immunoprecipitation of cell surface CUBN, cells were incubated in medium containing anti-FLAG antibody (1:1000) on ice for 1 h to label FLAG-mini-CUBN on the cell surface, and lysates were incubated with protein G Sepharose for 1 h at 4°C. Beads were washed three times with lysis buffer and bound proteins were eluted with SDS sample buffer.

Cell lysates were separated by SDS-PAGE and transferred to polyvinylidene difluoride membranes. Membranes were probed by the appropriate primary antibody; horseradish peroxidase-conjugated anti-mouse IgG, anti-rabbit IgG (GE Healthcare, Buckinghamshire, UK) or anti rat-IgG (abcam, Cambridge, UK) was used as secondary antibody. Peroxidase activity was detected using the ECL or ECL-Plus system (GE Healthcare).

Immunofluorescence staining was performed as previously described (Udagawa et al., 2018). HEK293T cells were cultured on glass slides in 24-well plates (Falcon; BD Biosciences, San Jose, CA, USA) under the conditions described above. For analysis of surface expression of CUBN in nonpermeabilized samples, cells were fixed with 2% paraformaldehyde in phosphate-buffered saline (PBS) for 5 min. For analysis of protein expression in permeabilized samples, cells were fixed with 3.7% formaldehyde in PBS for 10 min and permeabilized by 0.1% Triton X-100 in PBS for 10 min. Cells were incubated with 1% BSA in PBS for 30 min, followed by incubation with primary antibody for 1 h. Antibodies conjugated with Alexa Fluor 488, 555 or 647 (Invitrogen, Carlsbad, CA, USA) were used as secondary detection antibodies. Slides were mounted in a medium with DAPI (Life Technologies, Carlsbad, CA, USA). Stained cells were visualized using a confocal microscope (LV300; Olympus, Tokyo, Japan) or a fluorescence microscope (IX71; Olympus, Tokyo, Japan). Images were processed using ImageJ (National Institutes of Health, Bethesda, MD, USA). The 3D surface plots in Fig. 1G, Fig. 4C,D and Fig. 7A were generated using ImageJ.

HEK293T cells expressing FLAG-CUBN [mini-CUBN (1-4), miniCUBN(1-8) or full-length CUBN] and myc-AMN (wild-type or Sdel) were incubated in medium containing FITC-BSA (100 ng/ml) for 1 h at 37°C. To label FLAG-CUBN on the cell surface, cells were incubated in medium containing anti-FLAG antibody (1:1000) for 1 h on ice. Cells were fixed, permeabilized and stained with the secondary antibody [anti-mouse IgG antibody with Alexa Fluor 555 for microscopy or anti-mouse IgG antibody with Alexa Fluor 647 for flow cytometry (Invitrogen)]. Endogenous β-catenin was also stained with anti-β-catenin antibody conjugated with Alexa Fluor 647 using the Zenon Antibody Labeling Kit (Thermo Fisher Scientific, Waltham, MA, USA).

CACO2 cells were incubated in medium containing FITC-BSA (100 ng/ml) for 1 h at 37°C. Cells were immediately fixed and permeabilized. Cells were also stained with anti-β-catenin antibody, followed by staining with the secondary antibody (anti-mouse IgG antibody with Alexa Fluor 555).

FITC-BSA uptake and surface CUBN staining were performed as described above. Cell gating and data collecting were performed with FACS Gallios (Beckman Coulter Life Sciences, Brea, CA, USA). Data were analyzed using Kaluza (Beckman Coulter Life Sciences).

SILAC labeling and mass spectrometry analyses were performed as previously described (Udagawa et al., 2018). Briefly, HEK293T cells cultured in medium containing amino acids labeled with heavy isotopes (L-¹³C₆¹⁵N₄-Arg and L-¹³C₆-Lys) were transfected with myc-DDK-AMN and cells in ‘light’ medium (with L-¹²C₆¹⁴N₄-Arg and L-¹²C₆-Lys) were transfected with empty vector. Cell lysates were mixed and immunoprecipitated. After enzymatic digestion of protein samples, mass spectrometry analysis was carried out using data-dependent MS/MS with a Q-FT mass spectrometer (Q-Exactive, Thermo Fisher Scientific) equipped with a nano-HPLC system (Advance UHPLC, Bruker Daltonics) and an HTC-PAL autosampler (CTC Analytics) with a trap column (0.3×5 mm; L-column ODS, Chemicals Evaluation and Research Institute, Japan). The peptides and proteins were identified and quantified using Proteome Discoverer 1.4 software (Thermo Fisher Scientific) with the SequestHT algorithm.

Proximity ligation assay (PLA) was performed using the Duolink in situ PLA kit (Olink Bioscience, Sweden) as described previously (Tsurumi et al., 2016). HEK293T cells were cultured on coverslips coated with collagen type 1, fixed and permeabilized. After washing in PBS, cells were blocked in blocking solution (1% BSA in PBS) for 30 min. After washing, samples were incubated for 1 h at 37°C with diluted primary antibodies (mouse anti-FLAG antibody and rabbit anti-AMN antibody for FLAG-NVL2 and myc- or mycGFP-AMN; mouse anti-FLAG antibody and rabbit anti-HA antibody for FLAG-ARH and HA-NVL2), followed by incubation for 1 h at 37°C with the corresponding secondary antibodies conjugated to PLA oligonucleotide probes (Duolink PLA Rabbit PLUS and PLA Mouse MINUS proximity probes). Hybridization and ligation of connector oligonucleotides, rolling-circle amplification and detection of amplified DNA products using detection reagent Red were performed using the Duolink detection reagent kit according to the manufacturer's protocol. In the indicated samples, additional immunofluorescence staining for endogenous CNX or β-catenin was performed after the PLA assay. Samples were stained with DAPI before mounting.

Human kidney and intestine samples were obtained commercially (Cosmo Bio, Japan). Immunohistological analyses were performed as previously described (Udagawa et al., 2018). Paraffin-embedded samples were deparaffinized in xylene and rehydrated through a series of graded alcohols in H₂O, followed by heat-induced epitope retrieval by incubating in a target retrieval solution (S1699; Dako, Carpinteria, CA, USA) for 20 min at 121°C. Sections were cooled to room temperature and incubated with primary antibodies, followed by incubation with secondary antibodies (Invitrogen). For co-immunostaining of AMN and NVL, samples were first stained with NVL as described, followed by staining with anti-AMN antibody with Alexa Fluor 555 conjugated using the Zenon Antibody Labeling Kit (Thermo Fisher Scientific) and subsequent fixation with 3.7% formaldehyde in PBS. Confocal fluorescent images were obtained using an Olympus LV300 confocal microscope. Images were processed using ImageJ.

GST, GST-fused AMN-ICD and GST-fused NVL2-NT peptides were expressed in BL21 E. coli. Cells were precipitated by centrifugation, lysed in sonication buffer (1 mM dithiothreitol, 1% NP-40 in Tris-HCl-buffered saline) containing protease inhibitor cocktail (Nacalai Tesque, Japan) and sonicated (30 s×4 sets) on ice. Lysates were clarified by centrifugation and incubated with glutathione Sepharose beads for 1 h at 4°C. Beads were washed twice with washing buffer (1% NP-40 in TBS) containing 0.5 mM NaCl, washed three times with washing buffer and then incubated for 1 h at 4°C with lysates of HEK293T cells expressing FLAG-NVL2 or myc-DDK-AMN. After three washes with washing buffer, bound proteins were eluted with SDS sample buffer and analyzed by western blotting.

HEK293T and HCT116 cells expressing FLAG-tagged mini-CUBN (1-4) and myc-tagged AMN were incubated in medium containing anti-FLAG antibody (1:1000) on ice for 1 h to label FLAG-mini-CUBN on the cell surface. For analysis of CUBN localization before the internalization period, cells were immediately fixed, permeabilized and stained with the secondary antibody (anti-mouse IgG antibody conjugated with Alexa Fluor 555). For analysis of CUBN localization after the internalization period, cells were incubated in PBS for 1 h at 37°C, fixed, permeabilized and stained with secondary antibody. In indicated samples, cells were also stained for endogenous β-catenin with anti-β-catenin antibody with the Alexa Fluor 488 or 647 conjugated using the Zenon Antibody Labeling Kit (Thermo Fisher Scientific). Slides were mounted in medium with DAPI (Life Technologies). Stained cells were visualized using a confocal microscope (LV300; Olympus, Tokyo, Japan). Images were processed and analyzed using ImageJ.

Immunoprecipitation was performed using anti-FLAG antibodies in lysates of HEK293T cells expressing FLAG-tagged control protein (FLAG-KLHL3) or FLAG-NVL2. Diluted NVL primary antibody was incubated overnight with the immunoprecipitates at 4°C.

mRNAs were extracted from cells using ISOGEN-LS (Nippon Gene, Tokyo, Japan) and reverse transcribed into cDNAs using SuperScript III Reverse Transcriptase (Thermo Fisher Scientific). Quantitative PCR was performed with 400 ng of cDNAs, primer sets (5′-AACTGGAGCCAGAACCGGAC-3′ and 5′-AGGAGGCACTAGGCGGAAAG-3′ for AMN, 5′-TTCACTGCCGTCTTCCAGTC-3′ and 5′-CACAGCGGAACGAGCTTCTA-3′ for CUBN, and 5′-TGCACCACCAACTGCTTAGC-3′ and 5′-GGCATGGACTGTGGTCATGAG-3′ for GAPDH) and THUNDERBIRD SYBR qPCR Mix (Toyobo, Osaka, Japan). Amplification was run in a CFX Connect thermal cycler (BioRad, Hercules, CA, USA) at 95°C for 3 min followed by 40 cycles of 95°C for 15 s and 60°C for 1 min. Melting curve analysis was performed by heating the reactions from 65 to 95°C in 0.5°C intervals while monitoring fluorescence. Data were analyzed with Bio-Rad CFX manager 3.1 (BioRad).

ImageJ was used to quantify area size. Results are presented as mean±standard error of the mean (s.e.m.). The numbers of samples is given in the figure legends. Significant differences between means of two sample groups were evaluated using a two-sided t-test, with significance indicated for P<0.05.

We appreciate Dr Yasuhiro Yoshida, Department of Pediatrics, Graduate School of Medicine, University of Tokyo, for constructing the FLAG-NVL2 expression vector. We thank Life Sciences Core Facility, Graduate School of Medicine, University of Tokyo for supporting the flow cytometry analysis. We thank Edanz Group (www.edanzediting.com/ac) for editing a draft of the manuscript.

The peer review history is available online at https://jcs.biologists.org/lookup/doi/10.1242/jcs.242859.reviewer-comments.pdf