RILP Restricts Insulin Secretion Through Mediating Lysosomal Degradation of Proinsulin

Insulin secretion is a unique characteristic of pancreatic β-cells. The maturation of the insulin granule consists of a series of vesicular trafficking stages (1). Insulin biogenesis initiates with the synthesis of preproinsulin in the rough endoplasmic reticulum (ER), which is converted to proinsulin. Proinsulin is then transported to the Golgi apparatus and packaged rapidly into immature secretory granules (iSGs) at the trans-Golgi network (TGN) (2). Proinsulin in the iSGs undergoes proteolytic cleavage to produce insulin and C-peptide (3). Insulin is condensed/crystallized with zinc and calcium, resulting in the formation of the mature dense-core granules, which are ready for the regulated release of insulin upon stimulation (4).

The trafficking/sorting of the insulin granule is controlled by the membrane trafficking machineries. The Ras-like Rab guanosine-5′-triphosphateases are the master regulators for membrane trafficking, mediating vesicle budding, translocation, docking, and fusion events through interaction with the downstream effectors (5–7). Several Rab proteins have been proven to regulate the exocytosis of insulin secretory granules. Rab3 associates with the insulin granule and interacts with multiple factors to regulate the exocytosis of insulin granules by mediating granule docking to the plasma membrane (8). Rab27 regulates insulin granule docking and priming (9–11). In addition, Rab27 may mediate the movement of the granule along the actin filament (10,12), and its mutation caused diabetes in mice (13,14). Rab11 interacts with Rip11 to regulate insulin granule exocytosis in pancreatic β-cells (15). Rab2A mediates either insulin secretion or ER-associated degradation of proinsulin in β-cells (16).

RILP (Rab7 interacting lysosomal protein) serves as a downstream effector for Rab7 and is responsible for stabilizing Rab7 on the endosomal membrane and controlling microtubule minus-end-directed trafficking of vesicles via the dynein/dynactin complex (17,18). Rab7 and RILP both play important roles in the endocytic trafficking and tumorigenesis (19).

In contrast to the studies on RILP in regulating endocytosis and retrograde trafficking, few studies indicated that RILP participates in exocytosis. RILP mediates retrograde transport of secretory granules in mast cells (20) and induces perinuclear aggregation of melanosome in melanocytes (21,22). RILP and its cleaved RILP fragment (cRILP) mediate hepatitis C virus secretion (23,24). These studies suggest that RILP may be a new factor regulating the exocytosis of secretory vesicles.

Here, we investigated the function of RILP in insulin secretion in pancreatic β-cells. We found that RILP inhibited insulin secretion by regulating lysosomal degradation of proinsulin, uncovering a novel function of RILP in regulating insulin biogenesis and secretion.

INS-1, Min6, 293A, and 293t cell lines were from ATCC and maintained in the 5% CO₂ incubator at 37°C. INS-1 cells were cultured in RPMI 1640 medium supplemented with 10% FBS, 10 mmol/L HEPES, 1 mmol/L PyrNa, and 50 μmol/L β-mercaptoethanol. Min6 cells were cultured in DMEM supplemented with 10% FBS and 50 μmol/L β-mercaptoethanol. The transient transfection of plasmids into the cells was performed by using Lipofactamine 2000 reagents (Invitrogen) according to the manufacturer’s instructions. For the inhibitor treatments, cells were cultured in the media containing 20 mmol/L NH₄Cl, 100 µmol/L chloroquine, or 40 µmol/L MG132, respectively.

Guinea pig polyclonal antibodies recognizing total insulin (preproinsulin, proinsulin, and insulin, cat. no. ab7842) and rabbit polyclonal antibody for chromogranin A (cat. no. ab15160) were purchased from Abcam. Monoclonal antibody (mAb) specifically for proinsulin was from HyTest (cat. no. CCI-17). mAb specifically for proinsulin (cat. no. 8138 s) in Western blot and rabbit polyclonal antibody against insulin (cat. no. 4590 s) were purchased from Cell Signaling Technology. Rabbit polyclonal antibody against RILP was generated by GenScript (Nanjin, China). mAb against Rab7 was purchased from Sigma-Aldrich (cat. no. R8779), and mAb against GAPDH was from Proteintech. mAb for Lamp1 was obtained from Developmental Studies Hybridoma Bank. Cy5-conjugated goat anti-mouse IgG was from Thermo Fisher (Waltham, MA). Alexa Fluor 555–conjugated goat anti-rabbit/mouse IgG was from Thermo Fisher Scientific. Alexa Fluor 488–conjugated goat anti-guinea pig IgG was from Abcam (cat. no. ab150185). Streptozotocin (STZ), MG132, and chloroquine were purchased from Sigma-Aldrich.

Cell penetrating peptides (CPPs) were synthesized by GenScript, including CPP-RILP-1 (251–262aa: CPP-ILQERRNELKANV), CPP-RILP-2 (297–307aa: CPP-QRRKIKAKMLG), and CPP-control (CPP-PGHQHGQEPEWA). The CPP (YGRKKRRQRRR) was added to the N-terminus of all peptides. All peptides were acetylated at their N-termini and were amidated at their C-termini.

All oligonucleotides are listed in Supplementary Table 1.

The plasmids of green fluorescent protein (GFP)-RILP, GFP-RILP (199–401), and GFP-RILPm (304–306AAA) were described previously (25). mCherry-RILP was constructed by subcloning RILP cDNA into the pmCherry vector. GFP-Rab26WT was constructed by cloning Rab26 cDNA retrieved from mouse cDNA into the pEGFP-C1 vector. Rab26Q123L or Rab26T77N mutant was generated by the PCR-directed mutagenesis approach. myc-Rab26WT, myc-Rab26Q123L, and myc-Rab26T77N were generated by subcloning the correspondent cDNA into the pDmyc vector.

Adenovirus was prepared using the AdEasy system for generating recombinant adenovirus (26). Briefly, RILP or RILPm was cloned into the pAdTrack- cytomegalovirus vector and then linearized. The linearized plasmids were transformed into the competent AdEasier cells to generate recombinant adenovirus plasmids. The recombinant adenovirus plasmids were transfected into 293A cells to produce recombinant adenovirus (referred to as Ad-RILP or Ad-RILPm below).

For adenovirus-mediated gene knockdown, shRNA-RILP, shRNA-Rab7, and shRNA-Rab26 were cloned into pAdTrack-H1-U6 vector, respectively. The recombinant virus was prepared as described above. The shRNA sequences are listed in Supplementary Table 2.

RILP-deficient INS-1 cells were generated using the CRISPR/Cas9 system, as previously described (27). A single-guide RNA sequence (5-GCCACTAGTAGTGCGGGCGC-3) was used for disrupting the expression of RILP in INS-1 cells. RILP knockout (KO) cell lines were generated as described (27). The disruption of RILP gene was verified by genomic DNA sequencing and Western blot.

For insulin secretion in Min6 and INS-1 cells, cells were preincubated for 60 min with Krebs-Ringer bicarbonate HEPES (KRBH) buffer (containing 114 mmol/L NaCl, 4.7 mmol/L KCl, 1.2 mmol/L KH₂PO₄, 1.16 mmol/L MgSO₄, 0.5 mmol/L MgCl₂, 2.5 mmol/L CaCl₂, 0.25% BSA, and 20 mmol/L HEPES; pH 7.4) containing 2.8 mmol/L glucose, followed by incubation in 2.0 mL stimulation medium (KRBH containing 16.7 mmol/L glucose). At each time point (0–40 min), 2% of incubation medium was collected for insulin content measurement by ELISA kits (ImmunoDiagnostics Ltd, Hong Kong, China). For insulin secretion in vitro, mouse islets were isolated and cultured as previously described (28). Briefly, the pancreas was digested with collagenase V, and then the islets were handpicked under a dissection microscope. The freshly isolated mouse islets were cultured in RPMI medium containing 10% FBS and 2 mmol/L glutamine for 24 h and then infected by Ad-RILP or Ad-vector adenovirus for another 24 h. As described above, 100 islets were processed for the detection of insulin secretion. For the detection of insulin in serum, blood samples were collected and centrifuged at 2,000 rpm to obtain serum. The serum insulin concentration was detected by ELISA assay as described above.

A type 1 diabetic mouse model was generated, as previously described (29), and the mice received transplants of the freshly isolated islets. Approximately 200 islets infected by Ad-RILP or Ad-vector were suspended in 20 µL RPMI medium containing 5% FBS solution and then transplanted into the mice by injecting the islets between the capsule and renal parenchyma of the kidney, as previously described (30). To generate the type 2 diabetic rat model, male SD rats (160–180 g) were fed a high-fat diet (HFD: 60% fat) for 8 weeks, followed with a single intraperitoneal injection of low-dose STZ, as previously described (31). The rats with a nonfasting plasma glucose level of ≥16.7 mmol/L were determined as diabetic. BKS db/db diabetic mice were obtained from Model Animals Research Center of Nanjing University (no. T002407, BKS-lepr^em2Cd479/Nju) on a C57BLKS/J background.

Intraperitoneal glucose tolerance tests (IPGTTs) were performed by intraperitoneal injection of 2.0 g/kg glucose into mice or rats after fasting for 16 h. Glucose levels were measured with an automatic glucometer (Accu-Chek; Roche Diagnostics).

All animal experiments were performed according to the guidelines for the care of laboratory animals in strict compliance with the regulations of the Xiamen University Institutional Animal Ethics Committee.

INS-1 cells infected with Ad-vector or Ad-RILP for 48 h were processed for transmission electron microscopy (TEM) analysis using a Tecnai Spirit (T12) TEM (Thermo Fisher Scientific), as previously described (32).

Immunofluorescence microscopy was performed as previously described (33). Briefly, cells grown on coverslips were fixed with 3% paraformaldehyde, permeabilized, and then immunostained using the primary antibodies, followed by fluorophore-conjugated secondary antibodies. Mouse islets were embedded in 20% (w/v) gelatin (34) and then fixed in 4% paraformaldehyde on ice for 20 min, followed by three washes with PBS and three equilibrations in 30% sucrose/PBS for 3 h. The islets were cryosectioned, and the cryosections were permeabilized with 0.2% Triton X-100 and then blocked with 0.2% BSA and immunostained with primary antibodies, followed with secondary antibodies. The immunolabeled cells or tissues were analyzed with a Carl Zeiss LSM5 EXITER or Leica TCS SP8 STED laser scanning confocal microscope.

Min6 or INS-1 cells expressing the indicated plasmids were lysed in RIPA buffer to generate cell lysates for Western blot analysis. For the examination of RILP-Rab26 interaction, 293t cells were cotransfected with GFP-RILP and myc-Rab26WT, Rab26Q123L, or Rab26T77N, respectively. The resulting cell lysates were immunoprecipitated with anti-myc tag antibody (35). The precipitated proteins were detected by Western blot with the indicated antibodies. Western blot assay was performed as described (35).

RNA was extracted using the Trizol RNA purification system (Invitrogen), and 1.0 μg mRNA was reverse transcribed into cDNA using the PrimeScript RT Reagent Kit (TaKaRa). The expression of RILP transcript was analyzed by PCR or quantitative PCR using specific primers according to the manufacturer’s protocol.

The data are presented as the mean ± SD with GraphPad Prism 7.0. Statistical significance between two groups was assessed using Student t tests or among multiple groups using two-way ANOVA. P < 0.05 was indicated as statistical significance.

Data and resources are available from the corresponding authors.

RILP regulates the biogenesis and retrograde trafficking of the late endosomes/lysosomes and the lysosomal related organelles, such as secretory granules (20,22). To explore the role of RILP in the exocytosis of the insulin granules and insulin secretion in pancreatic β-cells, RILP was overexpressed in insulin-secreting β-cell lines Min6 or INS-1 through the adenovirus-mediated expression system, which gives high infection efficiency (>90% efficiency revealed by GFP) (Fig. 1A) and high expression level of RILP in β-cells (Fig. 1B). Insulin secretion was stimulated by 16.7 mmol/L glucose. Insulin secreted into the media was examined by ELISA. Figure 1C demonstrates the amount of secreted insulin of Min6 cells expressing Ad-RILP was significantly decreased. A similar result was obtained from INS-1 cells expressing Ad-RILP (Fig. 1D), indicating RILP overexpression inhibits insulin secretion in the pancreatic β-cells.

To test RILP on insulin secretion under a more physiological setting, freshly isolated islets from mice were efficiently infected with Ad-RILP and cultured in media (Fig. 2A and Supplementary Fig. 1A–C). Insulin secretion was stimulated by 16.7 mmol/L glucose in KRBH buffer, and the amount of insulin secreted was detected by ELISA at different time points as indicated in Fig. 2B. Again, the expression of RILP significantly inhibited insulin secretion in islets.

We next examined whether RILP influences the function of islets in vivo. A type 1 diabetic mouse model was generated by injecting STZ into the pancreas to destroy the islets (Fig. 2C). The disruption of islets resulted in the increase of the plasma glucose level (Supplementary Fig. 2A). The in vitro manipulated islets were transplanted back beneath the renal capsule of these diabetic mice. The islets beneath the renal capsule were observed to have normal function, as indicated by the expression of insulin (Fig. 2D). Physiological investigations revealed that the amount of insulin in the plasma was greatly decreased in mice transplanted with the Ad-RILP islets after feeding compared with that of Ad-vector (Fig. 2E). Furthermore, we monitored the concentration of blood glucose during 4 h after the intraperitoneal glucose injection and found the mice transplanted with Ad-RILP islets kept relatively high level of glucose compared with the control mice (Fig. 2F). The hypoglycemic effect of islets transplantation was monitored for 6 days. The results demonstrated that the blood glucose level of mice transplanted with Ad-RILP islets was maintained at >22 mmol/L, while the level of blood glucose in control mice was lowered to ∼10 mmol/L (Fig. 2G).

Our results revealed that overexpression of RILP in the islets resulted in their failure to rescue insulin secretion to recover glucose homeostasis in type 1 diabetic mice (Fig. 2E and G), consistent with that of in vitro experiments, suggesting that upregulation of RILP inhibits insulin secretion and probably results in the abnormal glucose homeostasis in mice.

Since RILP inhibits insulin secretion in β-cells, we tested whether there are alternations in the expression of RILP in animals suffering from diabetes. Thus, we generated the type 2 diabetic rat model by HFD feeding for 10 weeks, combined with injection of a low dose of STZ in the last 2 weeks. The body weight and insulin level increased in HFD-fed rats but dropped upon treatment with STZ (Fig. 3A and B), but the blood glucose was dramatically increased (Fig. 3C). In IPGTT experiments, the glucose tolerance was significantly impaired in HFD-fed rats compared with chow-fed rats during the GTT (Fig. 3D). The mRNA expression level of RILP was significantly elevated in islets from HFD-fed rats compared with chow-fed rats (Fig. 3E and F).

BKS db/db diabetic mice were used to examine the upregulation of RILP. The levels of glucose and insulin are maintained at higher levels in db/db mice than that of wild-type (wt)/wt mice in IPGTT experiments under normal feeding or the HFD condition (Fig. 3G and H) in addition to the higher body weight of db/db mice (Fig. 3I). Consistent with the results from the type 2 diabetic rats, the mRNA level of RILP was significantly upregulated in the db/db mice, especially under the HFD-fed condition (Fig. 3K and Supplementary Fig. 2B), suggesting that the expression of RILP is upregulated in the experimental diabetic rats or mice.

To test the effects of RILP on the distribution of insulin granules, INS-1 cells were transiently transfected with mCherry-RILP and immunolabeled with insulin/proinsulin antibody. Expression of RILP resulted in the perinuclear clustering of insulin granules in most of the INS-1 cells (>80%) in the normal culture stimulation condition (Fig. 4A) compared with the cells expressing vector (Supplementary Fig. 3A). However, a high concentration of glucose induces insulin granule translocation to the plasma membrane (indicated by arrow in Fig. 4A and Supplementary Fig. 3B) in nontransfected cells. This phenomenon was confirmed by transfection with GFP-RILP and marked by insulin/proinsulin (Supplementary Fig. 3B) or chromogranin A (Supplementary Fig. 3C), a prohormone cosorted with proinsulin (28).

Surprisingly, when cells were labeled with the antibody specifically recognizing proinsulin, the proinsulin-containing granules in most of the INS-1 cells (>80%) were greatly reduced under either lower or higher glucose conditions upon RILP overexpression (Fig. 4B) compared with that of vector expression (Supplementary Fig. 3D), regardless of granule clustering. As proinsulin is an iSG marker, the results suggest that expression of RILP results in the decrease of insulin iSGs, which probably causes the decrease of insulin release. As the insulin secretion was also inhibited in islets, we examined the insulin/proinsulin staining in the mouse islets and found that (pro)insulin signals were also reduced in the islets infected by Ad-RILP (Fig. 4C and D), while there was no major alternation in the islets infected by Ad-vector.

To verify the decreased number of secretory granules caused by the expression of RILP, we processed TEM for the Ad-RILP– or Ad-vector–infected INS-1 cells (Fig. 5A). Insulin granules appear as dense-core vesicles. The observation from different sections (indicated by nuclei of different sizes) revealed that cells expressing RILP have fewer secretory granules compared with the control cells. Further quantitative analysis demonstrated that mature secretory granules (mSGs) and iSGs are both decreased in the cells infected by Ad-RILP (Fig. 5B and C).

Taken together, these results suggest that RILP not only causes the retrograde movement of insulin granules to cluster but also regulates the biogenesis or degradation of iSGs, thus inhibiting the insulin secretion in the pancreatic β-cells.

Expression of RILP has no significant effects on the transcription of the insulin gene (Supplementary Fig. 4A and B); therefore, we tested whether RILP regulates the degradation of immature granules by examining the degradation of proinsulin, a marker for immature insulin granules.

INS-1 cells were infected by Ad-RILP or Ad-vector and stimulated with a high concentration of glucose (16.7 mmol/L), and proinsulin in the cells was detected by Western blot. The results showed that overexpression of RILP greatly decreased the amount of proinsulin (Fig. 6A and B), suggesting RILP actually promotes proinsulin degradation. We next examined the proinsulin degradation in INS-1 cells under different culture conditions (Fig. 6C and D). RILP promotes proinsulin degradation in Ad-RILP–infected INS-1 cells when cultured in the media containing a low (2.8 mmol/L) or high (16.7 mmol/L) concentration of glucose, even in KRBH buffer containing 16.7 mmol/L glucose, referred to as glucose-stimulated insulin secretion (GSIS) below. Further examinations for insulin secreted into the media/buffer demonstrated that Ad-RILP caused a significant decrease of insulin secretion accompanied with proinsulin degradation (Fig. 6E). These observations suggest that RILP may inhibit insulin secretion through promoting proinsulin degradation.

To complement our results so far based on a gain-of-function approach through RILP overexpression, we next used adenovirus-mediated expression of shRNA-RILP (Ad-shRNA-RILP) to deplete endogenous RILP in Min6 cells (Supplementary Fig. 4C). As shown in Fig. 6F and G, both shRNAs targeting to different regions of RILP mRNA can slightly increase the amount of proinsulin. Correspondingly, the insulin released into the KRBH buffer under the GSIS condition was increased significantly (Fig. 6H). To further confirm the role of RILP on insulin secretion, we knocked out RILP in INS-1 cells using PX458-mediated Cas9. As shown in Fig. 6I and J, two different RILP-KO cell lines exhibited the increased amount of proinsulin compared with the control cells, and insulin secretion was increased under GSIS condition in the KO cells (Fig. 6K).

These results taken together suggest a novel function of RILP in regulating GSIS through promoting the degradation of proinsulin.

As RILP regulates the late endosomal/lysosomal trafficking, RILP may regulate the lysosomal degradation of proinsulin in the immature insulin granules in the pancreatic β-cells. To test this hypothesis, we examined the effects of lysosomal (chloroquine and NH₄Cl) and proteasomal inhibitors (MG132) on proinsulin degradation in INS-1 cells. INS-1 cells were infected with Ad-RILP or Ad-vector and then incubated with the media containing chloroquine or NH₄Cl, and proinsulin was detected by Western blot. Interestingly, the RILP-caused decrease of proinsulin was reversed to the normal level after the treatments with chloroquine or NH₄Cl (Fig. 7A and B). However, a proteasomal inhibitor MG132 did not reverse the degradation of proinsulin induced by RILP (Fig. 7C and D). Furthermore, the lysosomal inhibitor chloroquine can recover the insulin secretion, which should be inhibited by RILP under the 2.8 mmol/L glucose GSIS condition (Fig. 7E) or 16.7 mmol/L glucose GSIS condition (Fig. 7F), but insulin secretion was still inhibited significantly by RILP upon MG132 treatment.

Immunofluorescent microscopy revealed that a small pool of insulin granules is associated with the Lamp1-marked lysosomes in vector-transfected INS-1 cells (Supplementary Fig. 5A). However, the insulin granules are in close contact with the clustered lysosomes in RILP-expressing cells, and few insulin puncta colocalized with Lamp1, which is probably due to the degradation of insulin/proinsulin in the lysosomes (Fig. 7G). Inhibition of lysosomal function by NH₄Cl treatment increased the colocalization of RILP/insulin/Lamp1 in either RILP-transfected cells (Fig. 7H) or vector-transfected cells (Supplementary Fig. 5B). In addition, GFP-RILP (199–401aa), a truncated form not inducing lysosomes clustering, colocalized well with insulin/proinsulin (Supplementary Fig. 5C). The results indicate that insulin granules can associate with the lysosomes and RILP-containing compartments.

Additional experiments demonstrated that RILP-mediated proinsulin degradation does not require autophagic function and that overexpression of RILP has no major effects on autophagy (Supplementary Fig. 6). These findings suggest that RILP regulates insulin secretion through promoting lysosomal degradation of proinsulin.

Rab7-RILP interaction plays a central role in the endocytic degradative pathway (17,18). Therefore, we examined whether Rab7 is required for RILP-mediated proinsulin degradation and insulin secretion.

Interestingly, expression of Rab7 alone did not have significant effects on either proinsulin degradation or insulin secretion (data not shown). However, adenovirus-mediated expression of shRNA-Rab7, which efficiently knocked down Rab7 in Min6 cells (Supplementary Fig. 4D), sufficiently inhibited the proinsulin degradation induced by the expression of Ad-RILP in Min6 cells (Fig. 8A and B). Consistently, Rab7 knockdown resulted in a significant increase of insulin secretion in Min6 cells infected with Ad-RILP under the GSIS condition (Fig. 8C), indicating that RILP inhibits insulin secretion through cooperation with endogenous Rab7 in β-cells.

The carboxyl terminus of RILP interacts with Rab7; therefore, we examined whether RILP (304–305AAA) (referred as RILPm), a mutant form of RILP not interacting with Rab7 (25), affects proinsulin degradation and insulin secretion. Ad-RILPm has no major effects on proinsulin degradation (Fig. 8D and E) and no significant effects on insulin secretion in INS-1 cells (Fig. 8F), while Ad-RILP markedly promoted insulin degradation and significantly inhibited insulin secretion under the GSIS condition. Immunofluorescence microscopy showed that proinsulin granules were not obviously altered by the expression of RILPm (Fig. 8G), further supporting the notion that RILP suppresses insulin secretion via promoting proinsulin degradation through its ability to interact with Rab7.

To further verify that RILP-Rab7 interaction is required for mediating proinsulin degradation, CPPs were used to deliver peptides capable of interfering with the RILP-Rab7 interaction. As shown in Fig. 8H and I, both RILP-targeting CPPs (CPP-ILQERRNELKANV and CPP-QRRKIKAKMLG) derived from the regions crucial for the Rab7-RILP interaction increased the amount of proinsulin in INS-1 cells infected with Ad-RILP in a dose-dependent manner. Collectively, these three lines of experiments—Rab7 knockdown, RILP mutant, and inhibitory peptides—established that RILP inhibits insulin secretion through promoting the lysosomal degradation of proinsulin in a Rab7-dependent manner.

Although we defined the RILP-Rab7 interaction as essential for promoting lysosomal degradation of proinsulin, we were curious about the mechanism of how RILP directs insulin granules to the lysosomal degradation pathway. We systematically screened Rab proteins interacting with RILP and found that several other Rab proteins potentially interact with RILP (Supplementary Fig. 7A). Since Rab26 has been demonstrated to participate in amylase release and play a role the biogenesis of the gastric zymogenic granule (36,37), we examined whether the Rab26-RILP interaction is involved in insulin secretion in β-cells. Specifically, the wild-type Rab26WT and constitutive active mutant Rab26Q123L, but not the dominant negative mutant Rab26T77N, interacted with RILP (Fig. 9A). RILP induced clustering of the Rab26-containing compartments at the perinuclear region (Fig. 9B). The expressed Rab26 was associated with insulin granules and induced enlarged insulin granules (Fig. 9C). Depletion of Rab26 by shRNA (Supplementary Fig. 7B) greatly increased the amount of insulin released into the media in the Ad-RILP–expressing Min6 cells (Fig. 9D), suggesting Rab26-RILP interaction is important for regulating insulin secretion. Thus, we propose that RILP interacts with insulin granule–associated Rab26 to direct insulin granules to the lysosomal degradation and consequently restricts insulin secretion (Fig. 9E).

RILP is emerging as a master integrator of endosomal trafficking through the interaction with multiple Rab proteins, such as Rab7, Rab34, Rab36, and Rab24 (38–41). Although RILP regulates melanosome exocytosis (22), its role in the secretory pathway remains elusive. In this study, we provide several lines of evidence obtained from gain-of-function as well as loss-of-function approaches in both β-cell lines and freshly isolated islets to demonstrate that RILP inhibits insulin secretion, suggesting a novel function of RILP in regulating insulin secretion, insulin granule biogenesis, and glucose homeostasis. Mechanistically, RILP regulates insulin secretion through mediating lysosomal degradation of proinsulin in a Rab7-dependent manner. We also showed that Rab26 associates with insulin granules and that RILP interacts with Rab26 in a manner dependent on Rab26’s guanine nucleotide–binding activities. Moreover, Rab26 knockdown greatly promoted insulin secretion in the Ad-RILP–expressing β-cells. These results suggest RILP binds to the insulin granule–associated Rab26 to direct insulin granules to the lysosomal degradation and consequently restricts insulin secretion.

In addition, our study also revealed that RILP was upregulated in the experimental diabetic rats or mice. Overexpression of RILP in freshly isolated islets rendered them incapable of rescuing diabetic phenotypes in the type 1 diabetic mouse model. These results imply a potential pathophysiological function of RILP in metabolic diseases such as diabetes.

To maintain homeostasis, the pancreatic β-cells will degrade excess insulin (proinsulin) in response to nutrient alternation. Autophagy is one of the major mechanisms for insulin granule degradation (42). However, other degradation machineries are engaged in insulin homeostasis. The insulin secretory granules can be digested via crinophagy (lysosomal degradation) when the islets are exposed to low or even physiological glucose concentrations (43–45). β-Cells use starvation-induced nascent granule degradation (SINGD) instead of autophagy during fasting to lower insulin release, which seems to occur independently of macroautophagy (46). When crinophagy or SINGD was largely suppressed in the absence of canonical autophagy, the Golgi membrane–associated degradation pathway is the compensatory pathway for the granule degradation (16,47). Crinophagy is a degradation pathway in which the insulin granules directly fuse with the lysosomes, which is also the underlying mechanism for SINGD. Our study revealed that RILP mediates the lysosomal degradation of proinsulin through interacting with insulin granule–associated Rab26 and lysosome-located Rab7, not requiring autophagy, suggesting RILP-mediated proinsulin degradation uses crinophagy machinery.

The trafficking of insulin granules consists of three stages: concentrating and packaging of proinsulin at the TGN (sorting for entry), followed by budding of the immature secretory granules; the maturation of insulin granules by “sorting by retention” and “sorting for exit” to allow the refinement and condensation of granule contents (3,4); and finally, the matured insulin granules dock to and fuse with the plasma membrane, which is mediated by SNARE proteins and Rab proteins (3,48). Protein kinase D is a crucial regulator in the formation of secretory vesicles from TGN (49) and also regulated the biogenesis of insulin granules and insulin secretion in β-cells (46,50,51). In light of RILP mediating proinsulin degradation in our study, RILP may be involved in the maturation of insulin granules via fine-tuning the level of iSGs, but not in the formation of granules. Also, the TEM data showed that overexpression of RILP caused a reduction in the number of secretory granules.

Our results suggest that RILP inhibits insulin secretion by promoting proinsulin degradation. However, these results do not exclude the possibility that RILP inhibits insulin granule exocytosis through mediating retrograde transport as described in melanocytes (21,22). We propose that nutrient alternation will upregulate the expression of RILP, and RILP will then inhibit the anterograde trafficking of insulin granules, resulting in the clustering of insulin/proinsulin granules; subsequently, RILP will drive the accumulated immature granules for lysosomal degradation (Fig. 9E). In this model, we also characterized RILP interacting with Rab26 to direct insulin granules for lysosomal degradation through interacting with lysosome-associated Rab7. The interaction of RILP with both Rab26 (on granules) and Rab7 (on endosomes and lysosomes) may promote the membrane fusion between the insulin granules and lysosomes. This study provides a novel role of the lysosomal degradation pathway in maintaining insulin homeostasis and offers an additional mechanistic basis for crinophagy.

Insulin is the most important hormone involved in maintaining glucose homeostasis. Insulin secretion disorder is believed to cause diabetes (52,53). The proteasomal degradation of the misfolded proinsulin at the ER can prevent diabetes (54,55). Although the degradation of insulin granules may result in the decrease of insulin release (39), there is no direct evidence linking insulin degradation to diabetes. Our study also shows that RILP is upregulated in diabetic rats or mice, suggesting the clinical significance of RILP in diabetes as well as RILP and the RILP-mediated trafficking pathway as potential therapeutic targets.

Acknowledgments. The authors are grateful to Tao Xu (College of Life Sciences, University of Chinese Academy of Sciences, Beijing, China) for providing the INS-1 cells.

Funding. This work was supported by National Natural Science Foundation of China (no. 31671478 and no. 31871423).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. Y.Zho. conducted most of the experiments and analyzed the results. Z.L. and S.Z. conducted the animal experiments. R.Z., H.L., X.L., X.Q., and Y.Zhe. provided technical support. M.Z. carried out work on Rab26-RILP interaction. L.L. provided technical support for the pancreas sectioning. W.H. was involved with the ideas and writing the paper. T.W. conceived the idea for the project, designed the experiments, and wrote the paper. T.W. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.