PGRN deficiency leads to accumulation of autophagosomes in OVA-challenged PGRN null GD model
We previously reported that PGRN deficient (PGRN KO) mice developed a typical GD cellular phenotype following ovalbumin (OVA) challenge, as evidenced by enlarged macrophages reminiscent of classic Gaucher-like cells and tubular lysosomes [14]. Leveraging this GD animal model (Fig. 1A), we first examined the levels of autophagy-associated molecules. One of the typical autophagosome marker is LC3, a microtubule-associated protein existing as two isoforms: LC3-I and LC3-II. A post-translational modification converts LC3-I into LC3-II, which is specifically associated with autophagosome membranes and is widely used as an autophagic marker [35, 36]. The level of LC3-II in PGRN KO OVA mice was significantly higher than that in WT OVA mice, suggesting the accumulation of autophagosomes in PGRN KO OVA mice (Fig. 1B, C).
Autophagy-related proteins 5 and 12 (ATG5 and ATG12) are two proteins that form a 50 kDa complex which are also involved in the formation of the autophagosome. Therefore, this dimer can be used as a detector of the presence of autophagosomes in an advanced stage of maturation of autophagosome and its accumulation is directly proportional to the accumulation of non-degraded autophagosomes [37, 38]. As shown in Fig. 1B and D, the two proteins can be observed as a complex (50 kDa) and are accumulated in tissues in PGRN KO OVA mice as compared to WT OVA mice. In addition, immunofluorescence staining of lung tissues from WT and PGRN KO mice challenged with OVA revealed much higher levels of LC3-II and ATG5 in PGRN KO OVA lung than in that in WT control as well (Fig. 1F, G).
During autophagy, damaged or misfolded proteins are ubiquitinated, and then colocalized with p62-SQSTM1 and delivered to the proteasome for degradation [39]. p62 binds to LC3-II and acts as a bridge between the substrate and the inner membrane of the autophagosome [40, 41]. p62 level is directly proportional to blockade of the autophagic flux since p62 is degraded with the cargo when the autophagic flux is completed, and its accumulation indicates the absence of autophagosome degradation [42]. We found that the expression level of p62 was markedly higher in tissues of PGRN KO OVA mice as compared to WT OVA mice (Fig. 1B, E). Electron microscopy (EM) images of p62 immunogold labeling in Gaucher cells derived from murine model lung tissues exihibited significant accumulation of p62 in PGRN KO OVA mice as well, indicative of defective autophagic flux in PGRN KO OVA mice relative to WT counterparts (Fig. 1H, I).
Progranulin deficiency aggravates the GD phenotype in GD patient fibroblasts
Coordination between the autophagic and lysosomal degradation pathways is critical for the cellular turnover of the proteins and organelles [2]. Accordingly, we next examined the effect of PGRN deficiency upon lysosomal storage in GD patient fibroblasts. PGRN was ablated in GD type 2 patient fibroblasts (L444P) using CRISPR-Cas9 technology (Fig. 2A) and the knockout efficiency was confirmed by western blotting (Fig. 2B). As shown in Fig. 2C and D, fluorescence signal intensity, indicative of lysosomal storage content, was significantly higher in L444P/PGRN KO relative to L444P control cells, especially following lipid stimulation. We previously reported that PGRN binds directly to GCase, functioning as an indispensable adaptor for the formation of a complex between Hsp70 and GCase/LIMP2, which contributes to appropriate lysosomal localization of defective GCase in GD [14]. Consistent with our previous report [14], GCase activity was dramatically decreased in L444P/PGRN KO, as measured by fluorescence intensity of released 4-methylumbelliferone (Fig. 2E). In addtion, β-glucosylceramide (β-GlcCer), the substrate of GCase, was further accumulated in L444P/PGRN KO fibroblasts (Fig. 2F).
PGRN deficiency causes autophagosome accumulation in GD patient fibroblasts
Given that autophagosome markers, such as LC3-II, ATG5-ATG12 complex and p62, accumulate in PGRN KO mice with OVA challenge (Fig. 1), we next employed western blotting to measure these molecular markers of autophagy in L444P GD patient fibroblasts in the presence or absence of PGRN. As shown in Fig. 3A-D, the LC3-II and ATG5-ATG12 complex levels were significantly increased in L444P/PGRN KO in comparison with L444P cells. Elevated LC3-II level could result from early stage initiation of autophagy or by blockade of autophagic flux at late stage. Notably, p62, one of the molecular markers of autophagic flux, was also significantly increased in L444P/PGRN KO relative to control L444P fibroblasts, which strongly suggested accumulation of autophagic substrates. Moreover, immunofluorescence staining supported western blotting results (Fig. 3E-H). In summary, the increased levels of LC3-II and p62 indicated that late stage of autophagy was defective in PGRN deficient L444P patient fibroblasts.
PGRN deficiency impairs autophagosome-lysosome fusion
The neoformed autophagosome goes through two stages of maturation before fusion, and EM could be employed to distinguish the double-membrane of an immature autophagosome from the single-membrane of the lysosome-fused late autophagosome [43]. EM images acquired from WT OVA murine macrophages highlighted the predominance of singe-membraned autophagic vacuoles, while macrophages isolated from PGRN KO OVA mice highlighted predominance of double-membraned initial autophagic vacuoles (Fig. 4A). These data demonstrated that the deficiency of PGRN resulted in defects in late stage autophagy. Accorrdingly, visualization of L444P and L444P/PGRN KO cells using EM revealed a greater abundance of autophagosome initial vacuole in L444P/PGRN KO compared with L444P (Fig. 4B, C).
To further confirm the association between PGRN deficiency and a defect of autophagosome-lysosome fusion, the subcellular localizations of the autophagic marker LC3 and the lysosomal protein LIMP2 were investigated with confocal microscopy in L444P/PGRN KO fibroblasts [44]. As shown in Fig. 4D and E, co-localization of LC3 and LIMP2 was significantly reduced in L444P/PGRN KO compared with L444P fibroblasts. PGRN KO C28I2 chondrocytes (Supplementary Fig. 1A) were also used to examine the co-localization of LC3 and LIMP2. Similarly, co-localization of LIMP2 and LC3 was significantly reduced in PGRN KO C28I2 cells as compared to control cells under rapamycin stimulation (Supplementary Fig. 1D). In addition, the LC3 accumulation was also observed in PGRN KO C28I2 cells (Supplementary Fig. 1B, C). These data suggested that PGRN regulation of autophagosome-lysosome fusion in autophagy may be a mechanism common across different cell types.
In addtion, we tested the autophagic flux, which represented autophagosome-autolysosome formation and degradation, in GD fibroblasts using mCherry-GFP-LC3 assay. Compartment specific differential quenching of GFP and mRFP fluorescence signals allows the use of mCherry-GFP-LC3 for assessment of the number of autophagosomes and autolysosomes, and quantification of autophagic flux [42, 45, 46]. As shown in Fig. 4F and G, green puncta, which represented the unfused autophagosome, were significantly increased in L444P/PGRN KO compared with L444P, especially after rapamycin stimulation. The numbers of red and green punta were quantified with normalization to cell count and the difference between red puncta and green puncta was taken as representative of the autophagic flux. From Fig. 4G, we can see the autophagic flux was increased after rapamycin stimulation in L444P, however, no clear change was observed in L444P/PGRN KO with or without rapamycin stimulation, which suggested that autophagosome-autolysosome formation was defect in PGRN deficiency L444P (Fig. 4F, G).
Progranulin binds to Rab2 and Grn E of PGRN is the major domain responsible for Rab2 interaction
After the maturation of the autophagosome, the autophagosome and lysosome move closer together and fuse. Once fusion is complete, the contents of the autophagosome are exposed to the lysosome and are degraded by the lysosomal hydrolases [47, 48]. Various tethering factors contribute to autophagosome-lysosome fusion. SNAREs are reported to be the core machinery for fusion, the autophagosomal Q-soluble N-ethylmaleimide-sensitive factor attachment protein receptor (SNARE) protein syntaxin 7 (STX17) interacts with endosomal/lysosomal R-SNARE VAMP8 to form a trans-SNARE complex, which mediates autophagosome-lysosome fusion [49]. Consecutive RAB-mediation is a critical tethering step for autophagosome-lysosome fusion. RABs are small GTPases, which regulate autophagic membrane traffic [50, 51]. Rab7 is most widely assumed to regulate autophagosome-lysosome fusion by interacting with various tethering factors [52, 53]. However, a growing body of reports suggest that Rab2 is another critical GTPase mediator of autophagy [54-56]. In our present study, PGRN deficient cells presented with impaired autophagosome-lysosome fusion. We thus hypothesized that PGRN may associate with these aforementioned critical tethering factors including VAMP8, Rab7, or Rab2. To test this hypothesis, we used Nickel beads to pull down His-tagged PGRN and then the precipitated His-tagged PGRN complex was immunoblotted onto a nitrocellulose membrane followed by detection with antibodies against VAMP8, Rab7, or Rab2. As shown in Fig. 5A, Rab2 was co-purified with PGRN, indicating the physical interaction between PGRN and Rab2. However, VAMP8 or Rab7 was not detectable in precipitated PGRN complex solution. In addition, co-immunoprecipitation (Co-IP) further demonstrated the interaction between endogenous PGRN and Rab2 in L444P patient fibroblasts (Fig. 5B).
To identify the domains of PGRN required for interacting with Rab2, serial GFP-tagged N-terminal deletions of PGRN were constructed (Fig. 5C). Co-IP was performed using the same amounts of these mutants to examine their bindings to endogenous Rab2. As shown in Fig. 5D, the binding of PGRN with Rab2 became more pronounced following deletion of Grn P (ND1); binding became weaker with further deletion of Grn F (ND3) and most weak following deletion of Grn B (ND2 and ND3, respectively). These results suggest that Grn P and B might act as regulatory domains for the binding to Rab2. The Grn E (ND7) fragment exhibited strong binding ability undistinguisbable to that of full-length PGRN. Interestingly, our previous study found that the Grn E domain was also required and sufficient for PGRN binding to GCase [14]. To further determine whether this Grn E domain is similarly critical for binding with Rab2, a PGRN mutant with E domain deletion (Δ-ND7) was constructed; Co-IP revealed markedly reduced binding between Δ-ND7 and Rab2 relative to the interaction observed between full-length PGRN and Rab2 (Fig. 5D). Collectively, PGRN has morn than one domain involved in the interactions with Rab2, but the C-terminal fragment of PGRN (amino acids 496-593), named ND7, is both required and sufficient for the full interaction with Rab2. In addtion, the association of PGRN and Rab2 was further confirmed by their colocalization in L444P fibroblats (Fig. 5E), which supported the interaction between PGRN and Rab2.
Rab2 deficiency accumulated autophagic markers and defected autophagic flux in L444P GD fibroblasts
Each stage of dynamic autophagic processing, including autophagosome formation, autolysosome formation and lysosomal degradation intracellular membrane dynamics, is regulated by members of the Ras-like GTPase superfamily - primarily comprised of Rab proteins [50]. Several specific autophagic roles of individual Rab GTPases have been identified and Rab2 has been recognized as a crucial regulator in formation of autophagosome and autolysosome [54, 57]. To clarify whether binding between Rab2 and PGRN contributes to the role of Rab2 in autophagosome-autolysosome formation in GD, we also established a stable Rab2 KO L444P fibroblasts (L444P/Rab2 KO) using CRISPR-Cas9 technology (Supplementary Fig. 2A). Expression levels of autophagic makers LC3 and p62 were analysed in L444P or L444P/Rab2 KO. As shown in Supplementary Fig. 2B-D, the levels of both LC3-II and p62 were increased in Rab2 KO deficiency L444P, which indicated that the Rab2 deficiency leads to the defect of the late stage of autophagy in L444P. Autophagic flux was also measured using tandem mCherry-GFP-LC3 assay. As shown in supplementary Fig. 2E and F, the red puncta, which represent LC3 activation, were increased in both L444P and L444P/Rab2 KO under rapamycin stimulation. These data suggested that the activation of autophagy was not affected by the loss of Rab2. However, the green puncta, which represent unfused autophagosomes, were much more abundant in L444P/Rab2 KO compared with control L444P with or without rapamycin stimulation. Quantification of red puncta and green puncta, illustrative of the autophagic flux, revealed significantly lower flux in L444P/Rab2 KO than in control L444P under rapamycin stimulation (Supplementary Fig. 2E, F). These data revealed that autophagosome-autolysosome formation was blocked in L444P/Rab2 KO GD patient fibroblasts, consistent with previous reports using U2OS and HEK293 cell lines [54].
As Rab2 was identified as a binding partner of PGRN, we performed WB and RT-PCR to examine the potential effect of Rab2 deficiency on the protein and mRNA levels of PGRN. Interestingly, the protein level of PGRN was decreased in L444P/Rab2 KO (Supplementary Fig. 2G), however, the mRNA level of PGRN was not affected by Rab2 deficiency (Supplementary Fig. 2H), which suggested that Rab2 may also regulate PGRN at post-transcriptional levels, such as translation and stability of PGRN, in addition to physical interactions to each other.
PGRN derived ND7 effectively ameliorates autophagy defects in GD patient fibroblasts
In our previous study, we found that Grn E is a critical domain for the binding of PGRN to GCase [14]. Interestingly, Grn E domain is also required for PGRN-Rab2 interaction (Fig. 5). Accordingly, we expressed and purified this C-terminal 98 amino acid fragment of PGRN containing Grn E, termed ND7 (Fig. 6A, B), and tested whether it is sufficient for the mediation of autophagosome-lysosome fusion. First, L444P or L444P/PGRN KO fibroblasts were transfected with GFP-fused vector, or GFP-fused full-length PGRN, or GFP-fused ND7. 24 h later, the transfected cells were stimulated with rapamycin for another 24 h. Western blotting reveals reduced levels of LC3-II and p62 in full-length PGRN or ND7 transfected L444P/PGRN KO fibroblasts relative to GFP vector transfected L444P/PGRN KO; moreover, the LC3-II and p62 expression were comparable between full-length PGRN or ND7 tranfected L444P/PGRN KO fibroblasts and the control L444P (Fig. 6C).
We next tested the effect of ND7 on the formation of autophagosome using EM. From Fig. 6D, E, we observed that L444P/PGRN KO fibroblasts treated with ND7 exhibited a shift toward prevalence of advanced degradative vacuoles to a degree comparable with L444P. mCherry-GFP-LC3 assay revealed that green puncta, which represent the unfused autophagosome, were significantly decreased in ND7 treated L444P/PGRN KO compared with untreated L444P/PGRN KO under rapamycin stimulation (Fig. 6F, G). Moreover, autophagic flux was also increased in ND7 treated L444P/PGRN KO (Fig. 6F, G). Taken together, ND7 administration could rescue the autophagic defect in PGRN KO GD L444P fibroblasts. Morever, in lung tissues of OVA-challenged PGRN KO mice, levels of LC3-II and p62 also decreased drastically after ND7 administration as assessed by western blotting (Supplementary Fig. 3A). In line with these results, immunofluorescence staining for LC3 also revealed that ND7 could significantly decrease the levels of LC3-II and P62 (Supplementary Fig. 3B).