Bidirectional interconversion between PtdIns4P and PtdIns(4,5)P₂ is required for autophagic lysosome reformation and protection from skeletal muscle disease

ABSTRACT

Autophagic lysosome reformation (ALR) recycles autolysosome membranes formed during autophagy, to make lysosomes and is essential for continued autophagy function. Localized membrane remodeling on autolysosomes leads to the extension of reformation tubules, which undergo scission to form new lysosomes. The phosphoinositides phosphatidylinositol-4-phosphate (PtdIns4P) and phosphatidylinositol-4,5-bisphosphate (PtdIns[4,5]P₂) induce this remodeling by recruiting protein effectors to membranes. We identified the inositol polyphosphate 5-phosphatase INPP5K, which converts PtdIns(4,5)P₂ to PtdIns4P is essential for ALR in skeletal muscle. INPP5K mutations that reduce its 5-phosphatase activity are known to cause muscular dystrophy, via an undefined mechanism. We generated skeletal muscle-specific inpp5k knockout mice which exhibited severe muscle disease, with lysosome depletion and marked autophagy inhibition. This was due to decreased PtdIns4P and increased PtdIns(4,5)P₂ on autolysosomes, causing reduced scission of reformation tubules. ALR was restored in cells with loss of INPP5K by expression of wild-type INPP5K, but not muscle-disease causing mutants. Therefore on autolysosomes, both PtdIns(4,5)P₂ generation and its removal by INPP5K is required for completion of ALR. Furthermore, skeletal muscle shows a dependence on the membrane recycling ALR pathway to maintain lysosome homeostasis and ensure the protective role of autophagy against disease.

Autophagy requires the fusion of cargo-laden auto-phagosomes with lysosomes, and lysosome replacement is critical for continued autophagy. Skeletal muscle has high basal autophagy, which ensures muscle heath, and is further increased during the physiological response to fasting or exercise. This places a significant demand on lysosome repopulation to ensure these adaptive responses.

Lysosome repopulation during autophagy can occur via de novo formation, whereby MiT/TFE transcription factors are master regulators of lysosomal genes. An alternate pathway is autophagic lysosome reformation (ALR), where lysosomes are re-derived from existing autolysosome membranes. The relationship between these processes is unknown but intriguing given both are regulated by MTOR, but with opposing effects; MTOR inhibition is a prerequisite for MiT/TFE-dependent lysosome biogenesis during autophagy, but ALR initiation requires amino-acid dependent MTOR reactivation on autolysosomes.

ALR is characterized by the localized budding of autolysosome membranes to form “reformation tubules” which undergo scission, generating new lysosomes. Successive generation of PtdIns4P, followed by PtdIns(4,5)P₂ on autolysosome membranes facilitates these membrane changes (Figure 1). PtdIns(4,5)P₂-rich microdomains recruit effector proteins including clathrin, KIF5B and WHAMM, driving localized membrane remodeling into a bud then a tubule. DNM2 (dynamin 2) is required for reformation tubule scission, but this stage is less well characterized.

Figure 1.

Autophagic lysosome reformation relies on the bidirectional interconversion between phosphoinositides PtdIns4P and PtdIns(4,5)P₂, to ensure lysosome homeostasis, autophagy function and protection from disease. During ALR, a hierarchical succession of phosphoinositide kinase and phosphatase enzymes is required for dynamic PtdIns4P and PtdIns(4,5)P₂ signaling on autolysosomes. First, PtdIns4P is made from PI by either PI4KB (phosphatidylinositol 4-kinase beta) or PI4K2A (phosphatidylinositol 4-kinase, type 2 alpha). In turn, PtdIns4P is used by two different PtdIns4P 5-kinases, PIP5K1A or PIP5K1B, to make PtdIns(4,5)P₂. This signaling is opposed by the PtdIns(4,5)P₂ 5-phosphatase INPP5K, which converts PtdIns(4,5)P₂ back to PtdIns4P. The generation of PtdIns(4,5)P₂ microdomains on the autolysosome membrane recruits effector proteins which drive specialized changes to membrane ultrastructure. The AP2-clathrin complex induces membrane budding, then the microtubule motor protein KIF5B and the actin remodeling protein WHAMM, exert forces to extrude the bud to form a membrane tubule. DNM2 is thought to facilitate tubule scission to generate lysosomes. When INPP5K function is lost, either by reduced expression or catalytically inactivating disease mutants, the dynamic association of PtdIns(4,5)P₂ effectors is dysregulated leading to the inability of reformation tubules to undergo scission to generate new lysosomes. This results in the enlargement of autolysosomes and the hyperextension of reformation tubules. The depletion of lysosomes when INPP5K function is lost, causes autophagy inhibition and muscle disease

INPP5K/SKIP, an inositol polyphosphate 5-phosphatase highly expressed in skeletal muscle, dephosphorylates PtdIns(3,4,5)P₃ to PtdIns(3,4)P₂ and with higher affinity PtdIns(4,5)P₂ to PtdIns4P. INPP5K mutations are a recently identified cause of congenital muscular dystrophy, where cataracts and intellectual impairment can also occur. INPP5K mutations reduce its catalytic function toward PtdIns(4,5)P₂ but how this causes disease was unknown. We generated a skeletal muscle-specific inpp5k knockout mouse, which developed early-onset, severe muscle disease [1]. Marked inhibition of autophagic degradation was apparent by the significant increase in SQSTM1/p62, ubiquitinated and LC3-II proteins in the muscle of knockout mice, associated with a severe block in autophagic flux.

INPP5K loss in muscle caused increased AKT-MTOR activation and we anticipated this caused the autophagy inhibition. However, treatment of inpp5k knockout mice with AKT (MK2206) or MTOR (rapamycin) inhibitors did not restore autophagy or reduce muscle disease, indicating the autophagy defect was independent of AKT-MTOR hyperactivation.

Lysosomes were depleted in inpp5k knockout muscle, and myoblasts with loss of INPP5K failed to sustain lysosome homeostasis during autophagy. We initially reasoned this may be due to suppression of MiT/TFE-dependent lysosome biogenesis, as a consequence of the MTOR hyperactivation observed in inpp5k knockout muscle. Indeed, nuclear localization of TFEB and lysosomal gene activation were reduced in inpp5k knockout muscle; however, this effect was only consistently observed following fasting. TFEB nuclear localization and lysosomal gene activation were reinstated in inpp5k knockout mice following treatment with the MTOR inhibitor rapamycin, but despite this, lysosome number and autophagy function were not restored and muscle disease was not reduced. This suggested defects in a different lysosome homeostasis pathway.

inpp5k knockout muscle and myoblasts exhibited prominent accumulation of enlarged autolysosomes indicative of ALR inhibition, which we examined further. ALR is induced by prolonged starvation-induced autophagy, and under these conditions autolysosome membrane reformation tubules were still present with loss of INPP5K, but were long-lived and hyperextended, suggesting a defect in their scission to form lysosomes. INPP5K localized to autolysosomes and lysosomes, where this 5-phosphatase converted PtdIns(4,5)P₂ to PtdIns4P. Elevated PtdIns(4,5)P₂ and decreased PtdIns4P on autolysosomes in inpp5k knockout muscle occurred with a marked increase in the PtdIns(4,5)P₂-effector proteins AP2 and clathrin. In myoblasts with loss of INPP5K, reformation tubules also showed increased clathrin-association and we propose this interferes with their scission to generate lysosomes. This regulatory role for INPP5K was dependent upon its PtdIns(4,5)P₂ 5-phosphatase function as expression of wild type INPP5K but not a catalytically inactive mutant, in INPP5K knockdown cells, restored lysosome homeostasis. Critically, expression of INPP5K muscular dystrophy mutants with reduced PtdIns(4,5)P₂ 5-phosphatase activity also did not promote ALR.

Finally, we examined the functional relationship between INPP5K and the PtdIns4P 5-kinases that generate PtdIns(4,5)P₂ on autolysosomes during ALR, PIP5K1A (phosphatidylinositol-4-phosphate 5-kinase, type 1 alpha) or PIP5K1B (phosphatidylinositol-4-phosphate 5-kinase, type 1 beta) (Figure 1). Co-depletion of PIP5K1B in INPP5K knockdown cells, restored ALR and lysosome homeostasis, by normalizing PtdIns4P and PtdIns(4,5)P₂ levels, revealing that PIP5K1B and INPP5K play opposing roles in regulating PtdIns(4,5)P₂ during ALR.

Sequential formation of PtdIns4P and PtdIns(4,5)P₂ on autolysosomes is known to be an important initiation signal for ALR. Our study reveals the conversion of PtdIns(4,5)P₂ to PtdIns4P is equally critical (Figure 1). This dynamic balance is achieved by the opposing functions of PIP5K1B which makes PtdIns(4,5)P₂ from PtdIns4P, and INPP5K which reverts PtdIns(4,5)P₂ back to PtdIns4P, allowing fine regulation of the recruitment of effector proteins such as clathrin to reformation tubules. This bidirectional phosphoinositide switch is critical for lysosome homeostasis, autophagy function and protection against muscle disease.

We were intrigued by the possibility that in muscle where autophagic flux is high, ALR may be highly efficient for replenishing lysosomes in muscle, since it utilizes preexisting autolysosome membranes formed during autophagy. Our study now establishes the dependence of muscle on ALR to maintain lysosome homeostasis, and when this fails severe disease and marked autophagy inhibition occurs. Lysosome homeostasis and autophagy could also not be restored in inpp5k knockout muscle by stimulating TFEB-dependent lysosome biogenesis. This raises a future question of whether different tissues show a dependence on lysosome biogenesis versus ALR, based on their distinct autophagy demands or specific physiological scenarios of increased autophagic flux.

Finally, the role of ALR inhibition in disease is now compelling and expanding, including our study of muscular dystrophy and the work of others on hereditary spastic paraplegia. This lays the foundations for screening other ALR regulatory genes as causative for disease.

Funding Statement

This study was funded by grants awarded to C.A.M and M.J.M. from the National Health and Medical Research Council, Australia [NHMRC; APP1024308 and APP1082253] and the Australian Research Council [ARC; DP190102499].

Disclosure statement

No potential conflict of interest was reported by the author(s).