TMEM41B functions with VMP1 in autophagosome formation

ABSTRACT

Macroautophagy/autophagy requires many autophagy-related (ATG) proteins. Most of the ATG genes were identified by genetic screening using simple model organisms. Recently, we performed a forward genetic screen in mammalian cells using the CRISPR-Cas9 system and our autophagic flux reporter GFP-LC3-RFP. One of the identified proteins was TMEM41B, an ER-localized multi-spanning membrane protein. TMEM41B has a characteristic transmembrane domain (the VTT domain), which is also found in VMP1, another protein involved in autophagy. Our results show that TMEM41B and VMP1 are physically and functionally associated.

Macroautophagy is a complex and dynamic process that requires many molecules. Currently, more than 40 ATG genes are known, including some that are required for specific types of autophagy such as pexophagy, mitophagy, reticulophagy, and nucleophagy. Most of these ATG genes were identified by genetic screening using model organisms including Saccharomyces cerevisiae, Komagataella phaffii, Caenorhabditis elegans, and Schizosaccharomyces pombe. RNAi-mediated screening has been performed in mammalian cells, but the recently developed CRISPR-Cas9 system allows knockout (KO)-based genome-wide screening to be performed efficiently. Indeed, DeJesus et al. found that the UFMylation pathway regulates the amount of SQSTM1/p62, a selective substrate of autophagy, and Goodwin et al. discovered an autophagy-independent lysosomal targeting pathway by CRISPR-based genome-wide screening.

Recently, we also performed genome-wide screening to search for novel autophagy regulators in mammals [1]. In this screen, we used GFP-LC3-RFP, an autophagic flux reporter established in our laboratory. This reporter is cleaved into GFP-LC3 and RFP by endogenous ATG4 family proteins. When autophagy is induced, GFP-LC3 is delivered to lysosomes by autophagy and quenched, whereas RFP stays in the cytosol and serves as an internal control. The signals of this reporter can be detected by flow cytometry and used in a 2-dimensional chart, enabling clear separation between normal and autophagy-deficient cell populations (Figure 1(a)). As a result, we succeeded in identifying most nonredundant canonical ATG genes (ATG3, ATG4B, ATG5, ATG7, ATG9A, ATG10, ATG12, ATG13, ATG14, ATG16L1, ATG101, and RB1CC1/FIP200) as well as genes encoding components of the homotypic fusion and protein sorting (HOPS) complex (VPS16 and VPS33A), negative regulators of MTORC1 (TSC1 and TSC2), and other known autophagy regulators (VPS15, EPG5, and EI24). In addition, TMEM41B, a novel ATG gene, was identified.

Figure 1.

Screening strategy and the structure of TMEM41B and VMP1. (a) Signals from the GFP-LC3-RFP probe can be detected by flow cytometry. A dual-color reporter that yields an internal control (RFP) reflecting the expression level of the reporter allows a clearer separation than a single-color reporter that lacks an internal control. Gray dots indicate autophagy-deficient cells; green dots indicate normal cells. (b) Hypothetical structures of TMEM41B and VMP1. The VTT domain is conserved in both proteins, which form a complex in the ER membrane. TMEM41B and VMP1 cooperate to form autophagosomes.

TMEM41B, like VMP1, is a multi-spanning membrane protein that localizes in the endoplasmic reticulum (ER). TMEM41B has a characteristic transmembrane domain in the central region. As similar domains are found in other proteins such as VMP1, yeast Tvp38, and even bacterial proteins, we named it the VTT (VMP1, TMEM41, Tvp38) domain. In TMEM41B-KO cells, autophagosome formation is blocked at an early step, as elongated phagophores are not detected. Lipid droplets accumulate in TMEM41B-KO cells. The phenotype of TMEM41B-KO cells is reminiscent of VMP1-KO cells, suggesting a close relationship between TMEM41B and VMP1.

In fact, TMEM41B physically interacts with VMP1 in vivo and in vitro. Moreover, in contrast to overexpression of TMEM41B in VMP1-KO cells, overexpression of VMP1 rescues the autophagy deficiency of TMEM41B-KO cells. According to classical genetic principles, these results suggest that VMP1 functions downstream of TMEM41B (Figure 1(b)). Thus, TMEM41B may assist the function of VMP1 in phagophore elongation.

At nearly the same time, 2 other groups also identified TMEM41B as a novel ATG gene. Moretti et al. identified TMEM41B in a CRISPR-based screen using endogenous SQSTM1/p62 and CALCOCO2/NDP52 as indicators. The autophagic phenotype that they observed was almost identical to the phenotype that we observed. They further characterized the lipid accumulation phenotype and discovered that mobilization and utilization of fatty acids were delayed in TMEM41B-KO cells. Shoemaker et al. also identified TMEM41B in a CRISPR-based screen using tfReceptor (RFP-GFP tagged autophagy receptor) probes.

What are the functions of TMEM41B and VMP1?

To answer this question, we need to know the function of the VTT domain. The VTT domain is widely conserved from bacteria to eukaryotes. Escherichia coli YdjX and YdjZ, which have VTT domains, are members of the DedA family. Although still speculative, DedA family proteins are predicted to be half-transporters. If this hypothesis is correct, TMEM41B and VMP1 may also be half-transporters and become full-transporters of ions or lipids when they form homo- or hetero-oligomers. To understand their precise functions, it is necessary to determine their structures and to perform in vitro experiments. Further information will provide new insights into the role of these VTT domain proteins and the ER in autophagosome formation.

Screening limitations

Our CRISPR-based screen was almost saturated, as evidenced by the fact that a nearly complete set of nonredundant ATG genes was identified and only TMEM41B was novel. However, there were 2 major limitations. First, genes essential for cell survival and growth could not be detected because cells lacking such genes were excluded during the 3 enrichment cycles. Shortening the selection period may have increased the number of surviving clones but would have resulted in a low signal-to-noise ratio. Second, our screening strategy could not identify genes that have functional homologs. Indeed, ULK1/2, ATG2A/B, and WIPI1, WIPI2, WDR45B/WIPI3 and WDR45/WIPI4 were not detected. This problem could be overcome by using multiplex gRNA libraries. Thus, it is still possible that more novel ATG genes exist.

Funding Statement

This work was supported by the Japan Science and Technology Agency [JPMJER1702].

Acknowledgments

This work was supported by Exploratory Research for Advanced Technology (ERATO, Grant Number JPMJER1702) from the Japan Science and Technology Agency (JST) to N.M.

Disclosure statement

No potential conflict of interest was reported by the authors.