Rab32/Rab38-positive Lysosome-Related Organelle degrades lipid droplet in hepatocytes by microautophagy

  1. Zidi Zhang
  2. Shiou-Ling Lu
  3. Yumiko Kato
  4. Tongxin Zheng
  5. Bohan Chen
  6. Yangjie Li
  7. Yu Usami
  8. Taki Nishimura
  9. Ryohei Sakai
  10. Tomohiro Kabuta
  11. Narikazu Uzawa
  12. Satoru Toyosawa
  13. Takeshi Noda

Reviewed by

Read the full article See related articles

Discuss this preprint

Start a discussion What are Sciety discussions?

Listed in

Log in to save this article

Abstract

Rab32 and Rab38 are paralogous small GTPases involved in the biogenesis of lysosome-related organelles (LROs), yet their roles in hepatic lipid metabolism remain poorly defined. Here, Rab32 and Rab38 double-knockout (DKO) male mice exhibited an age-dependent increase in body weight accompanied by hepatic lipid accumulation, suggesting impaired hepatic lipid processing. In AML12 hepatocytes, Rab32 and Rab38 localized to ring-like, LAMP1-positive structures characteristic of LROs, whose size increased with cell confluence.

Pharmacological inhibition of lysosomal acid lipase with orlistat led to the accumulation of lipid droplets (LDs) within Rab32/38-positive LROs, indicating that LD degradation occurs in these compartments. Additional treatment with bafilomycin A1 revealed invagination-like internal membrane structures within enlarged LROs. These processes were not affected by artificial inhibition of macroautophagy, highlighting the involvement of microautophagy. Ring-like signals positive for phosphatidylinositol 3-phosphate (PI3P) or phosphatidylinositol 3,5-bisphosphate (PI(3,5)P₂) were detected within or adjacent to LRO membranes, and LDs were frequently associated with these structures, suggesting a role for PI3P and PI(3,5)P₂ in internal membrane formation. Vps4B was also required for efficient LD incorporation. Consistently, Rab32/38 double-knockdown (DKD) AML12 cells exhibited increased lipid accumulation, indicating impaired LD engulfment.

Together, these findings identify Rab32/38-positive LROs as a structural platform for microautophagy-mediated lipid droplet degradation in hepatocytes.

Article activity feed

  1. Review Commons

    Note: This response was posted by the corresponding author to Review Commons. The content has not been altered except for formatting.

    Learn more at Review Commons

    Reply to the reviewers

    We sincerely appreciate the constructive and insightful comments on our manuscript.

    Both reviewers raised important concerns regarding our use of the term lysosome-related organelle. We fully acknowledge this criticism and will revise the terminology throughout the manuscript with greater care, referring to these structures as Rab32/Rab38-positive vacuoles where appropriate, and discussing their possible relationship to lysosome-related organelles in the Discussion.

    We believe that the remaining comments can be adequately addressed through additional experiments, including CLEM and three-dimensional reconstruction analyses. We therefore submit this revision plan and hope that it will be viewed favorably.

  2. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #2

    Evidence, reproducibility and clarity

    Summary:

    This manuscript employs overexpression and knockdown experiments in an immortalized hepatocyte cell line to probe roles for RAB32 and RAB38 in lipid metabolism by lysosomes or lysosome-related organelles (LROs). Using these approaches, the authors show that both RAB32 and RAB38 colocalize with LAMP1 on late endosomes/ lysosomes, that the appearance of enlarged, round lysosomal structures that they refer to as LROs scales with both RAB32 and RAB38 expression, and they provide some evidence to suggest that material from lipid droplets (LD) are taken up into these large rounded compartments in a manner that requires RAB32 or RAB38. Additional experiments are interpreted to suggest that macroautophagy is not required for this uptake but that PtdIns3-kinase, PtdIns5-kinase, and ESCRT complexes are required. Analyses of Rab32/Rab38 knockout mice shows an accumulation of white fat, and in liver an accumulation of what the author interpret to be lipofuscin. The authors conclude that lipid droplets are consumed by LROs in an autophagy-independent manner.

    Major Comments:

    While the topic of the paper is interesting, the conclusions of the paper are not supported by the data shown. No evidence is presented in this paper that the structures analyzed are actual LROs rather than lysosomes, other than their content of RAB32 and RAB38 - which are not limited in expression to LROs. The fact that lipid accumulates in the white fat and not the livers of double knockout mice and that hepatocytes express very little RAB32 and no RAB38 renders the model cell system studied here artifactual; the paper should start with the in vivo analysis and then progress with an appropriate cell type using a line that mimics the behaviour of the endogenous cells. Moreover, the only experiments documenting partial overlap of lipid droplet (LD) material - interpreted as uptake of LDs - into these structures is in cells that massively overexpress LAMP1-mRFP, RAB32, and/or RAB38; in untransduced cells, only a handful of LAMP1-containing structures are enlarged and there is no evidence that they overlap with LD material. Moreover, the only evidence that colocalization is independent of autophagy is that it is blocked by overexpression of a single dominant-negative autophagy component, ATG4B. Finally, the data quantification throughout the paper lacks sufficient power to support the conclusions. Thus, the none of the major conclusions from this paper are well supported, and the physiological significance of the observations for liver function is not at all clear. Altogether, the authors present an interesting idea for which the data are unconvincing.

    Below are detailed concerns throughout the paper.

    1. Abstract:

    i. Please explain why there was a reason to look at the involvement of Rab32/38 in hepatic lipid metabolism.

    ii. It seems rather unlikely that microautophagy can result in the engulfment of an entire lipid droplet in toto; is it more sensible to think of this as a means to transfer the contents of LDs, perhaps piece by piece, into lysosomes?

    1. Introduction:

    i. There is a vast literature on the roles of Rab32 and/or Rab38 in the biogenesis of other LROs besides melanosomes, including platelet granules, lamellar bodies in lung epithelial type II cells, and various non-vertebrate structures that should be cited.

    ii. The authors fail to cite the first papers describing roles of Rab32 or Rab38 in bacterial killing by macrophages (Spano et al 2012, PMID: 23162001 and several additional papers from the Galan/ Spano groups), and papers ascribing roles for Rab32 in mitophagy and perhaps other mitochondrial functions, including ER:mitochondrial contacts, prior to the authors' 2025 paper (various papers).

    iii. There have been quite a few papers addressing Rab32/38 effectors in pigment cells (see papers from the Di Pietro group) and other cell types (see Rab32 in mitochondria papers).These facts and at least some of the papers should be cited in the Introduction to better reflect the depth of understanding - and some of the confusion - surrounding Rab32 and Rab38 function.

    iv. Reference to the definition of LROs should also be cited.

    Results:

    1. In all experiments where quantification was done, the number of structures or cells analyzed is listed but not the number of experiments. Were these experiments repeated at least three times, and are the values and statistics calculated from the experiment to experiment variation? If not, the statistical values are inaccurate. In all, the number of structures or cells analyzed appears to be quite small. 4.Figure 1.

    i. How did the authors validate the specificity of the anti-Rab32 and anti-Rab38 antibodies used in Figure 1 and elsewhere? Data should be shown with individual knockdowns. Additionally, the overlap with LAMP1 seems too good to be true (it looks 100% and with similar labeling intensities in all cases) - were controls done to ensure lack of cross-reactivity of the secondary antibodies?

    ii. If anti-Rab32 and -Rab38 actually labeled all LAMP1-positive compartments, it seems likely that these are classical late endosomes/ lysosomes and not lysosome-related organelles. Rab32 is expressed by many cell types that do not harbor traditional LROs and may have more ubiquitous functions. The larger ring-like structures mentioned in the text only appear when Rab32 or Rab38 are overexpressed as GFP fusion proteins (compare Fig. 1A and B with 1C-F, and note that the scale bars are the same) and fail to overlap with smaller structures only when LAMP1-mRFP is overexpressed (compare Fig. 1A and B with S1A); these structures likely represent earlier endosomal intermediates illuminated by LAMP1 overexpression. The authors need to reconsider their interpretation of these data in light of these overexpression artifacts.

    iii. In Fig. 1C-F and Fig. S1, were cells transfected or infected with recombinant lentiviruses? This should be indicated in the figure legend.

    1. Figure 2. In Fig. 2E-G, cells depleted of Rab32 and/or Rab38 should be compared to cells transduced with a control shRNA, such as a non-coding shRNA, and not to untransduced cells. The quantification of these data "per field" is quite concerning, given that a field could have very different numbers of cells. The data should be normalized to cell number or cell area.
    2. Figure 3.

    i. It should be noted in the text that the Lipi- dyes fluoresce in high hydrophobic environments, and thus would indicate a cluster of lipid tails within a lysosome and not just an entire LD. Interpreting these spots as LD under lipase inhibitory conditions is a stretch.

    ii. The evidence that the Lipi-Blue labeled structures are actually inside of the lysosomal structures is not convincing. Three-D reconstructions would need to be done to be more convincing of this.

    1. Suppl. Fig. S2. In panel A, there is no obvious difference in intensity of p62 under any of the conditions, and this reviewer does not see any LC3-II in the gel- only LC3-I with a very slight smear underneath that may or may not be specific. The interpretation that autophagy is increased at higher confluency is thus not well founded. In panel B, I see weak labeling of the interior of the giant Rab38-GFP-containing compartments for LC3-mRFP, as if the mRFP was in the process of degradation. How this correlates with the biochemistry in panel A is unclear.
    2. Fig. 4 and Suppl. Fig. S3.

    i. All of the graphs in Fig. S3 require appropriate statistical analyses.

    ii. The interpretation of the size of the structures in the double DKD sample is complicated by their accumulation in the perinuclear area, which is very dense. If all samples look like the one in Fig. 3A, then it is not possible to measure their size by this technique and that sample should remain unanalyzed. It is misleading to refer to these as large when they appear to be clusters of small puncta.

    iii. The label on the image itself in Fig. 3C should indicate Lysotracker, not "LRO". This is misleading.

    iv. The same concern raised above that it is not clear whether the Lipi-Blue labeled structures are present within the lysosomal structures is true here. Indeed, in the unstransfected control, many of the LD structures appear to be present adjacent to (on one side of) the Lysotracker-labeled structures, as is also apparent in the shRab32 and shRab38 cells; those where they appear to be inside might simply be above them in these non-super-resolution images. This is a great example of how it is necessary to do 3D reconstructions to fully determine whether the Lipi-Blue structures are engulfed by or adjacent to lysosomes.

    v. Note, the LC3 flux experiment and identification of LC3-II and -I is correct in S4D, unlike the experiment in S2A.

    1. Fig. 5. The data in Figure 5A are incorrectly interpreted. PtdIns3P or PtdIns(3,5)P2 are present only on the cytoplasmic leaflet of endosomes and lysosomes; if those membranes were to be internalized, the phosphate would be removed. Thus, the presence of signal on the inside of the lysosomal structures does not indicate the presence of PtdIns3P or PtdIns(3,5)P2; it represents likely free mCherry, or perhaps the full conjugate with 2XFYVE, that has been engulfed by the lysosome and is no longer bound to its ligand. The observation that the mCherry signal accumulates near the Lipi-Blue signal in orlistat-treated cells thus cannot be interpreted as an interaction of the phosphoinositide with the LD or its content phospholipids or acyl chains. The disappearance of a punctate 2XFYVE signal is expected upon treatment with a PI3kinase inhibitor since it eliminates the ligand, and the failure of Lipi-Blue to accumulate in lysosomes of inhibitor-treated cells could reflect just about any defect in endolysosomal maturation since PtdIns3P is required for the early to late endosome transition as well as for several aspects of late endosome and lysosome biology. All this experiment shows is that uptake of Lipi-Blue labeled structures into lysosomes requires endolysosomal maturation. The same goes for the shVps4 experiments in Fig. 5B, which are also less convincing of any phenotype, and Fig. S5.

    Significance

    Because the conclusions are not supported by the data shown and because the authors exploit an immortalized cell type that does not mimic the behavior of the endogenous cells, the significance of the work as presented is very low. If the conclusions were justified, the advance could potentially be conceptual in showing that RAB32 and RAB38 redundantly functionalize lysosomes in some cell types to metabolize lipids through a mechanism distinct from macroautophagy. Such an advance would be of broad interest to investigators interested in the functions of lysosomes and lysosome-related organelles, as well as membrane trafficking machinery. However, the authors are unfortunately a long way from such an advance.

    My expertise is in the biogenesis of LROs, and I am considered a leading expert in the field. In my opinion, the authors require a functional readout unique to LROs to define the compartments shown as LROs. Otherwise, they might consider altering their language, abandon the LRO designation, and focus on mechanisms of fatty acid uptake promoted by RAB32 and/or RAB38 in appropriate cell types. Unfortunately, their own data show that the cell type used here is not such an appropriate cell type.

  3. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #1

    Evidence, reproducibility and clarity

    This study investigates the roles of Rab32 and Rab38 in hepatic lipid droplet metabolism. The authors propose that Rab32/38-positive lysosome-related organelles (LROs) mediate lipid droplet degradation through a mechanism independent of conventional macroautophagy. While the study addresses an interesting question, several conceptual and technical issues need to be addressed before the conclusions can be fully supported.

    Major Concerns

    1.The authors primarily define the Rab32/38-positive ring-like structures as "lysosome-related organelles (LROs)" based on their morphological characteristics and co-localization with LAMP1. However, this classification lacks biochemical validation. Would it be more appropriate to include a Lyso-IP assay to provide additional supporting evidence? 2.In hepatocytes, what is the operational definition of LROs? Beyond being "larger in size," how are these structures functionally distinguished from conventional lysosomes? If Rab32/38 defines LRO identity, why does GFP-Rab32/38 not co-localize with all LAMP1-positive structures (Figure S1A)? 3.In Figure 2A, the dextran pulse-chase experiment shows fluid-phase uptake into large vacuoles; however, dextran can enter any endocytic compartment after prolonged chase periods. What evidence supports that these structures are bona fide LROs rather than enlarged late endosomes or lysosomes resulting from long-term culture? What determines why only certain lysosomes become Rab32/38-positive? This heterogeneity is not explained. Does it imply that pre-existing lysosomes convert into LROs, or that LROs are newly formed under high-density stress? The developmental trajectory of these structures has not been explored. 4.The authors propose a microautophagy mechanism based on the "invagination-like" structures observed by light microscopy (Figure 3A). However, the resolution of light microscopy is insufficient to distinguish true membrane invaginations from lipid droplets that are closely apposed to, or partially wrapped by, the outer membrane of LROs in three-dimensional space. Would a CLEM experiment be necessary to confirm that lipid droplets are indeed located within the lumen of LROs, rather than in deep invaginations that remain connected to the cytosol? In addition, multilamellar membrane structures were observed after Bafilomycin A1 treatment (Figure 3A). Have these structures been validated by electron microscopy, or could they simply represent complex membrane infoldings within swollen lysosomes? The conclusions drawn from light microscopy alone appear somewhat insufficient. 5.The authors use ATG4B C74A overexpression to claim macroautophagy independence. However, while this mutant blocks LC3 lipidation, the study still lacks genetic evidence, such as ATG knockouts. In Figure S2B, the authors state that the "majority" of Rab38-positive LRO-associated lipid droplets are LC3-negative, but no quantitative data are provided. 6.The manuscript does not clearly distinguish the functions of Rab32 and Rab38. Although the authors describe these proteins as paralogs with overlapping roles, multiple data points indicate that they have differential effects on lipid droplet (LD) metabolism. Notably, Rab38-but not Rab32-significantly affects LD delivery to acidic compartments, exerts a stronger influence on LRO size, and responds more robustly to VPS4B perturbation. These observations suggest that Rab32 and Rab38 regulate distinct steps of LD metabolism rather than functioning redundantly. However, the manuscript does not clearly highlight these functional differences and lacks mechanistic validation. 7.Figure 5A shows that the PI3P probe (2×FYVE) forms ring-like structures inside or near the LRO membrane. However, LROs themselves are Rab5-negative (Figures 1C-E), and PI3P is typically generated by Vps34 on early endosomes. Where do these PI3P signals originate? Are they transported from other organelles, or is there a local PI3P-generating mechanism on the LRO membrane? If the latter, which kinase is responsible, and is Vps34 recruited to the LRO membrane? This issue is not discussed. If PI3P is indeed locally generated on LROs, it could represent a key feature distinguishing LROs from classical lysosomes.

    Minor Concerns

    1.The double-knockout mice exhibit obesity and fatty liver; however, Rab32 and Rab38 are expressed in multiple tissues. A whole-body knockout model cannot distinguish whether these effects are hepatocyte-autonomous or arise from contributions by adipose tissue or macrophages, emphasizing the need for liver-specific knockout animals or cell models. Serum TAG levels were unchanged, and the authors speculate that VLDL secretion may be impaired, but this was not directly tested. Furthermore, the authors do not address the observed sex-specific effects, which appear to be male-specific. 2.The concentration of Orlistat used is relatively high (50-200 μM) and may cause non-specific effects. Have dose-response experiments been performed, or have other LAL inhibitors (e.g., Lalistat) been tested? 3.LysoTracker reflects acidity rather than lysosome identity, and reduced acidification in DKD cells may affect co-localization analysis.

    Significance

    Assessment of Significance Overall Assessment

    Strengths:

    Conceptual novelty: Introduces lysosome-related organelles (LROs) into hepatic lipid metabolism, expanding the functional repertoire of Rab32/38 beyond pigment cells and macrophages.

    Mechanistic exploration: Links LD uptake to PI3P/PI(3,5)P2 signaling and VPS4B, providing molecular handles for future studies.

    In vivo validation: DKO mice show age-dependent obesity and HFD sensitivity, establishing physiological relevance.

    Weaknesses:

    Rab32 vs. Rab38 functions remain blurred: Data suggest differential roles (Rab38 in LD delivery, Rab32 in LD size regulation), but authors default to "redundancy" narrative.

    Microautophagy evidence incomplete: Relies on light microscopy; EM/CLEM needed to confirm true internalization.

    Model relevance unclear: High-confluence AML12 vacuoles lack clear physiological correlate in healthy liver.

    Audience

    Primary:

    Lysosome biologists

    Autophagy researchers

    Lipid metabolism researchers

    Secondary:

    Cell biologists

    Metabolic disease researchers

    Geneticists

  4. Version published to 10.64898/2026.02.13.705687 on bioRxiv