PREreview de Microglia extract neuronal proteolytic organelles via skoupocytosis

The manuscript by Pepper et al aimed to investigate how large proteolytic organelles are discarded from neurons, since these organelles may be too large to be trafficked back to the soma for lysosomal degradation. Using fluorescence and electron microscopy in in vitro models as well as 2-photon imaging in in vivo models, the authors identified a process, newly named skoupocytosis, whereby microglia, the resident immune cells of the central nervous system, contact neuronal processes, pinch off the portion containing the proteolytic organelle, and internalize it, without engulfing the entire process. Using synaptosome mass spectrometry and immunofluorescence imaging, the authors identified a phosphatidylserine (PS) lipase, ABHD16a, and the conversion of PS to lyso-PS as the initiation signal for skoupocytosis.

Altogether, the manuscript describes a localized process of microglial engulfment that maintains neuronal protein homeostasis without long-distance retrograde travel, which is an important mechanistic insight. While the in vitro (Figure 1J) and in vivo (Figure 2E-G) fluorescence imaging as well as the publicly available 3D volumetric electron microscopy (EM) datasets (Figure 4) strongly support the authors’ finding that microglia contact and internalize large, membrane-enclosed vesicles on various neuronal compartments, the authors’ claims that (1) skoupocytosis is a distinct and novel process from trogocytosis, (2) ABHD16a is necessary for the skoupocytosis process, and (3) the conversion of PS to lyso-PS is the attractant signal for microglia to neuronal processes is somewhat supported by quantification of microglial trogocytic vs skoupocytic events (Figure 4K) and in vitro pharmacological inhibition of ABHD16a in DIV21 hippocampal neurons and acute slices (Figure 6).

Major Points:

1. The authors report a novel process for microglial clearance of large proteolytic organelles, which they term skoupocytosis. The novelty of skoupocytosis is a major claim of the manuscript, however, the authors only explicitly address the difference between trogocytosis (trogo: dynamic, cellular “gnawing”) and skoupocytosis (skoupo: “garbage”) in a single panel (Figure 4K). The authors quantify the total number of trogocytic events, which they define in the text as microglial enwrapping of neurites without proteolytic organelles, and skoupocytic events, defined as microglial enwrapping of more than 50% of the proteolytic organelle or membrane containing proteolytic organelle. These definitions describe very similar processes, with skoupo adding specificity in target (proteolytic organelles). The data seem to suggest that skoupo events are a subset of trogo events that may be governed by the proposed ABHD16a/lyso-PS mechanism. Further, there are no statistical tests reported to show that the number of events is different between trogo and skoupo events.

   a. To strengthen the claim that skoupo is a novel pathway that differs from trogo, the authors may consider performing appropriate statistical tests (such as t-test) on the current dataset and adding significance notations, which would show the effect size of experimental conditions.

   b. Moreover, the authors may consider commenting on whether it has been reported if trogocytic immune cells have previously enveloped proteolytic organelles in neurons or other cell types, which would place the novelty of skoupo in context of other immune cells that perform trogocytosis.

   c. Additionally, the authors may consider conducting further assays that test whether aspects of trogo and skoupo are similar or different for each process including: molecular initiation signals/drivers, cytoskeletal dynamics of nibbling, and plasma membrane structure and fluidity during contact and internalization. Additional experiments would provide more evidence to identify whether skoupo is a subgroup of trogo or its own distinct process.

1. The authors report that ABHD16a activity is required for skoupocytosis and that lyso-PS, the byproduct of ABHD16a activity on PS, is a diffusible initiation cue for this process. They support this claim by (1) pharmacologically inhibiting ABHD16a with KC01, (2) labeling externalized PS heads with Annexin-V, which cannot differentiate PS from lyso-PS, and (3) measuring the ratio of PS intensity/HRS puncta (marking proteolytic organelles). While they find that KC01 treatment led to a significant increase in PS/HRS ratio, this finding does not directly show that lyso-PS diffuses from the neuron out into the extracellular space to recruit microglia to perform skoupocytosis. By inhibiting ABHD16a, one would expect that the levels of lyso-PS would decrease and PS would either increase or remain unchanged, whereas with functioning ABHD16a, PS is converted to lyso-PS, so PS levels would decrease and lyso-PS levels would increase.

   a. The authors may consider a more direct way of manipulating ABHD16a and measuring the amount of PS and lyso-PS intracellularly and/or extracellularly. One potential example could be to genetically inhibit ABHD16a (ex: knockout or dominant negative mutation) ensuring specific nullity of enzymatic active site, collecting neurons and neuronal media separately, and conducting a modified lipid extraction or high performance liquid chromatography / mass spectrometry to separate and quantify levels of PS and lyso-PS.

   b. The authors could also consider screening lyso-PS and other candidate signals in a microglial chemoattractant assay to better understand which signals are recruiting microglia. The caveat to in vitromicroglial assays is that it does not recapitulate all hallmarks of microglia in vivo, therefore, this lyso-PS signal would need to be validated in vivo.

   c. Moreover, the manuscript does not discuss how or why ABHD16a is localized to HRS+ puncta. Figure 5 shows that a majority of HRS sites colocalize with ABHD16a, but there is also ABHD16a present without HRS. To elucidate whether ABHD16a-mediated microglial recruitment is spatially restricted, the authors may consider quantifying the extent of colocalization throughout cellular compartments (soma/axon/dendrites) with correlative light and electron microscopy (CLEM). One next step would be to experimentally identify how ABHD16a is recruited to HRS+ organelles. Conversely, the authors could comment in the discussion on their hypotheses for ABHD16a recruitment.

Nonetheless, more direct evidence of the role of ABHD16a and lyso-PS is necessary to strengthen the claim that these are key players in initiating skoupocytosis.

Minor Points:

1. There are several different in vitro, in vivo, and slice methods, cell types, time periods, and brain regions used and analyzed in this manuscript. To provide greater context and relevance of surveying these different regions and cell types, the manuscript would benefit from additional rationale in the text to explain why the authors chose specific cell types and time periods. More discussion on whether microglia have a preference to use skoupo on one cell type or another or in one brain region versus another, even within cortex, would provide deeper mechanistic insight.

2. In Figure 5, the authors present a synaptosome proteomic analysis, where they pull out ABHD16a. To increase understandability of the plot, the authors should consider labeling the scale bar in panel D and highlighting a few top protein hits that are upregulated or downregulated in the volcano plot in panel B.

   a. Additionally, of the proteins plotted in panel D, ABDH16a does not show the most extreme differences between control and stimulated conditions. It may be useful to show the plot as a ranking of magnitude by stimulation effect, so that readers can understand which protein hits were most affected by stimulation.

   b. In the text, the authors do mention their interest in ABHD16a because of its association with PS, an “eat-me” signal. However, it would be interesting to know in the discussion if there were other proteins within the list of 112 membrane proteins that interacts with PS. This would provide insight and context into other proteins that could be candidate molecular cues for skoupo.

3. In Figure 3, the authors show aspects of proteolytic organelle uptake differ between hippocampus and somatosensory cortex. There are several plots (panels D-K) with a range of results. To increase visual clarity and home in on the central message, the authors may consider moving some of the plots that show no changes in age or region to supplemental information.

Competing Interests

The reviewer declares that she has no competing interests.

I reviewed this preprint as part of the Transparent and Accountable Peer Review training program. This report was written with the support of my research mentor.

Competing interests

The author declares that they have no competing interests.

Use of Artificial Intelligence (AI)

The author declares that they did not use generative AI to come up with new ideas for their review.