Xkr regulates phosphatidylserine transport via ER-PM contact sites to promote apoptotic cell clearance

Reviewer #1 (Public review):

Summary:

The authors characterize the phospholipid scramblase Xkr in Drosophila. They generate null mutants in both S2 cells and flies and find that phosphatidylserine (PS) exposure is reduced during apoptosis; they show reduced engulfment of apoptotic cells, and that the protein is localized partially within the cytoplasm, overlapping with the ER. They go on to identify Xkr binding partners and show that they overlap with plasma membrane-ER contact sites, suggesting that Xkr facilitates PS transfer from the ER to PM. Overall, this reveals a new role for Xkr and identifies new binding partners, which are valuable contributions to the field.

Strengths:

(1) The generation of new Xkr reagents in both S2 cells and flies to analyze its function. Tools are used to quantify both PS exposure and efferocytosis, and the effects of Xkr knockout are significant.

(2) The discovery of new binding partners of Xkr which also affect PS exposure and efferocytosis.

(3) The authors demonstrate that the binding partners are conserved in mammalian cells.

Weaknesses:

(1) Throughout the manuscript (e.g, lines 105, 165, 274 and discussion), the authors describe Xkr as being activated in a caspase-independent manner, and use this as the rationale for identifying binding partners. However, this is never shown in the manuscript or clearly referenced. Interestingly, there is a TEVDA sequence in the fly ortholog at the same location as the caspase cleavage site in C. elegans Ced-8 (Figure S1), suggesting the caspase cleavage site is conserved. This should be further investigated, or the statements regarding caspase independence should be modified. I don't think the N- and C-terminal GFP fusions indicate caspase independence, especially since apoptosis was not induced in Figure 1A, B. If cleavage occurred at the TEVDA site in Figure S1A, it would not lead to a noticeable change on the Western blot, although the size does look a bit smaller in Figure S2B at the 8 h time point.

(2) The authors examine overlap between tagged Xkr and cellular compartment markers and find substantial overlap with Lamp (and other vesicle markers to a lesser extent) (Figure S2). This is not addressed in the paper and could indicate engulfment of other cells since S2 cells are macrophages. To test this, the staining could be tested on the mixed cells (vesicle-GFP tagged S2 + apoptotic xkr-mcherry). Similarly, calreticulin is an eatme signal that gets translocated to the PM of apoptotic cells. This could affect interpretation of colocalization (Figure 2J), and ideally another ER marker should be used.

(3) There are some places where there is over- or incorrect interpretation, and these instances should be corrected.

Specific examples:

a) Line 342 "Relative expression analysis by RT-qPCR showed that all three mutants were likely null alleles." This does not make sense since there is still mRNA present. In Figure S7A, the tm9sf4 allele is expressed at 75% of the control. The others show a greater reduction, but this is not proof of a null allele.

b) Figure S3I - It looks like mCherry-Lact:C2 does get localized to the PM with AcD treatment in the xkr[ko], although the authors conclude "this disrupted PS localization to the PM could not be restored by apoptosis induction". However, the PM localization does look disrupted in the tm9sf4 and sac1 knockdowns.

c) Figure 3I. The control Lact:C2 staining looks very different from the staining in Figure 2J, with abundant Lact:C2 outside the cell. Given the variability in the staining, were the contact sites quantified? On lines 287-288, it is stated that "fewer ER-PM MCSs were detected in xkrko cells than in WT", but no quantification is provided.

d) Line 299-300 - "the interaction between Xkr and dORP9 was enhanced after apoptosis induction". The interaction does not look enhanced in Figure S5F, so this statement should be removed or data supporting the statement should be provided. The interaction between Xkr and dORP2 looks enhanced upon apoptosis induction, but also paradoxically looks even more enhanced when apoptosis is blocked.

e) The data in Figure S6 are highlighted in the abstract. If this is a major conclusion, it would be best to move it to the main text and provide quantification.

f) Lines 392-4. The concluding statement seems overstated given that there was only a modest inhibition of PS exposure in the osbpl5 knockdown (Figure 6A) and no defects in efferocytosis (Figure 6C). The osbpl8 showed a stronger effect on PS exposure but still a very modest effect on efferocytosis.

Reviewer #2 (Public review):

In this study, the authors investigate the mechanisms underlying phosphatidylserine (PS) exposure during efferocytosis in Drosophila. They first show that Xkr promotes PS exposure and apoptotic cell clearance in both S2 cells and Drosophila embryos. As Drosophila Xkr lacks the canonical caspase cleavage site found in mammalian XKR proteins, the authors further explore the underlying mechanism by which Xkr regulates PS externalization. Through protein interaction studies, they identify TM9SF4 as an interacting partner of Xkr that regulates PS distribution and show that non-vesicular PS transport contributes to apoptotic PS exposure and efferocytosis. Using protein interaction studies, they further demonstrate that Xkr interacts with the lipid transfer protein dORP9 at ER-PM contact sites to facilitate non-vesicular PS transport to the plasma membrane. Loss of these proteins affects PS externalization and efferocytosis in Drosophila. Finally, using human cells, they demonstrate that human OSBPL8 interacts with XKR8 to regulate apoptotic PS exposure. Overall, the study supports a model in which Xkr promotes efferocytosis by facilitating lipid transport in addition to its role as a phospholipid scramblase.

Reviewer #3 (Public review):

Summary:

The manuscript investigates the function of the Drosophila Xkr protein, a homolog of mammalian Xkr8 that lacks the canonical caspase-cleavage motif. The authors show that apoptotic stimuli increase Xkr protein abundance through a post-transcriptional mechanism and that Xkr promotes phosphatidylserine (PS) exposure during apoptosis. Using immunoprecipitation coupled with mass spectrometry, they identify TM9SF4 as an Xkr-interacting protein and further implicate TM9SF4, Sac1, dORP2, dORP9, and Vap33 in regulating apoptotic PS exposure and efferocytosis. Based on these findings, the authors propose that Xkr regulates PS transport at ER-PM contact sites. Similar observations are also presented in human cells.

Strengths:

Overall, this is an interesting study. The authors provide convincing evidence that Drosophila Xkr participates in apoptotic PS exposure and employ multiple complementary approaches to support the involvement of several proteins in this pathway. The identification of TM9SF4 as a potential regulator of Xkr-mediated PS exposure is likely to be of broad interest.

Weaknesses:

I am less convinced by the evidence supporting the proposed role of ER-PM contact sites, and several mechanistic conclusions appear to extend beyond the data presented. Addressing the following points would substantially strengthen the manuscript.

Major concerns:

(1) In Figure 2A and related text, it is unclear whether the mass spectrometry analysis was performed using untreated cells or AcD-treated cells. If the objective was to identify apoptosis-associated Xkr interactors, it would be helpful to clarify the experimental condition and explain whether apoptosis-specific interactors were analyzed separately.

(2) In Figure 2B, 2E, and several other co-IP results, a negative control of Flag tag only is required to exclude experimental errors like insufficient washing, etc.

(3) In Figure S3B, S3F, and several other BiFC results, an mVC-only negative control would be important to exclude nonspecific fluorescence complementation.

(4) In Figure 2G, the quantitative values appear inconsistent with the flow cytometry histograms. The peak shift following Sac1 knockdown appears smaller than that of TM9SF4 knockdown, whereas the quantified values suggest the opposite. Please clarify this apparent discrepancy.

(5) I find the interpretation in Lines 223-227 difficult to reconcile with the data. Knockdown of both tm9sf4 and sac1 impaired apoptotic PS exposure to a similar extent as xkr knockout. However, while xkr deficiency significantly reduced efferocytosis, sac1 knockdown produced only a modest, statistically insignificant effect. These observations suggest that impaired PS exposure alone may not fully account for the efferocytosis phenotype observed in xkr-deficient cells. These results appear difficult to reconcile with the proposed model, which needs careful discussion.

(6) In Lines 274-275, the authors state that 'increased Xkr may accelerate non-vesicular PS transport for efficient apoptotic PS exposure'. However, Xkr protein levels increase only ~8 h after AcD treatment, whereas PS exposure occurs much earlier. Thus, alternative explanations like Xkr relocalization (Figure S5C), rather than increased abundance, may also explain how Xkr mediates PS transport. An Xkr overexpression experiment could be helpful to support this statement.

(7) The interpretation of the MAPPER experiments requires further clarification. In Line 283, the authors refer to "the intracellular proportion of the signal for each protein overlapping with MAPPER." Since MAPPER is designed to label ER-PM contact sites, which are located on the plasma membrane, intracellular MAPPER fluorescence likely represents the ER network rather than bona fide ER-PM contacts. Throughout the manuscript (including Figure S6, etc.), intracellular MAPPER puncta appear to be interpreted as ER-PM contacts, which may not be appropriate. In contrast, the peripheral MAPPER puncta observed along the cell cortex (e.g., Figure S5C after AcD treatment) are more consistent with authentic ER-PM contact sites. It is also not obvious that these cortical MAPPER signals colocalize with Xkr(Figure S5C). Thus, while the data support a role for the ER, they do not yet convincingly demonstrate Xkr clustering at ER-PM contact sites.

(8) In the Xkr knockout cells, all fluorescence signals appear substantially low in intensity. Differences in protein distribution are difficult to interpret when overall probe expression also appears altered. It would be helpful to demonstrate that probe expression levels are comparable between conditions. Furthermore, as noted above, intracellular MAPPER signal may primarily represent ER rather than ER-PM contacts. Finally, despite the reduced signal intensity, the remaining MAPPER and PS signals still appear well colocalized in the knockout cells, similar to the observations in Figure 2J. The interpretation in Lines 285-288 should therefore be reconsidered.

Author response:

Public Reviews:

Reviewer #1 (Public review):

Summary:

The authors characterize the phospholipid scramblase Xkr in Drosophila. They generate null mutants in both S2 cells and flies and find that phosphatidylserine (PS) exposure is reduced during apoptosis; they show reduced engulfment of apoptotic cells, and that the protein is localized partially within the cytoplasm, overlapping with the ER. They go on to identify Xkr binding partners and show that they overlap with plasma membrane-ER contact sites, suggesting that Xkr facilitates PS transfer from the ER to PM. Overall, this reveals a new role for Xkr and identifies new binding partners, which are valuable contributions to the field.

Strengths:

(1) The generation of new Xkr reagents in both S2 cells and flies to analyze its function. Tools are used to quantify both PS exposure and efferocytosis, and the effects of Xkr knockout are significant.

(2) The discovery of new binding partners of Xkr which also affect PS exposure and efferocytosis.

(3) The authors demonstrate that the binding partners are conserved in mammalian cells.

Weaknesses:

(1) Throughout the manuscript (e.g, lines 105, 165, 274 and discussion), the authors describe Xkr as being activated in a caspase-independent manner, and use this as the rationale for identifying binding partners. However, this is never shown in the manuscript or clearly referenced. Interestingly, there is a TEVDA sequence in the fly ortholog at the same location as the caspase cleavage site in C. elegans Ced-8 (Figure S1), suggesting the caspase cleavage site is conserved. This should be further investigated, or the statements regarding caspase independence should be modified. I don't think the N- and C-terminal GFP fusions indicate caspase independence, especially since apoptosis was not induced in Figure 1A, B. If cleavage occurred at the TEVDA site in Figure S1A, it would not lead to a noticeable change on the Western blot, although the size does look a bit smaller in Figure S2B at the 8 h time point.

We thank the reviewer for pointing out that, as E/DXXD has been considered a conserved caspase-3 cleavage site, TEVDA has also been validated as a caspase-6 cleavage site, which we have missed. We will further confirm this using site-mutated expression vectors in S2 cells.

(2) The authors examine overlap between tagged Xkr and cellular compartment markers and find substantial overlap with Lamp (and other vesicle markers to a lesser extent) (Figure S2). This is not addressed in the paper and could indicate engulfment of other cells since S2 cells are macrophages. To test this, the staining could be tested on the mixed cells (vesicle-GFP tagged S2 + apoptotic xkr-mcherry). Similarly, calreticulin is an eatme signal that gets translocated to the PM of apoptotic cells. This could affect interpretation of colocalization (Figure 2J), and ideally another ER marker should be used.

We thank the reviewer for the suggestion. We will attempt to label Xkr-mCherry under apoptosis with other vesicles and change the ER marker to Cnx99A (Calnexin ortholog in Drosophila).

(3) There are some places where there is over- or incorrect interpretation, and these instances should be corrected.

We thank the reviewer for their careful reading, and we will correct the mistakes in the revised manuscript.

Specific examples:

a) Line 342 "Relative expression analysis by RT-qPCR showed that all three mutants were likely null alleles." This does not make sense since there is still mRNA present. In Figure S7A, the tm9sf4 allele is expressed at 75% of the control. The others show a greater reduction, but this is not proof of a null allele.

We agree with the reviewer’s opinion. These mutants from the BDSC are not completely deleted but partially deleted; therefore, the RT-qPCR assay may not be very accurate. We will detect the mRNA levels of tm9sf4, dorp9, and sac1 using RT primers from different cDNA regions to make the results more convincing.

b) Figure S3I - It looks like mCherry-Lact:C2 does get localized to the PM with AcD treatment in the xkr[ko], although the authors conclude "this disrupted PS localization to the PM could not be restored by apoptosis induction". However, the PM localization does look disrupted in the tm9sf4 and sac1 knockdowns.

We thank you for raising this intriguing hypothesis. Indeed, PM localization of Lact:C2 was reduced in xkr^ko cells, and the distribution could not be rescued after apoptosis. Unlike xkr^ko, tm9sf4, and sac1 RNAi-treated cells displayed weak PS disorder, which may be due to the efficiency of knockdown. However, the statistical results indicated that the ratio of PM/Cyto was reduced in tm9sf4 and sac1 RNAi-treated cells.

c) Figure 3I. The control Lact:C2 staining looks very different from the staining in Figure 2J, with abundant Lact:C2 outside the cell. Given the variability in the staining, were the contact sites quantified? On lines 287-288, it is stated that "fewer ER-PM MCSs were detected in xkrko cells than in WT", but no quantification is provided.

We thank for the reviewer’s suggestion. We will add the statistical results of Fig. 3I in the revised version.

d) Line 299-300 - "the interaction between Xkr and dORP9 was enhanced after apoptosis induction". The interaction does not look enhanced in Figure S5F, so this statement should be removed or data supporting the statement should be provided. The interaction between Xkr and dORP2 looks enhanced upon apoptosis induction, but also paradoxically looks even more enhanced when apoptosis is blocked.

We thank you for raising this intriguing hypothesis. We will delete the relevant statement to eliminate unnecessary misunderstandings.

e) The data in Figure S6 are highlighted in the abstract. If this is a major conclusion, it would be best to move it to the main text and provide quantification.

We thank for the reviewer’s suggestion. We will move this to the main text and provide quantification in the revised version.

f) Lines 392-4. The concluding statement seems overstated given that there was only a modest inhibition of PS exposure in the osbpl5 knockdown (Figure 6A) and no defects in efferocytosis (Figure 6C). The osbpl8 showed a stronger effect on PS exposure but still a very modest effect on efferocytosis.

We thank for the reviewer’s suggestion. We will weaken the statement in the Results section of Figure 6 and perform osbpl9 knockdown to observe efferocytosis in Raw264.7 cells, as OSBPL9 interacts with Xkr8 strongly.

Reviewer #2 (Public review):

In this study, the authors investigate the mechanisms underlying phosphatidylserine (PS) exposure during efferocytosis in Drosophila. They first show that Xkr promotes PS exposure and apoptotic cell clearance in both S2 cells and Drosophila embryos. As Drosophila Xkr lacks the canonical caspase cleavage site found in mammalian XKR proteins, the authors further explore the underlying mechanism by which Xkr regulates PS externalization. Through protein interaction studies, they identify TM9SF4 as an interacting partner of Xkr that regulates PS distribution and show that non-vesicular PS transport contributes to apoptotic PS exposure and efferocytosis. Using protein interaction studies, they further demonstrate that Xkr interacts with the lipid transfer protein dORP9 at ER-PM contact sites to facilitate non-vesicular PS transport to the plasma membrane. Loss of these proteins affects PS externalization and efferocytosis in Drosophila. Finally, using human cells, they demonstrate that human OSBPL8 interacts with XKR8 to regulate apoptotic PS exposure. Overall, the study supports a model in which Xkr promotes efferocytosis by facilitating lipid transport in addition to its role as a phospholipid scramblase.

Thank you for your comprehensive and generous assessment of our work and for the time and expertise you have devoted to reviewing our manuscript. We will revise the manuscript accordingly and provide a point-by-point response in the revised version.

Reviewer #3 (Public review):

Summary:

The manuscript investigates the function of the Drosophila Xkr protein, a homolog of mammalian Xkr8 that lacks the canonical caspase-cleavage motif. The authors show that apoptotic stimuli increase Xkr protein abundance through a post-transcriptional mechanism and that Xkr promotes phosphatidylserine (PS) exposure during apoptosis. Using immunoprecipitation coupled with mass spectrometry, they identify TM9SF4 as an Xkr-interacting protein and further implicate TM9SF4, Sac1, dORP2, dORP9, and Vap33 in regulating apoptotic PS exposure and efferocytosis. Based on these findings, the authors propose that Xkr regulates PS transport at ER-PM contact sites. Similar observations are also presented in human cells.

Strengths:

Overall, this is an interesting study. The authors provide convincing evidence that Drosophila Xkr participates in apoptotic PS exposure and employ multiple complementary approaches to support the involvement of several proteins in this pathway. The identification of TM9SF4 as a potential regulator of Xkr-mediated PS exposure is likely to be of broad interest.

Weaknesses:

I am less convinced by the evidence supporting the proposed role of ER-PM contact sites, and several mechanistic conclusions appear to extend beyond the data presented. Addressing the following points would substantially strengthen the manuscript.

We sincerely thank you for your careful reading and accurate summary of our manuscript. We appreciate the time, effort, and expertise you have dedicated to evaluating our work, and we will try our best to improve our manuscript according to your suggestions.

Major concerns:

(1) In Figure 2A and related text, it is unclear whether the mass spectrometry analysis was performed using untreated cells or AcD-treated cells. If the objective was to identify apoptosis-associated Xkr interactors, it would be helpful to clarify the experimental condition and explain whether apoptosis-specific interactors were analyzed separately.

We thank for the reviewer’s suggestion. We used AcD-treated S2 cells and untreated S2 cells to perform mass spectrometry. To clarify this, we will add a detailed method description in the method section.

(2) In Figure 2B, 2E, and several other co-IP results, a negative control of Flag tag only is required to exclude experimental errors like insufficient washing, etc.

We thank for the reviewer’s suggestion. We used anti-HA magnetic beads to perform immunoprecipitation, and single HA-TM9SF4 was used as a negative control.

(3) In Figure S3B, S3F, and several other BiFC results, an mVC-only negative control would be important to exclude nonspecific fluorescence complementation.

We thank for the reviewer’s suggestion, we will add the negative control for BiFC results in the revised version.

(4) In Figure 2G, the quantitative values appear inconsistent with the flow cytometry histograms. The peak shift following Sac1 knockdown appears smaller than that of TM9SF4 knockdown, whereas the quantified values suggest the opposite. Please clarify this apparent discrepancy.

We sincerely thank you for the careful consideration of our statistical results, which were obtained from 3 repeats. We will choose another flow cytometry histogram of tm9sf4 and sac1 to make the data and images more consistent.

(5) I find the interpretation in Lines 223-227 difficult to reconcile with the data. Knockdown of both tm9sf4 and sac1 impaired apoptotic PS exposure to a similar extent as xkr knockout. However, while xkr deficiency significantly reduced efferocytosis, sac1 knockdown produced only a modest, statistically insignificant effect. These observations suggest that impaired PS exposure alone may not fully account for the efferocytosis phenotype observed in xkr-deficient cells. These results appear difficult to reconcile with the proposed model, which needs careful discussion.

We sincerely thank the reviewer for their careful and thoughtful observations. Given the results we have observed, we will add this to the discussion section in the revised version.

(6) In Lines 274-275, the authors state that 'increased Xkr may accelerate non-vesicular PS transport for efficient apoptotic PS exposure'. However, Xkr protein levels increase only ~8 h after AcD treatment, whereas PS exposure occurs much earlier. Thus, alternative explanations like Xkr relocalization (Figure S5C), rather than increased abundance, may also explain how Xkr mediates PS transport. An Xkr overexpression experiment could be helpful to support this statement.

We thank for the reviewer’s suggestion. We will overexpress Xkr with or without AcD treatment to observe whether the localization or amount of Lact:C2 changes and to re-evaluate the role of Xkr in PS exposure.

(7) The interpretation of the MAPPER experiments requires further clarification. In Line 283, the authors refer to "the intracellular proportion of the signal for each protein overlapping with MAPPER." Since MAPPER is designed to label ER-PM contact sites, which are located on the plasma membrane, intracellular MAPPER fluorescence likely represents the ER network rather than bona fide ER-PM contacts. Throughout the manuscript (including Figure S6, etc.), intracellular MAPPER puncta appear to be interpreted as ER-PM contacts, which may not be appropriate. In contrast, the peripheral MAPPER puncta observed along the cell cortex (e.g., Figure S5C after AcD treatment) are more consistent with authentic ER-PM contact sites. It is also not obvious that these cortical MAPPER signals colocalize with Xkr(Figure S5C). Thus, while the data support a role for the ER, they do not yet convincingly demonstrate Xkr clustering at ER-PM contact sites.

We thank the reviewer for the suggestion, and we believe that the TIRF technique can help us demonstrate the ER-PM signal. Since our college has no TIRF microscope, we will try our best to seek cooperation from other colleges to achieve this experiment.

(8) In the Xkr knockout cells, all fluorescence signals appear substantially low in intensity. Differences in protein distribution are difficult to interpret when overall probe expression also appears altered. It would be helpful to demonstrate that probe expression levels are comparable between conditions. Furthermore, as noted above, intracellular MAPPER signal may primarily represent ER rather than ER-PM contacts. Finally, despite the reduced signal intensity, the remaining MAPPER and PS signals still appear well colocalized in the knockout cells, similar to the observations in Figure 2J. The interpretation in Lines 285-288 should therefore be reconsidered.

We sincerely thank the reviewer for this careful and thoughtful observation, and we agree that the interpretation in Line 285-288 is overstated. To explain this, we plan to detect the Lact:C2 and MAPPER signals in S2 and xkr^ko cells with or without AcD to confirm how Xkr regulates PS via ER-PM under apoptotic conditions.