HS1BP3 negatively regulates autophagy by modulation of phosphatidic acid levels

A fundamental question is how autophagosome formation is regulated. Here we show that the PX domain protein HS1BP3 is a negative regulator of autophagosome formation. HS1BP3 depletion increased the formation of LC3-positive autophagosomes and degradation of cargo both in human cell culture and in zebrafish. HS1BP3 is localized to ATG16L1- and ATG9-positive autophagosome precursors and we show that HS1BP3 binds phosphatidic acid (PA) through its PX domain. Furthermore, we find the total PA content of cells to be significantly upregulated in the absence of HS1BP3, as a result of increased activity of the PA-producing enzyme phospholipase D (PLD) and increased localization of PLD1 to ATG16L1-positive membranes. We propose that HS1BP3 regulates autophagy by modulating the PA content of the ATG16L1-positive autophagosome precursor membranes through PLD1 activity and localization. Our findings provide key insights into how autophagosome formation is regulated by a novel negative-feedback mechanism on membrane lipids.

experiments were included to strengthen the authors' data and provide a strong argument.
Reviewer #2 (Remarks to the Author) The formation of autophagosomes is a complex process which requires a complex machinery and multiple steps, involve both lipids, proteins and interdependencies between lipids and proteins. Holland et al have addressed the role of a lipid binding protein, HS1BP3, which binds PA. HS1BP3 is known to bind HS1 or cortactin. The authors demonstrate that HS1BP3 is a negative regulator of starvation-induced autophagy, which cololocalizes with Atg16L. HS1BP3 is proposed to act by inhibiting PLD1 activity on ATG16L1-positive and TfR-positive membranes. In addition, loss of HS1BP3 increases the synthesis of particular PA species in cells. The authors propose a model whereby HS1BP3 regulates PLD1's access to the Atg16L1, TfR-positive membranes thereby controlling the lipid composition, in particular PA levels of autophagosome precursor membranes. This is a revised version of a manuscript submitted to NCB. The authors have addressed some of the points raised in the previous review but they have done this primarily by removing large sets of data or replacing all the original problematic data. They have made a effort to identify an antibody to the endogenous protein but this has not provided insight into the mechanism. The manuscript, while much easier to understand, is now largely descriptive and provides no definitive mechanism to understand how HS1BP3 works to alter PA levels. Other major points: 1. What are the ATG16L-TfR-positive membranes? How do the authors believe they contribute to autophagosome formation if they only contain ATG16L1 and no up or downstream Atg proteins. 2. The lipid binding properties of HS1BP3 are now better defined. The lipid analysis is very well done but it is done in whole fed cells (as far as is clear from the Methods) the major effects of HS1BP3 are in starved conditions which must produce an altered lipid profile. The inhibitor studies on the lipid enzymes are done in starvation which can not really be correlated with the lipid analysis done in fed cells. 4. The data does not uncover how the loss of HS1BP3 alters PA levels and impacts on the composition or formation of the phagophore.
Reviewer #3 (Remarks to the Author) I reviewed the original version of this manuscript as reviewer 3, I am completely content with the changes made by the authors and their responses to my concerns and criticisms and I regard the manuscript as acceptable for publication without any change.

Reviewers' comments:
Reviewer #1 (Remarks to the Author): The authors have responded comprehensively to the reviews. Proper controls and additional for experiments were included to strengthen the authors' data and provide a strong argument.
We thank the reviewer for these encouraging comments.
Reviewer #2 (Remarks to the Author): The formation of autophagosomes is a complex process which requires a complex machinery and multiple steps, involve both lipids, proteins and interdependencies between lipids and proteins. Holland et al have addressed the role of a lipid binding protein, HS1BP3, which binds PA. HS1BP3 is known to bind HS1 or cortactin. The authors demonstrate that HS1BP3 is a negative regulator of starvation-induced autophagy, which cololocalizes with Atg16L. HS1BP3 is proposed to act by inhibiting PLD1 activity on ATG16L1-positive and TfR-positive membranes. In addition, loss of HS1BP3 increases the synthesis of particular PA species in cells. The authors propose a model whereby HS1BP3 regulates PLD1's access to the Atg16L1, TfR-positive membranes thereby controlling the lipid composition, in particular PA levels of autophagosome precursor membranes. This is a revised version of a manuscript submitted to NCB. The authors have addressed some of the points raised in the previous review but they have done this primarily by removing large sets of data or replacing all the original problematic data. They have made a effort to identify an antibody to the endogenous protein but this has not provided insight into the mechanism. The manuscript, while much easier to understand, is now largely descriptive and provides no definitive mechanism to understand how HS1BP3 works to alter PA levels.
We thank the reviewer for the thorough review of our manuscript. We understand that the reviewer still has some questions as to the exact mechanism by which HS1BP3 works to alter PA levels and autophagy. We clearly show that HS1BP3 regulates PLD1 activity and the localization of PLD1 to ATG16L1 positive autophagosome precursor membranes. We propose that HS1BP3 competes with PLD1 for binding to autophagosome precursor membranes as both proteins contain a PX domain with similar lipid-binding specificities and colocalize with ATG16L1. In this revised version of the manuscript we include new data which strengthen this model as described in more detail below.
Other major points: 1. What are the ATG16L-TfR-positive membranes? How do the authors believe they contribute to autophagosome formation if they only contain ATG16L1 and no up or downstream Atg proteins.  Fig. 3b,c).

We and others have previously shown that recycling endosomes contain ATG16L1 as well as other ATG proteins and that they provide membrane for autophagosome formation. The Rubinsztein lab has reported that fusion of
2) The HS1BP3-ATG9 positive membranes colocalize with TfR in U2OS (new Fig. 3b).
Taken together, we conclude that the HS1BP3 and PLD1 positive ATG16L1/TfR structures observed in this manuscript are the same as the ATG16L1-ATG9-ULK1-TfR positive structures observed by others and that these structures are involved in autophagosome biogenesis.
2. The lipid binding properties of HS1BP3 are now better defined. The lipid analysis is very well done but it is done in whole fed cells (as far as is clear from the Methods) the major effects of HS1BP3 are in starved conditions which must produce an altered lipid profile. The inhibitor studies on the lipid enzymes are done in starvation which cannot really be correlated with the lipid analysis done in fed cells.
We apologies that it is not clear from the text (or the methods section) that the lipidomics analysis of control and HS1BP3 depleted cells indeed was done with cells treated with amino acid starvation. This has now been corrected in the text (p9), methods (p22) and figure legend (p 31).
We agree that it is important to use the same conditions when comparing lipid analysis and inhibitor studies, which is the case in this study. We would however like to point out that HS1BP3 depletion increases autophagy both in fed and starved conditions and it is likely that HS1BP3 also in fed cells functions by inhibition of PLD1 activity.