Autophagy‐linked plasma and lysosomal membrane protein PLAC8 is a key host factor for SARS‐CoV‐2 entry into human cells

Abstract Better understanding on interactions between SARS‐CoV‐2 and host cells should help to identify host factors that may be targetable to combat infection and COVID‐19 pathology. To this end, we have conducted a genome‐wide CRISPR/Cas9‐based loss‐of‐function screen in human lung cancer cells infected with SARS‐CoV‐2‐pseudotyped lentiviruses. Our results recapitulate many findings from previous screens that used full SARS‐CoV‐2 viruses, but also unveil two novel critical host factors: the lysosomal efflux transporter SPNS1 and the plasma and lysosomal membrane protein PLAC8. Functional experiments with full SARS‐CoV‐2 viruses confirm that loss‐of‐function of these genes impairs viral entry. We find that PLAC8 is a key limiting host factor, whose overexpression boosts viral infection in eight different human lung cancer cell lines. Using single‐cell RNA‐Seq data analyses, we demonstrate that PLAC8 is highly expressed in ciliated and secretory cells of the respiratory tract, as well as in gut enterocytes, cell types that are highly susceptible to SARS‐CoV‐2 infection. Proteomics and cell biology studies suggest that PLAC8 and SPNS1 regulate the autophagolysosomal compartment and affect the intracellular fate of endocytosed virions.

membrane (rather than via endocytosis), particularly in lung cells. It would be interesting to know if PLAC8 affects infection of Calu3 cells or primary human lung cells that have been shown to primarily support the plasma membrane pathway of SARS-CoV-2 entry. Discussion of the basal levels of PLAC8 and SPNS1 in A549 and Calu3 cells would also be useful since these cell lines have been so extensively used in other SARS-CoV-2 publications. 2. The authors may also consider discussing whether their screen uncovered known restriction factors for SARS-CoV-2, such as Ly6E or CH25H, or also whether their data give insights into the controversial role of IFITM proteins in SARS-CoV-2 cell entry. 3. While the authors show that PLAC8/SPSN1 promote SARS-CoV-2 entry and that these proteins also affect autolysosomes, a cause-and-effect mechanism linking these two observed phenomena was not provided. Thus, it remains unclear how these factors enhance virus entry. 4. The supplementary tables need legends to make them interpretable. 5. Some of the asterisks indicating significance on various graphs throughout the manuscript are hard to distinguish from data points.
Referee #2: Ugalde et al. reported a novel entry factor for SARS-CoV-2. This factor, PLAC8, was identified through a genome-wide CRISPR screen in a human lung cancer cell line. The authors provided evidence to support the function of PLAC8 mediating SARS-CoV-2 entry. The RNA-seq analysis also confirmed the expression of PLAC8 in SARS-CoV-2 primary target cells in the human respiratory tract. Further, results suggested autophagy and lysosomal function may be regulated by PLCA8 in SARS-CoV-2 entry. The overall finding is potentially interesting. However, several important questions were not addressed which compromise the soundness and significance of this manuscript. Since PLAC8 is involved in multiple types of human tumors including lung cancer, PLAC8 role for viral infections in the lung cancer line may not be relevant unless confirmed in primary cells.
Major concerns: 1. Which step(s) of SARS-CoV-2 entry is affected in PLCA8/SPNS1 KO? Binding or internalization? The answer to this question will help understand the mechanisms of these two factors. 2. For SARS-CoV-2 entry, a very important factor, TMPRSS2, should be considered. This factor determines the entry pattern of SARS-CoV-2 in humans. The primary target cells always express TMPRSS2 and SARS-CoV-2 prefers to directly fuse with cell membrane but not through endocytosis. However, in the whole study, all the cell lines were lacking TMPRSS2 which might cause artificial effects. Therefore, observed effects should be confirmed in TMPRSS2+ cells. 3. Considering PLAC8 is also expressed in the cell membrane, it may provide a therapeutic target. Therefore, blocking antibody or purified PLAC8 protein could be used to block SARS-CoV-2 entry. This experiment will further validate the function of PLAC8. 4. Does PLAC8 directly interact with Spike? Many viral receptors/co-receptors interact with viral spike/glycoprotein to promote entry. 5. Although the authors put a lot of effort into RNA-seq data mining and found the expression pattern of PLCA8 matched the tropism of SARS-CoV-2, however, it is still essential to validate the function of PLCA8 in primary cells such as bronchial epithelial cells or nasal cells from humans. 6. How about other human coronaviruses like SARS-CoV-1, MERS-CoV? These viruses may also utilize these factors. 7. Fig 5F is not convincing because the ratio panel (merged images) is highly saturated as compared to their respective individual images. 8. levels of LC3 and p62 are different in Calu1ACE2 and H1299ACE2 cell lines and not much significant after knocking down of PLAC8?
Minor points: 1. Figure 3A. The y axis title should be "ratio", not "percentage" 2. Autophagy and lysosomal function are involved in SARS-CoV-2 replication and release but not entry. Not sure this figure could link the function of the two factors and viral entry.

Referee #3:
To identify targetable host factors for SARS CoV2 infections Ugalde et al performed a Crispr screen in human Calu1 lung cancer cells (overexpressing ACE2) using pseudotyped lentivirus. The authors pulled out two new factors, namely SPNS1 and PLAC8, that in loss of function experiments were confirmed to be involved in viral entry, including infections with the fully infectious SARS-CoV-2 virus. PLAC8 is highly expressed in ciliated and secretory cells from the respiratory tract and in gut enterocytes and its overexpression facilitates viral entry. Mechanistically the authors propose that SPNS1 and PLAC8 affect viral entry through regulation of autophagy and lysosomal function.
Though the proposed pathway is not novel and many questions remain on detailed mechanisms, this is certainly an interesting set of well conducted experiments that identify novel regulators of SARS-CoV-2.
I am not expert on Crispr screening (genome wide p values, proper controls, library used etc.) but the screen was done in any case to identify novel players in viral entry. The validation of the SPNS1 and PLAC8 knockout clones and their rescues, in particular, are very convincing. One question: in S. Fig. 1C, was the ACE2 clone (I assume the red bar is an ace2 wild type parent clone and the green bar an ace2 KO clone, would be good to describe this better in the figure legend) an incomplete knockout clone considering that viral entry is only reduced by about 40%.
For reporting the data in Fig. 2B etc, "Bars represent the average and standard error of the mean (SEM) of the percentage of infected cells in each condition (three replicates), normalized to the non-targeting control cell line NT596." Was the intensity of the zsGREEN signal also reduced in the remaining cells that were infected or is this an all or none phenomenon? If the latter is true, how do you explain this -incomplete knockouts? Please discuss this, unless I missed it somehow.
As for the SARS-CoV2 infections in Figure 3, it would be important to also assess viral RNA in addition to "immunofluorescence of nucleocapsid (N) protein 24 h post infection (p.i.)." After all, if the hypothesis of the authors is correct then there should be also a marked decrease in viral replication.
We would like to thank the referees for providing constructive and valuable suggestions to improve and strengthen our manuscript. We have used these comments as the basis for revising the manuscript. We have comprehensively addressed all concerns and provide below point-by-point list of responses to the reviewers' specific comments. Modifications of the original main text are marked with Microsoft Word change tracking in the revised manuscript.

Referee #1
In this manuscript, the authors performed a CRISPR screen in a lung cancer cell line to identify factors that affect cell infection with SARS-CoV-2 Spike-pseudotyped lentivirus. They newly identified SPNS1 and PLAC8 as cellular proteins that increase Spike-dependent infection of pseudotyped virus and confirmed this with authentic SARS-CoV-2 infection. These conclusions are overall well supported with both gain and loss of function experiments in multiple lung cell lines. They went on to analyze single cell RNA sequencing results to show that SPNS1 is widely expressed while PLAC8 is more selectively expressed in cell types that often co-express the virus receptor ACE2. They then performed quantitative proteomics on WT cells compared to PLAC8 or SPNS1 KO cells. Pathway analysis of differentially expressed proteins implicated a role for these proteins in autophagolysosome pathways and metabolic processes. Involvement of the proteins in controlling lysosome numbers and acidity in cells was confirmed. Overall, the experiments are well done, the provided data are convincing, and the discovery of these new factors in SARS-CoV-2 infection is an important addition to our fundamental knowledge of SARS-CoV-2 cellular entry.
Minor concerns:

The authors ignore the significant body of literature that suggests that SARS-CoV-2 enters cells via fusion at the plasma membrane (rather than via endocytosis), particularly in lung cells. It would be interesting to know if PLAC8 affects infection of Calu3 cells or primary human lung cells that have been
shown to primarily support the plasma membrane pathway of SARS-CoV-2 entry. Discussion of the basal levels of PLAC8 and SPNS1 in A549 and Calu3 cells would also be useful since these cell lines have been so extensively used in other SARS-CoV-2 publications.
We tested these cell lines in an initial round for suitable cellular models to study SARS-CoV-2. However, we discarded them because of the low infection efficiency using Spike-pseudotyped virus, even upon ACE2 overexpression. Nonetheless, as per your request, we have generated ACE2-RFP, GFP-PLAC8 and GFP-SPNS1 overexpression cell lines from Calu3, A549, and VeroE6 cells (another popular cell line in SARS-CoV-2 research from African Green Monkey) and tested them for infection efficiency using our flow-cytometry-based assays (see attached figure 1). These experiments produced again low ratios of infection in the parental cell lines, and ACE2-RFP overexpression failed to increase it. In contrast, the same batch of viruses yielded a 20% infection rate in Calu1 ACE2 cells. Notably, the infection efficiency with VSVG-typed virus is also very low in these cell lines. However, despite the overall low infection rate, GFP-PLAC8 overexpression resulted in a highly reproducible 3-to 6-fold increase in infection efficiency with S-typed viruses in A549 and VeroE6 cells. Interestingly, GFP-PLAC8 overexpressing VeroE6 cells without ACE2 overexpression also experienced that increase, probably because these cells have high basal expression of ACE2.
We do not know the reasons why Calu3 and A549 cells produce overall low infection rate in our infection assays with pseudotyped viruses. The main cell lines used in our work, as well as Calu3 and A549 were tested by STR genotyping and their identities were confirmed. There is a vast abundance of SARS-CoV-2 research literature; a detailed analysis of the experimental procedures used in other works that employed these cells might identify the origin of these discrepancies. We would like to point out that we used flow-cytometry as readout of infection, which allows to calculate the percentage of 23rd Jun 2022 1st Authors' Response to Reviewers infection cells, in contrast to luciferase reporters that yield a relative mean measurement of the whole population. A possible explanation could also be a different infection susceptibility to pseudotyped lentiviruses and full SARS-CoV-2 viruses. For example, plain VeroE6 cells were included as a control in our experiments with full SARS-CoV-2 viruses and showed the highest ratio of infection (around 10% more than Calu1 ACE2 cells), while their infection susceptibility to Spike-and VSVG-typed lentiviruses was very low (attached figure 1).
Nonetheless, we would like to point out that all the cellular models (up to 8 different cell lines) are lung cancer lines, like Calu3 or A549, and they have similar levels of TMPRSS2 expression ( Figure S2B in the revised manuscript, see point 2 from referee 2). The fact that GFP-PLAC8 overexpression in the attached experiment increases the infection rate of A549 and VeroE6 cells using Spike-typed lentiviruses support our conclusions and extend them to another primate species.

Figure 1: infection susceptibility of Calu3, A549 and VeroE6 cell with tagRFP or ACE2-RFP overexpression and GFP, GFP-PLAC8 or GFP-SPNS1 overexpression using Spike-typed (S-typed) or VSVG-typed lentiviruses. Bars represent the mean and SEM of three biological replicates.
2. The authors may also consider discussing whether their screen uncovered known restriction factors for SARS-CoV-2, such as Ly6E or CH25H, or also whether their data give insights into the controversial role of IFITM proteins in SARS-CoV-2 cell entry.
We are sorry if we missed some relevant SARS-CoV-2 host factors in our discussion. The SARS-CoV-2 literature grows very fast, and it is difficult to be up to date of all relevant discoveries. Ly6E ranks as the 62 nd most enriched knock-out gene in our screen, although it does not reach significance after multiple test correction. This observation would agree with the reported role of Ly6E in preventing SARS-CoV-2 membrane fusion. However, we do not see any change in the abundance of CRISPR guides against CH25H or IFITM proteins in our screen results. Nonetheless, Table S1 shows the screen statistics for every gene in our CRISPR-Cas9 screen to allow inspection of any gene of interest. Related to your point, we have included a short sentence in the discussion to point out that PLSCR1, a putative PLAC8 inhibitor, ranks 5 th in the list of most enriched knock-out genes in our screen.

While the authors show that PLAC8/SPSN1 promote SARS-CoV-2 entry and that these proteins also affect autolysosomes, a cause-and-effect mechanism linking these two observed phenomena was not provided. Thus, it remains unclear how these factors enhance virus entry.
In the revised manuscript, we have performed an extensive characterization of autophagy using more suitable techniques and experiments. We have replaced the Western-blot analyses of the original manuscript with a more complete immunofluorescence study of LC3B and p62, and autophagic flux analyses with the reporter GFP-LC3-RFP-LC3ΔG ( Figure 6C-F, S7 and S8 in the revised manuscript). In addition, we have extended these studies to cell lines with GFP-PLAC8 and GFP-SPNS1 overexpression. This set of experiments have helped us to reach a more detailed picture of the role of the autophagolysosomal alterations upon PLAC8 and SPNS1 gain-and loss-of-function in the context of SARS-CoV-2 infection. All these new developments have been discussed in the revised manuscript.

The supplementary tables need legends to make them interpretable.
This has been fixed in our revised manuscript.

Some of the asterisks indicating significance on various graphs throughout the manuscript are hard to distinguish from data points.
Thank you for your suggestion. We have increased the size of the asterisks in our revised manuscript to make them more clear. Most of our efforts in this round of revision were focused on trying to elucidate the molecular mechanisms underlying PLAC8 and SPNS1 function as SARS-CoV-2 host factors. We carried out binding and endocytosis experiments using recombinant SARS-CoV-2 RBD Spike protein ( Figure 5A-C in the revised manuscript). These experiments suggest that these two processes are not affected by loss-offunction of PLAC8 and SPNS1. Likewise, experiments with transferrin, which uses a different receptor, did not yield differences in the endocytosis rate upon loss-of-function of these two novel host factors. However, confocal microscopy analyses demonstrate that endosomes containing endocytosed Spike-RBD colocalize with vesicles labelled with ACE2-RFP, GFP-PLAC8 and GFP-SPNS1 ( Figure 5D-E), which provides molecular evidence of their functional relationship. In this revised manuscript we have also performed a detailed study of the autophagolysosomal compartment ( Figure 6C-F and S7-9 in the revised manuscript). Loss-of-function of PLAC8 and SPNS1 increases autophagic activity and produce an expansion and acidification of lysosomes. Despite their alterations, lysosomes seem to be functional, since autophagic flux experiments show no block in autophagy and the uptake and degradation of fluorogenic BSA is not affected. These experiments clearly indicate that PLAC8 and SPNS1 are molecularly connected to autophagolysosomal homeostasis. However, the observation that GFP-PLAC8 and GFP-SPNS1 overexpression does not affect this compartment in cell lines that experience increased infection susceptibility to SARS-CoV-2 does not support a causal relationship.

Ugalde et al. reported a novel entry factor for SARS-CoV-2. This factor, PLAC8, was identified through a genome-wide CRISPR screen in a human lung cancer cell line. The authors provided evidence to support the function of PLAC8 mediating SARS-CoV-2 entry. The RNA-seq analysis also confirmed the expression of PLAC8 in SARS-
Altogether, we think that our revised manuscript provides more clues about the molecular function of PLAC8 and SPNS1 in the context of SARS-CoV-2 infection, although we could not reach a fully understanding of the exact molecular mechanisms. All these new observations have been discussed in the revised manuscript more extensively.
2. For SARS-CoV-2 entry, a very important factor, TMPRSS2, should be considered. This factor determines the entry pattern of SARS-CoV-2 in humans. The primary target cells always express TMPRSS2 and SARS-CoV-2 prefers to directly fuse with cell membrane but not through endocytosis. However, in the whole study, all the cell lines were lacking TMPRSS2 which might cause artificial effects. Therefore, observed effects should be confirmed in TMPRSS2+ cells.
In our revised manuscript, we have included a Western-blot of TMPRSS2 in our main cellular models as well as in Calu3, A549 and VeroE6 cells for comparison (see minor point 1 from referee 1). As you can see in Figure S2B in our revised manuscript, TMPRSS2 shows an overall ubiquitous pattern of expression in all tested cell lines. These results differ from those of Koch et al (EMBOJ, 40: e107821, 2021), where they see very high expression of TMPRSS2 in Calu3 compared to VeroE6 and A549. We would like to point out that we employed the same antibody (Abcam ab92323) and that the main cell lines used in our work have been confirmed by STR genotyping. It might be possible that culture conditions or differences in experimental procedures explain this discrepancy. Nonetheless, in response to point 1 from reviewer 1 we have tested the infection efficiency of these 3 cell lines and found that GFP-PLAC8 overexpression boosts infection with Spike-typed lentiviruses both in A549 and VeroE6 (TMPRSSin Koch et al.).
3. Considering PLAC8 is also expressed in the cell membrane, it may provide a therapeutic target. Therefore, blocking antibody or purified PLAC8 protein could be used to block SARS-CoV-2 entry. This experiment will further validate the function of PLAC8.
We had also considered to explore this point as a potential therapeutic treatment. However, a previous study clearly demonstrates that PLAC8 localizes to the inner plasma membrane (Kaistha et al, Cancer Res (2016) 76 (1): 96-107), which dissuaded us from exploring this intervention. In addition, our revised manuscript now shows that binding and internalization of recombinant Spike is not affected by PLAC8 loss-of-function ( Figure 5 in the revised manuscript). Likewise, immunoprecipitation experiments of ACE2-GFP and GFP-PLAC8 in protein extracts that have been incubated with recombinant Spike protein did not support a direct interaction of PLAC8 with ACE2 or Spike protein ( Figure S5C-D in the revised manuscript). All these observations rule out that anti-PLAC8 antibody treatment would influence SARS-CoV-2 infection.

Does PLAC8 directly interact with Spike? Many viral receptors/co-receptors interact with viral spike/glycoprotein to promote entry.
As mentioned in the answer to your previous question, we have included immunoprecipitations experiments in the revised manuscript ( Figure S5C-D in the revised manuscript). Specifically, protein extracts from Calu1 cells that overexpress either ACE2-GFP or GFP-PLAC8 were incubated with recombinant Spike and subjected to immunoprecipitation with anti-GFP antibodies. While ACE2-GFP immunoprecipitation clearly brought down Spike protein, GFP-PLAC8 immunoprecipitation failed to retrieve recombinant Spike or ACE2, suggesting that PLAC8 does not function as an alternative receptor (also discarded by its inner membrane localization, see answer to previous point) and does not interact directly with ACE2. However, as discussed in the manuscript, it is possible that our immunoprecipitation experiments fail to detect weak or context-specific interactions.

Although the authors put a lot of effort into RNA-seq data mining and found the expression pattern of PLCA8 matched the tropism of SARS-CoV-2, however, it is still essential to validate the function of PLCA8 in primary cells such as bronchial epithelial cells or nasal cells from humans.
During this round of revision, we have tried to extend our observations to primary cells. Thus, we purchased HBEC cells and tried to perform functional studies. Unfortunately, these experiments take longer and are more complex than we anticipated. In addition to a delivery delay, these cells grow very slow and were very sensitive to antibiotics selection when trying to generate PLAC8 and SPNS1 gain-and loss-of-function models in a pilot experiment. Likewise, we attempted in parallel to use human colon organoids, but our lack of experience with this model precluded us to get results during this revision. We agree with you that a validation in human primary lung cells would further support our conclusions. However, we do not think this is an essential experiment for this manuscript. We have employed up to 8 different lung cancer cell lines and, in response to reviewer 1 point 1, we have also tested A549 and VeroE6 cells and confirmed our results about PLAC8. Moreover, during this round a revision another genome-wide CRISPR screen (in human hepatoma HuH7.5 cells) has reported that PLAC8 is a key host factor for the related coronavirus SADS-CoV (Tse et al., PNAS 2022 May 3;119(18):e2118126119).
6. How about other human coronaviruses like SARS-CoV-1, MERS-CoV? These viruses may also utilize these factors.
Although we agree with you that this would be an interesting point to test, when balancing the difficulty of the experiment and the relevance of the results, we decided to discard it. SARS-CoV-1 and MERS-CoV have very low population incidence and finding viral samples, obtaining permissions and access to an authorized laboratory would be very time and cost consuming. Moreover, new cellular models overexpressing the MERS-CoV receptor (DDP4) would be required. However, it is very likely that PLAC8 would also be relevant in SARS-CoV-1 and MERS-CoV infection, since a recent report has also identified PLAC8 as a main host factor for the related coronavirus SADS-CoV (Tse et al., PNAS 2022 May 3;119(18):e2118126119). Related to your question, a pilot experiment of some SARS-CoV-2 variants in H226-ACE2-RFP cells shows that GFP-PLAC8 overexpression boosts infection efficiency in all variants tested (see attached Figure 2), using our flow-cytometry based assays.

Fig 5F is not convincing because the ratio panel (merged images) is highly saturated as compared to their respective individual images.
H226-ACE2−RFP U n i n f e c t e d W u h a n W u h a n − ∆ 1 9 W u h a n − ∆ 1 8 W u h a n − D Thank you for pointing that out. We always take care of avoiding over-exposition. We double checked the images with the ImageJ HiLow look-up table and only few spots were overexposed. Nonetheless, we have included less exposed versions in the revised manuscript.

levels of LC3 and p62 are different in Calu1ACE2 and H1299ACE2 cell lines and not much significant after knocking down of PLAC8?
In the revised manuscript, we have performed an extensive characterization of autophagy using more suitable techniques and experiments. We have replaced the Western-blot analyses of the original manuscript with a more complete immunofluorescence study of LC3B and p62 and autophagic flux analysis with the reporter GFP-LC3-RFP-LC3ΔG ( Figure 6C-F, S7 and S8 in the revised manuscript). In addition, we have extended these studies to cell lines with GFP-PLAC8 and GFP-SPNS1 overexpression. This set of experiments have helped us to reach a more detailed picture of the role of the autophagolysosomal alterations upon PLAC8 and SPNS1 gain-and loss-of-function in the context of SARS-CoV-2 infection.
Minor points: 1. Figure 3A. The y axis title should be "ratio", not "percentage" We have fixed this issue in the revised manuscript.

Autophagy and lysosomal function are involved in SARS-CoV-2 replication and release but not entry. Not sure this figure could link the function of the two factors and viral entry.
As explained in response to point 8, our revised manuscript explores in much more detail the role of autophagolysosomal alterations upon PLAC8 and SPNS1 gain-and loss-of-function in the context of SARS-CoV-2 infection. All these new developments have been discussed in the revised manuscript.

To identify targetable host factors for SARS CoV2 infections Ugalde et al performed a Crispr screen in human Calu1 lung cancer cells (overexpressing ACE2) using pseudotyped lentivirus. The authors pulled out two new factors, namely SPNS1 and PLAC8, that in loss of function experiments were confirmed to be involved in viral entry, including infections with the fully infectious SARS-CoV-2 virus. PLAC8 is highly expressed in ciliated and secretory cells from the respiratory tract and in gut enterocytes and its overexpression facilitates viral entry. Mechanistically the authors propose that SPNS1 and PLAC8 affect viral entry through regulation of autophagy and lysosomal function.
Though the proposed pathway is not novel and many questions remain on detailed mechanisms, this is certainly an interesting set of well conducted experiments that identify novel regulators of SARS-CoV-2.
I am not expert on Crispr screening (genome wide p values, proper controls, library used etc.) but the screen was done in any case to identify novel players in viral entry. The validation of the SPNS1 and PLAC8 knockout clones and their rescues, in particular, are very convincing. One question: in S. Fig. 1C, was the ACE2 clone (I assume the red bar is an ace2 wild type parent clone and the green bar an ace2 KO clone, would be good to describe this better in the figure legend) an incomplete knockout clone considering that viral entry is only reduced by about 40%.
For reporting the data in Fig. 2B

etc, "Bars represent the average and standard error of the mean (SEM) of the percentage of infected cells in each condition (three replicates), normalized to the non-targeting control cell line NT596." Was the intensity of the zsGREEN signal also reduced in the remaining cells that were infected or is this an all or none phenomenon? If the latter is true, how do you explain this -incomplete knockouts? Please discuss this, unless I missed it somehow.
ZsGreen expression occurs in cells that have integrated the lentiviral plasmid delivered by the Spiketyped lentivirus. We used an intensity cutoff based on a non-infected sample to distinguish between infected and non-infected cells. While the histogram for non-infected samples has the classic bell shape, ZsGreen intensity histograms in samples from infected cells have a binormal distribution with two clear peaks (one that matches the single peak of a non-infected sample and a clearly shifted one that correspond to infected cells). Therefore, the readout is closer to an all or none phenomenon with a positional effect variability (differences in ZsGreen expression arising from the site of genomic insertion). The attached Figure 3 bellow is an example of a typical intensity histogram from cells coinfected with mCherry VSVG-typed and ZsGreen Spike-typed viruses.
Regarding the "incomplete knockouts", we do not see any discrepancy with the observed data. All the cellular models used in this work are polyclonal, not single clones. Although the knock-out efficiency is very high according to our Western-blot analyses, this technique only gives the average protein levels in the whole population, but there will be for sure wild-type or haploinsufficient cells that can be infected.
We hope we understood and addressed your question correctly.

As for the SARS-CoV2 infections in Figure 3, it would be important to also assess viral RNA in addition to "immunofluorescence of nucleocapsid (N) protein 24 h post infection (p.i.)." After all, if the hypothesis of the authors is correct then there should be also a marked decrease in viral replication.
We chose nucleocapsid immunofluorescence because it is a highly abundant protein in SARS-CoV-2 infected cells and there are polyclonal antibodies that provide an excellent signal. Nucleocapsid immunofluorescence has been used extensively throughout SARS-CoV-2 literature. In contrast, we reasoned that immunofluorescence of double strand RNA would yield a lower signal to noise ratio than nucleocapsid detection. In addition, since our screen only addresses viral entry, we did not find relevant to look at viral replication. While referee #1 is happy with the revised version, referee #2 has some remaining concerns with the revision. In light of these comments, I asked referee #1 to take a look at the points and have now heard back from the referee. Referee #1 appreciates the points raised by referee #2, but also find that you have responded to the raised concerns in a satisfactory manner. I have also looked at the concerns raised and I am in agreement with referee #1.
Given this I am therefore very pleased to let you know that we will accept this MS for publication here. No further experimental changes are needed. Before sending you the formal accept letter, there are just a few editorial points that we need to sort out: -Please include a point-by-point response to the remaining points of referee #2.
-Please add 3-5 keywords -COI needs to be re-labelled as Disclosure statement and competing interests -Please double check the grant number give for SAF2017-87655-R. Has different numbers in MS text vs online submission system.
-The ToC for the appendix needs page numbers adding. The nomenclature needs to be corrected to "Appendix Figure S#" and "Appendix Table S#" throughout. Appendix Table S2 is labelled "Appendix Table S4".
-In the data availability section please make the data sets available and remove referee tokes. Also please update the sentence" scRNA-Seq datasets of from lung , ileum and colon samples from healthy donors were downloaded from the UCSC Cell Browser (Karin: please add). The datasets of lung samples from COVID-19 patients were downloaded from the UCSC Cell Browser and from (Karin please add).
-We encourage the publication of source data, particularly for electrophoretic gels and blots but also for quantitative data, with the aim of making primary data more accessible and transparent to the reader. The source data should be uploaded as one file per figure. For blots please provide the original, uncropped and unprocessed scans and molecular weight markers; further annotation could be useful but is not essential. If multiple different types of files per figure zip them into one file per figure. The files should be labelled with the appropriate figure/panel number. The files will be published online with the article as supplementary "Source Data" files.
-Our publisher has also done their pre-publication check on your manuscript. When you log into the manuscript submission system you will see the file "Data Edited Manuscript file". Please take a look at the word file and the comments regarding the figure legends and respond to the issues.
-We include a synopsis of the paper (see http://emboj.embopress.org/). Please provide me with a general summary statement and 3-5 bullet points that capture the key findings of the paper. The authors have adequately addressed my concerns through experimental efforts. This study convincingly identifies new cellular factors involved in SARS-CoV-2 infections and links this effect to regulation of autophagy. This is a novel and important contribution to the field.
Referee #2: We appreciate the efforts authors made to address previously raised questions. However, major concerns were not resolved in this revision as outlined below.
1. The answers are disappointing. Purified RBD of Spike can't mimic the entry of SARS-CoV-2 because viral entry is mediated by the full length S1 and S2 with correct conformation of spike trimer. RBD alone (partial of S1) is not capable to mimic the functional full-length spike on virions. Therefore, the new data can't help making a solid conclusion about whether viral binding or internalization is affected when PLCA8/SPNS1 is knocked out.
2. These results are very confusing. This is a well-known fact in the field that TMPRSS2 is not expressed in A549 and Vero cells but highly expressed in Calu3 and differentiated primary NHBE cells. This phenomenon is confirmed by several labs and well documented in published literature. If the authors want to overturn this well-established conclusion, they should provide robust evidence like TMPRSS2 protein/mRNA evaluation in TMRPSS2 KO/KD conditions through multiple strategies. A simple WB is not sufficient and convincing.
Question 3 & 4: Because question 1 was not addressed in an appropriate way, there is not solid conclusion on it. Therefore, other associated questions are still there. 5: This question was not addressed.
6. This is the easiest and functionally relevant question to answer in the entire revision. Simply package MERS and SARS-1 pseudovirus and test them in the KO cells and we will know whether these two factors are involved in viral entry on the coronaviruses. Live virus is not required at all.

Referee #1:
The authors have adequately addressed my concerns through experimental efforts. This study convincingly identifies new cellular factors involved in SARS-CoV-2 infections and links this effect to regulation of autophagy. This is a novel and important contribution to the field.
We are pleased to hear that. We would like to thank the reviewer for his/her constructive criticisms and his/her positive comments on our work.

Referee #2:
We appreciate the efforts authors made to address previously raised questions. However, major concerns were not resolved in this revision as outlined below.
1. The answers are disappointing. Purified RBD of Spike can't mimic the entry of SARS-CoV-2 because viral entry is mediated by the full length S1 and S2 with correct conformation of spike trimer. RBD alone (partial of S1) is not capable to mimic the functional full-length spike on virions. Therefore, the new data can't help making a solid conclusion about whether viral binding or internalization is affected when PLCA8/SPNS1 is knocked out.
We are sorry to hear that. We are well aware that reductionist approaches have limitations.
One could also argue that soluble recombinant full-length trimeric Spike cannot recapitulate the biology of a virion. However, we do think that our experiments with Spike RBD provide helpful information to narrow down the underlying molecular mechanisms of these novel host factors.
Recombinant Spike RBD has been widely used in multiple works to study different aspects of Spike/ACE2 interaction (e.g Tai et al, 2020;Cao et al, 2022, 2). It is well-known that ACE2/SPIKE interaction is dictated by residues within the RBD, and as such, we think is fair to conclude from our binding experiments with S-RBD and the measurements of cell surface ACE2, that PLAC8 and SPNS1 do not play a role in this process. Regarding endocytosis, our experiments show that S-RBD is internalized in an ACE2-dependent manner and that PLAC8 and SPNS1 loss-of-function do not affect this process. This result probes that these novel host factors are not involved in ligand-triggered ACE2 internalization, the first step of SARS-CoV-2 infection. Further evidence of an intracellular role of these novel host factors is the fact that we see colocalization of internalized ACE2/S-RBD and PLAC8 and SPNS1. Other works have shown that binding of recombinant SARS-CoV or SARS-CoV-2 Spike RBD alone is sufficient to 5th Sep 2022 2nd Authors' Response to Reviewers induce the internalization of ACE2 (Bruno et al, 2018;Wang et al, 2008;Inoue et al, 2007) in a similar manner that full-length Spike (Bayati et al, 2021).
The main message of this paper is that we have unveiled two unknown novel host factors that were validated orthogonally with multiple loss-and gain-of-function experiments using full SARS-CoV-2 viruses in multiple cell lines. The functional relevance of these two genes, and specially PLAC8, is very solid. Science is built on cumulative knowledge and cooperation. We think that withholding our findings until we reach a full understanding of the underlying molecular mechanisms would impact negatively on science by hampering other labs to build from our discoveries. In this work we have identified PLAC8 and SPNS1 as novel host factors for SARS-CoV-2 infection. TMPRSS2 did not pop up in our screen. Although we could agree with the reviewer that studying the TMPRSS2 dependency of PLAC8 and SPNS1 might be interesting, we do not think that these would change the message of our work: PLAC8 is a key host factor for SARS-CoV-2 infection. It would probably be equally interesting to study the dependency on CTSL, which is a top hit in our screen and proposed to be highly important for Spike activation.

These results
Nonetheless, in this question the reviewer stated that none of our cell lines express TMPRSS2 (a reference supporting this statement would have been helpful) and therefore we included a Western-blot that shows otherwise, using a literature-validated antibody (Koch et al, 2021;Klouda et al, 2022;David et al, 2022). The bands fit with the expected molecular weight, and we see also cleaved TMPRSS2. We don't make any statement that try to overturn any wellstablished conclusion about TMPRSS2. In our manuscript we just state that TMPRSS2 was measured and was found expressed in most of our cellular models. A549, VeroE6 and Calu-3 were not vehicle in our manuscript and were only included for comparison in the revised manuscript.
We took this criticism very seriously and carried out an intensive search for evidence in the literature regarding TMPRSS2 expression in our cellular models. The figure below summarizes our findings. We apologize if we have overlooked any publication with relevant information on this matter. In general, it was hard to find works where the protein levels of TMPRSS2 were compared by western-blot in some of our cellular models. Surprisingly, we also found many works where statements about the expression levels of TMPRSS2 in their cellular models were not supported by experiments or citations (e.g. Hoffmann et al, 2020;Meng et al, 2022). As we mentioned in our previous response, Koch et al (Koch et al, 2021) show a black and white difference between A549 and Calu-3. A difference with this study is that they used Vero cells and ACE2-expressing A549, while we used Vero E6 and parental A549 (without ACE2 overexpression). However, we also found works that describe a different scenario. For example, Saccon et al. and Sagar et al. show similar levels of TMPRSS2 in Calu-3, A549 and VeroE6 cell lines (Saccon et al, 2021;Sagar et al, 2021). On the other hand, Kim et al. show higher levels of TMPRSS2 in Calu-3 than in A549, but they show strong signal in Vero cells (Kim et al, 2021).
Barret et al. detect higher levels of TMPRSS2 in A549 than in Vero cells (Barrett et al, 2021).
We also found pieces of evidence that contradict the idea that none of our cells express TMPRSS2. For example, Liu et al. use H1299 to study G-quadruplex regulation of TMPRSS2 expression and they show high levels of TMPRSS2 in untreated cells (Liu et al, 2022). Ou et al. also show signal for TMPRSS2 in H1299 cells, with levels similar to Calu-3 (Ou et al, 2021).
Finally, Pozzi et al. used Calu-1 cells to study TMPRSS2 inhibition by hydrogen sulfide and show multiple western-blots with strong signal for TMPRSS2 (Pozzi et al, 2021). In summary, the data we have found show important discrepancies within the literature, suggesting that TMPRSS2 expression and/or detection is influenced by the experimental procedures.
Nonetheless, to be on the safe side, we have repeated the Western-blot in Appendix Figure   S2C using new cell pellets and including additional controls: A549 and H358 cells expressing human full-length TMPRSS2 variant 1 (addgene plasmid #145843). These cells were generated at the beginning of the project when we were setting up a cellular model for the screen. As it can be seen in the figure below, the results are very similar than those presented in the revised manuscript: most of our cell lines have detectable TMPRSS2 and no large differences in TMPRSS2 protein levels are observed among the cell lines. However, increased levels of TMPRSS2 are very clear in A549 and H358 TMPRSS2-expressing cell lines, which demonstrates the specificity of the antibody. To have a better perspective, we also included LNcAP (Afar et al, 2001;Chen et al, 2010;Chu et al, 2014;Deng et al, 2021) and Caco-2 cells Cocozza et al, 2020;Chen et al, 2021;Zeng et al, 2022), which are supposed to express TMPRSS2. Please note that lysates from lanes 9-10 and 12-13 (parental and TMPRSS2-expressing A549 and H-358) were done separately with 1-day difference.
Question 3 & 4: Because question 1 was not addressed in an appropriate way, there is not solid conclusion on it. Therefore, other associated questions are still there.
Independently of question 1, questions 3 and 4 were addressed as follows.
In response to question 3, we provided detailed reasons of why it is very unlikely that PLAC8 antibodies would prevent viral entry. PLAC8 is located at the inner plasma membrane, which renders it inaccessible for an extracellular intervention with antibodies.
Regarding question 4, we performed immunoprecipitation experiments of ACE2 and GFP-PLAC8 in extracts that were incubated with full-length trimeric Spike protein (His-tagged).
While we captured the interaction between Spike and ACE2, we did not detect a direct interaction between PLAC8 and Spike.

5: This question was not addressed.
Indeed, this question was not addressed experimentally. While we agree with the reviewer that these experiments would provide with additional evidence about these genes, we think our work has multiple solid functional assays that probe that PLAC8, and SPNS1, are important SARS-CoV-2 host factors. We think that performing the requested experiments would unnecessarily delay several months the publication of our findings.
6. This is the easiest and functionally relevant question to answer in the entire revision. Simply package MERS and SARS-1 pseudovirus and test them in the KO cells and we will know whether these two factors are involved in viral entry on the coronaviruses. Live virus is not required at all. MERS-CoV uses a different receptor and therefore our ACE2-expressing cellular models would be meaningless. We would have to use DPP4 expressing cell lines or overexpress it in our models. In addition, we would have to order ORFs of MERS-CoV and SARS-CoV, and most likely generate C-terminal deletions to improve infection and to make it comparable to SARS-CoV-2 SpikeΔ19. Most importantly, although we agree that these experiments would yield useful information, they will not change the fact that PLAC8 and SPNS1 are host factors for SARS-CoV-2 infection, which is the message of our work. We think it is more important to readily share our findings with the research community so future follow-up studies can address all these associated questions.