A Potential Role of the CD47/SIRPalpha Axis in COVID-19 Pathogenesis

Round 1

Reviewer 1 Report

Authors have designed a study to evaluate upregulation of cell surface glycoprotein CD47 and SIRPalpha level to determine severity of Covid-19 in patients. They observed enhanced CD47 expression in SARS-CoV-2-infected primary human bronchial epithelial cells and Calu-3 lung cancer cells, and increased SIRPalpha levels in primary human monocytes.

Several other literature reports have been analyzed to demonstrate upregulation of CD47 with ageing, diabetes, and obesity, and its relationship with Covid-19 severity.

Despite demonstrating a direct relation between CD47, SIRPalpha upregulation and Covid-19 progression, this study looks a lot like a meta-analysis.

Also, a more detailed analysis of role of CD47 expression in pathogen recognition, including Covid-19, has already been reported (https://doi.org/10.1128/mBio.01293-20, reference 67).

However, this study still offers significant information to readers by compiling various literature reports and their own analysis. Thus, it can be considered after addressing following minor points.

1. The introduction section should be improved by including more information on role of CD47 expression in pathogenesis of other diseases. Mention reference 67 in the introduction section.
2. It would be nice if different figure and tables can be combined as one file in supporting information.

Author Response

Please find our responses below. All changes are highlighted in yellow in the manuscript.

1. The introduction section should be improved by including more information on role of CD47 expression in pathogenesis of other diseases. Mention reference 67 in the introduction section.

Authors’ response:

We have modified the last paragraph of the Introduction as follows to address this comment (p. 2, lines 81-89):

“CD47 is the receptor of thrombospondin-1 (THBS1) and the counter-receptor for signal regulatory protein-α (SIRPα). CD47 interaction with SIRPα inhibits the activation of macrophages and dendritic cells and thrombospondin-1/ CD47 signaling inhibits T cell activation [22,23]. Cellular surface levels of CD47 modulate immune responses in infec-tious diseases caused by parasites, bacteria, and viruses [22]. Typically, high CD47 levels prevent immune recognition of virus-infected cells [22,24]. Moreover, cancer cells have been described to avoid immune recognition by upregulating CD47 [22]. Here, we in-vestigated the potential role of the ubiquitously expressed cell surface glycoprotein CD47 in severe COVID-19.”

2. It would be nice if different figure and tables can be combined as one file in supporting information.

Authors’ response:

As requested, we have combined the supplementary figures into one file.

Reviewer 2 Report

The immune system evolved intricate mechanisms to control activity of macrophages. Especially ‘eat me’ and ‘do not eat me’ tags, that signal macrophages what to do with a cell displaying these markers. Among them, the CD47-SIRPA axis defines an important ‘don’t eat me signal’. Thus, the presence or abundance of CD47 defines whether a cell will be taken up by a macrophage or not. This is especially important e.g. for cancers, which are frequently known to overexpress CD47 on their surface to prevent macrophage-mediated killing and immune activation. However, CD47 is also an interesting target for a virus, that tries to avoid immune activation and elimination.

In this brief manuscript, McLaughlin and co-workers identify CD47 as a molecule induced by SARS-CoV-2 in several cell lines and primary cells, but also in patients that died of COVID-19. This is followed by a literature analysis concluding that COVID-19 risk factors like diabetes or increased weight may be associated with increased CD47 levels. In conclusion, they suggest that high baseline CD47 expression may contribute to severe COVID19.

While the concept of the study is overall interesting and suggests that a well-known, druggable molecule (CD47) could be a therapeutic target against SARS-CoV-2, a few aspects need to be improved:

Major issues:

• The experimental data of the manuscript needs to be strengthened, for example by including experiment that in co-culture of macrophages and infected cells assess the ‘eating’ of infected cells by the macrophages and whether this can be enhanced by blocking CD47 (e.g. using antibodies). Is macrophage activity decreased (Co-cultures) and consequently replication enhanced in cells that have high CD47 baseline expression (e.g. by overexpression) as less cells are eliminated? Alternatively, are SARS-CoV-2 infected cells with CD47 ko increasingly/better controlled/eliminated by macrophages?
• The post-mortem analysis with two samples is insufficient to draw any conclusion (Fig 1C) and the statistical analysis invalid. The literature search could be extended to include datasets of other people that looked at gene regulation in patients (e.g. Desai et al, 2020, Nat Comms) to have access to more patient-derived data.
• It is not completely transparent why certain studies were excluded from the literature search except that two independent authors did not think they were relevant or they did not contain any original data. Were there any distinct criteria for exclusion set?

Minor issues:

The introduction would benefit from a sentence or two about risk factors, e.g. which ones are the most important ones to later explain better why three risk factors were chosen for literature analysis.

Fig 1A: What is the unit of the y-axis? is it normalized? Why are levels of CD47 decreasing in non-infected cells? A later time-point (48h) could be included to get a more convincing increase of CD47 in infected samples.

Fig 1B+D: The infected samples could be labelled with ‘SARS-CoV-2’.

Fig 1B+2B: A size marker could be added.

Fig 1C: There is a grey bar next to the plot.

Fig 2B: Expression levels need to be quantified.

line 110-112: Induction of SIRPA could be by immune sensing of SARS-CoV-2 in the SN e.g. by TLRs, which would be a more convincing explanation.

The authors should discuss, why decreased immune activation via macrophages would lead to a more severe disease defined by excessive immune activation.

Is increased replication of SARS-CoV-2 and less elimination of SARS-CoV-2 infected cells by macrophages observed in serve cases?

Author Response

Please find our responses below. All changes are highlighted in yellow in the manuscript.

Major issues:

The experimental data of the manuscript needs to be strengthened, for example by including experiment that in co-culture of macrophages and infected cells assess the ‘eating’ of infected cells by the macrophages and whether this can be enhanced by blocking CD47 (e.g. using antibodies). Is macrophage activity decreased (Co-cultures) and consequently replication enhanced in cells that have high CD47 baseline expression (e.g. by overexpression) as less cells are eliminated? Alternatively, are SARS-CoV-2 infected cells with CD47 ko increasingly/better controlled/eliminated by macrophages?

Authors’ response:

We agree that these are important questions. However, they are beyond the scope of this initial study, which is meant to raise awareness for the potential involvement of CD47 and SIRP alpha in the COVID-19 pathology in the face of an ongoing pandemic. Notably, CD47 may have multifaceted roles in COVID-19, and it is not clear whether the virus-induced effects or increased CD47 levels in risk groups are more important. Therefore, we feel that the suggested experiments will not provide a definite answer on the involvement of CD47 and SIRP alpha in SARS-CoV-2 pathology. However, they would delay the presentation and distribution of this potentially relevant information and, in turn, prevent other research groups from addressing the role of CD47/ SIRP alpha in their ongoing research.

The post-mortem analysis with two samples is insufficient to draw any conclusion (Fig 1C) and the statistical analysis invalid. The literature search could be extended to include datasets of other people that looked at gene regulation in patients (e.g. Desai et al, 2020, Nat Comms) to have access to more patient-derived data.

Authors’ response:

We agree that the significance of these findings is limited. However, the data are clearly presented so that they enable every reader to come to their own conclusions on this. To gain further insights, a dataset would be needed in which the CD47 levels are directly compared in infected tissues and appropriate non-infected control tissues from the same patient. Such data were not available from the suggested study, and we were not able to identify a study with data that would have enabled such a comparison. However, we have identified two additional studies that suggest a role of CD47/ SIRP alpha interaction in severe COVID-19 based on patient data. The Discussion (p. 10, lines 306-309):

“Moreover, elevated SIRP-α expression was recently reported in blood mononuclear cells of COVID-19 patients [75], and CD47/ SIRP-α interaction was associated with lung damage in severe COVID-19 [76].”

It is not completely transparent why certain studies were excluded from the literature search except that two independent authors did not think they were relevant or they did not contain any original data. Were there any distinct criteria for exclusion set?

Authors’ response:

Only articles were considered that contained original data on the effect of the conditions of interest on CD47 levels. Annotated article lists are provided in the Tables S1-S4.

Minor issues:

The introduction would benefit from a sentence or two about risk factors, e.g. which ones are the most important ones to later explain better why three risk factors were chosen for literature analysis.

Authors’ response:

The first paragraph of the Introduction was amended as follows to address this comment (p. 1, lines 46-47):

“Older age, being male, and conditions such as diabetes, hypertension, and obesity are associated with an increased risk of severe COVID-19 [1,4].”

Fig 1A: What is the unit of the y-axis? is it normalized? Why are levels of CD47 decreasing in non-infected cells? A later time-point (48h) could be included to get a more convincing increase of CD47 in infected samples.

Authors’ response:

The unit is the normalized signal intensity. This information was added to the figure legend. The data are derived from an existing dataset, which does not include a 48h time point.

Fig 1B+D: The infected samples could be labelled with ‘SARS-CoV-2’.

Authors’ response:

This was done.

Fig 1B+2B: A size marker could be added.

Authors’ response:

This was done.

Fig 1C: There is a grey bar next to the plot.

Authors’ response:

This was removed.

Fig 2B: Expression levels need to be quantified.

Authors’ response:

A quantification of the Western blot results is now provided in the new Figure S5.

line 110-112: Induction of SIRPA could be by immune sensing of SARS-CoV-2 in the SN e.g. by TLRs, which would be a more convincing explanation.

Authors’ response:

Such a mechanism cannot be excluded. However, we are not aware of any data that would support this in the context of a viral infection. However, other viruses and bacteria have already been shown to increase SIRP alpha levels in infected cells (Immunol Res. 2016 Feb;64(1):115-22. doi: 10.1007/s12026-015-8729-y; Nat Immunol. 2020 Jun;21(6):636-648. doi: 10.1038/s41590-020-0673-x). We included these references to clarify this (p. 10, lines 305-306):

“Notably, other viruses and bacteria have previously been described to increase host cell SIRPα levels [73,74].”

The authors should discuss, why decreased immune activation via macrophages would lead to a more severe disease defined by excessive immune activation.

Authors’ response:

The Discussion was amended as follows to address this comment (p. 10, lines 310-313):

“High SARS-CoV-2 loads are associated with more severe COVID-19 and a higher risk of patient death [77,78]. Therefore, high CD47 and/ or SIRPα levels may affect initial virus control resulting in enhanced virus levels, which may eventually lead to hyperinflammation and immunopathology observed in severe COVID-19.”

Is increased replication of SARS-CoV-2 and less elimination of SARS-CoV-2 infected cells by macrophages observed in serve cases?

Authors’ response:

The Discussion was amended as follows to address this comment (p. 10, lines 313-316):

“Moreover, innate immune responses appear to be critically involved in the early control of SARS-CoV-2 and deregulation of monocytes and macrophages seems to be a factor contributing to severe COVID-19 [79,80].”

Round 2

Reviewer 2 Report

I was honestly a bit surprised by the authors answers to my comments. I do believe that especially during a pandemic, published data should be reliable and not rushed, so that future research and therapy may be safely based on it. An overinterpreted hint from two patients combined with weak, incoherent in vitro data is not very helpful. There are plenty of speculative pieces already out there proposing all sorts of targets and therapies. I also do not agree that in its current form a delay of the publication of this piece would “prevent other research groups from addressing the role of CD47/ SIRP alpha in their ongoing research”, especially since “two additional studies that suggest a role of CD47/ SIRP alpha interaction in severe COVID-19 based on patient data”. This is no reason to abandon proper sample numbers and do statistics on two samples. Considering the authors own statement that the value of the post-mortem analysis is limited, I wonder why it should be included in a manuscript in the first place?

I do not endorse publication of this manuscript in its current state, with the author’s responses in mind. However, in light of CD47 potentially being important as well as the solid literature analysis, I propose that the authors should - as mentioned in their response- dramatically tone down the conclusions of the experimental data of the manuscript (“Here, we show in a range of model systems and data from post mortem samples that SARS-CoV-2 infection results in increased levels of CD47, which is known to mediate immune escape in cancer and virus-infected cells. “) and clearly state (e.g. in the abstract) that their intention is to raise awareness that CD47 is a potential target in COVID-19 but that the significance of their findings is limited (‘future studies are needed to …’) and encourage the reader to come to their own conclusion/use it as a starting point for further research. In addition, the post mortem analysis with two samples should be removed.

Author Response

I do not endorse publication of this manuscript in its current state, with the author’s responses in mind. However, in light of CD47 potentially being important as well as the solid literature analysis, I propose that the authors should - as mentioned in their response- dramatically tone down the conclusions of the experimental data of the manuscript (“Here, we show in a range of model systems and data from post mortem samples that SARS-CoV-2 infection results in increased levels of CD47, which is known to mediate immune escape in cancer and virus-infected cells. “) and clearly state (e.g. in the abstract) that their intention is to raise awareness that CD47 is a potential target in COVID-19 but that the significance of their findings is limited (‘future studies are needed to …’) and encourage the reader to come to their own conclusion/use it as a starting point for further research. 

Authors’ response:
We agree with the Reviewer that we do not want to provide a wrong impression or oversell our findings, and hope that we have implemented the suggested changes in a way that puts our findings into an appropriate context:

Abstract (p. 1, lines 29-31):
“Here, we show that levels of CD47, which is known to mediate immune escape in cancer and vi-rus-infected cells, are elevated in SARS-CoV-2-infected Caco-2 cells, Calu-3 cells, and air-liquid interface cultures of primary human bronchial epithelial cells.”

Abstract (p. 1, lines 39-42):
“Further research will be needed to investigate the potential involvement of CD47 and SIRPalpha in COVID-19 pathology. Our data should encourage other research groups to consider the potential relevance of the CD47/ SIRPalpha axis in their COVID-19 research.”

The first sentence of the Discussion reads now (p. 10, lines 287-288):
“Here, we show that levels of CD47 are elevated in SARS-CoV-2-infected Caco-2 cells, Calu-3 cells, and air-liquid interface cultures of primary human bronchial epithelial cells.”

The Conclusion finishes now (p. 11, lines 349-352):
“Further research will be needed to define the roles of CD47 and/ or SIRPalpha in COVID-19 in more detail. Thus, our findings should also encourage other research groups to consider the potential relevance of these molecules in their COVID-19 research.”

In addition, the post mortem analysis with two samples should be removed.

Authors’ response:
This was done. The respective Methods section reads now (p. 4, lines 163-165):
“Raw read counts from SARS-CoV-2-infected Calu-3 cells were derived from a recent publication [31] via the gene expression omnibus (GEO) database (accession: GSE147507) and processed using DESeq2.”

The Results section was changed to (p. 4, lines 184-185):
“Analysis of transcriptomics data from another study also indicated increased CD47 levels in SARS-CoV-2-infected Calu-3 cells (Figure S2) [31].”

The post mortem data were removed from Figure 1.

The first sentence of the Discussion reads now (p. 10, lines 287-288):
“Here, we show that levels of CD47 are elevated in SARS-CoV-2-infected Caco-2 cells, Calu-3 cells, and air-liquid interface cultures of primary human bronchial epithelial cells.”

Round 3

Reviewer 2 Report

No further comments.