Post by Admin on Mar 8, 2021 1:03:40 GMT
Discussion
Disease-specific therapies are needed to reduce the morbidity and mortality associated with COVID-19 outcomes. In this large-scale, two-sample MR study of 931 proteins assessed for three COVID-19 outcomes in up to 14,134 cases and 1.2 million controls of European ancestry, we provide evidence that increased OAS1 levels in the non-infectious state are strongly associated with reduced risks of very severe COVID-19, hospitalization and susceptibility. The protective effect size was particularly large, such that a 50% decrease in the odds of very severe COVID-19 was observed per s.d. increase in OAS1 circulating levels. OAS proteins are part of the innate immune response against RNA viruses. They are induced by interferons and activate latent RNase L, resulting in direct viral and endogenous RNA destruction, as demonstrated in in vitro studies33. Thus, OAS1 has a plausible biological activity against SARS-CoV-2. Because therapies exist that activate OAS1, repositioning them as potential COVID-19 treatments should be prioritized.
In populations outside of Sub-Saharan Africa, the protective alleles at both rs4767027-T (the OAS1 pQTL) and rs10774671-G (the OAS1 sQTL) are found on a Neanderthal haplotype34, which was passed on to modern humans ~50,000–60,000 years ago35. The correspondence between the previously described gene flow35 from Neandertals at this locus and the haplotype associated with protection against COVID-19 in the GWAS22 was recently demonstrated34. Even though these two single-nucleotide polymorphisms (SNPs) share a haplotype, their evolutionary histories differ. The rs4767027-T allele is derived from the Neanderthal lineage, whereas, for the rs10774671-G allele, Neanderthals preserved the ancestral state. OAS1 alternative splicing regulated by the rs10774671-G allele increases the isoform p46, which has a higher enzymatic activity against viruses than the p42 isoform36 and is the only OAS1 isoform robustly upregulated during infection26. Although further studies are needed to fully elucidate the functional relevance of the pQTL and sQTL for OAS1, the antiviral activity of the gene products is higher for the Neandertal haplotype than the common haplotype in Europeans28. In Europeans, the Neandertal haplotype has undergone positive selection26, and the rs4767027-T allele reaches an allele frequency of 0.32. Using MR and measurements of circulating proteins, we demonstrated here that increased OAS1 levels of the Neandertal haplotype in modern-day individuals of European ancestry confer this protective effect.
Our MR evidence indicated that higher p46 isoform levels of OAS1 and higher OAS1 total protein levels, as measured by the SomaScan assay, had protective effects on COVID-19 outcomes. These results were strongly supported by co-localization analysis. Given the consistent co-localization between the sQTL and pQTL for OAS1, the lack of co-localization between the eQTL and pQTL for OAS1 and the evidence that the SomaScan assay likely measures p46 isoforms, it seems probable that the protective effect of OAS1 is derived from the p46 isoform. However, further investigations are required to specifically measure each isoform in circulation, and isoform activity assays will be required to better understand if the p46 isoform, rather than total OAS1 levels, is most protective against COVID-19 outcomes.
The ancestral OAS1 splice variant encoding the more active p46 isoform was lost in the modern human population that left Africa. Several scenarios might explain this loss of function—for example, loss of purifying selection during the out-of-Africa exodus, which might be due to changes in environmental pathogens or potential harm induced by OAS1 antiviral activity37. Unfortunately, we do not have sufficient data to test if the OAS1 p46 ancestral allele in Sub-Saharan Africans also offers protection against COVID-19. Nevertheless, these findings further emphasize the importance of the Neanderthal genome in COVID-19 risk modulation, because a risk locus on chromosome 3 has also been reported to be inherited from Neanderthals38.
OAS1, OAS2 and OAS3 share considerable homology. As an interferon-stimulated gene39, OAS1 polymorphisms have been associated with the host immune response to several classes of viral infection40,41,42,43,44. Given that OAS1 is an intracellular enzyme-activating RNase L leading to viral RNA degradation, it is probable that the circulating levels of this enzyme reflect intracellular levels of this protein. However, there is experimental evidence that extracellular OAS1 might also be important in the viral immune response33.
Molecules currently exist that can influence OAS1 expression. Interferon beta-1b, which activates a cytokine cascade leading to increased OAS1 expression45, is currently used to treat multiple sclerosis and has been shown to induce OAS1 expression in blood cells46. Interferon-based therapy has also been used in other viral infections47. However, recent randomized trials have shown inconsistent results. Although intravenous interferon beta-1b combined with lopinavir–ritonavir reduced mortality due to MERS-CoV infections48, in the unblinded SOLIDARITY trial49 there was no demonstrated benefit of intravenous interferon-beta-1b. On the other hand, a recent phase 2 trial testing the effect of inhaled nebulized interferon beta-1a (which is closely related to interferon beta-1b) showed improved COVID-19 symptoms in the treatment arm50. Although this study was not powered to show a difference in mortality, all deaths occurred in the placebo group. Inhaled nebulized interferon beta results in a much higher tissue availability in the lung and might result in improved antiviral activity. Moreover, timing of administration is likely to play a role, as the administration of a pro-inflammatory cytokine might not provide benefit during the inflammation-driven phase of the disease. However, data on timing of administration are currently unavailable in the SOLIDARITY trial, and conclusions cannot yet be drawn. Lastly, the effect of interferon supplement might vary across ancestral populations, as different ancestries have different amounts of the more active p46 isoform of OAS1. Our study was limited to individuals of European ancestry, a population with higher expression of the p46 isoform. Interestingly, the SOLIDARITY trial enrolled 78% of its patients in South Asia, the Middle East, North Africa and Latin America, populations that might have higher expression of the p42 OAS1 isoform, whereas the study on inhaled interferon beta comprised 80% White patients from the United Kingdom. It is possible that interferon beta-1b might have different effects in populations of different ancestry due to different frequency of genetic variants in different populations.
There is in vitro evidence that pharmacological inhibition of phosphodiesterase-12, which degrades 2′–5′ oligoadenylate synthesized by OAS1, potentiates OAS-mediated antiviral activity51,52. Interestingly, coronaviruses in the same family as SARS-CoV-2 have been shown to produce viral proteins that degrade 2′–5′ oligoadenylate and reduce RNase-L activity, leading to evasion of the host immune response53,54. Our findings are also consistent with recent experimental work55 showing that there are situations where SARS-CoV-2 is sensitive to OAS1-related antiviral defenses. Our findings motivate pharmacologic strategies to increase OAS1 levels or activity, as well as further evaluation of the possible antiviral activity of extracellular OAS1 (ref. 33). Thus, existing preclinical molecules that lead to increased OAS1 levels51 could be optimized and tested for their effect on COVID-19 outcomes.
Our MR analyses found that higher levels of OAS3 expression is associated with worse COVID-19 outcomes, which is an opposite direction of effect compared to OAS1. The discordant effects of the p46 isoform for OAS1 and OAS3 were also reported by a previous study26, which might reflect complex biology of OAS genes for innate immune response. In a recent transcription-wide association study from the GenOMICC program22, genetically predicted high expression of OAS3 in lungs and whole blood was associated with a higher risk of patients with COVID-19 becoming critically ill. Although further studies to assess the roles of OAS genes specific to SARS-CoV-2 are needed, it is likely that OAS1 is the main driver of the protective effect of the p46 isoform for COVID-19 outcomes given previous functional studies demonstrating the antiviral effect of OAS genes26.
This study had limitations. First, we used MR to test the effect of circulating protein levels measured in a non-infected state because the effect of the cis-pQTLs on circulating proteins was estimated in individuals who had not been exposed to SARS-CoV-2. Once a person contracts SARS-CoV-2 infection, levels of circulating proteins could be altered, and this might be especially relevant for cytokines such as IL10 (which binds to IL10RB) and OAS1. Thus, the MR results presented in this paper should be interpreted as an estimation of the effect of circulating protein levels when measured in the non-infected state. Ongoing studies will help to clarify if the same cis-pQTLs influence circulating protein levels during infection. Second, this type of study suffers a high false-negative rate. Our goal was not to identify every circulating protein influencing COVID-19 outcomes but, rather, to provide evidence for a few proteins with strong cis-pQTLs, because these proteins are more likely to be robust to the assumptions of MR studies. Future large-scale proteomic studies with more circulating proteins properly assayed should help to overcome these limitations. Third, most MR studies assume a linear relationship between the exposure and the outcome. Thus, our findings would not identify proteins whose effect on COVID-19 outcomes has a clear threshold effect. Fourth, the overall OAS1 levels measured by RNA sequencing (not only p46) might be biased by the effect of alternative splicing, and the role of overall OAS1 and OAS3 levels indicated by the association of the cis-pQTL of OAS1 in protection against COVID-19 are possible and not yet explored. We also could not completely exclude the possibility that measurement of OAS1 levels might be influenced by aptamer-binding effects. Last, all data presented in this paper pertain to individuals of European ancestry only—once again underlining the importance of genotyping efforts in other populations.
In conclusion, we used genetic determinants of circulating protein levels and COVID-19 outcomes obtained from large-scale studies and found compelling evidence that OAS1 has a protective effect on COVID-19 susceptibility and severity. Measuring plasma OAS1 levels in a case–control study demonstrated that higher circulating levels of this protein in a non-infectious state are strongly associated with reduced risk of adverse COVID-19 outcomes. Interestingly, the available evidence suggests that the protective effect from OAS1 in individuals of European ancestry is likely due to the Neanderthal-introgressed p46 OAS1 isoform. Known pharmacological agents that increase OAS1 levels51 could be explored for their effect on COVID-19 outcomes.
Disease-specific therapies are needed to reduce the morbidity and mortality associated with COVID-19 outcomes. In this large-scale, two-sample MR study of 931 proteins assessed for three COVID-19 outcomes in up to 14,134 cases and 1.2 million controls of European ancestry, we provide evidence that increased OAS1 levels in the non-infectious state are strongly associated with reduced risks of very severe COVID-19, hospitalization and susceptibility. The protective effect size was particularly large, such that a 50% decrease in the odds of very severe COVID-19 was observed per s.d. increase in OAS1 circulating levels. OAS proteins are part of the innate immune response against RNA viruses. They are induced by interferons and activate latent RNase L, resulting in direct viral and endogenous RNA destruction, as demonstrated in in vitro studies33. Thus, OAS1 has a plausible biological activity against SARS-CoV-2. Because therapies exist that activate OAS1, repositioning them as potential COVID-19 treatments should be prioritized.
In populations outside of Sub-Saharan Africa, the protective alleles at both rs4767027-T (the OAS1 pQTL) and rs10774671-G (the OAS1 sQTL) are found on a Neanderthal haplotype34, which was passed on to modern humans ~50,000–60,000 years ago35. The correspondence between the previously described gene flow35 from Neandertals at this locus and the haplotype associated with protection against COVID-19 in the GWAS22 was recently demonstrated34. Even though these two single-nucleotide polymorphisms (SNPs) share a haplotype, their evolutionary histories differ. The rs4767027-T allele is derived from the Neanderthal lineage, whereas, for the rs10774671-G allele, Neanderthals preserved the ancestral state. OAS1 alternative splicing regulated by the rs10774671-G allele increases the isoform p46, which has a higher enzymatic activity against viruses than the p42 isoform36 and is the only OAS1 isoform robustly upregulated during infection26. Although further studies are needed to fully elucidate the functional relevance of the pQTL and sQTL for OAS1, the antiviral activity of the gene products is higher for the Neandertal haplotype than the common haplotype in Europeans28. In Europeans, the Neandertal haplotype has undergone positive selection26, and the rs4767027-T allele reaches an allele frequency of 0.32. Using MR and measurements of circulating proteins, we demonstrated here that increased OAS1 levels of the Neandertal haplotype in modern-day individuals of European ancestry confer this protective effect.
Our MR evidence indicated that higher p46 isoform levels of OAS1 and higher OAS1 total protein levels, as measured by the SomaScan assay, had protective effects on COVID-19 outcomes. These results were strongly supported by co-localization analysis. Given the consistent co-localization between the sQTL and pQTL for OAS1, the lack of co-localization between the eQTL and pQTL for OAS1 and the evidence that the SomaScan assay likely measures p46 isoforms, it seems probable that the protective effect of OAS1 is derived from the p46 isoform. However, further investigations are required to specifically measure each isoform in circulation, and isoform activity assays will be required to better understand if the p46 isoform, rather than total OAS1 levels, is most protective against COVID-19 outcomes.
The ancestral OAS1 splice variant encoding the more active p46 isoform was lost in the modern human population that left Africa. Several scenarios might explain this loss of function—for example, loss of purifying selection during the out-of-Africa exodus, which might be due to changes in environmental pathogens or potential harm induced by OAS1 antiviral activity37. Unfortunately, we do not have sufficient data to test if the OAS1 p46 ancestral allele in Sub-Saharan Africans also offers protection against COVID-19. Nevertheless, these findings further emphasize the importance of the Neanderthal genome in COVID-19 risk modulation, because a risk locus on chromosome 3 has also been reported to be inherited from Neanderthals38.
OAS1, OAS2 and OAS3 share considerable homology. As an interferon-stimulated gene39, OAS1 polymorphisms have been associated with the host immune response to several classes of viral infection40,41,42,43,44. Given that OAS1 is an intracellular enzyme-activating RNase L leading to viral RNA degradation, it is probable that the circulating levels of this enzyme reflect intracellular levels of this protein. However, there is experimental evidence that extracellular OAS1 might also be important in the viral immune response33.
Molecules currently exist that can influence OAS1 expression. Interferon beta-1b, which activates a cytokine cascade leading to increased OAS1 expression45, is currently used to treat multiple sclerosis and has been shown to induce OAS1 expression in blood cells46. Interferon-based therapy has also been used in other viral infections47. However, recent randomized trials have shown inconsistent results. Although intravenous interferon beta-1b combined with lopinavir–ritonavir reduced mortality due to MERS-CoV infections48, in the unblinded SOLIDARITY trial49 there was no demonstrated benefit of intravenous interferon-beta-1b. On the other hand, a recent phase 2 trial testing the effect of inhaled nebulized interferon beta-1a (which is closely related to interferon beta-1b) showed improved COVID-19 symptoms in the treatment arm50. Although this study was not powered to show a difference in mortality, all deaths occurred in the placebo group. Inhaled nebulized interferon beta results in a much higher tissue availability in the lung and might result in improved antiviral activity. Moreover, timing of administration is likely to play a role, as the administration of a pro-inflammatory cytokine might not provide benefit during the inflammation-driven phase of the disease. However, data on timing of administration are currently unavailable in the SOLIDARITY trial, and conclusions cannot yet be drawn. Lastly, the effect of interferon supplement might vary across ancestral populations, as different ancestries have different amounts of the more active p46 isoform of OAS1. Our study was limited to individuals of European ancestry, a population with higher expression of the p46 isoform. Interestingly, the SOLIDARITY trial enrolled 78% of its patients in South Asia, the Middle East, North Africa and Latin America, populations that might have higher expression of the p42 OAS1 isoform, whereas the study on inhaled interferon beta comprised 80% White patients from the United Kingdom. It is possible that interferon beta-1b might have different effects in populations of different ancestry due to different frequency of genetic variants in different populations.
There is in vitro evidence that pharmacological inhibition of phosphodiesterase-12, which degrades 2′–5′ oligoadenylate synthesized by OAS1, potentiates OAS-mediated antiviral activity51,52. Interestingly, coronaviruses in the same family as SARS-CoV-2 have been shown to produce viral proteins that degrade 2′–5′ oligoadenylate and reduce RNase-L activity, leading to evasion of the host immune response53,54. Our findings are also consistent with recent experimental work55 showing that there are situations where SARS-CoV-2 is sensitive to OAS1-related antiviral defenses. Our findings motivate pharmacologic strategies to increase OAS1 levels or activity, as well as further evaluation of the possible antiviral activity of extracellular OAS1 (ref. 33). Thus, existing preclinical molecules that lead to increased OAS1 levels51 could be optimized and tested for their effect on COVID-19 outcomes.
Our MR analyses found that higher levels of OAS3 expression is associated with worse COVID-19 outcomes, which is an opposite direction of effect compared to OAS1. The discordant effects of the p46 isoform for OAS1 and OAS3 were also reported by a previous study26, which might reflect complex biology of OAS genes for innate immune response. In a recent transcription-wide association study from the GenOMICC program22, genetically predicted high expression of OAS3 in lungs and whole blood was associated with a higher risk of patients with COVID-19 becoming critically ill. Although further studies to assess the roles of OAS genes specific to SARS-CoV-2 are needed, it is likely that OAS1 is the main driver of the protective effect of the p46 isoform for COVID-19 outcomes given previous functional studies demonstrating the antiviral effect of OAS genes26.
This study had limitations. First, we used MR to test the effect of circulating protein levels measured in a non-infected state because the effect of the cis-pQTLs on circulating proteins was estimated in individuals who had not been exposed to SARS-CoV-2. Once a person contracts SARS-CoV-2 infection, levels of circulating proteins could be altered, and this might be especially relevant for cytokines such as IL10 (which binds to IL10RB) and OAS1. Thus, the MR results presented in this paper should be interpreted as an estimation of the effect of circulating protein levels when measured in the non-infected state. Ongoing studies will help to clarify if the same cis-pQTLs influence circulating protein levels during infection. Second, this type of study suffers a high false-negative rate. Our goal was not to identify every circulating protein influencing COVID-19 outcomes but, rather, to provide evidence for a few proteins with strong cis-pQTLs, because these proteins are more likely to be robust to the assumptions of MR studies. Future large-scale proteomic studies with more circulating proteins properly assayed should help to overcome these limitations. Third, most MR studies assume a linear relationship between the exposure and the outcome. Thus, our findings would not identify proteins whose effect on COVID-19 outcomes has a clear threshold effect. Fourth, the overall OAS1 levels measured by RNA sequencing (not only p46) might be biased by the effect of alternative splicing, and the role of overall OAS1 and OAS3 levels indicated by the association of the cis-pQTL of OAS1 in protection against COVID-19 are possible and not yet explored. We also could not completely exclude the possibility that measurement of OAS1 levels might be influenced by aptamer-binding effects. Last, all data presented in this paper pertain to individuals of European ancestry only—once again underlining the importance of genotyping efforts in other populations.
In conclusion, we used genetic determinants of circulating protein levels and COVID-19 outcomes obtained from large-scale studies and found compelling evidence that OAS1 has a protective effect on COVID-19 susceptibility and severity. Measuring plasma OAS1 levels in a case–control study demonstrated that higher circulating levels of this protein in a non-infectious state are strongly associated with reduced risk of adverse COVID-19 outcomes. Interestingly, the available evidence suggests that the protective effect from OAS1 in individuals of European ancestry is likely due to the Neanderthal-introgressed p46 OAS1 isoform. Known pharmacological agents that increase OAS1 levels51 could be explored for their effect on COVID-19 outcomes.