Surprising Effects of Antibodies in Severe COVID

Home , CD32, Fc receptor

a Pyroptosis Toxin-mediated Necroptosis Apoptosis followed b Cellular swelling in c Rupture of the death by secondary these types of death plasma membrane Ion necrosis H O DAMPs 2 Na+, K+- Gasdermin Toxin MLKL ATPase NINJ1

Macrophage

Figure 1 | Regulated rupture of the plasma membrane is an end point of inactivation of the Na+, K+-ATPase enzyme causes ion accumulation in dying multiple cell-death pathways. a, Human cells, such as macrophages, can cells10. b, The water entry causes cellular swelling, and bubble-like protrusions die by a range of mechanisms, including pyroptosis, toxin-mediated death, form. Kayagaki et al.1 report that rupture of the plasma membrane in these types necroptosis and apoptosis (followed by a process called secondary necrosis). of dying cell does not occur passively, as previously thought. Instead, it is an A common feature of these deaths is an increase in cellular osmotic pressure, active process that requires the protein ninjurin-1 (NINJ1). c, To mediate rupture presumably arising from an ionic imbalance that drives water entry. This of the plasma membrane, NINJ1 aggregates (oligomerizes). This rupture releases imbalance is triggered by ion movement through protein channels or pores, such cytoplasmic content, including molecules called damage-associated molecular as those formed by gasdermin proteins, toxins or MLKL channels. In apoptosis, patterns (DAMPs), which trigger inflammation in neighbouring tissue.

of apoptotic cells. Secondary necrosis occurs if NINJ1, or related proteins, that could convert 1. Kayagaki, N. et al. Nature 591, 131–136 (2021). apoptotic cells are not engulfed and removed necrotic death to a type of death with a less 2. Green, D. R. Means to an End: Apoptosis and Other Cell Death Mechanisms (Cold Spring Harbor Press, 2011). in a timely manner. Thus, NINJ1 is a common inflammatory outcome. Such treatments 3. Galluzzi, L. et al. Cell Death Differ. 25, 486–541 (2018). denominator at the end of many cell-death would thereby reduce the general level of 4. Zhang, Y., Chen, X., Gueydan, C. & Han, J. Cell Res. 28, 9–21 (2018). pathways. inflammation in tissue, presumably with posi- 5. Broz, P., Pelegrín, P. & Shao, F. Nature Rev. Immunol. 20, 8 NINJ1 is ubiquitously expressed , and is tive effects for chronic or acute inflammatory 143–157 (2020). evolutionarily conserved, from fruit flies disorders. 6. Vandenabeele, P., Galluzzi, L., Vanden Berghe, T. & Kroemer, G. Nature Rev. Mol. Cell Biol. 11, 700–714 (2010). to humans. How might this relatively small 7. Araki, T. & Milbrandt, J. Neuron 17, 353–361 (1996). (16 kilodaltons) protein mediate such striking Sebastian Hiller is at Biozentrum, University of 8. Araki, T., Zimonjic, D. B., Popescu, N. C. & Milbrandt, J. effects? Its structure is predicted to contain Basel, 4056 Basel, Switzerland. Petr Broz is in J. Biol. Chem. 272, 21373–21380 (1997). 9. Fink, S. L. & Cookson, B. T. Cell Microbiol. 8, 1812–1825 two transmembrane helices, as well as an the Department of Biochemistry, University of (2006). evolutionarily conserved extracellular helix Lausanne, 1066 Epalinges, Switzerland. 10. Dijkstra, K., Hofmeijer, J., van Gils, S. A. & van Putten, M. J. A. M. J. Neurosci. 36, 11881–11890 (2016). that is needed for NINJ1 to function properly. e-mails: [email protected]; Working out whether this helix senses a sig- [email protected] This article was published online on 8 February 2021. nal or serves to disrupt the membrane during cell death will require more study. Of note, this Coronavirus helix seems to have a mixed hydrophobic and hydrophilic (amphiphilic) character, a property similar to that of the helices found in other membrane-disrupting proteins, such Surprising effects of as melittin or BAX. Importantly, Kayagaki and colleagues’ antibodies in severe COVID findings will transform cell biology in a way that goes beyond just revealing NINJ1’s func- Matteo Gentili & Nir Hacohen tion. Their study underscores the enormous strength and resilience of the intact plasma Defects in the immune defences induced by the protein membrane. It also reduces the number of interferon are associated with some severe cases of COVID-19. events in cell biology considered to be non specific, highlighting how stringently organ- An analysis of patients’ blood samples sheds light on how isms control the fate of their cells until the very antibodies might contribute to these defects. See p.124 last moment of cellular existence. Many questions remain to be answered. What signal or property is sensed by NINJ1 Infection with the SARS-CoV-2 virus can lead immune cells, in blood samples from 21 people to activate its function in dying cells? What to diverse outcomes, ranging from no symp- with COVID-19 and 25 uninfected individuals mechanisms, if any, exist to prevent acciden- toms to varying degrees of disease severity, who were either healthy or had a lung injury or tal activation of NINJ1? It would be interesting spanning mild illness to death. What deter- breathing difficulties. They monitored gene to know whether NINJ1 requires other factors mines the degree of severity is unclear, but expression during the course of the infection when mediating membrane rupture. Do other mounting evidence points to exacerbated as patients went on to develop either what proteins with a similar function exist? And, and abnormal responses in the innate branch was categorized as mild–moderate COVID-19 of course, what is the structure of the mem- of the immune system as a main driver of (which required a short hospital stay without brane-rupturing entity that NINJ1 presumably major illness. Combes et al.1 present a study the need for intensive care or mechanical forms? on page 124 investigating the hallmarks of ventilation) or severe COVID-19 (requiring Answering these questions might lead to COVID-19 severity. intensive care and mechanical ventilation). new therapeutic strategies aimed at inhibiting The authors analysed cells, including The authors found that the cells of people with

News & views the induction of interferon-induced genes. a Mild–moderate COVID-19 b Severe COVID-19 However, given that most individuals in that study did not have anti-interferon antibod- Interferon Antiviral ies, the presence of such antibodies alone molecule SARS-CoV-2 could not fully explain the development of severe COVID-19. Indeed, Combes et al. found Interferon-regulated Anti-spike anti-interferon antibodies at a similarly low defence gene antibody CD32B-binding frequency in the samples they had obtained CD32B Fc domain from patients. In an attempt to explain the enigma of a dampened interferon response in severe Monocyte COVID-19, the authors considered various aspects of antibody function. An antibody Recruited antiviral consists structurally of two functional immune cell units: a variable region that recognizes the disease-causing agent and a constant region Lung epithelial (termed Fc) that engages Fc receptors on the cell surface of immune cells (Fig. 1). This latter interaction can help to shape the immune response. During the course of a disease, the Figure 1 | Antibodies that affect immune defence in severe COVID-19. Combes et al.1 analysed blood characteristics of the antibodies produced samples from people who had COVID-19 of differing degrees of severity. a, In what the authors classify change to regulate immune defences. One as mild–moderate cases of this illness, patients made antibodies that recognize the spike protein of the aspect of these changes is an alteration in the SARS-CoV-2 virus. The authors report that immune cells called monocytes from these individuals express antibody Fc component that affects which Fc genes that are regulated by the protein interferon. These genes encode molecules that aid antiviral defence receptors are engaged. For example, engage- by boosting immune responses or by suppressing the virus at sites of infection, which might include the ment with the Fc receptors CD64, CD16 and epithelial cells that line the lungs. The CD32B receptor, which is found on monocytes, can dampen such CD32 can regulate how the immune system immune responses if it is bound by another protein. b, The authors report that individuals with severe eliminates bacterial and viral infections5. COVID-19 have anti-spike antibodies in which a region of the antibody, called the Fc domain, binds to CD32B. This interaction hinders the expression of interferon-regulated genes. Combes and colleagues investigated whether Fc-receptor engagement has a role in blunting interferon-mediated responses mild–moderate COVID-19 expressed a distinct The authors next turned their attention to in severe COVID-19. Using immune cells from set of genes whose expression depends on antibodies. Antibodies against SARS-CoV-2 healthy donors exposed to IFN-α and plasma what are known as type I interferon proteins. have a protective role in the natural immune from people with severe COVID-19, they indi- Interferons, molecules that are also called response to this virus, and antibodies target- vidually blocked CD64, CD16 and CD32 Fc cytokines, drive the expression of genes that ing the virus have been used as COVID-19 treat- receptors. Only CD32 blockade enabled the have a role in antiviral defence. ments. Indeed, part of the rationale for using expression of interferon-regulated genes. This interferon-regulated gene-expression the vaccines currently available is to drive the The CD32 Fc receptor exists in two forms — signature was not observed in the cells of generation of such antibodies. The authors CD32A and CD32B. CD32A engagement acti- people with severe COVID-19. Instead, the cells found that the level of antibodies against the vates the immune system, whereas CD32B had a gene-expression signature described as SARS-CoV-2 spike and nucleocapsid proteins dampens immune responses6. Combes and an inflammatory S100A12 myeloid-cell pro- was higher in people with severe disease than colleagues showed that the inhibition of gram (S100A12 is a protein expressed as part in those with mild–moderate COVID-19. More- interferon-regulated gene expression was of this program). A S100A12 signature was over, high antibody levels were negatively due to CD32B. They thus conclude that people previously identified2 as being associated correlated with the presence of cells express- with severe COVID-19 develop antibodies that with severe COVID-19. Interestingly, a similar ing an interferon-regulated gene-expression engage with CD32B Fc receptors and thereby program is associated with another form of program. blunt interferon-mediated defence responses. severe disease called sepsis, which derives To search for a missing link between anti- In support of this conclusion, a previous from an aberrant immune response to bacte- body levels, interferon and COVID-19 severity, study7 demonstrated that people with mod- rial infection3. Combes and colleagues used an in vitro sys- erate and severe COVID-19 develop a diverse An interferon-regulated gene-expression tem. They took immune cells from the blood antibody response in terms of the Fc regions program can be crucial to defence against viral of healthy people, and exposed them to blood recruited, with the presence of spike-specific infection, so the lack of activation of this pro- plasma samples from people with mild–mod- antibodies that engage CD32B being a main gram in people with severe COVID-19 provided erate or severe COVID-19. The authors then predictor of disease severity. What determines a hint that defective initiation of this pathway stimulated the immune cells with IFN-α, to this difference in antibody type between might contribute to the observed differences determine whether an antiviral response devel- severe and moderate COVID-19 remains to be in disease severity. Combes et al. therefore oped. They found that the presence of plasma discovered. set out to determine the reason for the dif- from people with severe disease blocked the It is tempting to speculate that changing ferences. The first obvious suspect was the induction of interferon-responsive genes. the Fc domain to one that engages CD32B is level of an interferon protein (IFN-α) in blood However, if this plasma was treated to deplete a mechanism used by the immune system to plasma (blood lacking its cellular content). The it of antibodies, interferon-mediated gene shut down an intense immune response to authors found no notable difference in IFN-α expression was restored in these immune cells. SARS-CoV-2. It would be interesting to inves- levels with differing disease severity. However, Intriguingly, in a previous study4 of 987 indi- tigate whether such mechanisms are involved there are other types of interferon protein that viduals with severe COVID-19, 135 (13.7%) had in other types of viral infection, and, if so, the authors did not measure. anti-interferon antibodies that could blunt whether they have a detrimental or beneficial

role. Of note, there are also reported exam- develops during SARS-CoV-2 infection. Such speaking, kinetic insight has been availa- ples of SARS-CoV-2 infection generating anti- a comparison would help to reveal the checks ble only from in vitro studies using purified bodies that turn against the host. People with and balances used by the immune system to proteins; experiments in cells have been COVID-19 can develop antibodies that target help keep us alive during severe infection. able to identify the RNA targets of RBPs, but nucleic acids8 and host proteins9. lacked the precision to measure the kinet- It is important to remember that we do Matteo Gentili and Nir Hacohen are at ics of the interactions5. With the advent of not yet know whether, in people with severe the Broad Institute of MIT and Harvard, high-throughput sequencing methods, in vitro COVID-19, this antibody-mediated phenom- Cambridge, Massachusetts 02142, USA. approaches can now probe the kinetics of a enon is detrimental (by suppressing a natural N.H. is also at the Center for Cancer Research, protein’s interactions with tens of thousands antiviral pathway, allowing uncontrolled virus Massachusetts General Hospital, Boston. of RNA variants7. But these experiments are replication) or beneficial (by reducing toxic e-mails: [email protected]; still carried out on purified proteins in the effects of a potent antiviral response). That [email protected] absence of the cellular milieu. In the past said, these results raise the possibility that few years, a method called crosslinking and therapy to block CD32B would partially restore immunoprecipitation8 (CLIP) has become 1. Combes, A. J. et al. Nature 591, 124–130 (2021). interferon responses in people with severe dis- 2. Arunachalam, P. S. et al. Science 369, 1210–1220 (2020). a workhorse for the characterization of ease. However, before considering therapeutic 3. Reyes, M. et al. Nature Med. 26, 333–340 (2020). RNA–protein interactions in cells. In CLIP, applications, the following steps should be 4. Bastard, P. et al. Science 370, eabd4585 (2020). a protein in complex with an RNA molecule 5. Lu, L. L., Suscovich, T. J., Fortune, S. M. & Alter, G. taken. These results need to be confirmed in Nature Rev. Immunol. 18, 46–61 (2018). is covalently crosslinked to the RNA using a larger group of patients, the process should 6. Dhodapkar, K. M. et al. J. Exp. Med. 204, 1359–1369 (2007). ultraviolet light; the complexes are then iso- 7. Zohar, T. et al. Cell 183, 1508–1519 (2020). be examined in other types of tissue in which 8. Woodruff, M. C., Ramonell, R. P., Lee, F. E.-H. & Sanz, I. lated and the crosslinked RNA is identified by the virus is found (rather than just in blood Preprint at medRxiv https://doi.org/10.1101/2020.10.21.20216192 high-throughput sequencing. This approach samples), and a fuller explanation is needed of (2020). provides a catalogue of RNAs that bind to a 9. Wang, E. Y. et al. Preprint at medRxiv https://doi. the mechanisms that underlie these findings. org/10.1101/2020.12.10.20247205 (2021). specific RBP in the complex environment of With several anti-SARS-CoV-2 vaccines the cell, but it provides, at best, only a snapshot currently approved, it will be useful to deter- N.H. declares competing financial interests: see go.nature. of these interactions. com/3p7x5ex for details. mine the antibody profile that vaccination Sharma and colleagues now bridge the elicits, and to compare it with the profile that This article was published online on 16 February 2021. gap between in vitro strategies and CLIP by developing a type of CLIP that can determine Biochemistry the kinetic parameters of RNA–protein interactions in cells. The authors’ key insight was that certain technical aspects of previously Dynamics of RNA–protein reported CLIP methods precluded such approaches from being useful for capturing kinetic parameters. The most challenging binding probed in cells limitation is that crosslinking rates must be rapid to capture the rates at which proteins Olivia S. Rissland and RNA molecules associate and dissociate. Conventional UV sources cannot achieve An understanding of how quickly biomolecules bind and sufficiently rapid crosslinking, and so using dissociate in cells is crucial for developing quantitative models

of biology, but measurements of these kinetics were possible Proteins Ultraviolet only using purified proteins in vitro — until now. See p.152 pulse

As the geneticist Theodosius Dobzhansky proteins. Messenger RNA molecules are bound stated1 in 1973, “Nothing in biology makes by various RNA-binding proteins (RBPs), which Crosslink sense except in the light of evolution.” Many control almost every aspect of the mRNA life modern biologists might add that nothing in cycle — from the initial processing of newly RNA molecular biology makes sense except in the made RNAs to their eventual destruction3. Each light of biochemistry — without the quanti- RBP can bind to hundreds of RNA molecules, Figure 1 | A method for probing RNA–protein tative understanding that biochemistry pro- and, in turn, each RNA can be bound by dozens interactions in cells. Proteins that act on RNA vides, how can biologists predict the effect of of different RBPs4. Moreover, RNA–protein molecules rapidly associate and dissociate from a twofold reduction in the levels of a protein interactions are not static5,6. Instead, proteins their target binding sites. Measurements of the rates during the early development of an organism, can rapidly bind to their target RNAs and just as of association and dissociation are needed for a or of a tenfold increase in the concentration rapidly dissociate from them (Fig. 1), and these quantitative understanding of gene regulation, but of another protein in cancer cells? The chasm dynamics are at the core of gene regulation. In have been impossible to do in living cells. Sharma et al.2 describe a method called KIN-CLIP that uses between the streamlined experiments of bio- other words, the kinetics of RNA–protein inter- ultrafast pulses of ultraviolet light to generate chemistry and the messy complexity of the actions are a driving force of gene expression. covalent crosslinks between the bound proteins cell has long seemed unbridgeable. Now, on Defining the parameters of these kinetics in and RNA molecules in cells. This not only allows 2 page 152, Sharma et al. report a technique that cells is therefore crucial for fully understanding the RNA targets of the proteins to be identified (as enables the biochemical analysis of molecular the regulation of gene expression. was possible in previously reported crosslinking interactions in cells. Although RNA–protein interactions have techniques), but, owing to the rapidity of the The authors focused on the dynamics of been investigated for decades, their kinetics crosslinking process, also allows the kinetics of interactions between RNA molecules and in cells have not been characterized. Broadly association and dissociation to be determined.