Enterovirus D68 2A protease causes nuclear pore complex dysfunction and motor neuron toxicity

Reviewer #1 (Public review):

Summary:

Zinn and colleagues investigated the role of proteases 2A and 3C of enterovirus D68 (EVD68), an emerging pathogen associated with outbreaks of acute flaccid myelitis (AFM), a polio-like disease, on the nucleocytoplasmic trafficking in different systems, including human neurons derived from pluripotent cells. They found that 2A specifically cleaved Nup98 and POM121. Using reporter proteins and RNA synthesis and trafficking assays in cells expressing viral proteases, they showed that 2A induces broad loss of the nuclear pore barrier function, but, surprisingly, the RNA export appears to be minimally affected. Since nucleocytoplasmic trafficking defects are known to be associated with neuropatologies, they propose a hypothesis that 2A-dependent cleavage of nucleoporins in motoneurons underlies the development of EVD68-induced AFM. They further show that a 2A-specific inhibitor increases the survival of human neurons differentiated from stem cells upon EVD68 infection.

Strengths:

Use of multiple methods to investigate the effect of 2A and 3C expression on nucleoporin cleavage and nucleocytoplasmic trafficking.

Weaknesses:

Overall, the paper follows multiple others that extensively investigated the cleavage of nucleoporins by enterovirus 2As, so the results are of limited novelty. The hypothesis that infection of motoneurons is the cause of EVD68-induced neurological complications so far is supported by only one autopsy report. Other data suggest that infection of other cell types, such as astrocytes, and/or inflammatory cell infiltration in the CNS, are likely to be responsible for the symptoms. In any case, the claim that EVD68 is specifically neurotoxic because of the 2A-dependent cleavage of nucleoporins in neurons is unfounded, as the virus will be just as "toxic" for other infected cell types.

The paper also requires a more convincing presentation of the data.

Reviewer #2 (Public review):

Summary:

This manuscript investigates the role of EV-D68 proteases 2A and 3C in nuclear pore complex (NPC) dysfunction and their contribution to motor neuron toxicity. The authors demonstrate that both proteases cleave only a limited number of nucleoporins, with 2A^pro showing the strongest impact by inhibiting nuclear import and export of proteins and disrupting NPC permeability without affecting RNA export. Importantly, treatment with the 2A^pro inhibitor telaprevir reduced neuronal cell death in a dose-dependent manner, achieving neuroprotection at concentrations below those required to inhibit viral replication. The study addresses a relevant mechanism underlying EV-D68-induced neuropathology and explores a potential therapeutic intervention.

Strengths:

(1) Provides significant mechanistic insight into how EV-D68 proteases alter NPC function and contribute to neuronal toxicity.

(2) The use of recombinant 2A and 3C proteins allows clear dissection of the specific contribution of each protease.

(3) Demonstrates a therapeutic effect of telaprevir, with neuroprotection independent of viral replication inhibition, adding translational value to the findings.

(4) The topic is highly relevant given the association of EV-D68 with acute flaccid myelitis.

Weaknesses:

(1) Most experiments were performed with recombinant proteases, lacking validation in the context of viral infection, where both proteases act simultaneously.

(2) The conclusion that RNA export is unaffected requires confirmation during actual infection.

(3) The reduction of neurotoxicity by telaprevir does not fully demonstrate that the protective effect is solely mediated through NPC preservation; additional analyses of eIF4G cleavage, nucleoporin integrity, and stress granules are needed.

(4) The study would be strengthened by including another 2A inhibitor (e.g., boceprevir) to confirm the specificity of telaprevir's protective effects.

Reviewer #3 (Public review):

Summary:

The author showed expression of the viral proteases 2Apro and 3Cpro of EV-D68, which cleaved specific components of the nuclear pore complex (Nup98 and POM121 by 2Apro), and 2A but not 3C expression altered nuclear import and export. Similar nucleocytoplasmic transport deficits are observed in EV-D68-infected RD cells and iPSC-derived motor neurons (diMNs). 2A inhibitor telaprevir partially rescued the nucleocytoplasmic transport deficits and suppressed neuronal cell death after infection. While it's clear that 2A can cleave NPC proteins and affect nuclear transport, the link to neurotoxicity after EV-D68 infection is less convincing.

This study opens up a very intriguing hypothesis: that EV-D68 2Apro could be directly responsible for motor neuron cell death, mediated by POM121 and possibly Nup98 cleavage, that ultimately results in paralysis known as acute flaccid myelitis. This hypothesis notably does run counter to other published data showing that human neuronal organoids derived from iPSCs can support productive EV-D68 infection for weeks without cell death and that EV-D68-infected mice can have paralysis prevented by depletion of CD8 T cells, still with EV-D68 infection of the spinal cord. However, even if 2Apro is not ultimately responsible for motor neurons dying in human infections, that does not exclude the possibility that cleavage of nups could still disrupt motor neuron function. Notably, most children with AFM have some amount of motor function return after their acute period of paralysis, but most still have some residual paralysis for years to life. It is possible that 2A pro could mediate the acute onset of weakness, while T cells killing neurons could determine the amount of long-term, residual paralysis.

Strengths:

The characterization of nuclear pore complex components that appear to be targets of both poliovirus and EV-D68 proteases is quite thorough and expansive, so this data set alone will be useful for reference to the field. And the process by which the authors narrowed their focus to EV-D68 2Apro reducing Nup98 and POM121 as consequential to both import and export of nuclear cargo but not RNA was technically impressive, thorough, and convincing. As will be detailed below, when the authors move from studying over-expressed proteases in transformed cell lines to studying actual virus infection in both transformed cell lines and iPSC-derived neurons, some of the data only indirectly support their conclusions; however, the quality of the experiments performed is still high. So even if the claim that 2Apro causes neurotoxicity is circumstantial, the data certainly are intriguing and certainly justify further study of the effects of EV-D68 2Apro on the NPC and how this impacts pathogenesis. This is a convincing start to an intriguing line of inquiry.

Weaknesses:

This study falls a bit shy of actually showing that 2Apro effects are causing motor neuron toxicity because the evidence of this is fairly indirect. At points, the authors do admit these limitations, but at other times, they claim to have shown the link directly. The following are reasons why these claims are only indirectly supported:

(1) Cleavage of Nup98 and POM121 after EV-D68 infection in RD cells and diMNs is never demonstrated.

(2) Telaprevir was able to rescue nucleocytoplasmic transport in RD cells at low concentrations (Figure 4A). It is not shown if this correlates with its antiviral effect in RD cells, or could this correlate with inhibition of 2A cleavage of Nup98 or POM121, which is never measured.

(3) Building off of the prior point, the authors' claim that the neuroprotective effect of telaprevir is independent of its antiviral effect is not well-founded. Figure 4E (neuroprotection) was done with MOI 5, and Figure 4G (virus growth) was MOI 0.5. Telaprevir neuroprotection is not shown at MOI 0.5, nor is the neuroprotective effect correlated with inhibition of 2A cleavage of Nup98 or POM121.

(4) The use of mixed virus isolates only in the diMNs is problematic because different EV-D68 isolates are known to have drastically different effects on pathogenesis in mice. Since all initial data were generated with the MO isolate, adding the additional MD isolate to the diMN experiments actually adds uncertainty to the conclusions. It is not clear if the authors infected different cultures with the different isolates and combined the data or infected all cultures with a mixture of the two isolates. If the former, then the data should be reported separately to see the effect of each individual strain, which would be interesting to EV-D68 virologists. If the latter, then there is no way to know from these data whether one of the two isolates had increased fitness over the other and exerted a dominant effect. If the MD isolate overtook the MO isolate, from which all other data in this manuscript are derived, then we have much less of an idea how much the data from the first three figures supports the final figure.

Author response:

We thank the reviewers for their detailed and thoughtful comments on the manuscript. In general, the reviewers found the data supporting the role of Enterovirus D68 proteases in disrupting the composition of the nuclear pore complex, the 2A protease disrupting nucleocytoplasmic transport of protein cargoes, and the mechanistic dissection of this process to be convincing and potentially relevant to the pathogenesis of AFM. Reviewers requested additional experiments evaluating our observation that RNA export was not similarly impaired, particularly in the context of viral infection rather than solely expression of recombinant proteases. They also requested that cleavage of POM121 and Nup98 by 2A protease, which was demonstrated in 2A^pro transfected cells and in biochemical assays, also be demonstrated in motor neurons infected by EV-D68. Finally, reviewers noted that while suggestive, the evidence falls short of demonstrating that the toxicity of 2A^pro is mediated through nuclear pore complex dysfunction.

To address these critiques, we aim to do the following:

(1) Determine the impact of live virus infection on RNA export by repeating the ethinyl uridine pulse-chase assay in the setting of live virus infection. We will also provide representative images for these data and the previously reported data from transfection with GFP-2A^pro and GFP-3C^pro.

(2) Evaluate cleavage of POM121 and Nup98 in EV-D68-infected diMNs and inhibition of cleavage by telaprevir by Western blot.

(3) Present motor neuron survival data in figure 4 as separate graphs for each of the viral strains tested, rather than pooling the data. To clarify reviewer #3’s concern, these were not mixed cultures.

We agree that we have not demonstrated conclusively that the mechanism by which 2A^pro is toxic to motor neurons is via NPC dysfunction. Future work will determine the extent to which NPC dysfunction contributes to 2A^pro-mediated motor neuron toxicity versus other potential targets of 2A^pro. We feel that the additional experiments required to achieve this will be extensive and are beyond the scope of the present manuscript, which represents a key first step in this line of inquiry.

In addition to the above, there were several points of disagreement between reviewers. We would like to respond to those as follows:

Reviewer #1: “The hypothesis that infection of motoneurons is the cause of EVD68-induced neurological complications so far is supported by only one autopsy report. Other data suggest that infection of other cell types, such as astrocytes, and/or inflammatory cell infiltration in the CNS, are likely to be responsible for the symptoms.”

Reviewer #3: “This study opens up a very intriguing hypothesis: that EV-D68 2Apro could be directly responsible for motor neuron cell death, mediated by POM121 and possibly Nup98 cleavage, that ultimately results in paralysis known as acute flaccid myelitis. This hypothesis notably does run counter to other published data showing that human neuronal organoids derived from iPSCs can support productive EV-D68 infection for weeks without cell death and that EV-D68-infected mice can have paralysis prevented by depletion of CD8 T cells, still with EV-D68 infection of the spinal cord. However, even if 2Apro is not ultimately responsible for motor neurons dying in human infections, that does not exclude the possibility that cleavage of nups could still disrupt motor neuron function. Notably, most children with AFM have some amount of motor function return after their acute period of paralysis, but most still have some residual paralysis for years to life. It is possible that 2A pro could mediate the acute onset of weakness, while T cells killing neurons could determine the amount of long-term, residual paralysis.”

The infection of motor neurons is strongly supported not only by the aforementioned autopsy data[1], but also by mouse model data demonstrating replication of EV-D68 within motor neurons in the anterior horn of the spinal cord.[2 ] There are also extensive reports of electromyography and nerve conduction studies from human AFM patients demonstrating that the site of pathology is the spinal motor neuron.[3-10]. By contrast, infection of astrocytes has been demonstrated only in primary murine astrocyte cultures in which no neurons were present.[11] .Therefore, while the available data suggest that EV-D68 infection of astrocytes is possible, in the in vivo context of human and mouse spinal cord, tropism to motor neurons appears to be preferential. The relative contributions to toxicity of neuron-autonomous vs non-autonomous processes such as glial dysfunction and inflammatory cell infiltration remain to be elucidated, and are not mutually exclusive.

Our working hypothesis is more in line with that of Reviewer #3. Motor neuron dysfunction and motor neuron death may ultimately prove to have dissociable causes, each of which may be neuron-autonomous, non-neuron-autonomous, or a mixture thereof. The infection of motor neurons is likely the initiating event, with multiple downstream consequences. Much additional work will be required to resolve this controversy.

Reviewer #1: “Demonstrates a therapeutic effect of telaprevir, with neuroprotection independent of viral replication inhibition, adding translational value to the findings.”

Reviewer #3: “The authors' claim that the neuroprotective effect of telaprevir is independent of its antiviral effect is not well-founded. Figure 4E (neuroprotection) was done with MOI 5, and Figure 4G (virus growth) was MOI 0.5. Telaprevir neuroprotection is not shown at MOI 0.5, nor is the neuroprotective effect correlated with inhibition of 2A cleavage of Nup98 or POM121.”

The selection of MOIs for these two experiments was limited by technical considerations. If the viral growth curve were to be performed at MOI 5, it would be confounded by cell death. Further, a low MOI is required in order to allow multiple rounds of infection, replication, and spread within the culture, and is therefore more sensitive for assaying the effect of telaprevir on viral replication. On the other hand, at MOI 0.5 diMN death is very gradual, and in the neuroprotection assay we would have lacked the statistical power to determine whether a rescue of this small magnitude of toxicity is significant. The EC₅₀ of telaprevir is not expected to vary significantly at different MOIs.

References:

(1) Vogt, M. R. et al. Enterovirus D68 in the Anterior Horn Cells of a Child with Acute Flaccid Myelitis. N Engl J Med 386, 2059-2060 (2022). https://doi.org/10.1056/NEJMc2118155

(2) Hixon, A. M. et al. A mouse model of paralytic myelitis caused by enterovirus D68. PLoS Pathog 13, e1006199 (2017). https://doi.org/10.1371/journal.ppat.1006199

(3) Andersen, E. W., Kornberg, A. J., Freeman, J. L., Leventer, R. J. & Ryan, M. M. Acute flaccid myelitis in childhood: a retrospective cohort study. Eur J Neurol 24, 1077-1083 (2017). https://doi.org/10.1111/ene.13345

(4) Elrick, M. J. et al. Clinical Subpopulations in a Sample of North American Children Diagnosed With Acute Flaccid Myelitis, 2012-2016. JAMA Pediatr 173, 134-139 (2018). https://doi.org/10.1001/jamapediatrics.2018.4890

(5) Hovden, I. A. & Pfeiffer, H. C. Electrodiagnostic findings in acute flaccid myelitis related to enterovirus D68. Muscle Nerve 52, 909-910 (2015). https://doi.org/10.1002/mus.24738

(6) Knoester, M. et al. Twenty-Nine Cases of Enterovirus-D68 Associated Acute Flaccid Myelitis in Europe 2016; A Case Series and Epidemiologic Overview. Pediatr Infect Dis J 38, 16-21 (2018). https://doi.org/10.1097/INF.0000000000002188

(7) Martin, J. A. et al. Outcomes of Colorado children with acute flaccid myelitis at 1 year. Neurology 89, 129-137 (2017). https://doi.org/10.1212/WNL.0000000000004081

(8) Saltzman, E. B. et al. Nerve Transfers for Enterovirus D68-Associated Acute Flaccid Myelitis: A Case Series. Pediatr Neurol 88, 25-30 (2018). https://doi.org/10.1016/j.pediatrneurol.2018.07.018

(9) Van Haren, K. et al. Acute Flaccid Myelitis of Unknown Etiology in California, 2012-2015. JAMA 314, 2663-2671 (2015). https://doi.org/10.1001/jama.2015.17275

(10) Natera-de Benito, D. et al. Acute Flaccid Myelitis With Early, Severe Compound Muscle Action Potential Amplitude Reduction: A 3-Year Follow-up of a Child Patient. J Clin Neuromuscul Dis 20, 100-101 (2018). https://doi.org/10.1097/CND.0000000000000217

(11) Rosenfeld, A. B., Warren, A. L. & Racaniello, V. R. Neurotropism of Enterovirus D68 Isolates Is Independent of Sialic Acid and Is Not a Recently Acquired Phenotype. Mbio (2019). https://doi.org/10.1128/mBio