The arrival of SARS-CoV-2 in early 2020 changed the work environment for most of us. While we were collectively getting used to social distancing, empty offices, and online meetings, Dr. Benjamin Ostendorf was embarking on a new project.
While the breadth of the COVID-19 pandemic would unfold over time, it was very soon becoming apparent that SARS-CoV-2 infection had characteristic features – notably a very diverse presentation, with some patients being fully asymptomatic, and others developing serious and fatal illness.
While the interplay between innate immunity and Alzheimer’s disease through the immunomodulatory characteristics of Apolipoprotein E (APOE) has long been known, scientists in 2020 uncovered an association between APOE allelic variants and the severity of COVID-19.1
Dr. Ostendorf and colleagues have a well-documented knowledge of the role of APOE in immunity and cancer progression. This well-positioned them to investigate the impact of human genetic variants of APOE on murine COVID-19 mortality.
We recently had the opportunity to sit down with Dr. Ostendorf to discuss the results of this project published in Nature and to learn about the work he and his team are undertaking in his new lab at Charité-Universitätsmedizin Berlin using single cell technologies.2
When COVID-19 hit in March 2020, our paper exploring the role of APOE germline variation in melanoma had already been accepted for publication.3 Long studied for its role in Alzheimer’s disease, APOE exists as three alleles, ε2, ε3, and ε4, with allele frequencies of about 8%, 78%, and 14%, respectively. In that publication we show that APOE variants are sufficient to drive differential antitumor immunity in melanoma.
Combining the heterogeneity in disease presentation of SARS-CoV-2 patients with data linking it to differences in the immune response, we wondered if APOE variation could contribute to the broad spectrum of COVID-19 outcomes.
Since we were already studying APOE variants in a mouse model, we decided to look at APOE variation in the context of COVID-19.
While there has been much excellent work by multiple labs implicating different variants in differential outcomes in COVID-19, in vivo modeling has yet to be widely used. We were lucky because the mouse model already existed and has been studied thoroughly for Alzheimer’s disease.
In the APOE model, the gene had been replaced with one of the three human variants.
In Alzheimer’s disease, APOE variants ε2 and ε4 are at different ends of the risk spectrum. Variant ε3 carries medium risk of developing the disease, ε4 quite dramatically increases risk, while ε2 seems to have a protective effect.
In melanoma, we saw the same pattern, but inverted. Variant ε3 was in the middle with ε4 conferring a better outcome and ε2 conferring a worse outcome.
With COVID-19, we saw something interesting. When we looked at survival outcomes, both ε2 and ε4 mice fared worse than ε3 mice. When we looked at viral titers we saw the same pattern – both ε2 and ε4 mice showed higher levels of the virus in their lungs compared to ε3.
Flow cytometry revealed changes in ε2 and ε4 resected lungs that recapitulated what we see in humans with severe COVID-19 – relative depletion of “good” immune cell types like T cells and other lymphoid cells and relative expansion of “bad” myeloid cells. Variants ε2 and ε4 again appeared to fall on the same side of the spectrum, and we wondered why.
Flow cytometry lets us look for differences in the abundance of different cell types, but it is limited in its ability to distinguish functional differences, as one can only stain for fewer markers.
In contrast, single cell RNA-Seq (scRNA-Seq) enables us to look at the full transcriptional phenotype. We thought that single cell sequencing might be able to identify differences between ε2 and ε4 that we were not capturing by flow cytometry.
With single cell sequencing, we found that ε2 and ε4 are transcriptionally quite different.
Taking all the data together, we hypothesize that immune cell abundance is similar in ε2 and ε4 mice early on, right after infection. However, the different transcriptional states we see in ε2 and ε4 resected lungs may give rise to a differential immune response later.
Indeed, when we looked at virus-specific T-cell response later in infection, we saw quite dramatic differences in ε2 and ε4.
One big challenge of studying APOE is that it is pleiotropic. It is a secreted protein that acts on different cell types, binds to different receptors, and has multiple effects.
Going into this work, we knew that APOE is implicated in immune modulation, and we believe this is a significant part of its impact on COVID-19.
But clearly, there is more. We and others wanted to look at whether it might have an impact on the virus itself. Strikingly, when we infected cells in culture with SARS-CoV-2 we found that ε3 is able to suppress virus levels while ε2 and ε4 are not.
This is consistent with the data we saw in mice where early on after infection viral titers are higher in ε2 and ε4, which may be due in part to ε3 inhibiting viral infection.
Using the UK Biobank’s genomic data, we looked at the APOE genotype of individuals that had contracted COVID-19.
When we looked at survival, the data mirrored what we had seen in mice. Individuals with ε4/ε4 and ε2/ε2 homozygous genotypes showed drastically worse survival, with the ε4/ε4 differences being statistically significant (Fig 1).
The biggest challenge was that all of the work with the virus had to be conducted at biosafety level 3 (BSL3), without access to our usual arsenal of equipment. Droplet-based technologies would have required taking live cells out of the room to a sorter and then to the sequencing core, but this wasn’t possible for safety reasons.
What mostly got me interested in Parse Bioscience’s solution was the multiplexing. Being able to process 29 mice in parallel and later be able to distinguish which mouse each cell came from was a huge benefit.
Additionally, with the BSL3 restrictions, the ability to fix cells and then take them out of the room to process at a later point in time was also a huge plus.
The beauty today is that there are so many different solutions available. In our case, it was convenient to sequence a lot of mice, which was feasible with Parse’s multiplexing. In other experiments, you might want to sequence very few cells at a higher depth and might choose a different solution. My advice is to keep an open mind and consider which solution will best fit your specific experiment.
One challenge is cost. It is a mundane thing to say, but for many it is a limiting factor. Costs are coming down for kits and for sequencing, but at the same time, the number of cells and replicates we aim to sequence increases. It would be fantastic to be able to use single cell RNA-Seq for clinical applications, which will become more feasible as costs decrease.
In terms of opportunities and where we are headed, integrating more and more modalities will be important. scATAC-Seq is well established at this point, and spatial transcriptomics is a big focus right now. Integrating the temporal dimension might be the next frontier.
Although there are more questions that we could follow up on in our COVID-19 work, I will be returning to focus on studying the intersection of immunity and cancer. The overarching question we hope to contribute to understanding in the lab is why some cancer patients respond positively to immunotherapy while others do not.
There are numerous factors involved in modulating the response to immunotherapy, but there are two areas I would like to focus on. One is understanding the role that germline variation plays. The other is to look more directly at the mechanisms cancer cells employ to evade the immune system.
I did a lot of flow cytometry work as a postdoc. When you run your first single cell experiment, it is a lot like putting on new glasses, where suddenly you see so much more! It is really hard to beat, so I will definitely be using scRNA-Seq to help understand the mechanisms of antitumor immunity.
Thank you Dr. Ostendorf for taking the time to walk us through your latest research. We invite our readers to learn more about Dr. Ostendorf’s exceptional work and Parse Biosciences’ solution for instrument-free single cell RNA sequencing.