Harnessing omics technologies to improve COVID-19 outcome prediction

The outbreak of SARS-CoV-2 continues to keep the world gripped in uncertainty. Huge efforts within the scientific community are ongoing to rapidly further our understanding of this disease, and outlined below, I describe our own efforts in harnessing proteomics for patient risk stratification.
Harnessing omics technologies to improve COVID-19 outcome prediction

Since the first reported case of infection with Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) in humans late 2019, and the subsequent onset of Coronavirus disease 2019 (COVID-19), over three and a half million individuals have died as a result of infection (https://covid19.who.int/). Although there is now light at the end of the tunnel with the rapid development, approval and deployment of multiple vaccines, low and middle-income countries are still being severely affected. The ever-pressing worry of emergent strains with higher transmissibility and vaccine evading potential threatens further disruption.

During the first wave of the pandemic, when the decision from the UK government to lockdown the country came, all non-essential laboratory-based research ground to a halt. Due to our specialism in mass spectrometry-driven proteomics and biomarker discovery, our lab, led by Professor Manuel Mayr (vascular-proteomics.com), was in a unique and privileged position to be able to redirect its research efforts towards COVID-19. It became rapidly apparent, as the healthcare system within the UK was reaching its limit, that an improved risk stratification of patients with COVID-19 in intensive care was required. Particularly as conventional prognostic scores such as APACHE II (Acute Physiology and Chronic Health Evaluation) and SOFA (Sequential Organ Failure Assessment) proved unsuitable for prediction of outcome in COVID-19 patients. With this in mind we set out to determine whether, through the measurement of RNA and proteins in blood, we could identify biomarkers that would better predict risk than available clinical measures.

With our lab being based in south London, next door to King’s College Hospital and just a couple of miles south of Guy’s and St Thomas’ Hospital, it positioned us within the epicentre of COVID-19 critical care within the UK. Our long-standing collaborative ties with critical care physicians, spearheaded by NIHR clinician scientist and critical care consultant Professor Manu Shankar-Hari, enabled us to obtain precious blood samples from COVID-19 patients across numerous stages of the disease. A total of nearly 500 longitudinal blood samples were obtained for our study, including COVID-19 negative patients, hospitalised COVID-19 patients, patients with COVID-19 in intensive care, and, importantly, samples from sepsis patients negative for COVID-19. The latter allowed us to ask the question of how COVID-19 differed from another severe, critical care illness. An overview of the patient cohorts and also experimental approaches utilised in our study is shown below.

Overview of patient cohorts and experimental approaches

At the time of sample acquirement early reports in a small number of patients suggested that the presence and load of SARS-CoV-2 viral RNA, whether nasopharyngeal or in blood, may provide information on disease severity and prognostication. Therefore, our first goal was to determine whether viral RNA load within the blood, termed RNAemia, could predict COVID-19 outcomes. RNA measurements led by Dr. Clemens Gutmann, revealed RNAemia to positively associate with mortality. The only other measure to also significantly associate with mortality was age, further highlighting the need for novel biomarkers. Almost a quarter of COVID-19 patients in intensive care had detectable RNAemia, which was seen in 56% of deceased patients but only in 13% of survivors.

The evident capabilities of SARS-CoV-2 viral RNA to act as a prognostic biomarker were clear, and the next question was whether measurable circulating proteins could perform similarly or even better in their predictive power. Harnessing data-independent acquisition mass spectrometry (DIA-MS), alongside the addition of heavy-labelled peptide standards to allow a greater accuracy in quantification, we set out to quantify the circulating proteomes of our patient samples. Through the measure of hundreds of proteins, it became immediately apparent that COVID-19 led to a dramatic change in the circulating proteome that was dynamic, and importantly distinct to that of sepsis patients. With large scale data generation comes the need for skilled bioinformatics analyses. PhD student and data scientist Bhawana Singh led the effort to determine the best predictive signature for COVID-19 mortality from our datasets. Utilising a machine learning approach, it was observed that the combination of either RNAemia and Age or Age and protein marker Pentraxin-3 (PTX3) were the best binary combinations for the prediction of COVID-19 mortality.

Whether RNAemia is just a consequence of COVID-19 severity or whether viral dissemination contributes to a poorer outcome, remains to be established. Interestingly, our proteomics studies revealed differences in the circulating proteomes of patients with RNAemia which showed dysregulation of several members of the complement cascade and coagulation systems – hallmarks of COVID-19 pathophysiology. To obtain insights into the potential mechanistic relevance of RNAemia to COVID-19 outcome, Dr. Kaloyan Takov conducted experiments with SARS-CoV-2 spike glycoprotein as bait to identify binding partners in circulation of patients with COVID-19. Spike protein is critical for entry of the virus into the host cell. This approach discovered a number of expected interaction partners including various immunoglobulins and members of the complement system, but also identified novel spike-binding proteins such as galectin-3 binding protein (LGALS3BP).  Notably, LGALS3BP was the only protein to be retrieved to a greater extent with the spike glycoprotein from plasma of COVID-19 patients compared to control patients. LGALS3BP is prominently expressed in the lung, and through the elegant cellular and molecular approaches led by Dr. Hashim Ali we were able to highlight LGALS3BP as an inhibitor of SARS-CoV-2 spike-mediated cell-cell fusion, as well as having the ability to inhibit SARS-CoV-2 spike-pseudoparticle uptake. This work potentially sheds light upon a novel antiviral drug target that warrants further investigation.

Although I have described the roles played by the primary authors of this publication, we could not have made this work possible without the phenomenal efforts of everyone involved, and most importantly extending this gratitude to those patients volunteering in our study, to whom we are extremely thankful. Furthermore, the sheer breadth of approaches used to tackle the research questions put forward above could not have been possible without wide collaboration and expertise within our own British Heart Foundation Centre of Research Excellence at King’s College London, but also bringing together scientists and clinicians further afield, across disciplines to tackle a common goal.

Our article can be viewed here: https://rdcu.be/cl4iT

Please sign in or register for FREE

If you are a registered user on Nature Portfolio Microbiology Community, please sign in