Scientists are hopeful that a new drug — called EIDD-2801 — could change the way doctors treat COVID-19.
The drug shows promise in reducing lung damage, has finished testing in mice and will soon move to human clinical trials.
As of April 3, the novel coronavirus SARS-CoV-2 had infected more than 1 million people with COVID-19 and caused more than 58,000 deaths in a worldwide pandemic.
Currently, no antiviral drugs have been approved to treat SARS-CoV-2 or any of the other coronaviruses that cause human disease.
Researchers at the UNC-Chapel Hill Gillings School of Global Public Health are playing a key role in the development and testing of EIDD-2801. Virologists in the lab of William R. Kenan Jr. Distinguished Professor of epidemiology Ralph Baric, are working with colleagues in the lab of Mark Denison, Edward Claiborne Stahlman Professor of pediatrics at Vanderbilt University Medical Center (VUMC), and with George Painter, chief executive officer of the nonprofit DRIVE (Drug Innovation Ventures at Emory) and director of the Emory Institute for Drug Development (EIDD), where EIDD-2801 was discovered.
The results of the team’s most recent study were published online April 6 by the journal Science Translational Medicine.
The paper includes data from cultured human lung cells infected with SARS-CoV-2, as well as mice infected with the related coronaviruses SARS-CoV and MERS-CoV.
The study found that, when used as a prophylactic, EIDD-2801 can prevent severe lung injury in infected mice.
EIDD-2801 is an orally available form of the antiviral compound EIDD-1931; it can be taken as a pill and can be properly absorbed to travel to the lungs.
When given as a treatment 12 or 24 hours after infection has begun, EIDD-2801 can reduce the degree of lung damage and weight loss in mice.
This window of opportunity is expected to be longer in humans, because the period between coronavirus disease onset and death is generally extended in humans compared to mice.
“This new drug not only has high potential for treating COVID-19 patients, but also appears effective for the treatment of other serious coronavirus infections,” said senior author Baric.
Compared with other potential COVID-19 treatments that must be administered intravenously, EIDD-2801 can be delivered by mouth as a pill.
In addition to ease of treatment, this offers a potential advantage for treating less-ill patients or for prophylaxis — for example, in a nursing home where many people have been exposed but are not yet sick.
“We are amazed at the ability of EIDD-1931 and -2801 to inhibit all tested coronaviruses and the potential for oral treatment of COVID-19.
This work shows the importance of ongoing National Institutes of Health (NIH) support for collaborative research to develop antivirals for all pandemic viruses, not just coronaviruses” said Andrea Pruijssers, the lead antiviral scientist in the Denison Lab at VUMC.
Denison was senior author of a December 2019 study that first reported that EIDD-1931 blocked the replication of a broad spectrum of coronaviruses.
These interinstitutional collaborators, supported by an NIH grant through the University of Alabama at Birmingham, also performed the preclinical development of remdesivir, another antiviral drug currently in clinical trials of patients with COVID-19.
In the new Science Translational Medicine paper, Maria Agostini, a postdoctoral fellow in the Denison lab, demonstrated that viruses that show resistance to remdesivir experience higher inhibition from EIDD-1931.
“Viruses that carry remdesivir resistance mutations are actually more susceptible to EIDD-1931 and vice versa, suggesting that the two drugs could be combined for greater efficacy and to prevent the emergence of resistance,” said Painter.
Clinical studies of EIDD-2801 in humans are expected to begin later this spring.
If they are successful, the drug could not only be used to limit the spread of SARS-CoV-2, but also could control future outbreaks of other emerging coronaviruses.
“With three novel human coronaviruses emerging in the past 20 years, it is likely that we will continue to see more,” said first author Timothy Sheahan, a Gillings assistant professor of epidemiology and a collaborator in the Baric Lab. “EIDD-2801 holds promise to not only treat COVID-19 patients today, but to treat new coronaviruses that may emerge in the future.”
Therapeutic EIDD-2801 reduces SARS-CoV replication and pathogenesis
Given the promising antiviral activity of NHC in vitro, we next evaluated its in vivo efficacy using EIDD-2801, an orally bioavailable prodrug of NHC (β-D-N4-hydroxycytidine-5′-isopropyl ester), designed for improved in vivo pharmacokinetics and oral bioavailability in humans and non-human primates (15).
Importantly, the plasma profiles of NHC and EIDD-2801 are similar in mice following oral delivery (15). We first performed a prophylactic dose escalation study in C57BL/6 mice where we orally administered vehicle (10% PEG, 2.5% Cremophor RH40 in water) or 50, 150, or 500 mg/kg EIDD-2801 2hr prior to intranasal infection with 5E+04 PFU of mouse-adapted SARS-CoV (SARS-MA15), and then vehicle or drug every 12 hours thereafter.
Beginning on 3 days post-infection (dpi) and through the end of the study, body weight loss compared to vehicle treatment was significantly diminished (50 mg/kg) or prevented (150, 500 mg/kg) with EIDD-2801 prophylaxis (Two-way ANOVA with Dunnett’s multiple comparison test, P < 0.0001) (fig. S3A).
Lung hemorrhage was also significantly reduced 5 dpi with 500 mg/kg EIDD-2801 treatment (Kruskal-Wallis Test, P = 0.010, fig. S3B). Interestingly, there was a dose-dependent reduction in SARS-CoV lung titer (median titers: 50 mg/kg = 7E+03 pfu/mL, 150 mg/kg = 2.5E+03 pfu/mL, 500 mg/kg = 50 pfu/mL, vehicle = 6.5E+04 pfu/mL) with significant differences (Kruskal-Wallis with Dunn’s multiple comparisons test) among the vehicle, 150 mg/kg (P = 0.03) and 500 mg/kg (P = 0.006) groups. Thus, prophylactic orally administered EIDD-2801 was robustly antiviral and able to prevent SARS-CoV replication and disease.
Since only the 500 mg/kg group significantly diminished weight loss, hemorrhage and reduced lung titer to near undetectable levels, we tested this dose under therapeutic treatment conditions to determine if EIDD-2801 could improve the outcomes of an ongoing CoV infection. As a control, we initiated oral vehicle or EIDD-2801 2 hours prior to infection with 1E+04 pfu SARS-MA15.
For therapeutic conditions, we initiated EIDD-2801 treatment 12, 24, or 48 hours after infection. After initiating treatment, dosing for all groups was performed every 12 hours for the duration of the study. Both prophylactic treatment initiated 2 hours prior to infection and therapeutic treatment initiated 12 hours after infection significantly (Two-way ANOVA with Tukey’s multiple comparison test) prevented body weight loss following SARS-CoV infection on 2 dpi and thereafter (-2 hours: P = 0.0002 to <0.0001; +12 hours: P = 0.0289 to <0.0001) as compared to vehicle treated animals (Fig. 6A).
Treatment initiated 24 hpi also significantly reduced body weight loss (3-5 dpi, P = 0.01 to <0.0001) although not to the same degree as the earlier treatment initiation groups. When initiated 48 hpi, body weight loss was only different from vehicle on 4 dpi (P = 0.037, Fig. 6A).
Therapeutic EIDD-2801 significantly (Kruskal-Wallis with Dunnett’s multiple comparison test) reduced lung hemorrhage when initiated up to 24 hours after infection (-2, +12, and +24 hours P < 0.0001) mirroring the body weight loss phenotypes (Fig. 6B). Interestingly, all EIDD-2801 treated mice had significantly (Kruskal-Wallis with Dunnett’s multiple comparison test) reduced viral loads in the lungs even in the +48 hours group (All P < 0.0001, Fig. 6C), which experienced the least protection from body weight loss and lung hemorrhage. We also measured pulmonary function via whole body plethysmography (WBP).
In Fig. 6D, we show the WBP enhanced pause (PenH) metric, which is a surrogate marker for bronchoconstriction or pulmonary obstruction (27), was significantly (Two-way ANOVA with Dunnett’s multiple comparison test) improved throughout the course of the study if treatment was initiated up to 12 hours after infection (-2 hours: 2d pi to 5 dpi, P < 0.00
01 to 0.019, +12 hours: 2 dpi to 5 dpi, P < 0.0001 to 0.0192) although the +24 hours group showed sporadic improvement as well (3 dpi P = 0.002) (Fig. 6D). Lastly, we blindly evaluated hematoxylin and eosin-stained lung tissue sections for histological features of ALI using two different and complementary scoring tools (18), which show that treatment initiated up to +12 hours significantly reduced ALI (Kruskal-Wallis with Dunn’s multiple comparison test) (American Thoracic Society (ATS) Lung Injury Score: -2 hours P = 0.0004, +12 hours P = 0.0053, Diffuse alveolar damage (DAD) Score: -2 hours P = 0.0015, +12 hours P = 0.0004, Fig. 6E). Altogether, therapeutic EIDD-2801 was potently antiviral against SARS-CoV in vivo but the degree of clinical benefit was dependent on the time of initiation post-infection.

EIDD-2801 prophylactic and therapeutic efficacy correlates with increased MERS-CoV mutation rate
After obtaining promising in vivo efficacy data with SARS-CoV, we investigated whether EIDD-2801 would be effective against MERS-CoV. As the murine ortholog of the MERS-CoV receptor, dipeptidyl peptidase 4 (DPP4), does not support viral binding and entry, all in vivo studies were performed in genetically modified mice encoding a murine DPP4 receptor encoding two human residues at positions 288 and 330 (hDPP4 288/330 mice)(18, 28). Similar to our SARS-CoV data, all doses of prophylactic EIDD-2801 (50, 150 and 500 mg/kg) protected hDPP4 288/330 mice (fig. S4) from significant body weight loss (Two-way ANOVA with Dunnett’s multiple comparison test, P = 0.03 to < 0.0001), lung hemorrhage (Kruskal-Wallis with Dunn’s multiple comparison test, P = 0.01 to <0.0001), and virus replication which was undetectable (Kruskal-Wallis with Dunn’s multiple comparison test, P < 0.0001) regardless of drug dose following intranasal infection with 5E+04 PFU mouse-adapted MERS-CoV (fig. S4).
We then evaluated the therapeutic efficacy EIDD-2801 following the promising results of our prophylactic studies. Similar to our SARS-CoV study, EIDD-2801 treatment administered before or 12 hours after intranasal mouse-adapted MERS-CoV infection (5E+04 PFU) prevented body weight loss from 2 through 6 dpi (Two-way ANOVA with Tukey’s multiple comparison test, Fig. 7A, P = 0.02 to <0.0001) and lung hemorrhage on 6 dpi (Kruskal-Wallis with Dunn’s multiple comparison test, P = 0.0004 to < 0.0001, Fig. 7B), but treatment initiated 24 or 48 hours did not offer similar protection. Unlike body weight loss and lung hemorrhage data which varied by treatment initiation time, virus lung titer on 6 dpi was significantly reduced to the limit of detection in all treatment groups (Kruskal-Wallis with Dunn’s multiple comparison test, Fig. 7C, P < 0.0001).
Interestingly, when viral genomic RNA was quantified in paired samples of lung tissue, EIDD-2801 significantly reduced quantities of viral RNA (One-way ANOVA with Dunnett’s multiple comparison test, P <0.0001 to 0.017) in an initiation time-dependent manner for all groups except for +48 hours (Fig. 7D).
The discrepancy among infectious titers and viral RNA suggests that accumulated mutations render the particles non-infectious and undetectable by plaque assay, consistent with the MOA. To gauge the effect of EIDD-2801 treatment on lung function, we assessed pulmonary function by WBP. Mirroring the body weight loss data, normal pulmonary function was only observed in groups where treatment was initiated prior to or 12 hours after infection (Two-way ANOVA with Tukey’s multiple comparison test, -2hr: P < 0.0001 3 dpi, P = 0.0002 4 dpi, +12hr: P < 0.0001 3 dpi, P = 0.0008 4 dpi, Fig. 7E). Collectively, these data demonstrate that NHC prodrug, EIDD-2801, robustly reduces MERS-CoV infectious titers, viral RNA, and pathogenesis under both prophylactic and early therapeutic conditions.

To study the molecular mechanisms associated with drug performance in vivo, we investigated the correlation between infectious virus production and EIDD-2801-mediated mutagenesis of MERS-CoV RNA under therapeutic treatment conditions. Using Primer ID NGS, we measured the mutation rates of both viral genomic RNA (non-structural protein 10, nsp10) and host interferon stimulated gene 15 (ISG15) mRNA, a highly up-regulated innate immune-related gene after MERS-CoV infection (Fig. 7F).
Primer ID NGS measures the mutational frequency in single RNA molecules, each of which are represented by a single template consensus sequence (TCS) (25). Viral TCS were significantly reduced (Two-way ANOVA with Tukey’s multiple comparison test, -2 hours P <0.0001, +12 hours P = 0.0001, +24 hours P = 0.02) in a treatment initiation time-dependent manner (Fig. 7G) similar to viral genomic RNA measured by qRT-PCR. In contrast, the numbers of ISG15 TCS were similar (P = 0.2 to 0.8) for all groups indicating that neither vehicle nor drug treatment significantly affected the levels of or mutated ISG15 mRNA transcripts (Fig. 7G).
Similar to our TCS data in Fig. 6G, the total error rate in viral nsp10 was significantly increased (Two-way ANOVA with Tukey’s multiple comparison test) in groups where treatment was initiated prior to (-2 hours, median error rate = 10.5 errors/10,000 bases, P < 0.0001) and up to 24 hours post infection (12 hours, median error rate = 8.2 errors/10,000 bases, P < 0.0001 ; +24 hours, median error rate = 5.4 errors/10,000 bases, P = 0.0003) but the error rates in ISG15 remained at baseline for all groups (Fig. 7H).
In addition, nucleotide transitions observed in MERS-CoV genomes in vitro, were also observed in vivo in groups where treatment was initiated prior to and up to 12 hours post infection (Two-way ANOVA with Tukey’s multiple comparison test, P = 0.0003 to < 0.0001) (Fig. 7I).
Importantly, these transitions were not observed in host ISG15 mRNA (Fig. 7I). Lastly, the EIDD-2801 dose-dependent mutagenesis of viral RNA correlated with an increase in codon change frequency, including stop codons, in mice where treatment was initiated 12 hours or before (Two-way ANOVA with Tukey’s multiple comparison test, vehicle median = 3.4; -2hr median = 22.8, P = 0.0035; +12 hours median = 20.0, P = 0.0004, Fig. 7J).
Thus, approximately 20% of the mutations observed in the -2 hours and +12 hours groups resulted in a codon change and alteration of the nsp10 protein sequence. When extrapolating our results from nsp10 to the entirety of the 30kb MERS-CoV genome, EIDD-2801 likely causes between 15 (+24 hours treatment) and 30 (-2 hours treatment) mutations per genome, 10-20% of which result in amino acid coding changes.
Altogether, our data demonstrates that EIDD-2801-driven mutagenesis correlates well with the reductions in viral load, strongly suggestive of an error catastrophe-driven mechanism of action under therapeutic conditions.