INTRODUCTION
In June 2012, a previously unknown coronavirus was isolated from a patient who died from acute pneumonia and renal failure in Saudi Arabia (
1,
2). Since then, the virus, now known as the Middle East respiratory syndrome coronavirus (MERS-CoV) (
3), has been contracted by hundreds of others in geographically distinct locations in the Middle East, and evidence for limited human-to-human transmission has accumulated (
4). Travel-related MERS-CoV infections were reported from a variety of countries in Europe, Africa, Asia, and the United States, causing small local infection clusters in several cases (
http://www.who.int/csr/disease/coronavirus_infections/en/). About 200 laboratory-confirmed human MERS cases were registered during the first 2 years of this outbreak, but recently, for reasons that are poorly understood thus far, this number has more than tripled within just 2 months' time (April-May 2014 [
5]). This sharp increase in reported infections has enhanced concerns that we might be confronted with a repeat of the 2003 severe acute respiratory syndrome (SARS) episode, concerns aggravated by the fact that the animal reservoir for MERS-CoV remains to be identified with certainty (
6–9). Furthermore, at about 30%, the current human case fatality rate is alarmingly high, even though many deaths were associated with underlying medical conditions. MERS-CoV infection in humans can cause clinical symptoms resembling SARS, such as high fever and acute pneumonia, although the two viruses were reported to use different entry receptors, dipeptidyl peptidase 4 (DPP4) (
10) and angiotensin-converting enzyme 2 (ACE2) (
11), respectively.
Coronaviruses are currently divided across four genera:
Alphacoronavirus,
Betacoronavirus,
Gammacoronavirus, and
Deltacoronavirus (
12). MERS-CoV was identified as a member of lineage C of the genus
Betacoronavirus (
2), which also includes coronaviruses of bat (
13,
14) and hedgehog (
6) origin. Following the 2003 SARS epidemic, studies into the complex genome, proteome, and replication cycle of coronaviruses were intensified. Coronaviruses are enveloped viruses with a positive-sense RNA genome of unprecedented length (25 to 32 kb [
12,
15,
16]). The crystal structures of a substantial number of viral nonstructural and structural proteins were solved, and targeted drug design was performed for some of those (reviewed in reference
17). Unfortunately, thus far, none of these efforts resulted in antiviral drugs that were advanced beyond the preclinical phase (
18). The 2003 SARS-CoV epidemic was controlled within a few months after its onset, and since then, the virus has not reemerged, although close relatives continue to circulate in bat species (
14). Consequently, the interest in anticoronavirus drug development has been limited, until the emergence of MERS-CoV. Despite the modest size of this CoV outbreak thus far, the lack of effective methods to prevent or treat coronavirus infections in humans is a serious concern for the control of MERS-CoV or the next zoonotic coronavirus.
Antiviral research in the post-SARS era resulted in the identification of several compounds that may target coronavirus replication directly or modulate the immune response to coronavirus infection. For example, entry inhibitors targeting the coronavirus spike protein were developed (reviewed in reference
19). In addition, several of the replicative enzymes (including both proteases and the helicase) were targeted with small-molecule inhibitors, some of which can inhibit coronavirus infection in cell culture at low-micromolar concentrations (
20–26; reviewed in reference
27). Broad-spectrum antiviral agents, like the nucleoside analogue ribavirin and interferon (IFN), were tested for their ability to inhibit SARS-CoV infection and were—to a limited extent—used for the treatment of SARS patients during the outbreak (reviewed in references
28 and
29). In the case of ribavirin, mixed results were reported from studies in different cell lines, animal models, and patients. Also, the merits of treating SARS patients with immunomodulatory corticosteroids have remained a matter of debate (reviewed in references
28–30). For MERS-CoV, partial ribavirin sensitivity was observed in cell culture and in a macaque animal model, but only when using very high doses of the compound in combination with IFN-α2b (
31,
32). However, in a small-scale clinical trial, this combination therapy did not benefit critically ill MERS patients (
33). Nevertheless, the anticoronavirus effects of type I IFN treatment deserve further evaluation, in particular since MERS-CoV seems to be considerably more sensitive than SARS-CoV (
34,
35). Treatment with type I IFNs inhibits SARS-CoV and MERS-CoV replication in cell culture (
31,
34–41) and, for example, protected macaques against SARS-CoV (
36) or MERS-CoV (
32) infection. Based on experiments in cell culture, mycophenolic acid was recently reported to inhibit MERS-CoV infection (
41,
42), and we and others showed that low-micromolar concentrations of cyclosporine inhibit coronavirus replication (
34,
43–45).
We recently described (
34) a high-throughput assay for antiviral compound screening that is based on the pronounced cytopathic effect (CPE) caused by MERS-CoV infection in Vero and Huh7 cells. This assay was now further exploited to screen a library of 348 FDA-approved drugs for their potential to inhibit MERS-CoV replication. Chloroquine, chlorpromazine, loperamide, and lopinavir were found to inhibit MERS-CoV replication
in vitro at low-micromolar concentrations. In addition, these molecules appear to be broad-spectrum coronavirus inhibitors, as they blocked the replication of human coronavirus 229E and SARS-CoV with comparable efficacy. Since these compounds have already been approved for clinical use in humans, their anti-MERS-CoV activity merits further investigation, in particular in a small-animal model for MERS-CoV infection, of which a first example has recently been described (
46).
DISCUSSION
The ongoing MERS-CoV outbreak has made it painfully clear that our current options for treatment of life-threatening zoonotic coronavirus infections in humans are very limited. At present, no drug is available for the treatment of any of the human or zoonotic coronaviruses (reviewed in reference
58), despite the extensive research efforts triggered by the 2003 SARS outbreak (reviewed in references
26 and
27). The brevity of that epidemic is a major reason why, thus far, none of the prototypic coronavirus inhibitors was advanced beyond the (early) preclinical stage. Like SARS-CoV a decade ago and MERS-CoV at present, future emerging coronaviruses will likely continue to pose a threat to global public health. Therefore, the search for broad-spectrum inhibitors that may reduce the impact of coronavirus infections in humans remains a challenging research priority. Given the time-consuming nature of antiviral drug development and registration, existing therapeutics for other conditions may constitute the only immediate treatment option in the case of emerging infectious diseases. For most of these drugs, ample experience is available with dosing in humans, and their safety and absorption, distribution, metabolism, and excretion (ADME) profiles are well known.
At the time of this study, a MERS-CoV infection model in (small) animals was not available. For initial antiviral testing, we therefore used our cell culture-based screening assay (
34) to search for compounds that may inhibit MERS-CoV infection. We identified four FDA-approved compounds (chloroquine, chlorpromazine, loperamide, and lopinavir) that inhibit the
in vitro replication of MERS-CoV at low-micromolar concentrations (
Fig. 1 and
Table 1). While for some of these molecules the SI was limited (<10), for each of them we established at least one concentration at which MERS-CoV replication was inhibited by more than 80% without a detectable reduction of cell viability. The same four drugs were also found to inhibit, with comparable potency, the
in vitro replication of two other coronaviruses, i.e., HCoV-229E and SARS-CoV (
Fig. 2 and
3 and
Table 1).
CQ inhibited MERS-CoV replication with an EC
50 of 3.0 μM (
Fig. 1A) and blocked infection at an early step (
Fig. 4A and
B). CQ has a tendency to accumulate in lysosomes, where it sequesters protons and increases the pH. In addition, it interacts with many different proteins and cellular processes, resulting in the modulation of autophagy and the immune response (for a review, see reference
59). CQ has also been reported to inhibit the replication of multiple flaviviruses, influenza viruses, HIV (reviewed in reference
60), Ebola virus (
61), and Nipah-Hendra virus (
62), as well as several coronaviruses, including SARS-CoV, in cell culture (
51–55,
63,
64). Early reports showed that high doses of CQ inhibit an early step of the replication of the coronavirus mouse hepatitis virus (MHV). However, in SARS-CoV-infected BALB/c mice, systemically administered CQ did not result in a significant viral load reduction in the lungs. Intranasal administration of CQ (50 mg/kg of body weight) resulted in a minor reduction of viral titers in the lung (
65). When pregnant mice were treated with CQ (at 15 mg/kg), their newborn offspring were protected against a lethal challenge with HCoV-OC43 (
54). Likely, the accumulation of CQ in the milk glands, resulting in high drug concentrations in maternal milk, was a major factor in reaching a sufficiently high concentration of the drug in blood plasma. CQ was also shown to inhibit the
in vitro replication (EC
50, 2 μM) of the feline coronavirus infectious peritonitis virus (FIPV) (
55). Treatment of naturally infected cats with CQ resulted in a clinical improvement, which was, however, not attributed to a direct antiviral effect but likely due to the immunomodulatory properties of CQ. These results highlight that, e.g., the drug delivery route, virus strain used, and drug dosage might influence the outcome in animal models. In BALB/c mice, steady-state concentrations of 8 μM in plasma were observed following repeated administration of CQ at 90 mg/kg (
61), which is above the EC
50 of CQ for inhibition of MERS-CoV-induced CPE in this study. Levels of 9 μM in plasma were observed in humans following CQ treatment with 8 mg/kg/day for three consecutive days (
66).
The second FDA-approved drug found to block MERS-CoV infection was CPZ, the first antipsychotic drug developed for treatment of schizophrenia (
67). CPZ affects the assembly of clathrin-coated pits at the plasma membrane (
56) and has been reported to inhibit the replication of alphaviruses (
68), hepatitis C virus (
69), and the coronaviruses SARS-CoV (
70), infectious bronchitis virus (
71), and MHV-2 (
72). Our time-of-addition studies, however, suggest that CPZ inhibits MERS-CoV replication at both an early and a postentry stage, implying that an effect on clathrin-mediated endocytosis is unlikely to be the sole antiviral mechanism (
Fig. 4C and
D). CPZ plasma concentrations in patients treated for psychotic disorders range between 0.3 and 3 μM (
73), which is somewhat below the observed EC
50s observed here (which range between 2 and 9 μM).
The replication of MERS-CoV
in vitro was also inhibited by LPM, an antidiarrheal opioid receptor agonist that reduces intestinal motility (reviewed in reference
74). LPM also inhibits the replication of two other coronaviruses at low-micromolar concentrations (4 to 6 μM). Upon oral or intravenous administration, the molecule rapidly concentrates in the small intestine. Less than 1% of orally taken LPM is absorbed from the gut lumen, and its tendency to concentrate at the site of action is the probable basis for its antidiarrheal effect (
75). This property would very much limit systemic use for the treatment of respiratory coronavirus infections, although administration in the form of an aerosol might be explored. In the veterinary field, it would be interesting to test whether the compound has the potential to inhibit enteric coronaviruses, such as the porcine transmissible gastroenteritis coronavirus.
Finally, the HIV-1 protease inhibitor (PI) LPV was shown to inhibit MERS-CoV replication with EC
50s of about 8 μM, which is in the range of the LPV concentrations in plasma (8 to 24 μM) that have been observed in AIDS patients (
76). LPV was previously shown to block the SARS-CoV main protease (M
pro) (
24). This is somehow unexpected, since the retro- and coronavirus proteases belong to different protease families (the aspartic and chymotrypsin-like protease families, respectively). Since MERS-CoV and SARS-CoV are relatively closely related, LPV may also target the M
pro of MERS-CoV. However, several anti-HIV PIs are also known to influence intracellular pathways leading to side effects in patients undergoing highly active antiretroviral therapy, including lipodystrophy and insulin resistance (
77). The exact cellular targets of these PIs have not yet been identified, and most likely multiple pathways are involved. It remains to be investigated if the effect of LPV on these intracellular pathways is associated with the anti-CoV activity found here. Interestingly, no selective anti-CoV activity was found for two other HIV PIs in the compound library (atazanavir and ritonavir; see Dataset S1 in the supplemental material). During the SARS outbreak, treatment with LPV, in combination with ritonavir, was explored with some success in nonrandomized clinical trials (for reviews, see references
78 and
79).
The efficacy of the most promising compounds identified in this study, CQ and LPV, should now be evaluated in (small-) animal models for MERS-CoV infection, which are still in development. In a nonhuman primate model (macaques), only mild clinical signs developed, in contrast to the frequently severe clinical outcome in humans (
80,
81). Unfortunately, Syrian hamsters (
82), BALB/c mice (
83), and ferrets (
84) were found to resist MERS-CoV infection. A very recent study (
46) reported that mice can be rendered susceptible to MERS-CoV infection by prior transduction with a recombinant adenovirus that expresses human DPP4, a documented receptor for MERS-CoV entry (
10). Subsequent MERS-CoV infection resulted in severe pneumonia and high MERS-CoV titers in the lungs (
46). Despite some practical and conceptual limitations, this model may provide a useful starting point for further evaluation of inhibitors of MERS-CoV infection.
In 2003, the ∼10% mortality rate among SARS patients was one of the major reasons for the worldwide public unrest caused by the emergence of SARS-CoV. Clearly, and despite the recent sharp increase in the number of registered cases (
5), the course of the MERS-CoV outbreak has been quite different thus far. Although only about 650 laboratory-confirmed cases have been registered in the 2 years that have passed since the first documented human infections, in particular the ∼30% mortality rate within this group remains a grave concern. In this context, efficacious anticoronavirus drugs, administered alone or in combination, can constitute an important first line of defense. It typically takes over 10 years to develop a newly discovered molecule and obtain approval for clinical use. To the best of our knowledge, there are currently no potent and selective coronavirus inhibitors in (early or advanced) preclinical development. Hence, drugs that have been registered for the treatment of other conditions and that also inhibit MERS-CoV replication might be used (off-label) in an attempt to save the life of MERS patients. A combination of two or more of such drugs may cause a modest reduction in viral load, which might aid to control viral replication, slow down the course of infection, and allow the immune system to mount a protective response. In an
accompanying paper by Dyall et al. (
85), CQ and CPZ were identified as inhibitors of the MERS-CoV as well. Follow-up studies will include in-depth mechanism of action studies, including resistance development of MERS-CoV against the compounds identified. Furthermore, the efficacy of combinations of two or more of these drugs will be explored, also in combination with interferon. In particular, CQ and LPV may constitute valuable candidates for further testing in animal models or direct off-label use, since the concentrations needed to inhibit viral replication in cell culture are in the range of the concentrations that can be achieved in human plasma.