Difficulties and advancements in developing drugs for the treatment of dengue fever

Abstract

Dengue fever is caused by infections with dengue virus, the most prevalent mosquito-borne viral pathogen in humans. There are no drugs available to treat dengue virus infection. Here, we discuss the strategies and challenges to develop dengue drugs. Treating the disease pathology is hindered by a lack of understanding of the disease progression and the causes of disease symptoms such as vascular leakage. Most drug-discovery efforts have concentrated on viral targets, which have their own particular challenges. Despite various challenges, great progresses have been made towards the development of dengue therapy. Ultimately, the only way to know whether a direct-acting antiviral will be a viable approach for treating dengue patients is to test safe compounds with demonstrated antiviral activity in human trials.

Keywords::

• antiviral
• dengue virus
• drug discovery
• flavivirus
• inhibitor

Dengue virus (DENV) is a mosquito-borne flavivirus and is a major heath burden in tropical and sub-tropical regions of the world. Dengue fever is typically an acute and self-limiting febrile illness, but as many as 5% of infections lead to life threatening severe dengue, including symptoms of severe plasma leakage (causing shock or respiratory distress), severe bleeding or severe organ dysfunction. It is estimated that there are 390 million human infections each year and that 3.6 billion people are at risk of infection [1]. Despite the disease burden, there is still neither a licensed vaccine nor drug for its treatment. Development of vaccines and drugs to treat dengue is particularly challenging because DENV exists as four antigenically distinct, but co-circulating serotypes. The predominant serotype changes approximately every 2 – 4 years and all four serotypes may co-circulate in a single geographic location, so it is essential that antivirals are active against all four serotypes. However, the development of a single compound that is equally potent against all four serotypes is challenging. The same challenge has been encountered by the antiviral development of hepatitis C virus (HCV), a member of the same Flaviviridae family. HCV exists as seven genotypes that are similarly divergent to the four serotypes of DENV; the first two licensed drugs targeting the HCV protease are both genotype-1 specific. With the exception of nucleosides, the direct-acting antivirals (DAA) currently in clinical development for HCV do not inhibit all genotypes with equal potency.

Three strategies can be envisioned for the treatment of dengue patients. First, the most obvious strategy is a DAA against a viral target to prevent viral replication. It is known that patients with a higher viral load are more likely to develop severe dengue, with the difference between dengue fever and dengue shock syndrome being about 1 log viral load [2]. Therefore, it is assumed that lowering the viral load will decrease disease severity. Second, there may be host proteins that are essential for viral replication, but that are non-essential for host viability during the short-term treatment duration (less than a week). The advantages to this approach are that compounds are more likely to be active across the four serotypes and that there will be a higher barrier to the development of resistance. This approach is exemplified by celgosivir, a host α-glucosidase inhibitor, which is currently in a clinical trial for dengue. One concern with using celgosivir is its side effect of diarrhea, which could potentiate severe dengue because fluid loss from vascular leakage is a hallmark of the disease. Another example of this approach is lovastatin, which is also in a clinical trial for dengue [3]. Lovastatin can lower cholesterol levels, which is essential for DENV replication. In addition, lovastatin has anti-inflammatory effects at the endothelium, so it may have both antiviral and plasma-leakage prevention effects. Third, inhibitors can be developed to target the pathophysiological pathway of severe dengue disease. Perhaps the ultimate therapy is to combine two drugs, one DAA to suppress viremia and another drug to reverse the pathology of severe dengue. However, the disease pathology is still not well understood, so it is difficult to know which targets and approaches to pursue. During dengue infection, high levels of pro-inflammatory cytokines and vasoactive products are produced, but it is not known which of these factors have a direct role in increasing vascular permeability. For example, compounds that stabilize mast cells reduce vascular permeability in animal models, but increase viraemia [4], suggesting that modulating a single aspect of the disease pathology may not be sufficient to treat patients. For that reason, most drug-discovery efforts have concentrated on the viral enzymes in the non-structural proteins NS3 and NS5 and on phenotypic screens against viral replication in cell culture [5].

The most successful class of antivirals are nucleosides. The advantages of nucleosides include that they are broad spectrum so likely to inhibit all four serotypes of DENV [6]. Because the inhibitors are close analogues of natural substrates, resistant viruses are less likely to emerge. Nucleosides have been pursued for DENV, including a Phase II clinical trial using balapiravir, a nucleoside that was stopped for HCV clinical development. Unfortunately, balapiravir did not show any efficacy in dengue patients [7]. The negative outcome of the balapiravir trial raises fundamental questions for the development of DAAs against DENV. First, can a DAA work at all, given the acute nature of dengue viraemia? The only way to answer this question will be to test the best available compounds in the clinic. Balapiravir may not be the optimal nucleoside to answer these questions because some variability has been reported in its antiviral properties in different laboratories and in different cell lines. Balapiravir did not show any efficacy in the dengue AG129 mouse model (submitted for publication). Second, given nucleosides require conversion to their active triphosphate form, is the conversion too slow to exert its antiviral effect on an acute disease? Again, this can only be answered in the clinic with the right compound.

Another unknown for DENV drug development is the target organ and/or cell type where the majority of viral replication occurs. After a host is bitten by an infected mosquito, DENV first infects dendritic cells and spreads to lymph nodes where the virus amplifies. Later targets of infection are thought to be monocytes in the blood and macrophages in the liver, spleen and bone marrow [8]. Autopsies have identified DENV antigen widely distributed in different tissues, but this may be related to post-mortem tissues changes. Understanding the target organ is important when considering drug pharmacokinetics. If peripheral blood mononuclear cells are the major site of viral replication, drugs that have a low volume of distribution and retain a higher concentration in plasma will be more efficacious. If significant viral replication occurs in other tissues such as the liver, drugs that accumulate in the liver will be more efficacious. Although monocytes and dendritic cells were shown to be the major replication sites, it is likely that there are other sites for DENV replication. Therefore, a safe assumption is that a DENV antiviral will need to be widely distributed into tissues throughout the body. This lack of understanding is compounded by the lack of a disease model for DENV. Immunocompromised mice can be infected with DENV to produce a viraemic response, but these mice lack the disease pathology observed in humans, and so the sites of viral replication may be different from those in man. For these reasons, repurposing HCV inhibitors for DENV infection may be challenging because the HCV compounds are designed to target the liver, whereas DENV drugs need to enter systemic circulation and be distributed to the multiple viral-replication sites. Thus, novel compounds that are stable enough to enter systemic circulation are required for DENV.

NS3 and NS5 are multifunctional enzymes that are essential for viral replication. NS3 has an N-terminal protease and a C-terminal helicase and NTPase, while NS5 has an N-terminal methyltransferase and C-terminal polymerase. The polymerase and protease are attractive targets because these are proven targets of DAAs for other viruses, such as HIV and HCV. However, drug-discovery efforts against these targets have provided limited success. The crystal structures of the proteases from DENV and the related flavivirus, West Nile virus (WNV), showed that they have a very flat active site [9]. In addition, these enzymes have a preferred cleavage sequence that includes two consecutive Lys or Arg residues on the non-prime side of the scissile peptide bond. The combination of these factors ensures that the DENV protease is challenging to develop a drug against. Perhaps a more attractive target for DENV drug discovery is the polymerase. NS5 is the most conserved enzyme, increasing the likelihood of developing a drug to target all four serotypes of DENV. However, the polymerase also has its challenges. It shows low activity in vitro, is non-processive, and has limited stability, reducing its adaptability to high-throughput screening. In addition, an assessment of potential ligand-binding pockets in DENV and WNV polymerases found few druggable pockets; no inhibitors have been reported to bind these pockets. However, this assessment did not take into account conformational changes in the protein. A recent report shows that DENV polymerase is flexible and capable of binding drug-like molecules [10].

Phenotypic screening, using a DENV-infection assay or replicon cells, has been widely used to identify antiviral leads with varying success. The reason for this is that only 10 mature viral proteins are required for viral replication. In contrast, many more host proteins are required to support viral replication in cells. Therefore, most hits using this approach target susceptible host proteins. These proteins often do not translate into valid targets inside the human host. An example of this is the DHODH inhibitors that showed potent antiviral activity in cell culture, but the activity was completely lost in vivo due to exogenous uridine from plasma [11].

One successful example of phenotypic screens is the identification of NS4B inhibitors. Screens in different laboratories have identified compounds that target DENV NS4B [12,13]. These compounds are structurally diverse, indicating that they target different regions of the protein. NS4B is a membrane protein and a key component of the viral replication complex. Limited knowledge is available about the structure and function of flavivirus NS4B. The fact that this target has turned up independently suggests that it may be particularly susceptible. The druggability of NS4B may be equivalent to NS5A, a membrane protein in HCV which is a clinically validated target of potent compounds [14]. Another phenotypic screening success is the identification of a lead series targeting the capsid protein which showed efficacy in the dengue mouse model [15].

Despite the challenges highlighted above, several exciting new compounds have a chance to enter clinical trials in the coming years, including capsid, NS4B and polymerase DAAs. Reaching proof-of-concept in human trials will be a major milestone in dengue drug discovery and will finally tell us whether an antiviral is a viable approach for dengue.

Declaration of interest

The authors work for a pharmaceutical company and have received no payment in preparation of this manuscript.