Arboviruses and apoptosis: the role of cell death in determining vector competence

Abstract

A relatively small number of mosquito species transmit arboviruses such as dengue, yellow fever, chikungunya and West Nile viruses to hundreds of millions of people each year, yet we still lack a thorough understanding of the molecular factors that determine vector competence. Apoptosis has been shown to be an important factor in determining the outcome of virus infection for many viruses. However, until recently, it was not clear whether apoptosis plays a role in determining the outcome of arbovirus infections in mosquitoes. Recent work has begun to shed light on the roles of apoptosis in this important process.

Arthropod-borne viruses (arboviruses) including dengue (DENV), chikungunya and West Nile (WNV) viruses, as well as numerous others, infect hundreds of millions of people per year (Bhatt et al., 2013; Chancey et al., 2015; Weaver and Forrester, 2015). In most cases, in order for a population of mosquitoes to be capable of transmitting an arbovirus from infected to uninfected hosts (an ability defined as vector competence), replication of the virus must occur in the mosquito (reviewed by Tabachnick, 2013). After a female mosquito takes a blood meal from an infected vertebrate host, the first tissue that is infected is the epithelial cell layer of the mosquito midgut. If the virus is able to replicate in midgut epithelial cells, the virus must then pass across the basal lamina underlying the gut epithelium into the body cavity of the insect, known as the haemocoel. Once in the haemocoel, the virus has access to the other organs of the mosquito including the salivary glands. Productive infection of salivary glands is required in order for virus to be shed in saliva and injected into another vertebrate host during a subsequent blood meal.

Each of these infection steps in the mosquito vector requires penetrating physical barrier(s), binding to receptor(s) on target cells, successful replication in target cells and release of virus from the infected cells at high enough levels to complete the next infection step. Tissues such as midgut and salivary glands thus serve as potential barriers that the virus must breach in order to be transmitted (Franz et al., 2015). Overlying this entire infection process is the mosquito immune system, which can modulate infection through signalling pathways including RNA interference (RNAi), Toll, Imd and Jak/Stat (Sim et al., 2014; Blair and Olson, 2015). In addition, there is accumulating evidence that the gut microbiome and endosymbionts such as Wolbachia spp. can also affect the ability of mosquitoes to be infected by arboviruses (Hegde et al., 2015; Johnson, 2015). It is little wonder then that most of the more than 3000 mosquito species known to exist worldwide do not serve as virus vectors, for various reasons. Only around several dozen mosquito species can transmit arboviruses, and among these, there is a high degree of specificity regarding which viruses each mosquito species and population can transmit.

Although we have made important strides, we still lack a thorough understanding of the molecular basis for determining vector competence. If this portion of the arbovirus infection cycle could be interrupted, arboviruses would become extinct. It is thus imperative that we gain a thorough understanding of the factors that determine the outcome of infection in mosquitoes.

There are a number of known antiviral responses in insects and other animals, but perhaps the most dramatic of these is apoptosis. If an infected cell is able to commit suicide prior to virus replication, this obviously has negative consequences for the virus's ability to replicate and spread to new hosts. Apoptosis has been known to be an important cellular response to many different viruses in both invertebrates and vertebrates for more than 20 years (Hay and Kannourakis, 2002; Clem, 2007). In fact, many viruses encode proteins that inhibit apoptosis, and these are usually required for successful virus infection in an animal host (Hay and Kannourakis, 2002). However, apoptosis is not always antiviral. Examples exist of viruses whose pathology is increased by apoptosis, such as the bystander effect associated with human immunodeficiency virus infection. In addition, it appears that some viruses actually utilize apoptosis or caspases as part of their replication cycle, either because they must lyse the host cell to be released or because caspase cleavage of viral proteins is required for a viral replication step (Richard and Tulasne, 2012). Thus, it is not clear a priori that apoptosis must always be an antiviral response during arbovirus infections of mosquitoes.

Prior to the past decade, the significance of apoptosis during arbovirus infection of mosquitoes had not been the subject of much investigation. Apoptosis is not commonly observed in arbovirus-infected mosquitoes, and when it has been reported, its significance has sometimes been unclear. Nonetheless, over the years, a number of observations have hinted that apoptosis might be important. Pioneering studies focused on histological observations of pathology in midgut or salivary gland tissue of infected mosquitoes. An early report showed that Semliki Forest virus (SFV) caused cytopathic effects in salivary glands of Aedes aegypti (Mims et al., 1966). It is interesting to note that, as pointed out by the authors, A. aegypti is not a natural vector for SFV, suggesting the possibility that this could be a case of a successful antiviral response. It has also been reported that other togaviruses cause midgut pathology in their vectors, including sloughing of cells from the midgut epithelium and disruption of the basal lamina (Weaver et al., 1988, 1992). Whether this pathological effect benefits the virus by allowing more rapid access to the haemocoel, or conversely helps to control virus replication, is not clear. There have also been reports of apoptosis in salivary glands of Aedes albopictus infected with Sindbis virus (SINV) (Bowers et al., 2003; Kelly et al., 2012) and Culex pipiens quinquefasciatus infected with WNV (Girard et al., 2005), leading the authors to postulate that cell death in salivary glands could be a mechanism that decreases feeding behaviour and/or leads to decreased virus release in saliva; this hypothesis was supported by the observation that, while apoptosis increased over time, the proportion of WNV-infected mosquitoes able to transmit virus decreased (Girard et al., 2007). In another case, apoptosis was observed in the midguts of a refractory laboratory strain of Culex pipiens pipiens infected with WNV but not in uninfected midguts. This led the authors to suggest that apoptosis was responsible for the lack of successful replication in this refractory strain. However, the midguts of susceptible mosquitoes were not examined, and it was not clear whether apoptosis actually contributed to resistance or merely accompanied it (Vaidyanathan and Scott, 2006).

While these observations have provided correlative evidence suggesting that apoptosis might play a role in determining the outcome of arbovirus infection in mosquitoes, more recent experiments have sought to test this idea directly. A study using the model insect Drosophila melanogaster showed that the pro-apoptotic gene reaper was rapidly induced in Drosophila larvae after infection with baculovirus or flock house virus and, importantly, that eliminating this rapid apoptotic response caused increased virus replication and decreased host survival (Liu et al., 2013). Furthermore, although these authors did not carry out functional studies in mosquitoes, they also showed that a reaper homologue was induced by DENV infection in a refractory strain of A. aegypti but not in a susceptible strain.

Functional studies examining the importance of apoptosis in controlling arbovirus infection in mosquitoes have been done using two general approaches. The first, RNAi-mediated silencing of mosquito genes involved in regulating apoptosis by injection of dsRNA, affects the regulation of apoptosis in many cells and tissues, whether they are infected or not. In one study, silencing of the A. aegypti caspase gene dronc, which is important for apoptosis (Liu and Clem, 2011), was shown to cause an increase in infection prevalence in DENV-infected mosquitoes (Ocampo et al., 2013), consistent with a role for apoptosis in limiting infection. Another study, however, found that silencing dronc led to less virus infection in the midgut and other tissues when A. aegypti were infected with SINV (Wang et al., 2012). As these two studies were done with different arboviruses, it is difficult to know why seemingly opposite results were obtained. Moreover, Ocampo et al. (2013) examined infection prevalence within an experimental mosquito group but did not measure virus replication within individuals, while Wang et al. (2012) did not observe a change in infection prevalence, only a difference within individuals. It may be that Dronc activity is involved (directly or indirectly) in different steps (i.e. midgut infection and midgut escape) that differ in importance for DENV versus SINV in A. aegypti. Along these lines, caspase activity has been shown to be involved in midgut escape of baculoviruses (Means and Passarelli, 2010), suggesting the possibility that caspases may play a similar role in mosquitoes infected with arboviruses. Interestingly, Wang et al. (2012) also found that silencing IAP1, an inhibitor of apoptosis in A. aegypti (Liu and Clem, 2011), caused increased apoptosis and increased SINV infection. On the surface, this would seem to indicate that apoptosis actually aids SINV infection; however, the authors also observed increased mosquito lethality (as did Ocampo et al., 2013) and severe damage to the midgut when IAP1 was silenced, and speculated that this tissue damage may have affected the defences of the mosquito against infection in a non-specific manner.

The second approach taken to test the effect of apoptosis on arbovirus replication has utilized the ability of SINV to be engineered to express foreign genes. This approach has the advantage that gene expression is only altered in the infected cells. By inserting the Drosophila pro-apoptotic gene reaper into SINV, a recombinant virus was produced that efficiently induced apoptosis in mosquito cells (Wang et al., 2008). When this version of SINV expressing reaper was fed to A. aegypti, it was found to establish infection slower than control viruses (O'Neill et al., 2015). Furthermore, when the viruses that resulted from the infection were analysed by deep sequencing, a strong selection was observed against viruses that retained reaper expression. Meanwhile, there was no selection observed against the reaper insert if it was not expressed because it had been inserted in the antisense orientation. This strong selection against expression of reaper provides the best evidence to date that the induction of apoptosis in arbovirus-infected mosquitoes is detrimental to the virus.

If apoptosis is an effective antiviral response, why is it not commonly observed in infected mosquitoes? One reason is likely to be that it has been selected against in successful vector/virus combinations. When investigators have looked for apoptosis, it has usually been in a vector that is susceptible to infection with the particular arbovirus of interest. However, in a mosquito that vectors a particular arbovirus, the results seen by O'Neill et al. (2015) suggest that it is unlikely that apoptosis would be observed. A distinct possibility is that the level of virus replication, which is actively modulated by other defences such as RNAi (Campbell et al., 2008), is reduced to the point where it is not pro-apoptotic in susceptible vectors (Myles et al., 2008). It is also possible that arboviruses are able to somehow actively suppress apoptosis in mosquito cells in some situations. Although there is currently no evidence for this among arboviruses, it is known that enteroviruses, another type of positive-strand RNA viruses, can both induce and inhibit apoptosis (Harris and Coyne, 2014).

In conclusion, until recently it was not clear whether apoptosis played a significant role in determining the outcome of arbovirus infection in mosquitoes. For many years, the assumption has been that arboviruses cause little if any pathology in their invertebrate vectors, and this is undoubtedly true in most cases. However, a handful of reports over the years have reported cytopathology occurring in certain arbovirus/mosquito infections, and in some cases this has been correlated with resistance to infection or decreased transmission. More recently, investigators have attempted to directly determine whether inducing apoptosis has an effect on arbovirus replication. Using RNAi to transiently silence pro-apoptotic or anti-apoptotic genes has been one approach used, but the results have sometimes been difficult to interpret due to the possibility of indirect effects. More convincingly, when apoptosis was induced by expression of a pro-apoptotic gene from SINV, strong selection was observed against apoptotic gene expression. Based on all of the accumulated evidence, it seems that apoptosis can have strong antiviral effects in arbovirus-infected mosquitoes, but successful virus/vector combinations have probably evolved to avoid it. Apoptosis thus appears to be one of many defences that can be utilized by mosquitoes against arbovirus infection, and is probably a factor in determining vector competence.