Rotaviruses (RV) remain major causes of life-threatening diarrheal disease in children under 5 years of age, resulting in nearly half a million deaths annually. Although RV vaccines are available and successful in many developed countries, vaccine efficacy and delivery are suboptimal in developing countries where they are needed most. RV-induced disease is life-threatening compared to disease caused by most other enteric microbes, emphasizing the need to more fully understand the pathophysiology of RV gastroenteritis. RV infection, replication and pathogenesis are carried out by the viral encoded six structural and six nonstructural proteins and the viral genome. One intriguing protein encoded in the viral genome is nonstructural protein 4 (NSP4). RV NSP4 has been the focus of much attention in our laboratory since it was first described as a nonstructural glycoprotein with an uncleaved signal sequence and a glycosylation pattern that indicated it does not traffic to the Golgi apparatus (Arias et al., 1982; Both et al., 1983; Ericson et al., 1983). At that time, few viral glycoproteins with uncleaved signal sequences were known. More importantly, the existence of a nonstructural glycoprotein, which contained only high mannose glycans that are trimmed and produced by a nonenveloped virus was certainly intriguing.

The epidemiology and evolutionary patterns of RVs are being documented by studies of viruses from humans and a variety of animal species. RVs are broadly classified based on the serologic reactivity and genetic variability of the middle capsid structural protein VP6 into eight groups/species (RVA-RVH). RVA viruses are the most extensively studied and they are further classified based on antigenic and sequence differences of the two outer capsid proteins VP7 and VP4 into G (Glycosylated) and P (Protease-sensitive) types, respectively. Due to the unavailability of serological reagents for these neutralization antigens, and the increasing ease and availability of sequencing, including whole genome sequencing, each genomic segment of RVA is now classified based on sequencing data. A Rotavirus Classification Working Group (RCWG) established a system to identify and differentiate each genome segment according to particular cut-off points of nucleotide sequence identities. Each RV genome is designated by listing the segments in the order of VP7-VP4-VP6-VP1-VP2-VP3-NSP1-NSP2-NSP3-NSP4-NSP5/6 with their genotypes being represented by Gx-P[x]-Ix-Rx-Cx-Mx-Ax-Nx-Tx-Ex-Hx (Matthijnssens et al., 2008a,b, 2011) as discussed more thoroughly in Chapter 2.10.

A multiple amino acid sequence alignment of the current 19 E genotypes (Fig. 2.4.1B) provides information about similarities that may be related to protein function. Most mammalian NSP4 proteins contain 175 amino acids while NSP4s from avian, camel and fox species have smaller and larger proteins, respectively (Table 2.4.1 and Fig. 2.4.1B). This figure aligns NSP4 proteins from mammalian viruses on the top followed by those from mouse (E7), avian [pigeon (E4), chicken (E10), turkey (E11)], vicuna (E16), fox (E19) and camel (E15) at the bottom. Highly conserved residues in the mammalian viruses are evident that diverge in the NSP4 proteins from rodent, avian and camel viruses. Fig. 2.4.1C shows a schematic of the known functional domains in NSP4 in more detail. These figures illustrate that NSP4, which is synthesized as an ER transmembrane glycoprotein, consists of three hydrophobic domains (H1–H3) with two N-linked high mannose glycosylation sites oriented to the luminal side of the ER in the H1 domain (Chan et al., 1988; Bergmann et al., 1989). The H2 transmembrane domain traverses the ER bilayer and serves as an uncleaved signal sequence and aa 85–123 are reported to be involved in ER retention although this is likely to be an indirect involvement because these residues are located in the cytoplasmic domain of NSP4 (Mirazimi et al., 2003). The H3 domain consists of an amphipathic α-helix, which is part of the viroporin domain (Hyser et al., 2010). The remainder of NSP4 is an extended cytoplasmic domain containing a highly conserved coiled-coil region and a flexible C-terminus (Estes and Kapikian, 2007). The coiled-coil domain contains a calcium ion binding site coordinated by amino acids E120 and Q123, which are 100% conserved among all RVA viruses. This is consistent with the evidence showing that this region plays a critical role in conserved functions of NSP4 across different species. The number of multiple overlapping binding domains with distinct partners is striking (Fig. 2.4.1C). An interspecies variable domain (ISVD, aa 135–141) is clearly evident in the amino acid alignment. This alignment highlights that there are multiple other highly conserved domains without current functions, which suggest areas for future research.

The coiled-coil domain (aa 95–137) is the only domain whose structure has been determined by crystallographic studies. Challenges for obtaining crystals of the full-length protein are due to the presence of the N-terminal hydrophobic domains and a C-terminal flexible region, which leads to full-length NSP4 being highly insoluble in aqueous solution. Thus, crystallization studies have focused on the soluble domains. An initial study, which used a synthetic peptide composed of aa 95–137 of the simian strain SA11 (Bowman et al., 2000), and later studies using bacterially expressed protein (aa 95–146) from SA11 and a human I321 strain reported that the coiled-coil domain forms a tetramer. In spite of crystallizing longer proteins, only aa 95–137 have been resolved in the structures, and the amino acids after 137 are disordered (Bowman et al., 2000; Deepa et al., 2007). The first structure from the SA11 NSP4 peptide showed a tetramer with a calcium ion bound and coordinated by aa E120 and Q123. In contrast, the structure of the NSP4 coiled-coil domain from a human RV ST3 strain formed a pentameric coiled-coil that lacked a calcium ion at its core (Chacko et al., 2011). In these studies, the tetramer was crystallized at neutral pH while the pentamer was crystallized at low pH. It was unclear whether the two oligomeric forms of the coiled-coil domain from the two different strains represented a calcium-dependent, strain-dependent, or pH-dependent phenomenon. This was tested by determining the crystal structure of the NSP4 coiled-coil domain from a single strain SA11 (Sastri et al., 2014); these studies showed that at neutral pH this domain exists as a tetramer, which binds a calcium ion at its core (Fig. 2.4.2, structure on the left). However, mutation of the calcium- coordinating amino acids resulted in formation of a pentamer that lacks a calcium ion at its core (Fig. 2.4.2, structure on the right). Further solution studies showed that the NSP4 coiled-coil domain can directly bind calcium at physiological pH and exists as a tetramer. In contrast, at low pH, the coiled-coil domain does not bind calcium and forms a pentamer. This structural plasticity of NSP4 regulated by pH and calcium may be the basis for its pleiotropic functions during RV replication and pathogenesis.

The spectrum of responses to RV infection varies and can be asymptomatic, mild or severe, which results in a life-threatening dehydrating illness. The incubation time is short (<48 h), and illness often begins with a sudden onset of nausea and vomiting, followed by diarrhea that leads to a high frequency of dehydration and a mean duration of illness lasting 5–6 days. Several studies have shown infection leads to extra-intestinal spread of virus with antigenemia and viremia, and limited systemic replication in a variety of sites occurs frequently (Blutt et al., 2007). It remains unclear whether this systemic spread and replication cause any specific pathology in the normal host, but fever has been associated with antigenemia (Sugata et al., 2008), and antigenemia is reported to be associated with more severe clinical manifestations of acute gastroenteritis (Hemming et al., 2014).

The mechanism of diarrhea induction is best characterized through studies of the paracrine action of eNSP4 on uninfected intestinal epithelial cells. NSP4 binding to intestinal epithelial cell lines has been shown to initiate a signaling pathway that involves activation of phospholipase C (PLC), elevation in inositol 1,4,5-trisphophate (InsP₃) and release of ER Ca²⁺ stores that increases cytoplasmic Ca²⁺ resulting in Ca²⁺-dependent Cl⁻ secretion (Tian et al., 1995; Ball et al., 1996; Dong et al., 1997). Crypt cells isolated from mice also respond in a similar manner (Morris et al., 1999). Studies in mice lacking the cystic fibrosis transmembrane regulator (CFTR) channel showed that age-dependent diarrheal disease results from an event downstream from increased cytoplasmic Ca²⁺ (Morris et al., 1999). Since the CFTR channel is a c-AMP-regulated Cl⁻ channel, the finding that diarrhea is observed in CFTR knockout mice following RV infection or NSP4 treatment indicated that a Cl⁻ channel different than CFTR mediates this effect (Angel et al., 1998; Morris et al., 1999). Cl⁻ secretion is age-dependent in CFTR mice, indicating that age-dependent disease may result from an age-dependent induction, activation or regulation of the Cl⁻ channel, which was proposed to be the calcium-activated chloride channel (CaCC). This pathway would be one that is activated when eNSP4 is released from virus-infected cells and the affected cells are neighboring, noninfected secretory crypt or epithelial cells (Fig. 2.4.3). These studies also illustrate that NSP4 is a novel secretory agonist because CFTR knockout mice do not respond to classical secretory agonists (Morris et al., 1999). Confirmatory data for this proposed model came from showing NSP4 induces diarrhea by activation of the proposed CaCC TMEM16A and inhibition of Na⁺ absorption (Ousingsawat et al., 2011) and that CaCC inhibitors prevent RV secretory diarrhea in neonatal mice (Ko et al., 2014). NSP4 also does not have a direct, specific effect on chloride transport in brush border membranes, confirming the hypothesis that its actions are through triggering signal transduction pathways (Lorrot and Vasseur, 2006).

Many viruses induce apoptosis for release and dissemination of viral progeny. Apoptosis may also be a contributing factor to pathogenesis through destruction of differentiated absorptive enterocytes. In murine EDIM RV-infected mouse pups, apoptosis, shown by detection of caspase-3 in infected mature small intestinal enterocytes, was hypothesized to be responsible for the observed villous atrophy (Boshuizen et al., 2003). Further in vitro studies in human colon carcinoma HT-29 and fully differentiated Caco-2 cells indicated that RV infection induces apoptosis as demonstrated by peripheral condensation of chromatin and fragmentation of the nuclei (Superti et al., 1998; Chaibi et al., 2005). Apoptosis in RV-infected MA104 cells has been attributed to the activation of Bax, a proapoptotic member of the Bcl-2 family (Martin-Latil et al., 2007).

Perturbing the host Ca²⁺ signaling pathway is not only important in pathogenesis but also for viral replication and morphogenesis. RV infection induces dramatic changes in cellular calcium homeostasis causing a two- to fourfold elevation in cytoplasmic calcium ([Ca²⁺]cyto). Cellular calcium homeostasis is tightly regulated as calcium ions are ubiquitous intracellular signaling molecules responsible for controlling a plethora of cellular processes. To determine the viral protein(s) responsible for the elevation in [Ca²⁺]cyto, individual RV proteins were expressed in Sf9 insect cells or a variety of mammalian cell lines and [Ca²⁺]cyto was measured. NSP4 was identified as the sole RV protein responsible for the increase in [Ca²⁺]cyto by a PLC-independent mechanism (Tian et al., 1994,  1995; Berkova et al., 2007; Diaz et al., 2008,  2012). NSP4 expression in cells recapitulates all of the changes in calcium homeostasis observed in RV-infected cells, while silencing of NSP4 expression in infected cells inhibits these changes (Tian et al., 1994; Berkova et al., 2007; Diaz et al., 2008;  2012; Zambrano et al., 2008). A breakthrough in identifying the mechanism of iNSP4-mediated global disruption in cellular calcium homeostasis was attained when NSP4 was characterized as a viroporin in the ER (Fig. 2.4.2) (Hyser et al., 2010). Viroporins are small, hydrophobic proteins that contain a cluster of basic residues (Lys or Arg) and an amphipathic-α-helix that oligomerizes to create a transmembrane aqueous pore. The NSP4 viroporin domain is comprised of amino acids 47–90, which are critical for elevation of [Ca²⁺]cyto, since mutation of either the cluster of basic residues or amphipathic α-helix abolishes the observed elevation in [Ca²⁺]cyto (Hyser et al., 2010). Whether NSP4 viroporin functions as a channel or a pore to elevate cytoplasmic calcium requires further study. Although NSP4 mediates the release of calcium from the ER store into the cytoplasm, this alone is not sufficient to explain the dramatic increase in [Ca²⁺]cyto or the progressive increase of calcium permeability of the plasma membrane.

The ability of NSP4 to interact with multiple viral and cellular proteins may rely on the differential localization of this protein in infected cells during the RV replication process. In the ER, NSP4 interacts with the ER transmembrane chaperone calnexin, which accelerates NSP4 carbohydrate trimming by glucosidase I and II as well as affects the stability of infectious particles. TLPs produced in calnexin-silenced cells are less infectious, less dense in CsCl gradients (1.35 g/cm³) compared to control (1.36 g/cm³) and do not assemble properly suggesting that NSP4–calnexin interactions and ER carbohydrate quality control affect TLP assembly (Mirazimi et al., 1998; Maruri-Avidal et al., 2008).

Roles for NSP4 in innate immune responses have been described. NSP4 has adjuvant properties when coadministered with model antigens in mice (Kavanagh et al., 2010). This response may be through NSP4 functioning through pathogen-associated molecular pattern (PAMP) activity (Ge et al., 2013). Treatment of macrophages with NSP4 purified from the media of RV-infected Caco-2 cells induces secretion of the pro-inflammatory cytokines TNF-α and IL-6 as well as activation of p38 and JNK mitogen-activated protein kinases (MAPKs) and nuclear factor NF-κB (Ge et al., 2013) (Fig. 2.4.3). Secretion of cytokines was mediated through Toll-like receptor 2 (TLR-2) and may provide a mechanism for the production of proinflammatory cytokines associated with the clinical symptoms of infection in humans and animals.

Due to their limited coding capacities, most viruses encode multifunctional proteins that accomplish a variety of tasks during infection. NSP4 is a pleiotropic protein that performs many functions as described in this review. Our knowledge of the molecular mechanisms by which iNSP4 and eNSP4 mediate RV-induced pathogenesis, replication, and morphogenesis has increased over the last years. However, additional studies are needed to elucidate the different structural forms of NSP4 in order to augment biochemical and functional studies addressing the many functions that NSP4 performs during RV infection. Many questions remain for the structure of NSP4 and how it facilitates multiple functions of NSP4 including the functions of eNSP4 and iNSP4 in infection and pathogenesis (Table 2.4.2).

We gratefully acknowledge partial support of our rotavirus research from NIH grants R01 AI080656 and P30 DK56338 (M.K.E.) and R37 AI36040 (B.V.V. Prasad). We also thank Dr Beatrice Boniotti, Istituto Zooprofilattico Sperimentale della Lombardia e dell’ Emilia Romagna “Bruno Ubertini,” Brescia, Italy for sharing the sequence of the NSP4 from a fox prior to its publication.