INTRODUCTION

Top Previous Next References

Importance and Impact of Gastroenteritis

Acute gastroenteritis is one of the most common diseases of humans. In the United States, it is second only to acute viral respiratory disease as a cause of acute illness 61. Worldwide, more than 700 million cases of acute diarrheal disease are estimated to occur annually just in children under the age of 5 years 230. Gastroenteritis most commonly is manifested clinically as mild diarrhea, but more severe disease, ranging from upper gastrointestinal symptoms (nausea and vomiting) to profuse diarrhea leading to dehydration and death, may occur. The annual mortality associated with gastroenteritis has been estimated to be 3.5 to 5 million, with the majority of deaths occurring in developing countries 21, 97, 250.

Acute gastroenteritis is caused by a number of different agents, including bacteria, viruses, and parasites. Until recently, many cases were attributed to viruses because of the failure to identify a bacterial or parasitic pathogen. Although Reimann et al. 214 and Gordon et al. 77 suggested more than 50 years ago that viruses are a cause of diarrhea after inducing illness in volunteers with stool filtrates free of bacteria, it was only in 1972 that a virus (Norwalk virus) was definitively identified as a cause of acute gastroenteritis 139. Since that time, the number of viral agents associated with acute gastroenteritis has increased progressively.

Many different viruses have been found in the stools of persons with gastroenteritis. However, a causative role for each of these viruses has not been established. Criteria to define a virus as an etiologic agent of gastroenteritis include (i) identification of the virus more frequently in subjects with diarrhea than in controls, (ii) demonstration of an immune response to the specific agent, and (iii) demonstration that the beginning and end of the illness correspond to the onset and termination of virus shedding, respectively 148. Table 1 lists those viruses established as etiologic agents of gastroenteritis and other viruses found in stools that have not yet fulfilled the aforementioned criteria 9. Coronaviruses 34, 62, picobirnaviruses and picotrirnaviruses 35, 96, 174, pestiviruses 13, 257, and toroviruses 17, 18, 251 are candidate diarrheal agents because they are associated with diarrheal illness in animals and have been found in the stools of humans with gastroenteritis. However, these viruses have not fulfilled the criteria needed to establish them as diarrheal agents in humans 134, 148 and will not be considered further in this review. Non-group F adenoviruses and several enteroviruses (e.g., Coxsackie A and B viruses) are found in the stools of non-ill individuals with frequencies similar to those seen in ill individuals 148.

Of the viruses that have been shown definitively to be causes of acute gastroenteritis, only the human caliciviruses cannot be grown in cell culture. The study of rotaviruses, enteric adenoviruses, and astroviruses has been facilitated greatly by the ability to propagate these viruses in cell culture. The ability to cultivate these viruses has allowed the production of reagents for use in diagnostic studies, a better understanding of factors correlated with immunity to infection, and the elucidation of each virus's life cycle. Although human caliciviruses have defied numerous attempts to propagate them in cell culture to date, recent developments in their study by using molecular biology techniques have increased our understanding of this group of viruses. This article will review the new diagnostic assays that have resulted from these advances.

HISTORY OF HUMAN CALICIVIRUSES

Top Previous Next References

Recognition

In 1968, an outbreak of acute gastroenteritis (termed winter vomiting disease) occurred among students and teachers in a school in Norwalk, Ohio 1. The primary attack rate was 50%, with a secondary attack rate of 32%. Illness was characterized by nausea and vomiting in >90% and diarrhea in 38% of affected individuals, and the duration of illness was usually 12 to 24 h. Subsequently, organism-free filtrates of stools collected from affected individuals induced similar illness in human volunteers, and different treatments of the inoculum suggested that the causative agent was a small (<36 nm), ether-resistant (nonenveloped), relatively heat-stable virus 24, 57, 58. Attempts to propagate the agent in cell culture and organ culture were unsuccessful 24, 57.

In 1972, Kapikian et al. 139 used immune electron microscopy (IEM) to identify 27-nm viral particles, the Norwalk agent, in a fecal filtrate used to induce illness in human volunteers. Virus particles were precipitated in an antigen-antibody reaction using convalescent-phase serum from a volunteer who became ill following inoculation with the fecal filtrate. Antigen-antibody complexes were then visualized with an electron microscope. The assay was modified to quantify the amount of antibody in serum and demonstrated that significantly more antibody was present in convalescent-phase serum than in acute-phase serum. Because of these data, Norwalk virus (NV) was proposed to be the etiologic agent of the Norwalk, Ohio, outbreak of gastroenteritis 30. Later studies demonstrated that other small, round-structured viruses (SRSVs) morphologically similar to NV were associated with outbreaks of gastroenteritis 7, 59, 238, but NV remained the prototype of these fecal viruses.

Morphologically typical caliciviruses were first recognized in stool samples by Madeley and Cosgrove in 1976 176. These investigators found calicivirus particles in the fecal specimens of 10 children, but some of the children were asymptomatic, so that no conclusions as to the pathogenicity of the virus could be made. Later that year, Flewett and Davies 68 identified calicivirus particles in the small bowel from a fatal case of gastroenteritis, but because adenovirus particles also were present in large numbers, the significance of the calicivirus particles could not be determined. However, in the next several years, caliciviruses were definitively associated with several outbreaks of gastroenteritis 37, 38, 44, 45, 185 and became recognized as another virus group associated with gastroenteritis 33, 175.

Based on properties of NV and its appearance by electron microscopy, NV and other small round viruses were thought to be parvovirus-like 57, 137. However, in 1981 Greenberg et al. 93 published data suggesting that NV has a single structural protein with an estimated molecular mass of 59 kDa and proposed that Norwalk virus might be a calicivirus. Nine years later, Jiang et al. 216 provided molecular evidence that NV is a calicivirus by demonstrating that the viral genome consists of positive-sense, single-stranded, polyadenylated RNA. Subsequent elucidation of the complete sequence of NV and related SRSVs confirmed the genetic relatedness of these viruses to other caliciviruses 55, 108, 131, 153, 154, 225.

The classification of viruses responsible for acute gastroenteritis was first based on morphology (Table 2). For example, NV was the prototype of a group of agents initially called SRSVs. Recently, rapid advances in molecular biology allowed these viruses to be classified based on their genome characteristics, and most of the previously named SRSVs were shown to belong to the Caliciviridae.

Morphology. The virions are composed of a single major capsid protein. Structural analysis of virus particles from stool is limited by the small number of particles present in these samples. However, by negative-stain electron microscopy, the NV has an indistinct "feathery" outer edge and an indistinct surface substructure, although there is a suggestion of indentations on its surface (Fig. 1A). Expression of the capsid protein using the baculovirus system results in the self-assembly of the NV protein into virus-like particles (VLPs) (Fig. 1B). By negative-stain electron microscopy, these VLPs have a morphology similar to that of the native virus. The structure of these VLPs has been resolved by electron cryomicroscopy and computer image processing as well as by X-ray crystallographic methods 211, 212. The capsid exhibits a T = 3 icosahedral symmetry. The major structural protein folds into 90 dimers that form a shell domain from which arch-like capsomers protrude. A key characteristic of this architecture is 32 cup-shaped depressions at each of the icosahedral fivefold and threefold axes. These cup-like depressions are more prominent in some strains (particularly the Sapporo-like viruses [SLVs]), leading to the characteristic "Star of David" appearance from which caliciviruses get their name (Fig. lC). The name calicivirus is derived from the Latin calyx, meaning cup or goblet, and refers to the cup-shaped depressions visible by electron microscopy.

View larger version (63K): [in this window] [in a new window]   FIG. 1.   Electron micrographs of (A) NV, (B) baculovirus-expressed NVL particles, and (C) Sapporo virus. Bar, 100 nm.

Genomic organization and classification. In 1990, the genome of NV was cloned and characterized 124. More recent studies have characterized and completely sequenced this and other human viruses in the family 55, 108, 131, 153, 154, 170, 171, 225. Recently, the International Committee on the Taxonomy of Viruses established and approved four genera within the family Caliciviridae, including two human calicivirus (HuCV) genera 86. These nonenveloped viruses have a diameter of 27 to 40 nm by negative-stain electron microscopy and a buoyant density of 1.33 to 1.41 g/cm³, and they contain a positive-sense polyadenylated single-stranded RNA of approximately 7.6 kb (Fig. 2). The two HuCV genera currently have the tentative names of "Norwalk-like viruses" (NLVs; type strain, NV [Hu/NLV/Norwalk virus/8FIIa/1968/US]) and the "Sapporo-like viruses" (SLVs; type strain Sapporo virus [Hu/SLV/Sapporo virus/1982/JA]). The proposed nomenclature for HuCVs is species infected/virus genus/virus name/strain designation/year of isolation/country of isolation. Table 3 shows the strain name and proper name for common NLV and SLV strains and for the strains discussed in this review. Common names will be used for the human virus strains throughout the rest of the review.

View larger version (19K): [in this window] [in a new window]   FIG. 2.   Schematic of the genomic organization of viruses from the four different genera of the Caliciviridae. Strains [GenBank numbers] for which the entire genomic sequence is available are presented for each of the genera: NV [M87661] 108, 131, Manchester virus [X86560] 170, 171, rabbit hemorrhagic disease virus [M67473] 187, and feline calicivirus [M86379] 31. A nonstructural polyprotein (aa 1719 to 1789) is encoded by the 5' end of the genome. The major capsid protein (shaded area) is in frame for the SLVs and lagoviruses, while it is in the +1 frame for NLVs and 1 frame for the vesiviruses. A basic protein is encoded at the 3' end of the genome for all four genera. SLVs have another ORF (+1) overlapping the capsid ORF that is not seen in the other genera.

View larger version (14K): [in this window] [in a new window]   FIG. 3.   Unrooted phylogram generated using the entire capsid amino acid sequence of nine genogroup II (GII) strains and one genogroup I (NV) strain and the PAUP program in the Genetics Computer Group suite of programs 73. Strains within different genogroup II genetic clusters are indicated, with the genogroup I strain NV serving as an outgroup. GenBank accession numbers used for this analysis are as follows: NV, M87661; HV, U07611; SMA, U75682; Melksham virus, X81879; Oth-25, L23830; MX, U22498; TV, U02030; Camberwell virus, U46500; Bristol virus, X76716; and LV, X86557.

During the 1970s and 1980s, tests for the diagnosis of NV infections were designed using reagents from previously infected humans. The number of virus particles in the stools of infected subjects is sufficiently small that hyperimmune sera could not be produced in animals. (An exception was the use of partially purified Sapporo virus to produce hyperimmune sera in guinea pigs [192].) The inability to propagate NV and related viruses in cell culture also prevented the production of animal hyperimmune sera. Thus, stools of acutely infected individuals served as a source of virus antigen, and convalescent-phase sera from infected individuals were used as hyperimmune sera. These restrictions limited the general availability of these diagnostic tools to only a few research laboratories 40. However, a number of tests were developed with these reagents and used to begin to define the epidemiology of NV and other human caliciviruses.

Direct electron microscopy. Detection of enteric viruses in stool specimens using direct electron microscopy requires virus concentrations of at least 10⁶ per ml of stool 56. In many laboratories, stool is mixed with phosphate-buffered saline or tissue culture medium to form a 10 to 20% suspension before being clarified by low-speed centrifugation. Subsequent concentration of virus may be achieved by ultracentrifugation or precipitation with ammonium sulfate. Virus particles are negatively stained using one of a number of available electron-dense stains, including phosphotungstic acid, uranyl acetate, and ammonium molybdate 56. Viral particles must be distinguished from nonviral material, and this can be particularly difficult for the NLVs that do not have typical calicivirus morphology 56, 140. The small numbers of viral particles present in fecal samples make direct electron microscopy, even after concentration, relatively insensitive. Nevertheless, direct electron microscopy is used to screen fecal specimens for enteric viruses in the public health laboratories of many countries, and it is clear that talented electron microscopists who receive fecal samples collected early in the course of an infection can often detect viruses 52, 179, 246, 256. However, this method requires highly skilled microscopists and expensive equipment, making it not feasible for large epidemiological or clinical studies.

IEM . The use of immune serum to aid in virus identification was first described for tobacco mosaic virus in 1941 2. This method had been used for only a few human viruses 22, 49, 135 before Kapikian et al. 139 adapted it for the detection of NV in stool filtrates. Since then, it has been used to identify many other SRSVs in stool samples 27, 59, 141, 238, 242. A clarified stool suspension is incubated with a reference serum or saline (as a control) for 1 h at 37°C, and immune complexes are pelleted by ultracentrifugation 140. The pellet is suspended in a few drops of distilled water and negatively stained as for direct electron microscopy. The saline control is necessary to demonstrate the specificity of the immune serum because virus clumping may occur in the absence of antibody 56, 194. A positive sample has aggregates of viral particles which are absent in the saline control sample. Although IEM was first used to detect NV, it is positive on stool samples from only approximately half of volunteers who become ill following artificial challenge with NV 237. When IEM has been applied to outbreaks of gastroenteritis associated serologically with NV infection, it has been positive in only about 20% of the outbreaks, and just over one third of stool samples from affected individuals in these outbreaks are IEM positive 141. The lack of detection probably reflects the very low concentration of virus in many stool samples and the lack of collection of the first early diarrheal stool samples.

IEM was found to be a specific and reproducible method for antibody determination, but it is laborious, cumbersome, and time-consuming to perform. The immune adherence hemagglutmation assay (IAHA) was developed to allow the evaluation of Norwalk antibody levels in greater numbers of sera so that epidemiological studies of seroprevalence could be performed 138. Viral particles are purified from stool and used as antigen, and antigen-antibody-complement interactions are detected in a microtiter plate format by agglutination of sensitive human O erythrocytes. Kapikian et al. used the IAHA to demonstrate seroprevalence rates ranging from <20% in children to approximately 50% in adults in the fifth and sixth decades of life 91, 138. Although the assay has the advantage of requiring less antigen for its performance than complement fixation antibody assays, it was soon replaced with another assay, the blocking radioimmunoassay (RIA), which uses even less antigen and is more sensitive 91, 94, 136. IAHA also could not be adapted to detect viral antigen in stool specimens.

RIA was developed as an alternative to IEM for the detection of NV antigen in stool 91, 94. The assay is used in a microtiter format, and it detects both particulate and soluble antigen 133. Preinfection and convalescent sera from a volunteer experimentally infected with NV and known to have a high-titered antibody response (by IEM and IAHA) are used to capture virus antigen in duplicate wells. Immunoglobulin G (IgG) purified from a convalescent serum of a NV-infected chimpanzee or human volunteer and radiolabeled with ¹²⁵I is used as a detector system. The use of preinfection (negative) and convalescent (positive) serum differentially captures NV antigen and is indicated by greater radioactivity (counts per minute) in the convalescent than in the preinfection sample. Positive-negative (P/N) ratios greater than or equal to 2 indicate the presence of virus antigen in the sample 94. RIAs for the SMA and the morphologically typical HuCV were developed later 60, 191. In the latter assay, pre- and postvaccination sera from guinea pigs hyperimmunized with purified virus are used in place of human-derived reagents 191. For optimal specificity, positive samples must be confirmed using a blocking assay. Convalescent serum from a calicivirus-infected individual is incubated with captured virus antigen prior to addition of the radiolabeled IgG; a reduction in bound radioactivity of 50% or more indicates specific capture of virus antigen in samples with a P/N ratio of 2 or more 191. The RIAs using human-derived reagents do not require a blocking assay for confirmation of positive samples. All of these assays are specific (i.e., do not detect unrelated HuCVs or other enteric viruses) and 10- to 100-fold more sensitive than IEM 60, 94, 191. However, RIAs are negative when applied to stool samples from as many as one third of symptomatic volunteers experimentally infected with NV 231.

RIAs have been modified to detect virus-specific antibodies using a blocking format similar to that described above. A convalescent-phase serum from an HuCV-infected subject is used to capture partially purified virus antigen. Serial twofold dilutions of the serum to be tested for antibody determination are added next, and after an overnight incubation, ¹²⁵I-labeled, virus-specific IgG is added. If virus-specific antibodies are present in the serum being tested, binding of the radiolabeled IgG is blocked. A reduction in bound radioactivity of 50% or greater is used to define the presence of virus-specific antibodies, and the reciprocal of the last dilution at which 50% or greater blocking occurs is the titer of virus-specific antibody 94. RIA blocking assays have been developed for the detection of antibody to NLVs (NV and SMA and SLVs 60, 91, 94, 191. The RIA blocking assay for NV-specific antibody is 10 to >200 times more sensitive than the IAHA 91, 94.

RIA antigen and antibody detection assays were used to further characterize infection and illness in experimentally induced human infection 23, 48, 231, to perform seroprevalence studies in different populations 47, and to investigate outbreaks of gastroenteritis 14, 15, 46, 76, 92, 95, 98, 99, 141-144, 151, 169, 235, 253. The application of these assays helped identify NV and related viruses as a common cause of nonbacterial gastroenteritis outbreaks. For example, a review of 74 outbreaks of gastroenteritis investigated by the Centers for Disease Control between 1976 and 1980 showed that 42% of the outbreaks were associated with NLVs and an additional 23% were possibly associated with NLVs 141.

As the technology to perform immunoassays using nonisotopic reporters developed, enzyme immunoassays (EIAs) utilizing the same principles as the RIAs described above were developed for the detection of NV infection. Partially purified IgG is labeled with biotin or horseradish peroxidase in place of ¹²⁵I. The antigen detection EIAs detect NV in stool samples with a frequency similar to that seen with RIAs 72, 115, 178. Blocking EIAs for the detection of serum antibody are approximately twofold more sensitive than the comparable blocking RIA, although both assays detect fourfold or greater rises in serum antibody with similar frequency. An advantage of EIAs over RIAs is the increased stability of the reagents used to perform the tests. ¹²⁵I-labeled anti-NV IgG has a much shorter shelf life (several days to 2 weeks) than does anti-NV IgG labeled with biotin (3 months or more at 20°C) or horseradish peroxidase (6 months at 4°C) 72, 115. Other advantages of the EIA include the elimination of the use of radioisotopes and the decreased time needed to perform the blocking EIA compared to the blocking RIA (3 days and 6 days, respectively) 72.

EIAs for the detection of SMA and HV were developed later 178, 241, as were EIAs for the detection of IgM and IgA serum antibody responses 64, 65. The EIAs were applied in a fashion similar to RIAs, being used on specimens from experimental human infection studies 71, 177 and from outbreak investigations 16, 29, 114, 117, 249.

Another approach used for the evaluation of antibody responses has been the Western blot assay. For this assay, virus was partially purified from stool for use as an antigen 112. Multiple bands appeared on blots used to assay human serum specimens, but when acute- and convalescent-phase sera were tested, increasing reactivity with a protein of approximately 63 kDa was identified. The specificity of this reactivity was confirmed using virus purified by isopycnic cesium chloride density gradient centrifugation 112. In some assays convalescent sera also showed increasing reactivity with a second band with an approximate molecular size of 33 kDa 201, which may represent the soluble protein identified following proteolytic cleavage of the capsid protein 111. When applied to clinical specimens from outbreaks of gastroenteritis, the Western blot assay gave results comparable to those obtained using IEM, although antibody detected by Western blot persisted for a longer period of time than did that detected by IEM 112, 145. The Western blot assay may detect a larger number of epitopes than IEM because it is able to detect antibody to related viruses that is not recognized by IEM 112. While this assay has theoretical advantages for the confirmation of HuCV infections, it has not been used by many laboratories, probably because of high background reactivities and lack of a reproducible source of viral antigen.

The successful cloning of NV led to the development of new reagents and methods for the diagnosis of infections caused by HuCVs. When the NV capsid protein was expressed in a baculovirus expression system, VLPs were generated 130. These VLPs were subsequently shown to be morphologically and antigenically similar to native virus particles 88. The VLPs were used to immunize different animal species to produce polyclonal and monoclonal immune sera that could then be used to establish EIA-based diagnostic assays. Virus sequence was used to design primer pairs for the detection of HuCVs using reverse transcription (RT)-PCR. The development and application of these newer assays are described below.

EIA with hyperimmune animal sera. The production of NV VLPs provided sufficient quantities of viral capsid antigen to allow the generation of hyperimmune sera in mice, guinea pigs, and rabbits 130. Hyperimmune sera from these animals have NV-specific antibody titers of 1:256,000 to >1:1,000,000. Subsequently, VLPs have been produced for other HuCVs, including Mexico virus (MX), SMA, HV, Desert Shield virus, Toronto virus (TV), Grimsby virus (GRV), Sapporo virus, Southampton virus, and Lordsdale virus (LV) 55, 87, 102, 109, 126, 160, 161, 198. Polyclonal hyperimmune animal sera produced by immunization of different animal species with VLPs have been used to develop antigen detection EIAs for use in clinical specimens 102, 121, 128. These immune sera have been quite specific, detecting homologous recombinant VLPs in an EIA format but not reacting with heterologous VLPs.

EIA with MAbs. Monoclonal antibodies (MAbs) have been prepared using native NV, native SMA, and rNV VLPs 110, 113, 239. Similar to what was seen with polyclonal sera, these MAbs are often type specific, recognizing the capsid protein of the immunizing virus but not that of other NLVs. The MAbs have been evaluated in limited studies for the detection of virus in stool samples. One assay format used a pool of two MAbs for antigen capture and also antigen detection. The detector antibodies were conjugated to horseradish peroxidase. This assay was reported to have a twofold greater sensitivity (as measured by the amount of virus detected) than assays using polyclonal sera, detected NV in 15 of 15 stool samples from subjects infected with NV, and failed to detect HV in any of nine stool samples from infected subjects 113.

EIA with VLPs. The production of VLPs for NV and other NLVs has allowed the detection of immune responses to infection with these viruses. In these assays, VLPs are used to coat 96-well microtiter plates, and after blocking and washing steps, serial dilutions of human serum are added. Antibodies reacting with the VLPs are detected with goat anti-human immunoglobulin conjugated to an enzyme (e.g., horseradish peroxidase or alkaline phosphatase). The assay can detect total or class- or subclass-specific serum antibodies, depending on the reagents used to detect the bound human antibody 80, 116, 130, 240. The assay also has been modified to detect NV-specific IgA in fecal samples 203.

Two different approaches have been used to determine the amount of virus-specific antibodies in a serum specimen. The approach used by most laboratories is to perform a serial dilution of the serum specimen and to determine the last dilution that gives a reading above an empirically determined cutoff value 28, 41, 105, 116, 163. A second approach is to measure the amount of signal (e.g., optical density) from a single dilution of serum and to relate the measured signal to that measured using a standard reference serum 127, 189, 195, 207. Advantages of the latter method are that it allows a larger number of sera to be tested and is simpler and less expensive to perform. However, practical and theoretical problems limit the utility of testing a single dilution of serum. The practical problem is that well-characterized standard sera to be used as a reference reagent are not generally available, so that most laboratories cannot set up these assays. The theoretical problem is that the measurement of signal at a single dilution is affected by many factors besides the amount of antibody present in the sample, including the affinity of antibody in the sample for the test antigen and variability in the serum constituents that can affect antigen-antibody interactions. Thus, although antibody determinations using a single dilution of serum have been used in both seroprevalence studies 127, 207 and other epidemiological studies 195, most laboratories quantify virus-specific antibodies in serum using endpoint titration 28, 41, 105, 116, 163.

The VLP-based antibody detection EIA is more sensitive than assays using human reagents in RIA-, blocking EIA-, or IEM-based formats 88, 189. Total anti-NV antibody levels are 1.25- to >40-fold higher using an rNV VLP EIA assay than those obtained using RIA or blocking EIA. When applied to sera from experimental human or chimpanzee infection and from outbreaks of gastroenteritis, the rNV VLP EIA detects fourfold or greater increases in serum antibody levels more frequently than IEM and at least as frequently as RIA and blocking EIA 88, 189. The antibody rNV VLP EIA identified infection following experimental human challenge better than the antigen detection EIA, detecting 40 of 41 (98%) infections compared to 36 of 41 (88%) infections 79. Assays utilizing other VLPs, including rMX, rTV, rHV, rSouthampton virus, and rLV, have also been developed 127, 195, 206.

The antibody detection EIA has been used to characterize IgG, IgM, and IgA serologic responses following experimental human infection with NV 80. Eight of 13 infected subjects had fourfold or greater increases in virus-specific antibody levels between 8 and 11 days following infection; ill subjects (eight of nine) were more likely to have these early responses, while antibody rises were seen in asymptomatic subjects only after 15 days. All infected subjects (n = 14) developed virus-specific IgM serum antibody. Virus-specific IgM serum antibody was present as early as 9 days following infection, but it did not develop in some subjects until 2 weeks after infection. IgM serum antibody could still be detected 3 months later in some subjects. Fourfold or greater increases in virus-specific IgA serum antibody were detected in all nine symptomatic infections but in only two of five asymptomatic infections. Virus-specific geometric mean serum antibody levels of infected and uninfected subjects were similar 3 months after challenge 80. The kinetics of the IgM and IgA responses are similar to those seen in earlier studies using human reagents 64, 65. The presence of neither virus-specific serum IgG, IgM, or IgA nor of fecal IgA is associated with protection from infection 80, 203, 240. Low serum antibody levels appear to be associated with a decreased likelihood of infection following experimental human challenge and natural exposure 79, 226.

Virus-specific IgM serum antibody has also been detected using an IgM capture assay. In this assay, goat anti-human IgM is used to coat microtiter plates, and dilutions of the test serum are then applied. VLPs are added in the next step, and VLPs bound by virus-specific IgM are detected using hyperimmune anti-NV rabbit serum 240. An alternative method uses a virus-specific MAb for VLP detection 28. Antibody levels are four- to eightfold higher using the IgM capture assay compared to the assay in which IgM bound to VLPs is detected 240. In two different studies following experimental human infection, IgM capture assays detected IgM responses in 15 of 15 and 14 of 15 subjects, with infection documented by fourfold or greater IgG responses 28, 240.

Detection of infection caused by heterotypic HuCVs. The serologic responses measured using VLP-based antibody detection EIAs have been characterized using sera collected during studies of experimental human infection and during evaluations of gastroenteritis outbreaks. Heterologous rNV IgG responses occur following experimental human infection with HV or SMA although they are present at a lower frequency and magnitude than is seen following infection with NV 240 or when rMX (SMA-like) VLPs are used in the assay 206. Heterologous rHV IgG responses also can be demonstrated following NV infection 41. The results are similar to those obtained using human reagents in blocking EIAs, although heterologous seroresponses as measured by the older and newer assays occurred in different subjects 177, 240. Heterologous seroresponses appear to be limited to subjects who also have an IgG seroresponse to homologous viral antigen and who are ill. Heterologous IgM and IgA responses occur infrequently 240.

Hybridization assays. Only a few hybridization assays have been described for the detection of HuCVs. This is most likely due to the availability of the more sensitive RT-PCR assays at the time that cDNAs for HuCVs first became available. Jiang et al. 129 described a hybridization assay using ³²P-labeled cDNAs covering the same region of the genome as was amplified in RT-PCR assays. Stool suspensions (10 to 50%) were extracted with trichlorotrifluoroethane, and the viral nucleic acids were partially purified from the aqueous phase by digestion with proteinase K, extractions in phenol-chloroform and chloroform, and precipitation in ethanol. The hybridization assay detected NV with a sensitivity similar to that of RIA, but detected 100-fold less viral RNA and detected virus in stool samples 27% less frequently than an RT-PCR assay applied to the same samples. Thus, the hybridization assay was unable to detect NV when low titers of virus were present in a stool sample 129.

RT-PCR. The first RT-PCR assays were described within 2 years of the initial report of the successful cloning of the NV genome 53, 129. Since then a number of different RT-PCR assay formats have been developed, and these assays have become one of the principal means for the diagnosis of HuCV infections. A number of factors can affect the sensitivity and specificity of RT-PCR assays, including the sample being assayed, the method used for purification of viral nucleic acids, the primers used in amplification, and the method used for interpretation of test results. Approaches to addressing each of these factors are discussed below.

(i) Extraction methods. Clinical samples frequently contain substances that can inhibit the enzymatic activity of the reverse transcriptase and DNA polymerase enzymes used in RT-PCR assays. Thus, it is usually necessary to partially purify viral nucleic acids or otherwise prepare the sample prior to the performance of the RT-PCR assay. Two major considerations in selecting an extraction method are its efficiency of viral nucleic acid recovery and its ability to remove or inactivate RT-PCR inhibitors. Secondary considerations include the ease of performance of the method and the number of samples that can be processed at one time. A number of different approaches for viral nucleic acid purification from stool samples have been reported. Those that have been used for the detection of HuCVs are shown in Table 5.

(ii) Primer selection. The sensitivity and specificity of RT-PCR assays depend in large part on primer selection. Several factors affect the ability of a primer pair to detect a given NLV or SLV strain, including primer sequence, the amount of virus present in the sample to be assayed, and the temperature used for primer annealing during the PCR amplification process. The genetic diversity of NLVs and SLVs has made it difficult to select a single primer set with adequate sensitivity and specificity to detect all NLVs. In general, regions of the genome with the greatest degree of conservation between strains within the genera and within genogroups have been targeted for amplification and primer design (Fig. 4A). But even within these regions, the nucleotide identity can be as little as 36% (2C helicase region) to 53% (3D polymerase region) between strains of different genogroups 5, 43, 182, 197, 245, 248, 255, 256. Within a genogroup of NLVs, greater conservation is seen, but the nucleotide identity between strains can still be as little as 60 to 64% 245. These observations have led to the design and use of primer sets targeting multiple areas of the viral genome (Fig. 4A, Table 6).

View larger version (30K): [in this window] [in a new window]   FIG. 4.   (A) Schematic representation of the calicivirus genome and the regions amplified by common primer pairs. Modified from reference 220 with permission from Technomic Publishing Co, Inc., copyright 2000. See Table 6 for further details about primer sequences. hel, helicase; pol, polymerase. (B) Schematic representation of the NLV genome from which an internal standard control for genogroup I NLVs was made. The relative locations of selected primers and probes are noted for virus and internal standard (Int. Std.) RNA. The internal standard control RNA yields amplicons that are 123 bp shorter (347 bp) than those from NV genomic RNA (470 bp). A portion of the genomic sequence targeted by virus-specific probes (e.g., SR65 and p116) is not present in the internal standard control, allowing differentiation of virus-specific and internal standard amplicons by nucleic acid hybridization 221.

(iii) Other PCR conditions. The conditions (e.g., magnesium concentration and primer annealing temperature) of the RT-PCR assay are dictated, in part, by the primers used. The use of different DNA polymerases during the PCR amplification process has also been reported. Ando et al. 4 used a combination of Taq DNA polymerase and Pwo DNA polymerase for the amplification of a large (3 kb) fragment of the viral genome. Schwab et al. 222 utilized Tth polymerase in place of avian myeloblastosis virus reverse transcriptase and Taq polymerase and found that detection of NV in the Tth polymerase-based assay was comparable to that in the two-enzyme system. The use of a single enzyme for reverse transcription and DNA amplification also allowed the use of thermolabile uracil-N-glycolase (UNG) for the prevention of carryover contamination 222. dUTP is used in place of dTTP in the RT-PCRs, and the UNG degraded deoxyuracil-containing amplicons added (or carried over) to the reaction tube. The UNG is then inactivated by heating, and the viral RNA is amplified.

(iv) Confirmation of PCR products. A number of methods can be used to interpret the results of a PCR method. One of the simplest is gel electrophoresis. If a band of the size predicted from primer selection is seen following electrophoresis, the PCR is considered positive. This method has been used for NLV RT-PCR assays 10, 84, but it can yield false-positive results 11, 12. Nonspecific amplification of DNA occurs, particularly when more than 30 cycles of amplification are used, and the nonspecifically amplified DNA will occasionally migrate in a fashion similar to that expected for virus-specific amplicons. If this happens, the nonspecific amplification products may be misinterpreted as being virus-specific amplicons. Such nonspecific bands are frequently seen following the assay for NLVs in stool and shellfish samples 12. Because of the potential false-positive results in the visual interpretation of gels, it is essential to confirm the specificity of the amplicons by a second method.