Host genetic determinants have a major influence on an individual's risk of developing Type 1 diabetes mellitus (T1DM). At the same time, such environmental factors as foods and infectious agents are thought to play a role in the genesis of prediabetic autoimmunity or in the progression from persistent beta-cell autoimmunity to clinical diabetes (Yoon, 1990; See and Tilles, 1998). Immunity to one or more beta cell autoantigens, such as insulin, GAD65, or IA-2, may lead to destruction of beta cells and a loss of the capacity to produce insulin, ultimately resulting in clinical insulin-dependent diabetes mellitus. Postulated mechanisms by which infectious agents may trigger T1DM include:

1. direct cytolytic infection of beta cells, resulting in destruction of beta cells and loss of capacity to synthesize insulin;
2. a virus-induced immune response against infected beta cells, such as T-cell induced killing of virus-infected cells;
3. non-specific “innocent bystander” killing of beta cells through activation of non-specific immune mediators; and
4. induction of an autoimmune response to islet antigens by cross-reactivity with viral antigens (molecular mimicry) or disruption of normal immune tolerance mechanisms.

Several viruses have been proposed as infectious triggers of diabetes, but the enteroviruses (family Picornaviridae, genus Enterovirus) are the subject of the most intense scrutiny at present (Leinikki, 1998; Hyöty et al., 1998). Numerous studies have provided evidence for an association between enterovirus infection and prediabetic autoimmunity or clinical diabetes. Diabetes incidence has been epidemiologically linked to the incidence of enteroviral meningitis or enterovirus outbreaks (Karvonen et al., 1993). Serologic studies have shown that there is a correlation between enterovirus seroprevalence in patients with prediabetic autoimmunity or diabetes, compared to unaffected control individuals (Hiltunen et al., 1997; Helfand et al., 1995). Direct enterovirus detection in pancreas, blood, serum, or stool has suggested a temporal correlation between enterovirus infection and onset of diabetes (Yoon et al., 1979; Andreoletti et al., 1997; Clements et al., 1995).

Enteroviruses are among the most common of human viruses, infecting an estimated 50 million people annually in the United States and possibly a billion or more annually worldwide (Morens and Pallansch, 1995; Pallansch and Roos, 2001). Most infections are inapparent, but enteroviruses may cause a wide spectrum of acute disease, including mild upper respiratory illness (common cold), febrile rash (hand, foot, and mouth disease and herpangina), aseptic meningitis, pleurodynia, encephalitis, acute flaccid paralysis (paralytic poliomyelitis), and neonatal sepsis-like disease. Enterovirus infections result in 30,000 to 50,000 hospitalizations per year in the United States, with aseptic meningitis cases accounting for the vast majority of the hospitalizations (Pallansch and Roos, 2001). In addition to these acute illnesses, enteroviruses have also been associated with severe chronic diseases such as myocarditis (Martino et al., 1995; Kim et al., 2001), Type 1 diabetes mellitus (Leinikki, 1998; Rewers and Atkinson, 1995), and neuromuscular diseases (Dalakas, 1995). Enteroviruses are transmitted primarily by the fecal-oral route but respiratory transmission to close contacts may also be important. The incubation period between infection and onset of symptoms is usually 4–7 days. The intestinal mucosa or upper respiratory tract is the site of primary infection, with secondary spread to the central nervous system and other tissues. Viremia is usually short-lived, often waning before the onset of symptoms, except in very young children. Virus is excreted in the stool for up to 8 weeks (average 2–4 weeks) but maximal virus shedding occurs before the onset of symptoms. The maximum virus titer in stool is approximately 10⁴ infectious virus particles per gram.

Of the 89 recognized enterovirus serotypes, 64 are known to infect humans (Pallansch and Roos, 2001). In addition to the human enteroviruses, human pathogenic viruses are found in four other picornavirus genera: Rhinovirus (human rhinoviruses), Hepatovirus (human hepatitis A virus), Parechovirus (human parechoviruses 1 and 2, formerly echoviruses 22 and 23, respectively), and Kobuvirus (aichivirus, an agent of gastroenteritis). Most of the human enterovirus serotypes were discovered and described between 1947 and 1963 as a result of the application of cell culture and suckling mouse inoculation to the investigation of cases of infantile paralysis (paralytic poliomyelitis) and other central nervous system diseases (Committee on Enteroviruses, 1962; Panel for Picornaviruses, 1963). The human enteroviruses were originally classified on the basis of human disease (polioviruses), replication and pathogenesis in newborn mice (coxsackie A and B viruses), and growth in cell culture without causing disease in mice (echoviruses), but they have recently been reclassified, based largely on molecular properties, into four species, A through D (King et al., 2000). Sequences in various portions of the enterovirus coding-region correlate with species, but only capsid sequence correlates with serotype.

The neutralization test, long the gold standard for enterovirus typing, is generally reliable, but it is labor-intensive and time-consuming, and may fail to identify an isolate because of aggregation of virus particles or antigenic drift (the widely used standardized typing antisera were raised against prototype strains that were isolated 40 to 50 years ago [Lim and Benyesh-Melnick, 1960]). Anti-sera to all serotypes are not generally available and isolates that are not of a known human enterovirus serotype (new serotypes or serotypes that normally infect animals other than humans) would obviously also present difficulties in identification by antigenic means, as the neutralization method requires the use of serotype-specific reagents. In addition, neutralization requires virus isolation, which may require the use of multiple cell lines and adds to the time required to make an identification.

The application of PCR has improved the speed and accuracy of general enterovirus detection (Rotbart and Romero, 1995; Rotbart et al., 1997), and has found wide acceptance in the clinical diagnostic laboratory. Since the enterovirus serotype is rarely relevant to clinical case management, many clinical virology laboratories are bypassing virus isolation entirely, in favor of PCR detection of viral nucleic acid directly in clinical specimens such as cerebrospinal fluid, nasopharyngeal swabs, or tissue specimens (Rotbart and Romero, 1995). This approach uses genus-specific primers targeted to the 5′ non-translated region (see Figure 1-6), often coupled to probe-hybridization and detection of product in a microplate format (Rotbart and Romero, 1995). Specimens of choice for the direct detection of enteroviruses by RT-PCR are stool or rectal swab (stool is preferred because it contains a larger amount of fecal material and, hence, virus); oro- or nasopharyngeal specimens (throat swab, nasopharyngeal swab or aspirate, saliva); cerebrospinal fluid (if there is concomitant CNS disease); fresh-frozen or formalin-fixed tissue; and serum/plasma. Serum and plasma are generally only useful for RT-PCR in infants because viremia may still be present after onset of symptoms. If virus is detected only in a non-sterile site, such as stool or nasopharynx, a large number of patients are needed to establish the association between infection and disease.

FIGURE 1-6

Schematic representation of the enterovirus genome, indicating regions that have been targeted for development of PCR diagnostics. The genome is a positive-stranded, polyadenylated RNA of ∼7400 nucleotides, with a viral protein (3B/VPg) covalently (more...)

Despite the advantages of enterovirus detection by RT-PCR, challenges remain. In the case of chronic diseases, the virus may act indirectly (e.g., through immune-mediated pathology). The virus may be cleared well before disease onset or virus may be present in the patient but not in the diseased tissue. Even in acute illnesses, the titer is relatively low in all specimens. As a result, a conventional single-step RT-PCR amplification may not be sensitive enough for direct detection from the original clinical specimen. Designing a prospective study and collecting multiple specimens, at multiple time points throughout the duration of the study, may overcome some of these problems; however, the only way to solve the sensitivity problem is by increasing the sensitivity of the detection method. To address this issue, we have developed an enterovirus-specific semi-nested RT-PCR assay (5′ NTR RT-snPCR) that targets the conserved regions of the 5′ NTR (see Figure 1-6). Figure 1-7 shows the sensitivity of our standard, conventional RT-PCR (Yang et al., 1992) compared with that of the 5′ NTR RT-snPCR. Ten-fold serial dilutions of a virus isolate (10⁻¹ to 10⁻¹⁰) were prepared with uninfected cell extract as diluent. RNA was extracted using the QIAamp viral RNA mini-kit (Qiagen Inc., Valencia, CA) and reverse-transcribed using the antisense primer. PCR was performed using a single round of amplification (conventional PCR) or two rounds of amplification (semi-nested PCR). The second round of the semi-nested amplification used the same primers as the conventional PCR. Amplification products were visualized by polyacrylamide gel electrophoresis and staining with ethidium bromide. The RT-snPCR method (see Figure 1-7B) was approximately 10,000-fold more sensitive than the conventional RT-PCR (see Figure 1-7A). The 10⁻⁷ dilution corresponds to less than 20 infectious virus particles.

FIGURE 1-7

Sensitivity of pan-enterovirus RT-PCR methods. M-molecular weight marker. Virus dilutions are shown at the top of each panel. A. Titration of conventional two-primer RT-PCR. B. Titration of RT-semi-nested (three-primer) PCR.

Enterovirus infection elicits a serotype-specific immune response directed against epitopes on the surface of the viral capsid. Mucosal immunity is most important. Antibody alone fully protects from disease, probably by limiting virus spread from the gut, but antibody does not necessarily protect from infection. The virus-specific T-cell response, directed against epitopes on both the structural and non-structural proteins, is probably involved in virus clearance but it is not needed for protection. Antigenic sites are located in each of the three enterovirus structural proteins, VP1, VP2, and VP3 (Minor, 1990; Mateu, 1995), but the epitopes responsible for serotype specificity have not been identified. Since the picornavirus VP1 protein contains a number of immunodominant neutralization domains, we hypothesized that VP1 sequence should correspond with neutralization properties (serotype) (Oberste et al., 1999b). Due to the high frequency of recombination among picornaviruses (Kopecka et al., 1995; King, 1988; Santti et al., 1999), sequence information from non-capsid regions is of little value in characterizing new serotypes within known genera.

Practical criteria must be established before molecular sequence information can be applied routinely to picornavirus identification. A partial or complete VP1 nucleotide sequence identity of at least 75 percent (minimum 85 percent amino acid sequence identity) between a clinical enterovirus isolate and serotype prototype strain may be used to establish the serotype of the isolate (Oberste et al., 1999a,b, 2000). These criteria also appear to apply to comparisons among isolates of foot-and-mouth-disease virus (family Picornaviridae, genus Aphthovirus) (Vosloo et al., 1992), but a study directly comparable to the enterovirus studies has not yet been performed. A best-match nucleotide sequence identity of between 70 percent and 75 percent or a second-highest score of greater than 70 percent may provide a tentative identification, pending confirmation by other means, such as neutralization with monospecific antisera (Oberste et al., 2000) or more extensive sequencing. A best-match nucleotide sequence identity below 70 percent (less than 85 percent amino acid sequence identity) may indicate that the isolate represents an unknown serotype (Oberste et al., 2000, 2001). Sequencing of the complete capsid-coding region may be useful in confirming this result, but complete capsid sequences are available for less than half of the known enterovirus serotypes, limiting the utility of complete capsid sequence comparisons until more sequence becomes available. More extensive characterization, possibly including complete genome sequences, may be required for viruses that appear to represent previously unknown genera (Hyypia et al., 1992; Marvil et al., 1999; Niklasson et al., 1999; Yamashita et al., 1998).

Recognizing the technical difficulties and limitations inherent in the classic approach to enterovirus identification, we developed RT-PCR and sequencing primers that target the VP1 capsid gene and may be used to determine enterovirus serotype by sequencing of the amplicon and comparison to a database of the VP1 sequences of all enterovirus serotypes (Oberste et al., 1999a,b, 2000). These molecular detection and typing methods, when coupled with well-designed prospective studies, will be useful in addressing the potential causal relationship between enterovirus infection and development of prediabetic autoimmunity or progression from persistent autoimmunity to clinical diabetes.