Main

Although biliary atresia is the most common indication for liver transplantation during infancy, the cause of this serious disorder is unknown(1). During the first few weeks of clinically inapparent disease, liver histology is characterized by dense inflammatory infiltration with small numbers of polymorphonuclear leukocytes, lymphocytes, plasma cells, and eosinophils surrounding bile ducts(2). Later, mononuclear cells predominate. This presentation has prompted several investigators to seek a viral etiology, because the histology is consistent with a viral infection. Witzelben et al.(2) failed to find evidence of either CMV by viral culture or hepatitis B by immunofluorescence studies of fresh frozen sections of liver biopsy specimens from infants with the disorder. These authors recommended that an aggressive search for other viruses was warranted. A more recent study that used a sensitive PCR failed to find CMV in the majority of infants with biliary atresia(3). In addition, Balistreri et al.(4) reported negative serology for both hepatitis A and B in infants with biliary atresia.

In other studies, initial reports suggested that perinatal infection with reovirus 3 may be an initiating factor in both biliary atresia and idiopathic neonatal hepatitis(5). Indeed, infection in newborn mice with this virus resulted in periportal inflammation; however, bile duct epithelial damage was not observed(6). Other investigators failed to confirm a role for reovirus 3 in neonatal liver disease(7).

Recently, several lines of evidence have suggested an etiopathologic role for rotaviruses in biliary atresia. Rotaviruses are the most common cause of gastroenteritis in children and infants worldwide, with an estimated 140 million infected yearly and 1 million deaths(8). In addition, rotaviruses are the most common cause of nosocomially acquired gastroenteritis in children(9–11). Although GAR primarily infects mature enterocytes of the intestinal villi in humans, we reported the growth of simian and bovine GARs in a human liver cell line; however, these cells failed to support the growth of any human GAR(12). GAR (simian and human) produced extrahepatic biliary obstruction in orally inoculated newborn mice(13). Bile ducts from infected mice showed variable degrees of inflammation and fibrosis but without atresia of the extrahepatic biliary tract, characteristic of biliary atresia in human infants. In a recent report, GAR in liver samples from 14 infants with biliary atresia was not detected using PCR followed by liquid hybridization employing a probe for VP7, the outer capsid neutralization protein(14). However, 71% of the patients in the same study were positive for either gene 5 or 6 of GCR using a nested PCR or by liquid hybridization with radiolabeled probes to assay liver and bile duct samples.

The present study was undertaken to determine whether rotavirus A, B, or C could be detected in liver and bile duct remnants of biliary atresia patients from our geographic area. Detection tests, such as antigen immunoassay, immune electron microscopy, and PAGE are insensitive, generally requiring more than 107 virions/g of sample(15, 16). In clinical and subclinical infections, viral titers are well below this(17). Previous studies have shown PCR to be more sensitive for detection of a wide range of infectious agents, including rotavirus A(18–21). We have reported that detection of amplified DNA:biotinylated RNA hybrids using a quantitative EIA format (PCR-EIA) to be highly sensitive and specific for other pathogens(22–25). Therefore, we developed an individual RT-PCR-EIA for each GAR, GBR, and GCR to maximize our ability to detect these agents in liver samples from biliary atresia patients.

METHODS

Patient selection. Ten children with biliary atresia were selected consecutively. The diagnosis of biliary atresia was established by operative cholangiography, which demonstrated absence of a lumen in part or all of the extrahepatic biliary tract. All of the children had the“acquired” form of biliary atresia, without other anomalies. Liver biopsies were obtained either at the time of diagnosis (n = 6), at the time of orthotopic transplant (n = 3), or during an episode of ascending cholangitis (n = 1). Liver was obtained from all patients, and biliary remnants from five for a total of 15 samples. The “biliary remnants” in two patients were gallbladder tissue (from children in whom a gallbladder Kasai portoenterostomy procedure was performed), and in three patients they were the portal plates of scar tissue removed at the time of the Kasai procedure. The controls were children (n = 14) with a variety of liver diseases. Liver biopsies were obtained by percutaneous needle biopsy(n = 9) or intraoperatively at the time of orthotopic liver transplant (n = 5). Patient information is described in Table 1. None of the patients had a history of rotavirus or other gastroenteritides. The history of maternal infection during pregnancy is unknown. In all cases, liver samples were frozen when obtained by immersion in liquid nitrogen or dry ice and then stored at -70°C until analyzed. The study was approved by the Johns Hopkins Committee on Clinical Investigation, and informed consent was obtained from the parents.

Table 1 Clinical features of biliary atresia patients and liver disease controls
Full size table

Laboratory procedures. Tissue samples were thawed on ice, and a portion was minced cold in diethylpyrocarbonate water. Other aliquots were stored at -70°C. Total RNA was extracted using the guanidinium isothiocyanate-acid phenol method(26). Negative control buffers were included every third sample. Because rotaviral RNA is double-stranded, RNA was heated to 95°C for 5 min and then chilled. Reverse transcription was performed in a 20-μL final volume at 37°C for 1 h, boiling for 5 min, and snap-cooling on ice. RT reagents were RT buffer(50 mM Tris-HCl, pH 8.3, 50 mM KCl, 8 mM MgCl2), 1 mM RNasin, 5 μg of random hexamers, 100 U of Moloney murine leukemia virus RT, and 3 μL of boiled RNA. A second RNA aliquot, which was processed for β-actin mRNA, was not first preheated because β-actin mRNA is single-stranded. This aliquot was reverse transcribed in a separate RT-PCR. The β-actin sample served as a control for integrity of sample RNA.

Positive controls used for development of the RT-PCR-EIA were extracted MA-104 cells infected with rotavirus A (human strains ATCC VR-954, and Wa, ATCC VR-2018) (American Type Culture Collection, Rockville, MD), which were grown in our laboratory to moderate and high titers. Infected infant rat feces was the source of RNA for GBR (strain IDIR), and MA-104 cells were used to propagate a GCR porcine isolate (Cowden strain). The IDIR and Cowden strains were gifts from Drs. J. Eiden (Johns Hopkins University) and L. Saif (Ohio State University), respectively. In addition, a human strain of GCR isolated from a patient with diarrhea was also used as a positive control for the PCR.

PCR primers were chosen in areas of the genome which are conserved between animal species, including humans. Primers for DNA amplification are listed in the order of sense and antisense: GAR, gene 6, bp 1-20: GGCTTTTAAACGAAGTCTTC; bp 234-259: TCAACATAATTAGCGTCTAAGTTCA(27); GBR, gene 3, bp 898-920: ATCATGGAGGCCGGCCACAGACT; bp 1150-1175: CTAGGAAGT-ATCTATCTGTGCAAAGCC(28); GCR, gene 6, bp 994-1016: CTCGATGTCACTACAGAATCAG; bp 1327-1349: AGCCACATAGTTCACATTTTCATCC(29); β-actin: exon-1, bp 2-23: TGGATGATGATATCGCCGCG; exon-2, bp 467-489: CATCTTCTCGCGGTTGGCCTTG(30). Three microliters of cDNA were added to 47 μL of PCR buffer (10 mM Tris-HCl, pH 8.3, 50 mM KCl, 1.5 mM MgCl2), 0.2 mM of dNTPs, 0.5 μM final concentration of primers, and 1.25 U of Taq DNA polymerase (Amplitaq, Perkin-Elmer, Norwalk, CT) in a 50-μL final volume. The thermofile was an initial 5 cycles of 94°C, 55°C and 72°C for 1 min each, followed by 25 more cycles of 30 s for each temperature. Ten microliters of product were electrophoresed through 7.5% polyacrylamide gel(PAGE) followed by staining with silver nitrate(31).

Biotinylated RNA probes for each organism were transcribed from amplified DNA using primers completely nested to the external primers listed above. A standard transcription reaction described previously was performed on 500 ng of DNA using T7 RNA polymerase and biotin-11-UTP (Enzo, Farmingdale, NY)(22). Nested primers are listed as sense and antisense, and the sense primer contains the T7 RNA polymerase promoter, which is underlined: GAR: 48-60 bp TTAATACGACTCACTATAGGGT TGTCAAAGAACTCTTAAAG; 168-187 bp GGCAAATTACCCATTCCTCC; GBR: 929-948 bp, T7 promoter CAGTTCACGCTTCAACC; 1102-1122 bp, CCGCCCTTATTGGTTATTCCC; GCR: 1084-1118 bp, T7 promoter GGAAACAGTATTTCAGCCA; 1295-1316 bp, TCATCCACGTCATGCGCATTTG.

Solution hybridization of the amplified DNA with the biotinylated RNA probe was detected by an EIA as described previously(22). Briefly, biotinylated hybrids were captured in anti-biotin-coated microtiter plate wells and reacted to a MAb to RNA:DNA hybrids that was conjugated to alkaline phosphatase. Cleavage of the phosphorylated substrate yielded a fluorescent signal (FU), which was measured in a fluorometer. Samples were considered positive if the mean FU value of duplicate wells was greater than the mean of negative reagent controls plus 3 SD. This cutoff value derived from the reagent controls is subtracted from the sample values so that a value>0 FU would be considered positive.

Probes were tested with amplified DNA from approximately 105-106 plaque-forming units or tissue culture ID50 of GAR or GCR in homologous and heterologous hybridizations to determine specificity. Several dilutions of GBR were also tested. Nested primers were used to amplify each rotavirus stock strain in homologous and heterologous amplification reactions. Ethidium bromide-stained 2.0% agarose gel was used to analyze the results. Sensitivity of RT-PCR-EIA was determined on 10-fold dilutions of tissue-cultured stock strains of GAR and GCR that were each extracted and amplified.

Solid phase hybridization and detection was carried out for GCR only after Southern blot(32) of amplified samples using enhanced chemiluminescence (ECL, Amersham Corp., Arlington Heights, IL). An internal oligo probe (the nested antisense primer) was 3′-labeled with fluorescein-11-dUTP and hybridized according to the manufacturer's directions.

Several types of laboratory controls were included in this study. We monitored PCR contamination using both extraction and PCR-negative buffer and negative tissue cultured cells. If PCR products were detected concurrently in negative controls and test samples, the test was considered invalid. In this case, another frozen test sample from the same patient would be processed for reanalysis. To control for quality of tissue RNA and cDNA, a β-actin PCR was performed on each human sample, and the PCR was examined by silver-stained PAGE. Positive rotavirus samples were also included throughout the analysis, which would give a band intensity equivalent to 1 ng of DNA and a moderate EIA value of approximately 500 FU. The GCR PCR-positive control also served as the Southern blot-hybridization control.

RESULTS

Initial studies comparing sensitivity of tissue culture with RT-PCR-EIA on serial 10-fold dilutions of GAR and GCR rotavirus indicated that PCR was consistently 100-fold more sensitive than tissue culture. In addition, EIA fluorescence values were directly related to viral load and were linear over most of the 10-fold dilutions (Fig. 1). The cutoff was determined from the mean of 3 negative tissue culture samples plus 3 SD. Because GBR is not cultivable, we could not employ the same method of comparison. However, based on past RT-PCR studies by this laboratory on dilutions of purified GBR RNA using acrylamide gel detection, the limit of detection was 8 attamoles or 400 genome equivalents(21).

Figure 1
figure 1

Sensitivity of PCR-EIA for GAR and GCR. Each point represents a 10-fold dilution of the stock viruses used for positive viral controls (RNA extraction followed by individual RT-PCR). The positive cutoffs are indicated by the arrowheads. AMPLICON refers to the type of amplified DNA.

Full size image

Specificity was determined in two ways. First, each nested primer set reacted only with its homologous target (Fig. 2). Second, a target DNA, which was generated using external primers, reacted only with its homologous probe and not with the heterologous probes. Input target was approximately 103-102.5 plaque-forming units or tissue culture ID50 for GAR and GCR, respectively, and the cutoff was determined as mentioned previously.

Figure 2
figure 2

Specificity of nested PCR primers for DNA targets. Ethidium bromide-stained agarose gel contains phi X-HaeIII markers in lanes 1 and 14; lanes 2-4 contain GAR, GBR, and GCR cDNA amplified with GAR primers; lanes 6-8 contain the same sequence amplified with GBR primers; and lanes 10-12 with a repeat of the same sequence amplified with GCR primers. Lanes 5, 9, and 13 are negative controls for each PCR.

Full size image

β-Actin mRNA was present in all of the liver samples from both biliary atresia patients and control patients by RT-PCR as seen by detection of a 342-bp (as calculated from cDNA) amplified DNA band by gel. β-Actin mRNA was also detected in the biliary remants from biliary atresia patients 3 and 4 and gallbladder from biliary atresia patient 10 but not from the biliary remnants of biliary atresia patient 1 nor from the gallbladder of biliary atresia patient 5. However, despite the fact that β-actin mRNA was present in all of the liver samples, demonstrating that RNA was present in these samples, RT-PCR was negative for all of the rotavirus groups on all of the samples from both biliary atresia patients and control patients by both gel (Fig. 3A) and EIA detection (FU < 0). In comparison, EIA values for positive viral controls for GAR, GBR, and GCR were 454 ± 8, 395 ± 57, and 685 ± 10 FU, respectively. In addition to a PCR-EIA analysis on test samples for GCR, we also performed a Southern blot of the same amplified samples. Hybridization with an internal oligonucleotide was negative for the biliary atresia patients and liver disease control patients but strongly positive for the GCR control (Fig. 3B).

Figure 3
figure 3

Gel analysis and Southern blot-hybridization on biliary atresia samples amplified for GCR. (A) Silver-stained PAGE contains biliary atresia patient samples in lanes 3-9, 11-14, and 16-18; the 356-bp GCR control in lane 1; lanes 2 and 15 with buffer controls; and lanes 10 and 19 with phi X-HaeIII markers. (The gel contains the entire group of liver and biliary samples from the biliary atresia group except for the gallbladder sample from patient 5, which was inadvertently omitted.)(B) Southern blot hybridization is shown of the gel in A with the GCR positive control in lane 1.

Full size image

DISCUSSION

RT-PCR-EIA was developed for GAR, GBR, and GCR and found to be highly sensitive and specific. In previous studies on other viral and bacterial pathogens, we have shown EIA detection of amplified DNA to be consistently 100-fold more sensitive than silver-stained acrylamide gel detection and equivalent to radioactive hybridization of Southern blot(19, 22). In addition, the sensitivity and specificity of the biotinylated probe in a solution hybridization-EIA was equivalent to either a radiolabeled or chemiluminescent probe hybridization of a Southern blot. Moreover, the efficiency of RNA extraction and RT-PCR-EIA was monitored by inclusion of a rotavirus sample which would reproducibly give a moderately positive signal. Finally, in past studies on Chlamydia trachomatis using animal models(33) and in transmission studies between partners(34), and in human volunteer studies for influenzae A and parainfluenzae 3 vaccine efficacy(24, 25), we have demonstrated that it was possible to detect amplified DNA when traditional tests were negative. Therefore, we are assured that PCR-EIA is highly sensitive and specific.

The fact that our series of samples was negative for GCR by RT-PCR as well as by Southern blot contradicts the report by Riepenhoff-Talty et al.(14) of 71% positivity for this virus. Although both genes 5 and 6 of GCR were tested in the Riepenhoff-Talty study, it was reported that at least one gene was positive by either liquid hybridization with a radiolabeled probe or by nested PCR. If GCR were etiologic for the cases of biliary atresia studied by these investigators, and if both PCRs were equally efficient, one would assume that both genes should have been detected. Because Riepenhoff-Talty et al. did not identify their primers, we were unable to compare their primer sites with those chosen for our study. The primer sites that we chose for group C PCR were in areas that are known to be conserved between human, bovine, and porcine strains(35, 36).

In our experience, nested PCR is prone to contamination. Furthermore, in studies where we have reported much higher positivity rates than expected, or compared with “gold standard” detection methods, DNA sequencing of the PCR products from a gene segment with known variability has helped us to validate our findings in terms of discounting PCR contamination(34). An alternate possibility to explain the discrepancy between the Riepenhoff-Talty results and our own is that the scientific methodology used by both groups was equally accurate, but that GCR prevalence was defined by differing geographies. At least one group of investigators has noted time-space clustering of biliary atresia cases(37), suggesting that geographic and seasonal factors may play a role. In addition, it should be noted that there has been considerable controversy surrounding a putative association between biliary atresia and a variety of viral agents, including most notably reovirus-3, but also hepatitis B, CMV, and respiratory syncytial virus(38–42)

Our study could be criticized because four of the hepatobiliary samples were obtained at the time of transplant (13, 17, and 19 mo and 10 y of age), long after the initial putative viral insult. However, some investigators(43, 44) have proposed that, in studies of possible viral etiologies of biliary atresia, hepatobiliary samples should be studied for evidence of residual viral nucleic acids. The rationale for this approach is that progression of hepatobiliary pathology after portoenterostomy could be attributed to persistence of a virus and/or the immune response to it. Another criticism of our study is that β-actin could not be detected in the biliary remnants of patients 1 or 5, so consideration could be given to discounting the rotavirus PCR studies in those samples. On the other hand, it was possible to detect β-actin in the liver samples of those two patients. Because the biliary samples were harvested and stored the same way as the liver samples, neither the failure to detect β-actin nor the significance of this failure is easily understood.

A third criticism of our study is that the number of biliary remnants was small and that liver samples might not be of value in studying a disease that primarily affects biliary epithelium. On the other hand, we had previously demonstrated that it is possible to cultivate GARs in human-derived liver(HepG2) cells(12). Furthermore, Uhnoo et al.(45) showed that rotavirus could be detected in the livers of mice inoculated orally with rhesus rotavirus. Taken together, these findings suggest that liver samples as well as biliary samples might be useful in investigating the role of rotaviruses in hepatobiliary diseases such as biliary atresia.

In conclusion, our data, like those of Riepenhoff-Talty et al.(14), fail to support a role for GAR in the pathogenesis of biliary atresia. In addition, our data fail to support a role for either GBR or GCR as common etiopathologic agents in biliary atresia in human infants. Our findings reinforce the concept that large scale multicenter studies will be necessary before a firm link can be established between any putative viral agent and biliary atresia. In addition, the discrepancy between Riepenhoff-Talty et al.'s results and our own is entirely consistent with the “two-hit” model of biliary atresia proposed by Schreiber et al.(46). In this model, a variety of environmental factors such as diverse viral infections would act on a genetically susceptible host(47) to stimulate production of neoantigens by biliary epithelium(48), which would be followed by a cellular immune fibrosclerosing injury to the bile ducts. This model suggests that future investigations of the etiology of biliary atresia should include the host immune response as well as a search for etiologic agents.