History of the Discovery of Hepatitis A Virus

Abstract

Disease outbreaks resembling hepatitis A have been known since antiquity. However, it was not until World War II when two forms of viral hepatitis were clearly differentiated. After the discovery of Australia antigen and its association with hepatitis B, similar methodologies were used to find the hepatitis A virus. The virus was ultimately identified when investigators changed the focus of their search from serum to feces and applied appropriate technology.

Diseases resembling hepatitis A, both in individuals and in outbreaks involving groups, were reported in China as early as 5000 years ago. Hippocrates noted a disease he called benign epidemic jaundice in “De Morbis Internis” that certainly resembled hepatitis A. More accurate descriptions of hepatitis A began appearing in the 17th century often associated with military campaigns. The first outbreak recorded in the United States was in 1812 in Norfolk, VA, and the disease was common among the Union troops during the Civil War with more than 40,000 cases reported. The association of hepatitis A with war led to 19th-century terms such as “kriegsikterus” or “jaunisse des camps.” Hepatitis A continued to afflict troops on both sides during World War I and in the Second World War when there were estimates of 16 million cases of hepatitis among combatants and civilians (Sherlock 1984; Gust and Feinstone 1988; Fonseca 2010).

DIFFERENTIATION OF TWO FORMS OF VIRAL HEPATITIS

Viral hepatitis was a major problem for both the Allies and the Axis during World War II. Early in the war, an outbreak of hepatitis related to yellow fever vaccine, stabilized with human serum involving 49,233 clinically apparent cases (Seeff et al. 1987), prompted a major hepatitis research effort. As the records on the vaccinees were very good, the incubation period was defined accurately as between 60 and 154 days. This outbreak as well as the more widespread problem of infectious hepatitis caused both the British and the Americans to initiate studies of viral hepatitis. It became clear that one form of hepatitis spread quickly among troops and by 1943 hepatitis had become a real hindrance of the war effort in North Africa and the Mediterranean. U.S. Army epidemiologic studies showed that epidemic or infectious hepatitis had a much shorter incubation period than serum hepatitis—18 to 25 days compared with a mean of 90 days for hepatitis following the yellow fever vaccine (Havens 1968). Second, they found that some soldiers that had developed hepatitis after immunization with the yellow fever vaccine could still develop a second bout of infectious hepatitis. They also noted that officers were more likely to develop infectious hepatitis than enlisted men and that the disease often followed outbreaks of diarrheal disease, suggesting a fecal–oral mechanism of spread.

These epidemiologic studies were accompanied by experimental infections of humans performed by the Americans and British as well as the Germans (Voegt 1942; Neefe et al. 1944, 1946; MacCallum et al. 1951). Transmission to volunteers was not always successful. Preexisting immunity in a high proportion of volunteers, as well as the inability to know whether any given inoculum was actually infectious, caused the results of these studies to be difficult to interpret. Nevertheless, these studies pointed to two distinct diseases, one with a short incubation period transmitted by the fecal–oral route and the other transmitted by serum with a relatively long incubation period. The investigators also showed a lack of cross immunity between the two types of infections and that some of the physical characteristics of the causative agents were distinct. These two forms of hepatitis became known generally as “infectious hepatitis” and “serum hepatitis.” It was not until the early 1950s that the first description of these diseases as “type A” and “type B” viral hepatitis appeared in a report from an expert committee of the World Health Organization (MacCallum 1953).

Following the war, virology entered its golden era with the advent of tissue culture made possible by the development of defined media and antibiotics (Robbins and Enders 1950). Many viral agents were identified during this period but neither hepatitis agent was successfully propagated in cell culture. Saul Krugman, Robert Ward, and Joan Giles conducted studies between 1956 and the early 1970s at the Willowbrook State School for intellectually handicapped children on Staten Island, New York. These studies involved deliberate experimental infections of some children and were controversial even during the era in which they were conducted. They were justified by the investigators based on their belief that every child who entered the facility developed hepatitis soon after arrival. In the course of these studies, the investigators identified a child with two distinct bouts of hepatitis from whom serum samples had been collected during the acute and recovery stages of each bout. Additional studies with these two samples, termed MS-1 for short incubation (infectious) hepatitis and MS-2 for long incubation (serum) hepatitis, further defined two distinct hepatitis viruses and, most importantly, established a well-characterized, known infectious inoculum for each (Krugman et al. 1967). Krugman also determined the general period of infectivity for persons infected with these viruses, showed that serum hepatitis often became chronic, and that standard pooled human immune serum globulin was protective against infectious hepatitis (Krugman and Giles 1970, 1972). This latter finding confirmed earlier studies done by the U.S. Army during World War II (Gellis et al. 1945).

Because of the continuing threat to military operations posed by infectious hepatitis, the U.S. Army continued to sponsor research after the war under the aegis of the Armed Forces Epidemiological Board. To obtain larger quantities of the infectious hepatitis inoculum, the army contracted for studies in which the Krugman MS-1 inoculum was amplified by infecting volunteer inmates at the federal prison in Joliet, IL (Boggs et al. 1970; Melnick and Boggs 1972). Volunteers were inoculated with the MS-1 strain of hepatitis A, and the investigators collected stools and serum samples throughout the course of the illness, from the incubation period to acute disease, and then during convalescence. In addition, MS-1 infection was passed from one volunteer to another.

EARLY ATTEMPTS TO ISOLATE AN INFECTIOUS HEPATITIS A AGENT

In early quests to isolate an infectious agent, classic virologic methods, including inoculations of laboratory animals and embryonated chicken eggs with hepatitis-related material, did not yield positive results. With the development of cell culture as a routine laboratory technology, scientists could regularly propagate mammalian cells in vitro and identify many viruses known to be associated with human diseases (Robbins and Enders 1950).

Various cell culture methods were applied to the identification of the causative agent of hepatitis A (MacCallum 1972), but the most extensively reported study was performed by Wilton Rightsel and his colleagues at the Parke-Davis Company using a cell line termed Detroit-6, which was derived from human bone marrow (Rightsel et al. 1956, 1961). These investigators identified a cytopathic effect (CPE) in cells inoculated with various specimens thought to contain infectious virus. This CPE could be neutralized by convalescent serum from patients with hepatitis A–like illnesses. Although the neutralization studies were presumptive evidence for specificity, further studies were inconsistent for both the CPE and neutralization. The Detroit-6 cell culture studies were eventually abandoned at Parke-Davis, but Ruth Cole, a member of the Parke-Davis team, returned to Melbourne, Australia, where she continued studies with Detroit-6 cells. Working at Fairfield Hospital, where almost every patient with hepatitis in the region was admitted, Cole and colleagues performed a number of studies using various clinical inocula (Cole 1965; Ferris and Cole 1966). Again, CPE was observed but subsequent independent studies, done with coded samples including multiple control samples, failed to show that the observed CPE was specific for hepatitis A (Cross and Marmion 1966). A group led by Joseph Melnick also studied infectious hepatitis in the Detroit-6 system using an MS-1 inoculum from the Joliet studies, and concluded that the observed CPE was caused by a contaminant, most likely a parvovirus (Mirkovic et al. 1971).

ANIMAL STUDIES

Early studies suggested that the agent responsible for hepatitis A was unable to infect any common small laboratory animal. In 1967, however, Friedrich Deinhardt working at the Presbyterian-St. Lukes Hospital in Chicago, IL, reported that a human hepatitis virus could be passaged in marmoset monkeys (Saguinus sp., New World nonhuman primates now generally considered to be tamarins) and that infected animals developed evidence of liver disease (Deinhardt et al. 1967). One hepatitis-inducing inoculum, GB, came from a surgeon who had developed hepatitis without a known exposure. Deinhardt did extensive work to characterize the GB agent, but he could not specifically associate it with hepatitis A. Wade Parks and Joseph Melnick, at Baylor College of Medicine in Houston, TX, determined that the GB agent was likely a marmoset virus, but they also were unable to prove that these primates could be infected with the MS-1 virus (Parks and Melnick 1969; Parks et al. 1969). The story of GB resurfaced in 1995 when two viruses, GBV-A and GBV-B, were identified by molecular cloning in animals infected with the eleventh animal passage of the GB agent (Simons et al. 1995). Neither of these agents infect humans. A related virus, GBV-C, has been identified in humans, but it is of very low or nonexistent pathogenicity. This group of viruses are classified within the Flaviviridae family, distantly related to classic flaviviruses like dengue virus and to hepatitis C virus (HCV) (Stapleton 2003).

SEROLOGIC STUDIES

In 1965, while looking for serum protein polymorphisms that would distinguish population groups from around the world, Baruch Blumberg at the Institute for Cancer Research in Philadelphia, PA, and Harvey Alter and Sam Visnich at the National Institutes of Health (NIH) in Bethesda, MD, observed a precipitin band in an immunodiffusion gel reaction between serum from a multiply transfused hemophiliac (the potential source of antibody) and an Australian aboriginal man (Blumberg et al. 1965). They did not immediately understand the nature of the antigen, but within a few years it became clear that it was associated with hepatitis and then specifically with hepatitis B (Prince 1968). However, the idea that one could use a serologic approach to identify previously unknown and difficult-to-culture viral agents had been shown from these studies.

Similar antibody-based technologies were thus applied to search for a virus or a viral antigen associated with hepatitis A with variable results. The best characterized of these was the Milan antigen described by Salvatore Del Prete and colleagues (Del Prete et al. 1972a,b) at the University of Milan in Italy, who used an approach similar to Blumberg’s. Characterization of the antigen revealed it to have a density of only 1.06 gm/cm³ and electron microscopy (EM) studies revealed disk-like structures of varying size up to 50 nm, suggesting that Milan antigen was likely an abnormal serum lipoprotein often associated with liver disease (Fig. 1) (Taylor et al. 1972).

Figure 1.

Milan antigen showing disc-like structures associated with serum lipoproteins (Taylor et al. 1972).

The serology-based studies generally followed the hepatitis B work, looking for an antigen in the serum of patients with acute hepatitis A. In contrast, Geoff Cross, Alan Ferris, and Ian Gust at the Fairfield Hospital in Melbourne understood that feces were the best place to look for a virus responsible for hepatitis A, and they thus began a search for the virus in feces. Using acute stool samples as a source of antigen, they reacted it with sera from multiply transfused hemophiliac patients in a manner similar to the original studies that identified Australia antigen. In retrospect, although these investigators did understand where the virus was most likely to be found, it made little sense to use hemophiliac sera as the source of antibody. Hemophiliacs were not at increased risk for infectious, type A hepatitis. They had access to convalescent sera from hundreds of patients admitted with clinical hepatitis A that were likely to contain specific antibody. Nevertheless, they did identify single precipitin bands in an agar gel diffusion assay with two of 37 sera, and four of six fecal specimens. They then used this technique to track the antigen in sucrose density gradients spun in an ultracentrifuge. The partially purified antigen was then used to raise antisera in rabbits. After absorption, the rabbit sera also produced precipitin bands when reacted with stools from hepatitis patients. In a large study, a single precipitin band was identified in 90 of 220 patients with acute hepatitis, compared with five of 158 samples collected from patients with other diseases. The antigen was typically present in the first sample collected after hospitalization and became undetectable an average of 18 days after admission (Ferris et al. 1970; Gust et al. 1970, 1971).

Immune EM (IEM) performed on clarified fecal extracts revealed two different particles that appeared to be associated with small amounts of antibody (Fig. 2). One of these particles was 15 to 25 nm in diameter, seemed to have a virus-like morphology, and appeared to be present as both full and empty particles typical for negative-stained EM images of small nonenveloped viruses. The investigators thought these particles could resemble 22 nm subviral HBsAg particles associated with hepatitis B. The second, less common particle was 40 to 45 nm in diameter and was considered to be similar to the hepatitis B virion, otherwise known as the Dane particle (Dane et al. 1970). Antibody to fecal antigen was thought to react with true Dane particles but not with HBsAg, and antibody to Dane particle-enriched specimens seemed to aggregate the large particles observed in fecal antigen. Based on these results, Geoff Cross believed that there were two types of particles associated with hepatitis A, much like the various morphologic forms associated with HBV.

Figure 2.

Fecal antigen with both large particles of about 45 nm and small 20 nm particles. (Electron micrograph kindly supplied by Ian Gust.)

These studies show the complexity of searching for a specific virus or viral antigen in human stools by EM. Multiple particulate forms, many resembling viruses, are seen in every stool sample. Some of these particles may be breakdown products derived from food, but some probably are actual viruses or bacteriophages (Fig. 3). For this reason, any particle seen in stool must be critically evaluated for specificity before it can be associated with a specific infection. In the case of IEM studies, very specific antibody or well-characterized paired sera from patients with acute infections need to be used. Because IEM is somewhat subjective, samples should also be evaluated by a blinded observer.

Figure 3.

Electron microscopic images of particles in human feces. (A) Small 22–25 nm particles without antibody commonly observed in stool mixed with immune serum globulin. We generally called these Clarke particles, as they are similar to 22 nm particles described in 1973 by Clarke and colleagues that were considered to be parvoviruses (Paver et al. 1973). The particles could be bacteriophages. (B) Honeycomb structure that we often saw in tools. (C) Particles observed in stools from a patient with hepatitis on Palau Island in Micronesia. This group of particles was observed in a partially purified specimen not reacted with antibody. (D) The Palau particles following addition of immune globulin and showing very heavy coating with antibody. (Photo from the author’s private collection.)

IDENTIFYING HEPATITIS A VIRUS: A PERSONAL ACCOUNT

I came to the NIH in Bethesda, MD, in 1971 as a Research Associate in the United States Public Health Service (PHS). This was the height of the war in Vietnam and virtually every male doctor completing his internship was drafted. One alternative to military service was to join the PHS and work at one of the agencies for which that organization was responsible, including the NIH. This was obviously a very popular option and PHS officers of that era became known as the “yellow berets.” Acceptance to NIH was highly competitive and, for that reason, I never understood how I was selected. Applicants came to NIH for interviews in several laboratories and there was sort of a matching program for the available slots. I assume that I was selected because I had a long-standing interest in laboratory research and had written two papers from work I had done in medical school. I was absolutely thrilled when I was selected by the Laboratory of Infectious Diseases (LID) headed by Robert Chanock, one of the country’s leading virologists. As the two other new associates were pediatricians, they were assigned to projects dealing with infections of childhood, specifically respiratory syncytial virus and viral diarrhea. I was training as an internist and was assigned to the viral hepatitis group headed by Robert Purcell. As my previous research involved cellular immunity to viral infections, I was originally tasked to look at the importance of that arm of the immune system in hepatitis B infections. At that time, T-cell function was barely understood and I was somewhat lost, heading down many dead ends and using inadequate technology. I floundered in that project for more than a year.

By this time, the hepatitis B virus (HBV) envelope, HBsAg, had been identified, the Dane particle was shown to represent the virion, and the genome was known to be DNA. Bob Purcell and John Gerin had applied ultracentrifugation techniques to purify large quantities of HBsAg for use in high-sensitivity immunoassays for both the antigen and the antibody. Using the radioimmunoprecipitation technique (RIP), essentially every person who had been infected by HBV, become a carrier, or cleared the infection could be identified (Lander et al. 1971, 1972). This purified HBsAg also served as the basis for an experimental vaccine (Purcell and Gerin 1975). Using these assays, Bob was able to define much of the epidemiology of hepatitis B. The laboratory was very busy testing samples coming in from all over the world. High stacks of microtiter plates were always on the benches of Doris Wong, Bob’s nearly career-long master technician, and Jose Valdesuso, who could do more work than can be imagined in the shortest possible time.

All this work was going on in a tiny “containment” laboratory in the sub-basement of Building 7 on the NIH campus because there was some fear of this virus among others working in the building. Bob had a tiny, sealed cubicle inside the small laboratory where he used large amounts of radioactive iodine-125 to label the purified HBsAg that John Gerin produced in an off-campus laboratory. The sub-basement was a very tight space, but it was also fun with Doris, Jose, Bob, and me. However, I felt like I was beating my head on the wall working on cellular immunology, and when Bob asked me to take a crack at finding the virus responsible for hepatitis A in stools, I was very grateful to get a new project.

Albert Kapikian, who was head of the Epidemiology Section in LID, had just returned from a sabbatical with June Almeida in the United Kingdom. June had developed IEM as a technique for identifying viruses (Almeida and Waterson 1969). Al was interested in discovering novel, hard-to-grow viruses associated with acute respiratory infections. In fact, Al did identify the 692 strains of coronavirus in a harvest from a tracheal organ culture that had been inoculated with washings from an adult with an acute upper respiratory illness (Kapikian et al. 1973). Just next to Al’s laboratory was the group working on viral diarrhea. Periodically they would bring Al a sample that they thought might contain a virus and eventually, using IEM, they identified the Norwalk virus in stools from patients with acute diarrheal illness (Kapikian et al. 1972). These were beautiful studies, and Bob Purcell immediately saw the potential for IEM in the search for the hepatitis A virus (HAV) that he knew should be in stool in large quantities. At about this time, we also acquired new laboratory space where we had laminar flow hoods, a contained ultracentrifuge, and other equipment that was not available in the sub-basement laboratory. I believe Bob would have started the hepatitis A work sooner if we had a laboratory where we could safely work with this highly contagious virus. As I was working on my immunology project and another research associate, Jon Gold, had recently joined the laboratory and was working on hepatitis B polymerase, Bob asked both of us to share the hepatitis A project and keep our other work going as well. After the first stool extraction, Jon decided he was not interested in hepatitis A and I took it on full time.

The idea was to learn IEM from Al and work with him to try to find the virus in various samples that were available to us. We had one golden set of samples from the volunteer studies that the U.S. Army had sponsored at the Joliet Prison in Illinois (Boggs et al. 1970; Melnick and Boggs 1972). These volunteers had been infected with the MS-1 inoculum from the Krugman studies at the Willowbrook State School. The object of the Joliet studies was to produce large quantities of infectious material that could be used by both army and non-army investigators, and both stools and sera had been collected from the volunteers. One sample, termed K30 (collected from volunteer K on the 30th day after inoculation, while jaundiced), had been used to pass hepatitis to other volunteers and therefore was known to contain infectious virus. Al and I used standard immune globulin as a screening antibody, and particles that appeared aggregated by antibody were then checked for specificity using paired sera from the volunteers that was available in only small quantities. We looked at the K30 stool multiple times under different conditions, but never saw any particles that we could identify as specific for hepatitis A.

As we did not find the virus in the Joliet MS-1 stool, we started to examine stools from an outbreak of hepatitis on Palau Island in Micronesia in 1967. We quickly identified large quantities of an unusual particle in the stool of one of the patients studied in that outbreak (Fig. 3C,D). The particle had a core and what appeared to be an envelope and they were not only aggregated by the immune globulin, they were often so heavily covered with antibody that they looked like snowballs (Fig. 3D). In addition, the size of the cores and therefore the entire particle varied tremendously. We spent months studying this particle, and although we concluded that it was not the hepatitis A agent, we did have enough data to associate it with the outbreak on Palau. We even had a final draft of a manuscript with all our data. I was rather insecure about this particle, and we were not even sure that it was a virus with its very unusual and variable morphology. In addition, an enveloped virus did not fit well with what was known then about the virus responsible for hepatitis A. Even though we did not claim in the manuscript that the particle was the hepatitis A agent, I became very worried that many investigators in the field would think that we really did believe it was the virus. I spent some sleepless nights worrying about this and determined to go back to the Joliet samples and go through them in a systematic fashion. This of course meant making a lot of stool extracts—not the most desirable task and one that was not at all popular with the entire staff of Building 7. Working in laminar flow hoods, the stool samples were extracted, clarified, and filtered. Everything used for that work was then carefully bagged and put in a large autoclave to destroy the infectivity. The odors produced in that process pervaded the entire building, and everyone knew when it was stool extraction day.

Once we started this more systematic approach to the Joliet stools, the virus appeared in one of the first samples that I tested, a sample termed F33. Having spent about 1 year, and Al many years, looking at stool filtrates, we knew the common structures observable by EM like they were old friends. Among the many particles in stool, we typically saw several virus-like particles that might be bacteriophage, structures that were probably bacterial filaments, and honeycomb-like material for which I have no explanation (Fig. 3A,B). So, when I first saw the hepatitis A particle, it immediately popped out as different. It almost glowed as a result of the antibody covering the particles (Fig. 4). On that day, Al was actually out because of illness and I have always had regrets about that. Al had put in a huge effort on this project, and he should have been there on that day. However, with no evidence other than the visualization of these antibody-covered particles, I was just a little reluctant to tell Bob before doing more experiments. Bob always maintained healthy scientific skepticism and I might have had my enthusiasm quickly dampened. But I did run upstairs to tell Bob that I was nearly certain I had seen the HAV particle. When Bob saw it, I believe he was as excited as I was, and we immediately set out plans for proving it was indeed the virus responsible for hepatitis A. Al returned to the laboratory soon after that and we went to work (Fig. 5).

Figure 4.

Large immune aggregate of hepatitis A virus (HAV) reacted with immune globulin. Aggregates like this is the way we originally observed the virus. (Photo from the author’s private collection.)

Figure 5.

Robert Purcell, Al Kapikian, and the author, seated at the Siemens 1A electron microscope that Al used to identify coronavirus 692, the Norwalk gastroenteritis virus, and HAV. Al loved that scope and never got used to a new one he was eventually forced to use. (Photo kindly supplied by the National Institutes of Health.)

On a very personal note, the months that I spent with Al Kapikian in the basement of Building 7 on the NIH campus (recently razed), staring at the glowing phosphorescent screen of that antique Siemens 1-A EM (now on permanent display in the lobby of Building 50 on the NIH campus) were some of the best in my career. Al taught me so much about science, of course. But we had wonderful conversations about our mutual love for baseball, of which Al had great historical knowledge. We talked about politics, as this was the Watergate era. Al taught me a lot about opera and I am to this day a regular operagoer. Al Kapikian was a wonderful, kind, and generous man. I, like all his many friends, miss him tremendously.

Bob, Al, and I laid out a series of experiments to prove the specificity of the particle we had observed. We adopted two general approaches to show the specificity of this particle. One was to identify the particle in stool samples from other patients with acute hepatitis A. The second was to show a specific antibody response to the particle we had identified in the F33 stool. As IEM is really a subjective process, Al had tried to standardize it as much as possible. The sensitivity of IEM is somewhere on the order of 10⁶ particles/mL, and we understood that perhaps not all stools, even from acutely ill patients, would have sufficient numbers of particles to find the virus with a reasonable effort. Therefore, Al’s system was to look at five good-quality EM grid squares before we declared a sample negative. Of course, “negative” only meant that we did not find any particles in the five examined squares. We also always did these evaluations under code. Every test was done with an adequate number of controls to try to keep bias out of the system. Doris Wong coded the samples after I had prepared the EM grids, so that neither Al nor I knew what sample we were looking at. We found the virus-like particles aggregated by antibody in two out of four stools from acutely infected volunteers, but in none of their preinoculation stools.

The second method we used to show these particles were specific for hepatitis A was to test under code-paired sera for antibody directed to the 27 nm particles. We quantified the antibody on a 0 to 4+ scale (Fig. 6). A rating of “0” meant there were no particles detected. The virus particles in the F33 stool were present at such a level that if they were not coated with antibody or aggregated by antibody, we could not distinguish the HAV particles from the other material in the stools. If the antibody were rated 4+ on the scale, it meant that the particles were so heavily covered that the outlines of the particle were obscured. Sera containing antibodies rated 4+ also tended to produce single particles or doublets. The particles seen with lower-rated sera were usually in aggregates, but when the antibody was rated about 1+, the aggregates tended to be loose or appear to be almost falling apart. A serum antibody rating of 2+ to 3+ seemed to be near antigen/antibody equivalence, as we would see larger aggregates with those sera.

Figure 6.

Quantitation of antibody on hepatitis A virus (HAV) particles. (A) HAV particles purified and concentrated from stool without antibody. (B) An immune aggregate of HAV with a low level of antibody rated 1+. (C) An immune aggregate of HAV particles with the antibody rated 2+. (D) An immune aggregate of HAV particles with antibody rated 3+. (Photo from the author’s private collection.)

First, we tested six serum pairs from individuals who had been infected with the MS-1 strain. All six of these went from no antibody in the preinoculation sample to easily detected antibody in the convalescent sample. Because these six serum pairs were from volunteers infected with MS-1 and our “antigen” was also MS-1, we considered the possibility that we were looking at something that had been passed with the MS-1 inoculum that was not the hepatitis A virus. Therefore, we tested serum pairs obtained from patients with naturally acquired infection in two outbreaks of hepatitis A, one from Massachusetts and one from American Samoa. All six of these patients showed either seroconversion or antibody increases on the 4-point scale. As controls, we also tested paired sera from two patients with acute hepatitis B and two patients with Norwalk-type gastroenteritis. None of those patients seroconverted or had increases in antibody levels. Therefore, we felt confident that these particles were, at a minimum, virus-like antigens associated with hepatitis A and were most likely the virus itself (Feinstone et al. 1973).

We first saw the HAV particle on October 17, 1973. We published our results in Science on December 7. I believe it was nearly a speed record, but the rapid publication resulted from a lot of very intense work. We were getting new drafts of the paper out almost hourly and as soon as the new one was ready, Al, Bob, and I, as well as Dr. Chanock, would read each draft and edit it immediately. Luckily, our laboratory had an early word-processing machine so that the entire manuscript did not have to be retyped with each draft. We hand carried the final manuscript to the American Association for the Advancement of Science (AAAS) offices in downtown Washington, met with the editor, and went over the entire paper with her. I do not know if she even sent it out for review, but it was very quickly accepted and published just 3 weeks later.

On publication, NIH called a press conference and the finding was reported widely in the news media. We made the front page of the Washington Post early edition that comes out the evening before the publication date, but we were knocked off the front page in the later editions because of the resignation of Egil “Bud” Krogh, who was head of the plumbers investigating leaks from the Nixon White House. Appropriately, Krogh served prison time for his role in the Watergate scandal.

We worked so hard to get the paper out quickly, because we felt that there was some competition to find HAV. We knew the Australians had a similar approach to ours but had only published on fecal antigen so far (Ferris et al. 1970). If they were to change their experimental approach just slightly, they would find the virus. Stephen Locarnini (Fig. 7) was a graduate student working with Alan Ferris at Monash University in Melbourne. Ferris had parted with Geoff Cross and Stephen had decided to work with Alan. Their December 7 issue of Science had not yet arrived by March of 1974 when Stephen identified 27 nm virus-like particles using EM to examine a precipitin band formed in a gel diffusion assay with an acute stool extract and hyperimmune rabbit sera raised against acute phase stools. Stephen was quite excited, until he walked into Alan’s office and was given the just-arrived Science issue with the words, “These guys have discovered hepatitis A.” Obviously there was disappointment, but Stephen quickly confirmed our results by finding similar particles in naturally acquired, not experimental, hepatitis A. He also refuted a paper published in The Lancet by June Almeida, the mother of IEM, who suggested that acute liver disease could result in seroconversion to a multiplicity of particle types present in feces, and who urged caution in accepting that the particle we had discovered was in fact HAV (Almeida et al. 1974; Locarnini et al. 1974). Stephen went on to complete his Ph.D. in a record 3 years, followed by a medical degree and a long, highly productive career in hepatitis research.

Figure 7.

A young Stephen Locarnini in the laboratory at Monash University. Working with Alan Ferris and Ian Gust, Stephen was probably the second scientist to identify hepatitis A virus (HAV) particles and the first to visualize the virus in feces from a patient with naturally acquired hepatitis A. As he had not seen our paper before he first saw the virus, it can be said that he identified it independently. (Photo kindly supplied by S. Locarnini.)

I was slated to leave Bob’s laboratory at the end of June 1974 to complete my clinical training. Therefore, I had very little time to continue the HAV studies and I wanted to get a lot done as quickly as I could. We began a characterization of the virus, which included determination of its buoyant density. Working in John Gerin’s laboratory, we performed cesium chloride density gradient ultracentrifugation studies using stool filtrate in which it was reasonably easy to find particles by IEM. After ultracentrifugation and fractionation, I performed IEM on each fraction. I found a large number of particles in the fraction corresponding to a density of about 1.4 g/cm³ (Feinstone et al. 1974). Although some forms of HAV do band at that density, it was quickly determined that the major density for mature HAV is about 1.34 g/cm³ (Provost et al. 1975; Lemon et al. 1985). My guess is that after I found so many particles at 1.4 g/cm³, I was in a rush and simply did not spend adequate time on the other fractions. The fraction at 1.34 g/cm³ may also have contained so many particles that there was near antigen/antibody equivalence, resulting in most of the particles being sequestered in huge rafts that were not uniformly distributed, and causing me to miss them in looking at our standard five grid squares.

Because of the density that we found, we considered the possibility that HAV could be a parvovirus, although it was a bit larger than the diameters reported for most parvoviruses. Although we did not say it was a parvovirus or a DNA virus, we did suggest it was a possibility. Later, when the group at Merck led by Philip Provost identified HAV in samples from an outbreak in Costa Rica and determined the density to be ∼1.34 g/mL, they suggested that their virus was an enterovirus and therefore different from the virus that we described (Provost et al. 1975). However, all of the morphologically identical virus-like particles from these early studies were shown to be antigenically related, and the NIH particle, the Australian particle, and the Merck particle were all HAV.

IEM was a technique that could be used to detect the virus and also to measure antibody to it. There were so many things to do after the initial discovery, and as I was soon to leave the laboratory, I did not have time to develop a less labor-intensive antibody assay. I just continued to use IEM to do as many things as I could in that short time. As mentioned earlier, with the advent of highly sensitive assays for HBsAg and anti-HBs, virtually all patients with hepatitis B or who had once been infected with hepatitis B could be identified. One of the most puzzling problems was the fact that <50% of patients with transfusion-associated hepatitis could be shown to have hepatitis B. Some thought it was hepatitis A, but Bob Purcell recognized that the epidemiology of those posttransfusion cases, for example, the length of the incubation period, the lack of secondary cases, and the likelihood of chronicity, did not resemble hepatitis A at all. Therefore, one of the first things that we did was to study patients with non-B posttransfusion hepatitis for hepatitis A using our specific, though cumbersome IEM assay. In collaboration with Harvey Alter and Paul Holland at the NIH Blood Bank, where Bob Chanock and Bob Purcell had established a very close collaboration on transfusion-associated hepatitis, we obtained the appropriate samples to evaluate. We studied 22 open-heart surgery patients who had received multiple blood transfusions and who had developed hepatitis without any serologic evidence for hepatitis B infection. We had pretransfusion and posthepatitis sera from these patients and examined them under code for anti-HAV seroconversion or increases in antibody levels. We could not find evidence that any of them had hepatitis A (Feinstone et al. 1975). Thus, the concept of a third form of viral hepatitis, non-A, non-B hepatitis, now known as hepatitis C was born. We presently have five, well-established types of viral hepatitis, including hepatitis D or Delta, and hepatitis E.

The identification of HAV began decades of research in Bob’s laboratory on hepatitis A that culminated in the development of cell culture–adapted HAV and then vaccine development. Ian Gust came to Bob’s laboratory on sabbatical in 1978 and brought with him an Australian isolate called HM175 that he had collected from a family outbreak of hepatitis A (Gust et al. 1985). This virus was highly infectious in marmosets and chimpanzees and was adapted to cell culture better than our other laboratory strains (Daemer et al. 1981). It was the first strain of HAV to be molecularly cloned and sequenced (Ticehurst et al. 1983; Cohen et al. 1987), and indeed it remains a standard HAV strain used in many laboratories today. Formalin-inactivated, cell culture–adapted HM175 virus was protective against a heterologous virus challenge when used as an experimental vaccine by U.S. Army investigators in owl monkeys (Aotus trivirgatus), and was also the antigen in an army vaccine that was the first to be tested in humans (Binn et al. 1986; Sjogren et al. 1991). Eventually, it became the basis for the first licensed commercial hepatitis A vaccine (Innis et al. 1994). All of us who worked on the hepatitis A project in Bob’s laboratory remain proud of the vaccine that has significantly reduced the incidence of hepatitis A, wherever it has been broadly used.

The search for the cause of many diseases with a suspected viral etiology has depended on the technology available at the time. Cell culture was a tremendous boost to viral discovery in the 1950s and 1960s. IEM was a relatively new technology that was available to us in 1972. It was of sufficient sensitivity, and the addition of antibody to standard EM added the specificity that was needed to prove that a virus-like particle observed in stool filtrates from people with acute hepatitis A was indeed HAV. Almost 20 years later, HCV was identified with molecular techniques that were not available in the 1970s (Choo et al. 1989), and newer molecular technologies are of such sensitivity that virtually no virus can evade rapid detection today. However, the principles to assign specificity to any identified virus remain the same. Until specificity is proved, skepticism of any claim of etiology is wise as sensitivity and specificity still have an inverse relationship.

I was a research associate, basically a postdoc, at the time of the hepatitis A work described here. Graduate students and postdocs are the worker bees in most research laboratories, and when a significant or important result is obtained, the postdoc typically gets some credit, but the finding is usually most associated with the head of the laboratory who directed the research. For some reason, the discovery of hepatitis A is most often associated with my name and my career was launched on that one Science paper. Bob Purcell was instrumental in seeing that I received this credit and for that, I remain indebted.