Abstract
Porcine circoviruses (PCVs) belong to the genus Circovirus and the family Circoviridae, and they are the smallest known viruses that replicate autonomously in mammalian cells. They are nonenveloped, and they have characteristic single-stranded, negative-sense, circular DNA. Two types of divergent PCVs are recognized: PCV1 and PCV2. About 20 years ago, PCV2 began to emerge as a major pathogen of swine around the world, leading to burgeoning knowledge about the virus and porcine circovirus–associated diseases. However, much of the history of its discovery, including the controversy related to its importance, is not recorded. This review examines current issues related to the biology of PCV2 in the context of the original studies related to determining its causal association with disease and to the evolving understanding of the complex pathogenesis of PCV2 infections.
More is owing to what we call chance, that is philosophically speaking to the observation of events arising from unknown causes, than to any proper design, or preconceived theory in this business. Joseph Priestley, Experiments on Air (1790)
A perusal of the now extensive literature on PCV2, especially one focusing on review articles, can leave the impression that the current state of knowledge was arrived at through a logical progression of seamless, rational scientific investigation. However, as in many discoveries, 54 it was not that linear, and it was often impeded by groupthink and attendant entrenched dogmatic thinking. 31
Chance Favors the Open Mind
It has now been more than 20 years since Tracy Ward, a veterinarian in private practice, in Humboldt, Saskatchewan, and Chuck Rhodes, the swine medicine specialist at the herd investigation unit at the Western College of Veterinary Medicine (WCVM), investigated the first known outbreak of what came to be called postweaning multisystemic wasting syndrome (PMWS), the classic porcine circovirus–associated disease (PCVAD) entity. This incident occurred in a 40-sow farrow-to-feeder pig operation located in a remote part of northeastern Saskatchewan. In January 1990, the farm had been depopulated and restocked with breeding animals from a high-health multiplier herd. One year later, in 1991, the farmer reported a dramatic increase in nursery mortality, up to 12% to 15%. Clinically, there was jaundice, diarrhea, and respiratory disease, as well as associated ill thrift, icterus, and sudden death. Chuck recalls submitting both live and dead weanlings, along with blood samples. He said that Ted Clark, a pathologist at the diagnostic laboratory at the WCVM, “did an extremely thorough workup on these pigs. We considered everything from aflatoxins, toxicities, and infectious agents, and the histopathology looked liked immunodeficiency.” This made sense since Ted reported that lymphoid depletion was consistently observed microscopically. No known pathogens were identified in multiple diagnostic submissions, and no final diagnosis was obtained. Thereafter, the problem appeared to resolve on that farm.
A few years later in 1994, John Harding, at the time a swine consultant veterinarian in private practice, observed a similar disease syndrome in an intensive high-health swine operation, also in Saskatchewan. By definition, these “high-health” herds were free of common swine infections/diseases, including Mycoplasma hyopneumoniae, Actinobacillus pleuropneumoniae, swine dysentery, porcine reproductive and respiratory syndrome virus (PRRSV), and atrophic rhinitis. Although he and Tracy Ward worked in the same practice in 1991, he had no direct involvement in the initial outbreak, although he had a vague memory of her talking at the time about pigs with jaundice, a clinical sign that he specifically noted in the 1994 outbreak. The first submissions to the Diagnostic Laboratory at the WCVM relating to this new “outbreak” were made in October 1994. Chronic interstitial nephritis, severe focal hepatitis, and ileal muscular hypertrophy were found in 8- to 10-week-old pigs suffering loss of weight and body condition. The specific causes of these findings were unknown, but hepatitis associated with ascarid migration was proposed by Dr J. Orr, the attending diagnostic pathologist. Within months of these initial submissions, the farm experienced a fourfold increase in postweaning mortality, primarily in 2- to 4-month-old pigs associated with dyspnea, icterus, and weight loss. Numerous diagnostic submissions of carcasses, live animals, blood, serum, water, and feed were made over the course of the outbreak. All known swine pathogens were excluded, but with time, Ted Clark, the primary diagnostic pathologist who examined these further cases, began to recognize salient pathologic features common across the cases. He first documented many of what are now known as the classic, multisystemic lesions of PMWS and PCVAD, 2,70 including, notably, the presence of characteristic botryoid basophilic intracytoplasmic inclusion bodies in cells of monocyte/macrophage lineage.
On March 23, 1995, Lori Hassard, a technologist in the diagnostic virology laboratory at WCVM, which I supervised, reported that she had observed 17- to 19-nm dense particles that lacked the “hollow” appearance of parvoviruses, in electron micrographic preparations of material from John Harding’s cases. On the basis of her reading, she thought that they might be porcine circovirus (PCV). Ted and I discounted her seminal photographic evidence because we thought that the particles were probably parvoviruses; or, if they were circoviruses, “it was known”—or at least we vaguely remembered from the literature—that PCV was not associated with disease. 5 Nevertheless, Lori continued her efforts throughout 1995 and 1996 and demonstrated similar particles in more case material but not in normal pigs. In June 1995, Ted posted a brief description of the clinical disease and his morphologic findings on the American Association of Veterinary Laboratory Diagnosticians Listserv (@AAVLD.org) and, about the same time, sent pig-interested pathologists in North America and England representative slides. None of them at this stage reported seeing similar lesions. In December 1995, Barbara Daft, a pathologist at a regional diagnostic laboratory in San Bernardino, California, responded to Ted’s posting, indicating that she had seen three 8- to 15-week-old pigs with clinical signs similar to the those that Ted described. Two had bacterial bronchopneumonia, but one had interstitial pneumonia and a generalized lymphadenopathy with intracytoplasmic inclusion bodies similar to what Ted described and what she had associated with circovirus infection. Dr Daft had dealt with many exotic avian cases and so had seen botryoid inclusions typical of circoviral infections in birds. She therefore also suggested that Ted’s cases may be associated with circovirus infection. This was further supported by evaluations of slides by Craig Riddell, an avian pathologist at WCVM, who mentioned that the inclusion bodies were similar to those he had seen in pigeons with circoviral disease.
Concurrent with the initial pathologic and virologic investigations, representative samples of affected organs were submitted to the diagnostic bacteriology and immunohistochemistry laboratories at the WCVM for routine testing and screening for known swine pathogens. A number of bacteria and viruses were isolated or identified by immunohistochemistry, but none were consistently found in the case material. At the same time (May/June 1996) and following up on the lead provided by routine histopathologic examinations suggesting circoviral infection, Debbie Haines, who supervised the immunohistochemistry laboratory, identified 2 key investigators through cited publications on PCV: Irene Tischer, who had recently retired from the Robert Koch Institute in Berlin, and Gordon Allan, a virologist at Queen’s University and the diagnostic laboratory at Stormont, Belfast. These were the only 2 individuals who appeared repeatedly in the short list of references that she found. Both generously provided rabbit antisera and monoclonal antibodies raised against the virus that Dr Tischer had previously observed and isolated in porcine kidney cell (PK-15) cultures in the 1970s. 89,91 By mid-October 1996, using the rabbit hyperimmune antisera, Debbie had visualized PCV antigen in typical lesions in a lymph node from a PMWS-affected pig. This rabbit hyperimmune antiserum from Dr Allan—labeled A/S—had been raised against PCV recovered from contaminated PK-15 cells. Other donated antisera and monoclonal antibodies to the contaminant PCV were unreactive. Thus, while Dr Allan was excited, he was skeptical of the preliminary findings, since he had already spent years of effort on circoviruses and had obtained consistently negative results concerning a possible association of PCV with disease in swine. Regardless, he was interested enough to continue working with us and provide more reagents and expertise.
A few days later, armed with these preliminary immunohistochemical results, Drs Harding and Clark presented their preliminary clinical and pathologic findings at a regional meeting, the Western Canadian Association of Swine Practitioners in October 1996, and suggested the name of PMWS for the apparently new disease. 18,42 It was immediately following these presentations that Dr John Strokappe, of Red Deer, Alberta, reported seeing similar clinical signs in a number of herds in Alberta and submitted cases to the WCVM. This was the first indication that PMWS was present in more than a single farm. It is also noteworthy that the original 40-sow farm from 1991 was retrospectively diagnosed with PMWS after 1997, indicating that the PCV was associated with disease before the outbreaks that we were investigating in the mid-1990s.
Also in the fall of 1996, Barbara Daft, presented her findings on the pig with interstitial pneumonia and lymphadenopathy at the annual meeting of the American Association of Veterinary Laboratory Diagnosticians. 20 With her collaborators, avian pathologists from Georgia, she reported the presence of paracrystalline arrays typical of circovirues by electron microscopy and demonstration of circoviral DNA by in situ hybridization (ISH) in lesions from that pig. The ISH was done with probes based on the genome of the PCV from PK-15 cells and a “generic” circoviral probe. The probes also detected circovirus-specific sequences in the affected pig by Southern blot hybridization but not from control pigs. On the basis of their observations, they reasonably concluded that this virus (PCV) could be associated with disease in pigs. 20 This contrasted with Tischer’s earlier studies 90 based on experimental infections in the 1980s using the PK-15-origin PCV and with the contemporary work from Gordon Allan’s group 5 that included examinations of fetal material and experimental infections with the tissue culture–derived PCV. Since there was a lack of disease in pigs with documented PCV infections, both these investigators logically concluded that PCV was mildly or nonpathogenic in pigs.
On January 27, 1997, Lori Hassard, after much trial and error over more than 2 years, finally isolated a circovirus from a PMWS-affected pig. 30 She thought that it was “different” from the circovirus originally found by Dr Tischer. To increase the chances for isolation, she had worked with Debbie Haines to identify lymph nodes containing high amounts of PCV antigen on the basis of immunhistochemical staining. The conclusion that this PCV was different from the tissue culture–derived PCV isolated by Dr Tischer was based on differential staining of viruses from cases of disease compared to the staining of the PK-15 virus, with monoclonal antibodies and antisera obtained from Drs Tishcer and Allan. In addition, there was a strong reactivity of the PMWS-derived virus with convalescent serum from a Canadian pig (“Blue-2”). This pig had died of an inguinal hernia and was apparently free of PCV-associated disease. Furthermore, Lori demonstrated that this Blue-2 serum was free of antibodies to PRRSV and porcine parvovirus (PPV) and had questionable or no reactivity with the tissue culture derived PCV. These PCV isolates in cell culture also had in situ hybridization reactivity with a generic PCV probe prepared via the PCV cell culture contaminant. Further supporting the evidence that this was the first isolate of a new virus were the type of cells in which it was isolated. As luck would have it, we had previously (early 1990s) obtained PCV-negative pig kidney (“Dulac”) 25 cells to grow bovine viral diarrhea virus for ELISA antigen, free of the confounding PCV contaminate that was in virtually all readily available isolates of porcine kidney cells. Use of these cells greatly reduced potential confusion and the likelihood that the PCV being identified was not the “old” (contaminant) PCV. In summary, we had a virus growing in cell cultures inoculated with material from PMWS-affected pigs that had DNA homology with the PCV contaminant by ISH but a different reactivity pattern when immunostained with PCV-reactive antibodies. This first isolation was the vitally important evidence that a different (new) PCV was causally associated with the apparently new disease in pigs. Lori then went on to isolate the new PCV from other field cases.
In early 1997, shortly after a PCV was isolated from Canadian PCVAD cases, Gordon Allan, working closely with Francis McNeilly, isolated very similar viruses from Barbara Daft’s case material from the United States and from PMWS cases in Brittany, France, 7 and then from lesional material from other sites in Europe. 10 Antiserum raised against the first Canadian isolate, named “Stoon,” was used in addition to the monoclonal and polyclonal antibodies raised against the PK-15 PCV to clearly differentiate the phenotype of these new isolates from that of the tissue culture–derived PCV. As soon as isolates of the putatively new PCV were available in early 1997, Brian Meehan, a molecular biologist/virologist working with Gordon, provided the first, complete whole genomic sequences of PCVs from North America and Europe, indicating that the disease-associated PCVs shared >95% sequence identity but had <80% sequence identity with the PK-15 PCV. 63 These whole genomic sequences arguably comprised the most definitive data, which Gordon insisted were necessary to make the claim that we had “discovered” a new agent. This sequence information was provided to GenBank in October 1997, and we suggested for the first time in published form that Tischer’s tissue culture–derived, apparently nonpathogenic virus and genetically similar viruses be designated “PCV1” and the genetically disparate disease-associated agents be designated “PCV2.” 63
One investigator in our group (me), because of my background and bias, was not yet completely convinced of a primary causal relationship between PCV2 and the new disease syndrome and still clung to the “hope” of finding an unidentified porcine lentivirus, as these pigs presented with an AIDS-like syndrome. The obvious working hypothesis was that a lentivirus would be immunosuppressive and allow overt and disease-causing replication of an endemically but usually subclinically infecting agent—namely, PCV2. This hypothesis was (grudgingly) rejected shortly after the isolation of the new PCVs and upon further concurrent examination of case material for retroviruses in a collaborative effort with Ana Bratanich at the WCVM, Mike Lairmore at the Retrovirus Center at Ohio State University, and Walid Heneine at the Centers for Disease Control and Prevention. 14
In March 1997, Drs Clark and Harding re-presented their expanding findings—which included immunohistochemical evidence of PCV2 in many PMWS cases, although without the details of virus isolation or sequence information—at the annual meeting of the American Association of Swine Practitioners in Quebec City. 19,43 Many in that audience and subsequent audiences were less than receptive to the accumulating data describing an apparently new syndrome that was causally linked to this apparently “new” circovirus. The commonly held belief was that PCV was a “red herring” or was more aptly named “circus virus” and that the real culprit in these cases was yet another variant of PRRSV. 22,31 For some, this belief persisted, until the weight of accumulating evidence—often in the form of piles of PCV-laden hog carcasses in their neighborhood and, ultimately, the disease-sparing effect of single-component PCV2 vaccines—rendered their position completely untenable. Sadly, it was this skepticism that hindered financial investment into research to clarify the role of PCV2 in disease.
Nullifying Koch’s Postulates
After a PCV was directly associated with lesions in PMWS-affected pigs and the virus was obtained in apparently pure culture from case material, the next logical step in disease investigation was the fulfillment of Koch’s postulates, the conventional approach to establishing causality. Again, as luck would have it, Steve Krakowka, from The Ohio State University (OSU), was visiting the WCVM in March 1997 and was impressed by the lesions from a PMWS-affected pig. As we had isolated a virus by this stage, we discussed inoculation of gnotobiotic pigs with this newly isolated Stoon virus as the best way to demonstrate the primacy of PCV in a controlled and germ-free environment. This work was contracted through the Gnotobiotic Life Unit at OSU. Both Steve and I had had experience in reproducing disease in various “one agent–one disease” models for the purposes of evaluating vaccines and pharmaceutical products. Given the copious PCV2 antigen found in the lesions of natural cases and the high prevalence of viremia and disease that suggested an ease of transmissibility on affected farms, we assumed that there were not many subtleties operant in this obviously infectious disease process. In a phrase, our approach was “How hard can it be?” Very hard, as it turned out, or at least very complicated. But, initially, as we had smugly predicted, we reproduced a constellation of gross and histologic lesions characteristic of PMWS, and PCV2 was reisolated and identified in lesions through immunohistochemistry. This was accomplished in our first attempt, in November 1997, with a variety of inocula, including chloroform-extracted (to exclude lipid-enveloped viruses, including PRRSV) lymph node homogenates from a PCV2 field case of PMWS and cultured PCV2 (Stoon isolate). Self-congratulatory, we immediately wrote a manuscript describing the experiments. The manuscript was sitting on my desk, ready to submit pending one final control test result. We had decided to perform a hemagglutination inhibition test for PPV antibodies before and after infection, even though the inocula were PPV negative on the basis of standard fluorescent antibody testing before their use. Imagine our discomfort when the tests indicated PPV seroconversion in several of the inoculated pigs. We then went back and tested the inocula and tissues from infected pigs using polymerase chain reaction (PCR) and multiple passage isolation and found PPV in both. Our PCV2-containing tissue homogenates and in vitro–passaged isolates of PCV2 had low levels of PPV “contamination.” 32
After a brief bout of dismay, we next investigated the apparent relationship between PCV2 and PPV, retrospectively in field cases at WCVM, and in different experimental infections in gnotobiotes at OSU and in colostrum-deprived neonates at Stormont. Using a combination of PCR and immunohistochemistry, PCV2 was found in 69 of 69 randomly selected cases of PMWS from 25 different swine operations collected over 3 years in western Canada. 28 These results contrasted with a recent study based on PCR alone that detected PCV in a minority (about 15%) of naturally occurring PMWS cases in Manitoba. 67 Even though the authors suggested that there may be some differences in virulence between PCV “strains” and isolates, their results supported the contentions that PCV2 was not the (primary) pathogen in PMWS. 67
In contrast to our finding of the consistent presence of PCV2 in the PMWS cases and even allowing for some differences in the sensitivity of testing methods, PPV was found in only 12 of the 69 cases, whereas PCV1 was found in only 9. 28 These data provided the first evidence that coinfection (or at least concurrent coinfection) with PPV was not necessary to the development of naturally acquired PCV2-associated disease. As well, the low prevalence of PCV1 in case material conflicted with a contended relationship between PCV1 and PMWS, based on the isolation of PCV1 from field cases. 1,57 This was also the first indication that the previously reported high seroprevalence of antibodies to PCV1, 25,90 which was based on Brian Meehan’s sequence data, 63 was probably the result of the detection of antibodies to the highly conserved replicase that both viruses share and not indicative of infection with PCV1. Thus, PCV2 had probably been present in swine populations for quite awhile. Neither PCV2 nor PCV1 were found in non-PMWS-affected pigs with clinical Streptococcus suis infections, indicating that PCVs were unlikely to be endemic, persistent, potentially lifelong infections in pigs without PCVAD. 28
The results of the experimental infection studies with the same, now purified, isolates of PCV2 and PPV that had been used (PCV2) or found (PPV) in the first attempt at experimental reproduction of disease were essentially identical. 3,53 Pigs infected with either agent alone generally had no disease, minimal if any histologic changes, and low amounts of PCV2 in lymphoid tissue. These results provided insights into the occurrence and nature of subclinical PCV2 infections, which became more apparent later when vaccines were applied. In contrast, pigs coinfected with the 2 agents had variable but more severe and sometimes fatal disease and lesions with more abundant PCV2. Given these complementary field and laboratory results, we positioned PCV2 in the classic algorithm of causality: it is the necessary but not sufficient cause of PMWS (and other PCVAD). 3,53
Based on the swine industry’s focus on PRRSV and the recent report of PRRSV coinfection in some field cases of PMWS, 56 PRRSV was the next obvious agent to test in a similar series of experiments. As with PPV, coinfection with PRRSV enhanced the replication of PCV2, as was readily discernible in the PCV2 antigen content of immunohistochemically stained lymph nodes. The converse was not observed; PCV2 had no apparent effect on the replication of PRRSV. 6 For us, these results, admittedly to our initial chagrin, provided an insight into the complexity of the biology of PCV2 infections that broadened our appreciation for the pathogenesis of infectious disease. In contrast, for others, they proposed a confused reality and supported their view that PCV2 was not really a pathogen or, rather, a mere secondary invader.
Adding to this confusion and naysaying was the application of PCR to routine diagnostics in the 1990s, often as a replacement for the classic diagnostic approaches that we had used, including pathology, virus isolation, and electron microscopy. As mentioned, shortly after Ted’s original public descriptions of the disease, some diagnostic labs examined what they said were PMWS cases and reported the inconsistent detection of PCV based primarily or exclusively on PCR tests. 56,67 Others began to find PCV2 in blood by PCR in clinically normal animals, leading to the errant conclusion that PCV2 could not be pathogen. Steve Sorden’s original (and still used) case definition necessitated the presence of typical clinical signs, typical histopathologic changes, and, importantly, PCV2 antigen or DNA in lesions. These requirements for diagnosis of PMWS were published in a practitioner journal 86 in 2000 and added clarity but were ignored by many. Nevertheless, in the intervening years between the seminal experiments establishing the role of infectious cofactors in disease genesis and the present, we and many others have repeated and extended the original findings with various combinations of the same and different agents in various investigative scenarios involving the field and the laboratory. These findings have been sequentially reviewed. 29,69 In toto, the phenomenology fits a pattern: single infection with PCV2 can be associated with subclinical or mild disease, whereas coinfection (all other things being equal) can be associated with more severe disease.
An important nuance of the PCV2–infectious cofactor relationship, again first suggested in an early study by Gordon Allan and colleagues at Queen’s University, Belfast, is the effect of the timing of coinfections on the overall outcome of PCV2 infection. 9 In those first experiments based on a dual PCV2–PPV infection, they tediously and sequentially examined lymphoid and other organs using virus isolation and fluorescence staining. Even though both agents were given at the same time, the PPV apparently replicated more rapidly and declined, setting the stage for enhanced replication of PCV2. In more recent experiments, the important effect of timing and sequence of infection was prospectively demonstrated in torque teno virus–PCV2 infections where enhancement of PCV2 replication was observed only when torque-teno virus was given first. 27 Conversely, no effect on PCV2 infection was observed when PCV2 was inoculated concurrently with Mycoplasma hyopneumoniae. 82 Yet despite, the accumulated information, we are far from understanding the mechanisms involved in the synergy between PCV2 and other agents and how they may differ depending on the particular coinfecting agent. Proposed mechanisms that may enhance PCV2 replication during coinfections include the initiation of host cell replication (necessary for the S-phase-dependent PCV2 replication), altered (enhanced) cytokine response, immune dysfunction (immunosuppression) that reduces clearance of PCV2, and alterations (decrease) in virus-controlling innate (interferon-mediated) immune responses. 69
Rejecting Conventional Wisdom
After the pathologists and microbiologists among us had recovered from the shocking possibility that maybe Koch was wrong and even began to admit—at least behind closed doors—that epidemiology may be important, in other words acknowledging that a simple infection with PCV2 was probably not enough to cause fulminant disease in most pigs, the conventional wisdom would have been that coinfection resulted in immunosuppression that allowed enhanced, pathologic replication of PCV2. This would not have been a conceptual “leap” given that it was known that PPV and PRRSV were “immunosuppressive.” About that time, sometime in late 1999, Steve Krakowka asked himself and other members of our research group the rhetorical question “What if rather than suppressing the immune response, coinfection activated the immune system, thereby enhancing the replication of PCV2?” He then designed and performed a seminal experiment demonstrating for the first time the latter effect in the absence of other agents. 52 Using internal controls, Steve showed that PCV2 replication was significantly increased in lymph nodes ipsilateral to local immunostimulation with a prototype adjuvanted vaccine but not in contralateral unstimulated nodes. More dramatic PCV2 replication, severe clinical disease, and, often, death resulted when pigs were more systemically “immunostimulated” by intraperitoneal inoculation with thioglycollate broth concurrent with PCV2 infection.
Shortly after Steve completed this study, Gordon Allan presented the results at a meeting of swine consulting veterinarians in Italy. After his presentation, one veterinarian reported to Gordon that he thought that he had seen a similar enhancing effect on PCV2-associated disease following the administration of some commercial vaccines in the field. Lending credence to this clinical observation was the next set of experiments, demonstrating that administration of some commercial vaccines after PCV2 infection early in life (3 days)—notably, one containing an oil-based adjuvant—could result in severe and sometimes fatal PCV2-associated disease, 51 with the operative word being “could.” Between the completion of these experiments and a delayed publication, a similar phenomenon was demonstrated with modified-live PRRSV vaccine and other vaccines, immunostimulants, and adjuvants. 70 Most of these efforts focused on vaccines containing Mycoplasma hyopeumoniae because of their routine use in young pigs and in toto generally confirmed Steve’s original findings with the prototypical vaccine: immune activation could take a subclinical PCV2 infection to a fatal one. This opened a new Pandora’s box and clearly raised controversy, as it highlighted the inadvertent harmful effects of vaccines. Nevertheless, these experiments were another important piece in the complex pathogenesis of PCV2- induced disease. They complemented and extended the seminal work of Madec 60 and subsequently many others—notably Nicholas Rose, who identified management factors, essentially “stressors,” including husbandry practices, early weaning, mixing of pigs, and dietary changes, as well as early and frequent vaccination, which could modulate the outcome of PCV2 infections. 73 Ironically, all are factors of modern swine production.
One of the darkest corners of PCV2 epidemiology is the role of host genetics in disease expression. We first documented 33 and others confirmed 75 that PCV2 could infect and cause disease in the hairier, genetically disparate versions of Sus scrofa (wild boars). Ironically, there are more published data addressing the immunogenetics of PCV2 infection in wild boars 75 (which, interesting though they may be, are largely irrelevant economically) than there are for PCV2 in commercial swine. Of the now more than 900 papers concerning PCV2, only 5 have investigated the role of genetics of commercial swine in PCV2 biology. 73 Virtually all this work concerns an interaction between PCV2 and host genetics as related to breed differences. With acknowledgement of the great difficulty in investigating genetic connections to disease resistance/susceptibility, the approach that has been taken to date seems to be a rather blunt instrument. It is perhaps significantly more likely that line differences within breeds 74 hold the key to understanding any genetic links with PCV2 infections, especially given the artificial selection pressure for loin length and other performance traits, a selection process that is based primarily on pedigree and not breed. Studies carried out as part of a European Union–funded consortium (PVCD.net) using microsatellite markers and single-nucleotide polymorphisms failed to elucidate regions in the pig genome that could be definitively linked to susceptibility/resistance to PCVAD. Such investigations can be complicated by multigenic effects and limited by accessibility to sufficient numbers of animals with detailed pedigree and disease data. Therefore, examination of pedigree-related effects will require that pig-breeding companies embrace the transparency that is necessary to conduct this research.
One of the big open questions in the epidemiology of PCVAD is, where did PCV2 come from? and how did it apparently spread so fast? Recent data and modeling suggest that international trade and transport of animals (mostly breeding stock) may have been critical in the rapid dispersion of PCV2, including new variants. 96 Another important question, then, is whether this practice, combined with inadvertent “selection” for genetic susceptibility to PCV2 that could be in linkage disequilibrium with performance traits, 74 contributes to or ensures the emergence of PCVAD.
Given the gestalt of disease at the population level, it is almost logical to argue from the standpoint of “radical epidemiology” (ie, minimizing or negating the culpability of the defining etiologic agent in a disease process 101 ); that is, as poverty causes tuberculosis, 23 modern swine production causes PMWS. However, such arguments do not answer the question, how does PCV2 infection result in damage or disease at the level of the cell, 98 organ, and organism, in other words in the thing that ultimately composes the population?
More Than a “Trojan Horse”?
Beginning with Ted Clark’s original histopathologic descriptions, the monocyte/macrophage has been the cell of major interest in PCV2 biology. 18 This made morphologic sense, since even in the absence of extensive organ damage, changes in lymphoid tissue—including variable lymphoid depletion associated with large, often multinucleated, often inclusion-body-containing macrophages—were a characteristic, approaching pathognomonic, feature of PCV2 infection. With the first application of immunohistochemical staining, it became clear that PCV2 was in fact more broadly tropic, indeed pantropic, but the association of PCV2 antigen or DNA with monocytes/macrophages was even more eye-catching in immunohistochemically stained preparations. We first pictorially documented this broader tropism in an article describing the first isolation of PCV2. 30 The spectrum of target cells for PCV2 was later broadened to fetal tissues, in the first documentation of vertical transmission and PCV2-associated fetal damage, again with case material examined at the WCVM. 100 Apparently productive infection of cardiomyocytes was particularly prominent in infected fetuses. Nevertheless, we and others initially talked and wrote about the virus as being “monocyte/macrophage tropic”; in other words, those cells were primary targets of viral replication.
But there was an apparent niggling morphologic “disconnect” that was discussed in several meetings of our group, usually soporifically for the nonpathologists. The immunohistochemical staining in monocytes/macrophages was almost exclusively cytoplasmic, whereas in a variety of other cells that often appeared normal, it was often only nuclear or both nuclear and cytoplasmic. This was discordant with the traditional dogma that DNA viruses replicate in the nucleus and are associated with intranuclear inclusion bodies. Conceptually, the use of nuclear DNA replicative machinery would be especially necessary for PCV2, which, like PPV and other parvoviruses, codes for a limited repertoire of enzymes. In the quest to locate sites of cell replication, Deirdre Gilpin, a graduate student of Dr Allan, demonstrated for the first time, using conventional staining methods and at Brian Meehan’s suggestion, replicative intermediates as molecular evidence 63 that although PCV2 accumulated in blood monocytes and pulmonary alveolar macrophages, it did not replicate there. 39 Furthermore, virus was not observed at all in unstimulated lymphocytes. 39 Then, Ken McCullough and his graduate students and postdoctoral researchers at the Institute of Virology and Immunoprophylaxis in Mittlehausern, Switzerland, confirmed Deirdre’s results with monocytes, extended them to dendritic cells, and provided data that PCV2 was endocytosed rather than truly, productively infecting (replicating in) those cells. 97 Together, these data led to the still currently held view that PCV2 can persist in this large heterogenous family of myeloid cells without replicating or being degraded. The data concerning if or at which activational state PCV2 infects lymphocytes and directly affects lymphoid depletion are more conflicting and unresolved. 21 Nevertheless, available data collectively led to the suggestion that maybe the PCV2-containing cell of monocyte–macrophage–dendritic cell lineage is primarily a “Trojan horse” and not a true target for replicative PCV2 infection.
From the beginning of the PCV2 story, Annette Mankertz and her students and postdoctoral researchers at the Robert Koch Institute in Berlin have led the field in examining the molecular interactions between the virus (viruses) and host cells. 61 Complementing these studies is the work of Hans Nauwynck and his students in Ghent documenting that there is nothing particularly unique about how PCV2 gains entry into cells. 66 Like many other viruses, it uses cell surface sugars, the glycosaminoglycans, specifically heparin sulphate and chondroitin sulphate B, although entry into the target seems to be slower than in many viral infections and may be enhanced by type I and type II interferons. 66 The former observation is consistent with the apparent pantropism that the original immunohistochemical investigations indicated. The latter finding is consistent with observations regarding immune activation.
Since we now have considerable information on the interactions between PCV2 and target cells, how does PCV2 actually “cause” the debilitating disease that is characteristic of PMWS and other PCVAD? 61 Does the effect of PCV2 infection fit somewhere in the classical and divergent necrosis/apoptosis pathways? 26 Or, does PCV2 act in one of the growing list of variations on the cell death theme? 26,102 There are conflicting data concerning the role of apoptosis, historically the most studied mechanism of “programmed cell death.” In a recent review, Mankertz proposed that the ORF3 gene product of PCV2 induced apoptosis, an initially attractive hypothesis because it is this gene product that differs significantly in size and sequence between PCV1 and PCV2. But that hypothesis was not borne out in various experiments using ORF-null or chimeric viruses in pigs. 61 However, data support a central role for apoptotic events in the massive organ damage, notably in the liver, that is coincident with clinical PCV2 infection in vivo. 85
Allowing that PCV2 infection itself can lead directly to death of infected cells—for instance, fetal cardiomyocytes or activated lymphocytes—by some unknown mechanism, could it be that cell damage and death and, ultimately, systemic debilitating disease are really nothing more than or at least in large part “proinflammatory cytokine intoxication”—the kind of “poisoning” that occurs in chronic infections such as tuberculosis 72,92 or in terminal cancers of various sorts? 80 The early and consistent observation that PCV2 is minimally cytopathic in renal epithelial cells in vitro and apparently in many infected cells in vivo, 30,90,91 is another clue that, perhaps, it is the response to the infection rather than the infection itself that results in the most damage and death to infected and bystander cells and a general decline into catabolic state. 12,102 This scenario happens in many chronic infections. 12,72,87,92
So, are the putative intraisolate variations in virulence simply attributable to more rapid and efficient growth of some isolates in those animals with an optimal physiologic state for virus replication, resulting in more rapid and prevalent stimulation of mononuclear phagocytic cells and subsequent production of cytokines? In addition to early observations 39,97 related to the longevity without degradation of PCV2 in monocytes/macrophages, other data support this overall scenario. First, the work in Steve Krakowka’s PCV2-infected gnotobiotic pigs has repeatedly documented a simple association between the viral load and outcome of infection. 49,50,52 Much of this load is found in monocytes/macrophages and is often associated with massive cell death, including that of uninfected cells. 49,50,52 Validating these laboratory findings, an assessment of viral load is now being used as an in vivo prognosticator in the field. 77 Second and complementarily are the studies in Quim Segales’s laboratory and elsewhere documenting the production of various proinflammatory cytokines, including tumor necrosis factor, in response to PCV2 infection. 21 Together, are these elucidative of a final common pathway of necroapoptic events? 102
A Rose by Any Other Name
As mentioned, the seminal sequencing of the first PCV2 isolates from field cases by Brian Meehan showed that they were very closely related: a greater-than-96% intragroup nucleotide sequence identity. In contrast, the newly recognized PCV2 had a less-than-80% overall nucleotide sequence identity with the PK-15-derived PCV1. 63 These viruses have 2 major open reading frames: ORF1, which codes for 2 proteins that compose the DNA replicase, and ORF2, which encodes for the capsid protein. Analysis of ORF1 and ORF2 of PCV2 showed 83% nucleotide and 86% predicted amino acid identity with PCV1 for ORF1 and 67% nucleotide and 65% predicted amino acid identity with PCV1 for ORF2. 63 From the late 1990s to the middecade of the new millennium, more closely related PCV2 viruses were isolated or otherwise identified and were associated with a spectrum of PCVAD in various parts of the world. 78 Then something happened.
In 2004, Carl Gagnon and his colleagues at the diagnostic laboratories at the University of Montreal in Quebec, as well as Susy Carman and her colleagues at University of Guelph in Ontario, began to see PCVAD that was apparently different from the disease originally described by John Harding. 15,37 Older grower pigs were affected rather than weaners. Many of the lesions were the same as Harding and Clark originally reported, except that they were more severe and included prominent pulmonary edema with apparent acute pulmonary injury and, notably, splenic infarcts. Gagnon insightfully acknowledged that at least some of the differences in age of presentation, severity, and organ involvement may be due to differences in infectious doses or other cofactors and so did not lay all the blame on a potentially different PCV2. 37 PCR products were sequenced from cases and mainly segregated into a previously proposed PCV2b genogroup. 55 Carman used restriction fragment length polymorphism typing to show that the PVC2 viruses associated with these disease outbreaks were different from the PCV2 in previous outbreaks and that these differences essentially segregated into Carl’s 2b versus 2a designations. 15 About a year later, in late 2005, similar outbreaks occurred in the United States, in Kansas, Iowa, and North Carolina, and were also associated with the apparently “new” genotype, PCV2b. 17 Shortly after these outbreaks occurred, the increasingly confusing nomenclature of PCV2 subtypes was standardized with proposed designations PCV2a, b, c. 79
Analysis of molecular epidemiologic data from before and after these outbreaks indicated a shift in the predominance of PCV2 viruses circulating in pig populations worldwide, from PCV2a to PCV2b. 81 Nevertheless, the expected evolution in PCV2 led some to wrongly equate subtype (2b) with virulence, either forgetting or being ignorant of the very dead pigs in Alberta, Brittany, and Catalonia that had been infected with PVC2a virus in the original outbreaks of PCVAD. 2,7 Certainly, this (predictable) evolution of PCV2, with apparent differences in clinical presentation and lesions among isolates of the same genogroup, raised legitimate questions about possible differences in the virulence in the newly emergent strains. This prompted several prospective studies. This work, based on different isolates in different infection models, generally indicated no documentable difference in virulence between PCV2a and PCV2b subtypes (or isolates), with some exceptions. 94 A notable recent exception is a study with a putatively more virulent PCV2b variant, originally isolated in China but now probably also circulating in the United States. 41,103 However, the findings need to be interpreted with caution. Since there was no accounting for a “pen effect,” 41,76 the “n” in the experiment was 4, not 20; therefore, the claims relating to differences in virulence cannot be substantiated statistically. 76 An interesting, as yet not understood, twist on the virulence theme is the results of our study in which superinfection with a PCV2b variant 7 days after infection with a PCV2a isolate resulted in disease, whereas infection with either variant alone did not. 44 Minimally, these results further support the concept that a particular genogroup designation does not define virulence.
Once outbreaks of PCVAD occurred or were at least widely acknowledged in the United States, PCV2 somehow became real, even though some US workers, notably at Iowa State University, had been working with PCV2 since the late 1990s and were seeing field cases. After we had established the role PCV2 in disease, initial serologic surveys conducted in Gordon Allan’s laboratory indicated that pigs in Northern Ireland had antibodies to PCV2 in the early 1970s. 99 As part of that original effort, we reported seronegativity in a variety of other species, including pig veterinarians and our personnel, implying that PCV2 was exclusively or primarily a pig virus. 8,34 Subsequent surveys from various sites around the world have found a similar prevalence and extended that to at least the late 1960s. 4 In addition, retrospectively, there is documentation that isolated cases of PCVAD had occurred in the mid-1980s in both England 45 and Germany. 47
But what about before that? As Gordon Allan cited in a recent review, 4 Brian Meehan 62 and others 38 had proposed that PCVs initially evolved from a nanovirus DNA that was transferred from a plant to a vertebrate, possibly when a pig was exposed to sap from a nanovirus-infected plant. Furthermore, this event maybe involved recombination of the nanovirus with a calicivirus with the help of a reverse transcriptase from a coinfecting retrovirus. 38 More recent investigations suggested that circoviruses have been with their hosts for only about 500 years; PCV2a and PCV2b diverged from each other approximately 100 years ago and have been independently evolving while cocirculating since then. 35
So, since PCV2, though maybe not ancient, is apparently not a new virus, how come “epidemics” of PMWS have not occurred until recently? Certainly, the gamut of management changes already discussed undoubtedly played a role in the emergence of disease, but what about changes in PCV2 before the “epidemics” of the mid-1990s? Recently, Brian Meehan—using PCR fragments from tissues that Gordon Allan resurrected from the freezers in Belfast from 1970 to 1971—identified in the ORF2 nucleocapsid gene of archival PCV2 isolates a consistent sequence difference of 9 bases (positions 1331–1339) corresponding to amino acid residues 133–135 of a B-cell epitope in the capsid protein. 48,88 This altered sequence converted a hydrophilic region into a hydrophobic region. A reconstructed virus, based on the contemporary ORF1 backbone and this “archival” ORF2 sequence, was inoculated into gnotobiotic pigs and caused only a subclinical infection with minimal virus in only lymphoid tissue. 48 Whether differences in encapsidation potential or other properties resulted in the taming or attenuation of the contemporary PCV2 remains undetermined. However, this change or a similar one could be the missing link in the evolution of pathogenic PCV2. Of course, all the discussion and uproar about viral evolution, viral molecular genetics, and virulence differences, already arcane for many, become really academic for the veterinarian, the producer, and the pig, if there were vaccines that solved PCV2-related problems.
Nothing Confirms Like Success
Once a novel viral isolate existed with unique sequence information that was causally associated with an apparently new disease, the next step was the development of a vaccine. While this seems logical and likely to be supported internationally, again, the politics of research slowed progress. In April 1997, with Gordon Allan and Steve Krakowka, I wrote a brief research proposal outlining a series of experiments aimed primarily at solidifying the causal role of PCV2 in PMWS, including exclusion of retroviruses, and developing a challenge model that could be used in vaccine efficacy testing for this unique pathogen. At the time, we had the only or some of very few isolates of PCV2. We circulated this proposal to 3 of the largest veterinary biologics companies in the hope of collaboration. Two showed polite interest in the proposal but were noncommittal. This response was understandable given that significant numbers of people in areas of swine production and research had major questions about the validity and importance of our work. Moreover, and perhaps more important from a commercial standpoint, none of our group were recognized “opinion leaders” in the field of swine medicine. Fortunately, the third company, Rhone Merieux (Merial Ltd), was immediately interested and entered into a long-term research agreement in May 1997. Catherine Charreyre and Francois Joisel spearheaded the team at Merial, with Catherine in charge of development within Merial and Francois leading the “education process” about PCV2 and application of vaccines in the field. Their efforts were often reminiscent of Sisyphus and his rock.
Shortly after beginning the joint vaccine development project involving the University of Saskatchewan, Queen’s University, and Merial, the PCV2 story became more complicated with the discovery of the common role of infectious cofactors in the pathogenesis of PCVAD and the subsequent observation that stimulation of the immune system could potentially enhance disease. Primarily, because of the latter possibility and epidemiologic evidence that viral transmission early in life may be due to heavy environmental burden and robustness of PCV2, the team decided to take a conservative approach to immunoprophylaxis by creating a sow vaccine and a protocol to enhance passive immunization. The idea behind this approach was to extend the duration of passive immunity but to increase the chances for natural priming of acquired immunity while piglets had disease-sparing concentrations of maternal antibodies. Although this approach has been vindicated in several studies, 13 most recently in the context of its overall effects of reducing viremia in piglets, 68 sow vaccination to control PCVAD has been controversial. An observation from the field was that this approach simply delayed the appearance or onset of PCVAD. This makes sense biologically, if the timing or dose of exposure to a pathogen in the environment does not appropriately prime a passively immune individual. 65 To address this real concern, 3 other vaccines were designed for a more conventional use to stimulate acquired immunity in piglets. The original sow vaccine is now approved for use in piglets. There are conflicting data concerning the ability of the parenteral vaccines to prime in the presence of maternal antibodies. 13,36 This was probably predictable, since the latter is not an all-or-none phenomenon, depending on the passive immune status and vaccine antigen content. 84 The original 4 vaccines constitute the commercial market today, and it would be naïve not to acknowledge that controversy over vaccine protocols is, at least in part, due to a focus on fourth-quarter earnings by marketing groups with different label claims at respective manufacturers versus what combinations of available vaccines and routes of administration may improve overall efficacy.
All the currently available vaccines are based fundamentally on Brian Meehan’s original and patented sequence data. All therefore comprise capsid antigens from PCV2a. Of course, with emerging data on new variants of PCV2, primarily PCV2b, there are emerging concerns about viral escape mutants and possible vaccine failures. As luck would have it, again, postlaunch epidemiologic evidence directly addressed this concern at the population level. 13 At the time of the launch of the PCV2a-containing vaccines, the predominant circulating field strains were PCV2b. Therefore, the perceived and documented vaccine efficacy in reducing PCV2-associated disease evinced vaccine-stimulated protective heterologous immunity. 13 This cross-protection was reproduced in experimental infections in the laboratory. 13 Arguably, a more profound and potentially more important economic effect than sparing of clinical disease, which is only the tip of the epidemiologic iceberg, is the formerly anecdotal and now documented effect of PCV2 vaccines on improvement of production parameters at the herd level. This de facto documents the importance and physiologic cost of subclinical PCV2 infection. 13
The demonstrated vaccine efficacy has now been associated with a broad range of (neutralizing) antibody and cell-mediated immune responses that have been recently reviewed elsewhere, and do not differ significantly from those implicated in protection against more conventional acute viral infections. 13,21 However, there is an interesting twist on the antibody response to vaccine virus and natural infection reported in some animals. 93,95 It has been shown that pigs vaccinated with baculovirus-expressed capsid protein were more prone to develop a predominance of neutralizing antibodies to the largest capsid polypeptide (43-233) versus naturally infected pigs, which had fewer neutralizing antibodies reactive with both this polypeptide (43-233) and a smaller capsid peptide, DP (169-180). This latter oligopeptide is shared between PCV2a and PCV2b subtypes and putatively serves as a “decoy” that diverts antibody responses away from the larger protective epitope. The overall significance of these differential responses remains to be determined, especially since many naturally exposed animals may apparently clear PCV2 infections and do not develop disease and vaccines other than the baculovirus construct seem to be efficacious. Regardless, our group was surprised at how well the vaccines work in the field. We would not have predicted that, given the apparent overall complexity of PCVAD. Moreover, the apparent success of vaccination at the population level is perhaps the most profound confirmation of our original observations related to the primacy of PCV2 in an admittedly multifactorial disease process, which defied simple fulfillment of Koch’s postulates.
Just as Steve Sorden’s original case definition for PCVAD added clarity to a confused discussion, at least for those receptive to thinking, perhaps it is time to articulate a case definition for “PCV2 vaccine failure.” 86 With the recent cladistically driven PRRSV-esque fascination 64 for typing of PCV2 isolates and submitting “novel” small sequence differences to GenBank, 16,46,71,103,104,106 such a definition could add clarity to the perception of “failure”; specifically, is the “failure” truly due to the epitoptic constellation of a given vaccine or something else? A similar emphasis on sequence has occurred in many virus infections and beyond 58,59 in addition to PRRSV, notably in BVDV infections in cattle (BVDV1a and 1b) 105 and parvoviral infections in dogs (CPV2a, 2b, 2c). 11 This is not to diminish the critical role of sequence information in determining the molecular epidemiology of infection 11,81 and in identifying true escape mutation or major shifts in virus biology, such as real differences in virulence. Certainly, minor genetic and resultant amino acid changes can have profound effects, as famously demonstrated by carnivore parvoviruses that jumped species barriers from feline to canine 11 and perhaps in the case of our observations related to archival versus contemporary PCV2. 48 But, significant cross-protection has been demonstrated at least under controlled conditions in the laboratory in the cases of BVDV1a and 1b 102 and CPV2b and 2c 83 and at the population level in the case of PCV2a and 2b. 13 These viruses have similar degrees of cladistic distances within their respective populations. This success alludes to the probability that efficacious vaccines, by design or serendipity, target immune responses to enough of the conserved gene products that viruses cannot do without.
Relevant to the definition of a “vaccine break” in the case of PCV2 is a recent article reporting an apparent “vaccine break” in the context of a novel sequence variant of PCV2b. 71 The first paragraph of the discussion of that work noted but did not further address that “the ownership of the farms decided to switch to a program where all pigs were vaccinated twice with one ml of the (PCV2a) vaccine at 3 and again at 6 weeks of age and PCVAD has not been observed in subsequent groups of vaccinated pigs and the mutant PCV2b has no longer been detected.” A logical interpretation of those data would be that young pigs with good passive immunity may require a second dose of vaccine (and at a less stressful time?) to be truly immunized rather than just vaccinated. However, the authors implied another interpretation: that being that the current PCV2a vaccine does not protect against the latest PCV2b mutant. A variation of the former is usually the best explanation that applies in the case of perceived vaccine breaks in feedlots and kennels, where animals are also vaccinated at less-than-opportune times. It is almost trite to mention that there are multiple cofactors in vaccine efficacy, including passive immune status, level and frequency of challenge, the presence of coinfecting agents, overall health-related management practices, or, ominously, the involvement of isolates that are more virulent because their increased fitness is due to antigenically silent mutations. 40 At the risk of stating the obvious, again, all these possibilities have little, if anything, to do with overall antigenicity of a vaccine and would not be addressed even with a monthly reformulation of vaccines.
The now more fully documented mutability of PCV2, including possible recombination, approaches the range of genetic change generally considered characteristic of RNA viruses. 24,94 This, with the high replicative capacity in vivo that Brian Meehan excitedly noted to us in the course of original studies more than 15 years ago, mandate that we not be complacent about change in PCV2 populations. But going forward, neither should there be a Chicken Little response to the expected evolution of PCV2, absent a clear definition of a perceived versus true vaccine failure, as it relates to the latest PCV2 mutant.
Winning the Lottery
From an investigator’s standpoint, the circovirus story is really one of rare privilege and good fortune and, thus, a truly humbling experience: a privilege to see the biology of a new infectious disease and the investigation thereof play out on a world stage and, at the end of the day, a privilege to be left in a state of wonderment. Then there is the best of fortune—or, rather, in many ways, just plain dumb luck—manifested as rare instances of stellar alignment: a convergence of astute clinicians, a persistent diagnostic pathologist, dedicated technicians, and a major biologics firm that took a big risk on a small group of unruly, less-than-conventional investigators with no real presence in one of its major product areas—and, of course, the bug, the host, and the environment in just the right state of flux for something really big to happen. Together, this sort of gestalt just does not happen that often. We were also lucky to have detractors, powerful and entrenched ones, who provided not only the inspiration to stay the course, but a real-time insight into the way that “science” really works, as well as a lot of entertainment. All told, it just does not get any better; Hollywood could not have written a better script.
Footnotes
Acknowledgements
I thank Gordon Allan, Steve Krakowka, and Brian Meehan for critical review of the manuscript; Ted Clark, Barbara Daft, John Harding, Lori Hassard, Debbie Haines, and Chuck Rhodes for review of the chronology of early events; and all colleagues named and unnamed for making the circovirus story the adventure that it has been.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The author and collaborators, and the University of Saskatchewan receive royalty payments on the sale of all currently commercially available PCV-2 vaccines. The author and collaborators received funding for work on PCV2 from Merial Ltd, Wyeth Animal Health, Pfizer Animal Health, Natural Sciences and Engineering Research Council (Government of Canada), and the European Union.