---------------------------------------------------------------------------
Perspectives

Molecular Approaches to the Identification
of Unculturable Infectious Agents

Shou-Jiang Gao, Ph.D., and Patrick S. Moore, M.D., M.P.H., M.Phil.
School of Public Health, Columbia University,
New York, New York, USA
---------------------------------------------------------------------------


          New molecular biologic techniques, particularly
          representational difference analysis, consensus
          sequence–based polymerase chain reaction, and
          complementary DNA library screening, have led to the
          identification of several previously unculturable
          infectious agents. New agents have been found in
          tissues from patients with Kaposi's sarcoma, non-A,
          non-B hepatitis, hantavirus pulmonary syndrome,
          bacillary angiomatosis, and Whipple's disease by using
          these techniques without direct culture. The new
          methods rely on identifying subgenomic fragments from
          the suspected agent. After a unique nucleic acid
          fragment belonging to an agent is isolated from
          diseased tissues, the fragment can be sequenced and
          used as a probe to identify additional infected tissues
          or obtain extended portions of the agent's genome. For
          agents that cannot be cultured by standard techniques,
          these approaches have proved invaluable for
          identification and characterization studies. Applying
          these techniques to other human diseases of suspected
          infectious etiology may rapidly elucidate novel
          candidate pathogens.

Identifying the causative agent of an infectious disease is the cornerstone
for its eventual control. In recent years, a great deal of progress has
been made in identifying new agents associated with both well-known and
newly emerging infectious diseases. A number of syndromes exist, however,
in which infectious etiology is likely, but the pathogen resists
cultivation with standard microbiologic techniques. For emerging diseases,
rapid identification and characterization of the responsible agent are
crucial first steps for epidemic control.

The rapid identification of a hantavirus responsible for an outbreak of
severe pulmonary distress syndrome in the southwestern United States (1, 2)
demonstrated that applying molecular biologic approaches can accelerate the
identification of an unknown agent. With extensive nucleic acid and protein
databases readily available, isolating and sequencing genomic fragments
from an unknown agent can provide important clues regarding its origin and
biologic behavior. Once a new agent's phylogenetic relationship to other
known organisms is established, appropriate culture conditions, serologic
tests, and perhaps even therapeutic strategies can be rapidly developed.

Although they have revolutionized our ability to identify new pathogens,
innovative molecular biologic techniques must be applied in conjunction
with traditional epidemiologic procedures. Beral, Jaffe and colleagues, for
example, showed that AIDS-related Kaposi's sarcoma (KS) is likely to be
caused by a transmissible agent other than human immunodeficiency virus
(HIV), before any likely causative agent was isolated (3, 4). These
findings focused the attention of investigators on AIDS-KS, which resulted
in the isolation of viral DNA from AIDS-KS lesions and the description of a
new human herpesvirus (5, 6). Spurious associations between infectious
agents and diseases are common, however, and only through careful
epidemiologic studies can a causal link between an organism and a disease
be established. Epidemiologic criteria for causality (7-9), superseding
Koch's postulates, have been used for 30 years, and a critical phase in the
process of new pathogen discovery involves the unambiguous establishment
that an agent is central to the disease process.

The various molecular biologic approaches to agent identification differ in
technical detail, but all rely on isolating nucleic acid fragments
belonging to the agent's genome from diseased tissue. The formidable tasks
of identifying and separating small unique nucleic acid fragments from
human genomic material have been approached by various means, each with its
own particular strengths and weaknesses. The appropriate technique depends
on the type of infectious agent involved (e.g., bacterial or viral),
whether the disease occurs in a normally sterile site, and whether it can
be passaged through animals.

Once a fragment from the agent's genome is isolated and sequenced, standard
genomic walking techniques are used to extend the known sequence, and
computer homology searches can be used to identify the likely phylogenetic
relationship of the agent to other known organisms. In this article, we
highlight recent successful situations when molecular approaches were used
to identify and characterize unknown agents of infectious diseases.

Representational Difference Analysis

Representational difference analysis is one of the more robust methods of
identifying new agents since it does not require prior knowledge of the
agent's class (10). The technique is based on polymerase chain reaction
(PCR) enrichment of DNA fragments present in diseased tissue but absent
from healthy tissues of the same patient (Figure) [Figure not available in 
ASCII). Representational difference analysis is also an important tool 
for identifying polymorphic DNA sequences associated with noninfectious 
diseases (11).

[Figure]
Figure. Schematic representation of different molecular approaches to the
identification of unculturable infectious agents.

Representational difference analysis depends first on digesting DNA from
both healthy and diseased tissues by using a restriction enzyme and then on
separately “simplifying” the resulting genomes to reduce their sequence
complexity. This is done by ligating PCR primers to both sets of DNA and
nonspecifically amplifying the mixtures. Since PCR most efficiently
amplifies fragments of 150 to1500 bp, restriction fragments in this size
range are enriched, and the fragments of the sequence represented outside
this size range (90%) are reduced.

Unique strands of DNA from diseased tissue representing restriction
fragments of an exogenous agent are isolated in a subtractive hybridization
process coupled to PCR amplification. First, the priming sequences ligated
on DNA restriction fragments of both normal and infected tissues are
removed. New primer sequences are ligated only to the diseased tissue DNA
fragments. These are then hybridized with an excess of the healthy tissue
representation. Human DNA fragments common to both diseased and healthy
tissues reanneal to each other and, since the healthy tissue fragments are
in excess, any given human fragment derived from the diseased tissue will
reanneal to a complementary strand from the healthy tissue representation.
Thus, common human sequences found in both representations will only have
one PCR priming sequence or none present on two complementary strands.
However, DNA fragments from the infectious agent will not find
complementary strands in the healthy tissue representation and will
reanneal with each other. Only hybrids with both strands derived from the
diseased tissue representation will have priming sites and be able to
undergo subsequent exponential PCR amplification. Several rounds of
representational difference analysis are performed, which successively
enrich the mixture for unique DNA sequences present only in the diseased
tissue representation.

KS-Associated Herpesvirus and KS

The power of representational difference analysis and the difficulties
encountered in establishing the etiology of disease are illustrated by its
application to KS (5), a vascular neoplasm that frequently occurrs in
homosexual men with AIDS (3). Geographic clustering (12, 13) and
association with specific sexual behavior (4, 14) suggest that the disease
is caused by a sexually transmitted agent. Several agents have been
investigated, including human cytomegalovirus (CMV), human papillomavirus,
human herpesvirus 6 (HHV-6), and HIV (for review, see [15]), but no
convincing etiologic link has been established.

Using representational difference analysis, Chang and colleagues isolated
two unique DNA sequences (KS330Bam and KS631Bam) from a KS lesion in an
AIDS patient (5). These DNA sequences are homologous to portions of minor
capsid and tegument protein genes from EpsteinBarr virus (EBV) and
herpesvirus saimiri (HVS), a New World monkey herpesvirus. Both EBV and HVS
are gamma herpesviruses associated with neoplastic disorders in humans
(EBV) (16) and nonhuman primates (HVS) (17). These results suggested that a
new human herpesvirus, now called KS-associated herpesvirus (KSHV or HHV8),
could be the KS agent.

The strong association between KS and KSHV was demonstrated by using
Southern hybridization and PCR to amplify a 233-bp DNA fragment (KS330 233)
internal to KS330Bam from all 25 intact and amplifiable DNA samples from KS
lesions that were examined (5). Ninety tissues from patients without AIDS
or KS were examined, and none showed evidence of KSHV infection.

Although controversy remains over the role of KSHV in KS (18-20),
epidemiologic and biologic studies strongly suggest that the agent is a
causal factor in KS (21). KSHV has now been identified in 211 (94.2%) of
224 KS lesions examined by a number of groups around the world, and the
virus is found in all forms of KS, AIDS-related and non-AIDS-related
(22-29). Two independent studies have shown that KSHV is present in
peripheral blood mononuclear cells of AIDS-KS patients before onset of KS
(30, 31), indicating that the virus is unlikely to be a “passenger virus”
in KS lesions. Further, the virus has been localized to KS tumor tissues by
semiquantitative PCR (25) and by PCR in-situ hybridization (26). The virus
has only been found in 8 (1.8%) of 449 solid tissues from control patients
without KS (5, 22, 25, 27, 28, 32, 38) and thus appears to be specifically
associated with KS. Although early reports suggested that the virus is
present in skin tumors from transplant patients without KS (19), subsequent
studies have not confirmed this finding (33, 34). One study has found
evidence of viral DNA by nested PCR, but not unnested PCR, in semen from
(23%) of healthy donors; these results suggest that the prevalence of
infection in North America may be higher than that indicated by tissue or
lymphocyte studies (35). This intriguing finding has not been reproduced by
other groups (23, 36); however, it should be explored in large rigorous
studies.

Lymphoproliferative disorders are common secondary malignancies in KS
patients with and without AIDS (37). KSHV has also been found in a rare
subset of AIDS-related, body cavity-based lymphomas, which are manifested
as primary effusions (5, 38). In these AIDS-related lymphomas, tumor cells
are coinfected with EBV (38, 39) but, unlike tumor cells in EBV-related
Burkitt's lymphomas, these cells do not exhibit c-myc gene rearrangement.
EBV-uninfected, KSHV-related body cavity-based lymphomas have also recently
been identified (40, 43). Another lymphoproliferative syndrome associated
with KS is Castleman's disease in both AIDS patients and HIV-seronegative
patients. KSHV has been found in AIDS-related multicentric Castleman's
disease lesions at a high copy number and less frequently in tissues from
Castleman's disease patients without AIDS (41).

Identifying KSHV DNA sequences by representational difference analysis also
led to the quick identification of an in vitro culture system for growing
the virus. KS330Bam and KS631Bam probes were used to identify a B-cell line
derived from a body cavity-based lymphoma that was stably infected with
both KSHV and EBV (42). Extended sequencing and transmission studies have
been performed with body cavity-base lymphoma cell lines, which clearly
define KSHV as a new human herpesvirus of the genus Rhadinovirus (6). This
has recently been confirmed by detecting herpesvirus particles in a body
cavity-based lymphoma cell line by electron microscopy (43).

Discovery of a continuously infected cell line has allowed the first
generation of serologic assays for KSHV antibodies to be developed (6).
Second-generation immunoblotting assays, which appear to be both sensitive
and specific for detecting KSHV antibodies, indicate that KS patients are
infected with the virus months to years before the disease develops and
that few North American blood donors are infected (S-J Gao, P.S. Moore,
unpublished observation). Thus, identifying a virus associated with KS by
using molecular approaches has rapidly led to new assays that use
traditional serologic techniques for detecting infection.

HHV-6 and Multiple Sclerosis

The difficulties of using representational difference analysis for new
pathogen identification are also illustrated by the search for the agent
that causes multiple sclerosis (MS) (44). An infectious cause for MS has
been proposed (45-47), and geographic and household clustering of cases
consistent with an infectious process have been shown (45, 48-50).
Challoner and colleagues used DNA from sclerotic plaques in brain tissues
from MS patients and performed representational difference analysis against
pooled DNA from peripheral blood leukocytes of healthy donors. Use of
pooled lymphocyte DNA in representational difference analysis opens the
possibility for examining rare banked tissues for which healthy control
tissues are not available. A 341-bp representational difference analysis
band was isolated from MS tissues with near sequence identity to a region
of the major DNA-binding protein gene of HHV-6 (44). Detecting exogenous
DNA in diseased tissue by this technique, however, does not exclude the
possibility that a commensal agent may be identified that is not the cause
of the disease being examined. HHV-6 is a neurotrophic virus present in
brain tissues from healthy control patients (51), and more than (70%) of
both MS patients and controls were positive for the representational
difference analysis sequence by PCR (44). Although case-patients and
controls cannot be differentiated by the presence or absence of these HHV-6
sequences, only MS tissues showed nuclear staining of oligodendrocytes
surrounding plaques when monoclonal antibodies against HHV-6 viral proteins
were used (44). There may be a subtle difference in tissue distribution for
the virus in MS patients not found in controls, or the HHV-6 variant B
group 2 infecting MS plaques may be particularly prone to generating the
autoimmune response seen in this disease.

GB Hepatitis Viruses

Since representational difference analysis relies on PCR amplification, the
technique is suited for detecting agents with a DNA-based genome; sequences
can be amplified by generating a cDNA intermediate, but detecting RNA
viruses is problematic since mRNA expression patterns differ between
tissues and are likely to give spurious representational difference
analysis bands. Simons and colleagues overcame this problem by passing an
agent associated with non-A, non-B (NANB) hepatitis through primates and
using cell-free extracts of primate plasma for representational difference
analysis and discovered two new human hepatitis viruses (52, 53).

A novel form of NANB hepatitis was first identified in a physician who
became ill with hepatitis, and the disease was transmitted to primates by
injecting serum from patients into the animals (54). Extensive studies
demonstrated that the virus, named the GB agent, was different from all
known human hepatitis viruses (hepatitis A [HAV], hepatitis B [HBV],
hepatitis C [HCV], and hepatitis E) (55-57). Total RNA from the samples was
reverse-transcribed to obtain cDNA by using cell-free preinfection and
acute-phase plasma from infected monkeys. After cDNA synthesis,
representational difference analysis was performed, and seven cDNA clones
were found to be specifically associated with this form of hepatitis.
Sequence analysis and comparison with other known hepatitis viruses
identified two unique flaviviruses, GBVA and GBVB, as the GB agents (52,
53, 58). A third virus, GBVC, was subsequently identified using the known
GBVA, GBVB, and HVC consensus sequences (52b and see below). GBVA and GBVC
are closely related phylogenetically. Cross-challenge experiments showed
that GBVC probably originated in human hepatitis patients, whereas GBVA and
B may be tamarin monkey viruses that were inadvertently passaged along with
the human virus (53, 58).

In addition to having difficulty in identifying RNA viruses,
representational difference analysis has several other limitations.
Polymorphic human DNA can be amplified through this technique (5), and not
all bands generated by it belong to the suspected agent. Representational
difference analysis may produce DNA fragments from an agent whose genome
has not been sequenced (e.g., most bacteria and fungi), and sequence
homology searches may be unable to distinguish the agent's genomic DNA
fragments from unsequenced human DNA. Only normally sterile site tissues
are appropriate for this technique because normal flora that differs
between tissue sites could result in spurious amplification. Finally, for
viruses with small genomes, multiple restriction digests may be required to
identify a unique restriction fragment of the appropriate size that can be
efficiently amplified. Although the technique has been successfully used by
several groups to identify polymorphic DNA, the procedure is complex and
not uniformly reproducible.

Consensus Sequence–Based PCR

Consensus sequence PCR relies on the use of highly conserved DNA sequences,
such as ribosomal RNA (rRNA) gene sequences from known organisms, to
amplify DNA from related organisms not yet discovered (59). This technique
is simple and extremely successful in identifying new human pathogens.
Subunit rRNA genes evolve in a relatively slow and uniform manner, which
makes these sequences extremely useful for establishing phylogenetic
relationships (59). By using conserved DNA sequences from bacterial 16S
rRNA genes, at least two new bacteria associated with human diseases have
been identified (60, 61), and PCR amplification of conserved hantaviral
capsid DNA sequences resulted in the rapid identification of a new
hantavirus associated with an outbreak of severe pulmonary disease (1).

Unlike using representational difference analysis, using consensus
sequences to amplify DNA requires knowledge of the suspected agent's
phylogenetic relationship to other organisms. The technique generally
should be used on normally sterile site tissues if sequences from normal
flora are also likely to be amplified. Although the lack of broadly
amplifiable consensus primers limits use of this technique to prokaryotic
and eukaryotic pathogens, consensus sequences are to likely exist among
many of the classes of viruses that eventually could be used in screening
panels of diseased tissues when a viral cause is suspected.

Bartonella: Bacillary Angiomatosis and Cat-Scratch Disease

Phylogenetic studies of bacteria have relied on comparisons of highly
conserved rRNA gene sequences present in prokaryotes (62). rRNA genes are
present in all living cells and contain regions of highly conserved
sequences with intervening variable regions. Conserved sequences can be
used to amplify and sequence the intervening variable regions by PCR. This
technique was exploited by Relman and colleagues to identify the bacillus
associated with bacillary angiomatosis (60).

Bacillary angiomatosis is an inflammatory vascular proliferative process
that affects the skin, lymph nodes, and visceral organs of AIDS patients
(63). Although bacilli can be identified by Warthin-Starry staining of
lesions (64, 65), the suspected causal organism was resistant to
cultivation by standard techniques. Consensus oligonucleotide primers
complementary to the 16S rRNA genes of eubacteria were used to amplify 16S
rRNA gene fragments directly from tissue samples (60). Phylogenetic
analysis of the amplified DNA sequence showed that the organism belonged to
the genus Rochalimaea (now renamed Bartonella) (60). Bartonella organisms
were also cultured from bacillary angiomatosis lesions (66) and blood from
bacteremic patients (67, 68), and serologic analyses have associated the
organisms with cat-scratch disease (69). Current evidence suggests that
bacillary angiomatosis in HIV-seropositive patients results from infection
with either B. quitana or a newly described Bartonella species, B. henselae
(68, 70), whereas cat-scratch disease is primarily caused by infection with
B. henselae (71) (for review, see [72]).

Whipple's Disease

Consensus sequence PCR was also used to identify an organism associated
with Whipple's disease, one of the most persistent mysteries in
microbiology (61). Whipple's disease is a systemic illness, first described
in 1907, characterized by arthralgias, diarrhea, abdominal pain, and weight
loss (73). Rod-shaped bacilli were identified histologically in Whipple's
disease lesions in the early 1960s (74), but the suspected bacteria were
not culturable by standard techniques (73). The agent was found to be a
gram-positive actinomycete (Tropheryma whippelii gen.nov.sp.nov.),
unrelated to any characterized species when a bacterial 165 rRNA sequence
was amplified and sequenced directly from infected tissue (61, 75).

Hantavirus Pulmonary Syndrome

Consensus PCR primers have been successfully used to identify and classify
bacteria, and similar techniques can be used to diagnose new viral agents.
In May 1993, an outbreak of unexplained acute respiratory illness with a
high death rate occurred in the southwestern United States (76). In the
initial phases of the investigation, the cause of the syndrome was not
clearly known to be infectious. Serologic tests quickly detected
crossreactive antibodies to known hantaviral antigens in the serum of
patients; these results suggested that a previously unrecognized hantavirus
was the cause of the disease (77). PCR primers, based on consensus
sequences within the G2 protein coding region of the M segment of the
genomes of known hantaviruses, were designed and used in nested
reverse-transcription PCR to amplify a short segment of the viral genome
from diseased tissues (1).

Sequence analysis of the PCR products showed that the amplicon differed
from the other known hantaviral sequences by least 30%, and phylogenetic
analysis demonstrated that the new virus is most closely related to
Prospect Hill hantavirus (1), a zoonotic hantavirus endemic in North
America (78). Hantavirus antigens have been detected in endothelial cells
from patients (1, 79), and virus particles have been identified in infected
pulmonary endothelial cells and macrophages (80). Deer mice (Peromyscus
maniculatus) are the primary host for the virus (1), and the virus has been
passaged through laboratory-bred deer mice and cultured in Vero E6 cells
(2). This newly recognized virus was originally named Muerto Canyon virus
(2) but has since been renamed Sin Nombre virus in light of nomenclatural
disputes regarding the appropriateness of a descriptive name.

Complementary cDNA Library Screening

Identification of HCV and HGV

A third major approach to new organism identification relies on screening
cDNA libraries made from diseased tissues by using hyperimmune serum from
specimens. This method was successfully used to identify the cause of most
cases of NANB hepatitis, HCV. When serologic tests for HAV and HBV became
available, it became clear that most cases of transfusion-associated
hepatitis in the United States were not caused by either virus (81, 82).
Conventional techniques failed to identify the agent responsible for most
cases of NANB hepatitis (83), despite evidence that the disease was caused
by a bloodborne, small, enveloped virus readily transmissible to
chimpanzees (84, 85).

Since the virus was presumed to be an RNA virus, Choo and colleagues made a
cDNA expression library from RNA isolated from an infected chimpanzee's
plasma (86). While the GBVA, B, C were identified by direct detection of
viral cDNA through representational difference analysis, HCV cDNA was
identified by immunologic detection of cDNA that was encoding viral
protein. Viral antigens expressed from the cDNA library were identified by
immunoscreening with convalescent-phase human sera. A cDNA clone was
isolated encoding an antigen that could be used to screen
convalescent-phase sera from patients with NANB hepatitis. Identification
of the clone rapidly led to the production of a recombinant antigen used
for serologic screening to detect specific antibodies in infected
chimpanzees and patients with hepatitis after transfusion. Extended
sequence analysis demonstrated that the agent, now known as HCV, is closely
related to the family Flaviviridae and is the major cause of NANB hepatitis
throughout the world (86, 87). Recently, the same approach was successfully
used to identify HGV, which has been found to be almost identical to GBVC,
the human hepatitis virus identified by representational difference
analysis (88). The potential role of HGV and GBVC in human disease still
remains uncertain (88b).

Using convalescent-phase sera to screen cDNA libraries from diseased
tissues is a novel method for pathogen identification. It is a potentially
useful technique for diseases in which well-defined convalescent-phase sera
are available, and it requires tissues with a high titers of the agent. On
the other hand, constructing and screening DNA libraries are laborious, and
crossreactive antigens are likely to be detected, especially for diseases
in which autoantibodies are common.

Future Directions

Nucleic acid database information is rapidly expanding for all classes of
organisms, and a significant fraction of the human genome has already been
sequenced. Even small laboratories can exploit new and relatively
inexpensive molecular biologic technologies to search for new pathogens.
Once a small, unique nucleic acid fragment from a pathogen has been
identified, nucleic acid detection and serologic assays can often be
rapidly developed to establish an etiologic link with disease.

New pathogens are likely to be identified by some of these molecular
biologic approaches. A number of diseases have long been suspected to have
an infectious cause and are appropriate candidates for these techniques.
The eventual identification of infectious etiologic agents of diseases such
as sarcoidosis (89), Kawasaki's disease (90), and type I diabetes mellitus
is likely. New techniques, such as arbitrarily primed PCR and phage display
libraries (91), have not been used for pathogen discovery but show promise
for expanding the repertoire of techniques available for identifying
unculturable agents. Three years passed between the initial descriptions of
AIDS and the identification of HIV (92). Use of molecular biologic
approaches could lead to rapid identification and control of the next
pandemic caused by a newly emergent infectious disease.

Acknowledgments

We thank Drs. David Relman, Stanford University, and Yuan Chang, Columbia
University, for critical advice regarding new pathogen identification and
comments on the manuscript, and Francis Zappa for help in preparing the
manuscript.

Dr. Gao received his Ph.D. in virology from the University of Bordeaux,
Bordeaux, France. He is a research scientist in the School of Public
Health, Columbia University, New York, where his work focuses on the
serology and seroepidemiology of Kaposi's sarcomaassociated herpesvirus
(KSHV). He also works on the induction of lytic phase replication of KSHV
as well as genes that may be involved in the latent phase replication of
the virus.

Dr. Moore is an associate professor of clinical public health at Columbia
University. His current research is focused on delineating the relationship
between Kaposi's sarcoma and KSHV (HHV8) and understanding the tumorigenic
potential of this virus. Along with Dr. Yuan Chang, his laboratory
codirector and wife, he is working on both epidemiologic and basic
virologic studies of KSHV and other new potential pathogens.

Address for correspondence:  Shou-Jiang Gao, Ph.D., School of Public
Health, Columbia University, P&S 14-442, 630 West-168th St., New York,
NY 10032; fax: 212-305-4548; e-mail: sg192@columbia.edu.

References

1. Nichol ST, Spiropoulou CF, Morzunov S, et al. Genetic identification of
a hantavirus associated with an outbreak of acute respiratory illness.
Science 1993;262:914-7.

2. Elliott LH, Ksiazek TG, Rollin PE, et al. Isolation of the causative
agent of hantavirus pulmonary syndrome. Am J Trop Med Hyg 1994;51:102-8.

3. Beral V, Peterman TA, Berkelman RL, Jaffe HW. Kaposi's sarcoma among
persons with AIDS: a sexually transmitted infection? Lancet 1990;335:123-8.

4. Beral V, Bull D, Darby S, et al. Risk of Kaposi's sarcoma and sexual
practices associated with faecal contact in homosexual or bisexual men with
AIDS. Lancet 1992;339:632-5.

5. Chang Y, Cesarman E, Pessin MS, et al. Identification of herpesviruslike
DNA sequences in AIDS-associated Kaposi's sarcoma. Science 1994;265:1865-9.

6. Moore PS, Gao SJ, Dominguez G, et al. Primary characterization of a
herpesviruslike agent associated with Kaposi's sarcoma. J Virol
1996;70:549-58.

7. Hill AB. Environment and disease: association or causation? Proc R Soc
Med 1965;58:295-300.

8. Evans AS. Causation and disease: a chronological journey. Am J Epidemiol
1978;108:249-58.

9. Fredricks DN, Relman DA. Sequence-based identification of microbial
pathogens: a reconsideration of Koch's postulates. Clin Microbiol Rev
1996;9:18-33.

10. Lisitsyn N, Lisitsyn N, Wigler M. Cloning the differences between two
complex genomes. Science 1993;259:946-51.

11. Lisitsyn NA, Lisitsina NM, Dalbagni G, et al. Comparative genomic
analysis of tumors: detection of DNA losses and amplification. Proc Natl
Acad Sci USA 1995;92:151-5.

12. Archibald CP, Schechter MT, Craib KJ, et al. Risk factors for Kaposi's
sarcoma in the Vancouver Lymphadenopathy-AIDS Study. J AIDS 1990;3 (suppl
1):S18-23.

13. Beral V, Bull D, Jaffe H, et al. Is risk of Kaposi's sarcoma in AIDS
patients in Britain increased if sexual partners came from United States or
Africa? Br Med J 1991; 302:624-5.

14. Archibald CP, Schechter MT, Le TN, Craib KJP, Montaner JSG,
O'Shaughnessy MV. Evidence for a sexually transmitted cofactor for
AIDS-related Kaposi's sarcoma in a cohort of homosexual men. Epidemiology
1992;3:203-9.

15. Peterman TA, Jaffe HW, Beral V. Epidemiologic clues to the etiology of
Kaposi's sarcoma. AIDS 1993;7:605-11.

16. Harrington DS, Weisenburger DD, Purtilo DT. Epstein-Barr
virus-associated lymphoproliferative lesions. Clin Lab Med 1988;8:97-118.

17. Rangan SR, Martin LN, Enright FM, Abee CR. Herpesvirus saimiri-induced
lymphoproliferative disease in howler monkeys. J Natl Cancer Inst
1977;59:165-71.

18. Roizman B. New viral footprints in Kaposi's sarcoma. N Engl J Med
1995;332:1227-8.

19. Rady PL, Yen A, Rollefson JL, et al. Herpesvirus-like DNA sequences in
non-Kaposi's sarcoma skin lesions of transplant patients. Lancet
1995;345:1339-40.

20. Levy JA. A new human herpesvirus: KSHV or HHV8? Lancet 1995;346:786.

21. Strathdee SA, Veugelers PJ, Moore PS. The epidemiology of
HIV-associated Kaposi's sarcoma: the unraveling mystery. AIDS
1995;10(suppl):S51-7.

22. Moore PS, Chang Y. Detection of herpesviruslike DNA sequences in
Kaposi's sarcoma lesions from persons with and without HIV infection. N
Engl J Med 1995;332:1181-5.

23. Ambroziak JA, Blackbourn DJ, Herndier BG, et al. Herpeslike sequences
in HIV-infected and uninfected Kaposi's sarcoma patients. Science
1995;268:582-3.

24. Collandre H, Ferris S, Grau O, Montagnier L, Blanchard A. Kaposi's
sarcoma and new herpesvirus. Lancet 1995;345:1043.

25. Dupin N, Grandadam M, Calvez V, et al. Herpesviruslike DNA in patients
with Mediterranean Kaposi's sarcoma. Lancet 1995;345:761-2.

26. Boshoff C, Schulz TF, Kennedy MM, et al. Kaposi's sarcoma-associated
herpesvirus (KSHV) infects endothelial and spindle cells. Nature Med
1995;1:1274-8.

27. Chang Y, Ziegler JL, Wabinga H, et al. Kaposi's sarcoma-associated
herpesvirus DNA sequences are present in African endemic and
AIDS-associated Kaposi's sarcoma. Arch Intern Med 1996;156:202-4.

28. Schalling M, Ekman M, Kaaya EE, Linde A, Bieberfeld P. A role for a new
herpesvirus (KSHV) in different forms of Kaposi's sarcoma. Nature Med
1995;1:707-8.

29. Su I-J, Hsu Y-S, Chang Y-C, Wang IW. Herpesviruslike DNA sequence in
Kaposi's sarcoma from AIDS and non-AIDS patients in Taiwan. Lancet
1995;345:722-3.

30. Whitby D, Howard MR, Tenant-Flowers M, et al. Detection of Kaposi's
sarcoma-associated herpesvirus (KSHV) in peripheral blood of HIV-infected
individuals predicts progression to Kaposi's sarcoma. Lancet
1995;364:799-802.

31. Moore PS, Kingsley L, Holmberg SD, et al. KSHV infection prior to onset
of Kaposi's sarcoma. AIDS 1996;10:175-80.

32. Boshoff C, Whitby D, Hatziionnou T, et al. Kaposi's sarcoma-associated
herpesvirus in HIVnegative Kaposi's sarcoma. Lancet 1995;345:1043-4.

33. Adams V, Kempf W, Schmid M, Müller B, Briner J, Burg G. Absence of
herpesvirus-like DNA sequences in skin cancers of non-immunosuppressed
patients. Lancet 1995;346:1715.

34. Boshoff C, Talbot S, Kennedy M, et al. New herpesvirus and
immunosuppressed skin cancers. Lancet 1996;347:338-9.

35. Lin JC, Lin SC, Mar EC, et al. Is Kaposi's sarcoma-associated
herpesvirus detectable in semen of HIV-infected homosexual men? Lancet
1995;346:1601-2.

36. Gupta P, Mandaleshwar DS, Rinaldo C, et al. Detection of Kaposi's
sarcoma herpesvirus DNA in semen of homosexual men with Kaposi's sarcoma.
AIDS (in press).

37. Biggar RJ, Curtis RE, Cote TR, Rabkin CS, Melbye M. Risk of other
cancers following Kaposi's sarcoma: relation to acquired immunodeficiency
syndrome. Am J Epidemiol 1994;139:362-8.

38. Cesarman E, Chang Y, Moore PS, Said JW, Knowles DM. Kaposi's
sarcoma-associated herpesviruslike DNA sequences are present in
AIDS-related body cavity based lymphomas. N Engl J Med 1995;332: 1186-91.

39. Knowles DM, Inghirami G, Ubriaco A, DallaFavera R. Molecular genetic
analysis of three AIDS-associated neoplasms of uncertain lineage
demonstrates their B-cell derivation and the possible pathogenetic role of
Epstein-Barr virus. Blood 1989;73:792-9.

40.Nador RG, Cesarman E, Knowles DM, Said JN. Herpeslike DNA sequences in a
body cavity-based lymphoma in an HIV-negative patient. N Engl J Med
1995;333:943.

41. Soulier J, Grollet L, Oskenhendler E, et al. Kaposi's
sarcoma-associated herpesviruslike DNA sequences in multicentric
Castleman's disease. Blood 1995;86:1276-80.

42. Cesarman E, Moore PS, Rao PH, Inghirami G, Knowles DM, Chang Y. In
vitro establishment and characterization of two AIDS-related lymphoma cell
lines containing Kaposi's sarcoma-associated herpesvirus-like (KSHV) DNA
sequences. Blood 1995;86:2708-14.

43. Renne R, Zhong W, Herndier B, et al. Lytic growth of Kaposi's
sarcoma-asociated herpesvirus (human herpesvirus 8) in culture. Nature Med
1996;2:342-6.

44. Challoner PB, Smith KT, D. PJ, et al. Plaque-associated expression of
human herpesvirus 6 in multiple sclerossis. Proc Natl Acad Sci USA
1995;92:7440-4.

45. Kurtzke JF. Some epidemiologic features compatible with an infectious
origin for multiple sclerosis. Int Arch Allergy Appl Immunol 1969;36:59-81.

46. Cook SD, Dowling PC. Multiple sclerosis and viruses: an overview.
Neurology 1980;30:80-91.

47. Whitaker JN, Kingsbury DW. Viruses and the pathogenesis of multiple
sclerosis. Trends Neurosci 1984;7:57-60.

48. Kurtzke JF. Familial incidence and geography in multiple sclerosis.
Acta Neurol Scand 1965;41:127-39.

49. Kurtzke JF. On the fine structure of the distribution of multiple
sclerosis. Acta Neurol Scand 1967;43:257-82.

50. Kurtzke JF. Geography in multiple sclerosis. J Neurol 1977;215:1-26.

51. Luppi M, Barozzi P, Maiorana A, Marasca R, Torelli G. Human herpesvirus
6 infection in normal human brain tissue. J Infect Dis 1994;169:943-4.

52. Simons JN, Pilot-Matias TJ, Leary TP, et al. Identification of two
flavivirus-like genomes in the GB hepatitis agent. Proc Natl Acad Sci USA
1995;92:3401-5.

52b. Simons JN, Leary TP, Dawson GJ, et al. Isolation of novel virus-like
sequences associated with human hepatitis. Nature Med 1995;1:564-9.

53. Schlauder GG, Dawson GJ, Simons JN, et al. Molecular and serologic
analysis in the transmission of GB hepatitis agents. J Med Virol
1995;46:81-90.

54. Deinhardt F, Holmes AW, Wolfe L, Junge U. Transmission of viral
hepatitis to nonhuman primates. Vox Sanguinis 1970;19:261-9.

55. Tabor E, Peterson DA, April M, Seeff LB, Gerety RJ. Transmission of
human non-A, non-B hepatitis to chimpanzees following failure to transmit
GB agent hepatitis. J Med Virol 1980;5:103-8.

56. Feinstone SM, Alter HJ, Dienes HP, et al. Non-A, non-B hepatitis in
chimpanzees and marmosets. J Infect Dis 1981;144:588-98.

57. Karayiannis P, Petrovic LM, Fry M, et al. Studies of GB hepatitis agent
in tamarins. Hepatology 1989;9:186-92.

58. Zuckerman AJ. The new GB hepatitis viruses. Lancet 1995;345:1453-4.

59. Relman DA. The identification of uncultured microbial pathogens. J
Infect Dis 1993;168:1-8.

60.Relman DA, Loutit JS, Schmidt TM, Falkow S, Tompkins LS. The agent of
bacillary angiomatosis: an approach to the identification of uncultured
pathogens. N Engl J Med 1990;323:173-80.

61.Relman DA, Schmidt TM, MacDermott RP, Falkow S. Identification of the
uncultured bacillus of Whipple's disease. N Engl J Med 1992;327:293-301.

62.Lane DJ, Stahl DA, Olsen GJ, Heller DJ, Pace NR. Phylogenetic analysis
of the genera Thiobacillus and Thiomicrospira by 5S rRNA sequences. J
Bacteriol 1985;163:75-81.

63.Stoler MH, Bonfiglio TA, Steigbigel RT, Pereira M. An atypical
subcutaneous infection associated with acquired immune deficiency syndrome.
Am J Clin Pathol 1983;80:714-8.

64. Angritt P, Tuur SM, Macher AM, et al. Epithelioid angiomatosis in HIV
infection: neoplasm or cat-scratch disease? Lancet 1988;1:996.

65. LeBoit PE, Berger TG, Egbert BM, et al. Epithelioid haemangioma-like
vascular proliferation in AIDS: manifestation of cat-scratch disease
bacillus infection? Lancet 1988;1:960-3.

66. Koehler JE, Quinn FD, Berger TG, LeBoit PE, Tappero JW. Isolation of
Rochalimaea species from cutaneous and osseous lesions of bacillary
angiomatosis. N Engl J Med 1992;327:1625-31.

67. Slater LN, Welch DF, Hensel D, Coody DW. A newly recognized fastidious
gram-negative pathogen as a cause of fever and bacteremia. N Engl J Med
1990;323:1587-93.

68. Regnery RL, Anderson BE, Clarridge JE, RodriguezBarradas MC, Jones DC,
Carr JH. Characterization of a novel Rochalimaea species, R. henselae sp.
nov., isolated from blood of a febrile, human immunodeficiency
virus-positive patient. J Clin Microbiol 1992;30:265-74.

69. Regnery RL, Olson JG, Perkins BA, Bibb W. Serological response to
“Rochalimaea henselae” antigen in suspected catscratch disease. Lancet
1992;339:1443-5.

70. Welch DF, Pickett DA, Slater LN, Steigerwalt AG, Brenner DJ.
Rochalimaea henselae sp. nov., a cause of septicemia, bacillary
angiomatosis, and parenchymal bacillary peliosis. J Clin Microbiol
1992;30:275-80.

71. Koehler JE, Glaser CA, Tappero JW. Rochalimaea henselae infections: a
new zoonosis with the domestic cat as reservoir. JAMA 1994;271:531-5.

72.Regnery RL, Tappero J. Unravelling mysteries associated with cat-scratch
disease, bacillary angiomatosis, and related syndromes. EID 1995;1:16-21.

73. Dobbins WO III. Whipple's disease. Springfield, IL: Charles C Thomas,
1987.

74. Chears WCJ, Ashworth CT. Electron microscopic study of the intestinal
mucosa in Whipple's disease: demonstration of encapsulated bacilliform
bodies in the lesion. Gastroenterology 1961;41:129-38.

75. Wilson KH, Blitchington R, Frothingham R, Wilson JA. Phylogeny of the
Whipple's disease-associated bacterium. Lancet 1991;338:474-5.

76. Centers for Disease Control and Prevention. Outbreak of acute
illness—southwestern United States. MMWR 1993;42:421-4.

77. Centers for Disease Control and Prevention. Update: outbreak of
hantavirus infection—southwestern United States. MMWR 1993;42:441-3.

78. Leduc JW. Epidemiology of Hantaan and related viruses. Lab Anim Sci
1987;37:413-8.

79. Ksiazek TG, Peters CJ, Rollin PE, et al. Identification of a new North
American hantavirus that causes acute pulmonary insufficiency. Am J Trop
Med Hyg 1995;52:117-23.

80. Zaki SR, Greer PW, Coffield LM, et al. Hantavirus pulmonary syndrome.
Pathogenesis of an emergng infectious disease. Am J Pathol 1995;146:552-79.

81. Feinstone SM, Kapikian AZ, Purcell RH, Alter HJ, Holland PV.
Transfusion-associated hepatitis not due to viral hepatitis type A or B. N
Engl J Med 1975;292: 767-70.

82. Knodell RG, Conrad ME, Dienstag JL, Bell CJ. Etiological spectrum of
post-transfusion hepatitis. Gastroenterology 1975;69:1278-85.

83. Shih JW, Mur JI, Alter HJ. Non-A, non-B hepatitis: advances and
unfulfilled expectations of the first decade. Prog Liver Dis 1986;8:433-52.

84. Bradley DW, McCaustland KA, Cook EH, Schable CA, Ebert JW, Maynard JE.
Posttransfusion non-A, non-B hepatitis in chimpanzees: physicochemical
evidence that the tubule-forming agent is a small, enveloped virus.
Gastroenterology 1985;88:773-9.

85. He LF, Alling D, Popkin T, Shapiro M, Alter HJ, Purcell RH. Determining
the size of non-A, non-B hepatitis virus by filtration. J Infect Dis
1987;156:636-40.

86. Choo QL, Kuo G, Weiner AJ, Overby LR, Bradley DW, Houghton M. Isolation
of a cDNA clone derived from a blood-borne non-A, non-B viral hepatitis
genome. Science 1989;244:359-62.

87. Kuo G, Choo QL, Alter HJ, et al. An assay for circulating antibodies to
a major etiologic virus of human non-A, non-B hepatitis. Science
1989;244:362-4.

88. Linnen J, Wages Jr J, Zhang-Keck ZY, et al. Molecular cloning and
disease association of hepatitis G virus: a transfusion-transmissible
agent. Science 1996;271: 505-8.

88b. Alter HJ. The cloning and clinical implications of HGV and HGBV-C. N
Engl J Med 1996;334:1536-7.

89. Graham DY, Markesich DC, Kalter DC, Moss MT, Hermon-Taylor J,
El-Zaatari FA. Mycobacterial aetiology of sarcoidosis. Lancet
1992;340:52-3.

90. Rowley AH, Wolinsky SM, Relman DA, et al. Search for highly conserved
viral and bacterial nucleic acid sequences corresponding to an etiologic
agent of Kawasaki disease. Pediatr Res 1994;36:567-71.

91. Scott JK, Smith GP. Searching for peptide ligands with an epitope
library. Science 1990;249:386-90.

92. Barre-Sinoussi F, Chermann JC, Rey F, et al. Isolation of a
T-lymphotropic retrovirus from a patient at risk for acquired immune
deficiency syndrome (AIDS). Science 1983;220:868-71.