THE ETIOLOGY AND PATHOGENESIS OF GREEN TURTLE
LAWRENCE HENRY HERBST
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
Lawrence Henry Herbst
To my wife, Maria, and my two children, Thomas and Lisa,
who have supported me as I have pursued my dreams and who
have sacrificed the most to allow me that luxury. To my
parents who taught me the importance of family and the value
in doing what you love.
I want to express my deepest appreciation to Dr. Paul A.
Klein, chairman of my graduate committee, for the advice,
assistance, encouragement, support, patience, and friendship
that he has shown me throughout this project. I am also very
grateful to Dr. Elliott R. Jacobson, cochairman of my
graduate committee, for giving me the opportunity and
encouraging me to conduct this project. Special thanks also
go to Dr. Alan B. Bolten, Dr. Ammon B. Peck, Dr. Sheldon M.
Schuster, and Dr. John P. Sundberg, members of my graduate
committee, for their advice, assistance, and support.
A substantial part of this project would not have been
possible without the facilities and dedicated support
provided by Richie Moretti and Tina Brown of The Turtle
Hospital, Marathon, FL. Their sincere concern for the
environment and for endangered species has been amply
demonstrated by their commitment of time, labor, and money to
provide for the housing, husbandry, and rehabilitation of
marine turtles. I also thank veterinarians, Dr. Lisa Bramson
and Dr. Beatrice Lopez, for providing clinical care of
research turtles in my absence.
Many others have contributed in various ways to the
success of this research project. I especially thank George
Balazs (National Marine Fisheries Service, Southwest
Fisheries Laboratory, Honolulu, HI) for his continued
enthusiasm and interest in the fibropapilloma problem in
green turtles. Through his efforts, significant funding for
these studies were made available. I also thank Earl
Possardt, Ren Lohoefener, and Sandra McPherson of the U.S.
Fish & Wildlife Service for their support of this project.
The Florida Department of Environmental Protection (DEP)
provided permits to conduct these studies. In addition, the
cooperation and assistance of DEP personnel, especially
Barbara Schroeder, Allen Foley, Carrie Crady, and Blair
Witherington, in providing information about and access to
free-ranging or stranded turtles was extremely helpful.
Special thanks go to Dr. Llewellyn Ehrhart, University of
Central Florida, Orlando, FL, and his students and field
assistants, especially Dean Bagley and Steven Johnson, for
collecting blood, biopsy samples, and eggs. Eric Martin and
Barbara Schroeder also provided turtle eggs. Jim and Fern
Wood of Cayman Turtle Farm, Ltd, Grand Cayman, British West
Indies, made captive turtles available for antibody studies
and carried out immunizations. USAir Corporation transported
hatchlings from Gainesville to Marathon free of charge. Sea
World, Orlando, FL, provided access to turtles in their care
for blood sampling. Dr. Karen Bjorndal and Dr. Alan Bolten of
the Archie Carr Center for Sea Turtle Research provided
biological samples and allowed me to incubate eggs and house
hatchlings in their laboratory. Alan Bolten also provided
valuable assistance in negotiating the paperwork associated
with permits and work orders. I thank the personnel of the
Hybridoma Core Laboratory and Immunological Analysis Core,
Interdisciplinary Center for Biotechnology Research (ICBR),
especially Diane Duke, Linda Green, Cathy McKenna, and
Isabella Schumacher for their assistance. I also thank Dr.
Ratna Chakrabarty (Gene Expression Core, ICBR) for performing
the preliminary differential message display analysis. Dr.
Brian Gray (Cytogenetics Laboratory) karyotyped cell lines.
Dr. Ellis Greiner (Department of Pathobiology) sorted and
identified trematode parasites from green turtles. Dr. Bruce
Homer (Department of Pathobiology) allowed me to work at
times in his laboratory. Special thanks are due to Betty Hall
for teaching me immunohistochemistry techniques. Dr. Leonard
Shultz and Dr. John P. Sundberg (The Jackson Laboratory, Bar
Harbor, ME) provided immunodeficient mice for tumorigenicity
This project was supported by grants from SAVE-A-TURTLE,
Islamorada, FL, a joint contract from the U.S. Fish &
Wildlife Service, Department of the Interior and the National
Marine Fisheries Service, Southwest Fisheries Science Center,
NOAA, Department of Commerce (RWO No. 96), and a training
fellowship from the National Institutes of Health (National
Center for Research Resources RR07001).
TABLE OF CONTENTS
LIST OF TABLES
LIST OF FIGURES
. . . x ii
. . . x iv
. . . xv ii
2 REVIEW OF LITERATURE ON GREEN TURTLE
FIBROPAPILLOMATOSIS . .
Historical Perspective . .
Description of GTFP . .
Gross Pathology . .
Histopathology . .
Epizootiology of GTFP . .
Geographic Distribution . .
Prevalences . .
Seasonality . .
Demographic Patterns . .
Habitat Associations . .
Clinical Course, Morbidity, and Mortality
Possible Causes of GTFP . .
Environmental Factors . .
Ultraviolet light . .
Chemical contaminants .
Infectious Diseases . .
Viruses . .
Bacteria . .
Metazoan parasites . .
Genetic Factors .
Immune Dysfunction in GTFP Pathogenesis .
Conclusion . . .
3 DEVELOPMENT OF MONOCLONAL ANTIBODIES FOR DETECTING
CLASS-SPECIFIC IMMUNE RESPONSES IN THE GREEN
TURTLE . . .
: : : : :
Introduction . .
Materials and Methods . .
Turtle Plasma Samples and Turtle
Immunizations . .
Preparation of Turtle Immunoglobulins
Hybridoma Production . .
Monoclonal Antibody Screening Protocols
Monoclonal Antibody Purification
and Biotinylation . .
Cross Species Reactivity of Monoclonal
Antibodies . .
Verification of Monoclonal Antibody and
Turtle Antibody Specificity .
Results . .
Immunoglobulin Purification .
Production and Characterization of
Monoclonal Antibodies .
Verification of Monoclonal Antibody
Specificity . .
Discussion . .
4 EXPERIMENTAL TRANSMISSION OF GREEN TURTLE
FIBROPAPILLOMATOSIS . .
Introduction . .
Materials and Methods . .
Animal Housing and Maintenance .
First Transmission Study (1991) .
Materials for inoculation (1991)
Experimental turtles (1991)
Experimental treatments (1991)
Second Transmission Study (1992)
Material for inoculation (1992)
Experimental turtles and
treatments (1992) .
Third Transmission Study (1993) .
Material for inoculation (1993)
Experimental turtles (1993)
Experimental treatments (1993)
Histopathology . .
Transmission Electron Microscopy
Negative Staining Electron Microscopy
Results . .
First Transmission Study (1991) .
Second Transmission Study (1992)
Third Transmission Study (1993) .
Time to Tumor Development .
Histopathology of Experimentally
Induced Tumors . .
Transmission Electron Microscopy
Negative Staining Electron Microscopy
Discussion . .
Experimental Evidence for an Infectious
Etiology . .
Role of Spirorchid Ova . .
The GTFP-Associated Herpesvirus .
Attempts to Identify Viruses in
Donor Material . .
Variation in Transmission Success .
Environmental Influences on Tumor
Conclusion . .
5 INITIAL ATTEMPTS TO ISOLATE AND CHARACTERIZE
THE ETIOLOGIC AGENT . .
Introduction . .
Materials and Methods . .
Virus Isolation in Cell Culture .
Virus Extraction from Tumor Homogenates
Characterization Experiments .
Materials for inoculation .
Experimental turtles .
Transmission experiments with
tumor preparations .
Results . . .
Virus Culture . .
Virus Purification from Tumor
Homogenate . .
Characterization Experiments ..
Discussion . . .
Virus Culture . .
Virus Purification . .
Characterization Experiments .
6 HISTOPATHOLOGIC AND IMMUNOHISTOCHEMICAL
EVIDENCE FOR A VIRAL (HERPESVIRUS) ETIOLOGY
Introduction . .
Materials and Methods . .
Histopathology . .
Immunohistochemistry for Detection of
Herpesvirus Antigens .
Immunofluorescence Staining for
Immunoglobulin Deposition .
Results . .
Cutaneous Fibropapillomas of Florida and
Hawaiian Turtles . .
Epidermal folding .
Pathologic changes in the dermis
Pathologic changes in the epidermis 154
Inflammation . 163
Potential pathogens . .. 167
Experimentally Induced Cutaneous Tumors 173
Differences between spontaneous and
experimental tumors . 175
Early pathologic changes in
experimentally induced tumors 178
Herpesvirus Immunohistochemistry .. 180
Immunofluorescence for Antibody Deposition 181
Visceral Tumors . .. 181
Discussion . . 184
Florida Versus Hawaiian GTFP . 184
Spontaneous Versus Experimentally
Induced GTFP . 186
Pathogenesis of Tumor Progression .. 187
Pathogenesis of Dermal-Epidermal Clefts 190
Possible Routes of Dissemination .. 192
Can Herpesvirus Explain GTFP Pathology? 192
Can Papillomavirus Explain GTFP Pathology? .194
7 SEROLOGIC EVIDENCE FOR THE ASSOCIATION BETWEEN
SPIRORCHIDIASIS, HERPESVIRUS INFECTION, AND
FIBROPAPILLOMATOSIS . .. 196
Introduction . . 196
Materials and Methods . .. 198
Plasma Samples . .... ..198
Detection of Antibodies to the
GFTP-Associated Herpesvirus .. 199
Detection of Antibody Reactivity
to Spirorchid Trematodes .. 201
Results . . 204
GTFP-Associated Herpesvirus .. 204
Spirorchidiasis . 207
Discussion . . 218
GTFP-Associated Herpesvirus .. 218
Spirorchidiasis . .. 220
Conclusion . . 225
8 ESTABLISHMENT OF CELL LINES AND PRELIMINARY STUDIES
OF DIFFERENTIAL GENE EXPRESSION BETWEEN NORMAL
AND TUMOR-DERIVED FIBROBLASTS .. 226
Introduction . .. 226
Materials and Methods . .. 227
Cell Lines . . 227
In Vitro Assays .. . .. .. 229
Tumorigenicity Assay (In Vivo) 229
Detection of GTFP Cell-Associated Antigens 232
Detection of GTFP-Associated Changes in
Gene Expression . ... 233
Results . . 235
Cell Lines and In Vitro Characteristics 235
Tumorigenicity of GTFP-Derived Cell Lines 236
GTFP Cell-Associated Antigens .. 240
GTFP-Associated Changes in Gene Expression 244
Discussion . . 244
9 SUMMARY AND RECOMMENDATIONS . ... 251
REFERENCES . . ... 258
BIOGRAPHICAL SKETCH . ... 284
LIST OF TABLES
Table 2-1. Prevalences of GTFP among different
populations of free-ranging green turtles 19
Table 3-1. List of monoclonal antibodies specific for
green turtle immunoglobulins . .. 72
Table 4-1. Free-ranging green turtles with cutaneous
fibropapillomatosis used as fibropapilloma donors 93
Table 4-2. Fibropapilloma development at inoculation
sites in green turtles treated with filtered cell-
free fibropapilloma extracts . .. 101
Table 4-3. Frequency of fibropapilloma development at
injection and scarification sites in recipient
green turtles . ... 104
Table 5-1. Free-ranging green turtles with cutaneous
fibropapillomatosis used as fibropapilloma donors 125
Table 5-2. Summary of characterization transmission
experiments . . ... 132
Table 5-3. The effect of chloroform treatment on the
infectivity of the GTFP agent ... .134
Table 5-4. Partitioning of GTFP agent infectivity by
ultracentrifugation . ... .135
Table 6-1. Comparison of the gross and histologic
features of fibropapilloma biopsies from free-
ranging Florida and Hawaiian green turtles .. 150
Table 6-2. Prevalence estimates of spirorchidiasis and
herpesvirus infection among green turtle population
samples . . 169
Table 6-3. Comparison of the gross and histologic
features of spontaneous and experimentally induced
fibropapillomas . ... 176
Table 7-1. Seroconversion to anti-herpesvirus
immunoreactivity among green turtles with
experimentally induced GTFP .
Table 7-2. Post-mortem diagnosis of trematode infections
in green turtles used as controls in ELISA
development . . 208
Table 8-1. Tumorigenicicty of fibroblast lines derived
from green turtles . .
LIST OF FIGURES
Figure 2-1. Cutaneous fibropapillomatosis in the
green turtle, Chelonia mydas . .
Figure 2-2. Ocular fibropapillomatosis in green
turtles, Chelonia mda . .
Figure 2-3. Examples of visceral tumors found in
some green turtles with severe cutaneous
fibropapillomatosis . .
Figure 2-4. Variation in histologic appearance of
cutaneous fibropapillomatosis . .
Figure 2-5. Circumtropical distribution of green
turtle fibropapillomatosis (GTFP) .. 17
Figure 3-1. Fractionation of green turtle
immunoglobulins by gel filtration chromatography 67
Figure 3-2. Affinity purified turtle anti-DNP
antibody chains . . 69
Figure 3-3. Molecular masses of native 5.7S and
7S turtle immunoglobulins . ... 70
Figure 3-4. Sandwich ELISA demonstrating that putative
heavy chain specific Mabs bind plasma proteins with
immunoglobulin light chains . ... 74
Figure 3-5. Development of antibody responses to DNP
with time in 2 chronically immunized turtles
Figure 3-6. Inhibition of turtle anti-DNP antibody
activity by soluble antigen . .
Figure 4-1. Experimentally induced cutaneous
fibropapillomas in green turtles . .
Figure 4-2. Time course for experimental fibropapilloma
induction during the 1993 transmission study .
Figure 4-3. Experimentally induced fibropapilloma
showing characteristic benign epidermal hyperplasia
on broad fibrovascular stalks . .
Figure 4-4. Cytopathology in the epidermis of
experimentally induced green turtle
fibropapillomas . . .
Figure 4-5. Herpesvirus-like particles in experimentally
induced fibropapillomas . .
Figure 5-1. Subcellular particles identified in pooled
isopycnic gradient fractions prepared from
transmission-positive filtered tumor homogenate
Figure 6-1. Detection of herpesvirus antigens in green
turtle fibropapillomas by immunohistochemistry
Figure 6-2. Variation in the distribution of pigement
cells in cutaneous fibropapillomas
Figure 6-3. Typical cutaneous fibropapilloma in
the green turtle . .
Figure 6-4. Variation in the degree of epidermal
hyperplasia among cutaneous fibropapillomas
Figure 6-5. Degenerative changes observed in basal
epidermal cells in fibropapillomas .
Figure 6-6. Dermal-epidermal clefts in cutaneous
fibropapillomas . .
Figure 6-7. Degenerative changes observed in the
stratum spinosum of cutaneous fibropapillomas
Figure 6-8. Foreign body granulomas found in
fibropapillomas of free-ranging green turtles
Figure 6-9. Lymphocytic infiltrates observed in
green turtle fibropapillomas .
Figure 6-10. Herpesvirus-like intranuclear inclusions
Figure 6-11. Spirorchid trematode eggs found in
fibropapillomas of free-ranging green turtles
Figure 6-12. Bacteria, fungi, and algae found on
cutaneous fibropapillomas . .
Figure 6-13. Metazoan epibionts . .
Figure 6-14. Early pathologic changes in experimentally
induced fibropapillomas . .
Figure 6-15. Histologic features of visceral tumors
found in some green turtles with cutaneous GTFP
Figure 7-1. Spirorchid trematode ova recovered from
Florida green turtles . .
Figure 7-2. Plasma 7S IgY antibody responses of
controlturtles to crude adult spirorchid
antigen preparations . .
Figure 7-3. Relative frequencies of negative,
intermediate, and positive 7S IgY antibody
responses to Learedius learedii crude antigen
in turtle plasma samples from two sites .
Figure 7-4. Relative frequencies of GTFP negative
reef and lagoon green turtles having negative,
intermediate, and positive 7S IgY antibody
responses to Learedius learedii crude antigen
in plasma samples from GTFP-free turtles
from two sites . .
Figure 7-5. Relative frequencies of negative,
intermediate, and positive 7S IgY antibody
responses to Learedius learedii crude antigen in
plasma samples from GTFP-positive and GTFP-negative
Indian River lagoon turtles . .
Figure 8-1. Green turtle fibroblast cultures .
Figure 8-2. Tumorigenicity of GTFP-derived fibroblasts
in immunodeficient mice . .
Figure 8-3. Immunohistochemical detection of green
turtle fibroblasts in mouse ear fibromas .
Figure 8-4. Chromosomes of fibroblasts derived from
mouse ear fibromas . .
Figure 8-5. Differences in gene expression between
normal skin-derived and tumorigenic GTFP-derived
fibroblasts . . .
Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy
THE ETIOLOGY AND PATHOGENESIS OF GREEN TURTLE
Lawrence Henry Herbst
Chairperson: Paul A. Klein
Major Department: Department of Pathobiology
Green turtle fibropapillomatosis (GTFP) is a threat to
populations of Chelonia mydas worldwide. This project
attempted to characterize the etiology and to describe the
pathogenesis of GTFP. Transmission studies showed that tumors
could be induced in recipient turtles by inoculation with
twice frozen and thawed cell-free homogenates prepared from
spontaneous tumors. Tumors were not induced by inoculation
with intact spirorchid ova nor were spirorchid ova found in
any experimentally induced tumors. Oncogenicity of tumor
homogenates passed through 0.45 pm but not 0.2 im filters,
and was destroyed by chloroform. Some spontaneous and
experimentally induced tumors had epidermal eosinophilic
intranuclear inclusions, which contained herpesvirus-like
particles. Attempts to culture this virus on 2 reptilian cell
lines were unsuccessful. Particles resembling herpesvirus
were found in pooled isopycnic gradient fractions of one
transmission-positive tumor preparation, but were not
tumorigenic. Green turtle antibody class-specific monoclonal
antibodies, developed for the detection of turtle antibody
responses to putative GTFP agents, were used with a proven
herpesvirus-specific turtle antiserum, to demonstrate
herpesvirus antigens in spontaneous and induced tumors.
Tissue sections containing herpesvirus were also used to
screen plasma samples for antibody reactivity to herpesvirus
antigens by immunohistochemistry. Antibody reactivities to
herpesvirus developed in all experimental transmission-
positive turtles, but not in controls or transmission-
negatives. A strong association between antibody reactivity
to herpesvirus and clinical GTFP was also found in free-
ranging turtles. In contrast, antibody reactivity to
spirorchid trematodes was not associated with clinical GTFP.
The transformed phenotype of GTFP-derived fibroblast cultures
was demonstrated using tumorigenicity assays and preliminary
studies showed differences in mRNA expression between matched
pairs of normal skin- and GTFP-derived cell lines. Although
the pathogenesis of GTFP can be explained by herpesvirus,
proof that herpesvirus causes GTFP will require reproduction
of the disease in turtles with purified virus, or
demonstration of herpesviral gene sequences among these
differentially expressed messages in GTFP cell lines and in
transmission positive tumor homogenates, that can transform
normal fibroblasts to the tumorigenic phenotype.
Reports of neoplasia in chelonians are relatively
uncommon (Billups & Harshbarger, 1976; Jacobson, 1980; 1981a;
Machotka, 1984). For example, 24 neoplastic conditions in 19
turtle species, mostly represented by single case reports,
were compiled by Machotka (1984). Cutaneous papillomas,
fibromas, and fibropapillomas in green turtles, Chelonia
mydas, however, were reported commonly (Machotka, 1984).
These three proliferative lesions in Chelonia mydas are the
hallmarks of Green Turtle Fibropapillomatosis (GTFP) and the
number of observed cases has continued to increase while only
2 new case reports of neoplasia in other turtle species have
subsequently been published (Frye et al., 1988; Machotka et
al., 1992). Recent documented increases in GTFP prevalence
and the spread of GTFP to locations where it had not been
observed previously make GTFP the most common neoplastic
disease of reptiles and a significant threat to endangered
green turtle populations. Consequently, research to determine
the cause of GTFP and find ways to reduce the impact of this
disease has been listed as a priority in recovery plans for
green turtles (Balazs et al., 1990; National Marine Fisheries
Service & U.S. Fish and Wildlife Service, 1991).
More recently, fibropapilloma-like lesions have been
reported in other marine turtle species, including loggerhead
turtles, Caretta caretta (Llewellyn Ehrhart, University of
Central Florida, Orlando, FL 32816, pers. comm.; Barbara
Schroeder, Florida Marine Research Institute, Tequesta, FL
33469, pers. comm.), olive ridley turtles, Leoidochelvs
olivacea (Any Chaves, Universidad de Costa Rica, San Jose,
Costa Rica, pers. comm.; Pamela Plotkin, Texas A&M
University, College Station, TX 77843, pers. comm.) and
flatback turtles, Natator depressus (Limpus & Miller, 1994),
raising concerns about the potential impact of these diseases
on all marine turtle populations.
The purpose of this research project has been to
identify the cause of GTFP, to lay the groundwork for
understanding the pathogenesis of this disease, and to begin
to develop practical diagnostic tests for use in management
applications and in ecological studies (epizootiology) of
GTFP. This research has addressed the important questions
that will lead to understanding of the impact of this disease
on worldwide green turtle populations.
The specific objectives were as follows:
(1) To review current knowledge about GTFP and evaluate
the various proposed hypotheses about its etiology (Chapter
(2) To begin to develop the immunological tools needed
to study the immune response to fibropapilloma cells or to
putative etiologic agents. The initial focus was to produce
monoclonal antibodies specific for the known classes of green
turtle immunoglobulins (Chapter 3).
(3) To determine whether GTFP can be transmitted
experimentally and is therefore caused by an infectious
(transmissible) agent (Chapter 4).
(4) To identify and characterize the etiologic agent
(5) To rule out alternative infectious etiologies
(Chapters 4 & 5).
(6) To begin to describe the pathogenesis of GTFP based
on histopathology, experimental findings, and identified
serologic and epizootiologic associations (Chapters 6 & 7).
(7) To develop the techniques to distinguish cultured
GTFP-derived, i.e., transformed fibroblasts from normal
fibroblasts, as a basis for in vitro studies on the molecular
mechanisms of fibroblast proliferation, the most salient
feature of GTFP. In addition, to search for GTFP cell-
specific antigens or genes in these cell lines for use in
diagnostic test development (Chapter 8).
REVIEW OF LITERATURE ON GREEN TURTLE FIBROPAPILLOMATOSIS
Cutaneous papillomas, fibromas, and fibropapillomas were
first described by Smith and Coates (1938) in a captive green
turtle, Chelonia mydas, at the New York Aquarium that had
been captured near Key West, Florida, two years previously.
Two other green turtles and 2 loggerheads that were housed
with this animal did not have lesions. Smith and Coates
(1938) also found fibropapillomas in 3 of 200 free-ranging
green turtles (27-91 kg) that were captured off of Key West.
That same year, Luck6 described similar tumors from a green
turtle caught off Cape Sable, Florida (Luck6, 1938). Masses
were located on the tail, flippers, axillae, neck, eyelids,
and corneas. Schlumberger and Luck6 (1948) subsequently
described fibropapillomas from 3 Florida green turtles and
found numerous fibrous masses within the lungs of one turtle.
In 1958, Hendrickson noted the occasional occurrence of
fibrous masses on nesting females in Sarawak and Malaya
(Hendrickson, 1958). The first confirmed case of GTFP in
Hawaii occurred in 1958 and was a juvenile green turtle
captured by local fisherman in Kaneohe Bay, Oahu (Balazs,
1991). A survey of local fishermen conducted by Balazs (1991)
suggests that GTFP was rare to nonexistent prior to this.
Since this first report, green turtles with fibropapillomas
have been reported with increasing frequency from Hawaii
(Balazs, 1991; George Balazs, National Marine Fisheries
Service, Southwest Fisheries Center, Honolulu, Hawaii 96822,
pers. comm.). In 1980 an outbreak of fibropapillomatosis
occurred in a breeding group of adult green turtles at Cayman
Turtle Farm, Ltd, Grand Cayman, British West Indies
(Jacobson, 1981b; Jacobson et al., 1989). The outbreak began
in wild caught adults but subsequently developed over several
years in farm raised turtles as well. Ehrhart (1991)
documented the first cases of GTFP in the Indian River
Lagoon, Florida, in 1982. Netting surveys within the northern
portion of the Indian River Lagoon system (Mosquito Lagoon)
had been conducted since 1977 without encountering any green
turtles with fibropapillomas. However, when the study area
was shifted to the central portion of the system (Indian
River) in 1982, affected turtles were encountered
immediately. A review of late 19th century accounts of the
Florida east coast green turtle fishery and of reports on
Indian River Lagoon turtles published between 1978 and 1983
failed to yield any record of GTFP prior to this (Ehrhart,
1991; Ehrhart et al., 1986). Continued monitoring at this
site since 1982 has revealed GTFP prevalences around 50%.
Description of GTFP
Green turtle fibropapillomatosis (GTFP) is characterized
by single to multiple raised cutaneous masses ranging from
0.1 cm to greater than 30 cm in diameter. Individual masses
may be either verrucous or smooth and either sessile or
pedunculated. Large masses are often ulcerated. Cutaneous
fibropapillomas are usually found on the soft skin but may be
found anywhere on the turtle's body, including carapace and
plastron. Common sites for GTFP are the flippers, neck, chin,
inguinal and axillary regions, and tail base (Figure 2-1).
Ocular GTFP is common, with masses arising from the bulbar
conjunctiva, limbus, cornea, or mucocutaneous junction of the
eyelids (Brooks et al., 1994; Jacobson et al., 1989; Luck6,
1938; Smith & Coates, 1938) (Figure 2-2). Tumor pigmentation
is usually related to the pigmentation of the skin at the
site of origin.
Visceral tumors (Figure 2-3) have been found at necropsy
in some green turtles with cutaneous fibropapillomatosis
(Jacobson et al., 1991; Norton et al., 1990; Schlumberger &
Luck4, 1948; Williams et al., 1994). Schlumberger and Luck6
(1948) discovered numerous spherical 3-5 cm masses in the
lungs of one green turtle. Norton et al. (1990) observed
multiple firm white nodules in both kidneys from a juvenile
green turtle with extensive cutaneous fibropapillomatosis
collected in the Florida Keys. Jacobson et al. (1991)
Figure 2-1. Cutaneous fibropapillomatosis in the green
turtle, Chelonia mvdas. This juvenile stranded in
December 1993 near Key West, Florida in severely
debilitated condition as evidenced by the sunken
plastron. Multiple tumors were found on the neck, front
and rear flippers, axillary and inguinal areas,
perineum, and covering both eyes.
Figure 2-2. Ocular fibropapillomatosis in green turtles,
Chelonia mvdas. (Top) Left eye with multiple
fibropapillomas originating from bulbar and palpebral
conjunctiva, limbus, and cornea. (Bottom) Right eye of a
second turtle with large fibromas arising from the
palpebral and bulbar conjunctiva.
Figure 2-3. Examples of visceral tumors found in some green
turtles with severe cutaneous fibropapillomatosis. (Top)
Kidneys with multiple, irregularly shaped, firm, white
nodules (arrows) ranging from 0.5 to 3 cm in diameter
bulging from the surface. (Bottom) Lungs with multiple
nodules ranging from 0.2 to 5 cm in diameter. Lung
nodules are well demarcated, smooth, and either firm and
white (fibromatous), or gelatinous and translucent
(myxomatous). Myxomatous nodules (arrow) appear to arise
from fibromatous nodules.
examined 2 turtles with GTFP and, in one animal, found
several discrete firm, white foci up to 1 mm diameter on the
surface of one kidney and multiple discrete 1-4 cm diameter
nodules in the other. They also found similar nodules 1-2 cm
diameter within both lungs. Williams et al. (1994) found lung
and kidney nodules in 41% (7 of 17) of the green turtles
examined from Puerto Rico. Approximately 17% (9 of 52) of the
green turtles with severe cutaneous fibropapillomatosis
presented for necropsy at a rehabilitation center have been
found to have similar nodules in the lungs, kidneys, and
other viscera (Herbst, pers. obs.; Richie Moretti & Tina
Brown, The Turtle Hospital, Marathon, FL 33050, pers. comm.).
Several histologic descriptions of cutaneous GTFP have
been published (Aguirre et al., 1994b; Brooks et al., 1994;
Harshbarger, 1991; Jacobson et al., 1989; Luck6, 1938; Norton
et al., 1990; Smith & Coates, 1938, 1939; Sclumberger &
Luck6, 1948; Williams et al., 1994). Cutaneous GTFP is
described as papillary epidermal hyperplasia supported on
broad fibrovascular stromal stalks (Figure 2-4). The ratio of
epidermal to dermal proliferation varies among lesions.
Masses in which both tissues are hyperplastic are termed
fibropapillomas (Figure 2-4A) while others, comprised of
proliferating dermal components with relatively normal
epidermis, are termed fibromas (Figure 2-4B). Several authors
have postulated that there is a developmental progression
from papilloma (early lesions) through fibropapilloma, to
fibroma (chronic lesions) (Harshbarger, 1991; Jacobson et
al., 1989; Luck6, 1938).
Varying degrees of orthokeratotic hyperkeratosis and
acanthosis were consistent features in all studies (Aguirre
et al., 1994b; Brooks et al., 1994; Harshbarger, 1991;
Jacobson et al., 1989; Luck4, 1938; Norton et al., 1990;
Smith & Coates, 1938, 1939; Schlumberger & Luck6, 1948). The
degree of epidermal hyperplasia in GTFP varied from mild to
moderate (7-15 cells thick) on skin tumors to extensive (up
to 30 cells thick) on some conjunctival and palpebral masses
(Brooks et al., 1994; Jacobson et al., 1989). Fibropapillomas
with extensive epithelial hyperplasia often exhibit
anastomosing rete ridges extending deep into the dermis.
Epithelial cells in hyperplastic areas tend to be
hypertrophied (Brooks et al., 1994; Jacobson et al., 1989).
The fibrovascular stroma contains numerous well-
differentiated fibroblasts arranged in a ground substance
containing compact bundles of collagen fibers. Fibroblasts
and collagen bundles tend to be haphazardly arranged, are
more numerous than in normal dermis, and are more dense near
the basement membrane (Brooks et al., 1994; Jacobson et al.,
1989). Various amounts of collagen and mucopolysaccharide
; :. .
S.- :..I ..
P' -' .
Figure 2-4. Variation in histologic appearance of cutaneous
fibropapillomatosis. (A) Fibropapilloma showing typical
arborizing pattern of papillary epidermal hyperplasia
supported by fibrovascular stroma.(B) Fibroma showing
extensive fibrovascular proliferation covered by
relatively normal epidermis. (H&E, scale bars = 250 pm).
relatively normal epidermis. (H&E, scale bars = 250 Gun).
ground substance have been demonstrated in cutaneous tumors
by trichrome and alcian blue staining (Norton et al., 1990).
Nerves and numerous small blood vessels are found within the
stroma. Fibropapillomas examined in several studies show no
malignant or anaplastic changes and few mitotic figures
(Smith & Coates, 1938; Williams et al., 1994). The benign
nature of GTFP has been confirmed by flow cytometry studies
(Papadi et al., 1995).
Some histologic features identified in cutaneous tumors
have not been reported consistently and may be incidental
findings. Trematode (Spirorchidae) eggs surrounded by
epithelioid macrophages and multinucleate giant cells were
found within the dermal capillaries of some fibropapillomas
(Aguirre et al., 1994b; Brooks et al., 1994; Harshbarger,
1991; Jacobson et al., 1989; Jacobson et al., 1991; Norton et
al., 1990; Smith & Coates, 1939; Williams et al., 1994). In
some lesions containing eggs, eosinophilic granulocyte
infiltrates were also observed (Smith & Coates, 1939).
Epithelial cells in the stratum spinosum and outer layers of
epidermis were hypertrophic and vacuolated in some GTFP
specimens. In these areas, amphophilic intranuclear
inclusions were sometimes observed (Aguirre et al., 1994b;
Jacobson et al., 1989). Jacobson et al. (1991) found
eosinophilic intranuclear inclusions containing herpesvirus-
like particles within some superficial epidermal cells
undergoing intracytoplasmic vacuolation and ballooning
degeneration. Lymphocytic perivascular infiltrates have been
described in several studies (Jacobson et al., 1989; Smith &
Coates, 1938). Cleft formation at the dermal-epidermal
junction has also been noted in fibropapillomas examined by
Jacobson et al. (1989).
Visceral tumors were composed of proliferating fibrous
tissue compatible with the dermal component of cutaneous GTFP
(Jacobson et al., 1991; Norton et al., 1990; Schlumberger &
Luck6, 1948). Lung nodules described by Schlumberger and
Luck6 (1948) were covered by ciliated columnar epithelium.
The renal nodules described by Norton et al. (1990) were
sharply demarcated from surrounding renal tissue and covered
by renal capsule at the surface. Normal renal tubules were
found scattered throughout the proliferating connective
Epizootiolorv of GTFP
Epizootiology is the study of the temporal and spatial
patterns of disease expression in animal populations and
includes efforts to identify etiology, describe incidence and
prevalence, morbidity and mortality, routes of natural
exposure or transmission, and the conditions that lead to
disease outbreaks epizooticss). This information is needed to
understand the full demographic impact of GTFP on wild turtle
Epizootiologic studies of GTFP have been hampered by
several factors. First, there are no diagnostic tests to
detect exposure and early (preclinical) disease because an
etiologic agent has not been identified. Thus, all prevalence
data are based on observation of gross cutaneous tumors
Second, because green turtles are migratory and long lived,
taking between 20 and 50 years to reach sexual maturity
(Balazs, 1982; Frazer & Ehrhart, 1985; National Marine
Fisheries & US Fish and Wildlife, 1991), it is difficult to
sample specific life history stages such as the post hatching
pelagic phase and impossible to conduct longitudinal studies
of cohorts. Third, attempts to correlate disease prevalence
with assorted biotic and abiotic factors are hampered by the
geographic scale over which field surveys need to be
conducted and by limited human and fiscal resources.
Consequently, surveillance for GTFP and monitoring of
potentially relevant biotic and abiotic factors has been
Even in well-monitored sites, sampling methods introduce
biases that affect prevalence estimates. Most field studies
are conducted on feeding grounds or nesting beaches, and it
is therefore not surprising that post pelagic juveniles and
adult females are over represented in the prevalence reports.
Nesting beach surveys underestimate the true prevalence of
GTFP in adult females because debilitated turtles are less
likely to nest and are therefore not sampled (Limpus &
Miller, 1994). Field surveys and fisheries that employ tangle
nets tend to selectively sample larger turtles because small
turtles are not caught in the mesh. Surveys based on stranded
sea turtles may overestimate the prevalence of severe
debilitating disease. Cold stunning events, and sampling
methods that use direct, in-water captures, provide the least
biased population samples. The reader is asked to keep these
caveats in mind when evaluating the information presented
Presently, GTFP has a global circumtropical distribution
(Figure 2-5 and Table 2-1). GTFP has been reported from all
major oceans including the Atlantic (Florida, Bahamas,
Brazil), Caribbean (Cayman Islands, Puerto Rico, Virgin
Islands, Barbados, Venezuela, Colombia, Nicaragua, Costa
Rica, Panama, Belize), and Indo-Pacific (California, Hawaii,
Australia, Sri-Lanka, Seychelles, Sarawak, Malaya, Japan)
(Balazs, 1991; Balazs & Pooley, 1991; Gamache & Horrocks,
1991; Hendrickson, 1958; Jacobson, 1990; Jacobson et al.,
1989; MacDonald & Dutton, 1990; Limpus & Miller, 1994;
Williams et al., 1994; Karen Bjorndal & Alan Bolten,
University of Florida, Gainesville, FL 32611, pers. comm.;
Cynthia Lageaux, University of Florida, Gainesville, FL
32611, pers. comm.; Anne Meylan, Florida Marine Research
Institute, St. Petersburg, FL 33701, pers. comm.; Jean
Mortimer, University of Florida, Gainesville, FL 32611, pers.
-H 0 d 0
,-i E1 -i
i g0 0 0
~- I -
0 CL 0)
0 ( 1 H
4! 0 (U W$
-C El 4-Jca
u u- o
comm.). Unfortunately, there are insufficient data to
reconstruct the temporal and spatial pattern of disease
spread among regions. The early reports from Florida (Smith &
Coates, 1938) and Malaysia (Hendrickson, 1958) suggest that
the disease may have always had a worldwide, albeit sporadic
The prevalence of GTFP varies among locations and from
year to year. Table 2-1 summarizes the available prevalence
data from several field studies. The earliest published
prevalence estimate (1.5 %) was from a survey conducted in
1938 of turtles captured in the Key West, Florida fishery
(Smith & Coates, 1938). Most population surveys, however,
have been conducted since 1975. In these surveys, little or
no disease was found prior to 1982, but prevalences rose
rapidly in the 1980s and have remained elevated. Part of this
pattern may reflect an increased awareness of the disease,
but may also reflect a real increase in the prevalence and
severity of GTFP over time.
Prevalences in well monitored feeding ground sites range
from 0% in Inagua, Bahamas (Bjorndal & Bolten, pers. comm.),
Bermuda (Meylan, pers. comm.) and offshore reef sites in
Australia (Limpus & Miller, 1994), to 92% in Kaneohe Bay,
Hawaii (Balazs, 1991). Prevalences may vary greatly between
demographically matched populations over very short distances
(< 1 km), as seen when comparing the prevalence of GTFP in
m4 a r
M 0 0 (
420) 4 )
0 0 0
-H 0 0
42 U (0
- 0 0
0 o0 0O4
,0 I 0 c 0
,a 3 0
0 (DO WO
a m0 o ( o
,.-I u R u
m 04 mu F, E
44 C- -4
S3.H C -C c m o e
a4J -Ha)CD -H-H -H
S o l'0 C- 0
S <: 3i0 -HQ. -H
4 3 H g a
D 0d r
4-1 0a) CD 0) 4-q
aM CD CD E0 (D
a; ) 4 E-I -H a)-
CD '0 4-i
C14 i* -
CO Tl o1
0 s-4 U
0 0) 4-1
z (O o-A
(a- p S
r- r- H H
(0 Oa 1C 0a
0a -0 -
04 0m 0)
N N N N
o C1 ra r a (
H- 0) -- H H-1
(H r- 0 0r 0 0 a
r4 >4 r r-4
0 '0 4.J 90 -
CO Q C CO.
a0 0a) r. 0 Z 0
1u co o u au
*H a) C,
-H M )
- H 4J 4-4 ( 4-
S0 > m wa 0) m
m --I a) r-1 c-
CO 44 4 1 B M 4
C -H a
,a H- 4-4
0) 0) 0
a( a) -H
0 0 0 I
O 0 0
the Indian River (approximately 50%) with that from the
adjacent near-shore Sabellariid worm reef at Wabasso Beach
(0%) (Ehrhart, 1991; pers. comm.).
There is a seasonal pattern in the prevalence of GTFP
among stranded green turtles in Florida with more affected
turtles stranding in the winter months (Wendy Teas, Southeast
Fisheries Science Center, NMFS, Miami, FL, 33149, pers.
comm.). Anecdotal reports indicate that tumors grow rapidly
in summer and are quiescent in the winter in response to
water temperature (Moretti & Brown, pers. comm.). Thus,
tumors may grow rapidly in summer and may reach a size that
is debilitating by autumn. The onset of colder water
temperatures in winter may further stress GTFP affected
turtles sufficiently to cause the winter stranding peak.
GTFP appears to effect certain age and size classes of
turtles more than others. GTFP is rare (0-12%) among nesting
adult females and lesions tend to be focal and mild (Balazs,
pers. comm.; Ehrhart, pers. comm.), although these data are
probably biased (see above). In Hawaiian feeding ground
sites, intermediate sized turtles, measuring 40-90 cm
straight carapace length (SCL), were more frequently and more
severely affected than other size classes (Balazs, 1991).
Ehrhart (1991) and Schroeder (pers. comm.) found similar
results in Indian River and Florida Bay, respectively.
Turtles weighing between 10 and 30 kg were more likely to
have GTFP and they were more severely affected than larger or
smaller size classes (Ehrhart, 1991). Similarly, among
stranded turtles in Florida, 93-98% are between 30 and 69.9
cm SCL (Teas, pers. comm.).
Juvenile green turtles enter near-shore feeding grounds
after 2-5 years of pelagic existence. Too few pelagic stage
juvenile green turtles have been examined to provide any
information about GTFP prevalence in this life history stage.
The consensus, however, is that juveniles develop GTFP after
they have migrated into near-shore waters. This hypothesis is
supported by the very low prevalence estimates for off-shore
sites (Table 2-1) and the fact that GTFP has never been
observed in any recent recruits from the pelagic environment
to near-shore habitats (Balazs, 1986; Ehrhart, 1991; Limpus &
Miller, 1994). Newly arrived juveniles are recognized by
their small size (25-30 cm SCL, weight < 5 kg) and the lack
of epibiota (algae, bryozoans, leeches, etc) on their
carapaces as seen in older resident turtles (Ehrhart, 1991;
Bolten & Bjorndal, 1992; Limpus & Miller, 1994). Although
these small turtles are under-represented in net surveys,
relatively unbiased samples of the near-shore juvenile turtle
population have been obtained during cold stunning events.
During one such event in 1985 in the Mosquito Lagoon,
Florida, 145 cold stunned turtles were collected
(Witherington & Ehrhart, 1985). Twenty-eight percent of these
turtles were recent recruits (< 5 kg) and none had GTFP even
though GTFP prevalence in the overall sample was 29%.
There are several explanations for the absence of
clinical GTFP among turtles that have recently immigrated to
near-shore feeding grounds from the pelagic environment. One
is that affected pelagic juveniles do not survive long enough
to migrate to feeding grounds. Survivorship of healthy
turtles through the 2-5 years of pelagic existence is already
so low that it is unlikely that diseased turtles would
survive. A second explanation is that the disease has a long
latent period, so that clinical disease develops only in
older juveniles after they have moved inshore, even though
exposure to the etiologic agents) may occur in the pelagic
zone. A third hypothesis is that the causative agents) or
environmental conditions appropriate for disease expression
are found only in some near-shore habitats, so that exposure
occurs after juveniles are recruited to these sites.
The wide variation in GTFP prevalence among size/age
matched populations of juvenile green turtles lends support
to the last hypothesis. If the cause of GTFP were encountered
in the pelagic zone and the pelagic juveniles assorted
randomly among near shore sites, then one would expect the
distribution of GTFP prevalences among these near-shore sites
to be more uniform than is observed. Monitoring studies in
the Indian River system provide the most convincing data
because GTFP prevalence varies from 0% to approximately 50%
over a very short distance (< 1 km) between two
demographically similar populations and there is documented
movement of turtles from the low prevalence site (ocean) into
the high prevalence site (lagoon) but not vice versa
(Ehrhart, 1991; pers. comm.).
The field survey data summarized in Table 2-1 indicate
that GTFP is more prevalent in shallow, near-shore ecosystems
(lagoons, bays) and possibly most prevalent in areas that are
impacted by human activities such as agricultural,
industrial, and urban development within the catchment basin.
The strongest association of GTFP prevalence appears to
be with habitat type (embayments). These marine environments
may provide favorable physical conditions for either
infectious or non-infectious disease agents. For example,
certain sediment types may accumulate chemical contaminants
and, combined with low flushing rates, could increase the
level of exposure to chemical carcinogens or immunotoxins.
These same sediment properties and hydrodynamic conditions
may also favor the accumulation and maintenance of high
concentrations of infectious agents. More variable water
temperatures in shallow embayments could affect the rate of
xenobiotic metabolism, tumor cell proliferation, immune
system function, and pathogen replication. For example,
thermal stress has been shown to exacerbate virus infection
in hatchling green turtles (Haines & Kleese, 1977; Kleese,
1984) and modulate virus expression and tumor growth in Luck6
renal adenocarcinoma (McKinnell, 1981, 1984; Zambernard &
Vatter, 1966). Variable salinity in near-shore habitats may
have similar stress effects.
Near-shore habitats may also provide an optimum biotic
environment for survival and transmission of an infectious
etiologic agent. Disease transmission could be enhanced by
high population densities of vectors or intermediate host
species. Feeding grounds may attract a high density of
susceptible turtles which would facilitate the transmission
of pathogens in a density dependent fashion, as has been
shown for horizontally transmitted damselfish
neurofibromatosis (Schmale, 1991) and the Luck6 virus
(McKinnell, 1981, 1984). Recruitment of susceptible turtles
from many different breeding stocks into common foraging
grounds may allow the exchange of many diseases, including
GTFP, from exposed to naive individuals. Habitat differences
in levels of other stressors such as concurrent infectious
disease (parasites), and disturbance by human activity
(fishing, boating, dredging) may render turtles more
susceptible to or less able to recover from GTFP. These
hypotheses provide the conceptual framework for future
Clinical Course. Morbidity, and Mortality
Accurate estimates of the number of turtles that become
clinically affected following exposure to the agents) that
cause GTFP are unavailable. The duration and course of
clinical GTFP are poorly understood, primarily because
individual turtles with fibropapillomas of known duration
have not been available for longitudinal studies. A few green
turtles have been held in captivity long enough to provide
some generalizations about clinical course of the disease.
Jacobson et al. (1989) held 6 immature turtles with multiple
cutaneous GTFP in captivity for several months. Some tumors
on some animals decreased in size while others increased in
some animals when examined 4 months after capture. Ehrhart
maintained 3 green turtles with GTFP in captivity for
approximately 3 months (Ehrhart, 1991; Ehrhart et al., 1986).
During that time one animal lost several tumors, a second
developed 8 new tumors, and the third exhibited no changes in
tumor burden. In these captive observation studies, the
length of time that animals had the disease prior to capture
Field mark and recapture studies also indicate a
variable clinical course. For example, of 56 green turtles
recaptured in the Indian River, 7% had tumors when first
marked but had none at recapture, 14% contracted tumors
between first capture and recapture, 38% had lesions both
times, while 41% were free of lesions both times (Ehrhart,
1991). These data, while limited in number, support the
conclusion that the clinical course is prolonged and that
some individuals may spontaneously recover from disease.
Recapture rates are generally low, however, and there is no
control over the time interval between capture and recapture.
Studies of the temporal patterns of progression and
regression of experimentally transmitted GTFP in captive
turtles are needed.
Accurate estimates of the proportion of clinically
affected turtles that die are unavailable. Cutaneous
fibropapillomas can become large enough to interfere with
locomotion and are easily entangled in discarded line. Ocular
fibropapillomas (Figure 2-2) may occlude vision and those
invading the cornea may cause secondary panophthalmitis with
destruction of the globe (Brooks et al., 1994; Herbst, pers.
obs.). Visceral fibromas (Figure 2-3) grow by expansion
within the stroma of the affected organ and eventually
disrupt normal organ functions. Cardiac dysfunction, buoyancy
problems and respiratory compromise, hydronephrosis,
gastrointestinal obstruction have all been observed or
suspected causes of death in affected turtles (Balazs, pers.
comm.; Herbst, pers. obs.; Moretti & Brown, pers. comm.).
Many green turtles with multiple cutaneous fibropapillomas
become severely debilitated (Figure 2-1). Blood chemistries
and blood cell counts of severely affected green turtles
confirm a general pattern of debilitation (Jacobson, 1987;
Norton et al., 1990). Abnormalities include non-regenerative
anemia, hypoproteinemia, electrolyte imbalances, uremia, and
elevations in liver enzymes (Norton et al., 1990). The
cachexia may be caused by any combination of the following:
inability to locate, ingest, or digest food, excessive energy
demands for growth by proliferating tumors, increased
energetic costs for locomotion, the physiological effects of
certain cytokines such as tumor necrosis factor, mediated by
the immune system, and/or concurrent disease such as
spirorchidiasis. Whatever the mechanismss, a number of
animals become sufficiently debilitated by GTFP to strand
(Balazs, 1991; Teas, 1991). In one rehabilitation center
about 50% of green turtles that were still alive at stranding
died despite extensive rehabilitation efforts (Moretti &
Brown, pers. comm.).
Possible Causes of GTFP
The etiology of GTFP is unknown and its identification
is one of the aims of this research project. Hypotheses about
etiology can be proposed based on comparisons with similar
tumors of known etiology from other species, the association
of potential pathogens with GTFP, and from epizootiologic
patterns. However, demonstration of causation requires
rigorous experimentation to fulfill Koch's postulates. This
is most easily accomplished with diseases caused by
infectious agents. However, infectious disease expression may
depend on a variety of host-related, pathogen-related, and
environmental factors (Hanson, 1988).
GTFP has histologic features in common with benign
cutaneous neoplasia, found in other vertebrates, such as
papillomas and fibromas (Pulley & Stannard, 1990), as well as
hyperplastic conditions such as keloidosis (Caro & Bronstein,
1985) and exuberant granulation tissue (Smith et al., 1972).
Thus, GTFP is consistent with either neoplasia or
hyperplasia. The classification of GTFP as a neoplastic
disease has been controversial (Harshbarger, 1984). In
general, neoplasia may result from any of a variety of
derangements at any point in the complex signalling and
control network of normal cellular proliferation and
differentiation (Bishop, 1991). The pathogenesis of neoplasia
may involve multiple cumulative steps (Hunter, 1991; Peraino
& Jones, 1989), with early steps causing unregulated
proliferation and later events leading to malignancy.
Similarly, hyperplasia can be caused by a variety of stimuli.
Thus, identifying a single etiology and fulfilling Koch's
postulates for GTFP may be difficult if not impossible.
Some potential causes of neoplastic and hyperplastic
proliferative lesions in other vertebrate species include
abiotic agents (ultraviolet light, chemical contaminants),
and infectious biological agents (viruses, bacteria, metazoan
parasites), with or without predisposing heritable genetic
conditions. The following sections review and discuss
evidence for or against the involvement of each of these
factors in the etiology and pathogenesis of GTFP.
Smith and Coates (1938) were the first to suggest a role
for solar radiation in the pathogenesis of GTFP. Fifty years
later there is mounting concern that ozone depletion is
causing an increase in ultraviolet-B (290-320 nm) irradiation
(Kerr & McElroy, 1993) and that this may be having pervasive
effects in aquatic ecosystems (Hader, 1993) and on animal
health (Van der Leun & De Gruijl, 1993). UV-B produces direct
DNA damage by pyrimidine dimer formation (Anathaswamy &
Pierceall, 1990). This may lead to mutation in cellular
oncogenes and the development of neoplasia (Brash et al.,
1991). UV-B also causes immunosuppression in experimental
animals (Baadsgaard, 1991; Donawho & Kripke, 1991; Granstein,
1990; Noonan & DeFabo, 1992). The proposed mechanism involves
pyridine dimer formation (Applegate et al., 1989; Kripke et
al., 1992) and/or a trans to cis isomerization of urocanic
acid in the skin following UV-B absorption (DeFabo & Noonan,
1983; Noonan & DeFabo, 1992). For example, trout exposed to
levels of UV-B radiation within the ambient range recorded
for mid-latitudes developed skin damage and became
immunosuppressed, as evidenced by a high prevalence of fungal
skin infections (Fabacher et al., 1994).
Increased UV-B exposure could occur in the shallow
inshore waters where green turtles feed. However, GTFP
prevalence varies too greatly over very short distances (as
in Ehrhart's study area) for UV-B to be the cause of GTFP.
UV-B may be a cofactor in disease expression, however, and
the role of UV-B in modulating the immune system of turtles
deserves further investigation.
A variety of chemical compounds have been shown to cause
benign fibroepithelial proliferation and to have mutagenic
and carcinogenic properties under experimental conditions
(Anderson & Reynolds, 1989; Weisburger, 1989). The list of
compounds is extensive but they seem to act by either of two
basic mechanisms: (1) direct nucleic acid damage leading to
genetic mutation initiatorss), and (2) cellular damage or
irritation leading to proliferation (promoters). As mentioned
earlier, chemical effects may be one of many mechanisms
involved in multistep carcinogenesis.
The involvement of chemical contaminants in naturally
occurring neoplastic disease of lower vertebrates has been
documented best in fish. The prevalence of liver pathology,
including liver neoplasia in brown bullheads, Ictalurus
nebulosus, was higher at contaminated sites than at
relatively clean sites in several North American lakes and
rivers (Baumann et al., 1987; Black, 1983; Bowser et al.,
1990a). Disease prevalence was correlated with contaminant
levels in fish in one study (Baumann et al., 1987) and with
sediment levels in another (Black, 1983). Neoplastic lesions
were induced experimentally by treating bullheads with
sediment extracts (Black, 1983). Similar associations between
hepatic neoplasia, polluted sites, and sediment contaminant
concentrations have been found for mummichogs, Fundulus
heteroclitus in Chesapeake Bay (Vogelbein et al., 1990) and
various bottom fish in Puget Sound (Malins et al., 1984). A
similar association has been found between contaminated
sites, in vitro mutagenesis of water and sediment extracts
from those sites, and the prevalence of pigment cell
neoplasia (Chromatophoromas) in croakers, Nibea mitsukurii
(Kimura et al., 1984; Kinae et al., 1990). In addition,
experimental application of chemical carcinogens reproduced
the tumors in these fish. In most of these studies,
polycyclic aromatic hydrocarbon (PAH) concentration was a
major factor in the association between disease prevalence
with contaminant levels. Similarly, PAHs have been implicated
in the pathogenesis of cutaneous neoplasia in tiger
salamanders, Ambvstoma tirrinum from a polluted pond (Rose,
1981; Rose & Harshbarger, 1977). Cutaneous papillomas have
been experimentally induced in lizards, Lacerta aailis with
dimethyl benzanthracene (Stolk, 1963).
Chemical contaminants may also play a role in the
pathogenesis of certain neoplastic diseases by disrupting
immune functions that would otherwise allow the host to
eliminate transformed cells. The effects of various
immunotoxins have been reviewed by Dean et al. (1990).
Chemical effects on immune function in fish has also been
reviewed (Anderson et al., 1984; Dunier, 1994; Zeeman &
Brindley, 1981). Associations between chemical contaminants
and immune dysfunction have been shown (Lahvis et al., 1995)
and experimentally demonstrated in some marine organisms
(Arkoosh et al., 1994; DeSwart et al., 1994). Contaminants
may also disrupt the immune system indirectly by disrupting
neuroendocrine functions (Colborn et al., 1993).
The role of chemical contaminants in green turtle
fibropapillomatosis is unknown. As described previously,
there is a possible association between high GTFP prevalence
and near-shore marine habitats that have been impacted by
human activity (Table 2-1). Although it is possible that
environmental degradation and contaminants play a role in
disease expression, more objective documentation of these
impacts in high and low GTFP prevalence sites are needed.
Problems arise in how to document the contaminant
exposure of marine turtles. Few data are available for
comparing contaminant residue levels in water, sediment, or
benthic organisms from high GTFP prevalence areas with those
from areas where GTFP is rare. Similarly, data on contaminant
levels in green turtle tissues are scant and difficult to
obtain because of the endangered status of this species
(Aguirre et al., 1994a; Clark & Krynitsky, 1980; Hall et al.,
1983; McKim & Johnson, 1983; Rybitski, 1993; Thompson et al.,
1974). The few studies that have been published are difficult
to interpret in the context of GTFP. For example, while one
study in 1983 found significant amounts of hydrocarbons in 2
green turtles that stranded after a major oil spill (Hall et
al., 1983), most surveys of organochlorine and
polychlorinated biphenyl residues in green turtle tissues
including egg (Aguirre et al., 1994a; Clark & Krynitsky,
1980; McKim & Johnson, 1983; Rybitski, 1993; Thompson et al.,
1974) have yielded relatively low levels, often below the
limits of detection of the methods.
Where data exist, there are problems with relating
contaminant levels to disease prevalence. First of all, the
biologic effect (toxicity) of any particular residue level in
green turtles is unknown. Second, surveys of residue levels
are usually limited to those chemicals that persist in the
environment or bioaccumulate, although important toxic
effects such as genetic damage (in a multistage
carcinogenesis model) can result from transient exposures to
compounds that do not bioaccumulate. In addition, exposure to
a potent chemical carcinogen may occur transiently in a
completely different habitat from that being monitored.
Third, toxic effects may not be direct as in some
experimental models, but may involve complex interactions
with other abiotic and biotic factors. Thus, fulfilling the
criteria for implicating chemical contaminants as the primary
cause of GTFP or as cofactor could be extremely difficult
(Foster et al., 1993; Hanson, 1988). Finally, the same
biological effects may be caused by any number of different
compounds acting through several different mechanisms.
Decisions about which contaminant residues to measure should
be made with specific a Driori mechanistic hypotheses in mind
or in light of documented history of exposure to specific
Nevertheless, there is a need to conduct further
toxicological studies. Specifically, there is a need to
collect data on the water, sediment, and turtle tissue levels
of several classes of chemical contaminants (including known
chemical carcinogens and immunotoxins) from several carefully
matched marine sites with different prevalences of disease.
In addition, controlled experiments involving exposure of
turtles to water or sediment extracts from high and low
prevalence areas will be necessary in order to clearly
demonstrate a contaminant effect in the etiology and
pathogenesis of this disease.
The epizootiologic patterns observed among free-ranging
green turtle populations including the sudden appearance of
GTFP at new geographic sites, variation in prevalence over
relatively short distances, and temporal variation within a
locality are compatible with an infectious etiology. The
observation that some animals recover from GTFP is also
compatible with an infectious disease. In addition, an
infectious agent is the most plausible explanation for the
appearance and spread of GTFP among captive green turtles.
For example, the outbreak documented at Cayman Turtle Farm,
Grand Cayman, in 1980, began in wild caught adults and
subsequently developed in captive reared turtles over several
years. Once eliminated, GTFP has not recurred at Cayman
Turtle Farm despite little change in husbandry conditions
(Jacobson, 1981b; Jacobson et al., 1989). A similar outbreak
occurred in a head start facility in the Florida keys among 2
year old captive reared green turtles that had been held in a
pond where GTFP affected turtles were rehabilitated and
possibly had direct contact with affected turtles (Hoffman &
A number of virus families (Papovaviridae,
Herpesviridae, Adenoviridae, Poxviridae, Retroviridae) are
known to induce proliferative and or neoplastic lesions.
Papillomaviruses (Papovaviridae) are the documented cause of
papillomas, fibromas, and fibropapillomas in many mammalian
and avian species (Sundberg, 1987) and are associated with
malignant neoplasia as well (Sundberg & O'Banion, 1989; Zur
Hausen, 1989). Among reptiles, a papillomavirus has been
described from hyperplastic skin lesions of 5 Bolivian side-
necked turtles, Platemvs Dlatvcephala (Jacobson et al.,
1982), and papovavirus-like particles have been observed in
papillomas of green lizards, Lacerta viridis (Cooper et al.,
1982; Raynaud & Adrian, 1976). A polyomavirus (Papovaviridae)
of hamsters produces benign cutaneous neoplasia in these
rodents (Graffi et al., 1968) but other polyomaviruses of
rodents and primates do not produce disease in their natural
hosts (Eckhart, 1990). Herpesviruses have been associated
with cutaneous papillomas and or fibromas in green lizards,
Lacerta viridis (Raynaud & Adrian, 1976), african elephants,
Loxodonta africana (Jacobson et al., 1986b), carp, CyDrinus
carpio (Hedrick et al 1990; Sano et al., 1985), and several
salmonids (Kimura et al., 1981a, 1981b, 1981c; Sano et al.,
1983; Yoshimizu et al., 1987). Poxviruses are responsible for
fibroepithelial proliferative lesions in squirrels (Hirth et
al., 1969; O'Connor et al., 1980), rabbits (Pulley & Shively,
1973; Shope, 1932), and primates (Behbehani et al., 1968).
Retroviruses have been associated with or proven to be the
cause of dermal sarcomas in walleyes, Stizostedion vitreum
(Martineau et al., 1991), lip fibromas in angelfish (Francis-
Floyd et al., 1993), neurofibromas in damselfish, Pomacentrus
partitus (Schmale & Hensley, 1988), myxofibromas and
rhabdomyosarcomas of snakes (Lunger et al., 1974; Zeigel &
Clark, 1969), sarcomas in poultry (Benjamin & Vogt, 1990),
and fibromas and sarcomas in a variety of mammalian species
including cats (Hardy, 1985; Moulton & Harvey, 1990) and
primates (Gardner & Marx, 1985; Tsai et al., 1990). The
molecular mechanisms of virus induced proliferation and
oncogenesis vary, but all in some way disrupt the cells'
signal transduction network (Benjamin & Vogt, 1990; Moran,
1993; Zur Hausen, 1991). Examples include, the E5 protein of
certain papillomaviruses that binds and activates the PDGF
receptor (Kulke & DiMaio, 1991; Petti & DiMaio, 1992; Petti
et al., 1991), the adenovirus ElB protein and papillomavirus
E6 protein that bind and inactivate cell cycle checkpoint
protein p53 (Benjamin & Vogt, 1990; Moran, 1993), and
adenovirus ElA, polyomavirus T antigen, and papillomavirus E7
proteins that target the cell cycle control protein pRB.
Certain viruses produce proteins that act like growth factors
or their receptors. For example, poxviruses may produce
epidermal growth factor (EGF)-like peptides (Brown et al.,
1985), simian sarcoma virus (a retrovirus) produces a PDGF-
like peptide (Benjamin & Vogt, 1990), and certain
herpesviruses express a protein with protein kinase activity
similar to receptor kinases (Smith et al., 1992).
Evidence for viruses. Certain histologic features of
GTFP, including perivascular lymphocytic infiltrates, vesicle
formation, and epithelial degeneration are consistent with,
although not specific for, virus infection, and have prompted
investigators to search for viruses. Smith and Coates (1938)
failed to find virus-like inclusions within the tumors that
they examined. Jacobson et al. (1989) examined tumors from
six turtles from Florida and one turtle from Hawaii by light
and electron microscopy. In some sections, cells in the
stratum spinosum and outer layers of the epidermis were
hypertrophic and vacuolated and amphophilic intranuclear
inclusion bodies suggestive of herpesvirus infection were
occasionally seen. Ultrastructural examination revealed mild
acanthosis (3-6 cells thick) and intracytoplasmic vacuoles
containing 150-170 nm diameter granules of varying electron
densities were described within the superficial epidermis but
not identified (Jacobson et al., 1989). Aguirre et al.
(1994b) described basophilic intranuclear inclusions in
several lesions that they suspected to be nucleoli, but also
considered compatible with viral inclusions. However, viral
particles were not found within these inclusions when
examined by EM. They also observed intracytoplasmic electron
dense granules approximately 150 nm in diameter that were
morphologically similar to viral particles. These, however,
were found in both normal and GTFP epithelium. These
intracytoplasmic granules are now generally accepted to be
mucin granules that are produced and secreted by normal
turtle keratinocytes as they differentiate (Aguirre et al.,
1994b; Jacobson, 1989; Matoltsy & Huszar, 1972).
Herpesvirus-like particles were demonstrated by electron
microscopy in some fibropapillomas taken from two juvenile
green turtles housed in the same rehabilitation facility in
the Florida Keys (Jacobson et al., 1991). In 3
fibropapillomas examined from one turtle and 1 of 14 tumors
examined from the second turtle, focal areas of ballooning
degeneration of superficial epithelium were found to contain
eosinophilic intranuclear inclusions. Electron microscopy
demonstrated the presence of particles within the nucleus
conforming in size and morphology (icosahedral 77-90 nm
diameter) with immature herpesvirus and intracytoplasmic
particles conforming with mature enveloped herpesvirus (110-
120 nm diameter). This agent was not successfully cultured so
experiments to fulfill Koch's postulates could not be
Immunologic and molecular methods have been used to
search specifically for papillomaviruses in GTFP biopsies
because of the similarities between turtle fibropapillomas
and those of other vertebrates known to be caused by this
type of virus (Sundberg, 1987). Jacobson et al. (1989) were
unable to demonstrate the presence of papillomavirus group-
specific structural antigens in paraffin embedded sections of
fibropapillomas from 1 Hawaiian and 6 Florida green turtles
using peroxidase-antiperoxidase immunohistochemistry. Total
DNA extracted from portions of these same tumors was probed
under low stringency conditions with bovine papillomavirus
type 2 virion DNA. Finally, a reverse Southern blot was
performed in which radiolabelled tumor DNA from two turtles
was allowed to hybridize with blots containing 25 different
cloned papillomavirus genomes (6 bovine, 7 human, and dog,
rabbit, coyote, mouse, rat, and parrot papillomaviruses).
These screenings for papillomavirus yielded negative results
(Jacobson et al., 1989). Similarly, Marc Van Ranst (Einstein
Medical College, Bronx, New York, pers. comm.) had negative
results when he performed low stringency southern blot
analysis on DNA extracts of 11 tumors collected from a single
immature green turtle from the Florida Keys. Probes included
full genomic DNA from human papillomaviruses HPV-1, HPV-2,
and HPV-5, bovine papillomavirus BPV-1, canine oral
papillomavirus, and pygmy chimpanzee papillomavirus PCPV-1.
Preliminary experiments have also been conducted using
the polymerase chain reaction and degenerate PCR primers for
conserved sequences in the El and L1 mammalian papilloma
virus genes. These primers failed to amplify any sequences in
11 GTFP biopsies from one green turtle (Van Ranst, pers.
These negative immunohistochemical and molecular data
are insufficient to rule out a papillomavirus as a potential
etiologic agent because papillomaviruses are extremely
diverse and it is not unlikely that a reptilian papilloma
virus would fail to react with mammalian and avian probes,
primers, and antisera (O'Banion et al., 1992).
The possible role of oncogenic retroviruses has never
been investigated. As with the papillomaviruses, retroviruses
may cause neoplasia without ever developing a patent life
cycle in the host (Benjamin & Vogt, 1990; Coffin, 1990).
However, in the absence of electron microscopic evidence for
virus production and shedding from the tumor it is difficult
to implicate a retrovirus as the cause. Detection of
integrated retroviral genomes (provirus) within the green
turtle genome will require specific molecular probes that
will become available only after the agent has been
identified and portions of its genome sequenced. Until then
it is unlikely that non-specific retroviral probes would
yield conclusive results given the ubiquity of endogenous
retroviral sequences in vertebrates (Coffin, 1990).
Chronic bacterial infections may induce proliferative
lesions in some tissues. For example, intracellular
Camovlobacter-like organisms are associated with
proliferative enteritis in ferrets, hamsters, and swine (Fox
& Lawson, 1988; Lawson et al., 1985). An invasive spirochaete
has been observed in papillomatous foot lesions in cattle but
experiments to fulfill Koch's postulates have not yet been
conducted (Read et al., 1992). Numerous species of bacteria
have been cultured from the surfaces of cutaneous green
turtle fibropapillomas (Aguirre et al., 1994b). However,
bacteria are not seen within intact GTFP lesions, and little
inflammation is observed within tumors unless the surface is
ulcerated, suggesting that these are all secondary
An association between parasites and neoplasia has been
made in several species. Dogs infected with the nematode,
Spirocerca lupi, which encysts in the esophagus, may develop
a fibrosarcoma at the site (Bailey, 1963). Rats with tapeworm
Cvsticercus (Taenia) cysts were reported to develop hepatic
carcinomas with high frequency (Dunning & Curtis, 1946).
Schistosoma mansoni infection with egg shedding in the
urinary tract has been associated with bladder cancer in
humans (Hashem et al., 1961).
Marine turtles are host to a variety of digenetic
trematode species (Lauckner, 1985). Benign papillomatous
lesions in the gallbladder of green turtles have been
associated with flukes and eggs of Rhvtidodoides similis
(family: Rhytidodidae) (Smith et al., 1941). At least 12
species of spirorchid trematodes have been described in the
green turtle (Lauckner, 1985). Their natural history is very
similar to Schistosoma in that adult trematodes inhabit the
vascular system and eggs must migrate through tissues to
reach an outlet to the environment. Schistosoma mansoni egg
antigens can elicit a fibrotic response in the host (Lammie
et al., 1986; Phillips & Lammie, 1986; Wyler, 1983) and it
has been suggested that spirorchid eggs may induce fibromas
by similar mechanisms (Harshbarger, 1984).
Evidence for a metazoan parasite etioloav. The
association of spirorchid egg deposition with
fibropapillomatosis was first noted by Smith and Coates
(1939), who found eggs of Hapalotrema constrictum in sections
of over half of 230 fibropapillomas that they examined from
Florida green turtles. Later, Jacobson et al. (1989) did not
find ova in any sections of 28 tumor biopsies collected from
6 Florida green turtles but eggs were present in tumor
sections from 1 Hawaiian turtle. Norton et al. (1990) and
Jacobson et al. (1991) found eggs in many sections of tumors
from 3 Florida turtles. Williams et al. (1994) found eggs in
fibropapillomas examined from 39 Caribbean green turtles, and
Aguirre et al. (1994b) found eggs in biopsy sections from 8
of 10 Hawaiian turtles affected with GTFP.
On the other hand, spirorchid ova have been found in 16
of 21 (76%) wild green turtles, 3 of 10 (33%) oceanarium-
reared green turtles, and 3 of 102 (2.9%) farmed green
turtles from Queensland, Australia where the prevalence of
fibropapillomatosis was 0% (Glazebrook & Campbell, 1990a,
1990b). In addition, ova and lesions associated with ova have
been found within otherwise normal tissues of GTFP affected
turtles (Aguirre et al., 1994b; Norton et al., 1990).
The significance of cardiovascular trematodes in the
etiology of GTFP remains unclear. On the one hand, some
authors have characterized cutaneous fibromas from green
turtles as a hyperplastic response to spirorchid eggs
(Harshbarger, 1984) although the occurrence of spirorchid ova
and ova induced lesions in otherwise normal tissues of
turtles with GTFP and in those that do not have GTFP argues
against a direct hyperplastic or tumorigenic effect. On the
other hand, Smith and Coates (1939) concluded that trematode
eggs were incidental, accumulating passively in the
microvasculature of tumors. Other authors have also tended to
discount the eggs as the cause of GTFP (Jacobson et al.,
1989; Lauckner, 1985).
Finally, an argument has been put forward that
ectoparasites may have some role in the pathogenesis of
fibropapillomatosis. Nigrelli (1942) and Nigrelli and Smith
(1943) found leeches, Ozobranchus branchiatus, infesting the
folds of papillomas and suggested that the leeches may act as
vectors of the causative agentss. Most authors agree,
however, that leeches do not cause tumors directly, although
they may severely debilitate their host (Schwartz, 1974). In
addition, hirudin secretion may cause some increased
vascularization at leech attachment sites (Lauckner, 1985;
Nigrelli & Smith, 1943). Aguirre et al. (1994b) reported
finding mites attached to the surface of Hawaiian
fibropapillomas but did not speculate on the significance of
Neoplastic transformation, by definition, involves
permanent, often multiple, changes in the cell's genotype
leading to relatively unregulated proliferation and
differentiation (Sirica, 1989). Familial patterns of
neoplastic disease arise from a heritable (germline) genetic
lesion that renders individuals more susceptible to disease
development following subsequent somatic cell genetic damage.
For example, a germ line loss of function mutation in a tumor
suppressor locus would predispose an individual to neoplasia
following any event that disables the remaining functional
allele (Haber & Housman, 1991; Knudson, 1986; Marshall,
1991). Familial patterns of neoplasia are well documented in
humans. Examples include Li-Fraumeni syndrome, Wilm's tumor,
retinoblastoma, and neurofibromatosis (Haber & Housman, 1991;
Marshall, 1991). Laboratory mice show extensive strain
variation in susceptibility to experimental tumorigenesis
(DiGiovanni, 1989). A well documented example of genetic
susceptibility to neoplasia among lower vertebrates is found
in certain platyfish, Xiphophorus maculatus/Xiphophorus
helleri hybrids, which have high rates of spontaneous and
ultraviolet light induced melanoma due to loss of a
functional tumor suppressor gene (Anders et al., 1984;
Friend, 1993; Sobel et al., 1975). Heritable defects in DNA
repair mechanisms could also render individuals more prone to
neoplasia as is the case in xeroderma pigmentosa (Dresler,
1989; Kraemer et al., 1984). The familial pattern of
epidermodysplasia verruciformis is believed to involve
heritable defects in cellular immune function and an
inability to eliminate papillomavirus infection (Orth, 1987;
Shah & Howley, 1990). In rabbits, certain major
histocompatibility loci have been associated with the
regression or progression to malignancy of Shope papilloma
induced tumors (Han et al., 1992). In addition, individuals
of some species have genetic predispositions to exuberant
hyperplastic responses to wounding, e.g. keloidosis in humans
(Caro & Bronstein, 1985) and "proud flesh" in horses (Smith
et al., 1972), that can resemble benign neoplasia.
The possibility that some green turtles have a genetic
predisposition to develop GTFP must be considered. However,
there is no evidence that this is the case because
genealogical studies in this species are impractical and
methods to distinguish susceptible from resistant individuals
Immune Dysfunction in GTFP Pathoqenesis
The possibility that various biological and
environmental agents may be involved indirectly in GTFP by
causing immune system dysfunction has been alluded to in each
of the preceding sections. While immune suppression would not
be a necessary prerequisite for infection, if GTFP were
caused by a primary infectious agent, the disease would
probably be more persistent and severe (Chretien et al.,
1978; Duncan et al., 1975; McMichael 1967) and more likely to
progress to malignancy (Schneider et al., 1983) in those
individuals with compromised cellular immune function.
Implicating immune system dysfunction in the
pathogenesis of GTFP will be difficult because few turtle
specific reagents are available, immune function assays have
not been validated, and normal reference ranges have not been
established for green turtles. Immune system studies in sea
turtles are in their infancy. Collins (1983) provided initial
anatomic descriptions of green turtle lymphoid tissues. The
immunoglobulin classes of green turtles have been described
(Benedict & Pollard, 1972, 1977) and some preliminary
investigations of cellular immune functions have been
conducted (McKinney & Bentley, 1985). Matters are complicated
by the fact that the turtle immune system, as in other
poikilotherms, is influenced by both season and temperature
(Ambrosius, 1976; El Ridi et al., 1988; Muthukkaruppan et
al., 1982; Zapata et al., 1992). Eventually, systematic
surveys that compare immune function parameters among
apparently healthy turtles from populations with high and low
GTFP prevalences will be needed to test the hypothesis that
immune system dysfunction renders some green turtle
populations more susceptible to GTFP. If evidence for an
association between immune dysfunction and GTFP is found, it
will become important to identify those factors responsible
for immunomodulation in affected populations.
This chapter has brought together the available
published information about green turtle fibropapillomatosis
and provides the background for developing testable
hypotheses addressed in the following chapters of this
dissertation. It is clear from this review that the major
question to be addressed is, what is the etiology of GTFP?
Once the nature and identity of the etiologic agent are known
one can begin to develop strategies for monitoring and
preventing the spread of GTFP among turtle populations. With
techniques to monitor populations for exposure to the
causative agents) one can begin to model the long-term
demographic effects of this disease and initiate studies
designed to identify those factors that have allowed this
disease to become a worldwide epizootic.
DEVELOPMENT OF MONOCLONAL ANTIBODIES FOR THE DETECTION OF
CLASS-SPECIFIC IMMUNE RESPONSES IN THE GREEN TURTLE
There are 7 extant species of marine turtles, all of
which are threatened or endangered due to a variety of
factors such as over-harvesting, loss of nesting and feeding
habitats, marine pollution, and entanglement (National
Research Council, 1990). The impact of disease and the
relationship of disease susceptibility to environmental
factors are poorly understood, due in part to the lack of
diagnostic reagents with which to monitor the health status
of sea turtle populations. The importance of improved health
monitoring capabilities in wildlife conservation is becoming
increasingly recognized (Klein, 1993). The urgent need to
develop diagnostic tests for green turtles, Chelonia mydas,
stems in part from recent worldwide increases in the
prevalence and severity of green turtle fibropapillomatosis
and the need to better understand the epizootiology of this
disease (Chapter 2).
The development of standardized serodiagnostic tests for
green turtles would be facilitated by the availability of
monoclonal antibodies (Mabs) to specific turtle
immunoglobulin classes. Monoclonal antibodies are highly
specific and uniform reagents with reliable performance
characteristics that can be obtained in potentially unlimited
quantities. This paper describes the production and
validation of a battery of monoclonal antibodies specific for
each of the known immunoglobulin classes of the green turtle
(Benedict & Pollard, 1972).
Materials and Methods
Turtle Plasma Samples and Turtle Immunizations
Blood samples were collected into lithium heparin tubes
from the dorsal cervical sinus (Owens & Ruiz, 1980) of 4 green
turtles housed in a rehabilitation facility in Marathon,
Florida. The plasma obtained from these samples was used to
prepare immunoglobulins for use as antigen in hybridoma
production. Two juvenile captive-reared green turtles housed
at Cayman Turtle Farm, Grand Cayman, British West Indies were
immunized by biweekly subcutaneous inoculations with 250 tg
2,4-dinitrophenylated bovine serum albumin (DNP-BSA)
(Molecular Probes, Eugene, OR, USA) in Ribi's adjuvant (RIBI
ImmunoChem Research, Hamilton, MT, USA) for a total of 6
inoculations, followed by monthly inoculations of the same
DNP-BSA dose for another 8 months. A pre-immunization blood
sample was collected, followed by biweekly test bleedings
between each of the first 6 inoculations, and monthly
bleedings before each monthly booster inoculation. These
plasma samples were used to assess the ability of Mabs to
measure specific turtle anti-DNP responses and for affinity
purification of anti-DNP antibodies. In addition, pooled
plasma samples from loggerhead (Caretta caretta), olive
ridley (Lepidochelys olivacea), Kemp's ridley (LeDidochelvs
kempii), hawksbill (Eretmochelvs imbricata), and leatherback
(Dermochelvs coriacea) were obtained from various sources for
testing cross-species reactivity of the Mabs.
Preparation of Turtle Immunoalobulins
Several strategies were employed to isolate and purify
turtle immunoglobulins for use in mouse immunizations and
hybridoma screening. Initially, putative immunoglobulins were
identified and isolated according to their physico-chemical
properties. Later, additional approaches were taken, as
reagents and antigen specific antibodies from specifically
immunized turtles became available.
Globulins from a 50 ml sample of plasma from an
individual green turtle and from a 100 ml pooled sample from
3 green turtles from Marathon, Florida were precipitated with
saturated ammonium sulphate (SAS) (33% v/v ). The precipitate
was resuspended in PBS/az (0.01 M sodium phosphate buffer (pH
7.4) containing 0.15 M NaC1 and 0.02% NaN3) and the
precipitation repeated. The precipitate was dialyzed into
either PBS/az or 0.01 M Tris-HCl buffer (pH 8.0) and adjusted
to a final protein concentration of 2 mg/ml.
One portion (5 ml) of the globulin preparation (33% SAS
cut) in Tris buffer was applied to a diethylaminoethyl (DEAE)
anion exchange column and eluted in steps with 0.01 M Tris
buffer containing either 0.125 M NaC1, 0.25 M NaC1, 0.5 M
NaC1, or 1.0 M NaC1.
Another portion (18 ml) of the globulin preparation in
PBS/az was applied in 6 ml sample amounts to a 2.5 x 100 cm
Sephacryl S-300 column in order to separate proteins on the
basis of their size. Fractions were eluted with PBS/az at a
30 ml/hr flow rate and collected using a Gilson fraction
collector. Selected eluted protein fractions were reduced by
boiling for 5 minutes in Laemmli sample buffer (Laemmli,
1970) with 2-mercaptoethanol, and examined by SDS
polyacrilamide gel electrophoresis (SDS-PAGE) using a Phastgel
apparatus (Pharmacia LKB, Uppsala, Sweden). Fractions
containing similar protein composition were pooled and
concentrated in centrifuge filter concentrators (Amicon
CentriprepR-10, W.R. Grace & Co, Beverly, MA, USA). Selected
DEAE fractions were used to immunize mice and the gel
filtration fraction pools were used as antigen in preliminary
hybridoma screening protocols.
Immunoalobulin purification by anti-light chain affinity
column chromatographv. An affinity column was prepared using
2 mg monoclonal antibody HL673 which is specific for the
immunoglobulin light chain of the desert tortoise (Schumacher
et al., 1993). Mab HL673 which had been found to cross-react
strongly with putative light chain of the green turtle in
ELISA and Western Blots was covalently linked to a hydrazide
support gel (Affi-prepR Hz, Bio-Rad Laboratories, Richmond,
CA, USA) following manufacturers instructions. Briefly, 1 ml
of purified Mab HL673 (2 mg/ml) was oxidized with 20 pl of
sodium periodate stock solution (0.5 M NaI04) in oxidation
buffer (0.02 M sodium acetate, 0.15 M NaCl, pH 5.0) for 45
minutes at room temperature. The oxidation reaction was
stopped by addition of 5 p1 glycerol. The oxidized antibody
was dialyzed into coupling buffer (0.1 M sodium acetate, 1.0
M NaC1, pH 4.5) and incubated overnight with approximately 2
ml of settled hydrazide support beads. The antibody coupled
beads were then washed with 0.5 M NaC1, 0.01 M phosphate
buffer, pH 7.5 and stored at 40C. The column was prepared,
conditioned with elution buffer (0.1 M glycine, pH 2.7),
washed with PBS/az, and then 1 ml (2 mg) of green turtle
immunoglobulin rich preparation (33% SAS cut) was applied.
After washing the column the bound protein was eluted with
0.1 M glycine, pH 2.7. Fractions (1 ml) were collected and
neutralized with 45 p1 of 1.0 M Tris, pH 9.0. Eluted proteins
were concentrated and examined with SDS-PAGE. These proteins
were also used to immunize mice for hybridoma production.
Purification of anti-DNP antibodies by affinity column
chromatoaraphv. Turtle anti-DNP antibodies were purified
using affinity chromatography (Goetzl & Metzger, 1970; Wofsy
& Burr, 1969). N e-2,4-DNP-lysine (Sigma Chemical Co, St.
Louis, MO, USA) (2 mM in 0.1 M NaHCO3, pH 8.3) was coupled to
cyanogen bromide activated Sepharose 4B (Pharmacia LKB,
Uppsala, Sweden). The DNP-lysine coupled Sepharose was packed
into a column 2.5 x 13 cm (37 ml) and washed with 50 ml
methanol and then equilibrated in borate buffer (0.015 M
NaBO3, 0.15 M NaC1, pH 8.0). The column was further washed
with 1% bovine serum albumin (in PBS/az) and 25% acetic acid
followed by equilibration in high salt borate buffer (0.015 M
NaBO3, 0.5 M NaCi, pH 8.0) before use. Pooled plasma from the
two DNP-BSA immunized turtles was diluted 1:3 in high salt
borate buffer and applied to the column. The column was
washed until the optical density (OD28nm) returned to
baseline, and then any bound turtle anti-DNP antibodies were
eluted with 5 ml 0.1 M 2,4 DNP-glycine (pH 8.6). The eluted
fractions were pooled and concentrated to a 2 ml volume and
then extensively dialyzed against PBS/az. This solution,
containing highly purified turtle anti-DNP antibodies, as
judged by ELISA and SDS-PAGE, was used for final screening of
newly developed monoclonal antibodies. A small aliquot was
dialyzed against 50 mM Tris (pH 7.4) for mass
Mass spectrometrv. Affinity purified turtle anti-DNP
antibodies were submitted to the Protein Analysis Core,
Interdisciplinary Center for Biotechnology Research,
University of Florida for mass spectrometry (Vestec VT 2000,
Perseptive Biosystems/Vestec Mass Spectrometry Products,
Houston, TX, USA).
Mouse immunization protocols. One 6-8 week old female
BALB/c mouse was immunized subcutaneously with 6 Rg of HL673
affinity purified turtle immunoglobulin in Ribi's adjuvant.
Booster immunizations were repeated in two and four weeks.
The final booster was 17 ig of antigen intraperitoneally.
Fusion was performed 4 days after the last inoculation. Two
6-8 week old female BALB/c mice were immunized with a DEAE
fraction (50 gg total protein) containing both 5.7S and 7S
green turtle immunoglobulins (see results) in Ribi's adjuvant
at several subcutaneous sites. Booster immunizations were
performed at 2 weeks, 4 weeks (50 ig antigen per mouse).
Immunizations of both mice (100 gg antigen each) were
continued at 2 to 4 week intervals for a total of 7
immunizations using a combination of the 5.7S and 7S IgY rich
DEAE fraction (25-75 ig) and various IgM-rich preparations
derived from DEAE and Sephacryl S-300 chromatography runs
(45-100 gg). The two mice differed only in the last
immunization. One mouse was rested for about 4 weeks before
its final booster with both IgY and IgM whereas the second
mouse was rested for 10 weeks before its final booster with
turtle IgM alone. Serum anti-turtle titers were checked
periodically by ELISA.
Fusions. Monoclonal antibody production followed the
standard protocol of the Hybridoma Core Laboratory,
Interdisciplinary Center for Biotechnology Research,
University of Florida (Liddell & Cryer, 1991; Simrell &
Klein, 1979). Three independent fusions (one for each mouse)
were carried out. In general, four days following the final
booster immunization, mice were euthanized under
methoxyflurane anesthesia and their spleens removed.
Splenocytes were prepared by mechanical disaggregation,
washed, and fused with log phase SP2/0 mouse myeloma cells in
a 7:1 ratio using 50% polyethylene glycol 1500 media
(Boehringer Mannheim, Germany). After pelleting by
centrifugation at 400 x g for 8 minutes, cells were
resuspended in fusion media (D-MEM plus 1 x Antibiotic-
Antimycotic, 1 x HAT, 25% SP2/0 Conditioned Media, 20% Horse
Serum) (GIBCO, Grand Island, NY, USA) and seeded into 96 well
culture plates (Costar, Cambridge, MA, USA). Wells were
monitored microscopically for growth of hybridomas.
Screening was begun on growth positive wells 10-14 days
post fusion. The supernatants were removed and tested for
antibody reactivity against specific antigens (see below).
Hybridoma cultures of interest were transferred to 24 well
plates and expanded until they could be retested (about 7
days). Hybridoma cultures of interest were safeguarded by
cryopreservation in liquid nitrogen. Selected cultures were
isotyped using an isotyping kit (Amersham Mouse Monoclonal
Antibody Isotyping Kit, Code RPN.29, Amersham, UK) and cloned
by limiting dilution.
Monoclonal Antibody Screening Protocols
Hybridoma culture supernatants were screened against
each of the three turtle immunoglobulin rich pools derived
from the S-300 gel filtration column using enzyme linked
immunosorbent assays (ELISA). Secondary screening was done by
ELISA protocol. A standard ELISA protocol was used for
screening (Schumacher et al., 1993). Each well of a
microtiter plate (Maxisorp F96, NUNC, Kamstrup, Denmark) was
coated with 50 gl of antigen at a concentration of 10 gg/ml
in PBS/Az and incubated at 4C overnight. The wells were
washed four times with PBS/Az containing 0.05% Tween-20 (PBS-
Tween) by an automatic ELISA washer (EAW II, LT-Laboratories,
Salzburg, Austria) and then blocked with 300 il/well of 1%
BSA in PBS/Az at room temperature for 60 minutes or at 40C
overnight. After four more washes, 50 1l of hybridoma culture
supernatant was added to individual wells and incubated at
room temperature for 60 minutes. The wells were washed again
and 50 gl of a 1:1000 dilution of alkaline phosphatase
conjugated rabbit anti-mouse IgG whole molecule (Sigma
Chemical Co, St. Louis, MO, USA) was added to each well.
Following incubation at room temperature for 60 minutes, the
plates were washed 4 times with PBS-Tween and 100 gl of p-
nitrophenyl phosphate disodium (Sigma Chemical Co, St. Louis,
MO, USA) (1 mg/ml prepared in 0.01 M sodium bicarbonate
buffer, pH 9.6 containing 2 mM MgC12) was added to each well
and incubated in the dark at room temperature for 90 minutes.
The optical density of each well at a wavelength of 405 nm
was measured in an ELISA plate reader (EAR 400 AT, SLT-
Laboratories, Salzburg, Austria) at 30,60, and 90 minutes.
Positive and negative controls included on each assay plate
consisted of immune mouse serum and hybridoma cell culture
medium, respectively. Preliminary screens used selected gel
filtration fraction pools (either IgM-rich, 5.7S-rich, or 7S-
rich) as antigen (Figure 1). Later ELISA screens used
affinity purified turtle anti-DNP antibodies (2 gg/ml) as
Immunoblottina (Western blotting). Immunoblotting was
performed to help demonstrate the specificity of our
monoclonal antibodies for immunoglobulin chains. Immunoblots
were prepared following a published basic protocol
(Schumacher et al., 1993). Briefly, 100-150 gg of green
turtle globulins (33% SAS cut) were separated by SDS-PAGE
under reducing conditions, using a precast 10% Tris-glycine
gel (Novex, San Diego, CA, USA) as previously described
(Laemmli, 1970). The proteins were then electrophoretically
transferred from the gel to a nitrocellulose sheet (Schleicher
& Schuell, Keene, NH, USA) using a transfer apparatus (Novex,
San Diego, CA, USA). A Tris-glycine buffer (pH 8.3) in 20%
methanol was used as transfer buffer. Blotting time was 120
minutes at 30 volts. Once the transfer was complete, the
nitrocellulose was blocked immediately with 5% nonfat dry
milk in PBS/Az and incubated at room temperature on a rocker
overnight. The membrane was then washed three times (5
minutes per wash) with PBS-Tween and placed into a trough-
manifold (PR 150 Mini Deca Probe, Hoeffer Scientific
Instruments, San Francisco, CA, USA). Hybridoma culture
supernatants were loaded, 300 gl per channel, and incubated
on the nitrocellulose for 90 minutes at room temperature on a
rocker. The nitrocellulose membrane was washed 3 more times
and then incubated with 300 1l of a 1:1000 dilution of
alkaline phosphatase conjugated rabbit anti-mouse IgG whole
molecule for 90 minutes at room temperature. The membrane was
then removed from the manifold, washed 3 times and developed
with substrate buffer (0.1 M Tris-HCl, 1 mM MgC12, pH 8.8)
containing 44 p1 of nitroblue tetrazolium chloride (NBT) and
33 p1 of 5-bromo-4-chloro-3-indolylphosphate p-toluidine salt
(BCIP) per 10 ml of substrate buffer (Immunoselect, GIBCO
BRL, Gaithersburg, MD, USA). Immunoblots using biotinylated
Mabs followed the same basic procedure except that
biotinylated Mabs diluted in 1% BSA-PBS/az to 1 gg/ml
replaced hybridoma culture supernatants and strepavidin-
alkaline phosphatase (Zymed Laboratories, San Francisco, CA,
USA) replaced the alkaline phosphatase conjugated rabbit
Monoclonal Antibody Purification and Biotinvlation
Selected cloned hybridoma lines were injected i.p. into
Pristane-primed BALB/c mice and the resulting ascites fluid
containing the desired monoclonal antibodies was harvested.
Monoclonal antibodies were purified from ascites by passage
over a Protein G Sepaharose Fast Flow affinity column
(Pharmacia LKB, Uppsala, Sweden) and biotinylated.
Each purified Mab was dialyzed against 0.1 M NaHCO3, pH
8.0 and adjusted to a final concentration of 1.0 mg/ml
(Goding, 1986). Sulphosuccinimidyl-6-(biotinamido) hexanoate
(Immuno Pure NHS-LC Biotin, Pierce, Rockford, IL, USA)
dissolved in dimethyl sulphoxide at 1.0 mg/ml was added (120
gg biotin per mg of antibody) and the mixture was incubated
for 2 hours at room temperature. Following incubation, the
Mabs were dialyzed into PBS/az and stored at 40C.
Cross Species Reactivity of Monoclonal Antibodies
Supernatants from hybridomas producing Mabs specific for
green turtle immunoglobulins were screened by ELISA for
reactions with 33% SAS globulin preparations of 5 other sea
turtle species. The ELISA followed the general procedure but
used each 33% SAS cut at 5 pg/ml coating concentration as
Verification of Monoclonal Antibody and Turtle Antibody
The following experiments were conducted to prove
further that developed Mabs were specific for individual
turtle immunoglobulin classes and would react with turtle
Sandwich ELISA protocol. An antigen capture experiment
was designed to test whether the turtle plasma proteins bound
by each Mab possessed an immunoglobulin light chain. ELISA
plates were coated with 50 gl per well of selected purified
monoclonal antibody (5 gg/ml). Following incubation with
green turtle 33% SAS cut (2 gg/ml), the sandwich was
completed with 1 gg/ml biotinylated HL673 (anti-light chain)
and detected with strepavidin-alkaline phosphatase.
Detection of immune responses to DNP-BSA. Biotinylated
Mabs were used in an ELISA format to measure anti-DNP
antibody responses in turtles immunized with DNP-BSA. The
general ELISA protocol (described above) was used except that
Polysorp plates (Polysorp, NUNC, Kamstrup, Denmark) were
coated with 50 il per well DNP-BSA (1 gg/ml) and blocked with
2.5% casein (pH 7.0). Plasma samples from DNP-BSA immunized
turtles were diluted 1:50 in 1% BSA-PBS/az and serial two-
fold dilutions were tested. Plates were incubated with class
specific biotinylated monoclonal anti-turtle antibody (1
gg/ml in 1% BSA-PBS/az) followed by strepavidin-alkaline
phosphatase (Zymed Laboratories, San Fransisco, CA, USA).
Competitive inhibition assays. Competitive inhibition
ELISA's were used to verify that the turtle plasma proteins,
i.e. antibodies, detected by each Mab were DNP-specific.
First, plasma samples with peak anti-DNP responses from the 2
immunized turtles were serially diluted and incubated at 4C
overnight with increasing concentrations of soluble hapten
(2,4 DNP-glycine pH 7.4 in PBS/az final concentration range:
0-2 mM). These "inhibited" plasma samples were then assayed
for residual antibody activity by ELISA as described above.
Second, plasma samples with peak anti-DNP responses were
serially diluted and mixed with serial twofold dilutions of
rabbit anti-DNP antiserum (Sigma Chemical Co., St. Louis, MO)
with specific anti-DNP antibody concentrations ranging from 0
to 7.5 gg/ml. The residual turtle DNP-specific antibody
activity was then assayed by ELISA.
Turtle globulins eluted from the DEAE column in two
peaks corresponding to 0.125 M NaC1 and 0.25 M NaC1. Analysis
by reducing SDS-PAGE revealed that the 0.125 M NaCl peak
contained three major protein components with approximate
molecular weights of 23 kD, 38 kD, and 65 kD, consistent with
immunoglobulin light chain, 5.7S heavy, and 7S heavy chains
respectively (see Ambrosius, 1976). The 0.25 M peak contained
a mixture of proteins of various sizes, including a 70 kD
band consistent with IgM heavy chain (data not shown).
Turtle globulins separated on the Sephacryl S-300 column
also eluted in two major peaks: an early peak containing
putative IgM and a large late peak containing a mixture of
5.7S and 7S IgY (Figure 3-1A). Fractions were analyzed by
reducing SDS-PAGE and those with similar protein composition
were pooled and the resulting pools were designated as either
IgM-rich, 5.7S-rich, or 7S-rich (Figure 3-1B). These were
used for initial ELISA screening of hybridoma supernatants.
A very small amount of turtle immunoglobulin (< 100 pg)
was purified from the globulin preparation with an anti-light
chain immunoaffinity column. This material contained three
major components of approximately 65 kD, 38 kD, and 23 kD,
consistent in size with turtle 7S and 5.7S heavy chains and
light chain (data not shown).
Turtle anti-DNP antibodies eluted from the DNP-Sepharose
column showed 4 predominant bands on reducing SDS-PAGE. These
bands had molecular weights of approximately 23, 38, 65, and
70 kD as expected of the light chain, 5.7S, 7S, and IgM heavy
chains respectively (Figure 3-2A). This protein was used in
ELISA and western blotting to screen the monoclonal antibody
supernatants and purified biotinylated Mabs. Figure 3-2B
shows representative western blot results for these Mabs and
HL673 (anti-tortoise light chain) and demonstrates the
specificity of each Mab for its target immunoglobulin chain.
The control Mab, HL860 (anti-turtle non-immunoglobulin), did
not react with the affinity purified turtle anti-DNP antibody
An aliquot of affinity purified turtle anti-DNP
antibodies was examined by mass spectrometry. The mass
spectrometer detected two proteins with molecular weights of
120 and 175 kD corresponding to the expected molecular
weights of intact 5.7S and 7S Ig respectively (Figure 3-3).
Figure 3-1. Fractionation of green turtle
immunoglobulins by gel filtration chromatography.
Turtle globulins (33% SAS cut) were applied to a
Sephacryl S-300 column and eluted with PBS/az.
Fractions with similar protein composition by SDS-
PAGE analysis were pooled. Three fraction pools
were produced: IgM rich (fractions 8-18), 7S rich
(fractions 34-40), and 5.7S rich (fractions 50-62).
(A) Elution profile of turtle globulins
fractionated on S-300 column. Protein content of
each fraction was estimated by spectrometry
(OD28onm). (B) Reducing SDS-PAGE (10% Tris-glycine)
of green turtle immunoglobulin rich fraction pools,
stained with Coomassie Blue. Lane 1--molecular
weight markers (kD); lane 2--33% SAS cut; lane 3--
IgM rich pool; lane 4--7S IgY rich pool; lane 5--
5.7S IgY rich pool.
0 20 40 s0
1 2 3 4 5
97 p -
69 > <
1 2 1 2 3 4 5 6 7
S 974 4 974
S69 4 69
N 4 30 b
l4 21.5 I4 30
Figure 3-2. Affinity purified turtle anti-DNP antibody
chains. (A) Coomassie Blue stained 12% Tris-glycine
reducing gel showing turtle anti-DNP antibodies eluted
from a DNP-Sepharose affinity column with 0.1 M DNP-
glycine. Lane 1--turtle anti-DNP antibodies; lane 2--
molecular weight markers. (B) Immunoblot of selected
monoclonal antibodies (Mabs) on affinity purified turtle
anti-DNP antibodies. Each lane was incubated with a
different biotinylated Mab: either HL860 anti-turtle
non-immunoglobulin plasma protein (lane 1), HL846 anti
IgM (lane 2), HL857 anti-7S IgY heavy chain (lane 3),
HL814 anti-5.7S IgY heavy chain (lane 4), or HL673 anti-
tortoise IgY light chain (lane 5). Control lane
contained 1% BSA (lane 6). Lane 7 contained molecular
o o 0 o
o 0o o0 o
o 0 0 0
:) U? 1 8
J -I r- 4 O
-I > m -I O
an 7) Q)
0 4 rl H
,-i ( ~ .H
0 H 0
O, l 0. W .-4
S-r- OC) Or-
-,-1 4 > i (Q
0 (o H CU
44 r- -
-H 0r O o
(10 0 *- U
( 0 -1 01to
H 4 co
-r-i 1 ( A
H 0 04 -Y 0 Md
n0 W W ri M
( O3 3 O O
E0l 0 0 -H 0
L" (-i0 r i
(0 02 4 03-
02- 0H Q-H *>
02-H H C ) --1 (1
0 -20 0 5 E-i i
01 ( O) N 0
() H -A -H> *-H
0 01J *4
H3 u U -H 04 l) N a)
U CH e 3 02 0 rl U-
QY) (1) (3) ie
Ol )*HE* -l 4
OZ 04c 0
*4 Q0 *2H >-Q 0)
I CCO J
0. 0 -*H 0
Q)C 0O 15 C COE-
Native IgM could not be detected by mass spectrophotometry
because the size limit for the method is 200 kD.
Production and Characterization of Monoclonal Antibodies
Hybridoma screening by ELISA against various
immunoglobulin rich fraction pools and Western blotting
against turtle globulins (33% SAS cut) yielded 20 hybridomas
of interest which were retained for further study. The
initial selection of these hybridomas was based on the
specificity of their Mabs for turtle proteins of the
appropriate physical and chemical properties. However these
results were not sufficient proof that these Mabs were
specific for turtle antibodies. Further screening by ELISA
against affinity purified turtle anti-DNP antibodies showed
that only 15 were specific for turtle immunoglobulin classes.
Table 3-1 gives the specificities and isotypes of these 15
Mabs. Ten of these Mabs were specific for 7S IgY heavy chain,
whereas 2 Mabs each were specific for the immunoglobulin
light chain and IgM heavy chain, and only 1 Mab was specific
for the 5.7S IgY heavy chain.
Mabs from the 15 hybridomas that reacted positively with
affinity purified turtle anti-DNP antibodies were tested by
ELISA against serum globulin fractions (33% SAS cuts) from 5
other sea turtle species. Table 3-1 shows that several
monoclonal antibodies reacted with epitopes that are shared
broadly among sea turtle species. Nine of the 7s IgY heavy
chain specific Mabs cross-reacted with all sea turtle
u o u o o o U U
Q QQ QQ
U UU UU UU U
U U U OU U
I I I I I I I
I I 0D
i i o
H H H H H H H H
H H H H H 1-4 H H H H
o r- T
species. Both light chain specific Mabs cross-reacted with
all species except the leatherback turtle. The IgM and 5.7S
Mabs on the other hand seemed to be specific for green turtle
Verification of Monoclonal Antibody Specificitv
Three hybridomas were cloned and their monoclonal
antibodies purified. These were designated HL814 (anti-5.7S
IgY heavy chain), HL846 (anti-IgM heavy chain), and HL857
(anti-7S IgY heavy chain) respectively. These Mabs were used
for further validation experiments. Purified Mabs against
turtle immunoglobulin light chain were not prepared because
of the availability of HL673 (anti-desert tortoise light
Sandwich ELISA. An experiment was designed to test
whether the turtle plasma proteins being bound by each of
these Mabs could be identified as immunoglobulin by the
criterion of having an immunoglobulin light chain. Figure 3-4
shows that proteins, selectively captured from turtle plasma
SAS cut by Mabs HL814, HL846, or HL857, in turn bound
labelled light chain specific Mab HL673, whereas antigen
captured by HL860 (Mab specific for an unidentified 33 kD
turtle protein present in SAS cut) failed to bind Mab HL673.
Antibody Responses to Immunization. Figure 3-5A-D shows
that both HL814 and HL857 could detect rising anti-DNP
antibody responses in both experimentally immunized turtles.
A rise in 7S IgY anti-DNP activity was detected in both
0 0 -
HL814 HL857 HL846 HL860 PBS
Figure 3-4. Sandwich ELISA demonstrating that putative heavy
chain specific Mabs bind plasma proteins with
immunoglobulin light chains. ELISA plates were coated
with 5 pg/ml of each of the following purified Mabs:
HL814 (anti-5.7S heavy chain), HL857 (anti-7S heavy
chain), HL846 (anti-IgM heavy chain), and HL860 (anti-
non-immunoglobulin plasma protein). Following incubation
with 2 gg/ml turtle globulins (33% SAS cut), plates were
washed and incubated with 1 gg/ml biotinylated HL673
(anti-tortoise light chain) to complete the sandwich.
Binding of HL673 was detected with strepavidin alkaline
phosphatase. Data presented are ELISA reactivities
(OD405nm) after 90 minutes incubation with substrate.
turtles within 5 weeks of beginning immunizations and
remained high for the remainder of the experiment. The rise
in 5.7S IgY anti-DNP activity took up to 9 months to reach a
maximum in both turtles. Results for IgM (HL846) were less
clear. One turtle (#4624) showed a weak IgM peak at about 13
weeks (Figure 3-5F), whereas the other turtle (#3150)
appeared to have a high IgM anti-DNP response in the pre-
inoculation sample as well as subsequent samples (Figure 3-
5E). Various modifications of the ELISA protocol, such as
using high salt (0.5 M NaCl) buffer, failed to reduce the
pre-inoculation putative IgM anti-DNP signal. Neither turtle
developed detectable antibody titers to BSA after 10 months
of immunization with DNP-BSA.
Inhibition by soluble hapten. Figure 3-6 (A-C) shows
that the ELISA reactions of immune plasma having peak anti-
DNP antibody titers can be inhibited with increasing
concentrations of soluble hapten (DNP-glycine). Both the
degree of inhibition attained and the shape of the inhibition
curves varied between turtles and among antibody classes
assayed. Inhibition ranged from 72 to 97% with 1 pM DNP-
glycine. Because no anti-BSA antibody responses could be
detected in either turtle, it was impossible to test whether
peak anti-BSA responses would be refractory to inhibition by
soluble DNP-glycine. Nevertheless, results of this experiment
support the conclusion that these Mabs (HL814, HL846, and
HL857) recognize DNP-specific antibodies. The ELISA reactions
Figure 3-5. Development of antibody responses to DNP
with time in 2 chronically immunized turtles.
Serial two-fold dilutions of plasma samples,
collected periodically throughout a prolonged
immunization schedule (10 months), were tested by
ELISA for anti-DNP activity using various
biotinylated Mabs. The rise in OD405nm with time is
indicative of a rising DNP-specific antibody titer.
Turtle 7S IgY (A & B), 5.7S IgY (C & D), and IgM (E
& F) responses were detected by biotinylated HL857,
HL814, and HL846 respectively. Panels A, C, and E
show the responses of turtle #3150 and B, D, and F
show those of turtle #4624.
Figure 3-6. Inhibition of turtle anti-DNP antibody
activity by soluble antigen. Serially diluted
plasma samples with peak 5.7S, 7S, or IgM anti-DNP
antibody activity were incubated overnight with
final concentrations of 2,4 DNP-glycine ranging
from 0 to 1 pM. Samples were then tested by ELISA
for binding to DNP-BSA coated plates. Readings
(OD405n) taken at 60 minutes were plotted against
DNP-glycine concentration. The plasma samples used
were: week 41 for 5.7S and 7S (both turtles), week
1 for #3150 IgM, and week 13 for #4624 IgM (see
Figure 3-5). Data presented are for plasma diluted
1:400 for 5.7S IgY (HL814) and 7S IgY (HL857) and
diluted 1:50 for IgM (HL846). Solid bars (turtle
#3150); crosshatched bars (turtle #4624).
0 10 40 100 200 S00 1000
0 10 40 100 200 500 1000
0 10 40 100 200 500 1000
of immune plasma with peak anti-DNP antibody titers were also
strongly inhibited by increasing concentrations of rabbit
anti-DNP specific antibodies (data not shown).
Sea turtles have 3 major classes of immunoglobulins: a
17 S IgM, a 7S IgY, and a 5.7S IgY (Benedict & Pollard,
1972). IgM is believed to be produced transiently early in an
immune response, as in mammals (Benedict & Pollard, 1977;
Chartrand et al., 1971)- In reptiles, IgM may be the primary
immunoglobulin that is secreted onto mucosal surfaces (Portis
& Coe, 1975). The 7S IgY is believed to function as a serum
antibody like mammalian IgG. The role of 5.7S IgY is unclear,
but evidence suggests that it is a chronic immune response
globulin and that it is maternally transferred to egg yolk
(Benedict & Pollard, 1972, 1977; Chartrand et al., 1971).
The results presented here show that Mabs with
specificity for the light chain and each of the three heavy
chain classes of the green turtle have been produced. The
purification and screening strategies used were dictated in
part by the limited availability of turtles and antigen
specific plasma. Initially, plasma from specifically
immunized green turtles was unavailable, so the preliminary
immunoglobulin purification and hybridoma screening relied on
identification of plasma proteins with the physico-chemical
properties (solubility, size, and charge) consistent with
earlier reports on turtle immunoglobulins (Benedict &
Pollard, 1972, 1977; Leslie & Clem, 1972). Previous
structural studies of turtle antibodies (several species)
indicated that turtle light chains are approximately 22.5 kD
and turtle 5.7S IgY, 7S IgY, and IgM heavy chains are 35-38,
63-68 kD, and 70 kD respectively (Ambrosius, 1976; Benedict &
Pollard, 1977; Chartrand et al., 1971; Leslie & Clem, 1972).
Preliminary screening of fusions yielded a collection of 20
hybridomas that bound to turtle plasma proteins with the
appropriate physico-chemical properties. However, additional
screenings against antigen-specific turtle antibodies
(affinity purified turtle anti-DNP antibodies) revealed that
only 15 of these Mabs could be classified as immunoglobulin
specific. The turtle proteins recognized by the other 5 Mabs
had similar physico-chemical properties but could not be
shown to bind antigen.
Additional experiments were conducted to prove further
that the selected cloned hybridomas produced Mabs that were
specific for turtle antibodies. A sandwich ELISA, using an
anti-light chain Mab demonstrated that the turtle plasma
proteins recognized by each of the heavy chain specific Mabs
possessed a light chain, thereby confirming the specificity
of these Mabs for turtle immunoglobulins. Mabs specific for
turtle immunoglobulin classes should be able to detect an
increasing antibody titer in response to immunization with
specific antigen. Mabs HL814 and HL857 (5.7S IgY heavy and 7S
IgY heavy chain specific, respectively) were able to measure