CANINE HISTIOCYTIC DISEASES: UNRAVELING THE MECHANISMS OF
TUMORIGENESIS
By
Emmalena Gregory-Bryson

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
Comparative Medicine and Integrative Biology – Doctor of Philosophy
2014

ABSTRACT
CANINE HISTIOCYTIC DISEASES: UNRAVELING THE MECHANISMS OF
TUMORIGENESIS
By
Emmalena Gregory-Bryson
Dog breeds provide unique genetic pools for studying rare disorders that affect both dogs and
humans. Understanding the prevalence of these disorders is a key first step in addressing breedassociated diseases and establishing the utility of the dog as a spontaneous model of human disease.
Analysis of the community-based assessment of health status of Bernese mountain dogs (BMD)
was conducted utilizing an open-access database, the Berner-Garde Database. The Berner-Garde
Database was accessed on March 1, 2013 and outputted to SQL, a database management system,
for querying. Complete entries that included sex, age, and diagnosis were chosen, resulting in
7,262 individual dogs for analysis. Nearly 86% of these dogs died by ten years of age. Over 44%
of diagnoses were categorized as cancer; the most prevalent cancer was histiocytic sarcoma, which
comprised 37% of the cancer diagnoses, while lymphoma was the second highest cancer diagnosis at 20%. During the last twenty years cancer has surpassed dysplasias as the most important
diagnosis for BMDs. The prevalence of histiocytic sarcoma diagnoses has grown, which may be
partially due to improved diagnostic criteria. At the same time, the median age at death for BMDs
with histiocytic sarcoma has not improved, but in fact, has decreased over the last decade. Despite
improvement in diagnosis of histiocytic sarcoma, improvements in treatment of this disease have
not followed.
Histiocytic diseases encompass a spectrum of proliferative diseases in dogs and humans. Human histiocytic diseases are rare diseases that affect children and adults, but in BMDs, these diseases are common. Canine histiocytic diseases range from benign histiocytoma to malignant histiocytic sarcoma (HS), which in the disseminated form has a poor prognosis and does not respond
well to chemotherapeutics. The goal of this work is to unravel the mechanisms of tumorigenesis in

canine histiocytic diseases by probing the differential expression of microRNAs (miRNA) along
the spectrum of canine histiocytic diseases. As regulators of gene expression, miRNAs will provide insight into the pathways for tumorigenesis in histiocytic malignancies as compared to benign
histiocytic diseases. MiRNA profiling of canine histiocytic diseases was conducted on cases of
reactive histiocytosis, histiocytic sarcoma, and hemophagocytic histiocytic sarcoma. These samples were compared to normal canine histiocytes derived from peripheral blood and peritoneal
fluid. Analysis of the profiling data yielded several miRNAs, which were selected for validation
by quantitative reverse transcription PCR (qRT-PCR). Data analysis with unsupervised clustering
revealed sets of miRNAs that were overexpressed in BMDs or mixed breed dogs. Validation by
qRT-PCR showed significant upregulation of miR-125b and miR-152 in dogs affected with a histiocytic disease. MiRNAs that are critical for tumorigenesis represent targets for manipulation to
increase our understanding of these malignancies and discover novel and effective therapies.
As part of a concurrent approach to understanding histiocytic disease, the whole genome of a
female spayed BMD with histiocytic sarcoma was sequenced using three genomic DNA samples.
The first was a blood sample taken from this dog when she was 4 years of age, prior to diagnosis
with histiocytic sarcoma. The second was a sample of a mediastinal tumor mass and the third
sample was a cell line derived from an affected abdominal lymph node; both of these samples
were taken at necropsy. There were 29 spontaneous nonsynonymous mutations in the tumor that
were predicted to be deleterious and four of these variants were in genes associated with cancer
including SSH3, ITGB7, and FANCM. Six mutations potentially represented loss of heterozygosity including FAT1, MTUS1, and LRP1B, which are associated with cancer. Over one hundred
variants were considered germ line mutations, the majority of which were heterozygous in all the
samples and fifty-one were homozygous in all three samples. Using these data, future work will
involve evaluating the genes in independent affected and unaffected BMDs to validate the variants
as potential driver mutations of histiocytic sarcoma tumorigenesis.

This dissertation is dedicated to my mother, Pamela Gayle Gregory 1956-2014.

iv

ACKNOWLEDGEMENTS

This work was funded in part by the NIH as a subproject of the Canine Hereditary Cancer Consortium Grant: From Bark to Bedside (NIH UC2CA148149) and from CVM Endowed Research
Funds. I was supported by grant number NIH T32 8T32OD01127 and T32 RR018411 to Michigan
State University. I thank the Berner-Garde Foundation, Bernese mountain dog Club of America,
and Heart of Michigan Bernese mountain dog Club and all the dog owners and breeders for their
participation and support. I would also like to thank Dr. Tuddow Thaiwong for helpful discussions. The pHIT-YFP retroviral vector plasmid was kindly provided by Dr. Steve Suhr. Thank you
to Dr. Nikolaos Dervisis for sharing samples and the mRNA data. This work could not have been
completed without the BD cell line that was derived by Dr. Maciej Parys and the characterization
carried out in part by Dr. Marilia Takada. Bioinformatics assistance was provided by Dr. Matt
Scholz and Peter Schall.

v

TABLE OF CONTENTS

LIST OF TABLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viii
LIST OF FIGURES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

x

KEY TO ABBREVIATIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xv
CHAPTER 1

INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

CHAPTER 2
2.1
2.2

2.3

2.4

COMMUNITY-BASED ASSESSMENT OF HEALTH STATUS IN THE
BERNESE MOUNTAIN DOG BREED . . . . . . . . . . . . . . . . . .
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Materials and Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.2.1 Statistical Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.2.2 Ethics Statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.3.1 Diagnoses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.3.2 Recurrent Histiocytic Disease . . . . . . . . . . . . . . . . . . . . . .
Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1

. 5
. 5
. 6
. 7
. 8
. 9
. 9
. 25
. 25

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

30
30
32
32
34
35
36
37
37
37
45
49

.
.
.
.
.
.
.

.
.
.
.
.
.
.

52
52
53
53
56
56
63

CHAPTER 3
3.1
3.2

3.3

3.4

CANINE HISTIOCYTIC DISEASES: USING MIRNAS TO UNRAVEL
MECHANISMS OF TUMORIGENESIS . . . . . . . . . . . . . . . . .
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Materials and Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.2.1 Samples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.2.2 MiRNA transcriptome profiliing . . . . . . . . . . . . . . . . . . . . .
3.2.3 Validation by qRT-PCR . . . . . . . . . . . . . . . . . . . . . . . . . .
3.2.4 Statistical Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.3.1 Samples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.3.2 MiRNA transcriptome profiliing . . . . . . . . . . . . . . . . . . . . .
3.3.3 Validation by qRT-PCR . . . . . . . . . . . . . . . . . . . . . . . . . .
Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

.
.
.
.
.
.
.
.
.

CHAPTER 4
4.1
4.2

4.3
4.4

WHOLE GENOME SEQUENCING OF A BERNESE MOUNTAIN DOG
WITH HISTIOCYTIC SARCOMA . . . . . . . . . . . . . . . . . . . .
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Materials and Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.2.1 Samples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.2.2 Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

vi

CHAPTER 5

CONCLUSIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66

CHAPTER 6

FUTURE WORK . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68

APPENDIX . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
BIBLIOGRAPHY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90

vii

LIST OF TABLES

Table 2.1

Organization of diagnosis categories in Berner-Garde Database . . . . . . .

Table 2.2

Number of entries and dogs in the Berner-Garde Database available for
analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

Table 2.3

Overall most common diagnoses specified in the Berner-Garde Database . . 14

Table 2.4

Most prevalent cancers listed in the Berner-Garde Database . . . . . . . . . 15

Table 2.5

Congenital and developmental growth disorders in the Berner-Garde Database 18

Table 2.6

Cause of death for dogs less than one year of age at death

Table 2.7

Diagnoses stratified into five-year time increments

Table 2.8

The top diagnoses for each five-year time period indicate cancer incidence
is increasing and dysplasia is decreasing . . . . . . . . . . . . . . . . . . . . . 24

Table 3.1

Signalment of 28 samples analyzed by the miRNA array encompass hemophagocytic histiocytic sarcoma, histiocytic sarcoma, and reactive histiocytosis . . . 34

Table 3.2

MiRNAs are generally conserved among mammals . . . . . . . . . . . . . . 36

Table 3.3

Signalment of 138 samples used for miRNA validation by qRT-PCR encompass hemophagocytic histiocytic sarcoma, histiocytic sarcoma, reactive histiocytosis, and histiocytoma . . . . . . . . . . . . . . . . . . . . . . . . 42

Table 3.4

MiRNAs with significant differential expression in the disease groups . . . . 43

Table 3.5

Functional clusters of miRNA gene targets include immune signaling, migration, and proliferation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

Table 3.6

Results of qRT-PCR validation of miRNAs on samples of hemophagocytic histiocytic sarcoma, histiocytic sarcoma, reactive histiocytosis, and
histiocytoma . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

Table 4.1

Primer sequences used to verify the identity of BD cell line as a unique
cell line . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55

Table 4.2

Summary of variant analysis from whole genome sequencing . . . . . . . . . 58

Table 4.3

Spontaneous nonsynonymous mutations in tumor and cell line . . . . . . . . 60

viii

8

. . . . . . . . . . 19

. . . . . . . . . . . . . . 23

Table 4.4

Loss of heterozygosity variants resulting in nonsynonymous mutations . . . 61

Table 4.5

Germ line variants resulting in nonsynonymous mutations in cancer associated genes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62

Table 4.6

Variants found in the MTAP-CDKN2A locus . . . . . . . . . . . . . . . . . . 64

Table A.1 Samples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
Table A.2 Primer sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
Table A.3 Deletion mutations found in canine GISTs in this study . . . . . . . . . . . . 84

ix

LIST OF FIGURES

Figure 1.1

Schematic of histiocyte lineage and their corresponding markers for identification. Macrophages and dendritic cells develop from hematopoietic stem
cells (HSC) in the bone marrow via the granulocyte-macrophage progenitor
(CFU-GM = colony forming unit - granulocyte, monocyte). There are two
types of dendritic cells that differentiate from this progenitor: Langerhans
cells are in the epidermis for cutaneous immunosurveillance, and interstitial
dendritic cells are antigen presenting cells in other tissues. The cell surface markers that these cells express, listed below each cell type, are directly related to their function and can be used to distinguish these cells
by immunohistochemistry. All leukocytes express β 2 subunit of integrins
(CD18+). Macrophages express αD subunits (CD11d+), which distinguishes
them from the dendritic cells which express αX subunit (CD11c+). The two
types of dendritic cells can be distinguished by E-cadherin, which is an epithelial cell-cell adhesion protein that allows Langerhans cells to interact with
keratinocytes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

3

Figure 2.1

Kaplan-Meier survival plot of the 7,262 selected dogs shows reduced survival for dogs with histiocytic sarcoma. The cumulative survival of dogs
diagnosed with cancers other than histiocytic sarcoma (2,740 of the selected
dogs) are shown by the red line, and dogs with histiocytic sarcoma (1,185 of
the selected dogs) are shown by the blue line. Cancer impacts the survival
of older BMDs who are greater than 5 years of age, however BMDs with
histiocytic sarcoma die earlier than BMDs with other types of cancer. The
median lifespan of dogs with histiocytic sarcoma is significantly less (median
7.4 years) than dogs with other cancers (median 8.3 years) (P < 0.001). . . . . 10

Figure 2.2

Pie chart of the general categories of diagnoses shows a majority of diagnoses are neoplasms. There were 470 distinct disease diagnoses recorded
in the Berner-Garde Database organized into fourteen general system categories. Nearly half of the diagnoses among the 7,262 selected dogs were
categorized as neoplasms which included all cancers, tumors and benign
masses. The second prevalent category was the musculoskeletal category,
which included developmental growth disorders and injuries to bones or ligaments. The “General” category included infections and accidents. . . . . . . . 12

x

Figure 2.3

Pie chart focusing on specific cancers shows histiocytic sarcoma is the
major cancer diagnosis. The most important cancer among the 7,262 selected dogs was histiocytic sarcoma followed by lymphoma. The slice referring to “Other cancers" included fibrosarcoma, leukemia, transitional cell
carcinoma, hemophagocytic histiocytic sarcoma, and other specified cancers
all composing less than 1% each of the total cancer diagnoses. . . . . . . . . . 13

Figure 2.4

Box and whisker plot of the range and distribution of ages at diagnosis
for the top five most prevalent cancers among the 7,262 selected dogs.
Histiocytic sarcoma had a relatively narrow distribution of ages at diagnosis,
while the distribution for lymphoma was broader. Hemangiosarcoma and
osteosarcoma are more commonly diagnosed in older BMDs than the other
cancers. The median age at diagnosis for each cancer is shown by the diamond. Lymphoma had one outlier at 20.2 years of age. . . . . . . . . . . . . . 16

Figure 2.5

The incidence of cancer diagnoses peaks at six years of age. The incidences of the top five major cancer diagnoses are stacked across the age at
diagnosis. The overall peak incidence for histiocytic sarcoma, lymphoma,
mast cell tumor, hemangiosarcoma, and osteosarcoma occurred at six years
of age. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

Figure 2.6

Box and whisker plot of the range and distribution of ages at diagnosis
with a histiocytic disease. Histiocytoma is diagnosed at earlier ages while
histiocytic sarcoma is more variable for age at diagnosis. Among the 7,262
selected dogs, histiocytic diseases with known ages at diagnosis accounted
for 1,267 dogs. Hemophagocytic histiocytic sarcoma n = 20, histiocytic sarcoma n = 1083, histiocytosis, unspecified n = 78, histiocytosis, systemic n =
51, histiocytosis, cutaneous n = 5, histiocytoma n = 30. . . . . . . . . . . . . . 20

Figure 2.7

Incidence of histiocytic disease among the 7,262 selected dogs by age,
including histiocytic sarcoma, hemophagocytic histiocytic sarcoma, and
histiocytoma. Histiocytic sarcoma (left y-axis) shown with the orange line
had a single peak at 8 years of age at diagnosis with 22% of diagnoses being histiocytic sarcoma (219 dogs diagnosed with histiocytic sarcoma at 8
years of age over a total of 1,003 dogs with diagnoses at 8 years of age).
Hemophagocytic histiocytic sarcoma (right y-axis) shown with the blue line
had an overall lower and more variable incidence without a single obvious
peak. Histiocytoma (right y-axis) shown with the teal line peaked at an earlier
age of 1 year of age. Hemophagocytic histiocytic sarcoma n = 20, histiocytic
sarcoma n = 1083, histiocytoma n = 30. . . . . . . . . . . . . . . . . . . . . . 21

xi

Figure 2.8

Incidence of histiocytosis disease diagnoses among the 7,262 selected
dogs by age, including systemic histiocytosis, cutaneous histiocytosis,
and histiocytosis, unspecified. The systemic histiocytosis, shown with the
red line, has two peaks in incidence at 4 years of age and 8 years of age
at diagnosis. The cutaneous histiocytosis group, shown with the green line,
had a variable incidence, most likely due to the small sample size of only 5
dogs being diagnosed with cutaneous histiocytosis. The unspecified histiocytosis group, shown with the purple line, had a wide peak incidence from
6 to 8 years of age. The unspecified histiocytosis diagnosis may include the
previously used diagnosis of malignant histiocytosis, which is now termed
histiocytic sarcoma. Histiocytosis, systemic n = 51, histiocytosis, cutaneous
n = 5, histiocytosis, unspecified n = 78. . . . . . . . . . . . . . . . . . . . . . 22

Figure 3.1

Principal component analysis of the microRNA array data shows three
groupings among the samples based upon their miRNA expression.The
axes represent the first (x-axis) and second (y-axis) principal components and
indicate the direction of the greatest amount of variance among the samples.
The samples were color coded according to their disease type. The four
normal histiocyte samples from mixed breeds are shown in brown as one
group. The four normal histiocyte samples from BMD are shown in red
forming a second group. The disease samples together form a third grouping. . 38

Figure 3.2

Principal component analysis of the microRNA array data for just the
disease samples fails to display any subgroups. The axes represent the first
(x-axis) and second (y-axis) principal components. The normal histiocyte
samples were removed and PCA was conducted on only the disease samples.
No subgroups within the disease samples were found. . . . . . . . . . . . . . . 39

Figure 3.3

Heat map generated using unsupervised clustering shows the differential
expression of the 148 detectable canine microRNAs from the array data.
Red signifies increased expression and green signifies decreased expression
of microRNAs. The miRNAs cluster based upon breed and disease phenotype. On the far left the normal histiocytes cluster together. The affected
Bernese mountain dogs form a cluster in the middle. Other breeds of affected dogs, including golden retrievers and mixed breeds clustered together
on the far right. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

xii

Figure 3.4

Pathways regulated by miRNAs predicted to be important for histiocytic
disease. MiR-143 is predicted to target CD44 molecule, which is associated
with miRNA induced migration, and death-associated protein kinase 1, a positive mediator of programmed cell death. MiR-152, is of particular interest
because it has been shown to play an integral role in the differentiation of
dendritic cells, which are the cell of origin for histiocytic disease. MiR-152
was also over expressed and is predicted to target BCL2-like 11, an apoptosis
facilitator, colony stimulating factor 1 in macrophages, and integrin, alpha 5
(fibronectin receptor, alpha polypeptide), an integrin that if lost can result in
immune dysfunction. MiR-19a was had lower than normal expression in the
reactive histiocytosis samples and is predicted to target the epidermal growth
factor receptor (ERBB3) and SRY (sex determining region Y)-box 5, which
promotes cell survival. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

Figure 3.5

Results from qRT-PCR of selected canine miRNAs validates the upregulation of most miRNAs consistent with the array data. Values are fold
change relative to breed-matched normal canine histiocytes. Symbols denote
the groups which are significantly different from normal as determined by
two-tailed, random variance T-test, Bonferroni corrected: * = P < 0.05, # =
P < 0.005. HS = histiocytic sarcoma. . . . . . . . . . . . . . . . . . . . . . . 47

Figure 5.1

MiRNA regulated pathways. Predicted tumor suppressor mRNA targets of
cfa-miR-125b and miR-152. . . . . . . . . . . . . . . . . . . . . . . . . . . . 67

Figure A.1 Comparison of c-KIT mutations in human and canine GISTs. Human
mutations are in white; dog mutations are indicated in gray [Frost et al., 2003;
Kumagai et al., 2003; Taniguchi et al., 1999; Rubin et al., 2001; Heinrich
et al., 2002]. Codon numbering is based upon human amino acid sequence
[GenBank:NP_000213]. *These are the two deletion mutations found in this
study. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
Figure A.2 Histopathology of GIST. A Hemotoxylin and eosin stain of a GIST. Section
from a gastrointestinal stromal tumor analyzed in this study. B Immunohistochemical staining for KIT. KIT (brown deposits) in a section of a gastrointestinal stromal tumor. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
Figure A.3 Electrophoresis of PCR product of exon 11 of c-KIT, juxtamembrane domain. L = 100 bp ladder, 1 = case with normal exon 11 of c-KIT, confirmed
by sequencing, 2 = case with aberrant banding pattern, + = positive control
(normal dog spleen), - = negative control (water) . . . . . . . . . . . . . . . . 82
Figure A.4 Deletion mutations found in exon 11 of c-KIT. Sequence analysis results
for the six cases with the aberrant banding pattern illustrating the two deletion mutations present in these cases. This region of exon 11 of canine c-KIT
[GenBank:NP_001003181] differs by only one codon from humans [GenBank: NP_000213]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
xiii

Figure A.5 Single nucleotide polymorphisms (SNPs). SNPs, shown in boxed letters,
found in exon 11 of c-KIT and exons 12 and 14 of PDGFRA [GenBank: NC
0006595.2] in the canine GIST samples (R = A or G; Y = C or T). . . . . . . . 85

xiv

KEY TO ABBREVIATIONS

BMD = Bernese mountain dog; DIC = disseminated intravascular coagulation; FFPE: formalinfixed paraffin embedded; GI = gastrointestinal; GIST = Gastrointestinal stromal tumor; GDV =
gastric dilatation volvulus; GR = golden retriever; HCM = histiocytoma; HP = hemophagocytic
histiocytic sarcoma; HS = histiocytic sarcoma; mix = mixed breed dog; PDGFRA: platelet-derived
growth factor receptor, alpha polypeptide; RHC = reactive histiocytosis.

xv

CHAPTER 1
INTRODUCTION

Histiocytic diseases encompass a spectrum of proliferative diseases in both dogs and humans.
The clinical presentation, behavior, and response to treatment are quite dissimilar for the different
histiocytic diseases. In dogs, these diseases range from the mild proliferative lesion histiocytoma, which can spontaneously regress, to the aggressive neoplastic malignancy histiocytic sarcoma (HS), which in its disseminated form, can lead to death or euthanasia within 2 to 4 months
from diagnosis [Zavodovskaya et al., 2006]. Histiocytic diseases are relatively common in dogs.
The neoplastic malignancy HS is especially common among certain breeds of dogs including the
Bernese mountain dog, golden retriever, flat-coated retriever, and Rottweiler. Approximately 25%
of all tumors in the Bernese mountain dog (BMD) breed are diagnosed HS [Abadie et al., 2009].
Unfortunately, there is a dearth of effective treatment options available for histiocytic sarcoma
[Skorupski et al., 2007]. HS is usually disseminated at the time of diagnosis in the BMD and
response to chemotherapeutics is, at best, short-lived [Hafeman et al., 2010].
Human histiocytic diseases are rare and affect both children and adults. HS accounts for < 1%
of all non- Hodgkin’s lymphomas in humans [Ralfkiaer et al., 1990]. Human histiocytic diseases
are so rare that research funding and progress are slow due to inadequate sample sizes of human
patients to carry out randomized therapy trials. The dog is the only species that spontaneously
presents with similar diseases. For example, canine HS is most similar to the human malignancy
Langerhans cell histiocytosis (LCH). LCH is the most common of the human histiocytic diseases
and is estimated to occur in 1 per 200,000 children and 1-2 per 1,000,000 adults [Egeler and
D’Angio, 1995]. Human HS can be a very aggressive disease with a poor prognosis, progressing
to death within a year of diagnosis [Schmitz and Favara, 1998; Vos et al., 2005].
Histiocytic diseases arise from the abnormal accumulation of macrophages and dendritic cells
leading to tumor formation and potentially to organ damage. Macrophages and dendritic cells are

1

subsets of a larger family of cells known as histiocytes. Histiocytes develop from the primitive
CD34+ hematopoietic stem cell. A schematic of histiocyte lineage is shown in Figure 1.1. As
immune cells, the normal function of macrophages is to phagocytize pathogens and cellular debris,
while dendritic cells function as antigen-presenting cells. Since histiocytes are normal residents in
tissues throughout the body, many organ systems can be affected by histiocytic diseases. In dogs at
least four distinct histiocytic proliferative diseases have been recognized. The most benign of the
histiocytic diseases are cutaneous histiocytoma and reactive histiocytosis, the latter of which can
present as a cutaneous or systemic syndrome. The malignant histiocytic diseases encompass the
neoplasms histiocytic sarcoma and hemophagocytic histiocytic sarcoma. Padgett et al. determined
that HS is inherited in the Bernese mountain dog, but research groups are still working on finding
the precise mutation or set of mutations responsible for disease in this breed [Padgett et al., 1995;
Shearin et al., 2012].
More research is needed to understand the biology of histiocytes and the pathogenesis of histiocytic diseases. For this purpose, the dog holds great promise as a model for histiocytic diseases,
to the benefit of humans and dogs. Histiocytic diseases in dogs, although not the same as in humans, is similar enough that study of canine HS may permit insights into the pathogenesis of
this rare human disease, while providing the sample sizes necessary to provide significant data.
In addition, there is a wealth of information that can be gleaned from the stored formalin-fixed,
paraffin-embedded (FFPE) samples that are available from cases at Michigan State University Diagnostic Center for Population and Animal Health. Our lab hosts the Bernese mountain dog DNA
and Tissue Repository, which provides a unique opportunity to access blood and frozen tumor samples from the highly affected BMD. Fresh samples can be obtained from affected patients at the
Veterinary Teaching Hospital here on campus. Therefore, by exploiting these resources in dogs, a
study of the regulators of gene expression could be conducted. Investigation of the transcriptional
changes among these histiocytic diseases can be studied effectively in FFPE samples. Recent technological advances have provided valuable tools for the interrogation of transcriptomes of various
cancers [Barretina et al., 2010]. In this study, microRNA expression will be analyzed to probe

2

Figure 1.1: Schematic of histiocyte lineage and their corresponding markers for
identification. Macrophages and dendritic cells develop from hematopoietic stem cells (HSC) in
the bone marrow via the granulocyte-macrophage progenitor (CFU-GM = colony forming unit granulocyte, monocyte). There are two types of dendritic cells that differentiate from this
progenitor: Langerhans cells are in the epidermis for cutaneous immunosurveillance, and
interstitial dendritic cells are antigen presenting cells in other tissues. The cell surface markers
that these cells express, listed below each cell type, are directly related to their function and can
be used to distinguish these cells by immunohistochemistry. All leukocytes express β 2 subunit of
integrins (CD18+). Macrophages express αD subunits (CD11d+), which distinguishes them from
the dendritic cells which express αX subunit (CD11c+). The two types of dendritic cells can be
distinguished by E-cadherin, which is an epithelial cell-cell adhesion protein that allows
Langerhans cells to interact with keratinocytes.

the transcriptome of canine histiocytic diseases. Each microRNA regulates as many as 200 to 300
gene targets and have proven to play critical roles in the initiation of many cancers [Calin et al.,
2002]. This approach in canine histiocytic diseases will permit the analysis of microRNAs and the
pathways they regulate to increase our understanding of both the normal role of histiocytes and the
aberrant behavior of histiocytes in these tumors.
Since their relatively recent discovery in 1993 [Fire et al., 1998], microRNAs (miRNAs) have
become an active area of research. The miRNAs regulate many cellular processes including proliferation, differentiation, and apoptosis [Kloosterman and Plasterk, 2006]. The miRNAs themselves
are small non-coding RNAs of 21 to 25 nucleotides in length. The precursor pri-miRNAs are

3

frequently transcribed from the introns of protein-coding genes. Within the nucleus, DROSHA
functions to process the pri-miRNA into individual pre-miRNAs, which are approximately 70 nucleotides in length folded into a hairpin structure [Du and Zamore, 2005]. The pre-miRNA is then
exported from the nucleus where it is cleaved by DICER, a member of the Argonaute family of
proteins, to produce the double-stranded miRNAs. One of the strands of the mature miRNA is
assembled into the miRNA-induced silencing complex (miRISC) [Peters and Meister, 2007]. This
complex acts to inhibit protein action by either degrading the mRNA or inhibiting the translation
of mRNA by directly binding to it; in mammals, the predominant action of miRNA is to decrease
protein production by the degradation of the mRNA [Guo et al., 2010]. Recent studies have shown
that miRNAs can act as tumor suppressors or oncogenes, also known as oncomirs, by their regulation of critical genes [Esquela-Kerscher and Slack, 2006]. Previous work has shown that in FFPE
samples miRNAs remain stable due to their small size and can be effectively recovered as compared to the gold standard of RNA recovery from snap- frozen cells [Li et al., 2007]. In addition,
several microarray platforms are commercially available that can determine the expression of all
known miRNAs simultaneously for certain species including dog.

4

CHAPTER 2
COMMUNITY-BASED ASSESSMENT OF HEALTH STATUS IN THE BERNESE
MOUNTAIN DOG BREED

2.1

Introduction

The Bernese mountain dog (BMD) is considered a large breed of dog and is one of four varieties of
Swiss mountain dogs. Originally, the breed worked on farms pulling carts and driving cattle. Today
Berners, as they are nicknamed, are often family dogs, due to their gentle, easygoing personality.
The breed has become more popular over the last decade, climbing from the American Kennel
Club registration ranking from forty-seventh in 2003 to thirty-second in 2013. New owners of
BMDs soon become aware of their unfortunate feature: a short lifespan of only six to eight years,
which is limited primarily by the high incidence of cancers in this breed, specifically histiocytic
sarcoma [Affolter and Moore, 2002].
Histiocytic sarcoma is an aggressive malignancy of dendritic-cell origin. Histiocytic diseases
encompass a range of proliferative diseases in both humans and dogs, from the benign histiocytoma
to the malignancy, histiocytic sarcoma. Histiocytic diseases are very rare in humans; the most
common one, Langerhans cell histiocytosis, has a cumulative incidence of two to ten new diagnoses
per one million children each year [Salotti et al., 2009]. It peaks in frequency in children one to
three years of age, but it can be diagnosed at any age including adults [Nicholson et al., 1998].
The clinical outcomes are disparate; for example, papular skin disease is generally self-limiting,
and histiocytic disease of a single organ system has a 99% survival rate with treatment. However,
for children with multi-system disease, the mortality is estimated as high as 35% [Howarth et al.,
1999; Gadner et al., 2001; Vos et al., 2005]. The rarity of human histiocytic diseases hinders the
development of new therapies due to the inadequate sample sizes of patients. The dog is the only
species that spontaneously presents with a similar disease, in this case clinically and pathologically
resembling Langerhans cell histiocytosis [Affolter and Moore, 2002]. In dogs, histiocytic diseases

5

are more common, especially in certain breeds including the BMD where the diseases histiocytic
sarcoma and systemic histiocytosis were first described [Moore, 1984; 1986; Rosin et al., 1986;
Ramsey et al., 1996]. In a study of dogs affected by histiocytic sarcoma, BMDs were highly
overrepresented, only rivaled by Rottweilers, golden retrievers, Labrador retrievers, and flat-coated
retrievers [Affolter and Moore, 2002]. In dogs, the diseases of dendritic cells range from the
benign histiocytoma to the malignancy, histiocytic sarcoma. Disseminated histiocytic sarcoma at
the far end of the spectrum has a poor prognosis and lacks effective treatment options. It can
progress rapidly in dogs resulting in death or euthanasia within a few weeks to months. Dog breed
associated disorders provide a unique opportunity to study the genetic determinants of diseases
and evaluate the potential of novel treatments [Davis and Ostrander, 2014].
To improve the health and longevity of the breed, BMD breeders started the Berner-Garde
Foundation to document disease incidence among their dogs and join the data with pedigree information in an open-access computerized database, the Berner-Garde Database. This study provides
a community-based assessment of the health status of the BMD breed in congruence with the
Berner-Garde Foundation’s long-term goal of improving BMD health. Specifically, the database
was probed to identify the common disorders affecting the BMD breed and evaluate how the prevalence of these disorders has changed over time.

2.2

Materials and Methods

The Berner-Garde Database is composed of information submitted voluntarily by owners and from
public sources such as the Orthopedic Foundation for Animals (OFA) and previously from the
Canine Eye Registration Foundation (CERF). All BMD owners and breeders are urged to submit data to the Database via online reporting, starting at birth. The Berner-Garde Foundation
was incorporated in 1995 to steward the Database, but it contains data that extend back to the
1970’s. The Database can be publicly accessed through the Berner-Garde Foundation’s website
(www.bernergarde.org). The entire Berner-Garde Database as of March 1, 2013 was outputted to
SQL, a database management system. This analysis included database entries with known sex,
6

age, and diagnosis. The database is organized into individual dogs or entries; one entry contains
a recorded event in the dog’s life such as a specific diagnosis, and so one dog may have multiple entries. For the purposes of this study, some diagnoses were condensed under similar disease
terms. For example, the entries for the previously used diagnosis of malignant histiocytosis were
recoded to the contemporary equivalent histiocytic sarcoma. Layman and scientific terms for the
same diagnosis were combined, thus “dwarfism” was reclassified to “osteochondrodysplasia”. All
versions of the diagnosis for bloat were grouped under the diagnosis “gastric dilatation/volvulus.”
The diagnosis entries for “lymphosarcoma” were grouped together under the general “lymphoma”
diagnosis. To facilitate queries, the Berner-Garde Database organizes the diagnoses under general
system categories. In addition, the diagnoses were grouped under general categories reflecting the
pathological process (Table 2.1). The frequency of histiocytic disease recurring in the same dog
was evaluated with recurrent histiocytic disease defined as multiple different histiocytic disease diagnoses of different types, or the same histiocytic disease diagnosis made more than eight months
apart unless reported to be the progression of the same disease diagnosis.

2.2.1

Statistical Analysis

Survival plots were generated using the Kaplan-Meier method. Statistical analysis performed using R version 3.1.0. Wilcoxon rank-sum (Mann-Whitney) test was used for determining age differences between males and females. Differences between the lifespan of BMDs without cancer,
BMDs with cancer exclusive of histiocytic sarcoma, and BMDs with histiocytic sarcoma were
tested for significance using the log-rank or Mantel-Haenszel test and the Peto and Peto modification of the Gehan-Wilcoxon test. The odds ratio with Fisher’s Exact Ratio test were used to
compare the number of dogs diagnosed with a histiocytic sarcoma to those with recurrent histiocytic disease. A p-value < 0.05 was considered significant.

7

Table 2.1: Organization of diagnosis categories in Berner-Garde Database
Category assigned by the Database

General diagnosis grouping

General

ACCIDENT/TRAUMA/INJURY

Immune System
Behavior Disorders

AUTO IMMUNE DISEASE
BEHAVIORAL PROBLEM

Cancers/neoplasms/tumors

BENIGN MASS, specific cancers

Renal/Urinary
Cardiovascular
Cardiovascular
Reproductive
Digestive
Integumentary

URINARY DISORDER
CARDIAC DISORDER
COAGULATION DISORDER
CONGENITAL DISORDER
DENTAL DISORDER
DERMATOLOGICAL DISORDER

Musculoskeletal

DEVELOPMENTAL GROWTH DISORDER

Eye, Ear, Nose
Digestive
Reproductive
General
Based on body location

EYE DISORDER
GALL BLADDER DISORDER
GESTATIONAL DISORDER
INFECTION
INFLAMMATORY DISORDER

Digestive

INTESTINAL DISORDER

Renal/Urinary
Digestive
Respiratory

KIDNEY DISORDER
LIVER DISORDER
LUNG DISORDER

Nervous System

NEUROLOGIC DISORDER

Digestive
Reproductive
Nervous System
Nervous System

PANCREATIC DISORDER
PROSTATE DISORDER
SEIZURE DISORDER
SPINAL DISORDER

Cardiovascular

SPLENIC DISORDER

Endocrine
Reproductive

THYROID DISORDER
UTERINE DISORDER

Specific diagnoses from the Database
Orthopedic injuries, hit-by-car, poisoning, surgical accidents
Lupus, pemphigus
Aggressive, anxiety, biting, shy
Non-cancerous and cancerous proliferative disorders
Stones, crystals, cystitis, obstruction
All heart diseases excluding cancer
DIC, hemophilia, thrombocytopenia
Birth defects
Gum hyperplasia, poor tooth quality
Skin diseases excluding cancer
Defects in growth not including hip and
elbow dysplasias
All eye diseases excluding cancer
Cholecystitis, obstruction, stones, rupture
Problems in pregnancy
Any infection in any organ
Inflammation “-itis”
GI disorders and obstructions not including GDV or cancer
All kidney diseases excluding cancer
All liver diseases excluding cancer
All lung diseases excluding cancer
All nervous system disorders excluding
cancer
All pancreatic diseases excluding cancer
All prostate diseases excluding cancer
Epilepsy or seizure diseases
Orthopedic spinal problems
All disorders of the spleen excluding cancer
All thyroid diseases excluding cancer
All uterine diseases excluding cancer

The organization of diagnosis entries in the Berner-Garde Database is summarized in this table.
The first column on the left lists the general system categories that were assigned as part of the
internal organization of the Berner-Garde Database. The middle column shows the corresponding
diagnosis categorization that were assigned by this study to reflect the pathological process. The
last column on the right summarizes the specific diagnoses provided in the Database that were
included within each category. DIC = disseminated intravascular coagulation; GI =
gastrointestinal; GDV = gastric dilatation volvulus.

2.2.2

Ethics Statement

The Berner-Garde Database is an open-access database composed of voluntary submissions of self
reported records from BMD owners and breeders. No additional private information other than
that obtained from the database was included in this study. No direct animal experimentation was

8

performed.

2.3

Results

As of March 1, 2013, the Berner-Garde Database contained 92,055 entries belonging to 86,737
individual BMDs (Table 2.2). Selecting for dogs with known sex and date of death yielded 7,262
individual dogs for analysis, which corresponded to 10,690 entries for diagnoses made during the
dogs’ lives. The sex ratio was nearly 1:1 with 3,616 male (49.8%) and 3,646 female (50.2%). The
average age at death was 7.3 years and the median age at death was 7.7 years with a range of 0 to
20.2 years. There was a significant difference (P < 0.001) between males (median 7.4 years, range
0 to 14.8 years) and females (median 7.9 years, range 0 to 20.2 years) for age at death. Of these
dogs, nearly 86% of BMDs had died by the age of ten years (Figure 2.1).

2.3.1

Diagnoses

There were 470 distinct disease diagnoses recorded in the Berner-Garde Database for the 7,262
selected dogs. The Berner-Garde Database organized the diagnoses into fourteen general system
categories. As part of this study the diagnoses were also grouped into 136 general diagnoses
based upon the pathophysiology and affected organ system. Among the 10,690 diagnosis entries,
4,680 (44%) were in the general category of neoplasms followed distantly by 1,435 (13%) in the
musculoskeletal category, which included developmental growth disorders and injuries to bones
or ligaments (Figure 2.2). Excluding unspecific diagnoses like unclassified cancers and tumors,
and unspecified deaths described as old age, the most important specific diagnoses included histiocytic sarcoma, lymphoma, and hip and elbow dysplasias (Table 2.3). Among the cancer diagnoses,
histiocytic sarcoma represented 37% of cancer diagnoses and lymphoma represented 20% of the
cancer diagnoses (Figure 2.3). Together, histiocytic sarcoma and hemophagocytic histiocytic sarcoma composed 38% of the cancer diagnoses (Table 2.4). Among the dogs with cancer, the sex
ratio was 0.996 with 1,959 male (49.9%) and 1,967 female (50.1%) and for dogs with histiocytic

9

1.0
0.6
0.4
0.0

0.2

Cumulative survival

0.8

Other cancers
Histiocytic sarcoma

0

5

10

15

20

Age at death (years)

Figure 2.1: Kaplan-Meier survival plot of the 7,262 selected dogs shows reduced survival for
dogs with histiocytic sarcoma. The cumulative survival of dogs diagnosed with cancers other
than histiocytic sarcoma (2,740 of the selected dogs) are shown by the red line, and dogs with
histiocytic sarcoma (1,185 of the selected dogs) are shown by the blue line. Cancer impacts the
survival of older BMDs who are greater than 5 years of age, however BMDs with histiocytic
sarcoma die earlier than BMDs with other types of cancer. The median lifespan of dogs with
histiocytic sarcoma is significantly less (median 7.4 years) than dogs with other cancers (median
8.3 years) (P < 0.001).

10

Table 2.2: Number of entries and dogs in the Berner-Garde Database available for analysis
All entries in Database
TOTAL
Diagnoses
Death date
Diagnoses, death date, sex
Cause of death
Cancer
Specific cancer
Histiocytic disease

Entries
92,055
15,705
14,319
10,690
9,112
4,762
3,474
1,559

Proportion of entries
100%
17%
16%
12%
10%
5%
4%
2%

Dogs
86,737
10,689
10,193
7,262
7,932
4,352
3,132
1,504

Proportion of dogs
100%
12%
12%
8%
9%
5%
4%
2%

Selected entries
TOTAL
Cause of death
Cancer
Specific cancer
Histiocytic disease

Entries
10,690
5,490
4,297
3,207
1,412

Proportion of entries
100%
51%
40%
30%
13%

Dogs
7,262
5,161
3,926
2,887
1,361

Proportion of dogs
100%
71%
54%
40%
19%

The total numbers of entries and individual dogs from the entire Berner-Garde Database for all
dogs and all years inclusive are shown in the top half of the table with their corresponding
proportions of the total number of entries or dogs. In bold are the number of entries and dogs that
met the selection criteria for this study, which included complete entries with known sex, age, and
diagnosis. This yielded 7,262 individual dogs that are the focus of the bottom half of the table.
Dogs who lacked a recorded date of death, whether due to unknown date of death or the dog was
still alive at the time of the Database output, were not selected for this study.

sarcoma, the sex ratio was 1.07 with 602 male (51.6%) and 564 female (48.4%). For age at diagnosis with cancer, there was a significant difference (P < 0.05) between males (median 7.6 years,
range 0 to 14.1 years) and females (median 7.9 years, range 0.2 to 20.2 years). There was also a
significant difference (P < 0.05) for age at diagnosis with histiocytic sarcoma between males (median 7.1 years, range 1.1 to 12.6 years) and females (median 7.5 years, range 0.6 to 12.8 years).
However, there was no significant difference in survival with histiocytic sarcoma for males (median 0 months, range 0 to 3.3 years) or females (median 0 months, range 0 to 5.2 years). The age
distribution for the top five most common cancer diagnoses among the selected BMDs is shown
in Figure 2.4. The incidence of diagnosis of histiocytic sarcoma, lymphoma, mast cell tumor,
hemangiosarcoma, and osteosarcoma all peak at six years of age (Figure 2.5).

11

Figure 2.2: Pie chart of the general categories of diagnoses shows a majority of diagnoses are
neoplasms. There were 470 distinct disease diagnoses recorded in the Berner-Garde Database
organized into fourteen general system categories. Nearly half of the diagnoses among the 7,262
selected dogs were categorized as neoplasms which included all cancers, tumors and benign
masses. The second prevalent category was the musculoskeletal category, which included
developmental growth disorders and injuries to bones or ligaments. The “General” category
included infections and accidents.
The three most common congenital or developmental growth disorders in decreasing prevalence were osteochondritis dissecans, neonatal death, and cleft palate (Table 2.5). For dogs that
did not reach one year of age, the most common diagnosis besides accidents included neonatal
death, cleft palate, and parvovirus (Table 2.6). The most common specified cause of death, other
than cancer, for dogs that lived beyond on year of age was gastric dilatation/volvulus.
The Kaplan-Meier survival curve for BMDs with cancer drops precipitously after five years of
age, and then is nearly identical to the survival curve of BMDs with any diagnosis over the age of
ten years (Figure 2.1). The lifespan of BMDs with histiocytic sarcoma was significantly shortened

12

Figure 2.3: Pie chart focusing on specific cancers shows histiocytic sarcoma is the major
cancer diagnosis. The most important cancer among the 7,262 selected dogs was histiocytic
sarcoma followed by lymphoma. The slice referring to “Other cancers" included fibrosarcoma,
leukemia, transitional cell carcinoma, hemophagocytic histiocytic sarcoma, and other specified
cancers all composing less than 1% each of the total cancer diagnoses.

13

Table 2.3: Overall most common diagnoses specified in the Berner-Garde Database
Diagnosis
Histiocytic sarcoma
Cancer, unspecified
Lymphoma
Death, unspecified
Hip dysplasia
Elbow dysplasia
Gastric dilatation/volvulus
Tumor / mass, unspecified
Mast cell tumor
Accident/trauma/injury
Cruciate ligament rupture
Hemangiosarcoma
Osteosarcoma
Arthritis
Myelopathy, degenerative

Number diagnosed
1,201
776
652
499
498
410
364
322
297
297
250
243
222
155
123

Proportion of total diagnoses
11%
7%
6%
5%
5%
4%
3%
3%
3%
3%
2%
2%
2%
1%
1%

The most prevalent diagnoses among the selected 7,262 dogs are shown in decreasing frequency.
Diagnoses that affected less than 100 dogs are not shown. For each diagnosis, the number of
entries and the corresponding proportion of all 10,690 diagnosis entries are listed.

(median 7.4 years) compared to survival with other cancers (median 8.1 years) while death not
due to cancer remained relatively constant (median 7.1 years). The range and distribution of age
at diagnosis for the various histiocytic diseases is shown in Figure 2.6. Histiocytic sarcoma and
hemophagocytic histiocytic sarcoma peaked in incidence in dogs five to eight years of age, while
histiocytoma affected a younger set of dogs with a peak incidence at one year of age (Figure 2.7).
Systemic histiocytosis also tended to be diagnosed earlier with a peak in incidence at four years
of age and a second smaller peak at eight years of age (Figure 2.8). The selected dogs were
separated into five-year increments based upon the year they were born (Table 2.7). Over the last
twenty years, the incidence of hip and elbow dysplasias have decreased while cancer and histiocytic
diagnoses have increased in proportion of diagnoses. Qualitatively, this is also shown in Table 2.8,
which lists the top three diagnoses and their relative proportions of diagnoses for each time period.
Since the late 1980s, cancer has replaced dysplasias as the most important diagnosis for BMDs.

14

Table 2.4: Most prevalent cancers listed in the Berner-Garde Database
Diagnosis
Histiocytic sarcoma
Cancer / tumor, unspecified
Lymphoma
Mast cell tumor
Hemangiosarcoma
Osteosarcoma
Carcinoma
Adenocarcinoma
Sarcoma
Melanoma
Fibrosarcoma
Leukemia
Transitional cell carcinoma
Hemophagocytic histiocytic sarcoma
Spindle cell tumor
Squamous cell carcinoma
Round cell sarcoma
Chondrosarcoma
Synovial cell sarcoma
Prostate cancer
Basal cell carcinoma
Nerve sheath tumor
Myeloma
Leiomyosarcoma
Myxosarcoma
Mesothelioma
Mesenchymal neoplasia
Mammary tumor

Number diagnosed
1,201
1,090
652
297
243
221
78
64
59
55
37
36
31
22
21
21
21
15
14
11
11
11
7
6
6
6
6
5

Proportion of specified cancers
37.4%
20.3%
9.3%
7.6%
6.9%
2.4%
2.0%
1.8%
1.7%
1.2%
1.1%
1.0%
0.7%
0.7%
0.7%
0.6%
0.5%
0.4%
0.3%
0.3%
0.3%
0.2%
0.2%
0.2%
0.2%
0.2%
0.2%

The cancer diagnoses listed for the selected dogs are shown in decreasing frequency. Among the
7,262 selected dogs, 3,926 were diagnosed with cancer at some point in their lives. Cancers that
had less than five entries are not shown.

15

Age at Diagnosis (years)!

20!

15!

10!

5!

H

H

O
st
eo

sa

rc

om

m
ar
co
os
gi
an
em

M

a!

a!

!
or
lT
um
as
tC

el

ph
Ly
m

is

tio

cy

tic

Sa

rc

om

om
a!

a!

0!

Figure 2.4: Box and whisker plot of the range and distribution of ages at diagnosis for the
top five most prevalent cancers among the 7,262 selected dogs. Histiocytic sarcoma had a
relatively narrow distribution of ages at diagnosis, while the distribution for lymphoma was
broader. Hemangiosarcoma and osteosarcoma are more commonly diagnosed in older BMDs
than the other cancers. The median age at diagnosis for each cancer is shown by the diamond.
Lymphoma had one outlier at 20.2 years of age.

16

0.5!
0.45!
0.4!

Incidence!

0.35!
0.3!
Osteosarcoma!
Hemangiosarcoma!
Mast Cell Tumor!
Lymphoma!
Histiocytic Sarcoma!

0.25!
0.2!
0.15!
0.1!
0.05!
0!
0!

1!

2!

3!

4!

5! 6! 7! 8! 9! 10! 11! 12! 13! 14!
Age at diagnosis (years)!

Figure 2.5: The incidence of cancer diagnoses peaks at six years of age. The incidences of the
top five major cancer diagnoses are stacked across the age at diagnosis. The overall peak
incidence for histiocytic sarcoma, lymphoma, mast cell tumor, hemangiosarcoma, and
osteosarcoma occurred at six years of age.

17

Table 2.5: Congenital and developmental growth disorders in the Berner-Garde Database
Congenital or developmental disorders
Osteochondritis dissecans
Neonatal death
Cleft palate
Sub aortic stenosis
Umbilical hernia
Shunt, portosystemic
Panosteitis
Anomaly
Persistent pupillary membrane
Cryptorchid (undescended testicles)
Carpus valgus
Size anomalies/birth weights
Monorchid (one descended testicle)
Pulmonic stenosis
Maldevelopment
Patent ductus arteriosis
Sterility
Soft palate
Hypertrophic osteodistrophy
Osteochondrodysplasia
Spina bifida / myelodysplasia
Patent urachus
Congenital heart defect unspecified
Premature growth plate closure
Hermaphroditism

Number diagnosed
55
37
31
31
30
28
20
16
15
13
7
6
6
6
5
4
3
3
3
3
3
2
2
2
2

Proportion of diagnoses
0.51%
0.35%
0.29%
0.29%
0.28%
0.26%
0.19%
0.15%
0.14%
0.12%
0.07%
0.06%
0.06%
0.06%
0.05%
0.04%
0.03%
0.03%
0.03%
0.03%
0.03%
0.02%
0.02%
0.02%
0.02%

There were 350 congenital and developmental growth disorders diagnosed among the 7,262
selected dogs. The most prevalent congenital and developmental growth disorders are shown in
decreasing frequency with their corresponding proportions over all 10,690 diagnosis entries.
Congenital and developmental growth disorders that had only one diagnosis entry among the
selected dogs are not shown.

18

Table 2.6: Cause of death for dogs less than one year of age at death
Cause of death
Accident/trauma/injury
Neonatal death
Cleft palate
Hip dysplasia
Parvo
Renal disease
Cerebellar abiotrophy
Shunt, portosystemic
Death, unspecified
Intussusception
Herpes virus
Elbow dysplasia
Pneumonia
Poison
Renal dysplasia
Infection, unspecified
Intestinal obstruction
Gastric dilatation/volvulus
Sub aortic stenosis
Patent ductus arteriosis
Anomaly
Osteochondritis dissecans
Hemophilia
Cardiac disease
Hepatopathy
Intestinal disorder
Lymphoma

Number of dogs
66
30
21
18
19
13
12
10
9
9
9
8
7
7
6
6
6
5
5
4
4
4
4
4
3
3
3

Proportion of dogs
19.41%
8.82%
6.18%
5.29%
5.59%
3.82%
3.53%
2.94%
2.65%
2.65%
2.65%
2.35%
2.06%
2.06%
1.76%
1.76%
1.76%
1.47%
1.47%
1.18%
1.18%
1.18%
1.18%
1.18%
0.88%
0.88%
0.88%

The cause of death or reason for euthanasia for the selected group of dogs that died at less than
one year of age are shown in decreasing frequency. Among the 7,262 selected dogs, 5,161 dogs
had a known cause of death and 340 of these dogs died before reaching one year of age. Each
cause of death is listed with the corresponding number of dogs and proportion of the 340 dogs
that were less than one year of age at death (right column). Causes of death that affected only one
or two dogs are not shown.

19

Figure 2.6: Box and whisker plot of the range and distribution of ages at diagnosis with a
histiocytic disease. Histiocytoma is diagnosed at earlier ages while histiocytic sarcoma is more
variable for age at diagnosis. Among the 7,262 selected dogs, histiocytic diseases with known
ages at diagnosis accounted for 1,267 dogs. Hemophagocytic histiocytic sarcoma n = 20,
histiocytic sarcoma n = 1083, histiocytosis, unspecified n = 78, histiocytosis, systemic n = 51,
histiocytosis, cutaneous n = 5, histiocytoma n = 30.

20

0.03!

0.025!

0.2!

0.02!
0.15!
0.015!
0.1!
0.01!
0.05!

0.005!

Incidence of other Histiocytic Diseases!

Incidence of Histiocytic Sarcoma !

0.25!

Histiocytic Sarcoma!
Hemophagocytic HS!

0!

0!
0!

1! 2!

3! 4! 5! 6! 7! 8! 9! 10! 11! 12! 13! 14!
Age at diagnosis (years)!

Histiocytoma!

Figure 2.7: Incidence of histiocytic disease among the 7,262 selected dogs by age, including
histiocytic sarcoma, hemophagocytic histiocytic sarcoma, and histiocytoma. Histiocytic
sarcoma (left y-axis) shown with the orange line had a single peak at 8 years of age at diagnosis
with 22% of diagnoses being histiocytic sarcoma (219 dogs diagnosed with histiocytic sarcoma at
8 years of age over a total of 1,003 dogs with diagnoses at 8 years of age). Hemophagocytic
histiocytic sarcoma (right y-axis) shown with the blue line had an overall lower and more variable
incidence without a single obvious peak. Histiocytoma (right y-axis) shown with the teal line
peaked at an earlier age of 1 year of age. Hemophagocytic histiocytic sarcoma n = 20, histiocytic
sarcoma n = 1083, histiocytoma n = 30.

21

0.025!

Incidence!

0.02!

0.015!

0.01!
Unspecified!
0.005!

Systemic!
Cutaneous!

0!
0!

1!

2!

3!

4!

5!
6!
7!
8!
9! 10!
Age at diagnosis (years)!

11!

12!

13!

14!

Figure 2.8: Incidence of histiocytosis disease diagnoses among the 7,262 selected dogs by age,
including systemic histiocytosis, cutaneous histiocytosis, and histiocytosis, unspecified. The
systemic histiocytosis, shown with the red line, has two peaks in incidence at 4 years of age and 8
years of age at diagnosis. The cutaneous histiocytosis group, shown with the green line, had a
variable incidence, most likely due to the small sample size of only 5 dogs being diagnosed with
cutaneous histiocytosis. The unspecified histiocytosis group, shown with the purple line, had a
wide peak incidence from 6 to 8 years of age. The unspecified histiocytosis diagnosis may
include the previously used diagnosis of malignant histiocytosis, which is now termed histiocytic
sarcoma. Histiocytosis, systemic n = 51, histiocytosis, cutaneous n = 5, histiocytosis, unspecified
n = 78.

22

Table 2.7: Diagnoses stratified into five-year time increments
5 year
increments

Dogs

2003 - 2007

1,260

1998 - 2002

2,376

1993 - 1997

1,480

1988 - 1992

770

1983 - 1987

644

1978 - 1982

362

1973 - 1977

128

< 1973

48

All years

7,262

Age
5.8
years
8.1
years
8.6
years
7.8
years
7.4
years
8.0
years
8.8
years
7.9
years
7.6
years

Diagnoses

Hip
Elbow
Cancer
dysplasia dysplasia

Histio

HS

1,805

2%

2%

41%

16%

32%

3,609

3%

3%

45%

15%

30%

2,184

5%

6%

40%

12%

26%

1,113

6%

4%

33%

10%

25%

1,003

9%

4%

37%

11%

24%

514

11%

6%

38%

9%

21%

166

8%

4%

48%

8%

11%

64

19%

3%

30%

6%

16%

10,690

5%

4%

40%

13%

28%

Recurrent Overall Cancer
histio
survival survival
7
1
1%
months month
10
3
1%
months months
13
4
1%
months months
13
4
2%
months months
12
5
1%
months months
5
3
4%
months months
2
3
0%
months months
5
7
0%
months months
10
3
1%
months months

Histio
survival
1
month
2
months
3
months
8
months
5
months
3
months
0
months
6
months
3
month

The 7,262 selected dogs were stratified into five-year time periods based upon their date of birth. The number of dogs, their average
age, and the number of diagnosis entries for each time period are shown within each row with the average across all years shown in the
bottom row and most recent years in the top row. Specific diagnoses hip and elbow dysplasia and the general category of cancer and
histiocytic diseases (Histio) are shown as the proportion of total diagnoses for the given time period. Histiocytic sarcoma (HS) is shown
as the proportion of total cancer diagnoses for the given time period. The incidence of recurrent histiocytic disease in the same dog
(Reccurent histio) was calculated as the proportion of total histiocytic disease diagnoses for the given time period.

23

Table 2.8: The top diagnoses for each five-year time period indicate cancer incidence is increasing and dysplasia is decreasing
5 year
increments
2003 - 2007
1998 - 2002
1993 - 1997
1988 - 1992
1983 - 1987
1978 - 1982
1973 - 1977
< 1973
All years

#1 diagnosis

#1 diagnosis %

#2 diagnosis

#2 diagnosis %

#3 diagnosis

#3 diagnosis %

histiocytic sarcoma
histiocytic sarcoma
histiocytic sarcoma
histiocytic sarcoma
hip dysplasia
hip dysplasia
cancer, unspecified
hip dysplasia
histiocytic sarcoma

13.4%
13.6%
10.5%
8.0%
9.4%
11.3%
15.7%
18.8%
11.2%

cancer, unspecified
cancer, unspecified
cancer, unspecified
cancer, unspecified
histiocytic sarcoma
cancer, unspecified
lymphoma
cancer, unspecified
cancer, unspecified

7.8%
7.6%
6.9%
6.0%
8.8%
8.8%
10.8%
10.9%
7.3%

lymphoma
lymphoma
lymphoma
hip dysplasia
lymphoma
histiocytic sarcoma
hip dysplasia
lymphoma
lymphoma

6.0%
6.4%
5.8%
5.8%
6.8%
8.0%
8.4%
6.3%
6.1%

The three most frequent diagnoses among the 7,262 selected dogs for each five-year time period are shown with the corresponding
proportion of all diagnoses for the time period. Over all years the top three diagnoses were histiocytic sarcoma, unspecified cancers,
and lymphoma. Hip dysplasia has been among the top three diagnoses until the late 1980’s. Since 1993, cancer diagnoses have
encompassed the top three diagnoses in recent years.

24

2.3.2

Recurrent Histiocytic Disease

The risk of BMDs developing recurrent histiocytic disease was evaluated with the goal of determining if benign proliferative histiocytic disease predisposes a BMD to the more aggressive
malignancy histiocytic sarcoma. There were fourteen BMDs that were diagnosed with recurrent
histiocytic disease during their lives. Histiocytic sarcoma was listed as the diagnosis for 1,166
dogs, 9 of which had a history of other histiocytic diseases. Of the remaining 6,096 dogs, 186 dogs
were diagnosed with some other histiocytic disease besides histiocytic sarcoma during their life.
To estimate the risk of a dog developing histiocytic sarcoma given a history of histiocytic disease
the odds ratio of 0.26 (95% CI: 0.12-0.51) was calculated with Fisher’s Exact Ratio test p-value
= 3.658 ∗ 10−6 . One male dog had a cutaneous histiocytoma at 1 year of age that was surgically
removed and two years later had another histiocytoma surgically removed and then was euthanized
with T-cell lymphoma at ten years of age. Seven dogs, two females and five males, had histiocytoma and were diagnosed with histiocytic sarcoma eight months to nine years later. In one case,
a cutaneous histiocytoma was surgically removed from the right elbow of a 7-year-old neutered
male dog, then he was diagnosed with histiocytic sarcoma in the same location eight months later.
Five dogs were diagnosed with localized histiocytic sarcoma, either subcutaneous or periarticular,
and then developed disseminated histiocytic sarcoma six months to one year later. A female dog
was diagnosed with systemic histiocytosis and had a subcutaneous histiocytic mass on her neck
removed when she was four years of age, with no other masses found. At the time she was treated
with cyclosporine and prednisone and appeared to go into remission. Three years later she was
diagnosed with systemic histiocytosis in the face, after a wasp sting that resulted in her swollen
face and paw. Ultimately, she was unresponsive to treatment and was euthanized.

2.4

Discussion

The BMD has a shortened life expectancy, even compared to other large breeds of dogs like Doberman Pinschers which live to eleven years of age [Adams et al., 2010]. This is due primarily to high

25

incidence of cancers in BMDs [Brønden et al., 2010]. Due to the relatively regular influx of European BMDs to American BMD breeding stock, it makes sense that many features of the data
analyzed in our study are consistent with studies conducted on European BMDs [Quignon et al.,
2007]. The data presented here was gleaned from the Berner-Garde database as a communitybased assessment of health status, very similar to the Dutch BMD club’s reporting system in The
Netherlands used in the 2013 study by Erich et al. as opposed to the questionnaire-based studies of
Nielsen et al. 2010 [Nielsen et al., 2010; Erich et al., 2013]. The median age at death determined
for this group of BMDs was 7.7 years, slightly older than the 7.1 years calculated by Nielsen et
al. 2010 but younger than the 8.0 years determined by Erich et al. 2013 and Adams et al. 2010
[Nielsen et al., 2010; Adams et al., 2010; Erich et al., 2013]. Consistent with these previous studies, on average female BMDs lived longer than males, a finding that persisted even when calculated
without the female outlier at 20.2 years of age at death [Nielsen et al., 2010; Erich et al., 2013].
In the last decade,we found the average life expectancy of BMDs to be lower in this cohort with
86% of dogs dying by ten years of age, as compared to the Dutch population, where 84% of dogs
died by eleven years of age [Erich et al., 2013]. This lifespan is indeed shortened when compared
to a recent UK study of multiple breeds of purebred and mixed bred dogs that found 50% of dogs
survived beyond 12 years of age [O’Neill et al., 2014]. BMDs recorded as dying of “old age”
lived on average to 11.5 years, which is similar to other reports of BMDs succumbing to “old age”
at an average of 11 years [Nielsen et al., 2010; Erich et al., 2013]. Those BMDs affected with
histiocytic sarcoma had a shortened lifespan on average of 7.3 years, consistent with other reports
ranging from 6.3 to 7.7 years for affected BMDs [Abadie et al., 2009; Nielsen et al., 2010; Hedan
et al., 2011; Erich et al., 2013]. BMDs with other types of cancer lived almost a year longer to 8.1
years, comparable to the range cited in the literature of living 5 months to 1.5 years beyond their
histiocytic sarcoma affected counterparts [Nielsen et al., 2010; Erich et al., 2013].
Cancer affected 44% of the BMDs in our study, comparable to 42.2% of Danish BMDs and
45.7% of UK BMDs [Nielsen et al., 2010; Adams et al., 2010]. The most common cancer affecting
BMDs is still histiocytic sarcoma, which comprises 37% of the cancer diagnoses in our study as

26

compared to 25% described previously [Padgett et al., 1995; Abadie et al., 2009]. This finding
is consistent with a 2009 study of another breed predisposed to histiocytic sarcoma, flat-coated
retrievers, in which histiocytic sarcoma composed at least 40% of the tumors diagnosed [Dobson
et al., 2009]. A recent study on the Dutch BMD population found an even higher proportion
of tumors at 64.4% to be histiocytic sarcoma [Erich et al., 2013]. Consistent with these results,
Abadie et al. 2009 also did not find a sex predisposition among their group of BMDs affected with
histiocytic sarcoma. However, this study found that females were, on average, four months older
than males at diagnosis with cancer, which persisted among dogs with histiocytic sarcoma. This
sex difference in age at diagnosis did not exist for dogs with the other major cancers, lymphoma,
mast cell tumor, or melanoma. Other cancer diagnoses in our study of BMDs included lymphoma,
mast cell tumor, and melanoma, comprising 6%, 3%, and 0.5% of all diagnoses, respectively.
These results are similar to the Abadie et al. 2009 study of French and European BMDs, which
found lymphoma affecting 4%, mast cell tumor 6%, and melanoma 2% [Abadie et al., 2009]. In
the current study, cause of death was listed for 5,509 dogs (76%) and 822 (14.9%) of those listed
histiocytic sarcoma, which is consistent with the 15.3% found by Erich et al. 2013 [Erich et al.,
2013].
Nearly one-quarter of the cancer diagnoses in our study were unspecified cancer or tumors,
while previous studies listed unspecified cancers composing half or more of the cancer diagnoses
[Nielsen et al., 2010; Erich et al., 2013]. Further investigation of these cases revealed that a more
specific diagnosis was not possible due to insufficient diagnostics, limited by owner interest to pursue additional diagnostics or due to physiologic limitations in sampling including the potential for
uncontrollable bleeding or inaccessible anatomic location. At least 10% of these unspecified cancers listed histiocytic sarcoma as the top rule-out. Approximately 1% of the dogs with histiocytic
disease had distinct recurrent episodes. Our findings indicate that a BMD diagnosed with histiocytic disease is four times more likely to be diagnosed once rather than experiencing recurrent
forms of histiocytic disease. These data suggest that the risk of BMDs escalating to more malignant forms of histiocytic disease is relatively unlikely. A recent study of European BMD found that

27

previous joint disease was associated with the development of peri-articular histiocytic sarcoma in
the same joint [Van Kuijk et al., 2013]. Abadie et al. 2009 found 8% of histiocytic sarcoma cases
had a history of histiocytoma, and they also stated that although no environmental factor had been
identified, they had found two cases of exposure to toxic fertilizers and paint [Abadie et al., 2009].
The possibility of environmental factors interacting with the unique genetic profile of BMDs to
cause the high incidence of histiocytic disease in this breed warrants more investigation.
The top five cancers all peak in incidence at six years of age. The incidence of hip and elbow
dysplasia has been decreasing, likely due to the orthopedic genetic testing through the Orthopedic
Foundation for Animals (OFA) and controlled breeding. A recent study of 74 breeds over the
period of 1974 to 2009 found an overall incidence of hip dysplasia to be 0.83% (range 0.07% to
6%) and elbow dysplasia to be 2.08% (range 0.5%-8%) [Hou et al., 2013]. In our study, over
the whole study period the incidence of hip and elbow dysplasia was found to be 5% and 4%,
respectively, clearly on the higher end of the range found in that study, but, the BMD has appeared
to have made genetic improvement to the incidence of these dysplasias similar to the progress
other highly affected breeds made [Lavrijsen et al., 2014]. In one decade, while cancer diagnoses
increased at a rate of nearly 10%, the proportion of cancer diagnoses due to histiocytic sarcoma
increased by over 15%. This increase may be due to increased awareness and improved diagnostic
criteria. The median age at death for BMDs with histiocytic sarcoma has decreased over the last
decade. On average, BMDs survived one month after diagnosis with histiocytic sarcoma, similar
to the mean survival time of 49 days cited in the 2009 study by Abadie et al [Abadie et al., 2009].
In the Abadie et al. 2009 study, less than 10% of the dogs survived more than four months after
their diagnosis, while in our study less than 7% of dogs survived to four months [Abadie et al.,
2009].
This study is one of the largest sampling of BMDs published so far. These data were derived
from a large community-based database, which demonstrate the utility of such community sourced
data collection. With the advent of patient-accessible electronic health records, similar databases
will be easier to create for human disorders, which will be especially important for tracking inci-

28

dence and clinical progression for rare diseases. Cancer is an important disease for BMDs and is
predominately responsible for their short lifespan. While the incidence of hip and elbow dysplasias
are decreasing in this breed, cancer in general and specifically histiocytic sarcoma increased in incidence. Among neonatal diseases, the incidence of osteochondritis dissecans, and cleft palate
were among the top diagnoses noted in the BMD database, and can be addressed with further
genetic studies. These findings underscore the immediate need for more research to improve the
health of the BMD breed. Specifically, the genetic basis for histiocytic sarcoma in this breed needs
to be identified, and now with the advances in next-generation sequencing, this may be possible
in the near future. In addition, insights gleaned from the BMD regarding the pathophysiology of
histiocytic diseases, have the potential to be informative for understanding this rare set of human
tumors. Several research teams including our own are working diligently to address this need
[Shearin et al., 2012; Boerkamp et al., 2013].

29

CHAPTER 3
CANINE HISTIOCYTIC DISEASES: USING MIRNAS TO UNRAVEL MECHANISMS
OF TUMORIGENESIS

3.1

Introduction

Histiocytic diseases encompass a spectrum of proliferative diseases in both dogs and humans.
These diseases arise from the abnormal accumulation of macrophages and dendritic cells leading to tumor formation and potentially to organ damage. The clinical presentation, behavior, and
response to treatment are quite dissimilar for the different histiocytic diseases. In dogs at least
four distinct histiocytic proliferative diseases have been recognized. The most benign of the histiocytic diseases are cutaneous histiocytoma and reactive histiocytosis, the latter of which can
present as a cutaneous or systemic disease. The malignant histiocytic diseases encompass the neoplasms histiocytic sarcoma and hemophagocytic histiocytic sarcoma. Histiocytic sarcoma (HS)
in its disseminated form, can lead to death or euthanasia within 2 to 4 months from diagnosis
[Zavodovskaya et al., 2006]. While histiocytic diseases are rare in humans, these diseases are relatively common in dogs, especially among certain breeds of dogs including the Bernese mountain
dog, golden retriever, flat-coated retriever, and Rottweiler. Approximately 25% of all tumors in the
Bernese mountain dog (BMD) breed are diagnosed as HS [Abadie et al., 2009]. The diagnosis of
histiocytic diseases is often based on the histopathology of the lesions. Additional confirmation is
provided by immunohistochemistry for the leukocyte surface molecules such as the β 2 integrins
CD18, CD11c or CD11d, which are characteristic for the histiocytes present in these diseases.
Treatment of HS depends upon the presentation of the disease. Unfortunately, there is a dearth of
effective treatment options available for HS for both dogs and humans [Skorupski et al., 2007].
Localized HS or periarticular HS have been cured by early surgical excision [Klahn et al., 2011].
Disseminated HS, on the other hand, is better suited to chemotherapy due to the systemic nature of
this disease. The BMD presents more commonly with the disseminated form of HS and response

30

to chemotherapeutics is, at best, short-lived [Hafeman et al., 2010]. The prognosis for disseminated HS is quite poor, with the disease progressing rapidly resulting in death or euthanasia within
weeks to a few months.
The high incidence of HS in BMDs suggests a genetic component to this disease. Although
Padgett et al. determined that HS is inherited in the BMD, the genes involved in the tumorigenesis
are still unknown [Padgett et al., 1995]. Previous research has evaluated copy number variations
in the BMD and flat-coated retriever using genome wide array comparative genomic hybridization
and identified deletions of the tumor suppressor genes CDKN2A/B, RB1 and PTEN [Hedan et al.,
2011]. The first genome-wide association study of histiocytic sarcoma was conducted in BMD and
revealed a susceptibility locus encompassing MTAP and part of CDKN2A [Shearin et al., 2012].
Recently, several genes involved in DNA repair and replication pathways were associated with
histiocytic sarcoma through gene expression profiling of affected flat-coated retrievers [Boerkamp
et al., 2013].
Human histiocytic diseases are rare and affect both children and adults. HS accounts for < 1%
of all non-Hodgkin’s lymphomas in humans [Ralfkiaer et al., 1990]. Human histiocytic diseases
are so rare that research funding and progress are slow due to inadequate sample sizes of human
patients to carry out randomized therapy trials. The dog is the only species that spontaneously
presents with similar diseases. Human HS can be a very aggressive disease with a poor prognosis,
progressing to death within a year of diagnosis [Schmitz and Favara, 1998; Vos et al., 2005].
More research is needed to understand the biology of histiocytes, which are important components of allergic and infectious diseases, and the cell of origin for histiocytic malignancies. For this
purpose, the dog holds great promise as a model for histiocytic diseases, to the benefit of humans
and dogs. Histiocytic diseases in dogs, although not the same as in humans, is similar enough that
study of canine HS may permit insights into the pathogenesis of this rare human disease, while
providing the sample sizes necessary to provide significant data. In this study, a novel approach
used microRNA (miRNA) expression to probe the transcriptome of canine histiocytic diseases.
Since their relatively recent discovery in 1993, [Fire et al., 1998] miRNAs have become an active

31

area of research. The miRNAs can have as many as 200 to 300 gene targets and therefore regulate many cellular processes including proliferation, differentiation, and apoptosis [Kloosterman
and Plasterk, 2006]. Recent studies have shown that miRNAs can act as tumor suppressors or
oncogenes, also known as oncomirs, by their regulation of critical genes [Esquela-Kerscher and
Slack, 2006]. This discovery has led to a growing number of studies in cancer research to evaluate
miRNAs as regulators of pathways important to tumorigenesis, biomarkers of cancer, and novel
therapeutic targets [Calin et al., 2002]. We hypothesize that along the spectrum of canine histiocytic diseases, differentially expressed miRNAs will provide insight into the critical pathways
for tumorigenesis in histiocytic neoplasias as compared to nonmalignant histiocytic proliferative
diseases. This approach in canine histiocytic diseases will permit the analysis of miRNAs and the
pathways they regulate to increase our understanding of both the normal role of histiocytes and the
aberrant behavior of histiocytes in these tumors.

3.2
3.2.1

Materials and Methods
Samples

Canine histiocytic disease samples included formalin-fixed, paraffin-embedded (FFPE) samples
from the Michigan State University Diagnostic Center for Population and Animal Health collected
between 2006 and 2011. The FFPE samples comprised four canine histiocytic diseases and included cases from Bernese mountain dogs and other breeds: hemophagocytic histiocytic sarcoma,
histiocytic sarcoma, reactive histiocytosis, and histiocytoma. Fresh samples were obtained from
canine patients at the Michigan State University Veterinary Teaching Hospital. Normal mononuclear cells were isolated from approximately 16 ml fresh blood samples using ACCUSPINTM Tubes
and Histopaque -1077 (Sigma-Aldrich, St. Louis, MO) following the manufacturer’s procedure.
The isolated cells were washed three times with 10 ml isotonic phosphate buffered saline (PBS)
with 1:1000 Gibco

Gentamicin (Life Technologies) added and centrifuged 250xg, 10 minutes,

at room temperature. Each sample was then plated into 6-well plates in Gibco

32

RPMI Media

1640 (Life Technologies, Grand Island, NY) supplemented with 10% Gibco
(Life Technologies), 1:1000 Gentamicin, and 1:100 Gibco

Fetal Bovine Serum

Antibiotic-Antimycotic (Life Tech-

nologies). After incubating at 37 ◦ C, 5% CO2 for one to two hours, the media and any floating
cells were removed and discarded and fresh media was added. For macrophage derivation, cell
culture was continued for one week with media refreshed once after three days. For dendritic cell
derivation, 24 hours after plating the factors Human Flt3 Ligand Recombinant Protein (25 ng/ml
12.5 µl, Affymetrix 14-8513-80), Human Granulocyte/Macrophage-Colony Stimulating Factor
Recombinant Protein (800 U/ml, Sigma-Aldrich SRP3050), and Gibco

IL4 Recombinant Hu-

man Protein (500 U/ml 100 µl, Life Technologies PHC0044) were added to promote dendritic
cell differentiation and the cells were cultured for one week with the media and factors refreshed
once after 3 more days. Fresh tumor samples were dissociated using collagenase. Briefly, the
tissue was minced into 3-4 mm pieces, which were washed several times with Hanks’ Balanced
Salt Solution (HBSS), (Sigma-Aldrich). Collagenase, Type I (Sigma-Aldrich) was added to a final concentration of 200 U/ml in HBSS (Sigma-Aldrich) to the tissue pieces and incubated 37 ◦ C,
5% CO2 for 1 hour. The cell suspension was then filtered through a BD FalconTM Cell Strainer,
40 µm (BD, Franklin Lakes, NJ) and the strainer was washed three times with PBS. The cell
suspension was then washed three times with PBS and centrifugation. The final cell pellet was
resuspended in RPMI media supplemented with 10% Fetal Bovine Serum, 1:1000 Gentamicin,
and 1:100 Antibiotic-Antimycotic. After incubating 37 ◦ C, 5% CO2 for 2 hours, the media and
any floating cells were removed and discarded. Histiocyte cell identity for all fresh samples was
confirmed by flow cytometry for CD18 and CD11c cellular markers and evaluated for cellular
morphology by a board certified veterinary clinical pathologist. Total RNA including miRNA was
isolated from the FFPE samples using the QIAGENTM miRNeasy FFPE Kit and from the fresh
samples using the Life TechnologiesTM Ambion

mirVanaTM miRNA Isolation Kit, with phenol

per the manufacturer’s protocols. The isolated RNA was quantified by a NanoDrop spectrophotometer and RNA quality was evaluated by the Agilent Technologies 2100 Bioanalyzer system.

33

3.2.2

MiRNA transcriptome profiliing

MiRNA profiling was performed using the Affymetrix

GeneChip

miRNA Array (Santa Clara,

CA) which contained probes for 71 organisms including dog, for which it could detect 177 canine
miRNAs as acquired from the Sanger miRBase miRNA database v11 (April 15, 2008). The differential expression of miRNAs was compared among three canine histiocytic diseases: hemophagocytic histiocytic sarcoma, histiocytic sarcoma, and reactive histiocytosis. These diseases included
FFPE samples from the following four groups: BMD with hemophagocytic HS, BMD with HS,
golden retrievers with HS, and reactive cutaneous histiocytosis. A fifth sample was the purchased
canine HS cell line, DH82 [Wellman et al., 1988]. All five groups were compared to histiocytes
collected from peripheral blood or peritoneal fluid of breed matched unaffected dogs. Table 3.1
details the signalment data for the samples evaluated using the miRNA array.
Table 3.1: Signalment of 28 samples analyzed by the miRNA array encompass
hemophagocytic histiocytic sarcoma, histiocytic sarcoma, and reactive histiocytosis
Number of
samples

Disease

Breed

Location

5

Hemophagocytic
Histiocytic Sarcoma

Bernese mountain dogs

liver

5

Histiocytic Sarcoma

Bernese mountain dogs

4

Histiocytic Sarcoma

golden retrievers

5

Reactive Histiocytosis

mixed breeds

skin

1

Histiocytic Sarcoma
cell line

golden retriever

bone marrow

4

Normal histiocytes

Bernese mountain dogs

blood

4

Normal histiocytes

golden retriever,
mixed breeds

blood,
peritoneal effusion

liver, lung,
lymph nodes
lymph nodes,
subcutaneous

Age
Range
2-9
years
4-9
years
7-12
years
3-9
years
10
years
2-12
years
1-12
years

28 canine samples were evaluated for miRNA expression by Affymetrix GeneChip miRNA
Array. All samples were from FFPE except the cultured cell line and normal histiocytes. Males
and females were represented approximately equally in each disease.

The miRNA profiling data was normalized using the Affymetrix
34

miRNA QCTool and then

analyzed using Ingenuity

Pathway Analysis software (IPA Spring Release 2014). The miRNA

data was paired with mRNA data to visualize miRNA-mRNA relationships using the microRNA
Target Filter feature of IPA. The mRNA data came from other work in the lab using Affymetrix
GeneChip

Canine Genome 2.0 Array (Santa Clara, CA) on a set of hemophagocytic histiocytic

sarcoma and histiocytic sarcoma diagnosed samples of Bernese mountain dogs and other breeds
[Dervisis et al., 2012]. Gene targets that had the opposite expression pattern to the miRNA (i.e. the
gene target had decreased expression relative to normal samples while the corresponding miRNA
had increased expression or vice versa) were selected for further pathway exploration. The selected
gene list was inputted into DAVID Bioinformatics (http://david.abcc.ncifcrf.gov/) to identify functional clusters of pathways in these samples [Da Wei Huang and Lempicki, 2008].

3.2.3

Validation by qRT-PCR

Fourteen miRNAs were selected for validation by Quantitative Reverse Transcription PCR (qRTPCR) using the TaqMan

MicroRNA Assays available from Life TechnologiesTM . The ca-

nine miRNAs cfa-let-7c, cfa-miR-19a, cfa-miR-28, cfa-miR-29b, cfa-miR-34a, cfa-miR-125b, cfamiR-143, cfa-miR-146a, cfa-miR-146b, cfa-miR-148a, cfa-miR-148b, cfa-miR-152, cfa-miR-155,
and cfa-miR-200c were selected based upon their differential expression from the miRNA array
data and their role in cancer as predicted by a review of the literature. Consistent with many miRNAs, these miRNAs are highly conserved among mammals (Table 3.2 especially between the dog
and human [Lagos-Quintana et al., 2001; Meunier et al., 2013]. Samples for validation included
the samples used for the microarray, and additional FFPE samples for all diseases and included an
additional disease group of histiocytoma samples. A second canine histiocytic sarcoma cell line
(designated BD) derived in lab from a BMD with HS was also evaluated. Primary cultured cells
derived from fresh HS tumors of various breeds were included in the qRT-PCR analysis. Table 3.3
details the signalment data for all samples validated using qRT-PCR. All reactions were run in
triplicate with U6 snRNA used as the endogenous control and a negative water control included on
every plate. MiRNA expression was quantified by the comparative threshold cycle method (∆∆CT)

35

using a breed-matched normal histiocyte sample to compare to where possible (i.e. BMD disease
samples compared to the average of the four BMD normal histiocyte samples and other breeds
disease samples compared to the average of four mix breed and golden retriever normal histocyte
samples).
Table 3.2: MiRNAs are generally conserved among mammals
miRNA
let-7c
miR-19a
miR-28
miR-29b
miR-34a
miR-125b
miR-143
miR-146a
miR-146b
miR-148a
miR-148b
miR-152
miR-155
miR-200c

Dog
100%
100%
100%
100%
100%
100%
100%
100%
100%
100%
100%
100%
100%
100%

Cat
82%
100%
99%
100%
100%
68%
100%
100%
92%
100%
100%
100%
100%
100%

Cow
100%
100%
99%
100%
83%
97%
100%
95%
85%
100%
100%
95%
98%
100%

Human
100%
100%
99%
100%
91%
98%
100%
97%
80%
96%
100%
94%
95%
85%

Mouse
82%
99%
93%
99%
92%
85%
100%
91%
84%
94%
94%
86%
92%
91%

Rabbit
84%
99%
95%
88%
80%
69%
100%
97%
82%
93%
100%
74%
98%
91%

Rat
82%
98%
92%
99%
92%
97%
100%
91%
82%
93%
94%
93%
92%
55%

Sheep
82%
100%
99%
100%
83%
97%
100%
93%
85%
100%
100%
93%
98%
91%

Among the fourteen selected miRNAs, the percent of the dog miRNA sequence that matches the
miRNA homolog in other mammals are listed. MiR-19a, miR-28, miR-143, miR-146a,
miR-148a, miR-148b, and miR-155 were all highly conserved with more than 90% identity
between the dog sequences and the miRNA homologues in the mammals evaluated.

3.2.4

Statistical Analysis

The miRNA array data was visualized with principal component analysis (PCA) using Qlucore
Omics Explorer v.3.0, (Lund, Sweden). Heat mapping with unsupervised and supervised hierarchical clustering was conducted using using R version 3.1.0 to visualize the differences between
the samples and the normal histiocytes. This difference was calculated as the average of the breedmatched normal canine histiocytes subtracted from the normalized sample intensity. One-way
ANOVA was used to determine statistically significant differences between the miRNA profiles of
the disease groups.
36

3.3

Results

3.3.1

Samples

A total of 28 samples representing hemophagocytic histiocytic sarcoma, histiocytic sarcoma, and
reactive histiocytosis were evaluated by Affymetrix

GeneChip

miRNA Array and compared to

normal breed-matched histiocytes. An expanded set of 138 samples representing hemophagocytic
histiocytic sarcoma, histiocytic sarcoma, reactive histiocytosis, and histiocytoma were analyzed by
qRT-PCR for a selected set of 14 miRNAs. Among the FFPE samples an average of 24 µg (range:
1 µg-74 µg) of RNA was isolated with quality estimated by the RNA Integrity Number averaging
2 (range: 1-2.4). For the fresh samples an average of 10 µg of RNA (range: 0.27 µg-85 µg) was
obtained, with the RNA Integrity Number averaging 7.7 (range: 2.3-9.9).

3.3.2

MiRNA transcriptome profiliing

The miRNA expression profiles of six groups of cases representing hemophagocytic histiocytic
sarcoma in BMD, histiocytic sarcoma in BMD, histiocytic sarcoma in golden retrievers, the histiocytic sarcoma cell line DH82, reactive histiocytosis, and normal histiocytes were evaluated by
Affymetrix

GeneChip

miRNA Array for the expression of 177 canine miRNAs. In total, 148

(84%) canine miRNAs were detectable in the samples and 55 of these miRNAs had greater than
two-fold differences in the diseased samples relative to the normal canine histiocytes.
The application of principal component analysis (PCA) to the miRNA array data resulted in
three distinct groupings among the samples: normal histiocytes from mixed breeds, normal histiocytes from BMDs, and all of the disease samples together (Figure 3.1). PCA of just the disease
samples did not result in any apparent sub groupings (Figure 3.2). These data were also analyzed
with unsupervised hierarchical clustering, which revealed sets of miRNAs that were overexpressed
in Bernese mountain dogs or mixed breed dogs (Figure 3.3). The first cluster includes the normal
histiocytes grouping together regardless of breed. The affected dogs form two additional clusters
dependent on breed. The affected Bernese mountain dogs have a identifiable subset of overex-

37

Figure 3.1: Principal component analysis of the microRNA array data shows three
groupings among the samples based upon their miRNA expression.The axes represent the
first (x-axis) and second (y-axis) principal components and indicate the direction of the greatest
amount of variance among the samples. The samples were color coded according to their disease
type. The four normal histiocyte samples from mixed breeds are shown in brown as one group.
The four normal histiocyte samples from BMD are shown in red forming a second group. The
disease samples together form a third grouping.

pressed miRNAs (Table 3.4). Other breeds of affected dogs, including golden retrievers and mixed
breeds form another cluster with a different subset of overexpressed miRNAs. There is also a subset of overexpressed miRNAs that are shared among affected dogs regardless of breed. Supervised
clustering resulted in a similar expression pattern.
A list of gene targets was generated from the miRNA profiling data and corresponding mRNA
data using Ingenuity

Pathway Analysis (IPA). Those gene targets with inverse expression to their

miRNAs were put into DAVID Bioinformatics to find functional clusters of genes and identify
common gene families among the miRNA targets (Table 3.5). The canine histiocytic sarcoma cell
line DH82 was a common sample between the miRNA profiling and the mRNA data and thus the

38

Figure 3.2: Principal component analysis of the microRNA array data for just the disease
samples fails to display any subgroups. The axes represent the first (x-axis) and second (y-axis)
principal components. The normal histiocyte samples were removed and PCA was conducted on
only the disease samples. No subgroups within the disease samples were found.

39

Figure 3.3: Heat map generated using unsupervised clustering shows the differential
expression of the 148 detectable canine microRNAs from the array data. Red signifies
increased expression and green signifies decreased expression of microRNAs. The miRNAs
cluster based upon breed and disease phenotype. On the far left the normal histiocytes cluster
together. The affected Bernese mountain dogs form a cluster in the middle. Other breeds of
affected dogs, including golden retrievers and mixed breeds clustered together on the far right.

40

focus of the IPA microRNA Target Filter. This resulted in 477 highly predicted or experimentally
observed gene targets of 47 miRNAs. The functional annotation clustering algorithm determined
36 functional clusters among these genes, the most significant of which encompassed apoptosis,
growth factors, kinases, and several signaling pathways such as ERBB. The gene target list was
also classified into nine gene groups or families. These gene groups included the fibroblast growth
factors, WNT family members, kinases, SMAD family members, mitogens, ribosomal protein
kinases, calcium/calmodulin-dependent protein kinases, other growth factors, and integrins. A
subset of the paired expression data that has an inverse expression specifically in the BMD samples
revealed five miRNAs and nine gene targets as shown in Figure 3.4.

41

Table 3.3: Signalment of 138 samples used for miRNA validation by qRT-PCR encompass
hemophagocytic histiocytic sarcoma, histiocytic sarcoma, reactive histiocytosis, and
histiocytoma
Number of
samples

Breed

Location

Bernese mountain dogs

spleen, liver,
lymph node, lung

golden retrievers

spleen, liver

8

Hemophagocytic
Histiocytic Sarcoma

Labrador retrievers,
Rottweilers,
mixed breeds

23

Histiocytic Sarcoma

Bernese mountain dogs

12

Histiocytic Sarcoma

golden retrievers

spleen, liver,
lymph node,
lung
spleen, liver,
lung, lymph nodes
spleen, lymph nodes,
subcutaneous

9

Histiocytic Sarcoma

Labrador retrievers,
Rottweilers,
flat-coated retriever,
mixed breeds

spleen,
lymph nodes,
subcutaneous

2

Reactive Histiocytosis

Bernese mountain dogs

skin

3

Reactive Histiocytosis

golden retrievers

skin

23

Reactive Histiocytosis

25

Histiocytoma

10
5

1
1
7

Disease
Hemophagocytic
Histiocytic Sarcoma
Hemophagocytic
Histiocytic Sarcoma

Histiocytic Sarcoma
cell line
Histiocytic Sarcoma
cell line
Histiocytic Sarcoma
primary cultured cells

Various breeds,
mixed breeds
Various breeds,
mixed breeds

skin
skin

golden retriever

bone marrow

Bernese mountain dog

lymph node

Bernese mountain dog,
golden retrievers,
various breeds

lymph node, blood,
subcutaneous

4

Normal histiocytes

Bernese mountain dogs

blood

5

Normal histiocytes

golden retriever,
mixed breeds

blood,
peritoneal effusion

Expanded set of 138 canine samples evaluated for miRNA expression by qRT-PCR using Life
Technologies TaqMan MicroRNA Assays. All samples were from FFPE except the cultured
cell line, primary cells, and normal histiocytes. Males and females were represented
approximately equally in each disease.

42

Age
Range
2-9
years
6-9
years
5-12
years
4-13
years
4-12
years
6-11
years
4-9
years
2-8
years
2-12
years
1-12
years
10
years
8
years
6-12
years
2-12
years
1-12
years

Table 3.4: MiRNAs with significant differential expression in the disease groups
miRNAs
mir-574
mir-150
mir-155
mir-23a
mir-181a
mir-125b
mir-103
mir-20
mir-93
let-7a
mir-26a
mir-99b
let-7e

BMDs
decrease #
decrease #
decrease *
decrease *
decrease #
decrease *
increase *
increase *
increase #
increase #
increase #
increase #
increase #

Others
increase #
increase *
increase #
increase #
increase #
increase #
increase #
increase #
increase #
increase #
increase #
increase #
increase *

miRNAs
mir-146b
mir-23b
mir-1307
mir-146a
mir-27a
mir-99a
mir-27b
mir-532
mir-193a
mir-30b
mir-503
mir-106b
mir-130b
mir-30c
mir-423a
mir-25
mir-151
mir-30a

BMDs
decrease #
decrease #
decrease #
decrease *
decrease *
decrease #
decrease *
increase *
increase *
increase *
increase *
increase *
increase *
increase *
increase #
increase #
increase #
increase #

miRNAs
mir-191
mir-107
mir-24
mir-22
mir-16
mir-30d
mir-34a
let-7c
mir-320
mir-502
mir-140
mir-221
mir-185
let-7f
mir-106a
mir-92a
mir-378

Others
increase *
increase #
increase #
increase #
increase #
increase #
increase *
increase #
increase #
increase #
increase #
increase #
increase #
increase #
increase #
increase #
increase #

Among the detectable miRNAs there were subsets of miRNAs that were differentially expressed
in the disease samples in a breed dependent manner as identified on the heat map. The miRNAs
are listed down the columns in order of increasing difference relative to normal histiocytes. The
first segment of the table on the left lists significantly differentially expressed miRNAs that are
shared among affected dogs regardless of breed. The corresponding direction of the differential
expression is indicated for the BMDs and other breeds of dog as increase being increased
expression relative to normal histiocytes and decrease being decreased expression relative to
normal histiocytes. The subset of miRNAs defining the affected Bernese mountain dogs are listed
in the middle segment of the table. The last table segment on the right lists the significantly
overexpressed miRNAs in other breeds of affected dogs, including golden retrievers and mixed
breeds. Symbols denote significance as determined by two-tailed, random variance T-test: * = P
< 0.05, # = P < 0.005.

43

Table 3.5: Functional clusters of miRNA gene targets include immune signaling, migration, and proliferation
Term
Annotation Cluster 1

Count
Enrichment Score: 2.6

P-Value

Jak-STAT signaling pathway

8

6.38E-05

Natural killer cell mediated cytotoxicity

7

6.45E-05

Chemokine signaling pathway

6

7.96E-03

T cell receptor signaling pathway
Fc epsilon RI signaling pathway
ErbB signaling pathway
Annotation Cluster 2

5
4
4
Enrichment Score: 2.5

8.40E-03
1.65E-02
2.48E-02

Focal adhesion

6

1.32E-02

Regulation of actin cytoskeleton

6

1.65E-02

Annotation Cluster 3
Regulation of cell proliferation

Enrichment Score: 1.9
4

1.10E-02

Genes
IL2RA, GRB2, CREBBP, IFNG, CSF2RB,
PIK3R5, JAK3, PIK3R2
CASP3, GRB2, LCK, IFNG, PIK3R5,
FAS, PIK3R2
IL8, GRB2, GNG2, PIK3R5, JAK3,
PIK3R2
GRB2, LCK, IFNG, PIK3R5, PIK3R2
GRB2, PIK3R5, MAP2K6, PIK3R2
GRB2, TGFA, PIK3R5, PIK3R2
GRB2, PDGFRB, PIK3R5, PDGFD,
PIK3R2, KDR
FGF7, PDGFRB, PIK3R5, PDGFD,
FGF1, PIK3R2
IL2RA, CDKN1B, FGF7, KDR

The most significant pathways that resulted from DAVID Bioinformatics functional clustering are listed. For the miRNAs that were
potentially biologically significant (greater than 2-fold difference from the normal samples), their corresponding gene targets were
inputted to DAVID and three significant functional gene clusters resulted. Annotation Cluster 1 encompassed several immune system
signaling pathways. The second cluster included genes that regulate cell migration. The third cluster included genes responsible for the
regulation of cell proliferation.

44

3.3.3

Validation by qRT-PCR

From the array data, fourteen miRNAs were selected by validation by qRT-PCR. Among the selected miRNAs, miR-19a, miR-29b, miR-148a, and miR-148b were overexpressed in the cluster
of affected BMDs according to the array data. The miRNAs let-7c, miR-28, miR-34a, miR-125b,
miR-146a, miR-146b, and miR-155 were overexpressed in other affected breeds of dogs. The
miRNAs miR-143, miR-152, and miR-200c were overexpressed in all affected dogs on the array
regardless of breed. Validation was performed on an expanded set of samples representing multiple
breeds affected with hemophagocytic histiocytic sarcoma, histiocytic sarcoma, reactive histiocytosis, and histiocytoma, which were all compared to breed-matched normal histiocytes. Along with
the histiocytic sarcoma cell line DH82, the samples for qRT-PCR included a new cell line derived
in our lab from a BMD with histiocytic sarcoma. Figure 3.5 displays the qRT-PCR results for the
miRNA expression as the fold-change relative to breed-matched normal canine histiocytes.
Most disease groups showed upregulation relative to normal histiocytes for each miRNA evaluated as shown in Table 3.6. The largest fold changes were seen for miR-143, which had increased
expression in all disease groups regardless of breed, for an overall average of 10.5 fold increase
in the disease samples. MiR-152 expression was increased an average of 4.1 fold in all disease
groups regardless of breed. MiR-200c had increased expression more than 2-fold in most disease samples except in the HS cell lines and primary cell cultures where the fold change was less
than 2-fold. Let-7c had more than 2-fold increased expression in the hemophagocytic HS and HS
samples only regardless of breed. MiR-125b had an average of 6.9-fold increase in expression in
all affected breeds of dog, except for the BMD, which had an average of 3-fold decrease in expression. MiR-148a had increased expression more than 2-fold in hemophagocytic HS, HS and
histiocytoma samples regardless of breed. Hemophagocytic HS samples in all breeds and in the
HS samples from BMD. MiR-155 was increased 3.9-fold in most affected breeds of dog, except
for the BMD, which had an average of 1.4 fold decreased expression. MiR-148b was increased
more than 2-fold in BMD with hemophagocytic HS and in the HS primary cell cultures. MiR-34a
was increased more than 2-fold in breeds with hemophagocytic HS and histiocytoma, but not in

45

Figure 3.4: Pathways regulated by miRNAs predicted to be important for histiocytic
disease. MiR-143 is predicted to target CD44 molecule, which is associated with miRNA induced
migration, and death-associated protein kinase 1, a positive mediator of programmed cell death.
MiR-152, is of particular interest because it has been shown to play an integral role in the
differentiation of dendritic cells, which are the cell of origin for histiocytic disease. MiR-152 was
also over expressed and is predicted to target BCL2-like 11, an apoptosis facilitator, colony
stimulating factor 1 in macrophages, and integrin, alpha 5 (fibronectin receptor, alpha
polypeptide), an integrin that if lost can result in immune dysfunction. MiR-19a was had lower
than normal expression in the reactive histiocytosis samples and is predicted to target the
epidermal growth factor receptor (ERBB3) and SRY (sex determining region Y)-box 5, which
promotes cell survival.

BMD, which had more than 2-fold decrease in expression in the reactive histiocytosis samples.
MiR-146a was decreased more than 2-fold in BMD with hemophagocytic HS and in the cultured
cells, but was increased in other breeds of dog with hemophagocytic HS. MiR-19a was increased
2-fold in the hemophagocytic HS samples but decreased in the BMD HS cell line and in reactive
histiocytosis samples in dogs other than BMD. MiR-146b was decreased more than 2-fold in the
BMD HS cell line and increased reactive histiocytosis samples in dogs other than BMD. MiR-29b
was increased more than 2-fold in BMD cases of hemophagocytic HS and HS but decreased in
reactive histiocytosis samples in dogs other than BMD. MiR-28 was decreased more than 2-fold in
the histiocytoma samples only.

46

Figure 3.5: Results from qRT-PCR of selected canine miRNAs validates the upregulation of
most miRNAs consistent with the array data. Values are fold change relative to breed-matched
normal canine histiocytes. Symbols denote the groups which are significantly different from
normal as determined by two-tailed, random variance T-test, Bonferroni corrected: * = P < 0.05,
# = P < 0.005. HS = histiocytic sarcoma.

47

Table 3.6: Results of qRT-PCR validation of miRNAs on samples of hemophagocytic histiocytic sarcoma, histiocytic sarcoma,
reactive histiocytosis, and histiocytoma
Group
BMD
HP
Other
HP
BMD
HS
Other
HS
BMD
cell line
Other
cell
lines
BMD
RHC
Other
RHC
Other
HCM

miR143

miR152

miR200c

let-7c

miR125b

miR148a

miR155

miR148b

miR34a

miR146a

miR19a

miR146b

miR29b

miR28

12.42

5.25

2.22

4.39

-1.00

3.92

-1.62

2.52

-0.62

-3.39

2.56

-0.39

2.18

-0.36

14.57

6.32

4.16

8.44

9.15

3.75

5.99

1.58

4.76

2.72

6.43

1.75

-1.77

1.04

14.02

4.94

4.61

3.07

-0.53

2.31

-1.16

1.34

-0.51

-1.47

1.33

0.75

2.81

-0.44

13.27

4.35

6.11

3.25

6.68

1.53

3.29

0.92

1.61

0.80

-1.43

-0.25

0.49

-0.41

5.24

0.38

1.48

-0.29

-6.73

1.05

-4.05

-0.12

-0.47

-2.12

-2.02

-3.33

-0.11

-1.79

-1.00

7.92

-1.35

0.51

8.05

0.47

-0.16

2.58

1.82

-3.81

-0.87

-0.65

0.02

1.56

12.19

1.29

3.32

NP

-3.84

0.17

1.35

1.13

-2.98

-0.89

NP

1.50

NP

1.12

12.33

3.29

5.91

0.02

5.82

1.23

4.59

0.87

1.02

1.00

-4.38

2.11

-2.08

0.43

11.66

3.42

6.78

NP

5.03

2.12

5.55

0.75

2.32

0.12

NP

-0.02

NP

-2.00

The average fold-change for all miRNAs validated by qRT-PCR are shown for each disease group (listed vertically) and miRNA (listed
horizontally at top). The disease groups are also divided by breed: BMDs and other breeds of dogs. The miRNAs are in order of overall
fold-change across all groups, decreasing from left to right. Bolded numbers indicate P < 0.05 as determined by two-tailed, random
variance T-test. Other = all other breeds of dog beside BMD, HP = Hemophagocytic HS, Other cell lines = primary cell cultures from
breeds of dog besides the BMD, RHC = Reactive Histiocytosis, HCM = histiocytoma, NP = not enough sample for analysis to be
performed.

48

3.4

Discussion

The cases of histiocytic disease used in this study were isolated from FFPE and fresh samples.
While our histiocyte derivation protocol did limit the number of cells isolated from fresh samples
and therefore the quantity of RNA, the RNA quality was acceptable with an average RNA Integrity
Number of 7.7. The FFPE samples yielded more RNA on average, but the quality was substantially
lower with an average RNA Integrity Number of 2, which was expected from this type of sample.
Several studies have demonstrated that FFPE samples correlate well with matched fresh frozen
samples for miRNA expression [Li et al., 2007; Xi et al., 2007; Doleshal et al., 2008; Zhang
et al., 2008; Glud et al., 2008; Hui et al., 2009; Mortarino et al., 2010]. The cross-linking that
occurs during the processing of FFPE samples is well-known to inhibit total RNA yield and quality.
However, miRNAs appear to remain stable due to their small size and can be easily recovered from
FFPE samples as compared to the gold standard of RNA recovery from snap-frozen cells [Li et al.,
2007]. Therefore, despite the low RNA quality, these samples were considered suitable for miRNA
expression profiling.
For many of the miRNAs the array data and qRT-PCR data were in concordance with one another and varied mostly in the magnitude of differential expression relative to the normal canine
histiocytes. In general, the reactive histiocytosis group showed higher increased expression by
the array data than by qRT-PCR. Although both the Affymetrix GeneChip
the TaqMan

miRNA array and

MicroRNA Assay use the same probe sequence targeting these miRNAs, there are

mechanistic differences between the two methods resulting in these disconcordant data. This discrepancy between array and qRT-PCR data is not an uncommon finding in the literature [Rajeevan
et al., 2001]. The miRNA array was intended to be used for the discovery of potentially interesting
miRNAs and the qRT-PCR results are considered the gold standard.
The PCA results revealed a strong relationship between the miRNA expression and the diseased samples relative to the normal histiocytesdxfs. The heat map confirmed this relationship
and further identified breed-dependent differential miRNA expression among the disease samples.
Several miRNAs were selected for validation by qRT-PCR, nine of which were determined bio49

logically significant as predicted by a greater than two-fold change. The miRNAs with the highest
fold-change relative to normal histiocytes include miR-143, miR-152, and miR-125b. In the literature, miR-143 is most commonly associated with being a tumor suppressor miRNA, however
it is predicted by miRDB (www.http://mirdb.org) and TargetScan (http://www.targetscan.org/) to
target several tumor suppressors in the dog including IGFBP5 and TSHR, thus potentially fulfilling
an oncomir role. MiR-152, an immune regulatory miRNA, is of particular interest because it has
been shown to play an integral role in the differentiation of dendritic cells, which are the cell of
origin for histiocytic diseases. MiR-152 is predicted by miRNA prediction algorithms to target the
tumor suppressor gene CCDC6. MiR-125b, is overexpressed in several cancers, including prostate
cancer, leukemia, urothelial carcinomas and gastric cancer. It inhibits several important tumor
suppressor genes including BAK1 and E2F2. Of the validated miRNAs, miR-125b, miR-146a,
miR-155, miR-29b, and miR-34a have been associated with cancer in human studies [Lewis et al.,
2005; Wang et al., 2008]. MiR-125b, which had increased expression in nearly all disease samples,
also had decreased expression of the gene targets BAK1, CYP24, EGFR, and TP53 from the mRNA
array data. This pattern has been associated with migration, invasion and epithelial-mesenchymal
transition of prostate cancer and therapy resistance in breast cancer [Shi et al., 2007; Zhou et al.,
2010]. Increased expression of miR-155 leading to decreased expression of SOCS1 has a role in
cancer initiation in breast cancer [Iorio et al., 2005].Tumor suppressor miR-34a under expression
and corresponding SIRT1 over expression may have a role in early tumorigenesis of the prostatic
epithelial cell [Fujita et al., 2008].
The pathway analysis using Ingenuity

and DAVID resulted in several interesting clusters

among the miRNA gene targets. Genes involved in apoptosis and growth clearly play a role in
cancer development. The clusters of integrin genes and those involved in T cell receptor signaling
and cytotoxicity also have a functional role in immune cells such as histiocytes, the cell of origin
for histiocytic diseases. Examination of the paired expression data with inverse expression among
the BMD samples revealed nine interesting gene targets, including several cancer associated genes
such as DAPK1, CD44, SOX5, and ITGA5. DAPK1 is a positive mediator of programmed cell

50

death and has been shown to be decreased in several highly aggressive metastatic tumors similar to
the decreased expression observed in the BMD hemophagocytic histiocytic sarcoma samples [Inbal et al., 1997]. In a human breast cancer cell line, CD44 expression was inversely correlated with
metastasis as part of a miRNA induced migration phenotype [Huang et al., 2008]. Interestingly,
the BMD hemophagocytic histiocytic sarcoma samples in this study exhibited decreased CD44 expression similar what was described for the breast cancer cell line, which suggests that CD44 may
also play a role in the highly aggressive phenotype of hemophagocytic histiocytic sarcoma. SOX5
is a transcriptional regulator that when amplified, was correlated with reduced apoptosis in human
testicular seminomas [Zafarana et al., 2002]. SOX5 exhibited over expression in nearly all diseased
BMD samples analyzed and perhaps reduced apoptosis is also playing a role in these tumors. Previous work has demonstrated that the loss of integrins on leukocytes results in immune dysfunction,
a phenomenon that may also be occurring in the the histiocytic sarcoma samples in our study which
exhibited decreased expression of ITGA5 [Travis et al., 2007]. Interestingly, miR-125b and miR181a are predicted to target the p14/p19ARF tumor suppressor gene, also known as CDKN2A,
which has previously been associated with histiocytic disease in dogs [Hedan et al., 2011; Shearin
et al., 2012]. Identification of the pathway members with aberrant expression patterns will provide insight into the pathways responsible for tumorigenesis in histiocytic malignancies. MiRNAs
that are critical for tumorigenesis would potentially represent targets for manipulation to facilitate
the development of novel and more effective cancer therapies. By further exploring the pathways
regulated by these miRNAs, we expect they will prove to be instrumental in understanding the
tumorigenesis of histiocytic diseases for the benefit of dogs and humans.

51

CHAPTER 4
WHOLE GENOME SEQUENCING OF A BERNESE MOUNTAIN DOG WITH
HISTIOCYTIC SARCOMA

4.1

Introduction

Every dog breed can be easily identified by the physical traits defining that breed. Careful selection
and often inbreeding have been used for thousands of generations to create the dog breeds of today.
A specific set of physical and behavioral characteristics are determined as breed standard based
upon the historic use of the dog breed, for example hunting or herding or as a lap dog. However, in
the pursuit of the ideal breed specimen, sometimes rare genetic mutations and subsequent health
problems are inadvertently amplified in these restricted gene pools. While the health implications
of deleterious inherited traits must be addressed, dog breeds also provide a unique opportunity
to study what are otherwise rare genetic diseases in dogs and humans. The advent of relatively
inexpensive whole genome sequencing is a powerful new scientific tool that will facilitate this sort
of genetic study. Through the accrual of more genome sequences from carefully curated samples,
the genetic basis for inherited diseases in these dog breeds will be identifiable. In this work, we
present the whole genome sequencing of a Bernese mountain dog with histiocytic sarcoma.
Histiocytic sarcoma (HS) is an aggressive neoplasia arising from interstitial dendritic cells.
Approximately 25% of all tumors in the Bernese mountain dog (BMD) breed are diagnosed as HS
[Abadie et al., 2009]. HS is known to be familial in BMDs, and has been hypothesized to be a
polygenic trait [Padgett et al., 1995; Abadie et al., 2009]. Recent studies have utilized comparative genomic hybridization, genome-wide association, and gene expression profiling to identify the
genetic component of histiocytic sarcoma in BMDs and other highly affected breeds like the flatcoated retriever. Hedan et al. 2011 evaluated copy number variations in the BMD and flat-coated
retriever using genome wide array comparative genomic hybridization. This study identified deletions of the tumor suppressor genes CDKN2A/B, RB1 and PTEN present in affected dogs of both

52

breeds [Hedan et al., 2011]. The first genome-wide association study of histiocytic sarcoma was
conducted in BMDs of American and European origin and revealed a susceptibility locus encompassing MTAP and part of CDKN2A, one of the tumor suppressors also identified by the Hedan et
al. 2011 study [Shearin et al., 2012]. One of the latest research papers on this subject conducted
gene expression profiling of affected flat-coated retrievers and found several genes involved in
DNA repair and replication pathways associated with histiocytic sarcoma in these dogs [Boerkamp
et al., 2013]. Here we present the results of the first study using whole genome sequencing to probe
the genetic variations present in a Bernese mountain dog with histiocytic sarcoma.

4.2
4.2.1

Materials and Methods
Samples

The samples were derived from a spayed female Bernese mountain dog diagnosed with histiocytic
sarcoma in the lungs at 8 years of age. Initially, she was treated with the chemotherapeutic agent
Actinomycin D. She survived for an additional year before succumbing to disseminated histiocytic
sarcoma of the lung, liver, stomach, adrenal gland, mediastinum, omentum, and retroperitoneum
at 9 years of age. Three samples from the same dog were submitted for whole genome sequencing.
Genomic DNA was isolated from a blood sample taken from this dog when she was 4 years of age,
four years before her diagnosis with histiocytic sarcoma. A tissue sample was taken from a mediastinal mass at necropsy and snap-frozen in liquid nitrogen and kept at −80 ◦ C until genomic DNA
isolation. In addition, a cell line was derived from a sample of an abdominal lymph node taken
during necropsy. From the blood sample, genomic DNA was isolated from 200µl of whole blood
using 400µl lysis buffer (50 mM Tris pH 8.0, 0.5% Tween20, 1 mM ethylenediaminetetraacetic
acid) and 40µl proteinase K (15 mg/ml). Cell lysis and protein digestion occurred overnight at
42 ◦ C. The following day the proteinase K was inactivated by incubating at 95 ◦ C for ten minutes.
The DNA was extracted using a phenol:chloroform:isoamyl alcohol protocol. The frozen tumor
tissue was ground using a mortar and pestle and then genomic DNA isolation proceeded similar to

53

the blood sample. The cell line was established in-house and characterized as spontaneously immortalized CD11c positive histiocytes [Takada et al., 2014]. The cells were washed with phosphate
buffered saline and lysed directly on the 10 cm cell culture plate using 1000µl lysis buffer (50 mM
Tris pH 8.0, 0.5% Tween20, 1 mM ethylenediaminetetraacetic acid) and 100µl proteinase K (15
mg/ml). The cells were released with a disposable rubber spatula and the lysate was pipetted into
a tube. Cell lysis and protein digestion occurred overnight as previously described and genomic
DNA isolation proceeded as described for the blood sample. The microsatellite markers of the cell
line were determined using the primers listed in Table 4.1 and evaluated using gel electrophoresis
[Neilan et al., 1997; Irion et al., 2003].

54

Table 4.1: Primer sequences used to verify the identity of BD cell line as a unique cell line
Marker
1
2
3
4

Primer
F
R
F
R
F
R
F
R

Sequence
TTGGACCTGAGTTGTGGATG
CTGGGCTCCACTACATGGAA
GGGTTGATGTCAGGTCAGG
GGGTCCTCACAAGAGAGTATGG
CAAAAAATATGCAATGAAAAAGAGG
GGACAAGGATAGGAGGCAGAC
CCAAGGGAAGGAAGTTTAAGG
TTTTCTCTTTCTGTCTTTCTTGATG

Motif
(TTTC)n

Reference size
299

Location
chr4:66543470-66543646

(GAAA)n

372

chr5:61523795-61524073

(GAAA)n

390

chr9:37808480-37808650

(TCTA)n

239

chr14:21564592-21564721

Primer sequences used to verify the microsatellite markers of the BD cell line are distinct from other cell lines concurrently grown in
the lab. An M13 tail (AGGGTTTTCCCAGTCACGAC) was added to each F primer for a universal labeling. Motif is obtained from the
RepeatMasker track of the UCSC Genome Browser. Reference size (reference genome size) is based upon an in silico PCR from
CanFam3 and includes the length of the M13 tail. Location is for the tetraSTR as identified by RepeatMasker in CanFam3.

55

4.2.2

Analysis

Whole genome sequencing was performed on the three samples using Illumina’s HiSeq 2000 sequencing system (San Diego, CA) with paired-end reads of 100 base pairs in length to give an
average coverage of 30X. Quality trimming of the reads and removal of the adaptor sequence was
performed by Trimmomatic v.0.30 [Bolger et al., 2014]. The reads were aligned against the canine
boxer reference genome (CanFam 3.1.69) using the Burrows-Wheeler Aligner (bwa v.0.7.5a-r405)
[Li and Durbin, 2009]. The resulting variant call format (vcf) file was generated using SAMTools
v.0.1.19 [Li et al., 2009]. The predicted effect of variants on protein function was annotated using
the SIFT 4G v.1.0 software [Ng and Henikoff, 2001; 2002; 2003; Kumar et al., 2009]. The vcf files
were imported into a MySQL database for filtering and prioritizing. The list of variants for each
sample type were filtered to those that are novel in this Bernese mountain dog by removing those
common with the Single Nucleotide Polymorphism Database (dbSNP) and with other unaffected
breeds of dogs that were sequenced as part of other studies in the lab. Then the filtered list for
each sample were compared to each other to find variants that meet criteria for spontaneous mutations in the tumor tissue, loss of heterozygosity mutations, and germ line mutations. Spontaneous
mutations in the tumor tissue were identified as variants that were present in the cell line and may
be present in the tumor tissue but were not found in the blood sample. The loss of heterozygosity
mutations were variants that were heterozygous in the blood sample and homozygous in the cell
line and were present in the tumor tissue. Germ line mutations are variants that were shared among
all three samples. Perturbations in common pathways were determined by inputing the gene lists
into DAVID (http://david.abcc.ncifcrf.gov/).

4.3

Results

The pedigree of the selected Bernese mountain dog was investigated using the Berner-Garde Foundation Database (http://bernergarde.org/db/). No other dogs affected with a histiocytic disease
were found in this dog’s pedigree as far back as three generations (great-grandparents). Prior to

56

sequencing, the microsatellite markers of the cell line were analyzed to differentiate this cell line
from others concurrently grown in the lab. Gel electrophoresis confirmed that the microsatellite
motifs of the cell line were distinct from other canine cell lines and matched the motif sizes of the
blood and tumor tissue from the same dog as expected.
Prior to filtering, the blood, tumor tissue, and cell line samples had a total of 3.80 million, 3.84
million, and 3.86 million variants, respectively, resulting in 1 change every 632 bases, 625 bases,
and 622 bases, respectively (Table 4.2). Chromosomes X and 20 had the highest number of changes
for all three sample types. CD molecules composed the gene family with the highest number
of variants. Other gene families with a high number of variants included the immunoglobulin
superfamily and cadherins.

57

Table 4.2: Summary of variant analysis from whole genome sequencing
Number of variants
Change rate
X Chromosome with highest change rate
Chromosome 20 with next highest change rate
Single-nucleotide polymorphism (SNP)
Multiple-nucleotide polymorphism (MNP)
Insertion
Deletion
Missense
Nonsense
Silent

Blood
3,803,849
1 change every 632 bases
1,142
728
3,142,837
11
342,641
318,360
14,276
269
18,112

Tumor Tissue
3,841,485
1 change every 625 bases
1,105
732
3,191,720
12
338,221
311,532
12,674
130
17,696

Cell Line
3,862,932
1 change every 622 bases
1,158
793
3,228,985
18
320,146
313,783
12,726
150
16,284

Summary of variant analysis for blood, tumor tissue, and cell line samples. The reads were aligned against the canine boxer reference
genome. Across the three samples from this dog, an average of 3.8 million variants were identified resulting in a change rate of 1
change every 626 bases. Chromosomes X and 20 had the highest change rate for all three sample types. The majority of variants were
single nucleotide polymorphisms with inserts and deletions responsible for about 300,000 variants each. The maximum deletion size
was 42 bases and the maximum insertion size was 29 bases.

58

For the purposes of this study, the focus was on high impact variants. These were variants in
protein coding regions of known genes with predicted deleterious effects on the protein. There
were 29 variants that met the criteria for spontaneous mutations in the tumor (Table 4.3). Two
of these variants are in genes associated with cancer: SSH3 and ITGB7. 1% of the variants were
identified to be within the ECM-receptor interaction pathway. Six mutations show loss of heterozygosity (Table 4.4), and among these, three of the variants are in genes associated with cancer:
FAT1, MTUS1, and LRP1B. Over one hundred variants met the criteria for germ line mutations
(Table 4.5). The majority were heterozygous in all the samples and fifty-one were homozygous in
all three samples.

59

Table 4.3: Spontaneous nonsynonymous mutations in tumor and cell line
Chr
1
2
2
3
3
4
4
7
9
9
9
9
10
11
12
12
14
17
18
18
20
20
27
27
28
33
33

Pos
99577380
21723758
71919034
40432632
61335881
19158759
72771154
64513295
4394424
25027641
42583941
55311831
19808586
17927599
1560882
19306115
26363540
8318138
50130506
55493519
4917889
44758332
1994136
2342172
40673991
24961845
24969068

Ref
C
A
C
T
A
C
G
C
G
CAAAA
G
G
G
G
G
C
T
T
C
A
G
C
C
C
G
A
G

Alt
T
C
T
C
C
T
T
G
A
CAAAAA
A
T
A
A
A
T
C
G
G
G
C
T
G
T
A
G
T

AA
R305C
H520P
D277N
F373S
K175Q
G166D
P946T
L1084F
G522S
A277S
P24S
D421Y
G1605R
D257N
D268N
V3873M
D1358G
I1675M
Q316H
I87T
E862D
R1301C
R759G
A315V
A172V
V2604A
S548Y

Gene
ZNF837
DCLRE1C
TRNAU1AP
LINS
NOP14
HNRNPA1L2
SPEF2
LAMA3
QRICH2
TTLL6
IFT20
CIZ1
CELSR1
SLC27A6
EGFL8
PKHD1
THSD7A
GREB1
SSH3
MS4A15
FGD5
KIAA1683
ITGB7
KRT76
TTC40
GOLGB1
GOLGB1

Qual
222
53
94
36
85
117
170
50
49
142
73
41
25
30
15.1
37
66
32
126
52.1
222
139
16.1
113
87
52
142

GT
2/2
1/2
1/2
1/2
1/2
1/2
2/2
1/2
1/2
2/2
1/2
1/2
1/2
1/2
1/2
1/2
1/2
1/2
1/2
1/2
2/2
2/2
1/2
1/2
1/2
1/2
1/2

Gene function
transcriptional regulation
V(D)J recombination
selenocysteine biosynthesis
development
nucleolar processing of pre-18S rRNA
packaging pre-mRNA into hnRNP particles
axoneme development
attachment, migration, and organization
glutamine-rich protein-coding
mediates tubulin polyglutamylation
intraflagellar transport particles
subcellular localization of CIP/WAF1
cell/cell signaling
translocation of long-chain fatty acids
calcium ion binding
regulation of centrosome duplication
endothelial cell migration
estrogen-stimulated cell proliferation
Protein tyrosine phosphatase
signal transduction
activates CDC42, of the Ras family
protein-coding gene
adhesion molecule - lymphocyte migration
terminal cornification
tetratricopeptide repeat protein-coding gene
intercisternal cross-bridges of the Golgi
intercisternal cross-bridges of the Golgi

Chr = chromosome, Pos = position based upon CanFam3.1.69, Ref = reference allele, Alt = alternate allele, AA = amino acid change,
Qual = quality score of cell line, GT = genotype of cell line where 1 = reference allele and 2 = alternate allele thus 1/2 = heterozygous
variant and 2/2 = homozygous variant.
60

Table 4.4: Loss of heterozygosity variants resulting in nonsynonymous mutations
Chr

Pos

Ref

Alt

AA

Gene

16
16
19
21
36
X

41067620
44131177
42832330
42006008
10135919
55633452

G
C
G
C
G
C

T
G
A
T
A
T

P401T
P925R
S2899F
T585M
A208V
A36V

MTUS1
FAT1
LRP1B
NAV2
SLC38A11
ITGB1BP2

Qual GT
blood blood
55
14.2
42
133
176
65

1/2
1/2
1/2
1/2
1/2
1/2

Qual
cell
line
95
82.1
13.2
222
96.3
65.5

GT
cell
line
2/2
2/2
2/2
2/2
2/2
2/2

Gene function
inhibit ERK2 and cell proliferation
cellular polarization
bind and internalize ligands
helicase and exonuclease activity
amino acid/proton antiporter
maturation, organization of muscle cells

There are six variants that show loss of heterozygosity, i.e. heterozygous in the blood and homozygous in the cell line and tumor tissue.
Among these, three of the variants, MTUS1, FAT1, and LRP1B, are nonsynonymous mutations in genes associated with cancer. Chr =
chromosome, Pos = position based upon CanFam3.1.69, Ref = reference allele, Alt = alternate allele, AA = amino acid change, Qual =
quality score, GT = genotype where 1 = reference allele and 2 = alternate allele thus 1/2 = heterozygous variant and 2/2 = homozygous
variant.

61

Table 4.5: Germ line variants resulting in nonsynonymous mutations in cancer associated genes
Chr
1
1
1
1
1
3
4
6
6
9
9
10
14
16
16
16
18
19
22
24
38
38

Pos
108127452
111816850
111816940
111907507
111917369
54114371
57765782
5751604
55023326
25310309
52519537
43245707
40338217
1139218
41067620
44131177
52324714
42832330
58549714
20493619
1679038
23319736

Ref
C
G
T
A
C
G
G
G
G
T
C
C
G
C
G
C
A
G
G
C
T
G

Alt
T
A
G
G
G
A
A
A
A
G
T
T
A
T
T
G
G
A
A
T
A
A

AA
A393V
V125M
S155A
N155I
Q496E
R1813Q
E763K
T556M
S1099N
V88G
R257C
R1063C
L201F
R305Q
P401T
P925R
M360V
S3036F
P395L
R115H
S597C
T11M

Gene
LIG1
CEACAM28
CEACAM28
CEACAM1
CEACAM1
ALPK3
FAT2
GTF2I
ARHGAP29
PHB
RAPGEF1
REV1
HOXA9
ADCY1
MTUS1
FAT1
CDC42BPG
LRP1B
COL4A1
NRSN2
DSTYK
CD1B

Qual
140
57
11.3
114
140
74
140
36.1
117
95
37
106
217
24
95
82.1
222
13.2
109
55
61
185

GT
2/2
1/2
1/2
2/2
2/2
1/2
2/2
1/2
2/2
1/2
1/2
1/2
2/2
1/2
2/2
2/2
2/2
2/2
2/2
1/2
2/2
1/2

Gene function
DNA ligase
carcinoembryonic antigen-related cell adhesion molecule
carcinoembryonic antigen-related cell adhesion molecule
cell-cell adhesion molecule
cell-cell adhesion molecule
kinase, cardiomyocyte differentiation
migration of epidermal cells
formation of complex at C-FOS promoter
Rho GTPase activating protein
Prohibitin inhibits DNA synthesis
transduces signals to activate RAS
Deoxycytidyl transferase involved in DNA repair
sequence-specific transcription factor
calmodulin-sensitive adenylyl cyclase
inhibit ERK2 and cell proliferation
cellular polarization
effector of CDC42 in cytoskeletal reorganization
bind and internalize ligands
Type IV collagen, inhibits angiogenesis and tumors
maintenance and/or transport of vesicles
caspase-dependent apoptosis and cell death
antigen-presenting protein

List of variants in genes associated with cancer that met the criteria for potential germ line nonsynonymous mutations. Chr =
chromosome, Pos = position based upon CanFam3.1.69, Ref = reference allele, Alt = alternate allele, AA = amino acid change, Qual =
quality score, GT = genotype where 1 = reference allele and 2 = alternate allele thus 1/2 = heterozygous variant and 2/2 = homozygous
variant.

62

4.4

Discussion

This Bernese mountain dog was selected for sequencing because the lab had obtained multiple
high quality samples of sufficient quantity from her and, most importantly, a cell line had been
successfully derived from her tumor. Unfortunately, not enough information was known about the
health of this dog’s pedigree to determine if there were other relatives affected with histiocytic
sarcoma. However, interestingly, an unrelated BMD that shared the household was affected with
HS at approximately the same time and also did not have a known history of affected relatives
in the pedigree. The possibility of an environmental trigger of histiocytic sarcoma in these dogs
cannot be ruled out.
Filtering of the variants was based upon several assumptions. One was that the blood sample
was considered a normal sample for this Bernese mountain dog and variants identified in this
sample were germ line mutations in this dog. The second assumption was that the tumor tissue
sample was actually a mix of neoplastic cells and normal stromal cells and thus the allele status of
the variants may reflect that mixed population. The cell line was considered to be purely neoplastic
cells due to the clonal derivation utilized in the production of the cell line. While cell lines can
accrue spontaneous mutations during repeated subculturing, this cell line was less than a year old
and unlikely to have a substantial number of these mutations.
Chromosomes X and 20 had the highest base change rate among the chromosomes. Interestingly, the gene families with the highest number of variants included CD molecules, immunoglobulins, and cadherins - all gene families that are associated with the major facets of immune function
through signaling, function, or mobility. Nearly 10% of the genes associated with cancer did not
have variants identified and presumably are normal in the Bernese mountain dog. These genes
belong to over one hundred different gene families.
A 2011 study of BMDs and flat-coated retrievers suggested that deletions of tumor suppressors
on certain chromosomes including 2 and 16 may play a role in histiocytic sarcoma [Hedan et al.,
2011]. Our study revealed frameshift mutations in UBIAD1, a tumor suppressor on chromosome 2,
and CSMD1, a tumor suppressor on chromosome 16. Recent work using genome wide association
63

studies has found histiocytic sarcoma susceptibility associated with the MTAP-CDKN2A locus
[Shearin et al., 2012]. In our study we found fifteen variants shared between all three samples that
were within the same region of interest and two additional variants that were present only in the
tumor and cell line 4.6. These variants ranged in quality scores from 21 to 222 (corresponding to
a phred quality score of 13 to 23) with a depth coverage ranging from 5 to 23. However, all of
these variants were intergenic and the effect on the protein function of CDKN2A or MTAP was not
determined.
Table 4.6: Variants found in the MTAP-CDKN2A locus
Chr
11
11
11
11
11
11
11
11
11
11
11
11
11
11
11
11
11

Pos
44179537
44180236
44187367
44193532
44200738
44200767
44207948
44208197
44214225
44222716
44223736
44223794
44223907
44223919
44229492
44231570
44236979

Reference
CA
T
CTTTTTTT
G
TGGG
G
TG(AG)12
G
GAAAA
C
T
C
G
G
A
C
A

Alternate
CAA
G
CTTTTTT
A
TGG
A
TG(AG)11
A
GAAA
T
C
T
T
T
G
G
G

Qual
70.7
119
18.8
120
114
146
44
95.1
18.5
94
98
26
23
39
45
24
152

Depth
7
11
5
12
8
13
15
6
9
12
14
6
11
9
14
11
8

GT blood
2/2
0/0
2/2
2/2
2/2
2/2
2/2
2/2
1/2
1/2
1/2
0/0
1/2
1/2
1/2
1/2
2/2

GT tissue
2/2
2/2
2/2
1/2
2/2
2/2
2/2
2/2
1/2
1/2
1/2
1/2
1/2
1/2
1/2
1/2
2/2

GT cell line
2/2
2/2
2/2
2/2
2/2
2/2
2/2
2/2
1/2
1/2
1/2
1/2
1/2
1/2
1/2
1/2
2/2

List of variants in the region of interest from other studies. All variants were intergenic. Chr =
chromosome, Pos = position based upon CanFam3.1.69, Ref = reference allele, Alt = alternate
allele, AA = amino acid change, Qual = quality score, GT = genotype where 1 = reference allele
and 2 = alternate allele thus 1/2 = heterozygous variant and 2/2 = homozygous variant.

The causative genetic mutation for histiocytic sarcoma in Bernese mountain dogs has yet to be
identified. Future work will require experimental validation of the variants identified in this study.
Validated variants will be investigated to determine their association with histiocytic sarcoma and
their functional role in the tumorigenesis. The six variants that show loss of heterozygosity by

64

the whole genome sequencing can be validated quickly. Additional variants of interest include
the frameshift mutations in UBIAD1 and CSMD1. The whole genome sequencing has provided a
tremendous amount of data and is an important first step that will ultimately lead to the identification of the mutation that predisposes Bernese mountain dogs to histiocytic sarcoma.

65

CHAPTER 5
CONCLUSIONS

This work demonstrates the utility of dog breeds as unique genetic pools for studying rare disorders
that affect both dogs and humans. The focus here is on histiocytic diseases, a rare disease in humans
for which the dog may facilitate research on the genetic basis of these disorders. Historically, the
BMD is highly overrepresented among dog breeds affected by histiocytic diseases. By capitalizing
on a community sourced database of BMD health information, a contemporary assessment of the
health status of the BMD breed was conducted. The average life expectancy of a large breed
dog is around ten years of age, however, this analysis revealed that an overwhelming majority
of BMDs have a shortened lifespan of seven to eight years. Cancer is the major factor in the
shortened lifespan of BMDs, and among the cancers, histiocytic sarcoma was the most frequent
diagnosis. This investigation confirmed that histiocytic diseases remain an important disease for
the BMD, thus warranting further studies to improve the life expectancy of BMDs with this cancer,
and inform human studies where the rarity of the disease hinders the development of treatments.
Furthermore, histiocytic sarcoma can provide insight into the biology of dendritic cells, from which
these cancers arise.
To unravel the mechanisms of tumorigenesis in histiocytic diseases, a novel approach involving
miRNAs was utilized. The differential expression of miRNAs along the spectrum of canine histiocytic diseases was evaluated. MicroRNA expression was clearly different between the normal
histiocytes and the histiocytic diseases. Among the histiocytic disease phenotypes, the miRNA expression was more influenced by breed than specific disease diagnosis. Continued analysis of the
pathways (e.g., see Figure 5.1) regulated by these microRNAs may pinpoint critical targets in the
tumorigenesis of histiocytic sarcoma and allow us to design novel cancer therapies that are more
effective in humans and dogs.
The advent of relatively inexpensive next-generation sequencing has extensive applications,

66

Figure 5.1: MiRNA regulated pathways. Predicted tumor suppressor mRNA targets of
cfa-miR-125b and miR-152.
not the least of which has been in investigating the genetic basis of rare diseases. These studies
can now be more effectively conducted using whole genome sequencing. Three samples from one
BMD diagnosed with histiocytic sarcoma were submitted for whole genome sequencing. Several
terabytes of data were generated yielding an abundance of information that will continue to be
mined for answers to new questions in canine genetics. Results from this approach were compared to the pathways regulated by the miRNAs identified by the miRNA array analysis. Several
interesting gene targets were identified, including several cancer associated genes such as DAPK1,
CD44, SOX5, and ITGA5.
The importance of continued research into rare orphan diseases, such as histiocytic diseases,
cannot be overstated. These cancers may hold the key for discovering the mechanism of tumorigenesis in these cancers and potentially others as well.

67

CHAPTER 6
FUTURE WORK

In the course of this work, several miRNAs were identified that may have a critical role in the
pathology of histiocytes. These miRNAs will need to be evaluated as potential targets for downregulation, for example in histiocytic sarcoma, to determine if knockdown of these critical microRNAs to normal levels will reverse the oncogenic phenotype in affected cells. Work along these lines
was attempted and merits further attention. Although the miRNAs studied here did not display a
disease specific signature, the widespread miRNA upregulation in histiocytic diseases suggests
there are miRNAs playing a critical role in the pathology. Continued investigation of the pathways
regulated by these miRNAs may trace back to the core genetic perturbations responsible for the
development of disease. A few interesting cancer-associated gene targets were identified by comparing the miRNA results with gene expression data obtained through another study in the lab and
the gene variants identified in the whole genome sequencing data. These include DAPK1, a positive
mediator of programmed cell death, CD44, which is associated with miRNA induced migration,
SOX5, a transcriptional regulator correlated with reduced apoptosis, and ITGA5, an integrin that if
lost can result in immune dysfunction. Work is ongoing to verify the aberrant expression of these
genes, among others, in histiocytic disease samples. Histiocytic diseases have been studied in humans and dogs for decades with no specific genetic mutation found despite the clear inheritance in
BMDs. By undertaking a multi pronged investigation such as this, the confluence of data are more
likely to identify the critical players in these complicated diseases. Bioinformatic approaches in
combination with experimental analysis will be necessary to progress our understanding of these
diseases.

68

APPENDIX

69

APPENDIX A
CANINE AND HUMAN GASTROINTESTINAL STROMAL TUMORS DISPLAY
SIMILAR MUTATIONS IN C-KIT EXON 11

Published in BMC Cancer, 2010 [Gregory-Bryson et al., 2010].

A.1

Abstract

Background: Gastrointestinal stromal tumors (GISTs) are common mesenchymal neoplasms in
the gastrointestinal tract of humans and dogs. Little is known about the pathogenesis of these
tumors. This study evaluated the role of c-KIT in canine GISTs; specifically, we investigated
activating mutations in exons 8, 9, 11, 13, and 17 of c-KIT and exons 12, 14, and 18 of plateletderived growth factor receptor, alpha polypeptide (PDGFRA), all of which have been implicated
in human GISTs.
Methods: Seventeen canine GISTs all confirmed to be positive for KIT immunostaining were
studied. Exons 8, 9, 11, 13 and 17 of c-KIT and exons 12, 14, and 18 of PDGFRA, were amplified
from DNA isolated from formalin-fixed paraffin-embedded samples.
Results: Of these seventeen cases, six amplicons of exon 11 of c-KIT showed aberrant bands
on gel electrophoresis. Sequencing of these amplicons revealed heterozygous in-frame deletions
in six cases. The mutations include two different but overlapping six base pair deletions. Exons
8, 9, 13, and 17 of c-KIT and exons 12, 14, and 18 of PDGFRA had no abnormalities detected
by electrophoresis and sequencing did not reveal any mutations, other than synonymous single
nucleotide polymorphisms (SNPs) found in exon 11 of c-KIT and exons 12 and 14 of PDGFRA.
Conclusions: The deletion mutations detected in canine GISTs are similar to those previously found in the juxtamembrane domain of c-KIT in canine cutaneous mast cell tumors in our
laboratory as well as to those reported in human GISTs. Interestingly, none of the other c-KIT
or PDGFRA exons showed any abnormalities in our cases. This finding underlines the critical

70

importance of c-KIT in the pathophysiology of canine GISTs. The expression of KIT and the identification of these activating mutations in c-KIT implicate KIT in the pathogenesis of these tumors.
Our results indicate that mutations in c-KIT may be of prognostic significance and that targeting
KIT may be a rational approach to treatment of these malignant tumors. This study further demonstrates that spontaneously occurring canine GISTs share molecular features with human GISTs and
are an appropriate model for human GISTs.

A.2

Background

Gastrointestinal stromal tumors (GISTs) are one of the most common mesenchymal tumors that
arise from the wall of the gastrointestinal tract. Gastrointestinal stromal tumors occur in many
species including humans, dogs [LaRock and Ginn, 1997; Frost et al., 2003; Kumagai et al., 2003],
and horses [Del Piero et al., 2001; Hafner et al., 2001]. GISTs can metastasize to the liver and
peritoneal cavity, warranting a very poor prognosis. In humans, approximately 70% of GISTs
occur in the stomach and 20% occur in the small intestine [Miettinen et al., 2002; Agaimy et al.,
2008b], whereas in dogs the reverse is true with 76% of GISTs occurring in the small intestine and
colon, while 19% occur in the stomach [Frost et al., 2003].
The majority of GISTs are diagnosed by the demonstration of the expression of KIT (CD117),
a type III tyrosine kinase receptor encoded by the proto-oncogene c-KIT [Taniguchi et al., 1999;
Rubin et al., 2001; Croom and Perry, 2003; Frost et al., 2003; Wong et al., 2003]; although a small
proportion of GISTs do not exhibit CD117 immunoreactivity [Debiec-Rychter et al., 2004]. KIT
has critical roles in cell differentiation, proliferation and migration, especially in hematopoietic,
neural crest, and germ cell lineages [Ma et al., 1999b]. In addition, KIT along with its ligand, stem
cell factor (SCF), also known as steel factor, is necessary for the development of melanocytes, mast
cells, and interstitial cells of Cajal [Hulzinga et al., 1995]. It has been suggested that GISTs may
originate from the interstitial cells of Cajal, which are pacemaker cells responsible for regulating
peristalsis in the gastrointestinal tract [Kindblom et al., 1998; Steigen and Eide, 2009].
The KIT receptor is a cell surface receptor consisting of an extracellular domain, a transmem71

brane domain, and a cytoplasmic domain, which includes the juxtamembrane and kinase domains
[Yarden et al., 1987; Qiu et al., 1988]. The juxtamembrane domain (amino acid residues 543-580)
[Ma et al., 1999b] is a highly conserved region of KIT located between the transmembrane domain
(amino acid residues 521-543) [Yarden et al., 1987] and kinase domain (amino acid residues 581936) [Ma et al., 1999b]. The KIT juxtamembrane domain is primarily coded for by exon 11 of
c-KIT, while the split kinase domain are coded for by exons 12-18 of c-KIT [London et al., 1999].
The juxtamembrane domain regulates the enzymatic activity of KIT by preventing relative movement of the protein and thus inhibiting receptor dimerization [Roskoski, 2005]. In normal cells,
binding of the SCF ligand to the KIT receptor results in receptor homodimerization and subsequent
activation of the KIT receptor via cross phosphorylation of tyrosine residues on the opposite KIT
homodimer partner [Roskoski, 2005]. The phosphotyrosines become binding sites and activators
of several cell-signaling proteins including JAK2 and PI3K [Heinrich et al., 2002]. These particular proteins are the start of the JAK-STAT and JNK pathways, leading to a potent intracellular
signal for the cell to proliferate[Blechman et al., 1993; Heinrich et al., 2002]. Gain-of-function
mutations in KIT among human GISTs have demonstrated that the constitutive activation of KIT
in the absence of its ligand and without dimerization may play a critical role in GIST tumorigenesis
[Hirota et al., 1998].
In humans, mutations in c-KIT have been reported in more than 65% of GIST cases [Hirota
et al., 1998; Taniguchi et al., 1999; Roskoski, 2005; Agaimy et al., 2008a], and in GISTs with wildtype c-KIT, mutations of platelet-derived growth factor receptor, alpha polypeptide (PDGFRA)
were found in 35% of those cases [Heinrich et al., 2003]. PDGFRA codes for a transmembrane
type III tyrosine kinase receptor for members of the platelet-derived growth factor family, which
are mitogens for cells of mesenchymal origin. GISTs with c-KIT or PDGFRA mutations have
similar downstream signaling pathways, suggesting that PDGFRA mutations serve as an alternative
tumorigenic mechanism to c-KIT in GISTs [Heinrich et al., 2003]. Mutations have been found in
exons 11, 9, 13, and 17 of c-KIT in sporadic GISTs, with exon 11, the juxtamembrane domain,
being the most frequent site of mutations [Miettinen and Lasota, 2006; Ostrowski et al., 2009;

72

Steigen and Eide, 2009], comprising up to 90% of all c-KIT mutations [Longley et al., 2001].
Exon 8 of c-KIT has been reported to have mutations in other types of neoplasias. Most GISTs
are sporadic, but familial GIST syndromes presenting with multiple GISTs have been reported in
humans [Nishida et al., 1998]. Affected family members often harbor germline mutations of the
c-KIT gene in their tumors and leukocytes [Nishida et al., 1998; Hirota and Isozaki, 2006] and
there is a report of one family with a germline mutation in PDGFRA [Chompret et al., 2004]. The
familial GIST syndrome has been recapitulated in two knock-in mouse models, one designed with
a V558 deletion mutation in exon 11 of c-KIT, and the other carrying a K-to-E amino acid mutation
at position 641 in exon 13 of c-KIT [Sommer et al., 2003; Rubin et al., 2005].
All of the reported mutations in c-KIT could potentially lead to the activation of KIT in the
absence of its ligand. Constitutively activated KIT would then give rise to the development and/or
progression of gastrointestinal stromal tumors in dogs. A comparison of the currently documented
mutations found in c-KIT in humans and canines is presented in Figure A.1. The reported canine
c-KIT mutations have been associated with mast cell tumors [Zemke et al., 2002; Webster et al.,
2006a] as well as GISTs [Frost et al., 2003]. The purpose of this study was to evaluate the role of cKIT and PDGFRA in canine GISTs. While KIT immunopositivity has been demonstrated in canine
GISTs in two previous studies [Frost et al., 2003; Kumagai et al., 2003], only the study by Frost
et al. has explored mutations in c-KIT to date, where four archived canine GISTs were examined,
revealing mutations in exon 11 of c-KIT in two of the cases. Our present study investigates exons
8, 9, 11, 13, and 17 of c-KIT and exons 12, 14, and 18 of PDGFRA for mutations in a larger sample
set of canine GISTs, providing information on c-KIT mutational status in seventeen cases.

A.3

Methods

A.3.1

Cases

Forty-six cases of canine gastrointestinal smooth muscle tumors were submitted to the Michigan
State University Diagnostic Center for Population and Animal Health from 1991 to 2006. All

73

558

552

553

558

Rubin et al.,
2001

559

557

558 559 560

557

555

560

554

551

555

552

551

569

562

558

564

570

556

560

558

550

576
573

557

576

558
554

558
558

556

557 558

550

559
555

551

554

550

Taniguchi
et al.,
1999

551

559

556

559

555

557 559

555

557

550

556
553

550 551

560

558 560

556

558

560

556

561

556

561

557

561
559 561

557 558

557

551
551

552

Frost et al., 2003
*Found in this study

554

557

556

561

559
558

559

560

579

561

568

572

558

576

557 558
557

559

Juxtamembrane domain
544

585

586

581

589

642

820

Rubin et al., 2001

Kinase domain
582
Substitution

937
Nonsense

Deletion
Human

Insertion

Duplication

Dog

Figure A.1: Comparison of c-KIT mutations in human and canine GISTs. Human mutations
are in white; dog mutations are indicated in gray [Frost et al., 2003; Kumagai et al., 2003;
Taniguchi et al., 1999; Rubin et al., 2001; Heinrich et al., 2002]. Codon numbering is based upon
human amino acid sequence [GenBank:NP_000213]. *These are the two deletion mutations
found in this study.
74

tumors originated from surgical biopsies that were immediately fixed in 10% neutral buffered
formalin and embedded in paraffin within 24-48 hours, following routine protocols. From this pool
of cases eighteen tumors were diagnosed as GISTs confirmed by characteristic histomorphology
and positive KIT staining by immunohistochemistry and were included in this study (Figure A.2).
The age of the dogs in this study ranged from 4 to 15 years with a mean age of 10.9 years. Various
purebred and mixed bred dogs were included with a gender ratio of 72% female to 28% male dogs.
Tumor sites were distributed throughout the gastrointestinal tract from the stomach to the cecum
as indicated in Table A.1. A histologically normal, non-neoplastic tissue sample from each dog
was also analyzed to determine the c-KIT mutation status in constitutive DNA.

75

Figure A.2: Histopathology of GIST. A Hemotoxylin and eosin stain of a GIST. Section from a
gastrointestinal stromal tumor analyzed in this study. B Immunohistochemical staining for KIT.
KIT (brown deposits) in a section of a gastrointestinal stromal tumor.

76

Table A.1: Samples
Case
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17

Breed
Cockapoo
German Wirehaired Pointer
Cocker Spaniel
Labrador Retriever
American Pit Bull Terrier
Golden Retriever
Collie
Boxer
Cocker Spaniel
Irish Setter
Golden Retriever
Springer Spaniel
Labrador Retriever
German Shorthaired Pointer
Labrador Retriever cross
Mixed
Mixed

Age
13
10.5
12
4
15
13
12
10
US
11
10.5
10.6
8.5
9
10
9
15

Sex
FS
FS
FS
FS
FS
F
M
M
FS
FS
MN
FS
FS
FS
MN
FS
MN

Tumor Site
stomach
cecum
small intestine
duodenum
cecum
small intestine
jejunum
jejunum
small intestine
small intestine
cecum
cecum
jejunum
small intestine
ileocecocolic junction
cecum
jejunum

c-KIT Mutations
exon 11 deletion and SNP
exon 11 deletion
normal
normal
exon 11 deletion
normal
exon 11 SNP
normal
exon 11 SNP
normal
exon 11 SNP
exon 11 deletion
normal
normal
exon 11 deletion
exon 11 deletion
normal

PDGFRA Mutations
normal
exon 12 and 14 SNPs
normal
normal
normal
normal
normal
normal
normal
normal
normal
normal
normal
exon 12 SNP
normal
exon 12 and 14 SNPs
normal

Signalment and site of canine gastrointestinal stromal tumor cases analyzed in this study. FS = female spayed, MN = male neutered,
US = unspecified.

77

A.3.2

DNA isolation from formalin-fixed paraffin-embedded (FFPE) sections

Neoplastic tissue, less than 1 mm3 , was excised from each FFPE block to retrieve sections corresponding to KIT positive immunostaining areas. Similarly, sections of histologically normal tissue,
negative for KIT immunostaining were also collected from each case. From these tissue sections
DNA was isolated as described previously [Banerjee et al., 1995; Zemke et al., 2002]. The tissue
section was placed in 400 µl of digestion buffer (50 mM Tris, pH 8.5, 1 mM ethylenediaminetetraacetic acid [EDTA], 0.5% Tween) and heated to 95 ◦ C for 10 minutes to melt the paraffin. The
tissue solutions were then subjected to high power microwave irradiation twice for 30 seconds each
with vortexing after each heating step. After cooling, 5 µl of 15 mg/ml proteinase K was added
to each solution and incubated overnight at 42 ◦ C. Following protein digestion, proteinase K was
inactivated at 95 ◦ C for 10 minutes. The solutions were then centrifuged and 150 µl was aliquoted
to be used as DNA template in subsequent polymerase chain reaction (PCR).

A.3.3

Amplification of c-KIT juxtamembrane and kinase domains

Exon 11, coding for the juxtamembrane domain of KIT, and exon 17, coding for the kinase domain
of KIT, were amplified from these tissue sections via PCR using primers and conditions optimized
in earlier studies [Jones et al., 2004; Webster et al., 2006b]. Exon 8, 9, and 13 of c-KIT and exons
12, 14, and 18 of PDGFRA were also amplified (Table A.2). The PCRs were set-up in 25 µl total
reaction volume consisting of 50 ng of DNA template prepared as described above: 5 pmol of each
primer, 0.5 U of Taq polymerase (Invitrogen, Carlsbad, CA) and final concentrations of 80 µM
deoxynucleoside triphosphate, and 2 mM MgCl2 . Cycling conditions for the PCR were 94 ◦ C for
4 minutes; 40 cycles of 94 ◦ C for 1 minute, annealing temperatures averaging 58 ◦ C for 1 minute,
and 72 ◦ C for 1 minute; followed by a final elongation step at 72 ◦ C for 5 minutes. PCR products
were then subjected to electrophoresis on 2% agarose gels and visualized under ultraviolet light
after ethidium bromide staining.

78

Table A.2: Primer sets
Amplified Region
c-KIT exon 8
c-KIT exon 9
c-KIT exon 11
c-KIT exon 13
c-KIT exon 17
PDGFRA exon 12
PDGFRA exon 14
PDGFRA exon 18

Forward
5’-CAGCAGTCTGACCTATGGC -3’
5’-GATTGATTGATTGATTTTCCTAG-3’
5’-CATTTGTTCTCTACCCTAAGTGCT-3’
5’-CTGATTAAGTCGGATGCGGC-3’
5’-ATAGCAGCATTCTCGTGTTG-3’
5’-TTAATGGCTCTGATTGCTCAC-3’
5’-ACTGGTTTTGGTTCCCACAG -3’
5’-TAGCTCAGCCGTGGGTATG-3’

Reverse
5’-GCTCAGCTCCTGGACAGAAA-3’
5’-GCAGGCAGAGCCTAAACATC-3’
5’-GTTCCCTAAAGTCATTGTTACACG-3’
5’-CAAGCACTGTCGCAATGG-3’
5’-AACTAAAATCCTTCACTGGACTG-3’
5’-CACCCAGTGCTCATAACCTC-3’
5’-CAATGATTCGCAGCAACG-3’
5’-CACATGAGCAGAGATGTCAGG-3’

Primer sets used for the amplification of the indicated exons of c-KIT and PDGFRA during PCR.

79

A.3.4

Sequencing

Amplified fragments from all tissue sections were characterized by automated sequencing. The
PCR product for each section was submitted in 5 µl quantities with 30 pmol of the appropriate
primer to Michigan State University’s Genomics Technology Support Facility. This facility utilizes
the automated direct sequencing technique, which incorporates fluorescently labeled dideoxynucleotides during cycle sequencing and separates the resulting products by capillary electrophoresis
for detection on an ABI 3700 sequence analyzer (Foster City, CA).

A.4

Results

Of the eighteen KIT immunopositive cases, seventeen cases yielded amplification products. The
remaining case did not yield amplification products with any of the c-KIT or PDGFRA primer sets
or with primers for unrelated canine genes.
For exon 11 of c-KIT, six of these seventeen cases of canine GISTs displayed an aberrant
banding pattern upon gel electrophoresis of the PCR product (Figure A.3). The remaining eleven
cases displayed a band similar to the positive control on electrophoresis, and analysis confirmed
the sequence was identical to the wild-type exon 11 of c-KIT except for a single nucleotide polymorphism (SNP) located at base pair 50110905 C>T [GenBank:NC_006595.2] detected in four
cases. However, sequence analysis of the aberrant six cases uncovered a mixture of normal and
mutant alleles. Further examination identified short in-frame deletions (Figure A.4). The mutations included two different, but overlapping 6 base pair deletions, which translated to a deletion of
two amino acids in two of the cases and an amino acid change and a deletion of two amino acids in
the other four cases. The first mutation (canine codons 556-557) occurred in two of the cases and
consisted of the deletion of the sequence AGTGGA located at base pairs 50110838 to 50110843
of the canine genomic DNA [GenBank:NC_006595.2]. This mutation translated to a deletion of
two amino acids, tryptophan and lysine, at codons 556 and 557 of canine KIT, respectively [GenBank:NP_001003181]. The second mutation (canine codons 556-558) was discovered in four of

80

the cases and results in the deletion of the sequence GGAAGG located at base pairs 50110841
to 50110846 of the canine genomic DNA [GenBank:NC_006595.2]. This second mutation translated to a deletion of two amino acids, lysine and valine, at codons 557 and 558 of canine KIT,
respectively [GenBank:NP_001003181]. The deletion of the last two guanines of the codon 556
in this mutation combined with deletion of the next 4 nucleotides resulted in an amino acid change
from the tryptophan at codon 556 to a phenylalanine (Table A.3). In these six cases, analysis of
the normal tissue obtained from these dogs revealed sequences that were identical to the wild-type
exon 11 of c-KIT.

81

L 1

2 +

- L

A.3: Electrophoresis of PCR product of exon 11 of c-KIT, juxtamembrane domain. L

Figure
= 100 bp ladder, 1 = case with normal exon 11 of c-KIT, confirmed by sequencing, 2 = case with
aberrant banding pattern, + = positive control (normal dog spleen), - = negative control (water)

82

dog
human
wild-type KIT
deletion of canine
codons 556-557
deletion of canine
codons 556-558

552
553
Y

554
555
E

V

558
559

556
557
Q

W

K

V

560
561
V

E

562
563
E

I

564
565
N

G

566
567
N

N

568
569
Y

V

TATGAAGTACAGTGGAAGGTTGTTGAGGAGATCAATGGAAACAATTATGTT
Y E V Q
TATGAAGTAC
Y

E

V

Q

V V E E I N G N N Y V
AGGTTGTTGAGGAGATCAATGGAAACAATTATGTT
V

F

TATGAAGTACAGT

E

E

I

N

G

N

N

Y

V

TTGTTGAGGAGATCAATGGAAACAATTATGTT

Figure A.4: Deletion mutations found in exon 11 of c-KIT. Sequence analysis results for the
six cases with the aberrant banding pattern illustrating the two deletion mutations present in these
cases. This region of exon 11 of canine c-KIT [GenBank:NP_001003181] differs by only one
codon from humans [GenBank: NP_000213].

83

Table A.3: Deletion mutations found in canine GISTs in this study
Case
2
16
1
5
12
15

Codon
556-557
556-557
556-558
556-558
556-558
556-558

DNA Mutation
GTA CAG TGG AAG GTT GTT > GTA C AG GTT GTT
GTA CAG TGG AAG GTT GTT > GTA C AG GTT GTT
GTA CAG TGG AAG GTT GTT > GTA CAG T TT GTT
GTA CAG TGG AAG GTT GTT > GTA CAG T TT GTT
GTA CAG TGG AAG GTT GTT > GTA CAG T TT GTT
GTA CAG TGG AAG GTT GTT > GTA CAG T TT GTT

Amino Acid
VQWKVV > VQVV
VQWKVV > VQVV
VQWKVV > VQFV
VQWKVV > VQFV
VQWKVV > VQFV
VQWKVV > VQFV

The codon numbering is based upon the canine KIT amino acid sequence. The genomic DNA mutations were experimentally
determined by direct sequencing and the altered sequence is underlined in column three. The expected amino acid changes in the
protein are provided in column four. [GenBank:NC_006595.2, GenBank:NP_001003181].

84

Exon 11 of c-KIT
50110819
AAACCCATGT ATGAAGTACA GTGGAAGGTT GTTGAGGAGA
50110859
TCAATGGAAA CAATTATGTT TACATAGACC CAACACAGCT
50110899
TCCTTAYGAT CACAAATGGG AGTTTCCCAG AAACAGGCTG
50110939
AGCTTTG

50110858
50110898
50110938
50110945

Exon 12 of PDGFRA
49690336
GTCGGATCCT GGGATCTGGT GCGTTTGGGA AAGTGGTTGA
49690376
AGGAACTGCC TATGGATTAA GCCGCTCCCA GCCGGTCATG
49690416
AAAGTCGCRG TGAAGATGCT GAAAC

49690375
49690415
49690440

Exon 14 of PDGFRA
GCCCCATTTA
49691320
TTTGGTCAAC
49691360
AGCCGCCACC
49691400
TTGGATTGAA
49691440

49691359
49691399
49691439
49691473

CATCATCACC
TATTTGCATA
CRGAGAAGCC
CCCTGCTGAT

GAGTACTGCT
AGAATAGRGA
AAAGAAAGAG
GACAGCACAC

TCTATGGGGA
TAGCTTCCTG
TTGGACATTT
GGAG

Figure A.5: Single nucleotide polymorphisms (SNPs). SNPs, shown in boxed letters, found in
exon 11 of c-KIT and exons 12 and 14 of PDGFRA [GenBank: NC 0006595.2] in the canine
GIST samples (R = A or G; Y = C or T).

All seventeen cases were also amplified for exons 8, 9, 13, and 17 of c-KIT. Only the expected
single band, similar to the positive control, was observed after gel electrophoresis. Sequencing
of all PCR products obtained revealed no mutations in these GIST samples for exons 8, 9, 13, or
17 of c-KIT. Similarly, amplification of exons 12, 14, and 18 of PDGFRA in these GIST samples
revealed clear, single bands on electrophoresis and the PCR products were directly sequenced.
Three of the cases had a SNP located at base pair 49690424 A>G [GenBank:NC_006595.2] in
exon 12 of PDGFRA. Two of the cases also had a SNPs located at base pair 49691387 A>G and
49691411 G>A [GenBank:NC_006595.2] of exon 14 of PDGFRA (Figure A.5).

A.5

Discussion

This study was able to ascertain c-KIT and PDGFRA mutational status of seventeen of eighteen
KIT positive canine gastrointestinal stromal tumors, representing a good amplification success rate
of 94% from FFPE tissues. Significantly, the study identified two distinct but overlapping mutations in exon 11 of c-KIT in the juxtamembrane domain. This region appears to be a mutational

85

hotspot with an overall incidence of 35.3% in our study population of canine GISTs. The only
other study of c-KIT mutations in canine GISTs reported mutations in two of four (50%) GISTs
[Kumagai et al., 2003]. Human GISTs have higher incidences of c-KIT mutations ranging from
65% to 92% across exons 8, 9, 11, 13, and 17, a majority of which occur in the juxtamembrane
domain [Hirota et al., 1998; Taniguchi et al., 1999; Rubin et al., 2001; Tabone et al., 2005]. In our
study, no mutations were identified in exons 8, 9, 13, and 17 of c-KIT. None of our cases showed
mutations in PDGFRA. Only a single amplification product was noted from the corresponding normal tissue of each GIST case, with sequencing verifying the presence of only the wild type allele
in the normal tissue. These results indicate that all mutations observed arose somatically in each
tumor.
Interestingly, these deletion mutations are similar to those previously found in the juxtamembrane domain of c-KIT in canine cutaneous mast cell tumors in our laboratory [Zemke et al., 2002]
and others [Ma et al., 1999b]. In a previous study of 21 canine GISTs, DNA suitable for amplification was recovered from only four cases and then amplified for the KIT exon 11 of c-KIT,
juxtamembrane domain, and sequenced [Frost et al., 2003]. Sequencing revealed mutations in
two of the four canine GISTs, one with a 6 base pair deletion, TGGAAG, and insertion of CAG,
predicted to translate to a deletion of tryptophan and lysine and an insertion of glycine at codon
556 [Frost et al., 2003]. This deletion is quite similar to the mutation at canine codons 556-557
discovered in the canine GISTs in our study. The second mutation discovered by Frost et al. was
a substitution of T with C predicted to replace codon 575 leucine with proline [Frost et al., 2003].
We did not detect a similar mutation in our study population. The mutations observed in our study
population of GISTs were clustered at codons 556-558 of c-KIT. No gender predilection has been
reported in human GISTs, and the observation of 73% female to 27% male ratio in our study is
interesting, but its significance needs further evaluation.
A simple deletion identical to the mutation at canine codons 556-557 in our study has also been
reported in multiple cases of human GISTs [Taniguchi et al., 1999; Agaimy et al., 2008a]. In the
study by Taniguchi et al., a deletion and point mutation similar to the mutation at canine codons

86

556-558 in our study was also detected in one of the cases they analyzed [Taniguchi et al., 1999].
Rubin et al., found the same mutation as the deletion of canine codons 556-557 in our canine GISTs
in 2 of 48 human GISTs, and they reported the same deletion of canine codons 556-558 in 1 of
the 48 cases [Rubin et al., 2001]. In a study of human familial GISTs, germline deletion mutations
were discovered in the same region as the previously mentioned deletions [Nishida et al., 1998].
All four of the SNPs found in our canine GIST samples were silent mutations, with no change
predicted in the translated protein. The SNP in exon 12 of PDGFRA at genomic base pair 49690424
[GenBank:NC_006595.2], has been reported previously [Lindblad-Toh et al., 2005].
In humans, GISTs are rare neoplasms. The age-adjusted incidence of gastric mesenchymal
tumors was 0.31 per 100,000 population in 2002, of which 82% were classified as GISTs [Perez
et al., 2006]. The population incidence of GISTs is difficult to determine in dogs. Frost et al.
commented that in dogs, gastrointestinal neoplasias account for 12-120 cases per 10,000 neoplasia
cases, and in our study GISTs accounted for 39% of the total number of gastrointestinal tumors
collected during the study period.
Heterozygosity with regard to mutations in the tumor sections resulted in an easily detectable
aberrant banding pattern on agarose gels. While the gel electrophoresis used in this study does
not resolve the normal versus the mutant alleles, which differ by only six base pairs, the normal
and mutant alleles formed a heteroduplex, which contained a bubble created by the longer normal
allele. The heteroduplex structure is predicted to generate a drag during gel electrophoresis giving
rise to the higher band and allowing easy detection of this relatively small deletion [Bhattacharyya
and Lilley, 1989]. We cannot be absolutely certain that the tumor cells are heterozygous with
respect to the mutation, as the tumor sections contained some non-neoplastic components such as
blood vessels. Regardless, the aberrant banding is a useful screening tool for this set of mutations.

A.6

Conclusions

These data substantially expand the number of canine gastrointestinal stromal tumors evaluated
for mutations in c-KIT by previous studies [Frost et al., 2003]. The mutations we have found
87

are clustered and consistent with those shown to be activating mutations in the c-KIT gene of
human tumors [Hirota et al., 1998]. Based on these data, we can conclude that the nature of c-KIT
mutations in GISTs in dogs is similar to that observed in humans.
The juxtamembrane domain of the KIT gene is a highly conserved region among mammals
[Ma et al., 1999b]. This juxtamembrane domain acts as a negative regulator of KIT activation
and thus, when this particular domain is mutated, the autoinhibition is removed, allowing KIT to
be activated in the absence of the KIT ligand [Mol et al., 2004]. The residues we found to be
deleted in our cases of canine GISTs are the very same that Ma et al. determined to increase basal
receptor phosphorylation when mutated in c-KIT [Ma et al., 1999a]. The expression of KIT and the
presence of these mutations in c-KIT implicate KIT in the pathogenesis of these tumors pointing
to spontaneous GISTs in the dog being a relevant model for the human disease. Our results also
indicate that mutations in KIT may be of prognostic and therapeutic significance in canine GISTs
as they are in canine cutaneous mast cell tumors.
Numerous small molecule inhibitors that target specific tyrosine kinases, tyrosine kinase inhibitors (TKIs), have successfully been used for the treatment of human and canine cancers with
mutations in KIT [London, 2009; Pytel et al., 2009]. Imatinib mesylate (Gleevec, Novartis, Basel,
Switzerland), has been utilized for its ability to inhibit protein tyrosine kinases since the Food and
Drug Administration approved it in 2001 for the treatment of Chronic Myelogenous Leukemia
(CML) [Savage and Antman, 2002]. Early trials demonstrated imatinib was also highly effective
against GISTs [Croom and Perry, 2003]. Imatinib caused marked tumor response rates and dramatically increased survival times in most patients [Hornick and Fletcher, 2007] and has now become the standard of care in the treatment of patients with advanced GISTs [Schnadig and Blanke,
2006]. However, since with prolonged treatment clinical resistance can develop, most likely due
to secondary c-KIT mutations, a new generation of TKIs, such as sunitinib, have been successfully
introduced [Hornick and Fletcher, 2007; Rutkowski et al., 2008]. Research is ongoing to treat
GISTs with resistance to imatinib and sunitinib [Nilsson et al., 2009]. In dogs, the TKIs Palladia
(toceranib), Kinavet (masitinib), and Gleevec (imatinib) have been successfully used in numerous

88

neoplastic diseases [London, 2009] and toceranib (Palladia, Pfizer, New York, NY, USA) and masitinib (Kinavet, AB Science, Short Hills, NJ, USA) have been registered for the use in dogs with
cutaneous mast cell tumors [Gleixner et al., 2007; Isotani et al., 2008; Marconato et al., 2008]. In
a randomized trial, dogs with KIT mutations were much more likely to respond to Palladia than
those without KIT mutations [London et al., 2009]. To our knowledge there are no published data
on the treatment of canine GISTs with TKIs. Based upon the data presented here, we propose that
targeting KIT may be a rational approach to treatment of canine GISTs as well. In addition, we
put forward that canine GISTs are a relevant and accessible model for human GISTs, with shared
molecular pathways that can be targeted for therapy.

89

BIBLIOGRAPHY

90

BIBLIOGRAPHY

Jerome Abadie, Benoit Hedan, Edouard Cadieu, Clotilde De Brito, Patrick Devauchelle, Catherine Bourgain, Heidi G. Parker, Amaury Vaysse, Patricia Margaritte-Jeannin, Francis Galibert,
Elaine A. Ostrander, and Catherine Andre. Epidemiology, pathology, and genetics of histiocytic
sarcoma in the Bernese Mountain Dog breed. Journal of Heredity, 100(suppl 1):S19–27, Jul
2009.
VJ Adams, KM Evans, J Sampson, and JLN Wood. Methods and mortality results of a health
survey of purebred dogs in the uk. Journal of Small Animal Practice, 51(10):512–524, 2010.
V. K. Affolter and P. F. Moore. Localized and disseminated histiocytic sarcoma of dendritic cell
origin in dogs. Vet Pathol, 39(1):74–83, Jan 2002.
A Agaimy, S Dirnhofer, PH WUnsch, LM Terracciano, L Tornillo, and MP Bihl. Multiple sporadic
gastrointestinal stromal tumors (GISTs) of the proximal stomach are caused by different somatic
KIT mutations suggesting a field effect. Am J Surg Pathol, 32(10):1553–1559, 2008a.
A Agaimy, PH Wunsch, S Dirnhofer, MP Bihl, LM Terracciano, and L Tornillo. Microscopic
gastrointestinal stromal tumors in esophageal and intestinal surgical resection specimens: A
clinicopathologic, immunohistochemical, and molecular study of 19 lesions. Am J Surg Pathol,
32(6), 2008b.
S Banerjee, W Makdisi, A Weston, S Mitchell, and D Campbell. Microwave-based dna extraction
from paraffin-embedded tissue for pcr amplification. Biotechniques, 18(5):768–770, 1995.
Jordi Barretina, Barry S Taylor, Shantanu Banerji, Alexis H Ramos, Mariana Lagos-Quintana,
Penelope L DeCarolis, Kinjal Shah, Nicholas D Socci, Barbara A Weir, Alan Ho, Derek Y Chiang, Boris Reva, Craig H Mermel, Gad Getz, Yevgenyi Antipin, Rameen Beroukhim, John E
Major, Charles Hatton, Richard Nicoletti, Megan Hanna, Ted Sharpe, Tim J Fennell, Kristian
Cibulskis, Robert C Onofrio, Tsuyoshi Saito, Neerav Shukla, Christopher Lau, Sven Nelander,
Serena J Silver, Carrie Sougnez, Agnes Viale, Wendy Winckler, Robert G Maki, Levi A
Garraway, Alex Lash, Heidi Greulich, David E Root, William R Sellers, Gary K Schwartz,
Cristina R Antonescu, Eric S Lander, Harold E Varmus, Marc Ladanyi, Chris Sander, Matthew
Meyerson, and Samuel Singer. Subtype-specific genomic alterations define new targets for softtissue sarcoma therapy. Nat Genet, 42:715–721, Aug 2010.
A Bhattacharyya and DM Lilley. The contrasting structures of mismatched dna sequences containing looped-out bases (bulges) and multiple mismatches (bubbles). Nucleic Acids Res, 17(17):
6821–6840, 1989.
J.M. Blechman, S. Lev, D. Givol, and Y. Yarden. Structure-function analyses of the kit receptor
for the steel factor. Stem Cells, 11(2):12–21, July 1993.
Kim M. Boerkamp, Marieke van der Kooij, Frank G. van Steenbeek, Monique E. van Wolferen,
Marian J. A. Groot Koerkamp, Dik van Leenen, Guy C. M. Grinwis, Louis C. Penning, Erik
91

A. C. Wiemer, and Gerard R. Rutteman. Gene expression profiling of histiocytic sarcomas in
a canine model: The predisposed Flatcoated Retriever Dog. PLoS ONE, 8(8):e71094 EP –, 08
2013.
Anthony M Bolger, Marc Lohse, and Bjoern Usadel. Trimmomatic: a flexible trimmer for illumina
sequence data. Bioinformatics, page btu170, 2014.
L. B. Brønden, S. S. Nielsen, N. Toft, and A. T. Kristensen. Data from the danish veterinary cancer
registry on the occurrence and distribution of neoplasms in dogs in denmark. Veterinary Record,
166(19):586–590, 05 2010.
George Adrian Calin, Calin Dan Dumitru, Masayoshi Shimizu, Roberta Bichi, Simona Zupo, Evan
Noch, Hansjuerg Aldler, Sashi Rattan, Michael Keating, Kanti Rai, Laura Rassenti, Thomas
Kipps, Massimo Negrini, Florencia Bullrich, and Carlo M. Croce. Frequent deletions and downregulation of micro-RNA genes miR15 and miR16 at 13q14 in chronic lymphocytic leukemia.
Proc Natl Acad Sci USA, 99(24):15524–15529, Nov 2002.
A Chompret, C Kannengiesser, M Barrois, P Terrier, P Dahan, T Tursz, G Lenoir, and B BressacDe Paillerets. Pdgfra germline mutation in a family with multiple cases of gastrointestinal stromal tumor. Gastroenterology, 126:318–321, 2004.
K Croom and C Perry. Imatinib mesylate: In the treatment of gastrointestinal stromal tumours.
Drugs, 63(5):513–522, 2003.
Brad T Sherman Da Wei Huang and Richard A Lempicki. Systematic and integrative analysis of
large gene lists using DAVID bioinformatics resources. Nature protocols, 4(1):44–57, 2008.
Brian W Davis and Elaine A Ostrander. Domestic dogs and cancer research: A breed-based genomics approach. ILAR Journal, 55(1):59–68, 2014.
M Debiec-Rychter, B Wasag, M Stul, A Hagemeijer, and R Sciot. Gastrointestinal stromal tumours
(GISTs) negative for KIT (CD117 antigen) immunoreactivity. J Pathol, 202(4):430–438, 2004.
F. Del Piero, B. A. Summers, J. F. Cummings, G. Mandelli, and E. A. Blomme. Gastrointestinal
stromal tumors in equids. Veterinary Pathology, 38(6):689–697, 2001.
Nikolaos G Dervisis, Shawna L Klahn, Matti Kiupel, Maciej Parys, Emmalena Gregory-Bryson,
and Vilma Yuzbasiyan-Gurkan. Localized and disseminated histiocytic sarcoma complex: A
canine model [abstract number 5092]. In Proceedings of the 103rd Annual Meeting of the American Association for Cancer Research, Chicago, Illinois, Mar 31-Apr 4 2012. AACR, Philadelphia (PA).
J Dobson, T Hoather, TJ McKinley, and JL Wood. Mortality in a cohort of flat-coated retrievers in
the UK. Vet Comp Oncol, 7(2):115–121, Jun 2009.
Martina Doleshal, Amber A. Magotra, Bhavna Choudhury, Brian D. Cannon, Emmanuel
Labourier, and Anna E. Szafranska. Evaluation and validation of total RNA extraction methods
for microRNA expression analyses in formalin-fixed, paraffin-embedded tissues. J Mol Diagn,
10(3):203–211, 2008.
92

Tingting Du and Phillip D. Zamore. microPrimer: the biogenesis and function of microRNA.
Development, 132(21):4645–4652, Nov 2005.
R. Maarten Egeler and Giulio J. D’Angio. Langerhans cell histiocytosis. The Journal of Pediatrics,
127(1):1–11, 7 1995.
Suzanne A Erich, Gerard R Rutteman, and Erik Teske. Causes of death and the impact of histiocytic
sarcoma on the life expectancy of the Dutch population of Bernese mountain dogs and Flatcoated retrievers. The Veterinary Journal, 198(3):678–683, 2013.
Aurora Esquela-Kerscher and Frank J. Slack. Oncomirs — microRNAs with a role in cancer. Nat
Rev Cancer, 6(4):259–269, Apr 2006.
A Fire, S Xu, M K Montgomery, S A Kostas, S E Driver, and C C Mello. Potent and specific
genetic interference by double-stranded RNA in Caenorhabditis elegans. Nature, 391(6669):
806–11, Feb 1998.
D. Frost, J. Lasota, and M. Miettinen. Gastrointestinal stromal tumors and leiomyomas in the dog:
A histopathologic, immunohistochemical, and molecular genetic study of 50 cases. Vet Pathol,
40(1):42–54, 2003.
Yasunori Fujita, Keitaro Kojima, Nanako Hamada, Riyako Ohhashi, Yukihiro Akao, Yoshinori
Nozawa, Takashi Deguchi, and Masafumi Ito. Effects of mir-34a on cell growth and chemoresistance in prostate cancer pc3 cells. Biochemical and biophysical research communications,
377(1):114–119, 2008.
Helmut Gadner, Nicole Grois, Maurizio Arico, Valerie Broadbent, Adriana Ceci, Ake Jakobson,
Diane Komp, Jörg Michaelis, Stacy Nicholson, Ulrike Pötschger, et al. A randomized trial
of treatment for multisystem Langerhans cell histiocytosis. The Journal of pediatrics, 138(5):
728–734, 2001.
K Gleixner, L Rebuzzi, M Mayerhofer, A Gruze, E Hadzijusufovic, K Sonneck, A Vales, M Kneidinger, P Samorapoompichit, T Thaiwong, W Pickl, V Yuzbasiyan-Gurkan, C Sillaber, M Willmann, and P Valent. Synergistic antiproliferative effects of KIT tyrosine kinase inhibitors on
neoplastic canine mast cells. Exp Hematol, 35(10):1510–1521, 2007.
Martin Glud, Mikkel Klausen, Robert Gniadecki, Maria Rossing, Nina Hastrup, Finn C Nielsen,
and Krzysztof T Drzewiecki. MicroRNA expression in melanocytic nevi: The usefulness of
formalin-fixed, paraffin-embedded material for miRNA microarray profiling. J Invest Dermatol,
129(5):1219–1224, Nov 2008.
Emmalena Gregory-Bryson, Elizabeth Bartlett, Matti Kiupel, Schantel Hayes, and Vilma
Yuzbasiyan-Gurkan. Canine and human gastrointestinal stromal tumors display similar mutations in c-KIT exon 11. BMC Cancer, 10(1), 2010.
Huili Guo, Nicholas T. Ingolia, Jonathan S. Weissman, and David P. Bartel. Mammalian microRNAs predominantly act to decrease target mRNA levels. Nature, 466(7308):835–840, Aug
2010.

93

Scott Hafeman, Cheryl London, Robyn Elmslie, and Steven Dow. Evaluation of liposomal clodronate for treatment of malignant histiocytosis in dogs. Cancer Immunol Immunother, 59(3):
441–452, Mar 2010.
S. Hafner, B. G. Harmon, and T. King. Gastrointestinal stromal tumors of the equine cecum.
Veterinary Pathology, 38(2):242–246, 2001.
Benoit Hedan, Rachael Thomas, Alison Motsinger-Reif, Jerome Abadie, Catherine Andre, John
Cullen, and Matthew Breen. Molecular cytogenetic characterization of canine histiocytic sarcoma: A spontaneous model for human histiocytic cancer identifies deletion of tumor suppressor
genes and highlights influence of genetic background on tumor behavior. BMC Cancer, 11(1),
2011.
M Heinrich, C Corless, A Duensing, L McGreevey, C Chen, N Joseph, S Singer, D Griffith, A Haley, A Town, G Demetri, C Fletcher, and J Fletcher. Pdgfra activating mutations in gastrointestinal stromal tumors. Science, 299(5607):708–710, 2003.
Michael C. Heinrich, Brian P. Rubin, B. Jack Longley, and Jonathan A. Fletcher. Biology and
genetic aspects of gastrointestinal stromal tumors: KIT activation and cytogenetic alterations.
Human Pathology, 33(5):484–495, 2002.
Seiichi Hirota and Koji Isozaki. Pathology of gastrointestinal stromal tumors. Pathology International, 56(1):1–9, 2006.
Seiichi Hirota, Koji Isozaki, Yasuhiro Moriyama, Koji Hashimoto, Toshirou Nishida, Shingo Ishiguro, Kiyoshi Kawano, Masato Hanada, Akihiko Kurata, Masashi Takeda, Ghulam Muhammad Tunio, Yuji Matsuzawa, Yuzuru Kanakura, Yasuhisa Shinomura, and Yukihiko Kitamura.
Gain-of-function mutations of c-KIT in human gastrointestinal stromal tumors. Science, 279
(5350):577–580, 1998.
J Hornick and C Fletcher. The role of KIT in the management of patients with gastrointestinal
stromal tumors. Hum Pathol, 38(5):679–687, 2007.
Yali Hou, Yachun Wang, Xuemei Lu, Xu Zhang, Qian Zhao, Rory J Todhunter, and Zhiwu Zhang.
Monitoring hip and elbow dysplasia achieved modest genetic improvement of 74 dog breeds
over 40 years in usa. PLoS One, 8(10):e76390, 2013.
Douglas M. Howarth, Gerald S. Gilchrist, Brian P. Mullan, Gregory A. Wiseman, John H. Edmonson, and Paula J. Schomberg. Langerhans cell histiocytosis. Cancer, 85(10):2278–2290,
1999.
Qihong Huang, Kiranmai Gumireddy, Mariette Schrier, Carlos Le Sage, Remco Nagel, Suresh
Nair, David A Egan, Anping Li, Guanghua Huang, Andres J Klein-Szanto, et al. The micrornas
mir-373 and mir-520c promote tumour invasion and metastasis. Nature cell biology, 10(2):
202–210, 2008.
Angela B Y Hui, Wei Shi, Paul C Boutros, Naomi Miller, Melania Pintilie, Tony Fyles, David
McCready, Derek Wong, Kate Gerster, Igor Jurisica, Linda Z Penn, and Fei-Fei Liu. Robust

94

global micro-RNA profiling with formalin-fixed paraffin-embedded breast cancer tissues. Lab
Invest, 89(5):597–606, Mar 2009.
Jan D. Hulzinga, Lars Thuneberg, Michael Kluppel, John Malysz, Hanne B. Mikkelsen, and Alan
Bernstein. W/kit gene required for interstitial cells of cajal and for intestinal pacemaker activity.
Nature, 373(6512):347–349, 1995.
Boaz Inbal, Ofer Cohen, Sylvie Polak-Charcon, Juri Kopolovic, Ezra Vadai, Lea Eisenbach, and
Adi Kimchi. Dap kinase links the control of apoptosis to metastasis. Nature, 390(6656):180–
184, 1997.
Marilena V Iorio, Manuela Ferracin, Chang-Gong Liu, Angelo Veronese, Riccardo Spizzo, Silvia
Sabbioni, Eros Magri, Massimo Pedriali, Muller Fabbri, Manuela Campiglio, et al. Microrna
gene expression deregulation in human breast cancer. Cancer research, 65(16):7065–7070,
2005.
D. N. Irion, A. L. Schaffer, T. R. Famula, M. L. Eggleston, S. S. Hughes, and N. C. Pedersen. Analysis of genetic variation in 28 dog breed populations with 100 microsatellite markers. Journal
of Heredity, 94(1):81–87, 01 2003.
M Isotani, N Ishida, M Tominaga, K Tamura, H Yagihara, S Ochi, R Kato, T Kobayashi, M Fujita, Y Fujino, A Setoguchi, K Ono, T Washizu, and M Bonkobara. Effect of tyrosine kinase
inhibition by imatinib mesylate on mast cell tumors in dogs. J Vet Intern Med, 22(4):985–988,
2008.
C Jones, R Grahn, M Chien, L Lyons, and C London. Detection of c-KIT mutations in canine
mast cell tumors using fluorescent polyacrylamide gel electrophoresis. J Vet Diagn Invest, 16
(2):95–100, 2004.
L.G. Kindblom, H.E. Remotti, F. Aldenborg, and J.M. Meis-Kindblom. Gastrointestinal pacemaker cell tumor (gipact): gastrointestinal stromal tumors show phenotypic characteristics of
the interstitial cells of cajal. American Journal of Pathology, 152(5):1259–1269, May 1998.
Shawna L. Klahn, Barbara E. Kitchell, and Nikolaos G. Dervisis. Evaluation and comparison of
outcomes in dogs with periarticular and nonperiarticular histiocytic sarcoma. Journal of the
American Veterinary Medical Association, 239(1):90–96, 2011/07/02 2011.
Wigard P. Kloosterman and Ronald H. A. Plasterk. The diverse functions of microRNAs in animal
development and disease. Dev Cell, 11(4):441–450, Oct 2006.
Kazuyoshi Kumagai, Kazuyuki Uchida, Toru Miyamoto, Takahiro Ushigusa, Syusaku Shinohara,
Ryoji Yamaguchi, and Susumu Tateyama. Three cases of canine gastroinstestinal stromal tumors
with multiple differentiations and c-KIT-expression. Journal of Veterinary Medical Science, 65
(10):119–1122, 2003.
Prateek Kumar, Steven Henikoff, and Pauline C Ng. Predicting the effects of coding nonsynonymous variants on protein function using the sift algorithm. Nature protocols, 4(7):1073–
1081, 2009.

95

Mariana Lagos-Quintana, Reinhard Rauhut, Winfried Lendeckel, and Thomas Tuschl. Identification of novel genes coding for small expressed RNAs. Science, 294(5543):853–858, 10 2001.
R. G. LaRock and P. E. Ginn. Immunohistochemical staining characteristics of canine gastrointestinal stromal tumors. Veterinary Pathology, 34(4):303–311, 1997.
ICM Lavrijsen, HCM Heuven, BP Meij, LFH Theyse, RC Nap, PAJ Leegwater, and HAW
Hazewinkel. Prevalence and co-occurrence of hip dysplasia and elbow dysplasia in dutch purebred dogs. Preventive veterinary medicine, 114(2):114–122, 2014.
Benjamin P Lewis, Christopher B Burge, and David P Bartel. Conserved seed pairing, often
flanked by adenosines, indicates that thousands of human genes are microRNA targets. Cell,
120(1):15–20, Jan 2005.
Heng Li and Richard Durbin. Fast and accurate short read alignment with burrows–wheeler transform. Bioinformatics, 25(14):1754–1760, 2009.
Heng Li, Bob Handsaker, Alec Wysoker, Tim Fennell, Jue Ruan, Nils Homer, Gabor Marth,
Goncalo Abecasis, Richard Durbin, et al. The sequence alignment/map format and samtools.
Bioinformatics, 25(16):2078–2079, 2009.
Jinghuan Li, Paul Smyth, Richard Flavin, Susanne Cahill, Karen Denning, Sinead Aherne, Simone
Guenther, John O’Leary, and Orla Sheils. Comparison of miRNA expression patterns using
total RNA extracted from matched samples of formalin-fixed paraffin-embedded (FFPE) cells
and snap frozen cells. BMC Biotechnol, 7(1):36, 2007. ISSN 1472-6750.
K Lindblad-Toh, CM Wade, TS Mikkelsen, EK Karlsson, DB Ja e, M Kamal, M Clamp, JL Chang,
EJ Kulbokas, MC Zody, E Mauceli, X Xie, M Breen, RK Wayne, EA Ostrander, CP Ponting,
F Galibert, DR Smith, PJ deJong, E Kirkness, P Alvarez, T Biagi, W Brockman, J Butler,
CW Chin, A Cook, J Cu, MJ Daly, D DeCaprio, S Gnerre, M Grabherr, M Kellis, M Kleber,
C Bardeleben, L Goodstadt, A Heger, C Hitte, L Kim, KP Koepfli, HG Parker, JP Pollinger,
SMJ Searle, NB Sutter, R Thomas, C Webber, Broad Institute Genome Sequencing Platform,
and ES Lander. Genome sequence, comparative analysis and haplotype structure of the domestic
dog. Nature, 438(7069):803–819, 2005.
CA London. Tyrosine kinase inhibitors in veterinary medicine. Top Companion Anim Med, 24(3):
106–112, 2009.
Cheryl A. London, Stephen J. Galli, Toshifumi Yuuki, Zhi-Qing Hu, Stuart C. Helfand, and Edwin N. Geissler. Spontaneous canine mast cell tumors express tandem duplications in the protooncogene c-KIT. Experimental Hematology, 27(4):689–697, 4 1999.
Cheryl A. London, Phyllis B. Malpas, Stacey L. Wood-Follis, Joseph F. Boucher, Anthony W.
Rusk, Mona P. Rosenberg, Carolyn J. Henry, Kathy L. Mitchener, Mary K. Klein, John G.
Hintermeister, Philip J. Bergman, Guillermo C. Couto, Guy N. Mauldin, and Gina M. Michels.
Multi-center, placebo-controlled, double-blind, randomized study of oral toceranib phosphate
(SU11654), a receptor tyrosine kinase inhibitor, for the treatment of dogs with recurrent (either
local or distant) mast cell tumor following surgical excision. Clinical Cancer Research, 15(11):
3856–3865, 6 2009.
96

B. Jack Longley, Maria Jose Reguera, and Yongsheng Ma. Classes of c-KIT activating mutations: proposed mechanisms of action and implications for disease classification and therapy.
Leukemia Research, 25(7):571–576, 2001.
Y Ma, BJ Longley, X Wang, JL Blount, K Langley, and GH Caughey. Clustering of activating
mutations in c-KIT’s juxtamembrane coding region in canine mast cell neoplasms. J Invest
Dermatol, 112(2):165–170, 1999a.
Yongsheng Ma, B Jack Longley, Xiaomei Wang, John L Blount, Keith Langley, and George H
Caughey. Clustering of activating mutations in c-KIT’s juxtamembrane coding region in canine
mast cell neoplasms. Journal of Investigative Dermatology, 112(2):165–170, 1999b.
L Marconato, G Bettini, C Giacoboni, G Romanelli, A Cesari, A Zatelli, and E Zini. Clinicopathological features and outcome for dogs with mast cell tumors and bone marrow involvement. J
Vet Intern Med, 22(4):1001–1007, 2008.
Julien Meunier, Frédéric Lemoine, Magali Soumillon, Angélica Liechti, Manuela Weier, Katerina
Guschanski, Haiyang Hu, Philipp Khaitovich, and Henrik Kaessmann. Birth and expression
evolution of mammalian microrna genes. Genome Research, 23(1):34–45, 01 2013.
M. Miettinen and J. Lasota. Gastrointestinal stromal tumors: review on morphology, molecular pathology, prognosis, and differential diagnosis. Archives of Pathology and Laboratory
Medicine, 130(10):1466–1478, Oct 2006.
Markku Miettinen, Wa’El El-Rifai, Leslie H, and Jerzy Lasota. Evaluation of malignancy and
prognosis of gastrointestinal stromal tumors: A review. Human pathology, 33(5):478–483, 2002.
C Mol, D Dougan, T Schneider, R Skene, M Kraus, D Scheibe, G Snell, H Zou, B Sang, and
K Wilson. Structural basis for the autoinhibition and STI-571 inhibition of c-KIT tyrosine
kinase. J Biol Chem, 279(30):31655–31663, 2004.
Peter F. Moore. Systemic histiocytosis of Bernese mountain dogs. Vet Pathol, 21(6):554–563, Nov
1984.
PF Moore. Utilization of cytoplasmic lysozyme immunoreactivity as a histiocytic marker in canine
histiocytic disorders. Vet Pathol, 23(6):757–762, Nov 1986.
M. Mortarino, G. Gioia, M.E. Gelain, F. Albonico, P. Roccabianca, E. Ferri, and S. Comazzi.
Identification of suitable endogenous controls and differentially expressed microRNAs in canine
fresh-frozen and FFPE lymphoma samples. Leuk Res, 34(8):1070 – 1077, 2010. ISSN 01452126.
B A Neilan, A N Wilton, and D Jacobs. A universal procedure for primer labelling of amplicons.
Nucleic Acids Research, 25(14):2938–2939, 07 1997.
Pauline C Ng and Steven Henikoff. Predicting deleterious amino acid substitutions. Genome
research, 11(5):863–874, 2001.
Pauline C Ng and Steven Henikoff. Accounting for human polymorphisms predicted to affect
protein function. Genome research, 12(3):436–446, 2002.
97

Pauline C Ng and Steven Henikoff. Sift: Predicting amino acid changes that affect protein function.
Nucleic acids research, 31(13):3812–3814, 2003.
H. Stacy Nicholson, R. Maarten Egeler, and Mark E. Nesbit. The epidemiology of langerhans cell
histiocytosis. Hematology/Oncology Clinics of North America, 12(2):379–384, 4 1998.
L. Nielsen, S. N. Andreasen, S. D. Andersen, and A. T. Kristensen. Malignant histiocytosis and
other causes of death in Bernese mountain dogs in Denmark. Vet Rec, 166(7):199–202, Feb
2010.
B Nilsson, O Nilsson, and H Ahlman. Treatment of gastrointestinal stromal tumours: imatinib,
sunitinib - and then? Expert Opin Investig Drugs, 18(4):457–468, 2009.
Toshirou Nishida, Seiichi Hirota, Masahiko Taniguchi, Koji Hashimoto, Koji Isozaki, Harumi
Nakamura, Yuzuru Kanakura, Tomoyuki Tanaka, Arimichi Takabayashi, Hikaru Matsuda, and
Yukihiko Kitamura. Familial gastrointestinal stromal tumours with germline mutation of the kit
gene. Nat Genet, 19(4):323–324, 1998.
Dan G. O’Neill, David B. Church, Paul D. McGreevy, Peter C. Thomson, and Dave C. Brodbelt.
Prevalence of disorders recorded in dogs attending primary-care veterinary practices in England.
PLoS ONE, 9(3):e90501 EP, 03 2014.
J Ostrowski, M Polkowski, A Paziewska, M Skrzypczak, K Goryca, T Rubel, K Kokoszynska,
P Rutkowski, Z Nowecki, A Dobosz, D Jarosz, W Ruka, and L Wyrwicz. Functional features
of gene expression pro les differentiating gastrointestinal stromal tumours according to kit mutations and expression. BMC Cancer, 9, 2009.
GA Padgett, BR Madewell, ET Keller, L Jodar, and M Packard. Inheritance of histiocytosis in
Bernese mountain dogs. J Small Anim Pract, 36(3):93–98, Mar 1995.
EA Perez, AS Livingstone, D Franceschi, C Rocha-Lima, DJ Lee, N Hodgson, M Jorda, and
LG Koniaris. Current incidence and outcomes of gastrointestinal mesenchymal tumors including
gastrointestinal stromal tumors. J Am Coll Surg, 202(4):623–629, 2006.
Lasse Peters and Gunter Meister. Argonaute proteins: Mediators of RNA silencing. Mol Cell, 26
(5):611–623, Jun 2007.
D Pytel, T Sliwinski, T Poplawski, D Ferriola, and I Majsterek. Tyrosine kinase blockers: new
hope for successful cancer therapy. Anticancer Agents Med Chem, 9:66–76, 2009.
F H Qiu, P Ray, K Brown, P E Barker, S Jhanwar, F H Ruddle, and P Besmer. Primary structure
of c-KIT: relationship with the CSF-1/PDGF receptor kinase family-oncogenic activation of vKIT involves deletion of extracellular domain and C terminus. EMBO Journal, 7(4):1003–1011,
April 1988.
Pascale Quignon, Laetitia Herbin, Edouard Cadieu, Ewen F. Kirkness, Benoit Hédan, Dana S.
Mosher, Francis Galibert, Catherine André, Elaine A. Ostrander, and Christophe Hitte. Canine
population structure: Assessment and impact of intra-breed stratification on snp-based association studies. PLoS ONE, 2(12):e1324 EP –, 12 2007.
98

Mangalathu S. Rajeevan, Daya G. Ranamukhaarachchi, Suzanne D. Vernon, and Elizabeth R.
Unger. Use of Real-Time Quantitative PCR to validate the results of cDNA array and differential
display PCR technologies. Methods, 25(4):443–451, 12 2001.
E Ralfkiaer, G Delsol, NT O’Connor, P Brandtzaeg, P Brousset, GL Vejlsgaard, and DY Mason.
Malignant lymphomas of true histiocytic origin. A clinical, histological, immunophenotypic and
genotypic study. J Pathol, 160(1):9–17, Jan 1990.
IK Ramsey, JS McKay, H Rudorf, and JM Dobson. Malignant histiocytosis in three Bernese
mountain dogs. Vet Rec, 138(18):440–444, May 1996.
A Rosin, Peter F. Moore, and Richard R. Dubielzig. Malignant histiocytosis in Bernese Mountain
dogs. J Am Vet Med Assoc, 188(9):1041–1045, May 1986.
Robert Roskoski, Jr. Signaling by KIT protein-tyrosine kinase–the stem cell factor receptor. Biochemical and Biophysical Research Communications, 337(1):1–13, 2005.
Brian P. Rubin, Samuel Singer, Connie Tsao, Anette Duensing, Marcia L. Lux, Robert Ruiz,
Michele K. Hibbard, Chang-Jie Chen, Sheng Xiao, David A. Tuveson, George D. Demetri,
Christopher D. M. Fletcher, and Jonathan A. Fletcher. KIT activation is a ubiquitous feature of
gastrointestinal stromal tumors. Cancer Research, 61(22):8118–8121, 2001.
Brian P. Rubin, Cristina R. Antonescu, James P. Scott-Browne, Melissa L. Comstock, Yansong Gu,
Munir R. Tanas, Carol B. Ware, and Jessica Woodell. A knock-in mouse model of gastrointestinal stromal tumor harboring KIT K641E. Cancer Research, 65(15):6631–6639, 2005.
P Rutkowski, M Symonides, M Zdzienicki, and JA Siedlecki. Developments in targeted therapy
of advanced gastrointestinal stromal tumors. Recent Pat Anticancer Drug Discov, 3(2):88–99,
2008.
J A Salotti, V Nanduri, M S Pearce, L Parker, R Lynn, and K P Windebank. Incidence and
clinical features of Langerhans cell histiocytosis in the UK and Ireland. Archives of Disease in
Childhood, 94(5):376–380, 05 2009.
D Savage and K Antman. Imatinib mesylate-a new oral targeted therapy. N Engl J Med, 346(9):
683–693, 2002.
Laura Schmitz and Blaise E. Favara. Nosology and pathology of Langerhans Cell Histiocytosis.
Hematol Oncol Clin N Am, 12(2):221–246, 1998.
I Schnadig and C Blanke. Gastrointestinal stromal tumors: imatinib and beyond. Curr Treat
Options Oncol, 7(6):427–437, 2006.
Abigail L. Shearin, Benoit Hedan, Edouard Cadieu, Suzanne A. Erich, Emmett V. Schmidt,
Daniel L. Faden, John Cullen, Jerome Abadie, Erika M. Kwon, Andrea Gröne, Patrick Devauchelle, Maud Rimbault, Danielle M. Karyadi, Mary Lynch, Francis Galibert, Matthew Breen,
Gerard R. Rutteman, Catherine André, Heidi G. Parker, and Elaine A. Ostrander. The MTAPCDKN2A locus confers susceptibility to a naturally occurring canine cancer. Cancer Epidemiology Biomarkers & Prevention, 05 2012.
99

Xu-Bao Shi, Lingru Xue, Joy Yang, Ai-Hong Ma, Jianjun Zhao, Ma Xu, Clifford G Tepper,
Christopher P Evans, Hsing-Jien Kung, and Ralph W deVere White. An androgen-regulated
miRNA suppresses Bak1 expression and induces androgen-independent growth of prostate cancer cells. Proceedings of the National Academy of Sciences, 104(50):19983–19988, 2007.
KA Skorupski, CA Clifford, MC Paoloni, A Lara-Garcia, L Barber, MS Kent, AK LeBlanc,
A Sabhlok, EA Mauldin, FS Shofer, CG Couto, and KU Sørenmo. CCNU for the treatment
of dogs with histiocytic sarcoma. J Vet Intern Med, 21(1):121–126, Jan-Feb 2007.
Gunhild Sommer, Valter Agosti, Imke Ehlers, Ferdinand Rossi, Selim Corbacioglu, Judith Farkas,
Malcolm Moore, Katia Manova, Cristina R. Antonescu, and Peter Besmer. Gastrointestinal
stromal tumors in a mouse model by targeted mutation of the KIT receptor tyrosine kinase.
Proceedings of the National Academy of Sciences, 100(11):6706–6711, 2003.
Sonja E. Steigen and Tor J. Eide. Gastrointestinal stromal tumors (GISTs): a review. Acta Pathologica, Microbiologica et Immunologica Scandinavica, 117(2):73–86, 2009.
S Tabone, N Theou, A Wozniak, R Saffroy, L Deville, C Julie, P Callard, A Lavergne-Slove,
M Debiec-Rychter, A Lemoine, and J Emile. KIT overexpression and amplification in gastrointestinal stromal tumors (GISTs). Biochim Biophys Acta, 1741(1-2):165–172, 2005.
Marilia Takada, Maciej Parys, Emmalena Gregory-Bryson, and Vilma Yuzbasiyan-Gurkan. A
novel canine histiocytic sarcoma cell line provides a potential path to effective treatments with
relevance for translational and comparative studies in humans [abstract number 3931]. In Proceedings of the 105th Annual Meeting of the American Association for Cancer Research, San
Diego, CA, Apr 5-9 2014. AACR, Philadelphia (PA).
Masahiko Taniguchi, Toshirou Nishida, Seiichi Hirota, Koji Isozaki, Toshinori Ito, Taisei Nomura,
Hikaru Matsuda, and Yukihiko Kitamura. Effect of c-KIT mutation on prognosis of gastrointestinal stromal tumors. Cancer Research, 59(17):4297–4300, 1999.
Mark A Travis, Boris Reizis, Andrew C Melton, Emma Masteller, Qizhi Tang, John M Proctor,
Yanli Wang, Xin Bernstein, Xiaozhu Huang, Louis F Reichardt, et al. Loss of integrin alphaV/beta-8 on dendritic cells causes autoimmunity and colitis in mice. Nature, 449(7160):361–
365, 2007.
L Van Kuijk, K van Ginkel, JP de Vos, MJ Brearley, J Butinar, Ingrid Gielen, E van Garderen,
Koen Chiers, and PS Verhoeven. Peri-articular histiocytic sarcoma and previous joint disease in
Bernese mountain dogs. Journal of Veterinary Internal Medicine, 27(2):293–299, 2013.
Jeffrey A Vos, Susan L Abbondanzo, Carol L Barekman, JoAnn W Andriko, Markku Miettinen,
and Nadine S Aguilera. Histiocytic sarcoma: a study of five cases including the histiocyte
marker CD163. Mod Pathol, 18(5):693–704, Jan 2005.
Xiaohong Wang, Shuang Tang, Shu-Yun Le, Robert Lu, Janet S. Rader, Craig Meyers, and ZhiMing Zheng. Aberrant expression of oncogenic and tumor-suppressive microRNAs in cervical
cancer is required for cancer cell growth. PLoS ONE, 3(7):e2557–, Jul 2008.

100

Joshua D Webster, Matti Kiupel, and Vilma Yuzbasiyan-Gurkan. Evaluation of the kinase domain
of c-KIT in canine cutaneous mast cell tumors. BMC Cancer, 6(85), April 2006a.
Joshua D. Webster, Vilma Yuzbasiyan-Gurkan, John B. Kaneene, RoseAnn Miller, James H. Resau, and Matti Kiupel. The role of c-KIT in tumorigenesis: Evaluation in canine cutaneous mast
cell tumors. Neoplasia, 8(2):104–111, February 2006b.
ML Wellman, S Krakowka, RM Jacobs, and GJ Kociba. A macrophage-monocyte cell line from a
dog with malignant histiocytosis. In Vitro Cell Dev Biol, 24(3):223–229, Mar 1988.
NACS Wong, R Young, RDG Malcomson, AG Nayar, LA Jamieson, VE Save, FA Carey,
DH Brewster, C Han, and A Al-Nafussi. Prognostic indicators for gastrointestinal stromal
tumours: a clinicopathological and immunohistochemical study of 108 resected cases of the
stomach. Histopathology, 43(2):118–126, 2003.
Yaguang Xi, Go Nakajima, Elaine Gavin, Chris G. Morris, Kenji Kudo, Kazuhiko Hayashi, and
Jingfang Ju. Systematic analysis of microRNA expression of RNA extracted from fresh frozen
and formalin-fixed paraffin-embedded samples. RNA, 13:1668–1674, 2007.
Y Yarden, W J Kuang, T Yang-Feng, L Coussens, S Munemitsu, T J Dull, E Chen, J Schlessinger,
U Francke, and A Ullrich. Human proto-oncogene c-KIT: a new cell surface receptor tyrosine
kinase for an unidentified ligand. EMBO Journal, 6(11):3341–3351, November 1987.
Gaetano Zafarana, Ad JM Gillis, Ruud JHLM van Gurp, Per G Olsson, Fons Elstrodt, Hans Stoop,
José Luis Millán, J Wolter Oosterhuis, and Leendert HJ Looijenga. Coamplification of DADR, SOX5, and EKI1 in human testicular seminomas, with specific overexpression of DAD-R,
correlates with reduced levels of apoptosis and earlier clinical manifestation. Cancer research,
62(6):1822–1831, 2002.
Regina Zavodovskaya, Albert T. Liao, Cameron L. R. Jones, Becky Yip, May B. Chien, Peter F.
Moore, and Cheryl A. London. Evaluation of dysregulation of the receptor tyrosine kinases Kit,
Flt3, and Met in histiocytic sarcomas of dogs. Am J Vet Res, 67(4):633–641, Nov 2006.
D. Zemke, B. Yamini, and V. Yuzbasiyan-Gurkan. Mutations in the juxtamembrane domain of
c-KIT are associated with higher grade mast cell tumors in dogs. Veterinary Pathology, 39(5):
529–535, 2002.
Xiao Zhang, Jiamin Chen, Tom Radcliffe, Dave P. LeBrun, Victor A. Tron, and Harriet Feilotter.
An array-based analysis of microRNA expression comparing matched frozen and formalin-fixed
paraffin-embedded human tissue samples. J Mol Diagn, 10(6):513–519, 2008.
Ming Zhou, Zixing Liu, Yuhua Zhao, Yan Ding, Hao Liu, Yaguang Xi, Wei Xiong, Guiyuan Li,
Jianrong Lu, Oystein Fodstad, et al. MicroRNA-125b confers the resistance of breast cancer
cells to paclitaxel through suppression of pro-apoptotic Bcl-2 antagonist killer 1 (Bak1) expression. Journal of Biological Chemistry, 285(28):21496–21507, 2010.

101